Selbyana 26(1,2): 347-353. 2005. SPECIES DELIMITATION IN VANDA SECT. CRISTATAE LINDL. (ORCHIDACEAE): A MORPHOMETRIC ApPROACH LAUREN M. GARDlNER* University of East Anglia, Norwich, Norfolk NR4 7TJ, UK. Royal Botanic Gardens, Kew, Richmond, Surrey TW9 3DS, UK. Email: [email protected] JULIE A. HAWKINS Department of Plant Sciences, University of Reading, Reading, Berkshire RG6 6AS, UK. DAVID L. ROBERTS Royal Botanic Gardens, Kew, Richmond, Surrey TW9 3DS, UK. ABSTRACT. Species are arguably artificial units used for a variet~ o~ purpose~, ~~cluding class.ification, identification, and conservation, and authors have used different cntena for deluIlltmg new specIes. With more than 25,000 species already named and described in the Orchidaceae (often based on single or very few characters) and limited time and funding for conservation, priorities must be set to target resources. Ideally, observable morphological differences should distinguish genetically meaningful species. A mor phometric study of five species in Yanda section Cristatae (syn. Trudelia Garay) suggests three morpho metrically distinct species groups. Key words: Orchidaceae Yanda, Trudelia, species delimitation, morphometrics INTRODUCTION V. griffithii, as separate species, a single species, or a 'complex'. A comparison between morpho The species limits of the small group of or metric information on vegetative and floral char chids in John Lindley's Yanda section Cristatae acters in the group was made. (Lindley 1833), found in the Himalayas and Southeast Asia, are unclear. The species making Species Concepts up Yanda section Cristatae commonly are rec ognized as Yanda alpina Lindl. (Lindley 1853), With limited time and funding for conserva V. chlorosantha (Garay) Christenson (Christen tion and more than 25,000 orchid species de son 1992), V. cristata Lindl. (Lindley 1833), V. scribed, priorities are necessary to target re griffithii Lindl. (Lindley & Paxton 1851), and V. sources. The conservation of the widest possible pumila Hook.f. (Hooker 1890). Yanda griffith ii, diversity of taxa deserves to be the main target, however, sometimes is considered a synonym of but taxonomists must be able to provide quan V. alpina (M. Motes pers. comm., Roguenant & titative assessments of this diversity, if they are Chiron 2001, Senghas 1988). Another member to provide policymakers with meaningful 'units' of the section, Vandajainii (A.S. Chauhan), was for conservation, be they species, populations, described by Chauhan (1984). Very little mate or whole ecosystems. Approaches based on a rial of this species is available, certainly not in combination of morphological and molecular western European herbaria. Yanda jainii, de characters, incorporating an assessment of ge scribed as closely allied to V. cristata, was not netic diversity, need to be used in delimiting mentioned by L.A. Garay (1986), despite being new species and reassessing existing classifica described prior to his treatment of the group. It tions (Schuiteman & de Vogel 2003). The utility may be that Yanda jainii is a synonym of V. of the unit of 'species' would be increased, and cristata. Both V. jainii and also V. chlorosantha the selection and conservation of the largest are believed to be extremely rare, with only a gene pool facilitated. few herbarium collections of either in existence. The species concept(s) used by authors when This study investigates the confused species describing new species are rarely mentioned or limits of the section, with particular attention to discussed (Pridgeon 2003). The species concepts the status of Yanda alpina, V. chlorosantha, and and criteria used by a taxonomist to distinguish separate species significantly impact the number * Corresponding author. of species recognized. Although some taxono- 347 348 SELBYANA Volume 26(1,2) 2005 mists prefer to 'split' a group into different spe within Vanda section Cristatae, we analyzed cies on the basis of one or a few character dif both vegetative and floral characters using mul ferences, others prefer to 'lump' on the basis of tivariate statistics. Numerical taxonomy, includ similarities. For example, the Biological Species ing morphometric techniques, traditionally has Concept (Mayr 1942, Donoghue 1985) considers been used at or below species level; thus it is entities that can andlor do interbreed (by 'ordi suited to species