Diversification across biomes in a continental lizard radiation SUPPORTING INFORMATION (Methods 1.1–1.3, Results 1.1–1.2, Figures S1–5, Tables S1–12) Methods 1.1 Bioinformatic workflow The cleaned reads from Gehyra variegata group samples were assembled into contigs using a pipeline described by Bragg et al. (2016). The pipeline performs a de novo assembly of each target exon for each sample using Velvet v. 1.2.08 (Zerbino and Birney 2008), merges contigs using CAP3 v. date 08/06/13 (Huang and Madan 1999), and truncates the contigs to the exon boundaries using Exonerate v. 2.2.0 (Slater and Birney 2005). If multiple contigs were assembled for a particular target, the putative ortholog was identified with a reciprocal best BLAST criterion. The cleaned reads of each individual were then mapped onto the corresponding contigs (to avoid reference bias) using Bowtie 2 v. 2.2.2 (Langmead and Salzberg 2012), and heterozygous sites were identified using GATK release gb82c674 (McKenna et al. 2010). We inserted “N” bases where genotyping quality (GQ) was less than 20. Retained samples have a mean coverage of 36.42 bp with 1127 loci on average (Table S1). For exons with multiple heterozygous sites, phase was determined from overlapping reads (GATK), and one haplotype was retained from each exon and individual for phylogenetic analyses. The assembled haplotype sequences of each locus were aligned using MACSE v. 1.2 (Ranwez et al. 2011), which returns frame shift-corrected nucleotide and corresponding amino acid multiple sequence alignments. Terminal gap symbols and special characters referring to the occurrence of stop codons and frame shifts were replaced by "N" and "X" for nucleotide and amino acid alignments, respectively. In order to preserve the correct frame, we used the amino acid alignments to remove all columns containing only gap symbols. Subsequently, amino acid alignments were searched for randomly similar sequence sections with ALISCORE v. 2.0 (Misof and Misof 2009; Kück et al. 2010) using the maximum number of pairwise sequence comparisons and the default sliding window size. This information was then applied to the nucleotide alignments to remove the corresponding affected sections with ALICUT v. 2.3 (Kück 2009). Trailing ends were trimmed in each nucleotide alignment by removing codon columns consisting of more than 50% gap symbols. Methods 1.2 StarBEAST2 “divide and conquer” strategy The basic procedure in a divide and conquer StarBEAST2 analysis is to identify diverse subclades with many closely related species, and pull them out for smaller, subclade-specific StarBEAST2 analyses. With the subclades removed, a “skeleton tree” is estimated with StarBEAST2 (Ogilvie et al. 2017), and then the trees sampled from the posterior distribution of the dated skeleton tree are re-integrated with the trees sampled from the posterior of each subclade analysis. Here, the skeleton tree consisted of 15 of the 44 Gehyra lineages (G. punctataA2, G. variegataC1, G. punctataB1, G. girloorloo, G. pseudopunctata, G. einasleighensis, G. lazelliLP, G. moritzi, G. variegataC2, G. punctataB4, G. pulingka, G. punctataA1, G. purpurascens, G. variegataA, and G. versicolor5), with two outgroup species (G. CYsp, G. dubia). These taxa were chosen either because (1) they represented a pair (or quartet) of lineages with a common ancestor that delimited a large monophyletic subclade, or (2) they were taxa with some ambiguity in phylogenetic position, so they were allowed to “float” and have their position be estimated (Fig. S3). We conducted StarBEAST2 analyses on four subtrees, with a tight age prior of Normal (mean = 1, standard deviation = 0.001). Each subclade analysis also included an outgroup taxon to improve the likelihood that the rooting of the subclade would be consistent with the subclade’s position in the skeleton tree analysis. The four subclades were: 1. The “variegata+punctataA” subclade, with 16 lineages: G. punctataA2, G. variegataB1, G. variegataB2, G. variegataB3, G. punctataA5, G. punctataA4, G. punctataA3, G. minuta1, G. minuta2, G. montium, G. pilbara1, G. pilbara2, G. punctataA1, G. variegataA, G. versicolor4, G. versicolor5. The outgroup was G. variegataC1. In the skeleton tree, the common ancestor of this subclade was identified as the common ancestor of G. punctataA2, G. punctataA1, G. variegataA, and G. versicolor5. 2. The “punctataB” subclade, with 7 lineages: G. punctataB1, G. punctataB2, G. punctataB3, G. punctataB4, G. punctataB5, and G. punctataB6. The outgroup was G. variegataC1. In the skeleton tree, the common ancestor of this subclade was identified as the common ancestor of G. punctataB1 and G. punctataB4. 3. The “lazelli” subclade, containing G. lazelli, G. lazelliLP, and G. pulingka. The outgroup was G. moritzi. In the skeleton tree, the common ancestor of this subclade was identified as the common ancestor of G. lazelliLP and G. pulingka. 