The impact of plant restoration on soil aggregates in karst mountainous areas of southwestern China Authors: Qian Meng, Mingyi Zhang, RongYin Jin, Yunjie Wu, Runze Wang, Jianli Zhang, Zhaohui Sun, Guo Yuan, Yingchun Pu Published: 2026 ABSTRACT SummarySoil aggregates are the basic structure of soil. To investigate the distribution characteristics of karst soil aggregates in southwest China and the response of their stability to long-term vegetation restoration, we took karst areas and their restored successional transects in Caohai Nature Reserve of Weining, Guizhou Province, as the research subject. The result showed that: (1) soil water stable macroaggregates (R0.25), soil geometric mean diameter MWD, and mean weight diameter GMD were significantly reduced with vegetation restoration. (2) generalized linear mixed model (GLMM) and structural equation models revealed that vegetation type exhibited a highly significant effect on soil aggregates (p < 0.001), including both karst soil and vegetation-restored soil, during a homogeneous garden experiment. Vegetation type is the core factor affecting the formation and stability of soil aggregates, while topographic characteristics have no significant effect on soil aggregates during vegetation restoration.Subject areas: soil science, environmental science, environmental management, plant biology, plant ecology Introduction Karst regions are globally recognized as one of the world’s most ecologically fragile zones, with the most extensive distribution found in China. These areas are primarily concentrated in the southwestern provinces of Guangxi, Guizhou, and Yunnan, with a total area of approximately 540,000 square kilometers.1 Due to the predominantly carbonate-based soils and distinctive geomorphological characteristics of karst regions, these areas are particularly vulnerable to severe rocky desertification. Under the multiple stressors, this has precipitated a detrimental cycle of biodiversity loss, severe soil erosion, and significant soil quality degradation,2,3 ultimately affecting the economic development of Southwest China. A vital component of the Earth’s critical zone, soil constitutes a fundamental natural resource that underpins human survival and development.4,5 Soil aggregates, the fundamental units of soil structure, are crucial to both sustaining soil productivity and ensuring its structural stability.6 The quality, quantity, and composition of soil aggregates directly determine the physicochemical properties of soil structure, including soil fertility, erosion resistance, and carbon sequestration capacity.7 Based on particle size classification, soil aggregates are categorized as follows: those larger than 2 mm are classified as gravel fractions, those between 2 mm and 1 mm are classified as coarse sand fractions, and particles smaller than 0.25 mm are classified as the fine sand and very fine sand fractions.8,9 Vegetation restoration is a primary strategy for mitigating rocky desertification in karst regions,10 with demonstrated efficacy in reducing soil erosion and rehabilitating ecological balance. Long-term studies, conducted both in China and abroad, have systematically examined how diverse vegetation restoration approaches affect soil water retention capacity,11 content of active organic carbon components,12,13 microbial diversity, activity,14 and community structure.15 However, there is a clear consensus regarding the synergistic relationship between soil physicochemical property dynamics and vegetation restoration succession under identical karst environments. Notably, soil aggregates not only mediate various soil physicochemical characteristics but also serve as indicators of erosion susceptibility through their water stability and abundance. Therefore, investigating the correlation between soil quality and aggregate characteristics before and after vegetation restoration in karst regions would provide critical insights to address these knowledge gaps. This study selected Caohai National Nature Reserve, Weining, Guizhou Province (Southwest China) as the research area, conducting a common garden experiment with soil samples collected from pre- and post-vegetation restoration sites in karst terrain. Through particle size analysis and classification of soil aggregates, we employ path analysis to quantify direct and indirect effects of environmental factors on aggregate stability in both natural and restored karst landscapes. The study elucidates soil change patterns and their interacting drivers, aiming to advance understanding of how vegetation restoration influences soil fertility, erosion resistance, and carbon sequestration capacity in karst systems. We hypothesize that: (1) in karst soils, vegetation type, rather than topographic position, is the primary driver of aggregate stability, (2) restoration enhances stability by increasing soil organic carbon (SOC) and altering soil chemical properties (e.g., pH), and (3) native species promote the formation of more stable aggregates compared to commonly used restoration species. These findings provide scientific support for ecological rehabilitation and sustainable development in Southwest China’s karst regions under global climate change pressures. Results Regional overview and research methods The research area is located in Caohai National Nature Reserve (26°49 ′-26° 53′n, 1104°12' −104° 18′e) in Weining County, Guizhou Province, the central part of the Yunnan-Guizhou Plateau. As the largest natural freshwater lake in Guizhou province, Weining Caohai has the most unique and complete plateau wetland ecosystem in the plateau region. Its main water source comes from atmospheric precipitation, and its water depth is no more than 5 m, with an average water depth of 1.5 m. The overall topography of Weining Caohai gradually decreases from the southeast corner to the northwest corner, presenting a tableland of different heights. Its vulnerability, typicality, importance and biodiversity are representative of the few subtropical plateau karst wetland ecosystems in China, where aquatic plants are abundant.16 The selected site of the homogeneous garden is the slope belt of the scientific research base of the College of Ecological Environment Engineering of Guizhou