Placental mammals display a staggering breadth of morphological, karyotypic, and genomic diversity, rivaling or surpassing any other living vertebrate clade (1–3). This variation represents the culmination of 100 million years (Ma) of diversification and parallel adaptation to tumultuous changes in Earth’s environments, including catastrophic events such as the Cretaceous-Paleogene (K-Pg) bolide impact. These different measures of diversity have impeded a complete reckoning of how and why modern placental mammal orders suddenly appeared in the Paleocene with scant paleontological signal preceding the KPg impact.

Prior studies have produced conflicting results regarding the timing and sequence of interordinal and intraordinal cladogenesis. As many as five models of placental mammal diversification have been proposed (4, 5), each implying different degrees of causality between the K-Pg extinction event and ordinal diversification. Each model is supported with molecular analyses of different sequence matrices that have been heavily biased toward short, evolutionarily constrained protein-coding exons or ultraconserved noncoding sequences (6–10). Biased genomic sampling has hampered a full resolution of the placental mammal phylogeny and an understanding of the principal drivers of ordinal diversification.

Here, we report a comprehensive analysis of phylogenomic signals from investigations of multiple genomic character types assayed from a hierarchical alignment (HAL) of 241 placental mammal whole-genome assemblies (1, 11). The HAL samples all placental mammal orders and represents 62% of placental families. The process and data structure that generated the HAL provide a statistically vetted whole-genome assessment of synteny and sequence orthology, reducing the potential for phylogenetic reconstruction errors caused by ortholog misidentification observed in some previous studies (12). The resulting availability of per base estimates of genomic constraint (PhyloP scores) also allowed us to assess the impacts of natural selection on phylogenetic signal and enabled the rigorous application of coalescent approaches (13)...

Whole-genome phylogenies

We applied site pattern frequency–based coalescent methods implemented in the SVDquartets program to sample single-nucleotide polymorphisms (SNPs) spaced by a minimum of 1 kb to reduce the impacts of intralocus recombination and linkage. We estimated phylogenetic relationships for all species in the HAL alignment and for 65 taxon matrices that sample all ordinal lineages while minimizing missing data (table S1). We analyzed three versions of the 65-taxon alignment to mitigate the reference-bias of alignments that were extracted from the HAL (table S2): a human-referenced alignment (HRA), a dog-referenced alignment (DRA), and a root-referenced alignment (RRA) that was imputed from the inferred placental ancestor (1). Because of the absence of nonplacental outgroups in our alignment, the root position was assumed to be between Atlantogenata and Boreoeutheria (5) and remains an open question. To investigate the impact of selection, we also identified conserved, accelerated, and nearly neutral evolving SNPs from a distribution of HRA sites ranked by PhyloP conservation scores across the 241-species alignment (14).

The timing of placental mammal evolution.
Superordinal mammalian diversification took place in the Cretaceous during periods of continental fragmentation and sea level rise with little phylogenomic discordance (pie charts: left, autosomes; right, X chromosome), which is consistent with allopatric speciation. By contrast, the Paleogene hosted intraordinal diversification in the aftermath of the K-Pg mass extinction event, when clades exhibited higher phylogenomic discordance consistent with speciation with gene flow and incomplete lineage sorting.

Fig. 1. Placental mammal phylogeny based on coalescent analysis of nearly neutral sites.
(A) Fifty-percent Majority-rule consensus tree from a SVDquartets analysis of 411,110 genome-wide, nearly neutral sites from the human-referenced alignment of 241 species. Bootstrap support is 100% for all nodes. Superordinal clades are labeled and identified in four colors. Nodes corresponding to Boreoeutheria and Atlantogenata are indicated with black circles. (B) The frequency at which eight superordinal clades [numbered 1 to 8 in (A)] were recovered as monophyletic in 2164 window-based maximum likelihood trees from representative autosomes (Chr1, Chr21 and Chr22) and ChrX. Dotted lines indicate relationships that differ from the concatenated maximum likelihood analysis.

Fig. 3. Rare genomic changes. (A) Number of deletions recovered in the HRA, RRA, in both the HRA and RRA, and on the HRA ChrX in support of all potential laurasiatherian hypotheses. Within Euarchontoglires, hundreds of raw deletions were recovered for Euarchonta, a subset of which were further validated (table S7). Glires + Primatomorpha and Glires + Scandentia were unsupported by the deletion analysis. (B) The topology inferred from the Kuritzin-Kischka-Schmitz-Churakov (KKSC) analysis (50) of deletions for Cetartiodactyla, Perissodactyla, and Ferae (Carnivora + Pholidota) from the HRA, RRA, and HRA/RRA overlap datasets. In all cases, the corresponding KKSC bifurcation test was significant, indicating that a polytomy at this node was rejected. This topology was also recovered in an ASTRAL-BP analysis of the overlapping set of deletions (fig. S9). Bootstrap support values are shown for 500 replicates. (C) High-confidence chromosome breakpoints supporting the monophyly of select superordinal clades. No conflicting breakpoints were found for these nodes.

George Gaylord Simpson (47) predicted that “complete genetic analysis would provide the most priceless data for the mapping of this stream,” referring to the resolution of mammalian phylogeny, a classic and recalcitrant problem in evolutionary biology. Our comprehensive analysis of the 241-placental-mammal whole-genome alignment confirms Simpson’s prediction. It establishes a standard for phylogenomics that maximizes the value of genome sequences at deep taxonomic levels and moves beyond constrained, gene-centric approaches (1). On the basis of the preponderance of evidence across multiple variants of divergence time estimation, we propose that the combination of two major Cretaceous events played a fundamental role in the successful radiation of crown placental mammals in the Paleogene. First, increased continental fragmentation promoted lineage isolation (Fig. 4C), followed by the most rapid episode of land emergence during the Mesozoic (38). This second event would have set the stage for the emergence of morphologically diagnosable orders in the ecological vacuum that followed the mass extinction of nonavian dinosaurs 66 Ma ago. We envision a similar resolution of long-standing controversies across the tree of life with improved use of the historical information encoded within living genomes.