Modeling the Transport of Inorganic Arsenic Species through Field Soils: Irrigation and Soil Structure Effect

Abstract

Dissolved arsenic (As) may appear at the tile line level through preferential flow (PF), leading to contamination of shallow water bodies. Limited work on the movement of As forms in field soils urged the need for more research. The PF of arsenate (As(V)) and arsenite (As(III)) compared to chloride (Cl) at constant flow under saturated (10 mm), slightly unsaturated (−10 mm), and unsaturated (−40 mm) pressure heads was evaluated in replicated large field columns varying in subsoil structure. A solute containing As(V), As(III), and Cl was pulsed until the Cl concentration ratio in the drainage samples reached maxima and flushed with solute-free irrigation. HYDRUS-1D software version 4.15 was utilized to fit the breakthroughs of As(V) and As(III) in the dual-porosity physical non-equilibrium model (DP-PNE). The Langmuir equation was used to fit the As(V) and As(III) sorption isotherms, and blue dye staining was used for the marking of flow paths. Dye leaching was observed up to 50 cm or deeper in the soils. Under saturated conditions (+10 mm), Kotli, Guliana, and Mansehra soils showed chemical non-equilibrium (CNE) for As(V) and As(III); however, the extent of CNE was less under unsaturated conditions (−40 mm). These results implied that these well-structured soils had enough large macropores, which cause PF, but at the same time, they were also small enough to retain water and leach solutes under unsaturated conditions (−40 mm). It is concluded that irrigation of contaminated water or dumping solid waste on well-aggregated soil may exhibit PF of dissolved As during and after rains, and additionally As(III), which is more toxic and mobile under reduced conditions, has equal or greater potential for movement.

1. Introduction

Several natural processes and anthropogenic activities introduce a high level of As into the environment, which may leach into the subsurface water rather rapidly. A better understanding of As transport and retention in soils enhance the risk assessment of aquifers. In drinking water, the permissible limit for As is 10 µg L⁻¹, while in soil the permissible limit for As is 20 mg kg⁻¹ [1,2]. Since the oxidation state of As influences both its bioavailability and toxicity [3,4], it is critical to assess the risk of specific ionic forms leaching through preferential flow (PF) pathways. Most of the studies carried out used disturbed and repacked soil, and information on the movement of inorganic As species in field soils (undisturbed large columns) with variable physical non-equilibrium (soil structure and irrigation levels) is lacking, suggesting a need for more research.

Heavy metal and metalloid retention is high in the soils, except where PF is dominant. It has been found that, under saturated flow, the mobility of cadmium (Cd), copper (Cu), and lead (Pb) in field soil columns is comparable to that of chloride (Cl) [5,6,7]. In a recent study, Cd, nickel (Ni), and zinc (Zn) transport in undisturbed loamy soil columns was investigated, and it was observed that the analytical CXTFIT model performed slightly better than the numerical HYDRUS model under saturated conditions [8]. Research revealed that compared to arsenate (As(V)), arsenite (As(III)) is more mobile, less strongly sorbed, and leaches at a higher rate [9,10,11].

Non-uniform conditions affect the transport of inorganic As species in a heterogeneous soil environment, and asymmetrical breakthrough curves (BTCs) of As(V) are caused by both chemical non-equilibrium (CNE) and physical non-equilibrium (PNE) [12]. The potential for rate-limited transport relates to the non-linear As(V) adsorption isotherm [13]. It was observed that, due to a rate-limited mass transfer, As transport through sandy soil followed a non-linear pattern [14]. As(V) transport is influenced by sorption-related non-equilibrium [15], whereas As(III) transport in goethite-coated sand columns is influenced by pore water velocity under saturated conditions [16]. Despite the significance of PNE, As(V) and As(III) BTCs were studied only under CNE conditions [17,18]. Moreover, the available literature suggests that studies on As, As(V), and As(III) reactive transport and retention modeling were based on repacked soils [19,20,21], and we believe that this study is the first attempt to understand the As ionic species transport behavior in natural structurally variable soils under different moisture regimes.

Physical non-equilibrium induces the PF of solutes [22,23], which results in the rapid movement of various solutes through the soil [5,24]. Irrigation level governs PF, which occurs predominantly between matric potentials of 0 and −10 mm pressure head [25]. Numerous factors, including soil structure [24], water content [26], tillage [27], texture [28], and precipitation intensity and time, affect the extent of PNE during the transport and movement of solutes [29].

Solute transport under PNE conditions can be explained by two regions: the dual-porosity (DP) or mobile–immobile model (MIM) [30] and the convective dispersive flow equation (CDE), where water is partitioned into mobile and immobile regions and, in the immobile region, exchange between water and solutes occurs [31]. For modeling of solute transport, advection–dispersion equations (ADE) can be solved numerically, e.g., in HYDRUS-1D [32], which uses measured variables of saturated flow to perform Marquardt–Levenberg-type techniques for inverse estimation of solute transport and reaction parameters. The dye tracing technique monitors water and contaminant flow [33], where visible and non-hazardous dyes are used for the marking of flow paths. The objectives of this study were (i) to determine the BTCs of As(V) and As(III) in undisturbed large soil columns differing in subsoil structure at different irrigation levels and (ii) to predict the transport parameters for the inorganic As species in comparison with Cl.

2. Materials and Methods

2.1. Site Description, Sampling and Characterization

Two alluvial soils, i.e., Rawal and Kotli series, and two loess soils, i.e., Mansehra and Guliana series, which differed in subsoil structure, were selected (Figure 1). For laboratory leaching experiments, four large intact soil columns were manually dug from each soil. The excavated soil columns were shaped to loosely fit 50 cm long and 25 cm diameter PVC pipes. Before transportation to the laboratory, polyurethane expanding foam was injected to hold the soil column within PVC pipes and stabilized overnight. The loose soil samples were collected in replication from each horizon included in the columns and processed for batch sorption studies and physio-chemical analysis. The US soil classification, sampling location, brief description of the four soils, and physio-chemical parameters have been presented in Tables S1 and S2 of the supplementary data.

The dye application technique was used for flow path marking in each soil [33]. The soil surface was rained with 2 cm of brilliant blue dye (2%) to apply the dye one time, avoiding surface runoff. The dye was left to infiltrate overnight, the profile’s face was exposed in increments of 5 cm toward the dye application’s center, and each incremental face was photographed.

2.2. Leaching Experiments

Leaching of As(V), As(III), and Cl was performed for each soil column according to the procedure adopted by Akhtar et al. [24] and Rukh et al. [26] at constant flow rates produced by maintaining +10 mm, −10 mm, and −40 mm water heads. A complete experimentation setup for saturated and unsaturated water heads and a graphical representation of drainage and irrigation assembly are given in Figure S1 of the Supplementary Material.

The columns were switched to the solute pulse containing 0.1 mg L⁻¹ As(V) and As(III) and 3.5 mM LiCl once the flow rate was stable and continued there until the Cl maxima were reached. Once the Cl concentration was at the required level, the input solution was switched back to solute-free irrigation, i.e., distilled water. The sequence of leaching was: (i) leaching at +10 mm water head; (ii) leaching at −10 mm water head; and finally, (iii) leaching at −40 mm water head. The percolate sampling interval was adjusted to represent ≈0.1 pore volumes. The drained sample was passed through a 0.45 µm-sized cellular membrane, and 1 mL of 1 M H₃PO₄ was added and kept at 4 °C till analysis.

2.3. Batch Experiment for As(V) and As(III) Sorption

As(V) and As(III) sorption isotherms were generated, and data were fitted in the Langmuir equation for determining partitioning coefficients [34,35]. In triplicate, 3 g of soil was taken in a 50 mL centrifuge tube, and 30 mL of 0.01 M KNO₃ solution containing 0, 0.1, 2.5, 5, 8, 10, 15, 20, 25, 40, and 100 mg L⁻¹ As(V) concentrations from Na₂HAsO₄ and, separately, 0, 0.1, 0.5, 1, 3, 5, 7, and 8 mg L⁻¹ As(III) concentrations from NaAsO₂. The soil and solution suspension were shaken for 48 h at room temperature (25 °C), centrifuged at 5000 rpm, filtered through a cellulose membrane with a 0.45 μm pore size, and analyzed for total As. The amount of As(V) and As(III) adsorbed was the difference between the applied and recovered amounts.

2.4. Analysis

As(V) and As(III) species were determined in two steps. In the first step, total As was measured in the drainage samples employing an atomic absorption spectrophotometer AA-6300, Shimadzu, coupled with a hydride generation assembly HVG-1, Shimadzu [36]. For total As, a premix of 0.4% NaBH₄ and 0.5% NaOH solutions was mixed with 5 M HCl in the mixing chamber [37]. Using a w/v premix of 0.35 percent NaBH₄ and 0.30 percent NaOH combined with w/v solutions of 1.15 percent oxalic acid (H₂C₂O₄) and 0.60 percent sodium oxalate (Na₂C₂O₄), As(III) in the leachate sample was quantified in the second phase [38]. The difference between total As and As(III) was used to calculate As(V) [39]. Using PerkinElmer ICP-MS (ELAN 9000), the As(V) and As(III) in the batch experiment’s equilibrating solution were quantified. A Cl-specific ion electrode using a Cole Parmer (Model 6071N) benchtop meter was used for the determination of Cl in drainage samples [40].

