Chapter 1 General introduction Abildgaardieae Lye The tribe Abildgaardieae Lye is one of the tribes within the subfan1ily Cyperoideae (Table 1.1), in the family Cyperaceae and, depending on the classification used (Goetghebeur 1986; Bruhl 1995; Goetghebeur 1998; Simpson et al. 2005, in press), is composed of five, six or seven genera: Abildgaardia Vahl (TYPE genus), or as Fimbristylis Vahl section Abildgaardia Benth., Fimbristylis Vahl, Bulbostylis Kunth, Crosslandia W.Fitzg., Nemum Desv.ex Ham., Nelmesia Van der Veken and Tylocalya Nelmes, or as Fimbristylis nelmesii (Kern 1958, 1974; Simpson 1993; Goetghebeur 1998; World Checklist of Monocotyledons 2004). History of the tribe Abildgaardieae The genera Abildgaardia (expanded to include species of Bulbostylis) and jVe11lU11l were separated from Fi11lbristylis (Lye 1973) and Scirpus (RaYllalI973; Lye] 973) respectively, who placed them into his new tribe Abildgaardieae Lye. J\Tel11lesia was also considered for possible inclusion. Although Lye had formed the new tribe, the publication 'Sedges and Rushes of East Africa' (Haines and Lye 1983) retained the three genera in the tribe Scirpeae, simply commenting that several srnaJler tribes (including Abildgaardieae) would be a better option. In his systematic study of the family Cyperaceae, Goetghebeur (1986) included Crosslandia, Fi11lbristylis, TylocaJya, and Nelmesia in Lye's tribe Abildgaardieae. Bruhl (1995) supported the Table 1.1 Tribes of the Cyperaceae. Bruhl's classification shows the tribal and generic concepts used for this study. Altemative classifications provide comparative placement of genera studied within the confines of the present project. Some taxa from Schoenoplectus* (includes Schoenoplectiella Lye) were used as additional outgroup taxa for cladistic analyses. The number of genera and species, respectively, are shown in brackets after each tribe in the most recent studies. Bruhl (1995) Goetghebeur (1998) Goetghebeur(1986) Kern (1974) (for Malesia) Subfamily: Cyperoideae Tribes: Cyperaeae (17/878) Scirpeae (28/518) Genus: Schoenoplectus* Abildgaardieae (7/430) Genera: Abildgaardia Bulbostylis Crosslandia Fimbristylis Nemum Nelmesia TylocGf)'a Arthrostylideae (4/6) (name invalid) Genera: Actinoschoenus Arthrostylis Trachystylis Trichoschoenus Subfamily: Mapanioideae Tribes: Hypolytreae (9/130) Chrysitricheae (4/13) Subfamily: Cyperoideae Tribes: Scirpeae (6/60) Fuireneae (5/90) Genus: Schoenoplectus* Eleocharideae (3/200) Abildgaardieae (6/420) Genera: Fimbristylis (including TylocGf)'a) Crosslandia Bulbostylis Abildgaardia Nemum Nelmesia Tribes: Cypereae (19/900) Dulichieae (3/1 0) Schoeneae (29/700) Genera: Arthrostylis Actinoschoenus Trichoschoenus Trachystylis Subfamily: Mapanioideae Tribes: Hypolytreae Chrysitricheae Subfamily: Cyperoideae Tribes: Scirpeae Fuireneae Genus: Schoenoplectus* Eleocharideae Abildgaardieae Genera: Fimbristylis Tylocarya Crosslandia Bulbostylis Abildgaardia Nemum Nelmesia Tribes: Ficineae Cypereae Dulichieae Arthrostylideae (provisional name) Genera: Arthrostylis Actinoschoenus Trichoschoellus Trachystylis Rhynchosporeae Schoeneae Subfamily: Cyperoideae Tribes: Hypolytreae Cyperaeae Genera: Fimbristylis (includes Abildgaardia and Actinoschoenus as sections) Bulbostylis Scirpus Fuirena LipocGfpha Eleocharis Cypems Rhynchosporeae Subfamily: Caricoideae Tribes: Sclerieae Cariceae Genus: Schoenoplectus* Bentham and Hooker (1880) ----- Subfamily: Caricoideae Tribes: Rhynchosporeae (4/273) Schoeneae (27/379) Cryptangieae (5/92) Trilepideae (4/15) Cariceae (6/2089) Sclerieae (2/201) Bisboeckelereae (4/22) Hypolytreae (14/159) Subfamily: Scleriodeae Tribes: Cryptangieae (4/50) Trilepideae (4/15) Sclerieae (1/250) Bisboeckelereae (4/25) # Subfamily: Scleriodeae Tribes: Cryptangieae Trilepideae Scleneae Bisboeckelereae Suhfamily: C'aricoideae Tl'ibe: Cariceae (5/2150) I I I Subfamily: Caricoideae Tribe: Cariceae . I N.B. # Goetghebeur (1995) has monotypic genera Exochogyne and Koyamaea as incertae sedis Monoclines: Tribe: Scirpeae Genera: Fimbristylis (included Abildgaardia and Bulbostylis [= Oncostylis] as sections) Sci/pus sect. Schoenoplectus (ScllOenoplectus *) Tribes: Hypolytreae Rhynchosporeae Genera: Arthrostylis Actinoschoenus Diclines: Tribes: Cryptangieae Sclerieae Cariceae 3 placement of the genera within the tribe in his systematic study where vegetative anatomy was a focus. The tribe Fimbristylideae Cherm. ex RaYllallinked the related Fimbristylis and Bulbostylis, but was not widely used. Kukkonen (1991) did not recognise either tribe Fimbristylideae or Abildgaardieae, but retained the tribe Scirpeae. There has been little disagreement with the general boundaries of the tribe, however, the limits of some of the main genera within the tribe have been disputed and are still unresolved (Greuter et al. 1993; Simpson 1993; World Checklist of Monocotyledons 2004). General history of genera of the Abildgaardieae Finlbristylis, Abildgaardia, and Bulbostylis Vahl (1805) separated 21 species from Scirpus, placing 19 species plus four new species into his genus Fimbristylis, and two species, Abildgaardia mOl1ostachya (L.) Vahl (= A. ovata (Burm.f.) Kral or Fimbristylis ovata (Burm.f.) J.Ke:m) and A. tristachya Vahl (= A. triflora (L.) Abeywickr. or Fimbristylis triflora (L.) K.Schum. ex Engl.) into Abildgaardia. The generic status of Fimbristylis was widely accepted by the botanists of the day, but not Abildgaardia. Robert Brown (18] 0) collected extensively in Australia and described two new species of Abildgaardia, A. vaginata R.Br. and A. schoenoides R.Br. Botanists were tempted by the generic character of distichous-subdistichous glumes used to define Abildgaardia, with many new species assigned to Abildgaardia. Kunth (1837) certainly accepted the new genus, acknowledging 14 species in his circumscription. Mueller, however, preferred 4 Hasskarl's placement of Abildgaardia monostachya into Fimbristylis as F. monostachya (L.) Hassk., naming his new species F. oxystachya F.Muell. Initially Benthanl (1861) used Abildgaardia as a genus and included A. monostachya, A. eragrostis and A. fusca as species for the Hong Kong Flora. He comrnented on the habit being similar to Fimbristylis and that only the distichous glumes in Abildgaardia were different. It was in Bentham's later classifications (Bentharn 1878; Bentham and Hooker 1880) that Abildgaardia was demoted to a section within Fimbristylis, where he included F. oxystachya, F. macrantha Boeck., F. squarrulosa (syn. A. schoenoides), F. brownii Benth. (= A. vaginata) and F. dallachyi (= F. fimbristyloides) with F. monostachya. It was at this time that Bentham (1878) also denloted Bulbostylis to Fimbristylis section Oncostylis (F. barbata (RoUb.) Benth. and F. capillaris A.Gray), so that the expanded genus Fimbristylis then conlprised five sections: Heleocharoides, Dichelostylis, and ｔｲｩ｣ｨ･ｬｯｳｾｶ ｩ Ｌ Abildgaardia, and Oncostylis. Bulbostylis was reinstated to generic status when Clarke (1900, 19(8) recognised four sections of Fimbristylis: Eleocharoides, Dichelostylis, TrichelostyUs, and Abildgaardia. Clarke (1908) then assigned the species of Bulbostylis (some moved from Isolepis R.Br.) into five sections based on morphological features, narrowing groups within sections according to their broad geographical distribution (Table 1.2). The genera remained stable for some time, until Koyama (1961) incorporated Bulbostylis once again into Fimbristylis, although he chose the rank of subgenus. The move was generally not accepted and Van der Veken (1965) showed that species of Bulbostylis shared an embryo type that was different to that in species of Fimbristylis. Kern (1974) maintained Bulbostylis as a genus and Abildgaardia as a section of Fimbristylis. Table 1.2 Classification of Bulbostylis by Clarke (1908). Species sampled in the tribal study are indicated by an *. Section I. Stylus 2-fidus Section II. Stylus 3-fidus Section III. Stylus 3-fidus. Culmus l-cephalus Section IV. Stylus 3-fidus. Capitula umbellatae Section V. Stylus 3-fidus. Spiculae umbellatae B. humilis *(as B. striatella) Africanae B. festucoides B. renschii B. zambesica B. breviculmis* (as B. striatella) B. sphaerocarpus B. schlechteri B·funckii Americanae B. leucostachya B. conifera B.lanata B. paradoxa B. schaffneri B. pauciflora B.jloccosa B. aturensis Gerontogeae (old world) B. barbata* B. rarissima B. lichtensteiniana B. schoenoides B. scleropus B. parvinux B. cinnamomea B. collina B. filamentosa B.laniceps B. cardiocarpa B. erratica B. burkei B. astrosanguinea B. comorensis B. schimperiana B. fimbristyloides B. aphyllanthoides B. buchanani B. subspinescens Neogeae (new world) B. stenophylla B. warei B. subaphylla B. glaziovii B. fimbriata B. sphaerocephala Gerontogae B. zeyheri B. kirkii B. trabeculata B.japonica Neogeae B. vestita B. graminifolia B. consanguinea B. junciformis B. nesiotis Cosmopolitan B. capillaris* Gerontogeae B. puberula * B. taylori B. burchellii B. filiformis*(as B. hispidula) B. abortive B. biovini B. coleotricha B. johnstoni B. parva B. mucronata B. andongensis B. transiens B. melanocephala B. macra B. megastachys B. oritrephes B. trichobasis B. cylindrical Neogeae B·fendleri B. langsdorfiana B. scabra B. arenaria B. tene/la B. circinata B. paraensis B. jacobinae B. laeta B. asperula B. micans 6 In more recent times, Lye (1973; Haines and Lye 1983) placed Bulbostylis as a subgenus of Abildgaardia (Abildgaardia had nomenclatural priority), based on a similar embryo type that was distinct from that found in species of Fimbristyli3. Goetghebeur and Coudijzer (1984, 1985) disagreed with Lye's change and stated that Bulbostylis and Abildgaardia had evolved separately from Fimbristylis. Differences between the embryo types, in conjunction with nut epidermal (fruit wall) ornamentation and the presence of long white hairs at the mouth of the leaf sheath, factored in keeping Bulbostylis, Abildgaardia and Fimbristylis as distinct genera. RobYlls and Tournay (1955), and Kral (1971) had already adopted the equal generic rank in their regional studies. In later studies Lye (1995, 1996, 2000) accepted the generic rankings of Bulbostylis, Abildgaardia and Fimbristylis. Workers at Kew Herbarium (Hooper 1973; Ken1 1974; Simpson 1993; Simpson and Koyama 1998) have been steadfast in retaining Abildgaardia as a section of the genus Fimbristylis, while maintaining Bulbostylis as a separate genus. Kral (1971, 2002), Goetghebeur and Coudijzer (1984, 1985); Goetghebeur ＨＱＹＸＶｾＬ 1998), Greuter et £11. (1993), Gordon-Gray (1995), and Bruhl (1995) all accepted Abildgaardia as a genus, equal in rank with Bulbostylis and Fimbristylis. The widespread Abildgaardia ovata, plus the African species A. triflora, and A. hygrophila (Goetghebeur and Coudijzer 1984), and the American species A. mexicana, A. baeothryoJ1 and A. papillosa (Kral and Strong 1999) are currently assigned to the genus Abildgaardia. The Australian species Fimbristylis macrantha, F. oxystachya, and F. pachyptera, in section Abildgaardia of Fimbristylis, were provisionally placed in the reinstated genus Abildgaardia (Goetghebeur 1986), however, the new con1binations were never validly published. Fimbristylis squarrulosa (section Abildgaardia) was not included in Goetghebeur's study, but is accepted in the 7 Australian Plant Name Index (http://www.anbg.gov.au/cpbr/databases/apni.htrnl) as Abildgaardia schoenoides R.Br. Crosslandia W.Fitzg. was described as a monotypic genus (Fitzgerald 1918) from the Kimberley region in Western Australia, however, Goetghebeur (1986) proposed a second species be described, primarily on the variation of the inflorescence. The provisional name C. anthelata was never validly published. Nemum Desv. ex Ham. is another relatively old genus raised from Scirpus and has had varied acceptance. Kunth (1837) and Steudal (1855) included Nemum in their classifications, however, Clarke (1908) retained the species in Scirpus section Nemum (s. spadiceus Boeck. and S angolensis C.B.Clarke). Raynal (07··1973) reinstated Nemum to generic status as did Lye (11-1973), when he included the genus in the tribe Abildgaardieae. Nemum is currently generally accepted as a genus (Greuter et a1. 1993, Simpson 1993; World Checklist of Monocotyledons 2004). Nelmesia Van der Veken is a monotypic genus (Van der Veken 1955), with N. melanostachya known from a single collection in Africa (Goetghebeur 1986). Tylocarya Nelmes, described from Thailand, is also a monotypic genus, Tylocarya cylindrostachya Nelmes (Nelmes 1949), but was recombined as a species of Fimbristylis by Kern as F. nelmesii J.Kern (Kern 1958). Tylocarya has had mixed acceptance as a genus (Bruhl 1995), however, the most supported current view is as a species of Fimbristylis (Simpson 1993; Goetghebeur 1998). 8 Embryo morphology and Anatomy In Cyperaceae, the application of characters from embryo morphology and the type of photosYnthetic pathway in systematic studies has been well documented, especially when the use of plant morphology has its limitations. Van der Veken (1965) sampled embryos from 342 species across 16 genera, and found six different embryo types within Scirpus s.l., three of those types correlate specifically with Fimbristylis, Bulbostylis and Abildgaardia. Investigating the embryo type has proved useful in the placement of difficult taxa. For example, species of the now accepted Bulbostylis hispidula group were controversially placed in Fimbristylis, until it was found that sampled material had a Bulbostylis-type embryo. This embryo type determined the placement of members of the B. hispidula group in Bulbostylis. Metcalfe (1971) sampled the vegetative anatomy of the family extensively and related variation in leaf, culm and root anatomy to the classification popular at the time. The variation in anatomy, specifically the arrangement of tissues in vascular bundles, has since been associated with the photosYnthetic pathway, and has evolved many times in the family (Bruhl et al. 1987; Bruhl and Perry 1995; Soros and Bruhl 2000; Soros and Dengler 2001). In addition to the C3 photosYnthetic type of anatomy there are four C 4 anatomical types that are clearly associated with specific generic types (Figure 1.4). These types are the: rhYnchosporoid, ｣ｨｬｯｲ ｹｰ･ ｩ､ｾ fimbristyloid and eleocharoid types (Bruhl et a1. 1987; Soros and Dengler 2001). Some studies have related the type of anatomy to the classification of taxa at the broad tribal level (Bruhl 1995; Soros and Bruhl 2000), or at lower ranks e.g. the chlorocyperoid photosYnthetic pathway in Cyperus sect. Pinnati (Wilson 1991). 9 General aim The general aim of this thesis is to test monophyly of and within the tribe Abildgaardieae, using characters derived from plant morphology, embryo morphology and vegetative anatomy. To address the main aim of the thesis it has been necessary to define the limits and assess monophyly of genera (specifically Abildgaardia, Crosslandia and to a lesser extent, Australian Bulbostylis and Fimbristylis) within the tribe Abildgaardieae prior to the full tribal analysis that tests monophyly. The main aim is divided into minor objectives that are addressed as separate chapters to incorporate species level work where necessary (i.e. Crosslandia, Abildgaardia and Bulbostylis). The rninor objectives are given in the introductions of the individual chapters. Thesis outline The history of the tribe and taxonomic problem areas are outlined above (Chapter 1), followed by general materials and methods (Chapter 2). In Chapter 3, the lilllits of the genus Crosslandia are defined with the number of species within Crosslandia assessed. A cladistic analysis was used to determine if the species fOffil a monophyletic group. The limits of the disputed genus or section Abildgaardia are defined in Chapter 4, by determining the number of species that combine to fom1 the group, and the limits assessed across their global distribution. Cladistic analysis was used to test monophyly of all the species of Abildgaardia defined in the phenetic study and of selected samples of Fimbristylis. The number of species assigned to Bulbostylis is large and in need of revision globally. As only a small nunlber of species occur in Australia, the limits of the Australian species are assessed and 10 con1pared globally in Chapter 5. The final analysis in Chapter 6 brings together data frOlTI Chapters 3-5, and data for the remaining genera Nemum, Nelmesia and Tylocarya (= Fimbristylis nelmesii), to complete the analysis of the tribal group. Finally, the general conclusion attempts to synthesise the findings of all the da':a chapters and suggest areas where further research is needed. II Chapter 2 General materials and methods Plant Inaterial Data for phenetic and cladistic studies were obtained primarily frorn dried herbarium material. Field collections that supplemented loan material were focused in the regions of the East Kimberley in Western Australia, Kakadu National Park in the Northern Territory, far north Queensland, and to a lesser extent southeast Queensland and northeast New South Wales. Herbarium vouchers for all material collected are lodged with the N.C.W. Beadle Herbarium (NE) at the University of New England. Duplicate specimens will be distributed to other herbaria after manuscripts have been prepared for publication. Many field collections included fixed material for my anatomical studies and silica-gel samples for other projects (e.g. Ghamkhar et al. 2005, in press). Fixed specimens were placed in Fonnalin-Propiono-Alcohol (FPA), constituting proportions of Fonnalin, Propionic acid, and 70% ethanol (5:5:90), imnlediately upon collection. Specimens were held in fixative for at least one week to ensure adequate penetration of the tissues. In the laboratory, fixative was replaced with 700/0 ethanol (with 1% glycerol added) to facilitate long-tenn storage and safe handling. Specirnen loans were provided by Australian herbaria BRI, CANB, DNA, MBA, MEL, NSW, PERTH, QRS, and overseas herbaria EA, K, L, MO, NlT, P, PRE (Holmgren et al. 1990). 12 Taxa Studied Genera assigned to the tribe Abildgaardieae Lye (see Lye 1973; Goetghebeur 1986, 1998; Bruhl 1995) were sampled to test monophyly of the tribe. Representatives from Nemum, Nelmesia and Tylocarya (= Fimbristylis nelmesii), all non-Australian genera, were only included in the tribal cladistic analysis. Generic limits (Abildgaardia-Fimbristylis; Crosslandia) and contentious species limits (i.e. variation in Crosslandia setifolia, Bulbostylis pyriformis-B. hispidula complex, and the B. densa complex) required selection of particular taxa with the focus primarily on Australian material. In addition, putative new species (i.e. Bulbostylis sp. aff. barbata, Fimbristylis sp. aff. odontocarpa, Abildgaardia sp. aff. schoenoides, A. sp. aff. pachyptera), possible new combinations (i.e. Abildgaardia vaginala, Fimbristylis spiralis, F. macrantha, F. oxystachya, F. pachyptera), or taxa to be reinstated (A. schoenoides) were included in analyses to test species limits. The large number of species that are assigned to Bulbostylis and Fimbristylis prevented a comprehensive assessment of either genera. The study of species for Bulbostylis was restricted to those species that occur in Australia, including unnanled Australian collections, and some representative overseas species (i.e. 15 taxa, 10 occur in Australia). Nine representative species of Fimbristylis were selected from five ofKem's (1974) sections (excluding section Abildgaardia); sections ｆｩｭ｢ｲ ｳｾｹｬｩ Ｌ Fuscae, Leptocladae, Tenerae, and Trichelostylis. Fimbristylis depauperata R.Br. (= F. dichotoma (L.) Vahl subsp. depauperata (R.Br.) J.Keln) was selected to represent the TYPE section of Fimbristylis, Fimbristylis section Fimbristylis for cladistic analysis. Members from Fimbristylis section Fuscae were selected due to the history where species from both Fimbristylis sections Abildgaardia and Fuscae were sometimes combined in the one section due to the 13 distichous arrangement of the glumes. The remainder of the species (F. schultzii, F. blakei, F. sp. L, F. furva, and F. microcarya) were included in an attempt to capture some of the sectional variation not yet sampled. Specirnens were identified using regional floras (Rye 1992; Wilson 1993) and keys (Sharpe 1986; Latz 1990). Regional keys have obvious shortcomings when species occur outside the regional range. To offset this problem and to ensure correct assignment, i.e. when species limits were being assessed, Australian n1aterial was compared to TYPE specimens where possible. An extensive and all-inclusive global assessment of the tribe Abildgaardieae was not possible due to the large number of species that fall within the current circumscription (Chapter 1). In an attempt to accommodate the global shortcomings of a restrictive national study, overseas specimens were included in analyses when the range for a species was extended (i.e. Bulbostylis densa), cosmopolitan (i.e. Abi/dgaardia ovata, Bulbostylis barbata,) or previously unreported in Australia (i.e. Bulbostylis humi/is). Keys developed for specific countries or regions provided a basis for specimen identification of taxa on loan from international herbaria (Kral 1971; KetTI 1974; Haines and Lye 1983; Adams 1994; Gordon-Gray 1995). Not all taxa used in the cladistic analyses were included in the phenetic studies. Lists for specific taxa used in phenetic and cladistic analyses can be found in the relevant sections of the chapters dealing with generic groups, i.e. Crosslandia (Chapter 3), Abildgaardia (Chapter 