Influence of Humic Acid and Gypsum on Phosphorus Dynamics and Rice Yield in an Acidic Paddy Soil of Thailand

Abstract

Managing phosphorus (P) in acidic paddy soils is crucial for sustaining rice yields. However, the effects of combined humic acid (HA) and flue gas desulfurization gypsum (FG), a by-product of coal-fired power plants, on P forms remain poorly understood. This study examined P forms using a sequential extraction procedure and XANES spectroscopy following the application of HA, FG, and HA + FG. HA increased organic labile P, while FG and HA + FG promoted HCl-extractable Pi and humic Po, respectively. XANES data revealed that P associated with aluminum (Al) (hydr)oxides was dominant in acidic paddy soils. Brushite (CaHPO₄·2(H₂O)) accounted for 25% and 19% of total P in the FG- and HA + FG-treated soil, respectively. Iron (Fe)-bound P was absent in control and FG-treated soils but was present as strengite (FePO₄·2H₂O) in HA- and HA + FG-treated soils (23% and 30% of the total P, respectively). Inositol hexakisphosphate (IHP), a non-labile Po, was in HA- and HA + FG-treated soil (12% and 31% of the total P, respectively). Archerite (KH₂PO₄) was 40% and 20% of the total P in HA- and HA + FG-treated soil, respectively. HA alone is an effective soil amendment that enhances P cycling and availability by increasing organic P mineralization, boosting rice yield in acidic paddy soil.

1. Introduction

The yield of rice cultivated in acidic paddy soils is limited primarily by low phosphorus (P) availability [1]. P availability in acidic soils is largely influenced by soil pH and minerals, such as aluminum (Al) and iron (Fe) (hydr)oxides, as well as the clay mineral kaolinite [2,3,4]. Under acidic conditions, P fixation by Al and Fe (hydr)oxides increases and accounts for approximately 15% and 55% of the total P, respectively. Meanwhile, P associated with calcium phosphate (Ca-P) minerals remains low due to its solubilization in acidic environments. However, as soil pH approaches neutrality or the soil becomes alkaline, P solubility declines, as it precipitates with Ca to form Ca-P minerals [2,5]. Reducing soil P fixation can promote soil-available P for higher P uptake, crop productivity, and P use efficiency (PUE) [6]. Therefore, identifying soil management strategies that effectively reduce P fixation by Al and Fe (hydr)oxides while maintaining soil P availability for crop uptake is essential for sustaining crop production under acidic paddy soils.

Various soil management strategies have been introduced and developed to alleviate P fixation in acidic paddy soils, particularly through soil amendments such as agricultural lime and organic materials, namely manures, biochar, and leaf litter [2,7,8,9]. The addition of organic soil amendments can enhance P release from Al-P complexes by altering soil pH and increasing organic matter content [2,3,9]. Furthermore, the long-term incorporation (>22 years) of organic materials elevates available P concentration in the soil by increasing organic P (Po), particularly non-labile Po (e.g., humic Po and residual Po), which can form organomineral complexes with soil mineral surfaces [3]. However, increasing non-labile Po through organic material application occurs slowly over an extended period. This limitation can be addressed by directly applying exogenous humic substance products with high humic acid (HA) contents to the soil [10].

HA is a major humic substance derived from the decomposition of soil organic materials. Its structure contains more than 70% aromatic functional groups, including carboxyl (R–COOH), phenolic hydroxyl (R–OH), and carbonyl (C=O), which help prevent P fixation by competing for adsorption sites on mineral surfaces [11,12]. The addition of HA not only increases the availability of inorganic P (Pi) in the labile form by transforming insoluble Al-P and Fe-P into more soluble forms but also enhances the mineralization of organic P by stimulating microbial activity [13]. Consequently, HA application elevates soil-available P concentrations, thereby improving P uptake and crop yield [14].

In addition to P deficiency, calcium (Ca) deficiency is another major constraint in acidic soils [15]. Adding Ca-rich soil amendments, such as gypsum (CaSO₄·2H₂O), supplies exogenous Ca to the soil. Flue gas desulfurization gypsum (FG), a by-product of coal combustion, contains approximately 20–38% Ca [14,16]. Generally, addition of FG alleviates sodium (Na) toxicity in salt-affected soils [17] and increases soil Ca concentrations, including total Ca, soluble Ca, and exchangeable Ca in acidic soils [14,16,18]. Furthermore, FG reduces soluble P loss through leaching and runoff by forming Ca-P complexes [19]. When applied to acidic soils, particularly under waterlogged conditions, other types of gypsum, such as wallboard gypsum, facilitate the transformation of P fractions into Ca-P minerals. The sequential fractionation results from a previous study revealed that incorporating wallboard gypsum into flood-prone agricultural fields resulted in HCl-extractable inorganic P fraction of 65% of the total P [20]. X-ray absorption near-edge structure (XANES) analysis further confirmed that Ca-P species accounted for 16% of the total P in Oxisols under phosphogypsum application [21]. Ca-P species are considered a stable form of P that is not readily available for plant uptake [22]. The application of phosphogypsum in an Ultisol soil in Brazil reduced maize yields by inducing magnesium deficiency [23]. Similarly, FG has been shown to decrease crop yields, particularly in rice cultivated in paddy soils in Thailand, due to the decreased plant P uptake [14].

The coapplication of FG with HA improves soil properties and crop yields (such as rice and rapeseed) in salt-affected soils by improving various soil parameters such as the sodium adsorption ratio, exchangeable sodium percentage, electrical conductivity, total nitrogen, total organic carbon (TOC), porosity, and water-stable aggregates [24,25,26]. However, the previous studies did not report the results for soil-available P. Our earlier study demonstrated that HA applied alone increased rice grain yield but reduced available P concentrations in acidic paddy soil because rice plants absorbed the available P. Conversely, although FG applied alone or coapplied with HA increased available P in acidic paddy soil, it also reduced the grain yield due to lower P uptake [14]. However, no prior studies have examined the effects of FG and HA coapplication on P dynamics in acidic paddy soils.

This study hypothesized that the combined application of HA and FG would improve soil properties, including TOC, exchangeable Ca, and exchangeable sulfur (S). However, higher TOC and exchangeable Ca may promote the stabilization of organic P into less soluble forms. In addition, the increase in exchangeable Ca induces Ca-P precipitation. This combined effect can suppress P availability, resulting in reduced rice yield despite improved soil fertility. The objectives of this study were to: (1) evaluate the effects of HA, FG alone and their combination on soil chemical properties and rice yield; (2) analyze P speciation using a sequential extraction procedure (SEP) and P K-edge XANES; and (3) determine the relationships between P species, soil properties, and rice yield in acidic paddy soil. This study advances P management strategies in acidic paddy soils by providing a comprehensive understanding of P dynamics, which is essential for preventing losses of applied P through rapid fixation by Fe and Al oxides. This maintains the applied P in a readily available form for plant uptake. As a result, maximum crop yield can be achieved.

