Combining morphological and molecular characters through Bayesian total-evidence dating allows inferring the phylogenetic and timescale framework of both extant and fossil taxa, while accounting for the stochasticity and incompleteness of the fossil record. Such an integrative approach is particularly needed when dealing with clades such as sloths (Mammalia: Folivora), for which developmental and biomechanical studies have shown high levels of morphological convergence whereas molecular data can only account for a limited percentage of their total species richness. Here, we propose an alternative hypothesis of sloth evolution that emphasizes the pervasiveness of morphological convergence and the importance of considering the fossil record and an adequate taxon sampling in both phylogenetic and biogeographic inferences. Regardless of different clock models and morphological datasets, the extant sloth Bradypus is consistently recovered as a megatherioid, and Choloepus as a mylodontoid, in agreement with molecular-only analyses. The recently extinct Caribbean sloths (Megalocnoidea) are found to be a monophyletic sister-clade of Megatherioidea, in contrast to previous phylogenetic hypotheses. Our results contradict previous morphological analyses and further support the polyphyly of “Megalonychidae”, whose members were found in five different clades. Regardless of taxon sampling and clock models, the Caribbean colonization of sloths is compatible with the exhumation of islands along Aves Ridge and its geological time frame. Overall, our total-evidence analysis illustrates the difficulty of positioning highly incomplete fossils, although a robust phylogenetic framework was recovered by an a posteriori removal of taxa with high percentages of missing characters. Elimination of these taxa improved topological resolution by reducing polytomies and increasing node support. However, it introduced a systematic and geographic bias because most of these incomplete specimens are from northern South America. This is evident in biogeographic reconstructions, which suggest Patagonia as the area of origin of many clades when taxa are underrepresented, but Amazonia and/or Central and Southern Andes when all taxa are included. More generally, our analyses demonstrate the instability of topology and divergence time estimates when using different morphological datasets and clock models, and thus caution against making macroevolutionary inferences when node support is weak or when uncertainties in the fossil record are not considered.

中文翻译：

---

贝叶斯全证据约会重温树懒系统发育和生物地理学：形态时钟分析的警示故事


通过贝叶斯全证据测年结合形态和分子特征，可以推断现存和化石类群的系统发育和时间尺度框架，同时解释化石记录的随机性和不完整性。在处理树懒（哺乳动物：Folivora）等进化枝时，尤其需要这种综合方法，因为发育和生物力学研究表明树懒具有高度的形态趋同性，而分子数据只能解释其总物种丰富度的有限百分比。在这里，我们提出了树懒进化的另一种假说，该假说强调形态趋同的普遍性以及在系统发育和生物地理学推论中考虑化石记录和适当的分类单元采样的重要性。无论时钟模型和形态数据集如何不同，现存的树懒Bradypus始终被恢复为巨兽类，而Choloepus则被恢复为mylodontoid，这与仅分子分析一致。最近灭绝的加勒比树懒（Megalocnoidea）被发现是 Megatherioidea 的单系姐妹进化枝，这与之前的系统发育假设相反。我们的结果与之前的形态学分析相矛盾，并进一步支持“Megalonychidae”的多系性，其成员发现于五个不同的分支中。无论分类单元采样和时钟模型如何，树懒在加勒比海的殖民活动都与阿夫斯山脊沿线岛屿的挖掘及其地质时间框架相一致。总体而言，我们的总证据分析说明了定位高度不完整的化石的难度，尽管通过事后删除具有高比例缺失特征的类群恢复了稳健的系统发育框架。 消除这些类群通过减少多分体和增加节点支持来提高拓扑分辨率。然而，它引入了系统和地理偏差，因为这些不完整的标本大多数来自南美洲北部。这在生物地理重建中很明显，当类群代表性不足时，表明巴塔哥尼亚是许多进化枝的起源地区，但当包括所有类群时，亚马逊和/或安第斯山脉中部和南部。更一般地说，我们的分析证明了使用不同形态数据集和时钟模型时拓扑和发散时间估计的不稳定性，因此在节点支持较弱或不考虑化石记录中的不确定性时不要进行宏观进化推论。