The experimental site was located in the experimental monitoring area (34°26′-35°35′N, 103°52′-105°13′E) of Dingxi Research Institute of Soil and Water Conservation, Gansu Province, in the Loess Plateau in northwest China. It pertains to the typical V subregion of the semiarid loess hilly and gully area. The soil type is yellow meadow, with weak erosion resistance and low organic matter content. The study area is characterized by a temperate continental monsoon climate, with an average altitude of 1,900–2,250 m. The mean annual temperature was 6.3°C, and the average annual precipitation was 427 mm, while evaporation reached approximately 1,500 mm. Precipitation is concentrated in summer, primarily occurring as heavy rainfall events.

Before 1999, a significant portion of agricultural land in the study area was abandoned and naturally succeeded in S. bungeana grassland, forming a natural restoration ecosystem that has remained in a state of natural succession. In 1999, on this grassland, Prunus armeniaca L. forest, P. tabuliformis forest, and C. korshinskii shrublands were individually planted as artificial vegetation. In 2001, Medicago sativa L. was planted on the existing S. bungeana grassland. With S. bungeana as the control, sampling points were set up in the planting areas of P. armeniaca, P. tabuliformis, C. korshinskii, and M. sativa in July 2023 to study the effects of vegetation types on soil micro-food web structure and soil multifunctionality.

Soil samples were collected from the rhizosphere soil at a 0–30 cm depth using the five-point sampling method in July 2023, during the peak growth season of artificial vegetation in the Loess hilly area of Longzhong. Four soil samples were obtained from each point, and one composite sample was formed by blending the four individual samples. Each treatment plot had four replicates, and the sampling area was 7 × 7 m. After eliminating impurities from the soil samples, such as stones, gravel, and plant residues, they were thoroughly mixed to form a homogeneous composite sample. The sample was then passed through a 2 mm sieve and promptly preserved on ice. Each soil sample was partitioned into two aliquots for analysis. The first aliquot was placed in a sterilized centrifuge tube, immediately transferred to a foam box containing ice packs, and promptly transported to the laboratory. Then, the aliquot was stored in a −80°C refrigerator to extract the total soil DNA. The second aliquot was placed in a dedicated aluminum box to determine the soil moisture content. The residual soil samples were transported back to the laboratory and air-dried in a shaded area to measure the physical and chemical parameters of the soil.

A total of 10 soil physical and chemical parameters were determined: soil moisture, electrical conductivity, pH, soil organic carbon, total nitrogen, nitrate nitrogen, ammonium nitrogen, total phosphorus, available phosphorus, and available potassium.

Soil moisture was quantified using the oven drying approach at a temperature of (105°C ± 2°C). Electrical conductivity was measured with the METTLER TOLEDO FE38 benchtop conductivity meter. The pH value was ascertained using the glass electrode method. Soil organic carbon was determined using the chromic acid oxidation heating technique (Sanmanee and Suwannaoin, 2009). Total nitrogen was determined using the Kjeldahl nitrogen determination method. Nitrate and ammonium nitrogen were assayed by the colorimetric method in a continuous flow analyzer (Brookes et al., 1985). Total phosphorus was determined using the molybdenum-antimony dichromate colorimetric method (Levine et al., 1955; Li B. B. et al., 2021). NaHCO₃ extraction colorimetric method was used for available phosphorus (Bao, 1998). The available potassium was determined using the CH₃COONH₄ extraction-flame photometry method (Bao, 1998).

Four extracellular enzymes and one polyphenol oxidase associated with the soil carbon, nitrogen, and phosphorus cycles were analyzed. Glucosidase (β-1,4-glucosidase) and sucrase, both related to the soil carbon cycle, were quantified using p-nitrophenol colorimetry and 3,5-dinitrosalicylic acid colorimetry (Bell et al., 2013). Urease, which is linked to the soil nitrogen cycle, was measured using indophenol blue colorimetry (Tian et al., 2017; Jiao et al., 2018). Alkaline phosphatases connected to the soil phosphorus cycle were assessed using alkaline phosphonodisodium phosphate colorimetry (Li H. Z. et al., 2021). Polyphenol oxidase activity was determined using pyrogallol colorimetry (Ghiloufi et al., 2019).

Soil multifunctionality was assessed using the following variables: (1) soil environmental factors (soil moisture, electrical conductivity, and pH), (2) carbon cycling nutrients (soil organic carbon), (3) nitrogen cycling nutrients (total nitrogen, nitrate nitrogen, and ammonium nitrogen), (4) phosphorus cycling nutrients (total phosphorus and available phosphorus), (5) soil carbon cycling enzymes (ß-1,4-glucosidase and sucrase), nitrogen cycling enzyme (urease), and phosphorus cycling enzyme (alkaline phosphatases). The average method was used in this process, yielding similar results to the multi-threshold method (Jing et al., 2015). Each variable was normalized using a Z-score transformation, after which the versatility index was calculated by averaging the normalization rates of the variables (Ma et al., 2022). The Z-score conversion of variables was performed using SPSS version 19.0.

Soil DNA was extracted from a 0.5 g soil sample using the E.Z.N.A Soil kit (Omega Bio-tek, Norcross, GA, United States). The concentration and purity of the extracted DNA were evaluated with a NanoDrop 2000 UV-VIS spectrophotometer (Thermo Scientific, Wilmington, United States), and the quality of the extraction was assessed through 1% agarose gel electrophoresis (Porazinska et al., 2010; Tian, 2022).

Primers NF1 (5′-GGTGGTGCATGGCCGTTCTTAGTT-3′) and 18S r2bR (5′-TACAAA GGGCAGGGACGTAAT-3′) (Feng et al., 2017) were employed to amplify the V4 segment of soil nematode DNA (Du et al., 2020) under the following conditions: 95°C pre-denaturation for 3 min; 95°C denaturation for 30 s, 55°C annealing for 30 s, 72°C elongation for 45 s, for 35 cycles; 72°C elongation for 10 min, and storage at 4°C. For bacteria, the V3-V4 region of the 16S rRNA gene was amplified using the primers 515F (5′-GTGCCAGCMGCCGCGG-3′) and 907R (5′-CCGTCAATTCMTTTRAGTTT-3′) (Wang and Wang, 1996). The amplification conditions were as follows: pre-denaturation at 98°C for 2 min; denaturation at 98°C for 15 s, annealing at 55°C for 30 s, extension at 72°C for 30 s, final extension at 72°C for 5 min, with 30 cycles. For fungi, the ITS1 region was amplified using the primers ITS1F (5′-CTTGGTCATTTAGAGGAAGTAA-3′) and ITS1R (5′-GCTGCGTTCTTCATCGATGC-3′) (Yang et al., 2017). The amplification conditions were as follows: pre-denaturation at 95°C for 5 min, denaturation at 95°C for 1 min, annealing at 50°C for 1 s and extension at 72°C for 1 min, and final extension at 72°C for 7 min, with 15 cycles. PCR was conducted by employing Trans Start Fastpfu DNA Polymerase (Trans Gen AP221-02) on a PCR instrument (ABI Gene Amp^® 9700 type) (Fan et al., 2023).

The procedure was repeated more than three times for a single sample, and the PCR products obtained from the same sample were uniformly mixed and analyzed using 2% agarose gel electrophoresis. The PCR products were recovered utilizing the Axy Prep DNA Gel Recovery Kit (AXYGEN, United States), eluted with Tris-HCl, and re-evaluated through 2% agarose gel electrophoresis. Quantitative detection of the PCR products was conducted using the Quanti Fluor-ST Blue fluorescence quantification system (Promega, Beijing, China). Following the sequencing volume requirements for each sample, the PCR products were combined in appropriate proportions. The MiSeq library was constructed using polymerized DNA products and sequenced on Illumina MiSeq PE300, a high-throughput sequencing platform (performed by Shanghai Meiji Biomedical Technology Co., Ltd.). The resulting high-throughput sequencing data were subsequently utilized for further data analysis.

Significant differences were observed in the Shannon indices of nematode and fungal communities across different vegetation types (P < 0.05). Specifically, compared to P. tabuliformis, the Shannon indices of nematode and fungal communities in P. armeniaca increased by 113 and 62%, respectively. In comparison, those in C. korshinskii increased by 106 and 63%, respectively (Figures 1A–C). Moreover, vegetation type significantly influenced the community composition of soil microorganisms and nematodes (P = 0.001, stress < 0.15) (Figures 1D–F). In terms of relative abundance, omnivorous-predatory nematodes were most abundant in S. bungeana (relative abundance > 77%), whereas plant-parasitic nematodes predominated in other vegetation types (relative abundance > 72%) (Figure 1G). In M. sativa, the bacterial genus Arthrobacter exhibited the highest relative abundance (25%), whereas RB41 showed the highest relative abundance in all other treatments (> 19%) (Figure 1H). Among fungal genera, Inocybe exhibited the highest relative abundance in P. tabuliformis (82%), whereas Mortierella had the highest relative abundance in all other treatments (> 47%) (Figure 1I). Compared with S. bungeana, at the genus level of nematodes (Figure 1J), the proportion of Campydora in P. tabuliformis, M. sativa, and C. korshinskii treatments was significantly lower (P < 0.05); at the bacterial genus level (Figures 1K, H), the proportion of Microlunatus and Rubrobacter in P. armeniaca and Arthrobacter in M. sativa were significantly higher (P < 0.05), while the proportion of Microlunatus in M. sativa was significantly low (P < 0.05). Additionally, the proportion of Gaiella in M. sativa and C. korshinskii was significantly low (P < 0.05). At the fungal genus level (Figures 1I, L), the proportion of Chaetomium in M. sativa and Cladophialophora in P. armeniaca showed a significant increase compared to that in S. bungeana (P < 0.05) (Figures 1J, L).

As illustrated in Figures 2A–E, the soil microbial and nematode networks of C. korshinskii exhibited the highest number of connections, followed by M. sativa, P. tabuliformis, and S. bungeana, with P. armeniaca displaying the fewest connections. Additionally, all treatments had more positive connections than negative ones (Table 1). In S. bungeana, P. tabuliformis, and M. sativa treatments, bacteria primarily drove the degradation channel in the soil micro-food web. In P. armeniaca, fungi were the primary drivers. In C. korshinskii, bacteria, and fungi influenced the soil micro-food web (Figures 2F–J). Furthermore, the analysis of carbon flow in the soil micro-food web (Figure 3) showed that fungi contributed the most to soil organic carbon in C. korshinskii. In contrast, bacteria contributed the most to soil organic carbon in P. armeniaca.

The plant parasitic nematode metabolic footprint and the total nematode metabolic footprint were highest in C. korshinskii, at 26.44 μg⋅g^–1 and 28.65 μg⋅g^–1, respectively, which were 18 times and 2.2 times higher than those in S. bungeana. The omnivorous-predatory nematode metabolic footprint was 0 μg⋅g^–1 in both P. tabuliformis and M. sativa (Figure 4A). The metabolic footprint-based nematode faunal analysis (Figure 4B) showed that C. korshinskii, P. armeniaca, and S. bungeana were located in quadrant C, whereas P. tabuliformis and M. sativa were located in quadrant D. The food web energy flow analysis (Figure 4C) showed that C. korshinskii had the highest proportion in the plant energy flow channel (98.68%), which was 46% greater than that of S. bungeana. Furthermore, the ratio of the plant-parasitic index to the maturity index of M. sativa and the basic indices of P. tabuliformis (50) and M. sativa (49.25) were significantly higher than those of other vegetation types (P < 0.05) and the channel index of all five vegetation types exceeded 50 (Figures 4D–F).

The composition of soil nematodes and microbial communities was significantly correlated with pH, electrical conductivity, total nitrogen, sucrase, alkaline phosphatases, soil organic carbon, the ratio of carbon to nitrogen, and urease (0.01 < P < 0.05) (Figure 5A and Supplementary Figure 1). The ratios of carbon to nitrogen, soil organic carbon, and electrical conductivity were the primary factors influencing plant-parasitic nematodes, while nitrate nitrogen, total phosphorus, and available phosphorus were the primary factors for omnivorous-predatory nematodes. Urease was the primary factor for bacteria (P < 0.05), and total nitrogen, total phosphorus, the ratio of carbon to nitrogen, available potassium, and sucrase were the primary factors for fungi (Figures 5B–E). Redundancy analysis (RDA) showed that environmental factors explained 85.21% of the total variance in soil nematode community abundance (RDA1: 61.22%, RDA2: 23.99%), with pH and total phosphorus significantly affecting the community composition (P < 0.05) (Figure 5F). For soil bacteria, environmental factors explained 74.14% of the total variance (RDA1: 54.94%, RDA2: 19.20%), with soil organic carbon being the primary factor (P < 0.05) (Figure 5G). For soil fungi, environmental factors explained 89.58% of the total variance (RDA1: 49.49%, RDA2: 40.09%), with total phosphorus and the ratio of carbon to nitrogen having the most significant effects (P < 0.05) (Figure 5H).

As illustrated in Figure 6A, compared with S. bungeana, C. korshinskii enhances soil multifunctionality, whereas M. sativa exhibits a significant reduction (P < 0.05). In addition, the key variables for predicting soil multifunctionality included β-1,4-glucosidase, urease, soil organic carbon, total nitrogen, electrical conductivity, available potassium, sucrase, total phosphorus, alkaline phosphatases, and pH (Figure 6B). Soil pH and carbon cycling nutrients exhibited an extremely significant positive correlation with soil multifunctionality (P < 0.01). Nitrogen-cycling enzymes, phosphorus-cycling enzymes, and nitrogen-cycling nutrients showed a significant positive correlation (P < 0.05), while phosphorus-cycling nutrients had an extremely significant negative correlation (P < 0.01). Soil bacteria also exhibited a significant negative correlation (P < 0.05) (Figure 6C). Nutrients related to carbon cycling (0.487) and phosphorus cycling (0.124), as well as enzyme activities (0.440 and 0.382) were the primary positive effect factors influencing the structure of the soil micro-food web and the soil multifunctionality. In contrast, nutrients related to nitrogen cycling (-0.516) were the primary negative effect factors (Figure 6D).

Vegetation types differ in their root exudates, litter composition, and decomposition rates, influencing soil water and organic carbon distributions. These changes affect the composition and diversity of soil microbial and nematode communities, shaping the structure of the soil micro-food web (Lavelle, 1997; Johnson et al., 2003; Cesarz et al., 2013; Cai et al., 2022; Sha et al., 2023). Following artificial forestation in karst-degraded areas, the soil nematode community diversity index increased, and the nematode food web developed (Hu et al., 2016), which is consistent with this study. This study revealed that bacteria were the primary drivers of the degradation channel in the soil micro-food web in S. bungeana, P. tabuliformis, and M. sativa. At the same time, fungi were the dominant drivers of P. armeniaca. In C. korshinskii, both bacteria and fungi influenced the soil micro-food web, indicating that C. korshinskii and P. armeniaca have a greater capacity to protect carbon pools (Prescott and Vesterdal, 2021). Therefore, establishing C. korshinskii and P. armeniaca in Loess hilly areas not only facilitates the regeneration of understory herbaceous vegetation but also significantly improves soil structure and nutrient content, thereby enriching the soil with a considerable number of usable resources (Guan et al., 2015). However, as plantation age increases, the plant communities of C. korshinskii shrub forests degrade, reducing soil organic carbon, total nitrogen, and soil moisture, thus limiting the development of soil nematode communities (Li et al., 2015). In contrast, P. armeniaca arbor forest, characterized by favorable stand attributes and effective interception of rainfall through leaf gaps, mitigates soil erosion, enhances soil nutrients, and improves the structure of the soil micro-food web. Therefore, when restoring vegetation by planting C. korshinskii shrub forests in the region, it is imperative to establish appropriate P. armeniaca arbor forests on gentle slopes to safeguard the carbon pool.

The characteristics of soil nematode metabolic footprints provide effective methods and indicators for enhancing the study of soil multifunctionality by analyzing the carbon metabolic functions of different nematode groups and energy pathways within the soil micro-food web. This study revealed that the plant-parasitic nematode metabolic footprint and total nematode metabolic footprint of C. korshinskii were significantly higher than those of other vegetation types. This result indicates that the C. korshinskii shrub forests enhanced the metabolic activity of plant-parasitic and composite channels in soil nematodes. Given that plant-parasitic nematodes can effectively promote the allocation of photosynthate from plants to the rhizosphere, establishing artificial C. korshinskii shrub forests can not only enhance plant root exudates and microbial activity but also improve plant productivity (Ferris et al., 2001; Pan et al., 2021; Ruan et al., 2021).

Furthermore, metabolic footprint-based nematode faunal analysis revealed that C. korshinskii, P. armeniaca, and S. bungeana were located in quadrant C. This result contributed to maintaining soil ecosystem stability and enhancing the connectivity and complexity of soil micro-food webs. Conversely, P. tabuliformis and M. sativa are situated in quadrant D, exacerbating the degree of soil micro-food web disturbance in the area, leading to severe depletion of soil nutrients and structural degradation of soil micro-food webs. Studies by Cui et al. (2020) and Cai et al. (2022) also demonstrated that the soil micro-food web structure of shrubs and grassland vegetation types is more stable. However, P. tabuliformis, characterized by deep roots and strong allelopathic effects, is better suited for growth in fertile soil and deep soil layers. However, the soil in the study area was classified as yellow loess soil, characterized by its loose structure, low nutrient content, and severe erosion. Consequently, establishing P. tabuliformis in this region exacerbates soil erosion, leading to a decline in soil nutrients and a reduction in the diversity of soil microorganisms and nematodes. This, in turn, would seriously impede plant root exudation and microbial activity, thereby aggravating ecological degradation. In this study, the alfalfa meadow selected for M. sativa treatment had only been established for two years, resulting in low soil nematode metabolic activity, plant productivity, and significant soil micro-food web structure degradation. These findings differ from those reported in other studies (Hu et al., 2017).

Different vegetation types influence soil physical and chemical properties as well as the structure of soil micro-food webs (Maestre et al., 2012; Zhang R. et al., 2021), thereby affecting soil multifunctionality (Aponte et al., 2013; Wang et al., 2022; Xu et al., 2022). Han et al. (2022) and Li et al. (2022a) reported that bacterial diversity and community composition in subtropical forests and fungal diversity in northern forests are the primary drivers of soil multifunctionality, respectively. This study found that C. korshinskii and P. armeniaca significantly promoted soil multifunctionality compared to S. bungeana. Enzyme activities related to carbon metabolism (β-1,4-glucosidase, sucrase) and nitrogen metabolism (urease) were high in C. korshinskii and low in P. tabuliformis and M. sativa. This is because the stand characteristics and leaf gaps of P. tabuliformis hinder its ability to effectively intercept rainfall, leading to severe damage to the soil structure, reduced soil permeability, and increased soil degradation. Additionally, this results in substantial mortality of surface vegetation, significant loss of organic matter from both the soil surface and interior, and a reduction in the input of soil organic matter at the source (Agarwal, 1999). Consequently, this lowers the stability of soil micro-food webs and harms soil multifunctionality (Li et al., 2022b).

Therefore, establishing C. korshinskii can effectively mitigate soil erosion and enhance soil organic matter input in regions characterized by poor soil quality and severe soil erosion, such as Loess hilly areas. This restores soil nutrients, maintains the stability of soil micro-food web structures, and promotes soil multifunctionality (Figure 7). Additionally, P. armeniaca not only exhibits traits of drought resistance, tolerance to poor soils, and salt-alkali resistance but also produces litter that can be efficiently decomposed and absorbed by the soil, contributing to increased storage of soil organic carbon. Consequently, to better enhance soil multifunctionality, the appropriate integration of C. korshinskii and P. armeniaca is essential.

As the diversity of vegetation increases, the total rate of photosynthesis and the total carbon pool in the ecosystem also increase (Hu et al., 2017), leading to faster and more intense nutrient cycling, thereby promoting material transformation in the soil ecosystem (Zhang C. et al., 2016; Cui et al., 2020). Therefore, in regions with severe ecological degradation, such as Loess hilly areas, relying solely on a single type of vegetation for ecological restoration is insufficient. A comprehensive system comprising arbors, shrubs, and grasslands must be established and appropriately configured in space and time to effectively protect the ecological environment. The findings of this study can aid in predicting the responses of soil micro-food webs and soil multifunctionality to different vegetation types in plant production systems, thereby facilitating the identification of the most suitable vegetation types for specific areas and promoting the enhancement of the ecological environment.