Predicting Multiplex Networks from Embraces in Spider Monkeys: Computational Tools for Quantifying Social Bonds

Department of Psychology, Florida International University, Miami, FL, USAEmily R. Boeving and Eliza L. Nelson

IntroductionThe emerging field of network science has demonstrated thatsocial relationships emerge from structural connections known asa social network. Relationships emerge from social bonds acrossmultiple behavior types, and are represented using a multiplexnetwork. In using a multiplex approach, social bonds across andwithin behavior types may be examined. Spider monkeys show aphenotype that is opposite to many species in which grooming isnot utilized to regulate social relationships. Instead, embracesare thought to regulate relationships in spider monkeys. Previousstudies have investigated embracing and grooming, however nostudy has investigated the predictive power of any monoplexnetwork behavior and how it corresponds to a multiplex network.Here we investigated the hypothesis that embraces, but notgrooming, hold predictive power of social bonds in the multiplexnetwork by calculating edge weights between dyads.

Methods186 hours of observation data were collected from 15Colombian spider monkeys (Ateles fusciceps rufiventris) atMonkey Jungle in Miami, FL. Social interactive data werepooled as a global matrix and exported to Excel and uploadedto Cytoscape (Shannon et al., 2003) for multiplex networkanalyses. Edge weights, which represent strength of socialbonds, were calculated for each embrace and grooming dyad.The pearson correlation examined if embracing or groomingdyads were correlated with multiplex edge weights.

Results

Conclusion

Contact Information: Emily R. Boeving • eboev001@fiu.edu

• Network edge weights for embracing were correlated tomultiplex edge weights (r(4) = 0.93, p = 0.007.

• Network edge weights for grooming were not correlated tomultiplex edge weights r(4) = 0.0046, p = 0.994.

• We discuss these results in light of spider monkey ecologywith social network analysis as a unique tool for elucidatingthe structure and dimension of social behavior.

References

• Here we provide evidence that the embrace network predictsthe multiplex network in spider monkeys.

• In other words, relationships, which emerge across multiplebehaviors and not just one, are best predicted by embracingbut not grooming.

• This information may particularly useful for captivemanagement staff seeking tools to predict and increasecohesion in captive groups.

• This network analytic approach enabled empirical testing offield-dominated hypotheses regarding the regulation of socialbonds.

• Those wanting to examine social bonds should capture dataon embraces to accurately quantify social bonds in spidermonkeys.

Fig 3. Multiplex Network

Fig 1. Monoplex Network for separate social interactions

Available upon request.

EmbraceEdgeRankDyadsUva- MasUva- SunMas - BuSun- MasCleo - BuBu- Bon

GroomingEdgeRankDyads

Cleo- CaryCleo- CarmCleo- MolMol- BuBu- BonDusk- Sun

MutliplexEdgeRankDyads

Uva- MasUva- SunMas - BuSun- MasCleo - BuBu- Bon

Fig 3. Multi-layered network approach

Fig 2. Social group of Colombian spider monkeys (critically endangered)