Multiscale hybrid modelling framework for intercrops: the best of plant and crop models

I'm doing my PhD at CIRAD (Centre de coopération International en Recherche Agronomique pour le Développement) in Montpellier, in the AGAP (Amélioration Génétique et Adaptation des Plantes méditerranéennes et tropicales) and AMAP (botAnique et Modélisation de l'Architecture des Plantes) joint research units, on multi-scale modelling of associated crops, as part of a European project on associated crops. I've just graduated from INSA Lyon, I'm an engineer in bioinformatics and modelling, and I've done all my work placements in the world of research, particularly in the field of ecology/agronomy, which I really enjoyed, so I wanted to try my hand at a PhD. I'm working on the modelling of associated crops, an agro-ecological practice that involves growing at least two species together, and therefore interspecific interactions. However, existing models are either calibrated at plot level and do not take sufficient account of the heterogeneity of the system, i.e. crop models, or they calculate processes by taking account of the plant's architecture at organ level but are very complex, i.e. structure-function models (FSPM). The two approaches are complementary, so we're going to try to take advantage of both in order to eventually improve crop models so that they take account of interspecific interactions, which will improve their accuracy. We eat about three times a day and this simple act of life has a major impact on the environment, so I think it's important to work on reducing our environmental impact in this sector, so I'm really pleased to be able to use my skills to help optimise the development of agro-ecological practices. 
It's been almost a year since I started my thesis, and I've already developed a methodology for testing the hypotheses of crop models for associated crops using a multi-scale approach. I have also designed a structural and parametric plant model that is at the heart of this methodology, since it enables the growth of a crop simulated by a crop model to be represented in 3D, taking into account architectural features and a spatial arrangement that the latter model does not take into account. Thus, for a given ecophysiological process, we can test the uncertainty linked to these architectural and spatial features, and determine to what extent the corresponding formalism in the crop model should take them into account. If necessary, new formalisms can then be integrated into the crop model to make it more accurate, robust and generic.

• Starting date : 13 November 2023
• Research unit: Agap, Cirad
• University: Université de Montpellier
• PhD school: I2S – Information, Structures, Systèmes 
• Scientific field: Computer science
• Thesis management: Christophe Pradal, UMR Agap, Cirad
• Thesis supervisors: Marc Jaeger, UMR Agap, Cirad et Myriam Adam, UMR Amap, Cirad
• Funding: EU IntercropValuES project
• #DigitAg : Axe 6 : Modélisation et simulation (systèmes de production agricole), Axe 5 : Fouille de données, analyse de données, extraction de connaissances, Challenge 1 : Le challenge agroécologique, Challenge 8 : Développement agricole au Sud

Keywords: Intercropping, agroecology, multiscale modelling, FSPM, crop models

Abstract: Contemporary food systems face growing environmental and economic challenges, requiring more sustainable approaches to guarantee global food security. Agroecology is a promising way to achieve sustainability through crop diversification, such as intercropping. These practices involve growing two or more plant species on the same agricultural plot over a significant portion of their crop cycle (Vandermeer, 1989). 
Given the high cost of field experimentation, modelling appears to be a relevant solution for optimizing these complex agricultural systems. Among the various process-based models available for simulating the growth of intercrops, there are two main types: crop models and Functional-Structural Plant Models (FSPMs). On the one hand, the use of crop models enables in silico testing of the performance of a wide variety of combinations Crop genotype 1 x Crop genotype 2 x Environment (soil and climate) x Management (GxGxExM). These models represent crops at a degraded scale. On the other hand, FSPMs, at the organ scale, can provide a better understanding of the processes underlying plant interactions (Evers et al., 2019). 
However, both approaches have limitations and difficulties in answering key questions: a) how to model intercrops with crop models that consider a homogeneous distribution of plant canopy and root systems without spatial and temporal heterogeneity? b) how to simulate an intercrop over several seasons while being limited by the complexity of plant-plant and plant-environment interactions (FSPM) (Gaudio et al., 2022)? 
As part of the European Horizon IntercropValuES project, we are attempting to define a method for evaluating formalisms for intercrops in crop models, apply it to the STICS model, and if necessary propose new approaches to improve prediction quality. 
The hybrid multi-scale approach chosen in the project is a top-down one, which consists in constraining canopy growth by explicitly considering the spatial arrangement and architecture of plants at the organ scale, with a structural and parametric 3D model. The aim is then to simulate physiological and biophysical processes at this finer scale, and to use this simulation as a reference for evaluating the formalism of the growth model. The extent to which the consideration of plant architecture and explicit spatial arrangement can have an effect on processes in intercrops is therefore investigated. If the uncertainty of a process linked to architectural or spatial traits is significant enough, its formalism can be revisited in the crop model, by investigating new formalisms that are easy to integrate into crop models, i.e. generic, with few parameters, and acceptable robustness and accuracy.

Contact: oriane.braud [AT] cirad.fr - +33695549775
Social networks: LinkedIn - ResearchGate