A model for the impact of herbicide tolerance on the performance of oilseed rape as a volunteer weed

The introduction of genetically modified, herbicide‐tolerant, oilseed rape into the agricultural environment will have ramifications beyond weed control of the crop. Herbicide‐tolerant rape will undoubtedly become part of established volunteer weed populations that occur in many cereal rotations, but its longevity in these populations and its impact as a weed and contaminant of future oilseed rape crops is uncertain. A life cycle model of volunteer oilseed rape was therefore constructed, incorporating existing information on physiological processes such as emergence pattern, longevity of buried seed, death rates of various structures and flowering and seeding as functions of density. The model was designed to allow interaction with control factors such as harvesting efficiency, herbicide treatment, ploughing and the sequence of crops in the rotation. Many of the physiological parameters (including seed decay rates, fecundity at high density) are uncertain, simply through lack of information in the appropriate context. Other parameters such as harvesting efficiency and herbicide kill rates, are inherently variable in farming. Accordingly, a Monte‐Carlo approach, in which the model was run many times with different random realisations of parameter sets, was used to expose factors to which the seedbank was sensitive. Sets of 1250 realisations were compared for each of two extreme conditions: where herbicide could be used according to current intensive farming practice and where it was not an option (representing total herbicide tolerance). Modelled seedbank numbers after 5 yr ranged from 10‐3to 104m‐2, realistic values found in arable soils. The use of herbicide, together with efficient harvesting of seed, clearly has an important suppressive effect on the oilseed rape seedbank, keeping it lower than 102m‐2(a typical sowing rate) after 5 yr in more than 80% of realisations. In the absence of herbicide, seedbanks were invariably greater, but their absolute value depended strongly on harvesting efficiency and the extent to which high density of plants suppressed fecundity. Analysis of the time series from the simulations showed that the seedbank levels fluctuated by orders in magnitude from year to year in the absence of herbicide use. The sensitivity analysis of the life‐cycle model led to the development of a simplified model for the seedbank dynamics. The model shows that the essential features of the dynamics result from an interaction between density‐dependent fecundity and the perturbations due to management. Therefore predictions of the effect of herbicide tolerance on seedbank dynamics are highly uncertain without knowledge of the density dependence of fecundity. Furthermore, the sensitivity to management practices suggests that seedbank levels will be substantially more difficult to control if the efficacy of herbicide is compromised. It is concluded that the model and Monte‐Carlo approach have many potential uses in exploring the effects of management, cultivar physiology and the nature of the transgenes