4 DISCUSSION

Studies of population connectivity with mitochondrial markers provide critical information on gene flow and genetic relationships between neighboring populations (García, Vergara, & Gutiérrez, 2008; Turner, McPhee, Campbell, & Winemiller, 2004). Many studies showed that mitochondrial markers are highly effective for revealing marine fish genetic diversity and population connectivity (T. Gao et al., 2019; Lavergne et al., 2014; Machado et al., 2020). In this study onE. cardinalis , mtDNA sequence analysis of specimens from Beibu Gulf revealed no significant genetic differentiation among sampling sites, with low ΦSTvalues indicating genetic homogeneity.
In contrast to freshwater species, marine fish are usually expected to show low genetic differentiation across their distribution. This is mainly attributed to genetic exchange being maintained by adult mobility throughout Beibu Gulf during reproduction, and through the passive dispersal of eggs and larvae due to the lack of noticeable physical barriers in “open” oceans (Grant & Bowen, 1998; Hellberg, 2009; Machado et al., 2020). The dominant widespread haplotype H1 was found in all 11 sampling sites, which also indicated high dispersal potential of planktonic egg, larval, or adult stages of E. cardinalis in Beibu Gulf.
Previous studies suggested that E. cardinalis breed once a year in Beibu Gulf (Z. Z. Chen & Qiu, 2003; Hou, Feng, Lu, & Zhu, 2008).Evynnis cardinalis gonads begin to develop in November and spawning occurs from December to February. The population is concentrated in the northern Beibu Gulf during spawning. In the early spring, spawned fish mainly occur in the northeast of the gulf, and juveniles concentrate in shallow nearshore of this area in the late spring. Then, juveniles gradually migrate southwest and widely disperse in deep waters of Beibu Gulf in summer or early autumn (K. Zhang et al., 2020).
In addition, the dispersal pattern of E. cardinalis was also impacted by circulation in Beibu Gulf. In spring, the density gradient and monsoon wind drive the ocean current from northeast to southwest in the gulf. The surface current velocity reaches 30 cm/s, and the current in the middle layer is approximately 5–10 cm/s (J. Gao, Wu, & Ya, 2017). The direction of the spring currents roughly coincides with the migration of E. cardinalis . Therefore, the seasonal migration and ocean current may be responsible for gene exchange among different locations, and therefore why E. cardinalis shows low levels of genetic differentiation in Beibu Gulf. If we refer to the biological description of a stock as given by Ihssen et al. (1981), “a stock is an intraspecific group of randomly mating individuals with temporal and spatial integrity,” then the lack of distinct spatial boundaries and genetic substructure (low ΦST values) revealed by genetic analyses indicated that E. cardinalis in Beibu Gulf belong to a single stock.
The presence of a single stock in Beibu Gulf indicates that geographical isolation might block gene exchange between the Beibu Gulf stock and the other two E. cardinalis stocks, the Taiwan Strait and South China Sea stocks. In Beibu Gulf, the circulation, Hainan Island, and Leizhou Peninsula could act as barriers that impede free dispersal of E. cardinalis into this gulf from other areas of the South China Sea (J. Gao et al., 2017). The findings from our study and similar investigations conducted elsewhere demonstrated that marine fish that inhabit coastal waters usually constitute a single panmictic stock. For example, Rodrigues et al. (2008) revealed that Cynoscion acoupafrom northern Brazil represents a single stock, even though it occupies at least 1260 km of coastline. T. Gao et al. (2019) reported a high level of genetic homogeneity in the Pholis fangi population around Bohai Sea and Yellow Sea, and suggested it should be considered as a single panmictic stock. Hoolihan, Anandh, and van Herwerden (2006) also reported a homogeneous distribution of Spanish mackerel (Scomberomorus commerson ) throughout the Arabian Gulf, Gulf of Oman, and Arabian Sea on the basis of mtDNA analyses.
In addition, mtDNA sequence regions are particularly appropriate to infer historical processes that might be responsible for the contemporary geographic distribution of marine species because they are more prone to genetic drift than nuclear markers and have a smaller effective population size (Avise, 1994; Slatkin & Hudson, 1991). In our study, the haplotype network of E. cardinalis from Beibu Gulf exhibited a star-like and unstructured pattern with a predominance of scattered. The dominant haplotype (carried by 45% of the specimens) was in the central position of the haplotype network and surrounded by many haplotypes that diverged from the dominant haplotype by only few mutations. Most surrounding haplotypes were unique to each sampling site and showed few differences between them (Fig. 2). Similar star-like haplotype networks have been observed for other species in different coastal areas: Terapon jarbua , which consists of a panmictic stock from the Socotra Archipelago to the Hadhramout Coast along the wider Gulf of Aden (Lavergne et al., 2014); and Pogonias courbina , which did not display distinct structure along the coast of the southwestern Atlantic Ocean (Machado et al., 2020).
A star-like haplotype network pattern, high haplotype diversity, and low nucleotide diversity are often considered consequences of recent population expansion linked to the Pleistocene environmental changes (Craig, Eble, Bowen, & Robertson, 2007; Liu et al., 2011; Pereira, Márquez, Marin, & Marin, 2009). The recent demographic expansion ofE. cardinalis from Beibu Gulf is also supported by the unimodal mismatch distribution and significantly negative Tajima’s D and Fu’s Fs values. The population expansion of E. cardinalis in Beibu Gulf, which was directly estimated from the mismatch distribution, started 62–21 ka before present, which was during the late Pleistocene. BSP analysis indicated steady population expansion that started around 30 ka. Both two methods of estimated period of population expansion are consistent with the environment changes of the northern South China Sea in the Pleistocene.
Evynnis cardinalis is primarily distributed from 30–60 m depth, and spawns in coastal habitats and shallow shorelines. Therefore, theE. cardinalis distribution is closely correlated with fluctuating sea levels. When sea levels fell 120 m below the present level during the last glacial maximum of the Pleistocene, the northern South China Sea included Beibu Gulf, which was part of the South China continent, Hainan Island and Taiwan Island were connected to mainland China, and the entire South China Sea was separated from the Indian Ocean to form a semi-closed basin (Voris, 2000; P. Wang & Sun, 1994). Similar to other terrestrial species, E. cardinalis may have moved and survived in a potential glacial refuge during this period, such as the semi-closed South China Sea (Hewitt, 1999). In the late Pleistocene, the sea level was still 30 m below the present level, but the glaciation began to disappear and the sea water gradually poured into Beibu Gulf via the Qiongzhou Strait (Lu, Huang, Li, & Zhang, 2003). An initial population of E. cardinalis may have immigrated to Beibu Gulf from neighboring areas after it was filled with sea water and sufficiently deep. This initial panmictic stock quickly colonized the empty novel environment under the founder and priority effects, and experienced rapid population expansion when favorable conditions occurred (Boileau, Hebert, & Schwartz, 1992; Shulman et al., 1983).