Mean-field stochastic theory for species-rich assembled communities

A dynamical model of an ecological community is analyzed within a "mean-field approximation" in which one of the species interacts with the combination of all of the other species in the community. Within this approximation the model may be formulated as a master equation describing a one-step stochastic process. The stationary distribution is obtained in closed form, and is shown to reduce to a log-series or log-normal distribution, depending on the values that the parameters describing the model take on. A hyperbolic relationship between the connectance of the matrix of interspecies interactions and the average number of species exists for a range of parameter values. The time evolution of the model at short and intermediate times is analyzed using van Kampen's approximation, which is valid when the number of individuals in the community is large. Good agreement with numerical simulations is found. The large time behavior, and the approach to the stationary state, is obtained by solving the equation for the generating function of the probability distribution. The analytical results which follow from the analysis are also in good agreement with direct simulations of the model.