|Home | About | Journals | Submit | Contact Us | Français|
A recent phylogenetic study based on multiple datasets is used as the framework for a more detailed examination of one of the ten molecularly circumscribed groups identified, the Ophrys fuciflora aggregate. The group is highly morphologically variable, prone to phenotypic convergence, shows low levels of sequence divergence and contains an unusually large proportion of threatened taxa, including the rarest Ophrys species in the UK. The aims of this study were to (a) circumscribe minimum resolvable genetically distinct entities within the O. fuciflora aggregate, and (b) assess the likelihood of gene flow between genetically and geographically distinct entities at the species and population levels.
Fifty-five accessions sampled in Europe and Asia Minor from the O. fuciflora aggregate were studied using the AFLP genetic fingerprinting technique to evaluate levels of infraspecific and interspecific genetic variation and to assess genetic relationships between UK populations of O. fuciflora s.s. in Kent and in their continental European and Mediterranean counterparts.
The two genetically and geographically distinct groups recovered, one located in England and central Europe and one in south-eastern Europe, are incongruent with current species delimitation within the aggregate as a whole and also within O. fuciflora s.s. Genetic diversity is higher in Kent than in the rest of western and central Europe.
Gene flow is more likely to occur between populations in closer geographical proximity than those that are morphologically more similar. Little if any gene flow occurs between populations located in the south-eastern Mediterranean and those dispersed throughout the remainder of the distribution, revealing a genetic discontinuity that runs north–south through the Adriatic. This discontinuity is also evident in other clades of Ophrys and is tentatively attributed to the long-term influence of prevailing winds on the long-distance distribution of pollinia and especially seeds. A cline of gene flow connects populations from Kent and central and southern Europe; these individuals should therefore be considered part of an extensive meta-population. Gene flow is also evident among populations from Kent, which appear to constitute a single metapopulation. They show some evidence of hybridization, and possibly also introgression, with O. apifera.
The orchid genus Ophrys is famous for its distinctive floral morphology and remarkable pseudocopulatory reproductive strategy (Schiestl et al., 1999; Schiestl and Ayasse, 2002; Cozzolino and Scopece, 2008; Devey et al., 2008; Schiestl and Cozzolino, 2008; Cortis et al., 2009). It is also notorious for its controversial taxonomy and problematic species delimitation (cf. Delforge, 2005; Pedersen and Faurholdt, 2007; Devey et al., 2008; Schlüter et al., 2009; R. M. Bateman and D. S. Devey, unpubl. res.). This paper focuses on one subset of taxa within the genus, termed the Ophrys fuciflora aggregate (Fig. 1).
The O. fuciflora aggregate as a whole mirrors O. fuciflora sensu stricto (s.s.) in possessing a wide distribution, stretching from Syria in the south-east to south-east England in the north-west, and occurring on many Mediterranean islands. Populations most commonly occur close to sea level, though exceptionally they reach 1500 m a.s.l.
Ophrys fuciflora (s.s.) is one of four Ophrys species that occur in the UK, with O. sphegodes, O. apifera and O. insectifera (e.g. Harrap and Harrap, 2005). Together with O. sphegodes, O. fuciflora is listed as Vulnerable in the Vascular Plant Red Data List for Great Britain. Ophrys fuciflora is placed in category D1, indicating that there are currently fewer than 1000 mature individuals present in the UK (Cheffings and Farrell, 2005), where they are confined to Kent (Preston et al., 2002). Although scarce in Britain (there are only six clearly distinct populations) and elsewhere along the periphery of its distribution, O. fuciflora s.s. is not currently threatened in continental Europe. The species is widespread but local throughout the temperate and Mediterranean climate zones in calcareous habitats, particularly CG3 Bromus erectus grasslands and CG4 Brachypodium pinnatum calcareous grasslands, as designated under the UK National Vegetation Classification (Rodwell, 1993).
Members of the aggregate most frequently flower between March and June, peaking between mid-April and early May (early June in Kent); they are reputed to be largely bee pollinated. Average population size is small compared with many Ophrys species; most sites contain 10–100 individuals, though some populations reach 500 plants.
