Determinants of reforestation from 2000 to 2010 were mostly consistent with expectations (Table 1 and Supplementary Table 1), but there were still some non-negligible differences compared to the tropics. Reforestation increased on land that was flatter, at lower elevation, and with lower agricultural revenue which correspond broadly to the results of determinants of reforestation in the tropics (Busch et al., 2019). The highest amount of reforestation was achieved on land with medium forest cover. Reforestation was significantly greater near cities, which was also the case in Latin America (Supplementary Table 1). Specifically, reforestation was significantly greater near cites in the Western Economic Belt (Supplementary Table 1). However, the reforestation was not significantly greater near the cities in the Central Economic Belt and the Eastern Economic Belt (Supplementary Table 1), which were consistent with the findings of determinants of reforestation in the tropics (Table 1). Reforestation was significantly greater at lower elevation in the Western Economic Belt, which was not confirmed in the Central Economic Belt and Eastern Economic Belt. Moreover, unlike in the tropics, reforestation was not greater within multiple protected area and strictly protected area. This may be due to the fact that large-scale reforestation programs were not on protected land as there was no need or a lack of space for extensive reforestation within protected areas.

In general, response of reforestation to carbon price was lower in China than in the tropics (Figure 1). Response of reforestation to carbon price in China was relatively close to that of the tropical regions of Asia and Latin America. Specifically, the response of reforestation to carbon price in the Eastern Economic Belt was close to that of the tropical regions of Africa. What’s more, the response of reforestation to carbon price was gradually decreasing from Eastern Economic Belt, Central Economic Belt to Western Economic Belt. According to Table 1, the price salience coefficient of reforestation on potential agricultural revenue was −0.00014 in China, which indicated reforestation was negatively related to potential agricultural revenue. This result is broadly consistent with the findings of determinants of reforestation in Central Economic Belt, Eastern Economic Belt, Latin America, Africa and Asia (Table 2).

However, the coefficient of the price salience of potential agricultural revenue on reforestation was 0.00016 in the Western Economic Belt, which indicated reforestation was positively related to potential agricultural revenue. This result which confirmed the symbiosis of agriculture and reforestation in Western Economic Belt was broadly inconsistent with the findings of determinants of reforestation in Central Economic Belt, Eastern Economic Belt, and the tropics (Latin America, Africa and Asia).

According to Figure 2, the price salience of reforestation was greater in land with a higher proportion of private reforestation and output value of non-state enterprises. This is due to the fact that private reforestation is usually more responsive to price than government programs.

We projected 126.48 Mha of reforestation in 2020–2060 under a BAU scenario (Supplementary Figure 2). This projected reforestation corresponds to 29.59 GtCO₂ removals from reforestation from 2020 to 2060 (Supplementary Figure 3). The 93.82 Mha of reforestation from 2020 to 2050 in China was equal to 24.19% of the tropical reforestation (387.8 Mha) (Busch et al., 2019) over the same time period. Moreover, the 19.94 GtCO₂ removals of reforestation from 2020 to 2050 in China was equal to 19.45% of the tropical removals of reforestation (102.5 GtCO₂) (Busch et al., 2019) over the same time period.

A carbon price of US$20 tCO₂⁻¹ would motivate land managers to enhance reforestation in China by 3.35 Mha (2.65%) to 129.83 Mha from 2020 to 2060 compared to a business-as-usual scenario (Figure 3). This corresponds to enhancing CO₂ removals by 0.49 GtCO₂ (1.66%) to 30.08 GtCO₂, from 2020 to 2060 (Figure 4). The increased removal (0.29 GtCO₂) from reforestation at a carbon price incentive of US$20 tCO₂⁻¹ from 2020 to 2050 is much lower than in the tropics (5.7 GtCO₂ from 2020 to 2050) (Busch et al., 2019). However, the removal of 20.24 GtCO₂ from reforestation from 2020 to 2050 is still equivalent to 18.69% of that in the tropics (Busch et al., 2019) (108.3 GtCO₂ in 2020–2050) at a price incentive of US$20 tCO₂⁻¹. A carbon price of US$50 tCO₂⁻¹ would motivate land managers to enhance reforestation in China by 8.53 Mha (6.74%) to 135.01 Mha from 2020 to 2060 compared to a business-as-usual scenario (Figure 3). This corresponds to enhancing CO₂ removals by 1.26 GtCO₂ (4.26%) to 30.85 GtCO₂, from 2020 to 2060 (Figure 4). The removal of 20.69 GtCO₂ from reforestation from 2020–2050 is equivalent to 17.58% of that in the tropics (Busch et al., 2019) (117.7 GtCO₂ in 2020–2050) at a price incentive of US$50 tCO₂⁻¹.

