Comparing count data and cultivars in R

I'm trying to analyse an experiment with this setup:

A plant species with 16 different cultivars of 4-5 plants per cultivar (completely randomised within greenhouse, no blocks)
An insect species that was allowed to spread freely for 4 days on the plants.
The insect number per plant was counted.
So I have cultivar and insect number as variables and want to know if the cultivars host significantly different insect numbers.

My idea is to use a GLM with family quasi-poisson followed by an ANOVA (or better Analysis of Deviance) and a post-hoc test (which is possible with GLM as far as I know).

Anyway I use R and my approach seems to suck (not really experienced with R so far).
My code is very simple:

fit1<-glm(numberinsects ~ cultivar, data=mydata, family=quasipoisson())
anova(fit1, test="F")

summary(glht(fit1, mcp(cultivar="Tukey")))

Two problems occur:
The GLM works, but I'm not sure if it's correct since the cultivars are coded as 1,2,3 etc and might be estimated as numeric variable and not a factor (I guess they should be) and so the results might be just an artefact. Not sure if this data-type issue is really a problem.

Second, the post-hoc test ends up with an error:
"Error in mcp2matrix(model, linfct = linfct) :
Variable(s) ‘cultivar’ of class ‘integer’ is/are not contained as a factor in ‘model’.
(if I change the variable type to numeric the same occurs).

Can anybody help?
Thank you very much!
Last edited: