Abstract

Rapeseed transgenes can be dispersed by the way of pollen and grains. The resulting genetic pollution may cause serious problems for the weeding of the subsequent crops (herbicide resistance transgene) or for the quality of rape harvests (fatty acid transgenes). It is therefore necessary to evaluate transgene dispersal risks, to find, if possible, means to limit them.

The aim of our work is to integrate both dynamic and genetic aspects in order to model cropping system influence on gene flow from transgenic rape and to simulate the evolution of transgenic population on several neighbouring plots.

The field pattern comprises fieldedges and fields on which various crop types (non transgenic or transgenic rape, winter and spring crops, various kinds of set-aside) rotate. A crop rotation is set for each plot. Annual rape evolution (grains, seedlings…) as volunteer or crop plants, is modelled for each plot. The relationship between these stages depends on crop type and management. Pollen and grain exchanges between plots depend on plot areas and distances. The model output are the number of resistant and sensitive grains, seedlings and plants, as a function of successive years on each plot. The model structure and examples of simulation are presented.

Introduction

In the current discussion on transgenic herbicide-tolerant rapeseed varieties, the risk of gene transfer to rape volunteers is often brought up (Messéan, 1996). It is therefore essential to rapidly evaluate this risk and identify means to reduce it. Long-term field trials were set up by French technical institutes on several locations to evaluate this risk (Champolivier et al., 1997), but besides being too slow, these trials only comprise a limited number of cropping systems. Therefore, we built a model of the effects of cropping systems on gene flow from transgenic rapeseed varieties to rapeseed volunteers in neighbour plots and following crops; the first version of the model concerns herbicide tolerance transgenes. Hybridisation between rapeseed and Brassica weeds, though possible (Lefol et al., 1996; Chèvre et al., 1997), were not integrated. In this paper, only the general structure of the model is presented; details are given by Colbach and Meynard (1996), Clermont-Dauphin et al. and Colbach et al. Simulations performed with the model contribute to determine the effect of field plan, crop rotation and management on gene flow, to identify low- and high-risk situations, to propose agricultural practices minimising gene flow etc.

Model structure

The input variables of the model are: (i) the field plan of the region, comprising cultivated fields and field-edges or waysides (henceforth “borders”) consisting of spontaneous vegetation; (ii) the crop rotation of each field and (iii) the cultivation techniques (stubble breaking, soil tillage, sowing date and density, herbicide, cutting, date of harvest) applied to each crop. The major output variables are, for each field and year, the number per m² and the genotype proportions of adult rapeseed plants, of seeds produced and of the seedbank containing the viable rape seeds in soil.

Figure 1 shows the general organisation of the annual rape life-cycle simulated for each plot and year. For each life-stage such as seedlings, adults etc, the model calculates the number of individuals per m² and the proportion of the three possible genotypes AA, Aa and aa. The life-cycle is modulated slightly according to the crop grown on the simulated field; for instance, the compartment sown rape seeds only exists in rape fields and simulates the arrival of rape seeds at the sowing of the crop. The relationships between the various life-stages depend:

• on the crop grown in the plot. For instance, rape volunteers growing in a spring crop cannot complete their cycle because of insufficient vernalization and/or time and produce neither flowers nor seeds (relations 8 and 9).

• on the cultural techniques applied to the crops. For instance, the vertical seed distribution in the pre-sowing seedbank depends on soil tillage (relation 3); ploughing mixes seeds of various layers whereas chiselling little disturbs the initial seed distribution.

• on rapeseed characteristics. For instance, plant mortality after herbicide treatments strongly depends on plant genotype (relations 5 and 7).

The annual life-cycle of Figure 1 is simulated independently on each plot, but on two occasions, the various spatial units connect by exchanging pollen and seeds that are illustrated by the “pollen or seed import and exports” compartments. These exchanges are mostly caused by wind. The amount of pollen or seeds dispersed from plot i to plot j results from multiplying the amount produced (and not exported by harvest-combines in the case of seeds) in plot i with the proportion of pollen or seeds dispersed from i to j. This proportion is deduced by integration from equations giving the proportion of pollen or seeds as a distance from the parent-plant; it increases with plot areas and decreases with the distance between the two units. Besides calculating the number of pollen-producing or receiving flowers, it is also necessary to determine when these flowers are open. The flowering dates depend on the emergence date of the seedlings which depend, in the case of cropped rape plants, on the sowing date of the rape crops or, in the case of rape volunteers, on the sowing date of the crops sown in the plot where these volunteers emerge.

Figure 1. Annual rape life-cycle (for cropped and volunteer plants) simulated for each plot and year

GeneSys is currently evaluated, using surveys of farmers’ plots for the dynamic part and observations from the long-term OGM experimentses for the genetic and dispersal part. Preliminary results (Couturaud, 1998) show that the model correctly ranks cropping systems for rape volunteer infestation, but underestimates pollen and seed dispersal, probably because dispersal by insects and animals is insufficiently integrated into the model.

