www.sciencemag.org/content/355/6328/969/suppl/DC1

Supplementary Materials for

Late Pleistocene archaic human crania from Xuchang, China Zhan-Yang Li, Xiu-Jie Wu,* Li-Ping Zhou, Wu Liu, Xing Gao, Xiao-Mei Nian, Erik

Trinkaus*

*Corresponding author. Email: wuxiujie@ivpp.ac.cn (X.-J.X.); trinkaus@wustl.edu (E.T.)

Published 3 March 2017, Science 355, 969 (2017) DOI: 10.1126/science.aal2482

This PDF file includes:

Supplementary Text Figs. S1 to S30 Tables S1 to S14 References

1

Supplementary Materials for

Late Pleistocene archaic human crania from Xuchang, China Zhan-Yang Li, Xiu-Jie Wu, Li-Ping Zhou, Wu Liu, Xing Gao,

Xiao-Mei Nian, and Erik Trinkaus

correspondence to: wuxiujie@ivpp.ac.cn, trinkaus@wustl.edu

Supplementary Text and Methods

Supplementary Materials I: The Lingjing Site 1.1 The Lingjing Site

The Lingjing site (34° 04′ 08.6″ N, 113° 40′ 47.5″ E, elev. 117 m) is located in Lingjing town, northeast Xuchang county, Henan Province, northern China, about 120 km south to the Yellow River. The site consists of water-lain deposits, and it covers an area of >10,000 m2 (Fig. S1). It represents the infilling of a trailing edge depression of the Yinghe River, within the surrounding hilly topography. A number of small ponds were created in the Lingjing area from water welling up an underground river, through a series of springs, into depressions which lacked outflows. The Lingjing site represents the horizontally accumulated sediments from one of these springs and the gradual filling in of the pond (Figs. S1D, S2A). The levels, and particularly Layer 11 that yielded the human remains (Fig. S3), consequently mainly consist of fine-grained sediment with no evidence of large horizontal flow dynamics.

The site was discovered in the mid-1960s, when microblade tools and mammalian fossils were collected on the surface (23). The site was formerly saturated with water seeping from below; in April 2005, the spring stopped abruptly, and the original strata emerged. Systematic excavations were conducted from 2005 to 2015. The current excavation area is 551 m², consisting of two trial trenches (T1and T2), and twelve excavation trenches (T3 to T14). An abundant mammalian assemblage, tens of thousands of lithic artifacts, and human remains representing as many as five individuals, XUC 1 – XUC 5 (Table S1), were unearthed in Layer 11 in the T9 excavation area (Fig. S1).

2

Eleven stratigraphic levels were exposed and identified, with a total thickness over 9 m deep from modern surface (Fig. S3).The bottom of the stratigraphy has not been reached. The current stratigraphic sequence, from top to bottom, is as follows: Layers 1-4 (thickness: ≈2.9 m) are Holocence in age, yielding cultural materials representing the Neolithic to the Shang-Zhou Bronze Age. Layer 5 (thickness: 0.62–0.70 m) contains yellowish silty sediment and has yielded late Upper Paleolithic cultural materials including carved bone artifacts, microliths, perforated ostrich eggshells, hematite, and animal remains. Layer 6 is a flowstone layer (thickness: 0.4 m). Layer 7 is a yellowish silty soil (thickness: ≈0.85 m). Layer 8 is an archeologically sterile thin stratum of black ferruginous soils (thickness: ≈0.20 m). Layer 9 is an archeologically sterile and thick layer of brownish ferruginous silt, with a vertical root hole-like structure (≈2 m in thickness). Layer 10 is a similar brownish ferruginous silt with vertical root hole-like structures, containing lithic artifacts, animal bones and small pebbles (≈1.6 m in thickness). The lowest level, Layer 11, consists of sage-green silt with minimally inclined bedding; it contains the human fossils, abundant lithic artifacts, animal bones, and hematite inclusions (Fig. S2). Its total thickness is not known.

The mammalian and human remains are generally fragmentary, but there is no evidence of substantial horizontal displacement of the skeletal elements. Yet, considering surface runoff during the rainy season, oscillation in the pond level, and use of the locality by both mammals and human foragers with the associated trampling, fragmentation and minor displacement of the remains happened. Nonetheless, the human cranial pieces and the associated mammalian fragments and lithic artifacts were excavated in a small vertical range of the same layer. It has been possible to refit lithic artifacts and some of the mammalian remains, as well as the two human partial crania, each of which was found within a small area of the site (Fig. 1). There is no evidence for vertical transport of the remains, and even though some horizontal movement within levels of the archeological and paleontological materials in the Lingjing site apparently occurred, it appears to have been minor.

1.2 Dating the Lingjing Site

During the excavation, we collected 26 samples of charcoal in four groups from Layer 5 for 14C dating, and 8 samples from Layers 9 to 11 for optically stimulated luminescence (OSL) dating. The charcoal samples were analyzed using AMS at the Institute of Accelerator Analysis Ltd. (IAA) in Japan. Three of the sample are outliers, but the other 23 samples cluster at 11,300 ± 50 14C BP and 11,940 ± 50 14C BP, providing a mean age of 11,620 ± 10 14C BP. These dates calibrate to 13,409 – 13,494 cal BP, with a median probability value of 13,452 cal BP (24,25).

The eight samples from Layer 9 to 11, from a depth range of 6.35 to 9.55 m, were dated by means of the OSL method (see SI Section II for the OSL dating details). Four samples and the two samples were collected from the above and below human fossil layer, respectively. Two parallel samples were obtained from the same layer as the human fossils. Four different techniques applied to two types of minerals, quartz and potassium-feldspar, extracted from the sediment samples yielded early Late Pleistocene ages ranging from 78 ± 4 ka to 123 ± 10 ka. The K-feldspar in particular has median ages ≈88 ka and ≈95-100 ka for Layers 9 and 10. The parallel samples obtained from Layer 11 are dated to between ≈105 ka and ≈125 ka, corresponding to the last interglacial paleosol S1 in the Loess Plateau of China and the early Marine Isotope Stage 5 (MIS 5; MIS 5e to 5d).

3

Note that the ages obtained for the archeological Layer 10, overlying the human fossil Layer 11 for a combined depth of ≈2.0 m, also date to MIS 5. There is no evidence of vertical transport between Layers 10 and 11, but the similarity in age of these two levels, based on the OSL determinations, means that any mixing of materials between the levels would have little effect on the geological ages of these MIS 5 archeological and paleontological remains.

1.3 Faunal and Lithic Remains

The mammal fossil assemblage found at Lingjing is composed of 21 species, including 8 extinct species (Pachycrocuta cf. sinensis, Palaeoloxodon sp., Coelodonta antiquitatis, Stephanorhinus kirchbergensis, Sus lydekkeri, Axis sp. n., Megaloceros ordosianus and Bos primigenius), which account for 38% of the mammal species. Although four of theselarge mammal extinct species (Palaeoloxodon sp., C. antiquitatis, S. kirchbergensis, Sus lydekkeri) are present at the earlier Middle Pleistocene Zhoukoudian Locality 1, these species can also be found in the other, more recent comparative samples. For example, Palaeoloxodon sp. exists in the late Middle to early Late Pleistocene Dingcun and Salawus sites, C. antiquitatis is present in the Dingcun, Xujiayao and Salawus sites, Stephanorhinus kirchbergensis in the Dingcun assemblage, and Sue lydekkeri was found at Xujiayao. The extinct Bos primigeniusis also present in the Dingcun, Xujiayao and Salawus samples, but not at Zhoukoudian Locality 1.

The Lingjing assemblage does not contain any typical Early Pleistocene species, which were common in the Zhoukoudian Locality 1 faunal remains, such as Equus sanmeniensis, Hyaena sinensis, or Palaeoloxodon namadicus(26). Some typical Middle Pleistocene species, such as Sinomegaceros pachyosteus, Crocuta ultimaand Machairodus inexpectatus, are also not present in the Lingjing site.

Although the Lingjing collection has more extinct species, in common with the Zhoukoudian remains, than the assemblages from Dingcun, Xujiayao and Salawus, the most common mammals found in the Lingjing site, such as C. antiquitatis,E. hemionus, Cervus elaphus, Megaloceros ordosianus, Procapra przewalskii, Bos primigenius, are typical Late Pleistocene species. The faunal composition of the Lingjing site therefore has more species in common with Xujiayao and Salawus than with the Zhoukoudian Locality 1 and Dingcun samples. The Zhoukoudian fauna typically belongs to the Middle Pleistocene. The dating of the Dingcun, Xujiayao, Salawus and Upper Cave sites is ≈210-160 ka (27), ≈125-104 ka (28), ≈50-37 ka (29), and ≈34-29 ka (30). Therefore, the Lingjing fauna is typical of late Middle to early Late Pleistocene assemblages in China (31).

The lithic artifacts include stone cores, flakes, chunks, and animal bone tools (32). The primary technique used was hammer percussion, and a few of the quartz artifacts were made by bipolar flaking (32). The large amount of debitage and use-wear on the lithic remains suggests that the site was probably a tool making or using area (32-34).Modifications to the faunal remains reflects extensive processing of large herbivore carcasses (7,35).

Supplementary Materials II: Optically Stimulated Luminescence Dating 2.1. Sample Collection and Preparation

Eight samples were collected from the excavation section’s stratigraphic profile for OSL

4

dating in the luminescence laboratory of Peking University. Two samples (L1199, L1577) were obtained from Layer 11 at the same level as the human fossils. Four samples (L1434, L1436, L1271, and L1274) and two samples (L1202, L1578) were collected from the sediments above and below human fossils, respectively (Table S2).

The principal minerals currently used for luminescence dating are quartz and K-feldspar. There are respective advantages and disadvantages (36). In this study, we used both quartz and K-feldspar to determine the ages for the sediment samples.

Sample preparation was done under subdued red light in the dark room. Hydrochloric acid (HCl) (10%) and hydrogen peroxide (30%) were used to remove carbonates and organic matter, respectively. The samples were then washed with distilled water several times, followed by wet sieving to obtain polymineral coarse- and medium-grained fractions (90-125 μm and 45-63 μm). The polymineral fine-grained fraction (4-11 μm) was obtained by settling in 0.01 N sodium oxalate solution according to Stokes’ Law.

