Speciation of Iron and Aluminum in Relation to Phosphorus Sorption and Supply Characteristics of Soil Aggregates in Subtropical Forests

Abstract

Phosphorus (P) is one of the main limiting nutrients in subtropical forest soils. Both soil type and aggregate structure affect the P sorption capacity of soil; thus, determining soil P supply and leaching characteristics. However, the mechanism of their interactions on soil P sorption and leaching at an aggregate level remains unclear. We classified soil aggregates from red soils and limestone soils in a subtropical forest via wet-sieving and carried out P isothermal sorption experiments. The P sorption maximum (Q_m), P sorption strength (K_L, K_F), P sorption index (PSI) and maximum buffer capacity (MBC) were obtained by fitting to Langmuir and Freundlich isotherm equations. Moreover, different P fractions were determined to estimate the degree of P sorption saturation (DPS) of aggregates. The results showed that the Q_m of the two soils were similar, but the sorption strength (K_L, K_F) and MBC of the limestone soil were higher than those of the red soil. Higher contents of free iron (Fe) oxide and amorphous aluminum (Al) oxide in the limestone soil may enhance the P sorption capacity and, thus, reduce P availability, resulting in a higher total P retention capacity than in the red soil. A higher content of complex Fe in red soil may reduce P sorption and, therefore, play a role in maintaining the supply capacity of soil-available P. The 0.25–0.5 mm aggregates of the two soils had the largest MBC among all aggregate sizes, and their P sorption and buffering capacity were stronger than other aggregates. The DPS of different aggregate sizes were all low, indicating that the soils of subtropical forests were in a state of P deficiency; thus, the risk of P leaching was low. The <0.1 mm aggregate in red soil had relatively high DPS and significantly lower PSI than the other aggregate sizes, indicating that it was more prone to P leaching. The results provide further insight into forest management to improve P availability and reduce P leaching in subtropical forest soils.

1. Introduction

Phosphorus (P) is an essential nutrient for plant growth and development [1] and is one of the limiting elements determining the primary productivity of forest ecosystems [2,3]. The soil P cycle in forest ecosystems includes the processes of P input, transport and transformation and export [4]. Different fractions of soil P directly influence P bioavailability for forests, which, in turn, affects the transformation and retention of soil carbon storage [5,6], nitrogen [7], plant growth and microbial activity [8,9] in subtropical forests. A large amount of P is bound within iron (Fe) and aluminum (Al) oxides in soil aggregates as secondary mineral P, reducing the bioavailability of soil P and forest productivity in subtropical forests. A small proportion of P is lost through soil leaching and erosion to water [10], dramatically impairing aquatic ecosystems [11]. Consequently, P is the main limiting nutrient in subtropical forest soils [12]. Understanding P retention and the risk of P leaching in subtropical forests is important to evaluate the impact of P limitation on forest ecosystem productivity.

Aggregates are the basic units of soil structure that influence the myriad soil processes including P cycling. The aggregate stability is an important indicator of forest soil quality [13]. The physicochemical–biological components of soil aggregates vary considerably between particle sizes, which affects the soil’s P sorption capacity. Macroaggregates, microaggregates and clay particles are bound in different ways through soil organic matter and minerals [14,15]. Different sizes of aggregates have various specific surface areas, microbial compositions and abundances [16], suggesting the aggregates differ in absorbing and releasing of P [17].

Iron and Al minerals are present in large quantities in subtropical forest soils. Different fractions of Fe and Al oxides have various functions, such as sorption, desorption, oxidation and reduction, which can directly influence soil processes [18,19], and impact soil P supply and sorption capacity [20,21]. Furthermore, Fe and Al oxides have large specific surface areas, which not only provide substantial adsorption sites for phosphate anions, but can also form Fe and Al oxides or oxyhydroxides cutans to retain P, reducing P availability. Degree of P saturation (DPS) is mostly used to assess the environmental threshold of soil P and the risk of soil P leaching [22]. Different DPSs reflect the ability of different fractions of P to be released from soil. A higher DPS value indicates that there is more bioavailable P, and a greater P leaching risk from the soil. Currently, most studies on P sorption in subtropical forest soils have focused on bulk soils, but there is a lack of research on P sorption at an aggregate level and its mechanisms.

The objectives of this study were to examine P sorption and desorption mechanisms and how they affect P speciation, as well as to reveal P retention capacity and P leaching capacity from soil aggregates of subtropical forest soils. Special attention was paid to investigating the speciation of the Fe and Al controlling these mechanisms. This study provides further insight into the sustainable management of subtropical forests.

2. Materials and Methods

2.1. Study Sites

The study sites were located in Guilin National Forest Park (25°13′18.2′′ N, 110°14′53.3″ E) and Ludiyan (25°19′06.6″ N, 110°15′19.9″ E) in Guilin, Guangxi, China. The two sites are 12 km apart and have a subtropical monsoon climate. The precipitation and temperature are similar in both of the sites, with an annual average temperature of about 19.1 °C and an annual average rainfall of about 1887.6 mm. The soil of Guilin National Forest Park is made up of granite as the parent rock, the soil type is acidic red soil and the dominant tree species is broad leaf forest mixed with coniferous trees. The soil of Ludiyan is made up of limestone, with a neutral limestone soil and mixed evergreen and broad leaf forest being the dominant tree species.

2.2. Sampling and Analysis

Three sample plots were selected in each of the two study areas. Soil samples were taken in three randomly selected 50 × 50 cm squares at each site. After removing the surface foliage, the soil was stratified, according to the natural depth of the soil profile, and collected in O/A horizon (0–10 cm and 0–12 cm for Ludiyan and National Forest Park, respectively) and AB horizon (10–22 cm and 12–33 cm for Ludiyan and National Forest Park, respectively). After removing the stones and coarse roots, soil samples were air dried and then separated by wet sieving with a soil aggregate analyzer (DIK-2012, Daiki, Saitama, Japan) to classify five groups of soil aggregates of the following different sizes: 1–2 mm, 0.5–1 mm, 0.25–0.5 mm, 0.1–0.25 mm and <0.1 mm.

2.3. Phosphorus Sorption and Supply Characteristics

2.3.1. Soil Phosphorus Isothermal Sorption

Soil samples of 2.00 g were weighed in a 50 mL centrifuge tube, added 25 mL of 0.01 mg·L⁻¹ calcium chloride (CaCl₂) solution with P concentration (KH₂PO₄) of 0, 10, 20, 40, 80, 120, 150 and 200 mg·L⁻¹, and 3 drops of toluene were added to inhibit the activity of microorganisms [23]. After shaking at 25 °C and 220 r·min⁻¹ for 24 h, the solution was separated by centrifugation at 4000 r·min⁻¹ for 15 min, and the supernatant solution was aspirated and the concentration of P in the equilibrium solution was determined by means of the molybdenum antimony anti-colorimetric method. The adsorption data were fitted with two P adsorption models, Langmuir equation (monolayer adsorption) and Freundlich (multilayer inhomogeneous adsorption), to obtain the P sorption maximum (Q_m), the Langmuir adsorption coefficient (K_L), the Freundlich adsorption coefficient (K_F) and the adsorption constant (1/n). The higher the value of 1/n, the stronger the adsorption capacity. The adsorption constant (1/n) tended to be close to zero, indicating a stronger heterogeneity of the soil adsorption surface. An adsorption constant (1/n) between 0.1 and 0.5 indicates a strong adsorption capacity of the aggregates. The maximum buffer capacity (MBC) of soil P and the soil P sorption index (PSI) can be derived from the relevant parameters. Maximum buffer capacity reflects the sorption performance of the soil and provides a comprehensive assessment of the P supply capacity of the soil; the larger the value, the greater the P retention capacity of the soil [24]. The P sorption index can characterize the P retention capacity of the soil, as well as the P release capacity [25]. The equations are as follows [26]:

$${\mathrm{Langmuir} \mathrm{equation}: \mathrm{q}}_{\mathrm{e}}=\frac{{\mathrm{K}}_{\mathrm{L}}{\mathrm{Q}}_{\mathrm{m}}{\mathrm{C}}_{\mathrm{e}}}{1+\mathrm{K}{\mathrm{C}}_{\mathrm{e}}}$$

(1)

$${\mathrm{Freundlich} \mathrm{equation}: \mathrm{q}}_{\mathrm{e}}={\mathrm{K}}_{\mathrm{F}}{\mathrm{C}}_{\mathrm{e}}^{1/\mathrm{n}}$$

(2)

MBC = K_L × Q_m

(3)

$$\mathrm{PSI}=\frac{\mathrm{X}}{\mathrm{Log}{\mathrm{C}}_{\mathrm{e}}}$$

(4)

where: Q_m is the model theoretical P sorption maximum (mg kg⁻¹); q_e is total amount of P sorbed by the soil (mg kg⁻¹), C_e is the P concentration in the equilibrium solution (mg L⁻¹) (C_e in the PSI equation is the P equilibrium concentration at 200 mg L⁻¹ treatment); K_L is the Langmuir model coefficient; K_F is the Freundlich model coefficient; 1/n is the Freundlich model constant; X is the soil P sorption (mg kg⁻¹).

2.3.2. Degree of Phosphorus Saturation (DPS)

Degree of P saturation is widely used to assess the potential risk of soil P losses [24,27]. The readily bioavailable P is represented by CaCl₂ extracted P (P_CaCl2), the available P is represented by sodium bicarbonate (NaHCO₃) extracted P (P_Olsen) and Mehlich-3 solution extracted P (P_M3) [28]. Sodium citrate extracted P (P_citrate) represents organic acid activated P, while hydrochloric acid (HCl) extracted P (P_HCl) represents the maximum potential P pool for hydrogen proton activation [29]. A 1.00 g soil sample was mixed with 20 mL of 0.01 mol L⁻¹ CaCl₂, 0.01 mol L⁻¹ sodium citrate, 0.5 mol L⁻¹ NaHCO₃, and 1 mol L⁻¹ HCl as well as Mehlich-3 solutions (0.013 mol L⁻¹ HNO₃, 0.015 mol L⁻¹ NH₄F, 0.25 mol L⁻¹ NH₄NO₃, 0.2 mol L⁻¹ HOAc, 0.001 mol L⁻¹ EDTA), respectively. The mixtures were shaken at 25 °C for 3 h at 220 r min⁻¹ (30 min for P_Olsen extraction with NaHCO₃ solution) and then the molybdenum–antimony anti-colorimetric method was utilized to determine P_CaCl2, P_citrate, P_HCl, P_Olsen, and P_M3. The Al (Al_M3) and Fe (Fe_M3) in Mehlich-3 solutions were determined using a spectrophotometer (722 N, Youke, Shanghai, China) and an atomic absorption spectrometer (AA-6300, PerkinElmer, USA), respectively. The P activation index (PAC) reflects the potential ability of conversion of soil total P (TP) to available P (P_Olsen); the higher its value the greater the conversion capacity to available P [30]. It is calculated as follows [24]:

$${\mathrm{D}\mathrm{P}\mathrm{S}}_{\mathrm{M}3} \left(\%\right)=\frac{{\mathrm{P}}_{\mathrm{M}3}}{({\mathrm{Al}}_{\mathrm{M}3}+{\mathrm{Fe}}_{\mathrm{M}3})}\times 100$$

(5)

$${\mathrm{DPS}}_{\mathrm{Olsen}}(\%)=\frac{{\mathrm{P}}_{\mathrm{Olsen}}}{({\mathrm{P}}_{\mathrm{Olsen}}+{\mathrm{Q}}_{\mathrm{m}})}\times 100$$

(6)

$${\mathrm{DPS}}_{\mathrm{CaCl}2}(\%)=\frac{{\mathrm{P}}_{\mathrm{CaCl}2}}{({\mathrm{P}}_{\mathrm{CaCl}2}+{\mathrm{Q}}_{\mathrm{m}})}\times 100$$

(7)

$${\mathrm{DPS}}_{\mathrm{HCl}}(\%)=\frac{{\mathrm{P}}_{\mathrm{HCl}}}{({\mathrm{P}}_{\mathrm{HCl}}+{\mathrm{Q}}_{\mathrm{m}})}\times 100$$

(8)

$${\mathrm{DPS}}_{\mathrm{citrate}}(\%)=\frac{{\mathrm{P}}_{\mathrm{citrate}}}{({\mathrm{P}}_{\mathrm{citrate}}+{\mathrm{Q}}_{\mathrm{m}})}\times 100$$

(9)

$$\mathrm{PAC}(\%)=\frac{{\mathrm{P}}_{\mathrm{Olsen}}}{\mathrm{TP}\times 1000}\times 100$$

(10)

2.4. Determination of Soil Physical and Chemical Properties

Soil pH was determined using a PHS-25 pH meter with a water to soil ratio of 2.5:1. Total P was extracted using mixed sulphuric–perchloric acid digestion and determined using a continuous flow analyzer (Auto Analyzer III, Seal, Norderstedt, Germany). Total organic carbon (TOC) and total nitrogen (TN) were determined by means of an element analyzer (vario MICRO cube, Elementar, Langenselbold, Germany). The Fe and Al forms in the soil samples were determined by referring to the classic determination method [31]. The physical and chemical properties of the soil are shown in

Table 1. Physical and chemical properties (mean ± standard error) in the aggregates of red soil (RS) and limestone soil (LS) in two subtropical forests in southern China.

