Characterizing Sorption and Modeling Phosphorus Move- ment on Candler and Immokalee Fine Sand

Laboratory and computer experiments were conducted to 1) determine the effect of supporting electrolyte on mass distribution coefficient (KD) for predicting P movement at 30and 60-cm soil depths using HYDRUS-1D and 2) compare the temporal P concentrations as a function of soil type. The results on Candler fine sand at Lake Alfred showed that P contents for the KD estimated with 0.01 M KCl and 0.005 M CaCl2 were 10 to 15% higher than those predicted with a KD value measured with fertilizer mixture. The predictions on Immokalee fine sand showed that P contents for the KD estimated with fertilizer mixture and 0.005 M CaCl2 were 12 to 20% higher than those predicted with a KD value measured with 0.01 M KCl. The outputs with KD measured with 0.005 M CaCl2 appear to be close to those predicted with a KD measured with fertilizer mixture on Immokalee fine sand. However, the analysis of the KD values across all electrolytes on the two soils studied revealed that 0.01 M KCl is the electrolyte that yields KD values fairly close to fertilizer mixture on Candler fine sand and would be appropriate to use for the coated sand while 0.005 M CaCl2 was comparable with fertilizer mixture on Immokalee fine sand and would be an appropriate electrolyte to use for such uncoated sands. The results of study will help in improving estimation of KDs and description of P transport in citrus production systems of Florida sandy soils. This would provide necessary information for sustainable environmental management and reduce problems of eutrophication and prevent groundwater contamination.


Introduction
Adsorption is the mechanism that soil cations and anions (including various phosphorus (P) compounds) are retained by soil particles. The adsorption process tends to restrict compound's mobility and bioavailability [1]. Thus, the procedure for determining the P sorption isotherms could then provide information on their mobility in the soil. The supporting electrolyte concentration is chosen to mimic that of soil solution. Most commonly 0.01 M CaCl 2 [2,3], 0.01 N CaCl 2 [4], 0.005 M CaCl 2 [1], 0.1 to 3 mM KCl [5], 0.05 M KCl [6,7], and 0.01 M KCl [8,9] have been used as electrolytes in studies on P and K sorption. Nair, et al. (1984) [10] reported that P sorption varies with ionic strength and cation species of the supporting electrolyte. For example, [10] showed that P adsorption was generally lower with K + as the supporting electrolyte cation compared with Ca 2+ . These studies and others have not explained the rationale behind use of a particular electrolyte other than equilibrating the solutions in deionized or tap water.
The chemical characteristics of soils dominating the Flatwoods and Ridge regions of Florida are well described in [11] and some were also determined in this study. The soil at the Flatwoods consists of nearly level and poorly drained on the Flatwoods classified as Immokalee fine sand (sandy, siliceous, hyperthemic Arenic Haplaquods) with the spodic horizon lying within 1 m from the soil surface. The soil at the Ridge, classified as Candler fine sand (hyperthermic, coated Typic Quartzipsamments) is a well-drained sandy soil with no continuous layer limiting vertical water movement [11]. The Immokalee and Candler fine sand are moderately acidic (pH ranging from 4.9 to 5.6) with > 94% sand textural composition, have low organic matter content (ranging from 0.41 to 0.61% on Immokalee fine sand and from 1.6 to 1.96% on Candler fine sand) and low cation exchange capacity (CEC) (ranging from 2 to 6 cmol (+) kg -1 ), have inorganic N in the range of 8.2 and 11.2 mg kg -1 , moderate to very high P (in the range of 28.7 to 46.5 mg kg -1 for Immokalee sand and 112.8 to 115.8 mg kg -1 for Candler fine sand) and K in the range of 11.8 to 15.2 mg kg -1 for Immokalee fine sand and 23.0 to 29.7 mg kg -1 for Candler fine sand (Table 1) [12]. It has been accepted that soil CEC, particularly the CEC contribution from organic matter content (OMC) has a significant influence on the soil P movement [13]. This study sought to 1) determine the effect of supporting electrolyte on the mass distribution coefficient (K D ) for predicting P movement at 30-and 60-cm soil depths using HYDRUS-1D; and 2) compare the P concentrations with time as a function of soil type.

Laboratory adsorption study
The baseline soil chemical and physical properties for soils used in the study are described in (Table 1 and Table 2). Sorption isotherms on the disturbed soil samples (0-15 cm, 15-30 cm) were determined using the batch equilibration procedure [14]. The initial solution concentrations from potassium dihydrogen phosphate (KH-2 PO 4 ) for P in 0.005 M CaCl 2 were 10, 25, 50 mg PL -1 . To determine P sorption isotherms, soil samples were obtained from 5 random positions per site at two depths giving a total of 10 samples. Each sample was weighed in triplicates plus a blank control. A 10 g air-dried, subsample of soil < 2 mm particle size was placed in a centrifuge tube and equilibrated with 20 ml (soil solution ratio 1:2) of 3 initial concentrations of P solutions. The centrifuge tubes were shaken for 24 h, centrifuged for 20 min, and filtered through a filter paper (Whatman, #42). All these procedures were done at room temperature ~25 ± 1 o C as recommended by [14] but the filtrate was later stored at < 4 o C until analyzed for P.

Phosphorus sorption isotherm determination
The solutions from the adsorption study were analyzed for P using ICP-AES and calibration standards of 10, 30 and 50 mg PL -1 preparation in fertilizer mixture (with additions of 0.027 mM NH 4 NO 3 and 0.013 mM KCl), 0.01 M KCl and 0.005 M CaCl 2 . The amount of chemical sorbed to the soil was calculated from the difference between the initial and equilibrium solution concentration: Where S is the adsorbed concentration (mg kg -1 ); V o is the volume of initial solution (L); m is the soil mass (kg); C o is the initial concentration of the standard solution (mg L -1 ), and, C e is the soil solution concentration at equilibrium (mg L -1 ). Soil to water ratio = 1:2 (mass/volume); § OM: organic matter expressed as a percentage; ‡ CEC: cation exchange capacity expressed in cmol (+) kg -1 (CEC reported by Obreza and Collins; 2008); † Mehlich 1 P (mg kg -1 ); ‡ ‡ Mehlich 1 K (mg kg -1 ); ¶ ¶ IN: Inorganic N (mg kg -1 ). Sorption isotherms for P were calculated using the Freundlich equation: Where K f is the Freundlich sorption coefficient (mg 1-N kg -1 L N ) and N is an empirical constant related to adsorption phenomena [15]. The linearized form of the Freundlich equation was used to calculate K f and N: Where S is the adsorbed equilibrium concentration (mg kg -1 ); C e is the equilibrium concentration (mg L -1 ) and C max is the estimated maximum concentration (mg L -1 ) and K f and N are calculated from the intercept and slope of Equation 3. To find average linearized K D for the Freundlich isotherm, the integrated form of the equation was used:

