Meta-analyses of arsenic accumulation in Indica and Japonica rice grains

Arsenic (As) is a worldwide concern because of its toxic effects on crop yield and prevalence in the food chain. Rice is consumed by half of the world’s population and is known to accumulate As. The present study reviews the available literatures on As accumulation in different subspecies of rice grains (indica, japonica and aromatic) and performs meta-analyses for grain size and texture; these data include 120 studies conducted over the last 15 years across different parts of the world. Aromatic rice varieties accumulate less As with its 95% confidence interval (CI) being 73.90 – 80.94 μg kg-1 which is significantly lower than the As accumulation by either indica or japonica rice varieties with their overall 95% CI being 135.48 – 147.78 μg kg-1 and 204.71 – 212.25 μg kg-1, respectively. Japonica rice varieties accumulate higher As than indica rice grains and within each subspecies polished and/or shorter rice grains accumulated significantly lower As compared to larger and/or unpolished grains; 95% CIs for the polished indica and japonica rice varieties are seen to be 96.33 – 111.11 μg kg-1 and 203.34 – 211.09 μg kg-1, respectively, whereas the same for unpolished varieties are seen to be 215.99 – 238.18 μg kg-1 and 215.27 – 248.63 μg kg-1, respectively. This shows that rice-based As bioaccumulation in humans could be lowered by increased use of aromatic or polished indica rice varieties, followed by the cultivation of shorter polished grains of japonica rice. These findings will be important to inform policy on rice cultivation and dietary uptake of As for a large portion of the global population.