4), Bulbostylis (Chapter 5), Nemum, Nelmes/a, and Tylocarya (Chapter 6). The complete list of sampled taxa and specinlcns can be found in Appendix 1. A list of the more than 4400 loan specimens plus specimens 14 personally inspected at herbaria (BRI, CANB, DNA, MEL, MBA, NS\V, NE, QRS, and PERTH) during the course of the study may be obtained upon request. Sampling Dried herbarium samples provided the basis for scoring morphological characters. To assess variation in plant morphology, ten specimens (where available) were selected to cover the geographical range of species when species had well-defined boundaries. Taxa with uncertain limits required a larger sample size to encompass the greater variation and thus define the entities (e.g. Crosslandia). Duplicate sheets were examined where possible to ensure a more thorough sampling, especially for species that exhibit polYmorphism. Whenever possible, the collections chosen were whole plants that were in good general condition, with mature fruit, plus multiple culms and leaf blades (if present). If an individual specimen had all the necessary material to score all states, only that individual was sampled (i.e. single collected specimen on a herbarium sheet). Sometimes scoring from other individuals collected from the same site was needed - either from the same sheet or from duplicate specimens. When multiple individuals were scored, care was taken to ensure that all were the same taxon, as some sheets contained mixed taxa. If a duplicate sheet was required for sampling, the herbarium code and sheet information were included in the table of taxa (Appendix 1). Phenetic studies Exploratory pattern analyses were undertaken using the program PATN v 3.6 (Belbin 1993). The more recent PATN (for Windows) v 3.03 (2004) is limiting in that only a single similarity coefficient is allowed, preventing the use of merged data 15 sets and character weighting. In addition, Principal Components Analysis is not available, further limiting pattern exploration. Taxa for phenetic analyses Phenetic analyses were used to determine species boundaries. During assessrnent of species for a given genus, Operative Taxonomic Units (OTUs) forming welldefined species were selected as reference taxa. For example, in the analyses of species for the Crosslandia group, specimens from Abildgaardia or Bulbostylis that could not be assigned to a currently accepted species were excluded fron1 the Crosslandia analyses. Species lists for sampled specimens used in phenetic analyses are found in the relevant data chapters 3, 4, and 5 (see also Appendix 1 for specimen information). Characters for phenetic analyses Phenetic studies were based on morphological data. Both qualitative and quantitative characters were used in phenetic analyses. Qualitative characters consisted of both binary and multistate forms. Binary characters were scored as either present or absent (0/1) for the two states (attributes) of the character. Multistate characters, where more than two states were observed, were converted into individual presence/absence type data (i.e. binary form). Preparing multistate characters in this manner prevented loss of infonnation when a character had more than four states (Crisp and Weston 1993). Binary characters were given the weight value of ｬｾＬ while multistate attributes were weighted according to the nUll1ber of states (columns) per character, i.e. each character was given a weight value of 1. For example, a character with four states would have each attribute (column) assigned a 16 weight of 0.25. Attribute (column) weight values were updated with each phenetic analysis following the addition or exclusion ofOTUs. Removing OTUs to reanalyse subsets usually created invariant columns that were removed from the data set, altering the number of states per character and therefore the weight for each attribute column. Polymorphic characters were restricted in the phenetic analyses; hO\\Tever, when included these states were presented in separate colulllns of binary forn1 (0/1) and given a ,:veight of 0.5 for each of the two attribute columns formed. Two examples of polymorphic characters were variation in stamen number (e.g. in some species of Bulbostylis) and the presence or absence ofprophyllar buds within an inflorescencesYllflorescence (i.e. secondary floral growth from the axil of the prophyll as seen in some species of Bulbostylis). Most of the species under study exhibit regularity in the number of stamens per floret and the presence/absence of prophyllar bud growth; however, the observed variation required a realistic assessment in these otherwise stable characters. Quantitative characters consisted of measurement data that, where available, were the mean of at least five values. This sampling strategy was applied to prevent destruction of specimens, especially in species with few mature spikelets or ripe fruits. Loose material from herbarium sheets was scored when available in preference to removing nuts and glumes in situ. TYPE specimens were rarely included in analyses and when it was necessary (e.g. Fimbristylis odontocarpa S.T.Blake), only loose nuts and glumes were scored from the sheets. All measurements were recorded in millimetres for consistency. Some data were converted into ratio coefficients or proportions, i.e. glume (floral bract) l/(\V'idth!length), l/(nut stipe length! nut length) to provide additional information to 17 individual length or width measurements. Converted values were kept to a minin1Llln, and so do not dominate the analyses. Lists of the attributes used in phenetic analyses can be fc)und in the relevant chapters. Analysis preparation Data were stored within Microsoft Office Excel and files saved as comma delimited files (* .csv) for direct use within PATN. To accommodate the nature of mixed data sets, various types of data lnanipulation and association measures were tested for robustness of the results. Non-weighted characters were analysed and compared to separate analyses where characters were weighted. After setting parameters and prior to producing association matrices, characters were weighted using the data transformation and standardisation module (TRND option 10, under 'Manipulation'). The application of various coefficients to different types of data matrices produced association matrices. Gower's General Similarity coefficient, known to be useful for binary, multistate and quantitative data (Stuessy 1990), was used here for data matrices combining quantitative and qualitative data, and quantitative Inatrices used in 'merged' analyses. Qualitative matrices used for 'rrlerged' analyses \vere subjected to the KulcYllski coefficient which, although not generally used in phenetic analyses, is good for presence/absence data where polarity is expected (Crisp andWestol1 1993); % matches are not considered important (Belbin 1993). Data matrices analysed were grouped as: 1. whole data set, columns not weighted (cols = 1); ASO = Gower Similarity Coefficient; 18 2. whole data set, columns weighted (chars = 1); ASO = Gower Silnilarity Coefficient; 3. quantitative data set, columns not weighted (cols = 1); ASO = Gower Similarity Coefficient; 4. qualitative data set, columns not weighted (cols = 1); ASO = Kulcysnki Coefficient; and 5. qualitative data set, columns weighted (chars == 1); ASO = KuIcynski Coefficient. Data sets 3, 4 and 3,5 were merged within PATN (under 'Manipulation', Leftright & up-down merging). Merging data sets containing different types of data and using different association measures is more reliable than applying different association measures to one data set (Belbin 1993). The two association matrices were then standardised and added together in the association transforming and standardising (TRNA) module, located under 'Manipulation' option 9. The lowest stress values (in the ordination module) were obtained from merged data sets, however, weighted quantitative and qualitative data (option 2 using the Gower similarity coefficient) resulted in the same group formation of OTUs. Although it was necessary initially to explore differences between treatrnent of data and association coefficients, there was no great advantage in the time taken. Therefore, option 2 (whole data set, columns weighted i.e. characters =1, using the Gower metric similarity coefficient) was applied to all the final phenetic analyses found in chapters 3, 4, and 5. 19 Association matrices were then subjected to ordination, cluster and network analyses within PATN. Analyses Phenograms from cluster analyses are generally useful as rough guides to taxonomic structure based on similarity/dissimilarity rneasures, with more accurate and useful results obtained from ordination approaches based on multidimensional scaling (Stuessy 1990). Phenograms portray close 'relationships' rather than the distant 'relationships' of ordination diagrams (Stuessy 1990) based purely on the similarity of taxa and are best interpreted with no evolutionary foundation. Ordination, cluster, and network analyses were applied to association matrices. Combining these three techniques provides a thorough assessment for similarity/dissimilarity of taxa (Belbin 1993). Ordination Ordination reveals 'real' groups based on the underlying pattern in the data, whereas in cluster analysis taxa are forced to form groups. Ordination plots have been used as the basis for group formation (usually species), as emphasis is on the greater the distance between clusters the greater their dissimilarity. Cluster and network analyses were compared to ordinations to assess robustness of the data. Both Multidimensional Scaling (MDS) and Principal Components Analysis (PCA) were applied to association matrices. Gower's coefficient is an interval measure and not appropriate for Semi Strong Hybrid (SSH) multidinlensional scaling as SSH techniques apply to ratio data only. Therefore, care must be taken to select the 'interval' and not 'ratio' option when applying the ordination procedure. Crisp and 20 Weston (1993) performed ordinations using both interval and ratio strategies to avoid any loss of information, a practice followed here. However, only scatter plots using the 'interval' strategies were presented in the results. Gower's coefficient requires that association values be unimodal if the coefficient is to be a valid measure for the data (Belbin 1993). In addition, an appropriate ratioordinal cut value within the SSH module must be determined by the type of data analysed. For example, ordinal data should have a cut value less than the mininlum value within the association matrix, and interval data must have a cut value greater than the highest association value (Belbin 1993). Selecting 'HIST' from the analyses module (option 12) enables assessment of modality and maximum/minirnum association values. When combined quantitative and qualitative data showed bimodal histogran1s of association values, the data sets were rerun as separate matrices and merged. The resultant histograms of merged association values were usually unimodal. Principal Components Analysis (PCA) using Gower's similarity coefficient was compared with MDS of combined data, and merged quantitative and qualitative matrices that were suitable only for use with MDS type ordinations. Ordinations were run in two and three dimensions to achieve the lowest stress values, an indication of the level of best fit of the data to the number of axes used. Those that were 3-dimensional usually achieved acceptable recommended stress values (around 0.1), while higher stress values (0.15-0.17) were frequent in 2dimensions. Often the groups formed were similar (robust) for the various dimensions (2 or 3) and whether interval or ratio method was applied to the ordination. All 3-dimensional ordinations were observed using Statistica version 5.1B (Statsoft 1996); this program allows for rotation of the points around the axes and visual conformation of the groups formed. For ease of presentation, scatter plots 21 frOlTI 2-dimensional ordinations (often with borderline stress values) were presented in the results sections when grouping was similar to the more robust 3-dimensional ordinations. When 2-dimensional groups did not clearly represent the points in 3dimensional space, group borders were drawn onto the xy scatter plot to reflect the groups observed in the xyz plot. The stress values for both 2-dimensional and 3dimensional ordinations are given for comparison in the results section for each data chapter. Two-dimensional ordinations were plotted using Microsoft® Excel 2000 (Microsoft 1999). Table 2.1 Ordination stress values. Kruskal's goodness of fit associates stress levels to how well the data fit the number of axes used in any given ordination. From Belbin (1993 p: 133 of the 'Technical Reference'). Stress value Goodness of fit > 0.2 0.15> 0.2 0.1>0.15 0.05 > 0.1 < 0.05 poor be cautious fair (wish it were better) satisfactory . . ImpreSSIVe perfect! o Classification Clustering strategies may provide differing results depending on both the data and strategy selected. Various strategies were performed and compared with each other and the ordination scatter plot for each data set. Flexible unweighted pair group arithmetic averaging (UPGMA) or flexible weighted pair group arithmetic averaging (WPGMA) provided phenogrmTIs that most closely resembled the groups found in ordination space. Equal weight is given to objects (not groups) in the UPGMA strategy so that during the fusion process groups are 'weighted proportionally to the number of objects contained within each group. In 22 contrast, groups are weighted equally regardless of the number of objects in the flexible WPGMA fusion strategy (Belbin 1993). Assigning beta values allows for attribute space distortion, either contraction or dilation depending on the assigned sign. A beta value of -0.1 was used for UPGMA or WPGMA strategies, as negative ｾ is reported to aid known partition recovery (Belbin 1993), although changing the ｾ value did not alter OTU partitioning in these analyses. Both strategies frequently produced similar groups for OTUs with minor variations, however, the strategy closest to the groups formed in the ordination scatter plot was selected for presentation. Network analysis Minimum spanning trees (MST) from the network module accurately represent close objects/neighbours, with uncertainty proportional to the increase in object separation. Minimum spanning trees compliment the ordination where the greater the distance between objects the lower the affinity of those objects. When used in conjunction with the ordination, MST connections can confirm or refute close 'relationships' (Belbin 1993), i.e. of similar/dissimilar groups. Minimum spanning trees were not presented for the larger data sets due to their complexity, although they were examined and compared to the ordination scatter plot and phenogram for general OTU group robustness. Evaluation Attributes with the greatest influence for a given ordination were evaluated using the principal axis correlation (PCC) module by selecting ,SCAT' from the ordination menu, under 'evaluation'. Attributes are fitted to an ordination space using multiple- 23 linear regression (Belbin 1993). Those character states (attributes) with ＷＰＭＸ ｾＧｏ (depending on the number of attributes) or greater influence on the ordination were plotted to link with the matching scatter plot. Character states influencing cluster analysis grouping in the phenogram were firstly extrapolated using group definition (GDEF) and then box and whisper (OSTA) module applied (neither presented). The latter uses the non-parametric Kruskal-Wallis statistic for quantitative (continuous) data (option 1) and constancy percentages for nominal (presence/absence) data (option 2). Operative TaxonOlnic Units could be forced into larger groups using ODEF to observe major group boundaries and the attributes forcing the OTUs into the groups investigated. The attributes found in the evaluation module for classification were the same as those associated with the ordination. Therefore, only results from the PCC are presented. For any given data set, all taxa and characters were included in the initial analyses. Subsets of the first analysis were then rerun for unresolved groups, or finer analysis of the larger groups (i.e. generic groups formed) to assess species grouping. Invariant columns (character states) were removed prior to reanalysing. Characters that may have had undue influence in an analysis (i.e. presence of basal spikelets or amphicarpy) were removed and rerun to test the robustness of the analyses. Ordination scatter plots and cluster analysis phenograms provided groups used as 'terminal taxa' within cladistic analyses. 24 Cladistic studies Analysis preparation For cladistic analyses, data were collated from the multiple OTUs scored in phenetic analyses so that each defined group of OTUs was converted into data for each terminal taxon. Ingroup Ingroup taxa were based on the tribe Abildgaardieae as outlined by Goetghebeur (1986) and Bruhl (1995). All genera were included in the tribal analysis (Chapter 6). When species level assessment was required, subsets of genera and species were analysed initially as smaller groups (Chapters 3, 4 and 5). Outgroup Data were polarised using the outgroup method (Maddison and Maddison 1992). Some members from the provisional tribe 'Arthrostylideae' (Goetghebeur 1986), i.e. Arthrostylis (Queensland), Actinoschoenus (Northern Territory), and Trachystylis (Queensland), plus species from Schoenoplectus (Reich.) Palla and Schoenoplectiella Lye (tribe Scirpeae or Fuireneae) (Appendix 1) were selected as outgroup taxa based on the sister relationships with the Abildgaardieae (Bruhl 1995). As taxa from the 'Arthrostylideae' have been placed variously within genera of the ingroup, (Clarke 1908; Kern 1974; Latz 1990) species of Schoenoplectus provided some distinct character differences to polarise the data. 25 Characters Character states used in the phenetic analyses were converted into data suitable for cladistic study within the DELTA Editor v 1.04 (Dallwitz et al. 1999). Quantitave characters were included to provide additional infonnation for phylogenetic assessment (Poe and Wiens 2000), especially at the species level (Gonzalez-Elizondo et al. 1997). Quantitative characters were gap coded using a line graph to find gaps in the OTU mean data from the phenetic analysis to define the number of states, which were scored as ordered multistate characters within the DELTA data set. All other characters were scored as unordered. Spikelet width measurements were omitted fronl cladistic analysis and used for descriptive purposes only (not included), as gaps within the measurement data were not clear-cut. Although quantitative characters may be polYmorphic, the potential for added infonnation has been documented (Poe and Wiens 2000; Wiens 2000) and were therefore included in the present study. Polymorphism Taxa, mostly species, which exhibit a variable range across character states for a particular character, occur frequently within the study group, particularly in Bulbostylis and Fimbristylis. These polYmorphic characters present problems at a cladistic level where discrete states of homologous characters are required as a basic assumption in a phylogeny for any given group of taxa (Hennig 1979). There are numerous ways to code for polYmorphism (Wiens 2000), all producing different tree topologies, however, excluding polYmorphic characters and employing only fixed states provided the poorest tree topology of all methods when tested by Wiens. Breaking down the units of the species into separate samples is the best way to show polYmorphism within a species, as polYmorphism represented by one collated species 26 representative (n= 1) cannot be detected in an analysis (Wiens 2000). In PAUP* (Swofford 2001), only one state of a polymorphic character is specifically assigned in an analysis, thus affecting parsimony, and the resultant tree topologies. Data for species in this study were presented as a single line within the DELTA data set rather than multiple individual samples due to the large number of species being analysed and time constraints. Polymorphic characters were included in this study despi1e loss of some information during analyses, as many taxa were observed to be constant for some of these characters (e.g. inflorescence-sYnflorescence type; presence or absence of hairs on culms, leafblades, or glumes; style indumentum; etc.). Characters were given equal weight, as is the default option in PAUP* (Swofford 2001). Although a larger number of character states may give undue weight to a character, it was not possible to scale character states to provide differential weights, as the many odd numbers of states prevented real scale values (whole numbers) to be produced for use within PAUP*. Characters from embryo morphology plus leafblade and culm anaton1y were added to the data set at the terminal taxon stage (see Appendix 2). Leafblade and culm anatomy Anaton1ical characters were included in cladistic analyses as Metcalfe (1969, 1971) showed that the anatomical pattern found within the Cyperaceae on the whole might reflect generic relationships and provide an indication of tribal grouping. :'vfore recently, vascular bundle types, specifically from leaf blades, have been described and associated with specific photosYnthetic pathways (generally C 3 and C 4 ) (Bruhl et a1. 