INTRODUCTION
The East Asian monsoon is one of the largest monsoon systems on Earth (Ao et al., Reference Ao, Liebrand, Dekkers, Roberts, Jonell, Jin and Song2024), influencing the hydroclimatic conditions of regions comprising land on which a third of the world’s population depends. The modern climate of East Asia is dominated by a pronounced seasonal cycle, with prevailing wind directions producing dry and cold winters and wet, warm conditions during late spring and summer (Maher, Reference Maher2016). The intensity of the monsoon system is governed by interactions between continents, oceans, and the atmosphere, primarily forced by insolation (Lu et al., Reference Lu, Wang, Wang, Zhang, Yi, Wang, Stevens, Kurbanov and Marković2022). The East Asian winter monsoon (EAWM) is influenced by northerly winds generated by the Siberian High, which creates a northwest–southeast temperature gradient. In contrast, the East Asian summer monsoon (EASM) is driven by the thermal gradient between the warming land and comparatively cooler oceans. This gradient triggers the transport of moisture and heat from the low-latitude Indian and Pacific Oceans via southwesterly and southeasterly winds, respectively (Maher, Reference Maher2016; Chiang et al., Reference Chiang, Kong, Wu and Battisti2020).
Understanding the evolution and variability of the East Asian monsoon over time remains a significant challenge. However, geological records provide valuable insights into the characteristics of this climate system in the past. In terrestrial environments, reconstructions of East Asian monsoon dynamics have primarily been studied using cave speleothem records (Wang et al., Reference Wang, Cheng, Edwards, An, Wu, Shen and Dorale2001; Cheng et al., Reference Cheng, Edwards, Sinha, Spötl, Yi, Chen and Kelly2016; Cai et al., Reference Cai, Cheng, Cheng, Shi, Lu, Ma and Pérez-Mejías2023) and loess–paleosol sequences. In particular, the thick deposits of the Chinese Loess Plateau have been the subject of the most intensive studies into the East Asian monsoon dynamics (Kukla and An, Reference Kukla and An1989; Fang et al., Reference Fang, Ono, Fukusawa, Bao-Tian, Li, Dong-Hong, Oi, Tsukamoto, Torii and Mishima1999; An et al., Reference An, Kutzbach, Prell and Porter2001; Porter, Reference Porter2001; Sun et al., Reference Sun, Clemens, An and Yu2006; Liu et al., Reference Liu, Yang, Sun and Wang2011). Loess–paleosol sequences offer the advantage of (semi-)continuously recording the monsoon intensity over multiple glacial–interglacial alternations, as dust generally accumulates during the glacial periods, and soils form during warmer and more humid interglacial periods (Kukla and An, Reference Kukla and An1989; Ding et al., Reference Ding, Yu, Rutter and Liu1994; Fang et al., Reference Fang, Ono, Fukusawa, Bao-Tian, Li, Dong-Hong, Oi, Tsukamoto, Torii and Mishima1999; An et al., Reference An, Kutzbach, Prell and Porter2001; Porter, Reference Porter2001; Wang et al., Reference Wang, Wu, Li, Liu, Chang, Ding and Wu2008; Liu et al., Reference Liu, Yang, Sun and Wang2011; Lu et al., Reference Lu, Wang, Wang, Zhang, Yi, Wang, Stevens, Kurbanov and Marković2022). Various qualitative physical and geochemical proxies are used to reconstruct the variability and strength of the East Asian monsoon. Grain-size distributions reflect the energy of the winds that transport dust particles and can be used as a proxy for wind strength, thereby indicating phases of and transition to comparatively drier, windier winter monsoon conditions (reviewed in Lu et al., Reference Lu, Wang, Wang, Zhang, Yi, Wang, Stevens, Kurbanov and Marković2022). Magnetic susceptibility provides an indication of pedogenic processes and weathering intensity (Maher and Thompson, Reference Maher and Thompson1995; Maher, Reference Maher1998, Reference Maher2011, Reference Maher2016), whereas mineralogy and elemental geochemistry can provide additional information for past climate reconstruction. Dolomite and calcite concentrations indicate humidity and pedogenesis intensities, while elemental concentrations can serve as a proxy for weathering (e.g., Aquino et al., Reference Aquino, Scardia, Prud’homme, Dave, Lezzerini, Johansson, Marquer, Safaraliev, Lauer and Fitzsimmons2024). Variation in elemental geochemical parameters like Rb/Sr ratio can reflect precipitation fluctuations of the East Asian monsoonal system (Derbyshire et al., Reference Derbyshire, Kemp and Meng1995; Meng et al., Reference Meng, Liu, Wang, Balsam, Chen and Ji2018). On the other hand, quantitative reconstruction of past climate parameters such as temperature, humidity, and vegetation assemblages have been obtained using isotope ratios of carbon and oxygen in pedogenic carbonates (Zhang et al., Reference Zhang, Zhao, Lu and Faiia2003, Reference Zhang, Li, Yan and An2018; Rao et al., Reference Rao, Zhu, Chen and Zhang2006; Luo et al., Reference Luo, Wang, An, Zhang and Liu2020; Prud’homme et al., Reference Prud’homme, Scardia, Vonhof, Guinoiseau, Nigmatova, Fiebig, Gerdes, Janssen and Fitzsimmons2021; Da et al., Reference Da, Breecker, Li, Li, Lu and Ji2023a, Reference Da, Li and Ji2023b) and a range of organic compound concentrations (Zech et al., Reference Zech, Gao, Tarozo and Huang2012; De Jonge et al., Reference De Jonge, Hopmans, Zell, Kim, Schouten and Sinninghe Damsté2014; Schreuder et al., Reference Schreuder, Beets, Prins, Hatté and Peterse2016; Trigui et al., Reference Trigui, Wolf, Sahakyan, Hovakimyan, Sahakyan, Zech, Fuchs, Wolpert, Zech and Faust2019; Zeng and Yang, Reference Zeng and Yang2019; Wang et al., Reference Wang, An, Lu, Zhao and Liu2020).
Pedogenic carbonates are often found in loess–paleosol sequences and are particularly well preserved in paleosols. They are formed from dissolved ions (Ca2+ and HCO₃−) in the soil solution, through either biotic or abiotic processes, and are redistributed in soils via dissolution of soil inorganic carbon pools (Zamanian et al., Reference Zamanian, Pustovoytov and Kuzyakov2016). Therefore, the physical and geochemical characteristics of pedogenic carbonates provide an opportunity to reconstruct climate and potentially its seasonality, as they are well preserved in paleosols from loess profiles (Becze-Deák et al., Reference Becze-Deák, Langohr and Verrecchia1997; Barta, Reference Barta2011; Verrecchia, Reference Verrecchia, Reitner and Thiel2011).
The monsoon-driven Asian climate (wet–dry cycles) facilitates the formation of pedogenic carbonates, and their stable isotope ratios (13C/12C and 18O/16O expressed as δ13Ccarb and δ18Ocarb, respectively) can help us improve our understanding of this climatic system. In particular, δ18Ocarb values have traditionally been used to reconstruct temperature; however, these values also depend on rainfall amount, and the δ18O value of the soil solution (δ18Osw) from which it precipitates (Cerling and Quade, Reference Cerling and Quade1993). Previous investigations of pedogenic carbonates across the Chinese Loess Plateau have shown that modern δ18Osw values are predominantly influenced by the values of recent rainfall (Zhang et al., Reference Zhang, Li, Yan and An2018; Da et al., Reference Da, Li and Ji2023b). Such investigations support the argument that the δ18O of precipitation is the principal parameter controlling δ18Osw in regions influenced by the monsoon system, rather than crystallization temperature. Additionally, the precipitation of pedogenic carbonate in soils typically occurs in isotopic equilibrium with soil CO2, which is derived from root respiration and the microbial decomposition of soil organic matter (Cerling, Reference Cerling1984; Gocke et al., Reference Gocke, Pustovoytov, Kühn, Wiesenberg, Löscher and Kuzyakov2011). The δ13Ccarb values depend on the carbon isotope composition of soil CO2, soil respiration rate, pCO2, and the relative contributions of C3, C4, and CAM (crassulacean acid metabolism) plants (Farquhar et al., Reference Farquhar, Ehleringer and Hubbick1989; Hasinger et al., Reference Hasinger, Spangenberg, Millière, Bindschedler, Cailleau and Verrecchia2015; Rao et al., Reference Rao, Guo, Cao, Shi, Jiang and Li2017). In loess–paleosol sequences, the δ13Ccarb value has been frequently used to reconstruct the influence of monsoon dynamics on soil moisture (Sun et al., Reference Sun, Kutzbach, An, Clemens, Liu, Liu and Liu2015) and vegetation cover (Wang and Greenberg, Reference Wang and Greenberg2007).
In this study, we investigate the potential of calcified root cells (CRCs) as a paleoenvironmental proxy. CRCs are microscale carbonate that are well preserved in Chinese loess–paleosol sequences. They are mainly composed (90%) of low magnesium calcite (Verrecchia and Trombino, Reference Verrecchia and Trombino2021). Calcite crystals range from 60 to 80 µm and retain the structure of the original root cortex cells (Jaillard et al., Reference Jaillard, Guyon and Maurin1991). These calcifications, also known as root petrification (Klappa, Reference Klappa1980) or microcodium (Li and Li, Reference Li and Li2014; Zhang et al., Reference Zhang, Li, Yan and An2018), are present in calcareous sites associated with contrasting hydrological regimes. The calcification process likely protects roots against high concentrations of calcium and bicarbonate ions in the soil solution (Košir, Reference Košir2004) and might also be a strategy for acquiring mineral nutrients (Hinsinger, Reference Hinsinger1998; Košir, Reference Košir2004). Cortical cells calcify during root development, whereas the stele (central part of the plant’s root) might be preserved but never calcifies (Jaillard, Reference Jaillard1992; Lambers et al., Reference Lambers, Mougel, Jaillard and Hinsinger2009). Calcium carbonate in the soil matrix is dissolved through an exchange between protons (H+) and bicarbonate ions (HCO3−) and the release of organic acids (Becze-Deák et al., Reference Becze-Deák, Langohr and Verrecchia1997). Available Ca2+ absorbed by the root accumulates in the cell vacuole, gradually precipitating as calcium carbonate within the cells (Jaillard et al., Reference Jaillard, Guyon and Maurin1991; Becze-Deák et al., Reference Becze-Deák, Langohr and Verrecchia1997). This process results in the calcification of cortical cells by authigenic calcite, creating CRCs (Jaillard et al., Reference Jaillard, Guyon and Maurin1991; Becze-Deák et al., Reference Becze-Deák, Langohr and Verrecchia1997; Lambers et al., Reference Lambers, Mougel, Jaillard and Hinsinger2009; Li and Li, Reference Li and Li2014). Monitoring experiments have shown that the first calcite crystals appear within 24 hours (Jaillard, Reference Jaillard1992).
CRCs form mostly in living plants and are likely to be synchronous with the sedimentary matrix, making them a strategic target for reconstructing paleoclimatic conditions. To understand the paleoenvironmental significance of CRC concentrations and their use as paleoclimatic proxies, we analyzed CRCs extracted from a loess–paleosol profile, Fanshan, located northeast of the Chinese Loess Plateau, in the Yanqin-Huailai-Zhoulu Basin (Xiong et al., Reference Xiong, Ding and Liu2001; Wang et al., Reference Wang, Løvlie, Chen, Yang, Pei and Tang2014; Li et al., Reference Li, Song, Fitzsimmons, Dave, Liu, Zong, Sun, Liu and Orozbaev2022). This region is likely to be more susceptible to variations in the East Asian monsoon than sites farther south (Yang et al., Reference Yang, Ding, Li, Wang, Jiang and Huang2015), a fact that should be reflected in the variability of CRC concentrations and stable isotope ratios of CRCs. We present the first continuous record of CRC concentrations and stable isotope ratios down the stratigraphy of the Fanshan loess–paleosol sequence. We compare variations in CRC concentrations with grain-size proxies to explore how environmental conditions in the rhizosphere responded to past climate changes. Additionally, we investigate the oxygen and carbon isotope signatures of carbonates of CRCs, as well as the δ13Corg of organic matter preserved in the CRCs, as proxies for past hydroclimatic conditions and seasonality, with a particular focus on reconstructing the variability in East Asian Monsoon intensity. Our study evaluates the potential of CRCs as a paleoenvironmental proxy that can capture both short-term seasonal contrasts and the long-term effects of monsoon climate systems.
