JEQ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 5 January 2006
Published in J Environ Qual 35:253-267 (2006)
DOI: 10.2134/jeq2005.0059
© 2006 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (2)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Saffih-Hdadi, K.
Right arrow Articles by Barriuso, E.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Saffih-Hdadi, K.
Right arrow Articles by Barriuso, E.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Saffih-Hdadi, K.
Right arrow Articles by Barriuso, E.
Related Collections
Right arrow Field-Scale Studies
Right arrow Laboratory Column Studies
Right arrow Pesticides
Right arrow Coupled Flow/Transport Models
Right arrow Ecological Risk Assessment

TECHNICAL REPORTS

Organic Compounds in the Environment

A Model for Linking the Effects of Parathion in Soil to its Degradation and Bioavailability Kinetics

K. Saffih-Hdadia, L. Brucklerb,*, F. Lafoliea and E. Barriusoc

a Institut National de la Recherche Agronomique (INRA), CSE Bat. Sol, Site Agroparc, 84914 Avignon cedex 09, France
b Département "Environnement et Agronomie", Institut National de la Recherche Agronomique (INRA), Bat. CSE-Sol, Site Agroparc, 84914 Avignon cedex 09, France
c Institut National Agronomique de Paris-Grignon, EGC-sol, BP: 01, 78850 Thiverval-Grignon, France

* Corresponding author (Laurent.Bruckler{at}avignon.inra.fr)

Received for publication February 17, 2005.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 BASIC MODEL AND EQUATIONS
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Parathion is an insecticide of a group of highly toxic organophosphorus compounds. To investigate the dissipation and toxicological impact of parathion [O,O-diethyl O-(4-nitrophenyl) phosphorothioate] and its highly toxic metabolite, paraoxon, soil laboratory experiments were conducted in columns during a 19-d experiment under variably saturated conditions. Water and pesticide transport, sorption, and biodegradation of parathion were measured in three soil pools (soluble phase, weakly and strongly sorbed phases) using C-labeled pesticide. The effects of parathion and its metabolite on the mobility of soil nematodes were observed and then modeled with an effective variable, which combined pesticide concentration and time of application. Results showed that parathion was highly sorbed and slowly degraded to a mixture of metabolites. The parent compound and its metabolites remained located in the top 0.06-m soil layer. A kinetic model describing the sorption, biodegradation, and allocation into different soil pools of parathion and its metabolites was coupled with heat and water transport equations to predict the fate of parathion in soil. Simulated results were in agreement with experimental data, showing that the products remained in the upper soil layers even in the case of long-term (11-mo) simulation. The strongly sorbed fraction may be regarded as a pesticide reservoir that regularly provides pesticide to the weakly sorbed phase, and then, liquid phase, respectively. From both modeling and observations, no major toxicological damage of parathion and paraoxon to soil nematodes was found, although some effects on nematodes were possible, but at the soil surface only (0.01- and 0.02-m depth).

Abbreviations: EC50, effective concentration that inhibits the mobility of 50% nematodes • HPLC, high performance liquid chromatography • PASTIS, Prediction of Agricultural Solute Transfer in Soil


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
BASIC MODEL AND EQUATIONS
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
PERSISTENCE OF PESTICIDES in soil and its consequence for soil organisms is a worldwide subject of concern. Among the great diversity of organic pesticides used in crop protection, organophosphates are a group of highly toxic compounds that are used extensively as agricultural and domestic pesticides (Costa, 1988). The toxicity of the organophosphate insecticide parathion to nontarget organisms has been the subject of extensive research (Chang et al., 1997; Saffih-Hdadi et al., 2005). Under aerobic conditions, parathion is activated by oxidative desulfuration (Costa, 1988) to the oxygen analog paraoxon, which has a potent anticholinesterase effect (Eto, 1974) and is more toxic than the parent chemical (Guilhermino et al., 1996). In soils, nematodes are a relevant target of the toxicological impacts of parathion and paraoxon, because they function at various trophic levels in agroecosystems. Herbivorous species directly and indirectly affect plant growth and plant yield, while dwelling in the rhizosphere and puncturing roots (Freckman and Caswell, 1985; Yeates et al., 1993). Carnivorous species prey on other invertebrates including other nematodes (Hara and Kaya, 1983; Zimmerman and Cranshaw, 1990), and microbivores feed on bacteria and fungi and thus affect mineralization of organic matter (Ingham et al., 1985), and regulate its decomposition rate (Parmelee and Alston, 1986) and/or nitrogen dynamics (Didden et al., 1994).

There is a growing body of literature on the development of simulation models for predicting transport and fate of chemicals in the root zone of agricultural crops. The models show a large range in complexities, ranging from the most simple educational models (Nofziger and Hornsby, 1986) to the more or less sophisticated physically based models like PRZM (Carsel et al., 1984), LEACHM (Wagenet and Hutson, 1989; Wagenet and Rao, 1990), Pestfade (Clemente et al., 1993), and Agriflux (Larocque et al., 1998). To link pesticide transport with the distribution of the pesticide and/or its metabolites among various pools in soil (i.e., solid, liquid, gas), sorption has been treated as a rapid equilibrium, single-valued, reversible process (Weber et al., 1992), but in other studies, kinetic models have been developed (Wu and Gschwend, 1986; Brusseau et al., 1991; Ma and Selim, 1994; Shelton and Doherty, 1997). These models describe sorption in terms of a "two-site" model characterized by fast and slow binding sites. There are also some models that combine equilibrium with kinetic hypotheses (Xue and Selim, 1995; Xue et al., 1997). Among these approaches, dynamic models including a "two-site" representation of the sorption process are probably the most realistic when compared with independent experimental data describing the dynamics of pesticides in soil (Wu and Gschwend, 1986; Brusseau et al., 1991; Ma and Selim, 1994; Xue and Selim, 1995; Ma et al., 1996; Shelton and Doherty, 1997; Xue et al., 1997).

