Introduction

The mitochondria of the subclass Anthozoa (phylum Cnidaria) have a number of unique characteristics with respect to other Metazoa. These include the existence of many noncoding regions between genes and the presence of complete stop codons in all protein-coding genes (Beagley et al. 1998; Beaton et al. 1998; Fukami and Knowlton 2005; Pont-Kingdon et al. 1995, 1998; Tseng et al. 2005; van Oppen et al. 1999a, b, 2002), very slow nucleotide mutation rates suggesting the existence of a mitochondrial mismatch repair system (Medina et al. 1999, 2006; Shearer et al. 2002; van Oppen et al. 2002), a small number of tRNA genes (Beagley et al. 1998; van Oppen et al. 2002), and the existence of group I introns (Beagley et al. 1996; Fukami and Knowlton 2005; Medina et al. 2006; Tseng et al. 2005; van Oppen et al. 2002).

Group I introns have many characteristics, such as their self-splicing mechanism and their secondary and tertiary structure, that distinguish them from other classes of introns (Burke et al. 1987; Cech 1988; Michel and Westhof 1990). Although uncommon in metazoans, they are the most widespread class of introns. They are found in plant and fungal mitochondria (Dujon 1989) as well as in the genes of nuclei, chloroplasts, prokaryotes, and bacteriophages (Edgell et al. 2000; Kuhsel et al. 1990; Lambowitz and Belfort 1993; Michel and Westhof 1990; Reinhold-Hurek and Shub 1992; van Oppen et al. 1993).

Among the anthozoans, group I introns have been reported only in the sea anemone Metridium senile (Beagley et al. 1996) and in scleractinian corals, which include members of two distantly related clades, the robust and complex corals (Chen et al. 2002; Romano and Palumbi 1997; Romano and Cairns 2000). The complex corals with these introns belong to the families Acroporidae (Acropora tenuis, Montipora cactus, and Anacropora mathami), Agariciidae (Agaricia humilis and Pavona clavus), Poritidae (Porites porites), and Siderastreidae (Siderastrea radians) (Medina et al. 2006; Tseng et al. 2005; van Oppen et al. 2002); robust corals with these introns belong to the families Faviidae (the three members of the Montastraea annularis complex and Colpophyllia natans; note that recent molecular analyses do not support confamilial status for these corals [Fukami et al. 2004b]), Mussidae (Mussa angulosa), and Rhizangiidae (Astrangia danae) (Fukami and Knowlton 2005; Medina et al. 2006).

In both the sea anemone and the corals, one type of group I intron interrupts the NADH dehydrogenase subunit 5 (nad5) gene and encodes several other functional mitochondrial genes (Beagley et al. 1996; van Oppen et al. 2002). However, M. senile also has a cytochrome oxidase subunit I (cox1) group I intron of a different type; it encodes a homing endonuclease with the amino acid motif of LAGLI-DADG that could promote site-specific mobility of the intron by creating a double-stranded DNA break in the intronless allele (Belfort and Roberts 1997; Dujon 1989; Lambowitz and Belfort 1993). A fungal origin has been suggested for several group I introns of this type, including that of the sea anemone Metridium senile (Beagley et al. 1996; Cho et al. 1998; Vaughn et al. 1995) and the sponge Tetilla (Rot et al. 2006). Here we document for the first time in corals the widespread occurrence of this class of intron, which we found in the cox1 genes of 20 species belonging to the robust clade. This intron is distinct from another observed cox 1 group 1 intron (Medina et al. 2006; Chuang et al., in preparation) in insertion site, intron length, and DNA/amino acid sequence similarity.

Materials and Methods

Species

Most species analyzed in this study were identical to those used by Fukami et al. (2004b). However, Plesiastrea versipora, Blastomussa wellsi, and Physogyra lichtensteini were analyzed for the first time (two colonies each).

DNA Extraction

DNA extraction methods were the same as described by Fukami et al. (2004a,b). A small piece (1 × 1 cm) of tissue from each species was put in CHAOS solution (for details, see Fukami et al. 2004a,b), and total DNA was extracted from the solution using the phenol/chloroform extraction and ethanol precipitation method.

