Phylogenetics of tribe Orchideae (Orchidaceae: Orchidoideae) based on combined DNA matrices: inferences regarding timing of diversification and evolution of pollination syndromes

Abstract

Background and aims

Tribe Orchideae (Orchidaceae: Orchidoideae) comprises around 62 mostly terrestrial genera, which are well represented in the Northern Temperate Zone and less frequently in tropical areas of both the Old and New Worlds. Phylogenetic relationships within this tribe have been studied previously using only nuclear ribosomal DNA (nuclear ribosomal internal transcribed spacer, nrITS). However, different parts of the phylogenetic tree in these analyses were weakly supported, and integrating information from different plant genomes is clearly necessary in orchids, where reticulate evolution events are putatively common. The aims of this study were to: (1) obtain a well-supported and dated phylogenetic hypothesis for tribe Orchideae, (ii) assess appropriateness of recent nomenclatural changes in this tribe in the last decade, (3) detect possible examples of reticulate evolution and (4) analyse in a temporal context evolutionary trends for subtribe Orchidinae with special emphasis on pollination systems.

Methods

The analyses included 118 samples, belonging to 103 species and 25 genera, for three DNA regions (nrITS, mitochondrial cox1 intron and plastid rpl16 intron). Bayesian and maximum-parsimony methods were used to construct a well-supported and dated tree. Evolutionary trends in the subtribe were analysed using Bayesian and maximum-likelihood methods of character evolution.

Key Results

The dated phylogenetic tree strongly supported the recently recircumscribed generic concepts of Bateman and collaborators. Moreover, it was found that Orchidinae have diversified in the Mediterranean basin during the last 15 million years, and one potential example of reticulate evolution in the subtribe was identified. In Orchidinae, pollination systems have shifted on numerous occasions during the last 23 million years.

Conclusions

The results indicate that ancestral Orchidinae were hymenopteran-pollinated, food-deceptive plants and that these traits have been dominant throughout the evolutionary history of the subtribe in the Mediterranean. Evidence was also obtained that the onset of sexual deception might be linked to an increase in labellum size, and the possibility is discussed that diversification in Orchidinae developed in parallel with diversification of bees and wasps from the Miocene onwards.

INTRODUCTION

Circumscription of subfamily Orchidoideae (Orchidaceae) has varied extensively in recent decades from a broadly delimited definition comprising all monandrous orchids to a more restricted one in which the terrestrial habit, lack of leaf fibres and presence of root tubers became defining traits (Freudenstein and Rasmussen, 1999; reviewed by Pridgeon et al., 2001). The wide variation in the definition of Orchidoideae has impinged on the tribal number and arrangements proposed. For example, Garay (1972), based mostly on floral characters, divided the subfamily into three tribes, Orchideae (including Diurideae), Diseae and Disperideae, whereas Dressler (1990, 1993), based on vegetative and embryological traits, recognized Diurideae, Diseae (including Disperidae) and Orchideae. Molecular and morphological studies (Pridgeon and Chase, 1995; Kores et al., 1997, 2001) have indicated that some of the characters used by both authors are homoplastic. The abundance of alternative systems has led to considerable disagreement regarding not only subtribal and tribal concepts, but also subfamilial delimitation.

Chase et al. (2003) proposed a series of nomenclatural changes in Orchidoideae based on the results of phylogenetic studies conducted using plastid and nuclear data (Kores et al., 1997; Cameron et al., 1999; Douzerey et al., 1999). Four tribes, Cranichideae, Diurideae, Codonorchideae and Orchideae, were recognized in the subfamily. Chase et al. (2003) also recognized three subtribes in tribe Orchideae: Brownleeinae, Disinae and Orchidinae. This study focuses on the last group.

In Orchidinae, two broad-scale molecular analyses conducted by Pridgeon et al. (1997) and Bateman et al. (1997, 2003) using ribosomal internal transcribed spacer (ITS) regions enabled a new understanding of relationships among genera and species. There were three main taxonomic conclusions from these studies. (1) Orchis as previously defined, largely based on floral traits, was triphyletic. As a consequence, the authors reassigned species from Orchis sensu lato (s.l.) to expanded concepts of Neotinea and, especially, Anacamptis. Also, the monospecific genus Aceras was included in Orchis sensu stricto (s.s.). (2) Gymnadenia as previously defined was paraphyletic unless genus Nigritella was included, and it was therefore expanded. (3) Finally, Coeloglossum was nested within the Dactylorhiza clade and transferred to the latter.

Despite several morphological and molecular studies published on subtribe Orchidinae, a comprehensive phylogenetic study using molecular data from all three plant genomes is still lacking. Such a robust phylogenetic analysis could offer important insight into the evolution of this complex group of plants, especially regarding their pollination syndromes.

Orchids offer some of the most striking examples of plant–pollinator relationships, and this association has intrigued scientists for at least the last 200 years (e.g. Darwin, 1888; reviewed by Van der Cingel, 1995, 2001; Micheneau et al., 2010). Although plant–pollinator relationships are considered to be mostly mutualistic, with deception evolving independently in a limited number of species and families, deception is widespread in orchids. Thus, around one-third of orchid species use this strategy (e.g. Cozzolino and Widmer, 2005; Smithson, 2009). Orchids are also the only plants in which sexual deception of pollinators has evolved, and this strategy is present in orchid genera from Australasia, Europe and Central and South America (reviewed in Gaskett, 2011). Deception of pollinators (using sexual or food decoys as attractants) is clearly the norm among the Mediterranean and European orchids that constitute the core of subtribe Orchidinae. The evolutionary origin and significance of deception, both in this subtribe and in orchids more generally, as well as the influence of the different pollination syndromes in diversification and extinction rates in orchids, have been widely discussed throughout the last decade (e.g. Cozzolino and Widmer, 2005; Jersáková et al., 2006; Cozzolino and Scopece, 2008; Schiestl and Cozzolino, 2008; Smithson, 2009; Vereecken et al., 2010; Gaskett, 2011). As a result, several hypotheses have been proposed: (1) cheating is the ancestral state in Orchidinae, with rewards evolving independently several times, (2) pollination by deceit is one element that has triggered floral and taxonomic diversity in these orchids, (3) deceit as a strategy selected by each orchid lineage (sexual or food deception) has a strong effect on development of pre- or post-mating barriers and (4) sexual deception has evolved independently in different lineages of food-deceptive orchids, with alkene production playing a key role in the process (Schiestl et al., 2000).

Despite many studies published in the past few years on evolution of pollination systems in Orchidinae, there is still need for further studies that combine phylogenetic and ecological information (Schiestl and Cozzolino, 2008; Smithson, 2009). In particular, it is important to improve resolution and sampling of taxa and DNA regions in the phylogenetic studies of these taxa (Cozzolino and Widmer, 2005). The recent development of Bayesian methods of character reconstruction (Pagel et al., 2004; Ronquist, 2004; Pagel and Meade, 2006) opens up new insight into the evolution of pollination systems. Finally, new discoveries in the orchid fossil record, including fossil pollinia attributed to a Goodyerinae species and fossil leaves belonging to genera Dendrobium and Earina (Conran et al., 2009; reviewed by Gustafsson et al., 2010), have allowed dating of the main diversification events in the orchid phylogenetic tree (Gustafsson et al., 2010). This temporal framework should also be considered in studies of the evolution of pollination systems.

The purpose of this study is two-fold. First, we aim to unravel the phylogenetic relationships (including the temporal framework) within Orchideae (with special emphasis in subtribe Orchidinae) using the plastid rpl16 intron (sequences generated in this study), nuclear ribosomal (nr)ITS (Bateman et al., 2003) and mitochondrial cox1 (Inda et al., 2010). The different regions were used jointly and separately in the phylogenetic analyses in order to detect possible reticulation events in the evolution of Orchidinae. Second, the dated phylogenetic tree produced here was used to study the evolution of a series of characters of relevance to pollination biology of these taxa.

The plastid rpl16 gene is involved in the synthesis of a chloroplast ribosomal protein, and it consists of two exons separated by a relatively large group II intron (reviewed by Jordan et al., 1996). This intron was chosen for this phylogenetic study due to the high level of variability that it generally displays (Wolfe et al., 1987). In orchids, this region has been little used; Wallace (2006) used it to analyse the biology of some Platanthera populations, and Pillon et al. (2006) employed it to assess the relationship between taxonomic and phylogenetic diversity in the widespread Eurasian genus Dactylorhiza. The combined analysis of independent loci from different genomes (nuclear, mitochondrial and plastid) has been considered a good approach to analyse reticulate evolution events in groups of complex taxa (Linder and Rieseberg, 2004).

MATERIALS AND METHODS

Plant materials

In total, 118 samples from 103 taxa representing two tribes (sensu Dressler, 1993; Orchideae and Diseae) belonging to subfamily Orchidoideae were included in the survey. Collection data for the sampled species and voucher specimen numbers are presented in the Appendix. All vouchers cited in the Appendix are deposited in the Herbarium of the Royal Botanic Gardens, Kew (K). For the Dactylorhiza incarnata (diploid) species complex, we have indicated the subspecies names in the figures, but these are listed in the text and tables as, for example D. incarnata subsp. cruenta. Due to the complexity of how to treat the taxa sometimes treated as subspecies in the Dactylorhiza maculata complex and their allotetraploid progeny, we treat all these at the species rank to keep matters simple, but we acknowledge that this is not uniformly practised by all orchid taxonomists.

DNA isolation, amplification and sequencing

DNA was extracted from fresh or silica-dried leaves following a modification of the 2× CTAB procedure of Doyle and Doyle (1987). In this process, genomic DNA was first precipitated with 100 % ethanol or isopropanol, then chilled for at least 24 h at 4 °C and pelleted and purified by centrifugation through a CsCl₂-ethidium bromide gradient with subsequent dialysis. Finally, DNA was resuspended in Tris-EDTA buffer, pH 8, and were stored at –80 °C in the DNA Bank of the Royal Botanic Gardens, Kew (http://data.kew.org/dnabank/homepage.html).

