Elsevier

Ecological Complexity

Volume 19, September 2014, Pages 140-147
Ecological Complexity

Key players and hierarchical organization of prairie dog social networks

https://doi.org/10.1016/j.ecocom.2014.06.003Get rights and content

Highlights

  • We describe social networks characteristics of Cynomys gunnisoni.

  • The social network social groups we detected were consistent with those identified by more traditional behavioral approaches.

  • We found social sub-structuring and identified key players in prairie dog social groups.

  • We found substantial variation in the patterns of interactions across prairie dog social groups.

Abstract

The use of social network theory in evaluating animal social groups has gained traction in recent years. Despite the utility of social network analysis in describing attributes of social groups, it remains unclear how comparable this approach is to traditional behavioral observational studies. Using data on Gunnison's prairie dog (Cynomys gunnisoni) social interactions we describe social networks from three populations. We then compare those social networks to groups identified by traditional behavioral approaches and explore whether individuals group together based on similarities. The social groups identified by social network analysis were consistent with those identified by more traditional behavioral approaches. However, fine-grained social sub-structuring was revealed only with social network analysis. We found variation in the patterns of interactions among prairie dog social groups that was largely independent of the behavioral attributes or genetics of the individuals within those groups. We detected that some social groups include disproportionately well-connected individuals acting as hubs or bridges. This study contributes to a growing body of evidence that social networks analysis is a robust and efficient tool for examining social dynamics.

Introduction

In the study of social animals, there is growing interest in complex emergent properties of group structure. Social network analysis (SNA) has been increasingly used to study the social dynamics of animal systems (Bergmuller et al., 2010, Brent et al., 2011, Lusseau, 2003, Lusseau and Newman, 2004, Newman, 2003). It is a unifying conceptual framework that can be applied comparatively across all social taxa—from microbes to humans. Social networks can help to identify features of species that are indiscernible (or even invisible) based on studies of individuals or behaviors alone (Croft et al., 2004, Lusseau and Newman, 2004). In other cases, there exists substantial intra-specific variation among networks based, in part, on group attributes, individual differences, and ecological factors (Faust and Skvoretz, 2002, Guimarães et al., 2007, Bhadra et al., 2009, Madden et al., 2009). Furthermore, differences in social networks, whether among taxa or social groups, almost necessarily lead to differences in the spread of diseases, decision making strategies, information or, in some cases, food, through networks (Croft et al., 2004, Drewe et al., 2009, Hamede et al., 2009, Jacobs et al., 2011, Kasper and Voelkl, 2009, Madden et al., 2009).

A key challenge with SNA is how to relate their results to the much larger literature on social interactions that relies on other approaches to distinguish social groups. Prior to the widespread use of SNA, behavioral studies explored social interactions and social groups dynamics using informal clustering techniques (e.g., Hinde, 1976). Our understanding of the social systems of most organisms rests on such traditional approaches. Can the results from these earlier studies be related to those of social network analysis? This question seems to have not been well considered, particularly in the social mammals where research has tended to divide social groups into hierarchical categories. Such groups are constructed out of the existence of interactions among individuals but also the nature of those interactions and whether they are negative, positive, reproductive, relate to food sharing, or have some other defining features. The advantages of SNA are frequently highlighted (e.g., Proulx et al., 2005; Sueur et al., 2011; Wey et al., 2008), but whether SNA builds on, replaces, or conflicts with other approaches is unclear.

Gunnison's prairie dogs, Cynomys gunnisoni, are large, diurnal, highly social ground squirrels whose range is limited to the grasslands of the Colorado Plateau (Hall and Kelson, 1959). Gunnison's prairie dogs colonies contain a variable number of territories occupied by distinct social groups, ranging from 3 to 15 individuals (Travis et al., 1995, Verdolin and Slobodchikoff, 2010) akin to small groups of social insects (e.g., Temnothorax albipennis: Dornhaus and Franks, 2006), primate groups (Chapman and Chapman, 2000), or hunter gatherer societies (Hamilton et al., 2007). Traditionally, ecologists have distinguished prairie dog social groups using behavioral and spatial observations of known individuals over time (King, 1955, Slobodchikoff, 1984, Travis and Slobodchikoff, 1993, Verdolin, 2007), with a strong emphasis on negative interactions, where negative interactions among individuals imply those individuals are from different social groups (Slobodchikoff, 1984, Travis and Slobodchikoff, 1993, Verdolin, 2007). The designation of the size of groups and the identity of individuals within groups also often relies on data on mating behavior and behavioral time allocation (e.g., time spent being vigilant versus feeding; Slobodchikoff, 1984, Travis and Slobodchikoff, 1993, Travis et al., 1995, Verdolin, 2007). The resulting identification of distinct social groups within a site can be robust with regard to individual interactions, but tends to result in a categorical classification of groups, in which individuals either are or are not members of groups and any patterning in social structure above or below the standard social group is either not described or, if described, is in terms of the behavior of individual species and their histories.

Although SNA has been used recently for a variety of social species, its application has focused primarily on individual measurements or full network measurements. When SNA methods are used to find intermediate (within network) structure in the full networks, these methods are referred to as community detection (Leu et al., 2010, Lusseau, 2003, Lusseau and Newman, 2004, Maryanski, 1987). The use of community detection techniques in the analysis of social networks has recently gained traction (Porter et al., 2009). Often network structure is not obvious by simply looking at a list of interactions, or a resulting graph of interactions. Community detection permits a researcher to identify social groups by discerning which individuals in the network have more connections to the other individuals within the group than to individuals outside the group.

If a network-based approach to exploring the social dynamics of Gunnison's prairie dog—or any other species—produces social groupings similar to traditional methods, social network analysis can add to the insights of traditional approaches in several ways. First, comparing social network properties among groups may highlight subtle variation in social structure not readily observable or quantifiable by conventional behavioral studies (Faust and Skvoretz, 2002, Traud et al., 2011, Wolf et al., 2007). Second, network analyses can also reveal emergent properties of social groups, including identifying individuals with central roles—such as the dolphin social brokers—and characterizing variability in group cohesion or hubs, individuals who are connected to an unusually high number of other organisms (Bezanson et al., 2008, Croft et al., 2005, Gero et al., 2013, Lusseau, 2003, Lusseau and Newman, 2004, Madden et al., 2009, Naug, 2008). Third, SNA may provide a method for testing the hypothesis that individuals may group together based on similarities, differences, or random associations (Galef and Laland, 2005, Pedersen et al., 2006; Pepper, 2000; Reader and Biro, 2010, Rendella and Whiteheada, 2001, Ross, 2001). On the other hand, if SNA produces fundamentally different social group clusters than traditional behavioral approaches, it might imply that the two methods describe potentially distinct information and social processes.

Here, we generated social network matrices using data on positive social interactions of Gunnison's prairie dogs. We then used community detection analysis to discern distinct social groups simply from the network data and compared them to social groups identified by traditional behavioral approaches (Traud et al., 2011). Next, we used SNA to examine whether there were features of Gunnison prairie dog social behavior detectable only through SNA or behavioral studies alone. Based on the differences we found between SNA and traditional methods, we expanded our analysis to further explore aspects of sociality not detectable by traditional methods.

Section snippets

Study area

A detailed description of live-trapping, handling, and marking methods are available in Verdolin (2007). A Scientific Collector's Permit (Arizona Game and Fish Permit no. SP742094) was obtained prior to trapping and all procedures were in compliance with Stony Brook University IACUC (IACUC no. 2009-1745, Stony Brook University). Individuals were trapped with veterinary supervision from mid-February (upon emergence from hibernation) through August at two colonies, Country Club (CC) and Humane

Results

A total of 220 focal samples for 80 prairie dogs were collected. In addition, a total of 5, 5, and 4 social groups were identified using behavioral observations and spatial locations for populations HSI, HSII, and CCI, respectively. Network analysis resulted in three different weighted networks, where each connection between a pair of prairie dogs was weighted by the number of interactions between the prairie dogs in that pair (Fig. 1). Overall, CCI, HSI, and HSII, consisted of 46, 32, and 47

Discussion and conclusions

We found that the majority of the prairie dogs were placed in social network communities that were consistent with their traditional behavioral social group placement (Fig. 1). More importantly, the Social Network Analysis (SNA) approach also recovered additional structure within those groups, as well as previously undetected structure within those social groups. Within network-based social groups, individuals were subdivided into smaller subunits of individuals that mostly interact with each

Acknowledgements

JLV was supported by NESCent (EF-0905606). ALT was supported by the Statistical and Applied Mathematical Science Institute Complex Network Fellowship, NC State Mathematics Department and a NASA Biodiversity Grant (ROSES-NNX09AK22G). RRD was supported by a US DOE PER award (DE-FG02-08ER64510), a NASA Biodiversity Grant (ROSES-NNX09AK22G) and an NSF Career grant (0953390). For assistance in the field many thanks to Dr. David Washabau, Bill and Theresa Emig, Carolyn Parker, Perry Crompton, Kristen

References (75)

  • R. Wittig et al.

    Focused grooming networks and stress alleviation in wild female baboons

    Horm. Behav.

    (2008)
  • J.B.W. Wolf et al.

    Social structure in a colonial mammal: unravelling hidden structural layers and their foundations by network analysis

    Anim. Behav.

    (2007)
  • J. Altmann

    Observational study of behavior: sampling methods

    Behaviour

    (1974)
  • R. Bergmuller et al.

    Evolutionary causes and consequences of consistent individual variation in cooperative behaviour

    Philos. Trans. R. Soc. B: Biol.

    (2010)
  • M. Bezanson et al.

    Patterns of subgrouping and spatial affiliation in a community of mantled howling monkeys (Alouatta palliata)

    Am. J. Primatol.

    (2008)
  • V. Blondel et al.

    Fast unfolding of communities in large networks

    J. Stat. Mech.

    (2008)
  • C.A. Botero et al.

