Molluscan mito-genomics: chances and pitfalls in phylogenetics

The use of complete mitochondrial (mt) genomes ¿ their sequences and gene orders ¿ has raised high hopes of resolving important questions in molluscan phylogenetics. Yet, published data and analyses reveal only a limited power of resolution and several obvious nonsense-clades. Why is that so? In contrast to arthropods or vertebrates, mt-gene order is highly variable within molluscs, and changes in gene order often involve changes in strands the genes are encoded on. A change from the light to the heavy strand, or vice versa, also changes the nucleotide substitution skews of A and T, and C and G. This inhomogeneity leads to wrong model parameters in phylogenetic analyses and, therefore, to unreliable or erroneous trees. We present ten unpublished mt-genomes in a phylogenetic analysis with published data. A gene-by-gene comparison of strand skews reveals some pitfalls in the bivalve data set and possible sources of errors. In addition, we report near-plesiomorphic gene orders from an aplacophoran and a protobranch bivalve. This implies that extensive gene rearrangements have occurred independently in all major conchiferan lineages. Consequently, gene order data may be informative for lower level phylogenies but are unlikely to resolve the relationships among the classlevel taxa.