Microbial controls over soil priming effects under chronic nitrogen and phosphorus additions in subtropical forests

Abstract

The soil priming effect (PE), defined as the modification of soil organic matter decomposition by labile carbon (C) inputs, is known to influence C storage in terrestrial ecosystems. However, how chronic nutrient addition, particularly in leguminous and non-leguminous forests, will affect PE through interaction with nutrient (e.g., nitrogen and phosphorus) availability is still unclear. Therefore, we collected soils from leguminous and non-leguminous subtropical plantations across a suite of historical nutrient addition regimes. We added ¹³C-labeled glucose to investigate how background soil nutrient conditions and microbial communities affect priming and its potential microbial mechanisms. Glucose addition increased soil organic matter decomposition and prompted positive priming in all soils, regardless of dominant overstory tree species or fertilizer treatment. In non-leguminous soil, only combined nitrogen and phosphorus addition led to a higher positive priming than the control. Conversely, soils beneath N-fixing leguminous plants responded positively to P addition alone, as well as to joint NP addition compared to control. Using DNA stable-isotope probing, high-throughput quantitative PCR, enzyme assays and microbial C substrate utilization, we found that positive PE was associated with increased microbial C utilization, accompanied by an increase in microbial community activity, nutrient-related gene abundance, and enzyme activities. Our findings suggest that the balance between soil available N and P effects on the PE,  was dependent on rhizosphere microbial community composition. Furthermore, these findings highlight the roles of the interaction between plants and their symbiotic microbial communities in affecting soil priming and improve our understanding of the potential microbial pathways underlying soil PEs.

Introduction

Soil organic matter (SOM) is the largest terrestrial carbon (C) pool, and therefore, a slight change in SOM decomposition rates can greatly impact atmospheric CO₂ concentrations and global climate [1, 2]. The decomposition of SOM is mainly driven by soil microorganisms, and this process is strongly affected by exogenous labile C inputs [3, 4]. A sufficient increase in the amount of labile C can “jumpstart” decomposition, much like priming a pump. This positive feedback, where an increase in labile C can increase the rate of SOM decomposition, is consequently referred to as “positive priming effect (PE)” [5, 6]. On a global scale, PE can increase the decomposition rate of SOM by up to 60% and enhance the release of CO₂ from the soil by up to 50% [7, 8]. Given concerns about rising atmospheric C levels and increased global prioritization for the conservation of soil C sinks, it is essential to understand factors that drive SOM PE on soil C cycling, in order to improve our measurement and estimation of soil C dynamics [9, 10].

Although research on PEs has been increasing rapidly in recent years, it can be challenging to pin down its underlying mechanisms. Priming can vary in magnitude and direction among ecosystems, which makes it challenging to predict [4, 11, 12]. This substantial variation can be attributed to many factors, such as differences in microbial community composition and their functional traits [12,13,14,15]. Soil microorganisms, as the main decomposers of SOM, play a critical role in priming [4, 16]. The complexity of belowground microbial interactions makes predicting the influence of the soil microbiome difficult, but not impossible. For example, research suggests that the variation of microbial biomass and the ratio between r and K-strategists within microbial communities can strongly affect PE [11, 16, 17]. The increased availability of high-throughput sequencing data has allowed us to identify important taxa that may control decomposition. Recent work demonstrates that specific microbial taxa, such as Proteobacteria and Acidobacteria can contribute to SOM decomposition, and they may play a crucial role in regulating the soil PE [18, 19]. In addition, breakthroughs in microbial functional gene analysis suggest that microbial genes related to nutrient cycling (e.g., genes related to C, N, and P decomposition and acquisition) are closely linked to SOM decomposition and might be the key to understanding PE [20,21,22,23]. These interactions between microbial gene expression and the PE remain complex, but, demand further investigation. Moreover, variations in microbial biomass, function, and composition can impact SOM priming, and therefore, factors that impact microbial communities, such as soil nutrient availability, can also affect PE [3, 12, 24]. For instance, variation in available soil nitrogen (N) can have extensive effects on soil microbial biomass, activity, composition, and enzyme activity, and ultimately affect SOM decomposition [25, 26]. Microbial decomposition is often linked to Sprengel-Liebig’s Law of the Minimum (i.e., growth is dictated not by total resources available, but by the scarcest resource) or stoichiometric decomposition theory (i.e., microbial activity is highest and decomposition rates are greatest if the inputs of C and N match the microbial demand for substrate) [11, 27, 28]. As a result, microbial decomposition ability should be constrained by either the scarcest nutrient or imbalance of microbial nutrients. This principle is further complicated by global patterns of nutrient deposition, which have greatly influenced ecosystem nutrient balances, particularly in the tropics. For example, an increase in available N may not be accompanied by an increase in P, altering the stoichiometry between bioavailable N and P [25, 29]. However, we have a limited understanding of the influence of nutrient stoichiometry, such as the balance between N and phosphorus (P), on the PE.

Tropical and subtropical forest ecosystems account for a large part of the total global forest area and soil organic C stocks [30, 31]. These highly weathered ecosystems are often P-limited, and most P in the system is often bound to oxides and hydroxides of aluminum and iron, given the highly acidic soil and high levels of cation bridging [29, 32]. In addition, tropical and subtropical forests have been increasingly regarded as N deposition hotspots with potentially harmful influences on forest health and biodiversity due to the increased rapid consumption of N fertilizers and fossil fuels in recent decades [33, 34]. Increasing rates of N deposition may further exacerbate P deficiency as it likely disturbs a previously balanced rates of consumption between N and P in ecosystems [35, 36]. This increasing P limitation would suggest that P addition would dramatically improve growth in these systems, improving their C storage capacity. Furthermore, soil priming is also impacted by plant species, and species-specific microbial communities around plant roots may lead to species-level variation in priming responses [37, 38]. This physiological difference may extend to broad plant functional categories. Legumes, for example, can obtain additional N from the atmosphere through symbiotic N₂ fixation by rhizobia bacteria [39, 40]. In this case, soil priming may be less influenced by abiotic N than by abiotic P, as the N demands of the plant can be met through facultative mutualisms.

Under these assumptions, we would imagine a commensurately stronger influence of P in leguminous plants that are simultaneously capable of meeting an increased N demand through root-associative N-fixation. Therefore, we hypothesized that P addition would have a greater effect on PEs in legume soils. We further hypothesized that nutrient addition enhances microbial C utilization by influencing the microbial community structure and their activities, such as functional gene expression and enzyme activity, which in turn induces greater PEs. To test these hypotheses, we collected soil from one plantation dominated by an N-fixing tree species (Acacia auriculiformis) and another dominated by a non-N-fixing tree species (Eucalyptus urophylla) exposed to 10 years of N and/or P addition. We then conducted a soil incubation experiment, where ¹³C-labeled glucose was added as the labile C, to understand how N and P availability impact SOM priming and its underlying microbial processes in soils growing different plant functional types.

Materials and methods

Study sites and field treatments

The study site was located at the Heshan National Field Research Station of Forest Ecosystems (112°50′E, 22°34′N) in southern China. The region has a tropical monsoon climate, with a mean annual precipitation of 1580 mm, 80% of which occurs from April to September. The annual temperature is approximately 22 °C, ranging from 12.6 °C (January) to 28.0 °C (July) [41]. Study sites were located in two forest plantation types, dominated by either leguminous plants (LP) or non-leguminous plants (NLP). Both plantations sampled in this study are over 30 years old. The dominant species in the canopy layer of the LP plantation was Acacia auriculiformis. The predominant species of the NLP plantation was Eucalyptus urophylla. The soil in both plantations could be classified as acrisol (FAO, 2006) and soil properties are summarized in Table 1.

Table 1 Soil and microbial characteristics.

The N and P field addition experiment was established with a complete randomized block design in August 2010. The field experiment included a control plot without nutrient addition (henceforth CK), an N-addition treatment of 100 kg N ha⁻¹ yr⁻¹ (henceforth N), a P-addition treatment of 100 kg P ha⁻¹ yr⁻¹ (henceforth P), and an N + P-addition treatment of 100 kg of both N and P ha⁻¹ yr⁻¹ (henceforth NP). Treatments were applied to 10 m × 10 m plots in three separate locations (blocks) in each plantation (Fig. S1). Ammonium nitrate (NH₄NO₃) and/or dihydrogenphosphate (NaH₂PO₄) dissolved in 10 L water was used for the additional treatments, and these solutions were sprayed bimonthly. To keep water applications consistent between plots, the CK plot received 10 L of water at the time of nutrient addition treatments.

