Soil stoichiometric characteristics of protected cultivation systems in Northwest China

Abstract

Soil ecological stoichiometry is essential for understanding nutrient cycling and ecosystem functioning in agricultural systems. However, how planting duration and crop type affect soil stoichiometric characteristics in protected cultivation systems across different agroecological zones remains unclear. Therefore, we selected 189 protected cultivation systems (PCS) and 130 adjacent open-field systems (OFS) in southern Shaanxi (SS), central Shaanxi (CS), and northern Shaanxi (NS) of Northwest China. The topsoil (0–20 cm) samples were collected in common cultivation durations (short-term cultivation, ST; medium-term cultivation, MT; long-term cultivation, LT) and crop types (leafy vegetables, LV; solanaceous vegetables, SV; fresh fruits, FF) plots to measure soil organic carbon (SOC), total nitrogen (TN), total phosphorus (TP), total potassium (TK), and other soil properties. The average ratios of C: N, C: P, and N: P for OFS were 14.37, 11.86, and 0.93, which were significantly different from 7.71 (C: N), 33.35 (C: P), and 4.38 (N: P) for PCS (P < 0.01). The key soil stoichiometric ratios (C: P, N: P, C: K, N: K) in both PCS and OFS were strongly modulated by agroecological zone, following a gradient of SS > CS > NS, while the soil C: N ratio in PCS ranged from 2.69 to 16.17. Cultivation duration significantly affected soil stoichiometric ratios of PCS, and its effects varied among agroecological zones. In NS, the C: K, N: P, N: K and P: K ratios of PCS under LT were 1.80, 1.56, 1.98 and 2.02 times those under ST, respectively. Crop type alone had limited effects except for N: K, but its interactions with agroecological zone and cultivation duration significantly affected C: N, C: K, N: K, and P: K. Partial least square structural equation modeling (PLS-SEM) indicated that soil stoichiometric ratios were influenced by agroecological zone, cultivation duration, and crop type through both direct pathways and indirect pathways mediated by soil properties. Furthermore, soil properties exerted stronger effects on soil stoichiometric ratios than agroecological zone or management factors. Therefore, this context dependence complicates attempts to generalize stoichiometric relationships but also highlights potential management strategies (e.g., K fertilization) for regulating soil nutrient balances across agroecosystems.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-43026-5.

Introduction

Soil stoichiometry, reflecting the balance of key elements such as carbon (C), nitrogen (N), phosphorus (P), and potassium (K), plays a fundamental role in regulating nutrient cycling, microbial activity, and overall soil health in agricultural ecosystems¹. The ratios of these elements influence multiple ecological processes, including organic matter decomposition, nutrient use efficiency, and the structure and function of soil microbial communities². Agronomic practices (e.g., cultivation duration, fertilization management) can significantly modify soil nutrient stoichiometry by altering biogeochemical cycles through changes in nutrient inputs, outputs, and transformation rates^3–5. For example, excessive application of organic amendments and inorganic fertilizers may cause nutrient imbalances and shift stoichiometric ratios, potentially compromising soil fertility and ecosystem functioning⁶. Furthermore, different crop types have different nutrient uptake efficiencies^5,7, which affect soil nutrient accumulation. Such alterations can also influence greenhouse gas emissions, nutrient leaching, and the long-term sustainability of agricultural production⁸. Therefore, understanding how different agronomic practices influence soil stoichiometry is critical for sustaining crop productivity and mitigating environmental risks.

Considerable research has focused on stoichiometric changes in open-field and natural ecosystems^9–13. Sustainable agricultural systems are generally characterized by soil stoichiometric ratios within the following ranges: C: N between 5:1 and 10:1¹⁴, C: P between 100:1 and 200:1¹⁵, N: P between 5:1 and 7:1², N: K between 1:1 and 2:1^16,17, and P: K between 0.75:1 and 1:1¹⁸. Deviations from these ranges may indicate microbial and faunal nutrient immobilization or mineralization processes responding to stoichiometric imbalances³. These stoichiometric imbalances arise not only from agronomic management, such as fertilization practices¹⁹, but are also modulated by regional climatic factors, soil properties, and hydrological conditions²⁰. However, intensified production systems such as protected cultivation systems (PCS) have received comparatively less attention.

