Degradation and sorption parameters
Chapter 8, PLAP - Monitoring results May 1999 - June 2002
Information on degradation and sorption is of considerable importance for determining the fate of pesticides, including for the modelling of leaching. Site-specific information is usually sparse, however, and data from the literature often have to be used instead. To eliminate the uncertainty associated with the use of data from the literature and facilitate interpretation of the results of pesticide analyses, the present project incorporates studies on both half-life and K d (K oc ) in Danish soils to demonstrate degradation and sorption, respectively. Microbial biomass and microbial activity of the soils at the sites were also determined to clarify the level of microbial activity in the soil. The degradation and sorption parameters have been determined for combinations of seven pesticides and six soil types (both plough layer and subsoil) (Table 16). With fenpropimorph and flamprop-M-isopropyl, important degradation products were also investigated.
Table 16. Soil-pesticide combinations included in the degradation and sorption studies.
8.1 Materials and methods
8.1.1 Soil sampling
Degradation and sorption were determined in the laboratory using pooled soil samples. The samples were collected as short a time preceding pesticide application as possible from both the plough layer (0-20 cm) and the subsoil (80-100 cm). To avoid microbial and chemical contamination, the sampling equipment was cleaned with ethanol prior to use.
The plough layer samples were collected using a hand auger (2 cm inner diameter and 20 cm long). A sample based on 50 to 100 subsamples was collected from the plough layer within the test field at the Tylstrup, Silstrup, Estrup, and Slaeggerup sites. At the other two sites, Jyndevad and Faardrup, spraying had been carried out before soil sampling could be undertaken and the samples were therefore collected from the buffer zone surrounding the test field. Subsoil samples were collected from the walls of two 50 x 100 cm pits excavated in the buffer zone with the samples being collected horizontally. Each sample consisted of at least 2 kg of soil per substance per field per depth.
The samples were stored at 5°C until needed for the experiments. All results are expressed
on a dry weight basis. Prior to the experiments, the soils were homogenized and sieved (2 mm) to remove any stones and plants.
8.1.2 Microbial biomass and activity
Microbial biomass was measured using the substrate-induced respiration (SIR) method (Anderson and Domsch, 1978), which is a physiological method based on the increase in the respiration rate when glucose is added to the soil. The concentration of glucose yielding the most CO 2 was determined prior to the experiment. CO 2 evolution was measured by gas chromatography. Microbial activity was measured by the degradation of 14 C-labelled Naacetate. 14 C Na-acetate (5 µg/g) was added to the soil in an Erlenmeyer flask and the 14 CO 2 evolved was collected and counted using a scintillation counter. All studies were performed in quadruplicate.
8.1.3 Incubation of soil
The degradation studies were performed on mixed, homogenized soil from each field site. After homogenization, the water content of the soil was determined. The soil was air-dried and sieved. During the drying process the soil was mixed frequently to avoid excessive drying of part of the soil. For each degradation experiment, 10 replicates of each soil were prepared in Erlenmeyer flasks. An aquatic solution of the test pesticide was added to each flask and the water content adjusted to 40-60% of the water holding capacity (WHC). The initial pesticide content was 0.5 mg/kg dry soil. The plough layer (0-20 cm) and subsoil (80-100 cm) samples were incubated at 20°C and 10°C, respectively. The Erlenmeyer flasks were closed with rubber stoppers and hydrophobic cotton, which allowed diffusion of air and minimized desiccation of the soil during incubation.
At certain time intervals the incubation was discontinued for one replicate at a time, and the soil sample stored at -18°C until analysis. The time intervals were set for each pesticide according to the half-life reported in the literature, ensuring that the incubation period encompassed at least three half-lives. Each degradation experiment was performed in duplicate.
The extraction of pesticides was normally performed by ASE (Accelerated Solvent Extraction) at specific temperatures, pressures and duration. Exceptions were fenpropimorph and propiconazole, which were extracted and afterwards shaken on a Mastermixer (Spliid, 2000). Dimethoate was detected by means of GC/MS, whereas the other pesticides were detected by means of LC/MS (Tabel 17). Blanks and recovery were analysed in each run of the ASE apparatus and for each batch of shaking. If recovery differed significantly from 100%, the results were corrected to 100% recovery. Certificate standards were obtained from Dr. Ehrenstorfer in Germany. Stability tests were performed by adding the pesticides to soil samples and then storing them at -18°C for a period corresponding to the storage period of the test samples. If the recovery was low, the analytical results were corrected to 100% recovery. The detection limits for the pesticides and degradation products are shown in Table 17.
