CSIRO PUBLISHING www.publish.csiro.au/journals/asb Australian Systematic Botany, 22, 1–15 A molecular phylogenetic analysis of Diuris (Orchidaceae) based on AFLP and ITS reveals three major clades and a basal species James O. Indsto A,C,D, Peter H. Weston A and Mark A. Clements B A National Herbarium of NSW, Mrs Macquaries Road, Sydney, NSW 2000, Australia. Centre for Plant Biodiversity Research, National Botanic Gardens, Canberra, ACT 2601, Australia. C Institute for Conservation Biology, University of Wollongong, NSW 2522, Australia. Present address: Forensic Science Services Branch, Forensic Services Group, NSW Police Force. D Corresponding author. Email: inds1jam@police.nsw.gov.au B Abstract. Diuris is a terrestrial orchid genus of at least 61 and possibly more than 100 species, restricted to Australia except for one species endemic in Timor. Distinctive species groups have respective eastern and western centres of distribution. Although species affinities have been vaguely understood for many years, no formal infrageneric treatment has been undertaken as Diuris possesses few reliable morphological characters for a classification system. We have undertaken cladistic parsimony and Bayesian phylogenetic analyses of Diuris by using the ITS1–5.8S–ITS2 region of nuclear rDNA and morphological characters, with a subset of samples also studied by amplified fragment length polymorphism (AFLP) as an independent test of phylogenetic relationships. Four major clades with strong bootstrap support were resolved and are named here according to a recently published classification; D. sulphurea forms a lineage (subg. Paradiuris) of its own that is well supported as the sister to the rest of Diuris. Two other major eastern clades contained species related to D. maculata (subg. Xanthodiuris) and D. punctata (subg. Diuris), respectively. Although these latter two subgenera are genetically well resolved, there is minimal genetic variation at species level, consistent with recent, rapid speciation. A fourth clade (subg. Hesperodiuris) has a centre of distribution in Western Australia, and has more genetic and morphological variation than the eastern subgenera. Total evidence analysis provides support for the western clade being sister group to the two eastern subgenera Diuris and Xanthodiuris; however, this relationship was not resolved by molecular data. Hybridisation is known to be common among species within subgenera Diuris and Xanthodiuris. Instances of incongruence between different datasets were found suggestive of hybridisation events between species of different sections of Diuris. Introduction According to the compilation of Clements (http://anbg.gov.au/ cpbr/cd-keys/orchidkey/html/currentspecies.html), the genus Diuris (Orchidaceae) comprises at least 61 species (see Appendix 1, correct as at December 2008) that are restricted to Australia, with the exception of D. fryana, which is endemic to Timor. The circumscription of Diuris species is still an active process and numerous new taxa can be expected to be recognised in the near future. Species of Diuris are well represented in the southern parts of Western Australia and eastern Australia, separated by the Nullarbor Plain, with a few species found in tropical Queensland (Jones 2006). The eastern and western species mostly fall into morphologically distinct groups, suggestive of distinct phylogenetic lineages. However, a few species that are widely distributed in south-eastern Australia closely resemble western species (e.g. D. orientis cf. D. corymbosa). Although some eastern species, such as D. sulphurea, or representatives of eastern species groups, extend into South Australia, the centre of diversity of these species groups is eastern New South Wales and, to a lesser extent, Victoria and Tasmania. Many populations and species show
CSIRO 11 March 2009 considerable variability, making Diuris taxonomically challenging at species level. Several new species have been recognised in recent times and more are likely to be named in the near future. Dressler (1990) considered, on the basis of an informal phylogenetic analysis of morphological characters, that Diuris is the sister group of Orthoceras, and on this basis he grouped them together as the subtribe Diuridinae in the tribe Diurideae. This taxonomic decision was corroborated by subsequent phylogenetic analyses of both chloroplast and nuclear DNA sequences (Kores et al. 2001; Clements et al. 2002), although the circumscription of the tribe Diurideae has changed in other respects in response to these molecular results. Most Diuris species show a general resemblance in floral form to Australian legume shrubs, particularly those colloquially known as ‘egg and bacon peas’. These plants belong to several genera in the tribes Mirbelieae and Bossiaeeae of the family Fabaceae, including Pultenaea, Daviesia, Dillwynia and Bossiaea. The habitats of these orchids are typically open eucalypt forest or woodland communities, which in eastern Australia are extensively distributed in a mosaic pattern that 10.1071/SB08029 1030-1887/09/010001 2 Australian Systematic Botany would be expected to result in genetic isolation of many populations (Keith 2004). Most of the literature on pollination of Diuris is anecdotal (van der Cingel 2001), apart from two studies of the nectarless Diuris maculata s.l. (Beardsell et al. 1986; Indsto et al. 2006). Divergence in floral features from ancestral pea-flower mimicry may have occurred in some species, e.g. D. alba (Indsto et al. 2007). These studies provide evidence for Batesian-type floral mimicry of egg and bacon peas, with pollination mostly by native bees that specialise in collecting nectar and pollen from pea flowers. As the flowers of the majority of Diuris species resemble pea flowers, similarities in pollination mode may be expected in most members of the genus. A cladistic analysis of Diuris species on the basis of morphology alone is unlikely to be either well resolved or well supported, since the genus provides few taxonomically useful qualitative morphological characters. Moreover, extensive homoplasy in the floral morphology of deceptively pollinated orchids sometimes poses problems for cladistic analyses that are predominantly based on floral characters (see e.g. (Pridgeon et al. 2001)). Significant progress in reconstructing the phylogeny of Diuris is most likely to be made by sampling a large number of non-morphological characters and by including these in a cladistic analysis of the genus. The most cost-effective source of such independent phylogenetic evidence is DNA, sampled either through the determination and alignment of homologous DNA sequences or through the identification of shared, homologous DNA fragments from diverse genetic loci. The internal transcribed spacer (ITS) regions of the nuclear rDNA genes (rDNA), ITS1 and ITS2, have proved one of the most informative regions of variable DNA for phylogenetic analysis at the level of species relationships within genera (e.g. Cox et al. 1997; Clements 2003; Orthia et al. 2005). The ITS regions are non-coding DNA that is transcribed to RNA, but spliced out during ribosome assembly. Being non-coding, they accumulate DNA mutations much more rapidly than the 5.8S rDNA gene. Ribosomal genes, although present in high copy number, are usually homogeneous in DNA sequence within individuals through the process of concerted evolution and so are effectively equivalent to the study of variation of a single gene locus (Weider et al. 2005). The nuclear genomic ITS1, 5.8S and ITS2 regions of rDNA are contiguous and, being ~700 bases in total length, can be readily amplified by polymerase chain reaction (PCR) as a single unit and studied together. Several studies that have utilised plastid DNA to provide independent support for analysis of genomic sequences, have often resulted in conflicting phylogenies (e.g. Okuyama et al. 2004). This has generally been attributable to a consequence of introgression. AFLP (Vos et al. 1995) is an alternative technique based on genome-wide DNA sequences containing variable restriction sites. The AFLP technique has the advantage in many cases of distinguishing individuals at intraspecific level (Krauss 2000). A considerable database of ITS sequences has been lodged with GenBank; however, no comparable resource is available for AFLP profiles and they have been used only to a limited degree in orchid phylogenetic studies (e.g. Hedren et al. 2001; Mant et al. 2005). AFLP can clearly resolve variable ploidy levels, such as allotetraploidy, in plant lineages. Hedren et al. (2001) noted that AFLP produces highly reproducible results, including relative band intensities. This feature allows AFLP profiles to be readily J. O. Indsto et al. recognised as patterns or ‘fingerprints’ (Indsto et al. 2005). We find patterns of unique bands and altered band intensities to be much easier to interpret than polymorphic DNA sequences. We have chosen to include AFLP data to test the findings of ITS sequence analysis because of its value as an independent dataset that can potentially reveal incongruence with data from a singlelocus technique. Recent publications by Jones (2000) and Jones and Clements (2006) have detailed an infrageneric taxonomy of Diuris. These taxon names have been used in the present study where supported by our analysis. While morphological characters have been carefully analysed in these publications, unpublished ITS sequence data have also guided the authors in these works. Their total evidence approach combined with research into nomenclatural precedents has resulted in a classification scheme which we test against combined AFLP and ITS data, combined with taxonomically informative morphological characters. The data presented here are largely congruent with the taxonomic scheme of Jones