1. Introduction

Land pollution and ecosystem degradation from mining are major global environmental concerns, especially in developing countries (Wang et al., 2017). Worldwide, mining has affected approximately 400,000 km², with heavy metal contamination impacting more than 20,000,000 ha (Heydari et al., 2020). Following metal ore extraction, abandoned land often faces severe abiotic pressures, including high heavy metal concentrations, poor soil erosion control, unstable physicochemical properties, and nutrient deficiencies (Mi et al., 2019). Heavy metal contamination and nutrient scarcity severely threaten the survival, growth, and reproduction of plants, animals, and microorganisms in these areas (Liu et al., 2018; Hou et al., 2025). In moister mining regions, the lack of vegetation cover accelerates the spread of heavy metals through surface runoff, further intensifying pollution. These heavy metals can bio-accumulate and move up the food chain, ultimately endangering human health (Hamid et al., 2019; Food and Agriculture Organization of the United Nations FAO, 2024). Consequently, the ecological restoration and remediation of mining areas are critical global environmental challenges demanding immediate attention.

Environmental remediation of abandoned mining areas must mitigate the harmful effects of heavy metal pollution and land degradation (Fazlioglu et al., 2021) as well as improve ecosystem functioning. Research on remediation and restoration have generally focused on phytoremediation and reclamation, soil amendment and earthworks, chemical remediation, and microbial remediation (Li and Huang, 2015). However, these approaches primarily target short-term problems, such as decreasing heavy metal concentrations, lessening their toxicity, and stabilizing soil and nutrients, often overlooking long-term needs such as successful plant establishment, community sustainability, and effective ecosystem function recovery in these environments. Accordingly, Yuan et al. (2023) underscored that the ecological restoration of abandoned mining areas should align with natural successional processes to facilitate the establishment of stable and self-sustaining plant communities. Infrequent, low-intensity human disturbances can provide survival and colonization opportunities for tolerant plants, potentially forming the ecological basis for the development of plant communities and the restoration of ecosystem functions in abandoned mining areas (Yuan et al., 2022). Investigating mechanisms that influence stability of site-adapted plant species communities provides a foundation for simulating natural successional trajectories in abandoned mining areas and advancing net positive restoration outcomes in mining ecosystems (Wynberg et al., 2023).

Over time, spontaneously growing plants in abandoned mining areas gradually develop distinct survival strategies and dominant functional traits (Sulavik et al., 2021). Grime (1977) proposed the CSR strategy framework, which encompasses competitive, stress-tolerant, and ruderal strategies. This framework explains the adaptive ecological responses of plants to environmental stress and highlights the outcomes of plant–environment interactions. The expression of CSR strategies reflects the trade-offs that plant species make in allocating limited resources among functional traits, specifically balancing investments in competitiveness, stress tolerance, and resistance to disturbance (Pierce et al., 2017). The categories defined by the similarity of plant species in functional traits and in their responses to environmental stress are referred to as plant functional groups (PFGs) (Davison et al., 2020). These PFGs are closely linked to community composition and ecosystem functioning, and their stability is a key factor in the restoration of plant communities in abandoned mining areas (Lavorel and Grigulis, 2015). Different PFGs play various roles in ecosystem restoration, including photosynthesis, nutrient cycling, water regulation, and soil retention. Their unique ecological functions interact, collectively driving the recovery of degraded ecosystems (Martinez-Ramos et al., 2021). PFGs stability is closely linked to the reconstruction of ecological niches. In degraded ecosystems, plant niches can shift, altering the number and composition of PFGs (Petchey and Gaston, 2006).

The interactions between PFGs and environmental factors within plant communities vary, leading to different mechanisms by which each functional group is maintained (Lavorel et al., 1997). For example, Wen et al. (2022) showed that root functional traits, such as diameter, tissue density, and length, critically enhance phosphorus (P) and nitrogen (N) bioavailability in rhizosphere soils. These occur through different mechanisms, including direct physicochemical release for P and indirect microbe-driven activation for Liu et al. (2022) found that during the restoration of degraded alpine grasslands, categorizing plants into grasses, sedges, and forbs revealed increased N uptake by grasses and decreased uptake by forbs. These functional groups occupy distinct chemical ecological niches, decreasing N competition. In abandoned mining areas, interactions between plants and the soil are vital for the formation and development of spontaneous plant communities, and the key functional traits of tolerant plants are essential for adapting to stressful conditions (Gundale and Kardol, 2021; Yuan et al., 2022). However, few studies have examined how different PFGs in spontaneous communities within the extreme environments of abandoned mining areas respond to stress-induced changes and the mechanisms maintaining their stability.

We studied the spontaneous plant communities in an abandoned lead (Pb)–zinc (Zn) mine in the northeastern part of Yunnan Province, China. We hypothesized that local plant species in the area respond to the harsh environmental conditions of contaminated soils by adjusting their functional traits and survival strategies, thereby enhancing the stability of spontaneous plant communities through the development of distinct plant functional groups (PFGs). The objectives of this study were to: (i) elucidate the patterns of change in plant CSR strategies (competitive, stress-tolerant, and ruderal) along gradients of increasing contamination; and (ii) classify site-adapted plant species into distinct plant functional groups (PFGs) through cluster analysis of functional traits, and determine the mechanisms underpinning the stability of these groups. A comprehensive understanding of the key factors influencing these processes, and the underlying mechanisms, will provide a foundation for nature-based solutions to the remediation of abandoned mining areas.

2. Materials and methods 2.1. Study area

The study area was situated in Kuangshan town, Huize county, in northeastern Yunnan Province, China (26.64°N, 103.71°E) (Fig. 1), covering approximately 100 ha. This region exhibits a typical temperate plateau monsoon climate, characterized by minimal seasonal variation, mild summers, cold and dry winters, and distinct wet and dry seasons. The annual average precipitation is 858.4 mm, with an average elevation exceeding 2200 m. The region experiences an average of 225 sunny days per year, a total annual sunshine duration of 2100 h, a mean annual temperature of 12.7 ℃, an average relative humidity of 79%, and an annual mean wind speed of 2.5 m/s (Yuan et al., 2022). The native soil in the area is predominantly yellow-brown, with weakly acidic to neutral properties. The vegetation consists primarily of sparse herbaceous plants and shrubs, with patches of secondary forest in some areas (Yuan et al., 2023).

Mining activities had been carried out in the study area since the Ming Dynasty and ceased in 1994 (Li et al., 2019). Prolonged and unregulated open-pit mining has resulted in severe degradation of the surrounding vegetation. The site has remained abandoned for the past 30 years. Following abandonment, no artificial revegetation measures or soil amelioration techniques were implemented in the affected areas (Yuan et al., 2022). Furthermore, the traditional crucible lead-smelting technique, known as "precipitation smelting, " has led to major contamination of the soil with manganese (Mn), copper (Cu), cadmium (Cd), Zn, and Pb (Li et al., 2019; Hou et al., 2025).

A portable heavy metal analyzer (Vanta; Olympus, Tokyo, Japan) was used to conduct a preliminary assessment of three open-pit mines in the study area, revealing great spatial heterogeneity in contamination. To address this, the "random walk marker-throwing" method was applied, with 40 measurement points randomly selected in each pit. These points were spaced at least 30 m apart, with the starting and ending points overlapping, to ensure comprehensive detection of heavy metals. After excluding outliers, the data were analyzed and used to classify the three mining pits into three distinct pollution areas: A, B, and C. Additionally, a blank control sampling area (CK) was established in an unmined area using the same methodology (Fig. 1 and Table S1).

2.2. Plant identification, sample collection, and functional trait measurement

Twenty randomly selected 5 m × 5 m quadrats (with slopes less than 13°) were established in each sampling area, resulting in a total of 80 quadrats, to perform plant surveys and functional trait measurements. The plant species in each quadrat were identified in the field; for species that could not be identified onsite, we collected specimens and identified them in our lab (Wang, 2020). For this, we conducted morphological identification and sent specimens to the Kunming Institute of Botany, Chinese Academy of Sciences, for expert anatomical identification. A comprehensive list of the plant species is provided in Table S2. In each sampling area, 20 intact individuals of each plant species were selected for functional trait measurements, ensuring a total sample size of 80 plants per species. During sample collection, the entire plant was excavated to the depth corresponding to the maximum extent of its root system. Samples of the same species were placed in polyethylene self-sealing bags, air was expelled, and the bags were sealed for preservation. All samples from the same quadrat were consolidated into a single plastic bag. After transport to the laboratory, the samples were thoroughly washed with clean water, air-dried, and then used to measure root traits and biomass (Yuan et al., 2022).

This survey measured eight plant functional traits related to survival to extreme habitats: height (cm), leaf area (mm²), crown (cm), biomass (g), total root length (TRL; cm), total root surface area (TRSA; cm²), mean root diameter (MRD; mm), and number of root tips (NRT) (Neto et al., 2019). Height and crown were measured in the field using a tape measure and caliper. Leaf area was measured in the field using a portable leaf area meter (YMJ-B; Top Cloud-agri, Hangzhou, China). All root measurements were made using the WinRHIZO root analysis system (Regent Instruments, Quebec City, Quebec, Canada). Finally, the biomass of herbaceous plants, sub-shrubby herbaceous plants, and dwarf shrubs was determined using the drying method. The procedure involved placing plant samples in a constant-temperature air-drying oven (DZK-6025; JUCHUANG, Qingdao, China), initially heating them at 105 ℃ for 30 min for blanching treatment, followed by drying at 65 ℃ until the samples reached a constant weight (Yuan et al., 2023). Finally, the dry weight of the samples was precisely measured using an electronic balance (YP30002; Shanghai Yueping Scientific Instrument Manufacturing, Shanghai, China). For large shrubs (such as Coriaria nepalensis), biomass, TRL, TRSA, MRD, and NRT were estimated using empirical equations (Fig. S1 and Table S3).

2.3. Soil sample collection and measurement

Soil samples were collected using the five-point sampling method. Soil from five sampling points was combined into a single sample, with a weight ranging from 500 to 1000 g and a depth of 0–30 cm. At the center of each area, undisturbed soil was collected using a cutting ring (depth: 0–30 cm). After collection, excess soil at both ends of the cutting ring was removed, and external impurities were cleaned off. Then, the cutting ring was covered, and the total weight of the ring and the wet soil was measured on-site to determine the bulk density (SBD) and water content (SWC) of the soil (Yuan et al., 2023). In total, 80 soil samples and 80 cutting ring samples were collected across four sampling sessions. All soil samples were placed in polyethylene self-sealing bags and stored after air-drying. Upon returning the samples to the laboratory, stones, roots, and other debris were removed. After the natural drying process was complete, the samples were ground (2 mm) and sieved (0.149 mm) to determine relevant soil parameters.

Cutting ring samples were placed in a constant-temperature air-drying oven set at 105 ℃ and dried to a constant weight. Then, the total weight of the ring and the dried soil sample was measured using an electronic balance to calculate the SBD and SWC (Liu et al., 2010). A 10-g air-dried soil sample, passed through a 2-mm sieve, was weighed using an electronic balance and placed into a 100-mL centrifuge tube. Next, 25 mL of ultrapure water was added, and the centrifuge tube was placed on a shaker at 180 rpm for 30 min. After removal, the tube was allowed to stand for 30 min. Finally, the pH of the upper extraction liquid was measured using an OHAUS ST5000/F pH meter (OHAUS, Parsippany, NJ, United States) (Ciarkowska, 2017). An air-dried soil sample (0.03 g), passed through a 0.149-mm sieve, was weighed using an analytical balance (BCE224i-1CCN; Sartorius, Göttingen, Germany), wrapped tightly in aluminum foil, and placed into the automatic sample tray of a total organic carbon/total N analyzer (Elementar, Langenselbold, Germany) to measure total organic carbon (TOC) and total N (TN) in the soil (Yang et al., 2023). The average limit of detection (LOD) for TOC was 0.02 mg/L and average limit of quantification (LOQ) for TOC and TN were 0.002 mg/L and 0.02 mg/L, respectively. The average coefficient of determination (r²) value of the calibration curve was 0.9987. To determine total P (TP) through colorimetric measurement at a wavelength of 880 nm, a 0.1-g air-dried soil sample, passed through a 0.149-mm sieve, was dissolved in 5 mL of H₂SO₄ (reagent grade) and subjected to digestion using a graphite digestion system (OLB-U60; OLABO, Jinan, China). Then, the digested sample was analyzed using a multifunctional microplate reader (Epoch 2; BioTek, Winooski, VT, USA) (Zhang et al., 2013). The average LOD for TP was 0.000015 mg/L and the average LOQ was 0.0085 mg/L. The average r² value of the calibration curve was 0.9986. A 0.01-g air-dried soil sample, passed through a 0.149-mm sieve, was weighed using an analytical balance, and then 5 mL of concentrated H₂SO₄ (reagent grade) was added. The sample was digested at 375 ℃ for 4 h, followed by analysis on an inductively coupled plasma optical emission spectrometer (PlasmaQuant 9100; Analytik Jena, Jena, Germany) to determine the total content of heavy metals (Mn, Cu, Zn, Cd, and Pb) in the soil through elemental quantitative analysis (Yuan et al., 2024). The average LODs were Mn (0.000040 mg/L), Cu (0.000372 mg/L), Zn (0.000172 mg/L), Cd (0.000078 mg/L), and Pb (0.001046 mg/L); the average LOQs were Mn (0.0001 mg/L), Cu (0.0011 mg/L), Zn (0.0005 mg/L), Cd (0.0001 mg/L), and Pb (0.0031 mg/L); and the average r² values of the calibration curve were Mn (0.9998), Cu (0.9999), Zn (0.9997), Cd (0.9999), and Pb (0.9998).

2.4. Quality assurance and statistical analyses

All glassware was cleaned, then soaked in a 10% hydrochloric acid solution for 24 h, followed by rinsing with ultrapure water and drying. For every 20 samples, two consecutive continuous calibration standards were measured, and the best standard was selected for analysis. Additionally, the calibration curve was ensured to be linear, with an r² value greater than 0.990.

The contribution of different traits to competitive (C), stress-tolerant (S), and ruderal (R) strategies varies with environmental conditions, research goals, and the importance of specific functional traits (Wright et al., 2004). Therefore, directly applying fixed weights, as in the StrateFy method, often fails to fully capture this variability (Pierce et al., 2017; Li et al., 2024). Based on previous research findings, the CSR strategies of different plants were calculated after classifying the functional traits into 24 categories and assigning CSR strategy types and corresponding weight ranges (Table S4) (Pierce et al., 2005; Yuan et al., 2023). After calculating the CSR strategies of all plant species based on their weight, ternary plots were generated using the "ggtern" package to investigate the dynamic changes in CSR strategies within communities as pollution levels increased (Hamilton and Ferry, 2018).

To assess changes in the number of PFGs and species composition with an increase in pollution, cluster analysis of plant communities from different sampling areas was performed using Ward's method (Charrad et al., 2014). One-way analysis of variance was used to evaluate variation in eight functional traits across different PFGs, as well as the dynamic changes in the physicochemical properties and heavy metal levels of soils across different sampling areas. Multiple comparisons between different PFGs and sampling areas were performed using the least significant difference or Friedman tests in the "DescTools" and "rstatix" packages (Kassambara, 2021; Andri, 2022). Pearson correlation analysis from the "Hmisc" package was used to explore the relationships between plant functional traits and soil environmental factors (Harrell, 2022). To examine the impact of environmental factors on key functional traits of different PFGs, a random forest model was applied to analyze the contribution of influencing factors (Liaw and Wiener, 2002). Based on the results from the correlation analysis and random forest model, repeated modeling was conducted using piecewise structural equation modeling (SEM) to evaluate the mechanisms maintained by different PFGs and calculate the direct, indirect, and overall effects of different influencing factors (Lefcheck, 2016). All statistical analyses and graphical representations were conducted using the open-source software R (R Core Team, 2018).

3. Results 3.1. Differences in soil properties and plant survival strategies across different sampling areas

The four sampling areas showed a pollution gradient ranging from low to high, namely, a non-polluted area, low-pollution area, medium-pollution area, and high-pollution area (Table S1). In total, 36 plant species were identified in the study area. As pollution levels increased, the number of plant species decreased progressively, from 34, 34, 33, to 28, respectively (Table S2). The species Symphytum officinale and Buddleja crispa were more often observed in polluted areas (Table S2). Furthermore, TN, TP, total soil organic carbon (SOC) levels, and SWC gradually decreased as pollution increased, while soil pH increased and SBD remained constant (Table 1). The survival strategies of plants in the community changed significantly with increasing pollution levels (Fig. 2 and Table 2). Specifically, the number of stress-tolerant plants showed a fluctuating decline, ruderal plants exhibited a pattern of first decreasing and then increasing, and the number of competitor plants significantly increased.

3.2. Changes in PFG abundance and functional traits with increasing pollution levels

The optimal number of groups in the CK area was found to be three, while the optimal number in areas A, B, and C was four (Fig. S2). Consequently, in the non-polluted area, the plant community was divided into three functional groups (Group 1, Group 3, Group 4); in polluted areas, they were divided into four functional groups (Fig. 3). Analysis of differences in eight functional traits across the PFGs revealed that Group 1 had a larger leaf area, Group 2 had a greater root surface area, Group 3 had a lower degree of all eight functional traits than the other groups, and Group 4 had the greatest height, largest crowns, and highest values for biomass, TRL, MRD, and NRT (Fig. 4). These results suggest that Group 1 adapted to polluted environments by enhancing photosynthetic capacity and Group 2 increased the contact area of roots and soil. Group 3 showed the highest species diversity with greater functional redundancy and potential for recovery (Fig. 3). Finally, Group 4 demonstrated strong environmental adaptability and served as the dominant species. There were no significant differences in PFGs with an increase in pollution, suggesting that these PFGs formed via environmental filtering.

3.3. Differences in the effects of environmental factors on PFGs

The effects of environmental factors on plant group functional traits in abandoned mines varied (Fig. 5). For plants in Group 1, heavy metal levels were negatively correlated with crown, biomass, TRL, and TRSA, but positively correlated with leaf area and NRT. In Group 1 the functional traits with the strongest relationship with environmental factors were biomass and TRL; Group 1 functional traits were influenced by heavy metal levels, pH, TN, and TOC (Fig. 5a). For plants in Group 2, heavy metal levels were negatively correlated with height, biomass, TRL, and MRD, but positively correlated with crown, TRSA, and NRT. The relationships between biomass and TRSA and environmental factors were crucial, marking them as key functional traits (Fig. 5b). For plants in Group 3, heavy metal levels were negatively correlated with plant height, leaf area, and biomass. The relationships between leaf area and biomass and environmental factors were critical, serving as key functional traits, with the strongest correlations to heavy metals, pH, TN, and TP (Fig. 5c). For plant in Group 4, heavy metal levels were negatively correlated with crown, biomass, TRL, TRSA, MRD, and NRT, but positively correlated with plant height (Fig. 5d). However, the physicochemical properties and nutrient element content did not significantly influence the maintenance of Group 4. To assess the relative impact of soil environmental factors on key functional traits in Groups 1, 2, and 3, we used a random forest model (Fig. 6). The results indicated that heavy metal levels had the greatest impact on key functional traits across different PFGs, with Pb being the predominant contributing element.

