The field study was conducted at the USDA-ARS Dale Bumpers National Rice Research Center near Stuttgart, Arkansas in 2009. The soil on the site is a Dewitt silt-loam soil (fine, smectitic, thermic Typic Albaqualf). The experiment was arranged in a split-split plot design with four replications per treatment, in which MSMA-amended soils were main plots and water management as sub-plots. The main plots included MSMA-amended and no MSMA treatments. The MSMA-amended plots were established since 1980s for the straighthead resistance study (Yan et al., 2005), where MSMA was sprayed and incorporated into soil at a rate of 6.7 kg MSMA ha-1 prior to planting, as described by Yan et al. (Yan et al., 2005). The Native plot was located nearby where no MSMA was applied. Topsoil (0–20 cm) were collected from both MSMA-amended and Native plots before rice planting for soil characterization (Hua et al., 2011). Each main plot was divided into four completely randomized sub-plots for water treatments. The water treatment included four water management practices: one intermittent flooding and three continuous flooding. Intermittent flooding conditions were maintained by initially flooding the soil, and then allowing soil to dry until small, surface cracks were evident. At this point, the plots were irrigated up to 10 cm water depth with a nearby reservoir water (pH ∼7.0, <2 μg A L-1). Continuous floodings were conditions under which plots were drained at 10, 20, or 35 days after the heading date (50% panicle emergence from flag leaf). Each sub-plot contained six-row (1.8 m by 1.5 m) blocks in a completely randomized arrangement with four replications per cultivar (Pillai et al., 2010). Six rice cultivars were selected in this study: Cocodrie, Rondo and Wells from the south-central U.S.; Zhe 733 and GP-2 from China; and Spalcik from Russian Federation (Yan et al., 2008; Yan and McClung, 2010). These six cultivars included early, mid- and late-maturity and varied with the resistance or tolerance for As-induced straighthead disease (resistant, moderate, and susceptible) (Yan et al., 2008; Yan and McClung, 2010). This field study concurred with development of the practical solution to minimize grain-As accumulation in field conditions and primarily focused on Zhe 733 cultivar, since it has been identified as an As tolerant cultivar and widely studied for As-induced straighthead disease (Yan et al., 2005; Yan et al., 2008) and grain-As uptake (Yan et al., 2005; Pillai et al., 2010; Hua et al., 2011).

At maturity, rice grains from each plot were harvested, dried at 65°C to approximately 12% of moisture content, and stored at room temperature for digestion and then chemical analysis. Prior to digestion, grain samples were de-hulled using a ceramic mortar and pestle and oven dried at 65°C for 48 h and stored in a desiccator. 127.

A Multiwave 3,000 microwave digester (Anton Paar, Austria) was used for grain and soil sample digestion. In particular, EPA Method 3,052 was followed to digest rice grain samples, and slightly modified EPA Method 3051A was used to digest soil samples for the determination of the total concentrations of trace elements. For rice grain digestion, 0.5 g rice grain was placed in the acid cleaned Rotor 48 (MF50) digestion vessels and added with 8.0 mL optima grade nitric acid (HNO3) and 0.3 mL trace metal hydrogen peroxide (H2O2) (Fisher Scientific Co., Boston, U.S.). The digestion program consisted of a 10-min ramp to 100°C with a 10-min hold, followed by a 10-min ramp to 140°C, with a 15-min hold, then a 10-min ramp to 170°C, with a 10-min hold before cooling down.

For soil digestion, 0.5 g soil samples were placed in acid cleaned Rotor 8 (XF100) digestion vessels and digested with 9.0 mL trace metal grade HNO3 and 3.0 mL trace metal grade hydrochloric acid (HCl). The digestion program consisted of a 10-min ramp to 1100 W, with a 20-min hold before cooling down. All digested samples were allowed to cool down to <50°C before venting. Digested samples were rinsed from digestion vessels to acid-cleaned 50 mL conical tubes, and then diluted to 20 mL using Millipore water (18.2 MΩ-cm).

Standard reference materials (NIST, Gaithersburg, MD) for rice grain (SRM1568a) and soil (SRM2711a Montana II) were included in respective digestions for quality assurance and quality control (QA/QC). Additional digestion QA/QC included one sample blank, one sample duplicate, and one sample spike per 10 samples for rice and soil digestion. Spike concentrations were analyzed at an estimated two times of average sample concentrations.

Total As, Se, Mo, and Cd concentrations in digested grain samples and extractable Se, Mo, and Cd concentration in soil samples were determined using inductively coupled mass spectrometry (ICP-MS) (PerkinElmer Elan DRC-e, MDS Sciex, Concord, Ontario, Canada). Extractable As from soil was determined using graphite furnace atomic adsorption spectroscopy (GFAA) (PerkinElmer AAnalystTM 600, Shelton, CT, United States). Digested grain and soil samples were diluted using 10% and 5% HNO3 prior to chemical analysis, respectively. Sample method detection limits (MDLs) for all tested elements were determined for GFAA and ICP-MS, according to EPA Method 200.9 and 200.8, respectively. Estimated soil MDLs for As, Se, Mo, and Cd were 0.10, 0.08, 0.02, and 0.02 mg kg-1, respectively, while rice grain MDLs for As, Se, Mo, and Cd were 0.004, 0.031, 0.003, and 0.003 mg kg-1, respectively. Sample concentrations greater than 2–3 times MDL were required. QA/QC was validated if the following criteria were met: (a) relative difference of a duplicate sample less than 20%; (b) spike recovery between 85% and 115%; and (c) the accuracy of a calibration standard within 10% per 10 samples analyzed.

The analysis of variance (ANOVA) procedure that was appropriate for the split-split-plot design was accrued out using SAS 9.2 to estimate the impact of experimental variables on trace element uptake. The experimental variables (factors) were soil-As (Soil), irrigation management (Water), and rice cultivar (Cultivar), with Soil as the main-plot factor, Water as the sub-plot factor and Cultivar as the sub-sub-plot factor. ANOVA allows interaction terms representing the combined impact of different factors on element uptake to be tested, whereby a significant interaction term signifies that the impact of the changing levels of one factor is influenced by the level of the other factor(s). Least square means with Tukey’s adjustment were used to contrast means of specific Soil-Water combinations. Tukey’s means adjustment allowed experiment-wise error to be kept at the selected rate, α = 0.05 (Montgomery, 2009). Pearson’s partial correlations between grain-As and grain-Se, grain-Mo, and grain-Cd were also estimated.

Characterization and major physiochemical properties of two soils, Native and MSMA-amended soils, were presented in Hua et al. (Hua et al., 2011) and Yan et al. (Yan et al., 2008). The average As content in Native soil was 5.9 ± 0.25 mg kg^-1 while MSMA-amended soil 19.5 ± 0.36 mg kg^-1. Furthermore, Native soil contained 0.230 ± 0.004 mg Se kg^-1, 0.38 ± 0.04 mg Mo kg^-1, and 0.060 ± 0.003 mg Cd kg^-1 Cd, while MSMA-amended soil contained 0.53 ± 0.09 mg Se kg^-1, 0.63 ± 0.24 mg Mo kg^-1, and 0.060 ± 0.006 mg Cd kg^-1.

For each type of soil, total concentrations of the four trace elements analyzed were found not significantly different among three continuous flooding treatments (drain at 10, 20, and 35 days after heading), but MSMA amendment resulted in significantly higher grain-As and lower grain-Mo, as indicated in Table 1. Thus, the average value of three continuous flooding treatments was used when compared with that from the intermittent flooding treatment.

Elevated grain-As concentrations were reported in low-As soils under both continuous and intermittent flooding conditions, and the presence of trace MSMA quantities could strongly affect grain-As (Pillai, 2009). Yan et al. (Yan et al., 2005) showed increased grain-As and reduced grain yield due to MSMA-induced straighthead susceptibility under continuous flooding conditions.

The effect of soil-As on average grain concentrations of As, Se, Mo, and Cd under continuous and intermittent flooding for the As-tolerant Zhe 733 was presented in Figure 1. Data can be compared with current global As standards of the United Kingdom Food Statutory General Limit (1.0 mg kg-1 fresh weight in food) (Kingdom et al., 1959). Currently, there is no food standards established for As, Se, Mo, and Cd in United States. Zhe 733 cultivar grown under continuous flooding and MSMA-amendments significantly increased grain-As from 0.30 to 0.60 mg kg-1 (103%), grain-Se from 0.24 to 0.31 mg kg-1 (30%), grain-Cd from 0.014 to 0.016 mg kg-1 (18%), and decreased grain-Mo from 1.15 to 0.64 mg kg-1 (45%) (Figure 1). Under intermittent flooding, MSMA-amendments increased As from 0.10 to 0.22 mg kg-1 (115%), Se from 0.19 to 0.52 mg kg-1 (175%), Cd from 0.023 to 0.065 mg kg-1 (183%), and decreased Mo from 0.42 to 0.36 mg kg-1 (−13%). Thus, MSMA amendment resulted in the same trend on trace element concentrations in both irrigation managements, with a greater impact by intermittent flooding. The highest grain-As levels from MSMA-amended and continuous flooding were considerably less than the United Kingdom Food Statutory General Limit (Kingdom et al., 1959). However, rice consumption at the average U.S. consumer diet of ≥61.2 g rice grain day-1 (Batres-Marquez et al., 2009) and the average Asian-American diet of 400 g rice grain day-1 (Navarro-Alarcon and Cabrera-Vique, 2008) may result in an ingestion of approximately 0.52 and 3.4 μg A kg^-1 day^-1, respectively, which is substantially greater than the regulated intake amount of 0.29 μg A kg^-1 day^-1 through drinking water (EPA, 2001). The highest grain- Cd levels of 0.065 mg kg^-1 were found in MSMA-amended intermittent flooding conditions. Consuming such rice grain for the average U.S. consumer and the Asian-American may intake 0.4 and 2.6 μg kg^-1 wk^-1, which is less than the JEFCA Cd standard (FAO/WHO, 2003). Furthermore, increased soil-As level not only increased grain-As, but also impacted the uptake of Se, Mo, and Cd, especially under intermittent flooding conditions.

