Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
 The Chinese loess/paleosol sequence provides an excellent record of long-term variations in the East Asian paleomonsoon. For representative loess profiles, paleosol units have enhanced magnetic properties compared to loess units. However, at some depth intervals with special paleoenvironmental conditions, selective depletion of antiferromagnetic (AFM) minerals (hematite and goethite) could occur, resulting in complexities in accurately linking variations in magnetic properties to long-term fluctuations in paleoclimate. To resolve this problem, we directly measured the mass concentrations of these AFM minerals using second-order diffuse reflectance spectroscopy (DRS), and then correlated the DRS results to high-field isothermal remanent magnetization (HIRM) from paleosol unit S5 at the Luochuan profile, central Chinese Loess Plateau. Our results suggest that a low HIRM anomaly just below the sub-paleosol unit S5S1 is caused by fluctuations specifically in goethite content, while the hematite concentration exhibits a maximum over the same depth interval. This appears to rule out significant loss of hematite to reductive dissolution and further indicates that hematite and goethite may respond differently to changes in the paleoclimate conditions.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 It is well documented that loess units were deposited during cold/dry periods when the East Asian winter monsoon prevailed over the Chinese Loess Plateau (CLP) with a provenance from the deserts northwards of the CLP. In contrast during warm/humid periods, soils were developed when the East Asian summer monsoon brought moisture to the CLP through the equatorial Pacific Ocean. Thus, to first order, the interbedded loess/paleosol sequence reflects glacial/interglacial cycles.
 To quantify such periodic paleoclimate rhythms recorded in the Chinese loess/paleosol sequences, both magnetic and non-magnetic proxies have been proposed. Compared to the non-magnetic proxies, the magnetic techniques are non-destructive, inexpensive and efficient, and thus have been widely used to determine the basic pedostratigraphy from the onset of loess study. The most popular magnetic parameter is the low-field magnetic susceptibility (χ, mass-specific). For a characteristic loess profile, interglacial paleosol units have enhanced χ values compared to underlying glacial loess units. This is mainly due to the neoformation of extra fined-grained (in the superparamagnetic/SP, and single domain/SD grain size region) maghemite particles through pedogenesis. This pattern has been regarded as the standard pedogenesis model for the Chinese loess/paleosol sequence.
 However, soils can also suffer magnetic depletion if, for example, magnetic particles are reductively dissolved in anaerobic soil environments [Maher, 1998]. For example, Bloemendal and Liu  documented a significant decrease of the high-field remanent magnetization (HIRM) compared to the background trend within a narrow depth interval below the paleosol unit S5SS1 (where S5 and SS1 denote the soil unit 5, and the sub soil unit 1, respectively) at the Luochuan section, in the central CLP. The HIRM is a proxy of the concentration of hematite + goethite [Thompson and Oldfield, 1986; Maher and Dennis, 2001], and is defined as 0.5*(SIRM + IRM−300mT), where SIRM and IRM−300mT represent the saturation isothermal remanent magnetization and the remanence obtained after applying a field of −300 mT to reverse the SIRM, respectively. As a result, Bloemendal and Liu  proposed that antiferromagnetic hematite and goethite particles have been partly dissolved within this special interval.
 However, there are uncertainties about the interpretation of HIRM because both hematite and goethite contribute to the bulk HIRM. It is unclear whether the HIRM low is caused by the decrease in the concentration of goethite, hematite or both. Recently, Torrent et al.  used second-order diffuse reflectance spectroscopy (DRS) to quantify the absolute mass concentration of hematite and goethite particles in the Chinese loess. In this paper, we applied the DRS technique to quantify the mass concentrations of both hematite and goethite for the same set of loess samples examined by Bloemendal and Liu  in order to determine which mineral (hematite or goethite) dominates the HIRM signal. We then further discuss whether the investigated stratigraphic interval has suffered significant selective dissolution of hematite and/or goethite.
2. Sampling and Experiments
 Samples examined in this study from the Luochuan section (about 135 m in thickness at 35.8°N, 108.6°E, Figure 1) were first collected by Bloemendal and Liu , and then shipped to the University of Southampton, UK. To avoid surficial alteration or contamination effects, the outermost ∼50 cm thickness of sediment was removed. Samples were then obtained at 5 cm intervals. For this study, we focus only on the specific interval spanning loess unit L6, paleosol S5, and the lower part of loess unit L5.
χ was measured at two frequencies (low frequency, LF, 0.47 kHz; high frequency, HF, 4.7 kHz) using a Bartington Instruments magnetic susceptibility meter. The absolute frequency-dependent susceptibility χfd and the corresponding relative χfd (χfd%) were respectively defined as χLF−χHF and 100%*(χLF−χHF)/χLF, were calculated from these measurements.
 To quantify the mass concentrations of hematite and goethite, the DRS curves of fine powdered samples were acquired with a Varian Cary 1E spectrophotometer equipped with a BaSO4–coated integrating sphere 73 mm in diameter. Then the second derivative spectrum of the Kubelka-Munk (K-M) remission function was calculated according to Torrent and Barrón  and the intensities of the bands ∼425 nm (I425) and ∼535 nm (I535), which are proportional to the concentration of goethite and hematite, respectively [Scheinost et al., 1998] were measured. More specifically, the Hm/(Hm + Gt) mass ratio was obtained through the calibration curve Y = −0.133 + 2.871X − 1.709X2 (where Y is the Hm/(Hm + Gt) mass ratio and X is the I535/(I425 + I535) ratio). This calibration curve was based on 22 samples of soils from the Mediterranean region containing Fe oxides resembling those of our loess samples (and quantitatively determined using X-ray diffraction). Undoubtedly, goethite and hematite may differ in terms of crystal size, substituted cations, reflectance properties, etc [Torrent and Barrón, 2003]. Nevertheless, our calibration curve was based on actual soil samples and does not differ much from one soil type to another (e.g. soils of the Mediterranean region or from tropical regions); in addition, the I535/(I425 + I535) ratio is nearly independent of the matrix (e.g. the presence of more or less organic matter, etc.). The R2 of that calibration curve is 0.90; therefore, the error of the estimate was relatively small. The absolute concentrations of hematite and goethite were estimated by assigning all the citrate/bicarbonate/dithionite-extractable Fe [Mehra and Jackson, 1960] (hereinafter named Fed) to these two minerals. Total Fe (Fet) was determined by isotope-source X-ray fluorescence using a Metorex (Finland) XMET920 system with a 3–20 keV probe (109Cd source).
 Based on magnetic properties, the paleosol unit S5 can be divided into three sub-paleosol units S5S1-S5S3 in terms of the enhanced χ (Figure 1a), total Fe (Figure 1b), and χfd (Figure 1c). The sub-paleosol units are bracketed by the sub-loess units S5L1-S5L2 and the loess units L5 and L6. Amongst these units, the magnetic properties of S5S1 have been highly enhanced, which indicates that this sub-unit underwent the highest degree of pedogenesis.
χfd% can be used to indicate the grain size distribution (GSD) of fine-grained magnetic particles. Worm  and Worm and Jackson  suggested that samples with narrower size distributions of SP + SD particles have higher χfd% values than samples with a broader GSD. However, for the Chinese loess, χfd% can be significantly lowered due to effects from the aeolian inputs (with a background susceptibility of χo). χo can be estimated from the y-axis intercept from the plot of χLF versus χfd [Forster et al., 1994; Forster and Heller, 1997; Liu et al., 2005]. After removing χo from χLF, the corrected χfd% is almost independent of the degree of pedogenesis (Figure 1d), which indicates that the GSD of pedogenic particles is almost constant [Liu et al., 2004, 2005] within S5, especially for the interval with the low HIRM anomaly.
