Differential resource acquisition strategies of herbaceous and woody plants in temperate forests

Abstract

Background

Functional traits are pivotal indicators of plant ecological strategies, regulating resource acquisition and allocation through traits trade-offs. However, few studies have explored how trait coordination underlying resource-use strategies vary across plant life-forms from different trait dimensionalities. In this study, we examined 14 leaf functional traits in 81 plant species (27 herbaceous and 54 woody species) from the southwestern Qinling Mountains, including 6 structural traits (leaf area, leaf thickness, leaf mass per area, leaf dry matter content, leaf tissue density, and leaf volume), 4 chemical traits (leaf carbon, nitrogen, and phosphorus concentrations and nitrogen-to-phosphorus ratio), and 4 photosynthetic traits (maximum net photosynthetic rate, stomatal conductance, internal-to-ambient carbon dioxide ratio, and photosynthetic nitrogen use efficiency).

Results

Our findings revealed divergent distribution ranges between herbaceous and woody plants along the leaf economics spectrum, with herbaceous plants at the resource acquisition end (e.g., high nutrient concentration and photosynthesis rate) and woody plants at the resource conservation end (e.g., high tissue density). Furthermore, herbaceous plants exhibited higher connectivity in their trait networks, with leaf area as the hub trait, emphasizing efficient resource utilization through optimized light interception. In contrast, woody plants showed greater modularity, with leaf dry mass content as the hub trait, reflecting enhanced flexibility to environmental pressures by prioritizing maintenance and defense.

Conclusions

These findings advance our understanding of how trait coordination mediates plant adaptative strategies across multiple dimensions and underscore the functional importance of plant hub traits in modulating resource use strategy across plant life-forms.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-025-08026-z.

Introduction

Plant functional traits, comprising morphological, physiological, and phenological characteristics, can directly or indirectly affect the survival, growth and reproduction of individuals [1, 2]. Recent research has focused primarily on how traits respond to environmental gradients [3, 4], trade-offs in life history strategies [5, 6] and the role of traits in community assembly and biodiversity maintenance [7, 8]. Every plant organ possesses essential functional traits, while leaves are likely the most sensitive organ to climate change [9]. Leaf traits largely determine physiological processes and functions in plants, and their structural morphology, nutrient concentration, and photosynthetic capacity influence the survival strategies of plants in different habitats [3, 10]. Consequently, examining leaf functional traits can shed light on plant adaptation strategies to the environment under climate change.

Although individual functional traits may vary independently, plants typically display coordinated trait syndromes that reflect underlying resource trade-offs and ecological strategies [11, 12]. In particular, the concept of the leaf economics spectrum (LES) illustrates universal coordinated relationships among six key traits concerning plant growth and survival across diverse biomes and life-forms [13]. Along the economics spectrum, species vary from fast investment-return strategies, characterized by long leaf lifespan, low leaf mass per area, high leaf nitrogen and phosphorus and high photosynthetic rates, to slow investment return strategies, represented by short leaf lifespan, low nutrient concentrations, low respiration and photosynthetic rates [14, 15]. Studies across different geographical areas have substantiated the robust relationships among leaf traits [16–18]. Furthermore, the global spectrum of plant form and function extends this concept by integrating whole-plant traits, revealing two major ecological axes that shape plant adaptation strategies worldwide [5]. The first axis captures size-related variations, contrasting woody species with greater stature and seed size against smaller herbaceous plants with lighter organ mass. The second axis corresponds to the resource economics spectrum, balancing resource acquisition and conservation strategies across diverse plant forms [19, 20]. Trait plasticity along these axes enables different plant species to adjust their growth and survival strategies in response to varying environmental conditions, thereby enabling them to thrive in diverse habitats and to cope with environmental fluctuations [21].

