Introduction

Quaternary red earth sediments are widespread in the middle and lower reaches of the Yangtze River in southern China, and these deposits have experienced intense syndepositional pedogenesis under humid tropical to subtropical conditions1,2. Like the loess-paleosol sediments of northern China, the red earth sediments in southern China are also records of climatic and environmental changes during the period of their formation3,4,5,6,7.

The Jiujiang red earth section is located on the second terrace of the Yangtze River (Fig. 1), at 29°42′40.27″N, 116°00′13.7″E, 41 m above sea level2. The study area has a wet subtropical monsoonal climate that is warm and humid in summer and relatively cool and dry in winter, with mean annual precipitation (MAP) of 1000 to 1600 mm and mean annual temperatures (MAT) of 16–18 °C2. The red earth profile is 14 m thick and divided lithologically into three units (Fig. 2). The upper unit (0–3.9 m) consists of light orange yellow (7.5YR 9/8) to light moderate orange (2.5YR 7/8) sandy red earth, contains minor amounts of Fe-Mn cutans and Mn nodules, and exhibits a homogeneous texture without white veining. The middle unit (3.9–6.2 m) consists of light moderate orange (2.5YR 7/8) to strong reddish orange (7.5R 6/12) red earth and exhibits white fine net-like veins. The lower unit (6.2–14.0 m) consists of strong reddish orange (7.5R 6/12) to deep reddish orange (7.5R 4/12) red earth and exhibits white coarse net-like veins, which are more abundant and larger than those in the middle unit (Fig. 3). Optically stimulated luminescence (OSL) dating of samples at 0.8 and 2.2 m yielded ages of 40.8(±4.9) and 51.9(±5.8) ka, respectively, and electron spin resonance (ESR) dating of samples at 6.3, 8.1, 10.9, and 13.8 m yielded ages of 393(±45), 452(±43), 592(±77), and 685(±65) ka, respectively. The OSL method is usually more reliable for materials <100 ka and the ESR method more reliable for materials >100 ka. An age of 685(±65) ka for the base of the Jiujiang section indicates that the red earth accumulated after the middle Pleistocene8, and ages of 40.8(±4.9) and 51.9(±5.8) ka for the top of this section demonstrates that the upper unit was deposited during the late Pleistocene9. Average sedimentation rates are 12.6 cm/kyr for the upper unit (based on ages at 0.8 and 2.2 m) and 2.57 cm/kyr for the middle and lower units (based on ages at 6.3 and 13.8 m). Comparison with an earlier Sm-Nd isotopic study shows that the Jiujiang red earth sediments have a provenance similar to that of red earth deposits in the middle to lower reaches of the Yangtze River10. This suggests that all of the red earth sediments were generated within the middle to lower Yangtze River hydrologic basin.

Figure 1
figure 1

A generalized map showing the location of the study area. Adapted from a map of the National Administration of Surveying, Mapping and Geoinformation of the People’s Republic of China (http://bzdt.nasg.gov.cn/). The map is redrawn by Ke Yin using CorelDRAW X7.

Full size image
Figure 2
figure 2

Stratigraphy of the Jiujiang section. OSL ages and ESR ages from Hong et al. (2013). Ke Yin created this figure using CorelDRAW14.

Full size image
Figure 3
figure 3

Photographs of the Jiujiang red earth section. Hanlie Hong took photo of the section, and Ke Yin edited these photos using CorelDRAW14. (a) Full view of section; (b) Close-up of middle unit, consisting of net-like red earth with fine veins; (c) Close-up of lower unit, consisting of net-like red earth with coarse veins.

Full size image

After deposition, the Yangtze Basin red earth sediments underwent intense chemical weathering under warm and humid climatic conditions, as shown by high Rb/Sr ratios, intense depletion of mobile elements and concentration of immobile elements, as well as the well-developed net-like structure10. Although for the past few years the red earth sediments have attracted attention from many researchers in order to reconstruct paleoenvironments and paleoclimates in southern China, the climatic significance of these deposits is still debated. Studies of magnetic susceptibility and stable carbon isotopic compositions demonstrated the existence of eight depositional-pedogenic cycles in the Xiangyang section related to paleoclimatic changes11,12,13,14, although these findings were later challenged on the basis of an unreasonable degree of weathering of the “paleosols” and “loesses”15,16 and various soil parameters (Munsell color values, magnetic parameters, and stable carbon isotopic compositions)17. In view of these disagreements, further study of the southern Chinese red earth sediments is needed to evaluate the existence of multiple depositional-pedogenic cycles and their paleoclimatic significance.

