Establishing the pollen-mediated gene flow model for the simulated GM rice: a case study for Sinwu Township, Taoyuan County
| Autor: | Jung-Chiao Hung, 洪溶鍬 |
|---|---|
| Rok vydání: | 2015 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 103 If genetically modified (GM) crops outcross with wild relatives, cultivars and weeds, great impacts on the natural environment and biodiversity may arise. Rice is one of the most important crops in the world and the staple food for over half of the global population. Although rice is a self-pollinated crop, its transgenes can result in gene flow through the pollen dispersal. Up to now, no GM crop could be planted legally in the open field in Taiwan. Therefore, in order to provide the related information on the regulations of coexistence between GM and non-GM rice to the decision-maker, it is important to conduct the experiments of gene flow for GM rice. To simulate the pollen dispersal of GM rice, the field experiments were conducted at Taoyuan District Agricultural Research and Extension Station in Taoyuan County, Taiwan in 2012. The field experiments were a concentric circle design. In the center of the field, the non-glutinous rice named “Taikeng 14” was planted as the pollen donor. In addition, the glutinous rice named “Taoyuan glutinous 2” was planted as the pollen recipient in the circular ring of the field. The iodine test of glutinous and non-glutinous rice was used to distinguish the outcross seeds and self-pollinating seeds. Not only did our study use the log/log model, Gaussian plume model and M5’ model tree to establish the pollen dispersal model in rice, but the actual gene flow frequency was classified into “safe” and “excessive” by different methods and the entire accuracy rate of CART decision tree and Logistic regression model was also calculated. According to the results, the fitting ability of M5’ model tree was the best in estimating gene flow frequency, following by the Gaussian plume model and log/log model. Additionally, the performance of CART decision tree was better than Logistic regression model. As expected, the distance between the edge of pollen donor plot and pollen receipt was the first explanatory variable in M5’ model tree and CART decision tree. Similarly, it was also the important variable in Logistic regression model suggesting that distance was the important factor of gene flow frequency in rice. Furthermore, the prevailing wind direction in flowering period influenced the gene flow frequency in different directions of the experimental site. Wind tunnel and appropriated wind percent were selected as the classification variables. Similarly, the appropriated wind percent and average wind speed of prevailing wind were also the important variables in the Logistic regression model. These results suggested that wind direction and speed were indeed the important factors influencing the gene flow frequency in rice. Therefore, in order to reduce the gene flow frequency, the prevailing wind direction and speed during the flowering period should be concerned first, and the isolation distance could be set up. If the labeling thresholds were set at 0.1%, 0.3% and 0.5%, the isolation distances calculated by the Logistic regression model were 11.1 m, 6 m and 5.7 m, respectively. Moreover, the isolation distance can be applied to the area which microclimate is similar to Sinwu Township, Taoyuan County. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|