Ten soil samples were randomly collected from the upper 20 cm soil layer of each plot and mixed homogeneously into a single composite soil sample. In this fashion, a total of 24 (4 nutrient treatments × 2 plantations × 3 replicates) soil samples were collected. Soil samples were transported to the lab immediately and passed through a 2-mm sieve.

Incubation experiment

Soil samples were adjusted to 40% water-holding capacity and then pre-incubated in a climate-controlled room (~22 °C) for one week before the start of the incubation experiment to allow settling after the sampling and sieving disturbance [42, 43]. Pre-incubated soil from each treatment was weighed and divided into eight equal parts of 40 g each and placed into 50 mL vials. Vials were then placed into 2 L headspace chambers with lids. In total, 48 headspace chambers contained eight vials each (4 nutrient treatments × 2 plantations × 3 replicates × 2 parallels). The PE was calculated by adding demineralized purified water containing dissolved ¹³C-glucose to the vials (hereafter referred to as glucose treatment). This resulted in a C addition of ~450 µg C/g fresh soil, equivalent to the average microbial biomass C in all soil samples. Control vials received an equal amount of demineralized purified water (hereafter referred to as control). All headspace chambers were incubated in the dark in a climate-controlled room at ~22 °C, matching the annual average temperature at the field site [41]. After incubation for 0, 6, 18, 42, 66, 114, and 162 h, soil respiration and available N were measured. Soil C substrate utilization, microbial biomass, phospholipid fatty acids, and potential enzyme activities were measured at the end of the incubation (Fig. S2).

Soil respiration and priming effect

At each time point (0, 6, 18, 42, 66, 114, and 162 h), one vial was removed from the headspace for analysis. After being removed from the headspace chamber, the vial was placed into a 1-L airtight bottle equipped with a catheter. The headspace was purged with CO₂-free air before the bottle was closed with a lid. The bottles were incubated for 5–24 h at 22 °C in the dark, depending on the time required to obtain enough CO₂ [42]. Then, the gas was collected into an airbag, and the CO₂ concentration and its isotopic composition (¹³C/¹²C ratio) were measured using a Picarro G2201-i analyzer (Picarro Inc., Santa Clara, CA, USA). The released CO₂ that originated from SOM and glucose was calculated using the following equations:

$${f}_{{{{{{\rm{glucose}}}}}}}=({{{{{\rm{atom}}}}}}{ \% }^{13}{{{{{{\rm{C}}}}}}}_{{{{{{\rm{CO2}}}}}}}-{{{{{\rm{atom}}}}}}{ \% }^{13}{{{{{{\rm{C}}}}}}}_{{{{{{\rm{SOM}}}}}}})/({{{{{\rm{atom}}}}}}{ \% }^{13}{{{{{{\rm{C}}}}}}}_{{{{{{\rm{glucose}}}}}}}+{{{{{\rm{atom}}}}}}{ \% }^{13}{{{{{{\rm{C}}}}}}}_{{{{{{\rm{SOM}}}}}}})$$

$${f}_{{{{{{\rm{SOM}}}}}}}=1-{f}_{{{{{{\rm{glucose}}}}}}}$$

$${{{{{{\rm{R}}}}}}}_{{{{{{\rm{SOM}}}}}}}={{{{{{\rm{R}}}}}}}_{{{{{{\rm{total}}}}}}}\times {f}_{{{{{{\rm{SOM}}}}}}}$$

$${{{{{{\rm{R}}}}}}}_{{{{{{\rm{glucose}}}}}}}={{{{{{\rm{R}}}}}}}_{{{{{{\rm{total}}}}}}}-{{{{{{\rm{R}}}}}}}_{{{{{{\rm{SOM}}}}}}}$$

Where f_glucose and f_SOM are the fractions of released CO₂ derived from glucose and SOM, respectively, and atom% ¹³C_glucose and atom% ¹³C_SOM represent the atom% ¹³C in glucose and SOM, respectively. The total released CO₂ (R_total) was then partitioned into CO₂ released from SOM decomposition (R_SOM) and CO₂ released from the breakdown of glucose (R_glucose).

The PE was then calculated as the difference in the released CO₂ from SOM between samples with and without glucose addition:

$${{{{{\rm{Priming}}}}}}\,{{{{{\rm{effect}}}}}}({{{{{\rm{\mu }}}}}}{{{{{\rm{g}}}}}}\,{{{{{\rm{C}}}}}}\,{{{{{{\rm{g}}}}}}}^{-1}{{{{{\rm{soil}}}}}})={{{{{{\rm{R}}}}}}}_{{{{{{\rm{SOM}}}}}} \, {{{{{\rm{with}}}}}} \, {{{{{\rm{glucose}}}}}}}-{{{{{{\rm{R}}}}}}}_{{{{{{\rm{SOM}}}}}} \, {{{{{\rm{in}}}}}} \, {{{{{\rm{control}}}}}}}$$

$${{{{{\rm{Priming}}}}}}\,{{{{{\rm{effect}}}}}}( \% )=({{{{{{\rm{R}}}}}}}_{{{{{{\rm{SOM}}}}}} \, {{{{{\rm{with}}}}}} \, {{{{{\rm{glucose}}}}}}}-{{{{{{\rm{R}}}}}}}_{{{{{{\rm{SOM}}}}}} \, {{{{{\rm{in}}}}}} \, {{{{{\rm{control}}}}}}})\times 100 \% /{{{{{{\rm{R}}}}}}}_{{{{{{\rm{SOM}}}}}} \, {{{{{\rm{in}}}}}} \, {{{{{\rm{control}}}}}}}$$

where R_{SOM with glucose} is the CO₂ released from SOM with glucose addition treatment, R_{SOM in control} is the CO₂ released from SOM in the control without glucose addition.

Soil properties and microbial biomass

Initial soil properties were measured before the incubation. Soil pH was determined in a 1:5 ratio of soil and water slurry using a combination glass electrode meter (FiveEasyPlus FE28, Mettler Toledo, Switzerland). Soil organic carbon (SOC), soil total N (TN), soil inorganic N (IN), soil total P (TP), and soil available P (AP) concentrations were determined following protocols as described previously [44]. Microbial biomass C (MBC) and N (MBN) concentrations were determined by the chloroform fumigation method [45]. After the incubation, total microbial biomass was first determined by phospholipid fatty acid analysis (PLFA) using freeze-dried soil following soil PLFA protocols [46], with minor modifications [13]. Then the microbial ¹³C incorporated into biomass was calculated based on the atom% ¹³C excess in bacterial and fungal biomarker PLFAs determined by the GC-C-IRMS system [3].

Microbial DNA extraction and ultracentrifugation

Soil microbial DNA was extracted from 0.5 g soil using the DNA extraction kit (MP Biomedicals, Santa Ana, CA, USA) before and after the ¹³C-glucose addition incubation experiment. This DNA was further purified using a MO BIO purification kit (Carlsbad, CA, USA).

Ultracentrifugation was used to separate the ¹³C-DNA and ¹²C-DNA since the ¹³C-DNA is heavier than the ¹²C-DNA. Briefly, 3.0 μg microbial DNA was dissolved in 1.85 g/mL CsCl with the density adjusted by adding buffer (0.1 M pH = 8.0 Tris-HCl, 0.1 M KCl, 1.0 mM EDTA) or CsCl. The prepared DNA solution was transferred to a Beckman ultra-high-speed centrifugal tube and centrifuged at 177,000 × g at 20 °C for 44 h by an NTV-100 vertical rotor (Beckman Coulter, Palo Alto, CA, USA). The DNA solution was fractionated to 13 layers using a fixed velocity pump (New Era Pump Systems, Inc., Farmingdale, NY, USA) after centrifugation. The refractive index of each DNA layer was measured using a refractometer (Reichert Inc., Buffalo, NY, USA), and the DNA buoyant density was calculated based on the refractive index. All separated layer DNA was purified using polyethylene glycol 6000 precipitation and then stored in a freezer at −20 °C.

PCR amplification, DNA sequencing, and high-throughput quantitative PCR

Quantitative real-time polymerase chain reaction (qPCR) was used to quantify the DNA in each layer on a CFX96 Optical Real-Time Detection System (Bio-Rad, Laboratories Inc., Hercules, CA, USA) [47, 48]. The labeled DNA is expected to have a density of approximately 1.725 g/ml [48, 49], therefore, ultracentrifugal DNA fractions with a density between 1.69 and 1.75 g/ml were used for further sequencing and high-throughput quantitative PCR. Bacterial 16 S rRNA was amplified with primers F515 and R907 [50], and primers ITS1F and ITS2R [51] were used for fungal internal transcribed spacer (ITS). Sequencing was performed in the HiSeq 2000 System at Shanghai Majorbio Bio-pharm Technology Co., Ltd. Raw sequence data were filtered using Trim Galore software, and paired sequences were spliced using the FLASH2 software. The taxonomy of amplicon sequence variants (ASVs) was analyzed by QIIME2 against the 16 S rRNA database (SILVA v138). We followed a DADA2 pipeline to denoise the optimized sequences after quality control.

High-throughput qPCR was used to determine the abundance of genes involved in C, N, and P cycling by a SmartChip Real-time PCR system (Wafergen, Fremont, CA, USA) [52]. The protocol contained 66 primer pairs for nutrient-cycling genes (including 35 C-cycling genes, 22 N-cycling genes, and 9 P-cycling genes) and one 16 S rRNA gene primer (Table S1). The PCR amplification program followed the following steps. First, an initial denaturation of 95 °C for 10 min, followed by 40 cycles of denaturation at 95 °C for 30 s, followed by annealing at 58 °C for 30 s, and finally an extension at 72 °C for 30 s. The melting curve was automatically generated by the WaferGen software. The relative copy number of the genes was calculated as relative copy number = 10^{(31-CT)/(10/3)}, where CT represents the threshold cycle [52]. Then, the relative abundance of C-, N-, and P-cycling genes were expressed as copies/16 S rRNA gene.

Microbial carbon substrate utilization and potential enzyme activities

Soil microbial C substrate utilization was assessed as previously published with modifications [53]. The setup consists of two 96-well microtiter plates placed face to face. Fourteen different substrates, including five amino acids (L-alanine, L-cysteine, GABA, L-lysine, L-arginine); four carbohydrates (L-arabinose, D-fructose, D-galactose, yrehalose); four carboxylic acids (α-ketoglutaric acid, citric acid, L-malic acid, oxalic acid); one aromatic acid; and ultrapure distilled water (control) were added to four replicates of each soil sample and distributed to microtiter plates. The indicator plates were measured at 570 nm with a plate reader, before and after incubation for 6 h in the dark (Anthos 2010, Biochrom, Cambridge, UK). The C substrate utilization was calculated from the difference in CO₂ concentration between two measurement times and expressed as μg CO₂-C g⁻¹ soil h⁻¹.

Potential enzyme activities, including β-glucosidase, cellobiohydrolase, N-acetyl-β-D-glucosaminidase, leucine-aminopeptidase, acid phosphatase, phenoloxidase and peroxidase, were measured using a combination of fluorometric and photometric assays [54,55,56]. 2.0 g of fresh soil was dissolved in 100 mL of 100 mM sodium acetate buffer (pH 4.0). 200 µL aliquots of soil suspension were placed into 96-well plates, after which corresponding fluorescence-labeled substrates (Sigma Aldrich, CA, USA) were added. After incubation for 140 min, fluorescence intensity was measured by FLUO star Omega (BMG Labtech, Offenburg, Germany) with 365 nm excitation and 450 nm emission. Potential enzyme activities were then calculated according to the standard curve and expressed in nmol 4-methylumbelliferone (MUF) or 7-amino-4-methylcoumarin (AMC) g⁻¹ dry soil h⁻¹. Potential phenoloxidase and peroxidase enzyme activities were measured using 1 mL soil suspensions mixed with 1 mL 20 mM L-3,4-dihydroxyphenylalanine (DOPA) (Sigma Aldrich, CA, USA) as substrate. For peroxidase activity measurements, samples received 10 µL 0.3% (v/v) H₂O₂ as the additional substrate. Absorbance was measured at 450 nm with FLUO star Omega (BMG Labtech, Offenburg, Germany) after shaking and centrifuging before and after 20 h of incubation. The potential oxidative enzyme activities were calculated from the difference in absorbance between two times and expressed in nmol DOPA g⁻¹ dry soil h⁻¹.

