Microsatellite abundance across the Anthozoa and Hydrozoa in the phylum Cnidaria

Background Microsatellite loci have high mutation rates and thus are indicative of mutational processes within the genome. By concentrating on the symbiotic and aposymbiotic cnidarians, we investigated if microsatellite abundances follow a phylogenetic or ecological pattern. Individuals from eight species were shotgun sequenced using 454 GS-FLX Titanium technology. Sequences from the three available cnidarian genomes (Nematostella vectensis, Hydra magnipapillata and Acropora digitifera) were added to the analysis for a total of eleven species representing two classes, three subclasses and eight orders within the phylum Cnidaria. Results Trinucleotide and tetranucleotide repeats were the most abundant motifs, followed by hexa- and dinucleotides. Pentanucleotides were the least abundant motif in the data set. Hierarchical clustering and log likelihood ratio tests revealed a weak relationship between phylogeny and microsatellite content. Further, comparisons between cnidaria harboring intracellular dinoflagellates and those that do not, show microsatellite coverage is higher in the latter group. Conclusions Our results support previous studies that found tri- and tetranucleotides to be the most abundant motifs in invertebrates. Differences in microsatellite coverage and composition between symbiotic and non-symbiotic cnidaria suggest the presence/absence of dinoflagellates might place restrictions on the host genome. Electronic supplementary material The online version of this article (doi:10.1186/1471-2164-15-939) contains supplementary material, which is available to authorized users.


Background
As ancestral metazoans [1] (Figure 1), Cnidarians provide an interesting system within which to study mutational processes. Cnidarians do not possess a sequestered germ line [2] and their members have a large diversity of life cycles, main adult stages (colonial or solitary), reproduction strategies (sexual, asexual, brooding, broadcasting) and morphological plasticity. Cnidarians also frequently harbor intra-cellular symbionts, the presence of which may, on the one hand, increase the amount of mutagens (reactive oxygen species, ROS) in host tissues [3], but on the other might exert pressure to limit genetic change, ensuring continued communication between partners [4]. Some of the differences in longevity, life cycle, morphology and symbiotic relationships across the Cnidaria might be related to differences in genomic composition or mutation rates [5]. For example, Cnidarians with shortlived medusa stages have faster mitochondrial mutation rates than cnidarians with long life spans and no medusa stages [6].
Microsatellites or Simple Sequence Repeats (SSRs) are sequences of no more than 6 basepairs (bp), repeated tandemly [7]. Point mutations and substitutions are responsible for producing proto-microsatellites (short sequence precursor of microsatellites). Once proteo-microsatellites are formed mutation rates will increase with increasing repeat length, until new mutations or substitutions stabilize mutation rates and the microsatellite sequence [8,9]. DNA polymerase slippage and unequal recombination have been proposed to explain microsatellite expansion, and instability [10], while the frequency of microsatellites is associated with the presence of transposable elements [8,11,12].
Microsatellite loci are genomic features well suited for the study of mutational processes. Apart from high mutation rates, they influence a variety of phenotypic traits from skeletal morphology to host-pathogen interactions [13,14]. Recent studies suggest that microsatellites play an important, direct role in genome evolution because microsatellites may affect DNA replication, translation and gene expression [13][14][15]. For example, in some disease-associated microsatellites, increases in repeat length can originate replication and chromosome rearrangement. In microsatellites located within coding or promoter regions, mutations in the repeats can modify patterns of gene expression.
Microsatellites have been found in every genome studied [7], and taxon-specific variation in microsatellite types and distribution are common [10]. Previous studies concluded that tetranucleotides are more abundant than trinucleotides in vertebrates, while di-and trinucleotides are more abundant in invertebrates [10,16]. Some microsatellite features such as motif size and class can differ even among closely related species [10]; for example, extensive variation has been observed among Drosophila species [17]. The phylogenetic signal of microsatellites depends on the studied feature, for example, the relative abundance of AC and GC motifs and the average lengths of dinucleotides and trinucleotides appear to be conserved in most animals [17]. However, even when microsatellite composition is similar for closely related species, the relationship deteriorates as the evolutionary distance between species increases [18].
Comparisons of microsatellite abundances, lengths and repeat types within and among species have shed some light on the way genomes evolve [10,12,17]. Microsatellite types differ with their position in the genome, genome size does not correlate with the amount of microsatellites it contains, and mutation rates increase with microsatellite lengths. In Cnidaria, comparisons of the two sequenced genomes: Nematostella vectensis [19] and Hydra magnipapillata [20] showed noticeable differences in genome size, base composition, gene conservation and content of transposable elements; however, microsatellite abundance and frequency were not analyzed [5].
The objectives of this work were: 1) to describe the microsatellite abundance and composition in different members of the Cnidaria with focus on the Anthozoa and Hydrozoa, 2) to detect a phylogenetic signal in microsatellite abundance and composition and to 3) correlate the mitochondrial genome mutation rates and the presence of intra-cellular symbionts to microsatellite abundance and composition. We further provide resources for the design of microsatellite primers for future studies of cnidarian populations.

