Department of Biological Sciences, University of Alberta, Edmonton, Alberta, T6G 2E9, Canada.
Institute of Arctic Biology, University of Alaska, Fairbanks, Alaska, 99775, USA.
Ecology. 2017 Jan;98(1):32-47. doi: 10.1002/ecy.1615. Epub 2016 Nov 28.
The home-range concept is central in animal ecology and behavior, and numerous mechanistic models have been developed to understand home range formation and maintenance. These mechanistic models usually assume a single, contiguous home range. Here we describe and implement a simple home-range model that can accommodate multiple home-range centers, form complex shapes, allow discontinuities in use patterns, and infer how external and internal variables affect movement and use patterns. The model assumes individuals associate with two or more home-range centers and move among them with some estimable probability. Movement in and around home-range centers is governed by a two-dimensional Ornstein-Uhlenbeck process, while transitions between centers are modeled as a stochastic state-switching process. We augmented this base model by introducing environmental and demographic covariates that modify transition probabilities between home-range centers and can be estimated to provide insight into the movement process. We demonstrate the model using telemetry data from sea otters (Enhydra lutris) in California. The model was fit using a Bayesian Markov Chain Monte Carlo method, which estimated transition probabilities, as well as unique Ornstein-Uhlenbeck diffusion and centralizing tendency parameters. Estimated parameters could then be used to simulate movement and space use that was virtually indistinguishable from real data. We used Deviance Information Criterion (DIC) scores to assess model fit and determined that both wind and reproductive status were predictive of transitions between home-range centers. Females were less likely to move between home-range centers on windy days, less likely to move between centers when tending pups, and much more likely to move between centers just after weaning a pup. These tendencies are predicted by theoretical movement rules but were not previously known and show that our model can extract meaningful behavioral insight from complex movement data.
家域概念是动物生态学和行为学的核心，已经开发出许多机制模型来理解家域形成和维持。这些机制模型通常假设一个单一的、连续的家域。在这里，我们描述并实现了一个简单的家域模型，该模型可以容纳多个家域中心，形成复杂的形状，允许使用模式的不连续性，并推断外部和内部变量如何影响运动和使用模式。该模型假设个体与两个或更多个家域中心相关联，并以一定的可估计概率在它们之间移动。家域中心内和周围的运动由二维 Ornstein-Uhlenbeck 过程控制，而中心之间的转换则建模为随机状态切换过程。我们通过引入环境和人口统计学协变量来扩展这个基本模型，这些协变量可以修改家域中心之间的转换概率，并可以进行估计，以深入了解运动过程。我们使用加利福尼亚海獭（Enhydra lutris）的遥测数据来演示该模型。该模型使用贝叶斯马尔可夫链蒙特卡罗方法进行拟合，该方法估计了转换概率，以及独特的 Ornstein-Uhlenbeck 扩散和集中趋势参数。然后，可以使用估计的参数来模拟运动和空间使用，这些模拟与实际数据几乎无法区分。我们使用偏差信息准则（DIC）分数来评估模型拟合度，并确定风和繁殖状态都可以预测家域中心之间的转换。在大风天，雌性不太可能在家域中心之间移动，在照顾幼崽时不太可能在中心之间移动，而在刚断奶幼崽后更有可能在中心之间移动。这些趋势是由理论运动规则预测的，但以前并不为人所知，这表明我们的模型可以从复杂的运动数据中提取有意义的行为洞察力。
