Divergent blocks in D8 2017 - intermediate frequency in other ponds?

Driving question: Are blocks of the genome that show elevated divergence between SCA and SCB in D8 2017 in terms of fixed SNPs, more likely to be maintained at intermediate frequencies in other ponds? I took the windows that showed the highest divergence between SCA and SCB, and then graphed the distribution of the alternate allele frequency for all SNPs within those same windows versus a randomly chosen set of windows (matched sample size, but not matched beyond that).

Here is the result for D8 2012. We do see a shift in the expected direction (towards 0.5) in the divergent set of windows.

Here are the results for DBunk and DOily 2017. Again, we see a shift in the right direction, though both sets are farther from 0.5 to start with. Maybe because there are more lineages (or species) contributing to this pooled data.

DBunk

 Doily

However, once I look at D10, or more distant ponds (Bag, B1, W1, W6 so far), I don't see any signal.

However, I think that it is possible that some of the windows/blocks that are highly divergent in D8 are maintained at intermediate frequencies in D10, while others (many?) are not. But don't know if that is just due to chance, or that only a subset of blocks are maintained under balancing selection for longer periods of time.

For example, look at these prop alt distributions for D10 2012 and D8 2012 comparing the overall prop alt to prop alt for a specific scaffold. In this first set it seems like the scaffolds are being maintained at intermediate frequencies in both D8 and D10 (or more in D10 than in D8 if you look at the first one).

1359

1813

 1982

 2142

For this set, however, the distribution of prop alt in the scaffold doesn't look any different than the background distribution for D10, but it does for D8.

1310

 2507

2623

 2676