The polymineral coarse-grained fraction was treated with 40% hydrofluoric acid (HF) for 40-50 mins to dissolve feldspar and remove the alpha-irradiated layer of quartz grains, and it was subsequently etched with 10% HCl for 1 hour to obtain coarse-grained quartz. The fine-grained and medium-grained quartz grains were extracted by silica-saturated fluorosilicic acid (30%) for three days. After etching, any fluorides were dissolved using HCl (10%). The purity of the quartz extracted was checked by the ratio of IRSL to OSL and the characteristic 110°C TL peak of quartz. The potassium feldspar-rich fraction was obtained using sodium polytungstate solutions (ρ=2.58 g/cm3). The potassium feldspar-rich fraction is shortened here as K-feldspar. The fine-grained fractions were dispersed in acetone and then deposited onto 0.97 cm diameter aluminum discs. The medium-grained fractions were loaded onto aluminum discs using silicone oil.

2.2. Luminescence Measurements

All luminescence measurements were made with automated Risø TL/OSL readers (model 15 or 20) equipped with a 90Sr/90Y beta source for irradiation (37) and an EMI 9235QA photomultiplier tube. Blue light LED (470 ± 30 nm) stimulation set at 90% of 50 mW cm

-2 full power and 7.5mm Hoya U-340 filters (transmission between 290 and 370 nm) were used for the quartz OSL measurements. Medium-grained K-feldspar was stimulated with infra-red (IR) diodes emitting at 870 nm (90% of 280 mW cm-2) and a combination of Schott BG39/Corning 7-59 filters (transmission 320-460 nm) was used to detect the luminescence signals.

The concentrations of uranium (U), thorium (Th) and potassium (K) were measured using neutron activation analysis (NAA). For internal potassium content, a value of 12.5 ± 0.5% was used (38). Water contents were assumed to be 25 ± 5%, 30 ± 5% or 35 ± 5% according to the directly measured values and the sedimentary environment of the samples; the five samples from Layer 11 having higher water content reflects the situation observed at the time of sampling (see also SI 1.1 about the geomorphic condition of the site). Alpha efficiency values (a-value) of 0.04 ± 0.02 for quartz and 0.08 ± 0.02 for K-feldspar (39) were used. The dose rate calculation was performed using the DRAC (40). 2.2.1 Quartz OSL measurements

For the quartz OSL study, the single-aliquot regenerative-dose (SAR) protocol (Table S3) (41) was applied to the coarse-, medium-, fine-grained quartz of Sample L1199 and the medium-

5

grained quartz from Samples L1434, L1436, L1271, L1274 and L1202. Apart from the SAR protocol, the sensitivity-corrected multiple-aliquot regeneration dose (SC-MAR) protocol was employed to determine the equivalent dose, De, for the fine-grained quartz of Sample L1199, which was developed by Zhou and Shackleton (42) and successfully applied to a loess section at Luochuan by Lu et al. (43).

2.2.2. K-feldspar OSL measurements

K-feldspar has brighter signals and a higher saturation dose than that of quartz, and has been increasingly used in the recent years to date relatively old Quaternary sediments. However, the infrared stimulated luminescence (IRSL) signal of feldspars suffers from anomalous fading, and this kind of fading will cause age underestimation. To solve the problem, several correction methods have been proposed (e.g. 44-47), but each has limitations. Recent study showed that infrared stimulation of feldspar at an elevated temperature can reduce the laboratory fading rates in the blue region of the spectrum (48). Then a post-IR IRSL SAR protocol was introduced, with infrared stimulation of K-feldspar at an elevated temperature following a lower temperature IR stimulation (49,50), which can reduce or even eliminate laboratory fading of the IRSL signal. The post-IR IRSL SAR dating protocol, involving two experimental conditions with different preheating and stimulation temperature, was applied for equivalent dose determination on medium-grained K-feldspar for the eight samples. The first experimental condition includes the IRSL stimulation first at 50°C and then at 225°C, thus sampling the pIRIR225°C signals. To reduce the recuperation, infrared stimulation at 290°C for 40 s was applied at the end of each repeated cycle (51) (Table S4). The protocol is referred to as the pIRIR225°C protocol.

The second experimental condition includes a preheat temperature of 320°C for 60 s to natural/regenerative/test dose measurements, and stimulation with IR LEDs at 290°C for 200 s (i.e. sampling the pIRIR290 °C signal) following the IR stimulation at 50°C for 200 s (Table S5). The protocol is called pIRIR290 °C protocol. The pIRIR290 °C signal is considered to be stable and requires no anomalous fading correction (50). A ≈1.5 mm diameter aliquot was used to measure the coarse- and medium-grained quartz and feldspar.

2.3. Dating Results 2.3.1 Quartz OSL dating results

Linearity test (52) was carried out to check the validity of the sensitivity correction by a small test dose for fine-grained quartz SAR protocol for Sample L1199. With a preheat of 260°C for 10s and a 220°C cut-heat, a linear relationship between the regenerative and test dose OSL signals is observed with an intercept indistinguishable from zero (Fig. S4), indicating that the OSL test dose signal correlates well with the regeneration OSL signal in the SAR protocol.

As shown in Fig. S5, a De plateau is present for the preheat temperature in the range from 230°C to 280°C with a cut-heat temperature of 220°C for the fine-grained quartz SAR protocol for Sample L1199. Thus, a preheat of 260°C for 10 s combined with a cut-heat of 220°C for 0 s was employed. Dose recovery experiments (53) were performed with the SAR protocol for the fine-grained quartz of Sample L1199. The natural signal was bleached by the blue light for 100s at room temperature twice, with a delay of 104 s before the second one. The ratio of the recovered dose to the given dose of 245 Gy is 0.94 ± 0.02 (an average of eight aliquots).

The SAR and SC-MAR protocols were used to measure the equivalent dose for the fine-

6

grained quartz of Sample L1199. Ten aliquots were used to obtain De values with the single aliquot protocol, while the average OSL signals from three aliquots for each dose point were used for the multiple-aliquot protocols Dose response curves obtained using these protocols were fitted with a single saturating exponential function. Fig. S6 depicts the growth curves obtained and Table S6 shows the De and D0 (saturation dose) values determined using the SAR and SC-MAR protocols for the fine-grained quartz of Sample L1199.

The De values for the fine-grained quartz of Sample L1199 derived from the SAR and SC-MAR protocols are 234 ± 3 Gy and 222 ± 4 Gy, respectively. Coarse-grained quartz of Sample L1199, and medium-grained quartz of Samples L1434, L1436, L1271, L1274, L1199 and L1202 were measured with the SAR protocol (Table S3). Tables S6 and S7 show the dating results. The coarse-grained, medium-grained and fine-grained quartz are dated to 104 ± 7 ka, 94 ± 4 ka and 83 ± 4 ka with the SAR protocol for Sample L1199. The fine-grained quartz gives the youngest age, and the oldest age is produced by the coarse-grained quartz which has the smallest saturation dose. The De of the medium-grained quartz samples L1434, L1436, L1271, L1274 and L1202 obtained using the SAR protocol all exceed a value of 200 Gy, the ages range from ≈80 ka to ≈99 ka and increase with the depth except for Sample L1274.

2.3.2. Feldspar OSL dating results The medium-grained K-feldspar of Sample L1199 was chosen for the linearity test and

dose recovery experiment. LpIRIR225°C shows a relatively good relationship with TpIRIR225°C with an intercept indistinguishable from zero, and the correlation coefficient is ≈0.97. The result shows that the sensitivity change of the natural/regenerative pIRIR225°C signal can be corrected by the test dose. In the dose recovery experiments, the sample was bleached by the daylight for ≈4 hours, and was then given a laboratory dose of 265 Gy. Subsequently, the protocol in Table S4 was applied. The ratio of the recovered dose to the given dose for the medium-grained K-feldspar of sample L1999 is 0.99 ± 0.01 (an average value of five aliquots). The medium-grained K-feldspar pIRIR225 °C protocol is thus considered suitable for reproducible De determination.

Figure S7 depicts the IR-OSL decay curves for the medium-grained K-feldspar of Sample L1199. Dose response curve of the medium-grained feldspar with the pIRIR225 °C protocol is displayed in Figure S8.

The correction for anomalous fading of Huntley and Lamothe (45) was carried out for the K-feldspar signal with the pIRIR225 °C protocol. The anomalous fading test data are shown in Figure S9; the decay rate of anomalous fading was measured on the same aliquots with repeated post-IR IRSL SAR cycles and the g values are listed in Table S8. The dating results of the medium-grained K-feldspar are presented in Table S8 for samples obtained with the pIRIR225 °C protocol. After anomalous fading correction, the ages increase. The corrected ages are 88 ± 5 ka, 100 ± 8 ka, 95 ± 4 ka, 95 ± 4 ka, 109 ± 4 ka, 109 ± 5 ka, 108 ± 5 ka and 106 ± 6 ka for Samples L1434, L1436, L1271, L1274, L1199, L1577, L1202 and L1578, respectively. The ages of the samples increase with the sampling depth.

In the dose recovery experiment with the pIRIR290°C protocol, the sample (L1199) was bleached with a solar simulator (Hönle UVACUBE 400) for ≈10 hours. Following the bleaching, the aliquots were given 330 Gy β-dose. Initially, a high temperature IR bleach (325 °C, 40 s) was used at the end of each measurement cycle to remove any IR signal remaining

7

after previous dosing. However, this led to a dose recovery ratio (recovered/given dose) of 1.24 ± 0.08 (an average of six aliquots) for the medium-grained K-feldspar of Sample L1199. When removing the high temperature IR bleaching at the end of the sequence, the ratio of recovered to given dose became 1.07 ± 0.03 (an average of nine discs). The remaining dose using the protocol in Table S5 was 5 ± 1 Gy for the medium-grained K-feldspar of Sample L1199 after daylight bleaching for ≈10 hours, showing that the pIRIR290°C signals can be bleached down to a low level. No subtraction of this residual dose was applied to the samples measured with the pIRIR290°C protocol. The dating results are presented in Table S8. The pIRIR290°C ages agree well with the corrected ages obtained by pIRIR225 °C protocol for the medium-grained K-feldspar samples.

2.4. Discussions on OSL Dating Results and Conclusion

Nian et al (54) showed that seven different protocols involving the conventional OSL signal and recuperation OSL signal applied to the fine-grained quartz of Sample L1199 yielded a wide range of De values. As some of the protocols were shown either to have failed the linearity and dose recovery tests or to be affected by experimental artifacts, here we only report dating results with the relatively well established SAR and SC-MAR protocols for quartz and the pIRIR protocols for the K-feldspar (so as to avoid confusion with the unreliable results).

As the SAR protocol applied for De determination in Chinese loess was shown to suffer from the effect of saturation around 200 Gy (e.g., 49,52,55), we consider that there is also some degree of underestimation for the ages obtained with the SAR and SC-MAR protocols in this study. Although the coarse-grained quartz SAR protocol produced the oldest age (104 ± 7 ka) compared with those of the medium- (94 ± 4 ka) and fine-grained (83 ± 4 ka) quartz, it showed the lowest saturation dose D0 (98 ± 5 Gy). In this case, the Ln/Tn signal is considered to be too high on the dose response curve to be reliable. While the SAR OSL ages with the medium-grained quartz for Samples L1434, L1436, L1271 and L1202 increase with depth (except L1274), all of their De values are larger than 200 Gy and thus may underestimate the true ages.

The apparent ages of the medium-grained K-feldspar range from ≈80 to ≈98 ka using the pIRIR225°C protocol. After the anomalous fading correction with the Huntley and Lamothe method (45), the ages are increased by ≈10 ka. The two samples, L1199 and L1577, collected from the same layer yielded identical ages, i.e. 109 ± 4 ka and 109 ± 5 ka, respectively (Table S8).

The remaining dose of ≈5 Gy using the pIRIR290 °C protocol with the medium-grained K-feldspar of Sample L1199 confirms that the residual signal is negligible for relatively older samples. As shown in Table S8 and Figure S10, the medium-grained K-feldspar ages obtained with the pIRIR290°C protocol are highly consistent with the corrected ages derived from the pIRIR225 °C protocol.

In summary, considering the likely underestimation with the SAR and SC-MAR De values for the coarse-, medium- and fine-grained quartz and the indistinguishable ages using the medium-grained K-feldspar with the pIRIR225°C and pIRIR290°C protocols for the dated samples in this study (Table S8 and Fig. S10), we conclude that the depositional age of the sediment layer that contains the Xuchang human remains should be older than ≈90 ka. It is best considered to be between ≈105 ka and ≈125 ka.

8

Supplementary Materials III: The Human Fossil Remains: Preservation and Reconstruction 3.1. General Preservation and Procedure

The Xuchang human crania were excavated in situ in Layer 11, broken into pieces and dispersed over small areas of the excavation square T9 (Fig. 1C). The small distributions of the elements of each cranium, all except the right supraorbital torus and a small piece of cranial vault of Xuchang 1, indicate little dispersal of the pieces after breakage and deposition. The minimal dispersal of the preserved and conjoining elements is also indicated by the well-preserved surfaces of the individual elements and the tight joining of most of the pieces. The bone tissue of the individual pieces is solid. Surface detail is excellently preserved, as it is for the faunal remains. None of the remains exhibit cut marks, impact fractures, or evidence of gnawing, although the absence of the facial skeletons and postcrania implies some degree of dispersal and destruction of bone by trampling and/or scavenging. The breakage in situ of the neurocrania is most likely due to trampling into the soft sediment of dry bone combined with subsequent sediment pressure.

The largest piece is the largely complete right parietal of Xuchang 1 (piece 7001), with other pieces as small as 1-2 cm in maximum dimension. with minimal amounts of erosion. The breaks are mostly clean, dry bone breaks, and the adjacent pieces fit tightly together. The only tenuous joins are in the vicinity of the Xuchang 2 temporal bones.

The majority of the pieces derive from Xuchang 1 (26 elements), versus those for Xuchang 2 (16 elements) (Table S1). Three additional pieces, Xuchang 3 to 5, do not appear to fit with Xuchang 1 or 2, although the Xuchang 4 piece of supraorbital torus might derive from Xuchang 2.

To reconstruct the crania, initially the individuals pieces were molded, and replicas of the individual pieces were cast. The casts were then assembled to produce the primary reassemblies (see Figs. S17, S27); the individual fossil pieces remain separate, as excavated, for conservation (Figs. S11, S12). They were also used to produce an endocranial cast of Xuchang 1 (Fig. S13). All of the individual pieces were µCT scanned, and the crania were then virtually reconstructed, using the cast reconstructions as guides (Figs. 2, 3, S11, S12). The analysis is based on the morphologies of the individual pieces and the virtual reconstructions of the two crania, plus the isolated pieces.

3.2. Micro-CT scanning

The Xuchang 1 and 2 isolated fragments were μCT scanned using an industrial high precision μCT scanner (tube voltage: 150 kV; tube current: 110 μA; pixel size: 29-63 μm; error:

9

basis of the quantitative assessments (compare, for example, Figs. 2C, 3A and S17, and 2B, 3C and S27).

3.3. The Xuchang 1 Cranium: Preservation and Reconstruction

The Xuchang 1 cranium was reconstructed from 26 isolated fragments. Based on the continuity of the temporal line, articulated sutures, connected fractures and other aspects anatomical continuity within and between bones, 25 of the pieces articulate together. The additional piece is the right supraorbital torus (piece 14017), which is a close mirror image to the left one (Table S1; Fig. S11).

After reconstruction, Xuchang 1 is represented by an almost complete skull cap, consisting of lateral part of left supraorbital torus joining a large portion of the left frontal squama, the lateral part of right supraorbital torus, the complete right parietal bone, the greater part of the left parietal bone, the almost complete right temporal squama and petrous bone, the mastoid process and lower squamous portion of left temporal bone, and a large portion of the occipital bone including most of upper and middle part of occipital squamous portion but lacking the midline near inion. The morphological continuities between the left and right sides of the parietal, temporal and occipital bones, and between the frontal and parietal bones, are well preserved in coronal, sagittal and axial planes (Fig. S11B-E); the reconstruction is therefore reliable and can not be narrower in width.

The coronal, sagittal, lambdoidal, sphenoparietal, and squamosal sutures are almost complete with no traces of a beginning of obliteration. The Xuchang 1 fossil therefore probably represents a young adult individual. The large cranial size may suggest that it derives from a male.

3.4. The Xuchang 2 Cranium: Preservation and Reconstruction The Xuchang 2 cranium was reconstructed from 16 isolated fragments (Fig. S12), among

them four pieces of occipital bone (pieces 14021, 14022, 14023 and 14006) that can be joined together (Fig. S12B-E). Most of the nuchal plane, including inion, opisthion and the right posterolateral foramen magnum are preserved, along with the superior nuchal line and occipital squamous above it. In the region of asterion bilaterally and around the articulated lambdoidal, occipitomastoid and parietomastoid sutures, the occipital bone connects with the temporal mastoid regions (pieces 14027 and 14028) and the bilateral partial parietal bones (pieces 14011 and 14002). Bilateral temporal petrous portions (pieces 14004 and 14005), partial temporal squamous portions (pieces 14014 and 14015), and the pieces of the sphenoidal corner of right parietal bone can be positioned with respect to the posterior (mostly occipital) neurocranium.

Supplementary Materials IV: Xuchang Cranial Measurements 4.1. Linear Metrics and Indices

To quantify the dimensions and proportions of the Xuchang cranial remains, a series of standard linear osteometrics were taken on the original remains and the virtual reassembly of the

10

pieces. The measurements primarily follow Bräuer (56) and Howells (57). The same measurements were obtained, to the extent possible, on the comparative samples, from personal research and published descriptions of the original remains. The Xuchang 1, 2 and 4 measurements are in Tables S9 to S11.

4.2. Endocranial Morphometrics 4.2.1 Preservation and reconstruction

In order to assess endocranial proportions and to estimate the endocranial volume (ECV) of the Xuchang 1 cranium, an endocranial cast of the preserved neurocranium was made (Fig. S13), filling in missing portions using the existing contours. It consists of part of the left frontal lobe, the complete right and most of left parietal lobes, a large portion of right temporal lobe, bilateral and upper part of occipital lobe joining with two lateral small portions of the cerebellar hemispheres. Most of left temporal lobe, right frontal lobe and base are missing.

4.2.2 Endocranial measurements

A set of nine linear measurements were taken on the Xuchang 1 endocast (Table S12), using only those landmarks which derive from the preserved skeletal elements. The same set of measurements was taken samples of Pleistocene human and recent human endocasts to provide a framework for the estimation of the Xuchang 1 endocranial volume (Table S13); note that the samples differ from those employed for the exocranial metric comparisons, since this set of specimens was used only to estimate the endocranial volume of Xuchang 1. The selection of specimens was based on the preservation of endocasts where most of the parietal and occipital lobes exist. Specimens were measured at the Institute of Vertebrate Paleontology and Paleoanthropology (Beijing) and from the endocast collection of R.L. Holloway (Columbia University, New York). All measures were taken by XJW. Each endocast was measured three times, and the average was used as the final measurement. The ratio between the sagittal arc and chord of the parietal bones was used to quantify the degree of parietal bulging.

4.2.3 Endocranial volume (ECV)

The ECV of Xuchang 1 was estimated by using multiple linear regression between ECV (dependent variable) and the eight metric variables (independent variables; all of the ones in Table S12 except PAL) for all of the endocasts (Pleistocene and recent; n=88) and then for only the fossil specimens (n=38). Regression parameters were then used to estimate the Xuchang 1 cranial capacity, using the residuals between observed and expected values for each specimen to calculate the standard deviation of the uncertainty.

Using the total sample, the multiple regression provides an R2adj = 0.951, an ECV estimate for Xuchang 1 of 1,702 ± 56 cc, and a 95% CI of 1,636 – 1,767 cc. Using only the fossil sample the regression provides an R2adj = 0.971, an ECV estimate for Xuchang 1 of 1,801 ± 58 cc, and a 95% CI of 1,687 – 1,916 cc.

