Scientific collaboration networks are a special kind of social and communication networks in which nodes represents authors, and they are connected in pairs by an edge if they have worked together to publish a paper. Collaboration networks grow and change over time as new authors join the networks and new collaborative connections are formed between authors. However, the addition of nodes and edges is highly selective. Only authors who can produce acceptable papers can join the network and be connected only to their coauthors. Therefore, collaboration networks are not just growing but evolving under the evolutionary constraint that selection prefers particular authors and collaborative connections. To account for the structural change of scientific collaboration networks, this dissertation proposes an evolutionary model that incorporates the three fundamental mechanisms of evolutionary dynamics--replication, mutation, and selection. The replication and mutation processes were adopted to formulate the growth mechanism of collaboration networks. The growth mechanism describes the process in which new authoring teams are formed by replicating the membership structure of existing teams with mutation and generates an ensemble of possible networks by adding the new authoring teams to an ancestral network. Selection takes place at the team level where particular authoring teams are chosen from all the possible teams based on their capabilities to produce acceptable papers. The results of numerical experiments suggested that collaboration networks tend to show high levels of robustness against random failures which comes at the expense of the tolerance for targeted attacks but to be highly modular under strong selection, which is consistent with previous literature on network robustness and modularity. The proposed model was fitted against empirical collaboration networks. They were constructed based on the bibliographic information about the papers published in the eight leading journals in their fields across diverse disciplines, including a communication journal. The dataset included 168,557 papers and 161,797 authors. The structural changes of the collaboration networks were measured and decomposed into two parts: (1) the amount of the change due to the replication and mutation processes and (2) the amount of the change due to the selection mechanism based on the Price equation. The decomposition of the observed changes revealed that the structural changes of the empirical networks were not fully explained by the growth model alone. More specifically, the observed changes significantly deviated from the changes predicted by the replication and mutation processes. The analysis of the residuals of the growth model showed that the observed changes were consistently biased in the direction predicted by the selection mechanism. Further, a series of regression analyses found that the structural changes of the empirical collaboration networks were more fully explained when all three evolutionary mechanisms were included than when they assumed to grow only by the replication and mutation processes. A set of post-hoc analyses confirmed that the current findings were replicable when the assumptions of the team formation mechanism were relaxed, and when structural properties of collaboration networks were measured in different ways. The major contributions of the dissertation and their implications are discussed in the final chapter. The limitations of the current study as well as future research directions are also addressed. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://bibliotheek.ehb.be:2222/en-US/products/dissertations/individuals.shtml.]