Effects of climate, soil, and leaf traits on nutrient resorption efficiency and proficiency of different plant functional types across arid and semiarid regions of northwest China Authors: Jingkun Zhang, Wanying Yu, Yuan Wang, Zhenying Huang, Guofang Liu Published: 2024 ABSTRACT AbstractBackgroundPlant nutrient resorption is crucial for the efficient conservation of nutrients. However, the mechanisms through which abiotic and biotic factors control nutrient resorption remain controversial. We investigated leaf nitrogen (N) and phosphorus (P) resorption efficiency (NRE and PRE) and resorption proficiency, as well as the underlying mechanisms for each plant functional type (PFT: non-legume herbs, non-legume shrubs, and legumes) by collecting green and senescent leaves of 59 species covering 106 sites from arid and semiarid regions of northwest China.ResultsLegumes had much lower leaf NRE and much higher senesced leaf N than the other two PFTs; they had similar leaf PRE to non-legume shrubs. Non-legume herbs exhibited the highest leaf P resorption. Climate, particularly temperature, increased leaf N resorption in non-legume herbs; however, climate, particularly decreasing precipitation, decreased leaf P resorption in legumes. Leaf nutrient resorption in non-legume shrubs decreased with increasing soil fertility, but leaf NRE in legumes increased. Leaf traits contributed more to leaf N and P resorption than climate and soil. Senesced leaf N and P concentrations increased along the resource-conservative to resource-acquisitive strategy axis. There were strong negative relationships between leaf NRE and senesced leaf N concentration and between leaf PRE and senesced leaf P concentration, in which legumes had a lower slope than non-legumes.ConclusionsThese findings suggest that ecological strategies and N-fixing plant types modulate nutrient resorption. Plants with the resource-conservative strategy are highly proficient in nutrient resorption. We highlight the importance of leaf economics traits and spectrum in regulating leaf nutrient resorption in drylands in the context of global climate change, potentially modulating plant traits and community composition. The higher proficient and efficient N and P resorption of plant species suggests the crucial importance of nutrient resorption in the nutrient cycling of harsh drylands.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12870-024-05794-y.Keywords: Dryland, Leaf economics spectrum, Nutrient resorption efficiency, Nutrient resorption proficiency, Legumes, Plant functional type Introduction Nutrient resorption is a process by which plants redistribute nutrients from their senescent structures to other living tissues. Resorption can improve the nutrient (mainly nitrogen (N) and phosphorus (P)) utilization efficiency and reduce the dependence of plant growth on soil nutrient conditions [1–3]. Nutrient resorption also helps reduce nutrient loss after leaf senescence due to litter decomposition and leaching. Resorption can prolong the retention time of nutrients in plants, thereby enriching the material for plant growth and reproduction [4, 5]. Therefore, it is an efficient protective mechanism that improves nutrient resorption [4, 6–9]. Nutrient resorption enhances the adaptability of plants to nutrient-poor environments or difficulties in nutrient acquisition due to adverse conditions, such as drought [10, 11]. Nutrient resorption efficiency (NuRE) is best suited for quantifying the relative degree to which plants can conserve nutrients invested in the foliage [12]. Therefore, uncovering the underlying driving factors of nutrient resorption efficiency can aid in understanding plant adaptation to diverse environments and ecosystem nutrient biogeochemical cycling (i.e., nutrient resorption and litter nutrient release during decomposition) [13–15]. N and P are crucial mineral elements for plant growth and reproduction as they drive plant metabolic processes [12]. Therefore, N and P cycling has been extensively studied. A meta-analysis suggested that the global average resorption efficiencies of N and P are 62.1% and 64.9%, respectively [4]. Different plant functional types (PFTs) have different ecological strategies [16], resulting in their different nutrient resorption capacities [4, 17]. In general, herbaceous species have a higher relative growth rate than shrubs because of their larger specific leaf area [18, 19]. In terrestrial ecosystems, plants and microbes are closely linked, such as plant and N-fixing bacteria, arbuscular mycorrhizal fungi (AM), and ectomycorrhizal fungi (EcM) [20], which modulate nutrient resorption efficiency because specialized microbes can help host plants uptake mineral nutrients. Stewart et al. (2008) found that N-fixing plants had a lower N resorption capacity than non-N-fixing temperate woody species [21]. Similarly, Xu et al. (2021) found that legume plants had significantly lower leaf N resorption efficiency than non-legume plants, and that shrub leaves had higher N resorption rates than tree species [22]. As important abiotic drivers, both climate and soil affect leaf nutrient resorption, including precipitation, temperature [3, 23, 24] or soil fertility [6, 25–27]. Intense and chronic drought reduces N and P resorption efficiency in a typical steppe [24]. For example, in high-latitude regions with relatively low temperature and precipitation, leaf N and P resorption efficiencies are usually higher than in low-latitude regions [4]. Soil properties determine soil nutrient status or abundance, thus affecting plant nutrient resorption. Nutrient resorption tends to be higher in poor soil conditions, particularly under favorable precipitation conditions [25, 