The larger pooled sample furnishes a slightly smaller 95% CI for the Xuchang 1 estimate, given the larger sample size of the reference sample. However, the endocranial proportions of Xuchang 1 more closely approximate those of Pleistocene archaic humans, and they contrast with the higher and more rounded ones of the recent human samples. As a result, an estimated

11

ECV of ≈1,800 cc is primarily employed for the Xuchang 1 cranium in the comparisons, bearing in mind that the estimate may be modestly high.

Supplementary Materials V: Comparative Materials 5.1. Comparative Sample Designations

In order to provide a comparative morphometric framework for the Xuchang crania remains, data were pooled from personal research and the published literature for Middle and Late Pleistocene human remains. The comparative material is grouped into fourteen morphological, temporal and geographic paleontological samples, plus a recent human one. The purpose here is to provide a general morphometric context for the Xuchang crania, and not to resolve the morphological, taxonomic and phylogenetic issues regarding these paleontological remains. For this reason, the fossil crania for which data are available are separated primarily into anatomically modern versus non-modern groups and then each of these groups is subdivided by time and geography. Formal taxonomy is not employed given the lack of agreement on the taxonomy of Middle and Late Pleistocene non-modern members of the genus Homo (e.g., 58-62).

It is recognized that some of the fossil specimens could be included in alternative samples, given varying interpretations of some aspects of morphology and ambiguities in their chronological positions. However, the groupings here are sufficient to provide the broad comparative framework necessary for evaluation of the Xuchang human crania, and reasonable alternative groupings (short of making each site sample its own unit) would have little effect on the assessment of the Xuchang fossils.

To the extent possible given the fossil record, the chronological and morphological groups are divided into four geographical regions. The mid latitude eastern Eurasian sample derives principally from China, with the addition of early modern human remains from Okinawa and northern Laos. The low latitude eastern Eurasian sample is principally Indonesian, with the addition of the Middle Pleistocene Narmada cranium from India and Late Pleistocene modern humans from Australia (recognizing that Australia is not technically part of Eurasia). The African sample is from across the continent, given the general dearth of remains from it; it is recognized that this approach may well be obscuring regional variation within Africa (63). In addition, the initial Late Pleistocene African sample includes the southwest Asian early modern humans from Qafzeh and Skhul, given the “African” affinities of the Qafzeh-Skhul human remains and the associated fauna (64,65). The last regional samples are from western Eurasia, including Europe and southwest Asia. Although Middle and Late Pleistocene human remains are known across central Eurasia, only the Obi-Rakhmat 1 Neandertal (66,67) provides comparative data (temporal labyrinthine proportions); none of the mature crania furnish morphometrics for comparison with the Xuchang crania.

To provide a generally ancestral reference for the early Late Pleistocene Xuchang fossils, data are included for four regional Middle Pleistocene samples. Since there are diachronic trends in each region, they are separated into earlier and later Middle Pleistocene samples, at approximately 350 ka years ago. A couple of the fossils (e.g., the “Middle Pleistocene” Broken Hill and Reilingen, as well as the “Neandertal” Forbes’ Quarry) do not have stratigraphic contexts, and they are placed within chronological samples based largely on their comparative morphology. In addition, the large Ngandong sample has proven difficult to place accurately in time; recent assessments suggest, however, that a middle to later Middle Pleistocene age is most

12

likely for them (68), and therefore they are conservatively placed within the later Middle Pleistocene.

It should also be mentioned that the earlier Middle Pleistocene European crania exhibit considerable variability, with one group (Arago, Bilzingsleben and Ceprano) exhibiting more ancestral features and another (Atapuerca-SH and Swanscombe) suggesting the beginning of the Neandertal lineage (60,69). Although one can sort most of these remains into separate groups, the full sample shows some degree of morphological continuity. They are therefore kept together as a temporal and geographic sample. The western Eurasian later Middle Pleistocene sample reflects the emergence of the Neandertals, such that by end of that period most of the Neandertal features were established in the region.

The earlier Middle Pleistocene African sample consists of crania with generally ancestral morphology (63), but the later Middle Pleistocene ones show the emergence of early modern humans, as reflected in Omo-Kibish 1 and Herto 5 (19,70). The other specimens within this group are generally late archaic but are often considered modern due to their absence of Neandertal features (71). The sample probably represents the morphological mosaic expected during the transition from earlier archaic to later modern human populations.

These Middle Pleistocene samples are followed by Late Pleistocene samples, which are separated by morphology, geography and Marine Isotope Stage chronology (MIS 5–3b versus MIS 3a–2) associations. The earlier group is predominantly western Eurasian and consists of Middle Paleolithic / Middle Stone Age associated remains of early modern humans and Neandertals, plus the northern Chinese Xujiayao sample. The second group is of Late Pleistocene Upper Paleolithic / Later Stone Age sensu lato associated early modern humans.

The MIS 5–3b early modern human sample derives from east Africa and southwest Asia and exhibits a suite of craniofacial and postcranial derived modern human features (72). The second sample is of Neandertals sensu stricto with their complex mix of ancestral (archaic), uniquely derived, and shared (with modern humans) derived features (72). They represent a morphological continuum with western Eurasian later Middle Pleistocene humans, and span the Late Pleistocene into the middle of MIS 3 and the initial Upper Paleolithic.

The only other eastern Eurasian cranial remains securely dated to this time period are the MIS 3b early modern human Tam Pa Ling 1 (73) and the scattered remains from Xujiayao (6, 74-76). The former fossil does not provide data for the same regions as the Xuchang crania, but the latter ones are included in the comparisons. There are also earlier Late Pleistocene human remains from the sites of Huanglong, Zhiren and Fuyan in southern China, that have been attributed to early modern humans (21,22,77). They consist of isolated teeth and one partial mandible. Morphological comparisons can therefore not be made between them and the Xuchang crania.

There are also the Late Pleistocene “Denisovans” from the Altai, represented by a (destroyed) immature phalanx and two M3s, from which there are inferred DNA sequences (78). Their relationship to other Pleistocene humans is unknown (62), and the only diagnostic skeletal morphology known for them (very large M3s) most closely resembles that of a European early modern human (79). Given the absence of cranial data for the “Denisovans” and of dental remains or human aDNA from Xuchang, they are not considered in the assessment of the Xuchang crania.

The MIS 3a–2 Upper Paleolithic / Later Stone Age early modern human samples are from each of the four regions. The western Eurasian sample is primarily earlier (MIS 3a) Upper

13

Paleolithic in context, whereas the low latitude eastern Eurasian sample has primarily later (MIS 2) remains. The smaller mid latitude eastern Eurasian and the African samples are more distributed across the time period.

The fossils in the different samples vary considerably in terms of completeness and in the availability of morphometric data from them. In the comparisons, therefore, the representation of each of the samples varies considerably. Moreover, a few of the Middle Pleistocene samples are dominated single site samples (Zhoukoudian Locality 1 for the earlier mid latitude eastern Eurasian sample, Atapuerca-SH for the earlier western Eurasian one, and Ngandong for the later Middle Pleistocene low latitude eastern Eurasian sample). These comparative samples should nonetheless be sufficient to provide the necessary framework to evaluate the size and proportions of the Xuchang crania.

The samples therefore consist of four geographical groupings: Mid latitude eastern Eurasians (Mid Lat East), Low latitude eastern Eurasians (Low Lat East); African (includes southwest Asian early modern humans for the earlier Late Pleistocene); and western Eurasian (West Eurasia). They are arranged in the raw measurement and index plots by time period, bearing in mind that they are also sorted by morphology and geography within those chronological brackets: Earlier versus Later Middle Pleistocene; MIS 5-3b versus MIS 3a-2 Late Pleistocene; and Recent (late Holocene). The abbreviations for the bivariate plots are provided in their captions.

5.2. The Fossil Remains by Sample 5.2.1. Earlier Middle Pleistocene Humans

5.2.1.1. Mid latitude eastern Eurasian Hexian, Nanjing, Zhoukoudian Locality 1 5.2.1.2. Low latitude eastern Eurasian Sambungmacan 5.2.1.3. African Bodo, Broken Hill, Ndutu, Saldanha, Salé 5.2.1.4. Western Eurasian Arago, Aroeira, Atapuerca-SH (Sima de los Huesos), Bilzingsleben, Ceprano, Steinheim,

Swanscombe, Vértesszöllös 5.2.2. Later Middle Pleistocene Humans

5.2.2.1. Mid latitude eastern Eurasian Dali, Maba, Jinniushan 5.2.2.2. Low latitude eastern Eurasian Narmada, Ngandong 5.2.2.3. African Herto, Irhoud, Laetoli, Omo-Kibish 5.2.2.4. Western Eurasian Biache, Castel di Guido, La Chaise-Suard, Ehringsdorf, Fontéchevade, Petralona,

Reilingen, Zuttiyeh

14

5.2.3. Earlier Late Pleistocene Humans (MIS 5-3b) 5.2.3.1. Mid latitude eastern Eurasian Xujiayao 5.2.3.2. African and southwest Asian Aduma, Bouri 5, Qafzeh, Skhul 5.2.3.3. Western Eurasian (Neandertal) Amud, La Chaise-Bourgeois-Delaunay, Combe Grenal, La Chapelle-aux-Saints, Còva

Negra, La Ferrassie, Feldhofer, Forbes’ Quarry, Guattari, Krapina, Kůlna, Marillac, Monsempron, Palomas, La Quina, Saint-Césaire, Saccopastore, Šala, Salzgitter-Lebenstedt, Shanidar, Spy, Tabun, Vindija 5.2.4. Later Late Pleistocene (Modern) Human (MIS 3a-2)

5.2.4.1. Mid latitude eastern Eurasian Liujiang, Minatogawa, Tam Hang, Zhoukoudian-Upper Cave 5.2.4.2. Low latitude eastern Eurasian Cohuna, Coobool Creek, Keilor, Kow Swamp, Mossgeil, Nacurrie, Nitchie, Talgai,

Wajak, Wilandra Lakes (WLH) 5.2.4.3. African Fish Hoek, Hofmeyr, Kubbaniya, Nazlet Khater 5.2.4.4. Western Eurasian Arene Candide, Barma Grande, Brno, Cap Blanc, Caviglione, Chancelade, Cioclovina,

Cro-Magnon, Dolní Věstonice, Grotte-des-Enfants, La Madeleine, Mladeč, Muierii 1, Oase, Oberkassel, Ohalo, Ostuni, Paglicci, Parpalló, Pataud, Pavlov, Prĕdmostí, Sunghir, Zlatý kůn 5.2.5. Recent humans (RM)

Cranial metrics: Asia (n=70), Europe (n=10), Afirca (n=10). Labyrinthine morphometrics: Asia (n=26), Europe (n=101), Diverse (n=55)

Supplementary Materials VI: Comparative Morphology 6.1. Xuchang 1, 2 and 3 Endocranial Morphology 6.1.1 General endocranial morphology

Viewed superiorly (Fig. S13A), the endocranial cast is long, wide and low. The widest point is situated at the lateral superior border of the temporal lobes. The parietal areas show a flat superior profile and clear parasagittal depressions. In lateral view (Fig. S13B,D), the traces of the middle meningeal vessels show the dominance of the anterior branches with the network composed of many channels. The temporal region is wide. In posterior view (Fig. S13C), the occipital lobe forms a rounded elevation on the posterior aspect of the cast.

