Introduction

China has committed to achieving carbon neutrality by 20601,2. Increasing forest cover and restoring degraded forests have been recognized as cost-effective strategies for removing CO2 from the atmosphere and are key elements within national climate change mitigation framework3. Due to recent extensive forest restoration programs4,5,6, China now has the largest area of planted forests in the world, totaling 6.77 million ha7. Notably, more than 45% newly established forests are located in southern China, which have contributed to offsetting 33% of the region’s fossil CO2 emissions8, making this region a critical area for national carbon mitigation efforts.

Forest carbon uptake depends on the forest growth rate9,10 and forest cover expansion11,12,13. It is well established that the carbon accumulation rates generally decline as forests mature14,15. In China, most PFs are young and middle-age stands, with those in southern China being the youngest on average16,17. If left unmanaged, carbon accumulation rates are expected to decline gradually within about 15 years9. Manage plantation forest through selective harvest and replanting could optimize the age structure of the forests and, therefore, maintain relatively high carbon accumulation rate18,19. This approach presents a viable solution for enhancing China’s future carbon sink9. Given the rapid social-economic development and large-scale landcover transformation in recent years, land suitable for plantations in southern China is highly limited and fragmented20. There is a need to assess the efficacy of different forest management strategies for maximizing carbon uptake without plantation expansion21.

When exploring potential approaches to forest management, it is important to consider the impacts of environmental drivers and CO2 fertilization on forest carbon accumulation rates8,22. Spatial variability in forest growth rates has been found to strongly correlate with the differences in hydro-thermal conditions and disturbance history23,24,25. Previous studies showed that CO2 fertilization contributed greatly to global land carbon uptake. However, significant discrepancies persist among different projections and regions26,27. Favorable environmental conditions could enhance forest regrowth rates by up to 50% in different regions of the Brazilian Amazon28. Thus, quantifying climatic and CO2 dependency of forest carbon uptake is critical for predicting future forest carbon uptake under changing climate conditions20,29.

Forest cover in Guangdong province, China, has doubled since the 1980s, contributing to an estimated accumulated carbon sink of 0.31Gt carbon between 1980 and 2020 (Guangdong Forestry Administration; http://lyj.gd.gov.cn/). However, the extent to which these PFs will contribute to regional carbon neutrality by 2060 remains uncertain. Given that future forest cover expansion is constrained by other competing demand for arable land and projected climate change30,31, Guangdong Province serves as a model case for developing forest management practices for land-based carbon mitigation. In this study we aim to 1) explore spatial distribution and key drivers of the growth rates of newly planted forests, 2) quantify the contributions of these plantation forests to carbon sequestration, 3) estimate the carbon sink potential under different effective forest management, including varying rates of selective harvest and replanting. The findings provide valuable insights into the optimal forest management scenarios for maximizing the carbon sequestration potential of existing PFs without the need for forest expansion. These strategies also offer useful guidance for carbon sequestration in other similar regions worldwide.

Results

Distribution of regenerated plantation forests

Integrating multiple fine resolution satellite products, we identified a total of 2.67×105 ha PFs established between 1982 and 2018 in Guangdong (Fig. 1) with 41.55% in the North region, 30.26% in the PRD region, and 17.7% in the West and 10.49% in the East region. These PFs were composed of evergreen broad leaf forests (47.5%) and needle leaf forests (38.75%), with a small portion of mixed forests (7.2%) and bamboos (6.55%). Most of the PFs in Guangdong were young with an average age of 23 years. In addition, 23.1% of PFs were less than 10 years, 22.6% aged 10 ~ 20 years, and 54.3% of PFs were middle-aged (>20 years). Forests in the PRD region were the youngest. The East region had the most middle-age PFs (over 65%). A total of 7.1×105 ha suitable land for plantation forests in 2018 was identified. More than half of the suitable land areas were distributed in the North region. The West region accounted for over 1.54×105 ha (21.7%). In the East and PRD region, less than 1×105 ha of land was considered to be suitable for plantation forests.

Fig. 1: Distribution of identified PFs in Guangdong Province, China.
figure 1

Inserts for (a) aboveground and belowground biomass carbon storage, shown in light and dark colors, respectively, b total biomass carbon accumulation rate during the first 30 years and c percentage of PFs at different age stage, natural forest and potential area for plantation establishment in each region.