28, 29]. A meta-analysis showed that N enrichment decreases leaf NRE and PRE by 6.85% and 3.41%, respectively across Northern China’s grasslands [26]. According to a study of 14 populations of Stipa breviflora in northern China, soil characteristics were the primary environmental factor influencing leaf nutrient resorption [27]. Changes in nutrient resorption under different environmental conditions are advantageous for plant species for improving their growth and reproduction. However, the relationship between abiotic conditions and nutrient resorption in leaves remains unclear. Therefore, further studies across large spatial scales are needed to reveal how abiotic factors regulate leaf nutrient resorption. Plant functional traits, which represent distinct plant adaptation strategies, play crucial roles in nutrient resorption [4, 30, 31]. A global-scale study showed that the leaf resorption efficiency decreases with increasing leaf nutrient level [31]. Similarly, the leaf N, P, and K resorption efficiencies of Phoebe bournei were significantly negatively correlated with nutrients in green leaves in a forest ecosystem [32]. Wright & Westoby (2003) suggested that the leaf mass per area (LMA) and leaf lifespan of Australian evergreen plants were not strongly correlated with leaf N resorption efficiency, whereas senescent leaf N and P concentrations were negatively correlated with leaf lifespan [30]. These suggest that resource-conservative species have higher nutrient resorption capacity. In contrast, Ma et al. (2024) found that resource-acquisitive deciduous tree species tended to be higher in NRE than the resource-conservative evergreen tree species in subtropical forests. These indicate that the observed patterns could depend on ecosystem types and resource availability. Therefore, more empirical evidence on the effects of leaf economic traits and leaf or plant economic spectra on the regulation of leaf nutrient resorption should be encouraged to uncover the link between leaf nutrient resorption and plant ecological strategies [33]. Nutrient resorption proficiency (NuRP) measures the degree to which selection minimizes nutrient loss [7]. Thus, higher nutrient resorption proficiency corresponds to lower senesced nutrient concentrations. Killingbeck (1996) proposed that plants are highly proficient in nutrient resorption if they reduce the concentrations of N and P in senescing leaves to < 7 mg g− 1 and < 0.5 mg g− 1, respectively [7]. Plants growing in nutrient-poor environments have higher nutrient resorption efficiency, which may not be due to their stronger absorption capacity for soil nutrients, but rather to more complete nutrient resorption [13]. Nutrient resorption proficiency is more sensitive to increased N availability than efficiency [34]. Therefore, nutrient resorption proficiency may be a better measure of the evolution of this process because the selection force acts on traits rather than proportions expressed as resorption efficiency [7]. In general, there is a negative correlation between the nutrient concentrations of the senesced leaves and nutrient resorption efficiency [35–37]. These results indicate that both nutrient resorption and nutrient resorption proficiency can be used as supplementary metrics to qualify nutrient resorption capacity. In this study, we collected the leaves of 59 dominant plant species from arid and semiarid areas of the Ordos Plateau, China. We investigated their nutrient absorption efficiency and proficiency in response to climate, soil, and plant traits covering three plant functional types including non-legume herbs, non-legume shrubs, and legumes, and uncovered key abiotic and biotic controls in leaf nutrient resorption in water-limited arid and semiarid ecosystems. We formulated three hypotheses: (1) leaf nitrogen resorption in N-fixing species would be lower than that in non-N-fixing species; (2) leaf N and P resorption would be regulated by soil properties in arid and semiarid ecosystems; and (3) plant species with a resource-conservative strategy would be more efficient for leaf nutrient resorption. Materials and methods Study area This study was conducted in the arid and semiarid grasslands of the Ordos Plateau in northern China. The Ordos Plateau is a relatively independent geographical unit located in the southern part of the Inner Mongolia Autonomous Region along the northern edge of the Loess Plateau. The Ordos Plateau mainly includes the Mu Us sandy grassland and the Kubuqi desert, and is a multi-level ecological transition zone [38]. From east to west, it transitions from typical grasslands, desertified grasslands to grassland deserts with complex and diverse landscapes. This is a typical area for comparative studies of grassland and desert ecosystems. The study area is located on the western edge of a temperate continental climate. The latitude of sites ranges from 36°59′16″N to 40°38′19″N, and the longitude of sites ranges from 105°47′26″E to 111°3′45″E (Fig. 1, S1). Climate data with a resolution of 1 km × 1 km were downloaded from a global climate database (WorldClim, http://www.worldclim.org/), and climate information including the mean annual temperature (MAT) and mean annual precipitation (MAP) for the 106 sites was extracted using the raster package of R software (v4.3.1, R Core Team, 2023). MAT and MAP range 5.5 ~ 9.0 °C and 141 ~ 421 mm, respectively (Fig. 1). Precipitation mainly occurs from June to September, becoming more arid from southeast to northwest. The geographical map of 106 sites in the arid and semiarid regions of northwest of China Sample collection and leaf morphological trait measurements From July to August 2022, 106 sites were sampled, each 30 × 