An integrated multi-model perspective for Periphyton-based aquatic ecological assessment to enhance interpretability

As crucial bioindicators, periphyton communities play a significant role in watershed health assessments and ecological restoration. This study investigated periphyton in the mainstem Jialing River and its tributary (Xichong River) during autumn 2022 and spring 2023. By integrating partial least squares path modeling (PLS-PM), random forest models (RFMs), and the neutral community model (NCM), we elucidated the mechanisms underlying aquatic ecological conditions. The results revealed significant spatiotemporal variations in periphyton structure, with diatoms dominating. Sediment conductivity, organic matter, pH, and nutrients were key drivers. α diversity was significantly higher in autumn than spring ( P 0.5). PLS-PM and RFM confirmed that assembly mechanisms strongly explained ecosystem health across seasons and rivers ( P < 0.05). Anthropogenic disturbance promoted α-diversity in spring (SX: β = 0.9224， P < 0.001；SJ: β = 0.9954， P < 0.01), while nutrient influence strengthened from spring to autumn ( β = 0.8655, P < 0.01). In conclusion, seasonal fluctuations in nutrient concentrations and anthropogenic disturbances drove shifts in community assembly mechanisms between deterministic and stochastic processes, thereby shaping the periphyton community structure and ultimately resulting in the spatiotemporal pattern of “mainstem superior to tributary, autumn superior to spring” in aquatic ecosystem health. This study elucidates the reason for differences in watershed aquatic ecological statuses from a mechanistic perspective, providing a scientific basis for precise ecological restoration and management. • Machine learning quantifies spatiotemporal heterogeneity in basin aquatic health. • Validated periphyton metrics enable robust and multidimensional ecological assessment. • Community assembly shifts from spring deterministic to autumn stochastic processes. • Disturbance, nutrients and periphyton show a significant interactive relationship. • This cascade reveals the driving pattern of river aquatic ecosystem health.