‘MigrateR’: extending model‐driven methods for classifying and quantifying animal movement behavior
‘MigrateR’: extending model‐driven methods for classifying and quantifying animal movement behavior
复制标题
“MigrateR”：扩展用于分类和量化动物运动行为的模型驱动方法
DOI：
--
复制
发表时间：
2017
期刊：
影响因子：
--
通讯作者：
T. Stephenson
中科院分区：
文献类型：
--
作者：
D. B. Spitz;M. Hebblewhite;T. Stephenson
为了有用，动物运动行为（例如迁徙）的定义应该在数量上严格，足够灵活以适应物种生物学的变化（例如纬度与海拔运动），并足够普遍以允许不同物种之间的比较。最近的研究应用了一种模型驱动的方法来对来自全球定位系统（GPS）位置数据的动物运动进行分类和量化。我们通过以下方式改进这些方法：1）修改模型结构，以提供一个简单的生物学防御基础，以减少错误分类; 2）引入一个数据有效的工具，可用于量化和规避模型对起始位置的敏感性; 3）以海拔迁移为例，说明如何调整现有模型来描述短距离迁移。这些改进包括在'migrateR'中，这是一个开源的R包，它扩展并自动化了动物运动行为的模型驱动分类和量化。我们证明了软件和这些改进的方法，使用GPS项圈的位置数据，从一个长距离的移民，麋鹿马鹿，和一个短距离的海拔移民，塞拉利昂内华达州大角羊绵羊加拿大绵羊。我们提供了文本示例代码和补充脚本，说明如何修改默认选项以应对拟合运动模型中的几个常见挑战。
To be useful, definitions of animal movement behavior (e.g. migration) should be quantitatively rigorous, flexible enough to accommodate variation in species biology (e.g. latitudinal vs elevational movement) and sufficiently general to allow comparison among different species. Recent studies have applied a model‐driven approach to classifying and quantifying animal movement from global positioning system (GPS) location data. We improve upon these methods by 1) revising model structure to provide a simple biologically‐defensible basis to reduce misclassification; 2) introducing a data‐efficient tool that can be used to quantify and circumvent model sensitivity to starting location; and 3) illustrating how existing models can be adapted to describe short‐distance migration, using elevational migration as an example. These improvements are included in ‘migrateR’, an open source R package that expands and automates model‐driven classification and quantification of animal movement behavior. We demonstrate the software and these improved methods using GPS‐collar location data from a long‐distance migrant, elk Cervus elaphus, and a short‐distance elevational migrant, Sierra Nevada bighorn sheep Ovis canadensis sierrae. We provide in‐text example code and a supplementary script illustrating how default options can be revised to meet several common challenges in fitting movement models.