Reducing bias in density estimates for unmarked populations that exhibit reactive behavior towards camera traps

<p>Density estimates guide wildlife management, and camera traps are commonly used to estimate sizes of unmarked populations. Unfortunately, animals often alter their natural behavior in the presence of camera traps, which may bias subsequent density estimates. We simulated populations of known density to test several new and existing methods that aimed to reduce bias in density estimates from camera-trap distance sampling (CTDS) and the random encounter model (REM). Within our simulated populations, we introduced different behavioral reactions including attraction towards cameras, freezing when near cameras, and fleeing from cameras. CTDS and REM provided statistically unbiased density estimates for simulated populations with no reactive behavior. However, failure to implement a method to account for reactive behavior resulted in statistically biased density estimates when 30% of the simulated population potentially reacted by attraction to or fleeing from camera traps. We identified corrective strategies that produced confidence intervals which overlapped truth for every behavioral reaction except when individuals could flee before being detected by cameras. We provide empirically tested methods for reducing bias of density estimates. Wildlife managers requiring population estimates of animals that exhibit reactive behavior can use CTDS and REM in conjunction with our methods to reduce bias in density estimates. We encourage future studies to quantify behavioral responses to camera traps and to implement, test, and possibly extend our methods to reduce bias.</p>

<p>Several archived files each correspond to an .RData file that contains a list of 100 data frames containing detections of simulated movement paths and detections at camera traps.</p>

<p>The following denotes which composed populations are in each .RData file:</p>

<p>"Atraction10_Populations.RData" = camera-trap detections of simulated populations containing 90 animals, 10% (i.e., 9 animals) of which could potentially exhibit attraction towards camera traps. </p>

<p>"Atraction30_Populations.RData" = camera-trap detections of simulated populations containing 90 animals, 30% (i.e., 9 animals) of which could potentially exhibit attraction towards camera traps. </p>

<p>"combination10_pops.RData" = camera-trap detections of simulated populations containing 90 animals, 10% (i.e., 9 animals) of which could potentially attracted, 10% (i.e., 9 animals) of which could potentially flee if detected, and 10% (i.e., 9 animals) of which could potentially freeze. </p>

<p>"combination25_pops.RData" = camera-trap detections of simulated populations containing 90 animals, ~25% (i.e., 23 animals) of which could potentially attracted, ~25% (i.e., 23 animals) of which could potentially flee if detected, and ~25% (i.e., 23 animals) of which could potentially freeze. </p>

<p>"FleeIfDet10_Populations.RData" = camera-trap detections of simulated populations containing 90 animals, 10% (i.e., 9 animals) of which could potentially flee if detected by the camera trap. </p>

<p>"FleeIfDet30_Populations.RData" = camera-trap detections of simulated populations containing 90 animals, 30% (i.e., 9 animals) of which could potentially flee if detected by the camera trap. </p>

<p>"Flee_Regardless_Of_Detection10_Populations.RData" = camera-trap detections of simulated populations containing 90 animals, 10% (i.e., 9 animals) of which could potentially flee regardless of being detected by the camera trap. </p>

<p>"Flee_Regardless_Of_Detection30_Populations.RData" = camera-trap detections of simulated populations containing 90 animals, 30% (i.e., 9 animals) of which could potentially flee regardless of being detected by the camera trap. </p>

<p>"Freezing10_Populations.RData" = camera-trap detections of simulated populations containing 90 animals, 10% (i.e., 9 animals) of which could potentially freeze when near a camera trap. </p>

<p>"Freezing30_Populations.RData" = camera-trap detections of simulated populations containing 90 animals, 30% (i.e., 9 animals) of which could potentially freeze when near a camera trap. </p>

<p>"Nonreactive_Populations.RData" = camera-trap detections of simulated populations containing 90 animals, all of which never exhibited any reactive behavior towards camera traps. </p>

<p> </p>

<p>The following defines the columns found in each file:</p>

<p>"distance" = the distance between the detected animal and camera trap.</p>

<p>"x" = the x coordinate across the simulated landscape of where the animal was standing when detected. </p>

<p>"y" = the y coordinate across the simulated landscape of where the animal was standing when detected. </p>

<p>"Reaction" = denotes the reaction of the animal when detected. </p>

<p>"AnimalID" = the unique ID given to each simulated animal. </p>

<p>"DateTime" = the time step (unix time format). </p>

<p>"Sample.Label" = the unique identifier given to each camera trap simulated on the landscape. </p>

<p>"Detected" = denotes that the animal was detected. </p>

<p>"Area" = the area (km^2) of the simulated landscape. </p>

<p>"Effort" = the spatiotemporal sampling effort of the simulated camera trap. </p>

<p>"Region.Label" = a mandatory column used in the Distance package that names the region the analysis was being conducted. </p>

<p> </p>

<p> </p>