Steenhuizen seed dataset

Please find attached a dataset for analysing seed set in Protea amplexicaulos.
 
Sheet 1: 
Three flowerheads used per plant. Each flowerhead is either unmanipulated (control), selfed (receives self-pollen), or crossed (receives cross pollen). (3 treatments)
Protea flowerheads are made of numerous small flowers each with one ovule that can set a seed (florets).
Seed set is measured as number of fertile seeds per total number of florets.
I would normally run a GEE with negative binomial distribution and logit-link transformation with plant as the subject and treatment as within subject. (If using a GZLM, I would choose Wald statistics). I would also use the model to generate means and (unequal) standard errors and backtransform these for graphing.
 
Sheet 2:
Similar, just two treatments.
 
Sandy