2. Materials and Methods

2.1. Site Characteristics, Experimental Design, and Soil Sampling

The soil used in this study was collected from paddy fields in Khon Kaen, Thailand (UTM: 48Q 256,751 E, 1,827,796 N), at an elevation of 187 m above sea level. The study site had a tropical savanna climate, with average minimum and maximum temperatures of 21 °C and 32 °C, respectively. During rice cultivation, the cumulative rainfall reaches 412 mm [27]. The soil was classified as Aeric Kandiaquult according to Soil Taxonomy [28]. The initial soil used in this study was previously characterized by [14], who reported that it had a sandy loam texture and consisted of 583 g kg⁻¹ sand, 359 g kg⁻¹ silt and 58 g kg⁻¹ clay. The soil was acidic (pH H₂O 4.70) and contained low levels of TOC (1.56 g kg⁻¹) and cation exchange capacity (CEC) (2.58 cmol kg⁻¹). A study from [29] reviewed that TOC in paddy soil of Northeast Thailand ranged from 0.34 to 31.2 g kg⁻¹. The exchangeable Ca and S concentrations were 0.095 g kg⁻¹ and 0.037 g kg⁻¹, respectively. The total concentrations of various elements (g kg⁻¹) in the soil were as follows: Al, 1.02; Fe, 0.962; N, 0.577; Ca, 0.277; S, 0.144; K, 0.079; and P, 0.058 (

Table 1. Mean ± standard deviation (n = 3) of initial characteristics of the studied soil, humic acid (HA), and flue gas desulfurization gypsum (FG) used in this study [14].

Properties	Initial Soil	HA	FG
Sand (g kg−1) a	583	-	-
Silt (g kg−1) a	359	-	-
Clay (g kg−1) a	58	-	-
pH H2O a	4.70 ± 0.20	9.77 ± 0.06	7.67 ± 0.06
TOC (g kg−1) a	1.56 ± 0.08	291 ± 7.48	-
CEC (cmol kg−1) a	2.58 ± 0.16	56.6 ± 5.75	-
C/N	-	26.6 ± 0.13	-
Functional groups b	-	–COOH, –OH, and C=O	-
Exchangeable Ca (g kg−1) a	0.095 ± 0.02	-	-
Exchangeable S (g kg−1) a	0.037 ± 0.00	-	-
Al (g kg−1)	1.02 ± 0.03	20.3 ± 2.49	nd
Ca (g kg−1) a	0.277 ± 0.06	9.41 ± 0.88	383 ± 1.38
Fe (g kg−1)	0.962 ± 0.01	5.31 ± 1.02	0.158 ± 0.01
K (g kg−1)	0.079 ± 0.03	54.4 ± 8.99	0.057 ± 0.03
N (g kg−1)	0.577 ± 0.04	10.9 ± 0.33	0.123 ± 0.01
P (g kg−1) a	0.058 ± 0.00	0.062 ± 0.01	0.029 ± 0.00
S (g kg−1) a	0.144 ± 0.03	6.74 ± 0.12	199 ± 43.8

Abbreviations: TOC = total organic carbon, CEC = cation exchange capacity, nd = not detected. Elemental concentrations are reported as total content: Al = aluminum, Ca = calcium, Fe = iron, K = potassium, N = nitrogen, P = phosphorus, S = sulfur. Functional groups are carboxyl (–COOH), phenolic hydroxyl (–OH), and carbonyl (C=O). ^a,b Data were previously published in Hartina et al. [14] and Sukitprapanon et al. [12], respectively.

). The mineralogical composition of the bulk initial soil was dominated by quartz, with trace amounts of feldspars, muscovite, and kaolinite (

Table 2. Mineralogical composition of the bulk soil and flue gas desulfurization gypsum (FG) used in this study.

Sample	Gypsum	Quartz	Feldspars	Muscovite	Kaolinite
Initial soils	nd	xxxx	tr	tr	tr
FG	xxxx	nd	nd	nd	nd

Note: xxxx = dominant (>95%), tr = trace (<5%), and nd = not detected.

).

A greenhouse pot experiment using the collected paddy soil described above was conducted at the Soil and Fertilizer Research Station, Department of Soil Science and Environment, Faculty of Agriculture, Khon Kaen University, Thailand. The experiment followed a randomized complete block design with three replications. Four treatments were applied: (1) fertilizer alone without soil amendments (Control), (2) FG, (3) HA, and (4) HA combined with FG (HA + FG). All treatments, including the control and treatments (2 to 4), received a similar amount of fertilizer application to ensure uniform nutrient supply. The study used commercially sourced HA from leonardite, which was provided by Concurchem Co., Ltd., Bang Yai District, Nonthaburi, Thailand. HA was applied at the rate of 975 kg ha⁻¹, which was the optimal rate identified by [12]. Solid HA powder was uniformly incorporated into the soil before rice transplantation. The chemical properties of the HA used in this study were previously reported by [14], who indicated that it was strongly alkaline (pH H₂O 9.77) to enhance its solubility and had high amounts of TOC (291 g kg⁻¹), CEC (56.6 cmol kg⁻¹), and C/N ratio (26.6). Additionally, it contained N, 10.9 g kg⁻¹; P, 0.062 g kg⁻¹; K, 54.4 g kg⁻¹; Al, 20.3 g kg⁻¹; Ca, 9.41 g kg⁻¹; S, 6.74 g kg⁻¹; and Fe, 5.31 g kg⁻¹. Furthermore, the HA used in this study possessed functional groups such as carboxyl (–COOH), phenolic hydroxyl (–OH), and carbonyl (C=O) (Table 1). The FG applied in this study was received from Mae Moh Power Plant, Electricity Generating Authority of Thailand, Mae Moh District, Lampang Province, Thailand. The FG was applied at the rate of 636 kg ha⁻¹; this was determined based on the gypsum requirement (GR) for acidic soils as recommended by previous studies for the purpose of increasing soil Ca concentration [14,18]. For regional soil correction, the GR used in this study was determined using the clay content (106 g kg⁻¹) at a depth of 20–40 cm [12] to improve soil Ca concentration of subsoil layer. The GR was calculated by multiplying the clay content (g kg⁻¹) by 6 and dividing by 1000. The FG used in this study was neutral (pH H₂O 7.67) and had high total concentrations of Ca (383 g kg⁻¹) and S (199 g kg⁻¹) [14]. It also contained Fe, 0.158 g kg⁻¹; N, 0.123 g kg⁻¹; K, 0.057 g kg⁻¹; and P, 0.029 g kg⁻¹. The total Al concentration was not detected (nd) (Table 1). Additionally, gypsum (>95%) was the dominant mineral in the FG (Table 2).

