Is there minimum sample size for each group in 1-way ANOVA? if yes, please give a reference(s), If no, why there is not?
Technically speaking, there should be two or more observations in at least one cell (treatment). I.e., if there are n cells, you can perform analysis with (n + 1) observations. Moreover, if you use Poisson regression, it's enough to have just one observation per cell.