Recently, I posed the following question to the Evoldir community: ***** I would like to calculate a mutation rate per site, per generation, from a mtDNA control region divergence rate reported in the literature in units of divergence per million years. In my calculation, I am assuming that each species diverges at an equal rate, and I have a generation time of 5 years. The reported divergence rate is 4.8% per million years. To get the mutation rate, I divide 0.048/1,000,000, multiply by 5 for my generation time and divide by 2 to account for the fact that I have two lineages going different directions. This gives me a lineage-specific mutation rate of 1.2 x 10(-7) per-site, per-generation. That seems like an awfully slow rate for the control region. Where am I going wrong? I feel like my math is off somehow but I can't figure out where. ***** I think the consensus is that (a) there are a lot of caveats to calculating a mutation rate this way, but mathematically it is correct, and (b) 1.2 x 10(-7) is well within the range of control region evolution. Specific responses to the question are listed below. ----- The rationale behind your estimate of mutation rate (strictly speaking substitution rate) looks correct to me. A problem, though, might arise from equating short- and long-term evolutionary rates. There is an extensive literature on the slowing down of substitution rates over longer evolutionary timescales. For example, mutation rates estimated from pedigrees on human mtDNA are at least an order of magnitude higher than phylogenetic estimates. Similar conclusions have repeatedly been reached for microorganisms. One likely reason behind this discrepancy stems from purifying selection acting in the long term. Best wishes Dr François Balloux ----- in 1.2 x 10(-7) per-site, per-generation, is this % or base pairs... or something else? When I see your calculation, you just transformed 4.8 % per my to 0.048 /1.000.000 and I dont know if what you are really looking for is substitutions per generation (meaning, you have to account for the length of the sequence in focus)? But perhabs I am also wrong ----- It looks like you are starting with a relatively low divergence rate (4.8%) to begin with in your calculations. For example, some of the control region work that I have done have 14.27% substitution per site per million years. ----- 1.) There are different rates in different lineages so you really need to use a rate specific to your lineage. 2.) I am not sure where you are getting the rate, but if it is already in 4.8% per million years, it should be the rate for that branch [i.e., you would not have to divide by 2] - but double check where those figures came from... 3.) Your math looks ok, I think it looks low just the way you provide it.... 1.2 x 10(-7) per-site is really 12% per million generations, with 5 years per generation you are back to your 2.4 % per million years (but to reiterate you may not have to divide by 2). The 4.8% per million does seem slow for the control region, but again it depends what species that figure came from. I have found a lot of errors in published literature, so i would caution you against using previous numbers without checking where they came from. ----- I can't comment on whether this is a reasonable rate for the control region. In fact nobody could, because you didn't say what species you are working with. Presumably animals, which have a region called "control region", but are they humans, lizards, butterflies, horseshoe crabs, or what? Another problem is that what you are calculating is not the mutation rate, which is the rate at which a site actually changes from one base pair to another, but the substitution rate, which is the mutation rate multiplied by the fixation probability for that region in those organisms. The two are commonly confused in everyday language, but they are only equal for neutral mutations. ----- 1 x 10-7 seems actually pretty HIGH to me, compared to rates for COI and cytb (we usually estimate a rate of about 1x10-8). Maybe for the d-loop this is normal? You might look at Ho et al's paper in MBE (Mol. Biol. Evol. 22(7):1561-1568. 2005), for a discussion of mutation rates and clocks in general. It looks like for the d-loop in mammals they find rates between 10 and 40% per million years (or 1X10-7 per YEAR) ----- I guess it all depends on what kind of rate was reported in the literature exactly: If the reported rate is units of divergence per million years 'per site' your calculation is correct. If it was units of divergence per million years 'for the mtDNA control region as a whole' (so not per nucleotide site), you would need to further divide your rate by the number of nucleotide sites in that region to get the per-site rate. This means that your rate would become even slower. I don't know what rate values to expect from the mtDNA control region, but in comparison to mtDNA COI rates which are estimated around 1% per million years per site your rate is still comparatively high. Another thing to consider is that your calculation is only correct under the condition that the reported rate was for the same species (or at least a species with similar generation times). ----- Joel Anderson Natural Resource Specialist, Genetics Perry R. Bass Marine Fisheries Research Texas Parks and Wildlife, Palacios, TX Ph: 361-972-5483 Fax: 361-972-6352 Joel.Anderson@tpwd.state.tx.us Joel Anderson