Sparse camera-trap data

[tng]

I have very sparse camera trap data and am struggling with the best way to input it into Presence. I sampled 38 sites over the course of a year, but had only up to 5 cameras active at any given time. The average number of nights sampled was 30. Is it best to organize the data by Julian day (with many blanks), or treat "sample 1" as the first day that each site was sampled, regardless of what day it actually began? The first method results in 90% "no data" cells; the second method, 63%. In either case, I have season, but not Julian day, as a sampling covariate.

[darryl]

It shouldn't matter, you should get the same results either way if you define the models appropriately for the 2 different setups. The only difference is that entering the data in terms of survey 1, 2, 3, etc (rather than Julian date) should run faster because the total number of columns is smaller.

Do you think it's reasonable to assume that there's been no systematic changes in what your species has been doing over a 12 month period? That is, they've been using the same sort of areas the whole time and not responding to any changes in the climatic seasons?

Cheers
Darryl

[Bart]

So am I correct in thinking when assuming no temporal variability, (lets say all data were collected in one short season) it doesn't matter whether data collected for two sites on the same day are represented under different columns when entering in Presence? The reason I am asking is that Presence seems unable to compute my data with a lot of empty cells, (I posted about this issue in a different thread), but when I organize it like tng suggested Presence is able to deal with it, as long as I put the station with most records in the first line.

[darryl]

Provided you don't have any mistakes in your data file, PRESENCE should handle it in either format

[tng]

Hi Darryl,

Thanks for the response. And yes, my assumption is that the only climatic/temporal thing the animals are responding is wet versus dry season, which I have listed as a separate sampling variable.