Effects of Social and Environmental Characteristics on Coyote Human-Tolerance Behavior

Background

As urbanization expands and more wildlife adapt to living in these environments, managing human-wildlife interactions is increasingly important. Wildlife's tolerance for humans i.e., their willingness to overlap with humans spatially and temporally, is a key factor mediating the frequency and nature of an animal's interactions with humans. While intrinsic processes influencing human tolerance have been identified, little is known about the extrinsic causes. We explored the effect of urban environmental features in coyotes, a species of particular management concern due to their status as a relatively large predator. Using the GPS data of previously collared animals, we estimated human-tolerance behavior by quantifying their spatial and temporal overlap with humans and estimated the effect of environmental characteristics on human-tolerance behavior. Importantly, included human socioeconomic and demographic characteristics because of the strong relationship between these characteristics and aspects of the urban environment and human behavior important to coyote ecology.

Results

We found that environmental characteristics -- proportion of natural habitat, disturbed habitat, and agriculture -- had a negative effect on human-tolerance behavior. Median income had a positive effect on human-tolerance behavior. This study reinforces existing research indicating that suitable habitat is key for supporting sustainable wildlife populations and coexistence with humans and adds to a growing literature showing human social system attributes are important drivers of urban wildlife ecology.

Figure 1. Predicted effects of natural habitat (a), disturbed habitat (b), agriculture (c), median income (d), and proportion white residents (e) on human-tolerance behavior, i.e., relative selection strength at different values of human population density. Model predictions were generated using high (mean + sd) and low (mean - sd) values of the focal social or environmental variable. Variables not included in the focal interaction were set to their mean. Shading is 95% CI.

Files

hmm_behavior_states.R

Subset coyote GPS data to periods of high human activity and identify bursts of movement behavior using a hidden Markov model.

Data

• "gps_data.rdata": contains the GPS coordinates (long/lat) of multiple animals identified by their IDs, i.e. "COY_ID"

step_selection_analysis.R

Estimate human-tolerance behavior by estimating selection for human population density with a conditional logistic regression. Model includes interaction terms between environmental and socio-demographic characteristics and population density to estimate their effect on human-tolerance behavior.

Data

• "gps_ssf.rdata": output from hmm_behavior_states.R which includes movement data from periods of high human activity
• "raster.grd": raster containing environmental and socio-demographic geospatial data (see: https://github.com/zepedae/social-environmental-raster.git)

plot_results.R

Create interaction plots to visualize effects of environmental and socioeconomic/ demographic characteristics on human-tolerance behavior.

Data

• "ssf_data.rdata": GPS data and random location generated in step_selection_analysis.R
• "ssf_fit.rdata": conditional logistic regression model built in step_selection_analysis.R