The silt- and sand-amended peatland sites were located in Changbai Mountain, northeast China (Figure 1). The region has a temperate continental monsoonal climate, and the frost-free period was approximately 138 days during each year. The annual mean precipitation is 737.4 mm, and the annual mean temperature is approximately 5°C (Wang et al., 2021). The silt-amended peatland site is located in Dayishan Town, Huinan County, Tonghua City, Jilin Province (126°13′10″E, 42°16′35″N, 456 m a.s.l). The peat thickness was 0.5–1 m. C. schmidtii was the dominant species before farming. In the 1980s, most of the peatlands in these regions were drained and filled with silt soil to increase the load-bearing capacity of the soil and balance pH for agricultural purposes. As part of the soil amendment process, approximately 10 cm of silt was covered on the surface of the peatlands. The sand-amended peatland was located in Sipeng town, Tonghua City, Jilin Province (125°34′44″E, 41°51′22″N, 560 m a.s.l). The peat thickness in this study site ranges from 0.6 m to 1.0 m. Peatlands in these regions were drained and intensively reclaimed to paddy fields in the 2000s, with the introduction of sand in the plowed horizon and mixing it with peat while plowing. Peatland restoration was conducted in April 2019. At each peatland site, two types of experimental treatments were performed close to each other: topsoil removal (removing 10 cm topsoil layer from the area of 0.5 h m²) and control (no action on the topsoil on the area of 0.5 h m²). The water level was kept at 0–10 cm below the soil surface in all the treatments.

Field vegetation survey was performed in July 2021. Four replicate plots were randomly selected at each site. Within each plot (≈.1 h m²), five quadrats (1 m × 1 m) were randomly sampled, which were more than 10 m away from each other. We investigated plant community characteristics including total coverage, aboveground biomass, and the height, density, and percentage cover of each species. In addition, we recorded characteristics of dominant Carex species including height, basal diameter, root length, tiller number, and above- and below-ground biomass. Vegetation was harvested using a sickle and dried at 65°C, and the dried biomass was weighed.

Soil samples were collected from both treatment plots at each site. Four soil cores (depth 30 cm, interval 10 cm, and diameter 5 cm) were collected from each plot to measure the soil bulk density (BD), soil water content (SWC), soil organic carbon (SOC), soil total phosphorus (TP), soil total nitrogen (TN), and soil pH. BD was determined using the volumetric core method. SWC was determined using the gravimetric method (Jackson et al., 2000). SOC was determined using the dichromate oxidation method (Kalembasa and Jenkinson, 1973). TP and TN were measured using the Kjeldahl and molybdenum blue methods, respectively. Soil pH was determined using a glass electrode with a soil-to-water ratio of 1:10.

Species richness, the Shannon–Wiener Index, Pielou Index, Margalef Index, and Simpson Index were evaluated to describe plant diversity in each plot.

Species richness: R = S

Shannon–Wiener index: H = − ∑ i = 1 s P i   ln ⁡ P i

Margalef Index: F = ( S − 1 ) / ln N

Pielou Index: J = H / l n S

Simpson Index: D = 1 − ∑ i = 1 s P i 2

Where, S is the species number within the quadrat, N is the total number of individuals, and Pi is the importance value of the ith species; Pi is (the relative height of the ith species + relative cover of the ith species + relative abundance of the ith species)/3.

One-way ANOVA, followed by multiple comparisons using the LSD test, was performed to study the differences in soil properties and plant community characteristics between the two treatments at the same site and between the silt- and sand-amended sites for the same treatment. The significance level of p < .05 was considered statistically significant. Redundancy analysis (RDA) was conducted to study the correlation between vegetation composition and soil environmental factors using Canoco 5.0 (Microcomputer Power, Ithaca, NY, United States).

Soil environmental factors responded differently to topsoil removal between the two peatland sites. At the silt-amended peatland site, SOC, TP, TN, and SWC in the topsoil-removed plot were higher than those in the control plot (p < .05). No significant difference was observed in terms of BD, pH, and C: N between the two treatments.

Plant community coverage and aboveground biomass in the topsoil-removed plots were higher than those in the control plots at the silt-amended peatland site (p < .05; Figure 2A). No significant difference was observed in terms of plant community coverage and aboveground biomass between the two treatments at the sand-amended peatland site (Figure 2B).

At the silt- and sand-amended peatland sites, 23 and 18 species were recorded in the topsoil-removed plots and 32 and 22 species were recorded in the control plots, respectively (Table 2). Non-peatland species, such as Alopecurus aequalis and Echinochloa crus-galli, were not observed in the topsoil-removed plots at the silt-amended peatland site, and the dominant species C. schmidtii had a higher importance value (0.577) than the control plots (0.131). However, Carex species was not observed at the sand-amended peatland site. At this site, non-peatland species such as Eleocharis wichurae and Echinochloa crus-galli dominated in the control and topsoil-removed plots, respectively.

The species richness, Shannon–Wiener Index, Pielou Index, Margalef Index, and Simpson Index were lower in the topsoil-removed plots than in the control plots at the silt-amended peatland site (Figure 3). However, no significant difference existed between the two treatments at the sand-amended peatland site in terms of species richness and aforementioned indexes, except the Margalef Index (Figure 3).

The coverage, density, and biomass of Carex species in the topsoil-removed plots were higher than those in the control plots at the silt-amended peatland site (Figure 4). The basal diameter, root length, tiller number, and above- and below-ground biomass of Carex species in the topsoil-removed plots were higher than those in the control plots (Figure 5).

RDA demonstrated that environmental factors explained 62.3% and 67.5% of the total variation of species composition at the silt- and sand-amended peatland sites, respectively. At the silt-amended peatland site, the peatland species including C. schmidtii were positively related to SOC, TP, SWC, and TN. At the sand-amended peatland site, C:N, SOC, and pH were the main controlling factors that affected species composition (Figure 6).

Agricultural activities including reclamation, drainage, and fertilization significantly decreased the peatland area globally and changed BD, nutrient availability, and macropore space in farmed peatlands (Kløve et al., 2010; Smolczynski et al., 2021; Wang et al., 2022). In this study, soil environmental factors differed significantly between silt- and sand-amended peatlands and responded differently to the topsoil removal treatment (Table 1). At the silt-amended peatland site, mineral silt soil covered surface horizons of peatlands; therefore, the mineralization of peatlands can be hampered by the silted soil on the surface. This led to less oxygen availability and biological activity, thereby protecting peat below the surface against oxidation (Smolczynski et al., 2021). Therefore, such soils had higher SOC, TN, TP, and SWC at the silt-amended peatland site after removing top fine-textured material. However, coarse sand was mixed with peat soil, some of which leached into deeper soil layers (30–40 cm) at the sand-amended peatland site. It significantly changed physical and chemical properties of the soil and created an aerobic environment, leading to an increase in the decomposition of organic matter. Therefore, no significant difference was observed in terms of soil properties between the topsoil-removed and control plots at the sand-amended peatland site (Table 1).

Because of long-term agricultural cultivation, many non-peatland species such as Echinochloa crus-galli, Murdannia keisak, and A. aequalis were observed in the restored peatlands, particularly in the control plots at the sand-amended peatland site (Table 2). These annual weed species reproduced and grew rapidly through the “R” reproductive response (Ma et al., 2019). However, the species richness and important indexes of these non-peatland species were lower in the topsoil-removed plots than those in the control plots (Table 2). This indicated that topsoil removal was effective in removing propagules of weed species from the soil.

More peatland species were observed in the topsoil-removed plots at the silt-amended peatland site (Table 2). This indicated that these peatland species were retained in the soil seed banks at the silt-amended peatland site, even in the deeper soil layer (10–20 cm) under the mineral soil. Former seed bank studies mainly focused on the upper 10-cm layer of wetland soils (Wang M et al., 2017; Davies et al., 2018; Zhao et al., 2021). However, cool, wet, and anaerobic conditions with slow microbial activity are beneficial for seed survival in peatlands; the seeds could remain viable at soil depth up to 50 cm (McGraw, 1987). In the silt-amended peatland, the surface silt soil may protect the seeds of peatland species in the layers below from physical damage and aerobic decomposition. Meanwhile, the removal of the top mineral soil layer eliminated the physical barriers that inhibited seed germination and plant smothering and exposed the peat layer below for seed germination of the retained peatland species.

Successful settlement and establishment of seedlings of target species are the most important aspects of vegetation restoration (Wang et al., 2020), where interspecific competition is a major constraint on seedling germination, growth, and survival (Kotorová and Leps, 1999). At the silt-amended peatland site, topsoil removal significantly increased the occurrence of Carex species; the coverage, density, and biomass of Carex species were 2.4, 3.0, and 5.7 times higher in the topsoil-removed plots than those in the control plots (Figure 4). By removing the topsoil, the number of weed seeds significantly declined, which provided a regenerative niche for seed germination and seedling establishment for Carex species. Therefore, the topsoil removal enhanced seedling recruitment of Carex species by reducing the interspecific competition. However, at the sand-amended peatland site, Carex species was absent in both control and topsoil-removed plots, whereas non-peatland species such as Echinochloa crus-galli and Poa annua were the dominant species. The coverage and aboveground biomass of plant community and plant diversity indexes exhibited no significant difference between the control and topsoil-removed plots. This suggests that topsoil removal did not significantly affect vegetation restoration at the sand-amended peatland site. Soil seed banks have been destroyed in sand-amended peatlands, which is a challenge to restoring the native vegetation.

RDA revealed that soil environmental factors explained more than 60% of variations in species composition in both silt- and sand-amended peatlands (Figure 6), which indicated that the change in the soil environment influenced seed bank germination. Former studies reported that agricultural practices reduced SOC, which affected the soil moisture and water holding capacity and prevented seed germination (Kettenring and Galatowitsch, 2011). In addition, seed viability in wetlands is affected by storage conditions; dry storage conditions reduced the seed viability of Carex species by 95% (Leck and Schütz, 2005). Similarly, Carex species disappeared, and few peatland species were present in the sand-amended peatland site in this study. This is because the mixing of sand into soil deteriorated soil environmental conditions, which damaged the seed banks and led to the absence of Carex species at the sand-amended peatland site.

Inconsistent with previous studies (Emsens et al., 2015; Harpenslager et al., 2015), topsoil removal did not reduce soil nutrient levels at the silt-amended peatland site in this study. On the contrary, topsoil-removed plots exhibited lower BD and higher SWC and soil nutrient (SOC, TN, and TP) levels (Table 1). However, soil properties of the topsoil-removed plots at the silt-amended peatland site were close to those of the natural peatlands (Wang et al., 2021). This can explain the strong relationships between peatland species (e.g., C. schmidtii and Lysimachia davurica), soil nutrient, and SWC at the silt-amended peatland site (Figure 6A). Therefore, topsoil removal at the silt-amended peatland site improved soil nutrient levels and created suitable abiotic conditions for the establishment of Carex species, leading to high biomass of Carex species in the topsoil-removed plots.

Our study illustrated that silt soil covering at the farmed peatlands provided suitable storage conditions for Carex seeds. The mixing of sands created the aerobic environment and led to an increase in organic matter decomposition and a significant change in physical and chemical properties of soil (Zaidel′man et al., 2001). Therefore, despite using for farming for a shorter period, the soil environment in sand-amended peatlands had dramatically changed. The result illustrated that the soil amendment type significantly affected peatland revegetation, and the silt-amended peatland exhibited a higher revegetation ability than the sand-amended peatland. Moreover, our findings confirmed that topsoil removal is an effective method for vegetation restoration in silt-amended peatlands because it not only improved soil properties but also promoted the growth of peatland species including C. schmidtii. However, topsoil removal is not a suitable way to restore all degraded peatlands, such as sand-amended peatlands. A detailed examination of soil environmental conditions and soil seed-bank distribution is necessary for successful peatland restoration.