Sexual vs. asexual reproduction in an ecosystem engineer: the massive coral Montastraea annularis



    1. Marine Spatial Ecology Laboratory, School of BioSciences, University of Exeter, Prince of Wales Road, Exeter, EX4 4PS, UK; and
    Search for more papers by this author

    1. Division of Marine Biology and Fisheries, Rosenstiel School of Marine and Atmospheric Science, University of Miami, Miami, FL 33149, USA
    Search for more papers by this author
    • Present address: Department of Biology, The Pennsylvania State University, 208 Mueller Laboratory, University Park, PA 16803, USA


    1. Marine Spatial Ecology Laboratory, School of BioSciences, University of Exeter, Prince of Wales Road, Exeter, EX4 4PS, UK; and
    Search for more papers by this author

Nicola L. Foster, Marine Spatial Ecology Laboratory, School of Biosciences, University of Exeter, Prince of Wales Road, Exeter, EX4 4PS, UK. E-mail:


  • 1Long-lived sedentary organisms with a massive morphology are often assumed to utilize a storage effect whereby the persistence of a small group of adults can maintain the population when sexual recruitment fails. However, employing storage effects could prove catastrophic if, under changing climatic conditions, the time period between favourable conditions becomes so prolonged that the population cannot be sustained solely be sexual recruitment. When a species has multiple reproductive options, a rapidly changing environment may favour alternative asexual means of propagation.
  • 2Here, we revisit the importance of asexual dispersal in a massive coral subject to severe climate-induced disturbance. Montastraea annularis is a major framework-builder of Caribbean coral reefs but its survival is threatened by the increasing cover of macroalgae that prevents settlement of coral larvae.
  • 3To estimate levels of asexual recruitment within populations of M. annularis, samples from three sites in Honduras were genotyped using four, polymorphic microsatellite loci.
  • 4A total of 114 unique genets were identified with 8% consisting of two or more colonies and an exceptionally large genet at the third site comprising 14 colonies.
  • 5At least 70% of multicolony genets observed were formed by physical breakage, consistent with storm damage.
  • 6Our results reveal that long-lived massive corals can propagate using asexual methods even though sexual strategies predominate.


Clonal propagation, using fragmentation, budding or the production of asexual planulae, is considered to be an adaptation to both unfavourable local environmental conditions and relatively stable habitats. It allows a species to persist when it is unable to complete its sexual reproductive life cycle (Honnay & Bossuyt 2005) and it enables well-adapted genotypes to become dominant in the absence of moderate to high levels of disturbance (Miller & Ayre 2004). Examples of clonal propagation are found in many invertebrate and plant taxa from both aquatic and terrestrial environments; coast redwood trees (Douhovnikoff, Cheng & Dodd 2004), European aspens (Suvanto & Latva-Karjanmaa 2005), weedy perennial plants (Ceplitis 2001), epizoic anemones (Bronsdon et al. 1997), gorgonians (Gutierrez-Rodrigues & Lasker 2004) and stony corals (Ayre & Hughes 2000; Baums, Miller & Hellberg 2006; Severance & Karl 2006). In some species, sexual and asexual reproduction contribute equally to population growth (Sun, Gao & Cai 2001; Weppler, Stoll & Stocklin 2006) while in several species one mode of reproduction clearly dominates over another (Bronsdon et al. 1997; Ceplitis 2001; Magalon, Adjeroud & Veuille 2005). Furthermore, the contribution of asexual reproduction to recruitment can vary among populations of a single species across its range (Coffroth & Lasker 1998; Eckert 2001; Le Goff-Vitry, Pybus & Rogers 2004; Baums et al. 2006).

