Morphological variation in the atlantic genus SideraStrea ( anthozoa , scleractinia )

Siderastrea is a small genus of scleractinian corals, composed by zooxanthellate massive colonial forms. Besides contributing to the consolidation of the reef structure, represent a group with significant resistance to environmental stress. Until the 70’s, its taxonomy was a complicated task, limited by the difficulty in identifying morphotypes. More recently, interspecific boundaries were redefined for ‘Atlantic Siderastrea Complex’. However, despite new perspectives, including the occurrence of S. radians and S. siderea for Brazil, morphological patterns are still poorly studied and inconspicuous for several coastal areas. In order to characterize levels of intraspecific and interspecific morphological variation in S. stellata and S. radians, morphometric analysis were carried out in colonies from Bahia State coast. Samples were taken in Todos-os-Santos Bay and North Coast. We applied the Canonical Discriminate Analysis upon six corallite characters to evaluate the different levels of variation. The tests revealed that diameter of corallites, columella depth and number of septa varied among populations. Number of septa was the most important character for the species differentiation. Siderastrea stellata, with three different morphological patterns, was more variable than S. radians. Key-Words: Systematics; Modular organism; Morphotypes; Zooxanthellate coral; South Atlantic.

According to Laborel (1974), given the notable variation and overlap of diagnostic characters, the taxonomy of Siderastrea is a complicated task, limited by difficulties in identifying morphotypes and in determining interspecific boundaries.Recent studies with a broad taxonomic approach have maintained the identity of these species based on the following main characteristics: Siderastraea siderea is an hermaphroditic spawning specie while Siderastrea radians and Siderastrea stellata are gonogoric brooder species (Szmant, 1986;Duerden, 1904;Neves & Silveira, 2003); S. radians and S. stellata from Brazil presented exclusive alleles for each specie in a population genetic studies (Neves et al., 2008); divergence among the three species was revealed in a phylogenetic study based on ITS sequence (Forsman et al., 2005); and morphological traits also indicate differences among species (Neves, 2004;Neves et al., 2010).However, morphological variability is still an unexplored issue.Considering the presence of many quantitative characteristics in Siderastrea diagnosis and its overlap, identify morphotypes is an important task.
The morphological similarity between S. stellata and S. radians, in particular, are currently considered an important problem because of its occurrence in sympatry in Brazilian coast and the inability to in situ identification (Neves et al., 2010).Werner (1996) synonymizes S. stellata and S. radians based on statistical analysis of morphology.Santos et al. (2004) also studied the Brazilian and Caribbean species and founded high morphological variability.However, both cited studies did not consider the occurrence of S. radians and S. siderea in Brazilian coast.Supposedly, the observed variability could be related to interference of other species on their morphometric results, as suggested by Neves et al. (2010).
According to Gattuso et al. (1991), the morphological variation in Scleractinia depends of the adaptation to environmental heterogeneity and the mecha-nisms of genetic control.The modular development of these individuals is one important task to consider in morphological analysis, however usually ignored (Hageman, 2003;Davidson et al., 2004;Eble, 2005).Best et al. (1983) propose a new dimension to the analysis in Scleractinia taxonomy when suggest that in fossil record and current species, within a geographically limited area, the intraspecific variation can be 'intracolonial', 'intrapopulation' and 'interpopulation'.Menezes et al. (2013) observe that intracolonial variation is not an important problem to differentiate S. stellata from S. radians.In the present study, we focus on the others categories of morphological variation, describing the intraspecific and interspecific variation of S. stellata and S. radians in Brazilian reef systems.

