Climate data, including mean annual precipitation, mean annual temperature, and elevation, were extracted from WorldClim 2 (Fick and Hijmans, 2017). Soil pH data were obtained from the World Soil Database (WISE30sec: World soil property estimates at a nominal resolution of 30 by 30 arc sec) (Batjes, 2015), and the pH of the 0–20 cm surface soil was used to represent soil pH.

The N deposition data were extracted from the global N deposition dataset of Ackerman et al. (2019). The P deposition data were collected through a literature search combined with extrapolation (more details are provided in note S1). All selected predictors are summarized in Table 1 .

Independent analytical steps were used to quantitatively and qualitatively assess the impact of global change on the NPP of terrestrial ecosystems and identify the main driving factors. All environmental predictors were unified to a 1 km spatial scale using Kriging interpolation.

Based on our dataset, the mean NPP of global terrestrial ecosystems was 752 g C·m⁻²·yr⁻¹, with the EB category exhibited the highest NPP at 1,016 g C·m⁻²·yr⁻¹, and HB exhibited the lowest NPP at 531 g C·m⁻²·yr⁻¹ ( Table 2 ). Global N and P deposition were 8.18 and 0.29 kg·ha⁻¹·yr⁻¹, respectively, with CO being exposed to the lowest P deposition and generally lower leaf N and P contents than in the other PFTs (P < 0.001). In contrast, DB exhibited the highest N and P deposition and higher leaf N and P content (P < 0.001). The HB category generally received more N and P deposition (P < 0.001) and exhibited higher inherent leaf N and P contents than those in woody plants (EB, DB, and CO).

In all PFTs, both individually and at a global scale, both univariate and multivariate LMEs confirmed that NPP was correlated non-linearly with N and P deposition, except in the HB category ( Supplementary Tables S1 and S2 , Table 3 , Figure 1 ). In univariate LMEs, under N deposition, the NPP thresholds of EB, DB, CO, and all four PFTs together at a global scale were 8.71, 9.33, 8.32, and 8.32 kg·ha⁻¹·yr⁻¹, respectively ( Figure 1 ). However, a positive correlation was observed between NPP and N and P deposition in the HB ( Figure 1 ). Before reaching NPP_max, the tipping points of foliar N (N_max) and foliar N/P ratio (N/P_max) of EB, DB, and all four PFTs at a global scale were observed sequentially ( Figure 1 , Supplementary Figure S2 ). Under P deposition, the tipping points of the NPPs of EB, CO, and all four PFTs at a global scale were 0.30, 0.20, and 0.25 kg·ha⁻¹·yr⁻¹, respectively ( Figure 1 ). However, only limited threshold effects of P deposition were observed on leaf nutrient contents ( Supplementary Figure S3 ).

Our global maps reveal a clear pattern of N and P deposition ( Figure 2 ). In this pattern, the hotspots of N deposition were mainly in eastern Asia, Europe, eastern North America, and southern Brazil ( Figure 2A ). The highest N deposition occurred in central China at 27.57 kg·ha⁻¹·yr⁻¹. Atmospheric P deposition is strongly associated with deserts and semiarid regions ( Figure 2B ), mainly in North Africa, India, northern Australia, and northern China. The highest P deposition was 0.96 kg·ha⁻¹·yr⁻¹ seen in the Sahara Desert of North Africa.

At a global scale, the impacts of N and P deposition on the NPP of terrestrial ecosystems show clear N and P limitation patterns. Globally, 53.68 and 43.88% of the terrestrial ecosystem plants were negatively affected by N and P deposition, respectively ( Figures 2C, D ). In Europe, N deposition mainly exhibited an inhibitory effect on NPP ( Figure 2C ); however, relatively low P deposition in these regions alleviated P limitation and promoted vegetation growth ( Figure 2D ). The east coast of the United States also revealed that N deposition adversely affected NPP, whereas P deposition had a positive effect ( Figures 2C, D ). It is worth noting that in China, excessive N and P deposition significantly inhibited ecosystem NPP ( Figures 2C, D ).

