Manceau Marc, Lambert Amaury, Morlon Héléne
Muséum National d'Histoire Naturelle, 75005 Paris, France.
Center for Interdisciplinary Research in Biology, Collège de France, CNRS UMR 7241, 75005 Paris, France.
Syst Biol. 2017 Jul 1;66(4):551-568. doi: 10.1093/sysbio/syw115.
Models of phenotypic evolution fit to phylogenetic comparative data are widely used to make inferences regarding the tempo and mode of trait evolution. A wide range of models is already available for this type of analysis, and the field is still under active development. One of the most needed development concerns models that better account for the effect of within- and between-clade interspecific interactions on trait evolution, which can result from processes as diverse as competition, predation, parasitism, or mutualism. Here, we begin by developing a very general comparative phylogenetic framework for (multi)-trait evolution that can be applied to both ultrametric and nonultrametric trees. This framework not only encapsulates many previous models of continuous univariate and multivariate phenotypic evolution, but also paves the way for the consideration of a much broader series of models in which lineages coevolve, meaning that trait changes in one lineage are influenced by the value of traits in other, interacting lineages. Next, we provide a standard way for deriving the probabilistic distribution of traits at tip branches under our framework. We show that a multivariate normal distribution remains the expected distribution for a broad class of models accounting for interspecific interactions. Our derivations allow us to fit various models efficiently, and in particular greatly reduce the computation time needed to fit the recently proposed phenotype matching model. Finally, we illustrate the utility of our framework by developing a toy model for mutualistic coevolution. Our framework should foster a new era in the study of coevolution from comparative data.
适用于系统发育比较数据的表型进化模型被广泛用于推断性状进化的节奏和模式。目前已有大量这类分析模型，且该领域仍在积极发展。最急需的发展之一涉及能更好地解释分支内和分支间种间相互作用对性状进化影响的模型，这种影响可能源于竞争、捕食、寄生或互利共生等多种过程。在此，我们首先为（多）性状进化构建一个非常通用的比较系统发育框架，该框架可应用于超度量树和非超度量树。这个框架不仅囊括了许多先前关于连续单变量和多变量表型进化的模型，还为考虑更广泛的一系列模型铺平了道路，在这些模型中，谱系共同进化，即一个谱系中的性状变化受其他相互作用谱系中性状值的影响。接下来，我们提供一种在我们的框架下推导末端分支性状概率分布的标准方法。我们表明，对于一大类考虑种间相互作用的模型，多元正态分布仍是预期分布。我们的推导使我们能够高效地拟合各种模型，特别是大大减少了拟合最近提出的表型匹配模型所需的计算时间。最后，我们通过构建一个互利共生共同进化的简单模型来说明我们框架的实用性。我们的框架应该会开启一个基于比较数据研究共同进化的新时代。
