The study area is located at the critical area for mitigating desertification in the Ulan Buh Desert and Kubuqi Deserts. The Ulan Buh Desert is located in the middle reaches of the Yellow River and belongs to a transitional zone from warm-temperate semi-arid to arid climate (Li et al., 2018). The annual average precipitation is about 140 mm, and the annual average evaporation ranges from 2110 to 2995 mm (Li et al., 2024). The predominant soil type is aeolian sand (Tian et al., 2019), and the vegetation is dominated by xerophytic herbaceous plants and shrubs. The Kubuqi Desert is located at the northern edge of the Ordos Plateau, falling within the temperate arid and semi-arid regions. The annual precipitation is about 249 mm, and the annual average evaporation ranges from 2100 to 2700 mm (Yang et al., 2016; Chen et al., 2022). The predominant soil types are aeolian sand and gravelly sand (Dong et al., 2020; Sun et al., 2023), and the vegetation consists of steppe-like desert vegetation dominated by perennial grasses and small shrubs ( Figure 1 ).

We collected data from 24 survey plots in the Ulan Buh Desert and from 30 survey plots in the Kubuqi Desert from July and August in 2022 and 2023 ( Figure 1 ). In the Ulan Buh Desert, the main survey line was established along the east-west sand-crossing highway from Jilantai to Wuhai, where ten survey points were set up. An additional line extended from Jilantai through Dengkou to Wuhai, with fourteen survey points established along this route. In the Kubuqi Desert, the main survey line was laid out along the east-west direction of the Expressway, with branch lines set up in the north-south direction. The north-south branch line was divided into three regions: east, middle, and west, with 11, 7, and 12 survey points, respectively. At each survey point, we established a transect line of 100 to 200 meters. In each survey site, a nested design was implemented along the transect line, consisting of three 10 m × 10 m quadrats for moving sand dunes, six for revegetation, and three for natural woody vegetation. Within each of these larger woody vegetation quadrats, a smaller 0.5 m × 0.5 m quadrat was nested for the purpose of studying herbaceous vegetation. When survey points were distant from moving sand dunes or lacked revegetation, only natural vegetation plots were established. In the Ulan Buh Desert, a total of 99 quadrats were established (N=33 for revegetation, 57 for natural vegetation, and nine in moving sand dunes), and in the the Kubuqi Desert, a total of 185 quadrats were established (98 for revegetation, 60 for natural vegetation, and 27 in moving sand dunes) ( Figure 1 ; Supplementary Figure S1 ).

In each woody vegetation quadrat, we recorded the species, density and cover of woody plants, as well as the planting methods, row spacing (to evaluate the initial density of revegetation), and types of revegetation. In each herbaceous quadrat, we counted the number of species and estimated the vegetation cover. The above-ground biomass was collected using envelops in each herbaceous quadrat, and then was dried at 80°C in the laboratory for 48 hours, finally was weighed. The above-ground biomass was collected using envelops in each herbaceous quadrat, and then was dried at 80°C in the laboratory for 48 hours, finally was weighed. For both woody and herbaceous quadrats, we measured species richness (R) and calculated the Shannon-Wiener index (H’) and Simpson index (D) (Whittaker, 1972; Magurran and McGill, 2010; Tuomisto, 2010).

