Calcium Biogeochemical Cycle in a Typical Karst Forest: Evidence from Calcium Isotope Compositions

: In order to better constrain calcium cycling in natural soil and in soil used for agriculture, we present the δ 44/40 Ca values measured in rainwater, groundwater, plants, soil, and bedrock samples from a representative karst forest in SW China. The δ 44/40 Ca values are found to differ by ≈ 3.0‰ in the karst forest ecosystem. The Ca isotope compositions and Ca contents of groundwater, rainwater, and bedrock suggest that the Ca of groundwater primarily originates from rainwater and bedrock. The δ 44/40 Ca values of plants are lower than that of soils, indicating the preferential uptake of light Ca isotopes by plants. The distribution of δ 44/40 Ca values in the soil proﬁles (increasing with soil depth) suggests that the recycling of crop-litter abundant with lighter Ca isotope has potential effects on soil Ca isotope composition. The soil Mg/Ca content ratio probably reﬂects the preferential plant uptake of Ca over Mg and the difference in soil maturity. Light Ca isotopes are more abundant in mature soils than nutrient-depleted soils. The relative abundance in the light Ca isotope ( 40 Ca) is in the following order: farmland > burnt grassland > forests > grassland > shrubland. Our results further indicate that biological fractionation in a soil–plant system is a vital factor for Ca–geochemical transformations in soil surface systems.


Introduction
As an essential nutrient for plants, calcium (Ca) in organics-rich forest ecosystems (including mineral soils) is generally absorbed via the finer feeder roots of trees [1]. In particular, a major threat to forest ecosystem health is linked to the loss of environmental Ca [2]; therefore, many studies have focused on Ca sources and sinks in soils, plants, and the atmosphere across forest ecosystems [3][4][5][6]. Previous work has reported the use of 87 Sr/ 86 Sr (radiogenic Sr isotopic systems) and elemental ratios of Ba/Sr and Ca/Sr as typical tools for exploring Ca sources in forest ecosystems [7][8][9][10], while the application of Ca isotopes has been relatively limited due to high-precision measurements being difficult to perform. Radiogenic strontium isotope ratios, commonly used for distinguishing the sources of alkaline earth elements [8,[11][12][13][14] due to its similar behavior to Ca in plant uptake processes and in ionic radius (r Sr /r Ca = 1.13), are now more often applied in studies of the biogeochemical cycle in forest ecosystem [7,9,10,12,15]. Although Ca and Sr are both alkaline earth elements, previous studies have suggested that these two elements are different [16,17]. In addition, the ionic behaviors of Ca 2+ and Sr 2+ are different in the long-term development of a forest ecosystem [18][19][20][21].
Many studies have confirmed that large amounts of Ca are available in the processes that internally circulate organic matter, in vegetation, and in the soil of forest ecosystems [41]. The litter-decomposition rates determine the Ca mineralization rates in vegetation, which varies significantly among different plant species [42,43]. For example, previous studies indicated that the base cation (including Ca and Mg) demands of American basswood and sugar maple are relatively high [44], resulting in different element turnover rates. Currently, whether Ca and Mg vary significantly in the bottom soil and surface soil under different vegetation covers is still unclear. In the present study, the δ 44/40 Ca isotope compositions and Mg/Ca ratios of different forest ecosystem samples (rainwater, groundwater, soil, vegetation, and rock) were determined to understand the biogeochemical cycle of elements in Maolan areas, a representative subtropical karst forest in southwest China. In parallel, Ca in different pools were determined directly to further explore cycling processes involving Ca in karst forests. The main objectives are (1) to identify whether the stable isotope of Ca can be used to obtain quantified information about Ca uptake by plants during their life span and (2) to determine whether the Ca isotope composition of vegetation and associated soils can be used to better constrain Ca budgets in forests.

