Hi,

Here are a few references:

https://www.researchgate.net/post/How_the_sample_size_does_affect_the_Permanova_results

https://groups.google.com/g/pc-ord/c/Njcf0ufnUps?pli=1

Kelly BJ, Gross R, Bittinger K, Sherrill-Mix S, Lewis JD, Collman RG, Bushman FD, Li H. Power and sample-size estimation for microbiome studies using pairwise distances and PERMANOVA. Bioinformatics. 2015 Aug 1;31(15):2461-8. doi: 10.1093/bioinformatics/btv183

Analysis of similarities (ANOSIM) for 2-way layouts using a generalised ANOSIM statistic, with comparative notes on Permutational Multivariate Analysis of Variance (PERMANOVA). Paul J. Somerfield, K. Robert Clarke, Ray N. Gorley. https://doi.org/10.1111/aec.13059

Thank you very much for your answer. I will look into the reffereces you suggested.

In general PERMANOVA should not be restricted to equal sample sizes. However, you are building a distribution from the observed data. You are hoping that the observed data is representative of the underlying distribution. This is more likely as sample size increases. Building an underlying distribution assuming that the null-hypothesis is true only works if there are enough values to build an underlying distribution. There seems little point to any 2-way PERMANOVA when there are only two or three observations in each group. On the other hand if you have thousands of observations then there is no problem if one group has 687 observations and the other has 3562 observations.

Thank you very much for your answer Dr. Timothy A Ebert. It was really helpful.

In my case, I have 10 replicates per treatment, whereas 2 of them have only 9 replicates. So, I guess it won't be a big issue.

...Otherwise, maybe I'll just drop the 10º replicate from the other treatments and make them all even.

Again, thank you very much.