A general trend of soil-As on average grain trace element concentrations of other five cultivars under continuous and intermittent flooding was illustrated in Figure 2. For the five cultivars under continuous flooding, MSMA amendment increased As from 0.33 to 1.16 mg kg-1 (252%), Se from 0.34 to 0.44 mg kg-1 (29%), Cd from 0.016 to 0.021 mg kg-1 (31%), and decreased Mo from 1.19 to 0.59 mg kg-1 (50%). Under intermittent flooding, MSMA amendment increased grain-As from 0.26 to 0.38 mg kg-1 (46%), Se from 0.41 to 0.70 mg kg-1 (71%), decreased Mo from 0.67 to 0.50 mg kg-1 (25%) and Cd from 0.057 to 0.050 mg kg-1 (12%). Overall, increasing soil-As increased levels of grain-As and grain-Se, but decreased grain-Mo level for both irrigation management practices. However, MSMA amendment slightly increased grain-Cd under continuous flooding but slightly decreased grain-Cd under intermittent flooding.

Zhe 733 cultivar had been widely used as a resistant check cultivar for As-induced straighthead disease studies (Yan et al., 2005; Yan et al., 2008; Pillai, 2009; Pillai et al., 2010; Hua et al., 2011; Somenahally et al., 2011). Comparing with five other cultivars, Zhe 733 had much less grain-As concentrations in all of the irrigation-MSMA treatments: 9% less in Native-continuous, 48% less in MSMA-continuous, 62% less in Native-intermittent, and 42% less in MSMA-intermittent (Figures 1, 2). Data suggested that straighthead resistance be associated with less As uptake.

Also shown in Figure 1, in compared with the continuous flooding, Zhe 733 cultivar in Native soil under intermittent flooding significantly decreased grain-As from 0.30 to 0.10 mg kg-1 (66%), grain-Se from 0.24 to 0.19 mg kg-1 (21%), grain-Mo from 1.15 to 0.42 mg kg-1 (63%) and increased grain-Cd from 0.014 to 0.023 mg kg-1 (64%). The other five cultivars grown in same soil under intermittent flooding slightly decreased grain-As from 0.33 to 0.26 mg kg-1, slightly increased grain-Se from 0.34 to 0.41 mg kg-1, significantly decreased grain-Mo from 1.19 to 0.67 mg kg-1, and significantly increased grain-Cd from 0.016 to 0.057 mg kg-1 (Figure 2).

Soil redox potential had a large impact on the fate and transport of As, due to the presence of insoluble FeIII (hydr)oxides and, consequently, the formation of AsV-FeIII complexes in aerobic systems (Charlet et al., 2011; Wang et al., 2011). Soil redox measurements in intermittent and continuous flooding plots (339 ± 84 vs 160 ± 52 mV) supports the above analysis because intermittent flooding condition provided a more oxidized condition facilitating the formation of AsV-FeIII complexes, decreasing As solubility and subsequent plant uptake (Masscheleyn et al., 1991). In addition, under more oxidized conditions, As is likely in the less mobile AsV form (Cullen and Reimer, 1989).

For Zhe 733, intermittent flooding also significantly decreased grain-Se. However, the overall trend of the other five cultivars indicated a slight increase in grain-Se under intermittent flooding, confirming the results from Mikkelsen et al. (Mikkelsen et al., 1989). This further suggested that the Se uptake mechanism by rice was more of a function of cultivar. Therefore, cultivar selection based on minimizing grain-As may also have implications on other elements. Similar to grain-As, overall grain-Mo significantly decreased as a result of intermittent flooding. The primary uptake mechanism for Mo and As under aerobic conditions occurs via the phosphate pathway (Balistrieri et al., 1994). Thus, it is evident that the oxidized soil environment would regulate and reduce Mo bioavailability by the formation of immobile iron (hydr)oxide complexes, resulting in lower mobility and thus plant uptake (Moore and Patrick, 1991). Overall grain-Cd concentrations significantly increased under the intermittent flooding. Due to the oxidized conditions associated with intermittent flooding, insoluble CdS may have converted to the more soluble CdSO4, resulting in a greater Cd availability for rice uptake (Ito and Iimura, 1976).

