Direct and reverse pollen-mediated gene flow between GM rice and red rice weed

Several studies have reported transgenic rice transferring transgenes to red rice weed. However, gene flow also occurs in the opposite direction resulting in transgenic seeds that have incorporated the traits of wild red rice. We quantified this reverse flow being higher than the direct gene flow, nevertheless transgenic seeds carrying wild genes would remain in the spike and therefore most of it would be removed at harvesting. This phenomenon must be considered in fields used for elite seed production and in developing countries where there is a higher risk of GM red rice weed infestation increasing from year to year. Abstract Potential risks of genetically modified (GM) crops must be identified before their commercialization, as happens with all new technologies. One of the major concerns is the proper risk assessment of adventitious presence of transgenic material in rice fields due to cross-pollination. Several studies have been conducted in order to quantify pollen-mediated gene flow from transgenic rice (Oryza sativa) to both conventional rice and red rice weed (O. sativa f. spontanea) under field conditions. Some of these studies reported GM pollen-donor rice transferring GM traits to red rice. However, gene flow also occurs in the opposite direction, in a phenomenon that we have called reverse gene flow, resulting in transgenic seeds that have incorporated the traits of wild red rice. We quantified reverse gene flow using material from two field trials. A molecular analysis based on amplified fragment length polymorphisms was carried out, being complemented with a phenotypic identification of red rice traits. In both field trials, the reverse gene flow detected was greater than the direct gene flow. The rate of direct gene flow varied according to the relative proportions of the donor (GM rice) and receptor (red rice) plants and was influenced by wind direction. The ecological impact of reverse gene flow is limited in comparison with that of direct gene flow because non-shattered and non-dormant seeds would be obtained in the first generation. Hybrid seed would remain in the spike and therefore most of it would be removed during harvesting. Nevertheless, this phenomenon must be considered in fields used for elite seed production and in developing countries where farmers often keep some seed for planting the following year. In these cases, there is a higher risk of GM red rice weed infestation increasing from year to year and therefore a proper moni