Spatially explicit simulations of a microbial food web

A mechanistic model of a simplified microbial food web was implemented based on spatially explicit transport of nutrients and energy sources by molecular diffusion. Individual organisms were simulated to interact in a twodimensional arena consisting of 70 X 70 interconnected hexagonal compartments each of 2 nl in volume. Release of dissolved organic matter in conjunction with predation events resulted in patches of increased concentration lasting long enough to be consumed completely by bacteria before being dispersed. Individual bacteria encountered concentrations of dissolved organic matter varying up to loo-fold within time scales significant for growth. The simulations also predicted that given a chemotactic response to nutrient gradients, bacteria could grow 50% faster on average when gathered in loose clusters within patches. Extending the scenario to the three-dimensional case and looking at a l-ml volume, patch-generating events were estimated to occur several times per hour and spread to within a few millimeters in radius before being eroded by bacteria. The erosion time scale is probably longer than the time between patch appearances resulting in patches overlapping and creating a markedly inhomogeneous microenvironment. In such a scenario, foraging strategies enabling bacteria and predators to respond to elevated concentrations of food could represent a significant adaptive advantage. Within pelagic microbial food webs, nutrients must be recycled to sustain primary production. Bacteria and flagellates are the main agents for degrading organic matter to mineral components used by primary producers (Azam and Smith 1991; Azam et al. 1993). Carbon fixed by primary production flows from autotrophic organisms to bacteria, with an opposite flow of nutrients to autotrophs via bacterial activity. The flow occurs via a multitude of mechanisms operating on different temporal and spatial scales (Andersson et al. 1985; Azam et al. 1993; Fuhrman 1992; Hagstrom et al. 1988; Jumars et al. 1989; Larsson and Hagstrom 1979; Nagata and Kirchman 1992). The transport of matter and encounter frequencies between organisms are partly conAcknowledgments We thank G. A. Jackson, an anonymous reviewer, Grieg Steward, and David C. Smith for help and for comments on the manuscript. This work was supported by a Swedish NFR grant (B-A-U 04452-320) (A. H.) and NSF grant OCE 92-19864 (E A.). trolled by diffusion rates and swimming speeds. These mechanisms are implicit in constants defining MichaelisMenten uptake kinetics (Jumars et al. 1993) and clearance rates for flagellates (Fenchel 1987; Shimeta 1993) but are rarely modelled explicitly. Although the physics of diffusion and transport via shear is understood, it is rarely applied directly in evaluating ecological processes in microbial food web studies. Organic substances are often recycled through events such as point-source excretion and cell lysis. Although dissipated by diffusion and shear, such sources create a potentially inhomogeneous environment. This results in an analytical gap for ecologists studying the microbial food web; current modelling procedures generally assume spatial homogeneity although it is suspected that the microenvironment is nonhomogeneous (Azam and Smith 1991; Azam et al. 1993). The purpose of this paper is to summarize a simple theory for evaluating rates of physical transport in an ecological context and to demonstrate the variable structure of the mi-