PCS (including greenhouses and high tunnels) have expanded rapidly worldwide because they can enhance crop yield and extend growing seasons by manipulating microclimatic conditions^21,22. These systems typically rely on high levels of irrigation and fertilizers to maintain productivity, differentiating them fundamentally from rain-fed open-field agriculture systems (OFS). The combination of continuous cropping, high nutrient application rates, and a semi-closed environment markedly alters nutrient cycling, potentially creating distinctive stoichiometric patterns not commonly observed in conventional systems. Long-term intensive fertilization—often exceeding crop nutrient demand—can lead to nutrient imbalances, salinization, and soil degradation²³. In addition, the unique micro-environments in PCS, characterized by elevated temperatures and humidity, may accelerate nutrient transformations and further disturb elemental balances²⁴.

Within PCS, both crop type and planting duration may exert strong effects on soil stoichiometric balances. Crop species differ in nutrient uptake patterns and root exudation profiles, thereby differentially influencing the mineralization and immobilization of C, N, P, and K^25,26. For example, leafy vegetables with high nitrogen demand may accelerate N depletion²⁷, while fruit vegetables with longer growth cycles might promote potassium accumulation and alter K-related ratios²⁸. Planting duration, which often reflects cumulative management intensity, can either amplify or mitigate these effects over time²⁹. Short-term systems may show rapid nutrient buildup and elevated ratios, while long-term cultivation often leads to nutrient saturation, element-specific depletion, or accelerates nutrient turnover, further shifting stoichiometric equilibria and threatening the sustainability of PCS³⁰. Previous studies have highlighted the influence of cropping systems and regional conditions on soil nutrient availability in open-field and natural ecosystems^9,31, yet the interaction between region and cultivation system remains less well documented. Despite the agronomic and ecological importance of maintaining nutrient balance, the existing studies on soil stoichiometry in PCS focused on a single region or single factor, and lacked systematic analysis of the interactive effects of agroecological zone, cultivation duration and crop type across climate gradients.

Shaanxi Province provides an ideal setting for investigating these interactions due to its pronounced gradients in climate, soil types, and farming practices from south to north. Furthermore, Shaanxi Province is one of the most important provinces for PCS, which has been expanding rapidly at an average annual rate of 10,666.7 hectares. The cultivated area and comprehensive output value ranked first in Northwest China, contributing substantially to farmers’ income and regional food security. Currently, we examined soil stoichiometric characteristics (C: N, C: P, N: P, C: K, N: K, P: K) in Shaanxi. We hypothesized that planting durations and crop types profoundly impacted the soil stoichiometry across different agroecological zones (southern Shaanxi, SS; central Shaanxi, CS; northern Shaanxi, NS). The specific objectives were to analyze: (1) How do soil stoichiometric characteristics vary among protected cultivation systems across different agroecological zones in Shaanxi Province? (2) How do planting duration and crop type individually and interactively affect soil stoichiometric ratios in protected cultivation systems? and (3) To what extent are the effects of agroecological zone, planting duration, and crop type on soil stoichiometry mediated by fundamental soil properties. The findings provide insights into how agronomic practices and environmental contexts jointly shape soil nutrient stoichiometry and support the design of targeted fertilization strategies to mitigate nutrient limitations and enhance ecosystem stability in intensive farming systems.

Results

Variation of soil stoichiometric ratios across the agroecological zone

The average C: N, C: P, N: P, C: K, N: K, and P: K ratios for PCS were 7.71, 33.35, 4.38, 4.51, 0.61, and 0.18, respectively, whereas those for OFS were 14.37, 11.86, 0.93, 6.18, 0.46, and 0.58, respectively (Table S2). Overall, soil C: N, C: K and P: K in PCS were generally lower than those in the OFS, while soil C: P, N: P, and N: K were consistently higher in PCS than in OFS, especially in SS (P < 0.01) (Fig. 1). In particular, the soil C: N ratio in PCS was significantly higher in NS than in SS, whereas the corresponding value in OFS was significantly higher in SS than in NS (P < 0.01) (Fig. 1a). For both PCS and OFS, soil C: P, N: P, C: K, and N: K ratios showed a consistent spatial pattern across agroecological zones: SS > CS > NS (Fig. 1b-e). However, agroecological zone exerted significant effects on all four ratios in PCS (P < 0.01), but had no significant effects on N: P and N: K ratios in OFS. Furthermore, the effects of agroecological zone on soil P: K ratio also differed between OFS and PCS (Fig. 1f).

Fig. 1.

