Divergent non-structural carbohydrate allocation strategies in desert shrubs under coal mining-induced environmental gradients in Xinjiang, Northwestern China

Abstract

Background

Large-scale open-pit coal mining can create pronounced environmental gradients around mine sites. These gradients are often accompanied by coal-dust dispersion and elevated heavy-metal loads in surrounding soils, which may alter physiological and ecological processes of vegetation. As a key indicator of plant carbon balance, non-structural carbohydrates (NSC) provide valuable insights into plant adaptive strategies to mining-associated disturbances through their spatial distribution patterns. In this study, three typical desert shrubs, including Haloxylon ammodendron, Nitraria tangutorum, and Reaumuria songarica, were investigated in the Zhundong open-pit coal mining area of Xinjiang, China. Sampling transects were established in four directions, each comprising seven distance gradients, to collect branch and leaf samples for the analysis of NSC and its components, soluble sugars (SS), and starch (ST).

Results

The results revealed significant species- and organ-specific patterns in NSC dynamics. Variability was generally higher in branches than in leaves for NSC, SS, and ST, with R. songarica exhibiting the lowest overall variability. Spatially, NSC in branches and leaves exhibited pronounced species- and organ-specific differentiation along both directional and distance gradients. Branch NSC peaked upwind (N) in H. ammodendron and downwind (SE1; southeast transect 1) in R. songarica, reflecting their divergent adaptive strategies to heterogeneous mining environments. Furthermore, N. tangutorum consistently displayed higher NSC and component concentrations in both organs compared to the other species, while leaves universally contained higher concentrations than branches. Redundancy analysis and hierarchical partitioning identified the primary environmental predictors of NSC variation (based on the measured soil-derived variables) as heavy metals (Cr) for H. ammodendron, soil pH and nitrogen for N. tangutorum, and water and nutrient availability for R. songarica. Overall, branch NSC was mainly associated with heavy metals, whereas leaf NSC was more closely associated with water and nitrogen, exhibiting particular sensitivity to nitrogen availability. Importantly, this study elucidates the species- and organ-specific carbon allocation strategies of desert shrubs across mining-associated soil environmental gradients, providing physiological and ecological insights for vegetation restoration and ecological rehabilitation in arid open-pit mining regions.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-026-08238-x.

Introduction

Coal has long served as the cornerstone of China’s energy security, maintaining a dominant position in primary energy consumption [1, 2]. Owing to its high resource recovery efficiency, operational safety, and large-scale production capacity, open-pit mining has become the predominant method of coal extraction nationwide [3, 4]. With the westward shift of China’s energy-development strategy, Xinjiang, one of the nation’s most coal-rich regions, has gradually emerged as a key energy supply base [5, 6]. However, this region is characterized by arid and ecologically fragile desert ecosystems, where extensive surface mining has severely disrupted the inherently sensitive environment, triggering a cascade of ecological degradation and functional decline [7]. Previous studies have shown that mining activities accelerate dust deposition and heavy-metal accumulation in surrounding environmental media [8, 9], disturb groundwater balance and degrade soil structure and function [10], ultimately reducing community diversity and ecosystem stability [11, 12]. Although research on the ecological impacts of mining has gained significant momentum, emphasis has been placed on changes in soil physicochemical properties and community composition. Accordingly, the physiological mechanisms that enable individual plants to regulate internal carbon allocation and metabolic processes, thereby maintaining survival and function under multiple, interacting stresses, remain largely unresolved.

At the individual plant level, non-structural carbohydrates (NSC), comprising soluble sugars (SS) and starch (ST), represent a core physiological currency of carbon allocation and environmental adaptation. They dynamically shift between immediate energy provision and long-term carbon storage, thereby maintaining carbon balance and enhancing plant resilience under environmental stress [13, 14]. Soluble sugars not only act as respiratory substrates but also participate in osmotic regulation, signal transduction, xylem embolism repair, and reactive oxygen species (ROS) scavenging [15, 16]. Conversely, starch provides energy for metabolic processes through hydrolysis when carbon supply becomes limited, sustaining cellular homeostasis and promoting tissue recovery [17]. Consequently, both the concentration and the inter-organ distribution of NSC are key metrics for assessing the carbon source–sink balance, stress resistance, and survival strategies of plants [18, 19].

Notably, it is now understood that the responses of NSC to multifactorial stress are highly complex, resulting not only from the intensity of individual environmental factors but also from their interactive effects. In open-pit mining areas and surrounding ecosystems, multiple stressors often coexist and overlap, jointly shaping the physiological patterns of NSC dynamics in plants. Specifically, under drought stress, water availability is typically the primary limiting factor for plant growth, and its fluctuation can substantially modify plant responses to other stresses. For instance, severe drought commonly constrains photosynthetic carbon assimilation and leads to a decline in total NSC concentrations, whereas moderate water deficits often induce an increase in SS to enhance osmotic regulation and hydraulic safety, displaying a nonlinear response pattern [20]. Furthermore, changes in soil nutrients and pH can further regulate the balance between carbon and nitrogen metabolism. For instance, soil alkalization reduces nitrogen availability and inhibits structural growth, while simultaneously altering the allocation of assimilated carbon between growth and storage, thereby influencing the composition and accumulation dynamics of NSC [21, 22]. Meanwhile, previous studies have shown that dust deposition, a typical stressor in mining environments, can suppress carbon acquisition through light obstruction and stomatal blockage [23–25], while its alkaline constituents can modify cellular metabolism, inducing specific NSC accumulation patterns [26]. The interplay becomes more complex under concurrent heavy-metal stress. While exposure to Cd, As, or Cr can result in elevated SS and reduced ST levels [27, 28], in certain species such as Malus spp. rootstocks, Cd stress has been reported to cause contrasting effects, with higher starch but lower soluble sugars [29]. Overall, plant NSC responses to combined stressors depend not only on the type, magnitude, and duration of each factor but also on their synergistic or antagonistic interactions, leading to highly context-dependent patterns of NSC variation [30–32].

Furthermore, marked variations in NSC responses have been documented across species and organs. Leaves typically maintain higher NSC concentrations to sustain continuous metabolic demands and osmotic regulation, whereas branches and roots serve primarily as storage and remobilization organs, hydrolyzing stored starch and reallocating the resulting sugars to growing or stressed tissues when required [33, 34]. Deciduous and evergreen species exhibit distinct carbon source-sink trade-offs at the onset of the growing season [35]. Systematic differences in NSC storage and allocation have been documented across plant life forms, with the highest concentrations observed in herbs, followed by shrubs, and then trees [18]. Under environmental stresses such as drought, where photosynthesis is constrained and carbon balance becomes negative, plants with different functional strategies can adjust NSC storage and remobilization to buffer carbon starvation and sustain physiological activity [36, 37]. Collectively, these findings suggest that under the multifactorial stresses induced by open-pit mining, NSC likely act as a central physiological hub, maintaining plant function, facilitating recovery, and shaping adaptive strategies. Meanwhile, the divergent responses among species and organs indicate that NSC dynamics not only reflect a plant’s integrated adaptation to stress, but are also modulated by intrinsic factors such as functional group and organ specialization, ultimately resulting in distinct species- and organ-specific patterns.

The environmental consequences of mining operations are characterized by a distinct spatial heterogeneity. Dust emission, pollutant dispersion, and groundwater disturbance are commonly reported features of mining activities and can collectively generate directional and distance-dependent gradients radiating from the mine pit. Studies in the Hulunbuir open-pit coalfield have shown that mining-induced disturbances to soil moisture and nutrient content tend to recover with increasing distance from the pit [10], while heavy-metal concentrations exhibit a similar distance-decay pattern [38]. In the Baorixile open-pit mine, elemental correlations exhibit significant directional variation [39]. These spatial environmental templates may shape plant carbon allocation, particularly influencing the spatial distribution of NSC. Compared with trees or herbaceous species, desert shrubs possess unique morphological and ecological strategies. Their patchy distribution [40], dwarf stature, long-lived leaves, and deep-root systems [41] make them highly sensitive to near-surface environmental conditions [42]. Consequently, desert shrubs serve as ideal bioindicators for assessing environmental heterogeneity in mining-affected areas.

