The majority of the combined DNA-morphology analyses support the clades Afrotheria, Xenarthra, Euarchontoglires, and Laurasiatheria, as well as the placement of the Tertiary insectivoran-grade mammal Centetodon within Lipotyphla and the two Cretaceous eutherians (Ukhaatherium and Zalambdalestes) outside of Placentalia (Figs. 1, 2, 3). Using MP, the position of the placental root varies. With all data and gaps included and weighted equally (Fig. 1), or with third position transitions removed, it is at the Malagasy lesser hedgehog-tenrec Echinops, within a paraphyletic Afrotheria. A strict consensus in each case leaves the placental base unresolved (Fig. 1A) due to the variable position of Zalambdalestes. With third positions of protein-coding genes removed, it is at Xenarthra followed by Afrotheria with Cretaceous taxa outside of crown Placentalia (Fig. 2). Results from the Bayesian analysis using either living taxa and sequence data alone, or including three fossils (Zalambdalestes, Ukhaatherium, and Centetodon) plus morphology (Fig. 3), places the placental root at Afrotheria followed by Xenarthra. When included, Cretaceous taxa are again reconstructed outside of crown Placentalia.
Interestingly, MP applied only to extant taxa with all DNA characters, but without morphology, yields a placental tree rooted on murid rodents (Fig. 4B). Inclusion of morphology changes this signal to favour a root within Afrotheria, at the Malagasy tenrec Echinops (Fig. 4A). Removal of third positions favours a placental root at Xenarthra (Fig. 2) with or without morphological data. As evident by comparing Figs. 1B and 4A, exclusion of the 12 fossil taxa in the equally weighted MP analysis does not shift the root away from the afrotherian Echinops.
Table 1 summarizes the results of Templeton and Winning Sites tests using PAUP 4.0b10  evaluating competing hypotheses on the location of the placental root. Using MP applied to the combined dataset, and regardless of the treatment of third positions, the hypotheses of Glires or Erinaceus basal are rejected. With third coding positions excluded, these tests yield p-values close to but not consistently below 0.05 for both Atlantogenata and Muridae at the placental root. With all DNA-indel-morphology characters included, Atlantogenata is rejected and Muridae is not. Monophyletic, basal Afrotheria or Xenarthra is not rejected in any case (Table 1).
The position of the placental root influences the optimization of morphological characters throughout the placental tree. However, some morphological characters optimize at the root of Placentalia under a number of hypotheses. With either Afrotheria, Xenarthra, Atlantogenata, Glires, Muridae, or Erinaceus at the placental base, three morphological character states optimize as placental synapomorphies: #39-1 (single hypoglossal foramen), #48-0 (foramen rotundum confluent with sphenorbital fissure), and #159-1 (epipubic bones absent). With either Afrotheria or Atlantogenata basal, two additional morphological synapomorphies for Placentalia optimize unambiguously: #11-0 (presence of a sulcus for the internal carotid artery on the promontorium of the petrosal) and #105-1 (prominent lingual cusp on upper P3). A paraphyletic Rodentia at or near the placental base (following  or Fig. 4B) greatly increases the number of morphological characters that show unambiguous change on the branch leading to crown Placentalia, and requires significantly more homoplasy among morphological characters than the other hypotheses of rooting.
The placement of several fossils, namely Leptictis,Paleoparadoxia, Plesiorycteropus and Zalambdalestes, remains ambiguous in this study. However, when resolved, the latter taxon falls outside of crown Placentalia (Figs. 2, 3); this result has also been supported by other, independent datasets [20,21]. In the current study, the treatment of DNA third positions influences the topology of several fossils, a result that may appear counterintuitive since all DNA data are missing for these fossils. Nevertheless, this is a straightforward result based on the altered optimizations of morphological characters on those branches of the tree that are rearranged by addition of the sequence partition, which in turn can affect the influence of those characters on the placement of fossils .