Data analyses

For microbial characteristics, nutrient-related gene abundances, enzyme activities, and microbial C utilization, we calculated response ratios (RR) as the ratio between samples that received glucose and samples that did not receive glucose. Differences in soil properties and the RR of the microbial characteristics, nutrient-related gene abundances, enzyme activities, and microbial C utilization between treatments were identified by ANOVA with post hoc tests where appropriate, with p < 0.05 considered statistically significant. Repeated-measures ANOVA was used to test the main and interactive effects of glucose addition and incubation time on the cumulative SOM-derived CO₂, and three-way ANOVA was used to test the effect of plant species, N addition and P addition on the cumulative PE. A Spearman correlation, followed by post hoc tests, and the Mantel test were used to investigate the relationship between PEs and soil properties, the RR of microbial communities, nutrient-related gene abundance, enzyme activities, and microbial C utilization after incubation. The RR of bacterial and fungal communities was calculated as the absolute value of the difference in microbial taxa abundance between the control and glucose treatment. Principal co-ordinates analysis (PCoA) was performed to investigate the effect of plant species and field N and P treatments on the responses of bacterial and fungal communities. Structural equation modeling (SEM) was used to investigate the direct and indirect roles of soil N and P concentrations, microbial biomass, microbial diversity, enzyme activities, microbial C utilization, and nutrient-related gene abundance in driving the PE. All statistical analyses were performed using the R platform (version 4.2.2) and the online tool of the Majorbio Cloud Platform (https://cloud.majorbio.com/page/tools/).

Results

SOM-derived CO₂ and SOM priming effect

Our results show that glucose addition had a significant impact on the cumulative CO₂ emission derived from SOM, suggesting a positive PE beneath both leguminous plants (LP) and non-leguminous plants (NLP) (Fig. 1). Further, P treatments, both with and without N addition, significantly increased cumulative priming at the end of the incubation in LP soil, while N addition alone did not stimulate priming in these soils. In contrast, only the NP addition induced significantly higher priming in the NLP plantation.

Fig. 1: SOM-derived CO₂ and priming effects in different treatments.

Cumulative SOM-derived CO₂ fluxes (A, B) and priming effects (C, D) by treatment over the course of the incubation, and cumulative priming effects (E) at the end of the incubation (LP leguminous plant, NLP non-leguminous plant, S species, CK = no nitrogen or phosphorus addition, N = nitrogen addition, P = phosphorus addition, and NP = nitrogen and phosphorus addition). Different letters indicate significant differences at p < 0.05.

Soil microorganism responses to field and incubation treatments

As expected, prior to lab incubation, plant species and field nutrient addition significantly influenced both bacterial and fungal community compositions (Fig. S4). After incubations, microbial community responses to glucose addition (response ratios, RR) also significantly differed by dominant plant species and nutrient addition (Fig. 2). Specifically, principal co-ordinate analysis (PCoA) showed that the profiles of bacterial and fungal communities from different plant soils formed clusters (Fig. 2, ANOSIM <0.05). Further, community similarity analysis revealed significant differences in both bacterial and fungal communities among different nutrient treatments in both plant soils (Fig. 2). In addition, bacterial ¹³C biomass was higher in the NLP plantation than in the LP plantation and bacterial and fungal ¹³C biomass significantly increased with P addition (Table 2). Plant species identity also significantly affected the RR of bacterial and fungal biomass to priming. The RR of the bacterial and fungal diversity and fungal biomass was significantly different within the P treatment (Table 2).

Fig. 2: Principal co-ordinates analysis (PCoA) of the response ratio (RR) of bacterial and fungal communities among treatments (LP = leguminous plant, NLP = non-leguminous plant, CK = no nitrogen or phosphorus addition, N = nitrogen addition, P = phosphorus addition, and NP = nitrogen and phosphorus addition).

The RR of bacterial and fungal communities was calculated as the variation of the microbial taxa between glucose treatment and control.

Table 2 Response ratios (RR) of microbial biomass and alpha diversity at the end of the incubation experiment.

