Systematics of Vigna subgenus Lasiospron (Benth.) Maréchal et al. (Leguminosae: Papilionoideae: Phaseolinae)
Delgado, Alfonso; Lavin, Matt; Snak, Cristiane; Lewis, Gwilym (2021), Systematics of Vigna subgenus Lasiospron (Benth.) Maréchal et al. (Leguminosae: Papilionoideae: Phaseolinae), Dryad, Dataset, https://doi.org/10.5061/dryad.crjdfn35r
A taxonomic treatment is herein provided for the six primarily Neotropical species of Vigna subg. Lasiospron. This subgenus is distinguished, in part, by its close relationship with Old World Vigna species, but rather than having an Old World Vigna floral asymmetry where distal floral parts can have a right-hand curvature (from a face-view perspective), distal parts of subg. Lasiospron flowers, like those of American Phaseolinae, curve to the left. Our proposed taxonomy reflects published phylogenetic analyses that resolve the subg. Lasiospron species in a subclade of the primarily Old World clade of species of Vigna sensu stricto. Vigna subg. Lasiospron is therefore distinctive in having a primarily neotropical distribution. Three of the six subg. Lasiospron species have an amphi-Atlantic distribution, which we suggest was achieved naturally be the ability of these species to disperse and colonize coastal habitats. Nomenclatural and taxonomic synonyms are reported, along with species descriptions and the biogeographical, genetic, and morphological evidence that argues for each of the six Lasiospron species fitting a unified species concept.
We reanalyzed 160 ITS accessions representing the 67 species of the Vigna sensu lato clade resolved by Delgado-Salinas et al. (2011). This clade comprises the African genus Physostigma and the mostly African and Asian species of Vigna sensu stricto. The latter includes the focal subclade of this analysis, Vigna subg. Lasiospron, the American Vigna. The Vigna sensu stricto clade and Physostigma from the sister clade to the American Phaseolinae (Ancistrotropis, Cochliosanthus, Condylostylis, Dolichopsis, Helicotropis, Leptospron, Macroptilium, Mysanthus, Phaseolus, Oxyrhynchus, Ramirezella, Sigmoidotropis, and Strophostyles; Delgado-Salinas et al. 2011). We included conspecific samples of species from these genera to test the delimitation of species of Vigna subg. Lasiospron and compare patterns of coalescing conspecific genetic samples within subg. Lasiospron with such patterns in other subclades of Vigna sensu lato.
Phylogenetic Analysis—We conducted a Bayesian phylogenetic analysis of an ITS sequence data set from Delgado-Salinas et al. (2011) using BEAST 2.6.2 (Bouckaert et al. 2019). We did not redo the analysis of the chloroplast locus trnK/matK because those data do not included multiple conspecific accessions of subg. Lasiospron species (Delgado-Salinas et al. 2011) and because such studies as Raveenadar et al. (2018) find ITS sequence data to be significantly more informative at the species level in the genus Vigna than any of the commonly used cpDNA loci. The Bayesian analysis included partitioning the ITS region into three subregions (i.e., the internal transcribed spacer 1, the 5.8S locus, and the internal transcribed spacer). The nucleotide substitution model involved model averaging using bModelTest (Bouckaert and Drummond 2017) or using the general time reversible nucleotide substitution model, as in Delgado-Salinas et al. (2011).
Multiple MCMC chains each included 100 M generations sampled every 90 K generations such that a 10% burnin yielded 1000 sample points at likelihood stationarity. Using Tracer (Rambaut et al. 2018), we inspected likelihood scores simultaneously from multiple Bayesian runs, which invoked the same or differing combinations of settings (e.g., one versus three ITS data partitions, bModelTest versus general time reversible site models, and birth-death versus Yule tree models). In this manner we verified that all Bayesian runs achieved likelihood stationarity and determined the effect on the results of invoking different combinations of site and tree models.
The main tree prior included the root of Vigna sensu lato (the entire data set) with an age constraint set as a normally distributed density calibration of 8.0 ± 0.8 Ma (Lavin et al. 2005) rather than 9.1 Ma ± 1.0 (Delgado et al. 2011). This is because we wanted to bias ages toward conservative (minimum) estimates. With the root age so constrained, we made age estimates for the stem clades of Vigna sensu stricto plus Physostigma, the American Phaseolinae, all genera of the American Phaseolinae, and all the main subclades of Vigna sensu stricto including subg. Lasiospron.
The file is in Nexus format.