Using simulations to predict the genetic connectivity of the Mojave desert tortoise

The Mojave desert tortoise is a threatened species that is facing habitat fragmentation from human development. Understanding the impact of fragmentation on this species is critical for developing appropriate conservation actions, but the effects of habitat fragmentation are often delayed, making it difficult to assess the impacts of recent landscape change. One tool often used to predict the impacts of fragmentation are agent-based models, which simulate the behavior and life-history of individual “agents”. Agent-based models allow researchers to investigate the impacts of habitat fragmentation under many scenarios, which is useful for guiding conservation actions. However, because agent-based models are computationally intense, they are often limited to small spatial extents and low numbers of agents – while performing these simulations at large scales could lead to important insights, this is often infeasible.In this dissertation, I use a computationally efficient agent-based model to assess the impact of anthropogenic development on the range-wide genetic connectivity of the Mojave desert tortoise. In Chapter 1, I describe the quadtree R package, which implements the region quadtree data structure in C++ and makes it available to the R programming environment – using this data structure increases the speed of the agent-based model. In Chapter 2, I calibrate and validate an agent-based model for predicting desert tortoise genetic connectivity. In Chapter 3, I use the model to make range-wide projections of the influence of anthropogenic development on desert tortoise genetic connectivity