Climate change is altering the structure and dynamics of high mountain stream ecosystems globally, but multiple mechanisms drive biodiversity change that are difficult to isolate. Nested sampling designs with replication across multiple spatial scales, from microhabitats to whole watersheds, can help disentangle ultimate from proximal controls of community change. Here we investigated the potential drivers of macroinvertebrate community composition change in a California Sierra Nevada watershed during a severe supraseasonal drought (2019-2022), to assess (i) the mechanisms whereby climate change may alter communities, and (ii) the spatial scales and dependencies of these mechanisms of community change. We sampled 60 sites in Bull Creek following a nested spatial design that captured microhabitat, pool-riffle, and reach-level variation within the watershed. We also deployed 120 water and air temperature sensors to estimate future thermal vulnerability based on projected climate change in the watershed. We found that mainstem reaches experienced less variation in community composition than headwaters, likely due to mass effects influencing habitats in the lower network positions. Maximum water temperature, water velocity, and fine sediments all explained community variation within reaches, with pools resembling each other more than riffles. Highest diversity typically occurred in mainstem riffles that were both warm and had minimal fine substrate. Lastly, stream thermal sensitivity and downscaled projections of summer air temperature predict that coldwater habitat may decline by 25.5% by the end of the century, increasing organismal metabolic rates and thermal stress. Our results illustrate the multiple interacting mechanisms (thermal, hydrologic, and geomorphologic) linking climate change to biodiversity change in high-mountain streams. If climate change continues to reduce average snowpack and summer flow in streams, reducing water velocity and increasing fine sediment accumulation, loss of coldwater specialists and high rates of community replacement may be expected.