Look at the plots, there is one huge outlier in the small number of overall samples which is driving a lot of the results. I don’t even disagree with the conclusions, but this is not strong evidence for it.