Soil-Specific Redox Effects on Phosphorus Availability and Diagnostic Approaches in Flooded Paddy Soils

Abstract

Accurate evaluation of plant-available phosphorus (P) in flooded paddy soils requires consideration of redox dynamics and soil-specific properties. This study evaluated five soil P extraction methods, such as Truog, Bray 2, Mehlich 3, Olsen, and ascorbic acid-reduced Bray 2 (AR Bray 2), using soils collected from 20 paddy fields in a cold region of Japan that have received long-term fertilization. All four methods, except AR Bray 2, were conducted under air-dried and flooded incubation conditions. Additionally, we conducted pot experiments with the two rice cultivars to measure P uptake. Bray 2 extracted the highest amount of P (543.6–1045.4 mg P kg⁻¹). Incubation increased extractable P by factors of 2.4–4.9 with the Mehlich 3 and Truog methods, indicating enhanced P solubility under reduced conditions. The Olsen method showed minimal sensitivity to redox changes (−31.4 mg P kg⁻¹). Principal component and cluster analyses suggested three patterns of soil P behavior under changing redox conditions: (1) stable P extractability regardless of redox status; (2) increased P availability after incubation; and (3) P extractability depending on the extraction method used. These patterns were not explained by regional or taxonomic classifications. A comparison of soil extractions and P uptake indicated that no single method consistently predicted shoot P concentrations across all soils, suggesting that conventional P extraction methods may have limited ability in long-term fertilized paddy soils. Our findings demonstrate that soil-specific redox behavior and cultivar-specific P demand critically influence the effectiveness of standard P tests. Therefore, selecting diagnostic methods tailored to soil characteristics and crop requirements is essential for accurate P evaluation and sustainable fertilizer management in rice cultivation.

1. Introduction

Phosphorus (P) is an essential macronutrient that plays a critical role in plant physiological processes, such as nucleic acid synthesis, membrane formation, and energy metabolism, and it constitutes up to 0.2% of plant dry weight [1,2]. P deficiency is a widespread constraint on crop productivity, potentially reducing yield by 5–15% [3]. In rice, the typical deficiency symptoms include stunted growth, reduced tillering, dark green leaves, thin stems, delayed maturity, and poor grain quality [4]. For optimal growth, rice plants require between 1.8 and 4.2 kg P per ton of grain yield, with a widely cited average value of 2.8 kg P [5,6]. This wide range of optimal growth rates may be attributed to differences in cultivars, soil conditions, management, and environmental backgrounds.

This is particularly critical in regions like northern Japan, where early rice growth is frequently restricted by low temperatures, which limit P uptake and delay development [7]. Historically, large amounts of P fertilizer have been applied to mitigate this issue, especially in such regions, contributing to improved early growth and yields [8]. However, when P supply exceeds crop demand, a substantial portion of the applied P becomes fixed in the soil through interactions with iron and aluminum oxides, and organic matter, leading to its long-term accumulation rather than complete plant uptake. Consequently, rice cultivation in Japan consumes approximately 30% of the nation’s total P fertilizer input [9,10,11]. Therefore, excessive P in the soil has become a concern, with many rice fields exceeding the recommended P levels based on the Truog and Bray 2 methods, which are commonly used to evaluate available P in Japan [12,13], including Yamagata Prefecture [14]. Yamagata Prefecture was selected for this study because it is one of Japan’s major rice-producing regions and is representative of cold-climate paddy fields in northern Japan. To ensure sustainable fertilizer use, it is increasingly important to evaluate plant-available P in paddy fields and adjust inputs accordingly.

P availability in soils is governed by complex interactions between mineral phases, organic matter, and other environmental factors. Although soil solution P concentrations are typically low (0.01–3 mg L⁻¹), plant uptake is sustained through continuous resupply from labile solid-phase P pools [15,16]. Available P, which is a plant-available fraction, is commonly evaluated using various chemical extraction methods [16,17]. However, because of the high variability in soil properties, no single method is universally applicable and multiple extractants are often used, even in a single country [18,19]. Nevertheless, in practice, extraction methods have been largely standardized around a few representative types: the Truog method, which is widely used in Japan, targets Ca- and Mg-bound P under acidic conditions [20]; the Bray 2 method targets P associated with Ca, Al, and partially Fe [21]; the Mehlich 3 method extracts Ca-, Al-, and Fe-bound P using a multicomponent solution [22]; and the Olsen method, suitable for alkaline soils, primarily extracts Ca-bound P [23,24]. These methods are not directly comparable owing to the differences in the target P fractions [25]. In other words, the amount of available P determined by each method is difficult to compare directly because each extractant targets different soil P pools. Several recent studies have further elucidated the theoretical basis and limitations of these methods under diverse conditions (e.g., calcareous, acidic, flooded), such as soil-specific competition effect, comparative evaluations in organic and flooded soils [26,27], as well as Fe-bound P mobilization dynamics influenced by redox status in flooded soils [28,29]. Our work builds on these foundations by explicitly assessing extraction method sensitivity under varying redox conditions and cultivar-specific P demand in flooded paddy soils.

In paddy fields, recent efforts to refine soil P testing have emphasized the need to integrate redox-dependent dynamics driven by microbial activity and changing water regimes, particularly in flooded systems. Under reduced conditions, iron oxides undergo reductive dissolution, releasing Fe-bound P into the soil solution and substantially increasing P availability compared to aerobic conditions. These processes are central to P dynamics in submerged soils and must be considered when evaluating plant-available [30]. Additionally, notable attempts have been made to evaluate the P release potential of paddy soils [31]. Previous studies have developed chemical extraction methods specifically for paddy soils using Fe-reducing or chelating agents to better capture P availability under submerged conditions [30]. These approaches highlight the importance of considering redox-driven processes when selecting diagnostic methods for rice systems. Standard extraction methods may not reflect the solubilization of Fe-bound P under reduced conditions, which is common in paddy soils. Incubation-based methods that simulate submergence have provided valuable insights [32] but are time-consuming and labor-intensive. As an alternative, chemically reduced methods, such as ascorbic acid-reduced Bray 2 (AR Bray 2), offer a practical approach. This method has shown strong agreement with incubation-based extractions [33] and a high correlation with P extracted using the Bray 2 method after incubation (r = 0.92), supporting the potential of AR Bray 2 as a diagnostic tool for air-dried samples under reduced conditions [25].

In this study, we collected soils from 20 paddy fields and determined plant-available P using multiple extraction methods under both air-dried and flooded incubation conditions. The flooded incubation was intended to simulate the submerged environment that occurs in rice paddy fields. In addition, pot experiments using the same soils were conducted to estimate plant P content up to the heading stage, when P demand is highest. We hypothesized that the diagnostic reliability of P extraction methods varies with redox status, and that integrating soil extraction results with plant responses would improve the evaluation of the diagnostic performance of various extraction methods in cold-climate regions such as northern Japan. Accordingly, this study focuses on method applicability boundaries under high P conditions, rather than identifying a universally optimal method. Paddy fields in this area have typically received long-term fertilization and exhibit relatively high levels of plant-available P. This study was designed to provide conceptual insights that may contribute to improved P fertilizer management in such high P conditions rather than evaluating fertilizer practices directly under field conditions. The pot experiments were intended to assess the relationship between soil P availability and plant P content under controlled conditions, without examining yield responses to different fertilizer application rates. Therefore, the implications of our findings should be interpreted as method-specific strengths and limitations within the targeted high P conditions.

2. Materials and Methods

2.1. Study Site and Tested Soil Properties

Soil samples were collected from 20 paddy fields across four major rice-producing regions in the Yamagata Prefecture, Japan: Shonai, Mogami, Murayama, and Okitama (Figure 1). In each region, five fields representing typical paddy soils were selected. Candidate sampling areas were first identified by reviewing maps to maintain spatial intervals, and in each candidate area, permission was obtained from farmers after confirming their typical cultivation practices. Soil sampling was conducted in the tilled paddy fields in September and October 2023. Surface soil samples (0–15 cm) were collected from each field at five diagonally arranged points, including the center, and then composited into a single sample for each field. According to climate data from the Japan Meteorological Agency [34], the average annual temperatures in Shonai, Mogami, Murayama, and Okitama were 12.9 °C, 11.0 °C, 12.1 °C, and 11.4 °C, respectively [34]. The average temperature in May, when rice transplanting typically occurs, was 16.0 °C in Shonai, 11.1 °C in Mogami, 16.2 °C in Murayama, and 15.6 °C in Okitama. During the rice-growing season (May to September), the mean temperatures were 21.3 °C, 20.2 °C, 21.2 °C, and 20.7 °C, respectively (Figure 2). The corresponding annual precipitation levels were 2191 mm, 2006 mm, 1207 mm, and 1445 mm, respectively. The dominant soil types at the sampling sites were classified as Fluvisols and Lixisols, based on the World Reference Base for Soil Resources [35,36]. The locations of the sampling sites are shown in Figure 1.

All soil samples were air-dried and passed through a 2 mm sieve. For total carbon (TC) and total nitrogen (TN) analysis, fine-ground soil samples were oven-dried at 105 °C for 24 h and analyzed using a CN analyzer (SUMIGRAPH NC-220F; Sumika Chemical Analysis Service, Osaka, Japan). For total phosphorus (TP) analysis, 5 g of air-dried fine-ground soil was digested with 1 mL of sulfuric acid, 5 mL of nitric acid, and 20 mL of perchloric acid on a hot plate at 120 °C for 3 h. After digestion, 30 mL of 1 M hydrochloric acid and 50 mL of hot water were added, and the mixture was heated immediately to below the boiling point. The supernatant was filtered, and the P concentration in the digest was determined by the vanadomolybdate method using the same spectrophotometer. Available nitrogen (N) was estimated by incubating soil samples at 30 °C for 4 weeks, extracting with 2 M KCl, and analyzing the extract colorimetrically [37]. Phosphate adsorption capacity (PAC) was determined by adding 20 mL of 2.5% ammonium phosphate solution to 10 g of air-dried soil and allowing the mixture to stand for 24 h. The suspension was filtered through a filter paper (ADVANTEC No. 5C; ADVANTEC TOYO KAISHA, Tokyo, Japan). One milliliter of the filtrate was transferred to a 50 mL volumetric flask and diluted to volume with deionized water. A 2.5 mL aliquot was used for color development using the vanadomolybdate method, and absorbance was measured with a spectrophotometer (UV-2550; Shimadzu Corporation, Kyoto, Japan). The basic soil properties are presented in

Table 1. Sampling site information, including geographic coordinates (latitude and longitude), soil classification, and key soil chemical properties.

Region	Soil	Latitude	Longitude	Soil Classification	TC	TN	TP	Available N	PAC
					(g kg−1)	(g kg−1)	(g kg−1)	(mg kg−1)	(g kg−1)
Shonai	Sh1	139°52′ N	38°47′ E	Fluvisols	41.3	3.57	1.76	275.2	12.95
	Sh2	139°47′ N	38°45′ E	Fluvisols	19.8	1.55	0.99	51.1	6.04
	Sh3	139°55′ N	38°42′ E	Fluvisols	57.8	4.46	2.20	277.8	18.23
	Sh4	139°55′ N	38°51′ E	Fluvisols	21.4	2.02	1.38	153.5	9.66
	Sh5	139°55′ N	38°59′ E	Fluvisols	30.5	2.90	1.68	278.6	14.23
Mogami	Mo1	140°17′ N	38°46′ E	Luvisols	73.8	5.34	3.74	249.7	20.49
	Mo2	140°20′ N	38°49′ E	Luvisols	41.4	2.71	1.99	121.4	15.15
	Mo3	140°15′ N	38°51′ E	Luvisols	21.2	1.71	2.04	127.1	8.25
	Mo4	140°11′ N	38°44′ E	Luvisols	25.7	2.52	0.77	209.6	12.09
	Mo5	140°20′ N	38°41′ E	Luvisols	73.3	4.78	2.62	261.1	16.55
Murayama	Mu1	140°19′ N	38°20′ E	Luvisols	39.0	3.05	2.50	146.2	15.75
	Mu2	140°22′ N	38°32′ E	Luvisols	81.7	5.20	4.75	123.3	21.40
	Mu3	140°14′ N	38°22′ E	Luvisols	33.2	3.09	1.50	111.1	10.50
	Mu4	140°16′ N	38°15′ E	Luvisols	36.4	2.97	0.78	142.5	9.72
	Mu5	140°20′ N	38°28′ E	Luvisols	29.1	2.39	0.81	134.6	12.67
Okitama	Ok1	140°05′ N	37°53′ E	Luvisols	32.5	2.78	1.49	123.6	13.30
	Ok2	140°08′ N	37°57′ E	Luvisols	21.7	1.84	0.98	137.4	7.75
	Ok3	140°07′ N	38°02′ E	Luvisols	41.8	3.37	1.51	188.4	10.45
	Ok4	140°03′ N	37°59′ E	Luvisols	41.6	3.37	1.95	219.1	12.52
	Ok5	140°01′ N	38°07′ E	Luvisols	30.9	2.49	1.17	131.9	7.55

TC, total carbon; TN, total nitrogen; TP, total phosphorus; Available N, plant-available nitrogen; PAC, phosphate adsorption capacity. Soil samples were labeled using regional codes (Sh: Shonai; Mo: Mogami; Mu: Murayama; Ok: Okitama).

.

2.2. Plant-Available Phosphorus Analysis

Plant-available P was determined using five extraction methods: Truog, Bray 2, Mehlich 3, Olsen, and AR Bray 2. For the Truog method, 0.001 M H₂SO₄ was added to air-dried soil at a 1:200 soil-to-extractant ratio, allowed to stand for 30 min, and then filtered [20]. The Bray 2 method involved adding a mixture of 0.1 M HCl and 0.03 M NH₄F at a 1:20 ratio, followed by 1 min of shaking and filtration [21,38]. For the Mehlich 3 method, a mixed solution containing 0.2 M CH₃COOH, 0.25 M NH₄NO₃, 0.015 M NH₄F, 0.013 M HNO₃, and 0.001 M EDTA was added at a 1:10 ratio, shaken for 5 min, and then filtered [22]. For the Olsen method, 0.5 M NaHCO₃ solution (pH 8.5) was added to the soil at a 1:20 ratio, shaken for 30 min, and then filtered [23]. In the AR Bray 2 method, 10 mL of 0.057 M ascorbic acid was added to 1 g of air-dried soil, and the mixture was shaken for 16 h. Subsequently, 10 mL of double-strength Bray 2 extractant was added, and the mixture was shaken for 1 min and filtered [39]. All filtrates were analyzed for P concentration using the molybdenum blue method with a spectrophotometer (UV-2550; Shimadzu Corporation, Kyoto, Japan).

To simulate reduced conditions, soil was incubated according to a previously described method [40]. Approximately 10 g of air-dried soil was placed in an aluminum vapor-deposited pouch (LAMIZIP AL-11, 202 mm × 110 mm; Seisan Nipponsha, Tokyo, Japan) containing 100 mL of deionized water. The pouch was vacuum-sealed and incubated at 30 °C for 4 weeks. The soil incubation under flooded and vacuum-sealed conditions at 30 °C for 4 weeks is widely used as a diagnostic method for evaluating plant-available nitrogen under simulated flooded rice-growing conditions in Japan [41,42]. Additionally, previous studies have shown that soil redox potential (Eh) typically decreases to below −200 mV within several days after flooding and remains low thereafter [43,44,45,46]. Although Eh was not measured in this study, the redox status was likely shifted toward reduced conditions during incubation. After incubation, available P was extracted using the Truog, Bray 2, Mehlich 3, and Olsen methods.

2.3. Design of the Pot Experiments

Pot experiments were conducted in the experimental field of the Faculty of Agriculture, Yamagata University (38.74° N, 139.83° E), using 20 soil samples collected from four regions: Shonai, Mogami, Murayama, and Okitama. A total of 2.5 kg of air-dried soil from each site was placed in Wagner pots (1/5000 scale). Two rice cultivars were used: ‘Haenuki,’ a widely cultivated variety in Yamagata Prefecture, and ‘Tsuyahime,’ a premium-quality cultivar that has recently expanded in cultivation and is promoted under the Yamagata Prefecture certification program for reduced- or no-fertilizer cultivation [47]. Fertilization treatments followed the regional guidelines [48], except that no P was applied. Nitrogen and potassium were applied at fixed rates: 710.9 mg N and 238.6 mg K₂O per pot for Haenuki, and 474.0 mg N and 159.1 mg K₂O per pot for Tsuyahime. Ammonium sulfate and potassium chloride were used as sources of nitrogen and potassium, respectively. The fertilizers were mixed with air-dried soil, which was then saturated with water and thoroughly homogenized. Each treatment was replicated three times.

In both cultivars, seedlings were transplanted on 30 May 2023, and the experiments continued until the heading stage on 6 August 2023. The average air temperature during the growing period was 25.5 °C, and the average soil temperature was 24.7 °C. The average temperature in June at the experimental site was 23.4 °C, which was slightly higher than the regional averages for Shonai (20.1 °C), Mogami (19.3 °C), Murayama (20.3 °C), and Okitama (19.8 °C) (Figure 2). The pots were maintained under continuously flooded conditions, and standard practices were followed for weed and pest control. Because P uptake intensifies after root establishment and peaks around heading and flowering, this stage was considered optimal for our measurements. At this stage, the aboveground biomass was harvested, oven-dried at 75 °C for 48 h, and weighed to determine shoot dry weight (SDW).

To evaluate the P content, approximately 500 mg of dried, fine-ground plant tissue was digested. One milliliter of deionized water and 4 mL and concentrated sulfuric acid were added to a Pyrex test tube containing the sample, and the mixture was blended using a touch mixer. The tube was then heated at 250 °C for 30 min in an aluminum block. Hydrogen peroxide was added until the solution became clear, and the digest was diluted to a final volume of 100 mL [49]. The P concentration in the digest was determined using the molybdenum blue method and a spectrophotometer.

2.4. Statistical Analysis

Boxplots were generated to visualize the distribution of available P concentrations across the 20 soil samples for each extraction method. Pearson correlation coefficients were calculated to assess the relationships between the extraction methods, and the results are presented as a heatmap. Hierarchical cluster analysis was conducted to classify soils based on the available P profiles across the extraction methods, allowing for the evaluation of similarities and differences between the soils and methods. Principal component analysis (PCA) was performed using nine standardized variables representing the available P content determined by different extraction methods using air-dried and incubated soils. Loading vectors are displayed to indicate the contribution of each variable, and scores for the first two principal components are plotted to visualize the distribution of samples among the four regions (Shonai, Mogami, Murayama, and Okitama).

Two-way analysis of variance (ANOVA) was performed to evaluate the effects of soil, rice cultivars, and their interaction on SDW, shoot P concentration, and P uptake. Multiple comparisons were conducted using Tukey’s honest significant difference (HSD) test. To further examine the relationship between available P and plant response, linear regression analyses were performed between available P and shoot P concentrations for each cultivar. All statistical analyses were conducted using R version 4.5.0 [50].

3. Results

3.1. Plant-Available Phosphorus Distribution by Extraction Method

Figure 3 shows the distribution of available P determined using the five extraction methods under different conditions. The increase from air-dried soil (DS) to incubated soil (IS) indicated enhanced P solubility following incubation. The degree of change and variability differed between the extraction methods tested. Bray 2 exhibited high values and wide variation, particularly during incubation, with a mean available-P value of 543.6 and 686.3 mg P kg⁻¹ for DS and IS, respectively. Mehlich 3 also exhibited a strong response to incubation, with available-P value increasing from 141.9 (DS) to 549.7 mg P kg⁻¹ (IS). In contrast, Olsen showed almost no differences between DS (128.1 mg P kg⁻¹) and IS (96.7 mg P kg⁻¹), whereas Truog showed a significant increase in extractable P from 77.7 mg P kg⁻¹ (DS Truog) to 346.6 mg P kg⁻¹ (IS Truog) (p < 0.05). AR Bray 2 had the highest available-P value (1045.4 mg P kg⁻¹). The variability and interquartile range were particularly large for AR Bray 2, suggesting a substantial variation in P extractability across soils. The soils, which were also used in the pot experiment, exhibited substantial variability in their initial nutrient status (Table 1). Based on this heterogeneity, each soil was treated as an independent sample in the statistical analyses, and comparisons among extraction methods were conducted within each soil. These results suggest that the amounts of P extracted using the different methods responded differently under reduced conditions.

3.2. Correlations Among Total Phosphorus, Phosphorus Adsorption Capacity, and Plant-Available P in Different Extraction Methods

The Pearson correlation coefficients among the TP, PAC, and available-P values extracted by the different methods are shown as a heatmap in Figure 4. TP was strongly positively correlated with AR Bray 2 (r = 0.89, p < 0.001), DS Bray 2 (r = 0.81, p < 0.001), and DS Olsen (r = 0.77, p < 0.001). PAC was significantly correlated with AR Bray 2 (r = 0.71, p < 0.001) and DS Bray 2 (r = 0.55, p < 0.05). IS Bray 2 and AR Bray 2 showed a moderate correlation (r = 0.38), but this was not statistically significant. These results suggest a potential relationship between P mobilized during incubation and the chemically reduced fraction extracted by the ascorbic acid treatment, although this relationship is not clearly supported by the present data. Mehlich 3 methods showed weak or negative correlations with TP and PAC, with DS Mehlich 3 having a significant negative correlation with PAC (r = −0.52, p < 0.05). IS Mehlich 3 had no significant correlation with the TP or PAC. DS Olsen showed strong correlations with TP (r = 0.77, p < 0.001), DS Bray 2 (r = 0.79, p < 0.001), and AR Bray 2 (r = 0.88, p < 0.001). In contrast, Olsen (IS) exhibited lower correlation coefficients with these variables, indicating that incubation reduced the strength of these relationships. DS Truog moderately correlated with TP (r = 0.49, p < 0.05), whereas IS Truog had no significant correlation with either TP or PAC. However, IS Truog correlated strongly with IS Mehlich 3 (r = 0.95, p < 0.001), suggesting a similarity between these two methods under incubation conditions.

3.3. Cluster Analysis Based on Plant-Available Phosphorus Extraction Methods

Cluster analysis of soils based on P extracted using different methods is shown in Figure 5. Contrary to expectations based on soil classification, no clear regional clustering patterns were observed. Although soils from Shonai were classified as Fluvisols and those from Mogami, Murayama, and Okitama were classified as Lixisols, clustering did not reflect these taxonomic classifications.

However, the clustering pattern was more strongly influenced by the extraction method. The Mehlich 3 and Truog methods tended to group, with IS Mehlich 3 and IS Truog forming one cluster, and DS Mehlich 3 and DS Truog forming another cluster. This suggests that these two methods extract similar P fractions under both incubated and air-dried conditions. A similar but weaker relationship was observed between the Bray 2 and the Olsen methods. AR Bray 2, including the reduced fraction extracted using ascorbic acid, was expected to be similar to IS Bray 2. However, AR Bray 2 clustered closely with DS Bray 2.

Notably, the amount of P varied considerably among soils depending on the extraction method used. For example, Sh5, Mo4, Mu4, and Mu5 consistently exhibited low available-P value across all extraction methods. In contrast, Sh1, Mu1, Ok3, and Ok4 showed high available-P value using Mehlich 3, Truog, and Bray 2 under incubated conditions but low available-P value under air-dried conditions, suggesting the influence of redox status. Meanwhile, Mo5 and Mu2 showed low available-P value with Mehlich 3 under both conditions, but higher values with DS Bray 2 and AR Bray 2.

3.4. PCA on Plant-Available Phosphorus

PCA was conducted using data obtained from all extraction methods applied to the soils collected from the 20 paddy fields. The first two principal components explained 69.2% of the total variance, with PC1 accounting for 42.4% and PC2 for 26.8%. The variable loadings are shown in Figure 6. PC1 can be interpreted as representing variation among extraction methods, with loadings generally increasing in the order of the Mehlich 3, Truog, Bray 2, and Olsen methods. In contrast, PC2 can be interpreted as representing the variation associated with the redox status, as indicated by the positive loadings for DS and negative loadings for IS. AR Bray 2 behaved similarly to Olsen and Bray 2 on PC1 and was roughly positioned between the DS and IS on PC2.

Figure 7 illustrates the PCA scores of the soil samples from different regions across the two principal components. The Mogami soils are plotted as a narrow elliptical distribution along the DS ray 2 vector. The Shonai and Okitama soils are plotted toward the IS Mehlich 3 and IS Truog vectors, and the Murayama soils are plotted as a broader and more diffuse ellipse distribution with the centroid located similar to that of Mogami.

3.5. Shoot Biomass and Phosphorus Content

SDW, P concentration, and P content varied significantly among the soils, but no significant differences were observed between the two cultivars or in the cultivar × soil interactions (

Table 2. Shoot dry weight, P concentration, and P content of two rice cultivars (Haenuki and Tsuyahime) grown in soils from different sampling sites.

Cultivar	Soil	Shoot Dry Weight		P Concentration		P Content
		(g pot−1)		(%)		(mg pot−1)
Haenuki	Sh1	34.55 ± 2.50	abcde	0.44 ± 0.01	a	151.86 ± 13.17	ab
	Sh2	21.59 ± 2.52	e	0.36 ± 0.02	abcd	77.28 ± 3.96	f
	Sh3	44.07 ± 2.16	abcd	0.32 ± 0.01	bcd	140.22 ± 7.51	abc
	Sh4	40.70 ± 1.16	abcd	0.33 ± 0.01	bcd	136.19 ± 4.00	abcd
	Sh5	47.76 ± 4.07	ab	0.35 ± 0.03	abcd	163.17 ± 7.06	a
	Mo1	42.81 ± 2.91	abcd	0.31 ± 0.02	cd	132.54 ± 1.52	abcd
	Mo2	27.97 ± 1.58	de	0.34 ± 0.01	bcd	94.24 ± 5.43	cdef
	Mo3	28.27 ± 0.67	cde	0.33 ± 0.01	bcd	94.60 ± 5.18	cdef
	Mo4	39.65 ± 4.63	abcd	0.39 ± 0.02	abc	157.52 ± 24.28	ab
	Mo5	38.41 ± 4.82	abcde	0.34 ± 0.03	abcd	128.31 ± 9.64	abcde
	Mu1	45.44 ± 6.04	abc	0.34 ± 0.02	bcd	151.63 ± 10.43	ab
	Mu2	30.16 ± 1.01	cde	0.31 ± 0.01	cd	92.06 ± 1.73	def
	Mu3	41.57 ± 3.02	abcd	0.36 ± 0.01	abcd	147.42 ± 8.07	ab
	Mu4	39.66 ± 2.88	abcd	0.35 ± 0.02	abcd	136.79 ± 1.42	abcd
	Mu5	51.15 ± 2.25	a	0.33 ± 0.01	bcd	166.20 ± 2.24	a
	Ok1	31.29 ± 1.01	bcde	0.36 ± 0.01	abcd	112.51 ± 2.79	bcdef
	Ok2	29.32 ± 1.17	cde	0.29 ± 0.01	d	86.06 ± 3.72	ef
	Ok3	41.30 ± 1.93	abcd	0.38 ± 0.02	abcd	157.79 ± 3.19	ab
	Ok4	40.64 ± 4.57	abcd	0.41 ± 0.02	ab	165.44 ± 10.19	a
	Ok5	35.11 ± 5.68	abcde	0.40 ± 0.04	abc	137.38 ± 10.88	abcd
Tsuyahime	Sh1	40.13 ± 1.68	abcde	0.37 ± 0.01	a	147.53 ± 3.79	bcde
	Sh2	20.97 ± 1.57	f	0.35 ± 0.02	a	73.60 ± 3.37	h
	Sh3	45.09 ± 5.32	abc	0.32 ± 0.03	a	141.62 ± 7.10	cde
	Sh4	39.75 ± 3.09	abcde	0.33 ± 0.02	a	129.72 ± 4.15	cdef
	Sh5	45.34 ± 2.01	abc	0.36 ± 0.01	a	162.74 ± 11.31	abc
	Mo1	38.97 ± 1.90	abcde	0.35 ± 0.01	a	134.07 ± 2.32	cdef
	Mo2	29.28 ± 0.87	def	0.32 ± 0.01	a	93.00 ± 3.85	gh
	Mo3	29.18 ± 0.95	def	0.35 ± 0.01	a	100.64 ± 2.92	fgh
	Mo4	42.35 ± 3.28	abcd	0.36 ± 0.03	a	149.77 ± 0.73	bcd
	Mo5	36.57 ± 1.69	bcdef	0.35 ± 0.01	a	127.55 ± 3.09	def
	Mu1	39.88 ± 6.00	abcde	0.33 ± 0.04	a	128.79 ± 13.65	def
	Mu2	29.91 ± 0.74	cdef	0.30 ± 0.01	a	89.82 ± 2.05	gh
	Mu3	40.36 ± 4.78	abcde	0.39 ± 0.04	a	154.48 ± 3.09	abcd
	Mu4	40.50 ± 3.05	abcde	0.35 ± 0.01	a	139.63 ± 7.44	cde
	Mu5	54.52 ± 2.04	a	0.33 ± 0.02	a	177.75 ± 13.08	ab
	Ok1	30.04 ± 1.47	cdef	0.39 ± 0.03	a	115.13 ± 5.29	efg
	Ok2	24.84 ± 0.64	ef	0.30 ± 0.01	a	74.85 ± 2.10	h
	Ok3	43.04 ± 3.10	abcd	0.37 ± 0.01	a	158.09 ± 6.29	abcd
	Ok4	49.71 ± 1.91	ab	0.37 ± 0.01	a	183.83 ± 1.63	a
	Ok5	37.97 ± 4.15	bcde	0.39 ± 0.03	a	145.98 ± 5.94	bcde

Values are presented as means ± standard deviation. Different lowercase letters within each cultivar indicate significant differences among soils based on Tukey’s multiple comparison test at the 5% level. The soil samples were labeled with region codes: Sh (Shonai), Mo (Mogami), Mu (Murayama), and Ok (Okitama).

and

Table 3. Statistical significance of the effects of cultivar, soil, and their interaction on shoot dry weight, P concentration, and P content of two rice cultivars.

Source	Shoot Dry Weight		P Concentration		P Content
	F	p	F	p	F	p
Variety	0.146	0.704	0.604	0.439	<0.001	0.990
Soil	14.498	<0.001	5.797	<0.001	30.859	<0.001
Variety:Soil	0.705	0.803	1.006	0.464	0.429	0.653

Values represent F-values and corresponding p-values from the two-way ANOVA.

). For Haenuki, SDW varied between 22.63 g pot⁻¹ (Sh2) and 50.41 g pot⁻¹ (Mu5), whereas for Tsuyahime, it ranged from 20.97 g pot⁻¹ (Sh2) to 54.52 g pot⁻¹ (Mu5). Sh5, Mu5, and Ok4 consistently supported higher biomass production in both cultivars, whereas Sh2 had the lowest biomass. Shoot P concentrations exhibited relatively small variations. In Haenuki, P concentrations ranged from 0.31% (Ok2) to 0.44% (Sh1), whereas for Tsuyahime, they ranged from 0.29% (Ok2) to 0.41% (Ok3). No significant differences in P concentrations were observed between the two cultivars. Shoot P content represents the combined effects of biomass and P concentration. For Haenuki, P content varied between 75.03 mg pot⁻¹ (Sh2) and 175.80 mg pot⁻¹ (Mu5), while for Tsuyahime, it ranged from 73.60 mg pot⁻¹ (Sh2) to 183.83 mg pot⁻¹ (Ok4). Sh5, Mu5, and Ok4 resulted in higher P content in both cultivars.

3.6. Relationship Between Plant-Available P and Shoot P Concentration

The SDW in rice was partly affected by available nitrogen (N), as soils with low N produced lower SDW despite sufficient N fertilizer application (Table 1 and Table 2). This suggests that the limited mineralization of soil organic N during the growing period may have restricted biomass production and indirectly restricted P uptake. To reduce the confounding effect of N limitation, we focused on the relationship between available P and shoot P concentration, rather than shoot P uptake. Figure 8a–i shows the relationship between shoot P concentration and available P determined by five different extraction methods under air-dried and incubation conditions. When plotted against available P in DS, shoot P concentration did not show a clear increase with increasing available P (Figure 8a,c,e,g). In IS, although no sharp increase was observed, a positive trend was evident, particularly for Haenuki (Figure 8b,d,f,h), where shoot P concentration tended to increase with increasing available P. Among the extraction methods, AR Bray 2 (Figure 8i) showed a broader distribution of available-P values (500–2000 mg kg⁻¹) and a response pattern similar to that of DS Bray 2 (Figure 8a), whose values were mostly concentrated below 500 mg kg⁻¹. This broader distribution suggests that AR Bray 2 may provide a practical advantage as a rapid method for evaluating soil P availability across a wider range of soil types; however, this indication is preliminary.

4. Discussion

4.1. Phosphorus Extractability and Redox Sensitivity

Under both air-dried and incubated conditions, the Bray 2 method, including AR Bray 2, consistently extracted the highest values of P among all the tested methods and exhibited the widest range of extractable P across the soils (Figure 3). This indicates that Bray 2 had higher extractability than the other methods. Bray 2 extracted seven times more P than Mehlich 3 in volcanic ash soils, highlighting the strong influence of soil type on extraction efficiency [51]. In the same study, the Mehlich 3 method extracted approximately twice as much P as the Truog method, which is consistent with our results. Regarding the Olsen method, some studies have reported similar extractability to Mehlich 3 [52], whereas others have reported that Mehlich 3 extracts nearly twice as much P as Olsen, depending on the soil characteristics [53,54,55]. The Bray 2 and Mehlich 3 methods generally extracted larger amounts of P, whereas the Olsen method extracted significantly less [56]. These findings suggest that P extractability varies considerably, depending on the extraction method and soil properties. In the present study, the Bray 2 method exhibited the highest extractability among the tested methods in paddy soils.

In the three extraction methods, Bray 2, Mehlich 3, and Truog, the amount of P extracted in IS was significantly higher than that in DS (Figure 3). This increase may result from the reductive dissolution of Fe-bound P, which is known to become more soluble under the reduced conditions typical of flooded paddy soils. However, because Eh and Fe concentrations were not measured in this study, the extent to which reduced conditions were established and whether Fe-bound P dissolution contributed remain speculative and should be verified in future research. In contrast, P associated with calcium (Ca-bound P) and aluminum (Al-bound P) minerals tends to be less sensitive to redox changes [57,58]. The Truog method primarily targets Ca-bound P [59], whereas Mehlich 3 extracts both Ca- and Fe/Al-bound P owing to the combination of acetic acid and fluoride components [60,61]. Therefore, both methods showed marked increases in extractable P under reduced conditions, likely reflecting the enhanced solubility of Fe-bound P. Bray 2, which preferentially extracts Fe-bound P, extracted a part of the Fe-bound P pool even under oxidized conditions. As a result, the increase in extractable P from air-dried and incubated conditions was smaller for Bray 2 than for Mehlich 3 or Truog. The Olsen method, which primarily targets Ca-bound P, shows little to no increase in extractable P under reduced conditions. This is consistent with previous findings that Olsen P is largely unaffected by redox conditions and does not effectively extract Fe-bound P [16,62,63]. These results collectively highlight the importance of selecting P extraction methods that reflect soil redox status. In flooded rice systems, conventional aerobic extraction methods may underestimate the availability of Fe-bound P. Therefore, the use of reducing agents such as ascorbic acid or conducting anaerobic incubations can help simulate field-relevant conditions and improve the accuracy of P availability assessments.

Although AR Bray 2 has been proposed as a potential indicator of available P under reduced conditions, owing to its agreement with values obtained after flooded incubation [33], our results showed a weak correlation between AR Bray 2 and IS Bray 2 (Figure 4). This discrepancy may be attributed to the ability of AR Bray 2 to extract not only inorganic P (Pi) but also a portion of organic P (Po) [64]. Ascorbic acid reduces Fe(III)-bound P to Fe(II)-P, which becomes more soluble under acidic conditions, thereby enhancing P release to soil solution [62]. In addition, AR Bray 2 may mobilize Po to a greater extent than the other extraction methods, although the other methods may also extract Po. These mechanisms may partially explain the discrepancy and the higher extractable-P values observed with AR Bray 2.

Po can account for 20–80% of TP, depending on soil properties [65,66]. Although Pi is the primary form of P taken up by rice, Po, especially phosphate monoesters, may also be available through enzymatic mineralization [67]. The strong correlation observed between AR Bray 2 and TP may suggest that this method mobilizes a broader range of P forms. The chemical properties of AR Bray 2—particularly its reducing and acidic characteristics—could hypothetically facilitate the release of Po alongside Pi [64]. Furthermore, the significant correlation between the AR Bray 2 and PAC suggests the influence of soil organic matter, which affects both Po accumulation and PAC [68]. However, because Po was not directly measured in this study, this interpretation remains uncertain and should be quantitatively verified in future research.

4.2. Correlations with Extraction Methods

The strong correlations of Bray 2 and AR Bray 2 with both TP and PAC indicate that these methods reflect the influence of soil P pools and adsorption capacity. This suggests that Bray 2 and AR Bray 2 are sensitive to the chemical processes governing P availability in paddy soils, rather than simply reflecting TP. In contrast, the relatively low or negative correlations for Mehlich 3 and the low correlations for Truog under certain conditions suggest that these methods may be less reliable indicators of available P in reduced environments.

In the FAO/WRB classification, Mu5 is categorized as a Luvisol. However, according to the Japanese soil classification system, it is classified as an Andosol because of its high amorphous mineral content and strong phosphate retention capacity [69]. The low TP and high PAC of Mu5 resulted in lower available P in DS. Andosols, which are widely distributed in volcanic areas and occur mainly in uplands, although a small proportion is also used in paddy fields in Japan, are characterized by strong phosphate retention owing to the abundance of amorphous minerals, particularly allophane, which has a high affinity for phosphate [70]. Although soil organic matter can limit phosphate adsorption, the low TC content in Mu5 suggests its limited ability to mitigate P fixation. More than 60% of phosphate added to such soils can become occluded and nonreactive within 6 days. The long-term application of P may eventually saturate these adsorption sites, depending on soil properties and management [71]; however, the combination of high PAC and low TP in Mu5 indicates that the adsorption capacity remains active and unsaturated. PAC, which reflects the presence of amorphous Al and Fe compounds responsible for P fixation, has been proposed as a useful indicator of the P retention potential in Andosols [72,73]. Consistent with this, the P retention capacity of non-calcareous soils is correlated with oxalate-extractable Al, oxalate-extractable Fe, and dithionite-extractable Fe [71]. Thus, the high PAC in Mu5 suggests strong, ongoing P fixation typical of Andosols, which are often P-limited owing to the abundant reactive Al and Fe phases [74,75].

These findings underscore the need to evaluate P availability using indicators that reflect soil-specific characteristics, particularly in Andosols, where strong P fixation by amorphous minerals can mask actual P availability. In such highly adsorptive soils, both TP and PAC should be considered to accurately assess plant-available P under variable redox conditions Among the tested extraction methods, AR Bray 2 showed slightly better performance than the others within our dataset; however, none of the methods demonstrated sufficiently strong correlations, indicating that further refinement and validation are needed.

4.3. Distinct Extraction Method Characteristics and Redox Responses Revealed by Cluster Analysis and PCA

Cluster analysis and PCA collectively revealed systematic differences in the P extraction methods under varying redox conditions. As described in the Results Section 3.4, PC1 can be interpreted as representing variation among the extraction methods, with scores generally increasing in the order of the Mehlich 3, Truog, Bray 2, and Olsen. PC2 represents the variation associated with redox status, as indicated by the positive loadings for DS and negative loadings for IS (Figure 6). Because the data were standardized, these components reflect relative extraction patterns across soils rather than absolute extractable P. AR Bray 2, despite extracting on average 1.9 times more P than DS Bray 2 (Figure 3), clustered closely with DS Bray 2 in both the PCA and cluster analysis. This suggests that the extraction patterns across soils were similar despite quantitative differences (Figure 5).

Mehlich 3 and Truog are chemically distinct under oxidizing conditions because of their limited capacities to extract Fe- and Al-bound P from Truog [76]. These two methods showed low correlation in DSs; however, their correlation increased markedly after incubation (IS, r = 0.95; Figure 4), and they clustered closely in IS (Figure 5). This change likely reflects the increased solubility of Fe- and Al-bound P owing to microbial reductive dissolution [77,78,79]. In contrast, the Olsen method, which targets Ca-bound P, showed minimal differences between DS and IS and did not extract Fe-bound P, which is consistent with previous studies [16,62,63]. This was evident in Sh1, Mu1, Ok3, and Ok4, which exhibited increases in P with incubation in Truog and Mehlich 3 but not in Olsen.

Our findings provide two important insights. First, the extraction methods retained distinct characteristics even under redox changes, as shown by the consistent clustering patterns. Second, unexpected similarities—such as those between IS Mehlich 3 and IS Truog, and between AR Bray 2 and DS Olsen—suggest complex interactions among soil properties, redox status, and extractant chemistry. These methods are not directly interchangeable owing to differences in the extracted P fractions [25,80]. Previous studies have emphasized that assuming a direct relationship between Fe reduction and plant P uptake oversimplifies the system; instead, P availability is shaped by a multifactorial interplay of biogeochemical processes [81]. Based on our study, at least three patterns of soil responses were identified: (1) soils that were relatively unaffected by either the extraction method or redox status, showing little variation between DS and IS; (2) soils that were highly responsive to changes in redox status, showing large increases in extractable P after incubation; and (3) soils from which highly extractable P was obtained only with specific extraction methods. Recognizing these types can inform the selection of suitable extraction methods and improve P management under diverse paddy soil conditions.

4.4. Regional Variability in Soil Phosphorus Characteristics Identified by PCA

PCA based on the P extraction results revealed region-specific trends in soil P behavior (Figure 7). The shapes and orientations of the ellipses representing each region highlight the within-region variability and dominant directions of variation. For instance, Mogami soils were characterized by a smaller and narrower elliptical area than other regions, suggesting relatively homogeneous P behavior within the region, likely resulting from similar soil-forming processes and land use. In contrast, the elliptical areas for the Shonai and Okitama soils extended in a direction roughly orthogonal to that of the Mogami soil, indicating distinct patterns of variation related to extraction methods and redox status. Murayama soils showed no clear grouping and were widely scattered, reflecting greater heterogeneity in soil mineralogy, redox sensitivity, or land-use history.

Although regional grouping was initially expected, the observed patterns indicated that P availability was not determined solely by geographic origin or soil classification. This finding is consistent with those of previous studies that have emphasized the complexity of P dynamics in paddy soils. For example, the assumed direct relationship between Fe reduction and P availability is often oversimplified because plant P uptake is influenced by multiple interacting factors such as redox status, mineral composition, and root behavior [81]. Similarly, P availability varies substantially depending on the chemical form of the soil P pool and the chosen extraction method [82]. In addition, soils in temperate regions, unlike those in tropical climates, typically contain more reactive Fe (hydroxide) oxides, which play a dominant role in P solubility under changing redox conditions [83,84].

Moreover, differences in parent materials and pedogenic processes can lead to considerable heterogeneity in P behavior, even within seemingly similar environments [85]. Our findings support this view, as multiple soil types with contrasting P extraction responses were found to coexist within individual regions. Consequently, the predictive utility of regional or taxonomic classifications is insufficient for predicting P availability. Instead, effective P management in paddy systems should rely on site-specific assessments that consider soil properties and redox responsiveness.

4.5. Phosphorus Availability and Plant Uptake

In DS, the lack of a clear correlation between any extraction method and plant P concentration (Figure 8) reflects the high P values and narrow range of available P in the studied area. Most fields can already supply sufficient P to reach the plateau in plant P response, which is consistent with previous studies when soil P levels exceeded crop requirements [86,87,88]. This limitation is a characteristic of many long-term fertilized paddy systems in northern Japan. Rice fields account for approximately 30% of national P fertilizer use [9,10,11]. Surveys indicate that about 80% of rice fields exceed available P thresholds based on Truog-P in air-dried soils [13], including those in Yamagata Prefecture [14,25]. Therefore, the soils used in this study are representative of the rice fields to which diagnostic P tests are typically applied. Accordingly, the contribution of our study is not the identification of a universally “best” method, but rather the delineation of method applicability boundaries under high P conditions. Our results demonstrate a practical constraint in identifying a single best available-P extraction method in such paddy fields. These findings support ongoing efforts to reconsider P fertilizer rates and to explore management strategies that make better use of accumulated soil P while maintaining crop productivity.

In contrast, the incubated soils showed a slightly positive relationship between plant P concentration and soil P availability, particularly in Hanenuki, which has a higher nutrient demand than Tsuyahime [47]. This suggests that extractions under flooded incubation conditions may better reflect rice P status than those from air-dried soils. However, our results show that, even under standardized flooded incubation conditions, diagnostic performance remains limited when plant P status is already near the response plateau. Despite the use of ascorbic acid as a reductant, AR Bray 2 exhibited a relationship with plant P concentration that more closely resembled the pattern observed for DS. However, AR Bray 2 was more closely associated with rice shoot biomass and with the incubated available-P values. This indicates that AR Bray 2 may be more informative for evaluating growth responses rather than excessive P uptake, which is a valuable distinction for fertilizer management in high P paddy soils. Nevertheless, the current evidence is limited, and additional work is needed before AR Bray 2 can be considered a reliable diagnostic method.

When relating the results of available-P extraction methods to crop performance under such conditions, careful interpretation is required. In this study, we used simple linear regressions to evaluate the relationship between available P and plant P concentration; however, the relationship is not necessarily linear. Our investigation was conducted near the plateau where available P has accumulated to high levels. Datasets that include other soil P conditions, such as low P paddy soils, should be evaluated using nonlinear or segmented regression models. Furthermore, several soil and management factors can affect the apparent relationship between extraction results and plant response. Initial variation in soil properties among treatments (Table 1), including non-P nutrient status, may partly account for the differences in shoot biomass and P uptake summarized in Table 2. Under flooded cultivation, available P is also influenced by temporal changes in soil redox status [89,90]: dissolved reactive P can initially increase during the dissolution of Fe(III) oxides and subsequently decline as Fe(II) minerals form and re-adsorb P [90,91]. Thus, available-P extraction methods at one incubation time capture only part of the available-P dynamics, and the timing of measurement becomes an important consideration for any diagnostic protocol under flooded incubation conditions [92]. Although plant P uptake is a useful integrative indicator for evaluating P availability in extraction methods, its interpretation in high P paddy soils requires consideration of plant physiological traits, non-P nutrient availability, soil redox dynamics, and key soil properties such as texture, organic matter, and Fe/Al oxides.

5. Conclusions

This study evaluated multiple P extraction methods under both air-dried and flooded incubation conditions to determine their ability to predict rice P uptake in cold-region paddy soils that have received long-term fertilization. The results revealed that soil-specific responses to redox status and extraction method were more influential than regional or taxonomic classification. Cluster analysis identified three patterns of soil P behavior under changing redox conditions: soils with stable extractability regardless of redox status, soils with increased P availability after incubation, and soils in which extractability was strongly dependent on the extraction method. These patterns indicate that a one-size-fits-all approach is unsuitable, and that P assessment should instead be tailored to soil-specific characteristics and redox behavior. Additionally, in the pot experiments, no clear relationship was observed between any extraction method and plant P concentration because plant P status had reached the plateau of crop response. This provides practical evidence that, under high P conditions, soil P tests can have limited capacity to predict plant P status. Although AR Bray 2 showed preliminary promise as a rapid alternative method, none of the tested methods demonstrated consistent predictive performance across all soils. This outcome emphasizes the limitation of conventional extraction methods in intensively fertilized paddy systems. The agronomic guidance value of this study lies in decision-making risks when soil P test values are interpreted as direct indicators of plant response. Future work linking plant-available P obtained using different extraction methods to key soil properties such as amorphous Fe content may clarify P mobilization mechanisms and, in the long term, contribute to more efficient and sustainable P management in flooded rice production systems.