In the present experiments with D. magna we demonstrate that i) glyphosate and Roundup induce EC50 at concentrations typically below 10 mg/l in 48 h acute experiments, and ii) chronic exposure, particularly to formulated Roundup, causes serious reproduction damage at levels close to (1.35 mg/l) or even below (0.45 mg/l) accepted threshold values for glyphosate in surface waters in the United States in general (0.7 mg/l) and in the state of California specifically (1.0 mg/l)(California EPA 1997).
Acute toxicity of glyphosate and Roundup
There were only minor differences in tolerance to acute exposure of glyphosate and Roundup between laboratory clones, clones from natural ponds and clones taken from ponds surrounded by intensive agriculture (Fig. 1). A tendency towards higher tolerance to Roundup was observed, particularly in the Knokke 4-clone. This may be related to this clone’s origin, a pond surrounded by agriculture. Tests of carbaryl pesticide in 10 clones of D. magna, two of which were from the same lakes as two of our tested Knokke clones (Knokke 1 and Knokke 4), indicated an overall correlation between land use intensity (farming) and carbaryl tolerance (as EC50). These findings were attributed to a genetically based resistance, persistent through several generations of toxicant-free laboratory culturing (Coors et al. 2009). Our observed differences in clonal tolerance of glyphosate toxin can be interpreted as response to environmental differences. However, the biology of this response is unclear at present.
In this work we have shown a relatively uniform susceptibility to glyphosate and Roundup between clones of D. magna. This is in contrast to the extreme variation seen between published studies. Accordingly, the highly varying EC50 values in D. magna, and other species of daphnids reported in printed reviews (Melnichuk et al. 2007a, Pérez et al. 2012, Rico-Martínez et al. 2012), and online databases of pesticide exotoxicology such as the Pesticide Alert Network database (PAN 2011), and the US Environmental Protection Agency Ecotox database (US-EPA 2011), should not be primarily attributed to interclonal differences. D. magna toxicity tests are generally considered reproducible and representative, with only small variation between laboratories (Mark and Solbé 1998). Still, test conditions and laboratory environments or other contextual factors may cause the discrepant results. Furthermore, also the solubility of the different glyphosate-based chemicals may be a decisive factor in glyphosate toxicity testing of aquatic organisms.
In the literature the common name “glyphosate” is used somewhat indiscriminately, including chemical compounds that differ substantially from glyphosate-IPA salt (chemical identity CAS# 38641940), e.g. technical grade glyphosate, which has low solubility in water (CAS# 1071836). Toxicological data for technical grade glyphosate are not representative when assessing ecological effects of glyphosate herbicides, which for spraying need to contain a water soluble form of glyphosate, e.g. the IPA-salt, as the active ingredient (Dill et al. 2010). During our review of published studies we contacted the authors of 4 papers from groups that had recently published D. magna toxicity studies with unspecified glyphosate. These studies were performed in Korea (Le et al. 2010), Turkey (Sarigül and Bekcan 2009), Portugal (Pereira et al. 2009) and Mexico (Dominguez-Cortinas et al. 2008). Authors from 3 of these research groups kindly responded to our information request, confirming that the chemical substance tested was technical grade glyphosate, i.e. the non-soluble version of glyphosate.
Contrary to this, the glyphosate IPA and Roundup formulation tested in the present study is representative for glyphosate herbicides used in agriculture as active ingredient (glyphosate) and formulated product (Roundup). However, variations in toxicity levels may still be expected due to differences in adjuvants and other ingredients of individual formulations (Gasnier et al. 2009, Melnichuk et al. 2007a).
Contrary to the findings of Tsui and Chu (2003) the present work finds acute toxicity of Roundup formulation and active ingredient glyphosate expressed as EC50 (48 h) concentrations, to be in the same order of magnitude. This is in accordance with some published work in other aquatic invertebrates such as Hydra attenuata (Demetrio et al. 2012).
We have also shown that the D. magna tolerance for glyphosate and Roundup is enhanced with increasing age of the animals. This has also been shown for Roundup in other freshwater invertebrates, such as the freshwater shrimp Caridina nilotica (Mensah et al. 2011). Both these freshwater invertebrates have relatively low EC50-values as adults (22 and 25.3 mg/l for D. magna and C. nilotica, respectively). Such values are way below previously published results from acute glyphosate toxicity experiments in D. magna, even for juveniles. For example, Mcallister and Forbis (1978) presented an EC50-value of 759.7 mg/l with a sharp 95 % confidence interval (740.8-779.9).
