Non norma distribution I have a non-normal distribution (Kilograms ~ Years), so I can't use ANOVA test to reject the null hypothesis (that the tree means are equal).
There is a tendency of weight to be 100kg.
Is there a way to test if there are differences between the groups?

 A: There are basically two approaches to dealing with non-normal data: either transform the data or use a non-parametric test.
Data Transformations
You can try to transform your means all kinds of ways. If you search online people will recommend all kinds of transformations for different situations.
I recently had some success using the rank transformation method in Conover and Iman (1981). Take your data from all groups and place them in a single pool. Sort from smallest to largest and assign ranks such that 1 is the smallest number. In cases of ties (for example, duplicate values) assign the average rank for all tying-values.
Then use the normal parametric method on these ranked data.
Nonparametric Options
If you can't or don't want to transform your data consider a non-parametric option. In your use case you might consider the Kruskal-Wallis test, which is kind of a text-book answer for a non-parametric replacement for ANOVA. There are certainly other (and in some cases, better) tools too.
...or maybe nothing at all
You'd like to avoid these complicated options if you could, right? So just use the bog-standard ANOVA test. It's probably just fine even with non-normally distributed data (for example, Bianca et al. 2017). Pay more attention to the homogeneity of variances and not how the data is distributed.