Introduction
Arsenic (As) occurs naturally, in both soil and groundwater. The occurrence and distribution of As in natural environments is a function of geogenic origin and anthropogenic activities; regardless of origin, it poses serious ecotoxicological and human threats. In 2004, International Agency for Research on Cancer (IARC) labelled arsenic as a group-I carcinogenic compound for humans (IARC 2004). According to the World Health Organization (WHO), As levels in groundwater in many parts of the world occur above the permissible limit (10 μg l -1 ). High levels of As in groundwater can result in high levels in crops via irrigation and in humans via polluted drinking water. As contamination affects the food web, eventually creating serious threats for both humans and animals. The hazard index (HI) and estimated daily intake (EDI) of As have been used in numerous studies to assess carcinogenic and non-carcinogenic effects based on contamination of crops and water; results show that about 93% of the cumulative HI of As in crops comes from rice due to its rate of global consumption (Chanpiwat et al. 2019;Mondal et al. 2019;Rezaei et al. 2019;Liao et al. 2018;Ma et al. 2016;Ahmed et al. 2015). FAO, (2017) reported that rice yield/usage is highest in China, followed by India, Bangladesh, America, Indonesia and many other countries throughout the world.
Arsenic is preferentially taken up by rice and accumulates in the grain (50-400 μg kg -1 ) as compared to many other staple food crops (Lin et al. 2013;Su et al. 2010;Meharg et al. Responsible Editor: Elena Maestri 2009;Sun et al. 2008;Williams et al. 2007a). While much paddy rice (rain fed) is grown during the wetter months, there is also a tendency to grow rice using irrigation water during drier periods, such as January-April, in parts of West Bengal, India . In the Ganges Delta of West Bengal and Bangladesh, As ladened shallow aquifers supply rapid and significant amounts of As to paddy fields (Kumar et al. 2016;Neumann et al. 2011). Kumarathilaka et al. (2018a) demonstrated that extensive usage of As contaminated groundwater for rice irrigation increased As levels in paddy soil which subsequently induced As uptake and translocation into rice grain. Arsenic levels in rice grains are also positively correlated with As levels in the soil and levels in rice shoots (Matsumoto et al. 2015;Williams et al. 2007b). According to European Union (EU), the level of As should be restricted to 20000 μg kg -1 in agricultural fields (Hussain et al. 2021;Rahaman et al. 2013;Rahman et al. 2007). Japan recommends that As levels in paddy soil should not exceed 15000 μg kg -1 (Ministry of Environment, Japan, 2021). However, Zhao et al. (2010) noted that the level of As in unpolluted soil ranges from 100 to 10000 μg kg -1 globally, which makes it essential to control the accumulation of As in food crops, including rice cultivated in such As contaminated soil.
Elemental toxicity is often dependant of the biochemical species of a given element (Fitz and Wenzel, 2002). In the case of As, the inorganic form of As (arsenite, As-III)) is more poisonous than its organic form (arsenate, As-V) in terrestrial environments (Das et al. 2018;Suriyagoda et al. 2018). In the rhizosphere, the two-membrane bound silicon (Si) transporters viz., OsLsi1 and OsLsi2 supress the translocation of As (III) in rice shoots (Ma et al. 2008). Previous reports have shown that the rate of As uptake and its accumulation vary in different varieties of rice grains, possibly due to their genetic variations from species to species (Verma et al. 2020;Carracelas et al. 2019;Kumarathilaka et al. 2018b;Sandhi et al. 2017;Bhattacharya et al. 2013;Norton et al. 2009).
Meta-analysis is a specific statistical tool to systematically combine results from several similar primary experimental studies generating an overall conclusion. Such a meta-analysis often provides a more precise treatment effects by pooling the data from several past studies and reducing the biases present in the individual study results; see, e.g. Schwarzer et al. (2015) for more technical details. In biological sciences, when the same treatment (e.g. As level in a particular crop, a specific treatment for a particular disease in plants/animals/human, etc.) has been examined in different studies involving different samples (e.g. over different regions/countries), the results from individual studies often differ; appropriate meta-analysis procedure helps us to statistically combine all these individual results to yield an overall actionable inference. Mandal et al. (2021), used metaanalysis to predict levels of As reserves (200 μg kg -1 in polished rice and 350 μg kg -1 in husked rice) according to joint FAO-WHO Codex Alimentarius Commission in Asian paddy fields.. Some reports used meta-analyses based on sources and distribution of As toxicity to highlight serious health hazards (Shahid et al. 2018) like cardiovascular disease (Moon et al. 2017) and gestational diabetes mellitus (Salmeri et al. 2020). Majumder and Banik, (2019) evaluated the risk (health hazards) of As toxicity based on rice yield, consumption and dietary intake as a function of As occurrence and distribution in areas of Asia, US and Europe via meta-analyses.
However, to the best of our knowledge, there is no metaanalysis of worldwide As accumulation in different types of rice grains. This approach could be very valuable for evaluating the preferred rice types for cultivation in a given Aspolluted field in order to attenuate the flow of As into the food chain. This paper fills this gap by presenting a selective review and the corresponding (global) meta-analyses of grain As accumulation reported for the period 2005-2020 for different rice types produced in selected countries around the world: Australia, Bangladesh, Canada, China, Ecuador, India, Italy, Japan, Malaysia, Pakistan, Spain, Taiwan, Thailand and the USA. We have performed separate meta-analyses for As accumulation in each type of rice grain, along with their statistical comparisons for two major rice subspecies, i.e. Indica and Japonica, with different grain characteristics (longer vs. shorter) and textures (polished vs. unpolished). The Indica rice varieties mostly have longer (long or medium) grains while the japonica rice varieties are mostly short-grained and sticky. Aromatic rice is slightly different from these two subspecies due to genetic variations (Civáň et al. 2019) and are considered a separate group in our meta-analysis as compared to Indica and Japonica. Statistical analyses were used to distinguish between lesser and higher As accumulating rice grains by putting appropriately higher weights to the studies having greater reliability compared to variability, along with their comparisons at the desired significance level. The lesser As accumulating rice grain, as obtained from our results, could be used for future cultivation in worldwide As polluted fields, and also would be effective to store as germplasm for future studies. Further, our results can provide suggestions for farmers to cultivate the less As accumulating rice cultivars in As contaminated soil, although further extensive physiological and molecular based work is needed to understand the underlying mechanisms of As contamination in more detail.

