Remediation techniques and heavy metal uptake by different rice varieties in metal-contaminated soils of Taiwan: New aspects for food safety regulation and sustainable agriculture


Z.-S. CHEN, Department of Agricultural Chemistry, National Taiwan University, Taipei 10617, Taiwan. Email:


Rice is one of the most important staple foods worldwide. Soil contamination with heavy metals and food safety problems occur in many countries as a result of numerous human activities, particularly wastewater and solid waste disposal. This review paper provides a schematic summary of heavy metals in identification processes, transport in soil to different rice varieties, and soil remediation strategies and techniques surrounding the agro-environmental impact in paddy soils based on a description of Taiwan’s experiences and database. In terms of the soil control standard, heavy metals including As, Cd, Cr, Cu, Hg, Ni, Pb and Zn are regulated by the Soil and Ground Water Pollution Remediation Act of Taiwan. Owing to the heavy metal source from wastewater along irrigation systems, heavy metals not only accumulate in the surface soil (0–30 cm), but are also highly distributed at the main entrance of irrigation water into individual paddy fields. Moreover, sediments in the irrigation canal have to be dredged and the irrigation system needs to be isolated from the discharge system of wastewater to maintain soil quality. Cadmium in rice grains accumulates more significantly in Indica varieties than in Japonica varieties, and this accumulation exceeds the food quality standard. The best well-performing metal uptake models have been developed to predict Cd levels in rice grains for Indica and Japonica varieties using soil bioavailable Cd and Zn concentrations extracted by 0.01 mol L−1 CaCl2. Soil remediation techniques, including turnover and dilution, in situ stabilization by chemical amendments and phytoremediation, have been tested and recommended in Taiwan. Although the high background levels of As, Cr and Ni, which were higher the soil control standard in some paddy soils, are derived from andesite and serpentinites in Taiwan, rice quality and yield were not adversely affected by these metals when labile concentrations were very low. Overall, it is necessary to identify the bioavailability of heavy metals in different soil types from specific case studies to provide reliable parameters for health-based risk assessments and to further achieve the goal of food safety and sustainable agriculture.


Heavy metal contamination in soils is an increasingly urgent problem throughout the world, and the clean up of these soils costs time and is difficult. Unlike organic contaminants, heavy metals are generally immutable, not degradable and are persistent in soils. Soils have a natural capacity to attenuate the bioavailability and movement of heavy metals through them by means of different mechanisms (precipitation, adsorption process and redox reactions). When the concentrations of heavy metals become too high to allow the soils to limit their potential effects, the metals can be mobilized, resulting in serious contamination of agricultural products and the environment.

Rice (Oryza sativa L.), an important cereal crop in global agriculture, is second (more than 150 million ha) after wheat in terms of the planting area of cereal crops; however, rice production is mainly distributed in Monsoon Asia, with more than 90% of the world rice growing area occurring in this region (Kyuma 2004). Therefore, there are a great number of issues with regard to soil quality, food safety and human health for the agro-environmental sustainability of world rice supplies. Water resources for rice production in Taiwan have been contaminated by the illegal discharge of industrial parks and livestock wastewater, and paddy soil quality and rice safety have been adversely affected by heavy metal contamination.

In this review, human activities associated with wastewater and solid waste disposal are discussed to clarify the anthropogenic sources of contaminated soils in Taiwan. Survey processes and the ensuing regulations for soil contamination are recorded. In addition, the uncertainty and heterogeneity of heavy metals in paddy soils are characterized by the accumulation effect of sediment in the irrigation canal and spatial variation in the metals. We examine heavy metal uptake by rice in terms of extractability and bioavailability, accumulation in different cultivars, and predictions of model soil–plant systems. Three different soil remediation techniques with environmental compatibility were then carried out for sustainable use in agriculture, including soil turnover and dilution, chemical stabilization and phytoremediation. In addition, non-anthropogenic contamination with heavy metals in the paddy soils from parent materials were examined to ensure that rice uptake conforms to food safety regulations. Finally future research needs associated with a schematic summary of heavy metals in identification, transport and remediation processes are recommended.

Anthropogenic sources of heavy metals in paddy soils

Industrial and agricultural activities and waste production

Industrial and agricultural activities generate waste materials, and the amount of these materials tends to increase as the demand for quality of life increases. However, waste materials are the most visible environmental problem globally. Increasing population, changing consumption patterns, economic development, urbanization and industrialization result in the increased generation of solid waste and wastewater. Taiwan is a humid, tropical and highly industrialized country. The impact of disposed waste in Taiwan is composed of: (1) soil contamination through direct solid waste contact or leachate and wastewater discharge, (2) contamination of surface and ground waters, (3) air pollution through the burning of wastes, (4) spreading of diseases by different vectors such as birds, insects and rodents, and (5) odor in landfills.

Over 100,000 industrial plants are currently registered in Taiwan, and approximately one-fifth of these plants clearly produce hazardous wastewater. Taiwan’s government has obtained detailed information regarding solid waste disposal from 18,000 of these plants. Most of the plants generating wastewater and solid waste are located in over 90 industrial parks administered by the government. However, approximately 50% of these industrial parks offer a centralized system for wastewater collection and treatment. Approximately 30 wastewater treatment plants operate in the industrial parks, with a wastewater treatment capacity of 110,000 m3 day−1 from 2,103 factories. However, 2,037 factories are responsible for treating their own wastewater (158,000 m3 day−1).

Factories without registration and outside of these industrial parks were excluded from the above estimation of wastewater generation. The illegal discharge of swine wastewater has caused the accumulation of nutrients (N, P, and soluble salts) and heavy metals (Cu and Zn) in nearby water bodies. Paddy fields have also been contaminated by the above-mentioned wastewater around the irrigation canal systems. In addition, approximately 40% of the rivers in Taiwan have been contaminated by solid wastes and wastewater (Taiwan EPA 2009a).

With regard to the solid wastes from industry, most of these wastes are non-hazardous. One criteria for classifying solid waste hazardous materials was based on the results of a toxicity characteristic leaching procedure (TCLP) corresponding to heavy metal levels (US EPA 1997). It is estimated that 30 million tons of industrial solid wastes are generated in Taiwan each year. Hazardous wastes constitute up to 10% of the total. In addition to industrial solid wastes, the generation rate of municipal solid waste in Taiwan is 0.67 kg capita−1 day−1, which is higher than Malaysia, India and Vietnam (Shekdar 2009).

Heavy metal contamination in paddy soils

Although heavy metals are ubiquitous in parent materials, the major anthropogenic sources of heavy metals in the paddy soils of Taiwan are illegal wastewater discharged from industrial plants, such as chemical and electroplate plants, and pigment and swine wastewater (Chen 1992). For example, a famous event of Cd-rice occurred in Taoyuan county, northwestern Taiwan during the 1980s. Over 100 ha of paddy fields had been contaminated by Cd and Pb as a result of the illegal discharge of wastewater from a plastic-stabilizer producing plant close to the fields (Chen 1991). The household farmers found that the leaves of the rice plants were brown-dotted and affected by rice chlorosis and the whole plant died slowly before harvesting. After measurement of the soils and rice tissues, the Cd concentration was found to exceed 10 mg kg−1 in the soils and 0.5 mg kg−1 in the rice grains at the sites, respectively. This event had significant impacts on environmental regulation and food safety in Taiwan. As a result, four stages of a survey project of contaminated soils in Taiwan have been carried out by the Environmental Protection Administration of Taiwan (Taiwan EPA) since 1982. The objectives of the survey were to understand the heavy metal concentration in the rural soils of Taiwan. The major results of these stages are illustrated as follows (Taiwan EPA 2009b):

  •  Stage 1 (1983–1987): each representative survey unit is 1,600 ha. The final report was published by Taiwan EPA in 1987.
  •  Stage 2 (1987–1991): each representative survey unit is 100 or 25 ha. The results showed that approximately 790 ha of rural soils have higher concentrations of heavy metals than the heavy metal regulations announced by the Taiwan EPA in 1991 allow (As 60, Cr 16, Cd 10, Cu 100, Pb 120, Hg 20, Ni 100, Zn 80 mg kg−1). The soils were extracted by 0.1 mol L−1 HCl for the measurement of metals, except As and Hg, which were extracted by aqua regia.
  •  Stage 3 (1992–2000): survey lands with relatively high concentrations of metals in the soil were selected using the results of stage 2. Each representative survey unit was 1 ha.
  •  Stage 4 (2000–2008): this stage was conducted following the Soil and Groundwater Pollution Remediation Act (SGWPR Act) announced in 2000. The contaminated control or remedial sites were announced based on the criteria of the soil control stand (SCS) in the SGWPR Act.

During 1992–1997, the upper levels of the background concentration of the surface soil (0–15 cm) were determined using the database of heavy metal content in the rural soils of Taiwan (mg kg−1): As 18 (n = 9155), Cd 0.43 (n = 9160), Cr 2.65 (n = 9159), Cu 34 (n = 9149), Pb 18 (n = 9157), Ni 9.7 (n = 9170) and Zn 51 (n = 9135) (As and Hg are total content and the other elements are 0.1 mol L−1 HCl extractable) (Chen et al. 1998). A database of the total content of heavy metals in the rural soils of Taiwan was established in 1998. The upper levels of the background total concentration of heavy metals in the surface soil (0–15 cm) are estimated to be (mg kg−1): As 18, Cd 2, Cr 50, Cu 35, Pb 50, Ni 50 and Zn 120 (Chen et al. 2007).

