A new perspective on the taxonomy and systematics of Arvicolinae (Gray, 1821) and a new time-calibrated phylogeny for the clade

Background. Arvicoline rodents are one of the most speciose and rapidly evolving mammalian lineages. Fossil arvicolines are also among the most common vertebrate fossils found in sites of Pliocene and Pleistocene age in Eurasia and North America. However, there is no taxonomically robust, well-supported, time-calibrated phylogeny for the group. 

Methods. Here we present well-supported hypotheses of arvicoline rodent systematics using maximum likelihood and Bayesian inference of DNA sequences of two mitochondrial genes and three nuclear genes representing 146 (82% coverage) species and 100% of currently recognized arvicoline genera. We elucidate well-supported major clades, reviewed the relationships and taxonomy of many species and genera, and critically compared our resulting molecular phylogenetic hypotheses to previously published hypotheses. We also used five fossil calibrations to generate a time-calibrated phylogeny of Arvicolinae that permitted some reconciliation between paleontological and neontological data. 

Results. Our results are largely congruent with previous molecular phylogenies, but we increased the support in many regions of the arvicoline tree that were previously poorly-sampled. Our sampling resulted in a better understanding of relationships within Clethrionomyini, the early-diverging position and close relationship of true lemmings (Lemmus and Myopus) and bog lemmings (Synaptomys), and provided support for recent taxonomic changes within Microtini. Our results indicate an origin of ~6.4 Ma for crown arvicoline rodents. These results have major implications (e.g., diversification rates, paleobiogeography) for our confidence in the fossil record of arvicolines and their utility as biochronological tools in Eurasia and North America during the Quaternary.

Methods
This dataset is a portion of a chapter of Charles Withnell's doctoral dissertation at The University of Texas at Austin and an article published by PeerJ. Molecular data was obtained from GenBank (we did not sequence any genetic data ourselves). This resulted in three separate datasets; (1) a dataset of only taxa with mitochondrial data (n=146), (2) a dataset of only taxa with nuclear data (n=107), (3) a concatenated dataset that includes both mitochondrial and nuclear loci of n=146 species of extant arvicolines. Maximum Likelihood was computed in RAxML v8.2.12 (Stamatakis, 2014), and Bayesian Inference in MrBayes 3.2.7 (Ronquist et al., 2012). All analyses were conducted on the Cipres Cluster. Five fossil calibrations were used to constrain the phylogeny produced in MrBayes using all five genes.