Intercropping in silico: a process-based modelling framework to harness agrobiodiversity from field evidence to future scenarios

Feeding a growing population within planetary boundaries requires cropping systems that couple productivity with ecosystem-service delivery. Intercropping, particularly cereal–grain legume mixtures, offers such potential but remains underused in Europe due to design sensitivity, data gaps, and modelling limitations that hinder credible decision support. This thesis advances field- scale diversification by developing a generic, robust, and simple intercrop modelling framework to underpin a farmer-oriented decision support system (DSS). The research proceeds along three strands. First, to ground modelling in evidence and fill key gaps, we designed and implemented harmonized field trials over two seasons (2023–2024) at two contrasting European sites (Mediterranean Italy; continental Germany), testing durum wheat and oat with lentil and chickpea under low-input/organic management. High-frequency observations (phenology, biomass, LAI, canopy structure, soil water and mineral N, weeds, grain yield and quality) generated an extensive dataset for process understanding and model evaluation. Second, we adapted a structured, multi-variable calibration protocol to parameterize the MONICA crop model for lentil and chickpea using literature, direct measurements, and sequential/simultaneous optimization against multi-environment datasets, with an independent Europe-wide yield dataset for evaluation. This delivers transferable parameter sets for “minor” legumes and identifies priorities for improvement (e.g., biomass partitioning, maturity, soil detail). Third, we introduce MONICoSMo, which couples MONICA with a suitability-weighted community formalism (CoSMo) to simulate intercrops using one calibrated sole-crop instance per species interacting through a shared soil. Daily suitability functions partition light, water, and mineral N by suitability-weighted demand, adding only one structural parameter (inertia) while preserving sole- crop behaviour and avoiding dominant-crop allocation biases. The framework is evaluated across multiple species pairings and spatial arrangements (mixed, relay, row), with sensitivity analyses treating initial relative abundance and driver hierarchy as structural uncertainties. Together, these contributions establish the foundations of an operational intercrop DSS: (i) rich, multi-variable datasets; (ii) calibrated legume modules; and (iii) a parsimonious, mechanism- based framework capable of reproducing observed competition/facilitation patterns and screening intercrop designs for yield, land-use efficiency, and associated ecosystem services. The thesis closes by outlining co-design steps with farmers and data/model developments needed to scale decision support for Europe’s agroecological transition.