### Experimental Design and Social Covariates
We evaluated all community parks throughout the Phoenix-metro valley and determined which parks fell within one of four categories (High-Income, Low-Income, High-Latinx, Low-Latinx). We began the site selection process using the 2014 shapefile of parks and open spaces from the Maricopa Association of Governments, found and curated in the Arizona State University Geospatial Repository. We filtered out all parks that were not classified as a “community park”, which are parks that were not open to the public and managed by public funds. We also excluded desert parks and preserves and sites that were within 2 kilometers of such parks. We obtained census block data for Maricopa County from the American Community Survey (2017) and used to calculate the average median household income and percentage of Latinx residents within a 1-kilometer buffer radius of each community park. Within ArcMap utilizing the Buffer, Clip, Intersect, Dissolve, and Field Calculator tools, we created a 1km buffer zone around each camera site and calculated the weighted shape area, median household income, and percentage of Latinx residents within each census block. Each census block group was dissolved, and the average income and ethnicities were calculated by averaging the weighted values for each category
We calculated the top and bottom quartile for income and ethnicity to determine if a park would be classified as a High-Income, Low-Income, High-Latinx, or Low-Latinx site independently, if they fell into the top or bottom quartile for income, but not ethnicity and vice versa. Once each category was assessed, Low-Latinx parks were considered up to 15% Latinx residents, and Low-Income parks were considered up to $48,000 median household income/year. These conservative increases were done to increase the sample size of parks used for camera placement. All parks that had more than one standard deviation above the mean for non-white or non-Latinx residents were excluded as well, resulting in a gradient of Latinx residents and average median household income of residents ranging from 0.046%-0.777% and $25,835-$105171 respectively. These values will also be used as social covariates within all future models.
### Camera methods
We deployed an array of 28 non-baited motion activated cameras throughout neighborhood parks during Summer (June-Aug) 2021through Fall and into Winter (Sep-Dec) 2021. Sites that fell within the four parameters (High-Income, Low-Income, High-Latinx, Low-Latinx) were evaluated through google earth and in person, in which parks that were logistically challenging to access, were not approved by the city, or a dog park were excluded. Parks that had poor probability of wildlife detection or did not have a tree to secure a camera to were excluded as well. Ideal parks were those with tree cover or a wash within a neighborhood’s greenspace or a neighborhood park. Additionally, evidence of wildlife such as animal tracks, trails, or scat were used to place the camera within a site to maximize potential to detect mammals within the site. (Shannon, Lewis, and Gerber 2014; Burton et al. 2015). Each camera was placed on a tree approximately knee height at low sensor levels, two images captured per trigger at 1 second intervals, and with a 2-minute delay between a trigger sequence. The distance between each camera was at least 1 kilometer. Each camera was maintained by a team of undergraduates and checked bi-weekly to ensure proper functioning, collect photos, and to repair any damages to cameras. After retrieval of images, we identified the species within each photo utilizing the Colorado Parks and Wildlife Photo Warehouse database through the Fall and Winter of 2021 (Ivan and Newkirk 2016). Each photo was identified to species level by two observers and verified by a third “referee” observer if necessary.
### Citations
- Burton, A. Cole, Eric Neilson, Dario Moreira, Andrew Ladle, Robin Steenweg, Jason T. Fisher, Erin Bayne, and Stan Boutin. 2015. “REVIEW: Wildlife Camera Trapping: A Review and Recommendations for Linking Surveys to Ecological Processes.” Edited by Phil Stephens. Journal of Applied Ecology 52 (3): 675–85. https://doi.org/10.1111/1365-2664.12432.
- Ivan, J.S. and Newkirk, E.S. (2016), Cpw Photo Warehouse: a custom database to facilitate archiving, identifying, summarizing and managing photo data collected from camera traps. Methods Ecol Evol, 7: 499-504. https://doi.org/10.1111/2041-210X.12503