The European Commission (EC) working document on glyphosate (EC 2002), which forms the basis for European regulation in the context of health and environment, reports the EC50 (48) value of 930 mg/l in D. magna from Forbis and Boudreau (1981). The authors of the EC paper extrapolate this value into a general EC50 value for acute toxicity in aquatic invertebrates. Thus glyphosate is termed “harmless”. According to the 2009 WHO guidelines for pesticide classification (WHO 2009), glyphosate is in class 3; slightly hazardous (in relation to human health). The US EPA has defined glyphosate in Toxicity class 4: “Practically nontoxic”. For a review see Bates (2000).
Furthermore, in 1982 the agrochemicals producer Monsanto presented contrasting data for toxicity of Roundup formulations in Daphnia sp., by simultaneously giving LC50 (96 h) values of 5.3 mg/l for Roundup and 962 mg/l for glyphosate alone (Servizi et al. 1987). Already in 1979 it was pointed out that technical grade glyphosate had properties (notably reduced water solubility) totally different from those of the glyphosate isopropylamine salt (Folmar et al. 1979). This is, however, an important fact that has been commonly overlooked.
In contrast to other published toxicity data for formulated glyphosate-based herbicides, our results are comparable to those of Folmar et al. (1979) at 3 mg/l for Roundup in D. magna, and to 4 of the 6 formulations tested by Melnichuk et al. (2007a) at 4.2–10.2 mg/l. The most recent toxicity data presented by the producer, for the specific brand of Roundup that we have tested, is 11 mg/l EC50 (48) for D. magna (Monsanto 2011) and thus in accordance with our findings. A recent publication by Sarigül and Bekcan (2009) reports a much higher toxicity of a 48 % commercial Roundup formulation in D. magna, with LC50 (48 h) values of 0.012 mg/l. We have no explanations for this discrepancy.
Some difference in methods may partly explain published experimental test result variations. For example, Servizi et al. (1987) presented the LC50 (96 h)-value for Roundup in D. pulex as 25.5 mg/l, but this referred to the Roundup formulation including water. When the authors assessed only the active ingredient glyphosate IPA and the surfactant MONO818 respectively, LC50 (96 h)-values of 7.8 and 3.8 mg/l were recorded.
Chronic toxicity of glyphosate and Roundup
When D. magna were exposed to different concentrations in chronic life-cycle experiments, Roundup produced more serious adverse effects than glyphosate alone. This was the case for all tested end-points: survival, growth, fecundity, abortion rates and juvenile body size. Chronic exposure to 0.05 and 0.15 mg/l of Roundup significantly reduced juvenile body size compared to the control group (p < 0.001). The same was the case for glyphosate at 0.05 mg/l, but to a lower degree (p < 0.05). This is in accordance with findings of Papchenkova (2007) who found juvenile size significantly reduced (p < 0.05) by exposure to 2.0 mg/l (6 of 7 generations) and 0.2 mg/l (3 of 7 generations) a.i. concentrations of Roundup.
In our present study no other measured end-points were affected at these concentrations (Table 1). No significant effects on fecundity or abortion rates were seen at concentrations 0.05–0.45 mg/l for glyphosate IPA, but exposure to Roundup at 0.45 mg/l concentration significantly reduced fecundity and increased the abortion rate in addition to the reduced juvenile body size. Following exposure to Roundup at the 1.35 mg/l concentration, significantly impaired survival and growth was observed and reproduction failed completely: all eggs were aborted. A summary of the results from the chronic exposure tests is given in Table 1.
To put these results and concentrations in context: the US EPA general environmental guideline of 0.7 mg/l and the state specific California EPA environmental guideline limit of 1.0 mg/l glyphosate in surface waters are in between the 0.45 and 1.35 mg/l concentrations we use in our tests. The fact that, in the present study, D. magna subjected to 1.35 mg/l showed complete reproductive failure, aborting all eggs in early to late stages of embryonic development, indicates that the mentioned environmental guidelines may not be sufficiently restrictive to ensure viable populations of D. magna and other aquatic invertebrates.
Ronco et al. (2008) investigated pesticide levels in streams draining several sites with transgenic soybean (glyphosate-tolerant) cultivation in Argentina and found the levels to be; “often below 1 mg glyphosate/l, in Arrecifes tributary, although concentration ranges between 1.8 and 10.9 mg/l (…) were detected”. The authors concluded that non-target aquatic biodiversity (flora, insects, fish and amphibians) was adversely affected by the pesticide applications.
The levels of glyphosate accepted in surface fresh water vary between nations. As far as we have been able to ascertain, the highest tolerated concentrations are found in the earlier mentioned US-EPA guidelines, 0.7–1.0 mg/l, differing strikingly from the EU guideline limit of 0.0001 mg/l (=0.1 ppb), which seems to be the most restrictive. Canada enforces a limit of 0.065 mg/l (Struger et al. 2008), while Ukraine has set the environmental standard to 0.02 mg/l (Melnichuk et al. 2007a, b).
The results of the few other published studies on chronic exposure of daphnids to glyphosate or glyphosate-based herbicides are distinctly inconsistent.
An industry standard 21-day reproduction test of glyphosate in D. magna, based on test concentrations of 0, 26, 50, 96, 186 and 378 mg/l, was additionally reviewed and extrapolated by dr. Wayne C. Faatz, in a March 1983 report to the US EPA (McKee et al. 1982). Neither significant increase of mortality nor reduction of growth was observed in any of the test concentrations. For reproduction, the same report established 50 mg/l as NOEC, a level 100 times higher than the NOEC determined in our experiments.
In contrast, Papchenkova (2007) exposed seven generations of D. magna to 0.02, 0.2 and 2 mg/l a.i. glyphosate in Roundup. Significant reduction of endpoints related to fecundity, length of newborn juveniles and growth in first generation was recorded for D. magna exposed to a concentration of 2 mg/l. Significant effects on the same endpoints were observed also in subsequent generations for concentrations 2.0 and even 0.2 mg/l, but these effects were not consistent in all measured end-points through all of the 7 generations studied. A follow-up generational study of chronic toxicity in D. magna exposed to much higher concentrations of Roundup, i.e. 25 and 50 mg/l a.i. for four generations, showed a significantly reduced fecundity but no adverse effect on the survival of mother animals (Papchenkova et al. 2009).
A similar complexity is evident in a chronic effects study of the Fakel herbicide (48 % a.i. glyphosate IPA) in Ceriodaphnia affinis (Melnichuk et al. 2007b). Generational exposures to 10, 5, 2.5, 1 0.1, 0.01 and 0.001 mg/l Fakel established a NOEC of 0.001 mg/l. Even at the very low concentration of 0.01 mg/l, first and second generation fecundity was found to be significantly reduced compared to the control group. Temperature-dependent effects on end-points fecundity and abortion were recorded at test concentrations 1.0–0.1 mg/l. As temperatures were reduced, adverse effects decreased (Melnichuk et al. 2007b). C. affinis was shown to be more sensitive to glyphosate herbicide Fakel in acute LC50 (48) tests than D. magna (12.6 mg/l vs. 26.5 mg/l respectively (Melnichuk et al. 2007a).
The acidity of the aquatic environment (or laboratory medium) may also be a relevant factor for the toxicity of glyphosate-based herbicides. Chen et al. (2004) exposed the daphnid Simocephalus vetulus to glyphosate herbicide Vision® in sublethal concentrations 0.75 and 1.5 mg/l a.i. (acid equvivalent) under two different pH-regimes (pH 5.5 and 7.5). The authors found that survival, fecundity and juvenile maturation time was affected at both concentrations. The effects were more severe at neutral pH 7.5, versus the lower pH 5.5. Thus, the acidity of the experimental or environmental conditions must be taken into account, in particular when the buffering capacity of the artificial holding medium is low and the toxicants tested are acidic. In our experiments, the variation in acidity spanned a range of 2 pH-units. However, they were still within the pH 6–9 range defined as preferred experimental conditions for D. magna testing (OECD 2008).
The term “inert-ingredient” for Roundup formulation additives has been used for product labeling. This is problematic when published literature documents that additives may have significant direct or synergistic toxic effects. Numerous studies have demonstrated that surfactants, often called “adjuvants” or “inert ingredients”, used in Roundup formulations are the primary toxic agents, with toxicity notably higher than glyphosate (“the active ingredient”) alone (e.g. Benachour et al. 2007; Folmar et al. 1979; Gasnier et al. 2009; Melnichuk et al. 2007a).
Summary and conclusions
According to our experimental work and our literature reviews, we find that the previously published EC50 values of 780-930 mg/l for glyphosate (McAllister and Forbis 1978; Forbis and Boudreau 1981) are not representative. The classification of glyphosate as “practically nontoxic” to aquatic invertebrates is based on these non-representative values. The high EC50 values have demonstrated tenacious lives, been extensively referred to in the literature and have also found their ways into regulatory documents.
We have found the acute toxicity of glyphosate herbicide active ingredient to be substantially higher, with concentrations below 10 mg/l inducing immobility in D. magna within 48 h. If such more conservative EC50 values were used, glyphosate would be classified as “toxic” or “moderately toxic” to aquatic invertebrates.
In our chronic studies covering the whole life-cycle of D. magna, we demonstrated negative and serious effects at very low concentrations (see Table 1 for a summary), i.e. at levels that can be expected with use of the herbicide Roundup at prescribed dosages in agricultural practice.
The results of our acute and chronic toxicity tests with glyphosate-IPA and Roundup herbicide, in combination with our review of published data, warrant the conclusion that current European Commission and US EPA toxicity classification of these chemicals with regard to effects on D. magna and aquatic invertebrates in general, is based on non-representative evidence and needs to be adjusted.