Building on the above background, this study focused on three typical desert shrubs, including Haloxylon ammodendron, Nitraria tangutorum, and Reaumuria songarica, distributed around the Zhundong open-pit coal mine in Xinjiang, as representative species. Systematic sampling transects were established across multiple directions and distances from the mine pit. The concentrations of NSC and their components, including soluble sugars (SS) and starch (ST), were determined in both branches and leaves (or assimilating shoots), while plant morphological traits (height and crown width), soil physicochemical properties, and concentrations of multiple heavy metals (including Cr, Co, and Mn) were measured simultaneously. Specifically, this study aimed to address three scientific questions: (1) How NSC concentrations in branches and leaves differ spatially along distance and directional gradients; (2) How interspecific and inter-organ variability in NSC responds to these environmental gradients; and (3) Which environmental factors predominantly control the spatial variation of NSC. Through these analyses, we aim to elucidate the ecological adaptation strategies of desert shrubs under combined environmental stresses, focusing on carbon allocation and physiological regulation, thereby providing a scientific basis and practical reference for ecological restoration and vegetation reconstruction in arid-region open-pit mining areas.

Materials and methods

Study area

The present study was conducted near the No. 2 open-pit coal mine of the Jiangjun Gobi in the Zhundong Coalfield, Qitai County, Xinjiang, China (44°23′04″–44°38′05″ N, 90°05′31″–90°26′56″ E). The region lies in a mid-temperate continental arid to semi-arid climatic zone. The landscape consists of an eroded plain terrain with residual hills, with elevations ranging from 537 to 616 m and a gentle slope from southeast to northwest. The mean annual temperature is 5.4 °C, with a monthly mean of − 18.9 °C in January and 25.1 °C in July. Mean annual precipitation is only 269.4 mm, while potential evaporation reaches approximately 2141 mm, indicating severe water deficits. The site experiences a prevailing northwesterly wind, with an average speed of 2.9 m·s⁻¹. The dominant soil types are grey-brown desert soils and gypsum-bearing grey-brown desert soils, both of which are strongly affected by aeolian erosion and desertification. There are no perennial water bodies in the study area; runoff is primarily derived from sporadic rainfall events, forming seasonal flows that generally drain northwestward or accumulate in depressions, where evaporation produces micro-landforms such as silt pans and polygonal cracking surfaces. The vegetation is dominated by extremely xerophytic shrubs, with major community-forming species including H. ammodendron, H. persicum, Reaumuria songarica, Nitraria tangutorum, Calligonum mongolicum, and Atraphaxis spinosa.

Experimental design and sample collection

Plant sampling

Field investigations and sample collections were conducted during the peak growing season (June–August 2024), when foliage (or assimilating shoots) was fully developed and no visible senescence was observed at the time of sampling. Based on preliminary surveys, four transects (corresponding to four directions: north (N), west (W), southeast 1 (SE1), and southeast 2 (SE2)) were selected from eight pre-established routes around the open-pit mine for this study. Along each transect, seven sampling sites were established at distances of 0 km, 0.6 km, 1.5 km, 3.5 km, 5.5 km, 7.5 km, and 10 km from the nearest accessible point around the open-pit mine (Fig. 1). At each site, three large plots measuring 30 m × 30 m were established, with plot area adjusted slightly according to the minimum-area principle of the local shrub community. Within each large plot, two shrub quadrats (5 m × 5 m) were set along the diagonal corners and labeled as S1 and S2, respectively.

Fig. 1.

Sampling layout and target plant species in the Zhundong open-pit coal mining area, Xinjiang. (a) Geographic location of the study area and soil sampling design. The panels from top to bottom show: the regional position of the study area within China and Xinjiang; the layout of four sampling directions (N, north; W, west; SE1, southeast 1; SE2, southeast 2) and seven distance gradients around the open-pit mine; and a field view of the mine pit with the arrangement of shrub quadrats (S1, S2) and soil sampling locations in relatively bare-ground patches near the center of each 30 m × 30 m plot (0–10 cm). Soil samples were collected using a five-point quincunx (plum-blossom) pattern, and soils for heavy-metal analysis were collected using a pre-cleaned stainless-steel corer to minimize metal contamination. (b) Target desert shrub species and sampled organs. The study focused on three dominant desert shrubs, including Haloxylon ammodendron, Nitraria tangutorum, and Reaumuria songarica. The right-hand frames indicate the branch and leaf sampling positions for each species used in the analysis. Basemap approval no.: GS (2024) 0650

A total of 28 sampling sites were established, with shrub samples successfully collected from 20 sites. Statistical analysis of the collected samples indicated that H. ammodendron, N. tangutorum, and R. songarica were the dominant shrub species in the study area, while other species were sparsely distributed. Therefore, these three dominant species were selected as the focal species for this study. From each quadrat, where available, one mature and healthy individual of each target species was sampled to ensure spatial representation across the site. In cases where a species was absent from a given quadrat due to patchy distribution, additional individuals were sampled from other quadrats within the same site to maintain a consistent sample size of six replicates per species per site. For each sampled individual, plant height (H) and crown width (CW) were measured. From each sampled shrub, perennial lignified branches were clipped from the mid-canopy height on the southern exposure, and leaves (or assimilating shoots) were simultaneously collected. For H. ammodendron, “leaf” samples refer to current-year, green, photosynthetically active assimilating shoots (photosynthetic branchlets), which are the functionally equivalent foliar organs because true leaves are highly reduced; for N. tangutorum and R. songarica, fully expanded mature leaves were collected. All samples were placed in labeled paper bags and transported to the laboratory for the determination of NSC concentrations.

Field sampling was conducted in open desert areas surrounding the No. 2 open-pit coal mine of the Jiangjun Gobi in the Zhundong Coalfield, Qitai County, Xinjiang, China. According to local regulations, collecting small amounts of branches and leaves from common native shrub species in this region does not require a specific permit. All sampling activities were conducted with the knowledge and agreement of the local land management authorities. The plant species studied—H. ammodendron, N. tangutorum, and R. songarica— are neither protected nor listed in the Appendices of the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES). Species identification was strictly based on diagnostic morphological characters described in the Flora of China. No voucher specimens were deposited because all collected plant materials were immediately oven-dried and ground for biochemical analyses and therefore could not be preserved as long-term specimens.

Soil sampling

Soil sampling was conducted concurrently with plant collection. At each sampling site, three plot-level soil samples were collected (one per 30 m × 30 m plot). Within each plot, five subsamples were systematically collected from relatively sparsely vegetated patches (0–10 cm depth) using a quincunx (plum-blossom) pattern, while avoiding obvious disturbance sources such as ant nests and litter piles. Soils were collected using a pre-cleaned stainless-steel corer, and the tool was thoroughly rinsed with deionized water and air-dried between plots and sites to avoid cross-contamination [43]. Immediately after collection, samples were placed in sealed polyethylene bags, kept at 4 °C, and transported to the laboratory as soon as possible.

Measurement of variables

Three distinct categories of variables were measured in this study, including plant NSC components, morphological traits, and soil physicochemical and heavy metal indicators (Table S1). Physiological traits related to gas exchange and water relations (e.g., net photosynthetic rate, stomatal conductance, transpiration, and plant water status) were not measured because these variables are highly sensitive to short-term microclimatic fluctuations and require strictly standardized measurement windows across sites, which is difficult to implement in our landscape-scale field design. Therefore, NSC metrics were used as time-integrated indicators of plant carbon status, and morphological traits (plant height, H; crown width, CW) were used as integrative proxies of long-term growth status. The specific measurement methods are described as follows.