Polymerase Chain Reaction (PCR)

In a preliminary study, we determined the entire DNA sequence of cox1 from several species in the suborder Faviina and found the intron in the middle of cox1 in some species. Subsequently, for most species, only half of cox1, including the intron region, was amplified using the following primers: MCOIF (5′ TCT ACA AAT CAT AAA GAC ATA GG 3′) and MCOIR (5′ GAG AAA TTA TAC CAA AAC CAG G 3′) (see Fukami et al. 2004a). The protocol for amplifications of cox1 with the intron region was 94°C for 120 s for preheating, followed by 30 cycles at 94°C for 45 s, 55°C for 60 s, and 68°C for 4 min, with extension for 20 s for each cycle. The amplified fragment was separated by agarose gel electrophoresis and purified using Gene Clean (Promega). The recovered fragment was cloned with the pGEM-T System (Promega) and sequenced using the Dye Terminator Kit (Amersham Bioscience) for both strands. For the intron sequences, the following internal primers were used to determine DNA sequences: Mcox1F4 (5′ TGT TAG CGG GTG CAA TTA CT 3′), cox1_intF (5′ GCC GAC GAG GTC TTT TAA AAG TA 3′), and cox1_intR (5′ TAA CCA TCT GCA TCT AAA AAC CC 3′). For the three genera analyzed for the first time (see above), DNA sequences of two mitochondrial genes, cytochrome oxidase b (cob) and cox1, were determined as described by Fukami et al. (2004b).

RNA and DNA Analyses

The secondary structure of the group I intron was estimated manually using previously published folding data (Bhattacharya et al. 1994; Michel and Westhof 1990) and current understanding of the group I intron recognition process (Cech et al. 1994; Lisacek et al. 1994) and, also, by comparison with conserved regions (such as P, Q, R, and S) of the group I intron of the mitochondrial cox1 gene of Penicillium marneffei (Woo et al. 2003), whose DNA sequence has a high similarity to the coral cox1 intron (see Results). For less conserved regions, the RNA mfold server

(http://www.bioinfo.rpi.edu/applications/mfold/ma/form1.cgi [Zuker 2003]) was used. Sequence analysis was performed with Sequencher version 4.0 (Gene Codes Corp., Inc.). Open reading frames (ORFs) were translated in DNASIS version 2 (Hitachi Software Engineering) using the Acropora tenuis genetic code (van Oppen et al. 2002). PAUP 4.0b10 (Swofford 2002) was used to construct phylogenetic trees of DNA sequences using the neighbor-joining (NJ) and maximum likelihood (ML) methods. Topologies obtained by both methods were very similar, and only the ML trees are shown in this paper. Heuristic searches used TBR branch swapping and 10 random additions of taxa based on either the K81uf model with gamma parameter and proportion of invariable positions (for combined cob and cox1) (for details see Fukami et al. 2004b) or the HYK model with gamma parameter (for cox1 alone and for the cox1 intron). To find an appropriate model of evolution, Modeltest (Posada and Crandall 1988) was used. Indels (only for the cox1 intron) were deleted from the analyses, although the topology was largely unchanged with indels included. The final data set contained 630 (cox1) and 1053 (cox1 intron without indels) characters, including 58 and 51 (64 with sponge) parsimony informative sites for cox1 and the cox1 intron, respectively. Bootstrap values were estimated by the ML method with 100 bootstraps and the NJ method with 1000 bootstraps using the models above.

RT-PCR Analysis

To test the function of the cox1 intron and to determine the splicing site, RT-PCR was carried out. Total RNA was extracted from the tissue of Diploastrea using TRIzol reagent following a protocol recommended by the manufacturer (Invitrogen). The total purified RNA was treated with DNaseI and then reverse transcribed into cDNA using Superscript III (Invitrogen). The mitochondrial cox1 fragment was amplified using super Taq DNA polymerase (Protech) with the forward primer Mcox1F4 (5’ TGT TAG CGG GTG CAA TTA CT 3’) and the reverse primer McoxIRR (5’ CAA TAT CAA GAG AAC TAT TTG C 3’), which were designed based on the cox1-coding regions outside of the cox1 intron of Diploastrea. Amplification was carried out in a PC-9606 thermal sequencer (Corbett Research) using the following thermal cycle: 1 cycle at 94°C for 5 min, then 35 cycles at 94°C for 30 s, 46°C for 30 s, and 72°C for 1 min. The PCR products were checked by 1.2% TBE agarose electrophoresis, purified and subcloned into pCRII-TOPO vectors (Invitrogen), and then transformed into Escherichia coli TOP10 cells. They were sequenced on both strands by using T7 and SP6 primers to confirm that the PCR products were the cox1 gene of Diploastrea.

Sources of Gene Sequences

Coral sequences resulting from this study are deposited in DDBJ with accession numbers AB289561 for cox1 (P. versipora), AB289562–AB289563 for cox1 with cox1 group I introns (B. wellsi and P. lichtensteini,), AB289564–AB289566 for cob (P. versipora, B. wellsi, and P. lichtensteini), and AB289567–AB289584 for cox1 group I introns (other species). Names and GenBank accession numbers of the cox1 group I introns for other organisms used in our analyses are as follows: Tetilla sp. (Porifera) cox1-intron, AM076987 (Rot et al. 2006); Smittium culisetae (Fungi) cox1-intron4, AY863213 (Seif et al. 2005); Penicillium marneffei (Fungi) cox1-I3, AY347307 (Woo et al. 2003); Schizosaccharomyces octosporus (Fungi) cox1-I3, AF275271 (Bullerwell et al. 2003); and Metridium senile (Cnidaria) cox1 intron, U36783 (Beagley et al. 1996).

Results

Characteristics of the Coral cox1 Intron

The group I intron in cox1 was observed in 13 of 41 genera (20 of 73 species, including 3 members of the Montastraea annularis complex) in the Indo-Pacific, but not in any Atlantic corals (Fig. 1). The insertion sites of these introns into cox1, which were determined from the DNA sequence obtained by RT-PCR (Fig. 2), were identical in all species analyzed (Fig. 3). The coral insertion sites were the same as in a Mediterranean sponge Tetilla sp. (Rot et al. 2006), but they differed by eight nucleotides from the insertion sites of some fungi and were also distinct from that of the sea anemone Metridium senile (Table 1). Total lengths of the DNA sequences of the coral cox1 introns were 1077–1129 bases (Fig. 3), and these introns contained the regions P1 through P10 (except P2) and core regions P, Q, R, and S (Fig. 4); these characteristics were similar to those of the intron of the sponge Tetilla sp. (Rot et al. 2006). The data from RT-PCR showed that the splicing of these introns was functional (Fig. 2). These data are consistent with characteristics of group I introns. The open reading frame (containing 310 to 317 amino acids in 20 species) always contained the LAGLI-DADG sequence characteristic of this family of the group I intron-encoded, site-specific homing endonucleases (Fig. 3).

Fig. 1.
figure 1

Phylogenetic relationships of the scleractinian corals based on the cox1 and cob genes with ML analysis. Bootstrap values (estimated by NJ; ML bootstrap values were similar) are shown only on the main branches. Corals from the Indo-Pacific are indicated by white bars, and corals from the Atlantic are indicated by gray bars. Species with a cox1 group I intron are shown by black circles.

Full size image
Fig. 2.
figure 2

RT-PCR analysis of Diploastrea cox1 group I intron. Lane 1, amplification product (0.5 kb) from RT-PCR using the RT product of Diploastrea as a template. Lane 2, amplification product (1.5 kb) from PCR using genomic DNA of Diploastrea as a template. Lane M, 100-bp DNA molecular marker.

Full size image
Fig. 3.
figure 3

Alignments of DNA sequences and amino acids of the cox1 group I intron in four scleractinian corals and the sponge Tetilla (AM076987 [Rot et al. 2006]). Four types of indels are mentioned in the text: Type 1 (T1) shown here is from Diploastrea heliopora, Type 2 (T2) is from Echinophyllia aspera, Type 3 (T3) is from Scolymia sp., and Type 4 (T4) is from Caulastrea furcata. N (shown at beginning and end) is an example of the DNA sequence of the cox1 gene without the intron (Montastraea annularis from Fukami and Knowlton [2005]). Nucleotide differences among these DNA sequences are shown in gray. ID indicates indel, and C and S refer to coral and sponge DNA sequences, respectively (e.g., first indel for coral, ID-C1, and first indel for sponge, ID-S1). DNA sequences of the cox1 gene are represented by lowercase letters and the group I intron sequences are capitalized. Regions in boxes show possible repeat sequences of ID-C2 and ID-C3. Gaps are shown by dashes. Parts of the LAGLI-DADG motif, which is specific to group I introns, are shown in italics, with underlining for the core parts.

Full size image
Table 1. Insertion sites (indicated by ^) of cox1 group I introns with the LAGLI-DADG motif for corals, a sponge, the three fungi with intron sequences most closely matching coral sequences, and a sea anemone
Full size table
Fig. 4.
figure 4

Predicted secondary structure of the cox1 group I intron of Diploastrea heliopora.

Full size image

All cox1 introns in the corals were conserved and easily aligned (Fig. 3). The base composition of the introns and base composition of the ORFs within the introns (in parentheses) was 33.9% (34.7%) A, 13.6% (12.5%) C, 22.0% (21.5%) G, and 30.5% (31.3%) T, whereas that of the cox1 gene was 21.1% A, 16.2% C, 21.0% G, and 41.7% T. No statistically significant difference in base composition was found between cox1 and the intron (p = 0.20, χ2 = 5.28 [4.97 for ORF], df = 3).

Four types of sequences based on indel differences (that in each case also affect amino acid translation relative to the other types) were observed (Fig. 3). Type 1 (found in Diploastrea heliopora, Blastomussa wellsi, Physogyra lichtensteini, Oxypora lacera, Cynarina lacrymalis, and Scolymia vitiensis) is likely to be ancestral (see the following section). Relative to Type 1, Type 2 (observed in Echinophyllia aspera, E. orpheensis, E. echinoporoides, Symphyllia recta, S. radians, Lobophyllia hemprichii, and L. corymbosa) has one deletion of T at position 77. Type 3 (observed in one unidentified species of Scolymia) has an insertion at positions 83–123, which is likely to be a duplication of positions 43–82 (boxed sequence of T3 in Fig. 3). Type 4 (observed in Caulastrea furcata, Oulophyllia bennettae, Oulophyllia crispa, Mycedium elephantotus, Pectinia alcicornis, and P. paeonia) has an insertion at positions 170–219, which is likely to be a duplication of positions 220–291 (boxed sequence of T4 in Fig. 3). In addition, Type 1 has a start codon at positions 58–60, whereas Types 2–4 have a start codon at positions 36–38.

The entire cox1 intron from Diploastrea heliopora (1076 bases) was submitted as a query to BLAST (Altschul et al. 1997) and FASTA (Pearson and Lipman 1988) analyses in the DDBJ database (http://www.ddbj.nig.ac.jp); we used the results from the FASTA search because the BLAST search retrieved only very short sequences. The cox1 introns of D. heliopora had the highest similarity to the sponge mitochondrial cox1 putative group I intron in Tetilla sp. (78.8% in 684 bp; expectation = 1.1e-49), followed by the fungal mitochondrial cox1 group I introns in Smittium culisetae (58.0% in 785 bp; expectation = 3e-15), Penicillium marneffei (58.1% in 757 bp; expectation = 9.3e-14), and Schizosaccharomyces octosporus (56.9% in 706 bp; expectation = 1.7e-09). Notably, the similarity to the cox1 intron of the sea anemone Metridium was much lower (45% in 837 bp estimated using DNASIS). The sponge mitochondrial cox1 intron was very similar to the coral cox1 intron at the DNA level across the entire intron (Fig. 3), whereas the DNA intron sequences of the fungi and sea anemone were too divergent to be aligned with coral cox1 intron at several points. However, large parts of the amino acid sequences of the intron-encoded endonuclease are conserved among scleractinians, the sea anemone, the sponge, and the fungi, allowing us to align them using amino acid sequences for these regions (Fig. 5). In particular, parts of the LAGLI-DADG motif were highly conserved for all (although the cox1 intron from the sea anemone had a large deletion).

Fig. 5.
figure 5

Partial amino acid alignments of the mitochondrial cox1 group I intron among coral and other species (see Materials and Methods for the sequence sources for the latter). Positions are shown for the coral Diploastrea heliopora (positions 53–276 of 310 amino acids; formatted as 53–276/310 in the following), for the sponge Tetilla sp. (79–307/342), for the fungal species Smittium culisetae (15–242/279), Penicillium marneffei (36–258/294), Schizosaccharomyces octosporus (64–263/305), S. japonicus (25–250/293), and for the sea anemone Metridium senile (21–196/223). Dots indicate identical amino acids versus coral alignments, and dashes indicate deletions. Regions in boxes show highly conserved areas, and the region in the dashed box shows a variable area. The LAGLI-DADG motifs are shown in boldface for the coral sequence.