    Fluctuating environments, sexual selection and the evolution of flexible mate choice in birds

    PLoS One

    (2012)
  • L.J.N. Brent et al.

    Social network analysis in the study of nonhuman primates: a historical perspective

    Am. J. Primatol.

    (2011)
  • C. Chapman et al.

    Determinants of group size in primates: the importance of travel costs

  • D. Croft et al.

    Assortative interactions and social networks in fish

    Oecologia

    (2005)
  • D. Croft et al.

    Behavioural trait assortment in a social network: patterns and implications

    Behav. Ecol. Sociobiol.

    (2009)
  • D.P. Croft et al.

    Social networks in the guppy (Poecilia reticulata)

    Proc. R. Soc. Lond. Ser. B: Biol. Sci.

    (2004)
  • G. Csardi

    igraph: Routines for Network Analysis R Package

    (2005)
  • A. Dornhaus et al.

    Colony size affects collective decision-making in the ant Temnothorax albipennis

    Insectes Sociaux

    (2006)
  • J. Drewe et al.

    The social network structure of a wild meerkat population: 1. Inter-group interactions

    Behav. Ecol. Sociobiol.

    (2009)
  • E. Edgington et al.

    Randomization Tests

    (2007)
  • P. Erdős et al.

    On random graphs

    Publ. Math.

    (1959)
  • K. Faust et al.

    Comparing networks across space and time, size and species

    (2002)
  • J. Flack et al.

    Social structure, robustness, and policing cost in a cognitively sophisticated species

    Am Nat

    (2005)
  • R. Gadagkar

    The Social Biology of Ropalidia marginata

    (2001)
  • B.G. Galef et al.

    Social learning in animals: empirical studies and theoretical models

    Bioscience

    (2005)
  • S. Gero et al.

    Calves as social hubs: dynamics of the social network within sperm whale units

    Proc. R. Soc. B: Biol. Sci.

    (2013)
  • P.R. Guimarães et al.

    Vulnerability of a killer whale social network to disease outbreaks

    Phys. Rev. E: Stat. Nonlinear Soft Matter Phys.

    (2007)
  • E.R. Hall et al.

    The Mammals of North America

    (1959)
  • R.K. Hamede et al.

    Contact networks in a wild Tasmanian devil (Sarcophilus harrisii) population: using social network analysis to reveal seasonal variability in social behaviour and its implications for transmission of devil facial tumour disease

    Ecol. Lett.

    (2009)
  • M.J. Hamilton et al.

    The complex structure of hunter–gatherer social networks

    Proc. R. Soc. B: Biol. Sci.

    (2007)
  • R.A. Hinde

    Interactions, relationships and social-structure

    Man

    (1976)
  • Cited by (11)

    • The structure and temporal changes in brokerage typologies applied to a dynamic sow herd

      2022, Applied Animal Behaviour Science
      Citation Excerpt :

      Animal behaviour research has demonstrated that identifying influential or highly connected animals within a group is a useful tool for understanding social dynamics (Lusseau and Newman, 2004; Verdolin et al., 2014).

    • The stability of social prominence and influence in a dynamic sow herd: A social network analysis approach

      2021, Applied Animal Behaviour Science
      Citation Excerpt :

      Temporal behaviour changes at group and individual levels can be identified with the application of social network analysis (Davis et al., 2018), modelling the intricate patterns of social interactions (Croft et al., 2008). Social network analysis is also a mechanism for identifying critical animals; individuals with a disproportionate effect on group dynamics (Modlmeier et al., 2014) and whose removal or introduction may impart effects (Croft et al., 2008; Makagon et al., 2012; Verdolin et al., 2014; Kulachi et al., 2018a). With the potential to address welfare issues and further understand how sociality may have fitness consequences (Naug, 2008; Wilson et al., 2013), social network methods have transitioned into commercial pig research (Foister et al., 2018; Büttner et al., 2020; Turner et al., 2020).

    • Space matters: host spatial structure and the dynamics of plague transmission

      2021, Ecological Modelling
      Citation Excerpt :

      In addition to susceptibility to plague differing among prairie dog species (Russell et al. 2019), spatial structure may vary both within and between species (Tileston and Lechleitner 1966, Travis and Slobodchikoff 1993, Hoogland 2006. Verdolin et al. 2014). How spatial structure between individuals, pathogen reservoirs, and host susceptibility are represented in pathogen transmission models determines the force of infection and can have a substantial effect on model results (Malagon et al. 2020).

    • Animal Behavior

      2021, Animal Behavior
    • Friends of friends: Are indirect connections in social networks important to animal behaviour?

      2015, Animal Behaviour
      Citation Excerpt :

      Whether individuals succeed in dispersing can be influenced by network connections. In a study of network centrality in Gunnison's prairie dogs, Verdolin et al. (2014) found that males tended to have high degree, while females tended to have high betweenness. Because females tend to disperse within populations in this species, connections to members of other cliques (i.e. high betweenness) may facilitate dispersal (Verdolin et al., 2014).

    View all citing articles on Scopus
    View full text