Factors involved in SOM priming effects

Soil physicochemical and microbial properties showed different responses to chronic field treatments, where P addition appeared to have a greater effect on these basic soil properties (Table 1). The strength of the PE was significantly positively correlated with consumed N and soil P concentration, as well as bacterial and fungal biomass (Fig. 3A, and Figs. S7,  S8). In addition, the RR of microbial C utilization, nutrient-related gene abundance, and enzyme activities showed different responses to plant identity and nutrient addition (Table S2, and Figs. S5,  S6). For instance, C cycling genes are highly expressed in the NP addition in the NLP stands and in both P and NP additions in the LP stand (Table S2). These results match the cases where significant PEs were observed in our study. Correspondingly, significant positive correlations were found between the PE and the RR of nutrient-related gene expression, enzyme activities, and microbial C utilization (Figs. 3,  S9). Similarly, some microbial taxa also showed convergent patterns of variation with priming, leading to significant correlations between the PE and the RRs of the Proteobacteria, Acidobacteria, and Actinobacteria in our study (Fig. S10). Structural equation models (SEM) suggest that soil N and P concentrations affect PE indirectly by altering the microbial community, predominantly by influencing the proportion of nutrient-related genes, increasing C utilization, and promoting enzyme activity. (Fig. S11, Fig. 4).

Fig. 3: Correlation between priming effect, soil and microbial properties.

A Pearson correlations between the priming effect and microbial community and soil properties. B Correlations of priming effects with the response ratios (RR) of carbon utilization, nutrient-cycling gene expression, and enzyme activities after incubation. The RR of the microbial gene expression, enzyme activities, and microbial carbon utilization were calculated by the ratio between control and glucose treatment. TN total nitrogen, TP total phosphorus, AP available phosphorus, IN inorganic nitrogen in glucose treatments at the beginning and end of the incubations and the amount consumed; MBC microbial biomass carbon, MBN microbial biomass nitrogen.

Fig. 4: Conceptual diagram depicting factors and mechanisms underlying the soil priming effect.

Trapezoid thickness increases as the value of the variable increases.

Discussion

Nutritional balance regulates the priming effects

Our results demonstrate that both dominant plant species identity and long-term nutrient additions can impact soil and microbial properties that control the susceptibility of soil organic matter to priming in subtropical forests. Glucose addition resulted in increased SOM decomposition regardless of dominant overstory plant type or nutrient addition, an effect that can be attributed to “microbial co-metabolism”, where microorganisms in soil are C-starved and a proportion of living microbes remain inactive [57, 58]. The addition of labile C triggers microbial activation, enhancing their capacity to decompose SOM [3, 17, 59]. Our findings demonstrate this effect in that glucose addition increased microbial biomass, activity, and SOC utilization, resulting in increased CO₂ release from SOM and a positive PE.

Considerable work has proven that soil nutrient conditions influence SOM decomposition [60,61,62]. At our study site, long-term P addition showed a stronger effect on the microbial community and priming than N addition. This is unsurprising, as tropical and subtropical forests are often P-limited due to extreme weathering and highly acidic soils that promote cation bridging and the adsorption of P [25, 29]. Thus, low P availability could restrict microbial biomass and activity [63, 64], and P addition should have more profound effects on the microbial community than N addition in these ecosystems. While this pattern generally held in our study, interestingly, we found that P addition alone significantly increased the PE in the LP soil, but not in the NLP soil. This is likely due to the association of N-fixing microbes that can supply additional N to maintain microbial activity [39, 40]. As a consequence, when LP plantations are supplemented by P, they are capable of relying on their belowground mutualists to meet a growing need for N. In NLP plantations, however, this P supplementation may simply cause plants to be N-limited, which may explain why P supplementation was not sufficient to increase the PE.

Our results show that combined N and P addition resulted in positive priming in soils from both forest plantation types. Balanced and sufficient nutrient availability likely stimulates SOM degradation due to the increased potential decomposition capacity of microorganisms [11]. Balanced nutrient supplies lead to a healthier soil microbial community, which in turn promotes the production of hydrolytic enzymes that degrade SOM [11, 65]. This premise is consistent with our understanding of nutrient limitation theory. For optimal metabolism and growth of microorganisms under labile C inputs, N and P availability becomes the limiting factor in the capacity to activate microbial metabolism and growth to take profit of the labile C inputs [26, 27]. This may be particularly true in the tropics, where soil N and P availability can act together to affect SOM dynamics through their effect on soil microorganisms [66]. In this manner, the microbial decomposition of SOM and the ensuing PE follows Sprengel-Liebig’s Law of the Minimum [28]. In this case, decomposition is controlled by levels of essential soil nutrients like N and P, rather than labile C. This premise is consistent with our observations, such as the positive PEs, as well as the increased microbial biomass, extracellular enzyme activity, and the abundance of genes involved in nutrient cycling. This suggests that microbial health and nutrient limitation are crucial components influencing the strength of priming. Correspondingly, we observed that higher levels of soil P and N induced relatively high soil PE, and resulted in a relatively lower SOC than plots without NP addition.

Microbial mechanisms underpinning priming effects

Soil microorganisms play a crucial role in priming and are the main agents of SOM decomposition [67]. Several studies have established that soil priming is impacted by variations in microbial biomass and community structure, suggesting that the abiotic influence of soil nutrient conditions indirectly influences SOM decomposition [18, 68, 69]. Similarly, our study found that plant species and long-term nutrient addition significantly changed microbial community composition, and that these changes in microbial community composition influenced soil priming. For example, the response ratio of microbial biomass was significantly correlated with the measured PE. In addition, we found that soil communities in plantations with artificial nutrient addition responded differently to exogenous glucose supplementation than those in unfertilized plantations. Our results also show a positive correlation between the PE and the response ratio of the abundance of Proteobacteria and Acidobacteria, which are closely associated with litter decomposition and SOM mineralization [19, 70]. Similar results were found with significant correlations between the PE and microbial taxa composition in other recent studies [69, 71].

In our study, glucose addition increased the abundance of nutrient-related cycling genes, and the strength of the PE was positively correlated with the RR of the abundance of genes involved in nutrient cycling. These results suggest that microbial communities may enhance their utilization of soil organic C by altering the abundance of genes involved in SOM decomposition after receiving labile C inputs. High levels of expression in microbial functional genes suggest a commensurately high production of extracellular enzymes necessary for microbial organic C utilization and SOM decomposition [72, 73]. Generally, soil organic matter can be utilized by microorganisms only after it has been converted into smaller substances by enzymatic degradation [69, 74]. We observed that the input of labile C promotes the production of enzymes involved in the mineralization of SOM. This is consistent with our understanding of microbial SOM decomposition that the variation in the abundance of the microbial functional genes can be reflected in specific extracellular enzyme activities, and there is likely an intimate link between the changes in microbial functional gene abundance and enzyme-mediated soil C dynamics [20]. For instance, a large-scale study in Australia has found a significant correlation between the abundance of microbial functional genes and C-cycling-related enzyme activities [22]. These findings are consistent with our observations. We observed that labile C had the strongest influence on soil priming when microbial communities were least limited by other soil nutrients – such as N and P. In these cases, the introduction of glucose (labile C) strongly stimulated microbial biomass, activity, and the relative abundance of genes associated with nutrient cycling. In turn, this enhanced belowground activity facilitates positive feedback in SOM decomposition, which we understand as priming. Our results suggest that this PE was strongest when the microbial community was least limited by nutrient availability – either through natural rhizomatous mutualisms associated with LP plantations or through artificial nutrient addition.

Conclusion

Using a field nutrient-addition experiment together with a lab incubation experiment, we investigated the interactive effect of plant species and long-term N and P additions on SOM priming and its drivers in a subtropical forest. We found that the input of labile C enhanced SOM decomposition and reduced soil C storage potential by increasing soil microbial activity and altering microbial community composition, especially when the soil had balanced and sufficient N and P availability. In addition, soil priming was co-limited by N and P in non-legume soils, whereas priming was limited by P only due to the additional N supplied by N-fixing rhizobia in legume soils. Moreover, our study highlights the role of the interaction between plant species and soil nutrient availability in affecting the abundance of microbial genes and enzyme activities involved in soil nutrient cycling, which in turn drives the microbial utilization of soil organic C and determines the PE in this forest ecosystem. Overall, this study provides novel insights into our understanding of the functional effects of microbial communities on the soil PE and highlights how the interaction between plants and nutrient balance can impact SOM priming.