Arsenic Concentrations in Paddy Soil and Rice and Health

Apr 8, 2014 - Implications for Major Rice-Growing Regions of Cambodia ... Cambodia and coupled that data to soil-chemical factors based on extractions...
0 downloads 0 Views 4MB Size
Article pubs.acs.org/est

Arsenic Concentrations in Paddy Soil and Rice and Health Implications for Major Rice-Growing Regions of Cambodia Angelia L. Seyfferth,*,† Sarah McCurdy,‡ Michael V. Schaefer,‡ and Scott Fendorf‡ †

Department of Plant and Soil Sciences, University of Delaware, Newark, Delaware 19716, United States Department of Environmental & Earth System Science, Stanford University, Stanford, California 94305, United States



S Supporting Information *

ABSTRACT: Despite the global importance of As in rice, research has primarily focused on Bangladesh, India, China, and the United States with limited attention given to other countries. Owing to both indigenous As within the soil and the possible increases arising from the onset of irrigation with groundwater, an assessment of As in rice within Cambodia is needed, which offers a “base-case” comparison against sediments of similar origin that comprise rice paddy soils where As-contaminated water is used for irrigation (e.g., Bangladesh). Here, we evaluated the As content of rice from five provinces (Kandal, Prey Veng, Battambang, Banteay Meanchey, and Kampong Thom) in the rice-growing regions of Cambodia and coupled that data to soil-chemical factors based on extractions of paddy soil collected and processed under anoxic conditions. At total soil As concentrations ranging 0.8 to 18 μg g−1, total grain As concentrations averaged 0.2 μg g−1 and ranged from 0.1 to 0.37 with Banteay Meanchey rice having significantly higher values than Prey Veng rice. Overall, soilextractable concentrations of As, Fe, P, and Si and total As were poor predictors of grain As concentrations. While biogeochemical factors leading to reduction of As(V)-bearing Fe(III) oxides are likely most important for predicting plant-available As, husk and straw As concentrations were the most significant predictors of grain-As levels among our measured parameters.



Bengal basin.12,13 Owing to the ability to obtain additional rice crops with minimal infrastructure, groundwater may be used increasingly for dry-season rice production throughout the Mekong delta (Cambodia and Vietnam), and other agriculturally developing areas such as Myanmar may also follow. Even without the use of As-containing groundwater for irrigation, As uptake by rice will occur, especially when grown under flooded (vs nonflooded) conditions.14,15 Under flooded paddy conditions, reductive dissolution of As(V)-bearing iron(III) oxides results in the release of adsorbed As to the pore-water, which is subject to plant uptake.16 Arsenite dominates in flooded rice paddy soil17 and is taken up by rice roots due to its chemical similarity to silicic acid.18 The organic forms of As, DMA and MMA, are also transported into rice via the silicic acid pathway,19 but these chemical species are typically present in much lower amounts than arsenite and arsenate.17,20 Arsenate is taken up by rice roots due to its chemical similarity to phosphate.21 After uptake, As species are eventually transported to rice grains where they can impact human health upon consumption. Presently, groundwater is seldom used for rice irrigation in Cambodia.22 However, as Cambodia continues to develop with population increases, the rising demand for food and income is

INTRODUCTION Uptake of As by rice poses a threat to food security by impacting both the quantity (yield) and quality of rice.1,2 Arsenic is a known human carcinogen, and As-contaminated rice was suggested to elicit apparent genotoxic effects in humans who consumed rice with over 0.2 μg g−1 As.3 Arsenic in rice is a global issue impacting the lives of billions of rice consumers worldwide, especially when considering that rice is often exported from one country to another. Despite the global importance of As in rice, much of the research has been focused in Bangladesh, India, China, and the United States with little focus in other countries. Within Bangladesh, an epicenter of As contamination, Ascontaining groundwater from shallow tube wells is often used for dry-season rice irrigation.4 Because of dry-season irrigation, Bangladeshi farmers are able to produce up to three crops per year; however, this practice causes soil As concentration to rise,5 and hence an increase in As content in6 and toxic effects on7 rice, with a concomitant decrease in yield8 when irrigation water contains As. As a consequence, rice consumers are further exposed to unsafe levels of As in ricein addition to As that may be ingested through drinking water.2 It is noteworthy that the entire deltaic region of South and South East Asia is prone to As contamination.9 While groundwater use for irrigation is minimal in the Mekong delta in comparison to the Bengal basin,9 some farmers in the lower Mekong delta are increasingly pumping groundwater for irrigation,10 which may increase soil As and subsequent rice content,11 as has happened in the © 2014 American Chemical Society

