Factors affecting acoustic detection and site occupancy of Indiana bats near a known maternity colony

Indiana bats (Myotis sodalis) are capable of exploiting ephemeral resources in structurally complex, interior forests. Consequently, surveying for this species is difficult, and even in areas harboring known populations, documenting presence remains a challenge. We aimed to determine what factors affected probability of detection (p) and site occupancy (psi) for Indiana bats when sampling with acoustic bat detectors near an active maternity colony. From May to August 2012 and 2013, we passively sampled 71 random points near Indianapolis, Indiana, with Wildlife Acoustic SM2BAT+ detectors. On average, sampling occurred similar to 2.2 km from maternity colony roosts, which contained >= 150 bats by mid-July. Habitat and landscape covariates were measured in the field or via ArcGIS. We reduced model covariates with Pearson's correlation and principal component analysis. Indiana bat calls were identified using Bat Call ID software. We used Presence 6.1 software and Akaike's Information Criteria to assess models for detectability and occupancy. Detectability of Indiana bats increased as forest closure and mean nightly temperature increased, likely due to reduced clutter and increased bat activity, respectively. The null model best explained psi, but models were likely underpowered from few Indiana bat detections. Higher rates for psi were observed near agricultural lands, indicating this cover type may fulfill certain habitat requirements for Indiana bats within the study site. Overall, detection rates were low even when sampling near a concentrated population on its maternity grounds, likely due to limitations inherent to bat detector technologies. Managers should be aware of the limitations of using passive acoustic sampling as a means by which to document presence-absence for this endangered species.