Determination of NSC concentration

Stem and leaf (or assimilating shoot) samples were first oven-heated at 105 °C for 30 min to inactivate enzymes and then dried at 75 °C to a constant weight. The dried samples were ground into fine powder using a ball mill (MM400, Retsch, Haan, Germany) and sieved through a 60-mesh screen. For analysis, 0.1 g of the powder was accurately weighed into a 2 mL centrifuge tube, and SS and ST concentrations in the xylem and leaves (or assimilating shoots) were determined using the anthrone–sulfuric acid colorimetric method. Absorbance was measured at 620 nm using a multifunctional microplate reader (VarioSkan Flash, Thermo Scientific, USA). NSC concentration was calculated as the sum of SS and ST.

Determination of soil physicochemical properties

In the laboratory, soil moisture content (SMC) was determined individually for each of the five subsamples within each plot. Fresh weight (FW) was recorded immediately, and subsamples were oven-dried at 105 °C to constant weight to obtain dry weight (DW). SMC was calculated as SMC = [(FW − DW) / DW] × 100%. Plot-level SMC was expressed as the mean of the five subsamples, and the mean of the three plot-level values was used as the site-level SMC for subsequent analyses.

Except for the portion used for SMC determination, the remaining soil from the five subsamples within each plot was pooled and thoroughly homogenized to form one plot-level composite sample, which was passed through a 2 mm sieve to remove roots and coarse debris. The composite soil was divided into two subsamples: one portion was air-dried at room temperature and ground for the determination of soil physicochemical properties (pH, total nitrogen (TN), total phosphorus (TP), and available phosphorus (AP)), while the other portion was kept fresh at 4 °C for the determination of ammonium nitrogen (NH₄⁺–N) and nitrate nitrogen (NO₃⁻–N). For soil physicochemical properties and inorganic N, site-level values were represented by the mean of the three plot-level measurements, and plot-level raw data were retained to assess within-site spatial heterogeneity of soil variables.

Soil pH was measured potentiometrically in a 1:2.5 (w/v) soil-to-water suspension. TN content was determined using the Kjeldahl digestion method, and TP content was measured by the molybdenum–antimony colorimetric method. NO₃⁻–N and NH₄⁺–N were extracted with 2 mol L⁻¹ KCl solution, and AP was extracted with 0.5 mol L⁻¹ NaHCO₃ solution. The concentrations of nitrogen and phosphorus in the extracts were determined colorimetrically using a continuous flow analyzer (AA3, Seal Analytical, Jena, Germany), following the method described by Carter and Gregorich [44].

Determination of soil heavy metal contents

Heavy metal concentrations were determined in soil samples only and were used as environmental predictors; metal concentrations in plant tissues were not measured in this study. Soil sampling was conducted concurrently with plant sampling (as described in "Soil sampling" Section), and therefore corresponds to the same peak growing season period (June–August 2024). Air-dried, finely ground soil samples (0.1 g) were digested with a mixture of acids in accordance with the Chinese national standard method (HJ 832–2017). The digested solutions were filtered and analyzed by inductively coupled plasma mass spectrometry (ICP–MS; Agilent 7900, Agilent Technologies, USA) to determine the concentrations of chromium (Cr), manganese (Mn), cobalt (Co), nickel (Ni), copper (Cu), zinc (Zn), arsenic (As), strontium (Sr), molybdenum (Mo), and lead (Pb). To ensure analytical accuracy, procedural blanks, duplicate samples, and certified reference materials (GBW07401) were included for quality control in each batch [45]. Site-level heavy metal values were represented by the mean of the three plot-level measurements.

Data analysis

All abbreviations and measurement units for variables are listed in Table S1. The Shapiro–Wilk test for normality and Bartlett’s test for homogeneity of variances (package stats) were performed to evaluate data distribution characteristics. The response variables were log₁₀-transformed to approximate a normal distribution. To evaluate the effects of direction, distance, and species on NSC and its components in branches and leaves (or assimilating shoots), three-way interaction models (Direction × Distance × Species) were constructed. Type II analysis of variance (ANOVA; car package) was then applied to test the significance of main and interactive effects. Variation in NSC and its components between species and organs was quantified using the quartile coefficient of dispersion (QCD), a robust measure independent of normality and variance homogeneity assumptions.

To reveal spatial variation patterns of NSC and their components in branches and leaves across directional and distance gradients around the mine pit, data from the SE2 direction, which covered all combinations of species and distance, were selected for analysis. First, NSC levels and their components in branches and leaves were compared among the three species and across directions using ANOVA (stats package), followed by a least significant difference (LSD) post hoc test (agricolae package). At each spatial distance (0, 1.5, 3.5, 7.5, and 10 km), one-way ANOVA and the LSD test were used to assess interspecific differences in NSC and its components in branch and leaf organs. Besides, independent-sample t-tests (stats package) were conducted to examine branch–leaf differences within each species at each distance. To characterize variation trends of NSC and its components along spatial gradients, data for each species and organ were fitted with three models: linear, quadratic polynomial, and logarithmic regression [log₁₀(distance + 1) as the independent variable]. The optimal model was selected based on the corrected Akaike information criterion (AICc; MuMIn package).

Pearson correlation analysis (Hmisc package) was performed to evaluate relationships between biological factors (plant traits), environmental factors (soil physicochemical properties and heavy metals), and NSC and its components. Redundancy analysis (RDA, vegan package) was performed to elucidate the influence of environmental factors on the spatial distribution patterns of NSC. In addition, hierarchical partitioning (rdacca.hp package) [46] was employed to quantify the explanatory power and relative importance of biological and environmental factors in driving the variation of NSC and its components. All statistical analyses and visualizations were performed in R v4.3.0 (R Core Team, Vienna, Austria).

Transparency statement

All experimental procedures and sampling protocols were completed prior to any textual revision of this manuscript. The revisions made in this version were intended solely to improve descriptive clarity and transparency and do not represent any changes to the procedures as originally executed.

Results

Variation in branch and leaf NSC and its components among the three shrub species

The variation patterns of NSC and its components in branches and leaves across the three shrub species are shown in Fig. 2. Overall, all components showed greater variation in branches than in leaves, reflecting the higher physiological plasticity of carbon storage in branches. Interspecific differences were observed: N. tangutorum showed the highest variability in branch soluble sugar (BSS) and branch non-structural carbohydrate (BNSC) concentrations, whereas H. ammodendron exhibited the most significant variation in branch starch (BST). In contrast, leaf soluble sugar (LSS) and leaf non-structural carbohydrate (LNSC) concentrations varied most strongly in H. ammodendron, while N. tangutorum displayed the most significant variation in leaf starch (LST). R. songarica consistently exhibited the lowest variability across all organs and components, except for BSS, indicating a more stable carbon allocation strategy.

Fig. 2.

Branch and leaf NSC variation among three shrubs based on QCD

Spatial variation patterns of NSC and its components

Significant directional variation in branch and leaf NSC components was observed for all three shrub species (Fig. 3). All statistical comparisons were performed on log₁₀-transformed concentrations. In H. ammodendron, BSS, BST, and BNSC concentrations were significantly higher in the N direction than in W, SE1, and SE2 (P < 0.05) (Fig. 3a–c), with no significant differences detected among the latter three (P > 0.05). For N. tangutorum, BSS was highest in N and significantly greater than in SE2 (P < 0.05) (Fig. 3g); BST peaked in SE1 and was significantly higher than in W and SE2 (P < 0.05) (Fig. 3h). Besides, BNSC was lowest in W, significantly below all other directions (P < 0.05) (Fig. 3i). A significant spatial gradient was observed in R. songarica for BSS and BNSC, with concentrations highest in SE1, lowest in W, and intermediate in SE2 (P < 0.05) (Fig. 3m, o). BST in SE1 was also significantly higher than in all other directions (P < 0.05) (Fig. 3n). For leaves, H. ammodendron showed the highest LNSC concentrations in N and SE1, both significantly greater than in SE2 (P < 0.05) (Fig. 3f), whereas W did not differ significantly from the other directions (P > 0.05). In N. tangutorum, LSS and LNSC were highest in N and significantly greater than in SE2 (P < 0.05) (Fig. 3j, l), with W and SE1 showing intermediate values and no significant differences either between them or relative to N and SE2 (P > 0.05). In R. songarica, the highest LSS and LNSC were observed in SE1, significantly higher than in W and SE2 (P < 0.05) (Fig. 3p, r); concentrations in N were significantly higher than in SE2 (P < 0.05) but did not differ from W or SE1 (P > 0.05). LST in all three species (Fig. 3e, k,q), as well as LSS in H. ammodendron, showed no significant directional differences (P > 0.05) (Fig. 3d).

Fig. 3.

Directional variation in branch and leaf NSC and components among the three shrubs. (a–f) H. ammodendron, (g–l) N. tangutorum, and (m–r) R. songarica. Colors indicate sampling directions (W, west; SE1, southeast 1; SE2, southeast 2; N, north). Radial scales represent log₁₀-transformed mean concentrations (log₁₀[mg·g⁻¹ DW]). Uppercase letters denote significant differences among directions (P < 0.05)

The spatial variation patterns of NSC and their components along the distance gradient from the open-pit mine varied significantly across species and organs based on log₁₀-transformed concentrations (Fig. 4). In H. ammodendron, BSS, BST, and BNSC decreased significantly with increasing distance (P < 0.05; Fig. 4a–c), whereas no significant variation was observed in leaves (P > 0.05; Fig. 4d–f). In N. tangutorum, BSS and BNSC exhibited unimodal patterns, declining initially before increasing (P < 0.05; Fig. 4a, c), while BST remained stable (P > 0.05; Fig. 4b). In leaves, LSS declined significantly with distance (P < 0.05; Fig. 4d), while LST and LNSC displayed unimodal trends, increasing first and then decreasing (P < 0.05; Fig. 4e, f). In R. songarica, organ-level responses showed contrasting findings: BSS, BST, and BNSC showed no significant changes (P > 0.05; Fig. 4a–c), whereas all leaf components increased significantly (P < 0.05; Fig. 4d–f). To link the variation in total NSC to its components, the relationships between the response ratio (RR) of NSC and the RRs of soluble sugar and starch were assessed, as well as their relative contribution (Fig. S1).

Fig. 4.

Variation in branch and leaf NSC and components across a distance gradient. (a–c) Changes in branch concentrations of SS, ST, and NSC; (d–f) corresponding changes in leaves. Curves and shaded bands represent the best-fitting models selected according to the corrected Akaike information criterion (AICc). Coefficients of determination (R²) and significance levels of the fitted models are shown in each panel (log₁₀-transformed concentrations). Significance is indicated as *P < 0.05, **P < 0.01, ***P < 0.001

Interspecific and inter-organ differences in NSC and its component concentrations

Significant variation in SS, ST, and NSC concentrations (log₁₀-transformed) was observed along the SE2 direction, attributable to both species identity and plant organ across the sampled distances (Fig. 5). Overall, except for BST, N. tangutorum exhibited consistently higher levels of SS, ST, and NSC than H. ammodendron and R. songarica (P < 0.05; Fig. 5a–c). In branches, BSS concentrations in N. tangutorum were consistently highest across all distances (0, 1.5, 3.5, 7.5, and 10 km), significantly higher than those in H. ammodendron and R. songarica (P < 0.05), with a significant difference between the latter two species observed only at 7.5 km (P < 0.05; Fig. 5a). For BST, H. ammodendron showed the highest values at 0 km, significantly higher than R. songarica (P < 0.05; Fig. 5b), whereas N. tangutorum exhibited the highest BST at 7.5 km (P < 0.05); no significant interspecific differences were detected at 1.5, 3.5, or 10 km (P > 0.05). BNSC concentrations in N. tangutorum remained consistently highest across all distances (P < 0.05; Fig. 5c), with no significant differences between H. ammodendron and R. songarica (P > 0.05). In leaves, N. tangutorum also maintained significantly higher LSS, LST, and LNSC concentrations than the other two species at all distances (P < 0.05), except LSS at 7.5 km, where differences were not significant (Fig. 5a–c). Pairwise t-tests between organs revealed that, except for SS in N. tangutorum at 0 km and 10 km (not significant), SS, ST, and NSC concentrations were markedly higher in leaves compared with branches for all species (P < 0.01 or P < 0.001; Fig. 5a–c).

Fig. 5.

Interspecific and inter-organ comparisons of NSC and its components along the SE2 direction. (a–c) Changes in SS, ST, and NSC (log₁₀-transformed concentrations; mg·g⁻¹ DW) at different distances (0, 1.5, 3.5, 7.5, and 10 km). Data are shown as boxplots: boxes indicate the interquartile range (IQR), horizontal lines represent medians, whiskers extend to 1.5 × IQR, and dots denote outliers. Different colors represent species (H. ammodendron, N. tangutorum, and R. songarica), with paired adjacent boxes indicating branches (left) and leaves (right) of the same species. Symbols denote significant differences between branches and leaves according to t-tests (*P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant). Different lowercase and uppercase letters indicate significant interspecific differences within branches and leaves, respectively (P < 0.05; LSD post hoc test after Bonferroni correction)

Dominant drivers of NSC and their components

To identify the dominant factors driving variation in branch and leaf NSCs and their components among the three shrub species, correlation and ordination analyses were performed, integrating biological factors (H, CW; see Table S2) with soil physicochemical properties and heavy metal elements. Furthermore, hierarchical partitioning and stepwise regression models were employed to quantitatively evaluate the relative contributions of each factor on NSC and their components.

Overall, correlation analyses indicated strong coordination among NSC components within individual organs. However, the degree of branch–leaf coupling and the relationships between NSC components and environmental or plant traits variables were highly species-specific, defining distinct organ‑ and species‑dependent correlation patterns (Fig. 6a–c).

Fig. 6.

Branch and leaf NSC correlations with plant, soil, and metal factors. Pearson correlation heatmaps illustrate the relationships between branch and leaf NSC components (BSS, branch soluble sugar; BST, branch starch; BNSC, branch non-structural carbohydrate; LSS, leaf soluble sugar; LST, leaf starch; LNSC, leaf non-structural carbohydrate) and plant traits, soil physicochemical properties, and heavy metals. Analyses were conducted for (a) H. ammodendron, (b) N. tangutorum, and (c) R. songarica. Plant traits include height (H) and crown width (CW); soil physicochemical variables include soil moisture content (SMC), pH, nitrate nitrogen (NO₃⁻–N), ammonium nitrogen (NH₄⁺–N), total nitrogen (TN), total phosphorus (TP), and available phosphorus (AP); heavy metal elements include Cr, Mn, Co, Ni, Cu, Zn, As, Sr, Mo, and Pb. Colors represent Pearson correlation coefficients (r values), with red indicating positive and blue indicating negative correlations. Asterisks denote significance levels: * P < 0.05, ** P < 0.01, *** P < 0.001. Abbreviations and units are provided in Table S1

Significant positive correlations were observed among NSC components within both branches and leaves across all species (P < 0.01 or P < 0.001) (Fig. 6a–c). Specifically, in branches, BSS was significantly or highly significantly positively correlated with BST and BNSC, and BST was also positively correlated with BNSC (P < 0.05 or P < 0.001). Similarly, in leaves, LSS, LST, and LNSC were all significantly or highly significantly positively correlated with each other (P < 0.05 or P < 0.001). Correlation patterns between branches and leaves differed markedly among species. In this respect, in R. songarica, LSS showed significant or highly significant positive correlations with all branch components (BSS, BST, and BNSC) (P < 0.05 or P < 0.01) (Fig. 6c). In H. ammodendron, LST was significantly or highly significantly positively correlated with all branch components (P < 0.05 or P < 0.01), while LNSC was also positively correlated with BSS and BNSC (Fig. 6a). In R. songarica, LNSC was significantly or highly significantly correlated with BST and BNSC (P < 0.05 or P < 0.01) (Fig. 6c). Notably, no significant correlations were detected between any branch and leaf components in N. tangutorum (P > 0.05) (Fig. 6b).

