SLIGHT GENETIC DIFFERENTIATION BETWEEN WESTERN AND EASTERN LIMITS OF ASTROIDES CALYCULARIS ( PALLAS , 1776 ) ( ANTHOZOA , SCLERACTINIA , DENDROPHYLLIIDAE ) DISTRIBUTION INFERRED FROM COI AND ITS SEQUENCES

P. Merino-Serrais, P. Casado-Amezúa, Ó. Ocaña, J. Templado & A. Machordom. 2012. Slight genetic differentiation between western and eastern limits of Astroides calycularis (Pallas, 1776) (Anthozoa, Scleractinia, Dendrophylliidae) distribution inferred from COI and ITS sequences. Graellsia, 68(1): 207-218. Understanding population genetic structure and differentiation among populations is useful for the elaboration of management and conservation plans of threatened species. In this study, we use nuclear and mitochondrial markers (internal transcribed spacers -ITS and cytochrome oxidase subunit one -COI) for phylogenetics and nested clade analyses (NCA), thus providing the first assessment of the genetic structure of the threatened Mediterranean coral Astroides calycularis (Pallas, 1766), based on samples from 12 localities along its geographic distribution range. Overall, we found no population differentiation in the westernmost region of the Mediterranean; however, a slight differentiation was observed when comparing this region with the Tyrrhenian and Algerian basins.


Introduction
Polyspecific coral reefs disappeared from the Mediterranean Sea at the end of the Messinian, Late Miocene (Esteban, 1996).In spite of this, 33 species of scleractinian corals can currently be found in this sea (approximately half of them are colonial species).Among them, only one species, Cladocora caespitosa (Linnaeus, 1767), remains as a monospecific coral-reef builder.Although many of these species are affected by disruption or regression due to global climate change and human impact, none of them are included in the annexes of the Habitat Directive, and only one, Astroides calycularis (Pallas, 1766), has been included in Annex II of the Bern and Barcelona Conventions (list of endangered or threatened species).
Astroides calycularis is a colonial azooxanthellate coral commonly known as "Mediterranean orange coral" because of the deep orange colour of its coenosarc and polyps (Zibrowius, 1995).It belongs to the family Dendrophylliidae, which comprises nearly 170 species worldwide (Cairns, 2001) and represents approximately 23% of all azooxanthellate coral species (Picciani et al., 2011).Zibrowius (1995) suggested that A. calycularis resembles some species of Tubastrea, a genus that occurs worldwide in tropical seas, and in particular a relatively similar species found in the Cape Verde islands.More recently, based on skeletal biomineralization patterns and 28S rRNA sequences, it has been found that the monotypic genus Astroides forms a monophyletic group with the genera Balanophyllia and Tubastrea (Cuif et al. 2003).But this only demonstrates that the dendrophylliid taxa analysed were in accordance with the current taxonomy and classification, contrary to other families as, for instance, Caryophylliidae or Pocilloporiidae (see Kerr, 2005 or Kitahara et al., 2010).However, a dendrophylliid phylogeny based only on morphological data, showed Tubastrea and Turbinaria as the sister group of Astroides, Balanophyllia not being directly related to Astroides (Cairns, 2001).
A detailed description of this coral has been provided by Zibrowius (1980), and more recently, Goffredo et al. (2011) studied the colony and polyp biometric relationships and intra-colony polyp population size structure.This coral lives in shaded habitats (e.g., vertical walls, overhangs and cave entrances), prefers areas with high hydrodynamics and can be found from the water surface to a depth of approximately 50 metres; however, it is mainly found in shallow waters.In some places, this coral is the dominant species covering up to 80-90% of the surface of the walls.In places of high hydrodynamics, it typically forms massive colonies with polygonal corallites.In sheltered or deeper places, colonies tend to have a bush-shaped morphology with nearly circular corallites.The ecological importance of A. calycularis as a biobuilder has been previously shown in a study based on North African populations (Ocaña, 2005).
This coral has been characterised as gonochoric, both at the polyp and colony level, and as planulabrooder (Goffredo et al., 2010).Field observations performed by various authors have characterised the larvae as having negative buoyancy and a demersal behaviour, and thus crawling until finding a substrate to settle on (Lacaze-Duthiers, 1873;Goffredo et al., 2009).Some authors regard A. calycularis as an ancient Tethyan species (Ocaña et al., 2007(Ocaña et al., , 2009)).It is an indicator of Quaternary climate oscillations, since it is a warm-water species with a tolerance for a narrow temperature range (Bianchi & Morri, 1993).Based on fossil evidence, it was widely distributed throughout the Western Mediterranean Sea during certain periods of the Pleistocene (Zibrowius, 1995), but disappeared from the Northern Mediterranean areas during colder periods.This response is similar to that of other warm-water corals, such as the closely related genus Tubastrea (see Ocaña et al., 2007Ocaña et al., , 2009)).
Currently, the range distribution of A. calycularis is restricted to the south-central part of the Western Mediterranean Sea.In particular, it is found in the following regions: the southeastern Iberian Peninsula, from the Strait of Gibraltar to Cape Palos (Murcia); the northern coasts of Africa, from the Strait of Gibraltar to Cape Bonn in Tunisia; around Sicily and nearby islands; and the Gulf of Naples in the Tyrrhenian Sea (Zibrowius, 1995;Bianchi, 2007;Goffredo et al., 2010).In addition, some records report its existence in the Atlantic coast of southern Spain and northern Morocco (Zibrowius, 1980;Bianchi, 2007).Recently, A. calycularis has also been found in the Adriatic Sea, along the coast of Croatia (Grubelic´ et al., 2004) up to the Gulf of Venice (Casellato et al., 2007).The recent range expansion of this species into the Adriatic Sea seems to have been influenced by the warming of seawater, by the prevailing sea current system and by the rocky coastal configuration (Grubelic´ et al., 2004).However, currently, this species is disappearing in some places because of destruction or loss of habitat caused by human activities, such as coastal development, pollution, diving, angling and illegal fishing of the endolithic date-mussel Lithophaga lithophaga (Templado et al., 2004;Moreno et al., 2008).
To better understand the processes impacting regression of this coral during its history, its current expansion to the northeast and to establish the proper conservation plans, it is important to determine the genetic structure and the extent of gene flow among different populations along its geographical range.It is difficult to disentangle the effects of contemporary gene flow with those of historic population extinctions, expansions and colonisations, all of which have led to the present-day species distribution and population structure.To date, population genetic studies of Mediterranean scleractinian corals are limited to the solitary species Balanophyllia europaea (Goffredo et al., 2004) and Leptopsammia pruvoti (Goffredo et al., 2009); in both cases, these studies were based on allozyme electrophoresis.
Therefore, the main aim of this study is to provide the first characterisation of the population genetic structure of A. calycularis, contributing with this knowledge to the management and conservation of this species.