Full size image

Relationship Between Phylogenies of Corals and Their cox1 Introns

Phylogenetic analysis of coral DNA sequences of two combined mitochondrial genes (cox1 and cob; 1395 bases) yielded a tree with eight clades (in addition to two outgroups) (Fig. 1). Clades I–VI were apparent in an earlier analysis of the Faviina (Fukami et al. 2004b), and the three taxa analyzed for the first time (Blastomussa wellsi, Plesiastrea versipora, and Physogyra lichtensteini) form two additional clades (VII and VIII). The relationship of Blastomussa (conventionally assigned to the Mussidae) and Physogyra (conventionally assigned to the Euphyllidae) and the separation of Plesiastrea from other conventionally defined faviids (Fig. 1) were somewhat unexpected, but substantial incongruities between earlier taxonomic groupings based on morphology and recent phylogenies based on DNA sequences are common (e.g., Fukami et al. 2004b).

Thirteen Indo-Pacific coral genera from four different lineages (clades III, IV, VI, VIII) contained the cox1 intron (Fig. 1). In general, the phylogeny of the corals based on either the cox1 + cob DNA sequences (Fig. 1) or the cox1 sequences alone (in both cases excluding the cox1 introns) and the phylogeny of the cox1 intron sequences showed similar topologies (Fig. 6). There was, however, one point of discordance involving the most basal corals containing the intron. The intron of Diploastrea was genetically very distant from the introns of all other species, whereas Blastomussa and Physogyra were basal with respect to Diploastrea in the coral phylogeny.

Fig. 6.
figure 6

Comparison of phylogenetic trees constructed using sequences of the coral cox1 gene (A) versus the coral cox1 group I intron (B) (see legend to Fig. 1 for methodological details). The topology in B was largely the same when the sponge Tetilla intron DNA sequence was used as an outgroup (C). White bars show Type 2 introns, the black bar shows a Type 3 intron, the gray bar shows a Type 4 intron (for the positions of indels, see Fig. 3), and the cluster of bars indicates indels that distinguish coral and sponge introns.

Full size image

Discussion

In this study, cox1 introns were observed in 20 out of 73 species belonging to four phylogenetically distant coral lineages (Fig. 1). This might suggest several independent insertions of the intron into cox1 from one or more donors or, alternatively, a single insertion followed by multiple losses. However, comparison of phylogenetic relationships between the coral cox1 genes and their introns showed broadly similar topologies (Fig. 6). This suggests that the cox1 intron was acquired in an ancestral coral, maintained in some species by vertical inheritance, but lost in many genera. Interestingly, the type 2 intron (one deletion of T: indel-1 in Fig. 3) seems to have appeared twice—in the Echinophyllia species group and the Lobophyllia/Symphyllia species group (Fig. 6B). However, in the case of three basal corals (Diploastrea, Blastomussa, and Physogyra), topologies of the cox1 gene and cox1 introns were quite different (Fig. 6).

There are two possible explanations for this discordance. First, the intron in Diploastrea may have been acquired independently following loss of the intron originally acquired by the common ancestor of Diploastrea, Physogyra and Blastomussa. Alternatively, it is possible that an accelerated rate of evolution in cox1 in the latter two species has resulted in an inaccurate estimation of the phylogenetic relationships among the corals, caused perhaps by the phenomenon of long branch attraction (Felsenstein 1978). However, the distant relationship of Blastomussa and Physogyra is also supported by analysis of tubulin genes (unpublished data), suggesting that the coral phylogeny presented here is correct. Hence the most plausible scenario is that Diploastrea lost the original intron and experienced a second and independent insertion.

Regardless of the details of the history of this intron in corals, it seems highly unlikely that a single origin occurred in the common ancestor of corals and sea anemones. The cox1 group I introns of corals and sea anemones differ in several aspects: (i) the length of the cox1 group I intron of M. senile (672 bp for DNA sequences and 223 amino acids [Beagley et al. 1996]) is much shorter than that of corals (about 1100 bp and 310 amino acids; see also Fig. 5); (ii) the insertion site in M. senile differs from that observed in corals (Table 1); and (iii) the DNA sequence similarity of the cox1 group I intron ORF between M. senile and the corals is low (45%), apart from the characteristic LAGLI-DADG motif (about 66% against the motif-1 of corals at the amino acid level) and the P, Q, R, and S core regions (75%–93% at the DNA level). In contrast, similarity of the entire cox1 DNA sequence between M. senile and corals was high (about 80%). The absence of this type of intron from eight other more distantly related coral families in both the robust and the complex clades (unpublished data, based on the same region of cox1) also supports the idea that the cox1 intron was acquired by a basal member of the lineage analyzed here in detail (conventionally referred to as the suborder Faviina) but that sea anemones and corals acquired these introns independently. This conclusion is also supported by the fact that DNA sequences of the coral introns are much more similar to the cox1 intron of the sponge Tetilla sp. (78.8% [Rot et al. 2006]), as well as to some fungal introns (57%–58% with the mitochondrial cox1 introns of Smittium culisetae, Penicillium marnefferi, and Schizosaccharomyces octosporus [Bullerwell et al. 2003; Seif et al. 2005; Woo et al. 2003]) than to the intron in the sea anemone M. senile (45% [Beagley et al. 1996]).

