Comprehensive evaluation of the toxicity of the flame retardant (decabromodiphenyl ether) in a bioindicator fish (Gambusia affinis)

In recent years, concerns have increased about the adverse effects on health and the environment of polybrominated diphenyl ethers (PBDEs), especially BDE-209, the most widely PBDE used globally. These pollutants derive from e-waste and present different adverse effects on biota. In this work, a toxicological study on mosquitofish (Gambusia affinis) using BDE-209 (2,2′,3,3′,4,4′,5,′5′,6,6′-decabromodiphenyl ether) was carried out. Acute toxicity bioassays were conducted with daily renewal of solutions, using different concentrations of environmental relevance, ranged between 10 and 100 μg L−1 of BDE-209. At 48 and 96 h of exposure, several parameters were evaluated, such as mortality, individual activity (swimming), biochemical activity (catalase; thiobarbituric acid–reactive substances; and acetylcholinesterase), and cytotoxic responses (micronucleus frequencies). In addition, integrated biomarker response and multivariate analyses were conducted to study the correlation of biomarkers. The calculated Lethal Concentration-50 remained constant after all exposure times (24 to 96 h), being the corresponding value 27.79 μg L−1 BDE-209. Furthermore, BDE-209 induced effects on the swimming activity of this species in relation to acetylcholine, since BDE-209 increased, producing oxidative damage at the biochemical level and genotoxicity after 48 h of exposure to 10 and 25 μg L−1 BDE-209. The results indicate that BDE-209 has biochemical, cytotoxic, neurotoxic, and genotoxic potential on G. affinis. In addition, mosquitofish could be used as a good laboratory model to evaluate environmental stressors since they could represent a risk factor for Neotropical species.

In ecotoxicology research works, bioindicators are often used because they are organisms that identify longterm interaction of several environmental conditions, and they also react to a sudden change in a large combination of factors (Hamza-Chaffai 2014). In addition, in ecotoxicological assessments, biomarkers have been used as an early warning signal, being an adequate tool to detect adverse effects on the environment (Newman 2014). Specifically, cellular and biochemical biomarkers are useful as early warning signs of the exposure to environmental stressors in non-target organisms (Newman 2014;Larramendy 2017).
The combination of biomarkers and bioindicator species is used as a valid tool in laboratory studies as it simulates real exposure scenarios, considering actual concentrations of substances, which would cause oxidative damage and generate reactive oxygen species (ROS) (Beliaeff and Burgeot 2002;Newman 2014). Biochemical and cellular biomarkers may determine the potential of environmental stressors as ROS, assessing alterations in antioxidant enzymes, oxidative DNA damage, and physiological stress responses through toxicokinetic and toxicodynamic studies (Beliaeff and Burgeot 2002;Newman 2014;Larramendy 2017). Cellular biomarkers are similarly used in the study of clastogenic or aneugenic events with the determination of micronucleus frequencies (MNs). The damages in the DNA could trigger mutations or chromosomal aberrations, causing physiological dysfunctions, or even cancer (Newman 2014; Larramendy 2017). Although the evaluation of a single biomarker in a bioindicator may not be adequate, a comprehensive evaluation of biomarkers would detect early signals to prevent irreversible effects on ecosystems (van der Oost et al. 2003;Newman 2014). Therefore, cellular and biochemical biomarkers should be analyzed integrally (Adams and Ham 2011;Newman 2014), and biomonitoring programmes should include biomarkers integrated at different biological levels to detect organisms exposed to environmental stressors (Newman 2014;Rautenberg et al. 2015;Paniagua-Michel and Olmos-Soto 2016).
In the last century, the mosquitofish Gambusia affinis was introduced in many areas worldwide to control mosquito larvae. G. affinis is a cosmopolitan species which has a broad diet, an excellent physiological tolerance, with rapid population growth, genetic variability, adequate dispersal trends, and good adaptability to different settings (Rautenberg et al. 2015;Cabrera et al. 2017;Touaylia and Labiadh 2019). Due to these characteristics, the use of G. affinis has been suggested as a good bioindicator for laboratory model species to monitor changes at the different biological organization levels and adverse effects on enzymes, like catalase (CAT), acetylcholinesterase (AChE), and thiobarbituric acid-reacting substances (TBARS) (Rao et al. 2005;Rautenberg et al. 2015;Dang et al. 2017;Díez-del-Molino et al. 2018;Hou et al. 2018;Touaylia and Labiadh 2019). Some studies have examined biomarkers and potential adverse effects of BDE-209 on aquatic vertebrates in vivo (Ross et al. 2009;Jin et al. 2010;He et al. 2011;Zhao et al. 2011;Garcia-Reyero et al. 2014;Li et al. 2014;Xie et al. 2014;Yu et al. 2015;Chen et al. 2016;Zhu et al. 2016;Han et al. 2017;Anacleto et al. 2018;Wang et al. 2018;Espinosa Ruiz et al. 2019). It is worth mentioning previous studies on BDE-209 adverse effects on fishes, such as altered energy budget (Anacleto et al. 2018), disruption of the thyroid system (Noyes et al. 2011;Li et al. 2014;Han et al. 2017), alteration in thyroid and reproduction systems (He et al. 2011;Yu et al. 2015), changes in antioxidant systems, such as glutathione, catalase, and superoxide dismutase (Zhao et al. 2011;Xie et al. 2014), cholinergic and locomotor systems (Goodman 2009;He et al. 2011;Garcia-Reyero et al. 2014;Zhu et al. 2016;Wang et al. 2018), oxidative stress in proteins related to cell cycle (Espinosa Ruiz et al. 2019), cytotoxicity, apoptosis induction, and gene transcriptions (Li et al. 2014).
Although BDE-209 accumulation in sediments represents a long-term threat to biota, its decomposition products are even more dangerous since BDE-209 breaks down into very persistent, bioaccumulative, toxic, and mobile PBDE congeners in the environment (Ross et al. 2009;Hu et al. 2010;Kuo et al. 2010). Unfortunately, BDE-209 adverse effects have been scarcely studied. Similarly, their impact on biota in Neotropical regions, where BDE-209's presence has been reported, has not been investigated yet. The studies above mentioned dealt with the adverse effects in isolation, without considering the comprehensive multisystemic or integrative responses that occur when an organism is exposed to this kind of emergent environmental stressor. Thus, the aim of this work was to assess a comprehensive study of biomarkers at different biological levels to determine BDE-209 adverse effects before they become irreversible. In this context, Gambusia affinis was used as a vertebrate model, and as a representative species that inhabits in Neotropical regions. To this end, cytogenetic, biochemical, and behavioural biomarkers were used in an acute toxicity test for 96 h, at different concentrations of BDE-209.

