Modelling associations between animal social structure and demography

Abstract

Successfully navigating a complex social environment can influence fitness (i.e. survival and reproductive success), which can cascade to outcomes in demography and population dynamics. Extensive research has focused on the mechanisms and drivers of fitness variation that result from differences in social structure and interactions, and separately, the effects of demography on population change. However, there have been few attempts to address the effects of social structure on population dynamics through demography. Here, we first review the effects of social structure and individual social position on fitness. We then address knowledge gaps, including the relationship between social structure and population dynamics, the carryover effects of social conditions, and the differential effects of social variables on specific demographic rates. We also review statistical tools and data requirements for the analysis of social networks and demography. We then propose that knowledge gaps could be filled by using joint modelling approaches. We used a simulation study to highlight the potential use of social networks to inform survival, a key demographic parameter. We developed a model that combines social network and survival (Cormack–Jolly–Seber) analyses and evaluated its performance to place inferences under different group structures and sampling scenarios using simulated data. Our results show that valid inferences on social and survival parameters and their connections can be achieved with realistic sample sizes, but precision is improved with more complete information (i.e. fewer missing individuals). Based on our review and simulation study, we suggest that further development of integrative modelling approaches can yield greater understanding and improved power to make predictions about the effects of social environments on populations and the feedback of population dynamics on social structure. • Social structure influences demography through fitness. • We conducted a simulation combining social network analysis and survival modelling. • Models showed minimal bias and good precision but incomplete data reduced precision. • We suggest further work to overcome limitations associated with data integration.