Feeding habitat quality and behavioral trade-offs in chimpanzees: a case for species distribution models

Abstract

The distribution and abundance of food resources are among the most important factors that influence animal behavioral strategies. Yet, spatial variation in feeding habitat quality is often difficult to assess with traditional methods that rely on extrapolation from plot survey data or remote sensing. Here, we show that maximum entropy species distribution modeling can be used to successfully predict small-scale variation in the distribution of 24 important plant food species for chimpanzees at Gombe National Park, Tanzania. We combined model predictions with behavioral observations to quantify feeding habitat quality as the cumulative dietary proportion of the species predicted to occur in a given location. This measure exhibited considerable spatial heterogeneity with elevation and latitude, both within and across main habitat types. We used model results to assess individual variation in habitat selection among adult chimpanzees during a 10-year period, testing predictions about trade-offs between foraging and reproductive effort. We found that nonswollen females selected the highest-quality habitats compared with swollen females or males, in line with predictions based on their energetic needs. Swollen females appeared to compromise feeding in favor of mating opportunities, suggesting that females rather than males change their ranging patterns in search of mates. Males generally occupied feeding habitats of lower quality, which may exacerbate energetic challenges of aggression and territory defense. Finally, we documented an increase in feeding habitat quality with community residence time in both sexes during the dry season, suggesting an influence of familiarity on foraging decisions in a highly heterogeneous landscape.

Keywords: animal ecology; environmental heterogeneity; habitat selection; primate behavior; species distribution models..

Figure 1

Overview of study area, with main land cover classes, stream locations, and boundaries for 2 study communities of chimpanzees as estimated with 99% minimum convex polygons for recorded sightings for the period 2000–2009.

Figure 2

Environmental layers used for modeling chimpanzee plant food species distribution.

Figure 3

Example chimpanzee habitat quality layers in wet (left) and dry (right) season of 2006, at a spatial resolution of 10 m. Values represent the sum of proportions of feeding time across all modeled food species that were predicted to occur in a given grid cell.

Figure 4

Change in mean habitat quality with elevation, for 6 arbitrary latitudinal ranges throughout Gombe National Park. Curves are smoothed interpolation lines (lambda = 100) across all 10-m grid cells of a given elevation in each latitudinal range, as defined on the inset map.

Figure 5

Change in mean habitat quality with elevation (meters above sea level) within each of 3 main land cover classes used as feeding habitats of chimpanzees at Gombe. Curves are smoothed interpolation lines (lambda = 100) across all 10-m grid cells of a given elevation in each land cover class.

Figure 6

Effect of residence time in community on mean feeding habitat quality at which adult chimpanzees were sighted during the study period, in relation to season, sex, and swelling state. Quality is expressed as a standardized score in relation to the mean quality across all feeding party locations in a given season.