DIMACS Seminar on Math and CS in Biology


Computational Approaches to Competition in Ecology


Doug Deutschman
Cornell and Princeton Universities


Princeton Computer Science Department
Rm 401 (the library)
Princeton University


1:00 PM
Tuesday, December 5, 1995


Predicting ecosystem behavior remains a difficult problem in ecology. Early attempts to develop models of competitive systems relied on simplifying assumptions about the nature of individual interactions (i.e. all individuals are identical, the system is well mixed, the dynamics are deterministic) in order to remain tractable. The advent of fast computers has allowed the use of complex simulations to explore the consequences of these simplifying assumptions.

I will present a brief overview of the traditional models of competition in ecology and explore the insights gained from these models. I will then explore the consequences of incorporating mechanistic, spatially local interactions in models of competition among tree species in northeastern forests. Several different approaches for incorporating the spatial structure into forest models will be reviewed. The role of small-scale interactions controlling large-scale community behavior in forests will be explored in detail using SORTIE, a mechanistic, individual-based simulation model of forests in the northeastern United States. This work demonstrates that approximate spatial interactions are adequate to predict forest development, but only if spatial processes are retained.

Document last modified on November 29, 1995