Multiscale habitat modeling to improve spatial prioritization for mugger crocodile conservation in riverine landscapes

Assessing the species distribution and their detectability is crucial, holding wide-ranging implications for effective conservation planning and management initiatives. Although species-habitat relationships are inherently scale-dependent, few studies apply robust multivariate approaches to optimize spatial scale selection. We developed a scale-optimized habitat suitability model for the mugger crocodile (Crocodylus palustris) using binomial generalized linear models, evaluating each predictor across multiple spatial scales within the Cauvery River Basin (CRB) in southern India. Model selection based on the lowest Akaike information criterion scores identified the multiscale modeling as the best performing approach. Most predictors showed the strongest associations at finer (500 m) to moderate (1000-2000 m) scales, while a subset of topographic and hydrological variables was retained at broader scales (8000 m), showing the importance of incorporating scale heterogeneity in riverine species modeling. Key variables influencing the potential distribution of muggers across the CRB include isothermality, radius of gyration area-weighted mean of wetland, distance to roads, and terrain wetness, indicating mugger prefer stable temperature, low disturbance, and localized patches but well-distributed wetland habitats. The multiscale model estimated 2209.5 km² of potentially suitable habitat across the CRB of which only 38.12% lies within the existing protected area network. Integrating the best performing model into a systematic conservation planning framework that maximizes species target while minimizing human impacts, the solution identified 990 km² of priority regions, including five high-priority areas with a total area of 540 km², outside the current protected network. The study offers a robust and resource-efficient approach to habitat delineation and conservation prioritization, improving the performance of suitability modeling across spatially varying environmental factors.