Study plots
Out of the 49 independent grassland locations with active prairie dog colonies, three locations were selected for each grassland type, resulting in a total of 12 grassland locations. Locations were selected based on accessibility and spatial distance. We selected locations with at least 5 km separation, due to prairie dogs average dispersal distance (Garret and Franklin, 1988). Preference was also given to locations where community-based conservation projects had already been implemented or are currently implemented by local organizations to ensure feasibility of the study results for future conservation efforts within GPCA El Tokio. Cattle activity was observed at all selected locations and has been documented by Estrada-Castillón et al. (2010), but no detailed information was available on the number of cattle or stocking densities. In each of the 12 grassland locations, we selected two sites differing in prairie dog disturbance (WP = with active prairie dogs burrow section, WOP = without active prairie dog burrows section) and marked a randomly selected 30x30 m quadrat for each condition, considered as site (n=24). All sites were representative of the vegetation and placed congruent to the cardinal directions. WP and WOP sites within the same location had a minimum distance of 1 km, except for sites in one of the mountain locations, where WP and WOP sites were only 300 m apart due to lack of availability of alternative study areas. Within each site, six random 5x5 m plots, aligned along their edges following the cardinal directions, were delimited as well. Randomization was performed by blindly throwing six 60 cm diameter rings to fall at random. The burrow closest to the rings was selected as the center of the WP plots. Whenever the selected site had less than 6 burrows within it, all burrows were selected for plots and the leftover plots were randomly selected and assigned as non-burrow plots. To account for seasonal effects, the 5x5 m plots were further halved to create two 5x2.5 m subplots, which from here on are considered as seasonal subplots. Seasonal subplots were assigned as eastern (rainy season) and western (dry season; Figure 1b) halves. Data collection took place during August-September 2019 (rainy season) and during December 2019-January 2020 (dry season).