How do models deal with Date covariates? Do they just convert to a number? Why not use year as a covariate?
Consider demonstrating using elevation and other landscape metrics via landscapemetrics package
Shouldn't validation be done on calibrated predictions
Partial dependence is uncalibrated
2.5 vs. 3 km prediction surface
sampling plot, density instead of points
change experience observer to expert
why no year covariate in occupancy
need to add day_of_month to occupancy
subsampling, then split 80/20?

exploration suggested the potential need for a quadratic effect of day_of_month,

but preliminary explorations shows that a quadratic (i.e., `I(day_of_month^2)`)

was not needed: the top models produced using dredge to examine all possible

combinations of predictor variables rarely contained this quadratic term, and

its effect was unstable as its regression coefficient differed substantially

between the global model and the model-averaged value.

specify detection model