Fundamental to species conservation efforts are an understanding of habitat relationships, development of accurate distribution models, and risk factor assessments. Achieving these tasks is challenging for non-marque stream organisms where limited funding often necessitates compilation of incidental observations from multiple sources which lack an overall sampling design. Compounding matters, appropriate statistical techniques for flow directed networks like streams and the unique forms of spatial dependence that may arise among such observations are necessary. We aggregated a comprehensive presence-absence dataset for Idaho giant salamander (Dicamptodon aterrimus), a species of conservation concern that inhabits mountain streams across a restricted range in western North American and linked these data to geospatial habitat covariates. The dataset was modeled using spatial-stream-network models (SSNM) which account for autocorrelation and results were compared to non-spatial generalized linear models (GLM). The classification accuracy of salamander observations was higher with SSNMs than GLMs (90.8% versus 63.2%) and the spatial models identified fewer significant habitat relationships (four versus seven), which simplified model interpretation. The top-ranked SSNM and GLM were used to predict range-wide occurrence probabilities for scenarios representing historical baselines and futures associated with two significant model covariates (water temperature and riparian tree canopy density) that are changing with environmental trends in the study area. Baseline range estimates from the models were similar (13,090–14,114 stream km) and both predicted small range expansions (2.0% to 24.8%) with warming because many streams were sub-optimally cold for Idaho giant salamander. However, these expansions were partially or entirely offset in future scenarios which included decreases in riparian canopy density. Although the Idaho giant salamander does not appear to be at acute risk, a monitoring program for tracking future changes would be beneficial and could leverage the large dataset compiled for this study as well as spatially-explicit predictions from SSNMs.