Magnetostratigraphic dating of the hominin occupation of Bailong Cave, central China

Intermontane basins in the southern piedmont of the Qinling Mountains are important sources of information on hominin occupation and settlement, and provide an excellent opportunity to study early human evolution and behavioral adaptation. Here, we present the results of a detailed magnetostratigraphic investigation of the sedimentary sequence of hominin-bearing Bailong Cave in Yunxi Basin, central China. Correlation to the geomagnetic polarity time scale was achieved using previously published biostratigraphy, 26Al/10Be burial dating, and coupled electron spin resonance (ESR) and U-series dating. The Bailong Cave hominin-bearing layer is dated to the early Brunhes Chron, close to the Matuyama-Brunhes geomagnetic reversal at 0.78 Ma. Our findings, coupled with other records, indicate the flourishing of early humans in mainland East Asia during the Mid-Pleistocene climate transition (MPT). This suggests that early humans were adapted to diverse and variable environments over a broad latitudinal range during the MPT, from temperate northern China to subtropical southern China.

magnetite, which indicates that magnetite is the major contributor to the susceptibility. For some samples, there is a magnetic susceptibility decrease between ~300 °C and ~450 °C in the heating curves ( Fig. 2b-d), which is due to the conversion of metastable maghemite to hematite 30 . Two types of χ-T curves are evident. One type has cooling curves that are much higher than the heating curves, with susceptibility increasing significantly on cooling below ~585 °C (Fig. 2c-f). The significantly enhanced susceptibility after thermal treatment may arise from the neo-formation of magnetite grains from iron-containing silicates/clays, or from the formation of magnetite by reduction in the presence of combusting organic matter 29,31 . The other type has slightly enhanced susceptibility when cooling to room temperature (Fig. 2a,b).
These χ-T curves provide further evidence that magnetite and maghemite are the dominant ferrimagnetic minerals in the Bailong Cave deposits. Hematite, which is another important carrier of the natural remanence suggested by isothermal remanent magnetization (IRM) acquisition (Figs 3 and 4), hysteresis loops (Fig. 5), and progressive thermal demagnetization analyses (Fig. 7), is not well expressed in the χ-T curves because its weak susceptibility is masked by the much stronger contributions of magnetite and/or maghemite. . All the selected samples have similar IRM curves (Fig. 3). The rapid increase in the IRM acquisition curves below 100 mT indicates the dominant presence of magnetically soft components, such as magnetite and maghemite. However, the IRM of all the samples continues to increase above 300 mT, and the S-ratio 33 , which is defined as the ratio of IRM acquired at −0.3 T (IRM −0.3T ) to IRM acquired at 1 T (IRM 1T ), has relatively low values (generally below 0.8) (Fig. 3). This behavior suggests a significant contribution from high-coercivity minerals (hematite) which have a weak magnetization. Unmixing methods 32 were used to analyze the magnetic mineral composition. Derivatives of the IRM acquisition curves are plotted to illustrate the coercivity distributions ( Fig. 4), where one-to two-humped distributions illustrate distinct coercivity distributions with peaks at 20-30 mT and ~100 mT. The lower coercivity component is likely to be magnetite and/or maghemite, and the higher coercivity component represents hematite.
Hysteresis loops 28 and first-order reversal curve (FORC) diagrams 34,35 provide information about the coercivity spectrum and domain state of ferrimagnetic materials. All the selected samples have wasp-waisted hysteresis loops (Fig. 5), which are attributed to the coexistence of two magnetic components with strongly contrasting coercivities 28 . The low-coercivity component consists of magnetite and/or maghemite, and the high-coercivity component is mainly due to hematite, as suggested by the χ-T curves (Fig. 2) and progressive thermal demagnetization analyses (Fig. 7). FORC diagrams were obtained to provide a more detailed interpretation of the domain state of magnetic mineral assemblages. All samples have FORC distributions that are indicative of stable single-domain (SD) particles (Fig. 6). The vertical spread along the B c axis is mostly ~20 mT. The FORC diagram for sample B80 (Fig. 6b) suggests a low degree of magnetostatic interactions, as indicated by a ridge-like distribution along the B c axis, which suggests the dominance of non-interacting SD particles. All FORC diagrams are centered on the B c axis at 10-20 mT, which is consistent with the dominance of magnetite.
Paleomagnetic measurements. The characteristic remanent magnetization (ChRM) was isolated after removal of one or two soft secondary magnetization components (Fig. 7). Principal component analysis (PCA) was performed on stepwise demagnetization data using the PaleoMag software 36 . The principal component direction was computed using a least-squares fitting technique 37 . Demagnetization results for representative specimens, as shown in orthogonal diagrams 38 , indicate that both magnetite and hematite dominate the remanence, because a high-stability ChRM component persists up to 690 °C (Fig. 7a-c) or up to 60 mT (Fig. 7d-f). After the combined thermal and alternating field (AF) demagnetization, or thermal demagnetization only, 15 out of 18 and 11 out of 12 specimens with maximum angular deviation (MAD) values <15° yielded reliable ChRM directions, respectively. The ChRM vector directions yielded virtual geomagnetic pole (VGP) latitudes that were used to define the magnetostratigraphic polarity succession for the Bailong Cave section. A single, normal polarity zone is recognized (Fig. 8). In addition, two specimens recorded negative VGP latitudes, labeled a1 and a2 in Fig. 8d. These two anomalous paleomagnetic directions could represent short-period geomagnetic variations; however, we exclude them as possible geomagnetic excursions because they are based on a single specimen only.

