Difference between our estimators under a selective sweep. Mean value for our estimated Tajima’s D, for every 50 kb windows for 100 1 MB region for 25 samples. For the most progressive LRT cutoff some windows did not have data. This plot is based on a depth of 2× and an error rate of 0.5%. Figure a) is using the raw Tajima’s D estimate for the genotype calling methods. In Figure b) we have standardized the genotype calling methods relative to the estimates from a dataset of 100 1 MB neutrally evolving regions.