3.4. Key environmental factors maintaining the stability of different PFGs

Our integrated analytical approach established mechanistic frameworks through which distinct plant functional groups maintain stability. Group 1 exhibits three key regulatory pathways: (1) Soil pH (λ = −0.31) exerts direct suppression on community biomass; (2) pH (λ = −0.73) → TRL (λ = 0.39) → Biomass; (3) pH (λ = −0.73) → TRL (λ = 0.52) → TN (λ = 0.21) → Biomass. Notably, pH is the dominant edaphic driver (total effect = −0.58), while TN reflects the strongest nutrient mediation (total effect = 0.21) (Fig. 7a–d). Group 2 reveals two critical pathways: (1) pH (λ = 0.66) → Pb (λ = −0.32) → Biomass; (2) pH (λ = −0.49) → TN (λ = 0.27) → Biomass. Here, pH (total effect = −0.45) and Pb (total effect = −0.30) constitute primary stressors, with TN maintaining notable nutrient influence (total effect = 0.27) (Fig. 7b–e). Group 3 is regulated by three factors: (1) pH (λ = −0.41) → Biomass; (2) pH (λ = −0.66) → Pb (λ = −0.35) → Biomass; (3) pH (λ = −0.66) → Pb(λ = −0.66) → LA (λ = 0.19) → Biomass. This group shows heightened sensitivity to pH (total effect = −0.67) and Pb (total effect = −0.48), although the capacity of TN to mediate is lower (total effect = 0.07) (Fig. 7c–f). Our findings reveal that mineral extraction fundamentally alters soil pH profiles, initiating cascade effects through 1) enhanced heavy metal mobilization (particularly Pb); 2) disruption of root architecture and foliar development; and 3) modification of nitrogen cycling dynamics.

4. Discussion

Variation in mining intensity and duration significantly impact the physical structure and chemical properties of the soil, as well as its ability to accumulate nutrients, leading to distinct environmental conditions across different mining pits (Dong et al., 2020). In this study, three primary mining areas were classified according to heavy metal contamination levels, with an unpolluted natural area serving as a control. This study demonstrated that as pollution increased, plant species richness, nutrient levels, and soil moisture steadily decreased (Table 1, Table S2). Conversely, soil pH gradually increased. Prior research indicates a strong link between plant-soil feedback mechanisms and a negative correlation between soil nutrients, moisture, and heavy metal concentrations in abandoned mining areas (Yuan et al., 2022). Specifically, free nutrient ions compete with free heavy metal ions for cation exchange sites; nutrients can form insoluble salts, directly adsorb heavy metal ions, and enhance the soil adsorption capacity; and elements can precipitate out with heavy metal ions (Tardif et al., 2019). Additionally, during mineral development, soil substances such as calcium carbonate and aluminum oxide can bind heavy metal ions into stable complexes or precipitates, releasing hydroxide ions that increase soil alkalinity and pH. Simultaneously, heavy metals such as Zn and Cu react with soil carbonates to form insoluble carbonate precipitates, decreasing free metal ion concentrations and mitigating heavy metal acidification effects (Ewusi et al., 2022).

Heavy metal pollution significantly shifted the survival strategies of plants, where the abundance of stress-tolerant species fluctuated while declining, that of ruderal plants initially decreased and then increased, and that of competitor plants significantly increased (Fig. 2 and Table 2). This finding is consistent with the study by Li et al. (2022), which demonstrated that with increasing levels of heavy metal contamination (e.g., Cd, Cr, Pb, Zn, and Ni), plant survival strategies undergo marked shifts, transitioning from primarily stress-tolerant strategies to a typical CSR strategy without a clearly dominant component. However, Fazlioglu et al. (2021) reported that in abandoned mining areas, stress-tolerant species were more abundant, which contrasts with the results of the present study. In addition, Onipchenko et al. (2025) through the analysis of the competitive-stress-tolerant-ruderal (CSR) strategies of different plant species in nutrient-poor environments of the Caucasus Mountains, indicated that the dominant native species (such as Veratrum album, Polygonum bistorta, and Gentiana punctata) mainly adopted competitive strategies. Similarly, Li et al. (2024) conducted a controlled experiment on the Qinghai-Tibet Plateau and found that with increasing nitrogen addition, the C scores of plants increased significantly, suggesting that nitrogen input enhances the proportion of competitive strategies within plant communities. This observation supports our conclusion that nitrogen availability is a key factor in maintaining the stability of spontaneous plant communities. Taken together, these studies suggest that under stressful environmental conditions, competition for nitrogen resources may be a primary driver promoting the shift towards competitive strategies in plants.

This study showed that heavy metal pollution increased PFG richness. As a plant community recovered, representative characteristic plants emerged within different functional groups, potentially serving as dominant species (Fig. 3). Ntloko et al. (2024) showed that in artificially restored plant communities on abandoned land in the Lesotho diamond mining area, plant abundance significantly increased over time. This was closely linked to the distribution of 15 spontaneously established native species. Furthermore, the presence of native plant species in the topsoil seed bank and enhanced natural seed dispersal together promoted plant species diversity and accelerated recovery. Our results are in line with these findings, highlighting the crucial role of native species in enhancing biodiversity during the recovery of plant communities in extreme environments. Moreover, this study revealed that dynamic shifts in plant survival strategies and PFGs in response to changing environmental conditions aligned with the biogeochemical ecological niche hypothesis. This hypothesis posits that each plant species requires a specific nutrient ratio within its ecological niche and exhibits unique tolerance to heavy metals, enabling survival and reproduction under varying environmental pressures (Sardans et al., 2021).

Thomas et al. (2018) suggested that using cluster analysis or directly applying key functional traits could offer new insights into predicting changes in plant communities and their maintenance. Consequently, we followed these recommendations and found that different PFGs exhibited distinct functional trait characteristics (Fig. 4). Group 1 showed a larger leaf area, Group 2 had a larger root surface area, and Group 3 had a lower degree of all eight traits. Group 4 plants were the tallest with the biggest crowns and had the greatest biomass, TRL, MRD, and NRT. These differences reflect the different survival strategies among these groups, where nutrient-poor soil and severe pollution necessitate specific traits for survival (Tang et al., 2023). Plants in different functional groups enhance their competitiveness through distinct growth strategies, resource acquisition methods, and adaptive traits such as root depth, stress resistance, and growth rate (Salguero-Gomez, 2017; Yang et al., 2023). Additionally, plant genetics significantly influence the expression of functional traits (Fasola et al., 2015). Key functional traits, including stress resistance, reproductive strategy, and resource use efficiency, directly affect population dynamics, driving community recovery and species coexistence in extreme environments (Fulgione et al., 2022). Such changes are also closely linked to the composition of microbial communities in the soil, which shows a significant phylogenetic correlation with plant growth forms and photosynthetic pathways (Davison et al., 2020). Arbuscular mycorrhizal fungi (AMF) and plant growth-promoting rhizobacteria (PGPR) play a crucial role in regulating the bioavailability of heavy metals in soils, alleviating plant stress responses, and facilitating plant symbiosis (Liu et al., 2023). Asif et al. (2025) demonstrated that AMF, in association with mycorrhiza helper bacteria (MHB), can markedly enhance plant tolerance to high concentrations of heavy metals (such as Cr, Ni, and Cd) and nutrient-deficient conditions by modulating nitrogen, phosphorus, and potassium cycles, as well as metabolic pathways involved in metal resistance and detoxification. Furthermore, Chi et al. (2025) showed that elevated soil nitrate reductase (S-NR) activity stimulates amino acid metabolism in the rhizosphere, drives shifts in microbial life-history strategies, reshapes the nutritional microenvironment around the roots, and enhances nitrogen cycling efficiency. Collectively, these processes contribute to effectively mitigating the survival stress imposed by heavy metals on plants and to reshaping their adaptive strategies.

