Search

Recent advances in optical remote sensing (RS) technology in combination with lightweight Global Positioning System (GPS) tracking devices now make analyzing the multi-scale habitat selection (HS) of small mammals < 2 kg possible. However, there have been relatively few multi-scale HS studies integrating fine-scale RS data with data-rich, GPS-derived movement data from small mammals. This is critical because small mammals commonly select habitat features across multiple scales. To address this gap, we investigated the HS of a small mammal, fox squirrels (Sciurus niger), which are known to cover relatively large areas and select fine-scale environmental features. We specifically asked the following questions: (1) Do next-generation RS variables improve HS models at single spatial scales? (2) Do multi-scale HS models improve upon those at single spatial scales? Using data from 45 individuals, we constructed HS models at three spatial scales: 4 ha (210 m × 210 m), 0.09 ha (30 m × 30 m), and 0.01 ha (10 m × 10 m) using traditional and next-generation RS data. The 4-ha model, using traditional and next-generation RS data, produced the best single-scale model, explaining 58% of the variations in HS. However, the multi-scale model provided the most informative model, explaining 68% of the variations in HS. Our models provide evidence for the value of next-generation RS data when quantifying HS and additional support for the idea of studying HS at multiple spatial scales.

Increasing human–bear conflicts are a growing concern, and managers often assume bears in developed areas are food-conditioned. We examined the relationship between human–bear conflicts and food conditioning by analyzing isotopic values of hair from black bears (Ursus americanus floridanus) involved in research (n = 34) and conflicts (n = 45). We separated research bears into wild and developed subgroups based on the impervious surface within their home ranges and separated conflict bears based on observations of human food consumption (anthropogenic = observations; management = no observations). We initially assumed wild bears were not food conditioned and anthropogenic bears were. However, using isotopic values, we classified 79% of anthropogenic bears and 8% of wild bears as food conditioned. Next, we assigned these bears to the appropriate food conditioned category and used the categorizations as a training set to classify developed and management bears. We estimated that 53% of management bears and 20% of developed bears were food conditioned. Only 60% of bears captured within or using developed areas showed evidence of food conditioning. We also found that δ13C values were a better predictor of anthropogenic foods in a bear’s diet than δ15N values. Our results indicate that bears in developed areas are not necessarily food conditioned and caution against management actions based on limited observations of bear behavior.