Response of soil water hydrochemistry and δ¹³C_DIC to changes in spatio-temporal variations under different land covers in SW China karst catchment

Karst aquifer systems are becoming increasingly important as groundwater resources, covering approximately 20–25% of the world's demand (Ford & Williams 2007). In a karst aquifer, soil water plays a complex and vital role in the transformation of precipitation to groundwater. There are a large number of conduits, fissures and high matrix porosity, and sinkholes, which further facilitate the rapid transport of surface water to groundwater and decrease water storage capacity (White 1988; Yuan 2007). Although Southwest (SW) China's karst regions have a subtropical monsoon climate with a mean annual precipitation ranging from 1,000 mm to 2,000 mm (Wang et al. 2017), the karst habitat is deficient in soil water for vegetation growth because of the thin soil layer and relatively high hydraulic conductivity (Jiang et al. 2014; Chen et al. 2017). Thus, soil water is an important limiting factor for crop growth in SW China's karst regions (Jiang et al. 2014).

Soil water can dissolve organic and inorganic solutes of soil into underground aquifers. In the karst environment, soil CO₂ and dissolved organic acids can drive the chemical dissolution of carbonate rocks (Yang et al. 2012). Therefore, soil water has a great influence on the geochemical characteristics of karst groundwater (Yuan 2007). In recent decades, with the increase in population and the rapid development of economy in the karst areas of SW China, arbitrary discharge of wastewater, unreasonable disposal of solid wastes, and abuse of agricultural pesticides and fertilizers have caused severe soil and karst aquifer pollution (Liu et al. 2006; Gutiérrez et al. 2014; Kong et al. 2018). The soil water chemistry can provide useful information regarding the spatial and temporal distribution of nutrients, minerals, magnitude, and the movement rates of pollutants into groundwater (e.g., Schulz 2004; Schwarzenbach et al. 2006; Futter et al. 2009; Stehle & Schulz 2015). Consequently, an in-depth understanding of soil water chemistry in karst regions will help us to manage shallow groundwater resources and deepen our knowledge of water-related hydrogeochemical processes.

The hydrochemical behavior of karstic aquifers is very sensitive to the environmental changes (e.g., Ford & Williams 2007; Zhang et al. 2011; Hartmann et al. 2014; Xiao et al. 2015). Springs and underground rivers in the karst system respond rapidly to the changes in climatic and hydrologic conditions, such as extremely intensive to precipitation events (e.g., Liu et al. 2007; Chen et al. 2012; Yang et al. 2012; Lan et al. 2016; Luo et al. 2016; Kadić et al. 2017). Farmer & Williams (2001) revealed that chloroform concentrations in a karst spring can vary as much as 60-fold in a rainstorm cycle. Previous studies have reported that hydrochemical behavior in karst spring exhibits marked annual, seasonal, diurnal, and storm-scale variations (Liu et al. 2007). Recently, there is a growing interest in investigating dissolved inorganic carbon (DIC) concentrations, δ¹³C_DIC, as well as karstification related with hydrothermal activities, under different land covers because of its essential contributions of carbon sink to global carbon cycle (Liu et al. 2011; Zhao et al. 2015). Because different sources of DIC have different isotopic compositions, δ¹³C_DIC is a direct reflection of the karstification processes. Changes of physicochemical parameters in springs and underground rivers have been frequently reported in many karst regions. However, studies on the soil water quality are very limited because of constraints on soil water sampling and measurement (Xiao & Wang 2009). Although many studies have been carried out to identify the soil water movement parameters such as flow path, mean residence time recharge, and runoff generation, few data sets addressed the geochemical responses to the environmental changes (Ozyurt et al. 2014; Chen et al. 2017). A series of hydrogeochemical processes accompanied by the water–rock–gas–organism interaction will lead to the hydrochemical variations of groundwater (Liu & Dreybrodt 2015). Therefore, research on the effect of different land covers and climate processes on the wet–dry-seasonal and spatial geochemical variations in soil water is important.

In this study, soil water under different vegetation types, including grass, dry, shrub, reforestation, and bamboo lands in Qingmuguan karst catchment (QKC) were collected semi-monthly from July 2010 to July 2011. This study has two main objectives: (1) to examine wet–dry-seasonal and spatial variations of major ion concentrations, DIC concentrations and δ¹³C_DIC; and (2) to investigate the influences of lithology, soil CO₂, and other controlling factors/mechanisms including human activities on soil water chemistry.

The study was conducted at the Qingmuguan Observation and Research Station for Karst Dynamics in the Southwest University in the Beibei district of northwest Chongqing, SW China (29°40′40″–29°47′00″N, 106°16′48″–106°20′10″E) (Figure 1(a)). This research area was selected as a hydrogeological field supported by the IGCP513 ‘Global Study of Karst Aquifers and Water Resources’ project.

