Soil Phosphorus Partition and Transformations Under Diverse Land Uses

ABSTRACT

This study investigates soil phosphorus (P) partitioning and transformations through chemical fractionation and modeling across various land uses to elucidate their impact on P dynamics and the potential for eutrophication. Geo-referenced soils (0–30 cm at various depth intervals) were collected from diverse land uses, including corn, soybean, wheat stubble, modified relay intercropping, corn-soybean with rye under conventional tillage and no-till systems, and sheep pastures, alongside an adjacent forest (control). Processed soils were sequentially extracted to quantify soluble reactive (SRP), exchangeable (EP), Ca and Mg bound (CaMgP), Fe and Al bound (FeAlP), particulate organic (POP), residual (RP), and total P pools. Results showed significant effects of land use and soil depth on extracted P pools. Notably, corn-soybean with rye exhibited a higher SRP accumulation (2.8 mg/kg) compared to the forest. The labile (SRP + EP), moderately labile (FeAlP + CaMgP + POP), and non-labile P (RP) pools accounted for 1 to 2, 23 to 29, and 65 to 76% of total P, respectively. The structural equation model elucidated soil P transformation, indicating the contribution of labile, moderately labile, and non-labile P pools to SRP accumulation in various land uses. Our findings underscore the critical link between land use, soil P dynamics, and eutrophication risk, offering valuable insights for developing agricultural management practices aimed at mitigating edge-of-field P loss and protecting water quality against eutrophication.

KEYWORDS:

• Cover crop
• forest
• labile P
• phosphorous fractionation
• residual P
• structural equation modeling

Introduction

Phosphorus (P) is a crucial nutrient indispensable in agriculture due to its essential role in plant growth and productivity (White and Hammond 2008; Malhotra et al. 2018). Despite its abundance in the Earth’s crust, P availability to plants is often limited, leading to its excessive usage of P as chemical fertilizers in agriculture. However, the P distribution and its availability to plants are influenced significantly by soil depth and physicochemical properties (Niederberger, Kohler, and Bauhus 2019; Rahman et al. 2021). This variability necessitates a nuanced understanding of P dynamics to optimize its use in agriculture while mitigating environmental footprints. The transformation processes of P, crucial for its availability and environmental mobility, are shaped by complex interactions within the soil matrix. These processes are notably influenced by the specific land use and management practices (Shen et al. 2011). Understanding these transformations is vital for identifying effective P management strategies, ensuring sustainable crop production, and minimizing the risk of edge-of-field loss of P that contributes to the eutrophication of water bodies (Kleinman et al. 2011).

Soils under various land uses are routinely fertilized and/or amended with P fertilizers, rock phosphates, compost, or animal manures, municipal and industrial by-products (Jamal et al. 2023; Silva et al. 2023; Wierzbowska et al. 2020; Yadav et al. 2017). Yet, the P-use efficiency of these inputs by crops is low, ranging between 10 to 25%, depending on crop types (Hopkins and Hansen 2019; Roberts and Johnston 2015). Consequently, a significant portion of applied P is partitioned into different solubility pools, making it susceptible to edge-of-field loss. Soil P distribution among these pools is influenced by soil biological, chemical, and physical properties and associated reactions, which are in turn affected by agricultural management practices (Tiessen and Moir 1993).

Recent emphasis on recycling crop stubbles and introducing cover crops has shown potential for improving soil ecological functions and enhancing P availability (Huang et al. 2011; Islam & Sherman 2021; Ocio, Brookes, and Jenkinson 1991). These practices contribute to increased soil organic matter (SOM) and reduce the demand for P fertilization (Chan and Heenan 2005; Solangheisi et al. 2020). However, managing P in agroecosystems presents the challenge of optimizing P availability for crop productivity while minimizing its environmental footprint, particularly in light of the global issue of eutrophication. Despite these insights, there remains a gap in our understanding of how soil P fractionation and transformations are influenced by various land uses, and its comparison with natural forest. Previous findings suggest variations in soil P availability due to differences in agricultural practices such as corn-soybean with and without cover crops, organic amendments such as chicken manuring and dairy manuring, and tillage, no-till yet the overarching effect on soil P dynamics across diverse land uses requires further investigation (Rahman et al. 2021).

Investigations into soil P pools under specific agricultural management systems, such as long-term corn-soybean with tillage and P fertilization, have reported changes in labile and extractable P contents, highlighting the nuanced impact of various management practices on soil P dynamics (Shi et al. 2015). Furthermore, organic amendments with poultry litter, pig slurry, compost, and chemical fertilization in integrated crop-livestock systems have been shown to affect the distribution of labile and moderately labile P pools, as well as non-labile P contents, underscoring the complex interactions between management practices and soil P dynamics (Messiga et al. 2011; Rigo et al. 2019). The variability in P pools and their availability as influenced by management practices, soil properties, cropping systems, and environmental conditions, underscores the need for a comprehensive understanding of how agricultural management practices impact soil P transformation and partition. This understanding is crucial for developing management strategies that optimize P use efficiency of crops and minimize environmental risks.

Path analysis is a crucial multivariate technique that has been used to elucidate the understanding of the complex interactions and causal relationships between soil properties and management practices (Bayuelo‐Jiménez et al. 2020; Gama-Rodrigues et al. 2014; Hou et al. 2016; Tiecher et al. 2018; Wei et al. 2018). This statistical approach is particularly effective in evaluating the transformation of soil P pools as influenced by various agricultural management strategies. The transformation of soil P pools, including labile, moderately labile, and non-labile forms, is pivotal in determining the availability of P for plant uptake, its retention in the soil, and the risk of environmental loss. The hypothesis driving this research posits that multivariate path analysis of sequentially extracted P pools from soils under different land uses can reveal the underlying mechanisms of P transformation. Such an analytical approach allows for the dissection of direct and indirect effects of various land uses on soil P dynamics, offering insights into how these practices influence P availability to crops and susceptibility to environmental risk.

