This study genetically confirmed hybridization between the butterflyfishes C. trifasciatus and C. lunulatus at Christmas Island, determined the evolutionary consequences of this process and identified ecological conditions that favor hybridization. Direct field observations suggest that low abundances of both species and nonassortative mating may have promoted interbreeding. Furthermore, genetic analyses of mtDNA and microsatellite loci have confirmed that distinct and intermediate color morphs are hybrids, and shown that there is a unidirectional maternal contribution to hybridization and no introgression between the parental species. These findings suggest that the production of viable hybrids between C. trifasciatus and C. lunulatus is relatively rare. Apparently hybridization may facilitate persistence of the Pacific Ocean species at Christmas Island, but this may be at the expense of the Indian Ocean species at this isolated location.
Ecology of hybridization
The recently diverged, sister species C. trifasciatus and C. lunulatus (Bellwood et al. 2010) have largely allopatric distributions (Allen et al. 1998) and occur in sympatry at Christmas Island, where they form heterospecific pairs. In butterflyfishes, cases of hybridization between allopatric sister species that come into secondary contact, such as this one, represent the minority—13 of 35 (37%) (Hobbs et al. In press). Sympatry of Indian and Western Pacific Ocean species at Christmas Island has been considered a potential indication of westward dispersal of Pacific Ocean taxa (Allen and Steene 1979; Blum 1989), has set the scene for previously reported cases of reef fish hybridization (Marie et al. 2007; Hobbs et al. 2009) and is a precursor for hybridization between C. trifasciatus and C. lunulatus.
At Christmas Island, C. trifasciatus and C. lunulatus have very similar ecologies (dietary composition and habitat use), which increases the chance of heterospecific encounters. Ecological similarities and habitat overlap among recently diverged species are not uncommon, and act to increase (or at least do not limit) social interactions between hybridizing reef fishes (Randall 1956; Fischer 1980; Frisch and van Herwerden 2006; Yaakub et al. 2006, 2007; Marie et al. 2007). Both C. trifasciatus and C. lunulatus are known obligate corallivores (Harmelin–Vivien 1989; Pratchett et al. 2004; Berumen et al. 2005; Pratchett 2005; Cole et al. 2008) and accordingly, at Christmas Island, both species feed almost exclusively on scleractinian corals, mostly Porites, Galaxea, and Montipora. The dietary composition and feeding rates were not significantly different between these two species, indicating a high degree of dietary overlap (see also Randall 1956; Feddern 1968; Fischer 1980). If there were inconsistencies in the dietary composition, combined with broadly nonoverlapping distributions of respective prey, this might prevent pairing and subsequent reproduction between these species.
Chaetodon trifasciatus and C. lunulatus form heterospecific pairs at Christmas Island indicating that assortative mating has broken down between these two sister species, probably as a consequence of their local rarity. Pairing in this species complex most likely occurs for reproduction, because C. lunulatus has been deemed a monogamous breeder (Yabuta 1997) and pair formation corresponds with the onset of sexual maturity (Pratchett et al. 2006a).
The nominal hybrids in this complex—initially identified through aberrant markings, intermediate to those of their parents (Hobbs et al. 2009)—had spatial and dietary ecologies comparable to those of their putative parents. Moreover, hybrids were observed in breeding pairs with both parent species, potentially facilitating gene transfer between these species—if hybrids are fertile. Importantly, hybrids in this complex were relatively rare and the statistical power associated with their data is therefore limited. Nevertheless, rarity of hybrids is, per se, interesting and might simply reflect the relative rarity of their parents, or be indicative of some selective pressure on hybrids.