Pedogenic processes associated with intense chemical weathering produce soil minerals reflecting the prevailing climatic conditions. Iron minerals, in particular, have proven to be powerful tools for reflecting soil formation intensity and paleoclimatic evolution18,19. Goethite and hematite, which have similar thermodynamic stabilities, are the most common Fe oxides in soils. Soil factors such Eh, pH, temperature, and organic matter content influence the formation and stability of goethite and hematite19,20. Cooler and wetter conditions promote formation of goethite, which imparts a yellow-brown color to soils, whereas warmer and drier conditions favor hematite19. Hematite, especially the fine-grained variety, imparts a reddish color to soils, that masks the yellowish color of goethite21. Therefore, the ratio of hematite to goethite (i.e., hematite/(hematite + goethite, or Hm/(Hm + Gt)) is commonly used as a climate proxy in soil studies18,22. However, traditional analytical techniques such as visually measured soil color proxies, X-ray diffraction, and Mössbauer spectrometry are either not sufficiently precise or too time-consuming for determination of hematite and goethite concentrations in large numbers of samples22,23,24. The recent development of diffuse reflectance spectroscopy (DRS) has provided a faster, more precise, and nondestructive method of quantification of hematite and goethite in soils22,25. The presence of either hematite or goethite can be detected at less than 0.1% in mixtures with other soil minerals22,26. Rock magnetic techniques can also be used to obtain information about pedogenic processes and paleoclimate evolution from iron-oxide assemblages in soils27,28,29,30,31,32. The most common pedogenic magnetic minerals in red earth sediments of southern China are maghemite, hematite, and goethite, and transformations among them result in variable mineralogic concentrations and magnetic intensities in soils29. Thus, a combination of rock magnetic and spectroscopic techniques can be used to retrieve paleoclimatic information recorded in the Fe-oxide fraction of paleosols25,33.

In response to climatic warming and increased humidity, the red earth sediments in southern China were subject to relatively intense pedogenic alteration8, resulting in soil-mineral transformations, especially among iron oxides. The Jiujiang red earth sediments were accretionary in nature, reflecting some degree of syndepositional pedogenesis under tropical to subtropical conditions8. Therefore, we take it that post-depositional diagenesis played a very weak role in the red earth. The main goal of the present study was to investigate the Fe-oxide mineralogy of the Jiujiang red earth section, with supplementary information from silicate minerals, in order to reveal the paleoclimatic evolution of the middle to lower Yangtze River of southern China during the middle and late Pleistocene.

Results

Diffuse reflectance spectrophotometry (DRS)

In the Jiujiang red earth section, hematite content ranges from 0.04 to 11.16 g kg−1, whereas goethite content ranges from 0.03 to 3.78 g kg−1. The Hm/(Hm + Gt) ratio ranges from 0.01 to 0.98. Based on these proxies, three stages of accumulation of the Jiujiang red earth section can be identified. The lower unit (6.2–14.0 m) contains the highest concentrations of hematite (4.93 to 10.87 g kg−1; mean 9.28 g kg−1) and the lowest concentrations of goethite (0.20 to 2.85 g kg−1; mean 1.16 g kg−1), yielding the highest Hm/(Hm + Gt) ratios (0.75 to 0.98; mean of 0.89). The hematite content in the lower unit (6.2–14.0 m) is relatively stable although with minor fluctuations at depths of 11.4–12.0 m and a larger negative excursion at depths of 13.0–14.0 m. Its goethite content ranges from 0.20 to 2.85 g kg−1, with relatively higher content at depths of 11.0–14.0 m (mean 1.27 g kg−1). The middle unit (3.9–6.2 m) also contains more hematite (6.33 to 11.6 g kg−1; mean 8.46 g kg−1) than goethite (0.64 to 1.70 g kg−1; mean 1.14 g kg−1), although Hm/(Hm + Gt) ratios (0.81 to 0.91; mean of 0.84) are lower than in the lower unit. Within the middle unit, both the hematite and goethite content decline upsection, although the decline in hematite is relatively greater, imparting a pale yellow to yellowish-brown color to the upper part of this unit. The upper unit (0–3.9 m) exhibits the lowest hematite concentrations (0.04 to 6.43 g kg−1; mean 2.18 g kg−1) and the highest goethite concentrations (0.03 to 3.78 g kg−1; mean of 1.54 g kg−1), yielding the lowest Hm/(Hm + Gt) ratios (0.01 to 0.94; mean of 0.51). The contents of both hematite and goethite oscillate considerably, and neither proxy exhibits an obvious decreasing or increasing trend upsection. For the Jiujiang section as a whole, goethite content fluctuates more strongly than hematite content, which is relatively stable in the lower and middle units and oscillates only in the upper unit. From the base of the section upward, goethite exhibits an overall increasing trend and hematite a decreasing trend resulting in a pronounced decline in Hm/(Hm + Gt) ratios upsection (Fig. 4).

