Data from: Proactive cursorial and ambush predation risk avoidance in four African herbivore species
Abstract
Most herbivores must balance demands to meet nutritional requirements, maintain stable thermoregulation and avoid predation. Species-specific predator and prey characteristics determine the ability of prey to avoid predation and the ability of predators to maximise hunting success. Using GPS collar data from African wild dogs, lions, impala, tsessebe, wildebeest and zebra in the Okavango Delta, Botswana, we studied proactive predation risk avoidance by herbivores. We considered predator activity level in relation to prey movement, predator and prey habitat selection, and preferential use of areas by prey. We compared herbivore behaviour to lion and wild dog activity patterns and determined the effect of seasonal resource availability and prey body mass on anti-predator behaviour. Herbivore movement patterns were more strongly correlated to lion than wild dog activity. Habitat selection by predators was not activity level-dependent and, while prey and predators differed to some extent in their habitat selection, there were also overlaps, probably caused by predators seeking habitats with high prey abundance. Areas favoured by lions were used by herbivores more when lions were less active, whereas wild dog activity level was not correlated with prey use. Prey body mass was not a strong predictor of the strength of proactive predation avoidance behaviour. Herbivores showed stronger anti-predator behaviours during the rainy season when resources were abundant. Reducing movement when top predators are most active and avoiding areas with a high likelihood of predator use during the same periods appear to be common strategies to minimize predation risk. Such valuable insights into predator-prey dynamics are only possible when using similar data from multiple sympatric species of predator and prey, an approach that should become more prevalent given the ongoing integration of technological methods into ecological studies.
README: Data from: Proactive cursorial and ambush predation risk avoidance in four African herbivore species
https://doi.org/10.5061/dryad.w0vt4b8zr
The dataset contains R scripts and all the .csv files necessary to run the scripts.
Description of the data and file structure
Data were derived from GPS data collected from impala, tsessebe, wildebeest, zebra, African wild dog and lion in the Okavango Delta, Botswana, between 2014-2016.
Data are split into four folders: Distances, Predator habitat selection, Herbivore habitat selection and RSFs. Each folder contains the R script and all necessary .csv files to run the code and duplicate the results from the manuscript, which are presented in Notepad documents.
The Distances folder contains data relating to hourly distances covered by prey species during period of high and low activity. Files are split by herbivore species, in .csv format. Variables are Species (impala, tsessebe, wildebeest or zebra), ID (individual ID), Hour (Hour of the day), Pred (wild dog or lion), PredAct (whether predator activity was high or low) and season (rainy or dry).
The Predator habitat selection folder contains data used to determine whether predator habitat selection varied with their own activity, season, and whether wild dogs varied their habitat selection depending on the activity of lions (who will kill wild dogs if they can). The Excel file contains available habitat proportions (derived from home range outlines), used habitat proportions (derived from utilisation distributions), split by season and predator species, and habitat selection ratios (derived by comparing used and available habitat proportions). In each habitat proportion spreadsheet (e.g. Lions dry available, refers to habitats available to lions in their dry season home ranges), the variables are habitat types (floodplain, grassland, mixed and mopane). In each habitat selection sheet (e.g. Dogs manova), the variables are ID (individual ID), Season (dry or rainy), Lion (high or low activity), and the habitat types previously listed.
The Herbivore habitat selection folder contains data used to determine whether herbivores changed their patterns of habitat selection with season and activity patterns of predators. Similar to the previous folder, the Excel files, split by herbivore species, contain available habitat proportions (derived from home range outlines), used habitat proportions (derived from utilisation distributions), split by season and predator species and their activity levels, and habitat selection ratios (derived by comparing used and available habitat proportions). In each habitat proportion spreadsheet (e.g. Tsessebe lion rainy low available, refers to habitats available to tsessebe in the rainy season home ranges occupied when lion activity is low), the variables are habitat types (floodplain, grassland, mixed and mopane). In each habitat selection sheet comparing habitat selection ratios between periods of high and low predator activity (e.g. Tsessebe dogs dry activity MANOVA), the variables are activity (of the predator, high or low), and the habitat types previously listed. In each habitat selection sheet comparing habitat selection ratios between prey and predators (e.g. Tsessebe and dogs dry), the variables are Species (one of the 4 herbivores and one of the 2 predators) activity (of the species, high or low), and the habitat types previously listed.
The RSFs folder contains data used to test whether intensity of use of an area by herbivores varies with intensity of use by predators, depending on activity level of the latter. Data files are in .csv format, split by herbivore species and season. In each file, the variables are UI (utilisation intensity, derived from predator utilisation distributions), x and y (coordinates for each pixel in the utlisation distribution), Used (whether the location was used (1) or available (0); used locations were GPS fixes recorded from herbivores and available locations were generated randomly within herbivore seasonal home ranges), id (ID of the individual herbivore), Pred (dog or lion), Activity (of the predator, high or low), and Season (rainy or dry).
Code/Software
All scripts used to run the analyses on this dataset are included, to be run in R. Running them through RStudio will prompt for the installation of any packages necessary for the analyses. The working directory will need to be updated to the user's file system.
Methods
Data were collected using GPS-enabled satellite collars fitted to African wild dogs, lions, impala, tsessebe, wildebeest and zebra in the Okavango Delta, Botswana.
GPS data were used to extract information relating to hourly distances moved, habitat selection and resource selection functions.