Theses Doctoral

Social animals detecting danger: how social relations influence antipredator behavior in a noisy forest

Fuong, Holly

The risk of death by predation has been a major driver of group living in many prey animals. Animals must adapt to temporal and spatial variation in predation risk and would benefit from using relevant and reliable sources of information both from conspecifics and heterospecifics to better learn about danger. Research on the effects of group living on antipredator strategy has focused largely on group size. However, sociality is often more complex than simple amalgamations of individuals. Those living in groups are likely exposed to unequal levels of predation risk; some are exposed to more danger than others because of factors related to their age, sex, and spatial or social positioning. An individual’s antipredator strategy should reflect its perceived safety levels.

I studied antipredator strategies in blue monkeys (Cercopithecus mitis stuhlmanni) in the Kakamega Forest, Kenya. Blue monkeys are arboreal guenons that live in matrilineally-based social groups and form differentiated social relationships. These social relationships could affect how monkeys respond to variable predation risk. Blue monkeys live in dense, biodiverse rain forests and are preyed upon by both aerial and terrestrial predators. They have a well-developed acoustic communication repertoire and have been known to associate with other primates to reduce predation risk (Cords 1987). I conducted five playback experiments and two sets of observational studies, and used data gathered on social interactions among adult females to further our understanding of how group living affects antipredator strategies. I also used 14 years of social interaction data to explore the heritability of social tendencies.

In the first chapter, I present a comprehensive literature review of the connections between group living and antipredator behavior. I describe the effects of group size on antipredator behavior and how research on sociality has shifted towards focusing on individuals’ specific relationships and social connectivity. I then describe several ways in which social connectivity has been shown to influence antipredator behavior. I conclude with future directions and then introduce the dissertation.

In the second chapter, I focus on heterospecific eavesdropping. I identified the extent to which blue monkey adult females respond to playbacks of alarm and social calls of two syntopic non-predatory bird species—black-faced rufous warblers (Bathmocercus rufus) and joyful greenbuls (Chlorocichla laetissima). Blue monkeys responded differentially depending on both call type and species. I then evaluated differential responses to conspecific and heterospecific callers, hypothesizing that conspecific signals would trigger stronger anti-predator responses because conspecifics are more relevant signals of risk. I conducted a playback experiment in which adult females were presented simultaneously with one alarm or social call from both conspecifics and warblers (4 combinations of alarm and social calls), or ambient rain forest sound (control). Subjects did not differentiate their responses to simultaneous calls according to the type of playback stimulus. These findings suggest that blue monkeys do not differentiate their responses to alarm calls according to caller relevance. Heterogeneous results among different response variables also highlight the importance of examining multiple modes of antipredator behavior.

Next, I examine how an individual’s social connectivity influences its antipredator strategy, hypothesizing that more socially connected individuals would benefit from the proximity of more and closely bonded groupmates in enhancing predator avoidance. In Chapter 3, I evaluate the effects of social connectivity on acute antipredator responses, antipredator vigilance, and responses to signals related to various levels of predator-related threat. I first assessed how social connectivity affects the rate at which adult females exhibit acute antipredator responses (such as diving down in trees, climbing up trees, or alarm calling) and the proportion of responses that are major (lasting >30 s), statistically controlling for age, the presence of an infant, and 2-month “seasons”. I predicted that more socially connected individuals would exhibit less frequent acute antipredator responses because they would be better-informed about risk and therefore would exhibit fewer false alarms. I For the same reasons, I also predicted that they would exhibit more major (vs. minor) responses because false alarms are more likely to involve shorter responses (Cords 1987). Contrary to predictions, however, more closely connected individuals exhibited higher rates of acute antipredator responses, which might reflect their enhanced ability to learn about danger from surrounding groupmates, allowing them to detect more potential threats. There was no evidence that social connectivity was associated with the proportion of responses that lasted >30 s. I also found that the rate of acute antipredator responses and the proportion of responses that lasted >30 s varied with season. I then conducted 90-s focal vigilance follows, to assess how long females exhibit antipredator vigilance after controlling for other social and microhabitat factors (e.g., surrounding vegetation density), which can influence conspecific monitoring and exposure to potential predators. I predicted that more well-connected individuals would exhibit lower levels of antipredator vigilance in the absence of any imminent threats and after controlling for other social and spatial factors. More closely connected individuals who were in the spatial center of their social group did spend less time vigilant, but social connectivity was not associated with vigilance times when subjects were at the group’s edge, where exposure to predators and thus predation risk should be highest and antipredator vigilance should generally be higher. In the spatial center of the group, more closely connected individuals should be in a better position to observe their social partners’ antipredator behavior. Microhabitat also influenced antipredator vigilance in multiple ways, which highlights the spatial variation of perceived predation risk in a complex environment.