The significant impact of cultivars on the soil As and water management interaction in both Native and MSMA-amended soils was presented in Table 2. In compared to other cultivars, Zhe 733 grown in Native soil under intermittent flooding had the most significant decrease in grain-As. Among the six cultivars in Native soil under both continuous and intermittent flooding, cultivar selection had at least one significant impact on the uptake of selected elements. Results from this research could provide evidence that cultivar selection alone would significantly affect grain trace element concentration, varying with irrigation management practice.

Data in Figure 1 also showed the effect of water management on grain concentrations of As, Se, Mo, and Cd for Zhe 733 under MSMA-amended soil. In MSMA-amended soil and compared to continuous flooding, intermittent flooding significantly decreased grain-As from 0.60 to 0.22 mg kg-1 (63%) and grain-Mo from 0.64 to 0.36 mg kg-1 (44%), and significantly increased grain-Se from 0.31 to 0.52 mg kg-1 (68%) and grain-Cd from 0.016 to 0.065 mg kg-1 (3.1 times). Therefore, water management resulted in a similar effect on As, Mo, and Cd uptake in both Native and MSMA-amended soil, however, the effect on grain-Se was the opposite. For other five cultivars in MSMA-amended soil and compared to continuous flooding, intermittent flooding significantly decreased grain-As from 1.16 to 0.38 mg kg-1 and grain-Mo from 0.59 to 0.50 mg kg-1, and significantly increased grain-Se from 0.44 to 0.70 mg kg-1 and grain-Cd from 0.021 to 0.050 mg kg-1 (Figure 2). As compared to Native soil, intermittent flooding resulted in a greater decrease in grain-As for MSMA-amended soil. The grain-As concentrations determined in MSMA-amended soil were slightly different from those reported (Williams et al., 2007; Pillai et al., 2010), which was likely due to the cultivar variation used in this study.

The increase in grain-Se as a result of intermittent flooding in MSMA-amended soil in the current research confirms Mikkelsen et al. (Mikkelsen et al., 1989) and constant with Native soil results. Se is most prevalent in the mobile selenate form under aerobic conditions, therefore causing the increase in grain-Se in intermittent flooding (Mikkelsen et al., 1989). The increase in grain-Cd as a result of intermittent flooding discovered in the current research in MSMA-amended soil is consistent with Arao et al. (Arao et al., 2009). This increase in Cd concentrations of rice grain due to aerobic treatments makes it difficult to simultaneously decrease human exposure to both As and Cd elements based on irrigation management.

Among the six cultivars in MSMA-amended soil under continuous or intermittent flooding, cultivar selection had a significant impact on grain-Mo and grain-Cd (Table 2). Compared to other cultivars, GP-2 had the most significant increase in grain-Cd under continuous flooding and MSMA-amended soil, validating that cultivar can significantly alter grain element accumulation varying with water management. This is most likely due to cultivar specific uptake mechanisms. Though the cultivar impact on grain-As, Se, Mo, and Cd in MSMA-amended soil under intermittent flooding was determined to be negligible, Zhe 733 cultivar minimized grain-As in MSMA-amended soil, consistent with findings by Hua et al. (Hua et al., 2011). Overall, this result verified previous findings that cultivar selection is a useful practice to control trace element uptake in rice grain (Jian et al., 2008; Zhang et al., 2008; Norton et al., 2009).

Correlations among trace elements in all six cultivars grown from both Native and MSMA-amended soils were developed. The impacts of grain As uptake on Se, Mo, and Cd concentrations for Zhe 733 cultivar were presented Figure 3. Grain-As and grain-Se showed a positive correlation in Native soil, but a negative relationship in MSMA-amended soil. The correlation between grain-As and grain-Mo was positive in both soils. However, grain-As was negatively corelated with grain-Cd in both soils. Measurements contradicted the findings by Cheng et al. (Cheng et al., 2006) who reported a positive As-Cd correlation in rice root and straw.

The correlations of grain-As with Se, Mo, and Cd for other five cultivars indicated an overall negative trend between grain-As and grain-Se, but with a weak correlation. Similar to Zhe 733 cultivar, a strong positive grain As-Mo trends were found for Cocodrie, Rondo, and Wells cultivars. Dunn and Dunn (Dunn and Dunn, 2012) discovered that plant Mo was negatively correlated with the straighthead score, suggesting that grain Mo decreased with increasing the straighthead score and total grain-As, which contradicted to the overall positive grain As-Mo trend found in this study. A negative grain As-Mo correlation was also reported by Norton et al. (Norton et al., 2012). Strong neutral As-Cd trends were discovered for Cocodrie, Spalcik, Zhe 733 cultivars grown in MSMA-amended soil, and GP-2 in Native soil.