Data sources and quality controls
Literature from Google scholar, ResearchGate, Academia and PubMed on field-based As toxicity in rice were used here. We used the keywords, 'arsenic', 'rice', arsenic toxicity', 'arsenic contamination in soil', 'rice grain', 'paddy soil', 'arsenic accumulation in rice grain', etc., and found about 85 research articles from the year 2005 to 2020. We then narrowed the search using keywords 'arsenic accumulation in rice grain in agricultural fields' and discovered about 20 articles reporting grain As accumulations for different rice varieties. We screened titles and abstracts to sort out about 14 articles presenting 120 studies, after a thorough quality control (described below), which covered As aggregation in several subspecies and cultivars of rice grains from different countries across the globe (see Table 1 for a summary of all the selected studies).
The following quality control criteria were used to ensure the reliability of our subsequent statistical metaanalyses when selecting the 120 studies: • We rejected studies which lacked information (e.g., missing standard error, or having the range of As accumulation only) and/or graphical information (results are shown only in graphs, from which it was not possible to extract mean and standard errors of As accumulations). • We rejected studies for which the reported mean As accumulations were outliers to the pool of selected studies, either due to extreme irrigation conditions or for significantly different cultivars. This analysis has been done in a group-specific way based on the grain sizes and textures (since As accumulation values for longer and/or unpolished grains would look like outlier if considered along with the corresponding values for shorter and/or polished grains). For the sake of completeness, we have reported these studies separately in Appendix Table 3. • Studies with very high standard error (low reliability, mostly due to the pooling of significantly different rice varieties in the same study) were not included; more specifically, we did not include studies having coefficient of variation greater than one. Even if they were included, they would have almost zero weight in our final meta-analysis using the inverse-variance weighting scheme (see Subsection "Statistical analyses" for details of the meta-analysis method used in our paper).

Group assignment
Since grain As accumulation critically depends on the genetic variations of the rice species, we did the metaanalyses separately in the three rice subspecies: indica, japonica and aromatic. Studies on aromatic rice were pooled together in a separate group due to their significant genetic variations from both indica and japonica rice cultivars. Studies conducted on the same rice subspecies, on the basis of the reported names of rice cultivars from different countries, are pooled together to form the three groups and separate meta-analyses were performed within each group to understand the overall level of grain As accumulation and their variations for these three rice subspecies.
Within any rice subspecies, the grain As accumulation can vary significantly based on the size (long or short) and the texture (polished and unpolished) of the grains. For japonica subspecies, further subgroup meta-analyses were performed based on both grain size and texture. Two subgroups of polished and unpolished grains were considered based on the texture; mainly white rice grains have polished texture and brown rice grains are often unpolished (Williams et al. 2005). We also categorized two subgroups based on grain size -the subgroup for longer grain sizes consists of studies with rice cultivars having long, very long or medium sized grains, whereas those with shorter grain sizes are combined in a second subgroup. However, since most indica rice cultivars are composed of longer sizes, we did the subgroup meta-analysis based only on the grain texture (polished vs. unpolished). Subgroup meta-analyses was not performed for aromatic rice group due to lack of subgroup-specific information and/or sample sizes.

Statistical analyses
Pre-processing of data, their arrangement, and basic statistical analyses were performed with Microsoft Excel 2013. Statistical Meta-analyses were performed using RStudio with the R package meta (Schwarzer et al. 2015;Schwarzer, 2007). We used the fixed effect model for the meta-analyses within every subgroup, providing smaller confidence intervals (compared to the random effect models), and the effect sizes were estimated by the 'generic inverse variance method' following Borenstein et al. (2010) where the studies with higher variability (lower efficiency or reliability) get lower weights in estimating the overall effect sizes and the studies with smaller variability get greater weights. Reported means for the grain As accumulations are considered as the treatment estimates, along with their estimated standard errors, since the baseline value of the grain As accumulation should always be zero. Confidence intervals for the estimated treatment effects were obtained based on a normal approximation due to the relatively large number of studies available for each subgroup. Heterogeneity was observed within each subgroup as measured by the between-study variance (τ 2 ), estimated via the classical DerSimonian-Laird estimator (DerSimonian and Laird, 1986). All graphical representations were done using RStudio and Excel 2013.

Results and discussions
The Global Rice Science Partnership, (GRiSP; reported that the majority of the Asian population, about half a billion poor people, use rice as their daily food staple. The ICAR-NRRI Annual Report (2020), notes that 102 million tons of rice were consumed in India during 2018-2019 compared to 146.7 million tons in China during the same period. Beside India and China, many countries rely heavily on rice as a staple: Bangladesh, Indonesia, Philippines, Thailand, and Vietnam. These countries together are responsible for more than 80% of worldwide rice production (ICAR-NRRI Annual Report, 2020). In this study, we have conducted meta-analyses on As accumulation in rice grains from several published studies on indica and japonica rice cultivars; results are summarized in Table 2 for each rice subspecies and their subgroups, where we reported the estimated effect size and 95% CI along with the no. of studies used.

Indica group
The estimated average As accumulation in the grains of indica rice cultivars was (rounded) 142 μg kg -1 and 136 to 148 μg kg -1 at the 95% confidence interval (Table 2). Generally, inorganic As, which is more toxic than organic As, accumulates more in rice plants as compared to many other crops (Chen et al. 2018;Su et al. 2010). Previous studies have reported that the daily intake of As is higher via the consumption of cooked rice as compared to drinking water in some countries (Kumarathilaka et al. 2019;Rahman et al. 2011;Pal et al. 2009;Ohno et al. 2009); this is indeed true for India, especially West Bengal (Halder et al. 2013). According to the World Health Organization (WHO), the permissible limit of As in drinking water is 10 μg L -1 and the provisional tolerable daily intake (PTDI) for human health is 2.1 μg kg -1 of body weight (WHO, 2001). There exists a global threat where daily intake of As exceeds the WHO limit via the consumption of cooked rice grains (BBS, 2011;FAO, 2002). Meharg et al. (2008) and Williams et al. (2005) determined the levels of As in rice in different countries (in increasing amounts): 46 μg kg -1 As in rice from India; 65 μg kg -1 from Canada; 131 μg kg -1 from Bangladesh; 140 μg kg -1 from Thailand; 160 μg kg -1 in polished rice from Italy; 180 μg kg -1 in polished rice from Spain; 280 μg kg -1 in polished rice from France; and 383 μg kg -1 from China. Differences in rice grain As concentrations are highly dependent on genetic factors (Verma et al. 2020;Carracelas et al. 2019;Kumarathilaka et al. 2018b;Sandhi et al. 2017;Bhattacharya et al. 2013;Norton et al. 2009). Rice grain texture (polished vs unpolished) has a significant impact on As accumulation. According to FAO/ WHO and Codex Alimentarius Commission, the maximum permissible limit of As in polished rice is 200 μg kg -1 and in unpolished rice is 300 μg kg -1 (Atiaga-franco et al. FAO, 2014;JECFA, 2012). The European Commission has restricted the level of As in food for children and infants at 100 μg kg -1 (European Commission, 2015) and they note that eating rice grains is the main source of As toxicity in human health (Samal et al. 2010;Lee et al. 2008). This study shows that total As concentrations in the grain of indica polished rice was 104 μg kg -1 (estimated effect size), and as high as 227 μg kg -1 in indica unpolished rice; the respective 95% confidence intervals are 96 to 111 μg kg -1 and 216 to 238 μg kg -1 (Table 2). These data (95% CI) show that As is significantly higher in the indica unpolished rice as compared to the indica polished rice, at the 5% level of statistical significance, which is consistent with earlier studies (Li et al. 2015;Rahman et al. 2014;Meharg et al. 2008). Forest plots of different subgroups of Indica rice (Fig. 1) show that BRRI dhan 32 is a low As accumulator compared to BRRI dhan 30 and BRRI dhan 33, even though these three rice cultivars were developed in Bangladesh. The amount of As accumulation is highly variable for BRRI dhan 28 (from Bangladesh). The majority of Chinese rice, polished subgroup cultivars, are high As accumulators (Fig. 1A), whereas unpolished cultivars from China had greater variability across most studies (Fig. 1B). These Chinese cultivars were obtained from different regions and provinces of China, viz., Anhui, Chongqing, Fujian, Guangdong, Guangxi, Henan, Hubei, Hunan, Jiangxi, Jiangsu, Jilin, Liaoning, Sichuan, Yunnan, and Zhejiang (Li et al. 2015). Differences in As concentrations in rice grain within or between varieties and between different geographies may be due to geogenic settings, bioavailability based on As speciation, the intensity of anthropogenic activities which involve fluxes of As to near surface ecosystems and/or inherent differences in varietal genetics.

Japonica group
Japonica rice is a popular subspecies throughout Eastern and South-Eastern Asia. The most well-known japonica rice varieties are Arborio, Carnaroli, Ribe, Risotto, Roma, Sushi and Vialone Nano, etc., cultivated in Australia, China, Japan, Italy, UK and many other countries (Fransisca et al. 2015;Somella et al. 2013;Padovani et al. 2006). In this study, the overall estimated total As accumulation in the grains of japonica rice cultivars was 208 μg kg -1 and the corresponding 95% confidence interval was 205 to 212 μg kg -1 (Table 2), and therefore higher than Indica rice grain at the nominal 5% level of statistical significance.
The japonica rice group was subdivided with respect to grain texture and size to access their relative As accumulation. The total As level in Japonica polished rice grain was 207 μg kg -1 (effect size in the subgroup meta-analysis) and the corresponding 95% confidence interval was 203 to 211 μg kg -1 ; unpolished japonica rice grain accumulated 232 μg kg -1 As, and the associated 95% confidence interval 215 to 249 μg kg -1 ( Table 2). As with the indica rice, polished grains accumulate significantly less As compared to unpolished grains for the japonica species. According to Li et al. (2015), lower concentrations of As in polished samples is the result of mechanical removal of surface As during the polishing process. Among polished japonica rice, the Arborio rice varieties and a few Australian varieties showed greater variability in As concentrations, and sushi varieties from China and the USA were lower As accumulators ( Fig. 2A). Most white rice cultivars from the USA are high As accumulators, some accumulated greater than 350 μg kg -1 As in their grain. Fransisca et al. (2015) found that Chinese sushi rice accumulated less As compared to Australian sushi rice and higher variability was present for the USA sushi rice. Japonica unpolished rice cultivars have comparatively greater variability in As accumulation than japonica polished rice (Fig. 2B). Among unpolished japonica rice, the levels of As is higher in brown rice cultivars from the USA; which is significant in European countries and the USA, where the average is 140 g of brown rice consumed per day (Williams et al. 2005;Robberecht et al. 2002;Tao et al. 1999).
Subgroup meta-analysis based on grain sizes of japonica rice, show that As accumulation in longer grains is significantly higher than in shorter grains. The level of As in japonica rice varieties with longer grains is (on average) 232 μg kg -1 and its 95% confidence interval is 228 to 237 μg kg -1 (Fig. 3) while in shorter japonica rice grains the level is 163 μg kg -1 with the associated 95% confidence interval being 156 to 169 μg kg -1 (Table 2). These findings are consistent with other work (e.g., Fransisca et al. 2015) which demonstrated that the size of grains correlates with the level of As accumulation. A few cultivars of Australian rice (both white and brown) accumulated relatively less As although the corresponding studies (Fransisca et al. 2015;Rahman et al. 2014) had higher variability in the reported grain As accumulations (Fig. 3A). Most long grained cultivars of US white rice (Williams et al. 2005) had higher levels of As in their grain while cultivars from Italy like Ribe, Roma (Sommella et al. 2013) and a specific Spanish rice (Williams et al. 2005) were less As accumulators. As concentrations varied significantly for cultivars from China (Fransisca et al. 2015), for both shorter and longer grains. Within the pool of japonica rice varieties with shorter grain sizes, Chinese sushi rice accumulated less As whereas Texas and Australian sushi were high As accumulators (Fig. 3B).

