New Australian Taxa
Open-access, online, rapid taxonomy
This is an unpublished draft. Please do not distribute.
Phylogenetic relationships of north-eastern Australian earless dragons (Agamidae: Tympanocryptis spp.), with description of three new species.
Kirilee Chaplin (1,2), Steve K. Wilson (3), Joanna Sumner (1), & Jane Melville (1*)
(1) Department of Sciences, Museums Victoria, Melbourne, Australia.
(2) University of Melbourne
(3) Queensland Museum
Earless dragons (Tympanocryptis spp.) are found in most environments across the Australian continent, with the 21 currently described species inhabiting a variety of ecological niches, from stony desert to tropical woodland or cracking clay savannahs. Recent work has indicated a revision of the taxonomy of north-eastern Australian earless dragons is required. Focussing on this geographic region, we use the mitochondrial ND2 gene (987bp) to investigate the phylogenetic relationships among currently described earless dragons and newly delimited putative species, with an assessment of broad biogeographic divisions. We found significant structure across the north-eastern Australian lineages, with deep divergence between lineages occurring in the inland Great Artesian Basin region and more coastal Great Dividing Range. Regional diversification is estimated to have occurred in the late Miocene with subsequent Plio-Pleistocene speciation. The diversity of the north-eastern Australian earless dragon group is consistent with that seen in other areas of historic aridification with habitat expansions and contractions, including the Western Australian Tympanocryptis species group. Based on these data, we describe three new species of Tympanocryptis from the cracking clay grasslands of the Darling Riverine Basin and Queensland Central Highlands regions, and the stony open eucalypt woodlands on the Einasleigh Uplands. The revision of these north-eastern Australian earless dragon species provides further taxonomic clarity within the Tympanocryptis genus.
Publication date and citation
Earless dragons (Tympanocryptis spp.) are small, terrestrial agamid lizards endemic to Australia (Melville and Wilson 2019), with 21 species currently described. Although the genus is found in a range of habitats throughout the continent, each species is restricted to a certain ecological niche, including the stony regions of Western Australia, the arid interior, the Nullabor Plain, and tropical and temperate grasslands and woodlands (Shoo et al. 2008, Stevens et al. 2010, Melville et al. 2014, Doughty et al. 2015; Melville et al. 2019a; Melville et al. 2019b).
Earless dragons have diversified and adapted to region-specific arid conditions throughout Australia since the late Miocene, with broad scale radiations across the continent and subsequent geographically localised divergence of lineages in specific habitats or ecological niches (Shoo et al. 2008). Habitat corridors were a significant factor in the speciation of many arid interior and desert-based taxa during the late Miocene and Pliocene (Chapple and Keogh 2004, Byrne 2008, Mossop et al. 2015, Jobson et al. 2017), including several of the Western Australian Tympanocryptisspecies (Pianka 1972, Shoo et al. 2008, Doughty et al. 2015). Similarly, climatic oscillations during the Plio-Pleistocene drove expansions of the inland arid grasslands and contractions of the coastal humid rainforests in north-eastern Australia (Stocker and Unwin 1989, Kershaw 1994, Travouillon et al. 2009). Dated fossil evidence of Tympanocryptisspp. has been found as far east as the Mount Etna caves near Rockhampton (Hocknull 2005), suggesting that arid habitat corridors or grassland expansions had reached near-coastal areas prior to the late Pleistocene.
Several studies have focussed on species delimitation in earless dragon (Smith et al. 1999, Shoo et al. 2008, Melville et al. 2014, Doughty et al. 2015, Melville et al., 2019a; Melville et al. 2019b; Chaplin et al. 2019). However, a revision of the taxonomy of north-eastern Australian Tympanocryptisspecies is still required. There are currently five earless dragon species described in this region; T. intimadistributed throughout the inland arid stony desert areas, T. tetraporophora in stony shrubland and clay grassland areas, and T. wilsoni, T. condaminensis and T. pentalineata on cracking clay grasslands near Roma, the Darling Downs and Normanton, respectively.
Recent work (Chaplin et al. 2019), incorporating genomics & morphological assessments using geometric morphometrics, provided strong evidence there are three undescribed lineages (A, B and C) occurring in the Darling Riverine Basin, Queensland Central Highlands and Einasleigh Uplands, respectively (Fig. 1). However, this genomic work does not provide insight into the evolutionary relationships of these putative species within the context of the whole Tympanocryptis genus. Thus, we undertook an assessment of the phylogenetic relationships among all Tympanocryptis species, using mtDNA (987bp), including these new lineages, with detailed focus on north-eastern Australia. Finally, we provide a comprehensive taxonomic treatment of these three putative species, including morphological quantification.
