Possible Natural Hybridization of Two Morphologically Distinct Species of Acropora (Cnidaria, Scleractinia) in the Pacific: Fertilization and Larval Survival Rates

Natural hybridization of corals in the Indo-Pacific has been considered rather rare. However, field studies have observed many corals with intermediate interspecific or unusual morphologies. Given that the existence of F1 hybrids with intermediate interspecific morphologies has been proven in the Caribbean, hybrids may also inhabit the Indo-Pacific and occur more frequently than expected. In this study, we focused on two morphologically different species, Acropora florida and A. intermedia, and performed crossing experiments at Akajima Island, Japan. Results showed that these species could hybridize in both directions via eggs and sperm, but that fertilization rates significantly differed according to which species provided eggs. These results are similar to those reported from the Caribbean. Although all embryos developed normally to the planular larval stage, the developmental processes of some hybrid embryos were delayed by approximately 1 h compared with conspecific embryos, suggesting that fertilization occurred 1 h later in interspecific crosses than in intraspecific crosses. More successful hybridization could occur under conditions with low numbers of conspecific colonies. Additionally, a comparison of survival rates between hybrid and intraspecific larvae revealed that intra- and interspecific larvae produced from eggs of A. florida survived for significantly longer than those produced from eggs of A. intermedia. Considering these data, under specific conditions, hybrids can be expected to be produced and survive in nature in the Pacific. Furthermore, we identified one colony with intermediate morphology between A. florida and A. intermedia in the field. This colony was fertilized only by eggs of A. florida, with high fertilization rates, suggesting that this colony would be a hybrid of these two species and might be backcrossed.


