贝叶斯推论及其在百合目分子系统学中的应用

Bayesian Inference and Its Aqplication in the Molecular Phylogeny of Liliales

贝叶斯推论作为进化生物学的最新进展,在适用复杂模型、大型数据集、计算速度和结果容易解释等方面明显优于其它算法。本文简要介绍了贝叶斯推论原理及其在分子进化和系统发育研究中的重要性,并使用该方法对百合目主要类群的系统发育关系进行了重建。结果显示,百合目rbcL基因最适合的DNA进化模型为GTR+I+G,贝叶斯法与距离法和最大简约法构建的系统发育树拓扑结构相似,没有显著差异,但是分辨率和支持率明显比后者高。贝叶斯分析结果显示,百合目内划分的7个科,除Smilacaceae科外,其余各科均为高后验概率(PP=1·0)支持的单系类群;文中作者还对各科间的系统关系进行了探讨。

As a new method in evolutionary biology, Bayesian inference of phylogeny appears to have advantages over the other menthods in terms of dealing with complex models and large dataset, interpretation of the results, and computational efficiency. In this paper, we made a brief introduction to the significance of Bayesian inference in the molecular evolution and phylogenetie reconstruction, and its application in the molecular phylogeny of Liliales. The results indicate that the best model for DNA evolution of rbcL sequences in Liliales is GTR ＋ I ＋ G, and the topology of Bayesian tree is similar to those of Maximum parsimonious tree and Neighbor Joining tree, but giving more resolution and higher support values. The Bayesian tree provides strong support （PP= 1.0） for the circumscription of families in Liliales, except for Smilaeaeeae. The interrelationships between families are also discussed.