A social genetic effect (SGE) refers to the effect of an individual’s genotype on the phenotypic traits of its social partners. Such effects often work via behavioural interactions; a laying hen may, for example, affect the survival of her group mates via pecking behaviour. Bruce Griffing pioneered research on SGE between 1950 and 1980. Griffing found many important results, but his work was largely overlooked by both breeders and the scientific community. Inspired by the work of Griffing, Bill Muir of Purdue University executed several selection experiments for survival time in laying hens and quail showing cannibalistic behaviour. His comparisons of group and individual selection provide strong support for the theoretical work of Griffing. Muir and co‐workers subsequently developed a mixed model for the estimation of SGE. Inspired by a talk with Bill Muir, I started to work on the theory of SGE around 2004. At hindsight, several results had already been found by Bruce Griffing, but were presented in a conference proceeding.

Empirical work on SGE has included laying hens, pigs, mink, Eucalyptus trees, quail and cod. Studies on laying hens, quail and mink show strong evidence of substantial SGE for traits relating to behavioural interactions. The data structure in these populations, with many small groups and little confounding of groups with environmental factors, is near optimal for the estimation of SGE. Remarkably, the direct‐social genetic correlation shows a strong contrast between laying hens and mink. In laying hens, particularly in crossbreds, the direct‐social genetic correlation for survival time tends to be negative, indicating that “the aggressor survives,” In mink, in contrast, the direct‐social genetic correlation for bite mark score, measured on the inside of pelts after pelting, is strongly positive (~0.80), indicating that individuals benefit from not harming others. These results show that SGE is not always competitive, in which case the term “competitive effects” is quite misleading.

Estimation of SGE is more challenging in pigs, because the number of groups is smaller, and there is often more confounding of groups with environmental factors. Hence, genetic analysis of SGE in pig populations requires careful model comparison and validation. Nevertheless, analyses of large populations of Topigs Norsvin show convincing evidence of SGE for growth rate and feed intake, but not for backfat and loin depth. A one‐generation selection contrast for SGE on growth rate in pigs, performed by Camerlink et al., furthermore suggests that pigs selected for favourable SGE show better social behaviour.

In plants, the spatial structure provides good power to estimate SGE, using the reciprocal of distance as a weight for the SGE in the model, as suggested by Muir. The analysis of SGE on bark diameter in Australian Eucalyptus trees by Costa e Silva et al. indicates very strong competition among neighbouring trees, resulting in near zero heritable variation, despite an ordinary heritability of ~35%. In other words, any genetic increase in individual bark diameter resulted in a reduction of the sum of bark diameters of the neighbours by precisely the same amount. This result shows that SGE can almost completely remove the potential of a population to respond to selection. While SGE is probably more important in plants than in animals, remarkably little research has been done, with the exception of a few studies by Bruce Griffing and by Charles Goodnight.

Because the social environment is a very important component of animal welfare, improvement of SGE should be an integral component of strategies to improve the well‐being of our animals. On the one hand, the trend to larger groups with more behavioural freedom for the animals increases the importance of good social behaviour. On the other hand, however, it severely complicates the estimation of SGE. In small groups, such as traditional battery cages in laying hens, the number of social partners of an individual is limited, so that SGE can be teased out statistically based on the covariance between the phenotypes of the group mates of relatives. When data consist of a few large groups, however, this is impossible, and more information is needed on who interacts with whom.

The next step in the genetic improvement of SGE, therefore, will have to come from the automated detection of behavioural interactions between individuals, with the help of sensors and AI. We need to know who interacts with whom, how often and the consequences of each interaction. The availability of such data would not only greatly advance the breeding for SGE, but could also considerably increase our understanding of animal behaviour. Most importantly, it would allow transforming the study of animal behaviour into a quantitative field, with explicit quantitative models, potentially including genetic terms that provide testable quantitative predictions. The inclusion of genetic effects in such models may provide insights that cannot be obtained without genetic information. As an analogy, the study of SGE in laying hens, for example, demonstrated that ~1/3 of the genetic variation in survival time originates from the direct effect, that is, from the recipient of the pecks. Behavioural studies, in contrast, tend to interpret pecking behaviour purely as a trait of the individual performing the behaviour, ignoring any effect of the recipient.

In conclusion, the integration of recent developments in sensing and AI with quantitative genetics and with quantitative models of animal behaviour holds the promise to a much better understanding of the social interactions in our livestock, particularly when kept in modern animal‐friendly housing systems. Fully exploiting this avenue requires a close collaboration with specialists from the different disciplines, and the transformation of animal behaviour into a quantitative discipline.