Introduction
Hybridization occurs in many animals and plants. Especially in plants, reticulate evolution, in which hybridization and speciation are repeated in evolutionary time, has been reported [1,2]. In marine animals, several kinds of hybridization have been reviewed [3,4]. Veron [5] first proposed the hypothesis of reticulate evolution in zooxanthellate scleractinian corals (hereinafter referred to as corals). Many species of coral spawn synchronously on the same night and time in early summer (i.e., mass spawning) [6,7,8]. In the Indo-Pacific, such mass spawning of the most abundant corals, genus Acropora, might cause hybridization because large quantities of their eggs and sperm mix in mass spawning events. In fact, hybridization of Acropora species has been proven in crossing experiments [9,10,11,12,13]. In addition, molecular phylogenetic analyses have revealed that hybridization occurs only among genetically closely related species [11], suggesting the presence of the repeated hybridization of Acropora, i.e., reticulate evolution. Furthermore, study of Acropora chromosomes also supported a hypothesis of reticulate evolution [14].
In contrast to diverse Indo-Pacific Acropora, only three species of Acropora inhabit the Caribbean: A. palmata (Lamarck, 1816), A. cervicornis (Lamarck, 1816), and A. prolifera (Lamarck, 1816). Acropora prolifera has been recognized as a hybrid between A. palmata and A. cervicornis, based on genetic [22,23] and reproductive [24] studies. This hybrid species, A. prolifera, has been known to be nearly sterile, suggesting that reticulate evolution with repeating hybridization and speciation may not occur in the Caribbean [23]. Vollmer and Palumbi [23] proposed that no reticulate evolution occurs in the Indo-Pacific, either, because sterile F1 hybrids may have been produced in the Indo-Pacific, as in the Caribbean. Opposing this suggestion, Miller and van Oppen [25] strongly insisted on the possibility of reticulate evolution in the Indo-Pacific. Recently, Willis et al. [13] reviewed the hybridization of corals and suggested that hybridization may occur in the Indo-Pacific only in peripheral regions of a coral species' distribution, but not in main tropical and subtropical reefs, although Wei et al. [26] reported no hybridization of Acropora species in Taiwan, which is a marginal habitat of corals. Thus, since Willis et al. [13], hybridization in the Indo-Pacific, especially of Acropora, has been believed to be quite rare.
Compared with hybrid studies of Caribbean corals, a paucity of species-specific data on the hybridization and on the survival rates of hybrid larvae exists for the Indo-Pacific because of the high number (.100) of resident Acropora species [27]. Detailed analyses similar to those performed on Caribbean Acropora [24] and Montastraea [21] should be performed in the Indo-Pacific. In addition, Indo-Pacific hybrid morphologies remain largely unknown, and have been reported only for A. millepora (Ehrenberg, 1834) 6 A. pulchra (Brook, 1891) [13]. Furthermore, to date, insufficient data are available to judge the occurrence of natural hybridization in the Indo-Pacific. Willis et al. [13] performed sperm selection experiments, mixing intra-and interspecific sperm against intraspecific eggs, and indicated that hybridization does not occur under conditions in which intraspecific sperm exists. However, they also suggested that hybridization may occur in regions with a low number of conspecific colonies.
Around Okinawa, Japan, coral reefs are well developed and more than 300 species have been reported [28]. However, since the mass bleaching event of 1998, a large amount of corals in Japan had decreased dramatically, due to repeated bleaching and outbreaks of crown-of-thrown sea stars [29,30,31,32]. Around Akajima Island, where our study was performed, the number of Acropora colonies also decreased [32,33]. The condition may be assumed to be the peripheral region of distribution of some Acropora species, in which low opportunity of intraspecific fertilization may increase opportunity of interspecific fertilization.
In this study, we focused on two morphologically different species, A. florida (Dana, 1846) and A. intermedia (Brook, 1891) (previously recognized as A. nobilis (Dana, 1846), see [27] in detail). These species are very common and sympatric in the Indo-Pacific (see Fig. 1 for the species distribution), and Hatta et al. [11] reported their hybridization with high (.70%) fertilization rates. These species are useful in hybridization studies because the parent species clearly differ morphologically, with A. florida being hispidose (bottle-brush-like) and A. intermedia being arborescent (long branch-like), meaning that the morphologies of probable F1 hybrids would be easily detectable. These two species also belong to a genetically closely related group (uncorrected p distance 0-2.5% in the mini-collagen gene [11]), suggesting the high possibility of hybridization in the past [11]. Thus, to examine the possibility of natural hybridization, we investigated the fertilization rates, embryonic development, and larval survivorship of A. florida, A. intermedia, and hybrids of these two species.  Fig. 1; 26u129N, 127u179E) with permission to Akajima Marine Science Laboratory from the Okinawa Prefectural Government (# 24-17). After collection, colonies were kept in the sea (2007) or in a water tank with running seawater (2012) until spawning. When settings of bundles on polyp mouths were observed, colonies were transferred into individual buckets. Gamete bundles were then collected from individual colonies using Pasteur pipettes and allowed to break apart in a small volume of filtrated seawater (10 mm pore size) to yield free eggs and sperm suspensions. These experiments were conducted in a room maintained at 26uC. Eggs were washed twice with filtrated seawater, and sperm suspensions were diluted to adjust their concentrations. Eggs and sperm collected from individual colonies were then mixed in pairwise combinations. Two hundred to three hundred eggs were mixed with sperm in 60-ml vials within 2 h after spawning. Final sperm concentrations were 0.5-2610 6 /ml, which is an effective fertilization concentration for Acropora [10]. The numbers of fertilized and unfertilized eggs were scored at the 16-cell/morula stage 4-5 h after gamete mixing. Also, to compare differences in the development process between intra-and interspecific crosses, the developmental stage of fertilized eggs was observed every 1-2 h after insemination. Self-fertilization experiments were also conducted at the same time.

Crossing experiments
Average fertilization rates for intra-and interspecific crosses were analyzed statistically by Mann-Whitney U-tests and Kruskal-Wallis tests, and all pairwise multiple comparisons were performed by the Steel-Dwass method.

Measurement of planular survivorship
In 2012, thirty-five hours after the start of the experiment, we transferred 20 planular larvae from each crossing combination into each of four Petri dishes. Planulae in the Petri dishes were counted and their condition was checked every 24 h for 30 days. The mean number of planulae alive in the four replicates was considered to represent planular survivorship. We calculated a survival curve for planular larvae based on the Kaplan-Meier estimator to evaluate larval survivorship [34]. To compare survivorship between two colonies, we determined the distribution of each. If the distribution was exponential, we used a Cox-Mantel test. If distribution was Weibull, we used a Wilcoxon test. If distribution was neither exponential nor Weibull, we used a logrank test [35].

