Estimation of P retention capacity by the water content of soil kept with a saturated NaCl solution in a desiccator

ABSTRACT Phosphorus (P) retention capacity is an important soil parameter for site-specific nutrient management in croplands as well as a diagnostic criterion for classifying Andic properties. Air-dried soil water content is positively correlated with P retention capacity and has been proposed as a simple P retention capacity index; however, air humidity can affect the water content of air-dried soils. Therefore, in the present study, we 1) tested whether the water content of air-dried soils can be stabilized when they are kept with a saturated sodium chloride (NaCl) solution in a closed desiccator and 2) developed a model for predicting P retention capacity across various soil types. We tested 306 soil samples from paddy rice fields in Madagascar, exhibiting P retention capacities of 10.1–96.1%. Placing a saturated NaCl solution with the soil samples in a closed desiccator for one week kept the relative humidity inside the desiccator at 57–66%, regardless of the temperature and relative humidity outside the desiccator. The accuracy of the model predicting P retention capacity by soil water content was higher when soils were kept with the saturated NaCl solution [R2 = 0.870; root mean square error (RMSE) = 8.30] than without the solution (R2 = 0.812; RMSE = 9.96) at room temperature. The high correlation between P retention capacity and soil water content was mainly attributed to Al oxides. Additionally, the reproducibility of soil water content was enhanced by this method under different relative humidity conditions. However, compared with room temperature, controlling the temperature at 15°C and 30°C did not further improve the accuracy of the model when the soils were kept in the desiccator with saturated NaCl solution, although soil water content was higher at 30°C than at 15°C and room temperature. Therefore, we conclude that soil P retention capacity can be simply and accurately estimated with high reproducibility according to the water content of soils kept with saturated NaCl solution in a closed desiccator for one week at a room temperature.


Introduction
Phosphorus (P) retention capacity is an important agronomic soil parameter for determining crop responses to P fertilizer inputs under P deficient condition (Johnston et al. 2014;Nishigaki et al. 2021) and a diagnostic criterion of Andic properties for classifying volcanic soils (Soil Survey Staff 2014b;WRB 2015).Thus, the development of a simple and reliable methodology for estimating the P retention capacity of soils can provide important information about targeted fields.This is especially relevant for subsistent smallholder farmers in sub-Saharan Africa who rely on relatively heterogenous soils with high P retention capacity for food production (Batjes 2011;Ringeval et al. 2017).Additionally, these farmers often lack the financial capacity to purchase commercial fertilizer (Demay et al. 2023;Tsujimoto et al. 2019).Mogollón et al. (2021) emphasized that improving P management in sub-Saharan Africa is a critical challenge for not only achieving regional food security but also reducing environmental risks, such as cropland expansion and excess P outflows into freshwater and coastal ecosystems.
P retention capacity is known to be controlled primarily by oxalate-extractable aluminum (Al), such as poorly crystalline Al hydroxide and boehmite (Ando et al. 2021;Nidhi, Hesterberg, and Martin 2005), and other P-binding Fe oxides and clay minerals in soils (Gérard 2016).Given that the volume of micro-and small mesopores (0.7-4.0 mm) increases when the content of Al oxides/hydroxides increases (Watanabe 2017), a significant correlation exists between the oxalateextractable Al content and water content of air-dried soils (Nishigaki et al. 2021).Thus, a simple method of estimating P retention capacity through the water content of air-dried soils was proposed in previous studies (Kinoshita and Tani 2020;Nishigaki et al. 2021).In these studies, soil samples were pre-treated by drying them in an oven at 30-60°C for 3-7 days, resulting in less than 1% of soil water content.The methodology can be applied easily and safely, even in areas with little access to chemical wet analysis resources.However, a technical disadvantage of this procedure, which can lead to measurement error, is the effect of atmospheric humidity on the low water content of air-dried soils and its interaction with the temperature where the soil samples are dried (Andraski and Scanlon 2002).The correlation coefficient between P retention capacity and air-dried soil water content is also influenced by soil characteristics, such as organic matter, clay, and oxalateextractable Al contents.For instance, Kinoshita and Tani (2020) found that the regression slopes for volcanic and nonvolcanic soils differed due to differences in their organic matter contents and specific surface areas.Similarly, Hashimoto et al. (2012) reported that organic matter can have both enhancing and inhibitory effects on P sorption to Fe and Al minerals.However, no attempts have been made to investigate the factors that influence air-dried soil water content under controlled temperature and relative humidity conditions, or how these factors impact estimation models.This is crucial for developing a robust model that can be applied to various soil types and for the comparison of data obtained under different temperature and relative humidity conditions.
In the present study, we stabilized the water content of soil samples using the vapor equilibration technique (Scanlon, Andraski, and Bilskie 2002), involving the equilibration of soil samples with salt solutions of known osmotic potential in a closed desiccator.Unlike previous studies (Kinoshita and Tani 2020;Nishigaki et al. 2021) that required a drying oven for sample preparation, our approach offers an advantage for investigators without access to such experimental facilities.Saturated salt solutions maintain a constant relative humidity in a closed small chamber, which typically depends on temperature.However, the relative humidity with saturated sodium chloride (NaCl) solution is known to be less temperaturedependent, varying within 75.4%±2% over a range of 10°C to 40°C (Greenspan 1977).We hypothesized that the effect of ambient humidity and its interaction with temperature can be mitigated by keeping soils with a saturated NaCl solution and that P retention capacity predictions can be improved through the stabilization of soil water content.
The aims of this study were as follows: (1) to determine whether soil water content is stabilized when soils are placed in a closed desiccator with a saturated NaCl solution under different temperatures and relative humidities and 2) to develop a simple and reliable prediction model for P retention capacity that is applicable to various soil types.

