Scale Dependent Behavioral Responses to Human Development by a Large Predator, the Puma

Abstract

The spatial scale at which organisms respond to human activity can affect both ecological function and conservation planning. Yet little is known regarding the spatial scale at which distinct behaviors related to reproduction and survival are impacted by human interference. Here we provide a novel approach to estimating the spatial scale at which a top predator, the puma (Puma concolor), responds to human development when it is moving, feeding, communicating, and denning. We find that reproductive behaviors (communication and denning) require at least a 4× larger buffer from human development than non-reproductive behaviors (movement and feeding). In addition, pumas give a wider berth to types of human development that provide a more consistent source of human interference (neighborhoods) than they do to those in which human presence is more intermittent (arterial roads with speeds >35 mph). Neighborhoods were a deterrent to pumas regardless of behavior, while arterial roads only deterred pumas when they were communicating and denning. Female pumas were less deterred by human development than males, but they showed larger variation in their responses overall. Our behaviorally explicit approach to modeling animal response to human activity can be used as a novel tool to assess habitat quality, identify wildlife corridors, and mitigate human-wildlife conflict.

Top carnivores increase their kill rates on prey as a response to human-induced fear

“…Collars were programmed to record locations every 4 h, and location data were downloaded remotely via UHF once a month. We used a custom cluster generation algorithm integrated in the Geographical Information Systems program ARCGIS (v. 10; ESRI, 2010) using the programming languages R (v. 2.1.3.1; R Development Core Team, 2010) and PYTHON (v. 2.6; Python Software Foundation, 2010) to identify groups of locations in which each location was within 100 m of the cluster centre and 6 days of another location in the cluster (for full details on the algorithm, see [10]). We field-investigated clusters in reverse chronological order from their time of formation using the most recently downloaded GPS data.…”

“…Owing to the discrete nature of housing count data, each housing class was rounded up to the nearest number of houses, resulting in housing classes within 150 m of a cluster to be defined as 0 houses for no housing, one house for rural, two to nine houses for exurban and greater than nine houses for suburban. We used the 150 m buffer because this is the scale of development found to most impact puma hunting in our study area [10]. We constrained the response variables to only include behaviours we expected to be associated with risk aversion, which were narrowed down to P.NIGHT, P.ACTIVE, DUR, DIST and a final measure of prey consumption time (P.C.TIME) which was calculated as P.ACTIVE Â DUR.…”

The golden age of bio‐logging: how animal‐borne sensors are advancing the frontiers of ecology

“…Although forage and predation risk were both important in determining habitat selection, resident elk were found to make fine-scale adjustments in space use that substantially reduced their risk of predation. Similarly, in order to understand how puma behaviors varied in relation to human activity, Wilmers et al (2013b) characterized behavior-specific puma habitat selection with respect to covariates associated with human impact. Specifically, they used GPS collars on pumas to identify kill sites, nursery sites, movement areas, and communication sites (male pumas create ''community scrapes'' to which they return to every 1-4 weeks to urinate).…”

Early post-release movement of reintroduced lions (Panthera leo) in Dinokeng Game Reserve, Gauteng, South Africa

“…Daily distances travelled were calculated by summing up the distances between locations in each day. Distances to release sites and buildings were averaged across each day, using only the first location when animals remained stationary for more than one location (when distance to the next location was <100 m, a distance used to identify GPS clusters of potential feeding sites of carnivores; Tambling et al 2010;Wilmers et al 2013) to prevent pseudo-replication.…”