As described by Devillers and Devillers-Terschuren (1994) and illustrated in Fig. 1, members of the Ophrys fuciflora aggregate are characterized by elongated sepals and short, triangular lateral petals that are ciliate or abruptly auriculate. The labellum is large, with dense trichomes that form either a complete-marginal or sub-marginal wreath with a discrete tuft directly above the well-developed, often three-pronged appendix. The speculum varies from a simple collar above a single partial loop (together forming a distorted H) to a complex of three complete rings and three partial rings; it is always bounded by a relatively broad, pale margin. The connective has a pointed apex and the stigmatic cavity is bounded by incomplete labia. The ‘pseudo-eyes’ are often elongated transversally and positioned at the extremities of a well-marked angular crest, to which they are connected by short pedicels. Vestiges of the basal callosities are slightly raised from the sides of the basal field, and staminodial points are present in most species. The basal field is relatively large, and frequently differs in colour from that of the stigmatic cavity floor and the labellum.
Lindley (1840) first divided the genus Ophrys into two sections, Cornutae and Muticae, and Schlechter (1926) recommended further divisions into four sections. A more thorough evaluation by Godfery (1928), building on the work of Soó (1927), recircumscribed the genus to recognize sections Musciferae, Fuciflorae, Araneiferae and Apiferae. Devillers and Devillers-Terschuren (1994) partitioned ‘the O. fuciflora complex’ (more narrowly delimited than Fuciflorae) into three groups, based primarily on the density and location of trichomes on the labellum: the O. episcopalis group, the O. fuciflora group and the O. tetraloniae group. Faurholdt and Pedersen (2007) equated the O. fuciflora aggregate described here with O. fuciflora s.s., recognizing within this broadly circumscribed species ten subspecies that can be differentiated using leaf, sepal, petal and speculum characters. A list of all Ophrys taxa, as defined using the highly split taxonomy of Delforge (2005), is given in the Appendix for comparative purposes.
Building on previous work by Bateman et al. (2003), Devey et al. (2008) surveyed the entire genus Ophrys using internal transcribed spacer (ITS) sequencing, plastid sequencing and amplified fragment length polymorphism (AFLP) (Vos et al., 1995) analyses in an attempt to better resolve species-level relationships. Having obtained congruent results using all three methods, this more taxonomically focused study was approached with greater confidence, electing to employ the AFLP technique as it offered greater potential for elucidating infraspecific as well as interspecific patterns.
The AFLP technique was selected for genetic analysis over DNA sequencing or microsatellite variability analysis due to the suspected recent origin of the group (Devey et al., 2008) and the consequent close genetic similarities among the resulting taxa. AFLP markers sample restriction endonuclease sites widely across the nuclear genome by selective amplification (Remington et al., 1999; Arcade et al., 2000). Polymorphisms are evidenced as presence or absence of peaks when fluorescently labelled fragments are separated by capillary electrophoresis. AFLP markers were initially developed for population and species delimitation studies (Mueller and Wolfenbarger, 1999), though they have also been increasingly used to resolve species-level relationships within groups showing low genetic divergence (e.g. Richardson et al., 2003; Devey et al., 2008), in which AFLP results repeatedly proved acceptably congruent with the products of both plastid and nuclear DNA sequencing.
The primary objectives of the study were (a) to circumscribe minimum resolvable genetically distinct entities within the O. fuciflora aggregate, and (b) to assess the likelihood of gene flow between genetically and geographically distinct entities at species, subspecies and population levels.
Samples of O. fuciflora s.s. analysed for AFLP were collected from five sites in Kent, south-east England: four occurred along a 5-km-long strip of downland immediately west of Folkestone, whereas the fifth was located 11 km to the north-west at Bullstown, near Wye. Together, these sites spanned the full geographic distribution of O. fuciflora in the UK. A further 12 localities were sampled in continental Europe and the Mediterranean islands. Additional populations of other named taxa in the O. fuciflora aggregate were sampled in France, mainland Italy, Sardinia, Sicily, Austria, mainland Greece (Peloponnese) and Chios, a Greek island immediately west of Turkey. All accessions and vouchers used in this study are listed in Table 1, and their geographical locations are shown in Fig. 2.
Most field-collected specimens were dried in silica gel prior to DNA extraction, following the recommendations of Chase and Hills (1991), though some accessions from UK populations were extracted from fresh material. Extraction of genomic DNA followed the 2× CTAB protocol (Doyle and Doyle, 1987), but used a CsCl2/ethidium bromide density gradient (1·55 g mL−1) for purification of samples to be preserved for long-term storage. Extraction of samples not intended for long-term storage followed the 2× CTAB protocol with the following modifications: after precipitation with isopropanol and subsequent centrifugation, the DNA pellet was washed with 70 % ethanol, dried at 37 °C, then resuspended in TE buffer (20 mm Tris–HCl, 0·1 mm EDTA). No experimental differences were detected between DNA extracted using the two protocols.
A primer trial was conducted using 16 primer combinations to identify pairs of selective primers that would be most appropriate for this study. Primer combinations Mse1–AGG + EcoRI–CTAC and Mse1–ACA + EcoRI–CTAT (both 5 mm) were used following the manufacturers' instructions to produce AFLP profiles across all analysed accessions, since these combinations yielded suitable numbers of bands and levels of variation among loci. The addition of an extra base onto the EcoR1 primer to reduce the number of peaks produced was described by Vos et al. (1995), Fay and Krauss (2003) and Fay et al. (2005) for use in cases in which the genome of the plant in question is substantially larger than those of plants for which the AFLP kits are optimized.
The AFLP reaction products were visualized on an ABI 3700 Genetic Analyser, following the manufacturer's protocols. Fragment data generated were analysed using Genescan (version 2·02) and Genotyper (version 1·1) analysis software (Applied Biosystems). The AFLP traces were carefully compared by eye to ensure homology of peaks. Markers with evidence of ‘false-negative’ peaks (small, unscorable peaks in a size range where other samples showed larger, scorable peaks) were discarded from all samples. This screening strategy prevented the potential introduction of artefacts into the data due to uneven amplification among samples. Peaks ranging in size from 50 bp to 500 bp were scored as present or absent and entered into a binary matrix. Genetic relationships were analysed and ordinations produced in the statistics programs Le Progiciel R (Casgrain and Legendre, 2001) and Genstat (2007) by principal co-ordinates analysis (PCoA) (Gower, 1966), using the Jaccard similarity coefficient (Jaccard, 1908) to avoid grouping terminals on the basis of shared zeros.
The primer combinations generated 159 AFLP markers, each accession yielding between 54 and 73 peaks. The resulting binary matrix for all areas was first analysed in its entirety, before being narrowed by removal of the eastern Mediterranean populations. Both of the resulting PCoA were dominated by the first co-ordinate but also showed low levels of character correlation (i.e. they had high dimensionality), the first three axes accounting for only approx. 32 % of the total variance. As expected, the more geographically focused analyses of (a) UK populations only and (b) central and southern European (CSE) populations only showed lower dimensionality, the first co-ordinate being less dominant and the first three axes accounting for a greater proportion of the total variance (35 % and 45 %, respectively).
The PCoA analysis derived from AFLP data including all samples analysed (Fig. 3) shows partitioning of Ophrys species into two discrete clusters. One cluster contains individuals exclusively from the south-eastern limits of the geographical range of the genus (hereafter labelled group SE). Within the remaining PCoA cluster, which contains the accessions collected from the UK (group UK) plus the central European and southern European areas (group CSE), the spatial distribution of the O. fuciflora aggregate mirrors a geographical cline running from north-west to south-east Europe. Within this group, accessions that occurred in closest geographical proximity were genetically most similar. Although largely separated from the rest of this cluster, the accessions from the UK do not form a wholly discrete entity. Some members are genetically more similar to French individuals, in particular to the accession from Dijon (the continental sample site closest geographically to the UK), than to other UK accessions (individual 31, Table 1). The genetic variability collectively exhibited by the UK populations of O. fuciflora s.s. exceeded that evident within the remainder of PCoA clusters CSE or SE, despite the fact that both of the non-UK clusters contain several putative species (Fig. 1) and each represents a far greater geographical area. Individuals described as O. fuciflora s.s. are also present in both of these clusters, interspersed with other members of the O. fuciflora aggregate.
A reanalysis of the data omitting south-east European accessions (i.e. containing only groups UK and CSE) is shown in Fig. 4. Although no discrete clusters exist among accessions, a geographical cline is again evident. Samples from France and Sardinia are genetically more similar to those from the UK (to which they are closer in geographical proximity) than to those from collection sites in mainland Italy and Sicily. The genetic variation evident among samples from France and Sardinia does not overlap that among samples from Italy and Sicily. The single sample from Austria is genetically more similar to those from Italy than to any other populations.
One individual from the Shuttlesfield Farm population (UK) proved to be the most genetically distinct accession (individual 26, Fig. 4), forming an outlier on the PCoA plot. Two primary hybrids with O. apifera were recorded at this site (Fig. 1), and it is suspected that, despite the lack of morphological evidence, this aberrant individual reflects introgression with O. apifera. As in the broader analysis presented in Fig. 3, individuals assigned to O. fuciflora s.s. (Table 1) are distributed widely across the PCoA plot, rather than forming a discrete cluster.