delimitation questions. Multi nary means') to produce fertile progeny and that variate analysis can investigate underlying pat are reproductively isolated from other such terns and trends that may not be easily detect groups, as belonging to the same species. Such able to the taxonomist. Such analysis summariz concepts generally are not used in the Orchida es a range of variation in a relatively easily vi ceae because of the ability of phylogenetic ally sualized and practical manner. distant species to hybridize to produce fertile Classifications made using phenetic theory offspring. group entities are based on phenetic similarity, A phylogenetic species concept, with all and the methods adhere to a series of neo-Adan monophyletic, geographically distinct lineages sonian principJes, as described by Sneath and regarded as distinct species (Brasier 1997, Don Sokal (1973). The more characters included, the oghue 1985, Luckow 1995), cannot be applied more rigorous is the classification produced; and to Vanda and its related genera, as a full molec the greater the reduction in subjectivity, the ular analysis of the group has yet to be under greater is the increase in reproducibility of re taken. sults. Characters, scored for individual opera More 'practical' species concepts applicable tional taxonomic units (OTUs), are equally to groups not yet subject to molecular study in weighted and non-directional (i.e., no inference clude morphological species concepts, phenetic as to the probable direction of character evolu species concepts, and taxonomic species con tion). cepts; as well as more specifically, the Autapo Morphometric methods have not been used morphic Species Concept (Mishler & Brandon extensively in the Orchidaceae, but a study by 1987) and the Phylogenetic Species Concept Bateman and Denholm (1983) used the multi sensu Nelson and Platnick (1981). variate technique to classify British and Irish Taxonomic and morphological species con marsh-orchids and recommended that the group cepts (Mayden 1997, Cronquist 1978) are in be reduced to a single species with four subspe tended to be practical, morphological concepts cies. A study by Clifford and Lavarack (1974) useable with all groups of organisms. Propo compared the use of vegetative and reproductive nents of these concepts theorize that if an indi attributes of 93 Orchidaceae taxa by means of vidual or a population of individuals can be re peatedly and consistently distinguished from group-average clustering and found both types others by a taxonomist using 'ordinary means', of character "equally efficient predictors of the then it can be recognized as a distinct species. classification. " The diagnosis of a species may be based on only The most common multivariate statistical a few characters or even on a single character, methods used in taxonomic processes are Clus such as flower color. Phenetic species concepts ter Analysis and the ordination techniques, recognize species on the basis of overall simi which include the Principal Components Anal larity uniting taxa, in conjunction with their dis ysis (PCA) and its equivalent for handling con continuity from related taxa (Gornall 1997). tinuous and discontinuous data-the Principal Such concepts may involve only one or a few Coordinates Analysis (PCO or PCoA). The or key diagnostic characters or a range of attributes dination techniques represent all data collected may be taken into account. Concepts based on in multidimensional space and summarize it in numerical taxonomy, such as the Phenetic Spe 2- or 3-dimensional graphical plots. cies Concept sensu Sneath and Sokal (1973), use a range of attributes, attempting to take all avail MATERIALS AND METHODS able characters into consideration to detect un derlying patterns of similarity and discontinuity. A total of 30 spirit collections of flowers from Proponents of the concept find the species level Vanda section Cristatae from the University of is that at which distinct phenetic clusters can be Copenhagen (C), Royal Botanic Gardens Kew observed. The Phenetic Species Concept is used (K), and Royal Botanic Gardens Edinburgh (E) in this study to examine species delimitation were scored for 39 floral characters (see TABLE within Vanda section Cristatae. 