The Zuojiang River, originating in the mountainous area of northern Vietnam, flows some 450 km through the southwest part of Guangxi Zhuang Autonomous Region (GZAR), southern China. The Zuojiang River valley is characterized by a subtropical karst landform developed on a Carboniferous to Triassic limestone. This valley is renowned for the numerous red paintings preserved on steep cliffs along the Zuojiang River and its tributaries, namely the Lijiang, Mingjiang, and Heishui rivers (Fig. 1). Joint investigations of the rock paintings organized during the 1950s and 1980s by Chinese researchers, including geologists, archaeologists, historians, ethnologists, and art scientists, documented approximately 80 rock art sites stretching for more than 200 km (Qin et al., Reference Qin, Qin, Lu and Yu1987). The paintings are all drawn with ocher in a very similar style and range from 10 to 130 m above the active river surface. The motifs include mostly human figures, animals (e.g., horses, dogs, and birds), objects (e.g., boats, bells, drums, knives, and swords), and some designs probably representing the sun. The rock art at 38 sites, along with the surrounding rivers, terraces, and karstic mountains, forms the unique “Zuojiang Huashan Rock Art Cultural Landscape”, which was placed on UNESCO’s World Heritage List in 2016 (Fig. 1). Among these rock art sites, Mt. Huashan is the most spectacular, noted for the large scale and high density of motifs and excellent preservation (Huashan is called pay laiz in the local language, which literally means “a mountain with colorful paintings”).
The rock paintings in the Zuojiang River valley are widely believed to have been created between the Warring States period and the Eastern Han dynasty (475 BC to AD 220) on the basis of the comparison of the objects depicted (such as bronze drums, goat horn–shaped bells, and ring-head knives) with the archaeological findings from ancient tombs in southern and southwestern China (Qin et al., Reference Qin, Qin, Lu and Yu1987). The paintings are commonly attributed to the Luo Yue people, who are believed to be the ancestors of the Zhuang nationality and who inhabited this valley and the neighboring areas across present China and Vietnam from the Shang dynasty to the Han dynasty (1600 BC to AD 220) according to ancient documents (Qin et al., Reference Qin, Fu and Qin2015). These ancient paintings provide a precious record of the social life and religious beliefs of the Luo Yue people.
However, to determine a precise age on rock art is usually difficult because of the lack of stratigraphic context and datable material. When the pigment contains organic material, accelerator mass spectrometry (AMS) 14C dating can be used to date the paintings directly with less than 1 mg of carbon for analysis (e.g., Russ et al., Reference Russ, Hyman, Shafer and Rowe1990; Valladas et al., Reference Valladas, Cachier and Arnold1990, Reference Valladas, Cachier, Maurice, Bernaldo de Quirost, Clottes, Cabrera Valdés, Uzquiano and Arnold1992, Reference Valladas, Clottes, Geneste, Garcia, Arnold, Cachier and Tisnerat-Laborde2001; Watchman, Reference Watchman1993; Steelman et al., Reference Steelman, Rowe, Shirokov and Southon2002; Cole and Watchman, Reference Cole and Watchman2005; Smith et al., Reference Smith, Watchman and Ross2009; Morwood et al., Reference Morwood, Walsh and Watchman2010; David et al., Reference David, Barker, Petchey, Delannoy, Geneste, Rowe, Eccleston, Lamb and Whear2013; Quiles et al., Reference Quiles, Valladas, Bocherens, Delque-Kolic, Kaltnecker, van der Plicht and Delannoy2016). Alternatively, thin carbonate layers overlying or underlying the artwork can be used to establish the minimum or maximum age for the rock art by the 230Th/U method using thermal ionization mass spectrometry or multiple-collector inductively coupled plasma mass spectrometry (MC-ICP-MS) techniques, which require a very small amount of sample for analysis (10 to ~100 mg) (e.g., Pike et al., Reference Pike, Gilmour, Pettitt, Jacobi, Ripoll, Bahn and Munoz2005, Reference Pike, Hoffmann, Garcia-Diez, Pettitt, Alcolea, De Balbin and Gonzalez-Sainz2012; Aubert et al., Reference Aubert, O’Connor, McCulloch, Mortimer, Watchman and Richer-LaFleche2007, Reference Aubert, Brumm, Ramli, Sutikna, Saptomo, Hakim, Morwood, van den Bergh, Kinsley and Dosseto2014; Garcia-Diez et al., Reference Garcia-Diez, Hoffmann, Zilhao, de las Heras, Lasheras, Montes and Pike2013; Hellstrom and Pickering, Reference Hellstrom and Pickering2015). Nevertheless, secondary carbonate precipitates on the surface of bedrock, subjected to runoff over time, may experience postdepositional U migration (i.e., open system behavior). The coupled 230Th/U and AMS 14C approach allows the presumption of a closed system for the carbonates to be tested and thus is increasingly used in rock art dating studies (e.g., Plagnes et al., Reference Plagnes, Causse, Fontugne, Valladas, Chazine and Fage2003; Taçon et al., Reference Taçon, Aubert, Gang, Decong, Hong, May, Fallon, Ji, Curnoe and Herries2012; Fontugne et al., Reference Fontugne, Shao, Frank, Thil, Guidon and Boeda2013; Pons-Branchu et al., Reference Pons-Branchu, Bourrillon, Conkey, Fontugne, Fritz, Gárate and Quiles2014a; Corchón et al., Reference Corchón, Garate, Rivero, Valladas, Pons-Branchu, Murelaga, Ortega and Vicente2015).