and Clements (2006). Materials and methods Plant samples Table 1 contains information on plant taxa used in the present study, including accession details and source localities. Sequences for Orthoceras strictum (AF348048) and Eriochilus cucullatus (AF348030) were downloaded from GenBank and included in the sample as outgroup taxa (Clements et al. 2002). An additional sequence for Diuris sulphurea (AF348026) was also included from GenBank. Samples for AFLP were collected in the field by JOI, PHW, John Riley (Research Associate, Centre for Plant Biodiversity Studies, CSIRO, ACT) and Michael Batley (Honorary Research Associate, Entomology Department, Australian Museum, Sydney, NSW). Many of these samples were also used for combined ITS1–5.8S–ITS2 (ITS) analysis and combined with additional samples studied by MAC at the Centre for Plant Biodiversity Studies, CSIRO, ACT. Morphological analysis For this analysis, 10 parsimony-informative morphological characters (see Table 2) were coded for all of the sampled species. The data matrix was subjected to cladistic analysis under the Fitch parsimony criterion, by using PAUP v. 4.0b for Microsoft Windows (Sinauer Associates, Inc., Publishers, MA, USA). The PAUP analysis consisted of a heuristic search with equally weighted characters and TBR (tree bisection and reconnection) branch swapping, starting from 1000 trees, each produced from a random addition sequence of taxa. The full morphological dataset was then subject to successive approximations character weighting (Farris 1969), with the rescaled consistency index as the measure of fit for new character weights. The morphological analysis was then repeated with the two outgroup taxa, Eriochilus cucullatus and Orthoceras strictum, deleted. This reduced dataset was then subject to successive approximations character weighting as above. Strict consensus trees were constructed to summarise the shared components of equally parsimonious trees from all analyses. The morphological dataset was also subject to Bayesian A molecular phylogenetic analysis of Diuris by AFLP and ITS Australian Systematic Botany Table 1. Collection details of species samples used in molecular analyses Species Source locality Collection number GenBank accession Diuris abbreviata Diuris aequalis Diuris sp. aff. alba 1 Diuris sp. aff. alba 2 Diuris alba Diuris arenaria Diuris aurea 1 Diuris aurea 2 Diuris aurea 3 Diuris behrii Diuris brevifolia Diuris byronensis Diuris carinata Diuris chryseopsis Diuris monticola Diuris concinna Diuris conspicillata Diuris sp. aff. corymbosa Diuris drummondii Diuris flavescens Diuris fragrantissima Diuris goonooensis 1 Diuris goonooensis 2 Diuris goonooensis 3 Diuris goonooensis 4 Diuris goonooensis 5 Diuris goonooensis 6 Diuris goonooensis 7 Diuris laxiflora Diuris maculata 1 Diuris maculata 2 Diuris maculata 3 Diuris magnifica Diuris sp. aff. ochroma Diuris palustris Diuris pardina Diuris picta Diuris platichila 1 Diuris platichila 2 Diuris sp. aff. porphyrochila (ms) Diuris praecox Diuris punctata 1 Diuris punctata 2 Diuris punctata 3 Diuris punctata 4 Diuris sp. aff. punctata 1 Diuris sp. aff. punctata 2 Diuris semilunulata 1 Diuris semilunulata 2 Diuris nigromontana Diuris setacea Diuris sulphurea 1 Diuris sulphurea 2 Diuris tricolor 1 Diuris tricolor 2 Diuris venosa Eriochilus cucullatus Orthoceras strictum Cathedral Rock National Park, NSW Kanangra Boyd National Park, NSW Clarence Town, NSW Munmorah, NSW Yeppoon, Qld Nelson Bay, NSW Cultivation ANBG Munmorah, NSW Castlereagh Nature Reserve, NSW Stuart Mills Cemetery Mt Lofty Range, SA Pattersons Hill, Byron Bay, NSW Penelup Farm, Mt Lindsay, WA Ilford, NSW Kanangra Boyd National Park, NSW Mt Gibson, WA Esperance, NSW 1 km S of Marriot, WA Albany, WA The Bight Cemetery, NSW Cultivation Ex Tottenham, Vic Conimbla National Park, NSW Near Parkes, NSW Manna Mountain, NSW Bungambil State Forest, NSW Reefton, NSW Round Hill Nature Reserve, NSW Sims Gap, NSW Gordon Crossing, Albany Hwy, WA Kentlyn, NSW Scheyville National Park, NSW Lake Parramatta, NSW Murdoch University, Perth, WA Kings Hwy, E of Braidwood, NSW Hartley, SA Scotts Creek, SA Frog Rock, WA Dunedoo, NSW Halfway between Keith and Pinnaroo, SA Wellesly North Rd, WA Bobs Farm, NSW Mt Ainsley, ACT Penrose State Forest, NSW Bargo, NSW Tallong Cemetery, NSW Spring Mountain Rd, off Gwyder Hwy, NSW Mellong Swamp, NSW NSW Southern Tablelands, Braidwood–Nowra Rd Near Joadja Creek, NSW Black Mountain, ACT E of Cranbourne, WA NSW Central Coast, Wattagan Mountains Mellong Swamp, NSW Nangerbone State Forest, NSW 7 km S of Cowra, NSW Barrington Tops, NSW Mary Seymour Conservation Park, SA NSW South Coast, near Lynne NSW477113 NSW441910 NSW720057 NSW432988 NSW720059 NSW720050 DW1082 NSW720114 NSW719904 CANB668539 CANB619900 CANB648382 CANB668554 NSW720067 NSW441907 Cult ANBG NSW720073 CANB625050 CANB8102383 NSW720060 CANB940531 CANB732510 NSW720064 NSW720063 NSW720071 NSW720077 NSW720071 NSW720068 CANB647853 NSW720115 NSW719860 NSW720118 CANB625020 CANB634123 CANB619672 NSW720075 CANB8806717 NSW731415 NSW731440 CANB625042 NSW720061 CANB732531 NSW720113 NSW720088 NSW720099 CANB656698 NSW720039 CANB732515 NSW441899 NSW731460 CANB668557 CANB668538 NSW495371 NSW720053 CANB9813821 CANB668489 CANB619876 Jones 13790 DQ904011 DQ915945 DQ904027 DQ904012 DQ904020 DQ904028 AF348022 DQ904013 AFLP only EU595633 EU595637 EU595641 EU595626 DQ904029 DQ904019 EU595630 DQ904014 EU595636 EU595634 DQ904022 EU604809 EU595638 DQ904015 DQ904030 AFLP only AFLP only AFLP only AFLP only EU595642 DQ915944 AFLP only AFLP only EU595627 EU595625 EU595631 DQ904023 EU595635 DQ904031 DQ904021 EU595639 DQ904016 EU595643 DQ904017 DQ904024 AFLP only EU595629 DQ904032 EU595628 DQ904025 DQ904033 EU595632 AF348026 DQ904018 DQ904026 EU595640 EU595644 AF348030 AF348048 3 4 Australian Systematic Botany J. O. Indsto et al. Table 2. Morphological characters used in total evidence analysis Morphological character Character states 1. Tuberoid shape 2. Tuberoid orientation 3. Leaf number 4. Colour of predominant anthoxanthin pigments in flower 5. Relative length of lateral sepals and petals 6. Lateral sepal shape 7. Lateral sepal orientation 8. Relative length of labellum lobes 9. Labellum callus 0, Ovoid to ellipsoid; 1, forked ovoid-obovoid; 2, elongate ovoid-obovoid; 3, linear-terete 0, Vertical; 1, horizontal 0, One; 1, two to three; 2, more than three 0, Yellow; 1, white 10. Indumentum of labellum callus ridges 0, Lateral sepals about as long as petals; 1, lateral sepals at least twice as long as petals 0, Linear; 1, linear-spathulate 0, Semi-erect to erect; 1, horizontal to semi-pendent 0, Lateral lobes much shorter than mid-lobe; 1, lateral lobes about as long as mid-lobe 0, With one ridge; 1, with two slightly raised parallel ridges; 2, with two conspicuously raised, divergent ridges 0, Glabrous; 1, densely papillate phylogenetic analysis by using the standard discrete model of MrBayes 3.1.2 – one based on the Mk model reviewed by Lewis (2001). MrBayes 3.1.2 spawns two MCMC runs, and we ran each for 4
106 generations, sampling every 100th generation. A burn-in of 106 generations was found to be enough to get the average standard deviation of split frequencies well below 0.01 (a threshold recommended by the authors of MrBayes) for the two runs. For each MCMC run, we constructed a majority-rule consensus of the trees sampled after the burnin period in PAUP* v4.0b10 (Swofford 2003). When the majority-rule consensus trees differed between the two runs in the posterior probability of a branch, only the lower of the values is presented. AFLP analysis The AFLP procedure of Vos et al. (1995) was used with modifications. Fresh whole flowers, or several centimetres of healthy leaf tissue (leaves being thin and grass-like) were dessicated in a Zip-Lock bag with silica gel for ~10 days at room temperature and then stored at 20C until required. Dried samples, weighing ~10 mg were added to 2-mL Eppendorf tubes with a few grains of acid-washed sand. Open tubes were placed in 15-mL cryovials containing liquid nitrogen to ~20-mm depth and the frozen tissue was ground with an autoclaved bamboo skewer. The Qiagen Plant DNeasy Mini Kit (QIAGEN, Doncaster, Vic., Australia) protocol was followed without modification and DNA eluted into 200 mL of AE buffer. The DNA yield was estimated by agarose gel electrophoresis. AFLP reagents, including restriction enzymes EcoRI and MseI (New England Biolabs Inc., Beverly, MA) and AFLP adapters and primers (Sigma-Genosys, http://www.sigmaaldrich.com/life-science/custom-oligos.html, verified January 2009), were used as described by Vos et al. (1995), except that the EcoRI selective primers were 50 -HEX labelled. A combined restriction digest and ligation was carried out. A quantity of 200–500 ng of DNA in 10 mL TE0.1 (TE0.1 = 10 mM Tris, pH 8.0; 0.1 mM EDTA, pH 8.0) was added to 10 mL of reaction master mix containing, for 20 mL final volume, 0.5 mM EcoRI adaptor and 5 mM MseI adaptor (prepared according to Wolf Laboratory protocol: http://bioweb. usu.edu/wolf/aflp_protocol.htm, verified January 2009), 1
