Ecologie Systématique Evolution, Univ. Paris-Sud, CNRS, AgroParisTech, Université Paris-Saclay, 91400, Orsay, France.
Department of Botany and Biodiversity Research, University of Vienna, Rennweg 14, Vienna, A-1030, Austria.
Sci Rep. 2018 Jun 21;8(1):9473. doi: 10.1038/s41598-018-27750-1.
Ancestral state reconstruction is an important tool to study morphological evolution and often involves estimating transition rates among character states. However, various factors, including taxonomic scale and sampling density, may impact transition rate estimation and indirectly also the probability of the state at a given node. Here, we test the influence of rate heterogeneity using maximum likelihood methods on five binary perianth characters, optimized on a phylogenetic tree of angiosperms including 1230 species sampled from all families. We compare the states reconstructed by an equal-rate (Mk1) and a two-rate model (Mk2) fitted either with a single set of rates for the whole tree or as a partitioned model, allowing for different rates on five partitions of the tree. We find strong signal for rate heterogeneity among the five subdivisions for all five characters, but little overall impact of the choice of model on reconstructed ancestral states, which indicates that most of our inferred ancestral states are the same whether heterogeneity is accounted for or not.
祖先状态重建是研究形态进化的重要工具，通常涉及估计特征状态之间的转换率。然而，包括分类学尺度和采样密度在内的各种因素可能会影响转换率的估计，并间接地影响给定节点的状态概率。在这里，我们使用最大似然法测试了五个二分花被性状的速率异质性的影响，这些性状是在包括 1230 个来自所有科的物种的被子植物系统发育树上进行优化的。我们比较了由一个均等速率（Mk1）和一个双速率模型（Mk2）重建的状态，这些模型要么适用于整个树的一组速率，要么适用于分区模型，允许在树的五个分区上使用不同的速率。我们发现，对于所有五个性状，五个细分中都存在强烈的速率异质性信号，但模型选择对重建祖先状态的总体影响很小，这表明无论是否考虑异质性，我们推断的大多数祖先状态都是相同的。