Received: Revised: Accepted: Published: 4699

November 11, 2013 April 3, 2014 April 8, 2014 April 8, 2014 dx.doi.org/10.1021/es405016t | Environ. Sci. Technol. 2014, 48, 4699−4706

Environmental Science & Technology

Article

soils from these lowland sites developed from Mekong-derived sediment and have a greater risk for As in groundwater. In August 2011, 12 locations in Battambang, nine locations in Banteay Meanchey, and one location each in Kampong Thom and Siem Reap Provinces were sampled. Sites in both Battambang and Bantey Meanchey were lowland areas and soils were influenced by the Tonle Sap, a large inland water body that ebbs and flows with the monsoon (Figure 1), whereas sites from Kampong Thom and Siem Reap were upland sites with little influence from the Tonle Sap (Figure 1); Kampong Thom is the only province of these four with elevated risk of As in groundwater. Where possible, both soil and rice plants were obtained in the field. All sampling sites were small rice paddy fields (ca. 1000 m2) that were operated by homeowners. In some instances, only rice grains, husk, and straw were attainable, and in other sites only soil was attainable; thus, not every rice sample has a corresponding soil sample. There were a total of 23 sites from which both soil and rice samples were collected, and 31 sites from which grain was sampled. Soil. Soil was obtained from at least three locations in each field and included opposite ends and the middle of the paddy. Soil was obtained from ca. 0−20 cm depth using a soil probe that was pushed downward into the soil and was combined into a composite sample. Immediately after collection, the composite sample was placed inside of a gas-impermeable bag with two oxygen scrubbers and taped closed. Within 5 h of collection, fresh oxygen scrubbers were placed inside of all soil sample bags, which were completely vacuum-sealed. Soil sample bags remained vacuum-sealed (and oxygen-free) during air-transport back to Stanford University for analysis. Rice Tissues. Similar to soil samples, rice samples were obtained from at least three locations in each field and included opposite ends and the middle of the field site. In cases where plant material had recently been harvested and removed from the field, straw and husk + grain samples were obtained from piles on the homeowner’s land. In either case, three large handfuls of plants or plant materials were obtained and combined into a representative composite sample. Plant materials were transported to the Analytical Laboratory at Resource Development International (RDI) in Kandal Province 1 day prior to departure for the U.S. Where needed, straw was separated from pannicles and stored in aluminum foil. Husk + grain samples were stored in high-density polyethylene (HDPE) vials. Upon entry into the U.S., husk + grain samples were sent to the U.S. Department of Agriculture National Plant Germplasm Inspection Station in Beltsville, MD, where samples were heat-sterilized before being returned to the laboratory per conditions of the USDA importation permit. Rice Tissue Preparation, Digestion, and Analysis. Once in the laboratory at Stanford University, straw and husk + grain samples were allowed to air-dry and were further oven-dried at 65 °C overnight. Rice tissues were digested and analyzed per the procedures in Seyfferth and Fendorf.26 Briefly, grain and husk were separated, and straw, unpolished grain, and husk were ground, homogenized, and digested with HNO3/H2O2 for analysis of As with ICP-MS or plant nutrients with ICP-OES. For Si analysis of husk and straw, a second digest with NaOH was used with subsequent acidification and analysis of Si by ICP-OES. Duplicate analyses of plant tissue samples always agreed to within 10%, and digested NIST 1568a rice flour agreed to within 10% of the reported value; see Seyfferth and Fendorf26 for additional details on QA/QC.

expected to increase groundwater usage for dry-season rice irrigation.22 Only very recently have studies evaluated the As content of rice in Cambodia, and they have focused on regions near the Mekong River in sites where As has already impacted soil.11,23,24 An investigation of As content in rice across ricegrowing regions of Cambodia that are not yet impacted by irrigation with As-tainted groundwater is needed to assess both the current (potential) health burden arising from rice consumption as well as an ability to track changes in grain As content with likely changes in land use and agricultural intensification. Accordingly, here we evaluated the As content of rice from rice-growing regions of Cambodia in five provinces that vary in their risk of As in groundwater. We coupled rice samples to paddy soil samples, where possible, and used statistical analyses to develop predictive models for As content of rice tissues based on plant and soil-chemical factors.



