Habitat models that can predict the distribution of river organisms along environmental gradients are useful for basin environmental management. However, the effects of dams has not been considered in habitat models. In this study, we calculated dam indices representing the effects of dams based on GIS data and an Indicators of Hydrologic Alteration (IHA) based on a distributed hydrological model in the Omaru River catchment in southern part of Japan, and constructed species distribution models of stream animals using multiple machine learning methods (random forests, XGboost, LightGBM) and these new environmental metrics as predictor variables. We acquired seasonal fish distribution data using environmental DNA while invertebrates were sampled using a Surber sampler. By optimizing the number of tree in the two gradient boosting methods for each species, the prediction accuracy of the target species greatly improved. In general, important variables for predicting the distributions of invertebrates and fish were IHA and dam indices, respectively. Where dam indices and IHA were included into the predictor variables, the accuracy generally improved for the two life types (synonymous with habit), attachers and crawlers in invertebrates, and for the benthic fish species Pseudogobio esocinus. The number of dams per catchment area was an important variable for both invertebrates and fish, suggesting that the cumulative effect of multiple dams determines the distributions of many aquatic species.