The rpl16 intron was amplified for all taxa listed in the Appendix using the primers designed by Jordan et al. (1996). The PCR protocol comprised 28 cycles that included 1 min denaturation at 94 °C, 30 s annealing at 48 °C and a final extension of 1 min at 72 °C, starting with 3 min at 94 °C and ending with an extension of 7 min at 72 °C. Amplified products were purified using QIAquick (QIAgen, Crawley, UK) or NucleoSpin Extract (Macherey-Naggel, Düren, Germany) columns following the manufacturers' protocol. Cleaned products were then sequenced using the BigDye Terminator Cycle Sequencing Ready Reaction v3·1 kit (Applied Biosystems) in an Applied Biosystem 3710 automated sequencer. All sequences obtained are available in GenBank (Appendix).

Sequence alignment and phylogenetic analyses

The forward and reverse electropherograms were assembled and edited, and the edited sequences were tentatively aligned, using the software Sequencher v.4 (Gene Codes, Ann Arbor, MI, USA). Subsequently, rpl16 intron, ITS and cox1 intron sequences were aligned separately using the MUSCLE algorithm (Edgar, 2004) as implemented in the software SeaView v4 (Gouy et al., 2010). Alignments were later edited by eye with the software Se-Al v.2·0a11 (Rambaut, 2002). One hundred and eighteen new rpl16 intron sequences were generated for this study. Another 96 ITS and 94 cox1 intron sequences were available from previous studies (Bateman et al., 2003; Inda et al., 2010; the latter also available in GenBank; Appendix). In the combined data sets, missing sequences were coded as absent. Combined analyses were conducted with and without missing data. The combined data set without missing data comprises 92 sequences from 92 taxa.

Phylogenetic analyses were carried out on both individual and combined (plastid, nuclear and mitochondrial DNA) data sets using parsimony and Bayesian methods. The former was implemented using the software PAUP v.4·0b (Swofford, 2000) with Disperis lindleyana selected as outgroup (subtribe Coryciinae) based on the results of Douzerey et al. (1999), who showed that Disperis occupies the position of sister to a clade including subtribes Orchidinae, Disinae and Brownleeinae. All maximum-parsimony analyses were conducted with all sites equally weighted and 1000 random addition sequence replicates. Tree bisection–reconnection (TBR) branch swapping with a limit of ten trees per step was selected as the search method. Internal support was assessed by means of equal-weights maximum-parsimony bootstrap analyses (Felsenstein, 1985; 1000 pseudoreplicates, each consisting of a heuristic search using 1000 random sequence addition replicates).

Bayesian analysis was carried out using the software MrBayes v. 3·0B4 (Huelsenbeck and Ronquist, 2001). The GTR + I + G model of evolution was used in all individual analyses and all DNA partitions in the combined analysis were based on the software MrModelTest v2·3 (Nylander, 2004). Gaps were coded in the matrix as presence/absence variables following the simple method proposed by Simmons and Ochotorena (2000) as implemented in the software SeqState (Müller, 2005). Gaps were assumed to follow the binary model of evolution (Ronquist et al., 2005) and were included in the Bayesian analyses following Dwivedi and Gadagkar (2009). The analysis of each separate data set was performed with 5000 000 generations initiated with a random starting tree, sampling every 1000 generations and allowing the program to estimate the likelihood parameters required. Stationarity was assessed using the web-based software AWTY (Nylander et al., 2008). Results collected prior to stationarity were discarded as burn-in. The combined analysis was conducted with 10 000 000 generations and sampling every 2000 generations (other details as above). Results are presented as the 50 % majority-rule consensus tree constructed in PAUP.

Divergence dating analysis using BEAST

A Bayesian divergence analysis based on the relaxed molecular clock approach was conducted with the combined matrix (excluding missing data) using the software BEAST v1·6·1 (Drummond and Rambaut, 2007). Input data were compiled using the program BEAUTI v1·6·1. Three partitions (nuclear, plastid and mitochondrial) were included in the analysis. The GTR + I + G model was imposed in all partitions, with substitution models unlinked across partitions; an uncorrelated log-normal, relaxed-clock model with a Yule tree prior was used. The model priors were set as follows. (1) Age for the divergence between Orchis and Platanthera: normal prior distribution with mean 17·00 Ma and standard deviation of 5·65. (2) Age for the divergence between Habenaria and Orchis + Platanthera: normal prior distribution with mean 24·00 Ma and standard deviation 6·3. Secondary calibration points (means and standard deviations) were taken from Gustafsson et al. (2010) and A. L. S. Gustafsson (University of Oslo, Norway, pers. comm.). (3) Substitution rates for all partitions were established according to Wolfe et al. (1987). Lognormally distributed age priors were also tested following Ho and Phillips (2009), with no change in the results. Four independent analyses were run for a total of 80 000 000 generations. To test the influence of priors on posterior estimates, one additional chain was run for 80 000 000 generations without data following Popp et al. (2011). Log files were analysed using TRACER v1·5 (Rambaut and Drummond, 2007) to assess convergence and confirm that the effective sample size for all parameters was >200 (Drummond et al., 2007; Gustafsson et al., 2010). Resulting trees were combined using LogCombiner v.1·6·1 (Drummond and Rambaut, 2007) with a burn-in of 25 %. A maximum-credibility tree was then produced using treeAnnotator v.1·5·3 (Drummond and Rambaut, 2007).

Analyses Of Pollination System Evolution

We considered three characters in our analysis of the evolution of pollination systems, following Schiestl and Cozzolino (2008): labellum surface area (mm²), pollination strategy and predominant pollinator. Information on all characters was retrieved from the literature (Rose, 1948; Van der Cingel, 1995, 2001 and references therein; Bell, 1997; Cozzolino et al., 2001; Pridgeon et al., 2001; Kull and Arditti, 2002; Cozzolino et al., 2005; Delforge, 2006; Pedersen and Faurholdt, 2007; Valterová et al., 2007; Schiestl and Cozzolino, 2008 and references therein; Cheng et al., 2009; Efimov et al., 2009; Jacquemyn et al., 2009; Gaskett, 2011 and references therein). The data collected are included in Table 1. Estimation of labellum area (Table 1) was conducted following the techniques of Schiestl and Cozzolino (2008). We used the software BayesMultiState included in the package BayesTraits (Pagel et al., 2004) to reconstruct the ancestral pollination strategies and predominant pollinators in the main clades of the combined analysis. From the MrBayes analysis, 510 randomly selected trees were used as input. Disperis lindleyana, Huttonaea grandiflora and Holothrix scopularia were pruned from all trees using the R-based package APE v.2·7-3 (Paradis et al., 2004) to get a fully resolved root in all trees as required in BayesMultiState (Pagel et al., 2004). Pollination strategies were coded as: food deception (0); food reward (1); intermediate food deception/reward (Himantoglossum; 2); sexual deception (3); sleeping sites (Serapias; 4). As regards predominant pollinators, two coding strategies were tested: Coding 1: Lepidoptera (0); male bees (1); female bees (2); Hymenoptera (both sexes) (3); several insect groups (4); Coleoptera (5); self-pollination (6). Coding 2: Lepidoptera (0); Hymenoptera (1); several insect groups (2); Coleoptera (3); self-pollination (4). The two strategies differ in their treatment of Hymenoptera; we were interested in knowing if our more detailed method (coding 1) changed the overall results.

Table 1.

Pollination-related characters analysed: pollination syndrome (plus presence of nectaries), predominant pollinators and labellum area calculated as (length × width)/2

Genus	Species	Nectaries	Pollination syndrome	Pollinators	Labellum area (mm2)
Amerorchis Hultén	A. rotundifolia (Banks ex Pursh) Hultén	No	–	–	22
Amitostigma Schltr.	A. gracile (Blume) Schltr.	–		–	7·48
Anacamptis Rich.	A.boryi (Rchb. f.) R.M.Bateman, Pridgeon & M.W.Chase	–	–	–	48·87
	A. champagneuxii (Barnéoud) R.M.Bateman, Pridgeon & M.W.Chase	–	–	–	45·31
	A. collina (Banks & Sol. ex Russell) R.M.Bateman, Pridgeon & M.W.Chase	No	Food deception	Bees	–
	A. fragrans (Pollini) R.M.Bateman,	Yes	Food reward	Hymenoptera	26·25
	A. laxiflora (Lam.) R.M.Bateman, Pridgeon & M.W.Chase	–	–	–	116·37
	A. morio (L.) R.M.Bateman, Pridgeon & M.W.Chase	No	Food deception	Female bees	60
	A. palustris (Jacq.) R.M.Bateman, Pridgeon & M.W.Chase	–	–	–
	A. papilionacea (L.) R.M.Bateman, Pridgeon & M.W.Chase	No	Food deception	Male bees	148·75
	A. pyramidalis (L.) Rich.	Yes	Food reward	Lepidoptera
	A. sancta (L.) R.M.Bateman, Pridgeon & M.W.Chase	Yes	Food reward	Hymenoptera	29·69
Brachycorythis Lindl.	B. kalbreyeri Rchb.f.	–		–	135
Chamorchis Richard	C. alpina (L.) Rich	–	–	Ichneumoids, Diptera, Coleoptera	12·25
Cynorkis Thouars	Cynorkis sp.	–		Lepidoptera	–
Dactylorhiza Necker ex Nevski	D. alpestris (Pugsley) Aver.	No	–	Diptera, Coleoptera, Hymenoptera	75
	D. aristata (Fisch. ex Lindl.) Soó	No	–	–	–
	D. cordigera (Fr.) Soó	No	–	Diptera, Coleoptera, Hymenoptera	90
	D. elata (Poir.) Soó	No	–	–	54·9
	D. foliosa (Soland. ex Lowe) Soó	No	–	–	92·2
	D. fuchsii (Druce) Soó	No	–	Coleoptera	48
	D. iberica (M.Bieb. ex Wild.) Soó	No	–	–	31·87
	D. incarnata (L.) Soó	No	–	Bees	22·8
	D. incarnata	No	–	Bees	22·8
	D. incarnata subsp. cruenta (O.F.Muell.) P.D.Sell	No	–	–	22·7
	D. incarnata subsp. ochroleuca (Wüstnei ex Boll) P.F.Hunt & Summerh.	No	–	–	21·8
	D. incarnata subsp. ochroleuca	No	–	–	22·8
	D. incarnata subsp. ochroleuca	No	–	–	23·8
	D. incarnata subsp pulchella (Druce) Soó	No	–	Bees	21·9
	D. maculata (L.) Soó (in lit. as D. ericetorum (Linton) Aver.	No	–	Diptera, Coleoptera, Hymenoptera	42·5
	D. maculata	No	Food deception	Diptera, Coleoptera Hymenoptera	50·63
	D. praetermissa (Druce) Soó	No	–	Bees	46
	D. romana (Sebast.) Soó	No	–	Bees	68·25
	D. saccifera (Brongn.) Soó	No	–	Bees	55·12
	D. sambucina (L.) Soó	No	–	Bees	40
	D. viridis (L.) R.M.Bateman, Pridgeon & M.W.Chase	Yes	Food reward	Diptera, Coleoptera Hymenoptera	9·37
Galearis Raf.	G. cyclochila (Franch. & Sav.) Soó	–	–	–	21·25
	G. spectabilis (L.) Raf.	Yes	Food reward	Bees	73·5
Gennaria Parl.	G. diphylla Parl.	–	–	–
Gymnadenia R.Br.	G. conopsea (L.) R.Br.	Yes	Food reward	Lepidoptera	11·88
	G. conopsea (L.) R.Br. subsp borealis (Druce) F.Rose	Yes	Food reward	Lepidoptera	11·88
	G. nigra Rchb.f.	Yes	Food reward	Lepidoptera	12
Habenaria Willd.	H. arenaria Lindl.	Yes	Food reward	Lepidoptera	–
	H. delavayi Finet	Yes	Food reward	Lepidoptera	–
	H. procera Lindl.	Yes	Food reward	Lepidoptera	–
	H. socotrana Balf.f.	Yes	Food reward	Lepidoptera	–
	H. tibetica Schltr	Yes	Food reward	Lepidoptera	–
	H. tridactylites Lindl.	Yes	Food reward	Lepidoptera	–
	H. viridiflora R.Br	Yes	Food reward	Lepidoptera	–
Herminium L.	H. monorchis R.Br.	Yes	Food reward	Diptera, Coleoptera Hymenoptera	2·43
Himantoglossum Spreng	H. adriaticum H.Baumann	No/yes	Intermediate food reward/food deception	Female bees	163·8
	H. comperianum (Steven) P.Delforge	No/yes	Intermediate food reward/food deception	Female bees	–
	H. caprinum Spreng.	No/yes	Intermediate food reward/food deception	Female bees	267·3
	H. hircinum Spreng.	No/yes	–	Female bees	166·25
	H. robertianum (Loisel.) P.Delforge	No/yes	Intermediate food reward/food deception	Female bees	60·375
Holothrix Linld.	H. scopularia Rchb.f.	–	–	–	–
Neotinea Rchb f. & Poll.	N. lactea (Poir.) R.M.Bateman, Pridgeon & M.W.Chase	No	Food deception	Diptera	40·38
	N. maculata (Desf.) Stearn	Yes	Food reward	Autogamous	–
	N. tridentata (Scoo) R.M.Bateman, Pridgeon & M.W.Chase	No	Food deception	Bees	25·4
	N. ustulata (L.) R.M.Bateman, Pridgeon & M.W.Chase	No	Food deception	Diptera	16·82
Neolindleya Kraenzl	N. camtschatica (Cham.) Nevski	No	No	Autogamous	–
Neottianthe Schltr.	N. cucullata Schltr.	–	–	–	–
Ophrys L.	O. aesculapii Renz	No	Sexual deception	Male bees	63
	O. apifera Huds.	No	Sexual deception	Male bees	–
	O. araneola Rchb.	No	Sexual deception	Male bees	33·9
	O. bombyliflora Link	No	Sexual deception	Male bees	54
	O. cretica (Vierh.) E.Nelson	No	Sexual deception	Male bees	72
	O. fusca Link	No	Sexual deception	Beetle	129·5
	O. insectifera L.	No	Sexual deception	Wasps	42
	O. iricolor Desf.	No	Sexual deception	Male bees	143·4
	O. lutea Cav.	No	Sexual deception	Male bees	128
	O. regis-ferdinandii (Renz) Buttler	No	Sexual deception	Male bees	–
	O. heldreichii Schltr.	No	Sexual deception	Male bees	131·7
	O. speculum Link	No	Sexual deception	Wasps	–
	O. sphegodes Mill.	No	Sexual deception	Male bees	91
	O. spruneri Nyman	No	Sexual deception	Male bees	112·3
	O. tenthredinifera Willd.	No	Sexual deception	Male bees	93,7
Orchis L.	O. anatolica Boiss.	No	Indeterminate	Bees	52·5
	O. anthropophora All.	Yes	Food reward	Coleoptera	39·06
	O. brancifortii Bivona-Bernardi	No	Food deception	–	9
	O. italica Poir.	No	Food deception	Female bees	64
	O. mascula L.	No	Food deception	Female bees	78
	O. militaris L.	No	Food deception	Bees	62·5
	O. pauciflora Ten.	No	Sexual deception	Female bees	98
	O. purpurea Huds.	No	Food deception	Bees & Diptera	119·6
	O. quadripunctata Cirillo ex Ten.	No	Food deception	Diptera	25·5
	O. simia Lam.	No	Food deception	Bees	–
	O. sitiaca (Renz) P.Delforge	No	Food deception	Male bees	52·25
Pecteilis Raf.	P. sagarikii Seidenf.	Yes	Food reward	Lepidoptera	–
Platanthera Rich.	P. bifolia (L.) Rich.	Yes	Food reward	Lepidoptera	21·94
	P. chlorantha Cust. ex Rchb.	Yes	Food reward	Lepidoptera	18
	P. grandiflora (Bigelow) Lindl.	Yes	Food reward	Lepidoptera	175
	P. hyperborea Lindl.	Yes	Food reward	Lepidoptera	5·5
Pseudorchis Seguier	P. straminea (Fernald) Soják	Yes	Food reward	Lepidoptera	5·04
Satyrium Sw.	S. humile Lindl.	Yes	Food reward	–	–
	S. nepalense D.Don	Yes	Food reward	–	13·75
Serapias L.	S. cordigera L.	No	Sleeping site	Male bees	396
	S. lingua L.	No	Sleeping site	Male bees	126·5
	S. neglecta de Not.	–	–	–	192
	S. orientalis (Greuter) H.Baumann & Künkele	–	–	–	47·25
	S. parviflora Parl.	No	–	Autogamy	47,25
	S. vomeracea Briq.		–	–	47,25
Stenoglottis Lindl.	S. longifolia Hook.	–	–	Autogamous	–
	S. woodii Schltr.	–	–	Autogamous	–
Steveniella Schltr.	S. satyrioides Schltr.	No	Food deception	Wasps (eusocial)	19·5
Traunsteinera Rchb.	T. globosa Rchb.	No	Food deception	Bees, Coleoptera, Lepidoptera	39

Information was retrieved from the literature (Rose, 1948; Van der Cingel, 1995, 2001 and references therein; Bell, 1997; Cozzolino et al., 2001; Pridgeon et al., 2001; Kull and Arditti, 2002; Cozzolino et al., 2005; Delforge, 2006; Pedersen and Faurholdt, 2007; Valterová et al., 2007; Schiestl and Cozzolino, 2008 and references therein; Cheng et al., 2009; Efimov et al., 2009; Jacquemyn et al., 2009; Gaskett, 2011 and references therein). –, Unknown.

In BayesMultiState, we ran maximum-likelihood (ML) and Bayesian ancestral state reconstructions using the same input file. In ML and Bayesian analyses we followed the procedure described by Litman et al. (2011). Each Bayesian chain was run for 50 000 000 generations (burn-in 10 000 000). Each analysis was repeated four times, and the harmonic means of the likelihoods were averaged. Different models were compared using the average difference in likelihood and Bayes Factors as described in the BayesTraits manual. Decisions over models were taken following Raftery (1996) as cited in Barbeitos et al. (2010). Ancestral state reconstructions were carried out for nodes indicated in Fig. 1A.

Fig. 1.

(A) Summarized Bayesian tree obtained for the rpl16 intron data set (the complete tree is shown in Fig. 2). Posterior probabilities are indicated above branches. Letters represent the nodes where the ancestral state of the studied characters was assessed. (B) Summarized BEAST consensus tree (the complete tree is shown in Figure 4). Posterior probabilities are indicated above branches. Numbers in bold indicate the nodes where transition in labellum size was assessed.

The software Maticce (Hipp and Escudero, 2010) and OUCH (King and Butler, 2009) were used to assess the probability of clade-specific transitions in labellum area. This analysis was conducted in an information theoretic network, where transitions were modelled as shifts in the equilibrium value of a set of multiple-optimum Ornstein–Uhlenbeck models (Hipp, 2007; Escudero et al., 2010). Seven nodes at which transitions in labellum area equilibrium were permitted (Fig. 1B) were designated following Escudero et al. (2010). Overall, 2⁷ models were tested on each of 110 trees randomly subsampled from the Bayesian divergence analysis conducted with BEAST. These trees were selected to meet the ultrametricity requirement indicated in the Maticce manual. The input trees were pruned to eliminate all terminals for which the labellum area was unknown (66 terminals were included in the analysis; Table 1). Pruning of trees was conducted with the package APE (Paradis et al., 2004). The 128 models entailing all possible changes at these seven nodes were assessed following Escudero et al. (2010).