Assessment of plant morphological traits revealed that neither plant height (H) nor crown width (CW) showed significant correlations with any NSC components in branches (BSS, BST, BNSC) or leaves (LSS, LST, LNSC) of H. ammodendron and R. songarica (P > 0.05) (Fig. 6a, c). In contrast, in N. tangutorum, both LSS and LNSC were significantly positively correlated with H and CW (P < 0.05), whereas all other components showed no significant relationships with these traits (P > 0.05) (Fig. 6b).

The correlations between soil physicochemical properties and NSC components in branches and leaves exhibited distinct species- and organ-specificity (Fig. 6a–c). SMC showed a strong positive correlation with all branch components (BSS, BST, and BNSC) of H. ammodendron (P < 0.001) (Fig. 6a). In N. tangutorum, it was positively correlated only with LSS (P < 0.05) (Fig. 6b), whereas no significant correlations were observed for other components or for any branch or leaf variables of R. songarica (P > 0.05) (Fig. 6c). Soil pH was positively correlated with LSS in H. ammodendron (P < 0.05) (Fig. 6a). In contrast, it showed strong negative correlations with most branch and leaf components of N. tangutorum (except LST; P < 0.05–0.001) (Fig. 6b). In R. songarica, pH showed a strong negative correlation with BSS and BNSC (P < 0.01), but a positive correlation with LST (P < 0.05) (Fig. 6c). NO₃⁻–N was highly negatively correlated with LSS and LNSC in H. ammodendron and R. songarica (P < 0.01 or P < 0.001) (Fig. 6a, c), but positively correlated with BST of H. ammodendron and with LSS and LNSC of N. tangutorum (P < 0.05) (Fig. 6a, b). NH₄⁺–N was strongly positively correlated with BSS and BNSC of N. tangutorum and R. songarica (P < 0.01 or P < 0.001) (Fig. 6b, c), but showed no significant correlations with any branch components of H. ammodendron (P > 0.05) (Fig. 6a). Conversely, NH₄⁺–N exhibited a strong negative correlation with LST and LNSC of H. ammodendron and with LSS and LNSC of R. songarica (P < 0.05 or P < 0.01) (Fig. 6a, c), while no significant correlations were found with any leaf components of N. tangutorum (P > 0.05) (Fig. 6b). TN showed no significant correlations with any branch components in the three species (Fig. 6a–c). It exhibited a strong negative correlation with all leaf components of N. tangutorum (P < 0.05 or P < 0.001) (Fig. 6b), and with LST and LNSC of H. ammodendron and LSS of R. songarica (P < 0.05, 0.01, or 0.001) (Fig. 6a, c), but positively correlated with LST of R. songarica (P < 0.05) (Fig. 6c). TP demonstrated a strong positive correlation with all branch components of H. ammodendron (P < 0.001) (Fig. 6a), but a strong negative correlation with BST of N. tangutorum (P < 0.01) and with LSS and LNSC of R. songarica (P < 0.01 or P < 0.001) (Fig. 6b, c), while showing no significant correlations with most other components (P > 0.05) (Fig. 6a–c). Finally, AP showed significant but relatively weak correlations with LSS of N. tangutorum and R. songarica (P < 0.05 or P < 0.01) (Fig. 6b, c), but demonstrated a stronger negative correlation with LST of R. songarica (P < 0.001) (Fig. 6c).

Significant species- and organ-specific differences were also observed in the correlations between soil heavy metal elements and NSC components (Fig. 6a–c). Cr and As were negatively correlated with all branch components (BSS, BST, and BNSC) of H. ammodendron (P < 0.05 or P < 0.01) (Fig. 6a), whereas Mn, Co, Cu, Zn, and Sr were positively correlated with these branch components (P < 0.05, P < 0.01, or P < 0.001). In contrast, in N. tangutorum, Mn, Ni, Cu, As, and Pb were negatively correlated with BSS and BNSC (P < 0.05 or P < 0.01) (Fig. 6b), whereas Sr showed positive correlations with all branch components (P < 0.05, P < 0.01, or P < 0.001). In R. songarica, BST exhibited positive correlations with multiple heavy metal elements, including Mn, Co, Ni, Cu, Zn, Sr, Mo, and Pb (P < 0.01 or P < 0.001) (Fig. 6c). Overall, correlations between leaf components and soil heavy metals were weaker, with the only notable pattern observed in N. tangutorum leaves, where LSS showed positive correlations with several heavy metals (Co, Ni, Zn, Sr, and Mo) (P < 0.05, P < 0.01, or P < 0.001) (Fig. 6b).

To further assess the effects of key factors, three-way ANOVA revealed that direction, distance, and species all exerted significant effects on the concentrations of NSC and their components (SS; ST) in both branches and leaves (P < 0.05), although the strength of these effects varied among organs and components (Table 1). Specifically, BST was the only component not significantly influenced by distance (P > 0.05), whereas all other components exhibited significant main effects of the three factors (P < 0.05). Interaction analysis showed that, in branches, neither BSS nor BNSC responded significantly to the three-way interaction among direction, distance, and species (P > 0.05). In contrast, in leaves, LSS, LST, and LNSC were all significantly affected not only by the main effects but also by their two-way and three-way interactions (P < 0.05).

Table 1.

Three-way ANOVA of factors affecting NSC and components in branches and leaves

Variables	SS				ST				NSC
	df	F	P	ηp²	df	F	P	ηp²	df	F	P	ηp²
Branch
Direction	3	22.481	***	0.225	3	17.264	***	0.188	3	25.910	***	0.263
Distance	6	4.477	***	0.104	6	1.809	ns	0.046	6	3.304	**	0.083
Species	2	173.794	***	0.600	2	59.823	***	0.349	2	145.817	***	0.572
Direction × Distance	10	5.485	***	0.191	10	3.672	***	0.141	10	5.304	***	0.196
Direction × Species	6	4.694	***	0.108	6	5.248	***	0.124	6	4.944	***	0.120
Distance × Species	11	3.591	***	0.145	11	3.294	***	0.140	11	3.782	***	0.160
Direction × Distance × Species	9	1.267	ns	0.047	9	2.104	*	0.078	9	1.210	ns	0.048
Residual	232				223				218
Leaf
Direction	3	20.962	***	0.215	3	6.797	***	0.081	3	22.131	***	0.232
Distance	6	11.700	***	0.234	6	3.384	**	0.081	6	10.682	***	0.226
Species	2	69.835	***	0.378	2	576.545	***	0.834	2	359.804	***	0.766
Direction × Distance	10	6.462	***	0.219	10	11.939	***	0.342	10	11.997	***	0.353
Direction × Species	6	5.718	***	0.130	6	10.755	***	0.219	6	10.620	***	0.225
Distance × Species	11	7.188	***	0.256	11	5.074	***	0.195	11	8.607	***	0.301
Direction × Distance × Species	9	5.890	***	0.187	9	2.561	**	0.091	9	2.988	**	0.109
Residual	230				230				220

SS Soluble sugar, ST Starch, NSC Non-structural carbohydrate. ηp², partial eta squared (effect size), calculated as ηp² = (df_effect × F) / (df_effect × F + df_error). Significance levels: *** P < 0.001, ** P < 0.01, * P < 0.05, ns = not significant

