Deriving arsenic concentration guideline values for soil and irrigation water for rice cultivation

Mandal, Jajati

Abstract

Arsenic (As) is a naturally occurring, toxic trace element that can be found in irrigation water, soil, and crops. Rice accumulates higher concentrations of As in its grains than other cereals like wheat and barley. This leads to concern over dietary As exposure, especially in areas of India, Bangladesh, Nepal, Taiwan, Vietnam, and Thailand. This present study has been undertaken to manage the risk posed by rice grown in As-contaminated areas. The objectives of the study were to determine guideline values for total and bioavailable As in soil, as well as As levels in irrigation water, using predictability models. Additionally, the efficacy of biochar as an amendment for As-contaminated soils in rice cultivation was assessed through a meta-analysis.
Meta-analysis of a database compiled from an extensive literature review was undertaken using decision tree (DT) and logistic regression (LR) machine learning models to evaluate the relationship between As concentrations in rice grain, soil, and irrigation water. Soil total As was a stronger predictor of As in rice grain than irrigation water As. Both the DT and LR models successfully predicted the soil concentrations above which As in grain would exceed the Codex recommendation. Subsequent field studies in West Bengal, India in 2021 provided validation data, which demonstrated that 14 mg kg-1 of total As in soil was an appropriate guideline value for the safe cultivation of rice.
The concentration of bioavailable As in paddy soil was predicted using random forest (RF), gradient boosting machine (GBM), and LR models. The LR model was the better performing, identifying bioavailable As, total As, available iron (Fe), and organic carbon as significant predictors of grain As. Based on the LR model's partial dependence plots and individual conditional expectation plots, 5.70 mg kg−1 was the limit for bioavailable As in soil.
An incubation study was conducted using monolithic soil columns collected from 10 As-contaminated sites. Results were analysed using linear discriminant analysis (LDA) and LR, considering the As dose, soil pH, organic carbon, clay, available Fe, phosphorus, and total As. The LR model performed best, predicting 190  L-1 as the guideline value for irrigation water.
To support remediation of As-contaminated soil, biochar was evaluated as a potential soil amendment. A meta-analysis indicated that biochar could be an effective tool in the sequestration of As in soil, but further research is required under realistic field conditions.
This study has provided the first guideline values for As in soil and irrigation water and identified a potential management option (soil remediation using biochar). The findings have direct relevance to rice farmers and regulators, with the potential to deliver significant public health benefits in As contaminated regions.

Citation

Mandal, J. (2023). Deriving arsenic concentration guideline values for soil and irrigation water for rice cultivation. (Thesis). University of Salford

• Thesis Type: Thesis
• Deposit Date: Jul 12, 2023
• Publicly Available Date: Oct 30, 2023
• Keywords: arsenic, rice, soil, irrigation water, machine learning, logistic regression, random forest
• Award Date: Sep 29, 2023