Full size image

The total amount biomass carbon of the identified PFs in Guangdong Province is 0.22 ± 0.02 Pg C in 2018. Among the four regions, the North region was the greatest contributor, accounting for 0.1 ± 0.009 Pg C, followed by the PRD region with 0.05 ± 0.007 Pg C. The West and East regions accounted for less than 0.05 Pg C (Fig. 1a). The average accumulation rate of total biomass carbon during the first three decades since planted was estimated at 2.94 ± 0.05 Mg C·ha-1yr-1. Forests in the West region showed the highest carbon accumulation rate of 3.69 ± 0.04 Mg C·ha-1yr-1, closely followed by the North region at 3.14 ± 0.10 Mg C·ha-1yr-1. The growth rates in the PRD region were below 2.5 Mg C·ha-1yr-1. The East region with the oldest forests exhibited the lowest accumulation rate (Fig. 1b).

Drivers of biomass carbon accumulation rates

Stratified Chapman–Richards growth models revealed notable variations in carbon accumulation associated with distinct forest characteristics and environmental conditions (Fig. 2). For different forest types, mixed forests having the highest carbon accumulation rate, followed by needleleaf forests, broadleaf forests, and bamboo (Fig. 2a). Among the terrain variables, total biomass carbon accumulation rates increased with slope and elevation. In areas with steep slope (>15o), the carbon accumulation rate was nearly twice as high as in the areas with moderate slope (10o ~ 15o) and three times greater than in areas with gentle slopes (<10o) (Fig. 2b). Similarity, areas at higher altitudes (>400 m) showed almost double the carbon accumulation rate compared to lower altitudes (<200 m, see Fig. 2c). Under different hydrothermal conditions, carbon accumulation rates increased with precipitation and soil moisture (Fig. 2d, e). Forests in cooler areas accumulated more biomass carbon than those in warmer regions (Fig. 2f). Additionally, biomass carbon accumulation increased with incoming solar radiation (Fig. 2g), likely due to enhanced photosynthesis. For soil nutrient concentrations, areas with higher levels of soil organic carbon accumulating carbon faster (Fig. 2h). Similar disparities in carbon accumulation were found under different total soil nitrogen contents, with low (0 ~ 0.1%) and moderate (0.1 ~ 1%) nitrogen levels showing slower accumulation compared to areas with higher soil nitrogen content (>1%) (Fig. 2i).

Fig. 2: Total biomass carbon accumulation of PFs with growing stand age under different driving conditions.
figure 2

Drivers are (a) Forest type, b Slope (°), c elevation (DEM, m), d annual precipitation (mm), e soil moisture (mm), f mean annual temperature (Tmean, °C), g downward shortwave radiation (Srad, W·m−2), h soil organic carbon (OC, %) and i soil total nitrogen (TN, %). Shadings denote the 95% confidence interval of the growth models.

Full size image

Results from RF modelling show that topographic factors were the most important in influencing the carbon accumulation of PFs, followed by climate factors in the study area. Forest type and soil nutrients were the least important factors (Fig. 3). In the West and East regions, where the percentage of mature forest were higher (Supplementary Fig. 1), forest age played a more importance role (Fig. 3b,c). In the West region with highest annual rainfall (Supplementary Fig. 1), the relative importance of precipitation was the highest among the four regions.

Fig. 3: Average importance rankings of environmental drivers revealed by RF models.
figure 3

a The whole study area and the (b) West, c East, d North and e PRD region. Drivers include: forest type, slope, elevation (DEM), annual precipitation (Prec), soil moisture (Msoil), mean annual temperature (Tmean), downward shortwave radiation (Srad), soil organic carbon (OC) and total nitrogen (TN). Variable importance is ranked from the least important (1) to the most important (9). The average ranking was calculated following 99 iterations. The error bars denote the 95% confidence interval. The mean squared error (MSE) is shown by bar colors.

Full size image

Carbon sink potential under different forest management scenarios

Based on regional Chapman–Richards growth models, we projected the future carbon sink potential of existing PFs under five different forest management scenario, considering forest aging, future climate change and atmospheric CO2 conditions. Under the forest preservation scenario (S0), existing PFs are expected to supply an additional carbon sink of 0.15 ± 0.02 PgC by 2060 under the moderate development path SSP2-4.5 scenario (Fig. 4a). Since most forests were young or middle-aged, forest aging was the largest contributor to total biomass carbon increment (74.18%), followed by CO2 fertilization which accounted for 20.85% of the total carbon gains. As projections from multiple ESMs indicate that our study area will experience increased temperatures and precipitations in the future (Supplementary Table 6), climate change contributed positively, though it accounted for less than 5%.

