Exploring the ecological response of fish to flow regime by soft computing techniques

Ecosystem integrity has become a focus of watershed management. Recently much attention has been paid to the analysis of eco-hydrological data for developing a better understanding of how flow regimes influence the structures and functions of riverine ecosystems. This study explores ecosystem integrity in Taiwan rivers through characterizing the relationship between flow regimes (long-term continuous hydrological observational datasets) and fish communities (non-continuous ecological monitoring datasets) using soft computing techniques (self-organizing feature map (SOM) and clustering methods). The analysis is driven by the Taiwan Eco-hydrologic Indicator System (TEIS) statistics and biodiversity measures, in which data sets consist of 74 long-term monitoring records of flow and fisheries at 38 locations in 12 rivers in Taiwan. Results indicate that: (1) flow regimes of the 12 rivers can be related to watershed locations, namely upstream, midstream and downstream regions; (2) fish diversity is significantly higher in midstream regions than in up- and downstream regions; and (3) a consistent relationship between flow regime and fish community can be obtained when using presence-abundance data. Flow regime characteristics are consistent with respect to the hydro-geographical and ecological characteristics of streams, especially for midstream reaches, which provides further evidence of the importance of natural flow regimes in sustaining ecological integrity. These analysis results offer valuable insights into management options that support decision-makers in evaluation and classification of ecological status and provide further analyzed evidence to address ecological flow regime issues in water resources management.