Three to nine oriented block siltstone samples were collected from 16 sites of 15 different stratigraphic levels in the Nakali Formation for paleomagnetic analysis. At each site, samples were collected within a thickness of 0.3-3 m. Paleomagnetic analyses of two to four samples were performed by using a 2G superconducting magnetometer and a Natsuhara thermal demagnetizer. The results of demagnetization showed that all of the samples except those from one site exhibited a clear single-component behavior. Magnetic polarity results for the Nakali Formation are shown in Fig. 4. Progressive thermal demagnetization established that natural remanent magnetizations (NRM) showed a linear decay toward the origin between ~200 and 680°C (SI Fig. 8). Characteristic remanent magnetization (ChRM) component was isolated by principal component analysis (3). ChRMs from blocks at the same site show the same polarity, having consistent directions within a range of geomagnetic secular variation. All ChRM directions from Nakali Formation are plotted in SI Fig. 9. Samples from the uppermost horizon of the Lower Member and the lowest horizon of the Upper Member had a well defined characteristic remanent magnetization with full reverse polarity, with NRM intensity as strong as those of the other sites, suggesting that the reversal does not represent an excursion. This reverse polarity interval, combined with the ⁴⁰Ar-³⁹Ar ages, can be reasonably correlated with Chron C5n.1r, dated at 9.92-9.88 Ma (4). Consequently, the hominoid bed is located within the C5n.1n chronozone [9.88-9.74 Ma (4)]. Given the rapid sediment accumulation, the age of the hominoid bed is most probably 9.9-9.8 Ma.

The mammalian assemblages of the Nakali fauna (SI Table 4) closely resemble those from the Samburu Hills (Namurungule Formation) (5) and Ngeringerwa (6) in Kenya. These faunas were designated as Faunal Set VI of East African biostratigraphy [~10 Ma (7)]. Isotopic studies on the tooth enamel of proboscideans and equids show a greater tendency for transition toward a C₄ diet (8, 9). The richer primate fauna from Nakali, particularly small-bodied catarrhines, also supports this view because similar-sized (<3 kg) living primates occur exclusively in forests. Isotopic studies on tooth enamel of proboscideans from both sites indicate C₃ diets, but the Samburu Hills herbivores show a greater tendency for transition toward the C₄ diet than do those at Nakali (refs. 8 and 9 and T. Cerling, personal communication). The combination of active tectonism and higher rainfall together with a higher rate of supply in volcaniclastics (10) would have led to ongoing deterioration of the environment in this region after 10 Ma. This may account for the relatively open environment indicated by the younger Namurungule fauna compared with the Nakali fauna.

1. Golden M (1978) Ph.D. thesis (Univ of London, London), p 254.

2. Hyodo H, Kim S-W, Itaya T, Matsuda T (1999) J Miner Petr Econ Geol 94:329-337.

3. Kirschvink JL (1980) Geophys J R Astron Soc 62:699-718.

4. Cande SC, Kent DV (1995) J Geophys Res 100:6093-6095.

5. Nakaya H (1994) Afr Stud Monog Suppl 20:1-112.

6. Benefit BR, Pickford M (1986) Am J Phys Anthropol 69:441-464.

7. Pickford M (1981) J Hum Evol 10:73-97.

8. Cerling TE, Harris JM, MacFadden BJ, Leakey MG, Quade J, Eisenmann V, Ehleringer JR (1997) Nature 389:153-158.

9. Cerling TE, Harris JM, Leakey MG (1999) Oecologia 120:364-374.

10. Saneyoshi M, Nakayama K, Sakai T, Sawada Y, Ishida H (2006) Sed Geol 186:111-131.

Fig. 4. Location map showing Nakali and Samburu Hills in Kenya (A) and geological columnar sections with ⁴⁰Ar-³⁹Ar ages and magnetic polarity of the Nakali area (B). Magnetic polarity and chron scale are from ref. 1.

1. Cande SC, Kent DV (1995) J Geophys Res 100:6093-6095.

SI Figure 5

Fig. 5. Location map showing Nakali (A) and geological map and cross section of the Nakali area with ⁴⁰Ar-³⁹Ar ages (B).

