Relationships between stream ecosystem properties and landscape composition at multiple spatial scales along a heavily polluted stream in China: Implications for restoration

Restoration in seriously polluted streams has become an urgent issue in developing countries. In this study, we attempted to propose an effective restoration approach in a heavily polluted stream in China with the aid of relationships between ecosystem properties and landscape across multiple scales. Land cover and physical geometry at a fine gradient of spatial scales (doubling 60-960 m and whole catchment) were included as landscape predictive variables, while water quality and macroinvertebrate were involved as ecosystem response variables. The strong relationship between the stream properties and surrounding landscape indicated the stream still maintain its resilience, that is, the capacity to recover from disturbance without changing its fundamental functions and state, even undergoing severe pollution. Further, the primary predictors for water quality were agriculture and urban land, with most effective scale at local 60 m buffer, but independence tests pointed out most correlations between agriculture and water quality were attributed to urban land and spatial covariation. No direct association was detected between macroinvertebrate and surrounding land cover at all scales, but macroinvertebrate exhibited significant relationship with physical variables at in-stream habitat scale. Overall, it was quite possible for this heavily polluted stream to go back to its initial health state if appropriate restorations were taken. The extent and intensity of riparian urbanizations were high priority factors when performing water quality improvement, and increasing habitat complexity and heterogeneity at in-stream scale was crucial for increasing macroinvertebrates diversity.