Contents lists available at ScienceDirect

Environment International

journal homepage: www.elsevier.com/locate/envint

Assessment of the Sulfamethoxazole mobility in natural soils and of the riskof contamination of water resources at the catchment scale

D. Archundiaa,b,e,1, C. Duwiga,⁎, L. Spadinia, M.C. Morela,c, B. Pradof, M.P. Perezd, V. Orsagg,J.M.F. Martinsa

aUniv. Grenoble Alpes, IRD, CNRS, IGE, Grenoble, Franceb Consejo Nacional de Ciencia y Tecnologia (CONACYT), Mexico, D.F, Mexico.c CNAM, Laboratoire d'analyses chimiques et bioanalyses, Paris Cedex 3, FrancedUniversidad Mayor de San Andrés, Instituto de Hidrología e Hidráulica, La Paz, BoliviaeUniversidad Nacional Autónoma de México-Estación Regional del Noroeste, Mexicof Instituto de Geología, Universidad Nacional Autónoma de México, Coyoacán, Ciudad de México 04510, MexicogUniversidad Mayor de San Andrés, Facultad de Agronomía, La Paz, Bolivia

A R T I C L E I N F O

Handling Editor: Yong-Guan Zhu

Keywords:Antibiotic mobilityPollution riskReactive transportHYDRUS-1DRate-limited sorption

A B S T R A C T

Sulfamethoxazole (SMX) is one of the antibiotics most commonly detected in aquatic and terrestrial environ-ments and is still widely used, especially in low income countries. SMX is assumed to be highly mobile in soilsdue to its intrinsic molecular properties. Ten soils with contrasting properties and representative of the catch-ment soil types and land uses were collected throughout the watershed, which undergoes very rapid urbandevelopment. SMX displacement experiments were carried out in repacked columns of the 10 soils to exploreSMX reactive transfer (mobility and reactivity) in order to assess the contamination risk of water resources in thecontext of the Bolivian Altiplano. Relevant sorption processes were identified by modelling (HYDRUS-1D)considering different sorption concepts. SMX mobility was best simulated when considering irreversible sorptionas well as instantaneous and rate-limited reversible sorption, depending on the soil type. SMX mobility appearedlower in soils located upstream of the watershed (organic and acidic soils - Regosol) in relation with a higheradsorption capacity compared to the soils located downstream (lower organic carbon content - Cambisol). Bycombining soil column experiments and soil profiles description, this study suggests that SMX can be classified asa moderately to highly mobile compound in the studied watershed, depending principally on soil properties suchas pH and OC. Potential risks of surface and groundwater pollution by SMX were thus identified in the lower partof the studied catchment, threatening Lake Titicaca water quality.

1. Introduction

Antibiotics are considered as emergent contaminants, and soils canbe exposed to antibiotics in multiple field applications (use of sewagesludge or animal manure, irrigation with treated or untreated waste-waters and polluted surface waters, outflows of sewage drains and ofsewage treatment plants) (Mojica and Aga, 2011; Oppel et al., 2004).

Sulfamethoxazole (SMX) belongs to the sulfonamide antibiotics fa-mily. It is used worldwide since the 1960's for the treatment of animaland human diseases. Its excretion rate varies between 50% and 100%(Mojica and Aga, 2011) in humans and animals. It is commonly de-tected in waste and surface waters as well as in soils and manure (Hoaet al., 2011; Hu et al., 2010; Kim et al., 2011; Leung et al., 2012;

Michael et al., 2013; Zuccato et al., 2010).The mobility of sulfonamides in soils has been studied by a number

of authors in the last decade. Sulfonamides present generally high so-lubility in natural water. They present consequently low affinity forsolid surfaces and thus low solid-liquid distribution coefficients (Kd). Kd

values of sulfonamides in soils have been reported to fall in the range of0.6 to 7.4 L kg−1 (Sarmah et al., 2006). Park and Huwe (2016) foundSMX Kd values ranging between 1.1 and 1.39 L kg−1 for Korean soils atsoil pH (5.5). Srinivasan et al. (2013) reported Kd values for SMX inclay-loam, silt, and silt-loam soils from New Zealand ranging between 2and 4 L kg−1 with soil pH ranging between 4 and 8. Boxall et al. (2002)observed Kd values of sulfachloropyridazine varying between 0.9 and1.8 L kg−1 (in silty-clay and sandy-loam soils, respectively). Due to

https://doi.org/10.1016/j.envint.2019.104905Received 7 February 2019; Received in revised form 30 May 2019; Accepted 6 June 2019

⁎ Corresponding author.E-mail address: celine.duwig@ird.fr (C. Duwig).

1 Present address: Estación Regional del Noroeste (ERNO), Hermosillo, Sonora, Mexico.

Environment International 130 (2019) 104905

Available online 21 June 20190160-4120/ © 2019 Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).

T

these low Kd values, Sulfonamides are supposed to be highly mobile insoils (García-Galán et al., 2008) with high potential of groundwatercontamination (Archundia et al., 2017a; García-Galán et al., 2010;Kümmerer, 2009; Mojica and Aga, 2011; Tamtam et al., 2008). Themobility of Sulfadiazine studied with column displacement experimentsby Wehrhan et al. (2007) was quite high as it exhibited a retardationfactor of 2 to 5 compared to a conservative tracer (Cl−). Likewise, Hillet al. (2019) assessed the leaching of sulfonamides (including SMX) inundisturbed soil columns under unsaturated conditions observing thatall sulphonamides leached from the columns slightly after the con-servative tracer.

Kurwadkar et al. (2011) found that sulfonamides mobility was de-pendent on pH, soil charge density, and contact time; the effect of pHwas found to be most pronounced in sandy loam soils. In addition to theprocess of sorption with soil organic carbon, dissolved organic carbon(DOC) content was also shown to affect antibiotics mobility. Zhanget al. (2016) observed that sulfonamides are highly mobile in loamy-sand and silty-sand soils from China, and that irrigation with treatedwater increases their mobility as DOC increases antibiotics apparentsolubility. In parallel, DOC can form complexes with antibiotics thusreducing their mobility through mechanisms of co-sorption on solidphases (Chefetz et al., 2008).

For organic and charged compounds such as SMX, a large variety ofinteraction mechanisms is possible, such as complexation, ion bridgingor covalent binding, polar interactions as well as hydrophobic inter-actions, which can modify their sorption and mobility (Morel et al.,2014; Srinivasan et al., 2014). SMX is characterized by two acid dis-sociation constants (pK1=1.6 and pK2=5.7 (Boreen et al., 2004)).SMX anionic (SMX−) and cationic (SMX-H2

+) forms dominate atpH > pK2 and pH < pK1, respectively. At pK1 < pH < pK2 theneutral form (SMX-H) dominates. Overall, SMX speciation significantlyaffects its tendency to react with solid surfaces and to chelate metalions, and to undergo biodegradation or photolysis (Calvet, 1989; Dıaz-Cruz et al., 2003; Kim et al., 2011; Srinivasan et al., 2014; Srinivasanand Sarmah, 2014a). Tamtam et al. (2011) reported that SMX can beassociated with clay‑iron complexes through electrostatic adsorption,while neutral SMX can be easily adsorbed to soil organic carbon(Kurwadkar et al., 2011).

Mechanistic studies on the mobility of sulfonamides in natural soilsremain still limited. They are rarely associated with a modelling ap-proach, essential for identifying the relevant sorption processes con-trolling the observed transport behaviour. One way of explaining solutetransfer that deviates from classical convection dispersion equation is toconsider physical and/or chemical nonequilibrium. Generally, physicalnonequilibrium transport models assume a two-region or dual-porositytype formulation that partitions the liquid phase into mobile and im-mobile regions, while chemical nonequilibrium transport models as-sume kinetic interactions between solutes in the liquid and solid phases(Simunek et al., 2013).