Figure 5 shows the PCoA plot of AFLP results for UK populations of O. fuciflora s.s. only. There is no clear partitioning of these populations, though no overlap is evident between individuals collected from the Arpinge populations and those from the Bulltown and Channel Tunnel populations.
The corresponding PCoA plot of AFLP data for southern and central European O. fuciflora s.l. (group CSE) only is shown in Fig. 6. As with the other ordinations described here, the underlying pattern primarily reflects geography rather than taxonomic assignment. The plot shows greater discrimination than Fig. 3 in that Sardinian individuals are separated from French individuals, rather than distributed among them. There is some overlap, and therefore presumed gene flow, between French and southern Italian individuals. Sicilian accessions show some overlap with mainland Italian material, though they are wholly separated from Sardinian, French and Austrian individuals.
The most important outcome of this study is the partitioning of individuals along geographic lines, rather than those suggested by prior morphological delimitations of species and subspecies according to classical morphological taxonomy (Bateman and Devey, 2009). Within the O. fuciflora aggregate, accessions that were closer geographically were generally genetically more similar, indicating that when gene flow takes place, it typically occurs between proximal individuals of diverse morphologies rather than between individuals that are morphologically more similar yet located further apart. This finding challenges the highly specific ‘one orchid morphotype equates with one pollinator species’ theory favoured by many recent authors (e.g. Paulus and Gack, 1990; Delforge, 2005, Schlüter et al., 2009); it instead suggests that the pollination model is at best ‘leaky’ and at worst considerably more generalized than is frequently postulated. If pollination were indeed highly specific and if long-distance seed dispersal were the mechanism by which population gene flow was facilitated, this observation would be difficult to explain as each capsule can contain up to 10 000 dust-like seeds capable of travelling many hundreds of kilometres in air currents (new data and references in Fay et al., 2009). It is therefore possible to infer that it is pollen dispersal, more specifically pollinator error, that is primarily responsible for the extensive gene flow observed here (cf. Bateman et al., 2003; Soliva and Widmer, 2003; Bateman and Devey, 2009).
Figure 3 implies that the two distinct genetic clusters represent reproductively isolated lineages from different geographic areas, at least within the limits of the sampled material. The geographical distribution of the larger cluster (UK plus CSE) stretches from the UK in the north-west to Austria in the east and Sicily in the south. The contrasting group (SE) is localized in the south-eastern areas of the distribution, consisting of accessions collected from southern Greece and Chios, a Greek island located near the west coast of Turkey. This is a smaller geographical area than that covered by cluster UK and CSE, and so predictably contains a smaller proportion of the overall genetic variability. The four putative species investigated in group SE [O. candica and O. lacaena (including morphs resembling O. episcopalis, O. candica and O. heldreichii) from the Peloponnese and O. homeri and supposed O. fuciflora s.s. from Chios] represent only a single cohesive genetic species.
Metapopulations are assemblages of interconnected populations existing in a balance between extirpation and colonization (Levins, 1969); metapopulation models depict species occupying an array of habitat patches linked by migration (Barrett and Pannell, 1999). There is some evidence of restricted gene flow between the Kentish populations, notably the absence of overlap between individuals collected from the Arpinge versus the Bulltown and Channel Tunnel populations (Fig. 5). However, Bulltown and Channel Tunnel are two of the most geographically separated (by 16 km) Kentish populations, yet they generate convex hulls that generally overlap on Fig. 5. This lack of clear partitioning among the UK populations indicates continuing gene flow and suggests that this modest number of local interbreeding populations together constitute a metapopulation. Admittedly, there exists an alternative, technically based explanation for this lack of differentiation within UK accessions of O. fuciflora s.s. that reflects the rationale behind AFLP primer selection. Primers were optimized for use at the hierarchical level of the O. fuciflora aggregate (i.e. the initial focus of the investigation); consequently, primers that could potentially have differentiated more effectively between the more closely related members of O. fuciflora s.s. populations in the UK were not selected. Nonetheless, we believe that the PCoA plots accurately reflect relative levels of genetic differentiation in the major geographically circumscribed groups.