The Zuojiang rock paintings were made exclusively with red ocher without any organic pigment that could be used for AMS 14C dating. Fortunately, many paintings at the Mt. Huashan site are overlain by carbonate precipitates. They provide a great opportunity to date the rock art by the 230Th/U dating method. Yuan et al. (Reference Yuan, Chen and Hu1986) attempted to date these secondary carbonates using 14C with a liquid scintillation counting technique, and they constrained the rock art age to between 2370 and 2115 yr BP. Because of the limited number of samples and uncertainties in 14C age corrections, however, the authors pointed out that the real age range could extend from the Spring and Autumn period to the Eastern Han dynasty (770 BC to AD 220).
To determine a precise age on the rock paintings at the Mt. Huashan site, the present study carried out high-precision 230Th/U dating of 56 carbonate samples associated with the paintings. The AMS 14C method was applied on 9 of the samples dated by 230Th/U to test their respective reliability and consistency. The high-precision 230Th/U dating results are discussed in light of X-ray diffraction (XRD), stable isotope (δ18O and δ13C), and AMS 14C analyses to assess the reliability of the chronology.
MATERIALS AND METHODS
Rock paintings at Mt. Huashan
The Zuojiang River valley is on the fringes of the Yunnan-Guizhou Plateau to the northwest. This is a karstic area surrounded by isolated limestone hills, karst peaks, vast plains, and a complex of drainage systems. The hills are generally less than 500m above sea level (m asl), and the plains vary from 80 to 200 m asl with multiple terraces (Qin et al., Reference Qin, Qin, Lu and Yu1987). The Zuojiang River valley has a subtropical monsoon climate with an annual average temperature of ~22°C and rainfall of ~1250 mm.
The Mt. Huashan site (22°15.70′N, 107°00.61′E) is located on the bank of the Mingjing River about 20 km northwest of Ningming County, GZAR (Fig. 2). The painted cliff at Mt. Huashan is approximately 350 m long and 270 m high and is oriented north to south, leaning slightly forward over the Mingjing River. The painted panel is about 170 m wide and between 30 and 90 m above the river surface, covering an area of approximately 8000 m2. Although it has suffered from severe erosion, more than 1900 images are identifiable (Qin et al., Reference Qin, Qin, Lu and Yu1987). Human figures account for approximately 75% of the images; they are commonly 30 to 300 cm high, depicted in frontal or side view, in a semisquatting posture with the arms raised from the elbow. The typical grouping of the human figures is one large frontal human figure in the center surrounded by several smaller human figures. The large frontal human figure wears decorated headgear, carries weapons at the waist (such as a knife or sword), and has an animal at his feet (such as a dog), whereas the smaller human figures do not have decorations or weapons.
This rock art site was registered among the important units of cultural relics under national protection by the state council in 1988, was placed on the tentative World Heritage List of China by the state administration of cultural heritage in 2007, and became a World Heritage Site in 2016.