2.5. Parameter Estimation

The Langmuir equation was used to fit the As(V) and As(III) sorption isotherm data [41]:

$$X=\frac{bK{C}_w}{1+K{C}_w}$$

(1)

where C_w is the equilibrium concentration in solution (mg L⁻¹) and X is the sorbed amount of As(V) and As(III) (mg kg⁻¹ of soil). To obtain a linear relation, Equation (1) was rearranged as follows:

$$\frac{C_w}{X}=\frac{1}{kb}+\frac{C_w}{b}$$

(2)

A linear regression equation with a slope of 1/b and an intercept of 1/kb can be obtained by plotting C_w/X against C_w. The parameter b, which denotes maximum surface coverage, was calculated by taking the reciprocal of the slope. The parameter k, which denotes binding strength, was computed by substituting the value of b into the intercept.

The Kd, which denotes the adsorption partition coefficient, was calculated by identifying the tangent to the Langmuir adsorption isotherm at the applied concentration of 0.1 mg L⁻¹. The final equation was obtained by differentiating Equation (1) for C_w.

$$Kd=\frac{1}{1+k{C}_w}\left[bk-\frac{b{k}^2{C}_w}{1+k{C}_w}\right]$$

(3)

For modelling As(V), As(III), and Cl transport, version 4.15 of the HYDRUS-1D software was used [32]. First, the Cl BTCs data was fit to the DP (MIM) model with PNE [42]:

$$\frac{\partial {\theta }_m{c}_m}{\partial t}=\frac{\partial }{\partial z}\left({\partial }_m{D}_m\frac{\partial c_m}{\partial z}\right)-\frac{\partial q_m{c}_m}{\partial z}$$

(4)

$$\frac{\partial {\theta }_{im}{c}_{im}}{\partial t}=\alpha \left(c_m-c_{im}\right)$$

(5)

where D is dispersion coefficient (cm h⁻¹), α is mass transfer coefficient, c_m and c_im are concentrations of solutes in the MIM regions (mg cm⁻³), respectively, θ_m and θ_im denote MIM water content (cm³ cm⁻³), q_m is Darcy flux of the mobile phase (cm h⁻¹), z is depth (cm) and t is time (h). The estimated parameters for the Cl BTCs data were dispersivity (λ), θ_im, and α.

Equation (4) with PNE and CNE (DP with two-site sorption in the mobile zone), takes the form [42]:

$$\frac{\partial {\theta }_m{c}_m}{\partial t}+f_m\rho \frac{\partial {s^e}_m}{\partial t}=\frac{\partial }{\partial z}\left({\theta }_m{D}_m\frac{\partial c_m}{\partial z}\right)-\frac{\partial q_m{c}_m}{\partial z}-{\Gamma }_{s1}-{\Gamma }_{s2}$$

(6)

$$\frac{\partial {\theta }_{im}{c}_{im}}{\partial t}+\left(1-f_m\right)\rho \frac{\partial s_{im}}{\partial t}={\Gamma }_{s1}$$

(7)

$$f_m\rho \frac{\partial {s^k}_m}{\partial t}={\Gamma }_{s2}$$

(8)

$${\Gamma }_{s1}=\alpha \left(c_m-c_{im}\right)$$

(9)

$${\Gamma }_{s2}=\omega {\rho }_b\left({s^k}_{m,e}-{s^e}_m\right)$$

(10)

$${s^e}_m=\frac{f_{em}{K}_{d}c}{1+\eta c}$$

(11)

$${s^k}_{m,e}=\frac{\left(1-f_{em}\right)K_{d}c}{1+\eta c}$$

(12)

where Γ_s2 represents the mass transfer to kinetic sorption sites in the mobile region, and Γ_s1 signifies the mass transfer term for solute exchange between MIM regions; the sorbed concentration of the kinetic sites in contact with the mobile region is denoted by s^k_m,e; ω is the mass transfer coefficient (h⁻¹) between MIM phases; and s^e_m is the sorbed concentration in equilibrium with the liquid-phase concentration for the mobile region. The proportion of sorption sites in contact with the mobile water is denoted by f_m, while the fraction of sorption sites in the mobile region with instantaneous sorption is represented by f_em. Kd is the Langmuir sorption coefficient (cm³ mg⁻¹), and ρ_b is the bulk density (mg cm⁻³); λ, θ_im, and α are from fitted Cl BTCs. For As(V) and As(III) transport modeling, f_em, f_m, and ω were fitted parameters.

2.6. Statistical Analysis

The soil and water heads as the main effects and their interaction were used to analyze As(V) and As(III) transport parameters. Then soil-related variability was examined by taking the transport parameter data for heads as a multivariate. SAS/STAT software version 9.4 [43] was used for the analysis of the data after the transformation through Box–Cox transformation [44]. Tukey’s HSD test was used to compare the soil means at p ≤ 0.05.

3. Results

3.1. Flow Path Marking

Two to three cm of blue dye applied at the surface appeared between 35 and 90 cm of profile depth in all soils, but the flow pattern differed with the soil structure (Figure 2). In the Rawal soil, where the dye flowed along the dead root courses, biopores, and interaggregate surfaces, a higher concentration of blue dye was seen surrounding the macropores (>75 µm) and root paths, with dye movement reaching up to 90 cm. Due to firm peds, blue dye diffused a little radially and traveled through the fractures along the ped surfaces in the Kotli soil. In the Gulian soil, surface cracks seemed to be filled with the dye (where it diffused), to act as a funnel in the Bw horizon, and to appear well below 40 cm depth due to macropore flow in the Bt1 horizon, with a little radial diffusion. In the Mansehra soil, which had a 20 cm deep Ap horizon, the dye moved along the surface cracks, through a large number of macropores in the Bw horizon, and then through a few macropores and inter-pedal voids in the Bt horizon to a depth of 40 cm.

3.2. Non-Adsorbing Solute (Cl) Breakthrough Curves

The Cl BTCs were investigated in the previous studies [26] and are characteristic of non-equilibrium transport in both saturated and unsaturated flow regimes (Figure 2 in Rukh et al. [26]). The early and rapid arrival of Cl before one pore volume in the leachate indicated PF paths in all the soils. In the case of Kotli and Mansehra soils, the model-fitted parameters and Cl BTCs indicated PF due to PNE under saturated flow conditions that persisted but reduced under unsaturated flow (Table 5 in Rukh et al. [26]). The non-parametric indices (i.e., holdback factor, piston flow to solute velocity, and drainage to first 5%) showed that, under saturation in the Rawal and Guliana soils, strong PF of Cl occurred while, through Kotli and Mansehra soils, PF of Cl occurred to a lesser extent, especially under unsaturated conditions (Figure 3 in Rukh et al. [26]). In comparison to other soils, Rawal soil demonstrated greater non-equilibrium conditions at all saturation levels, while under unsaturated conditions, Guliana soil had more PNE conditions than Kotli and Mansehra soil. These observations were based on Cl transport parameters and BTCs.

3.3. As(V) and As(III) Breakthrough

The breakthrough data of As(V) and As(III) are well-fitted by the DP, two-site sorption model (Figure 3 and Figure 4). The BTCs of As(V) and As(III) at saturated flow (+10 mm water head) in all the soils were immediate and nearly as fast as Cl, even though the average peak concentration ratio reduced to 0.5 (as low, exceptionally, as 0.2 and as high as 0.7). During the injection phase, the rise in concentration was fast, and during the flushing phase, a rapid decline in the concertation ratio with a slight tailing was observed in most soil columns. The breakthrough of As(V) and As(III) at a slightly undersaturated flow (−10 mm water head) during the injection phase showed delay, a slow rise in concentration ratio and a reduction in the peak concentration, whereas, during the flushing phase, extensive tailing was observed. The Rawal and Guliana soils at all water heads showed extensive tailing, whereas the Mansehra and Kotli soils had an insignificant tail, both for As(V) and As(III). Overall, both As(V) and As(III) peak concentrations were lower than the peak Cl concentration under all water heads, showing a sorption effect. Breakthrough of As(V) and As(III) at unsaturated flow (−40 mm water head) occurred in the Rawal soil columns only, while leaching was effectively decimated in all other soils.

The fitted Langmuir isotherm (Equation (2)) sorption parameters, viz., b, k, and Kd at application rate for each horizon of all the soils, along with retardation coefficient (R) at saturated, slightly unsaturated, and saturated water, are presented in

Table 1. As(V) and As(III) sorption parameters obtained from Langmuir sorption model for each soil.