Concepts and governing equations for the simulations
The governing flow equations for water flow and nutrient transport are given by the [16] and convection-dispersion equations (CDE) [17][18][19]: Where θ is the volumetric water content [L 3 L -3 ], h is the pressure head [L], x i (I = 1, 2) are the spatial coordinates [L], t is time [T], A ij K are components of a dimensionless anisotropy tensor A K (which reduces to the unit matrix when the medium is isotropic), K is the unsaturated hydraulic conductivity function (LT -1 ), and s is a sink/source term [L 3 L -3 T -1 ], accounting for root water uptake (transpiration). The sink/source represents the volume of water removed per unit time from a unit volume of soil due to compensated citrus water uptake.
The equation (CDE) governing transport of independent solutes i.e. single-ion transport is given as: Where c 1 and c 2 are solute concentrations in the solid (MM -1 ) and liquid (ML -3 ) phases, respectively; q i is the i th component of volumetric flux density (LT -1 ), Ф is the rate of change of mass per unit volume by chemical or biological reactions or other sources (negative) or sinks (positive) (ML -3 T -1 ), respectively, providing connections between individual chain species, ρ b is the soil bulk density (ML -3 ), D ij is the dispersion coefficient tensor for the liquid phase [L 2 T -1 ]. The term r a represents the root nutrient uptake (ML -3 T -1 ) which is the sum of actual active and passive nutrient uptake. The solid phase concentration, c 1 , accounts for nutrient either sorbed in the solid phase or precipitated in various minerals. This is usually quantified by the adsorption isotherm relating c 1 and c 2 described by the linear equation of the form: Where K D (L 3 M -1 ) is the mass distribution coefficient of species 1. A tracer (e.g. bromide) are assumed to have a K D = 0 cm 3 g -1 . The first order decay constant ranges from 0.36-0.56 d -1 [20]. For P, K D is reportedly in the range of 19 to 185 cm 3 g -1 [21,22]. Bulk density for the soil is in the range 1.59-1.72 g cm -3 (Immokalee) and 1.55-1.93 g cm -3 (Lake Alfred) (T.A. Obreza, unpublished).
The sink term, s, for the Richards equation represents the volume of water removed per unit time from a unit volume of soil due to plant water uptake. Thus, s is defined as: Where the water stress response function ( ) h α is a prescribed dimensionless function of the soil water pressure head, b is the normalized water uptake distribution, L t is the width of the soil surface associated with the transpiration process and T p is the potential transpiration rate (LT -1 ) and w is the water stress index.
The nonlinear, predictive equations for the unsaturated hydraulic function in terms of soil water retention parameters are given by [23] as: Where 1 = 1-, n > 1 m n (11) Where θ r , θ s , K s and l are residual water content (L 3 L -3 ), saturated water content (L 3 L -3 ), saturated hydraulic conductivity (LT -1 ), and pore connectivity parameter (estimated to be an average of 0.5 for many soils respectively). Parameters α (L -1 ) and n are empirical coefficients affecting the shape of the hydraulic functions [23]. We estimated the hydraulic functions α and n after fitting the water content and matric potential data using the van Genuchten model in Community Analyses System (CAS) 2007 [24] developed for determination of soil hydraulic functions (Table 3).

Cumulative flux and concentration
The P concentration and flux were predicted at 30 cm soil depth. Assuming isotropy and a homogeneous profile, the concentration and fluxes at 60 cm soil depth were also predicted.

Characterizing the P isotherms
The results of P adsorption were described by a Freun-dlich model with linearized K D ranging from 0.50 ± 0.19 to 0.75 ± 0.13 kg L -1 for Immokalee fine sand and from   soil pH Ca-P minerals would precipitate while at lower pH (< 4.5), Al-P and Fe-P compounds dictate P adsorption. Barrow, et al. (1980) [25] also showed that at equal ionic strength below pH = 6, there was more phosphate adsorption from CaCl 2 than from NaCl on goethite. This phenomenon, according to Barrow and colleagues [25], is caused because high concentration of positive charges near the negatively charged soil surface may be induced by replacing a monovalent cation with a divalent one and also if the added divalent cation has a specific affinity for the adsorption surface. Addition of cations from the supporting electrolyte, unlike using the fertilizer, induced a greater positive charge for phosphate (and orthophosphate anion) adsorption. The higher K D s for Candler fine sand might be due to high organic matter and some Fe/ Al coatings that might bind P. The high K f value in the top 0-15 cm than the 15-30 cm layer is ascribed to higher organic carbon and organic matter in the former layer resulting in increased P adsorption. Our results are comparable to those reported on Margate sand and Immokalee fine sand by Muwamba, et al. (2016) [26].