Sampling collection
Colonies of A. calycularis were collected from 12 sites along its distribution range (Fig. 1, Table 1).We selected superficial colonies (0-5 m) from vertical walls and cave entrances where the population density was higher.At each site, colonies were removed from the rocky substrate with a knife.In order to avoid sampling the same colony twice, colonies were removed from different patches.

Sequence analysis
Following removal of the primer regions, DNA sequences obtained for each specimen and marker were aligned and checked using Sequencher 4.6 (Gene Code Corporation).All alignments were validated by eye.A Blast search was performed on GenBank sequences in order to determine the most similar sequences to use as outgroups; Tubastrea coccinea for COI (GenBank Accession No. DQ445807) and Tubastrea aurea for ITS (GenBank Accession No. AY722796) were chosen as outgroups.
The analyses were first performed for each gene separately.The congruence among tree topologies of COI and ITS genes was assessed by the partition homogeneity test in PAUP* (Swofford, 2002).Nucleotide saturation was evaluated by plotting transition and transversion changes against uncorrected (`p´) divergence values.Sequence analysis was based on the principles of maximum parsimony (MP), neighbour-joining (NJ) and maximum likelihood (ML), as well as Bayesian principles.The evolutionary model that best fit our data was selected using Modeltest 3.06 (Posada & Crandall, 1998).Maximum likelihood analysis was performed by Quartet Puzzling (using 1000 replicates) or heuristic search.Support in the phenetic and parsimony analyses was estimated by bootstrapping (1000 repetitions) (Felsenstein, 1985), and by posterior probabilities in Bayesian analyses.To test the genetic structure of the observed variation, we used nested clade phylogeographic analysis (NCPA) software (Panchal, 2006) to relate genetic structure to the geographic distribution of the samples.The haplotype network was manually converted into a series of nested clades, following the rules provided by Templeton (2004) and applied by the NCPA program (Panchal, 2006).A chisquared test of geographical association of clades and biological inference from nested clades was applied as the basis for the biological interpretation, according to Templeton (2004).