We collected latitude, longitude and altitude data of each sample plot using a GPS device. Meteorological data for the local area was obtained from the spatial interpolation dataset of China’s average meteorological elements using ArcGIS Desktop 10.8 software, based on the latitude and longitude of the sample plots (Spatial Interpolation Dataset of Average Conditions of Chinese Meteorological Elements: https://www.resdc.cn/DOI/DOI.aspx?DOIID=39). The acquired meteorological data mainly includes annual evaporation, annual average ground temperature, annual precipitation, annual average atmospheric pressure, annual average relative humidity, annual sunshine hours, annual average temperature, and annual average wind speed. We collected soil samples at nine depths (0, 0.1, 0.2, 0.4, 0.6, 0.8, 1.0, 1.2, 1.5 m) in the center of each woody quadrat. We placed the fresh soil into sealed bags and weighed them in situ. The soil samples were dried at 105°C for 24 hours and were weighed to get their dry weight in the laboratory. The gravimetric soil water content at different soil depths could be calculated. We collected surface (0-5 cm) soil samples using a 1×10^-4 m³ core in each quadrat and measured soil bulk density in the laboratory. To measure soil saturated water content, we soaked the soil samples in water for 24 hours and then weighed them before and after drying at 105°C for 24 hours. We collected soil samples from 0-20 cm using a 1×10^-4 m³ cutting ring and then divided them into two parts after air-drying. One part was passed through a 2 mm mesh sieve for measuring soil mechanical composition using the wet sieve and pipette method. The other part was passed through a 1 mm sieve to measure soil organic carbon, total carbon, total nitrogen, total phosphorus content, pH and conductivity. We tested the pH of a solution with a soil to water ratio of 1: 2.5 (g/ml) using a pH meter, and measured the electric conductivity of a solution with a soil to water ratio of 1:5 (g/ml) using a conductivity meter. Total carbon, total nitrogen, and total phosphorus contents were measured using an external heating-potassium dichromate method, a semi-micro Kjeldahl method, a sodium hydroxide melting-molybdenum antimony anti-colorimetric method, respectively (Pal, 2013; Haluschak, 2006).

In the Ulan Buh Desert, we found that revegetation led to an increase in the abundance, diversity and richness of herbaceous species, but a decrease in herbaceous biomass and woody abundance compared to natural vegetation. Similarly, compared to moving sand dunes, revegetation resulted in higher herbaceous abundance, richness, coverage and biomass; in addition, the soil water content in both shallow (0-0.3 m) and deep (0.4-1.5 m) layers decreased, while soil organic carbon, total nitrogen, total carbon, silt, clay, and electrical conductivity increased in both natural and revegetation communities ( Figure 2 ).

In the Kubuqi Desert, revegetation led to lower herbaceous abundance, richness, diversity cover, and biomass, but higher woody cover compared to the natural vegetation. There were higher herbaceous abundance, richness, and diversity, as well as higher woody abundance and cover in revegetated areas compared to moving sand dunes. In revegetated areas, the soil organic carbon, total nitrogen, total carbon, silt, clay contents, and electrical conductivity were lower than those in natural vegetation areas, while the organic carbon, total nitrogen, total carbon, clay content, and electrical conductivity were higher than those in moving sand dunes ( Figure 2 ).

We found that the natural Caragana korshinskii community, the Artemisia ordosica community and the Nitraria tangutorum-Haloxylon ammodendron community exhibited stronger vegetation-soil system coupling among the natural vegetation communities in the Ulan Buh Desert, and the coupling degree values (C) were 0.496, 0.492 and 0.486, respectively. Likewise, the natural Kalidium foliatum community, N. tangutorum community and N. tangutorum-H. ammodendron community exhibited better vegetation-soil system coupling and coordination. Their respective coupling coordination degree values (C_d) were 0.594, 0.515, and 0.471. In the revegetation community, the vegetation-soil system coupling was highest in the H. ammodendron, followed by the Hedysarum scoparium community, and the worst in the C. korshinskii community. Their respective C values were 0.495, 0.469 and 0.440. The vegetation-soil system coupling coordination state was highest in the C. korshinskii and H. ammodendron communities, but lowest in the H. scoparium, their C_d values were 0.494, 0.412, and 0.324, respectively. Compared to the natural C. korshinskii community, the coupling relationship between vegetation and soil of the revegetation C. korshinskii community was improved, and the vegetation-soil system also showed a good coupling state ( Table 2 ).

The coupling between vegetation communities and soil was higher in natural vegetation communities of the Kubuqi Desert. This included higher C values of natural Achnatherum splendens, Caragana tibetana, and A. ordosica communities were 0.500, 0.499 and 0.498, respectively. Likewise, the natural N. tangutorum-Tamarix chinensis, C. tibetana and A. ordosica communities had higher coupling coordination states of the vegetation-soil system, with high C_d values of 0.753, 0.694 and 0.682 respectively ( Table 2 ). In revegetated areas, the vegetation-soil system coupling in the A. ordosica, C. korshinskii, and Populus przewalskii-Salix psammophila revegetation communities was relatively higher, and the C values of them were 0.500, 0.498, and 0.490, respectively. The revegetated communities of A. ordosica, C. tibetana and P. przewalskii had the relatively higher vegetation-soil system coupling coordination states, with C_d values of 0.763, 0.612, and 0.484, respectively. The vegetation-soil system coupling and its coordination states of revegetated A. ordosica community were higher than those of natural A. ordosica community ( Table 2 ).

