Well, I guess the question is: "did between species trait variation (partly) arose as a result of natural selection or is all differentiation the result of drift? You can, with your data, potentially address this question by testing if morphological variation tracks variation in a known behavioural or abiotic measurement in a way that it implies functionality on part of the morphological feature. Say - and this will be a very classic example - your measurement is bill morphology in in Darwin's finches. If you can somehow test in a robust statistical framework, preferentially correcting for phylogenetic distance among the lineages involved, that the distribution of bill sizes tracks the distribution in seed sizes eaten by the finches you study, depending on the outcome of your analysis you could make an argument about natural selection for dietary specialization driving bill morphology. There is, on this specific example, but also on many other examples including including but definitely not limited to limb morphology and sprinting behaviour in lizards, body elongation and swimming capacity in several fish species, ample research that should be able to guide you towards asking the right question given the data you have. Just look at the classic work of the Grants on the finches or Jonothan Losos on the lizards to get going with the literature. On a side note, preferentially you turn around this process, where you first think of which question you want to answer, then how you want to answer it, then how your statistical framework should look, then collect your data. 

It is impossible to know if the trait occurred by natural selection because we cannot observe the process. How do you know it wasn't created? I don't know of any other evolutionary process apart from natural selection, but how about introduction by aliens a la Ridley Scott's Prometheus haha. My point is that natural selection is an enormous assumption on which to base any research imo lol. I usually avoid it. I recommended your question btw :)

Well, I guess the question is: "did between species trait variation (partly) arose as a result of natural selection or is all differentiation the result of drift? You can, with your data, potentially address this question by testing if morphological variation tracks variation in a known behavioural or abiotic measurement in a way that it implies functionality on part of the morphological feature. Say - and this will be a very classic example - your measurement is bill morphology in in Darwin's finches. If you can somehow test in a robust statistical framework, preferentially correcting for phylogenetic distance among the lineages involved, that the distribution of bill sizes tracks the distribution in seed sizes eaten by the finches you study, depending on the outcome of your analysis you could make an argument about natural selection for dietary specialization driving bill morphology. There is, on this specific example, but also on many other examples including including but definitely not limited to limb morphology and sprinting behaviour in lizards, body elongation and swimming capacity in several fish species, ample research that should be able to guide you towards asking the right question given the data you have. Just look at the classic work of the Grants on the finches or Jonothan Losos on the lizards to get going with the literature. On a side note, preferentially you turn around this process, where you first think of which question you want to answer, then how you want to answer it, then how your statistical framework should look, then collect your data. 

Indirect confirmation of natural selection is the parallelism.

If in different taxa we find similar signs (morphological, behavioral, physiological) in similar ecological conditions, we have reason to believe that these parallelisms are a consequence of selection.

Hello Ana

I would recommend you to follow the changes in frequency in time of the variants of the trait you are studying. Having three populations you may be able to compare the differences between those three and maybe relate those with the ecosystem, you should note though that natural selection can be decomposed in four elements: viability, fecundity, gametic selection and sexual selection, and the trait may be more influenced by one of them rather than all four.

The other forces you need to consider are genetic drift, mutation and gene flow; with your data I would say you can test how the trait (yes, you can use one) would evolve under each of the four, also you can model how the trait would change using models of migration-drift, migration selection, etc. I believe that any book on population genetics would provide you with the info about how to calculate the different forces. Having this "simulation" done it would be interesting to see if the changes in the trait after a few generations match what you predicted.

Hope this helped

Another thing you could do to test genetic drift and migration is to test for isolation by distance between the three populations, using Fst (if you have molecular data) or Pst (which may be useful as I think your trait is quantitative)

Dear Ana

You can test gene frequency for Hardy-Weinberg equilibrium, in the case of a population meet HW law then there is no evolution occur, and vice versa.

Good Luck

As I understand it, all species (and all individuals) are products of natural selection. All variation, including that associated with mutation, genetic drift et. al. is the expression of natural selection in action - i.e. evolution. From this orthodox Darwinist point of view this, and every other trait, evolved by natural selection.  QED

Hi Ana,

Extremely high or low genetic differentiation at very few gene loci suggests that diversifying forces are acting non-randomly (differential selection or non-recurrent mutation) or uniformly (similar selection regimes in all populations). 

Gregorius HR, Degen B, König A (2007) Problems in the analysis of genetic differentiation among populations a case study in Quercus robur. Silvae Genet. 56: 190-199.

Best wishes

Hi Ana,

I don't know if you can prove if the trait evolved by natural selection but I know you can evaluate if natural selection is currently occurring in the given trait. You need to measure the trait you are interested and measure some fitness proxy. With that kind of data you can use the Lande & Arnold 1983 model to evaluate this. I have used it to evaluate how natural selection is operating in morphological traits of an insect, if you want you can check it on my page. 

