Introduction

The orchid genus Dactylorhiza (Orchidoideae, Orchidinae) has an exceptionally chequered taxonomic history. The genus was segregated from Orchis (now known to be only a distant relative) as Dactylorchis by Vermeulen (1947), following rigorous morphological and chromosomal studies, but unfortunately his work was bracketed by scientifically trivial nomenclatural publications (Necker 1790ex Nevski 1935; Soó 1960, 1962) that conferred priority on the name Dactylorhiza. Within the genus, taxonomic controversies have been even more intense, prolonged, and by no means always rooted in genuine science. Once again, Vermeulen (1938, 1947) was the first author to suggest that allopolyploidy — hybridisation accompanied by chromosome doubling within a single generation — was the primary cause underlying many of these controversies.

Much of the research attention subsequently paid to this genus has been motivated by the desire to better understand the evolutionary significance of both allopolyploidy and autopolyploidy — goals that eventually allowed Dactylorhiza to become a model system for the study of whole-genome duplication. The genus features repeated unidrectional allopolyploidisation of the same two diploid (2n = 40) parental groups, D. fuchsii reliably operating as seed parent and D. incarnata as pollen parent. Moreover, the allopolyploidisation events have taken place at contrasting times and between subtly different habitat races of the parental species (e.g. Hedrén et al. 2008; Paun et al. 2010, 2011; Balao et al. 2016; Hawranek 2021; Wolfe et al. 2021; Eriksson et al. 2022; Thornton 2022). In addition to allopolyploidisation, the search for optimal species boundaries made the genus a pioneering case-study for population-level morphometrics. Early univariate approaches (Heslop-Harrison 1948, 1951, 1953, 1954; Roberts 1961a, 1961b) later gave way to computational multivariate techniques (Bateman & Denholm 1983, 1985, 1989; Dufrene et al. 1991; Pedersen 1998; Shipunov et al. 2004; Stahlberg & Hedrén 2008), in one case also employing landmark analysis (Shipunov & Bateman 2005).

Genetic studies of the genus have inevitably reflected the methodologies prevalent at the time each project was pursued. Molecular work began using allozymes in the mid-1990s in both Edinburgh, Scotland and Uppsala, Sweden (Hedrén 1996a, 1996b, 1996c, 2001), closely followed by typological phylogenetic studies spanning the genus that employed nrITS sequences (Pridgeon et al. 1997; Bateman et al. 2003). The 2000s began with analyses based on nuclear AFLPs (Hedrén et al. 2001; De Hert et al. 2012) and later plastid RFLPs (Devos et al. 2006), generated in parallel with more intensively sampled studies based on a combination of nuclear and plastid microsatellites (Hedrén 2003; Hedrén et al. 2007, 2011a; Pillon et al. 2007; Nordstrom & Hedrén 2007, 2009; Stahlberg & Hedrén 2008, 2010; Balao et al. 2016). During the 2010s, the repertoire of techniques successfully applied to the genus expanded further to include methylation-sensitive AFLPs (Paun et al. 2010, 2011) and gene expression patterns (Paun et al. 2011; Balao et al. 2017). The phylogeny of the genus was eventually established more firmly via the nuclear genome-wide RAD-seq approach (Brandrud et al. 2020), which allowed interpretation of the genome to begin to drill down to the level of ecophysiology (Wolfe et al. 2021) and genome dynamics (Hawranek 2021), including investigations of small RNAs (Eriksson 2022; Thornton 2022) and transposable elements (Eriksson 2022; Eriksson et al. 2022).

Synthesis of this veritable mountain of taxonomically relevant data is most parsimoniously (though by no means universally) interpreted as suggesting the presence of seven native species of Dactylorhiza in the British Isles. As determined by Bateman & Denholm (2012) and Bateman (2021, 2022a), these are:

Dactylorhiza viridis (L.) R.M.Bateman, Pridgeon & M.W.Chase (Frog Orchid), reputedly diploid; widespread, but local and decreasing in the south.

Dactylorhiza fuchsii (Druce) Soó (Common Spotted-orchid), diploid; common throughout most of the British Isles.

Dactylorhiza maculata (L.) Soó (Heath Spotted-orchid), autotetraploid; occurs throughout the British Isles but far more commonly in the north, particularly Scotland.

Dactylorhiza incarnata (L.) Soó (Early Marsh-orchid), diploid; widespread but local throughout the British Isles — intolerant of desiccation and divisible into fairly distinct ecotypes.

Dactylorhiza traunsteinerioides (Pugsley) R.M.Bateman & Denholm (Pugsley's Marsh-orchid), allotetraploid; widespread but local, occurring only north of a line connecting mid-Wales with the Humber. [Note that here, for reasons of nomenclatural priority explained below in the concluding section, titled Nomenclatural Postscript, we employ at species level the epithet 'francis-drucei' rather than 'traunsteinerioides', which with regret is demoted to a subspecies of D. francis-drucei (Wilmott) Aver.]

Dactylorhiza praetermissa (Druce) Soó (Southern Marsh-orchid), allotetraploid; frequent in England and Wales, absent from Ireland and Scotland but actively expanding northwestward.

Dactylorhiza purpurella (T.Stephenson & T.A.Stephenson) Soó (Northern Marsh-orchid), allotetraploid; frequent, occurring only north of a line connecting the Severn and Humber estuaries.

Dactylorhiza kerryensis (Wilmott) P.F.Hunt & Summerh. (Irish Marsh-orchid, syn. D. occidentalis), allotetraploid; confined to Ireland, where it is most frequent in the west.

Evidence has progressively accumulated showing that each of the four allotetraploid species is derived from a member of the Dactylorhiza fuchsii–maculata alliance as seed-parent and the D. incarnata clade as pollen parent, and that the two parental clades are only moderately closely related (e.g. Hedrén 1996b; Pillon et al. 2007; Brandrud et al. 2020), having diverged an estimated 8 Myr ago (Brandrud 2019; Hawranek 2021).

Fieldwork for the present study was confined to Scotland, in a focused investigation of 'boreal' dactylorchids that was conceived to address three of the most contentious issues that have long plagued the systematics of British (and indeed European) dactylorchids:

(1) Whether the Gordian Knot of several named taxa collectively known as the narrow-leaved marsh-orchids can ever be satisfactorily untangled (reviewed by Bateman 2011a, 2019; Bateman & Denholm 2012). Three epithets based on Scottish holotypes, francis-drucei (Wilmott 1936) and ebudensis/scotica (Nelson 1976; Wiefelspütz 1976; Landwehr 1977), have variously been treated as species in their own right or alternatively attributed to D. 'traunsteinerioides' (here conversely treated as a subspecies of D. francis-drucei — a species that may or may not be a British and Irish endemic), D. traunsteineri and/or D. lapponica and/or D. majalis (each of which may or may not be exclusively continental) (cf. Kenneth et al. 1988; Roberts 1988; Allan et al. 1993; Lowe 2003; Hedrén et al. 2011a; Bateman 2011a; Bateman & Denholm 2012; Eccarius 2016; Hedrén & Skrede 2018; Stace 2019). British populations of small, boldly-marked plants attributed to 'D. lapponica' (here treated as D. francis-drucei subsp. francis-drucei) were given the maximal conservation protection of being placed on Schedule 8 of the UK's Wildlife and Countryside Act in 1992.

(2) Whether allotetraploid populations of marsh-orchids in North Wales and Scotland that are often comparatively robust and include plants bearing leaf markings, and have been awarded the epithets cambrensis and majaliformis respectively, (a) represent the same taxon, as argued by Bateman & Denholm (1983, 2012), and if so, (b) whether that taxon is best treated as a full (and most likely endemic) species (e.g. Averyanov 1984) or an infraspecific taxon within either Dactylorhiza majalis (e.g. Roberts 1961b, 1966), D. kerryensis (e.g. Campbell 1937; Sell & Murrell 1996) or D. purpurella (e.g. Nelson 1976; Løjtnant 1979).

(3) Whether, despite its striking overall paucity of genetic variation, Dactylorhiza incarnata maintains populations in the British Isles that reliably differ genetically from those already studied in continental Europe, and also whether an unusual population of leaf-marked individuals of this species discovered in the Scottish Highlands in 1982 (Kenneth & Tennant 1984; Bateman & Denholm 1985; Allan et al. 1993) has been correctly attributed to D. incarnata subsp. cruenta (O.F.Müll.) P.D.Sell (Flecked-early Marsh-orchid), a taxon better known from the Alps and Scandinavia. This rare taxon also remains of legislative interest to both the British and Scottish conservation bodies.

Scottish exemplars of these taxa are illustrated in Fig. 1, and recently updated distribution maps are shown for the two allotetraploid species in Fig. 2. Authorites of taxa mentioned in the text are given in Appendix 1.