University for Nationalities (106.6144 E−26.3772 N). Experimental methods Soil collection will start in May 2022, and the dominant species of karst and vegetation restoration will be alternately planted for a homogeneous garden pot experiment. In karst regions, the native vegetation is predominantly composed of Elymus nutans (Table 1). As a perennial bunchgrass, it possesses a well-developed and deep root system, often penetrating the soil to depths exceeding 1 m. Its dense fibrous roots not only enable deep soil water absorption but also effectively anchor the soil, thereby enhancing its resistance to erosion. In contrast, the dominant species in restoration areas is primarily Artemisia argyi (commonly known as mugwort) (Table 1), a highly adaptable but moisture-preferring perennial herb from the Asteraceae family. Its root system is characterized by a combination of horizontally spreading rhizomes and shallow fibrous roots, demonstrating remarkable capabilities for propagation and expansion. The experiment will be started in early May 2022, and soil collection and plant seedling will be carried out. After the plants sprout, they will be transplanted into potted plants. Two kinds of plant growth environments will be constructed: the soil environment after vegetation restoration in the karst area and the soil environment in the karst area. In each plant growing environment, the corresponding environmental plants were cultivated in each cultivation environment, and the cross planting experiments were carried out for comparative cultivation, namely: soil in karst area after vegetation restoration-soil original plants in karst area after vegetation restoration, soil in karst area-soil original plants in karst area; soil in karst area after vegetation restoration-soil original plants in karst area, soil in karst area-soil original plants in vegetation restoration area. Each plant was tested in 3 replicates in each cultivation environment to ensure the reliability of the results. Destructive soil sample collection began in November 2022 after the plants completed a complete growth cycle, sealed and labeled in self-sealing bags, and cryopreserved for return to the laboratory. According to different experimental requirements, the samples were mixed evenly to determine physical and chemical properties after removing impurities (Table 2). The remaining soil samples were subjected to a series of pretreatment processes, including air-drying, crushing, and sieving, for subsequent analysis. Soil aggregates were determined by the wet sieving method: three 100 g air-dried soil samples were taken and separated by wet sieving with a six-layer sieve. The aggregates of each particle grade were collected and dried to determine the average mass. Table 1Location and vegetation status of the transecTransectLatitude longitudeAltitudeVegetation coverageSoil typePhytocoenosiumKarst104°14′0.33″-104°14′2.81″E 26°51′51.11″-26°52′0.38″N2182–218845Lime soilHerbs: to lamegrass (Elymus dahuricus)、pennisetum(Pennisetum alopecuroides)and Rat-tail millet(Sporobolus fertilis) MainlyVegetation restoration104°13′53.40″-104°13′7.0″E 26°51′56.86″-26°52′0.81″N2182–218865Lime soilarborcoenosis: To the elm(Ulmus pumila L) Mainly; Herbs: Wormwood (Artemisia argyi H. Lév. & Vaniot) pennisetum (Pennisetum alopecuroides (L.) Spreng.) Mainly Location and vegetation status of the transec Table 2Physical and chemical properties of surface soil from different transectsSample plotpHMoisture content (%)Soil bulk density (g/cm3)EC (μs/cm)TN (g/kg)SOC (g/kg)Karst - Uphill7.83 ± 0.1532.51 ± 2.761.42 ± 0.0444.90 ± 17.250.72 ± 0.038.73 ± 0.27Karst - Downhill7.40 ± 0.4135.86 ± 7.811.31 ± 0.1358.77 ± 8.731.24 ± 0.1511.38 ± 0.54Karst - Shore7.42 ± 0.4532.01 ± 4.871.41 ± 0.1062.27 ± 6.871.79 ± 0.0118.58 ± 0.20Restoration - Uphill7.01 ± 0.7924.12 ± 0.461.35 ± 0.1142.60 ± 14.271.05 ± 0.034.92 ± 0.66Restoration - Downhill6.93 ± 0.3626.02 ± 2.401.32 ± 0.0549.70 ± 15.951.24 ± 0.048.86 ± 0.52Restoration - Shore7.57 ± 0.3328.29 ± 0.291.36 ± 0.0545.70 ± 10.521.61 ± 0.0122.84 ± 0.64 Physical and chemical properties of surface soil from different transects Pretreatment and determination of samples The particle size composition of aggregates was determined by Elliott’s wet sieve method.5 A water-stable aggregate of >5 mm, 5-3 mm, 3-2 mm, 2-1 mm, 1-0.5 mm, 0.5–0.25 mm, <0.25 mm was obtained. With reference to the United States Department of Agriculture Soil Particle Size Classification Standard (United States Department of Agriculture, 1951), soil aggregate components with a particle size greater than 2 mm are defined as gravel. Those with particle sizes between 2 mm and 0.25 mm fall into the categories of very coarse sand, coarse sand, and medium sand (collectively referred to as coarse gravel in this study). Components with a particle size smaller than 0.25 mm are defined as fine sand and very fine sand (collectively referred to as fine sand in this study). Aggregate stability index: >0.25 mm water stable aggregate content (R0.25), mean weight diameter (MWD), geometric mean diameter (GMD), and fractal dimension (D) can be calculated by the following formula. In the formula, wi is the ratio of the mass of the i-th aggregate component to the total soil mass, x¯i is the average diameter of the particle size aggregate, M(r0.25 mm follows the descending order: riparian zone > lower slope position > upper slope position. In general, across all slope positions, the proportion of aggregates >5 mm is highest in the riparian zone, exceeding that in the upper slope position by approximately 20%. Meanwhile, aggregates >2 mm account for about 86%–88% of the total aggregate count and show significant differences compared to the contents of other aggregate fractions (Figure 1). Additionally, the distribution of soil aggregates varies between different regions (karst area vs. vegetation restoration area). In karst areas, the content of aggregates >0.25 mm is consistently higher than that in restored areas, whereas the content of aggregates <0.25 mm shows the opposite pattern. This indicates that karst soils contain a higher proportion of large aggregates, reflecting stronger soil aggregation capacity. Across different slope positions, the contents of various soil