T4 ligase buffer (New England Biolabs), 0.5 mg BSA, 50 mM NaCl, 2 U MseI, 5 U EcoRI and 20 U T4 DNA ligase (New England Biolabs). The mixture was incubated at 37C for 4 h. An aliquot of 10 mL was run on an agarose gel to check for complete digestion (all DNA evident as a visible smear), and the remainder was then diluted to 200 mL in TE0.1. An aliquot of 4 mL of diluted restriction/ligation mix was used as template for preselective PCR in 20-mL volumes containing 200 mM dNTPs, 20 ng each of EcoRI and MseI pre-selective primers, 0.5 mg mL1 BSA (Giambernardi et al. 1998), 50 mM KCl, 10 mM Tris at pH 8.5, 2.5 mM MgCl2, 2% formamide (Ranamukhaarachchi et al. 2000) and 1 U Taq DNA polymerase. A touchdown PCR protocol was employed with one cycle of 95C for 3 min, followed by successive cycles of 95C for 20 s, annealing for 30 s and 72C extension for 2 min, with the first annealing at 66C and then progressively reduced each cycle by 1C for the next 10 cycles, until an annealing temperature of 56C was reached, with 20 cycles carried out at this annealing temperature. This was followed by a final extension step of 72C for 10 min. An aliquot of 10 mL was run on an agarose gel to check for a visible smear, indicative of successful amplification and the remainder was diluted to 200 mL in TE0.1 to serve as template for selective PCR. A volume of 4 mL of diluted preselective PCR product was used as template for selective PCR. This was carried out with four combinations of 2-bp selective primers that were found to be most informative for Diuris: EcoRI-AC with MseI-CT, EcoRI-AA with MseI-CT, EcoRI-AA with MseI-CG and EcoRI-AC with MseI-CA. EcoRI selective primers were 50 -HEX fluorescently labelled. Reactions were carried out in 20-mL total volumes containing 200 mM dNTPs, 60 ng each of 50 -HEX EcoRI-XX and MseI-XX 2-base selective primer pairs, 50 mM KCl, 10 mM Tris at pH 8.5, 2.5 mM MgCl2, 0.5 mg mL1 BSA, 2% formamide and 1 U Taq DNA polymerase and by using the same PCR protocol as above, except that 25 cycles of PCR at 56C annealing were used. An equal volume of denaturing dye containing formamide with 10 mM EDTA at pH 8.0 and bromophenol blue was added and the samples were heatdenatured for 3 min at 95C and snap-chilled on ice. Aliquots A molecular phylogenetic analysis of Diuris by AFLP and ITS of 2–3 mL were loaded on a 5% 29 : 1 polyacrylamide gel containing 7.5 M urea and 0.6
TBE and run in 0.6
TBE at 40C and 900 V in a Corbett Gel-Scan 2000 DNA Analyser with He-Ne laser detection (Corbett Research, Mortlake, NSW, Australia). AFLP chromatograms were printed and compared manually. A D. maculata s.l. sample was used as the reference taxon run in each PCR and gel and a printout converted to a transparency. Bands of this species were numbered according to increasing size (gel retention time) on the transparency, which could be overlaid manually on other sample chromatograms. Taxa were then scored for loss of numbered D. maculata s.l. bands, or gain of bands, indicated by a letter, with position carefully marked in the transparency to 1-bp accuracy. Each of the four selective AFLP primer combinations listed above generated phylogenetically informative bands. AFLP peaks used in analysis ranged from ~90 to ~320 bp, and were each treated as separate characters, were scored for presence or absence and the combined character set was compiled with the software package Nexus Data Editor (NDE Version 5.0, http://taxonomy.zoology.gla.ac.uk/rod/NDE/nde.html, verified December 2008). The AFLP data matrix was subjected to cladistic analysis under the Fitch parsimony criterion, as for morphological data, and to Bayesian phylogenetic analysis, by using MrBayes 3.1.2. The PAUP analysis consisted of a heuristic search with equally weighted characters and TBR (tree bisection and reconnection) branch swapping, starting from 1000 trees, each produced from a random addition sequence of taxa. Bootstrap analysis with 2000 replicates was conducted with default heuristic search settings, generating a majority-rule consensus tree of nodes with >50% BP support. Two MrBayes analyses were conducted, one using the restriction site (binary) model with no absence sites, and the other with the standard discrete model. For the first analysis, two MCMC chains were each run for 6
106 generations, sampling every 100th generation with a burn-in of 1.5
106 generations, whereas for the second analysis, two MCMC chains were each run for 3
106 generations, sampling every 100th generation with a burn-in of 7.5
105 generations. Majority-rule consensus trees were constructed from the trees sampled after the burn-in period in each MCMC run, as in the morphological analysis. A total of 31 taxa, including 30 ingroup taxa, was studied for the total of 48 AFLP characters, of which 20 were parsimony informative. The AFLP profiles of the chosen outgroup taxa differed so much from Diuris that homologous bands could not be confidently identified, so these taxa could not be usefully included in the analysis. Consequently, unrooted trees were produced and re-rooted on the branch connecting D. sulphurea to the rest of the genus, in order to render the results readily comparable to those of our analysis of ITS sequences (see below). Combined ITS1–5.8S–ITS2 rDNA (ITS) plus indels analysis The contiguous ITS1–5.8S–ITS2 rDNA (referred to as ‘ITS’ hereafter) region was amplified as one unit, by using mainly the primer pair 17SE and 26SE (Sun et al. 1994; Gravendeel et al. 2001), although other combinations were also used (Clements et al. 2002). PCR reactions were performed with 20-ng sample Australian Systematic Botany 5 DNA in 50-mL volumes containing 10 mM Tris at pH 8.5, 50 mM KCl, 1.5 mM MgCl2, 200 mM dNTPs, 0.5 mg mL1 BSA (Giambernardi et al. 1998), 150 ng each primer and 3 U Taq DNA polymerase. The PCR protocol was one cycle of 95C for 3 min, followed by 25 cycles of 95C for 20 s, 58C for 30 s and 72C for 2 min and followed by a final extension of 72C for 10 min. PCR product yield was checked by agarose gel electrophoresis and purified for sequencing with the Qiagen QiaQuick kit (QIAGEN). PCR products of ~850-bp length were bidirectionally sequenced with Applied Biosystems Big DyeTerminator v3.0 chemistry and run on an Applied Biosystems ABI Prism 3100 Genetic Analyser. DNA sequence chromatograms were edited with Sequencher 3.0 software (Gene Codes Corporation, Ann Arbor, MI, USA) or the software package Finch-TV (Geospiza, Seattle, WA, USA). A text file of all taxa (50 in total, 48 ingroup taxa) in FASTA format was assembled and an alignment analysis performed with the software package Clustal X (Bio Pack 3.6, including also BioEdit and Treeview; Ibis Biosciences, Carlsbad, CA, USA), with default parameter settings. The aligned sequences were further processed to remove flanking DNA sequence using BioEdit, and then exported in Nexus format for analysis in PAUP v. 4.0, as detailed above for morphology and AFLP, except that the trees were rooted by using the designated outgroups, branch swapping started from 200 trees, each produced from a random addition sequence of taxa, maxtrees was set at 100, and only 1000 bootstrap replicates were completed. The analysis comprised 715 total characters, of which 122 were parsimony informative. In all, 11 indels were manually coded as additional characters by the simple indelcoding method of Simmons and Ochoterena (2000). The ITS dataset was also subject to Bayesian phylogenetic analysis with MrBayes 3.1.2. We used the Akaike information criterion (AIC) in MrModelTest 2.3 (available from J.A.A. Nylander’s website at http://www.abc.se/~nylander/ mrmodeltest2/mrmodeltest2.html, verified December 2008) to select an adequately parameter-rich model of nucleotide substitution for the ITS alignment, with indels coded as missing data. Indels were then added to the data matrix as an extra partition of binary characters (as for the parsimony analysis) and were analysed by using the standard discrete evolutionary model. We unlinked the sampling of state frequencies and substitution rates for the two data DNA partitions. The two MCMC runs were run for 3
106 generations, sampling every 100th generation. We found that a burn-in of 7.5
105 generations was sufficient to get the average standard deviation of split frequencies well below 0.01 for the two runs. Majority-rule consensus trees were constructed from the trees sampled after the burn-in period in each MCMC run, as in the morphological analysis. Total evidence analysis For this analysis, all morphological, AFLP and ITS characters were combined and analysed as a single dataset by using the parsimony criterion as described for the individual datasets above. The combined dataset was also subject to two Bayesian phylogenetic analyses with MrBayes 3.1.2, treating the ITS nucleotide sites, ITS indels, AFLP data and morphological Australian Systematic Botany J. O. Indsto et al. Results The parsimony analysis of morphological characters including all sampled species produced 402 equally parsimonious trees of length 19 steps. The strict consensus of these trees (not shown) included only one resolved component, a grouping of all five of the sampled species of Diuris subgenus Diuris section Pedunculatae (Appendix 1). The rest of Diuris and its outgroups formed a polytomy. Only 248 trees of length 13.61572 were produced by successive approximations character weighting and these yielded the same consensus tree as for the unweighted data. The two outgroup taxa, which were both scored as ‘unknown’ for several characters, were then deleted and the remaining taxa was reanalysed as before, to test whether the presence of unknown cells in the data matrix was responsible for the observed lack of phylogenetic resolution. This resulted in 6744 trees of length 18 steps for unweighted data and 3228 trees of length 14.81667 steps for weighted data. The strict consensuses of both sets of trees (not shown) were identical and included only one resolved component, the same as in the previous results produced with the full sample of species. The two majority-rule consensus trees from the Bayesian analysis had three clades resolved but two of these received posterior probabilities of less than 0.95. The clade corresponding to Diuris section Suffusae (Appendix 1) received a posterior probability of 0.95. The parsimony analysis of the AFLP dataset produced one tree of length 39 steps (see Fig. 1). This clearly resolved the subgenera Diuris and Xanthodiuris sensu Jones and Clements (2006), with Diuris sulphurea subg. Paradiuris 100/100 93/95 94/51 Diuris conspicillata Diuris chryseopsis Diuris abbreviata Diuris praecox Diuris goonooensis 2 Diuris goonooensis 4 Diuris goonooensis 3 Diuris goonooensis 6 62/- Diuris goonooensis 7 Diuris goonooensis 5 Diuris aequalis Diuris pardina Diuris maculata 1 Diuris maculata 2 Diuris maculata 3 Diuris semilunulata 2 Diuris nigromontana Diuris monticola sect. Pedunculatae sect. Suffusae subg. Hesperodiuris Diuris alba Diuris sp. aff. alba 1 Diuris sp. aff. alba 2 Diuris tricolor 1 Diuris arenaria sect. Purpureo-albae Diuris punctata 2 Diuris punctata 3 Diuris punctata 4 Diuris sp. aff. punctata Diuris flavescens sect. 66/58 Diuris aurea 2 Diuris Diuris aurea 3 subg. Diuris characters as four partitions and unlinking the sampling of state frequencies and substitution rates for each of the four data partitions. The two analyses differed in the model used for the AFLP partition; in the first we used the restriction-site (binary) model with no absence sites, whereas in the second the standard discrete model was used. The models used for the other three partitions were the same as those used in the separate analyses of these partitions. We ran the first MCMC analysis for 6