 HIRM is generally used as a proxy for the concentration of hematite and goethite [e.g., King and Channell, 1991]. Compared to the other proxies, variations in HIRM have a higher-frequency character (Figure 1e). The most notable feature is the HIRM minimum at ∼34.6 m, which coincides with the minimum of the goethite mass concentration at the same depth interval (Figure 1f). In contrast, the hematite mass concentration shows a broader similarity compared to the bulk susceptibility (Figure 1g).
 To further examine the contributions of hematite and goethite to the bulk HIRM, the bulk HIRM was decomposed into two components: the long-term trend obtained by 13-point smoothing (Figure 2a), and the corresponding residual (Figure 2b). The correlation coefficients between HIRM and hematite (Figure 3a), and between HIRM and goethite (Figure 3c) are improved when correlating the HIRM trend and the HIRM residual to the hematite and goethite mass concentrations, respectively (Figures 3b and 3d). This strongly suggests that the high-frequency component of HIRM is influenced by the goethite content, whereas the low-frequency trend is more controlled by the hematite concentration.
 Because the GSD of the pedogenic fine-grained maghemite particles remains relatively constant, χfd can be used as a proxy for the mass concentration of these pedogenic magnetic particles. The linear correlation between the hematite mass concentration and χfd strongly indicates that the hematite contents are not preferentially dissolved through pedogenesis (Figure 4).
Bloemendal and Liu  proposed that the low HIRM anomaly just below the sub-paleosol unit S5S1 is due to the selective destruction of antiferromagnetic materials. They further proposed that the maxima in other magnetic proxies (e.g., χ and SIRM) at this special depth interval suggested neoformation of the strongly magnetic ferromagnetic phases from antiferromagnetic phases. Because maghemite and magnetite have a magnetization about two orders of magnitude higher than that of hematite and goethite, the conversion from a significant amount of hematite + goethite to maghemite or magnetite would result in a large corresponding increase of χ and χfd. However, these proxies do not show anomalies at ∼34.5 m where the HIRM low occurs. Therefore, we can exclude the transformation from weakly antiferromagnetic minerals to strongly ferromagnetic minerals at this depth interval.
 Instead, our new results show that almost all pedogenesis-related proxies exhibit normal patterns except for goethite. For example, χfd% remains relatively constant from 32–38 m, suggesting that the GSD of pedogenic maghemite particles remains almost unchanged within S5. Moreover, the hematite concentration (Figure 1g) is positively correlated to the pedogenic maghemite concentration (Figure 4), indicating that hematite and maghemite were formed concomitantly in the CLP paleosols. The low goethite concentration at ∼34.5 m suggests that goethite and hematite could have very different responses to changes in paleoclimatic and soil-forming conditions within this depth interval possibly due to 1) reduced input of aeolian goethite, 2) reduced formation of goethite through pedogenesis, and 3) selective dissolution of goethite, which rarely occurs in soils containing paragenetic hematite and goethite.
 Because HIRM is controlled by variations in the concentration of both goethite and hematite, it is problematic to directly infer that low HIRM values are caused by selective dissolution of goethite and hematite. In this study, we quantified the mass concentration of goethite and hematite using a DRS technique. The new results show that the HIRM minimum at ∼34.5 m in the Luochuan profile is mostly caused by a low goethite concentration and is not reflected in the concentration of hematite and the grain size distribution and concentration of pedogenically-produced fine-grained maghemite particles. Therefore, no selective dissolution of hematite occurred within this specific depth interval. The lower goethite concentration at this depth interval could be due to selective dissolution, or to reduced input of aeolian goethite and/or reduced neoformation of goethite through pedogenesis. Regardless of the exact mechanism, this strongly indicates that goethite and hematite may respond differently to changes in paleoclimatic conditions.
 C. L. Deng was supported by NSFC grants 40221402 and 40325011 and CAS grant KZCX3-SW-150. J. Torrent was supported by Spain's Ministerio de Ciencia y Tecnología, project AGL2003–01510. Q. S. Liu was supported by a European Marie-Curie Fellowship (IIF), proposal 7555. J. Bloemendal was supported by the Natural Environment Research Council, UK.