Statistical inference for hierarchical models with applications in ecology

在作出任何政策决定之前，对各种生态和环境问题的统计证据进行量化是必不可少的。例如，艾伯塔省东北部的油砂开发对野生动物有影响吗？我们应该如何减轻这种影响？一个物种会在未来50年内灭绝吗？水产养殖对野生鲑鱼有什么影响？这些活动会增加附近社区的癌症发病率吗？生物多样性是否受到人类干扰的影响？在过去的几年里，我开发了一些统计方法，以定量的方式回答了其中的一些问题。本建议发展新的统计模型和客观统计推断程序来解决这些问题。考虑到空间和时间的依赖性，这些模型是复杂的。这些数据通常很大，但除了环境噪声外，还存在测量误差。我们开发的方法使用计算密集型方法克服了这些困难。拟议的研究结果将有助于科学家以有效和客观的方式分析他们的数据，也将有助于引导他们做出有效的管理决策。

Quantifying statistical evidence for various ecological and environmental problems is essential before any policy decisions can be made. For example, is there an impact of oil sands development on wildlife in North East Alberta? How should we mitigate such effect? Would a species go extinct in the next 50 years? What effect does aquaculture have on wild salmon? Would these activities increase cancer rates in nearby communities? Is the biodiversity being affected by human disturbances? In the last several years, I have developed statistical methods that answer some of these questions in a quantitative fashion. This proposal develops new statistical models and objective statistical inference procedures to address these issues. These models are complex taking into account spatial and temporal dependencies. The data generally are large but also have measurement errors in addition to the environmental noise. The methods we develop overcome these difficulties using computationally intensive methods. The results of the proposed research will help scientist analyze their data in an efficient and objective fashion and will also help lead them to make effective management decisions.