Geophysical Sciences, University of Chicago, Chicago, Illinois, USA.
PLoS One. 2013 Apr 25;8(4):e62312. doi: 10.1371/journal.pone.0062312. Print 2013.
Morphology-based phylogenetic analyses are the only option for reconstructing relationships among extinct lineages, but often find support for conflicting hypotheses of relationships. The resulting lack of phylogenetic resolution is generally explained in terms of data quality and methodological issues, such as character selection. A previous suggestion is that sampling ancestral morphotaxa or sampling multiple taxa descended from a long-lived, unchanging lineage can also yield clades which have no opportunity to share synapomorphies. This lack of character information leads to a lack of 'intrinsic' resolution, an issue that cannot be solved with additional morphological data. It is unclear how often we should expect clades to be intrinsically resolvable in realistic circumstances, as intrinsic resolution must increase as taxonomic sampling decreases. Using branching simulations, I quantify intrinsic resolution across several models of morphological differentiation and taxonomic sampling. Intrinsically unresolvable clades are found to be relatively frequent in simulations of both extinct and living taxa under realistic sampling scenarios, implying that intrinsic resolution is an issue for morphology-based analyses of phylogeny. Simulations which vary the rates of sampling and differentiation were tested for their agreement to observed distributions of durations from well-sampled fossil records and also having high intrinsic resolution. This combination only occurs in those datasets when differentiation and sampling rates are both unrealistically high relative to branching and extinction rates. Thus, the poor phylogenetic resolution occasionally observed in morphological phylogenetics may result from a lack of intrinsic resolvability within groups.
基于形态的系统发育分析是重建灭绝谱系之间关系的唯一选择，但往往会发现对关系的假设存在冲突。由此产生的系统发育分辨率不足通常可以用数据质量和方法学问题来解释，例如特征选择。之前有一个建议是，对祖先形态类群进行采样，或者对从长寿、不变的谱系中衍生出来的多个类群进行采样，也可能产生没有机会共享同源特征的进化枝。这种缺乏特征信息导致了缺乏“内在”分辨率的问题，这个问题无法通过额外的形态数据来解决。目前尚不清楚在现实情况下，我们应该期望内在分辨率出现的频率有多高，因为内在分辨率必须随着分类采样的减少而增加。通过分支模拟，我在几种形态分化和分类采样模型中量化了内在分辨率。在现实采样场景下对灭绝和现存类群进行模拟时，发现内在不可分辨的进化枝相对频繁，这意味着内在分辨率是基于形态的系统发育分析中的一个问题。测试了不同采样和分化速率的模拟，以确定它们与从充分采样的化石记录中观察到的持续时间分布的一致性，以及具有较高内在分辨率。只有当分化和采样速率相对于分支和灭绝速率都高得不切实际时，这种组合才会出现在这些数据集当中。因此，形态系统发生学中偶尔观察到的较差的系统发育分辨率可能是由于群体内部缺乏内在可分辨性所致。