Materials and Methods
Study site
The site of Fanshan is located approximately 50 km northwest of Beijing and 400 km northeast of the Chinese Loess Plateau (40.1963°N, 115.3975°E), at an elevation of 850 m above sea level within the Yanqin-Huailai-Zhoulu basin fault (Xiong et al., Reference Xiong, Ding and Liu2001; Li et al., Reference Li, Song, Fitzsimmons, Dave, Liu, Zong, Sun, Liu and Orozbaev2022). This region is ideal for investigating seasonal and long-term land–climate feedback due to its current cold, semiarid climate, which is driven by the interplay between the EASM and the EAWM associated with the Siberian High pressure system (Xiong et al., Reference Xiong, Ding and Liu2001). The impact of the westerlies is limited in this region (Kong et al., Reference Kong, Wang, Li and Pang2019; Li et al., Reference Li, Song, Fitzsimmons, Dave, Liu, Zong, Sun, Liu and Orozbaev2022). The local climate is influenced by the interaction between the EASM, which brings precipitation mainly in the summer from June to September, and the EAWM associated with the Siberian High pressure system, which exhibits a prevailing southeastward wind direction, bringing in cold and dry air from the northwest (Kong et al., Reference Kong, Wang, Li and Pang2019). The monthly average temperature ranges from −9.8°C in January to 23.5°C in August, and the average annual rainfall is ∼450 mm.
Stratigraphy, chronology, and grain size
The Fanshan profile is composed of two sections with a total thickness of 15 m that are located 40 m apart. A detailed description of the entire profile can be found in Li et al. (Reference Li, Song, Fitzsimmons, Dave, Liu, Zong, Sun, Liu and Orozbaev2022). Essentially, the profile consists of Section A, which extends from the surface to a depth of 7 m, and Section B, which continues the stratigraphic sequence from 7 to 15 m depth. Section B is composed of three horizons. A homogeneous loess unit with sparsely distributed carbonate nodules lies at the base, from 15 to 13.4 m. In this unit, the visible carbonate content increases upward toward the base of a strongly developed paleosol. This paleosol is observed between 13.4 and 11.2 m depths and is characterized by a high concentration of pedogenic carbonates. The upper part, from 11.2 to 7 m, consists of a second homogeneous loess unit characterized by a noticeable decrease in visible carbonates. The chronological framework of this sequence is based on optically stimulated luminescence (OSL) ages of four samples previously published by Li et al. (Reference Li, Song, Fitzsimmons, Dave, Liu, Zong, Sun, Liu and Orozbaev2022), which were collected during the same field campaign. As a result, there is no uncertainty regarding the stratigraphic position of the samples relative to our sequence.
To evaluate and compare the paleoenvironmental conditions spanning the entire transition across two loess accumulation units and the intercalating paleosol formation, we focused on the lower ∼4 m, from 10.6 to 14.6 m of the 8-m-thick Section B (Fig. 1). Continuous columns of samples were collected at a resolution of 10 cm down the Fanshan profile. Sampling was conducted on a freshly exposed section of the profile, obtained by cleaning back about 50 cm of overlying sediments, to ensure a clean surface unaffected by present-day vegetation and downward wash of loose sediment. CRCs were sampled continuously at 10-cm resolution by collecting 10 L of sediment in buckets. These were wet sieved in the field using a 0.5-mm-diameter sieve.
Sub–climate system in Asia and stratigraphy of Fanshan loess profile. (a) Main climate subsystems across Asia. Redpoint: Fanshan (FS). Blue arrows represent the northerly component of surface winds associated with the Siberian High (Sib.H; Persoiu et al., Reference Persoiu, Ionita and Weiss2019). Green arrows represent the westerlies, the Indian monsoon and the East Asian summer monsoon (EASM). The map was constructed using Generic Mapping Tools with the color map indicating the elevation (Wessel et al., Reference Wessel, Smith, Scharroo, Luis and Wobbe2013). (b) Lithostratigraphy of Fanshan profile II (FS 18 II) modified from Li et al. (Reference Li, Song, Fitzsimmons, Dave, Liu, Zong, Sun, Liu and Orozbaev2022). White lines: calcified root cells (CRCs), brown circle: carbonate nodules; the green stars from the top 4 m of the profile represent optically stimulated luminescence (OSL) ages on coarse-grain (6–90 µm) quartz, while the red star at the bottom of the profile represents a post-IR IR stimulated luminescence (pIRIR) age on coarse-grain (63–90 µm) K-feldspar (data from Li et al., Reference Li, Song, Fitzsimmons, Dave, Liu, Zong, Sun, Liu and Orozbaev2022).
OSL dating of the paleosol yielded ages spanning 135 ± 7 ka (at 14.0 m depth) and 62.5 ± 4.1 ka (11 m) (Fig. 1). These ages indicate that the paleosol formed during Marine Isotope Stage 5 (MIS 5) and may extend to MIS 4. The lower primary loess unit most likely accumulated before the MIS 5, probably during MIS 6. The upper primary loess layer, spanning ca. 11–7.5 m depth, dates to an age range of 62.5 ± 4.1 ka to 27.4 ± 2.9 ka (Fig. 1), indicating deposition during the glacial MIS 4, and possibly through the interglacial MIS 3.
The grain-size data of the entire profile at 2 cm resolution has been published by Li et al. (Reference Li, Song, Fitzsimmons, Dave, Liu, Zong, Sun, Liu and Orozbaev2022). Here, we make use of the grain-size data from the paleosol at 10.6 to 14.6 m depth. We calculated the Grain Size Index (GSI) (Antoine et al., Reference Antoine, Rousseau, Moine, Kunesch, Hatté, Lang, Tissoux and Zöller2009), defined as the ratio between the 52–26 µm (coarse silt) and the 0–26 µm (fine silt and clay) fractions. We also reprocessed the dataset using endmember modeling analysis (EMMA) following published methods (Dietze et al., Reference Dietze, Hartmann, Diekmann, IJmker, Lehmkuhl, Opitz, Stauch, Wünnemann and Borchers2012; Dietze and Dietze, Reference Dietze and Dietze2019), as grain-size distributions can provide key information on wind dynamics such as provenance and mechanisms of transport (Antoine et al., Reference Antoine, Rousseau, Moine, Kunesch, Hatté, Lang, Tissoux and Zöller2009; Vandenberghe, Reference Vandenberghe2013; Újvári et al., Reference Újvári, Kok, Varga and Kovács2016).
CRCs morphology and quantification
The Fanshan loess profile was sampled at 10 cm depth intervals for extracting CRCs. We quantified the number of CRCs with a diameter greater than 0.5 mm, manually extracting them under an optical microscope for each interval. Based on the established chronological framework, each 10 cm layer represents an average temporal resolution of approximately 240 years. This represents the first continuous CRC counting down a loess profile and enables us to assess whether concentration variations reflect environmental changes and how these may relate to stable isotope signals. The crystal structure and morphology of CRCs were analyzed using a Tescan Mira LMU scanning electron microscope (SEM) at the University of Lausanne. We imaged 11 CRCs selected from the depths between 11.00 and 14.60 m to obtain a representative overview down the length of the profile investigated in this study (Fig. 2).
Scanning electron microscope (SEM) images: (a) calcified root cell (CRC) surface; (b) close-up of the cells showing cell shapes totally replaced by the large crystallized vacuole in which calcium carbonate precipitated; (c) calcification of the cells and the intercellular space; (d) presence of needle fiber calcite of fungal origin at the CRC surface; (e and f) traces of dissolution of calcified cells. The perfectly preserved pattern of the calcite lattice demonstrates the common monocrystalline nature of CRC.
Stable isotope analyses of CRC
Carbon and oxygen isotope ratios of the CRC carbonate fraction were analyzed in two laboratories using different strategies. At the Stable Isotope Laboratory of the Max Planck Institute of Chemistry in Mainz, we selected 11 sampled levels, each level being 10 cm thick. We extracted 30 CRCs from each level and analyzed individual CRCs, thus analyzing a total of 330 individual CRCs. At the Stable Isotope Laboratory at the University of Lausanne, we decided to individually analyze 5 CRCs from each of the 41 sampled levels, for a total of 205 analyses. We analyzed individual CRCs to preserve natural variability and capture microscale environmental signals that bulk averaging would not identify. This approach also allows us to detect intra-annual isotopic variations and identify outliers or clusters that may reflect distinct climatic events. While average values are reported for comparison, analyzing distributions ensures these values are properly contextualized and representative.
Sample preparation and analyses followed the same protocols at both facilities. CRCs were extracted following cleaning in an ultrasonic bath with deionized water (2 × 30 seconds), to ensure well-preserved crystalline surfaces and absence of impurities. Before analysis, CRCs were crushed, and aliquots comprising 200 and 300 µg of carbonate powder were digested with phosphoric acid at 70°C in an exetainer vial, then flushed with helium gas in a Thermo Fisher Scientific GasBench II preparation device. The liberated CO2 was analyzed in a Thermo Fisher Scientific Delta V Advantage isotope ratio mass spectrometer at the Max Planck Institute of Chemistry in Mainz and a Delta V Plus isotope ratio mass spectrometer at the University of Lausanne. The long-term standard deviation of a routinely analyzed in-house CaCO3 standard was <0.1‰ (1σ) for both carbon and oxygen isotope ratios in both laboratories. Calibration of the in-house carbonate standard and assessment of the accuracy and reproducibility of the δ13C and δ18O values were undertaken by replicate analysis of international reference materials NBS19, NBS18, and IAEA 603. The δ13Ccarb and δ18Ocarb values are reported as per mil (‰) deviations relative to the Vienna Pee Dee Belemnite limestone standard (VPDB).
The carbon isotope composition of organic matter preserved in the CRCs (δ13Corg) was measured at the University of Lausanne. For each analysis, between 100 and 1000 mg of CRCs were used, so we analyzed one sample each per 10 cm. Samples were decarbonated using 10% HCl. Additionally, three samples were selected to extract the lipids and were cleaned with methanol and dichloromethane to determine whether the presence of lipids could influence the measured δ13Corg values in our dataset. The δ13Corg values were determined using an elemental analyzer (Carlo Erba 1108 EA) connected to a Delta V Plus isotope ratio mass spectrometer.
We calculated the isotopic difference between the inorganic and organic carbon in the CRCs (∆13Ccarb-org) using Equation 1 (Cerling and Quade, Reference Cerling and Quade1993):
\begin{equation}{{{\Delta }}^{{\text{13}}}}{{\text{C}}_{{\text{carb - org}}}}\,{\text{ = }}\,{{{\delta }}^{{\text{13}}}}{{\text{C}}_{{\text{carb}}}}\, - \,{{{\delta }}^{{\text{13}}}}{{\text{C}}_{{\text{org}}}}{\text{ }}\end{equation}where δ13Ccarb is the carbon isotope ratio of the inorganic fraction in CRCs, and δ13Corg is the carbon isotope ratio the organic matter preserved in the CRCs.
RESULTS
GSI and grain-size distribution
The variations in CRC quantities down the stratigraphy were compared with clay content (<8 µm) and GSI values (ratio between coarse silt and fine silt + clay). Clay content is influenced by the weathering of loess during soil formation, with higher clay percentages generally associated with the development of paleosols. The clay content varies between 9.7% during loess accumulation and 25.1% during soil formation (Fig. 3).
Variability of calcified root cell (CRC) abundance, stratigraphy, and grain-size analysis: grain-size distribution and end-member modeling results. (a) CRC amount, clay content (<4.5 µm), and Grain-Size Index (GSI) variations along the stratigraphy with the proportional contribution of modeled endmembers. (b) Endmember modeling of grain-size distributions, showing three dominant components: EM1 (coarse silt, ∼28 µm), EM2 (fine sand, ∼79 µm), and EM3 (clay and fine–medium sand, ∼4.5 and 126 µm).
The GSI ranges from 0.3 in the paleosol and 0.7 in the loess (Fig. 3). The GSI is sensitive to aeolian dynamics, with high GSI values implying strong winds during the deposition of loess units, and low GSI values indicating a decrease in dust accumulation which correlates with paleosols.
