Assistant Professor Kutztown University of Pennsylvania, United States
Abstract: Climate change and habitat destruction have impacted the habitats and migration patterns of many species. Migratory raptors respond to changes in their breeding and stopover sites and are good indicators of long term environmental health. To help understand the long-term trends of these various species, we used a 10-year dataset of raptor sightings during the migratory season to model the effect of weather, green vegetation, and the presence of other raptor species on raptor detectability. Raptor sightings data from Hawk Mountain Sanctuary included over 15 raptor species; weather data included temperature, cloud cover, and visibility; and green vegetation was incorporated as NDVI. We used conditional logistic regression to fit models for each raptor species. Predictor variables included weather data in both linear and quadratic formats to look for non-linear effects, NDVI on increasingly larger scales to identify the optimal scales for the raptors, and the presence or absence of other raptor species that had overlapping migratory patterns. We sampled this data across a 4-day window around each observation for the target raptor species. Temperature, cloud cover, and visibility in both linear and quadratic forms were all in some way deemed to be significant predictors. NDVI imagery was also deemed a significant predictor for each species at varying scales. For example, observations of turkey vultures, carrion birds that forage for food in open areas, were predicted by NDVI at a broad spatial scale. This study demonstrates the use of conditional logistic regression to predict raptor observations through time, which can be applied to other species with count data from a single location to aid in their study and conservation.
