I'm working with ADMIXTURE to explore populations and I noticed that 2 of my values are quite similar (the second lowest value is about 101% of the lowest, while the highest value is about 117%). These numbers are based on CV=5. I noticed that in the admixture guide, they had a similar phenomenon, where many of the surrounding values were arguably low themselves, though they only chose the lowest scoring k value to work with.

I'm wondering, when two k value CV tests are similar during CV testing, is it appropriate to report both of them as possible models? Since I'd like to become more familiar with the theory behind this analysis, I welcome additional resource recommendations.