ABI Innovation: Advanced mathematical, statistical, and software tools to unlock the potential of animal tracking data

该项目旨在为运动生态学的新兴领域建立一个严格的统计分析基础。对动物运动的理解可以为广泛的生物学主题提供信息，包括人口和社区生态学，动物生理学，疾病传播，基因流以及野生动物管理和保护。从历史上看，缺乏移动数据限制了进展，但跟踪技术的进步促进了收集越来越多物种的高质量移动数据。现在，关键的瓶颈是缺乏好的统计工具来从这些积累的数据源中提取信息。这项研究通过将物理学的运动建模技术与地质统计学的非独立数据的统计方法相结合来解决这一差距。这些下一代分析工具将使生物学家能够解决运动生态学中的四个关键分析类别：1）建模运动路径，2）估计运动学量，如速度和行进距离，3）确定资源和运动之间的驱动关系，以及4）量化动物空间使用。外联工作将针对收集和询问动物运动数据问题的各种生物学家。具体来说，一系列的教程论文将展示方法，涵盖四个分析类别的保护为重点的案例研究，包括非洲丛林大象在肯尼亚和濒危的呼兰在蒙古戈壁沙漠。此外，针对保护从业者和野生动物管理人员将提供深入的培训课程，项目成果将被纳入研究生和本科课程在马里兰州的大学。这项工作结合了最新的进展建模运动作为一个连续的空间，连续的时间随机过程与克里金技术从地质统计学。克里格法是一种统计学上的最佳方法，它可以在有限数量的自相关数据点之间进行概率“填空”。克里格法彻底改变了地质统计学，现在是自相关空间点观测值之间插值的黄金标准。类似地，将克里金法应用于运动数据将使研究人员能够从有限数量的位置观测中概率性地重建运动路径。动物穿越的连续路径的知识是将动物运动数据与广泛的生态信息和保护相关分析联系起来的关键纽带。因此，这些变革性的方法有可能消除目前阻碍运动生态发展的关键障碍。为了充分利用这一潜力，该项目将为R环境开发一套免费软件包，用于统计计算。这些软件包将使用户能够使用该项目开发的强大的基于克里格的分析工具回答各种物种的运动相关问题。项目结果将在http://biology.umd.edu/movement.html上公布。

This project aims to develop a statistically rigorous analytical foundation for the nascent field of movement ecology. An understanding of animal movement can inform a wide range of biological topics including population and community ecology, animal physiology, disease spread, gene flow, and wildlife management and conservation. Historically, a lack of movement data limited progress, but advances in tracking technology have facilitated the collection of high-quality movement data for an ever-growing number of species. Now, the key bottleneck is the dearth of good statistical tools for extracting information from these accumulating data sources. This research addresses this gap by combining movement-modeling techniques from physics with statistical methods for non-independent data from geostatistics. These next-generation analytical tools will allow biologists to tackle the four key analysis categories in movement ecology: 1) modeling movement paths, 2) estimating kinematic quantities such as velocity and distance travelled, 3) identifying driving relationships between resources and movement, and 4) quantifying animal space use. Outreach efforts will target the diverse array of biologists collecting and asking questions of animal movement data. Specifically, a series of tutorial papers will demonstrate methods covering the four analysis categories on conservation-focused case studies including African bush elephants in Kenya and the endangered khulan in the Mongolian Gobi desert. Additionally, an in-depth training course aimed at conservation practitioners and wildlife managers will be offered, and project results will be incorporated into the graduate and undergraduate curriculum at University of Maryland.This work combines recent advances in modeling movement as a continuous space, continuous time stochastic process with kriging techniques from geostatistics. Kriging is a statistically optimal method of probabilistically "filling in the blanks" between a limited number of autocorrelated data points. Kriging revolutionized geostatistics and is now the gold standard for interpolating between autocorrelated spatial point observations. Analogously, adapting kriging to movement data will allow researchers to probabilistically reconstruct movement paths from a limited number of location observations. Knowledge of the continuous path an animal traversed is the critical nexus linking animal movement data to a wide range of ecologically-informative and conservation-relevant analyses. These transformative methods therefore have the potential to remove the key roadblock that is currently holding movement ecology back. To fully capitalize on that potential, this project will develop an integrated suite of freely available software packages for the R environment for statistical computing. These packages will enable users to answer movement related questions for a broad range of species with the powerful krige-based analytical tools the project develops. Project results will be available at http://biology.umd.edu/movement.html.