To examine phylogenetic relationships between island, California and eastern gray foxes, additional publically available mammal mitogenomes were obtained from GenBank and aligned to the fox dataset. The alignment was run through JModelTest v.2.1 and the GTR+I+ Γ model was used to run 1000 pseudobootstrap replicates of the maximium likelihood tree program Garli as implemented on the Lattice grid computing system [52 (link),53 ].
To date the divergence between island and mainland foxes, Bayesian phylogenetic analysis was conducted in BEAST v.1.7.5. Each gene, as well as the entire alignment, were run through JModelTest v2.1 and PartitionFinder v1.1.1. Based on this analysis, no codon partitioning and empirical base frequencies were used with each gene fitting the HKY or the TN93 model. Both a lognormal relaxed and strict clock were tested with a coalescent of constant size. The earliest calibrated radiocarbon date was used as a prior estimate of the time to the most recent common ancestor for all island fox samples. An eastern gray fox sample was used as an outgroup as indicated by the maximum likelihood analysis. The eastern gray fox is deeply diverged from California foxes but this split must be younger than the oldest fossil date for Urocyon so we set the root length to the early Pliocene Urocyon fossil dating to 5.332–2.558 MYA [54 ].
All other priors were left to default settings and the MCMC was run in two independent runs of 100 million chains each, logging every 10,000 chains. An empty alignment was tested to sample for effects of the prior and the resulting poor posterior and prior ESS with values below 200 indicated that the priors were not strongly influencing the tree. Substitution rates were compared to other canids and mammals (seeS1 Text ) to examine how the radiocarbon date prior influenced divergence dates.
To date the divergence between island and mainland foxes, Bayesian phylogenetic analysis was conducted in BEAST v.1.7.5. Each gene, as well as the entire alignment, were run through JModelTest v2.1 and PartitionFinder v1.1.1. Based on this analysis, no codon partitioning and empirical base frequencies were used with each gene fitting the HKY or the TN93 model. Both a lognormal relaxed and strict clock were tested with a coalescent of constant size. The earliest calibrated radiocarbon date was used as a prior estimate of the time to the most recent common ancestor for all island fox samples. An eastern gray fox sample was used as an outgroup as indicated by the maximum likelihood analysis. The eastern gray fox is deeply diverged from California foxes but this split must be younger than the oldest fossil date for Urocyon so we set the root length to the early Pliocene Urocyon fossil dating to 5.332–2.558 MYA [54 ].
All other priors were left to default settings and the MCMC was run in two independent runs of 100 million chains each, logging every 10,000 chains. An empty alignment was tested to sample for effects of the prior and the resulting poor posterior and prior ESS with values below 200 indicated that the priors were not strongly influencing the tree. Substitution rates were compared to other canids and mammals (see
Free full text: Click here
