Molecular phylogeny and evolutionary history of the Eurasiatic orchid genus Himantoglossum s.l. (Orchidaceae)

Abstract

Background and aims: Lizard orchids of the genus Himantoglossum include many of Eurasia's most spectacular orchids, producing substantial spikes of showy flowers. However, until recently the genus had received only limited, and entirely traditional, systematic study. The aim of the current work was to provide a more robust molecular phylogeny in order to better understand the evolutionary relationships among species of particular conservation concern.

Methods: All putative species of Himantoglossum s.l. were sampled across its geographical range. A large subsample of the 153 populations studied contributed to an initial survey of nuclear ribosomal internal transcribed spacer (nrITS) ribotypes. Smaller subsets were then sequenced for four plastid regions and the first intron of the low-copy-number nuclear gene LEAFY. Rooted using Steveniella as outgroup, phylogenetic trees were generated using parsimony and Bayesian methods from each of the three datasets, supplemented with a ribotype network.

Key results: The resulting trees collectively determined the order of branching of the early divergent taxa as Himantoglossum comperianum > H. robertianum group > H. formosum, events that also involved significant morphological divergence. Relaxed molecular clock dating suggested that these divergences preceded the Pleistocene glaciations (the origin of the H. robertianum group may have coincided with the Messinian salinity crisis) and occurred in Asia Minor and/or the Caucasus. Among more controversial taxa of the H. hircinum-jankae clade, which are only subtly morphologically divergent, topological resolution was poorer and topological incongruence between datasets was consequently greater.

Conclusions: Plastid sequence divergence is broadly consistent with prior, morphologically circumscribed taxa and indicates a division between H. hircinum-adriaticum to the west of the Carpathians and H. jankae-caprinum (plus local endemics) to the east, a distinction also suggested by nrITS ribotypes. LEAFY phylogenies are less congruent with prior taxonomic arrangements and include one likely example of paralogy. Himantoglossum metlesicsianum fully merits its IUCN Endangered status. Potentially significant genetic variation was detected within Steveniella satyrioides, H. robertianum and H. hircinum. However, confident circumscription of the more derived species of Himantoglossum s.s., including local endemics of hybrid origin, must await future morphometric and population-genetic analyses.

Keywords: Dating; Himantoglossum s.l.; LEAFY; Orchidaceae; evolution; hybridization; incomplete lineage sorting; internal transcribed spacer; molecular phylogeny; molecular phylogeography; nrITS; orchid; topological incongruence.

Fig. 1.

Details of species sampled: nomenclature, distribution and chromosome numbers, together with names of major clades (left). More detailed taxonomic reviews were presented by Delforge (1999) and Bateman et al. (2003). ¹Sundermann and von der Bank, 1977; ²Ströhlein and Sundermann, 1972; ³D'Emerico et al., 1992; ⁴Bernardos et al., 2006; ⁵Löve, 1976; ⁶Sramkó et al., 2012 (chromosome numbers reviewed by Bateman et al., 2003, 2013a). NA, not available; ?, uncertain data.

Fig. 2.

Sampling sites of Himantoglossum s.l. and Steveniella (outgroup) for this study. Open circles represent populations included in the phylogenetic analyses, taxa being indicated by the enclosed letters. Solid symbols represent additional sample sites used to assess variability in the nrITS region.

Fig. 3.

Parsimony network of ribotype groups of Himantoglossum s.s. identified by TCS (A) and their geographical distribution (B), together with three length-polymorphic variants (L02, L05 and L10), a putative pseudogene (PG) and the ribotype group of the outgroup, H. formosum (OUT). Blank circles (A) represent undiscovered (i.e. hypothetical) ribotypes. The size of pie charts (B) is proportional to the number of samples analysed from the same population (maximum of four).

Fig. 4.

Parsimony network of cloned nrITS sequences of the Himantoglossum hircinum-jankae clade identified by TCS (A) and their geographical distribution (B). The TCS network (A) is coloured according to the two main groups identified; each cloned ribotype is coded as ‘CR’ followed by a number. Blank circles (A) represent undiscovered (i.e. hypothetical) ribotypes. The pie charts on the map (B) show the ribotype composition of a cloned sample (i.e. ribotypes cloned from a given specimen); their size is proportional to the number of clones sequenced. Colouring and numbering (without the acronym ‘CR’) of ribotypes on the pie charts follow those of the TCS network; the arrowed pie chart is located in south-west England.

Fig. 5.

Phylogram of Himantoglossum s.l. based on four concatenated plastid intergenic spacer sequences. Region of origin (displayed as country code) is shown in parentheses. Numbers above branches represent statistical support (bootstrap value/Bayesian posterior probability; values for nodes a and b are 87/– and 95/100, respectively).

Fig. 6.

Nuclear DNA phylogram of the genus Himantoglossum based on the first intron of the single-copy LFY gene. Region of origin (displayed as the country code) is shown in parentheses. Numbers above branches represent statistical support (bootstrap value/Bayesian posterior probability; values for nodes a and b are 86/100 and 87/100, respectively).

Fig. 7.

Topological incongruences between the contrastingly inherited genomes within the as shown by the cladogram of MP trees of four combined plastid spacers (A) and the nuclear LFY first intron (B). Statistically weakly supported branches (i.e. those with <74 % bootstrap support) are dashed, and the most conspicuous topological incongruencies are in bold.

Fig. 8.

Bayesian chronogram of nrITS sequences of a representative set of European orchids with enhanced sampling of the Steveniella-Himantoglossum clade. The tree is the maximum clade credibility tree obtained in a Bayesian analysis with a log-normal relaxed clock (uncorrelated) after 10 % burn-in and secondary calibration points derived from Gustafsson et al. (2010). The numbers at nodes represents mean divergence dates of key clades in million years before present (Ma); dashed branches have low (<0·83) posterior support.

Fig. 9.

Ancestral area reconstruction of the Steveniella-Himantoglossum clade based on the concatenated plastid regions and using a Bayesian DIVA approach. (A) The resulting tree has nodes displayed as pie charts denoting possible ancestral areas proportional to their Bayesian posterior probability, drawn after 250 000 cycles using ten chains and sampling every 100th generation with 10 % burn-in. Region of origin (displayed as country code) is shown in parentheses. (B) Initial classification of the geographical origin of samples (dots) used in this analysis.