Hello everybody !
I’m actually doing a master thesis on a bird, the Corn bunting, more specifically on the effect of agri-environment schemes (AES) on the distribution of territories of the bird during the breeding season in Belgium.
But now it’s time to work on the statistical analysis and I’m not the best on this field.
First, I’ve used quadrat sampling that are separate from each other to avoid spatial autocorrelation and each quadrats sampling have been visited only once. I’ve created a matrix in which I have, for each quadrat sampling :
- the surface area of 6 categories of fields,
- the number of Corn bunting encounter (and presence/absence),
- the surface area of a number of AES.
For the analysis, I’m considering the quadrat sampling as the terrritories of the birds inside it (it can biased the results but it’s necessary because of the lack of time).
I’m now wondering several questions :
- Is a GLM adequate for this analysis or should I use a GLMM ?
- If I’m working with the abundance of Corn bunting, is the Poisson Distribution the best one to use ? Is there any parameters to set ?
- If i’m working with the presence/absence, is the Binomial Distribution a good one ? Is there any parameters to set ?
- Is it better to work on the abundance or presence/absence ? I always heard that we are loosing informations when you change from abundance to presence/absence...
- I’m using R to do this statistical analysis, is the command « step » the best one to find the best model (by selecting « both » direction) ?
Thank you for your time and your precious help !