The occipital lobes are bulging and posteriorly projecting, especially with respect to the transverse sinus sulci, and hence the tentorium cerebelli. A similar configuration is present in the Xuchang 2 occipital region. There does not appear to be a lambdoid or supralambdoid flattening, especially in right lateral view (Fig. S13B). Xuchang 1 therefore does not have a distinct occipital bun, an exocranial protrusion from differential lambdoid sutural growth accommodating the occipital lobes (80); they are found in almost all Neandertals, about half of

15

the western earlier Upper Paleolithic modern humans, a few African late Middle Pleistocene humans and the occasional recent human (80,81).

The bi-lateral transverse sinus sulci of Xuchang 1 are wide, and they cross to the temporal bones along the superior margins of endoasterion. The inferior margins of the sulci are at endoasterion, and the middles of the sulci cross the posteroinferior parietal bones. The sulci for the transverse sinuses of Xuchang 2 follow the same path, marginally superior of endoasterion. For comparison, the sulcus for the sinus for the Xujiayao 3 occipital bone is at asterion, whereas that of the Xujiayao 15 temporal bone would have crossed above asterion and then across the parietomastoid suture to the sigmoid sinus. As such, the Xuchang 1 and 2 and the Xujiayao 3 sulci reflect a moderately low position for the tentorium cerebelli, but relatively above the infraasterion position of many Middle Pleistocene and Neandertal ones (82).

6.1.2 Endocranial volume (ECV)

The estimated ECV of ≈1,800 cc for Xuchang 1 (SI 4.2.3) places it at the very top of the Pleistocene human ranges of variation (Fig. S14). Only one specimen, the Upper Paleolithic early modern human Barma Grande 2, provides a higher estimate (≈1,880 cc). Using the smaller estimate of ≈1,700 cc for Xuchang 1 would still keep it at the top of Pleistocene ranges of variation, since only two other specimens have ECVs as high or higher than that estimate: the Upper Paleolithic Grotte-des-Enfants 4 (1,775 cc) and the Amud 1 Neandertal (1,736 cc). In the sample of 50 recent humans, only two have ECVs ≥1,700 cc and none are ≥1,800 cc.

Neither of the Xuchang crania is associated with a body mass estimate (no human postcrania have been identified from the site), but minimum encephalization quotients (EQs) can be calculated comparing the Xuchang 1 ECV estimates to the maximum body masses estimates (based on femoral head diameters (83)) available for Late Pleistocene humans: ≈81 kg for the La Chapelle-aux-Saints 1 Neandertal and the Barma Grande 2 Upper Paleolithic modern human; and ≈85 kg for the Tianyuan 1 early modern human. Following Martin (84), the lower body mass estimate provides EQs of 5.45 and 5.15 for the 1,800 and 1,700 cc estimates; the higher body mass value furnishes slightly lower EQs of 5.25 and 4.97 respectively. Even lower body mass values would produce slightly higher EQs. All of these estimates fall well within Late Pleistocene values, for both the Neandertals and early modern humans (8).

It is not possible to estimate the ECV of Xuchang 2, but it is apparent from its posterior cranial breadths that it was considerably smaller than Xuchang 1, closer to the other later Pleistocene values (Figs. S15, S16). The cranial base breadths of Xuchang 1 are large, with its maximum cranial and bi-auriculare breadths exceeding the comparative sample ranges. In contrast, these two cranial base breadths for Xuchang 2 are well within the comparative samples’ ranges of variations. Its maximum cranial breadth is modest for a Middle Pleistocene humans, in the middle of the earlier Late Pleistocene samples, and moderately high for a MIS 3a-2 human, and its bi-asterionic breadth is in the middles of all but the MIS 3a-2 and recent human distributions. Yet, its bi-auriculare breadth is well within the Middle Pleistocene and Neandertal distributions, but large for a Late Pleistocene non-Neandertal or recent human.

6.2. Xuchang 1 Overall Proportions

16

The maximum breadth of the Xuchang 1 cranium is located inferiorly, in the supramastoid region (Figs. 2, S17), and both that value and its bi-auriculare breadth exceed the values available for the comparative specimens (Figs. S15, S16). The Xuchang 2 cranium also has its widest point inferiorly, even though its maximum and bi-auriculare breadth values are smaller and unexceptional for a Middle or Late Pleistocene human. Their broad cranial bases are also reflected in an index between the maximum cranial breadth and the biparietal breadth, which provides of values of 102.1 and 101.1 for Xuchang 1 and 2 respectively. This pattern is most common among the eastern Eurasian Middle Pleistocene crania, with all of the earlier (n=10) and later (n=11) ones having indices >100. Among the western Eurasian and African earlier Middle Pleistocene crania, only a third (n=9) have indices >100. In the small later Middle Pleistocene western samples, 60% of the African ones (n=5) and 75% of the European ones (N=4) have broader cranial bases. With their laterally convex parietal profiles, all of the Neandertals (n=13) have the widest breadth on the parietal bones (hence indices of 100), as do the three sufficiently complete and undistorted MIS 5 modern humans and all but two of the MIS 3a-2 modern humans (n = 81). The inferior position of the maximum cranial breadth of the Xuchang crania is therefore more common among the earlier samples, but it is also occasionally present among later Pleistocene humans.

The cranial vault height of Xuchang 1 (Fig. S18A) is moderately high for an earlier Middle Pleistocene human and low for an early modern human, but it is not unusual for a later Middle Pleistocene human or a Neandertal. However, when it is scaled against its very broad cranial base (Fig. S18B), it has the next to lowest value for a Middle or Late Pleistocene cranium; only the estimated value for Narmada 1 is lower. Yet, as noted above (SI 3.3), the bones of the posterior neurocranium fit securely together along clean breaks, there is an even coronal contour across the parietal bones, and pieces of occipital bone (although incomplete) cannot be placed closer together, especially on the more complete right side (Fig. S17). Xuchang 1 therefore has an unusually broad and low neurocranium, most closely approaching the proportions of the earlier Middle Pleistocene Zhoukoudian sample and a few of the later Middle Pleistocene crania.

It is also possible to assess the proportions of the Xuchang 1 neurocranium using a more limited set of measurements and smaller comparative samples. It is done using both seven linear measurement and five indices. In the former assessment (Fig. S19A), Xuchang 1 is an outlier along PC-1, due largely to its overall large dimensions and especially its broad cranial base. With respect to PC-2, it is aligned principally with the eastern Eurasian Middle Pleistocene crania plus Petralona 1. It is well separated from all of the Late Pleistocene and recent human crania except the MIS 5 Tabun 1. PC-2 is influenced primarily by parietal length and vault height, higher values in both producing higher PC-2 values; the modest Xuchang 1 values for these measurements, moderately high for Middle Pleistocene humans and low for Late Pleistocene crania (Figs. S18, S22), account in part for its low position on PC-2.

The effects of size are largely removed by using a set of indices (Fig S19b), four of which are chord/arc indices and hence reflect parietal or occipital curvature. Xuchang 1 remains primarily with the eastern Eurasian Middle Pleistocene crania along PC-1, as well as with Petralona 1 and Tabun 1. The moderately high position of Xuchang 1 reflects its rather flat parietal (bregma-lambda) sagittal arc and its low and wide vault (see Figs. S18, S23). PC-2 does little to separate the samples, with Xuchang 1 falling relatively high primarily with respect to the eastern Eurasian and recent modern human crania. The relatively high position of Xuchang 1 reflects a relatively flat superior occipital bone in the sagittal plane and along the lambdoid suture.

17

6.3. Xuchang 1 and 4 Supraorbital Morphology

Xuchang 1 preserves the lateral portions of both supraorbital tori (the left one connecting to the frontal squamous portion), and Xuchang 4 retains a middle portion of the left one. All three pieces were broken medially at the lateral extents of their frontal sinuses, indicating the presence of frontal sinuses that extended to near mid-orbit.

The Xuchang 1 supraorbital tori are relatively even in thickness from midorbit to the frontozygomatic suture, with a modest increase in thickness laterally. They round superiorly into shallow but anteroposteriorly broad supratoral sulci. It is not possible to determine the nature of their form through the glabellar region. The Xuchang 4 toral fragment is similar in form, but its supratoral sulcus appears to be slightly shallower and not as deep anteroposteriorly. However, that assessment depends on the mediolateral placement of the toral fragment, given the normal mediolateral changes in supratoral sulcus form.

The midorbit thickness of the Xuchang 1 torus is modest for a Middle Pleistocene human, with only Nanjing 1 and Arago 21 presenting slightly thinner tori (Fig. S20A); it is in the middle of the Neandertal distribution, and among the thicknest of the early modern human ones (the very thick MIS 3a-2 torus is WLH 19). The lateral toral thickness of Xuchang 1 is relatively higher, among the thinner Middle Pleistocene ones, in the middle of the Neandertal distribution, but above all of the early modern human ones except WLH 18 and 69 (Fig. S20B). The Xuchang 4 midorbital thickness falls well within later Middle and earlier Late Pleistocene variation (Fig. S20A).

