Considering the complexity of forest ecosystem carbon cycle mechanism, and the data to support model building and validation, as well as initial model parameters need to be obtained from a large amount of measured data. Most forest carbon cycle models are parameterized or validated by gross primary productivity (GPP), net primary productivity (NPP) or net ecosystem productivity (NEP) from flux observations. Forest age is rarely taken into account, resulting in fewer climate-driven models being able to predict forest carbon sequestration rates (Cao and Woodward, 1998).

Here, we used a forest carbon sequestration (FCS) model on the basis of a classical logistic equation between forest age and biomass, which is parameterized and validated by large volumes of vegetation data from field surveys in China (He et al., 2017). The FCS model was established by measuring carbon sink data and forest age from more than 3300 sample plots of the Chinese Academy of Sciences Carbon Special Project as input parameters. The model was validated by data from 78 forest successions in China and combined with important factors such as temperature, precipitation and forest age. The construction, validation and predictions of the FCS model have been published in previous studies (He et al., 2017; Yan et al., 2020; Cai et al., 2022), making it possible to assess the carbon sequestration of forest ecosystems or afforestation in China rapidly and accurately. The model’s construction and operation process are briefly described below. A detailed description can be found in previous studies (He et al., 2017).

With forest development, forest biomass gradually reaches a relative equilibrium state. The relationship between vegetation biomass and forest age can be portrayed using a logistic growth equation (Figure 1) (Xu et al., 2010). Vegetation biomass can be calculated by (1):

where B_t is the forest vegetation biomass (Mg ha^-1); V₀ is the intrinsic growth rate, representing the maximum growth rate (%) when vegetative growth is not limited by the environment, nutrients, or disturbances; B_max is the maximum vegetation biomass under the mature forest scenario (Mg ha^-1); and t is the forest age (a).

Annual variation in soil carbon stocks is calculated by the annual input of organic matter minus the annual decomposition, which is the classical double pool of soil carbon cycle. When forest factors are relatively stable, the humification and mineralization processes will tend to balance the SOC level, as shown in (2):

where C_t is the SOC density (MgC ha^-1); I_t is the annual input of SOC (Mg ha^-1 a^-1); I_t = h (k₁ L_t), L_t is the litter content (Mg ha^-1); k₁ is the litter decomposition coefficient; k₂ is the SOC decomposition rate (a^-1); and h is the decay coefficient (0.3) (Foley, 1995).

The main parameters of the FCS model are the initial vegetation biomass (B₀), stand age (t), mean annual temperature (MAT, ℃), and mean annual precipitation (MAP, mm). The data for the existing forest vegetation biomass were obtained from the field survey of the “Strategic Priority Research Program of the Chinese Academy of Sciences” (XDA05050000), including the initial vegetation biomass, forest age, litter, and 0-20 cm SOC content in 3365 forest sample plots in China, which are the same as the special features of Proceedings of the National Academy of Sciences of the USA in 2018 (Tang et al., 2018) They included deciduous broadleaf forests (DBF, 806 plots), deciduous needle-leaved forests (DNF, 197 plots), evergreen broadleaf forests (EBF, 620 plots), evergreen needle-leaved forests (ENF, 1461 plots), and mixed needle-leaved and broadleaf forests (NBF, 281 plots). The plots were each 0.1 hm², and vegetation biomass was calculated by the relevant allometric equation using diameter at breast height and tree height (ECSP, 2015).

Climate data, including the mean annual temperature (MAT, ℃) and mean annual precipitation (MAP, mm), were obtained from the National Climate Center (http://ncc.cma.gov.cn/cn/), and simulated using the Regional Climate Model system (RegCM 4.0).The data was then output (spatial resolution of 1° × 1°) through one-way nesting of the Beijing Climate Center_Climate System Model Version 1.1 (BCC_CSM1.1) (Gao et al., 2012).

