Data from: Tree imbalance causes a bias in phylogenetic estimation of evolutionary timescales using heterochronous sequences

Phylogenetic estimation of evolutionary timescales has become routine in biology, forming the basis of a wide range of evolutionary and ecological studies. However, there are various sources of bias that can affect these estimates. We investigated whether tree imbalance, a property that is commonly observed in phylogenetic trees, can lead to reduced accuracy or precision of phylogenetic timescale estimates. We analysed simulated data sets with calibrations at internal nodes and at the tips, taking into consideration different calibration schemes and levels of tree imbalance. We also investigated the effect of tree imbalance on two empirical data sets: mitogenomes from primates and serial samples of the African swine fever virus. In analyses calibrated using dated, heterochronous tips, we found that tree imbalance had a detrimental impact on precision and produced a bias in which the overall timescale was underestimated. A pronounced effect was observed in analyses with shallow calibrations. The greatest decreases in accuracy usually occurred in the age estimates for medium and deep nodes of the tree. In contrast, analyses calibrated at internal nodes did not display a reduction in estimation accuracy or precision due to tree imbalance. Our results suggest that molecular-clock analyses can be improved by increasing taxon sampling, with the specific aims of including deeper calibrations, breaking up long branches and reducing tree imbalance.进化时间尺度的系统发育（phylogenetic）估计现已成为生物学研究中的常规手段，为诸多进化与生态学相关研究奠定了核心基础。然而，各类偏倚来源均可能影响这类估计结果。本研究聚焦于系统发育树中普遍存在的树不平衡（tree imbalance）现象，探究其是否会导致系统发育时间尺度估计的准确性与精确性下降。我们针对内部节点与末端节点均带有定标的模拟数据集展开分析，同时纳入不同的定标方案与树不平衡水平变量。此外，本研究还分析了树不平衡对两类实证数据集的影响：灵长类线粒体基因组（mitogenomes）与非洲猪瘟病毒的连续分离样本。在基于带定年信息的异时末端样本开展的定标分析中，我们发现树不平衡会对估计精确性造成负面影响，并引发整体时间尺度被低估的偏倚。在采用浅层定标的分析中，这一效应尤为显著。估计准确性降幅最显著的通常为系统发育树中层与深层节点的分化时间估计值。与之相反，以内部节点作为定标位点的分析并未因树不平衡出现估计准确性与精确性的下降。本研究结果表明，可通过优化类群采样策略改进分子钟（molecular-clock）分析，具体目标包括纳入深层定标位点、拆分长分支以及降低树不平衡程度。