The dataset consists of systematically recorded presences, from a vehicle, of medium to large mammal species within the Kruger National Park. There is 401 tar and 369 dirt road points within the dataset. Each dataset point has data on species presence, estimated proximity of the animals to the road and herd sizes. Each point also has the following data: estimated cloud cover percentage, surface wetness, rain, co-ordinates, vegetation biome (savanna), road surface, distance to the nearest natural water source and the total number of vehicles passing that point within a 10 minute period.

Data was collected during two field sessions, one in 2017 and one in 2019 (both of which in July, during the Austral winter). Neither rainfall nor temperature data differed substantially between the two field sessions. Visitor numbers to the park were also similar between the 12-month periods that spanned the field sessions.

Two observers systematically recorded the presence of medium-sized to large mammals, stopping every 1 km along tar and dirt roads, between 7h30 and 15h00. The main tar roads, connecting the main rest camps, were sampled, and equivalent distances of dirt road that ran, as far as possible, parallel to the tar road, were selected. Dirt road loops, which often lead to watering holes, were avoided. This ensured that there was no association between road type and distance to natural water sources (ANOVA, P = 0.85, f = 0.04, df = 1). Furthermore, none of the roads sampled crossed through any of the rest camps or other human settlements. Data was only recorded in the dominant vegetation type, namely savanna, as other vegetation types (e.g. forest and azonal vegetation) were too dense to sight animals more than a few meters from the vehicle. In general, savanna vegetation consists of open landscapes with good visibility and, within savanna vegetation, there was no noticeable difference in the range of sight between tar and dirt roads.

In total, 401 tar road records and 369 dirt road records were obtained over a total period of two months. At each stop all observed medium to large mammal species during a 10 minute interval were recorded. The distance of the closest individual of each species to the road was categorised into one of two distance categories – close (< 20 m from the road) and far (> 20 m from the road). All visible individuals were counted when fewer than 30 were visible, while larger group/herd sizes (>30) were estimated to the nearest interval of 10 animals based on sub-counts of groups of 10 individuals and extrapolating across individuals visible. Group/herd size was also categorised into one of two categories (< 6 individuals and >= 6 individuals), based on an evident binomial distribution in group/herd sizes. Traffic volume was recorded as the number of vehicles passing us from either direction during the 10 minute interval. Lastly, the GPS coordinate of the stop (Garmin GPSMAP 62s) was recorded and the percentage cloud cover was estimated. Afterwards, distances between sampling points and nearest natural water source were measured using ArcGIS 10.8 (ESRI, Inc.). Artificial water sources are not static and were ignored, since their locations and status (open/closed) could not be determined.

0m to the road was grouped into < 20m category.