Introduction

Tropical coral reefs are built mainly by scleractinian corals and provide habitat or protection for huge numbers of organisms (Kaiser et al. 2005; Cooper et al. 2008). In recent years, coral reefs have faced severe anthropogenic disturbances and it is expected these impacts will increase in the form of climate change and ocean acidification (Pandolfi and Kießling 2014; Bozec and Mumby 2015). The increase in anthropogenic disturbances can lead to local or even regional extinctions of coral species and can therefore reduce genetic variation on the level of species, population and individuals (Gattuso et al. 2015). Genetic variation, however, is a prerequisite to respond to environmental changes (Boulay et al. 2012). Analyzing the amount of genetic variation at different organizational levels (species, population, individual) is necessary to be able to estimate the potential of survival and recovering from disturbances.

The amount of genetic variation in a coral species is influenced by different processes. Primarily, of course, mutations are the source of novel alleles. The mode and time frame in which they are distributed within population/species may vary considerably (Orive 2001). An essential factor for mutation distribution is the proportion of sexual and asexual reproduction (Ayre and Hughes 2000). During asexual reproduction, coral genomes are inherited as units, no segregation and recombination take place, and new mutations are only passed on within the clonal lineage. Consequently, genotypic variability is low (Baums et al. 2006). Clonal populations can be well adapted to given (constant) environmental parameters and may allow for population persistence (Lasker and Coffroth 1999). The main form of asexual reproduction in corals is fragmentation of certain parts of a coral colony (Harrison 2011). Furthermore, sexual reproduction in corals is an important factor affecting genetic variation. Due to chromosome segregation and recombination, sexual reproduction leads to high levels of genotypic diversity and therefore may improve survival in a wider variety of habitats (Ayre and Hughes 2004).

In addition to the proportion of sexual and asexual reproduction, dispersal potential plays a major role in the genotypic variability of populations (Baird et al. 2009). The majority of corals are broadcast spawners that release bundles of gametes freely into the water column during mass spawning events. Contrary to broadcast spawning organisms, brooders release fully developed larvae into the water column (Harrison 2011). Additionally, the dispersal potential of a brooding and/or spawning corals can be directly influenced by regional oceanographic movements (Boulay et al. 2012). Although it is not necessarily possible to infer different dispersal potentials from spawning and brooding corals (Miller and Ayre 2008), generally the varying ability for long term or wide dispersal of a coral species can influence genetic variation of a population (Graham et al. 2008; Baird et al. 2009; Torda et al. 2013a).

Besides the reproductive mode, in corals another mechanism might influence genetic variation at the individual as well as population level, the phenomenon of intracolonial genetic variability. Intracolonial genetic variability describes the occurrence of two or more distinct genotypes within a single coral colony (Pineda-Krch and Lethilä 2004; Puill-Stephan et al. 2009; Maier et al. 2011). Two different processes are discussed to cause the phenomenon. Mosaicism arises from the occurrence of somatic mutations; chimerism is caused by the fusion of two or more partners (Santelices 1999). In mosaic colonies, mutations can be spread during polyp division and lead to a coexistence of several different genotypes within a coral colony (Orive 2001; van Oppen et al. 2011). Due to the potential lack of a separated germline in corals, these somatic mutations may also affect the genetic variability of the reproduction products (Otto and Hastings 1998; Schweinsberg et al. 2014). Chimerism mostly occurs in gregariously settled juvenile corals (Raymundo and Mapa 2004; Amar et al. 2008). During this developmental stage, an alloimmune system is not fully developed and self- or non-self-recognition is generally absent (Puill-Stephan et al. 2009). Intracolonial genetic variability can have advantages as well as disadvantages. Possible benefits of genotypic heterogeneity could be a more diverse genetic repertoire and better equipment for competition as faster growth rates (Rinkevich and Weissmann 1992; Santelices et al. 2010; Maier et al. 2011). The appearance of intraorganismal conflicts among genotypes, resulting in less endowment for competition with surrounding corals could be a disadvantage (Otto and Hastings 1998; Santelices 1999; Work et al. 2011).

The coral Porites lobata (Dana 1846) is a massive, slow growing species which is widely distributed in the South Pacific and South Atlantic oceans. It is a gonochoristic species with mainly broadcast spawning reproduction. However, asexual reproduction as fragmentation has been described (Glynn et al. 1994; Cortés 1997). P. lobata is a common reef building organism and dominates back reef margins, lagoons and some fringing reefs. It belongs to the main reef builders in Moorea, French Polynesia, especially in the Cook’s Bay. The aims of this study were to investigate genetic variation including genotypic diversity of the main reef building coral P. lobata in French Polynesia to better understand local genetic patterns. We analyzed tissue samples of three different size (age) classes of P. lobata to answer the questions: How diverse is the local genetic pattern of P. lobata? Do we find evidence for intracolonial genetic variability? And if so: Is intracolonial genetic variability represented in the different size (age) classes?

Materials and methods

Sample collection

The investigated P. lobata samples from eight sampling sites were collected in the Cook’s Bay in Moorea, French Polynesia, South Pacific (17°28′37.45″S 149°50′40.35″W) in December 2013 (Fig. 1a). To exclude the possibility of sampling, two different individuals as a single one, all colonies with different morphological types, colors and rejection lines within a single colony were omitted from sampling. All corals were sampled in water depths between one to three meters to ensure equal levels of potentially mutagenic UV radiation. As we were interested in small-scale genotypic patterns within Cooks Bay, colonies were randomly sampled in a 14-m radius (Fig. 1b, Table S2). The distance between the different sampling sites varied from 212 to 1296 m (Table S3). The sampling pattern included one adult colony (2 m in diameter) in the midpoint of the site, four young adult colonies (0.5–0.8 m) and ten juvenile colonies (<0.05 m) in a 360° circumference around the midpoint (Fig. 1b, Table S2). To get a more detailed view of each sampling site, we differentiated the sampled colonies according to their colony size. Colony size was used as a proxy for age, assuming three age classes: juvenile, young adult and adult. This calculation was based on a mean linear extension per year of 0.5–1 mm for Porites (Lough and Barnes 1992; Smith et al. 2008; De’ath et al. 2009; Lough and Cooper 2011). However, it is important to emphasize that for plants and corals size does not necessarily reflect the age of an organisms (Hughes 1984; Okubo et al. 2007) and consequently size is considered here as a proxy. Additionally, to test for the occurrence of intracolonial genetic variability, all colonies were sampled multiple times at the maximum possible distance (dependent on colony size). The maximum possible distance between samples was chosen to ensure optimal colony coverage and the highest potential to detect chimeras. Adult colonies were sampled ten times with a minimum distance of approximately 20 cm between tissue samples. Young adult colonies were sampled five times with a minimum distance of approximately 10 cm between tissue samples, and juvenile colonies were sampled three times with a minimum sample distance of 1 cm. Sample number was adjusted to colony size to ensure that the injury and tissue loss would not destroy the coral. The ten, respectively, and five tissue samples per colony were taken in a clockwise orientation of eight (four) sites of the periphery of every coral with an additional sampling of two (one) sites in the middle. In threefold sampled colonies, samples were taken from the points of a triangle All samples were preserved individually in absolute ethanol and stored at −20 °C until the molecular analyses in the laboratory in Bochum. Colonies were visually identified in the field by M. Schweinsberg based on corallite morphology (Veron and Stafford-Smith 2000), and taxonomic identity was later confirmed by light microscopy. All sampled colonies were similar in visual appearance. However, several studies have shown that identifying P. lobata based on morphology can be difficult (e.g., Forsman et al. 2009, 2015). We remark, that we cannot not fully exclude, that different cryptic lineages of P. lobata were sampled.

Fig. 1
figure 1

Map of the sampling area. a A detail view of the Cook’s Bay on Moorea, French Polynesia (17°29′10.75″S and 149°49′37.32″W). Shown are the eight different samplings sites marked as Site 1–Site 8. b Exemplary the small-scale sampling scheme for Site 1. The triangle corresponds with one adult coral colony; the squares correspond with four young adult coral colonies and the circles correspond with ten juvenile coral colonies