Geological Setting
In order to explore Ca biogeochemistry in a karst soil-plant system, a representative karst forest (Maolan National Natural Reserve Park, MNNRP) located in the southeast region of Guizhou Province (China) with a cover area of ≈213 km 2 and an elevation of ≈500 m (Figure 1) was selected. The terrain is high in the northwest and low in the southeast. Well known for its good vegetation and rich fauna, the study area has a hot and humid tropical monsoon climate with an annual mean air temperature of 15.3°C. The wet season lasts from April to October, while the dry season lasts from November to March. The annual precipitation is ≈1750 mm [45]. MNNRP is primarily covered by an evergreen broad-leaved forest (≈88%), with plants such as incense cedar and Taxus chinensis, that effectively reduces water loss and soil erosion. Moreover, MNNRP has a well-developed karst landform with jagged carbonate rock (shallow marine carbonate) as well as sporadic sandstones ( Figure 1). Black calcareous soil (shallow layer) dominates the study region due to lithological control (poor soil-forming capacity). Bare rocks on the ground lead to surface discontinuity, depleted water-holding capacity of the soil, and Ca-rich soil (salt base). Nevertheless, forest vegetation plays a key role in improving soil quality in MNNRP and in shaping the normal landscape and zonal vegetation.

Sampling Procedure
The calcium isotopic compositions of rainwater, groundwater, soils, forest tree leaves, crop leaves, and bedrock samples were measured. As shown in Figure 1, four rainwater samples were collected at the R site from June to September in 2008. Nine groundwater samples from springs and wells (≈5 m depth) were collected using a pre-cleaned sampler and bottles in July 2007 within the MNNRP (more details on the groundwater sites can be found in Reference [46]). Fifteen soil samples were collected from five typical soil profiles at different depths in the summer of 2007 [47], including farmland (LBP; 0, 40, and 110 cm), burnt grassland (LBM; 0, 40, and 70 cm), shrubland (LBS; 0, 40, and 90 cm), virgin forests (LBF; 0, 40, and 120 cm), and grassland (LBG; 0, 40, and 110 cm). Forest tree leaves, including the three most representative tree leaves from Nandina domestica, Folium Platycaryae, and Handliodendron bodinier, were collected during the summer of 2007 near the virgin forest soil site (LBF). Two crop leaves from soybean and corn were also collected at the farmland soil site (LBP) in 2007. A typical bedrock (dolomite rock) sample was collected near the virgin forest soil site (LBF) in 2007. The fresh plant samples were cleaned via ultrapure water, freeze dried, then put into a crushing device immersed in liquid nitrogen, and smashed into powder (<150 µm). All of the soil and bedrock samples were air-dried at ambient conditions and ground to less than 150 µm. The soils of five land use types had different pH and SOC values [47]. The cation concentrations of rainwater and groundwater were derived from our previous work [46,48].

Sampling Procedure
The calcium isotopic compositions of rainwater, groundwater, soils, forest tree leaves, crop leaves, and bedrock samples were measured. As shown in Figure 1, four rainwater samples were collected at the R site from June to September in 2008. Nine groundwater samples from springs and wells (≈5 m depth) were collected using a pre-cleaned sampler and bottles in July 2007 within the MNNRP (more details on the groundwater sites can be found in Reference [46]). Fifteen soil samples were collected from five typical soil profiles at different depths in the summer of 2007 [47], including farmland (LBP; 0, 40, and 110 cm), burnt grassland (LBM; 0, 40, and 70 cm), shrubland (LBS; 0, 40, and 90 cm), virgin forests (LBF; 0, 40, and 120 cm), and grassland (LBG; 0, 40, and 110 cm). Forest tree leaves, including the three most representative tree leaves from Nandina domestica, Folium Platycaryae, and Handliodendron bodinier, were collected during the summer of 2007 near the virgin forest soil site (LBF). Two crop leaves from soybean and corn were also collected at the farmland soil site (LBP) in 2007. A typical bedrock (dolomite rock) sample was collected near the virgin forest soil site (LBF) in 2007. The fresh plant samples were cleaned via ultrapure water, freeze dried, then put into a crushing device immersed in liquid nitrogen, and smashed into powder (<150 µm). All of the soil and bedrock samples were airdried at ambient conditions and ground to less than 150 µm. The soils of five land use types had different pH and SOC values [47]. The cation concentrations of rainwater and groundwater were derived from our previous work [46,48].

Chemical Analysis
For digestion of the plant, soil, and bedrock samples, (i) 100 mg of the sample powder was digested by 4 mL of 1:3 (v/v) HF and HNO3 in a PFA sample jar (Savillex co., Eden Prairie, MN, USA) and then heated at 140 °C on a hotplate (≈7 d) to ensure complete di-

Chemical Analysis
For digestion of the plant, soil, and bedrock samples, (i) 100 mg of the sample powder was digested by 4 mL of 1:3 (v/v) HF and HNO 3 in a PFA sample jar (Savillex co., Eden Prairie, MN, USA) and then heated at 140 • C on a hotplate (≈7 d) to ensure complete digestion of the fluorides, carbon, and silicates [49][50][51]. The samples were dried subsequently. (ii) The dried samples were dissolved again in 4 mL of 1:3 (v/v) HF and HNO 3 using the same digestion procedure repeatedly until the solution became limpid. (iii) After complete digestion and drying, pure HNO 3 (2 mL) was applied two times to remove the fluorin and then vaporized using a hotplate. (iv) The remaining substance after digestion and after drying was dissolved in 100 mL volumetric flasks with 2% HNO 3 . All of the digestion processes were performed in an ultra-clean laboratory at the Institute of Geochemistry, Chinese Academy of Sciences (IGCAS). The concentrations of major cations were measured using ICP-OES in IGCAS. The measurement of a procedural blank, the standard reference material (GSB071192 and GSB071193), and sample replicates were performed during whole analyzation processes to ensure quality control. The accuracy of cations was better than ±5% for the quality control standards. The measurements of rainwater and groundwater samples were the same as those in our previous work [46,48].

Calcium Isotope Analysis
The double spike method was applied to measure Ca isotope compositions following the previous studies [30,52]. The 43 Ca and 48 Ca double spikes were added into all samples before the purification procedure. In the mixed solution of the sample and spike, the 40 Ca  48 Ca came from the spike Ca. Thus, according to the measured ratios for 40 Ca/ 48 Ca, 44 Ca/ 48 Ca, and 43 Ca/ 48 Ca (spike), all of the non-naturally derived Ca isotope fractionations during purification and measurements could be corrected. Calcium was separated from other elements on cation exchange resin columns (BioRad AGW 50x8 resin) using 1.5N HCl as an eluent, concluding a Ca blank of <2 ng (<0.7%) [52]. After drying, the mixed solution of the purified sample (about 300 ng Ca) and the spike were re-dissolved using 2.2 N HCl (1 µL). All of the purification procedures were conducted at GEOMAR [52]. For the thermal ionization mass spectrometer (TIMS) measurements of the Ca isotopes, the "sandwich technique (TaCl 5 -activator solution)" and the out-gassed single filament (rhenium) were applied to preload the re-dissolved solution. Briefly, about 0.5 µL of TaCl 5 -activator solution was first added onto the filament and heated to near dryness. Then, the re-dissolved sample (1 µL) was added on top of the activator solution and dried again. Another 0.5 µL of the activator solution was then added onto the filament and dried [52]. Finally, the Ca isotopes were measured via Triton T1 TIMS in GEOMAR following the method described in a previous work [52]. Two data-acquisition steps were performed in measurement processes since the cup configuration of Triton T1 TIMS does not cover the Ca mass range (40 to 48 amu) in a single step. In other words, the data for masses 40, 42, 43, and 44 were collected in the first step while that for 43 and 48 were detected simultaneously in the second step. Moreover, 41 K was also measured to evaluate the potential influence on mass 40 K isobaric interferences. The details for the double spike correction of the Ca isotope data can be found in a previous study [52]. Typically, the Triton T1 TIMS presented a 40

Results
The Ca isotope compositions of the samples collected from MNNRP are summarized in Table 1 and Figure 2. The values measured from the forest ecosystem samples vary between −1.44‰ and 1.48‰, corresponding to a ≈3.0‰ natural variation in δ 44/40 Ca.

