Use of predictive distribution models to describe habitat selection by bats in Colorado, USA. Issue 2 (11th January 2022)

• Record Type: Journal Article
• Authors: Neubaum, Daniel J.; Aagaard, Kevin
• Is Part Of: Journal of wildlife management. Volume 86:Issue 2(2022)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-01-11
• Subjects: bats -- Bayesian regression analysis -- Colorado -- habitat selection modeling -- pinyon‐juniper -- presence‐only data -- probability of use; Wildlife management -- Periodicals; Zoology -- Periodicals; 333.954
• DOI: 10.1002/jwmg.22178
• Languages: English
• ISSNs: 0022-541X

Abstract

Abstract: Numerous processes operating at landscape scales threaten bats (e.g., habitat loss, disease). Temperate bat species are rarely examined at commensurate scales because of logistical and modeling constraints. Recent modeling approaches now allow for presence‐only datasets, like those often available for bats, to assist with the development of predictive distribution models. We describe the use of presence‐only data and rigorous predictive distribution models to examine habitat selection by bats across Colorado, USA. We applied hierarchical Bayesian models to bat locations from 1906–2018 to examine relationships of 13 species with landscape covariates. We considered differences in type of activity (foraging, roosting, hibernation), seasonality (summer vs. winter), and scale (1, 5, 10, and 15‐km buffers). These findings generated statewide probability of use models to guide management of bat species in response to threats (e.g., white‐nose syndrome [WNS]). Analysis of buffers suggest selection of land cover and environmental covariates occurs at different scales depending on the species and activity. Pinyon ( Pinus spp.)‐juniper ( Juniperus spp.) appeared as a positive association in the highest number of models, followed by montane woodland, supporting the importance of these forest types to bats in Colorado. Other covariates commonly associated with bats in Colorado include westerly longitudes, and negative associations with montane shrubland. Mechanical treatments within pinyon‐juniper and montane woodlands should be conducted with caution to avoid harming bat communities. We developed hibernation models for only 2 species, making apparent the lack of winter records for bat species in the state. We also provide a composite predictive surface of small‐bodied bats in Colorado that delineates where these species, vulnerable to WNS, converge. This tool provides managers with focal points to apply surveillance and response strategies for the impending arrival of the disease. Abstract : Presence‐only databases for bat species are often available for analysis but overlooked because of the biases that need to be addressed when examining habitat selection at landscape scales. We applied hierarchical Bayesian models to bat locations using a long‐term, robust dataset to develop probability of use models for 13 bat species across Colorado, USA. Models indicate the importance of some land cover types (e.g., pinyon‐juniper woodland) and highlight inadequate sampling efforts in others (alpine).