No host-associated genetic divergence was observed within C. galea: specimens failed to cluster by host species in the Bayesian analysis and genetic distances in snails were not significantly larger between host species than the distances within host species. Haplotypes were not correlated with host species. Johnston et al. (2012) also did not observe host-specific clustering in genetic data and only found a small genetic divergence between C. galea snails associated with Acropora palmata, Orbicella spp. and Mycetophyllia spp. Two ecotypes of the Mediterranean congener C. meyendorffii, which are associated with different host species, are also not genetically divergent (Oliverio and Mariottini, 2001b).
In contrast to C. galea, a host-associated genetic divergence was found within C. caribaea. Snails associated with alcyonaceans were genetically distinct from snails associated with scleractinians with one exception either way. The genetic divergence within C. caribaea was however relatively small. On the phylogenetic tree, the support value for the branch with nearly all snails associated with scleractinians was low. Haplotypes differed on a few loci between snails associated with scleractinians and alcyonaceans (two mutations between the haplotype cluster mostly associated with alcyonaceans and the other haplotypes for 12S, eight mutations for COI), and correlation of haplotypes with host order was not perfect. The K2P distance between specimens associated with alcyonaceans and scleractinians fell below the threshold between interspecific and intraspecific divergence as determined in the ABGD analysis. These observations all indicate that the divergence within C. caribaea is of relatively recent origin. Genetically diverged host races have mostly been described in insects (e.g. Feder et al., 1988; McPheron et al.; 1988; Powell et al., 2014), but also in sponge-associated shrimp species (Duffy, 1996). Formation of host races may be the first stage of sympatric speciation (Maynard Smith, 1966; Berlocher, 1998). Isolation among host races is required for full speciation to occur, as the host races will revert to a panmictic population in absence of reproductive isolation (Jaenike, 1981). While the genetic divergence observed in the present study suggest some reproductive isolation, more analyses are needed to assess the degree of reproductive isolation between the two putative host races found within C. caribaea (Jaenike, 1981).
Mathematical models suggested that host-associated selective forces are critical for snails to specialize (Kawecki, 1996, 1997). The absence of host-associated genetic divergence within C. galea is therefore consistent with the results from earlier prey-preference experiments, where only a weak preference for the native host species was found (Hayes, 1990b). This is especially true for species whose pelagic larval stage is long, as a long larval stage favours generalists and promotes plasticity instead of divergence and speciation (Sotka, 2005). The larval ecology of snails is therefore relevant too. Larval development varies among coralliophillid species, with planktotrophic development being considered the plesiomorphic state within the Coralliophillinae, while some evidence indicates a prolonged intracapsular or lecithotrophic development in several species (Richter and Luque, 2002; Oliverio, 2008). Veliger shells of both C. galea and C. caribaea have been illustrated by Abbott (1958) and Wells and Lalli (1977) (only C. galea). Both species are thought to have planktotrophic development, but the duration of the planktonic stage remains unknown, and is expected to be > 30 days for C. galea (Wells and Lalli, 1977; Richter and Luque, 2002; Johnston et al., 2012). Studies on the protoconch of snails may provide more insight in the larval ecology of Caribbean Coralliophila spp. (Oliverio, 2008).
For genetic divergence to be established, selection pressures must be strong enough to overcome homogenizing processes that increase gene flow between populations (Schluter, 2009; Johnston et al., 2012). The difference in host-associated genetic divergence between C. galea and C. caribaea suggests that selection pressures to specialize to either scleractinian or alcyonacean host species might be higher than the selection pressures to specialize to specific (groups of) species within these orders. Anti-predatory mechanisms of gorgonians might play a role here. Some generalist predators, such as the facultatively corallivorous polychaete Hermodice carunculata (Pallas, 1766), do feed on gorgonians (Marsden 1962; Preston and Preston 1975; Lasker 1985; Rotjan and Lewis 2008; Wolf et al., 2014), but most other species feeding on gorgonians, such as ovulid gastropods of the genus Cyphoma Röding, 1798, are specialized to this diet (Birkeland and Gregory 1975; Harvell and Suchanek 1987; Lasker and Coffroth, 1988; Lasker et al., 1988; Van Alstyne and Paul 1992; Burkepile and Hay, 2007; Chiappone et al., 2003; Reijnen et al., 2010; Schärer et al., 2010; Pinto et al., 2017; Reijnen and Van der Meij, 2017). Selection to overcome these mechanisms within C. caribaea might therefore have been strong enough to induce genetic divergence.
While specialization to (a group of) host species might have played a role in the intraspecific divergence observed within C. caribaea, the absence of a monophyletic Caribbean Coralliophila clade on the phylogenetic tree suggests that host-associated divergence was not the mechanism behind the divergence among species and that their common ancestor originated outside of the modern Caribbean. A similar pattern has been observed in the shrimp family Palaemonidae (Horká et al., 2016). Therefore, more extensive phylogenetic and phylogeographic analyses of Coralliophila spp. are needed to unravel the biogeographic patterns of this genus in the Caribbean.
The phylogenetic analyses in the present study suggest that the genus Coralliophila is polyphyletic, as C. caribaea seems to be closer related to the Indo-Pacific genus Leptoconchus than its own congeners (Fig. 15; Oliverio and Mariottini, 2001a; Oliverio et al., 2002, 2009). The genus Coralliophila is therefore in need of a taxonomic revision (Oliverio and Gofas, 2006; Oliverio, 2008; Oliverio et al., 2009).
Some overlap was observed between the distributions of intraspecific and interspecific divergence in the ABGD analysis, which is indicative of the absence of a universal threshold within the Coralliophilinae (Collins and Cruickshank, 2013). Only one other threshold value based on an ABGD analysis for muricid gastropods was found in literature: Barco et al. (2013) reported a threshold K2P distance between 0.020 and 0.025 based on a dataset of COI sequences of the genus Ocinebrina Jousseaume, 1880, which is lower than the value found in the present study. In addition, with the threshold value used in the present study, many of the Leptoconchus species included in the ABGD analysis could not be distinguished as separate species. This reinforces the lack of a universal threshold between intraspecific and interspecific divergence.