Full size image

Molecular analyses

Genomic DNA of the coral tissue samples was extracted using DNAzol® and polyacrylamide carrier (Molecular Research Center Inc., Cincinnati, USA) as described in Maier et al. (2001) and Schweinsberg et al. (2014). The tissue samples were amplified at nine different microsatellite loci for every colony (Table 1). Only primers which not amplified Symbiodinium DNA were chosen for further analyses to exclude symbiont contamination in the data. For each primer pair, the originally published sequence was adapted by adding a universal M13 tail (Oetting et al. 1995; 5′-CAC GAC GTT GTA AAA CGA C-3′) to the 5′ ends either to the forward or to the reverse primer (Table 1). PCRs were performed in a total volume of 10 μl with HotMaster Taq (5 Prime, Hamburg, Germany) and included 0.5/0.25 units of Taq, 1 μl extracted DNA template, 1× supplied PCR buffer, 2 mM MgCl2, 0.2 mM of each dNTP, 0.2 μM of sequence-specific untailed primer, 0.05 μM of sequence-specific tailed primer, 0.2 μM of fluorescently labeled universal M13 primer and 5% dimethylsulfoxide. Amplifications were done using a Flexcycler (Analytik Jena AG, Jena, Germany) with the following cycling protocol: Initial denaturation 94 °C for 300 s was followed by 40 cycles at 94 °C for 30 s, 50–60 °C for 30 s and 65 °C for 30 s and a final elongation of 65 °C for 300 s. Allelic size was determined with a high-resolution polyacrylamide gel electrophoresis and the software Saga 2GT (LI-COR, Biosciences, Bad Homburg, Germany). To avoid genotyping errors, all deviating samples were re-amplified a second or even third time. In addition, all genotypes were scored independently by a second person in order to confirm the deviating genotype. All peak profiles that were either faint or ambiguous were omitted from the analyses.

Table 1 Summary of microsatellite locus diversity in P. lobata
Full size table

Statistical analyses

To quality check the microsatellite markers, the most common genotype per colony was used to avoid artificial (multiple sampling induced) errors. Deviations from Hardy–Weinberg equilibrium at each locus and linkage disequilibrium were tested for each pair of loci using the program Arlequin 3.5 (Excoffier et al. 2010, Table 1). Multilocus genotypes were produced, reviewed for genotyping errors, and all genotypes were analyzed for uniqueness with Genclone 2.0 (Arnaud-Haond and Khalid 2007). Allelic diversity and genotypic diversity indices were calculated for every site. Here we differentiated between allelic diversity as measurement of variation of a single locus in a population and genotypic diversity as variation between multilocus genotypes of individuals.

Allelic diversity was estimated as average of all eight sites with the program HPrare (Kalinowski 2005) by number of alleles (A), number of private alleles (AP). The average observed (HO) and expected heterozygosity (HE) were estimated with the program Arlequin 3.5 (Excoffier et al. 2010).

For genotypic diversity, the genotypic richness (unique genotypes per sampling site), the evenness and the Simpsons diversity index were estimated for all samples. Genotypic richness is directly proportional to frequency of sexual recruitment, and evenness will approach 1 in cases of every genotype has equal numbers of colonies (Boulay et al. 2012). Those measurements provide an overview about comparable genotypic diversity per site.

To evaluate population differentiation, F ST values among all sites were calculated. A Mantel test was conducted with the genetic distances and the geographic distances between the eight sites to test for isolation by distance. All analyses were conducted with Arlequin 3.5 (Excoffier et al. 2010). Additionally, a principal component analyses (PCAs) were performed with the program PCAGEN (Goudet 1999) to test the multilocus genotypes for clustering.

The Bayesian clustering program STRUCTURE 2.3.4 (Pritchard et al. 2000) was used to estimate the number of discrete populations (K) for a more detailed population differentiation. Initial STRUCTURE runs used the default settings with a burn-in period of 10,000 and 10,000 MCMC repeats with 20 iterations per K (1–10). The results were then uploaded to STRUCTURE HARVESTER (Earl and von Holdt 2011). The cluster (K = 3) determined most likely was used to perform a second run in STRUCTURE with more iteration (100) and higher values for the burn-in period (100,000) and MCMC repeats (100,000), to increase the accuracy of the clustering. The resulting merged data file was then processed with the program CLUMPP 1.1.2 (Jakobsson and Rosenberg 2007) and visualized by DISTRUCT 1.1 (Rosenberg 2003). Further, the relatedness of all individuals was analyzed with the program MLrelate (Kalinowski et al. 2006). The MLrelate results were tested with Statistica V 3.5 for differences in mean relatedness among size classes.

