Department of Horticulture and Crop Science, The Ohio State University, 1680 Madison Av, Wooster, OH, 446591, USA.
Theor Appl Genet. 2023 Mar 23;136(4):70. doi: 10.1007/s00122-023-04279-0.
We documented changes in the wheat genome attributed to genomic selection including loss of diversity, and changes in population structure and linkage disequilibrium patterns. We conclude that training and prediction populations need to co-evolve instead of the use of a static training population. Genomic selection (GS) is widely used in plant breeding to shorten breeding cycles. Our objective was to assess the impact of rapid cycling GS on the wheat genome. We used 3927 markers to genotype a training population (YTP) and individuals from five cycles (YC1-YC5) of GS for grain yield. We assessed changes of allele frequency, genetic distance, population structure, and linkage disequilibrium (LD). We found 27.3% of all markers had a significant allele frequency change by YC5, 18% experienced a significant change attributed to selection, and 9.3% had a significant change due to either drift or selection. A total of 725 of 3927 markers were fixed by YC5 with selection fixing 7.3% of the 725 markers. The genetic distance between cycles increased over time. The Fst value of 0.224 between YTP and YC5 indicates their relationship was low. The number LD blocks decreased over time and the correlation between LD matrices also decreased over time. Overall, we found a reduction in genetic diversity, increased genetic differentiation of cycles from the training population, and restructuring of the LD patterns over cycles. The accuracy of GS depends on the genomic similarity of the training population and the prediction populations. Our results show that the similarity can decline rapidly over cycles of GS and compromise the predictive ability of the YTP-based model. Our results support implementing a GS scheme where the training and prediction populations co-evolve instead of the use of a static training population.
我们记录了与基因组选择相关的小麦基因组变化，包括多样性丧失，以及群体结构和连锁不平衡模式的变化。我们得出结论，训练和预测群体需要共同进化，而不是使用静态训练群体。基因组选择（GS）在植物育种中被广泛用于缩短育种周期。我们的目标是评估快速循环 GS 对小麦基因组的影响。我们使用 3927 个标记对一个训练群体（YTP）和五个 GS 循环（YC1-YC5）的个体进行了基因型分析，以评估其产量。我们评估了等位基因频率、遗传距离、群体结构和连锁不平衡（LD）的变化。我们发现，在 YC5 时，所有标记中有 27.3%的标记的等位基因频率发生了显著变化，18%的标记的变化归因于选择，9.3%的标记的变化归因于漂变或选择。在 YC5 时，共有 725 个标记固定下来，其中 7.3%的固定是由选择引起的。遗传距离随时间增加而增加。YC5 与 YTP 之间的 Fst 值为 0.224，表明它们之间的关系较低。随着时间的推移，LD 块的数量减少，LD 矩阵之间的相关性也随着时间的推移而降低。总的来说，我们发现遗传多样性减少，各循环与训练群体之间的遗传分化增加，LD 模式随循环而重构。GS 的准确性取决于训练群体和预测群体的基因组相似性。我们的结果表明，相似性在 GS 的循环中会迅速下降，从而影响基于 YTP 的模型的预测能力。我们的结果支持实施 GS 方案，其中训练和预测群体共同进化，而不是使用静态训练群体。