Materials and Methods
Soil samples were collected from 306 paddy rice fields in 6 districts, i.e., Faratsiho, Ambohibary, Ankazomiriotra, Antanifotsy, Behenjy, and Betafo, which are widely distributed in the central highlands of Madagascar (979 to 1945 m a.s.l.).The soil types in these paddy rice fields were classified as Fluvisols or Cambisols according to our previous soil profile survey (Nishigaki et al. 2020).In each field, a surface layer with a depth of 0-15 cm was collected as five composites.The collected soil samples were air-dried, sieved to pass through a 2 mm mesh screen, and kept in plastic bags until the following analyses and experiments were conducted.
Soil pH, particle size distribution, and total carbon (C) content were measured as the basic soil properties.Soil pH was determined in 0.01 M CaCl 2 at a soil-to-solution ratio of 1.0:2.5 (Hendershot, Lalande, and Duquette 2008).Soil particle size distribution was determined using the wet-sieving and pipette method (Gee and Bauder 1986).Total C content was quantified using the dry combustion method with an NC analyzer (Sumigraph NC-220F; Sumika Chemical Analysis Service, Ltd., Osaka, Japan).
P retention capacity was measured using a method described by Soil Survey Staff (2014a).Briefly, 5.0 g of soil was shaken in a 25 mL aliquot of 1000 mg L −1 P solution (containing KH 2 PO 4 , CH 3 COONa, and CH 3 COOH; pH 4.6), for 24 h.The aliquot was centrifuged and filtered, and the P concentration in the aliquot was quantified using a colorimetric method with nitric vanadomolybdate acid reagent at an absorbance of 466 nm.The P retention capacity was then determined as the initial P concentration minus the P remaining in the sample solution and represented as the percentage of P retained.Extractable Al and iron (Fe) contents were determined using the acid ammonium oxalate method (Al ox and Fe ox , respectively) described by Courchesne and Turmel (2008).The Al and Fe concentrations in the oxalate extraction were measured using an inductively coupled plasma mass spectrometer (ICPE-9000; Shimadzu Corp., Kyoto, Japan).
Two experiments were conducted using the soil samples in this study.The first experiment was aimed to evaluate the effect of placing soils in a closed plastic desiccator on soil water contents under different temperatures.Air-dried and sieved soils (3.0 g) were weighed in 15 mL porcelain crucible containers (C0; Nikkato Co., Ltd., Osaka, Japan), and the weights of the containers were recorded to a precision of 0.0001 g.The containers were placed inside and outside a closed plastic desiccator (35 × 35 × 40 cm, IWH; AS ONE Corporation, Osaka, Japan), respectively, in an incubator (CLE-303; Tomy Seiko Co., Ltd., Tokyo, Japan) (Figure 1).The weights of the containers were measured again in the same manner after keeping them for one week at 15°C, followed by another week at 30°C, respectively.In a preliminary trial, we confirmed that three days were sufficient time to reach a constant relative humidity (Supplemental figure).Another container was placed in a plastic desiccator outside the incubator and kept for one week at room temperature, and its weight was recorded in the same manner.In the desiccator, a glass beaker filled with approximately 200 mL of saturated NaCl solution was placed beside the soil samples to control relative humidity (Figure 1a).The saturated NaCl solution was prepared by adding a quantity of NaCl to distilled water that was apparently greater than its solubility (ca.72 g in 200 g water at 20°C) to prevent desaturation from water vapor absorption and to keep the solution saturated.Each crucible container was then oven-dried at 105°C for 24 h to determine its oven-dried weight.The soil water content (ω; g g −1 ) was calculated as the ratio of the water mass to the total mass of the oven-dried