Multi-agent modeling in plant biology

Traditional methods, such as systems of algebraic or differential equations, L-systems, or functional-structural models, are often unable to fully simulate the dynamic interactions of plants with their environment. Multi-agent systems allow the modeled object to be represented as a collective of autonomous agents representing individual functional parts, each of which follows local rules that ensure decision-making and interaction with the external environment.

Aim. The study is to analyze modern approaches to multi-agent modeling in plant biology. An analysis of several publications revealed that multi-agent modeling reproduces orange tree growth, root system architecture, the morphological adaptation of black alder, and the behavioral plasticity of animals in plant ecosystems, enabling the implementation of digital twins of wheat. The reviewed studies place particular emphasis on the emergent properties of the proposed models, which manifest themselves without explicitly defining global rules. The results of the analysis demonstrate the high potential of the multi-agent approach as a tool for modeling the morphological and physiological processes of biological systems, as well as its potential for digital farming, breeding, and yield forecasting in a changing climate. This approach is capable of accounting for spatial heterogeneity of the environment and temporal changes in conditions. The presented review of research shows that the approach based on multi-agent systems is successfully applied to modeling tree growth, root systems, population dynamics, and digital twins of agricultural crops.