Effect of Forest Greening on Carbonate Rock Weathering Carbon Sink in the Subtropical Humid Zone

Abstract

The karst inorganic carbon sink is crucial for carbon neutrality, but its trends and drivers in subtropical humid zones remain unclear. This study selected subtropical humid zones in China with significant forest greening, quantified the carbonate rock weathering carbon sink (CCS) using a thermodynamic dissolution model, and explored the effects of climate, vegetation, hydrology, and radiation energy on CCS through importance analysis. The results showed that from 1982 to 2020, the CCS flux was 12.40 t C km⁻² yr⁻¹, and the total carbon sink was 1188.54 × 10⁴ t C yr⁻¹. Normalized difference vegetation index, leaf area index, and CCS exhibited an increasing trend, with growth rates of 0.002, 0.01 m² m⁻², and 0.05 t C km⁻² yr⁻¹, respectively. Surface available water, precipitation, and evapotranspiration were the dominant factors affecting CCS. This study found that forest greening caused precipitation to increase faster than evapotranspiration, driving an increase in available surface water and ultimately promoting the karst carbon sink in subtropical humid zones. Our findings highlight forest greening as a vital strategy for carbon neutrality.

1. Introduction

The terrestrial carbon cycle is crucial to global climate change and the stability of terrestrial ecosystems [1,2]. Carbonate rock weathering processes can consume CO₂ from the atmosphere, and this carbon sink process has received increasing attention [3,4,5,6,7]. Limestone in carbonate rocks comes into contact with water and CO₂ and can be converted into dissolved inorganic carbon (HCO₃⁻), as expressed in Equation (1) [4,8]. According to current estimates, the carbon sink produced by carbonate rock weathering is 0.2–0.9 Gt C yr⁻¹ [1,5], accounting for about 7%–32% of the terrestrial carbon sink. Therefore, quantifying the total amount, pattern, changing trend, and drivers of carbonate rock weathering carbon sink (CCS) is of great significance for clarifying the carbon cycle process and diagnosing carbon neutrality.

$${\mathrm{CaCO}}_3+{\mathrm{CO}}_2+{\mathrm{H}}_2\mathrm{O}={\mathrm{Ca}}^{2+}+2{\mathrm{HCO}}_3^{-}$$

(1)

Several studies have recently explored the amount, pattern, drivers, and mechanisms of CCS. Their common understanding is that precipitation, temperature, evapotranspiration, and vegetation dynamics are important driving factors for CCS, but there are differences in their assessments of the trends in CCS changes [1,3,9,10]. Zeng et al. [3] showed that CCS in southwest China exhibited a downward trend from 1970 to 2013 due to rising temperature and reduced precipitation, while Du et al. [9] demonstrated that CCS in China exhibited an increasing trend of 0.16 t CO₂ km⁻² yr⁻¹ from 1991 to 2020 due to the increase in available surface water. The carbon dioxide fertilization effect, caused by increased atmospheric CO₂, has generally promoted vegetation greening since the Industrial Revolution [11]. Moreover, urbanization has reduced the pressure on rural land, making southern China a global greening hotspot [12,13]. Given the increasing prominence of forest greening in ecosystems, it can be further speculated that vegetation dynamic is the primary factor driving differences in CCS.

Recently, studies have shown that forest greening has a promoting effect on CCS. Local observation demonstrated that the positive succession of rock-bare soil-grassland increased the concentration of dissolved inorganic carbon by 0.08–0.62 mmol L⁻¹ and the inorganic carbon sink flux by 3.01–5.26 t C km⁻² yr⁻¹ [14]. Further studies reveal that forest restoration enhances the accumulation of CO₂ by stimulating the growth of roots and soil organisms, thereby increasing the CO₂ partial pressure at the soil-rock interface to 3–10 times the atmospheric level [15]. This significantly increases the underground dissolution rate of carbonate rocks [5], leading to a substantial increase in HCO₃⁻ [8,16]. According to the isotope mixing model, approximately 83% of dissolved inorganic carbon originates from soil CO₂, confirming that vegetation restoration promotes CCS by increasing soil CO₂ [17]. Although vegetation restoration increases the carbonate rock weathering rate by 3 to 10 times [5,15], some studies have concluded that precipitation and runoff are the primary factors for CCS [6,9,18]. That is, biological processes are important drivers of CCS, but it remains challenging to generate dissolution reactions in the absence of exogenous water. However, existing studies focus on the process by which biological action during vegetation restoration drives the increase in CCS by increasing soil CO₂ [8,14], but lack an analysis of the effect of vegetation restoration on CCS caused by water cycle-driven rainfall.

