We used the ITS phylogeny of Ipomoea as a single taxonomic character. a, We inferred a ML tree that included, when possible, multiple accessions of each putative species and interpreted it in three distinct ways, monophyly of accessions, non-monophyly and distantly related, non-monophyly but closely related. b, accessions of a given species are monophyletic (black triangles). Also, two existing species (I. piurensis and I. acanthocarpa) that we considered conspecific formed a clade of accessions that are intermingled (dashed-line green box), confirming our hypothesis. c, accessions of a given species are non-monophyletic and appear in very different parts of the Ipomoea phylogeny. In the example shown, our hypothesis based on morphology was a monophyletic I. squamosa but the ITS tree split the specimens, with some forming a clade in clade D and others forming a clade in clade A which is very distantly related to clade D. In such cases, we re-examined the morphology and usually found a mis-identified specimen(s). In other cases, specimens were similar on herbarium sheets but could be distinguished on closer inspection. In this case, the ITS tree alerted us to re-examine and have a closer look at specimens and subsequently describe the new species I. cryptica. d–e, in some parts of the ITS phylogeny there was a lack of resolution, with only a small number of subclades recognised and generally lacking support. Here we show an example with the clade including sweet potato. In such cases, we did not follow the ITS phylogeny but relied on morphology and genomic data when available. This decision was based on the fact that f, genomic phylogenies for the sweet potato clade demonstrate that nearly all species are monophyletic.