aggregate fractions also differ. In the upper slope soil layer, the content of aggregates >0.25 mm is generally lower than that in the lower slope soil layer, while the content of aggregates <0.25 mm exhibits the opposite trend. Distribution characteristics of soil aggregates before and after vegetation restoration Data are presented as mean ± standard deviation (SD), n = 3. As shown in Figure 1, the changes of soil aggregates at uphill and downhill sites were not significant; the above conclusions were further confirmed by electron microscope scanning at uphill and downhill sites (Table 3). Table 3Dominant plant species at different slope positions under karst native vegetation and vegetation restorationSlope positionKarst native vegetationVegetation restorationUphillPennisetum alopecuroides, Sporobolus fertilisArtemisia argyi, Elymus dahuricusDownhillPennisetum alopecuroides, Elymus dahuricusPennisetum alopecuroides, Artemisia argyi Dominant plant species at different slope positions under karst native vegetation and vegetation restoration Changes in GMD and MWD of soil aggregates after vegetation restoration The stability of soil aggregates refers to the ability of these aggregates to maintain their original form against external forces or changes in the external environment. The common indices of soil aggregate size distribution include MWD and GMD. In general, the larger the MWD and GMD values, the higher the degree of agglomeration and the better the stability of soil aggregates. These two indices are of great significance in soil structure assessment, and can reflect whether the size and structure of soil aggregates are healthy and stable. Therefore, by monitoring and analyzing the numerical changes of MWD and GMD, the physical properties of soil can be better understood, and a scientific basis for soil quality evaluation and management can be provided. According to the analysis, GMD values in karst areas are significantly greater than those in vegetation restoration areas under any slope, and GMD values in banks and wetlands are significantly greater than those in up-slope and down-slope locations. The GMD value of the subsoil of the downhill slope was significantly lower than that of the other two slopes. According to the average values of soil in two different regions (karst region and vegetation restoration region), the average GMD values of soil aggregates in the karst region were higher than those in the vegetation restoration region (Figure 2Aa). Soil MWD and GWD at each slope position before and after vegetation restoration (A) MWD of soil aggregates before and after vegetation restoration. (B) GMD of soil aggregates before and after vegetation restoration. Different capital letters indicate significant differences in particle size of soil aggregates before and after vegetation restoration (p < 0.05). Different lowercase letters indicate significant differences in particle size components of soil aggregates at different slope positions under the same land use type (p < 0.05). Data are presented as mean ± standard deviation (SD), n = 3. In addition, the MWD values of the karst areas with slopes at both uphill and downhill positions were significantly higher than those of the vegetation restoration areas, while the MWD values of the karst areas with slopes at banks and wetlands were significantly higher than those of the vegetation restoration areas (Figure 2B). Change the characteristics of the soil fractal dimension after vegetation restoration As evidenced in Figures 3A and 3B, both the fractal dimension (D) and water-stable macroaggregate content (>0.25 mm) consistently demonstrated significantly lower values (p < 0.05) in upper and lower slope positions compared to riparian zones, regardless of vegetation restoration status. The D and the content of water stable aggregates (R2.5) in the vegetation restoration area were significantly lower than those in the karst area (p < 0.05). However, there was no significant change in the D and the content of water stable aggregates (R2.5) in the vegetation restoration area and the karst area on the bank. Water-stable macroaggregates and fractal dimension across slope positions before and after vegetation restoration (A) Distribution characteristics of R0.25 under different land use patterns. (B) Fractal dimension D under different land use patterns. Different capital letters indicate significant differences in particle size of soil aggregates before and after vegetation restoration (p < 0.05). Different lowercase letters indicate significant differences in particle size components of soil aggregates at different slope positions under the same land use type (p < 0.05). Data are presented as mean ± standard deviation (SD), n = 3. Correlation analysis between soil aggregates and physical and chemical properties Correlation analysis between soil aggregates and physical and chemical properties The fitting analysis (Figure 4) revealed that in the karst region, electrical conductivity (EC) and TN were significantly positively correlated with coarse macroaggregates, whereas EC and TN exhibited significant negative correlations with fine macroaggregates (Figure 4A). Correlation analysis of different aggregate particle sizes and soil physicochemical properties (A) Karst native vegetation. (B) Vegetation restoration. GMD: geometric mean diameter; MWD: mean weight diameter; SOC: soil organic carbon content. The numerical value indicates the correlation coefficient, where positive values represent positive correlation and negative values represent negative correlation. ∗ indicates significance (p < 0.05), and ∗∗∗ indicates high significance (p < 0.001). Data are presented as mean ± standard deviation (SD), n = 3. As shown by the fitted data in Figure 4, in the karst region, EC and gravel content, as well as TN and gravel content, exhibit significant positive correlations with each other, whereas EC and coarse sand content, as well as TN and coarse sand content, show significant negative correlations with each other (Figure 4A). Following vegetation restoration, the significance between EC and gravel content, as well as between TN and gravel content, increased, both showing highly significant positive correlations. In contrast, EC and