MATERIALS AND METHODS
The study was carried out in two distinct geographical sections of Bahia State (Brazilian northeastern littoral): the Todos os Santos Bay (TSB) and the North Shore (NS).The TSB is the second largest coastal bay in Brazil with an area of approximately 1,200 km 2 , comprising a total of 91 internal islands, with coral reefs and mangroves as typical ecosystems (Cirano & Lessa, 2007).Founded in 1999 by the State Government, the 'Todos os Santos Bay Environmental Protected Area' aims to manage social and economic activities around the coast region (800 km 2 ).Nevertheless, despite its ecological importance, the bay has been critically impacted due to coastal development activities (Celino & Queiroz, 2006;Amado-Filho et al., 2008;Felizzola et al., 2008).NS is characterized by discontinuous coral banks distributed in 17 km along the coast in very shallow waters (Kikuchi, 2000).The resources have also been explored and are under degradation caused by domestic and industrial sewage, oil pollution, urban development, and intensive tourism (Costa et al., 2000;Leão & Kikuchi, 2005).
Sampling was conducted in nine localities from 2006 to 2008 (Table 1): five around the TSB and four along the NS.Due to similarities between S. stellata and S. radians, field identification was not performed.In practice, this procedure prevents taxonomical misidentification, but is not favorable to homogenize sampling of sympatric species.Detailed information concerning total number of specimens per locality (n), georeferenced coordinates (lat/long), and additional sampling information are listed in Table 1, Fig. 1.
Samples with at most 10 cm were haphazardly collected at a minimum distance of two meters.This measure was adopted to avoid the removal of clones due to the typical phylopatric pattern of S. stellata and S. radians during larvae recruitment.Colonies with evidence of disease or physiological senescence (irregular pigmentation, blemishes, whitening) were not considered.Samples were obtained from tide puddles of 0-30 cm deep or by free or SCUBA diving during low tide (no deeper than 3 m).Colonies were removed us-ing hammer and ferrule, and placed in plastic bags for transportation.For corallum and corallite structure analysis, the colonies were submitted to a 2,0% sodium hypochlorite solution.Identifications followed Neves (2004).Morphometric analyses were carried out with the use of ocular micrometric on NIKON SMZ1000 stereomicroscope and MITUTOYO 0,01-150 mm digital pachymeter.Specimens were deposited in the Cnidaria Collection at the Zoology Museum of Universidade Federal da Bahia (Table 1).Only mature corallites were examined ('old polyps' sensu Soong & Lang, 1992), i.e., those with, at least, the third septal cycle well formed.Fifteen corallites for each colony were selected from three regions: top; middle; and edge.The top zone was defined as the upper surface of the colony distal from the base, a flattened or dome-like area 1 to 3 cm in diameter.The edge was defined as the marginal area in contact with the substratum, including three and/or four corallite upper lines (the first of which was frequently buried in the sand).The middle was defined as the large area between the inner limits of the two previous zones.For small colonies (aprox.3 cm), were considered only 3 corallites in each region.Based on previously published studies (Milne Edwards & Haime, 1849;Verrill, 1868;Vaughan, 1919;Vaughan & Wells, 1943;Laborel, 1967Laborel, , 1969/70;/70;Budd & Gúzman, 1994;Neves, 2004;Foster, 1977Foster, , 1979Foster, , 1980Foster, , 1985;;Budd, 1990), six morphological characters were selected: corallite diameter (based on the mean of two greater diameters -corD); collumela diameter (based on the mean of two greater diameters -colD); septal number (sepN); theca thickness (tecThick); columellar depth (depth); and the distance among corallites (corDist).Septal number is a count and not a measure variable as the other characteristics.However, this data shows assumptions postulated by the Central Limit Theorem, which include a large number of independent random variables with finite mean and variance, not Bimodal and Poisson distribution, and absence of zeros (Underwood, 1997).As such, we considered the septal number a continuous characteristic.
To evaluate morphological differences between species and among localities where colonies where sampled, we performed a Hierarchical Permanova Analysis based in a Euclidian distance matrix and 9999 permutations.It is a multivariate analysis that uses permutation to test differences among discrete groups.To identify the most important traits that determinate de morphological variation between species and among localities, and better understand Siderastrea morphotypes, we performed a Discriminant Canonical Analysis (DCA).It is a multivariate analysis that can be used to discriminate a set of independent variables that better describe a priori defined groups (Hair et al., 2009;McCune & Grace, 2002).The relation of variables with canonical axis is represented on results by Pearson correlation analysis (Table 3).Significance adopted was 0,05, but it was changed to 0,025 by Bonferroni correction.Considering the absence of intracolonial influence in the variation of studied species (Menezes et al., 2013), these analysis were based in the average corallites of each colony because of the difficulty in observe the great number of data on graph.Data were homocedastic, presented normal distribution and, consequently, were not transformed.Additionally, a descriptive analysis was carried out with mean, standard deviation and data amplitude.The programs used were Statistica 7 (Stat-Soft®), Primer 6 and Excel 2007 (Microsoft © ).
S. radians also showed interesting morphotypes, despite the small number of colonies analyzed: (1) corallites with small diameter (< 3.0 mm) and shallow columella (< 1.1 mm) in Paramana and Ma-ria Guarda; and (2) corallites with large diameter (< 3.2 mm) and deep columella (< 3.5 mm) in Praia do Forte, Itacimirim and Guarajuba.Septal number was always small for this species.Probably, the improving of our sampling should help us to better describe these morphotypes.
Our results also shows that, in average, S. stellata had larger values of corallite diameter, columellar depth and, mainly, septal number than S. radians.This result is also supported by descriptive data on Tables 4 and 5.However, some overlap can be observed between the two species, particularly among colonies of S. stellata from Maria Guarda and Paramana and colonies of S. radians (Fig. 2, Table 3).In this case, the specific identification of these species is a more difficult task.