When the two supraorbital torus thickness measurements are combined into an index (Fig. S21), there is a variable but within-region consistent pattern of reduction in the midorbit to lateral orbital thickness. The proportions of the Xuchang 1 torus fall well within those of later Middle to later Late Pleistocene samples, with its proportions mostly closely matching those of the earlier Late Pleistocene humans.

6.4. The Xuchang 1 and 2 Parietal Bones

The Xuchang 1 parietal bones are generally average in sagittal length relative to Middle and Late Pleistocene human crania, being moderately long especially for the earlier Middle Pleistocene samples and moderately short for the Late Pleistocene modern human samples (Fig. S22A). At the same time, the oblique (bregma-asterion) dimension of the right parietal bone is one of the highest Pleistocene values (Fig. S22B), being exceeded only by Omo-Kibish 2 and four MIS 3a early modern humans. The contrast in the positions of Xuchang 1 in these two measures reflects a rather broad for length parietal bone.

Its parietal bones exhibit a relatively flat mid-sagittal (bregma-lambda) arc with a more pronounced one through the parietal eminence (reflected in its bregma-asterion arc). Both of these arcs are measured almost entirely on the complete and undistorted right parietal bone, specimen 7001; only 18 mm of the anterior sagittal suture is from another piece (7024), and it connects securely to the larger piece. There is therefore no estimation of the values.

The sagittal arc of the parietal bones of Xuchang 1 is largely flat from bregma for ≈70 mm toward lambda, and it then curves inferiorly to lambda. It remains convex along its entire length, such that there is no supralambdoid flattening. The resultant parietal index (96.1; Fig. S23A) is one of the highest known for Middle and Late Pleistocene humans. It is above that of Xujiayao 6

18

(94.5) and is matched or exceeded among eastern Middle Pleistocene humans only by Ngandong 5 and 10 and Sambungmacan 4. It is also above those of all of the western Eurasian and African Middle Pleistocene crania, and only two Neandertals (n=15), Còva Negra 1and Saccopastore 1, have higher indices. All of the early and recent modern humans have lower indices, with only those of Brno 2 (94.7) and Pavlov 1 (95.9) being close to the Xuchang 1 value.

At the same time, the parietal eminence of Xuchang 1 is moderately prominent, although it provides an even curvature in the coronal plane (Fig. S17). It lacks the external angulation at the eminence seen in most early modern and recent humans, or the angulation from the angular torus evident in some Early and Middle Pleistocene crania. This curvature is partly reflected in the bregma-asterion index (Fig. S23B), in which Xuchang 1 is aligned with all of the Late Pleistocene comparative samples but among the more curved of the Middle Pleistocene ones.

The external surfaces of the Xuchang 1 parietal bones are smooth, with only faint markings of the temporal lines. There is no evidence of an angular torus on the Xuchang 1 parietal bones, although there is a modest swelling extending posterosuperiorly from entomion on the Xuchang 2 right parietal bone. As noted above (SI 6.1.1), the transverse sinus sulci of both crania cross the lambdoid suture on the superior edge of asterion.

There is a general trend toward decreasing parietal thickness, measured at the eminence, through Middle and Late Pleistocene humans (Fig. S24A). The Xuchang 1 thickness (7.9 mm) is modest for an earlier Middle Pleistocene human, similar to those of the later Middle and earlier Late Pleistocene humans, and it is not unusual for a more recent modern human. The eminence region of Xuchang 2 is not preserved, but the preserved parietal superior margins towards its eminences suggest a similar parietal thickness (Fig. S12).

The parietomastoid sutures are preserved on both sides of each cranium (see Fig. S27). Those of Xuchang 1 are distinctly horizontal. The two of Xuchang 2 are less complete. However, the right one is horizontal, and the left one slopes only minimally toward asterion. It slopes distinctly inferoposteriorly on the Xujiayao 15 temporal bone and the Xujiayao 6 parietal bone. The orientation of this suture (horizontal versus posteriorly sloping) is variable within samples, but it tends to be horizontal particularly among the Neandertals (85-87).

6.5. The Xuchang 1 and 2 Occipital Bones

The Xuchang 1 occipital bone preserves major sections of the superior squamous portion (upper scale), although it lacks the midline of the superior nuchal line and hence inion. It is nonetheless possible to continue the two sides of the nuchal torus across the midline to approximate the position of inion for measurements. It retains only a small portion of the right nuchal plane. In contract, the Xuchang 2 occipital bone retains most of the nuchal plane, especially on the right side, and preserves both inion and opisthion. Superiorly it is broken off ≈40 mm above inion.

The squamous portion of the Xuchang 1 occipital bone is evenly convex in sagittal and transverse directions. It is moderately long relative to the breadth of the cranial base (despite its very broad posterior cranial base (Fig. S16B)). Its index (Fig. S25) is at the top of the Middle and earlier Late Pleistocene distributions, falling in the middle of the later Late Pleistocene and recent modern human values.

The Xuchang 1 occipital bone exhibits a relatively prominent nuchal torus, albeit one that is restricted to the middle two-thirds of the superior nuchal line (Fig. S17). Where best preserved on the right site, it is excavated on its inferior side for the superior nuchal line and the

19

semispinalis capitis fossa. There is also a modest sulcus superiorly between the torus and the squamous surface. In contrast, the Xuchang 2 occipital bone has little development of a nuchal torus. There are prominent fossae for semispinalis capitis on both sides producing a concavity (or partial sulcus), especially across the middle two-thirds of the superior nuchal line. However, there is no sulcus above the superior nuchal line, with only a slight thickening of the bone along the superior nuchal line and the external surface, then rounding smoothly onto the upper scale to where it was broken postmortem. What relief is present along the superior nuchal line is restricted, as with Xuchang 1, to the middle two-thirds of the superior nuchal line.

In the context of this nuchal torus variation, both Xuchang 1 and 2 have modest thicknesses across the superior nuchal line / nuchal torus (Fig. S24B). Their values of 12 and 11 mm respectively are similar to those of the Neandertals and more recent humans. Middle Pleistocene crania, with their more prominent nuchal tori, tend to have greater thicknesses, some substantially so; they are joined by the non-Neandertal MIS 5 crania.

Xuchang 2 lacks an external occipital protuberance, in that the midline curves smoothly from above to below the superior nuchal line, where the two sides of the superior nuchal line curve inferiorly to the midline around the semispinalis capitis fossae (Fig. S26A).

Just above where the external occipital protuberance would be, there is an ovoid depression, ≈12.6 mm high and ≈12.0 mm wide (Fig. S26A) (Xuchang 1 lacks inion and the suprainiac area, hence its original morphology in this area is unknown). It has irregular pitting within it. The depression is in the position of a suprainiac fossa, a trait developed to varying degrees in 100% of the Neandertals, both mature and immature (9,88,89). A similar depression, which may or may not be homologous, also occurs occasionally in other archaic humans, as well as among modern humans (11). Among the Neandertals, it is commonly a transversely elongated depression, delimited from the surrounding external occipital surface bone, and variably pitted or rugose within its floor. The Xuchang 2 depression corresponds to this morphological pattern, although it is subcircular rather than transversely ovoid.

In addition, Neandertal suprainiac fossae are associated with a maintenance of the external table cortical bone thickness through the fossa, such that the depression is described (9) as being created by a thinning of the underlying diploë. Among early and recent modern humans, the depression appears to be the result of an external table thinning. In the Xuchang 2 occipital bone, the external table maintains its thickness through the depression, such that the transverse external curve of the bone is reflected in the internal curve of the external table (Fig. S26B,C). It is not possible to describe Xuchang 2 has having “diploic thinning,” since the prominent endocranial crest for the superior sagittal sinus is opposite the fossa. The suprainiac “depression” of Xuchang 2 therefore follows the cross-sectional characteristics of a Neandertal suprainiac fossa.

The Xuchang 1 and 2 occipital bones together therefore exhibit all four aspects of the superior nuchal line region that have been considered distinctive of the Neandertals (9,11,88): a modest nuchal torus restricted to the middle of the superior nuchal line, the absence of an external occipital protuberance, the presence of a suprainiac fossa, and maintenance of external table thickness through the fossa. The only difference is with respect to the shape of the fossa, which is subcircular rather than transversely ovoid.

6.6. The Xuchang 1 and 2 Temporal Bones 6.6.1 External temporal morphology

20

The Xuchang temporal bones retain relatively complete inferior portions, with especially the petrous and retromastoid regions, portions of the mastoid process bases and variably complete temporomandibular articulations and auditory meati. In addition, the right one of Xuchang 1 has an almost complete squamous portion (Fig. S27). Only the left mastoid process of Xuchang 1 is intact.

An index of the temporal squamous height to length (Fig. S28) shows little difference across the comparative samples, although the earlier Middle Pleistocene eastern Eurasian samples have some of the lowest values. Xuchang 1 falls well within the overall variation, and among the temporal bones from the Late Pleistocene samples with the lower squamous portions. The modest temporal squamous index of Xuchang 1 is likely related to its overall low neurocranial vault.

The left mastoid process is relatively short (vertical height from porion ≈15 mm) and broad at its base (≈24 mm), The lateral surface is smooth, and there is no anterior mastoid tubercle (sensu 88,89). The lateral surface is angled inferomedially, as in Middle Pleistocene humans and most Neandertals, but unlike the more vertical lateral faces of early and recent modern human mastoid processes. The process extends little inferiorly of the floor of the digastric sulcus. The juxtamastoid area is not preserved, but even a small juxtamastoid eminence would have projected inferiorly of the mastoid process, as it does in many archaic humans but rarely in recent humans.

6.6.2. Temporal labyrinthine morphology

The Xuchang 1 right temporal bone and both Xuchang 2 temporal bones preserve their petrous portions sufficiently intact to permit analysis of their semicircular canal configurations (Fig. S29). Therefore, following standard procedures (10,90), the radii, proportions, and relative positions of the semicircular canals were quantified for Xuchang 1 and 2 (Table S14), which were then compared to two east Asian archaic humans and samples of Middle Pleistocene western Eurasian archaic humans, Neandertals, earlier and later Late Pleistocene modern humans, and recent humans (Fig. S30). Radii indicate the overall sizes of the canals; absolute and relative (%R) radii were calculated for the anterior (ASC), posterior (PSC), and lateral (LSC) canals of the Xuchang and comparative sample semicircular canals. The sagittal labyrinthine index (SLI) was calculated from the posterior and lateral canals using the equation: SLIi/ (SLIs +SLIi) × 100; it reflects the vertical position of the lateral canal relative to the posterior one.