This region belongs to a humid subtropical monsoon climate. The annual mean temperature and precipitation were 16.5 °C and 1,250 mm during 2010–2016, respectively, as recorded at the Qingmuguan meteorological station (Figure 2). The precipitation from May to October accounts for 77–84% of the precipitation for the entire year. The discontinuous soils have been developed from limestone and contain significant amounts of rock fragments. The soil depth in the depression varies from 20 to 190 cm. On most parts of the hillslope, there is a thin soil cover and exposed bedrock. The vegetation consists of grassland, subtropical forest (shrub), and crops (rice and vegetables). Land uses in the catchment are cultivated lands in depressions, shrub, or grass on the hillslope. The proportion of forest land, paddy fields, and dry fields accounts for 64%, 6%, and 28%, respectively. In addition, 70% of the agricultural area in this catchment lies within the Ganjiacao depression, at approximately 0.6 km². The fertilizer for rice and vegetables is mainly carbamide and farmyard manure. The season for fertilizer application is from May to early September.

The experimental site is a karst valley area covering 13.4 km², 12.7 km in length with a northeast–southwest structural trend (Figure 1(b)). The QKC's elevation ranges from 310 to 705 m. The north part of QKC has a generally higher elevation than the south part. The exposed strata are grey and black limestones and yellow sandstone. On the two sides of the catchment is Upper Triassic Xujiahe formation (T₃xj), a feldspar-quartz sandstone, siltstone, mudstone, and shale rock or coal, with a thickness greater than 700 m (Figure 1(c)). The outcropping strata in the middle part of the anticline include Lower Triassic Jialingjiang formation (T₁j) a grey, thick, massive limestone, dolomite, dolomite limestone, and brecciaous limestone, with a thickness greater than 600 m; the Middle Triassic Leikoupo formation (T₂l) argillaceous limestone, gypsum, with a thickness less than 80 m (Yang et al. 2010). The carbonate rocks of Jialingjiang formation and Leikoupo formation constitute the regional karst aquifer. The sand shale of Xujiahe formation is the relative water-resisting layer, and the mud shale of Feixianguan formation of Lower Triassic is not exposed is the aquifer floor (Yang et al. 2013). Due to the widely distributed limestones, a series of large or small depressions, caves, sinkholes, and funnels are distributed in the catchment. There is an underground river with a length of approximately 7.4 km, flowing from northeast to southwest (Figure 1(c)). In the upper stream, the surface water of Ganjiacao depression mainly recharges the underground river through the Yankou sinkhole. In the midstream, the underground river is also recharged from the other sinkholes and karst fissures. Downstream, the Jiangjia spring serves as the only outlet of the underground river (Figure 1(c)). There is a single conduit between Yankou sinkhole and Jiangjia spring tested by a high-resolution tracer experiment (Yang et al. 2010). The discharge of Jiangjia spring was perennial but varied greatly (0.5–3 m³·s⁻¹) in response to monsoonal precipitation distribution. The population in the catchment is approximately 6,000 and most drink water from Jiangjia spring.

To estimate the influence of different land covers on soil water, five sampling sites, including grassland (Q1), dry land (Q2), shrub land (Q3), reforestation land (Q4), and bamboo land (Q5), were chosen in QKC. The water sampling sites are illustrated in Figure 3 and Table 1. Soil waters were sampled at three different depths from 30, 60, and 90 cm in five sampling sites and sequentially collected semi-monthly. Soil water was sampled by using ceramic suction tension lysimeters (product model: 1900; Soil Moisture Co., Ltd, USA), which consisted of suction lysimeters, a porous ceramic cup, a santoprene stopper, a sampling bottle, tubes, and a pressure monitor (Huygens et al. 2008). An extraction kit was connected to the device. Percolating water in the sampler was drained by using an assorted syringe 3–5 days before sampling. The samples were compressed to negative pressure, and its neoprene tubes were bent and clamped to ensure that the vacuum was maintained in the sampler.

Field investigations included on-site measurement of pH, electric conductivity (EC), and concentrations of HCO₃⁻ and Ca²⁺. Conductivity and pH were determined by a portable water quality analyzer (Hach, USA), with precisions of 1 μS/cm and 0.01, respectively. The concentrations of HCO₃⁻ and Ca²⁺ were evaluated by using a calcium assay kit (Aquamerck, Germany), with precisions of 6 and 2 mg/L, respectively. A calibrated CDU440 CO₂ meter manufactured by Industrial Scientific, USA, was used to monitor the soil CO₂ concentration under different land covers. The measurement range is 0–60,000 ppmv, and the testing accuracy is 10 ppmv.

The water samples for hydrochemical analysis were stored in clean polyvinyl fluoride bottles and sealed with wax. All of the water samples were preserved at 4 °C until analysis, which was conducted within 10 days of sampling. In the laboratory, concentrations of Na⁺, K⁺, and Mg²⁺ were measured by using inductively coupled plasma optical emission spectrometry (ICP–OES Optima 2100 DV, with a lower limit of 0.01 mg/L). NO₃⁻, SO₄²⁻, and Cl⁻ concentrations were determined through ion chromatography (IC, with a lower limit of 0.01 mg/L). The precision of the IC and ICP–OES analysis was within ±5% for major elements.