This study found that heavy metal pollution had dual effects on PFGs, stressing some (Groups 2 and 3) while promoting others (Group 1) (Fig. 7). Stress-tolerant plants might enhance stress resistance or resource use efficiency, promoting certain traits, whereas competitor plants might experience declines due to inadequate adaptation or direct pollution effects (Deng et al., 2007; Fazlioglu et al., 2021). However, in our study, Group 4 (large shrubs) showed little impact from pollution (Fig. 5). This is in line with Germain et al. (2021), who provided strong evidence that once plants reach the large shrub stage, species coexistence mechanisms shift from early pioneer plant communities to more mature ecosystems. Furthermore, related studies have investigated the mechanisms underlying the strong adaptability of shrubs in mine tailings. For example, Nong et al. (2023) demonstrated that woody plants such as Broussonetia papyrifera and Koelreuteria paniculata effectively increase soil pH in manganese mining areas, thereby mitigating the detrimental effects of heavy metal contamination and promoting their stable growth. Moreover, Colin et al. (2019) found that functional traits related to leaf litter decomposition rates and root architecture in shrubs significantly enhance soil fertility in tailings, which in turn regulates the composition and structure of bacterial communities. These changes profoundly influence soil microbial assemblages and substantially improve resource availability. Nevertheless, since Group 4 contains only a single species, this may limit its ecological applicability.

This study also found that soil pH, heavy metal levels, and nutrient levels collectively influenced the maintenance of key functional traits and community stability within PFGs (Fig. 7). Among these factors, TN in the soil played a crucial role in mitigating heavy metal stress (Fig. 7). Li et al. (2016) demonstrated that N accelerates nutrient dissociation from minerals by increasing microbial respiration, thereby supplying essential nutrients to plant communities. Furthermore, certain microorganisms play a vital role in the formation of organic carbon in the soil. These microorganisms not only promote N fixation and the dissolution and transformation of P ions but also stabilize organic carbon by altering the soil structure (Kallenbach et al., 2016).

Vegetation cover improves the physical structure and elemental composition of soils in abandoned mining areas, thereby enhancing interactions between plants and soil (Xiao et al., 2021). This promotes the establishment of plants from various ecological niches, leading to the formation of multiple "functional survival islands." The connections among and complementary effects of these islands provide a foundation for the natural restoration of plant communities in abandoned mining areas (Yuan et al., 2022; Luza et al., 2023). Future research should focus on identifying similar functional traits across different PFGs, as these traits represent comparable ecological processes and are essential to the formation of natural plant communities in abandoned mining areas.

5. Conclusion

This study showed that mineral extraction significantly decreased the levels of essential nutrients and the water-holding capacity of soils in abandoned mining areas, while increasing soil pH. Heavy metal pollution caused changes in PFGs and plant community composition, leading to a fluctuating decline in stress-tolerant plants, an initial reduction followed by an increase in ruderal plants, and a significant increase in the number of competitor plants. However, different heavy metal levels did not significantly affect the functional traits of plants. Soil pH, heavy metal levels, and nutrient levels collectively influenced the maintenance of key functional traits and stability within PFGs. Notably, N in the soil plays a crucial role in mitigating heavy metal stress. This study lays the foundation for understanding the mechanisms by which natural plant communities are maintained in abandoned mining areas, providing a theoretical basis for nature-based solutions to remediation in such environments.

Acknowledgements

This research was supported by the National Natural Science Foundation of China (32371707 and 32260315), China Yunnan Provincial R&D Programs (202405AF140014, 202405AM340002, 202302AO370015, 202201AS070016, and 202201BF070001-002), and Yunnan University recommended exempted graduate research and innovation program (TM-23237006). This research expressed sincere gratitude for the strong support provided by the PhD program of the Youth Talent Promotion Plan of the China Association for Science and Technology (Chinese Society of Forestry).

Credit authorship contribution statement

Xin-qi Yuan: Writing, Conceptualization, Investigation, Data curation, Software, and Methodology. Yin-jie Li: Conceptualization, Data curation, and Investigation. Yao Zhao, Fu-xiang Peng, and Wen-jing Zhang: Investigation. Chang-qun Duan and Chang-e Liu: Funding acquisition, review, and editing. All authors contributed critically to the drafts and gave final approval for publication.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.pld.2025.08.007.