For example, Wehrhan et al. (2007) observed that kinetic sorptionincluding two reversible kinetics and one type of irreversible sorptionsites was the most relevant process for the transport of sulfadiazine in asilty loam soil. Martínez-Hernández et al. (2017) observed that SMXtransport in a sandy-loam soil can be well described with an equili-brium sorption model accounting for non-linear sorption. Engelhardtet al. (2015) observed that sulfadiazine transport in undisturbed andunsaturated sandy and silty-clay soils is efficiently modelled with adual-permeability model in relation with the presence of preferentialflow pathways.

However, we lack a global understanding of SMX mobility in dif-ferent types of soils. A limited number of studies have been conductedto date to evaluate its the mobility in natural soils, but none in parti-cular in arid soils from the Altiplano in the central Andes and at acatchment scale.

The objective of the present study was to evaluate and model thepotential mobility of the sulfonamide antibiotic SMX in natural soils

with contrasted properties and land uses by identifying the more re-levant processes leading to SMX retention or retardation. Displacementexperiments were conducted in repacked columns of 10 soils, of vari-able mineralogical and physico-chemical characteristics, collectedthroughout the Bolivian Altiplano (Katari watershed). The Katari wa-tershed supports one of the fastest growing cities in Latin America (ElAlto city). SMX was detected in soils at concentrations ranging from 0to 18 μg kg−1 (Archundia et al., 2017a). Experimental and field workpermitted to evaluate the potential of dissemination of SMX in thestudied zone and the vulnerability of surrounding water bodies such asthe Lake Titicaca and groundwater. Inverse modelling performed withHYDRUS-1D was efficiently applied to fit SMX column transport resultsconsidering different reactivity and hydrodynamic concepts.

2. Material and methods

2.1. Soils

2.1.1. Soils description and samplingSoils description was conducted in April 2014 during the wet

season, when soil profiles were opened at nine locations along the fourmain rivers of the studied watershed (Seke, Seco, Pallina and Kataririvers, described in Archundia et al. (2017b) according to the metho-dology proposed in the Guidelines for Soil Description (FAO, 2006).They represent the variety of soil types and soil use in the catchment.Soils 1 and 2 correspond to high altitude soils of non-urban influence.Soil 3 corresponds to a soil of low-urban influence. Soils 4 and 5 werecollected at the vicinity of the wastewater treatment plant (WWTP) ofPuchukollo, but only the profile of soil 5 was described. Soil 4 perma-nently receives untreated wastewater, and soil 5 is part of an agri-cultural trial receiving treated wastewater every week for 2months.Soil 6 was sampled in the lower Pallina River in a cropped barley plot.Soils 7 and 8 were sampled in a barley plot located in the upper andlower zones of the Katari River, respectively. Soils 9 and 10 weresampled on the shores of Lake Titicaca, in cropped quinoa and potatoplots respectively, which are partially flooded when the lake level ishigh. According to Bolivian FAO soil maps (FAO, 1998), soils 1 to 5 areclassified as Distric Regosols and soils 6 to 10 as Eutric Cambisols(Fig. 1).

At each profile, soil samples were taken in each horizon and com-posite soil samples were taken at the surface layer (0–15 cm) for thecolumn experiments, with the same amount of soils collected at 5 dif-ferent points separated by 5m within the same plot. In the laboratory,composite soil samples were air-dried and sieved (2-mm).

2.1.2. Soil chemical characteristicsFor the topsoil composite samples, chemical analyses were per-

formed at the INRA Analysis Laboratory (Arras, France). The soil ana-lysis for lower horizons was performed at IBTEN Laboratory (Bolivia).Total nitrogen (N) was analysed by dry combustion (norm NF ISO13878). Organic carbon (OC) was determined by direct determinationby dry combustion (norm NF ISO 10694). For iron oxides (Feoxy)contents, an extraction was conducted by the Mehra and Jacksonmethod (Mehra and Jackson, 1960), and the determination of Fe con-centrations was conducted by ICP-AES. pH was measured with a MettlerToledo electrode at a soil to water ratio of 1/5 (v/v) (norm NF ISO10390). Cu extraction was performed based on the US EPA 3052method by microwave-assisted extraction (MAE) as described inArchundia et al. (2017b). Final suspensions were analysed by ICP-AESwith external calibration (Limit of Detection, LD: 1.2 μg L−1 and Limitof Quantification, LQ: 3.7 μg L−1). Quality assurance and control (QA/QC) of the analysis were performed with Tibet Sediment (NIMGBW07323) certified reference material (CMR). Recovery for Cu was of71.6%.

D. Archundia, et al. Environment International 130 (2019) 104905

2

2.1.3. Soil physical characteristicsSoil texture analysis was carried out at the INRA Laboratory

(France) with a five particle size fractions granulometer without furtherdecarbonisation (norm NF X 31-107). Mineralogy was determined by X-ray diffraction (Siemens D5000 and Bruker Axs D8). Bulk density wasdetermined from 3 soil surface samples carefully recovered with a140 cm3 metal ring, after drying at 105 °C for 24 h.

To determine the in situ saturated hydraulic conductivity (Ksat), weused a tension disc infitrometer of 20 cm diameter set to subsequentpressure head, h, of −10, −5, and −0.5 cm at the surface of soils re-presentative of the upper (soils 2 and 3) and the lower (soils 6, 7 and 8)Katari catchment. Following the protocol described in Müller et al.(2012), we determined the relationship between hydraulic conductivityand tensions by assuming that the Darcy flux (q, cm min−1) under thedisc was given by Wooding's equation (Wooding, 1968) and that thehydraulic conductivity followed an exponential function with pressionhead (Gardner, 1958):

=K K esatαh( ) (1)

where Ksat is the saturated hydraulic conductivity (cmmin−1), h thepressure head (cm) and α a soil-dependent parameter (cm−1). We de-rived the parameters Ksat and α through a linear regression of the datapoints (h, ln q).

2.2. Column transport experiments

2.2.1. Column experimentsThe methodology was adapted from the OECD guideline “Leaching

in Soil Columns” (OECD, 2003). Experiments were done in duplicate foreach soil on the topsoil composite samples, under steady state and near-saturated conditions, using bromide (Br−) as nonreactive solute to es-timate soil hydrodynamic transport parameters. Experiments were doneat constant ionic strength with a 0.01M CaCl2 background solution and

at soil pH.The main experimental conditions are shown in SI.II. Air-dried

sieved soils were packed into glass columns of 2.6 cm internal diameterto a depth of about 10 cm at a bulk density close to the field ones,except for soils from the upper watershed, where it was impossible toreach the field bulk density because of the high organic carbon content.The column was then weighed and the soil bulk density calculated. Atthe bottom of the column a peristaltic pump pushed the solutes towardsthe top of the column. The column was first saturated by injecting abackground solution of 0.01M CaCl2 at a low flow rate of0.03 cm3min−1 during 1 pore volume (PV). Water flow was then in-creased to 0.3 cm3min−1 (equal to a Darcy flow of 0.9 10−5 m s−1).This flow was adjusted to field soil saturated hydraulic conductivities.At the outlet of the column, the leachates were sampled with an auto-matic fraction collector (Gilson FC203B). At the beginning of a columntransfer experiment, about two PVs of 0.01M CaCl2 solution wereleached through the column using a peristaltic pump to reach a steady-state flow rate and a stable electrical conductivity in the effluent. Thecolumn was weighed to determine the volumetric water content. Then0.01M bromide was injected during ~0.5 PV and leached with 2.5 PVsof 0.01M CaCl2. Then, one PV of 0.25mg L−1 SMX solution with a finalcontent of 0.1% MeOH (diluted from 1 g L−1 SMX-methanol using0.01M aqueous CaCl2) was injected and leached with 3 to 10 PVs of0.01M CaCl2 solution depending on the soil types.

During steady-state conditions, the flow rate was fixed at0.3 cm3min−1 and the working pressure was consequently highlyvariable between soils: the bottom-up flow ensured working in near-saturation conditions. The Eluted Mass (EM) was controlled byweighing both the input solution container (continuous measurement)and the collector tubes. Br− and SMX were measured in each tube, andpH in each 5th tube.