Results
The obtained sequences (a total of 67 for COI and 65 for ITS) were submitted to GenBank with the accession numbers JQ343061 to JQ343125 for ITS and JQ343126 to JQ343192 for COI (Table 1).We counted on 56 specimens in which both regions could be sequenced.Of the 1334 characters included in the matrices (658 for the COI gene and 676 for the ITS region), 1201 characters were constant, 99 variable characters were uninformative, and only 34 characters were informative (when gaps were treated as fifth state of characters, and only 24 when they were treated as missing data).No nucleotide saturation was found.According to COI codon positions, the most informative position was the third, and all of the substitutions were synonymous.The ITS region showed more variability than the COI gene (25 of the global 34 substitutions and 15/24 were due to changes in ITS regions, considering gaps as fifth state of characters and missing data, respectively).Base composition was homogeneous in all of the taxa analysed.The empirical proportions of the different nucleotides were as follows: A=0.232, C=0.209, G= 0.251 and T= 0.306.
The divergence between the outgroup (Tubastrea) and ingroup ranged from 1.28% to 7.33% (for COI and ITS respectively, with respect to specimens from Italy).The percentage of divergence between the 8 different A. calycularis haplotypes, excluding the outgroup, ranged from 0% for specimens belonging to the same population to 1.39% (for COI) and 2.06% (for ITS) between the Italian Massa Lubrense samples and the Murcia samples.Within the western samples (i.e., those from North Africa and the Iberian Peninsula, Alboran Sea basin), the population of Murcia (Algerian basin) presented the highest distances with respect to the other populations.
Given that the partition homogeneity test showed no significant differences (p= 0.29) between individual data corresponding to COI and ITS, the data from those sequences were analysed and presented together.The best-fit model for this global matrix was the "transversional model with equal base frequencies" (TVM+I, where I= 0.84).
The phylogenetic tree showed the relationships among the 8 different haplotypes (considering gaps as missing data) found in this study (Fig. 2).The main cluster (Fig. 2, node 6) represented samples collected along the Iberian Peninsula coasts and in North Africa (Alboran Sea basin).At the base of this group, the population from Algerian basin (southeastern Spain, Murcia) was found to be the most divergent with respect to the other populations.The Italian samples, Tyrrhenian basin, which clustered together based on their geographical origin, grouped a greater number of haplotypes in a smaller area, that is, both Eolie and Massa Lubrense samples were grouped together while the two samples from Capri appeared in different groups.Support values for the main nodes were strong for the terminal branches only (Fig. 2).The lack of support for other branches is likely due to the scarce variation of characters (only 34 informative characters including gaps or 24 characters excluding gaps).
The nested clade analysis (NCA) on the genetic haplotype network is shown in Fig. 3. Eleven different haplotypes were detected; 8 of these haplotypes corresponded to the ones previously defined in the phylogenetic trees.Three new haplotypes were detected when gaps were considered as a fifth character state, thus accounting for the discrepancy between analyses.
The null hypothesis of no association between the position of haplotypes in the network and geographical location was rejected (P< 0.05) for several clades.
The results of NCA showed a network of haplotypes and haplotypes/missing intermediate steps, which were organised in different levels ranging from level 1 (the most simple) to level 5 (the complete set of haplotypes).Based on the final phylogeographic network (Fig. 3), in level 1, we found that all of the haplotypes in the western group (Alboran Sea basin) were directly related, thus forming a homogeneous group with no interme-diate steps, except for haplotypes K (Tangier 6) and F (Murcia, Algerian basin), which had 1 and 8 intermediate steps, respectively.The largest difference was observed in the eastern group (Tyrrhenian basin) haplotypes in which from 5 to 14 intermediate steps were found.The two haplotypes most closely related between the two geographical areas had 10 intermediate steps.
In level 2, we found association between the haplotypes from Eolie (D) and one of the haplotypes from Capri (C).Some clusters included haplotypes from both sides of the Gibraltar Strait, such as cluster 2-13, which included haplotypes G (a widely distributed haplotype in the western group) and H (from the Moroccan locality Negro Cape 4), and cluster 2-4, which included two minority haplotypes observed in samples taken from around the Strait of Gibraltar, J (Tres Forcas, Tarifa, Ceuta) and K (Tangiers).
In level 3, all of the haplotypes from the western group, except for haplotypes I and F, appeared in cluster 3-6.The eastern group was divided into three different sets: the 5 clades grouped into clusters 3-1, 3-2 and 3-4.
The association observed in level 4 interestingly showed a geographical split.Although nearly all of the western haplotypes appeared in cluster 4-3, haplotype I was found in cluster 4-1, even though only one step differentiates haplotype I from haplotypes H or J, both of which are found in the same area as haplotype I.In fact, haplotype I was in cluster 4-1 with haplotypes D (from Eolie) and C (from Capri).Therefore, two of the three clusters represented haplotypes found only in Italy or only in Morocco/Spain, while the third cluster included samples from both of these geographical areas.
