COLONY SIZE-FREQUENCY DISTRIBUTIONS AMONG DIFFERENT POPULATIONS OF THE SCLERACTINAN CORAL SIDERASTREA STELLATA IN SOUTHWESTERN ATLANTIC : IMPLICATIONS FOR LIFE HISTORY PATTERNS *

Colony size-frequency distributions of reef corals may be used to infer growth potential and population responses upon environmental changes. The present paper compares the size structure of colonies of Siderastrea stellata Verrill, 1868, among 11 sites, six of them distributed along a gradient of sediment deposition in Abrolhos, Bahia, Brazil (18 S). Results indicated that the population structure is likely to be influenced by local conditions, rather than large scale factors, such as latitude. The 11 distributions, however, showed higher frequencies of small size classes. Class 1 (up to 2.5 cm diameter) was always present and the frequency of colonies from size class 3 (10 cm diameter) tended to decrease in all sites. Comparison among the six Abrolhos sites showed that S. stellata has advantages at sites with intermediate sedimentation, where colonies attain larger sizes, probably, reflecting a higher survivorship over time. The present study showed that, despite the influence of environmental conditions on parameters of the populations such as size of colony, the life history strategy of S. stellata reflects a local adaptation that allows its development and survivorship in shallow waters and horizontal substrates, sites characterized by high mortality rates.


INTRODUCTION
Different coral life history strategies derive from a balance among growth, survival and reproduction (Bak & Engel, 1979;Santangelo et al., 2003).These strategies are associated with population parameters, such as recovery and turn-over capacity, mortality, survivorship and intrinsic growth rates __________ (*) Paper presented at the 1 st Brazilian Congress of Marine Biology, on 15-19 May.Rio de Janeiro, Brazil.(Harvell & Grosberg, 1988;Goffredo & Chadwick-Furman, 2003).Coral life-history processes are strongly related to size, and size-frequency distribution of corals may be used as a simple and non-destructive method to analyze life history and dynamics of coral populations (Bak & Meesters, 1998;Meesters et al., 2001).
The inference of a dynamic process from a size-frequency distribution is made possible when other aspects of the species biology, such as the relationship between size, growth and reproduction, are known.In the case of a modular organism, such as a colonial scleractinian coral, partial mortality, fragmentation and fusion of colonies may disrupt the relationship between life processes and size.However, in massive corals, most partial mortality occurs as very small lesions that are regenerated within a few days (Bak & Meesters, 1998), and careful observation in situ to detect fusion and/or fragmentation, using knowledge of growth patterns and coral surface structures, can overcome the problem.
During the present study, size structure of the massive scleractinian coral Siderastrea stellata Verrill, 1868, was analyzed for 11 populations, along a latitudinal gradient of 20 o in the Southwestern Tropical Atlantic Ocean, comprising areas throughout almost the whole geographical distribution of the species.Growth rate and fecundity data of different size classes of colonies of S. stellata were used to discuss life history strategies based on size-frequency distributions.Six of the populations studied were distributed along a gradient of sediment deposition in the Abrolhos Reef Complex, Bahia (18 o S).The present study presents an analysis of the potential impact of this factor upon S. stellata populations based on size-frequency distributions obtained for each site.

MATERIALS AND METHODS
The Species Siderastrea stellata is an endemic and a common scleractinian coral species from Brazil (Laborel, 1970;Castro & Pires, 2001).It is a zooxanthellate, colonial and massive coral, occurring in all the Brazilian reef regions and in coral communities from Maranhão (00°53'S, 044°16'W) to Rio de Janeiro State (23°S, 042°W) (Castro & Pires, 2001) (Fig. 1).In some reef areas, such as in the Atol das Rocas (03 o 52`S, 033 o 49`W), it is the main reef building organism (Echeverría et al., 1997).It usually occurs in shallow waters (up to 10 m depth) and often occupies horizontal substrates (Segal & Castro, 2000).This species is considered relatively resistant to sedimentation, temperature and salinity variations and strong wave action (Hartt in Laborel, 1970;Chevalier, 1987).Siderastrea stellata is a gonochoric brooder, with a high female to male sex ratio and an annual reproductive cycle.Planulation occurs during January/February (Lins de Barros et al., 2003).

