Trait-Based Modeling of Larval Dispersal in the Gulf Of Maine
Benjamin Jones, Ph.D., 2017
Rubao Ji, Advisor
Population connectivity is a fundamental process that governs the dynamics of marine ecosystems and is often driven by dispersal during a planktonic larval phase. Accurate, affordable, and meaningful estimates of larval dispersal patterns are therefore a key aspect of understanding marine ecosystems. Individual-based models (IBMs) that couple ocean circulation and particle-tracking models provide a unique ability to examine larval dispersal patterns in detail. Obtaining accurate results with IBMs requires simulating a sufficient number of particles. The sequential Bayesian procedure presented in chapter 2 identifies when the number of particles is adequate to address predefined research objectives and optimizes the particle release locations to minimize the computational expense. The model in Chapter 3 further reduces computational expense by transferring some of the calculations to graphics processing units. Chapter 4 describes three algorithms that assist in interpreting IBM output by identifying coherent geographic clusters from population connectivity data, including a new algorithm that simultaneously considers multiple species. Finally, in chapter 5 we assess which species traits are likely to impact dispersal success and patterns in the Gulf of Maine. We conclude that traits influencing spawning distributions and habitat requirements for settlement are most likely to influence dispersal.