δ 44/40 Ca Ratios of Bedrock and Soils
The carbonate bedrock sample yielded a δ 44/40 Ca value of 0.36 ± 0.09‰ (Table 1). The Ca concentrations of the surface soil pools in different soil profiles varied from 20 to 106

δ 44/40 Ca Ratios of Bedrock and Soils
The carbonate bedrock sample yielded a δ 44/40 Ca value of 0.36 ± 0.09‰ (Table 1). The Ca concentrations of the surface soil pools in different soil profiles varied from 20 to 106 mmol/kg. Among the soil profiles, only the Ca concentration of grasslands increased with depth and relatively high Ca concentrations were found in the surface soil of farmland and virgin forests, while no obvious patterns were observed in the other profiles ( Table 1). As shown in Figure 4, the δ 44/40 Ca values of surface soil varied from −0.51 ± 0.25‰ to 0.29 ± 0.15‰, lower than that of rainwater, groundwater, and bedrock. However, the pools of Ca in bottom soil varied from 0.91 ± 0.18‰ to 1.48 ± 0.21‰, higher than that of bedrock and surface soils. The δ 44/40 Ca values of soil increased with depth in the soil profiles but were different for each soil profile. Simultaneously, it can be seen that the δ 44/40 Ca values of surface farmland is the lowest (Figure 5), indicating the depletion of heavier isotopes in surface farmland relative to other soils. In contrast, grassland and shrubland show the highest δ 44/40 Ca values, demonstrating 44 Ca enrichment compared with the virgin forests.
x FOR PEER REVIEW 7 of 12 mmol/kg. Among the soil profiles, only the Ca concentration of grasslands increased with depth and relatively high Ca concentrations were found in the surface soil of farmland and virgin forests, while no obvious patterns were observed in the other profiles ( Table  1). As shown in Figure 4, the δ 44/40 Ca values of surface soil varied from −0.51 ± 0.25‰ to 0.29 ± 0.15‰, lower than that of rainwater, groundwater, and bedrock. However, the pools of Ca in bottom soil varied from 0.91 ± 0.18‰ to 1.48 ± 0.21‰, higher than that of bedrock and surface soils. The δ 44/40 Ca values of soil increased with depth in the soil profiles but were different for each soil profile. Simultaneously, it can be seen that the δ 44/40 Ca values of surface farmland is the lowest (Figure 5), indicating the depletion of heavier isotopes in surface farmland relative to other soils. In contrast, grassland and shrubland show the highest δ 44/40 Ca values, demonstrating 44 Ca enrichment compared with the virgin forests.

δ 44/40 Ca Ratios of Vegetation
Vegetation yields the lowest δ 44/40 Ca values and the highest Ca concentrations in the forest ecosystem (Table 1)

δ 44/40 Ca Ratios of Vegetation
Vegetation yields the lowest δ 44/40 Ca values and the highest Ca concentrations in the forest ecosystem (Table 1)

Atmospheric Inputs of Calcium
In general, two distinct sources are suggested to be the main contributors of Ca in groundwater: one is atmospheric input, and the other is bedrock weathering. Although the lowest Ca concentrations were observed in rainwater (12.9-24.0 µmol/L) and the Ca concentration in groundwater (1163-2538 µmol/L) is tens of times greater than that in rainwater, the range of δ 44/40 Ca values in rainwater (0.55‰ to 1.01‰) is completely within the range of δ 44/40 Ca values in groundwater (0.40‰ to 1.08‰) ( Table 1). A previous study observed similar results in rainwater, snow, and throughfall samples in the Strengbach catchment [27]. Therefore, δ 44/40 Ca values in groundwater reflect atmospheric input to a certain extent. Carbonate rocks generally have high Ca concentrations and relatively low δ 44/40 Ca values [53][54][55][56]. In MNNRP, bedrock has the highest Ca concentration (6066 mmol/kg) and a δ 44/40 Ca value (0.36‰) close to that for groundwater. Thus, both local carbonate rock weathering and atmospheric inputs tend to be responsible for (a potential source of) the Ca isotope compositions of groundwater in MNNRP, which also indicates that carbonate dissolution does not give rise to significant Ca isotopic fractionation. Our 87 Sr/ 86 Sr data also support this finding (more details will be published elsewhere).