Globally, NPP showed significant correlations with biophysical stand properties and site quality indicators in both multivariate LMEs and RFMs in all four PFTs ( Table 3 ). Overall, soil pH correlated negatively with NPP ( Table 3 , Figure 3 ), through this relationship varied under nutrient-restricted conditions ( Supplementary Table S3 ). Significant correlations between leaf N and P contents and NPP were observed only in HB ( Table 3 ). In univariate LMEs, elevation did not exhibit a threshold effect but correlated negatively with the NPP of HB, EB, DB, and all four PFTs at a global scale ( Supplementary Table S1 ). However, more specific relationships between elevation and NPP were found in multivariate LMEs, except for the EB and DB categories ( Table 3 ).

The SEMs and RFMs based on the nutrient limitation hypothesis showed that PFTs under N limitation (N/P ≤10) exhibited an apparent positive response to the fertilization effects of N deposition, reflected by the positive correlation between NPP and N deposition. In this case, P deposition was negatively correlated with NPP. Under P limitation (N/P ≥20), the opposite was true for N limitation ( Figure 4 ). In addition, under the N/P balance (10< N/P <20), N and P deposition correlated negatively with NPP ( Figure 4 ). The RFMs revealed that P deposition was relatively more important than N deposition determining NPP under N/P balance and P limitation ( Figures 4C, F, I ). Among the PFTs, the response patterns of NPP to N and P depositions were relatively diverse. Specifically, in HB, under both N and P limitation, N and P deposition correlated positively and negatively, respectively, with NPP ( Supplementary Table S3 , Supplementary Figure S4 ); however, the results were not significant under N limitation alone (P >0.1). In EB, DB, CO, and all four PFTs together at a global scale, NPP exhibited similar response patterns to N and P deposition under the nutrient limitation hypothesis ( Supplementary Table S3 , Supplementary Figures S5–S7 ).

Additionally, the site quality indicators, soil pH and elevation, were strong predictors of NPP in the SEMs ( Supplementary Table S3 ). Soil pH correlated negatively with NPP (P < 0.001). However, the elevation did not show definite patterns in the SEMs ( Supplementary Table S3 ). In addition, our SEMs did not detect any direct correlations between foliar N and P contents and NPP, except in HB, where several pieces of correlated evidence were observed ( Supplementary Table S3 ).

Elucidating the effects of atmospheric N and P deposition on plant growth is crucial for biodiversity conservation and the maintenance of landscape heterogeneity. Numerous theoretical and empirical studies have demonstrated a “critical load” threshold in the response of plants to N deposition worldwide (Henttonen et al., 2017; Gentilesca et al., 2018; Rohner et al., 2018; Etzold et al., 2020). The most affected countries are the United States, Western Europe, Eastern Europe, South Asia, East Asia, Southeast Asia, and Japan (Dentener et al., 2006). Herein, we tested the non-linear relationships between atmospheric N and P deposition and terrestrial productivity (NPP) in the four PFTs. As expected, we found that N and P deposition non-linearly affected the NPP of woody plants globally. Studies have provided evidence for this threshold effect based on the perspectives of plant physiology and biochemistry. For example, moderate N deposition can promote plant growth; however, excess N deposition may inhibit plant photosynthesis (Li et al., 2022; Lu et al., 2023) by reducing stomatal conductance (Yao et al., 2016), chlorophyll content (Du et al., 2017), and enzyme activity (Chen et al., 2013). In addition, excess N deposition may change plant hydraulic architecture, thus affecting growth (Wang et al., 2016). Several N addition and plot control experiments (Braun et al., 2017; Salehi et al., 2021) have indicated that soil acidification and nutrient imbalances caused by excessive N deposition can inhibit plant growth. For example, in Italy and Europe, N deposition exhibited threshold effects on beech growth rates at 20 and 30 kg N·ha⁻¹·yr⁻¹ concentrations, respectively (Gentilesca et al., 2018; Etzold et al., 2020). A global meta-analysis based on ¹⁵N tracking also revealed that the N retention efficiency of forests begins to decline at 10–15 kg N·ha⁻¹·yr⁻¹, and N saturation is reached after exceeding 40–50 kg N·ha⁻¹·yr⁻¹ (de Vries et al., 2014a). Furthermore, when the N deposition in the temperate grasslands of North America (Clark and Tilman, 2008) and Inner Mongolia (Bai et al., 2010) reached 10 kg N·ha⁻¹·yr⁻¹ and 17 kg N·ha⁻¹·yr⁻¹, respectively, ecosystem productivity and biodiversity declined.

