Huelsenbeck J P
Department of Zoology, University of Texas, Austin 78712, USA.
Mol Biol Evol. 1995 Sep;12(5):843-9. doi: 10.1093/oxfordjournals.molbev.a040261.
The robustness (sensitivity to violation of assumptions) of the maximum-likelihood and neighbor-joining methods was examined using simulation. Maximum likelihood and neighbor joining were implemented with Jukes-Cantor, Kimura, and gamma models of DNA substitution. Simulations were performed in which the assumptions of the methods were violated to varying degrees on three model four-taxon trees. The performance of the methods was evaluated with respect to ability to correctly estimate the unrooted four-taxon tree. Maximum likelihood outperformed neighbor joining in 29 of the 36 cases in which the assumptions of both methods were satisfied. In 133 of 180 of the simulations in which the assumptions of the maximum-likelihood and neighbor-joining methods were violated, maximum likelihood outperformed neighbor joining. These results are consistent with a general superiority of maximum likelihood over neighbor joining under comparable conditions. They extend and clarify an earlier study that found an advantage for neighbor joining over maximum likelihood for gamma-distributed mutation rates.
通过模拟检验了最大似然法和邻接法的稳健性（对假设违背的敏感性）。使用Jukes-Cantor、Kimura和伽马DNA替代模型来实施最大似然法和邻接法。在三个模型四分类单元树上进行了模拟，其中方法的假设在不同程度上被违背。根据正确估计无根四分类单元树的能力来评估这些方法的性能。在两种方法的假设均得到满足的36个案例中，最大似然法在29个案例中优于邻接法。在180次模拟中的133次模拟中，最大似然法和邻接法的假设被违背，最大似然法优于邻接法。这些结果与在可比条件下最大似然法总体优于邻接法相一致。它们扩展并澄清了一项早期研究，该研究发现对于伽马分布的突变率，邻接法优于最大似然法。
