Measuring the benefit of a risk-induced trait response: vigilance and survival probability
Data files
Jul 11, 2024 version files 80.14 KB
-
README.md
2.56 KB
-
surv_DF_vigCov.csv
1.04 KB
-
Vigilance_Data.csv
76.54 KB
Abstract
Defensive traits are hypothesized to benefit prey by reducing predation risk from a focal predator but come at a cost to the fitness of the prey. Variation in the expression of defensive traits is seen among individuals within the same population, and in the same individual in response to changes in the environment (i.e., phenotypically plastic responses). It is the relative magnitude of the cost and benefit of the defensive trait which underlies the defensive trait expression and its consequences to the community. However, whereas the cost has received much attention in ecological research, the benefit is seldom examined. Even in a defensive trait as extensively studied as vigilance, there are few studies of the purported benefit of the behavior, namely that vigilance enhances survival. We examined if prey vigilance increased survival and quantified that benefit in a natural system, with white-tailed deer (Odocoileus virginianus) experiencing unmanipulated levels of predation risk from Florida panther (Puma concolor coryi). Deer that spent more time vigilant (as measured by head position using camera trap data) had a higher probability of survival. Indeed, an individual deer that were vigilant 75% of the time were more than three times as likely to be killed by panthers over the course of a year compared to a deer that were vigilant 95% of the time. Our results therefore show that within-population variation in the expression of a defensive trait has profound consequences to the benefit it confers. Our results provide empirical evidence supporting a long-held but seldom tested hypothesis, that vigilance is a behavior that reduces the probability of predation and quantified the benefit of this defensive trait. Our work furthers an understanding of the net effects of a trait on prey fitness and predator-prey interactions, within-population variation in traits, and predation risk effects.
https://doi.org/10.5061/dryad.2jm63xszb
Two datasets were used for this paper. The first dataset is vigilance data captured from the three camera grids deployed for the associated study. The second dataset is the survival data for the same deer captured on camera, taken from the telemetry tracking data.
Description of the data and file structures
VIGILANCE DATA
The first dataset (file name “Vigilance_Data.csv”) contains the vigilance data for the deer used for this study. This data was used to derive a vigilance index for each unique individual deer in the study. The first column (camID) identifies the camera location associated with that data point. The second column (did) gives the name of the unique identifier for an individual deer. The third column (DT_POSIX) provides the date and time of the photo. The fourth column (grpsize) supplies the total number of deer in the photo, i.e. group size. The fifth column (day) quantifies if the photo was taken during daylight hours (1) or at night (0). The six column (season) identifies the hydrologic season associated with the photo, either the wet season or dry season. The seventh column (beh) provides the vigilance behavior for the uniquely identified deer in the photo. “1” denotes vigilant behavior and “0” denotes non-vigilant behavior. The eighth column (JD) gives the days since the beginning of the study. The ninth column identifies the trail status of the camera location, i.e. on trail (1) or off trail (0). Finally, the tenth column identifies the camera grid associated with that particular camera location. “A” is the Addlands camera grid, “B” is Bear Island camera grid, and “F” is Florida Panther National Wildlife Refuge camera grid. The data provided is the result of thinning the photos of identifiable collared deer by 5 minute intervals to give intendent behavioral observations.
SURVIVAL DATA
The second dataset (file name “surv_DF_vigCov.csv”) contains the survival data frame for the deer used in this study. This data was used to test the effect of vigilance on survival. The first column (did) gives the name of the unique identifier for an individual deer. The second column (entry) gives the day that the individual entered the study. The third column (exit) gives the day that the individual exited the study. the fourth column (vigCov) gives the relative vigilance index derived from the vigilance dataset above.
Behavioral observations of deer were made with remote cameras, and survival data were collected using GPS collars deployed on the same deer. We captured and collared white-tailed deer between December 20th, 2014 and March 5th, 2017 via aerial helicopter net-gunning, rocket netting, or chemical immobilization via darting. We ear-tagged individual deer with a unique identifier and further marked collars with a unique combination of collar tags, symbols, and numbers to facilitate identification. If we received a mortality signal or suspected a mortality from visually inspecting daily movement data, we attempted to retrieve the collar and perform a field investigation within 24 hours. We assessed cause of death using field evidence gathered during investigations. We deployed 180 unbaited remote-sensing cameras across three grids from January 2015 to December 2017. Each 29-square-kilometer grid was separated by at least 13 kilometers and contained 60 cameras. Within each grid, we placed 40 cameras on human-made trails and we refer to these cameras as ‘on-trail’. On-trail cameras were placed approximately 700 meters apart. We placed the remaining 20 cameras on wildlife corridors or habitat edges, and we refer to these cameras as ‘off-trail’. Each of the 20 off-trail cameras were paired with an on-trail camera, and each off-trail camera was positioned 250 meters from any on-trail camera. We positioned all cameras 0.3 m off the ground and placed cameras to maximize wildlife detections. We visited cameras monthly for maintenance, data retrieval, and vegetation clearing. We categorized the vigilance behavior of female deer photographed by remote-sensing cameras. We examined images for the presence of a collared deer, and when possible, identified the individual. We categorized deer as vigilant or foraging based on the posture of the focal deer such that deer with the end of the rostrum above the proximal end of the metacarpal bone were classified as ‘vigilant’ and below were classified as ‘foraging’. Finally, to ensure independence of observations, we sorted images chronologically by individual and removed images if the time from the previous image was less than five minutes. To account for the gradient of environmental variables across camera locations, we created a metric to measure the vigilance of each individual deer.