Arsenic accumulation in aromatic rice grains
Generally, aromatic rice is medium to long grained. According to Civáň et al. (2019), this type of rice is not identical with indica or japonica rice based on genetic variability. There are many varieties of aromatic rice such as basmati, jasmine, badshabhog, tulsibhog, gobindobhog, radhunipagol, sona masuri, chinigura, kalijira, tulaipanji, etc. In this study, the estimated overall level of As accumulation in aromatic rice cultivars was 77 μg kg -1 with the corresponding 95% confidence interval being 74 to 81 μg kg -1 ( Table 2). The literature notes that aromatic rice varieties exhibit some typical characteristics in terms of amylose content, phenol reaction etc. and this type of rice could be an intermediary rice types between japonica and indica (Bhattacharjee et al. 2002). From our analyses, we found that the kalijira rice cultivars showed greater variability whereas basmati rice cultivars were lower As accumulators (Fig. 4). In general, the As levels do not affect the aromatic rice as much. Islam et al. (2017) demonstrated that the As level in the aromatic rice variety was only 58 μg kg -1 . According to Sandhi et al. (2017), the human consumption of aromatic rice is safer for human consumption due to its potentiality to accumulate less As. Aromatic rice grown in India, Bangladesh, and Pakistan were identified as low As accumulators and hence potentially safer for human consumption globally (Rahman et al. 2014).

A Comparative overview
In order to make an overall comparison of the grain effectsize impact on As accumulation for all three rice species (indica, japonica and aromatic) and their subgroups, we have represented the resulting 95% confidence intervals for each case in Fig. 5 (data from Table 2). It is evident from this comparison that the aromatic rice varieties are significantly lower As accumulators compared to both indica and japonica varieties, at 5% level of significance, and hence they are the safest option for human consumption to avoid As toxicity. Sandhi et al. (2017) also concluded from analysing the accumulation factor of As, that aromatic rice cultivars accumulated less As than other rice cultivars with high yield capacity. However, the consumption of aromatic rice is still significantly low compared to other common rice varieties (including indica and japonica) possibly due to its higher price. Figure 5 also shows that unpolished rice cultivars should be avoided, particularly when there is a potential risk of As contamination due to cultivation practices. Japonica longer grains do not differ significantly from the unpolished rice cultivars in terms of As accumulation, at 5% level of statistical significance; polished rice cultivars are a relatively safe option. Unless shorter grain size, even polished japonica rice cultivars are not necessarily a good option considering their As levels. Conversely, indica polished rice cultivars are far better with much lower As accumulation; consumptions of indica polished rice would be the next best option, after aromatic rice, to combat As toxicity from rice-based products.
Since, As accumulation in rice grains are also correlated with the soil As level, to make the above comparison of As accumulation in different rice groups, any confounding effect of soil As level needs to be taken into consideration. Although, the exact soil As levels were not reported in several of the studies considered in our meta-analysis, it has been generally seen that the soil As level ranges, on average, from 3700 to 120000 μg kg -1 for the studies involving indica rice varieties and from 500 to 42500 μg kg -1 for japonica rice varieties. For example, Rahman et al. (2007) reported the soil As in the studies involving the indica rice cultivars to be 15500 μg kg -1 whereas Fransisca et al. (2015) reported the soil As to be only 500 μg kg -1 in their studies involving japonica rice. In general, among all the studies considered in our meta-analysis, the average Total As accumulation in grain (μg kg -1 ) 58835 Environmental Science and Pollution Research (2023) 30:58827-58840 As level in soil are lower or equal for studies with japonica rice cultivars compared to those with indica rice cultivars. Since, the confounding effect of soil As level on grain As accumulation is believed to be positive, it indeed makes our conclusion even stronger that japonica rice varieties accumulate higher As than indica rice grains mostly due to genetic factors. However, to further understand these issues better, a more detailed studies correlating the exact levels of As in soil and rice grains within these cultivars would be important future research work.
We would like to note that, in contrast to our findings, the study of Roel et al. (2022) reported that total As level was more in indica rice cultivars than japonica rice cultivars in Uruguay due to different soil types.

