Volume 19, Issue 6 p. 1544-1554
Environmental Policy & Regulation
Open Access

Environmental exposure assessment of co-formulants in plant protection products under REACH

Claire McMillan

Claire McMillan

Cambridge Environmental Assessments, Cambridge, UK

Contribution: Conceptualization, Data curation, Methodology, Project administration, Software, Validation, Writing - original draft, Writing - review & editing

Search for more papers by this author
Sébastien Bonifay

Sébastien Bonifay

Corteva Agriscience, Production Agriscience Belgium BVBA, Brussels, Belgium

Contribution: Methodology, Validation, Writing - review & editing

Search for more papers by this author
Christopher Dobe

Corresponding Author

Christopher Dobe

Syngenta Crop Protection AG, Basel, Switzerland

Address correspondence to [email protected]

Contribution: Conceptualization, Methodology, Project administration, Supervision, Writing - original draft, Writing - review & editing

Search for more papers by this author
Ralph Fliege

Ralph Fliege

Bayer AG, Crop Science Division, Monheim, Germany

Contribution: Methodology, Validation, Writing - review & editing

Search for more papers by this author
Joachim D. Krass

Joachim D. Krass

BASF SE, Ludwigshafen, Germany

Contribution: Methodology, Validation, Writing - review & editing

Search for more papers by this author
Adrian Terry

Adrian Terry

Cambridge Environmental Assessments, Cambridge, UK

Contribution: Conceptualization, Methodology, Writing - review & editing

Search for more papers by this author
Matthias Wormuth

Matthias Wormuth

Syngenta Crop Protection AG, Basel, Switzerland

Contribution: Conceptualization, Methodology, Project administration, Validation, Writing - original draft, Writing - review & editing

Search for more papers by this author
First published: 02 March 2023

Abstract

It is a regulatory requirement to assess co-formulants in plant protection products (PPP) under the European Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) legislation. The standard environmental exposure assessment framework for chemicals under REACH is a multicompartmental mass-balanced model and, at the local scale, is designed for use with urban (wide dispersive) or industrial (point source) emissions. However, the environmental release of co-formulants used in PPP is to agricultural soil and indirectly to waterbodies adjacent to a field and, for sprayed products, to the air. The Local Environment Tool (LET) has been developed to assess these specific emission pathways for co-formulants in a local-scale REACH exposure assessment, based on standard approaches and models used for PPP. As such, it closes a gap between the standard REACH exposure model's scope and REACH requirements to assess co-formulants in PPP. When combined with the output of the standard REACH exposure model, the LET includes an estimate of the contribution from other nonagricultural background sources of the same substance. The LET is an improvement over the use of higher tier PPP models for screening purposes because it provides a simple standardized exposure scenario. A set of predefined and conservatively selected inputs allows a REACH registrant to conduct an assessment without requiring detailed knowledge of PPP risk assessment methods or typical conditions of use. The benefit to the co-formulant downstream user (formulators) is a standardized and consistent approach to co-formulant assessment, with meaningful and readily interpretable conditions of use. The LET can serve as an example to other sectors of how to address possible gaps in the environmental exposure assessment by combining a customized local-scale exposure model with the standard REACH models. A detailed conceptual explanation of the LET model is provided here together with a discussion on its use in a regulatory context. Integr Environ Assess Manag 2023;19:1544–1554. © 2023 BASF SE, Bayer AG et al. Integrated Environmental Assessment and Management published by Wiley Periodicals LLC on behalf of Society of Environmental Toxicology & Chemistry (SETAC)

INTRODUCTION

The European Regulation (EC) No. 1907/2006 (Registration, Evaluation, Authorisation and Restriction of Chemicals [REACH]) requires registrants to submit a chemical safety report (CSR) for all substances manufactured or imported in quantities greater than 10 tonnes/year. For substances classified as hazardous, and demonstrating effects on the environment, an environmental exposure assessment and risk characterization for all identified uses must be done. The active substance in a plant protection product (PPP) is considered registered if it meets the criteria of REACH article 15(1). This article does not apply to co-formulants that make up the ingredients of a PPP other than the active substance, such as solvents, colorants, antifoams, emulsifiers, fillers, and so on. Tank-mix adjuvants are a special case and consist solely of co-formulants. For brevity, co-formulants are referred to in this article as substances; however, it must be remembered that they are often, in fact, mixtures. CropLife Europe (CLE) has developed a standardized approach to support manufacturers and downstream users of co-formulants to readily assess the environmental and human exposure, find safe conditions of use and risk management measures to control risks adequately, and communicate these along the supply chain (Dobe et al., 20172020; Mostert et al., 2019). This article focuses specifically on the local environmental exposure assessment.