METHOD
At each site, the closest colony, from 0.5 m to 0.5 m or 1 m to 1 m, of two to three belt transects of 20 m each were measured, depending on the number of colonies in the area and the reef morphology.For each colony we measured the largest and the smallest diameters, and total surface area (size) was calculated using the formula for an halve spherical segment (A = 2 X [(D/2 X d/2) X π], where: A = area (cm 2 ), D = largest diameter (cm), d = smallest diameter (cm), π = 3.1419).Up to 80 colonies were measured at each site; however, in some sites, the total number of colonies in the area was less than 80.In "Guaratibas" and "Coroa Vermelha" only 73 and 58 colonies were measured, respectively.In the six Abrolhos Reef Complex sites, the presence of bioeroders, identified by holes on colonies surfaces, was quantified as present or absent for each measured colony.

Analyses of Data
The size structure of S. stellata populations was analyzed for each of the 11 sites by calculating the size-frequency distribution parameters: mean colony size and standard deviation, coefficient of variation (=standard deviation/mean), the largest colony size, maximum colony size (95 th percentile), commonest colony size and skewness (g 1 ).
Colony size data were log transformed to become approximately normal, tested by Kruskal-Wallis test.Also, we followed Bak & Meesters (1998) and Meesters et al. (2001), who argued that "living coral colony surface, which can be modelled as e.g.semi-spheres, increases as a power function in size" and that "population dynamic are first order processes, i.e. modelled by exponential functions".Transformed data also reduced heteroscedasticity, tested by Cochran test.Mean colony sizes were analyzed by ANOVA and Student-Newman-Keuls for a posteriori comparison.
Colonies were grouped into 21 size classes of 1.40 cm 2 each, with the exception of class 1, which included colonies with up to 0.70 cm 2 , unfertile or with an extremely low fecundity (Lins-de-Barros & Pires, in press).Size-frequency distributions were compared to each other using a 2-sample Kolmogorov-Smirnov test.For statistical analyses we used the program "Statistica 6.0", with p < 0.05.Life-history patterns of S. stellata in each site were inferred from size-frequency distributions, using data on growth rate and fecundity in each size class, obtained for colonies from Canudos reef (Linsde-Barros & Pires, in press).For the six sites at the Abrolhos Reef Complex, size-frequency distribution parameters were also compared to the frequency of colonies infested by bioeroders (collected during the present study), sediment deposition rates and species cover (extracted from Segal-Ramos, 2003) and, family recruitment rate (extracted from Alvarenga, 2004), using Pearson correlation analyses.

Size-Frequency Distributions
Colony size-frequency distribution of S. stellata varied among the 11 study sites (Figs 2 A-K).At some sites, a broader representation of size classes was observed, as at "Viçosa" and "Guaratibas" (Abrolhos), where eight and 12 classes were observed respectively, and at all other sites outside the Abrolhos Reef Complex, where more than 15 classes were observed, except at "Barra" (Tamandaré) (Fig. 2).The sites "Barra" (Tamandaré), "Leste" and "Coroa Vermelha" (Abrolhos) did not have any colony larger than size class 4 (Figs 2 C, F-K), while "Fernando de Noronha" sites and "Meio" (Tamandaré) had colonies up to class 17 (Figs 2 A, B, D).Size class 21 colony was only observed in "Canudos" (Fig. 2 E).Apart from the differences in size-frequency distributions, all sites had a higher percentage of the smaller size classes (classes 1 to 4).Size class 2 was the most frequent at almost all sites (Fig. 2), except at "Coroa Vermelha", where size class 1 was the most frequent (Fig. 2 H) and Inferno, where size class 3 was the most frequent (Fig. 2 B).
