Estimating prey activity curves using a quantitative model based on a priori distributions and predator detection data

Abstract

The impact of predators on prey activity patterns is routinely analyzed through the largely qualitative approach of comparing overlapping activity density plots. While this approach offers some insight into predator-prey dynamics, it precludes the direct estimation of a predator's impact on prey activity. We present a novel model that overcomes this shortcoming by using predator detections and an ideal prey activity curve to quantify the impact of predator activity on prey activity patterns. The model assumes that species strive to adhere to an ideal activity distribution and quantifies the degree to which a disturbance – in this case, a predator – prompts a departure from this ideal curve. We use spatially coincident camera trap records of mountain cottontail (Sylvilagus nuttallii), red fox (Vulpes vulpes), and coyote (Canis latrans) as a case study. We found that mountain cottontails limit their activity when red foxes are active, but do not alter their activity patterns to avoid coyotes. Critically, we also found that the model is sensitive to the a priori distribution used as an ideal activity curve. Therefore, preliminary testing of a priori distributions should be performed before running the model. This model improves our ability to quantify and predict predator-prey interactions as they pertain to activity patterns, but is presently limited to a single-predator system over a single active period.