Chemical analysis
Quantitative analysis of BDE-209 was carried out by Shimadzu Model 2010 GC-MS equipped with an AOC-20i auto injector (Shimadzu, Japan), using negative chemical ionization (NCI) in the selected ion monitoring (SIM) mode. A CP-Sil 13 CB (12.5 m × 0.25 mm i.d., 0.2 μm film thickness) capillary column was used. Ion fragments m/z 79, 81, 486.7, and 488.7 were monitored for BDE-209, according to the method proposed by Peng et al. (2009). The quantification, performed by external standard method, had a limit of 0.6 ng L −1 .

Model organism
Males of Gambusia affinis were obtained from an unpolluted permanent pond located in El Volcán, San Luis Province, Argentina (33° 15′ 01″ S, 66° 11′ 43″ W) with the authorization of the Ministry of Environment, Agriculture and Production of San Luis Province (Resolution 49-PMA-2019). Mature adults of G. affinis were transported and acclimated to the laboratory conditions (photoperiod 16:8; temperature 25 ± 1 °C; daily renewed dechlorinated tap water: pH 7.12, conductivity 412 µS.cm −1 , hardness 186 mg L −1 CaCO 3 , alkalinity 250 mg L −1 CaCO 3 , nitrate 0.6 mg L −1 ) for 2 weeks before conducting the experiments, according to the protocols of the Institutional Animal Care and Use Committee (CICUA, protocol Q-322/19), from National University of San Luis.

Ecotoxicity bioassay
Toxicity tests were performed following standardized methods proposed by the United States Environmental Protection Agency (USEPA 2002) with minor modifications for local species (Vera-Candioti et al. 2010). In all experiments, five individuals per replicate were placed in 1-L glass jars, with 1 g L −1 density of the organisms (n = 15 per treatment). Negative and positive controls, as well as increasing gradient concentrations of BDE-209 with test solutions, replaced every 24 h in acute exposure (96 h) were used. Fishes were starved for 24 h before starting the experiments, and were not fed during the bioassays, which were carried out under controlled conditions, such as photoperiod (16:8), daily renewal of the solutions, and bioterium temperature (25 ± 1 °C). Dechlorinated tap water (pH 7.12; conductivity 412 µS cm −1 ; hardness 186 mg CaCO 3 L −1 ; alkalinity 250 mg CaCO 3 L −1 ; chloride 7.1 mg L −1 ; nitrate 0.6 mg L −1 ; sulphate; 17 mg L −1 ; sodium 21.5 mg L −1 ; calcium 54.3 mg L −1 ; potassium 1.9 mg L −1 , and magnesium 8.1 mg L −1 ) was used for the experiments. BDE-209 lethal and sublethal concentrations for fish exposure were selected based on previous studies on zebrafish model (Garcia-Reyero et al. 2014;Zhu et al. 2016;Han et al. 2017), and considering the environmental relevant concentrations reported in the literature (Mackintosh et al. 2015;McGrath et al. 2017). Low concentrations (10 and 25 µg L −1 ) were determined according to Deng et al. (2016) and Kwan et al. (2013); and the highest concentrations (75 and 100 µg L −1 ) were defined pursuant the zebrafish experiments above mentioned (Garcia-Reyero et al. 2014;Zhu et al. 2016;Han et al. 2017). Thus, G. affinis fishes were exposed to concentrations equal to 10, 25, 50, and 100 µg L −1 of BDE-209 dissolved in methanol. The positive control group was prepared adding methanol (100 µg L −1 ). The solvent final concentration was the same in all the treatments, always lower than 0.1% (Kaviraj et al. 2004). All treatments (control and BDE-209 exposed groups) were triplicated. To evaluate biochemical and cellular biomarkers, fishes were euthanized by dissection at the operculum level. After that, blood samples were extracted and biochemically analyzed. All fishes were anesthetized and sacrificed after 48 and 96 h of exposure. The same procedures were followed in all the treatments.

Cytogenetical endpoints: MNs assays
Peripheral blood of each fish was smeared onto pre-cleaned slides according to Vera-Candioti et al. (2010). Then, slides with blood samples were air dried, fixed with methanol (4 °C), and stained with Giemsa solution (5%). Finally, the coded slides were quantified. The same researcher conducted all the experiments at microscopy 1000 × magnification. The MN frequency was determined by analyzing a total of 1000 mature erythrocytes, expressed as the total number of MN per 1000 cells (Fenech 2007).

Biochemical endpoints: CAT, TBARS, and AChE activity
Supernatants from homogenates of whole fish were obtained by applying the methodology proposed by Brodeur et al. (2017), with minor modifications. Briefly, post-mitochondrial supernatant was prepared (ice bath cooling) from a 1 mL homogenate of fish tissues with a buffer Tris 50 mM (pH 7.4) containing 1 mM EDTA and sucrose 0.25 M. The homogenate was centrifuged at 1 10 4 × g for 10 min at 4 °C. The supernatant was used to measure the following biochemical biomarkers: (i) the CAT activity was determined by measuring the decomposition of hydrogen peroxide at 240 nm (37 °C, 2 min), using a molar extinction coefficient of 43.6 M −1 cm −1 . The reaction mixture consisted of 20 μL of pure sample, 40 μL of H 2 O 2 (10%, v/v), and 1900 μL of PBS (pH 7, 100 mM); (ii) the lipid peroxidation was determined by the reaction of thiobarbituric acid-reactive substances (TBARS) according to the method of Buege and Aust (1978), with minor modifications for aquatic vertebrates. The lipid peroxidation in whole fish was determined by measuring the formation of the colour produced during the TBARS reaction. To this end, fish homogenate (20 μL) and 380 µL of the reaction mixture (trichloroacetic/thiobarbituric acid) were incubated at 90.0 ± 0.5 °C for 15 min; then the coloured product was cooled and centrifuged at 7500 × g for 8 min. Finally, the absorbance was measured at 530 nm. Lipid peroxidation or TBARS levels were expressed as mmol MDA mg −1 protein (Buege and Aust 1978); and (iii) AChE activity was determined by the method of Ellman et al. (1961). The reaction mixture consisted of 150 µL of PBS (100 mM, pH 8), 50 µL of acetylthiocholine iodide (1 mM), 150 µL of 5,5′-dithiobis-(2-nitrobenzoic acid) (0.5 mM), and 10 µL of pure sample. The change in absorbance was recorded at 412 nm (37 °C, 1 min). The enzymatic activity was calculated using a molar extinction coefficient of 14.150 M −1 cm −1 . Protein concentration was determined according to the Bradford method (Bradford 1976). All biochemical enzyme reactions and protein determinations were measured using a spectrophotometer (Rayleigh-Model UV2601 UV/VIS-Double Beam Spectrophotometer, China).

Individual endpoints
Mortality was considered the lethal endpoint. Fishes were examined daily, and mortality criterion was the presence of cadaverous appearance, non-cardiac movement in stereoscopic microscopy, and no movement detection after gentle touching with a glass rod, in contrast to control individuals. Dead individuals were fixed in formaldehyde (10%, v/v). The values of Lethal Concentration (LC-50), No Observed Effect Concentration (NOEC), and the Lowest Observed Effect Concentration (LOEC) were determined in fishes at each exposure time (48 h and 96 h).
The altered swimming activity in fishes was evaluated by direct observation according to parameters proposed by Little and Finger (1990), occurring frequently during toxicant exposure. The parameters were the following: changes in water column position (surfacing, resting on bottom), swimming posture (head-up swimming), body movements (increased or decreased waveform of body movement), or swimming patterns (frequent turns or spiralling). Extreme cases such as loss of coordination, convulsive movements, or loss of equilibrium were also considered. The all-or-none occurrence of activity level has been proposed as a sublethal biomarker, and it has been successfully used to describe swimming alterations after the exposure to several environmental stressors (Shuman-Goodier and Propper 2016).

Integrative response of biomarkers
To integrate the different results, two methods were reliably performed: (i) the integrated biomarker response (IBR) index calculated according to Baudou et al. (2019), considering the following biomarkers: CAT, TBARS, AChE, swimming activity, and MNs. The IBR provided a numeric value that integrated all these responses. High IBR values indicate high stress levels (Baudou et al. 2019); (ii) the principal component analysis (PCA) was used to determine the implications of each biomarker at each concentration level taken into account in this study. The significance of correlations was examined by simple linear regression, and correlation analyses were obtained with R software v.2.11.1. The level of significance was set at α = 0.05 for all tests, unless otherwise indicated.

Statistical analysis
LC-50 values, concentration response curves, and ecotoxicological parameters, such as slope and correlation coefficient at different sampling times (24 to 96 h) with 95% of confidence limits, were estimated using the USEPA Probit Analysis (Finney 1952) with the package "ecotoxicology" for R software v.2.11.1 (R Core Team 2010, October 14, 2015). All significance tests for regression and correlation were performed according to Zar (2010).
The proportion of fishes affected by test chamber was calculated for mortality, swimming activity, CAT, TBARS, AChE, and MNs. Angular transformation was used to analyze data. ANOVA one-way with Dunnett's test was performed to compare the different test concentrations in the control group, and to obtain NOEC and LOEC values. Homogeneity of variances and normality for ANOVA assumptions were corroborated by Barlett's test and x 2 test, respectively. When these assumptions could not be met, a non-parametric test was performed (Zar 2010).

Lethal effects
LC-50 values were determined by the mortality data obtained after fish exposure to BDE-209. They were evaluated at all times. With respect to our experiments, no mortality in fishes at the tested concentration of methanol and in the negative control groups was observed. LC-50 values remained constant after all exposure times (24 to 96 h), and the corresponding value was 27.79 μg L −1 BDE-209 (confidence interval 95% = 19.68-37.42 μg L −1 BDE-209; R 2 = 0.997, p < 0.05) for G. affinis (see Supplementary Information Section). In this case, NOEC and LOEC values were 10 and 25 μg L −1 BDE-209, respectively, at all exposure times.

Cytogenetical endpoints
It is important to note that the endpoints assessed were not obtained at 100 µg L −1 concentrations of BDE-209, since no fish survived after 48 h or 96 h of exposure. The results reveal that BDE-209 induced an increase in the frequency of MNs in G. affinis erythrocytes after 48 h (p < 0.05), see Table 1. The exposure to BDE-209 showed an acute adverse effect with a significant increase in micronucleus frequencies at sublethal concentrations, such as 10, 25, and 50 μg L −1 of BDE-209 (p < 0.05), in comparison with the negative control group. Likewise, at 48 h, the respective LOEC value was 10 μg L −1 of BDE-209. On the other hand, BDE-209 did not induce an increment of MNs frequencies in G. affinis erythrocytes after 96 h of exposure, at sublethal concentrations (p > 0.05).

ROS enzymes
The results in Table 1 indicate that all enzymatic systems were altered after 48 h of exposure to BDE-209, in comparison with the negative control group. Particularly, the following biochemical biomarkers were altered by the action of BDE-209 after acute exposure: the antioxidant response of carbohydrates and lipids measured through CAT and TBARS, respectively; and the cholinergic system measured by AChE response. The individuals exposed to 25 and 50 μg L −1 of BDE-209 showed a decrease in the catalase activity after 48 h of exposure (p < 0.05, Table 1) with respect to the negative control group. However, no significant differences were observed in the antioxidant response of CAT after 96 h at the concentrations tested (p > 0.05).
The evaluation of oxidative degradation of lipids or lipid peroxidation in fishes showed that 50 μg L −1 of BDE-209 concentration altered the activity of TBARS by inducing a significant decrease after 48 h of exposure (p < 0.05, Table 1), with respect to the negative control group. On the contrary, no significant differences were observed in the response of lipid peroxidation after 96 h at the concentrations tested in relation to the negative control group (p > 0.05).

AChE response
A BDE-209 concentration of 25 and 50 μg L −1 increased the activity of AChE after 48 h of exposure with respect to the negative control group (p < 0.001, Table 1). Additionally, an increment in the AChE activity was observed at the lowest

Individual endpoints: swimming activity
The analysis of swimming activity in G. affinis revealed that a BDE-209 concentration of 50 μg L −1 caused a significant increase of alterations in this endpoint after 48 h of exposure (p < 0.001), and this trend was also observed at 96 h (p > 0.001) with respect to the negative control group.

Principal component analysis
Biomarkers exhibited a different response when fishes were exposed to BDE-209 (Fig. 1). The 94.9% of the variability (PC1 = 70.5%, PC2 = 24.4%) was explained by two principal components (PC) obtained by the PCA correlations among the response variables, at 48 h of exposure. In addition, the reduction of the dimensionalities through the PCA demonstrated a concentration gradient, separating the lowest concentration of BDE-209 (10 μg L −1 ) from medium and highest concentrations (25 and 50 μg L −1 ). The activity of CAT and TBARS showed a positive correlation (r = 0.78), but negative correlations with AChE and MNs, which showed a positive correlation (r = 0.85). It is important to note that CAT and TBARS were negatively correlated with mortality (r = − 0.91). In addition, at the lowest concentration (10 μg L −1 ), only antioxidant biomarkers showed a response, while at a medium concentration (25 μg L −1 ), the response was given by AChE and MNs. Finally, at the highest concentrations assayed (50 and 100 μg L −1 ), the mortality was the prevalent effect. Furthermore, a gradient of adverse effects related to an increase in BDE-209 concentration could be observed. It was not possible to obtain the analysis of PCA at 96 h because AChE was the only biomarker to respond.

Discussion
The bioindicator fish G. affinis shows in our studies that BDE-209 produces toxicity at different levels until death. This contributes to the state of the art, due to the fact that few studies of LC-50 in non-vertebrate aquatic organisms have been reported (Davies and Zou 2012; Zhang et al. 2013; Xiong et al. 2018). An effective concentration (EC-50), similar to LC-50, was reported for the algae Heterosigma akashiwo and Karenia mikimotoi with values around 22.58 and 120.8 μg L −1 BDE-209, respectively (Zhang et al. 2013). Furthermore, in Daphnia magna, it was not possible to obtain LC-50 values after 48 h of exposure to a range 0.3-500 μg L −1 (Davies and Zou 2012). However, Xiong et al. (2018) reported that a mixture containing 125 μg L −1 (acute exposure) and 25 μg L −1 of BDE-209 (chronic exposure) caused 50% of immobility (IC-50). Finally, there was mortality evidence only in zebrafish (Danio rerio) when exposed to 300 μg L −1 for 96 h, although no LC-50 value was reported (Han et al. 2017). In this context, the present work shows the first LC-50 value for an aquatic vertebrate after acute exposure to BDE-209, resulting in a relatively high sensitivity value for the species tested. Similar LC-50 values have been estimated for other PBDE congeners in D. rerio embryos, such as 250, 350, 520, and 840 μg L −1 for BDE-28, BDE-47, BDE-99, and BDE-100, respectively (Usenko et al. 2011). Finally, it is important to highlight that BDE-209 concentration in water reported by Deng et al. (2016) (equal to 22 μg mg −1 ) is risky for the bioindicator species G. affinis, since the LC50 value obtained in this work was 27.79 μg L −1 .
In addition, this work shows the sublethal effects of BDE-209 separately, using the biomarkers as early warning signals during the first 48 h of exposure to BDE-209. An increase in cytotoxic and genotoxic effects was observed with increased MNs frequencies at all concentrations tested (48 h). These results are consistent with those of Jin et al. (2010), which showed that BDE-209 was cytotoxic and genotoxic for rainbow trout cell lines, producing apoptosis, metabolic activity alterations (MTT assay), and increase of ROS species.
According to Jin et al. (2010), BDE-209 generated ROS species in O. mykiss, as these authors corroborated it by the alteration of enzymes linked to oxidative stress such as CAT and TBARS. Moreover, the response observed in CAT activity against BDE-209 was consistent with that observed by Xie et al. (2014), where Carassius auratus was exposed for 96 h at concentrations between 1 and 5 μg L −1 of BDE-209. The inhibition of antioxidant enzymes, such as CAT and TBARS, was due to the loss of the function of the antioxidant system in overproduced ROS scavenging (Xie et al. 2014). Previous reports indicated that the decrease of enzyme activity could reflect a lack of a defence system response and antioxidant mechanisms to eliminate the highly reactive xenobiotics produced in cells. This situation could be considered a non-adaptive response to counteract ROS generation (Rautenberg et al. 2015;Touaylia and Labiadh 2019), caused by these emerging contaminants.
The reaction known as lipid peroxidation acts as a chain reaction, since the formed peroxyl radical becomes peroxide in contact with another fatty acid, forming, in this case, a new ROS (Esterbauer et al. 1992;Touaylia and Labiadh 2019). In addition, the peroxyl radical can release several toxic aldehydes, including MDA or hydroxynonenal (Touaylia and Labiadh 2019). The analysis of the lipid peroxidation effect shows a marked inhibition of TBARS only at the highest sublethal concentration tested (50 μg L −1 of BDE-209). In this regard, our results are not in agreement with Zhu et al. (2016), who reported an increase of TBARS after the exposure of zebrafish to BDE-209 as a protective response. Probably, as proposed by Touaylia and Labiadh (2019), the poly-unsaturated fatty acids are the target of attack by the hydroxyl radical, capable of extracting hydrogen from the carbons located between two double bonds to form a ROS. In addition, the peroxyl radical can release several toxic aldehydes, including MDA or hydroxynonenal (Touaylia and Labiadh 2019).
In summary, the biochemical analysis of some antioxidant systems indicates that BDE-209 can decrease and/or inhibit systems that act against free radicals in fishes such as G. affinis. In this respect, and according to Touaylia and Labiadh (2019), it is worth noting that the recorded adverse effects on the antioxidant system are related to two factors: the toxicity of the environmental stressor and the sensitivity of the target species. Some environmental stressors, such as BDE-209, may lead to an excessive stimulation of the cholinergic system (Wang et al. 2018). The environmental stressors can interfere with the catalytic process of the cholinergic system, based on their structural similarity with AChE (Xie et al. 2014;Wang et al. 2018;Touaylia and Labiadh 2019). This situation affects the swimming activity through tremors, convulsions, and erratic or lethargic swimming (Xie et al. 2014;Zhu et al. 2016), as observed in this study. Furthermore, if AChE is inhibited by these emergent pollutants, the neurotransmitter (e.g. acetylcholine) could be accumulated in the synaptic space, leading to muscle tetany and death (Wang et al. 2018). This type of AChE response has already been observed in G. affinis exposed to other environmental stressors, as reported by Rao et al. (2005), who related AChE alterations to locomotor and behavioural problems in the mosquitofish. In addition, the decrease in AChE and CAT response at 96 h shows that these biomarkers operate "down stream" from the original toxic lesion. Enzyme depletion occurs as a consequence of injury and damage to different cell types. Thus, CAT and AChE could be considered early biomarkers of adverse effects (Walker 2009). Finally, our results confirm that BDE-209 induces neurotoxicity in G. affinis, and AChE determination resulted in an excellent biomarker to evaluate the effects produced by BDE-209.
In recent years, Newman (2014) highlighted the importance of evaluating the correlation of biomarkers as a whole and not separately. This information helps us to understand not only the susceptibility of organisms to environmental stressors, but also their mode of action and toxicity, which can later be used as early warning signals in environments that are disturbed or contaminated by the presence of environmental stressors, as we previously highlighted (Pérez-Iglesias et al. 2020). Furthermore, these results show that the evaluated endpoints respond to the concept of biomarkers proposed by Walker (2009), who stated that the analyzed endpoints are useful biomarkers. The BDE-209 adverse effects induce alterations in the physiological responses, which are shown when evaluating biomarkers at different levels of biological organization. At this point, there is a progression of effects from cellular to individual level, ending in death. This situation is evidenced in the PCA analysis, where it is observed that there is a progression of adverse effects as the BDE-209 concentration increases. In addition, BDE-209 adverse effects are related to the levels of cell organization (effect at the genetic and cellular level). As observed in Fig. 1, there is a correlation with MNs, TBARS, CAT, AChE, and with low concentrations of BDE-209, while at higher concentrations of BDE-209, a response at organism level is observed, with irreversible effects, such as swimming activity or even death. This shows the importance of using several biomarkers at different levels, considering a dose-effect relationship in which the lowest doses produce adverse effects at cellular or cytogenetical level, accumulating biological impairments progressively with an increasing dose. Therefore, integrated analysis results in a diagnostic pattern of effects. Complementing this information, Beliaeff and Burgeot (2002) and Baudou et al. (2019) described that IBR was a useful tool to the health status of organisms subjected to stressful conditions in an easy and comprehensive way, providing a graphical summary and an integrated numerical value of the different biomarkers. In this regard, in the present study, the groups of tadpoles that did not die and were exposed to the highest concentrations of BDE-209 (50 µL) showed total IBR values greater than the control group (Fig. 2), indicating that exposure to the BDE-209 induces a greater impact on most of the biomarkers used, such as biochemical and cytogenetical. In sum, this approach provides a simple tool for a general description of the health status of organisms, combining the different biomarker signals and demonstrating that BDE-209 causes stress in the tested fishes that inhabit Neotropical regions. In conclusion, and in agreement with other authors (Van der Oost et al. 2003;Newman 2014), we recommend the use of this type of approach for ecotoxicological studies, since it is possible to differentiate the groups of anurans exposed to environmental stressors from those that are not, as seen in the IBR results. In this vein, as the ecotoxicological information that evaluates the correlation of adverse effects at different levels of biological organization is scarce, this work makes an important contribution, generating novel information on the biomarker concept, not provided by an individual and separate analysis of each biomarker.

Conclusion
This work has shown that the species G. affinis constitutes a good model organism for toxicological evaluations, and can be used as a bioindicator species. In addition, the studied biomarkers provide an integrated response to environmental stressors, such as BDE-209, and an early detection of the sublethal effects of emerging pollutants, like flame retardants. This situation highlights the importance of these biomarkers at higher levels of organization. However, further studies should be carried out in order to deepen this research work. Likewise, complementary studies about locomotor, behavioural, and swimming effects on this species are suggested in order to evaluate the effects of emerging pollutants on the Neotropical biota, due to their constant growth and the danger that they pose to aquatic ecosystems.
Funding This study was funded by Postdoctoral Internal Scholarship, granted by the National Commission for Scientific and Technical Research (CONICET), File No. 004121/17. It also received funds from the Research Project of National University of San Luis "Environmental quality of aquatic ecosystems: analytical methodologies for the determination of compounds of environmental interest. Line 2: Development and validation of sensitive and selective analytical methodologies for the determination of compounds of environmental interest" (UNSL-Ord. CS No 64/15). In addition, the authors received financial support from Research Project "Assessment of the quality and toxicological risk of water and soils, contaminated with analyts from electronic waste using native bioindicators from semiarid regions" (PICT-2018-01067), and "Evaluation of the ecological status of water resources with anthropic disturbance through multiple environmental quality indicators" (PICT-2018-01067) which are funded by National