为了评估转基因大豆向野大豆（Glycine soja Sieb. et Zucc.）基因漂移后可能产生的生态风险，选取抗草甘膦转基因大豆（携带外源基因cp4-epsps）与野大豆（来自内蒙古包头）杂交F₂代种皮颜色为黑色、褐绿色和褐黄色（与野大豆种皮颜色相近）的种子，对这些种子萌发产生的F₃代植株的19个定量性状和5个定性性状进行分析，在此基础上，将F₃代植株及其亲本进行聚类分析，并对不同组F₃代的定量和定性性状进行分析。结果显示：F₃代幼苗中抗性植株与非抗性植株的分离比符合“5∶1”的孟德尔遗传定律。定量性状中，结实率的变异系数最小（4.65%），其余18个性状的变异系数高于14%，其中，单株种子数的变异系数最大（55.68%）。不同种皮颜色F₂代种子萌发产生的F3代植株均以强缠绕型占比最高；F₃代植株中，90.7%的果荚为弯镰形或弓形，且80.0%的果荚颜色为黑褐色；黑色种皮F₂代种子萌发产生的F₃代种皮颜色无性状分离，褐绿色和褐黄色种皮F₂代种子萌发产生的F₃代种皮颜色均出现性状分离；F₃代种子中，82.7%的种子有泥膜。聚类结果显示F₃代植株被分为3组。不同组F₃代中，Ⅰ组的多数性状最低，且株高等5个性状显著低于野大豆；Ⅱ组的9个性状最高，且单株结荚数等15个性状显著高于野大豆，仅株高显著低于野大豆；Ⅲ组的部分性状最高，且单株结荚数等12个性状显著高于野大豆，株高等3个性状显著低于野大豆。综合比较认为，Ⅱ组的竞争能力最强，Ⅰ组的竞争能力最弱。Ⅰ组的不缠绕型植株占比最高，而Ⅱ和Ⅲ组的强缠绕型植株占比最高。3组F₃代的种皮颜色均以黑色占比最高。综上所述，供试的不同组F₃代植株均能完成生活史并产生后代。若抗草甘膦转基因大豆的基因通过花粉漂移到野大豆上并成功杂交，则会产生不同适应性杂交后代，增加生态风险。

In order to evaluate the potential ecological risk of gene flow from transgenic soybean to wild soybean (Glycine soja Sieb. et Zucc.), seeds of F₂ hybrids between glyphosate-resistant transgenic soybean (carrying foreign gene cp4-epsps) and wild soybean (collected from Baotou of Inner Mongolia) with black, brown green, and brown yellow seed coats (similar to the seed coat color of wild soybean) were selected, and analysis on 19 quantitative traits and 5 qualitative traits of F₃ generation plants germinated from these seeds was performed, on the basis, cluster analysis on F₃ generation plants and their parents was conducted, and analysis on quantitative and qualitative traits of different groups of F₃ generation was carried out. The results show that the segregation ratio of resistant plants and non-resistant plants of F₃ generation seedlings conforms to the Mendel’s law of “5∶1”. Among quantitative traits, the coefficient of variation of seed setting rate is the smallest (4.65%), those of the other 18 traits are higher than 14%, in which, that of seed number per plant is the highest(55.68%). Strong winding type accounts for the highest proportion in F₃ generantion plants germinated from F₂ generation seeds with different seed coat colors; among F₃ generation plants, 90.7% of the pods are bend sickle-shaped or bow, and 80.0% of the pods are black brown; the seed coat color of F₃ generation germinated from F₂ generation seeds with black seed coat has no trait segregation, while that of F₃ generation germinated from F₂ generation seeds with brown green and brown yellow seed coats all have trait segregation; among F₃ generation seeds, 82.7% of the seeds have seed-bloom. The cluster result shows that the F₃ generation plants can be divided into 3 groups. Among different groups of F₃ generation, most traits of group Ⅰ are the lowest, and 5 traits such as plant height are significantly lower than those of wild soybean; 9 traits of group Ⅱ are the highest, and 15 traits such as pod number per plant are significantly higher than those of wild soybean, while only plant height is significantly lower than that of wild soybean; some traits of group Ⅲ are the highest, and 12 traits such as pod number per plant are significantly higher than those of wild soybean, while 3 traits such as plant height are significantly lower than those of wild soybean. The comprehensive comparison indicates that the competitive ability of group Ⅱ is the strongest, while that of group Ⅰ is the weakest. Not winding type accounts for the highest proportion in group Ⅰ, while strong winding type accounts for the highest proportion in groups Ⅱ and Ⅲ. The proportion of black seed coat is the highest in three groups of F₃ generation. In conclusion, different groups of the test F₃ generation plants can all complete their life cycle and produce progenies. If the genes of glyphosate-resistant transgenic soybean flow to wild soybean via pollens and successfully hybridize, hybrids with different adaptabilities will be produced, and the ecological risk will be increased.