Materials & Methods
The three putative Tympanocryptis species (A, B and C) were sampled throughout their known ranges (Fig. 1). To determine the phylogenetic relationships within the genus we also incorporated samples from all other Tympanocryptis species groups, through additional samples from Museums Victoria, Queensland Museum and Western Australia Museum, or GenBank data from previously sequenced samples. Amphibolurus muricatus, Rankinia diemensis and Pogona vitticeps were used as outgroups. Locality data, museum registration numbers and GenBank accession numbers of all samples utilised in this study are listed in Supplementary Table 1.
Genomic DNA was extracted from either tail tissue or liver samples using the Qiagen Blood & Tissue Kit (Qiagen, Hilden, Germany) as per manufacturer guidelines. The mitochondrial gene ND2 (987bp) was amplified using the primers (Metf.1 and COIr.aga) and protocols described in Shoo et al. (2008) in a Bio-Rad MyCycler Thermal Cycler (Bio-Rad, California, USA). Negative controls were used in each PCR run. Amplification products were visualised on a 1.2% agarose gel with Sybr SAFE (Invitrogen, California, USA), then purified using ExoSAP-IT (Thermo-Fisher, California, USA) as per manufacturer guidelines, and sent to Macrogen (Seoul, South Korea) for sequencing. Sequence chromatograms were edited and aligned in Geneious 6.1.8 (Biomatters, Auckland, New Zealand).
Pairwise uncorrected genetic distances for the ND2 alignment were calculated in Mega7 (Kumar et al. 2016), with the codon frame set as the 3rd nucleotide position. Where more than one sample was present for a species, the clade was collapsed into a group prior to analysis. A Bayesian phylogeny of ND2 was produced using MrBayes (Huelsenbeck and Ronquist 2001) on the CIPRES Science Gateway (Miller et al. 2010), with two runs of four independent MCMC chains (each 50,000,000 generations long, sampled every 1,000 generations), under a GTR+I+G model with flat priors and no partitioning scheme (determined by the corrected Akaike Information Criterion (AICc) on PartitionFinder2 (Lanfear et al. 2017) on the CIPRES Science Gateway). Tracer v1.6 (Rambaut et al. 2014) was used to check for stationarity and convergence of the chain outputs. The trees were subject to a 25% burn-in in MrBayes, summarised and posterior probabilities obtained.
Seventeen meristic and metric characters previously used in Tympanocryptistaxonomy (Melville et al., 2014) and thought to be potentially diagnostic were recorded the three putative species. Electronic callipers were used for all morphological measures to the nearest 0.1mm and all bilateral counts and measurements were recorded on the left side (where possible) and analyses were run in SYSTAT v.13.2. A linear regression was performed on snout-vent length against all other measurements to standardise for size, and the residuals used in further analyses. A Discriminant functional analysis (DFA) was conducted to determine which morphological variables best discriminate between the three putative species.
All available specimens for the putative species (A, B and C) were examined. Seventeen meristic and metric characters previously used in Tympanocryptis taxonomy (Melville et al. 2014, Doughty et al. 2015) and thought to be potentially diagnostic were recorded (Table 3). Electronic callipers were used for all morphological measures to the nearest 0.1 mm and all bilateral counts and measurements were recorded on the left side only.
Previously published phylogenomic data, based on >8000 SNPs (Chaplin et al., 2019), provides strong evidence of the evolutionary independence of the putative species (A, B and C), however this published work did not provide the overall phylogenetic placement of these putative species within Tympanocryptis. Our phylogenetic analysis of all Tympanocryptis lineages and the three putative species (A, B and C) indicate all taxa are highly supported as monophyletic (Fig. 2), with uncorrected sequence divergence between taxa ranging from 3.4% to 15.5% (Supplementary Table 2). Genetic clades corresponded with previously published phylogenetic relationships (Fig. 2): the “Macra” group (T. macra); the Pebble dragons (T. centralis, T. intima, T. cephalus, T. fortescuensis, T. diabolicus, T. gigas and T. pseudopsephos); and the “Lineata” group (T. lineata, T. houstoni, T. pinguicolla, T. mccartneyi, and T. osbornei). A fourth lineage (the “Tetraporophora” group), which occurs in central and north-eastern Australia, is that covered in the current study.