After log transformation, the 11 sizefrequency distributions were statistically not significantly different from a normal distribution (K-S; p > 0.20).Kolmogorov-Smirnov analyses between pairs of size-frequency distributions showed that for the 54 pairs tested, 13 (24.07%)were not significantly different (K-S; p > 0.05).However, size-frequency distribution similarities and differences were not related to the geographic location of sites or to latitude.Distant sites showed similarities between their size-frequency distributions, while adjacent sites showed differences.
"Sebastião Gomes" and "Coroa Vermelha" (Abrolhos), sites that are close to each other, showed similar size-frequency distributions (K-S; p > 0.05), but they were significantly different from all other sites, including the other sites from Abrolhos ("Leste", "Lixa", "Guaratibas" and "Viçosa").These four sites were similar to each other, but also showed similarity with the "Barra" site (08 o S; 035 o W), which is located 2000 km away.However, the size-frequency distribution obtained for "Barra" was significantly different from those obtained for its closest site "Meio" (08 o S; 035 o W), which, instead, was similar to "Canudos" (16 o S; 039 o W).The only geographic pattern was observed at Noronha, where both sites, "Inferno" and "Sancho", were similar to each other (K-S; p > 0.05) and significantly different from any other site (K-S; p < 0.05).
The coefficient of variation was inversely related to mean colony size (r Pearson; r = 0.911; r 2 = 0.830; p = 9.520E-05) and directly related to skewness (r Pearson; r = 0.633; r 2 = 0.4008; p = 0.0365).This means that, in populations dominated by large colonies, the frequency of small colonies and the size variation are relatively small.Maximum Colony Size ("95 th percentile") and Largest Colony Size Maximum log colony size of S. stellata colonies averaged 2.14 cm 2 (± 0.47[s.d]),ranging from 1.52 cm 2 (or 20 cm in diameter) at "Coroa Vermelha" (Abrolhos) to 3.25cm 2 (or 44 cm in diameter) at "Canudos".Considering the growth rate defined for different size classes of colonies of S. stellata (0.5 cm in diameter per year) (Lins-de-Barros & Pires, in press), the oldest colony at "Coroa Vermelha" would be at least 10 years old, while at "Canudos" it would be approximately 88 years old.
Skewness (g 1 ) After log transformation, all size-frequency distributions showed a higher frequency of colonies above the mean colony size, as indicated by negative skewness values, excepting at "Viçosa" and "Sebastião Gomes" (Table 1).At "Viçosa", the skewness value was close to zero (g 1 = 0.019), indicating that most colonies present were approximately equal to or similar to the mean colony size of this population (1.44 ± 0.24[s.d]).At "Sebastião Gomes", the high positive skewness (g 1 = 0.224) indicates a high frequency of small colonies, below the mean colony size for this population (1.16 ± 0.28[s.d]).
Our results indicated that, in the six Abrolhos sites, mean sediment deposition rates were not directly related to the percentage of infested colonies (r Pearson; r = -0.076;r 2 = 0.0057; p = 0.887), mean S. stellata cover (r Pearson; r = -0.337;r 2 = 0.114; p = 0.513), Siderastreidae recruitment rates (r Pearson; r = 0.127; r 2 = 0.016; p = 0.811) or mean colony size (r Pearson; r = 0.377; r 2 = 0.142; p = 0.461).However, a strong positive relationship was observed between this variable and skewness (r Pearson; r = 0.912; r 2 = 0.832; p = 0.011).Thus, at sites with a high sediment deposition rate, a higher frequency of colonies smaller than the mean size of the population would be expected.

Frequency of Colonies Infested by Bioeroders
A strong positive relationship between the frequency of colonies infested by bioeroders and colony size was observed (Table 2) (r Pearson; r = 0.894; r 2 = 0.799; p = 0.0162) in the six Abrolhos sites studied.The largest frequencies of colonies infested were observed in "Viçosa", site that had the largest mean colony sizes (Table 1).
Siderastrea recruitment rate did not show a linear relationship with mean S. stellata cover (r Pearson; r = -0.426;r 2 = 0.181; p = 0.40), mean colony size (r Pearson; r = -0.366;r 2 = 0.134; p = 0.476) or the smallest colony size (r Pearson; r = 0.063; r 2 = 0.0039; p = 0.906).These results may be an indication that none of these factors seem to be strongly related to S. stellata recruitment in the Abrolhos sites studied.
