Yes imo because PERMANOVA can be used on percentages (for example, indices like JI) since it doesn't require normally distributed data, but I think it is still best to transform your data first. For example, you could use square root transformation :)