Allegheny woodrat and eastern small-footed bat Inventory at Cumberland Gap National Historical Park 2021 - Project Occupancy Analysis Code - NPS Inventory Program

Wildlife Specialists, LLC

Inventory project analysis code - contact the NPS Inventory Program for access. From report: https://irma.nps.gov/DataStore/Reference/Profile/2302513 Occupancy model methods: To address some of the issues with the design constraints, we increased the sample size to 16 by treating the rim and base of the original Sample Units as separate. We created a binary detection history of Allegheny woodrat detections (1) and non-detections (0) and pooled the data for spatial replicates for each night in each Sample Unit. We fit the data to Single Season Occupancy models in Program PRESENCE 2.13.18 (https://www.mbr-pwrc.usgs.gov/software/presence.html, accessed October 2021). A detection was defined as any five-photo burst in which an Allegheny woodrat was captured, and a site was considered occupied if 1 Allegheny woodrat detection occurred during a single night of sampling. Our candidate set of six probability of detection models for Allegheny woodrats included covariates for trap response, level of effort (number of sites/50 meters of survey length), habitat (whether the site was on the rim (1) or base (0)), and detection of larger species (i.e., bear or bobcat and raccoon or opossum) to account for their potential effects on detection, as well as null and global models (Table 4 in report https://irma.nps.gov/DataStore/Reference/Profile/ 2302513). Since cameras were baited, we expected there to be a trap response where detection probability increased based on whether detections occurred the previous night. The number of camera locations differed between Sample Units (from 1-2 at the rim and from 2-4 at the base) primarily to account for differences in size, but also due to presence of suitable rock features. Vegetative habitat was similar across the site with major differences between the rim and base in terms of canopy cover and composition, but the presence of suitable food sources was not a limiting factor in any Sample Unit. Removal of bait by other species was common across the base of the site, which we predicted could have a negative impact on detection. We did consider including some measure of rocky habitat and weather as covariates but ultimately decided this would have minimal impact. Cameras were only placed in areas with suitable rock features and the amount of suitable rock features in a Sample Unit was partly addressed by the level of effort. Weather was consistent across the entire survey period with low temperatures around 68 °F and only one night of rain on August 31. Detections were recorded on this night for all Sample Units affected. We determined the most parsimonious probability of detection model and included this covariate (trap response) in all subsequent occupancy models.