P. ludlowii is naturally found in the region spanning from Linzhi City to Shannan City in the Tibetan Autonomous Region of China. The wild species studied in this research are situated in Qunigongga Village, Bayi District, Linzhi City (29°50’44”N, 94°63’56”E), at an altitude of 3,046 m with a temperate monsoon climate. The average annual precipitation is 702.1 mm, the average annual relative humidity is 71%, the average annual temperature is 8.7°C, and sunshine hours are 3,019.5 h. The soil is sandy loam. This wild population thrives on roadside forest margins, sunny slopes, and middle slopes. There were approximately 300 plants in this population, all of which grew well ( Figure S1 ).

The cultivated species P. ludlowii can be obtained from Luoyang Paeoniaceae Ex-situ Protection Center, located in Xiaohong Village, Luanchuan County, Luoyang City, Henan Province (33°56’05”N, 111°21’36”E) ( Figure S2 ) at an altitude of 1,240 m with a continental monsoon climate. The average annual precipitation is 800–1,000 mm, average annual relative humidity is 66%, average annual temperature is 14.5°C, and sunshine hours are 2,301.7 h, and the soil is sandy clay. Currently, more than 200 flowering plants of P. ludlowii were sown and bred in the autumn of 2002. Seeds were collected from Nyingchi City wild species of P. ludlowii, which first bloomed in 2008. Following harvesting seeds and raising seedlings, the flowering plants bloom and bear fruit normally (Wang et al., 2013). Since 2015, the introduction region has been poorly managed, with weed removal taking place only once a year in the summer and no fertilization.

Fixed sample plots were established based on the terrain and topography at predetermined sampling locations. This study focused on the following three stage: flowering, fruiting and autumn foliage, and the sampling time was determined according to the phenological observation results of Xue and Luo (2014). The collection time of wild species of P. ludlowii in Nyingchi City was 2021-5-8(LZT1), 7-6(LZT2) and 9-8(LZT3), and cultivated species of P. ludlowii in Luanchuan County was 2021-5-17(LCT1), 7-17(LCT2) and 9-13(LCT3). P. ludlowii has a robust fleshy root system with dense absorbent roots on its surface, which are densely distributed near the surface, therefore, soil sampling was conducted at a depth of 0–20 cm (Gao et al., 2022). In each habitat sample plot, five P. ludlowii plants with similar growth patterns, spaced > 3 m apart, were selected as the five biological replicates. Three sampling points were established at a distance of 30 cm from the base of each P. ludlowii. After removing surface litter, soil samples were collected using a soil drill. At the laboratory, roots were removed from each sample and large pieces of attached soil were shaken off. The rhizosphere soil attached to the root surfaces was slightly brushed off. Thirty samples were collected from the two regions and divided into two parts. One part was stored at -80°C for microbial sequencing, and the other part was dried to determine soil physicochemical properties.

According to the soil DNA extraction kit (PowerSoil^® DNA Isolation Kit) instruction manual steps for extracting DNA from rhizosphere soil, with a sample DNA concentration higher than 10 ng/μL. OD260/OD280 is between 1.8 and 2.0 to ensure DNA quality. The ITS general-purpose primers gITS7 (5 ‘GTGARTCATCGARTCTTTTG 3’) (Ihrmark et al., 2012) and ITS4 (5 ‘TCCTCCGCTTTATTGTGC 3’) (White et al., 1990) were selected for DNA amplification. PCR amplification system (50 μL) comprised the following: 10× PCR buffer, 5 μL; dNTPs, 4 μL; forward and reverse primers, 1 μL; Taq DNA polymerase, 0.5 μL; DNA sample, 1 μL; and ddH₂O, 37.5 μL. Following PCR conditions were employed: pre-denaturation for 5 min at 94°C, denaturation for 30 s at 95°C, annealing at 56–62°C for 30 s, extension at 72°C for 30 s, fungal cycling for more than 35 steps, and finally extension at 72°C for 10 min. After separation and purification by 1% agarose gel electrophoresis, DNA concentration was detected using a NanoDrop 2000 (Thermo Fisher Scientific), and 30 samples were mixed according to the DNA concentration, such that the mixed samples contained 150 ng DNA. Finally, sequencing was performed using an Illumina MiSeq platform (Magigene, Guangdong).

After natural air drying, the soil samples were sieved, and the soil pH was measured using a pH meter (MP523-01) with a water-to-soil ratio of 2.5:1. Soil organic matter content was determined by potassium dichromate oxidation - external heating method; Kjeldahl method was used to determine soil total nitrogen (TN) content. The content of soil available phosphorus (AP) was determined by 0.5 M NaHCO₃ extraction with molybdenum antimony resistance spectrophotometry. The determination of soil AK content by ammonium acetate extraction - flame photometric method (Nie and Wang, 2019).

Leaf area, plant height, crown width, number of flowers, number of fruits, and fruit-setting rate were selected to characterize the growth of wild and cultivated P. ludlowii species. To determine the leaf area, five parietal lobules located 3–4 fully expanded to apical lobules upward from the base of flower branches were selected from each plant, and their maximum transverse and longitudinal diameters were measured. Additionally, five flowering branches were selected from each plant to determine the number of flowers and fruits per plant. The fruit-setting rate refers to the percentage of the number of fruits divided by the total number of flowers in P. ludlowii (number of fruits/number of flowers × 100%).

The sequencing results were uploaded to the Galaxy online analysis platform (http://mem.rcees.ac.cn:8080) (Feng et al., 2017), the FLASH tool was used to merge the same sequence upstream and downstream primers (Magoc and Salzberg, 2011), and Btrim was used to filter reads, to remove any sequences with degenerate bases and lengths less than 200 bp, retaining only 240–350 bp reads (Kong, 2011). Finally, UPARSE was used to remove chimeras and cluster sequences into operational taxonomic units (OTUs) in the sequences, with a 97% similarity threshold (Edgar, 2013). The total number of samples was 30; however, in this experiment, the number of reads obtained by sequencing one sample from LCT3 was < 10,000. After the second experiment, the sample was still disqualified; therefore, it was excluded from subsequent analysis. Based on the resulting 36,985 sequences, the OTU table was re-extracted, and species annotation was performed using the Unite database. Abundance and diversity indexes (Shannon, Simpson, and Chao1) were calculated, and based on Bray-Curtis distance, principal co-ordinate analysis (PCoA) and non-metric multidimensional scaling (NMDS) were used to analyze β diversity; Multi-Response Permutation Procedure (MRPP), analysis of similarities (ANOSIM), and permutational multivariate analysis of variance (PERMANOVA) were used to identify dissimilarities between groups. Mantel analysis and canonical correspondence analysis (CCA) were used to measure the relationship between the community structure and environmental factors. Fungi Functional Guild (FUNGuild) was used to predict the function of the rhizosphere soil fungal communities. All the analyses were performed using the Galaxy online analysis platform.

Using the Galaxy online platform (http://mem.rcees.ac.cn:8081), a molecular ecological network of the rhizosphere soil fungal communities of wild and cultivated species was constructed using the random matrix theory (RMT) method (Deng et al., 2012; Xiao et al., 2022). Topological roles, including within-module connectivity (Zi) and among-module connectivity (Pi), are classified as module hubs, network hubs, peripherals, or connectors (Olesen et al., 2007). The module hubs, network hubs, and many connectors in their respective modules are highly connected and can be considered as keystone species in the entire network (Deng et al., 2012). The results were visualized using Cytoscape 3.9.1 software (http://www.cytoscape.org/), and Student’s t-test was used to measure the significance between the empirical networks and random networks. Community assembly mechanisms were inferred using phylogenetic bin-based null model analysis on the Galaxy-iCAMP online platform (http://ieg3.rccc.ou.edu:8080) (Ning et al., 2020; Gu et al., 2022).

In terms of growth indicators for P. ludlowii ( Table 1 ), the leaf area and crown width of the wild species showed significant increases with seasonal changes, whereas the plant height increased relatively gently, indicating that P. ludlowii exhibited vigorous growth in the original region. However, the leaf area of cultivated species only increased slowly with the change in season, and was significantly higher at LCT3 than at LCT1; therefore, the growth rate of P. ludlowii in the introduction region was relatively slow. Although the number of flowers and fruits of the wild species was significantly higher than that of the cultivated species, there was no significant difference in the fruit-setting rate between the two regions.

The physicochemical properties of the rhizosphere soils of wild and cultivated peonies are shown in Table 2 . The pH of the rhizosphere soil in the two regions was slightly acidic to nearly neutral, and the pH of the cultivated species was significantly greater than that of the wild species only at the flowering stage. The soil organic matter content was similar between the two regions; the organic matter content in both regions was the lowest at the flowering stage, and significantly increased to the maximum during the fruiting stage. The difference between the two regions was that during the autumn foliage stage, the content of soil organic matter of cultivated P. ludlowii species was reduced to a level similar to that at the flowering stage, whereas the soil organic matter content of wild species remained relatively unchanged from the fruiting stage to the autumn foliage stage. The AP content of the wild species was significantly higher than that of the cultivated species at different growth stages (P<0.01). The soil AP content of wild species gradually increased significantly with the growth period, whereas the AP content of cultivated species increased significantly from the flowering to fruiting stage and remained stable from the fruiting stage to autumn foliage stage. The variation trend of soil AK content in wild and cultivated species was consistent, both of which first increased and then decreased with the change in the growth period. Although the soil AK content of the wild species was higher than that of the cultivated species during all three growth periods, the difference was only significant during the autumn foliage stage.