The grain-size distribution of the sequence is polymodal. While Li et al. (Reference Li, Song, Fitzsimmons, Dave, Liu, Zong, Sun, Liu and Orozbaev2022) applied a three-endmember model using the algorithm developed by Paterson and Heslop (Reference Paterson and Heslop2015), we used the MATLAB-based program of Dietze et al. (Reference Dietze, Hartmann, Diekmann, IJmker, Lehmkuhl, Opitz, Stauch, Wünnemann and Borchers2012). For our analysis, we selected samples from 10.6 to 14.5 m depth and restricted the grain-size range to 0.250–200 µm to minimize the large number of zeros that can cause errors in the MATLAB program. EM1 is unimodal, with a modal peak at 28.3 µm (medium silt), accounting for 31.4% of the dataset variance. EM2, explaining 37.8% of the variance, is unimodal, with a modal peak at 79.6 µm (fine sand). EM3 is bimodal, with peaks at 4.5 µm (clay) and 126.2 µm (sand), explaining 30.8% of the variance. The EMMA-based distribution of grain-size classes along the stratigraphic sequence is shown in Figure 3.
CRCs morphology and quantification
SEM images (Fig. 2) of the CRCs display a well-crystallized structure, clearly indicating root cells of ancient graminoids (Becze-Deak et al., Reference Becze-Deák, Langohr and Verrecchia1997). The cells of the calcified roots are generally well preserved, with calcification also present in the intercellular spaces. The morphology suggests that they are rootlets associated with a system of near-surface roots. The well-preserved crystals observed under the SEM and microscope suggest that the CRCs were not significantly affected by diagenesis. In addition, the preservation of CRCs after burial indicates that leaching was likely to be minimal (Becze-Deák et al., Reference Becze-Deák, Langohr and Verrecchia1997).
Continuous counting of CRCs down the profile reveals distinct variability between primary loess and paleosol units (Fig. 3). CRC abundance follows the stratigraphy, with the highest concentrations (up to 4 g/10 L) occurring in the paleosol, while the overlying and underlying loess units contain markedly lower amounts (0.012–0.454 g/10 L).
Stable isotope composition of CRCs
The CRCs of the entire section, as analyzed at the University of Lausanne, exhibit δ13Ccarb values ranging from −22.6 ± 3.0‰ to −16.0 ± 3.6‰, with a mean of −19.0 ± 2.7‰, and δ18Ocarb values ranging from −14.6 ± 1.4‰ to −11.7 ± 0.9‰, with an average of −13.6 ± 1.4‰. The δ13Ccarb values analyzed in Max Planck Institute for Chemistry yield a mean value of −18.9 ± 2.3‰, ranging from −19.9 ± 1.7‰ to −17.9 ± 2.7‰. The δ18Ocarb values yield a mean value of −13.5 ± 1.3‰, ranging from −14.1 ± 1.3‰ to −13.0 ± 1.0‰. The δ13Corg values range from −27.6‰ to −25.5‰ with a mean value of −26.3 ± 0.5‰. The three additional samples selected for lipid extraction showed no significant difference, indicating that lipids do not affect the δ13Corg composition of these samples. Mean and standard deviation values of each layer are compiled in Table 1. The ∆13Ccarb-org values range from 3.6‰ to 9.9‰ down the stratigraphic section, with a mean of 7.1 ± 1.6‰ (Fig. 4).
Variation of the stable isotopic composition of calcified root cells (CRCs) through the development of the Marine Isotope Stage 5 (MIS 5) paleosol. Lithostratigraphy with two optically stimulated luminescence (OSL) dates along with the amount of CRC (mg/10 L), δ13Ccarb, δ13Corg preserved in the CRC, ∆13Ccarb-org (difference between the δ13Ccarb and the δ13Corg), and δ18Ocarb values. The gray envelopes represent the deviation between the minimum and maximum values for δ13Ccarb and δ18Ocarb performed at University of Lausanne (UNIL); the black bullets with their whiskers refer to the mean and 1 SD (1σ) of 30 CRCs from each sample, performed at Max Planck Institute for Chemistry (MPIC). The red lines represent the moving average based on the UNIL dataset (δ13Ccarb, δ13Corg, and δ18Ocarb). Intervals 1–3 are described in the text.
Calcified root cell (CRC) counts and mean values of stable isotopic compositions of the CRCs (δ13Ccarb, δ13Corg and δ18Ocarb) extracted from Fanshan profile II along the depth (cm).
Based on the stratigraphy, CRC concentration, and stable isotopic composition, we identified three intervals within the targeted depth range of ∼10.6 to 14.6 m. We find that Interval 1 (13.15 to 14.55 m) is associated with the transition between the primary loess unit, which accumulated during MIS 6, and the beginning of the interglacial paleosol unit. This interval is characterized by low CRC concentration and high ∆13Ccarb-org values with a mean of 8.0 ± 2.2‰. Interval 2 (12.05 and 13.05 m) falls within the paleosol unit and is defined by high CRC concentration and low δ13Ccarb and ∆13Ccarb-org values (5.9 ± 1.1‰). Finally, Interval 3 (10.55 to 11.95 m) is associated with the transition between the paleosol and the overlying loess, which accumulated during MIS 4 and MIS 3. This phase is defined by low CRC concentration and high ∆13Ccarb-org values with a mean of 7.0 ± 1.3‰. Based on a nonparametric statistical test (Mann-Whitney U-test), the mean of ∆13Ccarb-org of Intervals 1 and 3 and Intervals 2 and 3 are not statistically different (P values = 0.0769 and 0.0846), whereas significant differences are observed between intervals 1 and 2 (P value = 0.0028).
DISCUSSION
Variability in CRC concentration as a proxy for paleoenvironmental change through time
Experimental observations indicate that, under optimal conditions, CRCs can form within a few days during plant growth phases (Jaillard et al., Reference Jaillard, Guyon and Maurin1991; Lambers et al., Reference Lambers, Mougel, Jaillard and Hinsinger2009). However, root calcification may not only occur during the plant’s lifetime but also after its death, when decay processes can result in calcite precipitation within the inner parts of the cells (Klappa, Reference Klappa1980; Barta, Reference Barta2011). On glacial–interglacial timescales, it can be reasonably assumed that CRCs precipitated within loess deposits or paleosols and are broadly contemporaneous with the deposition of the surrounding sediment. It is worth noting that several studies that undertook radiocarbon dating of CRCs directly yielded CRC ages younger than those of the surrounding sediment (Pustovoytov and Terhorst, Reference Pustovoytov and Terhorst2004; Luo et al., Reference Luo, Wang, An, Zhang and Liu2020; Reiss et al., Reference Reiss, Stüwe, Einwögerer, Händel, Maier, Meng, Pasda, Simon, Zolitschka and Mayr2022). However, those samples were collected within only 3 m of the surface and may have been contaminated with modern sources of carbon. A more recent study on fossil CRCs taken at depth from a Chinese loess sequence dated from the last glacial period obtained agreement between 14C dates on CRCs and OSL ages from the corresponding sediments (Luo et al., Reference Luo, Liu and An2024), suggesting that older fossil CRCs are less likely to be significantly contaminated by younger carbon and are effectively coeval in age with the surrounding sediment. The Fanshan loess–paleosol profile lies beyond the limits of radiocarbon dating; therefore, we cannot directly compare the 14C and OSL ages of the CRCs. Despite this limitation, the variability in the CRC concentration correlates well with the observed stratigraphy, exhibiting low concentrations in the loess units and high concentrations in the paleosol, and this correlation suggests contemporaneous formation of CRCs with sediment deposition.