To test all colonies for the occurrence of intracolonial genetic variability, the genotypes from all samples per colony were compared. For the comparison, we distinguished between a main genotype (the most often found genotype in the tissue samples) and deviating genotype(s) for each of the colonies. Exemplarily colony C13 consisted of three tissue samples. Two tissue samples showed an identical multilocus genotype, and the third tissue sample showed a multilocus genotype which deviated in one locus. In this example, the two identical multilocus genotypes were count as the main genotype and the deviating second multilocus genotype was count as deviating (intracolonial genetic variability) multilocus genotype. Following the differentiation of main and deviating genotypes per colony, every colony was analyzed for the occurrence of mosaicism and/or chimerism. For detection of mosaicism, we followed the theoretical assumption that genotypes which are caused by somatic mutations should only diverge from the main genotype in a few (1 or 2) loci or alleles and more than two independent somatic mutations are very unlikely in the same tissue. For detection of chimerism, the product of larval fusion of genetically different partners, we assumed that deviating genotypes should show a high variation compared to the main genotype of the colony (Orive 2001; Puill-Stephan et al. 2012; Maier et al. 2011; Schweinsberg et al. 2015). Therefore, a threshold of 60% multilocus genotypic difference to the main genotype was chosen to identify chimeras. Less than this 60% difference was referred to a mosaic (Schweinsberg et al. 2015). These theoretical assumptions were applied to our dataset. The deviating genotype of exemplarily colony C13 varied in one out of nine loci, which corresponds to a difference of 11.1% multilocus genotypic differences (Table 3). To visualize the deviating genotypes per coral colony, a STRUCTURE 2.3.4 (Pritchard et al. 2000) run was perform with equal settings as for the intraspecific analyses. Additionally, all results of the second run were compressed and uploaded to STRUCTURE HARVESTER (Earl and von Holdt 2011). The resulting merged data file was then visualized with the program CLUMPP 1.1.2 (Jakobsson and Rosenberg 2007) and processed by DISTRUCT 1.1 (Rosenberg 2003). Due to known problems with missing data in STRUCTURE, we substituted all missing loci with the main genotype per coral colony. This did not affect the amount of intracolonial genetic variability.

Results

Marker quality

DNAzol® DNA extraction was successful for all 120 colonies sampled. A total of 480 tissue samples were genotyped with (in most cases) nine different markers. All tested microsatellite loci showed high variability (between 6 and 21 alleles, Table 1). We found no consistent pattern of linkage between any of the tested loci per site. Microsatellites did deviate from the Hardy–Weinberg equilibrium. These deviations, however, were most likely caused by the relatively small sample size (Waples 2014). Interestingly, the loci PAGA24 and PL1868 showed alleles that differed from their regular repeat type. In both cases, alleles with a deviation of one base were detected. It has been described that point mutations in the flanking regions can occur in microsatellites (Ellegren 2000). Therefore, we assume a point mutation to be causal for the deviations in the two markers and counted them as separate alleles. Due to the relatively high number of occurrences of these deviations and therefore an increase in number of alleles in the two loci, we conducted all follow-up analyses twice, excluding and including PAGA24 and PL1868 the markers. The results of this test showed no differences, and we decided to conduct all analyses including all nine microsatellite loci.

Genotypic diversity and allelic diversity

Genotypic diversity was high in all tested sampling sites (Table 2). Only two clones were detectable in the whole sample set of 120 colonies. Two individuals with identical genotypes were found in sampling site four and two individuals with a different identical genotype were found in sampling site five. The distance between the clones was three meters (Site 4) and two meters (Site 5), respectively. Additionally, we calculated genotypic diversity including all samples of all eight sites. These results showed a genotypic richness of 0.988 and an evenness of nearly 1 (Table 2).

Table 2 Genotypic diversity and allelic diversity
Full size table

Allelic diversity was high in all eight sampling sites. Coral colonies carried on average 5.80–7.16 alleles per locus. Mean observed heterozygosity across all sites was 4.808 ± 0.138. A few private alleles were found (between 0.02 and 0.49 per site, Table 2).

Population structure and relatedness

The results of the Mantel test showed no evidence for isolation by distance. The PCA showed two separated clusters. Both axes (factors) were not significant, the x-axis explained 9.31% and the y-axis explained 7.85%. One cluster consisted of six adult and seven young adult colonies, whereas the second cluster consisted of all juvenile and the remaining two adult and eighteen young adult colonies (Fig. 2).