As traits influence multiple functions collectively [22, 23], adaptation strategies along environmental gradients create intricate networks of interdependence among traits [24], facilitating a higher resolution visualization and quantification of multidimensional trait relationships [25, 26]. Identifying highly correlated traits or functional modules within plant trait network is essential, given that these traits may wield greater influence in governing diverse plant functions [27], while the environmental selection of these traits can induce cascading effects on other traits, thereby impacting the whole plant phenotype [28, 29]. These relationships may arise from evolutionary constraints, shared developmental pathways, or coordinated physiological functions that optimize resource allocation and survival under specific environmental conditions [14]. Recent research using network analysis has delved into the coordinated relationships between traits in lianas [30], woody plants [31, 32], herbaceous plants [28, 33], submerged macrophytes [34], and specific taxa [35]. However, the structure of plant trait networks varies considerably across life-forms and environmental conditions. For example, in herbaceous perennial plants, biomass allocation traits and stem specific length were reported as hub traits [28], whereas in woody plants, leaf thickness was found to play a central role in the networks [26]. This variability reflects the distinct ecological strategies adopted by plants, which consequently influences the composition and strength of trait interactions [36]. The lack of consensus on the central traits shaping different life-forms underscores the need for further investigation into trait network variations to provide a clearer perspective on the ecological strategies that support plant persistence in diverse environments.

Distinctions in functional traits among plant life-forms are evident in aspects such as size, shape, and life history [37–39]. These distinctions not only reflect divergent adaptative strategies to the environment but also contribute substantively to shaping the functional diversity and vegetation dynamics of plant communities [40, 41]. In the global trait space, herbaceous and woody plants emerged in separate hotpots [5], indicating fundamental differences in morphology and physiology between these two functional groups [42]. Specifically, characterized by short life cycles, herbaceous plants prioritize rapid growth and reproduction by allocating resources to photosynthetically efficient structures, such as large, thin leaves with high nutrient concentrations [43]. This resource-acquisition strategy enables them to complete their life cycle quickly, and, therefore, this strategy is well-suited for environments with variable conditions [38]. In contrast, woody plants invest substantially in structural and defensive traits to support their longevity and resistance to environmental stressors [14]. These plants typically have thicker leaves with lower nutrient use efficiency and photosynthetic capacity, reflecting their strategy of resource conservation and long-term persistence [44]. Understanding these distinctions and integrating plant life-forms in trait-trait and trait-environment relationships could clarify how plant species adapt to and occupy different resource spaces, providing insights into underlying mechanisms of plant community dynamics [45].

The Qinling Mountains, known for their diverse ecosystems, rich biodiversity, and well-preserved natural resources, offer an ideal setting for studying the response of mountain ecosystems to global climate change [46]. However, most studies in this region have focused primarily on shrubs [46], trees [47, 48], or specific taxa such as Quercus [49, 50], with less attention given to herbaceous plants. Herbaceous and woody plants typically occupy distinct spatial niches and employ different resource-use strategies. Integrating single-trait comparisons, multi-trait relationships and trait network analysis, can provide a multidimensional perspective on how these life-forms coordinate and balance functional traits to optimize resource acquisition and utilization. In the present study, we examined variations in leaf functional traits across different plant life-forms and discern distinctions in the topological structure of trait correlation network. We investigated 14 key leaf functional traits, including 6 structural traits (leaf area, leaf thickness, leaf mass per area, leaf dry matter content, leaf tissue density, and leaf volume), 4 chemical traits (leaf carbon, nitrogen, and phosphorus concentrations and nitrogen-to-phosphorus ratio), and 4 photosynthetic traits (maximum net photosynthetic rate, stomatal conductance, internal-to-ambient carbon dioxide ratio, and photosynthetic nitrogen use efficiency), across 81 plant species (27 herbaceous and 54 woody species) in the southwestern Qinling Mountains of China. Specifically, we hypothesized that (1) different life-forms exhibit distinct resource-use strategies, with herbaceous plants characterized by higher acquisitive traits and woody plants by higher conservative traits; (2) the trait networks of herbaceous plants display higher connectivity, whereas those of woody plants exhibit greater modularity, potentially aiding their adaptation to various environments; and (3) the hub traits in these networks vary with life-forms.

Materials and methods

Study site

This study was conducted in the southwestern Qinling Mountains, China (32°53′–34°18′N, 105°34′–107°70′E), with elevation ranging from 484 m to 2742 m a.s.l. (Fig. S1). The mean annual temperature is 4.9–12.2 °C and the mean annual precipitation is 708–843 mm across the sampling sites, according to the WorldClim database (v.2) [51]. And the soil is categorized as yellow cinnamon soil. The region’s unique climatic conditions, which vary from temperate to subtropical, support a high level of biodiversity. The vegetation includes evergreen and deciduous broad-leaved mixed forests, deciduous broad-leaved forests, coniferous and broad-leaved mixed forests, coniferous forests, and alpine shrub meadows. The dominant plant families in this region are Pinaceae, Fagaceae, Caprifoliaceae, Poaceae, and Rosaceae, with representative species including Pinus tabuliformis, Pinus armandi, Quercus aliena, Quercus variabilis, Lonicera tangutica, Fargesia spathacea, and Fragaria vesca.