Many published papers describing the degradation of pesticides include models of biodegradation (Soulas, 1997) based on equations of chemical kinetics where only pesticide concentration limits degradation (Zimdhal et al., 1970) or taking into account the dynamics of microbial populations (Simkins and Alexander, 1984; Schmidt et al., 1985; Soulas and Lagacherie, 1990; Shelton and Doherty, 1997). However, none of the models found in the literature simultaneously take the dissipation of the parent compound in the soil into account and the fate and partitioning of its metabolites. Moreover, pesticide measurements in soluble and differently adsorbed pesticide soil fractions are not always available and prevent complete model verifications. Consequently, to analyze the dissipation of parathion in soil and its environmental effect on nematodes, we propose to use a sorption and decay sub-model that is an extended two-site model having sites arranged in series, with relatively fast kinetic sorption to Type-1 sites and slower sorption to Type-2 sites for both parent compound and metabolite, and with hysteresis of adsorption and desorption. In this model, unlike many other studies, variably saturated conditions will be employed, which require the simulation of variably saturated flow and the use of saturation-dependent sorption coefficients.

Thus, the present work has two main objectives: (i) to predict the fate of parathion and its metabolite in soil over short and/or long temporal scales by using a physically based model describing simultaneously transport, sorption, and biodegradation of parathion in soil (Saffih-Hdadi et al., 2003) and (ii) to predict in a more general way, and under various soil conditions, the toxicological effects of sorption, biodegradation, and transport of parathion and its metabolite in soil to nematodes.


   BASIC MODEL AND EQUATIONS
TOP
 ABSTRACT
 INTRODUCTION
 BASIC MODEL AND EQUATIONS
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The one-dimensional vertical mechanistic model PASTIS (Prediction of Agricultural Solute Transformations in Soils) described by Lafolie (1991) and Garnier et al. (2001) was adapted for predicting transport, sorption, and biodegradation of pesticides in soils. Basically, the model is built around water transport (Darcy's law for both saturated and unsaturated conditions), heat transport [conduction flux induced by temperature gradients and convective flux provided by the water flow itself (Chu et al., 1983)], and solute transport (convection–dispersion equation). We assumed that this physically based approach was necessary to accurately predict pesticide transport in soil. All the soil water was assumed to be mobile, and all the sources of dispersion were lumped into one parameter, the dispersion coefficient D({theta},{nu}), which is related to the pore water velocity {nu} and the effective molecular diffusion coefficient Dm({theta}) (m2 s–1) as follows:

[1]
where {lambda}1 (m) is the dispersivity coefficient for longitudinal dispersion (Bresler et al., 1982). The effective molecular diffusion coefficient is often very small as compared to {lambda}1|{nu}| and hence neglected in Eq. [1] (Feddes et al., 1974).

Sorption
A set of equations (i.e., Eq. [2]GoGoGoGo–[7] in Table 1 and Fig. 1 ) describing sink-source terms in the convection–dispersion equation, which account for sorption and biodegradation, was incorporated into the PASTIS model. Briefly, three phases are taken into account: (i) the "soluble phase" S1, (ii) the "weakly sorbed pesticide phase" S2 in which the pesticide is rapidly and weakly sorbed to the soil, and (iii) the "bound residue phase" S3 in which the pesticide is more slowly but more strongly sorbed to the soil as described in Saffih-Hdadi et al. (2003). The pesticide can move from one phase to another (Fig. 1) and pesticide allocation between the soil phases is possible through forward kinetic coefficients (k1 [s–1] for sorption between the "soluble phase" and the "weakly sorbed phase," and k3 [s–1] for sorption between the "weakly sorbed phase" and the "bound residue phase," respectively) and backward coefficients (k2 [s–1] and k4 [s–1], respectively) (Table 1). The transport from the "soluble phase" to the "weakly sorbed phase" corresponds to rapid sorption of the pesticide to mineral and organic solid particles, whereas the transport from the "weakly sorbed phase" to the "bound residue phase" describes the diffusion processes of the pesticide through the soil microporosity, combined with progressive sorption with organic and mineral compounds in the soil during the diffusion. Moreover, we assumed that the sorption rate was diffusion limited, and diffusion proceeded more slowly with decreasing water content or temperature (Lafolie, 1991). Consequently, for each ki (s–1) sorption or desorption coefficient the following temperature and water content relationship was chosen:

[8]
where ki is a sorption or desorption coefficient at absolute temperature T (K) and volumetric water content {theta} (m3 m–3) and kiref is the corresponding sorption or desorption coefficient at the reference absolute temperature Tref (301.2 K) and reference volumetric water content {theta}ref (m3 m–3, the soil water content at a water potential of –1 m H2O). This allowed taking into account the possible decreasing diffusion in the liquid phase and/or in the soil microporosity when the soil temperature and/or water content decreased.


View this table:
[in this window]
[in a new window]
  		Table 1. Basic equations of the model of sorption and biodegradation.

 


View larger version (11K):
[in this window]
[in a new window]
  		Fig. 1. Schematic representation of the model (S for the pesticide and M for the metabolite; k1, k2, k3, k4 and km1, km2, km3, km4 are kinetic coefficients of sorption of parathion and paraoxon, respectively).

 
Biodegradation and Metabolite Production
The model assumes that biodegradation of the pesticide occurs in the "soluble phase" only (Ogram et al., 1985) through a co-metabolism mechanism, as it was shown in a previous paper in the case of parathion and its metabolite paraoxon (Saffih-Hdadi et al., 2003). The total biomass B(t) of the soil, which dissipates the pesticide present in the soluble phase, is involved in the same way as for other various organic compounds in the soil. The rate of parathion biodegradation depends on the concentration of pesticide in the soluble phase, on the ratio B(t)/B0, where B0 is the initial biomass and B(t) is the biomass at time (t), and on a coefficient {alpha} (s–1). Biodegradation may be complete, thus directly leading to CO2 production through a total mineralization of the pesticide, or incomplete, thus leading to the production of a metabolite. Metabolite production efficiency resulting from parathion biodegradation is given by a parameter fpc (between 0 and 1) that controls the rate of paraoxon apparition. When a metabolite is produced in the liquid phase (here, paraoxon from parathion), convective and diffusive transport, sorption, and biodegradation pathways of the metabolite follow the same general pattern as for the pesticide (Fig. 1 and Table 1).

Metabolite allocation between the soil phases is possible through forward kinetic coefficients (km1 [s–1] for sorption between the "soluble phase" M1, and the "weakly sorbed phase" M2, and km3 [s–1] for sorption between the "weakly sorbed phase" and the "bound residue phase" M3) and backward kinetic coefficients (km2 [s–1] and km4 [s–1], respectively). Both forward and backward sorption coefficients of the metabolite depend on the soil temperature and water content as described before for the parent compound. Similarly, degradation by the microbial biomass is taken into account, but as only one metabolite is taken into account here, biodegradation of the metabolite, if it occurs, is assumed as complete biodegradation leading to CO2 production. The biodegradation rate depends on a coefficient {alpha}m (s–1), which is the analog of the {alpha} (s–1) coefficient of the parent compound. The dependence of the biodegradation coefficients {alpha} (s–1) and {alpha}m (s–1) on temperature and water potential is described in the same way as the microbial activity dependence on temperature and water potential (Garnier et al., 2001). Factors fT and fw describing the temperature and water potential dependence of the biodegradation coefficients are given as follows:

[9]

[10]

[11]

[12]
where ßT is a constant (0.115 K–1), {psi}ref is equal to –1.0 m H2O, and {psi}' (–758 m H2O) is the water potential at which microbial activity ceases (Garnier et al., 2001).

Although a simple first-order decay could be used for the prediction of the time evolution of the total biomass B(t), this evolution is calculated in the model PASTIS according to the scheme described in Garnier et al. (2003). Briefly, the model PASTIS contains the sub-model CANTIS (Carbon and Nitrogen Transformation in Soil), which simulates the transformations of carbon and nitrogen. It considers the decomposition of organic matter, mineralization, immobilization, nitrification, and humification, and consequently, models the biomass activity.

Numerical Solution
Finite-difference approximations were derived for spatial derivatives equations. Particular care was given to the approximation of convective terms. It is known that a fully upwind finite-difference approximation of the convective terms smoothes out spurious oscillations, but introduces artificial spreading of solute fronts. A procedure was developed to control the amount of artificial spreading necessary to avoid numerical oscillations. This procedure is based on a weighted upwind approximation of first-order terms and a fully implicit time-step marching scheme is used. The solution of equations governing biodegradation and pesticide exchange between compartments (Table 1) was performed with the time increment determined from the solution of Richards equation. Solution of those equations provided the sink-source terms used in the transport equation to account for pesticide exchange between the liquid and solid phases. In our case, because only co-metabolism was considered (Saffih-Hdadi et al., 2003), equations of Table 1 were linear with constant coefficients, and an explicit numerical scheme was chosen. The program is written in standard Fortran 77 and runs on a minicomputer (Sun, Unix) in few minutes to simulate a several-day experiment.

Model Parameterization
The main parameters used in the PASTIS model are listed in Table 2. Hydrodynamic parameter estimates for the soil column come from laboratory measurements of hydrodynamic properties. Realistic values of dispersivity and molecular diffusivity of 3.0 x 10–3 m and 5.5 x 10–8 m2 s–1, respectively, were assigned for solute transport. Initial biomass was based on the density of bacteria measured in the soil (400 mg kg–1). Basically, the kinetics coefficients for sorption and desorption came from estimations obtained in batch experiments with the same soil (Saffih-Hdadi et al., 2003). However, some of them (k1, k3, km1, km3, {alpha}, and {alpha}m) were decreased by manual calibration to take into account the specific experimental conditions using soil columns instead of batch experiments. Indeed, the soil columns, which are made of soil aggregates, probably present a lower affinity for the sorption and mineralization processes as compared with the soil–solution mixtures where shaking increases the contact and exchanges between the solid and liquid phases (Brusseau et al., 1991; Xue and Selim, 1995; Ma et al., 1996). Consequently, final values for the parameters k1, k3, km1, km3, {alpha}, and {alpha}m were 6.7 x 10–3, 6.0 x 10–7, 1.3 x 10–3, 1.5 x 10–6, 1.1 x 10–5, and 4.1 x 10–5 s–1 instead of 7.4 x 10–3, 8.5 x 10–7, 1.4 x 10–3, 2.1 x 10–6, 2.2 x 10–5, and 8.3 x 10–5 s–1, respectively (Saffih-Hdadi et al., 2003).


