I am developing a computational simulation for DNA barcoding. One of the parameters in my simulation is the number of subpopulation/demes, which I label as $K$. Most studies that use DNA barcodes tend to focus on a single geographic region.

My question is: is it valid to presume that a single geographic region is equivalent to a single subpopulation/deme?

For example, if the focus was on cichlids of Lake Malawi, would it be a safe assumption that Lake Malawi comprises a single subpopulation/deme?

Eventually, I would also like to incorporate migration rates into my simulation, but these are very difficult to estimate without accurate data on $F_{ST}$.

Any insight would be most helpful and is warmly welcomed.


For simulating DNA barcoding, it is valid to presume that a single geographic region is equivalent to a single subpopulation. Although there do occur cases where one or two nucleotides differ at the same location, that is rare, and these few differences would still cluster together in any analysis of barcode data.

| improve this answer | |
  • $\begingroup$ In terms of developing a simulation (simulations need not be perfect), over 50,000 specimens that I personally barcoded, I virtually never saw differences at the same location, and most data in BOLD are from single locations. "Region" in BOLD most often = single site. rstb.royalsocietypublishing.org/content/371/1702/20160025 $\endgroup$ – Karl Kjer Dec 22 '17 at 11:09
  • $\begingroup$ Thank Karl! Based on your advice then, I am thinking that a user of my finished simulation could simply set the number of demes (K) to be equal to the number of sampling sites. Other more-involved options include using the software program STRUCTURE (Pritchard et al., 2000) to determine K using a Bayesian framework. $\endgroup$ – compbiostats Dec 23 '17 at 5:15
  • $\begingroup$ Yes, I agree. Sampling in BOLD is supposed to be geographically diverse. But it is common to have a single individual barcoded. Sometimes however, hundreds were sampled at the same site, and these were almost always identical. When the sampling was designed for barcoding, sites are usually at the extremes of species distributions, and these almost always differ. In this case, sites are not regions like "Virginia", but places where a collector stoped the car and collected. I have heard of "STRUCTURE", but have not used it. It sounds good though. $\endgroup$ – Karl Kjer Dec 23 '17 at 12:43

It is impossible to answer your question in the general sense. Yes, detailed population structure may matter. It very much depends upon what your question is and what parameter space you are willing to answer but it is possible that assuming panmixis would be unrealistic.

Note that accurate data on $F_{ST}$ won't suffice to estimate migration rate $m$.

Note also that many simulation platform already exist and you might not have to reinvent the wheel (see Sequence evolution simulation tool).

| improve this answer | |
  • $\begingroup$ Thanks! I am not aware of any DNA barcoding studies that estimate migration rate; however an isolation-by-distance assumption would be required given population subdivision into demes. $\endgroup$ – compbiostats Dec 21 '17 at 23:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.