Phylogenetic relationships in the Eugongylini (Squamata: Scincidae): generic limits and biogeography

Taxon sampling

We obtained new squences for 139 samples, representing 76 species from 20 eugongylin genera, plus sequences from 80 species which we had previously used in two related studies (Smith et al. 2007; Chapple et al. 2009) from the Australian, Papua New Guinean, New Caledonian, New Zealand, Indian Ocean and Pacific Ocean region, which we combined with sequences from 66 species obtained from Genbank from these regions (Table 1). Our sampling, totalling 321 specimens from 198 species representing 39 eugongylin genera, encompasses all genera from this region except Geomyersia (two species from Papua New Guinea and the Solomon Islands), Geoscincus (a monotypic genus from New Caledonia not observed since the 1970s), Kuniesaurus (a monotypic genus from New Caledonia), Pygmaeascincus (three species from Queensland, two with very restricted distributions), Tachygyia (a monotypic genus from Tonga, presumed extinct) and Techmarscincus (a monotypic genus from Mt Bartle Frere in north Queensland, Australia) (Table 1). We also obtained sequences from Genbank representing two of the four African eugongylin genera (Leptosiaphos and Panaspis). We included samples representing as many as possible of the identified infrageneric lineages. For genera with multiple species, we used at least two species. Where there were nomenclatural issues, we obtained samples from the type species of nominal genera whenever possible. Where a number of species were represented within a lineage, we selected species that encompassed the extremes of morphological variation, or those that had been identified by previous studies as being the most genetically divergent. We included three of the Sphenomorphus species (bignelli, louisiadensis, minutus) that, based on morphological evidence, appear to be members of the Eugongylini rather than the Sphenomorphini. For a majority of species (118/198 eugongylin species), we obtained sequences from at least two individuals, as a check for errors in sequencing or identification.

Our tissue samples were obtained from museum collections in Australia (Australian Museum; Museum Victoria; Queensland Museum; South Australian Museum), New Zealand (Victoria University of Wellington; National Museum of New Zealand, Te Papa Tongarewa), and North America (Californian Academy of Sciences; Bishop Museum, Honolulu) (Table 1). To root our tree, we included 21 outgroup samples representing 19 species across 14 genera from other skink subfamilies and tribes covering the Acontiinae, Scincinae, Sphenomorphini, Lygosomini, Mabuyini, Sphenomorphini and Tiliquini (Table 1). Sequences from the remaining two tribes in the Lygosominae, the Ristellini (two genera) and Ateuchosaurini (one genus) were unavailable for the loci we used. For two genera (Ctenotus and Leptosiaphos), due to the lack of sequence data from the same species, we used composites of two species to create a single taxon for analysis (see Table 1).

DNA extraction, amplification and sequencing

Total genomic DNA was extracted from liver, muscle, toe or tail-tip samples using a Qiagen DNeasy Blood and Tissue Extraction Kit (Qiagen, Hilden, Germany). For each sample we sequenced portions of the ND2 mitochondrial gene (~600 bp) and two nuclear genes: recombination activating gene 1 (RAG-1, ~900 bp) and oocyte maturation factor (c-mos, ~550 bp). These regions were targeted because previous work on the Eugongylini has indicated useful levels of variability (Smith et al. 2007; Chapple et al. 2009). The primers used to amplify and sequence the three genes are provided in Table S1. PCR was conducted as outlined in Chapple et al. (2009). PCR products were purified using ExoSAP-IT (USB Corporation, Cleveland, Ohio USA). The purified product was sequenced directly using a BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems) and then analysed on an ABI 3730XL capillary sequencer.

Sequence data were edited using Geneious v5.4 (Drummond et al. 2011), roughly aligned by eye, and alignments optimised using MUSCLE (Edgar 2004). We translated all sequences to confirm that none contained premature stop codons. Sequence data were submitted to GenBank under the accession numbers provided in Table 1.

Among ingroup taxa, sequences for all three loci were available for 190 individuals (59%), for two loci for 32 individuals (10%) and for one locus for 99 individuals (31%). We obtained 777 of a possible 1026 sequences across all 342 individuals, for 76% coverage.

Phylogenetic and divergence time analyses

Our phylogenetic approach was designed to optimise the topological and temporal resolution of our mito-nuclear dataset given recent phylogenomic datasets of skinks. We started by downloading exon capture data focusing on Australian eugongylins (Brandley et al. 2015) and squamates (Burbrink et al. 2020) to build a consistent set of nuclear alignments and gene trees, and ultimately generate a time-calibrated backbone of the Scincidae. Data from different Anchored Hybrid Enrichment projects were combined using custom scripts which relied on metablastr to identify orthologous loci (blast_best_reciprocal_hit) (Benoit and Drost 2021), mafft to align them (--add, --keeplength) (Katoh et al. 2002), and AMAS to manipulate alignments (Borowiec 2016). We reconstructed individual genealogies for our exon-capture data (n = 272) under maximum-likelihood in IQTREE (Nguyen et al. 2015), allowing the program to assign the best-fitting model of molecular evolution using PartitionFinder, then perform 1000 ultrafast bootstraps (Hoang et al. 2018). We then estimated a species tree using the shortcut coalescent method ASTRAL III (Zhang et al. 2017) with IQTREE gene trees as input (Supplementary Fig. S1). To estimate divergence times among taxa we applied a series of primary and secondary calibrations (Table S2) and used the Bayesian divergence time software MCMCtre (Rannala and Yang 2007). We started by concatenating all loci and partitioning them into two partitions, first and second codons together, and third codons separately. We then used baseml to estimate approximate likelihoods and branch lengths before running mcmctree on the gradient and Hessian (in.BV file) for 10 replicate analyses (Reis and Yang 2011). We inspected mcmc files for stationarity and compared for convergence, then combined them using logCombiner (Bouckaert et al. 2019) and used this combined mcmc file to summarise divergence times on our tree (print = −1 in.ctl file).

