Are multi-season multi-method occupancy models possible?

[NicolaG]

Hi

I'm fairly new to occupancy modelling and have been running a few simple single-season models with detection and occupancy covariates. However, I think I should be running a multi-method model instead, but I'm struggling with the closure assumption as my methods weren't conducted at the same time (ranging from 1-7 months apart). I can't find much on multi-season, multi-method models, does anyone know if they are possible?

This is a rather long (sorry!) description of my data if that helps, I'm also debating if multi-season is even the best route.

I have 69 transects, spread over a mosaic landscape, where I set 13 Sherman traps (25m apart) per transect for a capture-mark-recapture study on mouse lemurs. I set the traps for 6 nights over a 2-week period on each transect, simultaneously trapping on 1-6 transects in each 2-week block. This ran over 16 months, and transects were not repeated. I also set up one camera trap on each transect, usually at the mid-point of the transect. These were recording for 12 nights at each transect before being moved to new transects. I only have 66 camera trapping sites due to some cameras failing. The camera traps were not set at the same time as the Sherman traps and varied from 40 days after the Sherman trapping was conducted on a transect, up to 232 days. I have a total of 29 transects where mouse lemurs were present; for 7 of these transects, they were captured using both methods, 11 transects were captured only in Sherman traps, and a different 11 transects where they were only captured on camera traps.

I was hoping to run a multi-method model as we had presences on quite a decent amount of transects with only one of the methods. I had been running the single-season models for each method separately as the transects were far enough apart (larger than an estimated home range size) that we never caught the same mouse lemur on two transects, so I treated the 2-week trapping session (or 12 nights for camera traps) as closed. I understand that the large gap between when the methods were completed would mean that it would be an open population for some transects. I want to account for the fact it would be open by running a multi-season, multi-method model. I'm not interested in the colonisation and extinction part of the multi-season model, just want to control for it. Also, the comparison of detection rates between methods isn't my main goal either, but it could be an interesting short paper. I have four different over-arching questions based on a variety of different landscape and anthropogenic occupancy covariates. My plan was to set the data up as follows: for sites where camera traps were set up shortly after Sherman trapping, I would consider these as a single season, so season 1. For sites where camera traps were set up months after the Sherman traps, I would treat these as separate seasons, so the Sherman trap data for these transects would be season 1, and the camera trap data would be season 2. This would mean the time of year for each method wouldn't be relevant. Does anyone know if this is the correct thing to do and is it possible?

Or would running a single-season, multi-method model and removing the season 2 camera trap data be better? I have 11 transects which would be classed as two separate seasons (out of 69) and only 2 of these where they were caught on camera traps, but they were also caught in Sherman traps along these transects, so I'm not losing the presence point completely for that location.

Any advice or help would be greatly appreciated! I'm going around in circles trying to figure out what to do.

Thanks!