For the experiment, soils with both P-surplus and P-deficient were selected. These soils were obtained from Danling County (30°04′ N, 103°53′ E), in Sichuan Province of China. The P-surplus soil was collected from a well-established citrus orchard, while the P-deficient soil was obtained from a recently established citrus orchard. Based on the classification by the United States Department of Agriculture Soil Taxonomy, both the experimental soils were classified as Alfisols. Soils used in the containers were taken from the top 20 cm of soil layer in October 2020. They were sieved to < 2 mm after air-drying and removing visible plant residues and stones. The fundamental characteristics of the soil samples were presented in Table 1 .

The citrus employed in the present study was of the first-year seedling stage, and the species is the Ehime mandarin 38^th. The legume and non-legume green manure residues were hairy vetch (Vicia villosa) and rattail fescue (Vulpia myuros (L.) C.C. Gmel.), respectively. The green manures were freshly harvested from the field in March, 2021 and then finely chopped into 2 mm in size. The nutrient content of the green manure residues was presented in Table 2 .

A pot-based experiment was conducted from March to September 2021 for the present study, which was carried out within a greenhouse facility located at Southwest University (29°81′ N, 106°42′ E), Chongqing, China. The experiment was designed using a completely randomized approach, considering two primary factors: (1) two treatments related to soil P levels, specifically P-surplus and P-deficient soils; (2) three treatments related to addition materials, including legume green manure (Leg), non-legume green manure (Non-Leg) and a control treatment with no addition (CK). A fixed quantity of 2 g·C·kg^-1 (dry soil) of green manure was added in all treatments. Due to the significant individual variations and potential systematic errors in fruit tree cultivation, each treatment was replicated five times. Each pot was filled with 10 kg of mixed soils, green manure residues and chemical fertilizers (40 mg·kg^-1 N, 40 mg·kg^-1 P₂O₅ and 80 mg kg^-1 K₂O were used as the foundation fertilizers, 40 mg·kg^-1 N were used as top-dress to ensure sufficient nutrition to citrus growth). The N, P and K fertilizers used in this study was urea, KH₂PO₄ and K₂SO₄, respectively. In addition, 0.5 g KH₂P¹⁸O₄ (containing 80‰ δ¹⁸O_p, diluted from KH₂P¹⁸O₄ (¹⁸O₄, 95%), purchased from Cambridge Isotope Laboratories) were applied to each pot to track the fate of exogenous P fertilizer. Ehime mandarin 38^th citrus seedlings were subsequently transplanted into pots and watered daily to 60% of the field water capacity. The position of pots switched once a week to minimize possible environmental effects.

Destructive sampling was conducted in September 2021, after citrus summer tips turned green. Plant samples were collected by separating the roots, stems, new leaves, and old leaves, following measurements of citrus height, stem thickness, and biomass. The collected plant samples were subjected to oven-drying at 105°C for 30 minutes, followed by further oven-drying at 70°C until a constant weight was achieved. Subsequently, the dried samples were ground and sieved through a 0.5 mm mesh size to determine the plant P concentration. The soil samples were divided into two portions. One portion was air-dried for the determination of total P and available P content, while the other portion was preserved at -80°C. The preserved samples were later used for analyzing phosphatase activity, the abundance of P-cycling genes, and the ¹⁸O_P values of different forms of P_i.

Soil total P and available P content were determined by molybdenum antimony anti-colorimetric method after digested with NaOH and extracted by NH₄F-HCl, respectively (Murphy and Riley, 1986). Different soil P fractions were sequentially extracted following a modified Hedley method (Hedley et al., 1982). Briefly, 0.5 g dried soil was sequentially extracted with 30.0 mL Milli-Q water (most labile P), 0.5 mol·L^-1 NaHCO₃ (pH 8.5) (labile and weakly adsorbed P), 0.1 mol·L^-1 NaOH (Fe/Al oxide-bound P) and 1.0 mol·L^-1 HCl (Ca-P minerals) after shaking overnight (16 h) at 25°C at 165 rpm. The supernatants were collected by centrifugation (10 min at 5000 rpm) and then filtered through 0.45 mm cellulose-acetate filters membrane to determine total P (P_t) and P_i content, the difference between P_t and P_i was taken as P_o content. Residual-P was determined by the molybdenum-antimony colorimetry after digested using the mixture of H₂O₂ and H₂SO₄.