Soil Types		Physical and Chemical Properties
		pH	TOC (g kg−1)	TN (g kg−1)	TP (g kg−1)	C/N	C/P	N/P
RS-O/A	1–2 mm	4.58 ± 0.10 Ba	37.00 ± 3.14 ABa	2.43 ± 0.13 Aa	0.357 ± 0.020 BCa	15.203 ± 0.466 Ab	105.36 ± 15.35 Aa	6.908 ± 0.805 Aa
	0.5–1 mm	4.34 ± 0.37 Ba	34.97 ± 4.57 Aa	2.33 ± 0.29 Aa	0.381 ± 0.026 ABa	15.016 ± 0.164 Ab	94.61 ± 19.81 Aa	6.306 ± 1.267 Aa
	0.25–0.5 mm	4.49 ± 0.29 Ba	30.37 ± 3.51 ABa	1.93 ± 0.19 BCab	0.377 ± 0.025 Aa	15.556 ± 0.345 Ab	82.66 ± 15.87 Aa	5.248 ± 0.902 Aa
	0.106–0.25 mm	4.71 ± 0.08 Ca	27.30 ± 3.65 Aa	1.60 ± 0.21 BCb	0.330 ± 0.024 BCa	16.931 ± 0.417 Aa	85.62 ± 18.81 Aa	5.011 ± 1.068 Aa
RS-AB	1–2 mm	4.78 ± 0.02 Ba	11.03 ± 0.58 Ca	1.10 ± 0.06 Ba	0.287 ± 0.003 Ca	10.047 ± 0.210 Dc	38.44 ± 1.79 Ba	3.835 ± 0.197 Aa
	0.5–1 mm	4.82 ± 0.04 Ba	10.97 ± 0.74 Ba	1.03 ± 0.09 Ba	0.318 ± 0.033 Ba	10.905 ± 0.161 Db	35.22 ± 3.95 Ba	3.319 ± 0.403 Aa
	0.25–0.5 mm	4.81 ± 0.03 Ba	12.50 ± 0.06 Ca	1.03 ± 0.03 Ca	0.266 ± 0.032 Aa	12.018 ± 0.250 Ca	48.27 ± 5.55 Aa	3.959 ± 0.332 Aa
	0.106–0.25 mm	4.85 ± 0.03 Ca	9.07 ± 0.32 Bb	0.73 ± 0.03 Cb	0.263 ± 0.034 Ca	12.608 ± 0.184 BCa	36.25 ± 6.79 Aa	2.941 ± 0.585 Aa
LS-O/A	1–2 mm	6.50 ± 0.14 Aa	42.70 ± 4.98 Aa	3.27 ± 0.44 Aa	0.687 ± 0.069 Aa	13.173 ± 0.198 Ba	62.83 ± 8.24 Ba	4.797 ± 0.682 Aa
	0.5–1 mm	6.53 ± 0.18 Aa	40.83 ± 5.22 Aa	3.10 ± 0.42 Aa	0.642 ± 0.050 Aa	13.123 ± 0.106 Ba	63.47 ± 6.63 ABa	4.821 ± 0.560 Aa
	0.25–0.5 mm	6.56 ± 0.19 Aa	43.47 ± 4.27 Aa	3.30 ± 0.36 Aa	0.469 ± 0.242 Aa	13.168 ± 0.263 Ba	69.71 ± 9.80 Aa	5.352 ± 0.846 Aa
	0.106–0.25 mm	6.32 ± 0.00 Aa	41.50 ± 4.56 Aa	3.17 ± 0.39 Aa	0.657 ± 0.069 Aa	12.985 ± 0.194 Ba	63.83 ± 7.65 Aa	4.870 ± 0.661 Aa
LS-AB	1–2 mm	6.28 ± 0.47 Aa	25.47 ± 5.95 Ba	2.10 ± 0.52 ABa	0.586 ± 0.132 ABa	12.196 ± 0.208 Ca	48.92 ± 18.51 Ba	4.027 ± 1.567 Aa
	0.5–1 mm	6.38 ± 0.42 Aa	25.20 ± 6.03 ABa	2.13 ± 0.52 ABa	0.578 ± 0.140 ABa	11.850 ± 0.119 Ca	48.92 ± 18.21 ABa	4.121 ± 1.534 Aa
	0.25–0.5 mm	6.30 ± 0.54 Aa	26.37 ± 6.82 BCa	2.23 ± 0.55 ABa	0.580 ± 0.141 Aa	11.547 ± 0.256 Ca	51.39 ± 20.24 Aa	4.341 ± 1.648 Aa
	0.106–0.25 mm	5.64 ± 0.00 Ba	26.40 ± 6.79 Aa	2.27 ± 0.52 ABa	0.561 ± 0.131 ABa	11.662 ± 0.354 Ca	52.88 ± 20.89 Aa	4.495 ± 1.624 Aa

Note: (mean ± standard error) Different lowercase letters illustrate significant differences among different aggregates in the same soil horizon, while different uppercase letters illustrate significant differences among different soils and horizons in the same aggregate. Aggregates <0.1 mm were not determined due to sample absence. TOC: total organic carbon, TN: total nitrogen, TP: total phosphorus.

. The speciation of soil Fe and Al in the aggregates are shown in

Table 2. Speciation of soil Fe and Al (mean ± standard error) in the aggregates of red soil (RS) and limestone soil (LS) in two subtropical forests in southern China.