Rice (Oryza sativa L.) seeds were sown in a plastic tray filled with the studied soil and incubated for 14 days. Seedlings with uniform growth were transplanted into pots (30 cm length × 30 cm width × 39 cm height) at a spacing of 20 cm. The soil was maintained under flooded conditions with 5 cm of water above the surface from transplanting until seven days before harvest. The total inputs of N, P₂O₅, and K₂O were 75, 38, and 38 kg ha⁻¹, respectively. The source of these major nutrients was chemical fertilizers, including urea (46:0:0), diammonium phosphate (18:46:0), and muriate of potash (0:0:60) at rates of 131, 81, and 63 kg ha⁻¹, respectively. Rice cultivation practices, including seed selection, seedling management, transplanting, weeding, pest and disease control, and harvesting, followed the recommendations of the Thai Department of Agriculture (DOA) [30]. Grain yield and straw biomass from each pot were recorded at harvest (128 days after planting, DAP) and converted to t ha⁻¹.

After the rice harvest at 128 DAP, soil samples were collected from each pot and divided into two subsamples: air-dried and field-moist soil. Air-dried soil samples were ground and sieved through a 2 mm mesh for chemical analysis. Field-moist soil samples were packed in plastic containers, immediately cooled to 4 °C to minimize microbial activity, and subjected to P sequential extraction (SEP) and P K-edge XANES spectroscopy.

2.2. Chemical Analyses

The pH (H₂O) of the initial soil and soil amendments (HA and FG) was determined using a 1:5 soil/soil amendment-to-water ratio [31]. Soil pH and redox potential (Eh) were measured at the harvesting stage (128 DAP) under field conditions using a pH meter (HI98103, Hanna Instruments, Woonsocket, RI, USA) and an oxidation–reduction potential (ORP) meter (HI 8424, Hanna Instruments, Woonsocket, RI, USA). These were measured by inserting the probe of the respective instruments into the soil at 5 cm depth. The particle size distribution of the initial soil was analyzed using the pipette method [32]. TOC and total concentration of N were determined by dry combustion with a CN analyzer (Multi N/C 2100s, Analytik Jena GmbH, Jena, Germany). Exchangeable Ca and CEC were measured using 1 mol L⁻¹ NH₄OAc at pH 7 as the extractant [33], while exchangeable S was determined using the Mehlich-3 solution [33]. Total concentrations of Al, Ca, Fe, P, and S in soil and FG were determined by aqua regia digestion (3:1 HCl:HNO₃ at 130 °C for 1 h). In HA, total concentrations were analyzed using hot acid digestion in a digestion block with 7 mL of HClO₄ at 190 °C until the suspension became clear [34]. All elemental concentrations were measured using inductively coupled plasma optical emission spectroscopy (ICP-OES) (Analytic Jena PQ 9000, Analytik Jena GmbH, Jena, Germany).

The mineralogical compositions of bulk soil samples were analyzed by Empyrean PANalytical X-ray diffraction (XRD) (Malvern Panalytical, Almelo, The Netherlands) using an XPert3 powder diffractometer with a Pixel detector (CuK radiation 45 kV, 40 mA).

2.3. Sequential Extraction Procedure

Field-moist soil samples were air-dried in plastic trays inside a refrigerator at 4 °C to minimize microbial activity. After drying, the samples were finely ground using an agate mortar and pestle before undergoing SEP analysis. Phosphorus fractionation in the finely ground soil samples was conducted using the SEP method as described by Jantamenchai et al. [3] and Kovar and Pierzynski [35]. Po and Pi were categorized into three fractions: labile (labile Pi and labile Po), moderately labile (HCl-extractable Pi, HCl-extractable Po, and fulvic Po), and non-labile P (NaOH-extractable Pi, humic Po, and residual Po).

The SEP for P fractionation was performed using a 1:40 soil-to-solution ratio [3]. Labile P was extracted first with 0.5 mol L⁻¹ NaHCO₃ at pH 8.5 for 16 h to determine the soluble P and P adsorbed onto exchangeable sites of soil constituents. After the first-step extraction, the following less labile fractions were extracted from the soil residues from the preceding extraction as follows: Moderately labile P, which is bound to minerals that dissolve at low pH, such as apatite (Ca₅(PO₄)₃(OH,F,Cl)) and polyphosphates, was extracted using 1.0 mol L⁻¹ HCl for 3 h. Non-labile P was extracted with 0.5 mol L⁻¹ NaOH for 3 h to determine the P associated with Al and Fe (hydr)oxides. The supernatant from the NaOH extract was then acidified with concentrated HCl until the pH dropped below 1.0, causing the precipitation of HA, thus allowing for the separation of non-labile Po associated with HA from moderately labile Po associated with fulvic acid. Finally, residual Po, identified as non-labile Po bound to organo-mineral complexes, was determined by ashing the soil residue from the NaOH extraction at 550 °C for 1 h. The resulting ash was dissolved in 1.0 mol L⁻¹ H₂SO₄ for 24 h.

The suspension of soil and extractant was shaken in an end-over-end tumbler at 15 rpm for the specified duration and then centrifuged at 5000 rpm for 10 min. The supernatants were filtered through a 0.45 µm nylon membrane filter. The Pi concentrations in the extracts were measured using ICP-OES (Analytik Jena PQ 9000, Analytik Jena GmbH, Jena, Germany). Total P concentrations in the NaHCO₃, HCl, and NaOH extracts were determined by digesting appropriate aliquots with 2.5 mol L⁻¹ H₂SO₄ and potassium persulfate (K₂S₂O₈). The total P concentrations in the extracts were measured using ICP-OES (Analytik Jena PQ 9000, Analytik Jena GmbH, Jena, Germany). Organic P concentrations were calculated by subtracting Pi from total P of each respective fraction. Total Pi and Po were determined as the sum of the labile, moderately labile, and non-labile fractions of Pi and Po, respectively [3,35].

2.4. Phosphorus K-Edge X-Ray Absorption Near-Edge Structure Spectroscopy

Field-moist soil samples were air-dried in a plastic tray placed in a refrigerator at 4 °C to minimize microbial activity. The dried samples were then sieved through a 45 µm mesh to obtain the silt and clay fractions for solid-phase P speciation analysis using XANES spectroscopy.

P K-edge XANES spectroscopy was conducted on the combined silt and clay fraction at beamline BL8 of the Synchrotron Light Research Institute (SLRI) in Nakhon Ratchasima, Thailand. The beamline operated at an energy of 1.3 GeV with a beam current ranging from 80 to 150 mA [36]. The decision to analyze P in the combined silt and clay fraction was based on two factors: (1) the low P concentration in the bulk soil (0.41–0.44 g kg⁻¹;

Table 3. Mean ± standard deviation (n = 3) of chemical properties of acidic paddy soil after treatment with humic acid (HA), flue gas desulfurization gypsum (FG), and their combination (HA + FG).