Horizon	Depth	As(V)						As(III)
		b †	k †	Kd ‡	Retardation for Three Heads §			b	k	Kd	Retardation for Three Heads §
	cm	mg kg−1	L mg−1	L kg−1	+10 mm	−10 mm	−40 mm	mg kg−1	L mg−1	L kg−1	+10 mm	−10 mm	−40 mm
Rawal soil series; silty, mixed, hyperthermic Typic Hapludalfs
A	0–10	307 (19)	0.07 (0.01)	21 (0.4)	58 (7)	75 (7)	85 (1)	22 (2)	9 (5)	50 (3)	132 (16)	174 (14)	197 (10)
Bw	10–18	248 (9)	0.14 (0.01)	34 (0.3)	108 (9)	133 (17)	145 (26)	36 (1.3)	6 (1.5)	86 (5)	271 (25)	337 (48)	370 (77)
Bt	18–30	275 (9)	0.16 (0.01)	42 (3)	138 (6)	171 (20)	193 (28)	43 (1)	14 (3)	104 (6)	341 (39)	422 (51)	476 (60)
Bk	30–46	131 (10)	0.57 (0.05)	67 (1.2)	248 (118)	319 (197)	350 (224)	37 (2)	22 (9)	79 (15)	264 (52)	331 (70)	364 (80)
Kotli soil series; fine, mixed, hyperthermic, Entic Haplusterts
Ap	0–10	153 (1.4)	0.43 (0.00)	61 (0.1)	255 (125)	310 (132)	502 (220)	42 (1.5)	10 (3)	104 (5)	290 (42)	357 (47)	580 (142)
Bw	10–18	227 (5)	0.20 (0.01)	44 (3.4)	144 (26)	163 (34)	205 (84)	42 (0.3)	14 (2)	101 (2)	357 (4.3)	404 (26)	500 (143)
C	18+	234 (35)	0.24 (0.05)	52 (3.5)	209 (26)	227 (47)	286 (52)	54 (1.0)	50 (20)	78 (21)	304 (88)	332 (124)	417 (146)
Guliana soil series; silty, mixed, hyperthermic Udic Haplustalfs
Ap	0–10	235 (27)	0.18 (0.04)	41 (5.1)	94 (11)	103 (16)	111 (17)	39 (0.75)	10 (0.75)	96 (11)	257 (16)	280 (32)	301 (38)
Bw	10–20	256 (26)	0.18 (0.03)	44 (11)	113 (21)	118 (12)	125 (11)	39 (1.3)	12 (1.3)	96 (7)	264 (91)	273 (74)	288 (75)
Bt1	20–30	248 (28)	0.32 (0.12)	71 (17)	233 (22)	242 (44)	260 (47)	61 (4)	38 (4.4)	100 (6)	336 (54)	344 (27)	370 (40)
Bt2	30–50	231 (13)	0.45 (0.05)	95 (4)	315 (83)	316 (31)	345 (32)	62 (0.5)	53 (0.51)	83 (1.3)	273 (27)	300 (34)	290 (42)
Mansehra soil series; silty, mixed, thermic Typic Hapludalfs
Ap	0–20	262 (27)	0.17 (0.04)	43 (5)	108 (25)	148 (34)	179 (75)	40 (0.5)	7 (0.35)	97 (0.9)	243 (41)	330 (48)	397 (132)
Bw	20–40	204 (11)	0.34 (0.09)	65 (12)	287 (157)	318 (124)	407 (115)	50 (0.4)	13 (1.7)	120 (2)	411 (19)	478 (63)	640 (193)
Bt	40–70	290 (21)	0.37 (0.04)	100 (5)	459 (240)	475 (252)	545 (258)	70 (0.6)	25 (4)	142 (10)	485 (37)	501 (30)	583 (42)

Notes: ^† b and k are Langmuir parameters (Equation (2)), which represent maximum adsorption capacity and binding strength, respectively. ^‡ Kd distribution coefficient was calculated at application concentration, i.e., 0.1 mg L⁻¹ using b and k in Equation (3), ^§ Retardation factor: R = 1 + [(ρ_bK_d)/θ].

. In general, the Langmuir isotherm fitted the data well with an r² of 0.90 or better (Figure 5). In soils, eight to ten times greater b values of As(V) compared to As(III) were observed, while As(III) had greater k values than As(V) (Table 1). The Langmuir parameters (b and k) both for As(III) and As(V) varied among horizons or soils nested within the soil type. The three soils, Rawal, Guliana, and Mansehra, are also statistically similar for b of As(V), and Kotli differs in having lower mean values of b. The k for As(V) was similar in Guliana and Mansehra soils, while Rawal soil showed significant (p < 0.0001) lower mean values of k for As(V). However, all the soils differed significantly for b of As(III), and Mansehra soil showed the highest mean values of b for As(III). The Guliana and Kotli soils varied statistically (p < 0.0001) for k of As(III) from the Mansehra and Rawal soils. The k for As(V) had a strong positive relationship with DOC, and the k for As(III) with the clay content (r = 0.60), CBD extractable Fe (r = 0.50), Al (r = 0.80), and (NH₄)₂C₂O₄ extractable Fe (r = 0.55) and Al (r = 0.78). The mean adsorption Kd varied significantly among the horizons or soils nested within the soil type at the application concentration of 0.1 mg L⁻¹. Generally, Kd mean values for As(III) are nearly twofold greater than As(V) and increased with the increase in depth in all soils except for Kotli soil. However, R significantly (p < 0.001) increased as a result of the change in the head from 10 to −40 mm, and the head effect also differed due to the soil (i.e., significant head × soil interaction at p < 0.001), as Rawal and Guliana soils statistically differed from Kotli and Mansehra soils, both for As(V) and As(III).

