Data from: Environmental controls on African herbivore responses to landscapes of fear

From 12 July to 18 September 2014, which is the late dry season in HiP, groups of two trained observers (one field ranger paired with a volunteer) walked 30 line transects ranging in length from 2 to 11 km, for a total of 232 km, and recorded the locations and group sizes of all herbivores encountered with the use of binoculars, a GPS unit, compass and Bushnell Yardage Pro 1000 range finder. The field ranger was largely responsible for the observations, while the volunteer recorded them into the database. This census is performed biennially in HiP and is highly standardized across years and between transects, making consistent observer bias highly unlikely. The bearing of, and distance to, each sighting (using herd centres when groups were encountered) were recorded to calculate the location of each observation relative to the transect line. In cases where animals moved after they had been sighted, observations were recorded where the animal (or herd) was first seen. If an animal (or herd) was not initially observed, but rather disturbed by the observers and then seen running off, it was not recorded. Effects of detectability and distance on group counts were accounted for using Distance software. Each transect was walked between five and thirteen times over the course of the survey, yielding a total survey distance of 2715 km. To avoid potential detection bias across different habitat types and vegetation densities, the datasets only include the observations recorded within 100 m of the observer (number of observations per species: red duiker = 52, grey duiker = 105, impala = 266, nyala = 287, kudu = 41, giraffe = 59, warthog = 101, wildebeest = 37, zebra = 82, buffalo = 145 and white rhinoceros = 326). Transects were evenly distributed throughout the park, apart from the access-limited, south-western wilderness area that excluded from the study due to no available lion spatial data, encompassing all main vegetation types and variability in rainfall and elevation. An equal number of random points, matching the maximum number of observations for any one herbivore species (i.e. 326), were generated within 100 m buffers around the surveyed transects to represent available habitat.