In contrast to the widespread attention paid to N deposition, the threshold effects of P deposition on vegetation growth have less frequently been reported. Recently, at the macro scale, it has been demonstrated that P limitation is not limited to tropical forests but is widely distributed globally, and its spatial pattern varies across continents, altitudes, climate zones, and aridity (Hou et al., 2020). Global-scale meta-analyses have suggested that P limitation exceeds N limitation to become the primary nutrient limitation in natural ecosystems (Du et al., 2020). At the microscopic scale, studies have shown that moderate P addition can significantly change the soil arbuscular mycorrhizal fungal community composition, leading to a decrease in mycorrhizal diversity (Camenzind et al., 2014; Duenas et al., 2020) and alleviate the inhibition of N deposition on photosynthesis (Hao et al., 2023). In this study, we demonstrated that P deposition also exhibited a threshold effect on plant growth. Our study further suggest an increasing P limitation in global terrestrial ecosystems, likely because of a relatively high atmospheric N/P deposition ratio, which enhances P limitation in the ecosystem (Duenas et al., 2020). Nevertheless, our results suggest that accounting for N and P deposition in a multiscale theoretical framework can provide a sound basis for sustaining terrestrial C sequestration. However, due to the small volume of atmospheric P deposition before 2000 (no more than 1.5 kg P·ha⁻¹·yr⁻¹, Pan et al., 2021), the tipping point of NPP caused by P deposition in this study may be smaller, but the observed threshold effect is real. In future studies, it is necessary to collect more samples of P deposition for model fitting.

The sequential threshold effects of leaf N/P ratio, leaf N and P contents, and NPP on N deposition at the leaf and individual plant scales indicate that the effects of N deposition on plant growth may be gradual and accumulated and that leaf nutrient status may be associated with plant growth. However, our results revealed that leaf N and P concentrations were significantly correlated with the NPP of HB alone ( Supplementary Table S3 ). This finding can be attributed to the rapid growth and metabolic rate in HB, maintained by several ribosomes and intensive protein synthesis, leading to an accumulation of rRNA in the cell and correspondingly higher P content and lower N/P ratio (Harris, 2003). Therefore, HB is more likely to show higher demand for N and excess P under atmospheric N and P fertilization. Our results support the productivity–nutrient allocation hypothesis, which states that herbs allocate relatively more nutrients for growth, whereas woody plants additionally allocate nutrients for morphological development (Tang et al., 2018); thus, the leaf nutrient concentrations of HB were more closely related to NPP.

N and P are the major limiting nutrients in ecosystems worldwide; thus, ecosystems can be generally classified as N- or P-limited or N/P-balanced ecosystems (Li et al., 2016; Zhu et al., 2022). For instance, N addition can significantly increase foliar N concentrations, and subsequently, P in plant leaves is assimilated to promote plant growth under N-limited conditions (Cleveland et al., 2013). However, under P-limited conditions, plants increase their P demand and uptake to maintain the stability of their internal chemical stoichiometry when foliar N concentrations are excessively increased by N addition (Marklein and Houlton, 2012). In addition, PFTs can regulate the stoichiometry of foliar N and P in response to N addition because of the different biological and ecological characteristics required for adapting to environmental changes (Townsend and Asner, 2013; Tian et al., 2016). For instance, herbaceous plants exhibit lower nutrient use efficiency but higher plant growth rates and plasticity, with relatively more sensitivity to N addition than observed for woody plants (Gilliam et al., 2016; Tian et al., 2016). A global meta-analysis based on the N/P threshold hypothesis indicated that increasing N deposition may aggravate plant P limitation under N-limited conditions but improve P limitation under P-limited conditions, and the effects are relatively more pronounced in herbaceous plants (You et al., 2018). Testing the N/P threshold hypothesis, we found that N and P deposition promoted the growth of woody plants under N and P limitation, respectively ( Supplementary Table S3 ), suggesting that land management decisions to control the N/P deposition ratio are critical for alleviating plant nutrient constraints and maintaining ecosystem productivity.

We found that soil pH correlated negatively with NPP at a global scale but varied considerably under nutrient restriction conditions ( Supplementary Table S3 ). This negative correlation between soil pH and NPP can be attributed to soil acidification, reducing species diversity and abundance but increasing N availability for plants globally (Gilliam, 2006; Du et al., 2014; Midolo et al., 2019). In general, soil acidification is regulated by different acid buffer compounds, including carbonate at pH >7, soil exchangeable base cations (EBCs) at 4.5< pH <7, and aluminum compounds at pH <4.5 (Bowman et al., 2008; Lieb et al., 2011). In this study, global soil pH values ranged from 5.72 to 6.92, indicating that EBCs are the main acid buffer compounds in the soil, and no apparent aluminum toxicity was observed. Therefore, a decline in soil pH and a rise in available soil N content promote plant growth. PFT-based variations may result from differences at finer spatial scales. For example, tropical forests generally exhibit the highest NPP but lower soil pH (acidic soil) than recorded for other forest types because litter in tropical forests is not converted into soil organic matter, with its decomposition completed by fungi and other benthic organisms. Long-term N deposition tends to reduce cation saturation in tropical forest soils, reducing their acid-neutralizing capacity, and does not cause aluminum toxicity (Lu et al., 2014). Our results indicate that soil acidification promotes tree growth globally and that it may be relatively more appropriate to incorporate actual microenvironments (i.e., N- or P-limited) at the regional scale.

Temperature limitations (low temperatures) associated with increases in elevation can directly inhibit plant photosynthesis (Li et al., 2008) and meristem activity (Streit et al., 2013) or indirectly impact plant growth through soil nutrient availability (e.g., soil mineralization correlates positively with soil temperature) (Carbutt et al., 2013). Distinguishing the direct and indirect effects of elevation on plant growth remains challenging. A few studies have shown that an increase in elevation results in higher foliar N concentrations (Peng et al., 2012) but lower P concentrations (Gerdol et al., 2017), which is consistent with the results reported in this study ( Supplementary Figure S8 ). These results support the biogeochemical hypothesis, which states that low temperatures at high elevations inhibit nutrient extraction in roots and reduce N availability by reducing organic decomposition and mineralization (Güsewell, 2004; Reich and Oleksyn, 2004). In this study, univariate LMEs showed that NPP exhibited negative responses with increasing elevation ( Supplementary Table S1 ); however, these relationships disappeared after considering additional predictors in multivariate LMEs, likely because when studies are conducted at relatively large scales, variations in the distribution patterns of water and heat (e.g., latitude patterns) may be more important than elevation.

In addition, PFTs may be another key factor in determining the effects of elevation on plant nutrient contents (Zhao et al., 2018; Kulkarni et al., 2023). For example, Korner et al. (1986) reported that leaf N concentrations in herbaceous plants, but not in evergreen woody plants, increase with increasing altitudinal elevation in New Zealand. In contrast, Soethe et al. (2008) reported that the N, P, K, and S contents in shrubs and herbs decrease significantly with increasing altitudinal elevation. These diverse results have been attributed to differences in the adaptation strategies of different PFTs in response to elevational changes because of differences in community clusters and altitudinal environments between studies (Munson et al., 2022; Zhang et al., 2022; Jaroszynska et al., 2023). However, the results obtained in our study indicate that leaf N and P contents in different PFTs respond weakly to changes in elevation ( Supplementary Figures S4–S7 ). An integrated study of other indirect effects of elevation on the growth of PFTs may provide additional observational perspectives not readily obtained in species-specific studies.