The objectives of this study are twofold. Firstly, we aim to sequentially fractionate and assess the distribution of phosphorus (P) at different soil depths, thereby providing insights into P availability and forms within the soil plow-depth. Secondly, we intend to utilize path analysis to elucidate the processes of soil P transformation in response to various land uses under similar soil and climatic conditions. By achieving these objectives, our research seeks to contribute to the optimization of soil P management across different agricultural systems, enhancing the efficiency of P use by crops, and minimizing environmental footprints.

Materials and methods

Site description

Geo-referenced composite samples of Blount silty clay loam (Fine, illitic, mesic Aeric Ochraqualfs) were collected from different land uses in the Northern region of Ohio (Crawford County), USA (USDA-SCS 1979). Blount soils are predominant soils in Illinois, Indiana, Michigan, Ohio, and Wisconsin, which are characterized as deep and poorly drained, with a low shrink-swell potential. These soils were formed in silty clay loam or clay loam till of Wisconsinan age at elevations ranging from 183 to 457 m above mean sea level, with slopes varying from 0 to 6% (USDA-SCS 1979). In the study area, most of the Blount soils are utilized for cultivation, with crops such as corn, soybeans, small grains, and meadows being grown. The native vegetation in the region consists of hardwood forests dominated by White oak (Quercus alba) and Northern red oak (Quercus rubra).

The study area experiences an annual precipitation ranging from 740 to 1070 mm, with a mean of 890 mm. The highest precipitation occurs in July, with an average of 103 mm, while the lowest is recorded in February, with an average of 47 mm. The mean annual temperature is approximately 10.6°C, reaching its peak in July at 28.6°C and hitting the lowest point in January at 1.1°C. The region receives an average annual snowfall of 850 mm, and the frost-free period typically lasts for 130–182 days.

Soil sampling and processing

Soil sampling was performed in various managed land uses, including conventionally-tilled and no-till corn-soybean with and without rye used as a cover crop, corn, soybean, and wheat stubble cropping, modified relay intercropping, and sheep pasture cropping. As a control, soil samples were also collected from the natural woodlot (forest). In the modified relay intercropping system, both winter wheat and soybeans were planted together. Subsequently, the winter wheat was harvested while the soybeans continued to grow. Once the soybeans reached maturity, they were then harvested.

At each geo-referenced replicated sampling fields under each land use, four microplots (5 m × 5 m) were selected for soil collection, following a systematic sampling technique. Six cores (2.54 cm internal diameter) were extracted using the JMC® stainless-steel soil environmental probe lined with a plastic tube immediately after crop harvest in October within each microplot, at depths ranging from 0 to 30 cm, and capped at both ends. A total of 420 cores (7 land uses x 3 replicated fields x 4 microplots x 5 soil cores) were collected. The plastic tubes containing cores were segmented into 0–2.5, 2.5–7.5, 7.5–15, 15–22.5, and 22.5–30 cm depth intervals and composited into 105 soil samples. The composited soils were placed in sealable plastic bags for temporary storage at 4°C until processed. A portion of the field-moist soil was air-dried for approximately 15 days under shade at room temperature (~25 °C), processed to pass through a 2-mm sieve, and analyzed for sequential P fractionation and selected chemical and physical properties.

Soil phosphorus fractionation and analysis

Air-dried processed soils were subjected to a sequential P fractionation procedure with various extracting solutions to separate different P pools, including water-soluble reactive P (SRP), salt-extractable P (EP), FeAlP, CaMgP, particulate organic matter P (POP), residual P (RP), and total P (Rahman et al. 2021). In this process, a 5-g soil sample was placed in a plastic tube and shaken with 20 mL of distilled deionized water for 2 hr. The mixture was then centrifuged at 2000 × g for 20 min, and the clear aliquot obtained from filtration was used for SRP analysis. Next, the residual soil was shaken with a 2 M KCl solution for 2 hr., followed by another round of centrifugation and filtration to obtain a clear aliquot for EP analysis. Subsequently, a 0.1 M dilute solution of NaOH was added to the residual soil and shaken for 6 hr. After centrifugation and filtration, the clear extract was collected for FeAlP analysis. For CaMgP analysis, the residual soil was shaken with 0.1 M HCl for 24 hr., then centrifuged and filtered to obtain the extract. Finally, a 1.0 g sample of the residual soil (oven-dried equivalent) was placed in a Folin-Wu glass tube and digested with a concentrated HCl: HNO₃ mixture (2:5) on a hot plate at 125°C for 60 min. After digestion, the digestate was diluted, centrifuged, and filtered to obtain the aliquot for RP analysis.

The particulate organic matter (POM) collected on filter paper after each extraction was combined, washed with distilled deionized water, and then dried in an oven at 55°C. After drying, the POM was ashed at 480 ± 5°C. The resulting ash was dissolved in a 0.1 M HCl solution, filtered, and then analyzed as particulate organic P (POP). The analysis of all P fractions was conducted using the Astoria® 310 continuous flow thru autoanalyzer.

Analysis of selected soil properties

Soil pH was measured using the glass electrode method in a 1:2 soil and deionized water suspension (Rahman et al. 2021). The electrical conductivity (ECe) was determined in a 1:1 soil and deionized water paste using an electrical conductivity probe (Rahman et al. 2021). Total soil organic C (SOC) and N (TN) were analyzed on finely ground (<125 µm) air-dried soil using the automatic dry combustion method with a FlashEA-1112 series CNHS-O analyzer®. Soil bulk density was determined using the standard core method (Blake and Hartge 1986).

