Sea-level-change-driven bottleneck enhanced population subdivision in the southern South China Sea
The genetic discontinuity between the “s-SCS” and “n-SCS” actually is not unique in A. corniculatum. The studies of mangrove speciesE. agallocha (W. Guo et al., 2018) and H. littoralis(Banerjee et al., 2020) have also shown similar genetic discontinuities. The previous studies suggested that these genetic discontinuities are attributed to the lack of suitable ocean currents to disperse fruits during the ripening season in the South China Sea (Banerjee et al., 2020; W. Guo et al., 2018). We argue that at least two observations inA. corniculatum are not compatible with the previous explanation. First, the differentiation between “n-SCS” and “Gulf of Bengal” is much lower than between “s-SCS” and “Gulf of Bengal”. How could “n-SCS” populations exchange genes with “Gulf of Bengal” populations without bridging by “s-SCS” populations? Second, the differentiation between “s-SCS” and “n-SCS” is comparable to between “s-SCS” and “Gulf of Bengal”, much higher than that between “s-SCS” and “Australasia”. It’s not feasible to assume that the influence of ocean currents in the SCS is even stronger than that of the Indonesian-through flow, to a level comparable with the land barrier of the Malay Peninsula. Hence, the break between the “s-SCS” and “n-SCS” populations of A. corniculatum cannot be simply attributed to the influence of ocean currents.
We hypothesized that bottleneck events could promote population subdivision by augmenting population differentiation. The “s-SCS” cluster was found to had undergone a bottleneck event. Our ABC simulations provided strong support that the “s-SCS” population originated from the “Australasia” cluster, with the spit occurring at ~1.5 Mya. The close relationship between “s-SCS” and “Australasia” was also evidenced directly by STRUCTURE and PCA clustering, and low FST values. During the split, the “s-SCS” population was reduced to a population size of ~824, from 10538 of the common ancestral population. The drastic reduction of genetic diversity in the populations Chai-ya and Kuching evidenced the bottleneck event. The average π values in the populations Chai-ya and Kuching are only 14.3~24.4% of those in the populations of “Australasia” subgroup and the average θ values are only 11.2~25.6% (Table 2). In other words, at least 75~90% of the ancestral polymorphisms were lost during the bottleneck event, which is consistent with the ~92% reduction in effective population size estimated by ABC modeling.
Although the time estimation using ABC computation may not be very accurate, it roughly dated the origination of the “s-SCS” cluster at the middle Pleistocene, when glacial periods alternated with interglacial periods repeatedly. During the glacial periods (Miller et al., 2005; Voris, 2000), most Southeast Asia populations should have been wiped out. The refugial populations might have contracted to the margins of the Sundaland with a range from Wallacea to North Australia. As sea level rose in the interglacial periods, a subset of the refugia population expanded to the current southern South China Sea range as the coastline advanced (Cannon, Morley, & Bush, 2009). Such a process could have repeated multiple times. However, the genetic data obtained today are powerless to distinguish them because the ancient genetic patterns had been reshaped by more recent events.
During the bottleneck processes, intensified genetic drift and relaxed purifying selection due to reduced population size should have contributed to the loss of ancestral polymorphisms and fixation of new mutations. The observation of haplotypes unique to the “s-SCS” cluster in many genes is consistent with this interpretation. This mechanism consequently generated the current deep genetic differentiation between “s-SCS” and “n-SCS”. Despite the bottleneck effect, the gene flow between “s-SCS” and “n-SCS” seems not completely blocked, evidenced by the occurrence of “n-SCS” haplotypes with relatively high frequency in “s-SCS” populations. Our ABC modeling also supported the existence of such gene flow. In contrast, gene flow in the reciprocal direction is much lower, indicated by that “s-SCS” haplotypes were rarely observed in “n-SCS” populations.
As mentioned before, the substantial genetic break between “s-SCS” and “n-SCS” has also been observed in E. agallocha (W. Guo et al., 2018) and H. littoralis (Banerjee et al., 2020), which may also be attributed to the mechanism we described above. However, the mismatch distribution analyses presented in the original papers provided no support for a sudden expansion model (Banerjee et al., 2020; W. Guo et al., 2018). Notably, obvious differentiation has been observed between the populations sampled from Northeastern Borneo and surrounding populations in S. alba (Y. Yang et al., 2017) and A. marina (Wang et al., 2021). Such population structures in the smaller geographical ranges are highly likely generated by a bottleneck process. Further studies may test this hypothesis in these species. In contrast, the populations of some mangrove species, whose distribution range covers both the northern and southern parts of the South China Sea, are found to be genetically continuous with confidential data. Such species include but may not limit to R. apiculata (Z. Guo et al., 2016),R. stylosa (Yan et al., 2016), C. decandra (Huang et al., 2012), B. gymnorhiza (Urashi et al., 2013) and L. racemose(J. Li et al., 2016). Comparing to those genetic structures attributed to geographical barriers, the population structure generated by bottleneck events appears to be rarer and more unpredictable.