Figure 4
figure 4

Variations of hematite and goethite content and Hm/(Hm + Gt) along the Jiujiang red earth section. Ke Yin created this figure using Grapher 9 and CorelDRAW14. a-δ18O records of the equatorial Pacific Core V28-238 adapted from Shackleton and Opdyke, 1973; b- magnetic susceptibility record of the Chinese loess obtained by averaging data from localities Luochuan and Xifeng (Kukla et al.56). (Luochuan County, Shanxi Province) and Xifeng (Xifeng District, Gansu Province) ~360 km to the west.

Full size image

Magnetic susceptibility (χlf)

The magnetic susceptibility (χlf) of the entire section ranges from 3.15 × 10−8 to 1.32 × 10−6 m3 kg−1 with a mean of 2.21 × 10−7m3 kg−1. χlf is most variable in the upper unit (1.01 × 10−7 to 1.32 × 10−6 m3 kg−1; mean 4.77 × 10−7 m3 kg−1), whereas it exhibits smaller fluctuations in the middle unit (7.34 × 10−8 to 2.70 × 10−7 m3 kg−1; mean of 1.70 × 10−7 m3 kg−1) and lower unit (3.15 × 10−8 to 2.23 × 10−7 m3 kg−1; mean of 1.02 × 10−7 m3 kg−1). χlf exhibits an overall increasing trend upsection (Fig. 5).

Figure 5
figure 5

Low-field magnetic susceptibility of Jiujiang red earth section. Ke Yin created this figure using Grapher 9 and CorelDRAW14.

Full size image

X-ray diffraction (XRD)

X-ray diffraction analysis of the Jiujiang red earth section shows that all bulk samples contain similar clay assemblages and other mineral species. The main non-clay minerals are quartz (identified from the 0.425 and 0.333 nm peaks) and minor feldspars (orthoclase and plagioclase identified from the 0.324 and 0.319 nm peaks, respectively) (Fig. 6a,b,c). The absence of other silicates suggests intense chemical weathering conditions since the middle Pleistocene. Feldspars (both orthoclase and plagioclase) have notably higher contents in the upper unit than in the middle and lower units (Fig. 7), a pattern suggesting decreasing weathering intensity towards the surface of the profile.

Figure 6
figure 6

XRD patterns of representative bulk samples from the upper (a), middle (b), and lower (c) units of the Jiujiang red earth section. I-illite; Kao-kaolinite; Q-quartz; Or-orthoclase; Pl-plagioclase. Ke Yin created this figure using Grapher 9 and CorelDRAW14. I-illite; Kao-kaolinite; Q-quartz; Or-orthoclase; Pl-plagioclase.

Full size image
Figure 7
figure 7

Plagioclase, orthoclase, and quartz content and Or/(Or + Pl) of Jiujiang red earth section. Ke Yin created this figure using Grapher 9 and CorelDRAW14.

Full size image

The main clay minerals in the Jiujiang red earth section are illite, kaolinite, and vermiculite, with trace amounts of mixed-layer kaolinite–smectite and mixed-layer illite–vermiculite, as determined from oriented clay samples in previous XRD studies8,34,35,36. Bulk-sample XRD data show that the clay assemblages of the lower and middle units are dominated by illite and kaolinite, whereas the upper unit contains illite, vermiculite, and kaolinite. Under acidic soil-water conditions, high-charge vermiculite can transform into kaolinite via mixed-layer vermiculite-kaolinite37,38,39. However, the amounts of the two mixed-layer phases in the Jiujiang section are quite low, as shown by an absence of characteristic peaks in bulk-sample XRD traces. The absence of vermiculite in bulk samples from the middle and lower units is due to their more highly weathered condition compared to the upper unit, reflecting warmer and more humid climatic conditions during their formation. Feldspars and vermiculite were more stable under the less intense weathering conditions that prevailed during accumulation of the upper unit. Hence, the clay mineralogy of the Jiujiang red earth section records a major paleoclimate shift from warm/humid to cool/dry conditions during the middle to late Pleistocene.