Soil Types		Contents of Fe and Al Forms (g kg−1)
		Fep	Feo	Fed	FeT	Alp	Alo	Ald	AlT
RS-O/A	1–2 mm	1.26 ± 0.23 Aa	2.99 ± 0.40 Aa	17.61 ± 1.27 Ba	30.31 ± 2.87 Ba	0.54 ± 0.08 Aa	4.08 ± 0.42 Ca	21.11 ± 1.40 Aa	24.32 ± 1.15 Ba
	0.5–1 mm	1.52 ± 0.20 Aa	3.36 ± 0.20 Aa	20.18 ± 0.67 Ba	19.43 ± 0.00 Bb	0.63 ± 0.08 Aa	4.84 ± 1.22 Ba	20.96 ± 4.39 Aa	11.39 ± 10.73 Ba
	0.25–0.5 mm	1.63 ± 0.20 Aa	3.22 ± 0.47 Aa	17.74 ± 0.98 Ba	18.89 ± 1.03 Bb	0.42 ± 0.11 Aa	3.59 ± 0.63 Ba	19.82 ± 0.70 Ba	6.81 ± 1.64 Ba
	0.1–0.25 mm	1.45 ± 0.17 Aa	2.95 ± 0.31 ABa	13.96 ± 0.63 Bb	8.97 ± 0.95 Bc	0.50 ± 0.07 ABa	3.44 ± 0.34 Ca	20.90 ± 0.35 Aa	16.21 ± 0.84 Ba
RS-AB	1–2 mm	1.15 ± 0.17 Ab	2.5 ± 0.32 Aab	20.09 ± 0.70 Ba	22.98 ± 0.63 Ba	0.37 ± 0.11 Aa	4.50 ± 0.88 Ca	31.2 ± 10.38 Aa	37.16 ± 0.00 Aa
	0.5–1 mm	1.22 ± 0.06 Aab	2.85 ± 0.28 ABab	19.03 ± 0.72 Ba	23.32 ± 0.00 Ba	0.47 ± 0.02 Aa	4.38 ± 0.39 Ba	25.84 ± 4.51 Aa	23.48 ± 0.00 Bab
	0.25–0.5 mm	1.50 ± 0.08 Aa	3.22 ± 0.23 Aa	19.47 ± 0.67 Ba	20.77 ± 1.31 Ba	0.53 ± 0.03 Aa	5.29 ± 0.54 Ba	23.19 ± 6.23 ABa	28.3 ± 8.33 ABab
	0.1–0.25 mm	1.20 ± 0.02 Aab	2.29 ± 0.12 Bb	15.67 ± 0.71 Bb	11.82 ± 1.30 Bb	0.43 ± 0.06 ABa	4.25 ± 0.52 Ca	25.40 ± 8.40 Aa	16.33 ± 2.94 Bb
LS-O/A	1–2 mm	0.38 ± 0.08 Ba	2.70 ± 0.21 Aa	50.5 ± 4.94 Aa	67.07 ± 7.44 Aa	0.49 ± 0.02 Aa	7.06 ± 0.37 Ba	33.63 ± 4.95 Aa	5.32 ± 2.38 Cb
	0.5–1 mm	0.49 ± 0.14 Ba	2.67 ± 0.25 ABa	51.84 ± 3.42 Aa	59.83 ± 4.51 Aa	0.45 ± 0.08 Aa	8.48 ± 1.28 Aa	28.03 ± 3.73 Aa	64.63 ± 4.86 Aa
	0.25–0.5 mm	0.41 ± 0.11 Ba	3.40 ± 0.31 Aa	50.71 ± 6.21 Aa	64.39 ± 7.67 Aa	0.51 ± 0.06 Aa	9.62 ± 2.94 Aa	46.18 ± 10.57 Aa	50.69 ± 14.76 Aa
	0.1–0.25 mm	0.39 ± 0.09 Ba	3.28 ± 0.31 Aa	43.03 ± 6.32 Aa	54.93 ± 9.83 ABa	0.60 ± 0.06 Aa	8.88 ± 1.31 Ba	32.49 ± 3.30 Aa	73.35 ± 11.3 ABa
LS-AB	1–2 mm	0.35 ± 0.11 Ba	2.49 ± 0.13 Aa	50.33 ± 5.46 Aa	54.11 ± 6.66 Aa	0.53 ± 0.06 Aa	9.59 ± 2.10 Aa	40.45 ± 8.54 Aa	27.98 ± 5.24 ABb
	0.5–1 mm	0.38 ± 0.10 Ba	2.46 ± 0.27 Ba	50.33 ± 6.37 Aa	53.47 ± 3.95 Aa	0.31 ± 0.18 Aa	10.49 ± 1.87 Aa	30.63 ± 4.73 Aa	61.00 ± 5.72 Aab
	0.25–0.5 mm	0.60 ± 0.26 Ba	2.80 ± 0.42 Aa	50.70 ± 8.13 Aa	62.37 ± 7.69 Aa	0.50 ± 0.08 Aa	10.18 ± 1.62 Aa	31.97 ± 7.19 ABa	56.14 ± 8.71 Aab
	0.1–0.25 mm	0.33 ± 0.09 Ba	2.89 ± 0.22 ABa	47.59 ± 6.33 Aa	82.91 ± 28.95 Aa	0.26 ± 0.14 Ba	11.30 ± 1.62 Aa	40.97 ± 17.78 Aa	104.58 ± 34.4 Aa

Note: Different lowercase letters illustrate significant differences among different aggregates in the same soil horizon, while different uppercase letters illustrate significant differences among different soils and horizons in the same aggregate. Fe_p and Al_p: complexed Fe and Al, Fe_o and Al_o: amorphous Fe oxide and amorphous Al oxide. Fe_d and Al_d: free Fe oxide and free Al, Fe_T and Al_T are total Fe and total Al. Aggregates <0.1 mm were not determined due to sample insufficient.

.

2.5. Data Statistics and Analysis

IBM SPSS Statistics 26.0 software (IBM SPSS Corp., Chicago, IL, USA) was used for data analysis. One-way ANOVA (one-way analysis of variance) and general linear models were applied to compare differences between different soil horizons and different aggregates in the red soil (RS) and limestone soil (LS), with Duncan’s method testing for significance of differences (p < 0.05). Figures were drawn using Origin 2019 and the R language (4.0.4) ggplot 2, vegan and rdacca.hp packages [32]. Data in the table are mean ± standard error, n = 3.

3. Results

3.1. Phosphorus Adsorption to Soil Aggregates

The P adsorption pattern for macroaggregates was mainly multi-layered and inhomogeneous, whereas the P adsorption pattern for microaggregates was mostly mono-layered (

Table 3. The parameters (mean ± standard error) of isothermal adsorption of aggregates in the O/A and AB horizons of red soil (RS) and limestone soil (LS) in two subtropical forests in southern China.