1). Flowers stored in preservative spirits unfor tunately lose color characters, which therefore Morphometric Analysis could not be scored. In all, 118 herbarium sheet Employing a morphometric approach to de specimens from Natural History Museum (BM), termine the species limits and number of species K, E, National Herbarium Nederland Leiden (L), SECOND IOCC PROCEEDINGS 349 TABLE 1. Floral character list for a morphometric study of five species in Yanda section Cristatae. No. Floral diagnostic character No. Floral diagnostic character Maximum breadth of flower (tip of tepal to 20 Labellum epichile length (straight line along tip of tepal) underside of labellum, from where epichile 2 Maximum height of flower (tip of dorsal se joins labellum midsection to end of label pal to tip of lowest lateral sepal) lum, excluding length of any apical projec 3 Maximum distance from top of corolla tube tions) to furthest tip of tepals 21 Labellum hypochile midpoint width 4 Positioning of tepals around labellum and col- 22 Labellum epichile midpoint width umn: open (0), cupped (1), strongly cupped 23 Labellum lateral lobes on hypochile: small (2) (0), large (I) 5 Dorsal sepal-length 24 Labellum apical lip thickness 6 Dorsal sepal-midpoint width (perpendicular 25 Labellum callus beneath epichile apex: nonel to sepal midline) indistinct (0), distinct (l) 7 Dorsal sepal-apical width (measured 1 mm 26 Labellum callus length from tip of sepal along midline) 27 Labellum callus width 8 Dorsal sepal-basal width (measured I mm 28 Type of labellum apical projections: none (0), along midline of sepal from point of at small bumps (l), large bumps (2), distinct tachment) hornsllobes (3) 9 Lateral petal-length 29 Number of labellum apical projections 10 Lateral petal-midpoint width 30 Maximum length of labellum apical projec 11 Lateral petal-apical width tions 12 Lateral petal-basal width 31 Ribbing on labellum upper surface shallow 13 Lateral sepal-length (0), intermediate (1), deep (2) 14 Lateral sepal-midpoint width 32 Presence and type of spur on labellurn hypo 15 Lateral sepal-apical width chile: none/indistinct bowl-shape (0), pro 16 Lateral sepal-basal width nounced deep hollow (1) 17 Total labellum length (straight line measured 33 Depth of labellum spur (from base of sinuses from point of attachment to furthest tip of of hypochilar lateral lobes to spur apex) labellum) 34 Midpoint width of labellum spur (side view) 18 Labellum hypochile length (straight line along 35 Column length underside of labellum, from point of attach- 36 Column midpoint width ment to rest of floral organs, to apex of 37 Angle between midline of dorsal sepal to that 'spur') of right- hand lateral petal 19 Labellum midsection length (straight line 38 Angle between dorsal sepal and lateral sepal along underside of labellum, from apex of 39 Angle between line straight through column 'spur' to angled point where epichile starts) and central line of labellum and Universitiit Wien (WU) were scored for 25 responding to all specimens previously identi vegetative characters (see TABLE 2). Each indi fied as Yanda cristata, all V. pumila specimens, vidual plant (1 OTU) was assigned an identifi and a group of the V. alpina, V. griffithii, and V. cation code, and measurements were made using chlorosantha specimens (see FIGURE 1). This a metal ruler with 0.5 mm divisions, cotton pattern was repeated when labellum characters thread for measuring curved structures, a grid of were excluded from analysis and when tepal 1 X 1 mm squares, and a protractor for measur characters alone were analyzed. Cluster Analysis ing angles. of the floral data shows the OTUs split accord The floral and vegetative data sets were ana ing to the same pattern, with V. cristata and V. lyzed with PCO (Principal Coordinates Analy pumila specimens dividing into two distinct sis) in the statistical program MVSP (Kovach clusters, similar more to each other than to the Computing Services 2002a). Analyses employed third cluster, composed of V. alpina, V. griffith ii, Gower's General Similarity Coefficient (Kovach and V. chlorosantha with no further intra-cluster Computing Services 2002b) and 10glO trans groupings (see FIGURE 2). Analysis of the veg formed the data. Cluster analyses were per etative morphometric data showed no groupings formed on the floral data, using UPGMA as the at all (see FIGURE 3). clustering method, Gower's General Similarity Coefficient, and 10glO transformation of the data. DISCUSSION RESULTS