Floral hosts of leaf-cutter bees (Megachilidae) in a biodiversity hotspot revealed by pollen DNA metabarcoding of historic specimens

spread of sampling sites across the regions studied would have been ideal, but could not be obtained here due to the limiting factor of pollen availability on historic specimens. Conclusion In summary, this study showed that pollen metabarcoding can be used to answer questions about floral choice in bees from three different geographic areas. It was possible to show that bees from the Succulent Karoo and Savanna do not differ significantly in the number of plant families or species that they visit, and that widespread bees tend to visit more plant taxa than bees restricted to a single biome. Clustering sequence reads into species OTUs allowed floral choice ranges to be investigated in more detail than if only family level classifications were considered. A comprehensive DNA reference library for all flowering plants in South Africa would greatly improve the conclusions drawn from this and future metabarcoding studies. Using pollen metabarcoding can be very useful in understanding interactions between threat- ened pollinators and plants in highly diverse regions of the world such as South Africa and will aid conservationists in their efforts to protect these biodiversity hotspots. One of the reasons for undertaking this study was to understand better whether climate change will have a greater effect on endemic species with narrower distributions and pollina- tion specificity. Although bee species endemic to a small biome, such as the Succulent Karoo, do not differ much from those endemic to a large biomes, the Savannah, both are less polylec-

tic than wide-spread species. This topic needs further investigation, concordant patterns across different pollinators would provide the best evidence for climate-driven shifts in pollinator services. Supporting information S1 Table. The national insect collection, ARC, South Africa collection information of Megachile venusta bee specimens from which pollen was collected for this study. Collection information, such as the date, province, GPS coordinates and nearest town are given for each sample, where available. The new province and town designations are given in brackets where they have been changed. (DOCX) S2 Table. Collection information from the national insect collection, ARC, South Africa, of Megachile murina bee specimens from which pollen was collected for the Succulent Karoo group in this study. Collection information, such as the date, province, GPS coordi- nates and nearest town are given for each sample, where available. The new province and town designations are given in brackets where they have been changed. (DOCX) S3 Table. The national insect collection, ARC, South Africa collection information of Megachile maxillosa bee specimens from which pollen was collected for the widespread group in this study. Available information regarding the specimen collection, such as date, province, GPS coordinates and collection locality are given for each sample. (DOCX) S4 Table. Collection information from the national insect collection, ARC, South Africa, of Megachile felina bee specimens from which pollen was collected for the Savanna group

in this study. Available information regarding the specimen collection, such as date, province, GPS coordinates and collection locality are given for each sample. (DOCX) PLOS ONE Floral hosts of Megachilidae revealed by metabarcoding PLOS ONE | https://doi.org/10.1371/journal.pone.0244973 January 21, 2021 13 / 16