Murcia (southeastern Spain) and the rest of the samples from the Alboran Sea.
The population of Murcia is placed in the Algerian basin and northeast of the Almeria-Oran oceanographic Front (AOF), which constitutes the eastern boundary of the Alboran Sea.The AOF is a thermohaline density front generated by the convergence between the inflow of Atlantic water through the Strait of Gibraltar and the Mediterranean water (Tintore et al., 1988).It has been regarded as an oceanographic barrier for dispersal of some species (Patarnello et al., 2007;Mokhtar-Jamaï et al., 2011).This oceanographic front appears to be more efficient in population isolation than the strong current of the Gibraltar Strait, which purportedly should prevent gamete interchange between the northern and southern parts of the Gibraltar Strait (Zane et al., 2000).However, in this study, we found that several clusters contained haplotypes originated from both sides of the Gibraltar Strait, such as clusters 2-13 and 2-4, which had two minority haplotypes observed in samples taken near the Strait of Gibraltar (Tangier and Hacho Mount-Ceuta, on the African side, and Tarifa, on the European side).
Indeed, the westernmost examined populations shared haplotypes, including between those from the Gibraltar Strait and the rest of the Moroccan and Iberian populations, thus showing no significant barriers for gamete interchange.Thus, it is difficult to determine a genetic structure in this area for A. calycularis, despite its being a barrier or transition zone for numerous species (Lo Brutto et al., 2004;Baus et al., 2005;González-Wangüemert et al., 2006;Atarhouch et al., 2007).Recent life histories of such populations may explain the low variation in the species studied here.For instance, the reduction in the effective size or regressions/expansions of populations may be due to previous glaciations (Duran et al. 2004;Lemaire et al., 2005) and present-day human activities.
Another factor that may explain this low level of variation is the selection of markers for analysis.Our genetic studies were based on the analysis of two commonly used genes: one nuclear region (ITS) (Diekmann et al., 2001) and one mitochondrial gene (COI) (Medina et al., 1999;van Oppen et al., 2002).Genetic data for scleractinian coral populations remains surprisingly limited, mainly due to the lack of adequate markers (Ridgway & Gates, 2006).Mitochondrial DNA of anthozoans is known to have a slow evolutionary rate (Hellberg, 2006;Costantini et al., 2010).Studies in Octocorals (Alcyonacea) suggest that the slow evolution of the mitochondrial genome may be due to a mitochondrial DNA mismatch-repair system encoded by the gene mtMSH (France & Hoover, 2002).In addition, ITS sequence markers may display intra-individual rDNA variation (Wei et al., 2006).Problems related to the ITS- -Geographical association of clades and biological inference from nested clade analysis.P is the probability of obtaining a χ 2 statistic larger than or equal to the observed statistic by randomly permuting the nested contingency 10,000 times.For each clade with significant geographical associations as detected by the permutation test, the inference chain and the biological interpretation, according to Templeton (2004), are indicated.Tabla 2.-Asociación geográfica de clados e inferencia biológica deducida de los análisis de clados encajados.P es la probabilidad de obtener por azar un valor de χ 2 mayor o igual al observado en una permutación de 10.000 réplicas.Para cada test en el que se encontró una asociación significativa se indican la cadena de inferencia y la interpretación biológica según las propuestas de Templeton (2004).(Costantini et al., 2007).In our study, ITS sequences had three times more changes compared to the sequences of the COI fragment, despite having sequenced a similar length for both regions.Several taxonomical and phylogeographical questions have been addressed with the use of this nuclear marker (Wörheide et al., 2002;Lam & Morton, 2003), in some cases using its secondary structure as a character (Chen et al., 2004).

Clade
Even though the variation rate of mitochondrial genes can be as much as 10 or 20 times slower than single-copy nuclear genes (McFadden et al., 2004), these genes continue to be used for phylogenetic and phylogeographical analysis (Medina et al., 1999, van Oppen et al., 2001), mainly in combination with other markers (Romano & Palumbi, 1997;Cuif et al., 2003), due to their usefulness or possible comparison with other studies.
Regardless, the near lack of genetic structure found in this study should be re-examined using other variable markers, such as microsatellites isolated for this species (Molecular Ecology Resources Primer Development Consortium et al., 2010), which will confirm if the observed lack of structure is real or derived from the genetic markers used.
Given the genetic data presented here (even with remarkably low variation), conservation strategy policies must focus on maintaining the continuity of the species habitat (threatened by human development) to ensure connectivity between adjacent populations.Some differentiated populations, such as in Murcia, in the western area, have to be especially protected due to their isolation and unique genetic characteristics.

Fig. 1 .
Fig. 1.-Map of sampling sites.The thick black line indicates the current species distribution range.Numbers are referred to in Table1.