Fig. 2
figure 2

Principal component analysis of genetic differentiation based on multilocus genotypes. The black triangles correspond with adult colonies; the black squares correspond with young adult colonies and the black circles correspond to juvenile colonies. The PCA showed two separated clusters. Both axes were not significant, the x-axis explained 9.31% and the y-axis explained 7.85%. One cluster consisted of six adult and seven young adult colonies, whereas the second cluster consisted of remaining juvenile and the remaining two adult and eighteen young adult colonies

Full size image

The STRUCTURE analyses also found no clear differentiation between the eight sampling localities (field sites) (Fig. 3). Genotypes from one field site were assigned to several different clusters. While no geographic clustering of genotypes was observed, individuals within one size class seemed to be related (Fig. 3). STRUCTURE sorted most of the adult and some young adult colonies into the same cluster. In total six out of the eight adult colonies, except C76 were assigned to this cluster (Fig. 3). Additionally, five young adult colonies and only one juvenile colony (C25) were also assigned to this cluster. The remaining 23 young adult colonies clustered randomly with the juvenile colonies.

Fig. 3
figure 3

STRUCTURE assignment plot for all multilocus genotypes (K = 3). Bar graph shows the cluster membership coefficient Q (color coded) on the y-axis. Sampling sites are displayed along the x-axis. Sampling sites are separated by vertical black lines. Age of the coral colonies is coded with light gray bars (adult colonies), gray bars (young adult colonies) and black bars (juvenile colonies). No differentiation among the eight sampling sites was detectable. Corals from similar age stages were mostly sorted to one cluster

Full size image

To compare the genetic structure in more detail, the relatedness of individuals in the three age classes was analyzed. MLrelate showed low values in all relationships tested, which supported the STRUCTURE cluster analyses. Low relatedness within every of the tested age clusters (0.0522 ± 0.1268 within adults, 0.0522 ± 0.0940 within young adults, 0.0686 ± 0.089 within juvenile corals) among every age clusters (0.0635 ± 0.1137 among adult and young adult, 0.0606 ± 0.0913 among adult and juvenile corals, 0.0602 ± 0.0904 among young adult and juvenile corals) was detectable. Further, the mean relatedness among all sampling spots was 0.0984 ± 0.1838, the relatedness between all sampling sites was 0.0890 ± 0.1925, and the mean relatedness within the sampling sites was 0.1485 ± 0.2289. Summarizing, low levels of relatedness were detectable between all spots and individuals. Both analyses, STRUCTURE and MLrelate, suggested a high genotypic diversity in all age classes and low relatedness between individuals.

Intracolonial genetic variability

To get an overview of intracolonial genetic variability in P. lobata, we compared the multilocus genotypes of all tissue samples per coral colony. Here we distinguished between the main genotype (the most often found genotype in the tissue samples) and deviating genotypes (genotypes which were found only in one or two tissue samples). Out of the 120 analyzed colonies, deviating genotypes could be found in 14 colonies. This corresponds to a proportion of 11.7% heterogeneous coral colonies.

The STRUCTURE plot displayed the deviating genotypes. Due to summarizing of separate STRUCTURE runs by CLUMPP and DISTRUCT not all deviating genotypes were visualized. However, in accordance with the chosen minimum of 60% assignment probability to the main genotype, all detectable deviating genotypes varied less than 60%. Therefore, all deviating genotypes were identified as mosaics (Fig. 4).

Fig. 4
figure 4

STRUCTURE assignment plot for all 14 colonies that showed more than one genotype. The x-axes show each tissue sample combined in the colonies they came from (colony numbers below the axes). The y-axes shows the cluster membership coefficient Q (color coded). Colonies C16 and C31 were sampled ten times, colonies C17–C107 were sampled five times and colonies C13–C111 were sampled three times each. Samples marked with M (mosaic) display deviating genotypes from main genotypes. In colony C16, three tissue samples showed deviating genotypes; in colony C18, two samples showed deviating genotypes. The remaining twelve colonies showed one deviating genotype

Full size image

Overall twelve out of the fourteen heterogeneous colonies consisted of two different genotypes, a rather common main genotype and a second deviating genotype. The colony C16 showed three distinguishable genotypes consisting of one common main genotype and two deviating genotypes (Table 3). Although all of those deviating genotypes were caused by allelic differences within the single coral colonies, no new alleles were identified, instead only deviating allele combinations could be identified (Table 3).

Table 3 Summary of intracolonial genetic variability
Full size table

The mosaics were distributed over all three size classes. Two heterogeneous colonies belonged to the adult colonies, five to the young adults and seven to juvenile colonies (Fig. 4). If we calculate relative values for each of the three age classes, 25% of the large adults, 15.6% of young adult colonies and 8.8% of juvenile colonies consisted of mosaics. Additionally, the mosaics were scattered randomly and occurred at seven of the eight sampling localities (Table 4).