Fig. 1. Study site and sampling points. e: refers to the entrance of the wastewater treatment plant (WWTP) of Puchukollo. o: refers to the outlet of the WWTP. CMe(Eutric Cambisol), RGd (Distric Regosol) and LPe (Eutric Leptosol) refer to soil types (FAO, 1998).

D. Archundia, et al. Environment International 130 (2019) 104905

3

2.2.2. Chemical analysisBr− was measured by ion chromatography (Metrohm 732/733 se-

paration center, MetrosepASupp 16–150mm, and Metrosep C2-150mmanion column) based on the USEPA Method 300.0. Calibration stan-dards were prepared from an Anion Multi-Element Standard (Roth).Limits of detection (LD) and of quantification (LQ) were determinedbased on a signal to noise ratio of 3 for both LD (2.66 10−3 mg L−1) andLQ (7.97 10−3 mg L−1).

After samples filtration (Chromafil Xtra PVDF, 0.45 μm) SMX wasanalysed by HPLC-UV (Spectra system UV 100, Thermo SeparationProducts) equipped with a C-18 (EC 125/3 NUCLEOSIL 100-5 C18,5 μm) column. The detection wavelength was 269 nm and the tem-perature was maintained at 40 °C. The mobile phase was composed ofACN (15%), MeOH (15%), ultrapure water (70%) and formic acid(0.02%) with a fixed flow rate of 0.5mLmin−1. Time analysis was fixedto 10min, with an injection volume of 20 μL and three injections persample. Quantification was carried out by external calibration andmeasurement of peak areas. The LD and LQ of SMX were 0.1mg L−1

and 0.5mg L−1, respectively (signal to noise ratio of 3 for both). Theanalytical method was validated by the method of accuracy profiles(Feinberg and Laurentie, 2010). No SMX and Br were detected in theleachates of non-contaminated soils.

2.2.3. Reactive transfer parametersIn modelling, the soil saturation condition was assumed. To confirm

the saturated condition, the soil water content was compared to theporosity n (SI.II), which was determined as:

= −nρρ

1 d

s (2)

where ρs is the solid particle density (g cm−3) and ρd is the soil bulkdensity (g cm−3). To consider the wide range of organic carbon content(Corg) of our soils, which can affect the value of solid density, the fol-lowing equation to estimate the solid particle density was taken fromRühlmann et al. (2006):

= −ρ C2.583 0.025s org (3)

The reactive transfer parameters (dispersivity, distribution coeffi-cient, sorption/desorption rate, fraction of instantaneous sorption sites,irreversible sorption first-order constant) were obtained by fitting anumerical solution of the convection dispersion equation (CDE) to theexperimental breakthrough curve (BTC) (SI·I Eq. (7)). The degradationconstant was fixed with experimentally obtained values from batchexperiments realized in 250mL glass bottles with soil slurry (data notpublished, Table 2) using experiments performed following the proce-dure described by (Martins and Mermoud, 1998).

The finite element code HYDRUS-1D (Simunek et al., 2008) hasalready been successfully used to study the fate and transport ofpharmaceuticals (García-Santiago et al., 2017; Martínez-Hernándezet al., 2017). We used it in inverse modelling mode. The dispersioncoefficient D (SI·I Eq. (7)) was first determined by fitting the numericalsolution of the CDE to the experimental bromide BTC. The bromide wasassumed to be inert and considered as a water tracer. The fitting of thebromide BTC allowed the determination of the dispersivity, λ (cm), foreach column experiment:

= D νλ / (4)

where v is the average pore water velocity (cmmin-1).This equation considers that molecular diffusion is negligible com-

pared to hydrodynamic dispersion, which is valid for Peclet numbers(Pe= νL/D) larger than five, according to Kutilek and Nielsen (1994).For the velocity flux range used in this study, the Peclet number wasalways larger than five.

The second step was to adjust the numerical solution of the con-vection-dispersion equation (CDE) with reactive solute transport to theexperimental SMX BTC.

To evaluate the fit goodness, we used different statistical criteria(Loague and Green, 1991):

The coefficient of determination (R2) was calculated as shown byEq. (5):

∑ ∑= − −= =

Oi O Pi OR ( ) / ( )i I

n

i I

n2 2 2

(5)

Model efficiency (EF) is given by Eq. (6). Perfect agreement of themodel and experimental data is given by EF =1.

∑ ∑ ∑= ⎛

⎝⎜ − − ⎞

⎠⎟

=−

= =−EF O O Pi Oi O O( ) ( ) / ( )

i I

n

ii I

n

i I

n

i2 2 2

(6)

where P are the predicted values; O are the observed values; n is thenumber of samples; and O is the mean of the observed data.

3. Results and discussion

3.1. Soil properties

For the upper Katari catchment a Ksat of 0.5 (± 0.1) 10−5 m s−1

was obtained and the average bulk density was 1.2 g cm−3. These va-lues were attributed to soils 1 to 5. In the lower Katari watershed, wefound a Ksat 1.7 (± 1.5) 10−5 m s−1 for freshly ploughed soils and 0.3(± 0.1) 10−5 m s−1 for soils left fallow, with an average bulk density of1.3 g cm−3. These values were attributed to soils 6 to 10.

Soil profiles description can be found in SI.III. Soils 1, 2 and 3 su-perficial horizons (0–15 cm) are organic of granular structure with highporosity, with extremely acid to highly acid pH (pH from 4.6 to 5.4) dueto the high OC content (3.6 to 8.2%). Below the upper 15–20 cm layer,the sand, gravel and stones contents increase considerably, favouringvertical water infiltration as well as lateral water flow in the slope di-rection. For soils 4, 5, 6, 7 and 8, the pH in the superficial horizon variesfrom highly acidic (5.4) to slightly alkaline (8.5). These soils aremoderately deep and the A horizon has loamy or loamy/sandy texture,with low OC content. Only soil 4 presents a high OC content, due tocontinuous arrival of waste waters from the WWTP. The sand andgravel contents increase with depth in soils 5 and 6, favouring poten-tially a good vertical water transfer. In contrast, in profiles 7 and 8, anincrease of the clay content with depth may limit water infiltration.Superficial horizons of soils 9 and 10 are of loamy texture and havemoderate fertility and slightly alkaline pH (8.3 and 8.4). Finer materialssuch as silt and clay increase in lower horizons of which soils, whichcould limit water infiltration compared to topsoil horizons.

3.2. Bromide transport

Table 1 presents the model estimated and calculated parameters aswell as EM (eluted mass) fractions of bromide BTCs. The bromide (Br−)mass balance varied between 97% and 110% (Table 1). No tailings wereobserved in bromide BTCs except in soils 9 and 10 (Fig. 2). The absenceof significant tailings indicates the presence of physical equilibrium, i.e.a mono-modal flow. In the case of soils 9 and 10, bi-modal flow (orphysical non-equilibrium) could have slightly affected bromide trans-port. Transport parameters derived from the bromide conservativetracer BTCs are shown in Table 1. The determination coefficients (R2)always indicated a good fit of the data (R2 ranging between 0.9 and0.99). Dispersivity values, λ, calculated for the 10 soils, varied between0.02 and 1.64 cm. Due to packing differences, column duplicates fromthe same soil did not have the same bulk density, leading to differencesin the fitted dispersivity values. For soils 1 to 4, with high organiccarbon contents, the process of packing was difficult and the soil bulkdensity in the columns did not reach the field ones (SI.II). As found byPerfect et al. (2002) observed OC contents and λ values were wellcorrelated (R2=0.5). This correlation could be due to the fact thatorganic carbon is involved in the binding of small soil particles into

D. Archundia, et al. Environment International 130 (2019) 104905

4

larger particles, which increases soil aggregation and thus soil tortu-osity for water pathways.

