Numerous methods have been suggested to incorporate crossbred (CB) phenotypes and genotypes into swine selection programs, yet little research has focused on the implicit trade-off decisions between generating data at the nucleus or commercial level. The aim of this study was to investigate the impact of altering the proportion of purebred (PB) and CB phenotypes and genotypes in genetic evaluations on the response to selection of CB performance. Assuming CB and PB performance with moderate heritabilities (⁠$h2=0.4$⁠), a three-breed swine crossbreeding scheme was simulated and selection was practiced for six generations, where the goal was to increase CB performance. Phenotypes, genotypes, and pedigrees for three PB breeds (25 and 175 mating males and females for each breed, respectively), F₁ crosses (400 mating females), and terminal cross progeny (2,500) were simulated. The genome consisted of 18 chromosomes with 1,800 quantitative trait loci and 72k single nucleotide polymorphism (SNP) markers. Selection was performed in PB breeds using estimated breeding value for each phenotyping/genotyping strategy. Strategies investigated were: 1) increasing the proportion of CB with genotypes, phenotypes, and sire pedigree relationships, 2) decreasing the proportion of PB phenotypes and genotypes, and 3) altering the genetic correlation between PB and CB performance (⁠$rpc$⁠). Each unique $rpc$ scenario and data collection strategy was replicated 10 times. Results showed that including CB data improved the CB performance regardless of $rpc$ or data collection strategy compared with when no CB data were included. Compared with using only PB information, including 10% of CB progeny per generation with sire pedigrees and phenotypes increased the response in CB phenotype by 134%, 55%, 33%, 23%, and 21% when $rpc$ was 0.1, 0.3, 0.5, 0.7, and 0.9, respectively. When the same 10% of CB progeny were also genotyped, CB performance increased by 243%, 54%, 38%, 23%, and 20% when the $rpc$ was 0.1, 0.3, 0.5, 0.7, and 0.9, respectively, compared with when no CB data were utilized. Minimal change was observed in the average CB phenotype when PB phenotypes were included or proportionally removed when CB were genotyped. Removal of both PB phenotypes and genotypes when CB were genotyped greatly reduced the response in CB performance. In practice, the optimal inclusion rate of CB and PB data depends upon the genetic correlation between CB and PB animals and the expense of additional CB data collection compared with the economic benefit associated with increased CB performance.

中文翻译：

---

杂交表型和基因型在细胞核选择程序中的包含率的影响

已经提出了许多将杂交 ( CB ) 表型和基因型纳入猪选择程序的方法，但很少有研究关注在核心或商业级别生成数据之间的隐式权衡决策。本研究的目的是调查在遗传评估中改变纯种 ( PB ) 和 CB 表型和基因型的比例对选择 CB 性能的反应的影响。假设具有中等遗传力的 CB 和 PB 性能 ( ?$H2=0.4$⁠ )，模拟了一个三品种猪杂交方案，并进行了六代选择，其目标是提高 CB 性能。模拟了三个 PB 品种（每个品种分别有 25 和 175 只交配雄性和雌性）、F ₁杂交（400 只交配雌性）和终端杂交后代 (2,500) 的表型、基因型和谱系。基因组由 18 条染色体组成，具有 1,800 个数量性状基因座和 72k 个单核苷酸多态性 ( SNP) 标记。使用每个表型分型/基因分型策略的估计育种值在 PB 品种中进行选择。研究的策略是：1) 增加 CB 与基因型、表型和父系谱系关系的比例，2) 降低 PB 表型和基因型的比例，以及 3) 改变 PB 和 CB 性能之间的遗传相关性 ( ?$r磷C$⁠）。每一个独一无二的$r磷C$场景和数据收集策略被复制 10 次。结果表明，包括 CB 数据提高了 CB 性能，而不管$r磷C$或与不包含 CB 数据时相比的数据收集策略。与仅使用 PB 信息相比，包括每代 10% 的具有父系谱系和表型的 CB 后代将 CB 表型的响应提高了 134%、55%、33%、23% 和 21%，当$r磷C$分别为 0.1、0.3、0.5、0.7 和 0.9。当同样 10% 的 CB 后代也进行基因分型时，CB 性能分别提高了 243%、54%、38%、23% 和 20%。$r磷C$与不使用 CB 数据时相比，分别为 0.1、0.3、0.5、0.7 和 0.9。当对 CB 进行基因分型时，包括或按比例去除 PB 表型时，在平均 CB 表型中观察到的变化最小。当对 CB 进行基因分型时，同时去除 PB 表型和基因型大大降低了对 CB 性能的响应。在实践中，CB 和 PB 数据的最佳包含率取决于 CB 和 PB 动物之间的遗传相关性，以及与提高 CB 性能相关的经济效益相比，额外的 CB 数据收集的费用。