106 generations, and the second for 4
106 generations, sampling every 100th generation in both cases. In the analysis that used the restriction-site (binary) model with no absence sites, the average standard deviation of split frequencies had reduced only to 0.012 after 6
106 generations; so, strictly speaking, it had not yet emerged from its burn-in phase. We concluded that this mixed model was intractably complex and terminated the analysis. By contrast, in the analysis that used the standard discrete model for the AFLP partition, the average standard deviation of split frequencies reduced to below 0.01 after 6.87
105 generations; thus, discarding the first 106 generations as the burn-in was more than adequate. Majorityrule consensus trees were constructed from the trees sampled after the burn-in period in each of the two MCMC runs of the analysis, by using the standard discrete model for AFLP data, as in the morphological analysis. Morphological-character phylogenies were reconstructed by mapping characters parsimoniously on the total evidence trees, using the ‘trace character history’ option in Mesquite version 2.01 (build j28; http://mesquiteproject.org, verified December 2008). subg. Xanthodiuris 6 Fig. 1. The one tree of length 39 steps produced by parsimony analysis of AFLP data, showing branch lengths estimated under the ACCTRAN algorithm. Above each branch are its bootstrap support index (left of the slash) and Bayesian posterior probability (right of the slash), both expressed as percentages. A total of 31 terminals, including 30 ingroup terminals, was studied for 48 AFLP characters, of which 20 were parsimony informative. A molecular phylogenetic analysis of Diuris by AFLP and ITS strong bootstrap support of 100 and 94, respectively. According to the AFLP parsimony analysis these subgenera form a clade with BP = 93 and are more closely related to each other than to the subgenera Hesperodiuris and Paradiuris, which together form an unresolved basal grade. Species within subg. Diuris, sect. Diuris, comprising species with yellow pigments as the dominant anthoxanthins in their flower, are resolved by just one AFLP band from species in sect. Purpureo-albae, which have white pigments as the dominant anthoxanthins in their flowers (bootstrap percentage (BP) = 66). Sections within subg. Xanthodiuris are poorly resolved in the AFLP parsimony analysis. The majority-rule consensus trees from the two Bayesian analyses of the AFLP dataset had identical topologies, despite the different evolutionary models that were used to produce them. They were also almost identical to the parsimony tree (see Fig. 1 for posterior probabilities), differing only in not resolving D. monticola as sister to the rest of subgenus Xanthodiuris. However, the clades corresponding to subgenera Diuris and Xanthodiuris and section Diuris received posterior probabilities of only 0.90, 0.51 and 0.58, respectively. Only the clade corresponding to Diuris subgenera Diuris plus Xanthodiuris received significant posterior probabilities, i.e. 0.95 in the analysis with the standard model and 0.97 with the restriction-site (binary) model with no absence sites. Australian Systematic Botany 7 The alignment of ITS sequences plus coded ITS indels, which consisted of 704 putatively homologous nucleotide positions for 50 taxa, plus 11 indel characters, provided 239 variable characters of which 122 were parsimony informative. The heuristic parsimony search with this dataset was stopped when >1.9 million trees of length 376 steps had been found. One of these is shown in Fig. 2. The 50% majority-rule bootstrap consensus tree is shown in Fig. 3. D. sulphurea (subg. Paradiuris) forms a sister group to the rest of Diuris (BP = 100). The subgenera Diuris, Hesperodiuris and Xanthodiuris form three well supported clades (BP = 100) with similar levels of genetic divergence from each other. However, within subg. Diuris, just one DNA base difference resolves sect. Diuris from sect. Purpureo-albae. Similarly, in subg. Xanthodiuris, one DNA base difference separates sect. Abbreviatae from sect. Xanthodiuris. Sect. Pedunculatae contains two DNA base differences from sect. Xanthodiuris; however, owing to polymorphisms suggestive of hybridisation in two species samples, this section has only 48% bootstrap support and, hence, is not resolved on this tree. By contrast, species within subg. Hesperodiuris show much greater genetic diversity. Section Suffusae (BP = 99) is well supported as circumscribed. Section Hesperodiuris shows a relatively high level of genetic heterogeneity, with two well supported internal Eriochilus cucullatus Orthoceras strictum Diuris sulphurea 1 Diuris sulphurea 2 sect. Paradiuris subg. Paradiuris Diuris magnifica sect. Diuris corymbosa Diuris conspicillata Suffusae Diuris porphyrochila Diuris concinna Diuris brevifolia subg. Hesperodiuris Diuris drummondii sect. Diuris laxiflora Hesperodiuris Diuris picta Diuris carinata Diuris setacea Diuris palustris sect. Palustres Diuris abbreviata sect. Abbreviatae Diuris praecox Diuris behrii Diuris venosa sect. Pedunculatae Diuris chryseopsis Diuris ochroma Diuris goonooensis 3 Diuris monticola Diuris pardina subg. Xanthodiuris Diuris semilunulata 1 Diuris maculata 1 Diuris sp. aff. semilunulata sect. Xanthodiuris Diuris goonooensis 2 Diuris goonooensis 1 Diuris aequalis Diuris semilunulata 2 Diuris platichila 2 Diuris platichila 1 Diuris alba Diuris sp. aff. punctata1 Diuris sp. aff. alba 1 Diuris sp. aff. alba 2 Diuris arenaria sect. Purpureo-albae Diuris fragrantissima Diuris sp. aff. punctata 2 Diuris punctata 1 subg. Diuris Diuris punctata 2 Diuris punctata 3 Diuris aurea 2 Diuris aurea1 Diuris flavescens sect. Diuris Diuris byronensis Diuris tricolor 1 Diuris tricolor 2 Fig. 2. One of >1.9 million trees of length 376 steps produced by the analysis of ITS nucleotide sites plus coded indels, showing branch lengths estimated under the ACCTRAN algorithm. A total of 50 terminals, including 48 ingroup terminals, was analysed. The analysis comprised 715 total characters, of which 122 were parsimony informative. Eleven indels were coded as additional characters. Australian Systematic Botany J. O. Indsto et al. 100/100 99/100 99/100 69/59 66/ 97 65/100 99/95 - /97 100/100 100/100 94/100 63/79 100/100 63/64/92 Diuris magnifica sect. Diuris corymbosa Suffusae Diuris conspicillata Diuris sp. aff. porphyrochila Diuris drummondii Diuris concinna Diuris brevifolia sect. Diuris setacea Hesperodiuris Diuris laxiflora Diuris picta Diuris carinata Diuris alba Diuris aff. punctata 1 Diuris aff. alba 1 Diuris aff. alba 2 sect. Diuris arenaria Purpureo-albae Diuris fragrantissima Diuris aff. punctata 2 Diuris punctata 1 Diuris punctata 2 Diuris punctata 3 Diuris aurea 1 Diuris aurea 2 sect. Diuris flavescens Diuris Diuris byronensis Diuris tricolor 1 Diuris tricolor 2 Diuris behrii Diuris venosa sect. Diuris ochroma Pedunculatae Diuris monticola Diuris chryseopsis Diuris aequalis Diuris semilunulata 1 Diuris semilunulata 2 Diuris maculata 1 Diuris pardina sect. Xanthodiuris Diuris platichila 1 Diuris platichila 2 Diuris goonooensis 1 Diuris goonoensis 3 Diuris nigromontana Diuris goonooensis 2 Diuris abbreviata Diuris praecox subg. Diuris 100/100 subg. Hesperodiuris Eriochilus cucullatus Orthoceras strictum Diuris sulphurea 1 subg. Paradiuris Diuris sulphurea 2 Diuris palustris sect. Palustres subg. Xanthodiuris 8 sect. Abbreviatae Fig. 3. Bootstrap 50% majority-rule consensus tree for a total of 50 terminals, including 48 ingroup terminals, produced by analysis of ITS nucleotide sites plus coded indels, which summarises >1.9 million equally parsimonious trees of length 376 steps. Above each branch are its bootstrap support index (left of the slash) and Bayesian posterior probability (right of the slash), both expressed as percentages. The analysis comprised 715 total characters, of which 122 were parsimony informative. Eleven indels were coded as additional characters. A molecular phylogenetic analysis of Diuris by AFLP and ITS clades represented by D. drummondii (BP = 99) and D. setacea (BP = 87). Section Hesperodiuris can be considered well supported, although the internal relationships of this section as represented in the consensus tree are only moderately supported (BP = 66). The majority-rule consensus trees from the Bayesian analysis of the ITS dataset were very similar to the bootstrap consensus tree from the parsimony analysis (see Fig. 3 for posterior probabilities), resolving only two clades differently, with both of these clades with posterior probabilities of <0.95. Trees for AFLP and ITS are highly congruent so far as resolution