Estimating carbon storage of desert ecosystems in China

ABSTRACT

Although desert ecosystems are commonly considered to have low carbon storage owing to their sparse vegetation and limited carbon sequestration capacity, they comprise 13.5% of China's terrestrial ecosystems and cover approximately 35% of the Earth's land surface, so they still possess significant carbon storage. The carbon storage of desert ecosystems in China was estimated using MODIS-NDVI data in combination with ground survey, soil census, and literature statistical data. The results indicated that the carbon storage of desert ecosystems in China was 7.063 Pg in 2020. One-meter soil carbon storage accounted for the largest proportion (89.514%), approximately 6.322 Pg, followed by vegetation carbon at 0.741 Pg, and biological soil crust carbon at 0.002 Pg. The carbon density of desert ecosystems was 33.895 Mg.ha⁻¹, with soil carbon density, vegetation carbon density, and biological soil crust carbon density at 30.330, 3.554, and 0.011 Mg.ha⁻¹, respectively. This research provides important evidence for the correct understanding of the carbon storage of desert ecosystems in China.

KEYWORDS:

• Desert ecosystems
• vegetation carbon storage
• soil carbon storage
• biological soil crust carbon storage

1. Introduction

Desert ecosystems are representative of arid regions. Harsh environmental conditions, such as high temperature, low rainfall, uneven seasonal distribution of precipitation, and strong soil erosion, have led to poor soil fertility, low soil organic matter content, and low vegetation productivity in desert ecosystems. Therefore, it is generally believed that their carbon sequestration capacities and potentials are very weak. However, desert ecosystems occupy approximately 35% of Earth's land surface, and their total carbon sequestration is still considerable, playing an important role in the global carbon cycle (Liu et al. 2022). Similarly, it is the primary ecosystem type in northwest China, accounting for about 13.5% of terrestrial ecosystems in China. Hence, their contributions to China's carbon cycle cannot be ignored (Xu et al. 2017). Some studies have suggested that desert ecosystems act as carbon sources, emitting more carbon than sequester (Conant, Klopatek, and Klopatek 2000). However, many studies also have indicated that arid regions sequester more carbon than they release, functioning as carbon sinks (Wohlfahrt, Fenstermaker, and Arnone 2008; Xie et al. 2009; Xu et al. 2023).

Desert ecosystems are composed of xeromorphic and ultra-xeromorphic small trees, shrubs, semi-shrubs, and associated animals and microorganisms and are dynamic systems of material and energy flow. This unique desert environment creates extremely sensitive and fragile ecosystem that differs from forests, grasslands, wetlands, and other ecosystems. Owing to the limitations of water resources and environmental conditions, the continuous distribution of vascular plants on the surface is greatly restricted, resulting in a patchy and sparse pattern. This provides favorable conditions for the widespread growth of biological soil crusts (BSCs) (Li et al. 2012; Zhang and Wang 2010), which can dominate the surface cover in some extremely dry areas, covering more than 70% of the surface (Li 2012). As a result, a mosaic distribution pattern emerges between the biological soil crust and vascular plants (Eldridge and Greene 1994; West 1990). Despite their low primary productivity, desert plants possess the ability to absorb and fix atmospheric CO₂ (Ma et al. 2021). Additionally, BSCs can also fix carbon through the photosynthesis of cryptogams, such as cyanobacteria, lichens, green algae, and mosses (Li et al. 2012; Su, Wu, et al. 2013). Soil microorganisms, an important component of BSCs, not only aid in the formation of BSCs and improve biogeochemical cycles, but also significantly enhance soil carbon storage (Liu et al. 2012). This makes them an important source of carbon sequestration in desert ecosystems (Elbert et al. 2012; Li et al. 2012; Petrie et al. 2015). In addition, the development of surface BSCs inhibits the release of soil CO₂ into the atmosphere, reduces soil carbon emissions, and affects the processes of carbon exchange at the soil-atmosphere interface (Su, Wu, et al. 2013; Xu et al. 2022). Consequently, BSCs contribute to regional carbon sequestration and have a significant impact on ecosystems carbon cycling.

Accurate estimation of carbon storage in desert ecosystems is advantageous for gaining a better understanding of their carbon sequestration capability, which is vital for comprehending the role of desert ecosystems in the global carbon cycle, achieving carbon peaking and carbon neutrality (Piao, Yue, et al. 2022). For a long time, carbon storage in forest ecosystems and grassland ecosystems has been intensively studied, but desert ecosystems have not been studied much. Most existing studies have focused on soil carbon storage and its dynamics (Duan, Liu, and Qn 1996; Feng et al. 2005; Jia et al. 2019; Lal 2004). Studies on estimation of vegetation carbon storage have primarily focused on semi-arid and sub-humid arid regions with better vegetation growth (Xu et al. 2022; Yan, Wu, and Wang 2015; Ye et al. 2018; Zandler, Brenning, and Samimi 2015), and results differ greatly, which severely hampers our understanding of the size of the carbon pool in desert ecosystems and the characteristics of terrestrial carbon sources or sinks.

However, the integration of big data and the generation and fusion of multiple remote sensing indicators provide necessary data sources and technical support for estimating regional productivity (Cui and Li 2022; Li, Lu, and Jia 2021). The aboveground biomass of vegetation can reflect the productivity of terrestrial ecosystems and is the basis for studying the relationship between carbon sources and sinks. Remote sensing technology has become the preferred approach for large-scale estimation of vegetation biomass and carbon storage because of its extensive coverage, reliable and repeatable. However, it is challenging to accurately estimate the aboveground biomass of vegetation in desert ecosystems because of the sparse vegetation coverage, significant spatial heterogeneity, low vegetation biomass, and prevalence of dominant species with needle-like or scale-like leaves. Moreover, the extraction of information on sparse desert vegetation is easily affected by the soil background, further reducing the estimation accuracy of the remote sensing model. Although existing research can identify vegetation growth trends and distribution patterns, it has been unable to systematically evaluate vegetation productivity or quantify the size of the carbon pool in desert ecosystems. The accuracy of remote sensing technology for estimating the above-ground biomass of sparse vegetation is limited by various factors, including the quantity of ground survey data, changes in vegetation growth, and spectral resolution. These factors contribute to errors in the remote sensing data and uncertainties in the key parameters of remote sensing models, resulting in low accuracy. Considering the strong correlation between NDVI index and observed data in desertification region, this study used a large amount of ground survey data and remote sensing data to establish a remote sensing estimation model for aboveground biomass of desert ecosystem vegetation based on the vegetation index (NDVI) (Cunliffe, Brazier, and Anderson 2016; Mu et al. 2016; Sternberg et al. 2011; Yan, Wu, and Wang 2015; Ye et al. 2018; 2022; Zandler, Brenning, and Samimi 2015), can effectively improve the accuracy of biomass estimation for sparse vegetation and more objectively reflect the actual status of carbon storage in China's desert ecosystems.

In this study, we utilized MODIS-NDVI data, combined with 485 valid field survey quadrat data, 45 soil type data, and a large amount of literature data to estimate the carbon storage and its constituent components in China's desert ecosystems. The objective of this study was to provide a scientific basis for accurately understanding carbon storage in China's desert ecosystems and accurately assessing the role of desert ecosystem carbon pools in addressing climate change.

2. Study area

Desert ecosystems in China are primarily located in inland basins and plateaus between 75–106°E and 35–50°N, from the western end of the Tarim Basin to Helan Mountain. They cover a total area of 2.08 × 10⁶km². Desert ecosystems are mainly distributed in Xinjiang, Qinghai, Gansu, Inner Mongolia, Ningxia, and other regions (Liu et al. 2020; Zulkar, Zhao, and Jiang 2021). There are eight major deserts, and a large area of Gobi, including Taklamakan Desert, Gurbantunggut Desert, Kumtag Desert, Qaidam Basin Desert, Badain Jaran Desert, Tengger Desert, Ulan Buh Desert, and Hobq Desert.

This region is located deep in the Eurasian continent and is less affected by the East Asian summer monsoon. Moist air from the ocean is blocked by mountains, resulting in a typical temperate dry arid desert climate. The area has an annual precipitation below 200 mm and is categorized as arid and extremely arid regions (Lu et al. 2022). Precipitation is unevenly distributed throughout the year, with the majority occurring between June and August (summer) and accounting for more than half of the total annual precipitation.

3. Data sources and methods

3.1. Method for estimating vegetation carbon storage aboveground biomass

3.1.1. Ground survey

Considering the distribution area of desert ecosystems and the difficulty of data collection, we carefully selected specific representative areas to collect ground vegetation data. These areas covered both arid regions and extremely arid regions, including the northeast region of the Ulan Buh Desert, the southeast region of the Tengger Desert, the surrounding areas of the Kumtag Desert, and the Hami Depression (Figure 1). The dominant vegetation in these areas consists of shrubs and semi-shrubs such as Nitraria tangutorum, and Reaumuria songarica. Additionally, herbaceous plants such as Agriophyllum squarrosum, Grubovia dasyphylla, Salsola laricifolia are also present in these regions.

Figure 1. Sample map of the study area.

A quadrat (30 m × 30 m) sample plot method was used for the field investigation. The investigation included topographical features, soil characteristics, elevation, GPS coordinates of the central point, and vegetation growth status. Additionally, the base diameter (D), crown length axis (C₁), short axis (C₂), and height (H) of each shrub within the sample plot were measured and recorded. We selected various healthy plants of different sizes and species from the study area. The stem diameter, height, crown length (axis), and short axis were recorded. Whole plants were clipped to the ground, weighed, and transported to the laboratory for drying in a ventilated oven at 80°C for 48 h. Based on ground sample plot survey data, the individual aboveground volume of each shrub was calculated using the basal diameter (D), canopy area (A = πC₁×C₂/4), and plant height (H) as predictor variables to construct an allometric equation between aboveground volume and biomass dry weight for each shrub species. The coefficient of determination (R²) and mean absolute error (MAE) were used to compare and evaluate the accuracy of the allometric equation, and an optimal allometric equation was obtained. The allometric equation was used to derive the unit area aboveground biomass of shrubs from the survey data of each sample plot (Yao et al. 2021; Ye et al. 2018; 2022). A total of 628 plots of ground vegetation were investigated in arid and extremely arid areas from August to September 2007 to 2020.

For plots with herbaceous plants, three 1 m × 1 m subplots were randomly placed along a diagonal line, and the aboveground biomass was measured using the harvesting method.

3.1.2. Preparation of remote sensing data

MOD13Q1 data derived from MODIS satellite imagery with a spatial resolution of 250 m and a temporal resolution of 16 days were used in this study. The annual vegetation information from 2007 to 2020 was extracted, representing the growing season from the 145th to 289th day (May to September) of each year. MRT software was used to extract NDVI within a confidence band after image stitching and mask processing. There were ten MOD13Q1 images taken during the growing season every year, and the highest value for each pixel was selected as the indicator of the vegetation index data of the year, which was expressed as MODIS-NDVI to estimate the aboveground biomass of desert vegetation. The acquisition time of all the remote sensing data was matched with the vegetation survey time.

3.1.3. Model construction and accuracy verification

The GPS coordinates of all the ground vegetation quadrats were superimposed on the MODIS-NDVI images in the same year as the survey time of the ground quadrats and compared with Google Earth images. Quadrats containing land cover types, such as cultivated land, water bodies, or towns were excluded from the pixels of the ground quadrats. Sample plots located in pixels containing land cover types, such as croplands, water bodies, or urban areas, were excluded after comparing them with Google Earth images. After image screening, 485 effective quadrats were finally determined from 628 ground vegetation quadrats, among which 259 were from 311 quadrats in the arid zone and 226 were from 317 quadrats in extremely arid zone. Data from 70% of the valid ground survey sample plots were randomly selected using regression analysis to establish a correlation between the vegetation index and the measured aboveground biomass. The estimation models of aboveground biomass for both arid and extremely arid regions were obtained (Figure 2), and the remaining 30% of the valid ground survey sample plots were used to evaluate the accuracy of the estimation results. The logarithmic model of MODIS-NDVI (p < 0.001, R² = 0.70) was determined to be the most suitable for estimating aboveground biomass in arid regions (Figure 2(b)), while the linear model of MODIS-NDVI (p < 0.1, R²= 0.67) was found to be optimal for extremely arid regions (Figure 2(a)).

Figure 2. Regression model between MODIS-NDVI and aboveground biomass of vegetation.

3.2. Methods for estimating soil carbon storage

Based on the Second Soil Resources Census in China, soil organic matter content, gravel (particle size > 2 mm) content, and bulk density data of soil profiles at different depths were collected in accurately located desert areas (Office National Soil Survey 1998). The soil thickness as the weighting coefficient, average thickness (with a maximum calculated thickness of 1 m), average organic matter, and average bulk density of all soil types and subtypes were obtained by the weighted average. After obtaining the corresponding parameters, the total organic carbon storage of each soil type was calculated using the following formula: $C=0.58\times S\times H_i\times O_j\times W_j$where j represents the soil type, C_j refers to the carbon storage (t), S_j represents the total area (hm²), H_j is the average thickness (cm), O_j is the average organic matter content (%), W_j represents the average bulk density (g.cm⁻³), and 0.58 is the conversion coefficient between soil organic matter and organic carbon. The area (S_j) for each soil type in the arid and extremely arid regions was obtained from the 1:1,000,000 soil-type distribution map of China from the Chinese Resource and Environmental Database.

3.3. Methods for estimating of biological soil crusts carbon storage

The estimation of biological soil crust carbon storage follows the same methodology as soil carbon storage. The carbon density of the biological soil crusts was based on published data. If thickness and bulk density data were unavailable from the literature, average values for algae, lichens, and mosses were used as substitutes. Previous studies have shown that biological soil crusts account for 40%−70% of fixed and semi-fixed sand dunes in desert ecosystems (Li 2012). In view of the lack of accurate statistical data on the crust coverage of the entire desert ecosystems, the distribution area of biological soil crusts of desert ecosystems in China was estimated using 40% of the area of fixed and semi-fixed sand dunes as a baseline, as provided in the Atlas of Chinese Deserts (Administration State Forestry 2018).

4. Results and analysis

4.1. Vegetation biomass

The aboveground biomass of desert vegetation was estimated separately in arid and extremely arid regions using the established remote sensing estimation model based on MODIS-NDVI data from 2020. The estimated aboveground biomass of desert vegetation in China is approximately 0.228 Pg, with approximately 0.203 Pg in arid regions and 0.025 Pg in extremely arid regions.

Huang (Huang et al. 2006) simulated the spatial distribution patterns of total vegetation biomass, above- and below-ground biomass, and root-to-shoot ratio in China using the Atmospheric Vegetation Interaction Model (AVIM2). Based on the ratio of roots, stems and leaves in different desert communities in China, the average root-shoot ratio of desert vegetation was determined to be 5.5 (Huang et al. 2006), which was used to estimate below-ground biomass in our study. In 2020, the total biomass of desert ecosystems in China was approximately 1.481 Pg, with arid areas contributing 88.86% (Figure 3).

Figure 3. Distribution of above-ground biomass of desert ecosystems in 2020.

4.2. Soil organic carbon storage

Organic carbon density for each soil type was calculated in arid and extremely arid regions using data from The Second Soil Resources Census. The soil organic carbon storage of desert ecosystems in China was approximately 6.320 Pg.

4.3. Biological soil crusts carbon storage

The carbon density of biological soil crust in desert ecosystems exhibits significant spatial heterogeneity, primarily influenced by topography and vegetation. The carbon density was higher in inter-dune areas and areas with vegetation cover, especially in fenced areas where artificial sand-fixing vegetation was planted (Table 1). In order to reflect more accurately the average carbon storage in biological soil crust, we collected crusts data from literature in desert ecosystems worldwide and calculated carbon density. Globally, biological soil crust carbon density ranged from 3.05 g.m⁻² to 181.58 g.m⁻², with a median value of 49.26 g.m⁻² and an average value of 61.66 g.m⁻². Desert ecosystems in China, the biological soil crust carbon density ranged from 20.88 g.m⁻² to 181.58 g.m⁻², with a median value of 60.65 g.m⁻² and an average value of 74.60 g.m⁻² (Table 1). A higher carbon density of biological soil crust appeared in vegetation areas of the artificially stabilized dune in arid regions, ranging from 46.95 g.m⁻² to 181.58 g.m⁻². Based on the average carbon density of global biological soil crust and 40% of the fixed and semifixed dune areas in the Atlas of Sandy Deserts in China (2.083 × 10⁶km²) (Administration State Forestry 2018; Li 2012; Li, Hui, and Zhao 2016), the carbon storage of biological soil crusts of desert ecosystems in China was approximately 0.002 Pg.

Table 1. The carbon density of biological soil crusts in desert ecosystems all over the world.

Region	Crust distribution	Biocrust type	Density (g.m^−2)	References
Gurbantunggut Desert, China	Between dunes	Algae、Lichen、Moss	20.88	(Zhang et al. 2009)
	Between dunes	Algae、Lichen、Moss	63.74	(Zhang, Zhou, and Zhang 2015)
	–	Moss、Mixed crusts	57.56	(Su, Wu, and Zhang 2012)
	inter-dune	Algae、Lichen、Mixed crusts	51.61	(Zhang et al. 2016)
	–	Algae、Lichen、Moss	29.85	(Li, Wu, et al. 2021)
Tengger Desert, China	Artificial sand-fixation vegetation	Cyanobacterial–algal crusts	46.90	(Su, Li, et al. 2013)
	Inter-space of plant in Artificial sand-fixation vegetation	Algae、Lichen、Moss、Mixed crusts	181.58	(Du et al. 2018)
	Artificial sand-fixation vegetation	Crust	143.29	(Xie, Li, and Li 2021)
	Windward slope, leeward slope, inter-dune in Artificial sand-fixation vegetation	Crust	75.59	(Li et al. 2002)
Hobq Desert, China	inter-space of plants under enclosure and afforestation	Algae、Moss	74.97	(Niu et al. 2017)
Chinese Median			60.65
Chinese Mean			74.60
Tabernas Desert, Spain	inter-space of natural perennial and annual plants	Algae、Lichen、Moss	173.93	(Chamizo et al. 2012)
	inter-space of dune slope annual plants	Algae、Lichen	5.99	(Cantón et al. 2014)
	inter-space of natural perennial and annual plants	Algae、Lichen	18.20	(Miralles et al. 2012)
Natural Park, Spain	inter-space of natural perennial and calcaric outcrop and stones	Algae、Lichen	108.67	(Chamizo et al. 2012)
Israeli–Egyptian border	inter-dune	Crust	28.24	(Ram and Aaron 2007)
Negev Desert, Israel	inter-dune and mid-slope	Algae	22.31	(Drahorad et al. 2013)
Sonoran Desert, USA	–	Cyanobacteria-dominated BSCs	3.54	(Beraldi-campesi et al. 2009)
Colorado Plateau, USA	–	Cyanobacteria-dominated BSCs	3.02	(Beraldi-campesi et al. 2009)
Global Median			49.26
Global Mean			61.66

4.4. Carbon storage in desert ecosystems

After applying a conversion coefficient of 0.50 to the total vegetation biomass, the vegetation carbon storage of desert ecosystems in China was estimated to be approximately 0.741 Pg by 2020. Additionally, the soil organic carbon storage was 6.320 Pg, while the biological soil crust carbon storage was 0.002 Pg. The total carbon storage of desert ecosystems in China was 7.063 Pg. Based on the total area of 2.083 × 10⁶km² of desert ecosystems in China, the carbon density of desert ecosystems was 33.895 Mg ha⁻¹. The carbon densities of vegetation, soil, and biological soil crusts were 3.554 Mg ha⁻¹, 30.330 Mg ha⁻¹, and 0.011 Mg ha⁻¹ respectively.

5. Discussion

China's terrestrial ecosystems have served and will continue to serve as CO₂ sinks for a long time in the future (Piao, He, et al. 2022), but the contributions of different ecosystems vary greatly (Yang et al. 2022). At present, the resource inventory method, a classical method for estimating terrestrial ecosystems carbon storage, is widely used because it can separate the contribution of vegetation and soil to the sources and sinks of ecosystems in a certain period. However, the coverage density of ground survey data is a crucial limiting factor for the accuracy of this method (Piao, Yue, et al. 2022). By integrating extensive ground survey, literature, and national soil census data, the carbon pools of desert ecosystems in China were accurately estimated in this study. The total carbon pool (7.063 Pg) in this study was significantly higher than that estimated using empirical models for sparse herbaceous deserts (3.450 Pg), which accounts for approximately 7%−8% of the carbon pool in China's terrestrial ecosystems (Wang et al. 2021). The soil carbon pool in this study was higher (6.320 Pg) than sparse herbaceous desert (2.480 Pg) and the vegetation carbon pool (0.741 Pg) was lower than sparse herbaceous desert (0.970 Pg). In addition to differences in vegetation type, vegetation coverage, and productivity, environmental conditions such as high temperature, low precipitation, and low soil moisture are important constraints in slowing down the decomposition process of soil organic matter in desert ecosystems (Yang et al. 2022). Although the carbon pool of desert ecosystems in China is significantly lower than that of forests and grasslands (Yang et al. 2022), it is not lower than shrubland ecosystems (6.700 Pg) (Tang et al. 2018) or wetland ecosystems (5.4∼7.3 Pg) (Zheng et al. 2013). Further analysis revealed that the desert ecosystem vegetation carbon pool (0.741 Pg) was higher than that in wetlands (0.22 Pg) and shrubland (0.4∼0.7 Pg) estimated by the inventory method in China. Even though over 90% of the carbon in desert ecosystems is stored in the soil, a relatively smaller value is the primary reason for their low contribution to the global terrestrial ecosystem carbon pool.

Vegetation carbon contributed 10.49% of the total carbon pool in desert ecosystems in China, which was 6.09% higher than the vegetation carbon contribution (4.40%) in global desert ecosystems, whereas the corresponding contribution of soil carbon storage was lower by 6.12% (Carvalhais et al. 2014). There are two reasons for this finding. First, a uniform soil depth of one meter was used to estimate global ecosystem carbon storage (Carvalhais et al. 2014), whereas this study considered different soil depths due to soil types. Affected by geomorphic and geological processes, desert soils in China are mostly less than 100 cm thick, with sierozem having the highest thickness (104.4 cm) and chisley soil the lowest (11.7 cm). Second, arid regions are the key area of ecological engineering construction in China, especially windbreak forests, sand-fixation forests, and farmland shelterbelts, which not only improve and protect the ecological environment but also play an important role in improving regional vegetation coverage and productivity (Lu et al. 2018). Vegetation restoration and ecological restoration carried out can not only increase vegetation carbon pool, but also improve soil texture to increase soil carbon pool, while the increase rate of soil carbon pool is much slower than that of vegetation. Arid regions with better climate conditions contributed 88.86% of the vegetation carbon storage to desert ecosystems in China. Therefore, the protection and restoration of desert ecosystems have a great impact on improving regional environment, increasing carbon sink capacity and enhancing the role of desert ecosystems in carbon peaking and carbon neutrality (Piao, Yue, et al. 2022; Lu et al. 2018; Yang et al. 2022). Future climate change scenarios, such as precipitation increase and CO₂ fertilization, will boost vegetation coverage and productivity in desert areas (Brandt et al. 2019; Donohue et al. 2013; Li et al. 2015). This will enhance the contribution of desert ecosystems to the carbon sources and sinks of terrestrial ecosystems, particularly because of their high sensitivity to extreme climate or significant changes in inter-annual precipitation (Deng et al. 2021; Poulter et al. 2014).

Biological soil crusts, as a main component of desert ecosystems, inlay vascular plants, forming a stable state through competition for limited resources. It has been widely proven that biological soil crusts play a significant role in carbon sequestration (Elbert et al. 2012; Li et al. 2012; Liu et al. 2020). At regional scales, the carbon density of biological soil crusts exhibited high spatial heterogeneity under different climatic conditions (Table 1). Topography and vegetation are the main influencing factors at a small scale (Li et al. 2010). The stabilized land surface, dust capture, precipitation redistribution, and organic matter input all contribute to the development of biological soil crust resulting in a relatively higher carbon density in the artificial sand-fixing vegetation areas (Table 1). In desert ecosystems, the biological crust is discontinuous, mostly distributed under shrubs, and black or gray under drought conditions. Remote sensing analysis is easily affected by the shrub canopy and soil background values. Large-scale remote sensing monitoring of biological soil crust coverage is relatively lacking, and there have been only exploration attempts in the Mu Us Sandland and Gurbantunggut Desert (Bu et al. 2016; Wang et al. 2022; Zan 2012). It is challenging to quantitatively estimate the carbon storage of biological soil crusts, despite their importance in desert ecosystem carbon cycling cannot be ignored. The carbon storage of biological soil crusts in desert ecosystems is approximately 0.002 Pg, accounting for less than 1/1000 of the entire desert ecosystem’s carbon pool (Tang et al. 2018). Their relatively low contribution is due to the low carbon density resulting from their thin layer (Li, Fu et al. 2021; Li, Hui, and Zhao 2016; Weber, Büdel, and Belnap 2016). Nevertheless, the thin layer plays an essential role in stabilizing the land surface, enhancing nutrient content, reducing erosion, and sequestering CO₂ (Li, Hui, and Zhao 2016; Weber, Büdel, and Belnap 2016), particularly in increasing soil carbon storage. It cannot be denied that biological soil crust is a key component of soil organic carbon storage in extreme arid areas with very minimal vascular plants cover and soil organic carbon content (Monus et al. 2023). Even a small change in biological soil crusts can affect carbon exchange between vegetation and soil and atmospheric interference by affecting its ecological functions, leading to changes in the carbon or sink capacity of the ecosystem. Therefore, it would not be appropriate to ignore biological soil crusts carbon storage contribution in the calculation of carbon budgets for desert ecosystems.

6. Conclusion

The carbon storage capacity of desert ecosystems in China was 7.063 Pg in 2020, with a carbon density of 33.895 Mg.ha⁻¹. Soil carbon storage accounted for the majority of the total, accounting for 89.48%, with a carbon storage of 6.320 Pg and a carbon density of 30.330 Mg.ha⁻¹. Vegetation was the second largest contributor, with a carbon storage of approximately 0.741 Pg and a carbon density of 3.554 Mg.ha⁻¹. Biological soil crusts had the smallest contribution, with a carbon storage of 0.002 Pg and carbon density of 0.011 Mg.ha⁻¹. Although desert ecosystems have sparse vegetation and a relatively low carbon density, their vast areas make their total carbon storage significant. Therefore, an accurate estimation of carbon storage is crucial for achieving China's carbon peak and carbon neutrality goals. Despite some uncertainties in the estimation results, they can still provide valuable insights into the carbon storage of desert ecosystems to a certain degree.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data supporting the findings of this study are available upon request from the corresponding author.

Additional information

Funding

This research was supported by the Key Special Projection ‘Science and Technology Promoting the Development of Inner Mongolia Autonomous Region’ [grant number KJXM-EEDS-2020006], and the State Key Research and Development Program of China [grant number 2016YFC0500800], and the National Natural Science Foundation of China (grant number 42371074).