Control of inorganic and organic phosphorus molecules on microbial activity, and the stoichiometry of nutrient cycling in soils in an arid, agricultural ecosystem

Study site

This study was carried out with samples obtained from an alfalfa crop plot located on the western side of the Cuatro Cienegas Basin (26°58′57″N, 102°5′10″W). The climate at Cuatro Cienegas Basin (CCB) is hot and arid, with an average yearly temperature of 21.9 °C and an average annual precipitation of 253 mm (Montiel-González et al., 2021). The dominant parent material in the west of CCB is calcium carbonate (Mc Kee, Jones & Long, 1990) and the dominant soil groups are Calcisols (García-Oliva et al., 2018), which is the soil group corresponding to the obtained samples.

Management at the farming plots consists of fertilization every 25 days, principally with MAP (monoammonium phosphate) technical grade, NPK 20−20−20 fertilizers, or NPK 11−42−0 fertilizers. Vermicompost leachate is often applied at a dose of 100 L ha⁻¹. Insecticides are used according to requirement and herbicides with the active compound clethodim are used to control grasses.

Soil sampling

Soil sampling from the alfalfa crop plot was conducted in August 2018. We established a 50 × 50 m plot within the alfalfa crop. Soil samples were taken along five transects chosen randomly on one side of 50 m. A subsample was taken each 10 m along each transect, obtaining five subsamples that were mixed homogeneously to produce one composite soil sample per transect. Soil samples were taken from the top 15 cm of the mineral soil with a soil core sampler, placed in black plastic bags, and stored at 4 °C until subsequent laboratory analysis.

Experimental design and incubation

An incubation experiment was conducted using soil amended with different phosphorus compounds (Fig. 1). The experimental design consisted of one factor, with six levels: five phosphorus compounds and one negative control. The added phosphorus compounds were chosen based on the most common organic P compounds found in farming soils, such as phosphate monoesters, phosphate diesters, and phytic acid. We used adenosine monophosphate (AMP) as a phosphate monoester, and torula yeast RNA (Sigma-Aldrich) as a phosphate diester. We also used two inorganic phosphate compounds: monoammonium phosphate (MAP; used at the study site as fertilizer) and monobasic calcium phosphate (Ca(H₂PO₄)₂), known as triple superphosphate and commonly used for fertilizer production. The concentrations of the phosphorus compounds added to the soil were calculated according to a maximum P concentration used as a fertilizer in the sampled site (16.5 kgP ha⁻¹). A concentration of 27.8 µg P g⁻¹ of soil was added, which corresponds to 89.87 µmol P (Table 1), calculated based on the PVC area of 0.0019 m². Phosphorus sources were added to the water used to adjust the soil samples to water holding capacity.

We included five replicates for each treatment, corresponding to each composite soil sample obtained from the field. Soil incubations were carried out over 19 days, (time defined by the obtained C mineralization rate data) at 28 °C. A soil sample of 100 g was added to sterilized PVC tubes with one extreme closed with a mesh (pore size <0.05 mm). Water was added to reach 90% of the field capacity. During the incubation period, soil water was maintained by weight measurement. PVC tubes were placed into 1 L glass flasks and sealed during the incubation, as shown in Fig. 1.

Potential C mineralization

Carbon mineralization was measured periodically during the 19 days of incubation. For this analysis, CO₂ traps were placed inside the glass flasks. These traps consisted of a vial containing 10 ml of NaOH 1N, which was titrated periodically each third day with HCl 1N and BaCl₂ and replaced with fresh NaOH solution. The HCl used for titration was used to calculate C mineralization rates (Coleman et al., 1977). The metabolic quotient for CO₂ (qCO₂) was determined according to Anderson & Domsch (1993), dividing the accumulated CO₂-C by microbial biomass C following incubation (Cmic).

Biogeochemical and enzymatic activity analyses

Before and after incubation, biogeochemical analysis and determination of enzymatic activities were performed. Soil moisture content was determined by gravimetric analysis, drying the samples at 100 °C until reaching constant weight. Active soil pH in deionized water (1:10 w/v) was measured using a digital potentiometer (Thermo Scientific Orion 3star Plus). The weight of the samples for all analyses was corrected with the fraction of dry soil obtained in the moisture content determination.

Total C, N, and P were quantified (TC, TN, and TP) using dry soil ground in an agate mortar. Total C (TC) and total inorganic C (TIC) were determined by coulometric detection (Huffman, 1977) in a total Carbon Analyzer (UIC model CM5012). Total organic C (COT) was calculated by the difference between TC and TIC. TN and TP were determined following acid digestion, in which TN was determined by the Kjeldahl macro method (Bremmer, 1996) and TP by the reduction of molybdate with ascorbic acid (Murphy & Riley, 1962). Both nutrients were measured by colorimetry in a Bran-Lubbe Auto Analyzer 3 (Norderstedt, Germany).