Principal components analysis and structural equation modeling

Principal Components Analysis (PCA) was conducted using OriginPro® 2022 to explore the relationship between SRP and other P pools in the soil.

Structural equation modeling (SEM), implemented using AMOS® v. 25 (IBM; SPSS Inc., Chicago, IL, USA), is a statistical multivariate model commonly known as path analysis, which enables the estimation of causal and direct or indirect relationships among multiple variables (Quinn and Keough 2002). SEM was employed to investigate ecosystems’ structure and function (Sutton-Grier, Kenney, and Richardson 2010), which is closely related to various types of statistical analyses, including regression, principal components analysis, and path analysis (Gama-Rodrigues et al. 2014). SEM allows for the examination of relationships among observed variables, latent variables, and residuals, providing a quantitative understanding of how independent variables influence dependent variables, including direct, indirect, and total impacts. The Maximum Likelihood method was used to estimate the model’s parameters, and the Chi-Square test (χ²) was applied to assess the overall model fit.

Statistical analysis

Both descriptive and multivariate statistics were utilized to assess significant differences among extracted P pools and their transformation in soils, attributed to the effects of contrasting land uses and varying soil depths. The data were analyzed using a factorial arrangement (7 land use x 5 soil depths) and the 2-way analysis of variance procedure of the SAS® (SAS 2012). In all statistical analyses, significant main and interactive effects of predictor variables on dependent variables were identified using the Least Significant Difference (LSD) Test at p ≤ .05, unless otherwise stated. Correlation analyses were conducted using OriginPro® 2022.

Results and discussion

Soil pH, total organic carbon, nitrogen, and phosphorus

Averaged across land uses, the soil pH narrowly ranged from 5.7 to 6.1 across the 0–30 cm depth, with the highest pH values observed at 22.5–30 cm depth (Table 1). Conversely, TN and SOC contents exhibited a decrease with soil depth. Specifically, SOC concentration showed a stratified pattern, with the highest concentration (3.2%) at the surface depth (0–2.5 cm) compared to the lowest concentration (1.22%) at 22.5–30 cm depth. A similar trend was observed for TN, where the surface soil contained 0.27% TN, whereas the deeper depth had a lower concentration of 0.11% TN. Similarly, TP content also decreased with soil depth.

Table 1. Soil properties, including pH, total nitrogen (TN), total soil organic carbon (SOC), and total phosphorus (TP) at different depths, averaged across various land uses.

Depth (cm)	Properties/unit	n	Mean ± std	Min	Max
0–2.5	pH		5.67 ± 0.55	4.62	6.71
	TN (%)	35	0.27 ± 0.10	0.13	0.69
	SOC (%)		3.16 ± 1.32	1.59	8.79
	C/N ratio		11.8 ± 1.3	9.0	13.7
	TP (mg/kg)		226.5 ± 56.7	113.4	373.2
2.5–7.5	pH		5.29 ± 0.54	4.15	6.44
	TN (%)	35	0.23 ± 0.07	0.13	0.48
	SOC (%)		2.68 ± 0.92	1.56	6.07
	C/N ratio		11.4 ± 1.2	8.8	14.1
	TP (mg/kg)		161.4 ± 49.4	64.7	265.0
7.5–15	pH		5.94 ± 0.52	4.82	6.78
	TN (%)	35	0.18 ± 0.04	0.11	0.24
	SOC (%)		2.02 ± 0.51	1.03	2.84
	C/N ratio		11.2 ± 1.2	9.3	14.1
	Total P (mg/kg)		143.5 ± 53.4	57.9	279.5
15–22.5	pH		6.06 ± 0.54	4.92	7.02
	TN (%)	35	0.14 ± 0.06	0.05	0.25
	SOC (%)		1.62 ± 0.71	0.51	2.95
	C/N ratio		11.0 ± 1.2	8.9	12.8
	TP (mg/kg)		132.0 ± 51.4	51.2	262.9
22.5–30	pH		6.13 ± 0.64	4.44	7.12
	TN (%)	35	0.11 ± 0.05	0.04	0.20
	SOC (%)		1.22 ± 0.64	0.61	4.25
	C/N ratio		10.7 ± 1.2	4.4	7.1
	TP (mg/kg)		116.8 ± 45.9	49.4	217.3

The significant differences in soil properties by depth were attributed to variations in land uses especially, tillage intensities, cropping systems, inputs, biological activity, and the leaching of nutrients. Previous studies have reported a similar trend of SOC and TN changes with soil depth (Sarker et al. 2018; Sundermeier et al. 2011). Likewise, Gama-Rodrigues et al. (2014), noted a decrease in P content with an increase in soil depth. Additionally, the concentration of SOC and TN decreased with increasing pH, attributing the effect of pH on soil properties.

Land use effects on soil phosphorus distribution

Averaged across land uses and soil depth, the concentration of sequentially extracted P pools was significantly affected by the effects of both land use and soil depth, without significant interactions observed, except for FeAlP (Table 2). Among the P pools, RP had the highest concentration, ranging from 58.5 to 151.2 mg/kg, followed by FeAlP with concentrations ranging from 7.5 to 47.1 mg/kg. CaMgP showed concentrations between 5.6 to 26 mg/kg, while POP ranged from 6.7 to 13.4 mg/kg. The pools with the lowest concentrations were SRP, ranging from 0.54 to 2.76 mg/kg, and EP, ranging from 0.2 to 1.15 mg/kg.

Table 2. Distribution of sequentially extracted soluble reactive (SRP), extractable (EP), calcium- and magnesium bound (CaMgP), iron- and aluminum bound (FeAlP), particulate organic (POP), residual (RP), and total phosphorus (TP) contents at different depths of soil under various land uses.

Managed ecosystem	Depth	SRP	EP	FeAlP	CaMgP	POP	RP	TP
	cm	mg/kg
Forest	0–30	1.08bc≠	0.95ab	29.6b	21.2ab	11.7ab	121.6b	186.1b
WS	0–30	1.46b	0.44bc	20.5c	6.5 cd	8.4bc	119.3bc	156.6c
MRI	0–30	1.08bc	0.20c	16.2c	5.6d	11.6ab	105.8d	140.3c
SS	0–30	0.54c	0.26c	20.4c	7.6 cd	6.7c	108.3 cd	143.8c
CS	0–30	0.86bc	0.58abc	13.2 cd	26.0a	10.7ab	100.0d	151.1c
CSR	0–30	2.76a	1.15a	47.1a	14.3bc	13.4a	151.2a	229.9a
SP	0–30	1.35b	0.75abc	7.5d	7.0 cd	10.2abc	58.5e	85.2d
Managed agroecosystems x depth
Forest	0–2.5	2.44	0.8	32.7	41.3	33.3	170.9	281.3
	2.5–7.5	1.38	0.76	28.9	22.6	13	124.8	191.4
	7.5–15	1.16	1.9	38.4	13.8	6.2	112.4	173.9
	15–22.5	0.26	0.54	25.6	12.6	3.4	100.8	143.2
	22.5–30	0.16	0.74	22.4	15.8	2.5	99.2	140.8
WS	0–2.5	2.32	0.27	38.1	8.2	21.8	150.4	221.1
	2.5–7.5	2.87	0.84	30.5	8.1	7.2	134.6	184.2
	7.5–15	1.16	0.32	14	5.5	4.9	110	136
	15–22.5	0.63	0.3	11.6	5.6	4.4	103.2	125.7
	22.5–30	0.33	0.46	8.2	5.2	3.6	98	115.8
MRI	0–2.5	3.22	0.26	24.9	7.7	34.9	127.5	198.3
	2.5–7.5	1	0.16	23	5.9	9	121	160
	7.5–15	0.52	0.26	11.1	5.6	5.2	108.2	130.9
	15–22.5	0.36	0.14	13.7	4.4	4.3	93.7	116.5
	22.5–30	0.3	0.18	8.1	4.3	4.3	78.8	96.1
SS	0–2.5	1.48	0.3	34.2	10.8	21.3	141.2	209.3
	2.5–7.5	0.46	0.22	11.6	6.6	4.9	116.1	139.8
	7.5–15	0.38	0.22	15.6	6.6	2.9	105.6	131.2
	15–22.5	0.28	0.34	17.9	6.8	2.3	98.1	125.7
	22.5–30	0.1	0.24	22.6	7.5	1.8	80.7	112.8
CS	0–2.5	1.76	0.66	18.2	32.8	30.1	134.5	217.9
	2.5–7.5	1.04	0.42	18.2	26.6	8.7	109.1	164.2
	7.5–15	0.76	0.32	13	19.4	5	95.4	133.9
	15–22.5	0.46	0.94	12.5	42.5	7	86	149.4
	22.5–30	0.3	0.56	3.9	8.5	2.6	74.6	90.4
CSR	0–2.5	4.82	0.36	46.1	34.1	34.9	175.9	296.1
	2.5–7.5	3.82	2.16	26.8	12.4	12.5	157.9	215.5
	7.5–15	2.8	0.88	54.2	10.3	9.3	151	228.5
	15–22.5	1.44	0.22	56.3	7.4	6.5	143.9	215.7
	22.5–30	0.94	2.14	52.3	7.3	4	127.1	193.8
SP	0–2.5	2.64	0.56	7.3	19.6	37	94.4	161.4
	2.5–7.5	1.54	0.76	6.1	2.4	7.1	57	74.8
	7.5–15	1.34	0.82	7.1	3.7	3.4	54.1	70.3
	15–22.5	0.87	0.23	7.7	4.8	1.9	46.7	62.1
	22.5–30	0.25	1.68	9.8	4.4	1.4	38.7	56.1
LSDp ≤.05
Soil depth		<0.001	0.36	0.014	<0.001	<0.001	<0.001	<0.001
Land use x depth	0.77	0.55	0.008	0.16	0.46	0.98	0.53

≠ Means separated by same lower-case letter under each column did not vary significantly at p ≤ .05 among managed land uses.

WS=Wheat stubble cropping, MRI=Modified relay intercropping, SS=Soybean stubble cropping, CS=Corn stubble cropping, CSR=Corn-soybean rotation with rye as a cover crop, and SP=Sheep pastures.

Soils under corn-soybean, with winter rye as a cover crop, exhibited the highest SRP concentration after the crop harvest (October), followed by wheat stubble (WS) corn-soybean and sheep pastures. In contrast, the lowest SRP concentration was observed under soybean stubble corn-soybean, as compared to the woodlot. A similar impact of land use was observed for EP, FeAlP, CaMgP, RP, and TP concentrations, except for POP. The POP concentration was highest in corn-soybean with rye, followed by the woodlot, modified relay intercropping, corn stubble, and sheep pastures (SP). The lowest POP concentration was found in soybean stubble. Regardless of the land uses, the concentration of all the P pools significantly decreased with soil depth.

When expressed as percent contribution, the concentration of various P pools exhibited significant variations across the land uses (Figure 1). The SRP concentration ranged from 5.9% to 30.2% across the land uses, with the corn-soybean with rye cover crop contributing the most and soybean stubble contributing the least to the SRP pool. Similarly, the rye cover cropped corn-soybean had the highest contribution to the EP pool, while the modified relay intercropping had the lowest contribution. The corn stubble contributed the highest to the CaMgP pool, while the modified relay intercropping had the lowest contribution. In contrast, the rye cover cropped corn-soybean contributed the highest to the FeAlP pool, while the sheep pastures contributed the lowest. The corn-soybean with rye had the highest contribution to the POP pool, while the soybean stubble contributed the lowest.

Figure 1. Distribution of soil phosphorus pools under diverse managed land uses (Ws=wheat stubble cropping, MRI=Modified relay intercropping, SS=Soybean stubble cropping, CS=Corn stubble cropping, CSR=Corn-soybean rotation with rye as a cover crop, FOR=Forest, and SP=Sheep pastures).

Land use effects on soil labile and non-labile phosphorus pools

Based on the conceptual P availability, soil P pools were grouped into labile (SRP + EP), moderately labile (FeAlP + CaMgP + POP), and stable or non-labile P (RP), respectively (Figure 2). The rye cover cropped corn-soybean exhibited a significantly higher labile P concentration across soil depth (0–30 cm) compared to other land uses, which was 3-folds higher than the soybean stubble system. Similarly, the rye cover cropped corn-soybean had more than 2-folds the moderately labile P concentration when compared to the sheep pastures at the same depth; however, the P distributions were decreased with soil depth a trend attributable to diminished root activity and organic matter inputs, which are more pronounced at the surface. Organic P constituted the largest fractions within the labile P (SRP + exchangeable P) and moderately labile P (FeAlP + CaMgP + particulate organic P) pools in surface soils. At the 0–2.5 cm soil depth, the concentration of labile P and moderately labile P is high because of the higher concentration of SRP, EP, FeAlP, CaMgP, and particulate organic P. This pattern aligns with the findings of Niederberger, Kohler, and Bauhus (2019), suggesting that surface soil layers, enriched with organic residues and root exudates, facilitate higher P availability.

Figure 2. Concentration of labile, moderately labile, and stable P pools across under various land uses (across soil depth) and at different soil depth (across land uses). WS=Wheat stubble cropping, MRI=Modified relay intercropping, SS=Soybean stubble cropping, CS=Corn stubble cropping, CSR=Corn-soybean rotation with rye as a cover crop, FOR=Forest, and SP=Sheep pastures.

The moderately labile- and stable P concentrations were significantly higher at 0–30 cm depth under the rye cover cropped corn-soybean than that of the other land uses. In contrast, sheep pastures had the lowest concentration of both moderately labile- and stable P contents compared to other land uses.

While the percent distribution of labile, moderately labile, and stable P pools varied significantly with the land uses of rye cover-cropped corn-soybean and sheep pastures (Figure 3(b)), the distribution of P pools was also affected by depth (Figure 3(a)). The highest labile P concentration was 2% at 2.5–7.5 cm depth under all the land uses, whereas other depths had a labile P concentration of 1% relative to the total soil P concentration (Figure 3(a)). The contribution of labile P ranged from 1 to 2% of the total soil P concentration. Moderately labile P was distributed in the range of 25 to 36%, with the highest concentration observed at 0–2.5 cm depth. In contrast, the highest distribution of stable P was observed at 22.5–30 cm depth, while the lowest distribution was observed at 0–2.5 cm depth in all the land uses.

Figure 3. Percent distribution of labile, moderately labile, and stable P concentrations (a) under various land uses (across soil depth) and (b) at different soil depths (across land uses). WS=Wheat stubble cropping, MRI=Modified relay intercropping, SS=Soybean stubble cropping, CS=Corn stubble cropping, CSR=Corn-soybean rotation with rye as a cover crop, FOR=Forest, and SP=Sheep pastures.

Soils under both modified relay intercropping and sheep pasture agroecosystems had the highest labile P concentration (2%), while the others had 1% at 0–30 cm depth (Figure 3(b)). In contrast, the woodlot had the highest moderately labile P content (34%) when compared to both modified relay intercropping and soybean stubble agroecosystems (24%). In contrast, wheat stubble had the highest stable P content (76%), and the woodlot had the lowest (65%). The distribution of labile and non-labile P pools decreased with depth under all the land uses (Figure 3(a)); however, the change was inconsistent.

Cereal rye, when used as a winter cover crop, takes up available P from the soil solution and slowly recycles the biomass contained organic P via decomposition. As a result, soils under rye cover crops contain the highest concentration of labile, moderately labile, and stable P pools (White et al. 2020). Crop stubbles, especially soybeans, are rich in P, thereby contributing to the soil labile P pool (Huang et al. 2011). However, due to various biochemical reactions, macropore flow and surface runoff, the P concentrations do vary with each stubble’s properties. Similarly, the sheep pasture experiences P-enrichment from annual inputs of animal manures and legumes, which contributes to the more labile P pool.

Relationship between soil properties and phosphorus pools

Correlation analyses have shown a significant positive linear relationship between SRP concentration with POP, POM, and SOC, indicating their interdependence in P solubility or vice versa (Table 3). The correlation between SRP and POP can be attributed to the release of various forms of P, especially SRP, due to the temporal decomposition of POM in the soil. Higher levels of POP, which represents P organically bound to POM, contribute to increased SRP concentrations. Similarly, the decomposition of POM, consisting of macroaggregate associated plant residues, microbial biomass and their metabolites, releases SRP into the soil solution, establishing a positive correlation between SRP and POM. Additionally, residues persist on the soil surface, with POM being subjected to annual freezing and thawing or wetting-drying cycles, leading to the lysis of plant cells and the subsequent release of SRP (Liu et al. 2019; Webb, Uemura, and Steponkus 1994). While C, N, and P are stoichiometrically linked in SOM which expected to influence their transformations in terrestrial ecosystems (Spohn and Stendahl 2022), a higher level of TN indicates greater N availability, which stimulates soil biological activities. Increased microbial activities are responsible for accelerated SOM decomposition, consequently influencing the release of SRP from organic P pools. As a result, the increased TN contents correlate with higher SRP concentrations. Additionally, labile SOC plays a crucial role in SOM decomposition and nutrient cycling (Islam and Sherman 2021). The interaction between C and P regulates microbial processes involved in the mineralization of organic P compounds, resulting in increased SRP concentrations with higher SOC contents.

Table 3. Pearson correlation among soil phosphorus pools and other properties.

P pool	SRP	EP	FeAlP	CaMgP	POP	POM	RP	TP	TN	SOC
pH	0.04	−0.10	0.04	−0.08	−0.03	−0.27	0.09	0.04	−0.15	−0.14
SRP		0.07	0.36*	0.27*	0.49*	0.30*	0.51*	0.59*	0.39*	0.31*
EP			0.12	0.01	−0.04	0.02	0.08	0.09	−0.05	−0.10
FeAlP				0.12	0.19*	0.18*	0.55*	0.70*	0.08	0.05
CaMgP					0.27*	0.33*	0.22*	0.47*	0.29*	0.30*
POP						0.66*	0.47*	0.61*	0.53*	0.51*
POM							0.46*	0.55*	0.58*	0.60*
RP								0.92*	0.48*	0.46*
TP									0.49*	0.47*
TN										0.97*

*Significant at the p ≤ .05 level.

SRP:Soluble reactive P; EP: Salt-extractable P; POP: Particulate organic P; POM: Particulate organic matter; RP: Residual P; TP: Total P; TN: Total nitrogen; SOC: Soil organic carbon.

The FeAlP and CaMgP pools, representing P bound to Fe-Al oxides and exchangeable Ca and Mg in the soil, respectively, demonstrated positive correlations with POP and other P pools (Table 3). These correlations suggest that under specific environmental conditions, such as changes in soil pH or redox potential, the bound P from FeAlP and CaMgP can become mobilized and soluble as SRP. Specifically, a decrease in pH can lead to the solubilization of P from FeAlP due to the increased solubility of Fe and Al oxides under acidic conditions. Furthermore, seasonal wetting-drying of soil influences the Fe oxidation-reduction to release SRP in soil. Conversely, an increase in pH at slightly acidic or neutral can promote P available in soil solution from CaMgP, as slightly acidic or neutral pH levels can weaken the bonds between P and Ca as well as reduced solubility of reactive Ca. Significant correlations among FeAlP, CaMgP, and TP pools were observed, indicating the complex interactions between P amendments and fertilizers with soil Ca, and as well as Fe, Mn, and Al oxides and hydroxides. The soluble inorganic anions (H₂PO₄⁻ and HPO₄^{− 2}) are prevalent in soils at a slightly acidic to slightly alkaline pH (5.6 to 8.5), often due to P amendments. In such environments, H₂PO₄⁻, HPO₄^{− 2} can react with Fe, Al, Mn, and Ca present in the soil, forming insoluble compounds such as Fe, Al, and Ca phosphates. However, the pH of our experimental soil ranges from 5.3 to 6.1, which is slightly more acidic than the optimal range for these reactions to occur extensively. This could still justify the observed correlations to some extent, as even at this lower pH range, limited reactions between H₂PO₄⁻ and soil minerals (Fe, Al, Mn, and Ca) to form insoluble phosphates may still occur, although the efficiency and extent of these reactions might be reduced compared to soils with pH closer to neutrality (Brady and Weil 1999; Rahman et al. 2021).

The above findings indicate that SRP is directly or indirectly correlated with POP, POM, SOC, FeAlP, and CaMgP to maintain its concentration in response to variable land uses and seasonal weather dynamics. Our study identified five core factors: FeAlP, CaMgP, POP, POM, and SOC, which collectively accounted for 52% of the total variations in SRP concentration (Table 4). Among these factors, POP emerged as the most significant contributor, explaining 15% of the variation, while other factors, namely FeAlP, CaMgP, POM, and SOC, contributed between 8 and 11% to the overall SRP concentration. The results highlight the significance of POP as a major contributor to SRP, indicating its potential role in soil P transformation and solubility.

Table 4. Effects of soil properties on soluble reactive phosphorus.

p-Pool	Correlation coefficient	p-value	% Contribution
pH	0.04	0.62	1.11
EP	0.07	0.39	1.97
FeAlP	0.36*	<0.0001	10.71
CaMgP	0.27*	<0.0001	7.99
POP	0.49*	<0.0001	14.83
POM	0.30*	<0.0001	9.13
RP	0.51	0.23	15.27
TP	0.59	0.60	17.86
TN	0.39	0.053	11.75*
SOC	0.31*	<0.0001	9.37

*Significant at the p ≤ .001 level.

SRP:Soluble reactive P; EP: Salt-extractable P; POP: Particulate organic P; POM: Particulate organic matter; RP: Residual P; TP: Total P; TN: Total nitrogen; SOC: Soil organic carbon.

Principal components analyses

Ten parameters, including pH, EP, labile-, moderately labile-, and stable-P, POM, TP, TN, and SOC, were selected for principal components analyses (PCA) to evaluate their contributory effects on SRP concentration (Figure 4). The PCA results have shown that the first principal component (PC1) and second principal component (PC2) accounted for more than 66% of the total variation in SRP concentration. Notably, the labile-, moderately labile-, and stable-P exhibited strong associations with the SRP. Labile-P (SRP + EP) and moderately labile-P (FeAlP, CaMgP, and POP) were positively associated with the SRP. Similar correlations and their contributions to the SRP were also observed (Table 3 and 4). Furthermore, a strong positive linear correlation was found between the SRP and labile-P, indicating that labile-P concentration had a significant relationship with the SRP than with the EP. Similarly, the moderately labile-P (FeAlP, CaMgP, and POP) showed close relationships with both stable- and total-P pools. Stable-P displayed a significant positive correlation with the SRP, labile-P, moderately labile-P, and TP pools. In contrast, TN and SOC were weakly associated with the SRP, labile-P, moderately labile-P, TP, and stable-P. However, the SRP did not associate with soil pH. This outcome can be attributed to the method used for extracting SRP, which involved shaking the slightly acidic soils with distilled deionized neutral water to produce or avoid extracting any artifacts. Consequently, P could be released into the water from soils enriched with cover crops, fertilizers and manures, which are significant contributors to SRP levels. It is well-established that soil pH plays a pivotal role in controlling P release mechanisms. These mechanisms are influenced by the presence of compounds such as FeAlP, CaMgP, POP, and SOC. However, the extraction method used did not specifically target these pH-dependent P release processes. Instead, it is likely that the SRP measured was more reflective of the existing or immediately available P pool, influenced by external inputs rather than the inherent soil properties controlled by pH. Furthermore, the lack of correlation between SRP and soil pH in our findings underscores the complexity of soil P dynamics, suggesting that factors beyond pH may have a more immediate impact on the solubility of P, particularly in soils subject to anthropogenic inputs. This emphasizes the need for a multifaceted approach to understanding soil P availability, considering both chemical processes and land management practices. Thus, the SRP concentration was significantly linked to the transformation of the FeAlP, CaMgP, POP, and stable-P within the TP pool in soil. Based on the above findings, we had the following reactions which were expected to associate with the SRP concentration at a specific soil pH:

(1) $\mathrm{Labile}-\mathrm{P}\left(\mathrm{Orthophosphate}\right)+{\mathrm{H}}_2\mathrm{O}\to \mathrm{SRP}$(1)

(2) $\mathrm{POM}+\mathrm{Microbial}\mathrm{decomposition}\left(\mathrm{mineralization}\right)\to \mathrm{SRP}$(2)

Figure 4. Principal components analysis to evaluate the relationships among the soil properties under diverse managed land uses.

(3) $\mathrm{Stable}-\mathrm{p}+\mathrm{Freezing}-\mathrm{thawing}+\mathrm{Acidic}\mathrm{pH}\to \mathrm{SRP}$(3)

(4) $\mathrm{FeAlP}+{\mathrm{H}}_2\mathrm{O}\left(\mathrm{pH}\right)\to \mathrm{Fe}{\left(\mathrm{OH}\right)}_3+\mathrm{Al}{\left(\mathrm{OH}\right)}_3+\mathrm{SRP}$(4)

(5) $\mathrm{CaMgP}+{\mathrm{H}}_2\mathrm{O}\left(\mathrm{pH}\right)\to \mathrm{C}{\mathrm{a}}^{+2}+\mathrm{M}{\mathrm{g}}^{+2}+\mathrm{SRP}$(5)

Path analysis of soil phosphorus

The relationship between SRP and other P pools was evaluated using a series of multiple regressions in a path analysis (Table 5). Path analysis showed that only the POP and RP pools had a significant direct influence on the SRP concentration (standardized β = 0.32, p ≤ .01 and β = 0.26, p ≤ .05, respectively). Similarly, the EP, FeAlP, and CaMgP pools had significant direct effects on SRP (β = 0.04, 0.14, and 0.11, respectively, at p ≥ .10). Furthermore, the FeAlP and CaMgP had an indirect positive influence on SRP concentration via RP (β = 0.47) and POP (β = 0.37) transformations, respectively. However, the EP had an indirect negative influence on SRP via POP transformation (β = −0.09) and a positive indirect influence on SRP via FeAlP (β = 0.10).

Table 5. Path analysis for direct (diagonal bold)- and indirect effects (off-diagonal) of different phosphorus pools on soluble reactive phosphorus.

Soil P pools	EP	FeAlP	CaMgP	POP	RP	r
EP	0.04	0.07	0.00	−0.07	0.03	0.61
FeAlP	0.10	0.14	0.02	−0.10	0.47	0.14
CaMgP	0.00	0.02	0.11	0.17	0.06	0.56
POP	−0.09	−0.09	0.21	0.32**	0.37	0.51
RP	0.06	0.59	0.10	0.49	0.26*	0.67

U = 0.79.

Significant coefficients (*p ≤ .05) and (**p ≤ .01) of direct paths and indirect paths with absolute values > 0.1 are shown in bold.

R2 = 0.37 (p ≤ .05). R (p < .05).

EP:Salt-extractable P; POP: Particulate organic P; RP: Residual P.

Factor analysis was used to reduce the dimensionality of the problem by employing fewer latent or unobserved variables to explain the variation in the measured variables (Gama-Rodrigues et al. 2014). From the factor analysis, two factors were identified that conceptually explained the grouping of P pools (Table 6). Factor one represented all the P pools (SRP, FeAlP, CaMgP, POP, and RP), while factor two represented only EP.

Table 6. Factor analysis of soil phosphorus pools.

	Factor
	1	2
Factor loading	–	–
SRP	0.78	–
EP	–	0.81
FeAlP	0.65	–
CaMgP	0.45	–
POP	0.70	–
RP	0.83	–

Only loadings ≥ 60% of the maximum are given except CaMgP.

EP:Salt-extractable P; POP: Particulate organic P; RP: Residual P.

Modeling of soil phosphorous transformation

The multiple regression and path analyses were able to distinguish between the direct and indirect effects of the soil properties, predicting the hypothetical distinction among the soil P pools. The levels of interdependence among the P pools were quantified using structural equation model (SEM) to understand the cause-and-effect relationship associated with the soil P transformation pathways. Figure 5 shows the structural model of the P transformation based on factorial analysis and known experience (Hou et al. 2016).

Figure 5. Structural equation model for the soil P cycle. All measured variables (in boxes) are represented as effect indicators associated with latent variables (in circles). The numbers correspond to the standardized parameters estimated (p < .001). Error variables (e1 to e7) are standardized values. Model χ² = 3.9, df = 3, p ≥.05.

In the conceptual model of the soil P cycle, it is assumed that SRP mediates the transformations among soil P pools (Johnson, Frizano, and Vann 2003; Stewart and Tiessen 1987; Tiessen, Stewart, and Cole 1984). Based on this logic, the model has shown that the SRP concentration is directly influenced by EP, FeAlP, CaMgP, POP, and RP, respectively. Therefore, SRP is closely related to soil labile, moderately labile, and stable P pools. Hence, the hypothetical structural model of the soil P cycle is composed of four P pools (latent variables), such as: (1) the EP in the labile P pool, (2) the FeAlP, CaMgP, and POP in the moderately labile P pool, (3) the RP in the stable P pool, and (4) the SRP in the existing or immediately available P pool (Figure 5).

The model fitting fulfills our assumption (χ2 = 3.94, df = 3, p = .26; NFI = 0.98; IFI = 0.99; CFI = 0.99; RMSEA = 0.04), reaching a satisfactory level (Table 7). The model clearly shows that the available P pool is directly dependent on the labile, moderately labile, and stable P pools. The standardized direct effect of the labile, moderately labile, and stable P pools on the available P pool is 0.044, 0.053, and 0.017, respectively. The indirect effects of the labile, moderately labile, and stable P pools are negligible or zero on the available P pool. Thus, the overall effects are the same as the direct effects, which are positive. The overall effects of these three P pools on the available P pool are positive (β = 0.044 + 0.053 + 0.017 = 0.11).

Table 7. The parameters and values required for model fitting.

Index Name	Evaluation Criterion	Model
The chi-square (χ^2)	The smaller the better	3.94
The ratio of chi-square and freedom (χ2/df)	1 ~ 3. When the ratio is less than 1, the model is over adapted, when the ratio is between 1 and 3, the model is well adapted, when the ratio is greater than 3, the model is poorly fitted.	1.31
Significant probability (p)	>0.05	0.26
Normative fit index (NFI)	0 ~ 1. A value greater than 0.7 is acceptable, the closer to 1 is better.	0.98
Incremental fit index (IFI)	0 ~ 1. A value greater than 0.7 is acceptable, the closer to 1 is better.	0.99
Comparative fit index (CFI)	0 ~ 1. A value greater than 0.7 is acceptable, the closer to 1 is better.	0.99
The root mean square error of approximation (RMSEA)	<0.05. The smaller the better	0.04
Akaike information criterion (AIC)	The smaller the better.	39.94
Bayes criterion (BCC)	The smaller the better.	41.45

Ref: Wei et al. (2018).

The structural model of soil P transformations suggests that the labile, moderately labile, and stable P pools act as a combined source for the available P pool associated with the SRP. Among the three P pools, the moderately labile-P pool (FeAlP, CaMgP, and POP) contributes (β = 0.05) more to the available P pool than the other two P pools (β = 0.04 and 0.02). Soil P from the stable-P pool contributes to the available P pool via the dissolution of inorganic P and/or mineralization of organically-bound P in the soil. The readily mineralized P (EP) from the labile P pool exerts a direct effect on the available P pool. However, the transformation of POP to the available P pool has a direct effect via the moderately labile-P pool. The contribution of the labile, moderately labile, and stable-P pools to the available P pool is further justified by their concentration and percent distribution in the soil. It was observed that the SRP comprises of the total available P pool, which is expected to originate from EP, FeAlP, CaMgP, POP, and RP, respectively. Therefore, our model is justified, considering the overall positive effect (β = 0.11).

Conclusions

Soil P pools and their distribution were significantly influenced by the effects of land uses and soil depth. Among the studied land uses, soils under the rye cover cropped corn-soybean exhibited the highest concentration of labile P compared to the natural forest, when used as a control. In contrast, soils under wheat, modified relay intercropping, and soybean stubble systems showed significant accumulation of stable P. The labile P pool, represented by the SRP, displayed a significant positive correlation with FeAlP, CaMgP, POP, RP, and TP in the soil. Path analysis revealed that POP had a direct influence on the SRP concentration, while FeAlP and CaMgP pools had indirect effects on SRP concentration. Structural equation modeling indicated that labile, moderately labile, and stable P pools are responsible for varying contributions to the SRP concentration, thus underscoring the complex interactions within soil P dynamics. Further studies are needed to explore the specific contributions of land uses on soil P transformation and its implications for freshwater eutrophication concerns. Understanding these relationships can aid in developing sustainable agricultural practices to manage soil P effectively and mitigate environmental impacts.

Disclosure statement

No potential conflict of interest was reported by the author(s).