The influence of coarse-scale environmental features on current and predicted future distributions of narrow-range endemic crayfish populations

Authors


Shannon K. Brewer, U.S. Geological Survey, Oklahoma Cooperative Fish and Wildlife Research Unit, Oklahoma State University, 007 Agriculture Hall, Stillwater, OK 74078, U.S.A. E-mail: shannon.brewer@okstate.edu

Summary

1. A major limitation to effective management of narrow-range crayfish populations is the paucity of information on the spatial distribution of crayfish species and a general understanding of the interacting environmental variables that drive current and future potential distributional patterns.

2. Maximum Entropy Species Distribution Modeling Software (MaxEnt) was used to predict the current and future potential distributions of four endemic crayfish species in the Ouachita Mountains. Current distributions were modelled using climate, geology, soils, land use, landform and flow variables thought to be important to lotic crayfish. Potential changes in the distribution were forecast by using models trained on current conditions and projecting onto the landscape predicted under climate-change scenarios.

3. The modelled distribution of the four species closely resembled the perceived distribution of each species but also predicted populations in streams and catchments where they had not previously been collected. Soils, elevation and winter precipitation and temperature most strongly related to current distributions and represented 65–87% of the predictive power of the models. Model accuracy was high for all models, and model predictions of new populations were verified through additional field sampling.

4. Current models created using two spatial resolutions (1 and 4.5 km2) showed that fine-resolution data more accurately represented current distributions. For three of the four species, the 1-km2 resolution models resulted in more conservative predictions. However, the modelled distributional extent of Orconectes leptogonopodus was similar regardless of data resolution. Field validations indicated 1-km2 resolution models were more accurate than 4.5-km2 resolution models.

5. Future projected (4.5-km2 resolution models) model distributions indicated three of the four endemic species would have truncated ranges with low occurrence probabilities under the low-emission scenario, whereas two of four species would be severely restricted in range under moderate–high emissions. Discrepancies in the two emission scenarios probably relate to the exclusion of behavioural adaptations from species-distribution models.

6. These model predictions illustrate possible impacts of climate change on narrow-range endemic crayfish populations. The predictions do not account for biotic interactions, migration, local habitat conditions or species adaptation. However, we identified the constraining landscape features acting on these populations that provide a framework for addressing habitat needs at a fine scale and developing targeted and systematic monitoring programmes.

Ancillary