We then turned our attention to the mito-nuclear dataset. We started by constraining a handful of nodes to match the topology of our ASTRAL exon capture species tree. We estimated the topology from our locus-partitioned concatenated mito-nuclear alignment using IQTREE, allowing PartitionFinder to choose and apply the preferred evolutionary model for each partition, and approximated topological support with 1000 ultrafast bootstraps (Supplementary Fig. S2). Following the methodology above, we then used this mito-nuclear species tree as input for MCMCtree, and split our alignment into four partitions (first and second mtDNA codons, third mtDNA codons, first and second nDNA codons, third nDNA codons). We applied a series of secondary calibrations (Table S2) and followed the procedure described above to estimate divergences between samples in our mito-nuclear dataset. This tree is presented twice: in Fig. 1 to highlight the intergeneric divergence times, and as a fully sampled phylogeny in Fig. 2.

Fig. 1.

Time-calibrated phylogeny of skinks with focus on topology and divergence times among the subfamily Eugongylini. This tree is based on the three-locus mito-nuclear dataset (ND2, RAG-1, c-mos) estimated using MCMCtree and shows the genus-level relationships within the subfamily as a subset of Fig. 2. Numbers at nodes indicate mean divergence times. Coloured bars at nodes indicate the 95% confidence estimates of divergence times. Nodes with red bars and a circle containing a letter show the placement of secondary calibrations used in the divergence time analysis (Table S2). Topological support values are shown on the tree in Fig. 2.

Fig. 2.

Fully sampled time-calibrated phylogeny of the tribe Eugongylini. The tree is split across two images---an inset on the left shows the corresponding portion of the tree. This tree is based on the three-locus mito-nuclear dataset (ND2, RAG-1, c-mos) estimated using MCMCtree and shows the species-level relationships within the subfamily. The topology of this tree was first estimated using IQTREE using a constrained guide tree based on a reliable phylogenomic hypothesis (see inset Phylogenomic Constraint). Topological support is indicated by the absence, presence, and colour of circles at nodes. Nodes with ultrafast bootstrap support values of >90 indicating strong topological support are not noted on the tree to simplify the visualisation. Nodes with moderate support (ultrafast bootstrap values of >70 < 90 are indicated by a grey circle, and poorly supported or equivocal nodes (ufbs <70) are noted by a white circle.

Monophyly of the Australian Eugongylini and its relationship to the Zealandia Lineage

Our study is evidence for a monophyletic Australian radiation, inclusive of Emoia s.s., within the Eugongylini (Figs 1 and 2) – this is the first time clear support has been provided for this relationship. The Zealandia Lineage is retrieved as the sister to the Sahul Radiation within the Eugongylini. Previous molecular studies on the New Caledonia (Smith et al. 2007) and New Zealand (Hickson et al. 2000; Chapple et al. 2009) eugongylin fauna hinted at this relationship, but limited sampling of Australian genera clouded demonstration of their affinity. A broader study of Australian skink phylogenetic relationships (Skinner et al. 2011) also recovered this relationship between the Australian and Zealandia eugongylin faunas, but that study included only a single New Caledonian and New Zealand species, and only six Australian species. These previous studies had shown the Zealandia Lineage to be monophyletic (Smith et al. 2007; Chapple et al. 2009; Skinner et al. 2011; Pyron et al. 2013), and we found comparable results with the broader taxonomic sampling employed in our study. Our results also retrieved some of the key phylogenetic relationships previously identified within Zealandia, including monophyly of the major New Caledonia radiation of endemic genera, apart from the genera Caesoris and Phasmasaurus as identified by Smith et al. (2007); and the New Zealand fauna as a sister lineage to the New Caledonia radiation as identified by Smith et al. (2007) and Chapple et al. (2009). However, our study placed Caesoris and Phasmasaurus progressively as the sister lineages to the New Caledonian + New Zealand radiation with high support (>0.90), rather than as the sister to the New Zealand species as in the study by Smith et al. (2007) with low support (<0.50 bootstrap), or with Phasmascincus nested within the New Caledonian radiation as recovered by Chapple et al. (2009) with high support (0.97) – Chapple et al. (2009) did not include Caesoris in their sampling. We found Oligosoma lichenigerum (O’Shaughnessy 1874), which lies between New Caledonia and New Zealand on the Norfolk Ridge and Lord Howe Rise, to be nested within a monophyletic New Zealand Oligosoma (though with low support for several earliest divergences for the genus), rather than as the sister to it, as found by Smith et al. (2007) with weak support (<0.50 bootstrap support), and in the more extensively sampled study of the New Zealand species by Chapple et al. (2009) with much higher support (0.96).

Relationships of the former Australian ‘Leiolopisma’

Early attempts by Greer (1974) at refining Leiolopisma Duméril & Bibron, 1839 resulted in the removal of window-eyed species that were outside the early concept of the Eugongylini (Scincella, Prasinohaema, Lipinia and Lobulia were erected or resurrected, and placed in what would become the Sphenomorphini), and provision of generic status for some obvious morphological clusters of window-eyed skinks within the Eugongylini (Carlia, Lampropholis, Anotis, Emoia, Panaspis), but Greer (1974, 1979) still retained ‘Leiolopisma’ as a speciose genus including taxa from Mauritius, Australia, New Caledonia and New Zealand. Hardy (1977) subdivided the New Zealand ‘Leiolopisma’ in resurrecting Cyclodina for a subset of species while still retaining ‘Leiolopisma’ for the remaining taxa, and proposed a detailed biogeographic history and dispersal routes for the New Zealand skink fauna. The concepts of ‘Leiolopisma’ as presented by Greer (1974) and Hardy (1977) were that it was paraphyletic, as they proposed the evolution of multiple genera and origins from within that genus.