30 m in size (Fig. 1). The plant types collected in this study were herbs and shrubs, which were divided into three categories based on their plant functional types (PFT): non-legume herbs (151 samples), non-legume shrubs (44 samples), and legumes (eight legume herbs and 30 legume shrubs), including 18 families and 59 species (details in Table S1), with a total of 233 samples. Newly senescent leaves of dominant species were collected randomly. The criteria were that completely dry and yellow leaves could be dropped by shaking the branches or the dry and yellow leaves that had just fallen onto the surface. The fresh and newly senesced leaves of each species were collected in triplicate, and the weight of each repeated collection was greater than 2 g. It was packaged in an envelope and brought back to the laboratory to determine the leaf traits. Newly senesced leaf litter was collected to remove the surface soil and other foreign matter. A portion of fresh leaves (2–10 randomly selected intact leaves per individual) was measured to determine their functional traits. The blade image was obtained using a scanner, and the leaf area was calculated using ImageJ software. Each batch of fresh leaves was soaked in clean water in the dark before measurement, and the surface moisture was absorbed using absorbent paper. Moreover, the surface contaminants of the treated leaves were removed, and their saturated fresh weights were measured. The dry weight of the treated leaves was determined after oven-drying at 70 °C for 48 h. Specific leaf area (SLA, cm2 g-1) was determined by dividing the leaf area by its dry weight; leaf dry matter content (LDMC) was determined by dividing the leaf dry weight by its fresh weight. We collected topsoil at a depth of 0–20 cm using a soil auger and five-point sampling method. The soil samples were cleared of all roots and stones and allowed to dry in a ventilated environment. A total of 318 soil samples were collected (three replicates per site). Measurements of physicochemical properties Green and senescent leaves and soil samples were ground in a ball mill (Retsch, model MM400, Restch GmbH). Soil pH was measured in a soil–water suspension using a pH meter (PB-10, Sartorius, Germany). We used a laser particle analyzer (Mastersizer 2000, Malvern, UK) to measure the soil mechanical composition (clay, silt, and sand content), in which only soil sand content was used as a proxy for soil texture because of the strong correlations between the three components. The C and N concentrations of green and senescent leaves and soil samples were analyzed using elemental analysis (Vario el cube chons elemental analyzer, Elementar Analysensysteme GmbH, Hanau, Germany). The amount of leaf samples used for C and N concentration (%) measurement for each sample type: We used 8 mg powdered dry sample for green and senescent leaves and 250 mg powdered dry sample for soil samples. The total phosphorus concentration of green and senescent leaf samples was measured using the H2O2-HNO3-ICP-OES method: Weighed 100 mg powdered leaf sample and placed it in a test tube, added 5 ml of HNO3 and 2 ml of H2O2 for digestion using a high-throughput closed microwave digestion device (MARS6), and measured it using inductively coupled plasma (ICP) mass spectrometry (iCAP 6300 ICP-OES spectrometer, Thermo Fisher, USA). The total phosphorus concentration of the soil samples was measured using HF-HClO4-HNO3-ICP-OES: Weighed 250 mg of soil sample and placed it in a four-fold polyethylene cover, added 5 ml HNO3, 2 ml HF, and 1 ml HClO4 for digestion using a graphite digestion instrument, and determined the total phosphorus concentration using an ICP mass spectrometer mentioned above. Nutrient resorption efficiency and proficiency At the species level, the nutrient resorption efficiency for N (NRE) and P (PRE) was calculated as the percentage of nutrient concentration reduced in senesced leaves (Nusen) relative to that in green leaf (Nugr) on a mass basis. The nutrient resorption efficiency (NuRE) was calculated using the formulae as follows [4]: NuRE, where Nugr and Nusen are the mass-based nutrient concentrations in green leaves and senescent leaves, and MLCF is the mass loss correction factor (the ratio of dry mass of senescent leaves to dry mass of green leaves). The MLCF of the woody plants and herbs used here were 0.784 and 0.713, respectively [4]. Leaf N and P resorption proficiency was expressed as senesced leaf N and P concentrations, respectively, with low resorption proficiency for plants producing higher senescent leaf nutrients. Data analysis Data on senesced leaf N and P concentrations were log10-transformed to satisfy normal distribution. One-way ANOVA was used to test whether green leaf N and P concentrations, N and P resorption efficiency and senesced leaf N and P concentrations were significant differences between plant functional types, including non-legume herbs, non-legume shrubs, and legumes, followed by Tukey’s HSD tests. The relationships between NuRE and abiotic factors such as soil N, P, pH, sand content, MAT, and MAP were tested using ordinary least squares (OLS) regression. Similarly, OLS was used to examine the relationship between NuRE and biotic factors such as N and P concentrations, SLA, and LDMC of green leaves. Linear regression was used to test whether there were significant relationships between nutrient resorption efficiency and senescent leaf N and P concentrations between PFTs and differences in their slopes. We carried out the principal components analysis (PCA) across leaf economic traits, including SLA, LDMC, leaf N and P concentrations, and then obtained the first axis of PCA (leaf PC1), representing the leaf