Plant and Soil Calcium
Generally, Ca is absorbed by plants via fine lateral roots, then transported, and redistributed to different organs through the xylem sap. After autumn, mineralized Ca in the litter of leaves and fruits is migrated to soil solutions, where it is partially reabsorbed by plants [41,57]. During pedogenesis, the contents of Ca and Mg in the soil decrease due to the combined influence of plant uptake (as important micronutrients for plants) and the rock weathering process (yielding elements/soluble components in the soil and further being potentially removed by groundwater [58,59]). The increasing Mg/Ca ratio shows a higher plant utilization rate for Ca than for Mg (or a potential higher groundwaterremoval rate of Ca than for Mg), probably indicating soil maturation and that pedogenesis started. Considering the comparable effects of the weathering process and the subsequent groundwater-removal process on soil Ca and Mg contents in a local region, it is reasonably assumed that the longer the pedogenic time, the higher the Mg/Ca ratio in soils may become under the impact of plants. The δ 44/40 Ca value of bottom virgin forests (120 cm) is as high as ≈1.48‰, and the δ 44/40 Ca value of soybean is as low as −1.44‰ (Table 1 and Figure 4). Therefore, the total range of δ 44/40 Ca values in the Maolan karst forest ecosystem is about 3.0‰. This observation is in line with the reported results. In other words, the Ca uptake by plant (as nutrient) is an important reason for Ca isotope fractionation on the Earth's surface [16,27,31,38].  A previous study also suggested that the total amount and the available pools of C, N, and P are the highest for forests and are less for shrublands and grasslands in the study area [60]. Moreover, the increasing δ 44/40 Ca values were observed to be accompanied by more soil development in a longer pedogenic process in Ohia tree-dominated Hawaiian rainforests [21].  (Table 1 and Figure 4). The lowest soil δ 44/40 Ca value observed in surface farmland may reflect the recycling of crop-litter-rich organic matter, which tends to be isotopically lighter than the underlying soil. This also reflects the result of the Ca nutrient demand in crops: the light Ca isotopes were preferentially extracted by the roots so that the soil was relatively enriched with heavy Ca isotopes. The biological fraction effects can be a reasonable explanation for the fractionation of Ca isotopes. This can also be supported by the results of a previous study which concluded that the isotope compositions of roots are dominated by the source variations and fractionation mechanisms, whereas the translocation mechanism is governed by fractionation processes [61]. Moreover, considering the different depths of plants roots, the δ 44/40 Ca of net fractionation processes would be higher if the roots spread within high δ 44/40 Ca value soil, which is an important direction of further research.

Local-Scale Ca Cycling Effects in Maolan Karst Forest Ecosystem
Our data confirm that the plant uptake of Ca (as nutrient) is an important source of Ca isotope fractionation on the Earth's surface [21,38,57]. Although we did not calculate the Ca budgets in the forest ecosystem, Ca-geochemical transformations occur in a surface system. The vegetation leaves are enriched with 40 Ca compared with soil: the light Ca isotopes are preferentially absorbed by plants, resulting in heavy Ca isotopes remaining in the soil. Soil mineral weathering-derived calcium presented a higher δ 44/40 Ca value (0.91‰ to 1.48‰) than bedrock (0.36‰). Therefore, mineral weathering cannot explain the heavy δ 44/40 Ca compositions of soil pools.
The Maolan forest catchment may only be representative of a typical karst ecosystem, but the significant vegetation-related biological uptake of Ca flux is observed. Therefore, its consideration in the future is necessary. As a function of ambient variations with time, the processes of Ca isotopic enrichment in different environmental systems (e.g., different plant-soil systems) is of great significance in correcting the Ca biogeochemical cycle model [57].

Conclusions
We observed that rainwater and groundwater are more enriched with heavy 44 Ca than bedrock and that soils and plants are more abundant in light 40 Ca. The greater abundance of 40 Ca in nutrient-rich soils than in nutrient-depleted soils reflects the decomposition and recycling of plant litter. The sequence of abundance in 40 Ca is as follows: farmland > burnt grassland > forests > grassland > shrubland. The evolution of Ca isotopes in karst forest ecosystems implies that intensely weathered soils have increased δ 44/40 Ca values and that the internal circulation of a nutrient pool controls the Ca cycle of forest ecosystems. The Ca isotope compositions of plants might record the fractionation produced via the dynamic equilibrium between soils and plants. Moreover, the soil Mg/Ca content ratio and the Ca isotope compositions probably reflect the preferential plants uptake of Ca over Mg and the different soil maturities. This study highlights that Ca isotopes may be useful in exploring (or quantifying in the future) the origins and sinks of Ca in forest ecosystems.

Data Availability Statement:
The data presented in this study are available in Table 1.