View this table:
[in this window]
[in a new window]
  		Table 2. Value and origin of the main parameters used in the PASTIS (Prediction of Agricultural Solute Transfer in Soil) model. Except where noted, the source is Saffih-Hdadi et al. (2003).

 
Boundary Conditions
Simulations for water, heat, and pesticides were applied to two contrasting situations: (i) simulations were compared with a 19-d laboratory experiment with soil columns and (ii) the model was used to simulate and extrapolate (without experimental validation for the pesticide dissipation) the transport and dynamics of the pesticide and its metabolite in a long-term simulation (11 mo). In the latter case, a field database where water and heat transport have been previously measured and simulated with the PASTIS model was used (Cannavo et al., 2004), assuming that the sorption and biodegradation parameters relative to the pesticide were similar to those of the laboratory situation. For both the laboratory and field cases the measured soil water potential and temperature profiles were used for the initial conditions, and an initial zero-concentration for pesticide was used. For water flow, two types of upper boundary conditions accounting for infiltration or evaporation were used for the long-term simulation (11 mo), a Neuman-type that imposes the surface flux, or a Dirichlet-type condition that provides pressure boundary conditions. For the 19-d laboratory experiment, a Neuman-type upper boundary condition was used. At the bottom of the soil profile, the model handles boundary conditions corresponding to no-flux conditions (impervious layer) for the 19-d laboratory experiment, and boundary conditions corresponding to a time-changing water potential for the long-term simulation (11 mo). For soil heat conduction, a constant temperature was imposed at the top and bottom of the soil in the laboratory experiment because the column was placed in laboratory conditions with roughly constant air temperature. For the field simulation, the soil surface temperature derived from a simplified energy budget was imposed (Cannavo et al., 2004), whereas a zero-temperature gradient was imposed at a 2-m depth. A flux condition was imposed on the pesticide applied at the soil surface.

Predicting Toxicological Impact on Nematodes
The PASTIS model can predict the dynamics of parathion and paraoxon in soil, and consequently, the concentrations of the parent compound and the metabolite in the liquid phase are available at each time and depth. Saffih-Hdadi et al. (2005) showed that the variable Ci x dt (where Ci [µg mL–1] is the concentration of the pesticide in the liquid phase at time ti [s], and dt [s] the time of application) was an effective variable for estimating toxicological impact. If the concentration Ci in the liquid phase is not constant in time (general case), then the variable {int}t0tnCi x dt must be considered. Here, t0 and tn are the initial and final times of the period of contact between the nematodes and the soil solution. This variable is relevant especially when the pesticide is bound through covalent bindings to the cells of the biological organisms, which is the case for paraoxon (Saffih-Hdadi et al., 2005). The variable {int}t0tnCi x dt is commonly divided by the term (tn – t0) to obtain a "time weighted average concentration" (Saffih-Hdadi et al., 2005).


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 BASIC MODEL AND EQUATIONS
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Chemical and Soil Properties
Carbon-14-U-labeled parathion (specific activity: 930 MBq mmol–1; radiopurity > 98%) was supplied by Dislab (Saulx Les Chartreux, France). Unlabeled parathion (analytical standards) was supplied by Sigma-Aldrich (Lyon, France). Parathion has a low vapor pressure (5 mPa at 20°C) and very low water solubility (11 mg L–1 at 20°C).

The soil (Typic Eutrudepts) was sampled in the surface layer (0–20 cm) of a bare experimental plot located at Caumont, France. Soil was air-dried and passed through a 5-mm sieve. Soil had a pH in water of 8.3, a water field capacity of 18% (w/w), and a textural distribution of 34.2% clay, 46.5% silt, and 19.3% sand, with 351 g kg–1 dry soil of lime (tot) and 15.5 g kg–1 dry soil of organic carbon. The bulk density of the soil columns was derived from soil mass, volume, and water content measurements. The mean bulk density was 1.29 Mg m–3 for the three soil columns.

For the soil water retention curve and unsaturated conductivity measurements, soil cylinders (height 72 mm, internal diameter 150 mm) were used, and were allowed to evaporate in the laboratory. Every 10 min, the sample was weighed using an automatic balance. Water potential profiles were recorded every 10 min by five miniature tensiometers connected to pressure transducers (Model CZ 5022/2; Schlumberger Industries, Vélizy-Villacoublay, France). Hydrodynamic properties were determined by measuring the water potential and estimating the water content in the sample at several depths and times, using the instantaneous profile method (Tamari et al., 1993). This allowed fitting the soil water retention curve and the saturated hydraulic conductivity (Table 2) in the van Genuchten model (van Genuchten, 1980).

