Comparative Large-Scale Mitogenomics Evidences Clade-Specific Evolutionary Trends in Mitochondrial DNAs of Bivalvia

Abstract

Despite the figure of complete bivalve mitochondrial genomes keeps growing, an assessment of the general features of these genomes in a phylogenetic framework is still lacking, despite the fact that bivalve mitochondrial genomes are unusual under different aspects. In this work, we constructed a dataset of one hundred mitochondrial genomes of bivalves to perform the first systematic comparative mitogenomic analysis, developing a phylogenetic background to scaffold the evolutionary history of the class' mitochondrial genomes. Highly conserved domains were identified in all protein coding genes; however, four genes (namely, atp6, nad2, nad4L, and nad6) were found to be very divergent for many respects, notwithstanding the overall purifying selection working on those genomes. Moreover, the atp8 gene was newly annotated in 20 mitochondrial genomes, where it was previously declared as lacking or only signaled. Supernumerary mitochondrial proteins were compared, but it was possible to find homologies only among strictly related species. The rearrangement rate on the molecule is too high to be used as a phylogenetic marker, but here we demonstrate for the first time in mollusks that there is correlation between rearrangement rates and evolutionary rates. We also developed a new index (HERMES) to estimate the amount of mitochondrial evolution. Many genomic features are phylogenetically congruent and this allowed us to highlight three main phases in bivalve history: the origin, the branching of palaeoheterodonts, and the second radiation leading to the present-day biodiversity.

Keywords: HERMES; bivalves; comparative mitogenomics; mitochondrial genomics; phylogeny.

Fig. 1.—

Main features of mitochondrial genomes. Five large-scale mitochondrial features are compared across four bivalvian subclasses (the fifth subclass, Anomalodesmata, was excluded because only one mitogenome was available): Op, Opponobranchia (purple); Pa, Palaeoheterodonta (blue); Pt, Pteriomorphia (red); He, Heterodonta (green). Length, median length of the genome; %UR, median percentage of Untranslated Regions; Annotated genes, mean number of annotated genes; A-T skew, median A-T skew; G-C skew, median G-C skew. Differences between subclasses were tested using the Krukal–Wallis and the Dunn's test; Anomaldesmata and Opponobranchia were excluded because of low sample size (N = 1 and N = 2, respectively). Asterisks above columns refer to levels of significance in both pairwise comparisons or in a single one (bracketed); *P < 0.05; **P < 0.01, ***P < 0.001.

Fig. 2.—

Conserved sites of protein coding genes. Single genes were aligned, masked using four different approaches, and finally only sites selected as phylogenetically informative by at least three softwares were kept. Whenever possible in terms of sample size, the same procedure was carried out for reduced datasets, corresponding to single subclasses (3 datasets) and single families (5 families). Each chart shows the proportion of datasets keeping single sites: this is 0 (discarded site) or 100 (kept site) at the class level, and ranges from 0 to 100 at the subclass and family level, because more datasets are available.

Fig. 3.—

Nucleotide and amino acid genetic distances. The mean genetic distance across the complete dataset ± one standard deviation is shown for each PCG; brown, nucleotides; green, amino acids. Differences in distance distributions were tested using the Kolmogorov–Smirnov test; ***P < 0.001.

Fig. 4.—

Example of four saturation plots. Pairwise p-distance is plotted over Jin–Nei/Kimura distance for nucleotides (above)/amino acids (below), respectively; nt, nucleotides; aa, amino acids. All saturation plots are available as supplementary file S11, Supplementary Material online.

Fig. 5.—

Distance PCA. Single-gene pairwise distance matrices were used as input data in order to investigate differences between PCGs.

Fig. 6.—

The mitochondrial phylogeny of bivalves. Ultrametric tree of 98 bivalve species computed using r8s starting from the best-known likelihood tree computed with RAxML and the consensus Bayesian tree. RAxML bootstrap values/MrBayes posterior probabilities are printed at each node (an asterisk means a bootstrap of 100 and a posterior probability of 1.00). Below the tree the phylogenetic informativeness for each PCG is shown. The geological timescale in Mya follows Cohen et al. (2013).

Fig. 7.—

Bivalve mitogenomics PCA. Input data are shown in supplementary file S5, Supplementary Material online (see text for details). Different subclasses are indicated with different shades: shades of blue, Palaeoheterodonta; shades of violet, Opponobranchia; shades of red, Pteriomorphia; shades of green, Heterodonta + Anomalodesmata; species of the same family are indicated with the same color.

Fig. 8.—

The HERMES index. To highlight HERMES differences between subclasses, species are horizontally listed by subclass and then in alphabetical order.