I am trying to work out the best approach for data analysis for my PhD and have gotten a bit stuck.
I am working in the field of ecology, trying to understand responses of fauna before and after a fire disturbance. I am using a BACI methodology, and have been using camera traps as a way to capture animal detections/activity. The issue I'm having is that the burn site I'm working with was very small, so I had to make a high density camera trap grid (i.e., 6 cameras at each site, 4 monitoring sites, cameras were spaced approximately 20 metres apart, so very close). I have so far accounted for animals occurring at the same camera within a 60 minute period (standard thing to do with camera trapping, to account for individuals who might captured multiple times on the camera in 60 minutes), but I am struggling to address the spatial autocorrelation between cameras at each site - i.e., the reality that animals probably moved between cameras within a similar time period of each other. I have read that I could treat each site as a replicate, rather than each camera within each site, so that even if there are animals moving between the cameras, I doesn't matter that they are being counted across multiple cameras within a 60 minute period. Does that sound legitimate?
Does anyone have any ideas of ways to address this through my analysis? For context, I am planning o using GLMMs (again, something lots of other people have used), and thought maybe I could use camera number as random effect? Wanting to see if anyone else has any ideas I could explore.
Thanks
