I'm trying to test whether in a particular location the density of invertebrates measured now is significantly different from the one measured a decade ago. I sampled the same two plots (A and B) that where sampled in the past, at the same times in the season as in the past (Early, Mid and Late season), with n=3 replicate cores at each of three sampling site (One, Two and Three) randomly picked within each plot.
Now my question are:
I) Is a four-factor nested ANOVA with factor Year(fixed, two levels: Past and Current), factor Time(fixed, nested within Year, three levels: Early, Mid, Late), factor Plot(fixed, two levels: A, B) and factor Site(random, nested within Plot, three levels: One, Two, Three) the correct design for the test? II) After square root and even fourth root, data as a whole are still non-normal. What should I do? III) If I want to see for which of the Times within Current year the density is different from the respective Time in the Past, is a post-hoc Tuckey's test the correct procedure?