RESULTS

Analysis of the rpl16 intron data set

The aligned rpl16 intron matrix of 118 samples (107 taxa) comprised 1810 characters, of which 1066 (58·6 %) were constant and 441 (24·3 %) were potentially parsimony-informative. Parsimony and Bayesian analyses (results not shown) yielded the same topology, but with lower bootstrap percentages (BP) than posterior probabilities (PP). The heuristic search found 2680 equally most-parsimonious trees that were 1771 steps long (consistency index = 0·46, retention index = 0·77). The strict consensus tree shows that most subclades were moderately to highly supported (Fig. 2).

Fig. 2.

Consensus tree of 2680 most-parsimonious trees obtained for the plastid rpl16 intron data set (118 samples, 1810 characters, informative characters 24·3 %). Length 1771 steps, consistency index = 0·46; retention index = 0·77. Numbers above branches indicate maximum-parsimony bootstrap percentages (BP)/Bayesian inference posterior probability support (PP). Dashed lines represent BP < 90. *BP and PP of 100.

Orchideae (sensu Dressler, 1993) received strong support in this analysis (95 BP). The expanded concept of Orchideae sensu Chase et al. (2003) was not directly tested in this study due to insufficient outgroups to provide an appropriate evaluation. Orchidinae sensu Dressler (1993) were moderately supported (90 BP), although several species that Dressler (1993) ascribed to Habenariinae are included in this clade (Neolindleya camtschatica, Amitostigma gracile and Pseudorchis straminea). In Orchidinae, all genera as redefined in Pridgeon et al. (1997) and Bateman et al. (2003) are monophyletic, apart from Platanthera which is unresolved. However, bootstrap support for these clades is fairly variable. Ophrys, Himantoglossum and Serapias are each strongly supported (100 BP), whereas genera such as Dactylorhiza (including Coeloglossum) and Gymnadenia (including Nigritella) receive moderate to low support (84 and 63 BP, respectively). Orchis s.s. and Anacamptis s.l. (sensu Bateman et al., 1997) received moderate to weak support (90 and 69 BP, respectively). Conversely, the expanded Neotinea is strongly supported (98 BP).

At generic and lower levels, most relationships exhibited in this analysis support results published previously (e.g. Bateman et al., 2003). Nevertheless, there are several groupings worth noting. In Anacamptis there is strong support (100 BP) for a clade including A. laxiflora and A. palustris, whereas the rest of the taxa are unresolved within the sister clade (90 BP). In Himantoglossum, a clade including all species with long, coiled lips (H. hircinum, H. caprinum and H. adriaticum) is strongly supported (100 BP), with taxa belonging to the former genera Barlia (88 BP) and Comperia (100 BP) being successively sister to that clade.

In Ophrys, several moderately to strongly supported clades can be observed in our results, most of them consistent with previous research based on plastid, nuclear and mitochondrial DNA (Soliva et al., 2001; Devey et al., 2008; Inda et al., 2010). Among them, it is worthwhile highlighting the association between O. speculum and O. regis-ferdinandii (98 BP), the latter considered a subspecies of the former by some authors (e.g. Pedersen and Faurholdt, 2007). In this analysis, O. insectifera is sister to the rest of the genus.

In Neotinea sensu Bateman et al. (1997), N. maculata is sister to the rest of the genus, whereas the three species formerly included in Orchis, namely N. ustulata, N. tridentata and N. lactea, are well supported (98 BP).

The groupings found in Orchis are consistent with the findings of Bateman et al. (1997). Thus, we recovered a clade comprising O. anatolica, O. brancifortii, O. sitiaca and O. quadripunctata (100 BP) consistent with the ‘O. anatolica’ subgroup of Bateman et al. (1997). We also recovered a strongly supported (100 BP) clade equivalent to the O. mascula clade of Bateman et al. (1997), including O. pauciflora and O. mascula. Finally, the O. militaris (anthropomorphic) group of Bateman et al. (1997) comprising O. simia, O. purpurea, O. militaris and O. anthropophora (the last often treated as Aceras) is also found in our results (100 BP). The position of the last member of the O. militaris group, O. italica, is unresolved with regard to the other two mentioned above. The pair Traunsteinera and Chamorchis (100 BP) is sister to Orchis sensu Bateman et al. (1997, 2003). The clade comprising Dactylorhiza (including Dactylorhiza viridis, former Coeloglossum viride) is not strongly supported (84 BP). Moreover, most internal groups in this clade are weakly to moderately supported, with two exceptions: (1) a ‘D. incarnata’ group (100 BP, in Fig. 2 labelled solely with the subspecies epithets), including D. incarnata subspp. incarnata, pulchella, cruenta and ochroleuca; and (2) a large clade (98 BP) including nine Dactylorhiza species (D. aristata, D. elata, D. foliosa, D. praetermissa, D. saccifera, D. fuchsii, D. maculata, D. alpestris and D. cordigera).

Combined analysis: ITS + rpl16 + cox1

There were no strongly supported incongruent results in the separate analyses of ITS nrDNA, rpl16 intron and cox1 intron matrices, and therefore we combined them in a joint analysis of 92 taxa. The combined data matrix included 4625 characters divided in four partitions (ITS nrDNA, 1–797; rpl16 intron, 798–2610; cox1 intron 2611–4100; gaps, 4101–4625). The combined data set was analysed using parsimony (results not shown, although bootstrap percentages from these results are provided in Fig. 3) and Bayesian methods (Fig. 3). Numbers above branches indicate posterior probabilities (PP). The tree recovered in the combined analysis is similar to that based on the rpl16 intron alone (Fig. 2). Nevertheless, resolution is much higher in the combined tree, especially polytomies near the basal nodes, although support in this sparsely sampled part of the tree is still low (Fig. 3). Internal clades are much better supported in the combined tree as compared with the rpl16 intron-alone analysis. Again the expanded concept of Orchideae of Chase et al. (2003) has not been tested in this analysis. However, the few representatives of former Diseae included in the analysis are well supported as non-monophyletic. The restricted concept of Orchideae (Dressler, 1993) receives lower support than in the parsimony analysis (78 PP vs. 95 BP), although support between different analyses/methods is not comparable. Subtribe Orchidinae sensu Dressler (1993) receives strong support (100 PP), whereas the expanded concept of Chase et al. (2003) receives moderate support (78 PP). The distinction between subtribe Orchidinae and former subtribe Habenariinae (sparsely sampled) is not supported (Habenariinae are paraphyletic to Orchidinae).

Fig. 3.

Bayesian tree of Orchideae based on combined nuclear (ITS), plastid (rpl16 intron) and mitochondrial (cox1 intron) DNA (92 samples, 4625 characters). Numbers above branches indicate maximum parsimony bootstrap percentages (BP)/Bayesian inference posterior probability (PP). Dashed lines represent BP < 90. *BP and PP of 100.

Within Orchidinae, genera as defined by Bateman et al. (1997, 2003) are all monophyletic and much more strongly supported than in the rpl16 intron analysis (Figs 2 and 3). Relationships among genera are more or less similar in these analyses, with a few exceptions. (1) Pseudorchis is sister to Amitostigma (<50 BP) in the Gymnadenia clade (also <50 BP) in the rpl16 intron topology, whereas in the combined analysis Pseudorchis is sister (weakly supported) to Neolindleya + Platanthera + Galearis, and Amitostigma is sister to the rest of core Orchidinae (excluding former subtribe Habenariinae; 100 PP). (2) The rpl16 intron analysis failed to resolve relationships among representatives of former Habenariinae, but it showed that the subtribe is not monophyletic. The combined analysis showed that Habenaria is paraphyletic as well. Gennaria, Pecteilis and Herminium would have to be included to make Habenaria monophyletic. (3) Traunsteinera and Chamorchis are sister to the Orchis clade in the rpl16 intron analyses, whereas they are sister to the Orchis + Neotinea + Ophrys + Himantoglossum + Serapias + Anacamptis clade in the combined analysis (100, 97 and 96 PP).

Within genera, the main differences between the rpl16 intron and the combined results are the stronger support and greater resolution in the latter. Nevertheless, there are some other notable contrasts. (1) In Anacamptis two additional clades can be distinguished, one composed of A. collina, A. fragrans and A. sancta and a second comprising A. papilionacea, A. boryi, A. champagneuxii and A. morio. The position of A. pyramidalis is not well resolved in this clade. (2) In Ophrys, two major clades roughly representing sections Pseudophrys and Ophrys (Fig. 3; originally published as Euophrys; Godfery, 1928) are observed, with section Ophrys being clearly paraphyletic (Soliva et al., 2001). (3) Orchis is almost equally well resolved in both analyses, the only difference being the position of O. italica, which joins the anthropomorphic (O. militaris) clade in the combined analysis. (4) Gymnadenia (including former Nigritella) and Dactylorhiza (including former Coeloglossum) received strong support in the combined analysis (100 and 99 PP, respectively). Within the latter, D. incarnata subsp. pulchella and D. viridis occupy positions successively sister to the rest of the genus. Resolution within Dactylorhiza is poor, with most clades showing only weak to moderate support. (5) In the combined analysis, the clades Dactylorhiza + Gymnadenia and Pseudorchis + Neolindleya + Galearis + Platanthera are sister clades (unresolved in the rpl16 intron analysis). Within the latter, Platanthera is sister to Galearis + Neolindleya, a relationship already observed by Bateman et al. (2009).

Time-calibrated phylogenetic tree

The maximum-credibility tree based on the relaxed molecular clock analysis of Orchideae using the results from Gustafsson et al. (2010) as secondary calibration points is shown in Fig. 4. In this analysis, the topology recovered was also highly supported and equal to that obtained in the Bayesian analysis of the combined data set. The few exceptions represent nodes where PP are low. These differences are restricted to three parts of the tree: (1) the nodes near the root of the tree; (2) the Dactylorhiza clade, in which D. aristata and D. romana occupy different positions compared with the MrBayes analysis; and (3) the Habenariinae clade, in which Herminium monorchis is placed in a different position with respect to previous results.

Fig. 4.

Maximum clade credibility tree of the Bayesian analysis of Orchideae with a relaxed molecular-clock model based on nuclear (ITS), plastid (rpl16) and mitochondrial (cox1) DNA (92 samples, 4625 characters). Secondary calibrations were conducted using the dates published in Gustafsson et al. (2010). Dashed bars represent support lower than 0·8. Divergence times for the main lineages (estimates, standard deviation and intervals) are indicated at nodes.

Our age estimates indicate that extant Orchideae shared a most recent common ancestor (MRCA) in the late Oligocene, around 23 Mya [95 % confidence interval (CI): 16–30 Mya]. This estimate indicates a more recent early diversification of Orchideae than given by Gustafsson et al. (2010). However, resolution and support at the root of our tree is poor (75 PP). As regards subtribe Orchidinae sensu Chase et al. (2003) (including former Habenariinae), the divergence between genus Habenaria and allies and the ‘core’ Orchidinae took place between the early Miocene and late Oligocene (∼22 Mya; 95 % CI: 16–28 Mya). Diversification within the Habenaria + allies clade occurred during the Miocene (∼18–9 Mya; 95 % CI: 24–5 Mya). Our analysis highlights a relatively recent divergence for taxa that are considered different genera within the former Habenariinae. Thus, the separation between Habenaria and Pecteilis took place ∼9 Mya (95 % CI: 5–13 Mya) and the Habenaria–Gennaria divergence happened during the mid-Miocene (∼13 Mya; 95 % CI: 8–19 Mya), although we note again that Habenaria is non-monophyletic and should probably include Pecteilis and Gennaria.

Within ‘core’ Orchidinae, our analysis detected an early successive divergence of the East Asian/Madagascan Cynorkis (∼20 Mya; 95 % CI: 14–25 Mya) and the Asian Amitostigma (∼17 Mya; 95 % CI: 12–21 Mya). Besides, the separation between the two main clades in ‘core’ Orchidinae (Dactylorhiza + Gymnadenia + Galearis + Neolindleya + Platanthera + Pseudorchis vs. Orchis and allies) took place in the mid-Miocene (∼15·6 Mya; 95 % CI: 11·4–20 Mya). The generic divergence within these two clades occurred mostly between the mid-Miocene and mid-Pliocene (∼14·8–3·6 Mya; 95 % CI: 21–1·1 Mya). Finally, our results indicate that the Euro-Mediterranean ‘core’ Orchidinae underwent a rapid species diversification between the late Miocene and mid-Pleistocene (∼8–0·4 Mya; 95 % CI: 11–0·2 Mya). Two genera are remarkable as regards their late diversification: (1) Ophrys, in which most species originated 3·8–0·4 Mya (95 % CI: 5·7–0·2 Mya); and (2) Serapias, in which most species diverged 1·1–0·4 Mya (95 % CI: 2·0–0·2 Mya). These last two genera exhibit different pollination systems compared with the rest of Orchideae.

Pollination system evolution

Ancestral state reconstructions on qualitative data (pollination systems and predominant pollinators) were carried out using ML (data not shown) and Bayesian methods with similar results. In most cases, reconstructions were non-significant or only marginally significant. Nevertheless, significant results were obtained for some nodes. Thus, for pollination systems, there is evidence that the common ancestor of ‘core’ Orchidinae (excluding Habenariinae; node C, Fig. 1A) had food deception as a pollination strategy [PP: 100; Bayes factors (BF): 4·76]. There is also evidence that food deception was the ancestral character in the most recent common ancestor of the clade Anacamptis + Serapias + Himantoglossum + Ophrys (node E, Fig. 1A; PP: 100; BF: 3·45). Finally, there is weak evidence that food deception might have been ancestral to Orchidinae sensu Chase et al. (2003) (including former Habenariinae, node A, Fig. 1A; PP: 62; BF: 2). As regards the trait ‘predominant pollinators’ we obtained significant results only when we used the second coding system [Lepidoptera (0); Hymenoptera without differentiating between males and females (1); generalist plants (2); Coleoptera (3); self-pollination (4)]. According to our results, pollination by Hymenoptera is the ancestral state both for Orchidinae in the broader sense (node A, Fig. 1A; PP: 62; BF: 2·42) and for ‘core’ Orchidinae (node B, Fig. 1A; PP: 100; BF: 3·02).

To test the hypothesis of the existence of clade-specific transitions in labellum area, we designated seven nodes in our tree following the criteria listed in Escudero et al. (2010) (Fig. 1B). Support values estimated using the Akaike information criterion (AIC), the cumulative small-sample Akaike information criterion (AICc) and the Bayes information criterion (BIC) weights highlighted transitions in labellum area (Table 2) at nodes 1 and 3 (Fig. 1B) with differing support. Node 3 is strongly supported, whereas node 1 is only marginally significant. The different models (transition in node 1 vs. transition in node 3 vs. transition in nodes 1 + 3 vs. no transition vs. Brownian motion model) were compared following Burnham and Anderson (2002) and M. Escudero (Morton Arboretum, Chicago, IL, USA, pers. comm.) using AIC values. A ΔAIC > 4 between the ‘no changes’ model and the best model (lowest AIC) was considered as an indicator of strong statistical support for the best model. The model implying two transitions in labellum area (nodes 1 and 3) was selected based of its ΔAIC with respect to the ‘no changes’ model (5·22; Table 3).

Table 2.

Relative support for a shift in labellum area at each of seven selected nodes

	Node at which labellum area transitions were assessed
	1	2	3	4	5	6	7
Cumulative AIC weight	0·8611	0·4524	0·9168	0·2879	0·1789	0·2993	0·6274
Cumulative AICc weight	0·8497	0·4739	0·9072	0·2741	0·1624	0·2708	0·6113
Cumulative BIC weight	0·7811	0·5520	0·8495	0·2182	0·1036	0·1700	0·5240

Support estimates are not conditioned on the existence of each node by model-averaging over all trees. Abbreviations: BIC, Bayesian information criterion; AIC, Akaike information criterion; AICc, small sample Akaike information criterion. Information criteria and weights were calculated following Burnham and Anderson (2002) and Escudero et al. (2010).

Table 3.

Comparison between the different models for labellum area shifts in the phylogenetic tree and the best fitting model (lowest AIC, transition in nodes 1 + 3)

	log-likelihood	d.f.	Sigma2	Theta/alpha	AIC	|ΔAIC|
Transition in nodes 1 + 3	–17·400	7	2·4042	11·4950	49·8	–
Transition in node 3	–21·2993	6	2·0906	8·5663	54·58	4·78
No transitions	–22·5134	5	1·8408	7·0364	55·02	5·22
Brownian model	–35·2180	2	0·7014	1·1736	74·43	24·63

ΔAIC is the difference between the best fitting model and each of the other models. |ΔAIC| values of 4 or more indicate strong support for the best fitting model. AIC, Akaike information criterion.

DISCUSSION

Phylogenetic analyses

In general, our results support the changes in classification and nomenclature proposed by Bateman et al. (1997, 2003) and Chase et al. (2003). At the tribal and subtribal level, we have not included enough samples belonging to former tribe Diseae to draw definite conclusions. However, the paraphyly indicated by those included supports the expanded concept for Orchideae of Chase et al. (2003). The paraphyly of former Habenariinae demonstrated in all analyses reinforces the expanded Orchidinae sensu Chase et al. (2003), and the expanded Orchidinae clade receives strong support in the combined analysis (Fig. 3).

All genera belonging to ‘core’ Orchidinae (excluding former Habenariinae) are monophyletic and strongly supported in the combined analysis. For the rpl16 intron, support is highly variable across genera, with those with the greatest morphological and pollination system differentiation (i.e. Ophrys, Spiranthes and Himantoglossum) receiving the strongest support.

Anacamptis sensu Bateman et al. (1997) was redefined based on the ITS nrDNA results to include all species of Orchis s.l. with 2n = 32 or 36. Our combined analysis shows that the expanded Anacamptis concept is strongly supported (100 PP and BP vs. 56 PP and 69 BP in Bateman et al. 1997, and the rpl16 intron analysis, respectively) and should be employed in taxonomic schemes for the subtribe, although our results do not end the debate over recognizing the A. laxiflora subgroup (see below) as a distinct genus. Within Anacamptis s.l., our results show the same subgroups proposed by Bateman et al. (1997), all of them with moderate to strong support in the combined tree and only modified by the inclusion of new species. Thus, we observe: (1) the A. laxiflora subgroup (A. laxiflora, A. palustris); (2) the A. fragrans subgroup (A. fragrans, A. sancta, A. collina); and (3) the A. morio subgroup (A. champagneuxii, A. morio, A. boryi, A. papilionacea). The position of A. pyramidalis in the clade is not well resolved in the combined or rpl16 intron tree (Figs 2 and 3). In the tree obtained using BEAST, which does not allow polytomies (Fig. 4), this taxon is sister to the A. fragrans clade, although with low support. A. pyramidalis is unique in the genus because it has a single viscidium bearing both pollinia (e.g. Delforge, 2006) and is pollinated by butterflies and moths attracted by nectar (Van der Cingel, 1995). All other members of Anacamptis are pollinated by Hymenoptera, and there are both rewarding and deceptive species (Van der Cingel, 1995). Further research on relationships of A. pyramidalis to the rest of the species is needed to infer the ancestral state of the genus with regard to pollination biology and structure. However, it is remarkable that in this recently diverged (∼9·2 Mya; 95 % CI: 6–12 Mya) clade of Orchidinae shifts in pollination strategies can be observed, highlighting the lability of these characters in orchids (e.g. Smithson, 2009) and in angiosperms more generally (Friedman, 2011). For Serapias, the results of the combined and the rpl16 intron analyses were identical, although, as expected, resolution and support were higher in the combined Bayesian tree. Two main well-supported groups are recovered, but with low levels of divergence: the first group (S. vomeracea, S. lingua and S. parviflora) is characterized by a epichile narrowing towards the base, whereas the second group (S. cordigera, S. neglecta and S. orientalis) has a cordate epichile at least as wide as the hypochile. The dated tree (Fig. 4) highlights a recent origin for all Serapias spp. included in the study (1·18–0·4 Mya; 95 % CI: 2–0·274 Mya), which may explain the weakness of the prezygotic barriers established between taxa (Bellusci et al., 2010). This weakness has led to extensive hybridization in the genus (Bellusci et al., 2010; Hršak et al., 2011), to such an extent that it raises questions about the wisdom of recognizing these as species.

