After the last two posts on unicellular organisms, I'm going to bravely leap to another end of phylospace and cover a mammal. Necrolestes patagonensis was a small fossorial animal from the Miocene of Patagonia that has always held a certain appeal for me, both because of its somewhat morbid genus name (it translates as "robber of the dead") and because of its enigmatic phylogenetic position (recent review by Asher et al., 2007).
Necrolestes was described by Florentino Ameghino in 1891. Ameghino seems to have left a few conundrums in his wake - he originally described the giant carnivorous bird Phorusrhacos (sometimes spelt Phororhacos - I'll have to explain that sometime) as a toothless mammal, and mistakenly described the "wingless" penguin Palaeoapterodytes (see here for an explanation). Ameghino seems to have favoured an association of Necrolestes with the African golden moles (Chrysochloridae) - a not unreasonable suggestion for the time. Since then, probably the majority of authors have felt that Necrolestes was a marsupial, but it has also been compared with edentates or suggested as a late survivor of an equally enigmatic group of South American fossil mammals called Gondwanatheria (see here). I recall seeing a nice little cartoon in one paper doubting a marsupial affinity for Necrolestes (I think it was Archer, 1984 but I'm not certain) showing a little Necrolestes being drop-kicked by an anthropomorphised borhyaenid out the door of a gathering of marsupial representatives (borhyaenids were a family of dog-like marsupial carnivores).
After a detailed redescription of the available material (which, among other things, introduced me to the glorious-sounding term schmelzmuster, which refers to the spatial arrangement of different enamel types within a tooth), Asher et al. (2007) attempt to shed some light on the position of Necrolestes by trying to match its characters with previously optimised trees for other mammals. This proves to be quite tricky - Necrolestes has a rather oddball combination of primitive and derived characters, and any suggested position requires a certain amount of convergence. Asher divide the possibilities into three main options - a position outside the Theria (the marsupials + placentals clade), a position close to or within marsupials (metatherians), and a position close to or within placentals (eutherians).
In regards to a position outside Theria, Necrolestes has an atlas (the first cervical vertebra after the skull) with the left and right halves not fused to each other, something unlike any adult therian. It also lacks many of the tooth apomorphies associated with Theria, though this may just be due to the simplified teeth of Necrolestes. However, Necrolestes does have a coiled cochlea, an astragalar neck and lacks a septomaxilla, so Asher et al. conclude it is most likely a therian. As Gondwanatheria is often regarded as non-therian, Asher et al. suggest that Necrolestes is probably not a gondwanatherian, but I feel that the non-therian nature of Gondwanatheria has not really been demonstrated.
In regards to whether Necrolestes is a metatherian or eutherian, Asher et al. don't really come to a firm conclusion. Patterson (1958) claimed that Necrolestes possessed epipubic bones, which are a primitive character retained in marsupials but absent from modern placentals (thought they were present in some stem eutherians). Asher et al., however, found no sign of epipubic bones. It also has a non-inflected mandibular angle, which is generally a eutherian character, but is also found in some derived marsupials. Necrolestes does share a number of characters with metatherians, most of them "absence" characters - lack of a stapedial artery sulcus, lack of a labial mandibular foramen, etc. It agrees with metatherians in having three premolars, but seems to have one too few molars (three instead of four), and shares a ball-shaped distal process on the ulna and transverse canal foramina on the basisphenoid with crown marsupials.
Characters shared with eutherians are a posteriorly small zygomatic process on the squamosal and small incisive foramina, as well as the aforementioned non-inflected mandible and lack of epipubic bones.
Overall, Asher et al. feel that a metatherian affinity for Necrolestes is most likely, which is appealing on biogeographical grounds (most South American insectivores and such at the time being marsupials). However, they admit that a eutherian affinity can't be ruled out, and I would certainly like to see this possibility further investigated. In particular, if Gondwanatheria are related to edentates (another South American group) as some authors have apparently suggested, the idea that Necrolestes is a late survivor of them may yet be reborn.
REFERENCES
Ameghino, F. 1891. Nuevos restos de mamiferos fosiles descubiertos por Carlos Ameghino en el Eoceno inferior de la Patagonia austral. Especies nuevas adiciones y correcciones. Revista Argentina de Historia Natural 1: 289–328.
Archer, M. 1984. Origins and early radiations of marsupials. In Vertebrate Zoogeography and Evolution in Australasia (M. Archer & G. Clayton, eds.) pp. 585–626. Carlisle: Hesperian Press.
Asher, R. J., I. Horovitz, T. Martin & M. R. Sanchez-Villagra. 2007. Neither a rodent nor a platypus: a reexamination of Necrolestes patagonensis Ameghino. American Museum Novitates 3546: 1-40.
Patterson, B. 1958. Affinities of the Patagonian fossil mammal, Necrolestes. Breviora Museum of Comparative Zoology 94: 1–14.