Collection of samples

Samples were collected at 17 sites during the growing season in August, 2021. The sites were selected in relatively undisturbed natural areas to minimize disturbance from human activities. Geographic information, including longitude, latitude, and elevation were recorded for each site (Table S1). Three random 10 m × 10 m plots were established at each site, and within each plot we focused on sampling the dominant woody and herbaceous species. Three individual plants of each sampled species were then selected randomly within the plot. Firstly, photosynthetic traits were measured during a non-cloudy morning, and, subsequently, 15–30 disease-free, mature, fully expanded and unshaded leaves were collected. The petioles were removed, and the leaves were placed in envelopes as one replicate. Fresh leaf weight, leaf area, and leaf thickness were measured within 24 h, while the remaining leaves were homogenized and transported to the laboratory for measurements of leaf dry weight and nutrient concentration. In total, 600 samples from 81 species, belonging to 65 genera and 39 families were collected (Table S2).

Measurement of leaf functional traits

We measured 14 leaf functional traits, which are closely related to plant morphology, nutrients, and photosynthesis (Table 1) [52]. Leaf area (LA) was calculated using ImageJ software after scanning the leaves (Epson scanner, V19, Tokyo, Japan) [53]. Leaf thickness (LT), avoiding main veins, was measured in triplicate using a vernier caliper to an accuracy of 0.01 mm. Leaf fresh mass was measured in the field using an electronic balance. The leaves brought to the laboratory were heated at 105 °C for 10 min and then dried at 75 °C for at least 48 h to determine dry mass. Leaf mass per area (LMA) was calculated as leaf dry mass/leaf area; leaf dry mass content (LDMC) was calculated as leaf dry mass/leaf fresh mass; leaf tissue density (LD) was calculated as leaf dry mass/leaf volume; and leaf volume (LV) was determined by multiplying LA by LT. The dried samples were ground into powder for elemental quantification. Leaf carbon concentration (LCC) was measured using the potassium dichromate volumetric method; leaf nitrogen concentration (LNC) using the Kjeldahl sulfuric acid-digestion method [54]; and leaf phosphorus concentration (LPC) using the molybdate colorimetric test [54]. Leaf N: P ratio (N: P) was calculated as LNC/LPC.

Table 1.

A list of trait abbreviations used in this study

Category	Trait	Abbreviation	Unit
Structural traits	Leaf area	LA	cm2
	Laef thickness	LT	mm
	Leaf mass per area	LMA	g m− 2
	Leaf dry mass content	LDMC	-
	Leaf tissue density	LD	g cm− 3
	Leaf volume	LV	cm³
Chemical traits	Leaf carbon concentration	LCC	g kg− 1
	Leaf nitrogen concentration	LNC	g kg− 1
	Leaf phosphorus concentration	LPC	g kg− 1
	Leaf N: P ratio	N: P	-
Photosynthetic traits	Area-based photosynthetic capacity	A area	µmol m⁻² s⁻¹
	Internal-to-ambient CO2 ratio	χ	-
	Stomatal conductance	gs	mol m⁻² s⁻¹
	Photosynthetic nitrogen use efficiency	PNUE	µmol g− 1 N s− 1

Photosynthetic traits were measured using the LI-6800 portable photosynthesis measurement system (LI-COR, Lincoln, NE, USA). Measurements were taken on clear mornings from 09:00 to 12:00, and leaves that grow in sunlight and can fill the leaf chamber were selected to be clamped in the leaf chamber. The light intensity was set to 1200 µmol m^− 2 s^− 1; the flow rate was set to 500 µmol s^− 1; the relative air humidity (RH-air) was maintained at 55%–60% to control the humidity status inside the leaf chamber; the CO₂ concentration was kept at 400 µmol mol^− 1 with CO₂ cylinders; the mixing fan speed was adjusted to 10,000 rpm and the temperature was controlled at approximately 25℃. After the conditions were stable, maximum net photosynthetic rate per unit area (A_area) and stomatal conductance (g_s) were measured in four replicates. Internal-to-ambient CO₂ ratio (χ) was calculated by the ratio of intercellular CO₂ concentration (C_i) to ambient CO₂ concentration (C_a) and photosynthetic nitrogen use efficiency (PNUE) was calculated as A_area/N_area.