economics spectrum. PCA was conducted on soil properties, including soil N, P, N/P, pH, and sand content, and on climate variables, including MAT and MAP, in which the first axes (soil PC1 and climate PC1) represented the soil physico-chemical characteristics and climate, respectively. Similarly, the OLS regressions were used to test the relationships between leaf PC1, soil PC1, climate PC1 and nutrient resorption efficiency and senesced leaf N and P concentrations. We used linear mixed-effects models to assess the influences of fixed terms including climate, soil, and leaf traits and random terms including plant species and site on leaf nutrient resorption efficiency and senesced leaf N and P concentrations for each PFT. The linear mixed-effects model was carried out using the lmer function of the lme4 package, in which the marginal and conditional R2, representing fixed and mixing effects, respectively, were calculated using the r.squaredGLMM function of the MuMIn package. The significance of fixed and random effects was tested using the Anova function of the car package and the ranova function of lmerTest package, respectively. Structural equation modeling (SEM) was used to assess the direct and indirect effects of influential climate, soil, and leaf traits on leaf N and P resorption efficiency and senesced leaf N and P concentrations for each PFT using AMOS software (IBM SPSS AMOS v21). All statistical analyses were conducted using the R software [39], and the significance level was set at p < 0.05 except for the χ2 test in the SEM at p > 0.05. Results N and P concentrations of green and senesced leaves and leaf nutrient resorption The N and P concentrations of green leaves in the arid and semiarid ecosystems were 26.02 mg g− 1 and 1.29 mg g− 1, respectively, and there were no significant differences among different PFTs (Table 1). N and P resorption efficiencies of leaves were 62.49% and 68.13%, respectively (Table 1). Legumes had much lower leaf NRE than non-legume PFTs (45.17% vs. 66.77% and 64.75%), non-legume shrubs and legumes shared lower leaf PRE than non-legume herbs (60.71% and 59.43% vs. 72.48%). Senesced leaf N and P concentrations were 12.83 mg g− 1 and 0.52 mg g− 1, respectively. Legumes had higher senesced leaf N than did non-legume herbs and non-legume shrubs, and legumes and non-legume shrubs had higher senesced leaf P than did non-legume herbs (Table 1). Table 1N and P concentrations of green and senesced leaves and N and P resorption efficiency overall and for different plant functional types (PFT). Values are presented as mean ± standard error. Different letters indicate significant differences between PFTs at p < 0.05Concentration (mg g− 1)Resorption efficiency (%)Senesced leaves (mg g− 1)nNPNPNPAll data23326.02 ± 0.491.29 ± 0.0462.49 ± 1.0168.13 ± 0.8312.83 ± 0.410.52 ± 0.02Non-legume herbs15125.23 ± 0.60a1.28 ± 0.05a66.77 ± 1.08a72.48 ± 0.80a11.36 ± 0.44b0.47 ± 0.02bNon-legume shrubs4427.54 ± 1.15a1.29 ± 0.09a64.75 ± 2.15a60.71 ± 2.10b12.62 ± 0.82b0.60 ± 0.05aLegumes3827.40 ± 1.18a1.34 ± 0.09a45.17 ± 2.11b59.43 ± 2.17b18.92 ± 1.03a0.65 ± 0.04a N and P concentrations of green and senesced leaves and N and P resorption efficiency overall and for different plant functional types (PFT). Values are presented as mean ± standard error. Different letters indicate significant differences between PFTs at p < 0.05 Relationships between nutrient resorption and climate and soil variables The leaf NRE of non-legume herbs and shrubs decreased with increasing MAT and the senesced leaf N concentrations of non-legume herbs and shrubs increased with increasing MAT (Fig. S2a, c). The leaf PRE of non-legume herbs decreased with increasing MAT, and the senesced leaf P concentration of non-legume herbs increased with increasing MAT (Fig. S2b, d). The senesced leaf N concentration of non-legume herbs decreased with increasing MAP, whereas the P concentration in the senesced leaves of legumes increased with increasing MAP (Fig. S2g, h). The PCA was conducted on the two climatic factors, in which climate PC1 captured 71.6% of the variance (Fig. S3). The climate PC1 was negatively correlated with MAP and positively correlated with MAT (Table S2). This represented a decreasing drought gradient (Fig. S3). The leaf NRE of non-legume herbs was negatively correlated with climate PC1, and the N concentration in senesced leaves of non-legume herbs was positively correlated with climate PC1 (Fig. 2a, b). The P concentration in the senesced leaves of legumes was negatively correlated with climate PC1 (Fig. 2d). Relationships between climate and leaf N and P resorption efficiency (NRE and PRE) and senesced leaf N and P concentrations for each PFT. NRE and climate PC1 (a), senesced leaf N and climate PC1 (b), PRE and climate PC1 (c), senesced leaf P and climate PC1 (d). The solid line denotes a significant relationship at p < 0.05. The dashed line denotes a marginally significant relationship at 0.05 < p < 0.1. The red, green, and blue colors represent non-legume herbs, non-legume shrubs, and legumes, respectively The leaf NRE and senesced leaf N in non-legume shrubs were marginally correlated with soil N, whereas leaf NRE in legumes was marginally positively correlated with soil N (Fig. S4a, b). The N concentration in the senesced leaves of non-legume herbs was marginally negatively correlated with soil P (Fig. S4d). Leaf NRE and the P concentration in senesced leaves of non-legume shrubs were positively correlated with sand content in the soil (Fig. S4e, S5f). The senesced leaf N in legumes marginally significantly increased with soil pH (Fig. S4h).The P concentration in the senesced leaves of non-legume herbs was positively correlated with the soil sand content (Fig. S5f). The P concentration in the senesced leaves of non-legume types was positively correlated with the soil pH (Fig. S5h). The PCA was conducted on the five soil factors, which captured 59.8% of the variance in the first component (Fig. S6). Soil PC1 was negatively correlated with soil sand and pH and positively correlated with soil N, P, and N/P (Table S2). This indicates an increasing fertility gradient (Fig. S6). The leaf NRE of legumes was positively correlated with soil PC1, but leaf NRE of non-legume shrubs was marginally negatively correlated with soil PC1 (Fig. 3a). The P concentrations in the senesced leaves of non-legume herbs and shrubs were negatively correlated with soil PC1 (Fig. 3d). Relationships between soil and leaf N and P resorption efficiency (NRE and PRE) and senesced leaf N and P concentrations for each PFT. NRE and soil PC1 (a), senesced leaf N and soil PC1 (b), PRE and soil PC1 (c), senesced leaf P and soil PC1 (d). The solid line denotes a significant relationship at p < 0.05. The dashed line denotes a marginally significant relationship at 0.05 < p < 0.1. The red, green, and blue colors represent non-legume herbs, non-legume shrubs, and legumes, respectively Relationships between leaf nutrient resorption and leaf traits The leaf NRE and PRE of non-legume shrubs were negatively correlated with SLA (Fig. S7a, e). The N concentrations in the senesced leaves of non-legume herbs and shrubs were positively correlated with SLA (Fig. S7b). The N and P concentrations in the senesced leaves of non-legume herbs were negatively correlated with LDMC (Fig. S7d, h). The P concentration in the senesced leaves was (marginally) positively correlated with SLA for each PFT (Fig. S7f). The leaf NRE of non-legume herbs and shrubs increased with increasing leaf N concentrations, but there was no significant relationship in legumes (Fig. S8a). The N concentration in the senesced leaves increased with leaf N concentration for each PFT (Fig. S8b). Leaf PRE and P concentrations in senesced leaves increased with the green leaf P concentration for each PFT (Fig. S8c, d). The PCA was conducted on the four leaf traits, which captured 48.8% of the variance in the first component (Fig. S9). Trait PC1 was negatively correlated with leaf N, P, and SLA, and positively correlated with LDMC (Table S2). It represented a strategy from the resource-acquisitive to resource-conservative end (Fig. S9). The N and P concentrations in the senesced leaves were negatively correlated with trait PC1 for each PFT (Fig. 4b, d). The leaf PRE of non-legume shrubs and legumes were (marginally) negatively correlated with leaf trait PC1 (Fig. 4c). Relationships between the first axis of principal component analysis representing the plant economics spectrum from the resource-acquisitive to resource-conservative end and leaf N and P resorption efficiency (NRE and PRE) and senesced leaf N and P concentrations for each PFT. NRE and trait PC1 (a), senesced leaf N and trait PC1 (b), PRE and trait PC1 (c), senesced leaf P and trait PC1 (d). The solid line denotes a significant relationship at p < 0.05. The dashed line denotes a marginally significant relationship at 0.05 < p < 0.1. The red, green, and blue colors represent non-legume herbs, non-legume shrubs, and legumes, respectively Effects of abiotic and biotic factors on leaf N and P resorption Linear mixed-effect models showed that fixed and random effects affected leaf N and P resorption differently (Table 2, S3). Climate PC1 affected the leaf NRE in non-legume herbs and non-legume shrubs, and the leaf PRE in legumes (Table 2). Soil PC1 predominantly regulated the leaf NRE in non-legume shrubs and legumes, whereas leaf trait PC1 predominantly regulated the leaf PRE in non-legume shrubs and legumes (Table 2). Leaf trait PC1 predominantly controlled senesced leaf N and P for each PFT (Table 2). Climate PC1 was the second most important predictor of senesced leaf N in non-legume herbs and senesced leaf P in legumes and soil PC1 affected the senesced leaf N in non-legume shrubs (Table 2). Table 2Results of linear mixed-effects models assessing effects of climate, soil, and leaf traits on N and P resorption efficiency (NRE and PRE) and senescent leaf concentrations of N and P. values in bold denote significant terms at p < 0.05; italicized p values denote marginally significant terms at 0.05 < p < 0.1PFTNREPRESenesced leaf NSenesced leaf Pχ2 p χ2 p χ2 p χ2 p Non-legume herbsClimate PC16.34 0.012 2.88 0.090 11.10 0.001 0.650.421Soil PC11.010.3151.830.1760.800.3720.070.789Trait PC12.490.1152.91 0.088 15.27 < 0.001 44.22 < 0.001 Marginal R20.060.030.210.31Conditional R20.450.340.690.66Non-legume shrubsClimate PC14.35 0.037 0.040.8410.990.3200.030.852Soil PC19.10 0.003 0.440.5086.74 0.009 0.300.583‘Trait PC13.46 0.063 3.86 0.050 6.17 0.012 10.18 < 0.001 Marginal R20.240.110.180.20Conditional R20.590.110.450.39LegumesClimate PC10.000.9624.31 0.038 0.060.8135.78 0.016 Soil PC199.52 < 0.001 1.310.2530.960.3270.130.716Trait PC114.95 < 0.001 3.56 0.059 10.65 0.001 20.99 < 0.001 Marginal R20.290.170.190.44Conditional R21.000.630.570.72 Results of linear mixed-effects models assessing effects of climate, soil, and leaf traits on N and P resorption efficiency (NRE and PRE) and senescent leaf concentrations of N and P. values in bold denote significant terms at p < 0.05; italicized p values denote marginally significant terms at 0.05 < p < 0.1 The SEM showed that climate PC1 had a direct negative effect on the leaf NRE in non-legume herbs (Fig. 5a; Table S4). Leaf trait PC1 had a direct negative effect on the leaf PRE in non-legume shrubs and legumes (Fig. 5c, e), whereas climate PC1 had a positive effect on the leaf PRE in legumes (Fig. 5e). Soil PC1 had a direct negative effect on the leaf NRE of non-legume shrubs but a direct positive effect on the leaf NRE of legumes (Fig. 5c, e). Climate PC1 increased the senesced leaf N of non-legume herbs, but decreased the senesced leaf P of legumes (Fig. 5b, f). Leaf trait PC1 consistently decreased the N and P concentrations in senesced leaves for each PFT (Fig. 5b, d, f). Soil PC1 increased the senesced leaf N in non-legume shrubs (Fig. 5d). There was a strong relationship between leaf NRE and PRE and between senesced leaf N and P concentrations for each PFT (Fig. 5). Structural equation modeling on direct and indirect effects of climate, soil, and leaf traits on N and P resorption and proficiency for each PFT. The red and blue lines denote positive and negative effects, respectively. a-b: non-legume herbs, c-d: non-legume shrubs, e-f: legumes. The dashed line denotes a non-significant path at p < 0.05 Relationship between leaf nutrient resorption efficiency and proficiency Leaf NRE was significantly negatively correlated with senesced leaf N for non-legume herbs, non-legume shrubs, and legumes (Fig. 6a; R2 = 0.59, 0.56, and 0.34, all p < 0.001, respectively), and legumes had a relatively weaker coordination than the other non-legume PFTs. Leaf PRE decreased with increasing senesced-leaf P concentrations (Fig. 6b; R2 = 0.26, 0.19, and 0.22, p < 0.001, 0.003, and 0.003, respectively). Relationships between leaf NRE and senesced leaf N and between leaf PRE and senesced leaf P for each PFT. NRE and senesced leaf N (a), PRE and senesced leaf P (b). The solid line denotes a significant relationship at p < 0.05. The red, green, and blue colors represent non-legume herbs, non-legume shrubs, and legumes, respectively Discussion In this study, we investigated the effects of biotic and abiotic drivers on leaf nitrogen and phosphorus resorption efficiency and proficiency in different functional types in the arid and semi-arid regions. Leaf nutrient resorption efficiency is higher in arid and semi-arid regions with poor soil fertility and low water availability, emphasizing the importance of nutrient resorption in harsh drylands. Plants with resource-conservative strategies have a high capacity for nutrient resorption, which improves the resilience of plant populations in nutrient-poor drylands. Lower nitrogen resorption efficiency by legumes benefits them efficiently fix exogenous nitrogen and rapidly release it to the soil, improving the entry of N induced by N-fixing plants into N-limited drylands. The negative correlation between nutrient concentration in senescing leaves and nutrient resorption efficiency emphasizes the importance of nutrient resorption in ecosystem nutrient cycling. N limitation improves the link between NRE and senescent leaf N for non-legume functional types in arid and semiarid ecosystems. Differences in leaf N and P resorption among PFTs According to the criteria of leaf resorption proficiency, plants are highly proficient in nutrient resorption if they reduce the concentrations of N and P in senescing leaves to < 7 mg g− 1 and < 0.5 mg g− 1, respectively [7]. Non-legume herbs were highly proficient in P resorption, whereas the other two PFTs were not highly proficient in leaf N and P resorption. Research has shown that under harsh environmental conditions, plants can effectively complete the nutrient absorption process by accelerating the transfer of mobile nutrients, thereby reducing their dependence on the ecological acquisition of nutrients [40]. We found that NRE and PRE were 62.49% and 68.13% in these arid and semiarid regions. This is similar to the global NRE and PRE (62.1% and 64.9%, respectively) [4]. Similarly, Lü & Han (2010) reported that leaf NRE of Agropyron cristatum, Achnatherum sibiricum, Stipa grandis, Potentilla bifurca are 57.8%, 67.3%, 61.4%, 57.0% and PRE of them were 60.0%, 56.9%, 59.0%, 64.4% under controlled conditions in typical steppes, respectively [28]. For tree species Populus euphratica in arid regions of Xinjiang Province of China, leaf NRE was 55.5% (female individual) and 62.8% (male individual), and leaf PRE was 60.8% (female individual) and 55.27% (male individual). Nutrient resorption efficiency in arid regions is much higher than that in the subtropical forests (101 species) [33] and in forests of eastern China (172 species) [41]. Altogether, in drylands characterized by poor soil fertility and low water conditions, plants develop a higher nutrient resorption efficiency to reduce their dependence on the surrounding soil nutrients, emphasizing that plants utilize absorbed nutrients from senescing leaves as an efficient strategy for nutrient conservation in harsh environments [6, 42]. Nutrient resorption differs significantly among plant functional types in arid and semiarid regions. We found that non-legume herbs and non-legume shrubs had high nutrient resorption efficiency and nutrient resorption proficiency, with relatively lower senesced leaf nutrients. Similarly, in a global meta-analysis, NRE and PRE of forbs were 70% and 77.4%, respectively [4]. Zhao et al. (2017) reported similar leaf NRE (73.5–74.4%) and PRE (72.6–87.4%) of herbs in the alpine grasslands of the Tibetan Plateau [17]. The high resorption characteristics of non-legume herbs and non-legume shrubs can reduce nutrient loss during leaf senescence and make plants less dependent on soil nutrients [43]. In addition, fast-growing plants have higher nutrient requirements. The P-resorption efficiency of non-legume herbs was significantly higher than that of non-legume shrubs. A previous study showed that fast-growing grasses have higher N resorption efficiency than slow-growing shrubs [44]. As one of the particular plant types, legumes can convert exogenous N into a form that plants use via N-fixing nodules [45, 46]. Therefore, a strong biological N fixation ability promotes the entry of N induced by N-fixing plants into the ecosystem. The N concentration in senesced leaves was half higher in N-fixers than in non-N-fixers in present and previous studies (Table 1) [4]. We found that legumes had much higher senesced leaf N and lower NRE than non-legumes, supporting our first hypothesis that the leaf nitrogen resorption of N-fixing species is lower than that of non-N-fixing species. Legumes fix N2 from the air, resulting in lower dependence on soil N availability [17]. Previous studies have shown that legumes have relatively high decomposition rates [47]. Those suggest that legumes can regulate soil N cycling by increasing N mineralization and N pools in nutrient-poor arid and semiarid ecosystems [48]. Leaf P resorption in legumes is comparable to that of non-legume shrubs, supporting the viewpoint that N resorption may be independent of leaf P resorption, which would be advantageous for N-fixing plants that have an increased demand for P in the energy-intensive process of N fixation [21]. Effects of climate and soil on leaf nutrient resorption We found that climatic factors differentially regulated leaf N and P resorption in non-legume herbs and legumes from the arid and semiarid regions (Fig. 5; Table S2). Climate PC1, particularly MAT, had a direct negative effect on leaf N resorption in non-legume herbs. Similarly, Tong et al. (2021) demonstrated that leaf N resorption in Chinese fir (Cunninghamia lanceolata (Lamb.)) decreased with increasing MAT across subtropical forests [49]. Vergutz et al. (2012) suggested that leaf N resorption decreases with MAT globally [4]. These findings highlight the general regulation of leaf nutrient resorption by MAT. In contrast, climate PC1, particularly decreasing MAP, decreased leaf P resorption in legumes, which was characterized by increasing PRE and decreasing senesced leaf P. Moreover, MAP had an indirect positive effect on senesced leaf P by shaping leaf traits. Similarly, Renteria & Jaramillo (2011) report that leaf resorption of P rather than N increases in years with low precipitation in a tropical dry forest in Mexico [50]. These suggest that water conditions determine P dynamics in dryland ecosystems [50]. Legumes have the physiological advantage of higher N via tight stomatal regulation of transpiration [51]. Thus, MAP can mitigate the P limitation of legumes and decrease leaf P nutrient resorption by improving P solubilization and acid phosphatase in soils [52, 53]. These results suggest that MAT and MAP differentially regulate leaf nutrient resorption in legumes and non-legumes owing to contrasting N and P limitations. Plants absorb nutrients from the soil or from their senescent tissues, a process involving a series of cost-related mechanisms [12, 34]. Soil nutrient conditions, which represent the amount of nutrients available to plants, affect plant nutrient resorption. In general, there is a trade-off between the cost of uptake resorption from the soil and the cost of nutrient resorption from leaves [30, 54]. Thus, plants cannot reabsorb nutrients from soils with poor nutrient content in arid and semiarid regions. Leaf nutrient resorption in non-legume shrubs decreased with an increase in soil fertility. Similarly, Zhao et al. (2017) demonstrated that soil N is negatively correlated with leaf N resorption in non-legumes along the precipitation gradient of the Tibetan Plateau (204–525 mm) [17]. However, leaf nutrient resorption in non-legume herbs did not respond to soil conditions. These findings partially support the second hypothesis that leaf N and P resorption is regulated by soil properties in arid and semiarid ecosystems. This is consistent with the finding that there is no relationship between total soil N and the nutrient resorption efficiency of two herbaceous Stipa species (S. grandis and S. krylovii) [55]. Urbina et al. (2021) reported a weak relationship between soil nutrients and leaf nutrient resorption in old tropical forests. Surprisingly, the leaf NRE of legumes increased with increasing soil fertility [56]. Similarly, there was an increasing trend in leaf NRE of legume species along a precipitation gradient on the Tibetan Plateau [17]. These suggest that the genetic difference between legume and non-legume species is a determinant of the capacity of legumes to resorb N under high fertility conditions, which may suppress nodulation and N fixation [21]. There was no relationship between soil fertility and senesced leaf N and P concentrations in non-legume herbs and legumes. Altogether, the relationships between nutrient resorption and nutrient status in the soil vary to some extent, depending on the biome, plant functional type, and other environmental conditions (i.e., input of precipitation). In arid and semiarid regions, soil nitrogen cycling is restricted because of low nutrient availability [57]. Thus, nutrients can be preserved in green biomass through internal nutrient cycling in nutrient-poor arid and semiarid regions [58]. Leaf economic traits regulate leaf nutrient resorption We found positive correlations between leaf nutrient status and nutrient resorption efficiency for each PFT, except for a lack of a relationship between leaf N and NRE in legumes (Fig. S6). This contradicts previous findings regarding the negative relationships between leaf nutrient status and global resorption efficiency [4, 31]. This evidence points to the positive influence of plant nutrient status on nutrient resorption efficiency in infertile habitats [6, 37]. Our results show a significant negative correlation between leaf PRE of non-legume shrubs and legumes and leaf trait PC1 representing the resource-acquisitive to resource-conservative strategy (Fig. 4c). Similarly, Ma et al. (2024) suggested nutrient resorption efficiency is greater in deciduous woody species with resource-acquisitive strategy [33]. Interestingly, we found a strong negative correlation between nutrient resorption proficiency and leaf economic spectrum, implying that nutrient resorption capacity is greater in resource-conservative species (Fig. 4). Our study also proves that plants with a resource-conservative strategy in arid and semiarid regions have stronger nutrient resorption proficiency for any PFTs, partially supporting the hypothesis that plant species with a resource-conservative strategy are more efficient for leaf nutrient resorption. This finding is consistent with that of a previous report on the herbaceous genus Helianthus [59]. The inconsistent findings between leaf economics spectrum and nutrient resorption efficiency and between the leaf economics spectrum and nutrient resorption proficiency indicate that the stable feature of nutrient resorption proficiency linking plant strategies can be a better metric for qualifying nutrient resorption capacity [7]. Together, these findings underpin the evidence that nutrient resorption proficiency, rather than nutrient resorption efficiency, is mainly controlled by plant economics spectrum [30]. Joswig et al. (2022) suggested that the joint effects of climate and soil variables and the independent soil effect regulate leaf economic traits globally [60]. Relatively fast warming trends over drylands and accelerated dryland expansion would result in soil degradation and reduced soil moisture [61]. Altogether, plant species in arid and semiarid ecosystems tend to evolve adaptively towards resource-conservative species with higher nutrient resorption proficiency. Linkage between leaf nutrient resorption and nutrient cycling There was a significant negative correlation between leaf nutrient resorption efficiency and senesced leaf nutrient concentration (Fig. 6), highlighting the supplementary features of nutrient resorption efficiency and proficiency in qualifying plant nutrient resorption. The link between NRE and senesced leaf N resorption is relatively weaker for legumes than for non-legume functional types because the N resorption efficiency of legumes with N-fixing capacity is less sensitive to lower soil N conditions than that of non-legumes. In addition, the link between NRE and senesced leaf N was stronger than that between PRE and senesced leaf P in non-legume types in this N-limited dryland ecosystems (Fig. 6). In N-limited mangrove ecosystems, the relationship between NRE and PRE tends to be stronger than that between PRE and senesced leaf P (R2 = 0.98 vs. 0.82; Lin et al., 2010). In the P-limited Mediterranean shrubland, the intensity of negative correlation between NRE and senesced leaf N and between PRE and senesced leaf P was similar [37].These imply that N or P limitation regulates the relationship between nutrient resorption efficiency and proficiency. The negative correlation between nutrient resorption and proficient nutrient in senesced leaves suggests a trade-off between the resorption and decomposition pathways at the ecosystem level, with the former being more efficient than the latter [12]. Thus, a trade-off between N resorption efficiency and litter N mineralization has recently been revealed [13–15]. This emphasizes the importance of nutrient resorption in ecosystem nutrient cycling [12]. Conclusions The results of this study suggest that ecological strategies and N-fixing plant types modulate nutrient resorption. Legumes have lower nitrogen resorption efficiency, enabling their crucial roles in semiarid and arid climates by efficiently fixing exogenous nitrogen and rapidly releasing it into soils. In arid and semiarid regions, leaf nitrogen and phosphorus resorption is mainly regulated by the nutrient status of the leaves and soil. Plants with a resource-conservative strategy are highly proficient in nutrient resorption, which improves the fitness of plant populations in nutrient-poor drylands. We highlight the importance of leaf traits in regulating leaf nutrient resorption in drylands in the context of global climate change, potentially modulating leaf economic traits and soil nutrient conditions [60, 61]. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Acknowledgements This research was funded by the Strategic Priority Research Program of the Chinese Academy of Sciences (XDA26010101). We are particularly grateful to the anonymous reviewers for their highly relevant comments on the manuscript. Author contributions GL and ZH designed the experiment. JZ and YW performed the field sampling. JZ and YW carried out laboratory measurements. JZ and GL wrote the first version of the manuscript. All authors contributed to the clarification and quality of the manuscript. Funding This research was funded by the Strategic Priority Research Program of the Chinese Academy of Sciences (XDA26010101). Data availability The data supporting the conclusions of this article is included in the article and its supplementary files. Declarations Ethics approval and consent to participate The collection of leaf and soil samples was approved by the Forestry Bureau of Ordos City, Inner Mongolia Autonomous Region, and Ningxia Hui Autonomous Region. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Supplementary Material 2 Data Availability Statement The data supporting the conclusions of this article is included in the article and its supplementary files.