Numerical analysis of the inter-relationships of some extinct and extant tax of Araucariaceae

The inter-relationships between extant and selected extinct taxa of Araucariaceae were explored using thirty morphological and anatomical characters. The sample of Araucariacae included all three extant genera of the family with three extinct species of Araucaria and the fossil genera Emwadea and Wairarapaia. The data were analysed using phenetic and cladistic methodology which revealed there was close agreement between the two when applied to extant taxa but not to extant plus extinct taxa. All analyses recognised that the araucarioid taxa with embedded seeds formed a group separate from the agathoid taxa whose seeds at maturity separate from the seed-scale. However, whereas the parsimony (cladistic) analyses failed to distinguish clades within Araucaria the phenetic analyses recognised four Sections within the genus and placed the three fossil species of Araucaria in Sect. Eutacta. The fossil genera Emwadea and Wairarapaia united with Agathis and Wollemia.  Araucariaceae, Wollemia, Emwadea, Wairarapaia, seed-cones, phylogeny. The description of Emwadea microcarpa Dettmann et al. (2012) based on permineralised seed-cones with preserved anatomy, from the mid-Cretaceous (late Albian) of western Queensland, adds to the data base of confirmed araucarian remains worldwide and supports the widely held view that during the Mesozoic and early Tertiary the family was more diverse than at present (Hill 1990; Cantrill 1992; Stockey 1994; Stockey et al. 1994; Pole 1995; Chambers et al. 1998; Hill & Brodribb 1999; Cantrill & Raine 2006; Dettmann et al. 2012) Whilst the araucarian affinities of many well preserved fossil seed-cones is not in doubt, their relationships with each other and with extant taxa has not been explored, until recently, by quantitative phenetic or cladistic analyses (Escapa & Catalano 2013). The extant Araucariaceae are represented by three genera Araucaria, Agathis and Wollemia (Farjon 2010), whose relationships have not been unambiguously established by cladistic studies based on gene sequencing data (Gilmore & Hill 1997; Stephanovic et al. 1998; Setoguchi et al. 1998; Codrington et al. 2002; Rai et al. 2008). Furthermore, these cladistic studies do not strongly support either the widely accepted four Sections into which extant Araucaria species were grouped by Wilde & Eames (1952) or the two Section grouping espoused by Laubenfels (1988). For example, whereas according to Setoguchi et al. (1998) Sect. Araucaria is the Clifford, Dettmann & Hocknull 28 Memoirs of the Queensland Museum | Nature  2015  59 sister group to the clade Sects Bunya and Intermedia according to Gilmore & Hill (1997) it is the sister group to Sect. Eutacta. Such disparity may be a consequence of the current Sections being based on morphological and anatomical data derived from extant taxa and so do not take into account the structure of Mesozoic seed-cones that may share characters with more than one extant Section of Araucaria (Stockey 1994; Stockey et al. 1994; Ohsawa et al. 1995). In view of the uncertainty of the interrelationship within Araucariaceae it was decided to investigate relationships between the three extant genera and five fossil taxa of the family incorporating morphological and anatomical data for all extant taxa and those fossils for which adequate descriptions are available. Both phenetic and cladistic analyses were undertaken. MATERIAL AnD METHODS Fourteen taxa, of which nine are extant, were selected for study. They were the genera Pinus, Podocarpus, Phyllocladus, Agathis and Wollemia together with the four currently accepted Sections of extant Araucaria (Wilde & Eames 1955). Following Farjon (2010) no subgeneric ranks were recognised within Agathis. The five fossil taxa, namely Emwadea microcarpa Dettmann, Clifford & Peters, Wairarapaia mildenhallii Cantrill & Raine, Araucaria mirabilis (Spegazinni) Windhausen, A. nipponensis Stockey, H. nishida & M. nishida and A. vulgaris (Stopes & Fujii) Ohsawa, H. nishida & M. nishida were chosen because the anatomical details of their ovule/seed-cones are available. Since the development of the seed-cones of most araucarian taxa has not been studied the homologies of their characters could not be determined directly. Instead, it was necessary to choose a theoretical model against which to make comparisons. The model accepted was that proposed by Florin (1944) as it provides a suitable framework for this purpose, notwithstanding it is predicated on the structure of mature cones. Allowance therefore has to be made for the considerable changes in structure that may occur following pollination (Tomlinson & Takaso 2002). For example, the ovules of young seed-cones of extant conifers are often initially orthotropous but are later inverted. Here it has been accepted that the ovules derive from an axillary complex which is subtended by a scale, and that each ovule is sessile or terminal on a more or less developed axis