Gene flow from transgenic crops

Abstract

Gene flow from crops to related wild species must be considered when assessing the potential environmental impact of cultivating genetically modified plants¹. Evidence of pollen dispersal within species has been found for several crops but little information is available on spontaneous gene flow from crops to related species with simultaneous flowering periods^2,3. To study the genetic mechanisms involved, we have developed an intergeneric model of gene flow from transgenic oilseed rape (Brassica napus L.; genotype, AACC; diploid chromosome number, 2n=38) containing one copy of the bar gene, which confers resistance to the herbicide Basta (glufosinate ammonium), to wild radish (Raphanus raphanistrum L.; genotype, RrRr; 2n=18), a widely distributed weed.

Main

We obtained oilseed rape×wild radish F₁interspecific hybrids⁴ and studied the four successive generations under field conditions. The hybrids were surrounded by wild radish plants.

All oilseed rape varieties used as mother plants produced seeds. In the first generation, female fertility of the F₁interspecific hybrids was poor, with only 28.6% producing seeds. The number of seeds was ten times higher at the second generation, and the increased female fertility was confirmed with the third-generation plants. The seed set (ranging from 0 to 5,673 seeds per plant) was, in some plants, close to the observed fertility of wild radish plants (500 to 10,000 seeds per plant; Table 1). In all generations, the percentage of seed germination never limited the production of successive progeny, which ranged from 74.2 to 88.9%.

Table 1 Gene flow analysis from transgenic oilseed rape to wild radish

In agreement with previous data⁴, 99.5% of first-generation interspecific hybrids had the expected genomic structure, ACRr, 2n=28, that is, half of the genome of each parent. Of the second-generation hybrids, 48% corresponded to the structure, ACRrRr, 2n=37. All others had either more chromosomes (33.4% of the total, mainly showing an amphidiploid structure, AACCRrRr, 2n=56) or less chromosomes (18.6% of the total, mainly showing the same chromosome number as the mother plant, ACRr, 2n=28). Female fertility was highly dependent on the chromosome number of the mother plant. Mother plants with 2n=56, 2n=37 and 2n=28 produced from 0.9, 8.4 and 59.7 seeds per plant, respectively. At the third generation, chromosome number generally decreased, with 83.6% of the plants having less than 28 chromosomes. The plants with the lowest chromosome numbers were the most fertile. We confirmed this tendency towards chromosome decrease in the fourth-generation plants, with 89.5% of the plants having less than 27 chromosomes. We observed a chromosome number close to that of wild radish (2n=18) in 25.4% of the plants (Table 1).

Basta resistance in the F₁interspecific hybrids displayed mendelian segregation, that is, a 1:1 ratio of resistant and susceptible plants, as the oilseed rape mother plants were heterozygous for the bartransgene. However, because of unreduced gametes, the bar gene transmission was high in first-generation interspecific hybrids. The bar gene transmission was dependent on the chromosome number of the mother plant and decreased in successive generations (Table 1).

It seems that intergeneric gene flow might mainly occur by transgene introgression within the genome of the weeds, but slowly and at a low probability under natural optimal conditions because four generationswere needed to provide herbicide-resistant plants with a chromosome number and morphology close to that of the weed. It is likely that under normal agricultural conditions this event is rare when the wild radish is the female parent⁵.