Statistical analyses

Descriptive statistical analyses and the coefficient of variation (CV) tested the extent of variation in the 14 leaf functional traits. To ensure the residuals approximated normal distribution, all trait data were standardized. LA performed logarithmic square root transformation (log_e), χ was logit-transformed (log_e [55], and the other traits were transformed using natural logarithms. To test differences in leaf functional traits between herbaceous and woody plants, we used linear mixed-effects models with lifeform as a fixed effect and sites and species as random effects. The models were fitted using the ‘lmer’ function in the ‘lme4’ package [56] in R [57]. The visualization of the results was conducted using Origin 2021.

Principal component analysis (PCA), using the ‘prcomp’ function in R, assessed multivariate associations among leaf functional traits and disparities in resource utilization strategies, with the results visualized using ‘factoextra’ [58] and ‘ggplot2’ [59] package. We quantified trait contributions to the principal components by computing Pearson correlation coefficients (r) between each trait and PC1/PC2 scores. To analyze the positions occupied by herbaceous and woody plants along the resource strategy axes, we used linear mixed-effects models with lifeform as a fixed effect and sites and species as random effects to examine differences in PC1 and PC2 scores.

Network analysis was applied to determine the topological discrepancy in multi-dimensional trait networks between plant life-forms. In leaf trait networks (LTNs), nodes represented leaf functional traits, while edges denoted relationships between traits [25]. Firstly, the matrix of correlation coefficients (r) for trait-trait relationships was calculated using Pearson’s correlations along with their statistical significance (p). Subsequently, a significance threshold was set to exclude illusory correlations among traits, setting relationships with p < 0.05 to 1 and other relationships to 0, yielding an adjacency matrix A = [a_{i, j}] and a_{i, j} ϵ [0,1] [25]. The connections between each pair of traits were weighted by absolute correlation strength (|r|). To derive uncertainty ranges for these parameters, we performed 1000 random resamplings of plant species, with the establishment of a LTN for each bootstrapping, and at least 75% of the species were selected randomly each time. Finally, the LTNs were visualized and the network parameters were calculated using the ‘igraph’ package [60] in R.

Four network parameters assessed the connectivity and centrality of the network, including two overall parameters (edge density and modularity) and two node parameters (degree and betweenness). Edge density is defined as the fraction of the number of edges actually present in the network compared to the maximum potential number of edges, and a higher edge density indicates better collaboration between the leaf functional traits [26]. Modularity is a metric used to measure the degree of internal cohesion within subnetworks (or modules) and the degree of separation between subnetworks (or modules), with a higher modularity representing more traits divided into different modules performing particular functions [31]. Degree is the number of edges connected to a node [27]. The more edges connected means the higher degree of the node, indicating that the trait occupies a more important position in the network and is referred to as a “hub trait” in the network. Betweenness is the number of all shortest paths through a node [25]. A trait with high betweenness indicates that the trait is a bridge between different functional modules and an indispensable “mediator” in the network, and the response of leaf functional traits to environmental changes is influenced by the coordination between different modules. All statistical analyses were performed using R 4.3.2.

Results

Variation of leaf functional traits in different plant life-forms

Substantial functional divergence was observed among species, as evidenced by coefficient of variation values ranging from 19% to 161% across the 14 leaf traits (Table S3). LV (range: 0.01 to 10.1 cm³) and LA (range: 0.59 to 227.5 cm²) showed large variation across species, with 1000-fold and 300-fold variations, respectively. LT (CV = 52%) and LD (CV = 47%) exhibited moderate variation across species. LCC exhibited the lowest variation (around 6-fold), varying from 112.8 to 701.6 mg g^− 1, which was a relatively stable trait. Using linear mixed-effects models, six leaf traits showed significant differences between plant life-forms (Fig. 1, Table S4). For structural traits, LT, LMA and LDMC were significantly higher in woody plants than in herbaceous plants. For chemical traits, LCC was significantly higher in woody plants, whereas LPC was significantly higher in herbaceous plants. For photosynthetic traits, PNUE was significantly higher in herbaceous than in woody plants.

Fig. 1.

