Rare earth element geochemistry in soils along arid and semiarid grasslands in northern China

Abstract

Background

Rare earth elements (REE) are a group of trace elements that behave geochemically coherently. REE fractionation patterns normalized to reference materials provide a powerful tool for documenting pedogenesis. In-soil processes are particularly difficult to illustrate with respect to contemporary and past climate conditions. In this study, we characterize the rare earth element (REE) contents in bulk soils and respective geochemical fractions (e.g., exchangeable, carbonate-bound, reducible, and oxidizable fractions) and to decipher the relationships between REE geochemistry components and climatic factors across a large-scale northern China transect (NCT).

Results

Across the NCT, bulk REE concentrations ranged from 55.2 to 241.1 μg g⁻¹ with a main portion in the residual fraction (49–79%), followed by oxidizable fraction (2–40%), reducible fraction (3–22%), carbonate-bound fraction (0.1–16%), and negligible exchangeable fraction. The REE contents of geochemical components (carbonate-bound, reducible, and oxidizable) in topsoils correlated to climate factors (mean annual precipitation, mean annual temperature, potential evaporation, and aridity index (AI)). The normalized abundances to the upper continental crust (UCC) composition show that the middle REE was generally enriched than the light REE and heavy REE in topsoils along the transect. The overall UCC-normalized bulk REE patterns in topsoils and subsoils were similar, characterized by weak negative Ce anomalies and positive Eu anomalies.

Conclusions

Our data in topsoils and depth profiles collectively suggest that cycling of REE was primarily regulated by abiotic processes in area with AI < 0.2, while the biological effect on REE circulation in soil played a more effective role in area with AI > 0.3. The similar UCC normalized patterns in topsoils suggest that the REE was originated from a common source with limited influences from other sources (e.g., atmospheric dusts and anthropogenic contributions). Our results to some extent provide evidence for climatic influence REE distribution patterns both in topsoils and subsoils across the continental-scale transect. Our investigation gives insights into future studies on vertical REE mobility and its associated biogeochemical pathways.

Introduction

The rare earth elements (REEs) are a set of 17 metallic elements as defined by the International Union of Pure and Applied Chemistry (IUPAC), including the lanthanide series elements (La, Ce, Pr, Nd, Pm, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, and Lu) plus yttrium (Y) and scandium (Sc). Despite their names, REEs (with the exception of short-lived radioactive promethium) are abundant and account for ~ 0.015% (w/w) of the Earth’s crust. For example, the contents of the lanthanides range from 63 μg g⁻¹ in Ce to 0.5 μg g⁻¹ in Tm (Tyler 2004). The REEs are widely applied in fertilizers to improve crop yields (Pang et al. 2001), and they are also essential for high-tech industries with an estimated global demand of 1200–1500 kilotons for expanding applications (e.g., Balaram 2019; Humphries 2010; Kumari et al. 2015; Mihajlovic and Rinklebe 2018).

Due to their common electronic structure of progressive filling of 4f orbit, the lanthanide elements (La to Lu) have similar physicochemical characteristics (e.g., ionic radii, 3⁺ valent state and electronegativity) and tend to exist together naturally rather than individually (Migaszewski and Gałuszka 2015; Tyler 2004). However, the lanthanides may be fractionated during different processes and under different environmental conditions. For example, Ce and Eu have variable electronic configuration; therefore, they have 4⁺ valent state and 2⁺ valent state under oxic and reducing redox conditions, respectively. In addition, the lanthanides are usually divided into three groups based on their subtle differences in electronegativity: light REE (LREE, from La to Nd), middle REE (MREE, from Sm to Ho) and heavy REE (HREE, from Er to Lu). The consequences and geochemical signatures of this fine distinction can be magnified and recorded in material circulation of the rock–soil–plant system. Thus, the geochemistry of REEs is widely used to trace parent rock lithology, weathering intensities, and redox changes in soil (e.g., Chapela et al. 2018; Chen et al. 2014; Laveuf and Cornu 2009; Liu et al. 2022; Liu et al. 2021a, b; Vermeire et al. 2016).

The REE contents in soil are principally the weighted average of the REE levels in geochemical components and the abundances of REE-bearing different constitutes. Five components, including exchangeable fraction, carbonate-bound fraction (highly mobile and bioavailable), Fe/Mn oxyhydroxide-affinity fraction (reducible), organic matter/sulfide-adsorption fraction (oxidizable) and the residue fraction in minerals (less mobility and non-bioavailable) can be operationally extracted and separated to show different forms of REE (e.g., Hu et al. 2006; Laveuf et al. 2012; Mihajlovic et al. 2014). Changes in environmental factors and edaphic factors such as temperatures, salinity, pH, and redox conditions have been documented to influence the REE redistribution and mobility into different geochemical components (Andrade et al. 2022; Cao et al. 2001; Davranche et al. 2011; Guénet et al. 2018; Pourret et al. 2007; Tyler, 2004). For instance, the releases of REE from Fe/Mn oxyhydroxide were found to enhance with increasing pH and decreasing redox potential in soils (e.g., Cao et al. 2001). However, the influence of long-range climatic factors on the distribution and mobility of soil REE on a continental scale is still unclear. Here, we address two important questions: (1) How would different REE geochemical fractions in soil vary along an aridity gradient? And (2) What biogeochemical and ecological information might those variations provide?