Himantoglossum s.l. constitutes a strongly supported clade in all analyses conducted, which is compatible with the inclusion of Comperia and Barlia in Himantoglossum. Himantoglossum comperianum (formerly Comperia comperiana) and H. robertianum (formerly Barlia robertiana) are successively sister to the other species included in the clade. According to Delforge (2006), these morphologically primitive forms of the genus arose at different times in separate areas of the Mediterranean (eastern Mediterranean basin in the case of H. comperianum). A more derived group comprising H. hircinum, H. caprinum and H. adriaticum, characterized by long lips with purple hairs in the middle part, diverged ∼2·2 Mya (95 % CI: 1·0–3·7 Mya). Of these, H. hircinum and H. adriaticum are recently diverged sister species with numerous morphological and karyological similarities (e.g. karyotypes with mostly metacentric chromosomes; D'Emerico et al., 1990, 1993).

The bee and spider orchids (Ophrys) constitute a clearly monophyletic and recently diversified clade (4·7 Mya; 95 % CI: 2·9–6·7 Mya) in both our plastid and combined trees, consistent with previous studies by Pridgeon et al. (1997) and Soliva et al. (2001). Our results only partially support the infrageneric classification established by Godfery (1928) based on the type of pollination (cephalic, section Ophrys, vs. abdominal, section Pseudophrys). According to Delforge (2006), section Pseudophrys (mostly comprising the O. fusca – O. lutea – O. omegaifera complex) is ancestral in the clade, whereas section Ophrys (as Euophrys) is derived. In our tree, section Pseudophrys, represented by O. lutea, O. fusca and O. iricolor is a well-supported clade, but it appears to have diverged from within section Ophrys (Figs 2 and 3), making the latter paraphyletic. These results are consistent with those published by Devey et al. (2008) based on nuclear/plastid DNA sequences and AFLP fingerprinting data. Other internal divisions of Ophrys based on morphology (Nelson, 1968; Devillers and Devillers-Terschuren, 1994; Delforge, 2006) are also not supported by our results.

Diversification within Ophrys is, according to our results, recent (4·6–0·47 Mya; 95 % CI: 6·7–0–4 Mya), with most taxa arising in the late Pliocene or early to mid-Pleistocene, which helps to explain the lack of clear genetic barriers within Ophrys, in which hybridization is extensive even among species belonging to different sections (Devey et al., 2008). In the combined tree (Fig. 3), Ophrys spp. are grouped in two main clades with strong support (100 PP). These groups are largely consistent with those in Soliva et al. (2001) using ITS nrDNA and plastid trnL-F spacer/intron data. The minor differences observed are most likely due to the different species selected in these studies. One of the groups (roughly equivalent to groups F–I in Devey et al., 2008) is fully resolved and has high support for all internal nodes. Ophrys insectifera, a widespread, wasp-pollinated taxon, is sister to this clade in the combined tree (Fig. 3), whereas it is sister to rest of the genus in the rpl16 intron tree (Fig. 2; >50 BP). This latter position is more consistent with the results from Devey et al. (2008). Self-pollinating O. apifera (Darwin, 1888) and the bee-pollinated O. araneola, O. cretica, O. spruneri, O. aesculapii and O. heldreichii are also members of this group. The second group (roughly equivalent to groups B–E in Devey et al., 2008) is less resolved, and some of its subclades have lower PP. Two main subclades can be distinguished, one composed of closely related O. speculum and O. regis-ferdinandii and another including O. tenthredinifera, O. bombyliflora and O. section Pseudophrys. O. speculum is pollinated by wasps of Scolioideae and Sphecoideae, which has led some authors (Kullenberg and Bergström, 1976) to link it to O. insectifera. Our results strongly reject this idea. Among the remaining taxa, Devillers and Devillers-Terschuren (1994) and Delforge (2006) classified O. bombyliflora and O. tenthredinifera in the ‘O. tenthredinifera group’ based on their broad and rounded sepals. Our work corroborates the close relationship between them, especially using the rpl16 intron (Fig. 2, weakly supported). In the Bayesian tree (combined data; Fig. 3), these two species are close to each other but successively sister to the same clade.

Our results highlight the plasticity of pollination in Ophrys. Up to four changes in pollinators can be inferred from our results for this taxon (wasp pollination in O. speculum and O. insectifera, self-pollination in O. apifera and beetle pollination in O. fusca; Pedersen and Faurholdt, 2007). Our results also suggest that these changes can occur rapidly (divergence of O. regis-ferdinandii, bee-pollinated, and O. speculum, wasp-pollinated, 1·1 Mya; 95 % CI: 2–0·4 Mya). According to other authors, these transformations are related to changes in the mixture of ‘pseudo-pheromonal’ compounds produced by the labellum of the flower (e.g. Schiestl and Cozzolino, 2008). Relationships between orchids and their pollinators in the framework of sexual deception continue to evolve rapidly, not just in Ophrys but in other genera, opening new possibilities for exploiting pollinator behaviour (e.g. Brodmann et al., 2009).

Both the rpl16 intron and combined trees (Figs 2 and 3) highlight a clear differentiation between the expanded Neotinea sensu Bateman et al. (1997) (including former Orchis tridentata, O. ustulata and O. lactea) and Orchis s.s. (including Aceras). Neotinea s.l. includes small plants with dense inflorescences of small flowers with three sepals and two petals being connivent forming a hood (mentum), much as in Anacamptis. The labellum is trilobed with the central lobe itself bilobed.

The combined tree (Fig. 3) shows two well-supported subclades in Orchis s.s. The first clade groups anthropomorphic species (O. italica, O. anthropophora, O. simia and O. militaris) and is consistent with the findings of Aceto et al. (1999) and Bateman et al. (2003). This group is characterized by the bract being much shorter than the ovary and the presence of an apiculum between the secondary lobes of the midlobe of the labellum. The second clade groups all Orchis spp. in which the perianth does not form a hood. Two subgroups within the second subclade can be distinguished: the O. mascula group (Bateman et al., 1997) comprising O. mascula and O. provincialis, two taxa with numerous morphological affinities (Soó, 1980); and the O. anatolica group (Bateman et al., 1997), comprising O. anatolica, O. brancifortii, O. quadripunctata and O. sitiaca.

The systematics and classification of Dactylorhiza, in which extensive polyploidization and reticulate evolution events are common, have been difficult and controversial (Heslop-Harrison, 1953; Devos et al., 2006a, b; Pillon et al., 2007). Neither morphological characters (Averyanov, 1990; Delforge, 2006) nor molecular analyses of ITS nrDNA (Bateman et al., 2003) clarified much due to the low levels of molecular and morphological divergence among the species. In our results, resolution within the clade is low, in both the rpl16 intron and combined trees. In the latter case, conflicting phylogenetic signals from different genomes might be preventing us from obtaining better posterior probabilties, but conversion of ITS nrDNA has been documented in nearly all cases to favour the copy type of the maternal parent (Pillon et al., 2007), so this would not cause incongruence with the data analysed in this study; the sole allotetraploid for which paternal conversion of nrITS was documented by Pillon et al. (2007), D. sphagnicola Höppner [=D. majalis subsp. sphagnicola (Höppner) H.A.Pedersen & Hedrén], was not included in our study. With the data collected in this study (maternally inherited for the plastid and mitochondrial loci or subject to concerted evolution/gene conversion, mostly to the maternal copy type, for nrITS), we did not expect to shed light on the problems caused by hybridization in Dactylorhiza.

The position that D. viridis (former C. viride) occupies in the combined tree (Fig. 3) highlights the importance of including it in Dactylorhiza. Differences in pollination and morphology, with molecular evidence based on nrDNA, have led some authors to keep D. viridis as an independent genus (Coeloglossum; e.g. Delforge, 2006; Devos et al., 2006a, b). However, molecular results based on nuclear and plastid data (this study; Bateman et al., 1997, 2003) and mitochondrial data (Inda et al., 2010) indicate that D. viridis has to be submerged to achieve monophyly for Dactylorhiza.

Gymnadenia and former Nigritella comprise a highly supported clade in both the plastid (Fig. 2) and combined (Fig. 3) analyses. These two taxa share several morphological traits that might support their merger, such as palmate-digitate tubers (also shared with Dactylorhiza and Platanthera), two lateral, lobe-like stigmas and two pollinia, each with a caudicle (Sundermann, 1980). Nevertheless, morphology is, in this case, rather ambiguous as other traits such as resupination (Aedo et al., 2005) support differentiation between these taxa. Further research is needed to clarify the relationships between these two orchid groups, but with just three accessions sampled we cannot comment further.

Pseudorchis occupies a different phylogenetic position in the rpl16 intron (Fig. 2) and combined trees (Fig. 3). In the former, it constitutes a weakly supported clade with Amitostigma as sister to Gymnadenia. In the latter analysis, it occupies a poorly supported position at the base of the Platanthera + Galearis + Neolindleya clade. Nevertheless, support increases when only the nuclear DNA is included (results not shown). Different authors have highlighted similar differences in their morphological interpretations of Pseudorchis. Thus, Summerhayes (1951) included Pseudorchis in Gymnadenia, whereas Luer (1975) included it in Platanthera. We suggest that Pseudorchis might have had an ancient hybrid origin between Gymnadenia and a member of the Platanthera + Galearis + Neolindleya clade. Divergence between Pseudorchis and the Platanthera + Galearis + Neolindleya clade took place ∼9·8 Mya (95 % CI: 6·3–13·4 Mya), during the late Miocene when connections between North America and Eurasia had already been broken. This fact, together with the present distribution of Galearis, Neolindleya and Platanthera, makes the last-named the most likely parent in such a hybridization (with Gymnadenia or Amitostigma as pollen donor). To confirm this hypothesis, analyses including low-copy nuclear markers should be conducted to clarify the position of Pseudorchis.