Treatment	pH	Eh (mV)	TOC (g kg−1)	CEC (cmol kg−1)	Exc. Ca (g kg−1)	Exc. S (g kg−1)	N (g kg−1)	P (g kg−1)	Ca (g kg−1)	S (g kg−1)
Control	6.83 ± 0.40	−85.4 ± 11.6 a	2.26 ± 0.01 b	3.13 ± 0.35	0.23 ± 0.00 c	0.05 ± 0.01	1.37 ± 0.32 b	0.42 ± 0.01	0.35 ± 0.02 bc	0.48 ± 0.08 b
FG	6.60 ± 0.27	−98.2 ± 9.17 b	2.23 ± 0.02 b	3.07 ± 0.46	0.25 ± 0.00 bc	0.05 ± 0.00	1.12 ± 0.11 bc	0.41 ± 0.02	0.50 ± 0.05 a	0.54 ± 0.09 b
HA	6.97 ± 0.51	−103 ± 12.7 b	2.26 ± 0.08 b	3.26 ± 0.32	0.30 ± 0.02 a	0.05 ± 0.01	1.67 ± 0.08 ab	0.44 ± 0.03	0.46 ± 0.07 ab	0.55 ± 0.04 b
HA + FG	6.70 ± 0.17	−114 ± 8.85 c	2.42 ± 0.06 a	3.27 ± 0.12	0.27 ± 0.01 ab	0.06 ± 0.00	1.96 ± 0.10 a	0.41 ± 0.01	0.31 ± 0.06 c	1.01 ± 0.02 a
p value	0.676	0.003	0.018	0.867	0.004	0.876	0.002	0.096	0.016	<0.001

Abbreviations: Control = chemical fertilizer alone, Eh = redox potential, TOC = total organic carbon, CEC = cation exchange capacity, Exc. Ca = exchangeable calcium, Exc. S = exchangeable sulfur, N = total nitrogen, P = total phosphorus, Ca = total calcium, and S = total sulfur. Different letters in the same column indicate significant differences (p < 0.05).

), which made it difficult to obtain a usable spectrum, and (2) the high Si concentration in the studied acidic paddy soil, mainly from quartz, a predominant mineral contained in sand fraction of highly weathered soil (Table 2), which could interfere with the total fluorescence signal of P [37].

The beamline was equipped with an InSb (111) double-crystal monochromator, providing a proton flux of 1.3 × 109 to 3 × 1011 protons s⁻¹ (100 mA) ⁻¹ and a beam size of 17.7 × 0.9 mm² [38]. The silt + clay-sized soil sample was ground and sieved through a 45 µm mesh to minimize thickness effects [38]. The sample was placed into a 0.5 × 1 cm² window, supported by 2 mm thick stainless-steel holders, and covered with a P-free polypropylene film (Kapton tape). P K-edge XANES was performed in fluorescence mode under He gas conditions. Fluorescence data were collected using a 13-element Germanium (Ge 13 Array) detector. The energy step sizes were 5 eV between 2205 and 2350 eV, 0.25 eV between 2135 and 2205 eV, and 5 eV between 2205 and 2350 eV. Due to the low P concentrations, data were recorded with time steps of 9, 18, and 9 s per energy step, respectively. Elemental P (black P powder, 2145.5 eV) was used to calibrate the edge energy. All XANES data were analyzed using Athena software (version 0.9.26, Amazon Web Service, Inc., Seattle, WA, USA) [39]. Multiple spectra (five scans) were normalized by applying a linear baseline, subtracting a linear function from the spectral region below the edge (−20 to −5 eV relative to E₀), and using a quadratic function to normalize spectra across the post-white-line region (+40 to +60 eV relative to E₀). Linear combination fitting (LCF) was performed over the energy range from −10 to +30 eV relative to E₀. The P standard spectra of phytic inositol hexakisphosphate (IHP), lecithin, apatite (Ca₅(PO₄)₃(OH,F,Cl)), hydroxyapatite (Ca₅(PO₄)₃(OH)), brushite (CaHPO₄·2H₂O), archerite (KH₂PO₄), struvite ((NH₄)MgPO₄·6H₂O), variscite (AlPO₄·2H₂O), poorly crystalline variscite, P-adsorbed gibbsite (Al(OH)₃), vivianite (Fe²⁺·3(PO₄)₂·8H₂O), strengite (Fe³⁺(PO₄)₃·2H₂O), P-adsorbed ferrihydrite ((Fe³⁺)₂O₃·0.5H₂O), and P-adsorbed goethite (FeOOH) were obtained from [38,40]. Additionally, the standard spectrum of P adsorbed onto kaolinite was obtained from [4].

2.5. Quality Control and Statistical Analyses

The soil chemical properties and elemental analysis results were assessed for accuracy and precision using a rigorous quality control system involving instrument calibration, the use of reagent blanks, and certified reference materials from the DOA Proficiency Testing Program for soil physical and chemical analysis. Prior to statistical analysis, all data for soil properties, P fractions, and rice yield after HA and FG incorporation were assessed for normal distribution. All data were analyzed using analysis of variance, followed by mean comparisons with the least significant difference test (p < 0.05). The relationships between soil properties, soil phosphorus fractions, and yield were examined using principal component analysis (PCA), regression analysis, and Pearson correlation analysis (p < 0.05). All data analyses and visualizations were conducted using Statistica (version 8.0, StatSoft, Inc., Tulsa, OK, USA) and OriginPro 2018 software (version 9.5, OriginLab Corporation, Northampton, MA, USA), respectively.

3. Results and Discussion

3.1. Soil Chemical Properties After HA and FG Applications

The chemical properties of acidic paddy soil after the incorporation of HA, FG, and HA + FG at rice harvest are presented in Table 3. These treatments did not significantly affect the soil pH (6.60–6.97), CEC (3.13–3.27 cmol kg⁻¹), exchangeable S (0.05–0.06 g kg⁻¹), or P (0.41–0.44 g kg⁻¹) (Table 3). Although soil pH did not significantly differ between treatments at harvest, it increased by 1.9–2.3 units compared to the initial soil pH of 4.70 (Table 1). The HA-treated soil (pH 6.97) was likely to have a higher pH compared to the other treatments (pH 6.60–6.83). The increase in soil pH is primarily attributed to proton consumption by Fe (hydr)oxides under waterlogged conditions in acidic paddy soils [41]. A higher soil pH observed after HA incorporation was likely due to the high pH of the HA used in the study, which was adjusted to alkaline (pH 9.77) to enhance its solubility [42].

The Eh in the experimental soils ranged from −85.4 to −114 mV, which indicated that the soils were under reducing conditions [43]. The HA (−103 mV) and HA + FG (−114 mV) treated soils exhibited lower Eh values than the other two treatments. Soil Eh was negatively correlated with TOC (R² = 0.60, p < 0.05) (Figure 1a), suggesting that TOC derived from HA further promoted reducing conditions. The TOC mineralization under anaerobic conditions utilized oxygen from Fe (hydr)oxides as electron acceptors to compensate for oxygen depletion, which led to the reduction of Fe³⁺ to Fe²⁺ and a further decrease in redox potential [44].