Boxplots displaying leaf traits in different life-forms. Structural traits (a-f): (a) LA, (b) LT, (c) LMA, (d) LDMC, (e) LD and (f) LV. Chemical traits (g-j): (g) LCC, (h) LNC, (i) LPC and (j) N: P. Photosynthetic traits (k-n): (k) A_area, (l) χ, (m) g_s and (n) PNUE. The box presents the interquartile range, the inner line represents the median; the solid dot represents the mean value. Significance was assessed using linear mixed-effects models with lifeform as a fixed effect, and sites and species as random effects (***p < 0.001, *p < 0.01, p < 0.05). For trait abbreviations refer to Table 1

Relationships of multiple traits of structural, chemical and photosynthetic traits across life-forms

Principal component analysis (PCA) revealed that the relationships among the 14 leaf traits aligned with the two-dimensional spectrum of global trait space, namely, the size spectrum and economics spectrum (Fig. 2a, Table S5). The PC1 axis, determined mainly by size traits (LA, LV and LT; see Table S5), explained 24.1% of the total variation; whereas, the PC2 axis, which reflected trade-offs in plant life history strategies between resource acquisition and resource conservation (LES), explained 18.6% of the total variation. Along the PC2 axis, species transitioned from having leaves with slow growth (high LMA, LDMC and LD) to leaves that acquired resources rapidly (high LPC, LNC and PNUE). There was a notable separation in the distribution ranges between herbaceous and woody plants along the PC2 axis, with the higher scores (p < 0.001) observed for woody plants (Fig. 3a; Table S6). This meant that herbaceous plants exhibited a strategy for resource acquisition, while woody plants employed a resource conservative strategy.

Fig. 2.

Principal components analysis (PCA) displaying multivariate coordination for (a) all traits, (b) structural traits, (c) chemical traits and (d) photosynthetic traits between different plant life-forms. Green and orange dots represent herbaceous and woody plants, respectively. Arrows represent the loadings of traits on PC1 and PC2. Ellipses indicate 95% confidence intervals. For trait abbreviations refer to Table 1

Fig. 3.

Differences in PC2 scores between herbaceous and woody plants based on linear mixed-effects models. (a) all traits PC2, (b) structural traits PC2, (c) chemical traits PC2, and (d) photosynthetic traits PC2. Significance was assessed using mixed-effects models with lifeform as a fixed effect, and sites and species as random effects (***p < 0.001, **p < 0.01, *p < 0.05)

For structural traits, PC1 explained 48.0% of the trait variation, with PC2 accounting for 30.9% (Fig. 2b, Table S5). LV, LT and LA were correlated positively with the PC1 axis. LMA, LDMC and LD were correlated positively with the PC2 axis, representing the direction of resource conservation. Herbaceous and woody plants differed (p < 0.01) in their positions along the PC2 axis (Fig. 3b; Table S6). Woody plants had higher PCA1 values, i.e., had more qualitatively dense leaves (high LMA, LDMC and LD) than herbaceous plants, which meant that woody plants tended to have a higher carbon investment. In contrast to woody plants, herbaceous plants had lower investment in construction costs.

For chemical traits, PC1 and PC2 axes explained 33.5% and 28.1% of the trait variation, respectively (Fig. 2c, Table S5). LNC and LPC were the main contributors to the first principal component, while LCC and N: P were the main contributors to the second principal component. The scores of principal component axis 1 exhibited significant differences (p < 0.01) between herbaceous and woody plants (Fig. S2; Table S6), with herbaceous plants having lower scores, indicating that herbaceous plant leaves had higher concentrations of nitrogen and phosphorus. In addition, PC2 scores also differed (p < 0.05) between herbaceous and woody plants (Fig. 3c; Table S6). The leaves of woody plants had higher carbon concentrations, indicating that they accumulated more photosynthetic products.

For photosynthetic traits, the first two principal components explained 55.3% and 32.3% of the trait variation, respectively (Fig. 2d, Table S5). A_area, PNUE and g_s were correlated positively with the PC1 axis, which represented the resource acquisitive strategy, while χ was correlated positively with the PC2 axis. The scores of herbaceous plants on the PC2 axis were higher (p < 0.001) than herbaceous plants (Fig. 3d; Table S6). Herbaceous plants had higher resource acquisitive traits (high A_area, PNUE and g_s) than woody plants, indicating a strategy of rapid growth, while woody plants adopted the opposite growth strategy.

