Hierarchical Modelling of Complex Ecological Data

生态系统因不同层次的过程相互作用而变得复杂。生物和环境现象可能影响整个社区、社区内的群体或群体中的个体，这些影响也可能随时间而变化。此外，在野外追踪个体的困难意味着这些系统的数据通常是通过反复观察有标记的个体来收集的。在传统的标记-再捕获实验中，从种群中捕获动物个体并用例如鸟腿上的条带或鱼鳍上的标签标记。最近，科学家们正在依靠自然方法来识别具有色素沉着模式的个体，这些色素沉着模式可以通过照片或皮肤、头发或粪便中DNA样本的基因型信息来识别。 我开发了用于分析生态数据的统计方法，帮助生态学家和野生动物管理人员研究人类影响和其他因素在个人，人口和社区层面的环境影响。我特别感兴趣的是创建层次模型，捕捉生态数据的多层次性质，并在实施复杂的计算机算法是必要的，我将在未来五年内在这一领域的三个主题。首先，我将扩展现有的模型，以考虑到通过计算机匹配照片识别个人时可能出现的错误，并开发新的计算方法，以使这些模型适用于大型数据集。第二，我将开发新的模型，利用标记个体的数据研究人群中的社会行为，并确定这种行为可能对个体移动和疾病传播产生的影响。最后，我将研究可以普遍应用于分析非常大的标记重捕获数据集的新方法。 我开发的工具将直接影响研究人员和野生动物管理人员研究野生动物种群的工作。它们将使这些研究人员能够回答有关这些种群动态和影响不同层次生态系统的因素的新问题。反过来，这将有助于管理受到气候变化、栖息地丧失和其他人类影响威胁的人口。

Ecological systems are complicated by the interplay of processes at distinct levels. Biological and environmental phenomena may affect entire communities, populations within a community, or individuals within a population, and these effects may also vary over time. Moreover, the difficulties in tracking individuals in the wild means that data from these systems are often collected by repeatedly observing marked individuals. In traditional mark-recapture experiments, animals individuals from the population are physically captured and marked with, for example, bands on a birds legs or tags in a fishes fin. More recently, scientists are relying on natural methods to identify individuals with pigmentation patterns that can be identified from photographs or genotype information available from DNA samples in skin, hair, or scat. I develop statistical methods for analysing ecological data that help ecologists and wildlife managers to study the effects of human impacts and other factors in the environment at the individual, population, and community levels. I am particularly interested in creating hierarchical models that capture the multi-level nature of ecological data and on implementing the complex computer algorithms that are necessary, and I will work on three themes within this area in the next five years. First, I will extend available models to account for possible errors when individuals are identified through computerized matching of photographs and develop new computing methods to fit these models to large data sets. Second, I will develop new models to study social behaviour within a population using data from marked individuals and to identify the impacts this behaviour might have on individual movements and the transmission of disease. Finally, I will work on new methods that can be applied generally to analyse very large mark-recapture data sets. The tools I develop will directly impact the work of researchers and wildlife managers studying populations of wild animals. They will allow these researchers to answer new questions about these populations dynamics and the factors that affect ecological systems at different levels. In turn, this will help to manage populations threatened by the effects of climate change, habitat loss, and other human impacts.