The low-cost abatement potential of reforestation enhanced from 2020 to 2060 (Figure 5). Enhanced removals at US$20 tCO₂⁻¹ are 109.52% higher in 2030–2040, 213.95% higher in 2040–2050 and 324.79% higher in 2050–2060 than in 2020–2030. Enhanced removals at US$50 tCO₂⁻¹ are 109.53% higher in 2030–2040, 213.93% higher in 2040–2050 and 324.73% higher in 2050–2060 than in 2020–2030. This percentage increment is clearly greater than in the tropics (Busch et al., 2019) (101, 188%) at US$20 tCO₂⁻¹.

Enhanced removals from reforestation are 0.01236 GtCO₂/yr. and 0.03149 GtCO₂/yr at US$20 tCO₂⁻¹ and US$50 tCO₂⁻¹ carbon prices from 2020 to 2060 in China (Figure 5 and Table 3), respectively. Although our projection of low-cost enhanced removals for China is lower than for the United States, Southeast Asia, South Asia, Sub-Saharan Africa, Brazil, and Canada, it is closer to that of Central America and higher than that of East Asia, Japan, Oceania, and Russia (Table 3; Austin et al., 2020). Enhanced removals from reforestation are 0.12875 GtCO₂/yr and 0.33865 GtCO₂/yr at US$ 20 tCO₂⁻¹ and US$ 50 tCO₂⁻¹ carbon prices from 2030 to 2060 in Brazil (Austin et al., 2020), which ranks first in the world. In comparison to tropical results, our projection of low-cost enhanced removals for China is only lower than for Brazil and Indonesia from 2020 to 2050, but is higher than the remaining 86 tropical countries such as India, Democratic Republic of the Congo, Mexico, and Australia (Busch et al., 2019). This result confirms that low-cost enhanced removals in China are a non-negligible part of global climate mitigation efforts in the future. Notably, total carbon dioxide removals through reforestation in China is 20.237 GtCO₂ from 2020 to 2050 (24.92 GtCO₂ from 2030 to 2060) at US$20 tCO₂⁻¹ carbon prices, which is close to and lower than the low-cost dioxide removal potential through reforestation of Brazil (Busch et al., 2019) (24.55 GtCO₂ at US$20 tCO₂⁻¹ from 2020 to 2050; Busch et al., 2019). This result confirms that total carbon dioxide removal potential from reforestation in China will be slightly inferior to tropical Brazil. China is likely to be one of the global leaders in mitigating climate through reforestation in the next 40 years, largely due to previous large-scale reforestation programmes that will continue to remove large amounts of carbon dioxide in the future, even if carbon prices are equal to 0 (removals from reforestation under BAU).

The Western Economic Belt (1.2767 GtCO₂) accounts for 42.44% of low-cost removals from 2020 to 2060 at US$20 tCO₂⁻¹ (Table 4), followed closely by Central Economic Belt (0.9495 GtCO₂, 31.56%) and Eastern Economic Belt (0.7823 GtCO₂, 26.00%). The Western Economic Belt (1.3098 GtCO₂) comprised 42.46% of low-cost CO₂ removals from 2020 to 2060 at US$50 tCO₂⁻¹, followed closely by Central Economic Belt (0.9717 GtCO₂, 31.50%) and Eastern Economic Belt (0.8035 GtCO₂, 26.04%). Western Economic Belt also comprised 43.32 and 43.21% of low-cost enhanced removals from reforestation respectively, followed by Central Economic Belt (29.15 and 29.07% respectively) and Eastern Economic Belt (27.73 and 27.72% respectively) at US$20 tCO₂⁻¹ and at US$50 tCO₂⁻¹ (Table 4).

The cost potential of enhanced CO₂ removal varied significantly across provinces (Figures 6, 7). The top 10 provinces for enhanced CO₂ removals (representing 73.19 and 73.22% of total removals) from reforestation between 2020 and 2060 at US$20 tCO₂⁻¹ and US$50 tCO₂⁻¹ were Yunnan, Sichuan, Guangxi, Guizhou, Hunan, Guangdong, Heilongjiang, Jiangxi, Fujian, Zhejiang (Figures 6, 7).