Mt. Huashan is composed of Carboniferous limestone and covered by evergreen shrubs. Because of the moist monsoon climate and subtropical vegetation, dissolution and precipitation processes here are very active. As rainwater percolates down through the soil, it is enriched with CO2 forming a weak carbonic acid, owing to respiration and organic decomposition. The acidic water, with a high partial pressure of CO2, dissolves the bedrock along the rock fractures. At various places on the limestone cliff, the water discharges from the bedrock, which causes degassing of CO2 and precipitation of secondary carbonates. Refer to White (Reference White1988) and Ford and Williams (Reference Ford and Williams2007) for detailed descriptions of the karst processes.
Secondary carbonates for 230Th/U dating were collected between the years 2013 and 2014 and benefited from the handrails that were built for a rock art conservation project, which make it possible to access all of the painted area. Only secondary carbonates that have a clear stratigraphic relationship with the rock art were sampled, including 34 samples overlying the paintings and 22 samples underlying the paintings (Fig. 3, Supplementary Table 1, and Supplementary Fig. 1). In addition, 2 bedrock samples were collected. In order to minimize the damage to the paintings, all samples were taken in a diameter less than 2 cm.
Located on the surface of the bedrock, the paintings, like the overlying calcite deposits, have been exposed to intense weathering. Preliminary analyses were performed on the carbonate samples to be dated by 230Th/U to characterize their nature and formation process.
Mineral and stable isotope (δ18O, δ13C) analyses
Aliquots of the samples collected from the Mt. Huashan site were analyzed by XRD in order to evaluate the mineralogy and to assist in the interpretation of isotopic values. The XRD analysis was performed on powdered samples using a Rigaku D/max 2500VL/PC XRD. The diffraction patterns (2θ=6°−95°) were recorded at room temperature using Cu Kα1 radiation (λ=1.5406Å) at a scanning speed of 5°/min and a step size of 0.02° (2θ). Diffraction patterns were matched using the International Center for Diffraction Data (ICDD, 2500) database.
The samples analyzed by XRD were recovered and submitted for measurements of isotopic compositions of oxygen and carbon using a Thermo-Finnigan MAT-253 mass spectrometer fitted with an online automated preparation system (Kiel Carbonate Device III). Samples were calibrated against the international calcite standard NBS-19 and were reported in per mil relative to the Vienna Pee Dee belemnite (VPDB) standard. Repeated measurements of this standard display an analytical precision of typically 0.08‰ (1σ).
AMS 14C analysis
Nine samples (aliquots of carbonate grains prepared for 230Th/U dating) were used to the AMS 14C analysis: five were prepared and analyzed by the Laboratoire de Mesure du Carbone 14 (LMC14) laboratory, Gif-sur-Yvette Laboratoire des Science du Climat et de l’Environment (LSCE) and Accélérateur pour la Recherche en sciences de la Terre, Environnement, Muséologie Installé à Saclay (ARTEMIS) AMS national facilities, in Saclay, France, and four at the Beta Analytic radiocarbon dating laboratory in Miami, USA. At LSCE, the samples were finely crushed and washed in distilled water using an ultrasonic bath. The grains were then lightly etched using 0.01N HNO3 and rinsed with distilled water. Under vacuum, the carbonate was reacted with phosphoric acid, and the evolved CO2 was reduced to obtain graphite targets prepared following the method described by Tisnérat-Laborde et al. (Reference Tisnérat-Laborde, Poupeau, Tannau and Paterne2001). Results are expressed as conventional ages following Stuiver and Polach (Reference Stuiver and Polach1977). The sample treatment and AMS measurement protocols used in the Beta Analytic laboratory are available on their website (http://www.radiocarbon.com).
The secondary carbonate pieces were mechanically cleaned by a dental drill to remove the yellowish surface layer and attached pigment. They were then gently crushed into small grains less than 5 mm in size and ultrasonically cleaned in 18.2 MΩ ultrapure water. The pure and well-crystallized grains were handpicked for 230Th/U dating.