It was also noticed that the Bt/Bk horizons had a high adsorption capacity, especially for As(III) at the lower end of the column, which leads to high R, except for Rawal soil, which allowed As(III) to leach through all the columns even under unsaturated conditions.

The As(V) and As(III) transport parameters f_m, f_em, and ω were fitted using the corresponding Cl BTC parameters (λ_m, α, and θ_im) as input from Rukh et al. [26] and the Langmuir exponents (b and k) along Kd calculated from each leaching.

Based on the findings of the Cl transport parameters and non-parametric indices, under both saturated and unsaturated flow, PNE conditions were evident, but in the Rawal soil they were much higher at all saturation levels and, under saturated conditions in the Guliana soil, stronger PNE conditions were present than in the Mansehra and Kotli soil columns [26]. The DP model with two-site sorption in the mobile zone also showed that Rawal and Guliana soil columns had greater PNE and CNE conditions compared with the Mansehra and Kotli soil columns. The DP model fits the data with R² = 0.95 or greater, where adjusted R² is the ratio of the sum of squares of residuals to the sum of squares subtracted from 1 (Figure 3 and Figure 4).

The f_m varied from 0.03 at +10 mm water head to 0.37 at −10 mm water head for As(V) and from 0.02 at +10 mm to 0.15 at −10 mm for As(III) which corroborated the Cl data that PNE conditions were present. However, f_m increased significantly in the soils at −10 mm water head for As(V) to a mean value of 0.23, 0.37, 0.17, and 0.23 for Guliana, Kotli, Mansehra, and Rawal soils, respectively. Similarly, f_m for As(III) at −10 mm water head was significantly higher compared to +10 mm in all the soils. According to Šimůnek and van Genuchten [42], smaller f_m values indicate greater kinetic sorption compared to lesser instantaneous sorption, which eventually results in an earlier solute arrival. Kinetic sorption is slower sorption, which results in higher concentration and earlier arrival in the drainage from MIM regions [42]. Overall, lower values of f_m were observed at +10 mm water head, which explains the early arrivals of As species under saturated conditions and, among As species, As(III) showed lower values of f_m than As(V) in all the soils. The values of f_em decreased with water heads in most soils. This decrease in instantaneous sorption f_em indicates reduced CNE with an increase in pressure head, which leads to a delay in solute arrival. This phenomenon occurred in all the soils, except in Mansehra soil for As(V) and Kotli soil for As(III), and Rawal soil at −40 allowed breakthrough for both As(V) and As(III).

The ω values ranged from 0.0025 to 0.10 h⁻¹ for As(V) and 0.012 to 0.43 h⁻¹. The lower values of ω under unsaturated conditions were due to a significant soil type (p < 0.001) factor. High ω values at +10 mm in Guliana and Rawal soils, both for As(V) and As(III), indicated excessive transfer between soil and mobile region under limited kinetic sorption. Although the ω values decreased in all the soils at −10 mm and −40 mm for both As(V) and As(III), As(III) remained higher than As(V) in all the soils. Consistent with earlier research, the non-equilibrium adsorption coefficient showed a strong correlation with the mobile water velocity, Vm (r = 0.72 (As(III) and r = 0.51 As(V)) [45].

The relative mass leached (M/M_o) is one of the clearest indicators of the PF [26]. In this study, M/M_o of As(V) and As(III) leached provide evidence of the PF presence in all the soils. Under saturated conditions, both As(V) and As(III) leached in high amounts (Figure 6 and Figure 7). Overall, the mass leached of As(V) and As(III) in Rawal soil under saturated conditions was higher as soil × head interaction was significant (p < 0.05), followed by Guliana soil, which had higher mass leached for As(V) compared to Kotli and Mansehra soils. It was also clearly noticed that, among As species, As(III) mass leached was significantly higher as compared to As(V) and, in the case of Rawal soil, it reached up to 60% under saturated conditions.

