Department of Biology, Winthrop University, Rock Hill, SC 29733, USA.
Mol Phylogenet Evol. 2013 Oct;69(1):1-3. doi: 10.1016/j.ympev.2013.05.010. Epub 2013 May 21.
To provide a baseline biological comparison to simulation study predictions about the frequency of rogue taxa effects, we evaluated the frequency of a rogue taxa effect using viral data sets which differed in diversity. Using a quartet-tree framework, we measured the frequency of a rogue taxa effect in three data sets of increasing genetic variability (within viral serotype, between viral serotype, and between viral family) to test whether the rogue taxa was correlated with the mean sequence diversity of the respective data sets. We found a slight increase in the percentage of rogues as nucleotide diversity increased. Even though the number of rogues increased with diversity, the distribution of the types of rogues (friendly, crazy, or evil) did not depend on the diversity and in the case of the order-level data set the net rogue effect was slightly positive. This study, assessing frequency of the rogue taxa effect using biological data, indicated that simulation studies may over-predict the prevalence of the rogue taxa effect. Further investigations are necessary to understand which types of data sets are susceptible to a negative rogue effect and thus merit the removal of taxa from large phylogenetic reconstructions.
为了对模拟研究中关于流氓分类单元效应频率的预测提供一个基线生物学比较，我们使用在多样性方面存在差异的病毒数据集来评估流氓分类单元效应的频率。使用四分树框架，我们在三个遗传变异性逐渐增加（同一病毒血清型内、不同病毒血清型之间和不同病毒科之间）的病毒数据集上测量了流氓分类单元效应的频率，以检验流氓分类单元是否与各自数据集的平均序列多样性相关。我们发现随着核苷酸多样性的增加，流氓分类单元的百分比略有增加。尽管随着多样性的增加，流氓分类单元的数量增加了，但流氓分类单元的类型（友好型、疯狂型或邪恶型）的分布并不依赖于多样性，并且在阶元数据集的情况下，净流氓效应略为正。本研究使用生物数据评估了流氓分类单元效应的频率，表明模拟研究可能高估了流氓分类单元效应的普遍性。有必要进一步研究哪些类型的数据集容易受到负的流氓效应的影响，从而需要从大型系统发育重建中删除分类单元。