The available forms of nitrogen (NH₄⁺ and NO₃⁻) were extracted from 10 g of fresh soil with 2 M KCl, followed by filtration through a Watman No. 1 paper filter and determined by colorimetry with the phenol-hypochlorite method (Robertson et al., 1999). Available inorganic P (HPO₄²⁻) was extracted from 5 g of fresh soil with 0.5M solution of NaHCO₃, adjusted to pH 8.5 and determined colorimetrically by the molybdate-ascorbic acid method (Murphy & Riley, 1962; Tiessen & Moir, 1993).

The dissolved organic nutrients were determined by the difference between the total dissolved nutrient (C, N, or P) and the dissolved inorganic nutrient. Dissolved organic C, N, and P (DOC, DON, and DOP) were extracted with deionized water (1:4 w/v) according to Jones & Willett (2006) and filtered through a Millipore 0.45 µm filter. Filtrates were used directly to measure inorganic dissolved N and P, and total and inorganic dissolved C. For total dissolved N and P, the filtrate was acid digested. Total and inorganic forms of dissolved N and P were quantified in a Bran-Luebbe Auto analyzer 3 (Norderstedt, Germany). For determination of the DOC, the total dissolved C (TDC) and dissolved inorganic C (DIC) were measured in a Carbon Autoanalyzer (TOC CM 5012).

The amounts of C, N, and P within the microbial biomass (Cmic, Nmic, and Pmic) were obtained by the method of fumigation with chloroform and incubation for 24 h at 27 °C (Vance, Brookes & Jenkinson, 1987). Cmic and Nmic were extracted using 0.5 M K₂SO₄, according to Brookes et al. (1985), and filtered with Whatman No. 42 and No. 1, respectively. Cmic was quantified using a Carbon Auto Analyzer (TOC CM 5012). C concentration was measured from each extract as total carbon (TCmic), using the module for liquids (UIC-COULOMETRICS), and as inorganic carbon (ICmic), using the acidification module CM 5130. For Nmic, the filtrate was acid digested and determined as TN by the Macro-Kjeldahl method (Brookes et al., 1985). The Pmic was extracted according to Cole et al. (1977), using a solution of NaHCO₃ 0.5M and adjusted to pH 8.5, shaken for 16 h, and passed through Whatman No. 42 filters. Filtrates were digested using 11 N H₂SO₄ and a 50% w/v solution of ammonium persulfate and neutralized following the acid digestion. Microbial P was determined colorimetrically by the molybdate-ascorbic acid method using an Evolution 201 Thermo Scientific Inc. spectrophotometer at a wavelength of 880 nm (Murphy & Riley, 1962). Nutrients in microbial biomass were calculated by subtracting non-fumigated sample data from that of fumigated samples and then dividing by the corresponding conversion factor. kEC (0.45) and kEN (0.54), determined by Joergensen (1996) and Joergensen & Mueller (1996), were used to calculate Cmic and Nmic, respectively, and a Kp correction factor of 0.4 (Hedley, Stewart & Chauhan, 1982; Lajtha & Jarrell, 1999) was used for the Pmic calculations.

The differences (Δ) between biogeochemical variables before and after incubation were calculated by subtracting the values at the beginning of the incubation from those at the end. Net nitrification was therefore calculated by subtracting the values of available NO₃⁻ after incubation from those of available NO₃⁻ before incubation.

The enzymatic activities of phosphomonoesterase (Phm), phosphodiesterase (Phd), phytase (Phy) beta-glucosidase (BG), N-acetyl glucosaminidase (NAG), and polyphenol oxidase (POX) were quantified. For these analyses, 2 g of fresh soil and 30 ml of modified universal buffer (MUB) at pH 8 were used for the ecoenzyme extraction. Three replicates and one control (sample with no substrate) were prepared per sample. Three substrate controls (substrate with no sample) were also included per assay, and all were incubated at 30 °C. The tubes were centrifugated after the incubation period and 750 µl of the supernatant was then diluted in 2 ml of deionized water and 75 µl NaOH 1N.

Measurements of the enzymatic activity of Phm, Phd, BG, and NAG are based on the spectrophotometric determination of p-nitrophenol (pNP) released from substrates linked to pNP, per unit of time (µmol pNP [g SDW]⁻¹ h⁻¹; Tabatabai & Bremner, 1969; Verchot & Borelli, 2005; Fioretto et al., 2009) and measured at 410 nm on an Evolution 201 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). The POX activity was determined by oxidation of the substrate 2,2′-azinobis-(-3 ethylbenzothiazoline-6-sulfononic acid) diammonium salt (ABTS), which was measured directly at a wavelength of 460 nm. Phy was quantified according to the method of phosphonatase enzymatic activity measurement described by Tapia-Torres et al. (2016), using phytic acid as a substrate and quantifying the released Pi using the ascorbic acid reduction method (Murphy & Riley, 1962) measured at a wavelength of 882 nm. Phy activity was expressed as micromoles of inorganic P released per gram of soil dry weight per hour (µmol Pi [g SDW]⁻¹ h⁻¹).

Specific enzyme activity (SEA) was calculated to determine how much enzyme is synthesized per concentration of nutrients immobilized in microbial biomass. SEA was calculated according to Waldrop, Balser & Firestone (2000) and Chávez-Vergara et al. (2016): ${\displaystyle \text{SEA}=\text{Enzymatic activity}/\text{Carbon in microbial biomass}}$ where enzymatic activity is expressed in units of µmol g SDW⁻¹h⁻¹ and C in microbial biomass is expressed in units of mg C g SDW⁻¹.

Homeostasis and the threshold element ratio (TER)

With the biogeochemical and enzymatic results obtained from the incubation experiment where different phosphorus compounds were applied to agricultural soil samples from the CCB, a homeostasis analysis was performed, performing simple linear regressions between the natural logarithm of DOC:DOP and the natural logarithm of Cmic:Pmic for C:P, and between the natural logarithm of DOC:DON and the natural logarithm of Cmic:Nmic for C:N. Taking the linear regression, it was assessed whether the slope differed from 0, which would mean a non-homeostatic microbial community. The elemental ratio thresholds (TER) were calculated in relation to the elements C:P (TER) and C:N (TER_C:N) according to Sinsabaugh, Hill & Foolstad Shah (2009), using the following equations: (1)${\displaystyle \text{TERc}:\mathrm{p}=\left(\left(\mathrm{BG}/\left(\mathrm{Phm}+\mathrm{Phd}\right)\right)\mathrm{Bc}:\mathrm{p}\right)/\rho \mathrm{o}}$ (2)${\displaystyle \text{TERC}:\mathrm{N}=\left(\left(\mathrm{BG}/\mathrm{NAG}\right)\mathrm{BC}:\mathrm{N}\right)/\mathrm{no}}$ where TER_C:P is the threshold elemental ratio for elements C and P; BG/(Phm+Phd) is the ratio of enzymatic activity for B-1,4-glucosidase (BG) and the sum of phosphomonoesterase plus phosphodiesterase (Phm+Phd); BC: P is the C:P ratio for microbial biomass (Cmic/Pmic) and ρo is a normalization constant. For elements C and N, the TER_C:N Eq. (2) is the threshold elemental ratio (dimensionless), (BG/(NAG)) is the ratio of enzymatic activity for β-glucosidase (BG) and N-acetyl glucosaminidase (NAG), B_C:N is the C:N ratio for microbial biomass (Cmic:Nmic), and n_o is a normalization constant. The normalization constants are the intercept calculated with a standardized major axis regression type II (SMATR). For the constant ρo, the regression is performed between the natural logarithms of the BG enzyme and the sum of the enzymes Phm and Phd. In contrast, for the constant n_o, the regression is calculated between the natural logarithms of the BG enzyme and the NAG enzyme. Equation (1) is modified from Sinsabaugh, Hill & Foolstad Shah (2009) since only the Phm enzyme is used in the original equation; however, the Phd enzyme has been included because of its importance and high activity in the soils of Cuatro Cienegas Basin (Tapia-Torres et al., 2016). The TER_C:P and TER_C:N results, converted to natural logarithms, were compared with the resource ratios (soil nutrients, DOC:DOP, and DOC:DON) using a Student’s t-test. This can reveal whether the soil microorganisms are limited by energy (carbon) or by nutrients (N or P).

Carbon use efficiency

Carbon use efficiency (CUE), in relation to N and P (CUE_C:N and CUE_C:P), was calculated using the formulas developed in Sinsabaugh et al. (2013) and Sinsabaugh et al. (2016): (3)${\displaystyle CUEc:x=CUEMAX\left(Sc:x/\left(Sc:x+Kx\right)\right)}$ where X represents element N or P; K_x is the mean saturation constant, which has a value of 0.5; CUE_MAX isthe upper limit for the efficiency of microbial growth, which has a value of 0.6 based on thermodynamic constraints; and S_C:X is calculated as follows: (4)${\displaystyle Sc:x=\left(1/\left(EEAc:x\right)\right)\left(Bc:Lc:x\right)}$ where EEA_C:X is the ratio of the enzymatic activities related to the nutrients C:X; L_C:X is the ratio of the substrates consumed, which in this case were the dissolved organic nutrients of the soil, and B_C:X the ratio of elements in microbial biomass. From the data obtained from CUE_C:P and CUE_C:N, the CUE calculation was performed using the formula suggested by Sinsabaugh & Follstad Shah (2012) and Sinsabaugh et al. (2016), as a best estimate for the CUE of the microbial community: (5)${\displaystyle CUE=\sqrt{CUEC:NXCUEC:P}}$

Statistical analysis

A one-way ANOVA was performed to determine the effect of the treatment on C mineralization, and on the biogeochemical and enzymatic variables, as well as on the differences between the beginning and the end of the incubation for enzyme activities and DOC, DON, DOP, NO₃⁻, PO₄⁻, NH₄⁺, Cmic, Pmic, and Nmic. Residual frequency distribution was assessed with a Kruskal-Wallis test to verify a normal distribution (García-Oliva & Maass, 1998). A Tukey HSD test was performed after the ANOVA to identify differences between treatments. An ANOVA was also performed for the results of SEA, and an LSD test was performed after the ANOVA for the SEA obtained with Cmic. A Pearson correlation was performed among post-incubation biogeochemical variables, enzymatic activities (post-incubation), accumulated C mineralization, qCO₂, and nitrification. Principal component analysis (PCA) was conducted to determine which variables explained variance in the results and to visualize the grouping of the different treatments. The data matrix was constructed using all biogeochemical and enzymatic data from all the samples, apart from those of SEA and qCO₂. The analysis was carried out using the function “prcomp” on R software. All statistical analyses were performed using R software (R Core Team, 2020). Given that three separate groups were observed in the PCA, Pearson correlation tests were conducted separately for each group.

