Environ. Sci. Technol. 2009 43, 9223–9229
Assessing Persistence and Long-Range Transport Potential of Current-Use Pesticides ¨ RG KLASMEIER, MICHAEL MATTHIES,* JO ANDREAS BEYER,† AND CHRISTIAN EHLING Institute of Environmental Systems Research, University Osnabru ¨ ck, D-49069 Osnabru ¨ ck, Germany
Received March 12, 2009. Revised manuscript received October 22, 2009. Accepted October 26, 2009.
Despite the fact that current-use pesticides (CUP) have different chemical properties to first-generation organochlorine products, the long-term and long-range environmental behavior of these chemicals is still unclear. Data for 45 active ingredients of CUPs were collected, most of which originate from the results of simulation tests submitted for authorization. According to the Stockholm Convention on persistent organic pollutants (POPs), two of the 45 CUPs exceed both screening level criteria for persistence and long-range transport potential (LRTP). Thirteen CUPs meet the persistence criterion and just one for LRTP. This classification is compared to the reference chemicals approach using overall persistence (Pov) and characteristic travel distance (CTD) calculated with a multimedia model. Although none of the 45 CUP have a CTD above the LRTP boundary line, three of them exceed the overall persistence criterion derived from legacy POPs for classification. Nineteen CUPs are transported over longer distances in water than in air. For such polar substances a LRTP boundary has yet to be defined. We recommend the multimedia model modeling approach to calculate Pov and LRTP as a second tier in persistence and LRTP assessment.
Introduction It is well recognized that the exposure and effect assessment of pesticides should not be restricted to the target area and its immediate neighborhood because this does not sufficiently cover possible hazards associated with the use of these products (1). Prolonged use of pesticides containing persistent active ingredients can lead to elevated concentrations due to accumulation in the environment and long-term exposure to nontarget organisms. Long-range transport (LRT) in air and water can result in the exposure of remote and particularly vulnerable ecosystems such as the arctic (http:// www.amap.no/) or pristine mountain regions (2). The ecosystem structure and the sensitivity of organisms in remote regions may deviate significantly from conditions in the application region. Hence international protocols such as the Stockholm Convention on Persistent Organic Pollutants (POPs) (3) and the UNECE Convention on Long-Range Transboundary Air Pollution (4) stipulate that anthropogenic chemicals may not enter remote regions at all. According to the new European pesticide regulation (5), an active sub* Corresponding author e-mail:
[email protected]. † Current address: Biotechnology Center (BIOTEC), Technical University Dresden, D-01062 Dresden, Germany. 10.1021/es900773u CCC: $40.75
Published on Web 11/09/2009
2009 American Chemical Society
stance shall only be approved if it is not considered to be a persistent organic pollutant. The criteria for persistence and potential for LRT are consistent with those given in the Stockholm Convention, with the exception of requiring DT50 rather than half-lives. DT50 is the time it takes for 50% of the initial mass or concentration to disappear from a compartment by dissipation processes (6). In order to assess the LRT potential, use should be made of monitoring data or environmental fate properties and/or model results that demonstrate the active substance’s potential for long-range environmental transport through air, water, or migratory species (3, 5). Consequently, current LRTP criteria already consider dependence on environmental distribution, whereas those for persistence assume that this is a sole substance property independent of environmental conditions. Recent publications suggest relatively simple measures for environmental persistence and LRTP based on multimedia box models (for an overview, see ref 7). These measures have the advantage of being independent of emission data, but include intermedia transport and partitioning among the environmental compartments. For this reason, their use enables the comparison of persistence and LRT behavior of chemicals according to their chemical properties. While there is consensus on the interpretation of overall persistence (Pov) as a measure of persistence (8–10), different approaches have been proposed to assess LRTP. However, it has been shown that most concepts yield similar rankings (7, 11). Here, the characteristic travel distance (CTD) is used as a measure of LRTP (12–14). Both Pov and CTD have already been used for the relative assessment of a limited number of chemicals. Most publications related to this field address halogenated aromatic compounds, including organochlorine pesticides (15–18). Current-use pesticides (CUPs) are more hydrophilic than many of the old organochlorine pesticides such as DDT. They tend to be washed out from the atmosphere by rain, and can also be transported with water currents if sufficiently persistent. In dry seasons, they are able to travel over much longer distances in the atmosphere than in wet seasons. Muir et al. (19) compared measurements of eight CUPs in Canadian Lakes along a south-north transect with LRT estimated using three generic models. The observations and model results suggest that, under the conditions prevailing in south-central Canada (relatively high latitude, low precipitation rates), many CUPs are able to undergo regional-scale atmospheric transport and reach lakes outside areas of agricultural application. This paper presents an approach regarding how to assess CUPs with respect to persistence and LRTP. We selected 45 CUPs for which data were already available from registration dossiers (see the Supporting Information (SI)). Estimation of half-lives in air, water, sediment, and soil is an important prerequisite for model simulations. Here, data from simulation tests in water, sediment, and soil, available from dossiers requested as information for the authorization of pesticides, have been converted into half-lives. The sensitivity of Pov and CTD to the selected half-lives in water is analyzed exemplary for four selected CUPs. For the 45 CUPs, we separately calculated Pov for emissions to air, water, and soil, and LRTP in air and water using a regional scale steady-state model. The objectives of the paper were to classify CUPs according to their maximum Pov and LRTP using boundaries derived from legacy POPs (9) and to compare the classification with the approach adopted by the Stockholm Convention on POPs (3) and the new EU pesticide regulation (5). VOL. 43, NO. 24, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
9223
Materials and Methods Overall Persistence and Characteristic Travel Distance. Both Pov and CTD metrics take into account the exchange of organic chemicals between the environmental media. The concept for calculating Pov and CTD assumes a steady-state situation, i.e., the input into the system equals the output (7, 14, 20). Pov in such a system is the ratio of total mass in the system mtot divided by the total loss (21). Since steady state is assumed, the total output can be replaced by the total input Itot, yielding POV )
mtot
∑mk
i i
)
mtot Itot
(1)
where mi and ki are the mass and the elimination rate constant in compartment i, respectively. The rate constants are lumped over all transformation and removal processes in the given compartment, assuming that all processes are of (pseudo) first order. A multimedia model is required to calculate the total mass in the system. This mass depends on the degradability of the given substance as well as the mass fractions present in the various media. It is commonly accepted to account only for permanent losses when calculating the overall persistence (7, 14, 16). Advective losses caused by wind or water currents are not considered, because they simply move the chemical from one site to another. On the other hand, permanent loss processes such as transport to the stratosphere, leaching or sediment burial can be included. The overall persistence can therefore be interpreted as the average residence time of a molecule in the system, excluding transport by wind and water currents. The characteristic travel distance is defined as the distance from a source at which the initial concentration drops to 1/e (or 37%) of its initial value (12). The characteristic travel distance in air (CTDA) is proportional to the residence time in air τA (13) CTDA ) u·τA ) u·
mA Itot
(2)
where u is the wind speed, which is assumed to be constant. Using the same approach, the characteristic travel distance in water CTDW can be calculated, taking into account the flow velocity and the residence time in water (22). No CTD has yet been defined for soil. Model. We use the ELPOS 2.2 model (Environmental LongRange Transport and Persistence of Organic Substances) (ref 14, http://www.usf.uos.de/usf/arbeitsgruppen/ASW/ ELPOS.en.html) to calculate Pov and CTD. ELPOS is a multimedia model that represents a simplified version of the regional scale model in EUSES-SimpleBox, a legally accepted tool for the assessment of new and existing substances in the European Union (23, 24). Since EUSESSimpleBox is a well-studied and evaluated model, its capabilities and limitations are comprehensively understood (25–28). The model includes the media air, water, and sediment, and three soil compartments for agricultural, natural, and urban/industrial soil. It is a steady-state fugacity model of the level III type, assuming constant environmental conditions and first-order processes. ELPOS was included in a model comparison study of nine multimedia models and their application for the screening assessment of LRTP and persistence (7, 9). Details on the differences between EUSESSimpleBox and ELPOS are given elsewhere (14). Pov and CTD strongly depend on the environmental medium in which the substance is emitted. ELPOS calculates Pov separately for continuous emissions into air, water, or soil and CTD for emissions into air or water, respectively. LRTP/Pov classification. Scheringer (15) introduced the plot of LRTP against Pov to display the relationship between 9224
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 24, 2009
the spatial and temporal properties of various substances. This approach was used by Klasmeier et al. (9) to classify substances as persistent and/or prone to long-range transport. They used the temporal and spatial remote states as a measure of persistence and LRTP, and defined boundaries for classification using a set of reference chemicals with wellknown environmental behavior. The remote states can be approximated by the maximum Pov and LRTP, resulting from emissions into the three compartments air, water, and soil (29, 30). These maximum Pov and CTD values were used for ranking and classification purposes. Using this approach, it is assumed that the objective of the assessment is to avoid chemicals similar to known POPs, such as PCBs. This “reference chemicals approach” is used to classify CUPs by applying the Pov and LRTP boundaries of 230 days and 5200 km derived for ELPOS (9). Physical-Chemical Substance Data. Physical-chemical data of 45 CUPs, measured at room temperature (20-25 °C), were obtained from registration dossiers, pesticide databases and the literature (31–33) (SI Table S1) and used without temperature correction. Measured organic carbon-water partition coefficients (KOC) were taken from publicly accessible pesticide databases and used to calculate soil-water and sediment-water distribution, assuming 2% organic carbon in soil and 4% in sediment. Partitioning between air and water is described by the dimensionless air-water partition coefficient (KAW), estimated from the vapor pressure of the solid state and the solid aqueous solubility. SI Table S1 and Figure S1 show that many CUPs exhibit rather low KAW values, making them prone to transport in water, e.g., 41 of the selected CUPs have a KAW below 10-4. Gas-particle partitioning in air is estimated by the KOA approach. For polar compounds, the polyparameter linear free-energy relationship (pp-LFER) approach may be superior if measured data for the solvation parameters are available (34, 35). Go¨tz et al. (36) compared the pp-LFER with the KOA approach for modeling the gas-particle partitioning of semivolatile organic chemicals, and included several polar pesticides. Using three different multimedia models, they also investigated how Pov and LRTP depend on gas-particle partitioning, and come to the conclusion that either of the gas-particle partitioning approaches can be used in generic models to screen chemicals for their persistence and LRTP. Half-Lives in Air. Reaction with hydroxyl radicals is the most important degradation process for pesticides and other organic compounds in air (37). Measured rate constants for degradation in air have not been reported for any of the 45 selected CUPs (38). We assumed that particle-bound chemicals are shielded from OH-radicals and are thus not degraded (21). Go¨tz et al. (36) compared alternative scenarios with and without the degradation of particle-bound compounds. They were able to show that the modeled concentrations of three exemplary pesticides in the arctic air are within the variability of measured concentrations. Most CUPs are in the gaseous phase and not particle-bound, as Go¨tz et al. (40) have already shown. Thus, our assumption only has a minor influence on the atmospheric residence time. Second-order reaction rate constants with OH-radicals in the gas-phase were calculated from the molecular structure using the AOPWIN computer program (39), and multiplied by the average atmospheric OH-radical concentration of 5 × 105 molecules cm-3 (38) to yield atmospheric half-lives. Half-Lives in Water. None of the 45 CUPs are readily biodegradable. One compound (dicofol) is susceptible to hydrolytic degradation (see SI Table S1), significantly reducing its half-life in water. Results from laboratory water-sediment simulation tests according to official guideline protocols (41, 42) were summarized by Sivapragasam (32) from dossiers submitted for authorization (SI Table S2). DT50 and DT90 values (disappearance or dissipation times 50% and 90%)
were reported for water, sediment, and the total system, usually for a river and a lake sample. A mass balance evaluation of water-simulation tests revealed that DT-values in water are inappropriate descriptors of degradation in water because the observed concentration decrease is almost exclusively due to transport from water into sediment (43). WethereforediscardedDTvaluesinwaterfromwater-sediment simulation tests and estimated biodegradation using BIOWIN4 (primary survey model) (39). The model output was converted into primary degradation half-lives in water using the regression equation derived by Arnot et al. (44) (SI Table S1). In an alternative scenario with four selected CUPs, we demonstrate the possible error introduced into Pov and LRTP assessment when using median DT50water values (SI Table S2) as degradation half-lives in water. Half-Lives in Sediment. Estimates of DT50 values in sediment are available for a subset of six substances from water-sediment simulation tests, which were used as halflives (SI Table S2). For 30 CUPs, only DT50 values of the total system and not for sediment were reported in registration dossiers. In this case, the half-life in sediment (HLsed) was estimated from the median DT50 of the total system under the following assumptions: (i) Dissipation in the total system was exclusively due to disappearance in the sediment and not in the water phase. (ii) The mass distribution ratio was almost constant over time after a rapid initial transport from water into sediment. Sediment half-lives are then related to the median DT50tot of the total system as follows: HLsed ) DT50tot·
msed mtot
(3)
The mass fraction in sediment is determined by the volumes of the water and sediment phase (Vw, Vsed) in the test vessel, the sediment sorption coefficient Kd, porosity ε, and the material density FF of the sediment: msed ) mtot
Vsed Vsed ·(1 - ε)·FF + ·ε VW VW Vsed Vsed 1 + Kd· ·(1 - ε)·FF + ·ε VW VW Kd·
(4)
(See SI for the derivation of eq 4.) For 13 further substances, neither DT50 in sediment nor in the total system were provided. Thus, DT50tot values of these CUPs were taken from the FOOTPRINT Pesticide Properties DataBase (33), and half-lives in sediment were again estimated using eqs 3 and 4. Half-Lives in Soil. DT50 and DT90 values estimated from simulation tests in laboratory soil were provided for most of the pesticides (SI Table S3). Data from field tests were only available for a subset, and were not used further. The procedure followed to determine the degradation rates in soil is described in the SI. The resulting half-lives in soil are portrayed in SI Table S1. It must be borne in mind that simulation tests in soil yield dissipation times rather than degradation half-lives. Substances disappear in soil not only due to degradation but also through the formation of irreversibly bound residues, which are often falsely regarded as being identical to nonextractable residues (6). Matthies et al. (45) were able to distinguish degradation from dissipation using a kinetic model to fit the experimental data of a few labeled CUPs from soil simulation tests under aerobic conditions. They showed that DegT50, i.e., the time needed for 50% true primary degradation, is usually greater than DT50. As long as the underlying mechanisms of the nonextractable residue formation are not known, degradation and dissipation should not be considered to be identical (46). This means that the DT values constitute a lower boundary for the degradation half-lives in soil.
Consideration of Intermittent Rainfall. Hydrophilic substances have an increased potential to be washed out by precipitation, and thus exhibit a lower atmospheric residence time and LRTP (40). Hertwich (47) investigated the effect of assuming continuous rain, and came to the conclusion that only chemicals with a low KAW (