Discussion

Goethite is generally the dominant Fe-oxide in loess-paleosol sections of northern China owing to locally cool and wet pedoclimatic conditions40, whereas Fe-oxides (especially hematite) dominate in the red earth sediments of southern China owing to warm and seasonally dry conditions10,41. Hm/(Hm + Gt) ratios have been used as a paleo-environmental proxy in reconstructing Quaternary climate changes in the loess-paleosol sections of northern China and in other soil types18,25,40. However, previous paleoclimatic studies of red earth sediments in southern China have focused on other features, e.g., grain-size characteristics3,42, environmental magnetism5,17, geochemical compositions43,44,45, and clay-mineral assemblages8,34,45, rather than Hm/(Hm + Gt). The Hm/(Hm + Gt) ratios of soils are closely related to MAT and MAP19 and, thus, can effectively record paleoenvironmental changes during soil formation18,22.

High temperature and moisture levels contribute to weathering of tectosilicates (i.e., quartz and feldspars)46,47. Quartz is generally resistant under most conditions, whereas orthoclase and plagioclase are leached much more rapidly than quartz in an acidic environment48. Hence, the (orthoclase + plagioclase)/quartz ratio, or [(Or + Pl)/Q], can serve as a proxy for the degree of chemical weathering of soils49.

Magnetic susceptibility (χlf) has been a useful proxy for reconstructing pedogenic histories and paleoclimatic changes in the loess–paleosol sequences of northern China50,51,52. However, χlf must be used cautiously as a paleoclimatic proxy in paleosols because of complications related to material sources, moisture regimes, pedogenic weathering intensity, and the presence of lithogenic ferrimagnetic minerals53,54,55,56. Especially in the red earth sediments of southern China, χlf does not accurately reflect paleoclimatic conditions because of the strong influence of post-depositional hydromorphic processes (i.e., related to the rise and fall of the groundwater table57). Although many ferromagnetic minerals contributing to the χlf signal are dissolved by such secondary processes, hematite and goethite are not significantly affected30. For this reason, the Hm/(Hm + Gt) ratio can be an effective paleo-environmental proxy in the red earth sections of southern China.

Whereas the Hm/(Hm + Gt) and (Or + Pl)/Q profiles at Jiujiang are consistent in indicating a climate shift from warm/humid to cool/dry conditions during the middle to late Pleistocene, the magnetic susceptibility curve offers an apparently contradictory signal. In the loess-paleosol sections of northern China, higher χlf reflects greater neoformation of fine-grained pedogenic ferrimagnetic particles, which is generally interpreted to represent more highly weathered soils resulting from warmer and more humid climate conditions25,50,58. At Jiujiang, an upward increasing trend in χlf (Fig. 8) thus appears to contradict findings from the Hm/(Hm + Gt) and (Or + Pl)/Q profiles. However, the red earth sediments from southern China have experienced more intense chemical weathering than the loess deposits from northern China2,30,45, and the warmer, more humid conditions of southern China contributed to breakdown of ferrimagnetic particles55, resulting in lower χlf values. Hence, the paleoclimatic information recorded by the χlf proxy is, in fact, consistent with interpretations of the Hm/(Hm + Gt) and (Or + Pl)/Q data at Jiujiang. The relatively smooth upward changes in these proxies through the lower and middle units indicate that the study area experienced gradual climate change from warm and humid to cooler and drier during the middle to late Pleistocene, and the pronounced oscillations in all three proxies through the upper unit reflect multiple climate cycles since the last interglacial period59,60,61.

Figure 8
figure 8

Comparison of proxies for Jiujiang red earth section with (a) δ18O records of the equatorial Pacific Core V28-238 (Shackleton and Opdyke, 1973) and (b) magnetic susceptibility from northern Chinese loess-paleosol sections (an average of data from Luochuan and Xifeng; Kukla et al.56). Ke Yin created this figure using Grapher 9 and CorelDRAW14.