Genetics of hybridization
The use of both mtDNA and microsatellite loci allowed confirmation of the hybrid status of intermediately colored individuals, and shed light on the potential evolutionary consequences of hybridization in this complex. The mitochondrial cyt b identified two distinct parental clades, separated by 5% (29 fixed substitutions out of 522 nucleotides). The genetic distance between the species clades is consistent with divergence of these species, some 2.5–4 Ma (Bellwood et al. 2010), given the cyt b mutation rate of 1–2.5% Ma−1 (McMillan and Palumbi 1997) and assuming a molecular clock. Hybridization is most common between species that diverge by less than 10% because the genetic similarity increases the chance of viable hybrid formation (Mallet 2005). Each parental clade contained all individuals of the respective species, from all geographical locations. Hybrids shared cyt b haplotypes with the Pacific Ocean C. lunulatus only, indicating a unidirectional contribution to hybridization, in which C. lunulatus appears to always be the mother.
Despite informing about the matrilineal contribution, the results from mtDNA analyses were not sufficient to confirm the hybrid status of the intermediately colored individuals, which could simply be an aberrant color pattern of C. lunulatus. However, nuclear microsatellite markers ruled out this scenario and showed that hybrid genotypes were distinct from (and intermediate to) those of C. trifasciatus and C. lunulatus. This confirms the hybrid status and suggests that intermediately colored individuals were most likely F1 hybrids. The unidirectional maternal contribution contrasts with findings of other reef fish hybridization studies (McMillan et al. 1999; van Herwerden et al. 2006; Yaakub et al. 2006; Marie et al. 2007), which found that maternal contribution was either bidirectional or biased toward the more abundant species. This may suggest that hybrids resulting from the opposite cross (C. trifasciatus females) are subject to some negative selection or that female C. lunulatus are actively choosing (i.e., female choice, Wirtz 1999) to mate with C. trifasciatus males, because of lack of conspecifics. This latter explanation seems more likely because, even though abundances of the parental species in this complex were not statistically different, C. lunulatus was observed less frequently than C. trifasciatus, and in previous censuses of the ichthyofauna of Christmas Island it was not recorded (Allen et al. 2007; but see Hobbs et al. 2009; Hobbs et al. 2010).
The apparent lack of introgression between C. trifasciatus and C. lunulatus could be explained by the extreme rarity of the hybrids or their infertility. F1 hybrids of this complex sampled at Christmas Island previously to this study were sexually mature (J-P. Hobbs, unpublished data), henceforth the rarity of these individuals seems to be the most likely explanation. Three conditions are required for introgression to take place: (1) a fertile hybrid with the mtDNA of one species must (2) pair with an individual of the other species and (3) produce viable offspring (see Fig. 8 in Yaakub et al. 2006). These three conditions have to be sequentially satisfied frequently enough for the genetic signal of introgression to perpetuate and be detected. The hybrids in this complex may not be abundant enough to meet the conditions for introgression. In addition, C. lunulatus appears to be a recent colonist to Christmas Island (Hobbs et al. 2010) and there may not have been sufficient time for introgression to be evident in the genetic composition of these species. Lack of introgression contrasts with the results of most reef fish hybridization studies (van Herwerden and Doherty 2006—northern hybrid zone; van Herwerden et al. 2006; Yaakub et al. 2006; Marie et al. 2007; but see van Herwerden and Doherty 2006—southern hybrid zone; Yaakub et al. 2007) and, more importantly, with the findings of other studies of butterflyfish hybridization (McMillan et al. 1999).
The different divergence ages of parental species might account for the dissimilarities found in the outcome of hybridization in reef fishes. At the Solomon Islands, C. punctatofasciatus hybridizes with C. pelewensis (McMillan et al. 1999) from which it differs by only 0.7% of the mitochondrial cyt b (McMillan and Palumbi 1995). Bidirectional introgression between these species is so extensive that, within the hybrid zone, hybrid phenotypes account for over 70% of the individuals (McMillan et al. 1999). Moreover, despite phenotypic differentiation of the two species evident outside the hybrid zone, their genetic homogeneity extends for thousands of kilometers beyond the hybrid zone (McMillan et al. 1999) as a result of extensive bidirectional introgression. Conversely, taxa in the C. trifasciatus complex show a clear mtDNA break and hybridization between species is unidirectional and lacks introgression, with apparently more localized, limited evolutionary consequences. From the data available, it appears that increased genetic divergence between hybridizing butterflyfishes signifies a higher fitness cost to hybridization and limits introgression.