The Taiwan EPA reported that areas of contaminated rural soil that were higher than the SCS, based on the results of 319 ha rural soil surveys in 2002, were 159 ha of Ni-contaminated soils, 148 ha of Cu-contaminated soils, 127 ha of Cr-contaminated soils, 113 ha of Zn-contaminated soils, 17 ha of Cd-contaminated soils, 4 ha of Pb-contaminated soils and 0.3 ha of Hg-contaminated soils. Moreover, the total area of contaminated rural soil was approximately 251 ha (Taiwan EPA 2009b).

Soil and Groundwater Pollution Remediation Act developed in Taiwan

The legislature confirmed that the SGWPR Act will fully address soil contamination and further establish a system where soil pollution sites will be divided into two categories. When the levels of soil contaminants exceed the SCS at a site, this site will be listed as a “control site”, and the competent authority will be charged with taking steps to avoid further deterioration or contamination. Control sites assessed to be a clear risk by a tiered approach will be further listed as “remediation sites”. Most remediation sites promulgated are gas stations and petroleum storage tanks. However, 88% of the total number of control sites were rural soil (paddy field) sites that has been mostly contaminated by heavy metals (Fig. 1). Petroleum products that have leaked from gas stations and large storage tanks are an additional and clear contamination type of soil in Taiwan.

Figure 1.

 Pollution control site type and number with percentage in Taiwan (verified 10 June 2009).

According to the survey of rural soils with potential contamination conducted by the Taiwan EPA in 2002, more than 300 ha of rural soils were contaminated by heavy metals. The SCS of heavy metals in the SGWPR Act are listed as follows (mg kg−1): As 60, Cd 5.0, Cr 250, Cu 200, Hg 5, Ni 200, Pb 500 and Zn 600, based on aqua regia and total digestion methods. The contaminated areas by the different metals that were higher than the SCS of rural soils were approximately 159 ha for Ni, 148 ha for Cu, 127 ha for Cr, 113 ha for Zn, 17 ha for Cd, 4 ha for Pb and 0.3 ha for Hg. Most of these sites (184 ha) were located in Changhua prefecture, central Taiwan, and were mainly contaminated by Cu, Zn, Ni and Cr (Taiwan EPA 2009b). The other seriously contaminated areas are Hsinchu 27.54 ha, Taoyuan 11.46 ha, Pingtung 6.9 ha, Taipei 1.62 ha, Miaoli 0.55 ha, Nantou 0.39 ha and Taichung 0.3 ha. According to the SGWPR Act, crops grown in the contaminated soils should be collected and burned by the government agency to avoid human health risks through the food chain. Although the heavy metal contents of the contaminated paddy soils were higher than those in the regulation of heavy metals, the effects of these total concentrations of metals on crop quality and human health require further examination.

Fate of heavy metals through the food chain

Residual metals in the sediments of irrigation canals from wastewater discharge

The total length of the irrigation canal system throughout Taiwan for paddy rice production is approximately 70,000 km (Taiwan COA 2007). Anthropogenic heavy metals in the contaminated paddy soils were inputted from wastewater discharge processes through irrigation water transportation. The paddy soils were not only accumulated by the metals, but also retained the residual metals from the wastewater by sediments in the irrigation canal system. Over the long term, an increase in the metal level in the sediments occurred through contamination of paddy soils and through metals directly transported into the rural soils from wastewater.

Chang (2009) investigated the levels of the eight hazardous elements (As, Cd, Cr, Cu, Hg, Ni, Pb and Zn) regulated in the SGWPR Act from sediments of different water bodies of contaminated irrigation canals, downstream rivers, harbors and reservoirs, and found that all metal concentrations in the sediments of irrigation canals were always much higher than those from other sediments in Taiwan. Moreover, the maximum level occurred in Changhua prefecture and was mainly attributed to the illegal discharge of wastewater from electroplating and metallic surface treatment industries over the past two decades. Although no regulation for sediment quality in Taiwan has been established, contaminated sediments in the irrigation canal have the ability to threaten paddy soils with their enriched metals from the wastewater.

Measurements of total heavy metal concentrations cannot determine precisely the bioavailability and/or toxicity of heavy metals in rural soils or sediments because of the different and complex distribution patterns of the metals among various chemical species or solid phases (Basta et al. 2005). The application of sequential extraction, as proposed by Tessier et al. (1979), outlines the possible mobility and bioavailability of heavy metals in soil and sediment environments. Such procedures provide information about the speciation of heavy metals and the origin, modes of occurrence, bioavailability, mobilization and transport of heavy metals (Alloway 1995). Only soluble, exchangeable and chelating metal species in the soils are the labile fractions available to organisms (Kabata-Pendias 1993). The total contents of the metals in the sediments sampled from the irrigation canals in Changhua prefecture in central Taiwan were Cd 52.4 mg kg−1, Cr 8530 mg kg−1, Cu 9560 mg kg−1 and Ni 4150 mg kg−1 (Chang 2009). These levels are considered to constitute highly contaminated levels compared with global soil or sediment environments.

The sequential extraction procedure by Tessier et al. (1979) was used to fractionate solid speciation of the metals to enable evaluation of the metal mobility in the sediments. The percentage distributions of Cd, Cr, Cu and Ni in the five fractions are shown in Fig. 2. The dominant amounts of Cd and Ni were associated with the exchangeable (F1), carbonate bond (F2), and Fe and Mn oxide bound (F3) fractions, which were approximately 90% of the total Cd and Ni. Most Cr and Cu were distributed in the F3 fraction and in the organic bound fraction (F4). With regard to Cr, the poor level of mobility (compared with Cd and Ni) was identified by the low amount of Cr in the exchangeable and carbonate bound fractions. It is estimated that a high affinity of organic matter to Cu results in the formation of stable complexes (Norvell 1991), so that 62% of the Cu was complexed with the organic bound fraction (Fig. 2). The residual fraction (F5), defined as metal mainly fixed in the silicate minerals, was very low in the sediments, indicating that the high potential of metal mobility in the sediments may contaminate the paddy soils. Therefore, sediments in the irrigation channel system have to be dredged and the irrigation system should be isolated from the industrial discharge system. By doing this, water and soil pollution will be lightened in the future.

Figure 2.

 Chemical fractions of heavy metals (Cd, Cr, Cu and Ni) by sequential extraction in the sediment of an irrigation canal for a paddy field in Changhua (F1, exchangeable form; F2, carbonate form; F3, Fe and Mn oxide bound form; F4, organic bound form; F5, residual form).

Spatial variation in heavy metals in paddy soils

A total of 19 paddy fields from Taoyuan and Hsinchu prefectures in northern Taiwan and from the Changhua prefecture of central Taiwan, contaminated with heavy metals by industrial wastewater discharge, have been investigated to determine spatial variation in the metals in the surface soils (0–25 cm) (Römkens et al. 2009a). According to this survey, the total metal levels varied considerably between and within fields, and the metal concentration ranged from a background level to high levels beyond the current SCS (Taiwan EPA 2009b). Not all paddy fields were equally contaminated; the survey indicated that the total Cd and Pb contents were particularly high in one field, whereas the total Ni and Zn contents in another field exceeded those of other locations. Aside from the differences between fields, a broad range in the total metal content was observed within each field. For example, the total soil Cd content decreased from 7 to below 0.2 mg kg−1 within 50 m of the irrigation entrance in a paddy field. Likewise, the total Cd contents in the other two fields, which ranged from 4.0 to 21.6 and from 6.2 to 29.4 mg kg−1, were highly variable. The distance from the main entrance of the irrigation water is crucial in relation to the metal levels observed in the soil. With regard to the spatial variation in the Cr levels in a paddy field contaminated by wastewater from nearby livestock plants in Changhua prefecture (Figs 3,4), Cr was concentrated in the surface soil (0–15 cm) and was relatively low in the subsurface soils (15–30 cm). In addition, the Cr level decreased with increasing distance from the entrance of the irrigation water from the upper right corner of the field. These survey examples indicate that soil monitoring programs should consider the heterogeneous nature of metal distribution between and within paddy fields to accurately assess the human health and agro-ecosystem risks related to metals in paddy soil near industrialized areas like those in the western alluvial plains of Taiwan (Römkens et al. 2009a).

Figure 3.

 Spatial variation in Cr (mg kg−1) in the surface soil (0–15 cm) of a paddy field in Changhua.

Figure 4.

 Spatial variation in Cr (mg kg−1) in the subsurface soil (15–30 cm) of a paddy field in Changhua.

Accumulation of heavy metals in different rice cultivars

The regulation levels of soil Cd total contents range from 1 mg kg−1 to 5 mg kg−1 throughout the world. The upper limit for background Cd contents of representative rural soil is 3 mg kg−1 in Taiwan. If the total soil Cd content is higher than 5 mg kg−1, most of the corresponding brown and polished rice is considered to be Cd-contaminated rice (Chen 1991). Liu et al. (1998) suggested that brown and polished rice is Cd-contaminated when the soil Cd concentration is higher than 2 mg kg−1 extracted by 0.1 mol L−1 HCl in the potentially contaminated soils of central Taiwan. The regulation of Cd in brown rice is 0.4 mg kg−1 in Taiwan. Kyuma (2004) and Ibaraki et al. (2009) indicated that Cd-contaminated rice in Taiwan and Japan can be found in areas with different soil properties and soil and water management, even though the total content of the Cd in the soil was <5 mg kg−1. Therefore, regulation of the total contents of Cd in Taiwan rural soils should be set at a lower level (4 mg Cd kg−1), down from 5 mg kg−1, for food safety (Hseu et al. 2007a).

