Microplastics (MPs) are persistent contaminants that accumulate in lake sediments, yet the long-term dynamics and driving mechanisms of this accumulation in shallow lakes remain poorly understood. Here, we reconstruct 80-year MP depositional histories from three sediment cores in Baiyangdian Lake, a large shallow lake in North China experiencing intensive anthropogenic disturbance. By integrating Rate of Change analysis, coupling models, Partial Least Squares Path Modeling, and random forest modeling, we identify two critical transitions in MP accumulation: an initial increase triggered by reduced hydrological connectivity following dam construction in 1963, and a subsequent accelerated phase driven by rapid socioeconomic development since ∼2000. Spatially, MP abundance exhibits marked heterogeneity (up to 37,230 items kg⁻¹ near inflow rivers vs. 3649 items kg⁻¹ in a nature reserve). Random forest modeling reveals that nutrient enrichment (TOC, TN, TP) and hydrodynamic intensity jointly govern this spatial heterogeneity. Our findings demonstrate that microplastic accumulation in shallow lakes is regulated by a dual mechanism-hydrological connectivity controlling retention capacity, and socioeconomic development controlling input intensity-with strong coupling between MPs and nutrients. The post‑2010 slowdown in accumulation rates shows that policy interventions can leave measurable imprints, while persistently high levels in semi‑enclosed zones highlight the need for targeted remediation. The proposed "switch‑tap" model provides a framework for lake management: effective mitigation requires both source control and, where feasible, restoration of hydrological connectivity.