Logistic regression analysis. The relationship between the proportion (p) of tag transmissions detected by a receiver, tag distance from a receiver, and environmental covariates. (A) Estimates of logit(p) averaged over intervals of distance (open squares ± SE), and logit(p) modeled as a quadratic function of distance (solid line). (B) Predicted number of pings detected per 5-min interval from linear model (dashed line) vs. model-weighted distance-only function (solid line), illustrating systematic underestimation by linear regression (dashed line) of detections expected within 350 m of receiver and overestimation beyond 350 m. Logistic function converted to pings detected using mean tag ping rate. (C) Scatterplot matrix of correlations between environmental covariates. Diagonal is a histogram of values for each covariate.