Genetic divergences and hybridization within the Sebastes inermis complex

Abstract

The Sebastes inermis complex includes three sympatric species (Sebastes cheni, viz Sebastes inermis, and Sebastes ventricosus) with clear ecomorphological differences, albeit incomplete reproductive isolation. The presence of putative morphological hybrids (PMH) with plausibly higher fitness than the parent species indicates the need to confirm whether hybridization occurs within the complex. In this sense, we assessed the dynamics of genetic divergence and hybridization within the species complex using a panel of 10 microsatellite loci, and sequences of the mitochondrial control region (D-loop) and the intron-free rhodopsin (RH1) gene. The analyses revealed the presence of three distinct genetic clusters, large genetic distances using D-loop sequences, and distinctive mutations within the RH1 gene. These results are consistent with the descriptions of the three species. Two microsatellite loci had signatures of divergent selection, indicating that they are linked to genomic regions that are crucial for speciation. Furthermore, nonsynonymous mutations within the RH1 gene detected in S. cheni and "Kumano" (a PMH) suggest dissimilar adaptations related to visual perception in dim-light environments. The presence of individuals with admixed ancestry between two species confirmed hybridization. The presence of nonsynonymous mutations within the RH1 gene and the admixed ancestry of the "Kumano" morphotype highlight the potential role of hybridization in generating novelties within the species complex. We discuss possible outcomes of hybridization within the species complex, considering hybrid fitness and assortative mating. Overall, our findings indicate that the genetic divergence of each species is maintained in the presence of hybridization, as expected in a scenario of speciation-with-gene-flow.

Keywords: Clustering analysis; Divergent selection; Genetic divergence; Hybridization; Microsatellite loci; Rhodopsin gene; Sebastes; Species complex.

Figure 1. Sampling sites along coastal waters of Japan.

Black points represent sampling sites.The reference map was retrieved from the Natural Earth website (https://www.naturalearthdata.com/). Data contained in this website is in the public domain.

Figure 2. Colouration patterns, meristic counts, and otolith weight~age relationships of the three rockfishes Sebastes cheni, Sebastes inermis, and Sebastes ventricosus, and the putative morphological hybrids between them.

The frequency distributions of the number of pored lateral line scales (SLL), number of gill rakers of the first arch (GR), and number of radials of the pectoral fin (PFR) are indicated in each species and putative morphological hybrid. Reference sizes for frequencies are indicated below the three variables. Points in the otolith weight~age plot were coloured to ease distinction of species and putative morphological hybrids. Arrows connecting specimens indicate the putative origin of each morphological hybrid. A scale of 3 cm was added next to each specimen as reference for size.

Figure 3. Genetic clusters inferred in the Sebastes inermis complex using ten (above and middle) and eight microsatellite loci (below).

Individuals are coloured based on their ancestry coefficients (Q-score) for each genetic cluster. Putative morphological hybrids are indicated as K (“Kumano”), BW (S. cheni × S. ventricosus morphotype), and RW (S. cheni × S. inermis morphotype).

Figure 4. Discriminant analysis of principal components (DAPC).

(A) Scatterplot of individuals in the discriminant functions (LD1 and LD2). Individuals are represented by pie charts showing their posterior probabilities. Colours of the genetic clusters are indicated in the legend. (B) Scatterplots of individuals projected into the space of principal components (PCs). Individuals were coloured based on their morphological features. (C) Inferred number of genetic clusters by k-means clustering considering the “elbow” point in the distribution of the Bayesian information criterion (BIC) scores.

Figure 5. Distribution of maximum Q-scores calculated in STRUCTURE using the observed and simulated individuals.

Central bold lines in the box plots indicate the medians; box limits represent the 1st and 3rd quartiles; Q-scores are drawn as black circles. Different colours indicate whether boxplots are from morphologically distinct or putative pure individuals, putative morphological hybrids, putative F1 hybrids, or putative backcrosses as represented in the legend above. B: black rockfish (S. ventricosus), R: red rockfish (S. inermis), W: white rockfish (S. cheni), K: “Kumano” morphotype, BW: black-white hybrids, BR: black-red hybrids, and RW: red-white hybrids. Putative backcrosses from simulations are represented with three letters, the first two indicate the putative F1 hybrid parent and the third one the putative pure parent. Thus, for example BRB: backcrosses from black-red F1 hybrids and pure black individuals.

Figure 6. Haplotype networks constructed from partial sequences of the mitochondrial control region (A) and the intron-free rhodopsin gene (B).

Colours indicate individuals assigned to a single species considering morphological and genetic information. BB, RR, and WW designate individuals identified as S. ventricosus (black rockfish), S. inermis (red rockfish) and S. cheni (white rockfish), respectively. BH, RH, and WH indicate individuals morphologically identified as black, red, and white rockfish, respectively, but genetically classified as putative hybrids. BW and RW indicate specimens classified as putative morphological hybrids of black-white and red-white hybrids, respectively. K is designated to individuals collected in Wakayama Prefecture that display the “Kumano” morphotype.