Preferential flow of phosphorus and nitrogen under steady‐state saturated conditions

Repeated broiler litter application on agricultural lands can cause nutrient enrichment of subsurface effluent, especially with the existence of preferential flow through soil macropores. Previous studies quantifying soil macropores have not attempted to establish a connection of soil macropore characteristics with the subsurface nutrient (nitrogen [N] and phosphorus [P]) losses, across different topographical locations in the field. This study investigated the effect of broiler litter application and preferential flow on subsurface nutrient transport (N and P) at different topographical positions (upslope, midslope, and downslope) in a no‐till pasture field located in Alabama, USA. Twelve intact soil columns (150 mm id and 500 mm length) were used, and the nutrient leaching measurements from laboratory experiments were linked to soil macropore characteristics quantified using X‐ray computed tomography image analysis and solute transport modeling. Treatments included surface broadcast broiler litter (5 Mg ha−1, on dry basis) and unamended control. Leachates were analyzed for dissolved reactive P (DRP), total P (TP), and nitrate + nitrite‐N (NO3− + NO2−–N). The bromide breakthrough curves provided evidence of preferential flow in all columns. Litter application significantly increased leachate P concentrations, and average TP and DRP concentrations were significantly higher in the leachate from upslope columns compared to those at downslope location. The NO3−–N concentrations in leachate exceeded the US EPA drinking water standard of 10 mg L−1 in all the treatment columns. The highest flow‐weighted mean concentrations of TP and DRP, at 2.7 and 2.5 mg L−1, respectively, were recorded in the upslope columns. Soil physicochemical properties and nutrient leaching losses varied substantially across topographical positions, indicating a need for variable litter application rates to reduce P build‐up and subsequent leaching in vulnerable locations within the field. The relevance of the effect of topographic position on nutrient leaching found in this study should be further tested by investigating a wider range of slopes and soil types in pastures.

Vadose Zone Journal upslope columns compared to those at downslope location.The NO 3 − -N concentrations in leachate exceeded the US EPA drinking water standard of 10 mg L −1 in all the treatment columns.The highest flow-weighted mean concentrations of TP and DRP, at 2.7 and 2.5 mg L −1 , respectively, were recorded in the upslope columns.Soil physicochemical properties and nutrient leaching losses varied substantially across topographical positions, indicating a need for variable litter application rates to reduce P build-up and subsequent leaching in vulnerable locations within the field.
The relevance of the effect of topographic position on nutrient leaching found in this study should be further tested by investigating a wider range of slopes and soil types in pastures.

INTRODUCTION
The Southeastern United States is the largest broilerproducing region, generating approximately 66% of the nation's broilers annually (USDA-NASS, 2021).The three leading broiler-producing states in the Southeastern United States are Georgia (15%), Alabama (13%), and Arkansas (12%).Broiler litter is a mixture of poultry manure and a bedding material such as pine shavings, peanut hulls, or rice hulls.Alabama generates about 1.8 million Mg of broiler litter each year.The broiler litter is typically used to fertilize agricultural areas within short distances of the production facilities to minimize transportation costs (Lin et al., 2018;Watts et al., 2011).Nitrogen (N) and phosphorus (P) are the two major nutrients in broiler litter (He et al., 2009).Repeated application of broiler litter to the same agricultural fields results in the accumulation of these nutrients in soils, increasing the susceptibility for off-site transport of nutrients to surface waters and groundwater (Bohara et al., 2019;He et al., 2009;Sharpley et al., 1993;Vervoort et al., 1998;Wood et al., 1996).
If present in excessive amounts, N and P can accelerate the eutrophication of surface waters, and nitrate (NO 3 − -N) in drinking water, which may be harmful to humans and animals (King et al., 2015).Therefore, a thorough understanding of P and N transport in broiler litter-amended agricultural systems is crucial.This knowledge informs optimal litter application rates and effective land management strategies for water quality preservation, pivotal for sustainable agriculture (Nguyen et al., 2013).
In agricultural landscapes, P transport can occur through surface or subsurface pathways.Extensive research efforts have focused on P transport via surface runoff and developing management strategies to reduce soil erosion (King et al., 2015;Kleinman et al., 2004;Verheyen et al., 2015).Increases in iron (Fe) or aluminium (Al) hydroxides with depth increase the subsoil's affinity to bind P (Lamba et al., 2013;Makris et al., 2006).However, there is growing evidence of significant P transport through subsurface flows over the last two to three decades (Grenon et al., 2021;Hively et al., 2006;Kianpoor Kalkhajeh et al., 2021;King et al., 2015;Sharpley et al., 2000;Turner & Haygarth, 2000).Subsurface transport of P can be significant, especially in areas with deep sandy soils, subsurface zones with preferential flow paths, soils with high organic matter, or with high soil P levels from long-term manure or fertilizer application (Kang et al., 2011;Simard et al., 2000;Sims et al., 1998;Yang et al., 2019).Kang et al. (2011) found that the load of dissolved reactive P (DRP) in the leachate was correlated to the water-extractable P (WEP) concentration in manure applied to soil columns.A study by Kleinman et al. (2005) found a significant increase in the TP and DRP concentrations in the leachate through eight intact soil columns after the application of poultry manure compared to the preapplication concentrations.The authors also reported that before the application of manure, DRP was only about 7% of the TP in the leachate.However, following the application, DRP was the dominant form of TP (∼72%) in the leachate, indicating that the majority of the rise in leachate P concentration following the manure application was possibly derived from the water-soluble P in the manure.
Extensive experimental data and groundwater monitoring suggest that preferential flow through soil macropores could be a principal mechanism facilitating the transport of these nutrients, as the macropore flow essentially bypasses the soil matrix and decreases the soil-nutrient contact times (Heathwaite & Dils, 2000;King et al., 2015;Li & Ghodrati, 1994;Makowski et al., 2020).No-till systems contribute to increased earthworm populations and dense root networks, which can lead to the formation of continuous and wellconnected pore systems.Additionally, the fact that no-till practices reduce runoff and lead to an increased amount of water available for percolation suggests that no-till practices may facilitate the movement of surface-applied agricultural chemicals to groundwater (Shipitalo et al., 2000;Shipitalo & Edwards, 1993;Zhang et al., 2018).Soils with macropore networks created by worm burrows and root channels have been observed to exhibit early breakthroughs, where solutes rapidly Vadose Zone Journal bypass the soil matrix and cause greater nutrient loss compared to soils with disrupted macropore channels (Domínguez et al., 2004;King et al., 2015;Singh & Kanwar, 1991).However, understanding the impact of preferential flow pathways on nutrient transport can be challenging due to the heterogeneity of macropores and the variability of flow paths over space and time.
In certain regions or under specific local conditions, subsurface transport of nutrients can be substantial.For instance, in a study conducted in the Sand Mountain region of North Alabama, USA, researchers found that more than 90% of rainfall infiltrated into the soil, while less than 10% contributed to surface runoff (Lamba et al., 2013).This suggests that significant subsurface flow may occur in this region.Furthermore, soils in this area are shallow to bedrock, sloping, and permeable to water and dissolved nutrients (He et al., 2009), thus, providing a high potential for nutrients to contaminate ground and surface water.Sen et al. (2008) studied the area's hydrology and also found a potential for significant subsurface flow in this region.In the study conducted by Lamba et al. (2013), the authors found that following the surface application of poultry litter at 5 Mg ha −1 (dry basis), the average loads of DRP and TP in the leachate were 878 and 983 g ha −1 , respectively.However, the authors reported lower average loadings of DRP and TP in the surface runoff, measured to be 305 and 563 g ha −1 , respectively.In another study conducted on a pasture field in this region, dense macropore networks were observed, which can play an important role in the subsurface transport of nutrients (Budhathoki et al., 2022b).Further, examining nutrient transport in the Sand Mountain region is important, as it is a major broiler-producing area in the state.For example, counties such as Cullman, Blount, DeKalb, and Marshall collectively produce nearly 43% of the state's broilers.Hence, it is vital to understand the fundamental processes that control the fate and transport of nutrients, especially the impact of preferential flow paths on nutrient transport in this region.This knowledge would facilitate the development of appropriate best management practices to mitigate nutrient losses from such agricultural landscapes.
Despite substantial measurements conducted using laboratory and field techniques to understand the impact of litter application on nutrient leaching, little is known about how the topographical positions (e.g., effect of slope position) across agricultural landscapes affect nutrient leaching dynamics (Julich et al., 2017;Ortega-Pieck et al., 2020;Raulerson et al., 2023;Sigua et al., 2011).To our knowledge, no study has been conducted in no-till pasture systems to evaluate the impact of litter application and preferential flow pathways on nutrient leaching at different topographical positions.Despite preferential flow promoting faster and greater subsurface nutrient losses, studies quantifying soil macropores using X-ray computed tomography (CT) image analysis have not attempted to establish a connection of soil macropore char-

Core Ideas
• Phosphorus and nitrogen concentrations in the leachate varied as a function of slope position.• Breakthrough curves and X-ray computed tomography (CT) analysis provided evidence of preferential flow.• Solute transport was well-fitted by the two-region physical nonequilibrium model.• Broiler litter application significantly increased phosphorus leaching under saturated conditions.
acteristics with the subsurface nutrient (N and P) losses.This approach can be implemented in regions across the country and globally where there is intensive animal production, areas that rely heavily upon animal manure application to provide nutrients for crops, and areas with high soil P levels.Furthermore, its adaptability can be extended to regions where subsurface flows are a significant mechanism, such as the karst landscapes or fields with tile drainage systems.Therefore, the primary objective of this research was to determine the effect of broiler litter application and preferential flow pathways on the leaching of P and N at three topographic positions (upslope, midslope, and downslope) in a pasture field using intact undisturbed soil columns.The secondary objective was to link the nutrient transport to macropore characteristics deduced from X-ray CT.The rationale behind these objectives was based on the previous investigation at the study site, which demonstrated that macropore characteristics varied at different topographical positions in the field (Budhathoki et al., 2022b).Quantification of nutrient transport in no-till pasture soils would foster implementing precision litter management to minimize the environmental impact of local animal production activities.Based on prior research in the Sand Mountain region of North Alabama, it was hypothesized that the nutrient leaching dynamics within the field after the application of broiler litter would vary as a function of topographical position and soil macropore characteristics.

Study site
The study site is a no-till pasture field located in the Sand Mountain region of North Alabama (34˚17′02.6″N,85˚57′51.8″W),a major broiler production area in the state (Lamba et al., 2019).This region is part of the Cumberland Plateau section of the Appalachian Plateau physiographic province (Sen et al., 2008) (Figure 1).The site (area 0. ).The primary soils in the study site are Hartsells fine sandy loam (fine-loamy, siliceous, thermic, Typic Hapludult) and Wynnville fine sandy loam (fine-loamy, siliceous, thermic, Glossic Fragiudult) (He et al., 2009;Sen et al., 2008).The Wynnville series consists of moderately well-drained soils that are slowly permeable and formed from weathered sandstone and shale (National Cooperative Soil Survey, n.d.).The Hartsells series consists of soils that are moderately deep, well-drained, and moderately permeable.These soils were formed from parent material that is loamy residuum weathered from acid sandstone containing thin strata of shale or siltstone.This field has not received any application of broiler litter for at least 13 years prior to the beginning of the study.A summary of baseline soil properties at the study site is reported in Table 1.
Based on a previous research conducted in this region on pastures fertilized with poultry litter (similar to our study site), labile inorganic P and labile organic P levels were approximately 66 and 60 mg kg −1 , respectively (Ranatunga et al., 2013).
T A B L E 1 Baseline soil properties for 0-25 cm before treatment applications at the study site.

Sample collection
In our study, we used 12 undisturbed intact soil cores (150 mm id × 500-mm depth), which were collected from the field using polyvinyl chloride (PVC) pipes.Four cores were collected from each topographical location (upslope, midslope, and downslope) (Figure 1).The cores were collected in May 2019 using a hydraulic cylinder device mounted on the front of a tractor (Prior et al., 2004).The system pushes and pulls the PVC core pipes with a hydraulic device.Additionally, six disturbed soil samples were collected adjacent to the location of undisturbed soil cores at depth intervals of 0-5, 5-10, 10-15, 15-20, 20-30, 30-40, and 40-50 cm.Therefore, a total of 42 disturbed soil samples were collected at each topographical location and subsequently analyzed for different chemical properties such as electrical conductivity (EC), pH, Vadose Zone Journal WEP, total P (TP), and nitrate + nitrite-N (NO 3 − + NO 2 − -N).Additionally, three soil samples were collected from each topographical location at different depth intervals (0-10, 10-20, 20-30, 30-40, and 40-50 cm) to determine bulk density, organic matter content, and texture.