Proportion of sequences with microsatellites
We partially sequenced the genomes of the following Cnidarians: black corals (Leiopathes glaberrima and Tanacetipathes sp.), candelabrum coral (Eunicea flexuosa), the octocoral Plumarella sp., strawberry anemone (Corynactis californica), giant mushroom anemone (Amplexidiscus fenestrafer), plumose anemone (Metridium senile), and fire coral (Millepora alcicornis). A total of 929,398 sequences with a mean length of 474 bp and a mean trimmed sequence length of 192 bp were obtained from these eight cnidarian species. The sequences represent a quasi-random sample of a fraction of the eight species' Figure 1 Classical (non-molecular) phylogeny of the Metazoa adapted from Ball et al. [1]. The split between the Cnidaria and the higher Metazoa occurred around 540 mya.
genomes. The overall proportion of sequences containing microsatellites varied among species, ranging from 1.3% in the octocoral E. flexuosa to 7.7% in the antipatharian L. glaberrima (Table 1).
DNA extracts from 3 of the species (Table 1) might have contained DNA from their intracellular symbionts, despite our efforts to isolate DNA from symbiont-free tissue (see methods). Thus, sequences containing microsatellites were aligned against a local database of Symbiodinium sequences. Sequences with more than 75% similarity to Symbiodinium were eliminated (if found). However, when aligning the putative Symbiodinium sequences against the full NCBI database the top hits were not to Symbiodinium. Regardless, putative Symbiodinium sequences were discarded from the analysis. The microsatellite cover was the same whether the putative Symbiodinium sequences were included or not. Microsatellite cover was also calculated for the Symbiodinium clades, their values being either similar (tetra-to hexanucleotides) or higher (di-and trinucleotides) than those of their hosts (Available in the Dryad Digital Repository: doi:10.5061/dryad.4k5st).
As a supplement to the eight species with partial genome sequences (PGS), scaffolds of whole genome sequences (WGS) of three other Cnidarians were processed as above, including WGS of N. vectensis (59,149 assembled scaffolds, with a mean length of 32,759 bp), H. magnipapillata (126,667 assembled scaffolds, mean length of 32,754 bp) and A. digitifera (29,765 assembled scaffolds, mean length 6,804 bp). The proportion of sequences containing microsatellites in the WGS species ranged from 0.8% in the scleractinian A. digitifera to 5.8% in the anemone Nematostella (Table 1).

Summary statistics
Mean GC content ranged from 33% to 43% ( Figure 2) in the 8 partial genome sequences (PGS) and from 23% to 42% in the WGS species, and differed between the PGS and the WGS (2 tailed t-test, p <0.001). GC content also differed among PGS species and among WGS species (Kruskal-Wallis One Way ANOVA, p <0.005, p <0.001).
The microsatellite lengths (number of repeat units) were also significantly different (One Way ANOVA, p <0.004) among species (Figure 4, Additional file 2). Mean repeat numbers ranged from 4 to 23 (repeats/microsatellite type) for the PGS species and between 4 and 86 repeats for the WGS species. However, sequences with long microsatellites (e.g. >23 repeats) were rare (mean of means =14.6 repeats, s.d. =5.4) in both WGS and PGS.

