Rooting phylogenetic trees with distant outgroups: A case study from the commelinoid monocots
Rooting phylogenetic trees with distant outgroups: A case study from the commelinoid monocots
复制标题
DOI：
10.1093/oxfordjournals.molbev.a003999
复制
发表时间：
2002-10-01
影响因子：
10.7
通讯作者：
Barrett, SCH
中科院分区：
文献类型：
--
作者：
Graham, SW;Olmstead, RG;Barrett, SCH
系统发育生根实验表明，来自鸭跖草类单子叶类群的两个叶绿体基因代表了梭鱼草科 Pontederiaceae 的现存近亲，保留了有关该科根位置的可测量信号。将叶绿体序列的生根偏好与人工序列的生根偏好进行比较，这些人工序列对应于外群的差异如此之大，以至于它们的信号已完全丢失。这些随机序列更喜欢无根内群拓扑中的三个最长分支，并且不会优先在真实外群序列所青睐的分支上扎根。然而，人工序列的生根行为并不是分支长度的简单函数。随机外群优先在长末端内群分支上扎根，但许多内群分支的长度与随机序列所青睐的分支相当，不会吸引或很少吸引命中。通常避免非末端内群分支，无论其长度如何。将序列强制到次优根的容易程度的比较表明，对于大约一半的最不简约的候选根，真正的外群比随机外群需要更大的生根惩罚。虽然这支持了真实外群中非随机信号的存在，但它也表明在最佳和接近最佳生根可能性之间进行选择的能力很小。基于似然的检验拒绝这样的假设：使用真实外群序列的子树的所有根都是对数据同样好的解释，并且还消除了大约一半的最不最优候选根。添加基因或外群可以提高区分不同根位置的能力。一般来说，对于更密切相关的外群体来说，根源歧视力更强，并且在不同的真实外群体、基因和最优标准之间高度相关。
Phylogenetic rooting experiments demonstrate that two chloroplast genes from commelinoid monocot taxa that represent the closest living relatives of the pickerelweed family, Pontederiaceae, retain measurable signals regarding the position of that family's root. The rooting preferences of the chloroplast sequences were compared with those for artificial sequences that correspond to outgroups so divergent that their signal has been lost completely. These random sequences prefer the three longest branches in the unrooted ingroup topology and do not preferentially root on the branches favored by real outgroup sequences. However, the rooting behavior of the artificial sequences is not a simple function of branch length. The random outgroups preferentially root on long terminal ingroup branches, but many ingroup branches comparable in length to those favored by random sequences attract no or few hits. Nonterminal ingroup branches are generally avoided, regardless of their length. Comparisons of the ease of forcing sequences onto suboptimal roots indicate that real outgroups require a substantially greater rooting penalty than random outgroups for around half of the least-parsimonious candidate roots. Although this supports the existence of nonrandomized signal in the real outgroups, it also indicates that there is little power to choose among the optimal and nearly optimal rooting possibilities. A likelihood-based test rejects the hypothesis that all rootings of the subtree using real outgroup sequences are equally good explanations of the data and also eliminates around half of the least optimal candidate roots. Adding genes or outgroups can improve the ability to discriminate among different root locations. Rooting discriminatory power is shown to be stronger, in general, for more closely related outgroups and is highly correlated among different real outgroups, genes, and optimality criteria.