Lastly, I conducted a playback experiment where I examined responses to signals from conspecifics and heterospecifics that are associated with different levels of threat. I predicted that more poorly connected individuals would respond strongly to all signals that might be associated with predators because they must identify personally whether danger is real, whereas more well-connected individuals would have more differentiated responses because they should be near social partners more frequently and can rely on their partners’ antipredator reactions to assess risk levels. However, although subjects did respond more to direct cues of the predator’s presence (its own calls) than to indirect cues of its presence (alarm calls by conspecifics and heterospecifics), there was no evidence that social connectivity affected responses to playbacks. As expected, stimulus type did affect responses—calls from predators (vs. alarm calls or social calls from non-predators) elicited increased looking responses from subjects, which suggests that stimuli that directly signal predator presence will elicit antipredator behavior regardless of the listener’s social connectivity. Overall, social connectivity seems to play a limited role in blue monkeys’ antipredator strategy but there was some evidence that more well-connected individuals were less vigilant when surrounded by groupmates.

The ability to distinguish alarm calls by individual callers has not been well-studied, but animals might benefit from making such distinctions if callers vary in how reliably they signal danger. For decades, researchers have tested whether animals can discriminate callers using the habituation-dishabituation paradigm. After habituating subjects by repeatedly presenting calls of one individual, A, they examine whether subjects dishabituate when they hear the calls of a different individual, B (test stimulus). In Chapter 4, I first review studies that used this paradigm to evaluate whether animals discriminate between conspecific callers and then report on two playback experiments which tested whether wild blue monkeys are capable of such discrimination. My review revealed much methodological variation, particularly in the habituation phase and criteria, statistical analysis, and controls. In experiments, I contrasted two methods of habituation, either presenting a fixed number of stimuli (set after pilot observations) or evaluating responses during the series before progressing to the test. Afterwards, I conducted Wilcoxon signed-rank tests to assess habituation statistically. In the first experiment where I played back a fixed number and rate of calls, it was statistically unclear whether subjects habituated to caller A, despite preliminary observations and similar studies that suggested that the experimental design would be appropriate. Because there was not strong evidence that subjects habituated, I did not evaluate statistically whether subjects differentiated between callers in the full dataset. However, in the second experiment where I assessed habituation during the trial, subjects did habituate to caller A and there was weak support that they dishabituated to caller B, which suggests that caller discrimination may occur. From my experiences, I propose an improved design for studies using the habituation-dishabituation paradigm.

Lastly, I explore the mechanisms that drive phenotypic variation in social tendencies (and in turn, social connectivity) in adult females. For natural selection to occur, there must be variation in traits, differentiated fitness benefits based on phenotypes, and heredity or a genetic basis underlying phenotypic variation. The previous chapters highlight the variation in and some of the benefits of social connectivity. In Chapter 5, I conducted an exploratory analysis to examine what factors account for phenotypic variance. Using animal models, I found that both environmental and additive genetic variance accounted for some of the phenotypic variance seen in traits associated with social tendencies (using social connectivity as a proxy). Variance in the social environment (i.e., environmental variance) played a large role in shaping observed phenotypic variation in social connectivity. However, all six of the social network measures examined were weakly heritable, which suggests that there is also a genetic basis for behavioral variation, allowing selection to occur.

This dissertation emphasizes the importance of examining both antipredator behavior and sociality using multiple experiments, observations, and measures, while also considering the importance of study species and habitat complexity. The relationship between antipredator behavior and social connectivity is not straightforward and can vary greatly between study systems. Although many of my predictions were not supported, I did find evidence that blue monkeys are receptive to heterospecifics, vary their acute antipredator responses and vigilance based on social relationships with conspecifics, adjust their antipredator vigilance according to spatial positioning, and potentially discriminate between alarm callers. The findings presented here expand our knowledge of how animals learn about predation risk by being attentive to conspecifics and heterospecifics.


  • thumnail for Holly Fuong_Dissertation Ch 4 - Supplementary Material S4.1.xlsx Holly Fuong_Dissertation Ch 4 - Supplementary Material S4.1.xlsx application/vnd.openxmlformats-officedocument.spreadsheetml.sheet 96.9 KB Download File
  • thumnail for Fuong_columbia_0054D_16872.pdf Fuong_columbia_0054D_16872.pdf application/pdf 5.01 MB Download File

More About This Work

Academic Units
Ecology, Evolution, and Environmental Biology
Thesis Advisors
Cords, Marina
Ph.D., Columbia University
Published Here
October 6, 2021