To the best of our knowledge, our study presents the most complete mitochondrial comparative- and phylogenomic analysis to date in the Pectinidae family, including the newly-assembled mitogenome of Mim. varia. The composition of the latter conforms to most other pectinid mitogenomes published so far. However, it represents a completely novel gene order, previously not described in any other species of Pectinidae.
Gene content
In addition to the 12 PCGs present in every bivalve mitogenomes, the atp8 gene was annotated in Mim. varia. This is not surprising, given that atp8 is being annotated in more and more newly published bivalve mitogenomes, as well as identified in mitogenomes where it was originally thought to be missing [22]. In our study, the atp8 gene was annotated in 25 mitogenomes in which it was not previously found, in two species (Miz. yessoensis, Pl. magellanicus) it was even detected in two paralogous copies, and in two accessions of Ar. irradians, only as an unusual version of the gene (alternative start codon and a total length of 222 bp)(Table 3). The atp8 gene is very short (135 – 222 bp in the Pectinidae), and is among the most variable mitochondrial PCGs, making it very difficult to detect. The discovery of a functional atp8 gene in most pectinids hints at it not being accessory, as previously proposed [48–50]. Duplicated genes in mitochondrial genomes are not unheard of [51–53], although whether all copies are functional remains unknown.
While it is common for metazoan mitochondrial genomes to contain two trnL and trnS genes, pectinid mitogenomes also commonly contain two trnM genes. In accordance, the mitogenome of Mim. varia contains two trnM genes, however, it contains only no trnS, one trnL genes, but two trnF and trnQ genes. The two trnF genes show low sequence similarity and trnF(AAA) shows low sequence identity to trnF genes in other species as well as every tRNA gene in Mim. varia. Although with our methodology, trnQ2 was annotated in Mim. varia, this might be an artifact.. We base this assumption on the fact that the two trnQ genes in Mim. varia are found next to each other, in the gene block (trnK, trnF, trnQ2, trnQ1, trnE), but the same gene block, containing trnS2 instead of trnQ2, is identified in Mim. senatoria [16] and Cr. gigantea [13], and revealed during reannotation in the current study in Mim. nobilis, Ch. farreri and Miz. yessoensis (Fig. 2, Additional file 3). Also, this trnQ2 gene shows higher sequence similarity to the trnS2 gene than to the trnQ gene in the aforementioned species. While we assume that this difference is the result of an artifact, it is not impossible that in Mim. varia, the trnS gene evolved into trnQ. This is supported by the fact that tRNA genes are often lost in mitochondrial genomes, and that remolding of tRNA anticodons is known to happen in mollusks [7, 16, 54].
Structure of the MNR
High A+T content and stem-loop secondary structures are the common diagnostic traits in identifying the mitochondrial control region [55–57]. Both can be observed in the MNR of Mim. varia (Fig. 1, Additional file 2), pointing towards it serving as the control region.
Annotation of pectinid genomes
The fact that we were able the annotate both the atp8 gene and some new tRNA genes in most pectinid accessions downloaded for this study show that the lack of these genes were most likely artifacts emerging from the difficulty of annotating these features in bivalves. With the expansion of annotation databases and the advancement of annotation tools, these difficulties are getting easier and easier to overcome, leading to progressively more precise annotations.
While variations in tRNA gene content presented by singular species (especially those with only one mitogenome annotated: Pe. albicans and Mim. varia) might well be resulting from sequencing errors, incomplete genomes, or simply misannotations. While the possibility of these cannot be completely excluded, even in the case of species that have multiple genomes published (Mim. nobilis, Miz. yessoensis, Pl. magellanicus), those presenting a phylogenetic pattern are most probably real. A peculiar case is that of trnM, as it is present in more than half of all studied species (Fig. 2). This gene is present in two copies in several invertebrate mitogenomes [58, 59], including some bivalves [20]. The invertebrate mitochondrial genetic code contains five start codons, most of which code for amino acids other than methionine in vertebrate mitochondrial genomes. This means that trnM must do ‘double duty’ as the tRNA for both methionine and formyl-methionine, matching five instead of two start codons [55], which could explain why some invertebrates, including some pectinids evolved to have more copies of it in their mitochondrial genomes.