Soil column preparation
After returning to the laboratory, each of the soil cores collected was oriented with its length vertically and was saturated from the bottom up with warm water, and each was allowed to remain saturated for 48 h for earthworm inactivation or removal, after which they were drained for about 3 days to ensure consistent soil moisture levels across all columns (Zopp et al., 2016).The columns were stored at 4˚C until the rainfall simulation experiments were started.A day before conducting the rainfall simulations, liquified petroleum jelly was injected in the gap between the soil and the internal walls of the PVC pipe and allowed to dry.The liquified jelly was injected to minimize preferential flow along the column walls under saturated flow conditions.Acid-washed sand was used to fill the vertical gap (3 cm) between the soil and the base cap.Three tensiometers (Soil Moisture Equipment) and three time domain reflectometry (TDR) probes (Campbell Scientific) were inserted on the opposite sides of each column at depths of 12.5, 25, and 37.5 cm to monitor matric potential and volumetric water content, respectively.The tensiometers and TDR probes were connected to a CR-1000X data logger (Campbell Scientific) to monitor the readings on a minute basis.

Rainfall simulation
The rainfall simulation experiments and laboratory analysis were conducted at the Soil and Water Laboratory and Biological Engineering Research Lab (BERL) at Auburn University.The rainfall simulation system consisted of the soil column, carboys, a peristaltic pump, a sealed PVC base cap, a leachate collection system, and a rainfall simulator cap consisting of hypodermic needles (Figure 2).The needles were pierced into a coiled rigid plastic tubing that was segmented into five pieces (Zopp et al., 2016).Uneven distribution of solution across the coil due to the pressure drop associated with each needle was prevented by supplying solution to each end of the segmented coil.A multichannel peristaltic pump (Masterflex L/S; Cole Parmer) was used to supply the solution to the rainfall simulator cap, and the cap was calibrated before the start of the simulation.Treatments for the rainfall simulation experiments were broadcast surface-applied broiler litter at a rate of 5 Mg ha −1 (dry basis) (typical litter application rate in Alabama for pastures) and control (0 Mg ha −1 ).Six columns were subjected to broiler litter treatment, while the remaining six were treated as control columns.Before starting the rainfall simulation experiment, the soil columns were subjected to capillary saturation from the bottom with 0.01 M KCl until 3-5 cm ponding was achieved on the top of the column.Then the soil columns were irrigated with a 0.01 M KCl solution from the top at a constant rate sufficient to maintain a head of 2-3 cm on the soil surface.The head was maintained to ensure saturated flow conditions.After the steady-state flow had been achieved at the bottom of the column, bromide (Br − ) as potassium bromide (KBr, 0.01 M) was applied as a conservative tracer for investigating breakthrough curves (BTCs) for preferential flow (Jinhua et al., 2014;Zopp et al., 2016).Solid broiler litter was applied to the treatment columns immediately prior to initiating the rainfall simulation to simulate the worst case scenario of rainfall following the manure application event.It should be noted that the time of storms relative to the time of manure application can significantly impact solute transport (Edwards et al., 1993;Shipitalo et al., 1990).Following that step, the KCl solution was applied again to wash out the bromide from the column.The effluent was collected in acid-washed plastic bottles placed underneath the soil columns through plastic tubing.The pore volumes collected were calculated as follows: where V is the volume of effluent collected (mL), V c is the bulk volume of the soil column (cm 3 ), ρ b is the bulk density (g cm −3 ), and ρ s is the particle density (2.65 Mg m −3 ) of the soil.Dimensionless or relative concentration (C/C 0 ) was calculated by dividing leachate Br − concentration by Br − concentration in the rainwater.The BTCs were plotted for each column representing the relative concentration of Br − in the effluent versus pore volume.During the rainfall simulations, three to six pore volumes of leachate were collected in 450-mL acid-washed bottles, depending on when C/C 0 for Br − reached a negligible level.The concentration of Br − ions in the leachate was measured using a bromide combination ionselective electrode (model HI4102, Hannah Instruments Inc.).Furthermore, the leachate samples collected were analyzed for EC, pH, and nutrients (DRP, TP, and NO 3 − + NO 2 − -N).

Solute transport modeling
The precise and accurate estimation of solute transport parameters is important to comprehend how solutes move through soil, enabling informed decision-making in agriculture and environmental practices.The one-dimensional steady-state transport of a nonreactive solute in homogenous soils is described by the equation of the convection-dispersion (CDE) model as follows (Bear, 1972): where C is the solute concentration, D is the hydrodynamic dispersion coefficient (cm 2 min −1 ), v is the pore water velocity (cm/min), and x and t are the distance and time, respectively (Shahmohammadi-Kalalagh et al., 2022).
For the soils with the presence of macropore flow, Equation (2) is not well-suited to describe solute transport.Therefore, a two-region mobile-immobile (MIM) nonequilibrium model could be used to characterize the solute transport in heterogeneous soils (van Genuchten & Wierenga, 1976;Zhang et al., 2015).The two-region model assumes that the liquid phase is divided into a mobile and an immobile region, and the solute could exchange between those two regions (Nsengumuremyi et al., 2021).Therefore, a conservative non-sorbing solute is transported by a convective-dispersive process in the mobile domain and enters and leaves the immobile domain as described by a first-order process (Casey et al., 1997;van Genuchten & Wierenga, 1976): where the subscripts m and im refer to mobile and immobile liquid regions, θ is volumetric water content (cm 3 cm −3 ), C m and C im are the concentrations in mobile and immobile regions, respectively, D m is the hydrodynamic dispersion coefficient of the mobile phase (cm 2 min −1 ), q is the Darcy flux (cm min −1 ), and α is the first-order mass transfer coefficient for solute exchange between the mobile and immobile regions (min −1 ) (Nsengumuremyi et al., 2021;Zhao et al., 2017).The Br − BTCs of all the columns were modeled using the CXTFIT 2.0 computer program in inverse mode using both the CDE and MIM models to estimate the solute transport parameters (Toride, 1995).For the CDE model, parameter D was fitted, and v (cm min −1 ) was calculated using q and porosity θ.For the MIM model, two parameters (D and ω) were fitted by holding parameters, v and β, fixed (Zhang et al., 2015).The dimensionless partitioning coefficient β represents the fraction of mobile water (θ m /θ) provided the solute is nonreactive (Zhang et al., 2015).When β = 1, the two-region model reduces to one-region model.The parameter β was estimated as the number of pore volumes required to reach the relative Br concentration of 0.5 (Singh & Kanwar, 1991).Then, the mass transfer coefficient α was derived from the dimensionless mass transfer coefficient ω as ω = αL/θv, where L is the length of the column (Herbert, 2011;Vervoort et al., 1999).The dispersivity (λ, cm) was calculated from the dispersion coefficient (D) and the pore water velocity (v) using λ = D/v.The performance of the model was evaluated by means of two measures, namely, the coefficient of determination (R 2 ) and the mean square error (MSE).

2.5
Physical and chemical analysis of soil, broiler litter, and leachate samples

Broiler litter and soil sample analysis
Broiler litter samples were collected from a poultry Farm in North Alabama.Before the litter application, the samples were stored at 4˚C and analyzed for pH, moisture content, DRP, TP, TN, NO 3 − -N, and NH 4 -N.The soil texture (percentage sand, silt, and clay) was determined using the hydrometer method (Bouyoucos, 1962;Carter & Gregorich, 2008).Organic matter content was determined by the weight loss on ignition method (Huisman et al., 2013).The bulk density of the soil was determined by dividing the mass of the oven-dried soil sample by the volume of the sample collected.
To prepare soil samples for the chemical analysis, the samples were dried at 60˚C for 48 h and sieved through a 2-mm sieve.Soil pH and EC were measured at a soil-to-water ratio of 1:10 using VWR Symphony electrodes (VWR Scientific Inc.) (Carter & Gregorich, 2008).
The WEP contents of soil samples were determined with a 1:10 soil/water ratio extract using deionized water and following the protocol of Kovar and Pierzynski (2009).Briefly, the soil water suspension was shaken on a mechanical shaker for 1 h.Subsequently, each sample was centrifuged, and the aliquots were filtered through a 0.45-μm membrane.After this step, the analysis of WEP was conducted using the molybdenum blue colorimetric method (Murphy & Riley, 1962) using a Seal Analytical AQ2 automated discrete analyzer (Seal Analytical Inc.).TP was measured on unfiltered samples using inductively coupled plasma-optical emission spectrometry (ICP-OES), following EPA Method 3051A (USEPA, 2007), after microwave-assisted digestion using a mixture of nitric and hydrochloric acid (1:25 soil/solution).For the analysis of NO 3 − + NO 2 − -N, 5 g of soil was placed into a 50-mL centrifuge tube and extracted with 25 mL of 2 M KCl for 1 h on a mechanical shaker.After each sample was centrifuged, the aliquots were filtered through a 0.45μm membrane.Subsequently, the samples were analyzed for NO 3 − + NO 2 − -N (US EPA Method 353.2) by cadmium reduction using the Seal Analytical AQ2 automated discrete analyzer (O'Dell, 1993).

Leachate samples
Filtered (0.45 μm) leachate samples were analyzed for DRP and NO 3 − + NO 2 − -N, and unfiltered samples were analyzed for TP.Analysis for DRP, NO 3 − + NO 2 − -N, and TP was conducted using the methods mentioned above.
Although the results for nitrate and nitrite analysis for soil and leachate samples were reported as the concentration of NO 3 − + NO 2 − -N in mg kg −1 and NO 3 − + NO 2 − -N in mg L −1 , respectively, the concentrations will be referred to as nitrate-N (NO 3 − -N) for the remainder of this article, as an insignificant amount of nitrite was present, compared to nitrate (Hoover et al., 2019;Richards et al., 2021).

Degree of phosphorus saturation
Soil samples were analyzed for oxalate-extractable Al (Al ox ), Fe (Fe ox ), and P (P ox ) (Gasparatos et al., 2006;McDowell & Sharpley, 2001).Five grams of soil was shaken with 25 mL of 0.2 M ammonium oxalate (pH 3) for 4 h, then centrifuged for 10 min at 1000 g and filtered through a 0.45-μm filter.The leachate was then analyzed using ICP-OES.The results were reported in mg kg −1 on a dry weight basis and converted to mmol kg −1 .From these data, the P sorption capacities (PSC), PSC =0.5(Al ox + Fe ox ) in mmol kg −1 , were calculated (Leinweber et al., 1999).Furthermore, the degree of saturation of the soil's P sorption capacity, that is, the degree of P saturation (DPS), which is a good indicator of a soil's potential to release P, was calculated as follows (Kovar & Pierzynski, 2009;Siemens et al., 2004): where [P ox ], [Feox], and [Al ox ] denote the concentrations of the elements in the oxalate extracts in mmol kg −1 .The dominator of Equation ( 6) approximates the TP sorption capacity of the soil (van der Zee & van Riemsdijk, 1986).

Image acquisition and processing
It is important to characterize the soil pore properties to improve our understanding of nutrient transport dynamics through soils.Therefore, in this study, we attempted to utilize X-ray CT-derived soil characteristics to explain the nutrient transport through the soil.The soil columns were imaged using a medical scanner located at the Bailey Small Animal Teaching Hospital within the Auburn University College of Veterinary Medicine.The details of the image acquisition and processing are provided in Budhathoki et al., 2022aBudhathoki et al., , 2022b)).Briefly, the soil columns were placed horizontally on the bench while scanning, and the machine would take up to 64 slices in one scan.The scanning produced images with a slice thickness of 0.625 mm, thus covering 40 mm in one scan.The images were produced with 512 × 512 pixels per slice with a field of view of 180 × 180 mm, thus producing voxels of 0.35 × 0.35 × 0.625 mm in the reconstructed image.The images were 16-bit (DICOM format).The images were processed and analyzed using an ImageJ software program version 1.52t (Rueden et al., 2017) to determine the pore circularity, macroporosity, interconnectivity, tortuosity, and connection probability.Based on the size, macropores have been defined variously in past studies, for example, as pores greater than 1 mm (Luxmoore, 1981) or pores larger than 3 mm (Germann & Beven, 1981).In this study, pores ≥0.75 mm were considered as macropores based on the resolution of the scanner.The macroporosity was calculated for each slice and then averaged to quantify the variation with soil depth.For the quantification of interconnectivity, node density was calculated by adding up the number of nodes (where at least two pore branches connect) and then dividing that by the volume of soil considered (Luo et al., 2010).For this calculation, the BoneJ plug-in in the ImageJ software was used (Katuwal et al., 2015;Yang et al., 2018).The actual macropore length was divided by the straight-line distance to get the tortuosity of macropores (Katuwal et al., 2015).The connection probability is defined as the likelihood that two randomly selected pore voxels in the region of interest belong to the same cluster, indicating that they are connected (Jarvis et al., 2017).It is zero if the pore space is very fragmented (Schlüter et al., 2020).Pore circularity is an important characteristic of pore morphology, and it reduces as the irregularity of the pore perimeter increases (Li et al., 2016).High pore circularity facilitates water and solute transport in soils (Yang et al., 2018).The circularity of macropores was calculated using Equation 7as follows (Soto-Gómez et al., 2018): where A is the pore area (mm 2 ) and P is the pore perimeter length (mm).

Statistical analysis
The statistical analyses were performed using SAS version 9.4 (SAS Institute Inc.).The PROC GLIMMIX procedure was used to assess the effects of treatment and slope position on the leachate concentrations of all nutrients.The unique id of the soil column was considered as a random effect to capture the temporal dependency of the leachate nutrient data in all columns, and slope and treatment were used as fixed effects.The nutrient leachate data were subjected to log transformation.The least-square means for each combination of treatment and slope position were obtained, and Tukey-Kramer's adjustment method was used to perform multiple comparisons with least-square means at a significance level of p ≤ 0.05.

Physical and chemical analysis of broiler litter and soil samples
Nutrient analysis and physical characteristics of the broiler litter are presented in Table 2.The TP concentration of this poultry litter (15.7 g kg −1 ) is slightly low in comparison with other poultry litter samples used in other studies in this region (19.3-26.07g kg −1 ) (He et al., 2008;Lamba et al., 2012;Sistani et al., 2010).
The soils at different topographical locations in the field differed in various properties that would be expected to impact the nutrient leaching losses (Table 3).The surface soils (0-10 cm) at the downslope location were finer and more compacted (sand content and bulk density as 53.9% and 1.31 g cm −3 , respectively) as compared to upslope (sand content and bulk density as 58% and 1.1 g cm −3 , respectively) and midslope (sand content and bulk density as 68.8% and 1.17 g cm −3 , respectively) locations.Compared to the midslope and upslope locations, where the organic matter in the surface soil was 3.93% and 4.75%, respectively, the downslope location had a higher organic matter of 6%.
EC was higher in the top 10 cm compared to the bottom layers of the soil profile at all topographical positions in the field (Figure 3).The average EC was 163 μS cm −1 , which was below the threshold value of 4000 μS cm −1 found to be detrimental for many crops (Kingery et al., 1994).The mean pH of the soil profile (0-50 cm) in the study site was 5.7.The slightly acidic pH of the soil was almost certainly due to the nature of the sandstone parent material.The pH in the subsoil layers was lower compared to the top 15 cm (Figure 3).
Depth distributions of background soil P (TP and WEP) and NO 3 −--N levels are illustrated in Figure 4.The average WEP content of surface soil in the study area was 3 mg kg −1 , with the highest concentration observed at the upslope location (5 mg kg −1 ).The average WEP content is similar to that observed by Lamba et al. (2013) (4.6 mg kg −1 ) in the study conducted in the Sand Mountain region.Soils at all locations exhibited evidence of P stratification: TP at 40-50 cm was lower than at the soil surface (p < 0.05).Phosphorus stratification was also reported in another study conducted in the Sand Mountain region of Alabama, where TP contents of 156, 109, and 90 mg kg −1 were observed at 0-to 20-, 20-to 40-, and 40to 60-cm depths, respectively (He et al., 2009).Researchers have found that soils with a history of receiving manure for a long time before the initiation of the study had greater background soil P levels.For example, Whittington et al. (2010) measured TP concentrations as high as 1000 mg kg −1 at the surface soil of their study site at the SMREC, which could be attributed to poultry manure application (6 Mg ha −1 ) for 10 years before the start of the study.He et al. (2009) reported a TP concentration of 847 mg kg −1 at 0-20 cm after 15 years of poultry litter application.
The lower TP content in soils at the downslope location as compared to soils from the upslope (Figure 4) could be attributed to the lower total iron and total aluminum content in soils at the downslope (Figure 5).However, it should be noted that additional factors such as the soil organic matter and soil texture can also have significant effects on the distribution of TP in soil.
The overall mean concentration of NO 3 − -N in this study area was 29.8 mg kg −1 for a soil depth of 0-50 cm.(2009) found the concentrations of NO 3 − -N to be 6.2 and 13.5 mg kg −1 for topsoil receiving 5 and 20 years of poultry litter application, respectively.In a study conducted in Oklahoma, Sharpley et al. (1993) reported the concentration of NO 3 − -N in untreated surface soil (0-5 cm) ranging from 0.6 to 34.5 mg kg −1 and from 10.7 to 110. 8 mg kg −1 in the soil treated with poultry litter.The high concentrations of NO 3 − -N on the surface soil could be attributed to high temperatures and relatively low rainfall around the sampling time (May 2019), which would reduce leaching losses.The onset of warmer weather stimulates the mineralization of organic N and increases the rate of biological processes.Many authors have also reported gradual increases in inorganic N in the soil during the dry summer season (Hardy, 1946;Hu et al., 2015;Korsaeth et al., 2002;Schofield, 1945).Harmsen and Kolenbrander (1965) also reported higher levels of inorganic N in the soil during this time (Spring and early Summer) in eastern US states.Furthermore, being a cool-season grass, the yield had declined around the sampling time, reducing the amount of NO 3 − -N removal via crop uptake (Harmsen & Kolenbrander, 1965).

