Widespread agrochemicals differentially affect zooplankton biomass and community structure: Comment

Determining the combined effects of multiple contaminants on freshwater ecosystems poses a major challenge to ecotoxicology but is vital to advance realism in the discipline and to ensure that risk assessments account for indirect effects and interacting factors on multispecies assemblages (Fleeger et al., 2003; Relyea, 2009). We therefore appreciate the approach by Hébert et al. (2021), who used outdoor mesocosms to investigate the individual and combined effects of a glyphosate-based herbicide (GBH) and the neonicotinoid insecticide imidacloprid across a range of concentrations under mesotrophic and eutrophic conditions. Using mesocosms introduces experimental artifacts, but what is lost in realism is arguably made up for through greater experimental control. It is this experimental control that supports causal attribution of observed experimental effects to the experimental treatment. Unfortunately, we contend that Hébert et al. (2021) make a critical error, namely, they attribute the observed effects of a commercial herbicide formulation on the zooplankton community entirely to a single ingredient, ignoring the large body of literature cautioning that the toxicity of GBHs might be largely due to their adjuvants. We consider this irresponsible because Hébert et al. (2021) conclude by calling for national guidelines on glyphosate to be revised and published this paper while the Federal Court of Appeals in Canada was hearing a case brought by Safe Food Matters objecting to Health Canada's Pest Management Regulatory Agency's 2017 re-evaluation of glyphosate. In their 8 February 2022 decision, the court ruled that Health Canada must reassess glyphosate. This emphasizes the extremely contentious nature of research into glyphosate and drives home the need for scientists to be diligent in designing their studies and communicating their results (Rivoalen et al., 2022; Robson, 2022). We contend that such overreach compromises an otherwise elegant experimental design and compounds other issues with Hébert et al. (2021) relating to (1) the lack of transparency regarding the product used, as well as the form of glyphosate and any adjuvants in it, (2) apparent use (apparent because the product's name is not accurately reported) of a domestic product to simulate agricultural applications, and (3) unrealistic “off-label” application directly to the water, bypassing weathering, soil sorption, bacterial degradation, and so forth. All mesocosm research faces challenges in extrapolating to real ecosystems, but Hébert et al.'s (2021) unreserved attribution of their experimental effects to a single ingredient in an undefined chemical mixture is what prompted us to write this comment. We hope it will spark greater scrutiny and rigor among our peers and readers of this journal. Glyphosate is the most used herbicide globally (Benbrook, 2016; Duke, 2018), serving as the active ingredient in hundreds of products, some intended for forestry or agricultural applications, some for domestic lawn and garden use, and some even for direct application to standing water for invasive species control. Consequently, traces of glyphosate are ubiquitous in the aquatic environment (e.g., Medalie et al., 2020; Okada et al., 2020; Silva et al., 2018), and its environmental toxicity has been extremely well studied (see reviews Annett et al., 2014; Giesy et al., 2000; Gill et al., 2018; Rodríguez-Gil et al., 2021). Indeed, glyphosate has been subject to renewed interest following the 2015 United Nations classification as a probable carcinogen (IARC, 2017). With this increased attention and accompanying research effort, you might expect that we have a thorough and coherent understanding of the toxicological risks presented by glyphosate, but such is not the case. As Mesnage and colleagues eloquently demonstrate in their 2019 review, the literature on the toxicity of GBHs is a mess (Mesnage et al., 2019). One cause of confusion is that the effects observed differ between studies testing glyphosate alone and those testing commercially formulated GBHs. For decades we have known that GBHs tend to elicit deleterious effects at much lower concentrations than when glyphosate is tested alone (Folmar et al., 1979; Mann & Bidwell, 1999). Confusion as to the actual driver of the observed effects is, however, widespread in the glyphosate literature, as noted by many researchers (Annett et al., 2014; Giesy et al., 2000; Mesnage et al., 2019; Rodríguez-Gil et al., 2021; Thompson et al., 2006). A second cause of confusion is that toxicity studies using GBHs commonly fail to report the form, formulation, appropriate concentration of active ingredient (i.e., in acid equivalents), and country where the product was procured. This is crucial information because products under the same trade name can have very different formulations from one country to another as companies adapt formulations to different crop uses or regulatory requirements. In fact, while the brand name Roundup has historically been synonymous with glyphosate, Bayer recently started commercializing Roundup-branded products for home and garden use in the European market that contain no glyphosate at all. Over the years, these issues have been raised by regulators and risk assessors asking the scientific community for help on providing a clearer picture of glyphosate's toxicity (Annett et al., 2014; Giesy et al., 2000; Mesnage et al., 2019; Rodríguez-Gil et al., 2021; Thompson et al., 2006). In fact, a 2005 study in this very journal (Relyea, 2005) resulted in a discussion about very similar topics (Relyea, 2006; Thompson et al., 2006). Seventeen years later, Hébert et al. (2021) makes these same mistakes. Hébert et al. (2021) report only that the source of