1987; Bruhl and Perry 1995; Soros and Bruhl 2000; Soros and Dengler 2001) and related to genera. In addition to the ancestral C 3 -type of anatomy, the Fimbristylis- 27 type was the only one of the four types of Kranz anatomy (C 4 ) described across Cyperaceae that applied to taxa within this study (Figure 2.1). PhotosYnthetic pathways were scored from leaf blades, or culms when plants were leafless. Although there are differences between leafblade and culm anatomy, the type of photosYnthetic pathway could still be assigned easily. Anatomical structure may vary with environmental influences, or be useful only for diagnostic purposes (Metcalfe 1971). However, generalleafblade and culm shape, vascular bundle arrangement, and photosYnthetic pathway are usually constant. Following Metcalfe's work, other anatomical features, such as shape of sclerenchyma and mesophyll chlorenchyma, number of vascular bundles, and the arrangement of vascular bundles and sclerenchyma within the organs, were included as potentially useful traits. Culm and leafblade anatomy were sampled from the mid-third area for three specimens from each species (where possible). Fixed material was used when available, or rehydrated' green' herbarium material provided satisfactory sections in most cases. Selected organ segments from dried herbarium sheets were placed in a beaker of cold water with a few drops of detergent, heated until boiling point was reached, then removed from the heat and allowed to cool. These segments were then ready to cut by hand using a double-sided razor blade, or were stored in 700/0 ethanol until needed. Hand cut sections were stained initially with Bismarck Brown, however, better tissue definition was obtained using the Astra Blue-Basic Fuchsin double-staining technique of Kraus et al. (1998), omitting bleaching, acetic acid pre-stain rinse and picric acid differentiation (picric acid has explosive properties when dry). Good C4 C3 NADP-ME rhynchosporoid o • • chlorocyperoid NAD-ME fimbristyloid o Primary Carbon Assimilation (PCA)/mesophyll Vascular Tissue Photosynthetic Carbon Reduction (PCR) Mestome Sheath Metaxylem Element - Suberin Lamella Parenchymatous Bundle Sheath eleocharoid Ontogenetic origin of PCR EIl Ground meristerm E88a Procambium Figure 2.1 Schematic representation of the 'type' of variation in the photosynthetic pathway that correlates with the arrangement of tissues within vascular bundles in Cyperaceae (from Soros & Bruhl 2000; Soros & Dengler 2001). Photosynthetic pathways that apply to this study are C 3 and C4 funbristyloid. 29 tissue differentiation of sections was obtained without these extra steps. Sections were then mounted for microscopic examination as semi-permanent slides using clear glycerine jelly (see Appendix 3 for recipe). Once familiar with the tissues, observations could be made without the use of stains and hand cut sections were mounted directly into 50% glycerol for speedier microscopic examination. These sections could be made permanent at a later date. Root anatomy was initially examined for useful characters, and although there appeared to be potential for some characters such as the endodermis shape and tissue layers, time constraints prevented a comprehensive assessment. These root characters were therefore eliminated from the analyses. Characters taken from anatomical studies were used in cladistic and not phenetic analyses. Embryo morphology MicrOlnorphology of the mature embryo has been widely used since Van der Veken's work in 1965 as an aid to assigning taxa to genera, or at least, to exclude taxa that do not fit within a given embryo-type group (Gordon-Gray 1971; Kenl 1974; Goetghebeur 1986). Van der Veken (1965) sampled embryos across the Cyperaceae, describing six main embryo types. Although the same embryo type can be be found among different genera, Van der Veken noted that different embryo types do not occur within a given genus. Embryological features, such as development, general size and shape, plus position of the primordial root and shoot, are known to be relatively stable and therefore taxonomically useful (Maheshwari 1964; Davis 1966; Johri 1992). 30 Sampling one or two embryos from a specimen can reflect the embryo luorphology of the whole plant, especially when compared with specimens of the Salue species. Selecting three specimens (where possible) across the range of each defined species group increased the reliability of the limited sample size (when compared to morphological characters). For consistency, embryo morphology was sampled for the same taxa used in the anatomical and scanning electron microscopy studies. Characters from embryo morphology were included in the cladistic studies only. Embryos were selected from wet (fixed material) or dried material (rehydrated as in anatOluical work) if the former was not available. Fruits were carefully dissected to release the embryo from the proximal portion of the ovary. To observe primordial root and shoot orientation for scoring, embryos were cleared. Chlorolactophenol was used initially as the clearing medium. Phenol is included as a preservative; however, this was later omitted, as it has known carcinogenic properties. In addition to increasing the refractive index of the medium, chloral hydrate also has preservative properties; the need for phenol as a preservative was redundant. The Chlorolactoglycerol (glycerol 10 mL, distilled water 10 mL, lactic acid 10 mL, chloral hydrate 1.6 g) was used to clear the remainder of the embryos. Whole embryos were placed directly into clearing fluid in a single cavity slide or double cavity slide, if small enough, and a no.1 coverslip placed over the cavity. It was necessary to make deep well slides to accommodate the larger embryos for species from genera such as Abildgaardia and thus allow manipulation of the elubryo during microscopy. Deeper well slides were made by affixing five 22x22 mm (no. 1) coverslips atop each other with Eukitt®, so that the coverslip piles on each side 31 partly covered the cavity of a single cavity slide. This created a slnaller, deeper area in which to place the larger-sized embryo. Prepared slides were left to dry for two days in a 40°C oven prior to adding the clearing medium, embryo, and final coverslip. It was often necessary to add additional clearing fluid from the front or back gaps to ensure that bubbles were removed from the cavity area before mIcroscopy. Frequently, embryos would not clear and remained clouded, obscuring the inner areas of the embryo (i.e. primordialleafblades and vascular tissue). Neither bleaching the embryos prior to clearing, or lengthening the time in the clearing medium was helpful. For this reason, some embryos were embedded using the paraffin wax method and sectioned using a rotary microtome. Wax embedded serial sections, which were prepared according to Johansen (1940), allowed me to become familiar with the embryo structure prior to assessing cleared material. These sections were stained with Safranin-O and Fast Green and made permanent by mounting in Eukitt®. Cleared embryos were best observed soon after the clearing medium was added. The longer the embryos were kept in the clearing fluid, the less visible were the inner areas as starch bodies burst and released their 'oily' contents, obscuring the young organs. Extra care was taken if embryos were kept in the clearing fluid for extended periods as the embryo tissues became very soft and were easily damaged when moved under the coverslip. It was difficult to assess the presence of second or third primordial leaves, especially in the smallest embryos where the tissues contain dense cytoplasm. There was inconsistency for scoring second and/or third leaf data, therefore these data were omitted. Experimenting further with various pre-treatrrlents 32 and clearing media is necessary to obtain greater definition and detail in embryos across the range of taxa. Embryos were scored according to the general embryo types outlined in Haines and Lye (1983) (Figure 2.2). Once familiar with the embryo structure, elllbryos were scored using a compound microscope with the stage diaphram shut down to increase contrast. Scanning electron microscopy Nuts n1ay exhibit surface patterning with: pits, tubercules, ridges and undulations; trichomes of various complexity; spines; or sometimes secondary or tertiary sculpturing (Lye 2000). Variation of the epidermal sculpturing has been a useful taxonomic character at the species level (Haines and Lye 1983; Gordon-Gray 1995), and conformity of cellular shape and arrangement has had some use at the generic level (Goetghebeur and Coudijzer 1984). Nuts selected from dried herbarium specimens for exarnination of anatomy and embryo morphology were prepared for scanning electron rnicroscopy (SEM) to provide comparable detail of the nut epidermis (see Appendix 1 for sampled specimens). When fruit availability was limited (i.e. from international loans, TYPE specimens - with prior permission, or when material was very scarce), some nuts used for SEM were rehydrated and the embryo dissected for embryo morphology. The success of rehydration depended on the condition of the nut prior to gold sputtercoating for SEM. One to three nuts (depending on size and availability) were affixed to stubs using double-sided tape and sputter-coated with gold for four minutes in the Polaron gold first leaf ", "/-0-" plumular ｾ ...ｾＬ ...:. - - ｾ indentation Schoenoplectus \ Car.ex type type ? °t plumular ..../.... strand B Cladium type E Eleocharis type ('-A-'l t_l:J_; Bulbostylis type Figure 2.2 General embryo types of the Cyperaceae (adapted from Haines and Lye 1983 after Van der Veken 1965). The Fimbristylis-, Schoenus-, Bulbostylis-, Carex- and Schoenoplectus-types as pictured were found in this study, plus the Abildgaardia-type (not pictured), that shares the same primordial shoot and root orientation as Bulbostylis-type, but is distinctly larger, has a broader cotyledon, and a well-developed second primordial leaf. 34 sputter-coater E51 00. Scanning electron microscopy was performed using a lEOL lSM-5800LY Scanning Microscope at 20 kY. Images were saved to disc in Tagged Image Format files (* .tit). Some rachillas, styles and pollen were also assessed using SEM. Problems obtaining resolution at higher magnification prevented detail of style papillae and pollen features to be captured. Scoring for these characters was therefore abandoned. Injlorescence-synjlorescence homology It is necessary to ensure that the characters scored are homologous if a given tree topology is to be accepted as the best fit for a given data set. Penet et al. (2005) reported that the monosulcate pollen grain found throughout the Asparagales was not homologous; they observed different developmental pathways during cytokinesis in the pollen tetrad of different genera. Scoring pollen simply as monosulcate would be misleading in a cladistic analysis. A similar situation occurs within the Cyperaceae sYllflorescence, where the commonly termed 'head' of sessile spikelets is not homologous across the species sampled. Differences in branching and arrangelnent of sessile spikelets that form a terminal 'head' was explored to ensure homology for inflorescence-sYllflorescence structure between scored taxa. Fixed or rehydrated sYllflorescences were examined for all species that exhibit 'heads' of sessile spikelets. Lateral branches were determined by the position of the prophyll that was always present within the sYllflorescence in all but solitary spikelets where only the terminal florescence was present. InflorescencesYllflorescence terminology follows that of Weberling (1989) and more specifically, 35 the anthelodium and paniculodiun1 synflorescence structures outlined by Vegetti (2003). Analyses A nexus file was generated from the DELTA data set from within DELTA to perform maximum parsimony analysis within PAUP* 4b 10 (Swofford 2001). The large number of taxa to be analysed prevented exhaustive searches (for less than 11 taxa), or branch-and-bound analysis (up to 22 taxa) that guarantee to find all the shortest trees for a data set (Swofford 2001). Smaller data sets were analysed initially to explore speciation of some groups prior to assessing all members of the tribe for monophyly. However, the data set with the smallest number of taxa analysed was still too large for exhaustive or branch-and-bound methods. Heuristic searches for optimal trees for simple and random addition-sequences of 10-2000 replicates using tree bisection-reconnection (TBR) branch-s\vapping, holding 1, 5, 10 or 100 trees at each replication were evaluated. Taxon order within DELTA data sets was also manually randomised (working with a newly saved .file to avoid program bugs corrupting the original data set) and rerun using simple and random addition-sequences as outlined above to explore further optilnal trees. Trees of the same tree length were recovered in all test analyses, the difference between analyses was that number of most parsimonious trees increased as the additionsequence replication was increased. Also, manually randomising taxa within DELTA prior to generating the nexus file, increased the number of trees found within an analysis, compared to data sets where the taxa were not manually randomised. Heuristic searches were performed using 1000 random addition-sequence replicates, holding five trees at each step, on data sets where the taxa were manually 36 randomised, thus saving computational time while optimising retrieval of the nun1ber of shortest trees. Evaluation Bremer support values and bootstrap frequencies are included with the relevant cladogram of each analysis, as different aspects of group support are obtained by the two methods (Ramirez 2005). Bootstrap analysis (Felsenstein 1985), where 'taxa are held constant and characters sampled with replacement to build a series of new data sets the same size as the original' (Swofford 1991 p: 62 PAUP 3.1 Users Manual), was used to provide statistical confidence to the relationship hypothesis provided by the heuristic search. Bootstrapping (100-1000 bootstrap replicates) was performed based on la, 100 or 1000 random addition-sequence replicates to assess the variation of results. Bootstrap analysis using 10 addition-sequence replicates and 1000 bootstrap replicates provided comparable values of support against analysis using 1000 random additionsequence replicates, but with much less computational time; Bootstrap values presented in Chapters 3, 4, 5, and 6 are based on 10 random addition-sequence replicates. Branch support was evaluated as >50<70% was weak support, >70<850/0 indicated moderate support, and >850/0 was strong support. Values that were <500/0 indicate no support and the values omitted. The majority-rule consensus constructed from the bootstrap replicates must be considered against the assumptions that the characters are independent and are representative of all the characters. On this basis, the non-statistical Bremer decay values were also used to assess the level of branch support. 37 Bremer support (Bremer 1994) or decay analysis, was obtained by expanding the tree length of the shortest trees(s) obtained from parsimony analysis by 1 step (s + 1, s+2, s+3, s+4, etc.), while maintaining the settings for the original heuristic search. A tree fron1 strict consensus determined the level of collapse or stability of branches, adding a value of 1 for each stable branch that remained after each analysis. Increasing the tree length by an extra step for each subsequent run and examining the strict consensus, provided the branch support values that were used in conjunction with bootstrap analysis. Usually five runs, increasing the tree length up to five steps, were sufficient to collapse most ifnot all of the main internal branches. As the number of taxa grew in subsequent analyses, the heuristic searches required too much computational time and were abandoned after the second or third extra step. BreIner support was not presented in Chapters 5 and 6 due to computational difficulties; Bootstrap support alone was presented on the cladogran1s. MacClade v 3.08 (Maddison and Maddison 1992) was used to trace characters on trees and explore branching. TREEVIEW v 1.6.6 (Page 1996) allowed trees to be viewed and saved into a format appropriate for presentation within this thesis. Photomicroscopy Images from anatomical and embryo morphological studies were initially captured using a Nikon Coolpix 990 digital camera attached to a Leitz Larborlux S con1pound microscope. More recent image capture, including phase contrast microscopy of embryos, was performed using an OlYmpus BH-2 compound microscope with the Nikon Digital Sight camera control unit (DS-L1), DS-5M camera head attached. 38 Lower magnification images were obtained with the same Nikon Digital Sight system attached to the WILD Photomakroskop M400. 39 Chapter 3 Crosslandia W.Fitzg.: a phenetic and cladistic study Introduction This chapter focuses on the specific delimitation of Crosslandia ｳ･ｴｾｦｯｬｩ｡ W.Fitzg., the provisional C. anthelata Goetgh., Fimbristylis spiralis R.Br., and Abildgaardia vaginata R.Br. and their generic placement. Fitzgerald's (1918) description of the monotypic genus Crosslandia, collected from Goody Goody Western Australia, is based on the capitate inflorescence stnlcture, male aerial floret, and the presence of female spikelets at the base of the plant. Crosslandia, therefore, is defined by male aerial florets and female basal spikelets. Hutchinson (1959), using Engler's system of classification, placed Crosslandia in the tribe Sclerieae, based on the male spikelets in capitate inflorescences on long slender peduncles, and numerous female basal spikelets among the leaves. While Fitzgerald (1918: p123) had noted in his protologue that 'the plant bears a close resemblance to some of the capitate Schoeni' , he den10nstrated the affinity Crosslandia setifolia has with members of Fimbristylis through nut and style features, and highlighted the differences for generic separation. Anatomical studies undertaken by Metcalfe (1969, 1971) revealed that the leaf and cuhn anatomy of Crosslandia setifolia (H. S. McKee 8432; Blain, Northern Territory) were similar to species of Fimbristylis, especially in the arrangement of vascular bundle sheaths. Crosslandia was subsequently placed nearer ｆｩｭ｢ｲｳｾｶｬｩ within the tribe Scirpeae (Hooper 1973). In Raynal's (1973) reconstruction of 40 phylogeny for the genera of the Cyperoideae Crasslandia was placed on the sanle lineage as Eleacharis, Fimbristylis (including Abildgaardia), Nelmesia, Nemum and Eulbastylis. Goetghebeur (1986) observed that some Crasslandia material similar to the TYPE, i.e. having spikelets forming dense heads, had female florets distally within the spikelet. He also noted that the inflorescence structure in some specimens was 'anthelate' (having one spikelet per ray), with some of the 'anthelate' rnaterial bearing bisexual aerial florets rather than male florets within the spikelets. Specimens with the 'anthelate' inflorescence, as distinct from the 'capitate' sOli, were given the provisional name Crasslandia anthelata. Goetghebeur, working on a broad study of Cyperaceae from Europe, had limited material of Crasslandia and was, therefore, not able to capture the full variation of the species at that time. The provisional name was never validated and a later treatment of Cyperaceae by Goetghebeur (1998) in 'The Families and Genera of Vascular Plants' did not mention the variation he had found previously. Fimbristylis spiralis is the only member of the genus Fimbristylis bearing female basal spikelets (Latz 1990), and apart from the spirally arranged glumes, seems to share with Crasslandia setifalia similar morphology for general habit (although smaller), basal spikelet shape and nut characters. The original collection by Brown was made fronl the remote Arnhem Bay (north-east Arnhem Land). No mention of the basal spikelets was made in Brown's (1810) protologue. Bentham (1878) too did not refer to basal spikelets in the description of F. spiralis in 'Flora Australiensis'. All collections, including the TYPE specimen, were made from the north-east to east coastal edge of Amhem Land, Northern Territory. The only other known collections are from Groote Eylandt, Northern Territory, R.L. Specht 235 (MEL 2048472, 41 CANB); Rose River, Gulf of Carpentaria (as '?Crosslandia setifolia'), Northern Territory, C.R. Dunlop 2957 (DNA 36442); and Cape Shield, Blue Mud Bay, ".JT, G.J Leach 3601 & I.D. Cowie (NSW 422184, CANB 478018). During this study, I observed female basal spikelets, at various stages of development, in herbarium specimens of Abildgaardia vaginata. These basal spikelets have a similar morphology to those found in both Crosslandia and Fimbristylis spiralis. Basal spikelets are not consistent with the generic delimitation of Abildgaardia and have not been noted previously in the literature. In addition, the nut of A. vaginata differs from A. ovata and A. oxystachya, A. macrantha, A. pachyptera, and A. schoenoides provisionally placed within Abildgaardia by Goetghebeur (1986); the nut resembles those from Crosslandia and Fimbristylis spiralis. Goetghebeur did not list A. vaginata with the species in his treatment of Abildgaardia. Abildgaardia vaginata R.Br. has a chequered nomenclatural past. Blake (1947) noted that Bentham transferred A. vaginata to Fimbristylis Section II Abildgaardia, with the name change to Fimbristylis brownii Benth., a move that seems illegitimate. The name F. vaginata (Boiv. ex C.B.Clarke) was not occupied until 1895; in 1915 Domin made the combination F. vaginata from A. vaginata, however, the prior use of the name by Clarke in 1895 made the combination illegitimate. Bentham's F. leptoclada (based on two collections: Dallachy, Rockingham Bay in far north Queensland and O'Shanessy, from Rockhampton) in 'Flora Australiensis' (Bentham 1878), non F. leptoclada Benth. in Flora Hong Kong (Bentham 1861), \vas assigned to Section IV Trichelostylis, Series I Oligostachyae along with F. spiralis. Fimbrisitylis leptoclada is now a sYnonYm of Abildgaardia vaginata or Fimbristylis brownii, depending on the classification system used. Clarke (1908) placed 42 F. spiralis and F. brownii in Section Trichelotylis Series A: Oligostachyae based on the 3-fid style and solitary spikelets, while the multi-spikeleted F. lepotoclada (with few spikelets according to Bentham) was in the same section, assigned to Series B. The studies by Kral (1971), and Goetghebeur and Coudijzer (1985) indicated that Abildgaardia should have equal rank to that of Fimbristylis and Bulbostylis. This view has been accepted by cyperologists in Australia and the U.S.A., and some cyperologists in Europe. Simpson (1993) chose to retain the broader generic concept and, therefore, accepted Fimbristylis brownii as the correct name. New information has led to questions whether Fimbristylis spiralis and Abildgaardia vaginata would be better placed within the genus Crosslandia. Also, is the variation within Crosslandia setifolia consistent with Goetghebeur's (1986) proposed recognition of a new species, Crosslandia anthelata? Materials and methods Taxa To assess species limits of Abildgaardia vaginata, Crosslandia setifolia and Fimbristylis spiralis, and to see if they form a monphyletic group, specin1ens from Abildgaardia, Crosslandia and Fimbristylis were sampled (Table 3.1). Species limits were set using phenetic analysis. To define the taxa within Crosslandia, the relationships of the species were tested to assess monophyly using cladistic analysis. Abildgaardia oxystachya, A. pachyptera, and A. macrantha, provisionally nan1ed in the genus Abildgaardia (Goetghebeur 1986), were included with A. schoenoides and A. ovata in this assessment. 