Transect and plot data from: Fruit trees drive small-scale movement of elephants in Kibale National Park, Uganda

Understanding what motivates animal spatial behavior requires discerning the relationship between animals and the landscape in which they move. Many studies have focused on large scale trends that drive migrations and movement across broad landscapes, but few studies have focused on drivers of local movement within smaller areas based on fine–scale landscape heterogeneity. This is especially true for African elephants (Loxodonta africana), whose movements in relation to landscape variables have frequently been studied across vast areas, but infrequently on a small scale. We report on how the local distribution of an important resource—fruit trees—shapes elephant movement patterns in Kibale National Park, Uganda. We hypothesized that the spatial location of elephant trails, and especially intersections in the network of trails, would be heavily associated with fruit trees, with trail density being higher in areas with greater fruit availability. Elephant trails and nearby fruit trees were mapped within a 64-km2 area within the interior of the Park. Trail and fruit tree sampling were conducted along sixteen evenly spaced one-km control transects and compared to 28.6 km of elephant trails arrayed throughout the park. Fruit tree basal area was higher along trails and at trail intersections. Trail density was also positively correlated with fruit tree density. Fruit trees may thus serve as a determinant of small-scale elephant movement and could be used to predict patterns of elephant spatial behavior across forested habitat, which has potential implications for elephant management in the area.Any blank rows with only a plot and grid label indicate that there were no trees meeting our requirements in that plot.Elephant trail data: We divided the 64 km2 study area into two-kilometer grid cells, resulting in sixteen squares (grid cells) of 4 km2 area which were each labeled by a letter. One elephant trail was mapped within each grid cell. Trails were followed for a minimum of one kilometer and a maximum of two kilometers. Each trail was divided into 200m segments, which are indicated by "Transect.Label" in the data. Along these trails, we recorded all trees of the study species >20cm DBH that lay within 10m of either side of the trail with approximate GPS location recorded in UTM. Trees that were greater than 20cm, but too small to be considered mature were removed in post processing. Vegetation was classified according to Wing & Buss (1970). Basal area was calculated in post processing. Control transect data: Control transects followed the method of "path of least resistance transects" (sensu Blake & Inkamba-Nkulu, 2004). These transects allow for minor deviations (40 degrees) around obstructions from the desired path (Blake & Inkamba-Nkulu, 2004). All trees of the study species >20 cm DBH (Table 1) within 10m of either side of the transect were recorded. Each transect was divided into 200m segments, which are indicated by "Transect.Label" in the data. Along these transects, we recorded all trees of the study species >20cm DBH that lay within 10m of either side of the transect with approximate GPS location recorded in UTM. Trees that were greater than 20cm, but too small to be considered mature were removed in post processing. Vegetation was classified according to Wing & Buss (1970). Basal area was calculated in post processing. Paired plot data: In each location (intersection and control site) we recorded all trees of the study species >10 cm DBH within a circular 400 m2 plot (11.3m radius). The minimum DBH was lowered to 10cm to gain a better understanding of the abundance of smaller trees. Trail intersections were identified along the trails previously mapped above. Trail intersections were only sampled within forest and woodland land cover. A maximum of one intersection per 200-m trail segment was sampled. Intersections were chosen by sampling the first major trail intersection crossed within that segment. Sample plots were located at the center of the intersection, as judged by eye. Paired controls to each trail intersection plot were placed 50m away in the direction most orthogonal from the trails at the intersection. Control plots may have been crossed by elephant trails. If the location of the control plot was deemed to contain the intersection of elephant trails with greater use or number of intersecting trails than in the recorded elephant trail plot, we returned to the initial trail intersection and measured 50m in the opposite direction to record an alternative control plot. The dataset includes trees from both the trail intersection and control plots which are distinguished by the "Plot.Type" column. Plot number indicates which pair the tree belongs to within the grid cell. UTM coordinates were taken from the center of the plot and do not accurately represent the location of specific trees. Trees that were greater than 10cm, but too small to be considered mature were removed in post processing. Basal area was calculated in post processing