社会遗传效应（SGE）是指个体的基因型对其社交伙伴的表型特征的影响。这种影响通常通过行为互动来发挥作用。例如，产蛋鸡可能会通过啄食行为影响同伴的生存。布鲁斯·格里芬（Bruce Griffing）在1950年至1980年间率先开展了SGE研究。格里芬（Griffing）发现了许多重要结果，但育种者和科学界都忽略了他的工作。受到格里芬（Griffing）工作的启发，普渡大学（Purdue University）的比尔·缪尔（Bill Muir）进行了多次选择实验，以研究蛋鸡和鹌鹑的成活时间，从而显示出食人行为。他对小组选择和个人选择的比较为格里芬的理论工作提供了有力的支持。随后，Muir和同事开发了一种混合模型来估算SGE。

SGE的经验工作包括蛋鸡，猪，貂，桉树，鹌鹑和鳕鱼。对蛋鸡，鹌鹑和貂皮的研究表明，与行为相互作用有关的性状有大量的SGE。这些人群中的数据结构，其中有许多小组，几乎没有与环境因素混淆的人群，对于SGE的估算几乎是最佳的。值得注意的是，直接社会遗传相关性显示了蛋鸡和貂皮之间的强烈对比。在蛋鸡中，特别是在杂种鸡中，与生存时间的直接社会遗传相关性趋向于负，表明“侵略者得以幸存”。相比之下，在貂皮中，叮咬分数的直接社会遗传相关性是根据皮发后，皮内强烈阳性（〜0.80），表明个人受益于不伤害他人。这些结果表明，SGE并不总是具有竞争性，在这种情况下，“竞争效应”一词极具误导性。

估计猪的SGE更具挑战性，因为组的数量较小，而且受环境因素影响的组的混淆通常更多。因此，对猪群中SGE的遗传分析需要仔细的模型比较和验证。但是，对Topigs Norsvin的大量种群进行的分析显示，令人信服的证据表明SGE可以促进生长速度和采食量，而不能用于背脂和腰肉深度。Camerlink等人对猪的SGE进行了一代代的选择对比，进一步表明，为获得良好的SGE而选择的猪表现出更好的社会行为。

在植物中，空间结构提供了很好的估算SGE的能力，使用距离的倒数作为模型中SGE的权重，正如Muir所建议的那样。Costa e Silva等人对澳大利亚桉树树皮直径进行SGE分析。表示相邻树木之间的竞争非常激烈，尽管普通遗传力约为35％，但可导致接近零的遗传变异。换句话说，个体树皮直径的任何遗传增加都会导致邻居的树皮直径总和减少恰好相同的数量。该结果表明，SGE几乎可以完全消除种群响应选择的潜力。尽管SGE在植物中比在动物中更重要，但很少进行研究，

由于社会环境是动物福利的一个非常重要的组成部分，因此改善SGE应该是改善动物福祉的战略不可或缺的组成部分。一方面，对动物具有更大行为自由的更大群体的趋势增加了良好的社会行为的重要性。但是，另一方面，它严重地增加了SGE的估算。在小组中，例如在产蛋鸡中使用传统的电池笼时，一个人的社交伙伴数量是有限的，因此可以基于亲戚同伴表型之间的协方差从统计学上挑出SGE。但是，当数据由几个大的组组成时，这是不可能的，并且需要更多有关谁与谁进行交互的信息。

因此，SGE遗传改良的下一步必须来自借助传感器和AI来自动检测个体之间的行为相互作用。我们需要知道谁与谁互动，每次互动的频率和后果。这些数据的可获得性不仅将大大促进SGE的育种，而且还可以大大增加我们对动物行为的了解。最重要的是，这将允许使用明确的定量模型将动物行为的研究转变为定量领域，并可能包括提供可检验的定量预测的遗传术语。在此类模型中包含遗传效应可能会提供没有遗传信息就无法获得的见解。例如，对蛋鸡中SGE的研究就是说，从啄的接收者那里。相比之下，行为研究倾向于将啄食行为完全解释为个体行为的特征，而忽略了接受者的任何影响。

总之，将感官和AI的最新发展与定量遗传学和动物行为定量模型相结合，有望更好地了解我们家畜的社会互动，尤其是当饲养在现代动物友好型住房系统中时。充分利用这一途径需要与不同学科的专家紧密合作，并将动物行为转变为定量学科。