The distribution of the south-eastern European O. fuciflora aggregate mirrors that of the O. umbilicata s.l. group first described by Devey et al. (2008), in which there is clear geographical and genetic separation between accessions from this area and those from further west and north-west. Within the genus, there are no prezygotic barriers to reproduction, including morphologically imposed isolation through pollinator behaviour (Scopece et al., 2007; Cozzolino and Scopece, 2008; Schiestl and Cozzolino, 2008), and karyological diversity is relatively low (D'Emerico et al., 2005). There must therefore be physical and/or ecological barriers that isolate this group from the remainder of the O. fuciflora aggregate. This topic will be explored in a future paper.
A potential and novel explanation for the relative isolation of the SE European cluster could be a result of the prevailing wind conditions in the boundary zone of the atmosphere (Fig. 7). Three dominant ‘wind patterns’, the Etesian, the Adriatic Bora and the Föhn, occur in this zone during the flowering and dehiscence period of the O. fuciflora s.l. aggregate.
The Etesian is a prevailing northerly monsoonal wind that occurs from early summer until early autumn, flowing across the Aegean Sea and into the eastern Mediterranean. The north-westerly flow is confined to the Aegean Sea by the Rhodope mountain range to the north (labelled A in Fig. 7), by the Boz Daglar Mountains to the east (labelled B in Fig. 7) and by the Pindus mountain range to the west (labelled C in Fig. 7).
The Adriatic Bora (named after Boreas, the Greek God of the North Winds) is a cold, north to north-easterly airflow. It originates when a layer of cold air forms above sloping ground; air close to the ground is colder than air occurring at the same altitude but further downslope, causing gravitational flow of the colder, denser air beneath the warmer, lighter air. The Adriatic Bora flows from the Balkan Peninsula and Hungary, through valleys or the mountain passages of the Dinaric Alps (known as ‘bora corridors’) and into the Adriatic Sea east of Italy (labelled D in Fig. 7). The Adriatic Bora is generally confined to the Adriatic Sea area, as the Alps to the north (labelled E in Fig. 7) channel the flow south-westward.
The Föhn is a prevailing wind blowing north from the Italian mainland over the Alps and into central Continental Europe. When air rises over the Alps, it expands due to the lower pressure. This expansion is accompanied by a reduction of temperature, termed adiabatic cooling, whereas air sinking on the leeward side undergoes adiabatic warming. Wind currents can be shaped as they are forced to rise, or are funnelled through valleys, greatly increasing their strength. This effect, which begins in spring and lasts throughout summer, is one of the main causes of alpine snow melt (Naval Research Laboratory, 2007).
We hypothesize that this combination of relatively strong prevailing winds blowing in contrasting directions separates both airborne seeds and pollinators, largely isolating the two geographical areas circumscribed by the present data and by other genetic data for Ophrys (Bateman et al., 2003; Devey, 2007; Devey et al., 2008). This barrier would hinder the movement of either pollinia-bearing insects or especially seeds, migration patterns of which dominate gene flow in orchids (Squirrell et al., 2001), thereby greatly reducing the probability of introgression between populations in the western and eastern provinces. Due to the variation and evidence of gene flow between populations evident throughout the remainder of the geographical distribution, this hypothesis suggests that the Alps and the Apennines provide less effective barriers to migration than do the prevailing winds, and raises the question of how long these prevailing winds are likely to have existed.
In light of the data described here, some taxonomic revisions are desirable within the O. fuciflora aggregate if the resulting groups are to possess comparable and appropriate levels of genetic cohesion. Adoption of a genetic species concept (as recommended by Devey et al., 2008) requires that all individuals described as O. fuciflora s.s. should show greater overall genetic similarity to each other than to members of any other species or subspecies. Although the phenetic analysis described by a PCoA plot may not be a perfect proxy for closeness of relationship, the congruence of DNA-based datasets and biogeographical conclusions justifies its use in this case. These taxa are present across both PCoA clusters, and interspersed with other members of the O. fuciflora aggregate that were afforded specific rank by Devillers and Devillers-Terschuren (1994) and Delforge (2005) (i.e. O. apulica, O. calliantha, O. candica, O. celiensis, O. dodekanensis, O. episcopalis, O. homeri, O. oxyrrhynchos; Fig. 1). Thus, O. fuciflora s.s., as traditionally morphologically delimited, is polyphyletic.
