From numerical ecology to landscape genetics and genomics

生态群落空间结构是环境条件、群落动态和历史动态共同作用的结果，引起了生态学家的极大兴趣。群落由多变量数据表表示；它们的空间结构可以使用一系列新的统计方法来研究，称为空间特征函数分析。使用这些方法，人们可以区分相互竞争、互补的过程的行动，如环境控制(基于利基)和社区动态(中立)，这些过程构成了社区。在我的研究中，我将群落生态学和统计学相结合，开发了分析自然群落数据的方法。下面是两个例子。(1)相关性分析使研究人员能够确定环境变量对物种的影响显著的规模(S)；我们将推广这一方法，以检验关于一系列环境变量对整个生物群落作用规模的假设。(2)第四角分析将物种的特征与发现它们的环境特征联系起来。我们将把这种方法推广到根据物种特征和环境特征来预测改良生态系统中物种出现的方法。我们的新R函数将允许世界各地的研究人员使用新方法。这项工作将包括一个新的R包，实施社区生态中的空间分析的各种方法。 我们的工作还涉及到系统发育问题。我们将开发探索植物杂交网络的算法，以确定树状和网状进化模型中哪一种最能代表数据。我们将扩展定向空间建模的方法，以模拟包括网状进化网络在内的系统发育关系，并测试系统发育模型中独立成分的重要性。最后，我们将把空间生态学中发展起来的多元分析方法转移到景观遗传学和基因组学中。我在群落生态学和系统发育分析方面的双重含义也将使我能够参与到社区系统发育的新兴领域。

The spatial structure of ecological communities is of great interest to ecologists because it results from the combined action of environmental conditions, community dynamics, and historical dynamics. Communities are represented by multivariate data tables; their spatial structure can be studied using a new family of statistical methods called spatial eigenfunction analysis. Using these methods, one can differentiate the actions of competing, complementary processes like environmental control (niche-based) and community dynamics (neutral) that structure communities. In my research, I combine community ecology and statistics to develop methods to analyse natural community data. Two examples follow. (1) Codependence analysis allows researchers to identify the scale(s) at which the effect of an environmental variable on a species is significant; we will generalize that method to test hypotheses about the scale of action of a whole set of environmental variables on a whole community of organisms. (2) The 4th-corner analysis links the traits of species to the characteristics of the environment where they are found. We will generalize that method to predict species occurrence in a modified ecosystem as a function of the species traits and the characteristics of the environment. Our new R functions will allow researchers around the world to use the new methods. The effort will include a new R package implementing a variety of methods for spatial analysis in community ecology. Our work also involves phylogenetic questions. We will develop algorithms for the exploration of hybridization networks in plants to determine which, of a tree-like and a reticulate evolution model, best represents the data. We will extend a method of directional spatial modelling to model phylogenetic relationships, including reticulated evolutionary networks, and test the significance of independent components of phylogenetic models. Finally, we will transfer methods of multivariate analysis developed in spatial ecology to landscape genetics and genomics. My double implication in community ecology and phylogenetic analysis will also allow me to participate in the emerging field of community phylogenetics.