allows comparison. We expected that AFLP would produce a more highly resolved tree than ITS since other studies had found AFLP to be sensitive enough even to resolve the parentage of plants in a population (Krauss 2000). For a given primer combination, the number of AFLP bands is expected to be proportional to genome size (Vos et al. 1995). As it was necessary to use 2-bp selective primers to obtain sufficient bands for analysis, we expect that the genome of Diuris is relatively small. The analysis of the combined dataset had to be stopped when it had produced more than 1.6 million trees of length 424 steps. The results are summarised in Fig. 4, the bootstrap consensus tree, which is more highly resolved than those produced from any of the data partitions. The total evidence tree shows moderate support (BP = 74) for the mainly western Australian subgenus Hesperodiuris being a sister group to the eastern subgenera Diuris and Xanthodiuris. Within subg. Hesperodiuris, Diuris palustris of the monotypic section Palustres appears as a sister group to sect. Hesperodiuris only, rather than sister to the whole of subg. Hesperodiuris as found in the ITS data. Within subg. Diuris, D. tricolor has equivocal placement in the total evidence tree as the floral characteristics are intermediate between sections Diuris and Purpureo-albae (yellow floral pigments of sect. Diuris with elongated lateral sepals of sect. Purpureo-albae). In addition, AFLP data suggest this species has affinity with sect. Purpureoalbae, while ITS evidence places it with sect. Diuris. Within subg. Xanthodiuris, sect. Pedunculatae is moderately well supported in the total evidence tree (BP = 82) as these species show clear morphological distinctions from sect. Xanthodiuris such as drooping lateral petals, densely papillate-hirsute callus ridges and higher leaf number. The majority-rule consensus trees from the Bayesian analysis of the combined dataset were very similar to the bootstrap consensus obtained with maximum parsimony (see Fig. 4 for posterior probabilities). Six clades were resolved differently by the different analyses; however, all of these involved branches with posterior probabilities <0.95. Discussion As we had expected, the results of our analyses of the morphological dataset were very indecisive, with parsimony consistently producing only one putative clade, composed of all five of the sampled species of Diuris subgenus Diuris section Pedunculatae and the Bayesian analysis producing only one significantly supported clade, composed of the sampled species of Diuris subgenus Hesperodiuris section Suffusae. This confirmed the need for additional evidence before the phylogeny of Diuris could be reconstructed with accuracy and precision. Analysis of two independent DNA datasets, ITS rDNA Australian Systematic Botany 9 sequences and AFLP resulted in highly congruent cladograms, both of which contained several consistently supported clades. This confirmed the wisdom of sampling molecular characters to supplement the morphological dataset. ITS analysis strongly supported the monophyly of Diuris and a basal split between D. sulphurea, the only species in Diuris subgenus Paradiuris, and a clade containing all other sampled species. Within the latter clade, both molecular datasets produced groupings corresponding to the remaining three subgenera of Jones and Clements (2006), although one of these, subgenus Hesperodiuris, was represented by only one species, D. conspicillata. The AFLP analysis strongly supported a sister group relationship between subgenera Diuris and Xanthodiuris in contrast to the ITS analysis, which could not resolve relationships between these clades and subgenus Hesperodiuris. Interestingly, neither of the molecular datasets resolved the only clade that emerged from the parsimony analyses of morphological characters, Diuris subgenus Diuris section Pedunculatae, which formed part of a larger polytomy in the ITS analysis and was weakly resolved as paraphyletic in the AFLP analysis. All three datasets therefore provided potentially complementary information on the phylogeny of Diuris. The analysis of the combined datasets provided the most highly resolved, best-supported bootstrap consensus tree of all of our analyses, even though it yielded such a large number of equally parsimonious trees that it needed to be stopped before completion of branch swapping. It also yielded the most highly resolved, best-supported Bayesian probability tree, which was highly congruent with the parsimony tree. These analyses supported the monophyly of all of the subgenera and all but two of the sections that had been proposed by Jones and Clements (2006). However, Diuris section Setaceae D.L.Jones & M.A.Clements was found to be embedded within section Hesperodiuris and, hence, is not supported by the present study. Species within Diuris sect. Pyrophilae D.L.Jones & M.A.Clements were not studied in the present analysis. To the extent that resolution allowed, ITS and AFLP datasets produced highly congruent trees. This confirmed the efficacy of the technique proposed by Indsto et al. (2005) of using AFLP profiles to identify pollinaria remnants collected from the bodies of putative pollinators. Surprisingly, the AFLP data were somewhat less phylogenetically informative than ITS data, which showed slightly higher resolution of inter-specific relationships. Even more surprisingly, the AFLP data showed almost no infraspecific variation across all sampled Diuris species within sections, in contrast to previous studies of other taxa, in which AFLP had been used at the intraspecific level for purposes demanding highly variable markers, such as paternity analysis (Krauss 2000; Coyle et al. 2003). A simplistic response to this result would be to suggest that in eastern Australia, species of Diuris have been circumscribed too narrowly and that species boundaries should be re-drawn at the limit of molecular phylogenetic resolution, if not higher. If implemented, this would reduce the number of eastern Australian species in our sample from 27 to 4, or fewer. This response, however, would be a mistake. First, it would treat as conspecific several entities that are both morphologically diagnosable and broadly sympatric and, in some cases, adapted to growing in distinctly different habitats. For example, D. chryseopsis and D. pardina would be treated as conspecific, even though they are broadly sympatric 4(0J1) 100/100 1(0J2) 97/100 8(0J1) 2(0J1) 1(0J3) 97/100 8(0J1) 99/100 63/99 50/1(0J2) 3(1J2) 9(0J1) 98/96 86/99 90/99 100/100 96/100 100/100 1(0J1) 4(0J1) 66/5(0J1) 72/81 X 5(1J0) 6(0J1) 74/- 3(1J0) 54/98 9(0J2) 57/- 61/8(0J1) 95/100 59/- 82/85 3(1J2) 7(0J1) 10(0J1) 4(0J1) 76/91 Eriochilus cucullatus Orthoceras strictum subg. Diuris sulphurea 1 Diuris sulphurea 2 Paradiuris Diuris magnifica sect. Diuris corymbosa Suffusae Diuris conspicillata Diuris porphyrochila (ms) Diuris palustris sect. Palustres Diuris drummondii Diuris concinna sect. Diuris brevifolia Hesperodiuris Diuris setacea Diuris laxiflora Diuris picta Diuris carinata Diuris tricolor 1 Diuris tricolor 2 Diuris alba Diuris aff. punctata 1 Diuris aff. alba 1 Diuris aff. alba 2 Diuris arenaria sect. Diuris fragrantissima Purpureo-albae Diuris aff. punctata 2 Diuris punctata 1 Diuris punctata 2 Diuris punctata 3 Diuris punctata 4 Diuris punctata 5 Diuris aurea 1 Diuris aurea 2 sect. Diuris aurea 3 Diuris Diuris flavescens Diuris byronensis Diuris abbreviata sect. Abbreviatae Diuris praecox Diuris aequalis Diuris semilunulata 1 Diuris semilunulata 2 Diuris maculata 1 Diuris maculata 2 Diuris maculata 3 Diuris pardina Diuris platichila 1 Diuris platichila 2 sect. Diuris goonooensis 1 Xanthodiuris Diuris goonooensis 3 Diuris goonooensis 4 Diuris goonooensis 5 Diuris goonooensis 6 Diuris goonooensis 7 Diuris nigromontana Diuris goonooensis 2 Diuris monticola Diuris behrii sect. Diuris venosa Pendunculatae Diuris aff. ochroma Diuris chryseopsis subg. Hesperodiuris 3(1J0) J. O. Indsto et al. subg. Diuris Australian Systematic Botany subg. Xanthodiuris 10 Fig. 4. Bootstrap 50% consensus tree for a total of 59 terminals, including 57 ingroup terminals, produced by analysis of a combined dataset of all sampled characters, which summarises >1.6 million equally parsimonious trees of length 424 steps. Datasets for AFLP, ITS sequence plus 11 indels, and 10 morphological characters were combined and analysed as a single dataset. Above each branch are its bootstrap support index (left of the slash) and Bayesian posterior probability (right of the slash), both expressed as percentages. The marks on the branches refer to morphological character-state changes listed below the marks, reconstructed by using Fitch parsimony (* denotes a unique forward change; | | specifies a parallel forward change;
denotes a reversal). Note that the character phylogeny for Character 8 is one of two equally parsimonious reconstructions. A molecular phylogenetic analysis of Diuris by AFLP and ITS in southern New South Wales, Victoria, Tasmania, and possibly in South Australia, and frequently co-occur even though they hybridise only sporadically, Second, AFLP variation is surprisingly low even within broadly defined species groups so it seems unlikely to be an artefact of excessive taxonomic ‘splitting’. Since AFLP provides dominant markers, it is unable to yield estimates of heterozygosity; however, it would be interesting to use a co-dominant marker system to test whether the low level of AFLP variation found in the present study reflects a generally low level of genetic variation, or just a small nuclear genome. Hybrids between species within the eastern subgenera Diuris and Xanthodiuris are well known. For example, D. aurea (sect. Diuris) and D. punctata (sect. Purpureo-albae) are sometimes sympatric and form the hybrid D.
