Predicting soil neonicotinoid content in agricultural landscapes using indirect indicators

Neonicotinoid insecticides are a major driver of pollinator decline. Due to their persistence and mobility in soil, they can contaminate non-target vegetation through runoff or dust, reaching pollinator resources. However, predicting soil contamination is challenging, especially where pesticide use data is lacking. This study assessed the potential of using proxies such as cropping history and landscape structure to predict neonicotinoid content in soils. We analyzed seven neonicotinoids in 86 sites in agricultural landscapes of Belgium. Neonicotinoids were detected in 78 % of sites, with imidacloprid and clothianidin being the most frequently detected (up to 16.3 µg kg⁻¹). In 69 % of sites, contamination occurred without recent recorded treatment and for 33 %, contamination occurred with no recorded treatment history. Risk exposure through soil contact was often high, particularly for clothianidin (44 % of hazard quotient values >1) and imidacloprid (19 %). In potentially treated sites, the total number of treatments was a better predictor of contamination than time since last potential application. Landscape structure poorly predicted contamination, whether when reflecting subsurface water flow or dust dispersion in sites without recorded treatment history. Our research shows that predicting soil contamination at large scales can be approached using treatment history but may be difficult without accounting for contamination in non-arable sites. Further research is needed to inform agri-environmental policies with realistic contamination patterns.