Life history and resource distributions determine bee sensitivity to landuse change

Background/Question/Methods

Bees are key pollinators in natural and managed ecosystems throughout the world and in many regions are threatened by anthropogenic landuse change.  Their sensitivity to landuse change is in part determined by ecological and life history traits, which define their need for different essential resources, such as nesting sites and forage.  These resources may be differentially distributed throughout the landscape such that trait-based bee guilds respond differentially to landuse change.  The result may be shifts in biodiversity, community composition, and pollination function.  To test for differential responses of bee guilds to landuse change, we surveyed bee communities throughout the growing season at semi-natural habitats and adjacent farm sites in a California agricultural landscape.  We assigned all bee species to different ecological guilds based on specific traits such as nesting substrate, trophic specialization and sociality.  We then used ANOVA to test for changes in the relative abundances of different guilds according to landuse.  To explore how nesting and floral resources interact with life history traits to differentially affect bee responses, we applied a spatially explicit model that coded the landscape based on suitability for nesting and forage resource for each bee species to predict relative abundance of different bee guilds across the landscape.

Results/Conclusions

Species traits strongly affected their sensitivity to landuse change.  In particular, bees that depended on wood for nesting substrate were relatively less abundant than ground-nesting species at agricultural sites than at adjacent natural habitats (F_2,13 = 17.92, P < 0.001) .  The landscape model revealed how the distributions of essential resource can provide a mechanism for bee responses to landuse change.  The model predictions matched observed abundances of wood-dependent species throughout the landscape (R² = 0.36, P = 0.01); however, it was less predictive of ground-nesting species (R² = 0.11, P = 0.20).  The results highlight the importance of species traits and resources distributions in filtering ecological communities among habitats including those modified by anthropogenic disturbance