4. Discussion

The patterns of blue dye movement relate to the occurrence of macropores and the grade and consistencies of soil aggregates. Dye marks the interaggregate surfaces, biopores, and dead root paths [46,47]. Very firm peds formed by the compression of aggregates under shrink–swell processes in the Kotli (Usterts) soil lead blue dye through interaggregate surfaces with a little radial diffusion. The surface cracks in the Guliana soil served as a funnel to the Bw horizon and, in the Bt1 horizon, macropores carried the dye with a little radial diffusion due to the clay skins [48]. In the Mansehra soil again, the surface cracks in the deep Ap horizon led blue dye to move through a large number of macropores in the Bw horizon, and in the Bt horizon flow occurred through a few macropores and inter-pedal voids. The Rawal soil site occurred under trees and a permanent rose plantation, unplowed for several years, and it had a large number of dead root paths and biopores distributed throughout the profile where the leaching dye concentrated. Although the brilliant blue dye has excellent visibility and low adsorption in soils [33], the adsorption differs with organic matter content and clay [49]. The use of dye to mark the PF paths in field soils is common [50,51].

The breakthrough of As(V) and As(III) was fast at saturated flow, though with a reduced peak concentration ratio. The breakthrough at −10 mm water head was delayed, the rise in concentration ratio was slow, peak concentration was reduced, and an extensive tailing occurred. The breakthrough at −40 mm water head occurred only in Rawal soil columns. The columns from Mansehra and Kotli soils at −10 mm water head had a lower peak concentration ratio, most likely because they have more nominal pores (≈375 μm), which resulted in increased contact with the matrix and ultimately large adsorption; at −10 mm water head, 1500 μm-size pores were excluded from the flow process [26]. At a water head of −40 mm, matrix retention of As(V) was evident from its minimal levels, except for the Rawal series. The massive silty clay loam soils effectively retain As species, and moderate medium subangular blocky silt loam and silty clay loam soils under saturated (+10 mm) and slightly unsaturated (−10 mm) flow conditions allowed breakthrough.

The occurrence of variability among the columns of the same soil remained the most striking feature of saturated flow. The adsorptive nature is recognized by the reduction in peak concentration ratio [17], and the presence of PF paths is predicted by the immediate breakthrough. Under unsaturated flow conditions, delay or no breakthrough for As(V) was attributed to the dominance of matrix flow as a result of reduced water content and flow rate. On the other hand, under saturated (+10 mm water head) and slightly unsaturated (−10 mm water head) flow rates, As(V) was effectively retained by massive loamy sand soil, and moderate medium subangular blocky silty clay loam soils allowed breakthrough [24]. The leaching experiments that used uniformly packed disturbed soil columns reported strong As(V) retardation indicated by slow release and extensive breakthrough tailings; however, As(V) breakthrough and an increase in effluent recovery had been observed with an increase in pore water velocity [17]. In disturbed soil columns, As(III) undergoes strong retardation with diffusive effluent fronts, followed by tailing during leaching or slow release [18]. Overall, the peak concentration of As(III) was higher than that of As(V) in all the soils. Despite the compact nature of aggregates in Kotli soil, greater clay contents probably caused retardation of As(III). The lower peak concentration of As(III) in the Guliana soil is indicative of retardation under both saturated (+10 mm water head) and unsaturated (−40 mm water head) flow. The Mansehra soil had an immediate and fast As(III) breakthrough under saturated (+10 mm water head) flow conditions. The decline in the concentration ratio suggested the presence of non-porous aggregate, while As(III) retardation under unsaturated (−40 mm water head) flow conditions was ascribed to a higher clay content. The b values for As(V), were eight to ten times higher than those of As(III) in the soils, while the k values for As(III) were greater than those of As(V) (Table 1), supporting the findings of Sondal et al. [52]. Various studies show that clay content plays a significant role in the adsorption of As(V) and As(III) [53,54]. The values for sorption parameters vary widely with soil properties [54,55]. The maximum coverage for As(V) had a strong positive relationship with DOC and for As(III) with the clay content (r = 0.60), citrate bicarbonate extractable Fe (r = 0.50), Al (r = 0.80), and ammonium oxalate extractable Fe (r = 0.55), and Al (r = 0.78). Several studies revealed that the metal oxides enriched material had higher adsorption maxima [53,56]. The sorption of As(V) on goethite [57], hematite [58], and soils [59,60,61] is reduced by dissolved organic carbon.