Simulations

Effect of cropping system in a single field

Figure 2 illustrates the effect of crop rotation on the evolution of the post-harvest transgenic seed bank. The basic rotation is the same for both simulations: rape/winter crop/spring crop/ set-aside/rape/winter crop/spring crop. The systematic growing of transgenic herbicide-tolerant varieties (of genotype AA with A conferring tolerance) results, of course, in the largest transgenic seed bank, with substantial increases after rape crops, because of seed loss before or during harvest, and reductions after winter and spring crops. If, except for the first year, only classic varieties are grown, then the seed bank decreases with time, with again slight increases after rape crops, despite these being classic varieties. Set-aside also increases seed bank because of conditions favouring seed production. After 20 years of simulation, there are still a few transgenic seeds left, despite there having been no transgenic variety since the first year.

The transgenic seed bank after four years is, of course, more important. If a classic variety is to be grown at the fifth year, minimising this seedbank is essential to limit the emergence of transgenic volunteers in the classic crop and thus avoid the pollution of the classic harvest by the unwanted gene which could lead to a refusal of a “non-transgenic” quality label for the harvest or to a decrease in price if certain fatty acids were required. The simulations show that set-aside management is crucial: sowing a set-aside field in spring (System 2 of Table 1) instead of leaving it unsown (System 1) considerably decreases the transgenic seed bank as rape volunteers cannot produce seeds in spring crops. Frequent cutting of set-aside plots is also necessary: a single cutting (System 3) only delays seed production without reducing it and thus increases the seed bank 40 times compared to two successive cuttings (System 1). Seed banks can also be regulated via winter and spring crop management. Tilling with a chisel (System 4) instead of a mouldboard plough (System 1) decreases the seed bank because most freshly produced seeds are left close to soil surface where seed survival rates are low. Delaying sowing (Systems 5 vs. 1) also decreases the seed bank via a reduction of both the viable seed bank at sowing and subsequent seedling emergence rates. Larger sowing rates (Systems 6 vs. 1) limit rape seed production in winter crops by favouring interspecific competition between rape volunteers and cropped plants.

Figure 2. Evolution of transgenic rape seedbank of a field cultivated once with the transgenic, herbicide-tolerant rapeseed variety (u) or every four to five years (n)

Table 1. Effect of cultural techniques for set-aside and winter and spring crops on transgenic post-harvest seedbank after four simulated years with a transgenic rapeseed/winter crop/spring crop/set-aside rotation

^a The first sowing date concerns winter crops, the second spring crops.

Conclusion

Despite the consistent results of the first evaluation and simulation steps, GeneSys is not yet ready to be used as a decision aid tool. The major sub-models needing improvement concern pollen dispersal processes and rape volunteer demographics in borders. Furthermore, the effect of rape genotype which is presently restricted to herbicide tolerance and selfpollination rates, should be extended to other rape characteristics (e.g. pod shattering, pollen production) to identify those features able to limit gene flow. Moreover, even though GeneSys was developed for herbicide tolerance dispersal, the model can be used for other genes, such as those coding for fatty acid content. The model therefore constitutes a reflection tool for those concerned with production, transformation and marketing of rapeseed, thus permitting a prospective analysis of the impact of changes in agricultural practices.

References

1. Champolivier J., Messéan A. et Gasquez J;, 1997. Crop management of transgenic rapeseed: risk assessment of gene flow. Bull. GCIRC, 14, 63-66.

2. Chèvre A. M., Eber F., Baranger A. et Renard M., 1997. Gene flow from transgenic crops. Nature, 389, 924.

3. Clermont-Dauphin C., Colbach N. et Meynard J. M. Modelling the influence of cropping system on gene escape from herbicide resistant rapeseed crops to rape volunteers. I. Temporal dynamic and genetic structure of the rape volunteer population in a field.

4. Colbach N. et Meynard J. M., 1996. Modelling the influence of cropping system on gene flow for herbicide resistant rapeseed. Presentation of model structure. 10^e Colloque International sur la Biologie des Mauvaises Herbes, Dijon 11-13 septembre 1996, 223-230.

5. Colbach N., Clermont-Dauphin C. et Meynard J. M. Modelling the influence of cropping system on gene escape from herbicide resistant rapeseed crops to rape volunteers. II. Spatial dynamic and genetic structure of the rape volunteer population in a small region.

6. Couturaud M C., 1998. Effet des systèmes de culture sur les risques de dissémination d’un transgène de colza dans l’environnement: validation et utilisation du modèle GeneSys. Rapport d’ingénieur agronome INA-PG, 50 p.

7. Lefol E., Danielou V. et Darmency H;, 1996. Predicting hybridization between transgenic oilseed rape and wild mustard. Field Crops Res., 45, 153-161.

8. Messéan A., 1996. Impact des plantes résistantes à un herbicide. OCL, 3, 5-9.