To further examine species- and organ-specific responses of branch and leaf NSC components to biotic and abiotic drivers, RDA was performed separately for each of the three desert shrub species. The results showed that both biotic and abiotic factors significantly accounted for the variation in branch and leaf NSC components across all species (H. ammodendron: P = 0.012; N. tangutorum: P = 0.001; R. songarica: P = 0.001), and distinct response patterns were evident among species and between organs in the ordination space (Fig. 7; Table S3). In H. ammodendron (Fig. 7a–c), the first two RDA axes together accounted for 94.8% of the total variation (RDA1 = 63.8%, RDA2 = 31.0%), with heavy metal Cr as the dominant contributor to RDA1 (54.96%) and TN to RDA2 (43.48%). Branch NSC components (BSS, BST, BNSC) were positively correlated with AP but negatively associated with Cr (P = 0.006), whereas leaf components (LSS, LST, LNSC) were negatively correlated with TN and NH₄⁺–N. In N. tangutorum (Fig. 7d–f), the first two RDA axes accounted for 91.4% of the total variation (RDA1 = 57.5%, RDA2 = 33.9%), with soil pH and TN as the dominant contributors to RDA1 (32.74%) and RDA2 (28.23%), respectively. Branch NSC components were strongly negatively correlated with CW and soil pH (P = 0.001), while leaf components were negatively correlated with TN and NH₄⁺–N (P < 0.01). In R. songarica (Fig. 7g–i), the first two RDA axes accounted for 83.6% of the variation (RDA1 = 60.1%, RDA2 = 23.5%), with SMC and NH₄⁺–N as the dominant contributors to RDA1 (27.61%) and RDA2 (47.91%), respectively. Branch NSC components were positively correlated with SMC and NH₄⁺–N but negatively correlated with pH (P = 0.001), whereas in leaves, NH₄⁺–N was positively correlated with LST and negatively correlated with LSS and LNSC. Collectively, the above results demonstrate pronounced organ-specific responses of branch and leaf NSC components to biotic and abiotic drivers across the three shrub species, with leaf components overall exhibiting higher sensitivity to nitrogen availability.

Fig. 7.

Redundancy analysis of branch and leaf NSC components and environmental factors. (a–c) H. ammodendron; (d–f) N. tangutorum; (g–i) R. songarica. Panels (a, d, g) display the ordination of branch (BSS, branch soluble sugar; BST, branch starch; BNSC, branch non-structural carbohydrate) and leaf (LSS, leaf soluble sugar; LST, leaf starch; LNSC, leaf non-structural carbohydrate) components constrained by environmental factors. Arrows are color-coded by variable type: blue represents plant traits (H, plant height; CW, crown width), green denotes soil physicochemical properties (SMC, soil moisture content; NO₃⁻–N, nitrate nitrogen; NH₄⁺–N, ammonium nitrogen; TN, total nitrogen; TP, total phosphorus; AP, available phosphorus), and pink indicates heavy metal elements (Cr, Mn, Co, Ni, Cu, Zn, As, Sr, Mo, and Pb). Different symbols represent species: circles = H. ammodendron, squares = N. tangutorum, and triangles = R. songarica. Different background colors represent sampling directions: green (W, west), orange (N, north), pink (SE1, southeast 1), and purple (SE2, southeast 2); the corresponding confidence ellipses represent the 95% confidence intervals, indicating the spatial distribution of samples from each direction in the RDA ordination space. Panels (b, e, h) and (c, f, i) show the top five environmental variables contributing to the RDA1 and RDA2 axes, respectively, with the red dashed line indicating the mean contribution of all variables included in the analysis. The significance of the full RDA models and the detailed contributions of each variable are provided in Table S3. Abbreviations and units are provided in Table S1

To quantify the independent effects of biotic and abiotic factors on NSC and its components, hierarchical partitioning and stepwise multiple regression analyses were conducted, based on the overall patterns revealed by RDA (Fig. 8). The results suggesting distinct species- and organ-specific responses, indicating divergent drivers of NSC variability. In H. ammodendron, branch NSC variation was predominantly explained by heavy metals (Fig. 8a–c), whose independent contributions ranged from 65% to 88%, far exceeding those of soil factors (12–35%). Hierarchical partitioning precisely quantified statistical contributions underlying the RDA1 axis: Cr was identified as the dominant negative predictor and key limiting factor for all branch components (highly significant independent effect, P < 0.001), while AP, which was positively associated with RDA1, showed significant positive independent contributions (P < 0.01–0.001). In contrast, leaf NSC variation was mainly explained by soil factors (79–90% contribution) (Fig. 8d–f), with good model fits (R² = 0.28–0.45), consistent with the RDA finding that leaf components were closely associated with soil nitrogen (TN, NH₄⁺–N, NO₃⁻–N) and moisture. In N. tangutorum, branch NSC variation exhibited the highest model fits (R² = 0.39–0.59) (Fig. 8g–i), with the variance attributed to soil properties (37–68%), plant traits (23–35%), and heavy metals (6–29%). Specifically, CW and soil pH showed significant negative associations (P < 0.05–0.001), whereas NH₄⁺–N and Cr emerged as significant positive predictors (P < 0.05–0.001), consistent with the RDA results. In contrast, leaf NSC variation was predominantly explained by soil properties (independent contribution ≥ 97%) (Fig. 8j–l), with moderate model fits (R² = 0.46–0.53). Specifically, NH₄⁺–N, TN, AP, and pH all exhibited significant negative correlations (P < 0.01–0.001). In R. songarica, branch models showed relatively low to moderate fits (R² = 0.15–0.43) (Fig. 8m–o), with variation primarily explained by soil properties (70–81%) and lesser contributions from heavy metals (19–28%). Among soil variables, NH₄⁺–N was strongly positively associated with all branch components (P < 0.001), whereas Cr showed strong negative associations with BSS and BNSC (P < 0.001) (Fig. 8m, o). Leaf models exhibited higher fits (R² = 0.34–0.54) (Fig. 8p–r), although the dominant drivers differed among components: LSS and LST were mainly explained by soil factors (55–100%) (Fig. 8p, q), whereas LNSC was primarily explained by heavy metals (59%) (Fig. 8r). Specifically, LSS and LNSC were positively associated with SMC and NO₃⁻–N but showed strong negative associations with Cr (Fig. 8p, r), while LST showed significant (P < 0.05) or highly significant (P < 0.001) negative correlations with NH₄⁺–N and AP (Fig. 8q). Collectively, these results reveal that the drivers of NSC and its component variation exhibit strong organ specificity across species, with leaf NSC components generally showing higher sensitivity to nitrogen availability (Fig. 8d–f, j–l, p–r).

Fig. 8.

Variance decomposition of branch and leaf NSC components in three desert shrub species. (a–f) H. ammodendron; (g–l) N. tangutorum; (m–r) R. songarica. Columns represent organs (branch vs. leaf) and NSC components (SS, soluble sugar; ST, starch; NSC, non-structural carbohydrate).The left panels show the independent explanatory power of environmental variables derived from hierarchical partitioning (green: plant traits; blue: soil physicochemical properties; purple: heavy metal elements; adjusted R² values are shown above bars). The right panels present standardized regression coefficients from stepwise-selected multiple linear regression models (error bars indicate standard errors). Significance levels: * P < 0.05, ** P < 0.01, *** P < 0.001

Overall, the dominant predictors of NSC variation differed among species and organs based on the measured soil-derived variables (Fig. 8). In branches, soil heavy metals, especially Cr, accounted for the largest independent fraction of variation in H. ammodendron, whereas in N. tangutorum the independent fractions were distributed among soil properties, plant traits, and metals, and in R. songarica branches the largest fractions were attributable to soil factors with additional contributions from specific metals. In leaves, NSC variation was more consistently associated with soil moisture and nitrogen availability, although the relative importance of soil versus metal predictors varied among components and species. Collectively, these results highlight pronounced organ- and species-specific statistical association patterns of NSC variability across the three shrubs.