Table 4 Sampling site distribution and size class distribution of detected mosaics
Full size table

Discussion

The main focus of our study was to analyze the small-scale genotypic structure in P. lobata to get information about genotypic diversity, allelic diversity and intracolonial genetic variability. Overall, our results showed high levels of genetic variation in the analyzed population in Cooks Bay in Moorea.

Genetic variation and genotypic diversity

The evaluation of allelic variability and genotypic diversity is essential to understand the relative contribution of sexual and asexual reproduction to local populations (Boulay et al. 2012). Our study of P. lobata showed high levels of genotypic diversity even at the very small geographic scale investigated with a mean genotypic richness of 0.98, evenness of 0.998 and a mean of 6 alleles per locus. The detected proportions of allelic variability are slightly higher than in an earlier conducted study of P. lobata from Panama (genotypic richness between 0.75 and 0.9, evenness 0.7–1.0 and 4 alleles per locus, Boulay et al. 2012). From these results, it can be inferred that the predominant mode of reproduction in P. lobata is sexual reproduction. This consideration is also supported by the low number of clones identified in our dataset. Fragmentation or other asexual reproduction modes have been documented in P. lobata by Cortés (1997) but seem to be rare. The massive growth form of P. lobata in Moorea seems to withstand most mechanical forces, e.g., fish or wave impact and asexual fragmentation is restricted to some rare cases. The high level of genotypic and genetic variation found is to emphasize because the reefs of Moorea had to deal with large-scale natural perturbations (mass outbreak of Acanthaster planci and hurricanes) during the last decades. Several studies suggest that those impacts could clearly have reduced the genetic variation within different coral species by impeding adult fertility and recruit survival (Omori et al. 2001; Adjeroud et al. 2009, 2014; Trapon et al. 2011).

Relatedness among age classes

Information about relatedness patterns among different age classes of a population can offer new insights into how sexual reproduction may increase genetic variation. The relatedness between individuals of a population can provide insights into recruitment patterns. Recruits can be of local or distant origins (Palumbi 2003). Depending on recruits’ origin, they can contribute in different ways to the genotypic diversity of a population. Local recruits mainly preserve existing variability, whereas distant recruits can lead to gene flow between populations and sometimes to the introduction of new alleles (Ayre and Hughes 2000). The comparison of the tested age classes (adults, young adult and juvenile) of P. lobata showed that adult colonies were more closely related to each other than other age classes. Additionally, less relatedness was detected between each of the age classes. The high level of genetic variation found within the tested population of P. lobata and the low relatedness between the adult and the juvenile colonies suggest a distant origin of recruits. Generally, genetic distance and variation between populations depend on regional oceanographic movements in combination with species-specific larva characteristics (Davies et al. 2015). Furthermore, local recruitment is influenced by settlement behavior and competency of larvae as well as water currents (Rinkevich 2006). Contrary to brooding species, where it has been shown that larvae often settle near to their parents, spawned larvae need more time for development and therefore might cover long distances till they settle (Underwood et al. 2007). According to water movement patterns in the Cooks Bay (see, e.g., Hench et al. 2008), recruits from distant locations can be introduced into the lagoon and increase the genotypic variability. Further, the spawning products and fertilized larvae produced by the local Cooks Bay population are likely to be washed out of the bay before they are ready to settle. Similar patterns have been described for P. lobata in Samoa and Panama (Barshis et al. 2010; Boulay et al. 2012).

A different interpretation for our datasets is the potential occurrence of two or more different cryptic species/lineages within in the sampled P. lobata colonies. The results of the STRUCTURE analysis showed three different clusters in the sampling set. STRUCTURE sorted most of the adult colonies within one cluster, whereas the juvenile colonies and the young adult colonies were sorted in two other clusters. Furthermore, the PCA showed two clusters. One cluster containing mostly adult colonies and a second cluster with mostly young adult and juvenile colonies. Several studies suggest that the identification of P. lobata is difficult and a potential misinterpretation of different species as P. lobata may occur (Forsman et al. 2009, 2015). Additionally, the occurrence of cryptic lineages has been described in numerous other coral species (e.g., Torda et al. 2013b; Schmidt-Roach et al. 2014; Warner et al. 2015) including Porites species (e.g., Jameson and Cairns 2012; Prada et al. 2014; Boulay et al. 2014). Although we cannot exclude the possibility that we sampled two or more cryptic lineages within our sampling set, we would have expected to see clear differences in marker amplification success and allelic ranges in the different size classes, as microsatellites are quite specific markers. We therefore assume that the allele frequencies differences among age classes are more likely to explain the differences found than the potential occurrence of cryptic lineages.