Population and gross domestic product (GDP) per capita data for each provincial-level region (Data for Hong Kong, Macao, and Taiwan were not included) were derived from China Statistical Yearbook 2020 (http:/www.stats.gov.cn/tjsj/ndsj/), and the CSR per capita and CSR per economic was calculated for each provincial-level region (Data for Hong Kong, Macao, and Taiwan were not included).

A coefficient of 0.5 had been used to transfer biomass to carbon density (Yu et al., 2014). First, the tool ‘Extract Multi Values to Points’ in ArcMap was used to extract MAT and MAP of each site under future climate scenarios, and the future carbon sequestration capacity of forest ecosystems in each provincial-level region (hereafter province) of China was predicted by FCS model. The changes in carbon density or CSR of vegetation and soil were calculated for every 10 from 2010-2060. The results were represented as averages of three future climate scenarios. Meanwhile, in order to meet the targets of carbon peaking in 2030 and carbon neutrality in 2060, the period 2010-2030 and 2030-2060 were used in the data processing for the analysis.

The spatial distributions of the forests and their carbon densities were mapped using ArcMap 10.2 (ESRI, Redlands, CA, USA). The data were analyzed and plotted on graphs using SPSS 25.0 (IBM Corp., Chicago, IL, USA) and Origin 2018 (Origin Lab, Northampton, MA, USA).

Forest ecosystems in different Chinese provinces are predicted to have the capacity to sequester carbon from 2010-2060 under different climate scenarios, with keep the existing forest area constant. The vegetation CSR varied between <0.01 TgC a^-1 (Shanghai) and 25.67 ± 0.91 TgC a^-1 (Heilongjiang Province) (Table 1). Soil CSR varied between <0.01 TgC a^-1 (Shanghai) and 15.36 ± 0.94 TgC a^-1 (Yunnan Province) (Table 2). Overall, the CSRs in forest ecosystems over the next 50 years were the largest in Heilongjiang Province (36.74 ± 1.35 TgC a^-1) and the smallest in Shanghai (<0.01 TgC a^-1). The carbon sequestration potential of forest ecosystems in most provinces varied significantly during the period, which caused inter-provincial differences in CSRs (Table 3).

Compared with the 2030-2060 period, CSR per unit area of the forest ecosystem in Chinese provinces larger varied from 2010-2030 (Figure 2). The provinces in the northwest (Xinjiang, Tibet, Qinghai) and southwest (Yunnan and Guizhou) had smaller changes in forest CSR per unit area (0.32-2.15 MgC ha^-1 a^-1), while the provinces in the eastern region are larger (Jiangsu and Anhui) (3.05-3.83 MgC ha^-1 a^-1). The result revealed that the per capita CSR of provinces varied from 0.01-1.50 MgC a^-1 per capita in the 2010-2030 period, and from 0.01-2.52 MgC a^-1 per capita in the 2030-2060 period (Figure 3), with significant differences among provinces.

Overall, provinces such as Tibet, Inner Mongolia, Heilongjiang and Jilin showed higher CSRs per unit of economic whereas the more economically developed cities on the eastern coast have lower CSRs (Figure 4).

There are significant negative correlations between the CSR per capita of each province and its GDP per capita. Provinces such as Heilongjiang and Inner Mongolia have a higher CSR per capita, while more developed ones (such as Beijing, Shanghai, Jiangsu and Tianjin) are lower (Figure 5).

During the 2010-2060 period, results show a large regional imbalance in the CSR between provinces in China, with the CSR of forest vegetation varying between 0.01 and 25.67 TgC a^-1 and that of soil varying 0.01-15.36 TgC a^-1 among provinces. The CSR in forest ecosystems was the largest in Heilongjiang Province (36.74 ± 1.35 TgC a^-1) and the smallest in Shanghai (<0.01 TgC a^-1). This is closely related to economic development and climatic conditions in each province. Economic development inevitably comes at the risk of damaging current forests, while climatic conditions determine whether forests are suitable for growth.