In the 19 Cd-contaminated paddy fields in Taoyuan, Hsinchu and Changhua prefectures of Taiwan, different rice cultivars were cultivated in 2005 and 2006 to evaluate rice uptake of Cd (Römkens et al. 2009b). In each field, 12 rice cultivars of Indica and Japonica varieties were planted on plots with 5–9 replicates for each cultivar to account for the high spatial variability in the Cd level in soil across the paddy fields. Each plot was sampled twice during the year in May and November. The total soil Cd concentrations ranged from <0.1 mg kg−1 to almost 30 mg kg−1, covering the range from background levels to heavily Cd-contaminated soils.

To minimize variation in the soil moisture regimes on Cd uptake by rice grains, a typical water management in planting paddy rice was carried out at the same condition of the field studies for comparison of cultivars and for predictions of heavy metal uptake (Römkens et al. 2009a,b). That is, the paddy field was submerged with a layer of water 2–10 cm deep on the soil surface until the heading period of the rice; the soils were then drained until the harvest period. Römkens et al. (2009b) found that the Cd levels in the rice grains were different among the rice varieties (Table 1). Higher concentrations of Cd were accumulated in the rice grains of the Indica species. For all Indica species, the median levels of Cd in the rice grains exceeded the food quality standard (FQS) of the European Union (0.2 mg kg−1) and the FQS used by the World Health Organization, Japan and Taiwan (0.4 mg kg−1). Hence, these Indica species are not suitable for cropping on paddy fields affected by Cd-contaminated soils (He et al. 2006). Cadmium levels in rice grains of the Japonica species were lower than those from the Indica species, although the median Cd grain levels were close to or in excess of the 0.2 mg kg−1 FQS of the European Union. Although the total soil Cd levels were below 0.3 mg kg−1, a large number of rice grain samples of Indica and Japonica varieties did not meet the FQS. For Indica varieties, the percentage of samples in which the Cd levels in the rice grains exceeded the FQS of 0.2 or 0.4 mg kg−1 at a given total soil Cd concentration ranged from 11.3% for 0.4 mg kg−1 FQS to 51.1% for 0.2 mg kg−1 FQS. As a result, Indica varieties cannot be grown safely in Cd-enriched soils without a considerable risk of exceeding the European Union FQS. This finding of the low bioaccumulation potential of Cd in Japonica species compared with Indica species was confirmed in a paddy field in Taiwan by Römkens et al. (2009b).

Table 1. Rice cultivars and heavy metals (mg kg−1) in soil with aqua regia extraction and in brown rice grains with the corresponding numbers of samples (n) for each cultivar in the 19 paddy fields of northern and central Taiwan (partly adopted from Römkens et al. 2009b)
Cultivar name (family)CdCuZn
nSoilRice grainnSoilRice grainnSoilRice grain
  1. Median value (minimum and maximum value). Bold and underlined numbers indicate the Cd levels exceeding the food quality standard.

Tainung No. 70 (Japonica)2770.60 (0.13–27.8)0.21 (0.02–4.57)74128 (58.7–615)4.83 (3.23–8.13)74345 (196–1160)27.4 (13.7–37.9)
Taiken No. 8 (Japonica)2780.61 (0.09–18.8)0.23 (0.01–6.00)74133 (63.7–670)4.39 (3.57–8.27)74325 (217–1080)26.6 (19.0–37.0)
Tainung No. 72 (Japonica)2780.65 (0.11–18.2)0.19 (0.02–2.98)74128 (64.5–721)4.12 (3.39–8.07)74317 (190–968)23.7 (15.6–39.9)
Kaohsiung No. 143 (Japonica)2780.64 (0.07–17.4)0.19 (0.01–4.47)74141 (67.2–711)4.84 (3.55–8.92)74367 (213–1160)25.5 (13.4–44.7)
Taitung No. 30 (Japonica)2770.59 (0.06–23.9)0.18 (0.02–3.32)74127 (60.6–770)4.52 (3.53–9.51)74338 (157–1110)25.6 (16.2–43.3)
Tainung Sen No. 20 (Indica)2770.71 (0.08–21.2)0.43 (0.02–12.6)74135 (65.1–758)5.25 (3.51–10.0)74340 (190–893)24.4 (16.3–38.8)
Tainung No. 71 (Japonica)2780.60 (0.13–25.9)0.20 (0.02–3.71)74122 (62.3–644)4.58 (3.01–9.44)74295 (193–1080)24.4 (16.0–36.4)
Tainung No. 67 (Japonica)2780.66 (0.08–26.6)0.19 (0.01–3.39)74145 (58.2–759)4.52 (3.07–9.24)74314 (196–998)24.3 (16.1–35.3)
Kaohsiung Sen Yu No. 1151 (Indica)2100.57 (0.09–25.9)0.44 (0.03–7.64)74123 (65.6–598)5.43 (3.80–12.5)74322 (203–1150)24.6 (15.8–48.0)
Taichung Sen Waxy No. 1 (Indica)2780.71 (0.14–25.8)0.60 (0.04–25.3)74131 (68.5–749)5.82 (3.35–12.5)74322 (215–1190)24.3 (15.8–42.0)
Taichung Sen No. 10 (Indica)2780.70 (0.14–22.7)0.37 (0.03–29.1)74131 (66.4–842)5.32 (3.86–10.8)74314 (191–1190)25.7 (20.3–42.1)
Kaohsiung No. 144 (Japonica)2110.59 (0.13–16.8)0.16 (0.01–3.72)74144 (68.9–904)5.36 (3.77–10.8)74355 (197–995)27.5 (19.9–46.6)

Furthermore, the control level of 5 mg kg−1 in soils by aqua regia digested in the SGWPR Act in Taiwan was not suitable for assessing the suitability or food safety of soils for rice production. The Japonica varieties, which accumulated significantly lower amounts of Cd, between 17% for 0.4 mg kg−1 FQS and 30% for 0.2 mg kg−1 FQS of all grain samples grown on soils with total soil Cd concentrations below 5 mg kg−1, did not meet the FQS. For Indica varieties, these percentages increased to 41% for 0.4 mg kg−1 FQS and 73% for 0.2 mg kg−1 FQS. These results clearly stress the need to develop alternative soil testing methods or soil–plant uptake models to predict Cd uptake by rice grains to identify soils in which rice can be grown safely with a low risk of the Cd levels in the rice grains exceeding the relevant FQS and in agreement with the conclusion of Hseu et al. (2007a).

Only Cd 0.5, Pb 0.2 and Hg 0.05 mg kg−1 were announced by the Taiwan Department of Health for the FQS in Taiwan, but Cu and Zn uptake by brown rice in the 19 paddy fields has also been demonstrated (Römkens et al. 2009b). The median Cu level value ranged from 122 to 145 mg kg−1 with a maximum value higher than 900 mg kg−1 in all survey soils, but the Cu content in brown rice was always lower than 6.00 mg kg−1 (Table 1). Like Cd, Cu was accumulated more in the Indica species than in the Japonica species, with less difference between the rice varieties compared with Cd uptake. The regulation levels of soil Cu total contents generally range from 100 to 1,500 mg kg−1 throughout the world. The soil total concentration of Cu ranged from 20 to 320 mg kg−1 in Taiwan; the upper limit for background Cu content in soil is 35 mg kg−1. The Cu content of polished rice ranged from 4 to 8 mg kg−1, but did not increase with increasing soil Cu content. However, the rice yield was significantly reduced by 30% in soils with a total Cu level up to 320 mg kg−1, and by 50% in soils with a total Cu content of 600 mg kg−1. The Cu content in polished rice was always lower than 17 mg kg−1 under the above reductions in rice yields (Liu et al. 1998).

Zinc concentrations are much higher than the concentrations of Cd and Cu metals not only in the contaminated paddy soils, but also in the brown rice grains (Table 1). The highest median value of Zn was 367 mg kg−1 in soils for the Kaohsiung No. 143 of Japonica variety, and the maximum Zn value was 1,190 mg kg−1 in the soils for the Taichung Sen No. 10 of Indica variety and Kaohsiung No. 144 of Japonica variety. However, no clear differences can be found in Zn uptake by rice grains and uptake ranged from 23.7 to 27.5−1 mg kg between all rice varieties in Taiwan. In addition, Liu et al. (1998) reported that the Zn content in polished rice from contaminated paddy soils with a total Zn concentration of 60–960 mg kg−1 in Changhua prefecture of central Taiwan, ranged from 20 to 80 mg kg−1. The regulation levels of soil Zn total contents generally range from 200 to 3,000 mg kg−1 throughout the world. The upper limit of background Zn contents of representative rural soil is 120 mg kg−1 in Taiwan. Rice yield was estimated to be reduced by 30% and the Zn level in polished rice ranged from 50 to 80 mg kg−1, corresponding to a total soil Zn content up to 500 mg kg−1. Rice yield was further reduced by 50% when the soil total Zn content rose to 800 mg kg−1, but the Zn level still remained lower than 30 mg kg−1 in the polished rice (Liu et al. 1998).

Dynamics of heavy metals in soil–plant systems

Extractability of heavy metals in contaminated paddy soils

Total metal pool is effectively estimated how much of metals was entered the contaminated site by comparing the total background level of metals. The soil regulation levels enacted in many countries including Taiwan use total metal contents rather than bioavailable or extractable metal levels to control the amounts of metals entering soil through anthropogenic activities. However, the total metal pool cannot represent the bioavailability of the metals in soil the system and fails to estimate the risks of heavy metal contaminated soil on human health and environmental quality. The relative metal pool represents metals absorbed by the surfaces of clay minerals, soil organic matter and Fe/Al oxides, and affects the metal concentration in soil solutions (Tipping et al. 2003; Weng et al. 2001). Previous field studies in Taiwan have shown that various extraction methods used to determine the reactive metal pool in Taiwan (0.1 mol L−1 HCl), the European Union (0.43 mol L−1 HCl) and in Australia (0.05 mol L−1 ethylenediaminetetraacetic acid [EDTA]) give comparable results for soil samples with variable metal levels, although the extraction ability of the extractants decreases in the order HNO3 > HCl > EDTA (Römkens et al. 2009a).