Phylogeny and ecology
A COI phylogeny of the studied species was constructed as a requirement to test evolutionary models of the microsatellite coverage, using a log-likelihood test. The COI tree showed the expected grouping of Hydrozoa, Octocorallia and Hexacorallia ( Figure 5). Nematostella vectensis was basal to the Hexacorallia, but the long branch in Nematostella suggested increased levels of sequence divergence between Nematostella and the other species. The Corallimorpharia clustered with the Scleractinian sequences while the Antipatharia clustered with M. senile (Actiniaria). The phylogeny also suggested that the Actinaria (N. vectensis, M. senile) are paraphyletic.
The log-likelihood test between phylogeny and coverage of microsatellite types suggested abundance follows a random walk evolutionary model (Table 6). Regressions (p =1.00, Table 6) between phylogeny and coverage of microsatellite types were not significant, indicating there is no phylogenetic signal in the coverage and types of microsatellites. Regressions (p =1.00) between phylogeny and microsatellite lengths yielded similar results.
The hierarchical cluster analysis revealed a weak phylogenetic signal for the coverage and length of microsatellites ( Figure 6). The significant clusters in each analysis had approximated unbiased p-values (AU) of >95% indicating that those clusters were supported by the data [21]. Millepora alcicornis, E. flexuosa, A. fenestrafer and C. californica grouped by microsatellite coverage (Figure 6a). Thus, clustering by microsatellite coverage discriminated between symbiotic and non-symbiotic corals (excepting C. californica). The corallimorpharians, A. fenestrafer and C. californica clustered by length of microsatellites (Figure 6b) making the corallimorpharia the only phylogenetic group recovered by the hierarchical clustering.
Non-symbiotic corals had about 6-fold higher microsatellite coverage than symbiotic corals (Mann-Whitney U Test, p =0.002, Additional file 1: Table S6a), but the microsatellites were similar in lengths (Mann-Whitney U Test, p =0.65, Additional file 1: Table S6b). Microsatellite coverage of Symbiodinium was higher than the microsatellite coverage found in their hosts (Kruskal-Wallis One alcicornis have fast mutation rates (superscript f ), the remainder of the species has slow mutation rates (superscript s ). Symbiotic state is indicated by Y = yes and N = no. The symbiotic species have lower numbers of sequences with microsatellites + , percent of sequences with microsatellites # , and a lower bound for the bootstrapped mean of total microsatellites^(t-tests, p <0.05, equal variances not assumed). Neither the total number of sequences not the mean raw sequence length differed between symbiotic and nonsymbiotic species (t-tests, p >0.05, equal variances not assumed).

Discussion
Here, we described microsatellite abundances and lengths in the phylum Cnidaria with a focus on the Anthozoa and Hydrozoa. We investigated if these characters were predicted by phylogeny, mitochondrial mutation rates or ecology (presence of symbiotic algae). When clustering the Cnidarian species by microsatellite lengths and microsatellite coverage, only the two closely related corallimorpharian species grouped together; no phylogenetic signal was found in the clustering of the remaining 6 species representing higher taxonomic levels. A phylogenetic signal in microsatellite abundance and length in the Cnidaria thus appears to be lost fairly rapidly, but more data is needed to support this finding. Interestingly, dinoflagellate-hosting Cnidaria had 6-fold lower microsatellite coverage and different microsatellite composition from those that do not host dinoflagellates. This result contrasted with the microsatellite coverage of their main associated dinoflagellates, which had also 5 to 6 orders of magnitude higher microsatellite coverage then their hosts. These observations prompted the hypothesis that the close association with dinoflagellates might place restrictions on microsatellite coverage and composition in Cnidarian genomes.