The results estimating the CSR of forest vegetation at the national scale differed significantly from earlier studies (Xu et al., 2010; Pan et al., 2011; Fang et al., 2014; Hu et al., 2015), but were similar to the results of recent studies (Tang et al., 2018; Yao et al., 2018). Varying data sources and methodologies are important reasons for the differences. None of the earlier studies at the national scale have systematically estimated CSRs of forest ecosystems in Chinese provinces under future climate scenarios, especially for soil. More importantly, most earlier studies estimated forest CSR at the national scale without providing inter-provincial data clearly (Xu et al., 2017; 2018a; 2018b; 2019), which made it impossible for us to compare between provinces. Regarding methodology, the FCS model was based on a classical logistic equation to establish the relationship between forest age and biomass, and was parameterized and validated using numerous field measurements in China, which could better simulate the changes in CSR during the natural succession of forest ecosystems (He et al., 2017; Yan et al., 2020; Cai et al., 2022). Meanwhile, the FCS model also considered the influence of future climate change on forest ecosystems by incorporating data under different future climate scenarios. However, the model only uses mean annual climate parameters, and is not sensitive to the overall response of climate data. As for the data source, the forest sample plots were obtained from a large number of field surveys conducted by the “Climate Change: Carbon Budget and Relevant Issues” of the Chinese Academy of Sciences, which were carried out in strict accordance with field survey specifications and were highly representative (ECSP, 2015). However, there are some uncertainties in this study; the definition of mature forest age is a widely debated scientific issue and we set the age of mature forest stands at 100 a (Liu et al., 2014), which may have allowed the carbon sequestration potential of forests to have been over or underestimated in some provinces. Although previous studies have shown that forests mature at 100-400 a (Guariguata and Ostertag, 2001), some studies showed that defining the mature forest across regions is difficult due to differences in species, forest type and climate (Martin et al., 2016). Theoretically, FCS model can be used to carry out more systematic simulations on basis of different forest types and species, but there are still uncertainties in the spatial data products available at the national scale for these parameters, which need to be improved and refined in future studies. As the economy and public awareness develop, forest area may increase from largescale afforestation/reforestation and ecological restoration or decrease in response to social development demands, but we assume that the existing forest area in each region will remain unchanged for the next 50 a, which may lead to underestimation of the results.

Forests in different Chinese provinces, especially Heilongjiang, Sichuan and Yunnan, have great potential for carbon sequestration over the next 50 years. However, to maintain and increase the carbon sequestration potential of the provinces’ forests, long-term effective forest management techniques and new socio-economic policies are needed.

In general, forest management practices to conserve and sequester C can be grouped into three major categories: (i) maintaining existing C pools (e.g., slowing deforestation and forest degradation), (ii) expanding existing C sinks and pools (e.g., increasing C density by modifying forest structure and growth processes), (iii) creating new C sinks and pools by expanding tree and forest cover (afforestation and reforestation) (Peng et al., 2008). As the CSR of forests may vary among provinces (Tables 1-3), strengthening the research and development of technologies with the function of enhancing forest carbon sinks according to its characteristics is urgent. When the CSR of forests is low, appropriate technical measures such as forest soil fertilization, soil improvement, forest thinning, and rational forest harvesting (Figure 6) can be used to not only increase CSR but also lengthen such periods. In addition to clarifying the principles and approaches for improving the capacity of forest carbon sinks, it is important to quantitatively assess the sustainability and consistency of forest carbon sinks, which is the new requirement for sink enhancement technologies for carbon neutrality. Carbon neutrality is a long-term goal; temporary and effective sink enhancement measures should be promoted with caution, because inappropriate forest management practices and short-term sink enhancement measures may also reduce the carbon sink of forests or even convert them to carbon sources (Hyvonen et al., 2007). Meanwhile, carbon neutrality should be integrated with the improvement of ecosystem quality. The development and promotion of carbon sink enhancement technologies that may cause damage to ecosystem quality should be treated with caution.

