Stochastic community assembly drives tilapia invasion in a subtropical reservoir revealed by environmental DNA

<jats:p>Understanding whether fish community assembly in large subtropical reservoirs is governed by deterministic or stochastic processes is critical for predicting and managing biological invasions. However, the mechanisms underlying tilapia invasions in highly regulated reservoir ecosystems remain poorly understood. This study aimed to quantify the relative contributions of stochastic versus deterministic processes to fish community assembly and to evaluate the niche characteristics and overlap patterns of three invasive tilapias (Coptodon zillii, Sarotherodon galilaeus, and Oreochromis niloticus) with native fishes. We employed environmental DNA (eDNA) metabarcoding to survey fish communities across four longitudinal zones, two microhabitats, and three water layers in Shuikou Reservoir, China, during July–August 2025. Among 55 detected fish species, C. zillii (21.73% of relative read abundance) and S. galilaeus (21.22%) dominated. Neutral and null model analyses revealed that stochastic processes—primarily ecological drift—dominated community assembly across all spatial dimensions, accounting for 66.04%–95.94% of assembly processes. Coptodon zillii (B′ = 0.6095–0.7919) and S. galilaeus (B′ = 0.5056–0.9465) exhibited moderate to broad niche widths, while O. niloticus showed restricted distribution (B′ = 0.3575). Despite stochastic dominance, extensive and statistically significant niche overlap occurred between invasive tilapias and native species, with particularly high overlap between C. zillii and S. galilaeus (Ojk = 0.6839–0.9784). These findings challenge traditional deterministic risk assessments that rely heavily on niche-based predictions and underscore the need to integrate stochasticity into invasion forecasting and management strategies for regulated reservoir ecosystems.</jats:p>