使用不同統計模式推估基改與非基改玉米共存之隔離距離

Using Different Statistical Models to Estimate the Isolation Distance of the Coexistence for GM and Non-GM Maize

連柏雁(Bo-Yan Lian) ； 郭寶錚(Bo-Jein Kuo)

《作物、環境與生物資訊》 13卷1期 (2016/03) Pp. 25-38

https://doi.org/10.30061/CEB.201603_13(1).0003

摘要

基改(genetically modified; GM)作物能透過花粉與非基改作物異交而造成基因流動(gene flow)，因此世界各國必須重視基改作物與非基改作物的共存(coexistence)，而最常用的共存措施為設置距離隔離。本研究利用2009年至2010 年間，在臺灣嘉義縣朴子市進行三期作玉米花粉飄散試驗之資料，以修正後的高斯煙羽模式(Gaussian plume model)模擬玉米花粉飄散，計算出相對花粉濃度，再藉由gamma 模式將相對花粉濃度進行轉換用來估計異交率，此估計方法稱之為高斯煙羽－gamma 法。經十折交叉驗證結果顯示，模式的決定係數(R2)值平均為0.71。此外，也利用logistic 迴歸模式配適轉換成類別變數的異交率，驗證的結果顯示，模式具至少85%的預測能力。進一步藉由拔靴法(bootstrapping method)及logistic 模式推估隔離距離之結果顯示，在歐盟當前容忍門檻值(the threshold)為0.9%下，則所需的隔離距離各為32.25 m 與47.25 m。這些資訊可提供給臺灣相關單位參考，將有助於未來建立安全的共存措施。

關鍵字

花粉飄散 ； 共存 ； 隔離距離 ； 高斯煙羽模式 ； 拔靴法 ； logistic迴歸模式

並列摘要

Genetically modified (GM) crops can cross-pollinate non-genetically modified (non-GM) crops by pollen. Therefore, this issue made countries around the world to concern about the coexistence between GM and non-GM crops. The commonly used approach of the coexistence is to set the isolation distance. The experimental data of three crop seasons collected in Puzih of Chiayi County from 2009 to 2010 were used. We simulated the pollen dispersal of maize by the modified Gaussian plume model and calculated the relative pollen concentration. Then we employed a gamma model to transform the relative pollen concentration to estimate the cross-pollination rate, called the Gaussian plume-gamma approach. The result of 10-fold cross-validation showed that the average coefficient of determination (R^2) of the model was 0.71. In addition, we employed logistic regression to fit the observed cross-pollination rates which have been transformed into categorical variable. The result of cross-validation showed that the average percentage of correct prediction was over 85%. Furthermore, the isolation distances were 32.25 m and 47.25 m at the threshold of 0.9% estimated by the bootstrapping method and logistic model, respectively. The information can provide a reference to relevant decision-making units in Taiwan for establishing a safe coexistence measure in the future.

並列關鍵字

Pollen dispersal ； Coexistence ； Isolation distance ； Gaussian plume model ； Bootstrapping method ； Logistic regression