Ieno, E.N., Zuur, A.F., Fuller, R.A., Piersma, T. & Saveliev, A.A. 2014. Poisson GAMM applied on ruddy turnstone data. Pp. 99-114 in: Zuur, A.F., Saveliev, A.A. & Ieno, E.N. (eds) A Beginner’s Guide to Generalised Additive Mixed Models with R. Highland Statistics, Newburgh, United Kingdom.
In their paper, Fuller et al. (2013) compared vigilance in habitat types that differed greatly in prey abundance and proximity to cover from which predators could launch surprise attacks. Results revealed that foragers formed larger and denser flocks in habitats closer to cover. Individual vigilance of foragers in all habitats declined with increasing flock size and increased with flock density. Nevertheless, vigilance by foragers in habitats closer to cover was always higher for a given flock size than vigilance by foragers in habitats further from cover, and habitat remained an important predictor of vigilance when comparing competing models including a range of potential confounding variables. Using these data as illustration, we discuss procedures of a generalised additive mixed model (GAMM) with a Poisson distribution. The results that we obtain in our analysis are similar to those presented by Fuller and collaborators, despite the fact that they used linear mixed effects models on log-transformed data.