The degree of water saturation varied between 0.8 and 1 (i.e. sa-turation conditions for most of the columns, SI.II) except for column 1A(0.72), for which this low water saturation level may have induced thepresence of immobile water (trapped by air bubbles) explaining theobserved low retardation factor of Br-. Variations in water saturationare due to specific behaviour of the soils and/or water content esti-mation errors, in the case of the column 1A water saturation was notreached.

As a fixed retardation factor of 1 (value for a water tracer) did notlead to good fits of the experimental bromide BTC for some soils, aretardation factor had to be fitted for all soils. The retardation factorvaried within 15% of the unit value for almost all columns indicating anacceptable conservative tracer behaviour of the bromide ion.

However, for soils 6, 9 and 10, the fitted retardation factor of Br−

was particularly low (0.64 to 0.76). It can be assumed that these low Rvalues relate to anionic exclusion or exchange processes, which areknown to affect the mobility of negatively charged chemicals in soil.Repulsion of Br- by negative surface charges (from organic matter sitesionised at high pH and clay exchanger sites) prevents Br− ions fromapproaching the slowly flowing surface water layers, thus reducing theeffective porosity seen by Br−. Significant reduction of effective por-osities for Br− were observed (10 to 50%; James and Rubin, 1986,Melamed et al., 1994, Gvirtzman and Gorelick, 1991). This reductiongenerates a corresponding diminution of fitted R values below one.Within the alkaline soils (soils 6 to 10), Br− in the soils 6, 9 and 10 mayhad encountered this situation as they have high OC and high claycontent, compared to soils 7 and 8.

3.3. Sulfamethoxazole transport

3.3.1. Experimental SMX breakthrough curvesThe recorded pH values in column leachates (Table 1) were com-

parable to the pH of bulk soils (SI.III) and remained stable during thewhole displacement experiment (standard deviations ranging between0.05 and 0.2).

The center of gravity of the SMX BTCs was delayed relatively to thebromide BTC in all studied soils (Fig. 2), with much higher SMX re-tardation in soils 1 to 4 than in soils 5 to 10. This is indicative of thepresence of retention processes linked with the reactivity of SMX and

with soils properties. Tailings were observed in all SMX BTCs except inthose for soils 6, 7 and 8. The BTCs tailings were generally much moreextended in soils 1 to 4 in which SMX retardation was higher, sug-gesting the existence of hysteresis in the desorption process.

3.3.2. Simulated SMX breakthrough curvesThe model estimated parameters as well as the EM fractions and

model statistics (Eqs. (5) and (6)) are presented in Table 2.The EM fractions were calculated from simulated BTCs, the EM

fractions varied strongly between 35% and 99%. They were higher(between 67% and 99%) for the weakly SMX-binding soils 5 to 10 thanfor soils 1 to 4 (between 67% and 35%). This mass deficit was con-sidered to be related to: (1) biodegradation and (2) irreversible sorp-tion. From independent microcosm experiments (Archundia, 2016),first-order degradation rates (μ) were determined as ranging between7.6 10−7 to 7.8 10−6 min−1, depending on the soil (Table 2). However,these rates were too small to explain the mass balance deficit. Conse-quently, the experimentally determined SMX depletion (i.e. the massbalance deficit) was represented as a sum of two parameters in theHYDRUS-1D modelling: a fixed biodegradation rate μ (fixed; minorcontribution) and an irreversible sorption rate β3 (fitted, major con-tribution; Table 2).

Table 2 shows the best model explaining SMX sorption behaviourfor each soil column. SMX BTC in soil 1 was best represented by the 3sites-2 rates- Freundlich sorption model (3-2-F in Table 2). SMX BTCs insoils 2 to 4 were best represented by the 3 sites-2 rates- linear sorptionmodel (3-2-L in Table 2). SMX BTC in soil 5 was best represented by the2 sites-2 rates- Freundlich sorption model (2-2-F in Table 2). SMX BTCsin soils 6 to 10 were best represented by the 2 sites-2 rates- linearsorption model (2-2-L in Table 2). In the 2 sites-2 rates- Freundlich/linear and irreversible sorption models, the instantaneous sorption sites(S1) were omitted (Wehrhan et al., 2007), while the 3 sites-2 rates-Freundlich/linear sorption models consider the three types of sorptionsites, considered to exist independently (i.e. instantaneous, rate-limitedand irreversible sorption as described in the theory of reactive solutetransport section of SI·I). Observed differences could be attributed tothe elevated OC contents of soils 1–4 (6 to 10%) which makes thesorption process more complex. Elevated OC content contributes to theheterogeneous distribution of SMX adsorption sites between the soilaggregates where instantaneous sorption will take place and the ag-gregates internal pores where rated-limited sorption will exist thus

Table 1Transport parameters, Eluted mass (EM) fractions of bromide-tracer BTCs and average pH of leachates during the experiment (standard deviation is given inparenthesis).

Soil Duplicate EM (%) R2 EF R v (cmmin−1) λ (cm) pH

1 A 0.97 0.98 0.98 0.85 0.11 0.08 4.8 (0.2)B 1.04 0.99 0.99 1.02 0.09 0.13 4.8 (0.2)

2 A 1.10 0.99 0.99 1.1 0.1 0.14 4.7 (0.1)B 1.06 0.9 0.93 1.1 0.11 0.18 5 (0.1)

3 A 0.99 0.90 0.90 0.93 0.08 0.06 6 (0.05)B 1 0.99 0.89 1 0.09 0.06 5.3 (0.2)

4 A 1 0.99 0.99 0.85 0.1 0.29 5.1 (0.2)B 1 0.93 0.82 0.91 0.1 0.54 5.9 (0.1)

5 A 0.99 0.99 0.99 0.95 0.11 0.02 6.3 (0.3)B 0.99 0.99 0.99 0.95 0.1 0.1 6.3 (0.3)

6 A 1 1.00 1.00 0.79 0.08 0.03 6.9 (0.02)B 0.99 0.96 0.95 0.72 0.1 0.06 7 (0.2)

7 A 0.99 0.96 0.96 1.07 0.11 0.03 7.5 (0.1)B 0.99 0.99 0.69 1.07 0.1 0.02 7.2 (0.1)

8 A 1.07 0.97 0.97 1 0.11 0.09 7.6 (0.08)B 1.02 0.99 1.00 0.92 0.12 0.03 7.5 (0.01)

9 A 0.98 0.95 0.96 0.76 0.11 1.64 7.8 (0.2)B 1.03 0.96 0.99 0.64 0.09 1.06 7.4 (0.2)

10 A 1.01 0.92 0.94 0.74 0.12 0.97 8 (0.2)B 1.09 0.93 0.98 0.72 0.12 0.77 7.8 (0.06)

EM: eluted mass fractions. EF: model efficiency. RMSE: root mean square error. R2: coefficient of determination. R: retardation factor for bromide (fitted). v: porewater velocity (obtained experimentally). λ: dispersivity (fitted).

D. Archundia, et al. Environment International 130 (2019) 104905

5

Fig. 2. Duplicate experimental (dots) and fitted (lines) BTCs for Br− tracer (gray line) and for SMX (black lines). The corresponding fitting parameters are presentedin Tables 1 and 2.

D. Archundia, et al. Environment International 130 (2019) 104905

6

retarding their migration (Pignatello, 2000).Despite the good fittings with model 2-2-L-I (see Table 2 for detailed

description) it failed to describe extended tailings of SMX BTCs for soils9 and 10, as shown by the lower R2 and EF values for columns 9A, 10Aand 10B (Fig. 2, Table 2). For these two soils, the SMX BTCs tails mightbe rather due to the addition of multi-modal flow processes, as detectedin the bromide BTCs tailings. The observed SMX tails for soils 9 and 10may thus be related to diffusive exchange between highly mobile so-lutes in preferential flow pathways and solutes in relatively stagnantwater (Johnston et al., 2005).