Forest dynamics are closely related to the water cycle and can affect CCS by regulating rainfall. The water cycle between the land and the atmosphere is weak in areas with sparse vegetation, which are prone to drought [2]. In contrast, forest restoration enhances the soil moisture-vegetation-precipitation feedback, thereby increasing rainfall [11,19]. Therefore, the changing trend of CCS is not only related to the increase in soil CO₂ triggered by forest restoration, but may also be related to the increase in rainfall regulated by greening. Before 2011, forest greening was not evident due to the small scale of rural-urban migration. To date, rural outmigration has contributed to China’s urbanization rate reaching 67%, effectively alleviating land pressure and promoting significant greening [13]. Forest greening has significantly changed the regional water cycle [20,21], which may be the reason why Zeng et al. [3] and Du et al. [9] found different CCS trends in similar regions. Related studies also reported that large-scale forest greening can impact the water cycle, enhancing the transfer of water from the land to the atmosphere and creating favorable conditions for rainfall [19,22]. Therefore, forest greening may drive precipitation growth faster than evapotranspiration by altering the water cycle [23], thereby promoting an increase in karst carbon sinks through enhanced surface water availability.

China’s subtropical karst zone has recently become a global greening hotspot and a crucial component of the Yangtze and Pearl River ecological barriers. The advantages in ecosystem structure, function, and service are emerging. Ecological indicators, such as water conservation, soil conservation, carbon fixation, and oxygen release, are improving. These conditions support the study of forest greening and the karst carbon sink. In view of this, this study selected the subtropical humid zone with significant forest greening as a case study and conducted a systematic discussion on three aspects. First, spatiotemporal trends of the normalized difference vegetation index (NDVI), leaf area index, and CCS from 1982 to 2020 were quantified to clarify the evolving relationship between forest indicators and CCS. Second, correlation analysis was used to determine the direction of influence between variables and establish the impact pathway of forest greening on CCS. Third, importance analysis was used to assess the magnitude of the impact of variables on CCS, revealing the dominant variables. Ultimately, this study aims to develop a study framework for forest greening, radiation energy, the water cycle, and CCS, and to reveal the driving mechanisms, thereby expanding the path of sink enhancement.

2. Materials and Methods

2.1. Research Framework and Innovation Point

This study hypothesized that forest greening regulates surface water availability by enhancing evapotranspiration, thereby promoting CCS. Previous research on CCS has focused primarily on two aspects (Figure 1a). On the one hand, some studies have emphasized that vegetation greening affects CCS by regulating HCO₃⁻ [8,14]. On the other hand, other studies have explored the effects of temperature, precipitation, evapotranspiration, soil moisture, and NDVI on CCS [3,9]. However, few studies have examined the cascade of forest greening, radiative energy, the water cycle, and CCS. This study aims to explore the impact of forest greening-driven radiation energy and the water cycle on CCS (Figure 1b).

