Analysis of Specific Habitat Conditions for Fish Bioindicator Species Under Climate Change with Machine Learning—Case of Sutla River

In studies of potential climate change (CC) impacts on freshwater ecosystems, water temperature is a primary abiotic factor. Still, it is insufficient to describe the specific habitat conditions that have changed for the biological elements of water quality affecting fish. In this study, special attention is focused on the fish bioindicator species, Barbus balcanicus. For two future scenarios of CC impact (RCP4.5 (2020–2050) and RCP8.5 (2070–2100)), in a Sutla River water body case study, fish life stage models are developed based on the fundamental abiotic factors (water flow, depth, velocity, temperature, and dissolved oxygen) to describe the ecological requirements of the selected fish bioindicator species. Two future CC impact scenarios and their results—water flow, dissolved oxygen, and nutrients, prepared by SWAT—have been analysed. To determine the most important abiotic factors, for water temperature, depth, and velocity, models have been developed by the machine learning tool Weka. The modelled biological elements of water quality were combined with previously calculated dissolved oxygen, flow, and E-flow values during dry periods and the spawning period. For both selected CC scenarios, the results indicate that in approximately 60–70% of the life stages of the bioindicator species Barbus balcanicus, the conditions are acceptable.