New chronology of the Chinese loess-paleosol sequence by leaf wax δD records during the past 800 k.y.

Loess-paleosol sequences are widely distributed in Europe, North America, and Asia, of which the Chinese Loess Plateau (CLP) is the most extensive loess deposition region (Liu et al., 1991; An, 2014). Loess deposits, as the longest continuous continental records, along with marine sediment, stalagmite, and ice records, play key roles for better understanding global paleoenvironmental changes from millennial to orbital time scales. Based on the correspondence between loess-paleosol stratigraphic alternations and glacial-interglacial cycles, it is known that loess formed during glacial periods and paleosol formed during interglacial periods, and so orbital parameter chronologies are commonly used to tune time series of loess-paleosol sequences (Kukla et al., 1988; Ding et al., 1994; Lu et al., 1998; Zhou et al., 2014). For example, magnetic susceptibility and grain-size records, which are regarded as indexes of East Asian summer and winter monsoon intensity, have provided reasonable time controls for loess-based climate and environmental studies (Yang and Ding, 2003; Guo et al., 2009; Sun et al., 2012). However, dissolution of magnetic minerals and translocation of loess particle sizes, especially during interglacial intervals, could seriously alter the original depths of these proxies, which challenges the integrity of these climate time series (Liu et al., 2014; Torrent, 2005). Consequently, the lack of independent and accurate time scales hinders the comparison between loess and other global records in better understanding the details of paleoclimate and environmental changes and associated mechanisms.

Asian stalagmite δ¹⁸O values record rainfall δ¹⁸O variations over the past 640 k.y. with highly precise, independent, and radiometric chronology (Wang et al., 2008; Cheng et al., 2012; Cai et al., 2015), which demonstrates a tight correlation with the summer insolation over large regions of the Northern Hemisphere (Cheng et al., 2016a). Moreover, stalagmite δ¹⁸O demonstrates that the rainfall δ¹⁸O varied consistently in different regions throughout glacial-interglacial cycles, defined as a “supraregional mode” (Cheng et al., 2016b). The loose soil structure inhibits pervasive forest cover over a large area of the CLP, where steppe and grasslands are common in the “Yuan” area (Li et al., 2003; Jiang et al., 2013; Sun et al., 2017). Loess hydrogen isotopic composition of plant wax (δD_wax) also inherits the δD signal of rainfall (Sessions et al., 1999; Sachse et al., 2012). Although δD_wax shows different fractionation between grasses and woody plants (Liu and Yang, 2008), the grassland-dominated plants of the CLP ensure that loess δD_wax is not significantly affected by vegetation changes. Thus, the loess δD_wax values mainly record the changes in rainfall isotope values, assuming that vegetation changes are not significant in order to change the fractionation in the δD_wax. Asian stalagmite δ¹⁸O records as an indicator of rainwater δ¹⁸O, and loess δD_wax records as an indicator of rainwater δD, should have a shared common origin of paleoprecipitation isotopes over monsoonal climate–affected regions, and over the CLP as well. Both δ¹⁸O variance and δD variance of rainwater events have a linear relationship when rain belts migrate with fractionation (Craig, 1961), which provides the basis for correlating chronologies of stalagmite δ¹⁸O and summer insolation to the loess-paleosol δD_wax sequence. Meanwhile, leaf wax is widely preserved in the loess sequences, which ensures that loess δD_wax records can be easily obtained. Therefore, in theory, tuning loess δD_wax time series by stalagmite δ¹⁸O and summer insolation chronologies can provide a more reliable and accurate method of determining loess chronologies than magnetic susceptibility and grain-size proxies, for establishing a highly accurate loess-paleosol time-climate sequence.

In this study, we report a δD_wax sequence spanning eight glacial-interglacial cycles to test the feasibility of establishing a new δD_wax chronology in order to provide a more reliable and accurate time-climate series for the Chinese loess-paleosol sequence for the past 800 k.y. We also aim to detect discrepancies between the newly established loess δD_wax record and the traditional magnetic susceptibility and grain-size time series, with a focus on sub–glacial-interglacial climate cycles.