Samples for δ¹³C_DIC were collected by using polyethylene bottles with 0.45 μm cellulose–acetate filter paper and then were added 0.2 mL saturated HgCl₂ solution to prevent biological activity. Analysis of δ¹³C_DIC was conducted by using a modified method of Atekwana & Krishnamurthy (1998). About 10 mL of water sample was injected into tubes that were pre-filled with 1 mL of 85% phosphoric acid. The CO₂ was extracted and purified after cryogenic removal of H₂O using a liquid nitrogen–ethanol trap. Finally, the CO₂ was cryogenically transferred into a tube for isotope measurement. The δ¹³C measurement has an overall precision of 0.1‰. A number of duplicate samples were measured and the results show that the differences were less than the range of measurement accuracy.

Soil moisture was measured via an alcohol combustion method. Soil carbonate content was determined through a HCl titration analysis. pH electrometric titration and the densimeter method were applied to perform pH determination and physical examination, respectively. Then, 10 g of fresh soil sample was extracted at a 1:5 (w/v) soil–to–solution ratio with 2 M KCl by shaking for 1 h. After the sample was centrifuged and filtered, the NO₃–N concentration was determined via the phenolate disulphonic acid method. Analyses were conducted in the Laboratory of Isotopic Geochemistry, Southwest University, China.

Table 2 shows the minimum, maximum, and mean values of the measured pH, EC, major cations and anions, as well as the calculated total dissolved solids (TDS) values in the soil waters and precipitation at each site. The pH of precipitation in the study area ranged from 4.21 to 5.53, with a mean value of 4.58 (n = 8). The EC (67–81 μS/cm) and TDS (23.13–46.79 mg/L) of precipitation were relatively low. The HCO₃⁻ and SO₄²⁻ concentrations in precipitation varied from 6.10 to 18.60 mg/L and from 9.96 to 15.85 mg/L, respectively.

Soil water pH ranged from 4.98 to 8.16. Most of the soil water samples were neutral to slightly alkaline, and their pH levels are higher than those of rainwater samples. For all soil water samples, HCO₃⁻ was the dominant anion, and Ca²⁺ was the dominant cation. There is a remarkable variation of HCO₃⁻, Ca²⁺, and TDS in soil waters under different land covers. The TDS of soil water samples vary from 18.62 mg/L (grassland) to 407.41 mg/L (shrub land), with a mean value of 188.76 mg/L for soil water, which was about five times higher than the mean value of 36.90 mg/L for precipitation. The mean concentrations of HCO₃⁻, Ca²⁺, and TDS were ranked as follows: shrub land > dry land > afforestation land > bamboo land > grassland. The concentrations of K⁺, Na⁺, and Cl⁻ in soil water were relatively lower than those of Mg²⁺ and Ca²⁺. The average NO₃⁻ concentrations in soil water from grass, bamboo, afforestation, shrub, and dry lands were 4.24, 10.47, 23.01, 42.71, and 96.68 mg/L, respectively. The [Ca²⁺]/[Mg²⁺] molar ratio of soil water was more variable, with a range of 2.73–24.98, and a mean value of 9.02. The [Cl⁻]/[Na⁺] molar ratios of most soil water samples were >1. The [Cl⁻]/[Na⁺] molar ratios of the soil water samples in shrub land yielded mean values (5.41).

The major ion concentrations of soil water and precipitation were shown as a hexa diagram in Figure 4. The soil water in this study area could be divided into three types. The concentrations of Ca²⁺ and HCO₃⁻ were high in soil water from dry land and shrub land under the Ca–HCO₃ type. The average SO₄²⁻ concentrations in soil water from grassland and bamboo land were 41.27 and 61.56 mg/L, respectively. These concentrations were significantly higher than those in the other soil water samples. These samples belonged to the SO₄–Ca type. HCO₃⁻ and (HCO₃⁻+ SO₄²⁻) accounted for 61% and 93% of the TZ⁻ on average in afforestation land and precipitation samples, respectively. Thus, these samples belonged to the HCO₃^.SO₄–Ca type.