2.2. Study Area

China’s subtropical humid zone (97.52–122.19° E, 21.40–34.12° N) encompasses the area south of the Qinling Mountains and the Huaihe River in China and north of the tropical monsoon climate type. The division of climate zones is based on the China Climate Zoning Map, compiled using climate data from 1951 to 1970 (www.resdc.cn). The study area covers an area of approximately 235.84 × 10⁴ km², of which carbonate rocks are exposed for about 95.85 × 10⁴ km², accounting for about 40.64% of the area of the subtropical humid zone (Figure 2). The vector data of carbonate rocks are derived from the 1:2.5 million geological map of the People’s Republic of China compiled by Ye et al. [24]. The study area’s average temperature was approximately 14.04 °C from 1982 to 2020, while average annual precipitation was about 1292.95 mm [25]. According to the 1:1 million soil map data of the People’s Republic of China (www.resdc.cn), the soil types in the study area are primarily red soil, yellow soil, and lime soil, with the vegetation type predominantly subtropical evergreen broad-leaved forest. In recent years, the area of karst desertification in southern China has been reduced from 12.96 × 10⁴ km² in 2005 to 7.22 × 10⁴ km² through the implementation of ecological projects [26,27]. From 2002 to 2017, diverse land use changes in southern China increased aboveground carbon storage by 0.11 ± 0.05 Pg C yr⁻¹ [28], providing a suitable background for studying the relationship between forest greening and CCS.

2.3. Data Source

The data sources in this study can be divided into two categories. Evapotranspiration, precipitation, and temperature were used to quantify CCS flux through the thermodynamic dissolution model. NDVI, leaf area index, albedo, net radiation, latent heat flux, and sensible heat flux were used to construct the effect of forest greening on CCS. The sources, resolutions, and uses of these data were described in detail in

Table 1. Data source and uses.

Variables	Data Source	Resolution	Uses
Evapotranspiration (ET)	National Earth System Science Data Center (https://www.geodata.cn/)	0.05° × 0.05°	These data are used to quantify CCS flux using thermodynamic dissolution models.
Precipitation (PRE)	National Tibetan Plateau Data Center (https://data.tpdc.ac.cn)	1 km × 1 km
Temperature (TEM)	National Tibetan Plateau Data Center (https://data.tpdc.ac.cn)	1 km × 1 km
Normalized difference vegetation index	ORNL DAAC (https://daac.ornl.gov/)	0.083° × 0.083° Approximately 9222 m × 9222 m	These data are used to construct the interaction process between forest greening and CCS.
Leaf area index (NDVI) (LAI)	National Earth System Science Data Center (https://doi.org/10.5194/essd-15-4181-2023)
Albedo	ERA5-land (https://cds.climate.copernicus.eu/datasets?q=era5-land)	11,132 m × 11,132 m
Net radiation
Sensible heat flux
Latent heat flux

. The data needs to be converted to a uniform resolution for statistical analysis due to the large differences in spatial resolution. Considering that converting all data to the maximum resolution will result in less spatial raster data and reduce the stability of the analysis. Therefore, ArcGIS 10.2 was used to convert the low-resolution raster data (NDVI, leaf area index, evapotranspiration, albedo, net radiation, latent heat flux, and sensible heat flux) into a coordinate point format. Subsequently, the point coordinates were converted into raster data in the form of pixels through the Kriging interpolation method, with a unified spatial resolution of 1 km × 1 km.

2.4. Quantification of Carbonate Rock Weathering Carbon Sink

There are several methods for estimating CCS, including the hydro-chemical runoff method, the inverse model, the coupled carbonate weathering model, and the GEN-CO₂ model [3,5]. Among them, the thermodynamic dissolution model proposed by Gombert [29] is the most widely used. Based on the dissolution equilibrium of carbonate rocks (mainly limestone) in the presence of water, temperature, and CO₂, a thermodynamic dissolution model was developed, known as the maximum potential dissolution model. The formula is as follows:

$$D_{\max }={10}^6(PRE-ET)\times {(K_s{K}_1{K}_0/(4K_2(\gamma {\mathrm{Ca}}^{2+}){(\gamma {\mathrm{HCO}}_3^{-})}^2))}^{1/3}{\left(pC{O}_2\right)}^{1/3}$$

(2)

where D_max is the maximum dissolution rate of carbonate rock under reaction equilibrium (mol km ⁻² yr⁻¹), PRE and ET are precipitation (mm) and evapotranspiration (mm), respectively. Ks, K₀, K₁, and K₂ are limestone solubility product constants, respectively, and the data are quantified by (3)–(6), respectively; they are expressed as the equilibrium constant of CO₂ hydration and dissociation of HCO₃⁻, the equilibrium constant of CO₂ dissolving in water, and the equilibrium constant of CO₃²⁻ formation; γCa²⁺ and γHCO₃⁻ are the activity coefficients of calcium ions and bicarbonate ions, respectively, and are quantified by (7). pCO₂ is the partial pressure of CO₂ in the soil aquifer (atm), which is quantified by (14).