Species identification and morphological analyses
All specimens used in crossing experiments were bleached for exact species identification and morphological analysis. Skeletal morphologies were observed using a digital microscope (Keyence). Species identification was mainly performed following literatures [27,36,37]. Based on taxonomic references, all specimens except one were identified as A. florida or A. intermedia ( Fig. 2A, 2B). For one colony, species identification was problematic for the following reasons: the colony shape was arborescent or nearly caespitose (Fig. 2C) and the colony appeared superficially to be A. intermedia or the branching type of A. florida. The axial corallites were large and dome-shaped ( Fig. 3C 1 ), which are also similar to A. florida and A. intermedia (Fig. 3A 1 , 3B 1 ). The radial corallites (Fig. 3A 2 , 3B 2 , 3C 2 ) were dimorphic and had long, tubular shapes with dimidiate or oblique openings [27], which suggested A. intermedia, as those of A. florida are appressed tubular shapes with round openings. Notably, this colony had many short secondary branches as well as incipient axial corallites (Figs. 2C 2 , 3C 1 ), but did not form the typical hispidose shape of A. florida. The microstructure of this colony, such as the shapes of the radial corallites and spinules of the coenosteum (Fig. 3C [4][5] were similar to those of A. intermedia ( Fig. 3B [4][5] ). Thus, this colony was intermediate in morphology, between A. florida and A. intermedia, and it was impossible to identify with certainty. Therefore, in this study, we named this morphology A. sp. ''int-flo'' (hereinafter, ''int-flo''). Fertilization rates are summarized in Table 1 and Fig. 4. Figure 5 shows examples of the crosses performed in 2012. Selffertilization was not observed (0%) in 19/20 colonies, but was noted in one colony (#AK63) of A. florida, where the rate of selffertilization was quite low (1.3%; see Fig. 5). Intraspecific crosses of the two species (A. florida and A. intermedia) showed high fertilization rates (82.1-88.1% on average; Table 1). Although the intraspecific crosses between two colonies each of A. florida and A. intermedia had 0% fertilization rates, these colonies were fertilized at nearly 100% with other colonies within species (see Fig. 5 for crosses between #AK50 and #AK64 of A. intermedia), suggesting that these colonies might have been clones or genetically similar [35].

Fertilization rate
Fertilization rates of interspecific crosses were recorded between A. florida and A. intermedia, and between A. florida and ''int-flo'' ( Table 1 ,Figs. 4,5). Notably, the fertilization rates differed with combinations of eggs and sperm. Crosses of A. florida eggs 6 A. intermedia sperm showed an average fertilization rate of 34.2%, but reciprocal crosses showed an average fertilization rate of 63.5%. Similarly, the fertilization rate of A. florida eggs 6 ''int-flo'' sperm was 34.5% on average, but 0% in reciprocal crosses. In the crosses shown in Fig. 4, the fertilization rates of most combinations of A. intermedia eggs 6 A. florida sperm exceeded 50% (maximum, 98.7%). In turn, the reciprocal crosses showed fertilization rates of ,50% in most combinations. Crosses between A. florida and ''intflo'' showed highly variable (6.0-95.3%) fertilization rates.
The average fertilization rates of intraspecific crosses between A. florida and A. intermedia did not differ significantly (Mann-Whitney U-test, Z = 2.57, p = 0.14), whereas the fertilization rates of interspecific crosses among combinations differed significantly intermedia eggs 6 A. florida sperm (x 2 = 3.03, p = 0.04). In addition, significant difference was found between A. florida eggs 6 A. intermedia sperm and the reciprocal cross (x 2 = 3.95, p = 0.00). For the crosses using ''int-flo'', A. florida eggs 6 ''int-flo'' sperm did not differ significantly against any other crosses although total number of crosses of this combination was too small (Table 1).

Development
All embryos of each intraspecific cross developed normally, as shown in [38], whereas the embryos of interspecific crosses were at different developmental stages at the same time for every cross, i.e., the two-, four-, and eight-cell stages were mixed within a cross.

Planular survivorship
None of the larval survival curves showed exponential or Weibull distributions, so we used log-rank tests to compare survivorship among crosses. However, we recalculated the significance levels based on the Bonferroni correction (a = 0.05/ 6 = 0.0083) when comparing the survival curves of more than four crosses. Figure 6 shows the survival curves of larvae resulting from intraand interspecific crosses. Two intra-and interspecific crosses (A. florida eggs 6 A. florida sperm and A. florida eggs 6 A. intermedia