Fig. 1.
figure 1

Representative plants of the Scottish marsh-orchid taxa analysed in the present study. A, B Dactylorhiza incarnata subsp. cruenta, Lochdroma, Wester Ross; C, D D. purpurella var. purpurella, Aberlady Bay, East Lothian; E, F D. purpurella var. cambrensis, Thurso, Caithness; G, H population previously regarded as D. francis-drucei subsp. traunsteinerioides, Applecross, Wester Ross; J, K D. francis-drucei subsp. francis-drucei s.s., Raasay, North Ebudes; L, M D. francis-drucei subsp. francis-drucei var. ebudensis, North Uist, Outer Hebrides. Enlarged images of flowers are reproduced at a constant scale of 22 mm in image width.

Full size image
Fig. 2.
figure 2

Distribution maps for A Dactylorhiza purpurella and B D. francis-drucei in the British Isles, as revised by the present authors and summarised in the latest British and Irish plant atlas (Stroh et al. 2023). Paler hectads have not been recorded since year 2000.

Full size image

Materials and Methods

Much of the work for this study was conducted in 1995 and 1996, utilising typical contemporary methods. The project was constructed around a combination of allozymes and character-rich multivariate morphometrics, sampling at the population level individual plants each of which provided data for both morphological and molecular analysis. Although the existence of our results from this project has occasionally received brief mention in earlier publications (Bateman 2001, 2011a; Hedrén 2002; Hédren et al. 2011a; Bateman & Denholm 2012), the data have not until now been presented or their implications rigorously explored. Their relevance has not diminished in the intervening years.

Fieldwork

During June – July 1995 and 1996, RMB gathered in situ morphometric measurements from 13 tetraploid marsh-orchid populations in Scotland: five populations of Northern Marsh-orchids (Dactylorhiza purpurella s.l.) and eight populations of narrow-leaved marsh-orchids (D. traunsteinerioides s.l.). The latter were collected under licence from Scottish Natural Heritage. Either 10 or, more often, 20 randomly chosen plants were scored for each population, and single leaves were also collected from each plant and field-chilled for subsequent allozyme analysis by LM and WC in the laboratory of RAE. Three further populations of Northern Marsh-orchid were sampled for allozyme analysis without accompanying morphometric data, whereas conversely, another population yielded morphometric data but no allozyme data. In addition, what was at the time the only known Scottish population of the diploid marsh-orchid D. incarnata subsp. cruenta was analysed for both allozymes (WC, RAE) and morphometrics (RMB), and compared with five populations of this subspecies sampled in west-central Ireland by RMB and ID between 1981 and 1997. Details of the study populations are given in Table 1.

Table 1. Details of Scottish and Irish marsh-orchid (Dactylorhiza) populations sampled for morphometric and allozyme analysis in the present study. 1 Allozyme data only. 2 Morphometric data only. 3 DNA sequence data available. 4 One week added to compensate for an unusually early flowering season. 5 One week subtracted to compensate for an unusually late flowering season. 6 Names of dactylorchid taxa: F, fuchsii; M, maculata; II, incarnata incarnata; IC, incarnata coccinea; ICr, incarnata cruenta; IP, incarnata pulchella; P, purpurella; D, francis-drucei. Taxon frequencies: vr, very rare; r, rare; o, occasional, f, frequent.
Full size table

In addition, small numbers of five species cultivated within the grounds of RBG Edinburgh — two diploids and three putative allotetraploids — were subjected to both allozyme and morphometric analyses in order to provide a broader taxonomic and geographic context for the present study (see also Bateman 2021).

Morphometrics

Character scoring

A complete list of the 52 characters scored by us was detailed by Bateman & Denholm (1985). While in the field we measured in situ 28 vegetative characters plus three floral characters; the remaining 21 characters (C1 – C17, C20 – C21) were recorded on the same data sheet in evening of the same day or subsequently in the microscopy laboratory (C26 – C27). Field measurements were made using a 15 cm steel rule bearing increments of 0.5 mm. A flower–bract unit for subsequent measurement was, wherever feasible, removed from a position one third to halfway from the base of the inflorescence, aiming to minimise the effect of the flower-size decreases from the base to the apex of the inflorescence that are evident in most Eurasian orchid species (Bateman & Rudall 2006). Each flower was initially placed in a numbered vial and later mounted onto double-sided adhesive tape attached to a filing card. Following measurement, these cards acted as compact herbarium vouchers. Metric characters for most floral organs were measured at a resolution of 0.1 mm, using a Leitz ×8 graduated ocular.

The colour of the distal half of each labellum was matched to the nearest one or two colour block(s) of the Royal Horticultural Society Colour Chart. The colours were later quantified through conversion to three CIE (Commission Internationale de I'Eclairage) coordinates. Two of these ('x' and 'y') define a position on a square grid superimposed onto a near-triangular array of colours that pale toward white at the centre of the triangle. The corners correspond with pure blue, pure green and pure red, respectively. Density of pigment was represented by a third coordinate (reflectivity or luminance, 'Y'), which decreases in value outward from the centre of the triangle (see figs 12 and 13 of Bateman et al. 2017; also http://hyperphysics.phy-astr.gsu.edu/hbase/vision/cie.html). A compound microscope was used to count bract marginal cells across three fields of view, each 1.5 mm in diameter, before their mean length was calculated and their average angularity was summarised.

The 52 characters scored describe the stem and inflorescence (5), leaves (12), leaf markings (7), bracts and ovary (7), labellum (14), spur (4) and sepals (3). They can alternatively be categorised collectively as metric (21), meristic (3), multistate-scalar (24) and bistate (4).

Data analysis

Our chosen approach to data analysis and interpretation was both detailed and experimental. Morphometric data for individual plants were summarised on an Excel v15.4 spreadsheet. Mean values, plus sample standard deviations and coefficients of variation for all metric and meristic characters, were calculated for every character in each study population. Univariate and bivariate analyses were summarised and presented using Deltagraph v7.1 (SPSS/Red Rock software 2013).

The morphometric matrix consisted of 209 individuals × 52 characters and contained only 0.3% missing values; no single character incurred more than 5% missing values. The derived matrix of population means consisted of 14 populations × 52 characters and lacked missing values. Both matrices were analysed by multivariate methods using Genstat v14 (Payne et al. 2011).

Of the 52 characters scored, four of the seven leaf-marking characters (C46 – C49) were omitted to avoid over-weighting a feature that appears to reflect only a single underlying gene, and the presence or absence of a basal leaf (C36) was omitted because such leaves had proven vulnerable to premature senescence. The remaining 47 characters were used to compute a symmetrical matrix that quantified the similarities of pairs of data sets (i.e. plants) using the Gower Similarity Coefficient (Gower 1971) on unweighted data sets scaled to unit variance. This similarity measure is comparatively effective when presented with a matrix of heterogeneous characters that includes missing values (Gower & Legendre 1986; Lloyd 2016; Bateman 2022b). The resulting matrix was in turn used to construct a minimum spanning tree (Gower & Ross 1969) and subsequently to calculate principal coordinates (Gower 1966, 1985) — compound vectors that incorporate positively or negatively correlated characters that are most variable and therefore potentially diagnostic. Principal coordinates are especially effective for simultaneously analysing heterogeneous suites of morphological characters and have the additional advantage of comfortably accommodating missing values. Such ordinations have proven invaluable for assessing relationships among orchid species and populations throughout the last four decades, employing a consistent analytical approach that was reviewed in detail by Bateman (2001).

Two separate multivariate analyses were conducted on the putative allotetraploids, the first being based on measurements for individual plants, whereas the second was based on mean values calculated for each analysed variable in each of the 12 study populations. For each of the two multivariate analyses, the first four principal coordinates (PC1 – PC4) were plotted together in pairwise combinations to assess the degree of morphological separation of individuals (and thereby of populations and taxa) in these dimensions, and pseudo-F statistics were obtained to indicate the relative contributions to each coordinate of the original variables.

In addition, the single putative population of Dactylorhiza incarnata subsp. cruenta then known in Scotland was compared morphometrically with Irish populations that have been attributed to this subspecies since they were first recognised as cruenta in 1948 (e.g. Heslop-Harrison 1949); five such populations were measured by RMB and ID between 1981 and 1997 in Cos. Galway and Mayo (Table 1).

Allozymes

Our study benefited considerably from preliminary allozyme studies already performed on the genus Dactylorhiza by Mikael Hedrén (published a year later as Hedrén 1996a, 1996b, 1996c). Of seven allozyme loci explored by Hedrén, the three involving phosphogluconate-group substrates (the dimeric 6-pgd and pgi [syn. gpi] plus the monomeric pgm) had been shown to offer an effective combination of both reliably discriminating between the parents of the western European allotetraploids (D. fuchsii/maculata and D. incarnata) but also showing some variation among populations of the same putative species of allotetraploids. Also, according to Hedrén (1996b), all three systems were competent to resolve allele dosage levels. We therefore elected to focus on these three loci.

