Measuring wildlife responses to anthropogenic activities often requires long-term, large-scale datasets that are difficult to collect. This is particularly true for rare or cryptic species, which includes many mammalian carnivores. Citizen science, in which members of the public participate in scientific work, can facilitate collection of large datasets while increasing public awareness of wildlife research and conservation. Hunters provide unique benefits for citizen science given their knowledge and interest in outdoor activities. We examined how anthropogenic changes to land cover impacted relative abundance of two sympatric canids, coyote (Canis latrans) and red fox (Vulpes vulpes) at a large spatial scale. In order to assess how land cover affected canids at this scale, we used citizen science data from bow hunter sighting logs collected throughout New York State, USA, during 2004–2017. We found that the two species had contrasting responses to development, with red foxes positively correlated and coyotes negatively correlated with the percentage of low-density development. Red foxes also responded positively to agriculture, but less so when agricultural habitat was fragmented. Agriculture provides food and denning resources for red foxes, whereas coyotes may select forested areas for denning. Though coyotes and red foxes compete in areas of sympatry, we did not find a relationship between species abundance, likely a consequence of the coarse spatial resolution used. Red foxes may be able to coexist with coyotes by altering their diets and habitat use, or by maintaining territories in small areas between coyote territories. Our study shows the value of citizen science, and particularly hunters, in collection of long-term data across large areas (i.e., the entire state of New York) that otherwise would unlikely be obtained.

This dataset "canid_data.csv" represents hunter observations of coyote (Canis latrans) and red fox (Vulpes vulpes) collected yearly from 2004-2017 across the state of New York, USA as part of the New York Bowhunter Sighting Log. We calculated the sum of counts of each species by wildlife management unit (WMU) and year. These counts are in dataset columns "coyote" and "redfox". The total number of hunter observation hours (in units of 100 hours) in each WMU and year is in column "hours". Columns "ag", "urbanL", and "para_mn" represent percent agriculture, percent low-density urban, and agriculture perimeter-area ratio for each WMU and year, derived from the National Land Cover Database (NLCD). The values in these columns were scaled for analysis so they had a mean of 0 and a standard deviation of 1. The following three columns "ag_unscaled", "urbanL_unscaled", and "para_mn_unscaled" are the unscaled versions of these variables. Finally, the ID column is simply the observation number (unique for each row) used to index the observation-level random effect.

To replicate the analysis in the paper, put the dataset ("canid_data.csv") and the R analysis file ("canid_paper_analysis.R") in the same directory and run the R code. The R file contains inline comments with more details on each step of the analysis and how the resulting output matches tables in the paper.