Introduction

Due to the impacts of natural disasters and human interference, many native ecosystems have been degraded, frequently leading to problems such as soil erosion and land desertification1,2. The synergistic effect of vegetation and soil is a hot topic in terrestrial ecosystem research3,4. Differences in soil properties under different vegetation restoration methods in turn affect vegetation5,6. In the process of vegetation ecological restoration, these restoration measures can promote positive succession of ecosystems by improving soil quality and increasing plant species diversity7,8,9. According to the degree of restoration, vegetation restoration methods can be classified as directional natural renewal, artificial promotion of natural renewal, partial reconstruction, or complete reconstruction10. Natural renewal emphasizes the use of natural forces and processes to restore ecosystems, focusing on maintaining the natural characteristics and biodiversity of ecosystems, while reconstruction emphasizes achieving rapid restoration goals through human intervention and control, which may introduce alien species or change the natural characteristics of ecosystems. Research shows that the resilience of forests has generally declined11. Therefore, studying the degree of improvement in soil quality under different vegetation restoration methods according to different restoration goals, environmental statuses, and costs is beneficial for optimizing vegetation restoration methods12,13.

Ecological restoration is a lengthy process, and changes in soil nutrients are an important part of characterizing whether long-term ecological restoration has achieved its intended effects. During the process of vegetation restoration, soil properties and vegetation characteristics exhibit a clear coevolutionary relationship13. A reduction in vegetation coverage caused by natural disasters and human activities in native ecosystems can have profound and lasting impacts on soil properties and functions, ultimately extending to deeper soils14. Vegetation restoration can have a significant impact on soil nutrient concentration and storage15, and different restoration methods have different impacts on soil16,17. Forest restoration contributes more to soil carbon sequestration than do farmlands and bare lands, but this contribution is generally smaller than or similar to that of grasslands and natural forests18. Soil nutrient conditions significantly improve along the vegetation restoration gradient13. The species diversity of vegetation communities is closely related to surface soil, and soil nutrients exhibit obvious surface aggregation17,19,20,21. In addition, the restoration of vegetation to soil is also mainly concentrated on the surface14. Analyzing the storage and stoichiometric characteristics of soil nutrients can reveal the impact of different vegetation restoration methods on the main elements of soil22. The soil quality index (SQI) integrates multiple indicators, making the evaluation of soil quality more comprehensive and reasonable23. The recovery of soil quality may take more than 30 years24, and the final soil properties obtained still deviate from those of the soil under natural forest conditions14. Therefore, accurately controlling the changes in soil nutrient content and quality during the long-term ecological system restoration process, formulating scientifically reasonable restoration and management measures, and transforming vegetation and soil conditions into near-natural conditions are conducive to promoting ecosystem restoration and forming a vicious cycle.

As the world's largest planted forest management area, the Saihanba Mechanized Forest Farm successively received the United Nations' Champions of the Earth Award and Land for Life Award in 2017 and 2021, serving as a paradigmatic success story in the restoration of degraded forest ecosystems. Currently, evaluations of afforestation achievements in Saihanba have focused primarily on forest quality, with limited research assessing the effectiveness of restoring degraded forest ecosystems from the perspective of soil characteristics. This study aimed to explore the impact of vegetation restoration on soil chemical properties and to fill the gaps in existing research. Our specific objectives include: (1) assessing the differences in soil quality evolution across various vegetation restoration methods in the forest–grassland–desert transition zone; and (2) comparing the relative merits of two types of artificial forests composed of native and exotic tree species in improving the soil chemical properties of degraded forest ecosystems. Our main hypothesis is that the use of native tree species for afforestation will more effectively enhance soil quality than the use of exotic species. Furthermore, we predict that soil quality indicators will exhibit specific patterns based on different methods of vegetation restoration. To test these hypotheses, this study will follow this framework: first, establish fixed monitoring points in the Saihanba Mechanized Forest Farm and the Ulan Buh Grassland; second, analyze the differences in soil quality evolution under the same climatic and soil historical conditions; and finally, directly address the research questions through statistical analysis and test whether the results conform to expectations. We hope that the findings of this study will provide valuable information and guidance for ecological restoration efforts under similar land conditions, such as the recovery and management of degraded forest ecosystems.

Materials and methods

Description of the study area

The study area is located at the junction of the Saihanba Mechanized Forest Farm and the Ulan Buh Autonomous Region Nature Reserve (Fig. 1). This area is situated on the southern edge of the Otindag Sandy Land and represents a transitional zone between forest–grassland and arid to semiarid regions. The elevation ranges from 1010 to 1939.9 m above sea level, with the prevailing climatic conditions expressing an average annual temperature of − 1.3 °C, an average of 7 months of snow cover, an average frost-free period of 64 days, and an average annual precipitation of 479 mm. The soil types affected by topography and climate mainly include brown soil, gray forest soil, black soil, sandy soil, meadow soil and swamp soil. The Saihanba Mechanized Forest Farm which was established in 1962, has experienced gradual restoration of its degraded forest ecosystem through years of afforestation and forest management, in stark contrast to its early state (Fig. 2). The region predominantly features two types of vegetation: forests and grasslands (Table 1). The grasslands are functionally divided into grazing lands and protected areas, while the forests are primarily composed of artificial plantations of Larix principis-rupprechtii, Picea asperata, and Pinus sylvestris var. mongolica, as well as some natural secondary forests such as Sorbus pohuashanensis, Quercus mongolica, and Betula platyphylla.

Fig. 1
figure 1

Location of the study area (above) and distribution diagram of the sample points (below). The maps were created in ArcGIS 10.8.1.

Full size image
Fig. 2
figure 2

Vegetation Restoration of Saihanba Mechanized Forest Farm. The black-and-white photo “a” was taken in 1962, while the color photo “b” was taken in 2022.

Full size image
Table 1 Classification of vegetation restoration methods.
Full size table

Sample points and collection

In late August 2022, soil samples were collected along the banks of the Hebei-Inner Mongolia border river, with a basic grid unit of 2 km × 2 km, and 63 grid intersection points were selected as sampling sites (Fig. 1), categorized by vegetation restoration method (Table 1). At each sampling point, soil samples from the 0 to 20 cm soil layer were collected, air-dried, and sieved before analyzing seven soil nutrient indicators. The total nitrogen (TN, g kg−1) content was determined using the concentrated sulfuric acid-perchloric acid digestion-Kjeldahl nitrogen determination method. The total phosphorus (TP, g kg−1) content was measured using the sodium hydroxide fusion-molybdenum antimony anti-colorimetric method. The total potassium (TK, g kg−1) content was assessed through the nitric acid, perchloric acid digestion-hydrofluoric acid decomposition-flame photometric method. The rapidly available phosphorus (AP, mg kg−1) content was estimated via the ammonium fluoride-hydrochloric acid solution extraction-molybdenum antimony anti-colorimetric method. The rapidly available potassium (AK, mg kg−1) content was determined using the neutral 1 mol L−1 ammonium acetate solution extraction-flame photometer method. The rapidly available nitrogen (AN, mg kg−1) content was quantified by the 1.8 mol L−1 sodium hydroxide melting-diffusion titration method. The soil organic carbon (SOC, g kg−1) content was measured using the potassium dichromate oxidation-spectrophotometric method.

Study methods

Finally, the sum of the normalized scalar of each indicator and the weighted sum were calculated to obtain the SQI23. The normalization method is as follows:

$$ \begin{array}{*{20}c} {x_{i} = \frac{{\left( {x - x_{min} } \right)}}{{\left( {x - x_{max} } \right)}}} \\ \end{array} $$
(1)

In formula (1), xi is the normalized pure quantity, and x is the original index.

The soil quality index was calculated as follows:

$$ \begin{array}{*{20}c} {SQI = \mathop \sum \limits_{i - 1}^{n} W_{i} \times x_{i} } \\ \end{array} $$
(2)

In formula (2), Wi is the weight of the index, xi is the normalized purity of the index, and n is the number of indicators to be calculated.

Based on the computational results, the SQI for each sampling point was categorized into three tiers according to magnitude. Points in the top 33.3% were designated “Excellent”, those in the middle 33.3% were classified as “Moderate”, and the remaining 33.3% were labeled “Poor”. The proportion of each vegetation restoration method across these tiers was compared, and the differences in the SQI among the various vegetation restoration methods were analyzed.

Data processing

Outliers were systematically identified and removed for each measurement indicator type using the R programming language, refining the data for further analysis. For each indicator under Level 1, independent sample t-tests and Mann‒Whitney U tests were conducted based on normality to determine the significance of nutrient differences between different categories. For Level 2 data, followed by one-way analysis of variance and multiple comparisons using the Scheffe method to test the significance of differences between each vegetation restoration. Seven soil nutrient indicators were taken as the full data set indicators. Principal component analysis (PCA) was used to select several principal components with eigenvalues greater than 1, determining the weight of each indicator in the principal components. The nutrient indicators were normalized through the membership function.

Results and analysis

Soil nutrient indices

According to the comparison of soil nutrients under the primary classification of vegetation restoration methods (Table 1), the forest TN, AN, and SOC contents were significantly greater than those in the grassland by 34%, 37%, and 52%, respectively (Table 2). However, the forest TK content was 6% lower than that of the grassland. This indicates that forests have more advantages in promoting soil N and C accumulation, while grasslands have a competitive advantage in improving soil TK content. Although the difference between forests and grasslands is significant according to the T-test, it is hypothesized that in the long term, both ecosystems could exhibit comparable levels of TK.

Table 2 The difference in nutrient indices between forests and grasslands. F indicates forest ,G indicates grassland.
Full size table

Figure 3 displays the results of multiple comparisons, indicating that the GO and FP treatments consistently showed significantly lower levels of TN, TP, AN, and SOC compared to the FL, FN, and GC treatments. The FN treatment had the highest average AN content (205.94 mg kg−1), which was 76% greater than that of both the GO and FP treatments. The average contents of TK, AK, and SOC in FN were greater than those in FL, FP, and GO by 4%, 86%, and 89%, respectively. However, the AP content in the FN was the lowest (37.63 mg kg−1), indicating that natural forests perform well in terms of soil N, K, and C retention but poorly in terms of soil P retention. The average contents of soil TN, TP, and SOC in the FL treatment were 85%, 56%, and 93% greater than those in the FP treatment, respectively. Although the average TK content was the lowest (16.29 g kg−1), there was no significant difference from that of the other restoration types, indicating that FL plays an important role in the retention of soil C, N, P, and K. The average contents of soil TN, AN, TP, and AK in the FP were the lowest among all the restoration types. Additionally, the average SOC content in the FP treatment was close to that measured in the GO treatment. In contrast, the average AP content in the FP treatment was the highest at 39.21 mg kg−1, but there was no significant difference compared to that in the other vegetation restoration methods. The soil TN, AN, TP, AK, and SOC contents in the GC were significantly greater than those measured in the GO.

Fig. 3
figure 3

Soil nutrient content under different vegetation restoration methods. FL, FN, FP, GO and GC refer to L. principis-rupprechtii artificial forests, natural forests, P. sylvestris var.mongolica artificial forests, open grasslands and enclosed grasslands, respectively. Different lowercase letters in the figure indicate that there are significant differences between different vegetation restoration methods (P ≤ 0.05).

Full size image

Ecological stoichiometry of the soil

The results of the Mann‒Whitney U test indicated significant differences in the ecological stoichiometric characteristics between forests and grasslands (Table 3). Specifically, the C:N, C:P, and N:P ratios for forests were greater than those for grasslands. The median values of C:P and C:N between forests and grasslands were significantly different, while there was no significant difference in the N:P between forests and grasslands. The median C:N of the forest exceeded that of the grassland by 11%, and the median C:P was 20% greater than that of the grassland. Among the different methods of vegetation restoration, there were significant differences in the C:P and N:P ratios, while the C:N ratio did not significantly differ among the different types of habitat (Table 4). Multiple comparisons revealed that in terms of C:P, this ratio in the FL was at least 26% greater than that in the FP and GO; in terms of N:P, it was 19% lower in the FP than in the FL, and 15% lower in the GO than in the GC.

Table 3 Median values of ecological stoichiometric characteristics between forests and grasslands (medians (lower quartiles, upper quartiles)).
Full size table
Table 4 Mean values of ecological stoichiometric characteristics under different vegetation restoration methods (means ± SDs).
Full size table

Soil quality index

Two principal components were selected according to the criterion of an eigenvalue greater than 1, and the cumulative variance interpretation rate was 81.4% (Table 5), indicating that more than 80% of the measured information can be explained by the selected principal components. The SQI was calculated by the linear combination coefficient of each principal component. The results showed that the SQIs of FL, FN and GC were significantly greater than those of FP and GO (Fig. 4), and the proportions of the three in the excellent category were significantly greater than those of the in GO category (Fig. 5). The proportion of the SQI in all the FP samples was 0 in the excellent category. In contrast, the largest proportion was identified in the GO group, followed by the FP group.

Table 5 Principal component analysis (PCA) results for the soil nutrient indices.
Full size table
Fig. 4
figure 4

Soil quality indices of different vegetation restoration methods. Different lowercase letters in the figure indicate that there are significant differences between different vegetation restoration methods (P ≤ 0.05). FL, FN, FP, GO and GC refer to L. principis-rupprechtii artificial forests, natural forests, P. sylvestris var.mongolica artificial forests, open grasslands and enclosed grasslands.

Full size image
Fig. 5
figure 5

The proportions of different vegetation restoration methods according to the soil quality index classification. Note: Excellent, moderate and poor indicate that the SQI is divided into three levels according to size. FL, FN, FP, GO and GC refer to L. principis-rupprechtii artificial forests, natural forests, P. sylvestris var. mongolica artificial forests, open grasslands and enclosed grasslands, respectively.

Full size image

Discussion

The storage and circulation of nutrients in terrestrial ecosystems are influenced by various factors, among which vegetation plays an important role15,25. The present study showed that the organic carbon content of forest soil was significantly greater than that of grassland (Table 2), indicating that the SOC content is influenced by the amount of vegetation returned and that the amount of forest returned was much greater than that of grassland. However, Jiao et al.26 reported that when the soil depth is greater than 5 cm, the SOC content in forests is significantly lower than that in grasslands, indicating that ecological restoration is a long-term process. In the short term, grasslands are more likely to improve the soil SOC content than forests are, while long-term vegetation succession can enhance the stability of the SOC pool and promote carbon sequestration27. The Saihanba afforestation project has a history of more than 60 years. Periodic pruning and thinning accelerate the return of forest carbon while ensuring the sustainability of vegetation restoration28. In addition, the TN and AN contents exhibited significant differences under the primary classification, and the distributions of N-type nutrients were relatively consistent among the vegetation types, indicating that the forests in the study area exhibited better performance in terms of improving soil N-type nutrients29. However, there was no significant difference in the TP, AP, or AK contents between forests and grasslands, indicating that the different vegetation restoration methods had no significant impact on the soil P content or AK content.

The difference in soil nutrients among the different methods of vegetation restoration was mainly due to the low nutrient content in the FP and GO (Fig. 3). The contents of AK, TN, TP, AN, and SOC in the GC restoration type were significantly greater than those in the FP and GO restoration types, indicating that enclosure management plays an important role in maintaining grassland soil nutrients. Open grasslands, which are subjected to severe human disturbances such as grazing and mowing, can increase the consumption of soil nutrients30,31. In this study, the standard deviations for the soil TN, AN, and SOC contents in the FN were the largest (Fig. 3a,b,g), indicating that the fluctuations in these three nutrients were greater among the different sampling points in the FN. The standard deviations of the TN, AN, and SOC contents in the FP and GO treatments were lower than those in the other restoration treatments , and the spatial distribution was more uniform. This may be because there is only one tree species in artificial pure forests, and the trends of the soil nutrient changes are similar. However, the tree species composition in natural forests is complex, and the trend of soil nutrient changes varies among different sampling points affected by different tree species mixtures32. The soil TN, TP, and SOC contents in the FL were the highest (Fig. 3), and there was no significant difference from those in the FN, indicating that the FL, which is composed of one of the main economic tree species locally, is more adaptable to the local environment. Measures such as tending and thinning during the growth process of trees have made important contributions to soil nutrient recovery. The FP, composed of an exotic tree species in the study area, is mainly planted for wind prevention and sand fixation, so it rarely operates, the stand density is high, and Scots pine is an evergreen tree. There is far less litter than in North China larch, leading to a decrease in the return speed of soil nutrients as well as consumption33. The FP, composed of a pioneer tree species, has greater adaptability to harsh environments. In the later stage of ecological restoration, native tree species can be considered for use in combination with afforestation under the canopy and natural renewal measures to regulate tree species structure. In the present study, the SOC content in each type of soil was greater than that in the northern edge of the continuous distribution area of karst in the southern region and in the Geqiugou River Basin in Inner Mongolia34,35, indicating that the long-term vegetation restoration effect was considerable; however, compared with that in the mixed coniferous and broad-leaved forests in Changbai Mountain, the soil nutrient level in the study area still exhibited high recovery potential36. The SOC content in the FP treatment (24.55 g kg−1) was close to the national average (24.56 g kg−1), while the SOC content in the GO treatment (23.6 g kg−1) was lower than the national average37. The TP content in both (the 0.25 and 0.27 g kg−1) treatments was lower than the TP content in the soil in karst areas (0.36 g kg−1)38, while the SOC and TP contents in the FL, FN, and GC treatments were all greater than the abovementioned levels, further indicating that the nutrient cycling modes of natural restoration and the reasonable operation of artificial restoration are more stable. Therefore, choosing appropriate management and operation measures is the key to maintaining soil nutrients for different methods of vegetation restoration39,40.

The soil ecological stoichiometric characteristics reveal the main nutrient elements that limit plant growth. Different methods of vegetation restoration and soil properties can affect the C, N, and P ecological stoichiometric characteristics of the soil41. According to the results, the C:N ratio of the forest restoration types was significantly greater than that of the other grassland types (Table 3). The C:N value of the surface soil of the GO was 14.63, which is slightly greater than the average C:N value of 14.4 in the organic matter enrichment layer of terrestrial soil in China37. The minimum C:N values of the other four types were all greater than 16 (Table 4). Taken together, these findings indicate that vegetation restoration notably contributes to carbon sinks, and the accumulation rate of SOC in the surface soil is greater than that in the soil C source, the organic matter decomposition rate, and the mineralization rate42. The C:P ratio of the soil indicates the effectiveness of P in the soil43 and is closely related to vegetation restoration. In this study, the average C:P value of the surface soil under different methods of vegetation restoration was lower than the average C:P value of the organic matter enrichment layer in terrestrial soil in China37, indicating that the effectiveness of soil P in the study area is high, the conversion rate of AP is fast44, and soil microorganisms are not restricted by P in the process of decomposing organic matter, which is consistent with the finding that there is no significant difference in soil AP content between different methods of vegetation restoration (Fig. 3). In contrast, the C:P ratios of FL and GC were significantly greater than those of FP and GO, indicating that P is a sensitive factor affecting the restoration of FL and GC. It should be monitored in real time whether P is deficient, and P should be supplemented in a timely manner34. The soil N:P ratio is an important indicator of whether N meets the needs of an ecosystem. The N:P value of the surface soil of forests and grasslands in the study area is lower than the average value in China37, indicating that there is a certain degree of N limitation in the study area45. Similarly, the N:P ratio under FL was significantly greater than that under FP, indicating that FL is more efficient at transforming and absorbing N than is FP. Considering the current background of increasing atmospheric N deposition, taking North China larch and other native tree species as the main cultivation targets in the study area may alleviate the current situation of N deficiency to a certain extent, but it is still necessary to monitor the soil for a long time and formulate precise fertilization plans according to the degree of deficiency and implement them.

The use of the soil quality index (SQI) is a method for objectively analysing and quantifying soil indicators. These models can be constructed according to different research objects. The greater the SQI is, the greater the soil fertility23,46, and the better the soil recovery. With respect to the choice of different restoration methods and management measures, the SQI exhibited different results. In this study, the SQIs of FL, FN, and GC were significantly greater than those of FP and GO (P < 0.05) (Fig. 4), and FL, FN, and GC together accounted for more than 90% of the top 33.3% of the sample points according to the SQI grade. The SQIs of FN and GC were greater than those of the other vegetation types, indicating that the soil recovery effect is similar under natural restoration conditions and that natural restoration has a long-term and positive impact on soil quality47. In the study area, with the FL (composed of a native tree species) being the main species under reasonable management, the soil quality can improve naturally, while the soil quality in response to the FP (composed of an exotic tree species) is poor. This shows that using native tree species for vegetation restoration has a more positive impact on soil nutrient restoration. Evidently, the study site, which is the provenance of the studied native tree species provides an advantage for enhancing the natural positive effects on improving soil quality. Roscher et al.48 and Mcguire et al.49 also noted in their study on the nutrient utilization characteristics of native tree species that the ecosystem element cycle of native tree species is more balanced. In addition, this study revealed that the soil quality of the GC treatment was significantly greater than that of the GO treatment (Fig. 4), which is consistent with the conclusion that enclosure time is positively correlated with the degree of soil improvement24,50.

Conclusions

The soil in the study area has a certain N limitation. This study revealed significant differences in soil quality among the different vegetation restoration methods. Forests have a much greater efficiency in retaining soil C and N than grasslands. Among these plantations, the L. principis-rupprechtii artificial forest (FL) has the greatest effect on maintaining or restoring soil N. Using native tree species for the restoration of degraded forest ecosystems can promote soil improvement and achieve natural restoration effects. Although vegetation restoration through exotic species has achieved certain results in terms of wind prevention, and sand fixation, including soil and water conservation, the effect on improving soil nutrients is minimal. It is advisable to consider the use of native tree species for afforestation under the canopy, in combination with long-term soil nutrient monitoring plans or campaigns, to gradually adjust the tree species structure and formulate precise fertilization plans.