Although Greer (1982a) continued to use Leiolopisma for the Australian species, he did suggest (following the lead of Rawlinson 1974, 1975) that the species known at the time fell into two groups on morphology, a small-scaled group with long limbs, low litter size, arboreal or saxicoline habits, and generally lacking chromatic hues (the ‘spenceri Species Group’), and a large-scaled group with short limbs, larger clutches/litters, terrestrial habits, and with chromatic hues usually present (the ‘baudini Species Group’). Subsequently, Greer (1989) used a single newly recognised osteological character, the fusion of the hemilaminae to the intercentrum of the atlas, to recognise a new grouping of eugongylin genera, the ‘Pseudemoia Group’. This character transected each of his previous spenceri and baudini Species Groups. The members of the former baudini Group were separated into two genera, Claireascincus for the cluster of species around entrecasteauxii (Duméril & Bibron, 1839), which possessed atlantal arch fusion and was hence grouped with a number of other Australian Eugongylini that comprised the Pseudemoia Group, and Eulepis (now Acritoscincus) for the cluster of species around trilineata (Gray, 1838) which lacked atlantal arch fusion. His former spenceri Group was also divided by this character, with Pseudemoia restored for the two species that showed atlantal arch fusion (spenceri (Lucas & Frost, 1894) and palfreymani (Rawlinson, 1974), thus restoring Rawlinson’s earlier concept of the genus but on different grounds), and Carinascincus for the Tasmanian snow skinks, reportedly lacking fusion.

Hutchinson et al. (1990) tested Greer’s (1974, 1979, 1980, 1982a) proposals with immunological comparisons of serum albumins, using microcomplement fixation techniques. They used four Australian ‘Leiolopisma’ species, entrecasteauxii, duperreyi (Gray, 1838), pretiosum (O’Shaughnessy, 1874) and palfreymani. The first two of these represented members of Greer’s (1982a) baudini Species Group, and the latter two were members of the spenceri Species Group. For these species they made two-way comparisons with three other genera, Emoia (represented by Emoia longicauda), Morethia (represented by M. adelaidensis (Peters, 1874)) and Lampropholis (represented by Lampropholis guichenoti (Duméril & Bibron, 1839) and Lampropholis challengeri (Boulenger, 1887b); it is worth noting that, based on the localities provided for their serum samples, their sample of the latter species represents what is now Saproscincus spectabilis (De Vis, 1888)). They found that the Australian Eugongylini sampled formed a monophyletic group, but the Australian ‘Leiolopisma’ were paraphyletic within this group, with Morethia and Lampropholis nested within it, and trilineatum (then as duperreyi) closely related to Morethia, supporting Greer’s (1980) early concept of affinities between these taxa. However, Greer’s (1982a) later hypotheses of the baudini and spenceri Species Groups were less congruent with the immunological results, as was the division created by the atlantal arch character of Greer (1989). In effect, the immunological study split the content of both Greer’s baudini and spenceri species groups, and was similarly not congruent with the taxa assigned to the Pseudemoia group on the atlantal arch character. To resolve the paraphyly inherent in the Australian Leiolopisma, Hutchinson et al. (1990) erected the genus Niveoscincus for the Tasmanian snow skinks (greeni (Rawlinson, 1974), microlepidotus (O’Shaughnessy, 1874), ocellatus (Gray, 1845), orocryptum (Hutchinson et al., 1988), pretiosus (O’Shaughnessy, 1874) and Leiolopisma palfreymani (in content Greer’s spenceri Group but without Leiolopisma spenceri itself), and also included Leiolopisma metallicum and L. coventryi (both formerly of the baudini Group). They resurrected the name Pseudemoia for several species in the baudini Group¹ (baudini (Greer, 1982a), entrecasteauxii, rawlinsoni (Hutchinson & Donnellan, 1988)) plus L. spenceri, and created the genus Bassiana for the remaining species that formerly comprised part of the baudini Group (duperreyi, platynota (Peters, 1881), trilineata). Complicating the nomenclature of this arrangement, Wells and Wellington (1985) had previously created the new generic name Acritoscincus for the species included in Bassiana by Hutchinson et al. (1990), and Carinascincus, Litotescincus, and Tasmascincus for the majority of species placed by Hutchinson et al. (1990) within Niveoscincus. There has been a gradual replacement of Bassiana with Acritoscincus in the herpetological literature, and a more recent transition from Niveoscincus to Carinascincus for the Tasmanian snow skinks.

For Pseudemoia, we included three of the six species, including the type species (Pseudemoia spenceri) and the type species of the synonym Claireascincus (P. entrecasteauxii). For Carinascincus we sampled six of the eight species, including the type species (C. greeni), which is also the type species of its synonym Niveoscincus, and also including the type species of two other synonyms, Litoteoscincus (Carinascincus metallicus), and Tasmascincus (Carinascincus palfreymani). We find high support for the species palfreymani as part of the group of snow skinks included in Carinascincus, and for Pseudemoia to include the type species spenceri and those taxa grouping around entrecasteauxii, rejecting earlier suggestions by Rawlinson (1974) and Greer (1989) that spenceri and palfreymani are congeneric. Further, these two genera are not sister-taxa within the Australian eugongylin radiation. This is congruent with phylogenies reconstructed by Brandley et al. (2015).

