Museum of Zoology, University of Michigan, Ann Arbor, MI 48109-1079.
Department of Ecology and Evolutionary Biology, University of Michigan, Ann Arbor, MI 48109-1079.
Evolution. 2018 Oct;72(10):2246-2256. doi: 10.1111/evo.13566. Epub 2018 Sep 10.
The software program BAMM has been widely used to study rates of speciation, extinction, and phenotypic evolution on phylogenetic trees. The program implements a model-based clustering algorithm to identify clades that share common macroevolutionary rate dynamics and to estimate parameters. A recent simulation study by Meyer and Wiens (M&W) argued that (1) a simple inference framework (MS) performs much better than BAMM, and (2) evolutionary rates inferred with BAMM are poorly correlated with true rates. I address two statistical concerns with their assessment that affect the generality of their conclusions. These considerations are not specific to BAMM and apply to other methods for estimating parameters from empirical data where the true grouping structure of the data is unknown. M&W constrain roughly half of the parameters in their MS analyses to their true values, but BAMM is given no such information and must estimate all parameters from the data. This information disparity results in a substantial degrees of freedom advantage for the MS estimators. When both methods are given equivalent information, BAMM outperforms the MS estimators.
BAMM 软件程序被广泛用于研究系统发育树上的物种形成、灭绝和表型进化的速率。该程序实现了一种基于模型的聚类算法，以识别具有共同宏观进化速率动态的分支，并估计参数。Meyer 和 Wiens（M&W）最近的一项模拟研究认为：(1) 一个简单的推断框架（MS）比 BAMM 表现要好得多；(2) BAMM 推断的进化速率与真实速率相关性较差。我解决了他们评估中影响结论普遍性的两个统计问题。这些考虑因素不仅针对 BAMM，而且适用于从数据中估计参数的其他方法，因为数据的真实分组结构是未知的。M&W 将其 MS 分析中大约一半的参数约束在其真实值，但 BAMM 没有得到这样的信息，必须从数据中估计所有参数。这种信息差异导致 MS 估计器具有很大的自由度优势。当两种方法都具有等效信息时，BAMM 优于 MS 估计器。
