ABSTRACT
Two general approaches exist for modeling populations with internal structure; i -space distribution models, such as Leslie matrix models or McKendrick-von Foerster and Sinko-Streifer partial differential equations, and i -space configuration models, such as computer simulation models of large numbers of individuals. The basic similarities and differences of these approaches are discussed. Each model type has certain advantages and disadvantages, depending on the nature of the population being modeled, the types of questions the model is supposed to address, and the type of data available. Small populations, populations that are subject to a high degree of temporal stochasticity in the environment, and populations in which environmental exposure and encounters with other individuals are likely to vary greatly within the population, are usually best described by i -space configuration models, which require computer simulation. Examples are presented showing that i -space distribution models may not accurately represent some of the subtleties of population behavior under such circumstances. Large populations in which all individuals are relatively similar are more likely to be satisfactorily modeled by i -space distribution models. The possibility of analytic solution of such models or of applying well-known numerical techniques means that the results of such models are more easily checked and are in forms that are more general and more convenient to use than those of i -space configuration models.
