Today, I looked at results of some runs I have been doing in BEAST to estimate divergence times and other parameters for my project on Least Killifish, Heterandria formosa (Poeciliidae). These analyses were multi-locus, based on cytb and RPS7 (or ‘S7’) sequence data (based on a small number of subsamples of our phylogeographic sampling). Essentially, no one knows the S7 mutation rate for freshwater fishes. In other words, it is poorly understood. However, by unlinking the rates of these two loci I was able to arrive at some estimates of the mutation rate of unexpressed portions of this nuclear gene based on variation in S7 intron 1 and intron 2.
In this analysis, I used a strict clock (a relaxed clock is inappropriate for genealogical data like the data we have) and a broad uniform ‘fish clock’ rate prior (0.08-0.028% Myr-1) for cytb; I partitioned by gene and used a codon model for cytb (1+2, 3) and I ran the MCMC for 108 generations, linking the tree models but not the clock models. Based on my experience reading the literature, nDNA rates in vertebrates are usually between 0.0005% and 0.005% Myr-1, so this was the prior for S7 in the analysis (like the cytb prior, this is very broad, thus more conservative and honest than a thinner interval). BEAST estimated the mean cytb rate to be somewhat high at 0.0198% Myr-1, and the mean S7 rate was 0.00282% Myr-1 (95% CIs were broad though).
I think we will report these values in our Heterandria manuscript (in prep.), but I’m going ahead and throwing the preliminary results up here in case someone is interested in S7 mutation rates.
~ J