Search

Casagrande M, Makowski D, Jeuffroy M-H, Valantin-Morison M & David C (2010). The benefits of using quantile regression for analysing the effect of weeds on organic winter wheat. Weed Research 50, 199–208. In organic farming, weeds are one of the threats that limit crop yield. An early prediction of weed effect on yield loss and the size of late weed populations could help farmers and advisors to improve weed management. Numerous studies predicting the effect of weeds on yield have already been conducted, but the level of uncertainty about weed effect is expected to be very high in organic crops. It is thus more appropriate to provide farmers and advisors with distributions of possible production levels, rather than with point values. The purpose of this study was to estimate the effect of early weed density at the end of the tillering stage of organic winter wheat on subsequent yield and on late weed density at flowering, by using quantile regression. Results showed that this method can be applied to a hyperbolic model and to an allometric density-dependent model, to describe the distribution of yield values and of late weed density respectively, as functions of early weed density measurements. Mechanical weed control showed no significant effect on the relationship between early weed density and grain yield, but it decreased late weed density. Yield and late weed density distributions derived by quantile regression provided sound information on the possible effect of weeds on organic winter wheat production. [ABSTRACT FROM AUTHOR]

Genotypes × environment interactions (GEI) are more fully analyzed when genotypes and environments are well characterized. The characterization of the environments via direct measurements seems somewhat immediate. However, this method generates too many variates that reduce their own significance in the GEI analysis. Several methods can be used to reduce this number, e.g., a crop diagnosis combined with probe genotypes and biological indicators. Genotypes can be characterized by several methods: (1) via direct measurements, (2) using crop modelling, or (3) by comparison to probe genotypes. All these methods will be illustrated in this chapter, mainly for winter wheat (Triticum aestivum L.), and their prospects for QTL detection will be discussed. [ABSTRACT FROM PUBLISHER]

Highlights • The current N strategy led to correct grain protein content for malting market but to high N losses. • Postponing by 10–15 days the first and second N applications led to similar or higher calibrated yields. • Postponed N applications also led to the required grain protein content for malting. • Best N strategies reaching high-calibrated yield, targeted grain protein content, and low N losses differed between regions. Abstract Nitrogen fertilizer applications are essential to achieve high yield and malting specifications in barley, but may also have a deleterious effect on the environment. Most strategies currently being implemented aim to optimize quantitative and qualitative production without taking environmental concerns into account. We used a barley crop model to pinpoint new nitrogen management plans maximizing the calibrated yields (yield of grains larger than 2.5 mm) and the grain quality whilst reducing N gaseous emission and N leaching. We compared the currently recommended N fertilization strategy in France and 44 new ones defined based on expertise knowledge, and differing in the amount of N applied, splitting patterns and time of application. The strategies providing the best compromise between the three criteria were identified by considering Pareto optimal solutions over 25 years in 35 French départements (equivalent to a county). We also identified Pareto optimal strategies for the seven years with the lowest yields and for years in which climatic conditions were unfavorable for efficient use of the early N supplies. The current recommended N fertilization strategy resulted in a high proportion of situations satisfying malting grain protein content requirements, but also to high N losses. We pinpointed new N strategies resulting in better compromise between the three outputs studied. Some Pareto optimal strategies were particularly efficient to reduce N losses in all tested environments regardless of the climatic conditions. They, however, also slightly reduced calibrated yield compared to the reference strategy. Others interesting strategies performed better than the reference simultaneously for all three studied outputs, but depended on the region considered. A common feature of these strategies was later application of smaller doses of N. Our results, thus demonstrated that low-N strategies are possible for malting barley. [ABSTRACT FROM AUTHOR]

In a context of economic and environmental concerns in agriculture, legumes appear to be suitable alternative crops to diversify current cropping systems and reduce their dependence on synthetic nitrogen (N) fertiliser and protein from imported soya bean. However, legume-based cropping systems may increase N losses through nitrate leaching if the N available from legumes does not coincide with subsequent crop requirements. To help agricultural advisers manage N in these systems, we adapted the decision-support system Syst'N®, designed to assess N losses in cropping systems, to simulate three annual and one perennial legume crops: pea, faba bean, soya bean and lucerne. To this end, we adapted and simplified existing submodels of legume functioning to include them in Syst'N, to keep the latter simple. We adapted the submodels "BNF" (i.e. biological N fixation) from the STICS model and "dormancy" from the CropSyst model. We also added the ability to enter the flowering date to improve predictions (improvement in N fixation's rRMSE from 57% to 41% and EF from 0.57 to 0.77). The equations and associated parameter set developed for the four legume crops yielded satisfying predictions of crop biomass (rMBE = 9%, EF = 0.82, rRMSE = 39%) and N content (rMBE = 5%, EF = 0.76, rRMSE = 37%). These performances support the philosophy of Syst'N® that requires minimising the number of additional parameters for users when representing new crops or processes. • The decision-support system Syst'N® was adapted to simulate diverse legume crops. • Submodels of legume functioning were adapted and simplified to keep Syst'N® simple. • The ability to enter the flowering date to improve predictions was added. • Predictions of soil water and N content were less accurate. [ABSTRACT FROM AUTHOR]