We found that the first axis (RDA1) and the second axis (RDA2) explained 56.2% and 17.5% of the influence of environmental factors on the vegetation-soil system coupling, respectively. Specifically, we found that vegetation-soil system had strong correlations with total nitrogen, total carbon, the cover, abundance, richness, and biomass of herbaceous species, as well as woody cover and richness, while it was negatively correlated with soil pH, sand content, wind speed and temperature. Furthermore, water status, vegetation growth and soil nutrient contents also influence the vegetation-soil system coupling ( Figure 3 ).

We compared vegetation and soil indicators, and vegetation-soil system coupling to select the proper revegetation species and configurations. In the Ulan Buh Desert, compared to the natural C. korshinskii community, the woody abundance, richness, and coverage of revegetated C. korshinskii community increased by 2.90%, 18.4%, and 15.8%, respectively; the soil organic carbon, total nitrogen, total carbon, silt, and clay contents increased by 39.5%, 31.3%, 30.7%, 70.8%, and 61.3%, respectively; the vegetation-soil system coupling also increased; while the sand contents decreased by 20.7%. Therefore, in the Ulan Buh Desert, we recommend two replanting species and their densities: C. korshinskii with 18-44 individuals per 100 m^-2, and H. ammodendron with 24-30 individuals per 100 m^-2 ( Table 2 ; Supplementary Table S1 ).

In the Kubuqi Desert, compared to the natural A. ordosica community, the herbaceous abundance, woody abundance, herbaceous coverage, and woody coverage of the revegetated A. ordosica community increased by 3.63%, 72.2%, 18.6%, 58.5%, respectively; the soil organic carbon, total nitrogen, total carbon, total phosphorus, silt, and clay contents increased by 57.7%, 27.6%, 10.7%, 21.7%, 27.7%, and 22.5%, respectively; the vegetation-soil system coupling also increased; while the sand content decreased by 4.83%. As a result, in the Kubuqi Desert, we recommend two replanting species and their densities: A. ordosica with 60-70 individuals per 100 m^-2 and C. korshinskii with 20-45 individuals per 100 m^-2 ( Table 2 ; Supplementary Table S1 ).

Consistent with our first hypothesis, plant and soil conditions improved compared to those of the moving sand dunes and became similar to those in natural vegetation areas following revegetation. This study finds that herbaceous species diversity significantly increased comparing to adjacent natural vegetation communities after the establishment of revegetation in the Ulan Buh Desert, whereas herbaceous biomass, woody richness, and soil water content significantly decreased. This may be due to shrub planting effectively stabilizing the surface of sandy soil and improving soil nutrient conditions, thus providing a suitable environment for the settlement and germination of herbaceous plant seeds (Angers and Caron,1998; Miao et al., 2015; Zhong et al., 2018). Meanwhile, the root turnover of woody plants also provides essential nutrients for the growth of herbaceous plants. Therefore, the abundance, richness, and diversity indices of herbaceous plants have increased (An, 2019). However, the growth of woody species also competed for some soil nutrients and water resources, leading to the reduction in the herbaceous biomass and woody richness (Rossatto et al., 2014). In addition, the root water uptake of woody vegetation resulted in the decline soil water content (Pellissier et al., 2008). The establishment of revegetation in the Kubuqi Desert, herbaceous indicators, soil nutrients, silt, and clay contents all declined when compared to natural vegetation communities, whereas woody abundance increased. These might be illustrated that the introduction of woody vegetation increased the abundance of woody species. Meanwhile, the competition for nutrients and water among these woody plants was relatively intensive, attenuating the resource availability and thus inhibiting the growth of original herbaceous plants (Cui et al., 2019; Zheng, 2022).