Soil stoichiometric ratios of two cultivation systems in three agroecological zones. PCS, OFS, SS, CS, and NS indicated protected cultivation systems, open-field agriculture systems, southern Shaanxi, central Shaanxi, and northern Shaanxi, respectively. CS and AZ indicated cultivation system and agroecological zone, respectively. ** denotes significant difference at the level of P < 0.01 among agroecological zones in PCS and OFS, while ns denotes no significant difference (P > 0.05 ) among agroecological zones in PCS and OFS. The same was shown as follows.

Variation of soil stoichiometric ratios along the cultivation duration

Cultivation duration exerted significant effects on soil stoichiometric ratios (P < 0.05), except for C: N and C: P (Table 1). Additionally, different soil stoichiometric ratios differed with the increasing of cultivation duration. For instance, the C: N ratio decreased with increasing cultivation duration (ST: 7.87, MT: 7.62, LT: 7.44), whereas the N: P ratio showed an opposite, increasing trend (ST: 3.95, MT: 4.58, LT: 4.99) (Table S2). The interaction between cultivation duration and agroecological zone exerted no significant effect on soil stoichiometric ratios. However, distinct regional trends in soil stoichiometric characteristics emerged with increasing cultivation duration in PCS (Fig. 2, Table S2). Soil C: N declined with increasing cultivation duration in SS, and the ratios in CS and NS were significantly lower in long-term cultivation plots compared to short-term ones (P < 0.05) (Fig. 2a). Soil C: P and N: P ratios increased with cultivation in SS and NS, whereas in CS they were lowest in the medium-term cultivation plots (Fig. 2b and d). The C: K, and N: K ratios increased with cultivation duration, especially with significant differences between LT and ST in NS (P < 0.05). The change of soil P: K along cultivation duration varied among agroecological zones. The P: K ratio in CS under MT was significantly higher than that under ST, while soil P: K in NS increased consistently with cultivation duration. Overall, in NS, the C: K, N: P, N: K and P: K ratios of PCS under LT were 1.80, 1.56, 1.98 and 2.02 times those under ST, respectively.

Table 1.

Effects of agroecological zones, cultivation duration, and crop types on soil stoichiometric ratios in PCS.

	Factor	F-value	P-value	ηp²		Factor	F-value	P-value	ηp²
C: N	AZ	4.939	0.008	0.057	N: P	AZ	15.543	< 0.001	0.164
	CD	1.292	0.277	0.016		CD	4.23	0.016	0.015
	CT	0.791	0.455	0.01		CT	0.669	0.514	0.007
	AZ*CD	0.605	0.66	0.015		AZ*CD	0.684	0.604	0.017
	CD*CT	2.34	0.057	0.055		CD*CT	0.867	0.485	0.021
	AZ*CT	0.499	0.737	0.012		AZ*CT	0.533	0.712	0.014
	AZ*CT*CD	2.703	0.008	0.118		AZ*CT*CD	1.701	0.102	0.078
C: P	AZ	8.382	0	0.094	N: K	AZ	15.832	< 0.001	0.166
	CD	0.266	0.767	0.003		CD	9.797	< 0.001	0.112
	CT	0.067	0.935	0.001		CT	5.612	0.004	0.066
	AZ*CD	0.97	0.426	0.023		AZ*CD	1.905	0.112	0.048
	CD*CT	1.83	0.126	0.043		CD*CT	4.486	0.002	0.099
	AZ*CT	0.372	0.828	0.009		AZ*CT	4.108	0.003	0.094
	AZ*CT*CD	1.776	0.085	0.081		AZ*CT*CD	3.437	0.001	0.15
C: K	AZ	10.945	< 0.001	0.12	P: K	AZ	6.525	0.002	0.075
	CD	4.696	0.01	0.056		CD	2.381	0.096	0.029
	CT	2.19	0.115	0.027		CT	0.78	0.46	0.01
	AZ*CD	0.567	0.687	0.015		AZ*CD	2.198	0.072	0.056
	CD*CT	1.63	0.169	0.038		CD*CT	1.778	0.136	0.042
	AZ*CT	2.542	0.042	0.059		AZ*CT	0.353	0.842	0.01
	AZ*CT*CD	1.535	0.149	0.075		AZ*CT*CD	2.102	0.038	0.094

Note: AZ, CD, and CT indicated agroecological zone, cultivation duration, and crop type, respectively.

Fig. 2.

Soil stoichiometric ratios of PCS along cultivation duration in three agroecological zones. ST, MT, LT indicated short-term, medium-term, and long-term, respectively. Error bars represent the standard error. Only significant differences were labeled in the figures. The same letter denotes non-significant difference among cultivation durations in the agroecological zone (P > 0.05), while the different letters denote significant difference (P < 0.05).