A close relationship of Acritoscincus (as Bassiana) with Morethia had been proposed on morphological evidence by Greer (1980, 1983), and on immunological evidence by Hutchinson et al. (1990) and Baverstock and Donnellan (1990). Greer (1980, 1983) further suggested that Proablepharus was also part of this lineage, as the sister taxon to Morethia (see also Fuhn 1969). Further, he later suggested that an unpublished cladistic analysis of morphological characters raised the possibility that the two genera were not reciprocally monophyletic (Greer et al. 2004). Early genetic studies either omitted representation of Proablepharus (Smith et al. 2007; Zheng and Wiens 2016), or were limited to inclusion of a single species (Skinner et al. 2011; Pyron et al. 2013). Couper et al. (2010) noted that character state incompatibility within Proablepharus rendered cladistic analysis of relationships difficult, and later (Couper et al. 2018) presented morphological (all described species) and genetic data (tenuis (Broom, 1896), reginae (Glauert, 1960) and kinghorni (Copland, 1947)) that identified paraphyly in the genus. As a result, Proablepharus was redefined to include only tenuis (type species of the genus) and reginae, and the genus Austroablepharus was created to accommodate kinghorni (type species for genus), and two recently described taxa, barrylyoni (Couper et al., 2010) and naranjicaudus (Greer et al., 2004; Couper et al. 2018).

Our results indicate that Acritoscincus, Morethia and Austroablepharus form a well-supported lineage within the Australian eugongylin radiation, with Proablepharus strongly supported as the sister, a relationship broadly in agreement with proposals initially based on morphological (Greer 1980, 1983) and immunological (Baverstock and Donnellan 1990; Hutchinson et al. 1990) evidence, and which has previously received some molecular support (Skinner et al. 2011; Ivan et al. 2022). We demonstrate that both Morethia and Acritoscincus are monophyletic, and support the results found by Couper et al. (2018) for polyphyly within earlier concepts of Proablepharus. However, our results offer only weak support for relationships among the three genera Australoblepharus, Morethia and Acritoscincus.

The Australian species (Carinascincus) coventryi was regarded by Greer (1979) as a member of Leiolopisma within the Eugongylus Subgroup, and later (Greer 1982a) as a member of the baudini Species Group (see above). One-way immunological comparisons by Hutchinson et al. (1990) found it to lie closest to the group of species now in Carinascincus, although it was outside the core of that group. It shared a live-bearing (viviparous) mode of reproduction with species in that group, but had morphological traits inconsistent with being placed with that group (it lacked the fused frontoparietal scales diagnostic of the group, and had lost the upper temporal opening on the skull, which was otherwise present in the group). Hence, Hutchinson et al. (1990) were uncertain of its relationship to this group of species, but tentatively placed it in the genus erected for them (then Niveoscincus). Their findings agreed in part with the earlier suggestion of Rawlinson (1975), who suggested affinities of coventryi with metallicus, and with placement of coventryi by Wells and Wellington (1985) in their genus Litotescincus. Greer later (1989) treated coventryi as congeneric with Harrisoniascincus zia, presumably following the earlier suggestion of Ingram and Ehmann (1981), which was largely based on colouration and ecological similarities. Stewart and Thompson (1998) found that the species did not possess the unique placental morphology of the species now assigned to Carinascincus, and this difference in placental morphology is mirrored by differences in placental physiology between coventryi and these taxa (Thompson et al. 2001). A sister relationship of coventryi with Carinascincus was also recovered by Brandley et al. (2015) and Ivan et al. (2022). Our results found that coventryi is not part of any described genus, but rather comes out as its own divergent lineage, most closely related to Anepischetosia and Pseudemoia, though with only moderate support, and as such warrants recognition as its own monotypic genus (see taxonomic implications section below).

Relationships of the ‘beta Group’ skinks

Greer (1974, 1979) proposed a subgroup of genera within the Eugongylini (variously as Group III of Greer (1974), and the Lampropholis Subgroup of Greer 1979) on the basis of a single morphological character, the development of a hook on the pterygoid bone to extend the secondary palate (the beta palate). Initially as Group III it consisted of the Australian genera Lampropholis, Carlia, Menetia, Notoscincus, and the Solomon Island genus Geomyersia, together with the African genera Afroablepharus, Cophoscincopus and Panaspis (the latter now further divided into Lacertaspis and Leptosiaphos – see Schmitz et al. (2005), and with Afroablepharus returned to synonymy of Panaspis – see Medina et al. 2016). Greer (1979) later removed Notoscincus from the Lampropholis Subgroup, transferring it to the Sphenomorphini (a placement that has received consistent support from later genetic studies: Reeder 2003; Rabosky et al. 2007; Skinner 2007; Skinner et al. 2011; Pyron et al. 2013; Zheng and Wiens 2016). Inclusion of the African genera by Greer in his Group III, or Lampropholis Subgroup, was not supported by the genetic studies of Pyron et al. (2013) and Zheng and Wiens (2016), although they did find these lie within a monophyletic group within the Eugongylini, but outside the Sahul Radiation. Our results similarly recover the African genera as basal within the Eugongylini.

Greer and Kluge (1980) recognised two lineages within Lampropholis (the delicata Group and the challengeri Group), of which the challengeri Group was later generically separated as Saproscincus (Wells and Wellington 1984; Greer 1989). Three lineages were later recognised in Saproscincus (Greer and Kluge 1980; Sadlier et al. 1993, 2005; Moussalli et al. 2005). Some studies have hypothesised (Schuster 1981) or provided evidence for recognition of Lampropholis and Saproscincus as sister taxa (Hutchinson et al. 1990 using microcomplement fixation), but other studies have not recovered such a close relationship exclusive of other taxa (Stuart-Fox et al. 2002; Moussalli et al. 2005; Smith et al. 2007; Pyron et al. 2013), and two-way immunological comparisons by Baverstock and Donnellan (1990) did not find any closer relationships between Lampropholis and Saproscincus species than to Niveoscincus, Pseudemoia or Carlia. We have been able to demonstrate monophyly of both Lampropholis and Saproscincus, consistent with previous studies (Greer 1989; Sadlier et al. 1993, 2005; Moussalli et al. 2005; Brandley et al. 2015; Ivan et al. 2022). As suggested by Schuster (1981), Hutchinson et al. (1990), Brandley et al. (2015) and Ivan et al. (2022), we found a well-supported sister-taxa relationship between Lampropholis and Saproscincus, contradicting the conclusions of several previous studies (Baverstock and Donnellan 1990; Moussalli et al. 2005; Smith et al. 2007; Pyron et al. 2013; Zheng and Wiens 2016).

