Microplastics drive both gradual and abrupt shifts in soil multifunctionality along concentration gradients.

Microplastics are increasingly recognized as emerging contaminants in terrestrial ecosystems, yet their cumulative impacts on soil multifunctionality remain poorly understood. Here, we evaluated the influence of two microplastic polymers, polyethylene terephthalate and polypropylene, on soil functioning by applying a gradient of 20 concentration levels from 0% to 0.95% w/w, equally spaced by 0.05 increments, to a plant-free soil microcosm experiment. We measured six variables representing soil physical, chemical, and biological functions. A statistical framework combining multi-model inference with threshold-like response detection and exploratory machine learning was implemented to characterize response shapes and identify predictors of soil multifunctional change. Most statistically-supported responses exhibited nonlinear or threshold patterns, primarily in physical properties, indicating that microplastic stress was associated with changes in soil structure, which may have contributed to subsequent changes in multifunctionality. We estimated two threshold-like shifts around 0.3% polypropylene and 0.55% polyethylene terephthalate w/w, which should be interpreted as approximate transition regions under controlled conditions rather than ecological tipping points. Random forest highlighted water-stable aggregates as the strongest predictor of soil multifunctionality under PP. Our findings provide experimental evidence of complex soil multifunctionality responses to microplastic pollution, with physical deterioration as a possible early-warning signal of microplastic disturbance.