Patterns and controlling factors of soil carbon sequestration in nitrogen-limited and -rich forests in China—a meta-analysis

Study sites description

China has a wide variety of climates due to its vast size. Summers are hot everywhere except the highlands and high mountains, and winters are severely cold in the north, mountains, and plateaus. Summer is the wettest season, with the exception of the vast desert areas in the West, where precipitation is limited and erratic (Tian et al., 2019). The southeast receives the greatest precipitation, while the northwest receives the least. The average annual temperature is 21 °C (range from 13 °C in January to 28 °C in July), and the climate is monsoonal humid tropical, with 1.930 mm of precipitation falling in a clear seasonal pattern (75% from March to August, and 6% from December to February). The soil type is lateritic red earth produced from sandstone (Oxisol) (Tian et al., 2019). Forests cover around 22% of China’s total land area. China has a total forest area of around 206.861.000 hectares, with approximately 5.6% (or 11.632.000 hectares) designated as primary forest. Between 1990 and 2010, China increased around 49.720.000 hectares (31.6%) of forest cover (Qin, Huang & Luo, 2010). Huge swathes of southern China are covered with forest, stretching from Fujian Province in the east to Sichuan and Yunnan Provinces in the west. Much of China’s far northeast is likewise forested. Forest cover declines considerably between the densely populated cities of Shanghai and Beijing and the distant western provinces of Xinjiang and Tibet (Zhu, 2017).

Data collection

We searched the Web of Science and Google Scholar for peer-reviewed journal publications published between 2000 and 2022 that investigated the patterns and controlling factors of soil carbon sequestration in Chinese forest (Table S1, Fig. S1). This period was chosen because it corresponds to the start date of the Gain for Green project (GFGP) or the Natural Forest Protection Program. We utilized terms like “patterns and controlling factors”, “soil carbon sequestration”, “nitrogen-limited and nitrogen-rich forests”, “nitrogen enrichment”, “nitrogen deposition”, “nitrogen addition”, and “nitrogen fertilization” as well as the keywords “forest” and “carbon”. Overall, 1,203 papers were obtained for this study, but only 61 were retained following the screening stage to exclude irrelevant publications (Fig. 1). There were eight variables in the database notably forest types, N addition forms (NH₄NO₃, Urea), N addition rates (30, 30–70, and >70 kg N ha⁻¹ yr⁻¹), N deposition rates (i.e., 10, 10–20, and >20 kg N ha⁻¹ yr⁻¹), N fertilization durations (1, 1–3, and >3 years), and N deposition rates were all compared. Soil depths (cm) ranged from 0 to 10, 10 to 20, and 20 to 40 cm, with experimental durations ranging from 6 to 15, 15 to 35, and >35 years. Data was collected as previously described in Supplement File and Ngaba et al. (2022).

Analysis and statistics

The meta-analysis approach outlined by Hedges, Gurevitch & Curtis (1999) was used to evaluate the patterns and controlling variables of soil carbon sequestration based on the natural log-transformed response ratio (RR) using the METAWIN version 2.1 (Sinauer Associates, Inc., Sunderland, MA, USA). In addition, we calculated the relative influence technique to calculate the relative importance of predictor factors of soil C sequestration (independent variables). The formulas applied has been used in previous works (Ngaba et al., 2022). Statistical analyses and graphs were determined with OriginPro 2021 and SPSS 26.0 software (SPSS, Inc., Chicago, IL, USA). Correlations were considered significant at p < 0.01 and p < 0.05.

Effect of N enrichment soil N use efficiency (NUE) and soil organic carbon (SOC) sequestration

In our meta-analysis, N addition significantly affected NUE (P < 0.05; Fig. 2A). The estimates of NUE varied between 0.24 and 13.3 kg C kg⁻¹ N depending on the forest type. Furthermore, NUE response to N addition decreased significantly in subtropical forests (P < 0.05), N form (NH₄NO₃), high C:N ratios (>15), and long-term application (>3 years) (all, P < 0.01, Fig. 2A). At the same time, it significantly increased in low application rate (P < 0.05). On the other hand, there was a positive relation between NUE and C/N ratio response after adding N fertilizer but had no significant effect (Fig. 2A). On the other hand, N addition increased soil organic carbon sequestration by an average of 6%, but the effect was insignificant (Fig. 2B). The enrichment was positive for tropical and subtropical forests (+7% and +2%, respectively). However, when the data were separated into low N fertilization rate (30 kg N ha⁻¹yr⁻¹; +2%) and high ambient N deposition rate (>20 kg N ha⁻¹yr⁻¹; +12%), N addition had a positive and significant effect (Fig. 2B; P < 0.05). The nitrogen use efficiency and mean annual precipitation (MAP) have significant linear relationships (R² = 0.34, P < 0.01) (Fig. S2).

