How about these:

Article Models based on individual level movement predict spatial pa...

Article Movement behavior explains genetic differentiation in Americ...

https://onlinelibrary.wiley.com/doi/full/10.1111/j.1365-294X.2011.05280.x?casa_token=s5I0g7TU9UgAAAAA%3Awcbno3FR0E1X1_SfM1NLjnIrPE9rCpCg0v3O0dKCLjbHRxszfVaITMTyHZksK98jgoC3nGZa-rjxvw

Hi Bart,

This is an interesting question, and I am interested to read studies that addressed it directly.

Thinking about it though, I would say that the argument is probably correct, but it seems to be more a question of resolution of the approach than a question of time scale. FST parameters measure the average genetic difference between groups, while the individual-based genetic distances give a distribution of these differences. So the latter should give a refined estimate of the genetic differences when compared to FST parameters. This could easily be tested through simulations.

The time scale is linked to that of course, but it comes as the by-product of an improved resolution with the individual-based genetic distances.

Best - Thierry

I was searching for some studies addressing this issue because I think it is a very interesting question, but I could not find anything. This calls for a simulation study!

I think, that the reviewer is somewhat right. I am not so sure how he understands "historic", but Fst measures the difference between the heterozygosity of the total population and the heterozygosity of the subpopulations. Hence, Fst should react to changes in the rate of geneflow between two subpopulations fairly quickly. However, since we are talking about reindeers, which I pressume have a rather long generation time (compared to the annual plant I work with) and overlapping generations, Fst should react with some delay. Compared to that, many individual-based measurements of distance can tell us what is going on right now (at the time of sampling). Nevertheless, I find individual-based measures of genetic distance a bit problematic, because they add noise to the data. Or the other way around, we use populations to get a better estimator of the "real" allele frequencies and avoid the noise that comes from individuals.

I guess the reviewer statement is partially correct... Fst doesn't care about time, just differences in nearness to fixation of alleles. But we can easily imagine cases in which the temporal context of isolation & migration would cause Fst versus genetic distances to respond differently. For example imagine a case where two populations (A and B) are 100% fixed for two different alleles, with only a single contemporary migrant moving from B to A after some time of prior isolation. If we increase the isolation time (=more mutations), Fst will stay the same regardless. So the migrant could be homozygous for an allele of 1%, 2%, 5%, 10% (etc) divergence and it wouldn't matter since Fst is just looking at expected heterozygosities based on the frequencies of those alleles. Mean within/ among pop genetic distances however would both increase as prior isolation increases. You could then pick out contemporary migrants as those with high gen distance to their geographical neighbors (and also low gen distance to some source population).

The part of the statement that I think is incorrect is that Fst mostly contains historical information, which doesn't make any sense to me. In the above scenario, Fst would easily allow to differentiate if we changed the rates of contemporary migration. It just doesn't contain any information about how divergent the alleles are that are being exchanged.

Hi all,

Thanks for a very interesting discussion! It is good to hear the viewpoints from others with longer experience on population genetics. I have given it some thought before I read your comments. Thank you Tyler for pointing out that FST indeed is a fixation measure and doesn't quantify the divergence of exchanged alleles.

Overall, I think individual-based statistics can be very useful to pick out contemporary migrants. Pairwise FST's reflect an average genetic distance between populations, but should nevertheless respond quickly to migration. So it is probably, as Thierry pointed out, a question of resolution.

My main concern is that individual-based statistics often have much stronger statistical power (sometimes two orders of magnitude) due to the number of observations. This can be advantageous if you have to make a trade-off between spatial and genetic (individual) sampling effort or in species with continuous distribution (see Luximon et al. 2014 Mol Ecol Resour), but can lead to highly significant results of very low or negligible effect sizes.

I think interpreting the differences in temporal scale of migration/gene flow between individual- and population-based metrics is difficult and should be avoided unless it is explicitly connected with the study aim. A simulation study would indeed be interesting.

I recommend this book: https://www.wiley.com/en-ca/Landscape+Genetics%3A+Concepts%2C+Methods%2C+Applications-p-9781118525289

There is also a landscape genetics course that is starting, if you are interested: https://sites.google.com/site/landscapegeneticscourse/

My colleagues and I have performed a study on populations of both European and American eel (Anguilla anguilla and Anguilla rostrata) along the coastlines of the US and Europe a while ago, which could be of relevance methodologically [Wielgoss et al. BMC Evolutionary Biology 2014, 14:61] Link: Article Introgressive hybridization and latitudinal admixture clines...

The background for our study: several papers found IBD in European eel, while others did not, we thought it might have to do with latitudinal introgression clines at least in the European eel, where introgression signals from American eels would get more and more pronounced from Southern to Northern ranges of the European eel (our study backs this hypothesis).

The genetic differentiation between the two species based on nuclear data is quite low (Fst = 0.0146; Rst = 0.0147), however, so in addition to relying on individual admixture signals derived from STRUCTURE, we also simulated potential introgression along the coastlines by simulating introgression (drawing from the gene pools of both species in a defined manner). In particular, the idea was that increasing the gene flow stepwisely along a gradient should give rise to an IBD signal if the allele frequencies were different enough between the eel species. Based on our real set of 12 samples from Europe, we added either 0%, +1%, +2%, +3%, +4% of the other species' alleles to each successive virtual population mimicking the entire range from Southern-most (Mediterranean) to the Northern-most (Icelandic) population, and inferred isolation-by-distance plots (using actual geographic data from the real populations). This worked well for us, maybe an approach like that could help visualising introgression trends in your study.

(Another comment, maybe you could try to detect hybrids. We relied on "NewHybrids" to detect first or later generation hybrids, am not sure if this is still used or not).

Does the Fst value specific to a paticular population can say whether it is ancestral? If yes, then which values? Higer specifc Fst or lower specific Fst?