Background

The herbicidal properties of glyphosate (n-(phosphonomethyl)-glycine, CAS: 1071-83-6) were first discovered in 1974 by J. E. Franz while he was working for Monsanto Company. As a broad-spectrum non-selective systemic herbicide which can be used for total control of weed plants [1], glyphosate quickly gained high importance in worldwide agriculture [2, 3] with dramatically increasing sales figures during the last decades [4, 5]. The intended mode of action of glyphosate is related to its influence on the shikimate pathway in plants via inhibition of the enzyme 5-enolpyruvylshikimate-3-phosphate synthase (EPSPS), resulting in a cessation of aromatic amino acid synthesis [6,7,8]. Since aromatic amino acids such as phenylalanine, tyrosine, and tryptophan are essential for cellular metabolism, their deficiency results in a perturbation of the metabolic homeostasis in plants [9], leading to growth arrest and ultimately to plant death. In addition to this disruption of the shikimate pathway, glyphosate has been shown to affect plant physiology by inhibiting photosynthesis, increasing oxidative stress, and modulating diseases, e.g., root infections [10]. Furthermore, numerous microorganisms are known to synthesize aromatic amino acids via the shikimate pathway, which makes them additional target organisms for glyphosate, e.g., in soils, river sediments, or in the microbiome of animals [11,12,13,14]. Considering a worldwide application of more than 750 000 tons of glyphosate per year, this chemical and/or its metabolites can now be detected in all ecosystem compartments worldwide [15,16,17]. Emerging evidence has shown that glyphosate and its commercial formulations have negative impact on fish [18, 19], including organ damage [20,21,22], teratogenic and genotoxic effects [23, 24], behavioral abnormalities [25], and disorders in reproduction [26,27,28].

Particular attention must be paid to the interactions between temperature and pollutant inputs to water bodies, especially in the context of climate change with rising air and water temperatures, increasing heavy rainfall events leading to increased pollutant drift from the surface, and increasing periods of drought leading to increased pollutant concentrations due to lower dilution ratios [29,30,31,32,33]. Regardless, pesticides are used throughout the year for different purposes and therefore at different ambient temperatures. Diffuse inputs of these substances into water bodies via the air or after rainfall also occur at different temperatures [34]. The corresponding water temperatures influence the material properties of the introduced substances, such as their solubility or volatility, their biotic and abiotic degradation in the water, and the physiological state and metabolism of poikilothermic aquatic organisms, which are continuously exposed to substances dissolved in the water or bound to particles throughout their lives [35, 36]. Water temperature therefore influences the uptake, biotransformation, and possible detoxification as well as the possible excretion or storage of substances in the organisms [37].

Based on its low octanol/water partition coefficient (Log Kow: − 3.4), a low accumulation potential of glyphosate in biota can be expected [38]. Nevertheless, the herbicide has been shown to bioaccumulate in the tissues of several organisms, after exposure through food, water, or air. For example, in Lumbriculus variegatus, the total amount of glyphosate that accumulated in the tissue increased to approximately 8 µg/g fresh weight with the concentration of glyphosate and Roundup® in the surrounding media (0.05–5 mg/l) [39]. Numerous studies have been conducted to measure glyphosate concentrations in the tissues of farm animals that serve as food sources. Up to now, glyphosate was not detected in meat, milk, and eggs, but in some sensitive organs [40]. In cows, the kidney and lung are the most susceptible organs with a glyphosate concentration of about 60–80 ng/g [41], while in chickens, the glyphosate concentration in the intestine was the highest at about 100 ng/g [42]. An average of 2 mg/kg glyphosate was detected in the livers of malformed piglets [43], and 5–16 mg/kg was found in the livers of pigs receiving glyphosate-contaminated food [44]. This phenomenon cannot be explained in general by passive diffusion of the compound and may be rather due to the active uptake of glyphosate into cells by amino acid transporters with a high affinity for glyphosate, which have been identified in mammalian cells [45].

Glyphosate-based herbicides (GBHs) like Roundup® always contain co-formulants additional to the active ingredient, which function as surfactants, diluents, or stabilizers, e.g., polyethoxylated tallow amine (POEA) [46], and improve the uptake of glyphosate into cells of target organisms [46, 47]. Against this background, it was of interest whether residues in biota exposed to pure glyphosate were lower than those in biota exposed to a commercial herbicide containing equimolar concentrations of the active ingredient. This is the more important because GBHs are often reported to be more toxic to non-target organisms than the herbicide itself [48,49,50,51].

Not only glyphosate but also its major metabolite AMPA has been shown to persist in the environment [52, 53]. It is assumed that the affinity of both glyphosate and AMPA to solid material, e.g., in soil, does not differ to a great extent [54], resulting in half-lives of 2–240 days, depending on the intrinsic chemical and physical properties of the soil particles or sediments that affect degradation [38, 55,56,57]. In general, the half-life of AMPA is somewhat greater than that of glyphosate, suggesting a slower degradation rate and making AMPA more persistent in the environment [57,58,59,60]. To date, data on AMPA accumulation in biota are scarce and, when available, mostly related to plants [61]. Recently, AMPA has been shown to bioaccumulate in the liver and muscle of broiler chickens [62] and in the armored catfish Hoplosternum littorale collected from a rice field [63], but further data on fish, especially related to possible adverse effects, are lacking.

Even though glyphosate residues can be expected in fish for human consumption, no threshold values for maximum residue levels have been determined for food fish so far, neither in German national legislation nor in European Union regulations. The present paper provides data on the bioconcentration of glyphosate and its main metabolite AMPA in brown trout, a commercially important fish species with high ecological relevance in freshwater ecosystems [64, 65] as a basis for a future derivation of such values.

Experiments required for the risk assessment of substances are usually conducted at standard temperatures and, thus, influences of temperature on the toxicity of chemicals are therefore often ignored. To address this shortcoming, in the present study mortality and bioconcentration of glyphosate and AMPA were measured in brown trout after exposure to different concentrations of glyphosate, AMPA, and Roundup®—at concentrations equimolar to the highest glyphosate test concentration—for two, three or four weeks at 7 °C or 15 °C. The selected test concentrations were based on the 2016 template for the annual average environmental quality standard published by the EU, which has defined a concentration of 56 µg/l glyphosate as a regulatory acceptable concentration (RAC) [66], and therefore 1, 10, and 100 times this concentration were chosen for the experiments. In the meantime, a RAC of 100 µg/l [67] and Environmental Quality Standards (EQS) were determined for surface waters including an AA-EQS (average allowed concentration) of 86.7 µg/l for freshwater not used for the abstraction and preparation of drinking water, and an AA-EQS of 0.1 µg/l for freshwater used for drinking water purposes [68].

Methods

Test organisms

Approximately six- and ten-month-old brown trout (Salmo trutta f. fario) were obtained from a commercial fish farm (Forellenzucht Lohmühle, 72275 Alpirsbach-Ehlenbogen, Germany). This breeding facility is classified as category I, i.e., disease free, according to the EC Council Directive 2006/88 [69]. Prior to the experiments, fish were acclimated to laboratory conditions for one week in a climate chamber at 7 °C or 15 °C, respectively, with a 10/14 h light/dark cycle. Experiments were conducted at two temperatures in order to simulate fish exposure at different seasons of the year (early spring and summer in Central Europe), resulting in different water temperatures in the field, which may influence the uptake, biotransformation, and possible detoxification as well as the possible excretion or storage of substances in an organism. Prior to the experiments, the fish were maintained in a 250-l aquarium with constantly aerated filtered tap water (iron, particle, and activated carbon filters) equipped with an external filter (CristalProfi, JBL, Germany). To avoid direct light exposure, two-thirds of the side glass panes and the top of the tank were covered with 0.5 mm black PVC. Fish were observed daily and fed commercial trout feed (Inico Plus, BioMar, Brande, Denmark).

Exposure experiments

All experiments were conducted under the same conditions in the same climate chamber. Three times six 25-l aquaria, each containing 15 l of the respective test solution, were placed in this climate chamber resulting in a three-block design. Also in these aquaria, the lower two-thirds of the side glass panes and the aquaria lids were covered with black PVC to prevent direct lighting. Between 30 and 40 six-month-old individuals or 10 ten-month-old individuals were exposed in each aquarium to account for possible age-dependent difference in body size. Fish were daily fed a defined portion of appropriate commercial food (Inico Plus, BioMar, Brande, Denmark) corresponding to 0.5–1% of the initial body weight. The health condition of the fishes was checked twice a day and dead fish were immediately removed. Every two to three days, half of the respective test solutions, feces, and food residues were removed from each tank and replaced by fresh test solution. Water parameters (NH4, temperature, oxygen content, pH, and conductivity) were controlled regularly, and although ammonium compounds were sometimes elevated despite even more frequent water changes, this had no effect on mortality or other measured variables. In addition, water samples were taken at the beginning and at the end of the exposure experiment to determine glyphosate and AMPA levels.

For the exposure experiments, glyphosate (n-(phosphonomethyl) glycine; Sigma-Aldrich, Merck KGaA, Darmstadt, Germany), Roundup® (Roundup® LB Plus, purchased at a local retail store), and AMPA (aminomethylphosphonic acid; Acros Organics BVBA, Geel, Belgium) were used. Stock solutions for each test medium were prepared with ultrapure water (glyphosate: 500 mg/l and 560 mg/l, respectively; Roundup®: 560 mg/l; AMPA: 366 mg/l) and diluted to the respective test concentrations. Table 1 provides information on the respective treatments/test concentrations and gives also further information about the test fish in the five experiments conducted. In a range-finding experiment, brown trout were exposed to 500 µg/l and 5000 µg/l glyphosate as well as Roundup® with 5000 µg/l glyphosate prior to the exposure experiments. In the following experiments 1 (E1) and 3 (E3), fish were exposed to an ascending series of glyphosate concentration, the highest concentration of glyphosate in Roundup®, and an equimolar amount of the major metabolite AMPA. The two experiments differed in the age of the fish (approximately 10 months and 6 months, respectively), the temperature of exposure (7 °C and 15 °C, respectively), and the duration of the experiment, as an increased total mortality of more than 10% at 15 °C necessitated termination of the exposure in this experiment (E3) after two weeks due to animal welfare reasons. Half of the survivors were sampled, and the other half was allowed to recover in pure filtered tap water for three weeks (experiment 4; E4). Again, water exchange was conducted every two to three days to remove feces and food residues. In the second experiment (E2), fish were exposed not only to 5600 µg/l glyphosate and 3666 µg/l AMPA but also to phosphoenolpyruvate (PEP; Biosynth Carbosynth, Bratislava, Slovakia), which is the glyphosate antagonist for the binding site on the EPSPS enzyme in the shikimate pathway in order to prove whether the addition of PEP may interfere with uptake and toxicity of glyphosate. PEP was added to the aquaria at a concentration equimolar to 5600 µg/l glyphosate. After exposure, fish were anesthetized and killed by 1 g/l tricaine methanesulfonate (MS-222; Pharmaq, Overhalla, Norway), buffered to a neutral pH level with NaHCO3, followed by a cut of the cervical spine. Tissue samples were taken for the assessment of several endpoints of toxicity (data not included in this manuscript) and the determination of glyphosate and AMPA residue concentrations (data presented here).

Table 1 Exposure experiments: brown trout at various ages with glyphosate, glyphosate in Roundup®, AMPA, and PEP
Full size table

Chemical analysis

At the beginning and at the end of the experiment, 45 ml of test solution was sampled from each aquarium and frozen at − 20 °C until further processing. After dissection of fish and removal of viscera for other analyses, the remaining fish tissues (mainly muscle tissue, head, backbone, and caudal fin besides fillets from the range-finding experiment) had to be pooled (up to 40 individuals per sample) in order to obtain the minimum tissue weight required for chemical analyses. In 2020 (time of the range-finding experiment), the required minimum weight per sample was lower than in subsequent years, so the fillets alone could be chemical analyzed as a pool (approximately 15 g of tissue). Later, the total weight of the fillets was not sufficient for analysis and samples containing muscle, head, backbone, and caudal fin tissue were pooled to reach the required amount of tissue. Tissue samples were stored at − 80 °C until further processing. Details on sample sizes and characteristics, etc., are provided in Additional file 1: Table S1.

The actual concentrations of glyphosate and AMPA in water samples from the test aquaria and in the exposed fish were determined by Eurofins (Eurofins Sofia GmbH, Berlin, Germany) according to their routine lab procedure to determine the concentration of glyphosate in food and liquids: After addition of 0.1 N HCl, shaking, and centrifugation, the tissue extract was taken and neutralized with 0.1 N KOH. After another centrifugation step, an internal, isotope-labeled standard was added and everything was derivatized with fluorenylmethyloxycarbonyl chloride (Fmoc-Cl), before HCl was added. A minimum of 20 g of tissue was required and the material matrix used was specific for plant material, food, and feed. For water analysis, an internal, isotope-labeled standard and HCl were added to 500 µl of sample. After derivatization with Fmoc-Cl, quantification in both cases was performed by liquid chromatography coupled to mass spectrometry (LC–MS/MS). The limit of quantification was 0.01 mg/kg in tissue and 0.05 µg/l in water, respectively. The results for glyphosate and AMPA concentrations were reported together with a generally anticipated and accepted measurement uncertainty of 50%, which is common in pesticide residue analysis (according to SANTE/11813/2017 [70]).

Finally, the bioconcentration factor (BCF) of the investigated compounds was determined by means of division of the concentration of glyphosate or AMPA in biota (in mg/kg) by the concentration of glyphosate or AMPA in water (in mg/l). Due to the small size of the individual fish, samples had to be pooled for analysis.

Statistics

In order to quantify the explanatory potential of the parameters ‘exposure concentration,’ ‘temperature,’ ‘age of fish,’ and ‘duration of exposure’ for the variation of the internal concentrations of glyphosate and AMPA as well as the mortality of the fish, we carried out multiple regression modeling using SAS JMP 16.2.0 for the data resulting from the experiments with continuous exposure (namely E1, E2, E3). To model the effects of glyphosate, all approaches with the pure substance, with Roundup® and the controls were analyzed (n = 14 mean values). With regard to AMPA, the modeling was carried out for the approaches with the pure substance and with the controls (n = 8 mean values). As the high mortality observed in E3 resulted in an unplanned, earlier termination of exposure (two instead of three weeks), we excluded the parameter ‘duration of exposure’ from the models generated to explain mortality.

Results

The mortality of the brown trout varied greatly between the experiments (Fig. 1). In E1, where the fish were the oldest at 10 months, no animal died during the exposure period at 7 °C. In the younger animals, mortality at 7 °C water temperature (E2) averaged 5.2% and increased dramatically at 15 °C exposure (E3), forcing the experiment to be terminated earlier. In the controls of experiments 2 and 3, mortality rates were comparable at the different temperatures. AMPA, equimolar to the highest concentration of glyphosate, resulted in a mortality rate similar to that of glyphosate alone, while Roundup® had the highest averaged mortality rate. In the recovery experiment with clean water (E4), only one animal died in each of the treatment groups previously exposed to the highest concentrations of glyphosate, AMPA, and Roundup® (E3), although the temperature was the same (15 °C). The effect of glyphosate on mortality was clearly greater at 15 °C than at 7 °C.

Fig. 1
figure 1

Mean mortality in percent ± standard deviation of brown trout in experiment 2 (E2, exposure at 7 °C), experiment 3 (E3, exposure at 15 °C), and experiment 4 (E4, recovery at 15 °C). Data of experiment 1 (E1, exposure at 7 °C) are not shown as mortality was always 0. Data are calculated as the means of three respective replicates of each treatment

Full size image

Multiple regression modeling revealed the variation in mortality in the glyphosate-exposed animals (n = 14) to be explained to 81.97% (p = 0.0005) by the parameters ‘temperature’ (p = 0.0035), ‘age of fish’ (p = 0.0310), ‘glyphosate concentration’ (insignificant contribution to the model, n.s.), and the intercept (n.s.). ‘Temperature’ alone explained 70.07% (p = 0.0002) of the variation in mortality; ‘age’ alone explained 56.03% (p = 0.0021). With increasing temperature, the mortality increased—this was already the case for 56 µg/l glyphosate, an environmentally relevant concentration. With increasing age of the fish, mortality decreased. The variation in AMPA-induced mortality (n = 8) could be explained by 88.97% (p = 0.0230) by the parameters ‘age of fish’ (p = 0.0192), ‘temperature’ (n.s.), ‘concentration’ (n.s.), and the intercept (p = 0.0491). ‘Age of fish’ alone explained 66.70% (p = 0.0130) of the variation in mortality with a significant intercept (p = 0.0029). The highest AMPA-induced mortality was associated with the highest temperature (15 °C), but due to this single data point, the contribution of the parameter 'temperature' to the model remained insignificant. As in the glyphosate experiments, also AMPA-induced mortality decreased with increasing age of fish.

The real water concentrations of glyphosate and AMPA were determined in the range-finding experiment at the end of the experiment (Additional file 1: Table S2). In the exposure experiments 1 to 3, samples were taken before and after the exposure experiments. Real concentrations were very close to the respective nominal concentrations in the different treatments (Table 2). In the control group, very low amounts of glyphosate and/ or AMPA were measured in some cases, which is most likely due to accidentally carryover during water exchange. The low levels of AMPA measured in the glyphosate and Roundup® treatment groups may indicate degradation processes of glyphosate. In the second experiment, a greater amount of glyphosate was found in the treatment with AMPA and PEP—as PEP is structurally very similar to glyphosate, re-metabolization may have occurred, but this requires further investigation. PEP did not interfere with the uptake of glyphosate.

Table 2 Real concentration of glyphosate and AMPA in water in the exposure experiments 1–3
Full size table

In all experiments, uptake of glyphosate was detected in the tissues of brown trout (Fig. 2). In the range-finding experiment, glyphosate levels were analyzed in the fillets of the fish only, and the detected concentrations of glyphosate were much lower than in the samples in the following experiments containing muscles, head, backbone, and caudal fin tissues. In the latter, concentrations of glyphosate residues increased with increasing exposure concentrations. Measurements in E1 and E2 originally consisted of multiple pools, but due to miscommunication the samples from each treatment in E1 were pooled again for analysis. Instead, in E2 each measurement consisted of two different pools so that a standard deviation could be calculated. In control fish, neither glyphosate nor AMPA was detected in all experiments (Additional file 1: Table S1). Ten-month-old fish in E1 contained less glyphosate and AMPA than the fish in E2 and E3, which were about four months younger. In the latter experiments, tissue from more individuals had to be pooled to achieve the minimum tissue weight required for chemical analysis. Although the age of the fish in E2 and E3 was comparable, the individuals exposed at 15 °C showed higher glyphosate and AMPA uptake than the fish exposed at 7 °C, even though the exposure time in E3 was one week shorter. Roundup® LB Plus resulted in the highest glyphosate uptake in fish in both experiments E1 and E3.

Fig. 2
figure 2

Concentration of glyphosate (purple) and AMPA (green) in the fillets of brown trout of the range-finding experiment, and in the samples consisting of muscle, heads, backbones, and caudal fin tissue. X-axis: exposure concentrations in the range-finding experiment (left; exposure at 7 °C), in experiment 1 (E1; exposure at 7 °C), experiment 2 (E2; exposure at 7 °C), and experiment 3 (E3; exposure at 15 °C); Y-axis: concentration of glyphosate or AMPA after exposure to the respective test concentration. Data are calculated as the means of three respective replicates of each treatment

Full size image

Varying internal concentrations of glyphosate in fish tissues (n = 14) could be explained to 96.11% (p < 0.0001) by the parameters ‘external glyphosate concentration’ (p < 0.0001), ‘age of fish’ (n.s.), ‘temperature’ (n.s.), ‘time of exposure’ (n.s.), and the intercept (n.s.). The glyphosate concentration in the water of the aquaria alone explained 93.14% (p < 0.0001) of the variation in tissue concentrations of glyphosate. When the parameters ‘external glyphosate concentration’ and 'temperature’ were crossed and included in a multiple regression model together with ‘external glyphosate concentration’ alone (p < 0.0001), ‘temperature’ alone (p = 0.0270), and the intercept (p = 0.0466), the crossed parameter ‘external glyphosate concentration × temperature’ contributed significantly to the model (p = 0.0247) which in total explained 97.07% (p < 0.0001) of the variation in internal glyphosate concentration in fish. With increasing temperature, glyphosate concentration in the water and 'temperature x water concentration' the internal glyphosate concentrations in fish tissue increased. A total of 94.21% (p = 0.0062) of the variation in internal AMPA concentrations in AMPA-exposed fish (n = 8) was explained by a model comprising the parameters ‘external AMPA concentration’ (p = 0.0014), ‘temperature’ (n.s.), ‘exposure time’ (n.s.), ‘age of fish’ (n.s.), and the intercept (n.s.). In this context, the AMPA concentration in the water of the aquaria was positively correlated with the internal AMPA concentration and, alone, explained 88.75% (p = 0.0005) of the variation in the concentrations of this compound in fish tissues. In our experiments, the highest AMPA concentration measured in fish tissues also was associated with the highest exposure temperature (15 °C) but, again, due to this single observation, the contribution of the parameter 'temperature' to the model was not significant.

After brown trout larvae in E3 had been exposed to glyphosate, Roundup® LB Plus, and AMPA for two weeks at 15 °C, about half of the fish were transferred to new, clean tanks and kept in pure filtered tap water (iron, particle, and activated carbon filters) completely lacking test solutions for three more weeks as a 'recovery experiment' (E4). Glyphosate and AMPA uptake was determined in the same tissue portions as directly after chemical exposure. It was evident that glyphosate or AMPA detected in the tissues directly after the two-week exposure persisted in the fish tissues after recovery in pure water for a longer period of time than they had previously been exposed to the test solutions (Fig. 3). At the lowest, environmentally relevant glyphosate concentration (56 µg/l), no more glyphosate was detected after three weeks of recovery, whereas at the highest glyphosate concentration (5600 µg/l), 42.42% of the previously measured concentration remained in the tissues. In the Roundup® treatment, 31% of the previously measured glyphosate concentration in the Roundup®-treatment remained in the tissues, and after AMPA exposure, 34.2% of the previously measured concentration remained. A small amount of AMPA was also detected after two weeks of exposure to 5600 µg/l Roundup®, suggesting that glyphosate is metabolized to AMPA [71], and therefore, tissue uptake of AMPA may have been possible.

Fig. 3
figure 3

Concentration of glyphosate (purple and pink) and AMPA (dark and light green) in the samples consisting of muscle, heads, backbones, and caudal fin tissue after an initial two-week exposure experiment followed by a three-week recovery period in pure filtered tap water. X-axis: exposure concentrations in experiment 4 (E4; recovery at 15 °C); Y-axis: concentration of glyphosate or AMPA after exposure to the respective test concentration and after three weeks of recovery, respectively

Full size image

Figure 4 shows the bioconcentration factors (BCFs) for glyphosate and AMPA in the respective exposure experiments. All BCFs were well below 1, indicating that glyphosate does not bioconcentrate or bioaccumulate in fish. However, while the glyphosate concentrations detected in the tissues of brown trout increased with increasing exposure concentrations (Fig. 2), it was apparent that higher exposure concentrations of glyphosate generally resulted in lower BCF values. The BCFs for glyphosate provided as Roundup® were always slightly higher than those for the parent compound. At 15 °C in the third exposure experiment E3, the BCF values for glyphosate were much higher than at 7 °C. In the first exposure experiment E1, the BCF for 56 µg/l glyphosate exposure was calculated 0 because the concentration in the tissues was below the detection limit, but in E3 with younger fish and at a higher exposure temperature, the BCF for 56 µg/l glyphosate was the highest of all. The BCFs for AMPA were not only in a similar range compared with those for glyphosate but also higher in younger fish and at a higher exposure temperature.

Fig. 4
figure 4

Bioconcentration factors (BCF values) of glyphosate (purple) and AMPA (green) in the fillets of brown trout of the range-finding experiment and in the samples consisting of muscle, heads, backbone, and caudal fin tissue. X-axis: exposure concentrations in the range-finding experiment (left), in experiment 1 (E1; exposure at 7 °C), experiment 2 (E2; exposure at 7 °C), and experiment 3 (E3; exposure at 15 °C); Y-axis: Bioconcentration factors for glyphosate and AMPA, after exposure to the respective test concentration

Full size image

Discussion

Although neither strong bioconcentration nor bioaccumulation of glyphosate in animal tissues is expected due to the high water solubility of this chemical, and as confirmed by the low BCF values, glyphosate is generally taken up by fish into their tissues and it is statistically proven that increasing concentrations of glyphosate and AMPA in the surrounding media lead to increased concentrations of these substances in brown trout tissue. The presence of glyphosate in tissues is relevant for the transfer into the food chain and for consumption of fish. Also for other mobile and hydrophilic substances similar BCFs in fish were reported, e.g., for sodium 4,4′-diaminostilbene-2,2′-disulphonate (CAS: 25394-13-2; Log Kow = − 3.99; BCF = 0.200), 2,7-naphthalenedisulfonic acid (CAS: 5460-09-3; Log Kow = − 6.33; BCF = 0.302), and 2-Naphthol-3,6-disulfonic acid (CAS: 15883-57-5; Log Kow = − 4; BCF = 0.350) [72].

Other previously conducted experiments with glyphosate have shown that both glyphosate and AMPA can accumulate to rather high levels in biota. For example, Wang et al. [73] measured the radioactivity after exposure of fish to different concentrations of radiolabeled glyphosate (50 µg/l and 5 µg/l, respectively) and thus the presence of glyphosate in the tissues of two fish, common carp (Cyprinus carpio) and tilapia (Oreochromis mossambicus), and showed that the maximum radioactivity was reached five to seven days after exposure (660 µg/l in carp and 1300 µg/l in tilapia, respectively), although the radioactivity in the surrounding water had decreased continuously. The bioconcentration factors (BCF) published in this study for common carp (22 ± 12) and tilapia (25 ± 19) exceeded the BCFs calculated in the present study by far. In addition to the general problem with radiolabeling, that the radiolabeled atom may have transferred to other molecules [74], the reason for this difference could be that the experiments of Wang et al. [73] were conducted at 22 °C, whereas brown trout in the present study were exposed only at 7 °C and 15 °C, respectively. This assumption is supported by our finding that more glyphosate or AMPA was detected in the tissues of animals exposed at 15 °C than in those exposed at 7 °C, and that bioconcentration factors were higher at 15 °C in the present study than at 7 °C. The ability of pollutants to bioaccumulate in organisms has been shown to depend on temperature, among other physico-chemical properties, i.e., substances may behave differently with increasing temperature. Thus, the toxicity of organic and inorganic contaminants may increase with increasing water temperature [75, 76]. This is particularly important in the context of environmental warming with temperature projections indicating that if climate change will continue unabatedly, water temperatures will increase as much as 4 °C by the year 2050 [77]. Furthermore, plant protection products are used throughout the year for different purposes and therefore also at different temperatures. Rising temperatures in combination with environmental pollution represents a major challenge for the future, and it is likely that the mutual effects will be additive or synergistic [78,79,80]. Recently, increased toxicity was observed after exposure of several GBHs at elevated temperatures (20 °C and 30 °C, respectively) in various marine crustaceans, such as Artemia franciscana or Sphaeroma serratum, and in the American oyster (Crassostrea virginica) [81, 82]. But on the other hand, significant effects of glyphosate on the development of Common toads (Bufo bufo) were observed at lower temperatures (15 °C) [83]. In the present study, mortality and tissue uptake of glyphosate were significantly higher at 15 °C compared to 7 °C. With increasing temperature, the mortality increased, and this was already the case for 56 µg/l glyphosate, an environmentally relevant concentration. The high mortality in E3 is due to the interaction between glyphosate and temperature, as mortality in E4 decreased substantially when glyphosate was no longer used in the water. The effect of glyphosate at 7 °C on mortality in E2 seems rather small, as mortality remained constant at the control level. Therefore, it is obvious that temperature can alter potential non-target effects of glyphosate or pollutants in general. This is notable because ecotoxicological risk assessment studies are typically conducted at one standard temperature, thereby perhaps not adequately examining effects at natural conditions above or below the standard temperature. The effects of temperature are also important in the approval of hazardous substances.

It became evident that 10-month-old fish (E1) took up less glyphosate, glyphosate formulated in Roundup®, and AMPA in their tissues compared to the 6-month-old fish (E2), even when the latter were exposed for one week less (E3). This may probably be due to a generally higher metabolic activity in older fish with higher body mass compared to younger ones [84]. In addition, glyphosate was still detectable in fish tissues after a three-week recovery period. The mechanism, by which glyphosate is processed in the body, is not yet fully known. It has been shown that it is excreted primarily as the unchanged parent chemical via the kidneys and urine [71, 85], with only a small fraction being metabolized. In general, the biotransformation of lipophilic xenobiotics is a biphasic process involving phase I and phase II enzymes, e.g., cytochrome P450 or glutathione-S-transferase. These enzymes work together to increase the body´s ability to eliminate mostly organic, hydrophobic contaminants, thereby preventing harmful accumulation in the body. The highly water soluble glyphosate is likely to be eliminated by a different pathway. However, there is evidence that glyphosate can inhibit cytochrome P450 enzymes and thus alter their detoxification potential for other substances [86,87,88]. An important enzyme of the phase II biotransformation system, glutathione-S-transferase (GST), also protects cells from effects of xenobiotics and endogenous substances and is considered to be beneficial in coping with stressful conditions like oxidative stress [89]. Studies conducted with several fish species have shown that glyphosate also affects this detoxification enzyme. In Anabas testudineus, Heteropneustes fossilis, Rhamdia quelen, and Carassius auratus, a significant reduction of GST activity has been shown after exposure to glyphosate [90,91,92]. In silver barb (Barbonymus gonionotus), however, GST activity was not altered after exposure to 10 mg/kg glyphosate for several days [93], and this was also the case in a study on Prochilodus lineatus, in which no change in GST activity was observed after exposure to 7.5 mg/l and 10 mg/l glyphosate for 24 and 96 h, respectively [94]. These different results make evident that further investigations concerning pathways of detoxification of glyphosate in fish are urgently required.

Our results clearly showed that the BCF values decrease with increasing glyphosate concentrations. Similar observations were made in a study conducted by Contardo-Jara et al. [39] with the earthworm Lumbriculus variegatus, where the BCF was significantly lower in the treatments containing 5 mg/l glyphosate (pure or as active ingredient in Roundup Ultra) than in the treatment with concentrations of 0.05 mg/l and 0.1 mg/l in the exposure medium. The authors suggest that the decreasing BCF value at high concentrations is associated with an increase in the activity of the biotransformation enzyme GST [39], which, however, as previously discussed, is unlikely to explain the result for fish in the present study. The BCF values for the metabolite AMPA were similar to the BCF values for the corresponding glyphosate concentration, whereas the BCF values for the Roundup® treatment were higher. In mammals, amino acid transporters have been reported to be involved in the transport of glyphosate across epithelial tissues [45]. Due to the surfactants and adjuvants in the glyphosate-based formulation, the membrane permeability increases, allowing for easier cellular uptake [51, 95, 96]. This may account for the higher bioconcentration potential of glyphosate when applied in the Roundup® formulation, as seen in the present study.

Farmed animals were shown to take up glyphosate via their food resulting in residues in various organs like intestines, liver, muscles, spleen, and kidneys [41,42,43, 62]. In muscle tissues, the accumulation of glyphosate was always lowest. Due to the small size of the brown trout, it was not possible to analyze solely fillets in the exposure experiments 1–3 due to analytical limitations. However, in the range-finding experiment prior to the other exposure experiments, pure fillets also exhibited lower glyphosate concentrations resulting in lower BCF values than in the bone-containing samples from the following experiments. So, the bioconcentration in our samples containing muscle tissue and skin, head, backbone, and caudal fin tissue was higher than it would be expected in fillets alone. This sounds plausible since it has already been shown that after oral administration of glyphosate to rats, most of the administered compound was found in their bones [97]. This bone-specific accumulation of glyphosate may result from its interaction with calcium [98, 99], possibly resulting in a negative impact of glyphosate on bone mineral density and quality in female Wistar rats as reported by Hamdaoui et al. [100]. At first glance, the lower concentrations of glyphosate in the fillets of fish compared to those in fillets plus head, bone, and caudal fin tissue may imply a lower risk for consumers of fish. However, on one hand, there are fish that are eaten as a whole (e.g., anchovies), and on the other hand, glyphosate residues in fillets and bones are relevant for fish-eating animals, and, consequently, the transfer of chemicals along the food chain. This relevance of glyphosate for terrestrial and aquatic food chains has been emphasized already by Gill et al. [101] or Thanomsit et al. [19], which is in contrast to previous statements of the glyphosate producer [102].

As residues of glyphosate and AMPA in food can be of potential toxicological concern, the so-called maximum residue limits (MRLs) have been introduced and recommended by the European Food Safety Authority (EFSA) [103], ranging from 0.05 to 30 mg/kg for various food sources. The glyphosate MRLs for animal products such as milk and eggs, and muscle, fat tissue, liver, and kidney of pigs, cattle, sheep, goats, horses, and poultry range from 0.05 to 2 mg/kg. Surprisingly, there are no MRLs for fish tissues consumed by humans [104], most likely because glyphosate is only rarely applied directly to water. Furthermore, its hydrophilicity does not urgently call for testing the bioconcentration of glyphosate in fish within regulatory requirements [105]. At 56 µg/l, the environmentally relevant concentration of glyphosate, the measured glyphosate concentration in fish did not reach the above mentioned MRLs for animal products. However, at the highest concentration tested (5600 µg/l), the values were in the range of 0.05 to 2 mg/kg. Due to the increasing consumption of fish [106], however, MRLs for fish samples should be established, especially since in many cultures small fish are eaten as a whole, i.e., with skin and bones.

It has been shown that the herbicide glyphosate and its major metabolite AMPA are present in juvenile brown trout after exposure and persist in the tissues even after three weeks of recovery. This is the most important finding of this study along with the evidence that these chemicals are associated with the skeleton. BCF determination should be used to determine whether bioaccumulation or bioconcentration is actually occurring. Although the BCF in this study was less than 1, we were able to demonstrate that glyphosate enters the tissues and is still detectable after three weeks.

Conclusion

Tissue analysis show that the controversially discussed herbicide glyphosate and its main transformation product AMPA do not bioaccumulate or bioconcentrate but are nevertheless taken up into the body of brown trout and remain detectable even after three weeks of recovery in clean water. The combination of glyphosate and elevated temperature resulted in significantly increased tissue uptake and significantly increased mortality. Although there is a vast number of data especially on the presence of glyphosate in humans and on the short-term effects of high and environmentally irrelevant concentrations of this herbicide in biota, long-term effects resulting from exposure to environmentally relevant concentrations are far from being understood and should be in the focus of future studies in order to realistically assess the risk that glyphosate poses to humans and the environment. Little data are available on AMPA in this regard, although the present study suggests a similar toxicity to the parent compound. In the context of the precautionary principle, the present data for glyphosate and AMPA toxicity and their residues in fish tissue, and their interaction with temperature should be regarded as a warning signal for the—so far largely disregarded—environmentally relevance of glyphosate and AMPA.