Hi Sacha, A few things: 1) If counts are large, a normal error structure may be fine as an approximation, large counts means conditional on the covariates, the counts in each cell are relatively 'large'. if there are not many 0's a log transformation might be appropriate. This will be a lot easier if you want to start adding spatial covariance structure. However, without seeing your particular data it is hard to say what is an appropriate assumption. 2) Spatial Poisson models are tricky. You have to be a bit careful in the specification. In a mixed model context, ie normal random effects I suspect the problems that you get in a Poisson auto regressive model are not the same. In the Poisson spatial auto regressive model on a finite lattice you can only get negative correlation. 3) The bell shaped abundance distribution is the old idea that a single species will have a Gaussian or unimodal distribution of abundance along a continuous environmental gradient, with the mode at the optimum. The species-abundance distribution is a different beast and tend to be highly skewed, the classic distribution is the lognormal but there are many others, see Engen, or many other references. 4) You are talking about empirical modelling of species abundance in a survey, so all the theoretical mumbo-jumbo above may not be that relevant, depending upon the goals of the study and interpretation of the data. So you have to think about how the survey was conducted, and construct your model according to the data collection process, some of the theoretical constructs can be tested as part of the systematic component of your model. Hoping this helps, Nicholas
Message: 1 Date: Fri, 18 Mar 2011 14:14:12 +0100 From: Sacha Viquerat <tweedie-d at web.de> To: r-sig-ecology at r-project.org Subject: [R-sig-eco] question about appropriate model for abundance studies Message-ID: <4D835AA4.40706 at web.de> Content-Type: text/plain; charset=ISO-8859-15; format=flowed hello! I did a count survey on a tropical fish species! Well, I didnt do it, Im just helping at the statistics-stage. We recorded some water parameters alogside each catch, such as no3, no2, po4 etc. as the data are count data (and require error terms due to spatial pseudoreplication), the glm with poisson error structure should be the method of choice. however, I do fear that in doing so, I will not be able to model fish abundance correctly. In my opinion (and, as far as I remember, in the opinion of those who gave ecology classes), the abundance of species should be sort of bell shaped, since there will always be, for example, an optimal temperature, pH-level and so forth. However, I have not yet seen such a discussion arise on one of the many forums. Am I missing the obvious??