Chinese lantern for early primates

A fossil skull from China, dating to 55 million years ago, provides much-needed substantial evidence of early primates in Asia. Interpretation of the creature's eye size and activity pattern will spark debate.

The first fossil evidence for 'primates of modern aspect' — the euprimates — appears abruptly in the northern continents at the beginning of the Eocene, about 55 million years ago. Eocene primates are well documented from sites in North America and Europe, but specimens from Asia have been scarce and fragmentary. That situation changes with the report by Ni et al. of a well-preserved partial primate skull, with lower jaws, from China (page 65 of this issue¹). The authors allocate the fossil to a new species of the genus Teilhardina, which was originally established for fragmentary material from Belgium².

The new skull has several euprimate characteristics, notably a bony ring around each eye-socket (orbit), forward rotation of the orbits and a relatively large braincase. But the skull is very small, with an overall length of about 25 mm, and the animal's body mass is estimated as 28 g — less than in any modern primate species. As expected for one of the earliest known euprimates, the dentition is relatively primitive. The upper and lower jaws both contain four premolars on each side (the maximum known for any primate) and the front premolars are not as reduced as in other Teilhardina species. Certain dental features, and its small body size, suggest that the animal was an insectivore.

The bigger picture is outlined in Fig. 1, which reflects the widely accepted view that there is a basic dichotomy in the primate evolutionary tree — one lineage leading to modern lemurs and lorises (strepsirrhines), the other to tarsiers, monkeys, apes and humans (haplorhines)^3,4,5. All or most Eocene primates are allocated to two major groups, lemuroids and tarsioids. These are comm-only linked to modern strepsirrhines and haplorhines, respectively (although both could possibly have arisen from an independent northern radiation of primates^6,7). Teilhardina is included among tarsioids as a potential relative of haplorhines.

Figure 1: Primate evolution in outline.

This tree incorporates the results of the analysis by Ni et al.¹ (see Fig. 3 on page 67; lemuroids are formally known as Adapiformes, and tarsioids as Omomyiformes). Along with Teilhardina belgica, the new species T. asiatica branches away first on the haplorhine side of the tree (T1). By contrast, T. americana (T2) is nested within the other Eocene tarsioids, calling into doubt its place in the genus Teilhardina. (Primate icons drawn by Lucrezia Beerli-Bieler.)

As well as the new species (T. asiatica), and the originally described T. belgica², the genus Teilhardina currently includes five other species from North America. On this basis, then, Teilhardina seemingly had a remarkably wide distribution. Using characters derived from the new skull, Ni et al.¹ repeated an earlier analysis of primate relationships⁵. The results confirm the basic primate dichotomy, and are striking for Teilhardina itself. Teilhardina asiatica and T. belgica together constitute the earliest offshoot from the haplorhine side of the tree, branching off before the split between other tarsioids and haplorhines (Fig. 1). By contrast, T. americana (the only American species considered) lies within the cluster containing all other tarsioids, reflecting a pronounced separation from T. asiatica and T. belgica. Inclusion of T. americana and other closely related North American species in the genus Teilhardina is therefore questionable, as is the apparent extensive distribution of the genus.

Another unusual feature of T. asiatica is the small size of the orbits relative to skull length. Analysis of the scaling of orbit size using a published data set⁸ indicates that this species falls on the best-fit line for modern diurnal primates, which generally have relatively smaller orbits than nocturnal species. From this, Ni et al.¹ infer that T. asiatica was diurnal. They go on to interpret the evolution of nocturnal versus diurnal habits among primates, and conclude that the last common ancestor of euprimates was a diurnal, visually oriented predator.

This is an important issue, one on which I have to disagree with the authors on both statistical and biological grounds. Prediction from regression lines beyond the range of the original data is suspect, and T. asiatica is smaller than any modern primate. In fact, Ni and colleagues excluded all large-bodied primates from the original data set, so altering the slope of the best-fit line for diurnal primates and strengthening the inference that T. asiatica was diurnal. Furthermore, no account was taken of the potential problem of 'phylogenetic inertia'⁹: closely related species may not be statistically independent, so estimates of probability may be open to doubt.

Biologically, one cannot assume that early primates (particularly if unusually small in size) showed the same functional patterns as modern primates — which themselves are very variable. Given that ancestral primates descended from ancestral mammals with smaller eyes, early primates presumably showed only moderate enlargement of their eyes, regardless of their habits. In early primates, the brain was less than half the size of the brains of their modern relatives³. So processing of visual inputs was probably more rudimentary, perhaps explaining why some Eocene lemuroids had far smaller orbits than modern diurnal lemurs.

Another source of information comes from a study by Kay and Kirk⁸. To increase visual sensitivity in dim light, nocturnal species show marked retinal summation of inputs from the photoreceptors, resulting in a narrower optic nerve. This is reflected by a narrower opening (foramen) for the optic nerve in the back of the orbit. For living primates, Kay and Kirk found a good match between activity pattern and relative size of the foramen. However, all Eocene primates have a small foramen, regardless of whether they have relatively small orbits (suggesting diurnal vision) or large orbits (indicating nocturnal habits). Thus, visual adaptations in early primates were clearly different from those in modern primates. Finally, a feature on the snout of T. asiatica, the infraorbital foramen, is markedly larger than in modern primates. The size of this foramen — large in primitive nocturnal mammals, typically reduced in diurnal mammals — indicates the degree of development of tactile whiskers for non-visual orientation.

I believe, then, that we remain in the dark with respect to the activity pattern of T. asiatica. Regardless of that, however, Ni and colleagues' discovery is notable not only for its implications for primate systematics but also from a biogeographical perspective. Until recently, it was widely held that direct migration of mammals between Asia and Europe around 55 million years ago was ruled out by a transcontinental marine barrier. The landmass of Eurasia was largely or completely split down the middle by a combination of the Western Siberian Obik Sea to the north and the Turgai Straits to the south. So it was suggested that the only possibility for interchange of mammals between Asia and Europe was indirect migration across the Bering Straits, through North America and across the Greenland landbridge, or vice versa. However, the presence of closely related Teilhardina species in China and Belgium (but not in America) in the early Eocene adds to evidence that migration between Asia and Europe did not necessarily involve such a roundabout route¹⁰.