What do these data suggest about the likely source of the coral introns? Although Beagley et al. (1996) suggested that prime donor candidates for the sea anemone cox1 intron would be ancestors of the endosymbiotic dinoflagellate algae, the existence of a group I intron with homing endonucleases with the LAGLI-DADG motif is unknown in any dinoflagellates (including Symbiodinium) based on sequences available in GenBank. Because the sponge intron is the most similar in sequence to the coral introns and is the only intron reported to date with the same insertion site as the coral introns, one might conclude that sponges and corals share a common ancestral insertion early in the history of the Metazoa with subsequent losses. However, only one of eight sponges, representing a diverse array of higher sponge taxa, possessed the intron (Rot et al. 2006) and the taxonomic distribution of the coral intron is also limited (see above).

Alternatively, corals and sponges may have independently acquired their introns from fungi (fungal donors have been suggested for the plant Peperomia and the sponge Tetilla [Adams et al. 1998; Rot et al. 2006; Vaughn et al. 1995]) or from each other (sponge to coral, or vice versa). Because sponges live surrounding and sometimes covering living corals (e.g., López-Victoria and Zea 2004; Schönberg and Wilkinson 2001) and fungi live in the skeletons of corals (e.g., Bentis et al. 2000; Le Campion-Alsumard et al. 1995; Raghukumar and Raghukumar 1991), such horizontal transfers are not unanticipated. The substantially closer relationship of sponge and coral introns suggests a direct transfer between them, but some as yet unstudied fungi may contain introns that are more similar to the coral and sponge introns than those from fungi studied to date.

Unlike the situation with the cox1 group I intron, the nad5 group I intron in the sea anemone Metridium and some corals does appear to represent a single transfer event. Beagley et al. (1996) searched for cox1 and nad5 introns in several cnidarians (Anthozoa [sea anemone, hard corals, soft corals], Hydozoa, Scyphozoa, and Cubozoa) but only found these introns in the sea anemone. More recently, however, several studies found a group I intron in nad5 in 11 genera from nine families (Fukami and Knowlton 2005; Medina et al. 2006; Tseng et al. 2005; van Oppen et al. 2002), all of which had the same insertion position as in the sea anemone but with different numbers of other coding genes inside of the intron. Our preliminary search for nad5 introns, which was done by sizing using PCR and partial sequencing, also revealed the existence of the nad5 intron (with the same size as in other corals previously reported) in all specimens in this study (data not shown). These findings support the suggestion by van Oppen et al. (2002) that the nad5 group I intron was acquired by a common ancestor of the Scleractinia and Actiniaria. The observed differences in the evolutionary patterns of nad5 and cox1 introns may be caused by a higher mobility of the cox1 intron associated with its LAGLI-DADG homing endonuclease (Bell-Pedersen et al. 1990; Loizos et al. 1994) and by the fact that functional mitochondrial genes (e.g., nad1, cob, nad2, rns for Acropora tenuis [van Oppen et al. 2000]) are encoded within the nad5 intron, which might reduce its mobility.

The reason for the higher prevalence of introns in anthozoan mitochondrial genomes compared to other metazoans remains unclear but may be related to another unusual aspect of anthozoan mitochondrial genomes, namely, their very slow rates of molecular evolution (Medina et al. 1999; Shearer et al. 2002; van Oppen et al. 1999a). For example, introns might be more abundant as a consequence of slow rates of molecular evolution by virtue of lower rates of elimination once inserted. The evolution of anthozoan mitochondrial introns deserves further study as more mitochondrial DNA sequences become available.