Soil Types		Langmuir Equation			Freundlich Equation			Adsorption Types
		KL	Qm (mg kg−1)	R2	KF	1/n	R2
RS-O/A	1–2 mm	0.041 ± 0.010 Ab	783.58 ± 69.74 Aab	0.943 ± 0.005	95.24 ± 12.98 Bb	0.402 ± 0.037 Ab	0.948 ± 0.007	multilayered and inhomogeneous
	0.5–1 mm	0.044 ± 0.003 Ab	1027.76 ± 72.73 Aa	0.961 ± 0.006	119.15 ± 9.09 Bb	0.428 ± 0.028 Ab	0.942 ± 0.019	monolayer
	0.25–0.5 mm	0.124 ± 0.041 Ba	878.15 ± 30.71 aAb	0.948 ± 0.020	219.2 ± 27.12 Aa	0.285 ± 0.028 Ab	0.861 ± 0.021	monolayer
	0.1–0.25 mm	0.013 ± 0.007 Bb	1137.89 ± 312.93 Aa	0.948 ± 0.020	31.15 ± 8.82 Bc	0.612 ± 0.088 Aa	0.963 ± 0.005	multilayered and inhomogeneous
	<0.1 mm	0.079 ± 0.019 Aab	467.08 ± 77.04 Bb	0.985 ± 0.001	90.08 ± 5.96 Cb	0.324 ± 0.033 Ab	0.922 ± 0.022	monolayer
RS-AB	1–2 mm	0.052 ± 0.01 Abc	667.92 ± 86.52 Abc	0.917 ± 0.006	102.08 ± 3.88 Bbc	0.359 ± 0.02 Abc	0.937 ± 0.014	multilayered and inhomogeneous
	0.5–1 mm	0.043 ± 0.006 Abc	1030.38 ± 133.26 Aa	0.928 ± 0.021	112.70 ± 18.56 Bab	0.438 ± 0.028 Aab	0.932 ± 0.023	multilayered and inhomogeneous
	0.25–0.5 mm	0.114 ± 0.02 B3 a	961.63 ± 88.48 Aab	0.890 ± 0.039	215.71 ± 12.7 A6 c	0.277 ± 0.018 Ac	0.867 ± 0.033	monolayer
	0.1–0.25 mm	0.020 ± 0.00 B6 c	933.81 ± 132.37 Aab	0.890 ± 0.039	54.73 ± 12.71 Ba	0.518 ± 0.060 ABa	0.977 ± 0.006	multilayered and inhomogeneous
	<0.1 mm	0.091 ± 0.03 Aab	408.37 ± 51.58 Bc	0.931 ± 0.022	81.01 ± 5.52 Cc	0.324 ± 0.030 Ac	0.930 ± 0.013	monolayer
LS-O/A	1–2 mm	0.103 ± 0.013 Ab	889.03 ± 94.31 Aab	0.900 ± 0.006	201.99 ± 15.6 Aab	0.307 ± 0.012 Aab	0.901 ± 0.009	multilayered and inhomogeneous
	0.5–1 mm	0.065 ± 0.014 Ab	1072.72 ± 241.25 Aa	0.931 ± 0.008	137.94 ± 11.07 ABa	0.410 ± 0.062 Aa	0.917 ± 0.027	monolayer
	0.25–0.5 mm	0.366 ± 0.122 Aa	424.98 ± 213.06 Bb	0.829 ± 0.027	202.41 ± 21.51 Ab	0.270 ± 0.033 Ab	0.823 ± 0.031	monolayer
	0.1–0.25 mm	0.058 ± 0.015 Ab	913.84 ± 36.01 Aab	0.902 ± 0.009	153.01 ± 15.9 Aab	0.354 ± 0.021 BCab	0.948 ± 0.017	multilayered and inhomogeneous
	<0.1 mm	0.079 ± 0.006 Ab	891.03 ± 82.57 Aab	0.930 ± 0.020	179.97 ± 16.76 Bab	0.319 ± 0.004 Aab	0.878 ± 0.010	monolayer
LS-AB	1–2 mm	0.146 ± 0.06 Aab	926.26 ± 156.58 Aa	0.892 ± 0.008	204.98 ± 8.99 Ab	0.314 ± 0.039 Ab	0.915 ± 0.024	multilayered and inhomogeneous
	0.5–1 mm	0.088 ± 0.02 A4 b	1228.75 ± 268.87 Aa	0.940 ± 0.031	169.30 ± 1.44 Aa	0.409 ± 0.045 Aa	0.941 ± 0.028	multilayered and inhomogeneous
	0.25–0.5 mm	0.235 ± 0.033 ABa	839.79 ± 74.32 Aa	0.910 ± 0.038	235.37 ± 21.24 Ab	0.279 ± 0.009 Ab	0.872 ± 0.007	monolayer
	0.1–0.25 mm	0.081 ± 0.003 Ab	801.65 ± 107.44 Aa	0.896 ± 0.032	168.98 ± 13.59 Ab	0.314 ± 0.011 Cb	0.945 ± 0.008	multilayered and inhomogeneous
	<0.1 mm	0.104 ± 0.013 Ab	946.57 ± 24.92 Aa	0.940 ± 0.007	214.71 ± 6.45 Ab	0.301 ± 0.009 Ab	0.876 ± 0.015	monolayer

Note: Different lowercase letters illustrate significant differences among different aggregates in the same soil horizon, while different uppercase letters illustrate significant differences among different soils and horizons in the same aggregate. For Langmuir equation, K_L: Langmuir model coefficient, Q_m: P sorption maximum, R²: Langmuir correlation coefficient; For Freundlich equation, K_F: Freundlich model coefficient, 1/n: Freundlich model constant, R²: Freundlich correlation coefficient.

). The Langmuir correlation coefficient was R² > 0.829 and the Freundlich correlation coefficient was R² > 0.823. The Freundlich model fitted better than the Langmuir model for the aggregates of 1–2 mm, 0.1–0.25 mm and 0.5–1 mm in the AB horizon of both forest soils. The Langmuir and Freundlich isothermal P adsorption of all sizes of aggregates had a similar pattern (Figure 1 and Figure 2). At the beginning of the adsorption, the slope of the soil aggregate P adsorption curve was greater and the adsorption rate was faster. As the adsorption proceeded, the adsorption sites gradually decreased, and the adsorption rate decreased and reached a dynamic equilibrium with the P content in solution. Phosphorus sorption maximum (Q_m) was greater for limestone soil aggregates >0.5 mm than for red soils (Table 3). Overall, the Q_m of the aggregates of 0.5–1 mm was higher in both soil aggregates than in the other aggregates. The Q_m of the <0.1 mm aggregates in the red soils was significantly lower than in the other aggregates. The trends in the K_L and K_F were similar (Table 3), suggesting a binding energy (K_L and K_F) for P was slightly higher in the limestone soils than in the red soils. The K_L and K_F of 0.25–0.5 mm aggregates were higher and those of 0.1–0.25 mm aggregates. The 1/n values in all aggregates varied between 0.01 and 0.31, indicating that all sizes of aggregates had a strong P adsorption capacity. Soil type, soil horizon and soil aggregates significantly influenced MBC and PSI (Figure 3). The MBC of limestone soils was higher than those of red soils. Among all aggregates, the MBCs of 0.25–0.5 mm aggregates were significantly higher than those of other aggregates (Figure 3a). Limestone soils generally had higher PSI values than red soils, and the AB horizon had higher PSI values than the O/A horizon. The PSI of 0.5–1 mm and 0.25–0.5 mm aggregates in the red soil were significantly higher than those of other aggregates < 0.1 mm, indicating red soils had limited P sorption and retention capacity. Limestone soils had higher PSI for 1–2 mm and 0.5–1 mm aggregates (Figure 3b).

3.2. Phosphorus Fractions in Soil Aggregates

The readily bioavailable fraction of P (P_CaCl2) and available P (P_Olsen and P_M3), the organic complexed P fraction (P_citrate) and PAC were higher in the red soil aggregates than in the limestone soil aggregates (

Table 4. The contents of phosphorus forms (mean ± standard error) of aggregates in red soil (RS) and limestone soil (LS) in two subtropical forests in southern China.

Soil Types		P Fractions (mg kg−1)						PAC (%)
		PCaCl2	POlsen	PM3	Pcitrate	PHCl	TP
RS-O/A	1–2 mm	1.17 ± 0.10 Aa	5.33 ± 0.74 Aab	7.57 ± 3.31 Aa	7.02 ± 1.27 Aa	2.46 ± 1.16 Aa	356.93 ± 19.99 BCa	1.53 ± 0.31 Aa
	0.5–1 mm	1.15 ± 0.15 Aabc	3.64 ± 0.05 Ab	4.37 ± 2.18 Aa	7.63 ± 1.82 ABa	2.34 ± 1.08 ABa	380.53 ± 26.14 ABa	0.97 ± 0.06 Aa
	0.25–0.5 mm	0.95 ± 0.02 Abc	4.38 ± 0.28 Ab	4.98 ± 2.10 Aa	5.77 ± 1.25 Aa	5.72 ± 1.30 Aa	377.11 ± 25.33 Aa	1.18 ± 0.14 Aa
	0.1–0.25 mm	0.86 ± 0.00 Ac	4.14 ± 0.03 ABb	4.93 ± 1.56 Aa	3.64 ± 0.82 Aa	4.34 ± 1.28 Aa	329.57 ± 24.43 BCa	1.27 ± 0.11 Aa
	<0.1 mm	1.25 ± 0.18 Aab	9.65 ± 3.63 Aa	5.84 ± 2.94 Aa	4.49 ± 1.02 Aa	5.47 ± 1.87 ABa	298.16 ± 24.79 Ba	3.51 ± 1.64 Aa
RS-AB	1–2 mm	0.86 ± 0.11 BCa	2.06 ± 0.48 Bb	2.39 ± 1.45 aABb	7.41 ± 1.95 Aa	5.91 ± 3.52 Aa	286.87 ± 3.24 Ca	0.71 ± 0.16 Bab
	0.5–1 mm	0.62 ± 0.04 Cb	1.61 ± 0.06 Bb	3.10 ± 2.19 Ab	8.19 ± 2.25 Aa	0.65 ± 0.13 Bb	317.68 ± 33.05 Ba	0.51 ± 0.05 Bb
	0.25–0.5 mm	0.63 ± 0.03 Cb	1.93 ± 0.14 BCb	1.60 ± 0.38 ABab	5.82 ± 1.52 Aa	2.56 ± 0.45 Aab	266.27 ± 32.03 Aa	0.74 ± 0.08 Aab
	0.1–0.25 mm	0.62 ± 0.03 Bb	2.15 ± 0.34 Cab	4.03 ± 2.20 Aa	2.95 ± 1.29 Aa	1.84 ± 0.46 Aab	262.97 ± 34.43 Ca	0.81 ± 0.04 Bab
	<0.1 mm	0.67 ± 0.02 Bb	2.70 ± 0.35 Ba	3.41 ± 1.52 Bab	4.11 ± 1.56 Aa	2.14 ± 0.56 Bab	231.80 ± 36.21 Ba	1.25 ± 0.32 ABa
LS-O/A	1–2 mm	1.09 ± 0.07 ABa	2.23 ± 0.22 Bb	0.74 ± 0.43 Bb	5.09 ± 0.64 ABa	1.61 ± 0.17 Ac	687.03 ± 68.93 Aa	0.33 ± 0.05 Bb
	0.5–1 mm	0.91 ± 0.03 ABb	3.74 ± 0.57 Aa	1.08 ± 0.89 Aab	5.57 ± 0.58 ABa	4.08 ± 0.98 Abc	642.40 ± 49.68 Aa	0.60 ± 0.13 Ba
	0.25–0.5 mm	0.84 ± 0.02 Bb	2.44 ± 0.24 Bb	0.86 ± 0.56 Bab	5.30 ± 0.52 Aa	6.48 ± 1.21 Ab	468.69 ± 242.38 Aa	0.32 ± 0.02 Ab
	0.1–0.25 mm	0.83 ± 0.02 Ab	4.66 ± 0.27 Aa	0.99 ± 0.66 Aab	3.47 ± 0.43 Ab	4.41 ± 0.54 Abc	657.28 ± 69.14 Aa	0.72 ± 0.07 Ba
	<0.1 mm	0.90 ± 0.07 Bb	3.58 ± 0.51 Bab	2.07 ± 0.58 Ba	5.29 ± 0.22 Aa	9.18 ± 1.65 Aa	650.33 ± 69.80 Aa	0.55 ± 0.05 Bab
LS-AB	1–2 mm	0.72 ± 0.07 Ca	1.51 ± 0.08 Bb	0.33 ± 0.04 Bb	2.69 ± 0.67 Aa	5.33 ± 2.29 Aa	585.76 ± 131.71 ABa	0.29 ± 0.06 Ba
	0.5–1 mm	0.64 ± 0.04 BCa	2.04 ± 0.30 Bb	0.19 ± 0.03 Ab	2.83 ± 0.19 Ba	0.89 ± 0.12 Bb	578.16 ± 139.79 ABa	0.41 ± 0.13 Ba
	0.25–0.5 mm	0.62 ± 0.02 Ca	1.67 ± 0.08 Cb	0.28 ± 0.01 Bb	2.34 ± 0.26 Aa	3.98 ± 1.80 Aab	579.59 ± 140.97 Aa	0.33 ± 0.08 Aa
	0.1–0.25 mm	0.59 ± 0.02 Ba	3.12 ± 0.64 BCa	0.35 ± 0.08 Ab	2.14 ± 0.25 Aa	2.12 ± 0.32 Aab	560.74 ± 130.80 ABa	0.60 ± 0.17 Ba
	<0.1 mm	0.68 ± 0.06 Ba	1.84 ± 0.33 Bb	2.42 ± 0.77 Ba	2.05 ± 0.16 Aa	2.38 ± 0.26 Bab	568.02 ± 141.41 Aa	0.33 ± 0.03 Ba

Note: Different lowercase letters illustrate significant differences among different aggregates in the same soil horizon, while different uppercase letters illustrate significant differences among different soils and horizons in the same aggregate.

). Limestone soils had higher P_HCl and TP. Soil type, soil horizons, soil aggregates and their interactions significantly influenced P_CaCl2 and P_Olsen (

Table 5. Results (p values) of general linear model for the effects of soil horizon, aggregates and their interaction on phosphorus content of different forest soils.

	PCaCl2	POlsen	PM3	Pcitrate	PHCl	TP	PAC
Soil horizon (H)	<0.01	<0.01	<0.01	<0.01	0.423	0.884	0.418
Aggregate (A)	<0.01	<0.01	0.070	<0.01	<0.01	<0.01	0.406
L×A	<0.01	<0.01	0.561	0.411	0.090	0.990	0.468

Note: L, soil and soil horizons. A, aggregates. Bold indicates that the corresponding p value reached a significant level (p < 0.01).

). The distribution of P fractions was similar in different soil horizons, with a decreasing trend with increasing soil depth. The PAC of soil O/A horizon was higher than that of AB horizon. The concentration of P fractions varied considerably between the different sizes of aggregates. P_citrate was more likely to be enriched in the macroaggregates. In limestone soils, the distribution of P fractions was more homogeneous across the aggregates. P_M3 tended to be enriched in the microaggregates. In the O/A horizon of both soils, P_HCl tended to be enriched in the microaggregates, while it tended to be highest in the 1–2 mm aggregates and lowest in the 0.5–1 mm aggregates of the AB horizon. The PAC of microaggregates was higher than that of macroaggregates in the red soil.

3.3. Degree of Phosphorus Saturation (DPS) in Soil Aggregates

Soil type, soil horizons, and soil aggregates, and their interactions, significantly influenced DPS_CaCl2 and DPS_Olsen (Figure 4). The highest DPS values for the two forest types did not exceed 2.04%, indicating that they were at low risk of P leaching. DPS_M3 and DPS_citrate were significantly higher in red soils than in limestone soils, while the saturation of other P fractions did not differ significantly between the two soils. The O/A horizon’s DPS_CaCl2 and DPS_Olsen in red soils and DPS_CaCl2 and DPS_citrate in limestone soils was significantly higher than those of the AB horizon, while the saturation of other P fractions did not differ significantly between the different soil horizons. Aggregates significantly influenced the DPS values, except for DPS_M3. The DPS_CaCl2, DPS_Olsen and DPS_HCl were significantly higher in the red soil O/A horizon <0.1 mm aggregates than in the other aggregates, suggesting a greater risk of P leaching from microaggregates in red soils. The DPS_CaCl2, DPS_HCl and DPS_citrate were lower in 0.5–1 mm aggregates than in other aggregates in limestone soils, which were higher in 0.25–0.5 mm and <0.1 mm aggregates of the O/A horizon.

3.4. Correlation and Redundancy Analysis

The speciation of Fe and Al were found to significantly affect the P sorption and P fractions in both forest soils on an aggregate level (Figure 5 and Figure 6). P_CaCl2 had significant negative correlations with amorphous Al oxide and total Al. P_M3 and P_AC were significantly and positively correlated with C/N ratio and complexed Fe, and had highly significant negative correlations with free Fe oxide, total Fe, amorphous Al oxide, total Al, pH and total P. K_L and K_F were significantly and positively correlated with free Fe oxide, total Fe, amorphous Al oxide and pH. Soil P sorption index was significantly and positively correlated with free Fe oxide, amorphous Al oxide and TP. There was a highly significant positive correlation between MBC and free Fe oxide, total Fe and amorphous Al oxide. Furthermore, P_CaCl2 and P_Olsen both had significant positive correlations with TOC, TN, C/N ratio, C/P ratio and N/P ratio. P_Olsen had highly significant negative correlations with pH.

Total Fe and free Fe oxide had the most significant effects on soil P sorption, explaining 18.22% and 11.44% of the coefficient of variation, respectively (

Table 6. Interpretation rates of explanatory variables between environmental factors and phosphorus adsorption characteristics in redundancy analysis.

	Unique	Average Share	Individual	I Perc (%)
Fep	0.0482	−0.0187	0.0295	7.21
Feo	−0.0105	0.0169	0.0064	1.56
Fed	−0.0137	0.0605	0.0468	11.44
FeT	0.1027	−0.0282	0.0745	18.22
Alp	0.0199	0.0161	0.0360	8.80
Alo	0.0051	0.0330	0.0381	9.32
Ald	−0.0043	0.0086	0.0043	1.05
AlT	0.0573	−0.0290	0.0283	6.92
pH	−0.0074	0.0273	0.0199	4.87
TN	−0.0049	0.0124	0.0075	1.83
TOC	−0.0047	0.0074	0.0027	0.66
C/N	−0.0156	0.0304	0.0148	3.62
TP	−0.0141	0.0252	0.0111	2.71
C/P	0.0149	0.0297	0.0446	10.90
N/P	0.0149	0.0297	0.0446	10.90

Note: Unique is the proportion of the total variation explained separately for each explanatory variable; Average share is the division of the explanatory portion common to each explanatory variable and other explanatory variables; Individual is the proportion of each explanatory variable in the total variation; I perc (%) is the proportion of each explanatory variable to the total explanatory rate change.

). Positive correlations were found between free Fe and amorphous Al oxide and adsorption parameters. Complexed Fe showed a negative correlation with adsorption parameters. Amorphous Al oxide and free Fe oxide had the most significant effects on P, explaining 16.22% and 12.53% of the coefficient of variation, respectively (

Table 7. Interpretation rates of explanatory variables between environmental factors and different phosphorus forms in redundancy analysis.

	Unique	Average Share	Individual	I Perc (%)
Fep	−0.0066	0.0311	0.0245	7.66
Feo	−0.0058	0.0071	0.0013	0.41
Fed	−0.0192	0.0593	0.0401	12.53
FeT	−0.0019	0.0284	0.0265	8.28
Alp	−0.0109	0.0005	−0.0104	−3.25
Alo	0.0004	0.0515	0.0519	16.22
Ald	0.0098	0.0053	0.0151	4.72
AlT	0.0035	0.0178	0.0213	6.66
pH	−0.0052	0.0312	0.0260	8.12
TN	−0.0046	0.0206	0.0160	5.00
TOC	−0.0048	0.0174	0.0126	3.94
C/N	0.0027	0.0251	0.0278	8.69
TP	0.0489	0.0077	0.0566	17.69
C/P	−0.0129	0.0180	0.0051	1.59
N/P	−0.0128	0.0179	0.0051	1.59

Note: Unique is the proportion of the total variation explained separately for each explanatory variable; Average share is the division of the explanatory portion common to each explanatory variable and other explanatory variables; Individual is the proportion of each explanatory variable in the total variation; I perc (%) is the proportion of each explanatory variable to the total explanatory rate change.

). The remaining factors, explained in descending order, were C/N ratio, total Fe and pH.

4. Discussion

4.1. Speciation of Iron and Aluminum Influence on Phosphorus Sorption and Supply

Compared to temperate forest soils [33], subtropical soils are characterized by the presence of large amounts of Fe and Al oxides, which are among the important cementing agents that promote soil aggregation [34]. Furthermore, the contents of Fe and Al oxides affect soil P adsorbing ability [27,35]. Previous studies found that less crystalline free and amorphous Fe–Al oxides could increase the soil P sorption capacity [20]. In general, the less crystalline Fe–Al oxides have a larger specific surface area [36] and, therefore, have greater sorption capacity. Our study showed that free, total and amorphous Fe oxides increased the buffering capacity of soil P mainly by increasing the P sorption strength (K_L, K_F), rather than by providing more sorption sites (Q_m). In contrast, the more crystalline Fe served to reduce the sorption strength and buffering soil P capacity. The limestone soil aggregates had higher total Fe, free Fe oxide and amorphous Al content than the red soils, while the complexed Fe content was lower than in the red soils, resulting in a higher P buffering capacity and a higher P sorption index (PSI). In the aggregates, the free Fe oxide in both soils and the amorphous Al in the red soils also showed a tendency to decrease with decreasing aggregate sizes, which may be one of the reasons for the lower P sorption capacity in the microaggregates. In addition, the positive surface charge of soils and Fe–Al oxides decreases with increasing pH [37], with a consequent decrease in the sorption of the negatively charged PO₄³⁻ [38,39]. A previous study suggested that the sorption capacity of soil for PO₄³⁻ decreased gradually with increasing pH in the pH range of 4–7, but the overall decreasing trend was not significant [39]. The increase of pH reduced Q_m in our study, shedding light on such a mechanism. On the other hand, increased pH promotes soil Al oxide and hydride contents [40]. We found a higher Fe content in the limestone soil than in the red soil. Therefore, a higher pH in the limestone soil promoted both P sorption strength (K_L, K_F) and buffering capacity (MBC). Overall, the P sorption capacity of limestone soils was higher than that of red soils. Furthermore, the complex formed by the combination of organic matter and Fe–Al oxides facilitates soil P sorption. The complex released anions that compete with PO₄³⁻ for sorption sites and reduce the sorption capacity [41,42]. We only found a weak correlation between TOC, TN and K_L, K_F, MBC, suggesting that the impact of soil organic matter was relatively low.

Bioavailable P (P_Olsen, P_M3) tended to enrich in microaggregates. Firstly, microaggregates had lower Fe and Al oxides and, thus, had weaker ability to retain soil bioavailable P (lower MBC, PSI). Secondly, microaggregates have a highly active outer surface, providing an ideal habitat for microorganisms [43]. In order to meet their own needs, microorganisms secrete more extracellular phosphate hydrolases to activate P in the soil to meet their own growth requirements [44], which promotes the dissolution of insoluble P-containing compounds and increases the available P. In addition, P_M3 and P_citrate were negatively correlated with the amorphous Al oxides and free Fe–Al oxides. The adsorption of phosphate by amorphous and free Fe–Al oxides formed a cutan that encapsulated the phosphate surface, immobilizing the P in the soil as Fe–P and Al–P by Fe³⁺ and Al²⁺ [3,45,46]; thereby, reducing the available P. Available P may be converted to more stable P, resulting in lower available P in limestone soils than in red soils. In contrast, Fe complexation was significantly and positively correlated with P_M3 and P_citrate, suggesting that Fe complexation may act as an anchor and protector for both P forms. The complexed Fe retains some of the phosphate, preventing the soil from sorption and retention of P [47]. However, soil microorganisms can secrete specific enzymes to hydrolyze the insoluble iron phosphates to release P and increase P content [48,49], which is one mechanism explaining the greater P transformation capacity in red soils.

4.2. Phosphorus Sorption Capacity of Soil Aggregates

The P sorption capacity of the limestone soil aggregates was stronger than that of the red soil. The Q_m of the two forest soils were similar, indicating that they had similar P sorption sites. However, the K_L and K_F of the limestone soil aggregates were higher than those of the red soil, indicating that the limestone soil aggregates had higher P sorption strength. The PSI of the limestone soil aggregates was also generally higher than that of the red soil. The difference in sorption capacity between the O/A and AB horizons was not significant, while the sorption and retention capacity of P differed significantly between the different sized aggregates. Overall, the chemical bonds formed by monolayer sorption were more stable, resulting in higher sorption strengths (K_L). Correspondingly, the 0.25–0.5 mm aggregates of the two forest soils with predominantly monolayer sorption had stronger P buffering capacity (higher MBC) and played a critical role in P retention. Conversely, the 0.1–0.25 mm aggregates, with predominantly multilayer inhomogeneous sorption, had relatively lower MBCs and weak P buffering capacity. The results of the combined PSI and MBC indicate that the macroaggregates (>0.25 mm aggregates considered as water-stable aggregates [50]) had higher P buffering capacity compared to microaggregates [51,52].

4.3. Phosphorus Fractions of Soil Aggregates and Potential Leaching

In this study, red soils were more abundant in bioavailable P and had greater activating P potential. The results implied that red soil aggregates had greater capacities for P conversion and P supply. The limestone soils stored more of the less bioavailable calcium- and magnesium-bound P fractions. Limestone soils retained a larger pool of soil P, but at a lower capacity for P supply. Both of the forest soils had higher P fractions in the O/A horizon than in the AB horizon, except for P_citrate and P_HCl in the 1–2 mm aggregates of red soils. The surface soil had a large accumulation of plant litter and its decomposition products, leading to high organic matter content and soil microorganism abundance [53], and higher soil P levels.

Both subtropical forest soil aggregates in this study had lower P_Olsen and P_M3, indicating that highly weathered subtropical forest soils had lower available P and were susceptible to P limitation [54]. P_CaCl2 and P_Olsen increased with increasing TOC, TN, C/N ratio and C/P ratio, indicating that the conversion of soil available P was closely related to the mineralization of soil organic P, which is one of the important sources of inorganic P in forest soils [55]. Soil organic matter provides sufficient substrate for P mineralization. The soil C–N–P cycle coupling mechanism regulates P content, with organic acids and extracellular phosphate hydrolases secreted by plants and microorganisms to meet P limitation stress (high C/P relatively). The process promotes the mineralization of organic P in the soil [56], which, in turn, increases the bioavailable P [57]. In addition, organic matter composition and content may influence soil P bioavailability by affecting soil P sorption capacity [38]. In general, a DPS of 25% was used as a threshold value, above which soil P is more likely to be lost by runoff or water infiltration [58]. In our study, the DPS was less than 2.04% for both forest soil aggregates, indicating that the risk of soil leaching was low in the subtropical forests. Compared to limestone soils, red soils had a weaker P sorption strength, higher effective P content and a relatively higher DPS, with a higher risk of P leaching in the microaggregates.

5. Conclusions

In this study, we found that the speciation of Fe and Al in different soil types were the key factors controlling soil P sorption capacity and P fractions of subtropical forest soil aggregates. The higher free Fe oxide and amorphous Al oxide in limestone soils increased the strength of P sorption, resulting in higher P sorption and buffering capacity in limestone soils compared to acidic red soils. A larger pool of P is retained by limestone soils, while red soils have a greater capacity to supply bioavailable P. Macroaggregates (0.25–0.5 mm) play a critical role as a P retention buffer in subtropical forest soils. In contrast, the microaggregates (<0.25 mm) had weaker P sorption and buffering capacity and higher available P. Both soils had a low DPS for all sizes of aggregates and a low risk of P leaching, with red soil < 0.1 mm aggregates prone to P leaching. Consequently, our results illustrate the complexity of the P sorption and supply characteristics on an aggregates level. We conclude that P sorption capacity and potential leaching strongly depends on the speciation of Fe and Al in subtropical forest soils. Identification of the influencing factors and the mechanisms at work in forest soil aggregates may help to further reduce P leaching and improve P supply by means of sustainable development and management of subtropical forests.