Further, introgression and bidirectional maternal contribution were found in hybridizing surgeonfishes Acanthurus leucosternon and A. nigricans, for which 1% genetic break was reported in the mtDNA COI marker used (Marie et al. 2007). In hybridizing groupers, Plectropomus leopardus and P. maculatus, separated by 1% genetic break based on two nuclear and two mtDNA markers (Craig and Hastings 2007), hybridization was highly introgressive, but the maternal contribution was unidirectional (van Herwerden et al. 2006). Based on the available data, it seems that for reef fishes a genetic break smaller than 2%—at a range of mtDNA loci other than HVR1—results in introgressive hybridization, which can have evolutionary consequences extending outside the hybrid zone and may result in increased genetic diversity and adaptability (e.g., McMillan et al. 1999). Conversely, a genetic break larger than 5%, albeit still lower than the proposed 10% cutoff for successful hybrid formation in the terrestrial environment (Mallet 2005), seems to result in a higher cost to hybridization and a reduction in the number of viable, fertile hybrids.
In addition to evolutionary relatedness, ecological factors may influence the occurrence of hybridization. Sustained pressure on coral reefs worldwide, which decreases available habitat and negatively affects the abundance of some reef fishes (e.g., Pratchett et al. 2006b; Graham 2007), might result in increased habitat overlap and local rarity of species—the ecological conditions most frequently ascribed a role conducive to hybridization in reef fishes. Further, ocean acidification negatively impacts mate recognition in reef fishes (Munday et al. 2009), increasing the chances of heterospecific breeding. While increased cases of hybridization between recently diverged reef fishes may prove beneficial to the adaptability of the species involved, a greater number of cases of hybridization among more distantly related species might have a detrimental effect and it may also result in reverse speciation (two species become one).
Evolutionary consequences of hybridization
The genetic diversity of C. trifasciatus is relatively low within the hybrid zone. Low genetic diversity can have a detrimental effect on the adaptability of species to novel environments (Rhymer and Simberloff 1996), and in this case might result in the local extinction of C. trifasciatus. For this species, hybridization might represent a significant cost. Indicative of this is the apparent lack of maternal contribution to hybridization from C. trifasciatus. Female C. trifasciatus form pairs with C. lunulatus males (J-P. Hobbs, unpublished data) and it seems therefore possible that the resulting hybrids may not be viable, and heterospecific mating may represent a significant reproductive cost. Moreover, the hybrids of this complex that do survive (those with C. lunulatus mothers) have the lowest genetic diversity and, compared to their parents, may be less well adapted to the environmental conditions of Christmas Island (as hinted at by their more limited resource use and significantly lower abundance). These findings are in sharp contrast with those of the only previous genetic studies of hybridization in butterflyfishes (McMillan et al. 1999). McMillan et al. (1999) found that, within the hybrid zone at the Solomon Islands and Papua New Guinea, hybrid phenotypes largely outnumbered the pure C. pelewensis and C. punctatofasciatus suggesting better survival (and hence fitness) of hybrids than parents within the hybrid zone environment.
Hybridization between C. trifasciatus and C. lunulatus at Christmas Island could be rare and relatively recent (and this could explain the lack of introgression), but, protracted through time, this may facilitate the local persistence of at least one of these rare species (C. lunulatus). Moreover, this study detected high levels of intrabasin connectivity in the Indian Ocean, with populations of C. trifasciatus from locations as far apart as Zanzibar and Christmas Island showing no significant genetic differentiation. According to previous studies of reef fish hybridization (Yaakub et al. 2006; Marie et al. 2007; Hobbs et al. 2009), hybrid individuals could stray further west to Cocos (Keeling) Islands. However, C. trifasciatus×C. lunulatus hybrids seem to be restricted to the hybrid zone, with limited, more localized consequences, a scenario also found in Caribbean wrasses of genus Halichoeres (Yaakub et al. 2007).