Discussion

Species- and organ-specific variability of NSC among the three desert shrubs

NSC function as central substrates in plant metabolism, serving not only as primary energy reserves but also as key regulatory intermediates under environmental stress. Our findings revealed pronounced species- and organ-specific differentiation in NSC variability among the three desert shrubs. R. songarica exhibited the lowest variability across all organs and NSC components except BSS, which reflects a typically conservative survival strategy. Such stability may be attributed to its well-developed deep-root system, which enables sustained access to subsurface water and nutrients [47], thereby maintaining strong internal homeostasis under heterogeneous desert conditions. In contrast, N. tangutorum showed the greatest variability in BSS and BNSC, consistent with an opportunistic physiology that enables rapid osmotic adjustment in response to environmental fluctuations [48]. Meanwhile, the high variability in LSS and BST in H. ammodendron indicates a strong dependence on water availability, characteristic of deep-rooted species. This pattern reflects the dynamic adjustment of soluble sugar and starch storage in response to changing moisture conditions.

At the organ level, all three shrub species exhibited greater variability in NSC and its components in branches than in leaves (Fig. 2), a pattern consistent with global observations [18] and indicative of the typical “source–sink” functional differentiation within plants. It is now understood that branches, which comprise woody tissues, function as long-term carbon reservoirs and play a pronounced buffering role in NSC dynamics. Studies on temperate tree species have shown that branches contribute approximately 40% of total plant NSC storage, far exceeding the 5% stored in leaves [33]. Interestingly, this structural advantage enables branches to mobilize reserves under carbon limitation, sustaining essential metabolism, and to actively accumulate carbon when supply exceeds demand, demonstrating substantial physiological plasticity. The inherently slow turnover rate of xylem-borne NSC further enhances its buffering capacity against environmental fluctuations. Notably, woody tissues also play important roles in water transport and mechanical support, in addition to serving as a carbon storage medium. Given that these functions impose distinct structural and resource requirements, natural selection may favor specific trait combinations that optimize trade-offs among storage, transport, and mechanical roles [49]. Recent evidence suggests that species with higher starch storage in sapwood tend to adopt dehydration-avoidance strategies, and that drought-deciduous species generally exhibit greater dehydration resistance than evergreens [49]. Taken together, these findings provide new insights into the functional coordination between carbon storage and water transport in woody organs.

In contrast to branches that function as long-term carbon reserves, leaves act as photosynthetic “source” organs, maintaining their NSC concentrations within a relatively stable dynamic range [50]. This homeostatic feature arises from the crucial metabolic role of leaves as the primary sites of photosynthetic assimilate synthesis, where stable substrate concentrations are required to sustain continuous carbon assimilation. In the peatland shrub Dasiphora fruticosa, soluble sugar concentrations in leaves have been reported to consistently exceed those in roots and stems [51]. Consistent with this pattern, our results showed that although leaf NSC concentrations were generally high across the three desert shrubs (Fig. 5), their overall variability was lower than that of branches (Fig. 2), corroborating the role of leaves as metabolic “source” organs. The homeostatic regulation of leaf NSC is achieved through a dual mechanism: the rapid responsiveness of photosynthesis to environmental cues, and active source-sink regulation. Within this framework, leaves prioritize assimilate export during periods of high carbon demand, while moderate accumulation during surplus periods acts as a buffer against short-term environmental variability. The complementary organ-level strategies of leaf “homeostatic maintenance” and branch “plastic adjustment” together form the physiological foundation for shrub adaptation to desert heterogeneity, representing a vital survival strategy under open-pit mining disturbance.

Spatial patterns of NSC in three desert shrubs under the influence of open-pit coal mining

Large-scale open-pit mining in the Zhundong Coalfield, Xinjiang, is associated with pronounced environmental gradients around mine sites. Coal dust dispersion and heavy-metal deposition have been widely reported as common features of such systems [23]. In this study, we interpret NSC spatial patterns primarily in relation to the measured soil variables (e.g., soil physicochemical properties and soil metal loads). As a key indicator of plant carbon balance, NSC offer insights into adaptive strategies of plants in response to mining-associated disturbances through their spatial distribution patterns [52].

The results of this study revealed significant directional differences, suggesting that the three shrub species exhibited distinct adaptive strategies. In H. ammodendron, BSS, BST, and BNSC concentrations were highest in the N direction (upwind) and significantly exceeded those in other directions (Fig. 3). This pattern was consistent with wind-shaped spatial heterogeneity in mining landscapes, where dust deposition and soil heavy-metal loads often show directional contrasts. Zhang et al. observed that natural dust deposition in this region was mainly concentrated within 8–10 km along both upwind and downwind corridors [53]. Tang et al. further confirmed that the mine interior and downwind sectors were the primary zones of heavy-metal contamination, with soil concentrations of Cr, Pb, and As exceeding standard limits in more than 70% of the samples [54]. The elevated BSS, BST, and BNSC concentrations in H. ammodendron along the N direction observed in the present study may reflect comparatively lower soil heavy-metal loads and more favorable soil conditions at upwind sites, consistent with wind-shaped spatial heterogeneity documented in mining landscapes. In contrast, R. songarica exhibited the highest BSS, BST, and BNSC concentrations in the SE1 direction (the main downwind sector and a zone with relatively higher soil metal loads in this study), suggesting a species-specific allocation response under metal-associated soil gradients. Previous studies have reported high Hg enrichment in R. songarica tissues, which is cited here only as background context rather than evidence from our samples [54].

The responses of NSC and its components along the distance gradient further revealed species-specific adaptive differentiation among the three desert shrubs. In H. ammodendron, branch NSC and its component concentrations declined linearly and significantly with increasing distance from the mine, indicating strong sensitivity to the disturbance gradient and suggesting its potential as a biological indicator of mining stress. This finding is consistent with the results of Wang et al., who reported moderate heavy-metal contamination (Cr, Cu, Zn, Hg) and a high potential ecological risk in mining-area soils [55], providing an environmental context for the distance‐dependent decline of NSC in H. ammodendron. In contrast, R. songarica exhibited a clear organ-level functional differentiation: BSS, BST, and BNSC remained relatively stable, whereas LSS, LST, and LNSC increased significantly with distance (Fig. 4), consistent with its conservative strategy of preferentially allocating limited resources to photosynthetic organs to maintain carbon supply. Wang et al. reported greater capacity for metal enrichment and translocation of Cr, Cu, Zn, and Hg in R. songarica than in H. ammodendron in local mine-affected vegetation [55], providing additional background on interspecific differences. Tang et al. also noted that despite exposure to heavy‐metal contamination, R. songarica maintained normal physiological metabolism through internal regulation, suggesting that this species can maintain physiological functioning under metal-contaminated conditions [54]. The nonlinear responses of branch and leaf NSC in N. tangutorum suggest the activation of a compensatory carbon‐storage mechanism under moderate stress, a pattern consistent with findings from various stress‐response studies [17, 52, 56].

Dominant drivers of NSC variation

The drivers of NSC variation showed clear species-specific differences among the three desert shrubs. In H. ammodendron, branch NSC variation was most strongly associated with soil heavy-metal indicators, with Cr emerging as a key predictor across analyses. Both redundancy analysis (Fig. 7) and hierarchical partitioning (Fig. 8) revealed a significant negative correlation between Cr and branch NSC, with heavy-metal factors independently accounting for 65–88% of its variation. This inhibition is consistent with documented toxic effects of Cr on photosynthetic metabolism, which can involve disruption of the PSII reaction center, the suppression of RuBisCO activity, and the interference with electron transport and carbon assimilation [57, 58]. Similarly, Stobrawa and Lorenc-Plucińska observed a marked reduction in fine-root NSC of Populus nigra L. growing near a copper smelter [59], whereas the heavy-metal-tolerant halophyte Sesuvium portulacastrum L. maintained metabolic homeostasis by enhancing amylase activity and promoting the conversion of starch to soluble sugars [60]. Collectively, these results highlight Cr as a key soil heavy-metal correlate of branch NSC variation in H. ammodendron across the mining-associated gradient.

