Factors influencing bumble bee persistence in a landscape fragmented by urbanization
Urbanization often reduces pollinator diversity. Although specific life-history traits may allow certain pollinator species to tolerate anthropogenic disturbance, even outwardly similar species can exhibit divergent responses to land-use change. Here, we use GIS to identify landscape-level correlates of occurrence for three bumble bee species (Bombus vosnesenskii, B. californicus, and B. melanopygus) in San Diego, CA. This heavily urbanized region still contains expansive areas of undeveloped scrub habitat as well as numerous scrub fragments. We established 57, one-hectare plots (100 x 100 m) divided evenly among three broad categories: urban areas, scrub fragments (< 500 ha), and large natural reserves (> 500 ha). We visited each plot multiple times in 2012 and 2013 to obtain presence / absence data for each species. We then used satellite imagery and GIS to quantify different landscape-level factors thought to be important in influencing bumble bee persistence: area of impermeable surfaces (e.g., pavement), area of sage (Salvia mellifera and S. apiana) cover, isolation (distance to nearest scrub habitat), proximity to coast, and proximity to a major freeway. For each species separately, we then used multiple logistic regression and model selection to test which independent variables best explained the observed pattern of presences and absences.
We observed approximately 1,000 individual bumble bee individuals across both years of the study. We found that B. vosnesenskii appeared to be the most prevalent species; we observed this bee at 74% of our plots, including 85% of reserve plots, 95% of fragment plots, and 39% of urban plots. Bombus californicus and B. melanopygus were found at 37% and 35% of our plots, respectively, and both species were present at roughly similar percentages of reserve and fragment plots (B. californicus: 50% of reserve plots, 58% of fragment plots; B. melanopygus: 45% of reserve plots, 42% of fragment plots). Model selection identified different correlates of presence / absence for each species. The strongest predictor of B. vosnesenskii presence / absence was the area of impermeable surface cover within 1.5 km of a plot (negative relationship). The strongest predictor of B. californicus presence / absence was the area of sage coverage within a site (positive relationship). The strongest predictor of B. melanopygus presence / absence was degree of isolation (negative relationship). These results suggest that different aspects of urbanization may influence how each bumble bee species responds to habitat fragmentation and urbanization.