Use of RAD sequencing for delimiting species

Abstract

RAD-tag sequencing is a promising method for conducting genome-wide evolutionary studies. However, to date, only a handful of studies empirically tested its applicability above the species level. In this communication, we use RAD tags to contribute to the delimitation of species within a diverse genus of deep-sea octocorals, Chrysogorgia, for which few classical genetic markers have proved informative. Previous studies have hypothesized that single mitochondrial haplotypes can be used to delimit Chrysogorgia species. On the basis of two lanes of Illumina sequencing, we inferred phylogenetic relationships among 12 putative species that were delimited using mitochondrial data, comparing two RAD analysis pipelines (Stacks and PyRAD). The number of homologous RAD loci decreased dramatically with increasing divergence, as >70% of loci are lost when comparing specimens separated by two mutations on the 700-nt long mitochondrial phylogeny. Species delimitation hypotheses based on the mitochondrial mtMutS gene are largely supported, as six out of nine putative species represented by more than one colony were recovered as discrete, well-supported clades. Significant genetic structure (correlating with geography) was detected within one putative species, suggesting that individuals characterized by the same mtMutS haplotype may belong to distinct species. Conversely, three mtMutS haplotypes formed one well-supported clade within which no population structure was detected, also suggesting that intraspecific variation exists at mtMutS in Chrysogorgia. Despite an impressive decrease in the number of homologous loci across clades, RAD data helped us to fine-tune our interpretations of classical mitochondrial markers used in octocoral species delimitation, and discover previously undetected diversity.

Figure 1

Comparison of locus detection for Stacks (a–f) and PyRAD (g–l). The number of loci, SNPs and indels detected for specimens separated by 0–16 mutations at the mitochondrial mtMutS gene are shown for the different read coverage (m parameter) and divergence levels (M and n parameters, see text). In PyRAD analyses, ‘s' corresponds to the ‘Wclust' parameter. A full color version of this figure is available at the Heredity journal online.

Figure 2

Information content of the locus catalog built by PyRAD for all 91 Chrysogorgia specimens. Wclust: percent divergence permitted between loci within and across specimens; in addition to the 93% Wclust level used to infer the Chrysogorgia phylogeny, the 89% Wclust level was tested here. A full color version of this figure is available at the Heredity journal online.

Figure 3

Maximum likelihood phylogenetic trees inferred using RAxML for the mitochondrial mtMutS data (a) and RAD loci (b–d). Bootstrap node support (1000 replicates for a, 500 replicates for b, c and 200 for d) is presented only for nodes with ⩾70% support. At the tips, colored dots, which represent mtMutS haplotype membership (each color represents a unique haplotype), are followed by specimen identifiers and haplotype numbers. Each tree was rooted to the Atlantic specimen (JAC1018, haplotype J). Genetic structure within clades 1, 2 and 3 were further investigated using a DAPC and TESS (see text and Supplementary Figure S1). Scale bars: substitution/site. A full color version of this figure is available at the Hypertension Research journal online.