43 Twenty-three specilnens of Crosslandia setifolia s.1. (Table 3.1) were sampled across the geographic range (Northern Territory and Western Australia) to encompass the variable inflorescence morphology and floret sex within the species. Sampling for Fimbristylis spiralis was restricted to three known collections (see Table 3.1), excluding the TYPE specimen housed at BM. Fragments of the ISOTYPE fr01n Arnhem Bay, Northern Territory, R. Brown (KEW) were available from Queensland Herbariun1 (BRI 340661), although they were not suitable for sampling. The remoteness of the habitat (far north-east Northern Territory) prohibited field collections during the course of this study. Representative taxa from Fimbristylis were included to compare Fimbristylis spiralis species limits with other members of the genus. Species of Bulbostylis were included due to moven1ent of taxa between Abildgaardia, Fimbristylis, and Bulbostylis. Herbarium material on loan from herbaria BRI, CANB, DNA, MBA, MEL, NSW, and PERTH supplemented NE collections (Holmgren et al. 1990). Phenetic study Quantitative (21) and qualitative (54) morphological attributes were scored fron1 165 samples (Operative Taxonomic Units, OTUs), from Crosslandia, Abildgaardia, Bulbostylis and Fimbristylis for the initial analyses. Characters Character state definitions are mostly self-explanatory (Table 3.2), with the more complex characters detailed in the text when necessary. Inflorescence- Table 3.1 Specimens sampled as the focus group in the assessment of the genus Crosslalldia. The label corresponds to phenetic analyses. N.T. = Northern Territory, W.A. = Western Australia, Qld = Queensland, N.S.W. = New South Wales. See Appendix 1 for specimen details. State Collector Species Label Crosslandia setifolia N.T. Chippendale G. 1268 Cl W.A. Poulton G. 5 C2 W.A. Wilson K.L. 4885 C3 N.T. Craven L.A. 7928, Whitbread G. C4 N.T. Blake S.T. 17420 C5 N.T. Cowie I.D. 4639 C6 W.A. Pullen R. C7 N.T. Wilson K.L. 5260 C8 W.A. Wilson K.L. 4859 C9 W.A. Wilson K.L. 4803 C10 C11 N.T. Lazarides M. 8, Adams L. W.A. Burbidge N. 5703 C12 C13 N.T. Dunlop C.R. 6789 C14 W.A. Clarke K.L. 166, Bruhl J.J., Wilson K.L. N.T. Blake S.T. 16585 C15 W.A. Van Rijn P.1. 19 C16 N.T. Clarke K.L. 155, Bruhl J.J., Wilson K.L., Cowie I.D. C17 N.T. Dunlop C.R. 6854, Wightman G. C18 N.T. Dunlop C.R. 3446 C19 C20 N.T. Bruhl J.1., Hunter J.T., Egan J. 1268 N.T. Dunlop C.R. 3408 C21 N.T. Wilson K.L. 5150, Dunlop C.R. C22 N.T. Thompson H.S. 403 C23 N.T. Specht R.L. 235 F1 N.T. Dunlop C.R. 2957 F2 N.T. Leach G. 3601, Cowie I.D. F3 Qld Blake S.T. 15540, Webb L.J. Av1 Av2 N.S.W. Floyd A.G.F. AGF2205 Av3 Qld BrassL.1.18362 Av4 N.S.W. O'Hara J. 3472 and Coveny R. Av5 Qld Blake S.T. 8598 Av6 N.T. Cowie I.D. 6801 Av7 N.T. Brennan K. 2588 Av8 N.S.W. Bell D.M. Qld Forster P.1. PIF9732 Av9 Av10 Qld Forster P.I. PIF16257 Av11 Qld Blake S.T. 8222 Av12 Qld Blake S.T. 22499 Av13 Qld Brass L.J. 1924 Av14 Qld Sharpe P.R. 5299 and Bird L. Crosslandia anthelata Fimbristylis spiralis Abildgaardia vaginata Table 3.2 Attribute codes and definitions used for the main phenetic analyses for Cross/alldia, including corresponding initial weight values. Weight values changed in subset analyses. Attribute Description charl Mean aerial spikelet width in mm (spikelets with mature fruit) at the widest point Mean aerial nut length in mm from base of stipe to nut apex (excluding persistent style base) Mean aerial nut width in mm at the widest point Aerial nut length:width (ratio I:W/L=x; convert to decimall/x) Mean aerial nut 'stipe' length in mm Stipe length/nut length (proportion) Mean aerial anther length in mm (including appendages) Mean aerial style length in mm (including style base to base of style arm junction) Mean aerial style width in mm (at mid third) Style length:width (1 :W/L=x; convert to decimal l/x) Mean aerial stylebase length in mm (from base to constriction at style junction) Mean aerial stylebase width in mm (at widest point) Style base length:width (1 :W/L=x; convert to decimal l/x) Mean aerial glume length in mm (from base of nerve to apical point) Mean aerial glume width in mm (at widest point) Aerial glume length:width (1 :W/L=x; convert to decimal l/x) Mean leaf width in mm (at mid third) Mean culm width in mm (at mid third) Mean root width in mm (one cm below plant base) Mean inflorescence-synflorescence length in mm (from base of main bract to furthermost point of spikelets) Stamen number (actual) Style base persists on nut even if temporarily, as style always separates from style base Style base falls in tact with style Style glabrous (processes absent) Style with fimbriolia 40-60/lm (somewhat flattened processes) Style fimbria 100-140 /lm (distinctly flattened processes) Basal spikelets O-absent: always only aerial; I-present: basal spikelets (morphologically distinct) as well as aerial spikelets Plant habit O-annual I-perennial Floret sex O-always bisexual I-mixed aerial floret sex: functionally male, female or some bisexual Nut outline elliptic Nut outline obovate (2: 1 or 3:2) to widely obovate Nut outline turbinate (top-like) Nut outline pyriform (pear-shaped) Nut outline obcordate Nut outline capitate or club shaped (with a prominent stipe) Nut epidermis without protuberances Nut epidermal cell walls raised while lumen appears sunken or flat Nut epidermis is sparse and irregularly puncticulate (from a central raised silica body in some cells) Nut epidermis puncticulate, as all cells with a central silica body char2 char3 char4 char5 char6 char7 char8 char9 charlO charll charl2 char 13 charl4 charl5 charl6 char 17 char18 charl9 char20 char24 char25 char26 char27 char28 char29 char33 char35 char36 char37 char38 char39 char40 char41 char42 char43 char44 char45 char46 Weight 1 0.5 0.5 0.33 0.33 0.33 1 0.167 0.167 0.167 0.167 0.167 0.167 0.1 0.1 0.1 0.1 Table 3.2 cont'd char47 char48 char49 char50 char51 char52 char53 char55 char56 char57 char61 char62 char63 char64 char65 char66 char67 char68 char69 char70 char71 char73 char74 char79 char80 char81 char82 char83 char84 char85 char86 char87 char88 char89 char90 char91 Nut epidermal individual cells raised indiscriminately (not multiple as seen in large warts) Nut epidermal individual cells raised evenly over nut Nut epidermis with warts (cluster of multiple raised cells) arranged in vertical rows along the face Nut epidermis with warts sparse and unevely distributed Nut epidermis with pronounced warts formed by clusters of raised cells that have dense distribution Nut epidermis is rugose (cells saised in horizontal waves) Nut 0- not winged 1- winged (flattened extensions from the nut sides, including any extended notching on nut 'margins') Pilose hairs at leaf /sheath junction O-absent I-present (at least in young plants) Inflorescence: solitary (1 spikelet only-I st order primary main florescence (HF) only) Inflorescence-synflorescence: main florescence (HF) and one to multiple primary coflorescences (Cot) that are 'rayed' spikelets (on lengthed epipodia) or sometimes sessile Multiple order 'rayed' spikelets (ie 2nd order or greater ramification) 'Head' of sessile spikelets (primary reduced anthelodium with epipodia highly reduced) terminal on the culm 'Head' of sessile spikelets on lateral branch additional to the main terminal 'head' Glumes arranged distichously (spikelet distinctly compressed as glumes arise opposite the previous glume in the same plane) Glumes sub-distichous (spikelet somewhat compressed as not all opposite pairs sit in the same plane, rachilla may twist distally) Glumes distichously spiral (spiro-distichous - glumes are opposite each other but ascending pairs are arranged spirally) Glumes tristichously spiral, arranged in an ascending spiral Aerial glume margins entire (all glumes) Aerial glume margins ciliolate (thin hair-like process) Aerial glume margins fimbriolate (somewhat flattened) Aerial glume margins fimbriate (distinctly flattened) Aerial glume margins ciliate (fine hairs 0.5 mm long) Stigma number (actual) Ligule O-absent I-present Leaf blades always present on an individual Some leaf blades present, some as subulate points in an individual Leaf blades always absent in an individual Inflorescence-synflorescence bracts absent (usually in solitary spikelets) Inflorescence-synflorescence bracts present and distinct Inflorescence-synflorescence bracts glume-like Inflorescence-synflorescence bracts leaf-like Main inflorescence-synflorescence bracts shorter than inflorescencesynflorescence length Main inflorescence-synflorescence bracts equals inflorescencesynflorescence length Main inflorescence-synflorescence bracts longer than inflorescence-synflorescence length Prophyllar buds present within the inflorescence-synflorescence (polymorphic) Prophyllar buds absent within the inflorescence-synflorescence (polymorphic) 0.1 0.1 0.1 0.1 0.1 0.1 1 0.2 0.2 0.2 0.2 0.2 0.25 0.25 0.25 0.25 0.2 0.2 0.2 0.2 0.2 1 1 0.33 0.33 0.33 0.5 0.5 0.5 0.5 0.33 0.33 0.33 05 0.5 47 synflorescence structure and floret sex variability are two cases that required extra attention. Schematic representations for inflorescence-sYnflorescence structure and floret sex variation found within the Crosslandia study group are presented in the results section for this chapter. Pattern Analyses Patterns within the data were explored using the program PATN v 3.6 (Belbin 1993), by subjecting the data to ordination, cluster and network analyses (see Chapter 2 for details). Gower's similarity coefficient applied to weighted data sets proved suitable for these data and was used here. Following the initial run, groups that were distinct and separated (i.e. Blllbostylis, Abildgaardia, Fimbristylis) were removed from the analyses and the subsets were rerun to explore the remaining Abildgaardia vaginata-Crosslandia-Fimbrist)'lis spiralis group more closely. Patterns were also explored after removing basal spikelet data to ensure that these data did not unduly influence the groups fom1ed in other analyses. Floret sex variation was expanded for the Crosslandia--F. spiralis data set (after removing Abildgaardia vaginata) to capture the full variation within the analysis. Cladistic study Ingroup For the cladistic study, data were collated from the multiple OTUs scored in phenetic analyses so that each species group was converted into data for a single taxon. Species from Abildgaardia and Bulbostylis that had well-defined species lirrlits in the phenetic study were used to represent genera in the cladistic analyses. 48 Representative species across the sections of Fimbristylis (Table 3.3; see also Appendix 1) complete the ingroup taxa included to test monophyly of Crosslandia setifolia, Fimbristylis spiralis and Abildgaardia vaginata at the generic level and assess their relationships. Embryo morphology and anatomy Goetghebeur (1986) reported that Crosslandia has a variant of the Fimbristylistype embryo. Embryos of Fimbristylis spiralis and Abildgaardia vaginata were dissected from the fruit and conlpared to embryos of Crosslandia and other taxa included in the study. Leafblade and culm anatomy were examined for specific anatomical features (see cladistic character list Appendix 2) to compare Crosslandia, Fimbristylis spiralis and Abildgaardia vaginata with the other study taxa that provided generic anatomical contrasts. Inflorescence-synflorescence structure The inflorescence-sYnflorescence structures are complex across some of the taxa smnpled. The variable inflorescence-sYnflorescence structure within the Crosslandia group required a detailed investigation, as the provisional C. anthelata was based primarily on a different inflorescence type to that of C. setifolia. Table 3.3 Taxa included in the cladistic analyses to assess the relationships of Crosslalldia setifolia, provisional C. allthelata, Fimbristylis spiralis and Abildgaardia vagillata. See Table 3.1 for Crosslandia specimen list and Appendix 1 for specimen details. Taxa No. of specimens sampled Ingroup Abildgaardia macrantha (provisional) Abildgaardia ovata Abildgaardia oxystachya (provisional) Abildgaardia pachyptera (provisional) Abildgaardia schoenoides Abildgaardia vaginata Bulbostylis barbata Bulbostylis densa Crosslandia anthelata (provisional) Crosslandia setifolia Fimbristylis blakei Fimbristylis cinnamometorum Fimbristylis depauperata Fimbristylis fimbristyloides Fimbristylis furva Fimbristylis microcarya Fimbristylis schultzii Fimbristylis sp L. (Flora of the Kimberley) Fimbristylis spiralis 10 11 10 11 11 14 12 10 5 18 2 5 2 4 2 2 2 2 3 Outgroup Actinoschoenus compositus (provisional) Arthrostylis aphylla Schoenoplectiella laevis Schoenoplectiella lateriflora Schoenoplectus tabernaemontani 4 4 5 5 3 50 PAUP* Analyses Data from 24 species and 156 characters for ingroup and outgroup taxa were subjected to parsimony analysis within PAUP* 4b 10 (Swofford 2001) using heuristic techniques (hsearch swap=TBR addseq=random nreps=l 000 hold=5 multrees=yes). When multiple parsimonious trees were retrieved, a strict consensus was compared to each tree. A single tree that nlost closely resembled the tree from the strict consensus was presented in the results section. Decay and Bootstrap analyses were applied to the data set to assess relative branch support (see Chapter 2 for details). Characters with the strongest branch association were plotted onto the cladogram, using MacClade v 3.08 (Maddison and Maddison 1992) to trace characters. Results Phenetic study Crosslandia, Abildgaardia, Fimbristylis, and Bulbostylis The OTUs from Bulbostylis, Fimbristylis, Abildgaardia, and Crosslandia revealed distinct groups in both 3-dimensional (stress value= 1.0) and 2-dimensional (stress value=0.17) ordinations (Figure 3.1). Fimbristylis spiralis OTUs formed a small group on the edge of the Crosslandia anthelata cloud that separated from the Crosslandia setifolia OTUs. A clear-cut group was formed by OTUs of Abildgaardia vaginata, separated from all other OTUs of Abildgaardia. In 3-dimensional space, species formed by the OTUs of Bulbostylis were separate from the retnaining OTUs. 5\ Sirnilarly, OTUs of Fimbristylis (excluding F. spiralis) and Abildgaardia (excluding A. vaginata) fonned separate clusters. Characters having the greatest correlation (>80%) with the ordination pattern were: style glabrous, style base persistent on nut, and pilose hairs present at leaf/sheath junction - features distinctly associated with species from Blllbostylis. The remaining characters: style length and width, aerial glume length and width, plus nut length and width separated taxa within the ａ｢ｩｬ､ｧ｡ｲｩＭｆｭ｢ｲｩｳｾｹｬ ﾭ Crosslandia group (Figure 3.2). Cluster analysis (Figure 3.3) produced groupings similar to the ordination pattern, with five definite groups being fonned and numbered to correspond with the ordination groups (Figure 3.1). The OTUs of Fimbristylis spiralis were associated with the OTUs of Crosslandia s.l. fonning a broad group (03), as no grouping for the C. anthelata OTUs (C18-23) was apparent. The three OTUs of Fimbristylis spiralis held together as a group within 03. The OTUs of Abildgaardia vaginata (02) grouped next to OTUs of Crosslandia s.l. and F.spiralis, but remained distinct from 03. The remaining OTUs clustered into generic groups of Fimbristylis (04), Abildgaardia (01), and Bulbostylis (05). The presence or absence of basal spikelets did not seem to influence the pattern of ordination and cluster analyses. The ordination and phenogram that resulted following removal of basal spikelet data, and re-·analysis, confinned that any association between Crosslandia setifolia, Fimbristylis spiralis and Abildgaardia vaginata was not due solely, or in particular, to the presence of basal spikelets. • C. setifolia • C. anthelata F. spiralis X A. vaginata ;t:: Abildgaardia • Bulbostylis + Fimbristylis- '----- x G2 ｸｾ + + + ＫｾＪ + + + + +++ : • +J•• ...... • • ａｾｘ Xx G4 • • x + G3 ...... . .\. .." ｾ .... ｉｾ ｾＮ . . . Ｎｾ ... • •• G5 Figure 3.1 MDS ordination in 2-dimensions (stress = 0.17). OTU groups for Crosslandia setifolia, C. anthelata and Fimbristylis spiralis (03), Abildgaardia vaginata (02), Abildigaardia spp. (01), Fimbristylis spp. (04) and Bulbostylis spp. (05). .2 .3 ｾＲＸ X 12 Style length & width, stylebase falls with style, aerial glume length, aerial glume width, nut length & width ｾ 11 .15 +25,55 -26 -14 ｾ <>27 <> Style glabrous, stylebase persistent on nut, pilose hairs present at leaf/sheath junction Figure 3.2 Correlation of attributes with ordination space in figure 3.1. Attributes with> 80% are shown; 27 (glabrous style), 25 (persistent style base on nut) and 55 (pillose hairs at leaf/sheath junction) separate OTUs of Bulbostylis from other taxa. See Table 3.2 for attribute definitions. 0.0118 C1 ( cn ( C13 C15 C18 C20 C21 C2 C4 C6 C12 C8 C3 C5 C9 C10 C17 C14 C7 C16 C19 C22 C23 F1 F3 F2 Av1 Av3 Av5 Av11 Av9 Av7 Av13 Av14 Av2 Av6 Av4 Av8 Av12 AvlO fcl fc2 fc5 fc3 fc·'1 ff'..ll ffi1 ffi3 ffi2 fdI fd2 ffl ff2 fsL1 fsL2 fbI fb2 ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( Figure 3.3 I 1) I 11) 0.1680 0.3243 I I 0.4805 1 13)=r_1 15)_1_ 18) I 20)_1_1 _ _ 21) I 2) 4) I 6)=r_ I 12)_1_1 8)_11_ 3) I 5)_1 I 9) I I 1o)=r_1 I 17)_11 I 14)_1_ I 7) 1 I 16)_1_1_ I 19) I I 22)=r_ I I 23) _ _ 1_1_1_ - 24) 1 26)_1_ I 25) 1_ _ 1 27) 29)_1_ 31) I 37)_1_1 35)_11 33) 39)_1_ 40) _ _ 1 28) 32)_1 _ _ 30) 134) I 38) _ _ _ 1_ I 36) I_I 149) 150)=r_ 153)_1 151)_1 152) _II 161) 1156) 1 158)_1 I 157)_1 _ _ 1 159) I I 160)_1 _ _ 1_ _ 1_ 154) I 155)_1 1162) 1 163)_1 I 164) 1 165)_1 I_- G3 G2 G4 0.6368 0.7930 I I Aov1 Aov11 Aov2 Aov4 Aov8 Aov5 Aov6 Aov3 Aov9 Aov10 Aov7 Aaffod Am3 Am10 Am9 Am8 Am4 Am5 Am6 Am7 Am1 Am2 As1 As11 As14 As8 As10 As12 As2 As3 As9 As4 As13 AsS AsS Aodb As7 Ao:x:1 Ao:x:3 Ame5 Aox10 Aox9 Aox7 Aox11 Aox12 Aox2 Aox6 Aox4 Ame8 ApI Ap3 Ap2 ApS Ap7 Ap4 Apll ApE) Ap9 Aox13 Aaffpa AplO baUba1 baffba4 baUba2 baffba3 baffba6 baffba7 baffba5 bbu1 bbu2 bbu3 bbu5 bbu4 41) 51)_1_ 42)_ 1 44)_1_ 1 48)_1_1 45)_11 46)_111_ 43) 1 ＴＹＩｾ｟ I 50)_1_1 47) 88)_1 68) ＷＵＩｾ 74)_1_ 73)_1 69) 70) 7l)=r_ 72)_1 _ _ 66) I 67)_1_1 52) 62)_1_ 65) _ _ 1_ 59) 1_ 61) _ _ 63)_1_1 53) 54)_ 1 60)_1 I _ 1 55)_1_1 64) _ _ 11 56)__ 57) _ _ 1_ 89) _ _ 1_ 58) 76) 78) 1 80) ＸＵＩｾ｟ 84)_1_ 82) 1 _ 86)_ 1 87)_1 1 ___ 77)___ I 81)_1_ I 79)_ I 1 _ 1 I 1 1_1 83)_1_1 __ 1 ｾ 1 90) 92)=r_ 91)_1_ 94) 1 96) 1__ 93) _ _ I 100) _ _ 1_ I 95) 1_1 98) II 97)_ I I 101)_1 1 I 99) 11 1 102) 105)=r_ 103)_ 1 104)_1 I 107) _II I 108) III I 106)_11_1 __ 135)_ I 136)_1_ 1 137)_1 1 139)_11 I 138)_11 1_ _ 1 Figure 3.3 1 1 Gl _ bba1 bba3 bba7 bba8 bba6 bba4 bba12 bba11 bba5 bba10 baffpu bt2 bt3 bt4 bUO bt5 bt7 bt8 baffbu bpy1 bpy6 bpy2 bpy4 bpy3 bpy7 bpy8 bcie1 bcie2 bcie9 bcie3 bde7 bde4 bde8 bde5 bde6 109) 110)=r 114)_1_ 115)_1_ 113) _ _ 1 111) 118)=r 117)_1_ 112) _ I 116) 1_1_ 140)_1 119) 120)=r_ 121)_ 125)_1_1 122) 123)_ I 124)_1_1 141)_11 142) 146)=r 143) 145)1 144) 1_ 147) 148)=r _ _ 126) 127)_1_ 134) _ _ 128) 132)=r_ 129)_1 133)_11_ 130)_ 131) _1_1 _ 1 G5 1 1 I 0.0118 --------c---------c,-------,--- I 0.1680 1 0.3243 1 0.4805 1 0.6368 0.7930 Figure 3.3 WPGMA phenogram ( rJ=-O.l) using the Gower metric similarity coefficient showing groups that correspond with the ordination (Figure 3.1). Group numbers correspond with the groups in the ordination: OTU groups Crosslandia setifolia s.l. and Fimbristylis spiralis (G3), Abildgaardia vaginata (G2), Abildgaardia spp. (G 1), Fimbristylis spp. (G4) and Bulbostylis spp. (G5). See Table 3.1 and Appendix 1 for OTU and specimen details. 