Predicting the effect of land use and climate change on stream macroinvertebrates based on the linkage between structural equation modeling and bayesian network

Land use and climate change are increasingly important stressors affecting freshwater ecosystems. Although their effects on freshwater organisms have been widely studied, most studies applied traditional statistical methods, which only focus on single stressor and simple cause-effect relationships without considering the complex interactions among stressors. Therefore, we developed an integrated method by combining Structure Equation Modeling (SEM) and Bayesian Networks (BNs) to estimate the interactive effect of land use and climate change on freshwater macroinvertebrates. A field investigation was conducted in August 2009 in Taizi River Basin, Northeast China, and samples of stream macroinverterbates and water chemistry were collected from 211 sites. The SEM-BN models were developed to explore the complex relationships among land use cover (crop, forest and residential land), water quality (total phosphorus, total nitrogen and dissolved oxygen), physical factors influenced by climate change (water temperature, flow velocity), other habitat characteristics (slope and substrate composition) and macroinvertebrate EPT (Ephemeroptera, Plecoptera and Trichoptera) indices (the percentage of EPT taxa, EPT richness, EPT abundance and the Shannon-Weiner Diversity Index of EPT). Three scenarios were designed to assess the possible responses of EPT indices to land use change, climate change and their interactions. Our results showed that when the change of land use and climate were considered alone, increasing crop and urban land led to declines in EPT indices whereas moderate rise of air temperature and more rainfall had opposite effects. However, the combined effect showed that the positive effects caused by climate change could weaken some negative effects, but land use change still had stronger effect on EPT indices. Our results provided more detailed understanding on how environmental stressors affect freshwater organisms, and further catchment management should integrate the combined effect of different environmental stressors on streams.