2.4.1. K_i Parameter Quantization

This study used the interpolated air temperature dataset from the meteorological station to quantify the K_i data. Ks, K₀, K₁, and K₂ are functions of Kelvin temperature TEM_k (K) [29,30]. The formula is as follows:

$$\log (K_s)=-171.91-0.08\times TE{M}_k+2839.32/TE{M}_k+71.60\times \log (TE{M}_k)$$

(3)

$$\log (K_0)=-14.02+2385.73/TE{M}_k+0.02\times TE{M}_k$$

(4)

$$\log (K_1)=-356.31-0.06\times TE{M}_k+21834.37/TE{M}_k+126.83\times \log (TE{M}_k)-1684915/TE{M_k}^2$$

(5)

$$\log (K_2)=-107.89-0.03\times TE{M}_k+5151.79/TE{M}_k+38.93\times \log (TE{M}_k)-563713.90/TE{M_k}^2$$

(6)

2.4.2. Quantification of Ca²⁺ and HCO₃⁻ Activity Coefficients

The activity coefficients of Ca²⁺ and HCO₃⁻ were calculated using the Debye-Hückel model [30] as follows:

$$\log ({\gamma }_i)=-A{Z}_i^2{\displaystyle \frac{\sqrt{I}}{1+B{a}_i\sqrt{I}}}$$

(7)

where a_i is the ionic radius of calcium and bicarbonate (Å), which are 6Å and 4Å, respectively; Z_i is the charge number of calcium ion and bicarbonate ion, which are 2 and −1, respectively; A and B are functions of temperature TEM (°C), quantified by (8) and (9); I is the ionic strength, quantified by (10).

The calculation formula for quantifying A, B, and ion intensity is as follows [30]:

$$A=0.4883+8.074\times {10}^{-4}TEM$$

(8)

$$B=0.3241+1.6\times {10}^{-4}TEM$$

(9)

$$I={\displaystyle \frac{1}{2}}{\displaystyle \sum _i{Z_i}^2}{C}_i$$

(10)

where C_i is the ion concentration of calcium and bicarbonate (mol L⁻¹), respectively; the data were measured data from 139 sampling points.

Subsequently, the point-scale γCa²⁺ and γHCO₃⁻ were converted to spatial continuous scale data using random forest. Random forest is a combined classification algorithm proposed by Breiman for classification and regression in machine learning [31]. It works by building multiple decision trees for a given training time and outputs the model as either the classification or the regression of each tree. We used γCa²⁺ and γHCO₃⁻ at the spatial point scale as dependent variables and 27 bioclimatic data in CHELSA (

Table 2. Bioclimatic data in CHELSA.

Shortname	Longtname	Unit
Bio1	Annual air temperature	°C
Bio2	Diurnal air temperature range	°C
Bio3	Isothermality	°C
Bio4	Temperature seasonality	°C/100
Bio5	Daily maximum air temperature of the warm	°C
Bio6	Daily minimum air temperature of the coldest month	°C
Bio7	Annual range of air temperature	°C
Bio8	Daily mean air temperatures of the wettest quarter	°C
Bio9	Daily mean air temperatures of the driest quarter	°C
Bio10	Daily mean air temperatures of the warmest quarter	°C
Bio11	Daily mean air temperatures of the coldest quarter	°C
Bio12	Annual precipitation amount	kg m−2 yr−1
Bio13	Precipitation amount of the wettest month	kg m−2 month−1
Bio14	Precipitation amount of the driest month	kg m−2 month−1
Bio15	Precipitation seasonality	kg m−2
Bio16	Monthly precipitation amount of the wettest quarter	kg m−2 month−1
Bio17	Monthly precipitation amount of the driest quarter	kg m−2 month−1
Bio18	Monthly precipitation amount of the warmest quarter	kg m−2 month−1
Bio19	Monthly precipitation amount of the coldest quarter	kg m−2 month−1
AI	Aridity index	-
CMI	Climate moisture index	kg m−2 month−1
GSP	Growing season precipitation	kg m−2 gsl−1
GST	Growing season temperature	°C
HUR	Near-surface relative humidity	%
NPP	Net primary productivity	g C m−2 yr−1
PET	Potential evapotranspiration	kg m−2 month−1
VPD	Vapor pressure deficit	Pa