In this study, the leaf area of LCT1 at the flowering stage was greater than that of LZT1 at the same stage, whereas the leaf area of the cultivated species increased slowly during the fruiting and autumn foliage stages. However, the wild species had a smaller leaf area before flowering, but showed the fastest growth rate from the flowering stage to fruiting stage, and still maintained a higher growth rate after the fruiting stage. Plants require more nutrients during the flowering stage and their roots absorb them vigorously, resulting in the lowest soil organic matter content and similar rhizosphere soil fungal community structures in the two regions.

The study revealed that the soil organic matter content of cultivated species varied significantly and was lowest at the flowering stage, increased significantly at the fruiting stage, and decreased to a similar level during the autumn foliage stage. The Mantel analysis revealed that only soil organic matter of the cultivated species significantly and positively correlated with the rhizosphere soil fungal community structure. In a study of rhizosphere soil microorganisms at the introduction region by Gao et al. (2022), soil pH was significantly positively correlated with rhizosphere soil bacteria, and soil organic matter, ammonium nitrogen, and AK content were significantly positively correlated with the AMF community but had no significant influence on the rhizosphere soil fungal communities. The difference in results may be due to differences in sampling time and soil layers. In this study, the sampling time was divided into three periods with a mixed soil layer of 0–20 cm, whereas the research conducted by Gao et al. (2022) included only stratified sampling of 0–10 cm and 10–20 cm in August. By decomposing complex plant residues, Mortierellomycota and Mucoromycota can increase soil phosphorus content and promote plant growth (Zhang et al., 2021; Fan et al., 2022). At the flowering stage of P. ludlowii, soil AP content of cultivated species was significantly lower than that of wild species, whereas the relative abundances of Mortierellomycota and Mucoromycota were significantly higher than those of the wild species. Such a contradictory conclusion may be explained by the following: (1) Due to the low phosphorus content in the mountainous soil at the introduction region, to meet the demand of phosphorus for plant growth, P. ludlowii may have recruited more Mortierellomycota and Mucoromycota to break down plant residues and release phosphorus. However, given the lack of soil background data before introduction, further research is required. (2) According to the research group’s investigation, there were few species of weeds in the introduced region, mainly Erigeron annuus (L.) Pers., Artemisia L. At the same time, weeds were removed every summer, and there were 78 species of accompanying plants of original habitat (Ni, 2009). Therefore, we concluded that the phosphorus content produced by the decomposition of Mortierellomycota and Mucoromycota was low, leading to a low AP content at the introduced region. Additionally, it was found that the soil AP of cultivated species was positively correlated with the rhizosphere soil fungal community structure, Therefore, low AP content at the introduced region is speculated to have affected plant growth, and the two types of fungi, Mortierellomycota and Mucoromycota, play an important role in promoting phosphorus absorption by the cultivated P. ludlowii species.

The effects of wild and cultivated species on the diversity of rhizosphere soil microbial communities are inconsistent. Compared to wild Agave salmiana, cultivated Agave tequilana has lower microbial diversity (Coleman-Derr et al., 2016), whereas other studies have shown that introduction and cultivation can improve the diversity of fungal communities in rice (Chang et al, 2021). In this study, although the α diversity of the cultivated species did not decrease compared with that of the wild species, the β diversity of the cultivated species was significantly different. Ascomycota and Chytridiomycota are mainly saprotrophic that, decompose organic matter in the soil into ionic states that can be absorbed by plants to help improve soil fertility (James et al., 2006; Beimforde et al., 2014). Among the cultivated species, Ascomycota consistently had the highest relative abundance and played an important role in decomposing dead leaves in introduced regions, providing nutrients for the growth of P. ludlowii (Gao et al., 2022).