There is considerable evidence that the sorption of organic com-pounds such as SMX has an instantaneous equilibrium component and aslower time-dependent component (Martins and Mermoud, 1998;Pignatello, 1989), leading to BTCs with tailings. Diffusion of SMXmolecules at the surface of soil organic carbon (for short times) andintra-organic carbon diffusion (for longer times) are possibly re-sponsible mechanisms for rate-limited sorption, as has been alreadyobserved for a wide range of other organic compounds such as theherbicide simazine (Fortin et al., 1997) or the pesticide parathion(Leenheer and Ahlrichs, 1971). Furthermore, the contribution of mi-neral surfaces to the observed rate-limited sorption has to be taken intoaccount, as it is well known that polar compounds, such as SMX, canhave significant interactions with inorganic soil components (Martinsand Mermoud, 1998). However, it cannot be excluded that at least apart of the observed tailing was caused by slow (rate-limited) SMXdesorption (Carrillo et al., 2016). Drillia et al. (2005) reported similarresults for SMX while Wehrhan et al. (2007) observed that the sorptionof sulfadiazine was also rate-limited.

For all BTCs, the best fits were obtained with the models that partlyconsider irreversible sorption (I term in the names of models inTable 2), which is in accordance with the calculated EM fractions(Table 2). Irreversible sorption can be attributed to the formation oforganic complexes between the chemicals in the irreversible compart-ment and the colloidal organic carbon of the soil. Furthermore, Kanet al. (2000) proposed that at least a part of the irreversible sorption ofhydrophobic organic compounds can presumably be caused by

entrapment in soil organic carbon increasing the constrictivity of thesolid phase to chemical diffusion.

3.4. Relation between transport parameters and soil properties

To facilitate the determination of the retardation factor for soils 1and 5, which were best described by the Freundlich model we fitted itagain by imposing a linear isotherm. The obtained Kd values are shownin Table 2. The fits were still acceptable, with R2 and EF varying be-tween 0.9 and 0.99, whereas they varied between 0.94 and 0.99 withthe Freundlich isotherm.

The comparison of SI.III and Table 2 shows that compared to soils5–10, SMX in soils 1–4 presents a markedly higher retardation factor(6.9 to 8.4 vs 1.1 to 1.8) probably related to the markedly higher OCcontent (3.6 to 10.3% vs 0.4 to 1.8%) and lower pH (4.6 to 5.6 versus6.7 to 8.5) of these soils. These findings indicate that the variable re-tention time of SMX in the soils is essentially a function of soil pH andOC content.

At pH > 5.7 the SMX is dominantly negatively charged. SMXcontains complexing amine groups, which suggests the preferred for-mation of inner-sphere surface complexes over weaker charge exchangerelated sorption processes (e.g. anion exchange). A specific processconcerns ion bridging, i.e. the complexation of SMX by positivelycharged trace metals sorbed on solid surfaces. This was observed withCu(II) by Morel et al. (2014), who found that the sorption of SMX in soilincreases strongly in the presence of copper, supporting the assumedformation of ternary SMX–Cu–soil complexes. All strongly sorbing andpoorly soluble bivalent/trivalent surface cations (Fe(III), Al(III), Zn(II),Ni(II), Cd(II)…) may be involved in SMX ternary surface complexationformation, but the complexation constants of copper for organics areknown to be particularly high. Therefore, the formation of Cu bridgingcomplexes could be specifically developed in soil 4, which presents thehighest content of Cu, originating from wastewater effluents, or for soil2, which has a high content of geogenic Cu. At pH < 5.7, i.e. in soils 1to 4, the sorption of SMX is relatively strong. Accordingly, SMX sorptionis found in the literature to increase at decreasing soil pH (Boxall et al.,

Table 2Fitting parameters of the best fitting model of SMX BTCs.

Soil Duplicate Best model EM (%) Model statistics Fitted parameters Fixed parameter

R2 EF Kd

(L kg−1)R F α

(min−1)β3(min−1)

μ(min−1)

1 A 3-2-F-I 50.6 0.98 1 4.16 8.35 3.6E-01 (0.01) 7.5E-03 (5.E-04) 6.1E-03 (2.3E-04) 0B 3-2-F-I 59.8 0.99 0.99 5.30 8.33 5.0E-01 (0.08) 8.4E-03 (3.E-04) 4.4E-03 (1.9E-02) 0

2 A 3-2-L-I 58.8 0.91 0.90 3.12 7.73 4.6E-04 (5.E-05) 3.9E-02 (3.E-03) 7.4E-03 (3.8E-04) 5.6E-6B 3-2-L-I 66.7 0.96 0.96 2.90 7.02 1.1E-02 (0.4) 4.1E-02 (2.E-02) 4.9E-03 (2.6E-02) 5.6E-6

3 A 3-2-L-I 40.0 0.97 0.95 4.41 7.69 5.9E-01 (0.01) 7.2E-03 (6.E-04) 8.9E-03 (2.9E-04) 7.8E-6B 3-2-L-I 48.0 0.89 0.89 4.24 6.90 4.3E-03 (4.E-03) 1.1E-02 (1.E-04) 6.4E-03 (9.9E-05) 7.8E-6

4 A 3-2-L-I 42.0 0.95 1 3.87 6.84 1.1E-01 (0.02) 8.6E-03 (5.E-04) 5.4E-03 (8.1E-04) 6.6E-6B 3-2-L-I 35.0 0.96 0.96 3.93 7.11 3.0E-01 (9E-03) 7.8E-03 (4.E-04) 7.5E-03 (1.8E-04) 6.6E-6

5 A 2-2-F-I 67.0 0.88 0.90 0.31 1.81 0 1.5E-01 (5.E-02) 3.9E-03 (7.1E-04) 2.6E-6B 2-2-F-I 82.5 0.99 0.99 0.4 2 0 9.2E-02 (4.12E-03) 1.7E-03 (1.7E-04) 2.6E-6

6 A 2-2-L-I 99.0 0.97 0.97 0.06 1.11 0 8.19E-02 (3.7E-02) 1.2E-04 (3.2E-04) 3.5E-6B 2-2-L-I 82.7 0.97 0.97 0.4 2 0 1.56E-01 (2.4E-02) 1E-06 (6.01E-17) 3.5E-6

7 A 2-2-L-I 98.0 1 1 0.04 1.10 0 2.E-01 (3.E-02) 2.1E-04 (3.5E-04) 7.6E-7B 2-2-L-I 90.0 0.99 0.98 0.06 1.14 0 1.E-01 (2.E-02) 1E-06 (4.1E-04) 7.6E-7

8 A 2-2-L-I 94.0 0.99 0.99 0.12 1.31 0 1 (1) 3.1E-03 (4.5E-04) 9.7E-7B 2-2-L-I 90.0 0.99 0.99 0.08 1.25 0 8.8E-01 (3.E-01) 3.7E-03 (4.4E-04) 9.7E-7

9 A 2-2-L-I 98.0 0.9 0.60 0.19 1.47 0 7.2E-03 (5.E-02) 2.6E-03 (1.8E-03) 9.7E-7B 2-2-L-I 96.0 1 0.78 0.07 1.18 0 1.1E-01 (9.E-03) 3.4E-04 (2.8E-04) 9.7E-7

10 A 2-2-L-I 82.9 0.86 0.86 0.19 1.57 0 1.29E-02 (5.2E-03) 2.6E-03 (1.2E-03) 3.8E-6B 2-2-L-I 83.0 0.80 0.76 0.11 1.32 0 1.8E-02 (2.E-03) 4.E-03 (9.7E-04) 3.8E-6

Model refers to the model that best fits the observed data. Names of models are composed as follows: number of sorption sites-number of rates-sorption concept (L:linear and F: Freundlich)-reversibility of sorption process (irreversible: I and reversible: R). EM: Eluted mass fraction, R2: coefficient of determination, EF: modelefficiency, R: retardation factor calculated with Kd value, Kd: distribution coefficient for the linear model, F: Fraction of adsorption sites available for instantaneoussorption, ß3: Irreversible adsorption rate coefficient, α: First order coefficient rate for one site or two sites non-equilibrium adsorption, μ: first-order degradation ratecoefficient determined from batch experiments (Archundia, 2016). Standard deviation is given in brackets.