This study shows that the five currently rec Analyses of the floral morphometric data con ognized species of this section are not distin sistently showed three floral morphotypes, cor- guishable on the basis of vegetative morpholo- 350 SELBYANA Volume 26(1,2) 2005 TABLE 2. Vegetative character list for a morphometric study of five species in Yanda section Cristatae. No. Vegetable diagnostic character No. Vegetative diagnostic character Height of plant measured from base of stem 15 Leaf lobe height difference (not incorporating roots) up to furthest 16 Average leaf length divided by average lobe point (usually tip of a leaf) height 2 Average root width measured 2 cm from 17 Leaves: lobes acutely ended (0), roundly end emergence from stem ed (1), blunt (2), mixed (3), acute and blunt 3 Stem length (4), rounded and blunt (5) 4 Stem width below first leaf 18 Leaves mucrone: absent (0), present narrowed 5 Stem midpoint width (l), present expanded (2) 6 Stem width below 4 youngest leaves 19 Leaves: number attached to plant 7 Average leaf length 20 Number of leaves excised from plant 8 Average midpoint leaf width 21 Number of excised leaves divided by number 9 Leaves: average apical leaf width measured 5 attached leaves mm back from apex of shortest leaf lobe 22 Total number of leaves 10 Length of longest leaf 23 Height of leaf sheath bases 11 Midpoint width of longest leaf 24 Height of stem (sheath bases without leaves) 12 Midpoint width of widest leaf 25 Total stem height divided by height of stem 13 Leaf length divided by midpoint width (sheath bases without leaves) 14 Leaf tapering: midpoint width divided by api cal width gy; however, three distinct floral morphotypes, U sing the Phenetic Species Concept sensu based on multiple characters, are Yanda pumila, Sneath and Sokal (1973), three species can be V. cristata, and the V. alpina groups. recognized. They are Yanda a/pina, V. cristata, The floral study broadly supports the currently and V. pumila, distinguishable on their floral accepted classification of section Cristatae, with morphologies. Intraspecific levels of variation a few changes at the ranking level of the taxa. occur within the species, corresponding to the i pumila 0.2 \' cristata . . N O.1 • . ~-~----------------------+----------------------------- : • alpina • I : I ii • chlorosantha ii i i -0.4 • griffithi i -0.5+---,--...,.----,--.----'-----,---,----,-----,----, -0.5 -0.4 ..Q.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 Axis 1 FIGURE 1. Analysis of floral characters: PCO axes 1 and 2 (Gower's General Similarity Coefficient used, data 10glO transformed). SECOND IOCC PROCEEDINGS 351 tially could remain at the species level, as they are all distinguishable by a few unique charac ters. Vanda jainii may be recognized as a subspe cies of V. cristata, which shows an extremely high level of floral morphological variation, from the number and length of the apical pro jections on the labellum, to the color and num ber of striations on the labellum upper surface, even on the same plant (M. Motes pers. comm.). Vanda jainii, only collected once from the wild, may be part of the natural phenotypic variation of the species. Where geographic distributions of species are very similar, where their distributions overlap, and where they share pollinators, we may not be looking at genetically distinct species. Instead we could be observing a complex of hybridizing entities with a shared gene pool. The taxa within the Vanda alpina complex may be interbreeding or, alternatively, may be gradually diverging. If Vanda section Cristatae has speciated rel atively recently to give the currently recognized species, and if few significant environmental dif ferences exist between habitats, then the vege tative parts of the plants may not have diversi fied sufficiently to be discriminating at the spe cies level by morphometric techniques. If recent speciation has been the result of isolation, per haps because of a change in pollinator/pollinator behavior, radiation in floral structures would pre cede radiation in the vegetative parts. The analyses presented here may suggest that Vanda alpina, V. griffith ii, and V. chlorosantha are indistinguishable and therefore closely relat ed, and it is not a priority to conserve all three. The three currently recognized species, however, are distinguished one from another by a