EXPERIMENTAL SECTION Sample Collection. Sampling of soil and/or rice occurred in March and August of 2011 in major rice-growing regions of Cambodia (Figure 1). In March 2011, six locations in Kandal Province and 9 locations in Prey Veng Province were sampled;

Figure 1. Rice-growing regions of Cambodia (A); sampling locations and grain-arsenic concentrations from five provinces in Cambodia (B). Panel A is modified from ref 25. 4700

dx.doi.org/10.1021/es405016t | Environ. Sci. Technol. 2014, 48, 4699−4706

Environmental Science & Technology

Article

Figure 2. Average (±standard deviation) grain (A), husk (B), straw (C), and total above ground (D) arsenic concentrations in rice collected from five Provinces in Cambodia. KD = Kandal (n = 6); PV = Prey Veng (n = 6); Ban M = Banteay Meanchey (n = 8); Bat = Battambang (n = 10); K Thom = Kampong Thom (n = 1); SR = Siem Reap (n = 1). Dashed horizontal lines in panel A represent the percentage of daily allowable arsenic intake (WHO, 2.1 μg kg−1 day−1) at the designated grain arsenic concentrations for a 65 kg adult consuming a typical Cambodian diet of 450 g rice day−1.

fluorescence spectrometry (Spectro XEPOS); absolute error of the As measurement was 0.3 μg g−1 (4 μmol kg−1), and analysis of a NIST certified reference soil (San Joaquin, 2709) gave As recoveries of 95%. Statistical Analyses. The mean grain, husk, and straw As concentrations were compared between Banteay Meanchey, Battambang, Kandal, and Prey Veng Provinces using ANOVAs and Tukey HSD tests. Soil-extractable As, P, Si, and Fe were similarly compared among provinces using ANOVAs and Tukey HSD tests. Correlation analysis, linear regressions, and multiple linear regressions between rice tissue concentrations and soil-extractable concentrations of As, Fe, P, and Si or total As were performed with all 23 sites from which both soil and rice tissues were obtained. All statistical analyses were performed with SPSS (v. 21) software.

Soil Preparation, Extraction, and Analysis. Within the laboratory (Stanford University), paddy soil samples were transferred to an anoxic chamber (95% N2, 5%H2); they were then removed from the sample bags and allowed to dry under anoxic conditions. Due to the high clay content of the samples, soil samples became hardened during drying and were thus broken-up using light-force with a mortar and pestle to pass through a 2 mm sieve. Using the 2 mm sieved portion, extractions with 0.01 M CaCl2 and 0.5 M acetic acid (HOAc) were conducted using a 1:10 soil/solution ratio (w/w) and 2 h shaking time at the University of Delaware. The CaCl2 extraction was chosen to estimate plant-available As, Si, and P using a single extraction so that direct comparisons between elements could be made, after Houba et al.,27 while acetic acid extractions provide an estimate of plant-available Si.28 A portion of the dried soil sample was further passed through a 150 μm sieve and was used for citrate−bicarbonate−dithionite (CBD) extraction and for acid−ammonium−oxalate (AAO) extraction according to standard methods;29 these methods nominally evaluate elements associated with reducible iron and poorly crystalline (and some small particle-sized crystalline) oxides of iron and aluminum, respectively. After dilution and acidification, where necessary, all soil extractions were analyzed with ICP-OES for P, Fe, Si, and As and quantified using matrixmatched standards. The CaCl2 and HOAc extractions were additionally analyzed with ICP-MS at the University of Delaware for As and quantified using matrix-matched standards. Total soil As was determined on 150 μm sieved soil using X-ray



RESULTS Plant As and Si Concentrations. Arsenic concentrations in rice grains from all five provinces sampled ranged from 0.10 to 0.37 μg g−1 (Figure 2A, Table SI 1, Supporting Information). The average grain-As concentration was 0.20 (±0.06) μg g−1, and the median grain As concentration was 0.21 μg g−1. GrainAs concentrations differed significantly among sites (P = 0.023) with Banteay Meanchey rice having significantly higher grain-As than Prey Veng rice; there was no difference between mean grain-As levels from Kandal or Battambang rice (Figure 2A). Husk-As concentrations varied little between or within provinces and ranged from 0.06 to 0.34 μg g−1 (Figure 2B). In 4701

dx.doi.org/10.1021/es405016t | Environ. Sci. Technol. 2014, 48, 4699−4706

Environmental Science & Technology

Article

Figure 3. Average (±standard deviation) CaCl2-extractable (A), acid ammonium oxalate (AAO)-extractable (B), citrate−dithionite−bicarbonate (CBD)-extractable (C), and total (D) arsenic concentrations in rice paddy soil collected from six Provinces in Cambodia. KD = Kandal (n = 6); PV = Prey Veng (n = 6); Ban M = Banteay Meanchey (n = 8); Bat = Battambang (n = 10); K Thom = Kampong Thom (n = 1); SR = Siem Reap (n = 1).

