Individual-based modeling of direct and indirect competition by the glass-box artificial intelligence

The main drawback of traditional ecological models is the inability to model local interactions between individuals. As a consequence, it was impossible to mathematically model individual-based mechanisms of direct and indirect competition, which hampers the understanding of ecological dynamics. To overcome this problem, we developed a fully transparent mechanistic analog of the classical Lotka-Volterra phenomenological model based on explicitly explainable artificial intelligence (AI) of the glass-box type. In this paper, on an individual-based model, we present (i) a rigorous mechanism of inverted competitive exclusion where, all else being equal, the weaker species in direct competition defeats the stronger species through indirect exploitative competition and (ii) a rigorous mechanism of coexistence of complete competitors resulting from a combination of accounting for microhabitat regeneration, moderate reproduction, a certain habitat size and boundary conditions. Both presented mechanisms go beyond the classical formulations of the competitive exclusion principle, as in our model two species with different competitiveness compete for a single limiting resource in one homogeneous bounded ecosystem. We have shown that individual-based modeling of direct and indirect competition mechanisms allows for a more detailed understanding of the causal mechanisms of interspecific competition. Such understanding is necessary to refine formulations of the competitive exclusion principle, to search for mechanisms of competitive coexistence, to predict ecosystem behavior, and to develop conservation strategies. Our approach allows us to study discrete systems with fundamentally discrete mathematics, which opens up new opportunities for achieving accuracy, clarity, rigor, and explainability in studies of complex systems.