D. Archundia, et al. Environment International 130 (2019) 104905

7

2002; Thiele-Bruhn and Aust, 2004). This is an expected behaviour fornegatively charged ligands, such as the negatively charged SMX specieswhatever the binding mode – surface complexation or anion exchange(Stumm, 1992). But in this pH range neutral aqueous SMX-H speciesdominate the deprotonated ones. This could lead to specific hydro-phobic interactions with soil organic carbon, meaning an even highersorption at low pH. Dissolved organic carbon (DOC) could potentiallyincrease ligand mobility as suggested for Naproxen (Chefetz et al.,2008). We did not measure DOC but expect its concentration to be toolow to become significant in this respect.

Altogether, these results are in accordance with Srinivasan andSarmah (2014b) and Kurwadkar et al. (2011), who found that SMXsorption is greatly enhanced in soils with low pH and high CEC, OC andclay contents. Overall, pH and OC contents are the parameters thatbetter explained the observed SMX retardation factors in all soils(R2= 0.83 for both).

Concerning the other fitted parameters, marked differences werealso noted between soils 1–4 and soil 5–10. In soils 1 to 4, F, the fractionof adsorption sites available for instantaneous sorption, varied between4.6 10−4 and 5.9 10−1 and were highly variable between duplicatecolumns for soils 2 and 3, showing the complexity of building realduplicates of repacked soil columns. These values varied in the samerange as in Wehrhan et al. (2007) studying sulfadiazine transport inEutric Cambisol. For soils 5 to 10, F was equal to 0, meaning that in-stantaneous SMX sorption was negligible. The irreversible sorption ratefactor, β3, varied from 4.4 10−3 to 8.9 10−3 min−1 for soils 1 to 4, andfrom 1 10−6 to 4 10−3 min−1 for soils 5 to 10, and was negativelycorrelated with the EM fraction (R2= 0.74). β3 was higher than μ by afactor of 102 to 104, demonstrating that irreversible SMX sorption (β3)was the main process explaining the EM fractions lower than 100%, andthat biodegradation was negligible during the time of the experiment.However, if we consider experimental and analytical uncertainty, theEM of soils 5 to 8 can be potentially equated to 100%.

3.5. SMX mobility at the watershed scale

The variable pedology in the studied watershed is reflected in avariable mobility of SMX. In the soils above El Alto city (soils 1, 2 and3) SMX showed low mobility (i.e. high retardation factors~7 to ~8),suggesting low vulnerability of groundwater to SMX contamination.According to the observed EM fractions, 40 to 66% of SMX are irre-versibly retained in the surface soil horizon. Nevertheless, infiltration tolower horizons and to the aquifers remains possible, especially in re-lation with the sandy and gravel texture of the deeper layers (see SI.IIISoil profiles properties), although deep SMX transport will be mostlycontrolled by its persistence in soil profile. This is especially critical forthe recharge zone of the El Alto Aquifer (soils 1, 2 and 3).

In soils near the WWTP (soils 4 and 5), SMX showed relatively lowmobility (retardation factors of ~2 to ~7). Here, SMX mobile fractions(~35 and ~80% in soils 4 and 5, respectively) are likely to be trans-ported below the soil surface layer in the dry season and at the be-ginning of the wet season. In this season, cracks are indeed often formedin soils down to lower horizons due to their high contents of smectiteclay, which induce soil shrinking phenomena and the establishment ofpreferential flows, which are known to favour pollutant transport(Keesstra et al., 2012; Martins et al., 2013).

In soils 6, 7, and 8 (Fig. 1), SMX is expected to be highly mobile(mean retardation factor of 1.3 and mean SMX mobile fraction: 92.2%).However, the higher clay content in lower horizons (see SI.III Soilprofiles properties) suggests that in wet season, SMX leaching towardsthe groundwater could be slowed down but could speed up when cracksform in clayey lower horizons, as similar to soils 4 and 5.

SMX also present high mobility in soils located on the shore of LakeTiticaca (soils 9 and 10) (mean retardation factor of 1.38 and meanSMX mobile fraction: 89.9%). In this zone, the vulnerability ofgroundwater to SMX pollution is clearly very high because of the

shallow groundwater levels in relation with to the lake's level fluctua-tion.

4. Conclusions

The transport of SMX in a variety of high altitude soils from theBolivian Altiplano was investigated using dynamic displacement ex-periments in repacked laboratory soil columns.

SMX eluted mass fractions were lower than 100% for all the studiedsoils, suggesting the existence of irreversible sorption and/or degrada-tion processes during the antibiotic transport in soil. This was supportedby HYDRUS modelling, as the best data fits were always obtained whenconsidering irreversible sorption for all studied soils in the modelling.In addition to irreversible sorption, considering rate-limited sorptionwith two sorption sites of variable affinity for SMX was required toproperly describe SMX transport These results indicate that the trans-port of SMX involves different chemical non-equilibrium and time-de-pendent sorption processes, which are related to soil properties andSMX pH-dependent speciation.

At the watershed scale, in soils located upstream (Regosols) SMXpresented low mobility as the consequence of its higher retention, en-hanced by higher soil OC contents and acidic pH. Nevertheless, animportant SMX fraction can potentially migrate to deep layers due tothe presence of coarse material in the lower horizons and to the highsmectite content, know to favour cracks formation in dry season andpreferential water flow paths. In soils located downstream (Cambisols)SMX showed high mobility in relation with a low retention capacity ofthese soils due to both their texture and OC content. In these soils, thepresence of clayey layers subject the formation of cracks in dry season,as well as the shallow groundwater levels enhance the risks ofgroundwater contamination by SMX. Our results also suggest that SMXcomplexation with metal ions (such as Cu+) also plays an importantrole in SMX retention in soil and consequently on its mobility.

Altogether, these results suggest that SMX can be classified as amoderately to highly mobile compound in the Bolivian Altiplano con-text, in agreement with its behaviour in other regions of the world. Inthe case of Altiplano, this important SMX mobility potential is clearlyunder the control of soil's texture, clay type and content, pH and OCcontent. Nevertheless, hydrological modelling at the watershed scale isrequired to properly estimate the pollution risk of surface waters andgroundwater. The important role of all these factors in SMX reactivetransport makes its prediction quite complex although the modellingwork conducted in the present study gave good results.

Declaration of Competing Interest

The authors declare that they have no known competing financialinterests or personal relationships that could have appeared to influ-ence the work reported in this paper.

Acknowledgements

This work was supported by the French national program EC2CO“Ecosphère Continentale et Côtière” (CNRS, France), the LABEXOSUG@2020 (ANR-10-LABX-56, ANR, France) and the NationalCouncil for Science and Technology (CONACYT, Mexico). We alsothank IBTEN (Instituto Boliviano de Tecnologia Nuclear, Bolivia) forsoil analysis.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.envint.2019.104905.

D. Archundia, et al. Environment International 130 (2019) 104905

8

References

Archundia, D., 2016. Etude du devenir et de l'impact des antibiotiques à l'échelle d'unbassin versant: application au bassin versant du Katari (Bolivie). Thèse de doctoraten Sciences de la terre et de l'univers, et de l'environnement. Université GrenobleAlpes, Grenoble, France.

Archundia, D., Duwig, C., Lehembre, F., Chiron, S., Morel, M.-C., Prado, B., Bourdat-Deschamps, M., Vince, E., Aviles, G.F., Martins, J.M.F., 2017a. Antibiotic pollution inthe Katari subcatchment of the Titicaca Lake: major transformation products andoccurrence of resistance genes. Sci. Total Environ. 576, 671–682. https://doi.org/10.1016/j.scitotenv.2016.10.129.

Archundia, D., Duwig, C., Spadini, L., Uzu, G., Guédron, S., Morel, M.C., Cortez, R.,Ramos Ramos, O., Chincheros, J., Martins, J.M.F., 2017b. How uncontrolled urbanexpansion increases the contamination of the Titicaca Lake Basin (El Alto, La Paz,Bolivia). Water Air Soil Pollut. 228. https://doi.org/10.1007/s11270-016-3217-0.

Boreen, A.L., Arnold, W.A., McNeill, K., 2004. Photochemical fate of sulfa drugs in theaquatic environment: sulfa drugs containing five-membered heterocyclic groups.Environ. Sci. Technol. 38, 3933–3940. https://doi.org/10.1021/es0353053.

Boxall, A., Blackwell, P., Cavallo, R., Kay, P., Tolls, J., 2002. The sorption and transport ofa sulphonamide antibiotic in soil systems. Toxicol. Lett. 131, 19–28.

Calvet, R., 1989. Adsorption of organic Chemicals in Soils. Environ. Health Perspect. 83,145–177.

Carrillo, M., Braun, G.C., Siebe, C., Amelung, W., Siemens, J., 2016. Desorption of sul-famethoxazole and ciprofloxacin from long-term wastewater-irrigated soils of theMezquital Valley as affected by water quality. J. Soils Sediments 16, 966–975.https://doi.org/10.1007/s11368-015-1292-2.

Chefetz, B., Mualem, T., Ben-Ari, J., 2008. Sorption and mobility of pharmaceuticalcompounds in soil irrigated with reclaimed wastewater. Chemosphere 73,1335–1343. https://doi.org/10.1016/j.chemosphere.2008.06.070.

Dıaz-Cruz, M.S., López de Alda, M.J., Barceló, D., 2003. Environmental behavior andanalysis of veterinary and human drugs in soils, sediments and sludge. TrAC TrendsAnal. Chem. 22, 340–351.

Drillia, P., Stamatelatou, K., Lyberatos, G., 2005. Fate and mobility of pharmaceuticals insolid matrices. Chemosphere 60, 1034–1044. https://doi.org/10.1016/j.chemosphere.2005.01.032.

Engelhardt, I., Sittig, S., Šimůnek, J., Groeneweg, J., Pütz, T., Vereecken, H., 2015. Fate ofthe antibiotic sulfadiazine in natural soils: experimental and numerical investiga-tions. J. Contam. Hydrol. 177–178, 30–42. https://doi.org/10.1016/j.jconhyd.2015.02.006.

FAO, 1998. Land and Water Digital Media Series n5; Soil and Terrain Database for LatinAmerica and the Carribean - 1:5 Million Scale (SOTERLAC). Food and AgricultureOrganization of the United Nations, Rome, Italy.

FAO, 2006. Guidelines for soil description, 4th ed. Food and Agriculture Organization ofthe United Nations, Rome.

Feinberg, M., Laurentie, M., 2010. Le cahier des techniques de l'Inra. Validation desméthodes d'analyse quantitative par le profil d'exactitude.

Fortin, J., Flury, M., Jury, W.A., Streck, T., 1997. Rate-limited sorption of simazine insaturated soil columns. J. Contam. Hydrol. 25, 219–234.

García-Galán, M.J., Silvia Díaz-Cruz, M., Barceló, D., 2008. Identification and determi-nation of metabolites and degradation products of sulfonamide antibiotics. TrACTrends Anal. Chem. 27, 1008–1022.

García-Galán, M.J., Garrido, T., Fraile, J., Ginebreda, A., Díaz-Cruz, M.S., Barceló, D.,2010. Simultaneous occurrence of nitrates and sulfonamide antibiotics in two groundwater bodies of Catalonia (Spain). J. Hydrol. 383, 93–101. https://doi.org/10.1016/j.jhydrol.2009.06.042.

García-Santiago, X., Garrido, J.M., Lema, J.M., Franco-Uría, A., 2017. Fate of pharma-ceuticals in soil after application of STPs products: influence of physicochemicalproperties and modelling approach. Chemosphere 182, 406–415. https://doi.org/10.1016/j.chemosphere.2017.05.049.

Gardner, W.R., 1958. Some steady-state solutions of unsaturated moisture flow equationswith application to evaporation from a water table. Soil Sci. 85, 228–232.

Gvirtzman, H., Gorelick, S.M., 1991. Dispersion and advection in unsaturated porousmedia enhanced by anion exclusion. Nature 352, 793–795. https://doi.org/10.1038/35279.

Hill, D.N., Popova, I.E., Hammel, J.E., Morra, M.J., 2018). Transport of potential manurehormone and pharmaceutical contaminants through intact soil columns. J Environ.Qual. 48, 47–56. https://doi.org/10.2134/jeq2018.06.0233.

Hoa, P.T.P., Managaki, S., Nakada, N., Takada, H., Shimizu, A., Anh, D.H., Viet, P.H.,Suzuki, S., 2011. Antibiotic contamination and occurrence of antibiotic-resistantbacteria in aquatic environments of northern Vietnam. Sci. Total Environ. 409,2894–2901. https://doi.org/10.1016/j.scitotenv.2011.04.030.

Hu, X., Zhou, Q., Luo, Y., 2010. Occurrence and source analysis of typical veterinaryantibiotics in manure, soil, vegetables and groundwater from organic vegetablebases, northern China. Environ. Pollut. 158, 2992–2998.

James, R.V., Rubin, J., 2010. Transport of Chloride Ion in a Water-unsaturated SoilExhibiting Anion Exclusion1. Soil Sci. Soc. Am. J. 50, 1142–1149. https://doi.org/10.2136/sssaj1986.03615995005000050010x.

Johnston, P.B., Atkinson, T.C., Odling, N.E., Barker, J.A., 2005. Models of tracer break-through and permeability in simple fractured porous media. Geol. Soc. Lond. Spec.Publ. 249, 91–102.

Kan, A.T., Chen, W., Tomson, M.B., 2000. Desorption kinetics of neutral hydrophobicorganic compounds from field-contaminated sediment. Environ. Pollut. 108, 81–89.

Keesstra, S., Geissen, V., Mosse, K., Piiranen, S., Scudiero, E., Leistra, M., van Schaik, L.,2012. Soil as a filter for groundwater quality. Curr. Opin. Environ. Sustain. 4,507–516. https://doi.org/10.1016/j.cosust.2012.10.007.

Kim, K.R., Owens, G., Kwon, S.I., So, K.H., Lee, D.B., Ok, Y.S., 2011. Occurrence andenvironmental fate of veterinary antibiotics in the terrestrial environment. Water AirSoil Pollut. 214, 163–174.

Kümmerer, K., 2009. Antibiotics in the aquatic environment – a review – part I.Chemosphere 75, 417–434. https://doi.org/10.1016/j.chemosphere.2008.11.086.

Kurwadkar, S.T., Adams, C.D., Meyer, M.T., Kolpin, D.W., 2011. Comparative mobility ofsulfonamides and bromide tracer in three soils. J. Environ. Manag. 92, 1874–1881.https://doi.org/10.1016/j.jenvman.2011.03.018.

Kutilek, M., Nielsen, D.R., 1994. Soil Hydrology. Geoscience Publisher, Catena Verlag,Cremlingen.

Leenheer, J.A., Ahlrichs, J.L., 1971. A kinetic and equilibrium study of the adsorption ofcarbaryl and parathion upon soil organic matter surfaces. Soil Sci. Soc. Am. Proc. 35,700–705.

Leung, H.W., Minh, T.B., Murphy, M.B., Lam, J.C.W., So, M.K., Martin, M., Lam, P.K.S.,Richardson, B.J., 2012. Distribution, fate and risk assessment of antibiotics in sewagetreatment plants in Hong Kong, South China. Environ. Int. 42, 1–9. https://doi.org/10.1016/j.envint.2011.03.004.

Loague, K., Green, R.E., 1991. Statistical and graphical methods for evaluating solutetransport models: overview and application. J. Contam. Hydrol. 7, 51–73.

Martínez-Hernández, V., Meffe, R., Kohfahl, C., de Bustamante, I., 2017. Investigatingnatural attenuation of pharmaceuticals through unsaturated column tests.Chemosphere 177, 292–302. https://doi.org/10.1016/j.chemosphere.2017.03.021.