Intraspecific variation
As expected, we observed different results depending of morphological variation category.The most important traits that determined the founded morphological variation was diameter of corallites, number of septa, and columellar depth, while colu-  mellar diameter, theca thickness and distance between corallites remained more homogeneous (Fig. 2, Table 3).Within each point, both species vary considerably when colony per colony is compared (Fig. 2, Table 3).In literature, variations in this level have been attributed to genetic factors.Studying species of Porites from the Caribbean region, Brakel (1977) attributed discrepant morphological patterns occurred in similar environmental conditions (subtracts with the same slope and exposure to light) to genetic differences.Similarly, Foster (1979), evaluating colonies of S. siderea from the same region, suggested that the variation within populations could be genetic because when colonies were transplanted between different environments, some of them show morphological variation, but others, not.
Moreover, as previously mentioned, genetic factors cannot be overlooked.According to Neves et al. (2008), S. stellata and S. radians have moderate genetic structure along the Brazilian coast, which could also explain the fact that nearby localities have colonies with higher morphological affinity than those more distant.But these are still speculation.Responses to morphological variation may reside in single or joint action of genetic mechanisms and environment.For TSB and LN, the speculation is greater because the results of Neves et al. (2008) did not extend to the coast of Bahia.In fact, molecular and morphological data for S. stellata and S. radians are still a gap for many sectors of the coast.