The pH of all soil water samples is slightly lower in the wet season (6.66) than in the dry season (6.95) in the study areas. The TDS in the soil water samples from all land types except grassland (Q1) was higher in the wet season than in the dry season. A detailed wet–dry-seasonal fluctuations for major ions are shown in Figure 5. The concentrations of Na⁺, K⁺, Ca²⁺, Mg²⁺, HCO₃⁻, and Cl⁻ in the soil water samples collected from bamboo land and grassland exhibited minor seasonal changes. The concentrations of these ions vary by <8%. This indicated that the chemical characteristics in soil water collected from bamboo land were stable in different seasons. By contrast, pronounced seasonal variations were observed in Ca²⁺, HCO₃⁻, NO₃⁻, and SO₄²⁻ in soil water collected from dry land. The average concentrations of Ca²⁺, HCO₃⁻, NO₃⁻, and SO₄²⁻ in the rainy season collected from dry lands were 18.78%, 8.60%, 104.25%, and 18.99% higher than those in the dry season, respectively. The NO₃⁻ concentrations in soil water samples collected from dry land in May, June, and September were 146.8, 157.4, and 187.8 mg/L, respectively. Monitoring time of these values was consistent with the time of precipitation peak and agricultural fertilization season. In contrast to the changes in NO₃⁻ in the soil water samples from dry land, remarkably high NO₃⁻ concentrations were detected in others types of soil water samples in the dry season, but low NO₃⁻ concentrations were observed in the wet season. The concentrations of Na⁺, K⁺, Mg²⁺, and Cl⁻ in the soil water samples from shrub land and afforestation land showed a slight seasonal variation, that is, their concentrations were slightly higher in the wet season than in the dry season. However, the concentrations of Ca²⁺ and HCO₃⁻ in the samples from shrub land and afforestation land exhibited remarkable seasonal variation.

The pH values of soil water from grass, shrub, reforestation, and bamboo lands all showed an increasing trend with the increase in depth (Table 2). By contrast, pH values in dry land samples showed a negligible change with an increase between soil profiles from 30 to 90 cm. In general, except for bamboo lands, the TDS of soil waters increased among the soil profiles. The concentration of mains ions in soil waters from grassland exhibited minor fluctuations within soil profiles (Figure 6). However, the concentrations of Na⁺, Ca²⁺, Mg²⁺, HCO₃⁻, NO₃⁻, and SO₄²⁻ in the soil water from dry land were notably lower at a depth of 30 cm than at depths of 60 and 90 cm. The concentrations of Ca²⁺, HCO₃⁻, NO₃⁻, and SO₄²⁻ in the soil water from afforestation land showed a gradual increase as depth increased. Other ions slightly changed with depth. The concentrations of Ca²⁺, HCO₃⁻, and SO₄²⁻ in the soil water from shrub land varied from 95.16 to 118.41 mg/L, from 189.30 to 235.96 mg/L, and from 46.59 to 64.19 mg/L, respectively, at depths of 30 and 60 cm. The concentrations of these ions were also lower than 125.19, 235.96, and 68.26 mg/L, respectively, at depths of 90 cm. The concentrations of Na⁺, K⁺, and NO₃⁻ in shrub land samples slightly varied with depth. The Cl⁻ concentrations decreased as the soil depth increased. For other vegetation types, the concentrations of Ca²⁺, Mg²⁺, K⁺, NO₃⁻, SO₄²⁻, and Cl⁻ in the soil water of bamboo land decreased as soil depth increased. Therefore, the ions accumulated in the topsoil.

At pH between 7 and 9, HCO₃⁻ was the dominant constituent of DIC in most karst water (Jiang 2013). Thus, under the observed pH conditions, concentrations of DIC were expressed as HCO₃⁻ in this study. The δ¹³C_DIC in soil waters were listed in Table 2. Soil water derived from grassland and bamboo land had low DIC concentrations and light δ¹³C_DIC values, with a mean value of 14.52 mg/L and −17.21‰, 16.53 mg/L and −16.56‰, respectively. Soil water derived from the dry, shrub, and afforestation lands had higher DIC concentrations and heavier δ¹³_DIC values with the averages of 154.63 mg/L and −12.30‰, 220.50 mg/L and −11.89‰, 127.11 mg/L and −11.65‰, respectively. The δ¹³C_DIC values of soil waters exhibited minor fluctuations within soil profiles (Figure 7). The median of δ¹³C_DIC in soil waters collected from shrub, reforestation, and dry lands decreased as the soil depth increased.

The isotope of dissolved inorganic carbon (δ¹³C_DIC) in soil waters varied seasonally for different types of soil waters (Figure 8). The mean δ¹³C values of the DIC for grass, dry, shrub, afforestation, and bamboo land samples were −20.13‰, −14.62‰, −13.64‰, −13.13‰, and −19.21‰ during the dry season, and −15.21‰, −9.75‰, −8.72‰, −9.92‰, and −15.65‰ during the wet season, respectively (Figure 8). Generally, these δ¹³C_DIC values of soil water were 4–5‰, which were higher in the wet season than those in the dry season.

The parent material of this study area is carbonate rock. The carbonate contents of the soil samples were summarized in Table 3. The carbonate concentrations in soils under different land covers are high, with a range of 4.57 to 31.93 g/kg (Table 3). The mean carbonate concentration in soils was ranked in the following order: shrub land > dry land > afforestation land > bamboo land > grassland. Shrub land yields the highest carbonate content of 31.93 g/kg at 30–60 cm. By contrast, grassland exhibits the lowest carbonate content of 4.57 g/kg at 0–30 cm. The carbonate content is higher in the deeper layer (30–60 cm) than in the upper layer (0–30 cm) of all land covers except bamboo land.

Figure 9 illustrates that there is a remarkable spatio-temporal variation of the soil CO₂ concentrations with different land covers. In general, the soil CO₂ concentrations were much higher than the CO₂ concentrations near the surface of the atmosphere. At soil depths of 0–80 cm from February to August, the CO₂ concentration changed from 1,000 to 27,100 ppmv, 1,700 to 42,300 ppmv, and 2,200 to 29,700 ppmv for grass, shrub, and bamboo lands, respectively. In this study area, seasonal pattern was represented by a yearly cycle of change, with remarkable changes in summer but slight variations in winter. In addition, the highest CO₂ concentration was detected in July for both grassland (with 27,100 ppmv at 60–80 cm) and bamboo land (with 29,700 ppmv at 60–80 cm). However, the highest CO₂ concentration was recorded in August for shrub land (with 42,300 ppmv at 60–80 cm) with a time lag of a month. The concentrations of CO₂ in the soil environment observed at various depths were increased with an increasing soil depth.

The soil water hydrochemistry and δ¹³C_DIC showed remarkable differences with the different land covers. It might be due to the manifold factors such as soil pCO₂, carbonate dissolution, precipitation, water residence time, and human activities. Previous studies demonstrated that soil water hydrochemistry was widely controlled by the interaction of water with soil minerals and host rock (Yuan 2002). In this region, the main lithology is limestone and dolomite, which are intercalated with gypsum. Under the control of parent material, carbonate is a prominent component of the soils (Table 3), which is consistent with results presented by Li et al. (2015). In the soil environment, carbonate minerals react with carbonic acid to produce bicarbonate (HCO₃⁻). As shown in Table 2 and Figure 4, Ca²⁺ and HCO₃⁻ are the dominant ions in the soil water samples. The correlation between Ca²⁺ and HCO₃⁻ in soil water was positive, with a coefficient of 0.85 (N = 225, ρ < 0.01). The carbonate content of soil was also positively correlated (R² = 0.809, ρ< 0.01) with the Ca²⁺ concentration in soil water (Table 4). Therefore, Ca²⁺ and HCO₃⁻ in soil water mainly originated from carbonate dissolution in the study area. The carbonate contents in soils from shrub, dry, and afforestation lands were obviously higher than those in soils from bamboo and grass areas (Table 4). Consisting of the carbonate contents data, the Ca²⁺ and HCO₃⁻ in soil water collected from shrub, dry, and afforestation lands were apparently higher than those in soil water from bamboo and grass areas. Meanwhile, on the soil profile, the carbonate contents in soils and the concentrations of Ca²⁺ and HCO₃⁻ in soil waters collected from shrub, dry, afforestation, and grass lands increased with an increasing depth; however, the carbonate content in soils and the concentrations of Ca²⁺ and HCO₃⁻ in soil waters collected from bamboo land decreased with an increasing depth (Figure 6). The results indicated that the spatio-temporal variability of Ca²⁺ and HCO₃⁻ in soil waters was determined by carbonate weathering, which was driven by soil CO₂ and carbonate content in soils under different land covers.

The concentrations of Na⁺ and K⁺ in the soil water samples exhibited minor spatial changes. The content of clay particles in the soil profile was negatively correlated with the concentration of K⁺ and Na⁺ in the soil water, with a coefficient of −0.653 (ρ < 0.01), −0.643 (ρ < 0.01), respectively (Table 4). Clay particles can strongly absorb K⁺ and Na⁺ in the soil (Li et al. 2010). Therefore, the more positive ions were absorbed by soil, the less is dissolved in the soil water.

On the two sides of the catchment is the Upper Triassic Xujiahe formation (T₃xj), which consists of thin coal strata (Yang et al. 2010). The Middle Triassic Leikoupo formation (T₂l) interbedded gypsum strata. As stated earlier (Table 2), the concentrations of SO₄²⁻ in soil water collected from grass and bamboo lands were significantly higher than those in soil water collected from shrub, dry, and afforestation lands despite being under the same climatic conditions. The primary cause of the hydrochemical differences may be subsurface mineralogy in the soil. SO₄²⁻ in soil water can be derived from the dissolution of gypsum and the oxidation of sulfide minerals (Lang et al. 2006). In addition to subsurface mineralogy origin, the precipitation was significantly enriched in SO₄²⁻ (average of 13.07 mg/L). Precipitation acts as the water source of soil water. Therefore, the concentrations of SO₄²⁻ in soil water partly originated from precipitation. In summary, the dissolution of gypsum and acid rain provides abundant Ca²⁺ and SO₄²⁻ ions for soil water from grass and bamboo lands; hence, these samples belonged to the SO₄–Ca type.

