Muséum national d'Histoire naturelle, Département Ecologie et Gestion de la Biodiversité, UMR CNRS UPMC 7204, Paris, France.
PLoS One. 2013;8(1):e54530. doi: 10.1371/journal.pone.0054530. Epub 2013 Jan 24.
We developed an approach for analysing the effects of two crossed factors A and B on the functional, taxonomic or phylogenetic composition of communities. The methodology, known as crossed-DPCoA, defines a space where species, communities and the levels of the two factors are organised as a set of points. In this space, the Euclidean distance between two species-specific points is a measure of the (functional, taxonomic or phylogenetic) dissimilarity. The communities are positioned at the centroid of their constitutive species; and the levels of two factors at the centroid of the communities associated with them. We develop two versions for crossed-DPCoA, the first one moves the levels of factor B to the centre of the space and analyses the axes of highest variance in the coordinates of the levels of factor A. It is related to previous ordination approaches such as partial canonical correspondence analysis and partial non-symmetrical correspondence analysis. The second version projects all points on the orthogonal complement of the space generated by the principal axes of factor B. This second version should be preferred when there is an a priori suspicion that factor A and B are associated. We apply the two versions of crossed-DPCoA to analyse the phylogenetic composition of Central European and Mediterranean bird communities. Applying crossed-DPCoA on bird communities supports the hypothesis that allopatric speciation processes during the Quaternary occurred in open and patchily distributed landscapes, while the lack of geographic barriers to dispersal among forest habitats may explain the homogeneity of forest bird communities over the whole western Palaearctic. Generalizing several ordination analyses commonly used in ecology, crossed-DPCoA provides an approach for analysing the effects of crossed factors on functional, taxonomic and phylogenetic diversity, environmental and geographic structure of species niches, and more broadly the role of genetics on population structures.
我们开发了一种方法来分析两个交叉因素 A 和 B 对群落功能、分类或系统发育组成的影响。该方法称为交叉 DPCoA，它定义了一个空间，其中物种、群落和两个因素的水平被组织成一组点。在这个空间中，两个特定物种的点之间的欧几里得距离是（功能、分类或系统发育）差异的度量。群落位于其组成物种的质心处；两个因素的水平位于与其相关的群落的质心处。我们为交叉 DPCoA 开发了两个版本，第一个版本将因素 B 的水平移动到空间的中心，并分析因素 A 坐标中具有最高方差的轴。它与以前的排序方法有关，如部分典范对应分析和部分非对称对应分析。第二个版本将所有点投影到由因素 B 的主轴生成的空间的正交补集上。当有先验怀疑因素 A 和 B 相关时，应优先选择第二个版本。我们应用这两种交叉 DPCoA 版本来分析中欧和地中海鸟类群落的系统发育组成。将交叉 DPCoA 应用于鸟类群落支持了以下假说，即在第四纪期间，异域物种形成过程发生在开放且分散分布的景观中，而森林栖息地之间没有扩散的地理障碍可能解释了整个西古北界森林鸟类群落的同质性。交叉 DPCoA 概括了生态学中常用的几种排序分析方法，为分析交叉因素对功能、分类和系统发育多样性、物种生态位的环境和地理结构以及更广泛的遗传对种群结构的影响提供了一种方法。