The balance between sexual and asexual reproduction within a species can be influenced by both biotic and abiotic factors. In marine environments, disturbance events can dramatically alter the contribution of asexual reproduction to recruitment (Henry & Kenchington 2004; Le Goff-Vitry et al. 2004; Rasheed 2004). For example, Le Goff-Vitry et al. (2004) documented an increase in the contribution of asexual reproduction to recruitment in the deep-sea, branching coral Lophelia pertusa in the presence of intensive fishing trawling. Disturbance may also affect the genotypic diversity of a species. Hunter (1993) described decreasing levels of genotypic diversity in populations of the finger coral Porites compressa with increasing levels of natural and anthropogenic disturbance. Interestingly, Coffroth & Lasker (1998) observed the greatest genotypic diversity in populations of the gorgonian Plexaura kura, in low and high disturbance environments and the lowest genotypic diversity in populations in intermediate disturbance environments. While these studies demonstrate the occurrence of asexual reproduction within a number of sessile organisms, they focus on species with a branching morphology. Detailed analysis of asexual dispersal in large, long-lived organisms with a massive (‘mound like’) morphology has not previously been provided.

Stony corals employ several methods of asexual reproduction. Whilst all corals grow using asexual budding of individual coral polyps (Jackson 1977), only branching species are believed to employ methods of asexual fragmentation for colony dispersal (Tunnicliffe 1981; Highsmith 1982; Lirman 2000). The branching morphology of these species makes them particularly susceptible to waves such as those caused by hurricanes. Wave energy can break apart colonies, distributing fragments across the reef, which later re-attach to the substrate and continue growth. In contrast, the dome-shaped morphology of many massive, long-lived coral species fits a tolerance model for withstanding hurricane disturbance (Massel & Done 1993). The low relief of such corals facilitates a laminar flow of water across the colony and minimizes the forces of lift that would otherwise uproot colonies during storms. After major storms, fragments of massive corals are far less common than those of branching corals (PJM, pers. obs.) although some massive colonies may be overturned (Bries, Debrot & Meyer 2004). The relative scarcity of colony fragmentation, together with morphological adaptations to withstand physical disturbance, has led to an implicit assumption that asexual mechanisms are unimportant for the dispersal of massive corals. This assumption is evident by the contrast of more than 30 years of research into asexual dispersal in branching corals vs. an absence of such studies on massive corals.

Here, we challenge the assumption that long-lived massive corals do not utilize asexual means of dispersal. We study clonality in the massive coral Montastraea annularis (Ellis and Solander), which is a major framework-builder of Caribbean reefs and forms dome-shaped colonies, often over 1 m in diameter. Many colonies within the population are estimated to be > 100 years old with an average growth rate of < 10 mm per year (Dustan 1975; Gladfelter, Monahan & Gladfelter 1978). Sexual reproduction occurs annually utilizing a mass-spawning event (Szmant 1991), yet despite this annual release of gametes, sexual recruits are infrequently observed on the reef (Bak & Engel 1979; Smith 1992; Mumby 1999).


study species

Colonies of Montastraea annularis may reproduce asexually using two methods. The first, intracolony fission, occurs because of partial-colony mortality (Hughes & Jackson 1985) and the second occurs when physical disturbance cleaves the colony resulting in the physical dislodgement of one or more fragments that then re-attach nearby (Highsmith 1982). In this study, we focus on the latter process which leads to dispersal of clones beyond the parent colony (i.e. the generation of structurally independent colonies rather than intra-colony fission).


Three sampling sites were established on the North coast of Roatan, Honduras in October 2004; Seaquest (16°17′39″N, 86°36′00″W), Sandy Bay (16°20′02″N, 86°34′04″W) and Western Wall (16°16′14″N, 86°36′16″W). A total of 146 colonies were sampled at the three sites. Honduras experiences intermediate frequencies of hurricanes, with the number of hurricane strikes being scaled by distance and storm intensity (Gardner et al. 2005). Within the last 30 years, one category 4 hurricane passed within 100 km of Roatan. Two category 3 hurricanes passed within 60 km and six tropical storms and category 1 and 2 hurricanes passed within 35 km ( No information is available on the impact the storms had on the sites. Sites were selected based on their prolific M. annularis populations and were located 1–3 km apart. Western Wall was located at the western tip of the island and approximately 1 km from Seaquest. Sandy Bay was a further 2 km east of Seaquest. Each site was located on the forereef at a depth of c. 5 m and a circular sampling plot was established with a radius of 5 m. Every M. annularis colony within each plot was tagged and its distance (to the nearest 5 cm) and bearing (to the nearest 5°) from the centre of the sampling plot (marked by a stake) were recorded. Colony size was measured as the length and width of the colony to the nearest 5 cm and colony condition was estimated as per cent of live tissue to the nearest 5%. One sample (1 cm × 1 cm) was taken from the edge of a lobe on each colony using a hammer and chisel. Each sample was split into two subsamples and placed in a labelled ziplock bag (only one per colony was used for analysis). On returning to shore each sample was preserved in 80% alcohol and stored at 4 °C prior to extraction.

dna extraction

DNA was extracted using the DNeasy kit (Qiagen, Qiagen House, Flemingway, Crawley, West Sussex). Approximately 0·5 cm2 of tissue was scraped off each sample with a sterile razor blade and placed in a 1·5 mL microcentrifuge tube. Extraction was performed overnight at 55 °C following the manufacturer's instructions. DNA was quantified using Nanodrop 3·0·0 spectrophotometer. DNA concentrations ranged from 10 to 45 ng µL−1.

microsatellite scoring

Six of the polymorphic microsatellite loci recently described by Severance, Szmant & Karl (2004) were used in the following protocol (Table 1).

Table 1. . Characteristics of microsatellite markers for Montastraea spp. (see Severance et al. 2004)
MultiplexLocusPrimer sequence (5′–′3′)MgCl2 (mm)Annealing temp. (°C)Product size (bp)

Two 10 µL multiplex polymerase chain reactions (PCR) were performed per sample to assay a total of six microsatellite loci (M-I and M-II). M-I consisted of 0·06 µm each of primer pairs maMS2-5, maMS11, maMS12 and 0·05 µm of primer pair maMS2-8, 1 × PCR Reaction Buffer (Promega, Southampton, UK), 2·5 mm MgCl2, 0·2 mm dNTPs, 2 U Taq DNA Polymerase (Storage Buffer B, Promega) and 4·6 µL H2O. M-II consisted of 0·06 µm each of primer pairs maMS2-4 and maMS8, 1 × PCR Reaction Buffer (Promega), 2·5 mm MgCl2, 0·2 mm dNTPs, 2 U Taq DNA Polymerase (Storage Buffer B, Promega) and 4·8 µL H2O. 2 µL of DNA was added to each reaction to a total volume of 10 µL. Thermal cycling was carried out with MJ Research (590 Lincoln Street, Waltham, USA) PT200 or PT100 cyclers. M-I cycling conditions consisted of an initial denaturation step at 95 °C for 2 min followed by 35 cycles of 95 °C for 1 min, 55 °C for 1 min, 72 °C for 2 min and a final step at 72 °C for 7 min. M-II cycling conditions consisted of an initial denaturation step at 95 °C for 2 min followed by 35 cycles of 95 °C for 1 min, 50 °C for 1 min, 72 °C for 2 min and a final step at 72 °C for 7 min.

PCR products were visualized using an ABI 3730 (Applied Biosystems) automated DNA sequencer with an internal size standard (Gene Scan 500-Liz, Applied Biosystems) for accurate sizing. Electropherograms were analysed using GeneMapper Software 3·0 (Applied Biosystems) and alleles were scored based on amplicon size. Owing to inconsistent scoring only four of the six microsatellites were used in the following analysis.



Of the 146 samples collected, 137 were successfully genotyped. Samples that had identical alleles at all four loci were identified as clonemates belonging to the same genet. Identical multilocus genotypes were never shared between sites, only within sites. The probability of identity (PID) was calculated to give a conservative estimate of the probability that two individuals sampled from the same population share a multilocus genotype by chance, not by descent (Waits, Luikart & Taberlet 2001). Biased and unbiased PID was calculated for each locus by GIMLET (Valiere 2002) and then multiplied across loci to give the combined PID (Table 4) (Waits et al. 2001). Due to the low probability of misidentifying colonies as clonemates when they are not, each distinct multilocus genotype was only included once in the data set in the following population statistical analyses (Baums, Miller & Hellberg 2005). Samples were tested for deviations from the expectations of Hardy–Weinberg equilibrium and the presence of heterozygote deficiencies and excesses were estimated for each locus within each population using Genepop ( Estimations of linkage disequilibrium between loci and calculations of the number of alleles per locus were conducted using FSTAT (Goudet 1995). Micro-checker (Van Oosterhout et al. 2004) was used to test for the presence of null alleles. Tests of Linkage Disequilibrium (data not shown) and deviations from Hardy–Weinberg equilibrium (Table 4) were not significant following Bonferroni corrections (test P < 0·003). Further tests failed to reveal null alleles for any of the four loci.