Summary statistics
GC content is correlated with genomic features relevant in determining genome function like gene density, distribution of transposable elements and gene expression levels [22,23]. Here, the average GC content in the microsatellite sequences varied between 23 and 43% (n =11 species). The GC content in the microsatellites did not differ among the studied species; thus the GC proportion in the microsatellites cannot explain differences in microsatellite abundance in the Cnidaria.
Microsatellite coverage in the partial genomes of eight species varied from 4.4 to 215, suggesting a wide range of microsatellite coverage for the phylum. These results mirror previous studies in which the densities of microsatellites vary within phyla [8,10,18,24].
Microsatellites might influence genome evolution and, at the same time, the processes generating and maintaining microsatellites might be altered during the evolution of the genome [10,25]. Clear signals in the abundance and type of microsatellite have been observed when comparing phyla and subphyla [10,17,24] and when comparing sister species [17], but microsatellite composition and coverage becomes more heterogeneous as the species divergence increases [18]. In this article, differences in the frequencies of microsatellite motif types were detected among the cnidarian species included. types was similar for some of the closely related species (L. glaberrima and Tanacetipathes sp; C. californica and A. fenestrafer), but differed for the other related group (Octocorallia).
All theoretically possible combinations of mono-(2), di-(4), tri-(10) and tetranucleotide (33)      The two most abundant motifs are in bold.       observed. Pentameric and hexameric nucleotides generally have longer sequences [26], and thus we might not have been able to detect them because read lengths from 454 GS FXL Titanium had means of 192 bp after trimming. Di-, tri-and tetranucleotides are the most abundant microsatellites in invertebrates [10,16,24], and this generally holds true for the Cnidaria. AC was the most frequent dinucleotide motif found in the studied species. This motif is also common in other metazoans, such as vertebrates and arthropods [10,18]. CG motifs are rare in Cnidaria as well as in other metazoans [10,16,18]. In this dataset, AAT, ATC, AAC and ACT were the most abundant trinucleotides motifs. Interestingly, ACT was reported as rare in most of the previously sequenced vertebrate and invertebrate taxa [10], but is frequent in the Cnidaria. For example, ACT constitutes 40% of the trinucleotides observed in Millepora (Table 4).
Repeat number, length (bp) and motif type are indicators of microsatellite mutation rates [8,11]. A greater number of Figure 4 Boxplots of the microsatellite repeat length (number of repeat copies) in the Cnidaria. T-test between whole genome sequenced (WGS) and partial genome sequenced (PGS) species were significant (p <0.001) for di-, tri-, tetra-, hexa-and pentanucleotides (p <0.02). For PGS species, analyses of variance were significant (p <0.05) for tri-, tetra-, penta-and hexanucleotides. For the WGS species, analyses of variance were significant (p <0.001) for mono-, tri-and tetranucleotides. NS = not significant. Notice the varying scales of the y-axis. See Figure 1 for an explanation of the box plots.  Figure 5 Bayesian genealogy of ten cnidarian species studied, based on partial COI sequences (approx. 700 bp). Posterior probabilities are shown next to the nodes. The scale bar represents 0.05% sequence divergence. Partial COI of E. tourneforti was used instead of E. flexuosa, Bathypathes instead of Tanacetipathes, and A. tenuis instead of A. digitifera.
repeats increases the probability of slippage and thus increases the mutability of the microsatellite [8,11]. Mapping the location and identity of our microsatellites is not possible and, therefore, we cannot directly estimate mutation rate. However, based on the fact that mutability increases with microsatellite length [11] we suggest that the significant differences in average repeat lengths among cnidarian species (Figure 4) might indicate differing mutation rates among species. Further studies are required to test this hypothesis.

Phylogenetic signal of microsatellite characteristics
Previous studies [10,17] suggested that some microsatellite features (i.e. dominant microsatellite motif ) might be related to the evolutionary history of the species and, therefore, those features should be concordant with the species' phylogeny. To test this hypothesis in the Anthozoa and Hydrozoa, we performed hierarchical cluster analysis and Bayesian regressions on the proportion of microsatellite motifs and the microsatellite lengths. The hierarchical cluster analyses suggested a weak phylogenetic signal for the coverage of microsatellite within the Cnidaria at the order level ( Figure 6). Additional sub-family sampling is required before conclusions can be drawn about the similarity of microsatellite coverage among members of the same family. However, other studies suggest that increasing the number of species in a phylogenetic group usually increases the heterogeneity of the microsatellite composition and coverage [18].
The phylogenetic regression showed no relationship between microsatellite abundance or microsatellite length and phylogeny, suggesting evolution of these microsatellite characters was random. However, the dominance of di-, tri-and tetranucleotide motifs was shared between the Anthozoa and the Hydrozoa, the most distantly related Cnidarians [1,5] (Figures 1, 3 and 5), suggesting that di-, tri-and tetranucleotide dominance might be the ancestral microsatellite state for Cnidaria. These results compare with the conclusions of Ross et al. [17] that the phylogenetic signal derived from microsatellite patterns depends on the microsatellite features under study.

