Modelling weed seed predation by carabids and its effects on crop production under contrasted farming systems
Abstract
To reduce herbicide use, different avenues of biological weed regulation are currently investigated. Among these, weed seed predation by carabid beetles appears promising. Though observed in different cropping systems and conditions in fields, there was to date no demonstration that this process actually influences weed dynamics over time and reduces weed harmfulness for crop production. Consequently, the objective of the present paper was to (1) model the impact of cropping system, field margin and pedoclimate on weed seed predation by carabids to complete the FlorSys model, which simulates weed dynamics and crop production from cropping systems, soil and weather, (2) evaluate whether including seed predation is needed to correctly predict weed dynamics in different cropping systems, (3) which components of the seed predation submodel are the most influential. The new seed-predation submodel calculates the daily predation rate for each weed species from seed traits, weather data, canopy state variables and management operations. In FlorSys, this predation rate is applied to the newly shed weed seeds on the soil surface. The equations and parameters were based on past publications from our team and other literature. Then, simulations were run with FlorSys over 13 years, with and without the seed predation submodel, using weather data and management operations from 10 fields from the INRAE Dijon-Epoisses experimental station. The resulting output in terms of weed and crop state variables (plant density, biomass, seed bank, yield) were compared to measurements from the 10 fields, showing that including weed seed predation in the simulations improved the model's prediction quality, by reducing the overestimation in weed-variable predictions. Finally, a sensitivity analysis to the components of the predation submodel was run, by repeating the simulations after successively switching off individual components of the submodel. This showed that daily incident radiation, light interception by plant canopy, harvest, carabid reproduction and daily temperature had the most influence on seed predation rates. The simulations showed that weed seed predation by carabids can indeed contribute to managing weeds, by reducing field infestation and improving crop yields, but with large variations among crops. However, to determine which cropping systems and field margins favour weed seed predation enough to noticeably contribute to biological weed regulation, a larger diversity of cropping systems and weather series needs to be explored by simulation.
Origin | Files produced by the author(s) |
---|