Chronology of the Bailong Cave sedimentary sequence and age estimation of hominin occupation.
We established the chronology of the Bailong Cave hominin-bearing sequence by combining the previously        Importantly, 26 Al/ 10 Be burial dating 39 of quartz samples from layers 4 and 6 in the lower part of the Bailong Cave sequence (Fig. 8) give a weighted mean burial age of 0.76 ± 0.06 Ma. Liu et al. 39 further concluded that cultural deposits at Bailong Cave site should be somewhat younger than the above date by considering possible biases introduced by the dating method, stratigraphic order, and the documented rapid sedimentation. Most recently, coupled ESR/U-series dating 40 was conducted on the fossil teeth of herbivores (Cervidae and Bovidae) from layers 1 and 2 in the upper part of the sequence, yielding a weighted mean age of 509 ± 16 ka for five fossil teeth from layer 2 (Fig. 8a). The 26 Al/ 10 Be burial and ESR/U-series ages, respectively obtained by Liu et al. 39 and Han et al. 40 , are within the Brunhes Chron, which provides stringent age control for the Bailong Cave sedimentary sequence. Given the robust chronological constraints from mammalian biochronology 26,43 , 26 Al/ 10 Be burial dating 39 and ESR/U-series dating 40 , the normal polarity magnetozone identified here in the Bailong Cave sequence must correlate with the early Brunhes Chron, which is close to the Early/Middle Pleistocene transition.

Geochronological implications. Bailong Cave is a Paleolithic hominin site in an intermontane basin along
the Hanjiang River in the southern piedmont of the Qinling Mountains. Available chronological data from a combination of detailed magnetostratigraphic analysis, optically stimulated luminescence dating, and pedostratigraphic correlation with well-dated loess-paleosol sequences of the central Chinese Loess Plateau indicate that hominins occupied the Hanjiang valley several times during the interval from 1.2-0.1 Ma [12][13][14][15][16][17] . Given the recognition of numerous Paleolithic sites on both the northern and southern sides of the Qinling Mountains, Sun et al. 16 proposed that the Hanjiang River valley was a probable hominin migration route through the Qinling Mountains between subtropical southern China and temperate northern China. Moreover, by ~1 Ma hominins (mostly Homo erectus) occupied a broad latitudinal range in North Africa, Europe, western Asia, and eastern Asia 3,44,45 , which indicates that early human populations had adapted to diverse climatic settings. We note especially that during the Mid-Pleistocene climate transition, which began at about 1.0-0.8 Ma and terminated at about 0.7-0.6 Ma 46,47 , early human populations had flourished and expanded in mainland East Asia, from the low latitudes of the Tropic of Cancer (e.g., the Bose Basin) to high northern latitudes (e.g., the Nihewan Basin) (Figs 1 and 9, Table 1).
From pre-to post-MPT, the dominant periodicity of high-latitude climate oscillations changed from 41 kyr to 100 kyr, leading to profound changes in the length and intensity of glacial-interglacial cycles [46][47][48][49] . The MPT was characterized by variable environments 50 , during which the increasing climate contrast between glacial and interglacial periods may have forced early humans to become increasingly resilient to glacial-interglacial cycling 51 .

