Analytical methodology as an overlooked key driver in predicting soil microplastic abundance across China.
Abstract
Differences in microplastic (MP) analytical methods represent a major but largely overlooked source of uncertainty in soil MP studies. Here, we developed a Geo-detector-aided machine learning framework that explicitly integrated methodological variables with environmental factors (i.e., soil properties, climate, and anthropogenic activities) to predict soil MP abundance across China. Analytical methodology was the dominant factor influencing the predicted MP abundance, with identification methods ranking first and extraction method ranking sixth among all the predictors. Different analytical method combinations resulted in remarkable difference in the national mean soil MP abundance, with the lowest (589 items·kg⁻1) being only half of the highest (1176 items·kg⁻1). The methodological divergence exhibited obvious spatial heterogeneity, being most pronounced in southern China due to its complex MP sources and humid climate that enhances the MP-soil interactions. This divergence further propagated into human exposure assessments, with predicted daily soil MP exposure for children ranging from 2.30 × 10⁻3 to 5.05 × 10⁻3 items·kg BW⁻1·d⁻1 under the lowest- and highest-yielding method combinations. These findings demonstrated that analytical workflows shaped both national-scale soil MP abundance and downstream exposure estimates. It also emphasized the need for standardized MP analytical methods that explicitly consider regional variation in soil properties and climatic conditions.