Platanthera is clearly monophyletic and sister to Galearis + Neolindleya, as indicated by Bateman et al. (2009). According to our results, diversification of Platanthera occurred first in North America (P. grandiflora; ∼7·7 Mya; 95 % CI: 4·8–11 Mya; P. hyperborea; ∼6·3 Mya; 95 % CI: 3·6–9·6 Mya), whereas diversification in Europe is much more recent. We agree with Bateman et al. (2009) that the divergence between P. bifolia and P. chlorantha is recent (∼1·3 Mya; 95 % CI: 0·5–2·5 Mya), despite the differentiation between them (Delforge, 2006) that has been attributed to a strong allometric component (Bateman et al., 2009).

Finally, regarding former Habenariinae (Habenaria and allies), we have only included a few taxa belonging to a widespread and diverse group of orchids. Nevertheless, a few conclusions can be drawn. First, Habenaria, which has not been studied in full since the work of Kränzlin (1897–1904), is grossly paraphyletic, with Gennaria, Herminium and Pecteilis embedded in it. Second, it is worth noting that some groupings in this clade are geographical (e.g. the Mediterranean/Macaronesian Habenaria tridactylites and Gennaria diphylla Parl. and the Asian Habenaria delavayi and Pecteilis sagarikii), suggesting repeated local diversification.

Pollination systems and evolution of Orchideae

Evolution of pollination systems in orchids and the influence of different pollination syndromes on diversification have received considerable attention in the last decade (e.g. Cozzolino and Widmer, 2005; Cozzolino and Scopece, 2008; Smithson, 2009). Some groups, such as Orchideae, that exhibit great diversity in pollination systems have also often been studied, and much information about these orchids is available (Smithson, 2009). We found evidence that food deception is ancestral in Orchideae, as highlighted by other authors (Cozzolino and Widmer, 2005; Smithson, 2009). In addition, we found that this trait continued to be prevalent throughout most of the tree, including several of the most species-rich clades, such as ‘core’ Orchidinae (excluding Habenariinae) and the Ophrys + Serapias + Himantoglossum + Anacamptis clade. According to our results, food rewards arose repeatedly. However, with the exception of former Habenariinae (mainly composed of the rewarding, lepidopteran-pollinated Habenaria and allies), these rewarding clades diverged more recently and are less diverse than the deceptive ones. According to Cozzolino and Widmer (2005), the higher outcrossing rates present in food-deceptive orchids may produce an increase of fitness despite the lower fruit-set. Moreover, pollination by deceit has been considered one of the keys to orchid species diversity (e.g. Cozzolino and Widmer, 2005; Peakall et al., 2010). As preferred pollinators, Hymenoptera are dominant throughout the tree, and we found significant evidence that pollination by bees and wasps is the ancestral state in both Orchidinae s.l. (including former Habenariinae) and Orchidinae s.s. Deviation from this pattern is common in the tree, with some clades shifting to other groups of insects (e.g. Lepidoptera in former Habenariinae and Platanthera + allies; Coleoptera, as reported in O. fusca and D. fuchsii; Gutowski, 1990) and others becoming generalists (e.g. D. viridis, H. monorchis). In most cases, this shift in pollinators has been accompanied by a change in pollination strategy from deception to food reward.

Hymenoptera show high behavioural diversity both among (e.g. social vs. solitary, hunters vs. pollen gatherers, parasitoids) and within (males vs. females) species, as well as complex chemical communication systems (e.g. Ayasse et al., 2001). The morphological and chemical plasticity present in orchid flowers, perhaps based on ‘deconstrained’ genetic regulation (Bateman and Rudall, 2006; Mondragón-Palomino and Theissen 2008, 2009), has allowed them to use the many opportunities opened up by diversity and complexity in Hymenoptera, triggering speciation in different orchid groups (Cozzolino and Scopece, 2008; Peakall et al., 2010; Griffiths et al., 2011). According to our dated tree (Fig. 4), most ‘core’ Orchidinae genera originated and diversified between the mid-Miocene and Pleistocene (∼14·8–1·0 Mya; 95 % CI: 19·0–0·4 Mya). Within this group, the Dactylorhiza clade diversified between ∼9·2 and 1·3 Mya (95 % CI: 12·8–1·0 Mya), whereas the mostly Mediterranean Orchis + Neotinea + Himantoglossum + Anacamptis + Serapias + Ophrys clade diversified between ∼14·8 and 0·4 Mya (95 % CI: 19·0–0·4 Mya). This diversification pattern parallels diversification of key groups of Hymenoptera in the Old World, particularly in the Mediterranean basin. Thus, bumble-bees (Hymenoptera, Apidae, Bombus), important in the pollination of Dactylorhiza, diversified in the Palaeartic mostly between ∼16 and 2 Mya (mid-Miocene to early Pleistocene; Hines, 2008). Moreover, some of the more relevant bee tribes for pollination of orchids in the Mediterranean (e.g. Eucerini and Meliponini in Apidae; Van der Cingel, 1995) widely diversified in the Miocene and Pleistocene (Danforth et al., 2006; Cardinal et al., 2010). Although a direct connection between diversification of orchids and their pollinators is difficult to make, and probably other events such as the end of the Messinian crisis (Krijgsman et al., 1999) and, especially, the onset of the Mediterranean climate between 16 and 2 Mya (Thompson, 2005; Salvo et al., 2010) had a large effect on orchid diversification, pollinator-driven speciation is a well-known phenomenon in Orchidaceae (e.g. Griffiths et al., 2011).

We found significant evidence that food deception is the ancestral trait in the mostly Mediterranean clade comprising Ophrys + allies (Ophrys + Serapias + Anacamptis + Himantoglossum). Moreover, we detected a change in labellum size for this clade (Table 3). Sexual deception and resting site mimicry (considered by some authors as another form of sexual deception; Vöth, 1980) originated almost exclusively in this clade of Orchidinae (Ophrys and Serapias). Schiestl and Cozzolino (2008) indicated that some taxa in this group, particularly Serapias, Ophrys and some Anacamptis, produce high levels of n-alkenes, considered an essential preadaptation for development of sexual deception. Schiestl and Cozzolino also indicated that, although abundant alkene production was clearly ancestral in Ophrys, lack of phylogenetic resolution prevented them from deciding if this trait was an evolutionary novelty or if it had been inherited from a common ancestor shared with Serapias. Although we have not included this trait in our analysis, our well-resolved phylogenetic tree indicates that abundant alkene production was probably ancestral in this clade, although this character has been lost on several occasions in the last 12 million years (in Himantoglossum and several Anacamptis spp.). It is worth noting that there is a clear relationship between alkene production and labellum size (Schiestl and Cozzolino, 2008), so the transition in labellum size we detected at this node might be also related to the onset of sexual deception. To clarify these questions completely, alkene production of more genera within the clade should be assessed. Also, pollination studies of Serapias are much needed.

Divergence in the Ophrys + Serapias + Anacamptis + Himantoglossum clade took place between the late Miocene and the Pleistocene (∼12·8–0·4 Mya; 95 % CI: 16·5–0·2 Mya). Anacamptis and Himantoglossum, pollinated mostly by food deception, were the first to diversify, whereas Ophrys and Serapias, with differentiated pollination systems (sexual deception and resting site mimicry, respectively), diverged early (∼12·8 and ∼10·5 Mya, respectively; 95 % CI: 16·5–9·1 and 15·5–8·4 Mya) but diversified only recently (Ophrys, ∼4·6–0·4 Mya; 95 % CI: 6·7–0·4 Mya; Serapias, ∼1·2–0·5 Mya; 95 % CI: 2–0·2 Mya). These divergence and diversification times coincided partially with two events that might have had a clear impact on diversification of Orchidinae: (1) the end of the Messinian crisis (∼5·3 Mya) and re-flooding of the Mediterranean basin (Krijgsman et al., 1999), which caused isolation of many populations in numerous groups of plants, triggering speciation; and (2) the orogenic events of the late Tertiary coincident with climate cooling, aridification and increased seasonality that gave rise to the Mediterranean climate 3–2 Mya (Thompson, 2005; Bruch et al., 2007; Salvo et al., 2010). According to some authors (e.g. Fiz et al., 2008), events of the late Tertiary (mid-Miocene and Pliocene) caused a loss of more specific-pollinating species, as a result of which some groups shifted towards generalist pollination (e.g. Geranium and Erodium, Fiz et al., 2008). Apparently, taxa of Orchidinae in the Mediterranean, especially in the Ophrys + allies clade, did not respond to this selective pressure in the same way. During the late Tertiary and the early Pleistocene orchids continued their rapid diversification and exploited more unusual aspects of the behaviour of their pollinators.

APPENDIX

Plant materials used in this study. Taxon, voucher information (Kew) and GenBank accession numbers (1st, rpl16; 2nd, cox1) are indicated for each sample.

