Impact Mechanisms of Different Ecological Forest Restoration Modes on Soil Microbial Diversity and Community Structure in Loess Hilly Areas

Abstract

The Loess Plateau, with a fragile ecological environment, is one of the most serious water- and soil-eroded regions in the world, which has been improved by large-scale projects involving returning farmland to forest and grassland. This work is mainly aimed at exploring a more reasonable and efficient ecological forest restoration mode and revealing synergistic restoration mechanisms. This study sampled typical Loess Plateau areas and designed the restoration modes for pure forests of Armeniaca sibirica L. (AR), Amygdalus davidiana (Carrière) de Vos ex Henry. (AM), Medicago sativa L. (MS), and mixed forests of apricot–peach–alfalfa (AR&AM&MS), using abandoned land (AL) as a control treatment. The effects of these modes on the physical and chemical properties and enzyme activities of various soils were investigated in detail. Moreover, the soil microbial diversity and community structure, functional gene diversity, and differences in the restoration modes were deeply analyzed by meta-genomic sequencing technology, and the inherent driving correlation and mechanisms among these indicators were discussed. The results showed that the soil water content and porosity of the AR, AM, and AR&AM&MS treatments increased significantly, while the bulk density decreased significantly, compared with AL. Moreover, the total carbon, total nitrogen, nitrate nitrogen, total phosphorus, available phosphorus, total potassium, and available potassium contents of the AR&AM&MS restoration mode increased significantly. Compared to CK, there was no significant change in the catalase content of pure forest and mixed forest; however, the contents of urease, phosphatase, sucrase, B-glycanase, and N-acetylglucosaminidase in the restoration mode of the mixed forest all increased significantly. The species diversity index of the restoration modes is similar, and the dominant bacteria in soil microorganisms include Proteobacteria, Acidobacteria, Actinobacteria, Bacteroidetes, and Gemmatimonadetes. The mixed forest restoration mode had the highest microbial abundance. The functional gene diversity of the different restoration modes was also similar, including kegg genes, eggNOG genes, and carbohydrate enzymes. The functional genes of the mixed forest restoration mode were the most abundant, and their restoration mechanism was related to the coupling effect of soil–forest grass. After evaluation, the restoration mode of mixed forest was superior to that of pure forest or pure grass. This is attributed to the fact that the mode can improve soil structure, retain soil moisture, enhance soil enzyme activity, optimize soil microbial community structure, and improve microbial diversity and functional gene activity. This provides key data for the restoration of fragile ecological areas, and the promotion of sustainable management of forests and grass in hilly areas of the Loess Plateau.

1. Introduction

The loess hilly region is an important part of the Loess Plateau. Over time, the local fragile ecosystem has been further damaged due to population growth and unreasonable human activities such as blind reclamation and overgrazing [1]. This has led to increased soil erosion, water shortages, degraded pastures, and desertification. In particular, soil erosion can lead to the thinning of the soil layer of arable land, a decline in fertility, and a reduction in crop yields [2]. A lot of sediment entering rivers may reduce water quality and cause further deterioration of the local ecological environment. Increasing ecological problems have fragmented the terrain of the region, making it one of the most seriously eroded areas in China and even in the world [2,3]. Ecological forest restoration is a key measure to control soil erosion and enhance ecological functions [4]. To control soil erosion in the region, China has successively implemented different ecological restoration projects. Particularly, the implementation of the Grain for Green Program for more than ten years has effectively restored vegetation in the region, significantly increased vegetation coverage, and significantly improved the ecological environment [5]. During the recovery process, although vegetation can provide better ecological benefits in most habitats, some areas have experienced poor growth and low ecological benefits [6]. Some reports suggest that land use change is one of the main causes of environmental degradation, and plays an important role in the increase of carbon dioxide concentration in the global atmosphere, global climate change, and the loss of ecosystem functional services [7,8]. In recent years, there has been growing global interest in assessing the impact of land use change and ecological forest restoration on soil quality [9]. During land use changes from forest and grassland to farmland, the soil microclimate is greatly affected due to the loss of perennial vegetation and changes in planting methods. This process initially disrupts the balance between the availability of plant-derived organic matter and microbial decomposition in the soil, and ultimately affects soil quality and health [10,11]. Therefore, the question of how to avoid the formation of this inefficient artificial vegetation and improve the function of existing inefficient vegetation by transformation is a key scientific issue that needs to be urgently solved for the restoration and reconstruction of vegetation in ecologically fragile areas [12].

Soil microorganisms are the main drivers of nutrient transformation and cycling in terrestrial ecosystems, and play an important role in promoting material exchange, energy flow, microbial metabolic activity, and plant growth [13]. Soil microorganisms can provide a good habitat for plants, improve soil texture and structure, and are the basis for the normal operation of the soil ecosystem. Therefore, studying the diversity of soil microbial communities is extremely important for understanding the structure, operating mechanism, and ecological functions of the ecosystem [13,14]. Previous studies have shown that soil microbial activity and community structure can change significantly under different land use patterns [14,15]. The vegetation in the restoration is generally grassland, forest land, and mixed forest and grassland. The root systems of grassland vegetation can reach only shallow depths in the soil, and the impact on the soil microorganisms in the soil layer is also less than that of forest land. However, the root systems of forest land plants are very deep in the soil, and their impact on soil microorganisms is bound to be wider [15]. Additionally, the soil quality and ecological function of mixed forest restoration methods are superior to monoculture plantations. The soil microorganisms in the cross-section will show their unique community composition, which will in turn affect the restoration of fragile ecosystems [16,17]. This means that geographical location, topography, environment, and vegetation are important factors influencing the structure and diversity of microbial communities. Meanwhile, the succession of vegetation causes significant changes in the structure, function, and diversity of soil microbial communities [14,18]. However, current research on soil ecosystem microorganisms mainly focuses on forest soil, degraded grassland soil, and soil under different orchard and mining conditions for different species, habitats, and depths [16]. There is less research on the composition and diversity of microorganisms in fragile ecosystems caused by differences in ecological forest vegetation. Specifically, in recent years, ecological forest restoration modes have seen a wide range of cultivated areas for artificial grasslands such as apricot (Armeniaca sibirica L., AR), peach (Amygdalus davidiana (Carrière) de Vos ex Henry, AM), alfalfa (Medicago sativa L., MS), and peach–apricot–alfalfa (AR&AM&MS). The impact of these restoration modes on the structure and diversity of soil microbial communities deserves further research.

Furthermore, soil enzymes are active participants in soil biochemical processes, and their activity can indirectly reflect the diversity of soil microbial functions. Enzymes produced by microorganisms and plants are closely related to soil energy flow and nutrient cycling, and they respond quickly to soil changes [19,20]. These enzymes are considered to be direct participants in the carbon, nitrogen, and phosphorus cycles [20]. Many studies have confirmed the high correlation between soil enzyme activity and bacterial and fungal communities [21,22]. Fungal communities have a greater impact on phosphatase, urease, and protease activities, while bacterial and actinomycete communities have a greater impact on sucrase. Microorganisms and plant root systems can secrete various enzymes to supply the soil, ensuring the smooth progress of various biological metabolic processes in the soil [21]. In addition, soil extracellular enzymes are special enzymes secreted by microorganisms into the soil and obtain energy and nutrients from the soil. They play a regulatory role in the decomposition of organic matter by soil microorganisms and the absorption and utilization of soil nutrients such as C, N, and P by microorganisms [23]. For example, β-1,4-glucosidase (BG) is mainly used to obtain carbon; β-1,4-N-acetylgl-ucosaminidase (NAG) is mainly used to obtain nitrogen; and acid or alkaline phosphatase (AP) is mainly used to obtain phosphorus [20]. Taken together, exploring the relationship between soil enzyme activity and soil organisms can help determine the response of microbial communities to the ecosystem, providing insight into ecosystem structure. Moreover, many scholars have used high-throughput sequencing technology to thoroughly and accurately investigate the distribution and diversity of soil microbial communities under different land use patterns [12,24]. This mainly involves metagenomic sequencing technology, among which Illumina Solexa technology is more commonly used [25].

Given this, this work aims to investigate the impact of different restoration modes on the physicochemical properties and the structure and diversity of soil microbial communities. Furthermore, the response of soil microorganisms to land use changes or soil resource availability will be explored, and the mechanism of the impact of different restoration modes on the structure of soil microbial communities will be further revealed. Based on the characteristics of vegetation distribution in the fragile area of the Loess Plateau, combined with the land use pattern, this work proposes using abandoned land (AL) as a control treatment, with AR, AM, MS, and AR&AM&MS as typical sample fields. The main highlight is the systematic study of the community structure and functional diversity of soil microorganisms (including bacteria and fungi) under different ecological forest restoration modes. The goal is the identification of the influencing laws on the physical and chemical properties and enzyme activity of the soil. The coupled relationship between soil microbial diversity and environmental factors is emphasized. This provides a decision-making reference for the optimization of forest and grassland vegetation, ecosystem management, and restoration in the hilly area of the Loess Plateau, to better guide ecological restoration in the area.

2. Materials and Methods

2.1. Research Area

The research area is located in Zhongzhuang Village, Pengyang County, Ningxia Hui Autonomous Region, China, as shown in Figure 1. The research area has a typical temperate continental climate and is located in the middle of the Loess Plateau. The terrain is hilly. The annual average precipitation in the research area is about 433.6 mm (22a), the annual average temperature is 7.4 °C, and the altitude is 1400–1900 m (these data come from the local meteorological department).

2.2. Experimental Design

This study selected pure forest (plantation), pure grass, forest + shrubbery + grass, and abandoned land (unfarmed land, 20a) with different plant configurations as research objects in each vegetation zone. Considering that AL is located within the study area and is not disturbed by man and is not cultivated, it provides scientifically valid background data for the present work. Typical vegetation types for farmland returning to the forest were selected, including the four ecological–forest restoration modes of Armeniaca sibirica L. (AR), Amygdalus davidiana (Carrière) de Vos ex Henry. (AM), Medicago sativa L. (MS), and the above mixed forests of apricot–peach–alfalfa (AR&AM&MS) as experimental plots. The selected forestlands were all mature forests. Five fixed standard plots (20 m × 20 m) were randomly set for each vegetation type, and five trees with straight trunks and full crowns were selected as standard trees. Five sample points are randomly selected within a sample plot 1 m from the base of a standard wooden tree trunk. After removing surface plant debris and stones, five soil samples are taken from 0 to 20 cm using a 5 cm diameter soil auger. The soil is mixed well and used as a sample. Each soil sample is sealed in a sterile plastic bag and stored quickly in an incubator (below 0 °C) for later analysis in the laboratory.

2.3. Soil DNA Extraction, Amplification, and Illumina MiSeq High-Throughput Sequencing

(1) Extraction of soil bacterial and fungal DNA: Soil microbial DNA was extracted using the Power Soil^® DNA kit (MOBIO Laboratories, Carlsbad, CA, USA), DNA was detected on 1% agarose gel, and DNA was purified using the Gel Recovery Kit (Tian Gen Biotech Co. Ltd., Beijing, China). The DNA was purified by Gum Recovery Kit (Tian Gen Biotech Co. Ltd.) and stored in the refrigerator at −40 °C after purification, with three replicates for each soil sample [6,7].

(2) qRT-PCR quantitative analysis: TransStartFastPfu DNA polymerase was used in the PCR process, and the PCR instrument used ABI GeneAmp^®9700 (Foster City, CA, USA). PCR reaction steps and conditions are as follows: denaturation at 95 °C for 30 s, to ensure DNA double-helix hydrogen-bond breakage, and the formation of a single-stranded DNA as a template for the reaction; 55 °C-annealed for 30 s, at the complementary region of primer and template binding to form a template-primer complex. After annealing at 55 °C for 30 s, the primer and the complementary region of the template form a template–primer complex; a 72 °C extension in implemented for 45 s, with the primer as a fixed starting point, under the action of DNA polymerase, to synthesize a new DNA chain [6,7]. The above three steps were repeated as one cycle a total of 35 times. The PCR products were quantified by QuantiFluor TM-ST blue fluorescence system, and then mixed according to the sequencing amount of each sample. Finally, the MiSeq library was constructed and sequenced.

(3) Illumina MiSeq High-Throughput Sequencing. Raw sequences were trimmed, merged, and assigned in QIIME software (v1.8.0, http://www.qiime.org/, accessed on January 2022–November 2024). Low-quality sequences (length < 150 bp) were removed. High-quality sequences were clustered into operational taxonomic units (OTUs) with 97% similarity. The number of OTUs, the α-diversity index, Chao1 index, Shannon, and Simpson indices were calculated in QIIME. Differences between treatments were also analyzed using ANOSIM analysis.

2.4. Determination of Basic Physical and Chemical Properties of Soil

Soil moisture content was determined using a common drying method. Soil bulk density was determined using ring knife sampling and weighing. Soil porosity was calculated according to the following formula: soil porosity = (1 − soil bulk density/soil density) × 100%. Soil aggregate structure was determined using a TPF-100 soil aggregate structure analyzer. Soil pH was determined using a pH meter (PHS-3E model, INESA Scientific Instrument Co., Ltd., Shanghai, China) with a soil-to-water ratio of 1:2.5. Soil total salt is measured using the conductivity method. Soil alkalinity, also known as soil exchange sodium percentage (ESP), is expressed as the percentage of exchangeable cations in the total exchangeable cations adsorbed by soil colloids. For detailed measurement methods for this indicator, please refer to (LY-T 1249-1999 [26], China). Soil total nitrogen is measured using the sulfuric acid digestion–Kjeldahl nitrogen analyzer method (microKjeldahl method, LY/T 1228-1999 [27], China). Soil alkali–soluble nitrogen is measured using the alkali-soluble diffusion dish method (LY/T 1229-1999 [28], China). For acidic soil, the soil-available phosphorus is determined using the hydrochloric acid-ammonium fluoride-leaching molybdenum–antimony anti-colorimetric method (LY/T 1233-1999 [29], China); for neutral and alkaline soil, the soil available phosphorus is determined using the sodium bicarbonate-leaching molybdenum–antimony anti-colorimetric method (LY/T 1233-1999, China). Soil available potassium is determined using ammonium acetate solution extraction–flame photometry. The specific steps of these test methods can be found in previously published papers [7,16,17,25].

2.5. Measurement of Soil Enzyme Activities

(1) Catalase: titration with potassium permanganate [19,22]. A 2 g air-dried soil sample sieved through 1 mm was weighed and placed in a 100 mL triangular flask and filled with 40 mL of distilled water and 5 mL of 0.3% H₂O₂ solution. The triangular flask was placed on a round-trip oscillator at 120 r min⁻¹ for 20 min, and then 5 mL of 3 mol L⁻¹ sulfuric acid was added immediately to stabilize the undecomposed hydrogen peroxide, and then the suspension in the flask was filtered through a quantitative filter paper. A total of 25 mL of the filtrate was aspirated and titrated with 0.1 mol L⁻¹ KMnO₄ to a pale pink endpoint.

(2) Urease: indophenol blue colorimetric method [19,22]. A total of 5 g of air-dried soil sample sieved through a 1 mm sieve was weighted and placed in a 50 mL volumetric flask. From there, 1 mL of toluene was added to the volumetric flask (for as long as it could completely wet the soil sample) and it was left for 15 min; then, 10 mL of 10% urea solution and 20 mL of citric acid buffer (pH ≈ 6.7) were added, and the result was mixed carefully and shaken well. The volumetric flask was then placed in a thermostat at 37 °C and incubated for 24 h. At the end of the incubation period, the suspension was diluted with purified water heated to 37 °C, shaken carefully, and then filtered through a triangular flask using quantitative filter paper. A total of 3 mL of filtrate was pipetted into a 50 mL volumetric flask, 10 mL of distilled water was added, it was shaken well, then 4 mL of sodium phenol was added; this was then mixed carefully, before adding 3 mL of sodium hypochlorite, shaking well, leaving to stand for 20 min, diluting with water to scale, and then the solution showed the blue color of indophenol. The color-developed solution was colorimetrically determined at 578 nm within 1 h using a spectrophotometer.

(3) Phosphatase: disodium phosphate colorimetric method [19,22]. Using 5% disodium phosphate as the hydrolysis substrate, 1 g of air-dried soil sample was put into a 50 mL conical flask, toluene was used to pretreat the soil, 20 mL of 0.5% disodium phosphate was added, and the sample was filtered at 37 °C for 24 h in the incubator. A total of 1 mL of the filtrate was taken into a 50 mL colorimetric tube, and 5 mL of borate buffer, 0.5 mL of 8% potassium ferricyanide, and 0.5 mL of 2% 4-aminoamino aminostilbene were added to it. Then, it was shaken well, the color development was fixed for 25 min, and the absorbance value was measured at 510 nm with a spectrophotometer. The control group was treated with distilled water instead of substrate, and 3 groups of parallel samples were set up. The results were expressed as the mass of phenol produced by 1 g of soil incubated at 37 °C for 1 h.

(4) Measurement of soil extracellular enzyme activity: BG, NAG, and sucrase activities were determined by fluorescence assay using 96 microtiter enzyme labeling plates [19,22]. A total of 2.75 g of fresh soil sample was weighed in a triangular vial, 91 mL of buffer (pH adjusted according to the measured soil samples) was added, and then it was shaken on a shaker at 200 r min⁻¹ for 30 min. A total of 250 μL of fluorescently labeled C, N, and P substrate solution at a concentration of 200 μmol L⁻¹ was inoculated into a deep-well plate as a substrate plate. We then inoculate 250 μL of 4-methylumbelliferone at different concentrations into the deep-well plate as a marker plate, and pipette 800 μL of soil suspension into a 96-well plate, seal the deep-well plate with self-adhesive sealing film after all the samples have been added, and invert the deep-well plate repeatedly to shake and mix. The plates were incubated in an incubator at 25 °C for 4 h. After incubation, the plates were centrifuged at 4 °C for 3 min at 4800 r min⁻¹ and 200 μL of supernatant was aspirated. The plates were excited at 365 nm and the fluorescence was detected at 450 nm.

2.6. Data Processing and Analysis

The data obtained from metagenomic sequencing were statistically annotated using the QIIM toolkit (version 1.8.0) to obtain the number of OTUs. Additionally, the analysis of the microbial community structure diversity and functional gene diversity was carried out using the BMKCloud platform (www.biocloud.net, ©2024 biocloud.net; accessed on January 2022–November 2024). This involved cluster analysis and principal component analysis (PCA) relevant models. A detailed description of the analytical models used for this work can be found on this cloud computing platform. Data analysis was performed using SPSS 22.0, with one-way analysis of variance (ANOVA) and least significant difference (LSD) multiple comparisons (p < 0.05). The bar charts were created using Origin 2019 software.

3. Results Analysis

3.1. Influence Analysis of Soil Physico-Chemical Properties

After investigation, the physical properties of the soil under different ecological forest restoration modes are shown in

Table 1. Physical properties of soils in different restoration modes.

Treatments	Soil Moisture (%)		Bulk Density (g cm−3)		Porosity (%)		Mechanical Composition (%)
							0.05–2.0 mm		0.002–0.05 mm		<0.02 mm
	Average	Error	Average	Error	Average	Error	Average	Error	Average	Error	Average	Error
AR	19.29ab	1.08	1.20b	0.04	54.63a	1.60	33.71a	2.53	59.42c	4.01	3.62a	0.22
AM	20.75a	1.71	1.19b	0.08	55.13a	2.87	32.17ab	4.00	61.86bc	6.74	3.58a	0.26
MS	17.77b	0.34	1.36a	0.08	48.77b	3.08	17.57c	4.25	77.64a	3.64	4.46a	1.45
AR&AM&MS	20.45a	1.31	1.15b	0.07	56.45a	2.83	29.38b	1.77	65.54b	2.20	4.27a	0.56
AL	18.43b	0.57	1.34a	0.04	49.45b	1.62	33.94a	1.60	61.52bc	2.65	3.98a	0.68

Note: Letters represent significant differences between different soil layers in the same stand (p < 0.05).

. Compared to abandoned land, the water content of all restoration modes has increased. Among them, the water content of the AR, AM, and AR&AM&MS treatments increased significantly. This means that the increase in soil water content is mainly related to the forest structure, such as involves trees. In terms of bulk density, except for the MS treatment, the bulk densities of the other restoration modes were significantly lower than those of AL (p < 0.05). This is consistent with the results of soil moisture content. A lower bulk density leads to more developed soil pores and therefore a higher soil moisture content. Additionally, soil porosity is highly correlated with this phenomenon. It can be seen from Table 1 that the porosity of the AR, AM, and AR&AM&MS treatments is significantly higher than that of the AL treatment and the MS restoration mode. Therefore, it can be inferred that AR&AM&MS is the best forest restoration mode, followed by AR and AM. In contrast, the pure alfalfa restoration mode has lower moisture content and porosity than AL, which shows that the alfalfa restoration mode is the worst. Of course, this is also related to the mechanical composition and texture of the soil. It can also be seen from the physical properties in Table 1 that the AR&AM&MS treatment has a higher content of silt and clay and a lower content of sand, and the soil texture is silty loam, which has a good effect on holding soil moisture. However, a similar phenomenon was also observed in the MS treatment, which may be related to the rhizosphere effect of the soil [30].

Additionally, the soil chemistry and nutrient properties under different forest restoration modes are shown in

Table 2. Chemical properties of soils in different restoration modes.

Treatments	pH		Total Salt (g kg−1)		Alkalinity (%)		Total Phosphorus (mg g−1)		Available Phosphorus (mg kg−1)		Total Potassium (mg g−1)		Available Potassium (mg kg−1)
	Average	Error	Average	Error	Average	Error	Average	Error	Average	Error	Average	Error	Average	Error
AR	8.35b	0.03	0.45bc	0.00	1.63ab	0.80	1.79a	0.34	16.34a	6.86	17.62b	0.25	136.00b	8.92
AM	8.36b	0.07	0.46b	0.03	0.63c	0.24	2.20a	0.61	14.82a	4.49	18.14a	0.23	155.20a	17.94
MS	8.43a	0.02	0.42c	0.01	2.47a	0.65	2.44a	0.22	10.81ab	8.95	18.14a	0.29	167.20a	8.17
AR&AM&MS	8.31b	0.06	0.50a	0.03	1.23bc	0.34	2.19a	0.86	10.17ab	13.10	18.02a	0.26	165.80a	7.95
AL	8.47a	0.06	0.45b	0.02	2.32a	1.04	1.75a	0.62	9.98b	8.32	17.64b	0.23	154.00a	14.20

Note: Letters represent significant differences between different soil layers in the same stand (p < 0.05).

and Figure 2. It can be seen from Table 2 that the pH of all soil samples is alkaline. Compared to CK, the pH of the soil in the AR, AM, and AR&AM&MS treatments decreased slightly, especially in the AR&AM&MS restoration mode. The total salt content of the AR&AM&MS treatment increased slightly compared to the other treatments, while the total salt content of the MS treatment decreased, indicating that different forest land soil restoration modes have a significant impact on soil chemical properties. A similar phenomenon was also observed for alkalinity. The AR, AM, and AR&AM&MS treatments all showed a significant decrease in content, while the MS treatment showed an increase in content compared to the AL treatment, but the difference was not significant. Moreover, forest soil is the substrate that maintains the healthy growth of trees, and its fertility characteristics affect and control the growth and health of plants and trees. Forest degradation is closely related to declining soil fertility [2]. Compared with AL, the total phosphorus, available phosphorus, total potassium, and available potassium contents of AM, MS, and AR&AM&MS treatments increased significantly (Table 2); however, the total potassium and available potassium contents of AR decreased, which may be due to interference from other factors, such as the sampling area and soil rhizosphere allelopathy [11,25].

Further, Figure 2a shows that the total carbon content was reduced in the MS treatment relative to AL, whereas the content was significantly increased in all three treatments, AR, AM, and AR&AM&MS, with significant differences (p < 0.05). For organic carbon (Figure 2b), all four restoration modes had reduced content relative to the abandoned land, but the difference was not significant (p < 0.05), implying that the vegetation in the four restoration modes utilized organic carbon. Figure 2c shows that the total nitrogen content of the AR, AM, and AR&AM&MS treatments was significantly higher than that of the AL treatment, and the difference was significant (p < 0.05); while the MS treatment was lower than the AL treatment, the difference was also significant (p < 0.05). This suggests that the restoration mode for tree treatments can significantly increase total nitrogen content, while shrub treatments may reduce total nitrogen content. Figure 2d shows that there is not much difference in the alkali-soluble nitrogen content of the AR, MS, and AL treatments. However, the alkali-soluble nitrogen content of the AM and AR&AM&MS treatments increased significantly, and the difference was significant (p < 0.05). This is consistent with the analysis in Table 1, and the restoration mode of AR&AM&MS and AM is significantly better than that of the other treatments. Moreover, as can be seen in Figure 2e for ammonium nitrogen, the differences between all treatments are not significant, and the content does not change greatly. This may be related to the metabolic action of soil microorganisms, and ammonium nitrogen is easily volatilized. The nitrate nitrogen data in Figure 2f show that, compared with AL, the nitrate nitrogen content of AR and AM treatments was significantly reduced, while the nitrate nitrogen content of MS and AR&AM&MS restoration modes was significantly increased, and the differences were significant (p < 0.05). This means that in the restoration mode of pure trees, the superimposed treatment of adding shrubs can comprehensively increase the cumulative content of nitrate nitrogen and enhance soil stability. Therefore, it can be inferred that the soil restoration of the AR&AM&MS restoration mode is relatively good, and its soil physical, chemical, and nutrient indicators are significantly better than those of abandoned land and other restoration modes.

3.2. Analysis of Differences in Soil Enzyme Activities

As can be seen in Figure 3a, there was no significant difference in catalase activity among the four restoration treatments. The catalase activity in the AR treatment was relatively low, which may be related to the enhanced trend of vegetation restoration succession and rhizosphere function (or litter). Compared with AL, the urease activity (Figure 3b) in the AM and AR&AM&MS treatments increased significantly (p < 0.05). However, the urease activity of the AR and MS treatments was significantly lower, and the difference was also significant (p < 0.05). This is consistent with the results of the physicochemical analysis of the soil, and the restoration effect of the mixed forest is better than that of the pure forest treatment, which increases the soil urease activity. A similar analysis can be seen in Figure 3c. Compared to AL, the phosphatase content of these four restoration modes increased, but the increase in the content of AR&AM&MS was greater, and the differences were all significant (p < 0.05). Furthermore, Figure 3d shows that the sucrase content decreased in the AR treatment relative to AL, but the difference was not significant. The sucrase content increased in the AM and MS treatments, but the difference was also not significant. Only the sucrase content in the AM and AR&AM&MS treatments increased with a significant difference (p < 0.05). It can be seen that the restoration mode of mixed forest with a combination of trees and shrubs is superior in the process of vegetation succession [15]. Figure 3e shows that the BG enzyme content of the MS treatment was lower than that of the AL treatment, and the difference was significant (p < 0.05). The AM treatment had an increased enzyme content, but the difference was not significant. Moreover, the AR and AR&AM&MS treatments lead to significantly increased enzyme content (p < 0.05). Figure 3f shows that the NAG enzyme content of AM treatment increased slightly relative to AL; AR, MS, and AR&AM&MS treatments also increased significantly, and the difference was significant (p < 0.05). Specifically, the NAG content of MS was relatively high, which was related to the physiological characteristics of alfalfa and the allelopathic effect of its root system. Therefore, it is clear from the analysis of differences in soil enzyme activity that the mixed forest restoration mode (AR&AM&MS) has more potential, and its various soil enzyme activities are relatively high and the restoration is good.

3.3. Analysis of Soil Microbial Diversity and Community Structure

The Venn diagram (Figure 4) shows that more than 3170 OTUs were detected in total in the soil samples of the five treatments in this study, of which the total number of OTUs was higher, reaching 3110. The Coverage index value of each sample was above 97%, and the sequencing results could accurately reflect the biological characteristics of the samples. In addition,

Table 3. Alpha Diversity Index.

Treatment	Observed OTUs 97%		Chao1 Index		Simpson Index		Shannon Index
	Average	Error	Average	Error	Average	Error	Average	Error
AR	3153	6	3114.84	9.96	0.96	0.0002	4.82	0.0051
AM	3159	20	3116.50	19.08	0.96	0.0001	4.82	0.0024
MS	3155	8	3119.60	13.89	0.96	0.0001	4.81	0.0031
AR&AM&MS	3150	12	3120.62	19.40	0.96	0.0002	4.82	0.0021
AL	3147	18	3108.87	18.04	0.96	0.0002	4.81	0.0038

shows the species diversity index of different forest restoration modes. Among them, the Shannon index and Simpson diversity index were basically similar under different restoration modes. However, the Chao1 index varied greatly, especially in the AR&AM&MS treatment.

Furthermore, the horizontal species distribution histograms under different ecological forest restoration modes, as shown in Figure 5, can visually show that the species composition and proportion of the five treatments are very similar. It can be seen from the relative abundance of species (Figure 5a) that the relative proportion of bacteria in all treatments in this study is above 99.8%. At the phylum level (Figure 5b), it can be seen that the dominant bacterial groups in soil microorganisms include Proteobacteria (about 38%), Acidobacteria (about 23%), Actinobacteria (about 11%), Bacteroidetes (about 6%), and Gemmatimonadetes (about 5%). Also, at the class level in Figure 5c, it can be seen that the five treatments in this study were most abundant in terms of Alphaproteobacteria (about 18%), Betaproteobacteria (about 8%), and Actinobacteria (about 5%). At the order level, as shown in Figure 5d, the five treatments in this study had the highest levels of Sphingomonadales (about 12%), Rhizobiales (about 8%), Burkholderiales (about 5%), and Xanthomonadales (about 4%).

Based on the species annotation and abundance information at the genus level for all soil samples, the genera with the top abundance rankings were selected in Figure 6. Using the abundance information for clustering, the results show that the microbial community structures under the four restoration mode species are significantly different. They are mainly divided into two categories, with AM and AR&AM&MS being more abundant. The main genus-level microorganisms include Solrubrobacter, Gaiella, Sphingosinicella, Bradyrhizobium, Lysobacter, Phenylobacterium, Sphingomonas, Steroidobacter, Streptomyces, and Variovorax.

Furthermore, Figure 7 analyzes the beta diversity of species under different forest restoration modes. Figure 7a uses the PCA model. The closer the two samples are to the PCA, the more similar the species composition of the two samples. As can be seen, the AL and MS treatments are close together, indicating that their compositions are similar, which is similar to the analysis results in Section 3.1. In addition, the AR and AR&AM&MS treatments have similar compositions. Figure 7b employs principal coordinates analysis (PCoA), a dimensionality reduction method similar to PCA. This means that AR and AM are most similar and that the species diversity processed by AR&AM&MS is richer. Figure 7c shows Non-Metric Multi-Dimensional Scaling (NMDS), which is a data analysis method that mainly simplifies research samples in multidimensional space to a low-dimensional space for positioning, analysis, and classification while retaining the original relationship between objects. It can also be confirmed that the species diversity of the AR&AM&MS treatment is the richest among all treatments.

3.4. Functional Diversity Analysis

3.4.1. Functional Gene Composition and Abundance

Figure 8 shows the composition and abundance of functional genes in different restoration modes. The kegg functional genes of the five differently treated soils are basically similar (Figure 8a), with an average of 7600 kegg Orthology genes and an average of 172 pathways. Additionally, genes with high abundance include those related to metabolic pathways, biosynthesis of secondary metabolites, and biosynthesis of antibiotics. Additionally, the eggNOG functional genes of the five different treated soil samples (Figure 8b) are basically similar, with an average of 54,500 annotated eggNOG orthologous gene clusters. Apart from those of unknown function, those with higher abundance are involved in amino acid transport and metabolism ([E]), general function prediction-only ([R]), and energy production and conversion ([C]). As can be seen in Figure 8c, the number of carbohydrate-active enzymes is 6 for the five different soil treatments, and the average number of annotated carbohydrate-active enzyme types is 317. Among them, GT4, GT2, and CE1 have the highest abundance, which can be classified as glycosyltransferases (GTs) and carbohydrate esterases (CEs). Finally, Figure 8d shows the functional genes related to antibiotic resistance information from the Comprehensive Antibiotic Research Database (CARD). The CARD contains information describing antibiotics and their targets, involving antibiotic resistance genes, related proteins, and antibiotic resistance mechanisms. It can be found that the types and proportions of the five treatments are similar in terms of the types and proportions of resistance genes, among which multidrug and Aminoglycoside resistance genes account for the largest proportion.

3.4.2. Functional Gene Beta Diversity Analysis

The beta diversity analysis of the functional genes of different forest restoration modes is shown in Figure 9. It can be seen from Figure 9a that the distances between the AR, AM, and AR&AM&MS treatments are very similar, which means that the composition of functional genes in these three restoration modes is also similar. Moreover, it can be seen from Figure 9b that the distance between the AR and AM treatments is even closer, which means that the composition of functional genes in these two restoration modes is also similar. Furthermore, the Beta diversity of the AR&AM&MS treatment is more dispersed, which means that the functional gene abundance is also relatively high. This is related to the synergistic restoration mechanism of forests, shrubs, and grass. This also matches the excellent vegetation restoration effect of AR&AM&MS (Section 3.1 and Section 3.2).

4. Discussion

4.1. Relationship Between Soil Microbial Community Composition and Functional Genes

In soil ecosystems, soil microorganisms are the dominant factor in the physical and chemical properties of the soil. Also, soil microorganisms depend on the substances formed by plant photosynthesis and respiration for growth and reproduction [2,16]. Additionally, different soil microorganisms lead to the presence of a variety of functional genes. The root exudates of plant communities also affect soil microorganisms, and different microbial groups, due to their different needs, also form competitive and interdependent relationships with each other [31,32]. Although this study found, in the above analysis, that the control treatment AL, AR, AM, MS, and the AR&AM&MS restoration modes have many similarities in terms of soil microbial diversity and functional genes, there are also differences. Figure 10a shows the heatmap of the abundance of species with differences at the genus level obtained by parameter testing. It can be seen that the heatmap distributions of AL and MS are similar, and the restoration modes of AR, AM, and AR&AM&MS are more similar. The heatmap distributions are more concentrated, which means that the species community structure and diversity of pure forest and the mixed forest restoration modes are similar, but there are also differences. The main difference is that the abundance of Phycicoccus, Gemmatimonas, Dactylosporangium, and Rubrobacter in the arboretum is significantly higher. Previous studies have shown that the phyla Proteobacteria and Actinobacteria are mainly involved in the decomposition of organic matter, while the phylum Chloroflexi is mainly involved in carbon and nitrogen fixation [33]. The relative abundance of Proteobacteria is positively correlated with soil carbon content and is higher in more nutrient-rich soil. Soil microbial diversity is positively correlated with plant community diversity. Woodland vegetation indirectly or directly affects the community structure and abundance of soil microorganisms by changing environmental factors such as soil nutrient content and pH [34]. More specifically, the AR&AM&MS restoration mode of the mixed forest had a higher abundance of the symbiotic bacterium Candidatus Entotheonella than the pure forest AR and AM, which may be related to the structure of the mixed forest. The advantage of the mixed forest AR&AM&M restoration mode is also evident from the differences in the functional genomes in Figure 10b. It is richer in functional genes, especially in the function of “metabolism of amino acid sugars and nucleotide sugars”. This suggests that the coupling of trees and shrubs in mixed forests maintains a relatively richer variety and number of microorganisms [24,33]. This is also related to the rhizosphere and allelopathic effects of woodland vegetation.

4.2. Relationships Between Physicochemical Properties, Enzyme Activities, and Soil Microbial Communities

The physical properties of the soil, such as soil moisture, porosity, and bulk density, and the chemical properties of the soil, such as organic matter, carbon, nitrogen, phosphorus, potassium, and other nutrient contents, are consistent with the succession process of forest land degradation. The higher the degree of forest land degradation, the lower the soil organic matter, C, N, P, K, and other nutrient contents [35]. Some studies have found that as forest degradation increases, the total soil carbon, organic carbon, and total nitrogen contents all show a downward trend, and the soil enzyme activity content also decreases relatively. In addition, like soil microbial biomass, soil enzyme activity is also an important biological indicator of soil quality, which can quickly indicate the trend of soil quality changes [23]. With the restoration of vegetation, the total carbon, organic carbon, total nitrogen, total phosphorus, and total potassium content of the soil increased significantly. Soil productivity gradually increased after the succession began, and gradually decreased in the later stage of succession. In addition, the impact of forest plant diversity indirectly affected soil enzyme activity by affecting soil microbial biomass [6]. The combined consideration of soil microbial biomass and enzyme activity in this study is the combined effect of vegetation screenshots and soil properties [35]. The physicochemical properties, enzyme activity, and soil microbial community diversity of the restoration mode of mixed forests were generally higher than those of other pure forest treatments, especially in terms of nutrients. This indicates that the restoration capacity of forest land varies among different modes, but the restoration effect of the mixed forest of mountain apricot–mountain peach–alfalfa is relatively good, and the soil fertility is relatively high [15]. Soil microorganisms under different forest restoration modes all showed relatively high abundances of the phyla Proteobacteria, Acidobacteria, Actinobacteria, Bacteroidetes, and Gemmatimonadetes [36]. The results of the species relative abundance cluster heatmap showed that the dominant genera in the soil were different under different management modes, which was speculated to be due to the influence of environmental factors [15,37]. The differences in soil microorganisms may be the cause of the differences in soil fertility, which means that the differences in the soil microbial community structure under different restoration modes result in differences in the restoration capacity of the forest land. Further in-depth discovery of the AR&AM&MS mixed forest restoration mode of enzyme activity and soil fertility-related total nitrogen, organic carbon content, etc. all have a high correlation. Different forest restoration modes result in different soil microbial community structures in the loess hilly–gully region, and different microbial community structures have different effects on soil fertility and forest restoration capacity.

4.3. Mechanisms for Ecological Forest Restoration Modes

After the artificial restoration of vegetation in the hilly areas of the Loess Plateau, the content of nutrients, enzymes, and microorganisms in the soil has changed, and the soil under artificial vegetation restoration is more active. On the one hand, the slope of the sparse tree shrubland is north facing, creating a shady slope, while the vegetation restoration of artificial planting is mostly on the southwest slope [2,30]. The sunny slope receives more solar radiation than the shady slope and the temperature is higher than the shady slope. The restored vegetation forms an arboreal layer and, to a certain extent, a litter layer, and the litter layer may affect the distribution of nutrients in the forest soil. Therefore, the soil nutrients in the artificially restored forest soil are more accessible, and the material cycle between vegetation and soil is faster, so the nutrient content in the soil is relatively low. The soil and soil microorganisms in the sparse tree shrub community are not very efficient at utilizing substances such as organic matter and total nitrogen after they have been converted, so most nutrients are stored in the soil [12]. On the other hand, the open grassland vegetation community is mainly composed of herbaceous plants and a very small number of shrubs, while in the artificially restored vegetation community, most of the nutrients in the soil are transferred to the tall trees, and the nutrients in the soil itself are relatively reduced [6]. At the same time, the nutrient content of the soil in different communities was higher in the root system than in the non-root system, indicating that the root exudates and fallen leaves produced by plants have a greater impact on the soil around the root system than on the soil far away from the root system. The direct and indirect effects of soil organic matter on plants affect the supply of nitrogen, phosphorus, sulfur, and other nutrient ions in the soil to plants, and then affect the activities of soil microorganisms and animals, which in turn affect the physical properties of the soil [7]. The differences in soil nitrogen, protease, and urease among the different ecological forest restoration modes in this study were consistent. Acid phosphatase can enzymatically hydrolyze the phosphoric ester bond of soil organic phosphorus compounds, thereby improving the bioavailability of soil phosphorus [2,24]. The effective phosphorus content in the restoration mode community of the mixed forest of mountain apricot–mountain peach–alfalfa is lower than that in the open grassland, but no direct relationship between the effective phosphorus content and acid phosphatase in the soil has been found. A possible reason is that the amount of acid phosphatase produced by vegetation in different communities is similar, and the utilization efficiency of phosphorus by artificially restored vegetation is higher, resulting in a decrease in the effective phosphorus content in the soil.

Furthermore, vegetation restoration is the main method for ecological restoration in the Loess Plateau region. The vegetation environment can change soil erosion problems. Different forest types and forest restoration modes will affect soil structure and nutrient conditions, producing different ecological benefits [7]. This study found that the soil moisture content in farmland conversion forest plots with different forest types generally showed a pattern of mixed forest being greater than pure forest. At the same time, the soil bulk density in the mixed forest of sea buckthorn and mountain apricots with different years of farmland conversion was the lowest. Overall, the soil organic matter showed a pattern of mixed forest being greater than pure forest and then abandoned grassland. Among the ecological forests with the same years of farmland abandonment, the mean soil total nitrogen value in the AR&AM&MS mixed forest restoration mode is the highest, and as the farmland abandonment time increases, the soil total nitrogen content in the same forest stands increases. A complex woodland structure is more conducive to improving soil structure and nutrients. Changes in the physical properties of the soil are a long-term process, and changes in soil nutrients are affected by the functional groups of organisms (plants, animals, and microorganisms) living in the soil [24,25]. Among the five restoration modes selected in this study, the overall effect of mixed forest restoration is stronger than that of pure forest restoration. At the same time, both mountain peach and mountain apricot are economic forest trees, which can also contribute to local economic development. The structure of soil microbial communities responds to soil nutrients and changes when soil nutrients change. In forest restoration, different types of trees are selected for planting according to different site conditions and natural environments, which leads to differences in the restoration of the soil’s physical structure and chemical composition, and thus differences in the structure and composition of soil microbial communities. Soil microorganisms are key drivers of elemental biogeochemical cycling processes and other soil ecological processes [37]. This study found both similarities and differences in the composition and structure of soil microbial communities across restoration modes, as well as differences in the species richness and diversity of soil bacterial and fungal communities. Therefore, maintaining forest land and newly restored forest land will help to stabilize the microbial community structure in the soil ecosystem while enhancing the function of the soil microbial community, allowing different microbial groups to perform their functions, enhancing the ecosystem’s resistance to interference, and increasing the internal stability of the ecosystem, thereby restoring the ecological environment of the Loess Plateau.

After comparison with restoration modes in other areas and in-depth discussion [4,16,24,38], it was found that this is closely related to the effectiveness of forest restoration with climatic conditions, geographic location, environmental factors, vegetation types, and management practices. The mixed forests of apricot–peach–alfalfa (AR&AM&MS) recommended in this study are more suitable for application in the hilly areas of the Loess Plateau. These research areas mainly include southeast China, the Qinghai-Tibet Plateau, the subtropical region of China, other regions of the world, and the Loess Plateau hilly region in this study [4,16,24,38,39]. In summary, this mixed forest restoration mode can improve soil structure, retain soil moisture, enhance soil enzyme activity, optimize soil microbial community structure, and increase microbial diversity and functional gene activity in the hilly areas of the Loess Plateau.

5. Conclusions and Outlook

This study investigated the differences in the physicochemical properties, enzyme activities, microbial diversity, community structure, and functional gene diversity of soils in different ecological forest land restoration modes in the Loess Hilly Region. The focus was on discussing the impact and restoration mechanism of different restoration modes. Compared with abandoned land treatment, the soil water content and porosity of the three restoration modes of AR, AM, and AR&AM&MS increased significantly, while the bulk density decreased significantly. Further, the soil water content and porosity of the MS treatment decreased, but not significantly, and the bulk density increased, but also not significantly. The mechanical composition of the AR&AM&MS treatment had higher contents of silt and clay. Compared with the control treatment, the pH alkalinity and organic carbon content of the AR&AM&MS restoration mode were significantly reduced, the total salt content increased slightly, and the total carbon, total nitrogen, alkali-soluble nitrogen, ammonium nitrogen, nitrate nitrogen, total phosphorus, available phosphorus, total potassium, and available potassium contents all increased significantly. The results show that the restoration mode of mixed forest is better than that of pure forest grass. Compared with the abandoned land treatment, there is no significant change in the catalase content of the pure forest and mixed forest, but the urease, phosphatase, sucrase, BG enzyme, and NAG enzyme content in the mixed forest restoration mode all increased significantly, followed by the restoration mode of the pure forest of mountain peach. The species diversity index of different ecological forest restoration modes is similar, and the difference is not significant. The dominant bacteria in soil microorganisms include the phyla Proteobacteria, Acidobacteria, Actinobacteria, Bacteroidetes, and Gemmatimonadetes. Correlation analysis showed that the mixed forest restoration mode had the highest microbial abundance, mainly including the genera Rhodobacter, Actinomyces, and Sphingomonas; its composition was closest to that of the pure forest AR. The functional gene diversity of different ecological forest restoration modes was also similar, but there were differences in the direct functional genes, and the functional genes of the AR&AM&MS restoration mode were the most abundant. Finally, the relationships and differences between the physicochemical properties of soil, enzyme activity, microbial community diversity, and functional genes were discussed in depth, with a focus on revealing the restoration mechanism of the mixed forest mode. Therefore, promoting forest land restoration, improving species composition, and enhancing soil microbiology activity through appropriate management and management measures is an important way to restore and maintain soil productivity. The results of this study provide useful insights into the restoration of the ecosystem on the Loess Plateau.

Based on the results of this work, it is further seen that the mixed forest restoration mode can improve soil structure, enhance soil physicochemical properties, retain moisture, optimize soil microbial community structure, and increase soil enzyme activity and microbial diversity. AR&AM&MS is especially suitable for application in the hilly areas of the Loess Plateau. It is hoped that these studies will arouse the interest of governmental decision-makers. While the restoration mode recommended in this paper may also be applied in forest restoration projects in ecologically fragile areas, this needs to be evaluated carefully. This requires enhanced and continued in-depth exploration of the importance of sustainable ecosystem management under this restoration mode. Additionally, the research on mixed forests improving the ecological environment, promoting ecological balance, and enhancing the ecological service functions of forests, including water conservation, soil retention, and air purification, is one future exploration direction. Moreover, the differences between mixed forests of different vegetation types in protecting microbial diversity, and the exploration of mixed forests in improving land utilization and enhancing disaster resistance will also be the focus of future research.