We collected samples from the S1 layer, which represents the last interglacial in the Xifeng loess section of the central CLP, at 5 cm increments, and samples from the L1 to S0 layers at 20 cm increments. The magnetic susceptibility record is consistent with previous reports (Fig. 1). According to the classical age-depth framework of the loess-paleosol sequence, stratigraphic units S1–S0 are linked to Marine Isotope Stages 5–1 (MIS 5–1). The average resolution of δD_wax of n-alkane C₂₉ in the Xifeng section was 0.9 ± 0.6 k.y. per increment, yielding a millennial-scale climate change record.

The millennial-scale δD_wax index in the Xifeng loess-paleosol sequence shows a series of periodic fluctuations with amplitudes obviously larger than those of the magnetic susceptibility and the grain-size variances. The fluctuations of the loess δD_wax since the last interglacial represented by the S1 layer are comparable with the clear marker peaks in the stalagmite δ¹⁸O curve (Wang et al., 2008; Cheng et al., 2012, 2016a; Cai et al., 2015) and the summer insolation since MIS 5e (Laskar et al., 2004). However, under the traditionally tuned time scale on the magnetic susceptibility and grain-size records, these marker peaks in the δD_wax time series at the same depth from the CLP clearly lag behind the marker peaks of the stalagmite δ¹⁸O index and summer insolation index (Fig. 1).

Following previous work (Liu and Huang, 2005; Z. Wang et al., 2018), we report new δD_wax data from loess layers L8 to L3 (in 25 cm increments) in the central CLP and compile a complete δD_wax data record from layer L8 upward in the central CLP. The δD_wax values oscillate uniformly around–185‰, with an amplitude of ±15‰, indicating strong periodic fluctuations of paleoprecipitation δD values for the past 800 k.y. (Fig. 2).

Loess δD_wax values reflect the δD of precipitation. Because vegetation ecosystem changes through time are minimal in the central CLP, and plant-affected isotopic fractionation would be minimal, loess δD_wax records could directly represent the variation of signals of paleoprecipitation isotopes.

Stalagmite δ¹⁸O also reflects changes in precipitation isotopes. Note that stalagmite δ¹⁸O values from southwest China (Cai et al., 2015), Central Asia (Cheng et al., 2012), and east China (Wang et al., 2008; Cheng et al., 2016a) all show variations of paleoprecipitation δ¹⁸O in phase with the summer insolation (Fig. 1).

The loess δD_wax fluctuations indicate at least six periodicities in Xifeng in the central CLP since the last interglacial (Fig. 1), which are synchronous with the loess δD_wax variance in Weinan in the southern CLP (Thomas et al., 2016). Assuming loess δD_wax varies synchronously within the CLP and with the stalagmite δ¹⁸O and summer insolation, we tuned the age chronology of the Xifeng loess δD_wax curve to match the stalagmite δ¹⁸O and summer insolation parameter chronology for S1–S0. The comparison shows that the average lag time for the new chronology on the loess δD_wax curve is 5.4 ± 3.6 k.y. behind the traditional magnetic susceptibility chronology, with a lag time range between–1.8 and 10.0 k.y. (Fig. 1). Because the resolution of δD_wax in the S1 layer is 0.9 k.y., there is a difference of 5 k.y. between the two chronologies of S1–S0 stratigraphy. For glacial-interglacial cycles, the age uncertainty of 5 k.y. is only 5% over a 100 k.y. period, which indicates that both the traditional magnetic susceptibility–based loess time scale and the loess δD_wax time series are consistent. However, on the sub–glacial-interglacial scale, e.g., on precession periods, the 5 k.y. age uncertainty is large enough to impact research results. We recommend use of the δD_wax chronology to replace the traditional magnetic susceptibility chronology for loess-paleosol sequences, especially for studying climate changes during the last interglacial interval.

We compared loess δD_wax records of layers L2–S2 in Xifeng (Z. Wang et al., 2018) and Weinan (Thomas et al., 2016), which show in-phase fluctuations (Fig. S2 in the Supplemental Material¹). Under the traditional chronology tuned by magnetic susceptibility values, the δD_wax records display a lag phase compared to the chronologies of stalagmite δ¹⁸O and summer insolation (Fig. S2). This means the chronological uncertainties of the traditional loess-paleosol sequence remain measurable not only in the younger, but also in the older, segments of the sequence.