In the 19 paddy field study by Römkens et al. (2009a), the metal pools in the soils were divided into total, reactive and directly available according to their availability. Microwave digestion using aqua regia was used to determine the total metal pool. The reactive metal pool was estimated by different extraction methods using 0.43 mol L−1 HNO3, 0.1 mol L−1 HCl or 0.05 mol L−1 EDTA as the extractants. The directly available metal pool was measured using the 0.01 mol L−1 CaCl2 extraction method. Samples of surface soil (0–25 cm) when the rice was fully mature were collected together at the same location in the studied fields in May and November of 2005 and 2006. The 12 rice cultivars of the Indica and Japonica varieties are shown in Table 1. The total numbers of soil and rice variety samples in this study were both 3,198. The results indicated that the size of the reactive metal pool relative to the total metal pool in the studied paddy soils increased as follows: Cr << Ni ∼ Zn < Pb < Cu < Cd. This suggested that Cd in the paddy soils was highly reactive, whereas Cr may be present in immobile chemical forms. The reason why the reactive Pb pool relative to the total Pb pool was higher than that of Ni and Zn was unclear because Pb is also believed to be a rather immobile metal (Tipping et al. 2006). The directly available metal pool can be associated with metal uptake by plants (Peijnenburg et al. 2007). The results from field studies in Taiwan show that the ratio of directly available metal pool/reactive metal pool reached ∼10% for Cd, Zn and Ni on average; however, it only reached ∼0.3% for Cu, Pb and Cr (Römkens et al. 2009a). This suggests that Cd, Zn and Ni in the paddy soils may be more bioavailable than Cu, Pb and Cr. Therefore, the effects of Cd, Zn and Ni contaminated soils on rice growth and food safety are of more concern in Taiwan. Water management has been recommended as a way of reducing Cd uptake by rice in Japan because flooding of the contaminated paddy fields for approximate 3 weeks before and after heading of rice significantly decreases Cd in the rice grain (Ibaraki et al. 2009; Kyuma 2004). It is not easy for Taiwanese farmers in the technology transfer of controlling the submerged time due to the difficulty on operation the heavy harvesting machinery.

Cadmium translocation from soil to rice grains

Of the eight heavy metals in the SCS in Taiwan, Cd is of major concern because of its high capacity for bioavailability and accumulation in rice at levels showing no phytotoxic syndromes, but that are toxic to humans and animals through the food chain. Copper and Zn are essential elements to crops as micronutrients and pose little risk to human health because phytotoxic syndromes appear before human consumption. Arsenic, Cr, Hg and Pb are strongly adsorbed by soil colloids and are not readily taken up by plants or translocated to their edible parts (Chaney 1980). Based on these characteristics of heavy metals in soil–plant systems, most studies concerning the food safety of crops produced from metal-contaminated soils should focus on Cd.

Even though the consumption of wheat products has increased in recent years, rice still dominates the daily intake of cereals in Asian countries. Approximately half of the arable land is currently used as paddy rice growing fields in Taiwan and two rice varieties are cultivated, including Indica and Japonica varieties, but the latter is the major one (approximately 90%) because of taste preferences. Many previous field surveys have shown that Cd-contaminated rice is still produced from fields with total soil Cd levels lower than 5 mg kg−1. The biggest challenge for government agencies is Cd contamination of rice grains because it is the major source of Cd toxicity by dietary intake. Kyuma (2004) indicated that the content of Cd in brown rice from Japan and Taiwan ranged from 0.04 mg kg−1 to nearly 0.10 mg kg−1, and was quite high compared with Brazil, Iran, Bangladesh, Philippines, Guatemala and Dominica. The standard for the tolerance of Cd in rice in Taiwan was reduced from 0.5 to 0.4 mg kg−1 in 2007 to follow the World Health Organization and Japan (CODEX 2008). Many studies have also been subsidized by government to assess the food safety of rice cultivated in Cd-contaminated soil.

The root oxidation ability of rice significantly affects the redox potential in the rhizosphere and Cd uptake by rice (Liu et al. 2000). The Indica rice variety has a higher ability for root oxidation than the Japonica variety and there is a significant linear relationship between root oxidation ability and Cd concentration in the rice plants (Liu et al. 2006). In addition, the ability of rice roots to excrete low molecular weight organic acids may also determine the amounts of Cd taken up by rice plants (Liu et al. 2007b). Field studies in Taiwan have found that Cd levels in rice roots do not differ significantly between Indica and Japonica varieties (Römkens et al. 2009b). The mean Cd concentrations in the roots of Indica and Japonica varieties were 30.4 and 30.6 mg kg−1, respectively, but the ratio of Cd in the rice grains to that in the root differed substantially between the Indica and Japonica varieties. The ratio ranged from 0.061 to 0.096 for Indica varieties and from 0.025 to 0.037 for Japonica varieties. Liu et al. (2003) also found that Cd levels of Indica and Japonica varieties did not differ significantly in rice roots at the ripening stage, but that Cd levels of the two varieties did differ significantly in rice grains. These findings suggest that the major difference between Indica and Japonica varieties is their ability to transfer Cd from the root to the rice grain. Liu et al. (2007b) further indicated that Cd levels in rice grains are controlled by the translocation of Cd from the shoot to the grain part. Ueno et al. (2009) collected 71 Indica varieties and 75 Japonica varieties in their study and reported that the major sets of genes detected on chromosome 11 played a key role in the specific translocation of Cd from rice roots to shoots.

Based on a database collected from field studies of Cd-contaminated soils in Taiwan, a well-performed regression equation was developed to predict Cd-root levels by soil available Cd and Zn concentrations: log[Cd–root] = 2.31 + 0.88 log[Cd–CaCl2] – 0.38 log[Zn–CaCl2] (r2 = 0.85, P < 0.01) (Römkens et al. 2009b). The negative coefficient of [Zn–CaCl2] indicated that soil available Zn suppressed the uptake of Cd into rice roots in all varieties because Zn can compete with Cd for sorption sites on the root surface (Liu et al. 2007b).

Cadmium concentration in rice grains

Field studies in Taiwan have demonstrated that Cd concentrations in rice grains differ among cultivars even when they are planted in soils with comparable soil properties and total soil Cd levels (Römkens et al. 2009b). Approximately 27% of the studied field area was defined as Cd-contaminated soil according to the SCS of Cd. Overall, median Cd concentrations in Indica variety rice grains are 2–3-fold higher than those of Japonica varieties irrespective of whether the rice is planted in low or high Cd-contaminated fields or in different climates in Taiwan (Fig. 5). He et al. (2006) also found that Cd accumulation in Indica variety brown rice was 1.54-fold higher than that of Japonica variety brown rice. This uptake characteristic of the rice varieties is important for selecting rice cultivars with a low Cd-accumulating ability in rice grains planted in slightly Cd-contaminated soil. Liu et al. (2007b) reported that Cd was not evenly distributed throughout the rice grain. The results of their pot experiments planting six rice cultivars (include Indica, Japonica, hybrid Indica and new plant types) in artificially Cd-contaminated soil showed that the average percentages of Cd accumulated in chaff, cortex (embryo) and polished rice were approximately 15, 40 and 45%, respectively. The cortex occupied only 9% of the dry weight of the grain on average, but polished rice occupied 71%; thus, the Cd concentration in the cortex is significantly higher than that in polished rice. These researchers suggested that polished rice produced from Cd-contaminated fields may be safer for consumers than brown rice, even though the latter has a higher nutrition value. However, Moriyama et al. (2002) reported that the Cd concentration in six Japonica rice varieties dropped only 3% on average after milling. A study using in situ synchrotron X-ray fluorescence to identify Cd distribution in brown rice produced from Bangladesh, China and the USA also showed that Cd is evenly distributed in brown rice (Meharg et al. 2008). These inconsistent findings may result from errors in the rice polishing process or from inherent differences in the Cd distribution in rice grains among rice varieties. More careful studies are required to clarify these inconsistent results.

Figure 5.

Relationship between the Cd concentration in two varieties of brown rice and the CaCl2-extractable Cd concentration in the soil in (a) low Cd-contaminated fields and (b) high Cd-contaminated fields harvested in May and (c) low Cd-contaminated fields and (d) high Cd-contaminated fields harvested in November.

Field studies in Taiwan have also shown that the percentages of rice grain samples exceeding the FQS in Taiwan (0.4 mg kg−1) increased as the soil total Cd level increased. At each total soil Cd level, the percentages of Indica rice variety grain samples with Cd levels exceeding the FQS were 2–3-fold higher than the percentages of Japonica varieties. More than 40% of Indica rice grain samples cannot meet the FQS, even when the soil total Cd concentration is less than 5 mg kg−1, the SCS enacted by the Taiwan EPA. At a soil total Cd concentration above 5 mg kg−1, more than 50% of Indica rice grain samples will exceed the FQS. It is clear that Indica varieties can accumulate more Cd in their rice grains than Japonica species, and easily exceed the FQS if planted in Cd-contaminated soil. Previous studies have shown similar trends (Arao and Ae 2003; He et al. 2006; Liu et al. 2007a; Morishita et al. 1987). For any Cd-contaminated rice-growing fields, Indica rice cultivars are not recommended for cultivation because of concerns regarding food safety.

