A comparison of methods for estimating substitution rates from ancient DNA sequence data.

BACKGROUND: Phylogenetic analysis of DNA from modern and ancient samples allows the reconstruction of important demographic and evolutionary processes. A critical component of these analyses is the estimation of evolutionary rates, which can be calibrated using information about the ages of the samples. However, the reliability of these rate estimates can be negatively affected by among-lineage rate variation and non-random sampling. Using a simulation study, we compared the performance of three phylogenetic methods for inferring evolutionary rates from time-structured data sets: regression of root-to-tip distances, least-squares dating, and Bayesian inference. We also applied these three methods to time-structured mitogenomic data sets from six vertebrate species. RESULTS: Our results from 12 simulation scenarios show that the three methods produce reliable estimates when the substitution rate is high, rate variation is low, and samples of similar ages are not all grouped together in the tree (i.e., low phylo-temporal clustering). The interaction of these factors is particularly important for least-squares dating and Bayesian estimation of evolutionary rates. The three estimation methods produced consistent estimates of rates across most of the six mitogenomic data sets, with sequence data from horses being an exception. CONCLUSIONS: We recommend that phylogenetic studies of ancient DNA sequences should use multiple methods of inference and test for the presence of temporal signal, among-lineage rate variation, and phylo-temporal clustering in the data.

The phylogenetics of the global population of potato virus Y and its necrogenic recombinants.

Potato virus Y (PVY) is a major pathogen of potatoes and other solanaceous crops worldwide. It is most closely related to potyviruses first or only found in the Americas, and it almost certainly originated in the Andes, where its hosts were domesticated. We have inferred the phylogeny of the published genomic sequences of 240 PVY isolates collected since 1938 worldwide, but not the Andes. All fall into five groupings, which mostly, but not exclusively, correspond with groupings already devised using biological and taxonomic data. Only 42 percent of the sequences are not recombinant, and all these fall into one or other of three phylogroups; the previously named C (common), O (ordinary), and N (necrotic) groups. There are also two other distinct groups of isolates all of which are recombinant; the R-1 isolates have N (5' terminal minor) and O (major) parents, and the R-2 isolates have R-1 (major) and N (3' terminal minor) parents. Many isolates also have additional minor intra- and inter-group recombinant genomic regions. The complex interrelationships between the genomes were resolved by progressively identifying and removing recombinants using partitioned sequences of synonymous codons. Least squared dating and BEAST analyses of two datasets of gene sequences from non-recombinant heterochronously-sampled isolates (seventy-three non-recombinant major ORFs and 166 partial ORFs) found the 95% confidence intervals of the TMRCA estimates overlap around 1,000 CE (Common Era; AD). We attempted to identify the most accurate datings by comparing the estimated phylogenetic dates with historical events in the worldwide adoption of potato and other PVY hosts as crops, but found that more evidence from gene sequences of non-potato isolates, especially from South America, was required.