A mitogenomic timetree for Darwin’s enigmatic South American mammal Macrauchenia patachonica

The unusual mix of morphological traits displayed by extinct South American native ungulates (SANUs) confounded both Charles Darwin, who first discovered them, and Richard Owen, who tried to resolve their relationships. Here we report an almost complete mitochondrial genome for the litoptern Macrauchenia. Our dated phylogenetic tree places Macrauchenia as sister to Perissodactyla, but close to the radiation of major lineages within Laurasiatheria. This position is consistent with a divergence estimate of ∼66 Ma (95% credibility interval, 56.64–77.83 Ma) obtained for the split between Macrauchenia and other Panperissodactyla. Combined with their morphological distinctiveness, this evidence supports the positioning of Litopterna (possibly in company with other SANU groups) as a separate order within Laurasiatheria. We also show that, when using strict criteria, extinct taxa marked by deep divergence times and a lack of close living relatives may still be amenable to palaeogenomic analysis through iterative mapping against more distant relatives.


Supplementary Note 1: Systematic Context of Macrauchenia patachonica
Overview of litoptern systematics. Litopterns were cursorial herbivores with mesaxonic limbs and several unusual cranial and dental features. In distinction to most other SANUs, litopterns primitively retained rooted teeth and developed varied cheektooth morphologies, including bunodonty (e.g., Megadolodinae), bunolophodonty (e.g., Proterotheriinae) and lophoselenodonty (e.g., Macraucheniidae) 1,2,3 . As the result of continuing discovery over many decades, we now know that litopterns were present in South America for most of the Cenozoic. One group (Sparnotheriodontidae) even reached West Antarctica when this was connected to southernmost South America prior to the appearance of the Drake Passage 4,5.6 .
Late Eocene (Mustersan SALMA) Polymorphis, the oldest known definite macraucheniid 14 , serves to date the origin of the family paleontologically. Differing from the horse-like proterotheriids, macraucheniids had a robust body structure more like that of modern camels, featuring three-toed autopodia, a long neck, and reduced nasals. In Late Miocene-Pleistocene macraucheniines, some of these features evolved in bizarre directions, as in the case of the retraction of the nasal aperture 21 . From a position at the end of the rostrum, as in typical mammals, this opening progressively moved dorsally. Pleistocene Macrauchenia, in which the nasal aperture is perched between the orbits, near the summit of the skull, represents a morphological extreme. It has long been speculated whether so drastic a modification implies that the snout was elaborated into a proboscis 8,[22][23][24][25] .
Macraucheniidae includes either two or three subfamilies, depending on whether Theosodontinae is separately recognized. Macraucheniinae and Cramaucheniinae contain most of the approximately 18 genus-level taxa currently recognized for the family (ignoring possible synonyms). Progressive increases in body size, cheektooth crown height, and snout length strongly marked the evolution of Macraucheniinae 1,21, 23,24,26-28 . From the standpoint of taxonomic richness, macraucheniine peak diversity was achieved in the Late Miocene. They became extinct early in the Holocene, Macrauchenia itself being the last surviving taxon (last appearance date: 8390 ± 140 14C yr BP 8,29 ). Macrauchenia (and its close relatives Macraucheniopsis and Xenorhinotherium) had a broad distribution across the continent, although as in the case of other South American native ungulates known fossils overwhelmingly come from the southern cone (Supplementary fig. 7, 8). Unlike toxodontid notoungulates, which managed to penetrate Central and southern North America after the completion of the Isthmus of Panama, Macrauchenia and other litopterns failed to participate in the Great American Biotic Interchange.
History of classification and phylogenetic relationships. The relationships of litopterns and other SANUs have been controversial since the first specimens were described 180 years ago 30,31 . Lydekker 32 included litopterns within "Ungulata", a wastebasket comprising fossil and extant ungulates generally, and in this he was followed by Osborn 33 , Scott 1 , and Schlosser 34 . Although Scott 1 presciently remarked on numerous resemblances between litopterns and perissodactyls, in the end he concluded that they were more likely the result of parallelism than actual close relationship. Ameghino 35 , by contrast, claimed that it was "absolutely certain" that macraucheniids and other litopterns had a common origin with perissodactyls, distinct from the ancestry of other SANUs.
Simpson 36 viewed litopterns as directly derived from Condylarthra, a heterogeneous group of early Cenozoic mammals structurally and presumably phyletically ancestral to later, more advanced ungulates, and grouped them with Notoungulata, Astrapotheria, and Tubulidentata in a paraphyletic supraordinal entity, Protungulata. Subsequently, Simpson 2,37 suggested that all SANUs (including Pyrotheria and Xenungulata) evolved in South America from a North American condylarth ancestral stock, which arrived in South America during the Late Cretaceous or Early Paleocene. This scenario has been accepted by many later authors (e.g., refs. 14,18,19,38,39 ).
McKenna 40,41 refined Simpson's proposal, contending that all SANUs (but not Tubulidentata) could be derived from a single North American condylarthran common ancestor. He proposed the name Meridiungulata for this putative monophyletic ensemble, but did not present an explicit cladistic analysis in support of it (see also ref. 42 ). Soria 3,43 also accepted a condylarthran origin for SANUs, but differed from McKenna in proposing a diphyletic origin for the latter. He argued that litopterns were explicitly related to certain North and South American condylarthrans and could thus be included within Protungulata (with content differing from Simpson's 36 concept). The remaining SANU orders were excluded from this grouping. A similar evolutionary scenario was proposed by Muizon and Cifelli 11 , who coined the term Panameriungulata for a group including litopterns, South American didolodontids, and Kollpaniinae, an endemic subfamily of North American mioclaenids. The monophyly of Meridiungulata sensu McKenna 40,41 was also rejected by Tong and Lucas 44 , Lucas 45 , and Kondrashov and Lucas 46 . Horovitz 47 likewise found that Meridiungulata was paraphyletic, based on a cladistic study of postcranial elements. She concluded that litopterns and notoungulates were sister taxa, allied with meniscotheriid and phenacodontid condylarthrans, but separate from astrapotheres. In a similar vein, the preferred tree of placental cladistic relationships published by O´Leary et al. 48 positioned Litopterna (represented by Protolipterna) within stem Pan-Euungulata while Notoungulata (represented by Thomashuxleya) grouped with Afrotheria, thereby denying once again the monophyly of Meridiungulata.
Muizon et al. 49 recently took up the question again, ruling in favor of meridiungulate monophyly but favoring close relationship with Artiodactyla. However, inasmuch as their data matrix did not include any perissodactyls, their analysis does not contradict the results of recent proteomic studies 50,51 , which conclude that both litopterns (represented by Macrauchenia) and notoungulates (represented by Toxodon) are more closely related to Perissodactyla than to any other extant placental group (see also ref. 52 ). In the absence of consensus in phenomic analyses, and lack of molecular information for other SANU orders, it remains unsettled whether all South American native ungulates are related monophyletically.