Sequence Analysis
Pollen libraries were analyzed using JAMP (https://github.com/VascoElbrecht/JAMP). In summary, the pipeline demultiplexed the sequences using the assigned custom tags, trimmed the primers using cutadapt (v. 2.4; Martin 2011), filtered by length (184 +/- 10 bp) and expected error (1), and denoised using Usearch (Edgar 2010). The resulting exact sequence variants (ESV) were queried against a custom rbcL library (Braukmann et al. 2017, Kuzmina et al. 2017) using MegaBlast (Tan et al. 2006) in Geneious (ver 9.1.1; Kearse et al. 2012). The extracted hits were then queried against the ESV using the classify sequences command in Geneious with a minimum 99% identity match and 0.5% to the next best hit. A 99% threshold was chosen to allow more sequences to be included, as rbcL markers are distinct at the family-level. Singletons and ESVs below 0.01% were excluded as these are likely not represent true diversity but rather sequencing or PCR errors.