Fig. 4: Potential biomass carbon accumulation (in petagrams) of existing PFs under various effects.
figure 4

Effects inlude: a forest aging, climate change and CO2 fertilization under SSP2-4.5 scenario and b different forest management scenarios.

Full size image

Across the four regions, the North region contributed the most to the future carbon accumulation with 0.05 ± 0.01 PgC (44.24%), followed by the PRD region with 27.02%. The West and East region contributed 19.63% and 9.1%, respectively. Under the sustainable scenario SSP1-2.6 or higher-emission scenarios SSP5-8.5, existing PFs will provide an additional carbon storage of 0.12 ± 0.02 PgC and 0.19 ± 0.03 PgC by 2060, respectively. The effects of CO2 fertilization varied significantly, ranging from 7.52% for SSP1-2.6 to 33.68% for SSP 5-8.5 (Supplementary Figs. 5 and 6). Climate change impacts were similar to those of SSP2-4.5, contributing 1.78% and 7.8% for SSP1-2.6 and SSP5-8.5, respectively.

We estimated the influence of various forest management scenarios, involving preservation of existing forests and reforestation on suitable land (Fig. 4b). With all suitable land planted (S1), the total carbon storage increment could reach 0.23 ± 0.02 PgC by 2060. Alternatively, if all PFs were clear-cut and allowed to recover naturally without human disturbance (S2), a total of 0.42 ± 0.01 PgC biomass carbon would accumulate by 2060. With S3, where trees over 30 years are selectively logged, the existing PFs will gain a total of 0.41 ± 0.01 PgC by 2060 without additional plantation established. Furthermore, if only young forests (less than 20 years) are preserved (S4), existing PFs will provide a substantial carbon stock of 0.53 ± 0.01 PgC by 2060. The results of SSP1-2.6 and SSP5-8.5 scenarios showed similar trends (Supplementary Figs. 6 and 7).

Discussion

In this study, we investigated the spatial patterns and driving factors of carbon accumulation rate in newly regenerated PFs in southern China. Additionally, we assessed the carbon sink potential under five different forest management scenarios, considering the effects of forest aging, climate change and CO2 fertilization. Our results show that the carbon accumulation rates in the West and North regions of Guangdong Province were higher than those in the PRD and East, due to synergistic effects of young age structure and favorable environmental conditions. We also found that the impacts of different management scenarios far outweights that of climate or rising CO2 levels on the carbon sequestration potential of the existing PFs over the next four decades.

Among the time-invariant drivers, topographic variables were key determinants of carbon biomass accumulation (Fig. 3), corroborating the results obtained from the eXtreme gradient boosting (XGBoost) models (Supplementary Fig. 8) and after the removal of outliers (Supplementary Fig. 9). Biomass carbon accumulation rates of PFs were positively associated with rising elevation and slope (Fig. 2), consistent with findings from partial dependence analysis (Supplementary Fig. 10) and previous studies on observed carbon density10. This relationship may be attributed to recent government-funded efforts to expand the forested area onto steep slopes in the study region32,33. A large number of forests were planted in high elevation and steep areas8,34,35, where forest age generally declined with increasing elevation and slope (Supplementary Fig. 11a, b). This could explain the North region with relatively high elevation and cooler temperature has higher carbon accumulation rate of PFs. Additionally, the high carbon sequestration potential of PFs in the West region could be due to the ample water supply in this region (Supplementary Fig. 1).

Comparing the contributions of different factors to future carbon sink potential helps identify where efforts should be focused to maximize forest carbon uptake. In addition to constant drivers such as terrain and soil conditions, forest age was found to be the main determinant of carbon accumulation20,36. The effects of climate and CO2 fertilization, while present, are relatively small with high uncertainty. The additional carbon stock provided by new PFs is also limited (equivalent to half of that in existing PFs), due to the limited area suitable for PFs in Guangdong37.

As the most economically developed province in China, Guangdong faces the severe challenge of mitigating rising carbon emissions while balancing competing land use demands across various sectors38,39,40. Guangdong’s annual carbon emission reached 585.81 million tons in 2019 (Carbon Emission Accounts and Datasets; https://www.ceads.net.cn/), is projected to grow steadily until 2025 or 203038,41,42. This necessitates exploring measures to enhance forest carbon sequestration through appropriate management strategies without expanding the plantation area21. Our results suggest that progressive forest management is an effective strategy for maximizing carbon uptake, as harvest and replanting can alter forest age structure while maintaining or even increasing accumulation rates. For instance, Scenario S3 shows a 1.7-fold increase in carbon sequestration compared to Scenario S0. Unlike clear-cutting, partial cutting can maintain high soil carbon storage while also supporting timber production and ecosystem services43. This scenario also offers a viable solution for restoring forest cover in areas where economic benefits are a consideration44. We acknowledge that our estimates of potential carbon storage might be overestimated, as we did not account for likely carbon loss from harvested wood products (HWPs). HWPs play a key role in climate change mitigate, but estimates of their impact vary significantly among different studies45,46. Our estimate represent the upper limit of potential carbon storage, various practical trade-offs and end uses of HWPs are likely to prevent these scenarios from reaching their maximum potential.

Maintaining the current area of PFs unchanged has significant ecological implications, since large-scale afforestation programs can alter ecosystem water balance by increasing water demand and evapotranspiration47, even in southern China where water supply is typically sufficient48. Additionally, monoculture PFs generally support lower biodiversity, soil erosion control, and water yield compared to native forests49. Thus, preserving primary and naturally regenerated forests is essential to maintaining ecosystem services and mitigating negative influences50. Applying different forest management scenarios tailored to specific regions could offer a solution for multi-purpose forest management. Our initial attempt shows that this approach could contribute to considerable carbon sink potential while balancing biodiversity conservation and land-use constraints (Supplementary Note 7).

In our study, the identified plantation forests lack those older than 40 years due to the absence of satellite imagery. Although this does not affect our key findings, it is suggested that forest height products or forest inventory data could be utilized to identify older plantation forests for more accurate estimates of future carbon sink potential9,51. Changes in soil organic carbon under different forest management scenarios were not accounted for. Afforestation or reforestation can enhance soil carbon accumulation52 or offset soil carbon losses over a rotation53, though there is significant uncertainty due to factors such as tree species composition, developmental stage, soil type, and management practices54,55. Moreover, we did not consider forest cover change resulting from illegal logging and small-scale encroachment by adjacent cropland, activities that are often undetectable with most satellite observations56,57. Our estimations could be refined by using higher-resolution remote sensing data, which is become increasingly available58,59,60.

Conclusions

This study provides a comprehensive assessment of plantation forests (PFs) in Guangdong since 1982, highlighting their spatial distribution, environmental drivers of carbon accumulation, and potential carbon sinks under various influences. We found that topographic factors significantly influenced carbon sequestration, with the North and West regions exhibiting higher carbon accumulation rates due to favorable environmental conditions and younger forest age structures.

Our scenario projections indicate that forest management scenarios will have a more substantial impact on future carbon sink potential than climate change or CO2 fertilization. By implementing selective logging and reforestation scenarios, existing PFs can optimize carbon sequestration by maintaining a younger forest age structure, without the need for expanding forested area. This approach offers a promising solution for balancing carbon mitigation with the ecological and economic realities of southern China, where land-use conflicts and water resource management are pressing concerns. While Guangdong was chosen for its specific characteristics and available data, the methodologies and findings from this study are intended to be applicable to other regions with similar ecological and climatic conditions.

Methods

Study area

The study region, Guangdong province (20°13′ ~ 25°31′N, 109°39′ ~ 117°19′E) is located in the southern part of mainland China, covering a land area of 1.8×107 ha (Fig. 1). Mean annual precipitation varies from 1300 to 2500 mm, and mean annual surface air temperature from 19°C to 24°C. Guangdong province is characterized by predominantly mountainous and hilly terrain, which accounts for nearly 60% of its total area. The northern region situated at higher in elevation, is drier and cooler compared to the southern region. For this study, Guangdong Province was divided into four sub-regions: the North, West, East and Pearl River Delta (PRD) regions, based on differences in topography, soil types, and climate conditions-factors that are crucial for forest growth (Fig. 1, Supplementary Note 1).

Identifying planted forests and stand age with associated biomass carbon

The Spatial Database of Planted Trees (SDPT), which compiles and synthesizes national and regional data to provide global spatial information on planted forests and tree crops61, was used to identify areas covered by PFs. Tree crops, such as apple orchards, which include fruits, nuts or other tree products were excluded from our analysis. Only planted forests designated for wood production or ecosystem services were considered. Additionally, we used the spatial database of planted forest62 to filter out natural forests, defined as areas where the percentage of planted forests is less than 50%. The excluded area was minimal, accounting for less than 0.01% of the total planted forest area.

To estimate total (aboveground and belowground) biomass carbon of the identified PFs, we used the aboveground biomass estimates from the European Space Agency (ESA) Climate Change Initiative (CCI) biomass products for 2018, with a spatial resolution of 100 m63. Belowground biomass was calculated using vegetation type-specific allometric models, which relate the below-ground biomass to above-ground biomass64. A carbon fraction of 0.456 was applied for biomass-carbon conversioz65. To ensure accuracy, we filtered out pixels showing a decreasing total biomass trend due to forest degradation and small scale logging66,67 by comparing biomass estimates in 2010, 2017, and 2018. This filtering ensured that our biomass growth model reflected natural growth, or a non-decreasing trend in accumulated biomass as the forest aged.

The stand ages of planted forests were estimated using a global map of plantation years with a spatial resolution of 30 m68, which provides the years since the most recent planting event, ranging from 1981 to 2019. This dataset has also been used to estimate forest age in previous studies69,70. To reduce uncertainty, we compared the estimated forest age with another recent 30 m forest age dataset51. Only pixels where the differences in estimated forest age was less than five years were included in further analysis (Supplementary Fig. 2, Supplementary Note 2). Clusters of pixels with a total area >9000m2 (approximately 1 pixel of the ESA-CCI product) were selected for developing the forest growth model (see next paragraph). For each selected cluster, we extracted the mean age, which ranged from 1 to 34 years old. To standardize the biomass carbon data and align with reality, we adjusted the values to start at or near 0 Mg C ha−1 for 1-year-old forests by subtracting the smallest value from all age class.

Following Heinrich et al. 28, we used a space-for-time substitution method to develop a forest growth model that relates forest stand age to total biomass carbon, using the Chapman–Richards model71. The average biomass carbon accumulation rate was calculated by dividing total forest biomass carbon by stand age (Supplementary Fig. 3). We validated the growth curves and carbon accumulation rates using recent field survey data (Supplementary Fig. 4, Supplementary Note 3).

Identifying the dominant drivers of forest growth

We explored 14 potential drivers, grouped into five categories, that can influence forest growth and carbon accumulation: (1) forest type derived from the SDPT database61; (2) terrain variables, such as elevation, slope and aspect, derived from the GMTED2010 dataset; (3) climate variables, including mean annual temperature (Tmean), accumulated precipitation (Prec), downward shortwave radiation (Srad), soil moisture (Msoil), Vapor Pressure Deficit (VPD), Palmer Drought Severity Index (PDSI), sourced from the TerraClimate dataset72; (4) soil fertility, covering soil organic carbon (OC), total phosphorus (TP) and total nitrogen (TN), obtained from the Soil Database of China for Land Surface Modeling73; (5) primary drivers of forest cover loss, such as commodity-driven deforestation, shifting agriculture, wildfire, forestry and urbanization74. These datasets have varying spatial resolutions (Supplementary Table 1) and were resampled to 100 m to match the pixel size of PFs. We conducted a multi-collinearity test to avoid including strongly correlated variables, which could result in misleading inference (Supplementary Fig. 5, Supplementary Table 3, Supplementary Note 4). Two common metrics were used for this test: the Pearson correlation coefficient and variance inflation factor (VIF)75. Variables with a Pearson’s R greater than 0.8 and a VIF exceeding 10 were excluded from the analysis (i.e., soil total phosphorus, VPD, PDSI, and aspect), while their physical significance was also considered (Supplementary Note 4). Additionally, we found that forest loss drivers were primarily concentrated in a single category, namely forestry (defined as large-scale forestry operations), and these variables were excluded from further analyses. As a result, nine drivers were selected for random forest modeling: forest type, elevation, slope, mean annual precipitation, soil moisture, mean annual temperature, shortwave radiation, soil organic carbon and total nitrogen.

Each environmental driver was reclassified into three groups (33th and 66th percentiles), and their effects on forest growth and carbon accumulation were tested by modeling total biomass carbon against forest age using the stratified Chapman–Richards model across these groups. We also applied the random forest (RF) algorithm to test the relative importance of these drivers in the whole study area and different regions. RF is a robust ensemble learning method known for handling complex interactions and non-linear relationships among variables. To account for potential spatial autocorrelation in geographic data, we implemented a spatial cross-validation strategy to ensure that training and testing datasets were spatially independent. Due to the large size of the geographic dataset, we followed an approach similar to that of28,76. A relative small sample size (n = 2600, equivalent to 1% of the total sample size) was used, and we performed multiple iterations (n = 99) with a reduced number of trees (n = 500) to balance between computational efficiency and result stability. Model performance was assessed using 10-fold cross-validations, with Root Mean Square Error (RMSE) and R2 used to evaluate model accuracy. To validate spatial autocorrelation, Moran’s I tests were conducted on model residuals to ensure that no spatially structured errors influenced the results (Supplementary Note 4). Variable importance was quantified using a permutation-based approach, which measures the change in model accuracy (RMSE) when each variable was perturbed (n = 10).

Estimating effects of climate change and CO2 fertilization on forest growth

The influence of climate change on carbon accumulation in PFs was estimated by varying Chapman–Richards growth model parameters according to their climatic dependency (Supplementary Table 4). We investigated the climatic dependency of these regional models’ parameters on various climate variables using a step-wise multiple regression analysis, considering the current environmental conditions averaged from 1982 to 2018 for each region. Following Chazdon et al. 12, we performed each regression model twice to account for the influence of mixed pixels with lower biomass carbon. In the first run, pixels identified as outliers (with negative residuals≤0.5*STD and positive residuals≥0.5*STD) were removed, and the regression was rerun to ensure robustness. Due to the inherent non-linearity of the Chapman-Richards model, small changes in parameters might lead to amplified changes in carbon accumulation. We conducted a parameter sensitivity analysis and the results show our estimation is rather robust (Supplementary Note 6, Supplementary Fig. 12).

To assess the influences of future climate on forest biomass accumulation, we used the WorldClim (v2.1) dataset77, which provides downscaled future climate data from the Coupled Model Intercomparison Project (CMIP6) at 30 arc seconds spatial resolution. We extracted mean projections of 13 Earth System Models (ESM) for three Shared Socioeconomic Pathways (SSPs) 1-2.6, 2-4.5 and 5-8.5 (Supplementary Table 5). To address biases in the ESM outputs for the present climate, we applied a trend-preserving approach to correct the future climate projections (Supplementary Note 5).

The effect of CO2 fertilization on carbon accumulation was estimated using the results of 1 pctCO2 experiments, which keep other climate forcing constant but assuming a 1% annual increase in atmosphere CO2 concentration from preindustrial level78. Following Xu et al. 20, we calculated the ratio of future vegetation carbon, derived from ESM outputs (Supplementary Table 5), to current vegetation carbon. The changes in PFs carbon were estimated by multiplying 1- ratio with the biomass carbon attributed solely to forest aging. The projected future forest biomass carbon is the cumulative result of three factors: forest age, climate change, and CO2 fertilization (Supplementary Note 6).

Carbon uptake of different forest management scenarios

To evaluate how different forest management approaches affect the carbon sink potential of existing PFs, we developed five different scenarios by altering harvest and regeneration strategies (Table 1, Supplementary Note 7). They are: Preservation (S0): all existing PFs and suitable land remain unchanged; Preservation and Expansion (S1): preserve all existing PFs and establish new plantations on all suitable land since 2019 using the dominant forest type within a 10-kilometer radius. Forestry-suitable land consisted of grasslands and barelands derived from a 30 m satellite-based landcover map79. Complete Harvest and Reforestation (S2): all existing PFs and suitable areas were clear cut simultaneously, then planted with the original or nearest forest types. Selective harvest of mature PFs (S3): annually harvest and replant PFs older than 30 years from 2019 to 2060 with the original forest types; selective harvest of middle-aged PFs (S4): similar to S3, but only harvest and replant trees over 20 years old, while preserving younger PFs. For all scenarios, it is assumed that the biomass carbon of harvested forest is preserved as accumulated carbon without decaying. Carbon loss from wood products and replanted soils is not considered, representing the maximal potential for CO2 accumulation. Optimizing rotation period is a common forest management strategy to enhance carbon sink potential. For subtropical forests, rotation periods typically range between 20 to 40 years8,80,81,82. The preservation approach, which avoids conversion, is recognized as a sustainable long-term strategy that prevents deforestation, protects biodiversity, and maintains carbon storage. In contrast, clear cutting method is a common practice in industrial forestry that offers short-term economic benefits but is associated with increased carbon emissions and environmental degradation43.

Table 1 Forest management scenarios
Full size table