Premise of the study: Early plant taxonomists formed hypotheses about relationships among taxa based on characters such as morphology, anatomy, phytochemistry, ecology, and geography. Modern molecular systematic methods, based on DNA sequence variation, augment early methods and provide an additional line of evidence by which to evaluate taxonomic hypotheses. In North America north of Mexico, wild onions (Allium, Amaryllidaceae) are represented by 84 native species, 81 of which belong to subgenus Amerallium. On the basis of morphology, these species have been divided into eight informal taxonomic "alliances" hypothesized to represent shared evolutionary history among species. The main aim of this research was to test the monophyly of the alliances with molecular phylogenetic methods.
Methods: We sampled 74 Amerallium species north of Mexico and two Mexican endemics and constructed a molecular phylogeny of subgenus Amerallium in North America based on predominantly noncoding sequences from two nuclear ribosomal RNA regions (ITS and ETS) and two plastid regions (trnL-F and rpL32-trnL).
Key results: Most clades are well supported in analyses of nuclear data and when nuclear and plastid data are combined. However, the plastid data alone did not produce a well-resolved or well-supported tree. Morphological alliances were sometimes congruent with groups recovered in the molecular phylogeny, but strict monophyly was observed in only three of eight alliances.
Conclusions: We propose an infrageneric classification that recognizes two sections in New World Amerallium. Because there is substantial incongruence between morphological and molecular groups, we advocate retaining informal alliances rather than adopting formal subsections until further morphological and molecular analyses can be carried out.