coarse sand content, as well as TN and coarse sand content, both exhibited highly significant negative correlations. Additionally, the relationships between pH, CEC, SOC, and gravel content shifted from non-significant to significant. Among these, the changes in CEC and gravel content, as well as SOC and gravel content, were the most pronounced, with both pairs showing high significance (Figure 4B). Main controlling factor of soil aggregates -- structural equation model To further elucidate the direct and indirect effects of the aforementioned soil factors on soil aggregates and quantify their relative importance, significant changes in SOC, TN, pH, CEC, and EC were observed during the vegetation restoration process (Table 2). Therefore, SOC, TN, pH, CEC, and EC were selected as latent variables to investigate the mechanisms by which vegetation restoration influences soil aggregates. Three SEMs were constructed based on the correlation results between these factors and soil aggregates. The SEMs demonstrated good fit indices (χ2/df = 1.85, CFI = 0.96, RMSEA = 0.04). The SEM results indicate that vegetation restoration, soil SOC, TN, pH, CEC, and EC collectively exerted positive total effects on soil gravel content, while showing negative total effects on coarse sand and fine sand contents. In contrast, slope position exhibited a negative total effect on gravel content but positive total effects on coarse sand and fine sand contents. Among all factors, TN had the strongest total effect on gravel content (0.748), followed by SOC (0.339) (Figure 5). Structural equation model (SEM) of soil aggregate dynamics (A) Coarse macroaggregate formation. (B) Fine macroaggregate formation drivers. (C) Microaggregate dynamics. Solid blue lines indicate significant positive effects; dashed blue lines indicate non-significant positive effects; solid red lines indicate significant negative effects; dashed red lines indicate non-significant negative effects. Significance is defined as p < 0.05. Data are presented as mean ± standard deviation (SD), n = 3. Discussion Effects of vegetation restoration on soil aggregates in karst regions Water-stable macroaggregates (>0.25 mm) serve as fundamental structural units in soil systems, whose mass proportion and spatial distribution profoundly influence key physical properties.17 Generally, a higher mass percentage of macroaggregates (R0.25 mm) indicates greater soil structural stability. However, this study found that the MWD and GMD of soil decreased significantly after vegetation restoration in the Karst rocky desertification area (Figures 2A and 2B). This reduction reflects a more developed and porous microstructure in soil aggregates during the initial restoration stage (0–5 years) (Figure 3A), a feature that is more pronounced in areas with higher vegetation coverage. This phenomenon is in sharp contrast to the linear cognition of “vegetation restoration promotes soil structure improvement” in classical ecological restoration theory, suggesting that there may be a unique restoration trajectory in the Karst fragile ecosystem at the initial stage of vegetation restoration. Further analysis suggested that, despite increasing land cover, tree introduction may have transiently suppressed soil structural development via multiple interacting mechanisms. On one hand, the rhizosphere provides a specific microhabitat defined by complex interactions that influence plant growth and resource competition.18 Drying cycles can stimulate aggregate formation through matrix forces, water surface tension and heterogeneous swelling, the physical penetration process of roots in shallow karst soil layer directly destroyed the structural integrity of the original fragile aggregates.19 On the other hand, the litter input with high lignin and high C/N ratio stimulated the “co-metabolism” pathway dominated by r-strategy microorganisms, accelerating the decomposition of original mineral-bound organic matter (MAOM),20 and loss of SOC (Table 2),21 and resulting in a decrease in cementation and aggregate stability. At the same time, organic acids in root exudates help to increase soil cation exchange capacity (CEC) (Table 2) and promote calcium-mediated weathering-cementation processes, but their short-term depolymerization of calcium-organic matter complexes further weakens the stability of large aggregates.22 SEM models show that vegetation restoration has a significant positive effect on pH, which in turn has a significant positive effect on soil aggregate microstructure (Figure 5A), along with a significant increase in EC of about 25% (Table 2). This alkalization trend is closely related to the complex balance of carbonate dissolution and organic acid metabolism in a high calcium background. The increase in pH enhances the negative charge on the surface of soil colloids,23 while the synchronous increase of Ca2+ concentration promotes colloid flocculation by compressing the electric double layer.24 However, the aggregates formed by the inorganic flocculation process lack sufficient biological cementing materials at the initial stage, which indicates the decoupling of physical stability and biological stability.25 Microbial community analysis further revealed that concurrent synergistic changes in pH-EC-Ca2+ selected specific microbial functional groups to promote the proliferation of calcium-tolerant actinomycetes and fungi. The extracellular polymers produced by these microorganisms differed significantly in composition and function from the hyphal network in traditional vegetation restoration systems. The seemingly contradictory responses, i.e., CEC enhancement and aggregate stability decline, reflect the staged reorganization and adaptation strategies of vegetation-soil-microbial systems in the early stages of restoration in extremely degraded karst ecosystems. In habitats with high rock exposure and discontinuous soil cover, organic matter input and root distribution present highly heterogeneous “patchy” patterns,5 resulting in strong excitation effects in local microdomains. This complex response mechanism of micro-deconstruction and macro-improvement highlights the spatiotemporal heterogeneity and nonlinear characteristics of the restoration path of the karst