I posted this on cross validated but haven't received an answer. Just thought someone here might be able to help:

How to tell if my data samples are independent and whether multiple Kruskal-wallis tests will increase family wise error rate?

I ran multiple K-W tests on my data and realized that I didn't account for family-wise errors. However, everything I've looked at have said that I won't have an increase in error rate if my data samples are independent (if they come from unrelated populations and the samples do not affect each other). How do I know if my data samples are independent? Examples I've seen have not been clear.

I am looking to see if there is a pattern in various fruit traits based on one fruit trait. For example, let's say I have fruits from different species (but within the same genus) that are either red or blue. If I'm looking to see whether fruit length, width, number of seeds, etc., differ between red and blue fruits, would I be able to do multiple K-W tests or are my samples not independent?