Assume that we are studying population genetic structure of a species and demographic events using nuclear markers and mtDNA sequences. Nuclear gene sequences give information about further past events compared to mtDNA genes as nuclear DNA has a higher effective population size, and slower mutation rates. What about comparison of mtDNA and microsatellites? There are contradictory information in literature. Some researchers say mtDNA recover more ancient events as mutation rate is slower than microsatellites. Some other researchers say, as being nuclear, microsatellites recover more ancient information due to the larger effective population size despite their high mutation rate.Which one is correct?
Dear Yeserin,
Nuclear DNA, as you have seen has about four times (assuming balanced sex ratio) the effective population size than mitochondrial DNA in diploid organisms. This extends the expected age of the most recent common ancestor -MRCA- of their genealogies in the same amount, and with it the time frame in which it is possible detectable events that affect genetic variation. So, the general rule is that nuclear markers have the four-fold advantage over mitochondrial markers. The problem with microsatellites is their extremely fast mutation rate which can be three-five orders of magnitude above nuclear and even mitochondrial genes. Such mutation rates provokes a problem of saturation-homoplasy that effectively erases the correlation between the number of mutations and the difference in motif repetitions between two alleles. In other words, the excess of mutations produce that information about past events, which is stored in the statistical patterns of mutations, get effectively lost. In mitochondrial DNA that rarely occurs since there are multiple genes to choose from, and despite some genes could face saturation, other more conserved genes remain useful. As you can guess, there are situations in which a combination of not so large mutation rates in microsats combined with moderate-small populations sizes maintain microsatellites useful and advantageous over mtDNA, while in many others (large mutation rates, large populations) that doesn't occur.
Have a nice day.
Thank you for your answer. I am working with a sea slug species which is an obligate out-crossing simultaneous hermaphrodite. It copulates reciprocally, too. So, population effective size of its mtDNA should be double compared to many other higher eukaryotes. This makes me think even more on this issue.
The microsatellite loci that I am using should be nuclear as many individuals have two alleles.
Dear Edson and Jean-Baptiste,
Thank you both for your helpful answers.
If you want fast-evolving nuclear DNA that can be logically combined in sequence matrices with mtDNA (considering Edson's caveats mentioned above) you might want to try introns.
In spite of Edson's caveats, nuSSRs, specially EST-SSRs, can be used to infer very old events. Selective-sweeps and/or population bottlenecks will cause both a reduction in heterozygosity and the lost of rare alleles (this is an advantage of SSRs over other markers with less homoplasy). The recovery of "normal" heterozygosity levels will occur much faster than the recovery of the rare alleles distribution. As a consequence, genome scans using the lnRV test usually detect many more outliers than genome scans using the lnRH test.
I have observed such behavior in my last publication (attached file, you can get the Supplementary Files from my ResearchGate page), where we could observe divergent selection footprints from events taking place around 4.5 My ago in oaks (very large population size).
Have a nice we
Hi Pablo,
The article is very intriguing. Thank you a lot. I have a small question about "discrepancy between D and GST" part. These two values between species look very different for your data.But if P-values are significant for both estimators at most of the loci, can we still say that there is discrepancy? I am asking because I observed some differences between these two values for my data even though significance levels were similar. Differences are not as high as your data, though.
Hi Michael,
It would be useful to sequence some introns. Unfortunately, I don't have time to do any more analysis.Thank you though as it is great to get ideas that I can consider for future.
No problem Yeserin.
Indeed D and Gst (or Fst) measure very different things. You can take a look at Jost's paper (and others) for details. However, let me put a simple example: Imagine two subpopulations with a very high heterozygosity at one SSR marker, but no alleles in common. Hs is approximately 1 and therefore Gst is approximately 0 (Gst=1-(Hs/Ht)). However the subpopulations have no alleles in common and differentiation should be 1!! (which is the result given by Jost's D). There are many ways to express the differences (i.e., Gst value is bounded by the heterozygosity, etc.), but that simple example clearly shows Gst (and Fst) can not be taken as a real differentiation measurement.
On the other hand, you are right at considering the significance values, but shouldn't forget about the actual differentiation values. Let me put another example. Imagine you have twol subpopulations from an endangered species, with Gst=0.01 (p
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation