Fixed or Random?

I am trying to make a model to see the variation of a variable (preferred temperature) between seasons. Two different treatments have been made (different rise of the temperature per time). The experiments have been repeated for 6 years. It has not been made both treatments for every year, and there is not every season for every Year. For that I used the lme function in nlme package.

As I want to know if there is any difference between seasons and not between treatments (that I know there are), my doubt is if I should include the treatment (rate) as a fixed or as a random effect as below.

modseason<-lme(Temp~Season+rate, random=~1|Year, data=data)

modseason2<-lme(Temp~Season, random=list(~1|Year,~1|rate), data=data)