The effect of model choice on phylogenetic inference using mitochondrial sequence data: Lessons from the scorpions

Chelicerates are a diverse group of arthropods, with around 65,000 described species occupying a wide range of habitats. Many phylogenies describing the relationships between the various chelicerate orders have been proposed. While some relationships are widely accepted, others remain contentious. To increase the taxonomic sampling of species available for phylogenetic study based on mitochondrial genomes we produced the nearly complete sequence of the mitochondrial genome of the scorpion Mesobuthus gibbosus. Mitochondrial gene order in M. gibbosus largely mirrors that in Limulus polyphemus but tRNA secondary structures are truncated. A recent analysis argued that independent reversal of mitochondrial genome strand-bias in several groups of arthropods, including spiders and scorpions, could compromise phylogenetic reconstruction and proposed an evolutionary model that excludes mutational events caused by strand-bias (Neutral Transitions Excluded, NTE). An arthropod dataset of six mitochondrial genes, when analyzed under NTE, yields strong support for scorpions as sister taxon to the rest of Chelicerata. We investigated the robustness of this result by exploring the effect of adding additional chelicerate genes and taxa and comparing the phylogenies obtained under different models. We find evidence that (1) placement of scorpions arising at the base of the Chelicerata is an artifact of model mis-specification and scorpions are strongly supported as basal arachnids and (2) an expanded chelicerate dataset finds support for several proposed interordinal relationships (ticks plus mites [Acari] and spiders plus whip spiders plus whip scorpions [Araneae + Pedipalpi]). Mitochondrial sequence data are subject to systematic bias that is positively misleading for evolutionary inference and thus extreme methodological care must be taken when using them to infer phylogenies.