Hybridization of Acropora
In this study, fertilization rates between A. florida and A. intermedia differed significantly between combinations of eggs and sperm (34.2% and 63.5% on average). The fertilization rates of A. intermedia eggs 6 A. florida sperm were higher than those of the reciprocal crosses. This result is very similar to that of a crossing experiment between A. florida and A. intermedia in 1998 reported by Hatta et al. [11]. The similar data obtained in 1998 [11], 2007, and 2012 (this study) suggest that the fertilization rates of these two species usually differ between combinations of eggs and sperm. This pattern is also similar to that of Caribbean Acropora species, in which the fertilization rates of two species, A. palmata and A. cervicornis, differ between combinations of eggs and sperm [24].
All embryos in the intraspecific crosses developed normally, as shown in [38], whereas gametes from interspecific crosses developed unevenly (see Results). Overall, many (not all) embryos of interspecific crosses were developmentally delayed by 30-60 min compared with those of intraspecific crosses, although all embryos of intra-and interspecific crosses had developed to the morula stage by 5 h after insemination. Considering that embryos of interspecific crosses developed normally (same as intraspecific embryos) after two cells stage, one main reason of delayed development of interspecific crosses could be a delay of egg-sperm fertilization. The reasons for delayed fertilization remain unclear, but it may be a survival mechanism by hybridization for Acropora species in severe conditions, such as those under which depleted conspecific populations may be found; i.e., when few gametes of conspecific colonies are present for fertilization, hybridization with gametes of sympatric particular species may occur. Morita et al. [39] showed that the sperm did not respond to the eggs among species that cannot hybridize. At present, the response of sperm to eggs among species that can hybridize remains unknown, but this work may provide clues to understanding the underlying mechanism.
The tendencies of planular survivorship were reversed from those of fertilization rates. Planular survivorship between A. florida eggs 6 A. intermedia sperm (34.2% fertilization) was as high as that of intraspecific crosses of A. florida, whereas planular survivorship between A. intermedia eggs 6 A. florida sperm (63.5% fertilization) was lower than those of A. florida eggs 6 A. intermedia sperm and intraspecific crosses of A. florida (Fig. 6). In acroporids, almost all planulae show settlement 5-8 days after fertilization [40,41] and .50% of planulae die by 30 days [40]. Meanwhile, some planulae remain competent to settle 30 days after fertilization [42]. In this study, of twenty planulae, eight to twelve planulae showed normal metamorphosis and settlement after 8 days in every crossing experiment when there was conditioned plate to be able to be settled by planulae. More detailed studies of settlement are required, but hybrid planulae may have high survival potential during the planular stage because their survivorship was higher than that of intraspecific crosses at 5-8 days. In scallops (among marine invertebrates; [43]), higher growth and survival have been  reported for hybrids than for intraspecific crosses in early-life stages. Similarly, the larvae of A. florida 6 A. intermedia hybrids may have greater potential for survival than do larvae produced from intraspecific crosses. Thus, hybridization does not appear to negatively affect early life stages, and hybrids can survive at least these stages.

