Occupancy and Maternity Roost Selection of Northern Long-Eared Bats in the Lake States Region

Abstract

Among the bat species most impacted by White-nose Syndrome (hereafter; WNS) is the northern long-eared bat (Myotis septentrionalis), which has resulted in its recent listing as federally threatened under the Endangered Species Act. Currently the specific roosting habitat requirements of northern long-eared bats in the Lake States region is poorly understood. The recent occurrence of WNS in Wisconsin affords an important opportunity to establish baseline data on the northern long-eared bat for comparison to studies done elsewhere in the species’ distribution both before and at later stages of WNS occurrence. Information also is still needed on specific detection and occupancy patterns for this species in the Lake States Region to allow for the implementation of better management practices in the face of WNS and to better document imminent population declines. Misidentification of bat calls, especially those within the Myotis genus, is extremely common. This misidentification can introduce bias into estimates of occupancy and lead to erroneous interpretation of results i.e., causing conservation measures or mitigations to occur when in fact the species is actually absent. Utilizing models that allow for false-positive and false-negative error rates produces models with better support and more accurate site occupancy estimates. The objectives of my research were to determine if: (1) selection of maternal roost sites by northern long-eared bats in Wisconsin is influenced by characteristics of individual roost trees and site-level habitat conditions, (2) roost selection is attributed to overall availability within the stand, and (3) occupancy of northern long-eared bats is influenced by specific large-scale and site-level habitat characteristics using site occupancy models, which account for error that can be introduced by misclassification rates. I captured bats via mist netting at three known pre-WNS maternity sites in south-central Wisconsin in 2015 and 2016 and affixed radio transmitters to 39 female northern long-eared bats. I tracked them daily to 53 confirmed day roosts and recorded information on site and tree characteristics. I used conditional logistic regression and an information theoretic approach and found that the best models contained variables related to relative tree size and age (DBH, canopy class, and decay stage). Oak (Quercus spp) species were used most frequently as roosts at Sandhill and Black River Falls, but were used in proportion to availability. Black locust (Robinia pseudeoacacia) trees were used most commonly at Governor Dodge, and were used more than expected based on availability. I deployed 20 zero-crossing, frequency division bat detectors at each of 9 sampling sites in 2015 and 2016 stratified by habitat type. Detectors were left out for 10-15 capture nights. I compared two automated classifiers to identify my call files and then used the unmarked package in the R statistical program to run a false-positive occupancy analysis. I recorded 51,064 bat calls across both field seasons. The best supported single-season model for 2015 showed that elevation had the strongest effect on probability of occupancy for northern long-eared bats. The best supported single-season model for 2016 contained habitat type (wet vs. dry) as the best covariate for detectability, and distance to road as the best predictor of occupancy. The multiyear models resulted in year being the best predictor for detectability and distance to road having the most effect on site occupancy for northern long-eared bats. The results from this research provide region-specific data for the conservation of northern long-eared bats, along with baseline data for this species in the Lake States region to serve as a comparison to other regions at varying levels of WNS occurrence. Results from this research should also provide guidance regarding forest management practices that contribute to the conservation of this species, specifically increasing availability of potential day roosts on the landscape (i.e. snag retention). Increasing summer habitat and availability of roost trees during the reproductive period may improve recruitment for this species, which might mitigate the overall detrimental population effects of WNS during hibernation. It is important to continue to contribute invaluable ecological information on this species; however, it is also crucial to continue to evaluate methods used for bat call identification, as well as occupancy estimation. Proper evaluation and use of these methods will ensure that only quality data contributes to the management of this sensitive species.