The northern China transect (NCT) is a 3700 km West–East grassland transect from Xinjiang to Inner Mongolia in northern China. The ecosystem types along the NCT shift from the typical steppe and meadow steppe from west to east, an ideal region to investigate elemental biogeochemistry along large-scale climatic gradients (e.g., Wang et al. 2014; Feng et al. 2016; Luo et al. 2016a, b). Previous studies have shown that cycling of nitrogen, phosphorus, sulfur, and carbon could be loosely decoupled and tightly coupled across this transect, shifting from geochemical to biological control on the nutrient elemental cycling over a narrowly defined aridity index (AI, the ratio of precipitation to potential evapotranspiration) threshold (AI = 0.20 to 0.30, Wang et al. 2014; Feng et al. 2016; Luo et al. 2016a, b).

REE contents and their mobility in soils are linked to weathering intensity of parent materials, the storage capacity of authigenic soil phases (e. g. carbonates, secondary oxide and oxyhydroxide, kaolinite and smectite and organic ligands) and the fluctuations of redox conditions. As all above mentioned factors are affected by climatic parameters (such as temperatures and precipitations), it is reasonable to hypothesize that the REE mobilization, redistribution, and fractionation will be impacted by climate change. The aims of this study are to (1) quantify the abundances and geochemical fractions of REE in surface soils; (2) examine the relationships between REE abundances and climate factors/edaphic factors and explore their implications; (3) compare the difference of REE distribution in the depth profiles to understand their potential vertical mobility.

Materials and methods

Study sites

A total of 21 study sites were selected along the northern China transect (NCT, Fig. 1). The NCT locates in the mid-latitude temperate region, spanning from the arid to semi-arid continental climate, and the representative types of vegetation are alpine meadow, desert steppe, typical steppe and meadow steppe from west to east. The terrains along the transect are generally flat with tablelands and gently rolling hills (Luo et al. 2016b), a unique feature that allows to deconvolve the climate effects from the minor/diminished topography effects in pristine soils at or near steady state with respect to elemental cycling along a well-constrained climate gradient.

Fig. 1

Map of sampling localities on a continental scale. Across northern China from west to east, the sampling sites are labeled S01–S21 in order. Based on the aridity index (AI), the sites are divided into three climate subtypes: arid areas (0 ≤ AI < 0.2, represented by light blue circles), semi-arid areas (0.2 < AI ≤ 0.5, represented by blue circles), and dry sub-humid areas (AI ≥ 0.5), represented by dark blue circles)

Mean annual precipitation (MAP) ranges from 34 to 406 mm along the NCT, with > 77% of rainfall occurring during June–September. The annual air temperature (MAT) ranges from − 3 °C to 10 °C from west to east. During November–March, the average monthly temperatures are below 0 °C at all sites, while the highest average monthly temperatures are 15–25 °C during June–August. The annual potential evapotranspiration (PET), a complex function of MAT, varies from 611 mm at the western sites to 1202 mm at the eastern sites (Wang et al. 2014). An overall negative correlation between annual MATs and annual MAPs exists among the sites along the transect (Feng et al. 2016). In this study, we use an integrated parameter of aridity index (AI = MAP/PET) to reflect the climate variability and change. Accordingly, the studied NCT region is divided into three subtypes, including hyper-arid and arid (0 ≤ AI < 0.2), semi-arid (0.2 ≤ AI < 0.5) and dry sub-humid (AI ≥ 0.5) (Middleton and Thomas 1997). Specifically, the studied sites (S02, S03, S04, S05, S06, S07, S08, S09, S10) are in the arid area, and other sites are in the semi-arid (S01, S12, S13, S14, S15, S16, S17, S18) and dry sub-humid areas (S19, S20, S21).

The soil is classified into three types: Haplic Calcisols (S01–S08), Calcic Cambisols (S09–S12), and Calcic Kastanozems (S13–S21) (Feng et al. 2016). The Haplic Calcisols represents an initial stage of soil formation with significant accumulation of secondary calcium carbonate; Calcic Cambisols have minimal B-horizon development, and are considered to be in an intermediate stage of development; Kastanozems has a deep, dark-colored surface horizon with an accumulation of organic matter, and an accumulation of calcium carbonate within 100 cm of the soil surface (FAO 1993, 2014).

Soil sampling was conducted in regions far away from the road to minimize anthropogenic interference. For each site, two 50× 50 m plots were selected and divided into five 1 × 1 m subplots in each corner and center. In each subplot, twenty random soil samples were collected from 0 to 10 cm below the surface (topsoils). According to the climate subtypes and locations, we also selected four representative sites (S03, S09, S13, S19) to study vertical REE geochemical behavior. For sites S03, S09, S13, and S19, soil samples from the underlying soil (10–20, 20–40, 40–60, and 60–100 cm) were collected.