Table 2: Frequency of S. stellata colonies infested by bioeroder organisms in six Abrolhos sites, distributed along a sediment depositional gradient (extracted from Segal-Ramos, 2003).
Legend: Inferior and superior limits -smallest and largest log colony size (cm 2 ), comprising the size range of each size class; G = Guaratibas, PL = Pedra de Leste, PX = Pedra Lixa, SG = Sebastião Gomes, CV = Coroa Vermelha, V = Viçosa; --= presence of colonies of the size class but none infected; 0 = absence of colonies of the size class; ratio = number of colonies infected/number of colonies measured of the size class; % colonies infected = (number of colonies infected/N) X 100.

DISCUSSION
Although the commonest size class of colonies was similar among the 11 sites (class 2), the size-frequency distributions of colonies of S. stellata varied among the sites (Fig. 2).However, such variation did not show any regional or latitudinal pattern.Some sites that were located near to each other had different size frequency distributions (i.e.: "Barra" and "Meio", both at Tamandaré), while similar size frequency distributions were observed between some sites (i.e.: "Meio" and "Canudos").The same result was observed in relation to the mean size of colonies (Table 1).These results indicated that the population structure is likely to be influenced by local environmental conditions, rather than being determined by large scale factors, such as latitude.
The largest colony sizes were observed at the two sites at Noronha, and these sites had the highest frequencies of colonies larger than the mean (g 1 Sancho = -0.616;g 1 Inferno = -0.505(Table 1).The relationship between mean colony size and skewness (g 1 ) was also documented for other coral species by Bak & Meesters (1998), Meesters et al. (2001) and Langmead & Sheppard (2004).
At the Noronha sites, the maximum colony size ("95 th percentile") was larger than in the others sites and the smallest ones were observed at the six sites at Abrolhos and at "Barra", Tamandaré (Table 1).Rinkevich & Loya (1987) and Kramasky-Winter & Loya (1998), concluded that shallow habitats would be exposed to more frequent disturbances, such as salinity and temperature variations, wave action and dessication, which are likely to decrease the life-span of organisms that inhabit such environments.Noronha sites were the deepest, being around 7 m, while the depths of the others sites were above 4 meters.The larger colony sizes observed at the Noronha sites are probably the result of greater colony longevity and/or less partial mortality (Meesters et al., 2001).The Abrolhos Reef Complex (18 o S) is considered to be the largest and most diverse reef area of the South Atlantic Ocean (Castro & Pires, 2001).However, Segal-Ramos (2003) reported extremely high sediment deposition rates at this region, considered by some authors as being catastrophic for reef development (Pastorak & Bilyard, 1985apud Brown, 1997).More recently, other studies have registered the occurrence of rich coral communities in turbid waters around the world (Anthony & Larcombe, 2000).
Sedimentation affects all life phases of corals, and its influence seems to be related to colony size.Recruits and small coral colonies may be easily buried by sediment (Bak & Engel, 1979;Babcock & Smith, 2000;Miller et al., 2000), and only a few corals manage to survive burial for a long time period.In contrast, large colonies less likely to be killed by sediment may be reduced in size by partial colony mortality (Bak & Meesters, 1998;Goffredo & Chadwick-Furman, 2003).By affecting the size classes differentially (Dodge & Vaisnys, 1977;Langmead & Sheppard, 2004), sedimentation changes the size structure of populations, affecting their maintenance and recovery potentials (Dodge & Vaisnys, op.cit;Fabricius, 2005).
Survivorship in coral populations exposed to sedimentation, depends on their capacity to become adapted to different conditions, and the intensity and duration of the stress (Anthony & Fabricius, 2000;Fabricius, 2005).Analysis of the responses of the populations of the main reef building corals among areas characterized by different sedimentation patterns provides important insights into understanding how this factor may affect reef coral population dynamics.The apparent adaptation of the Brazilian coral species to high rates of sediment deposition, as proposed by Leão et al. (1988), provides an interesting opportunity to study the relation between corals and long term sedimentation.