As stated above (Figure 9), there were remarkable differences in measured soil CO₂ with the different vegetation types. In general, soil CO₂ inputs controlled by biological activity, available soil moisture and precipitation, and dissolution of carbonate (Zhao et al. 2015). Shrub land had a higher carbonate content and a strong bioactivity, due chiefly to the root respiration and microbial activity (Hanson et al. 2000). Root respiration is enhanced, leading to an increase in soil CO₂ production. Therefore, the soil CO₂ concentrations in shrub land were more significant than those in grass and bamboo lands. In karst systems, a small part of the soil CO₂ diffuses to the atmosphere, whereas most parts are dissolved in the infiltration water (Jassal et al. 2004). Soil CO₂ dissolves in soil water to form carbonic acid, which is a significant chemical driving force for carbonate dissolution (Jiang et al. 2014). A higher soil pCO₂ of the shrub land led to a higher carbonate dissolution. Therefore, high contents of TDS, Ca²⁺, and HCO₃⁻ in soil water in the shrub land corresponded to a high CO₂ dissolution in water and a high carbonate content in soil, in which more limestone and dolomite are dissolved. The bamboo land has the lowest carbonate content results in lower soil CO₂ concentration. Furthermore, the bamboo land located in sloping land, with a thin soil layer, and then soil CO₂ in the bamboo land can readily degas. Thus, there was a low soil CO₂ concentration gradient in bamboo lands. With a low soil CO₂ concentration gradient in bamboo lands, a relatively weak vertical hydrochemical variation is observed in the soil water.

The CO₂ concentration also showed remarkable temporal variations, with a higher level in summer and a lower level in winter. The catchment experiences a subtropical monsoon humid climate with a higher temperature and abundant precipitation in summer (June to August). In summer, root respiration in the soil is strong, with the lush growth of plants; hence, larger amounts of CO₂ are generated (Atkin et al. 2000). By contrast, soil CO₂ generation decreased as most weeds wither and die, and a low soil temperature was detected in winter. Therefore, temperature and humidity are the most important factors that determine the bioactivity in soils and make a significant contribution to soil CO₂ generation. Similarity, Ouyang & Zheng (2000) reported that precipitation determines the soil water content available for biological respiration and the air-filled pore spaces available for CO₂ flux. Yang et al. (2012) revealed that pCO₂ (partial pressure of dissolved CO₂ in water) in karst water changed synchronously with soil CO₂.

The similarities of insignificant seasonal variations between soil pCO₂ and the concentrations of Ca²⁺ and HCO₃⁻ in soil waters indicated that the soil water hydrochemistry was mainly determined by carbonate weathering, which is driven by soil pCO₂, the latter varying with both precipitation and temperature. In summer rainy seasons, more corrosive CO₂^– water solution is generated as the soil CO₂ concentration increases; as a consequence, erosion strongly occurs and high TDS and HCO₃⁻ levels can be detected in soil water. This variation shows that the soil CO₂ stimulates the seasonal hydrochemical variations in karst areas. Liu & Zhao (2000) demonstrated similar seasonal hydrochemical variations of the epikarst springs in Guilin Karst Experimental Site. They found that Ca²⁺, HCO₃⁻, and pCO₂ in water showed almost a simultaneous significant change as soil CO₂ partial pressure varies remarkably within a year. Yuan (2002) revealed that the soil CO₂ concentrations varied from 800 to 7,400 ppmv at 0–20 cm depths from February to June. The HCO₃⁻ concentrations changed from 201.3 to 286.7 mg/L. The soil CO₂ produced by plant root respiration in soil affects the diurnal hydrochemical change in the karst springs (Zhao et al. 2010; Yang et al. 2012). These differences indicate that climatic factors, particularly precipitation and temperature, and vegetation control the production, migration, and conversion of CO₂ in the soil–aquifer system and the intensity of karst processes (Zhang 2011). Hence, changes in temperature, precipitation, and vegetation can effectively regulate the soil CO₂ changes and result in wet–dry-seasonal and spatial variations of geochemistry in soil waters.

