Habitat quality influences pollinator pathogen prevalence through both habitat–disease and biodiversity–disease pathways

The dilution effect hypothesis posits that increasing biodiversity reduces infectious disease transmission. Here, we propose that habitat quality might modulate this negative biodiversity–disease relationship. Habitat may influence pathogen prevalence directly by affecting host traits like nutrition and immune response (we coined this as the ‘habitat–disease relationship’) or indirectly by changing host biodiversity (biodiversity–disease relationship). We used a path model to test the relative strength of links between habitat, biodiversity, and pathogen prevalence in a pollinator–virus system. High-quality habitat metrics were directly associated with viral prevalence, providing evidence for a habitat–disease relationship. However, the strength and direction of specific habitat effects on viral prevalence varied based on the characteristics of the habitat, host, and pathogen. In general, more natural area and richness of landcover types were directly associated with increased viral prevalence, while greater floral density was associated with reduced viral prevalence. More natural habitat was also indirectly associated with reduced prevalence of two key viruses (black queen cell virus and deformed wing virus) via increased pollinator species richness, providing evidence for a habitat-mediated dilution effect on viral prevalence. Biodiversity–disease relationships varied across viruses, as prevalence of sacbrood virus was not associated with any habitat quality or pollinator community metrics. Across all viruses and hosts, habitat–disease and biodiversity–disease paths had effects of similar magnitude on viral prevalence. Therefore, habitat quality is a key driver of variation in pathogen prevalence among communities via both direct habitat–disease and indirect biodiversity–disease pathways, though the specific patterns varied among different viruses and host species. Critically, habitat–disease relationships could either contribute to or obscure dilution effects in natural systems depending on the relative strength and direction of the habitat–disease and biodiversity–disease pathways in that host–pathogen system. Therefore, habitat may be an important driver in the complex interactions between hosts and pathogens.

Methods
We collected pollinator community species richness and abundance data by sampling pollinators from each community with hand nets and pan traps. At each collection site, we measured the density of flowers and species richness of flowering plants (including native and non-native plants) to assess local habitat quality along the same transects that pollinators were collected from. To assess landscape habitat quality for pollinators, we used the USDA cropland data layer to quantify the proportion of natural area (including forest, wetland, and grassland) and the landscape richness (the number of different landcover types) within a 1000 m radius of each collection site. Landowner permission was given for pollinator collection at each site. 

A subset of Apis mellifera, Bombus impatiens, Lasioglossum spp. and Eucera pruinosa was tested for the presence of deformed wing virus (DWV), black queen cell virus (BQCV), and sacbrood virus (SBV). Viral data was recorded as binary presence or absence of each virus for each individual tested. Each bee in the study was tested for all three viruses.

We used a path model to test the relative effects of habitat quality characteristics and pollinator community characteristics on patterns of viral prevalence. The path model allows us to rigorously disentangle the effects of habitat characteristics on patterns of pathogen prevalence through the newly proposed habitat–disease relationship and the well-established biodiversity–disease relationship.