for always ensuring my library was well-stocked, and for fostering my enduring love of the natural world. Animal populations are often highly structured, with individuals differing in terms of whom they interact with and how frequently they do so. The resulting pattern of relationships constitutes a population's social network. In this dissertation, I examine how environmental variation can shape social networks and influence information flow within them. In Chapter I, I review the history of social network analysis in animal behavior research, and discuss recent insights generated by network approaches in behavioral ecology. I focus on the fields of: social learning, collective behavior, animal personalities, and cooperation. Animal network studies are often criticized for a lack of replication at the network level and an over-reliance on descriptive approaches in lieu of hypothesis testing. Small, shoaling fish may provide a means to address these concerns, as manipulative experiments can be conducted on replicate social groups under captive conditions. Chapters III-V examine the impacts of environmental variation on the social networks of Trinidadian guppy (Poecilia reticulata) shoals, the social dynamics from which they emerge, and information diffusion within them. In the experiments described in Chapter III, I manipulated shoal composition in terms of within-group familiarity.Mixed shoals of familiar and unfamiliar fish exhibited greater homogeneity in network vi structure relative to other groups, which likely contributed to the rapid diffusion of foraging information observed within them. In the experiments discussed in Chapter IV, Imanipulated the within-shoal mixture of personality types. In addition to impacting frequencies of partner switching and patterns of phenotypic assortment, individual-and group-level personality variation had strong effects on the initial acquisition of novel foraging information and the speed of its transmission through a group. In the experiments in Chapter V, I manipulated the ambient predation risk perceived by groups.High-risk conditions were associated with shifts in network structure consistent with attempts to minimize predation risk. High ambient risk also impeded the acquisition and subsequent transmission of foraging information, likely due to heightened neophobia and/or an increase in the perceived costs of personal sampling. I conclude in Chapter VI by considering the broader implications of my work and highlighting promising avenues for future research.vii