Martins, J.M., Mermoud, A., 1998. Sorption and degradation of four nitroaromatic her-bicides in mono and multi-solute saturated/unsaturated soil batch systems. J.Contam. Hydrol. 33, 187–210.

Martins, J.M.F., Majdalani, S., Vitorge, E., Desaunay, A., Navel, A., Guiné, V., Daïan, J.F.,Vince, E., Denis, H., Gaudet, J.P., 2013. Role of macropore flow in the transport ofEscherichia coli cells in undisturbed cores of a brown leached soil. Env. Sci Process.Impacts 15, 347–356. https://doi.org/10.1039/C2EM30586K.

Mehra, O.P., Jackson, M.L., 1960. Iron Oxide Removal from Soils and Clays by aDithionite-Citrate System Buffered with Sodium Bicarbonate.

Melamed, R., Jurinak, J.J., Dudley, L.M., 1994. Anion exclusion–pore water velocity in-teractions affecting transport of bromine through an Oxisol. Soil Sci. Soc. Am. J. 58,1405–1410. https://doi.org/10.2136/sssaj1994.03615995005800050018x.

Michael, I., Rizzo, L., McArdell, C.S., Manaia, C.M., Merlin, C., Schwartz, T., Dagot, C.,Fatta-Kassinos, D., 2013. Urban wastewater treatment plants as hotspots for the re-lease of antibiotics in the environment: a review. Water Res. 47, 957–995. https://doi.org/10.1016/j.watres.2012.11.027.

Mojica, E.-R., Aga, D., 2011. Antibiotics Pollution in Soil and Water: Potential Ecologicaland Human Health Issues. Elsevier.

Morel, M.-C., Spadini, L., Brimo, K., Martins, J.M.F., 2014. Speciation study in thesulfamethoxazole–copper–pH–soil system: implications for retention prediction. Sci.Total Environ. 481, 266–273. https://doi.org/10.1016/j.scitotenv.2014.02.040.

Müller, K., Duwig, C., Prado, B., Siebe, C., Hidalgo, C., Etchevers, J., 2012. Impact of long-term wastewater irrigation on sorption and transport of atrazine in Mexican agri-cultural. J. Environ. Sci. Health B 47, 30–41.

OECD, 2003. OECD Guidelines for the Testing of Chemicals: Leaching in Soil Columns.Oppel, J., Broll, G., Loffler, D., Meller, M., Rombke, J., Ternes, T., 2004. Leaching be-

haviour of pharmaceuticals in soil-testing-systems: a part of an environmental riskassessment for groundwater protection. Sci. Total Environ. 328, 265–273. https://doi.org/10.1016/j.scitotenv.2004.02.004.

Park, J.Y., Huwe, B., 2016. Effect of pH and soil structure on transport of sulfonamideantibiotics in agricultural soils. Environ. Pollut. 213, 561–570. https://doi.org/10.1016/j.envpol.2016.01.089.

Perfect, Sukop, M.C., Haszler, G.R., 2002. Prediction of dispersivity for undisturbed soilcolumns from water retention parameters. Environ. Pollut. 213, 561–570.

Pignatello, J.J., 1989. Sorption Dynamics of Organic Compounds in Soils and Sediments.SSSA Special Publication, vol. 22. Soil Sci. Soc. Am. Inc, Madison WI, pp. 45–80.

Pignatello, J.J., 2000. The measurement and interpretation of sorption and desorptionrates for organic compounds in soil media. Adv. Agron. 69, 1–73.

Rühlmann, J., Körschens, M., Graefe, J., 2006. A new approach to calculate the particledensity of soils considering properties of the soil organic matter and the mineralmatrix. Geoderma 130, 272–283. https://doi.org/10.1016/j.geoderma.2005.01.024.

Sarmah, A.K., Meyer, M.T., Boxall, A.B.A., 2006. A global perspective on the use, sales,exposure pathways, occurrence, fate and effects of veterinary antibiotics (VAs) in theenvironment. Chemosphere 65 (5), 725–759. https://doi.org/10.1016/j.chemosphere.2006.03.026.

Simunek, J., Sejna, M., Saito, H., Genuchten, M.Th.V., 2008. The HYDRUS-1D SoftwarePackage for Simulating the One-Dimensional Movement of Water, Heat, and MultipleSolutes in Variably-Saturated Media. Dep. Environ. Sci. Univ. Calif, RiversideRiverside Calif.

Simunek, J., Jacques, D., Langergraber, G., Bradford, S.A., Šejna, M., van Genuchten,M.T., 2013. Numerical modeling of contaminant transport using HYDRUS and itsspecialized modules. J. Indian Inst. Sci. 93, 265–284.

Srinivasan, P., Sarmah, A.K., 2014a. Dissipation of sulfamethoxazole in pasture soils asaffected by soil and environmental factors. Sci. Total Environ. 479–480, 284–291.https://doi.org/10.1016/j.scitotenv.2014.02.014.

Srinivasan, P., Sarmah, A.K., 2014b. Assessing the sorption and leaching behaviour ofthree sulfonamides in pasture soils through batch and column studies. Sci. TotalEnviron. 493, 535–543. https://doi.org/10.1016/j.scitotenv.2014.06.034.

Srinivasan, P., Sarmah, A.K., Manley-Harris, M., 2013. Co-contaminants and factors af-fecting the sorption behaviour of two sulfonamides in pasture soils. Environ. Pollut.180, 165–172. https://doi.org/10.1016/j.envpol.2013.05.022.

Srinivasan, P., Sarmah, A.K., Manley-Harris, M., 2014. Sorption of selected veterinaryantibiotics onto dairy farming soils of contrasting nature. Sci. Total Environ. 472,695–703. https://doi.org/10.1016/j.scitotenv.2013.11.104.

D. Archundia, et al. Environment International 130 (2019) 104905

9

Stumm, W., 1992. Chemistry of the Solid-Water Interface: Processes at the Mineral-Waterand Particle-Water Interface in Natural Systems. Wiley.

Tamtam, F., Mercier, F., Le Bot, B., Eurin, J., Tuc Dinh, Q., Clément, M., Chevreuil, M.,2008. Occurrence and fate of antibiotics in the Seine River in various hydrologicalconditions. Sci. Total Environ. 393, 84–95. https://doi.org/10.1016/j.scitotenv.2007.12.009.

Tamtam, F., van Oort, F., Le Bot, B., Dinh, T., Mompelat, S., Chevreuil, M., Lamy, I., Thiry,M., 2011. Assessing the fate of antibiotic contaminants in metal contaminated soilsfour years after cessation of long-term waste water irrigation. Sci. Total Environ. 409,540–547. https://doi.org/10.1016/j.scitotenv.2010.10.033.

Thiele-Bruhn, S., Aust, M.O., 2004. Effects of pig slurry on the sorption of sulfonamideantibiotics in soil. Environ. Contam. Toxicol. 47.

Wehrhan, A., Kasteel, R., Simunek, J., Groeneweg, J., Vereecken, H., 2007. Transport ofsulfadiazine in soil columns — experiments and modelling approaches. J. Contam.Hydrol. 89, 107–135. https://doi.org/10.1016/j.jconhyd.2006.08.002.

Wooding, R., 1968. Steady infiltration from a shallow circular pond. Water Resour. Res. 4,1259–1273.

Zhang, H., Zhou, Y., Huang, Y., Wu, L., Liu, X., Luo, Y., 2016. Residues and risks ofveterinary antibiotics in protected vegetable soils following application of differentmanures. Chemosphere 152, 229–237. https://doi.org/10.1016/j.chemosphere.2016.02.111.

Zuccato, E., Castiglioni, S., Bagnati, R., Melis, M., Fanelli, R., 2010. Source, occurrenceand fate of antibiotics in the Italian aquatic environment. J. Hazard. Mater. 179,1042–1048. https://doi.org/10.1016/j.jhazmat.2010.03.110.

D. Archundia, et al. Environment International 130 (2019) 104905

10