) as predictor variables [32]. The number of decision trees generated in the random forest is 500, and 3-fold cross-validation was used to reduce the evaluation bias caused by random data partitioning. Afterward, a random forest was used to convert the point-scale data into raster data at a continuous spatial scale.

$$Y_i=argave({\displaystyle \sum _{\mathrm{t}}^N{h}_t(x_i\left|(f_j)\right.)})$$

(11)

where Y_i represents the prediction result of the i-th sample in the overall prediction sample; argave() represents the mean; h_t() is the prediction result of the t-th decision regression tree.

To evaluate the prediction accuracy of the model, this study used out-of-bag sample data (OOB) that did not participate in forest construction and calculated the mean square error (MSE) of the out-of-bag data:

$$MS{E}_{OOB}={\displaystyle \frac{1}{n}}{\displaystyle {\sum }_{k=1}^n{(y_i-{{\widehat{y}}_i}^{OOB})}^2}$$

(12)

$${R_{rf}}^2=1-{\displaystyle \frac{MS{E}_{OOB}}{{\widehat{{\delta }_y}}^2}}$$

(13)

where y_i represents the actual value of the i-th sample in the overall prediction sample; ŷ_i^OOB is the out-of-bag sample data, which represents the predicted value corresponding to the model.

2.4.3. Quantification of CO₂ Partial Pressure

The CO₂ partial pressure was quantified by Brook et al. [33] model as follows:

$$\log (\rho C{O}_2)=-3.47+2.09\times \left(1-e^{-0.00172ET}\right)$$

(14)

2.5. Correlation Between Factors and Carbonate Rock Weathering Carbon Sink

Pearson correlation was used to calculate the correlation coefficient between variables. The formula is as follows:

$$R_{xy}={\displaystyle \frac{{\displaystyle {\sum }_{i=1}^n(x_i-\overline{x})(y_i-\overline{y})}}{\sqrt{{\displaystyle {\sum }_{i=1}^n{(x_i-\overline{x})}^2}}\sqrt{{\displaystyle {\sum }_{i=1}^n{(y_i-\overline{y})}^2}}}}$$

(15)

where R_xy is the correlation coefficient between variables x and y; n is the number of samples; $\overline{x}$ and $\overline{y}$ are the means of variables x and y.

2.6. Statistical Analysis

In this study, Origin 2018 was used to fit the trend changes in NDVI, leaf area index, and CCS. The “Trend” package in R 4.4.2 was used to analyze the changing trends of CCS and factors [34]. Before running the “Trend” package, this study prepared yearly data, enabled 8-core parallel computing, and selected the Sen slope to estimate trend values.

Structural equation modeling was used to construct the impact pathways of forest greening on CCS. Specifically, annual data for each variable were extracted and imported into Amos 24.0. Then, path correlation parameters within the model were extracted, and trend analysis was used to quantify the cumulative magnitude of changes in the variables within the impact pathways. Furthermore, the “Glmulti” package was used to quantify the importance of variables affecting CCS [35]. This study used the corrected AIC as the selection criterion, retained a maximum of 128 optimal model selections for calculation, and did not consider the interaction effect.

3. Results

3.1. Vegetation Change Pattern and Trend

From 1982 to 2020, the spatial trend of NDVI in China’s subtropical humid zone was greater than 0 in 95.93% of the areas (Figure 3a); from the time series, NDVI increased from 0.56 to 0.63, with a linear trend of 0.002 (Figure 3b). From 1982 to 2020, the spatial trend of the leaf area index was greater than 0 in 91.35% of the areas, while only 8.65% of the areas were less than 0 (Figure 3c); from the time series, the leaf area index increased from 1.81 to 2.18, with a linear trend of 0.01 (Figure 3d). In summary, the region has experienced significant greening of vegetation, especially after 2011. This may further alter the regional water cycle and, in turn, influence CCS.