The current environmental risk assessment framework under REACH is based on the previous risk assessment frameworks used in Europe, otherwise known as the EU TGD approach (European Chemicals Agency [ECHA], 2016a; European Chemicals Bureau [ECB], 2003). The underlying multicompartmental or nested multimedia mass balance model was developed originally to assess environmental exposure arising from chemical use at industrial sites (point sources) and from wide-dispersive uses in the catchment of a municipal wastewater treatment plant (RIVM [National Institute of Public Health and the Environment], 19962004a) before being adopted for European regulatory risk assessments (RIVM, 2004b; Vermeire et al., 19972005). Based on the conceptual models, several environmental exposure and risk assessment tools have been developed to conduct assessments for regulatory purposes, including USES (RIVM, 2002), EUSES (RIVM, 2004b; Vermeire et al., 2005), ECETOC TRA (European Centre for Ecotoxicology and Toxicology of Chemicals [ECETOC], 200420122014), and ECHA's Chesar (ECHA, 2017). Petrorisk was developed as an improvement on the standard EU TGD-based models for the assessment of complex substances found in the petrochemical industry (Redman et al., 2014).

Three spatial scales are considered in the EU TGD approach: continental, regional, and local, with the regional scale providing a background concentration to the local scale. A regional predicted environmental concentration (PEC) is estimated for each environmental compartment from the combined annual emissions in kg/year resulting from all uses of a given substance. Local-scale concentrations are estimated based on daily local emissions in kg/day for each individual use of the substance coming either from a point source or wide dispersive use (ECHA, 2016a). The sum of regional PEC and local concentration gives the local PEC for each environmental compartment. Risk characterization compares local or regional PEC against the Predicted No Effect Concentration (PNEC) for that compartment, and the resulting PEC/PNEC ratio is the risk characterization ratio (RCR; ECHA, 2016b).

The EU TGD exposure models are mass balance- (tonnage) based and key assumptions at the local scale are that release to water will be via either an industrial wastewater or municipal sewage treatment plant before release to surface water (e.g., a river). At the regional scale, direct releases to air, water, and soil are considered for both industrial and wide dispersive uses (ECHA, 2016a). However, at the local scale, only indirect releases to agricultural soil (via atmospheric deposition and application of sewage sludge) are considered. Direct releases to agricultural soil (e.g., from PPP uses), in fact, are explicitly stated to be out-of-scope of the EU TGD. As a result, this local exposure assessment approach does not consider substances applied directly onto agricultural soil, direct or indirect emissions to surface water, and the resulting contribution to exposure of humans via the environment and secondary poisoning. This is relevant to substances with wide-dispersive use as co-formulants in PPP, which are expected to be applied in nonurban, rural environments with direct exposure to soil, particularly for granules and treated seeds and with some unintended exposure to freshwater waterbodies via spray drift, runoff, or drainage processes.

The European environmental exposure assessment framework developed for PPP authorization under Regulation (EC) No. 1107/2009 uses models that assess surface water exposure to a given active substance and its metabolites according to the application rate of the representative formulation in kg/ha. The approach is tiered, starting with a very conservative Step 1 assessment and progressing with increased levels of realism through to Step 4. At each step, exposure is modeled for an edge-of-field waterbody with direct input from spray drift and indirect input from drainage and runoff (Forum for the Co-ordination of Pesticide Fate Models and their Use [FOCUS], 2001). Step 2 provides a reasonable and REACH-compatible level of complexity (data input requirements), calculating spray drift loadings according to the number of applications, and runoff and/or drainage loadings according to season of application, EU geographical region (North or South), and crop interception. Steps 3 and 4 apply a suite of models and further refinements that require more data and use more details than are normally available for substances to be registered under REACH. In contrast, the calculation of soil exposure in the PPP framework is a relatively simple approach that considers application rate, crop interception, and mixing depth (FOCUS, 1997).

A detailed explanation of the conceptual model for the Local Environment Tool (LET; currently available as version 4.0) is provided here together with a discussion of its use in a regulatory context. Further details on the model are available in the Supporting Information and LET model documentation.

METHOD

The approach developed for the sector-specific assessment of environmental exposure to co-formulants consists of two components. First, integration of environmental emissions of a co-formulant in the regional exposure assessment conducted with the existing nested multimedia fate models is achieved through use of Specific Environmental Release Categories (SpERCs; Ahrens et al., 2017; Reihlen et al., 2016; Sättler et al., 2012), and thus keeps the advantages of a tonnage-based approach when considering the big-picture use of commodity chemicals. Use of the CLE (formerly known as ECPA—European Crop Protection Association) SpERCs maintains this link to the rest of the REACH environmental risk assessment (Dobe et al., 2020). The second component, presented in this article, effectively constructs a use-specific, stand-alone replacement for the local-scale assessment, consequently named the CLE LET, and is listed in ECHA guidance R.16 as a tool for the environmental exposure assessment for co-formulants (ECHA, 2016a). Regional PECs are imported from an existing EU TGD model using the CLE SpERCs (e.g., Chesar, ECETOC TRA, EUSES). Local PECs in the LET are calculated based substantially on existing methodologies. The FOCUS Step 2 calculator was used as the basis to assess surface water and sediment exposure. A hybrid PPP–EU TGD approach was used to assess soil and secondary poisoning exposure. The EU TGD approach to assess exposure of humans via the environment was modified with a new module for the prediction of co-formulant residues in treated crops (Figure 1).

Details are in the caption following the image
Schematic summary of the local scale assessment approach developed for co-formulants used in plant protection products and assessed under REACH

Four generic exposure scenarios describing professional and amateur use of substances as a co-formulant in PPP were developed (Dobe et al., 2017); however, for environmental exposure, only two reasonable worst-case scenarios need be differentiated: the spraying of diluted PPP or the direct application of granular products or treated seeds to soil (Dobe et al., 2020).

Soil model

The amount of a co-formulant reaching the soil surface depends on the application rate, application method (spraying, granules, treated seeds), vapor pressure, and crop interception (the relative portion of applied product retained on plant surfaces). Where the co-formulant is included in sprayed formulations, in alignment with the SpERC 8d.2.4 (Table 1), the initial application rate that reaches the soil can be reduced significantly because of volatilization from spray droplets (Delmaar & Bremmer, 2009; FOCUS, 2008; RIVM, 1995) and from surfaces within the first 24 h (Guth et al., 2004; RIVM, 1995; Van Wesenbeeck et al., 2008). A conservative vapor-pressure threshold ≥0.01 Pa is set, above which rapid loss caused by volatilization is assumed. It should be noted that, before Version 4 of the LET and SpERCs, the initial application rate reaching the soil was reduced incrementally for substances with a vapor pressure greater than 1 × 10−5 Pa. No vapor-pressure dependence on the application rate reaching the soil is foreseen for granules or treated seeds, due to the likelihood of burial (Table 1). No reduction in the release factor to soil due to crop interception is taken as default worst case for all application types.

Table 1. Release factors for CropLife Europe SpERCs Version 4, corresponding to default release values implemented in the LET version 4.0
Vapor pressure (Pa) Fair Fsoil Fsurface water
Release factors spray applications (SpERC 8d.2.4), including volatile solvents with vapor pressure ≥0.01 Pa used in liquid seed treatment formulationsa
≥0.01 1 0 0.002b
0.001–<0.01 0.5 1 0.002b
0.0001–<0.001 0.2 1 0.002b
0.00001–<0.0001 0.1 1 0.002b
<0.00001 0.01 1 0.002b
Release factors for application as granules or treated seeds (SpERC 8d.1.4)
≥0.01 0a 1 0
<0.01 0 1 0
  • Abbreviations: LET, Local Environment Tool; SpERC, Specific Environmental Release Category.
  • a Volatile solvents with vapor pressure ≥0.01 Pa in liquid seed treatments are lost on coating before field application of the dried treated seed. Unlike the SpERC used at the regional scale in mass balance models, volatile solvents in liquid seed treatments should not use the spray release factors in the LET (emission to air is spatially separated from the agricultural field).
  • b Rounded value corresponds to the default 15.7% drift rate.

Once a single application of co-formulant reaches the soil surface, it is assumed to be homogeneously mixed with an upper layer of soil. The default mixing depth for agricultural soil (plowed) in the EU TGD framework is 0.20 m (ECHA, 2016a); however, this was not considered conservative for a co-formulant applied directly to untilled soil (e.g., orchards). The default mixing depth was set at 0.05 m, which is the value used for PPP risk assessments (FOCUS, 1997).

The local concentration in soil (Clocalsoil) is calculated as a 30-day time-weighted average (TWA), with dissipation through biodegradation, volatilization, and leaching included, in line with the EU TGD approach. The concentration of a substance in soil at the start of the application is calculated by assuming its application over the previous 10 consecutive years, following the same approach used for sewage sludge application on agricultural soil (ECHA, 2016a). The PEClocalsoil is calculated as the sum of Clocalsoil and PECregionalagricultural soil. This deviates from the EU TGD approach, which instead uses the PECregionalnatural soil as the background concentration, because this only receives input from air deposition and does not lead to double counting of sewage sludge application (i.e., at both the local and regional scale). However, since the co-formulant use does not result in release to a sewage treatment plant under regular conditions of use because PPP labels prohibit disposal via the public sewage system, the concentration in sludge arising from the local scale calculation for the co-formulant use will always be zero. Therefore, in the LET, the PECregionalagricultural soil was selected to ensure that contributions via sewage sludge arising from nonagricultural uses of the same substance are considered. The equations used in the soil model are reported in the Supporting Information.

The new PERSAM (Persistence in Soil Analytical Model) tool has been developed to predict concentrations of active substances in soil (VITO NV, 2019) under the EU PPP regulation. This new model had not yet been adopted for regulatory use at the time of writing this article, and consequently it has not been included in the latest version of LET. However, it will be investigated if a modification of the LET is required in the future to maintain alignment with PERSAM and the approach to assess PPPs in the EU.

Surface water and sediment model

The unintended direct emissions to an edge-of-field waterbody depend on the application rate, the application method, and the crop type. A spray drift value of 15.7% of the application rate is used as default for spray applications, in alignment with the value used to derive the SpERC 8d.2.4 release factor (Table 1). This corresponds to the regulatory accepted 90th percentile spray drift values determined for one application and a 3-m buffer zone for citrus, olives, and late applications to pome and/or stone fruit (FOCUS, 20012003) and represents orchard scenarios where high spray drift is expected. No spray drift is considered for applications of granular formulations or treated seeds (Table 1).

