Threshold of trap nights for camera trapping

[Pentatome]

Hello,

I am working on an arboreal camera trapping project on Binturongs to identify habitat preferences. We have used 30 cameras originally setup for ~90 trap nights. However multiple cameras constantly triggered by leaves have had their battery depleted within days or even hours.

To limit the loss of power in the model I decided to arbitrarily select cameras with 10 or more trap nights, this leaves 19 out of 30 cameras, which represent 969 out of 983 total trap nights.

Do think this is a sound method?



Additionally how many covariates can I test in the model? I assume no more than three?

Thanks in advance!

[Pentatome]

I intentionally did not check whether the cameras left out had detected the species to avoid being biased.

But after inspection, none of the cameras excluded from the models had detected any binturongs (they barely detected anything, as a matter of fact).