Conclusion
We conclude that aromatic rice varieties are the lowest As accumulators as compared with both indica and japonica rice varieties. Aromatic rice cultivars are different from indica and japonica rice cultivars on the basis of genetic disparity, aromatic nature and chemical properties which may be the difference in As accumulation. The levels of As were high in indica and/or japonica unpolished rice but low in the case of indica and/or japonica polished rice which is probably due to the reduction of As during the polishing process. Based on literature findings, most indica rice cultivars were medium to long grained; but some japonica rice varieties are short or medium grained. Consequently, we conducted further sub-group analyses of japonica rice varieties based on their grain sizes -longer vs. shorter grains; we observed that the As contents were high in longer grains as compared to that of the shorter grains of japonica rice species as intuitively expected. Some cultivars of sushi rice, mainly from China, were low As accumulators.
Based on comparisons of grain As accumulation across different rice species and their subgroups, we propose the best option for sustainable agriculture to cultivate aromatic or polished indica rice varieties in As contaminated agricultural fields; this would significantly reduce the As uptake by humans from consumption of rice and its products. If required, polished japonica rice cultivars with shorter grain size could be used in As contaminated fields, but the cultivation of unpolished rice cultivars should definitely be avoided if there are potential risk of As contamination from soil and/or ground water during irrigation. However, these conclusions were drawn based only on statistical meta-analysis based on the data from previous studies and so need more laboratory and field experiments before adapting them in practice.
In this respect, we would like to highlight that our present study considered total As values. However, it is well known that the most dangerous component is the inorganic As and there are reported studies that shows that proportion of inorganic As from total is not a fixed value. There can be a cultivar with high total As with low inorganic concentration and also the opposite. So, this issue of inorganic As concentration must be studied in more detail in future works before promoting aromatic and polished Indica rice varieties which turn out to be the best candidate in terms of total As concentrations in our present study.
An unwanted limitation of the present study was the availability of uniform secondary data with detailed information on cultivars, grain types, texture and other morphological features of the rice grains. We tried to classify available studies appropriately as far as possible, which also led to the rejection of several studies that could have been included in our analysis provided they had given the necessary information about their studies. This led to including some studies which were done on mixed cultivars and hence had relatively high variability of As accumulation. It would be important for future work to further refine the collection of studies, with a specific target species or grouping in mind, to create a pool of secondary data having similar characteristics which would lead to further improvement on our results. Similar meta-analyses can also be performed for other crops (wheat, maize, pulses, millets etc.) or on the effects of other hazardous materials (Cd, Pb, Hg, Ba, Cr etc.) and on their amelioration. Appendix A: Studies with extreme/outlying results Authors contributions SD conceptualized, designed, collected the data and prepared the original draft of the manuscript. AG helped in data collection, visualization and validation of statistical part, also helped in reviewing of the manuscript. MAP reviewed and edited the manuscript. PB helped in conceptualization and supervised the work. All authors read and approved the final version of the manuscript.
Data availability Not applicable.

Declarations
• The authors would like to declare that they have no conflict of interests.
• The authors have no relevant financial or non-financial interests to disclose.
• All authors certify that they have no affiliations with or involvement in any organization or entity with any financial interest or non-financial interest in the subject matter or materials discussed in this manuscript.
• The authors have no financial or proprietary interests in any material discussed in this article.
Ethical approval Not applicable.

Consent to participate Not applicable.
Consent to publish Not applicable.