Comparison of leaf trait networks between herbaceous and Woody plants

Leaf trait networks facilitated the visualization of interdependencies among traits of different plant life-forms (Fig. 4), and, in addition, their distinctions were quantified by the calculation of network parameters. The proportion of connectedness among traits was higher (p < 0.001; Fig. S3a) in herbaceous than in woody plants (edge densities were 0.47 and 0.44, respectively), indicating better connectivity among leaf functional traits in herbaceous plants. However, the modularity of the trait correlation networks for woody plants were higher (p < 0.001; Fig. S3b) than that of herbaceous plants. Similarly, the component and number of modules of the networks varied between life-forms. Three modules were observed in the leaf trait network of herbaceous plants (Fig. 4a), while four modules were observed in the leaf trait network of woody plants (Fig. 4b). Compared to herbaceous plants, the leaf trait networks of woody plants were more modularized.

Fig. 4.

Leaf trait networks for herbaceous plants (a) and woody plants (b). Different colors in the network represent different modules. Black edges and red edges indicate intra-module and inter-module connections, respectively. The width of the line indicates the strength of the correlation. The size of the circle indicates the degree of the trait. For trait abbreviations refer to Table 1

“Hub traits” in the trait network were typically those traits that play pivotal roles within the network, possessing higher connectivity to other traits. The trait with the highest degree was LA in the trait networks of herbaceous plant (Fig. 5a), whereas, LDMC was the most important trait in the network of woody plants (Fig. 5c). Traits with higher centrality (i.e. betweenness) in the trait network were identified as “mediator traits”. The trait with the highest betweenness in the trait network of herbaceous plants was LA (Fig. 5b), whereas the most central trait in the trait network of woody plants was LDMC (Fig. 5d). Therefore, within trait networks, a specific trait could function as both a hub and a mediator.

Fig. 5.

Variation in the node parameters of the leaf trait network (LTNs) across plant life-forms. (a) Degree and (b) betweenness of 14 plant traits for herbaceous LTNs; (c) degree and (d) betweenness of 14 plant traits for woody LTNs. For trait abbreviations refer to Table 1

Discussion

Two main dimensions of leaf functional traits variation

In line with previous studies, nearly half of the leaf trait variation was captured by two independent dimensions. The first dimension was associated closely with size-related traits, including LA, LV, and LT. Larger leaves contribute to plants gaining an advantage in the competition for light resource, thereby enhancing their photosynthetic capacity [44]. Coupled with higher LT, this fortifies plants against environmental stressors like wind and potential predators, therefore this combination of traits enables plants to better store water and nutrients in adverse environments [61]. Although Díaz et al. reported clear differentiation along the size axis, we found no significant difference between herbaceous and woody plants [5]. This can be largely attributed to the dominance of perennial herbs (76% of the sampled herbaceous species), which tend to have thicker and longer-lived leaves than annual herbs, enhancing structural persistence and resistance to environmental stress [62]. Furthermore, environmental factors such as climate and soil conditions may act as strong filters on plant functional traits, causing individuals to evolve similar traits in response to comparable environmental constraints [63].

The second dimension displayed pronounced differences between fast-growing species (with high acquisitive trait values) and slow-growing species (with high conservative trait values), known as the LES. Species at the acquisitive end of the spectrum typically have higher nitrogen and phosphorus concentrations and greater photosynthetic efficiency [64], allocating resources primarily for growth and nutrient uptake [21]. This fast-return strategy maximizes carbon gain in high-resource environments but increases vulnerability to abiotic stresses [14]. In contrast, species at the conservative end exhibit denser tissues and reduced nutrient concentrations, allocating resources to long-term survival through lignified cell walls and phenolic compounds that enhance herbivore resistance and tissue longevity [3, 45]. These two orthogonal dimensions revealed that plants evolved general ecological trade-off strategies in functionally diverse trait space [5, 7]. Variation of different species along the spectrum may lead to niche segregation, reducing competition by adapting to different functional strategies, which can benefit ecosystem stability and diversity [39].

