Widespread introgression in deep-sea hydrothermal vent mussels
Studies of interspecific hybridization in animals have a long-standing history in terrestrial, coastal and shallow-water research, but virtually nothing is known about contact zones in deep-sea environments. We used 18 diagnostic SNP markers and the mitochondrial ND4 gene to analyse the hybrid zone between the deep-sea mussels B. azoricus and B. puteoserpentis [10, 11]. With this panel of diagnostic markers we were able to reliably distinguish hybridization from incomplete lineage sorting and to estimate the extent of admixture in individuals from eight localities of the MAR. In contrast to former studies we found that evidence for introgression was widespread and not restricted to the geographically intermediate BS locality. Moreover, the frequency of hybrids was much higher than described previously (30% in [11] compared to at least 60% in this study). Gene flow occurred across the sampled geographic range from MG to SM, but in a highly asymmetric fashion. The population at BS consisted mostly of F2–4 generation hybrids with a minority of B. puteoserpentis-like genotypes, whereas no B. azoricus-like individuals were found. By contrast, mussels at vent fields to the north of BS carried predominantly B. azoricus genotypes, while mussels in the south were predominantly B. puteoserpentis with declining proportions of backcross genotypes. No individuals that matched an F1 genotype were detected at any locality. Although different statistical approaches varied in their estimates of the number of hybrids at the investigated vent locations, even the most conservative program (STRUCTURE) detected admixed individuals outside the BS hybrid zone. As many backcross genotypes were often identified as pure species and as the use of diagnostic markers might underestimate the level of introgression [27], these observations indicate that gene flow is possibly more extensive than suggested by our analyses.
Different, mutually non-exclusive explanations could account for the asymmetry in admixture patterns and the differential abundance of genotype classes among vent locations. Bathymodiolus azoricus and B. puteoserpentis appear to segregate by depth along the MAR [10, 11]. Such differential depth associations could prevent the two species from meeting and forming hybrids in certain regions. Or, if dispersal into non-native ranges occurs and the two species differ in their bathymetric tolerances, individuals with a high genomic proportion of only one parent might experience a selective disadvantage in depths that are characteristic for the other ancestral lineage. Both scenarios would agree with the differential fixation of B. azoricus and B. puteoserpentis ND4 haplogroups in the northern and southern regions due to geographic premating barriers resulting in genetic drift or due to co-evolution of nuclear and cytoplasmic genomes [47, 48]. Alternatively, the apparent lack of B. azoricus genotypes at the intermediate BS locality could simply be a matter of sampling bias. Firstly, mussels are very difficult to distinguish from surrounding rock at this vent field as they are coated with sulphide-enriched sediments [10]. Secondly, hydrothermal vents are highly dynamic habitats, where physico-chemical conditions can vary over small spatial scales, thereby providing opportunities for niche segregation [49]. Consequently, it is possible that we failed to sample B. azoricus genotypes, if they are relatively infrequent or adapted to environmental patches at BS that we did not explore during our cruises. Likewise, this could be the case for F1 hybrids, which were surprisingly not observed in our sample. Given that we were able to detect B. azoricus at other localities and that the assignment certainties for F1 individuals were very high, it is improbable that our marker panel did not have enough power to detect these genotypes at BS. Therefore, it is justified to conclude that our results mirror a true pattern rather than a statistical artefact.
Another explanation for the observed patterns could be a southward movement of the contact zone from an unknown, possibly now extinct vent locality north of BS. Under this scenario, the early generation hybrids would be immigrants that reproduced locally. Both hybrid zone movements and potential sampling gaps would be consistent with the composition of thiotrophic symbionts in BS individuals [50]. Based on their analyses of internal transcribed spacer sequence variation, Won et al. [50] observed either single infections by the B. puteoserpentis-specific bacterial strain or double infections by symbionts from both species. By contrast, no single infections by the B. azoricus-specific type were found. Mixed infections would only be possible, when both pure species co-occur at BS (sampling gap hypothesis) or when the first contact happened at another locality (hybrid zone movement hypothesis). Both hypotheses seem to be plausible for several reasons. On the one hand, secondary contact zones between recognized species often emerge under environmental conditions to which parental genotypes are not well adapted [1, 51–53]. In such areas increased genetic diversity due to interspecific mating is likely to promote adaptability in hybrids and allow them to exploit these unoccupied niches [1]. As we did not sample environmental parameters from the investigated vent fields and were not targeting genes that might be involved in local adaptation, we cannot test this hypothesis with our current data set. However, the dynamic nature of hydrothermal vents, depths differences and the association with different chemosynthetic symbiont strains could favour ecological differentiation between B. azoricus and B. puteoserpentis and lead to hybrid advantage in environmentally intermediate vent areas. Addressing this aspect will be helpful to identify the relative importance of selective forces and neutral processes, which can cause the same introgression and admixture patterns as natural selection [54].
On the other hand, a recent study by Breusing et al. [28] indicates that hydrothermal vents or equivalent chemosynthetic habitats are much more abundant on the MAR than previously thought. In combining biophysical modelling approaches with molecular analyses the authors found that known mussel populations are connected over thousands of kilometres, although veliger larvae seldom reach suitable settlement sites that are more than 150 km away from their natal locality. These results are in agreement with current predictions from physical and geochemical measurements about the frequency of hydrothermal habitats on the MAR [55]. Thus, it is also possible that the centre of the hybrid zone originated at another, so far undetected vent locality close to BS.
Based on information from two loci (ND4 and Lap), Won et al. [11] reported that the parental genotypes greatly exceeded hybrid genotypes in abundance. They hypothesized that the scarcity of mussels at BS results from a deficit of local reproduction and episodic recruitment of immigrants from the parental regions—a pattern that supported the tension zone model. The present multi-locus data cannot confirm this hypothesis, as most specimens at BS were of mixed ancestry and fell into a variety of hybrid genotype classes. Furthermore, no deviations from random mating expectations could be found, thereby providing no evidence for heterozygote deficiencies. However, the finding of significant linkage disequilibria could imply that some nuclear incompatibilities exist in hybrids. Alternatively, linkage disequilibria might be due to a recent origin of hybridization at BS, as it has been suggested that vents at this locality were lately re-activated and re-colonized after experiencing a period of dormancy [56, 57].