I am using EEMS to analyze a sample of rodents (Calomys venustus) genotyped at microsatellite loci, living in an agricultural landscape. According to diagnostic scatterplots provided by EEMS, data do not fit the mode. For between demes comparisons, R2 shows very low values==> R2=0.011 for Ndemes=100, to 0.001 for Ndemes= 1200. For within demes comparison, fit gets somewhat better, but it's still not good: R2=0.335 (Ndemes=0.335) and R2=0.234 (Ndemes=1200).
My question is: It is correct to consider the results obtained with EEMS in this situation?
I could'nt find a paper using this method that reports that data do not fit the model (many of them do not report this at all). The only reference I found is this sentence in the original paper by Petkova et al. (2016):
"EEMS attempts to explain observed genetic dissimilarities using an approximation to the stepping stone model. Some datasets may contain features that are not captured by this model, such as recent long distance migrants". In other part of the paper, the authors state that when this happens, EEMS represents this as a barrier, but "cannot represent the fact that these migrants are genetically similar to some very distant individuals".
Therefore, I would say that the lack of fitting does not prevent to use the method, but it is indicating the presence of some particular characteristic, like some long distance migrants. This may be indeed the case in this species, where movement distances as large as 1.6 km were recorded for an individual using CMR.
However, I've doubts about this interpretation due to the lack of comparable works.
Can anyone give me some feedback?
Thank you!!
Marina
If there is no correlation, you can transform your data
You may kindly repeat the experiment, if possible and verify the same before final conclusion
- Badges
- Science topic
- Similar topics
- Correction