During the present study, no linear relationship was apparent between sediment deposition rate and colony size (mean and maximum), mean cover, or number of recruits per square meter.These results are similar to those obtained by Segal-Ramos (2003), who also did not detect a direct relationship between sediment and these aspects of coral community structure.However, Segal-Ramos (2003) suggested an indirect effect of the deposition of sediment upon coral communities.In the present study we also considered the hypothesis that sedimentation may be causing some influence upon the life history of S. stellata, as indicated by the size structure of colonies at the six sites of Abrolhos.
The only significative linear correlation was a strong positive correlation between sediment deposition rate and skewness (g 1 ) (Pearson r; r = 0.912; r 2 = 0.832; p = 0.011).This result suggests that at high sedimentation conditions, a high frequency of colonies smaller than mean would be present, i.e. high and positive skewness values.Gúzman & Holst (1993) analyzed the size structure of S. radians populations affected by oil pollution and also found a higher frequency of small colonies.However, our results contrasted with those by Meesters et al. (2001), who observed negative skewness (i.e. higher frequency of large colonies) at sites affected by sedimentation.These results indicated that different population structures may arise depending on the species and the impact factor analyzed.
The "Sebastião Gomes" site had the highest frequency of small colonies (the most positive skewness value -Table 1), which may be explained by a high recruitment rate.Considering that S. stellata has an annual gametogenic cycle, releasing larvae by the end of January-early February, the critical period for larvae recruitment would be February-March (Lins de Barros et al., 2003).Despite the high mean sediment deposition rate observed at this site, during this critical recruitment period (February-March), sedimentation is not at its highest level (Segal-Ramos, 2003), hence it may not be a significantly limiting recruitment.However, the number of recruits per square meter observed at "Sebastião Gomes" was not particularly high (Table 1), hence recruitment alone does not explain why this site had the highest frequency of small colonies.Reis & Leão (2000) studied S. stellata colonies from the North of Bahia State, and observed a growth rate similar to that estimated for this species at a Southern reef in Bahia State (Lins-de-Barros & Pires, in press).However, the highest sediment deposition rates registered at "Sebastião Gomes" were very high and could be causing effects on the species biology, such as reducing growth.Riegl & Branch (1995) observed that an experimental increase in sedimentation decreased the ratio of productivity/respiration and increased the mucus production of many coral species.Sedimentation can also cause the initiation of cleaning mechanisms by ciliary action, tentacular movement and polyp expansion (Sanders & Baron-Szabo, 2005).These events represent expenditure of energy that in other conditions could be used for growth (Dodge & Vaisnys, 1977;Riegl & Branch, 1995;Babcock & Smith, 2000;Nughes & Roberts, 2003a).
Finally, the high frequency of small S. stellata colonies at "Sebastião Gomes" may be reflecting some limitation to growth, perhaps by competition with another organism that gains advantage in high sediment habitats.An example for the Brazilian reefs is the zoanthid Palythoa caribaeorum, which has the capacity to incorporate sediment into its tissue during tissue growth (Haywick & Mueller, 1997).Segal-Ramos (2003) did not detect a linear relationship between mean sediment deposition rate and Palythoa cover, but observed a higher cover of this organism at "Sebastião Gomes", the site with the highest mean sediment deposition rate.This author also observed a negative correlation between Palythoa and Scleractinia cover.Following Bastidas & Bone (1996), in their study from Venezuela, P. caribaeorum presents a higher competitive capacity than the other corals, such as S. siderea and Agaricia agaricites.At the sites studied in this project, Palythoa occupies the same habitat as S. stellata, on shallow and horizontal substrates.Therefore, these organisms could be potential competitors.
