CAS Key Laboratory of Genomic and Precision Medicine, Beijing Institute of Genomics, Chinese Academy of Sciences, Beijing, China.
Department of Genetics, Evolution and Environment, University College London, London, United Kingdom.
Mol Biol Evol. 2018 Jan 1;35(1):159-179. doi: 10.1093/molbev/msx277.
The phylogenetic relationships among extant gibbon species remain unresolved despite numerous efforts using morphological, behavorial, and genetic data and the sequencing of whole genomes. A major challenge in reconstructing the gibbon phylogeny is the radiative speciation process, which resulted in extremely short internal branches in the species phylogeny and extensive incomplete lineage sorting with extensive gene-tree heterogeneity across the genome. Here, we analyze two genomic-scale data sets, with ∼10,000 putative noncoding and exonic loci, respectively, to estimate the species tree for the major groups of gibbons. We used the Bayesian full-likelihood method bpp under the multispecies coalescent model, which naturally accommodates incomplete lineage sorting and uncertainties in the gene trees. For comparison, we included three heuristic coalescent-based methods (mp-est, SVDQuartets, and astral) as well as concatenation. From both data sets, we infer the phylogeny for the four extant gibbon genera to be (Hylobates, (Nomascus, (Hoolock, Symphalangus))). We used simulation guided by the real data to evaluate the accuracy of the methods used. Astral, while not as efficient as bpp, performed well in estimation of the species tree even in presence of excessive incomplete lineage sorting. Concatenation, mp-est and SVDQuartets were unreliable when the species tree contains very short internal branches. Likelihood ratio test of gene flow suggests a small amount of migration from Hylobates moloch to H. pileatus, while cross-genera migration is absent or rare. Our results highlight the utility of coalescent-based methods in addressing challenging species tree problems characterized by short internal branches and rampant gene tree-species tree discordance.
尽管使用形态学、行为学和遗传学数据以及全基因组测序进行了多次尝试，但现生长臂猿物种之间的系统发育关系仍未得到解决。重建长臂猿系统发育的主要挑战是辐射物种形成过程，这导致物种系统发育中的内部分支非常短，以及基因组中广泛的不完全谱系分选和广泛的基因树-物种树分歧。在这里，我们分析了两个基因组规模的数据集，分别有约 10000 个假定的非编码和外显子基因座，以估计主要长臂猿类群的种系发生树。我们使用贝叶斯全似然法 bpp 在多物种合并模型下进行分析，该模型自然适应不完全谱系分选和基因树的不确定性。为了进行比较，我们还包括了三种启发式合并方法（mp-est、SVDQuartets 和 astral）以及串联法。从这两个数据集，我们推断出现存的四种长臂猿属的系统发育关系为（Hylobates、（Nomascus、（Hoolock、Symphalangus）））。我们使用真实数据指导的模拟来评估所使用方法的准确性。虽然 astral 不如 bpp 高效，但即使在存在过度不完全谱系分选的情况下，它在估计物种树方面表现良好。当物种树包含非常短的内部分支时，串联法、mp-est 和 SVDQuartets 是不可靠的。基因流的似然比检验表明，有少量从 Hylobates moloch 到 H. pileatus 的迁移，而跨属的迁移是不存在的或很少的。我们的结果强调了基于合并的方法在解决具有短内部分支和猖獗的基因树-物种树分歧的具有挑战性的物种树问题方面的实用性。