57 Network analyses (not presented) supported the general generic/species groupings found using ordination and cluster techniques. Crosslandia, Fimbristylis spiralis and Abildgaardia vaginata Analyses following the removal of OTUs of Bulbostylis, Abildgaardia (not A. vaginata) and Fimbristylis (not F. spiralis) resulted in the formation of four groups in the 3-dimensional ordination (stress values 3D=0.08, 2D=0.11). The OTUs of Crosslandia anthelata (G3), C. setifolia (G2) and Fimbristylis spiralis (G4) grouped more closely to each other than to OTUs of Abildgaardia vaginata (G 1) (Figure 3.4). Characters having greater than 80% correlation in separating OTUs within the ordination were generally associated with the two major groups formed (Figure 3.5). Annual habit and the characters: leaf blades always present, and inflorescence bracts leaf-like, were correlated with Crosslandia and Fimbristylis spiralis; perennial habit and the characters: inflorescence bracts glume-like and the low prevalence of basal spikelets, were correlated with Abildgaardia vaginata. Similar groups were revealed in both the phenogram and ordination, with one major difference in pattern between the two analyses. In the phenogram, OTUs of Fimbristylis spiralis (G4) were nested within the OTUs of Crosslandia anthelata (G3) (Figure 3.6). Crosslandia complex Operative taxonomic units of Abildgaardia vaginata were omitted ｦｩｾｯｭ subsequent analyses, resulting in the formation of three groups (stress values 30= • C. setifolia • C. anthelata • F. spiralis X A. vagi nata G4 • •• • • • •• • • • • • • • ｾ • • x x x x x G2 Xx x x x x x x x Gl Figure 3.4 MDS ordination in 2-dimensions (stress = 0.11) showing groups formed when Abildgaardia vaginata is included within the Crosslandia complex. Group separation for C. anthelata, C. setifolia and Fimbristylis spiralis was distinctly resolved in three dimensions (stress = 0.08) as indicated by the group borders. .35.47,80 A86 X50 ;t(85 .19 +62 -88.3 <>33 .57.4 Leaf blades always present Nut cells raised indiscriminately Inflorescence bracts leaf-like Aerial nut L:W • Anthelodia • Basal spikelets absent Inflorescence bracts glume-like <> X + Inflorescence bracts equals main inflorescence 'Head' of sessile spikelets • RootW Aerial nut W Perennial habit Nut surface warts sparse and inconsistent Figure 3.5 Correlation of characters that fit the ordination space in Figure 3.4. Characters with greater than 80% influence on the ordination are shown for the Crosslandia s.l. L=length, W=width. See Table 3.2 for attribute definitions. C1 C15 C7 C16 C3 C5 C9 C10 C13 C17 C14 C2 C6 C8 C11 C4 C12 C18 C23 C20 C19 C22 C21 F1 F3 F2 Av1 Av14 Av5 Av11 Av9 Av2 Av3 Av6 Av13 Av7 Av4 Av8 Av10 Av12 0.0835 0.2402 0.3969 0.5536 0.7103 0.8670 I I I I I I 1) 15) _ _ 1_ 7) I 16)_1 _ _ 1_ 3) I 5)_1_1 9) 10)1_ 13)_1 17)_1_ 14)_1 _ _ - 2) 1 6) - - I I 8) - I 1 I 11)1_1_1_ I G2 4) I I 12)=r_1 I 18) 23)=r_ 20)_1_ 19) 1 22)1_ _ 1_ _ 21 ) 1 24) 26)_1 25 ) 27) 40)_1_ 31) I 37)1_ I 35)_1_1_ 28) 1 29)_1_1_ 32) 1 39)_1_ I 33) _ _ 1_1 30 ) I 34) I I 36) I_I- I 38) I_I I 0.0835 I I I I G3 I I I ｾ G4 I I I_I I I I I Gl I I I 0.2402 0.3969 0.5536 I 0.7103 0.8670 Figure 3.6 UPGMA phenogram Ｈｾ］ＭｏＮｬＩ using the Gower metric similarity measure showing f()ur groups. Group numbering corresponds to the ordination in Figure 3.4 after the removal of Bulbostylis, Fimbristylis and Abildgaardia (excluding A. vaginata). See Table 3.1 and Appendix 1 for OTU and specimen details. 61 0.13, 2D=0.19). The OTUs of Crosslandia anthelata (with lengthened epipodia within the anthelodium, i.e. rayed) formed a group (G2=C18, C19, C20, C21, C22, C23), which separated more distinctly than in previous analyses from OTUs of Crosslandia setifolia (with capitate sYnflorescences, i.e. sessile spikelets) (G 1). Fimbristylis spiralis formed a group (G3) that was distinct from the Crosslandia clusters (Figure 3.7). Characters with the highest correlation with the ordination pattern that separated OTUs of Fimbristylis spiralis, from the 'rayed anthelodium' type of Crosslandia, and the typical 'capitate' type of Crosslandia were sYnflorescence related (chars 56, 57 and 62, respectively - see Figure 3.8). In the network analysis, all OTUs that were collected from the Northern Territory were connected, with OTUs of Abildgaardia vaginata and Fimbristylis spiralis diverging from the OTUs of Crosslandia anthelata (Figure 3.9). CJluster analysis (Figure 3.10) revealed groups consistent with those formed in the ordination in Figure 3.7 and MST (Figure 3.9). The clusters recovered from the phenetic analyses correspond to Crosslandia setifolia, C. anthelata, Fimbristylis spiralis, and Abildgaardia vaginata. To assess their relationships, these entities were used as terminal taxa in cladistic analyses. Cladistic analysis Four most parsimonious trees were retrieved from a heuristic search (tree length=725, CI=0.5862, HI=0.4138, RI=0.5757, RC=0.3375). Differences for terminal taxa within the Crosslandia clade were displayed in two of the four trees. Trees 1 and 3 showed Abildgaardia vaginata and Crosslandia setifolia nested within I • G2 G3 • • • • • • • • •• • Gl • •• \ ｾＭ • • • • • • • • • • • • c. setifolia • C. anthelata • F. spiralis Figure 3.7 MDS ordination in 2-dimensions (stress = 0.19) showing three groups: Crosslandia setifolia (G 1), Crosslandia anthelata (G2), and Fimbristylis spiralis (G3). The boundaries indicate the tighter grouping that was observed in the 3-dimensional scatter plot (stress = (0.13). .62 .10 89 88 X57 .56,67. Rayed anthelodium ;( Style length:width Main inflorescence bracts> inflorescence length Solitary spikelets Floral bracts tristichous • • • Sessile spikelets in a 'head' Main inflorescence bract=inflorescence length L _ Figure 3.8 Correlation of attributes with the ordination in Figure 3.7. Attributes having greater than 800/0 influence on separating OTUs into three groups are presented. G2 was separated from the other groups due to the rayed anthelodium (char 57). Fimbristylis spiralis characters were mainly solitary spikelets (char 56) and glumes spirally arranged (char 67). Capitate synflorescence (char 62) pulled the typical Crosslandia setifolia into G1. See Table 3.2 for attribute definitions F3 F1 C1 Jr.oiI---..JII"C C2 :• C. setifolia I :• C. anthelata 1& F. ｳｬ｡ｾｰｳ I . . Figure 3.9 Minimum spanning tree (MST) for OTU linkages of Crosslandia setifolia, C. anthelala and Fimbrislylis spiralis that correspond to the ordination in Figure 3.4. See Table 3.1 and Appendix 1 for specimen details. 0.1240 C1 C15 C7 C9 C16 C2 C3 C5 C6 C8 C4 C12 C11 C13 C10 C17 C14 C18 C23 C19 C22 C20 C21 F1 F3 F2 0.2122 0.3004 0.3886 0.4768 I I I I 1 1) 15) 7) 9) _ _ 1 16) 2) 3) 5) 6) 8) 4) 12) 11) 13)_1 10) 17) _ _ 1_ 14) I 18) 23) 19) 22) 20) 21) 24) 26) _ _ 1 25) I 1 1 I I I 1 1 1__ 1_1I I I I_- I I 1 Gl 1 I I 1_1- 1I I T 0.1240 0.5650 G2 1_I G3 1 0.2122 1 0.3004 I 0.3886 I 0.4768 I 0.56:;0 Figure 3.10 UPGMA phenogram Ｈｾ］ＭＰＮＱＩ using the Gower metric similarity measure that corresponds to the ordination in Figure 3.7. The separation of OTUs of Crosslandia setifolia as G 1 (same synflorescence structure as the TYPE specimen), Crosslandia anthelata with solitary or small groups of sessile spikelets on lengthened epipodia (rays) as G2 and Fimbristylis spiralis as a distinct group G3 are supported. See Table 3.1 and Appendix 1 for specimen details. 66 Fimbristylis spiralis, which in tum was nested within Crosslandia anthelata. Trees 2 and 4 both placed Crosslandia setifolia and C. anthelata as sister to Abildgaardia vaginata and Fimbristylis spiralis . Strict consensus of the four trees did not indicate branch stability for the terminal branches of the Crosslandia clade, however, the position of Crosslandia setifolia and C. anthelata as sister terminal taxa is most likely due to their close similarity. Therefore, tree 2 is presented in the results (Figure 3.11). Crosslandia setifolia, C. anthelata, Fimbristylis spiralis, and Abildgaardia vaginata formed a monophyletic group, sister to the remaining species of Abildgaardia. The Abildgaardia-Crosslandia clade was sister to the FimbristylisBulbostylis clade (containing F. depauperata, the representative of Fimhristylis section Fimbristylis that contains the TYPE species for the genus). Fimbristylis was rendered non-monophyletic by the exclusion of two members (not counting F. spiralis) of Fimbristylis: F sp. Land F. blakei. Bremer support and Botstrap analyses indicated that branches containing clades Abildgaardia spp. (Decay=4 Bootstrap=98%), Bulbostylis spp. (Decay 4 Bootstrap=94%), Actinoschoenus compositus and Arthrostylis aphylla (Decay=.3 Bootstrap=93%), plus Schoenoplectiella laevis and S. lateriflora (Decay=2 Bootstrap=88%) have strong support (Figure 3.11). Support for the Crosslandia clade was weaker (Decay=l Bootstrap=58%), while the internal branch to the Abildgaardia-Crosslandia clade was moderately supported (Decay=2 Bootstrap=6 70/0). Despite weak support, the Crosslandia group was consistently retrieved in the most parsimonious trees of subsequent reruns using different RSEED values, nreps, nchuck and chucksore values. ,....--------t-....---------- Fimbristylis blakei 13-1,139-4 .....- - - + - - - - - - - - - - - - - - 13-1 Fimbristylis sp. L Fimbristylis cinnamometorum .....-+ 13-2 .....+------ Fimbristylis furva A 13-1 Fimbristylis microcarya 1 Fimbristylis schultzii 86 13-1 Fimbristylis depauperata ｾ .i····· ｾ Ｕﾷ 2 13-1 87 26-1 27-2 ｾＱＰＭ Fimbristylis fimbristyloides : Bulbostylis densa • Bulbostylis barbata 138-4 139-3 Abildgaardia pachyptera Abildgaardia oxystachya ..... Abildgaardia ovata B 13-1 139-4 Abildgaardia schoenoides 3 67 Abildgaardia macrantha ••• ｾ •• .:- Crosslandia anthelata 80-1,3: C ross Ian d"la se t"f I" S 6 7 -lOla '+1++-----1 :at; Fimbristylis spiralis =-+ Abildgaardia vaginata 80-1,4 " ........ Actinoschoenus compositus Arthrostylis aphylla 26-2 27-1 Schoenoplectus tabernaemontani 3 Schoenoplectiella 2 66_2lateriflora 88 Schoenoplectiella laevis Figure 3.11 Cladogram for tree 2 of 4 shortest trees (tree length=725) to assess monophyly for Crosslandia. Crosslandia setifolia, C. anthelata, Fimbristylis spiralis and Abildgaardia vaginata form a monophyletic group sister to species of Abildgaardia. Bootstrap support values are given below the branch and decay indices above. The dashed lines indicate collapsed branches obtained from the strict consensus. See Appendix 1 for specimen details and Appendix 2 for characters used in analysis. o c c.O ., o C "'0 68 Support for Actinoschoenus and Arthrostylis as outgroup taxa was not strong; with the internal branch connecting the Actinoschoenus-Arthrostylis and Schoenoplectus- Schoenoplectiella as sister groups collapsing early in Bremer support analysis. Noteworthy characters Vascular bundle anatomy (including photosynthetic pathway) and embryo characters had significant association with group formation as indicated by branch support for the internal branching of Abildgaardia, Crosslandia, Actinoschoelllls- Arthrostylis, and Schoenoplectus-SchoenoplectieUa (Figure 3.11). Outgroup and ingroup were largely separated on characters relating to differences in photosynthetic path1Nay. The observed differences in embryo morphology between the generic groups, which linked Abildgaardia vaginata, Fimbristylis spiralis and Crosslandia proved to be important within the cladistic analysis (see Figure 3.11). Observations Species that form the Crosslandia complex display considerable morphological variability - specifically in Crosslandia setifolia and C. anthelata inflorescence-synflorescence structure and floret sex distribution within the spikelets. Inflorescence-synflorescence structure In:f1orescence-synflorescence structure within the Crosslandia group was highly variable (Figures 3.12-17). A simple solitary spikelet or one to two primary lateral rayed spikelets are common in Fimbristylis spiralis (Figure 3.12). Solitary spikelets were also observed in Abildgaardia vaginata, along with primary lateral rays that may be substituted by one or two primary sessile spikelets. Occasional secondary A Hf1 Cof1 B Figure 3.12 Highly reduced anthelodium A. Fimbristylis spiralis (ISOTYPE fragment BRI AQ34II94) with bisexual florets that are spirally arranged (scale bar=IO mm). B. Diagramatic representation of image A. HF I=main florescence; Cofl =primary coflorescence which bears a prophy11 indicating the lateral growth. The lowest bract subtends the coflorescence. Bar across the coflorescence indicates that the spikelet is 'rayed' as the epipodia is lengthened. c Figure 3.13. Inflorescence-synflorescence variation observed within Abildgaardia vaginata. A. Sample Av7 collected from the Northern Territory with open primary anthelodia and depauperate secondary florescences. B. Multiple spikelets that may be sessile or on lengthened epipodia (rays) produce a congested inflorescence (AvI3). C. Open reduced anthelodia are seen in Queensland material (Avll) with primary florescences. D. Geminate spikelets (primary coflorescence is sessile) or a single rayed primary coflorescence in Av5 fits the TYPE description by Brown (1810). E. Extremely lengthened epidodia are unique and seen in Av6 from the Northern Territory F. The simplest inflorescence type, solitary spikelets (Avl0), are common in material collected from SE Queensland and NSW. Drawn scale bars=10 mm HF1 Cof1 B Figure 3.14 The reduced anthelodium. A. Crosslandia anthelata (C20) showing the simple structure of the inflorescence formed from the main florescence (HF 1) and two primary rayed (indicated by the bar) coflorescences (Cofl) as represented in B (scale bar=5 mm). A HF1 Cof1=HF2 8 Figure 3.15 Highly reduced secondary anthelodium. A. Crosslandia anthelata (CI9) inflorescence with multiple rayed coflorescences (Cofl), indicated by the bar in the schematic diagram B. Some sessile secondary paracladia (2°) are present on two of the rayed coflorescences that now represent the secondary main florescence (HF2) (scale bar=8 mm). HF1 Cof1 B Figure 3.16 A terminal head of spikelets as sessile ramified reduced anthelodia. A. Crosslandia setifolia (C14) has the synflorescence structure that is typical for the TYPE collection from Goody Goody, Western Australia (scale=2mm). B. All primary (HP1 and Cofl) and secondary (2°) spikelets are sessile in the schematic for the synflorescence of A. See Appendix 1 for specimen details. HF1 HF2 (Cof1) B Figure 3.17 Lateral head of sessile spikelets. A. In Crosslandia setifolia lateral heads of spikelets may be formed in addition to the terminal head (e.g. in C14 scale=8 mm). The extended growth of one (rarely two) epipodium or 'ray' (marked with a bar) produces sessile secondary coflorescences (Cof2) with sessile tertiary paracladia (30) as represented in the schematic diagram B. See Appendix 1 for specimen details. 75 lateral branches (secondary paracladia) arising from the prin1ary co florescence lnay also be present (Figure 3.13). The reduced anthe10dia of primary rayed spikelets (primary coflorescences) seen1ed to separate specimens of Crosslandia collected in Kakadu National Park, Northern Territory (Figure 3.14) easily from material producing 'heads' of sessile prirrLary coflorescences collected in Western Australia (the origin of the TYPE) (Figure 3.16). Sampling across the range, however, revealed sYnflorescences with: 1. usually simple rayed spikelets as 2-4 coflorescences (Figure 3.14); 2. a mixture of 4-6 coflorescences, sessile or rayed, bearing solitary, or 2-3 sessile spikelets as primary coflorescence paracladia (Figure 3.15); 3. heads of sessile spikelets formed from primary florescences, i.e. main florescence plus primary coflorescences and their paracladia (Figure 3.16); and 4. characters as for 3, with occasional lengthened primary ray or rays, bearing a lateral 'head' of spike1ets (Figure 3.17). Although no specimens from Western Australia were observed to have the simplest inflorescence-sYnflorescence structure (see type 1 above), samples frOlu the Northern Territory displayed all four types. Floret sex Of the 23 specimens of Crosslandia setifolia s.1. sampled, 19 have felnale florets distal within some aerial spikelets (Table 3.4 and Figure 3.18). When present, feluale florets always occur in the distal position of aerial spikelets (Figure 3.18 B-E) and Table 3.4 Floret sex distribution seen in aerial spikelets for sampled specimens in Crosslandia setifoiia and the provisional C. anthelata. W.A. = Western Australia, N.T. = Northern Territory. See Table 3.1 and Appendix 1 for OTU label and specimen details. Species OTU State label Floret sex Floret sex Floret sex bisexual male female 1(distal) l(mid only with distal female) 1(proximal or all) 1(proximal) l(proximal or all) 1(proximal or all) 1(distal) 1distal a 1(distal) I (distal) C. setifolia C. setifolia C2 C3 W.A. W.A. C. setifolia C. setifolia C7 C9 W.A. W.A. C. setifolia CIa W.A. I (proximal, distal or all) C. C. C. C. C12 CI4 C16 CI7 W.A. W.A. W.A. W.A. C. setifolia C. setifolia C. setifolia CI C4 C5 N.T. N.T. N.T. 1(proximal) 1distal) 1(mid) C. C. C. C. C. C. C. C. C. C. C. C6 C8 C11 CI3 C15 CI8 C19 C2a C21 C22 C23 N.T. a a a a setifolia setifolia setifolia setifolia setifolia setifolia setifolia setifolia setifolia anthelata anthelata anthelata anthelata anthelata anthelata N.T. N.T. N.T. N.T. N.T. N.T. N.T. N.T. N.T. N.T. 1(mid) a 1(proximal or all) a I (proximal) a 1(proximal or all) a I (mid) a a I (proximal) or all florets I (proximal) or all florets I(proximal or all) I(proximal or all) a 1(proximal) or all florets 1(proximal or all) 1(proximal or all) 1(proximal) or all florets I (proximal or all) 1(proximal or all) 1(all) 1(all) 1(proximal or all) I a 1(proximal l(proximal I (proximal I (proximal or all) or all) or all) or all) I(distal) l(distal) I(distal) I (distal) I(distal) I(distal) 1(distal) 1(distal) I (distal) 1(distal) 0 0 1(distal) 0 1(distal) I (distal) () I (distal) I(distal) A. B. c. D. ｾ Female floret ｾ ｾ E. F. Bisexual floret Male floret G. Figure 3.18 Floret sex variation exhibited in aerial or basal spikelets within Crosslandia as defined here. Crosslandia setifolia, C. anthelata (A-G see Table 3.3 for sample list and floret sex distribution) Fimbristylis spiralis (A), and Abildgaardia vaginata (A and rarely C). The all female florets in spikelet F. occur only in basal spikelets. are easily recognised by the change in shape of the spikelet due to the narrowly elongated glumes. These distal female aerial portions usually resemble the all female basal spikelets that also have narrowly elongated glumes, and are conspicuously clustered at the base of the plant. Female florets (aerial or basal) appear to have never possessed rudimentary androecia, as no male parts were visible in mature female florets and filament scars were not observed. Bisexual florets within aerial spikelets occurred in 13 of the 23 specimens sampled. Distribution of the bisexual florets varied considerably, from being positioned proximally (following the non-reproductive glumes with female florets distal) (Figure 3.18 C), at mid-spikelet (following male functional florets with female florets distal) (Figure 3.18 D), or distally (following male functional florets) (Figure 3.18 B). In four of the 13 samples, some spikelets were observed to have only bisexual florets (excluding the lowest empty glumes). When spikelets of Crosslandia setifolia and C. anthelata consisted of all bisexual florets, it appears that the fenlale proximal florets were undeveloped, as other spikelets in the same samples had female distal florets present. The bisexual floreted spikelet in Crosslandia setifolia and C. anthelata may easily revert to unisexual female florets distally and therefore does not consistently produce bisexual only spikelets as seen in Abildgaardia, Fimbristylis and Bulbostylis. The pattern of sexuality seen in the TYPE specimen of Crosslandia setifolia with aerial spikelets consisting of all functionally male florets (Figure 3.18 G) was represented by only three of the 23 specimens sampled. Those three specimens, however, were not restricted to Western Australian material. In the other nlaterial sampled, male only spikelets regularly occurred amongst plants with fenlale distal or 7<) bisexual mix types. Rudimentary female organs were always present in the 111ale florets of all specimens. It was possible then for a single plant to possess aerial spikelets with floret sex arranged as: all male, male proximal and bisexual distal, male proximal and female distal, bisexual proximal and female distal, or all bisexual. In contrast, bisexual florets were constant in aerial spikelets in specimens of Abildgaardia vaginata and Fimbristylis spiralis (Figure 3.18 A), except for one of the sampled specimens of A. vaginata (Av14), where female florets were present in some spikelets. Floret sex in basal spikelets was more constant -- usually bearing all-female florets in a spikelet (Figure 3.18 F). Some basal spikelets in Crosslandia setifolia (not observed in C. anthelata), F. spiralis and A. vaginata contained minor variations in floret sex. Basal spikelets with all bisexual florets (CIS, Fland AvI2), or bisexual proximal and female distal florets (C17) were concealed amongst the usual allfemale spikelets. No functionally male-only florets were seen in basal spikelets in any of the material examined. No correlation was apparent between the distribution of floret sex within aerial or basal spikelets and inflorescence pattern within Crosslandia setifolia or C. anthelata. Basal spikelets The presence of basal spikelets fOffi1s one of the main consistent features for the group (Figure 3.19). Basal spikelets were observed in 10 of the 103 specin1ens of A. vaginata examined for this study, and were usually poorly developed and easily overlooked. Three of these 10 specimens were included in analyses; only specirnen Av8, collected from New South Wales, produced a mature nut in a well-developed 80 basal spikelet (Figure 3.20). The other two sampled specimens where basal spikelets were observed (AvIO and AvI2) were collected from Queensland. The morphology of basal spikelets in all species that form the Crosslandia group was distinct from aerial counterparts in the shape of spikelets, and floret features such as glume shape and length, style length, and nut size (Figures 3.20-21). With the exception of Abildgaardia vaginata, the nuts in the plants examined were produced mainly from the conspicuous basal spikelets (Figure 3.22). Embryo morphology Elnbryo morphology for Abildgaardia vaginata, Crosslandia and Fimbristylis spiralis was the Crosslandia-variant of the Fimbristylis-type (Figure 3.23 A-E). The maturity of the embryo is crucial when defining the type; immature embryos show the Schoenus-type embryo formation (Figure 23 F). Vegetative anatomy Abildgaardia vaginata occasionally produces leaf blades and these were compared with species of Crosslandia and Abildgaardia. Leafblade and culm anatomy for Fimbristylis spiralis and Abildgaardia vaginata were of the general C 4 fimbristyloid type, as was Crosslandia. Transverse sections revealed some differences among the other species assessed for the Crosslandia complex. There were three to four hypodermal layers in the leaf blade of Abildgaardia vaginata, while these cells may be absent from Crosslandia and Fimbristylis spiralis or present as one or two rows (the second often an incornplete row) (Figure 3.24). Abildgaardia ovata also had hypodermal tissue formed from one complete row and some cells in an incomplete second row, but the Figure 3.19 Basal spikelets that show different morphology to aerial spikelets observed in Crosslandia s.l. Some of the variation observed in basal spikelets of A. Crosslandia setifolia (C14), B. C. anthelata (C21), C. Abildgaardia vaginata (Av8) and D. Fimbristylis spiralis ISOTYPE fragment (BRI AQ341194) (scale=5 mm). See Appendix 1 for OTU and specimen details. B D Figure 3.20 Comparative floral parts from aerial and basal spikelets of Abildgaardia vaginata (Av8). A. Basal spikelet with mature nut. B. Aerial glume (left) and basal glume (right) showing size difference. C. Aerial style is distinctly shorter than the basal style. D. Aerial nut (left) is distinctly smaller than the basal nut (right). Scale=l nun. Figure 3.21 Scanning electron micrographs of nuts from the provisionally defined Crosslandia s.l. Similarities in the nut features for provisional Crosslandia anthelata (C 19) A. basal nut and B. aerial nut; Fimbristylis spiralis (F 1) C. basal nut and D. aerial nut; and E. aerial nuts from Abildgaardia vaginata (AvI2). Scale=500 lJIIl. See Appendix 1 for specimen details. Figure 3.22 Basal spikelets at the plant base of Crosslandia setifolia (C12 collected from Western Australia). Many aggregated sessile spikelets are distinct at the plant base and exhibit different morphology to the aerial spikelets. Most of the nuts produced by the plant are from these basal spikelets. Scale bar=125 mm. See Appendix 1 for specimen details. A B c o E Figure 3.23 Variation in embryos of Crosslandia setifolia and Abildgaardia vaginata. A. and B. Crosslandia setifolia, C17 collected from Western Australia, C. and D. Abildgaardia vaginata Av2 collected from New South Wales, and E. Av3 (collected from Queensland), share the Fimbristylis-type embryo. F. Abildgaardia vaginata Av6 (embryo not fully developed - collected from the Northern Territory) has the Schoenus-type embryo, where the primordial shoot and root are situated sub-basally (see text p 80). Scale bar=200 IJ.m. See Table 3.1 and Appendix 1 for specimen details. A B c Figure 3.24 Leaf and/or culm transverse sections showing C4 fimbristyloid anatomy. A. Crosslandia setifolia (C17) leaf and B. culm; C. Fimbristylis spiralis (F3) leaf and D. culm; E. Abildgaardia vaginata (Av6) culm. Scale bar=200 Jlm. See Table 3.1 and Appendix 1 for specimen details. 87 hypodennis is not a constant feature for species of Abildgaardia (see Chapter 4 for details). Culm shape for the Crosslandia complex was either elliptic to wavy (sometimes deeply) elliptic with the number of sclerenchYma strands constantly less than the nU111ber of vascular bundles. In contrast, sclerenchYma strands in species of Abildgaardia equalled the number of vascular bundles within the culm. Discussion The four separate species (Crosslandia anthelata, C. setifolia, Fimbristylis spiralis and Abildgaardia vaginata) within Crosslandia as found in phenetic analyses, fonned a monophyletic group in the cladistic analysis. Evidence from phenetic and cladistic studies indicates that all four taxa should be recognised as species in the genus Crosslandia. Cladistic analysis clearly finds that Fimbristylis is non-monophyletic, with the con1bined Crosslandia-Abildgaardia and Fimbristylis s.s.-Bulbostylis clade nested within two species currently accepted in the genus Fimbristylis. The nested species of Bulbostylis also renders the Fimbristylis s.s. clade non-monophyletic. The position of Bulbostylis is possibly an artifact of the small sample size of Bulbostylis used in this analysis (see Chapters 5 for greater detail on Bulbostylis). The genus Fimbristylis is large and morphologically heterogeneous and its taxonomy is unresolved. Although Crosslandia is nested within taxa currently accepted as Fimbristylis, sinking Crosslandia into Fimbristylis is not recommended while the generic limits for Fimbristylis are uncertain. It will take some time to 88 ･ｶｾｯｳ ｲ taxonomic issues due to the large number of species assigned to Fim bris(vlis, and the great amount of variation present across those species. Meanwhile, defining the species and generic limits for Crosslandia could only benefit future work on Fimbristylis. The presence of basal spikelets across the Crosslandia group, plus the floret sex variation in Crosslandia setifolia and C. anthelata do not fall into the current generic limits for Fimbristylis; this adds to the justification of maintaining Crosslandia. New combinations can now be made for the species of Crosslandia, based on the phenetic and cladistic data. This study supports Goetghebeur (1986) in that the Crosslandia 'rayed' and 'capitate spikeleted' forms ofinflorescence-synflorescence, although highly variable, can be separated into two entities. The TYPE specimen for Crosslandia setifolia from Western Australia represents the densely spikeleted form, which is consistent for the State. Inflorescence variation is greatest in the Northern Territory for the current Crosslandia setifolia as recognised here. The occurrence of all four species in the Northern Territory indicates that this area is the centre of diversity for the Crosslandia group. Although there were lilnited samples of Fimbristylis spiralis, this species appears to be morphologically uniform and apparently geographically isolated in coastal north-east Arnhem Land. Abildgaardia vaginata shows a broader range of distribution, occurring in the Northern Territory, Queensland, and coastal New South Wales. The dinlinished capacity to produce basal spikelets, the presence of bisexual florets and the extended distribution reflects the adaptability of this species compared to other species of Crosslandia. The synapomorphy of functionally unisexual florets in the aerial spikelets of the mixed floret sex seen in Crosslandia setifolia and C. anthelata indicates evolution from the bisexual condition, and distinguishes the two species 89 froln Abildgaardia vaginata and Fimbristylis spiralis, both of which regularly have bisexual aerial florets (see Figure 3.18 and Table 3.3). A concurrent study tracing the molecular phylogeny of some members of the tribe Abildgaardieae (Ghan1khar et a1. 2005, in press) showed strong support for the genetic closeness of Abildgaardia vaginata and Crosslandia setifolia. However, the restricted smnple size used in the study (one specimen of Abildgaardia vaginata - from New South Wales; five Crosslandia s.l. specimens - three from Northern Territory, two from Western Australia; and no Fimbristylis spiralis samples) provided no additional insight into the relationships of the Crosslandia group to that already obtained from plant morphology, anatomy and embryo morphology of this study. More detailed gene analyses on populations for all four species would be necessary to assess fully the genetic diversity and thereby the genetic relationships within the Crosslandia group. Additional sampling, generally between the Western Australian and Northern Territory gradient, is also recommended, including sampling closer to the Fimbristylis spiralis sites. The current description of inflorescence variation for Crosslandia setifolia is clearly inadequate. Also, the observed variation of mixed floret sex within spikelets in C. setifolia and C. anthelata (see also Goetghebeur 1986) has not been formally published. Unisexual flowers are common in some groups of the Cyperaceae and have independent origins (Bruhl 1995; Goetghebeur 1998). The mixed floret sex seen in Crosslandia is exceptional within the tribe Abildgaardieae, and the spikelet sex arrangement is similar to the expressed floret sex in species of Scleria; explaining the placement of Crosslandia into Sclerieae by Hutchinson 1934, 1959). Terminal female spike1ets with no male rudiments occur in several species of Scleria (includes 90 Acriulus), and intermediates occur between the bisexual spikelet, male spikelet and the strictly female spikelet (Kern 1963). Scleria differs from Crosslandia in having spikelets that consist of a single fertile floret (spikelets being female, male or bisexual), rather than the multi-floreted spikelet with mixed floret sex in Crosslandia. There are no structural discrepancies within the Crosslandia spikelet to indicate that the variable sex expression results from a composite floral structure such as a 'spike' e.g. as seen in the Cariceae (Kern 1958; Timonen 1998; Starr et a1. 2004), or 'sYnanthium' e.g. as in Mapania (Simpson 1992); on the contrary, all florets conform to the typical base plan of the scirpoid floral arrangement (Vrijdaghs et a1.. 2005). Functional male florets possessing sterile vestigial female organs are 'type l' unisexual flowers according to Mitchell and Diggle (2005); 'type II' unisexual flowers are unisexual from inception (the floral meristem initiates only the male or female organs, omitting the hermaphroditic stage). Female florets in aerial or basal spikelets of Crosslandia appear to be type II flowers. Mitchell and Diggle (2005) mapped unisexual flower types onto a composite angiosperm phylogeny for dioecious taxa and then analysed the information to determine the number of evolutionary origins (Mitchell and Diggle 2005). A similar exercise could be applied to taxa with unisexual florets within the Cyperaceae, to see where Crosslandia and the type I and type II florets fit into the evolution of unisexual florets within the family. In addition, examining the developmental stages of both the functionally staminate and pistillate florets within Crosslandia could determine whether androecia in pistillate flowers become sterile at a very early stage and, therefore, these pistillate flowers could actually be type I unisexual flowers. 9\ The attenuated glumes of female florets in Crosslandia have different nl0rphology to glumes in male and bisexual florets. Some grass species (e.g. maize and Tripsacum), although grass spikelets are not homologous with those in Cyperaceae, may show glume morphology that correlates with unisexual floret sex expression; grass florets in the Andropogoneae always having type I unisexual florets, i.e. with vestigial male or female organs (Le Roux and Kellogg 1999). In Cyperaceae, type II female florets occur singly or in small clusters at the plant base in Trianoptiles, with these amphicarpous plants producing aerial spikelets with bisexual florets (Levyns 1943). The female florets are enclosed by a glumaceous 'utricle' vvhere modified glUlnes tightly hold the nut (Haines and Lye 1977). In Crosslandia, multiple female florets occur in basal (usually solely female) or aerial spikelets (as mixed sex), where the glume (one per nut) also tightly encloses the nut; the attenuated apex of the glurrle wraps around and supports the long, thin, narrow style. Delimiting Crosslandia setifolia s.l. into two species is consistent with the phenetic analyses and reflects the inflorescence-synflorescence variability across the sampled geographical range. All observed specimens with fewer spikelets on rays (lengthened epipodia) occur within the Northern Territory (apparently concentrated in the Kakadu National Park area), and specimens with many sessile spikelets forming distinct 'heads' occur in Western Australia; this provides support for the separation. Despite the unavoidably small sample size, combining Fimbristylis spiralis into the genus Crosslandia and retaining species level distinction seems justified based on: differences in nut size (smaller overall), only bisexual aerial florets present, and glumes always spirally arranged (Table 3.5). Plants of Fimbristylis spiralis were found to be consistently smaller in habit than those of Crosslandia setifolia or the 02 provisional C. anthelata; however, the extreme conditions of the coastal envirol1111ent where Fimbristylis spiralis grows may affect the plant size. Further sampling from the remote locations in the Northern Territory is needed to test this hypothesised classification. Abildgaardia vaginata should be placed in Crosslandia. This move is supported by Abildgaardia vaginata having an embryo type similar to that of ｃｲｯｳ ｬ｡ｮ､ｩ｡ｾ not the Abildgaardia-type embryo observed for other members of Abildgaardia in this study. Leafblade and culm anatomy in Abildgaardia vaginata are also similar to Crosslandia, as are basal spikelets, when present. New combinations are now provisionally put forward prior to valid publication. N olnenclature of Crosslandia Genus: Crosslandia W.Fitzg. TYPE: Crosslandia setifolia W.Fitzg. 1. Crosslandia anthelata Goetgh. ex K.L.Clarke J.J.Bruhl & K.L.Wilson sp. nov. ined. 2. Crosslandia setifolia W.Fitzg. 3. Crosslandia spiralis (R.Br.) K.L.Clarke & K.L.Wilson comb. nov. ined. Fimbristylis spiralis R.Br. SYn. nov. (photograph and fragments of the Queensland Herbarium, BRI 340661 were seen) Iria spiralis (R.Br.) Kuntze Scirpus spiralis (R.Br.) Poir 4. Crosslandia vaginata (R.Br.) K.L.Clarke comb. nov. ined. TYPE Abildgaardia vaginata R.Br. Fimbristylis brownii Benth. F. leptoclada Benth (non. Fl. Hongk.) F. vaginata (R.Br.) Domin (non Boiv. ex C.B.Clarke in Dur. & Schinz) F. vaginata f. leptoclada (Benth.) Domin F. vaginata (R.Br.) Domin f. vaginata held at Table 3.5 Comparison of species to be assigned to Crosslandia. Characters taken from phenetic and cladistic analyses. Crosslandia setifolia is the species of the genus. W.A. = Western Australia, N.T. = Northern Territory, Qld = Queensland, N.S.W. = New South Wales. TYPE Crosslandia setifoiia Crosslandia anthelata Abildgaardia vaginata Fimbristylis spiralis Inflorescence-synflorescence distinct 'heads' of sessile spikelets, sometimes with lateral 'heads' rayed solitary spikelets or rayed solitary plus some sessile spikelets solitary spikelet or rayed solitary spikelets sometimes with sessile spikelets solitary spikelet or rayed solitary spikelets Primary coflorescence number 5-8 2-4 0-4 0-4 Floret sex within a spikelet mixed mixed bisexual bisexual Aerial nut range: length/width 1.25/0.65 to 1.911.0 1.35/0.8 to 1.75/1.25 1.2/0.9 to 1.911.4 1.2/0.7 to 1.35/0.75 Glume arrangement subdistichous subdistichous to spiral subdistichous tristichously spiral Basal spikelets (amphicarpy) always present always present sometimes present always present Plant habit annual annual perennial annual Distribution \V.A. and N.T. N.T. N.T., Qld, N.S.W. N.T. 94 Chapter 4 .Abildgaardia Vahl: a phenetic and cladistic study Introduction In this chapter I focus on the limits and relationships of taxa within Abildgaardia Vahl to determine the most appropriate taxonomic rank of Abildgaardia - either as a separate genus or as a section of Fimbristylis. Species currently circumscribed under Abildgaardia (as a genus or as Fimbristylis section Abildgaardia), are concentrated in Australia (Blake 1969; Goetghebeur 1998). Australian species occur mostly in northern regions of Australia. Abildgaardia ovata (= Fimbristylis ovata (Burm.f.) J.Kern) is the only species found in temperate coastal regions of eastern Australia (extending south into New South Wales). KIal (1971) reinstated Abildgaardia as a genus, elevated from Fimbristylis section Abildgaardia. Abildgaardia ovata (Burm.f.) Kral and A. mexicana (Palla) Kral were recombined for the move. Goetghebeur and Coudijzer (1984, 1985) followed the generic elevation, recognising A. hygrophila (Gordon-Gray) Lye (1971) and A. triflora (L.) Abeywickr. (syn. A. tristachya Vahl), as species in addition to A. ovata. Species that formed Abildgaardia were defined by the following characteristics: perennial habit, leaf sheath orifice glabrous, one to few large spikelets, spikelets laterally compressed, glumes distichous, fruits stipitate-capitate, and 3-fid style (Goetghebeur and Coudijzer 1984). Australian endemics under 95 Fimbristylis section Abildgaardia, F. macrantha, F. oxystachya (Bentham 1878) and F. pachyptera (Blake 1940), were provisionally named as species of the genus Abildgaardia (A. macrantha, A. oxystachya, A. pachyptera respectively) by Goetghebeur (1986) in his doctoral dissertation. The provisional names were never fomlalised. Abildgaardia schoenoides R.Br (sYn. Fimbristylis squarrulosa) and Fimbristylis odontocarpa S.T.Blake were not included in Goetghebeur's studies, but fit the general characteristics of species of Abildgaardia. Bulbostylis parvinux C.B.Clarke and Fimbristylis variegata Gordon-Gray were transferred by Lye (1971) to Abildgaardia. Gordon-Gray (1995) did not accept the transfer while embryographic studies were lacking. Goetghebeur and Coudijzer (1985) decided that B. parvinux truly belonged within Bulbostylis and retained Fimbristylis variegata, rejecting a transfer to Abildgaardia. Bruhl (1995) found that Abildgaardia variegata and A. hygrophila possessed C 3 anatomy, and referred to thern as C 3 Abildgaardia. More recently Bruhl and Wilson (2005, in press) have included the former under Fimbristylis and the latter under Abildgaardia. Kral and Strong (1999) reinstated Fimbristylis bahiensis Steud. into Abildgaardia as A. baeothryon St Hi!. and described a new species A. papillosa Kra1 & M.Strong, which they regarded as being close to A. baeothryon. The limits of the genus Abildgaardia were extended to incorporate the two species. I find the three species Abildgaardia baeothryon, A. papillosa, and A. hygrophila, more closely resemble taxa from Fimbristylis in their nut size, colour, and epidermal scuJlpturing; and glume characteristics, although superficially similar to species of Abildgaardia, also resemble glumes in some species of Fimbristylis and Crosslandia. 96 Some Australian collections with intermediate morphology did not fit into current species descriptions or keys (Sharpe 1986; Latz 1990; Rye 1992). Intennediates were observed for A. oxystachya, A. pachyptera, and A. schoenoides (A. sp. aff. schoenoides 1: AsS and As6, A. sp. aff. schoenoides 2: As7 and required assessment as to their placement. The putative new species A. sp. aff. pachyptera and F. sp. aff. odontocarpa also required attention. Prior to testing the monophyly of Abildgaardia, it was necessary first to test and set species limits within Abildgaardia. Materials and Methods Taxa As the previous main phenetic analysis in chapter 3 recovered Abildgaardia as a distinct group, only the species from Abildgaardia under study were included in the phenetic analyses for this chapter. All Australian taxa provisionally placed in the genus Abildgaardia (A. oxystachya, A. pachyptera, A. macrantha) by Goetghebeur (1986) plus A. ovata, A. schoenoides and Fimbristylis odontocarpa, were sampled to allow species level assessment. Specimens of Fimbristylis sp. aff. odontocarpa, and A. sp. aff pachyptera were included in the group requiring phenetic analysis. A total of 62 specimens for Abildgaardia formed the basis for the phenetic study (Table 4.1). Specimens sampled were selected to encompass morphological variability over the geographic range of available material. Abildgaardia ovala is the only cosmopolitan species of the genus occurring in Australia. Abildgaardia vaginata was not included in this study; this species was removed fron1 Abildgaardia Table 4.1 Specimens sampled as the focus group in the assessment of Australian Abildgaardia. The 'OTU' corresponds to the sample used in phenetic analyses. States are given for Australian collections. N.T. = Northern Territory, W.A. = Western Australia, Qld = Queensland, N.S.W. = New South Wales. See Appendix I for specimen details. Species Abildgaardia ovata Abildgaardia schoenoides A. sp. aff. schoenoides 1 Abildgaardia macrantha Abildgaardia oxystachya OTU State Aov1 Aov2 Aov3 Aov4 Aov5 Aov6 Aov7 Aov8 Aov9 Aov10 Aov11 As1 As2 As3 As4 As7 As8 As9 As10 As11 As12 As13 As14 As5 As6 AmI Am2 Am3 Am4 Am5 Am6 Am7 Am8 Am9 Aml0 Aox1 Aox2 Aox3 Aox4 Aox5 Aox6 aox7 aox8 aox9 Qld Qld Qld N.S.W. N.S.W. N.S.W. N.S.W. N.S.W. Qld Qld Qld W.A. W.A. W.A. N.T. Qld Qld Qld WA N.T. W.A. N.T. N.T. N.T. N.T. W.A. N.T. N.T. N.T. Qld Qld Qld N.T. N.T. N.T. W.A. Qld N.T. N.T. N.T. W.A. Qld W.A. W.A. Collector Specht R.L. 408, Reeves R.D. Batianoff G.N. 11056 O'Shanessy P.A. 1656 Wilson K.L. 5818 Johnson L.A.S. Rodd A.N. 2277 Mueller F. Rodd A.N. 2434 Stanley T. 8019 Neldner V.1. 