Drying in stream ecosystems is consequential: it alters local community composition by filtering sensitive taxa, and disrupts metacommunity dynamics by fragmenting and isolating stream habitats. Although ecological theory predicts that drying location (e.g., headwaters vs. lower reaches) should be consequential, limited empirical work has tested this hypothesis. Here, we sought to determine (i) how drying acts at the metacommunity scale through network-wide fragmentation and local disturbance, (ii) how network-level responses change as a function of where drying occurs (i.e., predominantly in headwaters vs. lower reaches), and (iii) whether the relationship between drying spatial structure and community diversity is contingent on climate. To this end, we generated networks that dry predominantly in the headwaters, in the lower reaches, or randomly. Using a full factorial design combining the different drying spatial structures, fragmentation levels, and disturbance severity, we then simulated stream invertebrate metacommunities. We found that the spatial signature of drying altered the effects of fragmentation and local disturbance. Fragmentation had strong effects in all spatial patterns, but downstream drying uniquely reduced alpha diversity across all connectivity levels. Local disturbance only affected sites with extremely low connectivity, and had the greatest effects in upstream-drying watersheds. Our results predict that drying has the greatest watershed-wide effects when it is mainstem biased. Results from the modeled metacommunities are being compared to statistical relationships based on ten metacommunities sampled across the U.S., spanning from mesic to arid climates. Our results show how biodiversity responses to drying stress and disturbance depend on dominant climate and drying signatures. From an applied perspective, understanding how drying carries different consequences depending on where it occurs could help inform and prioritize conservation and restoration efforts at the watershed scale.