Phosphorus Dynamics in Long-Term Flooded, Drained, and Reﬂooded Soils

: In ﬂooded areas, soils are often exposed to standing water and subsequent drainage, thus over fertilization can release excess phosphorus (P) into surface water and groundwater. To investigate P release and transformation processes in ﬂooded alkaline soils, wheat-growing soil and vegetable-growing soil were selected. We ﬂooded-drained-reﬂooded two soils for 35 d, then drained the soils, and 10 d later reﬂooded the soils for 17 d. Dissolved reactive phosphorus (DRP), soil inorganic P fractions, Olsen P, pH, and Eh in ﬂoodwater and pore water were analyzed. The wheat-growing soil had signiﬁcantly higher ﬂoodwater DRP concentrations than vegetable-growing soil, and ﬂoodwater DRP in both soils decreased with the number of ﬂooding days. During the reﬂooding period, DRP in overlying ﬂoodwater from both soils was less than 0.87 mg/L, which was 3–25 times less than that during the ﬂooding period. Regardless of ﬂooding or reﬂooding, pore water DRP decreased with ﬂooding days. The highest concentration of pore water DRP observed at a 5-cm depth. Under the effect of fertilizing and ﬂooding, the risk of vertical P movement in 10–50 cm was enhanced. P diffusion occurred from the top to the bottom of the soils. After ﬂooding, Al-P increased in both soils, and Fe-P, O-P, Ca 2 -P decreased, while Fe-P, Al-P, and O-P increased after reﬂooding, When Olsen P in the vegetable-growing soil exceeded 180.7 mg/kg and Olsen P in the wheat-growing soil exceeded 40.8 mg/kg, the concentration of DRP in pore water increased signiﬁcantly. Our results showed that changes in ﬂoodwater and pore water DRP concentrations, soil inorganic P fractions, and Olsen P are signiﬁcantly affected by fertilizing and ﬂooding; therefore, careful fertilizer management should be employed on ﬂooded soils to avoid excess P loss.


Introduction
The loss of phosphorus (P) from agricultural soils to surface water and groundwater has been a worldwide concern for many years, but losses of P from agriculture still contribute to nonpoint pollution [1][2][3][4][5][6]. P is a major contributor to eutrophication [7,8], and its loss from agricultural nonpoint sources can contribute to this process.

Soil Sampling and Preparation
Two types of soil were used in this experiment: a vegetable-growing soil and a wheat-growing soil. This experiment is for yearly soil fertilization when plants are not growing intensively. Soil samples were collected from six different depth intervals using core tubes: 0-5 cm, 5-15 cm, 15-25 cm, 25-35 cm, 35-45 cm, and 45-60 cm. Each sample consisted of six subsamples, which were mixed together and homogenized to obtain a single sample. A portion of each soil sample was air dried, passed through a 0.15-mm sieve. The remaining soil was used to fill the soil columns.
The soil columns were made of Perspex, with a height of 1 m and an internal diameter of 20 cm. They were filled with 45-60 cm, 35-45 cm, 25-35 cm, 15-25 cm, 5-15 cm, and 0-5 cm soils. To reduce the disturbance of the soil, the backfill soil was filled into the column according to the wet bulk density, which is 1.27-1.80 g/cm 3 . Below 45-60 cm, the soil was fine sandy, and its particle diameter was larger than that of the upper soils in the column. There was also a nylon wire netting between the 45-60 cm soil and fine sand. At the bottom of the sandy section, there was a water-tap to drain the column water.

Experimental Set-up
Two replicates of each soil were used and there were a total of four soil columns. Values are means of two replicates. Before flooding, the four columns were saturated with tap water from the bottom up to remove entrapped air. Next, fertilizer was added: 1.38 g KH 2 PO 4 was dissolved in 4 L tap water, and this solution, in which the PO 4 -P concentration was 34.5 mg/L, was put into the soil column. Three days later, we added tap water into the soil column until the water in the column was 20 cm deep. The soil columns were covered with a plastic cover to minimize evaporation and kept at room temperature (26 ± 2 • C).
Flooding, drying, and reflooding treatments were used to all soil columns: (i) long-term flooding-draining, which involved flooding the soil columns to a depth of approximately 20 cm for 35 days and then draining the floodwater and drying the soils for 13 days and (ii) reflooding, which involved reflooding the soil columns to the same depth of 20 cm for 16 days. Because of evaporation and water-sampling, 250 mL tap water was added into soil columns every 3 days during the flooding period.
During flooding and reflooding, pore water in each soil layer were collected with Rhizon samplers (Rhizosphere Research Products, Wageningen, The Netherlands) placed horizontally at 5,10,20,30,40,50, and 60 cm depth in the soil profile. Rhizon samplers extract small volumes of pore water from soil, in an easy, non-destructive way. They are made of inert polymer with no ion exchange properties. Other advantages of Rhizons are their small diameter and low dead volume. The pore size of the porous part is 0.12-0.18 µm, hence filtering the samples before analyzing is unnecessary. Pore water samples were collected using these samplers every two or three days, and analyzed for pH, Eh, and DRP. Grab samples of standing flood water were collected manually during the flooded periods.
On the first day of draining and redraining, soil samples were collected from each layer on one side of the soil column, opposite from the Rhizon samplers. Soil samples were air dried, sieved to 0.15 mm, and analyzed for TP, Olsen P, and inorganic phosphorus fractions.

Water and Soil Analysis
Soil samples were used to measure TP, Olsen P (soil available phosphorus), and inorganic P fractions. The analysis method of P forms in soil was given in Table 1. We used different extractant to analyze P forms and all extracted solutions were analyzed with molybdenum blue method [32]. The Olsen P was measured using the method described by Olsen [33]. The concentrations of inorganic P fractions were measured using methods described by Lu [34]. The soils were subjected to a sequential extraction: (i) Ca 2 -P (CaHPO 4 ·2H 2 O); (ii) Ca 8 -P (Ca 8 H 2 (PO 4 ) 6 ·5H 2 O); (iii) Al-P; (iv) Fe-P; (v) occluded P (O-P); and (vi) Ca 10 -P (Ca 10 (PO 4 ) 6 (OH) 2 ).  [34] The contents of sand, silt, and clay were measured with a Laser particle size analyzer (LS-POP (6)); OMEC, Zhuhai, Guangdong, China). Soil samples were extracted by potassium dichromate and H 2 SO 4 and then analyzed colorimetrically for organic matter [34]. Soil pH was determined in solution (1:1 soil/water) using a PHS-3C pH meter (Leici, Shanghai, China).
Floodwater was extracted with a 25-mL syringe, and pore water was extracted with Rhizon samplers. Then, water samples were filtered (0.45-µm membrane filters) and measured for pH, Eh, and DRP.

Statistical Analysis
SPSS 17.0 was applied for statistical analysis. Pearson correlations were used to discuss the relationship between floodwater P and pore water P.

General Soil Properties
Both soil types were alkaline, with pH values ranging from 7.31 to 7.96 (Table 2). According to the international classification of soil texture, both soils were silt loams at the 0-45 cm depth, whereas the 45-60 cm depth interval was a loam in the vegetable-growing soil (VS) and a sandy loam in the wheat-growing soil (WS). The clay content varied from 34.6% to 59.2% in VS and from 17.23% to 60.7% in WS and was higher in VS. Therefore, the specific surface area of VS was larger, and its P adsorption should be higher. TP, organic matter (OM), and Olsen P in both soils decreased progressively with soil depth. TP and Olsen P were higher in VS than WS. Furthermore, Olsen P in VS was 106.9-274.2 mg/kg at 0-35 cm. The content of Olsen P in surface soil was 79.87-210.32 mg/kg, which was higher than that in the research of Amarawansha et al. [18]. This will lead to increased soil P leaching into flood water and pore water and increased the risk of eutrophication of rivers and lakes from agricultural non-point source pollution. Previous studies in flooded soils found that the P content of standing floodwater was low in unfertilized soils. Tian et al. found that in fertilized acid soils, the P content of floodwater was higher than in unamended soils [15]. Amarawansha et al. also found that the P content of floodwater and pore water in manured alkaline soil was significantly higher than in unamended alkaline soil [18]. The effects of long-term flooding, draining, and reflooding on pH, Eh, and DRP in floodwater are shown in Figures 1 and 2.
During the flooding period, floodwater pH increased significantly in both VS and WS, from 7.61 to 8.55 and from 7.48 to 8.35, respectively ( Figure 1). Floodwater pH increased during the flooding and reflooding periods due to long-term exposure to the atmosphere [18]. After draining, the floodwater pH, which was initially higher than 8, decreased slightly during the reflooding period. Due to higher pH in floodwater, ion exchange occurred between OHand phosphate in soils, which released P into floodwater. Floodwater Eh of both soils decreased over time during flooding, which led to the release of P banded by iron [21,23,35].

Standing Floodwater P Following Long-Term Flooding, Draining, and Reflooding
Previous studies in flooded soils found that the P content of standing floodwater was low in unfertilized soils. Tian et al. found that in fertilized acid soils, the P content of floodwater was higher than in unamended soils [15]. Amarawansha et al. also found that the P content of floodwater and pore water in manured alkaline soil was significantly higher than in unamended alkaline soil [18]. The effects of long-term flooding, draining, and reflooding on pH, Eh, and DRP in floodwater are shown in Figures 1 and 2.
During the flooding period, floodwater pH increased significantly in both VS and WS, from 7.61 to 8.55 and from 7.48 to 8.35, respectively ( Figure 1). Floodwater pH increased during the flooding and reflooding periods due to long-term exposure to the atmosphere [18]. After draining, the floodwater pH, which was initially higher than 8, decreased slightly during the reflooding period. Due to higher pH in floodwater, ion exchange occurred between OHand phosphate in soils, which released P into floodwater. Floodwater Eh of both soils decreased over time during flooding, which led to the release of P banded by iron [21,23,35].  During the flooding period, the DRP concentrations in floodwater decreased significantly over time, likely due to the effect of soil adsorption. In the early period, the concentration of DRP was high: 11.88 mg/L and 21.92 mg/L in VS and WS, respectively. However, it was only 1.39 mg/L and 3.28 mg/L, respectively, on day 35. DRP decreased 88.3% in VS and 85.0% in WS. The result was

Standing Floodwater P Following Long-Term Flooding, Draining, and Reflooding
Previous studies in flooded soils found that the P content of standing floodwater was low in unfertilized soils. Tian et al. found that in fertilized acid soils, the P content of floodwater was higher than in unamended soils [15]. Amarawansha et al. also found that the P content of floodwater and pore water in manured alkaline soil was significantly higher than in unamended alkaline soil [18]. The effects of long-term flooding, draining, and reflooding on pH, Eh, and DRP in floodwater are shown in Figures 1 and 2.
During the flooding period, floodwater pH increased significantly in both VS and WS, from 7.61 to 8.55 and from 7.48 to 8.35, respectively ( Figure 1). Floodwater pH increased during the flooding and reflooding periods due to long-term exposure to the atmosphere [18]. After draining, the floodwater pH, which was initially higher than 8, decreased slightly during the reflooding period. Due to higher pH in floodwater, ion exchange occurred between OHand phosphate in soils, which released P into floodwater. Floodwater Eh of both soils decreased over time during flooding, which led to the release of P banded by iron [21,23,35].  During the flooding period, the DRP concentrations in floodwater decreased significantly over time, likely due to the effect of soil adsorption. In the early period, the concentration of DRP was high: 11.88 mg/L and 21.92 mg/L in VS and WS, respectively. However, it was only 1.39 mg/L and 3.28 mg/L, respectively, on day 35. DRP decreased 88.3% in VS and 85.0% in WS. The result was During the flooding period, the DRP concentrations in floodwater decreased significantly over time, likely due to the effect of soil adsorption. In the early period, the concentration of DRP was high: 11.88 mg/L and 21.92 mg/L in VS and WS, respectively. However, it was only 1.39 mg/L and 3.28 Water 2017, 9, 531 6 of 13 mg/L, respectively, on day 35. DRP decreased 88.3% in VS and 85.0% in WS. The result was consistent with those of Amarawansha et.al [18] but different from those of Tian [36]. Tian et al. found that DRP in floodwater decreased with time of flooding in amended acid soils. While, Amarawansha et al. found that DRP in surface floodwater increased within a few weeks of flooding, reached a peak, and then decreased slightly with time of flooding in most amended soils [18]. This could be a result of differences in fertilizing methods or maybe the soil texture and characteristics. In the experiment of Amarawansha, soils were amended first and then kept at room temperature for 4 weeks with field moisture before flooding. So, amended P was mostly adsorbed by soils in the early days and then released with the flooding days. In contrast, in our experiment the soils were amended by KH 2 PO 4 solution for 3 days and then flooded to simulate the effect of flooding on soils. Within a few weeks of flooding, DRP in floodwater decreased. This could be the result of soil adsorption, and adsorption was likely the dominant process.
During the flooding period, DRP was higher in WS than in VS at the same fertilization rate. This meant that the P adsorption capacity was higher in VS than in WS, which could be due to the higher clay content of VS and thus larger specific surface area, or maybe because of metal ions, such as Fe, Al and Ca et al. [11,37]. During reflooding, DRP concentrations in standing floodwater on both soils were less than 0.87 mg/L, which was 3-25 times less than those during the flooding period ( Figure 1). DRP increased slightly during reflooding. It was probably because of P being released into floodwater by soil. Furthermore, DRP concentrations were higher in VS than in WS during reflooding. That was to say that P content release from VS was more than that from WS. So, it could be noted that the ability of VS to release P was greater than that of WS.

Phosphorus Distribution in Pore Water
Regardless of WS or VS, DRP concentrations in floodwater and pore water were positively correlated (r = 0.41-0.81, p = 0.01) (Figure 3). During both flooding and reflooding periods in WS, floodwater DRP was higher than pore water DRP at 5 cm, which induced P diffusion from floodwater to pore water. In contrast, in VS, pore water DRP was higher than floodwater DRP after 28 d of flooding, including during reflooding. There was efficient P diffusion from pore water to floodwater as was also shown by Young et al. [10] and Amarawansha et al. [18]. Young and Ross found that pore water P increased with increasing Fe 2+ concentration. This was because Fe-P in flooded soil dissolved under anaerobic conditions, which promoted P release into pore water [38][39][40].
Water 2017, 9,531 6 of 13 consistent with those of Amarawansha et.al [18] but different from those of Tian [36]. Tian et al. found that DRP in floodwater decreased with time of flooding in amended acid soils. While, Amarawansha et al. found that DRP in surface floodwater increased within a few weeks of flooding, reached a peak, and then decreased slightly with time of flooding in most amended soils [18]. This could be a result of differences in fertilizing methods or maybe the soil texture and characteristics. In the experiment of Amarawansha, soils were amended first and then kept at room temperature for 4 weeks with field moisture before flooding. So, amended P was mostly adsorbed by soils in the early days and then released with the flooding days. In contrast, in our experiment the soils were amended by KH2PO4 solution for 3 days and then flooded to simulate the effect of flooding on soils.
Within a few weeks of flooding, DRP in floodwater decreased. This could be the result of soil adsorption, and adsorption was likely the dominant process. During the flooding period, DRP was higher in WS than in VS at the same fertilization rate. This meant that the P adsorption capacity was higher in VS than in WS, which could be due to the higher clay content of VS and thus larger specific surface area, or maybe because of metal ions, such as Fe, Al and Ca et al. [11,37]. During reflooding, DRP concentrations in standing floodwater on both soils were less than 0.87 mg/L, which was 3-25 times less than those during the flooding period ( Figure 1). DRP increased slightly during reflooding. It was probably because of P being released into floodwater by soil. Furthermore, DRP concentrations were higher in VS than in WS during reflooding. That was to say that P content release from VS was more than that from WS. So, it could be noted that the ability of VS to release P was greater than that of WS.

Phosphorus Distribution in Pore Water
Regardless of WS or VS, DRP concentrations in floodwater and pore water were positively correlated (r = 0.41-0.81, p = 0.01) (Figure 3). During both flooding and reflooding periods in WS, floodwater DRP was higher than pore water DRP at 5 cm, which induced P diffusion from floodwater to pore water. In contrast, in VS, pore water DRP was higher than floodwater DRP after 28 d of flooding, including during reflooding. There was efficient P diffusion from pore water to floodwater as was also shown by Young et al. [10] and Amarawansha et al. [18]. Young and Ross found that pore water P increased with increasing Fe 2+ concentration. This was because Fe-P in flooded soil dissolved under anaerobic conditions, which promoted P release into pore water [38][39][40].  During reflooding, DRP in pore water decreased over time. In both WS and VS, DRP in pore water decreased gradually along the soil profile. The effect of diffusion was less during the reflooding period than during the flooding period because of the lower DRP concentration in floodwater. Pore water DRP was lower during the reflooding period than during the flooding period. The observed pattern was very different in VS. Furthermore, pore water DRP was higher in During reflooding, DRP in pore water decreased over time. In both WS and VS, DRP in pore water decreased gradually along the soil profile. The effect of diffusion was less during the reflooding period than during the flooding period because of the lower DRP concentration in floodwater. Pore water DRP was lower during the reflooding period than during the flooding period. The observed pattern Water 2017, 9, 531 7 of 13 was very different in VS. Furthermore, pore water DRP was higher in VS than in WS. Those differences were probably due to differences in soil adsorption and desorption.
On day 12 of flooding, a higher DRP concentration was observed in pore water at 5 cm and 30 cm, both of which decreased on day 19. In contrast, DRP at 10 cm and 40 cm increased on day 19. There was an opposite trend of DRP between days 12 and 19. There was a higher DRP concentration on day 28 in 20-cm pore water, which was greater than that on days 12 and 19. In contrast, DRP in pore water at the 30-60 cm soil depth was very low and was the same on day 35. Regardless of the flooding treatment, DRP in pore water decreased at the end of flooding.
In VS, after 19 d of flooding, pore water DRP was highest at 5 cm (greater than 1 mg/L), and relatively lower at 20 cm and 60 cm (less than 0.5 mg/L; Figure 4). DRP decreased over time during flooding except at 10, 40, and 50 cm. The slow diffusion of P was also observed. For example, DRP at 5 cm decreased over time during flooding, whereas DRP at 10 cm increased. There was diffusion of P from 5 cm to 10 cm due to a concentration gradient. Under fertilized and flooded conditions, the risk of vertical P movement was increased.
Floodwater DRP was higher in WS than in VS during the flooding period. Due to its larger surface area, P adsorption and release capacities were greater in VS than WS (Tables 1 and 2). During the reflooding period, pore water DRP in WS was relatively lower, which may be due to the effect of soil texture. Clay content was higher in VS than in WS. The porosity of VS was relatively uniform, which promoted P diffusion.
Water 2017, 9, 531 7 of 13 VS than in WS. Those differences were probably due to differences in soil adsorption and desorption. On day 12 of flooding, a higher DRP concentration was observed in pore water at 5 cm and 30 cm, both of which decreased on day 19. In contrast, DRP at 10 cm and 40 cm increased on day 19. There was an opposite trend of DRP between days 12 and 19. There was a higher DRP concentration on day 28 in 20-cm pore water, which was greater than that on days 12 and 19. In contrast, DRP in pore water at the 30-60 cm soil depth was very low and was the same on day 35. Regardless of the flooding treatment, DRP in pore water decreased at the end of flooding.
In VS, after 19 d of flooding, pore water DRP was highest at 5 cm (greater than 1 mg/L), and relatively lower at 20 cm and 60 cm (less than 0.5 mg/L; Figure 4). DRP decreased over time during flooding except at 10, 40, and 50 cm. The slow diffusion of P was also observed. For example, DRP at 5 cm decreased over time during flooding, whereas DRP at 10 cm increased. There was diffusion of P from 5 cm to 10 cm due to a concentration gradient. Under fertilized and flooded conditions, the risk of vertical P movement was increased.
Floodwater DRP was higher in WS than in VS during the flooding period. Due to its larger surface area, P adsorption and release capacities were greater in VS than WS (Tables 1 and 2). During the reflooding period, pore water DRP in WS was relatively lower, which may be due to the effect of soil texture. Clay content was higher in VS than in WS. The porosity of VS was relatively uniform, which promoted P diffusion.

Soil P Following Long-Term Flooding, Draining, and Reflooding
The soil O-P fraction was the smallest inorganic P fraction in both soils (less than 35 mg/kg), followed by Fe-P and Al-P, which accounted for 2.6-14.3% and 3.0-13.1% of total inorganic P, respectively ( Figure 5). Total Ca-P ranged from 67.7% to 91.1%. Ca-P was the dominant fraction in alkaline soils, and Ca compounds in alkaline soils played an important role in P retention and most

Soil P Following Long-Term Flooding, Draining, and Reflooding
The soil O-P fraction was the smallest inorganic P fraction in both soils (less than 35 mg/kg), followed by Fe-P and Al-P, which accounted for 2.6-14.3% and 3.0-13.1% of total inorganic P, respectively ( Figure 5). Total Ca-P ranged from 67.7% to 91.1%. Ca-P was the dominant fraction in alkaline soils, and Ca compounds in alkaline soils played an important role in P retention and most of inorganic P was bound to Ca compounds [41,42]. The highest concentration of the Ca 10 -P fraction (37.1-70.0% of total inorganic P) was observed in WS. The largest fraction in VS was Ca 8 -P (14.2-51.0%), followed by Ca 2 -P and Ca 10 -P. As a result of long-term fertilization and irrigation, total inorganic P content was significantly higher in VS than in WS (0.37-1.71 g/L and 0.38-0.68 g/L, respectively). In both WS and VS, inorganic P fractions decreased along the soil profile. The highest concentrations of inorganic P were observed at 0-10 cm.
After the KH 2 PO 4 solution was added, the Al-P and Ca 10 -P fractions increased during the flooding period, but other soil P fractions decreased in WS ( Figure 5). The Al-P content during the flooding and reflooding periods was 1.7-118.9% higher than before flooding. During both flooding and reflooding periods, Ca 8 -P was 31.9-88.9% and 45.6-93.5%, respectively, lower than before flooding. Except at 30 cm and 40 cm, Ca 10 -P was 14.6-61.9% and 10.8-56.5%, respectively, higher than before flooding.
A different pattern was observed in VS ( Figure 5). The Al-P and Ca8-P fractions increased, while other P fractions decreased. The Al-P content during the flooding and reflooding periods was 18-97% more than that before flooding. During both flooding and reflooding periods, Ca8-P at 10-40 cm was 16-64% and 11-27% higher, respectively, than before flooding. Ca10-P at 20-40 cm was 7-17% lower than before flooding but increased at 5, 10, 50, and 60 cm.
The increased concentrations of Al-P, Ca8-P, and Ca10-P may be due to fertilizer inputs and transformations among these fractions. After fertilizing, DRP was tied up by Al and Ca, which increased the concentrations of Al-P and Ca-P. Flooding and draining changed conditions in the soil-water system, which greatly affected the transformations among inorganic phosphorus fractions, especially calcium phosphate. Ca2-P and Ca8-P were the effective P sources in alkaline soils and were readily soluble [43]. Leher et al. [44] and Lindsay et al. [45] reported that calcium phosphate occurred mainly as Ca2-P, Ca8-P, and Ca10-P. Ca2-P can transform into Ca8-P, and Ca8-P can transform into Ca10-P. The increase in Ca10-P in WS may be due to the transformation of Ca2-P and Ca8-P. However, total Ca-P at 0-30 cm in both soils decreased slightly after reflooding, compared to flooding.
In both soils, Fe-P decreased after flooding. When soil is subjected to reducing conditions, Fe 3+ is reduced to Fe 2+ , and the associated phosphorus is released, so Fe-P decreases [21,23,36]. After 10 d of draining, the concentrations of Fe-P, Al-P, and O-P during the reflooding period were greater than those during the flooding period. Several studies have examined P transformations in flooded-drained soils [9,14,40,46,47]. Furthermore, Ca-P was different in the four soils and increased only in the soil with high initial Ca-P. The differences in these findings may be due to differences in soil management, fertilizer type, incubation length, and initial soil characteristics.

Relationships between Olsen P and Dissolved P in Pore Water
Olsen P is an important index for evaluating the capacity of a basic (high pH) soil to supply P, and it can predict P losses to surface water and groundwater [48,49]. Changes in Olsen P are shown in Figure 6. The initial Olsen P concentration was higher in VS than in WS: 12.0-274.2 mg/kg versus 10.1-92.5 mg/kg. In both soils, Olsen P was highest at 0-10 cm and decreased with depth. After  The increased concentrations of Al-P, Ca 8 -P, and Ca 10 -P may be due to fertilizer inputs and transformations among these fractions. After fertilizing, DRP was tied up by Al and Ca, which increased the concentrations of Al-P and Ca-P. Flooding and draining changed conditions in the soil-water system, which greatly affected the transformations among inorganic phosphorus fractions, especially calcium phosphate. Ca 2 -P and Ca 8 -P were the effective P sources in alkaline soils and were readily soluble [43]. Leher et al. [44] and Lindsay et al. [45] reported that calcium phosphate occurred mainly as Ca 2 -P, Ca 8 -P, and Ca 10 -P. Ca 2 -P can transform into Ca 8 -P, and Ca 8 -P can transform into Ca 10 -P. The increase in Ca 10 -P in WS may be due to the transformation of Ca 2 -P and Ca 8 -P. However, total Ca-P at 0-30 cm in both soils decreased slightly after reflooding, compared to flooding.
In both soils, Fe-P decreased after flooding. When soil is subjected to reducing conditions, Fe 3+ is reduced to Fe 2+ , and the associated phosphorus is released, so Fe-P decreases [21,23,36]. After 10 d of draining, the concentrations of Fe-P, Al-P, and O-P during the reflooding period were greater than those during the flooding period. Several studies have examined P transformations in flooded-drained soils [9,14,40,46,47]. Furthermore, Ca-P was different in the four soils and increased only in the soil with high initial Ca-P. The differences in these findings may be due to differences in soil management, fertilizer type, incubation length, and initial soil characteristics.

Relationships between Olsen P and Dissolved P in Pore Water
Olsen P is an important index for evaluating the capacity of a basic (high pH) soil to supply P, and it can predict P losses to surface water and groundwater [48,49]. Changes in Olsen P are shown in Figure 6. The initial Olsen P concentration was higher in VS than in WS: 12.0-274.2 mg/kg versus 10.1-92.5 mg/kg. In both soils, Olsen P was highest at 0-10 cm and decreased with depth. After flooding and reflooding, Olsen P increased in VS at all soil depths. Olsen P was lower after reflooding than after flooding, except at 0-20 cm. In WS, Olsen P increased after flooding and reflooding (relative to initial values) at 0-20 cm but decreased below 20 cm. Furthermore, Olsen P was higher after reflooding than after flooding.
Olsen P is also a good index for representing soil nutrient status of high pH soils [47]. According to the UK Agricultural Development and Advisory Service (ADAS), Olsen P concentrations greater than 46 mg/kg indicate a high nutrient status [50]. Figure 6 shows that Olsen P at 0-20 cm in WS and 0-40 cm in VS was greater than 46 mg/kg, which indicated a high P status. Relationships between Olsen P and DRP in pore water can be seen in Figure 7. When Olsen P in VS exceeded 180.7 mg/kg, and when Olsen P in WS exceeded 40.8 mg/kg, the concentration of P in pore water increased significantly, which increased the risk of P leaching. When Olsen P was accumulated to a certain degree, the release of soil P into floodwater and pore water would reach a threshold value. Edwards et al. proposed the concept of the break point of soil P [51], which has been studied by many researchers from North America and Europe [52]. When Olsen-P exceeds this break point, the release of P into floodwater and pore water would increase significantly.
The critical depth in VS for the risk of soil P release was 50 cm during flooding and 28 cm during reflooding. The critical depth in WS was 18 cm during flooding and 20 cm during reflooding (Figure 7). The ability of VS to release P was greater than that of WS during flooding. In VS, the risk of P release may increase during flooding, whereas it may decrease during reflooding after draining. Wright et al. also found that P availability was significantly higher in flooded soils and decreased after draining [53]. Olsen P at 20-60 cm decreased during flooding and reflooding. This was due to P re-adsorption by amorphous Fe and Al minerals [54]. Below this critical depth, the risk of P release may increase significantly. This could accelerate the transport of pore water P to deeper soils and increase the likelihood of groundwater P pollution.

Conclusions
Based on the results of this study, we concluded that flood regimes strongly influence P release and transformations in flooded soils. Under flooded conditions, DRP in floodwater decreased significantly. During all periods, the concentration of DRP in pore water was highest at 5 cm soil depth. Under fertilized and flooded conditions, the risk of vertical P movement in 10-50 cm soil depth was increased. Flooding increased the concentration of Al-P and decreased the concentrations of Fe-P, O-P, and Ca2-P. Reflooding increased the concentrations of Fe-P, Al-P, and O-P. Olsen P was closely related to DRP in pore water. When Olsen P exceeded 180.7 mg/kg in VS and 40.8 mg/kg in WS, the release of soil P into floodwater and pore water increased significantly.
These findings have substantial implications in understanding how flooding and draining affect soil P release and transformations in flood zones. After fertilizing and then flooding, vertical P movement increased at deeper soil depths, which increased the potential for groundwater pollution. Further research is needed to better estimate the P loss potential, P vertical diffusion, and its contributions to eutrophication. 41202179; 51309126) and the Research Foundation of Jiangsu Normal University (09XLR22). The experimental work was conducted in Jiangsu Normal University and data analysis was done while JT was hosted as a visiting scientist in the Biological and Agricultural Engineering Department at Texas A&M University. The authors acknowledge three anonymous reviewers for providing constructive critique that improved the final quality of the manuscript.

Conclusions
Based on the results of this study, we concluded that flood regimes strongly influence P release and transformations in flooded soils. Under flooded conditions, DRP in floodwater decreased significantly. During all periods, the concentration of DRP in pore water was highest at 5 cm soil depth. Under fertilized and flooded conditions, the risk of vertical P movement in 10-50 cm soil depth was increased. Flooding increased the concentration of Al-P and decreased the concentrations of Fe-P, O-P, and Ca2-P. Reflooding increased the concentrations of Fe-P, Al-P, and O-P. Olsen P was closely related to DRP in pore water. When Olsen P exceeded 180.7 mg/kg in VS and 40.8 mg/kg in WS, the release of soil P into floodwater and pore water increased significantly.
These findings have substantial implications in understanding how flooding and draining affect soil P release and transformations in flood zones. After fertilizing and then flooding, vertical P movement increased at deeper soil depths, which increased the potential for groundwater pollution. Further research is needed to better estimate the P loss potential, P vertical diffusion, and its contributions to eutrophication. 41202179; 51309126) and the Research Foundation of Jiangsu Normal University (09XLR22). The experimental work was conducted in Jiangsu Normal University and data analysis was done while JT was hosted as a visiting scientist in the Biological and Agricultural Engineering Department at Texas A&M University. The authors acknowledge three anonymous reviewers for providing constructive critique that improved the final quality of the manuscript.

Conclusions
Based on the results of this study, we concluded that flood regimes strongly influence P release and transformations in flooded soils. Under flooded conditions, DRP in floodwater decreased significantly. During all periods, the concentration of DRP in pore water was highest at 5 cm soil depth. Under fertilized and flooded conditions, the risk of vertical P movement in 10-50 cm soil depth was increased. Flooding increased the concentration of Al-P and decreased the concentrations of Fe-P, O-P, and Ca 2 -P. Reflooding increased the concentrations of Fe-P, Al-P, and O-P. Olsen P was closely related to DRP in pore water. When Olsen P exceeded 180.7 mg/kg in VS and 40.8 mg/kg in WS, the release of soil P into floodwater and pore water increased significantly.
These findings have substantial implications in understanding how flooding and draining affect soil P release and transformations in flood zones. After fertilizing and then flooding, vertical P movement increased at deeper soil depths, which increased the potential for groundwater pollution. Further research is needed to better estimate the P loss potential, P vertical diffusion, and its contributions to eutrophication.
Author Contributions: J.T. conducted the experiment, analyzed the data, and wrote the first draft of the manuscript; G.D. assisted in data analysis and edited the manuscript; R.K. edited the manuscript and provided critical evaluation of the work; L.L. edited the manuscript; R.D.H. edited the manuscript and provided valuable inputs to analysis.

Conflicts of Interest:
The authors declare no conflict of interest.