terminating in a pair of bracts fused marginally to form an integument around the nucellus. The axis may or may not bear lateral appendages below the integumentary bracts. If present, these appendages may generate secondary axes. Such a modular construction of the cone is supported by the recent studies of developmental genetics reviewed by Mathews & Kramer (2012). Although all ovules are postulated to arise directly from the axils of bracts or from axillary complexes, due to the activity of intercalary meristems at the complex or bract bases, they may appear to arise from the adaxial surface of the bract rather than its substanding axis. The interpretation of the bract-ovule complex can be resolved only through a study of its ontogeny. Although the pattern of vascular traces in the mature complexes may reflect their ontogeny, this assumption cannot be justified a priori because primordia, at least those of ovules, may develop from almost any tissue and generate their own vascular tissues (Bouman in Johri 1984). Furthermore, the formation of adventitious buds on wound callus tissues and the development of ovules from single epidermal cells, both of which may become vascularized (Romberger et al. 1993), suggests that the arrangement of the vascular tissues may not always be phylogenetically informative. However, the situation is much less clear with the interpretation of the ‘ligule’ which is restricted to araucarian seed scale where the ovule is always inverted. Although generally accepted as arising from the ovule stalk it has recently been reinterpreted as an extension of the chalaza (Dettmann et al. 2012) or a stigma (Krassilov & Barinova 2014). To distinguish between these hypotheses the development of the ligule must be determined, but as cautioned by Tomlinson & Takaso (2002, p. 1251), ‘If part-for-part Extant and extinct taxa of Araucariaceae Memoirs of the Queensland Museum | Nature  2015  59 29 equivalence is assumed, one has to invoke both heterochrony (i.e. changes in developmental timing among parts) and heterotopy (i.e. spatial transference of characters ), but only with considerable mani pulation of the original model.’ Due to such developmental flexibility, ‘plants become so transformed by meristematic invocation that to expect to be able to identify all structures of a putative ancestor is unrealistic.’ (Tomlinson & Takaso 2002, p. 1272). An example of heterochrony such as that postulated by Tomlinson & Takaso (2002) is the reversal of the sepaline and petaline whorls in Xyris and other monocot flowers with a double perianth (Remizowa et al. 2012). The seeds of many conifer species are accompanied by accessory structures variously described as teeth (Cryptomeria), appendages (Cunninghamia), arils (Taxus and Phyllocladus) or ‘ligules’ (Araucaria). As these structures, with the possible exception of the ligule, arise from immediately below the integuments they are accepted as homologous. The difficulty of interpreting characters is furthermore compounded by the lack of a definite sister group for the conifers (Taylor et al. 2009, pp. 870-871) which, in the literature, has led to conflicting reports of character states. The two following examples illustrate the problem. The cotyledon numbers of Araucaria species are given as 4 by Kindel (2001), 2-4 by Laubenfels (1988) or in 2 free and 2 fused and 2 fused pairs, with 4 free, or 4 fused into 2 pairs at the base. (Farjon 2010, p. 185). A similar diversity of ovule number per ovuliferous scale has also been reported for the genus. Whereas Araucaria species usually bear only one ovule per scale, both 2 and 3 ovules have been reported (Wilde & Eames 1955; Mitra 1927). numbers of ovules in excess of 1 per scale may be teratological malformations and so may be ignored if not regarded as atavistic. Because the best preserved fossil taxa are represented by ovuliferous cones these provided most of the characters studied. For each of the 14 taxa included in the analysis information, where available, was collated for 30 characters, of which at least one was known for each fossil taxon. This stricture was introduced so as to ensure the fossil and extant taxa are not ab initio members of unrelated taxa. The characters and their states are given in Appendix 1 and the taxa together with their character scores are listed in Appendix 2. Due to the paucity of character states available for the fossil taxa evidence of structure within the data matrix was investigated using only simple phenetic and phylogenetic methods. The former were based upon a Similarity Index (S.I.) defined as the percentage of characters shared by two taxa and so varies from zero when they share no character states to 100% when they are identical. Two types of phenetic analyses were under taken. One analysis constructed a Constellation Diagram in which those taxa with arbitrarily high similarity values were linked to each other; the other was the formation of a dendrogram using a simple distance measure and group average as the clusterin