Plant P concentration was measured by the molybdovanado phosphate method after digested in concentrated H₂SO₄ and H₂O₂ (Thomas et al., 1967).

Soil acid phosphatase (ACP) and alkaline phosphatase (ALP) activities were determined by the disodium phenyl phosphate colorimetric method (Jin et al., 2016). The ACP was extracted using acetate buffer at pH 5.0, while the ALP was extracted using borate buffer at pH 10.0.

The fresh soil samples (0.5 g) was subjected to whole genomic DNA extraction using FastDNA ^® Spin Kit (MP Biomedical, Santa Ana, USA). DNA purity and quality were detected by NanoDrop ND-2000 spectrophotometer and nucleic acid integrity was determined by agarose gel electrophoresis. The tested P-cycling genes, including pqqC (pyrroloquinoline-quinone synthase), phoD (alkaline phosphatase D), phoC (acid phosphatase C) and bpp (β-helical phytase), as well as 16S rRNA gene as a reference for bacterial abundance were quantified by quantitative PCR (qPCR) reations via QuantStudioTM 6 Flex Real-Time System (Applied Biosystems, USA). All primers used in this study were listed in Table 3 . The amplification program initialized with 95°C for 10 min, following with 40 cycles of 95°C for 5 s, 58°C for 30 s and 72°C 1 min. Each PCR assay was performed in triplicate, with amplification efficiencies between 91-100%. The copy number of each target gene in soil DNA was calculated based on the standard curve and the mean Ct value. The relative abundance of P-cycling genes was calculated as the ratio of cope numbers of respective P-cycling gene and 16S rRNA gene.

The measurement of oxygen isotope ratios in phosphate was followed the methods of Bi et al. (2018) and Jiang et al. (2017). In brief, 25.0 g freeze-dried fresh soil was sequentially extracted with H₂O, NaHCO₃, NaOH and HCl. Frist, the magnesium-induced co-precipitation (MAGIC) method was used to enrich PO₄ ^3- and added DAX-8 macroporous resin to eliminate the organics in samples. Then, the coprecipitation of ammonium phosphomolybdate (APM) and magnesium ammonium phosphate (MAP) method was used to further separated and purified PO₄ ^3-. After anion cation resin purified, PO₄ ^3- was converted into Ag₃PO₄ precipitation by ammonia volatilization method. Finally, the mineral structure of Ag₃PO₄ was detected by X-ray diffractometer (XRD) and compared with the standard pattern to check the purity of the sample and the soil ¹⁸O_P values were measured after the sample passing the test.

The values presented in the figures and tables are given as means ± standard errors. One-way analysis of variance (ANOVA) tests was used to examine the significant changes in soil P fractions, enzyme activities and the microbial gene abundance, citrus growth parameters and citrus P uptake under CK, Leg and Non-Leg treatments in P-surplus and P-deficient soils, respectively. Spearman correlation analysis was conducted to reveal the relationship between soil P fractions, the δ¹⁸O_P values of P_i pools and phosphatase activities, P-cycling gene relative abundance in P-surplus and P-deficient soils, respectively. All the statistics analysis were performed with IBM SPSS Statistics 22.0.

To evaluate the direct and indirect factors (including green manure addition, soil phosphatase activities and P-cycling genes abundances) affecting soil P sources transformation and citrus P uptake under different treatments in P-surplus and P-deficient soils, partial least squares path models (PLS-PM) were constructed using the R 4.2.3 packages “vegan” and “plspm”. And all figures in the present study were plotted using GraphPad Prism 8.0 and Adobe Illustrator 2023.