Phosphorus movement with microsprinkler irrigation as function of time
Phosphorus movement was predicted using three different K D s estimated with fertilizer mixture, 0.01 M KCl and 0.005 M CaCl 2 for of 21 days, assuming no rainfall events (Figure 3). The assumption is that a K D value ob-1.73 ± 0.15 to 4.43 ± 0.50 kg L -1 for Candler fine sand using a C max of 15 mg L -1 (Figure 1 and Figure 2). Phosphorus adsorption was linear for the P concentration range (0 to 50 mg P L -1 ) studied on the Immokalee sand using fertilizer mixture with K D averaging about 0.44 ± 0.10 kg L -1 (Table 4). Freundlich sorption coefficients (K f ) were lower for Immokalee fine sand than for Candler. High coefficients were observed on Candler fine sand with K f values eightfold greater than that of Immokalee fine sand. The K f value obtained with 0.005 M CaCl 2 was approximately twofold that obtained with 0.01 M KCl and threefold that obtained in the fertilizer mixture suggesting the influence of the cation effect on P adsorption than with water. According to [7], the lower Freundlich sorption coefficients (K f ), indicate low P retention capacity at low P concentrations suggesting that the potential risk of subsurface P movement and leaching would be high when the concentration of P in surface soils is high. The K f and K D values reported in were generally lower than those reported for carbonatic soils in south Florida [7] where K D ranged from 14.8 to 76.3 L kg -1 and K f from 12 to 58 mg 1-N kg -1 L N . However, the results in this study agree with those of other researchers [10,13,25]. According to [13], divalent cations on the CEC enhance P adsorption relative to monovalent cations because they increase the accessibility of (+)-charged edges of clay minerals to P. This occurs at pH < 6.5 (the pH of our soils ranged from 4.9 to 5.6, Table 1, because at greater tained using fertilizer mixture typifies that of field conditions with regard to chemical processes. The results on Candler fine sand at Lake Alfred showed that that P contents for the K D estimated with 0.01 M KCl and 0.005 M CaCl 2 were 10 to 15% higher than those predicted with a K D value measured with fertilizer mixture. The predictions on Immokalee fine sand showed that that P contents for the K D estimated with fertilizer mixture and 0.005 M CaCl 2 were 12 to 20% higher than those predicted with a K D value measured with 0.01 M KCl. The outputs with K D measured with 0.005 M CaCl 2 appear to be close to those predicted with a K D measured with fertilizer mixture on Immokalee fine sand. Thus, the analysis of the K D values across all electrolytes on the two soils studied revealed that 0.01 M KCl is the electrolyte that yields K D values (0.42 to 0.69 L kg -1 on Immokalee fine sand and 0.42 to 1.06 L kg -1 on Candler fine sand) fairly close to fertilizer mixture on coated sands while 0.005 M CaCl 2 tends to give K D values (2.44 to 4.66 L kg -1 on Immokalee fine sand and 2.81 to 4.93 L kg -1 on Candler fine sand) two to threefold in magnitude to those determined with fertilizer mixture on sands with Fe coatings. This suggests that 0.005 M CaCl 2 would tend to overestimate P sorption and retardation during unsaturated or saturated flow on coated sand while estimating the adsorption process satisfactorily on uncoated sands than 0.01 M KCl. It appears the addition of a supporting electrolyte with a divalent or monovalent cation, unlike fertilizer mixture, increases the surface charge for adsorption of orthophosphate anions on sands with coatings. The use of 0.01 M KCl appears not to be influenced by presence of sand coatings compared with CaCl 2 and, thus, would present an appropriate supporting electrolyte for Candler fine sand while CaCl 2 would be appropriate for the Immokalee fine sand.

Conclusion
The results show that P adsorption in the top 0-15 cm was greater for Candler than Immokalee sand using the fertilizer mixture, 0.005 M CaCl 2 and 0.01 M KCl. The mass distribution coefficients (K D ) for P estimated using 0.01 M KCl were similar to values determined using fertilizer mixture for Candler fine sand. The K D values determined using 0.005 M CaCl 2 as the supporting electrolyte were two-to threefold greater than the K D of the fertilizer mixture on Candler fine sand suggesting that divalent Ca might result in overestimation of P sorption on Candler sandy soils. On Immokalee fine sand, fertilizer mixture and 0.005 M CaCl 2 were comparable and resulted in greater P adsorption than 0.01 M KCl. Thus, it would be appropriate to use 0.01 M KCl as supporting electrolyte for Florida's Candler fine sand and 0.005 M CaCl 2 on Immokalee fine sand.