Reliable model established depending on soil properties to assess arsenic uptake by Brassica chinensis

Generally, prediction of arsenic (As) bioavailability, mobility and its transfer from soil to plant is very important with respect to management of environment and food safety. In this study, pakchoi (Brassica chinensis) was sown in a greenhouse to evaluate the As transfer characteristics from different soils to plant system, and to investigate the possible prediction equations and key factors involved in As bioavailability. The results showed that As uptake of plant and soil As concentration was significantly and positively correlated (R2 = 0.778; P < 0.01). A log-transformed data provided a better correlation (R2 = 0.901; P < 0.01). Results obtained from stepwise multiple linear regression (SMLR) showed that soil pH and total As were important variables involved in the contribution of As transfer to plant. The As accumulation in plant exhibited a positive correlation with soil As content and pH. Various prediction equations were obtained from different As sources, whereas the most favourable equation was screened by root mean square error (RMSE) between the measured and predicted Log [plant As] content. The prediction model (Log [plant As] =1.34 Log [soil As] +0.18pH-1.25) showed the greatest accuracy of R2 = 0.978 and RMSE = 0.11, by combining the data of three As treatments (45 observed data points). These current findings are quite useful and could be used for predicting the As transfer from soil to plant system.