Data were collected in the laboratory of RAE by LM in 1995 and by WC in 1996. Chilled leaf tissues were prepared satisfactorily as a crude buffer extract with no elaborate purification or concentration steps; optimisation of pH values proved to be the most crucial methodological challenge (e.g. Wendel & Weeden 1989).

For each individual analysed, approximately 1 cm2 of leaf was ground in 80 μl of a Tris-HCl extraction buffer (Soltis et al. 1983), modified by replacing ß-mercaptoethanol with dithiothreitol. Extracts were absorbed onto paper wicks and proteins were separated on horizontal starch gels at 60 mV for 30 mins until removal of the wicks, after which current was increased to 70 mV for a further 3 – 4 h. Pgi alleles were resolved on the lithium-borate tris-citrate buffer system of Ashton & Braden (1961), as modified according to Lonn & Prentice (1990), whereas a histidine-citrate buffer system (Wendel & Weeden 1989) was used to separate alleles of pgm and 6-pgd. Staining recipes followed Wendel & Weeden (1989) with only minor modifications. Gel patterns were recorded immediately, both graphically and photographically, prior to immersion in a methanol-acetic acid fixative.

Each 20-lane starch gel included extracts from 16 allotetraploid individuals bracketed at either end by extracts from "standard" plants representing the diploid parental genomes. The fuchsii standard was derived from a small population maintained in cultivation in RBG Edinburgh (1984/1618: originally gathered in 1984 from a coal bing at Gorebridge, Midlothian), whereas the incarnata standard was derived directly from a natural population located 21 km east of RBG Edinburgh in extensive dune-slacks at Aberlady Bay (this population was also subjected to morphometric analysis: Table 1). Numbers of individuals analysed per population ranged from five to (more often) 16.

When scoring the resultant gels (cf. Weeden & Wendel 1989), alleles were designated by lower-case letters, beginning with the most rapidly migrating allele. 'Missing letters' denoted alleles found in populations of Dactylorhiza outside the present study, but as summarised for European populations by Hedrén (1996a) rather than later coding employed for a broader spectrum of Eurasian populations by Hedrén (2001). Routine use of the diploid parental 'standards' contributed appreciably to accurate identification of specific alleles; nonetheless, some gels incurred sufficient ambiguity to discourage us from presenting the tentative results (denoted by 'f' in Table 2).

Table 2. Allozyme allele frequencies determined in A natural populations and B cultivated populations of Scottish dactylorchids (f = gel judged inadequate, * = diploid populations routinely employed as allozyme standards). Allele annotation follows Hedrén (1996). For (B), + indicates the reliable presence of the specified allele, whereas ++ indicates consistent double dosage (i.e. homozygosity) for all three loci in Dactylorhiza foliosa.
Full size table

Results

Allozymes

Table 2 details the allele frequencies obtained for the three studied loci for Scottish populations of Dactylorhiza francis-drucei, D. purpurella, D. incarnata and D. fuchsii, together with qualitative results for four additional species held in cultivation at RBG Edinburgh. Letters denoting specific alleles of the three loci follow Hedrén (1996a, 1996b, 1996c).

Diploid species parental to the allotetraploids

Dactylorhiza incarnata yielded greater allozyme diversity, both within and among populations (Table 2A), than was predicted through extrapolation from Hedrén's (1996a) Scandinavian data. Of the four subspecies investigated, three yielded only the b allele for the pdg locus, whereas all three populations of subsp. cruenta included individuals characterised by possession of the a allele. Only two populations of D. incarnata yielded reliable data for the pgi locus, producing the expected result of reliable homozygosity for the slow e allele. For the pgm locus, subspp. incarnata and coccinea (including Aberlady, here used as the allelic yardstick for incarnata) proved consistently homozygous for the c allele, whereas populations of subspp. pulchella and cruenta yielded mixtures of the b and c alleles.

Analysis of Dactylorhiza fuchsii was confined to the single 'yardstick' accession from Gorebridge (Table 2A). This population provided no surprises; it contained the expected pairing of b and c alleles for pgd, the b allele only for pgi, and was, after some debate, judged to bear the d and e alleles for pgm (cf. Hedrén 1996a, 1996b).

Allotetraploids

In the case of Dactylorhiza francis-drucei, pgi profiles were dominated by alleles a and c, though in some populations a minority of plants replaced the a allele with the b allele (Table 2A). Results for pgi were reliably balanced between the fuchsii-derived allele b and incarnata-derived allele e, and pgm typically provided equal frequencies of the b and d alleles, though a few plants in the Aonaich and Kernsary populations replaced the b allele with the c allele. No allelic patterns distinguished between the named taxa traunsteinerioides s.s., francis-drucei s.s. ('lapponica' sensu Kenneth et al. 1988; Stace 1997) and ebudensis.

Results for Dactylorhiza purpurella were straightforward for the pgd and pgi systems. Six of the seven populations studied contained only the b allele for pgd, though the Dunnet population also maintained allele a at a frequency of approximately 7% (Table 2A). Balanced heterozygosity of the b and e alleles was consistent for pgi and of the c and e alleles for pgm, thus contrasting with D. francis-drucei at this locus.

Other RBG Edinburgh accessions

Of the four non-British species cultivated in RBG Edinburgh, to the best of our knowledge only Dactylorhiza majalis s.s. had previously been subjected to allozyme analysis (Hedrén 1996a, 1996b, 1996c). Our results were consistent with previous work on this species, yielding the a and b alleles for pgd, b and e alleles for pgi, and b and d alleles for pgm (Table 2B). Dactylorhiza alpestris surprisingly diverged from the morphologically similar D. majalis by replacing the b allele with the c allele for pgd. A third allotetraploid species, D. elata, matched D. majalis and D. alpestris in pgi and pgm profiles, but consistently possessed the b and c alleles for pgd. The Madeiran island endemic D. foliosa, a diploid more closely related to D. fuchsii than to D. incarnata (e.g. Pillon et al. 2007), maintained only one allele at each locus: the c allele for pgd, the a allele for pgi and, surprisingly, the b allele for pgm — an allele that is characteristic of D. incarnata rather than D. fuchsii. Multiple genetic lines with contrasting geographic origins investigated within three of these four cultivated species consistently yielded identical, species-specific results (Table 2B).

Morphometrics

Tetraploid marsh-orchids: individual plants

The first two principal coordinates for 209 individual plants of Dactylorhiza francis-drucei s.l. and D. purpurella s.l. accounted for 29% of the total variance. We anticipated that the first coordinate would distinguish between the two molecularly circumscribed species, but in fact this role fell to the second coordinate, which permitted only slight overlap of the two species (Fig. 3). This coordinate largely represented the greater vegetative vigour of D. purpurella, being dictated by flower number, the number, length and especially width of the sheathing leaves, and the diameter of the stem. Subsidiary contributors to this coordinate that favoured D. francis-drucei included more angular bract-margin cells, markings more widely distributed across the labellum, a greater frequency of diffuse anthocyanins below the inflorescence, and the widest leaf also being the longest (in the majority of D. purpurella plants the widest leaf was that located immediately below the longest leaf).

Fig. 3.
figure 3

Plot of the first two principal coordinates for 47 diverse morphological characters measured in 209 plants of 14 Scottish populations of the tetraploid marsh-orchids Dactylorhiza francis-drucei and D. purpurella. Parenthetic percentages represent the proportion of the total variance accounted for by each coordinate. Characters contributing significantly to each coordinate are listed in order of decreasing importance, with arrows indicating the direction of increase in value; boldface characters were dominant.

Full size image

The first coordinate reflected the extensive variation evident within both species in a wide spectrum of discrete anthocyanin markings on both floral and vegetative organs. Discrete spots on the bracts and leaves are often accompanied by leaves that tend to be a slightly darker, bluer green, and by bold loop markings (both solid and annular) on the lateral sepals. Weaker positively correlated characters included greater frequency of diffuse anthocyanins on both bracts and stem. The main effect of this axis was to present both species as marginally overlapping horizontally elongate ellipses in Fig. 3, each grading from anthocyanin-poor on the left to anthocyanin-rich on the right. Interestingly, flower colour was far less variable than the remaining anthocyanin-based characters and played no meaningful part in dictating positions of plants on the plot. Taxonomically, the first coordinate was surprisingly effective at largely separating var. cambrensis (Fig. 1E, F) from Dactylorhiza purpurella s.s. (Fig. 1C, D). Less surprisingly, PCo1 also placed Applecross — the only study population of D. francis-drucei to wholly lack leaf-marked individuals (Fig. 1G, H) — at the anthocyanin-low extreme of the francis-drucei ellipse, though it was not resolved as a discrete entity. The third and fourth coordinates were much weaker and lacked taxonomic structure. Significantly, none of the first four coordinates suggested any distinction between ebudensis (Fig. 1L, M) and the far more geographically widespread D. francis-drucei subsp. francis-drucei (Fig. 1J, K).