TOC was significantly higher in the HA + FG-treated soil (2.42 g kg⁻¹) compared to the other treatments (2.26, 2.23, and 2.26 g kg⁻¹ for the control, FG, and HA-treated soils, respectively). The increase in TOC in the HA + FG-treated soil resulted from the synergistic effect of HA and FG. HA addition increases soil TOC due to its high organic carbon content (Table 1) and FG releases Ca, which potentially reacts with organic carbon (OC) to form Ca-OC complexes, thereby accumulating and preserving OC in the soil [45]. Additionally, Ca can stabilize OC in the soil through Ca-OC complexation, thereby protecting OC from degradation and decomposition by soil microbes [46].

CEC was higher in the HA and HA + FG-treated soils (3.26 and 3.27 cmol kg⁻¹, respectively) than in the control and FG-treated soils (3.13 and 3.07 cmol kg⁻¹, respectively) (Table 3). The bivariate analysis revealed a positive association between TOC and CEC (R² = 0.49, p < 0.05) (Figure 1b). This suggests that the addition of HA increases CEC by providing negative charges from the functional groups of HA, particularly carboxyl and phenolic groups [11,12].

The combination of HA and FG in acidic paddy soil increased the total N (Table 3). The highest N concentration (1.96 g kg⁻¹) was observed in the soil treated with both HA and FG, although it was not significantly different from the HA-treated soil (1.67 g kg⁻¹). Total N was positively correlated with TOC (R² = 0.90, p < 0.05) (Figure 1c). Both HA and FG supplied N to the soil, as they contained this element (Table 1). This suggests that adding HA, whether applied alone or in combination with FG, enhances N accumulation in paddy soils.

The total S concentration (g kg⁻¹) in the soils significantly increased under the HA + FG treatment (1.01 g kg⁻¹), followed by HA (0.55 g kg⁻¹), FG (0.54 g kg⁻¹), and control treatments (0.48 g kg⁻¹). Although the exchangeable S concentrations were not significantly different among the treatments, they exhibited a similar trend to those of total S in the soils (Table 3). Exchangeable S was highest in the HA + FG-treated soil (0.06 g kg⁻¹). The bivariate analysis revealed a positive association between the total S and TOC (R² = 0.65, p < 0.05) (Figure 1d). A positive relationship was also observed between the total and exchangeable S (R² = 0.73, p < 0.05) (Figure 1e). These results indicated that the S concentration originated from both HA and FG, which contain sulfur (Table 1). Moreover, the increase in S concentration was associated with the increase in TOC, suggesting that TOC positively influences soil S accumulation, thereby increasing the exchangeable S. This finding aligns with that of a previous study that reported TOC accumulation significantly contributed to a higher S pool in the soil, particularly total S and exchangeable S [47].

The total Ca (ranging from 0.31 to 0.50 g kg⁻¹) and exchangeable Ca (ranging from 0.23 to 0.30 g kg⁻¹) were significantly increased by the soil amendments (Table 3). The FG-treated soil (0.50 g kg⁻¹) exhibited higher Ca concentrations than the control (0.35 g kg⁻¹), HA-treated (0.46 g kg⁻¹), and HA + FG-treated soils (0.31 g kg⁻¹) (Table 3). Exchangeable Ca concentrations were higher in the soil treated with HA (0.30 g kg⁻¹) or HA + FG (0.27 g kg⁻¹) than in the control (0.23 g kg⁻¹) and FG soils (0.25 g kg⁻¹) (Table 3). The bivariate analysis revealed a positive association between exchangeable Ca and CEC (R² = 0.62, p < 0.05) (Figure 1f), indicating that the higher total Ca concentration in the FG-treated soil is attributed to the greater Ca input from FG addition (Table 1). However, the higher Ca concentration in the FG-treated soil did not contribute to increases in exchangeable Ca because the added Ca from FG was not adsorbed onto the exchange sites in the soil but precipitated with P. Exchangeable Ca was increased in the HA- and HA + FG-treated soils because they not only supplied Ca to the soil but also enhanced CEC, which provides more negative charges to adsorb the added Ca. Consequently, HA + FG-treated soils maintain higher levels of exchangeable Ca. However, the elevated exchangeable Ca likely facilitates higher Ca uptake by rice [48], which in turn reduces total soil Ca through crop removal.

Total phosphorus concentrations ranged from 0.41 to 0.44 g kg⁻¹ (Table 3). HA, FG, and HA + FG additions did not significantly affect total P concentrations because of crop removal of P. As the pH shifted after prolonged waterlogged conditions, it facilitated the release of P into readily available P form for rice uptake. Total P uptake plays a crucial role in the formation of grain and straw biomass in P-deficient soils, such as acidic paddy soils [14]. Hartina et al. [14] reported that rice cultivated in HA-treated soil had the highest total P uptake, which was not significantly different from the coapplication of HA and FG. The lowest P uptake occurred when FG was applied alone. These are the reasons for the non-significant changes in total P concentrations in the soil across the treatments.

3.2. Phosphorus Fractionation Determined by a Sequential Extraction

The soil P fractions in acidic paddy soil after the incorporation of HA, FG, and their combination are presented in Figure 2. The majority of soil P existed in the organic form, with concentrations ranging from 73% to 74% of the total P (299–319 mg kg⁻¹) (Figure 2a,b), indicating that organic P (Po) is the predominant form of P in acidic paddy soils. This result is contradictory to the general findings, which reported that acidic soil (pH 4.5–6.9) is dominated by inorganic P [3]. However, our findings reflected the biochemical process of waterlogged paddy soil conditions at the soil sampling time (128 DAP). The soil pH rose to 6.60 to 6.97 after prolonged waterlogged conditions, mainly due to the reduction condition of Fe oxides. The soil pH shift allowed P fixation (e.g., Fe-P and Al-P) to release soluble P to the soil, while Po mineralization was hindered under reducing conditions, resulting in Po forms that are likely to persist longer than inorganic forms.

Labile P constituted the largest P pool, particularly labile Po, which ranged from 31% to 32% of the total P (126–141 mg kg⁻¹) (Figure 2a,b). This was approximately three times higher than labile Pi (12% of the total P) (49.1–51 mg kg⁻¹) (Figure 2c,d). The HA-treated soil had the highest proportion of labile Po (32% of the total P) (141 mg kg⁻¹) and Pi (12% of the total P) (51 mg kg⁻¹) among the treatments (Figure 2a–d). This finding is consistent with that of previous studies, which reported that HA addition promoted labile P, especially organic P, in soils across a range of pH values (5.5 to 8.5). This is because HA alters P dynamics in the soil by forming HA-metal-phosphate complexes, which prevent phosphate fixation by metal ions (such as Al/Fe hydroxides) and reduce P adsorption onto Al and Fe (hydr)oxide minerals (such as goethite), as the carboxyl groups of HA compete with P adsorption sites [49].

Moderately labile P was present in both organic and inorganic forms. Moderately labile organic P (Po) consisted of HCl-extractable Po (HCl-Po) and fulvic acid-bound Po (fulvic Po). The HCl-Po fractions ranged from 13% to 15% of the total P (55.8–61.3 mg kg⁻¹). Fulvic Po was the dominant form of moderately labile Po in all studied soils (21% of the total P) (85.3–91.9 mg kg⁻¹) for all treatments (Figure 2a,b), which was likely due to the sandy loam texture of the soils used in this study. Several studies have reported that fulvic Po is a major pool of moderately labile Po in loamy soils, as it is adsorbed onto silt and clay fractions [3,50,51].

The moderately labile inorganic phosphorus (Pi) ranged from 6.2% to 6.4% of the total P (25.6–27.5 mg kg⁻¹) (Figure 2c,d). The highest proportion of HCl-extractable Pi (6.4% of the total P) was observed in the FG-treated soil, thereby suggesting that the FG-treated soil contained a higher amount of Ca-bearing P minerals. Further identification of Ca-P species is discussed in Section 3.3. This is attributed to the higher Ca content derived from FG in the soil. Lower HCl-extractable Pi was observed in the HA combined with FG-treated soil (6.3% of the total P) compared to the FG-treated soil alone (Figure 2c). This reduction is likely due to HA preventing the formation of Ca-P complexes by interacting with Ca derived from FG through the oxygen-based functional groups of HA (e.g., carboxyl and phenolic groups) [52].

The incorporation of HA and FG significantly affected NaOH-extractable inorganic phosphorus (NaOH-Pi), which is considered a non-labile form of P. NaOH-Pi in the soil was notably increased by HA addition (8.8% of the total P, 38.2 mg kg⁻¹) (Figure 2c,d), indicating higher amounts of P associated with Al and Fe (hydr)oxides in the HA-treated soil. This is attributed to HA supplying both Al and Fe, as shown in Table 1, which can bind with P and accumulate Al- and Fe-associated P in the soil.

The non-labile organic phosphorus (Po), comprising humic Po and residual Po, was a minor Po fraction in the studied acidic paddy soil. Humic Po ranged from 0.3% to 1.5% of the P, whereas residual Po ranged from 5.9% to 6.4% of the total P (Figure 2a). The HA + FG-treated soil exhibited the highest humic Po (1.5% of the total P) but the lowest residual Po (5.9% of the total P). Higher humic Po concentrations observed in the HA-treated (1.0% of the total P) and HA + FG-treated soils (1.5% of the total P) were associated with the accumulation of HA through HA application. Jantamenchai et al. [3] reported that non-labile humic Po is found in soils with the incorporation of low-quality organic residues. Low-quality organic residues, characterized by low nitrogen (N) concentration and higher lignin and polyphenol contents, are recalcitrant and exhibit slow decomposition rates. In addition, humic substances can form complexes with soil minerals, protecting organic matter from degradation by soil microbes. Thus, the incorporation of low-quality organic residues potentially contributes to higher humic substance accumulation, such as HA [3]. HA is a component of soil organic matter that can chelate metal ions, such as Ca derived from FG addition, which often forms complexes with P, resulting in the formation of humic-associated P [3,52]. Residual Po refers to more recalcitrant forms of Po that are less readily available for plant uptake [53]. The application of HA stimulates soil microbial populations and activity [54], leading to the mineralization of residual P and a subsequent decline in residual Po after HA application.

Therefore, this study highlighted the fact that the application of HA alone promoted organic labile Po and that the application of FG alone induced the formation of Ca-bound P minerals. Conversely, the coapplication of HA and FG suppressed Ca-P formation, reduced residual Po, and enhanced humic Po in acidic paddy soils. This study suggested that in acidic paddy soil after prolonged waterlogged conditions where P fixation is a major constraint, HA application maintains soil-available P in the form of organic P. Consequently, HA improves the efficiency of applied P fertilizer by increasing P availability, contributing to sustainable and effective nutrient management strategy.

3.3. Phosphorus Speciation Determined by P K-Edge XANES Spectroscopy

The P K-edge XANES spectra of the reference materials and those of silt + clay fraction of the soil samples after the incorporation of HA, FG, and their combination are presented in Figure 3a and Figure 3b, respectively. The LCF data showed that P-adsorbed gibbsite (Al(OH)₃) was the predominant P species found in the control soil and accounted for 55% of the total P species (Figure 4a). Additionally, variscite (AlPO₄·2H₂O) and P-adsorbed kaolinite were present in both the control and FG-treated soils. The P-adsorbed kaolinite accounted for 21% and 37% of the total P species in the control and FG-treated soils, respectively. The Al-P mineral, identified as variscite, corresponded to 24% in the control soil and 38% in the FG-treated soil (Figure 4a,b). Brushite, a Ca-P mineral, accounted for 25% of the total P species in the FG-treated soil (Figure 4b).

In the HA- and HA + FG-treated soils, archerite accounted for 40% and 20% of the total P species, respectively (Figure 4c,d). Strengite, an Fe-P mineral, accounted for 23% and 30% of the total P species in the HA- and HA + FG-treated soils, respectively. IHP, a non-labile organic phosphorus (Po), accounted for 12% and 31% of the total P species in the HA- and HA + FG-treated soils, respectively. Brushite (19%) was detected in the HA + FG-treated soil, but at a lower proportion compared to the FG-treated soil. Lecithin was observed only in the HA-treated soil, which accounted for 25% of the total P species.

The P K-edge XANES revealed that P transformation in the studied acidic paddy soil is primarily driven by the waterlogged condition and the application of soil amendments (HA, FG and their combination). Untreated paddy soil was dominated by Al compounds (i.e., variscite and P-adsorbed gibbsite), indicating that Al (hydr)oxides are the primary hosts of P in acidic paddy soils. These findings are consistent with previous studies on acidic agricultural soil and tropical regions [2,4,55,56]. The absence of measurable Fe-bound P (e.g., strengite and P-adsorbed ferrihydrite) may be attributed to the reductive dissolution of Fe (hydr)oxides during flooding in rice cultivation [41,57], resulting in Al-P being the predominant inorganic P form.

HA applied both alone and combined with FG in acidic paddy soil promoted the formation of archerite (K, NH₄) (H₂PO₄) by supplying a high K concentration (Table 1). In addition, HA promoted the NH₄-N in paddy soil [58], and the application of DAP fertilizer to the soil accommodated the formation of archerite, as NH₄ is the structure of archerite [59]. While waterlogged conditions lowered the Fe-P form in untreated paddy soil, XANES analysis found Fe-P in the form of strengite in HA- and HA + FG-treated soils. This is likely due to the direct Fe input derived from HA (5.2 kg ha⁻¹) and HA + FG (5.3 kg ha⁻¹) (Table 1), which may facilitate the formation of Fe-P mineral.

The occurrence of IHP (phytic acid), a stable form of P in the HA- and HA + FG-treated soils, was associated with the higher proportion of non-labile organic phosphorus (Po) discussed in Section 3.2. This is due to the addition of HA to the soil. Rigon et al. [56] showed that soils under long-term crop rotation contained substantial amounts of monoester-enriched TOC, such as IHP. Jantamenchai et al. [3] reported that phosphate monoesters are the predominant organic phosphorus form in acidic soils following the long-term (>22 years) annual incorporation of organic residues, particularly those with high lignin and polyphenol content. Furthermore, humic substances play a crucial role in influencing the fate of organic phosphorus in the soil, especially by regulating the hydrolysis of IHP through the action of the phytase enzyme [60]. The proportion of IHP was relatively higher in the HA + FG-treated soil compared to that in the HA-treated soil. This increase in IHP in the HA + FG-treated soil can be attributed to both HA- and FG-derived Ca, thereby facilitating the IHP stabilization. HA stabilizes IHP through strong complexation between IHP and HA (IHP-HA), especially with hydrogen bonding or sorption [61]. The Ca-IHP stabilization prevents its degradation or transformation into a more soluble form [62].

Furthermore, the effect of IHP stabilization in HA + FG-treated soil that involved Ca led to lower brushite in the HA + FG-treated soil. The Ca in HA + FG-treated soil had a higher affinity for forming complexes with IHP than P. Additionally, Ca-HA complexes were likely formed, contributing to TOC accumulation and stabilization, as discussed in Section 3.1. Ca can form organo-metal (HA-Ca-P) complexes with HA or act as a bridge in bidentate complexation between HA and P [52]. Consequently, fewer Ca ions remained in the HA + FG-treated soil to react with P to form brushite. Brushite is more likely to form under conditions such as higher Ca and P concentrations and slightly acidic to near-neutral soil pH, as found in FG-treated soil [63].

The occurrence of lecithin was found only in the HA-treated soil. To clarify this point, we emphasize that lecithin was employed as a representative organic P standard in the LCF analysis. The fitting results indicated that lecithin contributed exclusively to the HA-treated soil, implying the possible presence of macromolecular organic P compounds in the studied soil; nevertheless, the precise molecular identity of these species remains uncertain and should be confirmed by complementary spectroscopic approaches, such as solid-state ³¹P NMR. It should be noted that organic P cannot be unequivocally identified using P K-edge XANES, as this technique provides limited sensitivity to specific organic bonding environments. Moreover, phospholipid-type organic P, such as lecithin, generally constitutes only a minor fraction—typically a few percent—of the total P in agricultural soils. According to current methodological advances, P L-edge XANES could, in principle, provide more direct information on organic P speciation [64]. However, its application to soil systems remains technically challenging, as it requires exceedingly high total P concentrations (≈4000 mg kg⁻¹) that are rarely encountered in sandy soils characterized by low mineral phases capable of retaining P.

This study highlights the fact that HA addition effectively increased the accumulation of archerite and lecithin, which play a crucial role in enhancing soil P availability for plant uptake. On the contrary, the application of FG, whether alone or in combination with HA, reduced P availability in acidic paddy soils as P was transformed into more insoluble forms, such as brushite and IHP. The lower solubility and the consequent less availability of brushite and IHP reduced P uptake by rice under prolonged waterlogged conditions.

3.4. Rice Yield

The rice straw biomass and grain yield results are presented in Figure 5. The incorporation of HA, FG, and their combination significantly affected both parameters. The HA-treated soil exhibited higher straw biomass (15.8 t ha⁻¹) and grain yield (7.74 t ha⁻¹). Notably, the FG- and HA + FG-treated soils produced lower straw biomass (14.4 and 14.5 t ha⁻¹, respectively) and grain yield (6.83 and 7.0 t ha⁻¹, respectively) compared to those in the control (16.3 and 7.33 t ha⁻¹, respectively) and HA-treated soil (Figure 5).

Crusciol et al. [65] reported that phosphogypsum application increases rice yield by improving soil properties, such as exchangeable Ca and S in acidic upland rice soil (initial pH 4.8, harvest pH 4.9–5.6), while Goiba et al. [48] showed that applying sludge-based gypsum at a rate of 750 kg ha⁻¹ increased rice yield in an acidic paddy soil in India (initial pH 4.6, harvest pH 4.0 to 4.7). These contrasting results with our study were attributed to differences in soil pH at harvest. In previous studies, soil pH remained acidic (ranging from 4.0 to 5.6), whereas in this study, it was altered to near-neutral (6.83–6.97) (Table 3). Near-neutral pH conditions can induce brushite formation, particularly when Ca²⁺ and H₂PO₄⁻ concentrations in the soil solution are high [63]. As brushite is insoluble, it limits P availability and restricts P uptake by rice, leading to reduced yield. A previous study also reported that soil treated with FG had lower yield than soil receiving chemical fertilizer alone or HA alone, due to reduced P uptake [14]. Therefore, this study underscores the requirement for careful consideration when incorporating FG into flooded acidic paddy soils or other flooded paddy soils with neutral pH, as it promotes the formation of brushite (a Ca-P mineral), which reduces soil P availability and, consequently, lowers rice grain yield and straw biomass.

3.5. Relationships Between Phosphorus Fractions, Phosphorus Speciation, Soil Chemical Properties, and Rice Yield

The results of the principal component analysis (PCA) of the standardized average data of P fractions, P speciation, soil properties, and rice yield in acidic paddy soil are presented in Figure 6. The PCA results showed that the control soil was associated with Eh, HCl-Po, fulvic Po, residual Po, and P-adsorbed gibbsite (Figure 6a,b). The higher proportion of moderately labile Po, including both HCl-Po and fulvic Po, in the control soil compared to other treatments was attributed to its association with clay minerals (e.g., kaolinite) in the studied soil. Additionally, fulvic Po was positively correlated with residual Po (r = 0.55, p < 0.05) (Figure 7), suggesting a potential linkage between these P pools. This correlation indicated that transformations between residual Po and fulvic Po occurred over time, possibly through microbial or physicochemical processes. However, residual Po represented a stable form with low turnover. The absence of external organic amendments in the control soil resulted in lower TOC, implying limited microbial activity, which hindered the breakdown of stable Po compounds that contribute to residual Po accumulation [66]. Consequently, residual Po tends to be higher in the control soil. The PCA also showed that P adsorbed onto gibbsite was associated with the control soil (Figure 6). Pearson correlation analysis indicated that gibbsite was positively correlated with Eh (r = 0.83, p < 0.05) (Figure 7), thereby suggesting that higher Eh provided suitable conditions for gibbsite formation. Gibbsite, a dominant Al hydroxide mineral in acidic paddy soils, has a stronger P adsorption capacity than clay minerals [67].

The FG-treated soil was associated with HCl-Pi, brushite, variscite, P associated with kaolinite, and labile Pi (Figure 6). HCl-Pi was positively correlated with brushite (r = 0.79, p < 0.05) (Figure 7), likely due to the high amount of Ca released from FG, which induced brushite formation. Ca-P minerals primarily form in soil through reactions between Ca and P ions, particularly under slightly acidic to alkaline conditions [63]. Additionally, Wang et al. [68] confirmed that Ca-P precipitation is more pronounced in neutral and slightly alkaline conditions. The pH of the FG-treated soil was slightly acidic (pH = 6.60), thus facilitating brushite precipitation. Furthermore, the addition of Ca from FG, in the absence of organic amendments, allowed Ca ions to freely precipitate P into Ca-P minerals. This finding aligns with previous studies reporting that Ca-P mineral formation occurs in acidic soils following FG incorporation in Canada, southwestern Brazil, and the USA [16,20,21].

