Data from: Using spatially-nested hierarchical species distribution models to estimate current and future distributions of a cryptic species at a regional scale

Understanding a species’ conservation status requires evaluating its
ecological relationships, contemporary distribution, and vulnerability to
future environmental change. Species distribution models (SDMs) are widely
used for these purposes, but regional-scale applications often suffer from
extrapolation and niche truncation, reducing model transferability.
Spatially-Nested Hierarchical SDMs (N-SDMs), which integrate data across
multiple spatial scales, offer a promising solution but remain
underutilized in regional conservation research. Crawfish frogs
(Rana areolata) are a cryptic grassland species reliant on crayfish
burrows that have experienced declines across their range and are data
deficient in Oklahoma. This study combines comprehensive regional field
surveys across Oklahoma with large-scale occurrence data from GBIF using
an N-SDM framework to characterize the species’ current regional
distribution, identify factors influencing habitat suitability, and
forecast future range shifts under climate and land-use change.
Additionally, we compared the performance of N-SDMs to regional-only and
rangewide SDMs, and assessed how niche truncation and extrapolation
influence model performance and transferability under future environmental
conditions. We documented R. areolata at 303 survey locations and found no
evidence of historical county-level extirpations, with our models
suggesting large amounts of suitable habitat in eastern Oklahoma. Our
rangewide SDM lacked the resolution and regional predictive performance
necessary for regional conservation planning. While our regional-only SDM
had higher predictive performance, it suffered from substantial
extrapolation and niche truncation, leading to predictions of significant
habitat loss under future conditions. In contrast, our N-SDMs had the
highest regional predictive performance, and mitigated the effects of
niche truncation and extrapolation, and projected no change or a slight
increase in future habitat suitability. Our findings highlight the
advantages of N-SDMs for improving model predictions and informing
conservation assessments. Failure to account for niche truncation and
extrapolation can lead to poor predictions and misguided conservation
decisions. We advocate for the broader adoption of this approach in
regional-scale studies to improve predictions of species responses to
environmental change, and more effectively assess species status at a
regional level.