Numerous pectinid mitogenomes are longer than the usual metazoan mitochondrial genome (~16 kbp), for example Aequipecten opercularis 21–28.2 kbp, Chlamys hastata 23.9–27.2 kbp, Chlamys islandica 22.5–25 kbp, Cr. gigantea 22.8–24.8 kbp [60], Mim. varia 20.4 kbp (the current study), Miz. yessoensis 20.5–21 kbp [12], Pl. magellanicus 30.7–40.7 kbp [15], and Pe. maximus [17, 60, 61], although the available mitogenome for the latter species is only 17.2 kbp long. This is mostly due to inflated non-coding regions, as demonstrated by Ghiselli et al. [55] in the whole Mollusca phylum. We identified tandem repeats of total length larger than 500 bp in the non-coding regions of Pl. magellanicus, Ch. farreri (compare with [11]), Mim. varia and Am. pleuronectes (Additional file 5) and repeats of around 500 bp length scattered throughout the mitogenome of Miz. yessoensis (Additional file 3). Gjetvaj et al. [60] also found that most of the pectinid mitogenomes they studied, contain tandem repeats of various sizes and repeat numbers in their non-coding region, and they found no significant sequence similarity among these repeats. This supports the assumption that the repeats arose independently in every lineage. There is also apparent intraspecific variation in mitochondrial genome size within the Pectinidae [60–62].
While it is likely that correlation exist between the number and length of tandem repeats and genome size, the available number of pectinid mitogenomes does not allow to establish correlations between these traits.
The presence of multiple repeats make it seem very likely that tandem duplication is frequent in pectinid mitogenomes, perhaps enabled by the fact that all genes are coded on the “+” strand. However, this does not seem to influence the mitogenome architecture of other marine bivalves, most of which also code all genes on the “+” strand [11]. In plants, which usually have significantly larger mitogenomes than animals and contain numerous repeated sequences, it is suggested that these repeats promote recombination [63–65]. This might play a role in the evolution of the mitochondrial genome in the Pectinidae as well.
Mito-phylogenomics
The phylogenies presented in our study are in accordance with previous results, both with those using a few genes, and those using the complete set of mitochondrial PCGs. One novel result is that in this group, the two rRNA genes seem to have little power to resolve deep divergences, as seen in the different placement of the Pectininae in both the ML and Bayesian approaches. Similarly to Puslednik et al. [36] and Alejandrino et al. [27], we found that Mim. varia is not placed into one monophyletic clade with the other two Mimachlamys species. Generally, although Chlamydinae itself is monophyletic, its lower taxonomic levels are not well-resolved, and several genera are paraphyletic, as it was presented in previous studies [17, 23, 27, 36, 66]. In contrast to Puslednik et al. [36], who concluded Aequipectini to be the basal clade in Pectinidae (similarly as shown in our rRNA datasets, G and H), we have found that Aequipectini always form a monophyletic group with Pectininae in both of our ML and Bayesian approaches, when we included PCGs in the analysis. It appears that while mitochondrial rRNA genes and the nuclear H3 gene are in accordance [36], mitochondrial PCGs paint a slightly different picture, as can also be seen in, for example, Marín et al. [17] and Yao et al. [23]. However, this could at least be partly caused by the small number of taxa involved in our study, and the different choice of outgroups, as both taxon sampling and outgroup selection are known to influence topologies [36]. Despite the Pectinidae being one of the largest families in Bivalvia, containing around 350 species [67], existing phylogenies usually contain only a handful of species, with the most making up only 31% of all species [27]. Chlamydinae is an especially problematic group within the Pectinidae, perhaps partly because of the low sampling relative to the high number of species in the group. While Puslednik et al. [36] conclude that Chlamydinae —in contrast to Waller’s hypothesis (i.e., that Chlamydinae is paraphyletic, and have provided the ancestral stock for Palliolinae and Pectininae)— is in fact the crown group of the Pectinidae. However, their more recent study [27], doubled taxon sampling and used one more nuclear gene (28S), and subsequently came to the opposite conclusion, confirming Waller’s original hypothesis. Another, recent study by Smedley et al. [66] included yet another nuclear gene (18S), and—as they were studying the Pectinoidea—numerous outgroup species to Pectinidae. They revealed a similar topology to Alejandrino et al. [27], but with substantially lower support. This shows the importance of marker selection and appropriate taxon sampling on pectinid phylogenetics.
