| The diversity of Mentzelia section Bartonia (Loasaceae), a monophyletic group that is widely distributed in arid regions of the American West, has presented considerable taxonomic difficulties, including misapplied names, vague descriptions that circumscribe multiple entities, a poor sense of geographic distributions, undescribed diversity, and confusing morphological variation in complexes of species. We take a phylogenetic approach, along with extensive museum and field studies, to reassess the diversity in the section. We applied nrDNA nucleotide sequence data for phylogeny reconstructions that recovered two well-supported clades that included (1) a clade of the Great Plains M. decapetala and its sister, which is restricted to taxa from the intermountain region and included species previously denoted as subshrubby, and (2) a more widespread clade that included taxa from the Chihuahuan and Sonoran deserts as well as the Great Plains, Rocky Mountains, and intermountain region. Notable results include modern treatments of M. multiflora as highly polyphyletic, and we suggest a narrow circumscription of this taxon to include only populations in the Southern Rocky Mountains. Our results call to attention undescribed diversity, including ten species described here. We summarize all previous studies and phylogenetic results in a monograph.;Divergence times of section Bartonia and its relatives were estimated with molecular dating methods. Penalized likelihood analyses suggested an origin of Mentzelia 60--27.5 million years before present (mybp), however, crown ages of the sections were much more recently derived. For example, section Bartonia was estimated to have diverged 11--1.5 mybp. Furthermore, we quantified the influence of substitution models on divergence time estimates with an empirical data set for Cornales that had 16 calibration point constraints in maximum likelihood analyses using 19 different substitution models. Discrepancy in divergence time estimates among corresponding nodes was small in most cases (within 95% confidence intervals); however, we recovered instances of nodes varying as much as 23.7%. We estimated that, on average for all nodes within a tree, divergence times vary 1.0--3.6%. Most discrepancy was associated with long branches, although short branches were also a source of discrepancy. We found no differences in disparity among nodes that were reconstructed in deep- mid- or shallow-level regions of the topologies. |
