ContextGlobal climate change poses a significant threat to the habitat connectivity of cold-water-adapted organisms, leading to species extinctions. If gene flow can be modeled by landscape variables, changes in connectivity among populations could be predicted. However, in dendritic and heterogeneous stream ecosystems, few studies have estimated the changes in gene flow from genetic data, in part due to the difficulty in applying landscape genetics methods and accessing water temperature information.ObjectivesInferring the determinants and future changes of the gene flow in the cold-water adapted fluvial sculpin Cottus nozawae using a recently developed model-based riverscape genetics technique and a hydrological model for estimating water temperature.MethodsThe strength of gene flow on each stream section was modeled by watershed-wide riverscape variables and genome-wide SNP data for C. nozawae in the upper reaches of the Sorachi River, Hokkaido, Japan. Future changes in gene flow were inferred by this model and hydrologically estimated water temperatures under the high greenhouse gas concentration scenario (IPCC RCP8.5).ResultsStream order, water temperature, slope, and distance were selected as riverscape variables affecting the strength of gene flow in each stream section. In particular, the trend of greater gene flow in sections with higher stream order and lower temperature fluctuations or summer water temperatures was pronounced. The map from the model showed that gene flow is overall prevented in small tributaries in the southern area, where spring-fed environments are less prevalent. Estimating future changes, gene flow was predicted to decrease dramatically at the end of the twenty-first century.ConclusionsOur results demonstrated that the connectivity of cold-water sculpin populations is expected to decline dramatically in a changing climate. Riverscape genetic modeling is useful for gaining information on population connectivity that does not fully coincide with habitat suitability.