Symbiotic signal of microsatellite features
Members of the Cnidaria have diverse symbiotic relationships and this diversity might be related to differences in genomic composition or mutation rates [5]. We explored if, presence of intracellular algal symbionts in the genus Symbiodinium, and host mtDNA mutation rates were associated with microsatellite coverage and/or microsatellite length. The symbiotic corals had 6-fold lower microsatellite coverage than the non-symbiotic corals. Similarly, the microsatellite coverage in Symbiodinium was up to 6-fold higher than the microsatellite coverage in their coral hosts. Therefore, difference in microsatellite coverage cannot be attributed to contamination by Symbiodinium sequences (Additional file 1: Table S3). The methods used for tissue collection and DNA extractions were biased against Symbiodinium, reducing the chances of overrepresentation of Symbiodinium sequences: DNA was extracted from Symbiodinium-free eggs (Eunicea), bleached colonies (Millepora), or the base of the anemone's foot (Amplexidiscus). Thus the number of symbionts was expected to be low [27,28]. The studied coral sequences were then aligned against sequences from various Symbiodinium Clades (provided by LaJeunesse et al., unpublished data) to remove potential contaminants. The frequency and type of microsatellites found in the only published Symbiodinium genome [29] was higher or similar to that of the studied corals. Therefore, we concluded that a potential contamination with Symbiodinium sequences would have increased microsatellite coverage, instead of decreasing it.
The intracellular nature of the symbiosis demands tight communication between the partners [30] and thus it is conceivable that the symbiotic state selects for stable host genomes. Alternatively, factors that co-vary with symbiosis state might explain the differences in microsatellite coverage. The symbiotic species included in this study occur in high temperature and light environments. High light and temperature combined with the mutation stress from reactive oxygen produced as a by-product of photosynthesis by their symbionts might select for efficient DNA repair mechanisms in these long-lived species. To further investigate if microsatellite coverage is influenced by symbionts or light/temperature stress, future work would benefit from the addition of a non-symbiotic tropical cnidarian (such as Tubastrea coccinea).
Previous studies suggested differences between nuclear and mitochondrial DNA mutation rates across the Cnidaria [31][32][33]. For example, the mitochondrial genomes of the Anthozoa have low mutation rates, while the mitochondrial genomes of Medusozoa (including the Hydrozoa) have mutation rates similar to more complex animals [32]. Slow mutation rates in the mitochondrial DNA point to efficient DNA repair mechanisms [31][32][33]; however, cnidarian nuclear allozymes and microsatellite markers show high heterozygosity indicative of high mutation rates [33][34][35].
We considered WGS and PGS species together to compare microsatellite abundances and lengths between Cnidaria with slow mtDNA mutation rates (L. glaberrima, Tanacetipathes sp, Plumarella sp, C. californica, E. flexuosa, A. fenestrafer, M. senile, N. vectensis, A. digitifera) and Cnidaria with fast mtDNA mutation rates (M. alcicornis and H. magnipapillata), and found no difference (p =0.61 for microsatellite coverage, Additional file 1: Table S7a; p =0.99 for microsatellite length, Additional file 1: Table S7b). While additional samples from the Meduzosoa and Hydrozoa (fast mtDNA mutation rates) would strengthen the comparison, our results lend some support to the idea that DNA repair is efficient in the Medusozoa and Hydrozoa but mitochondrion-specific.

Conclusions
This study broadens our understanding of microsatellite distributions in non-model invertebrates. Almost every theoretically possible microsatellite motif was found. Trinucleotides and tetranucleotides were the most common microsatellites in the Cnidaria, challenging the convention that penta and hexanucleotides are most frequent in all nonvertebrate taxa [10].
Cnidarian species showed differences in microsatellite densities, between symbiotic and non-symbiotic cnidarians. Lower microsatellite coverage, although not lengths, in symbiotic corals suggest that either the symbiotic state itself or factors that co-vary with it, such as high light and high temperature, place restrictions on the host genome. Inclusion of non-symbiotic tropical Cnidarians will be informative in this regard. In addition, the whole genome sequence scaffolds from Nematostella vectensis [19], Acropora digitifera [37] and Hydra magnipapillata strain 105 [20] were obtained from GenBank. Whole genome sequences (WGS) were generated from symbiont-free tissues (larvae for N. vectensis and sperm for A. digitifera) [19,37] except Hydra for which contaminant sequences were removed manually after assembly [19,20,37].