Precipitation of CRCs requires pronounced seasonality, characterized by alternating wet and dry periods. Their presence indicates a relatively short vegetation-growing season under an arid conditions marked by strong hydroclimatic fluctuations, dominated by herbaceous plant cover (Jaillard et al., Reference Jaillard, Guyon and Maurin1991; Becze-Deák et al., Reference Becze-Deák, Langohr and Verrecchia1997). The gradual increase in CRCs observed at the base of the interglacial MIS 5 paleosol suggests that the environmental conditions were favorable for the expansion of grass-like vegetation, driven by seasonal contrasts in moisture availability. By contrast, the markedly lower CRC concentrations in the overlying and underlying primary loess units (MIS 6 and MIS 4, respectively) point to reduced vegetation activity and a likely decrease in plant cover during these colder climatic phases (Fig. 3). Furthermore, the shift in EM between the two loess units indicates changes in dust provenance and/or wind strength, with the younger loess horizons characterized by coarser grain sizes.
Within the paleosol, internal variability in CRC concentrations aligns with grain-size variability. Clay content increases, while GSI decreases, both reaching their maxima at 12.7 m, overlain by a reversal of this trend toward the top of the paleosol at 11.2 m, suggesting several episodes of pedogenesis. This shift is also reflected in the endmember modeling, where at 12.7 m the fine sand component (EM2) dominates over the medium silt (EM1), while the clay-rich fraction (EM3) remains present. This change suggests a transition to stronger winds capable of transporting coarser material. The persistence of the clay fraction, however, indicates that background suspension settling continued, confirming the gradual development of the paleosol unit. The association of CRC concentration with grain-size variability and stratigraphy suggests that environmental changes within the rhizosphere are most likely driven by fluctuations in moisture availability and atmospheric humidity, which may reflect variations in EASM intensity. Likewise, other studies of MIS 5 sediment characteristics in both loess and speleothems in China suggest environmental variability during the MIS 5 interglacial (Rao et al., Reference Rao, Liu, Hua, Gao and Chen2015; Zhang et al., Reference Zhang, Yang, Zhang, Zhang, Yan, Zuli and Wang2021).
Hydroclimatic insights from δ1⁸Ocarb values of CRCs in Fanshan
While CRC abundance reflects stratigraphic and moisture-related variability, we hypothesize that the stable isotopic ratios provide complementary information on climate and, more precisely, hydroclimate and/or seasonality. To understand how much information can be extracted, we examine the δ18Ocarb and δ13Ccarb values of CRCs from the Fanshan profile and compare these with published records from other sites in Eurasia, as relatively few studies to date have investigated CRC isotopic variability (Fig. 5). We observe that the δ18Ocarb values of CRCs display a continental gradient spanning Eurasia, with progressively depleted values from west (Mediterranean region) to east (China). This pattern follows the expected Rayleigh distillation process, whereby air masses progressively lose heavier isotopes as they move inland and undergo successive rainout events. The west–east gradient of δ18Ocarb in CRCs suggests that their isotopic signatures are sensitive to large-scale atmospheric circulation patterns, making them useful for reconstructing past hydroclimate. The δ18Ocarb values from the easternmost site at Fanshan are the lowest, with a mean value of −13.6 ± 1.4‰ (Fig. 5). The standard deviation of δ18Ocarb values of CRCs at Fanshan ranges between 0.3‰ and 2.5‰ (1σ), indicating isotopic variability within a 10 cm layer. Although average δ18O values from the MIS 5 paleosol and the MIS 6 and 4–3 loess horizons are not statistically different, extreme values do indicate differences (Fig. 4). This observation suggests that while hydroclimatic conditions were relatively stable throughout MIS 5, MIS 6, and MIS 4–3, the frequency or intensity of extreme events—such as particularly wet or dry seasons or changes in precipitation source—varied among these intervals. Assuming that variability within the layers may reflect seasonal differences, we estimated the δ18O value of precipitation at mean temperatures of 20°C and 0°C following the equation of Kim and O’Neil (Reference Kim and O’Neil1997). The 0°C case is presented solely to evaluate the lower extreme of possible δ1⁸O values, whereas 20°C represents a more realistic warm-season formation temperature. For a mean temperature of 20°C, we obtained a mean δ18O value of −12.3 ± 0.6‰, with a range of −16.5 to −9.3‰. At 0°C, the mean δ18O value was −16.8 ± 0.6‰, ranging from −20.9 to −13.8‰.
Comparison of stable isotopic signatures of calcified root cells (CRCs) from different environments: oxygen and carbon stable isotope composition of CRC extracted from Fanshan (this study) analyzed at (1) Max Planck Institute for Chemistry (MPIC) and (2) University of Lausanne (UNIL); (3) data from Marine Isotope Stage 2 (MIS 2) Xifeng loess profile (Luo et al., Reference Luo, Wang, An, Zhang and Liu2020); (4) modern CRC from southern France (Morin, Reference Morin1993); (5) Suttö loess profile in Hungary deposited between MIS 6 and MIS 2 (Koeniger et al., Reference Koeniger, Barta, Thiel, Bajnóczi, Novothny, Horváth, Techmer and Frechen2014); (6) Skala Podil’ska loess profile in Ukraine dated from MIS 2 (Łacka et al., Reference Łacka, Łanczont and Madeyska2009); and (7) Kammern-Grubgraben loess profile in Austria deposed during MIS 2 (Reiss et al., Reference Reiss, Stüwe, Einwögerer, Händel, Maier, Meng, Pasda, Simon, Zolitschka and Mayr2022).
