Collaborative Research on Bayesian Semiparametric Population Dynamics Modeling

人口调控的机制很少得到很好的理解，而且经常引起激烈的争论。原则上，聪明的实验可以揭示种群数量波动的原因，但对大多数物种来说，这样的实验即使不是不可能，也是不切实际的。因此，需要确定这些机制的间接方法。将开发使用半参数贝叶斯方法的统计模型，允许从观测到的丰度波动推断未知机制。种群动态模型通常用于制定保护和管理方面的重要决策。然而，目前使用的所有模型都有一个很大的缺点，即无论这些机制是否已知，它们都需要对所有模型组件进行规范。模型规格上的错误可能会给采矿者造成数百万美元的损失，或者导致濒危物种的灭绝。根据这项奖励将开发的方法将使保护和管理界能够作出可靠的人口预测，并将大大增加其成功的机会。这项工作将是生态建模和贝叶斯非参数思想的新颖综合。为了促进这些方法的应用，项目管理人员将开发易于使用的软件，这些软件将通过保护和管理会议及其专业网站积极分发。

The mechanisms of population regulation are rarely well understood and often hotly debated. In principle, clever experiments could reveal why populations fluctuate but, for most species, such experiments are impractical, if not impossible. Indirect methods for identifying these mechanisms are therefore needed. Statistical models that use semiparametric Bayesian methods will be developed that allow unknown mechanisms to be inferred from observed fluctuations in abundance. Population dynamics models are routinely used to make important decisions in conservation and management. However, all models currently in use have the substantial drawback that they require the specification of all model components whether or not these mechanisms are known. Mistakes in model specification may result in millions of dollars lost to harvesters or the extinction of threatened species. The methods that will be developed under this award will enable the conservation and management communities to generate robust population projections and will greatly increase their chances of success. This work will be a novel synthesis of ideas from ecological modeling and Bayesian nonparametrics. To foster applications of these methods the PIs will develop easy to use software that will be actively distributed through conservation and management meetings and their professional websites.