Ph.D. Candidate: Atish Agarwala
Research Advisor: Daniel S. Fisher
Date: Wednesday, June 5, 2019
Time: 1 pm
Location: Clark S360
Title: High dimensional dynamics in ecology and evolution
Technological advances have ushered in a new age of quantitative biology. With the rise of quantitative methods has come a need for theory, to understand how to analyze data, guide data acquisition and, more broadly, to develop questions. I will present two stories in evolution and ecology, where physics-style toy models illuminate unintuitive but important effects.
The first is a simple toy model of evolution with epistasis - interactions between mutations. I first define a family of random fitness landscapes - functions from genome to fitness - which encode tunable amounts of epistasis. I then present an algorithm which allows for efficient simulation of evolutionary dynamics on these landscapes. This method can also be used to gain an analytic understanding of dynamics, which reveals that evolution takes populations to atypical places on the fitness landscape - strongly constraining the potential for future evolution.
The second story concerns the generation and maintenance of fine-scale diversity in interacting ecosystems. Physicists have tried to develop stable ecological models which maintain diversity since the 1970s. However, most approaches have either failed or require unrealistic assumptions. I present a model which maintains diversity for long periods of time using a combination of structured interactions and spatiotemporal chaos. The model exhibits similar single-time statistics to neutral models of ecology (whose distributions are often fit to real data), while having a rich temporal dynamics. When evolution is added to the model, the diversity increases without bound, provided that the ecosystem is seeded with enough initial diversity.