nebulosa, which is intermediate in floral form between these species and readily identifiable in the field. Hybrids have also been commonly observed between species within sections Xanthodiuris and sect. Pedunculatae. Hybrids between species of different subgenera are rare. The frequency of natural hybridisation observed between sympatric species fits well with expectations based on the molecular distances between species presented here. A couple of instances of incongruence between datasets in the placement of individual species are presumably the results of either past hybridisation or incomplete lineage sorting. One of these involves the phylogenetic position of D. tricolor, the only species that has both elongated lateral sepals typical of section Purpureo-albae and yellow pigments as its dominant floral anthoxanthins. These morphological characters are incongruent with the ITS tree in which D. tricolor forms part of Diuris subgenus Diuris section Diuris, a yellow-flowered clade that mostly has short lateral sepals. The AFLP results are potentially congruent with either ITS or morphology because they place D. tricolor as part of a large polytomy with other species of section Purpureo-albae plus section Diuris. The observed patterns of character distribution are consistent with D. tricolor having evolved through hybridisation between ancestral species that belonged to sections Purpureo-albae plus section Diuris. Alternatively, if D. tricolor really is the sister species of section Purpureo-albae, its anomalous ITS sequence could be explained by retention of an ancestral ITS polymorphism in the most recent common ancestor of these taxa, followed by fixation of alternative alleles in its immediate descendants. The parsimony tree for AFLP is incongruent with that for morphology with respect to the placement of D. monticola. It placed this species as sister to the rest of sections Xanthodiuris, Abbreviatae and Pedunculatae, whereas morphology alone placed this species in the only putative clade that it decisively resolved, section Pedunculatae. The ITS results are potentially congruent with either AFLP or morphology because they include all of the species of sections Pedunculatae and Xanthoidiuris in one large polytomy with section Abbreviatae. These observed patterns of character distribution are consistent with either hybridisation or lineage-sorting processes, as described above for D. tricolor. Of the 10 morphological characters that we included in our analyses, five are completely consistent with the consensus tree Australian Systematic Botany 11 derived from our parsimony analysis of the combined dataset. Four of these provide synapomorphies for small species groups, whereas only one, labellum callus, was reconstructed as changing state near the base of the tree. A labellum callus with one ridge was resolved as the ancestral condition in Diuris, which subsequently transformed to one with two slightly raised parallel ridges in the clade comprising sections Hesperodiuris sensu lato and Palustres and to one with two conspicuously raised, divergent ridges in subgenus Diuris. The phylogenetic signal of this character has been overwhelmed by the conflicting signals provided by the five homoplasious characters in the morphological analysis. Other clades marked by synapomorphies that were consistent with the total evidence analysis were section Suffusae (horizontal tuber orientation), section Diuris (lateral sepals linear-spathulate) and section Pedunculatae (lateral petals horizontal to semi-pendent; labellum callus ridges densely papillate). Within subgenus Xanthodiuris, leaf number is distinctly higher in section Pedunculatae than in sections Xanthodiuris and Abbreviatae (>3 in Pedunculatae cf. 1–2 in sections Xanthodiuris and Abbreviatae); however, this feature also evolved in parallel in section Hesperodiuris. Within subgenus Diuris, distinctive forked tubers are found, although their presence is polymorphic in some species. Species within sect. Purpureoalbae have white floral anthoxanthins and long lateral sepals, both of which appear to be synapomorphies for this section. Within Diuris subgenus Xanthodiuris section Pedunculatae there are some species such as D. venosa which also have white floral anthoxanthins and this appears to have arisen independently in these species. Long lateral sepals comparable to those of section Diuris are also found in D. tricolor; however, as discussed above, this character maps equivocally on our best estimate of phylogenetic relationships. The large lateral labellum lobes of section Xanthodiuris are a conspicuous synapomorphy for this clade, but this feature is also found in section Suffusae. What can be inferred about the age of Diuris and evolution over time? Molecular dating analysis cannot yet provide us with a calibrated chronology because we have no fossils of Diuris to provide age constraints and very few for the Orchidaceae as a whole (Ramirez et al. 2007). However, inspection of the distribution of relative branch lengths in our molecular trees reveals some striking contrasts in macroevolutionary pattern. Within the predominantly Western Australian subgenus Hesperodiuris, several clades are well resolved according to ITS data and sections within this subgenus have strong bootstrap support. The shape of the ITS phylogram for this subgenus appears to be consistent, with approximately constant rates of diversification. In contrast, within the predominantly south-eastern subgenera Diuris and Xanthodiuris, there is minimal DNA evidence to distinguish species and to provide support for sections, a pattern that is also shown in the AFLP phylogram. The shape of the molecular phylograms for these clades resembles a pair of rakes, with long, unbranched ‘handles’ and broad ‘heads’ bearing short ‘tines’. This suggests that the macroevolutionary processes that have shaped the phylogeny of Diuris have differed significantly on either side of the Nullarbor Plain. The Western Australian taxa seem to have evolved by a process of phyletic gradualism, whereas evolution of the south-eastern clades seems to have involved either high rates of extinction or very sporadic 12 Australian Systematic Botany rates of speciation, or both, as would be expected under the model of punctuated equilibria (Gould and Eldredge 1972). The latter model proposes that rapid species radiation follows disturbance and species extinctions, or evolutionary innovation (Gould and Eldredge 1972). This rapid change is then followed by periods of relative stasis. The contrasting patterns observed in Diuris suggest that environments in which Diuris species live have been much more stable in the south-west than in the south-east. The flowers of all species of Diuris that have yellow anthoxanthins appear to mimic pea flowers of genera within the tribes Mirbelieae and Bossiaeeae (‘egg and bacon peas – see Indsto et al. 2006) and this state is decisively reconstructed as the ancestral condition within the genus. According to the model of punctuated equilibria invoked above, early Diuris species could have evolved to fill a niche opportunity soon after the appearance of the Mirbeliae and Bossiaeeae more than 35 million years ago (Wojciechowski 2003), which can be considered a best estimate in the absence of molecular dating. The pea flowers of these groups have strongly conserved floral form, despite extensive evolutionary radiation. Consistent with the punctuated equilibria theory is the possibility that the pea and Diuris orchid flowers recognisable today may have existed in similar form for tens of millions of years. Recent evidence challenges the idea that orchids are recently evolved. The oldest fossil orchid (Ramirez et al. 2007) 15–20 million years old, shows close resemblance in