soil.Hereafter, the water content of the soils kept under different conditions are referred to as ω(15, In), ω(15, Out), ω(30, In), ω(30, Out), ω (Room, In), and ω(Room, Out), respectively, according to the combination of the temperature treatment and the NaCl treatment placing inside or outside a plastic desiccator (Table 1).The water content of soils after air-drying and sieving without any pre-treatment corresponds to ω(Room, Out).The temperature and relative humidity inside the plastic desiccator were continuously measured with a temperature-humidity sensor (THA-3001; T&D Corporation, Nagano, Japan) and recorded every 30 min via a recorder (TR-72Ui; T&D Corporation).The relative humidity inside the incubator but outside the desiccator was intermittently measured and recorded in the same manner, whereas the temperature was continuously measured with a thermometer (TR-5106; T&D Corporation) and recorded every 30 min via a recorder (TR-52i; T&D Corporation).Owing to the limited volume of the incubator, the soil samples were divided into four batches for incubator-based measurements, and soil water content was successively measured in these batches.
The second experiment was conducted to evaluate the effect of placing soils in a plastic desiccator on soil water contents under different relative humidities and its correlation with P retention capacity.We selected 20 soil samples with varying P retention capacity (ranging from 19.5% to 94.1%) out of 306 samples.Air-dried soils were weighed into 15-mL crucible containers, and they were placed inside and outside a closed plastic desiccator containing a saturated NaCl solution, respectively, in a growth chamber (LH-411PFQDT-SP, Nippon Medical & Chemical Instruments Co., Ltd., Osaka, Japan) (Figure 1b).The temperature in the chamber was set at 20°C throughout the experimental period, and the relative humidity in the growth chamber was kept at three different levels, i.e., low (41%), middle (52%), and high (64%) for 72 h respectively.The weight of each container was recorded at the end of each relative humidity level.Each container was then oven-dried at 105°C for 24 h, and the soil water content was calculated.The temperature and relative humidity inside and outside the plastic desiccator were continuously measured with temperaturehumidity sensors (THA-3001) and recorded every 5 min via a recorder (TR-72Ui).
Statistical analyses were performed using JMP14 software (SAS Institute Inc., Cary, NC, U.S.A.).A Pearson correlation analysis was performed to assess the relationship among the soil properties and between the soil water contents and the soil properties, at a significance level of p < 0.01.Analysis of variance (ANOVA) was conducted to determine the single and interactive effects of the temperature and NaCl treatments on soil water content in the first experiment.A post-hoc mean comparison of the water contents of soils kept under different conditions was performed using Tukey's honestly significant difference (HSD) test.Linear regression models were also developed considering the water contents of soils kept at room temperature and 15°C or 30°C and placed either inside or outside the desiccator.An exponential function model, which defines an asymptote, was used to determine P retention capacity as follows; where Pret is soil P retention capacity, a is an asymptote, b and c are coefficients, and X is any measured soil water content and Al ox .The accuracy of the models was assessed using the coefficient of determination (R 2 ) and root mean square error (RMSE) or root mean square percentage error.Coefficient of variation was calculated for soil water contents at different relative humidities (low, middle, high) for each soil sample in the second experiment.Each sample was analyzed without replication for all measures.

