It must be shown that the potential impact of a transgenic plant has been carefully analyzed and that if it is not neutral or innocuous, it is preferable to the impact of the conventional agricultural technologies that it is designed to replace (Campbell and Cooke 1993; May 1999; Toledo et al. 1995).
Given the limited use of transgenic plants worldwide and the relatively constrained geographic and ecological conditions of their release, concrete information about their actual effects on the environment and on biological diversity is still very sparse. As a consequence there is no consensus as to the seriousness, or even the existence, of any potential environmental harm from GM technology. There is therefore a need for a thorough risk assessment of likely consequences at an early stage in the development of all transgenic plant varieties, as well as for a monitoring system to evaluate these risks in subsequent field tests and releases.
Risk assessments need base line information, including the biology of the species, its ecology and the identification of related species, the new traits resulting from GM technology, and relevant ecological data about the site(s) in which the transgenic plant is intended to be released. This information can be very difficult to obtain in highly diverse environments. Centers of origin or diversity of cultivated plants should receive careful consideration because there will be many wild relatives to which the new traits could be transferred (Ellstrand et al. 1999; Mikkelsen et al. 1996; Scheffler et al. 1993; Van Raamsdonk and Schouten 1997). For special environments, transgenic plants can be developed using technologies that minimise the possibilities of gene flow via pollen and its effects on wild relatives, through the use of male sterility methods or maternal inheritance resulting from chloroplast transformation (Daniell 1999; Daniell et al. 1998; Scott and Wilkinson 1999).
Studies of gene transfer from conventional and transgenic plants to wild relatives and other plants in the ecosystem have so far concentrated on species of economic importance such as wheat, oilseed rape and barley. A virtual absence of data, particularly for