Carlia has similarly been subject to generic division (Greer 1974; Ingram and Covacevich 1988; Stuart-Fox et al. 2002; Zug 2004, 2010; Dolman and Hugall 2008). The expanded concept of this genus created by Greer (1974) included a small species, burnetti Oudemans, 1894 (now foliorum De Vis, 1884, following reallocation of an earlier name of uncertain identity) with an ablepharine eye in a genus otherwise with a ‘spectacled’ lower eyelid. This species, which had for many years been placed in Ablepharus (Oudemans 1894; Copland 1949), had later been considered to be of uncertain affinity (Greer and Parker 1968; Fuhn 1969), but its transfer to Carlia placed it with a number of small species of similar phenotype, which were later grouped and formally recognised as the genus Lygisaurus (Ingram & Covacevich, 1988) on morphological criteria. However, Stuart-Fox et al. (2002), in a genetic analysis of relationships among Carlia species using a single mitochondrial gene (ND4), were unable to identify a monophyletic Carlia as sister to Lygisaurus. While the species assigned to Lygisaurus formed a discrete lineage, which also included one species that had recently been described in Carlia (Carlia parrhasius Couper et al., 1994) due to its lack of several of the purported diagnostic morphological characters of Lygisaurus, the Lygisaurus lineage was nested in Carlia, and there was little resolution of relationships among the remaining Carlia species. Hence, they returned Lygisaurus to the synonymy of Carlia. Zug (2004) presented a cladistic analysis of a limited number of morphological characters for this expanded Carlia, and was able to recover a clade representing 28 of the 30 species of Carlia (including parrhasius), but this clade formed an unresolved polytomy with the other two species of Carlia and the two Lygisaurus species, although Zug’s analysis did not include one internal character (number of premaxillary teeth) that had been identified as a putative synapomorphy of Lygisaurus by Ingram and Covacevich (1988), and his analysis found very low support for many of the clades recovered within Carlia. Indeed, removal of two characters that showed some intraspecific variation resulted in nearly complete collapse of the cladogram into a large polytomy (including the outgroups) with only two small clades identified among species assigned to Carlia. A later genetic analysis, incorporating wider species sampling and more genes (two mitochondrial, two nuclear introns: Dolman and Hugall 2008) provided much more resolution, with three groups recognised among Carlia sensu lato. Although their analyses were unable to unequivocally determine the relationships among these three lineages, the high support for each lineage led to them recognising each as a distinct genus: Carlia, Liburnascincus (for several saxicoline species that had formerly been placed in Carlia), and Lygisaurus (including parrhasius). Similar results were obtained by Ivan et al. (2022), though with a lesser range of species. While Dolman and Hugall (2008) identified lower support for relationships within Carlia, Zug (2010) mapped morphological characters onto their gene tree to diagnose nine species groups, one of which (the Carlia peronii Species Group) was unsampled in the earlier genetic analyses. Subsequently, Pyron et al. (2013), while using mostly the same sequences as Dolman and Hugall, were not able to replicate the results of that study. Instead, they found that Lygisaurus, while still monophyletic, was nested within a broader paraphyletic Carlia, with Liburnascincus remaining outside this, a result also recovered by Zheng and Wiens (2016). Pyron et al. (2013) also found the species timlowi Ingram, 1977, initially described in Menetia, later transferred to Lygisaurus by Ingram and Covacevich (1988), then back to Menetia by Greer (1991), to be nested among the Carlia group genera, but not within Lygisaurus. This finding prompted Couper and Hoskin (2014) to create the genus Pygmaeascincus, with Menetia timlowi as type species for the genus, and also including the species Pygmaeascincus koshlandae and Pygmaeascincus sadlieri described by Greer (1991) which had formerly resided in Menetia, but identified as close to timlowi by Greer. Pygmaeascincus (represented by two of the three species) was recovered as sister to the other genera within the Carlia group by Ivan et al. (2022), with high (>95) bootstrap support.

Our study provides strong support for the reciprocal monophyly of the Carlia group, comprising Carlia, Lygisaurus and Liburnascincus, with the relationships among the genera well resolved. This result confirms previous suggestions, from Ingram and Covacevich (1988) and Dolman and Hugall (2008), for the recognition of multiple genera within the Carlia group.

The relationships of Menetia have been particularly problematic, with a lack of concordance between its assignment to beta palate group of skinks of Greer on morphological criteria (Group III of Greer (1974), and the Lampropholis Subgroup of Greer (1979)), and its affinities to other genera in genetic studies. The genus was resurrected from within Ablepharus by Fuhn (1969), and was suggested by Greer (1974, 1979) to be derived from Carlia, and later (Greer 1991) as closely related to Carlia. At the time of these initial studies, the genus was monotypic, with the sole species, Menetia greyii Gray, 1845, having a unique and very distinctive head scalation. However, the discovery of additional species (Storr 1976; Ingram 1977; Rankin 1979; Sadlier 1984) required a broadening of the morphological diagnosis and the loss of some of these apomorphies. Immunological studies by Baverstock and Donnellan (1990) found M. greyii to be closest to Acritoscincus and Morethia, and distant to the other beta-palate skinks sampled (Carlia, Saproscincus, Lampropholis). Smith et al. (2007) did not recover a relationship of M. greyii to the other Australian beta palate genera sampled (Lampropholis, Saproscincus and Lygisaurus) but did not include Carlia in their analysis. Instead, they found it was more closely allied to a species of Emoia, Emoia cyanura. Pyron et al. (2013) added a second species, Menetia alanae Rankin, 1979, and similarly identified a sister-group relationship of Menetia s.s. to a more extended group of Emoia, Their findings were replicated by Zheng and Wiens (2016), but with much lower support. However, those analyses were limited to a single shared gene (ND4) for the Menetia species included in that study. Most recently, Ivan et al. (2022) have included four species of Menetia. Their main published tree (their Fig. 1) recovered Menetia as sister to Cryptoblepharus. However, although the tree provided in their Accessory fig. 3 found strong (>95% bootstrap) support for a relationship of Menetia with the genera Acritoscincus, Austroablepharus, Morethia and Proablepharus, they did not include Emoia in their analysis.

Our study, which included three species, M. greyi, M. alanae and M. surda Storr, 1976, placed these as the sister to a broadly sampled Emoia s.s. within the Sahul Radiation, similar to the affinities suggested by Smith et al. (2007), Pyron et al. (2013) and Zheng and Wiens (2016), and not as having any relationship to previously suggested genera (e.g. Saproscincus, Lampropholis, Carlia, Lygisaurus, Liburnascincus, Acritoscincus, Austroablepharus, Morethia, Proablepharus or Cryptoblepharus). These results again indicate two independent evolutions of the beta-palate within the Australian radiation of eugongylin skinks.

Relationships of the monotypic genera currently recognised among the Australian Eugongylini

Our phylogenetic analyses resolve the taxonomic uncertainties related to the monotypic genera Anepischetosia Wells & Wellington, 1985 (type species S. maccoyi Lucas & Frost, 1894), Eroticoscincus Wells & Wellington, 1984 (type species L. graciloides Lönnberg & Andersson, 1913) and Harrisoniascincus Wells & Wellington, 1985 (type species Leiolopisma zia Ingram & Ehmann, 1981). Mittleman (1952) and Greer (1974) on morphological criteria placed the species graciloides and maccoyi in Anotis, a genus at that time otherwise restricted to New Caledonia. Schuster (1981) suggested there were undescribed members of this genus in north Queensland, but his statements have not been subsequently corroborated. Czechura (1981), in reviewing graciloides, noted the homonymy of Anotis with an insect taxon, and resurrected Nannoscincus for this genus. Sadlier (1987) included maccoyi as part of Nannoscincus, but excluded graciloides from that genus, noting that it was ‘only distantly related’. Further explanation was provided by Sadlier (1990) with a rediagnosis of Nannoscincus, noting that graciloides showed a different pattern of phalangeal reduction to Nannoscincus, although its further affinities were left unresolved. The affinities of graciloides have since been largely unexplored. Wells and Wellington (1984) proposed the generic name Eroticoscincus for it, and Hutchinson (1993) suggested that it was phenetically similar to Saproscincus czechurai, though lacking the beta-palate of the species of that genus, and tentatively transferred it to Saproscincus. In contrast, Greer (1991) considered graciloides to form a lineage with Carlia, Lygisaurus and Menetia, based on a shared phalangeal formula of a loss of the first digit of the manus. The continued retention of maccoyi in Nannoscincus by Sadlier (1990), albeit in a different subgenus (Nannoseps) to the other species, created a biogeographic anomaly, this being the only New Caledonian genus with Australian representatives. Smith et al. (2007) confirmed that Nannoscincus was part of an otherwise endemic New Caledonian clade, but did not include Nannoscincus maccoyi in their study. Sadlier et al. (2006b) resurrected the Wells and Wellington (1985) genus Anepischetosia (itself a replacement name for the preoccupied Anepischetos Wells & Wellington, 1984) to replace his subgenus Nannoseps for the Australian species maccoyi, but did not otherwise offer any argument in support of their action in raising Anepischetosia to generic status

Hutchinson et al. (1990), with only one-way immunological comparisons possible for this part of their analysis, found that ‘Nannoscincus’ maccoyi, together with two other problematic species, Leiolopisma zia Ingram & Ehmann, 1981 and L. jigurru Covacevich, 1984, represented relatively early offshoots from a group also including Carinascincus (as Niveoscincus) and the beta-palate skinks Lampropholis and Saproscincus, but were uncertain about their placement within this group. They erected monotypic genera, Cautula for zia and Bartleia for jigurru, although earlier names, Harrisoniascincus and Techmarscincus, were available respectively for these species (Wells and Wellington 1985; Couper et al. 2006) and have subsequently been applied. The comparisons made by Hutchinson et al. (1990) did not include taxa within other Australian eugongylin genera that were a part of the Lampropholis Subgroup (Carlia, Lygisaurus, Liburnascincus and Menetia), nor did they include graciloides, nor any of the New Caledonian Nannoscincus, and hence were limited in hypothesising affinities, although they did suggest that maccoyi and zia might be related. Subsequent molecular studies have variously reported a close relationship of Harrisoniascincus zia to Lampropholis (Moussalli et al. 2005), or the Carlia group (Smith et al. 2007; Pyron et al. 2013; Zheng and Wiens 2016), but these studies had only sparse sampling of other Australian eugongylin genera. Schuster (1981) suggested that zia was closer to Lampropholis and Saproscincus than to maccoyi. Brandley et al. (2015) recovered zia as either basal (along with Acritoscincus) to other Australian eugongylins sampled, or sister to Pseudemoia at the base of the tree in different analyses.

Our study clearly demonstrates that A. maccoyi is not part of the New Caledonian Nannoscincus genus, but rather represents a divergent lineage within the Sahul Radiation. Our analyses place it within a moderately supported group that also includes (Carinascincus) coventryi and Pseudemoia, with a divergence time to its nearest relative (Pseudemoia) of ~28 mya, an age over twice that of the species within Pseudemoia, and with a morphology and physiology very different to Pseudemoia. Our results support the retention of Anepischetosia as a monotypic genus within the tribe. We also found Eroticoscincus graciloides and Harrisoniascincus zia each represent a highly divergent lineage within the Sahul eugongylin radiation, with a moderately supported sister-taxa relationship. Given the extent of genetic divergence between the two taxa (dating back ~22 mya) and the extent of morphological differentiation, we here retain them as monotypic genera.