Patterns of soil C sequestration in response to N addition

The results showed that N addition significantly increased C sequestration across Chinese forest ecosystems (P < 0.05; Fig. 2C), with values ranging between 6.8 and 11.2 kg C kg⁻¹ N (Fig. 2C). Also, C sequestration response was significantly higher in temperate forests (10.6:8.37–13.68 kg C kg⁻¹ N) compared to subtropical forests (5.87:4.83–7.16 kg C kg⁻¹ N). Figure 2C also highlights that C sequestration was dependent on N form, N addition levels, (R² = 0.34, P < 0.01) rate, and N application duration (P < 0.05; Fig. 2C). In particular, this study showed that NH₄NO₃ form, low N rates, high NUE levels, and short-time N application strongly influenced C storage processes (Fig. 2C).

Soil C sequestration in response to N addition rates and duration

The rate of fertilization influenced the responses of the C dynamic to N addition. In general, a low application rate (<30 kg N ha⁻¹yr⁻¹) did not significantly increase soil C concentration in Chinese forests (+4%). While high application rates (>70 kg N ha⁻¹yr⁻¹) reduce soil C concentration by 7% (P < 0.01), especially in NH₄NO₃ form (−9%) and long-term application (>3 years, −2%) (Fig. 3A; P < 0.01). However, C accumulation was more pronounced in temperate forests (−3%) compared to subtropical (−7%) (Fig. 3A, P < 0.01). When the data was subdivided by N application duration, we found that short-term application increased soil C concentration in Chinese forests by +5% (Fig. 3B; P < 0.01), while long-term application decreased by −10% (Fig. 3B; P < 0.01).

Controlling factors of soil C sequestration

Overall, we found a negative correlation between SOC concentration, bulk density, pH, and sand content (P < 0.01; R² =  − 0.59, R² =  − 0.94, R² =  − 0.58 for Figs. 4A, 4D and 4F, respectively). While it was positive with total nitrogen (TN), dissolved organic carbon concentration (DOC), clay, silt content, and soil aggregate size (R² = 0.94, R² = 0.94, R² = 0.13, R² = 0.67, R² = 0.66 for Figs. 4B, 4C, 4E, 4G and 4H, respectively). However, SOC concentration was higher in macro-aggregates than in micro-aggregates (Fig. S3). Our meta-analysis showed that although soil carbon concentration varied with season types, it increased with forest age and elevation but decreased with soil depth layers (Fig. 5). Concerning climate factors, SOC concentration had a negative and significant relationship with mean annual temperature (MAT) but a positive relationship with mean annual precipitation (MAP) depending on their magnitude (P = 0.91, Fig. S4). In addition, the relative influence analysis indicated that N availability (40%) is one of the major factors controlling C accumulation, followed by DOC > MAP > MAT > clay > pH > sand > silt (Fig. 6).

Ecosystem types, N addition levels, and duration drive C sequestration

Understanding the changes in forest nitrogen (N) deposition rates has important implications for C sequestration. Our meta-analysis showed that N addition significantly increased soil C sequestration in the temperate forest compared to the subtropical forest (Fig. 3), as Fleischer et al. (2015) demonstrated. This could result in how N availability relates to forest productivity because ecosystem productivity and species composition vary widely in China as a result of regional variations in temperature, precipitation, and other physiographic factors. For instance, forest productivity is determined by the amount of N available in the soil for plant use (Binkley & Fisher, 2019); thus, high rates of N availability result in rapid forest development. Evidence of this trend can be seen in Fig. 2; soil C sequestration increases with the level of NUE. According to Zhu et al. (2021), NUE induced by N addition in Chinese terrestrial ecosystems is mainly influenced by the aridity index and precipitation.

