The key vigilance and behavioural predictors of looking in a group of gray-footed chacma baboons
exhibiting inter-individual differences in tolerance to observers

Abstract

In behavioural ecology, vigilance has proved a popular area of research focus over the preceding
decades. Although primates have received relatively less attention than other mammals or birds,
primate vigilance research has also grown considerably. In this thesis my primary aim was to identify
the main drivers of vigilance use in a habituated group of gray-footed chacma baboons (Papio
ursinus griseipes) at the Lajuma Research Centre, in the Western Soutpansberg. My review of
primate vigilance literature (chapter 2) found extensive variation in terms of vigilance definition and
sampling methodology that combined with other methodological inconsistency made cross-study
comparisons challenging. I explored the implications of this in chapter 3 and found that different
vigilance definitions can vary in their inter-observer reliability and produce varied results both within
and across observers through definition and interpretation effects. Although there was no single
definition that removed interpretation issues entirely, there was some evidence that more complex
operationalised definitions may help remove some of the ambiguity in definitional interpretations.

Although my review of primate vigilance literature highlighted that certain themes were consistently
investigated, such as the group-size effect on vigilance and sex differences, observer-effects on
vigilance were largely overlooked; I therefore elected to explore habituation in the next study
(chapter 4). Research on wild animals, particularly primates, has often relied upon habituation of
study subjects to ensure researchers are able to observe animals directly. However, habituation is a
process of declining response to a consistent stimulus as opposed to a state, and in many cases
throughout behavioural ecology, it seemed to be an implicit assumption that researchers are a
‘neutral’ stimulus and that study subjects are ‘equally’ tolerant of researcher presence; however,
neither factor had received much empirical attention. I explored whether these implicit assumptions
had merit in the study group of baboons at Lajuma by quantifying the visual orientation distance
(VOD) and flight initiation distance (FID) of all non-infant group members. The results suggested
evidence of a potential personality component to the outcome of habituation processes, with
individuals displaying consistent but individually distinct responses to both measures. The results of
this work allowed for the extraction of individual level estimates for visual and displacement
tolerance (conditional modes) that were utilised in my remaining chapters to explore the role this
trait and observer proximity/behaviour had on baboon behaviour. The baboon’s behavioural
responses to our approaches were incredibly passive and similar to their typical responses to
approaching social threats, suggesting the baboons likely considered observers as equivalent to a
high-level social threat as opposed to a neutral stimulus.

During the process of assessing VODs and FIDs in the study group an adult male group member was
predated by a leopard. The remaining group members exhibited an intense alarm response and
gathered around the deceased animal for some time afterwards. Once this situation had begun to
calm down, I assessed whether the stressful event had altered the typical VOD and FID responses of
a subsample of individuals (approx. 25% of group members) during the remainder of the day
(chapter 5). The individual VODs, FIDs, and individual tolerance estimates were largely unchanged,
suggesting that despite the stressful event that the habituated baboons do not alter their fear
perception towards researchers. FID research typically assumes that FIDs are a proxy for predation
risk; however, this may not be the case if habituation processes have begun. In an increasingly
urbanised world, it may become increasingly unlikely that such assumptions continue to have merit.

In chapter 6, I used focal samples collected on the baboon group between June 2018 and June 2019
to explore whether researcher proximity influenced the inter-individual proximity patterns of the
habituated baboon group. I found that the interaction between individual displacement tolerance
(derived from FID measures) and the distance with which I stood from a focal animal had a strong effect on how likely animals were to be neighbours of a focal animals. When I was close, the number
of intolerant animals occurring as neighbours of the focal animal was lower than that of more
tolerant animals this effect appeared neutralised when I was further away. Together these results
suggest observers have the potential to influence the inter-individual association patterns of
habituated animals, tolerance should therefore be discussed as important methodological
information in these research areas, particularly studies using social network analyses.

Finally, using the looking framework I proposed for primate vigilance research in my literature
review, I investigated the potential risk/vigilance drivers of looking and the specific behaviours and
tasks that may constrain or promote its use in the study group at Lajuma. The risk drivers included
threats posed by leopards and other groups of baboons (both preemptive and reactionary), withingroup
group threats, and the interaction these variables had with the baboon’s physical and social
environment, e.g., habitat visibility, spatial position, and group cohesion. I also used the visual
tolerance estimates (from VOD assessments) to investigate how the study animals responded to
observer distance and movement. My analytical approach to these questions weighted the rival
hypotheses alongside one another in the same analysis and revealed that the baboons increased the
duration of their looking behaviours in response to encounters with other groups and ongoing
events linked with social threats (i.e., wahoo bouts and within-group conflict). Both the duration and
frequency of looking bouts also had strong positive associations with the number of threatening
group-members nearby. However, models exploring their specific behaviours and foraging
success/items held the greatest prediction accuracy, suggesting the baboons have a propensity to
utilise compatible looking time during certain behaviours and tasks, thus are likely to have up-todate
and reliable information on their threat environment at most times. In chapter 3, I also found
that the baboons visually oriented towards our approaches very quickly if they were already looking
around, and that this ability was not substantially hindered during engaged behaviours (e.g.,
digging); highlighting that the baboons were adept at detecting localised threats regardless of their
behaviour. Collectively, these results suggest the study group may not need to be routinely or preemptively
vigilant in non-threatening scenarios, and that the combination of their sensory capacity
and attentiveness to their environment make them proficient at detecting threats if they are there.

Overall, this thesis highlights the importance of detecting and investigating methodological
assumptions in published research, as they may not always be applicable to all study animals and
groups. Within primate vigilance literature alone I found many factors that were consistently
overlooked and no clear research framework to enable reliable cross-study comparisons. It may be
important going forward that researchers work to consolidate methodologies at all levels to improve
the comparability of results and further explore definition and interpretation effects more generally.
We additionally found strong evidence that habituation may not remove the fear study animals have
towards observers and therefore these topics may require additional exploration going forward. My
findings highlight that it is possible to explore a range of risk and vigilance hypotheses without
making habituation assumptions and without attempting to sample specific subcomponents of
vigilance. Adoption of such a framework may therefore offer ways of minimising between study
differences in methods without losing the ability to gather evidence supporting complex hypotheses.