Monophyly of Emoia

The intra- and interspecific relationships of Emoia Gray, 1845 have received only limited morphological and genetic investigation. Greer (1974) was uncertain of the affinities of Emoia, although he treated it as a single unit, and suggested an origin either from a Eugongylus-like stock or a L. spenceri-like stock. Thereafter the genus received only cursory mention by Greer (1979) as a member of the Eugongylus Subgroup of the Eugongylus Group. The genus was defined by Brown (1991) on morphological criteria that included a suite of plesiomorphic characters, either at the level of the Lygosominae generally or within the larger part of the Eugongylini. Within Emoia, Brown (1991) proposed eight species groups. Five of these (the adspersa, atrocostata, baudini, physicae and cyanogaster Groups) shared a common synapomorphy in osteology, two of the remaining three species groups (the cyanura and monotypic ponapea Groups) showed one or more unique apomorphic states in osteology within the genus, and the remaining group (samoensis Group) was identified largely on the basis of plesiomorphic features, and as being the only species group within the genus that does not have a fixed clutch size of two for its species (Greer 1968; Schwaner 1980; Brown 1991). However, the robustness of these species groups is confounded by Brown (1991) not indicating which species he had examined osteologically, and it is likely that not all species were examined for the diagnostic criteria to be sure of correct group allocation. Regardless, the work by Brown remains the only comprehensive overview of the genus.

Early immunological (Hutchinson et al. 1990) and genetic studies (Austin and Arnold 2006; Smith et al. 2007; Skinner et al. 2011) were limited in their utility to interpret the intra- and intergeneric relationships of Emoia by consequence of including only one or two species in their sampling, apparently assuming monophyly for the genus and/or as a legacy of the unavailability of samples for analysis. Hutchinson et al. (1990) found the Emoia cyanogaster Group (represented by E. longicauda) to lie well outside the group of Australian genera sampled. Austin and Arnold (2006) similarly found the E. cyanura Group (represented by Emoia impar) and the Emoia physicae Group (represented by the nominotypic species) to be weakly supported as sister to Mauritian Leiolopisma outside the Australian and New Zealand endemic eugongylin genera. Smith et al. (2007) did not retrieve a monophyletic Australian lineage, and found the E. samoensis Group (represented by Emoia loyaltiensis) to be sister to Leiolopisma s.s., and the E. cyanura Group (represented by E. cyanura) to be sister to Menetia (represented by M. greyii), but with only moderate Bayesian posterior probabilities (0.75 and 0.70 respectively) and bootstrap support (<50%). However, these groupings each belonged to a branch in a polytomy of several across which the non-Zealandia skink genera sampled were distributed. While Smith et al. (2007) gave no evidence for any close relationship of the E. cyanura Group or E. samoensis Group to each other, their analysis did suggest a relationship between the E. cyanura Group and Menetia which is reflected in the later studies by Pyron et al. (2013), Zheng and Wiens (2016) and our study, and for the affinities of the E. samoensis Group to lie outside the endemic Australian and Zealandia Lineage genera.

The study by Pyron et al. (2013) provided a more extensive coverage of Emoia taxa than earlier genetic studies, and included 13 species across six species groups. They found a relationship between the samoensis Group species E. loyaltiensis (using the sequences from Smith et al. 2007) and a group that included Leiolopisma s.s. and several African region genera (Panaspis, Lacertaspis, Leptosaiphos). However, two species of the samoensis Species Group not previously sampled, E. concolor (Duméril & Duméril, 1851) and E. tongana (Werner, 1899), were external to the Eugongylini, instead being sister to the Lygosomini. A similar placement, this time with these two species embedded within the Lygosomini, was identified by Zheng and Wiens (2016). This unexpected result may be an artefact of sampling, as the latter two species were represented in the analysis by only a single gene, which was not one of the three available for E. loyaltiensis. Both studies also found the other species of Emoia included, representing those previously studied by Austin and Arnold (2006), Smith et al. (2007) and Skinner et al. (2011), together with three other species in the cyanura Group (Emoia schmidti, Emoia pseudocyanura and Emoia isolata), and representatives of the Emoia baudini Group (Emoia jakati) and Emoia atrocostata Group (nominotypic species), to form a monophyletic group. This group clustered with Cryptoblepharus and Menetia as sister to the Australian group of genera, with the New Zealand and New Caledonian radiations external to this, albeit with relatively low support for these higher relationships.

A more recent genetic sudy focusing on the intra- and interspecific relationships of E. atrocostata (Richmond et al. 2021) included two species in the E. atrocostata Species Group not included in previous studies, boettgeri and nativitatis. It also included members of Brown’s (1991) adspersa Species Group, adspersa (Steindachner, 1870) and lawesii (Günther, 1874), also not represented in previous studies, and as an outgroup utilised several taxa in the cyanogaster Species Group, cyanogaster (Lesson, 1829), kordoana (Meyer, 1874), and longicauda (the latter two of which had not been included in previous studies). This study found E. atrocostata to be paraphyletic with respect to other members of the species group sampled, and the atrocostata Species Group to be polyphyletic with respect to inclusion of species from the adspersa Species Group. It also found the three outgroup species included from the cyanogaster Species Group to form a well supported lineage sister to the lineage in which the atrocostata and adspersa Group species resided. While this study lacked representation of taxa from the other five species groups within Emoia identified by Brown, or other taxa in the Eugongylini, its relevance here lies in the extent of apparent genetic affinity between the atrocostata Species Group and the adspersa Species Group, the latter of which has been regarded as unusual within the genus in having much smaller body scales than species in other groups within Emoia.

Within Emoia, we sampled 16 species representing six of the eight putative species groups of Brown (1991), including the type species of Emoia (atrocostata), and three species in the E. samoensis Group (E. loyaltiensis, Emoia nigra and Emoia sanfordi), the nominotypic species of the Emoia ponapea Group, and two species in the E. cyanura Group (Emoia rufilabialis and Emoia taumakoensis), not previously reported. We found clear evidence that the widespread and diverse Emoia genus is not monophyletic, but rather represents two major lineages within the Eugongylini. One lineage, represented in our study by three (of 13+) species in the E. samoensis Group (E. loyaltiensis, E. nigra, E. sanfordi), represents a basal lineage within the tribe outside the Australian radiation. For E. loyaltiensis, this result is consistent with the results of previous molecular studies by Smith et al. (2007), Pyron et al. (2013) and Zheng and Wiens (2016), although as noted above the latter two studies found two other taxa in the species group to be placed within the Lygosomini.

The other lineage found in our study, represented here by three (of six) taxa from the E. atrocostata Group (E. atrocostata, Emoia boettgeri and Emoia nativitatis), one (of 21) species from the E. baudini Group (E. jakati), two (of six) species from the E. cyanogaster Group (E. longicauda and Emoia tetrataenia), the sole species in the E. ponapea Group, and eight (of 13) species from the E. cyanura Group (Emoia caeruleocauda, E. cyanura, E. impar, E. isolata, E. pseudocyanura, E. rufilabialis, E. schmidti and E. taumakoensis), forms part of the Sahul Radiation of Eugongylini. For some taxa, their inclusion in this lineage and within the Australian eugongylin radiation has previously been suggested (Austin and Arnold 2006; Smith et al. 2007; Skinner et al. 2011; Pyron et al. 2013; Zheng and Wiens 2016), but their affinitites within this radiation have varied. Our study has provided strong support for these species as a monophyletic lineage within the Sahul Radiation, and by virture of inclusion of the type species, as representing Emoia s.s. Within this redefined Emoia s.s. two well-supported sublineages were retrieved, one comprising taxa from the E. cyanura + E. ponapea Groups, and the other taxa from the E. atrocostata + E. baudini + E. cyanogaster Species Groups. However, the content of this redefined Emoia s.s., and of the sublineages within, may change with the inclusion of other taxa from species groups not (Emoia adspersa Species Group, E. physicae Species Group) or poorly (E. baudini Species Group) represented in our study.

Our study places Menetia as the sister genus to the Emoia s.s. lineage, a relationship consistent with the previous findings of Smith et al. (2007), Pyron et al. (2013) and Zheng and Wiens (2016), but with broader taxon sampling and strong support.

Cryptoblepharus as part of the Sahul Radiation

Cryptoblepharus Wiegmann, 1834 was suggested by Fuhn (1969) to be related to Emoia, and Greer (1974) suggested it evolved from within Emoia, particularly from the E. cyanura Species Group (Greer 1983). However, Greer (1989) later considered Cryptoblepharus to be part of his Pseudemoia Group, along with most of the other Australian eugongylins, with Emoia outside that lineage, a hypothesis incompatible with his previous view. Neither of the genetic studies by Austin and Arnold (2006), or Smith et al. (2007) found any evidence for a relationship of Cryptoblepharus with Emoia species in their analyses. Horner (2007) provided the most comprehensive modern review of the genus, based primarily on morphology, but also utilising genetic data from allozyme electrophoresis (Horner and Adams 2007). Horner did not speculate on the relationships of Cryptoblepharus, but suggested an origin of the genus in south-east Asia. Pyron et al. (2013) and Zheng and Wiens (2016) did recover a sister-taxon relationship between Cryptoblepharus and Emoia s.s. + Menetia. Blom et al. (2016) provided an assessment of genetic relationships of 24 Australian species to elucidate the mechanisms driving evolution across the continent, and extended the study more broadly in taxonomic and regional representation (Blom et al. 2019) with regard to trans-oceanic dispersal. Both studies appear to have assumed monophyly of Cryptoblepharus, and did not extend beyond exploring the intrageneric relationships of the genus.

Our study provides strong evidence for the monophyly of Cryptoblepharus as sampled (although we did not include any of the Indian Ocean taxa) and strong support for this genus being part of the Sahul eugongylin radiation. The genus was found to be part of a group that includes the beta-palate genera (Saproscincus/Lampropholis/Lygisaurus/Liburnascincus/Carlia), diverging from them ~30 mya. This result suggests that the genus originated in Australia before spreading out across the Pacific.

Affinities of the Sphenomorphus species S. aignanus, S. louisiadensis, S. minutus and S. bignelli to the Eugongylini

These four species, by virtue of their generic placement within Sphenomorphus, are currently part of the Sphenomorphini, despite having been recognised as members of the Eugongylini for more than half a century (Greer and Parker 1968; Greer 1974, 1977, 1979). Our study provides confirmation that three ‘Sphenomorphus’ species (S. louisiadensis, S. minutus and S. bignelli) are indeed part of the Eugongylini, a result consistent with previous morphological evidence (Greer and Parker 1968; Greer 1977, 1979). Our unpublished ND2 data indicate that S. aignanus is also part of this Sphenomorphus clade within the Eugongylini. Our study found these four Sphenomorphus species were the sister to Eugongylus, and provides evidence for this group as one of several basal lineages within the tribe.

Geographic origins of the Eugongylini and the ages of the various lineages

The Eugongylini is one of seven tribes (Lygosomini, Ateuchosaurini, Eugongylini, Mabuyini, Ristellini, Sphenomorphini, and Tiliquini) in the subfamily Lygosominae, as recently refined by Shea (2021). While monophyly of the Lygosominae, and of most tribes, is well supported in the taxon-rich study of genetic relationships within the Scincidae by Pyron et al. (2013)