The range of Kd, (20 to 100 L kg⁻¹ for As(V) and 30 to 100 L kg⁻¹ for As(III)) is close to those reported by Bonis et al. [62]. An averaged Kd value was used for modeling the solute transport, although pore water velocity controls the contact time between sorbing surfaces and solutes, as opposed to almost 48 h in the batch experiment. The Bt horizons, the lower end of the column, had high As(III) adsorption capacity except for Bk in the Rawal soil, where the same soil allowed As(III) to leach through all the columns even under unsaturated (−40 mm water head) conditions. The lower Kd value of As(III) in the Bk horizon can be explained by the fact that As(III) sorbs less on calcite than As(V) [63]. The validity of the use of Kd from batch sorption experiments to explain breakthroughs in field columns needs further investigation [64].

The transport of As species is affected by PNE and CNE, as indicated by the low values of f_m, f_em, and ω, low M/M_o for As(V) and As(III) when fitted in the DP two-site sorption model (Figure 5 and Figure 6). The values of f_m less than one are indicative of PNE and CNE for both As(V) and As(III), which increased with the flow at unsaturated conditions when macropore flow had reduced and solute solid interaction increased. By comparing the BTCs, it was noticed that CNE diminishes under unsaturated flow as peak concentrations and M/M_o were greatly reduced at negative water heads, especially at −40 mm. The effect of water heads was clear on the As species transport parameters and as f_m increased, f_em and ω decreased in most soils. However, M/M_o for As species clearly showed that CNE persisted in Rawal soil even at −40 mm water head and to a lesser extent at −10 mm water head in all soils. The size of the aggregates as a structural unit in Rawal and Guliana soils is medium blocky compared to coarse blocky in Mansehra and Kotli soils, reducing the channels for gravity flow under unsaturated conditions. The role of aggregate clay skin, which restricts the rate of solute exchange between the aggregate matrix and mobile flow path [48], would greatly increase retention through adsorption once the active macropores are emptied. Clay content increased the adsorption of both As forms. In nearly all soils, an increase in saturation resulted in an increase in ω for As(V) and As(III), suggesting slower sorption kinetics at lower velocities and unsaturated flow conditions. Multiple studies have established a linear relationship between pore water velocity and ω [45,65,66]. Solute transport for PNE and CNE was simulated by Avila and Breiter [67] using f_m = 0.88 and f_em = 0.5. Additionally, Köhne et al. [68] found that, for pesticide transport studies, fitting f_m = 0.258 using an inverse DP two-site sorption model yielded the best match for breakthrough. Similarly, fitted f_em = 0.2 was noted by Ladu and Zhang [69] when simulating atrazine transport with a DP two-site sorption model.

The results of this study show that, in undisturbed field soil columns, the sources of the PF of As species are aggregate in nature, especially macropores, which under saturated conditions cause bypass flow. However, under unsaturated conditions, macropore flow diminishes, leaving behind the sorption-related non-equilibrium (f_em > f_m).

5. Conclusions

In this study, four soils varying in subsurface structure, derived from loess and alluvial parent material, indicated the PF of As(V) and As(III) at and near saturation in large field soil columns. The As(V) and As(III) breakthroughs under saturated and unsaturated conditions for all soils were fitted well by a dual porosity model with chemical non-equilibrium using HYDRUS-1D software (version 4.15). As(V) and As(III) breakthrough showed that the concentration ratio and mass leached were reduced under negative pressure heads and only Rawal soil allows the leaching of As(V) and As(III) at −40 mm pressure head. The PF associated with Rawal and Guliana soils was ascribed to a weak coarse sub-angular blocky structure, although an increase in desaturation caused a delayed breakthrough. The HYDRUS-1D successfully fit the data, indicating that the exchange rate coefficients varied significantly with soil and flow rate variation. As(V) and As(III) leaching in all soils suggested that only a fraction of the sorption sites were in equilibrium with the liquid phase in the mobile region. Moreover, the more toxic As form As(III) showed higher mobility and mass leached quantities in all the soils. Overall, under saturated and, to a lesser extent, under unsaturated conditions, these well-structured soils result in more PF. This study highlights the need to prevent untreated wastewater irrigation and the accumulation of solid waste disposal on soil surfaces that are strongly aggregated in the subsoil, particularly fine-textured soils. In the case of rainfall and irrigation, such soils may exhibit PF. The result of this study urges not only the need for wastewater treatment and its safe disposal, but also the leachate management of solid waste dumping sites through the installation of impermeable liners and collection systems. To improve management, these results may also aid in the prediction of the movement of other adsorbing and non-adsorbing contaminants, as well as fertilizer nutrients, through these soils. In further studies related to aquifer risk assessment, contaminant transport should consider the PNE along with the CNE, instead of the CNE alone. Groundwater management and protection plans can benefit from the findings of this study.