Predictive modeling of plant uptake of Pb and Cd: Implications of aerial deposition and the origin of parameterisation data

Abstract

We developed ordinary least squares regression models to predict uptake of cadmium and lead, two metals that are of public health significance because of their toxicity, in the edible tissues of lettuce. Models were parameterised using data on soil metal concentration, pH, and organic carbon. To assess the impact of physical contamination in form of aerial deposition and soil-splash on the metal concentration in lettuce, separate linear regression models were parameterised for indoor- and outdoor-grown lettuce, assuming the physical contamination to be negligible for indoor conditions. Both Cd models showed high model fit and strong predictive performance, when tested on an independent dataset, suggesting uptake via roots to be dominant. For Pb, the indoor model performed better than the outdoor model, indicating that physical contamination, contributes significantly to metal concentration in lettuce leaves. Our results highlight the importance of the parameterisation data when developing uptake models for predictions and risk assessment. Regression models for predicting Pb concentration in lettuce based on indoor data should not be used for predicting lettuce concentrations cultivated in outdoor conditions unless the contribution of physical contamination is explicitly accounted for.

Keywords

• Urban gardening
• Contamination routes
• Metal uptake
• Predictive regression models
• Model parameterization
• Model evaluation