A simulation study of the medium-term effects of field patterns on cross-pollination rates in oilseed rape (Brassica napus L.)

Gene flow between cultivars within a landscape may lead to impurities that reduce harvest value. In OSR, as for most crops, impurity rates are expected to depend on the spatial distribution of crops over the landscape. However, in contrast to other well-studied crops such as maize, OSR crops generate seed banks in European agro-ecosystems. Gene flow is thus a spatio-temporal process which depends on cropping systems. We therefore aimed at identifying spatial variables that have an effect on regional or local harvest impurities, taking account of the time since the introduction of OSR crops in the regions and of cropping system. Gene flow was simulated over 36 field patterns cultivated with either 15% or 30% of OSR fields, among which 10% or 50% were GM, for three contrasted cropping systems, with the GeneSys software already used for EU co-existence studies. Through regression analyses, we determined spatial and agronomic factors that most affected harvest impurity rates of non-GM OSR after one or seven years of OSR cultivation. The cropping system was the main factor explaining regional harvest impurity rates. Its importance increased after six years of OSR cultivation. For a given cropping system, the regional impurity rate after one year increased linearly with the current proportion of GM crop. In contrast, impurity rates after six years largely depended on the proportions of OSR crop (GM or not) in the two preceding years. During the first year of OSR cultivation, local impurity rates were mostly explained by the distance to the closest GM field. After six years, these rates were mostly explained by the density of GM volunteers in the analysed field and, to a lesser degree, to that of volunteers in neighbour non-OSR fields. Cropping systems were most important in determining impurity rates and the way impurity rates related to regional or local factors. Determination of isolation distances to ensure harvest purity should thus consider past history of OSR cultivation in the area and, in particular, how current or future cropping systems will manage volunteers. Regression quantiles were fitted to the simulated data to determine regional rules (i.e. the maximum regional area of GM OSR and isolation distances between GM and non-GM crops) as a function of the risk accepted by the decision-maker (i.e. the % of situations exceeding harvest impurity thresholds), the cropping system and the volunteer infestation.