To apply the method of tuning to older loess δD_wax chronology, we correlated the chronologies of stalagmite and summer insolation to the δD_wax record in Xifeng up to the L8 loess layer (Fig. 2). Based on the loess-paleosol stratigraphy corresponding to glacial-interglacial cycles, the loess δD_wax and summer insolation records are synchronous (Fig. 2). Although the average 3.4 k.y. resolution of 25 cm increments is too low to identify climate events (indicated by low value peaks of the indexes in the lower paleosol stratigraphies), the possibility of synchronous climate events between the summer insolation and δD_wax records up to layer L8 is still measurably clear.

Because stalagmite δ¹⁸O records only date back to 640 ka, corresponding to loess-paleosol stratigraphy back to the bottom of layer S5, the summer insolation chronology was used to establish the chronology of the loess δD_wax proxy below the S5 unit. We found that the loess δD_wax variation, particularly, better matches the stalagmite δ¹⁸O and summer insolation during interglacial periods than during glacial periods. Therefore, the loess δD_wax time series has higher reliability in paleosol than in loess layers. Under the traditional chronology, an average 5 k.y. time lag occurs between low value peaks of δD_wax and the corresponding maximum values of summer insolation, which is consistent with the 5 k.y. lag time above layer S1 (Fig. 2). Thus, the new chronology has significance for studying longer loess-paleosol sequences in the CLP. When the vegetation changes between grass and trees through time are taken into account, we recommend using loess δD_wax time series as a more accurate age-controlled monsoon precipitation proxy to study loess-paleosol deposits and associated climate and environment changes.

The analysis of loess records based on orbital cycles, such as eccentricity or precession, is a common method to demonstrate the driving factors of climatic or environmental changes in the CLP. The wavelet power spectrum analysis indicates that δD_wax variation in Xifeng has clear periods of 23 k.y. and 41 k.y. in the newly tuned time series (Fig. S3). This means that loess-paleosol sequences contain consistent periodic characteristics similar to stalagmite δ¹⁸O and other important global records.

In previous studies, phase analysis has been used as basic evidence to discuss climatic forcing mechanisms. For instance, the phase differences between the loess records in the CLP and marine or stalagmite records in other regions are between–120° and 130°, which were interpreted as a regional difference in comprehensive climatic forcing (Peterse et al., 2014; Thomas et al., 2016). However, the new δD_wax chronology shows that the traditional chronology of the loess-paleosol alternation has an ∼5 k.y. time lag (Fig. 2), which leads to uncertainty of the phase difference by 80° on precession cycles. In some loess stratigraphy, the phase uncertainty due to the error of the traditional chronology (developed from either magnetic susceptibility or grain-size proxies) becomes as high as 180°. On the sub–glacial-interglacial scale, the inherent differences between the loess record and other records may have been partially or completely covered by the uncertainty of the traditional loess chronology. The error of the traditional chronology itself has actually upset the interpretation of existing climate-driven mechanisms.

The strong hydrophobicity of leaf wax ensures that loess δD_wax will not be affected by precipitation leaching, which is an advantage over the traditional proxies. We believe that the high sensitivity of δD_wax variations to monsoon rainfall strength over the CLP makes it the best proxy for tuning the Chinese loess-paleosol chronology, which will eliminate the uncertainty for comparison studies that seek the true inherent climate characteristics of the CLP. It also implies that magnetic susceptibility or grain-size chronologies are no longer the best option for loess and climate change research on sub–glacial-interglacial scales, and need to be replaced by the new δD_wax chronology.

At present, a growing body of evidence shows that the δ¹⁸O and δD_wax values on the Tibetan Plateau (Chen et al., 2020), adjacent to the CLP, also vary in a similar pattern as the stalagmite δ¹⁸O, though the local temperature and humidity obviously differ from region to region (Rao et al., 2016; Chen et al., 2020). This indirectly indicates that the precipitation isotope is the best proxy to link the chronology of geological records in different regions over semicontinental scales. It also suggests that the application of the δD_wax chronology can even be extended to wider areas for discussion of alternative climate-driven environmental changes and associated mechanisms on sub–glacial-interglacial scales.

This work was supported by the Strategic Priority Research Program of Chinese Academy of Sciences (grants XDB40000000 and XDA2007020202) and the National Natural Science Foundation of China (grant 41773010).