Predictions of heavy metal uptake by rice using soil–plant transfer models

Previous studies have shown that the total soil Cd concentration cannot be used reliably to determine whether rice grains are safe for consumers (Adriano 2001). Rice varieties and soil characteristics, such as soil pH, redox potential, cation exchange capacity (CEC), texture and soil organic matter, are important factors affecting the concentration of Cd in rice grains. To determine whether a rice-growing field can produce safe rice grains with Cd levels below the FQS, it is necessary to develop a simple and reliable soil test to predict the Cd concentration in rice grains.

Field studies in Taiwan have compared different extraction methods (0.01 mol L−1 CaCl2, 0.1 mol L−1 HCl, 0.43 mol L−1 HNO3 and 0.05 mol L−1 EDTA) to assess which method is the best for predicting Cd levels in rice grains under typical water management plans in paddy rice in Taiwan (Römkens et al. 2009b). The best regression equations to predict Cd levels in rice grains of Indica and Japonica varieties were developed using soil-available Cd and Zn concentrations determined by 0.01 mol L−1 CaCl2, as shown in Table 2. The CaCl2-extractable Zn is also included in the equation because it is able to compete with Cd for plant uptake and reduce the toxic effects of Cd (Choudhary et al. 1995; Hassan et al. 2006; Liu et al. 2007a).

Table 2. Soil–plant transfer models developed to predict the Cd concentration in brown rice grains using soil analysis data
Soil–plant transfer modelReference
  1. Japonica variety. Indica variety. CEC, cation exchange capacity.

log[Cd–rice] = 0.94 + 0.78 log[Cd–CaCl2] − 0.30 log[Zn–CaCl2], R2 = 0.73Römkens et al. (2009b)
log[Cd–rice] = 0.60 + 0.82 log[Cd–CaCl2] − 0.28 log[Zn–CaCl2], R2 = 0.86Römkens et al. (2009b)
log[Cd–rice] = 1.20 + 0.76 log[Cd–NaNO3]  − 0.17 pH − 0.32 log[CEC], R2 = 0.74Römkens et al. (2009b)
log[Cd–rice] = 0.97 + 0.74 log[Cd–NaNO3] − 0.18 pH − 0.43 log[CEC], R2 = 0.81Römkens et al. (2009b)
Cd–rice = −14.129 + 2.044 pH + 1.393 log[Cd–CaCl2], R2 = 0.638Simmons et al. (2008)
log[Cd–rice] = 0.174 + 0.491 log[Cd–CaCl2], R2 = 0.281Brus et al. (2009)
log[Cd–rice] = 5.023 + 1.036 log[Cd–NaNO3] − 0.437 pH − 1.735 log[clay] − 1.084 log[SOM], R2 = 0.661Brus et al. (2009)

For all rice cultivars tested in the studies in Taiwan, the regression equations accounted for more than 70% of the variability in rice grain Cd. These regression equations may be used in Taiwan as a screening tool to predict Cd concentrations in rice grains before rice cultivation. As shown in Table 3, critical concentrations of CaCl2-extractable Cd in soils under different levels of soil CaCl2-extractable Zn are constructed for farmers and authorities in Taiwan to avoid the production of Cd-contaminated rice using the regression equations in Table 2. The concentration of CaCl2-extractable Zn in the soil generally ranged from 0.1 to 50 mg kg−1 when the total soil Zn concentration was less than 600 mg kg−1, the SCS enacted by the Taiwan EPA. According to the regression equations, less Cd will be accumulated in rice grains if the soil CaCl2-extractable Zn is higher. Therefore, only the critical concentrations of CaCl2-extractable Cd in soil under soil CaCl2-extractable Zn levels lower than 50 mg kg−1 are presented in Table 3. If the measured soil CaCl2-extractable Cd is higher than the critical value, it is possible to produce rice grains with Cd concentrations exceeding the FQS (0.4 mg kg−1). Further studies need to be conducted to validate the practicality of these regression equations.

Table 3. Critical concentrations of CaCl2-extractable Cd in soils under different levels of soil CaCl2-extractable Zn in two rice varieties
Rice varietyCaCl2-extractable Zn in soil (mg kg−1)
  1. Cadmium concentrations in the rice grains will exceed 0.4 mg kg−1 if the measured soil CaCl2-extractable Cd is higher than the critical concentration.


To predict the Cd concentration in rice grains, Simmons et al. (2008) developed a regression equation using soil pH (1:5) and CaCl2-extractable Cd determined from field-moist samples collected during the grain-filling period (Table 2). This equation can predict Cd concentrations in unpolished rice grains with an r2 value of 0.638. If air-dried soil samples are used for Cd–CaCl2 and pH determination, the regression equation cannot explain variability in the Cd levels in the rice grains. Air-drying may affect soil sample conditions to the extent that CaCl2-extractable Cd cannot represent the Cd availability in the soil compared with extracts collected from field-moist soil. However, the soil samples used to develop the regression equations in the Taiwanese study were air-dried and collected during the rice harvest period, an easier pretreatment for soil samples that is more suitable for routine monitoring.

Brus et al. (2009) recently developed a multiple regression model using 0.43 mol L−1 HNO3-extractable Cd, pH, clay and soil organic matter to predict Cd levels in rice grains from Fuyang, Zhejiang, China (Table 2). This model performed much better (r2adj = 0.661) than the linear model using only 0.01 mol L−1 CaCl2-extractable Cd as a predictor (r2adj = 0.281). The field study in Taiwan also developed a multiple regression model using 0.43 mol L−1 HNO3-extractable Cd, pH and CEC to predict Cd levels in rice grains (Table 2). Although the model used more factors to reflect the effects of pH and CEC on the bioavailability of Cd, it did not perform much better (r2 = 0.81 and 0.74 for Japonica and Indica varieties, respectively) than the model using 0.01 mol L−1 CaCl2-extractable Cd and Zn as predictors (r2 = 0.86 and 0.73 for Japonica and Indica varieties, respectively). Therefore, the latter simpler model is preferred for validation and use in Taiwan. As different environmental and soil factors affect the accumulation of Cd in rice grains in different ways and to differing extents, prediction models developed using local data will be more reliable for use in that specific area.

Remediation techniques for paddy soils

A number of soil remediation techniques for metal-contaminated soils have been developed, but some of these techniques are not efficient in terms of time, cost and environmental compatibility (Dermont et al. 2008). For agro-environmental sustainability, three remediation techniques are illustrated in this review with regard to the remediation experiences in the contaminated paddy soils of Taiwan: soil turnover and dilution, in situ stabilization by chemical amendments and phytoremediation in the field.

Soil turnover and dilution technique

Soil attenuation is an alternative technology for returning heavy metal contaminated soils to their original function in agriculture. This method adequately mixes contaminated surface soils (0–30 cm) and clean subsurface soils (30–100 cm) in situ using mechanical forces to reduce the heavy metal concentrations through the soil pedon from the surface to a depth of 1 m. This technology process has the advantage of low cost in terms of time and budget, rapid reduction in the total concentration of heavy metals, and minimizing effects to crop production after treatment compared with other soil remediation techniques, such as acid washing and chelating agent extraction techniques. Cropping must cease in all agricultural soils that have been identified as contaminated according to the SCS. All currently remediated sites in Taiwan are fallow and dry. Nevertheless, the soil turnover was proposed to perform in dry and fallow seasons of cropping rice during October to next March for avoiding clear changes of moisture from subsurface to surface soil. However, this technology is restricted to relatively low levels of heavy metal soil concentrations at rural land sites and geology settings (i.e. water table, stone proportion) in Taiwan.

The turnover and dilution method has been the most popular remediation strategy for most metal-contaminated paddy soils in Taiwan since 2000. This technical process has, however, reduced the levels of organic carbon and other nutrients in the cultivated layers of paddy soils. Consequently, compost and mineral fertilizer need to be applied to the treated soils to recover soil fertility in crop growth and soil function. In addition, plow pans (approximately 30–50 cm depth) in the paddy pedons were simultaneously destroyed by the attenuation practices and thus re-building of the plow pans was requested by the land owners to allow cultivation of paddy rice.

The spatial distribution of heavy metals has shown that metal concentrations tend to be high near the entrance of irrigation water within individual fields, and that the concentrations gradually decrease with increasing distance away from the entrance. Therefore, Liu et al. (2008) applied the soil with relatively low concentration of Zn considered as the source of attenuation by horizontal direction for a field-scale remediation practice of paddy soil with Zn up to 4000 mg kg−1 in Changhua, central Taiwan. Furthermore, the amount of Zn was concentrated within a depth of 60 cm in the paddy soil, but decreased with increasing soil depth. Soil below 60 cm was considered to be a source of attenuation in a vertical direction. Deep plowing and consequently mixing the two layers well can significantly decrease the Zn level to meet the SCS of Zn (600 mg kg−1) in Taiwan (Fig. 6). However, the potential drawback to this is that the total mass of Zn is consistent throughout the soil pedon before and after treatment; thus, the bioavailability of the metal is needed to evaluate crop production.

Figure 6.

 Mean (± standard deviation) Zn concentration in the contaminated paddy soils at different depths before and after the turnover treatment.

Chen and Lee (1997) used the soil turnover and dilution method to effectively decrease Cd and Pb concentrations in Taoyuan prefecture, northern Taiwan. Soils in the surface layer (0–20 cm) were mixed with the subsoil (20–40 cm) and then planted with lettuce, water celery and Chinese cabbage. These crops were harvested after 4 weeks. Even though mixing of the surface and subsurface soils decreased the total soil Cd and Pb concentrations, the mixed soils were still higher than 2.10 mg Cd kg−1 and 9.07 mg Pb kg−1 with 0.1 mol L−1 HCl extraction because the initial metal concentrations were not significantly different in the surface 60 cm. The soil turnover and dilution method can decrease Cd and Pb concentrations in crops (Table 4), but the total concentrations of these metals are still too high for rice production (Chen 1991).

