Vigilance response of a key prey species to anthropogenic and natural threats in Detroit

Rapid urbanization coupled with increased human activity induces pressures that affect predator-prey relations through a suite of behavioral mechanisms, including alteration of avoidance and coexistence dynamics. Synergisms of natural and anthropogenic threats existing within urban environments exacerbate the necessity for species to differentially modify behavior to each risk. Here, we explore the behavioral response of a key prey species, cottontail rabbits (Sylvilagus floridanus), to pressures from humans, domestic dogs, and a natural predator, coyotes (Canis latrans) in a human-dominated landscape. We conducted the first camera survey in urban parks throughout Detroit, Michigan in 2017-2020 to assess vigilance response corresponding to a heterogeneous landscape created from variation in the occupancy of threats. We predicted a scaled response where cottontail rabbits would be most vigilant in areas with high coyote activity, moderately vigilant in areas with high domestic dog activity, and the least vigilant in areas of high human activity. From 8,165 independent cottontail rabbit detections in Detroit across 11,616 trap nights, one-third were classified as vigilant. We found vigilance behavior increased with coyote occupancy and in locations with significantly high domestic dog activity, but found no significant impact of human occupancy or their spatial hotspots. We also found little spatial overlap between rabbits and threats, suggesting rabbits invest more in spatial avoidance; thus, less effort is required for vigilance. Our results elucidate strategies of a prey species coping with various risks to advance our understanding of the adaptability of wildlife in urban environments. In order to promote coexistence between people and wildlife in urban greenspaces, we must understand and anticipate the ecological implications of human-induced behavioral modifications.

While prey modify their behavior to avoid attempted predation, predators modify their behavior 75 to account for prey behavior and to increase the likelihood of success of their predation attempts. 76 Specifically, prey are forced to modify their behavior spatially or temporally to avoid threats 77 from humans as well as associated domestic animals or natural predators (Fenn and Macdonald Highly adaptable species and those with relatively smaller body sizes are more successful 83 at coexisting with humans in urban areas (Bateman and Flemming 2012     To determine the level of risk from each of our three potential predator focal species, we 177 used two method to capture their spatial variation in parks across Detroit. First, we used kernel 178 density analysis to construct utilization distributions from rabbit, human, coyote, and domestic 179 dog camera triggers in ArcMap (v. 10.6.1). To test for significant spatial clustering (i.e., 180 hotspots), we applied the Getis-Ord-GI* statistic to species triggers, which summarizes spatial 181 autocorrelation with resultant high positive z-scores indicating clustering and low negative z-182 scores indicating dispersion (Getis and Ord 1992). Specifically, significant trigger hotspots and 183 coldspots are derived from z-scores greater than 1.96 and less than 1.96 (α < 0.05), respectively.

184
Finally, we overlaid significant trigger hotspots for rabbits with associated threats to determine if 185 rabbits avoided hotspots for humans, dogs, or coyotes across the city. In other words, we 186 assessed whether trigger hotspots for rabbits were congruent with any of the threats. Evidence of 187 spatial avoidance may represent a sufficient evasion strategy that necessitates less vigilance 188 behavior.

189
Second, we constructed single-species, single-season occupancy models for humans,   independently for vigilance by at least two members of the Applied Wildlife Ecology Lab at the University of Michigan. Any discrepancies that were not resolved resulted in classifying the 238 image as unknown. 239 We calculated multiple metrics of vigilance as a response variable to each risk factor.  17.4% moving, 1% active, 1.8% sniffing, and 1% eating. Over a quarter of the total images were 292 either unknown or out of frame, with these categories both being removed from analysis.

293
Models further support differential effects of threats on rabbit vigilance ( Table 2). The  Harris (in press) found no response of human occupancy on carnivore occupancy throughout 318 Detroit in the same parks we surveyed here to access rabbit vigilance behavior. We also found 319 that rabbit vigilance was significantly higher with more vegetation cover, which could be a 320 response to lower visibility to detect predators. did not influence rabbit or squirrel occupancy across an urban-rural gradient study. We also 342 found that as rabbits moved further away from water their vigilance level increased in the urban 343 parks we sampled, which could reflect increased exposure to more disturbed areas in the urban 344 matrix. Urban systems represent a novel landscape for rabbits that requires dynamic changes in 345 vigilance based on the environment and threats of specific locations within the landscape.

346
Our hotspot analysis indicated very little spatial overlap between species, with domestic 347 dogs and rabbits being the only two species to have significant densities at the same camera 348 location in the same year. As a result, we conclude that generally, rabbits are investing more in 349 spatial avoidance, requiring less effort for vigilance. By mostly avoiding their predators, rabbits 350 may be better able to maintain constant levels of vigilance across the landscape rather than 351 heightening vigilance in areas their predators occupy at significant densities. These hotspots of activity might also be confounded by other factors affecting vigilance that were not incorporated 353 in our models. For example, rabbits might be selecting environments based on proximity to 354 housing, overall vegetation density, or grass cover that might be less desirable for their predators, 355 allowing the rabbits to spend less time being vigilant. 356 Notably, our analysis was limited in scope by only examining behavior in areas where 357 these species co-occur. It is entirely possible that spatial or temporal partitioning plays a larger 358 role in mediating predator-prey interactions than vigilance solely in prey. We examined 359 interactions within patches in the city, but neglected to examine the amount of interaction 360 occurring between these spaces. Quantifying the level of risks between patches in the city could 361 be the next step in examining threat impacts on prey behavior. Furthermore, seasonality may      Top models (< 2 Δ AICc) explained rabbit vigilance behavior using detection data from camera