ecosystem. Furthermore, the D serves as a crucial indicator for evaluating the influence of water-stable aggregate content on soil structure and stability trends. Its magnitude reflects the soil’s resistance capacity against water erosion and erosive forces.26 Following vegetation restoration, the D values of soil at both upper/lower slope positions and riparian zones showed significant decreases (p < 0.05), demonstrating that vegetation restoration positively enhanced the structural stability of soils at lower slope positions and riparian areas. These findings provide further mechanistic support for the aforementioned conclusions. Influence of vegetation succession-soil type interaction on soil aggregates Although the mechanisms by which vegetation restoration affects soil aggregates have been elucidated, the specific relationships among newly established vegetation following restoration in karst landscapes, native karst vegetation, and various soil types remain poorly understood. To further investigate this complex mechanism, common garden experiments would represent an effective approach. These controlled experiments would enable systematic observation and analysis of different vegetation types growing under identical edaphic conditions, thereby providing more precise insights into the interactive relationships between vegetation succession and soil aggregate formation. This study employed a common garden experiment to investigate the influence of vegetation restoration on soil aggregates in karst ecosystems. Taking the dominant vegetation types in different regions and different slope positions as fixed factors, and karst soils and vegetation-restored soils as random factors, the interaction between vegetation and soil on soil aggregates was investigated. The final GLMM showed a marginal R2 of 0.65 and a conditional R2 of 0.78. The results demonstrate that native vegetation consistently promotes superior soil aggregation in its original habitat compared to introduced restoration species, revealing a specialized plant-soil coadaptation mechanism (Figure 6). Specifically, karst soils cultivated with native vegetation exhibited significantly higher water-stable macroaggregate content than those with restoration species (Figure 6A), while restored soils similarly showed better aggregation when grown with their adapted vegetation (Figure 6B). This pattern likely stems from evolved soil legacy effects, where native plants shape distinct microbial communities that enhance aggregate formation and create favorable growth conditions for subsequent vegetation. These findings highlight the importance of considering evolutionary plant-soil-microbe interactions in restoration ecology, as the long-term coadaptation between vegetation and edaphic properties fundamentally governs soil structure development in fragile karst systems.27 Specifically, the native vegetation in karst regions has adapted to thrive under nutrient-poor soil conditions with limited water availability. Consequently, when vegetation restored from other environments is transplanted into these karst soils, the edaphic conditions fail to meet the plants’ growth requirements. Under such survival stress, the transplanted vegetation allocates most resources to basic metabolic maintenance rather than secreting sufficient exudates to promote the formation of water-stable macroaggregates. This specialized plant-soil interdependence is characteristic not only of the original karst ecosystems but also of newly restored vegetation systems, where successful establishment depends on developing similar co-adaptation mechanisms over time.28 Distribution characteristics of soil aggregates in common garden experiments (A) Karst native vegetation. (B) Vegetation restoration. F denotes degrees of freedom; P denotes significance level, where p < 0.05 indicates significance and p < 0.01 indicates high significance. Data are presented as mean ± standard deviation (SD), n = 3. Furthermore, these findings underscore the intricate and nuanced interactions between plants and their soil environments - highlighting the critical importance of plant-soil feedback (PSFs). As a key ecological mechanism regulating plant populations and community dynamics, PSFs operate through species-specific plant modifications of their soil environment, which in turn differentially affect the performance of subsequent or neighboring vegetation.29,30 Comparative analysis revealed significantly lower soil moisture content in restored vegetation areas relative to native karst soils (Table 2). Under these xeric conditions, plant-associated symbionts (e.g., mycorrhizal fungi) may enhance host drought tolerance and nutrient acquisition capacity, thereby potentially mitigating nutrient-mediated negative PSFs.31,32 In humid environments, negative PSFs may be amplified by increased abundance and mobility of natural enemies (e.g., pathogenic microbes and herbivorous nematodes). Consequently, the vegetation-driven soil modification mechanisms appear less pronounced in native karst ecosystems compared to restored systems, where more robust positive soil-vegetation feedback mechanisms dominate. However, for species with enhanced resistance to natural enemies (e.g., ectomycorrhizal [EcM] vegetation, which represents the native karst flora), PSFs may remain largely independent of water availability33; consequently, the co-evolved interactions between karst ecosystems and their native vegetation can effectively mitigate partial negative PSF effects. When discussing the significance of vegetation and slope-position soil on soil aggregates, this study adopted a GLMM for analysis (Table 4). Soil and slope position were used as fixed factors, while karst and native vegetation after vegetation restoration were used as random factors to explore the interaction relationship between vegetation-soil and soil aggregates. The results showed that in karst areas, vegetation, soil, and their interactions all had extremely significant effects on soil aggregates. The soil in karst areas is nutrient-poor2 and has a high proportion of micro-aggregates (Figure 1), which, to some extent, limits the structural stability and fertility.34 Moreover, the specific root length (SRL) value plays a positive role in soil