At present, the Fanshan region is affected by seasonal variations, with moisture deriving from the EASM and minimal impact from the midlatitude westerlies (Kong et al., Reference Kong, Wang, Li and Pang2019). We compared our values with a compilation of Global Network Isotopes in Precipitation (GNIP) data from two sites (Tianjin and Batou) and from the CHNIP Station in Beijing (Table 2; Liu et al., Reference Liu, Song, Yuan, Sun and Yang2014). Monthly values of δ18O precipitation range from −17.3‰ (January) to −5.5‰ (May), with a mean of −9.9 ± 3 .6‰ and temperatures between −7.1 ± 4.0°C (January) and 23.9 ± 4.8°C (July). Today, the δ18O values of precipitation reflect the EASM source, with values starting to decrease in May, coinciding with the onset of the monsoon (Table 2). The δ1⁸O value of precipitation estimated from the average value of δ1⁸OCRC is approximately 5‰ lower than today’s value during the main vegetation growing season (May to September). This value is even more negative than the estimated δ1⁸O for the soil water from last glacial loess in the Chinese Loess Plateau (between −12 and −6‰) (Luo et al., Reference Luo, Liu and An2024). The measured 18O depletion (i.e., lower δ1⁸O values) suggests an intensification of the EASM during the period from MIS 6 through MIS 5 and into MIS 4, compared with the present, which is consistent with a stronger EASM producing lower rainwater δ1⁸O values (Dansgaard, Reference Dansgaard1964). We suggest that CRC formation would have occurred during the seasons most influenced by rainfall seasonality, driven by the monsoon circulation that strengthened northward over the summer months, coinciding with weakening of the northerlies (Siberian High) (Chiang et al., Reference Chiang, Kong, Wu and Battisti2020).
Average value and standard deviation of δ18O of precipitation (δ18Op) and monthly temperature of Global Network Isotopes in Precipitation (GNIP) (Tianjin, Batou) and CHNIP stations (Beijing) (Liu et al., Reference Liu, Song, Yuan, Sun and Yang2014).
Insight into the carbon cycle from the δ13Ccarb values of CRCs
The mean δ13Ccarb values of CRCs in existing records range from ∼ −24‰ in Austria to ∼ −10‰ in southern France and Hungary, indicating significant regional variability in isotopic composition. These variations likely reflect differences in vegetation types, soil productivity, and local hydroclimatic conditions.
The δ13Ccarb values of CRCs extracted from the Fanshan loess profile are surprisingly low, with a mean of −18.9 ± 2.4‰, compared with typical values reported for pedogenic carbonates in soils and paleosols (Boguckyj et al., Reference Boguckyj, Łanczont, Łacka, Madeyska and Zawidzki2006; Łacka et al., Reference Łacka, Łanczont and Madeyska2009; Koeniger et al., Reference Koeniger, Barta, Thiel, Bajnóczi, Novothny, Horváth, Techmer and Frechen2014; Luo et al., Reference Luo, Liu and An2024, Reference Luo, Wang, An, Zhang and Liu2020). This offset suggests that specific formation mechanisms may influence the incorporation of soil-derived CO2 into CRCs. To further constrain the origin of these 13C-depleted carbonates, we measured, for the first time, the δ13Corg preserved within the CRCs. The resulting δ13Corg values (−26.3 ± 0.5‰) are consistent with the isotopic signature of C₃ vegetation, indicating that C₃ taxa dominated the vegetation cover at the time of CRC formation. The mean isotopic difference between δ13Ccarb and δ13Corg (∆13Ccarb-org) is 7.1 ± 1.6‰, which indicates high discrimination against 13C that deviates substantially from the 14–17‰ isotopic difference predicted by the classical 1D production–diffusion model for soil-respired CO₂ in pedogenic carbonates (Cerling and Quade, Reference Cerling and Quade1993; Nordt et al., Reference Nordt, Hallmark, Wilding and Boutton1998; Cerling, Reference Cerling, Thiry and Simon-Coinçon1999). Comparable values have been reported from Xifeng on the Chinese Loess Plateau, where the difference between the δ13Ccarb of CRC and the δ13C of the soil organic matter varied between −8.0 and −6.8‰ (Luo et al., Reference Luo, Wang, An, Zhang and Liu2020, Reference Luo, Liu and An2024). Such consistently low ∆13Ccarb-org values strongly suggest that kinetic isotope effects play a non-negligible role during CRC precipitation.
We infer that the primary carbon source for CRCs at Fanshan derives from CO₂ released by C₃ plant respiration and microbial decomposition of organic matter, with negligible contribution from primary sedimentary carbonates or secondary soil carbonates. Several mechanisms may account for the relatively low isotopic signatures. Previous studies have proposed that mineralization may occur before the complete diffusion of dissolved CO₂ in plant cells, thereby eliminating the CO2 diffusion of 4.4‰ between 12CO2 and 13CO2 in the soil pores (Cerling, Reference Cerling1984; Luo et al., Reference Luo, Wang, An, Zhang and Liu2020). Alternatively, CRC precipitation may incorporate 13C-depleted organic compounds (Łacka et al., Reference Łacka, Łanczont and Madeyska2009). In addition, the rhizosphere-respired CO2 and CO2 released from microbial decomposition of soil organic matter might also have influenced the isotopic composition of CRCs during the growing season (Wang and Greenberg, Reference Wang and Greenberg2007). During microbial respiration, the preferential 13C depletion in respired CO2 is balanced by 13C enrichment in the microbial biomass. Therefore, the CO2 that diffuses out has lower δ13C compared with the organic matter.
