Geomorphic setting and seasonal variation shape benthic macroinvertebrate communities, but their effects on community patterns and assembly processes remain unclear. We surveyed 49 sites across seven Sichuan rivers during spring and autumn 2025. Reaches were classified as Mountain, Hilly, or Plain. Principal coordinate analysis (PCoA), random forest models, a neutral community model (NCM), and the modified stochasticity ratio (MST) examined community patterns, predictors, and assembly processes. Diptera had high relative abundances in Mountain and Hilly reaches in spring, whereas the proportion of Ephemeroptera increased across reach types in autumn. Taxa richness and Shannon and Margalef indices were higher in Hilly and Plain reaches. Community composition differed significantly among reach types, with reach type explaining 10.3% and 11.4% of variation in spring and autumn, respectively. For Ephemeroptera, Plecoptera, and Trichoptera (EPT) richness, random forest models ranked dissolved oxygen, channel width, elevation, and ammonia nitrogen highly in spring, and the permanganate index, dissolved oxygen, and flow velocity highly in autumn. NCM fit and Nm were higher in spring and Plain reaches. Mean MST values below 0.5 suggested deterministic dominance, with the highest estimated deterministic contribution in Mountain reaches (80.97%). These findings highlight seasonal differences in benthic responses to environmental gradients and potential stressors and may inform river bioassessment that accounts for geomorphic and seasonal context.

With the continuous expansion of hydropower development in China, basin-scale river regulation has gradually evolved from the operation of single diversion dams to coordinated cascade regulation of multiple dams. Nevertheless, the mechanisms underlying cumulative disturbance induced by cascade diversion dams in riverine benthic ecosystems remain poorly understood. Taking the Baoxing River Basin, a mountainous headwater basin in Southwest China, as the study area, this study analyzed the coupled response characteristics among hydrology, water quality and benthic index of biotic integrity (B-IBI), based on spring-autumn field monitoring data from 15 sampling sites and discharge records of upstream diversion dams during 2020–2024, so as to evaluate the comprehensive benthic ecological impacts of cascade diversion dams. The results showed that: (1) Cascade regulation was associated with spatially heterogeneous hydrological conditions. The mean cumulative number of upstream hydropower stations was 0.8 in the upstream section and 11 downstream, while mean nearest upstream releases in the downstream section were 12.49 m3·s−1 in spring and 9.49 m3·s−1 in autumn. (2) Driven by such hydrological gradients, the structure of benthic macroinvertebrate communities shifted: the proportion of sensitive EPT (Ephemeroptera, Plecoptera and Trichoptera) taxa decreased, while Diptera and other tolerant taxa became dominant. Correspondingly, the mean B-IBI declined by 28.6% from 4.39 to 3.13. (3) Random forest analysis identified turbidity as the strongest predictor of B-IBI, followed by hydrological regulation and other environmental variables. (4) Covariance-based structural equation modeling (CB-SEM) revealed season-specific patterns of association in separately fitted spring and autumn models. The spring model highlighted associations involving turbidity, dissolved oxygen, and biomass, whereas the autumn model emphasized physicochemical conditions and tolerance-related community attributes. Overall, this study provides insights into the associations among hydropower release, water-quality conditions, and seasonal benthic macroinvertebrate responses under cascade hydropower regulation. These findings provide scientific support for ecological flow management and adaptive ecological assessment of mountain rivers subjected to cascade hydropower impacts.