Collaborative Research on Bayesian Semiparametric Population Dynamics Modeling

人口调节机制很少被很好地理解，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，需要间接的方法来确定这些机制。将开发使用半参数贝叶斯方法的统计模型，以便从观察到的丰度波动中推断出未知的机制。 种群动态模型通常用于保护和管理中的重要决策。然而，目前使用的所有模型都有很大的缺点，即它们需要所有模型组件的规范，无论这些机制是否已知。模型规格中的错误可能会导致收割机损失数百万美元或受威胁物种的灭绝。根据该奖项开发的方法将使保护和管理社区能够产生强大的人口预测，并将大大增加他们的成功机会。这项工作将是一个新的合成的想法，从生态建模和贝叶斯非参数。为了促进这些方法的应用，项目执行人将开发易于使用的软件，通过养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.