3.2. Pattern and Trend of Carbonate Rock Weathering Carbon Sink Flux

From 1982 to 2020, CCS flux in China’s subtropical humid zone was 12.40 t C km⁻² yr⁻¹ (Figure 4a). From a spatial perspective, CCS flux increased with increasing longitude (Figure 4b) and decreased with increasing latitude (Figure 4c). From the perspective of China’s subtropical zone, CCS flux in Taibei is the highest, at 29.06 t C km⁻² yr⁻¹, followed by the Oujiang-Minjiang-Nanling region, at 14.99 t C km⁻² yr⁻¹, and the Jiangnan region ranks third, at 16.79 t C km⁻² yr⁻¹ (Figure 4d). Moreover, the total CCS in this area is 1188.54 × 10⁴ t C yr⁻¹, while the study by Zeng and Liu [36] showed that the total amount in China is 1760 × 10⁴ t C yr⁻¹. This area contributes 67.53% of the total CCS, with 37.74% of China’s carbonate rock area.

From the time series, CCS flux exhibited an increasing trend from 1982 to 2020. The subtropical humid zone of China showed a spatially growing trend, with a growth rate of 0.05 t C km⁻² yr⁻¹ (Figure 5a). From a spatial perspective, CCS flux initially increased and then decreased with increasing longitude and also decreased with increasing latitude. The overall trend of change manifested a decreasing trend from southeast to northwest, with the most significant decrease in the southwest region (Figure 5b–d).

3.3. Correlation Analysis Between Variables and Carbonate Rock Weathering Carbon Sink

The spatial proportion of the correlation coefficients between variables and CCS was quantified through correlation analysis. On the one hand, the spatial correlation coefficient of NDVI–albedo was less than 0 in 84.62% of the area, indicating that an increase in NDVI can reduce the albedo (Figure 6a). The spatial correlation coefficient of albedo–net radiation was less than 0 in 65.17%, manifesting that a reduced albedo will increase net radiation (Figure 6b). The spatial correlation coefficient of net radiation–sensible heat flux was greater than 0 in 99.34% of the area, indicating that increased net radiation can promote an increase in sensible heat flux (Figure 6c). Similarly, the spatial correlation coefficients of sensible heat flux–temperature and temperature–evapotranspiration were greater than 0 in 82.42% and 93.45% of the areas, respectively, indicating that increased sensible heat flux can enhance evapotranspiration by increasing temperature (Figure 6d,e).

On the other hand, the spatial correlation coefficients of leaf area index–latent heat flux, latent heat flux–evapotranspiration, and evapotranspiration–precipitation were greater than 0 in 88.34%, 92.20%, and 91.11% of the areas, respectively, indicating that the enlarged leaf area index increases precipitation by promoting latent heat flux and evapotranspiration (Figure 6f–h). Increased rainfall can enhance the availability of surface water and ultimately promote the improvement of CCS (Figure 6i,j). Although the correlation analysis showed the influence of variables, it did not clarify the multi-factor process path.

3.4. The Effect of Forest Greening on Carbonate Rock Weathering Carbon Sink

Through the path relationship among multiple factors, the direct and indirect effects of variables on CCS flux were further inferred. Forest greening promoted an increase in vegetation greenness and leaf area index by 0.07 and 0.35 m² m⁻², respectively, which affect CCS through both physical and physiological effects. In terms of physical effect, the increased greenness of the vegetation reduced the albedo by 0.01, leading to an increase in net radiation reaching the ground by 11.35 W m⁻². This process boosted the sensible heat flux by 0.40 W m⁻² and ultimately raised the air temperature by 1.43 °C.

From a physiological point of view, the expansion of the leaf area increased the latent heat flux by 10.21 W m⁻², and the warming and the expansion of the leaf area jointly drove the evapotranspiration to increase by 92.55 mm. The increased water transport from land to the atmosphere contributed to a 147.06 mm increase in precipitation. Finally, the growth rate of precipitation exceeded that of evapotranspiration, driving an increase in available surface water of 57.55 mm, thereby boosting CCS by 1.88 t C km⁻² (Figure 7a). Moreover, importance analysis revealed that surface available water, precipitation, and evapotranspiration were key factors affecting CCS (Figure 7b).

4. Discussion

4.1. Mechanism of Forest Greening Promoting Carbonate Rock Weathering Carbon Sink

Through a systematic investigation of forest change and CCS, this study established a framework for studying forest greening, radiative energy, the water cycle, and CCS, revealing their cascading relationships. Meanwhile, this study found that forest greening indirectly affects CCS by altering radiative energy and the water cycle.

On the one hand, forest greening can alter the radiation energy through physical effects, driving a temperature rise that hinders CCS. This study showed that NDVI and leaf area index increased by 0.07 and 0.35 m² m⁻², respectively, from 1982 to 2020, and significant greening altered the surface roughness. Compared to bare ground, greened vegetation areas have a lower albedo. This lower albedo causes the surface to absorb more radiation energy, which in turn increases sensible heat flux and leads to surface warming [22]. Studies have shown that climate warming is a crucial driving force for increased transpiration in the Northern Hemisphere, except in tropical regions [22,37]. Therefore, forest greening promoted the increase in evapotranspiration through the warming caused by physical effects in this study. Meanwhile, the increased evapotranspiration further enriches the water vapor content in the atmosphere, facilitating the formation of clouds. Longwave radiation reflected from the surface is captured by more clouds, promoting surface warming [11]. Research suggests that a temperature range of 10–15 °C is the threshold for maximum carbonate rock dissolution [5,38], and exceeding this threshold after warming would be detrimental to CCS. The importance analysis of this study also revealed that temperature and evapotranspiration have a negative impact on CCS, which is consistent with previous research findings [3,9].

On the other hand, forest greening can also alter the distribution of water between land and the atmosphere through physiological effects, a process that offsets the negative effects of physical effects and ultimately promotes CCS. From 1982 to 2020, the expansion of leaf area directly increased latent heat flux and evapotranspiration by 10.21 W m⁻² and 92.55 mm, as confirmed by multiple studies [39,40]. Atmospheric moisture in inland areas is derived from transport by evapotranspiration, of which plant transpiration can account for 60%–90% of the evapotranspiration [22,41]. That is, forest greening can increase the transfer of land water into the atmosphere through transpiration and increase precipitation within the region or downwind. Several studies have illustrated that the precipitation cycle rate is significantly improved after afforestation, and both the external and internal branches of the atmospheric water cycle are strengthened, contributing to precipitation generation [19,42]. Ultimately, the increase in rainfall was faster than the evapotranspiration driven by warming and leaf area expansion, which brings more surface water to participate in the karst inorganic carbon sink reaction. As a result, CCS is promoted, in agreement with the study by Bai et al. [5].

4.2. Comparison and Verification of Research Results

Different results were compared to verify the stability of the study. CCS flux in China’s subtropical humid zone was calculated using a thermodynamic dissolution model to be 12.40 t C km⁻² yr⁻¹, with a total carbon sink of 1188.54 × 10⁴ t C yr⁻¹. Zeng and Liu [36] concluded that the CCS flux in China was 6.93 t C km⁻² yr⁻¹, totaling 1760 × 10⁴ t C yr⁻¹. Due to sparse vegetation, low temperature, and little precipitation, CCS in the Qinghai–Tibet Plateau and the arid northwest region is low. Therefore, CCS flux at the Chinese scale is lower than that in the subtropical humid zone, which is consistent with the objective law. At the provincial level, China’s subtropical humid zone encompasses the provinces of Hunan, Fujian, and Jiangxi. CCS fluxes in the three provinces are 11.83 t C km⁻² yr⁻¹, 13.83 t C km⁻² yr⁻¹, and 14.13 t C km⁻² yr⁻¹, respectively [5], which are consistent with the findings of this study. At the global subtropical scale, the study by Bai et al. [5] concluded that CCS flux was 18.60 t C km⁻² yr⁻¹ (

Table 3. Different study results and verification.

Study Area	Research Methods	Karst Carbon Sink Flux/t C km−2 yr−1	Karst Area/ 104 km−2	Total Carbon Sink/104 t C yr−1	Reference
China	Thermodynamic dissolution model	6.93	253.97	1760	Zeng and Liu [36]
Hunan		11.83	8.39	98.98	Du et al. [9]
Fujian	Thermodynamic dissolution model	13.83	0.82	11.30	Du et al. [9]
Jiangxi		14.13	3.21	45.38	Du et al. [9]
Subtropical	Thermodynamic dissolution model	18.60	−	−	Bai et al. [5]
China’s subtropical		12.40	95.85	1188.54	This study

), which is consistent with the results of this study.

4.3. Supplements and Uncertainties to Existing Knowledge

This study found that forest greening indirectly promoted CCS in the subtropical humid zone, supplementing previous understandings appropriately. Previous studies have focused on the process by which biological action in vegetation restoration drives an increase in CCS by increasing soil CO₂ [8,14]. This study investigated the impact of forest greening on CCS from the perspective of regulating radiation energy and the water cycle. The reason why Zeng et al. [3] and Du et al. [9] observed different trends in CCS in similar regions is that forest greening is at various stages. Despite the implementation of ecological engineering in 2000, China’s greening efforts significantly increased after 2011 [13,20], as confirmed by the NDVI and leaf area index time series in Figure 3. In short, this study found that the rapid greening of forests can alter the radiation energy distribution process and the water cycle process, indirectly enhancing CCS, which supplements previous understanding. Furthermore, the study also found that forest restoration can not only convert related inorganic matter into organic carbon for storage through photosynthesis but also promote CCS by strengthening the exchange of water between the land and atmosphere.

Our study only established a bottom-up framework, lacking a top-down cascade. We found that forest greening is only one crucial driver of increased precipitation. In fact, the factors affecting the increase in precipitation are diverse, and the monsoon from the ocean is also an essential factor. The study has shown that the water vapor transmission power in the south of China is weak when strong westerly winds and weak East and South Asian monsoons, making it difficult to penetrate deep into the continent, resulting in an abnormal decrease in precipitation in the East Asian monsoon region; on the contrary, it leads to an abnormal increase in rainfall [43]. Therefore, how to effectively separate the precipitation contributed by the monsoon and the precipitation contributed by forest greening needs to be considered in the future. Moreover, land use changes have an impact on soil carbon storage [44,45]. In particular, vegetation restoration can increase soil carbon and also affect the stability of karst carbon sinks [46], a direction that needs to be focused on in the future. It should also be noted that the low R² of the CCS time series was primarily due to a severe drought from 2009 to 2011 in five southwestern provinces of China (Yunnan, Guizhou, Guangxi, Sichuan, and Chongqing) [47]. This drought caused a sharp fluctuation in CCS, which directly affected R².

5. Conclusions

The pattern and trend of CCS were quantified in the subtropical humid zone of China, and the mechanism of vegetation, climate, hydrology, and radiation factors on CCS was further analyzed. The study concludes that forest greening indirectly promotes CCS in the subtropical humid zone by altering the distribution of radiation energy and the water cycle.

Specifically, NDVI and leaf area index increased by 0.002 and 0.01 m2 m⁻², respectively, from 1982 to 2020. Meanwhile, CCS increased by 0.05 t C km⁻² yr⁻¹. The specific mechanism of this phenomenon is that forest greening intensifies the transfer of water from land to the atmosphere through transpiration, resulting in increased regional rainfall and more available surface water. This process offsets the enhanced evapotranspiration driven by warming and leaf area expansion, and thus, the increased availability of surface water contributes to CCS. This study emphasizes that forest greening is not only an essential means of biological carbon fixation but also a meaningful way to promote karst carbon sinks, which is of great significance to the carbon neutrality strategy.