4. The “nana group” subclade, containing 12 lineages: G. girloorloo, G. kimberleyi, G. multiporosa, G. nana1, G. nana2, G. paranana, G. nana4, G. pseudopunctata, G. granulum, G. nana7, G. pluraporosa, and G. occidentalis. The outgroup was G. lazelliLP. In the skeleton tree, the common ancestor of this subclade was identified as the common ancestor of G. pseudopunctata and G. girloorloo. Methods 1.3 Morphological data collection Ideally, 20 adults (preferably males) were measured per Gehyra variegata group lineage (average = 13). The two lineages within each of G. pilbara and G. minuta were combined to increase sample size, while two lineages known only from single specimens were excluded from analyses (G. pluraporosa and G. punctataB4). One person (LGA) measured 11 ecologically relevant morphological traits, following Doughty et al. (2012) with some additional variables (indicated with *): snout-vent length (SVL), trunk length (TrunkL; body length between legs), trunk width* (TrunkW; width between ventral skin folds of forelegs), foreleg length (ForelegL; elbow to base of wrist), hindleg length (HindlegL; knee to heel), head length (HeadL; anterior edge of ear to tip of snout), head depth (HeadD; deepest part posterior to eyes), head width (HeadW; widest part posterior to eyes), snout length* (SnoutL; anterior edge of eye to tip of snout), snout depth (SnoutD; deepest part anterior to eyes) and number of lamellae on fourth right toe pad (Lamellae; excluding distal wedge) (Table S3). Tail length was not measured as tails were often missing or regenerated; precloacal pore count was recorded but not used for morphological analyses. Measurements were made on the right hand side of the body (dorsal view) to the nearest 0.1 mm using digital callipers. Results 1.1 Lineage phylogeny comparison Both the RAxML and ASTRAL analyses (Fig. 2) and the StarBEAST2 species tree (Fig. 3) strongly infer paraphyly of G. punctata as currently described, with punctata A sister to a diverse variegata clade rather than punctata B. Furthermore, G. variegata is polyphyletic, with one pair of lineages (G. variegata C1–2) being highly divergent from the remainder. Within the variegata clade, “true” G. variegata (variegataA) is sister to G. pilbara and a closely related group of lineages including additional variants of G. variegata (variegataB1– 3), G. montium, G. minuta and G. versicolor. Some relationships within the main AAZ clade are not strongly supported, or are conflicting: variegataC1 and variegataC2 are in different positions in each tree. Relationships among lineages within the punctata A clade are poorly resolved, but in all three trees relationships among the punctata B lineages are well resolved. Results 1.2 Lineage diversity validation Nearly all (28 of 30) of the candidate lineages across the Gehyra variegata group AAZ species are found to be monophyletic in the RAxML concatenated tree (Fig. 2A). Using the tr2 gene tree concordance approach for lineage discovery, and conditional on the topology of the RAxML tree used, all candidate lineages are supported and some are divided yet further (e.g., G. lazelli, G. einasleighensis, G. punctataB6, G. montium, G. versicolor; Fig. S4). All candidate lineages are also confirmed as evolutionarily independent using BPP (Table S2); posterior support for each lineage is pp = 1.0, except some lineages within G. variegata (variegataB1, B2, B3: pp = 0.88, also with uncertain species tree topologies). Lineage diversity is especially high within G. variegata (six lineages) and G. punctata (11 lineages). Populations of G. lazelli with clearly different body sizes collapse into a single taxon, but only when each individual sample is treated as a separate species (Table S2; see also Sistrom et al. 2012). The G. versicolor4 and the G. versicolor5 lineages are supported as independent using BPP; however, we do not consider them genetically distinct as they group by geography rather than lineage in the RAxML phylogenetic analysis. Consistent with previous analyses (Sistrom et al. 2014), two sister lineages of G. minuta are nested within G. versicolor, and are also supported as two separate lineages using BPP. The two samples representing G. montium are paraphyletic with G. variegataB3, yet are still supported as separate using BPP; we treat these two lineages as distinct based on Kealley et al. (in press). Thus the 28 AAZ lineages, plus the 12 AMT nana group lineages (Moritz et al. 2018), gives a total of 40 lineages in the variegata group. Supporting Information References Bragg, J. G., S. Potter, K. Bi, and C. Moritz. 2016. Exon capture phylogenomics: efficacy across scales of divergence. Mol. Ecol. Resour. 16:1059–1068. Doughty, P., R. Palmer, M. J. Sistrom, A. M. Bauer, and S. C. Donnellan. 2012. Two new species of Gehyra (Squamata: Gekkonidae) geckos from the north-west Kimberley region of Western Australia. Rec. West. Aust. Mus. 27:117–134. Huang, X. and A. Madan. 1999. CAP3: a DNA sequence assembly program. Genome Res. 9:868–877. Kealley, L., P. Doughty, M. Pepper, S. Keogh, M. Hillier, and J. Huey. in press. Conspicuously concealed: revision of the arid clade of the Gehyra variegata (Gekkonidae) species group in Western Australia using an integrative molecular and morphological approach, with the description of five cryptic species.