Soil properties
The soil samples exhibited a range of soil properties (Table 2).P retention capacity of 306 samples substantially varied among the samples.Most soil samples were acidic, and soil texture varied from clay to sandy loam.Total C content varied from an extremely low (0.3%) to high (13.4%)level.The Al ox +0.5 Fe ox was >20 g kg −1 , i.e., the diagnostic criteria of Andic properties (WRB 2015), in 12.4% of samples.The correlation analysis showed a significant positive correlation between P retention capacity and silt content, total C, Al ox , and Fe ox , while there was a negative correlation with sand content (Table 3).The correlation coefficient between P retention capacity and Al ox was higher than that with Fe ox .

First experiment
The temperature was consistently within ± 1°C of the set temperature in all treatments while samples were kept either inside or outside the desiccator in the incubator.
The temperature inside the plastic desiccator kept at room temperature was 23.3-26.2°Caccording to the daily fluctuation in the external temperature.The relative humidity inside the desiccator had reached a steady state at the termination of preparation, varying from 57% to 66% among the measurement batches.In contrast, the relative humidity outside the plastic desiccator but inside the incubator varied substantially from 27% to 89% during incubation.Room temperature did not affect the relative humidity inside the plastic desiccator.
According to ANOVA results, the single and interactive effects of temperature and saturated NaCl solution treatment on soil water content were significant.According to Tukey's HSD test, ω(30, In) had the highest soil water content followed by ω(30, Out), and the other soil water contents had no significant difference among them (Figure 2).The soil water contents under different temperature conditions were highly and linearly correlated when the humidity was controlled inside the desiccator (Figure 3a).Although the mean value of ω(30, In) was 2.21 times higher than that of ω(15, In), the values between ω(15, In) and ω(Room, In) were clearly fitted with a 1:1 linear regression line.The correlations among the different temperature conditions were weakened outside the desiccator (Figure 3b).All the soil water contents had significant correlations with P retention capacity and other soil properties except clay content (Table 4).The soil water contents were more strongly correlated with Al ox than Fe ox .
The exponential function model of P retention capacity exhibited high R 2 and low RMSE with ω(Room, Out) (Figure 4a).Additionally, model accuracy was improved when the soil samples were placed inside the desiccator, ω(Room, In), compared with outside the desiccator, ω (Room, Out) (Figure 4b), the R 2 and RMSE values of which were comparable to those of the model using Al ox (Figure 4e).However, model fitting was not further improved when temperature was controlled at 15°C and 30°C,i.e.,ω(15,In) and ω(30, In) (Figure 4c,d).

Second experiment
The soil water contents were greatly affected by relative humidity when the soils were placed outside the desiccator (Figure 5); the higher the relative humidity, the higher the soil water content with the same value of P retention capacity.Their coefficient of variations (CVs) ranged from 4.6% to 13.2% with the mean of 9.2%.The CVs were substantially reduced when the soils were placed inside the desiccator, ranging from 0.9% to 2.9% with the mean of 1.8%, indicating high reproducibility under a range of relative humidity conditions.The average relative humidity inside the desiccator was 43%, 50%, and 53% when the relative humidity outside the desiccator was set at low (41%), middle (52%), and high (64%), respectively.The temperature remained constant within ± 0.5°C of 19.5°C inside and outside the desiccator.