The phylogeny reconstructed from gene order data shows a similar, although not identical topology to the one reconstructed from sequence data. The difference between the two trees within the Pectininae can be explained with the likely incompleteness of the Pe. albicans mitogenome, missing a few tRNA genes. Similarly to our phylogenomic results, the gene order phylogeny fails to properly resolve Chlamydinae, which, again, can be attributed to low taxon sampling within this group.
The lack of sampling is even more prominent if we look at phylogenomic studies, as the number of published pectinid mitogenomes is currently low and only slowly growing. Although our study does not resolve this problem, it extends the list of published pectinid mitogenomes, shows the utility of mito-phylogenomics within the Pectinidae and expands our knowledge of the evolutionary history of mitochondria within this remarkable group of bivalves.
Genome rearrangements
Gene collinearity
According to the ‘punctuation model’ [68], mitochondrial genomes are generally transcribed as a single polycistronic RNA from each strand, followed by the enzymatic removal of tRNAs, leading to gene-specific mRNAs [69]. In pectinids, major genes are not always separated by tRNAs (Fig. 6). It is possible that these cistrons are punctuated by secondary structures instead of tRNAs, or that they remain bicistrons. However, given that LCBs are the units of genome rearrangement, and most of them contain one major gene and some tRNA genes, it is probable that these are all separated during mRNA maturation. The one exception from this rule is LCB7, that contains the nad6 and cob genes in every studied pectinid species but contains the atp8 gene in addition to these in Mim. varia and Pectininae+Aequipectini, although in different position, and atp8 together with atp6 in Pl. magellanicus (Fig. 6, Additional file 3). Nevertheless, the variance in the composition of this block points towards it not being a single unit in rearrangements. The order of LCBs also shows a clear phylogenetic pattern. The divergence from this pattern in the case of the three outlier species, Mim. varia, Miz. yessoensis and Pl. magellanicus is most likely stemming from limited taxon sampling, i.e. they are relatively divergent from their closest relatives in our sampling, and their true close relatives would most likely show a similar gene order to theirs, as evidenced in other, better sampled groups in our study (e.g. Argopecten).
Common interval analysis
To the best of our knowledge, our study is the first attempt to reconstruct, at least partially, the ancient pectinid mitogenome. The CREx method is widely used in other animal groups, for example insects, e.g. [70, 71]. While the ancestral gene order could be inferred for Insecta, shared by most groups, and some lineages showing some rearrangement, this is not the case for Mollusca. Katharina tunicata [72, 73] and Solemya velum [26] are hypothesized to carry a mitochondrial genome similar in organization to the ancestral Molluscan and Conchiferan mitogenomes, respectively. However, pectinid mitogenomes are very divergent from these two species, and from closely related groups with available mitogenomes, which severely limits the effectiveness of using these as outgroups, given the large chance for homoplasy. Our method tries to overcome this difficulty, with first reconstructing the putative gene order of internal nodes, including the common ancestor of the Pectinidae, followed by the inference of rearrangement events. Although some internal nodes were not consistent, we recovered the putative gene order of the common ancestor of the Chlamydinae, a monophyletic group in most phylogenetic analyses, including ours. While most previous studies focusing on mitochondrial genome rearrangements in the Pectinidae involved only a few species, we compare genome rearrangements among all available pectinid mitogenomes, with methods not previously used in this group. Confirming the assumption of Marín et al. [17], we have found that TDRL (Tandem Duplication/Random Loss) events are equally important as transpositions in pectinid mitogenome rearrangements.