Antipredator behavior affects prey fitness, prey demography, and the strength of ecological interactions. Although predator-prey interactions increasingly occur in habitats that experience multiple forms of human-generated disturbance, it is unclear how different forms of disturbance might affect antipredator behavior. Fire is a contemporary disturbance that has dramatic effects on terrestrial habitats. Such habitats may have also experienced past disturbances, like agricultural land use, that leave lasting legacies on habitat structure (e.g., overstory and understory composition). It is unclear how these past and present disturbances affect the use of different antipredator behaviors, like temporal avoidance and vigilance. We examined whether variation in disturbance regimes generates differences in ungulate antipredator behavior by using cameras to measure white-tailed deer vigilance and activity time across 24 longleaf pine woodlands that vary in past land use and contemporary fire regime. Regardless of land-use history, woodlands with high fire frequencies had 4 times less vegetation cover than low-fire woodlands, generating riskier habitats for deer; however, deer responded to fire with different antipredator strategies depending on land-use history. In nonagricultural woodlands, fire affected deer activity time such that activity was nocturnal in low-fire woodlands and crepuscular in high-fire woodlands. In post-agricultural woodlands, fire affected vigilance and not activity time such that deer were more vigilant in high-fire woodlands than in low-fire woodlands. These results suggest that ungulate antipredator behavior may vary spatially depending upon past land use and contemporary fire regime, and such disturbances may generate “landscapes of fear” that persist for decades after agricultural use. --

This study was conducted at the Savannah River Site (SRS; Aiken, SC). Sites that were farmland in 1951 were classified as “post-agricultural woodlands,” and sites that were forested were classified as “nonagricultural woodlands.” The number of fires since 1991 was determined from annual fire records, and sites were characterized as low (five or less burns) or high (more than five burns) fire frequency. Sites were not burned the year of the study. At each of our 24 sites, we deployed an unbaited, motion-activated camera trap between June 8 and July 9 in 2018 for a total trapping period of 33 days. For every photo capturing deer activity during an independent foraging bout, the observer recorded the date and time, the sex of the individual, whether or not it was in a group, group size, and if the individual was foraging (1) or being vigilant (0) as a binomial variable. If the individual’s head was up in a non-feeding posture, then the photo was classified as vigilant, and if the individual’s head was down in a feeding posture, then the photo was classified as foraging. We characterized independent foraging bouts as any sequence of deer photos of the same sex captured within 30 minutes at a site. Vertical vegetation cover was measured by a single observer using a density board that estimated percent visibility across 1 x 1 foot quadrats. The density board was placed 15-m from the observer standing at the camera-trap station. The observer took four measurements (one in each cardinal direction from the station) at each site.