This study was conducted at the Xiaowenshan Forest Farm (established in 1962) in the Guandishan Forest Region of Shanxi Province, northern China ( Figure 1 ). The total area covers approximately 1.99 × 10⁴ ha, with altitudes ranging from 1,460 to 1,610 m. This region is located in the inner continental mountain monsoon climate zone. The mean annual temperature is 4.2°C, the mean annual precipitation is 822.6 mm, and the relative humidity is 70.9%. The soil is a typical Alfisol, which is covered by a 3–7-cm humus layer. P. tabuliformis is a typical dominant tree species, with a few associated species such as Larix principis-rupprechtii, Picea wilsonii, Betula platyphylla, and Quercus mongolica. The understory species primarily comprise Spiraea salicifolia, Rosa xanthine, Corylus mandshurica, Rosa bella, and a few Lespedeza bicolor (Zhang et al., 2021).

In 2020, 42-year-old (42-year), 55-year-old (55-year), and 65-year-old (65-year) stands of P. tabuliformis forest were selected as experimental sites at the forest farm (111°24′E–112°37′E, 37°41′N–37°54′N) ( Figure 1 ). A randomized block design with three replicate blocks of five levels of N application at each successional age was established before the experiment: N₀ (0 kg N m⁻² year⁻¹, N₁ (5 kg N m⁻² year⁻¹), N₂ (10 kg N m⁻² year⁻¹), N₃ (15 kg N m⁻² year⁻¹), and N₄ (20 kg N m⁻² year⁻¹). Five 201-m² circular plots (radius = 8 m) with a buffer zone (Song, 2017; Pretzsch, 2022) were included in each block. A backpack sprayer was used to apply various concentrations of urea solutions (CH₄N₂O) to the corresponding N application plots (N₀, 0 g CH₄N₂O; N₁, 6,608 g CH₄N₂O; N₂, 8,913 g CH₄N₂O; N₃, 11,217 g CH₄N₂O; and N₄, 15,826 g CH₄N₂O; each added to 20 L of water) continuously before rain in mid-May, July, and September of each year (Geng and Jin, 2022). The details of the experimental sites are listed in Table 1 .

In June 2022, after the continuous application of N for 1.5 years, three standard trees were randomly selected to collect the paired non-rhizospheric and rhizospheric soils, as well as the living fine roots (Ø ≤ 2 mm) in each plot of each stand age. Three 8 cm inside diameter soil cores (10 cm depth) were randomly extracted under the canopy area of each standard tree to ensure sufficient rhizospheric soil and living fine root samples following careful removal of understory plants and surface litter. The living roots of P. tabuliformis can be distinguished based on their features, including shape, color, taste, and elasticity. The fine roots were gently shaken to collect the adhered rhizospheric soil (Phillips et al., 2011). The fine roots were collected and placed in polyethylene bags. Finally, soil without any attachment to the roots was collected, which was regarded as non-rhizospheric soil. The living fine roots and paired non-rhizospheric and rhizospheric soils of three standard trees collected from each plot were pooled for a homogenized sample.

A cooler box was used to store the collected rhizospheric and non-rhizospheric soil samples, and it was ensured that their transport to the laboratory was complete within 24 h of analysis. Each soil sample was divided into two fractions. One fraction was stored at 4°C for <1 week and subsequently used for the analysis of soil enzyme activity. The other fraction was sieved using a 0.25-mm sieve after air-drying at room temperature (20°C–25°C) to analyze its chemical properties, including the concentration of SOC, total N (TN), total P (TP), NH₄ ⁺–N, NO₃ ⁻–N, and available P (AP). The root samples were first gently rinsed in distilled deionized water over a 0.5-mm wet sieve and then dried to a constant mass at 75°C for 48 h after heat-killing at 105°C for 15 min. Finally, the dried root samples were ground into fine powder to determine their chemical properties [TN, TP, and total C (TC)].