In comparison to the control treatment (CK), the addition of green manure residues did not yield a statistically significant impact on the growth of citrus plants in P-surplus soil, except for a 5.5% increase (p < 0.05) in stem thickness observed under the Leg treatment ( Figure 1E ). However, both citrus biomass and stem thickness exhibited significant increases (p < 0.05) under both Leg (19.0% and 11.5%, respectively) and Non-Leg (20.0% and 16.5%, respectively) treatments in P-deficient soil ( Figures 1B, F ). Additionally, plant height also experienced a significant increase (p < 0.05) under the Non-Leg treatment, with an 11.1% increase ( Figure 1D ).

In P-surplus soil, the addition of legume green manure residues significantly increased P uptake of citrus roots and new leaves by 29.6% and 80.1% (p < 0.05), respectively ( Figures 2A, G ). And, a 21.3% increase (p < 0.05) of P uptake in citrus roots by Non-Leg treatment was also observed ( Figure 2A ). In P-deficient soil, there were significant increases (p < 0.05) in P uptake observed in citrus roots and old leaves under both the Leg and Non-Leg treatments, compared to the control treatment (CK). Specifically, there was a 21.7% increase in citrus roots and a 46.7% increase in old leaves under the Leg treatment, while the Non-Leg treatment showed a 42.1% increase in citrus roots and a 37.9% increase in old leaves ( Figures 2B, F ). Additionally, P uptake in new leaves significant increased (p < 0.05) 36.3% by Leg treatment ( Figure 2H ).

In P-surplus soil, the H₂O-P_i and NaHCO₃-P_o content demonstrated a significant increase (p < 0.05) under both Leg (13.6% and 8.9%, respectively) and Non-Leg (9.2% and 8.5%, respectively) treatments. Furthermore, the NaOH-P_i and NaOH-P_o content also significantly increased by 9.5% and 30.0% (p < 0.05) under the Leg treatment, respectively ( Table 4 ).

In P-deficient soil, the NaHCO₃-P_i and NaHCO₃-P_o content showed a significant increase (p < 0.05) under both Leg (66.3% and 34.8%, respectively) and Non-Leg (299.0% and 132.6%, respectively) treatments. Additionally, the H₂O-P_i and NaOH-P_i content also significantly increased by 150.0% and 59.0% (p < 0.05) under the Leg treatment, respectively ( Table 4 ).

The content of HCl-P_i and Residual-P did not significantly change among the different treatments in both P-surplus and P-deficient soils.

Green manure residues addition changed the phosphate oxygen isotope ratios of different P_i pools in both P-surplus and P-deficient soils ( Figure 3 ). In P-surplus soil, compared with CK, the δ¹⁸O_P values of NaHCO₃-P_i and NaOH-P_i under the Leg treatment, as well as NaHCO₃-P_i under the Non-Leg treatment, tended to approach or shift toward the isotopic equilibrium zone ( Figure 3A ). This indicated that the biological cycling of these P_i pools may be rapid and related to the oxygen isotope equilibration. In addition, the δ¹⁸O_P values of HCl-P_i under the Leg treatment would be lighter than the other treatments.

In P-deficient soil, the phosphate oxygen isotope ratios of various P_i pools exhibited similar variations under both Leg and Non-Leg treatments. Compared to CK, the δ¹⁸O_P values of NaHCO₃-P_i showed a considerable increase (shifted toward the isotopic equilibrium zone) following the addition of green manure residues ( Figure 3B ). This observation suggests that microorganisms play a crucial role in the turnover of NaHCO₃ pools.

In P-surplus soil, the activities of ACP and ALP exhibited a significant decrease (p < 0.05) under the Leg than CK treatment, while no significant change was observed by treated with Non-Leg ( Figures 4A, C ). Conversely, in P-deficient soil, the activities of soil ACP and ALP significantly increased (p < 0.05) under both Leg and Non-Leg than CK treatments, with a higher increase observed under the Leg treatment ( Figures 4B, D ).

