Factors affecting the presence of Arctic charr in streams based on a jittered binary genetic programming model

• A novel evolutionary model (JBGP) was developed. The JBGP classifies the presence of Arctic charr in Teno River catchment. is superior to jittered decision tree model. Macroinvertebrates, mires and slope are dominant indicators. one fish species most sensitive climate change but studies on their freshwater habitat preferences limited, especially fluvial environment. Machine learning methods offer automatic objective models for ecohydrological processes based observed data. However, i) number ecological records often much smaller than hydrological observations, ii) measurements over long-term costly. Consequently, datasets scarce imbalanced. To address these problems, we propose binary genetic programming detect parameters affecting occurrence across tributaries within large subarctic catchment, northernmost Finland Norway. We quantitatively assessed accuracy proposed compared its performance with that classic (GP), (DT) state-of-the-art jittered-DT methods. achieves highest total classification 90% a Heidke skill score 78%, showing superiority counterparts. Our results showed factors contributing include higher density macroinvertebrates, lower percentage catchment iii) milder stream channel slope.