WheatSim: A process-based wheat model linking organ development, plant architecture, and field-scale responses under environmental variability

Wheat ( Triticum aestivum L.) underpins global food security, yet process-based simulations of wheat growth remain challenging under variable weather and management conditions. Traditional crop growth models (CGMs) mainly aim to reproduce field-scale outcomes but often simplify organ growth dynamics and plant architecture, whereas functional-structural plant models (FSPMs) represent detailed organ-level growth but are typically confined to single stems and vegetative stages. Explicit plant architecture can integrate organ-specific details from FSPMs into the production-oriented framework of CGMs, thereby establishing a computational pathway that links organ-level responses to plant architecture and field-scale functioning. In this study, we developed WheatSim, a new process-based framework that explicitly integrates organ-level morphology and physiology, plant-level architecture, and field-level production. In WheatSim, organs are encapsulated as self-contained modules with organ-specific models for emergence, growth, and senescence, while multiple organs are organized and coordinated through a tiller hierarchy that defines a botanically realistic plant architecture. A representative plant approach is established alongside the hierarchical tillers to provide a simplified averaging bridge from organ-level simulation to field-scale outcomes. WheatSim couples biochemical photosynthesis and stomatal regulation to simulate carbon assimilation and transpiration, and exchanges state variables with external soil and climate modules for interoperable simulations. Across APSIM NextGen field experiments spanning diverse environments, management practices, and stress regimes, WheatSim showed good agreement with observations for both morphological properties (leaf number, tiller population, and leaf area index) and biomass-related properties (aboveground and grain biomass and nitrogen mass), with relative mean absolute errors typically ranging from 0.1 to 0.3. Temporal trends were also reasonably reproduced, as indicated by linear regression slopes between simulated and observed values that were generally close to 1. The combination of explicit plant architecture and organ-specific models also provides a useful basis for interpreting integrated crop responses to interacting environmental drivers in selected field sites. WheatSim therefore provides an effective computational framework to simulate wheat growth across organ, plant, and field levels, and may also inform future development of process-based models for other cereal crops.