Sequential extraction procedure and chemical analysis

Soil samples were first passed through a 2-mm polyethylene sieve to remove rocks and plant residue in situ. The air-dried soil samples were then ground into a fine powder in the laboratory and fully mixed and homogenized to be used as a representative sample. Soil pH values, clay contents, total organic carbon concentrations were determined by Wang et al. (2014), and were briefly described below. The soil pH value was determined in a soil–water ratio of 1:2.5 using the pH electrode. The soil organic carbon (SOC) concentrations were measured using an elemental analyzer (2400II CHN elemental analyzer; Perkin–Elmer, USA) at the Stable Isotope Facility of the University of California, Davis after removing carbonates using HCl (0.5 M) (Wang et al. 2014). The total inorganic carbon was measured by measuring the volume of CO₂ released from air-dried soil with HCl (2 M) at room temperature (Luo et al. 2016a).

A BCR sequential extraction procedure modified from Li et al. (2020) was also used to extract the geochemical components of REE and the general protocol was briefly described below (also see the flowchart of REE extraction protocol in Additional file 1: Fig. S1):

• Step 1: approximately 500 mg soil was weighed in a polypropylene centrifuge tube and 25 ml 0.05 M CaCl₂ solution was added to the tube. The tube was then shaken for 8 h at ambient temperature (22–23 °C), and centrifuged to obtain the residue and supernatant;

• Step 2: 20 ml 0.11 mol L⁻¹ acetic acid solution was added to the residue from step 1, and shaken for 16 h, and extracts retained as in step 1;

• Step 3: residues from step 2 mixed with 20 ml 0.5 mol L⁻¹ hydroxylamine hydrochloride (pH = 2) and shaken for 16 h at ambient temperature (22–23 °C). The extracts retained as in step 2;

• Step 4: residues from step 3 reacted with 5 ml 30% H₂O₂ for 1 h at ambient conditions (22–23 °C). The mixtures were evaporated (repeated twice). 25 ml 1 mol L⁻¹ pH = 2 CH₃COONH₄ solution was then added to the dried residues, and shaken for 16 h at ambient temperature (22–23 °C).

• Step 5: residues from step 4 were digested with HF–HNO₃–HCl solution in closed Teflon™ beakers for 3 days (Li et al. 2019).

According to the BCR sequential extraction method (Mittermüller et al. 2016; Rauret et al. 1999), the supernatants of step 1, step 2, step 3, and step 4 are exchangeable fraction, carbonate-bound fraction, reducible fraction, and oxidizable fraction, respectively. The bulk REE contents were calculated as the sum of geochemical fractions extracts. REE chemical extraction was conducted at the laboratory of deep-sea geobiology, Institute of Deep-Sea Science and Engineering, Chinese Academy of Sciences (IDSSE-CAS). The contents of reducible Fe, Mn and REE were determined using an inductively coupled plasma-mass spectrometry (Thermo Fisher Scientific ™ iCAP ™ RQ ICP-MS) at the analytical center of IDSSE–CAS. Internal standards, including Be, Ge, Rh, and Ir in 2% HNO₃, were used to correct instrumental drifts. The calibration standard (BCR667 and GBW07986) and blank samples were inserted every 20 samples and were routinely analyzed to evaluate the analytical accuracy and precision. The external relative precisions associated with bulk REE measurements of BCR667 and GBW07986 were 4.2 and 2.5%, respectively.

The mineralogical compositions of soils are deduced from by X-ray diffractogram. The general procedure of X-ray diffraction analysis is briefly described as follows: First, the air-dried soil sample was ground into a fine powder by an agate mortar and pestle and passed through a 200 mesh sieve. Then the powders were determined by X-ray diffractometry (XRD) (Ultima IV, Rigaku, Japan) with Cu Ka radiation (40 kV and 40 mA) at a scanning rate of 1 min⁻¹ and 0.02° step size at the Institute of Soil Science, Chinese Academy of Sciences.

Statistical analysis

Pearson’s correlation analyses were applied to study the relationships between rare earth elements (bulk and geochemical components) and climate factors as well as soil properties. The strength of the association between the variates was evaluated by P values: P < 0.05, representing the correlation is significant at 95% confidence interval, P < 0.01, representing the correlation is significant at 99% confidence interval. The statistical analysis was done using the software of PASW Statistics 18.

Results

Geochemical characteristics of REE in topsoils

Bulk REE contents in topsoils (0–10 cm) along the NCT varied significantly from 55.2 μg g⁻¹ at S04 to 241.1 μg g⁻¹ at S02 with an average value of 118.3 ± 36.8 μg g⁻¹ (SD) (Additional file 2: Table S1). The exchangeable fractions at all sites were practically negligible. The most mobile and bioavailable REE (15.3–79.0 μg g⁻¹), including carbonate-bound, reducible, and oxidizable fractions, accounted for 21–51% of the bulk REE contents among all sites (Fig. 2a) and had a wide range of values from 0.1 to 12.0 μg g⁻¹, from 5.6 to 31.3 μg g⁻¹, and from 1.5 to 63.0 μg g⁻¹, respectively (Additional file 2: Table S3). At the arid sites of S02, S03, S04, S05, S06, S07, S08, S09, S10, the carbonate-bound fractions (3.5 to 12.0 μg g⁻¹) were generally higher than those at other sites (Additional file 2: Table S3). The absolute amounts of reducible REE contents had no obvious difference among all the sites, but the relative fractions of these geochemical components were generally higher at arid sites (Fig. 2a). The oxidizable fractions (9.3 to 63.0 μg g⁻¹) dominated over the reducible fractions (5.6 to 16.1 μg g⁻¹) at the sites of S1, S14, S15, S16, S17, S18, S19, S20, S21.

Fig. 2

a Relative geochemical fractions (carbonate-bound, reducible, oxidizable and residue) of rare earth elements (REE) vs. aridity index (AI) in topsoils (0–10 cm). b Upper continental crust (UCC) normalized patterns of bulk REE in topsoils (0–10 cm) along the northern China transect (S01–S21)

The REE concentrations are usually normalized relative to the upper continental crust (UCC, refer to Taylor et al. 1981) to eliminate the “even–odd” effect and show the distribution patterns of individual REE (Vermeire et al. 2016). The normalization and parameters defined below facilitate comparisons of REE patterns graphically between sites. La_N/Yb_N ratios, La_N/Gd_N ratios, and Gd_N/Yb_N ratios are represented as the fractionations of LREE to HREE, LREE to MREE, and MREE to HREE, respectively (Brito et al. 2021). The Ce and Eu anomalies are defined as Ce/Ce* = Ce_N/(La_N × Pr_N)^0.5 and Eu/Eu* = Eu_N/(Sm_N × Gd_N)^0.5, respectively. The subscript N in the parameters above denotes REE concentrations normalized against UCC (Liu et al. 2019; Taylor and McLennan 1985).

UCC-normalized bulk REE patterns (Fig. 2b) were generally similar, characterized by MREE enrichment relative to LREE and HREE (La_N/Gd_N = 0.6–1.1, Gd_N/Yb_N = 1.2–1.9), weak negative Ce anomalies (Ce/Ce* = 0.9–1.0) and positive Eu anomalies (Eu/Eu* = 1.0–1.3) (Additional file 2: Table S1). In contrast, UCC-normalized REE patterns of geochemical fractions showed distinguished features (Fig. 3, Additional file 2: Table S4): (1) the carbonate-bound fraction was characterized by HREE depletion (La_N/Yb_N = 1.2–4.5, Gd_N/Yb_N = 1.9–3.8), pronounced negative Ce anomalies (Ce/Ce* = 0.3–0.9) and positive Eu anomalies (Eu/Eu* = 1.0–1.9), except for small Eu anomaly of 0.9 at site S02); (2) the reducible fractions generally showed HREE depletions (La_N/Yb_N = 0.8–3.1, Gd_N/Yb_N = 1.7–2.4) and negative Eu anomalies (Eu/Eu* = 0.6–1.0, except for the site S08 with a small positive Eu anomaly of 1.1); (3) the oxidizable fractions revealed clear MREE enrichments (La_N/Gd_N = 0.3–1.7, Gd_N/Yb_N = 1.2–2.1), and moderate Eu anomalies (Eu/Eu* = 0.7–1.2); and (4) the patterns of residue fractions were characterized by HREE depletion (La_N/Yb_N = 0.7–2.0, Gd_N/Yb_N = 1.1–1.9), with positive Eu anomalies (Eu/Eu* = 1.0–1.6).

Fig. 3

Upper continental crust (UCC) normalized patterns of a carbonate-bound REE, b reducible REE, c oxidizable REE, and d residue REE in topsoils (0–10 cm) along the northern China transect

Geochemical characteristics of REE in depth profiles at selected sites

The depth-profile REE patterns showed site-specific variations in carbonate-bound, reducible and oxidizable fractions, but with dominant residue fractions (Fig. 4, Additional file 2: Table S5, 63.1–94.5 μg g⁻¹ at S03, 81.0–110.1 μg g⁻¹ at S09, 76.1–122.3 μg g⁻¹ at S13, and 85.1–166.5 μg g⁻¹ at S19). In the arid site S03, carbonate-bound fractions in upper soil (13–15% in the layers of 0–10, 10–20, and 20–40 cm) were significantly higher than that in deep soil (0.4–2% in the layers of 40–60 and 60–100 cm), while the reducible and oxidizable fractions varied over a small range in the depth (13–17 and 2–7%, respectively). At the semi-arid site S09, the carbonate-bound, reducible and oxidizable fractions did not show noticeable depth changes, varying within 3–5, 15–20, and 3–6%, respectively. At the semi-arid site S13, the variations in carbonate-bound, reducible and oxidizable fractions were also within small fluctuations (0.2–2, 14–22 and 3–8%, respectively). At the dry sub-humid site S19, the carbonate-bound fractions were extremely low (0.1–0.3%), while the reducible fractions generally increased with depth (7–18%) and the oxidizable fractions decreased with the depth (18–40%).

Fig. 4

Depth profiles (0–10, 10–20, 20–40, 40–60, and 60–100 cm) of relative geochemical fractions in soil at selected study sites. a S03, aridity index (AI) = 0.03, b S09, AI = 0.10, c S13, AI = 0.22, d S19, AI = 0.51)

The UCC-normalized patterns for bulk REE and geochemical fractions of selected sites also showed various characteristics with depth (Fig. 5, Additional file 2: Table S5 and S7). However, the overall patterns of bulk REE were similar in topsoils and subsoils at all sites, showing HREE depletions (La_N/Yb_N = 0.8–1.6, Gd_N/Yb_N = 1.0–1.5), positive Eu anomalies (Eu/Eu* = 1.1–1.5).

Fig. 5

Upper continental crust normalized patterns of carbonate-bound REE, reducible REE, oxidizable REE, and residue REE in soil depth profiles (0–10, 10–20, 20–40, 40–60 and 60–100 cm) at selected sites. a S03, aridity index (AI) = 0.03, b S09, AI = 0.10, c S13, AI = 0.22, d S19, AI = 0.51

Correlations among REE geochemical fractions, climatic factors and soil properties

Across all sites, the carbonate-bound fractions were negatively correlated with aridity index (AI) (Table 1, r = – 0.81, P < 0.01) and mean annual precipitation (MAP) (Table 1, r = – 0.85, P < 0.01), but positively correlated with mean annual temperature (MAT) (Table 1, r = 0.83, P < 0.01) and potential evapotranspiration (PET) (Table 1, r = 0.86, P < 0.01). The reducible fractions had no correlations with AI and MAP, but had but had positive correlations with MAT (Table 1, r = 0.48, P < 0.01) and PET (Table 1, r = 0.46, P < 0.01). In contrast, the oxidizable fractions showed positive correlations with AI (Table 1, r = 0.81, P < 0.01) and MAP (Table 1, r = 0.77, P < 0.01), and negative correlations with MAT (Table 1, r = – 0.71, P < 0.01) and PET (Table 1, r = – 0.68, P < 0.01).

Table 1 Pearson correlation coefficients between climate factors and soil properties with REE concentrations (μg g⁻¹) of bulk and geochemical fractions in topsoils (0–10 cm) along the transect. Locations of the sites (S01–S21) are plotted in Fig. 1

There were no discernible relationships between carbonate-bound REE concentrations and total Fe–Mn abundances and clay contents (Table 1). There were positive correlations between carbonate-bound fractions and pH values (Table 1, r = 0.52, P < 0.05), total inorganic carbon (TIC) concentrations (Table 1, r = 0.46, P < 0.05), and the negatively correlation with total organic carbon (TOC) (Table 1, r = – 0.58, P < 0.01). The reducible fractions were only positively correlated with pH values (Table 1, r = 0.45, P < 0.05). The oxidizable fractions presented no evident relationship with TIC but had a positive relationship with total organic carbon concentrations (Table 1, r = 0.89, P < 0.01), reducible Fe–Mn concentrations (Table 1, r = 0.57, P < 0.01) and clay contents (Table 1, r = 0.71, P < 0.01), and presented a negative correlation with pH (Table 1, r = – 0.62, P < 0.01).

Discussion

Overall, our bulk REE data (55.2 to 241.1 μg g⁻¹) in topsoils (0–10 cm) along the NCT were well within the range of the reported Chinese surface soil REE contents (79.5 to 354 μg g⁻¹, Cao et al. 2000; Chen and Yang 2010; Liu et al. 2021a, b; Miao et al. 2008), except for a few samples with lower abundances. The general similarity of normalized bulk REE patterns in topsoils and depth profiles at selected sites (Figs. 2b and 5) suggests that the soil REE source along the NCT was originally from a source with the same lithology. Our normalized bulk REE patterns and bulk REE parameters were significantly different from those seen in Asian dust (e.g., from Taklamakan desert, Qaidam basin, Tengger desert, Indian Thar desert, Loess, and Tibetan Plateau), which have distinguished features of LREE depletion and negative Eu and Ce anomalies (Ferrat et al. 2011). It is reasonable to expect somehow significant influence of atmospheric dust inputs on REE behavior in the surface soils. However, the consistent continental-scale REE patterns across the NCT strongly suggest that the studied soils across the NCT may weather from similar parent rocks and did not acquire other significant sources (e.g., anthropogenic or natural contributions). This provides context for understanding the fate of REE in soil along the NCT. In the following sections, we first discuss the general REE behaviors in the topsoils. Next, we discuss REE contents and normalized patterns of different geochemical fractions. Finally, we briefly explore the implications of REE exchange in the soil depth profiles.

REE behaviors of geochemical fractions in topsoils

Our data indicated that the residue fraction comprises the majority of bulk REE (49–79%) at all sampling sites (Fig. 2a). This finding is comparable with earlier results of different soil types reported by Davranche et al. (2011), Mihajlovicet al. (2014), and Sukitprapanon et al. (2019) (e.g., 78% in wetland soil, 45–63% in marsh soil, and 60–90% in agricultural land). It is generally thought that this fraction is fixed in the lattices of minerals and/or strongly adsorbed onto clay minerals, and is not biologically available (Hu et al. 2006). The residue fractions had no relationship with soil properties or climatic factors (Table 1), suggesting that REE contents of residue fractions mainly inherit from parent materials in place.

As shown in Fig. 2a, the carbonate-bound fractions accounted for 0.1–16% of the bulk REE in surface soil. The carbonate-bound fractions in arid sites (AI < 0.2) were generally higher than those in other sites of this study, in soils of mining area in southern China, and of agricultural land and forest belt in Russia (2–3%) ( Liu et al. 2021a, b; Shatrova et al. 2021). Studies have shown that REE can be adsorbed on the surface of carbonate minerals due to electrostatic attraction, but mostly exist in the lattice by replacing Ca²⁺ ions during secondary carbonate formation (Polyakov and Nearing 2003). High evapotranspiration conditions were suggested to facilitate the formation of evaporated minerals including carbonates and upward transfer of soluble REE from deeper soil to the surface (Khorasanipour and Jafari 2018). Therefore, the arid environments (AI < 0.2) may favor higher REE retention by enhanced REE incorporation into carbonates in surface soil. The observed strong positive or negative correlations between carbonate-bound fractions and climate factors (P < 0.01, AI, MAP, MAT and PET, Table 1) in our data set are likely explained by the proposed mechanism above.

The reducible fraction is generally adsorbed on amorphous Fe–Mn (hydro)oxides and amorphous Fe sulfides (Cao et al. 2001; Rankin and Childs 1976). This fraction is not usually available for plants unless under reducing conditions (Šmuc et al. 2012). The percentages of this fraction ranged from 3 to 22% with an average of 12%, much smaller than values found in soil that had similar Fe and Mn abundances (Mihajlovic et al. 2019) and in Bohai Bay of China (Zhang and Gao, 2015). A positive correlation between soil pH and reducible REE contents was also observed (P < 0.05, Table 1). This might be due to pH-dependent adsorption and desorption kinetics that controls the budget of this fraction of REE, as well as the stability of Fe–Mn (hydro)oxides and sulfides. More interestingly, the reducible fraction does not show an expected positive correlation with the total Fe–Mn concentration. There are two possibilities to explain this observation: (1) part of REE could be adsorbed on the negative surface of organic matter, and they are released under the reducing condition (Davranche et al. 2011; Du Laing et al. 2009); (2) Fe–Mn (hydro)oxides are preferentially embedded into the organic matter matrix, forming the OM–metal nanoparticles that are difficult to be attacked to release REE under the reducing condition (Davranche et al. 2011; Pédrot et al. 2015), which is part of the reason a positive correlation between oxidizable fractions and Fe–Mn contents was observed.

The oxidizable fractions are those complex, absorb, and/or chelate with various forms of organics and sulfides (Mihajlovic and Rinklebe 2018). This component is usually considered to be associated with stable high-molecular-weight organic carbon. Thus it is less mobile and available compared with the carbonate-bound and reducible fractions (Filgueiras et al. 2002; García-Miragaya and Sosa 1994). Strong positive correlations between the oxidizable fractions and total organic carbon concentrations in soil were observed (P < 0.01), suggesting that organic matter is a major sink for REE as documented in numerous studies (Brito et al. 2021; Mihajlovic et al. 2014). We found that climate factors and the oxidizable fractions are correlated, showing strong positive correlations with AI and MAP (P < 0.01), but negatively correlations with PET and MAT (P < 0.01). This observation can be explained by the climatic-controlled preservation and decomposition of organic matter. At sites with higher MAP and AI, soil organic matter preservation is likely greater, thus leading to higher REE adsorption and less mobility. In contrast, higher PET and MAT may promote rates of organic matter decomposition and enhance REE cycling into other geochemical components and/or lost to the soil system.

As discussed above, the fractions of carbonate-bound and reducible REEs are bound to inorganic minerals, such as secondary carbonates and Fe/Mn oxides, whose formation and stability are primarily controlled by geochemically abiotic processes. The fraction of oxidizable REE is mostly associated with organic carbon, whose formation and stability are mainly controlled by biologic processes. The ratios were less than 1 at the sites of AI ≥ 0.32, while the ratios were significantly larger than 1 at the sites of AI ≤ 0.22. This observation suggests that the cycling of organic matter at sites of AI ≥ 0.32 significantly influences REE budget and transfer in soil, while abiotic processes (e.g., mineral dissolution, formation and translocation) play a more important role in controlling REE distribution at the site of AI ≤ 0.22. Our findings highlight a transition from abiotic to biotic control on rare earth elements in the soil along the large-scale climosequence. Moreover, we observed a significant correlation between AI and ratios of the sum of carbonate-bound and reducible REE contents to oxidizable REE contents (R² = 0.59, P < 0.05, Fig. 6) across the large-scale transect. The correlation indicated the importance of the biotic and abiotic processes on REE related to AI. The intertwined effects of abiotic and biotic processes on rare earth elements translocation within soils across the large-scale climatic transect can be broadly deconvoluted via the relation between geochemical parameters and climatic parameters, to reveal how REE cycling in soils would respond under global climate change, and provide a scientific basis of the sustainable development of the ecologically fragile areas.

Fig. 6

The correlation of the ratios of the sum of carbonate-bound and reducible REE content to oxidizable REE content (unless) against aridity index (AI) (R² = 0.59, P < 0.05)

Normalized REE patterns of geochemical fractions in topsoils

Overall HREE depletions compared to MREE and LREE were observed in the UCC-normalized REE patterns of carbonate-bound (La_N/Yb_N = 1.2–4.5, Gd_N/Yb_N = 1.9–3.8) and reducible components (La_N/Yb_N = 0.8–3.1, Gd_N/Yb_N = 1.7–2.4). This depletion may arise from progressive enhanced complexation and solubility in soil solutions from LREE to HREE (Khorasanipour and Jafari 2018), thereby leading to disproportionately fractionation and loss of the three different groups of REE in topsoils. The signature of the HREE depletion may partly inherit from their parent rock, which is also seen in the residue fraction (Fig. 3d, Additional file 2: Table S4). In addition, fractionations between LREE and MEEE (represented by La_N/Gd_N) in carbonate-bounded and reducible components (P < 0.01, Additional file 1: Fig. S2) were positively correlated to the climatic factor of MAP. The shift from MREE enrichment (La_N/Gd_N < 1.0) to depletion (La_N/Gd_N > 1.0) relative to LREE suggest factors/properties other than complexation and solubility influence MREE affinity with carbonate and Fe–Mn (hydro)oxides, but not specifically identified in this study partly due to lack of kinematics data.

Moderate MREE enrichments (Gd_N/Yb_N = 1.2–2.1) compared to HREE were also observed in the UCC-normalized patterns of oxidizable fractions (Fig. 3c, Additional file 2: Table S4). Stronger complexation of MREE with convex carboxylic acids and high stability of organic matter complexes in soil may contribute to their enrichments in oxidizable fractions (Marsac et al. 2013; Zhang et al. 1998; Laveuf and Cornu 2009). A complementary moderate LREE depletions compared to HREE (La_N/Yb_N = 0.5–0.9, Additional file 2: Table S4) were observed in oxidizable fractions at arid sites (AI < 0.22, except for S02), reflecting negligible loss of REE to soil through leaching. In contrast, no pronounced LREE depletion or enrichment compared to HREE (La_N/Yb_N = 0.8–1.2) were found at semi-arid and dry sub-humid sites (AI > 0.3), suggesting no preferentially bound to organic matters under these environmental conditions.

Ce anomalies (Ce/Ce* = 0.9–1) in bulk REE were close to 1; however, large negative Ce anomalies (Ce/Ce* = 0.3–0.9) were shown in the carbonate-bound fractions (Fig. 3a, Additional file 2: Table S4). In soil solutions, the common Ce³⁺ ion can be readily oxidized into Ce⁴⁺ ion under oxidative conditions. The Ce⁴⁺ ion may be subsequently absorbed on the insoluble particles, such as Fe/Mn (hydro)oxides and organic matter in topsoils (Laveuf & Cornu, 2009; Leybourne & Johannesson 2008). This is also consistent with overall positive and complementary Ce anomalies in reducible (Ce/Ce* = 0.6–1.1) and oxidizable fractions (Ce/Ce* = 0.7–1.2). Note that Ce anomalies values (Ce/Ce* < 1.0) in reducible and oxidizable fractions may reflect the loss in the open system of REE in topsoils to some extent.

Positive Eu anomalies were found in bulk REE (Eu/Eu* = 1–1.3). The positive Eu anomalies were likely inherited from their parent rocks. Feldspar is known to contain fewer amounts of REE except for Eu (Compton et al. 2003), because Eu²⁺ preferentially replaces Ca²⁺ in feldspar, while all other rare earth elements are trivalent (except for Ce). Therefore, the consistently positive Eu anomalies (Eu/Eu* = 1.0–1.6) in residue fractions likely reflected the presence of abundant plagioclase feldspar in these sites. The mineralogical compositions in soils verified the existence of feldspar (Additional file 2: Table S2). The moderate Eu positive anomalies also were observed in the carbonate-bound (Eu/Eu* = 0.9–1.9) and oxidizable REE (Eu/Eu* = 0.7–1.2) as shown in Fig. 3. The element of Eu is a valence variable element that could be reduced from Eu ³⁺ to Eu²⁺. The Eu ²⁺ prefer to be sequestered into the extracting phase and convey to other layers or substances (Chang et al. 2016; Prudêncio et al. 2011). However, the overall negative Eu anomalies (Eu/Eu* = 0.6–1.1) also were shown in the reducible fractions, which may be due to the incomplete extraction of hydroxylamine hydrochloride solution (Leybourne and Johannesson 2008).

REE exchange in soil depth-profiles

REE abundances in natural soils are generally controlled by their parent materials, intensities of weathering, and pedogenic processes. The REE behaviors in the soil depth-profile depend mostly on climatic factors and the local edaphic properties. In addition, root biolifting, soil animal activity, and soil water transport all together exert a complex influence on the distribution of REE (Stille et al. 2009; Vermeire et al. 2016; Wen et al. 2006). Specifically, the biogeochemical cycling of REE in soils can be viewed as mass transfer within different geochemical fractions and soil horizons or mass fluxes into and/or out of soils. Constraints on changes in the REE abundances and UCC-normalized REE patterns may provide critical clues to decipher in-soil processes and evaluate the states (e.g., fully open-system condition, semi closed-system condition and/or transient steady state) of the studied soil system.

As shown in Fig. 4, both absolute REE contents and proportions in geochemical fractions of soil depth profiles at respective sites of S03, S09, S13, and S19, varied in a wide range. However, the UCC normalized patterns in bulk REE and geochemical fractions (Fig. 5, Additional file 2: Table S5 and S7) were greatly similar features. These results suggested that REE was partitioned into the different geochemical fractions during pedogenesis and successive processes within soils, but no significant fractionations among the three groups of REE (LREE, MREE and HREE). Thus, the soil system studied here can be viewed at a transient steady state. Note that the inferred steady-state REE cycling does not imply REE contents in soils were constant over the course of soil formation and development, but imply the response time of REE due to addition/loss processes match the residence time of REE in soils. In the following discussion, only qualitative generalization can be drawn from the limited data set, mostly due to the paucity of in-situ soil, parent materials and water data in this study.

At the hyper-arid and arid sites of S03 and S09, proportions of carbonate-bound fractions were significantly higher than those at the semi-arid and dry sub-humid sites of S13 and S19 (Fig. 4). As shown in Fig. 4 and Additional file 2: Table S6, carbonate-bound REE abundances in the upper soil (0–40 cm) at the site of S03 (11.3–12.0 μg g⁻¹, 0–10, 10–20, and 20–40 cm) was significantly larger than the deeper soil (40–100 cm) (0.4–1.0 μg g⁻¹, 40–60 and 60–100 cm). The greater carbonate-bound fractions in the upper soil may be due to the upward transfer of REE from deeper soil by evaporation in arid sites soil (Khorasanipour and Jafari 2018). This also can be applied to interpret the observation of REE contents at different soil layers in reducible fractions at S03 (Fig. 4, Additional file 2: Table S6).

Another noticeable observation is REE contents in topsoils and oxidizable fractions were significantly greater than that in the underlying soil, as the aridity index increased (Fig. 4 and Additional file 2: Table S6). For example, REE abundances in oxidizable fractions at the site of S19 had shown progressively upward-increasing (from 16.3 to 63.0 μg g⁻¹, Fig. 4). This observation is consistent with the suggestion that organic matter would be the most important carrier of REE in soil depth profiles, while the Fe /Mn oxide and carbonate have minor contributions (Jin et al. 2017). Under wetter environments, plant productivity and microbial biomass may be enhanced, thereby promoting the production and preservation of organic matter in topsoils.

Besides the upward transfer of REE in the soil depth profiles, a downward transfer of complexations of Fe/Mn oxides and organic matter may occur. For example, REE contents in reducible fractions and oxidizable fractions slightly increased with the depth at the arid and semi-arid sites of S09 and S13 (Fig. 4, Additional file 2: Table S6). The observed trends may be related to the downward transfer of those REE geochemical complexes in the soil depth profiles, reflecting a minor contribution of biological effects on the REE redistribution, but a major role of abiotic effects on REE mobility.

Implications for sustainable development

Sustainable development is a key concept for our modern civilization and future prospects, which mainly concerns the harmonious development among humans, the economy and ecosystems on our planet. Evolution and development of all life, environments and our society on earth rely on the cycling of critical elements (such as C, N, S, O, P, Fe, Mo, Cu, Zn and REE) in a dynamically interconnected subsystem of the Earth (e.g., atmosphere–plant–soil subsystem). In this study, we provide a preliminary understanding of how REE in less-disturbed pristine soils fractionated and redistributed, and also provide a basic hypothesis of how REEs in soils respond to a large-scale climate gradient. Our results demonstrated that a significant portion of rare earth elements is labile that could circulate in the terrestrial ecosystem. Such results would help evaluate the positive and adverse impacts on plant growth with the increasing use of REE as agricultural fertilizers in human-impacted systems, and would be potentially beneficial to sustainably manage and mine REE in the terrestrial ecosystem for industry applications (Adeel et al. 2021; Chen et al. 2020; Dushyantha et al. 2020; Tripathee et al. 2016).

Conclusions

Our data among the first to characterize REE distributions and behavior of bulk and geochemical fractions in soils and study the relationship between REE geochemistry and climatic factors along a large-scale climosequence in arid and semiarid grasslands of northern China. Similar upper-continental-crust (UCC) normalized REE patterns of bulk and respective geochemical components suggest that REE in soils may inherited from common parent materials. These results indicate a transition from geochemical to biotic control on soil REE cycling with decreasing aridity across a threshold of 0.2–0.3. Our findings highlight the importance of climate factors (e.g., MAT and MAP) and soil properties (e.g., organic carbon and clay contents) on elemental cycling and carry implications for sustainable development and ecosystem services. In addition, our results open up new opportunities to further study factors and processes that control the REE cycle in soils. These include: (1) studying mineral-scale selective adsorption and affinity of REE in soils; (2) performing extensive research on the REE cycle along toposequences and chronosequences from a wide range of climates; (3) developing non-steady state models by incorporation of soil properties (e.g., salinity and pH value) to trace REE mass balance in the rock–soil–plant system and quantify the role of underlying processes. Such studies will continue to improve our understanding of the ecosystem REE cycling across tempo-spatial scales under global environmental changes.