Bromide breakthrough curve parameters
The Br − breakthrough curves were fitted using both singledomain CDE and two-region MIM models.The singledomain CDE model could not fit the breakthrough curves well (data provided in the Supporting Information), and the two-region nonequilibrium model was able to fit the Br − BTCs well for all the columns.Fitted parameters derived from the two-region model are presented in Table 4.The performance of the model for all columns showed R 2 values ranging from 0.97 to 0.99 and MSE <0.0029.Figure 6 shows the observed and fitted Br − breakthrough curves using the two-region MIM for all 12 columns.The Br − recoveries were more than 93% for all columns.These curves were evaluated for possible evidence of preferential flow.A soil with macropore flow can rapidly transport a solute resulting in a relative concentration (C/C 0 ) of 0.5 much earlier than one pore volume (Garrett et al., 2004;Jinhua et al., 2014).All columns showed an early breakthrough for Br − , that is, Br − in the effluent reached a C/C 0 of 0.5 well before one pore volume was drained, indicating rapid flow through macropores (Figure 6) (Singh & Kanwar, 1991).Bouma and Wösten (1979) attributed the rapid breakthrough of a conservative tracer in undisturbed soil columns to the presence of large, continuous pores occupying only a small volume.Therefore, the rapid transport of Br − in all the columns in this study could be associated with preferential flow through the soil macropores.Similar observations have been reported elsewhere (Luo et al., 2010;Zhang et al., 2015).
The initial breakthrough of Br − occurred at an average of 0.35 pore volumes in upslope soil columns compared to an average of 0.48 and 0.52 pore volumes for midslope and downslope columns, respectively, indicating a higher macropore flow in upslope columns.A wide range of immobile water content indicates that immobile regions played an important role in the Br − transport and that the physical non-equilibrium cannot be ignored (Wang et al., 2019).Compared to the downslope columns (λ: 8.5 ± 3.8, α: 0.0013 ± 0.001, ω: 0.2 ± 0.06), those collected from the upslope area had a higher average λ (17.9±8) and lower average values of α (0.0006 ± 0.0003) and ω (0.11 ± 0.07).The greater average λ and lower exchange coefficient values are associated with an increasing T A B L E 4 Solute transport parameters determined from bromide breakthrough curves (v and β) and estimated by CXTFIT (D and ω) using the two-region model.The parameter λ was calculated from D/v.The parameter α was calculated using the dimensionless mass transfer coefficient ω since ω = αL/θv.

Slope position Column
Flux (cm min −1 ) Average θ (cm 3 cm −3 ) degree of preferential flow (Herbert, 2011;Vervoort et al., 1999).For example, Luo et al. (2010) used λ values in addition to the solute breakthrough to assess the extent of preferential flow in different soil columns and concluded that higher λ values are associated with greater preferential flow.Higher dispersivity in soil promotes enhanced mixing of the solute, facilitating the rapid spread of the solute front in the soil profile.A lower value of ω indicates a lower mass transfer between the mobile and immobile phases (Herbert, 2011).Further, the low values of β (high immobile pore-water fraction) for columns from upslope is indication of a higher degree of preferential flow (Comegna et al., 2001;Singh & Kanwar, 1991).Our findings are consistent with previous research indicating that columns with lower α values exhibit greater preferential flow (Comegna et al., 2001;Luo et al., 2010;Mahmood-Ul-Hassan et al., 2011).These observations underscore the importance of understanding the variability in solute transport parameters to assess the fate and transport of nutrients in soil.

X-ray CT image analysis
The detailed X-ray CT image analysis was performed as a part of a larger study (Budhathoki et al., 2022a(Budhathoki et al., , 2022b)).In this study, macropore characteristics were quantified for the 12 columns used.The X-ray CT image analysis yielded additional evidence for the presence of macropore networks in columns collected from all slope locations.Table 5 shows macropore characteristics for the control and treatment soil columns.For the treatment columns, these characteristics were determined before litter application to the columns.
Figure 7 shows the three-dimensional visualization of macro-pore networks in the selected columns used in this study.The soil macropore characteristics inferred from X-ray CT image analysis varied between cores collected from different topographical locations and also varied depending on the depth of the soil profile.In the columns collected from the upslope location, macroporosity was the highest for the surface layers.However, in columns collected from the downslope location, dense macropore networks were observed in the subsurface layers (>25-cm depth) compared to those collected from the upslope and midslope (Table 5).Furthermore, the interconnectivity was the lowest at the downslope position for the surface layers (0-25 cm) compared to midslope and upslope (p < 0.05) (Table 5).The circularity of macropores in columns at the upslope location was larger than at columns from the downslope location (p > 0.05).Furthermore, the circularity increased from surface to subsurface layers, which could facilitate the movement of water and nutrients into deeper soil layers.It was also observed that the upslope soil columns are less tortuous compared to those from downslope and midslope (p > 0.05).Increased tortuosity increases the residence time for the nutrients to interact with the soil matrix (Ronayne, 2013).

Phosphorus breakthrough curves
Figure 8 shows the breakthrough trends of all nutrients in the 12 columns.   of columns at all topographical locations (p < 0.05), identifying applied litter as a significant source of P in the leachate.
The highest flow-weighted mean TP concentration found in leachate samples from the control columns was 0.039 mg L −1 and ranged from 0.21 to 2.7 mg L −1 for the treatment columns.The average TP levels in all the treatment columns are in excess of the eutrophic criterion (0.1 mg L −1 as TP) established for streams or other flowing waters not discharging directly into lakes or impoundments (McDowell & Sharpley, 2001).The TP loss in leachate was about 17.2%, 1.35%, and 1% of the TP applied through broiler litter for the treatment columns from upslope, midslope and downslope positions, respectively.However, the corresponding DRP loss in leachate was much higher and about 90%, 5%, and 3% of the DRP applied through broiler litter for the upslope, midslope, and downslope columns, respectively.The percentage loss of DRP and TP was determined by subtracting the average mass of nutrients recovered in the leachate for control columns from the mass recovered for treatment columns, multiplying by 100, and then dividing by the mass of nutrients added with litter to the treatment column.The immediate transport of DRP in the leachate in response to litter application suggests the existence of preferential flow pathways that connect the soil surface (to which manure was applied) to the subsurface horizon (Kleinman et al., 2009).The TP levels increased significantly in the leachate collected right after manure application and then declined and stabilized.Such an immediate transport of P (first flush) in leachate to litter application is consistent with other studies (Kleinman et al., 2005;Zang et al., 2011).Without litter application, DRP/TP ratio (based on average concentrations) was very low.However, after manure application, DRP accounted for the majority of TP.For example, from upslope treatment columns, approximately 92% of TP loss in leachate was recorded in the dissolved form, that is, DRP.In comparison, treatment columns at midslope and downslope transported about 78% and 52% of TP as DRP, respectively.Therefore, the inorganic form of P appeared to be the dominant form of P loss in all the treatment columns.These results are similar to Lamba et al. (2013), who found DRP averaging 35% and 91% of leachate TP from untreated soil and soil with surface-applied poultry litter, respectively.It has been reported that the dissolved form of P is more susceptible to leaching in sandy or organic soils with a low P retention capacity (Schelde et al., 2006).
Average DRP and TP concentrations were significantly higher in the leachate from the upslope treatment columns compared to those at the downslope position (p < 0.05).There could be three possible explanations for these observations.First, the interconnectivity and macroporosity of the surface soil (0-25 cm) at the downslope location were significantly less (p < 0.05) as compared to the upslope position (Table 5).The higher macroporosity combined with a higher interconnectivity in the surface layers of the upslope columns would facilitate the rapid infiltration and transport of DRP from the litter applied on the soil surface to lower depths without substantial contact with the soil matrix.Second, the average dispersivity estimated from the MIM model for the upslope columns was higher compared to downslope columns.The greater dispersivity values correspond to increasing degree of preferential flow (Luo et al., 2010).Higher dispersivity in soil promotes enhanced mixing of the solute, facilitating the rapid spread of the solute front in the soil profile.Third, soil WEP and TP contents of surface soil were significantly higher in the upslope than downslope (p < 0.05).Additionally, the DPS in soils varied among slope positions and soil depths (Figure 9).The lowest DPS value was obtained for the surface soil from the downslope location.The surface soils had significantly higher DPS than soils collected between 30and 50-cm depth at all slope locations, indicating that the subsurface soils still possessed sufficient capacity to sorb P.
F I G U R E 9 Degree of phosphorus saturation (DPS) calculated using oxalate extractable Al, Fe, and P at different depths in soils collected from upslope, midslope, and downslope locations.
However, it has been found that even in soils with high subsoil P sorption capacities, elevated leaching losses may still occur due to the presence of preferential flow, which can reduce the buffering effect of the subsoil layers (Djodjic et al., 2004), similar to what we found in our study.It has been well established that P losses from agricultural soils through leaching or runoff increase as the DPS increases (as higher DPS results in P release from sorption sites with lower affinities for P) (Butler & Coale, 2005).Soil DPS values above 25% are indicative of a greater risk of P losses via leaching (Abdala et al., 2015;Butler & Coale, 2005;Casson et al., 2006;Pautler & Sims, 2000).
Therefore, a lower DPS coupled with a lower macroporosity and interconnectivity in surface layers could have possibly caused less P transport in columns collected from the downslope location.Ortega-Pieck et al. ( 2020) also reported that subsurface transport of DRP varied as a function of slope position and was strongly influenced by the variability of soil P content at different topographic positions within a landscape.They attributed the significantly higher DRP leaching in soil cores collected from toe slope areas (downslope areas) compared to the top slope to the differences in soil P content between the slope positions.

Nitrate breakthrough curves
In all treatment and control columns, NO 3 − -N breakthrough occurred shortly after the simulation started owing to the preferential flow (Figure 8).This was followed by a long tail, indicating that the concentration decrease was slow and gradual during the flushing phase.This extended tail sug-gests that the solute that had entered the matrix pores and seeped back to preferential flow paths during the flushing phase (Mahmood- Ul-Hassan et al., 2008, 2010).The NO 3 − -N peaks appeared slightly early or at the same time as the DRP and TP peaks in the treatment columns.Additionally, we observed that the litter application did not significantly increase the flow-weighted mean NO3 − -N concentrations in all columns.It could be possibly due to form in which N is present in organic manures.A considerable portion of N in the poultry litter was in organic form, which requires time to undergo conversion to NO 3 − -N, given the duration of the simulation experiments.However, the mean NO 3 − -N concentrations in most of the columns (Table 6) were greater than the 10 mg L −1 , the USEPA-recommended maximum (Montgomery et al., 2014).The average flow-weighted concentrations of NO 3 − -N in the control and treatment columns were 22 ± 4.2 and 23 ± 9.2 mg L −1 , respectively.The NO 3 − -N loss in leachate was more than 90% of the NO 3 − -N applied through broiler litter for all the columns.In a study conducted in a fescue pasture in Arkansas, the investigators reported the highest NO 3 − -N concentrations as 13 and 54 mg L −1 from suction-cup lysimeters at 60-cm depth in soil with poultry litter application at 10 and 20 Mg ha −1 , respectively (Adams et al., 1994).The concentrations remained above 10 and 30 mg L −1 for soil with poultry litter application at 10 and 20 Mg ha −1 , respectively, for most of the sampling period.
Average leachate NO 3 − -N concentrations from all control and treatment columns in this study are substantially higher than reported in Lamba et al. (2013), nearly 8.6 and 7.6 times for control and treatment soils, respectively.We also observed that the flow-weighted mean NO 3 − -N concentrations in leachates were influenced by the background soil NO 3 − -N levels.Unlike DRP, we did not observe elevated NO 3 − -N concentrations in the leachate from upslope columns compared to downslope columns.This may be attributed to the lower adsorption or retention potential of NO 3 − -N in soil.
The solubility and mobility of NO 3 -N is substantially higher (Dong et al., 2022) compared to P, which tends to adsorb more easily to soil.Therefore, we did not observe distinct trends for NO 3 − -N with respect to soil macropore properties.
Therefore, to better understand NO 3 − -N mechanisms, futures studies are needed to link physical characteristics with NO 3 − -N leaching.The highest NO 3 − leachate concentrations were observed for the midslope columns (Table 6), corresponding to the highest background soil NO 3 − -N level at this position (Figure 4).Additionally, the flow-weighted mean nitrate concentration in the treatment column at the midslope location was significantly higher (p < 0.05) than the downslope treatment column.Overall, the high leachate nutrient concentrations are not surprising because we used the worst case scenario in this study, that is, saturated soil and litter application before the onset of rainfall simulation accompanied by the presence of preferential flow pathways.

Vadose Zone Journal
Although the simulation of biogeochemical process in the column system is beyond the scope of this study, for future long-term studies, it is important to examine the influence of biogeochemical transformations on P and N transport in systems, which are dominated by preferential flow.

Implications of the study
Overall, the results of our study support our hypothesis that slope position could be an important factor impacting the variability of soil physical and chemical properties in the field and, hence, the nutrient leaching trends.Our results have shown that soils at the upslope location had the highest concentration of soil P and the highest degree of P saturation of surface soils, when compared with other slope locations in the field.Further, the flow-weighted mean concentrations of DRP and TP in the leachate from the upslope treatment columns were significantly higher compared to those from the midslope and downslope columns.
Litter addition significantly increased the leaching of DRP and TP, and the dissolved inorganic form was the dominant fraction of the TP in the leachate at all slope locations in the field.This suggests that with the litter addition, the leaching of dissolved inorganic P might be a more widespread environmental problem in sandy loam no-till pasture soils with the existence of macropores than the leaching of particulate form of P.
Future experiments that explore and evaluate other factors such as rate of litter application, duration of rainfall simulation experiments, type of manure (such as dairy or swine manure), and nutrient loss in other forms (such as colloidalfacilitated) are needed to understand the overall dynamics of nutrient transport.In the future, experiments could be conducted on field sites with varying history of litter applications such as short-term (<∼6 years), medium term (∼10-15 years), and long term (>∼25 years) to understand the influence of duration of litter application on nutrient transport dynamics.We would also encourage more efforts to further investigate and continuously test relationships between soil properties, flow and transport parameters, and nutrient leaching across a wide range of soil types, land uses, and management practices.It should be noted that this study was conducted on soils collected from a field with a 3.4% slope.Therefore, studies on a wider range of slope categories are required to comprehend better the effect of soil physical and chemical properties on nutrient transport.Further, fields with spatial variability in soil P and N levels should be targeted by precise variable rate application of fertilizers.The P index developed for evaluating risk of P loss should also include subsurface loss, especially in fields implementing no-till conservation practice and areas with dense network of macropores or where subsurface flow can be an important contributor to P loss.

CONCLUSIONS
This study investigated the effect of broiler litter application and soil macropores on the leaching of P and NO 3 − -N at different topographic positions in a tall fescue pasture system.Soil macropore structures were quantified using X-ray CT and the characterization of the existence of the preferential flow was performed by Br − BTCs.A two-region MIM model was applied and transport parameters were determined by inverse modeling using a curve fitting program, CXTFIT.The shape of Br − BTCs suggests the occurrence of preferential flow through macropores in these undisturbed soil columns from no-till pasture soils.The X-ray CT image analysis provided evidence of a dense network of soil macropores at all topographic locations in the field.Results of Br − and nutrient leaching show that soils at different topographical locations, with variable physical and chemical properties, resulted in considerable differences in the values of solute transport parameters, and these should be considered for nutrient management studies.
The addition of litter significantly increased the flowweighted mean concentrations of TP and DRP in the leachate from all columns.The analysis also indicates that 10%-92% of TP was associated with DRP in all columns, highlighting the importance of the dissolved form of P in the leachate from these no-till sandy loam soils.
The litter application did not significantly increase the flow-weighted mean concentrations of NO 3 − -N in the leachate, but the concentrations remained higher than the recommended limit for drinking water (10 mg L −1 ) in all the treatment columns.Overall, linking the X-ray CT image analysis and solute transport modeling with the nutrient leaching offered valuable insights into the subsurface nutrient transport dynamics.Since this study was conducted under saturated conditions, it is important to conduct future studies with unsaturated flow conditions to further investigate the importance of macropores and their role in water and nutrient transport under varying degrees of saturation.

F
Location of the study site at the Sand Mountain Research and Extension Center (SMREC) (DeKalb County) in North Alabama.
40 ha and 3.4% slope) was located at the Sand Mountain Research and Extension Center (SMREC), one of the field research and Extension units of the Alabama Agricultural Experiment Station.Kentucky 31 tall fescue (Festuca arundinacea Schreb.), a cool-season grass, along with bermudagrass [Cynodon dactylon (L.) Pers.] was grown on the study field.The area receives an average annual precipitation of 1474 mm and has a mean annual temperature of 15.6˚C (NCEI-NOAA, n.d.

F
I G U R E 2 Illustration of the rainfall simulation setup.TDR, time domain reflectometry.

F I G U R E 3
Soil profile distribution of average (a) pH and (b) electrical conductivity (EC) in the study area.Each half bar is 1 standard error.F I G U R E 4 Soil profile distribution of (a) water extractable phosphorus (WEP), (b) total P, and (c) nitrate + nitrite-N before the experiments.Each half bar is one standard error.F I G U R E 5 Soil profile distribution of (a) total Fe and (b) total Al before the experiments.Each half bar is one standard error.However, compared to soil NO 3 − -N concentrations in the current study, other investigators reported lower concentrations in the Sand Mountain region soils.For instance, the mean NO 3 − -N concentration (103.6 mg kg −1 ) at the soil surface (0-5 cm) in our study was far greater than the mean value of 13.22 mg kg −1 reported by Lamba et al. (2012).He et al.

7
Three dimensional visualization of the macropore networks (>0.75 mm) in the selected columns collected from (a), (a)* upslope; (b), (b)* midslope; and (c), (c)*downslope (* indicates treatment columns).For the treatment columns, macropore characteristics were determined before litter application to the columns.Breakthrough curves for total phosphorus (TP) and dissolved reactive phosphorus (DRP), and nitrate-N for all columns collected at upslope, midslope, and downslope.Note the different scales for the x-axis and y-axis.T A B L E 6 Flow-weighted mean concentrations of nutrients (total P [TP]), dissolved reactive phosphorus [DRP], and nitrate-N [NO 3 − -N]) in the leachate for all columns at different topographical locations (n = 2).
Composition of broiler litter applied to treatment soil columns for rainfall simulation (mean ± standard deviation).
T A B L E 2 Average organic matter, bulk density, and sand, silt, and clay of the soil at different depth increments.
Average macropore characteristics at different depths for the columns collected from upslope, midslope, and downslope locations determined using X-ray computed tomography (CT).
Table 6 lists the flow-weighted mean concentrations of nutrients in the leachates collected from all columns throughout the simulation.Litter application significantly increased the transport of TP and DRP in the leachate T A B L E 5 :Values shown in parentheses are standard errors.Within each column, within each topographic location, means followed by the same lowercase letters are not significantly different at α = 0.05.Within each column, within each treatment type, means followed by the same uppercase letters are not significantly different at α = 0.05.a Most of the leachate TP concentrations in the control columns were below the detection limit.Hence, the values presented in the table represent the average concentrations calculated based on the detected values. Note