A next generation approach to species delimitation reveals the role of hybridization in a cryptic species complex of corals

Abstract

Background: Our ability to investigate processes shaping the evolutionary diversification of corals (Cnidaria: Anthozoa) is limited by a lack of understanding of species boundaries. Discerning species of corals has been challenging due to a multitude of factors, including homoplasious and plastic morphological characters and the use of molecular markers that are either not informative or have not completely sorted. Hybridization can also blur species boundaries by leading to incongruence between morphology and genetics. We used traditional DNA barcoding and restriction-site associated DNA sequencing combined with coalescence-based and allele-frequency methods to elucidate species boundaries and simultaneously examine the potential role of hybridization in a speciose genus of octocoral, Sinularia.

Results: Species delimitations using two widely used DNA barcode markers, mtMutS and 28S rDNA, were incongruent with one another and with the morphospecies identifications. When mtMutS and 28S were concatenated, a 0.3% genetic distance threshold delimited the majority of morphospecies. In contrast, 12 of the 15 examined morphospecies formed well-supported monophyletic clades in both concatenated RAxML phylogenies and SNAPP species trees of > 6000 RADSeq loci. DAPC and Structure analyses also supported morphospecies assignments, but indicated the potential for two additional cryptic species. Three morphologically distinct species pairs could not, however, be distinguished genetically. ABBA-BABA tests demonstrated significant admixture between some of those species, suggesting that hybridization may confound species delimitation in Sinularia.

Conclusions: A genomic approach can help to guide species delimitation while simultaneously elucidating the processes generating coral diversity. Results support the hypothesis that hybridization is an important mechanism in the evolution of Anthozoa, including octocorals, and future research should examine the contribution of this mechanism in generating diversity across the coral tree of life.

Keywords: Anthozoa; Coral reefs; Hybridization; Octocorallia; Phylogenetics; RADSeq; Taxonomy.

Fig. 1

Maximum likelihood cladograms for Dongsha Atoll Sinularia a clade 4 and b clade 5C. Each colored cell denotes an individual’s species assignment based on RAD data (R), a morphospecies assignment (Sp), or a molecular operational taxonomic unit (MOTU) based on mtMutS and 28S rDNA (M + 28). Colors match RAD clades in Figs. 3 and 4. * = holotype or paratype. Morphospecies names are also included. See also Additional file 4 for clade and MOTU assignments

Fig. 2

Maximum likelihood phylogeny of Dongsha Atoll Sinularia clade 4 constructed using RAxML rapid bootstrapping (200 b.s. replicates) on the concatenated c 0.85, m 0.75, p 0.25 locus dataset. * denotes 100% b.s. support. Distruct plots are included and show the probability of individual membership into different K clusters (K = 4 for clade 4 and K = 2 for the S. tumulosa group). Colors denote different species. S. tumulosa is designated as ‘a’ and ‘b’ as results strongly suggest this clade consists of two species

Fig. 3

Maximum likelihood phylogeny of Dongsha Atoll Sinularia clade 5C constructed using RAxML rapid bootstrapping (200 b.s. replicates) on the concatenated c 0.85, m 0.75, p 0.25 locus dataset. * denotes 100% b.s. support. Distruct plots are included and show the probability of membership into different K clusters (K = 8 for clade 5C, and K = 2 for the S. slieringsi/penghuensis group). Colors denote different species. S. slieringsi/penghuensis is designated as ‘a’ and ‘b’ as results suggest this clade may consist of two species

Fig. 4

Species trees of Dongsha Atoll Sinularia clades 4 and 5C. Cloudograms illustrate the best species delimitation models (DAPC+ 1) for both clades inferred from bi-allelic SNP data [a clade 4: 6,236 SNPs and b clade 5C: 8,022 SNPs); m0.75 datasets] using SNAPP species tree analyses. The maximum clade credibility tree and congruent trees are in blue. Trees with different topologies are in red and green. Posterior probabilities at internal nodes > 95% unless indicated

Fig. 5

Discriminant analysis of principal components (DAPC) plots for Dongsha Atoll Sinularia a clade 4 and b clade 5C. Genetic clusters representing different morphospecies are color coded to match the phylogenetic trees in Figs. 2, 3 and 4. Species (S. tumulosa in (a) and S. slieringsi/penghuensis in (b) encircled in dotted lines) that were suggested to be further divided into two species by Bayes Factor Determination are also denoted

Fig. 6

D-statistic tests for admixture in Dongsha Atoll Sinularia clade 4. Test numbers are listed on the right for each 4-taxon test (((p1, p2), p3), p4). Horizontal bars below the tips of the tree indicate which taxa were included in each test. S. humilis was set as the outgroup for all tests (indicated by gray bars). Tests are configured to ask whether P3 (black bars) shares more derived SNPs with lineage P1 (green bars) relative to P2 (orange bars). As illustrated to the left, Z scores are bar plots and D-statistics are histograms. Histograms are green for significant gene flow between P1 and P3 (BABA) and orange for significant gene flow between P2 and P3 (ABBA). D-statistics that were not significant are gray. Significance was assessed at an alpha level of 3.0 (i.e., when D deviates > 3.0 standard deviations from zero)

Fig. 7

D-statistic tests for admixture in Dongsha Atoll Sinularia clade 5C. Test numbers are listed on the right for each 4-taxon test (((p1, p2), p3), p4). Horizontal bars below the tips of the tree indicate which taxa were included in each test. S. humilis was set as the outgroup for all tests (indicated by gray bars). Tests are configured to ask whether P3 (black bars) shares more derived SNPs with lineage P1 (green bars) relative to P2 (orange bars). As illustrated to the left, Z scores are bar plots and D-statistics are histograms. Histograms are green for significant gene flow between P1 and P3 (BABA) and orange for significant gene flow between P2 and P3 (ABBA). D-statistics that were not significant are gray. Significance was assessed at an alpha level of 3.0 (i.e., when D deviates > 3.0 standard deviations from zero)