| Summary: | 碩士 === 國立中興大學 === 農藝學系所 === 107 === With the rapid increase of events and cultivated areas of genetically modified (GM) crops, several countries in the world gradually began to pay attention to the studies on the coexistence of GM and non-GM crops. Nowadays many pollen-mediated gene flow models of maize have been developed ,which can be used to predict the GM adventitious presence (AP) at field scale, and to set the proposed isolation distance to make sure AP of non-GM products below official threshold. Using uniform isolation measures for coexistence strategy might effectively reduce the pollution caused by pollen of GM maize, but lead to restrict the use of fields. In recent years, the research of GM and non-GM coexistence mature gradually, the develop of coexistence towards the landscape scale simulation, in order to provide more flexible coexistence strategy to give farmers the freedom to choose in the future. As part of the PRICE project, a new flexible and user-friendly Decision-Support Tool (DST) has been designed, which simulates for the various factors affecting maize cross-pollination and accounts AP at the landscape. The model of DST built in Bayesian framework to provide quality of prediction. In DST, the model consider the bare isolation zones as buffer zones which crop non-GM maize. Because of Taiwanese agriculture classified as small area and intensive farming, which fields adjacent or only separated by road or ridge, if regarded the ridge or road as a buffer to predict could lead to underestimate of AP.
To establish a decision support tool in Taiwan, this study used the Cauchy weighted model, which is suitable for predicting the effects of bare isolation zones and also considered flowering time-lag to be used for asynchronous flowering strategy. For testing fitted ability of Cauchy weighted model, we compared it with log/log model, 2Dt model and compound exponential model with three experimental data with different design in Taiwan Puzi during 2009 to 2010, designed as: all recipient bordered by pollen source, half recipient bordered by pollen source and set 7.5m isolation distance respectively. The results of the study indicated that the Cauchy weighted model had the best ability among the models, and fitted well for all three experimental data. In addition, the Cauchy weighted model could be used to estimate the posterior distribution of AP in a Bayesian framework, so the final model not only could fit the variability of experimental data, but also provide quality of prediction. Finally, we established Taiwanese corn decision support tool T-DST, which used the Cauchy weighted model as the calculation core to consider and explain the isolation effect, time isolation and meteorological factors in small agricultural areas in the hope of providing a more flexible coexistence strategy.
|
|---|
