Abstract

Simple SummaryFloodplain lakes are biodiversity hotspots, yet understanding how water level changes shape species communities remains limited. Using eDNA technology and sonar, we studied species in China’s largest freshwater lake without catching them. Results revealed that random chance largely determines which species live in the lake overall, but water depth acts as a local filter that sorts species by body size and ecological roles. This means protecting lake biodiversity requires two things: maintaining natural water level fluctuations and preserving areas with varying depths. Managers can use these insights to set better environmental flow targets and design more effective conservation strategies for floodplain lakes worldwide.Hydrological fluctuations drive community dynamics in floodplain lakes, yet their integration into water resource management remains limited. Here, we integrated environmental DNA (eDNA) metabarcoding with hydroacoustic surveys to investigate vertebrate community assembly in China’s largest freshwater lake (Poyang Lake) during the winter dry season. We detected 65 vertebrate species, with Cypriniformes dominating. Beta-diversity partitioning revealed that turnover dominated taxonomic and functional dissimilarity, while phylogenetic beta diversity was characterized by nestedness, which is consistent with environmental filtering. Functional richness declined with water depth, coinciding with hydroacoustic vertical size stratification, indicating niche partitioning along depth gradients. Null model analysis showed stochastic processes (ecological drift) dominated regional assembly (72.97%), whereas joint species distribution modeling attributed explained variation to environmental factors (28.9%, notably water depth) and species associations (29.7%) at local scales. This hierarchical framework, regional stochasticity shaping the species pool and local deterministic filtering structuring communities, reframes environmental flow conceptualiz