Wild bee communities in grassland habitats of the Central Valley of California: Drivers of diversity and community structure

Recent research has revealed a trend of decreasing pollinator abundance and diversity in regions throughout the world. This highlights the need to understand factors influencing patterns in bee community structure and the drivers of bee diversity and abundance patterns. My dissertation uses several methods to determine factors structuring bee communities with regards to diversity and abundance. I selected 10 sites in different regions of the Central Valley of California that differ with regards to land use and floral diversity. Bee communities at each site were sampled for diversity, abundance, and bee-floral host relationships. Sampling bee communities is often done using only bee bowls because netting is time consuming and prone to sampler bias. In chapter one the methods used in this study were detailed and the use of bee bowls and netting in capturing a representative sample of the bee community were compared using the Sørensen's similarity index and the Bray-Curtis dissimilarity index. It was determined that sampling using one method alone would miss approximately 40% of the species richness of the community. Further, there were biases in using bee bowls and nets; the bee bowls sampled certain species more than nets and vice versa. This chapter provides evidence that to adequately sample a bee community both bee bowls and netting must be used. Chapter two focuses on bee biodiversity and the correlation between bee species richness and plant diversity. Patterns of diversity in bee communities of the Central Valley indicate that the family Apidae was more speciose than other families. However, on a species level, those from the family Halictidae far exceeded species from Apidae in abundance. This could have reflected a sampling bias given that pan traps tend to sample individual bees from Halictidae more than Apidae. Chapter two also focused on temporal variability. There was considerable temporal variability in the abundance of one of the more abundant species, Lasioglossum incompletum. This highlighted the need for studies of longer duration in order to account for natural stochasticity in bee populations. Several different diversity indices were used to assess the biodiversity of the different study locations; Putah Creek sites were found to be more diverse than the San Joaquin sites. A correlation analysis was used to determine that a positive relationship between plant diversity and bee species richness did exist for 2005 but not 2006. This indicated that plant diversity may be one of the factors driving bee species richness and community structure.Another factor possibly accounting for variation in bee species richness and abundance is land use. Chapter three used non-metric multidimensional scaling and generalized linear mixed effects model to test for associations between differences in land use patterns and bee species richness and abundance. While there was no direct association between these factors, the ordination did show that the Putah Creek sites, San Joaquin sites, and Cosumnes sites clustered together. Therefore, sites that shared similar land-use patterns were related along a gradient. These cluster patterns were used to group the study locations for the other analyses performed in this project. The Putah Creek sites were characterized by agriculture and urban land use whereas, San Joaquin was semi-natural and Cosumnes Preserve was semi-natural and agricultural. Chapter four is an analysis of the pollinator networks of Putah Creek, San Joaquin, and the Cosumnes Preserve. Pollination webs, matrices, and gplots were used to visualize the networks, while network and species-level indices were used to assess asymmetry, specialization versus generalization, and connectance. It was determined through these analyses that the connectance of the network decreased with increasing species richness and the complexity and composition of the network varied between the three regions of the Central Valley. Further, the San Joaquin Refuge sites, which were characterized as seminatural land use, contained a higher number of oligolectic species than other sites dominated by agricultural and urban land use. The focus of this project was to use different methods to determine drivers of bee species diversity and abundance in different bee communities of the Central Valley of California. Three conclusions can be drawn from the analyses presented; 1) Given temporal variability in bee populations, studies of longer duration must be conducted to determine factors affecting bee community structure from that of natural population variability, 2) Floral diversity is positively correlated with bee species diversity and abundance but it is not the only factor influencing bee community structure. and 3) Land use change may be a factor influencing bee-plant networks but studies that compare networks across space and time are needed to determine the nature of this relationship