The samples (n=42) collected in 2014 were analyzed at Nanjing Normal University. The clean samples (50–200 mg) were weighed and dissolved in 7N HNO3 in a Teflon beaker containing a known quantity of a 229Th-233U-236U triple spike. After sample dissolution, HClO4 (~30 μL) was added to decompose organic material. The sample-spike mixture was heated on a hot plate overnight to equilibrate. U and Th were then separated from each other and from other cations by passing the sample solution through a U-TEVA resin column according to Douville et al. (Reference Douville, Salle, Frank, Eisele, Pons-Branchu and Ayrault2010). Sample matrix elements were first eliminated by rinsing with 3N HNO3. Subsequently, Th was eluted using 3N HCl, and finally, U was eluted using 1N HCl. HClO4 (~30 μL) was added again to the sample solutions to remove the organics dropped off from the U-TEVA resins. The U and Th fractions were then dried and diluted in a mixture of 0.5N HNO3 and 0.01N HF for U and Th isotopic analysis.
The U and Th isotopic measurements were carried out on a Neptune MC-ICP-MS. U was measured statically by 233U, 235U, 236U, and 238U on Faraday cups and 234U simultaneously on a secondary electron multiplier (SEM). Th was analyzed with 230Th and 229Th alternately on the SEM and 232Th on a Faraday cup. The U isotopic ratios of the HU-1 standard, which is a commonly used uraninite standard at secular radioactive equilibrium, were measured before and after every U and Th isotopic measurement for samples. Mass fractionation was corrected by comparing the measured 238U/235U to the value of 137.760 for the HU-1 standard and to the value of 137.818 for natural samples (Hiess et al., Reference Hiess, Condon, McLean and Noble2012) with an exponential law (Russell et al., Reference Russell, Papanastassiou and Tombrello1978). The mass fractionation for U and Th are assumed to be equivalent. The yield of SEM to Faraday cup was assessed by the δ234U measured in the HU-1 standard. Hydride interferences, machine abundance sensitivity, and amplifier gains were evaluated every day prior to sample measurements.
The samples (n=14) collected in 2013 were prepared at the LSCE laboratory and analyzed on a Neptune plus MC-ICP-MS fitted with a jet pump interface. The procedure used for U/Th chemical separation and isotopic measurement is detailed in Pons-Branchu et al. (Reference Pons-Branchu, Douville, Roy-Barman, Dumont, Branchu, Thil, Frank, Bordier and Borst2014b).
230Th/U ages were calculated using half-lives of 75,584 yr and 245,620 yr for 230Th and 234U, respectively (Cheng et al., Reference Cheng, Edwards, Shen, Polyak, Asmerom, Woodhead and Hellstrom2013). The 230Th/U age uncertainty was estimated by Monte-Carlo simulations with consideration of all sources of analytical uncertainty.
Mineralogy and stable isotopes
The results of XRD analysis on the secondary carbonates, bedrock, and pigment are presented in Table 1 and Figure 4. The XRD analysis on the secondary carbonates showed that 33 samples contain pure calcite, 2 are made of aragonite, and 21 are a mixture of aragonite with calcite and/or hydromagnesite [Mg5(CO3)4(OH)2(H2O)4]. XRD analysis on the 2 bedrock samples identified calcite and dolomite. The red pigment attached on the surface of sample NM-40 (Fig. 3) was carefully scraped off using a sharp blade. The XRD spectrum of this pigment shows the presence of hematite (Fe2O3), quartz, and calcite (Fig. 4). This analysis confirms the previous XRD analysis: the major color-generating element is Fe2O3, but calcite, quartz, and kaolin are also present, indicating that the pigment was derived from natural red ocher mixed with clay (Guo et al., Reference Guo, Han, Huang, Lan and Xie2010).
a The type XRD of Ara./hyd./cal./dol. is a mixture of aragonite, hydromagnesite, calcite, and/or dolomite. The analytical errors were given at the 2σ level.
b U/Th ages were calculated with the decay constants of λ230=9.1705×10−6/yr and λ234=2.82206×10−6/yr (Cheng et al., Reference Cheng, Edwards, Shen, Polyak, Asmerom, Woodhead and Hellstrom2013). The U/Th ages were corrected with assumption of the initial 230Th/232Th atomic ratio of 4.4±2.2×10−6, a value for a material at secular equilibrium, with the bulk earth 232Th/238U value of 3.8 and with an assumed error of 50%.
c The corrected ages are given at the “BP” scale, before AD 1950.