3905 Clarke K.L. 99, Bruhl J.J. Wilson K.L. 4888 Dunlop C.R. 7838 Mitchell A.A. 2129 Dunlop C.R. 8651, White Bruhl J.J. 487 Jacobs S.W.L. 5903 Clarke K.L. 70, Bruhl J.1. Clarke K.L. 157, Bruhl J.1, Wilson K.L. Clarke K.L. 216, Bruhl J.J, Wilson K.L., Cowie J.D. Clarke K.L. 120, Bruhl J.J, Wilson K.L Clarke K.L. 230, Bruhl J.J, Wilson K.L., Cowie J.D. Perry R. 222 Bruhl J.J. 1261, Hunter J.T., Egan J. Dunlop C.R.5863, Craven L.A. Hartley T.G. 14405 Cowie J.D. 6202, Booth R. Wilson K.L. 4971 Dunlop C.R. 4102 Clarkson J. 8324 Wilson K.L. 8073, Clarkson J., Jacobs S.W.L. Clarkson J. 6624 Cowie ID 5260, Taylor S. Dunlop C.R. 3453 Clarke K.L. 249, Bruhl J.J, Wilson K.L., Cowie J.D. Latz P.K.4038 Blake S.T. 15725, Webb Bruhl J.J. 1252 Latz P.K. 8667 Wilson K.L. 5369 Clarke K.L. 124, Bruhl J.J, Wilson K.L. Blake S.T. 19620 Cane S. 53 Hartley T.G. 14357 Table 4.1 cont'd Fimbristylis sp. aff. odontocarpa Fimbristylis Odol1tocarpa Abildgaardia pachyptera Abildgaardia sp. aff. pachyptera Aox10 Aox11 Aox12 Aox13 Aaffod W.A. N.T. Qld N.T. W.A. Carr G.W. 4377, Beauglehole A.C. Booth R. 618 Blake S.T. 13611 Wightman G. 424 and Dunlop C.R. Carey J. Aod1 Aod2 ApI Ap2 Ap3 Ap4 Ap5 Ap6 Ap7 Ap9 Ap10 Ap11 Aaffpa Qld Qld N.T. N.T. N.T. W.A. N.T. N.T. N.T. N.T. N.T. Blake S.T. 13582 Turpin G.P., Thompson E.J. Hunter J.T. 1547, Bruhl J.J. Dunlop C.R. 9041 Jones M., Booth R. 24 Dunlop C.R. 5339 Wilson K.L. 5109, Taylor S. Wilson K.L. 5207 Chippendale G Clarke K.L. 253, Bruhl J.J, Wilson K.L., Cowie LD. Clarke K.L. 181, Bruhl J.J, Wilson K.L., Cowie J.D. Adams L.A. 1715 Clarke K.L. 201, Bruhl J.J, Wilson K.L., Cowie LD. N.T. <)9 and placed in Crosslandia (Chapter 3). Abildgaardia vaginata, is provisionally called Crosslandia vaginata. Representative specimens of Abildgaardia triflora, A. mexicana, A. hygrophila, A. baeothryon, and Fimbrisytlis variegata were sampled only for cladistic analysis. Phenetic Study Pattern Analyses Invariant character states were removed from the initial PATN data set used to assess the Crosslandia group, and the revised data subjected to multivariate analyses, where characters (not states) were given equal weight (see Chapter 2 for full details). Twenty-one quantitative and 16 qualitative morphological characters (Table 4.2) were scored from 62 samples (Operative Taxonomic Units - OTUs) of Australian Abildgaardia. Data were subjected to ordination, cluster and network analyses as detailed in Chapter 2, and the combined data set analysed using the Gower metric similaritycoefficient is presented in the results section. Groups that were defined in the first run of analyses for all OTUs of Abildgaardia in this chapter were relTIoved and data re-analysed to explore grouping of the relnaining taxa. As with the previous study, 2-dimensional scatter plots adequately displayed the separation of taxa and were preferred for presentation purposes. Groups defined in the phenetic study formed the terminal taxa used in cladistic analysis. Table 4.2 Attribute codes and definitions used in the main phenetic analyses for the Australian Ahildgaardia, including corresponding initial weight values. Weight values changed in subset analyses. Attrilbute Description charI Mean aerial spikelet width in rum (spikelets with mature fruit) at the widest point Mean aerial nut length in rum from base of stipe to nut apex (excluding persistent style base) Mean aerial nut width in rum at widest point Aerial nut length:width (ratio I:W/L(x) (to decimalllx) Mean aerial nut 'stipe' length in rum Stipe length/nut length (proportion) char2 chad char4 char5 char6 char7 char8 char9 Weight Mean aerial anther length in rum (including appendages) Mean aerial style length in nun (including style base to base of style arm junction) Mean aerial style width in rum (at mid third) charI 0 charIl Style length:width (ratio 1:W/L(x) to decimal lIx) Mean aerial stylebase length in nun (from base to constriction at style junction) char12 Mean aerial stylebase width in nun (at widest point) charl3 Style base length:width (ratio 1:W/L(x) to decimal lIx) char14 charl5 charl6 char17 charl8 charJl9 char20 Mean aerial glume length in rum (from base of nerve to apical point) Mean aerial glume width in nun (at widest point) Glume length:width (ratio 1:WIL(x) to decimal lIx) Mean leaf width in nun (at mid third) Mean culm width in nun (at mid third) Mean root width in nun (one cm below plant base) Mean inflorescence-synflorescence length in rum (from base of main bract to furthermost point of spikelets) Stamen number (actual) Plant habit O-annual I-perennial Nut outline obcordate 0.5 Nut outline capitate or club shaped (with a prominent stipe) 0.5 Nut epidermis without protuberances Nut epidermis with some warts that are sparse and unevenly distributed Nut epidermis with pronounced warts formed by raised multiple cell clusters that are dense and evenly distributed Nut 0- not winged 1- winged (flattened extensions from the nut sides. Includes any extended notching on nut 'margins') Glumes I-distichous to sub-distichous (spikelet somewhat compressed, but rachilla may twist distally); O-glumes tristichously spiral Leafblades always present in an individual Some leaf blades present, some as subulate points in an individual Inflorescence-synflorescence bracts absent (usually in solitary spikelets) Inflorescence-synflorescence bracts present and distinct Inflorescence-synflorescence bracts glume-like Inflorescence-synflorescence bracts leaf-like Main inflorescence-synflorescence bracts I-shorter than the inflorescencesynflorescence; O-equal or longer than main inflorescence-synflorescence 0.33 0.33 0.33 char24 char35 char4l char42 char43 char50 charSl char53 char65 char80 char81 char83 char84 char85 char86 char87 1 1 0.5 0.5 0.5 0.5 0.5 0.5 1 101 Cladistic study Ingroup l\1embers of Fimbristylis, Crosslandia and Bulbostylis used in Chapter 3 were also included in the cladistic component of this chapter. Overseas specimens for the cosJmopolitan A. ovata were compared to the A. ovata group defined in the phenetic analyses and five overseas samples added to the cladistic data set to cover the extended geographical range. Additional species assigned to Abildgaardia occurring outside of Australia: A. triflora, A. mexicana, A. baeothryon, including the C 3 species Abildgaardia hygrophila and Fimbristylis variegata (sYn. Abildgaardia variegata) (Bruhl 1995), were added to the ingroup to assess monophyly of the genus (Table 4.3). No specimens of A. papillosa were available. Embryo morphology The Abildgaardia-type embryo was first observed by Van der Veken (1955) based on material of A. ovata and A. oxystachya (at the time assigned under Fimbristylis section Abildgaardia). In his study, Van der Veken commented on the possibility of reinstating genera, including Abildgaardia, based on the embryo types observed. Goetghebeur (1986) expanded on Van der Veken's work and found that embryos from A. macrantha (= F. macrantha) and A. pachyptera (= F. pachyptera) were also of the Abildgaardia-type. Embryo morphology was sampled for all species included in the cladistic analysis. Table 4.3 Taxa included in the cladistic analyses to assess the relationships of species in AbUdgaardia. Species from Crosslandia included here were defined in Chapter 3. See Appendix 1 for specimen details. Taxa No. of specimens sampled Ingroup Abildgaardia baeothryon Abildgaardia hygrophila Abildgaardia macrantha (provisional) Abildgaardia mexicana Fimbristylis odontocarpa (section Abildgaardia) Fimbrisytlis sp. aff. odontocarpa Abildgaardia ovata Abildgaardia oxystachya (provisional) Abildgaardia pachyptera (provisional) Abildgaardia schoenoides Abildgaardia triflora Abildgaardia sp. aff. pachyptera Abildgaardia sp. aff. schoenoides 1 (As5, As6) Abildgaardia sp. aff. schoenoides 2 (As7) Bulbostylis barbata Bulbostylis densa Crosslandia anthelata (provisional) Crosslandia setifolia Crosslandia spiralis (provisional) Crosslandia vaginata (provisional) Fimbristylis blakei Fimbristylis cinnamometorum Fimbristylis depauperata Fimbristylis fimbristyloides Fimbristylis furva Fimbristylis microcarya Fimbristylis schultzii Fimbristylis sp L. (Kimberley Flora) Fimbristylis variegata 4 2 10 5 2 1 16 13 11 11 4 1 2 1 20 15 5 18 3 14 2 5 2 4 2 2 2 2 1 Outgroup Actinoschoenus compositus (provisional) Arthrostylis aphylla Schoenoplectiella laevis Schoenoplectiella lateriflora Schoenoplectus tabernaemontani Trachystylis stradbrokensis 4 4 5 5 3 7 103 Anatomy Variation in vegetative anatomy Gordon-Gray (1971) and photosynthetic pathways (Gilliland and Gordon-Gray 1978; Bruhl 1995) for species of Abildgaardia (C4 fimbristyloid and C 3 ) prompted further investigation here. Leaf blade and culm anatomy were, therefore, assessed for species being tested for placement in Abildgaardia. Anatomical data were included in the cladistic analysis. PA UP* Analyses Data for ingroup and outgroup taxa, comprising 34 terminal taxa and 149 characters, were subjected to parsimony analysis within PAUP* (Swofford 2001) using heuristic techniques (hsearch swap=TBR addseq=random nreps==l 000 hold=5 multrees=yes) (see Chapter 2 for full details). Branch support was determined using Bremer support and Bootstrap analysis. Characters were traced using MacClade v 3.08 (Maddison and Maddison 1992) and the most relevant characters plotted onto the cladogram selected for presentation. Results Phenetic study Operative Taxonomic Units formed disjunct groups that represent the genera Abildgaardia, Crosslandia, Fimbristylis and Bulbostylis (see also chapter 3), with some species groups apparent within Abildgaardia at this broad level (Figure 4.1). W"hen OTUs from the Abildgaardia group were analysed separately, OTU group boundaries were slightly relaxed, revealing broad variation within some species groups (Figure 4.2). Group formation in figure 4.2 was correlated with nut outline, • Crosslandia IlllI A. ovata I. A. schoenoides X A. aft. schoen 1 • ;( A. aft. schoen 2 • A. macrantha G1 + A. oxystachya <> F. aft. odontocarpa F. odontocarpa • G2 .• •... • ＮｾＬ ·. ..,•.,...• ••• • ••• • • • A. pachyptera o A. aff. pachyptera 6 Bulbostylis Fimbristylis G3 G4 Figure 4.1 MDS ordination in 2-dimensions for OTUs of Abildgaardia, Crosslandia, Fimbristylis and Bulbostylis. OperativeTaxonomic Units that divide into generic groups Abildgaardia (G 1) Crosslandia (G2), Fimbristylis (G3) and Bulbostylis (G4) are clearly shown within the ordination. Groups of OTUs that form species in Abildgaardia are discemable at this broad level. Stress value=O.17. See Chapter 3 for PCC and classification that correspond with this ordination. <> <> <> <> <> <> ..• , • • • 0 <> + • • <> ++ <> + +-i:t:-l+ • •• ••• # • • • • <> + + + .A.ovata • A. schoenoides As13 • •• A. aff schoenoides 1 As? • • A. macrantha + A. oxystachya - A. aft odontocarpa • • • • I I - A. odontocarpa <> A. pachyptera o A. aff pachyptera Figure 4.2 MDS ordination (stress = 0.13) showing OTUs forming broad species groups within Abildgaardia. As13 (A. schoenoides) with perianth nests within the main group, and As? falls on the outer limits. OTUs for A. aff schoenoides 1 are separate from the remaining A. schoenoides OTUs. The OTU of A. sp. aff. pachyptera is distinctly separated from all other groups. See Table 4.1 and Appendix 1 for OTU and specimen details. aerial style W perennial habit inflorescence bract leaf-like nut outline capitate nut winged and obchordate in outline • style L:W • + aerial style L aerial stylebase L .53 1_11 I I I 9 ,X41 ;(42 .35 ,+8 -86 -, 10 ｾ '--------------' Figure 4.3. Correlation of attributes with ordination space in Figure 4.2. Attributes with ＾ＸＰＨｾ influence on OTU group formation are shown. See Table 4.2 for attribute definitions. L=length, W=width ｾＭ ｾＭ ｾＭ \ As5 .A. ovata ilA. schoenoides A. aff schoenoides 1 • A. macrantha +A. oxystachya - A. aff odontocarpa As10 ｾａｓＴ As3 ｾＢＬａＮ odontocarpa <> A. pachyptera o A. aff pachyptera As12 Figure 4.4 Minimum spanning tree (MST) from network analysis that corresponds to the ordination in Figure 4.2. See Table 4.1 and Appendix 1 for OTU and specimen details. 1) ( ( ( ( ( 2) ( 4)--1 ( 8)_1_1_1 II ( 5) _ _ ( 6 ) _ 1 _ 1 1 _ _-,- ( 3) ( ( 9) 10) ( 7) I I 1 1_1 ----,-1 II 1_11 48) 16) 17) 49) ---;- -----,1 _ 1 50)_1 1 _ 36) ( 42) _ _ ( 44) _ _ ( 46) 1 I 47) _ _ 1_1_1_ 58) 1 62) 1 26) 27) 28) 35)1_ 38)_1 --,- 43)_1_ 40) _ 41) 1_ 1 I 45)_1_1_ ( ( ( ( ( 1 ( ( ( 33)_1__ 30) _ _ 1 29) 31)_ 1 ( 32)_1 __ ( 12) ( ( 22) _ _ ( ( I I I I I 1 I I 1 1 -.,.. -----,- 1 1 1_1_ 1_ _-,- 1 1 I ---.,.- 23)___ 1 I 13) ｾ 20) _ _ 1_ 14) I 15) I 1 11_ 24) 18) 51) I 1 _ _---.,.- 1 1 1 _ 1 ( ＵＩｾ ( ( 59) 52) ( 60) ( ( 57) I 61) _ _ 1_1__ ( ---I 1_ _ 1__ 25) 19) 21) ( ( ( _ -----,-1 ( ( I 1_1_1_ 34)_1_ ( ( ---.,.- 37) 39) ( ( ( ( ( ( I I I ( ( ( ( ( ( ( ( ( ( 0.6250 11)_1 __ ( ( 0.5084 1 I I 1 Aov1 Aov11 Aov2 Aov4 Aov8 Aov5 Aov6 Aov3 Aov9 Aov10 Aov7 Aaffod As:; AsEi Aod1 Aod2 Aox1 Aox3 Aox2 Aox4 Aox8 Aox5 Aox6 Aox10 Aox7 Aox9 Aox11 Aox12 Aox13 Aaffpa AmI Am2 Am3 Am10 AmS' Am8 Am5 Am4 Am6 Am7 As1 As11 As14 As8 As10 As12 As2 As9 As3 As4 As13 As7 Ap1 Ap5 Ap9 Ap2 Ap10 Ap7 Apll Ap4 Ap3 Ap6 0.3918 0.2753 0.1587 0.0421 1 1_1_ I 1_ 1 1_ I 54) 1 1 I 1__ 53) 1 56)--1 1 -I0.0421 1 0.1587 1 1 I 0.2753 1 0.3918 I 0.5084 1 0.6250 Figure 4.5. WPGMA (P=-O.I) phenogram for the full Abildgaardia analyses, using the Gower metric association measure, corresponds to the ordination in Figure 4.2. Abildgaardia sp. aff. schoenoides lOTUs are separated from A. schoenoides s.s. in the phenogram and As13, As7 nested withinA. schoenoides. The OTU A. sp. aff.pachyptera is nested within A. oxystachya and not A. pachyptera. The two outlying A. macrantha OTUs (AmI, Am2) are nested with the main A. macrantha group. See Table 4.1 and Appendix 1 for OTU and specimen details. 109 nut wing presence/absence, style and style base length, and plant habit (Figure 4.3). The broad species groups observed in the ordination were consistent in network (Figure 4.4) and cluster analysis (Figure 4.5). Abildgaardia schoenoides group Minor internal groups ofOTUs that broadly form the species Abildgaardia schoenoides were consistent in the ordination scatter plot (Figure 4.2), phenogram (Figure 4.5), and MST (Figure 4.4). The minor group of OTUs As 1, As8, As 11 and As 14 are spread across the geographical range, and do not appear to be consistent with any other morphological, anatomical, or embryological feature that could Justify separation of these OTUs from the remainder of the main group. In the ordination scatter plot the OTU A. sp. aff. schoenoides 2 (As7) is placed at the extreme edge of the A. schoenoides limits (Figure 4.2), and links A. sp. aff. schoenoides 1 and other OTUs A. schoenoides in the MST (Figure 4.4); As7 is nested with OTUs of A. schoenoides in the phenogram (Figure 4.5). The placement of As7 is therefore considered to fall within the limits of A. schoenoides. The one OTU (AsI3) collected from the edge of the Arnhem Land Plateau with perianth bristles present (see observations section) was distinctly nested within the A. schoenoides cluster in all analyses. Specimens of A. sp. aff. schoenoides 1 (As5, As6) remain separated from OTUs of A. schoenoides, consistent with the habit being annual and not obviously perennial as in the TYPE specimen of Fimbristylis squarrulosa. 110 Abildgaardia pachyptera - A. oxystachya group Operative Taxonomic Units that grouped as the species A. pachyptera were distinct in the ordination due to their winged nuts and obcordate nut outline (Figure 4.3). Two of the specimens sampled Ap3 and Ap6, appear transitional between A. pachyptera and A. oxystachya (Figure 4.2, 4.4, 4.5), with nut attributes intermediate to the two species. Cluster analysis groups Ap3 and Ap6 together as a minor group nested within A. pachyptera (Figure 4.5), and the MST links A. o),ystachya to A. pachyptera them within the other OTUs of the species (Figure 4.4). The OTU of the collection labelled A. sp. aff. pachyptera is distinctly separated from the main group of A. pachyptera OTUs in the broad ordination of all OTUs for Abildgaardia (Figure 4.2 and 4.4), and in the subset analyses for ordination (2-D stress=O.l; 3-D stress=O.07) (Figure 4.6), but is grouped with A. oxystachya in the phenogram of the same analyses (Figure 4.7) as it was in the full Abildgaardia cluster analysis (Figure 4.5). Operative taxonomic units of Abildgaardia oxystachya also exhibited a minor group (aox 1, aox3, aox 11) within the ordination (Figure 4.2) that was not substantiated in the phenogram (Figure4.5) or by MST linkages (Figure 4.4). Again an intermediate specimen (aox 13) stretches the A. oxystachya limits and pushes towards the limits of A. pachyptera within the ordination scatter plot; any connection between the two species was not seen in subsequent analyses (Figures 4.6-7). Analyses of the data subset containing OTUs for A. oxystachya, A. paclzyptera and A. sp. aff. pachyptera confirmed the three species groups in the ordination (Figure • • • • • • • • • • • • • • • •• • •• • • • • A Qxystachya I i • A pachyptera A aft pachyptera Figure 4.6 Two dimensional MDS ordination (stress = 0.1) showing OTUs grouped as Abildgaardia oxystachya, A. pachyptera and A. sp. aff. pachyptera. The drawn border separating A. sp. aff. pachyptera indicates the stronger separation of this OTU from the remainder of the OTUs in 3-dimensional analysis (stress = 0.07). See Appendix 1 for specimen details. • A oxystachya • A pachyptera Ap7 ;Y5 Ap10 A aft pachyptera ｾｰａ Figure 4.7 Minimum spanning tree from network analysis that corresponds to the ordination in Figure 4.6. See Table 4.1 and Appendix 1 for OTU and specimen details. Aoxl Aox3 Aox2 Aox5 Aox6 Aoxl0 Aox4 Aox8 Aox7 Aox9 Aoxll Aox13 Aaffpa( Aox12 ApI Ap5 Ap9 Ap2 Apl0 Ap7 Apll Ap4 Ap3 Ap6 0.0885 0.1820 0.2755 0.3690 0.4625 0.5560 I I I I I I 1) --;- 3) -----;- 2) I 5)1_ 1 6)_1___ 10) _ _ 1___ 4)__ 8) 1 1 I I 1 1 1 I 7) 9) 11) I 1 _ _ 1 _ _ 1 _ _------,-- --,-- 20) 24) 12) 13) 17)_ 1 _ _ 1 _ _ I I I 21)_1_1___ I I 14) 22) 19) 23) 16) 15) 18) I 1__ 1 1 1 1_ _ 1__ 1 1__ 1 1 1 J 1 I I 1 I I 1 0.0885 0.1820 0.2755 0.3690 0.4625 0.5560 phenogram, using the Gower metric association measure, that Figure 4.8 WPGMA Ｈｾ］ＭｏＮｉＩ corresponds to the ordination in Figure 4.4. OTUs of Abildgaardia oxystachya, A. pachyptera and A. sp. aff. pachyptera form distinct groups separate to A. oxystachya, in which A. sp. aff. pachyptera is nested. See Table 4.1 for OTU specimen information and Appendix 1 for specimen details. 114 4.6), observed when all OTUs for Abildgaardia were analysed. Within the n1inilTIum spanning tree the OTU for A. sp. aff. pachyptera terminates the A. oxystachya chain ofOTUs, and is not nested within the group (Figure 4.7), however, the single OTU for A. sp. aff. pachyptera was nested within OTUs of A. oxystachya in the phenogram (Figure 4.8). Abildgaardia macrantha group OTUs of A. macrantha: Am3-1 0 formed a tight cluster in all ordination analyses with Am 1 and Am2 as distinct outliers from the main group (Figure 4.2). Phenograms show the outlier OTUs as a minor internal group in both the broad Abildgaardia group (Figure 4.5) and subset analyses (not presented). Network analysis was less consistent, with Am 1 and Am2 linked to OTUs of A. macrantha at separate ends of minor chains in the full Abildgaardia analysis (Figure 4.4), while in a subset analysis these samples were together, linked at the end of the A. macrantha chain (not presented). Abildgaardia ovata - F. odontocarpa group 'Within ordination and cluster analyses for the full Abildgaardia data set the cluster of OTUs that combine as A. ovata included the single sample of Fimbristylis sp. aff odontocarpa on the outer limits of the group and within the group respectively (Figures 4.2 and 4.5). Operative Taxonomic Units for Fimbristylis odontocal]Ja were placed near to the Abildgaardia ovata group in the phenogram, although distinctly separated within the ordination. Subset analyses showed Fimbristylis sp. aff. odontocarpa separate to, but near the perimeter of the A. ovata group in the ordination scatter plot (Figure 4.9), and in the phenogram was retained within the •• • • • ••• • • Am2 Am] .Aovata A macrantha >< A aft odontocarpa ' L- XAodontocarpa Figure 4.9 MDS ordination in 2-dimensions (stress = 0.17) showing OTUs grouped as Abildgaardia ovata, A. macrantha, F. odontocarpa, and F. sp. aff. odontocarpa .. The border drawn around F. sp. aff. odontocarpa indicates the stronger separation of this OTU from the remainder of the OTUs in 3-dimensions (stress=0.09). AmI and Am2 were more broadly separated from each other in the 3-dimensional ordination. See Table 4.1 and Appendix 1 for OTU and specimen details. Aov1 Aov11 Aov2 Aov4 Aov5 Aov6 Aov8 Aov7 Aov3 Aov9 Aovl0 Aaffod Aodl Aod2 AmI Am2 Am3 AmlO Am9 Am8 Am5 Am4 Am6 Am7 0.0701 0.1647 0.2593 0.3538 0.4484 0.5430 1 I I 1 1 I 1} -,- 11} 1 2} _ _ 4} 1 _ _----,- I I I I I_I 5} I I_I 6) 8) ｾ 7) 3) 9) 1 __ I T I 10) 1 22) 23} --;24) 12} 13} 14} 21} 1_ 20) _ _ 1 _ _-,19) 1_____ 16) 1 15) ｾ 17} ｾ 18) ----,--- I 0.0701 I 0.1647 I 1 _ I 1 _ --;- I I I I I I .....,- -----,- 1 I I I I I I I 1 ---,---- I 0.2593 --;-1 I 0.3538 I -.,.. I 0.4484 1 I 0.5430 Figure 4.10 WPGMA (P=-O.I) phenogram, using the Gower metric association measure, that corresponds to the ordination in Figure 4.6. OTUs of Abildgaardia ovata (including F. sp. aff. odontocarpa), F. odontocarpa group separately to OTUs of A. macrantha. AmI and Am2 form a minor inten1al group within OTUs of A. macrantha. OTUs for F. odontocarpa form a distinct group and F. sp. aff.. odontocarpa is nested within OTUs of A. ovata. See Table 4.1 and Appendix 1 for OTU and specimen details. 