Behavior of the Chiral Herbicide Imazamox in Soils: Enantiomer

Article Cite This: Environ. Sci. Technol. 2019, 53, 5733−5740

pubs.acs.org/est

Behavior of the Chiral Herbicide Imazamox in Soils: Enantiomer Composition Differentiates between Biodegradation and Photodegradation Ignaz J. Buerge,* Roy Kasteel, Astrid Bächli, and Thomas Poiger Plant Protection Chemistry, Agroscope, CH-8820 Wädenswil, Switzerland

Downloaded via NOTTINGHAM TRENT UNIV on August 13, 2019 at 02:36:45 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: Imazamox is a chiral herbicide that, in laboratory experiments in the dark, exhibits pronounced enantioselective biodegradation in certain soils. Imazamox also shows rapid photodegradation. However, which processes are predominant in the field is not clear. We conducted a set of soil incubation experiments under natural sunlight (and corresponding dark controls), using enantioselective LC− MS/MS analysis as a probe to distinguish biodegradation and photodegradation. Under dark conditions, imazamox was degraded enantioselectively. In contrast, degradation was nonenantioselective and 2× faster when the soil was exposed to sunlight, suggesting that biodegradation (in the dark) and photodegradation (under sunlight) were the predominant degradation processes. We also investigated the effectiveness of strategies that were proposed to exclude photodegradation in field studies, covering of soil with sand or irrigation after herbicide application. The sand cover did not prevent photodegradation. On the contrary, degradation was 10× faster than in the dark and nonenantioselective. Computer simulations supported the explanation that imazamox was transported upward by capillary flow due to evaporation onto the sand surface, where it was rapidly photodegraded. Irrigation postponed but not completely prevented photodegradation. For mobile substances susceptible to photodegradation, upward transport to the soil surface thus needs to be considered when deriving rates for biodegradation from field studies.

■

INTRODUCTION Laboratory and field studies on pesticide degradation in soil each come with their specific advantages and limitations. In laboratory incubation studies, many experimental parameters, such as soil temperature and moisture, can easily be controlled and standardized. Use of radio-labeled compounds provides means to derive route and rate of degradation and to establish a mass balance, including formation of metabolites, CO2, organic volatiles, and bound residues. A gradual decline in microbial activity upon handling and storage of soils is, however, one of the potential disadvantages of the artificial conditions in the laboratory.1 In field studies, loss of microbial activity generally is not an issue, and in fact, degradation rates of pesticides are often faster than under laboratory conditions (after normalization to a reference temperature and moisture). On the other hand, data typically show more scattering as a result of inhomogeneous application or spatial variability in soil properties. Furthermore, in the field, processes other than microbial degradation may contribute to the observed dissipation of pesticides from soils, e.g., uptake by plants, leaching to deeper soil layers, surface runoff, or further loss processes on the soil surface, such as volatilization or photolysis. These additional loss processes have to be taken into account when deriving © 2019 American Chemical Society

(microbial) degradation rates. Uptake by plants is usually excluded by maintaining the test plots free from vegetation throughout the field trial. The influence of leaching is assessed based on the analysis of soil samples from deeper layers. Possible strategies to minimize surface processes include incorporation of the pesticide in the soil, application of a sand layer on the soil surface, or irrigation.2 For the design of new field dissipation studies, the European Food Safety Authority (EFSA) recommends an incorporation depth of 7 cm. Alternatively, soil can be (i) covered with a fine sand layer of at least 3 mm or (ii) sufficiently irrigated after application to reach an average penetration depth of the pesticide of 10 mm. For the assessment of old studies (legacy studies), EFSA proposes to exclude data points before a cumulative rainfall and/or irrigation of 10 mm had occurred after application. This recommendation was, for example, followed in the kinetic reevaluation of legacy field studies with the herbicide imazamox.3 In more recent field studies with this compound, Received: Revised: Accepted: Published: 5733

December 21, 2018 April 11, 2019 April 24, 2019 April 24, 2019 DOI: 10.1021/acs.est.8b07210 Environ. Sci. Technol. 2019, 53, 5733−5740

Article

Environmental Science & Technology Table 1. Selected Properties of the Clay Loam Soil, the Sand, and the Florist Foam texture (%) clay loama sanda florist foam

sand

silt

clay

pH (CaCl2)b

Corg (%)c

ρb (g/cm3)d

θv(pF 2.5) (cm3/cm3)e

27 95

43 3.1

30 1.7

6.9 7.5

3.8 290 nm), a half-life of only 5 h was determined at pH 7; whereas no degradation was observed in the dark within 15 d.3 However, the observed photolysis in water seems to be inconsistent with studies on the photolysis of imazamox on soil surfaces. Upon irradiation of a 1- to 2-mm thick, air-dried soil layer with a xenon-arc lamp, imazamox was only slowly degraded with a half-life of 65 d (again no degradation in the dark within 30 d).3 In another study, a 2-mm thick, moist soil layer was irradiated with a xenon lamp. Imazamox was degraded with a half-life of 28 d, but in the dark control, degradation was faster (half-life, 20 d). It was thus concluded that photodegradation would not play a significant role in soils.3 Imazamox is a chiral compound and consists of a pair of enantiomers. In our accompanying study, we found that in pHneutral soils under dark conditions, (+)-imazamox was degraded faster than (−)-imazamox, leading to enrichment of (−)-imazamox with proceeding degradation.7 While the enantioselective degradation observed in these soils points to a microbially mediated process, photodegradation would, by nature, not be enantioselective.8−10 The enantiomer composition may thus serve as a probe to differentiate between microbial and photochemical degradation in soil. The aim of the present study was to investigate whether imazamox nonetheless undergoes photodegradation on soil surfaces and whether photodegradation could be prevented by irrigation or application of a sand layer. For that, we devised a set of six small-scale mesocosm experiments using a clay loam soil in which microbial degradation in the dark was shown to be enantioselective (preliminary experiments). Degradation of imazamox was followed in soils exposed to natural summer sunlight and soils that were kept in the dark under otherwise identical conditions, and we studied the possible influence of irrigation or coverage with a sand layer prior to sunlight exposure. Residues of imazamox were determined with enantioselective LC−MS/MS (liquid chromatography−tandem mass spectrometry) with the aim to distinguish between microbial and abiotic degradation.7 Additionally, simulations were run with a numerical water flow and solute transport model to support our interpretation of the observed degradation of imazamox in the different experiments.

■

MATERIALS AND METHODS Chemicals. Imazamox (purity, 99.9%, 2-[(RS)-4-isopropyl4-methyl-5-oxo-2-imidazolin-2-yl]-5-methoxymethylnicotinic acid) and imazapic (99.9%, 2-[(RS)-4-isopropyl-4-methyl-55734

DOI: 10.1021/acs.est.8b07210 Environ. Sci. Technol. 2019, 53, 5733−5740

Article

Environmental Science & Technology

cylinders (typically, 22−28 g wet weight) was then transferred into 50 mL polypropylene centrifuge tubes and stored in a freezer at −20 °C until extraction. Samples were taken immediately and 2, 4, 7, 20, 29, 53, 77, 103, 164, and 236 h after application. After sampling, the empty cylinders remained in the soil/florist foam (see TOC graphic). Extraction and Enantioselective LC−MS/MS Analysis. All samples of a particular experiment were processed on the same day. The extraction followed the procedure described in the accompanying paper,7 with the difference that twice the volume of extracting solvent (20 mL 0.5 M NaOH) and internal standard (2 μg imazapic in 200 μL water) was used to account for the larger soil samples. Imazamox and its metabolites were then analyzed with enantioselective LC− MS/MS.7 Kinetic Analysis. Kinetic parameters of imazamox degradation were determined using the software KinGUII.12 Total concentrations (sum of enantiomers) were fitted with the DFOP (double first-order in parallel) model.13 This biphasic model assumes two compartments in which the compound is degraded according to first-order kinetics but with different rate constants, k1, and k2 c(t ) = c(0)(ge−k1t + (1 − g )e−k 2t )

c(t) is the total concentration at time t, and g is the fraction of the initial concentration c(0) applied to compartment 1. However, since the slower rate constant (k2) was not significantly different from zero in all experiments, it was fixed to zero; i.e., in one compartment, it was assumed that no degradation occurred. Initial concentrations were adjustable. For fitting, the iteratively reweighted least-squares optimizer was selected. Numerical Simulation of Water Flow and Solute Transport. Simulations were primarily aimed at supporting the discussion of the findings in the different degradation experiments and not at an exact description of the measurements by an inverse parameter optimization. The finite element software package Hydrus-1D (version 4.16)14 was used to numerically solve the one-dimensional water flow (Richards’ equation) and solute transport (advection-dispersion equation). A constant nodal distance of 0.3 mm was used for the spatial discretization in all simulations. The iteration criteria were chosen in such a way that the relative error in the water and solute mass balance of the entire flow domain was always below 1%. The hydraulic properties, i.e., the water retention characteristic and the hydraulic conductivity function, were parametrized by the Mualem−van Genuchten model15 and are summarized in Table 2 for the clay loam soil, the sand, and the florist foam. The parameters for the sand were estimated based on texture and dry bulk density and for the clay loam

Figure 1. Setup of six degradation experiments (A−F) with different orders of herbicide application, irrigation, and sand coverage, with or without sunlight exposure. A photograph of the experimental setup is shown in the TOC graphic.

The sequence in which herbicide application, irrigation, and sand coverage were performed, differed in the six experiments (Figure 1). For the three dark controls, shading was accomplished using somewhat larger polypropylene buckets wrapped in aluminum foil, which were placed upside down over the buckets holding the soil (Figure 1). This setup ensured efficient shading while still allowing air exchange. The other three buckets were exposed to natural sunlight in Wädenswil (47°13′19′′N/8°40′37′′E; altitude, 480 m). These buckets were only covered during the night or when it rained (only once, 13 mm). Apart from this single rain event, the experiments were conducted during a sunny period with 7−15 h sunshine per day (on average, 12 h). The mean soil temperature 2 cm below the soil surface was 21.0 °C in covered buckets and 22.5 °C in sunlight exposed buckets. The soils were always moist (also in the sunlight exposed experiments up to the surface) but not saturated with water due to the height of the florist foam layer. For sampling, 5.5 cm long plastic cylinders with an inner diameter of 2.5 cm were cut into the soil and the first few millimeters of florist foam (Figure 1). The soil in these

Table 2. Parameters of the Water Retention Characteristic and the Hydraulic Conductivity Function for the Clay Loam Soil, the Sand, and the Florist Foam clay loam sand florist foam

θra (cm3/cm3)

θsb (cm3/cm3)

αc (1/cm)

nvgc (−)

Ksd (cm/h)

τe (−)

0.08 0.05 0.0

0.45 0.43 0.99f

0.0078 0.036 0.01

1.44 3.25 2

0.55 39 20

0.5 0.5 0.5

a Residual water content. bSaturated water content. cCurve shape parameter. dSaturated hydraulic conductivity for the soil matrix. eFactor accounting for tortuosity and pore connectivity in the hydraulic conductivity function. fMeasurement based on difference in weight of a piece of florist foam with known volume before and after immersion in water.

5735


Article


and concentration distribution obtained after the setting time as an initial condition for water flow and solute transport, respectively.

additionally on the estimated volumetric water content at pF 2.5 (Table 1), using the pedotransfer function approach (Rosetta) implemented in the software package. For the packed clay loam soil, we additionally assumed a macro-pore domain starting at a volumetric water content of 0.445 cm3/ cm3 with a saturated hydraulic conductivity of 8 cm/h, following the approach of Vogel and Cı ́slerová.16 The hydraulic properties of the florist foam were based on expert judgment and are not decisive, as long as it can sustain the required flux imposed by the atmospheric demand (bare soil evaporation). The molecular diffusion coefficient of imazamox was set to 0.018 cm2/h, which is a general estimate for compounds with a molecular mass of 200−250.17 The dispersivity for the clay loam was set to 0.05 cm, in line with the dispersivity of 0.06 cm reported by Unold18 for a repacked silty loam, which implicitly included spreading of the tracer by molecular diffusion. A value of 0.05 cm was also used for the sand cover. A soil-specific Freundlich distribution coefficient, KF, of 0.49 μg1−1/n cm3/n g−1 was calculated for the clay loam (Corg, 3.8%) using an organic carbon-normalized Freundlich adsorption coefficient, KFoc, of 12.9 μg1−1/n cm3/n g−1 with a Freundlich exponent 1/n of 0.938 (geometric and arithmetic mean from 13 soils, respectively).3 The rate constant of microbial degradation in the clay loam was taken from the dark control experiments (mean k1 = 0.0245 h−1 from experiment A and C) and was assigned to both, the water and the solid phase. In the sand layer, however, adsorption and microbial degradation were neglected. Photodegradation in sunlight exposed soils (B, D, and F) was assumed to occur only in the water phase, at a rate of 0.231 h−1 (corresponding to a half-life of 0.125 calendar days (3 h) for natural sunlight at 40°N, end of May),3 through a depth of 1 mm for the clay loam soil and 3 mm for the sand. The penetration depth of light in soil was reported to be in the order of 1 mm or less19 but was expected to be higher in the sand (here, we considered the whole thickness of the sand layer). At the start of the simulations, the 14-cm tall florist foam was at hydrostatic equilibrium with a water level at the bottom. The soil had a uniform matric potential distribution of −500 cm and was void of imazamox. The lower boundary was characterized by a seepage face; i.e., water could only leave the florist foam when a local water saturation occurred at the bottom. The soil was irrigated for 5 min at a rate of 12 cm/h (10 mm of water), followed by the application of imazamox with 0.1 mm of water (experiments A, B, E, and F) or vice versa (experiments C and D). The input concentration was 40 μg/cm3. Experiments E and F were run with an additional 3 mm sand cover, which was placed on top of the soil directly after the application of imazamox. The sand was void of imazamox and had an uniform initial matric potential distribution of −200 cm, mimicking that its initial moisture status was very dry. After a setting time of 15 min, during which only microbial degradation was allowed, we assumed constant evaporation of water for 250 h, with a higher flux density of 0.03 cm/h in the sunlight exposed experiments, which resembles the daily maximum evaporation rate for an average sunny day in August,20 compared to 0.01 cm/h in the dark controls. A constant water level was maintained 14.1 cm below the soil surface (experiments A to D) or 14.4 cm below the surface of the sand cover (experiments E and F). All experiments were run starting with the appropriate simulated matric potential

■

RESULTS AND DISCUSSION Herbicide Distribution at the Start of the Experiments. The vertical distribution of imazamox in the soil column at the beginning of the experiments was expected to depend on the experimental conditions; i.e., whether the soil was irrigated prior to or after herbicide application or covered with a sand layer. This initial distribution was estimated by computer simulation (Figure 2). Imazamox shows low

Figure 2. Simulated concentration distributions in the soil and sand cover after a setting time of 15 min (before exposure to sunlight) in experiments with irrigation before or after herbicide application and with sand coverage. Depth 0 represents the surface of the clay loam soil, while positive values represent the sand cover. The inset shows the concentration distribution in the experiment application/irrigation after 0, 7, 20, and 64 h of sunlight exposure assuming no microbial degradation and no photodegradation, indicating upward transport back to the soil surface.

adsorption to soil with organic carbon-normalized Freundlich adsorption coefficients (KFoc) of 2−130 mL/g.3 Above pH 3.3, the compound is anionic and quite hydrophilic and nonvolatile.3 Consequently, irrigation with 10 mm of water after herbicide application caused significant leaching down to a depth of ≈1.5 cm (Figure 2, experiments C and D). However, imazamox did not reach the florist foam. In contrast, when imazamox was applied after irrigation, the compound was located primarily in the top 1−2 mm of soil (experiments A and B). Finally, in experiments E and F with sand coverage, the difference in matric potential between the dry sand and the wet soil induced an upward water flow and transport of imazamox into the sand cover (64% after 15 min). Comparable Half-Lives in Dark Control Experiments. The soil used in this study was selected because it had shown rapid and enantioselective degradation in preliminary batch incubation experiments (data not shown). In the definitive experiments under dark conditions, imazamox was degraded with half-lives of 37−44 h (Figure 3, left side, experiments A, C, and E and Table 3). These half-lives are among the shortest half-lives observed in our laboratory batch incubation experiments.7 Irrigation (experiment C) or sand coverage (experiment E) after herbicide application apparently did not notably affect the half-lives under dark conditions. 5736


Article


Figure 3. Degradation of imazamox in soil under various conditions. The left panels show measured concentrations (symbols; in % of the fitted initial concentration) and kinetic fits (lines) of experiments under dark conditions (full symbols and continuous lines) and experiments exposed to natural sunlight (empty symbols and dashed lines). Grayed out data points were considered as outliers and were not included in the fitting. The insets show the change of the enantiomer fraction during incubation. The right panels show simulated concentrations, without (black lines) or with (blue lines) temperature-dependent degradation.

Table 3. Half-Lives (DT50) of Imazamox in Six Degradation Experiments Kinetically Analyzed with the DFOP Model experiment

DT50 (h)

A, irrigation/application/dark B, irrigation/application/sunlight C, application/irrigation/dark D, application/irrigation/sunlight E, irrigation/application/sand cover/dark F, irrigation/application/sand cover/sunlight

37 16 41 38 44 3.3

c(0)a,b 28.5 29.2 24.7 24.5 24.4 27.4

± ± ± ± ± ±

0.9 2.3 1.5 1.0 1.6 2.6

k1 (1/h)a,c 0.024 0.051 0.025 0.019 0.016 0.23

± ± ± ± ± ±

0.003 0.021 0.006 0.004 0.005 0.05

g (−)a

χ2 error (%)

± ± ± ± ± ±

5.8 17 12 8.7 15 24

0.86 0.90 0.77 0.97 1.00 0.95

0.03 0.06 0.06 0.06 0.12 0.05

Mean ± standard deviation. bArbitrary scale. cDegradation rate constant k2 was statistically not different from 0 (based on 95% confidence interval) and was therefore fixed to 0. a

Differences between the three dark controls were only observed toward the end of the experiments. After 236 h, ≈80% degradation was achieved in experiments A and C (irrigation before and after application, respectively); whereas, in experiment E (with sand), more than 90% were degraded at

this time (Figure 3, left side). Degradation was thus clearly biphasic in experiments A and C and could be described reasonably well by the DFOP model (χ2 error, 5.8 and 12%, respectively, Table 3). For experiment E, an acceptable fit would also result with the simple first-order model (χ2 error, 5737


Article


degradation of imazamox in the dark control (Figure 3, right side, simulation A). The average temperature of 19.5 °C during the first 4 days, when most degradation occurred, was close to the reference temperature of 20 °C, and the amplitude of the temperature variation was relatively small (on average, 3.6 °C during the first 4 days at 2 cm below the soil surface). However, the effect of temperature was more pronounced in sunlight-exposed soil (Figure 3, right side, simulation B), since average temperature and amplitude were both higher during the first 4 days (20.4 and 6.2 °C, respectively). Nonetheless, on the basis of these simulations, the influence of temperature is too small to explain the observed differences between experiments A and B and thus confirms a significant contribution of photodegradation in experiment B. In the dark, the two major soil metabolites, a diacid and an acid-hydroxy metabolite,3 were both found in maximum amounts of ≈10% after 10 d (based on semiquantitative analysis).7 In sunlight exposed soil, however, the diacid reached only 1%, and the acid-hydroxy metabolite was not detectable at all. These findings are in line with studies on the direct photolysis of imazamox in water, where these two metabolites were also not detected.4−6 Further evidence for photodegradation comes from the observed changes in enantiomer composition. In the insets of Figure 3 (left side), we plotted the enantiomer composition of imazamox residues, expressed as enantiomer fraction (defined as EF = [+]/([+] + [−]), where [+] and [−] are the concentrations of (+)- and (−)-imazamox, respectively). In the dark control experiment A, the EF value continuously decreased from ≈0.5 to ≈0.4 after 236 h. Degradation was thus enantioselective with a preference for the (+)-enantiomer, as previously observed in other neutral soils.7 This change of the enantiomer composition points to a microbially mediated process. In the corresponding sunlight exposed experiment B, the enantiomer fraction did not change notably. The slopes of linear regressions EF vs time for experiments A and B were statistically different based on a 95% confidence interval. Faster degradation, insignificant formation of the major metabolites, and negligible enantioselectivity thus provide evidence that photolysis was a major degradation pathway in experiment B. Influence of Irrigation on Photodegradation. In experiments with irrigation after application, a considerable portion of imazamox was transported away from the soil surface as shown by the simulations (Figure 2). The initial, average penetration depth, calculated as the first spatial moment, was 9 mm and thus slightly below the penetration depth of 10 mm, as required by guidance of EFSA to minimize the impact of surface processes, such as photolysis.2 Half-lives were similar, both in the dark control and sunlight exposed experiment (≈40 h, Figure 3, left side, experiments C and D and Table 3). At later sampling times, however, concentrations decreased faster in the sunlight exposed soil than under dark conditions. In addition, the enantiomer fraction did not change in the sunlight exposed soil, suggesting that photolysis may have contributed notably to the overall degradation of imazamox (note that slopes of linear regressions EF vs time for experiments C and D were again statistically different based on a 95% confidence interval). It thus seems that the hydrophilic compound was transported back to the soil surface, where it was again exposed to sunlight. Upward water movement was likely to occur in the soil.21 The surface of the soil was always moist, and the water level in the trays had to be adjusted regularly (experiments were performed at

15%). A reason for the differences at later sampling times, however, cannot be given. Overall, the data showed more scattering than those measured in the laboratory batch experiments (χ2 errors, typically

Behavior of the Chiral Herbicide Imazamox in Soils: Enantiomer

Recommend Documents