E Ecosystem Modeling toward Ecosystem Modeling on Computing Grids

of challenges. Much of classical ecological theory originates from very simple differential equations in which a single variable represents population densities. Researchers typically analyze the solutions mathematically and then compare them to abundance estimates from field or lab observations. Although highly influential in ecological theory, the models’ aggregated form is particularly difficult to relate to observational biology. Applying them to complex natural systems with spatially and temporally varying environmental factors, for example, typically produces analytically intractable models that must be investigated numerically. Recently, researchers have begun emphasizing integrated, multicomponent ecosystem models. These models involve complex interactions between some or all of an ecosystem’s trophic layers (feeding levels), resulting in models that link multiple components that researchers can model using several different mathematical approaches. One implication of these efforts is that monolithic software development within a traditional computing framework is hindering further sustained innovation in complex, highly integrated model simulation. Traditional ecosystem modeling is sequential and not spatially explicit. Grid computing is a natural means to address the spatially explicit requirements of ecosystem models with multiple components. Researchers use different grid service modules1 for different ecosystem model components, based upon the most efficient partitioning of the system’s spatial domain. Herein, we present a conceptual model for ecosystem modeling, and use Across Trophic Level System Simulation (ATLSS; see www.atlss.org) as an example to explain the key design considerations and efforts for such spatially explicit modeling on a computing grid at the University of Tennessee.