In the rhizosphere soil fungal community of the wild species P. ludlowii, Ascomycota had the highest relative abundance only at the flowering stage; however, according to a molecular network analysis (Jiang et al., 2017), Ascomycota is the key group of the rhizosphere soil fungal community. Chytridiomycota had the highest relative abundance (2.99%) at the fruiting stage, which was significantly higher than that of the cultivated species. As a perennial flowering plant, P. ludlowii requires more nutrients during its flowering and fruiting stages. These two fungi decomposed dead branches, leaves, and other complex organic substances for plant absorption and utilization at different growth stages. Basidiomycota significantly increased in the autumn foliage stage, and became the most dominant species of the wild species P. ludlowii (Lu et al., 2020a), mainly due to the significant increase in Solicoccozyma, which contributed to improving the accumulation of medicinal components in the roots of the wild species P. ludlowii; it is speculated that the medicinal properties of the cultivated species P. ludlowii would be weakened (Schmidt-Dannert, 2016). The relative abundance of Mortierella in the cultivated species was significantly higher than that of the wild species, and the relative abundance of Humicola in LCT2 was significantly higher than that during the other stages and was also significantly higher than that in LZT2. Some Mortierella and Humicola fungi have biological control functions that can inhibit pathogenic bacteria such as Fusarium oxysporum, Alternaria brassicicola, and Phytophthora capsici (Ko et al., 2011; Edgington et al., 2014; Li et al., 2014; Yang et al., 2014). Consequently, the cultivated P. ludlowii species may be more susceptible to some pathogens, especially during the fruiting stage, P. ludlowii enhances its ability to resist pathogens by recruiting Mortierella and Humicola fungi. The relative abundance of Pseudogymnoascus was the highest during the flowering stage and then decreased significantly. This genus can decompose cellulose, which is beneficial for degrading dead branches and leaves, and improving the content of soil organic matter (Thormann, 2011). Pseudogymnoascus fungi are enriched during the flowering stage of P. ludlowii because of the high nutrient demand at this stage, thus becoming the most dominant genus in both the wild and cultivated species.

In mixed Pathotroph-Saprotrophs, LZT1 and LZT3 were significantly higher than LCT1 and LCT3. Mixotrophic fungi exhibit a diverse fungal functional profile, that is conducive to maintaining ecological stability (Fan et al., 2022). Thus, wild species have more stable fungal ecosystems than cultivated species. Although there was little difference between Pathotrophs in the wild and cultivated species, the Soil Saprotroph of LCT2 was significantly higher than that of LZT2, and the Mixed Saprotroph of LCT1 was significantly higher than that of LZT1. These Saprotrophs are prone to enrichment in decaying and diseased plants and are more prone to causing plant diseases (Zhang et al., 2022). Rhizosphere soil fungi and bacteria exhibited similar patterns, compared to cultivated varieties, wild beet (Beta vulgaris ssp.) and wild soybean (Glycine soja Sieb. et Zucc) have diverse rhizosphere microbial communities, and microbial communities isolated from native soil have antibacterial activities that can protect plants against environmental stress (Zachow et al., 2014; Ma et al., 2019).

The aggregation structure of rhizosphere soil fungal communities was similar between wild and cultivated species and was dominated mainly by stochastic processes. The most important ecological process was drift (40.28–57.55%), followed by dispersal limitation (17.28–37.00%), with no significant differences. Drift is a stochastic process in the soil microbial community caused by factors such as reproduction and gradual death (Zhou and Ning, 2017). Dispersal limitation increases community variation and diversifies the microbial community structure (Stegen et al., 2013). Both are important ecological assemblages that drive soil fungi in multiple regions (Aslani et al., 2021; Gu et al., 2022). The difference is that drift and dispersal limitations exist stably in the cultivated species, whereas the opposite trend is observed in wild species. Drifts gradually increased with the growth period, whereas dispersal limitation gradually decreased, showing a significant difference from LZT1 to LZT3. However, the specific reasons for this phenomenon require further exploration. Homogeneous selection (16.84–28.79%) drove the deterministic ecological process of the fungal communities of wild and cultivated species, which enabled fungal communities to maintain a stable state after being disturbed by the environment, and the community structure was more similar (Dini-Andreote et al, 2015).

In this study, ecological network analysis was used to visualize the interactions between rhizosphere soil fungi in wild and cultivated species (Fuhrman, 2009). Based on the module network topology parameters, wild species exhibited more complex molecular network modules than did the cultivated species, indicating that wild species of P. ludlowii can respond more stably to changes in the external environment. Although the growth of the cultivated P. ludlowii species did not show a significant decline in this study, in the long run, as harmful fungi increase, beneficial fungi decrease, and network connections between fungi decline, plant–soil negative feedback is likely to occur (Martín-Robles et al., 2020). Studies have reported that the introduction and domestication of many food crops such as soybean, rice, and tomato, reduce network complexity and plant resistance to pathogenic fungi, resulting in microbe-mediated soil negative feedback (Carrillo et al., 2019; Chang et al., 2020; Chang et al., 2021).