Table 4. Accumulated Cd and Pb concentrations (mg kg−1) in different crops after planting in the turnover and dilution method treated soil for 4 weeks (Chen and Lee 1997)
TreatmentWater celeryLettuceChinese cabbage
After dilution18.42.7739.63.4434.06.94

In situ stabilization by chemical amendments

Current technologies for soil remediation are time consuming and costly. Therefore, it is imperative to develop techniques that can treat and stabilize heavy metals in situ in an efficient and cost-effective manner (Wenzel et al. 1999). Chemical stabilization is one of the most cost-effective soil remediation techniques for heavy metal contaminated sites (Chen et al. 2000). Sorption, ion exchange and precipitation are the principle mechanisms used to convert soluble and pre-existing potentially soluble solid phase forms of heavy metals to more biogeochemical stable solid phases, reducing the heavy metal pool for root uptake in soil by natural occurring or artificial additives such as lime material, phosphate, zeolite, bentonite, clay, hydrous iron and manganese oxides, and organic matter (Chen et al. 2000; Cheng and Hseu 2002; Mench et al. 1994; Phillips 1998; Rebedea and Lepp 1994; Shuman 1999). Therefore, it is valuable to understand the levels and mechanisms of control on heavy metal solubility by different soil additives for remediation practices.

The effects of recommended amendments for soil remediation are often attributed to an increase in the CEC, the addition of materials with a strong preference for particular ions, and precipitation as a solid phase (Czupyrna et al. 1989; Phillips 1998). For example, hydrous ferric oxide has a high surface area and CEC and plays a significant role in the control of solution levels of heavy metals such as Cd, Cu, Co, Pb and Zn (Chubin and Street 1981; Davis and Leckie 1978; Swallow et al. 1980). Because of the specific retention of phosphate, the zero point of charge of ferric oxide incorporated with phosphate shifts downward to increase the adsorption capacity for heavy metals (Sposito 1981). Yuan and Lavkulich (1997) also reported that the adsorption capacity of a soil for Zn was reduced by 72% when 11% of the organic C content was lost.

Two paddy soils (A and B) contaminated by Cd and Pb from industrial wastewater in northern Taiwan were used to evaluate the effect of different chemical treatments on changes in speciation and extractability of Cd and Pb and in phytoavailability to wheat (Triticum aestivum) by Chen et al. (2000). Six amendments including calcium carbonate, calcium phosphate, hog composts, iron oxide, manganese oxide and zeolite, were used to immobilize Cd and Pb in the contaminated soils using pot experiments. Wheat was planted for a further 1 month to evaluate the effectiveness of the treatments on the uptake of Cd and Pb by wheat shoots. The results indicated that the addition of calcium carbonate, manganese oxide and zeolite considerably reduced the extractability of Cd or Pb in both soils (Tables 5,6), and significantly reduced the uptake of Cd and Pb by wheat shoots (Table 7). Changes in the extractability and metal sequential fractionations indicate that the exchangeable (or available) form of Cd and Pb in the two soils can be transformed into unavailable forms after these amendments.

Table 5. Cadmium concentrations (mg kg−1) using different extractions in two paddy soils spiked with various amendments from northern Taiwan (Chen et al. 2000)
TreatmentCadmium concentrations with different extractions
  1. Different letters in a column indicate significant differences between treatments at P = 0.10 using Duncan’s multiple range tests. EDTA, ethylenediaminetetraacetic acid; HOAc, acetic acid.

Soil A
 Control4.31 a4.09 a3.75 a4.56 a
 Iron oxide4.11 b4.09 a3.13 bc4.45 ab
 Manganese oxide4.11 b3.86 ab2.92 c4.23 bc
 Calcium carbonate3.95 b3.42 c3.13 bc4.17 c
 Calcium phosphate3.34 b3.98 ab3.34 b4.34 abc
 Hog compost3.13 b3.87 ab3.13 bc4.56 a
 Zeolite4.05 b3.76 b2.92 c4.17 c
Soil B
 Control13.7 a15.1 a17.0 a15.3 a
 Iron oxide13.7 a15.1 a16.0 bc15.3 a
 Manganese oxide13.7 a14.0 ab15.3 c15.8 a
 Calcium carbonate13.7 a11.3 c16.2 c15.8 a
 Calcium phosphate13.2 a14.0 ab16.2 bc15.8 a
 Hog compost13.2 a12.4 bc14.9 c15.3 a
 Zeolite13.7 a12.9 bc17.0 ab15.8 a
Table 6. Lead concentrations (mg kg−1) using different extrac-tions in two paddy soils spiked with various amendments from northern Taiwan (Chen et al. 2000)
TreatmentLead concentrations with different extractions
  1. Different letters in a column indicate significant differences between treatments at P = 0.10 using Duncan’s multiple range tests. EDTA, ethylenediaminetetraacetic acid; HOAc, acetic acid.

Soil A
 Control14.6 a12.2 a1.27 a11.3 a
 Iron oxide12.9 b10.5 ab1.27 a11.3 a
 Manganese oxide8.59 c6.13 c1.27 a6.14 b
 Calcium carbonate12.9 b11.3 ab1.27 a10.5 a
 Calcium phosphate13.3 ab12.2 a1.27 a11.3 a
 Hog compost14.2 ab10.5 ab1.27 a11.3 a
 Zeolite12.9 b8.72 b1.27 a10.5 a
Soil B
 Control347 a415 a282 a398 a
 Iron oxide343 a371 b282 a398 a
 Manganese oxide330 b381 b235 c398 a
 Calcium carbonate343 a381 b282 a398 a
 Calcium phosphate343 a389 b275 b398 a
 Hog compost339 ab398 ab242 c398 a
 Zeolite347 a372 b275 b372 b
Table 7. Uptake of Cd and Pb (based on dry matter) in wheat shoots grown in two contaminated soils with various amendments (Chen et al. 2000)
TreatmentSoil A (mg pot−1)Soil B (mg pot−1)
  1. Different letters in a column indicate significant differences between treatments at P = 0.10 using Duncan’s multiple range tests.

 Control0.05 a0.13 a
 Iron oxide0.02 bcd0.13 ab
 Manganese oxide0.02 bcd0.10 b
 Calcium carbonate0.01 d0.10 b
 Calcium phosphate0.04 ab0.12 ab
 Hog compost0.03 abc0.12 ab
 Zeolite0.02 cd0.06 c
 Control0.14 a0.31 a
 Iron oxide0.02 b0.25 ab
 Manganese oxide0.05 b0.21 b
 Calcium carbonate0.05 b0.15 c
 Calcium phosphate0.02 b0.23 b
 Hog compost0.02 b0.18 bc
 Zeolite0.02 b0.14 c

Cheng and Hseu (2002) used different soil amendments, including 1% zeolite, 1% bentonite, 5% Penghu calcareous soil (PHS), 5% Penghu calcareous soil plus 1% manganese oxide (PHS + MO), 1% MO and 1.5% silicate slag (SS), to immobilize Cd and Pb in two contaminated paddy soils evaluated by single and sequential extractions and by uptake of Chinese cabbage (Brassica Chinensis L.) in Taiwan. The two soils (C and D) were collected from paddy fields that had been contaminated by Cd and Pb in wastewater discharged from a chemical plant in central Taiwan. Penghu calcareous soil, derived from basalt at an offshore island west from Taiwan, offers promise as an amendment to immobilize heavy metal because of its desirable chemical and mineralogical properties. The PHS is highly weathered and dominated by kaolinite. In addition, the PHS is characterized by high pH and CEC values and high carbonate and free oxide contents. The results indicated that PHS and MO significantly reduced the Cd and Pb levels with 0.1 mol L−1 HCl and 0.05 mol L−1 EDTA extractions in the two contaminated soils (P < 0.05) (Table 8). None of the amendment treatments changed the organic matter fraction and residual fractions of Cd and Pb in the soils, but the PHS and PHS + MO treatments significantly reduced the exchangeable fractions of Cd and Pb as measured by metal sequential extraction. The combination of PHS and MO amendments was associated with a high pH value and this negative soil surface charge showed the best immobilizing efficiency of Cd and Pb in the soils. Not all of the soil amendments investigated increased the dry matter weight of the plant, but most of them decreased the uptake of Cd and Pb, particularly the PHS and MO treatments. The PHS and MO treatments reduced the extractability of Cd and Pb in the two soils and their uptake by plants, but only the Pb content in Chinese cabbage from the amended soils was less than the background levels of heavy metals in leafy vegetables in Taiwan (Table 9).

Table 8. Cadmium and lead concentrations (mg kg−1) extracted by 0.1 mol L−1 HCl and 0.05 mol L−1 ethylenediaminetetra-acetic acid in two paddy soils spiked with various amendments from central Taiwan (Cheng and Hseu 2002)
0.1 mol L−1 HCl0.05 mol L−1 EDTA0.1 mol L−1 HCl0.05 mol L−1 EDTA
  1. Different letters in a column indicate significant differences between treatments at P = 0.05 using Duncan’s multiple range tests. EDTA, ethylenediaminetetraacetic acid.

Soil C
 PHS + MO12.0c2.64cNDd7.64c
Soil D
 PHS + MO1.99b0.38cd88.5c30.7b
Table 9. Uptake of Cd and Pb (mg kg−1 based on dry matter) in the shoots of Chinese cabbage grown in two contaminated soils treated with different amendments (Cheng and Hseu 2002)
  1. Different letters in a column indicate significant differences between treatments at P = 0.05 using Duncan’s multiple range tests. ND, not detectable.

