Prioritizing stream network connectivity for species conservation and sustainable planning

Ecological connectivity is crucial for amphibian conservation amid habitat fragmentation from human development. This study assessed stream network connectivity for two stream-breeding amphibians, Rana sauteri and Odorrana swinhoana , across an entire river catchment using the MaxEnt modeling framework. Multiscale stream order predictors represented stream network spatial structure across terrestrial habitats, while citizen-science occurrence and roadkill data serve as proxies for distribution and movement, enabling model construction and cross-validation without direct movement data. Most models performed significantly better than random expectations (p < 0.05–0.001), with high predictive power (AUC = 0.79–0.93; CBI = 0.74–0.97). Tree cover dominated predictor importance (41–70%), while stream order contributed substantially (17–23%) and showed strong predictive power, underscoring the importance of stream networks. R. sauteri responded at broader spatial scales across both occurrence and roadkill models (stream order 50–750 m; tree cover 150–2000 m), whereas O. swinhoana responded at intermediate scales (150–500 m; 250–1500 m), reflecting differences in habitat preferences and breeding behavior. In cross-dataset validation, occurrence models effectively predicted roadkill distributions (AUC = 0.88–0.92), while roadkill models transferred less successfully (AUC = 0.68–0.81), confirming their complementary ecological signals. These results demonstrate that stream order effectively captures the spatial hierarchy of amphibian connectivity across an entire catchment. Our framework enables connectivity analysis in data-limited contexts. The approach can be incorporated into Environmental Impact Assessments to support environmental planning and design by identifying scales of connectivity influence, altitudinal ranges, and topographic features such as valley-bottom zones where habitat suitability and roadkill risk overlap. Ultimately, this approach promotes stream network connectivity that sustains the movement and persistence of multiple taxa, contributing to biodiversity conservation and sustainable development.