At Fanshan, we assume that the low ∆13Ccarb-org implies that CRC formation occurred under quasi–closed system conditions, with soil organic matter serving the main carbon reservoir during root respiration. Furthermore, the enhancement of biological activity associated with denser vegetation cover, as evidenced by paleosol development and a gradual increase in the concentration of CRCs within the paleosol, would have led to more effective organic matter mineralization, resulting in lower δ13C values (Kohn, Reference Kohn2016, Reference Kohn2010; Luo et al., Reference Luo, Liu and An2024). However, we should note that the range of δ13Ccarb values at Fanshan is particularly wide, with standard deviations of 4.4‰ to 0.9‰ observed within a single 10 cm layer. This heterogeneity cannot be explained simply by varying contributions of CO₂ from C₃ and C₄ plants, as the measured δ13Corg values from CRC exclude C₄ plants (all δ13Corg values are lower than −25‰). Instead, this variability may reflect the relative contributions of atmospheric CO2 and soil-respired CO2, the latter being primarily influenced by root and microbial respiration on seasonal timescales. The magnitude of this relative contribution likely depends on climatic conditions, suggesting that δ13Ccarb values were controlled not only by formation mechanisms but also by climate, particularly through the interaction between the EASM and the Siberian High over the MIS 6 to MIS 4–3 period.
The strong seasonality recorded in China has a direct influence on vegetation and the rate of soil respiration, which controls the isotopic composition of carbonate precipitation and the formation of CRCs (Cerling and Quade, Reference Cerling and Quade1993; Quade et al., Reference Quade, Rech, Latorre, Betancourt, Gleeson and Kalin2007). The presence of CRCs has been identified as a proxy for arid environments; however, their stable isotope values and variable concentration suggest that they formed in an environment characterized by strong seasonal contrasts with wet and dry cycles. This seasonality has an impact on soil respiration, which is primarily regulated by soil moisture, resulting in a low rate during the dry season and a high respiration rate during periods of intense rainfall (Licht et al., Reference Licht, Kelson, Bergel, Schauer, Petersen, Capirala, Huntington, Dupont‐Nivet, Win and Aung2022); thus, it also affects the δ13Ccarb values. In an arid environment, the δ13Ccarb values are generally more positive due to an increase in the contribution of atmospheric CO2 (δ13Catm ∼ −8‰) in the soil CO2, resulting in a ∆13Ccarb-org higher than the expected value based on the 1D production–diffusion model (Chen et al., Reference Chen, Cui and Wang2016; Da et al., Reference Da, Zhang, Li and Ji2020). By contrast, when soil moisture is higher, vegetation is growing, and δ13Ccarb values are characterized by lower values (Caves Rugenstein and Chamberlain, Reference Caves Rugenstein and Chamberlain2018).
Variability in the stable isotope composition of CRCs as a proxy for past hydroclimate
Based on the stratigraphy, grain size, CRC concentration, and CRC stable isotopic signatures at Fanshan, we identified three distinct intervals that were likely formed under different interactions between the EASM and the Siberian High. During the accumulation of primary loess units (Intervals 1 and 3), arid and cold conditions prevailed. The high GSI and the EM modeling both indicate strong wind dynamics. Furthermore, the change in the proportions of the grain-size EM 1 toward EM2 dominance in the two loess units indicates a change in dust sources and/or in wind dynamics. During both intervals, the clay content decreases and is associated with low CRC abundances, confirming limited pedogenesis and sparse vegetation cover. Under arid conditions, reduced soil respiration likely increased the contribution of atmospheric CO2 to the soil, leading to higher δ13Ccarb and ∆13Ccarb-org values. In addition, the relatively high δ18Ocarb extremes may reflect enhanced evaporation under arid conditions (Quade et al., Reference Quade, Rech, Latorre, Betancourt, Gleeson and Kalin2007). Taken together, the loess accumulation during MIS 6 (Interval 3) and MIS 4 (Interval 1) was most likely associated with a strengthening of the EAWM associated with the Siberian High.
In contrast, the development of the paleosol (Interval 2) was characterized by increased clay content and a decrease in GSI during the interglacial MIS 5, suggesting wetter conditions during that period, as also observed in the Chinese Loess Plateau (Zhang et al., Reference Zhang, Yang, Zhang, Zhang, Yan, Zuli and Wang2021). The lower GSI reflects reduced wind intensity, while the increase in CRC concentration points to enhanced vegetation cover. Higher biological activity would have promoted root respiration, thereby reducing the relative contribution of atmospheric CO₂ to CRC formation, as soil-respired CO₂ from roots became more effective than during loess accumulation. Overall, the formation of paleosols, supported by multiple proxies, indicates a wetter environment, likely sustained by a strengthening of the EASM, as further suggested by the low δ1⁸O of precipitation estimated from δ18Ocarb values. It is important to note that the stable isotope composition of the CRCs also reflects internal variability within the paleosol, consistent with the patterns highlighted by CRC concentrations and grain-size variations. This observation confirms that the paleosol formed over multiple episodes and was influenced by varying climate and paleoenvironmental conditions.
CONCLUSIONS
This study demonstrates that CRCs serve as valuable paleoenvironmental proxies within loess–paleosol sequences. At Fanshan, their abundance and stable isotopic compositions provide complementary insights into vegetation dynamics, soil processes, and hydroclimatic variability associated with the East Asian monsoon. Variability in CRC concentration reflects changes in stratigraphy, with higher abundances in the MIS 5 paleosol indicating enhanced vegetation activity during wetter conditions, and reduced abundances in the MIS 6 and MIS 4–3 loess units signifying sparse plant cover during colder, drier phases.
Stable isotope data reveal that CRCs capture both regional- and local-scale climatic signals. The δ1⁸Ocarb values show a continental west–east gradient across Eurasia and indicate enhanced EASM conditions at Fanshan compared with the present, aligning with increased rainfall seasonality. The δ13Ccarb and δ13Corg values highlight the prevalence of C₃ vegetation and suggest that quasi–closed system soil respiration processes, influenced by monsoon-driven seasonality, play a crucial role in the formation of CRCs. The consistently low ∆13Ccarb-org values indicate a significant kinetic isotopic effect during carbonate precipitation, further distinguishing CRCs from other pedogenic carbonates. Notably, internal isotopic variability within the MIS 5 paleosol point to multiple episodes of pedogenesis and fluctuating hydroclimatic conditions, emphasizing the sensitivity of CRCs to short-term environmental changes.
Overall, our findings show that the abundance and isotopic composition of CRCs reflect both long-term glacial–interglacial dynamics and short-term hydroclimatic variability. This assessment highlights their potential as robust paleoenvironmental indicators, complementing traditional loess proxies for the reconstruction of past monsoon intensity and seasonality. Future research integrating CRC records from additional sites across Eurasia will further refine their application for tracing spatial gradients and temporal dynamics of the monsoon system.