Column Experiments
Three inox cylinders (diameter 0.15 m, length 0.21 m) were packed with sieved soil. Before starting the leaching experiments, all soil samples were irrigated with 0.005 M CaCl2 water solution (16 mm h–1 until outflow occurred) to ensure the same initial water content. Three days after and before incubation, soil columns were treated with 1 mg kg–1 dry soil of radio-labeled parathion (dose similar to the usual agricultural practice). The pesticide was pipetted in a fine grid to give a homogeneous spatial distribution. A steady flow of 10 mm h–1, during 2 h, was applied at the beginning of the experiment just after pesticide application using a simulated rainfall device having small needles placed regularly above the soil surface. Soil moisture was measured at four depths (0.01, 0.02, 0.05, and 0.1 m) using time domain reflectometry (TDR). Soil temperatures were measured at two depths (0.02 and 0.15 m). To evaluate water potential, tensiometers were installed horizontally in the soil columns at depths of 0.01, 0.02, 0.05, 0.1, and 0.15 m starting from the top of the columns. The experiments were performed in an air-conditioned chamber at a controlled temperature of 25 ± 1.5°C. The columns were closed at the top to limit evaporation and were closed at the bottom. Vials containing 5 mL of 1 M NaOH were placed in each column to trap CO2. The 0- to 0.10-m soil layer of one of the columns was sliced into 14 layers 5 h after the beginning of the incubation, the second column 5 d after, and the last column was sliced 19 d after, with a second irrigation of 10 mm h–1 for 1 h on the fifth day of incubation. To determine the distribution of parathion and its metabolites in different layers of the columns, soil from each layer was well mixed, three replicates of 10 g of each soil layer were weighed in a glass vial and extracted. A quantity of soil from each layer was kept to measure the soil moisture of the soil at the end of each experimentation.

Analysis
The evolved 14CO2 trapped in NaOH was directly measured by liquid scintillation counting with UltimaGold XR (Packard, Meriden, CT) as liquid scintillate. Extractable residues in soil layers were analyzed after extracting triplicate 10-g soil samples in a 150-mL glass centrifuge bottle with a Teflon cap. This sample was first extracted in 30 mL of 0.01 M CaCl2 water solution ("soluble phase"). After 24 h of shaking and centrifugation (15 min at 8000 x g), the radioactivity in the extract was measured. The soil pellet was then extracted three times in succession with 30 mL of methanol, using the same procedure as for water extraction, and the radioactivity in the methanol extracts was measured ("sorbed phase"). The soil containing the non-extractable 14C residues was air-dried and finely ground, and its radioactivity measured ("strongly sorbed phase") after combustion of triplicate 200-mg aliquots using a Sample Oxidizer 307 (Packard). High performance liquid chromatography (HPLC) analysis of the 14C residues in the water and methanol extracts was performed for each incubation time. The extracts from the three replicates were pooled. The water extracts were concentrated by solid phase extraction with an Env+ cartridge (200 mg; Isolute, Glamorgan, UK), previously activated with methanol and MilliQ water, and eluted with 10 mL of methanol. The methanol extracts were concentrated by evaporation to near dryness under vacuum. The residues were dissolved in 2 mL of the solvent used for the HPLC analysis, and filtered through a Cameo 13N syringe nylon filter (0.45 µm; MSI, Westboro, MA). High performance liquid chromatography analysis was performed using a Waters appliance (600E Multisolvent Delivery System, 717 Autosampler, and a Novapak C18 column of 5 µm and 4.6 x 250 mm; Waters-Millipore, Milford, MA) equipped with a photo diode array detector (Waters 996; Waters-Millipore) coupled on line with a radioactivity continuous flow detector (Packard-Radiomatic Flo-one A550; Packard). The mobile phase was methanol and water (50/50) at 1 mL min–1, and the injected volume was 200 µL. Under these conditions, retention times were 35 min for parathion and 16 min for paraoxon. In the chemical analysis described here, the mineral and organic soil pools were not separated.

Toxicological Experiments
Soil sieved at 5 mm was sampled in the surface layer (0–20 cm). It was not dried and was mixed well for toxicity tests. The soil water content was adjusted to the water content of the soil columns. A water solution with parathion, at a dose similar to that found in the soil columns (at the layer 0–0.005 m and the layer 0.035–0.04 m), was applied to the soil. Nematodes were then extracted and their mobility was scored for 5 h and 5 d after the application of parathion. For all toxicity tests, a control test without pesticide was performed.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 BASIC MODEL AND EQUATIONS
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Water Flow
Measured and simulated water potentials against time are shown for different depths in Fig. 2 . Because the initial condition corresponded to a wet situation and no or limited evaporation occurred at the soil surface, the column remained wet during the 19 d of the experiment, with the water potential and the volumetric water content varying in the range [0; –0.50 m H2O] and [0.40; 0.45 m3 m–3], respectively. Taking into account the total water potential H = {psi} – z, where z is the distance (m) considered to be positive downward and {psi} is the soil water potential, data indicated that the soil column was near equilibrium relative to the total soil water potential. Nevertheless, water potentials exhibited a continuous decrease over time due to the combination of both slow vertical downward gravitational water flow and possible small evaporation of the soil column. The small oscillations in the simulated pressure heads came from the experimental water potential at the bottom of the soil column (0.15 m), which was used as an imposed boundary condition. Following the first and second irrigations (Days 1 and 5 of the experiment), the water potential rapidly increased and then decreased, and a good agreement between measured and simulated water potentials and water contents was observed. The accurate water content simulation was a prerequisite for predicting transport and biotransformation of the pesticide both depending directly or indirectly on the soil water content.



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 2. Measured (symbols) and simulated (continuous lines) water potentials at different depths in a soil column.

 
Overall Sorption and Biodegradation of Parathion in Soil
Table 3 provides the overall partitioning of parathion and its metabolites in the soil expressed as percentage of the initially applied radioactivity. Following the kinetic pattern of parathion breakdown estimated by the 14CO2 evolution, data show that 0, 0.7, and 2.8% of parathion was mineralized 5 h, 5 d, and 19 d after the beginning of incubation, respectively. The remaining products in the soil layers of the columns are shared between the soluble (water-extractable fraction), weakly sorbed (methanol-extractable fraction), and strongly sorbed (non-extractable fraction) pools (Table 3). The soluble fraction decreased as the residence time in the soil increased, from 18.4% of the recovered radioactivity after 5 h of incubation to 9.6 and 3.4% after 5 and 19 d, respectively. This resulted from the combination of biodegradation and adsorption to the mineral and organic particles of the soil. The weakly sorbed fraction rapidly increased just after pesticide input and then decreased (Table 3). This was due to the continuous formation of the strongly sorbed residue fraction. A confirmation of the transfer from the weakly sorbed phase to the bound residue phase is provided by the fact that the sum of the "weakly sorbed plus bound residue phases" was roughly constant in time, that is, 92.9 and 94.3% after 5 and 19 d of incubation, respectively. The mass balance was generally satisfactory (103.3 and 100.5% after 5 and 19 d of incubation, respectively), except at the beginning of the experiment (124.8%). This corresponded to the first soil column for which no reasonable explanation was found. For this reason, results in the next sections will be mostly based on the results obtained in the second and the third column.


View this table:
[in this window]
[in a new window]
  		Table 3. Mass balances for parathion and its metabolites for the three soil columns (quantities are expressed as percentages of initial applied radioactivity).

 
Transport, Partitioning, and Metabolites of Parathion in Soil
Total radioactivity was detected in the top of the soil (0 to 0.06 m), whereas a trace of radioactivity was detected below. The 14C metabolites of parathion were detected in both water and methanol soil sample extracts, and HPLC data indicated that parathion, ethylphosphates, paraoxon, p-nitrophenol, and some unspecified other products were present in both aqueous and methanol extracts (Fig. 3 ).



View larger version (41K):
[in this window]
[in a new window]
  		Fig. 3. Parathion and metabolites as percentage of radioactivity (by layer) in the soluble phase and in the weakly sorbed phase 5 and 19 d after the beginning of incubation in two soil columns in the top 5 cm.

 
In the aqueous extracts (soluble phase, Fig. 3), parathion rapidly disappeared from the surface soil layers, whereas mostly ethylphosphates appeared. The sum of "parathion plus ethylphosphates" was generally dominant (more than 90% of the total radioactivity in aqueous extracts), at least during the first 5 d of incubation. Conversely, total paraoxon and p-nitrophenol were generally lower than 10% of the total radioactivity, except at the end of the experiments (Day 19) where they reached up to 30% (or more) of the total radioactivity in deeper soil layers. This was due to the transformation of paraoxon into p-nitrophenol, which accumulated progressively in time, thus leading to a paraoxon percentage lower than 10% for every depth and time.

In the methanol extracts (weakly sorbed phase) parathion dominated after 5 h of incubation (>95% of the total radioactivity in the methanol extracts), because of its rapid sorption. Ethylphosphates, paraoxon, and p-nitrophenol appeared in small quantities (<5%) a few hours after the beginning of the experiment, thus confirming that production of ethylphosphates, paraoxon, and p-nitrophenol was rapid. For longer times, results were more or less variable, but indicate that parathion sorption in the weakly sorbed phase was generally dominant. Nevertheless, sorption affected both parathion and its metabolites, with sorption of metabolites arising from (i) parathion desorption following irrigation and then production and sorption of metabolites or (ii) transport through the soluble phase of products coming from parathion biotransformation and then followed by sorption.

Modeling and Comparison with Experiments
Figures 4a and 4b show the measured and the simulated total (parathion + metabolites) concentration profiles, for the soluble, weakly sorbed, and strongly sorbed fractions, 5 and 19 d after the application of parathion, respectively. The model predicts correctly that parathion and its metabolites are mostly located in the top 0.06-m soil layer, whereas in deeper layers, parathion and its metabolites are found but at very small levels. No parathion or metabolites are found below 0.1 m, accordingly with both theoretical and experimental results. The model describes well the general shape of the profiles, as well as the order of magnitude of each phase relative to the others. After 5 d of incubation and convective–diffusive transport in the soil column, the soluble phase represents the smallest component, whereas the weakly sorbed phase is the greatest after rapid sorption on the mineral and organic fractions of the soil. After 19 d, the soluble phase remains the smallest pool, whereas the strongly sorbed phase becomes dominant due to continuous accumulation of bound residues into the soil aggregates. The model underestimates the total concentration (parathion + metabolites) in the soluble phase in the 0- to 0.005-m layer, but generally overestimates the concentrations for the weakly and strongly sorbed phases, especially at the front of the pesticide penetration. This can result from both transport processes (molecular diffusivity and/or dispersivity of the different products), and from the combination of biological (biodegradation) and physical (sorption at the liquid–solid interface) factors that interact to determine production and allocation of the pesticide and of the metabolites to the different phases of the soil.



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 4. Depth profiles of measured (symbols) and predicted (continuous lines) (parathion + metabolites) concentration profiles (a) 5 and (b) 19 d after the beginning of incubation in two soil columns. Error bars are not presented because they are less than 0.0008 (x10–6 g g–1 of soil) and not visible.

 
Figure 5 shows the evolution of parathion and of its metabolite distributed between the soluble, sorbed, and strongly sorbed pools in two soil layers (i.e., 0–0.01 and 0.03–0.04 m). For parathion, the results show:



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 5. Times series of measured (symbols) and predicted (continuous lines) concentrations for parathion (a, c) and its metabolite (b, d) in the layers 0 to 0.01 and 0.03 to 0.04 m during the 19 d of the experiment. For metabolite concentrations, the experimental points correspond to the total (paraoxon + p-nitrophenol) concentration.

