Biogeographic history predicts bee community structure across floral resource gradients in south-east Australia

Aim: Plant populations are declining in their native ranges around the globe through the expansion of agriculture, urbanization, and plant invasions. We test the hypothesis that animal species that have spent more of their evolutionary history in a region are more dependent on native plants, particularly those plants that have spent more of their evolutionary history in the region, and are therefore more negatively impacted by native plant decline. Location: Yarra Valley landscapes, Australia. Methods: We test the presence and pattern of phylogenetic signal in native bee community responses to local flower density of ancient Australian plant lineages and the amount of native vegetation in the surrounding landscape across farm and native vegetation sites. We also test phylogenetic signal in the frequency of bee visitation to flowers from ancient Australian plant lineages. We compare the patterns of phylogenetic signal to the current understanding of bee biogeographic histories to evaluate our hypothesis. Results: There was significant phylogenetic signal in responses to flower density of plants from ancient Australian lineages, and the frequency of visitation to these flowers, with most species from the ancient Australian bee Glade being positively associated with these flowers. This is consistent with our hypothesis. Significant phylogenetic signal in response to native vegetation in the surrounding landscape was driven primarily by the more recently arrived bee linages, with ancient lineages able to persist on some farms where ancient Australian flowers were present (e.g. on roadsides). Main conclusions: Bee species that have spent more of their evolutionary history in Australia are more dependent on ancient Australian plant lineages and so most negatively impacted by the decline of these plants. This may be a broader phenomenon because phylogenetic conservatism in host plant use, the main assumption underlying our hypothesis, is common among herbivorous arthropods (similar to 500,000 species).