pollinaria structure to modern Goodyerinae. A fossil bee 65–70 million years old, now placed in the genus Cretotrigona, has been noted to be remarkably modern in appearance and to be deeply nested within a eusocial clade (Michener and Grimaldi 1988; Engel 2000). Perhaps one day a fossil Australian bee will be found, ancient yet modern in appearance, with an orchid pollinarium very like that of modern Diuris species attached to its face. Acknowledgements This work has been undertaken as part completion of a Master of Science degree by JI. Financial support for the project was from a Hermon Slade Orchid Trust grant. The Westmead Millennium Institute (University of Sydney) kindly provided laboratory facilities. John Riley in particular gave much advice on Diuris colony locations and flowering times. Michael Batley provided several Diuris samples used in this study. We also thank Debby McGerty for editing of graphics. References Beardsell DV, Clements MA, Huthchinson JF, Williams EG (1986) Pollination of Diuris maculata R.Br. (Orchidaceae) by floral mimicry of the native legumes Daviesia spp. and Pultenaea scabra R.Br. Australian Journal of Botany 34, 165–174. doi: 10.1071/BT9860165 Clements MA (2003) Molecular phylogenetic systematics in the Dendrobiinae (Orchidaceae), with emphasis on Dendrobium section Pedilonum. Telopea 10, 247–298. Clements MA, Jones DL, Sharma I, Nightingale ME, Garratt MJ, Fitzgerald KJ, MacKenzie AM, Molloy BPJ (2002) Phylogenetics of Diurideae (Orchidaceae) based on the internal transcribed spacer (ITS) regions of nuclear ribosomal DNA. Lindleyana 17, 135–171. Cox A, Pridgeon AM, Albert VA, Chase MW (1997) Phylogenetics of the slipper orchids (Cypripedioideae, Orchidaceae): nuclear rDNA ITS sequences. Plant Systematics and Evolution 208, 197–223. doi: 10.1007/BF00985442 J. O. Indsto et al. Coyle HM, Palmbach T, Juliano N, Ladd C, Lee HC (2003) An overview of DNA methods for the identification and individualization of marijuana. Croatian Medical Journal 44, 315–321. Dressler R (1990) ‘The orchids: natural history and classification.’ (Harvard University Press: Cambridge, MA) Engel MS (2000) A new interpretation of the oldest fossil bee (Hymenoptera: Apidae). American Museum Novitates 3296, 1–11. doi: 10.1206/0003-0082(2000)3296<0001:ANIOTO>2.0.CO;2 Farris J (1969) A successive approximations approach to character weighting. Systematic Zoology 18, 374–385. doi: 10.2307/2412182 Giambernardi TA, Rodeck U, Klebe RJ (1998) Bovine serum albumin reverses inhibition of RT–PCR by melanin. BioTechniques 25, 564–566. Gould SJ, Eldredge N (1972) Punctuated equilibria: an alternative to phyletic gradualism. In ‘Models in paleobiology’. (Ed. TJM Schopf) pp. 82–115. (Freeman, Cooper and Co.: San Francisco) Gravendeel B, Chase MW, de Vogel EF, Roos MC, Mes THM, Bachman K (2001) Molecular phylogeny of Coelogyne (Epidendroideae; Orchidaceae) based on plastid RFLPs, matK and nuclear ribosomal ITS sequences: evidence for polyphyly. American Journal of Botany 88, 1915–1927. doi: 10.2307/3558367 Hedren M, Fay MF, Chase MW (2001) Amplified fragment length polymorphisms (AFLP) reveal details of polyploid evolution in Dactylorhiza (Orchidaceae). American Journal of Botany 88, 1868–1880. doi: 10.2307/3558363 Indsto JO, Weston PH, Clements MA, Whelan RJ (2005) Highly sensitive DNA fingerprinting of orchid pollinaria remnants using AFLP. Australian Systematic Botany 18, 207–213. doi: 10.1071/SB04009 Indsto JO, Weston PH, Clements MA, Dyer AG, Batley M, Whelan RJ (2006) Pollination of Diuris maculata R.Br. (Orchidaceae) in the Sydney region, NSW by male Trichocolletes venustus bees. Australian Journal of Botany 54, 669–679. doi: 10.1071/BT05146 Indsto J, Weston PH, Clements MA, Dyer AG, Batley M, Whelan RJ (2007) Generalised pollination of Diuris alba (Orchidaceae) by small bees and wasps. Australian Journal of Botany 55, 628–634. doi: 10.1071/BT06207 Jones D (2000) ‘Natural groups in Diuris Sm.’ Fourth Australasian native orchid conference and show, Karralyka Centre, Melbourne. (Australasian Native Orchid Society) Jones D (2006) ‘Complete guide to native orchids in Australia: including the Island Territories.’ (New Holland Publishers Australia: Frenchs Forest) Jones D, Clements MA (2006) The infrageneric classification of Diuris Smith (Orchidaceae). The Orchadian 15, 203–206. Keith D (2004) ‘Ocean shores to desert dunes: the native vegetation of New South Wales and the ACT.’ (NSW Department of Environment and Conservation: Sydney) Kores PJ, Molvray M, Weston PH, Hopper SD, Brown AP, Cameron KM, Chase MW (2001) A phylogenetic analysis of Diurideae (Orchidaceae) based on plastid DNA sequence data. American Journal of Botany 88, 1903–1914. doi: 10.2307/3558366 Krauss SL (2000) Patterns of mating in Persoonia mollis (Proteaceae) revealed by an analysis of paternity using AFLP: implications for conservation. Australian Journal of Botany 48, 349–356. doi: 10.1071/BT98082 Lewis PO (2001) A likelihood approach to estimating phylogeny from discrete morphological character data. Systematic Biology 50, 913–925. doi: 10.1080/106351501753462876 Mant J, Peakall R, Weston PH (2005) Specific pollinator attraction and the diversification of sexually deceptive Chiloglottis (Orchidaceae). Plant Systematics and Evolution 253, 185–200. doi: 10.1007/s00606-005-0308-6 Michener CD, Grimaldi DA (1988) The oldest fossil bee: apoid history evolutionary stasis, and antiquity of social behavior. Proceedings of the National Academy of Sciences, USA 85, 6424–6426. doi: 10.1073/pnas.85.17.6424 A molecular phylogenetic analysis of Diuris by AFLP and ITS Okuyama Y, Noriyuki F, Wakabayashi M, Kawakita A, Ito M, Watanabe M, Murakami N, Kato M (2004) Nonuniform concerted evolution and chloroplast capture capture:heterogeneity of observed introgression patterns in three molecular data partition phylogenies of Asian Mitella (Saxifagaceae). Molecular Biology and Evolution 22, 285–296. doi: 10.1093/molbev/msi016 Orthia L, Crisp MD, Cook LG, de Kok RPJ (2005) Bush peas: a rapid radiation with no support for monophyly of Pultenaea (Fabaceae: Mirbeliae). Australian Journal of Botany 18, 133–147. doi: 10.1071/SB04028 Pridgeon A, Solano R, Chase MW (2001) Phylogenetic relationships in Pleurothallidinae (Orchidaceae): combined evidence from nuclear and plastid sequences. American Journal of Botany 88, 2286–2308. doi: 10.2307/3558390 Ramirez S, Gravendeel B, Singer RB, Marshall CR, Pierce NE (2007) Dating the origin of the Orchidaceae from a fossil record record with its pollinator. Nature 448, 1042–1045. doi: 10.1038/nature06039 Ranamukhaarachchi DG, Kane ME, Guy CL, Li QB (2000) Modified AFLP technique for rapid genetic characterization in plants. BioTechniques 29, 858–866. Simmons MP, Ochoterena H (2000) Gaps as characters in sequence-based phylogenetic analyses. Systematic Biology 49, 369–381. doi: 10.1080/10635159950173889 Sun Y, Skinner DZ, Liang GM, Hubert SH (1994) Phylogenetic analysis of Sorghum and related taxa using internal transcribed spacers of nuclear ribosomal DNA. Theoretical and Applied Genetics 89, 26–32. doi: 10.1007/BF00226978 Australian Systematic Botany 13 Swofford DL (2003) ‘PAUP*. Phylogenetic analysis using parsimony (*and other methods). Version 4.’ (Sinauer Associates: Sunderland, MA) van der Cingel N (2001) ‘An atlas of orchid pollination: America, Africa, Asia and Australia.’ (AA Balkema Publishers: Rotterdam, The Netherlands) Vos P, Hogers R, Bleeker M, Reijans M, van der Lee T, Hornes M, Frijters A, Pot J, Peleman J, Kuiper M, Zabeau M (1995) AFLP: a new technique for DNA fingerprinting. Nucleic Acids Research 23, 4407–4414. doi: 10.1093/nar/23.21.4407 Weider LJ, Elser JJ, Crease TJ, Mateos M, Cotner JB, Markow TA (2005) The functional significance of ribosomal (rDNA) variation: impacts on the evolutionary ecology of organisms. Annual Review of Ecology and Systematics 36, 219–242. doi: 10.1146/annurev.ecolsys.36.102003.152620 Wojciechowski M (2003) Reconstructing the phylogeny of legumes (Leguminosae): an early 21st century perspective. In ‘Advances in legume systematics’. Part 10. Higher level systematics. (Eds B Klitgaard, A Bruneau) pp. 5–35. (Royal Botanic Gardens, Kew: London) Manuscript received 17 June 2008, accepted 24 November 2008 14 Australian Systematic Botany Appendix 1. J. O. Indsto et al. Diuris species list compiled by Mark Clements, current as at December 2008 (see http://anbg.gov.au/cpbr/cd-keys/orchidkey/html/ currentspecies.html) Diuris Sm., Trans. Linn. Soc. London 4: 222 (1798). Type species: Diuris aurea Sm. subgen. Diuris Sect. Diuris Synonym: Diuris sect. Flaviflorae G.Don in Loudon’s Hortus Britannicus 368 (1830). Type species: Diuris aurea Sm., fide Jones and Clements (2006). Diuris aurea Smith, Exotic Bot. 1: 15, t. 9 (1805). Diuris byronensis D.L.Jones, Orchadian 14(3): 132–133, f. 1, t. (2003). Diuris chrysantha D.L.Jones et M.A.Clem., Proc. Roy. Soc. Queensland 98: 130–132, f. 7 (1987). Diuris disposita D.L.Jones, Austral. Orch. Res. 2: 55, f. 69 (1991). Diuris flavescens D.L.Jones, Austral. Orch. Res. 2: 56, f. 71 (1991). Diuris luteola D.L.Jones et B.Gray, Austral. Orch. Res. 2: 57–58, f. 73 (1991). Diuris unica D.L.Jones, Austral. Orch. Res. 5: 82, f. 3.13, t. (21 Dec. 2006). Diuris secundiflora Fitzg., Austral. Orch. 1(4): [t. 9] (1878). Diuris tricolor Fitzg., J. Bot. 23: 137 (1885). sect. Purpureo-albae G.Don in Loudon’s Hortus Britannicus 368 (1830). Type species: Diuris alba R.Br., fide Jones and Clements (2006). Diuris alba R.Br., Prod. 326 (1810). Diuris arenaria D.L.Jones, Orchadian 12(12): 567–568, f. 1, t. (1999). Diuris callitrophila D.L.Jones, Orchadian 14(3): 133–135 (2003). Diuris curta D.L.Jones, Austral. Orchid. Res. 5: 76–77, f. 3.9 (21 Dec. 2006). Diuris daltonii (C.Walter) D.L.Jones et M.A.Clem., Orchadian 14(8): Sci. Suppl. xiv (June 2004). Diuris dendrobioides Fitzg., Austral. Orch. 1(7): [t. 3] (1882). Diuris fragrantissima D.L.Jones et M.A.Clem., Austral. Orch. Res. 1: 68 (1989). Diuris minor (Benth.) D.L.Jones et M.A.Clem., Orchadian 14(8): Sci. Suppl. xiv (June 2004). Diuris oporina D.L.Jones, Austral. Orch. Res. 2: 59–60, f. 77 (1991). Diuris parvipetala (Dockrill) D.L.Jones et M.A.Clem., Proc. Roy. Soc. Queensland 98: 132 (1987). Diuris punctata Sm. var. punctata Exotic Bot. 1: 13, t. 8 (1804). Diuris punctata Sm. var. sulphurea Rupp, Proc. Linn. Soc. New South Wales 69: 73 (1944). subgen. Xanthodiuris D.L.Jones et M.A.Clem., Orchadian 15(5): 203 (2006). Type species: Diuris maculata Sm. sect. Xanthodiuris Diuris aequalis F.Muell. ex Fitzg., Austral. Orch. 1(2): [t. 6] (1876). Diuris bracteata Fitzg., Austral. Orch. 2(4): [t. 2] (1889). Diuris cuneilabris Rupp, Proc. Linn. Soc. New South Wales 73: 134, f. 1 (1948). Diuris goonooensis Rupp, Victorian Naturalist 72: 110 (1955). Diuris maculata Sm., Exotic Bot. 1: 57, t. 30 (1804–05). Diuris nigromontana D.L.Jones, Orchadian 15(12): 550–551, t. (June 2008). Diuris pardina Lindl., Gen. sp. orchid. pl. 507 (1840). Diuris platichila Fitzg., Austral. Orch. 2(4): [t. 3] (1891). Diuris semilunulata Messmer in Rupp, Orch. New South Wales 139–140 (1943). sect. Abbreviatae D.L.Jones et M.A.Clem., Orchadian 15(5): 203–204 (2006). Type species: Diuris abbreviata Benth. Diuris abbreviata Benth., Fl. Austral. 6: 329 (1873). Diuris exitela D.L.Jones, Austral. Orch. Res. 2: 55–56, f. 70 (1991). Diuris praecox D.L.Jones, Austral. Orch. Res. 2: 60, f. 78 (1991). sect. Pedunculatae D.L.Jones et M.A.Clem., Orchadian 15(5): 204 (2006). Type species: Diuris pedunculata R.Br. Diuris basaltica D.L.Jones, Austral. Orch. Res. 5: 75–76, f. 3.9 (21 Dec. 2006). Diuris behrii Schldl., Linnaea 20: 572 (1849). Diuris chryseopsis D.L.Jones, Austral. Orch. Res. 3: 74–75, f. 4.1 (1998). Diuris eborensis D.L.Jones, Austral. Orch. Res. 5: 77–78, f. 3.10, t. (21 Dec. 2006). Diuris fucosa D.L.Jones, Austral. Orch. Res. 5: 78, f. 3.11, t. (21 Dec. 2006). Diuris gregaria D.L.Jones, Austral. Orch. Res. 5: 79–80, f. 3.12 (21 Dec. 2006). Diuris lanceolata Lindl., Gen. sp. orchid. pl. 508 (1840). Diuris monticola D.L.Jones, Austral. Orch. Res. 3: 76–77, f. 4.3 (1998). Diuris ochroma D.L.Jones, Muelleria 8(2): 182–184, f. 2 d-f (1994). Diuris pedunculata R.Br., Prod. 316 (1810). Diuris protena D.L.Jones, Austral. Orch. Res. 5: 81–82 (21 Dec. 2006). A molecular phylogenetic analysis of Diuris by AFLP and ITS Australian Systematic Botany Diuris subalpina D.L.Jones, Orchadian 15(12): 551, t. (June 2008). Diuris venosa Rupp, Proc. Linn. Soc. New South Wales 51: 313, f. 1–6 (1926). subgen. Hesperodiuris D.L.Jones et M.A.Clem., Orchadian 15(5): 204 (2006). Type species: Diuris laxiflora Lindl. sect. Hesperodiuris Synonym: Diuris subgen. Hesperodiuris sect. Setaceae D.L.Jones et M.A.Clem., Orchadian 15(5): 204 (2006). Type species: Diuris setacea R.Br. Diuris brevifolia R.S.Rogers, Trans. & Proc. Roy. Soc. South Australia 46: 148 (1922). Diuris carinata Lindl., Gen. sp. orchid. pl. 510 (Sep. 1840). Diuris concinna D.L.Jones, Austral. Orch. Res. 2: 53–54, f. 67 (1991). Diuris drummondii Lindl. in Edwards’s, Bot. Reg. 1–23: Swan Riv. Append. li (1840). Diuris emarginata R.Br., Prod. 316 (1810). Diuris filifolia Lindl. in Edwards’s, Bot. Reg. 1–23: Swan Riv. Append. li (1840). Diuris heberlei D.L.Jones, Austral. Orch. Res. 2: 56–576, f. 72 (1991). Diuris immaculata D.L.Jones, Austral. Orch. Res. 5: 80–81 (21 Dec. 2006). Diuris laxiflora Lindl. in Edwards’s, Bot. Reg. 1–23: Swan Riv. Append li (1840).D. Diuris micrantha D.L.Jones, Austral. Orch. Res. 2: 58–59, f. 75 (1991). Diuris pauciflora R.Br., Prod. 316 (1810). Diuris picta J.Drummond in Hooker’s, J. Bot. 5: 347 (1853). Diuris setacea R.Br., Prod. 316 (1810). sect. Pyrophilae D.L.Jones et M.A.Clem., Orchadian 15(5): 204 (2006). Type species: Diuris laevis Fitzg. Diuris laevis Fitzg., Gard. Chron. (new ser.), 17: 495 (1882). Diuris purdiei Diels, J. Proc. Mueller Bot. Soc. Western Australia_1(11): 79 (1903). sect. Palustres D.L.Jones et M.A.Clem., Orchadian 15(5): 204–205 (2006). Type species: Diuris palustris Lindl. Diuris palustris Lindl., Gen. sp. orchid. pl. 507 (1840). sect. Suffusae D.L.Jones et M.A.Clem., Orchadian 15(5): 205 (2006). Type species: Diuris longifolia R.Br. Diuris amplissima D.L.Jones, Austral. Orch. Res. 2: 53–54, f. 65 (1991). Diuris brumalis D.L.Jones, Austral. Orch. Res. 2: 53, f. 66 (1991). Diuris conspicillata D.L.Jones, Austral. Orch. Res. 2: 54–55, f. 68 (1991). Diuris corymbosa Lindl. in Edwards’s, Bot. Reg. 1–23: Swan Riv. Append. li (1840). Diuris longifolia R.Br., Prod. 316 (1810). Diuris magnifica D.L.Jones, Austral. Orch. Res. 2: 58, f. 74 (1991). Diuris orientis D.L.Jones, Austral. Orch. Res. 3: 77–78, f. 4.4 (1998). Diuris porrifolia Lindl. in Edwards’s, Bot. Reg. 1–23: Swan. Riv. Append. li (1840). Diuris pulchella D.L.Jones, Austral. Orch. Res. 2: 61, f. 79 (1991). Diuris recurva D.L.Jones, Austral. Orch. Res. 2: 61–62, f. 80 (1991). subgen. Paradiuris D.L.Jones et M.A.Clem., Orchadian 15(5): 205 (2006). Type species: Diuris sulphurea R.Br. Diuris sulphurea R.Br., Prod. 316 (1810). subgen. Timordiuris D.L.Jones et M.A.Clem., Orchadian 15(5): 205 (2006). Type species: Diuris fryana Ridl. Diuris fryana Ridl. in H.O.Forbes, Nat. Wand. East. Archip. 519 (1885). Natural hybrids Diuris
fastidiosa R.S.Rogers, Trans. & Proc. Roy. Soc. South Australia 51: 6–7 (1927). Diuris
nebulosa D.L.Jones, Austral. Orch. Res. 2: 59, f. 76 (1991). Diuris
palachila R.S. Rogers, Trans. & Proc. Roy. Soc. South Australia 31: 209–210 (1907). Diuris
polymorpha Messmer in Rupp, Orchids New South Wales, Suppl. 142 (1943). http://www.publish.csiro.au/journals/asb 15