aggregate stability. Due to its high specific root length, the root system forms a dense network that physically entangles microaggregates. In contrast, the taproots of native vegetation may contribute more to the formation of deep soil fissures and the creation of microbial habitats. This suggests that the selection of restoration species should take root system architecture characteristics into account to target specific aggregation mechanisms. Plants provide the necessary substrate that fuels microbial growth and metabolism, thereby strongly promoting the development of the soil microbial community. The activities of these microorganisms not only help regulate nutrient cycling but also significantly improve soil fertility and nutrient availability.35 Moreover, ecosystems in karst areas are highly vulnerable. Karst landscapes are characterized by extensive bedrock outcrops and, where vegetation is lacking, by hot, rainy summers; consequently, soil erosion is common,36 with upslope positions being the most vulnerable. This phenomenon of soil erosion not only leads to a large loss of topsoil and exposed bedrock but also seriously reduces biodiversity and causes a series of ecological problems. More importantly, the washed soil aggregates accumulate in the downhill position, which further aggravates the significant difference in the content of water-stable large aggregates between upper and lower slope positions.37 In addition, there is a close coupling relationship between vegetation and soil, and vegetation enhances soil scour resistance through its roots and other structures, thus improving the stability of soil structure.38 At the same time, the improvement of soil quality provides better conditions for the growth of vegetation, forming a virtuous cycle, and this relationship is particularly significant in karst areas. Table 4Location and vegetation status of the transeckarstvegetation restorationFPFPVegetation44345.291>0.0010.0420.001Slope position286832.245>0.0016.0850.839Vegetation ∗ vegetation3.443>0.0010.2820.067 Location and vegetation status of the transec For the vegetation restoration area, this study found that vegetation had a significant effect on soil aggregates (Table 4). This change may be attributed to the significant increase in water-stable macroaggregates in karst regions after vegetation restoration (Figure 3A). Soil macroaggregates (gravel) exhibit higher oxygen content than microaggregates (fine sand particles). The increased oxygen availability not only provides more sufficient oxygen and moisture for plant roots, promoting vegetation survival and growth, but also reduces competitive pressure on plants in natural ecosystems.39 This optimization of growth conditions further stimulates vegetation to produce more root exudates and litter, which in turn enhances the formation and stabilization of water-stable soil macroaggregates.40 Notably, the native vegetation across all slope positions in the karst region was exclusively dominated by herbaceous species with filamentous root systems. In contrast, the vegetation that naturally established following restoration efforts primarily consisted of fibrous-rooted herbaceous plants (Table 1). Comparative analysis demonstrated that the SRL of filamentous-rooted species was significantly greater than that of fibrous-rooted species. Moreover, a significant negative correlation was identified between SRL values and PSF parameters.41 These findings collectively indicate that these two distinct root systems morphotypes exert significantly divergent influences on soil properties within identical edaphic conditions.42 However, slope position and its interaction with vegetation showed no significant effects on soil aggregate formation (Table 3). This finding suggests that while slope position represents an important topographic variable, it did not significantly influence the formation and stability of soil aggregates in this study. Following vegetation restoration, both upper and lower slope positions exhibited significant increases in water-stable macroaggregates. This improvement not only enhanced soil fertility but also substantially improved soil structural stability. Concurrently, soil water retention showed marked improvement, effectively mitigating soil erosion. Specifically, in restored karst areas, soil translocation from upper to lower slopes was rarely observed. Furthermore, the increased canopy density and enhanced litter accumulation following vegetation restoration effectively intercepted rainfall, reducing surface runoff by approximately 35–40% (based on field measurements) and minimizing direct raindrop impact on soil surfaces. These mechanisms collectively contributed to the preservation of macroaggregates and the stabilization of soil structure.43 In contrast to slope positions, riparian zones exhibited no significant changes in soil aggregate composition following restoration (Figure 3B). Consequently, no statistically significant differences were observed in water-stable macroaggregate content between upslope, downslope, and riparian soils. Although the common garden experiment controlled for variability, it could not fully replicate complex field interactions (e.g., microbial community dynamics). Secondly, the one-year cycle captured initial responses rather than long-term stability. Future studies should include longer-term field monitoring and investigate the role of specific soil biota. Finally, sampling was limited to the 0–20 cm layer; deeper soil layers may exhibit different responses and warrant further investigation. The soil aggregate paradox in early karst vegetation restoration: Mechanisms and implications based on common garden experiments 1In the initial stage of vegetation restoration in karst rocky desertification areas (0–5 years), the mass percentage of water-stable macro-aggregates (>0.25 mm) showed a significant decline, which contradicts conventional understanding: despite increased vegetation cover, soil aggregate stability (as characterized by MWD and GMD) decreased significantly. In karst landscapes, the restoration of native vegetation is a core driver for improving soil aggregate stability. This phenomenon of “microstructural deconstruction coexisting with macroscopic improvement” reflects the nonlinear and stage-specific adaptive processes during the early restoration of fragile karst ecosystems. In the initial stage of vegetation restoration in karst rocky desertification areas (0–5 years), the mass percentage of water-stable macro-aggregates (>0.25 mm) showed a significant decline, which contradicts conventional understanding: despite increased vegetation cover, soil aggregate stability (as characterized by MWD and GMD) decreased significantly. In karst landscapes, the restoration of native vegetation is a core driver for improving soil aggregate stability. This phenomenon of “microstructural deconstruction coexisting with macroscopic improvement” reflects the nonlinear and stage-specific adaptive processes during the early restoration of fragile karst ecosystems. 