Fig. 6. Age spectra of anorthoclase grains from the Nakali Formation. Slight excess argon can be recognized in the low-temperature fraction.

Fig. 7. Age spectra of plagioclase phenocrysts and groundmass from the basalt of the Nasorut Formation.

SI Table 2

Fig. 8. Vectorial endpoint diagrams for selected samples from the Nakali Formation. Open circles show projections onto the vertical plane, and filled circles show projections onto the horizontal plane.

Fig. 9. Equal area plot of ChRM directions for all samples from the Nakali Formation. Open circles indicate projections in the upper hemisphere, and filled circles indicate projections in the lower hemisphere.

Fig. 10. Relative enamel thickness of extant and fossil hominoids. Number of specimens is in parentheses. Comparative data are from refs. 1-8.

1. Martin L (1985) Nature 314:260-263.

2. Grine FE, Martin LB (1988) in Evolutionary History of the "Robust" Australopithecines, ed Grine FE (Aldine de Gruyter, New York), pp 3-42.

3. Andrews P, Martin L (1991) Philos Trans R Soc London B 334:199-209.

4. Beynon AD, Clayton CB, Rozzi R, Reid DJ (1998) J Hum Evol 35:351-370.

5. Schwartz GT, Liu W, Zheng L (2003) J Hum Evol 44:189-202.

6. Smith TM, Martin LB, Leakey MG (2003) J Hum Evol 44:283-306.

7. Chaimanee Y, Jolly D, Benammi M, Tafforeau P, Duzer D, Moussa I, Jaeger J -J (2003) Nature 422:61-65.

8. Kunimatsu Y, Ratanasthien B, Nakaya H, Saegusa H, Suganuma Y, Fukuchi A, Udomkan B (2005) Primates 46:65-69.

SI Table 3

Fig. 13. Scatter plot of upper canine length vs. breadth. The mesiodistal length of the upper canine crown is plotted against the buccolingual breadth. The mesiodistal length is measured approximately parallel to the mesial and distal crests and the buccolingual breadth perpendicular to the length. In most cases, these length and breadth correspond to the maximum length and perpendicular breadth. The isometric line is drawn passing through the Nakali upper canine (KNM-NA47594). Compared with extant great apes, the Nakali upper canine is similar in breadth to male chimpanzees and female orangutans, but it is as small as that of female chimpanzees in its length. In other words, the Nakali upper canine has relatively short and broad crown proportion. In some cases, such as some (but not all) upper canines of Sivapithecus, the long axis of the root cross section is oriented fairly buccolingually. Consequently, the canine crown is buccolingually broadened, but the upper canines of Sivapithecus, including presumedly female ones, do not show the characteristic lingual morphology of N. nakayamai described in the text. The upper canine of the Afropithecus type (KNM-WT16999) is as broad as long, but this specimen is presumably male, and the combination of the robust, much larger canine with smaller cheek teeth is clearly a different pattern from that in N. nakayamai. Because the type maxilla of Samburupithecus kiptalami (sex unknown) does not preserve the upper canine, the mesiodistal and buccolingual dimensions of the canine socket are plotted. If these dimensions broadly reflect the crown dimensions, S. kiptalami would have had a buccolingually compressed upper canine. Most of the measurements were taken from original specimens by Y.K. Measurements of Sivapithecus were taken from the casts at the National Museums of Kenya. Data for Ankarapithecus were taken from refs. 1 and 2, and for two specimens of Ouranopithecus macedoniensis (RPl208 and RPl209) from ref. 3.

1. Alpagut B, Andrews P, Fortelius M, Kappelman J, Temizxoy I, Çelebi H, Lindsay W (1996) Nature 382:349-351.

2. Begun DR, Güleç E (1998) Am J Phys Anthropol 105:279-314.

3. de Bonis L, Koufos GD, Guy F, Peigné S, Sylvestrou I (1998) C R Acad Sci Paris 327:141-146.