Agricultural land use modifies plant-pollinator interaction networks

Background/Question/Methods

Although the impact of agriculture on species loss and homogenization of biological communities (beta-diversity) are relatively well documented, its effect on the pattern of functional interactions among species, such as those between plants and insect pollinators, is less well known. Furthermore it is unclear the extent to which differences in such interaction networks result from habitat-based filtering of species (plants and pollinators) versus changes in visitation preferences.  We sampled wild bees from flowering plant communities at 21 sites in North Central California.  Sites included natural habitats isolated from agriculture, natural habitats adjacent farms, organic farms adjacent natural habitat and conventional farms adjacent natural habitat. We used these data to quantify the effect of agricultural land use on the composition of wild bee and plant communities and on the structure of their interactions.  Differences in bees, plants and quantitative visitation networks were tested using permutational ANOVA and potential homogenizing impacts of agriculture were quantified by comparing differences among sites within habitat types.  We used partial Mantel tests and a series of null models to explore whether differences in interaction networks are due to habitat-specific changes in composition of bees and plants or additional changes to visitation patterns.

Results/Conclusions

Agricultural land use affected plants, bees and patterns of linkage between the two.  Plant communities differed significantly between natural and farm sites, but not between farm types or between natural sites.  Consistent with their mobility, bee communities were most distinct between isolated natural sites and other habitats, but also differed between organic farms and other adjacent habitat types.   Linkage patterns showed similar between habitat differences as bees.  Agriculture also strongly homogenized bee communities and linkage patterns among sites.  Bee communities were significantly less distinct among farms than they were among natural habitat sites. Interactions showed a similar amount of homogenization among farms. Homogenization of plant communities was weaker; only conventional farms showed reduced among site differences. Null models and partial Mantel tests suggest that most of the observed differences in linkage are explained by differences in both plant communities and bee communities. The ability to partition out habitat specific visitation preferences proves difficult for quantitative network data.  That said, shift of bees’ visitation to widespread alien plant species may strongly affect the observed homogenization of networks in agricultural landscapes.