Effective wildlife monitoring requires innovative methodologies to ensure accurate population assessments for conservation planning. This study focuses on snow leopards (Panthera uncia), a species classified as Vulnerable by the IUCN due to habitat loss, climate change, and over-exploitation. Employing spatial capture-recapture (SCR) models and AI-assisted individual identification, we analyzed camera trapping data from 2012-2013 and 2017-2019 to assess changes in snow leopard density and abundance in Wakhan National Park, Afghanistan. Our analysis used single-session and multi-session models to provide a comprehensive view of population dynamics. Single-session models revealed a substantial increase in density, with the full model estimate rising from 1.38 (0.39 SE) to 3.57 (0.88 SE), while right-only and left-only models showed increases from 1.43 (0.53 SE) to 3.38 (1.01 SE) and from 1.04 (0.36 SE) to 2.21 (0.72 SE), respectively. Our findings also highlight the importance of incorporating bilateral asymmetry into population estimates to avoid overestimation. For single-session models, the full model density estimate was 11.7% higher in 2012-2013 and 27.8% higher in 2017-2019 compared to the average of the left-only and right-only estimates. For multi-session models, the full model density estimate was 13.0% higher in 2012-2013 and 46.9% higher in 2017-2019. This emphasizes the importance of addressing asymmetry to avoid potential overestimation. This study demonstrates the potential of combining advanced analytical frameworks with AI to improve wildlife population assessments. By refining SCR methodologies and addressing key biases, these methods provide critical insights for snow leopard conservation and broader applications to wildlife monitoring. While the results indicate a significant increase in snow leopard populations, continued monitoring and long-term studies are essential to account for changes due to environmental and anthropogenic factors.
Bohnett et al. (2025) studied this question.