Abstract
The parameters in the B and C matrices represent potential interactions between functional groups or between an external driver and a functional group. All interactions were allowed to occur if the data supported the parameter. Model selection was conducted by iteratively fitting every possible combination of the B-matrix and selecting the model with the lowest AICc (Ives et al. 2003; Viscido and Holmes, 2010; Holmes and Ward 2012). The most reduced model possible would contain only density-dependent terms (diagonal of the B-matrix) while the full model had an estimated parameter for every interaction possible in the B-matrix. Parameters were estimated for the complete matrix of external drivers (C-matrix) in every model fit due to computing constraints. A single survey would require over a million runs to work through every combination of the B and C matrices, which was simply not feasible.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