"Coroa Vermelha" had a skewness value close to zero (g 1 = -0.045-Table 1), a high frequency of size class 1 colonies (Fig. 2 C), and the smallest size of colonies (mean and maximum) (Table 1).Colony size-frequency distribution obtained for this site was similar to that obtained for "Sebastião Gomes" (K-S; p > 0.05), but different from the other sites.The small size of colonies at "Coroa Vermelha" could be explained by a high recruitment rate.The highest number of recruits per square meter was registered at this site (Table 1 -extracted from Alvarenga, 2003).However, despite increased recruitment, colonies at this site, apparently, did not attain large sizes.The mean sediment deposition rate documented at "Coroa Vermelha" was similar to that of "Guaratibas" and "Viçosa" (Table 1).However, Segal-Ramos (2003) registered the highest water turbidity at "Coroa Vermelha".So, the high degree of sediment in the water column could be causing the death of recruits by abrasion or displacement.It is possible that while the "Coroa Vermelha" conditions are good to recruitment, they may not be good for the adult population (see Babcock & Mundy, 1996).The low coral cover observed at "Coroa Vermelha" (Table 1) could be indicative of mortality during postrecruitment phases.
At "Coroa Vermelha", the small colony size and high frequency of size class 1 colonies could be the result of high recruitment, recruit mortality and partial mortality of large colonies.These results indicated that similar distributions, such as those from "Sebastião Gomes" and "Coroa Vermelha", could reflect different environmental conditions and selective factors acting upon the community.
Sediment deposition rates varied among "Guaratibas", "Lixa" and "Leste"."Lixa" and "Leste" presented the lowest rates.The low frequency of small colonies at these sites could be the result of low recruitment rates, which were low indeed (Table 1).Despite the low recruitment rate, the recruit survivorship rate appears to be high, which can be indicated by the high S. stellata cover at "Lixa" (Table 1).However, if "Leste" and "Lixa" presented high frequency of large colonies, they did not present the largest colony sizes, indicating a possible effect of partial mortality events.In the present study, the frequency of colonies infested by bioeroders increased with colony size.A hypothesis could be that, at "Lixa" and "Leste", colonies did not attain large sizes because of partial mortality events caused, for example, by the borers infestation.We may conclude that the populational structure of S. stellata at "Lixa" and "Leste" could be the result of a balance among low recruitment, recruit survivorship and partial mortality of large colonies.
Size-frequency distribution parameters obtained for "Guaratibas" and "Viçosa" indicated that these sites present optimum environmental condition for S. stellata population development, being characterized, among the Abrolhos sites, by the largest mean colony sizes and the best size class representativeness (larger coefficient of variation)."Viçosa" also presented the lowest skewness (g 1 = 0.019 -Table 1), indicating similarity to a normal distribution, while at "Guaratibas" it was the most negative (g 1 = -0.388-Table 1), indicating a high frequency of colonies larger than the mean.This size structure difference could be explained by differences in recruitment rates, since at "Viçosa" it was nine times higher than at "Guaratibas" (Table 1).However, the presence of many size classes observed at these two sites may indicate a balance among recruitment, survivorship and mortality.
We concluded that sedimentation may be causing an indirect effect on the structure of S. stellata populations, acting synergistically with other factors and community parameters.In the case of the genus Siderastrea, its tolerance to sedimentation has already been documented for S. stellata (Laborel, 1970), S. siderea (Nughes & Robert, 2003b) and S. radians (Lirman et al., 2002).Our results suggested that S. stellata has advantages at sites with intermediate sedimentation.At these sites, S. stellata colonies attain larger sizes, probably, reflecting a higher survivorship over time.Also, the influence of sedimentation on life history of S. stellata in Abrolhos reefs is probably occurring mainly during post-recruitment phases.
The present study showed that, despite of the influence of environmental conditions on parameters of the populations such as size of colony, the life history strategy of S. stellata reflects a local adaptation that allows its development and survivorship in shallow waters and horizontal substrates, at sites characterized by high mortality rates.

Fig. 2 .
Fig. 2. Size-frequency distribution (log scale) of the Brazilian reef coral Siderastrea stellata at 11 sites along the species geographical distribution.

Table 1 :
Resume of Siderastrea stellata size data for 11 sites along the geographical distribution of the species and abiotic and biotic factors analyzed for the Abrolhos sites.