Combining literature-based and data-driven fuzzy models to predict brown trout (Salmo trutta L.) spawning habitat degradation induced by climate change

Abstract

A fuzzy rule-based system combining empirical data on hydraulic preferences and literature information on temperature requirements was used to foresee the brown trout (Salmo trutta L.) spawning habitat degradation induced by climate change. The climatic scenarios for the Cabriel River (Eastern Iberian Peninsula) corresponded to two Representative Concentration Pathways (4.5 and 8.5) for the short (2011–2040) and mid (2041–2070) term horizons. The hydraulic and hydrologic modelling were undertaken with process-based numerical models (i.e., River2D© and HBV-light) while the water temperature was modelled by assembling the predictions of three machine learning techniques (M5, Multi-Adaptive Regression Splines and Support Vector Regression). The predicted rise in the water temperature will not be compensated by the more benign lower flows. Consequently, the suitable spawning habitat will be reduced between 15.4–48.7%. The entire population shall suffer the effects of climate change and will probably be extirpated from the downstream segments of the river.