Draft:Functional-Structural Plant Models

FSPMs draw on a lineage of formal models of plant branching. In 1968 the biologist Aristid Lindenmayer introduced L-systems, a class of string-rewriting grammars that could generate the branching patterns of filamentous organisms and, by extension, higher plants.^[6] Przemysław Prusinkiewicz and Lindenmayer's 1990 monograph The Algorithmic Beauty of Plants demonstrated how L-systems could produce realistic computer graphics of plant form and became a reference point for later work.^[7]

Through the 1990s, researchers began coupling these architectural descriptions with physiological process models—for example, attaching radiation-transfer calculations or carbon-balance equations to a 3D branching structure—rather than treating form and function separately.^[1][2] Godin and Sinoquet identified this coupling as the defining characteristic of the FSPM approach in a 2005 editorial for New Phytologist.^[1] Improvements in computing power, 3D imaging, and high-throughput plant phenotyping during the 2000s and 2010s made it feasible to parameterise and validate increasingly detailed models.^[2][8][4]

An FSPM typically represents a plant as a graph of connected modules—internodes, leaves, root segments, fruits—each assigned a position, orientation, and geometry in three-dimensional space.^[2] One or more sub-models then calculate how those organs interact with the environment: a common configuration pairs a ray-tracing or radiosity light model with a leaf-level photosynthesis model and a set of carbon-allocation rules that feed assimilates back to growing organs.^[2][3] Because the geometry is explicit, architectural features such as leaf angle, phyllotaxis, or the degree of mutual shading between neighbouring plants can directly influence simulated carbon gain.^[2][4]

A broad distinction is drawn between static and dynamic FSPMs. Static models fix the plant's architecture (often digitised from a real specimen) and compute quantities such as light absorption or transpiration at that single point in time; they are useful for analysing how a given canopy structure performs under particular conditions.^[2] Dynamic models, by contrast, simulate development: organs appear, expand, senesce, and shed according to developmental rules, and the changing architecture feeds back into the physiological calculations at each time step.^[2][9][4]

Some FSPMs adopt an agent-based formulation in which each organ or meristem is an autonomous entity following local rules; whole-plant behaviour then emerges from the interactions among these agents.^[10]

FSPMs have been applied in greenhouse horticulture to optimise crop spacing and light distribution, in field-crop research to evaluate ideotype designs for cereals, and in intercropping studies where the spatial arrangement of different species determines resource partitioning.^[2][3][11] In forestry, they have been used to model crown development and competition for light in mixed-species stands.^[3][4] More recently, FSPMs have been coupled with genetic algorithms or evolutionary computation to explore how plant traits may have evolved under particular selective pressures, an approach sometimes called "evolutionary FSPM".^[12]

The field has also intersected with plant phenotyping: because FSPMs generate synthetic 3D canopies, they can serve as training-data generators or as tools for interpreting phenotyping measurements, a connection highlighted in several reviews.^[5][11]

Several open-source and academic platforms have been developed for building FSPMs:

• L-studio / vlab – developed at the University of Calgary, one of the earliest FSPM environments, based on L-system formalisms.^[13]
• OpenAlea – a Python-based visual-programming platform developed at CIRAD and INRAE in France, designed for component-based assembly of plant models.^[14]
• GroIMP – a Java-based platform from the University of Göttingen that extends L-systems with relational growth grammars.^[15]
• CPlantBox – a C++/Python framework focused on root–shoot interactions and soil–plant water relations.^[16]
• Helios – a C++ framework emphasising scalability and GPU-accelerated radiative transfer.^[17]
• OpenSimRoot – a C++ model for simulating root architecture and nutrient uptake; it can function as a stand-alone model or be linked to other platforms.^[18]
• Virtual Plant Laboratory (VPL) – a recent Julia-based framework.^[19]

• Crop simulation model
• L-system
• Plant physiology
• Plant phenotyping
• Agent-based model