I struggle with running a GLM with abundance data (continuous, zero to 24) in R. I've tried various error structures (Poisson, Gamma) with appropriate links (log, inverse), but it seems to never work. I receive many warnings, but the GLM goes through. I see a number of significant correlations, but for a backward model selection, I need the AIC, which is displayed as "inf" in all cases. What do I mean? Regardless of whether I include one or five explanatory variables (most of them are continuous, one of them is categorial), I do not get an AIC needed for model selection. Can anybody help?

The code looks like this:

m_Cyc<-glm(data=Cyc,Abundance_ind_m3~
                Year+Season+MEI+MEI.1+MEI.2+NPGO+NPGO.1+NPGO.2+Distance_land_km+
                Bottom_depth_m,family=Gamma(link="inverse"))
summary(m_Cyc)

The warning that I receive is the following:

Error: no valid set of coefficients has been found: please supply starting values.
In addition: Warning message: In log(ifelse(y == 0, 1, y/mu)) : NaNs produced