In Africa, burgeoning human populations promote agricultural expansion and the associated demand for water. Water abstraction for agriculture from perennial rivers can be detrimental for wildlife, particularly when it reduces water availability in protected areas. Ruaha National Park (Ruaha NP) in southern Tanzania, one of the largest parks in Africa, contains important wildlife populations, including rare and endangered species. The Great Ruaha River (GRR) is the main dry-season water source for wildlife in the Park. Water offtake from this river for large-scale irrigation and livestock production up-stream of the Park has caused large expanses of this formerly perennial river within the Park to dry out during the dry season. The dry season distribution of a species in relation to surface water is considered an indicator of its dependence on water and ability to cope with the loss of surface water. We investigated how diminishing surface water availability during three dry seasons (2011–2013) affected herbivores’ distance to water in Ruaha NP. The distance held by herbivores to water is shaped by a range of factors including dietary category. We determined changes in the locations of available surface water throughout the dry season using standardized ground transects, close to and leading away from the GRR, to map the locations of nine herbivore species. Functional responses of herbivores, i.e. their change in distance to water between early and late dry season, indicated that distance to water was (i) shortest in buffalo and waterbuck (grazers), (ii) similar for plains zebra (grazer), elephant and impala (mixed feeders), (iii) larger in giraffe and greater kudu (browsers) and (iv) largest in generalist feeders (warthog, common duiker). The substantial species’ differences in surface water dependence broadly fit predicted species differences in their ability to cope with anthropogenic reduction in surface water in Ruaha NP.

The study was conducted during three dry seasons (2011–2013) within Ruaha NP in central Tanzania, which is located at the transition between the East African Acacia-Commiphora zone and the southern African Brachystegia miombo zone (Barnes 1983).

Counts of individuals of each species sighted along 10 ground transects that used an existing game viewing tract were conducted at intervals of approximately 14 days from June to October and once in November during the dry seasons of 2011, 2012 and 2013; thus each transect was surveyed 11 times per year. To visualize how the progress of the dry season was associated with changes in the distribution of species in the study area, transect data were divided into early (June-Aug, each transect surveyed 6 times) and late dry season (Sept-Nov, each transect surveyed 5 times). Transects were approximately 20 km (mean transect length 19,880 m ± 115.3 m [standard error of mean, S.E.M.]) in length. Five transects were alongside the GRR (A transects 6, 7, 8, 9 and 10), and five transects lead away from the GRR perpendicular to the course of the river (P transects 1, 2, 3, 4 and 5). This resulted in two transect categories (‘alongside’, ‘perpendicular’) that differed in their distance to the GRR, vegetation type and water availability at the start of the dry season. Hence, the main differences between transects influencing detectability of species are encapsulated in these two categories, whereas available surface water is a variable that changed during the dry season and was updated at each survey.

To investigate the effect of the availability of surface water on herbivore distribution, we recorded as dynamic variables the locations of species sightings and their counts and the locations of available water surface in the following way: Data along transects were collected between 07:00 and 11:00 hours in the morning. Sightings of the nine most numerous herbivore species were recorded, including buffalo, waterbuck, zebra, giraffe, kudu, impala, elephant, warthog and common duiker. For each sighting, we also recorded species counts, i.e., the number of individuals of a species per sighting (group sizes), to account for large aggregations in areas containing larger sources of surface water. We did not include the hippopotamus because this species’ dependence on surface water in Ruaha NP is reported elsewhere (Stommel et al. 2016b). We restricted our sightings to animals within 100 m of the survey vehicle to maximize detection and assumed equal bias in detection across the study area. We could not formally account for imperfect detection and hence animals that potentially were missed, but we assume any bias to be similar in the two transect categories. For each sighting of an animal or a group of animals, the perpendicular distance between the animal sighting to the transect line was measured with a laser range finder (Bushnell, PinSeeker 1500 7x26). However, only the GPS coordinates (latitude, longitude) of the locations of each animal or group were taken when animals were first sighted to minimize possible error due to animals shifting position in response to an approaching vehicle. Thus locations were not routinely taken when animals were perpendicular to the vehicle, neither was perpendicular distance recorded. As the maximum distance for recording any animal to the right and left of the vehicle along all transects was set at 100 m, this method resulted in a maximum bias of ± 100 m for locations.

The location of all available sources of surface water in the study area was also monitored from the beginning to the end of the dry season as detailed by Stommel et al. (2016a). That is, while driving along each transect, not only species identity and group size were recorded, but also the GPS location (latitude, longitude) of available water sources. We cannot exclude the possibility that we may have missed some small, ephemeral water sources such as small wildlife-dug waterholes, but we do not think we missed any major sources of surface water as these are well known to rangers in the dry season.

To analyse whether herbivores change their location in relation to the location of surface water, which shifts spatially during the course of the dry season due to evaporation and the digging of water holes by wildlife, we calculated the Euclidean distances (m) from the species’ GPS location to the GPS location of the nearest surface water in the respective sampling week. Please note that both, species location and the position of surface water, can shift from one sampling period to the next, as the pools and water courses dry out. To calculate the nearest distance between species coordinates and water coordinates, we used the statistical software package R, version 4.2.1 (R Core Team 2022) and the function ‘dist’ for matrix calculations and selected the minimum distance to any water location for a respective species location.