In P-surplus soil, the addition of green manure residues did not have a significant effect on the relative abundances of soil P-cycling genes, except for the phoC gene. Specifically, under the Leg treatment, the relative abundance of the phoC gene exhibited a significant increase (p < 0.05) of 576.6% compared to CK ( Figures 5A, C, E, G ). In P-deficient soil, compared to CK, the relative abundance of pqqC, phoC, phoD and bpp genes under the Leg treatment significantly increased (p < 0.05) by 48.1%, 42.9%, 21.7% and 27.4%, respectively. And the relative abundance of pqqC, phoD and bpp genes under the Non-Leg treatment significantly increased (p < 0.05) by 45.1%, 33.3% and 18.6%, respectively ( Figures 5B, D, F, H ).

In P-surplus and P-deficient soils, the effects of green manure addition on soil P fractions and citrus P uptake were correlated with the majority microbial parameters in P-cycling, including enzymes activities and P-cycling genes abundances ( Figure 6 ).

In P-surplus soil, the addition of green manure residues had a positive impact on the turnover of soil P fractions through direct effects (path coefficient = 0.59). Additionally, the influence of green manure residues on citrus P uptake was mediated indirectly through alterations in soil phosphatase activities, including both ACP and ALP ( Figure 6A ).

In P-deficient soil, the addition of green manure residues had a positive influence on the turnover of soil P fractions through both direct effects (path coefficient = 0.55) and indirect effects, mediated by alterations in soil phosphatase activities (path coefficient = 0.49). These changes in soil phosphatase activities or soil P pools, in turn, affected citrus P uptake (path coefficient = 0.40). Additionally, citrus P uptake was also directly influenced by the addition of green manure residues (path coefficient = 1.46) ( Figure 6B ).

This study confirmed green manure residues incorporation had a positive effect on the enhancement of soil P pools ( Figure 6 ; Table 4 ). This effect was mediated by both green manure species and initial soil P levels, characterized that legume green manure had a superior capacity to regulate soil P pools compared to non-legume green manure, and was more prominent in P-deficient soil.

Green manure residues contain considerable amounts of nutrients and their decomposition results in the release of these nutrients into the soil, which can assist in fulfilling the nutritional requirements of crops (Calegari et al., 2013; Jamal et al., 2023). Hinsinger et al. (2011) and Pavinato et al. (2008) indicated that organic forms of P released during residues decomposition were less prone to strong adsorption on the functional groups of Fe and Al oxides and hydroxides compared to inorganic forms. And organic decomposition products can also compete for adsorption sites, thereby increasing bioavailable of soil P pool (Varela et al., 2017; Soltangheisi et al., 2020). Maltais-Landry and Frossard (2015) also reported that comparable transfer of P from green manure residues and water-soluble mineral fertilizer to the soil was observed. Consistently, Hallama et al. (2019) confirms the observation of this study, that legume green manure was found to be more nutrient efficient than non-legume green manure. This can be primarily attributed to the following reasons. Firstly, legume green manure exhibits a higher capacity for P uptake and accumulation due to its abundant biomass (Arrobas et al., 2015). In regard to the green manure materials supplied by this study, when subjected to equal carbon input (2 g·C·kg^-1 soil), the P input of legume green manure was 270 mg·P·pot^-1, while the P input of non-legume green manure was 140 mg·P·pot^-1 ( Table 2 ). Consequently, there is a greater release of P during decomposition, thereby replenishing the labile soil P pools and promoting the growth of the main plant. Secondly, legume green manure with lower C/N and C/P ratios can more rapidly facilitate microbial decomposition and utilization, as well as stimulate the activity of P-cycling microorganisms (Truong and Marschner, 2020). This leads to the formation of greater microbial biomass P (Benitez et al., 2016), which is considered a potential active P pool in the soil (Hallama et al., 2019). These two factors collectively increased soil active P fraction (e.g. H₂O-P_i, NaHCO₃-P_o, NaOH-P_i and NaOH-P_o in P-surplus soil and H₂O-P_i, NaHCO₃-P_i, NaHCO₃-P_o and NaOH-P_i in P-deficient soil ( Table 4 ) and enhanced the growth and P nutrition of citrus plants ( Figures 1 , 2 ).

The impact of green manure on soil P fractions were greatly influenced by the soil condition, particularly the size of the easily and sparingly available P pool (Damon et al., 2014). Similar to the finding of this study, the meta-analysis conducted by Hallama et al. (2019) reported that the benefits of green manure in terms of soil biological P were more pronounced in soils that had limited availability of P compared to sites with higher P availability. This study provided evidence that the contribution of green manure residues to labile soil P pools and citrus growth was comparatively limited in the context of the already abundant P availability in the soil ( Figures 1 , 2 , 6 ; Table 4 ). This perspective was also confirmed by the research of Rick et al. (2011). Their study demonstrated that incorporation of green manure and input of rock P fertilizer had no significant impact on labile soil P fractions, wheat biomass, or P concentration due to high soil initial P availability, which already satisfied the requirements for crop growth.

There is a consensus that the incorporation of green manure residues into soil can serve as a source of C and energy for soil microorganisms (Huang et al., 2021), which in turn promotes increased metabolic activities of P-cycling microbial populations and facilitates the conversion of bio-unavailable P into bio-available forms (Pavinato et al., 2017; Zhou et al., 2018; Arruda et al., 2021). In the current study, the effects of microorganisms on soil P cycling were investigated using oxygen isotope labeling on exogenous chemical P fertilizers. This labeling technique allowed for tracking and monitoring the behavior and transformations of P_i within soil, and the closer isotope value of a P pool was to the theoretical equilibrium, the faster that the P pool was biologically cycled, indicating higher bioavailability. Conversely, if the isotope value deviated from the equilibrium, it suggested slower biological cycling and lower bioavailability of P (Bi et al., 2018; Tian et al., 2020). This study proved that incorporating green manure increased the conversion of chemical P fertilizer by microorganisms. In P-surplus soil, microorganisms quickly transformed P sources into H₂O-P_i and NaHCO₃-P_i pools under the Leg treatment and into NaHCO₃-P_i pools under the Non-Leg treatment ( Figure 3A ). In P-deficient soil, both Leg and Non-Leg treatments promoted the conversion of P sources into NaHCO₃-P_i and NaOH-P_i pools ( Figure 3B ). These findings provide direct validation of the pivotal role played by green manure in the mobilization of soil P through the activation of soil microorganisms. This activation is achieved by increasing the abundance of P-cycling bacteria and enhancing the activities of P-cycle enzymes (Piotrowska-Dlugosz and Wilczewski, 2020; Hallama et al., 2021). By promoting these microbial processes, green manure contributes to the efficient cycling and availability of phosphorus in the soil ecosystem.

In the present study, green manure residues addition had a positive effect on pqqC, phoC, phoD and bpp genes abundances and ACP and ALP activities, which was more prominent in P-deficient soil ( Figures 4 , 5 ). In response to a P_i-limited soil environment, microorganisms tend to release a higher quantity of phosphatases (Bergkemper et al., 2016). These enzymes are capable of effectively hydrolyzing ester-phosphate bonds in various phosphate-esters, thereby promoting the mineralization of soil insoluble P. This process enhances the overall effectiveness of P in the soil and facilitates its availability for biological uptake and cycling (Lu et al., 2022; Touhami et al., 2023).

Moro et al. (2021) also observed that in the soil with low P_i availability, microorganisms would be stimulated, especially bacteria, to synthesize large amounts of phosphatases, including acid phosphomonoesterase and phosphodiesterase, which promoted the hydrolysis of soil P_o, replenished soil active P pools and supplied plant growth. In contrast to the expected response, the incorporation of legume green manure in P-surplus soil, as observed in this study, resulted in a decrease in ACP and ALP activities ( Figures 4A, C ). Indeed, the study by Spohn and Kuzyakov (2013) indicated that in soils with ample P availability, soil microbial biomass P was primarily influenced by the release of plant-available P as a by-product of C mineralization. The incorporation of legume green manure in P-surplus conditions lead to an increased availability of plant-available P, reducing the need for microbial phosphatase activity ( Table 4 ). This finding could potentially explain the observed decrease in phosphatase activities under the Leg treatments in P-surplus soil of present study ( Figures 4A, C ).

