The Way They Move: Towards a General Framework for Understanding Animal Movement in Changing Environments

几乎所有的动物都需要移动才能找到食物、配偶、住所和其他生活必需品。然而，在移动过程中，它们正在消耗能量，并将自己暴露在竞争对手和捕食者等危险之中。移动的动物也可能在不知不觉中使其他动物暴露在危险之中/例如，它们可能携带传染病(例如，想想禽流感或獾的牛结核病)。想一想整个动物种群，它们都在决定下一步搬到哪里：很明显，单个动物使用的决策策略对种群的动态有着根本性的影响。同样，当多个物种在群落中相互作用时，个体动物的运动策略将从根本上影响群落的动态。因此，动物运动是将空间野生动物系统联系在一起的粘合剂。了解自然环境中动物运动的最好方法之一是在动物身上安装一个轻型无线电或卫星发射器。这项技术发展非常迅速，现在已经有可能对个人进行长距离和长时间的跟踪。在一些物种中，由于内置了GPS(地理定位系统/类似于汽车导航系统)的标签，这些位置可以以惊人的精度进行测量。跟踪技术的快速发展还不能与分析方法的类似发展相匹配。首先，大多数目前的分析侧重于描述数据，根据在不同栖息地移动的距离、家的范围大小等/而不是试图理解产生移动的过程。其次，大多数目前的方法忽略了由于收集数据的方式而产生的问题，例如测量误差、测量的时间，以及经常在少数人身上进行多次重复测量的事实。最后，当前试图处理这些问题的方法在面对目前可用的巨大数据集时往往会失败。我们工作的目标是：1)开发新的方法来建模和理解动物运动。我们的前提是：i)动物运动的复杂性可以分解为几个一般的运动策略；ii)动物在这些策略之间切换，因为它们受到内外环境变化的影响。以这种方式思考运动的一个好处是，我们可以预测由于环境变化而导致的运动行为的变化，例如通过气候变化。2)开发先进的统计工具，使这些模型能够应用于真实数据。我们建议利用和扩展通常被称为蒙特卡罗方法的先进的计算机模拟技术。这些方法可以很好地处理复杂模型和大数据集的分析。由于可用计算能力的快速增长，它们是统计研究的重点，我们打算以最新的发展为基础，例如允许多个计算机处理器并行应用于同一问题的方法。我们的想法和方法将在三个案例研究中得到检验。第一项是对120只无线电领麋鹿的活动进行长期研究，这些麋鹿成群结队或单独活动的时间长短不一。这使我们能够研究人口流动以及最终的人口动态如何受到社会力量的影响。第二个是一个高分辨率的数据集，显示了单个麋鹿和捕食性狼的活动。我们认为这是一片恐惧的景象，捕食者正在调整它们的行动和景观使用，以最大限度地减少与猎物的接触，而这反过来又试图变得不可预测。最后，我们有迁徙的塞伦盖蒂角马的活动数据，以及相应的饲料和降雨量数据，这将使我们能够研究这些动物如何决定使用哪条迁徙路线/一个具有重要保护意义的问题。

Almost all animals need to move in order to find food, mates, shelter and other necessities of life. Yet in moving they are expending energy, and exposing themselves to dangers such as competitors and predators. Moving animals may also unwittingly expose others to danger / for example they may carry infections diseases (think of avian flu or bovine TB in badgers, for example). Think of a whole population of animals, all making decisions about where to move next: it is clear that the decision strategies that individual animals use has a fundamental effect on the dynamics of the population. Similarly, when multiple species interact in communities, individual animal movement strategies will fundamentally affect the community dynamics. Animal movements are therefore the glue that keeps spatial wildlife systems together.One of the best ways to find out about animal movement in a natural environment is to attach a lightweight radio or satellite transmitter to an animal. This technology has been developing very rapidly and it is now possible to track individuals over long distances and time periods. In some species the locations can be measured with amazing accuracy, thanks to tags with built-in GPSs (Geographic Positioning Systems / like those in car navigation systems). The rapid development in tracking technology has not yet been matched by similar developments in analysis methods. Firstly, most current analyses focus on describing the data, in terms of distances moved in different habitats, home range sizes, etc / rather than attempting to understand the processes that generate the movements. Secondly, most current methods ignore issues that arise due to the way the data were collected, such as measurement error, timing of measurements and the fact that often many repeated measurements are made on few individuals. Lastly, current methods that do attempt to deal with these issues often break down when faced with the enormous datasets that are currently available.The goals of our work are :1) To develop new ways to model and understand animal movements. Our premises are that i) the complexities of animal movement can be dissected into a few general movement strategies; and ii) animals switch among these strategies as they are affected by changes in the internal and external environment. One advantage of thinking of movement in this way is that we can predict changes in movement behaviour as a result of changes in environment, for example through climate change.2) To develop advanced statistical tools that allow these models to be applied to real data. We propose to harness and extend advanced computer simulation techniques known generally as Monte Carlo methods . These methods can cope well with the analysis of complex models and large datasets. They are an intense focus of statistical research due to the rapid increase in available computing power, and we intend to build upon the latest developments, such as methods that allow multiple computer processors to be applied in parallel to the same problem.Our ideas and methods will be tested in three case studies. The first is a long-term study of movement of 120 radio collared elk, who spend variable amounts of time in groups or alone. This allows us to study how movement, and ultimately population dynamics, is affected by social forces. The second is a high-resolution dataset showing movements of individual elk and predatory wolf. We think of this as a landscape of fear where the predators are adjusting their movements and landscape use to maximise their encounter with prey which in turn tries to become unpredictable. Lastly, we have movement data for migrating Serengeti wilderbeest, as well as corresponding forage and rainfall data that will allow us to study how these animals determine which migration route to use / a question with important conservation implications.