Possibility of hybrids
In this study, we identified an ''int-flo'' colony with an intermediate morphology between A. florida and A. intermedia. The species identification of this colony was problematic due to its unusual morphological characters. Although we did not focus on taxonomy in this study, we will discuss briefly some taxonomic characters of this morphotype for future studies. When we collected the colony in the field, we first tentatively identified it as a morphological variant of A. intermedia or A. florida, due to the colony shape with thick branches and its sympatry with these species. However, at least two skeletal morphological characters (radial corallites, and numerous sub-branches) differed from the typical skeletal morphology of A. intermedia or A. florida (see method). In total, ''int-flo'' differed morphologically from A. intermedia and A. florida, but would be phylogenetically closely related to both of these species due to the shared characteristic of dome-shaped large axial corallites (Fig. 3A 1 , B 1 , C 1 ). In addition, the peculiar characteristics of numerous sub-branches and incipient axial corallites on a main branch have also been reported in a few species; e.g., A. samoensis (Brook, 1891) (senior synonym of A. wallaceae Veron, 1990), shown in [27,37], or the Indian and Red Sea species A. forskali (Ehrenberg, 1834) and A. pharaonis (Milne Edwards & Haime, 1860) [44]. These species, however, differ morphologically and ecologically from ''int-flo'' in some aspects (i.e., radial corallite structure, colony shape, and species distribution). At present, we cannot determine whether ''int-flo'' is an undescribed species or a morphological variant of a known species. Morphological analyses and taxonomical studies using more samples of this morph may yield a more precise identification of ''int-flo.'' However, considering the difficulty of species identification, the morphological similarities to A. florida and A. intermedia, and its hybridization with eggs of A. florida (summarized in Fig. 7), ''intflo'' is highly likely to be a hybrid produced by combinations of A. florida and A. intermedia. Richards et al. [45] also suggested that rare Acropora species in the Indo-Pacific are probably hybrids. As mentioned above, ''int-flo'' had a mixture of morphological patterns of A. florida and A. intermedia. Such mixed morphologies have been reported in the A. humilis group (A. humilis, A. gemmifera, A. monticulosa, and A. digitifera) classified by [27]. Based on morphological, reproductive, and genetic analyses of the A. humilis  group, Wolstenholme [46] suggested that two morphs, ''monthum'' (a mixture of A. monticulosa and A. humilis) and ''terete-mont'' (an unusual shape of A. monticulosa), had hybrid origins derived from A. humilis and A. monticulosa. Further molecular analyses of ''int-flo'' or the growth of the hybrids produced in this study to mature colonies are necessary to clarify the possible hybrid status of ''int-flo''. In the Caribbean, fertilization rates of A. palmata eggs 6 A. cervicornis sperm were 5-10% on average, while those of the reciprocal crosses were 50-70%. In addition, spawning time sometimes overlaps in these two Caribbean species, but A. palmata spawns about 10 to 20 min earlier than A. cervicornis, a pattern similar to that observed for A. florida and A. intermedia; i.e., A. florida spawned 10-40 min earlier than did A. intermedia (Fig. 7). Nevertheless, in the Caribbean, the colony shape of the F1 hybrid A. prolifera differs depending on which species provides eggs, suggesting that natural hybridization can occur in both directions, despite differences in spawning timing and low (5-10%) fertilization rates in one direction. Thus, considering data from Caribbean Acropora, it is highly possible that A. florida and A. intermedia hybridize in the field, at least around Akajima, Okinawa Japan. ''Int-flo,'' which had intermediate morphology between A. florida and A. intermedia, is likely to be a hybrid.
Although this study did not provide direct evidence of natural Acropora hybrids in the Pacific, the larval survival experiments inspired us to examine the role of hybrids. Survivorship rates were higher in intraspecific larvae of A. florida and hybrid larvae of A. florida eggs 6 A. intermedia sperm than in intraspecific A. intermedia larvae and hybrid larvae of A. intermedia eggs 6 A. florida sperm, suggesting that the eggs of A. florida induce larval survivorship more effectively than do those of A. intermedia, even though the larvae were hybrids (Fig. 6). Nevertheless, the survivorship of larvae from A. florida eggs 6 ''int-flo'' sperm was the lowest among all comparisons. This finding suggests that crosses of A. florida eggs 6 ''int-flo'' sperm might be backcrosses with reduced larval survivorship.
Willis et al. [13] suggested that the hybridization of corals may be more frequent at peripheral boundaries of species' ranges, which means lower number of conspecific colonies. A decrease in coral may cause a higher incidence and/or survivorship of hybrids. Around Akajima Island, a large amount of corals had been decimated by the outbreak of the crown-of-thorns starfish, Acanthaster planci during recent 2001 to 2006 [47], or coral bleaching there since 1998 [33]. These situations might increase hybridization rates, and ''int-flo'' may be a product of this process. More field research into the existence of ''int-flo'' and morphological, reproductive, and genetic comparisons of the artificial hybrids of A. florida and A. intermedia would provide us with more critical data in the near future.
In conclusion, to prove directly the existence of natural hybrids in the Pacific, growing hybrids produced in crossing experiments to adult colonies is absolutely necessary to examine their morphologies and reproductive capacities. To date, we have maintained the hybrids produced from the A. florida 6 A. intermedia crosses. In the near future, we hope to present further data to increase the understanding of Acropora hybridization. Figure 7. Summary of fertilization rates in crossing experiments, spawning times, and morphological characteristics. A) this study and B) the Caribbean. All data of the Caribbean were referred from [24]. Arrows show eggs fertilized by sperm (dash, nearly 0%; thin line, ,50%; thick line, .50%). Question marks mean no data available. Roman numerals and letters show the following morphological characters; I, colony form (a, hispidose; b, arborescent); II, radial corallite structure (a, appressed tubular with round opening; b, dimorphic, long tubular with dimidiate or oblique opening). doi:10.1371/journal.pone.0056701.g007