provinces (P = 0.006), with Kandal soil having higher values and Battambang having lower values (Figure 3B). Citrate− bicarbonate−dithionite (CBD)-extractable As ranged from below detection to nearly 80 μmol kg−1, which represented from 0 to 71% of total soil As; Prey Veng soils had significantly higher values than the other provinces (P = 0.001) (Figure 3C). Soil extractable Si, P, and Fe are summarized by province in Figures SI 2−4. Of note, soils from Siem Reap and Kampong Thom had lower extractable Si and Fe than the other provinces. CaCl2-extractable P differed significantly among provinces (P < 0.0001) with Prey Veng soil having significantly higher values than the other provinces (Figure SI 3A). Both AAO- and CBDextractable P also differed significantly among provinces (P < 0.0001), where in both cases Kandal soil had higher values than Banteay Meanchey and Battambang soils (Figure SI 3B,C). Both AAO- and CBD-extractable Fe did not significantly differ among provinces (Figure SI 4). Relation between Plant As and Si and Plant and Soil Parameters. Pearson correlation analyses were performed between plant As, plant Si, and soil-extractable As, Si, P, or Fe (Table SI 2). There was a strong positive correlation between grain As and both husk As (r = 0.731, P < 0.0001) and straw As (r = 0.583, P = 0.004); however, none of the soil parameters evaluated were significantly correlated with grain As (Figures SI 5, 6 and Table SI 2). There was a negative correlation between straw As and CaCl2-extractable Si (r = −0.424, P = 0.049) and straw Si (r = −0.411, P = 0.058) (Figure SI 7 and Table SI 2). Straw Si was negatively correlated with CaCl2-extractable As (r = −0.567, P = 0.006) and positively correlated with CaCl2-, and HOAc-extractable Si (r = 0.554, P = 0.007; r = 0.505, P = 0.017;

contrast, straw-As concentrations varied widely both between and within provinces with the lowest (0.28 μg g−1) and highest (4.2 μg g−1) straw-As concentrations found within the same (Kandal) province (Figure 2C). There were no significant differences in husk, straw, or above-ground (i.e., sum of straw, husk, grain) As concentrations among provinces (Figure 2B− D). There were, however, significant differences in straw-Si concentrations among provinces (P = 0.047) with Kandal having significantly lower straw-Si values. Despite differences in straw-Si concentration among the five provinces, they were all relatively low and ranged from 2% to just over 4%, whereas husk Si concentrations were higher and ranged 4 to 7.5% (Figure SI 1, Table SI 1, Supporting Information). Total As and Soil-Extractable As, Si, P, and Fe. Total soil As concentrations ranged from 0.8 to 18 μg g−1 (11−238 μmol kg−1) and differed significantly among provinces (P < 0.0001), with Banteay Meanchey and Battambang having lower soil As than Kandal and Prey Veng provinces (Table SI 1, Figure 3). Calcium chloride-extractable As ranged from 0.003 to 0.28 μmol kg−1, representing between 0.004 to 2.6% of the total soil As (Figure 3). Soils from Kandal province had significantly higher CaCl2-extractable As than Banteay Meanchey and Prey Veng provinces (P = 0.004) (Figure 3A). The highest CaCl2-extractable As was found in Kampong Thom and Siem Riep provinces; however, these were not included in the ANOVA because only one sample from each province was obtained. Acid ammonium oxalate (AAO)-extractable As ranged from 4 to nearly 50 μmol kg−1 and represented from 10 to 46% of the total soil As. The pattern of AAO-extractable As was similar to total soil As, differing significantly among 4702

dx.doi.org/10.1021/es405016t | Environ. Sci. Technol. 2014, 48, 4699−4706

Environmental Science & Technology

Article

Figure 4. Regressions (n = 23) between grain As and husk (A) or straw (B) As, and between straw As and CaCl2-extractable (C) or straw (D) Si from five provinces in Cambodia. Ban M = Banteay Meanchey (n = 8); Bat = Battambang (n = 7); K Thom = Kampong Thom (n = 1); KD = Kandal (n = 5); PV = Prey Veng (n = 2). ***P < 0.0001; **P < 0.005; *P ≤ 0.05.



