Leaders of war: the evolution of collective decision-making in the face of intergroup conflict

大自然中有很多令人惊叹的动物集体行为的例子：鸟群在傍晚的阳光下起伏，鱼群紧紧地绕在一起以避免被捕食，蚂蚁沿着沿着复杂的小径探索和利用它们的环境。在过去的几十年里，生物学家已经发现了简单的行为决策规则如何产生这些复杂的群体行为。类似的决策规则已经被发现可以预测人群和交通堵塞中人类的集体运动，这引起了巨大的跨学科兴趣，认为简单的普遍规则可以解释非常广泛的自然现象，并且这些规则可以通过动物行为的研究来解释。对于形成稳定社会群体的动物来说，集体行为的核心问题是如何在去哪里和做什么上达成一致，尽管群体成员之间存在利益和偏好冲突。在许多动物中，这一问题通过领导者的出现得到解决--领导者对其他动物的行为和运动具有不成比例的影响力。例如，面对在哪里寻找食物或如何避免被捕食的不确定性，个体动物可能会尽最大努力跟随一个群体中最有经验或最知情的成员，或者那些对自己的健康最有价值的人。到目前为止，关于集体决策和领导的研究几乎只集中在单个群体上，而与其他群体隔绝。但现在有广泛的证据表明，群体间的相互作用可以对社会动物的群体内行为产生强大的影响，从而影响可能的决策模式。例如，考虑到协调对战斗中的群体成功至关重要，我们可能会期望与觅食群体相比，战斗群体的决策会变得更加迅速和独裁。另一方面，考虑到所涉及的风险，追随者应该对他们追随的个人进行严格的选择，这可能有利于在这些重要决定上更均匀地分享权力。群际冲突可能对领导和追随的模式以及群体能够就统一行动达成一致的速度产生深远的影响，但迄今为止，对这些群际冲突对集体决策的影响的性质几乎一无所知。我们将通过发展新的理论来预测群际竞争如何塑造移动的动物群体中领导和追随的演变来解决这一知识缺口;并通过使用无人机（无人驾驶飞行器，或UAV）对一个理想的野生哺乳动物系统--带状猫鼬进行测试，来验证我们的预测。带状猫鼬生活在高度领土的群体中，经常与邻近群体发生暴力冲突。与猫鼬、黑猩猩或（据我们所知）其他非人类哺乳动物相比，带状猫鼬的群体间冲突更为严重。然而，群体之间和整个生殖周期的群体间冲突的激烈程度以及男性和女性之间的群体内冲突程度也有很大差异。群体间和群体内冲突的这种变化提供了一个机会，可以利用实验和对群体间冲突之前、期间和之后集体运动的自然观察来检验我们的理论预测。在这个过程中，我们将展示如何将无人机视频与新的“深度学习”人工智能方法相结合，用于自动识别单个动物并跟踪它们在野外的运动-为野生动物管理和畜牧业的许多潜在应用打开大门。该项目的成果将是提高对集体行为和群体间冲突的理解，具有超越进化和行为生态学的广泛跨学科影响，例如在经济学、心理学、政治学和计算机科学方面。

Nature abounds with breathtaking examples of collective animal behaviour: bird flocks undulating in the evening light, schools of fish swirling tightly together to avoid predation, ants teeming along intricate trails to explore and exploit their environment. In the last few decades biologists have discovered how simple behavioural decision rules produce these complex group-level behaviours. Similar decision rules have been found to predict the collective movements of humans in crowds and traffic jams, generating huge interdisciplinary interest in the idea that a very wide range of natural phenomena can be explained by simple universal rules - and that these rules are discoverable through research on animal behaviour. For animals that form stable social groups, the central problem of collective behaviour is how to agree on where to go and what to do, despite conflicting interests and preferences among group members. In many animals this problem is solved through the emergence of leaders - individuals with disproportionate influence on the behaviour and movement of others. For example, in the face of uncertainty about where to find food or how to avoid predation, individual animals may do best to follow the most experienced or best informed members of a group, or those that are most valuable to their own fitness. Up to now, research on collective decision-making and leadership has focused almost exclusively on single groups in isolation of all others. But there is now widespread evidence that intergroup interactions can exert a powerful influence on within-group behaviour in social animals, and hence the likely patterns of decision-making. For example, given that coordination is vital to group success in combat, we might expect a shift to more rapid, dictatorial decision-making in groups that are fighting compared to groups that are foraging. On the other hand, followers should be extremely choosy about which individuals they follow into battle, given the risks involved, which could favour more even power sharing over such vital decisions. Intergroup conflict is likely to have profound impacts on patterns of leadership and followership and the speed with which groups are able to agree on unified action, but to date almost nothing is known about the nature of these intergroup impacts on collective decision-making.We will tackle this gap in knowledge by developing new theory to predict how intergroup competition shapes the evolution of leadership and followership in mobile animal groups; and by testing our predictions using drones (unmanned aerial vehicles, or UAVs) on an ideally suited wild mammal system, the banded mongoose. Banded mongooses live in highly territorial groups which engage in frequent violent conflicts with neighbouring groups. Intergroup conflict is more severe in banded mongooses than in meerkats, chimpanzees, or (to our knowledge) any other non-human mammal. Yet there is also great variation in the intensity of intergroup conflict between groups and across the reproductive cycle, and in levels of within-group conflict between the males and females. This variation in inter- and intragroup conflict provides an opportunity to test our theoretical predictions using experiments and natural observations of collective movement before, during, and after intergroup encounters. In the process we will demonstrate how video from UAVs combined with new 'deep learning' artificial intelligence methods can be used to automatically recognise individual animals and track their movements in the wild - opening the door to a host of potential applications in wildlife management and husbandry. The outcome of the project will be an improved understanding of collective behaviour and intergroup conflict, with broad interdisciplinary implications beyond evolutionary and behavioural ecology, for example, in economics psychology, political science, and computer science.