Estimating pesticide environmental concentrations in Latin America: The importance of developing local scenarios

Abstract

Data to assess pesticide exposure in soil and water are scarce and unevenly distributed in Latin America, especially due to the size of the region and the vast agricultural landscape. This makes it difficult to assess associated environmental risks. We suggest that the lack of pesticide exposure or monitoring data can be addressed by using validated models to provide estimated pesticide exposure concentrations in soil and water bodies. This exposure modeling approach has been used by regulatory agencies in other countries and regions such as the United States, Canada, and the European Union. In order to properly estimate pesticide exposure concentrations, we advocate for the development of local scenarios containing local weather, soil, and crop data to be used in the existing models. A sensitivity analysis of the models can be performed to determine parameters that are sensitive and therefore inputs to these parameters are derived locally. We believe the development of local scenarios in the region is attainable and can be a pragmatic approach for developing a more comprehensive picture of potential pesticide exposure in the region. Integr Environ Assess Manag 2021;17:901–904. © 2021 Syngenta Proteção de Cultivos Ltda

CHALLENGES OF PESTICIDE MONITORING DATA IN LATIN AMERICA

In Latin America, most of the regulatory decisions regarding pesticides are based on hazard instead of risks. The Andean countries (Bolivia, Colombia, Ecuador, Peru, and Venezuela) are an exception because they adhere to the Andean manual as a regulatory framework (SGCAN 2002) to assess the risks from the agricultural use of pesticides. In 2014, Carriquiriborde et al. (2014) described the issues of developing reliable risk assessment schemes in the region, specifically an aquatic risk assessment, and recommended several actions such as harmonization of risk assessment frameworks with other regions (e.g., North America and Europe), data sharing within Latin America and with other regions, as well as the characterization of pesticide use in each country. Regarding the development of risk assessment schemes, in 2017 the Brazilian Environmental Agency published its first risk assessment guideline focusing on pollinators (Ibama 2017).

As in other risk assessment schemes, calculating the estimated environmental concentration (EEC) is critical to properly evaluate the risks to the environment and nontarget organisms. The effort on either estimating or measuring pesticide concentrations in soil and water bodies has been gaining attention from local academia (De Gerónimo et al. 2014; Hunt et al. 2016, 2017; Etchegoyen et al. 2017; Alvarez et al. 2019; Caldas et al. 2019) and government agencies (Casara et al. 2012; Souza et al. 2016), focusing on analyzing and monitoring residues in surface water and soils. Nonetheless, monitoring data generation in the region is still generally scarce (Carriquiriborde et al. 2014) because monitoring residues need proper guidelines, infrastructure, and well-trained personnel, which are not readily available in the region. In Argentina, most of the monitoring data come from the Pampa region (De Gerónimo et al. 2014; Hunt et al. 2016, 2017; Etchegoyen et al. 2017; Alvarez et al. 2019), which is the country's most important agricultural landscape. In Chile, data generation has been infrequent: Palma et al. (2004) analyzed several pesticide residues in surface water, while it was not until 2017 that similar work was done for organochloride pesticides (Montory et al. 2017). As for other countries, such as Colombia (Marrugo-Negrete et al. 2014; Pérez-Holguín et al. 2016; Lans-Ceballos et al. 2018), Ecuador (Decknock et al. 2019), and Mexico (Hernández-António and Hansen 2011; Leyva-Morales et al. 2017), the scenario is similar, where residues in water bodies are monitored in a very limited part of each country's most relevant agricultural areas, indicating further efforts are necessary to understand environmental exposure of pesticides.

Similar to Argentina, Brazil has a vast agricultural region, and monitoring data have been generated in regions with very different climatic conditions, such as southern Brazil (Caldas et al. 2019), the Brazilian semiarid region in the Northeast (Souza et al. 2016), and the central region in the State of Mato Grosso (Casara et al. 2012). A recent review by Albuquerque et al. (2016) summarizes the occurrence of pesticides in surface waters in Brazil. In general, the authors observed the lack of data, indicating that residue data in surface waters were found for only a small number of registered pesticides in Brazil (~11%). It was also indicated that available data were obtained from a few sampling sites, which were concentrated in a small number of Brazilian states (5 of 27 states). Albuquerque et al. (2016) suggested that it is important to generate a more comprehensive data set by implementing a nationwide monitoring program.

PESTICIDE MODELING AS AN ALTERNATIVE TO FULFILL THE GAP ON PESTICIDE MONITORING DATA

In the absence of pesticide monitoring data, mathematical models may be used to predict the fate, transport, and exposure of pesticides in the environment and under different environmental conditions (Scorza and da Silva 2011; D'Andrea et al. 2020). One could challenge that a predictive model may not be as reliable as targeted monitoring data; however, a mathematical model that has been well validated and widely used can provide a realistic but conservative exposure estimate that regulators can base their decisions on. Therefore, before selecting a model for predicting pesticide exposure concentrations, it is important to consider a model or tool that has been extensively reviewed, validated, and used to ensure model assumptions, conceptual models, and supporting data remain scientifically sound. Models such as Pesticide Root Zone Model (PRZM) and Variable Volume Water Model (VVWM), which are components of the United States Environmental Protection Agency (USEPA) Pesticide in Water Calculator (PWC) tool, have been extensively used by the USEPA and the Canadian Pest Management Regulatory Agency (PMRA). This tool is readily available, user friendly, regularly reviewed, and recently updated to its version 2.001 (USEPA 2020). The updates included changes not only in its interface, but also in the conceptual model and algorithm, crop scenarios, and weather files (USEPA 2020; Young and Fry 2020).