Tetraploid marsh-orchids: population means

Reducing individual-level data-sets to population means inevitably decreases dimensionality within the data and so allows the first two coordinates to encompass a greater proportion of the total variance — in this case, 47% (Fig. 4). The two coordinates from the individual analysis are essentially transposed at population level, PCo1 being a “vigour” coordinate; it represents most of the characters that contributed to PCo2 in the plot of individuals (Fig. 3). The second coordinate resembles the first coordinate from the individual plants plot in that it represents each of the two species as an elongate ellipse and distinguishes between Dactylorhiza purpurella purpurella and D. purpurella cambrensis. However, the spectrum of characters underlying the axis is somewhat altered, those representing diffuse anthocyanins being promoted at the expense of those representing localised anthocyanin markings. More importantly, in outline the longest and lowest leaves tended to be more rounded in the Thurso and nearby Scrabster subpopulations of cambrensis and hence had more-or-less planar rather than hooded apices. These plants grew in taller vegetation and therefore bore their leaves roughly evenly spaced along the stem (Fig. 1F), but showed unusually low levels of diffuse anthocyanins on stem and bracts. In contrast, the nearby Dunnet population of D. purpurella cambrensis, which occupied exposed and grazed stabilised dune-slacks and so was environmentally dwarfed, is placed close to the D. francis-drucei cluster on Fig. 4, though the two species are connected by a reassuringly weak link on the minimum spanning tree. The strongest links in the tree connect pairs of sampled populations that were either subpopulations of what was effectively one extensive metapopulation (the Scrabster and Thurso populations of cambrensis, and the Suenish and Hornish populations of ebudensis) or occupied near-identical habitats (the dune-slack populations of D. purpurella s.s. from Aberlady and Robach).

Fig. 4.
figure 4

Plot of the first two principal coordinates for 47 diverse morphological characters measured in 14 Scottish populations of the tetraploid marsh-orchids Dactylorhiza francis-drucei and D. purpurella, analysed as population mean values. Parenthetic percentages represent the proportion of the total variance accounted for by each coordinate. Characters contributing significantly to each coordinate are listed in order of decreasing importance, with arrows indicating the direction of increase in value; boldface characters are dominant. Populations are linked by a minimum spanning-tree representing maximum Gower Similarity values.

Full size image

The third and fourth coordinates were weak and offered only very limited discriminatory power among populations of Dactylorhiza francis-drucei. The third coordinate separated the Borrodale population from the remainder on account of its longer, slightly more curvaceous spurs and the light spotting observed on the underside of the leaves of some plants. The fourth coordinate weakly separated ebudensis according to its spurs, which were slightly more saccate than those of francis-drucei s.s.; spur widths measured halfway along spur length were only slightly less than the comparable widths obtained at the spur mouth.

Diploid marsh-orchids: individual plants

The ordination of 60 plants of six populations of Dactylorhiza incarnata cruenta (Fig. 5) yielded a strong first coordinate that largely reflected positive correlation between three characters likely to share expression of a single set of genes: discrete markings on the upper and lower surfaces of the leaves and on the bracts. This axis generated two crude clusters, one notably richer in these vegetative markings and annular markings on the lateral sepals. Five of the six populations sampled contributed individuals to each of the two clusters; for example, of the ten plants sampled at Lochdroma, eight are placed in the markings-rich category and two in the markings-poor category (these plants were later tentatively reassigned to subsp. pulchella). Only the comparatively markings-deficent Bunny population is confined to a single cluster, reflecting a more general trend within Ireland for plants of subsp. cruenta from Co. Clare to be less likely to be anthocyanin-rich than are plants from Co. Mayo (distinguished by blue vs purple symbols respectively in Fig. 5).

Fig. 5.
figure 5

Plot of the first two principal coordinates for 47 diverse morphological characters measured in a total of 60 plants representing one Scottish population (Lochdroma) and five Irish populations of the Flecked-early Marsh-orchid, Dactylorhiza incarnata subsp. cruenta. Parenthetic percentages represent the proportion of the total variance accounted for by each coordinate. Characters contributing significantly to each coordinate are listed in order of decreasing importance, with arrows indicating the direction of increase in value; boldface characters are dominant.

Full size image

The Scottish (i.e. Lochdroma) plants of subsp. cruenta are placed toward the negative end of the appreciably weaker second coordinate, which otherwise discriminates poorly among the five Irish populations. It is largely a vigour coordinate, dictated by several characters that reflect the sizes of both floral and vegetative organs. Lochdroma features labella that are unusually narrow and thus longer than wide; the majority of plants have labella that are entire rather than shallowly three-lobed and are sufficiently small that the markings cover most of the labellar surface rather than leaving an unmarked border (Fig. 1A). Spurs are small, lateral sepals are dominated by annular markings, and most plants also have spotted bracts. Stems and inflorescences are short and narrow (Fig. 1B), and compared with the Irish populations, Lochdroma plants have on average one fewer sheathing leaf. In addition, Lochdroma leaves are only half the length and two-thirds the width of the Irish populations, and most Lochdroma plants have leaves that are spotted on both surfaces (as they are in the Irish population from Keelbridge, which resembles Lochdroma on PCo1). The even weaker third and fourth coordinates served only to largely separate the three study populations from the Irish "Lake District" of Co. Mayo (Mask, Carra and nearby Keelbridge).

Discussion

Comparison of British allozyme profiles with their Scandinavian equivalents

Our limited allozyme study was inspired by extensive research using seven loci conducted by Hedrén (1996a, 1996b, 1996c), primarily sampling Scandinavian dactylorchid populations. On the basis of his results (as summarised in the present Table 3), we selected our three study loci to fulfil two contrasting purposes.

Table 3. Allozyme allele frequencies presented in previous studies of western European dactylorhizas. Sources: Hedrén (1996b) for Dactylorhiza purpurella, Hedrén (1996a) for the remaining taxa; data were derived from Scandinavian populations, except those for D. praetermissa (two populations from SC England). A few minor alleles have been omitted to facilitate comparison with data for the present project summarised in Table 2.
Full size table

In Hedrén’s data, the pgi locus reliably distinguishes the diploid Dactylorhiza fuchsii and its autotetraploid derivative, D. maculata (both dominated by allele b) from D. incarnata, dominated by allele a, and from the six allotetraploid species that possess approximately equal frequencies of alleles a and b — alleles that they acquired through hybridisation between paternal D. incarnata and maternal members of the D. fuchsii clade. The pgd locus was selected because of its apparent ability to subdivide, through contrasting frequencies of the a, b and c alleles, the six allotetraploids analysed by Hedrén into three groups: traunsteineri plus lapponica plus sphagnicola (dominated by a + c), majalis s.s. plus praetermissa (dominated by a + b), and purpurella (predominantly b). Lastly, the pgm locus was selected primarily because of its reputed ability to distinguish purpurella from the remaining allotetraploids through favouring c plus e alleles over the faster pairing of b plus d alleles (Table 3).

Perhaps the most striking feature of Hedrén's (1996a, 1996b, 1996c) data for Dactylorhiza incarnata was the fact that analysis of several populations each of three subspecies nonetheless consistently yielded fixed homozygosity of a single allele for all three of the loci used here (Table 3), helpfully simplifying its identification as the paternal parent of the allopolyploids (though precluding identification of the paternal subspecies). Features of Hedrén's data that were least readily explained primarily involved D. purpurella; specifically, its apparent enrichment in the b allele of pgd and its co-dominance of the c allele of pgm, which on the basis of Scandinavian material was assessed as absent from D. purpurella's supposed pollen-parent D. incarnata and rare in its supposed seed-parent D. fuchsii (Table 3: Hedrén 1996a).

Our results for the pgi locus are entirely consistent with those of Hedrén, usefully serving to demonstrate the allopolyploid nature of both Dactylorhiza francis-drucei and D. purpurella. This locus need detain us no further. However, both the pgd and pgm loci yielded results that were both surprising and informative.

