Parameter estimates for standardized variables modeled for cormorant presence and positive count values at roosting locations in the Mississippi Delta.

The presence model was done using generalized linear mixed effects model with a binomial distribution based on cormorant presenece and absence. The count model was done using a generalized linear mixed effects model with negative binomial regression on counts greater than zero. For both models roost ID was set as a random effect. Data was collected at multiple time points over the winter months (October–April) from 1996–2010, and 2015–2017.