Potential agricultural revenue and percent forest cover showed significant effects on reforestation in the robustness test, which coincided with the results of the regression analysis, although distance to city, slope, elevation, multiple protected area, and strict protected area did not significantly affect reforestation (Table 1).

A relative comparison between our estimates and those in reference (Busch et al., 2019) was used for calibration, according to the methodology of reference (Lu et al., 2022). Our calibration results (Figure 8) show that carbon removals (InterQuartile Range: 0.00 ~ 135.77 tCO₂ ha⁻¹ and 0.00 ~ 135.74 tCO₂ ha⁻¹, respectively, in US$20 tCO₂⁻¹ and US$50 tCO₂⁻¹) from reforestation at carbon prices of US$20 tCO₂⁻¹ and US$50 tCO₂⁻¹ are close to those (InterQuartile Range:16.89 ~ 120.82 tCO₂ ha⁻¹ and 32.89 ~ 123.10 tCO₂ ha⁻¹, respectively, in US$20 tCO₂⁻¹ and US$50 tCO₂⁻¹) in Busch et al. (2019). Notably, the range (20.45 ~ 85.18 tCO₂ ha⁻¹) of our estimated Mean ± SD is slightly lower than that (40.90 ~ 91.89 tCO₂ ha⁻¹) of reference at carbon prices of US$20 tCO₂⁻¹. Moreover, the range (20.65 ~ 86.58 tCO2 ha⁻¹) of our estimated Mean ± SD is also slightly lower than that (47.46 ~ 101.44 tCO2 ha⁻¹) of reference (Busch et al., 2019) at carbon prices of US$50 tCO₂⁻¹. This may be due to the fact that unit carbon accumulation rates in Hainan are generally lower than the average values of carbon accumulation in typical tropical rainforests (Amazon Basin, Congo, and Southeast Asia), given the marginal location of the tropical climate (Busch et al., 2019). The difference between the Chinese secondary forest growth samples and the tropical secondary forest growth samples database caused this bias (Anderson-Teixeira et al., 2021). Moreover, the difference in response of reforestation to carbon prices on state-owned land and privately owned land is also an important cause of this bias.

(1) Cross-spatial validation of carbon removals. We validated carbon removal using Kling-Gupta Efficiency (KGE) based on carbon removal through reforestation estimated by land-use data, given that KGE addresses several perceived shortcomings in the Nash-Sutcliffe efficiency (NSE) (Knoben et al., 2019). Moreover, given the lack of forest cover data for 2010–2020, our estimates were validated by carbon removal estimated by 2010–2020 land use data. According to Table 5 and Supplementary Table 4, the value of KGE about cross-spatial carbon removals through reforestation in 2010–2020 is 0.43. (2) Cross-temporal validation of carbon removals. The value of KGE is 0.67 and 0.080, respectively, in 2035 and 2055 (Table 6). These results confirm that this study is a “satisfactory” model simulation (Knoben et al., 2019; Vergopolan et al., 2021; Koppa et al., 2022).

Among the global actions and plans for climate mitigation in the past decades, the tropical rainforests of Brazil, Indonesia, and Democratic Republic of the Congo have attracted lots of global attention due to their vast forests and carbon pools (Gullison et al., 2007; Murdiyarso et al., 2015; Rochedo et al., 2018; Pennisi, 2020; Staal et al., 2020; Heinrich et al., 2021; Koh et al., 2021; Masuda et al., 2021; Steckel et al., 2021; White et al., 2021; Koch and Kaplan, 2022). However, our projection of low-cost enhanced removals (0.01236 GtCO₂/yr. at US$ 20tCO₂⁻¹, 0.03149 GtCO₂/yr. at US$ 50tCO₂⁻¹) for China in the future is higher than all tropical countries except Brazil and Indonesia, such as Democratic Republic of the Congo, India, Mexico (Busch et al., 2019; Austin et al., 2020). Our projection of low-cost removals through reforestation in China from 2020 to 2060 is 30.08 GtCO₂ at US$20 tCO₂⁻¹ carbon prices, which is 1.23 times that of Brazil from 2020 to 2050 (24.55 GtCO₂ at US$20 tCO₂⁻¹; Busch et al., 2019).

Previous studies speculated that reforestation projects on public land (China and India), might not be as responsive to price signals (Busch et al., 2019). That government-led reforestation policies do not usually respond to cost/price signals is accepted by lots of Chinese researchers and mainstream views (Feng et al., 2021), which constrains the replication of cost/price studies in China. They believe that reforestation will not respond to price incentives under land ownership by the state and a strict land regime in China (Wang et al., 2018; Wu et al., 2018; Yang et al., 2021). However, the reality is not entirely so. First, government-led large-scale reforestation projects are the main driver for enhancing forest cover. Despite the inherently policy-driven nature of these reforestation projects (the Natural Forest Conservation Program, Three-North Shelter Forest Program, National Key Ecological Function Area Program, etc.), limited government investment forced project managers to incorporate cost considerations. The government is less willing to undertake reforestation in regions with high potential agricultural income, due to higher costs of arable land acquisition and food security (Cao, 2011; Yan, 2019; Tong et al., 2020; Winkler et al., 2021). Moreover, land management rights are possessed by enterprises, farmers, foresters, farms, and forest farms, who are the direct managers of the land, although all land is owned by the state in China (Chow, 1997; Feng et al., 2021). We have found that these enterprises, farmers, foresters, farms, and forest farms are responsive to price signals (or cost compensation). Second, commercial timber plantations responding to the demand for wood are also an important driver of forest cover change. Due to the natural forest conservation policy, commercial timber plantations have extensively cultivated fast-growing species such as poplar and eucalyptus to generate profits and fulfill the timber demand. Third, the Sloping Land Conversion Program is also a significant driver of forest cover change. As this program provides subsidies for converting farmland into forests, many farmers opt for reforestation on their land to reap greater economic benefits.

The cost potential of carbon dioxide removal through reforestation in China has been overlooked in the past. This study not only confirms that MAC curves about carbon dioxide removal can also be applied to China as well, but that China has a huge cost potential to CO₂ removal through reforestation. In fact, the cost potential of CO₂ removal in China may not be limited to this. The large-scale reforestation programs implemented in the past 30 years will continue to remove large amounts of carbon dioxide in the future, even if the carbon price is equal to 0 (removals through reforestation under BAU), which offers significant potential for zero-cost CO₂ removal through reforestation. As per capita income crosses the inflection point of the Environment Kuznets Curve, the structure of economic growth and natural environmental needs have changed significantly in China (Wang et al., 2019; Yang et al., 2020). The idea of protecting nature has been widely accepted in China (O'Meara, 2021). These trends will broaden the cost potential of carbon dioxide removal through reforestation in China.

This study estimated the spatial extent and the potential of low-cost CO₂ removal through China’s reforestation. We estimate that reforestation could remove 0.9848, and 1.0158 GtCO₂ yr⁻¹ at US$20 tCO₂⁻¹, and US$50 tCO₂⁻¹ in China from 2050 to 2060, respectively. These cost potentials are separately equal to 9.56 and 9.86% of CO₂ emissions of China (10.3 GtCO₂ yr⁻¹) in 2019 (Jiang et al., 2020; Shan et al., 2020; Yang et al., 2020; Liu et al., 2022; Tang et al., 2022). Moreover, the potential for removals through reforestation (1.0 GtCO₂ yr⁻¹ from 2050 to 2060) at US$50 tCO₂ are separately equal to 67.72 and 35.03% of CO₂ emissions of China under the 1.5°C (1.5 GtCO₂ yr⁻¹) and 2°C (2.9 GtCO₂ yr⁻¹) goal in 2050 (group, P. s. r. w., 2020; Duan et al., 2021).

Our estimates indicate that reforestation has greater low-cost abatement potential in 2060 relative to direct CO₂ capture & storage (0.03–0.3 GtCO₂ yr⁻¹ at US$ 390–415 tCO₂⁻¹) and bioenergy with carbon capture and storage (BECCS) (0.3–0.7 GtCO₂ yr⁻¹ at US$ 390–415 tCO₂⁻¹) (Zhang et al., 2022). Although removals (1.0158 GtCO₂ yr⁻¹) through reforestation at US$ 50 tCO₂⁻¹ are equivalent to conventional fossil fuel carbon capture and storage (0.9–1.5 GtCO₂ yr⁻¹) at US$ 390–415 tCO₂⁻¹ in 2060 (Zhang et al., 2022), the cost of removal through reforestation is much smaller than conventional fossil fuel carbon and storage.

Regions vary in price salience of reforestation for mitigating climate change. Price salience intensity of reforestation in China (0.00014) was lower than in the tropics (0.00026). The price salience of reforestation to carbon price is gradually decreasing from the Eastern Economic Belt, Central Economic Belt to the Western Economic Belt. The Eastern Economic Belt is the region with the most developed economy and the most active private economy in China. A large number of free trade zones, special administrative zones, special economic zones, and coastal open cities gather in the Eastern Economic Belt (Chen et al., 2018). The percentage of the output value of non-state enterprises is 79.17, 64.87, and 52.73%, respectively, in the Eastern Economic Belt, Central Economic Belt, and Western Economic Belt. Moreover, the percentage of private reforestation areas is 66.33, 49.88, and 54.95%, respectively, in the Eastern Economic Belt, Central Economic Belt, and Western Economic Belt. In fact, in the central and western regions, the state economy is extremely important, while the private sector is often weaker than that of the Eastern Economic Belt (Li et al., 2018). The active degree of the private economy is an important reason for the evolution of reforestation price salience.

The price salience coefficient of reforestation on potential agricultural revenue is 0.00016 in the Western Economic Belt, which is broadly inconsistent with the findings of determinants of reforestation in tropical (Latin America, Africa, and Asia), Central Economic Belt, and Eastern Economic Belt (Busch et al., 2019). An important reason for this result is the difference in natural conditions. The default non-agricultural ecological land cover in most of the tropics, Central Economic Belt, and Eastern Economic Belt is forest. In wetter areas such as the tropics, Central Economic Belt, and Eastern Economic Belt, where natural conditions are adequate for forestry and agriculture, the expected income becomes the key to the choice between forest or agriculture. Thus, there is a trade-off between agriculture and reforestation on the same land in the Eastern Economic Belt, Central Economic Belt, and the tropics. However, most of the Western Economic Belt is located in ecologically fragile areas with very low precipitation and temperature and over 3,000 meters in altitude (Herzschuh et al., 2019; Wu et al., 2020). This fragile natural condition imposes additional constraints on agriculture and forests. That is, regions, where natural conditions cannot satisfy forests, are usually difficult to satisfy agriculture. This constraint forces a symbiosis between agriculture and vegetation in ecologically fragile areas (Ran et al., 2013). Due to the fragility of ecosystems, sustainable agriculture depends on stable ecosystems covered by vegetation. Revegetation resulting from carbon price incentives will reduce wind and sand, soil erosion, and vegetation degradation, and enhance ecosystem stability (Wu et al., 2020), which will simultaneously enhance the sustainability of agriculture and ultimately farm income (Wen et al., 2019). However, the choice of vegetation and restoration methods (plantation or natural regrowth) is still uncertain in ecologically fragile areas, which is important for both carbon dioxide removal and agriculture.

Notably, the price salience of reforestation may face potential challenges and constraints that cannot be fully anticipated. Some studies around the world (White et al., 2018; Grainger, 2020) pointed out that not owning land may inhibit long-term investment such as reforestation, unless the carbon compensation to offset the reforestation investment is received in the period of the management right. Notably, the term of private contracts for forests across China is 30–70 years (Xu and Hyde, 2019). In addition to all the other potential challenges of long-term investment (economic, financial, policy, etc.) that private operators need to face, China’s forest operators will be exposed to the risk of international carbon price fluctuations in the future, which may not be easily coped with by their domestic operational experience from the past. These risks will need to be estimated and quantified in future studies. Carbon compensation payments (purchasing forest-based carbon offsets) have additionality in Brazil (Sonter et al., 2015), Congo (Hubau et al., 2020), Indonesia (Steckel et al., 2021), etc., as these countries face substantial deforestation and forest degradation. The additionality makes carbon compensation payments attractive. However, carbon compensation payments may not have significant additionality in China, due to the widespread presence of government-led large-scale afforestation (Zhang et al., 2000; Wang et al., 2020; Winkler et al., 2021).

Despite our efforts to incorporate major factors (Griscom et al., 1794; Busch et al., 2019) (biome type, elevation, slope, protected areas, forest age, price salience, etc.) influencing land use in the estimates, it’s important to note that personal preferences, religions, cultures, political instability, economic uncertainties, and other socioeconomic factors, which are challenging to quantify, may also play a role in affecting land-use change (Lu et al., 2022). These factors may introduce additional uncertainty into our estimates. Nonetheless, with carbon neutrality being identified as a long-term national strategy (Mallapaty, 2020), carbon removal through forests has become one of the goals of forest management in state forests (Tong et al., 2020). The spatial extent and potential for low-cost CO₂ removal through China’s reforestation can help operators (state or private) prioritize the most cost-effective actions.