Abstract: China is the largest rice producer worldwide and is one of the origins of Asian cultivated rice as well. Along with the rapid development of transgenic rice in China, the potential impact of rice transgene flow on the environment and food safety has become one of the major concerns. Gene flow is an important parameter in the risk assessment and regulation of transgenic rice on the scientific basis. In accordance with this situation, we have formed a team and systematically studied the rice transgene flow since 2002. The results obtained in recent ten years are as following: (1) the patterns of transgene flow and the major biological and meteorological factors controlling rice gene flow have been elucidated. Following the prevailing wind direction in rice flowering period, a rectangular design of field experiments were conducted at 3 locations（Sanya, Hainan Island; Guangzhou, Guangdong; and Hangzhou, Zhejiang）in 2-3 years by using a homozygous transgenic line L201 or B2 (sister lines) with bar gene inserted, resistant to herbicide Basta, as a pollen donor, and totally 19 non-transgenic rice as recipients, including male sterile (ms) lines, common rice cultivars (CRC), F1 hybrid rice, and common wild rice (Oryza rufipogon). Results indicated that the frequency of transgene flow to ms lines was the highest, while gene flow to CRC and F1 hybrids was the lowest (less than 1% or 0.1% at parallel plantation). The frequency of transgene flow to O. rufipogon was in between. By comparison, the maximum frequency of gene flow to ms lines is one to three orders of magnitude higher than that to O. rufipogon and CRC. Gene flow frequency decreased exponentially as the distance increase, with a sharp cut-off point at about 1-2 m in Guangzhou and Hangzhou, while it was approximately 5m in Sanya. It indicates that the sharp cut-off point is closely related to the wind speed during rice flowering period at a given location. By using a concentric circle design of field experiment and an ms line BoA with higher outcrossing rate as a recipient, we have been able to clearly quantify the relationship between the gene flow frequency and the wind direction. On the basis of the cumulative data in eight compass sectors, 90%–96% of the cumulative gene flow events occurred in the four downstream prevailing wind sectors, while it was only 4%–10% in the four lateral and upstream prevailing wind sectors. In short, a general conclusion is that the order of magnitude of transgene flow frequency is basically the same as the outcrossing rate of the CRC (generally less than 1%), which means the gene transfer has not added a new additional risk. (2) By using historical meteorological data as an input, a regional applicable model of rice pollen dispersal and gene flow has been established, which is successively used to predict the maximum threshold distances (MTDs) of gene flow in 17 provinces of southern rice growing area in China. The feature of spatial distribution of MTDs shows: from east to west, MTDs gradually decrease; from north to south, MTDs first decrease in the hilly region and then increase again along the southeast costal region. Reason for it is that the spatial distribution is dramatically influenced by the southeast monsoon (seasonal wind direction from the southeast) and the landform structure. (3) We have artificially constructed two mixed populations of O. rufipogon with F1 hybrids of CRC/O. rufipogon derived from transgene (either Bt or bar) flow to investigate the long-term fate of the transgene integrated into common wild rice. It was found that the F1 hybrids of CRC/O. rufipogon totally disappeared within 3-5 years and the Bt or bar gene was not detectable in the mixed population. It is reasonable to speculate that the common wild rice possesses a mechanism of self-protection. (4) The effectiveness in reducing transgene flow by using flowering isolation or a protective cloth-screen in small-scale field trials of transgenic rice was studied and the results discussed. To investigate the degree of flowering synchronization of CRC and O. rufipogon populations in adjacent plantation, a survey has been conducted in Hainan, Guangdong and Guangxi provinces which allow us to establish a corresponding database. In order to further eliminate the rice transgene flow, a biological containment measure - gene split approach has been established. Based on the data obtained in this study and a survey on the internationally published data of gene flow from major crops, we have proposed to use principles of classification management and threshold-value management in the risk assessment and regulation of transgenic rice. Meanwhile, the progress and prospective of rice gene flow are also discussed in this paper.

Key words: rice (Oryza sativa L.) , common wild rice (O. rufipogon) , transgene , pollen dispersal , gene flow , risk assessment , risk management