Divergent resource acquisition strategies in herbaceous and Woody plants

We found that woody plants had higher conservative trait values (high LMA, LDMC, LD) and herbaceous plants had higher acquisitive trait values (high LPC, LNC, PNUE) on the PC2 axis (i.e., leaf economics spectrum) (Figs. 3 and 6), which supported the first hypothesis. This suggests that plant life-forms occupying different ecological niches within a community employ different trade-off strategies in response to the environment. Similar patterns have been documented in early successional communities, where differences in leaf traits between herbaceous and woody plants were associated with competitive substitution of life-forms [44].

Fig. 6.

A comprehensive framework illustrating plant functional strategies, integrating single trait characteristics, two-dimensional spectrum and multi-dimensional network structures for herbaceous and woody plants. For trait abbreviations refer to Table 1

There were differences in morphological structure, nutrient concentration and photosynthetic characteristics between herbaceous and woody plants. For leaf structural traits, LT, LMA, LDMC and LD were higher in woody than in herbaceous plants. Thicker leaves provide mechanical support for woody plants to withstand environmental stresses in the canopy; while also providing protective tissue against frost and radiation damage [65]. LMA correlates with LD due to the higher proportion of cell wall mass in denser tissues [66, 67]. The higher LMA and LD of woody plants can be attributed to a carbon-expensive investment in support tissues that favors leaf defense and a longer life span [66]. In addition, woody plants had higher LDMC values than herbaceous plants, aligning with results from a study that compared herbaceous and woody angiosperms across 530 sites in China [43]. Variations in the structural attributes of plant life-forms imply differences in resource allocation strategies, with herbaceous plants tending to use resources for rapid growth, while woody plants invest more for maintenance, storage and defense [45, 68].

In contrast to woody plants, herbaceous plants exhibited low LCC, consistent with low construction investment, but higher LNC and LPC. These findings reinforce the premise that fast-growing species produce more leaves with higher nutrient concentrations to improve photosynthetic rates, as nitrogen and phosphorus are essential for photosynthesis, whereas slow-growing species exhibit the opposite trend [69]. Although leaf N: P was higher in herbaceous than in woody plants, their mean values were both lower than 14 (13.60 and 12.40, respectively; see Table S7), suggesting that plant growth is limited mainly by N in this region [70].

PNUE can well evaluate the contribution of nitrogen to photosynthesis [71]. In nitrogen-restricted conditions, plants with higher PNUE can utilize available resources more efficiently. In the present study, herbaceous plants had higher PNUE, indicating that herbaceous plants used nitrogen more efficiently than woody plants. Leaves with longer life span will allocate more nitrogen to cell walls, while those with shorter life span allocate more nitrogen to the photosynthetic apparatus [72]. Furthermore, the high density or thickness of leaves in woody plants, characterized by more mesophyll cell layers and smaller intercellular spaces, hinders CO₂ diffusion, thereby limiting photosynthetic rates [73]. In contrast, herbaceous plants typically maintain higher stomatal conductance (g_s) to increase CO₂ uptake and mitigate overheating to enable rapid growth and reproduction in their shorter life spans [61, 74], . In summary, differences in structural attributes, nutrient acquisition, and photosynthetic capabilities between herbaceous and woody plants underscore not only their morphological and physiological distinctions, but also their adaptation and trade-off strategies under varying environmental conditions [75].

Leaf trait networks vary across plant life-forms

Through the coordination of multiple traits, plants reshape their trait networks to optimize resource acquisition and utilization, thereby enhancing their ability to adapt to the environment [25]. In the present study, herbaceous plants exhibited higher edge density in their leaf trait networks compared to woody plants (Figs. 4 and 6), which was consistent with our expectations. Increased edge density within the trait network implies that plants can acquire and mobilize resources more efficiently [31]. Hence, in resource-limited environments, such as light availability, herbaceous plants are inclined towards higher proportions of trait connectivity, promoting their competitiveness and ability to thrive. However, establishing and maintaining high trait connections at the expense of costly construction, therefore, to a certain extent, trait independence facilitates the flexibility of plants to adjust their functions under variable conditions [76]. Modularity identifies the extent of separation among trait modules. Woody plants displayed higher modularity in the trait network than herbaceous plants, consistent with previous research on adaptation strategies among plant life-forms in drylands of China [77]. In harsh environments with high radiation and strong wind, woody plants organize multiple traits into different functional clusters to form optimal combinations, as environmental pressures may only affect certain traits [30].