Soil C
 Unamended3.20 ab0.50 a
 Zeolite3.50 a0.50 a
 Bentonite2.48 bc0.50 a
 PHS1.98 cdND
 PHS + MO1.50 dND
 MO2.48 bcND
Soil D
 Unamended15.1 a7.69 a
 Zeolite13.1 a2.19 b
 Bentonite13.9 a2.18 b
 PHS3.74 c0.80 bc
 PHS + MO3.50 c0.93 bc
 SS5.97 b0.91 bc


Phytoremediation has been shown to be a feasible method for treating contaminated paddy fields that have large areas and low to medium levels of heavy metal concentration (Lai and Chen 2005). However, most hyper-accumulator plant species used to remove heavy metals had low biomass in the shoot part and a low growth rate. Enhanced agents were applied to increase the bioavailability of the metals and also increase the amounts of heavy metal taken up by the plants. The application of EDDS (ethylene diamine disuccinic acid), HEDTA (hydroxyethyl ethylenediaminetetraacetic acid), EDTA or NTA (nitrilotriacetate) significantly increased the concentrations of various heavy metals in the shoots of different plants (Meers et al. 2005). The transfer of Cd and Pb from the root to the shoot (translocation factor [TF]) increased in the EDTA treatment (Chen et al. 2004).

However, the results of most previous studies have shown that chemical agents have a negative effect on the growth of Indian mustard, sunflower and corn and thus decrease the total removal of heavy metals by plants (Blaylock et al. 1997; Hsiao et al. 2007; Meers et al. 2004; Turgut et al. 2004; Wu et al. 2004). After the application of chemical chelating agents, there is a potential risk of groundwater contamination because solubility and leaching of the heavy metals has increased (Jiang et al. 2003; Lai and Chen 2004, 2005; Wu et al. 2004). For lands with a sandy texture or high water table, chemical agents should be carefully applied because there is a health risk associated with groundwater quality (Lai and Chen 2006; Wu et al. 2004).

Two field-scale case studies have demonstrated phytoextraction of paddy soils in Taiwan. A large-scale phytoextraction of Cd-contaminated paddy soils was conducted in northern Taiwan (Chen and Lee 1997). Eight garden flower species were planted to assess their accumulating capacity, including azalea (Rhododendron spp.), osmanthus (Osmanthus spp.), star cluster (Pentas lanceolata), scarlet sage (Salvia splendens), common cosmos (Cosmos bipinnatus), zinnia (Zinnia elegans), cockscomb (Celosia argentea) and garden verbena (Verbena tenera). Ten seedlings of each species were planted in a 2 m × 1.5 m plot with three replicates. The experimental results showed that cockscomb, star cluster and garden verbena accumulated high Cd concentrations. The increases were approximately 30–40-fold compared with the initial concentration of Cd in the plant tissues. Although a high Cd concentration (86 mg kg−1) was accumulated in cockscomb, chlorosis symptoms with brown spots on the leaves and veins were clearly observed, and probably resulted from high soil Cd toxicity (approximately 10–30 mg kg−1). The increases in Cd in common cosmos, scarlet sage and zinnia were approximately 10-fold compared with the initial concentration of Cd in the plant tissues. The color of the zinnia leaves became yellow when the Cd concentration was higher than 10 mg kg−1. However, osmanthus and azalea only absorbed a low amount of Cd, ranging from 2 to 4 mg kg−1. Table 10 shows the suitability of the garden flowers planted at the Cd-contaminated sites based on an assessment of growth condition, Cd accumulation capacity and economic benefit (Chen and Lee 1997).

Table 10. Phytoextraction test in Cd-contaminated soils in northern Taiwan (Chen and Lee 1997)
Plant speciesCd concentration in the leaves (mg kg−1)Growth conditionEconomic valueAssessment§
  1. a, most suitable; b, suitable; c, not suitable. a, high; b, medium; c, low. §aa, most suitable, bb or ba, suitable; cb, not suitable.

Celosia argentea86cbcb
Pentas lanceolata44aaaa
Verbena tenera42aaaa
Cosmos bipinnatus13bbbb
Salvia splendens12cbcb
Zinnia elegans10cbcb
Osmanthus spp.4cbcb
Rhododendron spp.3baba

A large field contaminated site in Changhua, central Taiwan, with an area of 1.3 ha and contaminated with Cr, Cu, Ni and Zn, was selected in 2006 to examine the feasibility of phytoextraction in Taiwan (Lai et al. 2009). Based on the results of a pre-experiment at this site, approximately 20,000 seedlings of 12 species were selected from 33 tested plant species to be used at this field scale site. A comparison with the initial metal concentration in 12 plant species before planting demonstrated that most species accumulated significant amounts of Cr, Cu, Ni and Zn in their shoots after growing in this mixed-metal contaminated site for 31 days. Among the 12 plant species, the following accumulated higher concentrations of metals in their shoots: Garden canna and Garden verbena (45–60 mg Cr kg−1), Chinese ixora and Kalanchoe (30 mg Cu kg−1), Rainbow pink and sunflower (30 mg Ni kg−1), and French marigold and sunflower (300–470 mg Zn kg−1). The roots of the plants of most of the 12 species can accumulate higher concentrations of heavy metals than the shoots. Extending the growth time from 1 month to 2 months promotes the accumulation of heavy metals. These plant species can accumulate higher concentrations of each heavy metal, except Zn, in their roots than in their shoots.

The phytoavailability of soil metals declines with time during phytoextraction, increasing the period required for effective remediation. Under ideal conditions and based on the above experimental results, it takes 3–20 years to reduce the current concentrations of Cr, Cu, Ni and Zn to the regulation levels based on the SGWPR Act of Taiwan. The results of this large-area field scale experiment also show that planting suitable and high-potential phytoextraction plants in heavy metal contaminated soil can enable the contaminated sites to recover to their natural condition and generate economic benefit because the plants can be sold.

Rice with high level metals

In addition to anthropogenic sources of heavy metals in paddy soils, geochemical origin is another source of heavy metals/metalloids, such as As, Cu and Ni. No metal mining activities exist in Taiwan, but some parent rocks of the paddy soils are enriched with these metals that are further concentrated in the soils. For example, As concentration is high in soils from andesite, and Cr and Ni are high in soils from serpentine group minerals (ultramafic rocks). The levels of these endogenous heavy metals have exceeded the SCSs (Cheng et al. 2009; Hseu 2006; Hseu et al. 2007b; Su and Chen 2008). The occurrence of large amounts of these metals could considerably increase potential risks to crop safety and agro-environmental quality. In this review, we examined the Guandu plain in northern Taiwan and the alluvial valley of the Coastal Range in eastern Taiwan to explore the enrichment of As, Cr and Ni in the paddy soils and their uptake by rice.

Arsenic in paddy soils from andesite at Guandu plain

Arsenic levels in the paddy soils on the Guandu plain varied widely from 12.4 to 535 mg kg−1. Although As concentrations in the soils were much higher than the SCS of the SGWPR Act in Taiwan (60 mg As kg−1), the corresponding brown rice only accumulated <0.35 mg As kg−1 (dried-weight basis) and no adverse effect was observed during rice growth. The standard for the tolerance of heavy metals in rice excluded As in Taiwan. According to the statutory limits of As concentration in cereals or food crops constructed in different countries (Table 11), rice from the As-enriched soils of the Guandu plain was still safe for consumers after a health risk assessment.

Table 11. Statutory limits of arsenic concentration in cereals or food corps announced by the World Health Organization and various different countries
Country/instituteRegulation itemStatutory limitReference
  1. Limit for inorganic As. DW, dry weight; FW, fresh weight.

AustraliaCereals1 mg kg−1 FWMcLaughlin et al. 2000
CanadaFood crops1 mg kg−1 FWZandstra and De Kryger 2007
ChinaRice0.15 mg kg−1 DWMHPRC 2005
New ZealandCereals1 mg kg−1 FWMcLaughlin et al. 2000
SwitzerlandFood crops4 mg kg−1 DWGulz et al. 2005
United KingdomFood in sale1 mg kg−1 FWWarren et al. 2003

The global normal range of As concentration in rice is 0.08–0.20 mg kg−1, according to the database (n = 411) (Zavala and Duxbury 2008). These researchers also found that As levels in rice produced from Asia (lower than 0.098 mg kg−1) were significantly lower than those from the USA or the European Union. Compared with the findings of Zavala and Duxbury (2008), As levels in rice grains from the Guandu Plain are clearly higher than the normal global range, even though they did not exceed the statutory limits. However, a pot experiment using paddy rice also showed that the As concentration in brown rice ranged from 0.1 mg kg−1 to as high as 0.4 mg kg−1, even when the rice was cultivated in soils contaminated with only <25 mg kg−1 As (Lin et al. 1999). Therefore, it is necessary to undertake a full survey of the As taken up by the different rice varieties produced in Taiwan to identify normal levels of As in rice and to allow for further comparisons with data from Guandu plain near Taipei city.

Chromium and nickel in paddy soils from serpentinites in eastern Taiwan

Serpentine soils are not derived from ultramafics, but rather from soils with hydrothermal alternation of ultramafic minerals and serpentine minerals (McGahan et al. 2008). However, serpentine soils are often of ecological or environmental importance owing to high levels of heavy metals, such as Cr and Ni (Cheng et al. 2009; Hseu 2006; Oze et al. 2004a,b). Throughout the world most complex ophiolite sites related to serpentine soils are close to convergent boundaries of tectonic plates. Serpentinities and serpentine soils are generally present in areas within the Circum-Pacific margin, such as Japan, Taiwan and the Philippines; paddy rice production mainly occurs in these regions.