Two taxonomic solutions are possible. The first treats O. fuciflora s.l. as a single species containing two subspecies: O. fuciflora subsp. fuciflora (UK and central Europe) and O. fuciflora subsp. candica (the earliest subspecies epithet available for the SE Europe cluster). Any variation within these clusters would be attributed to local variation. The alternative, and in our opinion preferable, solution would accept two species, O. fuciflora and O. candica (the earliest epithet for the SE Europe cluster). Again, further variants within these species would then be attributable to either local variation or aberrant pollination events causing hybridization. Should further division be considered necessary, the segregates could be assigned subspecies rank; for example, the Sicilian O. oxyrrhynchos would become O. fuciflora subsp. oxyrrhynchos (Fig. 1).
Although, for both systematic and conservation purposes, it would be remiss to ignore a biologically significant species, it is a severe hindrance to conservation bodies to have to factor into their plans poorly substantiated taxa. In some respects this situation mirrors that of Epipactis (Orchidaceae), as reviewed by Squirrell et al. (2002), in which the authors described how the choice of taxonomic treatment can profoundly affect the number of rare and endemic species perceived to occur within a taxonomically problematic group.
In order to best conserve the phylogenetic diversity already present within the O. fuciflora aggregate, as well as the potentially ongoing processes of speciation (Ennos et al., 2005), the focus of conservation needs to shift away from supposed endemics of geographical or ecological islands, as the delimitation of such taxa demonstrably has little basis in genetic structure or reproductive isolation. Instead, disparate populations that contain a variety of morphotypes, yet also span the geographic distribution of the aggregate, need to be protected. It is only by ensuring that sufficient populations of plants (and pollinators) are maintained throughout the geographic continuum of the aggregate, and in sufficiently close proximity for short-range gene flow to continue, that the pool of genetic diversity (from which all the morphotypes, local variants and often transient ‘prospecies’ presumably arise) can be maintained.
The genetic variability exhibited by the UK populations of O. fuciflora s.s. (group UK) unexpectedly exceeds that exhibited by the remainder of the central European cluster (Fig. 5). If the postglacial recolonization route had been north-west across Europe, as suggested by various studies of other groups of Mediterranean species, genetic variability would be expected to decrease away from the presumed origin(s) of the expansion (e.g. Hewitt, 1999; Pillon et al., 2006). The UK populations currently contain a surprisingly large proportion of the total genetic diversity present within the O. fuciflora aggregate, despite being the northern-most outlier of the distribution of not only the species but also the species group. These populations would have occupied among the last areas of the current distribution to become available for recolonization following the retreat of the ice sheets at the end of the last glacial period. They are therefore highly unlikely to constitute the original centre of genetic diversity; this was more likely located in one or more glacial refugia in or even beyond the southern areas of the present distribution of the species. However, the high levels of genetic variability exhibited by UK populations may be the result of more than one recolonization event by members of different genetic lineages.
Despite ongoing uncertainty about the underlying cause(s) of the larger than expected variability, the UK populations should receive higher conservation priority than other areas if resources are limited. The much greater taxonomic diversity recognized in southern and eastern Europe (e.g. Delforge, 2005; Pedersen and Faurholdt, 2007) is not reflected in greater genetic diversity. This outcome contrasts strongly with the conclusions reached by Pillon et al. (2006) and Pillon and Chase (2007) for another problematic genus of Eurasian Orchidinae, Dactylorhiza, in which greater taxonomic diversity recognized in north-west Europe conflicted with greater genetic diversity documented towards the south and east.
We thank Mark Chase, Paula Rudall, Salvatore Cozzolino and Laure Civeyrel for insights and helpful comments, and Paula Rudall for comments on the manuscript. Barry Tattersall, Richard Manuel and other members of the UK Hardy Orchid Society, and Giuseppe Tosi in Italy, provided invaluable plant samples and local knowledge. The Royal Botanic Gardens, Kew wishes to acknowledge the financial support of Natural England. Natural England also kindly granted permission to sample the Kentish localities. D.S.D. gratefully acknowledges financial support provided by the H. F. Layzell Fellowship of the J. S. Lewis Foundation.
List of all Ophrys species recognized by Delforge (2005) using his narrow species concept. Although Delforge used ‘section Euophrys’, this has been corrected to ‘section Ophrys’, following the International Code of Botanical Nomenclature, Division II, Chapter III. Section 3, Article 22·1 (http://ibot.sav.sk/icbn/main.htm).
|Ophrys species||Section||Delforge group|
|heldreichii var. calypsus||Ophrys||scolopax|
|heldreichii var. pseudoapulica||Ophrys||scolopax|
|heldreichii var. sclechterana||Ophrys||scolopax|
|heldreichii var. scolopaxoides||Ophrys||scolopax|
|panormitana var praecox||Ophrys||exaltata|