One-way ANOVAs were performed to compare the results of TER_C:P, TER_C:N (using the natural logarithm of TER), CUE_C:P, CUE_C:N, and CUE between treatments. Residual frequency distribution was assessed with a Kruskal–Wallis test to verify a normal distribution (García-Oliva & Maass, 1998). The Tukey HSD test was performed to identify the treatments with significant differences, except for the analysis conducted for the CUE in which no results were obtained with the Tukey HSD test, and an LSD analysis was performed. Student’s t-tests were performed to identify differences between the TER values and the ratios between the dissolved organic nutrients. For the TER calculation, type II linear regressions were performed among the enzyme activities using the SMATR package. All statistical analyses were performed using R software (R Core Team, 2020).

Incubation experiment and metabolic quotient for CO₂ (qCO₂)

After 19 days of incubation, adenosine monophosphate (AMP) and RNA additions of P organic treatments presented the highest C mineralization (950 and 863 µg CO₂-C g⁻¹, respectively), while the Ca(H₂PO₄)₂ addition and the control treatments had the lowest C mineralization values (781 and 739 µg CO₂-C g⁻¹, respectively; Table 2). Cmic was lowest for the phytic acid treatment and therefore the calculated qCO₂ was higher for this phosphate ester treatment (1.9 ± 0.56) compared to the control (0.59 ± 0.03), suggesting a lower metabolic efficiency of the soil microbial community fertilized with phytate (Table 2). Cmic values in MAP, Ca(H₂PO₄)2, RNA and AMP were lower after the incubation compared to the Cmic values obtained before incubation (Table 3).

Post-incubation biogeochemical analysis: changes in organic and inorganic C, N, and P pools and microbial P immobilization

AMP and RNA additions in P organic treatments produced higher NO₃⁻ concentrations than the other treatments, as well as nitrification (Table 2). In contrast, NH₄⁺ and HPO₄²⁻ presented no significant differences between treatments (Table 2). Dissolved organic C (DOC) was significantly greater for the treatment with Ca(H₂PO₄)₂ than for those with RNA and AMP (Table 2). Moreover, the Ca(H₂PO₄)₂ andRNA treatments had higher dissolved organic N (DON) concentrations than the other treatments (Table 2).

Both organic (AMP and RNA) and inorganic Ca(H₂PO₄)₂ treatments increased dissolved organic P (DOP) (Table 2). Therefore, the control samples had higher DOC:DON and DOC:DOP ratios than the AMP and RNA treatments (Table 4). In addition, the control had a higher DON:DOP ratio than the monoammonium phosphate (MAP), AMP, RNA, and phytic acid treatments (Table 4). The control, MAP, and Ca(H₂PO₄)₂ treatments presented higher Cmic concentrations than the AMP, RNA, and phytic acid treatments (Table 2). In contrast, treatments with RNA, AMP, and Ca(H₂PO₄)₂ immobilized significantly more P compared to the control treatment (Table 3). The organic treatments favored N immobilization in microbial biomass given that the Cmic:Nmic ratio was lower in these P organic treatments (RNA, AMP, and phytic acid) than in the control treatment (Table 4). These results suggest that the organic P treatments favored P and N immobilization in microbial biomass (Pmic) and high dissolved organic P (DOP) as well as higher available nitrate.

Enzyme and specific enzyme activity

Enzyme activity was only significantly higher for NAG, where in the Ca(H₂PO₄)₂ treatment (Table S1). The specific enzyme activity obtained by enzyme activity normalization using Cmic differed significantly among the enzymes POX, NAG, and Phd (Table 5).

The RNA and phytic acid treatments had higher POX SEA than the inorganic P treatments (MAP and Ca(H₂PO₄)₂). Organic P treatments also produced higher Phd SEA than the inorganic P treatments and the control (Table 5) suggesting that the microbes used phosphodiesterase to obtain P from these substrates.

The phytic acid-treated samples had the highest NAG SEA values while the RNA, AMP, and control treatments had the lowest (Table 5). N-acetyl glucosaminidase is one of three enzymes that catalyze the hydrolysis of chitin, which is important in carbon (C) and nitrogen (N) cycling in soils. It participates in chitin conversion to amino sugars, which are major sources of mineralizable N in soils.

Increases in DOC, DOP, and Pmic after the incubation experiment

The Ca(H₂PO₄)₂ and the AMP treatments presented the highest and lowest increases (Δ) in DOC concentration after incubation, respectively (Table 6). Similarly, the Ca(H₂PO₄)₂ treatment had the highest DON increase, but the lowest increase was in the MAP treatment. In contrast, the Ca(H₂PO₄)₂, RNA, and AMP treatments had higher increases in DOP than the control, which had negative values (Table 6).

Among microbial nutrients, only Pmic presented a significant increase after incubation. Among treatments, the RNA and control presented the highest and lowest values, respectively (Table 6).

A complex dynamic observed from the application of inorganic and organic P fertilization

In the PCA, the first and second components explained 26% (eigenvalue = 3.69) and 17% (eigenvalue = 2.32) of variance, respectively (Table S1). NO₃⁻ and the NAG enzyme were the variables with greater weight in the first component, while HPO₄²⁻ and the POX enzyme better explained the variance of the second component (Fig. 2). The treatments clustered into three groups: only control samples on the left side of the first component and negative values of the second component; the Ca(H₂PO₄)_2, MAP and phytic acid-treated samples in the middle of the figure; and the samples with AMP and RNA organic treatments on the right side of the first component (Fig. 2).

Pearsons correlation tests, for the control samples, indicated that the microbial community requires more energy to acquire phosphorus than the samples fertilized with other treatments, as shown by the positive correlation between microbial P and the enzymes BG (r = 0.88, p = 0.046), POX (r = 0.89, p = 0.044), and Phy (r = 0.97, p = 0.007; Fig. 2A). These correlations were not observed in the other groups of treatments (the MAP, Ca(H₂PO₄)₂, the phytic acid group, and the AMP and RNA group) (Figs. 3B, 3C). However, POX activity was correlated with Phm activity in the AMP and RNA group (r = 0.89, p = 0.004, Fig. 3C).

In correlations of the cluster of treatments with Ca(H₂PO₄)_2, MAP, and phytic acid, the samples showed greater microbial growth when they were able to immobilize more phosphorus, as indicated by the positive correlation of Cmic with Pmic (r = 0.58, p = 0.037; Fig. 3B). Pmic also correlated negatively with qCO₂ (r = −0.58, p = 0.046; Fig. 3B), which is an indicator of the lower metabolic efficiency of microorganisms when there is insufficient phosphorus within their biomass. However, for this cluster of treatments, Phd, a phosphorus-acquiring enzyme, correlated negatively with Cmic (r = −0.61, p = 0.017; Fig. 3B).

Finally, in the third group, negative correlations between qCO₂ and Pmic were also significant for the AMP and RNA group samples (r = −0.63, p = 0.041 Fig. 3C), and Nmic and qCO₂ presented the same correlation for these treatments (r = −0.72, p = 0.017). For the same treatment group, a negative correlation was found between DOC and NO₃⁻ (r =  − 0.85, p = 0.0035), and a positive correlation between NO₃⁻ and CO₂-C (r = 0.8, p = 0.04). The latter also correlated negatively with the NAG enzyme (r =  − 0.61, p = 0.04) and with DOC (r =  − 0.73, p = 0.018; Fig. 3C).

Homeostasis, threshold element ratio, and carbon use efficiency

In most of the treatments, the microbial community is a homeostatic community estimated by a standardized linear regression. A slope that did not differ from 0 (p > 0.05) according to the standardized linear regression performed for the control treatments, monobasic ammonium phosphate (MAP), calcium phosphate (Ca(H₂PO₄)₂), RNA, and adenosine monophosphate (AMP) treatments (Figs. 4 and 5) suggest homeostasis. In contrast, the samples treated with phytic acid (phytate) as a source of P comprise a non-homeostatic community, given that in the regressions performed, it exhibited a slope that differed from zero (Figs. 4E and 5E). The microbial community of these soil samples tends to decrease the C:P and C:N ratio of its microbial biomass (immobilization of nutrients) while increasing the C:P and C:N ratio of the resource.

TER_C:P analysis showed significant differences between treatments. TER_C:P was higher for the samples with the control and MAP treatments, followed by the treatments with Ca(H₂PO₄)₂ and RNA, while the TER_C:P was lower for samples with the AMP and phytic acid treatments (Fig. 6). Compared with the dissolved nutrient ratios (DOC:DOP), the TER_C:Pwas found to be lower than this ratio for the control treatments, Ca(H₂PO₄)₂, AMP, RNA, and phytic acid, but the same for the treatment with MAP.

TER_C:N was higher for the control, followed in equal measure by the samples treated with AMP, Ca(H₂PO₄)₂, MAP, and phytic acid, but lower for the samples treated with RNA (Fig. 7). Compared to the dissolved nutrient ratios, the TER_C:N was lower than the DOC:DON ratio for MAP- and RNA-treated soil, while it was higher for the control treatment. For the other treatments (AMP, Ca(H₂PO₄)₂, and phytate), the TER_C:N was the same as the DOC:DON ratio.

Carbon use efficiency, in relation to phosphorus (CUE_C:P), did not vary significantly between all treatments (Fig. 8A). However, the value of carbon use efficiency in relation to nitrogen (CUE_C:N), differ between treatments (Fig. 8B). CUE_C:N was higher for the samples treated with Ca(H₂PO₄)₂, intermediate for the samples treated with MAP and phytate, and lower for the samples treated with AMP and RNA, and for the control samples (Fig. 8B). The total CUE (Fig. 9) showed a similar trend to CUE_C:N.

Phosphorus sources effect on soil C and N dynamics

All the evaluated phosphorus sources stimulated microbial C mineralization. This suggest that P limits the activity and growth of microbial communities in the selected soil, as was reported in previous studies for the study site (Perroni et al., 2014; Tapia-Torres et al., 2015a). As hypothesized, labile organic treatments, such as adenosine monophosphate (AMP) and RNA, promoted microbial C mineralization. It has been previously reported that soil bacteria from CCB prefer DNA as a phosphorus substrate over inorganic phosphorus, such as potassium phosphate and calcium phosphate when isolates are grown in culture media (Tapia-Torres et al., 2016). The degradation of both DNA and RNA requires phosphodiesterase enzymes, while the substrate AMP can be seen as a monomer from the decomposition of nucleic acids and requires phosphomonoesterase (Lehninger, Nelson & Cox, 2005). Both treatments contain not only P but also C and N, which suggests phosphorus colimitations with C and N in the soil, and microbial activity is promoted when these nutrients are added. However, the other organic treatment, phytic acid, did not have the same expected effect.

The microbial community and nutrient dynamic response to phytic acid was similar to that for inorganic P substrates, as shown by the principal component analysis and with the accumulated C mineralization results. Moreover, the phytic acid treatment had the highest metabolic quotient (qCO₂) value, suggesting that the microbial community is undergoing metabolic stress (Anderson & Domsch, 1993) or is a microbial community with high energy requirements (Carpenter-Boggs, Kennedy & Reganold, 2010). The metabolic quotient (qCO₂) also showed a negative correlation with Pmic in the principal component group of inorganic treatments and the AMP and RNA group, suggesting that metabolic stress has an inverse relationship with the amount of P immobilized in microbial biomass. These results suggest that phytic acid is not a readily available source of P and C for soil microorganisms. Higher energy requirements may be due to phytic acid interactions with soil since it is strongly bound to soil clay and soil organic matter and can react with soil minerals, such as calcium, favoring precipitation and adsorption reactions (Dalai, 1977; Stewart & Tiessen, 1987; McKercher & Anderson, 1989; Wan et al., 2016) and becoming less susceptible to microbial attack. This is important in the CCB soils given their high sorption capacity, which is greater where higher concentrations of organic compounds are found (Perroni et al., 2014). These tend to be higher in agricultural fields than in natural soils, because the continuous water and nutrient inputs increase total organic carbon and dissolved organic phosphorus in soils, compared to that of native grasslands (Hernández-Becerra et al., 2016). As a consequence, phosphorus acquisition from phytic acid molecules is a two-step process, which demands more energy from soil microorganisms. First, insoluble and mineral-bound phytate compounds must be solubilized by bacteria or fungi capable of synthesizing organic acids and chelates (Hill & Richardson, 2007). Free and soluble phytic acid can then be hydrolyzed by phytases (Lim et al., 2007); specifically, B-propeller phytase, the active phytase type in neutral and alkaline soils, which breaks down each bound monoester to release inorganic phosphate (Gontia-Mishra & Tiwari, 2013; Cotta et al., 2016). Only after the complete dephosphorylation of the molecule is phytic acid transformed into myo-inositol, which can then be used as a carbon source by soil microbes (Cosgrove, Irving & Bromfield, 1970) as it has been reported for some bacteria (Chen et al., 2020; Rothhardt et al., 2014; Yuan et al., 2019). These results suggest that the molecular structure plays an important role in its decomposition, rather than simply the concentration of C, N, or P.

Addition of labile organic molecules (AMP and RNA) can also affect soil N dynamics, promoting nitrification rates and thus increasing the susceptibility to soil N losses (Tapia-Torres et al., 2015b). Two processes can explain this result. The first addition of labile organic molecules could prime the microbial community (Garcia et al., 2017; Scotti et al., 2015a; Scotti et al., 2015b). In our case, the organic treatments AMP and RNA acted as these OM inputs and served as an initial energy source for microorganisms capable of mineralizing soil OM. The measured DOC concentration was therefore lower in the AMP and RNA treatments at the end of the experiment, probably because of the high rate of depletion of C sources, as well as Cmic for both treatments, while the Ca(H₂PO₄)₂ treatment contained the highest concentration of DOC at the end of the experiment. These findings suggest that lower DOC results from higher C mineralization rates since it is negatively correlated, producing the depletion of labile organic matter in the soil. As a consequence, the carbon use efficiency, in relation to nitrogen (CUE_C:N), was lower in the AMP and RNA treatments than with monobasic calcium phosphate (Ca(H₂PO₄)₂). These CUE results can be explained by the biogeochemical analysis performed at the end of the incubation period when the DOC was consumed by the microbial community. CUE_C:N is expected to decrease when C is a limiting resource and the remaining organic matter for decomposition has higher recalcitrance (Sinsabaugh et al., 2013). Second, a decrease in DOC at the end of the incubation are related to an increase of the activity of nitrifier. Our results showed that the DOC concentration correlated inversely with nitrate in both of these treatments, which suggests enhanced nitrifier activity. These chemoautotrophs obtain energy from oxidizing NH₄ to NO₂⁻ and NO₂⁻ to NO₃⁻ (Fenchel et al., 2012). Moreover, these bacteria present optimum activity in a neutral to alkaline pH (Prosser, 1990) and nitrification rates increase with higher pH (Li et al., 2020), which is coincident with the soil studied herein. A decrease of COD at the end of the incubation in these soils could make these bacteria increase and be competitive with the heterotrophic bacteria. In a pulse of carbon, such as that created by the application of organic treatments AMP and RNA, rapidly growing heterotrophic (r strategist) microbes immobilize nutrients and grow faster, which may occur during the first days of the incubation. However, enhanced growth of these organisms may induce a rapid depletion of labile carbon sources, giving place to a reduction of r strategist bacteria, and an increase in k strategist and chemoautotrophic bacteria (Montaño Arias & Sánchez-Yañez, 2014), which also explains the reductions in microbial C. A decrease of NOD in the AMP treatment while NO₃⁻ increases is an indicator that heterotrophic bacteria are mineralizing organic matter containing N, and yielding NH₄⁻ as a result of organic C limitation (Chapin III, Matson & Vitousek, 2011). The NH₄⁺ is then rapidly used as a substrate for the nitrifiers.

These two processes suggest that, while organic labile substrates such as ARN and AMP may favor microbial respiration, it could be important to consider a constant supply of organic amendments in agricultural practices to avoid soil N losses.

Effect of phosphorus addition on P availability

In this study, we hypothesized that AMP and RNA treatments would promote the availability of soil nutrients, particularly phosphorus. However, we did not find an increase in concentrations of available HPO₄⁻³ at the end of incubation, but instead found higher DOP and Pmic concentrations in both labile organic treatments (RNA and AMP). Increases in Pmic are crucial because the microbial community is retaining labile forms of P in actively cycling biological pools, and reducing the rate at which labile inorganic P would otherwise be permanently lost via adsorption onto soil particles or leaching (Cleveland, Townsend & Schmidt, 2002). On the other hand, organic phosphorus compounds are an essential fraction of the total P in the soil since, in the CCB grasslands, they can represent about 50% of the total P (Perroni et al., 2014), and dissolved organic phosphorus is composed principally of products of microbial metabolism (Cleveland, Townsend & Schmidt, 2002).

Besides the changes in organic and microbial P pools, the specific enzyme activity (SEA) of the phosphorus enzymes differed among treatments. Organic treatments, whether AMP, RNA, or phytic acid, stimulated phosphodiesterase activity per unit of microbial biomass, as shown with the SEA of Phd, whereas the phosphomonoesterase enzyme was unaffected. Phosphomonoesterases and phosphodiesterases are parts of the phosphate regulon (Pho) in bacteria, which is responsible for phosphorus uptake and responds to P starvation (Santos-Beneit, 2015). Lower concentration of inorganic phosphate but higher availability of organic P in the organic treatments at the beginning of the experiment may have enabled the production of the Phd enzyme. Extracellular enzymes can persist in soil, associated with clay and organic matter particles, and remain active (Nannipieri et al., 2011). It is therefore possible that Phd could have persisted until the end of the experiment.

Nucleic acids, such as RNA and DNA, are released by dead cells in the environment and constitute an important labile source of nutrients such as C, N, and P (Tani & Nasu, 2010), particularly for bacteria from oligotrophic environments (Tapia-Torres et al., 2016). In CCB soils Phd activity tends to be higher than Phm activity (Tapia-Torres et al., 2016; Montiel-González et al., 2017), demonstrating that phosphodiester uptake plays a prominent role in phosphorus cycling in these soils (Tapia-Torres et al., 2016). These studies agree with Turner & Haygarth (2005), who determined, in pasture soils, that phosphodiesterase activity is the rate-limiting step that regulates P turnover because P availability depends on the degradation of fresh organic materials, which are abundant in phospholipids and nucleic acids, cellular components that are sources of phosphate diesters.

Phosphorus turnover is highly important in agricultural systems because inorganic phosphorus tends to be lost or become unavailable to crops due to lixiviation or occlusion processes. Although inorganic P is the immediate source of P for vegetation, it is necessary to promote an increase in labile organic P molecules and microbial P pools to prevent these losses, and an increase in the enzymes that hydrolyze organic P compounds, such as phosphomonoesterases, phosphodiesterases, phytases, and phosphonatases, to allow a slow but constant release of inorganic P.

Effect of phosphorus addition on C, N, and P stoichiometry

In most treatments, the microbial community was homeostatic, i.e., the C:N:P ratios in the microbial biomass remained constant despite changes in these ratios in the resources (Elser & Sterner, 2002). Nevertheless, the microbial community in the phytic acid treatment was non-homeostatic. A common premise used in ecological stoichiometry studies is that heterotrophic organisms are strictly homeostatic, while autotrophs can present a changing stoichiometry (Persson et al., 2010; Fanin et al., 2013), although there are some scenarios in which members of a microbial community can change their stoichiometry according to that of their resource, thus becoming non-homeostatic. Non-homeostatic behavior is a mechanism by which to reduce stochiometric imbalances between the resources and microbial biomass (Mooshammer et al., 2014) because it can occur through the microbial storage of nutrients in excess or by shifts in microbial community structure and therefore shifts in the biomass stoichiometry (Mooshammer et al., 2014). We reported a lower value of the Cmic:Pmic ratio compared to that of the control, suggesting greater P immobilization with P addition; however, this difference was present in all phosphorus treatments, not just that of phytic acid. Fanin et al. (2013) suggested that non-homeostatic behaviors are the result of changes in microbial community composition rather than shifts in the microbial biomass of individual microorganisms since they found that the bacteria:fungi and gram positive:gram negative ratios change along with changes in homeostasis. For example, the reported fungal C:N:P ratio is 250:16:1 (Zhang & Elser, 2017), while the bacterial C:N:P ratio is 46:7:1 (Cleveland & Liptzin, 2007). In the phytic acid treatment, the average C:N:P ratio was 58:5:1 and thus closer to the bacterial biomass ratio, or the average soil microbial biomass ratio (60:7:1) suggested by Cleveland & Liptzin (2007). However, in the pre-incubation samples, the microbial biomass stoichiometry was closer to the fungal biomass stoichiometry (246:16:1, Table 1), as was the case in the control samples (310:10:1; Table 4). This suggests that the homeostasis imbalances are due to the microbial community changing to different microbial groups with the addition of fertilizers. Regarding the phytic acid treatment, bacteria are the main producers of B-propeller phytase, the active phytase type in neutral and alkaline soils, while fungi are the producers of acid phytases (Jain, Sapna & Singh, 2016).

The threshold element ratio (TER) is the elemental proportion that corresponds to balanced microbial growth, with no limitation by C or nutrients (Sinsabaugh et al., 2016). It represents the critical ratio at which organisms transition from net nutrient immobilization to net nutrient mineralization (Mooshammer et al., 2014) and it defines whether the community is limited by nutrients (N or P) or by energy (C). When resource C:N or C:P ratios are greater than the TER, the system is limited by nutrients and immobilization processes dominate. However, when these resource ratios are lower than the TER, then the system is limited by energy, and nutrient mineralization occurs (Sinsabaugh et al., 2013). In this study we selected the treatment MAP because it was used as a fertilizer in the agricultural plots from which the soil was obtained. This treatment did not show differences between the DOC:DOP ratio and TER and it can therefore be considered that the soil microbial community is co-limited by phosphorus and energy (Sterner & Elser, 2002). In contrast, for the Ca(H₂PO₄)₂, AMP, RNA, and phytic acid treatments, there is a limitation by P for the soil microbial community because TER was lower than the soil DOC:DOP ratios. In this case, the microbial community is inclined to immobilize available phosphorus. This coincides with previous studies conducted with non-managed soils from CCB, at the eastern side of the valley (Pozas Azules), where low concentrations of DOC, a trend for phosphorus limitation, and low values of TER_C:Pwere found (Tapia-Torres et al., 2015a).

Regarding nitrogen, the TER_C:N was lower than the DOC:DON ratio for MAP and RNA-treated soil, suggesting that the microbial community is limited by N, which indicates a tendency to immobilize N, although this is not shown in the microbial biomass. It was also found that, in a CCB site from the western side of the valley (Churince) with higher DOC values as well as in our MAP treatment, there was a limitation by N (Tapia-Torres et al., 2015a). On the other hand, the TER_C:N was higher for the control treatment, implying limitation by C or energy and a preference for mineralizing organic nitrogen compounds to obtain C and release NH₄⁺ while immobilizing more C and reducing its losses through mineralization. The CUE_C:N discussed previously (‘Phosphorus sources effect on soil C and N dynamics’) reflected this carbon limitation since the AMP and RNA treatments had the lowest CUE. These results all suggest that the addition of labile organic molecules with P (MAP and RNA) acts to increase microbial N limitation, probably through the increased demand for N by the growing microbial community brought about by the priming effect discussed previously (‘Phosphorus sources effect on soil C and N dynamics’).

The results of this study show that the soil microbial community responds differently to different phosphorous molecules. These effects show differences both between organic and inorganic molecules and among the same groups of molecules with different chemical compositions. This can have implications when conducting fertilization with organic matter in field crops since the chemical structures of the molecules that make up composts and manures are usually unknown. Although the most labile organic compounds (AMP and RNA) favored C mineralization, they also showed a rapid decrease in DOC, implying a reduction in microbial biomass and an increase in chemoautotrophic microorganisms such as nitrifying bacteria, indicating that when fertilizing with labile organic sources, the periodicity of application must be carefully considered to avoid soil N losses.