As previously documented (6,10), the western Eurasian Neandertal temporal labyrinths are frequently distinct from those of both ancestral Early Pleistocene humans and modern humans in the relative sizes and positions of their semicircular canals, especially a small anterior canal, a larger lateral one, and a more superior position of lateral one relative to the posterior one (Fig. S30A). Moreover, Middle Pleistocene Europeans share many of these features with the Neandertals (91). The northern Chinese Xujiayao 15 labyrinth follows the Neandertal pattern, and the Hexian 1 labyrinth is intermediate between the Neandertals and early/recent modern humans. The Xuchang labyrinths are also close to that of the central Asian Obi-Rakhmat 1 Neandertal (66).

The Xuchang 1 and 2 labyrinths fall in the small overlap zone of the Neandertals and modern humans, especially in proportions of the anterior and lateral radii and the sagittal labyrinthine index (Fig. S30). They are not as much in the Neandertal distribution as Xujiayao 15, but they remain close to the Neandertals and their European Middle Pleistocene predecessors

21

and distinct from the vast majority of the early and recent modern humans. The Xuchang labyrinthine proportions thus further document the variation in this feature in eastern Eurasian archaic humans and reinforce the presence of a derived, “Neandertal,” configuration across Eurasian among late archaic humans.

It has been suggested (91) that the Neandertal labyrinthine configuration may be related to combinations of cranial base proportions and patterns of encephalization, and hence not necessarily of relevance to population relationships. It is certainly possible that some aspects of the Xuchang 1 and 2 labyrinthine morphology are secondarily related to their cranial configurations. However, the suite of differences between the Xuchang cranial proportions and those of the Neandertals noted here (especially their wide cranial bases), as well as those between the Xujiayao remains and the Neandertals (6,75,92), make such an interpretation inadequate to explain the morphometric similarities between these eastern and western earlier Late Pleistocene Eurasian archaic human temporal labyrinths.

6.7. Xuchang Morphology Discussion

These overall and regional morphological comparisons of the Xuchang cranial remains provide a mosaic pattern with respect to other samples (defined morphologically, regionally and temporally) of Middle and Late Pleistocene humans. The primary comparisons of interest are to the earlier samples of principally archaic humans and to the approximately contemporaneous early Late Pleistocene humans. The former samples are variably represented across the Old World, but should be sufficient to provide a general ancestral framework for the Xuchang human remains. The latter samples include both late archaic and early modern humans.

In the western Old World, the early Late Pleistocene sample contains the well-known Eurasian Neandertals and early modern humans from eastern Africa and southwestern Asia. In the eastern Old World, they include one fragmentary late archaic human sample (Xujiayao in northern China), two samples with a mixture of archaic and modern human features (Huanglong and Zhiren in southern China), and possibly one with only recent human features (Fuyan in southern China). The Xujiayao sample provides a number of individual comparisons to the Xuchang crania, but the absence of largely complete neurocrania limits overall assessments. The southern Chinese samples retain only isolated teeth and a partial mandible, and hence provide little basis for comparison (the same applies to the human remains from Denisova – see SI 5.1); it cannot be determined whether they are closely related to the Xuchang remains, even though they appear to be distinct dentally from the Xujiayao sample (76).

In this context, the Xuchang crania present a combination of features that variably align them with different earlier and contemporaneous samples. They share with earlier eastern Eurasian remains, both mid and low latitude samples, in having low, sagittally flat cranial vaults and especially neurocrania that are widest inferiorly. This pattern is evident in overall and parietal measurements and in their occipital profiles. At the same time, they share with late Middle Pleistocene and Late Pleistocene humans (both archaic and modern) the culmination of Middle Pleistocene trends in encephalization; Xuchang 1 has one of the highest Pleistocene endocranial volumes known, and Xuchang 2 should be in the middle of the Late Pleistocene variation based on its preserved measurements. They also share with these later Pleistocene humans a general gracilization of the cranium, which is reflected in small to absent nuchal and angular tori, modest cranial vault thicknesses, and reduced supraorbital torus thicknesses. With

22

respect to the last, they show in particular the reduction of the midorbital thickness relative to the lateral one.

In addition, they exhibit two features that closely approximate the derived configurations evident in the Neandertal lineage, best expressed in the Late Pleistocene ones: temporal labyrinthine proportions and occipital suprainiac configuration. The presence of these “Neandertal” features in the Xuchang crania should not be surprising, given the presence of a distinctly Neandertal temporal labyrinth in the late archaic Xujiayao 15 and “Neandertal” DNA in the early modern human Tianyuan 1 (6,18), as well as Neandertal aDNA and dental morphology in central Asia (17,67).

This mosaic of morphological configurations in the Xuchang crania, in the context of the current human fossil record, lends strong support to a complex model of regional population continuity and extensive interregional interactions through the later Middle and early Late Pleistocene of Eurasia, albeit in the context of similar trends in overall human biology. Given the dearth of relatively complete fossil human remains for much of central Eurasia, most of which can only be ascribed to “late archaic” or “early modern” generally, and the probable extensive movements of human populations in the context of Middle and Late Pleistocene climatic and ecozonal fluctuations, it is difficult to provide detailed scenarios for the extensive intra- and interregional interactions implied by the Xuchang and other later Pleistocene humans remains. The combinations of features present among them are nonetheless sufficient to indicate the unity of these human populations despite the evidence for morphological diversity at a continental scale.

23

Fig. S1. The Lingjing site. A: the Lingjing site, outlined in red, by satellite image; B: the Lingjing site before excavation in 2005, the black arrow showing the excavation area; C: excavation areas from 2005 to 2015; D: excavation work in 2009.

24

Fig. S2. Excavation at the Lingjing site. A: the surface of Layer 11 in the T14 excavation area and the overlying Pleistocene levels in profile; sticks mark the positions of mammalian fossils and stone artifacts during excavation; B: abundant lithic artifacts and animal bones found in situ; C: the Xuchang 1 right temporal bone (7019) discovered at Layer 11 in the T9 excavation area.

25

Fig. S3. Schematic stratigraphy of the Lingjing site and the OSL sample positions taken in situ.

26

Fig. S4. OSL signal of repeated regenerative dose of 245 Gy versus the signal measured following a test dose of 19.6 Gy using the SAR protocol for Sample L1199.

27

Fig. S5. Equivalent dose, De, as a function of preheat temperature. The solid line denotes the mean of the De values between 230°C and 280°CC for sample L1199.

28

Fig. S6. SAR and SC-MAR dose response curves of sensitivity-corrected OSL for Sample L1199. The dose response curves were calculated using the initial 0.64 s of the OSL decay curves. De: equivalent dose which corresponds to the intercept on the dose axis.

29

Fig. S7. IR-OSL decay curves for medium-grained K-feldspar of Sample L1199 by the pIRIR225 °C protocol. The dotted line: IR-OSL decay curve for the stimulation at 50°C; the solid line: IR-OSL decay curve for the stimulation at 225 °C after depleting the IR50°C signal.

30

Fig. S8. Dose response curve of sensitivity-corrected pIRIR225 °C signal for medium-grained K-feldspar (L1199) by the pIRIR225 °C protocol.

31

Fig. S9. Anomalous fading data for medium-grained K-feldspar pIRIR225 °C signals of samples L1434, L1436, L1271, L1274, L1199, L1577, L1202 and L1578. Given dose: 265 Gy, test dose: 21.2 Gy. t*: time interval between half-time after irradiation and the next stimulation time, normalized to a measurement delay time (tc) of 2 days (46). Ix= LpIRIR225 °C/TpIRIR225 °C, and the value was normalized to the first point. The g values are given in Table S8.

0.01 0.1 10.90

0.95

1.00

L1434

0.01 0.1 10.90

0.95

1.00

L1436

0.01 0.1 10.90

0.95

1.00

L1271

I x /

I 1

0.01 0.1 10.90

0.95

1.00

L1274

0.01 0.1 10.90

0.95

1.00

L1199

0.01 0.1 10.90

0.95

1.00

L1577

0.01 0.1 10.90

0.95

1.00

L1202

log (t*/tc)0.01 0.1 1

0.90

0.95

1.00

L1578

32

Fig. S10. The depth-OSL age plot. Closed squares: medium-grained K-feldspar corrected ages with the pIRIR225 °C protocol; closed circles: medium-grained K-feldspar ages with the pIRIR290 °C protocol. The OSL ages are presented with a one sigma error range.

33

Fig. S11. The Xuchang 1 human fossil fragments (A) and 3D virtual reconstruction along sutures, fractures and anatomical continuity. B: superoanterior view; C: basal left view; D: posterior right view; E: posterior left view. Red arrows show preserved sutures; blue arrows show the connected temporal line.

34

Fig. S12. The Xuchang 2 human fossil fragments (A) and 3D virtual reconstruction. B: posterior view; C: occipital internal view; D: posterior right view; E: left view. Red arrows show preserved sutures; blue arrows show the sigmoid sinus in the right temporal bone.

35

Fig. S13. Reconstruction of Xuchang 1 endocast. (A) superior view; (B) right lateral view; (C) posterior view. (D) left lateral view.

36

Fig. S14. Endocranial volume (ECV) values for Xuchang 1 (XUC1) and the comparative samples. Both the more likely Xuchang 1 estimate of ≈1,800 cc and the smaller one of ≈1,700 cc are provided.

37

Fig. S15. Maximum cranial breadth (XCB) for the Xuchang 1 and 2 crania (XUC1, XUC2) and the comparative samples.

38

Fig. S16. Inferior cranial breadths, bi-auriculare (A) and bi-asterionic (B), for the Xuchang 1 and 2 (XUC1, XUC2) crania and the comparative samples.

