Tracking Ideal Varieties and Cropping Techniques for Agroecological Weed Management: A Simulation-Based Study on Pea.

Pea or Pisum sativum L. is a key diversification crop, but current varieties are not very competitive against weeds. The objective was to identify, depending on the type of cropping system and weed flora, (1) the key pea parameters that drive crop production, weed control and weed contribution to biodiversity, (2) optimal combinations of pea-parameter values and crop-management techniques to maximize these goals. For this, virtual experiments were run, using FLORSYS, a mechanistic simulation model. This individual-based 3D model simulates daily crop-weed seed and plant dynamics over the years, from the cropping system and pedoclimate. Here, this model was parameterized for seven pea varieties, from experiments and literature. Moreover, ten virtual varieties were created by randomly combining variety-parameter values according to a Latin Hypercube Sampling (LHS) plan, respecting parameter ranges and correlations observed in the actual varieties. A global sensitivity analysis was run, using another LHS plan to combine pea varieties, crop rotations and management techniques in nine contrasting situations (e.g., conventional vs. organic, no-till, type of weed flora). Simulated data were analyzed with classification and regression trees (CART). We highlighted (1) Parameters that drive potential yield and competitivity against weeds (notably the ability to increase plant height and leaf area in shaded situations), depending on variety type (spring vs. winter) and cropping system. These are pointers for breeding varieties to regulate weeds by biological interactions; (2) Rules to guide farmers to choose the best pea variety, depending on the production goal and the cropping system; (3) The trade-off between increasing yield potential and minimizing yield losses due to weeds when choosing pea variety and management, especially in winter peas. The main pea-variety rules were the same for all performance goals, management strategies, and analyses scales, but further rules were useful for individual goals, strategies, and scales. Some variety features only fitted to particular systems (e.g., delayed pea emergence is only beneficial in case of herbicide-spraying and disastrous in unsprayed systems). Fewer variety rules should be compensated by more management rules. If one of the two main weed-control levers, herbicide or tillage, was eliminated, further pea-variety and/or management rules were needed.

Validation of a crop field modeling to simulate agronomic images.

In precision agriculture, crop/weed discrimination is often based on image analysis but though several algorithms using spatial information have been proposed, not any has been tested on relevant databases. A simple model that simulates virtual fields is developed to evaluate these algorithms. Virtual fields are made of crops, arranged according to agricultural practices and represented by simple patterns, and weeds that are spatially distributed using a statistical approach. It ensures a user-defined Weed Infestation Rate (WIR). Then, experimental devices using cameras are simulated with a pinhole model. Its ability to characterize the spatial reality is demonstrated through different pairs (real, virtual) of pictures. Two spatial descriptors (nearest neighbor method and Besag's function) have been set up and tested to validate the spatial realism of the crop field model, comparing a real image to the homologous virtual one.