Phosphorus Sorption by Purple Soils in Relation to Their Properties: Investigation, Characterization, and Explanation

Abstract

Improved soil phosphorus (P) management can be achieved through an understanding of regional soil–P interactions and their relation to soil properties. To this end, soil samples from different use types (paddy soils, dry farmland soils, forest soils, and urban green land soils) were collected from 10 sites across the west to the east of Sichuan Basin, China. These samples were analyzed to determine their P sorption properties and physical and chemical compositions. P sorption was described using a modified Freundlich equation. The results demonstrated a gradient in P sorption within the basin, characterized by higher values in urban areas and a west-to-east decrease trend, along with the null-point values of soil sorption–desorption equilibrium. This variation was linked to the extensive use of P fertilizer, which altered soil particle surface conditions and significantly reduced both the quantity and rate of subsequent fertilizer sorption. Furthermore, P sorption was found to be correlated with the soil clay fraction, amorphous aluminum oxides, and soil organic matter contents. Urban expansion and accelerated erosion of productive agricultural land increase mean soil particle size and may decrease soil P holding and retention capacity. As preliminary deterioration in soil properties was found, conservative soil management is needed to address the potential threats of soil degradation in the central Sichuan Basin.

1. Introduction

Over the past few decades, human activities have been the drivers of ecosystem transformations through the conversion of natural landscapes into agricultural and urbanized lands, causing significant evolutions in soil physical and chemical properties [1,2,3]. The behavior of phosphorus (P) is affected by its reactions with soil, including sorption reactions that determine its movement towards plant roots and removal from soil by leaching [4]. Measuring the amount of P sorbed and summarizing the sorption curves facilitates the interpretation of the sorption properties, and these values can be used to predict the fertilizer requirements [5].

Phosphate reacts with specific sites on the surface of variable charge particles through a process known as “specific adsorption”. This reaction is reversible and influenced by the electric potential of the reacting surfaces. Adsorption can be plotted against concentration to demonstrate the incomplete nature of the reaction. A subsequent slower reaction likely occurs due to solid-state diffusion into the particles. The entire process is referred to as “sorption”. The Langmuir equation, widely used in describing soil P sorption behavior [6,7,8], allows for the calculation of a maximum sorption [7,9]. However, the fitted value for the maximum is usually exceeded by the data; this occurs because the equation does not describe observed sorption by soil very closely. The equation also faces issues in describing ion sorption in soils because it assumes a uniform surface and no alteration of surface properties except for site occupancy [5]. Yet, soil surfaces are typically non-uniform, and reactions introduce negative charges, reducing further reactions. In experiments studying sorption kinetics, solution concentration decreases over time as sorption increases, resulting in varying shapes of sorption over time and concentration over time plots depending on the degree of sorption, which are traditionally characterized using the Elovich equation or combinations of first-order rate equations [10,11,12]. However, recent research has demonstrated that these diverse shapes can be effectively described by a single simplified equation.

In the realm of modeling external variables like time, phosphate concentration, temperature, and pH, simple equations provide accurate representations, playing a vital role in comprehending the underlying processes and facilitating cross-soil comparisons. A fundamental component of this modeling approach involves employing the Freundlich equation to elucidate concentration effects. Despite its lack of a mechanistic foundation, it arises as a result of soil heterogeneity [13,14]. The Freundlich equation offers improved sorption description by postulating heterogeneous surfaces with a logarithmic distribution of binding constants. When considering reactions with soils, heterogeneity is more effectively characterized through the assumption of a normal distribution, permitting the depiction of cation and anion reactions at the negative and positive ends of the distribution, respectively. Notably, for minimal sorption quantities, such as with selenite, the Freundlich equation proves inadequate due to the fractional power’s increase with rising sorption [15]. Conversely, for phosphate, typically measured at relatively high sorption levels, the Freundlich equation serves as a suitable approximation.

