Detecting changes in habitat-scale bee foraging in a tropical fragmented landscape using stable isotopes

As the body of research on the ecosystem service of pollination grows, our ability to tackle a range of agricultural, conservation, and land management issues is limited by our understanding of pollinator foraging patterns and requirements. In particular, better knowledge of which habitats bees utilize for foraging over their lifetime would inform a range of applied and theoretical questions. Traditional methods of studying foraging are either impractical for insects (e.g. radio tracking) or else are limited spatially or temporally (e.g. observations by researchers). Here we describe a method for using stable isotopes of carbon and nitrogen from bee tissues to gain an integrated signal of which habitats a bee has foraged in over its lifetime, using three species of social stingless bee (Apidae: Meliponini) in a fragmented tropical forest landscape in southern Costa Rica as a test case.There were strongly significant differences in the carbon and nitrogen isotope signals of bees collected in the landscape extremes: the largest forests and largest open pastures in the region. We could not estimate proportions of dietary carbon and nitrogen derived from different habitats, however, due to the high sample variance. Still, isotope ratios reflect a signature of the habitat that a bee has foraged in, and we found significant relationships between foraging patterns and season, landscape context (proportion of forest surrounding sample sites) and local biotic context (bee and plant diversity and abundance). Landscape context was the most consistent factor among the analyses, in line with the idea that meliponines do not preferentially forage in deforested habitats. Bee and plant diversity and abundance showed various significant relationships with the isotopic values of the different species, but the patterns were inconsistent. These results are potentially consistent with habitat switching from pasture to forest in areas of greater competition, but the extreme spatiotemporal variability in resource density likely blurs these patterns.