Transcript
Taxonomic reassessment of the British andIrish tetraploid marsh-orchids R. M. Bateman * 1 , I. Denholm 2 1 Jodrell Laboratory, Royal Botanic Gardens Kew, Richmond, Surrey, TW9 3DS, UK, 2 School of Life Sciences,University of Hertfordshire, Hatfield, Herts AL10 9AB, UK Spasmodic progress in understanding the morphology of European allotetraploid marsh-orchids of thegenus Dactylorhiza achieved in the second half of the twentieth century has in the twenty-first century beenlargely superseded by rapid progress in the molecular systematics of the group. However, a grandsynthesis is still needed to integrate the re-circumscriptions of taxa achieved primarily through moleculardata with improved identification ability derived from corresponding morphometric datasets. In anticipationof such a synthesis for the British Isles, we here summarise the taxonomic consequences of recentbiosystematic research into the genus. These changes better accommodate infraspecific variation thatreflects either local ecologically specialised races or especially regional contrasts in the frequency anddegree of expression of anthocyanin pigments. We review recent nomenclatural changes, supporting suchchanges – and enacting further changes – only where they are clearly wholly unavoidable. Dactylorhiza traunsteinerioides (Pugsley) Landwehr ex R.M. Bateman & Denholm ssp. francis-drucei (Wilmott) R.M.Bateman & Denholm, comb. nov., encompasses all Scottish (and probably some northern Irish)populations of D. traunsteinerioides , including those formerly attributed to D. ‘lapponica’ p.p. and D.‘ebudensis’ ; the latter becomes D. traunsteinerioides ssp. francis-drucei var. ebudensis (Wiefelspu¨tz exR.M. Bateman & Denholm) R.M. Bateman & Denholm, comb. nov. Also, Dactylorhiza praetermissa (Druce)Soo´ ssp. schoenophila R.M. Bateman & Denholm, ssp. nov., is established to encompass populationslocated in southeast England, south of a line from the Severn to the Wash, that were formerly attributed to D.traunsteinerioides on the basis of their gross morphology and habitat preference but have genotypescharacteristic of D. praetermissa ; varietal level may eventually prove more appropriate. A relatively broadmorphological circumscription is recommended for D. purpurella (T. & T.A. Steph.) Soo´ var. cambrensis (R.H. Roberts) R.M. Bateman & Denholm. The evolutionary histories and present conservation status ofBritish and Irish tetraploid marsh-orchids are also summarised. Keywords: allotetraploid, dactylorchid, Dactylorhiza majalis , DNA, ecotypes, infraspecific taxonomy, morphometrics, pigmentation Introduction Three decades ago, building on a fine tradition of morphometric analysis of the UK’s dactylorchid taxa(e.g. Heslop-Harrison, 1954; Roberts, 1966), we beganour morphometric monograph of the genus Dac-tylorhiza in the British Isles (Bateman & Denholm,1983 et seq .) by systematically revising the tetraploidmarsh-orchids, D. majalis (Rchb.) P.F. Hunt &Summerh. sensu lato (2 n 5 80). Our analytical approachwas based on quantifying a large number (initially 51,later 52) of morphometric characters for ten plantseach sampled from a modest number (15) of popula-tions that, nonetheless, encompassed all of the speciesandsubspecies,andthemajorityofvarietiesandforms,then recognised (though the study did not includeScottish populations). Although much of our formalclassification was based on quantitative analysis,limited sampling of populations meant that we wereobliged to rely on a thorough literature review todescribe, rank and, in cases where it appearednecessary, synonymise the remaining named infraspe-cific taxa previously reported from Britain and Ireland.Our study was further limited by exclusive reliance onphenotypic data; genotypic analyses based on DNAsequencing were still in their infancy in the early 1980s.A few subsequent studies obtained detailed mor-phometric datasets from the tetraploid marsh-orchidsof Scotland (e.g. McLeod, 1995), Belgium (e.g.Tyteca & Gathoye, 1989), and Scandinavia (e.g.Pedersen, 1998, 2004, 2007, 2010a, b). However, thisperiod of investigation was notable primarily forprogressively added, contrasting genotypic datasetsto supplement (or, in a disappointingly high propor-tion of studies, to substitute for) phenotypic data.Allozymes (McLeod, 1995; Hedre´n, 1996a, b) weresucceeded by genome fragmentation techniques Corresponding author, email:
[email protected] ß Botanical Society of the British Isles 2012 DOI 10.1179/2042349712Y.0000000004 New Journal of Botany 2012 VOL . 2 NO . 1 37 (Hedre´n et al ., 2001; Devos et al ., 2006), then nuclearand plastid DNA sequencing (Bateman et al ., 2003;Pillon et al ., 2007), then nuclear and plastid micro-satellites (Nordstro¨m & Hedre´n, 2009; Hedre´n et al .,2011a). The latest approaches were based on epige-netics (Paun et al ., 2010) and gene expression (Paun et al ., 2011).In a recent review, Bateman (2011a) argued thatthese recent technical innovations, combined withimproved geographical sampling, have greatlyimproved our understanding of the evolution – andthereby of the systematic relationships and taxo-nomic circumscriptions – of the tetraploid marsh-orchids. However, he also observed that agreementamong scientists actively researching these issues hasnot led to similar agreement on optimal taxonomicstatus. Most notably, the ‘British School’ currentlyviews the four native tetraploid marsh-orchids (andalso, importantly, the two native spotted-orchids) asfull species (e.g. Bateman, 2006a, 2011a, b; Stace,2010). In contrast, the ‘Scandinavian School’ viewsthe four tetraploid marsh-orchids as subspecies of anexceptionally morphologically and ecologically vari-able (and undoubtedly polyphyletic) ‘superspecies’, D. majalis sensu latissimo , and the two spotted-orchids as subspecies of an exceptionally morpholo-gically and ecologically variable (diploid-polyploidaggregate) ‘superspecies’, D. maculata sensu latissimo (e.g. Pedersen, 1998 et seq .; Hedre´n, 2002; Hedre´n et al ., 2008; Nordstro¨m & Hedre´n, 2009).Thus far, attempts to summarise these biosyste-matic advances in order to better classify andconserve British and Irish dactylorchids have focusedon re-circumscription of species (as recognised by theBritish School: Bateman, 2006a, 2011a), acknowl-edging that the species constitutes the most mean-ingful of all taxonomic levels. However, this decisionalso implicitly recognised that operating below thespecies level introduces into any such discussion anintimidating plethora of taxa regarded as infraspecificby both the British and Scandinavian schools. Thesetaxa remain embedded in an even more extensivemorass of prior names that seriously challengeattempts at circumscription, description and identifi-cation, as well as often bringing the over-valuedformal law of nomenclatural priority into conflictwith the under valued informal law of nomenclaturalfamiliarity (cf. Pedersen et al ., 2003; Kreutz, 2004;Bateman et al ., 2010; Pedersen, 2011).In order to address these issues, through the last30 years we have spasmodically increased to 52 thenumber of tetraploid marsh-orchid populationsanalysed morphometrically by us; also, from 1996onward we have stored dried tissue samples for DNAsequencing. In theory, we are now in a position tointegrate phenotypic and genotypic data for almostall formally named taxa, however seemingly trivial,though in practice we have not yet conducted ourmorphological ‘grand synthesis’ using multivariatemorphometrics. In addition, we have consistentlybeen reluctant to make piecemeal taxonomic changes,because we believe strongly that formal taxonomictreatments are better conducted as the final elementof a research project – after multiple sources of datahave been painstakingly gathered, thoroughly ana-lysed, and their systematic implications carefullyconsidered (Bateman, 2009, 2011a). However,experience has taught us that some of our colleaguesreach taxonomic judgements far more hastily, in theirunderstandable (if sometimes counter-productive)enthusiasm to achieve fuller and more detailedtaxonomic coverage (compare the highly contrastingclassifications of tetraploid marsh-orchids summarisedin Table 1). Indeed, it is the multiple, Europe-widemonographic treatments that are currently in prepara-tion that have encouraged us to accelerate our plans toreview the taxonomy and nomenclature of the Britishand Irish tetraploid marsh-orchids, in advance of completing our own morphometric monograph. Weemphasise that our taxonomic decisions remainsubject to future change as yet more data accumulate. Rationale of the present taxonomic treatment Comparison of extensive morphometric and molecu-lar data gathered from European dactylorchidsdemonstrates that the single most frequent categoryof taxonomic error has been to over-estimate thesystematic value of anthocyanin-based characters.These pigmentation characters are both exceptionallyvisually striking and capable of being expressedthroughout the body of the plant, thereby influencingnot only flower colour and markings but also bothdiffuse pigments and localised markings of all of theabove-ground vegetative organs of the plant. Thus,many of the most visually striking features of adactylorchid tend to be positively correlated andare hypothesised to reflect contrasts in expression of very few underlying genes. As there is inevitably atemptation to separate such striking variation inpigmentation into anthocyanin-poor and anthocya-nin-rich morphs and to formally name them, thissuite of characters has strongly influenced, arguablydominated (Bateman, 2006a) the taxonomy of Dactylorhiza through the centuries.Each of the four allotetraploid species found in theBritish Isles encompasses at least one named antho-cyanin-poor morph (usually the nominate race) andat least one anthocyanin-rich morph. In the cases of at least two of the species, D. traunsteinerioides and D. purpurella , the anthocyanin-rich morphs increasein frequency, relative to the conspecific anthocyanin-poor morphs, toward the north and west. Moreover, Bateman and Denholm Tetraploid marsh-orchid systematics 38 New Journal of Botany 2012 VOL . 2 NO . 1 the distribution of the relatively anthocyanin-rich D.kerryensis / occidentalis is also predominantly westerlywithin Ireland (Bateman, 2006a, 2011a, b; Hedre´n et al ., 2011a). However, it is uncommon for popula-tions of any of these three species to be composedwholly of either anthocyanin-rich or anthocyanin-poor individuals. These facts cause serious problemsof identification: Should the field botanist applynames to individuals or to populations? How muchvariation in pigmentation was permitted in thesrcinal description (protologue) of the named taxon?And should subsequent botanists feel empowered toexpand the srcinal circumscription in order to betteraccommodate such variation, or is it preferable tocoin yet another formal epithet to circumscribe therelevant phenotype (noting in passing that epithetsare by no means in short supply in the byzantinerealm of dactylorchid taxonomy)?From our earliest work (Bateman & Denholm,1983) onwards, we have viewed all taxonomic ranksbut forma as being applicable primarily to popula-tions rather than individuals. We have also preferredto expand previous diagnoses, and where necessary toimplement new combinations, rather than coin newepithets (cf. Bateman, 2009). Here, for the first time,we have broken this self-imposed constraint, feelingobliged to coin one new name in order to finally layto rest long-standing controversies surrounding thecircumscriptions of infraspecific taxa that havebeen widely regarded as belonging to that mostproblematic of allotetraploid dactylorchids, D. traun-steineri(oides) . Much-debated taxonomic problemssurrounding D. ‘lapponica’ and D. ‘ebudensis’ areremedied via new combinations, and several purelynomenclatural issues are also addressed, thoughnot all are satisfactorily resolved. It will be evidentfrom Table 1 that these taxonomic decisions divergemoderately from the circumscriptions and classifica-tion given in edition 3 of the New Flora of the BritishIsles (Stace, 2010) and substantially from those inedition 2 (Stace, 1997), which formed the frameworkfor the national plant atlas (Preston et al ., 2002). Ourclassification has remarkably little in common withthe treatment, rich in infraspecific taxa, which waslaid out in the Flora of Great Britain and Ireland (Sell& Murrell, 1996).We have organised the following taxonomicaccount under the four main epithets that weretreated first as subspecies of D. majalis by Bateman &Denholm (1983) and later as species by Bateman(2006a, 2011a; contra Hedre´n et al ., 2011a). Formaldiagnoses also follow the style laid down by Bateman& Denholm (1983): taxa are compared via a chosenthreshold value for each character; this approximatesa trough