In developed countries, pesticide exposure modeling is one of the most common approaches by regulatory agencies such as the USEPA and the European Food Safety Authority (EFSA). Therefore, we understand that mathematical models may also help regulators in Latin America to estimate exposure associated with pesticide use on agricultural crops. Pesticide modeling may also help us understand where mitigation measures are necessary in order to reduce exposure and associated risks. In this sense, several Latin American researchers attempted to shift toward pesticide exposure modeling, and therefore several publications in this field started to emerge (Scorza and da Silva 2011; Rämö et al. 2018; Cambien et al. 2020; D'Andrea et al. 2020).

CREATING LOCAL SCENARIOS FOR USE IN EXISTING PESTICIDE MODELING TOOLS

Although some researchers tried to develop local models to evaluate the fate and transport of pesticides (Embrapa 2009, 2010), another approach is to adapt existing models to local conditions (Scorza and da Silva 2011; Rämö et al. 2018; Cambien et al. 2020; D'Andrea et al. 2020). In the latter case, there is an advantage to using a model that is already validated and well recognized (Wolt et al. 2002). As mentioned in the previous section, USEPA models such as PRZM and VVWM, which are components of the USEPA PWC tool, may be used. In the Latin American context, an additional important step would be to develop local crop, soil, and weather scenarios to provide EECs that are representative of local environmental conditions.

Soil, weather, and hydrogeological data are key inputs in multiparametric pesticide exposure models such as PRZM and VVWM. In many, if not most, Latin American countries, these data are not readily available at a national scale and therefore this could be perceived as a challenge for developing local scenarios. Countries such as Argentina, Brazil, and Chile have a reasonable network of field stations from which these data can be retrieved and used as inputs to the models. Not surprisingly, most publications in this area come from Argentina and Brazil, where there have been efforts to stimulate such data generation (MCC 2018; ARG 2020), and therefore, local scenario development should be attainable for these countries.

IMPORTANCE OF SENSITIVITY ANALYSIS TO DETERMINE PARAMETERS OF IMPORTANCE

A sensitivity analysis indicates which parameter influences the output (e.g., EECs) and to what extent (Saltelli 2004; Scorza and da Silva 2011; D'Andrea et al. 2020). For instance, Scorza and da Silva (2011) evaluated the Pesticide Emission Assessment at Regional and Local scales (PEARL) model for Brazilian conditions, and found that soil characteristics influence the output while the water flux does not. On the other hand, D'Andrea et al. (2020), when using the PWC tool, noticed that the output depends mostly on the environmental fate properties of the tested pesticide (e.g., half-lives in soil and water, binding to soil). Moreover, the authors indicated some the most sensitive parameters were not the same for each pesticide because the tested compounds have different soil binding characteristics. Similar sensitivity analysis approaches can be used and tested for other regions to improve model parametrization, resulting in more realistic estimates of pesticide exposure concentrations in soils and water bodies while still maintaining conservatism for wildlife protection. By knowing which parameter is more relevant for estimating the output, efforts should focus on obtaining inputs for such parameters locally, while inputs for parameters that do not influence the output can use default values, available guidance, or recommendations.

Sensitivity analysis of input parameters is not only very helpful to identify the relative importance of a given parameter; it also allows the evaluation of the effectiveness of the selected model. Multiparametric models are often nonlinear in their output response and use multidimensional parameters to define the geographically distributed properties of a natural system. Therefore, it is very important to define which of the parameters are more relevant and have a greater influence on the output in order to properly parameterize the chosen model (Saltelli et al. 2019).

CONCLUSION

In the context described in the present communication, we believe that developing local scenarios for Brazil and other Latin American countries is an attainable effort for regulatory agencies, crop protection industry research groups, and academia. It will improve the prediction of the pesticide exposure in the environment such that it is representative of local conditions, enhance the availability of environmental data, and may improve both local regulatory and research groups' capabilities. It may also contribute to development of local expertise in a region that is one of the world's leading agricultural areas. Therefore, it is an effort that should be encouraged and supported by the local research community.

Acknowledgment

The authors declare no conflicts of interest. The authors declare that there was no funding to support this work. We acknowledge the contribution of Peter Campbell, Syngenta Crop Protection, UK, in the review of this manuscript.

Open Research

Data Availability Statement

Because this Brief Communication represents an opinion, there is no experimental data to be provided. The sources of information are mentioned in the References section.