The results of oxygen and carbon isotopic analyses on these carbonates are presented in Table 1 and Figure 5. Overall, the data show a strong positive correlation between the δ18O and δ13C values, with correlation coefficients of 0.34 for the aragonitic samples, 0.66 for the calcitic samples, and 0.77 for all of the data (except for the HS-S3 sample). It can also be observed that the aragonitic samples are more enriched in both 18O and 13C compared with the calcitic ones. The δ18O values in the aragonitic samples range from −5.94‰ to −2.83‰ and have a mean of −4.76‰, whereas in the calcitic samples they vary from −9.37‰ to −4.48‰ with a mean of −6.66‰, except for the HS-S3 sample, which is extremely depleted in 18O with a δ18O value of −16.70‰. The distribution of δ13C values is in the range of −1.03‰ to +3.49‰ for the aragonitic samples and in the range of −9.71‰ to +2.02‰ for the calcitic samples. Oxygen and carbon isotopic analyses on the two bedrock samples yielded consistent results with a mean of −6.59‰ for δ18O and +2.68‰ for δ13C, which are in the typical δ18O and δ13C value ranges for Carboniferous marine limestone and dolomite (Veizer et al., Reference Veizer, Ala, Azmy, Bruckschen, Buhl, Bruhn and Carden1999).
These mineralogical and isotopic results show that the secondary carbonate layers associated with the paintings are present as different mineralogical phases and are likely to have different chronologies and/or chemical behavior.
230Th/U dating results
A summary of the MC-ICP-MS U/Th analytical results completed on the 56 samples is presented in Table 1. The aragonitic and calcitic samples are distinguishable in U concentration and U isotopic compositions. The average U concentration is 3.20 μg/g in the aragonitic samples and 1.94 μg/g in the calcitic ones. The measured δ234U values range from 145.5‰ to 461.9‰ with a mean of 277.1‰ in the aragonitic samples, and from −46.8‰ to 163.5‰ with a mean of 81.4‰ in the calcitic samples.
The average 232Th concentration is 8.6 pg/g in the aragonitic samples and 14.8 pg/g in the calcitic samples. The measured 232Th implies the presence of at least some 230Th when the carbonates were formed. The present-day 230Th/232Th activity ratios are as low as 1 in the young samples, whereas values as high as 250 were measured in the old ones. Correction for initial 230Th was made for each sample on the assumption of an initial 230Th/232Th atomic ratio of 4.4±2.2×10−6 (or activity ratio of 0.82±0.41), which is a value for a material assumed to be at secular equilibrium for 230Th, 234U, and 238U, with the bulk crustal 232Th/238U concentration ratio of 3.8±1.9.
The samples overlying the paintings have corrected ages in the range of 1728±41 to 14±64 yr BP (before AD 1950), and the samples underlying the paintings are between 1856±16 and 148,385±1883yr BP. Replicate U/Th analyses on 19 samples found slight differences in U concentration and isotopic compositions (Supplementary Table 2), but the 230Th/U ages are consistent (Fig. 6), indicating that these samples behaved as closed systems with respect to uranium.
AMS 14C results
Radiocarbon values are reported in Table 2 as conventional age (using Oxcal 4.2 software [Ramsey, Reference Ramsey2009]; ages reported at 95.4% confidence interval) after normalization to δ13C of 25‰; they were corrected for backgrounds (accelerator analysis and chemical treatment) and isotopic fractionation using the δ13C measured during the AMS analysis (Stuiver and Polach, Reference Stuiver and Polach1977). Ages are widely scattered, ranging from 1290±30 to 38,910±630 14C yr BP. Calibrated ages using the IntCal 13 (Reimer et al., Reference Reimer, Bard, Bayliss, Beck, Blackwell, Bronk Ramsey and Buck2013) are also reported, which represent the ages assuming no dead carbon within the samples. They range between 1180 cal yr BP (sample HS-S1) and ca. 44,000 cal yr BP (sample HS-S3), the latter being at the limit of the method. The 230Th/U ages allow calculations of the dead carbon percentage (DCP) values (Genty et al., Reference Genty, Massault, Gilmour, Baker, Verheyden and Keppens1999) for the paired U/Th and 14C analysis (Table 2). These DCP values are unusually high with a mean of 31.5% and values up to 57.4%, with generally higher values obtained for aragonitic samples.
The mechanism for sample deposition in open-air or well-ventilated sites such as CaCO3 layers on shelter or tufas could be different from speleothems in caves. Ihlenfeld et al. (Reference Ihlenfeld, Norman, Gagan, Drysdale, Maas and Webb2003) and Garnett et al. (Reference Garnett, Andrews, Preece and Dennis2004) found that tufa can be deposited in conditions close to isotopic equilibrium between CaCO3 and water for oxygen and carbon. In contrast, Lojen et al. (Reference Lojen, Trkov, Scancar, Vazquez-Navarro and Cukrov2009) observed that isotopic composition of tufas can be largely influenced by nonequilibrium precipitation, which favors the incorporation of isotopically depleted C and O into the carbonate.
In Mt. Huashan samples, isotopic equilibrium between CaCO3 and waters could not be verified. The δ18O and δ13C data are not used here for paleoclimatic studies, but in order to constrain the origin of calcitic samples (primary or neomorphic), as the mechanism for this transformation could be similar in different environments (including caves and open-air sites).
Data on both calcitic and aragonitic samples follow the same O/C trend (Fig. 5). The main difference in the O and C isotopic composition between the two minerals is that δ18O and δ13C values are generally higher within aragonitic samples than in calcitic ones (Fig. 7).
In a previous study on speleothems, Zhang et al. (Reference Zhang, Cai, Tan, Qin and An2014) highlighted that the aragonite/calcite transformation alters the δ18O signal (with lower δ18O values for secondary calcite), but not the δ13C. In another case, Woo and Choi (Reference Woo and Choi2006) found that the δ18O and δ13C values of neomorphic calcite are similar to the isotopic composition of primary aragonite. They also observed that unaltered aragonite speleothems are enriched in 13C with respect to the primary calcitic speleothems. They argued that this could indicate a stronger limestone imprint and/or a higher soil water component. A similar conclusion was reached by McMillan et al. (Reference McMillan, Fairchild, Frisia, Borsato and McDermott2005) during their study of two speleothems from Clamouse Cave (France): they reported higher δ18O and δ13C values in aragonitic layers than in primary calcite within the same speleothems.
Following these studies, the higher δ18O and δ13C values observed in Mt. Huashan aragonitic samples than in calcitic ones provide further evidence of the original calcitic samples, with a strong host rock imprint within aragonitic samples. At this site, the mechanism proposed for the deposition of aragonite involves strong water/host rock interaction with limestone dissolution in closed conditions (low exchange with the atmosphere) leading to high δ13C and DCP values. This mechanism is strengthened by the high δ234U composition in the aragonitic samples (Fig. 7), which is higher in dry periods or regions because of the longer interaction time between water and host rock compared with wetter periods or regions (e.g., Kaufman et al., Reference Kaufman, Wasserburg, Porcelli, Bar-Matthews, Ayalon and Halicz1998; Ayalon et al., Reference Ayalon, Bar-Matthews and Kaufman1999; Hellstrom and McCulloch, Reference Hellstrom and McCulloch2000; Robinson et al., Reference Robinson, Henderson, Hall and Matthews2004; Zhou et al., Reference Zhou, Lundstrom, Fouke, Panno, Hackley and Curry2005; Göktürk et al., Reference Göktürk, Fleitmann, Badertscher, Cheng, Edwards, Leuenberger, Fankhauser, Tüysük and Kramers2011; Maher et al., Reference Maher, Ibarra, Oster, Miller, Redwine, Reheis and Harden2014). In contrast, the low δ13C, δ18O, DCP, and δ234U values determined for calcitic samples indicate lower water/host rock interaction, congruent dissolution of the host rock (lower δ234U, closer to equilibrium), and higher CO2 exchange with the atmosphere (resulting in lower DCP). The development of both minerals on the same cliff could be because of the existence of two water pathways, one within rock fractures close to the cliff surface (relatively rapid water transfer in the case of calcite formation) and another within interstitial pore spaces (slower water transfer in the case of aragonite).
Reliability of the 230Th/U ages
Open system behavior for the 230Th/U chronometer could result in U gain or loss, but in the case of speleothems or thin calcitic and/or aragonitic layers, the most common phenomenon is U loss (e.g., Railsback et al., Reference Railsback, Dabous, Osmond and Fleisher2002; Plagnes et al., Reference Plagnes, Causse, Fontugne, Valladas, Chazine and Fage2003), leading to anomalously old 230Th/U ages (Pons-Branchu et al., Reference Pons-Branchu, Bourrillon, Conkey, Fontugne, Fritz, Gárate and Quiles2014a; Hellstrom and Pickering, Reference Hellstrom and Pickering2015). In addition, the initial 230Th introduced by detrital materials can also make apparent 230Th/U age older than the true age. For 14C dating, an open system behavior is expected to introduce young C producing apparent 14C ages that are too young (e.g., Holmgren et al., Reference Holmgren, Lauritzen and Possnert1994; Goslar et al., Reference Goslar, Hercman and Pazdur2000).
As aragonite is a metastable mineral, its transformation to calcite is commonly observed. This transformation within speleothems and secondary carbonates could result in U loss and thus a possible age bias toward an apparent age that is older than the real age (e.g., Ortega et al., Reference Ortega, Maire, Deve and Quinif2005; Lachniet et al., Reference Lachniet, Bernal, Asmerom and Polyak2012). Thus, the validity of the 230Th/U dating on the calcitic samples relies on their origin: primary calcite (i.e., not produced by aragonite transformation) can yield reliable 230Th/U ages, whereas neomorphic calcite (derived from aragonite transformation) could result in erroneous ages. McDermott et al. (Reference McDermott, Frisia, Huang, Longinelli, Spiro, Heaton and Hawkesworth1999) and McMillan et al. (Reference McMillan, Fairchild, Frisia, Borsato and McDermott2005) observed for two speleothems primary calcitic and aragonitic layers within the two stalagmites, deposited during a brief dry period. Aragonite and calcite can therefore be deposited in the same karstic context.
For the coupled 230Th/U and 14C analysis, excluding HS-S3, which is at the 14C dating limit, the 14C ages are systematically older than 230Th/U ages (Table 2). The difference between the two chronometers is most likely because of the presence of dead carbon (nonatmospheric carbon, originating from soil and bedrock), and not to an open system behavior or uncorrected detrital Th contamination, which would result in older 230Th/U ages. In the present study, the comparison of 230Th/U and 14C results validates the 230Th/U ages on the aragonitic samples and underlines an unusually large DCP within these samples that could be because of closed system calcareous dissolution and/or the presence of old organic matter (Noronha et al., Reference Noronha, Johnson, Hu, Ruan, Southon and Ferguson2014). This comparison also indicates that the 230Th/U ages of the calcitic samples may be reliable, and the DCP fraction deduced from 14C analysis is within the range of classical values for speleothems (3.5% to 13.5% for the three samples HS-S1, HS-S7, and NM-18, and 41.3% for NM-17).
The general convergence of the ages obtained by the two methods (14C and 230Th/U) suggests a chemically closed system, allowing the validation of ages. For Huashan carbonates, the 14C and 230Th/U ages do not converge; nevertheless, 14C activities suggest that the system has remained chemically closed.
The age of the rock paintings
So far, the antiquity of rock art in China has not been widely studied with absolute dating methods. Using 14C dating of carbonates, the age of Mt. Huashan rock paintings has been estimated to be between 2370 and 2115 cal yr BP (Yuan et al., Reference Yuan, Chen and Hu1986), and the paintings at Cangyuan, Yunnan Province, were dated to between 3100 and 2960 yr BP (Wang, Reference Wang1984; Bednarik and Li, Reference Bednarik and Li1991). However, these 14C ages are likely too old because of the DCP input. Recently, 230Th/U and 14C dating methods were applied to a flowstone formed above and below a large painted deer at Baiyunwan, Yunnan Province, and yielded a maximum age of 5740 cal yr BP and a minimum 230Th/U age of 2050 cal yr BP for the rock art (Taçon et al., Reference Taçon, Aubert, Gang, Decong, Hong, May, Fallon, Ji, Curnoe and Herries2012).
From the motifs of bronze drums, goat horn–shaped bells, and ring-head knives in the Mt. Huashan rock paintings, an indirect age between the Warring States period and the Eastern Han dynasty (475 BC to AD 220) can be inferred from comparison with the objects excavated from ancient tombs in southern and southwestern China (Qin et al., Reference Qin, Qin, Lu and Yu1987). In the present study, the 56 calcitic/aragonitic samples yielded 230Th/U ages ranging from 148,385±1883 cal yr BP to 14±64 yr BP. The age distribution shows that the number of analyzed samples generally increases toward recent times, and there is an apparent gap between 1600 yr BP and 920 yr BP (Fig. 8A). This time interval generally corresponds to a weakening monsoon event centered around 1600 yr BP in East Asia, according to a stalagmite δ18O record from the Dongge Cave (Fig. 8B; Wang et al., Reference Wang, Cheng, Edwards, He, Kong, An, Wu, Kelly, Dykoski and Li2005), approximately 350 km northwest of the Mt. Huashan site. It also corresponds to a distinct China-wide cold period between 1720 and 1150 yr BP, revealed by multiple paleoclimate proxy records including ice cores, tree rings, lake sediments, and historical documents (Yang et al., Reference Yang, Braeuning, Johnson and Shi2002). Beyond sampling bias, the absence of samples in this age gap probably is related to the relatively dry and cold climate, which may have suppressed the formation of secondary carbonates on the cliff. On the other hand, the large number of samples dated to the past 200 yr BP may correspond to the strengthening monsoon and the increasing temperature since roughly AD 1800 (Fig. 8B; Yang et al., Reference Yang, Braeuning, Johnson and Shi2002; Wang et al., Reference Wang, Cheng, Edwards, He, Kong, An, Wu, Kelly, Dykoski and Li2005).
The 230Th/U dating in combination with mineral, stable isotopic, and 14C analyses on these carbonate samples strongly indicate that the aragonitic samples were formed in a closed carbonate system and that the calcitic samples were formed in a relatively open condition. These analyses do not show evidence of open system behavior after carbonate precipitation. Therefore, the 230Th/U ages obtained are able to represent the age of the mineral formation. Accordingly, the maximum age of the rock paintings can be estimated to be 1856±16 yr BP or AD 94±16 (by NM-29, the youngest of the samples underlying the paintings), and the minimum age can be estimated to be 1728±41 cal yr BP or AD 222±41 (by NM-21, the oldest of the samples overlying the paintings), corresponding to the middle to the end of the Eastern Han dynasty (AD 25 to AD 220).
The XRD analyses show that the secondary carbonates formed on the painted cliff at the Mt. Huashan site can be separated generally into calcitic (>55%) and aragonitic (~40%) varieties. Isotopic (δ18O, δ13C, and 14C) and 230Th/U analyses on these carbonate samples did not exhibit clear evidence of postdepositional chemical alteration. Based on the 230Th/U ages, the age of the Mt. Huashan rock paintings can be bracketed between 1856±16 and 1728±41 yr BP, corresponding to the middle to the end of the Eastern Han dynasty (AD 25 to 220), and possibly even to the early part of the Three States period (AD 220 to 280), considering the uncertainties in 230Th/U ages. The results suggest that the rock painting practices lasted probably more than 1 century in the Mt. Huashan site, and that this rock art is younger than that at Baiyunwan and Cangyuan in Yunnan Province by 1 to 10 centuries according to the available numerical dating results. This study also shows that the indirect dating of rock art using secondary carbonate deposits over- or underlying paintings may produce a large range of ages, sometimes far away from the age of the painting itself. Because the delay between the carbonate formation and the rock art practice remains unknown, dating of a large number of samples, as performed in this study, may be necessary to constrain the timing of rock art by the U/Th dating method. This study does not exclude the possibility of older rock paintings existing at other sites in the Zuojiang River valley. Thus, further numerical dating research on the rock paintings in this valley is warranted.
This research was carried out in the framework of the application for World Heritage for the “Zuojiang Huashan Rock Art Cultural Landscape” organized by the Guangxi Zhuang Autonomous Region Department of Culture. We thank Lan Ri-Yong, Zhou Han-Cheng, and Hu Peng-Cheng for their help in the field with sampling. We thank Eric Douville and François Thil for their help on the development of U/Th dating techniques at NNU. We thank Jeff Pigati and two anonymous reviewers, who provided insightful comments/suggestions to improve this paper. This research was supported partly by the National Natural Science Foundation of China (grant 41302136) and the China Postdoctoral Science Foundation (grant 2013M530263).
To view supplementary material for this article, please visit https://doi.org/10.1017/qua.2017.24