Table 4.  Characteristics of Montastraea annularis microsatellite markers for three sites in Honduras. Given are the number of samples per site (N), the number of observed heterozygotes (HO), the number of expected heterozygotes (HE) and the number of alleles (A) per locus and site. The presence of heterozygote deficits (HD) and heterozygote excess (HE), and their associated P-value, for each locus at each site was estimated using Genepop ( The probability of identity (PID) was calculated using GIMLET (Valiere 2002). Only unique multilocus genotypes were included in the analysis
LocusCharacteristicSBSQWWPID BiasedPID Unbiased
  1. SB, Sandy Bay; SQ, Seaquest; WW, Western Wall.

HD P-value 0·818 0·212 0·364  
HD SE 0·026 0·026 0·023  
HE P-value 0·205 0·812 0·672  
HE SE 0·027 0·023 0·030  
HD P-value 0·521 0·111 0·225  
HD SE 0·031 0·016 0·022  
HE P-value 0·518 0·846 0·851  
HE SE 0·032 0·017 0·019  
A1013 8  
HD P-value 0·575 0·726 0·838  
HD SE 0·018 0·033 0·011  
HE P-value 0·468 0·215 0·187  
HE SE 0·019 0·030 0·013  
maMS8HE 710 20·7030·698
HO 5 9 2  
A 4 5 3  
HD P-value 0·130 0·196 1  
HD SE 0·007 0·013< 0·001  
HE P-value 0·984 0·941 0·982  
HE SE 0·003 0·007 0·002  
Combined PID    8·7 × 10−66·6 × 10−6

genotypic diversity

Genotypic richness, normalized to sample size, was calculated as Ng/N (Coffroth & Lasker 1998), where Ng is the number of unique genotypes (genets) and N is the number of colonies genotyped. Genotypic richness equals one when all colonies in a population are unique and approaches zero when a population is dominated by a single genet. Genotypic evenness was calculated as Go/Ng (Coffroth & Lasker 1998), where Go is the observed genotypic diversity. Go was calculated as


where pi is the frequency of the ith genotype in the population (Stoddart & Taylor 1988). Genotypic evenness equals zero in a population dominated by a single genet and one where each genet is represented by an equal number of individuals.

spatial distribution of colonies

The spatial distribution of colonies at each site was mapped on to polar plots using the radial sampling coordinates. XY distances were then calculated for each colony and the pairwise distances between clonemates and non-clonemates were calculated. To discriminate the mechanism by which potential clonemates arose, we assumed that storm-induced colony fragmentation must have occurred when the separation of clonemates exceeded that of normal adult colony size. If the sum of colony widths and their separation was less than the width of a normal adult colony (52 cm wide ± 0·025 cm; based on the average size of colonies in the three populations) it was not possible to discount origins of partial-colony mortality (though severe colony erosion to the colony base only occurs rarely, Mumby pers. obs.).

size distribution

The size (area, cm2; calculated as colony length multiplied by colony width) distribution of clonemates vs. non-clonemates was analysed across sites and within sites using one-way analysis of variance. Data were normally distributed (Anderson–Darling test, P > 0·05) with homogeneous variances (Levene's test, P > 0·05).


genotypic diversity

A total of 137 colonies were successfully genotyped from the three sites identifying 114 individual genets. Over 90% of genets were represented by a single colony (Fig. 1). Small genets of two to three colonies comprised an additional 8% of the overall sample and one genet, at Western Wall, was composed of 14 colonies (Fig. 1).

Figure 1.