Interespecific variation
One of the main goal of this study was to provide additional data for the recognition of the interspecific limits between S. stellata and S. radians.In agreement with the arguments exposed primarily by Neves (2004), significant differences were obtained between these congeners.According to the results, within the morphological characters analyzed, the number of septa was the most informative to distinguish these species.It was possible to observe that there are two groups of S. stellata that are totally distinct from S. radians and, also, one group that overlap morphological groups (Fig. 2, Table 3).In the latter, it becomes practically impossible to differentiate the species through the used characters.For these morphotypes, only presence of the fourth complete cycle of septa, characteristic indisputably relevant to the identification of S. stellata, can be used for differentiation.
The complete fourth cycle of septa for S. stellata has been cited by diverse authors such as Verrill (1868Verrill ( , 1901)), Gravier (1909), Vaughan (1919), Yonge (1935), and posteriorly by (Neves, 2004;Neves et al., 2010 andMenezes et al., 2013).However, it is important to highlight a common confused interpretation.Siderastrea stellata colonies can also include some corallites with an incomplete fourth cycle.It is distinctly different from S. radians, in which a complete 4 th cycle never occurs.Thus, S. stellata and S. radians cannot be confused because of the presence of an incomplete S4 in both species, but should be distinguished by the absence of a complete S4 in S. radians.
Morphological overlays between S. stellata and S. radians were also observed by other authors.Comparing morphological characteristics of Siderastrea, Santos et al. (2004) found that the average diameter of the corallites and septa number of S. stellata from Pernambuco (dcor = 3.43; Nsep = 36.84)are similar to values found for S. radians from Panama (dcor = 3.42;Nsep = 35.19).In addition, these sites were very close in a grouping statistical analysis.One important factor probably influenced this result: the author doesn't consider the occurrence of S. radians in the Brazilian coast.Indeed, Neves (2004) founds more than 60% of S. radians on the Pernambuco samples and records the species for reefs from Formoso, Santo Aleixo, Porto de Galinhas, and São José da Coroa Grande.Thus, noise may have been assimilated to the results.Unlike Santos et al. (2004), who didn't question the taxonomic status of S. stellata, Werner (1996) tried to synonymize the congeners based in a morphometric statistical analysis.However, the author doesn't realize a wide taxonomical review of these species and doesn't consider the occurrence of S. radians for the Brazilian coast.
Finally, despite the feasible morphological differentiation of species, its identification cannot be performed in situ.Diameter of the corallites is the only trait easy to see in the field.However, this characteristic is one of the most variable morphological traits, with great potential for overlap among species.Currently, many scleractinian visual identification guides have been published (e.g., Humman, 2002).But the margin of error for species complexes, without the scrutiny coral and corallites structures is considerable.Observation of diagnostic characters is essential to scleractinian coral identification.Therefore, the use of 'Siderastrea spp.' (see Cruz et al., 2008;Leão et al., 2008) has gained a degree of acceptance as a secure practice for reporting findings of the genus by those other than expert taxonomists.

FIgURE 2 :
FIgURE 2: Graph of Discriminate Canonical Analysis for morphological comparison between S. radians and S. stellata.CV = canonical variable.

TABLE 1 :
Detailed data of collecting sites within two major sections of the Bahia State (Brazilian northeastern coast): the Todos os Santos Bay (TSB), and the North Shore (BS).Lat/Long = georeferenced coordinates, VCN = Vouchers catalogue number of deposited at the Zoology Museum of Universidade Federal da Bahia, Sp(n) = number of specimens per locality (Ss = S. stellata, Sr = S. radians).

TABLE 3 :
Results of Discriminate Canonical Analysis for S. stellata and S. radians.CV1 and CV2 point to about 50% and 32% of the variation, respectively.CV = canonical variable; * = the most heavily weighted characters.

TABLE 5 :
Descriptive statistics: mean (µ), standard deviation (s) and range of morphological variables analyzed for S. stellata and S. radians.General values including all localities.Metric values in millimeters

TABLE 4 :
Descriptive statistics: mean (µ), standard deviation (s) and range of morphological variables analyzed for S. stellata and S. radians localities.Metric values in millimeters.
Laborel (1969/70).:Morphologicalvariation in SideraStrea observed in Boa Viagem Beach and Yacht Club, both areas of sandy substrate; and the morphotypes 3 and 2 of S. stellata and S. radians, respectively, were taken from the sandy bottom pools located on the plateau of the NS reefs (Guarajuba, Itacimirim, Praia do Forte and Genipabu).The place Porto da Barra, a region of sandy substrate presented colonies with intermediate morphology.Curiously, this region is located in a transition between the Todos-os-Santos Bay and the open sea region.Only two morphotypes were previously described byLaborel (1969/70): one, similar to S. radians (corallites 2-3 mm in diameter, incomplete fourth cycle of septa) and another similar to S. siderea (corallites with 5 mm, presence of the fifth cycle of septa), which seem to fit the patterns founded in the present study.But the change was considered 'progressive' by the author, particularly associated with the depth.