Investigating and modelling the morphological plasticity of weeds

Weed dynamics models are essential for assessing crop management techniques and cropping systems. They must include a competition model that accounts for the particular characteristics of crop:weed stands, i.e. heterogeneous canopies consisting of contrasting species presenting a high morphological plasticity and resulting from successive emergence flushes. In a previous study, we developed a model which predicts light incidence and absorption in each location of a 3D, individual-based canopy as a function of plant morphology, latitude and seasonal solar height. Plant morphology was simplified to a set of five plant traits (plant height, diameter, leaf vs. non-leaf biomass ratio, specific leaf area and median leaf distribution height). Here, the objective was to propose and parameterize a new modelling approach for predicting these simplified morphological variables as a function of two effects that are notoriously difficult to separate (1) plant biomass which results from past light absorption and determines a large part of the current competitive ability and (2) past and current shading to integrate changes in morphology resulting from plant response to shading. A field experiment was sown with oilseed rape in September. Gaps were artificially created in the canopy by removing seedlings after emergence; additional bare-soil plots were created. Target plants of one grass weed species (Alopecurus myosuroides), three broadleaved weed species (Galium aparine, Sinapis arvensis, Stellaria media), and one broadleaved crop species (oilseed rape) were transplanted into the various canopy scenarios. Target and canopy plants were sampled five times from October to April to determine the morphological variables. A shading index was calculated from incident light averaged over each target plant from predictions with the light availability submodel. A single equation was successfully fitted to each morphological variable, estimating parameters in shadeless conditions, a correlation parameter with plant biomass (for height and diameter only), and a parameter representing sensitivity to shading for each variable, species and date. Plant biomass decreased in the shade but plants were taller with more stem biomass (except in grass weeds) and larger/thinner leaves concentrated towards the top of the plant. The grass species was least plastic; among the broadleaved species, the crop species presented the least plasticity. The light availability and the plasticity submodels were combined to simulate the effect of crop sowing densities and patterns (row-sown vs. broadcast, varying interrow widths) on weed morphological variables. The latter was more altered in dense and in row-sown vs. broadcast canopies but its range of variation was smaller in the former, pointing to a more selective and less diverse environment. The consequences for weed adaptation to crops and for optimizing crop and cultivar choices were discussed.