Technical Comments

Comment on “Permanent human occupation of the central Tibetan Plateau in the early Holocene”

See allHide authors and affiliations

Science  11 Aug 2017:
Vol. 357, Issue 6351, eaam9231
DOI: 10.1126/science.aam9231


Meyer et al. (Reports, 6 January 2017, p. 64) suggest a new chronology for permanent human occupation of Tibet based on their dating of the travertine and colluvium deposits that contain or are deposited near fossil human handprints and footprints. However, misinterpretations in both stratigraphic reconstruction and dating data may have caused the newly proposed age of these human imprints to have been seriously underestimated.

The Tibetan Plateau is the world’s highest plateau and a famously harsh environment. The timing of human adaptation to this environment and colonization of the region is of great interest to both academia and the public. We discovered 19 human handprints and footprints (imprints, hereafter) and a fossil hearth on travertine sheets at the Queshang (Chusang) hot spring site in central Tibet (1). The imprint-bearing and hearth samples were dated to around 20 thousand years before the present (B.P.) using optically stimulated luminescence (OSL) techniques. Meyer et al. (2) recently suggested a different age (between 7.4 and 12.67 thousand years) for these imprints based on an analysis of the reconstructed stratigraphic profile and travertine and colluvium dating results from three different methods. We believe that problematic stratigraphic reconstruction and misinterpretation of the data sets and dating methods might have resulted in some serious errors in the imprint age deduced.

A thorough understanding of the sedimentology and petrology of Chusang travertine is key to the development of an accurate chronology for travertine and the imbedded handprints and footprints (2). However, Meyer et al. (2) reconstructed the stratigraphic profile [figure 2 of (2)] by collecting samples from different facies of deposits in various locations in a sloped area with many fossil spring outlets (travertine cones), two active spring outlets, and mass movements (2, 3) [figures S1 and S2 of (2)]. The active and fossil springs precipitated travertine around the outlets that scattered on the slope surfaces during the different periods. The travertine layer with the imprints [figure 2, QS-T-2, of (2)] is stratigraphically separated from the other layers in this profile, which might be the precipitation from different spring outlets. The 14C dating samples [figure 2, P12, of (2)] were actually taken from the colluvium in the Bathing House [figures S1A and S3E of (2)], 120 m away from and more than 8 m higher than the position of the imprint-bearing travertine (QS-T-2), and not “directly below the imprints,” as indicated in figure 2 of (2). Both CS-T-2 (OSL dating sample, clastic travertine, 100 m away) and CS-T-3 are horizontally and vertically separated from the imprint-bearing layer (QS-T-2, porous layered travertine). Samples P2, P4, P10, and P11 from the lower slope are even farther from the imprints, at distances ranging from 150 to 400 m [figures 2 and S1 of (2) and figure. 1 of (3)]. There is scant evidence that the QS-T-2 position has a colluvium layer deposited directly below. Consequently, the actual sedimentological positions of the colluvium and travertine formation do not support a successive relationship of the reconstructed stratigraphic profile. This also indicates that the imprint-bearing layer (QS-T-2) is not stratigraphically attached with the travertine layers of CS-T-2, CS-T-3, and the upper layer of P12 [figure 2 of (2)], and the dated ages of these travertine and colluvium layers, of course, cannot be used to support the age of the imprint-bearing layer (QS-T-2).

The 230Th/U dating samples used in the study were porous and recrystallized travertine from an open environment, which may create problems for this method (4). Some authors of the research (2) acknowledged that “a precise 230Th/U chronology will probably not be achievable” for the imprint-bearing travertine because of these problems (3).

In the study (2), the impure carbonate samples were dissolved by 7 M HNO3, and the supernatant solutions were then used for 230Th/U dating [supplementary materials of (2)]. This procedure is problematic and can lead to unpredicted apparent ages (5, 6). In addition, most dating samples had very low 230Th/232Th activity ratios (~2) [table S1 of (2)], regardless of the use of supernatant, which could potentially have increased the apparent 230Th/232Th ratio. As such, these 230Th/U dating results do not appear to provide a robust timing of the formation of the travertine.

The only two samples with higher 230Th/232Th ratios (QS-T-2A-01/03 and P11) gave relatively reliable ages (~7 and 13 thousand years B.P., respectively). However, both samples were secondary calcites, and thus they provide only the minimum age of the host travertine. In this regard, the results of Meyer et al. are consistent with, rather than contradictory to, the earlier study that determined that the travertine formed around 20 thousand years B.P. (1).

The dated OSL age (2) from the travertine layer, which is nearly 100 m away and stratigraphically separated from the imprints, might have been substantially underestimated due to misinterpretation of the equivalent dose (De) data. The OSL age is derived from the De divided by the dose rate. The authors explained their De distribution as resulting from incomplete resetting of the OSL signals before deposition [figure S6C of (2)], based on the overdispersion (OD) value (37 ± 6%) when the central age model of Galbraith et al. (7) is applied. However, based on the radial plot [figure S6C of (2)], only two of 86 grains have statistically significant larger De values. If we ignore these two grains, the remaining De values (84 grains) appear to be randomly distributed around a central value. The OD value could be significantly reduced to a value expected for a fully bleached sample (~20% or even less). In fact, OD values of ~20% are not uncommon for well-bleached sediments (e.g., 711). Furthermore, the authors did not mention how the minimum age model (MAM) is implemented. One of the most crucial requirements of an MAM is estimation of the sigma value (equivalent to the OD value expected for a well-bleached sample), which was totally ignored by the authors in the description of their MAM. The results of the MAM could be significantly influenced by the choice of this value. The OD value of the dose recovery data set can provide a minimum estimate of the sigma value used for the MAM, although it does not guarantee a reliable estimate (12). Unfortunately, the authors do not provide enough information on the OD value for the dose recovery data and the De after ignoring the two data points. If a central age model is applied to the sample, the De value is ~12 grays. The OSL age of the sample would be in the range of 15.6 to 16.2 thousand years for maximum and 20.2 to 21.0 thousand years for minimum mean dose rate scenarios. Rather than being contradictory, the results are again consistent with the OSL ages of Zhang and Li (1).

Incomplete bleaching was not a problem in the OSL dating of the hearth material from Zhang and Li’s study (1), because the OSL signal was completely reset when it was heated. The consistent ages from the hearth and travertine (heated and unheated) materials indicate that the quartz grains in the travertine were completely bleached by sunlight before deposition (1). A central/mean age model is suitable for the samples (13). The OD could be a result of microdosimetry differences among the quartz grains (14), particularly given that the mean dose rate is dominated by the contribution from cosmic rays. Any subtle difference in surrounding grains could lead to significant differences in the dose rates of individual grains, giving rise to the dispersion of the De value, even if they are perfectly zeroed in deposition.

References and Notes

  1. Acknowledgments: Most of the comments on 230Th/U dating were originally made by H. Cheng at the Institute of Global Environmental Change, Xi’an Jiaotong University. We thank him for permission to quote his idea. This study was financially supported by grants to S.-H.L. from the Research Grant Council of the Hong Kong Special Administrative Region, China (7033/12P and 17303014).
View Abstract

Navigate This Article