Trait-based modelling to explore processes underlying diversity-functioning relationships

目前在耶拿实验中自信地预测个人、群落和生态系统表现的问题表明，我们对植物特征如何转化为功能的理解有限。我们的总体目标是在耶拿实验的背景下，将植物物种特征与草原生态系统的功能联系起来。我们提出了三种互补的建模方法，易于在耶拿实验的背景下使用关于特征和功能的丰富信息来解释潜在的多样性-功能关系的机制：(1)我们将建立一个统计框架，探索特征和特征相关性对在不同整合水平上操作的过程的影响。为此，我们将使用非线性降维估计函数权衡，并使用得到的流形来计算混合物的互补性、一致性和区分性。将使用方差划分和分层贝叶斯建模来探索在个人、社区和生态系统层面上出现的功能之间的关系。(2)我们将探索由特征信息和实验数据参数化的Lotka-Volterra模型中以相互作用和连接矩阵为特征的相互作用网络结构如何引起不同的多样性-功能机制；(3)将应用一个机械模型(Gemini)来开发一个预测系统，以预测特征如何转化为功能模式，以及特征相关性如何沿互补梯度做出可塑性反应。新实验的设计是为支持这三种方法而量身定做的。

Current problems in confidently predicting individual, community and ecosystem performance in the Jena Experiment point to our limited understanding of how plant traits translate into functioning. Our general objective is to link plant species traits to the functioning of grassland ecosystems in the context of the Jena Experiment. We propose three complementary modelling approaches that are apt to use the abundantly available information on traits and functions in the context of the Jena experiment to elucidate mechanisms underlying diversity-functioning relationships: (1) We will develop a statistical framework that explores the effect of traits and trait correlations on processes operating at different levels of integration. To this aim, we will estimate functional tradeoffs using non-linear dimensionality reductions and use the resulting manifolds to calculate complementarity, identity and distinctiveness of mixtures. The relationships to functions emerging at the individual, community and ecosystem level will be explored using variance partitioning and hierarchical Bayesian modelling. (2) We will explore how the structure of interaction networks characterized by interaction and connectivity matrices in Lotka-Volterra models parameterized from trait information and experimental data evoke different diversity-functioning mechanisms and (3) will apply a mechanistic model (GEMINI) in order to develop a system of predictions how traits translate into patterns of functioning and how trait correlations respond plastically along complementarity gradients. The design of the new experiment was tailored to support these three approaches.