mtpc analyses plots not showing null permutations at one specific threshold

[Charly]

Hello,

I have run a mtpc analysis (testing for weighted global efficiency difference) with 5000 permutations across 6 thresholds and have noticed that one threshold almost only gets 1 to 2 maximum null statistics.

I tried running 3000000 permutations which does increase the amount at thresh .26 but in a markedly smaller range compared to the other thresholds. Is that an expected behaviour or is there something wrong happening with the plotter? 

I tried looking at the matrices thresholded at .26 and there is nothing wrong with them.

I tried looking at the null.dist, its summary statistics, and if I'm not mistaken I should expect a contrast in the second to last 5000 permutations (as .26 is second to last threshold), but nothing unusual appears:

Similarly, looking into the stats ($DT) of the mtpc object does not suggest anything changes at that specific threshold. I also tried looking at the E.glob.wt values per subject across each threshold, and global efficiency barely changes .22,.26.30 thresholds. Everything lets me think there's nothing special about that threshold. 

Is it simply a product of the randomness of graph data? Or could the R plotter have difficulties with it?

Best regards,

Charly

[Charly]

(The second plot was local efficiency,  hence the x axis label, sorry for the mistake. However exact same homogeneity is visible for global efficiency, so null.dist does not suggest anything special)