few au tapomorphies each-mostly characters related to the labellum and flower color. Since these are only a small number of characters out of the large number measured in this study, especially since color characters were not included because of the nature of the material studied, these dif ferences among the taxa are not seen in the anal yses performed. The species could be genetical ly closely related and in the process of diverging into three separate lineages, in which case, they FIGURE 2. Analysis of floral characters: clustering may constitute relatively low priority for con analysis (UPGMA clustering method, Gower's General servation. Alternatively, they could be distantly Similarity Coefficient used, data loglO transformed). related, but morphologically similar either be cause of convergence or retention of the ances currently recognized species V. griffithii and V. tral morphology of the group, in either case, chlorosantha, as subspecies of V. alpina divisi constituting a higher priority for conservation. ble on the basis of one or a few floral characters Phylogenetic study with wide sampling and sur (mostly related to flower color). If either the tax veys of distribution and pollinators-whether onomic or autapomorphic species concept was the species are sympatric or allopatric and applied to the section, then all the taxa poten- whether gene flow is likely to occur between 352 SELBYANA Volume 26(1,2) 2005 0.35 " 'I 'I( 'I II! 0.28 " cristata , ': v 'I( 1\ 0.21 , Ii) , If cristata? A ~ " ~ ~ II @ alpina ~---~------ A pumila indet -0.14 + griffithii 'In: -0.21-+----,----,----,--L-~--__r--___,__--__,__--__,__--, -0.36 -0.27 -0.18 -0.09 0.00 0.Q9 0.18 0.27 0.36 0.45 Axis 1 FIGURE 3. Analysis of vegetative characters: pca axes 1 and 2 (Gower's General Similarity Coefficient used, data log,o transformed). them-would provide greater insight into the re IDENTIFICATION KEY OF VANDA lationships and dynamics of the group. SECTION CRISTATAE For Yanda section Cristatae, a morphological L Labellum base open saccate hollow. Labellum species concept based on floral characters ap equal in length or shorter than tepals. Small flowers, pears to be an appropriate concept to distinguish less than 2.5 cm diameter. . ............. . species, with floral autapomorphies used to rec ................... Vanda alpina complex ognize intraspecific levels of variation. If such 2. Strongly cupped flowers (tepals held closely autapomorphic individuals consistently and per around Jabellum). ... . . . . . . .. "V. alpina" sistently are identified and seen to form isolated 2. More open flowers (tepals cupped slightly for ward but not held closely around labellum). populations with different distributions from the 3. Flower pale yellowish green tepals, all same rest of the species, then grounds for the eleva color. Labellum unstriped ........... . tion of such taxa to specific level may be justi .................. "V. chlorosantha" fied. Yanda chlorosantha, for example, may 3. Flower bright greenish. Labellum with purple warrant remaining as a separate species from V. striations. ............. "v. griffithii" alpina, if it meets such criteria. In combination 1. Labellum base with distinct spur. Labellum longer with genetic and biogeographic distribution than tepals. Large flowers, greater than 2.5 cm di ameter. studies, a revision of the section will be pro 4. Strongly cupped flowers (tepals held duced. closely around labellum). Thick, fleshy la The whole genus Yanda needs considerable bellum without distinct callus below apex work in order to produce a phylogenetic frame of labellum, slight apical swellings of work around which to base a more meaningful varying size on edge of labellum. White classification of the group. flowers with reddish striations on label- A simple identification key is presented, dis lum. ........ Vanda pumila complex 4. More open flowers (tepals cupped slightly tinguishing the three floral morphotypes, Yanda forward but not held closely around la alpina, V. cristata, and V. pumila, and the sub bellum). Distinct callus below apex of la morphotypes corresponding to the currently rec bellum and distinct apical projections of ognized species epithets, 'alpina', 'cristata', varying size on edge of labellum. ..... 