Variation of soil stoichiometric ratios among the crop types

Overall, crop types alone had no significant effects on soil stoichiometric ratios except that the average value of N: K for SV (0.70) was significantly higher than for FF (0.53) (Table 1, Table S2, Fig. 3). However, crop types were involved in several significant interaction effects. In particular, the interaction between crop type and agroecological zone had significant effects on soil C: K (P < 0.05, ηp² = 0.059) and N: K (P < 0.05, ηp² = 0.094) ratios, while the three-way interaction among crop type, agroecological zone and cultivation duration significantly affected soil C: N ( P < 0.01, ηp² = 0.118) and P: K (P < 0.05, ηp² = 0.094) ratios. The results indicated that the influence of crop type on these stoichiometric ratios depended on the agroecological context. For example, Soil C: P and N: P ratios in SS exhibited significantly higher values than those in CS and NS. Specially in SS, soil C: N ratio in plots with FF (8.05) was significantly higher than that in plots with SV (7.14), while no significant differences among crop types were observed in NS (Fig. 3a). In CS, soil N: P and N: K ratios in plots with SV were significantly higher than those in plots with FF, increasing by 33.48% and 32.99%, respectively (P < 0.05).

Fig. 3.

Soil stoichiometric ratios of PCS among different crop types in three agroecological zones. SV, LV, and FF indicated solanaceous vegetables, leafy vegetables, and fresh fruits, respectively. Error bars represent the standard error. Only significant differences were labeled in the figures. The same letter denotes non-significant difference (P > 0.05).among crop types in the agroecological zone, while the different letters denote significant difference (P < 0.05).

Effects of agroecological zone, cultivation duration, crop type on stoichiometric ratios

Different agroecological zones, cultivation durations, and crop types not only affected directly stoichiometric ratios, but also indirectly affected stoichiometric ratios through influencing soil properties (Fig. 4, Table S3). Specially, soil properties exerted a dominant and significant influence on soil stoichiometric ratios, with their role as mediating variables providing the strongest explanatory power. The total effects of agroecological zones on soil stoichiometric ratios were stronger than cultivation durations, and crop types. We further analyzed the relationships between stoichiometric ratios and soil properties (Fig. 5). Soil C: N was positively correlated with pH, but negatively correlated with CEC. Conversely, soil C: K and N: K ratios were positively correlated with CEC, but negatively correlated with pH. Soil P: K exhibited strong positive correlations with AK, while C: P and N: P showed negative correlations with AK (Fig. S2). Furthermore, agroecological zones, cultivation duration, and crop type modified the correlations between soil stoichiometric ratios and soil properties (Fig. S3–S5).

Fig. 4.

Directed graph of the partial least squares path model (PLS-PM) showing the effects of agroecological zone, cultivation duration and crop type on the stoichiometric ratios. Indicated values are the path coefficients. Full lines indicate a positive effect, whereas dotted lines indicate a negative effect. The grey lines indicate no significant effect. The original PLS-PM included all indicators involved in soil properties indicators and stoichiometric ratios in this study. In the fitting processes, the indicators with low outer loading (< 0.7) and high collinearity (Variance inflation factor, VIF > 10) were eliminated. The final result only showed the indicators with significant effects (P < 0.05).

Fig. 5.

Redundancy Analysis (RDA) between soil properties and stoichiometric ratios.

Discussion

Effects of cultivation system on soil stoichiometric ratios

Our findings established the agroecological zone as the primary determinant of regional-scale soil stoichiometric patterns. The significant differentiation in most ratios (C: N, C: P, N: P, C: K, N: K, P: K) and the consistent gradients of (SS > CS> NS) across both PCS and OFS pointed to a fundamental control exerted by broad-scale environmental filters. Warmer, more humid, and potentially productive conditions (such as SS) appeared to accumulate relatively more N and C in relation to P and K. This may be linked to zone-specific differences in climate, soil parent material, vegetation inputs and management practices, which together regulate the rates of organic matter decomposition and nutrient mineralization, and leaching losses^32–34. Notably, the absence of significant differences in P: K ratio across agroecological regions suggests that the dynamics of total phosphorus and potassium were regionally synchronized. Effects of regional gradients and management practices appeared to offset in a way that maintained a relatively stable P: K balance.