glyphosate in their mesocosm study was a product called “Roundup Super Concentrate (Monsanto).” They give no information about the form of glyphosate used or the other chemicals contained within this formulation, nor do they provide the registration number under the Pest Control Products Act (PCPA), which would confirm the product's identity. This failure in reporting hinders transparency and prevents other researchers from reproducing their work. There is no product registered under this name for sale in Canada. After scouring Health Canada's Public Registry and consulting Bayer directly (A. Russell, Regulatory Affairs Manager, Bayer Crop Science Inc., personal communication, September 24, 2021), we determined that the closest match was Roundup Super Concentrate Grass & Weed Control (PCPA Registration [Reg.] No. 22759), registered to Monsanto Canada Inc., now owned by Bayer. If this is the product Hébert et al. (2021) used, then it contained glyphosate as an isopropylamine salt (CAS No. 38641-94-0) at 356 g a.e./L (approximately 50.2% by weight) (Bayer Crop Science Inc., 2020). This information is important. Different forms of glyphosate (pure acid, potassium, ammonium, trimesium, or isopropylamine salts) have different molecular weights, complicating dosing and exposure calculations (Mesnage et al., 2019). They also have different toxicities, with the acid form altering pH in a manner that increases its toxicity to exposed aquatic organisms (Edginton et al., 2004; Tsui & Chu, 2003). Although the concentration of the glyphosate component can be standardized to acid equivalents, as Hébert et al. (2021) do, this does not tell us about the concentration of other ingredients in the product they applied: What is the chemical composition and toxicity of the other 49.8%? The inactive ingredients might include adjuvants like surfactants to make the product spread on leaf surfaces, colorants, anti-foaming agents, and so forth. The composition of adjuvants in most pesticide products are proprietary, as is the case with Reg. No. 22759, yet these chemicals can be more toxic than the active ingredient (Currie et al., 2015; Folmar et al., 1979; Mesnage & Antoniou, 2018; Rodriguez-Gil et al., 2017, 2021). For example, the most common class of surfactants historically in GBHs used in agriculture and forestry are polyoxyethylene amines (POEAs). Traditionally the most common of these POEAs has been polyoxyethylene (15) tallow amine (POE-T), which is known to be highly toxic to aquatic life (Rodriguez-Gil et al., 2017). Comparison of acute aquatic species-sensitivity distributions (SSDs) for commercial products containing POE-T against SSDs for the active ingredient alone show HC5 values (the concentration causing the studied effect in 5% of the tested species) approximately 10 times lower for formulated products (480 μg/L) than for the active ingredient alone (4490 μg/L) (Rodríguez-Gil et al., 2021). In other words, traditional GBHs containing POE-T are ~10 times more toxic to aquatic organisms than the active ingredient alone when directly exposed in the water column. While newer formulations have moved to less toxic POEAs or even other classes of surfactants (Currie et al., 2015; Mesnage et al., 2019; Rodríguez-Gil et al., 2021), it remains essential to consider adjuvants in commercial formulations a confounding factor. Hébert et al. (2021) acknowledged the need to determine interactions of co-occurring agrochemicals—it is the main objective of their study—but how can this be achieved when constituents of tested products are unknown? But it is not the use of a commercial formulation rather than pure glyphosate that we object to. It is Hébert et al.'s (2021) widespread reference to “glyphosate” throughout the manuscript when referring to the GBH product they actually tested. Attributing observed effects to active ingredients alone is misleading when it is well known that formulated products can be significantly more toxic, that toxicity varies among formulations, and there is uncertainty regarding which formulation Hébert et al. (2021) used. Clearly their conclusion that “Overall, glyphosate was the most influential driver of aggregate community properties of zooplankton” must be reconsidered. Certainly, their assertion “…that national guidelines should be revised, especially in agricultural areas” goes too far. Their approach prevents causal attribution of observed effects to any individual ingredient, and because there are so many different forms and formulations of glyphosate used, the practical implications of their study of the whole-mixture effects of a single product are extremely limited. The argument might be made that ecotoxicology studies using pure ingredients in controlled lab conditions lack realism, and we must study whole mixtures if we are to understand the environmental risks presented by commercial products. Indeed, Mesnage and Antoniou (2018) make this case very compellingly. Yet causal attribution requires that studies of unknown mixtures incorporate a control containing just the active ingredient to isolate its effects from those of the adjuvants, which Hébert et al. (2021) fail to do. The results of a recent workshop addressing issues related to the risk assessment of multicomponent substances and unknown or variable composition, complex reaction products, or biological materials (UVCBs) (Salvito et al., 2020) highlight the need to understand the fate of the individual components of a mixture such as a pesticide formulation to understand exposure and, hence, risk. Risk assessment frameworks for multicomponent substances start by asking an important question: Is exposure to this mixture “as is” likely in a particular environmental compartment? In this context, we further counter that Hébert et al. (2021)'s approach fails to achieve realism in two critical