Furthermore, the FG-treated soil was positively correlated with variscite and P-adsorbed kaolinite (Figure 6a,b) because Ca released from FG replaced Al³⁺ on exchange sites, resulting in displacement of Al³⁺ into the soil solution and increasing its availability [69]. The released Al³⁺ then reacted with soluble P, forming variscite. Additionally, the FG-treated soil was associated with P-kaolinite. Kaolinite, a clay mineral with negatively charged surfaces, can interact with Ca²⁺ ions derived from FG, which bridge or neutralize surface charges, facilitating the adsorption of P via electrostatic attraction and ligand exchange mechanisms. Furthermore, Ca has a strong affinity for P ions, enhancing P adsorption onto kaolinite [70].

The HA-treated soil was associated with soil pH, total P, labile Po, NaOH-Pi, lecithin, and archerite (Figure 6a,b). Labile Po was positively correlated with total P (r = 0.98, p < 0.05) and soil pH (r = 0.82, p < 0.05) (Figure 7), thereby indicating that total P and pH play a crucial role in increasing labile Po. A higher total P input supplied more labile Po to the soil because the added P served as a source for labile Po. During the P cycling process, organic P continued to be mineralized, replenishing depleted soil-available P. Soil pH and total P also played a pivotal role in improving P availability [2,3,68]. The increase in soil pH in the HA-treated soil stimulated microbial activity, enhancing the breakdown of organic P compounds and releasing more labile Po. Although P K-edge XANES was not effective at quantifying individual organic P, as discussed in Section 3.3, labile Po was positively correlated with lecithin (r = 0.98, p < 0.05) (Figure 7), suggesting that labile Po originated from microbial biomass in the form of lecithin. Lecithin (phosphatidylcholine) is a membrane phospholipid primarily found in microbial cells, and its presence in soil is often linked to microbial biomass turnover. As microbial biomass decomposes, lecithin is degraded and releases P, thereby increasing the soluble P [71]. In addition, lecithin can serve as an important source of soluble P following its mineralization by soil microbes [71]. The HA-treated soil was also associated with NaOH-Pi because HA addition contributed total Fe input to the soil, as discussed in Section 3.3.

The HA + FG-treated soil was associated with TOC, exchangeable Ca, humic Po, IHP, and strengite. The concentration of non-labile organic P, such as humic Po, was higher in HA + FG-treated soil than in the other treatments (Figure 2b). Humic Po was positively correlated with TOC, exchangeable Ca, and IHP (r = 0.65, 0.73, and 0.94, respectively; p < 0.05) (Figure 7), suggesting that Ca released from FG stabilized humic substances by forming HA-Ca complexes [72]. These complexes stabilized pre-existing Po, particularly as the IHP and humic Po, by acting as a bridge in HA–metallic–P complexes [52]. The formation of HA-Ca-P complexes restricted microbial access to organic P, thereby slowing its mineralization [73]. As a result, more humic Po remained in the HA + FG-treated soil.

Moreover, PCA revealed that FG-treated soil was partially separated from HA + FG-treated soil along the factor 1-axis because the former contained higher levels of brushite and labile Pi. Labile Pi was enriched in FG-treated soil because rice absorbed less labile Pi in this treatment than it did in the HA treatment. As a result, higher labile Pi concentrations were found in FG-treated than HA-treated soils. Hartina et al. [14] reported that P uptake in rice was higher under HA treatment than that under FG or HA + FG treatments, resulting in greater labile Pi accumulation in FG-treated soil compared to other treatments. Furthermore, the higher proportion of strengite in HA + FG-treated soil was associated with the greater proportion of NaOH-Pi in these soils, likely due to the Fe constituents in the added HA (Figure 2) (Table 1). Additionally, HA applications with FG induced interactions between HA and Ca derived from FG, thus preventing the formation of Ca-P compounds such as brushite. Consequently, brushite concentrations in HA + FG-treated soil were lower than those in the FG-treated soil.

From an agronomic and environmental perspective, labile Po is an important source of soluble P that, once mineralized, replenishes depleted labile Pi as a readily available P form for rice absorption [66]. Labile Po was positively correlated with grain yield (r = 0.82, p < 0.05) (Figure 7). Additionally, rice grain yield was strongly associated with straw biomass and total P (r = 0.85 and 0.90, respectively, p < 0.05) (Figure 7) but negatively correlated with HCl-Pi and brushite (r = −0.51 and −0.91, respectively, p < 0.05) (Figure 7). This study found that grain yield was associated with HA-treated soil, as HA application enhanced rice yield by increasing available P for plant uptake. Hartina et al. [14] reported that enhanced P uptake is associated with higher PUE. P fertilizer applied under HA treatment was effectively utilized by rice, which can potentially minimize the environmental risk linked to excessive P application. In contrast, grain yield declined with higher HCl-Pi proportions, which were linked to FG application (Figure 6a,b). This study suggests that Ca derived from FG application promotes brushite formation in flooded acidic paddy soils—a process that may also occur in paddy soils with neutral pH. Furthermore, brushite formation reduces soil-available P, leading to lower rice yields.

The findings indicate that HA offers sustainable P management, enhancing nutrient cycling and agronomic performance by reducing P fixation. As a result, HA increases P availability and facilitates P uptake, leading to improved rice yield. FG applied alone or combined with HA induced more insoluble P form that limits P uptake for crop production. This suggests the importance of selecting P management strategies that promote available P in acidic paddy soils. However, as this present study was conducted on a single acidic paddy soil specifically classified as Aeric Kandiaquult, further study is required to evaluate the effect of HA, FG and their combined application on P cycling across multiple soil types and diverse locations.

4. Conclusions

This study investigated the effect of HA, FG, and their combination (HA + FG) on P dynamics and rice yield in an acidic paddy soil. HA application enhances P cycling to maximize rice yield by increasing organic P (Po) mineralization (e.g., labile Po and lecithin), stimulating soil microbial biomass, and increasing soil pH and total P. In contrast, FG, either applied alone or combined with HA, promoted stable Ca-P mineral brushite, leading to a reduction in soluble P concentration and a lower yield. Although coapplication of HA and FG improved soil nutrient-related parameters (TOC, N, S, and exchangeable Ca and S), elevated TOC causes the stabilization of non-labile organic P (e.g., humic Po and IHP), ultimately limiting available P and decreasing yield. This shows conclusively that HA alone is an effective soil amendment for improving P cycling and sustaining rice production in the studied acidic paddy soil. While this study demonstrated the positive effect of HA application as a soil amendment on P dynamics and rice yield in a single type of acidic paddy soil, further research is needed to explore P dynamics under applications of HA, FG, and their combinations in other acidic paddy soils as well as multiple other soil types, particularly under a long-term or multi-year field experiment.