Soil stoichiometric ratios differed substantially between OFS and PCS across agroecological zones, indicating that soil nutrient cycling was strongly governed by both agronomic practices and environmental conditions. Overall, the C: P and N: P ratios in both OFS and PCS were lower than the national (C: P = 136, N: P = 9.3)³⁵ and global² (C: P = 186, N: P = 13.1) averages. The observed pattern may be attributed to the following reasons. Firstly, microorganisms in Shaanxi Province may preferentially mineralize organic matter to release more phosphorus, thereby increasing soil P content. Secondly, the disproportionate loss of C and P elements-driven by both differential plant uptake and soil erosion-could directly reduce the soil C: P ratio. Regarding C: N, however, only the PCS with rapid decomposition of organic matter fell below the reference values (China: 11.9; Global: 14.3)^2,35 whereas OFS showed a similar or higher ratio.

The significantly higher soil C: N ratio in OFS compared to PCS can be attributed to differences in organic matter input and decomposition rates. OFS typically received greater input of crop residues and experienced lower decomposition rates due to more variable microclimatic conditions (e.g., rainfall, temperature), leading to the accumulation of organic carbon³⁶. In contrast, intensive management in PCS, including higher fertilizer input and frequent irrigation, promotes microbial activity and accelerates the mineralization of organic matter, resulting in a lower C: N ratio^23,37. The consistently higher C: P, N: P, and N: K ratios in PCS than OFS, particularly in SS, pointed to a fundamental decoupling in the accumulation of soil elements. PCS management appeared to favor a faster net accumulation of soil organic carbon and nitrogen relative to phosphorus and potassium. This decoupling effect was more pronounced in SS, suggesting that a warm, humid climate amplified the underlying mechanism. The observed lower P: K ratio in PCS than in OFS might reflect different nutrient management strategies, such as relatively higher potassium inputs in PCS. Our findings confirmed that climate, vegetation, and soil jointly impacted soil stoichiometry³⁸.

Effects of cultivation duration on soil stoichiometric ratios in PCS

The distinct regional trends of soil stoichiometric ratios with increasing cultivation duration indicated that soil nutrient dynamics were not linear and jointly affected by management intensity and agroecological factors. The decline in soil C: N ratio with increasing cultivation duration, particularly in SS, together with lower C: N ratios in LT than in ST plots in CS and NS, implied a relative enrichment of N compared with C under long‑term PCS. This pattern was consistent with sustained N fertilization, enhanced mineralization of soil organic matter and improved N retention under intensive, enclosed production³⁹. This pattern might indicate an emerging risk of carbon imbalance under prolonged cultivation. The consistent increase of C: P and N: P over time in SS and NS showed the accumulation of soil organic matter (increased C and N) outpaced phosphorus, pointing to a growing phosphorus limitation relative to carbon and nitrogen⁶.

Unlike most previous studies that focused on the effects of cultivation duration on C: N:P stoichiometry^19,40, our study further found that K-related ratios (C: K, N: K, P: K) are more sensitive to cultivation duration, which highlights the risk of K limitation in long-term PCS. The increases in C: K and N: K with cultivation duration, particularly the pronounced difference between LT and ST in NS, indicated a progressive depletion or insufficient replenishment of K relative to C and N. This is consistent with the high K demand of many intensively managed crops and the tendency for K to be removed in harvested biomass without proportional K inputs²⁶. In soils or regions with inherently lower K reserves, such as NS, this imbalance can emerge more rapidly, causing a marked rise in C: K and N: K and signaling an emerging K limitation. The relative balance between P and K was also strongly shaped by regional soil properties, fertilization strategies and historical management⁴¹. Taken together, these results indicated that cultivation duration under PCS systematically altered soil C-N-P-K relationships and differed among agroecological zones. Long-term cultivation tends to decrease C: N while increasing C: P, N: P, C: K and N: K, pointing to a gradual shift from N limitation towards P and K limitation in many situations. Therefore, fertilizer regimes and soil amendment strategies should be tailored to local agroecological conditions and cultivation histories to prevent the development of severe P and K imbalances and to sustain soil fertility under plastic-covered systems.

Effects of crop type on soil stoichiometric ratios in PCS

Unlike agroecological zone and cultivation duration, crop type alone had only limited direct effects on soil stoichiometric ratios, with a significant main effect observed only for N: K, which did not support our hypothesis and other studies^5,42,43. This discrepancy may be attributed to the strong buffering effect of intensive management (e.g., uniform fertilization and irrigation) in our PCS, which offsets crop-specific differences in nutrient uptake and residue quality. However, the significant two-way (crop type and agroecological zone, crop type and cultivation duration) and three-way (crop type, agroecological zone, and cultivation duration) interactions indicated that the influence of crop type was highly context-dependent. For instance, soil C: N was significantly higher in FF plots than in SV plots in SS, but no differences among crop types were observed in NS. Likewise, in CS, N:K and P: K were higher in SV than in FF plots.

In contrast, the absence of significant differences among crop types in NS suggests that strong environmental constraints or relatively uniform management may override crop‑specific effects in this zone. Together, these patterns implied that crop-specific traits (e.g., rooting depth, nutrient uptake efficiency, residue composition) could substantially modulate soil nutrient balances^44,45, but this modulation cannot be generalized across regions or cultivation stages; instead, it depends on the interaction between crop traits, agroecological conditions and cultivation duration. Therefore, recommendations on crop rotations or crop substitution for maintaining balanced C-N-P-K stoichiometry in the PCS should consider not only species traits but also the agroecological setting and cultivation history.

Mediating role of soil properties in regulating stoichiometric ratios in PCS

The PLS-SEM results further emphasized that soil properties were central mediators linking agroecological zone, cultivation system, crop type, and cultivation duration to soil stoichiometry. This implies that inherent soil conditions form the fundamental basis for the effectiveness of agronomic practices. The strong pathway from agroecological zone to soil properties underscored the need for region-specific soil management strategies. Although the direct effects of cultivation duration and crop type were weaker, their long-term cumulative impacts should not be overlooked (Table S3). Persistent imbalances between nutrient inputs and outputs will eventually lead to significant shifts in stoichiometry, as evidenced by the trends in C: K and N: K ratios with planting duration observed in this study. Compared with other indicators, N:P and C: P showed weaker or more complex relationships with the measured soil properties. In contrast, several stoichiometric ratios were tightly associated with key soil attributes⁴⁶.

As soil C-N-P-K stoichiometric ratios are derived from SOC, TN, TP, and TK, the consistent positive correlations of C: K, N: K, and P: K with SOC, TN, and TP were obviously expressed (Fig.S2). These ratios are driven more by the accumulation of organic matter and phosphorus than by changes in the potassium pool. Results may show that soil organic matter is essential to maintaining a balanced nutrient stoichiometry^47,48. In particular, soil pH was the primary factor driving the distribution and function of microorganisms to SOC accumulation^49,50. Accordingly, soil pH influenced stoichiometric ratios, through its fundamental control on nutrient availability⁵¹. Increasing pH promoted the accumulation of soil organic carbon, leading to a positive relationship between C: N ratio and pH. By contrast, higher pH enhanced soil potassium availability and retention, thereby reducing C: K and N: K ratios. Moreover, agroecological zone, cultivation duration and crop type modified the correlations between stoichiometric ratios and soil properties, underscoring that the same soil property could exert different effects on stoichiometry under different environmental and management conditions⁵². Therefore, this context dependence makes it difficult to generalize stoichiometric relationships, but it also highlights potential strategies for regulating soil nutrient balances.

Implications for nutrient management and sustainable cultivation

Overall, our findings indicated that soil stoichiometric ratios were shaped by the combined effects of agroecological zone, cultivation system, cultivation duration, and crop type, with soil properties playing a dominant mediating role. For example, Soil CEC was positively correlated with C: K and N: K ratios, mainly because higher CEC was closely associated with greater soil organic matter accumulation⁵³, which substantially increased soil C and N concentrations but had a weaker effect on K content. The observed increases in C: P, N: P, C: K and N: K with cultivation duration, together with the strong link between stoichiometry and pH, CEC and AK, suggested that long-term intensive cultivation might gradually shift nutrient limitation from N to P and K and altered the balance between C storage and nutrient supply.

Compared with previous regional studies on soil stoichiometry^54,55, we systematically compared PCS and OFS across a complete agroecological gradient, and we clarified the context-dependent effects of crop type and cultivation duration mediated by soil properties. From a management perspective, these results highlight the need to adopt region-specific and system-specific nutrient management strategies^47,56,57. In zones and systems where C: P and N: P continue to rise, enhancing P inputs may be critical to avoid P limitation. Where C: K and N: K are consistently high, additional K fertilization or strategies to increase K retention may be required. Furthermore, optimizing crop combinations and rotations in line with local agroecological conditions could help maintain more balanced stoichiometric ratios, thereby supporting both soil fertility and long-term productivity.

Materials and methods

Study area and investigation information

The study was in Shaanxi Province, Northwest China (from 31°42′ N to 39°35′ N; longitude from 105°29′ E to 111°15′ E). Shaanxi Province comprises three agroecological zones: (1) southern Shaanxi (SS), characterized by a subtropical continental humid monsoon climate; (2) central Shaanxi (CS), with semi-humid warm continental monsoon climate; (3) northern Shaanxi (NS), with temperate continental semiarid climate. SS receives approximately 1,000 mm of rainfall annually, with mean monthly temperatures ranging from 3℃ to 25℃. Soils are generally acidic soils (pH < 7) and contain on average 15.2 g organic matter kg^− 1. The primary soil types are Hapli-Udic Argosols, Ferri-Udic Argosols, and Hapli-Stagnic Anthrosols. CS receives about 740 mm of annual rainfall, with mean monthly temperatures from − 2℃ to 22℃. Soils are alkaline (pH 7.96–8.75) and contain 3.0–27.1 g organic matter kg^− 1. NS has an average annual rainfall of 548 mm and mean monthly temperatures from − 7℃ to 18℃. Dominant soil types are Earth-cumuli-Orthic Anthrosols in the semi-humid zone and Loessi-Orthic and Ust-sandic entisols in the temperate zone⁷.

We selected PCS with different cultivation duration and crop types across three agroecological zones. To study the effect of cultivation duration on soil stoichiometry in the PCS, we designated three groups following: a short-term group (ST, 1–4 years of cultivation), a medium-term group (MT, 5–9 years) and a long-term group (LT, more than 10 years)⁷. PCS were multi-cropping systems with crops such as kidney beans, chilies, cucumber, tomato, eggplant, and muskmelon. To study the impact of crop type on soil stoichiometry in PCS, we designated three groups: leafy vegetables (LV), solanaceous vegetables (SV), and fresh fruits (FF).

Furthermore, adjacent conventional open-field systems (OFS) were selected as controls to provide a baseline for soil nutrient status and stoichiometric characteristics. Although their crop types and cultivation durations did not exactly match those of the PCS, these OFS still represent local soil nutrient conditions and stoichiometric characteristics under conventional management within the same agroecological zones.

Soil sampling and analysis

Prior to soil sampling, permission for soil sampling at each site was obtained from the relevant landowners. The sampling complied with national regulations. At each location, five sub-samples from 0 to 20 cm depth were collected using the five-point diagonal sampling method, which is the standard soil sampling method for agricultural fields in China, and then composited into a 1 kg sample. A total of 189 soil samples were collected from protected cultivation systems with different cultivation durations and crop types across three agroecological zones, including 50 samples in SS, 89 in CS, and 50 in NS. Approximately 130 soil samples were collected from OFS across three agroecological zones as controls (Fig. S1). All soil samples were placed in polyethylene bags, covered with cotton bags, and then transported to the lab. Samples were air dried at room temperature. Plant roots, stones, and other debris were removed before analysis. The air-dried soil samples were prepared for different soil analysis.

Soil samples were analyzed for pH, electrical conductivity (EC), cation exchange capacity (CEC), soil organic carbon (SOC), total nitrogen (TN), total phosphorus (TP), total potassium (TK), available phosphorus (AP), and available potassium (AK). Soil pH was measured in a 1:2.5 soil: water suspension. EC was determined in a saturated soil paste. CEC was measured using ethylenediaminetetraacetic acid (EDTA) at 1 mol L^− 1 and 0.005 mol L^− 1. SOC was calculated from organic matter by dividing by a conversion factor of 1.724. Soil organic matter was determined by a wet oxidation method using a mixture of K₂Cr₂O₇ and H₂SO₄. TN and TP were measured by the micro-Kjeldahl method and ascorbic acid-molybdenum blue method, respectively. Soil AP was analyzed by the sodium bicarbonate method and determined by the molybdenum blue method at 700 nm. Soil TK was determined using a flame photometer after a complete digestion with hydrofluoric acid-perchlorate. AK was extracted with NH₄OAc and determined using a flame photometer. All the methods were detailed in the analytical methods of soil and agricultural chemistry⁵⁸. All soil property analyses were performed with three technical replicates for each sample. The coefficient of variation of parallel determinations was required to be less than 10%, and abnormal values were re-measured and verified. Blank control experiments were set up for all chemical analysis processes to eliminate reagent and instrument interference. Summary statistics for soil properties are presented in Table S1.

Statistical analysis

The sampling map was prepared using ArcGIS 10.7 (ESRI Inc., Redlands, CA, USA; URL: https://www.esri.com). Prior to analysis, the assumptions of normality (using the Shapiro-Wilk test) and homogeneity of variances (using Levene’s test) by the “car” package were checked for stoichiometric ratios. Soil stoichiometric ratio data passed the Shapiro-Wilk normality test (P > 0.05) and Levene’s variance homogeneity test (P > 0.05), which met the basic assumptions of analysis of variance (ANOVA). Two-way ANOVA was used to analyze the effects of cultivation system (PCS vs. OFS) and agroecological zone (SS, CS, NS) on soil stoichiometric ratios using the “agricolae” package. The effects of agroecological zone, cultivation duration, and crop type on soil stoichiometric ratios in PCS were analyzed using multi-way ANOVA in R software. The Least Significant Difference (LSD) test at a 5% probability level (P < 0.05) was performed using the “agricolae” package in R to assess pairwise differences in soil stoichiometric ratios. The heatmaps were generated using the pheatmap package in R to visualize Spearman correlations between soil properties and soil stoichiometric characteristics across different agroecological zones, cultivation durations, and crop types. The redundancy analysis (RDA) using the “vegan” package further intuitively illustrated the relationship between soil properties and stoichiometric ratios. Partial least squares-structural equation modelling (PLS-SEM), implemented in R software using the plspm package, was used to assess the associations between nutrient ratios and soil properties, cultivation duration, crop type, and agroecological zones, and to identify the key drivers of stoichiometric characteristics. In this model, we included agroecological zones, cultivation duration, and crop type as the exogenous variables, whereas soil properties as the endogenous variables and soil stoichiometric ratios as the response factor. The statistical analyses and graphs were performed on R software (version 4.5.2) (http://www.r-project.org/). The significance level was set at P < 0.05. As soil C-N-P-K stoichiometric ratios are derived from SOC, TN, TP, and TK, we excluded these base nutrients to avoid redundancy and informational overlap when analyzing the relationships between soil properties and stoichiometry in RDA and PLS-SEM.

Conclusion

This study systematically analyzed the soil C-N-P-K stoichiometric characteristics, especially under the PCS system. Soil C: N, C: P, N: P, C: K and N: K ratios showed clear spatial patterns across agroecological zones and differed among cultivation systems. In both PCS and OFS, C:P, N:P, C:K and N: K consistently followed the order SS > CS > NS. Compared to OFS, PCS exhibited higher C: P, N: P, and N: K ratios, indicating that soils in PCS tended to accumulate more C and N rather than P and K. In PCS, cultivation duration significantly affected soil N: P, C: K and N: K ratios, and its effects were strongly context dependent. Long-term cultivation tends to decrease C: N while increasing C: P, N: P, C: K and N: K, pointing to a gradual shift from N limitation towards P and K limitation in many situations. Crop type alone had limited effects on soil stoichiometric ratios, except for N: K. However, crop type interacted significantly with agroecological zone and cultivation duration, indicating that its influence on stoichiometry depended on the agroecological context. Overall, agroecological zone, cultivation duration and crop type influenced soil stoichiometric ratios both directly and indirectly by altering soil properties. Soil properties exerted the dominant and mediating control on stoichiometry. Our findings highlight the fundamental role of soil properties in maintaining nutrient balance. So, differentiated fertilization strategies are recommended: appropriate N reduction combined with P and K supplementation in SS, balanced C-N-P-K management in CS, and priority P and K inputs to alleviate nutrient limitation in NS. However, the study is a cross-sectional investigation, long-term multi-site observations are still required to reveal the temporal evolution of soil stoichiometry under continuous cropping. Therefore, ongoing monitoring is required, and soil management strategies should be tailored to specific agroecological zones to optimize soil nutrient balances to sustain productivity and ecosystem stability.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Guanghua Jing: Conceptualization, Methodology, Formal analysis, Writing-Original Draft. Biao Huang: Resources, Investigation. Liyan He and Yingmei Li: Methodology, Formal analysis. Weixi Wang: Visualization, Investigation. Qiangqiang Lu and Zhao Zhang: Validation, Investigation. Zhikun Chen: Conceptualization, Methodology, Supervision, Writing-Review, Funding acquisition. All authors reviewed the manuscript.

Funding

This work was funded by the Shaanxi Provincial Department of Science and Technology (NO. 2021 PT-018 & 2023KJXX-098), the Basic Research Project of Shaanxi Academy of Sciences (NO. 2022 K-02 & 2024P-17 & 2025k-35).

Data availability

All data generated and/or analyzed during this study are included in this published article. The datasets are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.