Methods
Geological setting. Bailong Cave (32°59′40.0′′N, 110°31′33.6′′E, elevation 550 m) (Fig. 1) is situated on the northwestern margin of the Wudang Uplift in the Qinling Orogenic Belt. Mesoproterozoic metamorphic volcanic and sedimentary rocks, comprising the Wudangshan Group and Neoproterozoic carbonate and sedimentary rocks comprising the Yaolinghe, Doushantuo, and Dengying Formations, occupy a large part of the area. The Neoproterozoic carbonate rocks form a karst topography controlled by the regional hydrologic system. The Neogene Shaping Formation consists of conglomerate and conglomeratic mudstones, and Quaternary sedimentary deposits unconformably overlie Mesoproterozoic and Neoproterozoic strata 52 .
Bailong Cave developed in the Neoproterozoic carbonate and Neogene sedimentary rocks. The cave deposits are divided lithostratigraphically into 8 sedimentary layers (Fig. 8), which were described in detail by Wu et al. 26 and Dong 43 . Layer 2 is fossiliferous and mainly composed of brownish-red clay with occasional calcareous concretions and gravels (Fig. 8). Eight hominin teeth and associated mammalian fossils and stone artifacts were unearthed from this layer 26,27,53,54 . Archeological setting. Bailong Cave archeological site was discovered in 1976 and three systematic excavations were conducted subsequently in 1977, 1982, and 2007-2009. So far, 29 taxa of vertebrate mammals (Table 1) and 38 stone artifacts were unearthed, which were reported in detail by Wu et al. 26   which were assigned to Homo erectus 25,27 , were recovered from the Bailong Cave, including two found by farmers in 1976, four by excavation in 1977, one by excavation in 1982, and one by excavation in 2008 27 . The Bailong Cave lithic assemblage is essentially an Oldowan-like industry (i.e., Mode 1 core and flake technologies). Like other Oldowan-like industries in China, the Bailong Cave stone assemblage is characterized by a simple technological design, a low degree of standardization, and casually retouched flakes. Technologically, the Bailong Cave lithic assemblage includes 4 cores, 4 flakes, 10 retouched tools, and 20 chunks and debris fragments. The utilized stone raw material is primarily vein quartz, which can be obtained from local Precambrian outcrops. The principal flaking technique was simple direct hard hammer percussion, followed by bipolar percussion. The cores were moderately exploited, probably due either to the difficulties of flaking low-quality vein quartz, or to the short distance of these rocks to the hominin site 26 . Sampling. Due to possible disturbance, the uppermost 0.3 m of the cave sedimentary sequence was removed before sampling. A total of 18 oriented block samples were collected with a magnetic compass at 5-25 cm stratigraphic intervals. Cubic specimens with dimensions of 20 mm × 20 mm × 20 mm were obtained from those block samples in the laboratory.

Mineral magnetic measurements.
Το determine the magnetic mineralogy, a total of 6 representative samples were selected for mineral magnetic measurements, including χ-T curves, IRM acquisition curves, backfield IRM demagnetization curves, hysteresis loops, and FORC diagrams.
χ-T curves were obtained by continuous exposure of samples through temperature cycles from room temperature to 700 °C and back to room temperature with a ramping rate of 2 °C/min, using an AGICO MFK1-FA  Table 2). (f) ATNTS2012 56 . The shaded area represents the MPT, which began at about 1-0.8 Ma and terminated at about 0.7-0.6 Ma 46,47 .
SCIEntIfIC REPoRTS | (2018) 8:9699 | DOI:10.1038/s41598-018-28065-x equipped with CS-3 temperature control system. To minimize the possibility of oxidation, the samples were heated and cooled in an argon atmosphere. For each sample, we subtracted the contribution of the sample holder and thermocouple to the magnetic susceptibility.
Hysteresis loops, IRM acquisition, back-field demagnetization curves, and FORCs were measured with a Princeton Measurements Corporation MicroMag 3900 vibrating sample magnetometer (VSM) up to a maximum field of 1 T. FORC diagrams were calculated using the FORCinel software package 55 . Magnetic components were analyzed using the unmixing programs written by Egli 32 . Paleomagnetic measurements. To establish the magnetic polarity stratigraphy, all specimens were subjected to stepwise demagnetization. To confirm the paleomagnetic results, two sets of parallel specimens were measured on the Bailong Cave samples. First, all 18 specimens were subjected to combined thermal and AF demagnetization at a peak field up to 60 mT at 5-10 mT intervals after stepwise thermal demagnetization at 80 °C, 120 °C, and 150 °C, with a Magnetic Measurements thermal demagnetizer with a residual magnetic field less than 10 nT. Then, the second set of 12 parallel specimens was subjected to stepwise thermal demagnetization up to 690 °C (21 steps with 10-50 °C temperature increments). Both methods are capable of isolating the ChRM after removal of a soft secondary component of magnetization. The remanence measurements were made using a 2-G Enterprises Model 760-R cryogenic magnetometer installed in a magnetically shielded space with background field of <300 nT. Data availability. The datasets generated and/or analyzed during the current study are available from the corresponding author on request or from the Magnetics Information Consortium (MagIC) database (http:// earthref.org/MAGIC).