For serpentinitic parent materials, the release of Cr and Ni into the agro-ecosystems and particularly into natural water systems during mineral weathering may occur (Becquer et al. 2006; Oze et al. 2008). The extractability of Cr and Ni in serpentine soils varies from site to site and indicates that the metals come from a source of non-anthropogenic contamination. Fernandez et al. (1999) found that the foliage of agricultural crops such as sugar beet, cabbage and pasture grown on serpentine soils in north-western Spain can accumulate significant quantities of Cr and Ni, despite low to moderate EDTA-extractable amounts in these soils.

The Cr and Ni contents in serpentine soils are much higher than the levels recorded in soils formed from other parent materials, and vary considerably between sites and regions (Table 12), probably reflecting different degrees of chemical weathering in these serpentinitic rocks (Bonifacio et al. 1997). In all rocks of the Earth, mafic and ultramafic rocks are richest in Cr, containing up to 3,400 mg kg−1 of Cr along with 3,600 mg kg−1; however, the average concentrations of Cr and Ni in world soils are approximately 84 and 34 mg kg−1, respectively (Alloway 1995). A Cr concentration of 19,000 mg kg−1 has been found in serpentine soils of New Caledonia (Becquer et al. 2006). Some serpentine soils have measured Ni concentrations higher than 10,000 mg kg−1 (Oze et al. 2004a). However, Cr and Ni in serpentine soils are primarily associated with the residual compartments (>90%) measured by metal sequential extraction, indicating that the original Cr and Ni were primarily fixed in the primary minerals (e.g. spinel groups and silicates) and did not really extend into the organic and Fe/Mn-bound fractions of the soils (Hseu 2006).

Table 12. Concentrations (g kg−1) of global Cr and Ni in serpentine soils
Costal Range, Taiwan0.4–3.30.4–5.8Hseu (2006) and Cheng et al. (2009)
Hokkaido, Japan5.22.6Suzuki et al. (1971)
Samar, Philippines<0.1–1.6<0.1–0.96Navarrete et al. (2007)
Maryland, USA0.1–6.00.1–20Rabenhorst et al. (1982)
Oregon, USA<0.1–4.2<0.1–4.5Burt et al. (2001)
Southwestern Oregon, USA0.6–2.31.1–3.4Alexander et al. (2007)
California, USA2.0–3.51.5–2.5Lee et al. (2004)
California, USA<0.1–2.1<0.1–3.9Oze et al. (2008)
British Columbia, Canada0.8–5.50.2–1.7Bulmer and Lavkulich (1994)
Santa Elena, Costa Rica1.4–3.63.2–7.2Reeves et al. (2007)
Niquelândia, Brazil5.1–>172.3–6.0Garnier et al. (2006)
Savona, Italy1.0–8.30.6–5.1Bonifacio et al. (1997)
Aosta Valley, Italy0.9–2.80.3–1.1D’Amico et al. (2008)
Trás-os-Montes, Portugal0.2–4.3<0.1–1.6Díez et al. (2006)
Szklary Massif, Poland2.7–4.21.3–4.1Kierczak et al. (2007)
France0.1–3.90.2–3.6Massoura et al. (2006)
New Caledonia17–197.0–8.5Becquer et al. (2006)
New Zealand7.96.1Robinson et al. (1999)
Transvaal, South Africa0.5–2.62.0–5.0Noble and Hughes (1991)

With regard to food safety and the uptake of Cr and Ni by rice grown on serpentine soils, Wu (2009) used a Vertisol from eastern Taiwan to plant paddy rice amended with chemical fertilizer and oilseed compost in a pot experiment. The pot treatments in the open field were: unfertilized soil treatment (control), full recommended dose of chemical fertilizer (CF), half recommended dose of chemical fertilizer + 2.5 ton ha−1 oilseed compost (1/2 CF + 2.5 OC), 5.0 ton ha−1 oilseed compost (5.0 OC), and 10 ton ha−1 oilseed compost (10 OC). The study examined the uptake of Cr and Ni by the plant tissues of rice and revealed that the concentrations of Ni in the roots, stalks and brown rice grains were always higher than those of Cr in all treatments (Table 13). A significant increase in the metal concentrations of brown rice grains in the fertilizer treatments was observed, particularly when the amount of compost applied was increased. However, these Cr and Ni levels can be considered to be within the normal range when compared with the concentrations of Cr and Ni in the database (n = 120) of brown rice and polished rice established by Liu et al. (1998) for uncontaminated paddy soils in central Taiwan. Nevertheless, quality issues for other crops and Cr oxidation potential in the serpentine ecosystem need to be examined in future studies.

Table 13. Chromium and nickel concentrations (mg kg−1) in the plant tissues of paddy rice grown on a Vertisol from serpentine in eastern Taiwan (Wu 2009)
TreatmentRootStalkBrown rice grains
  1. Different letters in a column indicate significant differences between treatments at P = 0.05 using Duncan’s multiple range tests. Control, unfertilized soil treatment; CF, full recommended dose of chemical fertilizer; 1/2 CF + 2.5 OC, half recommended dose of chemical fertilizer + 2.5 ton ha−1 oilseed compost; 5.0 OC, 5.0 ton ha–1 oilseed compost; 10 OC, 10 ton ha−1 oilseed compost.

 Control16.2 a1.38 c2.43 c
 CF12.6 b1.60 bc3.34 b
 1/2 CF + 2.5 OC14.7 ab1.99 b3.58 b
 5.0 OC16.7 a1.99 ab4.48 a
 10 OC16.4 a2.51 a4.03 a
 Control143 c8.08 a4.90 c
 CF146 c7.11 b5.55 b
 1/2 CF + 2.5 OC163 b7.44 b5.85 b
 5.0 OC145 c6.39 c4.95 c
 10 OC198 a6.60 bc6.71 a

Prospects and future research needs

Rapid civilization makes soil contamination an inevitable problem and a big challenge to scientists and environmental policy makers. For healthy and sustainable future generations, the soil resource should be protected against slow and insidious poisoning by heavy metals released from industrial and agricultural activities in the world. Although there are lots of available studies on soil pollution, the guidelines established by individual countries worldwide to control the pollution of agricultural soils are not consistent and are not standardized, pointing to the complexity of heavy metal behavior in agro-environmental systems, various climatic, geological and hydrological conditions, and the political and non-scientific factors affecting the establishment of regulations in any country.

We provide a schematic summary of heavy metals in identification, transport and remediation surrounding the agro-environmental impact in paddy soils in Taiwan (Fig. 7). Considering the central sequence (solid boxes) of processes under the environmental monitoring system and according to the regulation of soil pollutants in the SGWPR Act in Taiwan, the source, fate and remediation of heavy metals for food safety is a dynamic consequence of the time-integrated scientific information. However, site-specific and health-based risk assessments are deemed to be the most reliable and practical approaches in resolving problems of soil pollution. More scientific evidence from case-specific research, particularly from long-term field trials involving all types of key conditions and factors are necessary to understand the bioavailability of heavy metals in various soil types after long periods of time and to provide reliable parameters for health-based risk assessments. Eventually, the collection of reliable databases on heavy metal concentrations in soils and different crops must be given utmost importance. From these databases, a variety of useful information and methodologies can be developed toward achieving the ultimate goal of providing safe and high-quality food for humans in the future.

Figure 7.

 Schematic summary of the identification, transport and remediation of heavy metals in paddy soils in Taiwan.


In addition to biogeochemical processes, industrial and swine wastewater and hazardous solid wastes are the main anthropogenic sources of heavy metals in the paddy soils of Taiwan that threaten food safety and quality of life. Consequently, heavy metal levels in rural soils have been surveyed by the Taiwan EPA since 1982. Currently, the rural soil quality is regulated by the SGWPR Act according to the SCS of As, Cd, Cr, Cu, Hg, Ni, Pb and Zn; rural soils dominant the control sites announced by the Taiwan EPA. Owing to the heavy metal sources from wastewater by irrigation canal systems, heavy metals are not only concentrated in the surface soil, but are also highly distributed at the main entrance of irrigation water into paddy fields.

Cadmium in rice grains is significantly accumulated in Indica varieties compared with Japonica species, and these levels exceed the FQS. Therefore, Indica varieties are not recommended for cultivation in Cd-contaminated soils. According to the 19 paddy field study, the size of the reactive metal pool relative to the total metal pool increased as follows: Cr << Ni ∼ Zn < Pb < Cu < Cd. However, uptake of Cd into rice roots for all cultivars was inhibited by an increase in soil bioavailable Zn.

With respect to soil remediation techniques, turnover and dilution by mixing surface-contaminated soil and subsurface clean soils has the advantage of low cost in terms of time and budget. In situ stabilization by adding low-cost amendments is an alternative technique in soil remediation; however, long-term transformation of heavy metals should be re-evaluated by the partition of metals in the soils. Field-scale practices of phytoextraction enable contaminated sites to recover to their original function and generate economic benefit for farmers; however the soil remediation time of phytoextraction is considerably longer than the other techniques.

Despite enriched As, Cr and Ni in some paddy soils from andesite and serpentinites of Taiwan, these metals were not clearly accumulated in rice. Nevertheless, issues for food quality of other crops and ecological toxicity potential should pay more attention by monitoring network system in Taiwan. It is necessary to identify the bioavailability of heavy metals in various soil types and crops in order to provide reliable parameters for health-based risk assessments. The collection of reliable data on heavy metal speciation in soils and crops is critically important and further studies are needed to determine the contribution of bioavailable metals to improving soil quality to achieve food safety and sustainable agriculture.