I'm analyzing abundances from various ecological surveys. So data was collected across multiple independent years from three sites within each year. The years fall under two temperature conditions(warm or cold). In order to analyze differences between the two conditions a Mann Whitney U test was used in place of a t.test since the data is not normally distributed. However, I want to see if there is any significant difference across a factor of year. If ANOVA can't be used, is a Kruskal-Wallis test appropriate for this kind of data? I'm getting mixed responses from other sources and I am not very familiar with non-normal data tests. The data isn't ranked data, which is why I'm hesitant to Kruskal test for it. I plan to use the test to also check for significant differences in richness, Shannon diversity, and evenness index.