In comparison, a gradual decrease in nitrogen deposition in the future will cause a drop in the net carbon sequestration. In general, the sink for soil carbon sequestration is often linked to plant productivity and aboveground litter inputs, which is the case in temperate forests (Lu et al., 2021). The low sequestration rate observed in subtropical forests can result from excess N accumulation and N leaching in the soil because soil microorganisms do not assimilate inorganic excess N added. According to Matson, Lohse & Hall (2002) and Wu et al. (2022), over-enrichment with N can increase N loss via trace-gas emissions (e.g., denitrification, volatilization) and solution leaching. As a result of the loss of N inputs, NUE will decrease and lead to low N retention rate fractions in subtropical ecosystems (Bai, Houlton & Wang, 2012).

On the other hand, forest productivity increases when increasing water availability is involved. Temperatures and rainfall in China tend to follow the same pattern from southeast to northwest, with higher rates in subtropical forests than in temperate forests (Li et al., 2022b). Together, the high availability of carbon in N-rich forests and the high water supply could induce high nutrient uptake by the rhizosphere. This process is aided by nutrient recycling, which converts organic to plant-available mineral forms; nutrient storage in SOM; and physical and chemical processes that regulate nutrient sorption, availability, displacement, and eventual losses to the atmosphere and water (e.g., nitrogen fixation) (Canton, 2021). These conditions favor faster decomposition and less SOM accumulation (Cusack et al., 2011) in the soil and, consequently, soil C sequestration (Frey et al., 2014).

Our meta-analysis reported that soil C sequestration was less pronounced in soils with high N fertilization rates (>70 kg N ha⁻¹ yr⁻¹) (Fig. 3), implying that soil C accumulation efficiency declines with N addition added (Hyvönen et al., 2007). Reduced C allocation due to N enrichment may be due to root physiology, exudation modification, and rhizosphere priming effects (Janssens et al., 2010; Kuzyakov, 2002). Considering all available evidence, it appears probable that elevated N addition decreased loss of soil C in temperate forests. Sustainable forest management practices could store the most carbon and keep the quality and productivity of the forest ecosystem (Adam Langley, Chapman & Hungate, 2006; Rillig & Allen, 1999) and result in positive or negative effects on plant growth. Several mechanisms could explain soil carbon gains following N deposition in N-limited temperate forests: the effect of N enrichment on soil acidification and mineralization of soil organic C by microorganisms by influencing nutrient demand (Lu et al., 2021); plant growth stimulation, as well as shifts in the composition and functions of the fungal community (Tian et al., 2019). For instance, acidic deposition is known to influence the growth of plants, and the quality of soil suffers as a result (De Vries et al., 2009). A reduction in soil pH promotes an increase in forest floor thickness at numerous locations since roots are unable to penetrate the acidic mineral soil, and organic residues decompose more slowly on the soil surface than in the soil (Nieder & Benbi, 2008). Therefore, soil acidity changes will alter the equilibrium between the metabolism of the soil microorganisms and plant roots. Considering all evidence findings, it appears probable that elevated N addition decreased loss of soil C in temperate forests. Sustainable forest management practices could store the most carbon and keep the quality and productivity of the forest ecosystem (Diao et al., 2022; Xue, Kimberley & McKinley, 2022).

Overall, our results showed that short-term N enrichment enhanced soil C sequestration, implying that microbial composition and function activities enhance fast carbon cycling. However, the evidence was only shown in the temperate sequestration rate. Our findings contradict those found by Tian et al. (2019). The study revealed that high-and long-term N-enrichment resulted in an increase in soil organic carbon sequestration due to the reduction in relative abundances of bacteria and genes that degrade cellulose and chitin. The large discrepancy suggests that (i) N forest status may influence the response of C allocation to belowground under N input; (ii) fungal and bacterial responses differ following specific site conditions. Indeed, there is a long debate about whether N addition increases or decreases bacterial biomass in subtropical forest soils, fungal/bacterial ratios, and its significant effect on complex organic SOC compounds (Wang, Liu & Bai, 2018; Zhu et al., 2014). Moreover, forest productivity is stimulated by N input until N saturation is reached. The response of C sequestration to N enrichment is also a function of the other N input sources. It has been well demonstrated that N input in soil has three major sources: atmospheric N deposition, weathering of soil minerals, and, for nitrogen, biological N fixation (Binkley & Fisher, 2019). The sequestration of carbon in soils depends on a number of abiotic or biotic factors.

Abiotic and biotic controlling factors and implications for soil C sequestration

According to the hierarchy concept, soil aggregate-related OC regulates SOM dynamics, which is the primary agent for soil aggregation (Six & Paustian, 2014). Likewise, determining SOC sequestration in different sized aggregates helps evaluate soil quality or degradation of soils. In general, the accumulation of SOC is mainly attributed to the physical protection of OM (and thus OC and N) from environmental and microbial attack or mineralization (Six & Paustian, 2014). Previous studies have shown that SOC sequestration in the forest ecosystem strongly correlates with soil aggregate stability, as reflected by the MWD in this study (He et al., 2021). Similarly, our findings showed that SOC was significantly higher in macroaggregates than in microaggregates in Chinese forest types because of the OM binding microaggregates into macroaggregates (Li et al., 2020). This result suggests a slower turnover rate and lower decomposability of SOM in macroaggregates (Ngaba, Bol & Hu, 2021). It highlights the high rate of root decomposition within macroaggregates, increasing C sequestration. According to Liu et al. (2018), macroaggregates contain more C and young labile C than microaggregates because the new input OC is first associated with microaggregates and then incorporated into macroaggregates. However, our findings contradict those found by He et al. (2021). This difference could be attributed to the surface area size and the binding agents because soil microorganisms could not break down and utilize the OM associated with the micro-aggregates. For example, Six et al. (2004) argued that the increased specific surface area of soil micro-aggregates might improve OM adsorption from root exudates and litter residues.

Soil depth, season changes, time, and elevation were among the most important controlling factors of SOC sequestration in the Chinese forest (Figs. 5 and 6). Generally, SOC sequestration decreased with soil depth in all aggregate fractions. Our study fully agrees with this trend. We found that the pattern of SOC decreased gradually with the increase of soil depth layers, indicating that climate effects are predominantly in the topsoil (close to the surface, 0–20 cm). This decrease could be attributed to the reduction of environmental impacts in deep soil layers through increasing soil buffering ability (Yang et al., 2010); the variation of plant litter C input (both the surface litter and below the fine ground root) (Tang et al., 2011) or the slower-cycling SOC in deep soil layers (Jobbágy & Jackson, 2000). Climate gradients and altitude influenced SOC distribution by affecting litter generation and breakdown processes (Ngaba et al., 2022; Zhao et al., 2020). However, tree species and forest age are two key variables affecting carbon sequestration in forest ecosystems through their impact on the balance between C input and output.

In agreement with Yu et al. (2014), our study found a negative association between SOC and bulk density (BD), pH, and sand content (Fig. 4). This outcome can be explained by numerous causes, including increased soil porosity, (ii) root growth, and (iv) soil aggregate stability (Liu, Shao & Wang, 2011; Ngaba, Bol & Hu, 2021; Yu et al., 2014; Liu, Shao & Wang, 2011; Ngaba, Bol & Hu, 2021; Yu et al., 2014). On the other hand, their combined influence has a direct effect on SOC. According to Xiao (2015), soil aeration pattern is determined by the interactive influence of clay content and texture, total soil porosity, or soil bulk density, which affects soil microbial respiration. Increases in soil BD, for example, have been shown to be deleterious to tree growth due to the way they alter other soil properties. As a result, soil C accumulation is often negatively related to BD, possibly due to greater rooting (root growth) in improved systems. This trend has highlighted the importance of soil BD for tree growth and, by extension, C sequestration as illustrated by Binkley & Fisher (2019) in a savanna region of central Brazil. Because of its effect on soil water and oxygen availability, the bulk density of the soil influences SOC decomposition via mineralization (Baldock & Skjemstad, 1999).

The processes of decomposition and nitrification are regulated by the pH of the soil (Jackson et al., 2005; Zhou et al., 2019). Soil microorganism activity is dependent on the structure and diversity of the soil microbial population. The breakdown processes of SOM were considerably influenced by soil pH, resulting in a negative correlation between SOC and pH (Jackson et al., 2005). Soil organic matter buildup, in particular, would accelerate the rate of soil acidification. This finding suggests that the buildup of SOC in the soil profiles that follow can be aided by the substantially lower pH values of the soil. Soil erosion influences sand content, making it a useful indicator of soil degradation across a variety of land-use types (Zhou et al., 2019). Therefore, the SOC pattern can be changed by any factor that affects soil quality. Similarly, compacted soils are more resistant to being broken down by root systems. The activities of roots, aerobic microbes, and animals can all be suppressed by reduced aeration in compacted soils (Kuzyakov, 2002; Tian et al., 2019). Compacted soils have lower water infiltration rates and may develop anaerobic conditions in puddled areas.

Climate factors such as mean annual temperature (MAT) and mean annual precipitation (MAP) are widely regarded as the most important determinants of soil organic carbon accumulation at large geographical scales and over time because they regulate inputs from litter production and outputs from decomposition. Likewise, NPP and OM input into the soil are both influenced by climate. Microbial activity in soil is also an essential factor in the climate’s influence on SOC storage (i.e., the SOM decomposition rate) (Jackson et al., 2005; Jobbágy & Jackson, 2000). Our meta-analysis showed that soil C sequestration increases with precipitation and decreases with temperature (Fig. S5) because the increasing precipitation stimulates vegetation growth and productivity as a result of an increased SOC sequestration rate (Chen et al., 2022; Liu, Shao & Wang, 2011). Our findings were consistent with global trends (Jobbágy & Jackson, 2000). Previous studies have demonstrated that SOC sequestration varied significantly with vegetation types. For example, during an assessment of the patterns and environmental controls of soil organic carbon, according to Chen et al. (2016), the average SOC sequestration in P. crassifolia forest and grassland at a soil depth of 0–50 cm was 23.68 and 10.11 kg m⁻² respectively.

Our findings showed that SOC was low at 250 mm compared to 500 mm, indicating that SOC accumulation in Chinese forests is mainly restricted by water availability. Namely, increasing precipitation may reduce SOC decomposition by reducing microbial activity because high soil water availability may limit microorganisms’ access to oxygen and hinder SOC decomposition (An et al., 2022; Poeplau et al., 2018; Yu & Zhuang, 2020). In addition, SOC sequestration decreases with temperature due to SOM decomposition temperature dependence. Increased soil temperature accelerates the decomposition of soil organic matter, which decreases the amount of OM stored in soils by transferring soil C into the atmosphere (Davidson et al., 2008; Ngaba, Ma & Hu, 2020). Hence, lower temperatures could result in reduced SOC breakdown. SOC sequestrations tend to accumulate along the temperature gradient, according to Ge et al. (2020), since the temperature-induced increase in soil carbon decomposition is substantially smaller than the increase in vegetation production.

Implications and uncertainties of the study

The balance between input and production of C in the soil is critical to soil C sequestration. In the temperate and subtropical forests of China, the response to nitrogen deposition was estimated to be between 5.9 and 10.7 kg C kg⁻¹ N. The magnitude of other forest ecosystem types has been reported in previous studies. De Vries, Du & Butterbach-Bahl (2014), in tropical forests, they reported that soil C sequestration was 5.4 kg C kg⁻¹ N, which is in line with the value range suggested by Hermundsdottir & Aspelund (2021) of 8.6 to 10.5 kg C kg⁻¹ N. Soil C sequestration varied between 3 and 25 kg C kg⁻¹ N in tropical forests (Frey et al., 2014; Janssens et al., 2010), and between 10 kg C kg⁻¹ N in an N-limited forest (Maaroufi et al., 2015). The study conducted by Du & De Vries (2018) will bring more clarity by specifying that N deposition leads to a larger C sink in the temperate forest (0.11 Pg C yr⁻¹) than in the tropical forest (0.08 Pg C yr⁻¹), which is in line with our findings.

Uncertainties might remain in our study when determining the carbon sequestration rate due to the lack of information on the fraction of N retention in Chinese forest ecosystems. Indeed, no study to date has determined this parameter, which led us to use the values obtained in Europe (50% and 15% for temperate and subtropical forests, respectively) (De Vries, Du & Butterbach-Bahl, 2014; De Vries et al., 2007). It is well known that biological processes control N retention in forest ecosystems, but these processes vary from one site to another. Therefore, although its values were determined in the same forest types, several factors (such as the nitrogen deposition rate, tree species types, climate conditions, vegetation, and microbial uptake) due to the specific sites conditions and environment could influence N retention rate. Moreover, heterotrophic demand for N, soil properties, nonbiological retention of N, land-use change, or plant community can also significantly impact N retention, for instance (Barrett & Burke, 2002; Johnson, Cheng & Burke, 2000; Johnson, 1992; Silver et al., 2005; Templer et al., 2008). In addition, experimental results on root respiration, microbial respiration, and SIC were relatively scarce and shed limited insights on the actual N deposition effect in belowground C forest ecosystems. Nevertheless, our results represent the best knowledge we can derive and give an overview of carbon sequestration in response to the nitrogen deposition in Chinese forest ecosystems.