1

I wonder if anyone would like to give me some guidance (I am a beginner in R). My data comes from a forest, where they performed a conservation fire. We measured the volume of trees before the fire (2007) and after (2013 and 2016) in eight plots. I want to analyze if there is any difference in tree volume between the years, for each species.

Random effect = plot, fixed effects = species and year, response = volume.

The data is poisson distributed. One species (pine) dominates, and other species are more rare, which means that there are many zeros in the dataset. I think I can use generalised linear mixed modelling (GLMM).

I wrote the code like this:

modelglmm <- glmm(as.integer(round(Volume)) ~ Species*Year, 
                  random = list(~Plot), 
                  data = data, 
                  family.glmm = poisson.glmm, 
                  m = 10^4, 
                  varcomps.names = c("plots"))

summary(modelglmm)

And after that:

data_pine = data[data$Species == "Pine",]
modelglmmpine <- glmm(as.integer(round(Volume)) ~ Year,
                      random = list(~Plot), 
                      data = data_pine, 
                      family.glmm = poisson.glmm, 
                      m = 10^4, 
                      varcomps.names = c("plot"))

summary(modelglmmpine)

What do you think? I'm unsure about this since the data is unbalanced. Should I use a different method, or change the code? I hope I am clear enough.

Annie
  • 11
  • If "volume" is a synonym of number, I don't understand why you need to round. If it is not, I don't understand why you assume a Poisson distribution. If you have a zero inflated Poisson distribution, you should account for that. – Roland Dec 12 '16 at 12:25
  • Volume is numbers, but not integers. I thought integer was required. How do I account for the zero inflated Poisson distribution? – Annie Dec 12 '16 at 12:43
  • Yes, integer is required, but Poisson data should be integer to begin with. If it isn't it's not Poisson data. – Roland Dec 12 '16 at 13:13

0 Answers0