In the present study, we found that the establishment of revegetation significantly increased the diversity of herbaceous plants, biomass, and soil nutrients, which is attributed to the woody vegetation creating more suitable growth conditions for herbaceous plants and increasing the content of silt and clay in the soil, thereby improving soil structure and stability (Ou et al., 2020; Scotton and Andreatta, 2021; Kumar et al., 2022). Vegetation restoration enhanced soil organic matter content through the input of roots and residues, which in turn increased soil fertility, beneficial for the growth of woody plants, and reduced wind erosion, providing a more stable growth environment for plants and promoting their reproduction and spread (Lajtha et al., 2014; Mitchell et al., 2018; Zhang et al., 2020; Zhu et al., 2010; Mensah, 2015; Maiti et al., 2021; Luo et al., 2020). Meanwhile, vegetation restoration improved the physical, chemical, and biological characteristics of the soil, increased soil nutrient content and water retention capacity, and enhanced species diversity (Xu et al., 2010; Chen et al., 2023; Zhou et al., 2023; Sheoran et al., 2010; Ploughe et al., 2021). In the early stages of vegetation restoration, effective soil coverage reduced water evaporation and soil erosion, helping to maintain soil nutrients (Li et al., 2004; Zhang et al., 2018). Overall, the vegetation subsystem, soil subsystem, and the vegetation-soil system coupling all increased after restoration, and these improvements collectively promoted the favorable status and interaction of the vegetation and soil systems (Li and Liber, 2018; Guo et al., 2021).

Our findings provided scientific evidence for windbreak and sand-stabilization in these two deserts, and gave important guidance to future vegetation restoration in deserts. Traditionally, the success of restoration was determined by monitoring plant cover, individual counts, and soil characteristics after revegetation (Cudjoe, 2011; Song, 2018) as well as by selecting suitable species by greenhouse, field experiments, and seedling survival tests (Jusaitis and Pillman, 1997; Gastauer et al., 2020). Here, we considered not only the vegetation and soil systems, but also the vegetation-soil coupling in different vegetation types to select the most suitable revegetation strategy.

Consistent with our second hypothesis, we found that different community types exhibited variations in the vegetation-soil system coupling in these two deserts, which indicted that species selection was extremely important for the sustainability ecological restoration in deserts. The consistence of the changes between vegetation and soil indicators elucidated that plant growth and soil properties were closely correlated. There is a feedback mechanism between vegetation and soil. The plants exert a huge influence on soil conditions, which in turn support plant growth (Bever et al., 1997; Ehrenfeld et al., 2005; Miki, 2012; Maiti and Ghosh, 2020). Vegetation alters soil properties through physical and chemical processes during the restoration process (Cao et al., 2008; Ghestem et al., 2011; Wu et al., 2019). Meanwhile, vegetation roots increase soil porosity, and the decomposition of plant material adds organic matter to soil (Wu et al., 2016). In a word, different community types have varying impacts on soil properties, leading to disparities in the vegetation-soil systems coupling of different community types. The improvements of soil properties and vegetation growth enhance soil aggregate stability and resistance to wind erosion (Dou et al., 2020), thereby creating a positive feedback loop within the vegetation-soil system.

In this study, we found that soil nutrient content, herbaceous and woody indicators were positively correlated with the coupling degree of the vegetation-soil system. These were consistent with our hypothesis that the vegetation-soil system coupling was mainly affected by vegetation and soil characteristics. Generally, superior soil conditions promote plant growth, which in turn improves soil structure and nutrient cycling, thus forming a positive feedback loop. The mutual promotion and dependence between vegetation and soil constituted a well-coupled ecosystem, demonstrating a positive coupling state between vegetation-soil systems (Van der Putten et al., 2013; Hobbie, 2015; Lehman et al., 2015; Wang et al., 2020; Aqeel et al., 2023). Simultaneously, this feedback loop suggests that healthy soil and abundant vegetation maintain ecosystem stability and functionality (Powlson et al., 2011; Mensah, 2015). Our findings indicated that unsuitable soil pH, high sand content, high wind speed, and high temperature adversely influenced plant growth and soil conditions (Zuo et al., 2012). These factors can also lead to soil degradation, nutrient loss, and reduce water availability, which in turn reduce plant growth and survival. The plant degeneration subsequently reduced vegetation cover and richness, exacerbating soil degradation (Duniway et al., 2019; Huang and Hartemink, 2020; Zhang et al., 2021; Ferreira et al., 2022). Taken together, coupling coordination of vegetation-soil system is crucial for maintaining the stability of ecosystems after the establishment of revegetation, as it directly affects the normal functioning of ecological processes and the long-term health of the ecosystem.