Compared with those collected from other land use covers, soil water samples collected from dry land were relatively high in Na⁺, K⁺, NO₃⁻, SO₄²⁻, and Cl⁻ contents. Especially, the average NO₃⁻ concentration in soil water from dry land was 96.68 mg/L, which far exceeded WHO's maximum contaminant level (10 mg/L) (Table 2). In a pollution-free area, natural nitrate concentrations in groundwater are very low (Debernardi et al. 2008). However, many karst aquifers are characterized by elevated NO₃⁻ concentrations from anthropogenic pollution (Musgrove et al. 2016; Xiao et al. 2016). In the investigated area, some crops, such as corn, potato, and capsicum dry, were planted in dry lands. Fertilizers containing 225 kg of urea, 750 kg of ammonium bicarbonate per hm², and a certain proportion of potassium dihydrogen phosphate were applied to the dry land from May to June (Yang et al. 2010). Excessive N fertilization in the dry land has resulted in severe nitrate pollution. NO₃⁻–N of soil is easily dissolved in water, which leads to the significant positive correlation (R² = 0.845) between the concentration of NO₃⁻ in the soil water and the content of NO₃⁻–N of soil through long-term eluviation (Table 4). Therefore, the higher NO₃⁻ contents in soil water under dry land may be introduced by intense farming activities, such as synthetic chemical fertilizers. This is supported by the fact that the NO₃⁻–δ¹⁵N values of soil water samples collected from dry land ranged from −3.74‰ to 5.56‰ (Wang et al. 2009). Normally, the NO₃⁻–δ¹⁵N values of chemical fertilizers ranged from −4‰ to ±4‰, while the NO₃⁻–δ¹⁵N values of mineralized soil organic nitrogen ranged from 2‰ to 9‰ (Heaton 1986). Previously published studies report that many sources of nitrate contaminants include human and animal excreta and bio-combustion, the nitrification of organic N and NH₄⁺, and agricultural fertilizers (Heaton 1986; Galloway et al. 2008).

Usually, the time for agricultural fertilization is from May to September in this study area. Thus, evident seasonal variations in NO₃⁻ were observed in soil waters from dry land with a higher NO₃⁻ in the rainy season, and vice versa during the dry seasons. The soil waters collected from dry land had significantly higher NO₃⁻ contents in the upper layer (0–30 cm) than in the deeper layer (30–60 cm). This is mainly due to the fact that surface soil is more susceptible to nitrate pollution.

Conventionally, the DIC in water generally has three sources, including atmospheric CO₂, CO₂ from organic respiration, and carbonate dissolution (Aucour et al. 1999). As infiltrating waters dissolve soil CO₂, some of the dissolved CO₂ is hydrated and dissociated into HCO₃⁻ (bicarbonate) and CO₃²⁻ (carbonate), and the distribution of DIC is set by pH (Clark & Fritz 1997). The contribution of atmospheric CO₂ is minor because of the high partial pressure of soil CO₂ and low pH values (lower than 5.6) of rainwater. Thus, DIC in soil water studied here may have two primary sources, soil CO₂, and weathering of carbonate minerals.

δ¹³C_DIC as a natural tracer in karst systems can help to differentiate material originated. δ¹³C_CO2 depends on both the source material (C₃ or C₄ plants) and the rate of CO₂ diffusion. The C₃ photosynthetic pathway dominates in most terrestrial ecosystems and produces organic matter with δ¹³C values that ranged from −30‰ to −24‰ (Vogel 1993). C₄ plant-derived lipids exhibit δ¹³C values ranged from −20‰ to ±2‰ (Chikaraishi & Naraoka 2006). As suggested by Cerling et al. (1991), molecular diffusion of CO₂ in the soil can result in an isotopic enrichment of +4.4‰. Thus, it can be assumed that the δ¹³C of soil CO₂ for biogenic carbon is approximately −23 ± 3‰ (Li et al. 2008). Marine carbonate has a relatively narrow range of C isotope ratios, with δ¹³C values close to 0‰ by definition (Telmer & Veizer 1999; Clark & Fritz 1997). The δ¹³C_DIC related to the dissolution of carbonate by carbonic acid should be around −12‰ (Jiang 2013). The soil waters displayed remarkable variations in δ¹³C_DIC values in the range of −20.68‰ to −6.90‰ under different land covers in the study area (Table 2). This suggested that the δ¹³C_DIC of soil waters under different land covers was controlled by different geochemical processes. The soil water samples collected from shrub, afforestaion, and dry lands had δ¹³C_DIC values approaching −12‰, suggesting that DIC in soil waters mainly originated carbonate dissolution by carbonic acid. The δ¹³C_DIC values in soil waters collected from grass and bamboo lands were lower than −12‰. This is mainly due to the δ¹³C_DIC values which were determined by the mixing ratio of soil CO₂ and carbonate bedrock, and the ratio of soil CO₂ was higher than carbonate bedrock.

In a closed system, carbonate weathering generally by carbonic/sulfuric acid could lead to water chemistry characterized by [HCO₃⁻]/([Ca²⁺] + [Mg²⁺]) molar ratios of 2 and 1, and [SO₄²⁻]/([Ca²⁺] + [Mg²⁺]) molar ratios of 1 and 0.5, respectively. The following equations can express these processes.