ways, both pertaining to their “off-label” use of the glyphosate-containing product. Product labels dictate the terms of their legal use in Canada under the Pesticide Control Products Act, and product applicators in the real world follow these label restrictions. Off-label use in an experimental context cannot be taken as representative of real-world use, and hence the resulting exposure scenarios cannot be considered environmentally relevant. This limits the validity of any statements of risk that emanate from work conducted under off-label conditions. In fact, it was the off-label use of the tested product that critiques of Relyea (2005) centered around. Hébert et al. (2021), state their objective is to look at globally relevant agricultural pollutants, but the product Reg. No. 22759 relates to domestic lawn and garden use (Bayer Crop Science Inc., 2020), contradicting Hébert et al.'s (2021) claim to have used an agricultural formulation, and their general framing of their experiment in the context of agricultural pollution. This claim exacerbates our confusion regarding what formulation Hébert et al. (2021) used. To extrapolate to exposure scenarios and inform risk assessments of agricultural practice, one must use a product approved for use in agriculture because the composition and, hence, toxicity profile of the products can vary considerably. Then, Hébert et al. (2021) applied the herbicide directly to the water in their mesocosms. This unrealistic application contradicts the product label instructions and would be subject to fines under the Agriculture and Agri-Food Administrative Monetary Penalties Act, which are typically $10,000 or more. The 2017 re-evaluation of the registration of glyphosate by Health Canada required that all product labels be updated to require buffer zones to protect nontarget organisms, including aquatic invertebrates and fish (PMRA, 2017). For POEA-containing GBHs, these restrictions have been in effect in several jurisdictions since the 1990s (NRA, 1996). Even the Material Safety Data for Reg. No. 22759 specifies that it should be kept out of drains, sewers, ditches, and waterways and not applied if rain is forecast within 2 h (Monsanto Company, 2017). The glyphosate we detected when monitoring aquatic ecosystems in agricultural areas (e.g., Medalie et al., 2020; Okada et al., 2020; Silva et al., 2018) did not get there by direct application but by well-studied mechanisms of leaching and run-off, including desorption from treated plants and soils (Kanissery et al., 2019). During this process, the product is exposed to UV light, air, and microbial decomposition. In fact, the historically common POE-T surfactant is known to readily bind to soil and sediment particles (Rodriguez-Gil et al., 2016, 2021; Tush et al., 2018; Tush & Meyer, 2016), limiting its water-column half-life to less than 6 h in the presence of sediment. Notably, the mesocosms used by Hébert et al. (2021) lacked sediment, undoubtedly altering the fate of the GBH. Regardless, with constituents of the GBH all having different half-lives, sorption coefficients, water solubilities, and decay pathways, it is extremely unrealistic to assume that direct, over-water application of a GBH intended for domestic terrestrial use is representative of the toxicity profile experienced by zooplankton communities in agricultural environments, even if the experimental concentration of glyphosate measured on an acid equivalents conversion were representative of environmental monitoring results. Zooplankton in the environment would be exposed to a completely different mixture. What really motivated us to comment on Hébert et al. (2021) is that they end their paper by calling for a reassessment of environmental protection benchmarks for glyphosate despite its recent re-evaluation (CCME, 2012; PMRA, 2017; US EPA, 2020) and during a period where this re-evaluation was being contested in the Federal Court of Appeals in Canada (Rivoalen et al., 2022). All pesticides are re-reviewed periodically as part of continuous registration for use in almost every country. In Canada, the Pest Management Regulatory Agency conducts reviews every 15 years, and in the United States the Environmental Protection Agency conducts reviews every 10 years. Water quality guidelines are revised under similar schedules. We agree that periodic review of benchmarks in light of new toxicological evidence is important, but our resources for ecotoxicological research are limited, and we argue that we are misallocating those resources as a field. We point to the excellent work by Kristiansson et al. (2021), which clearly shows our field is overly focused on a small number of chemicals, while we have virtually no knowledge about the ecotoxicological effects of literally thousands of commonly used chemicals in our ever expanding chemosphere. Several of the shortcomings (e.g., limited ability to incorporate single-compound controls, lack of sediment in mesocosms) are the result of common logistical challenges in ecotoxicology research that all studies face, but others (e.g., off-label application method, uncertainty surrounding the composition of the formulation) are the result of design decisions that could have easily been avoided by taking the time to understand current risk assessment needs. It is irresponsible to ignore the constraints these decisions place on our interpretation of results in a risk assessment context, and we must be diligent in attributing observed effects. As a community of authors, reviewers, and editors in ecotoxicological research, we need to do better in enforcing standards of transparency and reproducibility, and we must guard against overreach when offering opinions on regulatory decisions. The authors declare no conflicts of interest.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.