Abstract: In this article, we introduce a new method allowing the computation of the distribution associated to the number of complex structures in plants deriving from GreenLab type growth model. In order to use this method, we set a new mathematical framework based on combinatorics. We show how plants can be seen as plane rooted trees and how their topology can be described by Dyck words. Moreover, we integrate plant growth in the formalism by adapting stochastic F0L-systems to the framework. This new representation enables not only the computation of the distribution associated to all types of organ but also the analysis of particular patterns thanks to a symbolic method. This approach gives new kinds of applications such as estimating the age of a plant from a set of botanical data.
Introduction
In functional-structural plant modeling (FSPM), researchers utilize models like GreenLab or L-systems to simulate plant development. While these models effectively simulate "typical" plant growth, they often struggle to account for stochastic variability. Specifically, calculating the full probability distribution of organs—such as the number of fruits or leaves—or complex arrangements like bud patterns across a population of randomly growing plants is mathematically intensive.
Traditional methods frequently rely on "moments" (mean and variance), which provide an incomplete picture of variability. Furthermore, these methods often fail to account for "transitory" structures, such as apexes (growing tips), which evolve into different structures in subsequent growth cycles. This paper bridges this gap by applying tools from computer science and discrete mathematics to plant biology.
A Combinatorial Framework for Plant Topology
The core of the proposed framework involves representing plant architecture as Plane Rooted Trees (PRT). In this view, a plant is treated as a collection of nodes and edges where the branching order is significant. To make these structures algorithmically readable, the authors employ Dyck words. By "walking" the tree—recording a specific character when moving away from the root and another when backtracking—a unique string is created that perfectly describes the plant's skeleton.
This approach integrates stochastic F0L-systems (parallel rewriting grammars) into a combinatorial framework, offering greater flexibility than standard branching process theories.
The Symbolic Method and Pattern Analysis
The "symbolic method" from analytic combinatorics allows researchers to translate botanical rules directly into functional equations known as Generating Functions. For instance, a rule stating a bud has a 20% chance of death and an 80% chance of branching can be modeled as a functional equation. Solving these equations yields the exact probability distribution for any structure of interest.
By treating plants as recursive data structures, the method can identify and track "transitory" patterns. For example, a specific sequence like "internode-internode-bud" can be treated as a single pattern, and its distribution can be tracked over time, even if those specific buds eventually bloom or die.
Key Results and Age Estimation
The authors demonstrate that for permanent organs, this symbolic method is mathematically equivalent to existing generating function methods but offers significantly more generality. A primary application of this framework is solving inverse problems, such as estimating a plant's age. By knowing the probability distribution of apexes for a plant at various ages, researchers can use Bayes' Theorem to work backward from a physical specimen's apex count to determine its most likely age.
The method successfully produces the full probability mass function for the number of fruits in a stochastic GreenLab model, moving beyond simple averages to reveal the true "shape" of biological variability.
Limitations and Future Directions
Currently, the study focuses on 2D plant representations. While the transition to 3D is theoretically straightforward, it would significantly increase the complexity of the "alphabet" used to describe the plant. Additionally, the method assumes that underlying growth probabilities are already known; in practice, these must be estimated from noisy real-world data.
Future work may involve using this framework for advanced statistical tasks, such as sensitivity analyses on growth models or improving parameter estimation accuracy in complex biological systems.
Summary updated