Ecological integrity assessment and degradation diagnosis are used globally to evaluate the health of water bodies and pinpoint critical stressors. However, current studies mainly focus on separate evaluation or diagnosis, leading to an inadequate exploration of the relationship between stressors and responses. Here, based on multiple data sets in an urban lake system, a synchronous evaluation-diagnosis model with quantitative stressor-response analysis was advanced, aiming to improve the accuracy of evaluation and diagnosis. The weights for key physicochemical stressors were quantitatively determined in the sequence of NDAVIadj > CODMn > TP > NH4+-N by the combination of generalized additive model and structural equation modeling, clarifying the most significant effects of aquatic vegetation on the degradation of fish assemblages. Then, sensitive biological metrics were screened by considering the distinct contributions of four key stressors to alleviate the possible deviation caused by common methods. Finally, ecological integrity was evaluated by summing the key physicochemical stressors and sensitive biological metrics according to the model-deduced weights instead of empirical weights. Our system's diagnosis and evaluation results achieved an accuracy of over 80% when predicting anthropogenic stress and biological status, which highlights the great potential of our multiple-level system for ecosystem management.
Read full abstract