The surface water and sediment model from FOCUS Step 2 was implemented (FOCUS, 20012003), which provides a greater degree of realism than at FOCUS Step 1 (particularly as used for refinement, modeling loadings from spray drift and runoff and/or drainage separately). It also allows for a simple spreadsheet implementation without the use of a suite of dedicated models and consideration of various emission and use scenarios, as is required for FOCUS surface water Steps 3 or 4. The Step 2 approach assumes a 0.1 ha waterbody next to a 1 ha field from which it receives runoff or drainage water (a field:water surface area ratio of 10). The waterbody in the standardized scenario has a depth of 30 cm (volume of 3 × 105 L) overlying sediment of 5 cm depth (sediment density of 0.8 g/cm3 and an organic carbon content of 5%). The sediment properties represent relatively vulnerable sediment layers that may be present in ditches adjacent to agricultural areas. The input from spray drift to the edge-of-field waterbody is evaluated as a single loading on the day of application.

The co-formulant soil loading available for runoff, erosion, and/or drainage will be a function of the amount of co-formulant in the soil, geographical region, and season of application. In FOCUS Step 2, runoff, erosion, and/or drainage loadings are evaluated after a heavy rainfall event four days after application. Consequently, the calculated co-formulant concentration will be a function of application type (spray or granular solid), vapor pressure (if spray application), crop interception, and degradation in soil until the rainfall event. Degradation in soil is assumed to follow first-order kinetics. Other dissipation processes such as leaching and volatilization are not evaluated before the runoff event. The worst case for runoff was set at 5% of soil loading, corresponding to the most conservative estimate in the FOCUS scenario for Northern Europe during October and February. The co-formulant entering the waterbody via runoff can enter in either the water phase or sediment phase, which depends on soil adsorption (KOC) of the co-formulant (FOCUS, 2003).

Spray drift input enters the surface water on the day of application (Day 0) without any distribution to sediment. Distribution of the co-formulant between the sediment and water layers is calculated at the end of Day 0 (i.e., the day of application and spray drift entry) with surface water distribution assumed to be across two theoretical compartments, one available and one unavailable for sorption to sediment (FOCUS, 2003). The fraction available for sorption before the runoff event is assumed to be two-thirds of the total mass in surface water, which is the default used in Step 2 assessments (FOCUS, 2001). After Day 0, daily concentrations are calculated for surface water and sediment considering first-order kinetics degradation and the initial input (before surface water and/or sediment distribution) from any drift and runoff and/or drainage events, which occur four days after application. Instantaneous loading to the system and immediate distribution between water and sediment after the runoff or drainage event is assumed, with full equilibrium established within 24 h. Clocalsurface water and Clocalsediment are selected from the maximum calculated daily concentrations, respectively. These are added to PECregionalsurface water and PECregionalsediment to calculate PEClocalsurface water and PEClocalsediment, respectively. The equations used in the surface water and sediment model are reported in the Supporting Information.

Air model

Version 4 of the LET introduced a calculation for Clocalair in support of a local-scale exposure of humans via the environment assessment based on a model in USES version 4.0 (RIVM, 2002). The concentration downwind from a 1 ha agricultural field treated with a PPP is calculated with an atmospheric plume model, the Dutch version of the American PAL model (Petersen, 1978). The default parameters for the model are taken from USES 4.0 and include inter alia the values used to derive the CLE SpERC release factors to air (Table 1). The emission of a co-formulant from the treated field is estimated from its application rate and the release factor. Clocalair is combined with PECregionalair to obtain the PEClocalair and daily inhalation exposure to the PEClocalair of a co-formulant is assumed, that is, no dilution factor is considered. The equations used in the air model are reported in the Supporting Information.

Marine water and marine water sediment model

The local concentrations for marine water and marine-water sediment were calculated by applying a dilution factor of 10 to the local concentrations calculated for freshwater and sediment, in line with the EU TGD practice. The local concentrations for marine water and marine water sediment are added to the appropriate regional PECs to calculate local marine PECs.

Secondary poisoning model

This assessment is required if the substance can bioaccumulate, has low biodegradability, and can cause toxic effects if accumulated (ECHA, 2016a). An assessment of secondary poisoning for terrestrial predators (earthworm eating), aquatic predators (fish eating and marine fish eating), and marine top predators was implemented according to the EU TGD framework. The concentration in food of the freshwater aquatic predator is calculated as a function of the PEClocalsurface water, bioconcentration factor (BCF) for fish (representing uptake from the aqueous phase), and biomagnification factor (BMF; representing bioaccumulation from food consumed by the fish; ECHA, 2016a). It is expected that the foraging area of a freshwater aquatic predator is larger than an edge-of-field waterbody; therefore, it is assumed that 50% of a predator's diet comes from the local environment and 50% from the regional area in alignment with the EU TGD approach (ECB, 2003). The PEClocalsurface water used to calculate the concentration in the local diet of a predator is based on a 21-day TWA local concentration and the regional PEC surface water (dissolved). The annual average was not considered practical to derive here without arbitrary assumptions on the emission rate (kg/day); for example, depending on tonnage, point source, or area emitted to, and so forth. The 21-day TWA was selected in line with the aquatic secondary poisoning assessments for PPP (EFSA, 2009). The same approach was adopted for marine (fish eating) predators but using PEClocalmarinewater. The secondary poisoning assessment for the marine top predator is a function of the PEClocalmarinewater, BCF for fish, BMF1, and BMF2 (representing biomagnification in the tissue and organs of the predators; ECHA, 2016a). It was assumed that marine top predators prey mainly on organisms from the regional marine environment, with 10% of the diet from the local scale and 90% from the regional environment (ECB, 2003).

