This paper studies how a strict predator–prey ordering changes which species can live together for the long run. The authors extend a well‑known ecological toy model (the May–Leonard model) so that species sit in a directed chain: each species eats the one below it, and the top has no predator while the bottom has no prey. They add the usual birth and death processes, then use equations for average densities and computer simulations to follow what happens over time.

At a high level, the team used two complementary tools. First, mean‑field analysis — a set of differential equations that track average population densities — gives predictions about which steady states (sets of coexisting species) are possible and which are stable. Second, Monte Carlo simulations — repeated stochastic computer experiments that include random birth, death and predation events — test how those predictions play out when chance matters, especially in small populations.

The main finding is that a state with all species present is usually unstable. Instead of staying in a full, multi‑species mix, the system tends to lose species and settle into lower‑dimensional coexistence states that contain only some species. Which subsets survive is not arbitrary. The authors show that the long‑lived groups correspond to independent sets of the interaction network — that is, sets of species that do not directly prey on one another. This provides a simple graph‑based rule that links the model’s wiring (who eats whom) to which species combinations the dynamics will select.

The simulations also revealed important size effects. In small populations, random extinctions are common and can drive species loss that is more severe than the mean‑field equations predict. As system size grows, the stochastic runs more closely follow the mean‑field outcomes, and the graph‑based selection of coexistence states becomes clearer. The authors analyze a five‑species example in detail and note a simpler three‑species case in the appendix to illustrate the mechanism.