117 A. ovata group (Figure 4.10). The minimum spanning tree links Fimbristylis sp. aff. odontocarpa between A. ovata and A. oxystachya, and Fimbristylis odontocarpa connected as a terminal branch in the full Abildgaardia analysis (not presented). Hov/ever the subset minimum spanning tree reveals Fimbristylis sp. aff. odontocarpa and Fimbristylis odontocarpa as chains linked to Abildgaardia ovata at opposite ends of the A. ovata chain (not presented). Abildgaardia ovata, A. macrantha, A. oxystachya, A. pachyptera, A. aff. pachyptera, A. schoenoides, A. aff schoenoides 1, Fimbristylis odontocarpa, and F .sp. aff odontocarpa, as defined in the phenetic analyses, formed tenninal taxa for the Australian species of Abildgaardia. Cladistic Analysis There were 126 most parsimonious trees retrieved (tree length=987, CI=0.4863, HI=0.5137, RI=0.5848, RC=0.2844) from the heuristic search. Tree number 23 was selected to show branch support and character placement, as it was one of the trees with essentially the same topology as the strict consensus. All species assessed as Abildgaardia did not form a monophyletic group (Figure 4.11). Abildgaardia ovata, A. schoenoides, A. sp. aff schoenoides 1, A. macrantha, A. oxystachya, A. pachyptera, A. sp. aff pachyptera, A. triflora, A. mexicana, Fimbristylis odontocarpa and F. sp. aff odontocarpa formed a monophyletic group with strong branch support (Decay=3 Bootstrap=87%). The Abildgaardia s.s. group was sister to A. h'ygrophila which does not have the typical Abildgaardia embryo or nut characters of other taxa from Abildgaardia (see observations section) and therefore is not considered as a species of Abildgaardia. Synapomorphies for the Abildgaardia clade 1 r t t - - -... 63-4; 42-2 ＱｾＲＷＭ r--t+f_ _" 145-6 72 Actinoschoenus compositu Arthrostylis aphylla 1 7-3; 63-3 Trachystylis stradbrokensis Schoenoplectus tabernaemontani 42-3 27-2 12-1 Schoenoplectiella lateriflora Schoenoplectiella laevis Fim bristylis va riegata 145-6; 127-1; 42-3 Abildgaardia baeothryon 145-6; 127-2; 42-3 Fimbristylis furva 1 Fimbristylis cinnamometorum ···tH··············..············..············..··: 63-2 27-2 12-1 145-1 127-1 42-1 1 Fim bristylis m icrocarya 12-2 Fimbristylis fimbristyloides 27-1 : >5:wr: Ｙｾ fi. 145-3·116-1· ' , Bulbostylis barbata I . Bulbostylls densa B 127-2: 127-1:(2-3: 42-2 : 1 12-2 1 63-3 ,..1 42-1 ..II Fimbristylis depauperata Fimbristylis schultzii Fimbristylis sp. L =+-Fimbristylis blakei 42-2 145-1 127-3 ｊｉｾ Abildgaardia hygrophila :14'5-1; 127-3; 42-3; 27-2 :125-2 r+------ Abildgaardia macrantha :124-2 60-2 ..... 1 Abildgaardia ovata : ＸＰｾ . 64 :60-2 1 :i : 74-4 ｾ .. : 54-3 : . : 63• -.- - - - - Abildgaardia oxystachya 60- •• • 1 F. sp. aft. odontocarpa 54-1 • • : 124-3 ••• : ••• ｾ," Fimbristylis odontocarpa : "1 4· : 51 JJ-Abildgaardia schoenoides • :60-2 87 .. i Ｖｾ Abildgaardia sp. aff. 145-2 124-5 schoenoides 127-3 Abildgaardia pachyptera 1 125-5 124-4 --7Abildgaardia sp. aff. 0 --....l[ 91-5 pachypte ra 80-4 Abildgaardia triflora 2 42-2 • 79 Abildgaardia mexicana : .1 Crosslandia vaginata : .145-1,6 ". ﾷｾｉ ". "....•..•.....•... i C I d· . I' 145-1 65 : : - - ross an la spira IS 127-1 42 -1 •• • 55 : . : Crosslandia setifolia - Crosslandia anthelata F1 119 that separate the Abildgaardia s.s. clade from A. hygrophila are: Abildgaardia-type embryo (145-2), nut length greater than 1.35 mm, most greater than 2mlTI (124- 3,4,5), nut width greater than 1.5 mm (124-5), glume nerve area broad (91-5), glume margins indistinct from glume back (80-4), culm sclerenchyma strands equal the number of vascular bundles in the outer ring (42-2), and C 4 photosynthetic pathway (27-1). In addition, there is no branch support for the combined clade Abildgaardia s.stricto plus Abildgaardia hygrophila (Figure 4.11). Within the Abildgaardia clade all Australian species including the cosmopolitan A. ovata formed a group sister to A. triflora and A. mexicana, with both internal branches receiving weak support (Bootstrap=64%; Decay=l) (Figure 4.11). Species currently accepted in the genus Fimbristylis were polyphyletic, with the bulk of Fimbristylis species (including F. depauperata from Fimbristylis section Fimbristylis of the TYPE species), and nested species of Bulbostylis, grouped as sister to the broad group containing Abildgaardia s.s. and Crosslandia. Abildgaardia baeothryon and Fimbristylis variegata occur basally in the tree, with A. baeothryon nested within F. variegata and sister to all terminal taxa of Fimbristylis, Abildgaardia and Crosslandia. Observations Inflorescence-synflorescence structure Species of Abildgaardia s.s. possess the simplest inflorescence structure of all the genera within the tribe Abildgaardieae. All Australian species of Abildgaardia 120 consistently have the most reduced inflorescence type, seen as a solitary spikelet (Figure 4.12 A, B). Occasionally, Abildgaardia ovata Inay have a single primary coflorescence (rarely 2) present in addition to the main florescence, as seen in specimens across the global range of this cosmopolitan species (Figure 4.12 C). The African A. triflora and American A. mexicana have the most complex inflorescence structure for Abildgaardia with one to four primary coflorescences (Figures 4.12 D, E). Spikelet coflorescences in A. mexicana are always sessile and are therefore the only species of Abildgaardia with depauperate 'heads'. Perianth Perianth bristles were observed in a specimen (As13) collected from the edge of the Amhem Plateau in Kakadu National Park, that otherwise resembles the other OTUs of Abildgaardia schoenoides. The presence ofperianth in this material is unique among taxa of the Abildgaardieae. The two bristles with antrorse prickle hairs (Figure 4.13) were obscure within the depauperate spikelets and were not observed in all florets. Nut shape and pattern Nuts within the Abildgaardia s.s. are large (> 2mm), frequently capitate or club shaped, and have a distinct narrowed stipe (Figure 4.14-15). The exception in nut shape and stipe presence is in A. pachyptera, where nuts may have winged extensions on the adaxial margins of the plano-convex nut (Figure 4.15). Intennediate fOnTIS between wide wings and shorter notched margins are seen between specimens of A. pachyptera (Figure 4.15) and and to some extent in samples A c Figure 4.12 Variation of the inflorescence-synflorescence structure seen in species of Abildgaardia. Solitary spikelets are most common in Abildgaardia, especially in the Australian species as seen in A. A. macrantha (Am8) and B. A. sp. aff pachyptera (Aaffpach). C. A. ovata (Aov9) occasionally produces one or two coflorescences (rays), while multiple coflorescences are common in D. A. triflora (P.J Greenway 1859). E. In A. mexicana (CJ Pringle 3127) one or two (rarely 3) coflorescence spikelets are sessile. See Table 4.1 and Appendix 1 for OTU and specimen details. Figure 4.13 Perianth in Abildgaardia. Specimen of Abildgaardia schoenoides (As 13) collected from Arnhem Land Plateau, showing A. and B. one of two perianth bristles. B. At higher magnification the antrorse barbs are evident. Scale bar A=500 ｾｭ and B=50 ｾｭＮ See Table 4.1 and Appendix 1 for OTU and specimen details. Figure 4.14 Scanning electron micrographs (SEM) and light micrographs (LM) of nuts in species of Abildgaardia. A. SEM of nut in A. ovata (Aov2), B. and C. A. macrantha (Am3 and Am7), D. and E. A. oxystachya (Aox5 and Aoxl), F. Fimbristylis sp. aff. odontocarpa (Faffod), O. and H. LM of nut from F. odontocarpa (Fod2) with a hom-like apical protusion (0) and large tubercules in view (H). 1. Nut from A. sp. aff. pachyptera (Aaffpach) showing the distinctive projections from the face margins. Scale bars A-F=500 Jlm, 0-1=1000 Jlm. See Table 4.1 and Appendix 1 for OTU and specimen details. Figure 4.15 Scanning electron micrographs (SEM) of nuts in species of Abildgaardia. A. A. schoenoides nut (As9), B. A. sp. aff. schoenoides nut (As6) and C. epidermal cells of the nut in B at higher magnification. D., E. and F. The nut of A. pachyptera (Ap4, Ap3 and Ap2) is variable in the size of the nut 'wings', 'wing' notching and epidermal sculpturing. G. A. triflora (F. Malaisse 400 & P. Goetghebeur) nut that is almost smooth as large tubercules are absent. H. Nut of A. mexicana (c. G. Pringle 3127) and 1. epidermal cells of the nut in H. at higher magnification. Scale bars A-B, D-H=500 Jlm; C-I=1000 Jlm. See Table 4.1 and Appendix 1 for OTU and specimen details. Figure 4.16 Scanning electron micrographs for Abildgaardia hygrophila and Fimbristylis variegata. A. Abildgaardia hygrophila (CJ Ward 2794) nut at magnification (35x) comparable to other species of Abildgaardia, showing rugose sculpturing and B. at higher magnification (45x). C. Nut in A and B at higher magnification to view epidermal cell shape and cell walls. D. Fimbristylis variegata (J Browning 834) at 35x magnification and E. at higher magnification (65x). F. Nut epidermal cells for F. variegata differ in shape, being almost rectangular, and have sinuose cell walls. Scale bar A, D=500 ｾｭ［ C, D=50 ｾｭＮ See Appendix 1 for specimen details. 126 of A. oxystachya (not shown). In A. sp. aff. pachyptera nut margins are deeply notched in all three planes (Figure 4.15), and in F. odontocarpa prominent horns at the apex of the nut may be present (Figure 4.14). Tubercules are common as multicelled protuberances from the nut epidermis and are most prominent in F. sp aff. odontocarpa as large flat-topped protusions (Figure 4.14). Distinct tubercules are absent in some A. pachyptera (Figure4.15 F), and almost so in other samples of A. pachyptera, A. macrantha, and A. triflora (Figure 4.14-15). The nut epidermis in A. hygrophila is rugosely sculptured, although the cell shape is hexagonal and the cell walls are straight (Figure 4.16). In A. baeothryon the sn1all white nut has circular epidermal cells but the cells walls are sinuose. Ovate to circular nut epidermal cells are present in Fimbriszvlis variegata and the cell walls also are sinuose (Figure 4.16). Embryo The Abildgaardia-type embryo was present in all species that formed the monophyletic group Abildgaardia. Embryo size and shape was consistent within a species and varied between species (Figure 4.17-18). Embryos for A. baeothryon and Fimbristylis variegata were of the Schoenus-type, and the Fimbristylis--type in Abildgaardia hygrophila (Figure 4.18). Anatomy Leaf blade and culm anatomy confirmed that all species of the Abildgaardia s.s. group shared the C4 photosynthetic pathway of the Fimbristyloid-type. Anatomical structure for Abildgaardia was consistent across the species (Figure 4.19-20). Leaf G Figure 4.17 Light micrographs of mostly whole cleared embryos for Abildgaardia-type embryos in species of Abildgaardia. A. Embryo of A. ovata (Aovll), B. A. oxystachya (Aox5) and C. longitudinal section (Aox12) showing well developed first and second primordial leaves surrounding a small third leaf. D. A. sp. aff. schoenoides (As5), E. and Abildgaardia schoenoides (As7) are shown from a top view of the basal orientated root and shoot, F. side view of embryo in E. G. In F. sp. aff. odontocarpa (Faffod) the well-developed Solid arrow=root, open second leaf is clearly visible (thin arrow). Scale bars=100 ｾｭＮ arrow=shoot, thin arrow =second primordial leaf. See Appendix 1 for specimen details. A E F H G Figure 4.18 Light micrographs of whole cleared embryos for some species assigned to Abildgaardia. A. Abildgaardia macrantha (AmI 0) B. A. sp. aff.pachyptera (Aaffpach), C. A. triflora (P.J Greenway 1859) D. A. mexicana (e.G. Pringle 3127 - embryo damaged) all share the Abildgaardia-type embryo. E. Abildgaardia hygrophila (K.L. Tinley 307) has a Fimbristylis-type embryo (embryo partially damaged) and F. Fimbristylis variegata (J Browning 834) (previously in Abildgaardia) and G. Abildgaardia baeothryon (J Almeida de Jesus 1466) share the Schoenus-type embryo. The orientation of the sub-basal root and shoot (of G) are shown in top view in H. Scale bars=l 00 /lm. Solid arrow=root, open arrow=shoot, thin arrow=second primordial leaf. See Appendix 1 for specimen details. A B D Figure 4.19 Culm and leaf blade transverse sections for two species of Abildgaardia showing the typical outlines, arrangement of sclerenchyma strands per vascular bundle, and C4 fimbristyloid anatomy. A. Abildgaardia ovata (Aov11.) culm and B. leaf blade sections at low magnification, with C. culm and D. leafblade at higher magnification. E. Abildgaardia triflora (P.J Greenway 1859) culm and F. leaf blade sections at low magnification with G. culm and H. leaf at higher magnification, showing a second semi row of vascular bundles in the leaf section. Scale bars=1 00 Jlm. See Table 4.1 and Appendix 1 for OTU and specimen details. A B E F Figure 4.20 Culm and leaf blade transverse sections for species of Abildgaardia showing the typical outlines, arrangement of sc1erenchyma strands per vascular bundle, and C4 fimbristyloid anatomy. A. A. sp. aff. pachyptera (Aaffpach) culm and B. leaf blade sections. C. Fimbristylis odontocarpa (Fod2) culm and D. leaf blade sections. E. Abildgaardia oxystachya (Aox4) culm and F. leaf blades sections. Scale bars=200 j.!m. See Table 4.1 and Appendix 1 for OTU and specimen details. 131 blades and culms mostly have sc1erenchYn1a equalling the number of vascular bundles. The exception is seen in the leaf of A. triflora, where a second layer of vascular bundles occurs; although the outer bundles are usually in line with a sclerenchYma strand (Figure 4.19). Leafblade shape in transverse section is generally crescentiform and a hypodermis may be present as 2 or 3 cell layers; well developed in A. triflora. Culms are mostly elliptic or sometimes hexagonal, as in A. pachyptera and A. sp. aff. pachyptera (Figure 4.20). In contrast, Abildgaardia hygrophila and Fimbristylis variegata both share C 3 photosYnthetic pathway and vary in general vegetative anatomy, however the sections were not suitable for photography. Discussion Abildgaardia, as defined by the sYnapomorphies of C 4 fimbristyloid photosYnthetic pathway, Abildgaardia-type embryo, and nut length, and is composed of 11 species. Nine species are found in Australia, eight of these are endelnics. The caution by Bruhl and Wilson (2005, in press) regarding the inconsistency of the photosYnthetic pathway in Abildgaardia is refuted by this study, which shows that all species in the Abildgaardia clade share the C 4 fimbristyloid photosYnthetic pathway. This study finds that Abildgaardia baeothryon and A. hygrophila have been misplaced in Abildgaardia as indicated by the embryo type (Schoenus- and Finlbristylis-types respectively) and the C 3 anatomy of A. hygrophila. Fimbristylis hygrophila and A. baeothryon (= Fimbristylis bahiensis) nut characters are atypical 132 for the genus Abildgaardia, and along with F. variegata, resemble those of some species of Fimbristylis. Evidence from this study does not support the inclusion of Abildgaardia hygrophila or A. baeothryon in Abildgaardia and confirms the exclusion of F. variegata (Goetghebeur and Coudijzer 1984, 1985; Goetghebeur 1986; Gordon-Gray 1995; Bruhl and Wilson 2005, in press). Therefore, Fimbristylis bahiensis, F. hygrophila, F. variegata and should all be excluded from Abildgaardia Detailed studies are needed to ascertain the correct placement of these three species, which is not necessarily in Fimbristylis. Kral and Strong (1999) described their new species A. papillosa as being close to A. baeothryon (= FimbrisZylis bahiensis Steud.). Based on this similarity and the nut characters described in the protologue, it seems reasonable now that A. papillosa does not belong in the genus Abildgaardia. Assessing the embryo and anatollly of A. papillosa is a necessary step to confirm the correct placement of this species. Abildgaardia papillosa should therefore also be excluded from Abildgaardia. Goetghebeur and Coudijzer (1985 p:209) regarded Abildgaardia and Bulbostylis as 'genera with offspring that were derived from a more primitive fimbristylidoid stock'. The perianth bristles found in some florets from a single collection of Abildgaardia schoenoides (As 13 - KLC230) supports their statement, as perian1h has been lost from all other members of the Abildgaardieae and appears to be a remnant feature retained in this sample. The uncontroversial placement of As 13 within the A. schoenoides s.s. limits, indicate that other characters for this specimen are consistent with the species; the only difference between the sample As 13 and the other OTUs is the presence ofperianth bristles. 133 The two samples AsS and As6 that form A. sp. aff. schoenoides 1 and the main group of A. schoenoides need to be compared to the TYPE for A. schoenoides collected by Robert Brown. The TYPE specimen for Fimbristylis squarrulosa was assigned by Ferdinand von Mueller for a collection from Victoria River and was not based on the Robert Brown specimen collected from the Gulf of Carpentaria. Both TYPE collections do not necessarily represent the same species and need to be compared. Brown's (1810) protologue does not give an adequate description to ascertain the similarity between his specimen and that used by Mueller. The TYPE specimen for Fimbristylis squarrulosa is distinctly perennial and corresponds to the specimens currently included as A. schoenoides s.stricto. The correct application of the name A. schoenoides to one of the delimited groups (A. schoenoides or A. sp. aff. schoenoides 1) must be resolved before relegating Fimbristylis squarrulosa as synonym of Abildgaardia schoenoides, or a new species described (i.e. AsS and As6). Although Abildgaardia sp. aff. pachyptera and Fimbristylis sp. aff. odontocarpa are from single collections, they form discrete entities and will be formally described subsequently in a valid publication. This study indicates that Abildgaardia should have generic status given that Fimbristylis is non-monophyletic, and that F. depauperata (from the TYPE section Fimbristylis section Fimbristylis) falls within a clade that is sister to the clade that contains the species of Abildgaardia. Given that Crosslandia and Abildgaardia form monophyletic groups and that Fimbristylis is non-monophyletic, it seen1S best at present to accept all three as genera. The relationships of Fimbristylis need to be assessed in a separate study. This analysis therefore supports the trend in accepting Abildgaardia and Fimbristylis as equally ranked genera (Gordon-Gray 1971; KIal 134 1971; Lye 1973; Goetghebeur and Coudijzer 1984, 1985; Goetghebeur 1986; Gordon-Gray 1995; Lye 1995; Kral and Strong 1999). Recent phylogenies based on DNA sequence data consistently retrieved Abildgaardia as sister to Fimbristylis (Muasya et al. 2000; Ghamkhar et al. 2005, in press; Simpson et al. 2005, in press). Sample size for Abildgaardia used in all studies was small with studies by Muasya et al. (2000) and Simpson et al. (2005, in press) incorporating only one representative, A. ovata. In Simpson et al. (Simpson et a1. 2005, in press) A. ovata and Arthrostylis aphylla formed an unresolved group. New combinations as proposed by Goetghebeur (1986) can now be formally published. Species that comprise the genus Abildgaardia in this study are presented here prior to formal publication. 135 Nomenclature of Abildgaardia Abildgaardia Yahl, Enumeratio Plantarum 2.296 (1806). LECTOTYPE: Abildgaardia monostachya (L.) Yahl 1. Abildgaardia nlacrantha (Boeck.) Goetgh. ex K.L.Clarke, K.L.Wilson, J.J.Bruhl comb. nov. ined. Fimbristylis macrantha Boeck. 2. Abildgaardia nlexicana (Palla) Kral Fimbristylis mexicana Palla Fimbristylis crassipes Boeck. (non Flora 41: 602 1858) 3. Abildgaardia odontocarpa (S.T.Blake) K.L.Clarke comb. nov. ined. Fimbristylis odontocarpa S.T.Blake 4. Abildgaardia oxystachya (F.Muell.) Goetgh. ex K.L.Clarke, K.L.Wilson, lJ.Bruhl comb. nov. ined. Fimbristylis oxystachya F .Muell. 5. Abildgaardia ovata (Burm.f.) Kral Cyperus monostachyos L. Abildgaardia monostachya (L.) Yahl, Fimbristylis monostachya (L.) Hassk., Iriha monostachya (L.) Kuntze, Fimbristylis ovata (Burm.f.) J.Kem 6. Abildgaardia pachyptera (S.T.Blake) Goetgh. ex K.L.Clarke, K.L.Wilson, J.J.Bruhl comb. nov. ined. Fimbristylis pachyptera S.T.Blake 7. Abildgaardia schoenoides R.Br. Fimbristylis squarrulosa F.Muell. (although need to check the A. schoenoides to confirm synonomy) TYPE for 8. Abildgaardia triflora (L.) Abeywickr. Cyperus triflorus L. Abildgaardia tristachya Yahl Schoenus cyperoides Retz. Finzbristylis triflora (L.) K.Schum. ex Engl. Abildgaardia triflora (L.) Lye (superfluous) Abildgaardia sp. aff. schoenoides 1 (AsS, As6) is pending comparison 9. ｧｮｩｭ｡ｾｬ with the A. schoenoides TYPE specimen. 10. A. sp. aff. odontocarpa, and 11. A. sp. aff. pachyptera are yet to be named.