Methods
Several steps were taken to avoid/minimize sequence contamination with symbiotic dinoflagellate algae in the zooxanthellate corals (E. flexuosa, A. fenestrafer and M. alcicornis). When available, DNA was extracted from Symbiodinium-free larvae (E. flexuosa). Amplexidiscus DNA was extracted from the base of the anemone's foot, which contains lower concentrations of symbionts [27]. Millepora DNA was extracted from bleached colonies which also feature a significantly reduced symbiont density [28]. In addition, the Partial Genome Sequences (those containing both flanking regions) were aligned to a custom database containing sequences from three Symbiodinium species: 454 sequences of clade C (Wham et al. unpublished) and assembled EST sequences of clades A and B [38], using BLASTn [39] and BLASTx [40] programs to check for the presence of Symbiodinium sequences. Sequences with more than 75 percent identity, alignment lengths larger than 50 bp and e-values lower than 1e −05 were filtered out of the cnidarian sequences because they represented putative Symbiodinium DNA and aligned against the NCBI database (Additional file 1: Table S3).
Cnidarian sequences were imported to the Tandem Repeat Finder (TRF) database [41] and processed using the default alignment parameters as follows: Match: 2; Mismatch: 7; Indels: 7. Sequences were categorized as having at least one flanking region or having two flanking regions (of at least 6 nucleotides) and run in the program SciRoKo [42] to extract all perfect tandem repeats between sizes two and six, containing at least three consecutive repeats. Microsatellite search parameters in SciRoKo were as follows: Search mode: Mismatched, Fixed Penalty; Mismatched Search Setting: Required score: 15; Mismatch penalty: 5; SSR seed minimum length: 8; SSR seed minimum repeat: 3, Maximum mismatches at once: 3. High error rate in homopolymer regions have been observed for Roche 454 [43]; for this reason mononucleotides sequences were excluded from the analyses.
Microsatellite coverage and GC content were calculated for each species based on the full data set, using SciRoKo [42]. Because only one representative of each species was sequenced, the coverage of microsatellite types for each species was bootstrapped using the boot function in R [44], to assign a measure of confidence to the coverage value. The subset of sequences with both flanking regions was used to calculate microsatellite length and repeat number. Analysis of Variance (ANOVA) was performed to compare microsatellite lengths among species using SPSS version 19.0 (IBM). Sequencing methodologies varied between species for which whole genomes are available (N. vectensis and H. magnipapillata: Sanger, A. digitifera: Roche 454GS-FLX and Illumina Genome Analyser IIx) and those that were sequenced in this study likely resulting in different sequencing biases between these two data sets [45]. Thus, WGS and PGS data sets were tested for differences due to sequencing methodology and were only combined when sequencing methodology did not influence the patterns.
For the phylogenetic analysis, COI sequences for each species were downloaded from Genbank (Additional file 1: Table S4), translated to proteins and aligned in Geneious version 5.5.4 [46]. Bayesian phylogenies were generated in Geneious with Mr. Bayes [47] using the mixed amino acid model with gamma distributed variation rates, a uniform branch length clock, and MCMC settings of 4 heated chains for 1000000 generations. A maximum clade credibility tree was constructed in TreeAnnotator v 1.6.2 in the BEAST package [48]. Regressions of the phylogeny and the microsatellite relative abundance and length were performed with BayesTraits [49] using Model A and B, and followed by a log likelihood test, to test for a relationship between phylogeny and microsatellite traits. Species were grouped based on microsatellite abundances and microsatellite lengths using hierarchical clustering in R, with the function hclust from the pvclust package [21].
COI alignment and phylogenetic tree available from the Dryad Digital Repository: doi:10.5061/ dryad.4k5st.
Microsatellite search results for the cnidarian and Symbiodinium species available from the Dryad Digital Repository: doi:10.5061/dryad.4k5st.

Additional files
Additional file 1: Supporting tables. Table S1: Proportion of pentanucleotides motifs found in Cnidaria. Most abundant motifs in bold. Table S2: Proportion of hexanucleotides motifs found in Cnidaria. Most abundant motifs in bold. Table S3: Results from the BLASTx alignments of Cnidarian sequences and Symbiodinium sequences. Table S4: Accession numbers of the sequences (700 bp) used to construct the Cytochrome Oxidase I genealogy. Table S5: Average microsatellite length (average nucleotide length/microsatellite type) found in Cnidaria. Table  S6: Mann-Whitney Rank Sum Test between microsatellite cover (A) and microsatellite length (B) of symbiotic and non-symbiotic Cnidarians. Table S7: Mann-Whitney Rank Sum Test between microsatellite cover (A) and microsatellite length (B) for species with fast or slow mitochondrial evolution.