Full size image

The existence of multiple climate subcycles with periodicities of ≤100 kyr during the middle and late Pleistocene has been demonstrated by magnetic susceptibility profiles from loess-paleosol sections in northern China (e.g., Luochuan and Xifeng) as well as by δ18O records from the equatorial Pacific (Fig. 8)58,62. Luochuan (Shanxi Province) and Xifeng (Gansu Province) are typical loess-paleosol sections of the Chinese Loess Plateau that are subdivided (from base to top) into the Wucheng Loess, Lower Lishi Loess, Upper Lishi Loess, Malan Loess, and Black Loam58,62. Although the Jiujiang section is stratigraphically equivalent to paleosols S1-S6 and loess units L1-L7 of the northern Chinese sections (Fig. 8), it does not show the same climate subcycles. The smooth pattern of climate change recorded by multiple proxies at Jiujiang between ~390 and 690 ka is due to the intense chemical weathering under warm, humid climate conditions experienced by this section. Consequently, the shorter-term (≤100-kyr) climate signals recorded in the loess-paleosol sections of northern China were overprinted in the red earth sections of southern China, which, thus, do not offer the same degree of paleoclimate detail. The controversies over paleoclimate interpretations of the red earth sediments of southern China are largely the result of use of different paleoclimate proxies in key studies11,12. The present study provides new insights into the implications of multiple proxies for understanding the paleoclimatic history of southern Chinese red earth sediments.

Methods

The Jiujiang section was sampled at 10 cm intervals. For diffuse reflectance spectrophotometry (DRS), the powders were pressed into black plastic holders at a pressure of >500 kPa. Reflectance spectra of samples were analyzed in a TU-1901 double-beam UV/VIS-spectrophotometer with a diffuse reflectance attachment (reflectance sphere) from the near-ultraviolet to the near-infrared (380–750 nm). Data processing was restricted to the visible spectrum (400–700 nm), which is the most sensitive region for Fe-oxide minerals with discrete measurements at 1-nm intervals63. The DRS method based on the second-derivative calculation was applied to quantify goethite and hematite via the band intensity in the second-derivative curves between the ~415-nm minimum and the ~445-nm maximum for goethite and between the ~535-nm minimum and the ~580-nm maximum for hematite64 (Fig. 9). The second-derivative method has an advantage over redness ratings in that it helps simultaneously to predict hematite and goethite contents, and it is also a fast, precise, and non-destructive method that has been widely used in many studies18,64. The concentrations of hematite and goethite in red earth samples were quantified using the second-derivative approach proposed by Scheinost et al.64 based on the following equations:

$${\rm{Hematite}}\,({\rm{g}}\,{{\rm{kg}}}^{-{\rm{1}}})=-{\rm{0.09}}+{{\rm{402Y}}}_{{\rm{1}}}({{\rm{r}}}_{{\rm{2}}}={\rm{0}}\mathrm{.85};\,{\rm{n}}=\mathrm{40};\,{\rm{p}}({\rm{\alpha }}) < {\rm{0.001}})$$
(1)
$${\rm{Goethite}}\,({\rm{g}}\,{{\rm{kg}}}^{-{\rm{1}}})=-{\rm{0.06}}+{{\rm{268Y}}}_{{\rm{2}}}({{\rm{r}}}_{{\rm{2}}}={\rm{0}}\mathrm{.86};\,{\rm{n}}=\mathrm{40};\,{\rm{p}}({\rm{\alpha }}) < {\rm{0.001}})$$
(2)

where Y1 stands for the second-derivative amplitude between the ~535-nm minimum and the ~580-nm maximum of hematite, and Y2 stands for the second-derivative amplitude between the ~415-nm minimum and the ~445-nm maximum of goethite.

Figure 9
figure 9

Second-derivative curves of representative samples from the upper, middle, and lower units of the Jiujiang red earth section. Ke Yin created this figure using Origin 8.0 and CorelDRAW14.

Full size image

For rock magnetic measurements, the dried powder samples were transferred to plastic boxes and subsequently compressed and fixed with cotton wool before securely fastening the lid in order to prevent movement of sediment particles during the measurements. The magnetic susceptibility was measured at room temperature with a MFK1-FA magnetic susceptibility meter (AGICO, Brno) and is given as mass-specific susceptibility (χlf), using an alternating field of 200 A/m at 976 Hz measurement frequency. The susceptibility (χlf) depends on the concentration and grain-size distribution of magnetic minerals18.

For X-ray diffraction analysis (XRD), bulk samples were air-dried and then crushed and ground manually to powder in an agate mortar with a pestle. The XRD patterns of the bulk samples were collected using a Panalytical X’Pert PRO DY2198 diffractometer at the Laboratory of Geological Process and Mineral Resources, China University of Geosciences (Wuhan). The instrument was operated at 40 kV and 40 mA with Ni-filtered Cu Kα radiation. It was measured from 3° to 35° 2θ at a scan rate of 4° 2θ/min and a step size of 0.02° 2θ. Relative abundances of quartz, plagioclase, and orthoclase were estimated semi-quantitatively using the weighting factor method with reference material of corundum65.