Tribe Orchideae, Subtribu Orchidinae, Amitostigma Schltr: A. gracile (Blume) Schltr., Lou sn (7917), EU176087, EF143193; Anacamptis Richard: A. boryi (Rchb. Fil.) Bateman, Pridgeon & Chase, Bateman 29, EU176019, EF143108; A. champagneuxii (Barneoud) Bateman, Pridgeon & Chase, Chase O-706, EU176017, EF143106; A. collina (Banks & Solander ex Russel) Bateman, Pridgeon & Chase, Bateman 10, EU176024, EF143113; A. fragrans (Pollini) R.M.Bateman, Bateman (22846), EU176021, EF143110; A. laxiflora (Lam.) Bateman, Pridgeon & Chase, Chase O-909, EU176025, EF143114; A. laxiflora, Chase O-1141, EU176026, EF143115; A. morio (L.) Bateman, Pridgeon & Chase, Chase O-712, EU176018, EF143107; A. palustris (Jacquin) Bateman, Pridgeon & Chase, Hedrén (5557), EU176027, EF143116; A. papilonacea (L.) Bateman, Pridgeon & Chase, Chase O-935, EU176023, EF143112; A. pyramidalis (L.) Richard, Chase O-563, EU176020, EF143109; A. sancta (L) Bateman, Pridgeon & Chase, Chase O-912, EU176022, EF143111; Chamorchis Richard: Ch. alpina (L.) Richard, Chase 5558, EU176086, EF143191; Cynorkis Thouars: C. sp., Bateman (22842), EU176090, EF143197; Dactylorhiza Necker ex Nevski: D. alpestris (Pugsley) Aver., Bateman 48 (O-963), JQ581531, –; D. aristata (Fisch. Ex Lindl.) Soó, Dick (RMB 366) (15160), DQ022904, EF143165; D. cordigera (Fries) Soó, Dick (RMB 108) (15156), JQ581532, –; D. incarnata subsp. cruenta (O.F. Müll.) P.D.Sell, Bateman 115 (5549), EU176077, –; D. elata (Poir.) Soó, Chase O-718, EU176072, EF143168; D. maculata (L.) Soó, DNA-bank 6506, DQ022901, –; D. foliosa (Solander ex Lowe) Soó, Chase O-537, DQ022898, –; D. foliosa (Solander ex Lowe) Soó, Bateman 608, EF176075, EF143177; D. fuchsii (Druce) Soó, Bateman 139 (15166), DQ022896, EF143172; D. fuchsii, Hedrén 97037 (3971), DQ022897, –; D. fuchsii, (S. Gotland) (O-1375), DQ022895, –; D.iberica (M-Bieb. ex Wildenow) Soó, Chase O-960, DQ022910, EF143176; D. incarnata subsp. incarnata, (N. Gotland) (O-1378), DQ022917, EF143179; D. incarnata subsp. pulchella (Druce) Soó, Bateman & Rudall (17049), EU176076, EF143163; D. incarnata subsp. pulchella (Druce) Soó, Bateman 56 (988), DQ022915, –; D. maculata (L.) Soó, Bateman sn (14065), DQ022900, EF143173; D. maculata, Hedrén 97214 (3995), EU176074, EF141775; D. maculata (L.) Soó, Pinto sn (11795), DQ022899, EF141774; D. praetermissa (Druce) Soó, Chase O-1124, EU176073, EF143169; D. incarnata subsp. ochroleuca (Wüstnei ex Boll.) P.F.Hunt & Summerh., Clarke (RMB462) (15172), DQ022916, EF143181; D. romana (Sebastiani) Soó, Rossi sn (O-760), DQ022905, EF143164; D. romana, Bateman/ Rudall (RMB 522) (15177), DQ022906, –; D. saccifera (Brongniart) Soó, Manuel (RMB 74) (15986), DQ022903, EF143171; D. sambucina (L.) Soó, Lowe (RMB 315) (15168), DQ022908, EF143167; D. sambucina, No voucher (14610), DQ022909, -; D. sambucina, (S. Gotland) (1374), DQ022907, –; D. viridis (L.) Bateman, Pridgeon & Chase, Bateman 66 (15982), DQ022912, EF143164; D. viridis, Hartwell sn (O-579), DQ022911, –; D. viridis (L.), Lou & Lou 667 (13079), DQ022913, –; D. viridis, Hedrén 98033 (15979), EU176070, –; D. viridis, Dick (RMB 71) (15985), EU176071, –; Galearis Raf.: G. cyclochila (Franch. & Sav.) Soó, Inouye (O-441), EU176080, EF143185; G. spectabilis (L.) Raf.ex Ruth, Davis (O-727), EU176079, EF143184; Gennaria Parlatore: G. diphylla (Link) Parl., Chase O-886, EU176101, EF143206; Gymnadenia Brown: G. conopsea (L.) Brown, Chase O-574, DQ022919, –; G. conopsea, Fay 573A (14752), DQ022920, EF143159; G. borealis (Druce) R.M. Bateman, Pridgeon & MW Chase, Bateman 64 (O-991), DQ022921, EF143160; G. nigra (L.) Reichenbach, Hedrén 97322 (5561), DQ022924, EF143162; G. nigra (L.) Reichenbach, Bateman 67 (1122), EU176069, –; Habenaria Willdenow: H. arenaria Lindley, Chase O-1135, EU176095, EF143201, H. delavayi Finet, Lou 56 (8055), EU176097, EF143203; H. procera Lindle,yO-594, EU176100, –; H. socotrana Balfour, Bateman (22843), EU176096, EF143202; H. tibetica Schltr, Bateman (22844), EU176094, EF143200; H. tridactylites Lindley, Bateman (22841), EU176098, EF143204; H. viridiflora Span., Chase 15935 (15935), EU176099, EF143205; Herminium L.: H. monorchis (L.) Brown, No voucher (O-1381), EU176091, EF143198; H. monorchis (L.) Brown, Bateman (22845),EU176092, –; Himantoglossum Koch: H. adriaticum Baumann, Rossi sn (O-757), EU176052, EF143140; H. comperianum (Steven) Delforge, Bateman (22833), EU176049, EF143137; H. caprinum (M.Bied.) Sprengel, Bateman 27 (O-943), EU176053, EF143141; H. hircinum (L.) Sprengel, Chase O-898, EU176051, EF143139; H. robertianum (Loiseleusr) Delforge, Chase O-550, EU176050, EF143138; Holothrix Linldley: H. scopularia Rchb., Kurzweil 1821 (O-675), EU176106, EF143212; Neotinea Reichenb.: N. lactea (Poiret) Bateman, Pridgeon & Chase, Bateman 21 (O-944), EU176057, EF143146; N. maculata (Desf.) Stearn, Chase O-548, EU176054, EF143143; N. tridentata (Scopoli) Bateman, Pridgeon & Chase, Chase O-914, EU176055, EF143144; N. ustulata (L.) Bateman, Pridgeon & Chase, Rossi sn (O-755), EU176056, EF143145; Neolindleya Kraenzl.: N. camtschatica (Cham.) Nevski., Bateman (22835), EU176078, EF143183; Ophrys L.: O. aesculapii Renz, Chase O-901, EU176046, EF143135; O. apifera Huds., Chase O-536, EU176042, EF143131; O. araneola Rchb., Chase O-701, EU176045, EF143134, O. bombyliflora Link, Bateman 22 (16017), EU176039, EF143128; O. cretica (Vierh.) Nelson, Chase O-706, EU176047, EF143126; O. fusca Link, Chase O-711, EU176037, EF143123; O. insectifera L., Chase 19422, EU176034, EF143127; O. iricolor Desfontaines, Chase O-903, EU176038, EF143125; O. lutea Cav., Chase O-904, EU176036, EF143130; O. regis-ferdinandii (Renz) Buttler, Chase O-905, EU176041, EF143132; O. heldreichii Schltr, Bateman 13 (O-948), EU176043, EF143129; O. speculum Link, Chase O-902, EU176040, EF143136; O. sphegodes Mill., Chase O-896, EU176048, EF143123; O. spruneri Nyman, Bateman (22847), EU176044, EF143133; O. tenthredinifera Willd., Chase O-906, EU176035, EF143124; Orchis L.: O. anatolica Boiss, Chase O-907, EU176065, EF143154; O. anthropophora (L) Allioni, Chase O-549, EU176058, EF143147; O. brancifortii Bivona-Bernardi, Bateman (22835), EU176066, EF143155; O. italica Poir., Chase O-908, EU176059, EF143148; O. mascula L., Chase O-1138, EU176063, EF143152; O. militaris L., Chase O-939, EU176061, EF143150; O. pauciflora Tenore, Chase O-710, EU176062, EF143151; O. purpurea Huds., Rossi (O-759), EU176068, EF143157; O. quadripunctata Cirillo ex Ten., Chase O-545, EU176067, EF143156; O. simia Lam., Chase O-705, EU176060, EF143149; O. sitiaca (Renz) Delforge, Bateman 28 (O-1139), EU176064, EF143153; Pecteilis Raf. : P. sagarkii Seidenf., 21779, EU176093, EF143199; Platanthera Richard: P. bifolia (L.) Richard, Bateman 62 (O-993), EU176083, EF143188; P. chlorantha (Custer) Reichenbach, Rossi sn (O-758), EU176084, EF143189; P. grandiflora (Bigelow) Lindl., VAA (O-351), EU176082, EF143187; P. hyperborea (L.) Reichenbach, Lights (O-407), EU176081, EF143186; Pseudorchis Seguier: P. straminea (Fernald) Soó, Liden sn (5556), DQ022926, EF143182; Satyrium Sw.: S. humile Lindl., Chase 5959, EU176103, EF143208; S. nepalense Don, Chase O-539, EU176102, EF143207; Serapias L.: S. cordigera L., Chase O-899, EU176029, EF143118; S. lingua L., Chase O-557, EU176028, EF143117; S. neglecta de Notaris, Chase O-961, EU176030, EF143119; S. orientalis (Greutier) Baumann & Kunkele, Bateman 23 (O-1143), EU176032, EF143121; S. parviflora Parlatore, Bateman 7 (O-945), EU176033, EF143122; S. vomeracea (Burman) Briquet, Bateman 40 O-942), EU176031, EF143120. Stenoglottis Lindl.: S. longifolia Hook., Bateman (22839), EU176088, EF143195; S. woodii Schltr, Bateman (22840), EU176089, EF143196; Traunsteinera Reichenbach: T. globosa (L.) Reichenbach, Bateman 68 (16026), EU176085, EF143190; Subtribe Disinae Benth, Disa Bergius: D. tripetaloides N.E.Br., Chase O-492, EU176104, EF143210; Huttonaea Harv.: H. grandiflora (Schltr) Rolfe, Goldblatt & Manning 11047 (8990), EU176107, EF143213; Subtribe Brownleeinae Linder & Kurzweil, Disperis Sw.: D. lindleyana Rchb., Chase 5925, EU176105, EF143211;