The [HCO₃⁻]/([Ca²⁺] + [Mg²⁺]), [NO₃⁻]/[HCO₃⁻] and [SO₄²⁻]/([Ca²⁺] + [Mg²⁺]) molar ratios in soil waters have large amplitude of variation (Figure 10). The δ¹³C values of DIC display a positive relationship with the [HCO₃⁻]/([Ca²⁺] + [Mg²⁺]) ratio and a negative relationship with [NO₃⁻]/[HCO₃⁻] and [SO₄²⁻]/([Ca²⁺] + [Mg²⁺]) and [SO₄²⁻]/[HCO₃⁻] ratios, with a coefficient of 0.58 (ρ < 0.01), –0.06, –0.68 (ρ < 0.001), and –0.66 (ρ < 0.001), respectively (Figure 10). This suggested that carbonic acid could not be a unique dissolving agent and sulfuric/nitric acids played an important role in the carbonate dissolution in the investigated area (Jiang 2013). Specifically, the soil waters derived from grass and bamboo lands had lower [HCO₃⁻]/([Ca²⁺] + [Mg²⁺]) molar ratio values and had higher [SO₄²⁻]/([Ca²⁺] + [Mg²⁺]) and [SO₄²⁻]/([Ca²⁺] + [Mg²⁺]) molar ratio values (Figure 10(b)). This suggested that carbonate dissolution by sulfuric acid may be insignificant in the grass and bamboo lands (Lang et al. 2006). It was demonstrated in previous studies that sulfuric acid from oxidation of pyrite and acid rain might be implicated in chemical weathering of carbonate rocks in SW China (Li et al. 2008; Jiang et al. 2018).

The soil waters collected from dry land had high NO₃⁻ concentrations (∼132.07 mg/L) and high [NO₃⁻]/[HCO₃⁻] ratios (Figure 10(b)). There is no distinct correlation between [NO₃⁻]/[HCO₃⁻] ratios and δ¹³C_DIC values. In the investigated area, nitrate–nitrogen in the soil waters is mainly derived from the fertilizer mixture and soil organic nitrogen. The oxidation of organic fertilizer and NH₄⁺ fertilizers can produce acid (Equations (3) and (4)). The nitrification of NH₄⁺ fertilizers can enhance soil degradation and contribute to carbonate dissolution (Semhi et al. 2000). This suggests that weathering of carbonate rocks by nitric acid may be insignificant in a dry land.

As mentioned above, carbonate dissolved by carbonic acid, sulphuric acid, and nitric acids has made important contributions to the changes in hydrochemistry and δ¹³C_DIC in soil waters. A conceptual model with geochemical variations and its related controlling mechanisms in soil waters under different land covers is shown in Figure 11. It highlights that soil water hydrochemistry and δ¹³C_DIC are widely controlled by both land covers and carbonate lithology. The weathering of carbonate bedrock controlled by mixed acids in this study area. The various mechanisms can be described by the following Equations (5) and (6).

Semi-monthly hydrochemical data and δ¹³C_DIC in karst soil water samples were obtained to reveal the controlling mechanisms on geochemistry and δ¹³C_DIC under different land covers, namely grass, shrub, dry, afforestation, and bamboo lands, in Qingmuguan, a small karst catchment in Chongqing Province, SW China. Results show that there is a remarkable spatio-temporal variation of hydrochemistry and δ¹³C_DIC in soil waters under different land covers. The concentrations of TDS, Ca²⁺, and HCO₃⁻ in soil waters showed significant variation under different land covers. The average concentrations of TDS, Ca²⁺, and HCO₃⁻ in soil water ranked as shrub > dry > reforestation bamboo > grass lands. SO₄²⁻ was dominant in soil water collected from grass and bamboo lands. Hence, the soil water belonged to SO₄–Ca type. The concentrations of Ca²⁺ and HCO₃⁻ in the samples exhibit remarkable seasonal variation, with a higher level in summer and a lower level in winter from dry, shrub, and afforestation lands. However, the concentrations of Na⁺, K⁺, Ca²⁺, Mg²⁺, HCO₃⁻, and Cl⁻ in the soil water samples from bamboo and grass lands exhibited minor seasonal changes. There are clear seasonal variations in NO₃⁻ where observed in the soil water samples from dry land, with a higher NO₃⁻ level in the rainy season, and vice versa during the dry seasons. Except for bamboo lands, main ions in soil waters accumulated in deep soil layers. The soil waters displayed remarkable variations in δ¹³C_DIC values in the range of −20.68‰ to −6.90‰, and the δ¹³C_DIC values were 4–5‰ lighter in the wet season than in the dry season.

DIC increased due to the increase of soil CO₂ resulting from stronger microbial activities and root respiration in summer and wet seasons. The seasonal variation of soil water hydrochemistry and δ¹³C_DIC mainly controlled by carbonate dissolution related to carbonate content in soils, soil CO₂ input related to vegetation cover and climatic conditions and human activity, etc., and according to the range of δ¹³C_DIC and the relationship between δ¹³C_DIC and elemental ratios, carbonate dissolution by carbonic acid and other strong acids, such as sulphuric acid and nitric acids.

This research was supported by the Open Project Program of Chongqing Key Laboratory of Karst Environment (Grant No. Cqk 201702), the National Natural Science Foundation of China (Grant No. 51709127), the Natural Science Foundation of Guangdong Province, China (Grant No. 2017A030310172).