Purple soils in southwest China are highly fertile. Especially in the Sichuan Basin, 70% of the arable land is on purple sandstone and shale, easily weathered and generally coarse in texture. They developed on Mesozoic (Triassic, Jurassic, and Cretaceous) and Tertiary sedimentary rocks and are classified as entisols in the USDA Soil Taxonomy [16,17]. The region is densely populated and has a long history of agriculture; it produces 10% of the country’s food and livestock feed on 7% of China’s arable land [16]. Many researchers have been focusing on P sorption and leaching processes in the Sichuan Basin and the associated factors that influence P transport and transformation processes [18,19,20,21]. Soil P loss exceeding 1.0 kg ha⁻¹ has been recorded during heavy and storm rain events [22,23,24]. Prior reaction with P adds a negative charge to the reacting surfaces and this decreases the P buffering capacity; prior reaction with P fertilizer also slows the diffusive movement of phosphate into the adsorbing particles.

The phenomena of P sorption and leaching in purple soils remain elusive in experimental observations and model interpretations. Without a comprehensive understanding of these effects, there is a risk of the over-application of P fertilizer, potentially leading to water pollution. Thus, the present study aimed to (1) characterize the P sorption by purple soils in relation to their properties and (2) assess the relationship between sorption parameters and soil P status, soil use types, and soil degradation trends.

2. Materials and Methods

2.1. Soil Sampling and Treatment

The sampling sites were distributed in a significant agricultural belt with a latitude range of 29°58′ N–30°52′ N and a longitude range of 102°59′ E–107°37′ E within the Sichuan Basin in Sichuan Province. The sites were oriented west-to-east through the hilly region of west and central Sichuan, the Chengdu Plain, and the parallel ridge and valley area of east Sichuan, China. The study area has a subtropical monsoon climate, characterized by year-round warm and humid weather, high cloud cover, few sunny days, an annual average temperature of 16–18 °C, a daytime temperature of ≥10 °C for 240–280 consecutive days a year, a small diurnal temperature range and large annual temperature range, warm winters and hot summers, and a frost-free period of 230–340 days. The annual precipitation is 900–1200 mm, with 90% occurring from May to October. Soils of the study area consist of highly fertile, purple-brown forest soils, which rapidly adsorb and lose nutrients but are highly prone to erosion. Non-calcareous alluvial soils and paddy soils are widely distributed in the study area. These soils are the most crucial soil class groups for agriculture owing to their high fertility and are mainly formed from fertile black soils eroded from the peripheral areas of Tibet. Soil samples were collected from the cities of Ya’an (YA), Qionglai (QL), Chongzhou (CZ), Chengdu (CD), Ziyang (ZY), Suining (SN), Nanchong (NC), and Dazhou (DZ) (Figure 1). The coordinates of the sites are listed in the Supplementary material (Table S1).

Based on the investigation report for the General Plan of Land Use of Sichuan Province from the Sichuan Provincial People’s Government [25], sampling sites were chosen to represent the local agricultural or natural soil conditions, including paddy soils, dry farmland soils, forest soils, and grassland soils. Although the limited sampling sites in this study could not cover the entire Sichuan Basin, variations in the interaction pattern between P and soil due to changes in the physical and chemical properties of the soil can be detected at large scales.

2.2. Physical and Chemical Analysis of Sampled Soils

Surface soils within a depth of 20 cm were collected with a stainless soil auger and sealed in separate sampling bags. The samples were air-dried for 21 days, and then ground by a ceramic mortar and pestle, sieved through 2 mm sieves, and stored in sample bags before testing. The collected soil samples were tested for grain size composition, total phosphorus (TP), total nitrogen (TN), amorphous iron oxides (Fe_[amo]), amorphous aluminum oxides (Al_[amo]), soil organic matter (SOM), and pH. Grain size distribution was determined using a Mastersizer 3000 laser diffraction particle size analyzer (Malvern Panalytical, UK), and the soil texture was determined according to the soil taxonomy of USDA [26]. Five grams of the sieved sample was mixed with 25 mL of deionized water, shaken at 160 times min⁻¹ using a reciprocating, constant-temperature water-bath shaker (SHA-B, Yoke Instrument, Shanghai, China), and set aside for 30 min. The pH of the supernatant was measured using a pHS-3C pH meter (INASE Scientific Instrument Co., Ltd., Shanghai, China). Soil TP was measured following the perchloric acid digestion procedure with minor revision [27,28]. For each sample, 0.2 g of the ground and sieved soil sample was placed in a digestion tube and 3 mL of 70% HClO₄ was added. The tube was placed in an aluminum digestion block and the sample digested at 200 °C for 75 min; a funnel was placed atop the tube throughout the digestion to ensure refluxing of HClO₄. Following digestion, the digest was allowed to cool and then diluted with distilled water to 50 mL. The tube was stoppered, inverted several times to mix the contents, and allowed to stand overnight; the residue was removed from the extract by centrifugation at 5000 rpm for 15 min, thereby permitting P analysis. The digestion procedure was also used to determine the soil TN [29]. The P and N contents of the digested soils were measured via colorimetry using a Proxima continuous flow analyzer (AMS Alliance, Frepillon, France). For Fe_[amo] and Al_[amo], extraction was performed using an oxalic acid–ammonium oxalate buffer solution [30], and the Fe_[amo] and Al_[amo] concentrations were measured by inductively coupled plasma–optical emission spectrometry using the 720-ES ICP-OES system (Agilent Technologies, Inc., Santa Clara, CA, USA). The SOM content was measured using the Walkley–Black method [31].

2.3. Sorption Experiments for Sampled Soils

For each soil, there were two sets of measurements of phosphate sorption; in one, sorption was measured after 24 h for a range of initial concentrations, and in the other, sorption was measured after different periods at one initial concentration. The protocol was adapted from Barrow and Debnath [32]. For both sets, 2 g of the ground and sieved soil sample were weighed, dispersed in 0.01 M KCl solution, and made up to a volume of 30 mL with appropriate volumes of KH₂PO₄. For the first set, the concentrations were 5, 10, 20, 30, 40, 60, and 80 mg·L⁻¹. For the second set, an initial concentration of 40 mg·L⁻¹ was used for most soils, but 60 mg·L⁻¹ was used for soil YA because of its higher sorption; the periods of mixing were 0.25, 0.5, 1, 2, 5, 10, 24, 32, and 48 h. The pH of each solution was adjusted to 7 using a 0.01 M HCl solution and 0.01 M NaOH solution, and one drop of 0.1% chloroform solution was added to eliminate the influence of microbial activity on the experiments. After the soil suspension had been shaken with a reciprocating shaker at an amplitude of 6 cm and a frequency of 150 oscillations per minute at 25 ± 1 °C for 24 h, centrifugation was performed at 5000 rpm for 15 min. The resultant liquid was passed through a 0.45 μm membrane filter and analyzed with a Proxima continuous flow analyzer to measure the P concentration in the supernatant. Three replicate samples were tested for each soil type, and the average value was used as the final measurement result of the amount of P sorbed. Consequently, we considered that measurements after 24 h of sorption were adequate to characterize differences between soils.

2.4. Model Characterization

In order to describe the effect of solution concentration and time on sorption, the measured data were described by a modified Freundlich equation, including the effect of time [33], which is a 4-parameter model, as shown by Equation (1).

$$S=a{c}^{b_1}{t}^{b_2}-q{t}^{b_2}$$

(1)

The variables of Equation (1) are not independent, as the sorption was calculated from the observed solution concentration. Equation (2) was used to predict values for any pair of values for the initial concentration and the solution concentration simultaneously at a given soil/solution ratio.

$$S=\left(ci-c\right)Rss$$

(2)

where S (mg·kg⁻¹) is the amount of P adsorbed by the soil at the solution concentration c (mg·L⁻¹) and at time t (h); q is formally equal to the P that could be desorbed if the solution concentration could be maintained at zero; ci (mg·L⁻¹) is the initial concentration, and Rss is the solution/soil ratio (Rss = 15 in this study). The parameter a reflects both the binding strength of P at the sorption sites and the number of sorption sites.

In the present study, the effects of concentration were measured after one period of reaction, and the effects of the period of reaction were measured at one initial concentration. There were therefore insufficient data to fully explore the concentration–time surface. We therefore used a simplified equation in which the same index term (b₂) was used for both time components of the equation. This may be explained as follows. The “null-point” (the concentration at which neither sorption nor desorption occurs) is given when S = 0 and $a{c}^{b_1}{t}^{b_2}=q{t}^{b_2}$. When the equation is specified in this way, the null-point does not change with time. This form of the equation is appropriate for soils in which phosphate has had a long time to react; further changes are slow.

Parameter b₁ describes the curvature of plots of sorption versus solution concentration. Its main determinant is the heterogeneity of the reacting surfaces [33]. The parameter b₂ describes the curvature of plots of sorption versus time. It is an important determinant of phosphate effectiveness.

The BASIC program [34] was used to fit the equations. For any given value of the parameters, the program finds the solution to the simultaneous equations by a series of extrapolations. A subroutine based on the simplex algorithm then adjusts the values of the parameters to find the values that minimize the sums of squares for the deviations between the observed values and the predicted values.

When fitting data with Equation (1), the coefficients are correlated—that is, variation in one can be partly compensated by variation in another. We decided to use an equation with the value of b₁ common to all soils because this gave us more consistent values of the other parameters. We tried different values of this parameter and chose the value (0.30) that gave the smallest residual sums of squares. We preferred this simplification even though it caused a significant increase in the residual sums of squares (

Table 1. Comparing the analysis of variance results of 3-parameter and 4-parameter fitting.

	Residual Sum of Squares	Degree of Freedom	Mean Square	VR
3-parameter	398.6300	130	3.3219
4-parameter	325.1200	120	2.7093	2.7100
Improvement	73.5100	10	7.3510

). A detailed comparison of the fitting results between common and separate values of b₁ is listed in the Supplementary material (Table S2).

3. Results

3.1. Phosphorus Sorption Properties

The sorption behavior of P was assessed using the modified Freundlich equation, affirming the effectiveness of this equation in describing the sorption characteristics of the sampled soils. Individual sorption curves, as depicted in Figure 2a–j, exhibited variations across different locations. Figure 2k provides a comparative analysis of these curves, revealing that soils YA and DZ demonstrated the highest phosphate sorption capacities, as indicated by the solid lines. The Supplementary material (Table S3) contains the corresponding values for the a parameter. In contrast, soils around Chengdu City (QL, CD1, CD2, CD3), represented by the blue dashed lines, displayed lower phosphate sorption capacities. Soils from other sites in the central and eastern regions, represented by the orange and red lines, exhibited the lowest phosphate sorption capacities.

3.2. Soil Physical and Chemical Properties

Within the transect of the study area, the physical and chemical properties of the soil are derived from their basic lithological features: the purple soil, which inherits many of the characteristics from its parent materials, has a coarse texture, low SOM content, and is prone to erosion [35]. The transect across the Sichuan Basin (Figure 3) showed that soil organic matter had a generally U-shaped distribution, with high values at the edges and lower values in the middle. The clay-plus-silt had a similar distribution though somewhat less marked. Consequently, the sand content was distributed oppositely, with the highest values towards the center of the transect. Several soil properties also showed a general U-shaped distribution with the highest values at the edge (Figure 4). In the case of soil aluminum, the values for the soils at the edge of the distribution were more than twice as high as for some of the soils in the center of the distribution.

4. Discussion

4.1. Historical Perspectives of Soil Particle Property Variation

Examining the historical variation in soil particle properties reveals significant changes in fine particles and their impact on soil phosphate sorption capacity. Soil fine particles play an important role in P retention, and the increase in the sorption on soil particles with decreasing grain size has been previously observed [36,37,38], and it is explained by the higher specific surface area-to-volume ratio and an increase in the concentration of soil surface functional groups [39]. Low contents of soil fine particles in central parts of the study area were found (Figure 3) and were accompanied by a decline in P sorption capacity in this study. According to field visits and related literature [40], farmers tended to meet the P demand of their crops by increasing fertilizer application to improve soil fertility. However, it caused an increase in TP in soils around Chengdu City (CD2, CD3, ZY, SN), accompanied by a large amount of P loss, which disrupted the balance of P circulation in the ambient environment [40,41]. This phenomenon has also been confirmed in the research of Li et al. [42].

Phosphate sorption decreases with increasing phosphate application, indicating that soil properties and land use both play a role. Rapid physical weathering of purple rocks and associated pedogenesis, sedimentation, and urbanization have affected erosion-related soil losses [20,35,43] and formed the current pattern of soil textural distribution within the study area. The loss of fine soil particles, especially in Chengdu City and its surrounding areas (CZ, CD1, CD2, CD3, ZY), is particularly significant in modern times. In previous investigations of historical conditions in the study area (Figure 5), it was found that the mean grain size gradually increased from west to east [44,45,46]. The mean grain size of soils at the west and east of the study area (YA, QL, NC, DZ) maintained their paleogeological state. However, with the soil disturbance in modern times, the soil mean grain size in the central basin has increased significantly (especially for CD3 and ZY), indicating that modern human activities could contribute to fine soil particle loss.

This case study served as an illustrative example of the intricate interplay between urban expansion and soil degradation within the region. The examination of Chengdu City’s historical expansion history [42,47] revealed profound repercussions on both natural and agricultural ecosystems. This analysis elucidated that the rate, process, and spatial distribution of urban area expansion closely aligned with the observed variations in soil properties, as illustrated in Figure 5. Despite the area’s extensive history of agricultural cultivation spanning approximately 4500 years, recent investigations of anthroposols development highlighted a noteworthy shift in soil mean size, a phenomenon primarily manifested within the past four decades. Importantly, it was essential to recognize that this change was intrinsically linked to the ongoing urban expansion within the region. Of significant concern was the substantial impact of modern agricultural practices within Sichuan Province, which had engendered a marked acceleration in the erosion of productive agricultural land [48]. The analysis of the rapid transformations in land-use patterns highlighted an elevated susceptibility to soil and water losses, particularly in densely populated central and eastern regions, including peri-urban areas [47]. The conspicuous prevalence of a high sand content, as evidenced by the CD1, CD2, and CD3 samples, provided support for these findings. Particularly noteworthy were the locales of CD1 and CD2, which were situated within urbanized green land areas characterized by a dearth of vegetation cover. This environmental context caused the erosion of clay during the ongoing modern urbanization processes in the Chengdu Plain [49,50,51].

4.2. Interpretation for the Sorption Parameters

In contrast to YA and DZ, lower values in P sorption capacity were observed in CD1 and CD2 soils. The focus of this analysis is on the concept of null-point values, which represent the concentration at which neither sorption nor desorption occurs, providing additional insights into soil characteristics. The results suggested that the soils in the central basin (QL, CD1, CD2, and CD3) have received large applications of P fertilizer for decades or have been exposed to P pollution during urbanization [22,23]. Figure 6 shows high null-point values for CD1 and CD2, indicating heavy use of P fertilizer and probable loss in drainage. The samples were measured in dilute sodium chloride solution, which would lead to lower null-point values than if they had been measured at a lower ionic strength (such as would occur at field capacity). It is quite feasible that water leaching through the soils would have a high phosphate content. Such soil conditions are considered to be a harbinger of soil degradation from the perspective of soil ecology [52,53]. Overloaded foreign P induces an initial reaction with soil, which significantly decreases both the amount and the rate of sorption of subsequent fertilizers. Freshly adsorbed P is more difficult to displace with anions due to sorption site occupation.

To compare the curves, the values of a were investigated using a common value of b₁. However, since most of the data were for sorption after 24 h, $a·{24}^{b_2}$ is a better value. Figure 6 shows that the highest values were observed for the outlier sites (YA and DZ), with generally lower values within the basin proper. Additionally, Figure 6 reveals that the null-point values were high for CD1 and CD2. These samples were taken from urban green spaces and suggest a heavy input of phosphate. Apart from these exceptions, there was a general downward trend from west to east. The values for the parameter b₂ ranged from 0.265 to 0.420, which were much higher than explained by the diffusion mechanism [33]. Mutual abrasion of the particles is considered to have caused these higher values, leading to much higher rates of reaction than would be observed in the field.

Consider a plot of sorption against concentration raised to the power (Figure 7a). Such plots will be linear. If the null-point is similar, the plot tends to be steeper and its interpolation will cut the vertical axis at a lower point. The value of the desorption term (q) in the model was correlated with a value at 24 h. So q would then increase as the a value increased, indicating that the soil surface properties largely decided the P sorption–desorption abilities, and the value indicated the height of the sorption curve [34]. The data appeared to fall into two groups, with the outlier group consisting of the two soils (YA and DZ) collected from the margin with much higher sorption. As the soil samples mostly came from areas fertilized with phosphate, prolonged contact between phosphate and soil could result in the penetration of the adsorbed molecules into the particles, thereby increasing the negative charge, making it more difficult to displace with other anions.

Bivariate analysis was used to analyze the correlation between the model parameters and the properties of the samples (

Table 2. Correlation between model parameters and soil properties.

Parameters	a·24b2	b2	q
Clay	0.820 **	0.043	0.656 *
Silt	0.644 *	0.087	0.526
Sand	−0.693 *	−0.080	−0.564
Mean grain size	−0.595	−0.353	−0.390
pH	−0.463	0.390	−0.458
Al[amo]	0.910 **	−0.030	0.726 *
Fe[amo]	0.708 *	0.136	0.439
TN	0.792 **	0.272	0.466
TP	−0.579	0.098	−0.524
SOM	0.813 **	0.221	0.569

Correlation is significant at the ** 0.01 or * 0.05 level (two-tailed test).

). The a value at 24 h had significant positive correlations with soil clay, Al_[amo], TN, and SOM contents. An especially highly significant positive correlation (p = 0.018) was found between the a value and the clay content. The Al_[amo] content was significantly positively correlated with the q value (p = 0.017) (Figure 7b), this could be attributed to the fact that Al_[amo] in soil particles mainly consists of amorphous inorganic polymers with huge surface areas and many ion binding sites [54,55], and the amorphous aluminum oxides could be a predictor of soil P affinity. These correlations indicated that the fine fractions exerted considerable effects on P sorption within the few soil samples used in this study. However, correlation cannot prove causation because these correlations may not be significant in a wide range. As the a value reflects the amount of reacting surfaces present in a given sample of soil and their affinity for P [33,55], the applicability of this interpretation to purple soil requires more evidence.

We expressed concerns because the data from this study indicate the loss of fine particles, incorporating SOM and Al_[amo] reduction, which potentially causes future soil deterioration. Heavy use of P fertilizer and subsequent changes in soil particle surface condition will eventually affect P sorption. Furthermore, urban expansion has caused severe soil pollution in agricultural areas, aggravated soil and water losses, and caused a continuous rise in soil intrusions [35,40], resulting in the occupation of sorption sites in soil by substantial amounts of foreign substances [56,57,58]. It is worth noting that P-containing detergent is widely used as the major constituent of synthetic detergents and has become the most common pollutant in the soil. As an accumulation, the surface properties of the soil may vary significantly, consequently reducing the soil ionic strength and sorption amount for beneficial nutrients [59]. These foreign substances affect the balance of soil and P interaction, affecting agricultural productivity and posing potential ecological and health risks.

5. Conclusions

This study examined P sorption and the physical–chemical properties of purple soils in the Sichuan Basin, encompassing paddy, dry farmland, forest, and urban green land soils. A modified Freundlich model was employed to analyze the sorption data, revealing decreasing sorption capacity from west to east, with high levels at the basin margins and in the central urban area. High P fertilizer use and changes in soil particle surface conditions in urban green land soils significantly affected P sorption. Reduced fine soil particles, soil organic matter, and aluminum content, possibly due to hydraulic erosion during recent rapid development, correlated with diminished soil P retention. These findings highlight preliminary soil property deterioration, emphasizing the need for specialized soil management protocols to address potential degradation threats.