Skyline Interpretation

[Dylan O'Hearn]

Hello,

I'm analyzing a small wildlife population (7 individuals all collected from within a few miles of each other) using the Extended Bayesian Skyline Plot on RAD loci.  I'm using both haplotypes, so a total sample size of 14, and 50 loci, each of which has at least four variable sites. 

My results are coming out like this (see figure below; time in units of Myr).  This troubles me.  I do expect a priori that there has been a population bottleneck event since the LGM, so the overall pattern makes sense, but the population size estimate for the present is unrealistically high.  Given my reasonable assumptions about generation time, this equates to an Ne of ~200,000.  This may be realistic for the pre-bottleneck portion of the plot that corresponds to the species as a whole, but for the present, it is much too high for a small disjunct population of a rare species. 

I wonder if the sharp rebound toward the present is not the result of an actual inference but is simply something you see due to insufficient signal, in which case I couldn't claim that this is actually evidence of a bottleneck, since it could just be a overall decline or something else.  Can someone shine some light on this situation?  Thanks!