The PEC in the food of terrestrial predators is assumed to be the concentration in earthworms after bioaccumulation in worm tissue and adsorption of the co-formulant to soil present in the earthworm's gut (ECHA, 2016a). Bioaccumulation in earthworm tissue is calculated from BCF and soil porewater which, at the local scale, is calculated from the 180-day TWA concentration in soil. This is also used to calculate gut content of the earthworm at the local scale. The PECregionalagricultural soil is used to estimate the background concentration at the local scale rather than the PECregionalnatural soil (used by EU TGD) in accordance with the approach to the calculation of soil concentrations. The terrestrial predator is unlikely to consume 100% of its food from directly exposed agricultural soil. It was therefore assumed that 50% of a predator's diet comes from the local scale and 50% from the regional environment (ECHA, 2016a).

Exposure of humans via the environment

Version 4 of the LET introduced an assessment of exposure of humans via the local environment, and generally followed the EU TGD scheme with regards to exposure pathways and routes, while calculating concentrations by the relevant LET approach. The detailed description of this new approach is beyond the scope of this article. Briefly, for dietary exposure, the EU TGD scope was expanded to cover direct application of co-formulants to food and forage crops to calculate potential maximum residual concentrations in edible crops, using the safe application rate in kg/ha and average yields of crops in the EU in tonnes/ha. The calculations of concentrations in local drinking water, fish, milk, and meat followed the EU TGD approach (ECB, 2003; ECHA, 2016a), using the adaptations for calculation of local PECs of a substance used as a co-formulant in PPP described in previous sections. Where regional PECs are provided as input values to the LET, local PECs for each route are calculated (i.e., the sum of the local concentrations and regional PECs).

Data requirements

Data necessary to run the model were limited to the REACH Annex VIII standard requirements (molecular weight, water solubility, vapor pressure, octanol–water partition coefficient [KOW], biodegradability classification, and freshwater aquatic PNECs). The KOC is also a standard requirement of REACH Annex VIII, and therefore needed as an input parameter. For convenience in nonregulatory contexts, the KOC can be estimated using a Quantitative Structure Activity Relationship (QSAR) based on KOW and chemical class (Sabljic & Güsten, 1995). Methods used in the REACH risk assessment framework to estimate the soil and sediment PNECs and the higher tier endpoints (Annexes IX and X) degradation rate and BCF were also included in the LET. Experimental data for these endpoints can be entered in the tool manually.

It is expected that, for most substances, only screening data on biodegradation (e.g., ready or inherent biodegradability tests) will be available. As a result, the biodegradation rates in soil, surface water, and sediment are inferred based on partitioning behavior and biodegradability screening tests (ECB, 2003; ECHA, 2016a). Experimental degradation rates can be entered manually when such data are available for a substance. Due to the presence of anoxic layers in sediment, the inferred half-life for sediment is assumed to be a factor of 10 higher than in soil (ECHA, 2016a). The maximum inferred DT50 for all compartments was set at 1000 days in line with the approach used in PPP exposure assessment (FOCUS, 20062014).

Where experimental values for the BCF in fish are unavailable or waived for convenience in nonregulatory contexts, these can be predicted within the model from the relationship between KOW and BCF (Veith et al., 1979). Two QSARs are used, one for log KOW < 6 and one for log KOW > 6.

As a minimum, the freshwater aquatic PNEC for a substance is required. In this case, PNEC values for the other compartments (except for secondary poisoning) are calculated using standard assessment factors or the equilibrium partitioning method.

Model implementation

The LET is a Microsoft Excel-based tool programmed using Visual Basic for Applications and freely available for download from the CLE website (https://croplifeeurope.eu/pre-market-resources/reach-in-registration-evaluation-authorisation-and-restriction-of-chemicals/). Outputs from the LET are the maximum safe application rate for a target RCR, as well as local PECs and RCRs for each compartment. The LET can be run in two assessment modes, Default and Refinement. The Default mode was developed as a generic screening approach to co-formulant manufacturers, with reasonable worst-case parameterization, which is aligned either conceptually or directly with the release factors in the CLE SpERCs. Detailed justification for the selection of the default values used in the SpERCs release factors has been given by Dobe et al. (2020). The screening approach does not require any detailed information on end-use product type and application parameters. These parameters (number of applications, crop type, crop interception, incorporation in soil, location, and season of application) were deliberately fixed to represent a realistic default worst case, because making changes to their values without a clear understanding of the use pattern of the PPP formulations could impose arbitrary restrictions on conditions of use and not represent the authorized PPP. The Default scenario is run initially with an application rate of 1 kg/ha, used internally by the tool for a subsequent iterative calculation of the maximum safe application rate (i.e., the maximum dose at which the RCR for the most sensitive environmental compartment is equal to a given target RCR). After completion of the iteration, PECs and RCRs are reported at this application rate, and the most sensitive environmental compartment is indicated. Identification of the maximum application rate per use is thus a key feature of the LET. The maximum application rate is intended to be communicated to the downstream user as one of the conditions of use defining safe use. The target RCR can be varied by the user, depending on the specifics of the substance risk assessment (e.g., toxicity profile, dataset uncertainty, magnitude of the calculated maximum application rate, etc.).