39

Fig. S17. Posterior views of the Xuchang 1 (left) and 2 (right) crania (reconstructed casts).

40

Fig. S18. A: Cranial vault height, from auriculare to vertex, for Xuchang 1 (XUC1) and the comparative samples, and B: index of cranial vault height to maximum cranial breadth.

41

Fig. S19. Principal components analysis (PCA) comparisons of the Xuchang 1 (XUC1) neurocranial proportions to those of the comparative samples preserving sufficient measurements. A: PCA using seven linear measurements (XCB, ASB, AUB, AVH, PAC, PASC, BRAC – see Table S9 for measurement codes). B: PCA using five indices (AVH/XCB, BRAC/BRAA, PAC/PAA, LASC/LASA, LINC/LINA). For the symbol key: regional color codes are for mid latitude eastern Eurasia (MLEE), low latitude eastern Eurasia (LLEE), Africa plus southwest Asia MIS 5 modern humans (A/SWA), and western Eurasia (WE); temporal samples are earlier Middle Pleistocene (EMPl), later Middle Pleistocene (LMPl), earlier Late Pleistocene (MIS 5-3b) (ELPl), and later Late Pleistocene (MIS 3a-2) (LLPl).

42

Fig. S20. Supraorbital torus thickness at midorbit (A) for Xuchang 1 and 4 (XUC1, XUC4) and at the lateral orbit (B) for Xuchang 1 versus comparative samples (no recent human data are included).

43

Fig. S21. The index of mid-orbit to lateral orbit supraorbital torus thicknesses for Xuchang 1 (XUC1) and Middle to Late Pleistocene humans.

44

Fig. S22. Sagittal (bregma-lambda) (A) and oblique (bregma-asterion) (B) parietal chords for Xuchang 1 the comparative samples.

45

Fig. S23.Sagittal (bregma-lambda) (A) and oblique (bregma-asterion) (B) parietal curvature indices (chord/arc) for Xuchang 1 (XUC1) and the comparative samples. Higher indices reflect flatter parietal arcs.

46

Fig. S24. Parietal thickness at the parietal eminence (A) and occipital torus maximum thickness (B) for Xuchang 1 and 2 (XUC1, XUC2), relative to the comparative samples.

47

Fig. S25. The index of the lambda-inion chord (reflecting occipital squamous height) versus bi-asterionic breadth (reflecting cranial base breadth) for Xuchang 1 (XUC1) and the comparative samples.

48

Fig. S26. A: Posterior view of the Xuchang 2 midline and right superior nuchal line and adjacent occipital bone (pieces 14022 and 14023). The arrow indicates the suprainiac fossa. B: μCT image of the inferior half of the central occipital piece (14022) with the suprainiac fossa indicated. C: Transverse μCT slice through the occipital bone at the level of the suprainiac fossa, with the fossa indicated. Note the absence of external table thinning through the level of the fossa.

49

Fig. S27. Lateral views of the Xuchang 1 (A) and 2 (B) right temporal bones (reconstructed casts). Both external auditory meati possess auditory exostoses, especially that of Xuchang 2.

50

Fig. S28. The relative height of the temporal squamous portion, as the index of the maximum height from porion to the length from krotaphion to asterion, for Xuchang 1 (XUC1) and the comparative samples.

51

Fig. S29. The Xuchang petrous bones (left) and the μCT extracted temporal labyrinths of the Xuchang 1 right bone and the Xuchang 2 right and left bones (right).

52

Fig. S30. Scatter plots of the index of the anterior to lateral semicircular canal radii versus the sagittal labyrinthine index (SLI) (A) and the first two principal components (B) for the Xuchang 1 and 2 (XUC1, XUC2) temporal labyrinths and those of Middle and Late Pleistocene plus recent humans. The PCA includes the SLI, canal radii, and canal radial percents. the east Asian Middle Pleistocene Hexian 1 (HEX) and early Late Pleistocene Xujiayao 15 (XJY) are indicated. The “West Mid Pleist” sample includes European earlier and later Middle Pleistocene remains, primarily from Atapuerca-SH. The “MIS 5 Modern” sample includes Middle Paleolithic modern humans, and the “MIS 3a-2 Modern” sample includes pan-Old World Upper Paleolithic sensu lato remains. Comparative data from Wu et al. (6), Quam et al. (91), and references therein.

53

Table S1. Inventory of the Xuchang (XUC) cranial remains. Original No. Discovery

date Preservation

XUC 1 Frontal bone

7013, 7012, 7006 2007 Middle-lateral of left supraorbital torus joining with posterior frontal squama

7025 2007 A small piece preserving the base of frontal sinus

703X 2007 A small piece of left frontal squama near bregma

14033 2014 A large part of left squama 14017 2014 Lateral part of right supraorbital torus Parietal

bones 7001, 7024 2007 Almost complete right parietal bone

7002, 7003, 7007, 7010, 7011, 7017

2007 Most of left parietal bone

Temporal bones

7019, 7021, 7027 2007 Right temporal squama and temporal petrous bone

7016 2007 Mastoid and low squama part of left temporal bone,

Occipital bone

14001 2014 Upper and middle part of occipital squamous portion

7005, 7020, 7014, 7030 2007 Large part of right occipital bone 7022, 7015 2007 Large part of left occipital bone XUC 2 Parietal

bone 14002/14019 2014 Mastoid corner joining with upper and lateral

part, and sphenoidal corner of right parietal bone

14011 2014 Occipital corner of left parietal bone Temporal

bones 14004, 14012, 14013, 14015, 14027

2014 Right temporal bone with partial squamous portion and missing mastoid process, inferior tympanic

14005, 14010, 14014, 14028

2014 Left temporal bone missing mastoid process, tympanic part and most of the squamous portion

Occipital bone

14006, 14023, 14021, 14022

2014 Occipital bone with partial nuchal plane, largely complete right and partial left occipital plane.

XUC 31

Parietal bone

14007 2014 A thick posteroinferior left parietal bone section

XUC 42

Frontal bone

14029 2014 Medial part of left supraorbital torus

XUC 51

Parietal bone

14030 2014 Partial bilateral middle parietal bones

1 XUC 3 and 5 do not derive from XUC 1 or 2, based on anatomical overlap and cross-sectional morphology. 2 XUC 4 may derive from XUC 2, but the absence of other portions of the anterior neurocranium make confirmation

of that association not possible.

54

Table S2 Information and the radioactive element data for the OSL samples at Lingjing site

Sample code Lab No. Depth (m) U (ppm) Th (ppm) K (%) Water

content (%)a

LJ08 OSL17 L1434 6.35 2.21±0.10 11.90±0.35 2.12±0.04 25±5

LJ08 OSL19 L1436 7.30 2.28±0.10 11.70±0.34 2.13±0.04 25±5

LJ08 OSL1 L1271 8.10 2.18±0.08 11.10±0.32 2.00±0.04 30±5

LJ08 OSL4 L1274 8.80 2.72±0.09 11.00±0.31 1.88±0.04 35±5

07block_up L1199 9.00 1.94±0.08 10.90±0.30 2.14±0.07 35±5

LJ08 OSL5 L1577 9.00 1.94±0.08 10.90±0.30 2.14±0.07 35±5

07block_down L1202 9.25 1.99±0.08 10.90±0.33 2.11±0.04 35±5

LJ08 OSL6 L1578 9.55 2.08±0.09 12.30±0.36 2.22±0.07 35±5

a Water content of the samples was assumed to represent an average condition during burial for each sample as

explained in SI 1.1 and SI 2.2.

55

Table S3 SAR and SC-MAR protocols Step Treatment

1 Give dose, Dia 2 Preheating at 260°C for 10 s 3 Blue-light stimulation at 130°C for 40 s 4 Give test dose, Dt (19.6 Gy) 5 Cut heat at 220°C for 0 s 6 Blue-light stimulation at 130°C for 40 s 7b Return to 1

a: For the natural sample, i=0.

b Step 7 is omitted for SC-MAR. For the multiple-aliquot, the average OSL from three aliquots were applied in each

regenerative dose. The natural signal is first measured for the set of aliquots before various regenerative doses administered on different aliquots.

56

Table S4 K-feldspar pIRIR225 °C protocol

Step Treatment Observed 1 Give dose, Dia 2 Preheating at 250°C for 60 s 3 Infrared stimulation at 50 °C for 100 s LpIRIR50 °C 4 Infrared stimulation at 225 °C for 100 s LpIRIR225 °C 5 Give test dose (21.2 Gy) 6 Preheating at 250°C for 60 s 7 Infrared stimulation at 50 °C for 100 s TpIRIR50 °C 8 Infrared stimulation at 225 °C for 100 s TpIRIR225 °C 9 Infrared stimulation at 290 °C for 40 s 10 Return to 1

a For the natural sample, i=0.

57

Table S5 K-feldspar pIRIR290 °C protocol

Step Treatment Observed 1 Give dose, Dia 2 Preheating at 320°C for 60 s 3 Infrared stimulation at 50 °C for 200 s 4 Infrared stimulation at 290 °C for 200 s LpIRIR290 °C 5 Give test dose (71 Gy) 6 Preheating at 320°C for 60 s 7 Infrared stimulation at 50 °C for 200 s 8 Infrared stimulation at 290 °C for 200 s TpIRIR290 °C 9 Return to 1

a For the natural sample, i=0.

58

Table S6 OSL dating results for quartz of different sizes for Sample L1199 with the SAR and SC-MAR protocols

Protocol Grain size (μm)

D0 (Gy)

De (Gy)

Disc No.

Environmental dose rate (Gy ka-1) Age (ka) Alpha Beta Gamma

Cosmic rays

Total

SAR 90-125 98±5 253±13 13 — 1.45±0.08 0.90±0.04 0.08±0.01 2.42±0.09 104±7 SAR 45-63 109±3 244±6 21 0.11±0.04 1.51±0.08 0.90±0.04 0.08±0.01 2.59±0.10 94±4 SAR 4-11 153±7 234±3 10

0.30±0.11 1.56±0.08 0.90±0.04 0.08±0.01 2.83±0.14 83±4

SC-MAR 4-11 129±8 222±4 — 78±4

59

Table S7. SAR OSL dating results for the medium-grained quartz of Samp