The SOC and TC of the roots were quantified using a TOC/TN analyzer (Multi N/C 3100; Analytik Jena, Jena, Germany). The TN and TP concentrations in the paired rhizospheric and non-rhizospheric soils were quantified using Kjeldahl digestion and molybdenum blue colorimetry, respectively. Root TN and TP contents were determined by initial digestion with H₂SO₄–H₂O₂ (Nelson and Sommers, 1973; Parkinson and Allen, 1975). The soil samples were extracted using a 2 M KCl and 0.5 M NaHCO₃ solution under shaking and then passed through filter paper to determine the available N and P (NH₄ ⁺–N, NO₃ ⁻–N, and AP) of paired rhizospheric and non-rhizospheric soils (Olsen et al., 1954; Bremner and Keeney, 1966). These parameters were determined using an automated discrete analyzer (SmartChem 450, AMS Alliance, Rome, Italy).

The activities of four soil extracellular enzymes, i.e., β-1,4-glucosidase (BG), β-1,4-N-acetylglucosaminidase (NAG), l-leucine aminopeptidase (LAP), and acid phosphatase (ACP), representing one C-acquiring, two N-acquiring, and one P-acquiring enzymes, respectively, were determined using standard fluorometric techniques and enzyme calibration (M Plex, Tecan, Männedorf, Switzerland) (Cui et al., 2019, 2021; Huang et al., 2022). The enzymatic activity was expressed as nmol g⁻¹ h⁻¹.

The REs of each soil nutrient variable were calculated as the percentage difference between paired rhizospheric and non-rhizospheric soil samples using the Equation 1 (Phillips and Fahey, 2006):

where C_R and C_N are the concentrations of the measured variables that include one soil C-related indicator (SOC), three soil N-related indicators (TN, NH₄ ⁺–N, and NO₃ ⁻–N), and two soil P-related indicators (TP and AP) in the paired rhizospheric and non-rhizospheric soil samples, respectively.

Extracellular enzyme N:P stoichiometry in the rhizospheric soil was calculated using the Equation 2:

where N:P_SEE represents the natural logarithm of the ratio of relative enzymatic activity, which indicates N acquisition and P acquisition in rhizospheric soil (Cui et al., 2018).

The vector analysis of soil extracellular enzymes was widely applied to determine the responses of soil microbial nutritional constraints of C, N, and P to internal plant succession and external soil environmental changes, which are always calculated with Equations 3 and 4 (Cui et al., 2021; Huang et al., 2022; Kang et al., 2022; Xu et al., 2022):

The vector length (dimensionless) quantified the relative microbial C limitation. The greater the vector length, the more severe the C limitation experienced by the microbes. The magnitude of the vector angle (°) determines the relative microbial N or P limitation. Vector angles (> 45° or < 45°) indicate that the microbes experienced relative P or N limitation, respectively. The greater the vector angle, the more P-limited the microbes and vice versa for N-limited microbes. Thus, we developed a new indicator to define microbial N limitation, as the formula, Equation 5:

where Degree represents the tangent angle of the line extending from the origin of the coordinates to the points with horizontal and vertical coordinates of Ln BG/Ln ACP and Ln BG/Ln (NAG + LAP), respectively.

The N:P stoichiometry of fine roots was calculated as described by Elser, with Equation 6 (Elser et al., 2010):

where N:P_CON represents the mass ratio (g N/g P) of TN: TP in the fine roots.

A new vector analysis was developed to determine the nutritional limitations of N and P in plants, similar to that in a previous study (Xu et al., 2022), and was calculated with Equation 7:

The magnitude of the vector angle (°) determines the relative N or P limitation of the fine roots. A vector angle (> 4.09° or < 3.58°) indicated that the roots were experiencing relative N or P limitation, respectively, and (3.58° < vector angle < 4.09°) indicated that the relative P or N was free from limitation in fine roots. The greater the vector angle, the more serious the N-limitation experience for the roots and vice versa for the P limitation. The degree represents the tangent angle of the line extending from the origin of the coordinates to points with the horizontal and vertical coordinates of the TN and TP contents, respectively.

