ABSTRACT: Sequences of the nrDNA ITS region have been obtained from 88 named taxa of family Orchidaceae to investigate the phylogenetics of subtribe Orchidinae. The first paper in this sequence (Pridgeon et al., 1997) emphasized intergeneric relationships, but the focus here is within genera. Overall, the monophyly of most genera is well supported, whereas support within genera is more variable. Beginning with the derived, globose-tubered genera, Ophrys, Serapias, and Himantoglossum–Barlia are well supported as genera but have short internal branches, reflecting their controversial morphologically based classifications. Orchis as currently widely delimited is triphyletic, prompting extensive taxonomic revisions below. A heterogeneous group characterized by 2n = (32–)36 is placed in an expanded Anacamptis; it contains four monophyletic groups based on “Orchis” laxiflora, “O.” coriophora, “O.” papilionacea–A. pyramidalis, and “O.” morio. Species of 2n = 42 form two clades. The smaller and more derived clade, based on “O.” ustulata and “O.” tridentata, is placed in an expanded Neotinea. The larger clade contains Orchis s.s. (including the now synonymized, molecularly and morphologically similar Aceras) and, more tentatively, Traunsteinera. Orchis s.s. can be divided into an anthropomorphic grade (e.g. “Aceras”, O. militaris) and two monophyletic groups based respectively on O. mascula and O. anatolica (perhaps including the O. patens aggregate). Passing down into the digitate-tubered grade, the Platanthera s.l. clade also encompasses Galearis and, more tentatively, Pseudorchis. Platanthera s.s. is well supported, and species relationships suggest specific migration patterns. Dactylorhiza is also well supported, but surprisingly also encompasses the now synonymized Coeloglossum viride. Interspecific branches are fairly short, and internal conflicts understandably characterize data from the reticulate allotetraploid complex. The supposedly primitive diploid D. iberica is shown to be a derived member of the spotted-orchid group. “Nigritella” is nested within (and thus synonymized into) Gymnadenia, revealing far more morphological than molecular divergence between the two previously recognized genera; their relationship to Dactylorhiza remains ambiguous.
Overall, relationships among species within the revised genera show ITS disparities of 0–95 steps, and those among genera 36–165 steps. In most cases there appears to be a strong positive correlation between molecular and morphological disparities, though this has yet to be quantified. Relationships within species cannot be assessed using ITS, requiring instead a combination of morphometric and population genetic techniques plus range-wide pollinator surveys. Historical review reveals that the pre-Victorian concept of Orchis was gradually dismantled into a series of monophyletic genera delimited by morphological synapomorphies, leaving Orchis as a plesiomorphic (and thus phenotypically cryptic), triphyletic residuum. It is not surprising that past attempts at hierarchical classifications within this still species-rich genus show little congruence with the ITS phylogeny, especially as they were highly typological. Their relative accuracy depended on levels of homoplasy in the chosen morphological character suite (low homoplasy for tubers, moderate for gynostemium, high for spur dimensions and perianth hooding). Attempts to study the taxonomy and biology of the artificial construct Orchis s.l. have been further hampered by misinformation (e.g. erroneous reports of hybrids and chromosome counts) and misconceptions (e.g. that contrasts among species in the relative amounts of floral pigments could be arranged in a primitive > derived polarity). Patterns shown by these character sets are far more explicit and informative in the phylogenetic context provided by the ITS data, as are several notable morphological parallelisms and convergences.