that separates two peaks corresponding tothe modal values among individuals of the pair of taxa under comparison. Here, the descriptions arerestricted to characters thought to provide statisti-cally significant distinctions between the two taxa.Adjectives describing the frequency of that characterstate in that particular taxon represent the followingfrequencies:rarely ( 20%,occasionally 5 20–50%,often 5 51–80%, usually > 80%. The most convincingly diag-nostic characters are italicised. Dactylorhiza traunsteinerioides A recent historical review of the systematics of thenarrow-leaved marsh-orchids (Bateman, 2011a) delib-erately focused on the species level, emphasising inparticular the recent molecular evidence that arguesstrongly for reassignment to D. praetermissa ( q.v .) of all populations that were previously attributed to D.traunsteineri(oides) but occur southofaline extendingfrom the Severn to the Wash (Hedre´n et al ., 2011a).Formalinfraspecifictaxonomywas deliberately evadedby both Bateman and Hedre´n et al .The earlier decision of Bateman & Denholm (1983)to recognise D. traunsteinerioides as a species dis-tinguishable from the Alpine D. traunsteineri wasexplicitly tentative, being based primarily on per-ceived subtle morphological differences betweenmorphometric data on British and Irish populationsversus traditional descriptions published in the classi-cal taxonomic literature. Other authors remainedconvinced of the conspecificity of British and Alpinenarrow-leaved marsh-orchids (e.g. Roberts, 1988;Foley, 1990, 1994). Nonetheless, several moleculardatasets acquired more recently (Pillon et al ., 2007;Nordstro¨m & Hedre´n, 2008; Paun et al ., 2010;Hedre´n et al ., 2011a) have encouraged us to continueto distinguish at species level between narrow-leavedallotetraploid marsh-orchid taxa found in threedisjunct regions: the Alps ( D. traunsteineri [Sauter]Soo´), Scandinavia ( D. lapponica [Laest. ex Hartman]Soo´), and the British Isles ( D. traunsteinerioides [Pugsley] Landwehr ex R.M. Bateman & Denholm).Although the morphometric distinctions betweenthese taxa remain subtle, the cumulative molecularevidence supports arguments that they had indepen-dent evolutionary origins in their three respectivegeographical enclaves (Bateman, 2006a, 2011a; Pillon et al ., 2007; Nordstro¨m & Hedre´n, 2008, 2009; Paun et al ., 2010; Hedre´n et al ., 2011a).The name D. traunsteinerioides is based on ‘ Orchismajalis ssp. traunsteinerioides ’ Pugsley (1935, p. 124),the holotype originating from a population nearWicklow in EC Ireland that is believed to no longerbe extant. Several permutations of the epithet weresubsequently devised, reviewed by Bateman & Den-holm (1983), but only recently did Clive Stace informus that the combination Dactylorhiza traunsteiner-ioides (Pugsley) Landwehr was illegitimate because the Bateman and Denholm Tetraploid marsh-orchid systematics 40 New Journal of Botany 2012 VOL . 2 NO . 1 basionym was specified incorrectly by Landwehr(1975, p. 80). We therefore corrected this legalisticerror at Clive’s behest (Bateman & Denholm in Stace,2006, p. 20).Given the clear importance of geography in dic-tating the species-level taxonomy of the D. traun-steineri group as a whole (Bateman, 2011a), this is theobvious principle from which to begin examining D. traunsteinerioides in order to determine whetherinfraspecific subdivision is merited. The species hasfour centres of distribution in the British Isles:Ireland, North Wales, northeast England and westernScotland (Carey & Dines, 2002). As noted byBateman & Denholm (1983), at least some popula-tions of D. traunsteinerioides have been given sepa-rate epithets in three of these four areas; only thoseof North Wales have remained steadfastly classifiedunder D. traunsteineri(oides) alone, from Lacey(1955) onward. Populations across its full rangeshare an affinity for neutral and alkaline flushes(often tufaceous and surrounded by more acidic soils)that are dominated by Schoenus nigricans .Populations in these four areas show remarkableconsistency in all molecular datasets (Pillon et al. ,2007; Hedre´n et al ., 2011a; Bateman et al ., unpub-lished). Plastid haplotypes suggest modest introgres-sion with other dactylorchids in the Angleseypopulations, but the only substantial genetic devia-tion detected affected the Loch na Bron population,near Mullagh Mhor, Co. Clare, western Ireland,where both plastid and nuclear microsatellites indi-cate possible introgression with D. kerryensis/occi-dentalis (cf. Hedre´n et al ., 2011a). Thus far, we havedetected no genetic markers within D. traunsteiner-ioides to correspond with the clear north- andwestward decreases in the average sizes of all organsand, nearing the coast, increase in the proportion of plants exhibiting comparatively dense spotting on theleaves (Bateman, 2006a, 2011a).Although the type locality for traunsteineroides s.s. resides in Co. Wicklow, east-central Ireland (Pugsley,1936, 1940; see also Vermeulen, 1947), at least someIrish populations were distinguished from D. traun-steinerioides as D. traunsteineri ssp. hibernica , withsparing justification, by Landwehr (1975). ‘Sub-species hibernica ’ has as its type locality ScraughBog in Co. Westmeath but also encompasses ourstudy site at Pollardstown Fen, Co. Kildare (cf.Landwehr, 1977; Bateman & Denholm, 1983). We areconfident that Ireland does not support bona fide D.traunsteineri in addition to D. traunsteinerioides. Yorkshire plants found near Rievaulx Abbey weregiven the epithet eborensis by Godfery (1933) andtransferred to a variety of traunsteinerioides byPugsley (1939); see also Roberts & Gilbert (1963)and Tennant (1979). The probable type populationremains virtually unchanged today (Bateman, 2006b),but unfortunately, the only plant from the localityanalysed by us thus far is suspected to be a hybridwith D. fuchsii . Narrow-leaved marsh-orchid popula-tions from several nearby localities fit well within D.traunsteinerioides (Bateman, 2011a).Equally petite plants from Wester Ross, Scotlandwere first described as a full species, ‘Orchis’ francis-drucei , by Wilmott (1936: Fig. 1), but the epithet wassubsequently largely ignored until it was downgradedto a subspecies of D. traunsteineri by Soo´ (1962).Although the probable type locality of francis-drucei at Loch Maree was recently refound, most of thescientific work conducted on this taxon has utilised asimilar population located 8 km to the north, at LochKernsary (Bateman, 2011a). Admittedly, these extantpopulations contain a majority of leaf-marked plants,whereas Wilmott’s (1936) srcinal description, photo-graph, and line drawing of francis-drucei specifiedunmarked leaves, as well as an unusually palebackground colour to the labellum. Ssp. francis-drucei ( < ‘lapponica’) Perhaps the greatest taxonomic controversy that post-dated Bateman & Denholm (1983) was the assignmentto D. lapponica by Kenneth et al. (1988) of at least nineallotetraploid dactylorchid populations known fromthe western seaboard of Scotland (though not at thattime known from the headquarters of francis-drucei in Wester Ross). Dactylorhiza lapponica (Laest. exHartman) Soo´ is a boreal taxon that has its head-quarters in Scandinavia – the source of the holotype – but has also been reported from the Alps (e.g.Reinhard, 1985; Nordstro¨m & Hedre´n, 2008).Kenneth etal. (1988)aggregateddatafor20 charactersfrom seven Scottish populations in order to demon-strate their similarity to aggregated data for 31 bonafide D. lapponica populations previously measured inboth Scandinavia and the Alps by Reinhard (1985).The archetypal Scottish population assigned to D.lapponica , first found in 1963, is located at Knapdale,Kintyre (Kenneth et al ., 1988). Although it has sincebeen analysed for morphometrics and allozymes(McLeod, 1995; Bateman et al ., unpublished), earlierstudies of the population were of necessity confined totraditional morphological comparisons. The popula-tionwasfirst ascribedto D. traunsteineri (Cunningham& Kenneth, 1979), then to D. kerryensis/occidentalis (Tennant & Kenneth, 1983), then back to D. traun-steineri (Lowe et al ., 1986) before taking up longer-term residence in D. lapponica (R.H. Roberts, pers.comm., 1987; Kenneth et al ., 1988; Allan et al ., 1993;Stace, 1997; Carey & Dines, 2002). These rapidtaxonomic shifts graphically reflect the subtleties of the morphological distinctions among taxa and com-plexity of the related nomenclature. Bateman and Denholm Tetraploid marsh-orchid systematics New Journal of Botany 2012 VOL . 2 NO . 1 41