2Common garden experiments indicate a close interaction between vegetation succession and geomorphology. Geomorphic features significantly influence vegetation growth and succession, while changes in vegetation types contribute to the stabilization of landforms. Although topography alone exhibits no independent effect, it plays a regulatory role: at lower slope positions, the positive influence of vegetation on aggregates is enhanced due to more favorable moisture conditions. This experiment provides an effective approach for elucidating vegetation-geomorphology interaction mechanisms and offers a scientific basis for vegetation restoration practices in karst regions. Common garden experiments indicate a close interaction between vegetation succession and geomorphology. Geomorphic features significantly influence vegetation growth and succession, while changes in vegetation types contribute to the stabilization of landforms. Although topography alone exhibits no independent effect, it plays a regulatory role: at lower slope positions, the positive influence of vegetation on aggregates is enhanced due to more favorable moisture conditions. This experiment provides an effective approach for elucidating vegetation-geomorphology interaction mechanisms and offers a scientific basis for vegetation restoration practices in karst regions. In summary, based on common garden experiments and GLMM analyses, this study preliminarily clarifies the mechanisms through which vegetation restoration affects soil aggregates in karst regions. The research reveals that in highly degraded karst ecosystems, soil structure reconstruction may undergo a critical transitional phase characterized by “deconstruction first, followed by reorganization.” This stage is marked by a temporary predominance of microbial metabolic shifts driven by new carbon inputs over physical protection mechanisms of organic matter, as well as the prioritization of physical reconstruction of root networks over their biogenic cementation functions. From a practical perspective, it is recommended that deep-rooted native shrubs and perennial herbaceous species be prioritized in the early restoration stage to rapidly build soil structure, followed by the gradual introduction of tree species to promote long-term ecosystem development. By establishing a multi-scale evaluation framework that integrates spatiotemporal dynamics and by optimizing vegetation allocation and the regulation of key soil processes, ecosystems can be guided toward a systematic transition from morphological rehabilitation to functional restoration. Limitations of the study Several limitations of this study should be acknowledged. First, while the common garden experiment allowed us to isolate the effects of vegetation type from topographic factors, it cannot fully replicate the complex interactions present in natural field conditions. The controlled environment may not capture all the spatial heterogeneity and temporal dynamics of soil aggregation processes occurring in actual karst ecosystems. Second, this study focused on soil aggregate stability at a specific successional stage following vegetation restoration. Long-term temporal dynamics, including seasonal variations and multi-year trends in aggregate formation and turnover, were not captured. Future studies with extended monitoring periods would help elucidate the temporal trajectory of soil structure improvement under different vegetation restoration strategies. Third, although we identified vegetation type as the dominant driver of aggregate stability, the underlying microbial mechanisms remain underexplored. Soil aggregate formation is closely linked to microbial community composition and activity, yet detailed analyses of microbial contributions (e.g., fungal hyphae and bacterial extracellular polymeric substances) were beyond the scope of this study. Integrating microbial community profiling with aggregate fractionation could provide deeper mechanistic insights. Finally, our study was conducted in a representative karst region of Southwest China (Weining Caohai Nature Reserve). While the findings offer valuable insights for similar ecosystems, caution should be exercised when generalizing these results to other karst areas with different climatic conditions, bedrock types, or disturbance histories. Comparative studies across multiple karst regions would strengthen the broader applicability of our conclusions. Despite these limitations, our study provides robust evidence that vegetation type, rather than topography, is the primary factor regulating soil aggregate stability during karst vegetation restoration. These findings contribute to a mechanistic understanding of soil recovery processes and offer practical guidance for ecological rehabilitation in fragile karst landscapes. Resource availability Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Yunjie Wu (wyjzxcvbn@163.com). Materials availability This study did not generate new unique reagents. Data and code availability Raw data were deposited on Mendeley Data: https://data.mendeley.com/10.17632/DCM4hxbnd7.1. This study did not use any code. Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request. All data generated or analyzed in this study are available from the corresponding author upon reasonable request. Acknowledgments This research was financially supported by the Guizhou Provincial Science and Technology Projects (grant no. ZK[2023]157) and the Guizhou Minzu University Science and Technology Projects (grant no. GZMUZK[2022]YB13), for which the authors are deeply grateful. Author contributions The authors contributed as follows: Q.M. and Y.W. were responsible for study conceptualization and methodology development. Q.M., M.Z., R.Y.J., R.W., and Y.P. conducted the investigation and data collection. Q.M., M.Z., and R.W. performed data analysis. Q.M. and Y.W. wrote the original draft. J.Z., Z.S., and G.Y. participated in manuscript revision and editing. M.Z. and R.Y.J. prepared the visualizations. Y.W. and J.Z. provided supervision and project administration. Y.W., J.Z., and Z.S. acquired funding. ∗Correspondence should be addressed to Y.W. Declaration of interests The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. STAR★Methods Key resources table REAGENT or RESOURCESOURCEIDENTIFIERDeposited dataOriginal datasetThis studyMendeley Data:https://doi.org/10.17632/DCM4hxbnd7.1Software and algorithmsIBM SPSS StatisticsIBM Corp.Version 27.0IBM SPSS AmosIBM Corp.Version 28.0RR Foundation for Statistical ComputingVersion 4.3.1lme4 R packageBates et al., 2015v1.1-34vegan R packageOksanen et al., 2020v2.6-4QGISQGIS Development Teamv3.32.0OtherSoil aggregate wet sieve analyzerNanjing Soil Instrument Co., Ltd.WS-2002Constant Temperature Drying OvenShanghai Boxun Industrial Co., Ltd. Medical EquipmentGZX-9240MBEAnalytical BalanceShenzhen Mumei Technology Co., Ltd.M2000TB-BKjeldahl Nitrogen Determination ApparatusHaineng Future Technology GroupK1160 Method details Detailed methodology: Study area The study area is located within the Caohai National Nature Reserve in Weining County, Guizhou Province, in the central part of the YunnanGuizhou Plateau (26°49′–26°53′ N, 104°12′–104°18′ E). As the largest natural freshwater lake in Guizhou, Weining Caohai features the most unique and intact plateau wetland ecosystem in the region. It exhibits terraced landforms at varying elevations and holds representative significance in terms of its fragility, typicality, importance, and biodiversity among the relatively few subtropical plateau karst wetland ecosystems in China. The wetland is rich in aquatic vegetation. Soil sampling Soil collection began in May 2022. After removing impurities, the soil samples were thoroughly mixed and transported back to the laboratory for analysis of their physical and chemical properties (Table 2). Additionally, dominant plant species for karst-vegetation restoration were interplanted for a common-garden pot experiment. Soil aggregates were determined using the wet-sieving method: three 100g subsamples of air-dried soil were taken and separated via wet sieving through a six-layer sieve stack. The aggregates from each size fraction were collected, oven-dried at 40°C, and their average mass was measured. Experimental treatment Soils were selected from the karst area and the vegetation restoration area within the Jiangjiawan region of the Caohai Nature Reserve in Guizhou, including samples from uphill, downhill, and shoreline locations, as well as the dominant native plant communities from both areas. At the scientific research base of the College of Ecological and Environmental Engineering, Guizhou Minzu University, located on a slope belt (106.6144°E, 26.3772°N), two types of plant growth environments were constructed: soil from the karst area after vegetation restoration, and soil from the original karst area. In each plant growth environment, corresponding native plants from that environment were cultivated, and a cross-planting experiment was conducted for comparative cultivation. Specifically, the following pairings were established: vegetation-restored karst soil with native plants from the vegetation restored area, and original karst soil with native plants from the original karst area; as well as cross-pairings: vegetation-restored karst soil with native plants from the original karst area, and original karst soil with native plants from the vegetation restored area. Each plant type was replicated three times in each cultivation environment to ensure the reliability of the results. The experiment commenced in early May 2022 with soil collection and plant seedling cultivation. After germination, the plants were transplanted into pots. Following a complete growth cycle, destructive soil sampling began in November 2022. The collected soil samples were sealed in zip-lock bags, labeled, stored at low temperature, and transported back to the laboratory. Depending on the experimental requirements, a portion of the soil samples were immediately placed in a −80°C freezer to preserve microbial activity, providing fresh samples for subsequent microbial analysis. Another portion was stored in a −24°C freezer for experiments requiring fresh soil. The remaining soil samples underwent a series of preprocessing steps, including air-drying, grinding, and sieving, to facilitate subsequent analyses. Quantification and statistical analysis Analysis of soil physicochemical properties Soil pH and electrical conductivity were measured using a pH/EC analyzer with a soil-to-water ratio of 1:5. Soil organic carbon was determined using a TOC analyzer. Total soil nitrogen was measured using the Kjeldahl method. Soil bulk density was determined using the core method. Soil moisture content was determined by drying at 105°C. Water-stable aggregates were measured using the wet-sieving method, following procedures described by Cambardella and Elliott as well as Huang et al. In brief, a 50 g soil sample sieved through a 1 cm mesh was transferred to a set of sieves and immersed in deionized water to a depth of 1 cm for 5 min to allow dispersion. Subsequently, the sieve set was moved up and down in an aggregate stability analyzer at a rate of 40 cycles per minute for 2 min, yielding the particle size distribution of the aggregates. Indicators of aggregate stability included: the content of water-stable aggregates >0.25 mm (R0.25), mean weight diameter (MWD), geometric mean diameter (GMD), and fractal dimension (D), which were calculated using the formulas provided below. In the formula, wi is the ratio of the mass of the i-th aggregate component to the total soil mass, x¯i is the average diameter of the particle size aggregate, M(r