Our analyses recovered two clades within the “Tetraporophora” group: one associated with the inland Great Artesian Basin (T. tetraporophora, Species A, T. condaminensis, T. wilsoni and T. pentalineata) and a second with the coastal Great Dividing Range (Species B and C). Mean uncorrected pairwise sequence divergence between the Great Artesian Basin (GAB) and Great Diving Range (GDR) clades was 10.6%, with deep divergences between putative species within each clade. The GDR clade is moderately supported as monophyletic (Bayesian posterior probability of 0.95; Fig. 2), with 13.5% sequence divergence between taxa B and C (Table 2). Conversely, the phylogenetic relationships between lineages within the GAB clade are not fully resolved (low Bayesian posterior probabilities, Fig. 2), although each putative species is strongly supported as monophyletic (Bayesian posterior probability of 1.00).
The GAB species group has relatively low sequence divergence between lineages compared with the GDR clade. T. condaminensis is the most basal taxon of this group with the lowest pairwise divergences of approximately 4% , while T. pentalineata has the highest levels of divergence, including 7.0% from the most closely related sister lineage to this species: T. wilsoni (Table 2, Fig. 2). Each of these species has similar intra-specific lineage divergence, observed in the length of the collapsed species clades, except for T. tetraporophora, which exhibits substantial intra-specific clade divergence (Fig. 2).
Significant differences in cranial skull shape has already been documented (Chaplin et al., 2019) in the putative species (A, B, and C), with Species A have a more spherical orbital and temporal region, while Species B and C have flatter medial regions. Additionally, Species C has a blunter snout that Species B, which has a flatter more elongated head. As well as these differences in cranial skull shape, visual inspection of external morphology of putative species (A, B and C) revealed clear differences in dorsal scalation. There are differences in number and shape of enlarged spinous scales and also whether or not scales are imbricate (further detailed in taxonomic section of this paper). In addition, ventral scales on the torso differ between the putative species with Species C having strongly keeled scales, while the other two have weakly keeled or smooth scales. There is also differentiation based on femoral pores, with Species A having two femoral pores (one on each side), which are absent in the other two species. Finally body patterning differed between putative species, with differences in the presence/absence and nature of the dorsolateral, vertebral and lateral stripes and also the width of the dark crossbands on the body (detailed fully in the taxonomic section).
We also undertook multivariate analyses of external morphological measures to discriminate between putative species. A Discriminant Function Analysis (DFA) of the three putative species (A, B and C) significantly distinguished groups based on external morphology (Wilks' λ8, 2, 56 = 0.071, F16,98 = 16.860, p<0.001). The DFA correctly classified 97% of animals (2/59 specimens) into the assumed a priori species classes, with only two Species A animals being incorrectly classified, one as Species B and one as Species C (figure 6). Canonical factor 1 (93.1% of variance), when corrected for within group variance, was associated with head width, neck width and hindlimb length, where Species C had wide heads, narrow necks and short hindlimbs, while Species B had narrow heads, wide necks and long hindlimbs. On Canonical factor 1, Species A was intermediate between Species B and C. Canonical factor 2 (6.9% of variance), which did not separate species significantly, was associated with tail length and neck width, where high scores had long tails and narrow necks.
Our data, which incorporates a mtDNA phylogeny of Tympanocryptis and external morphological analysis of putative species, combined with previously published evidence, including multilocus phylogeography, phylogenomics (SNPs), geometric morphometric analysis of cranial shape using micro x-ray CT scans and external morphological assessment of other Queensland species (Melville et al., 2014; Chaplin et al., 2019; Melville et al., 2019b), shows that the putative species (A, B, and C) are paraphyletic and the current taxonomy does not recognise these lineages. Using an Integrative Taxonomic assessment criteria (ITAX), previously used in Tympanocryptis (Melville et al., 2014), we identify the four existing named taxa and three additional unnamed species (figure 8). We used the criteria for assessment outlined in Melville et al. (2014), incorporating not-necessarily sympatric sister species, where at least two lines of independent evidence were required to delimit taxa. For all lineages covered in our ITAX assessment we provide at least four lines of evidence (mtDNA, SNPs, external morphology and geometric morphometric analysis of cranial shape). Based on this assessment, we recognise three new species of earless dragons: Tympanocryptis darlingensis sp. nov. (Species A), Tympanocryptis hobsoni sp. nov. (Species B) and Tympanocryptis einasleighensis sp. nov. (Species C).