Best wishes

In plants, a useful tool has been common garden studies, comparing the populations' fitness in the respective parental environments to test for local adaptation, and evaluating whether the traits you are interested in are correlated with fitness -- assuming they're heritable in the first place.  The latter is typically done with the Lande and Arnold approach Daniel recommends, but more sophisticated tools such as the Aster modeling approach developed by Shaw and Geyer are preferable in many cases.

Dear Ana,

there are tons of different models and test that you can try, and I suggest you this reading. A lot of variables are involved in your question, so it is not easy to answer properly. Take into account previous answers regarding population genetics and of course the type of molecular marker you are working with.

Good luck.

https://www.ncbi.nlm.nih.gov/pubmed/16285858

Hi,

In addition to the other suggestions, You could try to estimate population genetic differentiation for morphological traits (Qst) between your populations and compare with corresponding Fst estimate (using any molecular marker available). If Qst is significantly larger than Fst you could consider that the difference is caused because natural selection is acting on your morphological trait. There are several studies using Qst/Fst and you can get a typical experiemntal design from them.

Alternatively you may study variation on morphological trait and estimate its relation with fitness (any component) following Arnold and Wade 1984 methodology. If fitness is working on your trait you should find  higher fitness estimate for certain range of the trait distribition...

Best

All Genetic Algoritms are the tests. Passed ;-)

But for your problem specialized simulation is needed, probably.

Thank you all for your answers,

Emilio Rolán-Alvarez, I have been reading about QST-FST and trying to remember about this estimation, that I studied some years ago in a course of genetics or population genetics. However, biology and genetics are not my field and my knowledge about this is very basic (in this part I should mention that I am a physical anthropologist, and my data are linear dimensions of human teeth), so I have some basics doubts that I would really appreciate if you could clarify me.

First, I think I can not use QST since I don't have genetic data on the traits I am interested in (I have heritability estimations though), but I have read that PST is the phenotypic equivalent for QST(I read it in Grant and Grant, 2010 book entitled: In Search of the Causes of Evolution: From Field Observations to Mechanisms) and that the only genetic information you need for calculating it is heritability (which I have). If this is correct, do you think my data is appropriate for PST estimation?

As for FST, you say that I could use any molecular marker available, but really any? even molecular data not related to teeth?

Hi,

Pst is hardly the analougous of Qst, because it has confounded environmental and genetic effects in an unknown way, if you have not information of different families forget about Qst/fst comparison. Try another way to show indirectly the effect of natural selection. Can you correlate/regress the teath variability aganst individual fitness estimate (number of sons, age?, etc) using Arnold and Wade gradient analysis?

Good luck! Emilio

Thank you for the answer!

I understand the problem with Pst, but I should mention that heritability in teeth dimensions are really high (>80%). So, considering this, is still Pst a bad option?

There is no information of individual fitness estimate mainly because the sample is from archaeological collections, and most of the individuals are represented just by their teeth (and only by one tooth in some cases).

If I have to forget about more "complex" estimations like those you mention, I think maybe using simply the coefficient of variation and skewness of the data may be an option for evaluation directional selection. I am saying this because I have read Wolpoff (1969), and some works in archaeology based their conclusions of natural selection using just the coefficient of variation: 

"If relaxed selection is to be proposed as the cause of hominid dental reduction, an associated increase in dental variability must be shown for at least the period of reduction" (Wolpoff, 1969)

But, I wonder if this paper is too old and if this calculation is enough to make conclusions on this subject.

Hi

Heritability is the proportion of additive genetic variation within population while Qst is the genetic differentiation BETWEEN populations, they do not need to be related or be of similar size, they are absolutely disconnected measurements.I can give you two hypothetical alternative scenarios using two populations A and B as an example.

Scenario 1; most phenotipic variation within the two population is genetic (but genetic differentiation between populations A and B is near zero)

Scenario 2: most phenotypic variation within opulation is environmental (but the small genetic variation differ completelyt between populations A and B)

Therefore, Pst is less than nothing but actually can be biassed, you will never know the true magnitude of Qst, as PST is the phenotypic differentiation (but the population differences could be caused because phenotypic plasticity; i.e. being environmentally induced)... 

Hi! There are some useful but difficult to interpret tests, I mean sometimes could be difficult to isolate NS as the only oxplanatory factors, unless you are very familiar with the study case and have strong expectations. For example: Lande (1979), Ackermann & Cheverud (2004).

Gracias Lumila :)

An analysis of the causes of natural selection – how and why selection operates – is essentially an ecological question. Observational studies are not sufficient to identify the causes of the selection. To certify that a phenotypic trait is being selected, it is necessary to conduct experimental manipulations in the field. (Of course, this is not always possible. In these cases, we rely on extrapolations...) It’s not quick or simple, which in part explains the origin of some controversies among scholars.

The literature on the subject, however, does not stop growing. At the risk of sounding repetitive, I think it would be worth consulting: Wade & Kalisz. 1990. The causes of natural selection. Evolution 44: 1947-55; and MacColl. 2011. The ecological causes of evolution. Trends in Ecology and Evolution 26: 514-22.