Comparing coyote (Canis latrans) density estimates from camera traps and genetic spatial capture recapture

<p>Density estimates assist conservation and management decisions. Unfortunately, management of elusive mesocarnivores occurring at low densities has often relied on qualitative metrics or population indices. Proliferation of camera trap sampling of wildlife over the last 2 decades provides an opportunity to apply recently developed statistical methods for estimating density of unmarked animals including mesocarnivores. Our goals were to estimate coyote (Canis latrans) density with camera-trap distance sampling (CTDS) and compare it to an established method, genetic spatial capture-recapture (SCR), within three large regions of Indiana, USA. Based on previous research, we predicted smaller coyote densities in regions heavily modified by human land use (i.e., agriculture and development). From 2019 to 2021, we deployed >1000 camera traps and sampled > 900 km of transect for coyote scat. Using pairwise differences in bootstrapped density estimates from CTDS, we detected lower coyote densities in landscapes experiencing more agriculture and development, supporting our prediction. Coyote density estimates from CTDS in each RMU covaried inversely with contemporaneously estimated ratios of young white-tailed deer:adult female deer (Odocoileus virginianus). Despite our large sample size of camera traps, the precision of our coyote density estimates was mediocre to poor (coefficients of variation range: 0.24 to 0.34), which reduced power to detect changes in density across all the regions. Incorporation of spatial covariates holds promise for improved precision of camera-based estimates of density. Precision tended to be even worse for genetic SCR estimates (range: 0.18 to 0.48) and thus failed to reveal differences in regional density despite yielding point estimates with rankings identical to CTDS estimates. Future managers should consider CTDS in situations where individual identity cannot be accurately determined but management actions based on estimates of density or abundance are desired.</p>

<p> </p>

<p>The "Coyote_CTDS_PURR.csv" file contains data used to estimate coyote density via camera-trap distance sampling. The following describe each column within the "Coyote_CTDS_PURR.csv" file:</p>

<p>"Sample.Label" = the unique identifier given to each camera location.</p>

<p>"Date" = the date of the coyote detection (month/day/year).</p>

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

<p>"RMU" = the Regional Management Unit the camera was deplyed in.</p>

<p>"TL" = the 2x2 mi area the camera was deplyed in.</p>

<p>"Region.Label" = the broader area the camera as deployed within that we are interested in estimated density within. Designates whether the camera was deployed in concealed or open areas as well as the Regional Management Unit the camera was deployed within. </p>

<p>"Effort" = the spatiotemporal effort of the camera sampling at each camera location. </p>

<p>"veg" = Whether the camera was deployed in concealed (forest/wetland) or open (grass/agriculture) areas.</p>

<p>"Area" = the area (km^2) of the Region.Label (above) that we are interesting in estimating density within. </p>

<p> </p>

<p>The "data_SCR_multi.RData" file contains data used to estimate coyote density via genetic spatial capture-recapture. The data within "data_SCR_multi.RData" is structured as a list of 6 elements containing a collection of data for each session (site and year):</p>

<p>    1: List of data for Regional Management Unit 3 during 2020</p>

<p>    2: List of data for Regional Management Unit 4 during 2020</p>

<p>    3: List of data for Regional Management Unit 9 during 2020</p>

<p>    4: List of data for Regional Management Unit 3 during 2021</p>

<p>    5: List of data for Regional Management Unit 4 during 2021</p>

<p>    6: List of data for Regional Management Unit 9 during 2021</p>

<p>Within each of these 6 elements, the following describes the columns (each element contains the same columns):</p>

<p>"RMU" = the Regional Management Unit the samples were collected within</p>

<p>"Year" = the year samples during the session were collected</p>

<p>"n" = the number of unique individual coyotes captured during the session</p>

<p>"y" = a matrix with the frequency each unique individual coyote was encountered at each trap during the session</p>

<p>"X" = a named matrix of trap locations in KM. Columns are named "Easting_km" and "Northing_km"</p>

<p>"K1D" = a vector with the number of occasions every trap was operable</p>

<p>"K" = the maximum number of occasions the traps were operable during the session</p>

<p>"poly" = a matrix of the UTM x- and y-coordinates representing the polygon vertices in KM</p>