Arguably the greatest surprise was that British and Irish populations of Dactylorhiza incarnata proved to be capable of maintaining allelic diversity, which was observed in two of the four subspecies analysed by us. With the exception of subsp. cruenta, all of the British and Irish populations proved to be characterised by the b allele for pgd, rather than the a allele reported by Hedrén (1996a, 1996b) as being the one and only allele in all Scandinavian populations (this important allelic distinction was checked by running a mixture of British and Scandinavian plants of D. incarnata on the same electrophoretic gel in the laboratories of both RAE and M. Hedrén). The only exceptions to this British–Scandinavian dichotomy were the three populations of subsp. cruenta, each of which contained both the a and b alleles. In the case of the Scottish population of cruenta (Lochdroma), all plants with leaf markings contained the Scandinavian a allele, whereas admixed plants lacking leaf markings (hence provisionally allocated to subsp. pulchella) contained the characteristically British and Irish b allele. Unfortunately, we do not know whether the same correlation between leaf markings and specific pgd alleles characterised the Irish populations of subsp. cruenta at Carra and Gelain; although these populations similarly contained both leaf-marked and unmarked plants, the two morphs were not distinguished reliably when analysed for allozymes.

Similarly, the pgm locus showed the presence of both b and c alleles in both of the populations of Dactylorhiza incarnata subsp. cruenta successfully analysed by us for this locus, but in this case, both of these alleles were also detected in our two sparsely sampled populations of subsp. pulchella. At Lochdroma, the putative pulchella plants were admixed among bona fide plants of subsp. cruenta, but this was not true of the single pulchella plant from Kernsary that reputedly bore the b allele.

Moving on to consider the allotetraploids, our results for the three named morphs of Scottish Dactylorhiza francis-drucei (formerly D. traunsteinerioides s.l.) were consistent with the results for the closely related Scandinavian taxa D. traunsteineri and D. lapponica for all three loci (cf. Tables 2, 3); in both geographic regions, a minority of populations contained a minority of plants bearing the b allele of pgd or the c allele of pgm. In the case of D. purpurella, a minority of Scandinavian plants included the a or c alleles at the pgd locus, but Hedrén (1996b) received a greater surprise from his results at the pgm locus; Scandinavian populations of D. purpurella proved to be uniquely dominated by the c and e alleles of pgm. We found that this pattern was consistently mirrored in all six pgm data-sets for Scottish populations of D. purpurella, irrespective of whether they were attributed to var. purpurella and var. cambrensis. Among the four additional dactylorchid species cultivated at RBG Edinburgh, only D. majalis was also analysed by Hedrén (1996a) from natural populations, the two sets of allozyme results proving congruent.

Comparison of allozyme profiles with DNA sequencing profiles

Hedrén et al. (2011a) sampled allotetraploids widely across the British Isles, though material from Scotland was limited to single populations each of Dactylorhiza francis-drucei traunsteinerioides, francis-drucei s.s. and ebudensis, together with four populations of D. purpurella s.s. confined to the Outer Hebrides plus the Applecross population located on the west coast of the mainland. As with the present allozyme data, D. purpurella proved to be the most genetically variable of the four British and Irish allotetraploid species in both nuclear and especially plastid microsatellites, whereas all three named forms of D. francis-drucei proved similar in terms of nuclear microsatellites and virtually indistinguishable using plastid microsatellites. For both genomes, D. francis-drucei was shown to be more similar to D. praetermissa than to D. purpurella, a pattern mirroring that evident in allozyme results. In terms of nrITS sequences, its approximately equal frequencies of ribotypes III, V and X led to an intermediate placement of D. francis-drucei between D. purpurella (ribotypes X and V) and D. praetermissa (ribotypes III and V). These three ribotypes are readily sourced in putative diploid parents of the allopolyploids: D. incarnata is characterised by ribotype X whereas D. fuchsii commonly features ribotypes III and V (Pillon et al. 2007).

Within Dactylorhiza purpurella, no obvious genetic differences were observed by either Pillon et al. (2007) or Hedrén et al. (2011a) between D. purpurella cambrensis (represented largely by Welsh rather than Scottish material) and D. purpurella s.s. — conclusions that are congruent with the allozyme data presented here. Similarly, within D. francis-drucei, Hedrén et al. (2011a) failed to find any meaningful genetic distinction between ebudensis, francis-drucei s.s. and traunsteinerioides s.s. However, they did detect possible introgression into ebudensis from either admixed D. purpurella or D. incarnata, perhaps explaining the unfortunate recovery of an atypical plastid microsatellite profile from the single plant of ebudensis that was analysed by Pillon et al. (a topic discussed in detail by Bateman 2011a, 2019; Bateman & Denholm 2012). Subsequent in-depth exploration of RAD-seq data confirmed a relatively high frequency of gene flow into ebudensis from intermingled plants of D. purpurella on North Uist (Hawranek 2021).

The data-rich RAD-seq-based phylogenetic study of Brandrud et al. (2020) encompassed mostly continental samples and included only five Scottish plants: single representatives of the diploids Dactylorhiza fuchsii and D. incarnata from the Outer Hebridean island of North Uist, plus a single plant of D. purpurella s.s. from Suenish (also North Uist), single plant of supposed D. francis-drucei traunsteinerioides from Applecross, and a single plant of undoubted D. francis-drucei francis-drucei from Kernsary, close to the type locality for the species. The main conclusions to be drawn from their study were that both D. purpurella and D. francis-drucei s.l. are fairly similar genetically but nonetheless each was resolved as monophyletic, as were D. praetermissa and the wholly continental D. majalis. One conclusion that Brandrud et al. (2020) did not emphasise, but that is clearly evident from their Fig. S3 (Bateman 2019), is that the British and Irish D. francis-drucei is also monophyletic and potentially sister to D. traunsteineri s.s., a species now arguably better viewed as incorporating the former D. lapponica and as being confined to continental Europe (Bateman 2019). This conclusion is also supported by the epigenetic data of Paun et al. (2010) and by extensive nuclear microsatellite data acquired by Balao et al. (2016, their fig. 2), who showed D. francis-drucei to be approximately as genetically distinct from D. traunsteineri as it is from D. majalis.

Overall, the more detailed recent genetic investigations (e.g. Brandrud et al. 2020; Hawranek 2021) have downplayed the relatively distinct and diverse genetics found in Dactylorhiza purpurella during the present study and earlier by Hedrén et al. (2011a). In contrast, RAD-seq has enhanced present and past allozyme data in more reliably distinguishing D. francis-drucei from D. praetermissa within the British Isles, and more importantly, has proved more successful than allozymes and microsatellites in discriminating British and Irish D. francis-drucei from continental D. traunsteineri.

Comparison of morphometric with genetic divergence levels

In summary, the many diverse sources of genetic data now available, including allozymes (Tables 2, 3), are able to readily distinguish and reliably circumscribe both Dactylorhiza purpurella and D. francis-drucei, but they fail to identify any meaningful genetic structure among Scottish populations of either of these allopolyploids. It is therefore of considerable interest that the ordination of individual allotetraploid plants (Fig. 3) shows morphological variation to be marginally greater within these species than between them. The first coordinate effectively separates infraspecific taxa primarily on a range of characters that are dependent on anthocyanin pigments, separation of the two species being relegated to the second coordinate using 'vigour' characters, supported by features such as more angular bract cells and more widely distributed lip markings but less frequent notching of the lateral lobes on the lip in D. francis-drucei. Two similar axes, supported by broadly similar characters, were found in the corresponding morphometric analysis conducted at the population level (Fig. 4), but here the first two axes are transposed in their respective strengths, species distinction taking precedence over distinctions among infraspecific taxa. Nonetheless, the wide scatter of points across the plot, particularly of D. purpurella populations, emphasises both the considerable morphological diversity present within the two allopolyploids and the comparative subtlety of the morphological distinction between the two species. In this particular case, detailed molecular data are more discriminatory at the species level than are detailed morphological data, helping to explain why earlier taxonomic circumscriptions based only on morphology tended to become seduced into over-weighting anthocyanin-based characters, thereby incurring an unacceptably high risk of circumscribing artificial taxa (Bateman 2011a, 2019; Hedrén et al. 2011a; Bateman & Denholm 2012).

The dominance of pigmentation characters predictably extends into the morphometric analysis of six Scottish and Irish populations of Dactylorhiza incarnata subsp. cruenta (Fig. 5), where it dictates the first coordinate to an extent where one might speculate that two morphologically distinguishable taxa are present. Only the fact that five of the six populations have placed at least two individuals in each of the two clusters shows that all of these plants are conspecific. However, it is less clear that they are genuinely consubspecific, because some evidence has accumulated to suggest that there exists genetic structure within this group, in contrast with the inexplicably low levels of genetic variation that characterise the remaining subspecies of D. incarnata (e.g. Hedrén 1996a, 2001, 2003; Pillon et al. 2007; Balao et al. 2016).

Likely origin of Scottish populations of Dactylorhiza incarnata subsp. cruenta

Focusing on the classic Scottish population of cruenta (Lochdroma), all eight plants present in the anthocyanin-rich cluster proved to have the pgd-a allele and pgm-b allele that characterise Scandinavian Dactylorhiza incarnata (Hedrén 1996a), whereas the two admixed plants placed in the comparatively anthocyanin-deficient cluster had the pgd-b allele and pgm-a allele that are here shown to characterise D. incarnata in the British Isles. Similar mixtures of plants bearing either Scandinavian or British/Irish allele profiles, and of plants bearing or lacking discrete leaf markings, were found in the two Irish populations analysed here for allozymes (Carra and Gelain: Table 2A). These results are elegantly congruent with the nuclear and plastid microsatellite-based study of the Gelain incarnata population conducted a decade later. Specifically, Hedrén et al. (2011b) detected strong genetic differentiation between plants with and without leaf markings, the rarer leaf-marked plants showing limited gene-flow with the admixed unmarked plants and greater evidence of inbreeding.

Thus, leaf-marked plants of Dactylorhiza incarnata and the pgd and pgm alleles that are dominant in continental Europe are also positively correlated in the British Isles, but here both this phenotype and this genotype are rare. These results are consistent with comparatively recent (presumably post-glacial) arrival of seed of leaf-marked D. incarnata subsp. cruenta from mainland Europe, followed by limited introgression into pre-existing populations of unmarked subsp. pulchella. If so, it is likely that colonisation of west-central Ireland, where cruenta is now locally frequent, occurred earlier than establishment of the isolated outpost at Lochdroma in west-central Scotland. Admittedly, there exists a potential source of the characteristically continental incarnata alleles pgd-b and pgm-c in the form of D. francis-drucei, but this species would also have been obliged to donate the pgd-c and pgm-d alleles, yet these alleles are absent from all analysed populations of D. incarnata. Also, the closest known locality of D. francis-drucei to Lochdroma is situated 25 km to the south (BSBI DDb 2022). The Lochdroma cruenta population was first found in 1982 (Kenneth & Tennant 1984). Two further supposed populations have since been discovered in Scotland: a small population in West Sutherland (recorded 1998 – 2002) and a larger population on Hoy in the Orkney Islands (2019 onward), though improved images sent to RMB suggest that the latter may actually represent depauperate plants of D. francis-drucei — a species that also occurs within a kilometre of the Sutherland cruenta site. On balance, we consider highly improbable an origin of the cruenta populations through gene-flow from D. francis-drucei; certainly, the Lochdroma plants show no morphological evidence of hybridity (Fig. 1A, B) of the kind observed by Aagaard et al. (2005) in Scandinavia.

More broadly, it might prove instructive to compare allozyme profiles with plastid and nuclear microsatellites for populations assigned to Dactylorhiza incarnata subsp. cruenta across Europe, because Hedrén (2009) showed that Alpine cruenta share with most British plants of D. incarnata plastid haplotype A, whereas cruenta populations in Scandinavia (the type region for cruenta) are dominated by the typically continental B haplotype. It is therefore possible that neither British and Irish nor Alpine leaf-marked populations should strictly be assigned to cruenta.

Likely origin of Dactylorhiza purpurella

The discovery that the pgd-b and pgm-c alleles dominate Dactylorhiza incarnata in the British Isles has even more profound implications for our understanding of the origin of D. purpurella, which is not only uniquely dominated by, but is also homozygous for, pgd-b. Even more tellingly, the data presented in Table 2A suggest that D. purpurella is stably heterozygous for pgm-c and pgm-e, which characterise D. incarnata and D. fuchsii respectively within the British Isles. Admittedly, pgm-b is present at low frequencies in some Scandinavian populations of D. purpurella (Hedrén 1996b), where a ready souce for the b allele can be found in the typical Scandinavian genotype for D. incarnata (Table 3).

Following its original description (Stephenson & Stephenson 1920), Dactylorhiza purpurella was initially regarded as endemic to the British Isles, until suspicions were raised that sporadic populations along the North Sea coasts of northern Denmark (together with the Faroe Islands) and southern Norway might also be attributable to this species (e.g. Pedersen 2007; Eccarius 2016). A dactylorchid population on the Dutch Frisian island of Schiermonnikoog also briefly masqueraded as D. purpurella before being awarded its own highly questionable species epithet, D. vadorum (cf. Kreutz & Dekker 2016a, 2016b).

Significantly, a well-sampled RAD-seq survey of European Dactylorhiza incarnata by Brandrud (2019) revealed a strong separation of British populations from all continental populations, mirroring our allozyme results. Within the British Isles, the degree of genetic divergence from continental populations increased from southeast to northwest. The one exception to this rule was western Norway, where typically British genotypes were detected in D. incarnata using RAD. There is thus an almost perfect coincidence between the geographic distribution of the British/Irish genotype of D. incarnata and the distribution of D. purpurella which, uniquely among the allotetraploids, shares the same distinctive alleles.

These observations suggest that Dactylorhiza purpurella originated within the British Isles, through allopolyploidy between the British/Irish genotypes of D. fuchsii and D. incarnata. Given that the climate of the British Isles was periglacial as little as 11,500 calibrated years ago, it seems likely that D. purpurella originated more recently and was pre-adapted for life in the postglacial landscape of the glaciated northern and western regions of the British Isles (Bateman 2011a; Hedrén et al. 2011a). Given this timescale, emigration to the Faroes, Norway and Denmark is likely to have occurred very recently, presumably through wind-borne or bird-borne seed. The unusual distribution of D. purpurella in Ireland, concentrated in the north and the southeast (Bateman & Denholm 2023) but "inexplicably missing from the Midlands" (Curtis & Thompson 2009: 75), could also indicate relatively recent emigration from mainland Britain. It potentially represents two separate migrations, the first from southwest Scotland to northern Ireland and the second from South Wales to southeast Ireland.

One further aspect of Dactylorhiza purpurella that is particularly intriguing is the fact that, irrespective of the kind of genetic analyses being performed, it is reliably resolved as the allotetraploid that most closely resembles its pollen parent, D. incarnata, rather than its seed parent, D. fuchsii. This outcome is mirrored in morphometric comparisons (Bateman & Denholm, unpublished) and remains in need of a cogent explanation.

Likely origin of Dactylorhiza francis-drucei

Bateman (2006, 2019, 2020) argued that several categories of molecular evidence conspired to suggest that populations commonly assigned to Dactylorhiza francis-drucei s.l. (as D. traunsteinerioides s.l.) had separate evolutionary origins in the Alps, Scandinavia and the British Isles, and should therefore be treated as distinct species. However, recent modelling of RAD-seq data for the allopolyploids in the D. traunsteineri s.l. and D. majalis s.l. groups, performed separately against the ancestral fuchsii and incarnata subgenomes, suggested otherwise (Brandrud 2019). As expected, RAD-seq data indicate that these two allopolyploid lineages had separate origins, majalis emerging first (estimated at 3,000 – 10,000 yr, compared with 2,000 – 5,000 yr for traunsteineri), but a subsequent modelling exercise comparing British vs continental populations of D. traunsteineri s.l. concluded that the traunsteineri group had a single allopolyploid origin, presumably somewhere within continental Europe.

However, this conclusion appears to contradict the unrooted tree generated from the same body of RAD-seq data (fig. S3 of Brandrud et al. 2020), which suggests that British Dactylorhiza francis-drucei populations constitute a separate species from continental D. traunsteineri that may even be marginally more closely related to D. majalis or D. praetermissa. Moreover, ordination of the RAD data showed D. traunsteineri s.s. to be more closely similar to D. majalis than either is to the more discrete cluster of plants representing British D. francis-drucei (Brandrud 2019, fig. 2.3), echoing results obtained earlier from analyses of nuclear microsatellites (fig. 2 of Balao et al. 2016), small non-coding RNAs (Thornton 2022) and methylation (Paun et al. 2010). More recently, a STRUCTURE analysis performed within a study that regrettably assumed monophyly of the narrow-leaved marsh-orchids suggested greater RAD-seq divergence between D. francis-drucei and continental D. traunsteineri/lapponica than was evident between D. francis-drucei and D. purpurella, subsp. francis-drucei populations typically showing greater "purity" than those of subsp. traunsteinerioides (fig. 4 of Hawranek 2021). In contrast, the RAD-seq data of Brandrud et al. (2020) and Hawranek (2021) failed to convincingly distinguish the Scandinavian lapponica (sampled by them in Norway, Sweden, Finland and Estonia) from Alpine traunsteineri (sampled in Switzerland, Austria and Germany) (see also the microsatellite study by Nordström & Hedrén 2008).

We summarise these various studies as strongly suggesting that Dactylorhiza francis-drucei is best treated as a species separate from D. traunsteineri (including the former D. lapponica), and that it speciated more recently than D. majalis (and D. praetermissa) but earlier than D. purpurella (and, we suspect, much earlier than the Irish endemic D. kerryensis: Bateman & Denholm, unpublished). However, it remains uncertain whether D. francis-drucei is an allopatric derivative of D. traunsteineri or arose through a separate allopolyploidy event, both speciation events congruent with D. fuchsii as 'mother' and D. incarnata as 'father'.

The present allozyme data are consistent with either hypothesis, but do usefully contradict the suggestion put forward by Bateman (2006, 2011a) that Dactylorhiza francis-drucei could have originated through an allopolyploidy event that occurred within the British Isles; they also weaken (though not fatally) his argument that D. francis-drucei may never have occurred south of the line demarking the glacial maximum at approximately 20,000 yr. Nonetheless, there is little doubt that populations formerly attributed to D. francis-drucei (as D. traunsteinerioides) but occurring south of the glacial maximum have correctly been reassigned to D. praetermissa as subsp. schoenophila (Bateman & Denholm 2012; Bateman 2019).

Dactylorhiza francis-drucei reliably yields the pgd and pgm alleles that are characteristic of D. incarnata populations in mainland Europe rather than those in the British Isles, and thus most likely had a continental 'father'. On balance, an allopatric origin in Britain from within D. traunsteineri soon after its own origin in mainland Europe currently appears to be the most likely scenario. The presence among the plants analysed through RAD-seq by Brandrud et al. (2020) and Hawranek (2021) of a single Norwegian plant attributed by them to D. traunsteineri that bore a genotype typical of British D. francis-drucei suggests the possibility of secondary migration of this lineage from Scotland to Scandinavia (Bateman 2019, 2022a), thus mirroring geographically the likely emigration to Norway of D. purpurella discussed above. If confirmed, the presence of D. francis-drucei in Norway would challenge its current status as strictly endemic to the British Isles (Bateman 2022a; Bateman & Denholm 2023).

Conclusions

(1) In the case of Scottish marsh-orchids, a series of genetic studies of increasing technological sophistication has both optimised their taxonomy and deepened our understanding of their evolutionary patterns and processes.

(2) The present results arguably endorse all of the taxonomic conclusions put forward for the diploid marsh-orchids by Bateman & Denholm (1985) and for the tetraploid marsh-orchids by Bateman & Denholm (2012). During the last decade, all four bona fide tetraploid marsh-orchid species native to Britain and Ireland have been re-circumscribed taxonomically in the light of molecular and, to a lesser degree, morphometric reappraisal as certain genotypes, phenotypes and/or regional ecotypes were transferred from one named species to another. The resulting taxonomic circumscriptions have largely been followed by subsequent authors in both Britain (e.g. Harrap & Harrap 2009; Stace 2019; Cole & Waller 2020; Stroh et al. 2023) and continental Europe (e.g. Delforge 2016; Eccarius 2016). Nonetheless, we predict that debates will continue regarding whether the species concept applied to the allopolyploids should prioritise having broadly the same parental species (thus yielding an exceptionally broadly circumscribed Dactylorhiza majalis) or, as here, we should give priority to multiple origins from different ecological races evident within the two parental species. In our opinion, also still undecided is the important question of whether D. francis-drucei emerged relatively recently from within D. traunsteineri/lapponica or alternatively represents an independent polyploidy event.

(3) Scotland supports a single diploid species, Dactylorhiza incarnata, containing four formally named infraspecfic taxa: subspp. incarnata, coccinea, pulchella and cruenta. Whether the first three taxa should be viewed as subspecies or varieties remains debatable (cf. Haggar 2004; Cole & Waller 2020), but the combination of genetic and morphological distinctiveness definitely justifies subspecies status for cruenta.

(4) Scotland currently hosts two tetraploid marsh-orchid species, Dactylorhiza purpurella and D. francis-drucei, which are genetically distinct, morphologically separable, differ in ecological preferences, and have separate evolutionary origins. According to Swainbank (2022), the two species can even be distinguished when in fruit, through the comparatively elongate pods and longer seeds of D. purpurella. Bateman & Denholm (2012) ascribed relatively anthocyanin-rich populations of D. purpurella to var. cambrensis (syn. majaliformis, encompassing only a minority of Scottish populations of the species) and those of the species then named D. traunsteinerioides to subsp. francis-drucei (a taxon encompassing all Scottish populations of the species: Bateman 2022a; Bateman & Denholm 2023). In both cases, the anthocyanin-rich mode forms a morphological continuum with the less anthocyanin-rich race. In both cases, the two taxa together span an approximately equal range of morphological variation. And in neither case do the anthocyanin-rich and anthocyanin-poor populations appear readily distinguishable genetically. A legitimate argument could therefore be put forward for re-equilibrating var. cambrensis and subsp. francis-drucei to equal rank. However, if D. francis-drucei subsp. traunsteinerioides (and thereby the equivalent nominate subspecies, francis-drucei) were to be demoted to varietal status, we would then also be obliged to demote the North Uist 'endemic' ebudensis — viewed as a full species by Bateman (2006) but shown here to clearly be a morphological subset of subsp. francis-drucei — from a variety to a mere forma. Given that the narrow-leaved marsh-orchids lie at the epicentre of this perennial taxonomic Gordian Knot, we do not propose to engineer further taxonomic changes (preferably driven by science rather than mere nomenclatural priority) until a broader morphometric survey, spanning all taxa and the whole of the British Isles, has been completed (Bateman & Denholm, unpublished).

(5) Similarly, the equally complex taxonomic and nomenclatural issues surrounding Dactylorhiza incarnata cruenta (cf. Vermeulen 1947; Heslop-Harrison 1949; Summerhayes 1951; Bateman & Denholm 1985; Haggar 2004; Curtis & Thompson 2009; Hedrén et al. 2011b; Eccarius 2016) will not be solved until high-throughput sequence data and detailed morphometric data are gathered from populations scattered across Europe. For the present, we prefer to continue using the long-recognised epithet cruenta.

(6) Combining genetic approaches with detailed in situ morphometrics has proved to be an especially powerful approach to taxonomic circumscription, offering the opportunity to assess robustly the all-important degree of congruence between genotype and phenotype.

(7) There has been much recent discussion of 'cryptic speciation', when genotypic divergence is hypothesised to precede phenotypic divergence (e.g. Monro & Mayo 2022). However, in circumstances when speciation occurred fairly recently and involved closely similar parental lineages, it is alternatively possible that relative levels of genetic divergence will generate more accurate taxonomic circumscriptions than will relative levels of phenotypic divergence. In the case of Dactylorhiza purpurella and D. francis-drucei, infraspecific variation in anthocyanin-related characters is approximately as great as phenotypic differences that genuinely distinguish the two species and reflect their independent origins through allopolyploid speciation.

(8) Deciding whether to award pre-eminence to genotypic or phenotypic data is especially relevant to determining whether or not Dactylorhiza francis-drucei should be treated as a species separate from continental D. traunsteineri. The two taxa can be genetically circumscribed but our ongoing morphometric comparisons (Bateman & Denholm, unpublished) indicate that there are no reliable morphological features competent to distinguish British and Irish D. francis-drucei from Alpine D. traunsteineri or Scandinavian populations often attributed to D. lapponica.

(9) From the viewpoint of conservation, application of IUCN criteria has meant that both Dactylorhiza francis-drucei (as D. traunsteinerioides) and D. purpurella have consistently, and correctly, been designated Least Concern in both the UK (e.g. Cheffings et al. 2005) and Ireland (Wyse Jackson et al. 2016). This statement also applies to D. incarnata subsp. cruenta in Ireland, whereas in Scotland cruenta rightly returned to its original designation of Endangered in 2010 (Leach 2010) after spending the previous five years languishing in the bureaucratic doldrums of Data Deficiency. Despite its Least Concern rating, D. francis-drucei should be taken seriously as a good indicator species for relatively biodiverse habitats in Scotland, particularly for slopes that support calcareous flushes featuring reliable groundwater movement (Cowie 1999).

(10) April 2022 witnessed submission to the UK government of recommendations for the seventh quinquennial review of Schedule 8 of the Wildlife and Countryside Act. In total, 16 orchid species and subspecies figured among the 306 vascular plants that were considered for Schedule 8 status (JNCC 2022). As a result, the only orchid 'species' recommended for removal from Schedule 8 is Dactylorhiza francis-drucei subsp. francis drucei (a taxon formerly mis-assigned to D. lapponica) — a decision that we support in the light of greatly improved knowledge of its distribution within Scotland. It will also end the irony of offering protection to a species that in truth does not occur in Britain and that, in any case, may not be a valid species, either biologically or nomenclaturally. Unfortunately, the application of the single Scottish population of D. incarnata subsp. cruenta to join this exclusive club has been recommended for rejection, in stark contrast with the recommended acceptance into Schedule 8 of the two East Anglian populations of the less genetically distinct D. incarnata subsp. ochroleuca (Bateman 2022a). There is much to be said for consistency of decision-making.

(11) Comparison of the 2000 and 2020 plant atlases for Britain and Ireland (cf. Preston et al. 2002; Stroh et al. 2023) reveals not only improved mapping of some species, particularly Dactylorhiza francis-drucei (Fig. 2), but also considerable distributional changes among Dactylorhiza species. Most notable is the rapid and inexorable northward march of the Southern Marsh-orchid, D. praetermissa, which recently reached Cumbria and Northumbria and now appears poised to invade Scotland (Bateman 2022a; Bateman & Denholm 2023). Migration of D. praetermissa seed across the Irish Sea to Ireland (perhaps also of D. kerryensis seed outwards from Ireland) might also be predicted in the near future. Thus far, although adaptable, D. praetermissa does not seem to be invading those habitats preferred by D. francis-drucei, but limited evidence suggests that it is an enthusiastic hybridiser with D. purpurella (e.g. Stace et al. 2016). It will therefore be interesting to see how rapidly these current genetic boundaries become blurred between these two closely related species as the contact zone expands into Scotland. Credible predictions of the likely consequences of migration will require detailed study of existing multi-species colonies using modern technologies.

Nomenclatural Postscript

In his 20-year review of research into British and Irish orchids, Bateman (2022b, p. 362) re-asserted his regret that a purely nomenclatural law, rather than any scientific argument, had forced a nomenclatural change of the Irish endemic Dactylorhiza occidentalis to the less well-known epithet D. kerryensis, a taxon formerly generally viewed as an infraspecific taxon of D. occidentalis (Bateman & Denholm 2009, 2012). Previously, a detailed case for nomenclatural conservation, co-authored by most of the authorities then working on this genus in Europe (Bateman et al. 2010), had been summarily rejected by the ultra-conservative panel appointed to police such requests on behalf of the International Code of Nomenclature (Brummitt 2011).

Bateman's (2022b) published comment encouraged orchid enthusiast Felix Benoit to contact him in August 2022 in order to point out the existence of a parallel situation that regrettably afflicts Dactylorhiza traunsteinerioides. This unfortunate taxon has long been treated as the ultimate 'taxonomic football', having incurred changes of epithet and/or rank on a regular basis; some of these name changes were motivated by scientific advances but others were purely legalistic. As reviewed by Bateman (2011a), formalised by Bateman & Denholm (2012) and subsequently updated by Bateman (2022b), Pugsley's Marsh-orchid, D. traunsteinerioides, was perceived as a tetraploid species endemic or near-endemic to Britain and Ireland that shows sufficient morphological variability to warrant division into two subspecies: a more southerly nominate subspecies inhabiting Ireland, North Wales and northern England, and a typically smaller-bodied, smaller-flowered, often more intensely marked subspecies that is characteristic of Scotland and is named subsp. francis-drucei. However, this hierarchical relationship between the epithets traunsteinerioides and francis-drucei has now been subjected to a purely nomenclatural challenge.

The epithets traunsteinerioides (Pugsley 1936) and francis-drucei (Wilmott 1936) were formally established in successive papers published in the same issue of the Proceedings of the Linnean Society, the former epithet preceding the latter by just four pages. Unfortunately, francis-drucei was established by Wilmott at species level, whereas traunsteinerioides was established at subspecies level; Pugsley only raised traunsteinerioides to species level four years later, a decision taken under implicit pressure from fellow contemporary dactylorchid enthusiasts, not least Wilmott himself (Pugsley 1940). Thus, as noted in litt. by Benoit, the presently accepted relationship between the epithets traunsteinerioides and francis-drucei should strictly be reversed; as a bona fide species, Pugsley's Marsh-orchid should strictly be attributed to the epithet first employed at species level, namely D. francis-drucei (Wilmott) Aver. (Averyanov 1984).

Accepting the blanket dictat that is the (often infuriating) law of nomenclatural priority therefore requires the well-known epithet traunsteinerioides to be demoted to a southerly subspecies of a newly promoted species, Dactylorhiza francis-drucei. As currently religiously applied, the law of priority overrides all scientific and pragmatic arguments that favour the epithet traunsteinerioides over the epithet francis-drucei — that traunsteinerioides has finally achieved stability for this most unstable of taxa, that it helpfully indicates its closeness of relationship to the Alpine D. traunsteineri (and without employing an irritating internal hyphen), that its protologue was arguably more competently prepared, and that its holotype is more typical of the morphology of the species as a whole (the morphologically extreme holotype of francis-drucei was illustrated by Bateman & Denholm 2012, p. 42).

Moreover, the most popular flora of the British Isles (Stace 2019), the accompanying hybrid flora (Stace et al. 2016), and the latest UK plant atlas (Stroh et al. 2023) — a tome that is set to summarise the distribution of that flora for the next 20 years — all employ the previous nomenclature, and so will inevitably provide passive resistance against the altered names. We would certainly be grateful if the venerable custodians of the International Code of Nomenclature would consider developing a more liberal approach to scientifically-based applications for nomenclatural conservation (cf. the discussions of Dactylorhiza occidentalis vs D. kerryensis in Bateman 2011b, 2022b; Bateman & Denholm 2012).

Dactylorhiza francis-drucei (Wilmott) Aver., Bot. Zhurn. 69: 875 (Averyanov 1984).

Basionym: Orchis francis-drucei Wilmott, Proc. Linn. Soc. London 148: 128 (1936).

Type: 'West Ross; slopes above Loch Maree, 23 June 1935, coll. A.J. Wilmott’ (Wilmott, 1936, p. 128) (holotype BM: fig. 1 of Bateman & Denholm 2012).

Synonyms (selected): Orchis traunsteinerioides (Pugsley) Pugsley, J. Bot. (London) 78: 179 (1940); Dactylorhiza traunsteinerioides (Pugsley) Landwehr, Orchideeën 37: 79 (1975), ex R.M.Bateman & Denholm in List Vasc. Pl. Brit. Isles (D. H. Kent) Supp. 3: 20 (2006); Dactylorhiza traunsteineri (Saut.) Soó subsp. traunsteinerioides (Pugsley) Soó, Nom. Nov. Gen. Dactylorhiza 6 (1962).

Dactylorhiza francis-drucei subsp. francis-drucei var. ebudensis (Wief. ex R.M.Bateman & Denholm) R.M.Bateman & Denholm, comb. nov.

http://www.ipni.org/urn:lsid:ipni.org:names:77308347-1

Basionym: Dactylorhiza majalis (Rchb.) P.F.Hunt & Summerh. var. ebudensis Wief. ex R.M.Bateman & Denholm, Edinburgh J. Bot. 52: 57 (1995). Type: Scotland, North Uist, Lingay Strand, in dunes near Newton Hotel, 4 June 1974, W. Wiefelspütz, DM 37 (lectotype HEID).

Synonyms (selected): Dactylorhiza majalis (Rchb.) P.F.Hunt & Summerh. subsp. scotica E.Nelson, Mon. Ikon. Orch. Dactylorhiza: 90 (1976), nom. nud.; D. majalis (Rchb.) P.F.Hunt & Summerh. subsp. ebudensis (Wief. ex R.M.Bateman & Denholm) M.R.Lowe, Eurorchis 15: 81 (2003); D. ebudensis (Wief. ex R.M.Bateman & Denholm) P.Delforge, Naturalistes Belges 81: 397 (2000).

Dactylorhiza francis-drucei subsp. traunsteinerioides (Pugsley) R.M.Bateman & Denholm, comb. et stat. nov.

http://www.ipni.org/urn:lsid:ipni.org:names:77308335-1

Basionym: Orchis majalis Rchb. subsp. traunsteinerioides Pugsley, Proc. Linn. Soc. London 148: 124 (1936). Type: Ireland, Co. Wicklow, Newcastle, coll. H. W. Pugsley (#530).

Synonym (selected): Dactylorhiza majalis (Rchb.) P.F.Hunt & Summerh. subsp. traunsteinerioides R.M.Bateman & Denholm, Watsonia 14: 372 (1983).

The formal descriptions of these taxa given in Bateman & Denholm (2012) remain adequate to support the two subspecies and the variety, pending a future morphometric analysis richer in data and broader in taxonomic scale; although the status of their respective names has changed, their circumscriptions have not.