Table 17. Detection limits for the pesticides and degradation products included in the degradation and sorption studies.
8.1.5 Degradation kinetics
The registration procedures for pesticides and many published degradation studies assume that the degradation of pesticides follows simple first-order degradation kinetics. The halflife is thus estimated and used for further evaluation. A number of recent publications have shown that a two-compartment 1 st + 1 st order model better describes the degradation processes (Fomsgaard, 1999). In a two-compartment model, one part of the added pesticide is rapidly degraded, while another part is adsorbed to the soil, and thus degraded much more slowly.
Once a sufficient number of data points had been obtained, a curve-fitting analysis was performed comparing the use of a simple 1 st order model and a two-compartment 1 st + 1 st order model. The modelling was performed using the software TableCurve 2D. The mathematical expressions are:
8.1.6 Determination of sorption
Sorption was determined in both plough layer and subsoil samples. The soil samples were sieved (2 mm) and homogenized. To reduce microbial activity the soils were irradiated with 10 Kgray. Sorption experiments were carried out in a manner similar to that described in OECD (1997). The ratio between soil and 0.01 M CaCl 2 was fixed on the basis of literature values for K d as described in Table 18. The ratio was selected in order to obtain an acceptable concentration ratio after equilibration. All experiments were performed in triplicate using the same concentration of unlabelled pesticides. After shaking the soil with 0.01 M CaCl 2 for 24 hours the suspension was centrifuged, and the concentration of the pesticide in the aqueous phase determined by LC/MS. The pesticide concentration sorbed on soil was then calculated, and the constants K d and K oc calculated as follows:
Table 18. Pesticide concentrations and water:soil ratios used in the sorption experiments.
8.2 Results and discussion
8.2.1 Soil characteristics
As could be expected, the microbial biomass and the content of total organic carbon were significantly greater in the plough layer than in the subsoil at all test sites (Table 19). The biomass was highest in the soil from Silstrup (641 mg biomass C/kg) and lowest in the sandy soil from Tylstrup and Jyndevad (142 and 194 mg biomass C/kg, respectively). The high microbial biomass at Silstrup might be due to the frequent application of manure at the site in previous years (Lindhardt et al., 2001). The microbial activity is expressed as the percentage 14 C evolved in the form of 14 CO 2 from 14 C-labelled acetate during 2 and 96 hours of incubation (Table 19). The evolution from plough layer soil was fastest in soil from Estrup (20% evolved after 2 hours) and slowest in soil from Faardrup (9% evolved after 2 hours). After 96 hours, almost the same percentage had evolved from all soils. In the subsoil, 14 CO 2 evolution after 2 hours amounted to less than 2% in all soils, thus confirming the low microbial biomass in these soils. On the other hand, more than 40% of the 14 C from 14 C-labelled acetate had evolved after 96 hours, thus indicating the potential for degradation of the very easily degradable acetate.
Table 19. Organic carbon, microbial biomass and microbial activity determined in the plough layer (0-20 cm) and the subsoil (80-100 cm) at the PLAP sites.
All the experiments were performed on homogenized soil samples. To confirm that the soils were properly mixed, 14 CO 2 evolution from 8 individual samples was determined after addition of acetate (2 replicates). The evolution was almost identical during the whole experimental period. Even though the 14 C-Na-acetate method is not very sensitive to minor differences in soil microbial activity, the identical evolution of 14 CO 2 from the individual soil samples indicates that homogenization of the soil samples was satisfactory.
The results of the sorption studies are summarized in Table 20.
Table 20. Site-specific sorption coefficients for each of the pesticides analysed. K d is the mean of triplicate measurements ±SD. The organic carbon content of the soils is also shown. Literature K oc ranges are included for comparison.
Comparison of K d for the individual pesticides in different soils reveals that compounds such as flamprop-M-isopropyl, metamitron and propiconazole exhibit increasing adsorption with increasing soil organic matter content. This correlation was less obvious for dimethoate, fenpropimorph, ioxynil and bromoxynil.
Compared with the literature values, K oc was relatively low for bromoxynil and ioxynil, whereas K oc for dimethoate and flamprop-M-isopropyl was high. With fenpropimorph, K oc was lower than the literature range on sandy soils, but higher on loamy soils. K oc of metamitron was within the literature range, while that of propiconazole was at the higher end of the literature range.
The data confirmed that sorption is generally very low in the subsoil, probably due to the very low organic matter content (Table 19). Due to the very low carbon content, K oc has not been calculated for the subsoil analyses. K d in subsoil was negative for flamprop-Misopropyl in Estrup soil, for ioxynil in Slaeggerup soil and for bromoxynil in both Faardrup and Slaeggerup soil (data not shown). K d could not be calculated for fenpropimorph in the subsoils.
The degradation parameters for 6 pesticides in plough layer and subsoil are shown in Table 21. In the plough layer, the half-lives for the very short-lived pesticides bromoxynil, ioxynil and dimethoate varied little between soils, being less than one day for bromoxynil and ioxynil and up to a few days for dimethoate. With fenpropimorph, flamprop-M-isopropyl and propiconazole the inter-pesticide variation was much greater, as was the inter-site variation for the individual pesticides.
With propiconazole the first order half-lives varied from 106 to 444 days in these laboratory experiments in the order Faardrup < Jyndevad < Tylstrup < Slaeggerup. The difference in rate of degradation between the Faardrup, Jyndevad and Tylstrup soils may be due to the decreasing biomass (372, 194 and 142 mg C/kg, respectively) and increasing adsorption (11.8, 21.3 and 40.0, respectively). The correlation between degradation and biomass/sorption is less clear for the Slaeggerup soil, however. With fenpropimorph the degradation rate was also highest in the Faardrup soil and decreased in the order Faardrup > Jyndevad > Tylstrup. As mentioned above, the biomass decreased in the same order. As K d was 60.7 at Faardrup, 29.3 at Jyndevad and 37.7 at Tylstrup, sorption did not seem to be very important for the degradation rate of fenpropimorph.
With flamprop-M-isopropyl, the half-life was 16 days in Slaeggerup soil and 125 days in Estrup soil. The difference did not correlate with the biomass, which was higher in Estrup soil (430 mg C/kg) than in Slaeggerup soil 346 mg C/kg), but is probably attributable to the fact that K d was higher in Estrup soil (28.3 ml/g) than in Slaeggerup soil (6.8 ml/g), thus delaying bioavailability and hence pesticide degradation at Estrup.
From Table 20 it can be seen that degradation is considerably lower in subsoil than in plough layer (DT 50 is generally much longer in subsoil). The half-life of the short-lived pesticides bromoxynil, ioxynil and dimethoate ranged from less than one day for bromoxynil in Faardrup subsoil to 70 days for dimethoate in Estrup soil. These short DT 50 values indicate that these three pesticides degrade in subsurface soils, thus making it unlikely that they will cause groundwater pollution. With fenpropimorph and propiconazole, the degradation rates in subsoils are so slow that their half-lives could not be calculated during the 300-day incubation period. In most cases DT 50 exceeded 300 days. This indicates that these pesticides will be rather stable in the subsoil. On the other hand, as Table 20 shows high adsorption of fenpropimorph and propiconazole in plough layer (11.8 and 60.7 ml/g), the risk that they will leach seems to be low.
Table 21. Degradation parameters for the analysed pesticides. Half-lives are either estimated from drawn curves (DT 50 - Read), calculated using a simple 1 st order model (DT 50 - 1 st order) or calculated using a two compartment 1 st + 1 st model (DT 50 - 1 st + 1 st order). 1 st , 2 nd and 3 rd refer to the first, second and third half-lives
determined using the two-compartment 1 st + 1 st mode. Literature DT 50 ranges are shown for comparison.
The correlation between the rate of degradation and adsorption/biomass is thus unclear. In several cases, degradation was fastest at high biomass content and seemed to be influenced by adsorption - decreasing at increasing K d . In other cases the correlation was solely explicable by either biomass or by adsorption.
Fenpropimorphic acid, an important degradation product of fenpropimorph, was identified in the plough layer soil (Table 22), but not in the subsoil. After 240 days the metabolite accounted for less than 5% of the applied parental compound.
Table 22. Concentration of fenpropimorphic acid in plough layer from Tylstrup, Jyndevad and Faardrup incubated for up to 240 days following application of fenpropimorph (0.5 mg/kg). Values are in mg/kg.
The analyses for the degradation product of flamprop-M-isopropyl (flamprop-M-isopropyl acid) was only performed in the plough layer. At Slaeggerup, but not at Estrup, small amounts were detected and after 270 days the metabolite accounted for about 5% of the 0.5 mg/kg of the parent compound applied (Table 23).
Table 23. Concentration of flamprop-M-isopropyl acid in plough layer from Estrup and Slaeggerup incubation for up to 270 days following application of flamprop-M-isopropyl (0.5 mg/kg). Values are in mg/kg.
8.2.4 Degradation kinetics
In mathematical descriptions of the degradation of pesticides, the single first-order model (SFO) has been dominant for decades. From a conceptual point of view, this model is a logical choice since the rate of degradation is assumed to depend solely on the number of molecules expressed in terms of concentration or absolute mass. It is thus a well-known kinetic model. The single first-order (SFO) model can be expressed in both an integrated and a differential form. In its differentiated form, the SFO model can be used in dynamic leaching models that include a change in pesticide concentration over time. The equation can be analytically solved, and end-points (DT 50 , DT 90 ) can easily be calculated. These endpoints are typically used to assess whether a specific substance can be approved, or whether further studies have to be performed.
Practical experience shows, however, that the best mathematical description of the kinetics of chemical decomposition frequently differs from single first-order kinetics. This applies both to the degradation of pharmaceuticals in living organisms and - as in the present case - to the degradation of chemicals in soil (Beulke and Brown, 2001; Reid et al., 2000).
There may be many reasons why a single first-order model does not provide the best description of the degradation of a chemical. Both soil and water/sediment are complex environments where populations of degrading microorganisms vary considerably. Many chemicals can be degraded by different degradation pathways that may involve both chemical and microbiological steps. Furthermore, chemicals are distributed between soil and water by complex adsorption/desorption mechanisms that influence the availability of the chemical to microbial degradation.
A frequently used alternative model for describing degradation kinetics is the twocompartment 1 st + 1 st order model, Double First Order in Parallel (DFOP). Unfortunately, this model cannot be described in a differential form, and DT 50 and DT 90 can only be calculated by an iterative procedure. In cases where the DFOP model undoubtedly provides the best mathematical expression of a set of data, consideration must be given to whether an SFO model with a poor fit is preferable for the purpose of obtaining a result that can be used directly in a dynamic leaching model, or whether a DFOP or other alternative model with a better fit shall be used, even if the results from such a model cannot be used directly in the dynamic leaching models. To address this question, the EU has established a work group on degradation kinetics (FOCUS work group on degradation kinetics) to provide regulatory guidance for kinetic analyses in pesticide degradation studies.
In the two-compartment model, the first compartment - rapid degradation - is expected to occur within the soil water phase, where microorganisms have easy access to the pesticide. In the second compartment, degradation is slow. Here, the pesticide is expected to be adsorbed to soil particles or to be located in micropores in the soil matrix, with the degradation rate being governed by the slow desorption-diffusion processes. The distribution of pesticide between compartments is governed by the structure of the pesticide as well as by the amount and type of organic matter present in the soil. The speed at which the pesticide is transformed in the two compartments is expressed by the rate constants k 1 and k 2
The results from the curve fitting analysis are shown in Figure 51-Figure 56. In each figure, the measured data are shown together with the curves simulated by the SFO model and the DFOP model (provided fits were obtained). The parameters a, b, k 1 and k 2 for each model are shown in Appendix 10 together with the correlation coefficient for each curve. A more comprehensive evaluation of the goodness of fit including an evaluation of standard errors for each parameter and further testing will be presented in future publications. With bromoxynil in Slaeggerup subsoil (Figure 51), dimethoate in Estrup and Slaeggerup subsoil (Figure 52) and ioxynil in Faardrup subsoil (Figure 55), the DFOP model was not notably better than the SFO model. With dimethoate in Estrup plough layer, no fit could be obtained due to the low concentration detected on day 0. With dimethoate in Slaeggerup plough layer, only the SFO model (Figure 52) could be used. With the remainder of the plough layer samples the degradation processes were best described by the DFOP model, while the SFO model provided a less satisfactory description of the degradation processes.
Figure 51. Degradation of bromoxynil in the plough layer (0-20 cm) and the subsoil (80-100 cm) from Slaeggerup. No corresponding figure is shown for Faardrup soil as insufficient data are available. Closed circles indicate the experimental data, while solid lines indicate the fitted curve for a 1 st order model (black) and a two-compartment 1 st + 1 st order model (red).
Figure 52. Degradation of dimethoate in the plough layer (0-20 cm) and the subsoil (80-100 cm) from Estrup and Slaeggerup. Closed circles indicate the experimental data, while solid lines indicate the fitted curve for a 1 st order model (black) and a two-compartment 1 st + 1 st order model (red).
Figure 53. Degradation of fenpropimorph in the plough layer (0-20 cm) and the subsoil (80-100 cm) from Tylstrup, Jyndevad and Faardrup. Closed circles indicate the experimental data, while solid lines indicate the fitted curve for a 1 st order model (black) and a two-compartment 1 st + 1 st order model (red).
Figure 54. Degradation of flamprop-M-isopropyl in the plough layer (0-20 cm) from Estrup and Slaeggerup. Closed circles indicate the experimental data, while solid lines indicate the fitted curve for a 1 st order model (black) and a two-compartment 1 st + 1 st order model (red).
Figure 55. Degradation of ioxynil in the plough layer (0-20 cm) and the subsoil (80-100 cm) from Faardrup and in the plough layer (0-20 cm) from Slaeggerup. Closed circles indicate the experimental data, while solid lines indicate the fitted curve for a 1 st order model (black) and a two-compartment 1 st + 1 st order model (red).
Figure 56. Degradation of propiconazole in the plough layer (0-20 cm) and the subsoil (80-100 cm) from Tylstrup, Jyndevad, Faardrup and Slaeggerup. Closed circles indicate the experimental data, while solid lines indicate the fitted curve for a 1 st order model (black) and a two-compartment 1 st + 1 st order model (red).
With the DFOP model, half-lives can only be obtained if they lie on the curve or can be calculated through an iterative process. The SFO equation can be solved analytically such that DT 50 = ln2/k, which means that the half-life will be the same irrespective of the stage in the process. In contrast, half-life determined with the DFOP model increases with time because the rate constant for the second compartment becomes increasingly dominant with time. The error that is introduced by using a simple 1 st order model for the calculation of half-life varies among the soils and compounds. Guidelines for the evaluation of these errors will be established in the above-mentioned FOCUS group. In similar, long-term degradation studies, the biological activity after 200 days of incubation of selected samples was controlled to assure that the decline in degradation was not due to a substantial decrease in biological activity.
If the pesticides in the second compartment were only mobilized due to the use of a strong extraction technique, they would never have been available for use by the microorganisms or to leach to the groundwater. As a consequence, the use of a simple 1 st order model would overestimate pesticide persistence. If, on the other hand, the pesticides were truly available to the soil microorganisms or to leach to the groundwater, then the use of a simple 1 st order process would underestimate pesticide persistence. For example, 3 half-lives for fenpropimorph in Faardrup soil is only 45 days calculated using the 1 st order model, while the twocompartment 1 st + 1 st order model more correctly yields 62 days (4+22+36), as indicated in Table 21.
The distribution of pesticides between compartments is expected to be chiefly governed by the sorptive capacity of the soil for each compound. At the same time, microbial activity is expected to govern the rate of degradation in the soil water compartment. However, it can be expected that the degradation rate in the soil water compartment will also influence the distribution. Thus complex patterns are expected to govern the overall process. With a view to elucidating all the relationships between compartments and degradation rates, modelling studies are currently being performed. The results will be presented in subsequent reports. Further modelling will take into consideration the guidelines being elaborated by the EU FOCUS work group on degradation kinetics.
Sorption and degradation parameters were determined on various combinations of pesticides and soil types representative of the PLAP. The results confirmed the low microbial activity, sorption, and degradation rates generally found in subsoil. Both degradation rates and sorption differed markedly between soils, thus stressing the importance of having sitespecific parameters when modelling the leaching of pesticides.
Compared with published values, sorption determined in the present study was higher for dimethoate and flamprop-M-isopropyl, lower for bromoxynil and ioxynil and within the published range for fenpropimorph, metamitron and propiconazole. The DT 50 of ioxynil and bromoxynil were remarkably low, ranging from <1 day in the plough layer and from <5 to 12 days in the subsoil.
In some cases the degradation rate was better described by a two-compartment 1 st + 1 st order model than by the usual 1 st order model. As degradation often involves one initial fast degradation rate with a short half-life followed by slower degradation rates with longer halflives, an error is introduced if the simple 1 st order half-life is used in the evaluation of pesticide persistence. Further analysis of the significance of the introduced error for risk assessment of pesticide leaching is thus required.