 
The relative orders of magnitude of the pesticide or metabolite quantities differed greatly both in time and depth. Setting at 1 a relative index of concentration of parathion in the surface soil layer, the order of magnitude of the relative concentration of paraoxon at the same depth equaled approximately 10–1 to 10–2, whereas the relative concentrations of parathion and paraoxon in the 0.03- to 0.04-m soil layer were approximately 10–3. Consequently, paraoxon was present at approximately 0.1 to 1% of the applied initial concentration of parathion. Moreover, the paraoxon to parathion ratio increased with depth because paraoxon in deeper layers had two origins: (i) the biotransformation of the parathion previously transported in these layers and (ii) the transport of paraoxon produced itself in the upper soil layers and then moved by convective–diffusive flow. Comparison with experimental data indicated that the relative position and/or orders of magnitude of the different fractions were correctly predicted, whereas some discrepancies were observed between the calculated and observed values. Although differences were noted in the absolute values, this corresponded mainly to small fractions if expressed in terms of percentage of initially applied pesticide. Because the model does not accurately predict the front of the pesticide penetration, disagreement between predicted and observed concentrations at a given soil depth may be, indeed, the consequence of a biased prediction of a few millimeters in the pesticide front position.

Long-Term Simulation
The rainfall, soil water potential (at depths 0.2 and 1 m), and daily mean soil temperature (at depths 0.2 and 1 m) are presented in Fig. 6 (Cannavo et al., 2004). This simulation corresponded to a Mediterranean climate (mean annual rainfall and temperature of 640 mm and 13.6°C, respectively, with the monthly temperature varying between 5 and 23°C), with hot and dry summers, and most rainfall occurring in spring and autumn. By fitting the hydrodynamic properties and the thermal conductivity of the soil, a good agreement between measured and simulated water potentials and temperature is observed, which provides an efficient basis for predicting transport and biotransformation of the pesticide both depending on the soil water content and temperature. Figures 7a and 7b show the simulated total (parathion + metabolites) concentration profiles 1 and 11 mo after the application of parathion, respectively, for the soluble, sorbed, and strongly sorbed fractions. These results confirm the rapid sorption of parathion and its low diffusion through the soil profile leading to significant concentrations in the top 0.1-m soil layer only. After 1 and 11 mo, the soluble phase represented the smaller component, whereas the strongly sorbed phase became dominant because of a continuous accumulation of bound residues in the soil aggregates. Figure 8 describes the pesticide and metabolite evolution in time in two layers (0–0.01 and 0.1–0.2 m). For parathion and paraoxon and for both layers, similar patterns to those previously described in the 19-d experiment were observed. Significant sorption and desorption pulses occurred in the sorbed and strongly sorbed phases (see for example Fig. 8c and 8d), due to successive precipitation events that induced water and pesticide flows from the upper soil layers. However, a decrease of parathion and paraoxon in the strongly sorbed phase appeared 40 d after the application of parathion. This was the consequence of the decreasing concentration in the weakly sorbed phase, thus leading to a negative net balance for the strongly sorbed fraction. The strongly sorbed fraction may be regarded as a pesticide reservoir that regularly provides pesticide to the weakly sorbed phase, and then, the liquid phase.



View larger version (20K):
[in this window]
[in a new window]
  		Fig. 6. (a) Rainfall, (b) water potential, and (c) soil temperature versus time in an 11-mo simulation. The continuous lines correspond to the model predictions and the symbols correspond to the measurements performed during the experiment.

 


View larger version (20K):
[in this window]
[in a new window]
  		Fig. 7. Predicted (parathion + metabolites) concentration profiles (a) 1 and (b) 11 mo after the application of parathion.

 


View larger version (26K):
[in this window]
[in a new window]
  		Fig. 8. Predicted concentrations for parathion and paraoxon at two depths, during an 11-mo simulation.

 
Toxicological Impact of Parathion on Nematodes in Soil
The variable {int}t0tnCi x dt, where Ci is the predicted total concentration of parathion and paraoxon in the soluble pool, was calculated at four depths (0.01, 0.02, 0.05, and 0.10 m) during the 19-d period following the application of parathion. Figure 9a shows the evolution with time of this variable, as well as the experimental EC50 value of 17 for the nematode Caenorhabditis elegans as found by Saffih-Hdadi et al. (2005), corresponding to the effective concentration that inhibits the mobility of 50% nematodes. These results indicate that no toxicological impact is expected below 0.02 m. This was confirmed by the experiment, which showed that parathion applied to the soil at a dose comparable to that found in the columns at a depth of 0.04 m had no effect on nematodes after 5 h and 5 d of application, and that the mobility of nematodes extracted from the contaminated soil and from test control soil was similar. However, at depths of 0.01 and 0.02 m, predictions indicated that parathion may induce a toxicological effect on nematodes after 4 and 14 d, respectively. This was qualitatively in agreement with the toxicological observations that showed that one species of nematodes had slower movements in treated soils than in controls 5 d after application. For the 11-mo simulation period (Fig. 9b), the model predicts that parathion may have a toxicological impact on nematodes at the soil surface only (i.e., upper 0.05 m), but no short-term damage to other deeper subsurface soil layers. Moreover, for long-term prediction of biological damage of the pesticide to nematodes, the life time of nematodes must also be considered. If the life time is short, the toxicological damage will be lower than in the case of a long life time. With a short life time, each generation will receive a moderated cumulative dose of pesticide, and may be able to resist the potential toxicological effect of the pesticide.



