The distribution of wild bee species along a Latitudinal gradient in northern Europe depends on their flower preferences

Aim
The functional diversity of bees contributes to the maintenance of plant biodiversity because different species of wild bees prefer and pollinate different plants. Many bees, in particular species with narrow flower preferences or specialized habitat requirements are threatened by landscape homogenization and climate change. Nonetheless, we still lack an understanding of large-scale impacts of anthropogenic stressors on the distribution of wild bee species with different flower preferences.

Location
Northern Europe: Norway, Denmark and Germany.

Methods                                                                             
We combine a dataset comprising ~30 000 observations of presences or absences of bee occurrences from structured surveys at 269 sites in northern Europe to investigate if flower preferences modulate species distributions across multiple environmental gradients. Bees were assigned a continuous functional trait separating preference for short vs tubular flowers.

Results
We observe that bee flower preference for either tubular flowers (Fabaceae) or plants with shallow flowers (including Apiaceae and Brassicaceae) can be described by a continuous flower preference trait score. The likelihood of observing a bee along a latitudinal gradient – encompassing variation in temperature, atmospheric N deposition and elevation – is dependent on its flower preference trait score. Specifically, bees with preferences for tubular flowers has a higher likelihood of occurrence with higher latitudes, while bees with preference for non-tubular flowers increase towards the south.

Main conclusions
Our results improve our understanding of how species-specific variation in flower preferences drives community-wide shifts in diversity and can therefore help devise region-specific conservation strategies.

Methods
We combined datasets from six different wild bee surveys conducted within northern Europe. The combined dataset consisted of a total of 269 sites sampled between 2017–2022, located in roadsides (Sydenham et al., 2024; Sydenham et al., 2023) and semi-natural grasslands (Sydenham et al., 2022a; Sydenham et al., 2022b; Sydenham et al., 2022c).

We used binomial GLMMs (R package ‘lme4’) with the occurrence of a species (presence/ absence) as response variable to test for interactions between flower preference niche score (DCA1) and the bioclimatic environmental gradients (PC1 and PC2) while controlling for landscape composition (PC1 and PC2).