Distribution models combined with standardized surveys reveal widespread habitat loss in a threatened turtle species

Turtles are among the most threatened vertebrate groups, and reconstruction of population and distributional trends can be important for evaluating their status and defining conservation targets. Using a two-phase modeling approach along with occurrence records and field surveys, our objectives were to: estimate the potential and current distributions of the wood turtle (Glyptemys insculpta), a semi-terrestrial fluvial specialist of range-wide conservation concern; assess how climate, geomorphology, and land-use relate to its distribution; and estimate habitat loss. Using 1061 occurrences in the northeastern US with stream geomorphology and climate data, we created distribution models at sub-regional scales to estimate that 24% of stream km in the region are potentially suitable for wood turtles. Suitable stream habitat is generally lower gradient, higher sinuosity, and higher flow than random, but varies significantly by subregion, suggesting that distribution or habitat models built at the regional or range-wide level may miss important local variation in habitat and climatic needs at finer scales. We then used landscape composition at multiple spatial scales to predict results from standardized field surveys from 78 sites across 9 states to estimate habitat loss. We estimate that 58% of potential habitat is degraded based on landscape composition. Results support previous observations of decline and habitat loss in the wood turtle, driven by landscape degradation at scales larger than the typical home range, necessitating conservation intervention at multiple spatial scales, and underscoring the importance of models that address locally varying and scale-dependent relationships between species and habitats.