View larger version (16K):
[in this window]
[in a new window]
  		Fig. 9. Product (concentration of parathion and paraoxon x time of application) during (a) a 19-d experiment and (b) 11 mo of simulation at different depths. The horizontal line indicates the EC50 value (effective concentration that inhibits the mobility of 50% nematodes) for the nematode Caenorhabditis elegans.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 BASIC MODEL AND EQUATIONS
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
CONCLUSIONS
 REFERENCES
 
Overall, simulations elucidated the main pathways of parathion dissipation. Sorption effects were dominant and continuous production of metabolite at a low rate was observed as well as a significant amount of bound residues. Water flow induced desorption and sorption pulses in the upper and lower soil layers, respectively. Formation and then desorption of bound residues was possible, and finally, the pesticide movement was reduced, even in the case of a long-term scenario.

Model Analysis
A model was proposed to predict the dissipation of parathion and its metabolites in soil, by linkage of water and heat transport, biodegradation, sorption, and then dynamic allocation of products into three soil pools. The sorption and decay sub-model is an extended two-site model having sites arranged in series with decay in the liquid phase to a metabolite, relatively fast kinetic sorption to Type-1 sites, and slower sorption to Type-2 sites for both parent compound and metabolite, and hysteresis of adsorption and desorption. Unlike many other studies, variably saturated conditions were employed, which require the simulation of variably saturated flow and the use of saturation-dependent sorption coefficients. The approach presented here can be regarded as a reference mechanistic model used as a basis to test simpler models. To our knowledge, there is no model in the literature that takes the proposed linkages into account in such a way. Indeed, the chosen approach differs from the simplified models using transfer-function or piston-flow hypotheses (Walker and Barnes, 1981; Carsel et al., 1984; Nofziger and Hornsby, 1986; Walker and Welch, 1989). These models are based both on a specific half-life of the pesticide biodegradation, which does not take into account the biomass dynamics, and on the instantaneous equilibrium sorption hypothesis, which assumes that sorption is entirely described by a unique adsorption coefficient Koc or Kd. However, these simplified approaches are useful for environmental diagnosis, and especially for comparative screening of environmental impacts when several pesticides are used together. Our approach also differs from previous mechanistic models based on Darcy's flow and convective–diffusive transport equation for water and solute movements, which generally assume instantaneous equilibrium sorption (Wagenet and Hutson, 1989; Piver and Lindstrom, 1990; Boesten and van der Linden, 1991; Clemente et al., 1993) without dynamic allocation of the pesticide and/or metabolite into different soil pools.

Metabolite Production
Taking into account the metabolite production and its own biodegradation and sorption behavior allowed us to predict more accurately the possible toxicological damage of parathion and paraoxon, for which one of the metabolites (paraoxon) is known to have more dangerous effects on nontarget biological organisms (Guilhermino et al., 1996) than the parent compound.

However, we considered in our model only one metabolite, whereas the analysis of aqueous or weakly sorbed extracts shows that parathion is transformed into different products, such as ethylphosphates, paraoxon, and p-nitrophenol. As a consequence, what we call "metabolite" in the model should be considered as a mixture of different products, paraoxon being the most toxic. Consequently, the predicted concentrations of the "metabolite" in our model can be considered as the maximum concentration of paraoxon in the soluble phase, since paraoxon is mixed with other products.

Ecotoxicological Impact
Moreover, we tried in this work to predict the biological damage of parathion and paraoxon to nematodes, by using the experimental EC50 value previously estimated by Saffih-Hdadi et al. (2005). As pointed out by these authors, toxicological predictions must be carefully drawn. Indeed, the EC50 value they used is in theory only valid for one species of nematodes, although toxicological observations were made in our experiments for a large number of species of nematodes. The different species of nematodes living in soil can have different responses to the same toxic compound, and consequently, other EC50 values estimated for a large range of species (herbivorous species dwelling in the rhizosphere and puncturing roots, carnivorous species preying on other invertebrates including also other nematodes, and microbivores species feeding on bacteria and fungi) would be useful to draw more general conclusions.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 BASIC MODEL AND EQUATIONS
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
REFERENCES
 
From a practical point of view, the model predicts a maximum transient production of paraoxon at a concentration of 1% of the quantity of parathion initially applied. The main conclusion to be drawn is that high sorption limits parathion fluxes toward the water table. Parathion and its biodegradation products remain at the soil surface, there is no major risk for the transport of parathion and its metabolites through the soil profile, and there is no major effect of parathion on soil nematode mobility when applied at doses comparable to usual agricultural practices, except maybe at the soil surface. Preferential flow can exist under natural conditions, mainly in clay and sandy soils; then a part of the applied parathion and/or its metabolite can be transported to greater depths in the soil profile. Another environmental risk arises from the formation of a bound residue fraction in the soil aggregates, but the reversibility of this pool remains in question, and more data over longer periods of time are needed to draw any clear conclusion about the impact of these residues. Moreover, a more complete modeling describing the production of all metabolites of parathion would be useful.


   ACKNOWLEDGMENTS
 
The authors are grateful to Dr. Pierre Benoît and Dr. Valérie Pot (UMR INRA-INAPG Environnement et Grandes Cultures) for helpful discussions, Mr. Rolland Lorrain for his experience in nematode observations, and Mr. Dominique Renard and Mr. Christophe Labat (INRA, Unité Climat, Sol et Environnement; UMR INRA-INAPG Environnement et Grandes Cultures) for their technical assistance. We also thank Dr. Patrice Cannavo for the data relative to the 11-mo experiment.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 BASIC MODEL AND EQUATIONS
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES