I would suggest starting with some exploration of the data. Something like this:
library(ggplot2)
ggplot(data = your_data, mapping=aes(x=factor(TRAT), y=GER)) +
geom_boxplot() +
facet_grid(~factor(ACESSOS))
your_data needs to be a dataframe and I'm assuming that the categorical variables are strings so I cast them to factors in the ggplot function call.