Road bias for deer density estimates at 2 national parks in Maryland

Abstract

Estimating the population density of deer is an essential task for public agencies that plan a herd reduction. Distance sampling has been increasingly utilized to estimate population density, and is used by the National Park Service to estimate white‐tailed deer (Odocoileus virginianus) densities throughout the eastern United States. Many of these surveys are conducted along public roads due to limited resources and accessibility, which may violate a critical assumption of distance sampling and potentially introduce sampling bias. We used infrared cameras to confirm deer activity with respect to survey roads at 2 national parks in Maryland, USA (Catoctin National Park and Antietam National Historic Battlefield), during 2005 and 2006 and compared results with the predicted distributions. The number of deer observed during road surveys declined with distance intervals at Catoctin, but there was a similar amount of deer activity at each distance interval. At Antietam, survey observations maintained a constant level of activity beyond 200 m from the survey route, while deer activity was inconsistent between distance intervals. The mean number of deer photographs/day/sample point did vary significantly across distance intervals from the survey route at Antietam, but not at Catoctin. In Antietam, the uneven distribution of agricultural fields and public roads were significant predictors of deer activity detected during the camera surveys. At Catoctin, the fit of the detection function was improved by expanding the first distance interval. Although density estimation using DISTANCE can account for most sources of error introduced by use of public roads, our data indicate bias is likely to occur in landscapes with high road densities and long sight distances. © 2011 The Wildlife Society.

Using spotlight observations to predict resource selection and abundance for white‐tailed deer

“…Distance sampling estimates may be biased when the detection probability is variable across space and time. Detection probability in our study was estimated to be <1, which is consistent with other spotlighting efforts (Collier et al , McShea et al ). LaRue et al () and Beaver et al () reported that distance sampling overestimated density for deer using spotlight observations in comparison with other survey designs, which is consistent with our results.…”

“…Past efforts have evaluated environmental factors (e.g., weather, seasonal, land cover types) influencing spotlight counts (Progulske and Duerre , Rybarczyk , McCullough ), the accuracy of counts, and the precision of population indices (McCullough , Fafarman and DeYoung , Cypher , Whipple et al ). Although spotlight methods are capable of producing reliable indices of relative abundance for ungulates in some landscapes (Focardi et al , Garel et al , McShea et al , Anderson et al , Corlatti et al ), the effectiveness of spotlight counts to produce reasonable and precise estimates of density has remained inconsistent because of bias associated with study design, environment, or animal behavior (Focardi et al , McShea et al , Anderson et al ).…”

“…First, spotlight surveys are most often conducted at local scales (e.g., a forest or park property) along non‐linear road‐based transects (Naugle et al , Focardi et al , Roberts et al , Stapp and Guttilla , Collier et al ). The use of non‐linear roads may violate assumptions for random or systematic sampling in distance sampling studies, and may further bias detection rates and density estimates if roads follow landscape features that structure animal populations (McShea et al , Beaver et al ) or animals select for or against features created by road openings (e.g., in highly forested landscapes; McCaffery and Creed , Ward et al ). Second, observation rates for deer across landscapes are rarely constant and may vary according to factors (e.g., observer, topography, land cover type; Corlatti et al ).…”

Distance Sampling: Methods and Applications

“…Further, roads are unlikely to be placed independently of topography (they avoid rugged terrain) or habitat (they favour dry ground), which may create larger-scale differences in density between areas near roads and areas far from roads. A recent study by McShea et al (2011) suggests bias in estimates of deer density can arise from non-random transect placement. Track-based surveys may need supplementary data, e.g.…”

“…4.1). The assumption of representativeness needs to be tested by the investigator as demonstrated by Erxleben et al (2011);Hilário et al (2012); McShea et al (2011).…”