The results of this study, except for the phoC gene in P-surplus soil, showed no significant differences in the abundance of other P-cycling genes between the incorporation of legume and non-legume green manure treatments ( Figure 5 ). This finding contradicts our understanding that organic matter with a low C/P ratio, typically found in legume green manure, has a greater capacity to stimulate P-microbial abundance compared to organic matter with a high C/P ratio, typically found in non-legume green manure (Benitez et al., 2016; Erinle and Marschner, 2020; Erinle et al., 2020). It is worth noting that the C/P ratio of the legume green manure and non-legume green manure used in this study was 73 and 141, respectively, both below 200 ( Table 2 ). These ratios fall within the same range, which could explain the limited differences observed in P-cycling genes abundances. Alamgir et al. (2012) pointed out that crop residues with a C/P ratio above 200 tend to induce net P immobilization and depletion of P pools, while residues with a C/P ratio below 200 increase P availability and P pools. Additionally, Blanco-Canqui et al. (2015) suggested that the increase in effective phosphorus (P_o and P_i) pools through the regulation of microbial biomass and enzyme activities is primarily associated with an overall increase in soil organic matter. In other words, the exogenous inputs of C play a significant role in activating P-cycling microbial abundance. Since the amount of C inputs from different green manure sources in this study were consistent, the effects on functional gene abundances may tend to converge as well ( Figure 5 ; Table 2 ).

In P-deficient soil, after the incorporation of green manure residuals, the increase in soil active P pools, along with the enhanced abundance and activity of microorganisms, indirectly promoted the growth, development, and P nutrition status of citrus plants ( Figures 1B, D, F , 2B, D, F, H , 6B ; Table 4 ). However, the current research does not provide a consistent conclusion regarding the growth and P nutritional status of citrus plants when different green manure varieties are incorporated. Previous studies (Shackelford et al., 2019; Khan et al., 2022; Ullah et al., 2023) have demonstrated that non-legume green manure incorporation did not favor plant growth, while legume green manure incorporation did. The main reason for the inconsistency observed is that the chemical N supply in this study was sufficient, which may have hindered the beneficial N supply from legume green manure. Secondly, citrus plants are perennial crops and have a slower growth and development compared to annual grain crops. Their slower growth rate and development can limit the observable differences in response to different green manure treatments.

In the P-surplus soils of the present study, the addition of green manure residues did not show a significant improvement in the growth and P nutrition status of citrus plants, except for the P uptake by citrus roots ( Figures 1A, C, E , 2A, C, E, G ). This unexpected result can be attributed to the adequate availability of pristine soil nutrients and chemical fertilizers and the antagonistic effect of trace element uptake by citrus plants in P-surplus soils. Previous research has consistently demonstrated that high availability of P in the soil can hinder the uptake of essential trace elements such as zinc, copper, and iron by plants (Aboyeji et al., 2020; Singh et al., 2021; Stanton et al., 2022). This interference in trace element uptake can result in reduced yields and overall crop performance in various crops.

This study has provided a scientific foundation for quantifying the contributions of green manure to soil P pools and P nutrient of citrus plants, thereby enhancing production efficiency and fostering sustainable development in orchards. However, certain limitations were encountered during the experimental process. Firstly, it is essential to acknowledge that the annual citrus seedlings were employed in this experiment, which differs from the mature citrus trees typically found in actual orchard production. As a result, factors such as the kinetic uptake of soil P and the capacity to regulate the quantity, species, and activity of soil microorganisms vary between seedlings and mature trees. Consequently, the obtained results can only offer limited insights into seedling management within citrus orchards. Secondly, the environmental conditions of the pot experiment were deliberately standardized and controlled, unlike the intricate and dynamic situation present in actual orchard production. The use of potting apparatus may have restricted the growth of plants, particularly impeding root development. Research has demonstrated that P deficiencies induce changes in plant root architecture, leading to an increase in the root-to-shoot ratio, root hairs, root topsoil foraging, and root morphology (Zhang et al., 2013; Su et al., 2014; Iqbal et al., 2020). Hence, in future research endeavors, it is imperative to direct greater attention toward investigating the P uptake status of mature orchards under natural field conditions.