Discussion
A saturated NaCl solution placed with soil samples in a closed desiccator improved the regression models of soil water content for predicting P retention capacity (Figure 4).Furthermore, the variability in the water content of the soil samples was clearly mitigated in the closed desiccator under un-controlled room condition and different relative humidities (Figures 3 and 5).Therefore, the accuracy and the reproducibility of modeling P retention capacity by soil water content were improved when the soil samples were placed with saturated NaCl solution in a closed desiccator.As a reagent, NaCl is safer than other strong acids or alkalis that may also be used to keep a constant water potential (Winston and Bates 1960), and producing a saturated NaCl solution is simple.Therefore, we recommend placing a saturated NaCl solution with soil samples in a closed desiccator to stabilize the water potential of the soil samples.
Regarding estimating P retention capacity by ω(Room, In), the exponential function model was as accurate as the Al ox model, which represents the main sorbent of phosphate in soils (Nidhi, Hesterberg, and Martin 2005) (Figure 4b,e).The linear regression models used in previous studies were soil-type specific, e.g., volcanic and non-volcanic soils (Kinoshita and Tani 2020), hampering the application of the models to sets of soil samples with varying properties.In contrast, the exponential model applied in the present study more suited to predict P retention capacity across a wide range of soil types than a linear regression model previously adopted.This is because P retention capacity cannot exceed 100% due to the single-point adsorption method used to determine it (Soil Survey Staff 2014a).
Compared with room temperature conditions, i.e., ω (Room, In), controlling temperature at 15°C and 30°C did not further improve model accuracy when the soil samples were kept with the saturated NaCl solution (Figure 4b-d).Although soil water content was higher in soil samples kept at 30°C than in those kept at 15°C, the model accuracy was comparable between these conditions.Thus, given the high accuracy of the model with ω(Room, In), we propose using room temperature without requiring strict temperature control and ω(Room, In) as the index for estimating soil P retention capacity.However, if it is not possible to maintain room temperature, specifically within the range of 15-26.2°Ctested in the current study, calibration of the estimation model would be necessary.
The ω(Room, In) had a significant positive correlation with Al ox +0.5 Fe ox but not with clay content (Table 4), suggesting a significant effect of clay mineralogy rather than clay content on water content of air-dried soils.The high correlation between water content of air-dried soils and Al ox content was likely due to increasing volume of micro-and small mesopores (0.7-4 nm) with increasing Al oxides/hydroxides (Watanabe 2017).The strong adsorption of water molecules to the surface of clay minerals was experimentally shown in previous studies (e.g., Goates and John Bennett (1957)), and thus water vapor adsorption used to be a measure of specific surface area in soils (Newman 1983).There were also significant correlations between the water contents of air-dried soils and total C, indicating the possible contribution of organic matter in increasing water adsorption.The higher correlation of Al ox than Fe ox with the water content of air-dried soils suggested the preferential water adsorption of Al minerals over Fe minerals, which is in agreement with our previous study (Nishigaki et al. 2021).
In conclusion, P retention capacity can be accurately estimated by measuring the water content of soils kept with a saturated NaCl solution in a closed desiccator at room temperature for one week.This method demonstrates high reproducibility irrespective of relative humidity conditions, allowing for comparisons of data sets obtained at different times and locations.When combined with microwave oven-drying (Topp and Ferré 2002), it would become possible to estimate P retention capacity without the need for a drying oven.Therefore, this simple but robust model for predicting P retention capacity will facilitate soil analysis toward tailored P fertilizer management at a cropland scale in areas of developing countries with little access to chemical analysis at laboratory.Further investigation is required to determine the applicability of this model to soils other than those in paddy rice fields and with a neutral to high soil pH level, i.e., soils that were not included in the sample set of the present study.Al ox +0.5Fe ox (g g −1 ) (g g −1 ) (g g −1 ) (g g −1 ) (g g −1 ) (%) (%) (%) (%) (%) (g kg −1 ) (g kg −1 ) (g kg −1 ) ω(15, In) (g g −1 )   The relative humidity in the growth chamber was set at three different levels (low, middle, and high); the average of the relative humidity during 72 hours at each level being 41%, 52%, and 64% inside the closed desiccator, and 43%, 50%, and 53% outside the closed desiccator, respectively.Crosses represent the coefficient of variation (CV) of the soil water contents at each P retention capacity.

Figure 1 .
Figure 1.Overview of soil sample preparation in the first experiment (a) and the second experiment (b).

Figure 2 .
Figure 2. Box plots of soil water contents prepared under different conditions (n = 306).Soil water contents are referred as ω(15, In), ω(15, Out), ω(30, In), ω(30, Out), ω (Room, In), and ω(Room, Out), respectively, according to the combination of the temperature treatment (15°C, 30°C, Room temperature) and the NaCl treatment placing inside or outside a plastic desiccator.The cross in each box represents the mean, the central vertical bar shows the median, the box represents the interquartile range, the whiskers show the location of the most extreme data points that are still within a factor of 1.5 of the upper or lower quartiles, and the points are values that fall outside the whiskers.Different letters indicate a significant difference between the soil water contents at the level of p = 0.01 by Tukey's HSD test.

Figure 3 .
Figure 3. Relationships of the water contents of soils kept at room temperature and 15°C or 30°C inside (a) and outside (b) a closed desiccator.

Figure 4 .
Figure 4. Relationships of P retention capacity with the water contents of soils kept under different conditions [ω(Room, Out), ω(Room, In), ω(15, In), and ω(30, In); a -d] and with oxalate-extractable Al content (Al ox ; e).Solid lines represent the regression curves given by the equations above each figure.

Figure 5 .
Figure 5. Relationship between P retention capacity and soil water contents of soil samples placed inside (a) and outside (b) a closed desiccator in a growth chamber.The relative humidity in the growth chamber was set at three different levels (low, middle, and high); the average of the relative humidity during 72 hours at each level being 41%, 52%, and 64% inside the closed desiccator, and 43%, 50%, and 53% outside the closed desiccator, respectively.Crosses represent the coefficient of variation (CV) of the soil water contents at each P retention capacity.

Table 1 .
Abbreviations of soil water contents as the combination of temperature and relative humidity.

Table 3 .
Correlation matrix among the soil properties of all samples.

Table 4 .
Correlation matrix between the soil water contents and the soil properties of all samples.