Evaluating the performance of anchored hybrid enrichment at the tips of the tree of life: a phylogenetic analysis of Australian Eugongylus group scincid lizards

Abstract

Background: High-throughput sequencing using targeted enrichment and transcriptomic methods enables rapid construction of phylogenomic data sets incorporating hundreds to thousands of loci. These advances have enabled access to an unprecedented amount of nucleotide sequence data, but they also pose new questions. Given that the loci targeted for enrichment are often highly conserved, how informative are they at different taxonomic scales, especially at the intraspecific/phylogeographic scale? We investigate this question using Australian scincid lizards in the Eugongylus group (Squamata: Scincidae). We sequenced 415 anchored hybrid enriched (AHE) loci for 43 individuals and mined 1650 exons (1648 loci) from transcriptomes (transcriptome mining) from 11 individuals, including multiple phylogeographic lineages within several species of Carlia, Lampropholis, and Saproscincus skinks. We assessed the phylogenetic information content of these loci at the intergeneric, interspecific, and phylogeographic scales. As a further test of the utility at the phylogeographic scale, we used the anchor hybrid enriched loci to infer lineage divergence parameters using coalescent models of isolation with migration.

Results: Phylogenetic analyses of both data sets inferred very strongly supported trees at all taxonomic levels. Further, AHE loci yielded estimates of divergence times between closely related lineages that were broadly consistent with previous population-level analyses.

Conclusions: Anchored-enriched loci are useful at the deep phylogeny and phylogeographic scales. Although overall phylogenetic support was high throughout the Australian Eugongylus group phylogeny, there were nonetheless some conflicting or unresolved relationships, especially regarding the placement of Pseudemoia, Cryptoblepharus, and the relationships amongst closely-related species of Tasmanian Niveoscincus skinks.

Figure 1

The distribution of parsimony-informative characters (PICs) for anchored hybrid enrichment (AHE) markers. a) The distribution of parsimony-informative characters (PICs) across the length of the 415 AHE loci for the full-taxon, Lampropholis, and Saproscincus data sets. Colored dots indicate how many total loci had at least one PIC at that nucleotide position. The zero position is the center of the anchor sequence for all loci. Grey dots indicate the distribution of locus length by plotting number of loci (right y-axis) with any nucleotide at that position. b) The distribution of PICs across the length of the 415 AHE loci in the full-taxon data set “normalized” by dividing the number of PICs at each nucleotide position by number of alignments of that length.

Figure 2

Results of the (a) RAxML maximum likelihood and (b) STEAC species tree analysis of 415 anchored enriched (AHE) loci for all taxa. Numbers above or below nodes indicate bootstrap proportions from 1000 pseudoreplicates. Outgroup not shown.

Figure 3

Results of the (a) RAxML and (b) STEAC analyses of all 1650 loci selected by our transcriptome mining (TM) analysis and the 415 anchored enriched (AHE) loci with taxa pruned the 10 ingroup and one outgroup taxa present in the TM data set. Numbers above the nodes in the RAxML tree indicate bootstrap proportions from 1000 pseudoreplicates. Clade support was identical between the TM and pruned AHE analyses, with the exception of the Lampropholis + Saproscincus clade that was supported by a bootstrap proportion of 100 and 96, respectively. We used PHYML maximum likelihood analyses to infer trees from 1000 bootstrap pseudoreplicates per locus. We used 1000 bootstrapped trees per locus (1,650,000 total trees) as input trees for the STEAC analysis, and numbers above or below the nodes indicate the proportion of times STEAC inferred that clade. Clade support was identical between the TM and pruned AHE analyses. Outgroup not shown.

Figure 4

Correlations between different metrics for divergence, shown with one-tailed p-values. a) Correlation between nucleotide diversity at anchored enrichment (AHE) loci and transcriptome (TM) loci. b) Correlation between tau estimated by AHE loci using 3s and nucleotide diversity at AHE loci. c) Correlation between tau estimated by AHE loci using 3s and tau as previously estimated from population genomic data [31].