Polar plots of Montastraea annularis populations at: (a) Sandy Bay, (b) Seaquest and (c) Western Wall. Each mark represents a colony. Genets represented by a single individual are indicated in red. Individuals of the same genet are indicated by the same colour. Size classes are denoted by shape. Radial axis represents distance in m; angular axis represents angle in degrees. Number of colonies is 42, 52 and 43, respectively. All Montastraea annularis colonies present in the circle were sampled.

The density of colonies at each site was similar with 48, 53 and 45 colonies/78·5 m2 per site at Sandy Bay, Seaquest and Western Wall, respectively. However, the amount of clonal replication within populations differed significantly between the three sites (anova F = 18·33, P = 0·003), with the population at Western Wall having a higher degree of asexual recruitment compared with the populations at Sandy Bay and Seaquest (Tables 2 and 3). The index of genotypic evenness (Go/Ng) approached a value of 1 for both Sandy Bay and Seaquest indicating that colonies were evenly distributed among genets (Table 2). At Western Wall, however, this index was less than 0·30, indicating that one or more genets were represented by a large number of colonies, which implies a higher degree of asexual recruitment within the population (Table 2). Genotypic richness was almost 1 at Sandy Bay and Seaquest (0·88 and 0·92, respectively), whereas richness fell to 0·67 at Western Wall indicating fewer colonies with unique genotypes (Table 2).

Table 2.  Genotypic diversity summary of Montastraea annularis colonies sampled at three sites in Honduras. Total area sampled at each reef always equals 78·5 m2
RegionReef nameNo. ColCol DensGenet DensNNgNg/NGoGo/Ng
  1. No. Col, number of colonies within sampling plot; Col Dens, number of colonies m−2; Genet Dens, number of genets m−2; N, number of colonies genotyped; Ng, number of unique genotypes (genets); Ng/N, genotypic richness; Go, observed genotypic diversity; Go/Ng, genotypic evenness.

HondurasSandy Bay 480·610·47 42 370·8833·920·92
HondurasSeaquest 530·680·61 52 480·9243·610·91
HondurasWestern Wall 450·570·37 43 290·67 8·150·28
Total3 reefs146  137114   
Average  48·70·620·48 45·7 380·8228·60·70
SD   4·040·060·12  5·51  9·540·1318·30·37
Table 3.  Clonal structure summary of Montastraea annularis colonies sampled at three sites in Honduras. Genet size represents number of individual colonies per genet
Genet sizeFrequency of genet size
Sandy BaySeaquestWestern Wall
  1. Min, minimum value; max, maximum value; SE, standard error.

14 0 0 1
3 0 1 0
2 5 2 1
Mean no. of colonies per genet 1·14 1·08 1·48
Distance between clonemates (m)
 Mean ± SE 1·01 ± 0·61 3·30 ± 1·16 2·01 ± 0·10
 Min 0·15 0·17 0·27
 Max 3·38 6·94 4·60

distance between clonemates

The distance between clonemates ranged from a minimum of 0·15 m to a maximum of 6·94 m (Table 3) and did not differ among sites (Mood's χ2 = 2·04, P = 0·360). Conservatively, we estimate that at least seven (70%) of the clonal replication events involved breakage of the colony and dispersal of fragments. The large genet at Western Wall was likely formed through the splitting of a single colony (Fig. 1c) but several of the daughter colonies may have split further by partial mortality.

size distribution

No differences were found between the colony size of clonemates and non-clonemates at Seaquest and Sandy Bay (anovaF = 0·78, P = 0·381, for pooled data, though the same result was obtained for individual sites). Clonemates were significantly smaller than non-clonemates at Western Wall (0·37 ± 0·23 m2 and 0·74 ± 0·25 m2, respectively; anova F = 5·40, P = 0·025). This result was caused by the large (14 colony) genet at this site as no differences occurred between clonemates and non-clonemates of the remaining colonies (anova F = 2·27, P = 0·143).


The recent isolation and development of polymorphic microsatellite loci for Montastraea annularis (Severance et al. 2004) enabled us to quantify the incidence of clones in this long-lived coral for the first time. All three sites exhibited clonal replication with 8% of genets comprising two to three colonies and a single genet consisting of 14 colonies. While sexual reproduction appears to be the predominant mode of reproduction in M. annularis, this massive coral can propagate asexually in a manner consistent with colony breakage during storms.

