Extreme behavioural shifts by baboons exploiting risky, resource-rich, human-modified environments

A range of species exploit anthropogenic food resources in behaviour known as ‘raiding’. Such behavioural flexibility is considered a central component of a species’ ability to cope with human-induced environmental changes. Here, we study the behavioural processes by which raiding male chacma baboons (Papio ursinus) exploit the opportunities and mitigate the risks presented by raiding in the suburbs of Cape Town, South Africa. Ecological sampling and interviews conducted with ‘rangers’ (employed to manage the baboons’ space use) revealed that baboons are at risk of being herded out of urban spaces that contain high-energy anthropogenic food sources. Baboon-attached motion/GPS tracking collars showed that raiding male baboons spent almost all of their time at the urban edge, engaging in short, high-activity forays into the urban space. Moreover, activity levels were increased where the likelihood of deterrence by rangers was greater. Overall, these raiding baboons display a time-activity balance that is drastically altered in comparison to individuals living in more remote regions. We suggest our methods can be used to obtain precise estimates of management impact for this and other species in conflict with people.


Additional information (Methods: capture)
For five days, cages were baited with corn and placed within the core home range of the study troop so that the baboons readily entered and consumed the food (corn). On the day of trapping, once the baboon was settled and feeding, the door was shut by an observer pulling a cord that released it. The baboons were anaesthetized using low impact 1.5ml plastic dart deployed from a pole (Daninject or Telinject), with 0.6ml of Ketamine (2mg/kg). Once sedated, the collar was fitted to the baboon and adjusted accordingly. The collar was made sufficiently tight so that it may not be pulled over the baboons head, nor get caught in vegetation, but sufficiently loose as to be comfortable and not hinder the animal's respiration or ingestion and rotate around its neck freely. The smooth inner surface of the collar minimised risk of rubbing. Once the collar had been fitted the baboon was administered with 0.6ml of Medetomidine (0.02 mg/kg) in order to reverse the effects of the sedation. The baboon was then placed back into the cage until it had regained full mobility and awareness. Once recovered, it was released from the cage and re-joined the troop.

Additional information (Methods: collar failure)
Whilst we completed robust testing in the lab for our custom-designed collars, we experienced a high rate of collar failure. Two of the collars did not record and several stopped working due to water damage or due to a failure in the memory cards (details provided in Table S1). These prototype collars were significantly enhanced after this study; learning from our mistakes we created a second version that we describe in Fehlmann et al. S1 . If you are looking to replicate this work, or do something similar, please follow the full end-to-end process we describe in the supplementary material of Fehlmann et al. S1 in order to avoid encountering the same issues we had during the study we present here.

Fig. S1. Map of study area (Methods: Risk and returns).
Map of the area given to the rangers during the interviews. Rangers were asked to colour in areas between the white lines where, in their opinion, the baboons were allowed to be: at any time (green), some of the time (orange), or never (red). Areas were defined by specific landmarks such as farm boundaries, roads or tree lines. The satellite image is issued from Google Earth (access date: 6/6/2014).

Fig. S2. Sensitivity analysis for VeDBA (Methods: Space use and activity).
Sensitivity analysis to select the time interval around a GPS fix (n=6,273) for the computation of the mean VeDBA (g). The sensitivity analysis has been carried in each different habitat type fynbos, trees, meadows, vineyards and urban (red, green, blue, cyan, magenta, respectively). The vertical line represents the time interval chosen.  The collars were intended to record for 50+ days but stopped working due to water damage or due to a failure in the memory cards (details provided). These prototypes were significantly enhanced in a second version that we describe in Fehlmann et al. S1 . Overall, collars recorded 7,573 GPS fixes, among which 7,428 were associated with activity data and 6,322 were recorded during daytime. From those, 6,274 GPS fixes met our accuracy criterion. The collars were removed after 3 months.

Source code
This document provides all R codes used for all analyses of data presented in:

Extreme behavioural shifts by baboons exploiting risky, resource-rich, human-modified environments
To use these scripts requires the following R packages:

To test differences in energy availability and VeDBA according to habitat type (prediction 1)
Loading the data grid=read.csv("EnvironmentData.csv")

To investigate relationships between the risk of deterrence by rangers and habitat type (prediction 2)
Generating distance matrices: