Maternity roost selection of Indiana bats (myotis sodalis) and occupancy of two threatened myotine bat species on national wildlife refuges in northern Missouri

Abstract

North American bat species face a range of environmental stressors which have negatively impacted recovery of the endangered Indiana bat (Myotis sodalis) and led to inclusion of the northern long-eared bat (Myotis septentrionalis) as a federally threatened species. Historic threats like disturbance of winter hibernacula and habitat loss continue to imperil both species, but the introduction of the fungal disease white-nose syndrome (WNS) into North America in 2006 resulted in substantial population declines in several species of Myotine bats in the eastern and central portions of the United States. Prior to the emergence of WNS, M. septentrionalis populations were estimated in the millions, and the rangewide M. sodalis population had experienced almost a decade of steady recovery. However, since the onset of WNS, M. sodalis populations have declined steadily and M. septentrionalis now faces extirpation from much of its range. Additionally, the development of wind power facilities across large portions of the central U.S. has increased the likelihood that critical habitat will be lost or fragmented and pose a new threat of large-scale mortality caused by collisions between bats and turbine blades. Objectives of this study were to 1) quantify maternity habitat characteristics of the endangered Indiana bat (Myotis sodalis) in northern Missouri to identify factors that drive selection and 2) identify local- and landscape-scale habitat characteristics associated with occupancy of M. sodalis and M. septentrionalis. To identify drivers of maternity roost selection of M. sodalis in Northern Missouri, we used mist nets to capture pregnant and lactating females during the summers of 2017 and 2018 and applied radio transmitters to individuals with sufficient body mass. We tracked 24 M. sodalis for an average of 5.8 days and identified 21 roost trees. We conducted emergence counts at each roost to classify them as primary or alternate and collected habitat data for each tree and the surrounding area. We then collected the same habitat data at available roosts and used discrete choice models to compare selected roosts with available trees within the study area. The top ranked model for primary roosts included tree diameter (DBH), tree height, and canopy cover while the top ranked model for alternate roosts included DBH, snag basal area, and canopy cover. Our results indicate that the probability of primary roost selection was greatest for trees with DBH ~ 50 cm and height of ~ 17 m. Roost site selection probability decreased with canopy closure, falling to 0 above ~ 75% closure. The probability of selection for alternate roosts was associated with greater canopy closure (~ 75%), smaller trees (~ 35 cm) and was positively associated with snag basal area. Land managers who wish to promote maternity habitat for M. sodalis could preserve existing snags, implement techniques to create new large-diameter snags, and, when possible, acquire additional bottomland hardwood forests to ensure the availability of an extensive network of available roost trees. To identify the local- and landscape-scale factors associated with occupancy of M. sodalis and M. septentrionalis, we used ANABAT SD1 acoustic detectors to record the echolocation calls of passing bats at 87 sites during the maternity seasons of 2017 and 2018. We deployed three detectors at each site for a minimum of two consecutive nights and recorded a total of 581 detector nights. Calls were identified to species, and detection histories of M. sodalis and M. septentrionalis were used to develop single-season occupancy models which used environmental covariates to estimate the probability of detecting each target species and evaluated the effects of local and landscape habitat characteristics on occupancy probability. The top detection model for M. sodalis included minimum temperature, barometric pressure, average wind speed, and moon minutes. The top detection model for M. septentrionalis included amount of precipitation. Our models indicate that the primary drivers of M. sodalis occupancy were percent of landscape (3 km) composed of wooded wetlands, distance to nearest wooded wetland, forest connectivity, forest shape, and wetland connectivity. Occupancy was positively associated with the proportion of wooded wetlands on the landscape, forest shape, and wetland connectivity and negatively associated with distance to nearest wooded wetland, forest connectivity, and proportion of forest on the landscape. The primary factors associated with M. septentrionalis occupancy were proportion of wooded wetlands on the landscape and wetland connectivity. Occupancy was positively associated with proportion of wooded wetlands and negatively associated with the degree to which wetlands were connected. We recommend land managers preserve riparian forest habitat and enact measures to reduce clutter and stem density in upland forests to improve overall habitat suitability and increase the likelihood that forests in this region can support foraging Myotis bats.