At least 70% of multicolony genets observed must have been caused by physical breakage and dispersal, providing the first genetic evidence of this process in massive corals. This mechanism of clonemate creation is likely to involve considerable trauma and potentially generate damaged areas of tissue that could have elevated susceptibility to disease. If the formation of clonemates was associated with a decrease in growth rate due to the metabolic expense of repairing the trauma, the average size of clones would be expected to be less than that of non-clonemates. However, as the size distributions of clonemates and non-clonemates were generally indistinguishable (with the exception of the large genet at Western Wall), our data suggest that clones can still attain normal adult size through asexual reproduction. It will be interesting, in future, to compare the size-based survival of asexually produced colonies (ramets) to those generated by sexual reproduction and recruitment.

The apparent scarcity of sexual recruits, combined with their robust morphology and high longevity, have led to speculation that massive corals may employ storage effects (Edmunds 2000), whereby the persistence of a small group of adults maintains the population when recruitment fails (Murphy 1968; Warner & Chesson 1985). Low adult mortality allows strong year classes to persist through time until a favourable recruitment period occurs (Warner & Chesson 1985). However, the outcome of employing storage effects may become increasingly questionable given environmental and ecological changes occurring throughout the Caribbean. Coral larvae are unable to settle on macroalgae (Diaz-Pulido & McCook 2004) and the abundance of such plants is increasing markedly in many parts of the region (Gardner et al. 2003) due to a region-wide decrease in grazing levels (Lessios 1988; Hughes 1994). Coral mortality rates have increased, largely through outbreaks of disease (Aronson & Precht 2001) and mass coral bleaching (McField 1999). Therefore, favourable conditions for the recruitment of sexual propagules are likely to have declined in the Caribbean and may become even less frequent given anticipated increases in the incidence of mass coral bleaching as the oceans continue to warm (Hoegh-Guldberg 2004). Importantly, reductions in the success of sexual reproduction in massive corals may result in a shift in the importance of dispersal methods towards asexual mechanisms. The dispersal of colonies through asexual fragmentation confers several advantages to the population, including: (1) daughter colonies are substantially larger than sexual recruits and, being elevated above the substrate away from macroalgae, may increase their chances of survival, and (2) daughter colonies do not require macroalgal-free areas in order to settle (i.e. a coral fragment can fall on established macroalgae and much of the upper coral tissue can continue to grow without macroalgal contact – PJM pers. obs.).

We have shown that asexual reproduction can occur within populations of the massive coral M. annularis, but the limited scale of this study prevents us understanding the overall importance of this process in different environments. Corals were sampled within 10 m plots largely because we did not expect clonemates to be distributed over more than a few metres (if at all). However, the spread of some genets was relatively large and it is likely that several genets extended beyond plot boundaries. Our observations of the spread of colony fragments are therefore conservative. Nevertheless, M. annularis exhibits an unexpected clonal structure. Intriguingly, the large, 14-colony genet was found at the site with the greatest wave exposure. We hypothesize, therefore, that the importance of asexual reproduction will be positively correlated to the incidence of physical disturbance such as hurricanes on Caribbean reefs. It is interesting to note, however, that Caribbean-wide variation in clonal structure of the branching coral Acropora palmata is not correlated to the incidence of hurricanes (Baums et al. 2006) but rather is correlated with habitat factors. The effects of acute disturbance upon massive corals may be as complex and only significant under particularly stressful conditions. Given current changes in climatic conditions and unprecedented anthropogenic disturbance to reef ecosystems that may reduce the success of sexual reproduction, it is now necessary to revisit the role of asexual dispersal in massive corals, particularly across taxa and gradients of environmental disturbance.


We thank the Royal Society and Natural Environment Research Council for funding (grant NER/A/S/2001/01127 to PJM and a PhD studentship to NLF). Financial support for IB came from the NOAA-Fisheries Coral Reef Initiative through the Cooperative Institute for Marine and Atmospheric Studies (CIMAS). Thanks to Emily Severance and Chad McNutt for sharing microsatellite protocols and to Lysa Johnston for great work in the laboratory. We also thank Steve Box for his generous assistance in the field.