and r = 0.430, P = 0.046, respectively) (Figure SI 8 and Table SI 2). Regression analyses were performed to find potential relationships between grain, husk, or straw As (dependent variables) and husk As, straw As, straw Si or soil-extractable As, Si, P, or Fe (independent variables). Positive linear relationships were found between grain As and husk As (r2 = 0.535, P < 0.0001) or straw As (r2 = 0.340, P = 0.004) (Figure 4A,B), whereas a weak, negative exponential relationship was found between grain As and CaCl2-extractable P (r2 = 0.192, P = 0.036) (data not shown). Aside from CaCl2-extractable P, grain As was not predictable by any other extractable soil parameter inclusive of extracts of As and Si. Weak and negative linear relationships were found between straw As and CaCl2extractable Si (r2 = 0.180, P = 0.049) or straw Si (r2 = 0.169, P = 0.058) (Figure 4C,D). Stepwise multiple linear regression analysis was performed to predict grain As or straw As from soil-chemical characteristics and other plant portions. Parameters used to predict grain As were husk As, straw As, straw Si, CaCl2-extractable Si, As, and P, acetic acid-extractable Si, oxalate-extractable Fe, As, P, and total As. With the exception of plant-As levels, the same parameters were used to predict straw As. For grain As, the prediction model only contained two of the predictors, husk As and straw As, and was statistically significant (P = 0.003); however, this model only explained about 47% of the variance of grain As (R2 = 0.473). For straw As, the prediction model only contained CaCl2-extractable Si and was statistically significant (P = 0.049); however, this model only explained about 18% of the variance of straw As (R2 = 0.180).

DISCUSSION

Levels of As and Si in Rice. Arsenic concentrations in rice grains vary widely as a result of differences in plant uptake between varieties and differences in soil-chemical factors.30,31 Total As concentrations in unpolished Cambodian rice grains in the present study, ranging from 0.10 to 0.37 μg g−1, were within the range that has been reported in market- and fieldsurveys worldwide.1,30−36 The highest grain As concentrations that have been reported (up to 1.8 μg g−1) were found in fieldcollected samples from districts of Bangladesh where Ascontaminated groundwater is used for rice irrigation,30 but levels above ∼0.6 μg g−1 are not common.37−42 While the grain As levels in Cambodia were not considered highly elevated, the As contents of ≥0.2 μg g−1 arguably could contribute negatively to human health upon chronic exposure.3 In addition, our data provide a baseline of “normal” grain-As concentrations in a country where appreciable agricultural intensification, and most notably irrigation with groundwater, has not yet taken place. The average concentration of grain As in our study (0.20 μg g−1) is similar to recent studies in Prey Veng and Kandal provinces11,23 and is at the upper limit of the “normal” range reported by Zavala and Duxbury.42 Similar grain-As levels were found in Bangladeshi paddies not irrigated with groundwater,43 and our results are a factor of 2−3 lower than rice grains from paddies irrigated with As-impacted groundwater.33 Surprisingly, there was no difference in average grain As between Kandal and Battambang provinces, despite soil from Kandal having higher AAO-extractable As and being considered more severely impacted by As owing to deposition of Himalayan-derived Mekong River sediments. 4703

dx.doi.org/10.1021/es405016t | Environ. Sci. Technol. 2014, 48, 4699−4706

Environmental Science & Technology

Article

As as well as the amount of labile carbon available for microbial metabolism and the depth and duration of soil flooding.14,51 Predictive models based solely on dry soil-chemical conditions that do not consider the redox dynamics of flooded paddy soil are thus expected to be poor predictors of As uptake by rice. While redox processes dominantly control the dissolved As concentrations in flooded rice paddies,17,50 varietal differences in As uptake at the root, and transfer to the straw and grain are also important.26,31,52 The most prominent factors predicting grain As concentrations measured in our study were the husk As or straw As concentrations; each had a significant and positive linear relation with grain As but only explained 53% and 34%, respectively, of the variation in grain As. We expect that including the evolution of pore-water As concentrations among sites would help to explain a large portion of the variation in grain As. In our study, the straw-to-grain As transfer decreased exponentially with increasing straw As concentrations (Figure 5), which is a finding similar to previous work

The low amount of husk As we found in Cambodia (