The Refinement Option was implemented to allow downstream users (i.e., formulators) to conduct calculations equivalent to FOCUS surface water Step 2 (FOCUS, 2003), for the purposes of scaling, or downstream user assessment. Scaling in this context means a check by the downstream user if the conditions of use are inside the boundaries of a communicated exposure scenario (ECHA, 2014). Parameters available for modification include the number and interval of applications. In the first instance, multiple applications of a co-formulant can be considered by a downstream user simply by scaling the maximum application rate to 1/(number of applications). Should actual use rates require a slightly less conservative value, the LET refinement options can be used to scale, which considers dissipation processes between the applications, as well as slightly reducing drift rates to avoid cascading conservatism. The crop type can also be modified, which defines the release factor to surface water via spray drift. The crop interception, location, and season of application affect the amount in runoff and/or drainage events. Crop interception and incorporation into soil can also affect soil loading. Vapor-pressure-dependent release factors to air and soil can be modified for borderline cases but would need to be fully justified. The application rate is defined by the user; the PECs and RCRs are calculated for the specified application rate. Use of the Refinement Options is expected only by downstream users (i.e., formulators) who have detailed knowledge of the use patterns of products containing the co-formulant (i.e., the critical Good Agricultural Practice). The use of any of the refinement options offered by the LET would additionally restrict the operational conditions of use of the co-formulant beyond a single maximum application rate. A summary of the relevant conditions of use for Default and Refinement Options is given in Table 2.

Table 2. LET inputs (exposure determinants) and corresponding conditions of use
Assessment type
Condition of use Default Refinement optiona
Application rate (kg/ha) Calculated by LET for a target RCRa Set by user
Number of applications per season 1a,b, a,b Set by user
Application interval (days) n/a Set by user
Application type Spray or granule and/or seed treatmenta Spray or granule and/or seed treatment
Crop Drift rate for reasonable worst-case crop is set (15.7%) Crop set by user (drift ranges between 2.76% and 33.2%)
Soil incorporation Worst case of no incorporation is set (0.05 m mixing depth) Yes, increases mixing depth to 0.2 m
Interception type Worst case of no interception is set Application rate to soil reduced by a factor linked to crop type and stage of growth defined as minimum crop cover, medium crop cover, or full canopy
Geographical region and season of application Worst case of Northern Europe in winter is set Set by user
Release factor soil Defined by SpERC Set by user if it can be justified for the specific end use
Release factor air Defined by SpERC Set by user if it can be justified for the specific end use
  • Abbreviations: LET, Local Environment Tool; RCR, risk characterization ratio; SpERC, Specific Environmental Release Category.
  • a These inputs become conditions of use in the default refinement mode of the tool.
  • b Modifiable by the downstream user using simple linear scaling.

DISCUSSION

The LET was developed to allow a REACH-specific environmental exposure assessment for substances used as co-formulants in PPP to be conducted easily, especially for registrants not having detailed knowledge of the specific conditions of the agricultural use of co-formulants. This use can result in direct exposure of agricultural soil at the local scale, direct and indirect exposure of edge-of-field waterbodies, and dietary exposure of humans via the environment and predators via secondary poisoning caused by direct application of co-formulants to soil or treated crops. These particular exposure routes are outside the scope of the current models used for environmental risk assessment under REACH, which were intended primarily to address industrial and urban sources of exposure. The LET and the “default” assessment scenario have been developed to address this identified gap in the REACH standard models and support registrants of substances used as co-formulants in their obligations under the REACH legislation. This article illustrates how a downstream industry sector has adapted the regulatory assessment framework for industrial chemicals by using an exposure assessment element from another European regulatory domain (i.e., the regulation of PPPs). As part of this approach, the agrochemical industry sector has identified the conditions of use that are key to conservatively assessing environmental exposure to co-formulants for the purpose of registration of a substance under REACH. These are defined default inputs with corresponding phrases for communication to ensure that consistent, and realistic, worst-case environmental assessments can be conducted by registrants. This type of approach draws on expert knowledge of the agrochemical industry and the tools used in the EU regulatory approval of PPP. Appropriate inputs and conditions of use are defined for the environmental risk assessment conducted by upstream users lacking this specific knowledge (i.e., registrants of co-formulants). This enhances the communication with subsequent downstream users who can receive environmental risk assessments and conditions of use that are appropriate to their industry sector, or conduct their own assessment if necessary.

As described above, the environmental exposure routes relevant to co-formulants (i.e., direct exposure of agricultural soil and edge-of-field waterbodies at the local scale) are explicitly outside the scope of the standard EU TGD-based environmental exposure assessment models. The existing models used for PPP authorizations are designed to assess active substances and their metabolites, and thus assume that an extensive dataset for environmental fate is available, as well as a detailed knowledge of a single formulation's use patterns. The LET has been developed in response to the need for a simple screening level tool for the assessment of co-formulants under REACH that addresses out-of-scope exposure routes but could also cover the large variety of PPP and their different use patterns, all in a single standardized assessment using only REACH Annex VIII data (the minimum available when a CSR is also first required). Further constraints on the development of the model were that it must integrate with the rest of the REACH risk assessment done for a given substance and, where possible, make use of existing REACH and PPP exposure models with established regulatory acceptance. The models and approaches to the assessment of PPP constantly evolve, as demonstrated above in the example of the new PERSAM model for the calculation of concentrations in soil. Such new models will be evaluated and considered for implementation in future versions of LET, where technically possible, to ensure continuous alignment with the PPP approaches.

The LET approach requires a regional exposure assessment to be conducted using the CLE SpERCs in an appropriate nested multimedia fate model (e.g., ECETOC TRA, EUSES, Chesar). The regional PECs are then imported into the LET, to conduct a local-scale assessment in accordance with the REACH requirements. This is driven by application rate. In the standard assessment approach, a daily use rate is usually calculated from tonnage and release to the local scale estimated according to a release factor and number of use days. This type of tonnage-based approach is not appropriate to estimating local releases of co-formulants, and the adoption of an application-rate approach is a key difference between the standard environmental risk assessment approach adopted for REACH and the risk assessment framework developed for PPP. Much of the LET development was based on the latter but was further tailored to fit co-formulants. The maximum application rate of a co-formulant (in kg/ha) is communicated to formulators of PPP and can be interpreted easily by them because the approvals of PPP formulations in the EU are also based on application rates. The maximum application rate for a single application per season and/or year is thus reported as the simplest way in which to communicate a multivariate limitation on operational conditions, with the expectation it will be scaled as required. Scaling in the LET is considered to meet the ECHA's requirements for scaling through its use of an invariant target RCR, and all scalable exposure determinants are contained in the algorithms as default values (ECHA, 2014). Restrictions on invariant emission rates for scaling purposes are not relevant in this case because the LET is not a mass balance model.

Co-formulants are primarily included in PPP for a technical function (e.g., solvents) and thus processes such as volatilization during application are potentially more applicable than for the active substance (where volatility is usually not a desired property). Volatility has been considered in the development of vapor-pressure-dependent release factors to air, which are similar to those employed in USES 4.0 and the release factor to soil for volatile co-formulants used in spray applications.

To align the standard PPP approach with the REACH framework, it has been coupled with a regional scale assessment (or background concentration). As typically commodity chemicals, co-formulants are also frequently used in a wide variety of applications other than PPP and, as such, it is important to consider the possible input from these other uses leading to a regional background concentration, for example, by atmospheric deposition to agricultural soil, releases to a waterbody further up the catchment, and application of sewage sludge to agricultural soil.

Furthermore, calculations for marine water and sediment were included to meet the REACH requirements. Although this is not considered a likely target compartment, it nevertheless requires a mandatory risk characterization for the CSR under REACH. The exposure assessment for the marine environment is conservative because the additional dilution factor, compared with the freshwater compartment, is only 10. Further dilution may occur caused by currents resulting from tidal influences where dilution factors of more than 500 have been determined from model simulations (ECHA, 2016a).

One essential outcome of the LET, and an improvement over using existing higher tier PPP models for screening purposes, is to provide a simple, standardized exposure scenario. The LET effectively provides a set of appropriately selected inputs to existing models that allows a registrant to conduct an assessment without detailed knowledge of PPP risk assessment methods or typical conditions of use of PPP under Good Agricultural Practice (e.g., application type, treated crops, timing of application). The benefit to the downstream user (and regulators) is a standardized and consistent approach to co-formulant assessment, with meaningful and readily interpretable conditions of use defining safe use boundaries (i.e., the co-formulant maximum application rate). The LET also provides output that can readily be incorporated into the CSR, and the extended safety data sheet for communication to downstream users.