Highly connected traits within the plant trait network are crucial in understanding plant strategies, given their role in regulating vital functions or participating in physiological processes [78, 79]. There is ample evidence that hub traits are not universally consistent but vary across different life-forms or combinations of traits [25, 27, 30]. In the present study, whether as “hub” or “mediator” traits (i.e., the highest degree or betweenness), LA and LDMC were central traits in the herb and woody trait networks, respectively. LA is related closely to photosynthesis, and leaf size can directly affect the plant’s ability to intercept light and acquire carbon [80]. Concurrently, leaf area influences the balance between leaf temperature and transpiration rate, thereby affecting the exchange of heat and water within leaves [80]. In shaded and humid conditions of the understory, larger leaves facilitate herbaceous plants to obtain more light energy, which promotes their growth and development [81]. Conversely, LDMC is considered to primarily reflect leaf defense and drought resistance capabilities [43, 82]. Woody plants generally prioritize the connectivity between structural traits to enhance adaptability to unfavorable environments, investing more carbon in thicker and denser leaves to minimize water and nutrient loss and to shield the foliage from high radiation damage [83]. Within trait networks, different traits function as essential “hubs” or “bridges”, connecting multiple modules and underscoring the adaptation and survival strategies of different plant life-forms [28].

Conclusions

The present study sheds light on multivariate trait relationships and trade-off strategies in herbaceous and woody plants as they adapt to varying environmental conditions. At the leaf level, approximately half of the variation in 14 functional traits was captured by two main dimensions: the size spectrum and the economics spectrum. Herbaceous and woody plants occupy divergent positions along the resource economics spectrum, with herbaceous species at the acquisitive end and woody plants at the conservative end. The trait network further reinforced the different life strategies between these two plant groups. Herbaceous plants exhibited denser trait networks, characterized by high edge density and low modularity, with leaf area (LA) as the hub trait, reflecting efficient light interception that supports rapid growth in light-limited understory environments. Woody plants exhibited looser trait networks, characterized by low edge density and high modularity, with leaf dry mass content (LDMC) as the hub trait, reflecting costly carbon investment in the leaf structure for persistence and defense. In summary, these findings demonstrate how life-form differences are reflected in multidimensional trait coordination, offering new insights into how plants balance resource use and survival across contrasting environments.

While the findings advance our understanding, several limitations remain. Firstly, although the study encompasses extensive variation in leaf-level traits, it neglects belowground and stem traits, which are also critical for capturing the full spectrum of resource-use strategies and ecological trade-offs. Secondly, the study infers ecological strategies from trait variation but does not incorporate direct assessments of performance metrics such as survival or productivity, limiting the interpretation of functional significance. Overall, future studies should integrate traits across organs to explore potential shifts in existing strategy spectrum, thereby enhancing our understanding of plant form and function across different environmental conditions.

Authors’ contributions

JD and JR conceived the ideas and experimental design; J. Luo, YS and SX collected samples in the field; J. Luo, DT, FL, QH and YD performed laboratory measurements; J. Luo and J. Li analysed the data with suggestions from XW, JX, JP, JR and JD. J. Luo, J. Li and JD wrote the manuscript. AAD, MA, WH and LD contributed to the subsequent revision of the manuscript. All authors provided substantial comments on the final manuscript.

Funding

This study was supported by the National Key Research and Development Program of China (2023YFF0805602), the National Natural Science Foundation of China (32225032, 32271597, 32501402), the Natural Science Foundation of Gansu Province (21JR1RA138, 22JR5RA525, 23JRRA1157), the Key Research and Development Programs in Gansu Province (23ZDKA0010, 23ZDNA009, 24ZD13NA016), the Gansu Provincial Youth Science and Technology Fund (25JRRA710) and Research Grant Support from Lanzhou City (127000-563224111).

Data availability

Data from this study can be requested from the corresponding author.

Declarations

Ethics approval and consent to participate

The collection of leaf samples was approved by the Forestry Bureaus of Gansu Province and Shaanxi Province. Field studies and sample collection strictly complied with Chinese law as well as local legislations of the collection areas. All collection activities were conducted in a sustainable and non-destructive manner to minimize disturbance to natural populations and their habitats. No species listed under the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES) were collected in this study. Plant specimens were formally identified by Ying Sun and Jiali Luo, and voucher specimens have been deposited in the Herbarium of the School of Life Sciences, Lanzhou University under accession number LZU98859. We confirm that all aspects of the study comply with institutional, national, and international guidelines.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.