The variation in NSC of N. tangutorum was most strongly associated with soil physicochemical properties and nutrient availability. Both soil pH and TN emerged as significant negative predictors, suggesting that soil alkalization and nitrogen excess may indirectly constrain carbon acquisition through impaired root physiological function. This pattern was consistent with the adaptive strategy of N. tangutorum under arid conditions, where intensified drought imposes co-limitations of water and phosphorus availability. Under such conditions, N. tangutorum adjusts the accumulation and allocation patterns of NSC among organs, enhancing nitrogen and phosphorus use efficiency to cope with environmental stress [61].

The present study revealed that in R. songarica, NSC dynamics were closely associated with the interplay of soil moisture, nutrient availability, and heavy metals. SMC emerged as a significant positive predictor of LSS and LNSC concentrations, contrasting with reports of NSC decline under extreme drought [62], thereby reflecting species-specific differences in carbon allocation strategies. In R. songarica, soluble sugars accounted for a markedly higher proportion of NSC than ST (Fig. S1), consistent with the drought-induced starch hydrolysis mechanism that sustains metabolic demands [63]. Moreover, interactions between water and nutrient availability were evident. The effects of nitrogen addition on NSC vary across plant functional types [64]. Under nitrogen-sufficient conditions, plants tend to increase carbon storage to enhance stress tolerance [65]. In the mining-affected environment, R. songarica maintains elevated soluble sugar levels under improved water and nutrient conditions, reflecting an adaptive strategy that prioritizes metabolic maintenance over long-term storage.

At the organ level, NSC variation in branches was most strongly associated with soil heavy metals, particularly Cr, whereas leaf NSC variation was more closely associated with the combined effects of water and nitrogen. This pattern is consistent with the “water–nitrogen interaction” theory [66], which posits that the two factors synergistically enhance photosynthetic carbon assimilation by coordinating stomatal conductance and nitrogen use efficiency. The observed pattern also corroborates the high sensitivity of leaf NSC to water fluctuations [17]. Long-term nitrogen addition has been shown to significantly increase NSC storage in trees [67], confirming the rapid responsiveness of leaf NSC to nitrogen availability. However, nitrogen effects on NSC can plausibly operate through competing pathways. On the one hand, increased N supply may enhance leaf N investment and photosynthetic capacity, thereby increasing carbon gain and promoting NSC accumulation when assimilate supply exceeds growth (sink) demand [64, 67, 68]. On the other hand, N enrichment may stimulate growth and sink activity, drawing down NSC pools if newly fixed carbon is preferentially allocated to structural biomass and maintenance rather than storage [64, 65, 68]. In our mining-affected transects, the close association between leaf NSC and water–nitrogen availability may reflect the balance between N-driven carbon gain and N-driven carbon consumption under water-limited conditions [66].

At the same time, leaf NSC composition can shift under metal-associated stress. Under heavy-metal stress, plants adjust their NSC composition as part of physiological acclimation: Yang et al. found that leaf soluble sugar, total NSC, and sugar-to-starch ratios increased by 21.6%, 11.6%, and 55.9%, respectively, whereas leaf and root starch decreased by 10% and 23.3%, respectively [52]. Such responses can be influenced by leaf habit, stress duration and intensity, soil pH, and heavy-metal type, and they tend to intensify with increasing stress severity. Mechanistically, Cr toxicity can constrain carbohydrate availability by impairing chloroplast integrity and photosynthetic functioning (e.g., disruption of PSII and suppression of electron transport/Calvin-cycle capacity), thereby limiting triose-phosphate supply for sucrose biosynthesis and downstream storage formation, while simultaneously elevating oxidative stress and the associated respiratory/detoxification costs that consume soluble sugars [57, 58, 69], providing a plausible explanation for lower SS and total NSC.

Although foliar dust deposition was not quantified in this study, in arid and windy open-pit mining landscapes, dust deposition on leaf surfaces may provide an additional pathway influencing leaf carbon balance and NSC dynamics, alongside soil-mediated effects. This may be particularly relevant in the Zhundong mining region, where frequent strong winds can enhance dust dispersion and deposition. Species differ in their capacity to retain particulate matter on leaf surfaces, which can alter light interception and gas exchange [70]. In addition, particle-associated metals deposited on foliage may contribute to aboveground metal inputs and have been widely used in biomonitoring studies [71]. Thus, leaf NSC responses in mining landscapes could involve the combined influence of soil water–nitrogen conditions and atmospheric deposition processes.

The influence of biotic factors was relatively limited. In both H. ammodendron and N. tangutorum, branch and leaf NSC and their components were not significantly correlated with H or CW, consistent with the findings of Chen et al., who reported no significant vertical variation in NSC content among assimilation branches in mature H. ammodendron at different canopy heights [72]. Wang et al. further showed that soluble sugar and total NSC concentrations in succulent halophyte leaves were significantly lower than in non-succulent species, with spatial variation primarily governed by soil salinity rather than climate [73]. This finding aligned with the mechanistic insights of Tian et al., who emphasized that drought stress triggers a shift from environment-driven regulation to NSC-mediated physiological control in plants [74]. Collectively, these findings suggest that inherent species traits (e.g., succulence) exert a stronger influence on NSC patterns than individual morphological variation. Functionally, the decoupled responses between branches and leaves may reflect a whole-plant strategy in which branches act as longer-term carbon reserves supporting survival and post-stress regrowth, whereas leaves prioritize rapid metabolic adjustment to maintain carbon supply under fluctuating environments [34, 75].

Conclusions

This study systematically elucidates the response mechanisms and adaptive strategies of desert shrubs in carbon allocation under the influence of open-pit coal mining. The main conclusions are as follows: (1) Distinct species- and organ-specific carbon allocation strategies were observed among the desert shrubs. Branches, functioning as long-term carbon reservoirs, exhibited higher plasticity, with NSC showing greater variability than in leaves. In contrast, leaves, as photosynthetic source organs, maintained relatively stable NSC levels, reflecting their functional role in sustaining continuous carbon assimilation. (2) Spatial patterns of NSC effectively indicated environmental stress gradients. In H. ammodendron, branch NSC concentrations were highest in the upwind (N) direction and declined significantly with distance, indicating high sensitivity to mining-associated environmental gradients. R. songarica maintained elevated NSC accumulation even in the downwind polluted areas, suggesting a distinct allocation pattern under metal-associated soil gradients. N. tangutorum displayed a nonlinear response pattern, suggesting a compensatory carbon-storage mechanism under moderate stress. (3) The dominant predictors of NSC variation exhibited multi-level differentiation. At the species level, NSC variation in H. ammodendron, N. tangutorum, and R. songarica was most strongly associated with soil heavy metals (especially Cr), soil physicochemical properties (pH and nitrogen), and water–nutrient availability, respectively, based on the measured soil-derived variables. At the organ level, branch NSC was more strongly associated with soil heavy metals, whereas leaf NSC was more closely associated with water and nitrogen availability. Because foliar particulate loading and plant-tissue metal concentrations were not quantified, the relative importance of soil–root versus foliar pathways cannot be resolved in this study and remains a key question for future research. Overall, this study provides a carbon economy-based framework to explain the differentiated adaptive mechanisms of desert shrubs under combined mining-induced stresses, offering theoretical insights for vegetation restoration and species selection in arid mining-reclamation areas.

Authors’ contributions

J.Z., B.Y., and Y.Z. were responsible for the overall research design and project planning. J.Z. also obtained the project funding. X.S. and J.Z. conducted data analysis and manuscript writing. X.S., L.P., and H.S. performed the experiments. J.Z., X.Z., and Y.Z. were responsible for manuscript revision and language polishing. All authors read and approved the final manuscript.

Funding

This research was supported by the Major Science and Technology Projects of Xinjiang Uygur Autonomous Region (2023A01002-1), the Project of Stable Support for Youth Teams in Basic Research Field of the Chinese Academy of Sciences (YSBR-119), and the West Light Foundation of the Chinese Academy of Sciences (2021-XBQNXZ-006).

Data availability

The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author.

Declarations

Ethics approval and consent to participate

All sampling and experimental procedures strictly complied with local and national regulations governing the collection and study of biological specimens in China. No endangered or protected species were involved in this study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.