Beardtongues of the genus Penstemon are largely pollinated by hymenoptera such as bees and wasps, but there have been repeated independent transitions to pollination by hummingbirds. With these shifts come alterations in floral color, morphology, nectar traits, and pollen dispensing. A typical "bee" Penstemon is blue or sometimes yellow/white with wide corollas and concentrated nectar. A typical "hummingbird" Penstemon is red with narrow corollas and dilute plentiful nectar.
The floral diversity and repeated shifts in pollination have inspired a number of studies looking at adaptive evolution in the genus. But many of the species relationships have been unresolved and hampered by the very thing that makes studying the genus so appealing. The rapid speciation and evolutionary radiation that makes studying adaptation in these groups so interesting have also meant the species relationships have been difficult to resolve.
A new study now demonstrates the utility of multiplexed shotgun genotyping (MSG), a variation of restriction site-associated DNA sequencing (RADseq), to infer phylogenetic relationships within a subset of species in this genus.
We sampled genomic DNA, primarily from herbarium material, and subjected it to MSG library preparation and Illumina sequencing. The resultant sequencing reads were clustered into homologous loci, aligned, and concatenated into data matrices that differed according to clustering similarity and amount of missing data.
The MSG approach resolved species relationships and confirmed both the ancestral state of hymenoptera pollination and several independent transitions to hummingbird pollination. The study also confirmed that relationships within Penstemon are difficult to resolve due to gene tree discordance which is likely the result of incomplete lineage sorting.
This study demonstrates that phylogenomic approaches yielding thousands of variable sites can greatly improve species-level resolution of recent and rapid radiations. Similar to other studies, we found that less conservative similarity and missing data thresholds resulted in more highly supported