Statistical Methods and Computational Tools for Marine Animal Movement, Distribution and Population Size

在海洋环境中收集的复杂时空数据越来越多。这使海洋科学家（包括生态学家、自然资源保护者和渔业科学家）能够在比以往更精细的空间和/或时间尺度上提出问题，例如，检查特定渔业管理区内而不是区域之间的丰度变化。这些新的数据源，有些来自快速发展的数字技术（例如，用于海洋动物跟踪的全球定位系统）和来自迄今未充分利用的公民科学努力的其它系统（例如，夏季在新斯科舍省海滩上看到水母），以及伴随它们的重要问题要求在时空建模方面取得进展。管理和保护海洋动物具有挑战性，因为它们会迁移，随着时间的推移改变它们的分布，并且难以观察。 我的研究项目侧重于开发方法和计算工具，以回答与海洋保护和管理有关的重要科学问题。我的研究的具体短期目标是：1）开发用于选择正确数量的行为状态和评估适合于动物运动的状态空间模型（SSM）和隐马尔可夫模型（HMM）框架的拟合优度的方法; 2）将空间（a）直接将（和其他）信息纳入种群动态模型，例如更准确地确定副渔获物热点;（3）发展近亲标记重捕法（CKMR），以估计具保育价值的物种的种群数量。这些目标将通过推进标准方法来实现（例如，动物运动的SSM），批判性地评估竞争方法（例如，用于捕获和/或描述空间依赖性的那些），以及面对处于其婴儿期的具有挑战性的统计推断设置（例如，通过精心挑选的案例研究。将尽一切努力确保对离群值和其他偏离模型假设的小偏差的稳健性。将开发的方法将广泛适用，但我将集中在物种的保护和管理的关注在加拿大大西洋。我研究的长期目标是不断提供统计方法和支持计算工具，确保加拿大能够谨慎管理其海洋资源。

Increasingly large amounts of complex spatiotemporal data are being collected in the marine environment. This has enabled ocean scientists (including ecologists, conservationists, and fisheries scientists) to ask questions at finer spatial and/or temporal scales than previously possible, for example to examine variation in abundance within a particular fisheries management area as opposed to between areas. These new data sources, some derived from rapidly advancing digital technologies (e.g., global positioning systems for marine animal tracking) and others from hitherto under-utilized citizen science efforts (e.g., summer jellyfish sightings on Nova Scotia beaches), and the important questions that accompany them demand advancements in spatiotemporal modelling. Managing and conserving marine animals is challenging because they migrate, change their distributions through time, and are difficult to observe. My research program focuses on developing methodology and computational tools to answer important scientific questions related to marine conservation and management. The specific short-term objectives of my research are to: 1) develop methods for choosing the correct number of behavioural states and assessing goodness of fit suitable for both state-space model (SSM) and hidden Markov model (HMM) frameworks for animal movement; 2) incorporate spatial (and other) information into population dynamics models directly, for example to more accurately identify by-catch hotspots; and 3) develop Close Kin Mark Recapture (CKMR) methods for estimating population size for particular species of conservation concern. These objectives will be achieved by advancing standard approaches (e.g., SSMs for animal movement), critically assessing competing approaches (e.g., those for capturing and/or describing spatial dependence) through simulation, and confronting challenging statistical inference settings that are in their infancy (e.g., CKMR) by way of carefully chosen case studies. Every effort will be made to ensure robustness to outliers and other small departures from model assumptions. The methods to be developed will be broadly applicable, but I will focus on species of conservation and management concern in Atlantic Canada. The long-term objective of my research is to continually make available statistical methods and supporting computational tools that ensure Canada can steward its ocean resources with care.