For both methods, landscape diversity and edge density best described differences in the indices and had positive estimated effects on the relative fox abundance. The predicted relative abundances were similar between the two methods, suggesting that the two indices were closely related to the true abundance of red foxes. Further, we compared the costs and the precision of the two field methods. We compared the estimated regression coefficients of the landscape parameters and the predicted abundance of the two methods. Applying negative binomial regression models, we identified landscape parameters that influence red fox abundance, which we then used to predict relative red fox abundance. We used data obtained from camera traps and feces counts to independently estimate relative abundance of red foxes in the Black Forest, a forested landscape in southern Germany. Using two different field methods and comparing the results can help to detect possible bias or a non monotonic relationship between the index and the true abundance, providing more reliable results. I have also changed the method in the bam model from REML to GCV.Due to time and financial constraints indices are often used to obtain landscape-scale estimates of relative species abundance. Gam_dredge<-dredge(gam_global,evaluate=TRUE,fixed=c('offset(offset)','s(StationID,bs='re)'),rank='AIC') Gam_dredge<-dredge(gam_global,evaluate=TRUE,fixed=c('offset(offset)','s(StationID)'),rank='AIC') I tried the following two approaches with no success: #genertric random effect name as shown in model summary output I also need to retain the offset term in all models.Īfter conducting a web search, I tried the following based on this post 1 I am not interested in determining if the random effect is needed and need to keep it in all models as the random effect. When I run the dredge function on the global model the random effect is dropped from a subset of the models as dredge is running through the different model options with and without certain variables. The random effect is s(StationID, bs = 're') gam_global<-bam(Numberpertow ~ s(interval,k=6)+Stratum+ClosArea+CruiseID+s(interval, by=CruiseID,k=6)+offset(offset)+s(StationID, bs = 're'),data=l.data,method = 'REML',family=nb(),na.action = 'na.fail') The global model using bam from the package mgcv is below. I am using the dredge function from the MuMin package for a gam with a random effect:

Mumin dredge report se of coefficients