Divergence times between these earless dragon species of north-eastern Australia are probably similar to other Tympanocryptis (Melville et al. 2007, Hugall et al. 2008, Shoo et al. 2008, Doughty et al. 2015). Using a rough estimate of 2% divergence per million years (Brown et al. 1982, Wilson et al. 1985), regional divergence times likely occurred during the late Miocene and between lineages during the Plio-Pleistocene (Fig. 2), which corresponds to previous estimates within the genus (Shoo et al. 2008, Melville et al. 2014, Doughty et al. 2015). During the Plio-Pleistocene, climatic fluctuations in north-eastern Australia would have shaped diversification and speciation of Tympanocrpytis within this region (Price 2012). The aridification and expansion of grasslands from inland to coastal areas resulted in the restriction of rainforest-specialist taxa to refugia, and vice-versa as the environmental oscillations continued (Martin 1982, Joseph et al. 1995, James and Moritz 2000). The extinction of many species occurred during this period of instability, although the diversification of more suitably adapted species (especially arid and semi-arid taxa) replaced these losses in the community (Hocknull et al. 2007). These shifts in community composition and species turnover have been well-documented throughout north-eastern Australia, including through dated fossil studies of faunal assemblages (Hutchinson and Mackness 2002, Hocknull 2005, Price et al. 2011). Fossil evidence of Tympanocryptis spp. near Rockhampton on the eastern coast indicate that previous earless dragon species or populations were widely distributed (Hocknull 2005), and the ranges of the extant Tympanocryptisspecies in north-eastern Australia are likely to be a relic of these Plio-Pleistocene climate oscillations.
The two GDR species form one of the most highly diverged clades within Tympanocryptis, with particularly long branches in the mitochondrial phylogeny indicating a deep divergence. These two species are the only north-eastern Australian species to be found on the eastern side of significant upland areas of the GDR (Fig. 3). While T. condaminensis is found on the Toowoomba plateau of the GDR at higher elevations (up to 500m above sea level) than either Species B or C, the Darling Downs plains slope steadily westwards to the lowlands of the Darling Riverine Plains and Mulga Lands, encompassing the distributions of T. wilsoni, Species A and T. tetraporophora, with no abrupt elevational variations throughout these areas. The lowlands on the western side of the GDR also extend north through the Mitchell Grass Downs to the Gulf Plains, including the distribution of T. pentalineata. Thus, it is probable that divergence within and between the GAB and GDR lineages is highly correlated with geology and topography of the region. Although there have been countless studies of geographic barriers of GDR taxa along latitudinal gradients (James and Moritz 2000, Chapple et al. 2011, Pepper et al. 2014), there is surprisingly little literature published on longitudinal phylogeographic patterns of GDR taxa with sister lineages in the western plains or GAB regions.
There is deep divergence and geographic isolation within the GDR clade. Species B exists on the Central Highlands of Queensland, with rugged escarpments and notable areas of high elevation (including the Drummond Range to the west, Carnarvon Range to the south, and Expedition Range to the east) encircling the species’ distribution. In contrast, Species C is distributed across the Einasleigh Uplands region of northern Queensland. The Einasleigh Uplands is a rugged plateau with several escarpments and varied elevation, extending from the Atherton Tablelands in the east to the Gulf Plains in the west (White 1965, Whitehead 2010). The deep molecular divergence and disjunct distributions of Species B and C are consistent with biogeographic breaks in north-eastern Australia, such as the Burdekin Gap and St Lawrence Gap, known to be associated with phylogeographic structure of a range of other herpetofauna (James and Moritz 2000, Chapple et al. 2011, Edwards and Melville 2011, Smissen et al. 2013). Although the paleoenvironmental fluctuations between arid and rainforest conditions in this region are likely to have contributed to the geographic isolation and molecular divergence of the two GDR species, it is possible that the deep branch lengths of these taxa are indicative of intermediate population extinctions, with only remnant lineages still extant, consistent with similar documented Plio-Pleistocene community composition shifts and species turnover (Hocknull 2005).
Figure 1. Distributions of north-eastern Australian Tympanocryptis species, including inset maps of locality of specimens included in current study for three putative species (A, Band C).