Investigating floral choice in bees (megachilidae) using pollen metabarcoding.

Doctor of Philosophy in Genetics. University of KwaZulu-Natal, Durban 2017Interactions between plants and their pollinators are often poorly understood, specifically in a species-diverse country such as South Africa. Traditional methods of studying plant-pollinator interactions are time-consuming and imprecise. This study aimed to develop a technique that uses genetic analyses to identify pollen provenance directly from bees (Megachilidae) housed in a historic collection and to apply this technique to investigate floral choice differences in species of megachilid bees from three regionally important areas in South Africa: the Succulent Karoo, Savanna, and a widespread group, with bees occurring throughout the country. To develop the technique to accurately identify provenance, pollen was sampled from Megachile venusta specimens in the collection. Three DNA barcode regions were amplified and sequenced on an Illumina MiSeq instrument: the internal transcribed spacer 1 (ITS1) and internal transcribed spacer 2 (ITS2) regions, and the ribulose-1,5-biphosphate carboxylase (rbcL) gene. Sequenced reads were compared to sequence reference databases that were generated by extracting sequence and taxonomic data from GenBank. ITS2 reads were also compared to an established ITS2 database for Viridiplantae. More diverse plant classifications were obtained with ITS2 compared to ITS1. Amplification and sequencing of rbcL was inconsistent on pollen sampled from historic specimens. To study how floral choice differed in three South African regions, ITS2 was sequenced on Illumina HiSeq and MiSeq from pollen sampled from two different bee species from each region. Sequence reads were compared to the previously published ITS2 sequence reference database. Generalised linear models (GLM) indicated that the mean number of both plant families and species varied significantly between bee species. No significant effect of the time since bee collection was found. Taxon identifications were only confidently interpreted on family-level due to very limited local plant representation in sequence reference databases. DNA metabarcoding of mixed-origin pollen samples provided a faster, more accurate method of determining pollen provenance, without the need for expert palynologists. The use of historic collections to sample pollen directly from pollinators provided additional value to these collections. Sampling pollen from historic collections can also provide the spatial and temporal scales for investigations into changes in plant community structure or pollinator floral choice in the face of global climate change