Abstract: With increasing discharges of rural domestic sewage and livestock wastewater, eutrophication in water bodies of subtropical hilly regions has become increasingly severe. As an ecological restoration technology, artificial wetlands play an important role in water quality improvement, and benthic animal communities can serve as effective indicators of aquatic environmental health. This study examined artificial wetlands from 2019 to 2022 to compare water quality variation under different vegetation configurations and to analyze the responses of benthic animal communities to water environmental conditions. The results showed that the effluent water quality of artificial wetlands generally met Class Ⅲ to Ⅳ surface water standards. Wetlands planted with Canna glauca and Pontederia cordata exhibited the highest removal efficiencies for total nitrogen (TN), total phosphorus (TP), and chemical oxygen demand (COD), with significantly better performance in summer and autumn than in spring and winter (P < 0.05). A total of 36 benthic species were identified, and both community abundance and biomass showed pronounced seasonal variations. With increasing concentrations of TN, TP, and COD, pollution tolerant taxa became dominant and community diversity declined. In contrast, under improved water quality conditions, the proportion of pollution sensitive taxa increased and the Shannon–Wiener diversity index rose significantly. Structural equation modeling indicated that TN was the key environmental factor driving changes in benthic community structure. The dominance of the pollution tolerant species Limnodrilus hoffmeisteri can be used as an indicator of nutrient pollution and water quality degradation, whereas higher community diversity and an increased proportion of pollution sensitive taxa reflect improved water quality in artificial wetlands. These findings provide a clear biological basis for water quality assessment of artificial wetlands based on benthic animal indicators.