Differences in the parameters representing the metabolic limitations of root and rhizospheric soil microbes and the REs between different N application levels and successional ages were tested using a two-way analysis of variance (ANOVA). Multiple comparisons among the different N application levels and successional ages were conducted using the least significant difference (LSD) test (p < 0.05). Relationships between paired nutrient REs were determined using Pearson’s correlation analysis. The relationships between root and rhizospheric soil microbial nutrient limitations and REs were identified using linear regression analysis.

The relative importance of TC content in fine roots, rhizospheric soil microbial nutrient limitation, and REs in the regulation of root nutrient limitation was demonstrated using random forest analysis and the random forest R package. The complicated directions of multiple variables and the direct and indirect effects of the predicted variables on the metabolic limitations of the roots were quantified by constructing structural equation models (SEMs) based on our fundamental knowledge in conjunction with the results of linear regression analysis and random forest modeling (Zhao et al., 2022). Moreover, we used five metrics containing low chi-square value (χ²), high p-values (> 0.05), high Jöreskog’s goodness fit index (GFI > 0.95), and comparative fit index (CFI > 0.95), as well as low root mean square error of approximation (RMSEA < 0.05) to verify an ideal model fit. Construction and examination of the SEM model were based on the maximum likelihood estimations and performed through the R “lavaan” package (Zhao et al., 2022). All analyses were performed using the R version 4.2.2 (R Core Team, 2022).

Pearson’s correlation analysis revealed that the root TC and TN contents were negatively correlated with the REs of SOC or REs of TN, respectively, but positively correlated with the REs of NO₃ ⁻–N. However, these relationships were only significant in the REs of TN with root TC and TN contents (p < 0.05) ( Figure 8 ). Root TN content also tended to decrease with the REs of NH₄ ⁺–N; however, these relationships were not significant (p > 0.05) ( Figure 8 ). In addition, the REs of NO₃ ⁻–N were negatively correlated with the REs of NH₄ ⁺–N (p < 0.05) ( Figure 8 ).

Linear regression analysis indicated that REs of SOC, TN, TP, NH₄ ⁺–N, and NO₃ ⁻–N, as well as rhizospheric soil microbial N limitation, were linearly correlated with root N limitation, and their correlation coefficients were 0.30, 0.34, −0.44, 0.35, −0.33, and 0.34, respectively; p < 0.05) ( Figures 9B, C ). Among the variables, successional age (20.5% IncMSE), REs of TP (4.9% IncMSE), and microbial N limitation in the rhizosphere (3.8% IncMSE) were the three strongest predictors of root N limitation ( Figure 10 ).

The results of the SEM model demonstrated that the explanatory factors cumulatively explained 64 variances in root N limitation ( Figure 11A ). Succession, which had the highest overall effect, was the most critical factor driving root N limitation, both directly and indirectly. The N limitation of microbes in the rhizosphere and REs of NH₄ ⁺–N availability, with relatively high total effects, ranked following succession and had direct positive effects on root N limitation ( Figures 11A, B ). Succession had an indirect effect on root N limitation by manipulating root TC content to regulate the REs of NO₃ ⁻–N availability and the N limitation of microbes ( Figure 11A ). In addition, we found that the REs of TN directly and positively regulated microbial N limitation in the rhizosphere ( Figures 9A , 11A ). However, the REs of TN and N levels had slight indirect effects on root N limitation ( Figures 11A, B ).

Initially, we found that the roots were limited by soil N, which is consistent with our previous study on branches and leaves (Yan et al., 2023). This indicates that both the aboveground growth and underground growth were consistent. In the present study, root N limitation was significantly alleviated when N addition rates reached 20 kg N m⁻² year⁻¹ in the 42-year and 55-year stands. However, N application increased root N limitation in the 65-year stands ( Figure 2I ). As a fast-growing tree species prior to maturity, the increase in exogenous N in P. tabuliformis directly promoted the accumulation of root TN in the 42-year and 55-year stands ( Figure 2C ), which is consistent with previous studies (Kou et al., 2018; Liu et al., 2021; Geng et al., 2023). However, because the increased growth rate of P. tabuliformis caused by improved N availability exceeds the nutrient uptake rates during the maturation period, dilution effects have been observed (Deng et al., 2017; Geng et al., 2023).

Notably, the microbial vector angles were (< 45°), indicating N restriction rather than P restriction for microorganisms in the rhizospheres of stands at the three successional ages ( Figures 3C, D ). Similarly, application of N generally alleviated microbial N limitations in the 42-year and 55-year stands but exacerbated them in the 65-year stand, although they were not significantly affected by the application of N due to their strong homeostasis (Figures 3C, E, G) (Zhang et al., 2019). This correlated with root N limitation and was consistent with the SEM results, which suggested a key role of microorganisms in plant nutrient limitation and competitive utilization of resources between roots and rhizospheric microbes ( Figure 11A ). Roots and microbes in the rhizosphere are always in progressive N limitation ( Figures 2I, J , 3C, D ), which is attributed to the formation of strong depletion zones around the roots by continuous N uptake by plants (Kuzyakov and Xu, 2013). Similar patterns were also observed for root and microbial N limitation during succession, which increased with succession, consistent with previous studies (Zhou et al., 2017; Zhang et al., 2022), indicating that succession would increase N limitation in the root–soil–microbe system in the rhizosphere under the same level of soil N in this forest ecosystem.

The REs of SOC and TN decreased following N application, although sensitivity to N varied in P. tabuliformis stands of different ages ( Figures 6A, C ). Soil, a strong source of plant N acquisition, provides N resources to meet its high demand and conserves N for low demand and vice versa (He et al., 2021). The application of N reduces the mineralization of soil N in the rhizosphere, which is aligned with C inputs from root exudates (Dijkstra et al., 2013; Cheng et al., 2014; He et al., 2021). Thus, the improvement in soil N availability resulting from exogenous N input (i.e., fertilization or atmospheric N deposition) would enable plants to obtain N from the soil immediately rather than investing more C in belowground tissues and root exudates to stimulate microorganisms to decompose SOM for N acquisition (Phillips et al., 2009). However, the magnitude of the REs of SOC and TN in the 65-year stand was the greatest ( Figures 6B, D , 12 ) in response to exacerbated root and microbial N limitations caused by lower root TC and TN contents in the roots of 65-year stands, in contrast to that in the 42-year and 55-year stands (Zhu et al., 2022) ( Figure 8 ). This was consistent with some previous studies that showed that a dilution effect related to greater nutrient accumulation and biomass production would lead to lower N and P concentrations in plants (Guiz et al., 2016; Lü et al., 2019; Zhang et al., 2022). In addition to root exudates or rhizodeposits, microbes obtain N by assimilating the intractable decomposition of SOM to meet their elevated N requirements under intense N-limited conditions (Dijkstra et al., 2013; Kirkby et al., 2014; He et al., 2021), which is consistent with the preferential substrate utilization hypothesis (Phillips and Fahey, 2008; Dan et al., 2023).

The decreased REs of NH₄ ⁺–N availability with the development of P. tabuliformis forests after the application of N was due to decreased root TN content ( Figure 8 ), which is similar to previous findings (Hogberg et al., 1998; Jiang et al., 2020). However, positive REs of the NH₄ ⁺–N availability in the 65-year stand and REs of the NO₃ ⁻–N availability of all stands ( Figures 6G–G , 7D, E ) positively direct root TC content to the REs of the NO₃ ⁻–N, positively correlate between the REs of the NO₃ ⁻–N and root TN content, and positively affect the indirect REs of the NO₃ ⁻–N to NH₄ ⁺–N ratio induced by grave root N limitation ( Figures 8 , 11A ) and demonstrated that differently aged natural secondary P. tabuliformis stands exhibited similar N preferences and dominantly relied on NO₃ ⁻–N for its inorganic N nutrition. This was due to the greater mobile properties of NO₃ ⁻–N than NH₄ ⁺–N in soil solution (Kuzyakov and Xu, 2013). Our results also suggested that P. tabuliformis forests in 65-year-old absorbed NH₄ ⁺–N from soils under severely N-limited conditions as a strategy to make full use of soil N sources, which enhances the primary productivity of forest ecosystems (Houle et al., 2014; Gao et al., 2020). Similarly, from this perspective, the REs of TN and NH₄ ⁺–N availability alleviated root N limitation.

The REs of TP increased linearly with the N level, leading to a generally insignificant response signal of the REs of AP, which was further free from P-limited roots and microbes in the rhizosphere ( Figures 6K , 7C ). Lower soil pH owing to the application of N promoted phosphatase activity, which is consistent with the findings of Phillips and Fahey (Phillips and Fahey, 2006).

As shown by root exudation and nutrient absorption, REs control N cycling in the soil. The REs of the soil N indicators were susceptible to soil N in P. tabuliformis stands at the three successional ages. Numerous studies have shown that applying N to soil has variable effects on each RE indicator of N indicators (Cai et al., 2023; Dan et al., 2023; Gao et al., 2023). Moreover, REs affect the nutrient supply to plants and are essential drivers for alleviating root N limitation (Sun et al., 2014; Han et al., 2023). Linear regression and RF analyses revealed the importance of the REs of soil N indicators and the N limitation of microbes in the rhizosphere for root N limitation ( Figures 9B, C , 10 ). SEMs revealed that the root TC content indirectly mitigated root N limitations by regulating the REs of NO₃ ⁻–N availability and N limitations of microbes in the rhizosphere ( Figure 11A ). Thus, we put forward a conceptual diagram to summarize the effects of succession on the N limitation of roots and rhizospheric soil microbes ( Figure 12 ). We determined the characteristics of stands of three successional ages by comparing their REs, microbial C and N limitations, and root N limitation among the five levels of N application using radar charts in P. tabuliformis forests after N input ( Figure 13 ).

Stand development enhanced the fine root biomass production, resulted in weaker fine root nutrient foraging ability, and caused lower soil N nutrient availability (Ryan et al., 2004; Lü et al., 2019; Zhu et al., 2022). These changes directly decreased root TC and TN concentrations, increased investment in belowground photosynthetic products such as root exudates, and increased microbial N limitation in the rhizosphere symbiosed with roots when plant growth was N-constrained ( Figures 2B, D , 3F, H ) (Guiz et al., 2016; Geng et al., 2021; He et al., 2021; Zhang et al., 2022; Geng et al., 2023). However, the increase in microbial N limitation with succession is insignificant owing to their faster turnover rates and strong homeostasis (Zhang et al., 2019, 2024). Root exudate inputs induced by succession promote the growth and activity of microorganisms and further stimulate stronger REs of SOC and TN and theoretically stronger REs of NH₄ ⁺–N and NO₃ ⁻–N availability to obtain more N resources ( Figures 6B, D ) (He et al., 2021; Zhu et al., 2022). In fact, continuous available N uptake by roots and microbes in the rhizosphere and the increased competitive utilization of resources between them led to strong depletion zones near the roots (Kuzyakov and Xu, 2013; Zhang et al., 2024), which ultimately decreased the REs of NH₄ ⁺–N and NO₃ ⁻–N availability with succession ( Figures 6H, J ). In addition, REs of NH₄ ⁺–N also directly and positively drove the root N limitation and directly controlled the transformation from REs of NO₃ ⁻–N ( Figure 11A ). This was because of the poor mobile properties in the soil solution of NH₄ ⁺–N compared with NO₃ ⁻–N and was not quickly absorbed by microbes in the rhizosphere but could be easily uptake by roots, making the roots outcompete in the uptake of N in the rhizosphere under N-limited condition (Kuzyakov and Xu, 2013; Huygens et al., 2016; Zhang et al., 2024).

The results of the radar charts showed that root N limitation of stands of the three ages and REs of TP in the 42-year stand generally increased with N levels, whereas root N limitation and REs of SOC greatly increased with succession ( Figure 13 ). This result indicates that the adverse effect of N application on root N limitation in the 42-year stand was the smallest among the three stands of different ages. Therefore, the afforestation activities of the natural secondary P. tabuliformis forest in this area should prioritize middle-aged forests (40–50 years old) to adapt to global N deposition or the distribution heterogeneity of bioavailable soil N.