Intracolonial genetic variability

Next to the detection of genetic variation, our study aimed to describe the phenomenon of intracolonial genetic variability and its potential influences on the genetic variation found. Our results showed high level of intercolonial genotypic variability as well as intracolonial genotypic variability in all three size classes in P. lobata. The detected fourteen heterogeneous colonies consisted of seven juvenile colonies, five young adult colonies and two adult colonies. Our small-scale sample approach showed that in seven out of the eight sampling sites heterogeneous colonies could be found. According to the differentiation approach, all fourteen heterogeneous coral colonies were identified as mosaics. The average proportion of intracolonial genetic variability in all three age classes was 11.6% and therefore less than in previous studies on scleractinian corals. In the Great Barrier Reef up to 45.0% of local, coral populations consisted of heterogeneous coral colonies (Maier et al. 2011; Schweinsberg et al. 2015). Particularly for the massive coral Porites australiensis 24.5% of heterogeneous, colonies were identified (Schweinsberg et al. 2015). Due to the fact that the majority of our samples were small colonies and adult organisms were limited to a few individuals, the average amount of heterogeneous colonies is only partially comparable with the GBR study, which was conducted on 25 adult P. australiensis colonies. However, if we compare the relative proportions of our adult and young adult colonies with the found amount of heterogeneous colonies in P. australiensis, the adult colonies showed 25% and the young adults 15.6% mosaic colonies and showed therefore at least in the adult colonies similar values in P. lobata. An interesting finding is the occurrence of mosaics in juvenile corals. Although it is commonly assumed that mosaics (or somatic mutations) accumulate over time and therefore are much more frequent in adult colonies (Orive 2001; Puill-Stephan et al. 2012) already 8.8% of the juvenile colonies were heterogeneous.

Furthermore, all deviating genotypes found in the mosaic colonies showed unique genotypes. Due to the occurrence of new unique genotypes within intracolonial genetically variable colonies, those new genotypes increase the genotypic richness and evenness. Consequently, the genotypic diversity increases, too. If we include the occurrence of intracolonial genetic variability into allelic diversity of the tested P. lobata population, the phenomenon did not influence the single locus diversity. Even though, we did not find novel alleles in our mosaic genotypes, theoretically, new alleles could arise from mutations and being passed on to the next generation, thereby increasing the genepool of the population. Hence, they could affect the genetic variation not only in individual organisms but also in a population (Otto and Hastings 1998).

Overall, our results show that intracolonial genetic variability seems to be common in P. lobata. With those findings, new questions arise. Do novel genotypes with deviating alleles persist, particularly in juvenile corals, during development or will they be overgrown by the main genotype? The occurrence of mosaics in adult colonies suggests that a persistence and accumulation of mutations could be possible. Consecutively, additional questions about the fitness advantages of those deviating genotypes have to be investigated.

Our observation of mosaics being more common than chimeras is not surprising as it confirms earlier studies (e.g., Schweinsberg et al. 2015). However, future analyses should try to also investigate chimeras and test whether chimerization success might be correlated with the level of relatedness (kin or conspecific) between settling larvae.

Further studies have to be done to evaluate the effect of geographic and local influences of the frequency of intracolonial genetic variation. Our results did not show any differences in occurrence of intracolonial genetic variability within the eight sampling sites; however, the overall level of intracolonial genotypic diversity in Moorea was low compared to other field sites (Great Barrier Reef). Nonetheless, the influences of locally specific habitat conditions as well as the influence of competing organisms to the formation of intracolonial genetic variability need to be tested in more detail. Only with this information, we will be able to understand the causes for harboring multiple genotypes within a single coral colony and the potential influences on genetic variation.

Conclusion

This study presents a first overview of genotypic variability on intercolonial and intracolonial genetic variability in P. lobata from Moorea. Overall, high levels of genotypic variability at individual and population level were found. Our results show that sexual reproduction is the dominant mode of reproduction in P. lobata. The detected intracolonial genotypic variability occurred in all three age classes and indicates that mosaics exist in juvenile corals, too.