Post date: Mar 27, 2017 8:41:50 PM
To better evaluate the limited migration low Ne (drift) hypothesis for local adaptation in L. melissa I generated contemporary Ne estimates for the 26 (including ABM) populations using the LD method proposed by Waples and Do (2008). The input was the same 500 random SNPs I used for BayesAss. These were a random subset of the entropy SNPs (common and somewhat spaced out) and thus should be mostly unlinked as required. I used NeEstimator 2.01 for the actual analysis (a java program I ran on my desktop), and chose the random mating option. I moved the results back to the UofU cluster and they are in /uufs/chpc.utah.edu/common/home/u6000989/projects/lmelissa_hostAdaptation/demog/Ne/.
The file out_melissa_ne.txt contains the full results (population numbers are alphabetical), and the results for the MAF > 5% for each population are summarized in LDest.txt, where the point estimate and 95% CIs are given. NA denotes infinity. A plot of Ne across populations is varNe.pdf and the code to make it is nePlot.R.
The mean Ne across populations was 176 (median = 115), which argues for substantial drift.