Through natural restoration and growth, forest ecosystems can not only improve the ecological environment and forest health, but also achieve the long-term forest carbon sink, which is highly desirable (Jin et al., 2020). However, we must pay great attention to their long-term sustainability, disturbance factors and potential risks. Forest management strategies must be developed for each province according to local conditions. Clay et al. (2019) observed that reasonable harvesting and appropriate fires would increase the carbon sequestration capacity of forests, while excessive logging and extreme fires would do the opposite. A study on Pinus sylvestris found that heavy intercutting reduces photosynthesis as well as biomass and soil carbon fluxes, thereby reducing the positive effect of fertilizer application on carbon sink potential, so the interaction between intercutting measures and fertilizer application can be used rationally and is important in guiding management to increase forest carbon sink (Jörgensen et al., 2021). Moreover, forest fire policies should be revisited to optimize the fire-disturbing properties of long-term carbon sinks. Moreover, low-intensity fires must be actively employed to improve forest structure, enhance forest productivity, and boost carbon sequestration (Wright et al., 2020). Extensive outbreaks of pests and diseases also have a significant impact on the structure and quality of forest ecosystems, which can greatly threaten the long-term carbon sink of forest ecosystems (Hyvonen et al., 2007). Biological protection should be fully utilized and strengthened to avoid the loss of carbon sinks caused by pests and diseases as far as possible. In conclusion, to realize the carbon sequestration potential of forests in each province from 2020-2060 (or even longer), we must attach importance to the harmonious development of human beings and nature and actively promote forest conservation. We must follow the two-pronged approach of “increasing” and “preserving” sinks while also adopting scientific and reasonable long-term forest management measures that are appropriate to each province.

Most of the regions with high contributions to carbon neutrality are economically underdeveloped or less economically developed provinces. Therefore, we must not expect to rely on traditional carbon trade (e.g., Clean Development Mechanism, CDM) to make up for the huge imbalance in the inter-provincial carbon sinks. Firstly, compared with the European Union (EU) and other countries, China’s carbon trade market is not perfect and its policies are less flexible. Secondly, there are strong regional differences in China’s naturals, society and economy, which are completely different from those of developed countries in Europe and the US. Thirdly, the carbon trade only captures a small portion of carbon compared to ecosystem carbon sinks; it is therefore difficult to reflect its contribution to carbon neutrality targets. Besides strengthening the carbon trade system, we should also take special measures such as carbon sink compensation or an eco-environmental compensation tax to address the regional imbalance of carbon sinks (Figure 6). It must be ensured that people in western regions or underdeveloped provinces are willing to establish long-term forest carbon sinks, consciously protect and enhance them, and prevent the development of these provinces to achieve the national ecosystem carbon sink target. This will result in poverty in these areas due to the long-term protection of forest carbon sinks and the restriction of economic development. Once the phenomenon of “carbon sink poverty” emerges, it will reduce the subjective willingness of people in the region to protect forests and carbon sinks, making it difficult to realize the long-term carbon sequestration potential of forest ecosystems and affect the national strategic goal of carbon neutrality (Tong et al., 2020). In conclusion, the national and local governments should focus on the regional imbalance based on the CSR per capita, economic development level, climate and soil characteristics. Novel measures such as national carbon trading markets and inter-regional carbon sink compensation in policy making should be optimized. We recommend choosing a pair or pairs of provinces and cities with low GDP but high forest CSRs and high GDP but low forest CSRs, and compensating in both directions for common development. We also recommend providing effective development support to economically underdeveloped provinces and increasing the carbon sequestration potential of the forest ecosystems so that they can make greater contributions to achieving carbon neutrality. This will promote synergy among several strategic objectives such as national rural revitalization, coordinated regional development and common prosperity to help achieve the country’s double hundred goals.