'griffithii', 'pumila', and also 'jainii'. . . . . . . . . . . .. Vanda cristata complex SECOND IOCC PROCEEDINGS 353 5. Apical projections very distinct hom Garay, L.A. 1986. Trudelia, a new name for Vanda shaped lobes, commonly 2-3, often di alpina. Orchid Digest 50(2): 73-77. varicating. Labellum longer than lateral Gornall, RJ. 1997. Practical aspects of the species sepals, 5-6 carunculate ridges on upper concept in plants. Pp. 171-190 in M.P. Claridge, surface of labellum. Distinct, deep H.A. Dawah and M.R Wilson, eds. Species: The spur. .............. "V. cristata" Units of Biodiversity. The Systematics Associa 5. Two erect lobules at apex of labellum. tion Special Volume Series 54, Chapman & Hall, Labellum shorter than lateral sepals. 8- London. 9 shallow purple ridges on upper sur Hooker, J.D. 1890. The Flora of British India 5(17): face of labellum. Spur obtuse and shal 49-54. lower. . . . . . . . . . . . . .. "v. jainii" Kovach Computing Services. 2000a. MVSP: Multi Variate Statistical Package. KCS, Anglesey, Wales. ACKNOWLEDGMENTS ---. 2000b. Gower's General Similarity Coeffi cient, KCS, Anglesey, Wales. We thank Dave Simpson at Royal Botanic Lindley, J. 1833. Genera and Species of Orchidaceous Gardens Kew for help with multivariate analy Plants. pp. 215-217. sis, also Martin Motes, of Motes Orchids, Flor ---. 1853. Folia Orchidacea 4(10): 1-11. J. Mat ida, and Phillip Cribb at RBG Kew for advice thews, London. on this study. Claire Moyle at RBG Kew assist Lindley, J. and J. Paxton. 1851. Paxton's Flower Gar ed with loans and digitized type images; and den 2: 22. Cassell, Petter, Galpin & Co., London. herbaria staff at BM, C, E, K, L, P, and WU Luckow, M. 1995. Species concepts: assumptions, loaned preserved plant material. We thank Brent methods and applications. Syst. Bot. 20(4): 589- Emerson at University of East Anglia (UBA) for 605. Mayden, RL. 1997. A hierarchy of species concepts: advice with the writing of this article and Sue the denouement in the saga of the species prob Mott and Sue Rumsey at Reading University for lem. Pp. 381-424 in M.P. Claridge, H.A. Dawah assistance with plant material loans and equip and M.R Wilson, eds. Species: The Units of Bio ment. diversity. The Systematics Association Special Volume Series 54, Chapman & Hall, London. Mayr, E. 1942. Systematics and the Origin of Species. LITERATURE CITED Columbia University Press, New York. Bateman, RM. and I.A. Denholm. 1983. A reappraisal Mishler, B.D. and R. Brandon. 1985. Individualism, of the British and Irish dactylorchids, 1. The tet pluralism, and the phylogenetic species concept. raploid marsh-orchids. Watsonia 14(4): 347-376. BioI. & Phil. 2: 397-414. Brasier, C.M. 1997. Fungal species in practice: iden Nelson, G. and N.I. P1atnick. 1981. Systematics and tifying species units in fungi. pp. 135-170 in M.P. Biogeography: Cladistics and Vicariance. Colum Claridge, H.A. Dawah and M.R Wilson, eds. Spe bia University Press, New York. cies: The Units of Biodiversity. The Systematics Pridgeon, A.M. 2003. Modern species concepts and Association Special Volume Series 54, Chapman practical consideration for conservation of Orchi & Hall, London. daceae. pp. 43-53 in K.W Dixon, S.P. Kell, RL. Chauhan, A.S. 1984. Vanda jainii Chauhan-a new Barrett and P.J. Cribb, eds. Orchid Conservation. orchid species from Meghalaya. J. Econ. Bot. Natural History Publications, Kota Kinabalu, Sa 5(4): 977-979. bah. Christenson, E.A. 1992. Notes on Asiatic orchids. Roguenant, C. and G. Chiron. 2001. Les vandas-de Lindleyana 7(2): 88-94. courverte d'un group d'orchidees spectacu1aires. Clifford, H.T. and P.S. Lavarack. 1974. Role of vege Tropicalia, Saint-Genis-Lava1, France. tative and reproductive attributes in classification Schuiteman, A. and E. de Vogel. 2003. Taxonomy for of Orchidaceae. BioI. J. Linnean Soc. 6(2): 97- conservation. Pp. 55-68 in K.W. Dixon, S.P. Kell, 110. RL. Barrett and P.J. Cribb, eds. Orchid Conser Cronquist, A. 1978. Once again, what is a species? Pp. vation. Natural History Publications, Kota Kina 3-20 in L.Y. Knutson, ed. Biosystematics and Ag balu, Sabah. riculture. Allenheld Osmun, Montclair, New Jer Senghas, K. 1988. Trudelia Garay. Orchideen (Ham sey. burg) 1(19,20): 1211. Donaghue, M.J. 1985. A critique of the biological spe Sneath, P.H.A. and R.R Sokal. 1973. Numerical Tax cies concept and recommendations for a phylo onomy. W. H. Freeman and Company, San Fran genetic alternative. Bryologist 88(3): 172-181. cisco, Calif.