The default parameter selection and conceptual model implemented in the LET is highly conservative, and it is worth summarizing what this selection means. It assumes that, for surface water, a given substance is applied at the maximum rate cumulating the total seasonal amount into a single application, with high drift (15.7% as in an orchard), in Northern Europe, during winter, and this drift occurs to a nearby 30 cm deep static waterbody. Apart from volatile substances for spray applications, it is assumed that all of the applied substance reaches the soil (i.e., none of the product is actually intercepted by the target crop) and is only mixed into the upper 0.05 m soil layer. The co-formulant in the soil is then available for runoff and/or drainage to the same waterbody (at the conservative rate expected for winter, rather than spring and/or summer). For the calculation of the soil concentration, the same co-formulant is applied at this maximum rate for 10 consecutive years before the application considered in the local scale assessment. The secondary poisoning calculation assumes that 50% of an aquatic predator's diet comes from the same edge-of-field waterbody, at concentrations equivalent to the 21-day TWA aquatic concentration following the maximum safe application rate (and without consideration of any dilution). The more conservative 21-day TWA, compared with the EU TGD's annual average, was selected in line with the aquatic secondary poisoning assessment for PPP (EFSA, 2009). Furthermore, the LET does not consider the use of risk mitigation measures such as spray buffer zones, drift reducing nozzles, or vegetative filter strips that are often adopted to reduce environmental exposure to the active substance included in the PPP. It has been reported that approximately 95% of active substances examined at the European level required some form of mitigation to demonstrate environmental safety (Alix et al., 2017). This suggests that risk mitigation measures play an important role in demonstrating environmentally safe use of PPP, and these would also be expected to mitigate co-formulant exposure, especially to edge-of-field waterbodies. Obviously only a small fraction, if any, of the applied co-formulant tonnage could be expected to meet all of the above conditions. The maximum application rate calculated is thus limited by the above conservatism and therefore considered to be most suitable for the communication of safe conditions of use of a co-formulant under REACH. However, the real-world maximum application rate for a given substance will depend on the technical function (usually limiting the expected concentration in the product, e.g., solvents vs. colorants) and the diversity of products affecting the product application rate, for example, insecticides versus herbicides, cereals versus stone fruit, and so forth.

Version 4.0 of the LET introduces for the first time a novel screening approach, extending the scope of the exposure of humans via the local environment assessment to include direct application of substances to agricultural soil and crops. Sufficient conservatism was built into this new module. The concentration in local air is calculated based on the maximum safe application rate of a co-formulant and is assumed to be constant throughout the year, that is, it reflects daily use of the substance. The concentration in drinking water is the higher of the two concentrations predicted for the ditch adjacent to an agricultural field and the groundwater directly below the treated field, which is the soil porewater concentration in agricultural soil calculated with the simplistic EU TGD model. Very limited dissipation of a co-formulant is assumed in the calculation of residual concentrations in treated crops, which are used in the calculations of dietary exposure. Due to these simplifications, the approach must not be used for active substances in PPP. This new component of the LET will be discussed in detail in a separate publication, which is currently being prepared.

CONCLUSION

The assessment of environmental exposure to substances used as co-formulants in PPP is a requirement under REACH. A novel approach incorporating the standard REACH and the agrochemical risk assessment frameworks was developed resulting in the CLE SpERCs (for regional scale exposure) and the LET (for local scale exposure assessment). The approach adopted in the LET is similar to FOCUS surface water Step 2 for surface water and sediment, with an additional marine assessment based on the standard EU TGD approach, and a combination of FOCUS and REACH procedures for the assessment of soil and secondary poisoning concentrations. A novel screening assessment for exposure of humans via the environment has been introduced in the latest version. The LET addresses exposure routes currently out-of-scope of the models typically used to perform REACH risk assessments. It exemplifies how a gap in the REACH risk assessment methodology can be closed by using elements from another regulatory domain (i.e., the regulation of PPP). It provides a standardized assessment for registrants with limited knowledge of how the co-formulant will be used, and a framework for downstream users potentially requiring a co-formulant risk assessment but unfamiliar with the requirements of REACH.

AUTHOR CONTRIBUTION

Claire McMillan: Conceptualization; data curation; methodology; project administration; software; validation; writing—original draft; writing—review and editing. Sébastien Bonifay: Methodology; validation; writing—review and editing. Christopher Dobe: Conceptualization; methodology; project administration; supervision; writing—original draft; writing—review and editing. Ralph Fliege: Methodology; validation; writing—review and editing. Joachim D. Krass: Methodology; validation; writing—review and editing. Adrian Terry: Conceptualization; methodology; writing—review and editing. Matthias Wormuth: Conceptualization; methodology; project administration; validation; writing—original draft; writing—review and editing.

ACKNOWLEDGMENT

The authors thank Kevin Harradine for his contributions to the early stages of the project. Cambridge Environmental Assessments received direct financial support for the development of the LET model from CropLife Europe (formerly the European Crop Protection Association). The remaining authors are employed by CropLife Europe member companies, which produce and market products that are subject to regulation by REACH.

    DISCLAIMER

    The LET was developed by an expert group formed from the member companies of CropLife Europe and tasked with developing a methodology for assessing co-formulants under REACH. All authors participated in the expert group during the normal course of their employment. The authors have responsibility for the writing and contents of the article, and the views expressed in this article are those of the authors and do not necessarily represent the views or policies of CLE or their respective employers.

    DATA AVAILABILITY STATEMENT

    The CropLife Europe Local Environment Tool (LET) is freely available for download from the CropLife Europe website at: https://croplifeeurope.eu/pre-market-resources/reach-in-registration-evaluation-authorisation-and-restriction-of-chemicals/.