Acute toxicity of three herbicide formulations of Astyanax altiparanae (Characiformes, Characidae), an emerging neotropical fish model species

ABSTRACT Herbicides are used in agriculture to control harmful crop weeds, prevent algae proliferation, and enhance macrophyte growth. Herbicide contamination of water bodies might exert toxic effects on fish in different development stages. Sperm, embryos, and adults of Astyanax altiparanae were used as a model to examine the detrimental effects of the following herbicide formulations: Roundup Transorb® (glyphosate), Arsenal® NA (imazapyr), and Reglone® (diquat). The lethal concentration 50 (LC50) values for adults using glyphosate and imazapyr were 3.14 mg/L and 4.59 mg/L, respectively, while the LC50 was higher than 28 mg/L for diquat. For the initial stages of embryo development, LC50 values were 16.52 mg/L glyphosate, 9.33 mg/L imazapyr, and 1084 mg/L diquat. Inhibition of sperm motility was noted at 252 mg/L glyphosate, 137 mg/L imazapyr, and 11,300 mg/L diquat, with an average sperm viability of 12.5%, 73.2%, and 89.3%, respectively, compared to 87.5% detected to control. A. altiparanae exhibited different sensitivities to the herbicide formulations investigated in the developmental stages evaluated. Roundup Transorb® exposure was more toxic for adults, while Arsenal® NA was most harmful for early embryonic development and inhibited sperm motility. Reglone® demonstrated low toxicity for A. altiparanae compared to Roundup Transorb® and Arsenal® NA. A. altiparanae may be considered an emerging fish model for toxicological studies for the neotropical region due to its wide distribution and biological characteristics.


Introduction
Herbicides are chemicals applied in agriculture to control weeds present in crops and combat uncontrolled proliferation of algae or aquatic weeds, termed chemical management (Abbas, Naveed, and Kremer 2018;Gage, Krausz, and Walters 2019). Americans apply the highest level of pesticides in agriculture globally, averaging more than 1 megaton (Mt)/year. The highest pesticide levels noted were per cropland area (2.83 kg/ha per year). The quantity of herbicides used in the Americas increased from 362 to 852 kilotonnes from 1990 to 2020 (https://www.fao.org/3/ cc0918en/cc0918en.pdf).
Only Sonar AQ NA (fluridone) is authorized (IBAMA n° 2910/2000) for aquatic vegetation management in Brazil. Fluridone is inefficient to enhance various macrophyte growth attributed to genetic variability in aquatic plants due to differences in phytoene desaturase gene, the enzyme involved in carotenoid biosynthesis (Benoit and Les 2013). Therefore, population genetic studies of within-and among-population genetic variability are essential to managing submerged aquatic plant species (Thum et al. 2020). Other herbicides were also used experimentally to determine their efficacy in controlling aquatic vegetation, including Reglone® (Esteves et al. 2020;Pitelli et al. 2011), Arsenal® NA and other imazapyr formulations (Carvalho et al. 2005;Dugdale et al. 2020), and Roundup Transorb® and Roundup Original® (Maria et al. 2020;Souza et al. 2020).
Different herbicides are registered for aquatic use in other countries. Several herbicides were registered by the U.S. Environmental Protection Agency (EPA) and the Florida Department of Agriculture and Consumer Services (FDACS) for treatment and management of aquatic weeds, including imazapyr, diquat dibromide, glyphosatebased herbicides, and others (https://plantsarchive.ifas.ufl.edu/manage/developingmanagement-plans/chemical-controlconsiderations/herbicides-registered-for-use-inflorida-waters/). Diquat is registered for general aquatic use in Canada. However, other herbicides, such as glyphosate and imazapyr, are only permitted under emergency registration (Breckels and Kilgour 2018).
The use of herbicides that might be more effective in controlling macrophytes requires knowledge of the toxicity of these herbicides on the aquatic community. The applications of Roundup Transorb® and Reglone® are not permitted at a distance of less than 500 m from the drinking water supply and 250 m from water bodies. In comparison, Arsenal® NA is recommended to maintain a containment strip 30 m away from water bodies without applying the product. However, aquatic toxicity tests are essential to establish safety since different sources including industry, agriculture, human recreational activities, and traffic are known to contaminate aquatic environments (Amoatey and Baawain 2019). Other contamination sources include immediate release from treated or untreated wastewater, run-off, atmospheric deposition (including spray drift), or indirectly due to leaching (Holt 2000;Riter et al. 2021;Vryzas 2018). In aquatic ecosystems, the deposit of herbicides is related to the water flow on the soil surface during precipitation, carrying herbicide molecules adhered to soil particles (Vryzas 2018). Water contamination includes directly applying to the water bodies for aquatic vegetation control (Sesin et al. 2018). In the aquatic ecosystem, pesticides are found in soluble and adsorbed forms, and the relations between forms can change in new geochemical and biological conditions in the environment, once epilithic biofilms can interact with pesticides, adsorbing soluble compounds or incorporate contaminated sediments in their extracellular polymeric matrix (Rheinheimer dos Santos et al. 2020).
Assessing the potential off-target effects of herbicides on the aquatic community is of both ecological and health importance. Herbicide formulations may affect non-target organisms such as bacteria, earthworms, insects, and amphibians (Li et al. 2022;Strilbytska et al. 2022;Trudeau et al. 2020;Zaller and Brühl 2019), including fish and humans (Costa et al. 2022;Milesi et al. 2021). The mechanisms of action and effects of herbicides on fish systems and organs were reviewed by Ribeiro et al. (2022). Briefly, herbicides enter mainly through respiratory, gastrointestinal, and dermal routes, which subsequently adversely affect physiological functions in the organism. Frequently, herbicides inhibit metabolic enzymes, disrupt electron transport in mitochondria, and increase reactive oxygen species (ROS) and nitrogen reactive species (RNS) production, leading to oxidation of lipids, proteins, and DNA. In addition, herbicides alter gene expression (Gaaied et al. 2019). Herbicides produce morphological and metabolic changes, including impaired cellular function, change in cell number, modify the expression of antioxidant enzymes, and cause tissue damage (Marino et al. 2021). Herbicides may affect human health depending upon the chemical class, concentration, time, and exposure route (Marin-Morales, de Campos Ventura-Camargo, and Hoshina 2013).
Atrazine is an endocrine-disrupting chemical primarily targeting the neuroendocrine system and associated axes, acting as a reproductive toxicant by attenuating the luteinizing hormone (Stradtman and Freeman 2021). Glyphosate exerts cytotoxic and genotoxic effects, induces inflammation, and affects lymphocyte functions and the interactions between microorganisms and the immune system (Costa et al. 2022;Peillex and Pelletier 2020). Diquat irritates the mucous membranes of the eyes, nose, and upper respiratory tract, which may lead to lung disorders. Diquat intoxication produces nausea, vomiting, chest pain, development of cardiogenic shock, and toxic effects on the liver and kidneys (Yastrub, Omelchuk, and Yastrub 2020).
Previously Gonçalves et al. (2018) reported that glyphosate-based herbicides decreased sperm motility at 50 µg/L glyphosate and at concentrations from 300 to 1800 µg/L the herbicide reduced sperm viability and survival rates in Astyanax lacustris utilizing a glyphosate formulation diluted in distilled water. However, Gonçalves et al. (2018) did not indicate the minimal concentration of glyphosate in which sperm motility was observed. Lopes et al. (2014) found that exposure of Danio rerio to pure glyphosate at 5 and 10 mg/L for 24 or 96 hr diminished sperm motility at both concentrations and exposure periods, while mitochondrial functionality and membrane and DNA integrity were affected only at the highest concentration during both exposure times. These findings suggest that glyphosate and glyphosate-based herbicides might affect the potential of fertilization and, consequently, fish reproduction (Webster et al. 2014).
Teleost fish rely on external fertilization, which involves the release of gametes into the surrounding medium. In this manner, gametes are exposed to contaminants present in water, such as herbicides, which may reduce sperm motility (Gonçalves et al. 2018). Fish exposed directly to herbicide formulations exhibited altered sperm quality. Sánchez et al. (2017) noted that Jenynsia multidentata exposed (24 or 96 hr) to 0.5 mg/L of glyphosate using Roundup Original®, Roundup Transorb®, and Roundup WG® formulations, a marked reduction in spermatozoa motility mediated by all formulations. Exposure of Odontesthes humensis to 7.8 mg/L glyphosate (Roundup®) for 24 hr produced a significant decrease in sperm concentration and motility parameters as well as increased membrane fluidity, ROS production, lipid peroxidation, and diminished mitochondrial functionality, lowering the fertilization potential of spermatozoa (Silveira et al. 2019).
Herbicides may affect fish at different stages of development, from gamete to embryo and mature adult (Cruz et al. 2015;Flach et al. 2022;Folmar, Sanders, and Julin 1979;Lopes et al. 2014;Paul et al. 1994). If one of these life stages is exposed to risk in a given environment, the survival and reproduction of fish might be endangered. Based upon this observation, it is crucial to determine toxicological sensitivities at each stage of development.
In several studies, herbicide formulations were used rather than pure compounds. It is essential to note that certain surfactant substances in these formulations may be more toxic than the active herbicidal components themselves (Folmar, Sanders, and Julin 1979;Howe et al. 2004). Experiments using pure herbicidal compounds were employed to examine the toxicity in fish and amphibians. Santos et al. (2019) used a sublethal concentration of a glyphosate-based herbicide (0.1 or 1 mg/L) in O. mykiss. Data demonstrated that no marked adverse effects on embryonic or larval survival or developmental abnormalities were observed. However, Xenopus laevis exposed to the same compound at sublethal concentrations resulted in abnormal larval development, including a reduced body length and mobility of embryos as well as smaller eyes, brains, and cranial cartilages, resulting in shorter cranial nerves. In addition, the herbicide affected cardiac development as evidenced by reduced heart rate and atrial size and a reduced expression of marker genes in different tissues and developmental stages (Flach et al. 2022). Differences in herbicide sensitivity using Cyprinus carpio and D. rerio exposed to pure glyphosate were identified during the early developmental stages, especially at high concentrations from 10 to 50 mg/L (Fiorino et al. 2018). Folmar, Sanders, and Julin (1979) assessed the acute toxicity of pure glyphosate, Roundup formulations, and surfactants for different fish species. Surfactant and Roundup formulations presented similar toxicities, while technical glyphosate was found to be a relatively less toxicant. The 96 hr LC 50 values for pure glyphosate were higher than LC 50 for surfactants, in the order of 70-fold higher in O. mykiss, 97-fold in Pimephales promelas, 10-fold in I. punctatus, and 47-fold in Lepomis macrochirus. Roundup increased toxicity when fish were exposed to higher temperatures and alkaline pH values (Folmar, Sanders, and Julin 1979). The toxicities of 19 adjuvants, commonly employed as surfactants in aquatic herbicide applications, were determined in juveniles of Lepomis macrochirus (Haller and Stocker 2003). Two surfactants, classified as members of the ethoxylated tallow amine group, were found in some formulations of glyphosate-based herbicides and are high toxicants with LC 50 values of 1.6-2.9 ppm. Glyphosate's Appendices and the given LC 50 values of technicalgrade glyphosate and herbicide formulations indicated that pure compound is less toxic to fish than the herbicide formulations (Durkin 2010).
Environmental contamination might occur from exposure to surfactants, pure herbicidal compounds, or both. These compounds exert distinct mechanisms of toxicity associated with factors such as differing environmental distributions and degradation schemes. Glyphosate was detected in surface water as well as in sediments. Ronco et al. (2016) quantified glyphosate levels in the main tributaries of the Paraná River Basin, the natural environment of A. altiparanae. These investigators found glyphosate in 15% of the water samples at an average of 0.6 µg/L. Glyphosate was detected at even higher levels in surface water from areas of extensive agriculture, with concentrations ranging from 100 µg/L to 700 µg/L (Peruzzo, Porta, and Ronco 2008). In several regions, the application of herbicides occurs during the reproductive season for local fish, exposing fish sperm, embryos, and larvae to high glyphosate levels for a short period. The lack of data on the sensitivities to herbicides in neotropical fish species across different stages of development requires future research efforts to address the lethal and sublethal effects of commercial herbicides on these populations, including studies into reproduction efficiencies, abnormal developmental rates, gamete quality, and survival rates.
Glyphosate levels in sediment are higher than surface water (Peruzzo, Porta, and Ronco 2008;Ronco et al. 2016). Wang et al. (2016) examined glyphosate degradation in a water-sediment experiment and determined that glyphosate levels decrease rapidly in water while accumulating in sediments during the initial 10 th day. Approximately 50% was degraded on the 20 th day, and approximately onethird of glyphosate was detected in water, while most compounds were found in sediment. On the 40 th day, low glyphosate quantities were detected in water. Less than 20% of the initial levels remained in the sediment. A continuous fall was observed until the 80 th day. Wang et al. (2016) identified sediments as critical in glyphosate degradation.
Imazapyr was shown to contaminate both water and sediments. Patten (2003) examined dissipation of imazapyr in the environment. After direct herbicide application to surface water, the maximal level of imazapyr found in water and sediment was 3.4 mg/ L and 5.4 mg/kg, followed by an exponential decay, which approached the zero asymptote at 40 hr for water and 400 hr for sediment, with a half-life of <0.5 days and 1.6 days, respectively. Imazapyr was used to control Spartina alterniflora in a submerged area, and residual imazapyr in the water was detected on the day of application at 268 µg/kg and 19.01 µg/kg on day 7. In the soil samples, imazapyr was found on the application day at higher concentrations than in water, but was not detected 21 days after application (Peng et al. 2022). In Brazil, imazapyr has been detected in streams of the Paraná basin, spring, and groundwater at concentrations ranging from 0.3 to 0.5 µg/L (Santos 2013). In another study, Mulder and Schmidt (2011) noted imazapyr present in surface water and groundwater samples at concentrations of <0.011 to 4.7 µg/L. Diquat dibromide was applied to the aquatic environment, and herbicide dissipation was determined. In a field study, diquat was applied at 2.5 mg/L and subsequently, measurement indicated 0.81 mg/L diquat after 24 hr (Grzenda, Nicholson, and Cox 1966). In reservoirs treated with 1 mg/L diquat, the concentration fell to 0.06 mg/L after 4 days (Yeo 1967). After applying 4.48 kg diquat cation/ha in the lake, the herbicide dissipated to 0.08 mg/L in 24 hr (Sewell 1970). Diquat in water is adsorbed rapidly by lake sediment, which is persistent for a long time (Simsiman and Chesters 1976). Diquat concentrations were measured in the Steilacoom Lake outlet following Reward® applications. Data demonstrated that diquat concentrations ranged from 0.5 µg/L to 10.3 µg/L following applications, persisting in the water at detectable levels for 12 days (Serdar 1998). Diquat dibromide was detected at 180 to 400 µg/L a few hr post-exposure to Reward® Aquatic Herbicide was determined in a small pond and a slough in the Sacramento-San Joaquin River Delta by Siemering, Hayworth, and Greenfield (2008) using fiberglass tanks with submerged plants, where 0.37 mg/L diquat was applied to tanks without water exchange and simulated flowing water conditions. Skogerboe, Getsinger, and Glomski (2006) found that diquat half-life of 27 hr for static conditions was 2.5 to 4.5 hr with water flow.
Rodriguez-Gil et al. (2017) determined the concentration of surfactants (MON0818 and POAE) and glyphosate at different water depths, confirming a differing surfactant distribution. Surfactant levels were highest at 0.15 m but decreased with rising depth. Furthermore, glyphosate concentrations appeared to be uniform across all sampled depths. Similar to glyphosate, surfactants were noted to interact with sediments. Wang et al. (2005) observed a rapid fall in MON0818 in water in the presence of sediment. Reduced toxicity was also observed if sediments were present, indicating that surfactant toxicity and concentration are diminished in shallow water due to interactions with sediments. The degradation of surfactants depends upon the type and abundance of microbes in the water and sediments. The half-life of POEAs was assessed in the water column (3.2 to 5.3 hr), suspended solids (fast-phase ranged from 0.71 to 1.3 hr and slow-phase from 18 to 44 hr) and sediment (5.8 to 71 days). Rodriguez-Gil et al. (2016) suggested that aquatic organisms are exposed to POEAs in the aqueous phase for a few hr following an over-water application (Rodriguez-Gil et al. 2016). Hence, knowledge of the environmental behavior of herbicides and surfactants is essential to know and helps provide context for toxicological studies with these compounds.
Furthermore, studying a model organism that is ecologically relevant across several environments is essential. Fish from the genus Astyanax are widely distributed across the Americas, from southern Argentina to North American aquatic habitats. Astyanax species are ecologically important because of their role as secondary consumers. This genus serves as a large biomass reservoir that maintains carnivorous species and thus influences population dynamics. Due to their small size, adaptability to aquarium life, and easy breeding procedures, Astyanax species have recently been utilized in lab studies, especially the blind cavefish Astyanax mexicanus (Stahl et al. 2019) and the yellowtail tetra Astyanax altiparanae (Yasui et al. 2022).
Using different fish models is essential in toxicological studies to support critical levels of the contaminants and their legal restrictions. Cichlasoma paranaense a teleostei found in Brazilian rivers was employed as a model to determine agrochemical-induced genotoxicity (Francisco et al. 2023). Danio rerio has been used as a model fish for herbicide toxicity studies (Costa et al. 2022). However, this species does not represent native species from different regions and ecosystems. A. altiparanae is distributed in the Upper Paraná River Basin, located in central-southern Brazil, with an area of 900,480 km 2 , and drains rivers from 6 Brazilian states and a small portion of Paraguay. This basin includes the Paraná River, considered the second largest river in South America and the most socioeconomically important basin in South America due to its significant contribution to economic sectors, including agriculture, livestock, energy, and urban and industrial water supply (Abou Rafee et al. 2019). The Upper Paraná River Basin hosts large numbers of fish species, and A. altiparanae might serve as an emergent model species due to being readily found in all rivers and tributaries of this basin, with easy domestication, precocity, culture, and reproduction. A. altiparanae was previously used in different studies, with 106 results found in PubMed (March 2023), including toxicological studies (Fernandes et al. 2019). Another important finding was the generation of gynogenetic individuals (Do Nascimento et al. 2020), which produce fully inbred lineages, and these homozygous animals are used as models (Krause and von Brand 2016). A review of studies performed with A. altiparanae aiming to include it as an emerging fish model was recently published (Yasui et al. 2022). These observations added to the high impact of the Upper Paraná River Basin agricultural sector and emphasize the importance of A. altiparanae as a fish model for pesticide toxicity studies.
The yellowtail tetra A. altiparanae is a neotropical characiform fish used as a sentinel fish to understand the toxicity underlying exposure to three herbicide formulations (Roundup Transorb®, Arsenal® NA, and Reglone®) to adults and embryos and their inhibitory effects on sperm motility. The results are essential for establishing toxicological herbicide concentrations that affect reproduction, development, and adult survival in natural environments in the rivers of the Neotropical region.

Materials and methods
All experiments were carried out following the Ethical Committee of the National Center for Research and Conservation of Continental Aquatic Biodiversity -CEPTA/ICMBio (CEUA 02031.000057/2018-22). The LC 50 experiments for adults were performed under OECD 203 and ABNT NBR15088, and for embryos, OECD 236 Guidelines for the Testing of Chemicals. However, some modifications were performed in the adult LC 50 experiments, including a system without water renewal, considering possible partial herbicide degradation as part of the experiment. Static non-renewal tests were performed under EPA guidelines.
The nominal concentrations of the active ingredients from the herbicide formulations were evaluated by HPLC analysis performed on a Shimadzu equipment model LC-10AT VP equipped with a Shimadzu SCL-10A VP Controller, Shimadzu SIL-10AF Autosampler, and Shimadzu VP Series SPD-10A UV-Vis Detector. It used a Kromasil C18 column (#K08670349, Supelco) with particles of 10 μm, pore size 100 Å, 250 mm in length, and 4.6 mm in diameter. The sample volume applied was 20 μl, and the flow rate was 1 ml/min. The diquat concentration was measured as previously described (Hara et al. 2007) using the isocratic mobile phase constituted by a mixture (1:4) of methanol (#34860, Sigma-Aldrich) and 200 mM phosphoric acid (phosphoric acid solution 85% #W290017, Sigma-Aldrich) containing 100 mM diethylamine (#31730, Sigma-Aldrich) and 12 mM sodium 1-heptanesulfonate (#H2766, Sigma-Aldrich). The column eluent was monitored at a wavelength of 290 nm. The imazapyr concentration was determined as previously described (Liu, Pusino, and Gessa 1992), and the isocratic mobile phase consisted of a mixture (30:70) of acetonitrile (#1000302500, Supelco) and water acidified with formic acid (#F0507, Sigma-Aldrich) at pH 2. The column eluent was monitored at a wavelength of 250 nm. The glyphosate concentration was assessed according to the methodology described by Khrolenko and Wieczorek (2005). For the analysis, the samples were derivatized using 1 ml of sample with 0.5 ml phosphate buffer at pH 11 and 0.2 ml p-toluenesulfonyl chloride (#240877, Sigma-Aldrich) solution (10 mg/ml acetonitrile) and incubated at 50°C for 10 min. The isocratic solution used in the analysis consisted of a mixture of 60 mM KH 2 PO 4 buffer (#P8709, Sigma-Aldrich) at pH 2.3 and acetonitrile (85:15). The column eluent was monitored at a wavelength of 240 nm. The concentration of active ingredients in the herbicide formulations was measured using standard curves performed using imazapyr (PESTANAL®, analytical standard #37877, Supelco), diquat dibromide monohydrate (PESTANAL®, analytical standard # 45422, Supelco), and glyphosate (PESTANAL®, analytical standard #45521, Supelco) in the range of 0.05 to 1 µg/ml. The herbicide formulations were diluted in type 1 ultrapure water to 1 mg/L nominal active ingredients and analyzed by HPLC. The nominal concentrations of the active ingredients provided by manufacturers were used for dilutions, and the values were adjusted from HPLC analysis.

LC 50 of herbicides on adults
Animals were randomly collected at the National Center for Research and Conservation of Continental Aquatic Biodiversity (CEPTA), Pirassununga, SP, Brazil, and acclimated for 7 days in 60 L glass aquariums containing potable water dechlorinated, with constant aeration at 28°C. Following this acclimation period, fish were subjected to food deprivation for 24 hr. The fish were anesthetized using eugenol, weighed to adjust the herbicide concentrations to be added concerning fish total weight, and distributed into 40 L aquariums. The amount of water in each aquarium was adjusted to provide 1 L per gram fish (Clesceri, Greenberg, and Eaton 1998). Prior to determining the LC 50 for herbicides in adults, previous experiments were performed to assess the concentration necessary to kill all fish. Based upon these previous results, the concentrations used for Roundup Transorb® were 0 (control), 1.05, 1.995, 2.94, and 3.885 mg/L glyphosate ae. The concentrations of Arsenal® NA used were 0 (control), 1.1, 4.18, 4.62, and 5.06 mg/L imazapyr ae. For Reglone®, the concentrations were 0 (control), 5.65, 11.3, 16.95, 22.6, and 28.25 mg/L diquat ai. The herbicide concentrations were calculated using data on the amount of active ingredient reported by the manufacturer, as reported in the section Herbicide formulations, and the amount of water used in each aquarium. The values of active ingredient concentrations used in the experiments were adjusted from the data of HPLC analyses. The LC 50 experiments were performed using 96 hr herbicide exposure.
During the experiment, the pH and dissolved oxygen content were monitored daily using a multiparameter probe Model U-50 (Horiba, Japan). The pH and dissolved oxygen values were 7.3 and 6.3 mg/L, respectively. The temperature remained from 25°C to 28°C. Dead fish were removed from aquariums to maintain water quality and were frozen at −20°C until appropriate disposal. Residual water from the experiments was stored in open outdoor tanks for 40 days for herbicide degradation and disposed of in the soil. The period for herbicides degradation was based upon the half-life of glyphosate (Souza et al. 2017), imazapyr (Shaifuddin et al. 2017), and diquat (Negrisoli et al. 2003) in water tanks. LC 50 values were calculated using the Toxicity Relationship Analysis Program (TRAP), Version 1.3a (Erickson 2015), applying the best-fitting non-linear regression model (logistic 2 parameters or threshold sigmoid 2 parameters), and are presented as the mean ± standard error. All the experiments were conducted in triplicate, with each replicate consisting of an aquarium containing three fish; therefore, nine animals were used for each herbicide concentration. The results from the derivation of LC 50 values for adults are illustrated as the mean ± standard error.

LC50 of herbicides on the embryos
Embryos of A. altiparanae were obtained using an induced spawning and in vitro fertilization protocol established (Yasui et al. 2015). Freshly fertilized eggs were placed individually into 96-well plates containing different glyphosate, imazapyr, or diquat concentrations from Roundup Transorb®, Arsenal® NA, and Reglone®, respectively. All experiments were performed in triplicate, with each replicate consisting of 32 embryos from a given couple, with 96 embryos for each herbicide concentration.
The herbicide concentrations used in the experiments were based on previous experiments to determine the concentration necessary for total hatching inhibition. The concentrations used for Roundup Transorb® were 0 (control), 10.50, 13.125, 15.75, 18.375, 21.0, or 23.625 mg/L glyphosate ae. The concentrations for Arsenal® NA were 0 (control), 5. 50, 7.15, 9.35, 11.55, or 13.75 mg/L imazapyr ae. For Reglone®, the concentrations were 0 (control), 565, 961, 1503, or 1695 mg/L diquat ai. Early development was observed for 17 hours at 27°C. This period includes all developmental phases from gamete activation until hatching, following several critical stages of development: the 2-cell, blastula, gastrula, and somite stages. Dead embryos were removed during all experiments. The % normal and abnormal hatched embryos were recorded, and LC 50 calculated using the same procedures described for adults. The results from the derivation of LC 50 values for initial embryonic development are presented as mean ± standard error.

Effects of herbicides on sperm motility
Sperm was obtained using hormonal induction by carp hypophysis extract injection in five adult males of A. altiparanae not exposed to the herbicides. The procedure to extract sperm and sampling procedure was performed as previously described (Yasui et al. 2015). The sperm were collected and transferred to 400 µl modified Ringer's solution (128.3 mM NaCl, 23.6 mM KCl, and 3.6 mM CaCl 2 , and 2.1 mM MgCl 2 ). Samples were maintained at 25°C for later sperm analysis upon herbicide exposure.
Motility was analyzed using ImageJ software (Rasband 1997(Rasband -2018 with modifications (Gonçalves et al. 2018) in conjunction with computer-assisted sperm analysis (CASA) (Wilson-Leedy and Ingermann 2007) following species standardization (Neumann et al. 2013). Relevant parameters, including motility (%), VCL (curvilinear velocity), and VLR (straight-line velocity), were evaluated. During the determination of sperm motility, a small aliquot containing 0.5 µL of sperm was pipetted into a Makler counting chamber (Self-Medical Instruments, Haifa, Israel), where both the chamber and the cover slide had previously been coated with a solution of 0.1% bovine serum albumin. This treatment was performed to prevent sperm attachment to chamber surfaces. Sperms were activated by diluting 20-fold with distilled water (control) or herbicide solutions. The cover slide was placed over the relevant solutions, and sperm motility was visualized using a trinocular microscope (Nikon Ni, Tokyo, Japan) at 200× magnification. Video sequences of sperm motility were captured using a CCD camera (Nikon DS-Fi, Nikon Tokyo, Japan) connected to the microscope. Video sequences were recorded using Nis-Ar Elements software (Nikon, Tokyo, Japan). The duration of sperm motility was measured using video, with the initial time point corresponding to sperm activation and the final time point corresponding to sperm motility lower than 5% of initial motility.
The viability of each sperm sample was assessed using flow cytometry. Each sperm sample was activated with the herbicide solutions and water as the control group and immediately stained using a live/ dead sperm viability kit (Molecular Probes, Thermo Fisher Scientific Inc., USA) based upon SYBR-14 and propidium iodide -PI (516 nm and 617 nm, respectively). The stained samples were analyzed using a flow cytometer (Accuri C6, BD Biosciences, San Jose, USA) and a relevant filter set. Gates containing live cells (516 nm) and dead cells (617 nm) were expressed as % to determine sperm viability.
All sperm analyses were performed in triplicate from each sample of the adults. The results from the sperm motility are presented as the mean ± standard error. Data were checked for normality using the Lilliefors test (5%) and subjected to ANOVA followed by Tukey's multiple range tests to determine significant differences. Statistical analysis was performed using STATISTICA software (v. 10, StatSoft, Tulsa). A significance threshold of 0.05 was used in all experiments.

Results
The concentrations of active ingredients from HPLC analysis were similar to the nominal concentrations provided by the manufacturers. In the HPLC analysis, using the dilution of 1 mg/L nominal active ingredients from herbicide formulations, the following concentrations were obtained: 1.05 mg/L glyphosate ae, 1.1 mg/L imazapyr ae, and 1.13 mg/L diquat ai. Therefore, the nominal concentrations of active ingredients were used to perform the experiments, and the concentrations were adjusted following concentrations from HPLC analysis.

LC 50 of herbicides on adults
The observed LC 50 values for adult individuals of A. altiparanae exposed to glyphosate ae and imazapyr ae were obtained from lethality rates (Figure 1). Glyphosate ae at 3.885 mg/L was lethal for all individuals exposed. In comparison, concentrations of 1.05 mg/L and 1.995 mg/L did not result in death (Figure 1a). These results suggest that concentrations from 2.94 mg/L to 3.885 mg/L glyphosate ae may be lethal for A. altiparanae. Similarly, a 5.06 mg/L imazapyr ae concentration was lethal to all exposed fish. At the same time, no deaths were observed in fish exposed to 1.1 mg/L or 4.18 mg/L (Figure 1b). These results indicate a narrow lethality window from 4.62 to 5.06 mg/L imazapyr ae. Despite high concentrations from 5.65 to 28.25 mg/L, no lethality was detected in the LC 50 experiments with diquat. Based upon LC 50 data, Roundup Transorb® seems to be the most toxic to adults of A. altiparanae. The calculated LC 50 value for this herbicide was 3.14 mg/L glyphosate ae after 96 hr exposure. Data for Arsenal® NA illustrate an observed LC 50 value of 4.59 mg/L imazapyr ae. Thus, the LC 50 value for imazapyr ae (95% confidence interval 4.44 to 4.73 mg/L) was slightly higher than the LC 50 value for glyphosate ae (95% confidence interval 2.86 to 3.44 mg/L). In the experiment in which adults were exposed to Reglone®, the value of LC 50 for diquat ai could not be evaluated in the used concentration range after 96 hr treament, indicating that the LC 50 was higher at the greater concentration (28.25 mg/L diquat ai) used in the investigation. This herbicide appears less toxic than the other herbicides investigated and requires concentrations higher than 28.25 mg/L for any lethal effects to occur in adults of A. altiparanae.

LC 50 of herbicides on early embryonic development
Herbicide toxicity was evaluated from gamete activation until the hatching stage. The % survival at different stages of embryonic development, including hatching, for the different concentrations of the herbicides Roundup Transorb® (glyphosate ae), Arsenal® NA (imazapyr ae), and Reglone® (diquat ai) are described in Tables 1, 2 , and 3, respectively. The general aspects of the hatching rates of larvae exposed to different concentrations of herbicides are illustrated in Figure 2. Comparisons between Figure 2 and Tables 1, 2 , and 3 indicate that the last two points of greater herbicide concentrations observed the highest hatching inhibition (Tables 1, 2 , and 3), the larvae are abnormal, and surviving larvae may be compromised. The practical survival of the larvae may be below the LC 50 calculated from the hatching % (Figure 2). The hatching rate of normal larvae at 10.5 mg/L glyphosate ae was slightly affected (88.54%) similar to controls (91.66%); however, an increase in % of abnormal larvae was detected at this concentration (Table 1). The total lethality number for embryos was verified at 23.625 mg/L glyphosate ae. A gradual rise in Figure 1. Acute toxicity test of herbicide formulations using adult A. altiparanae. Average of three repetitions of lethality rates for adult individuals exposed to different concentrations of herbicides (n = 9). a-Roundup Transorb® (glyphosate ae) and b-Arsenal® NA (imazapyr ae). For Reglone® (diquat ai), there was no mortality at any tested concentration. The mortality rate illustrated in the curve for the higher concentration of glyphosate (5.25 mg/L) and imazapyr (6.05 mg/L) was obtained from preliminary tests and included in the sigmoidal fitting. The circular yellow point and the dashed lines indicate the 96 hr LC 50 . lethality was noted from 13.125 mg/L to 21 mg/L glyphosate ae (Figure 2a). The hatching rate for normal larvae was 64.58% at 5.5 mg/L imazapyr ae, which was lower than controls (91.66%).
In contrast, abnormal larval hatching was approximately 27-fold higher than that of control (Table 2). A gradual increase in lethality was observed for imazapyr ae exposure from 7.15 mg/ L to 11.55 mg/L (Figure 2b), and total mortality number was found at 13.75 mg/L. Reglone® demonstrated the lowest embryonic toxicity compared to Roundup Transorb® and Arsenal®, with 100% embryo mortality prior to hatching at 1695 mg/L diquat ai (Figure 2c). The lower used concentration of 565 mg/L diquat ai resulted in a hatching rate of 76.04%, and all larvae were abnormal (Table 3). This result indicates the low toxicity of Reglone® in the early embryo development of A. altiparanae using hatching as a parameter to evaluate. However, the lower diquat ai concentration used resulted in abnormal hatched larvae.
Based upon the obtained LC 50 data, Reglone® may be considered the least toxic herbicide for embryos, in which the LC 50 value was approximately 65 and 116-fold higher than the LC 50 for Roundup Transorb® and Arsenal® NA, respectively. The LC 50 values obtained were 1084 mg/L for diquat ai (95% confidence interval 996 to 1172 mg/L), 9.327 mg/L for imazapyr ae (95% confidence interval 8.983 to 9.670 mg/L), and 16.515 mg/L for glyphosate ae (95% confidence interval 15.984 to 17.04623 mg/L). Roundup Transorb® seems to be the least toxic than Arsenal® NA during the initial stages of development. In the analysis of abnormal larval rates, it was verified that Roundup Transorb® at 15.75 mg/L glyphosate ae (Table 1) resulted in a higher % abnormal hatched larvae (37.5%) than normal hatched larvae (15.65%). All hatched larvae were abnormal at concentrations higher than 18.375 mg/L glyphosate ae. Arsenal® NA was considered the most 13.54 ± 6.51 12.5 ± 8.27 0 ± 0 9.37 ± 6.25 23.625 0 ± 0 0 ± 0 0 ± 0 0 ± 0 0 ± 0 0 ± 0 0 ± 0 toxic herbicide assessed for A. altiparanae embryos. High % abnormal larvae (46.9%) was found at 7.15 mg/L imazapyr ae (Table 2). All hatched larvae were abnormal at 11.55 mg/L imazapyr ae. For Reglone®, the lower concentration of 565 mg/L diquat ai resulted in high instances of abnormal larvae, an average of 76.04% (Table 3), and all hatched larvae were abnormal from this concentration. Higher rates of abnormal larvae were noted at concentrations near the LC 50 of the herbicides. Normal and abnormal larvae are illustrated in Figure 3.

Effects of herbicides on sperm parameters
The sperm parameters at different concentrations of the herbicides are presented in Table 4. In the experiments using Arsenal® NA, a gradual decrease in sperm motility was observed from imazapyr ae concentrations of 34.6 to 91.3 mg/L, resulting in reduced sperm motility rates from 42% to 13% compared with the control (61%). Total inhibition of motility was achieved at 137 mg/L imazapyr ae. Using Roundup Transorb®, sperm motility rates were decreased from 43% to 9% at concentrations of 100.8 to 168 mg/L glyphosate ae, and no sperm motility was detected at 252 mg/L glyphosate ae. For the Reglone® group, the fall in sperm motility rates ranged from 31% to 10% using 1510 to 9080 mg/L diquat ai, and total inhibition of sperm motility was achieved at 11,300 mg/L diquat ai. Reglone® was the herbicide that showed a lower capacity to inhibit sperm motility. The curvilinear velocity (VCL) and path velocity (VAP) parameters were not markedly altered at the concentrations at which reduced sperm motility was observed for all herbicides used. No marked differences in sperm viability were detected for all imazapyr ae and diquat ai used; sperm viability decreased only at the highest glyphosate concentrations (Table 4). 10.42 ± 4.77 9.40 ± 3.13 0 ± 0 9.40 ± 3.13 13.75 100 ± 0 0 ± 0 0 ± 0 0 ± 0 0 ± 0 0 ± 0 0 ± 0

Discussion
Roundup Transorb® was selected for this study because this herbicide may be employed during rainy periods attributed to faster absorption by plants than Roundup® (Rodrigues and Almeida 2018), suggesting that Roundup Transorb® may be more conducive than Roundup® in controlling aquatic plants if allowed by law. LC 50 analysis verified that adult A. altiparanae are more sensitive to herbicides than early embryonic stage to establish toxicity. These observations may be attributed to differences in intervals of herbicide exposure for LC 50 evaluation, 17 hr for embryos and 96 hr for adults. The LC 50 values observed for embryos were 9.327 mg/L imazapyr ae and 16.515 mg/L glyphosate ae, while for adults, the LC 50 values were 3.14 mg/L glyphosate ae and 4.59 mg/L imazapyr ae. Information regarding the sensitivities of different stages of life to herbicides is essential for determining safe levels of these compounds in aquatic ecosystems. Thus, this information enables the reliable maintenance of a given species in its native environment. Folmar, Sanders, and Julin (1979) reported that Roundup LC 50 values for fish vary according to the stage of development and that larvae and juveniles are more sensitive to herbicides than adults. The LC 50 for A. altiparanae embryos during the initial embryonic development, from zygote to hatching, was assessed with an interval of approximately 17 hr. Rapid embryonic development makes it impossible to examine the 96 hr LC 50 for a specific stage. The 96 hr LC 50 for Roundup in the eyed egg stage and large fish Oncorhynchus mykiss were 16 and 8.3 mg/L glyphosate, respectively (Folmar, Sanders, and Julin 1979). These results are in agreement with our findings, where adult A. altiparanae were more sensitive to herbicides than to initial embryonic development. However, further developmental stages, such as sac fry (LC 50 3.4 mg/L), swim-up fry (LC 50 2.4 mg/L), and fingerling at 1 g (LC 50 1.3 mg/L), were more susceptible to Roundup formulation than larger O. mykiss (Folmar, Sanders, and Julin 1979). Similar results were observed for Ictalurus punctatus, in which the sac fry and swim-up fry exhibited 96 hr LC 50 values of 4.3 and 3.3 mg/L glyphosate, respectively, while large fish displayed an LC 50 of 13 mg/L glyphosate (Folmar, Sanders, and Julin 1979). Data suggested that the eyed egg stage is the least sensitive to glyphosate. In this study, the 96 hr LC 50 values observed in adult A. altiparanae for Roundup Transorb® (3.14 mg/L glyphosate ae) showed an intermediate LC 50 value related to Pimephales promelas (2.3 mg/L) and Lepomis macrochirus (5 mg/ L) using the Roundup formulation (Folmar, Sanders, and Julin 1979). However, the composition difference between Roundup used by Folmar, Sanders, and Julin (1979) and Roundup Transorb Table 4. Sperm parameters of A. altiparanae after activation with the herbicides imazapyr (Ima), glyphosate (Gly), and Diquat (Diq). Different superscript letters within a column denote significant differences by Tukey's multiple range tests (P = 0.05). MOT: motility; VCL, curvilinear velocity; VLR, straight-line velocity. Viability: Percentage of live cells. used in this study is found in the concentration of glyphosate salt and acid equivalent, as well as the other ingredients (surfactants) not specified by manufacturers. These differences in the herbicide formulations might account for differing LC50s for each product. Several subsequent investigators confirmed that early life stages are more sensitive to herbicides than adult fish. Paul et al. (1994) determined the LC 50 for the Diquat HA formulation using three fish species post-hatching days. These investigators verified that the earlier stages were more sensitive to the herbicide than noted in previous studies. The most sensitive stage for Stizostedion vitreum was 8-10 days old (96 hr LC 50 0.75 mg/L diquat), 6-8 days old for Micropterus dolomieu (96 hr LC 50 3.9 mg/L diquat), and 9-13 days old for Micropterus salmoides (96 hr LC 50 4.9 mg/L diquat). Our study found a significantly higher value of LC 50 for Reglone® (1,084 mg/L diquat ai) during the initial embryonic development of A. altiparanae compared to the early life stages of S. vitreum, M. dolomieu, and M. salmoides using the Diquat HA formulation. However, hatching was employed as a parameter to evaluate the LC 50 experiments during the embryonic development of A. altiparanae, and the lower concentration (565 mg/L diquat ai) resulted in all abnormally hatched larvae, which might compromise survival of the larvae. In addition, the exposure interval was approximately 17 hr, from zygote to hatching than the 96 hr usually used. The LC 50 for diquat is highly variable between fish species, such as 96 hr LC 50 values of 2.1 mg/L to Stizostedion vitruem (Gilderhus 1967), 35 mg/L to Lepomis macrochirus (Gilderhus 1967), and 289 mg/L to Gambusia affinis (Leung, Naqvi, and Leblanc 1983). The acute toxicity of imazapyr was determined for different juvenile fish, such as Hyphessobrycon eques, Brachydanio rerio, Phallocerus caudimaculatus, and Piaractus mesopotamicus, with 96 hr LC 50 values ranging from 3.9 to 8.19 mg/L imazapyr ae (Cruz et al. 2015). The LC 50 observed for adult A. altiparanae was 4.59 mg/L imazapyr ae, similar to the LC 50 found in juveniles of B. rerio (4.35 mg/L imazapyr) and P. caudimaculatus (5.3 mg/L) (Cruz et al. 2015).
The comparison of acute and chronic sensitivity to chemicals in fish and amphibians demonstrated that toxicity data are highly correlated and that fish are generally more sensitive than amphibians (Weltje et al. 2013 (Folmar, Sanders, and Julin 1979) (using Roundup®) and S. vitreum (Paul et al. 1994) (using Diquat HA). In comparison, adults and in the initial embryonic development of A. altiparanae were more sensitive to Arsenal® NA (4.59 and 9.327 mg/L, respectively) than NF48 X. laevis (32.8 mg/L) (Babalola and van Wyk 2018), as well as to juveniles H. eques, B. rerio, P. caudimaculatus, and P. mesopotamicus (Cruz et al. 2015). However, it was verified that Arsenal® NA at environmental concentrations (0.5 to 3.5 mg/L imazapyr ae), following a Xenopus metamorphosis assay, resulted in a significantly delayed tadpole development, reduced hind-limb length, and increased whole-body mass at 3.5 mg/L imazapyr ae. Histology of the thyroid displayed elevated epithelium height and reduction of colloidal area, even at lower concentrations. Therefore, Arsenal® NA is thyroid-active in X. laevis at environmentally relevant concentrations (Babalola and van Wyk 2021).
The enhanced toxicant sensitivity during the early stages of development is related to several factors, including (1) low levels of accumulated fat for lipophilic substance storage, (2) greater uptake of toxins from the environment, (3) presence of underdeveloped organs that cannot properly engage in detoxification and elimination of herbicidal compounds, and (4) immature immune systems and underdeveloped homeostatic systems (Mohammed 2013). The initial embryonic development stage was less sensitive to the surveyed herbicides than adult A. altiparanae, corroborating with previously published studies for the eyed egg stage of O. mykiss and I. punctatus (Folmar, Sanders, and Julin 1979). Fish larval stages are the most sensitive to herbicides. Herein, the acute toxicities of the herbicides during the larval stages of A. altiparanae were not evaluated. Future analyses are essential to confirm the most sensitive stage to herbicides in A. altiparanae.
Determining LC 50 values for herbicides is essential for establishing the maximal permitted levels (MPLs) for these chemicals in aquatic environments. In drinking water, the MPL for pesticides, including organic herbicides, is 0.1 µg/L in Europe (Council of the European Union 1998). However, levels of glyphosate were approximately 14-fold higher than the European MPL in groundwater and approximately 5-fold higher in urine of farmers from areas of intensive agriculture (Rendonvon Osten and Dzul-Caamal 2017). Other countries have adopted different MPLs for pesticides in water. In Brazil, the established MPL with higher glyphosate and surfactant concentrations in water than the European MPL, with permitted levels in drinking water at 500 µg/L (Ordinance MS 518/ 2004of the ministry of health 2004). However, the National Environment Council (CONAMA) established a glyphosate concentration from 65 to 280 µg/L in water destined for human consumption, conservation unity, irrigation, aquaculture, and bodies of water (RESOLUÇÃO CONAMA Nº 357, DE 17 DE MARÇO DE 2005).
The herbicide manufacturers do not provide the concentration of surfactants in the herbicide formulations or the specific surfactant used. Information regarding the compounds utilized in these formulations is essential to assess the toxicity of the herbicide formulations and the application of specific MPLs to different surfactants. In this study, the concentrations of herbicides employed in LC 50 experiments with adults, embryos, and sperm were higher than permitted by the European Union and Brazilian MPL.
A worst-case scenario is conceivable, in which greater herbicide dosages recommended by manufacturers (Reglone® 3.5 L/ha, Arsenal® NA 10 L/ha, and Roundup Transorb® 5.625 L/ha) might be used in a standard water body 15 cm deep (usually used as a conservative reference). Under these conditions, the herbicides might reach high concentrations in the water, such as approximately 1.8 mg/L glyphosate ae, 1.66 mg/L imazapyr ae, and 0.467 mg/L diquat ai. In this scenario, glyphosate ae, imazapyr ae, and diquat ai levels might be lower than values found for LC 50 in adults (3.14 mg/L glyphosate ae, 4.59 mg/L imazapyr ae, and >28.25 mg/L diquat ai), initial embryonic development (16.515 mg/L glyphosate ae, 9.327 mg/L imazapyr ae, and 1,084 mg/L diquat ai), and sperm motility inhibition (252 mg/L glyphosate ae, 137 mg/L imazapyr ae, and 11,300 mg/L diquat ai). At these hypothetical concentrations of herbicides, no observed lethality in adults of A. altiparanae was detected. However, the hypothetical imazapyr level (1.66 mg/L) would be near the lower imazapyr ae concentration (5.5 mg/L) employed in the experiment, increasing the number of abnormally hatched larvae. The LC 50 for larvae A. altiparanae was not determined, but considering reports from other fish, the larvae of A. altiparanae may be more sensitive than adults (LC 50 3.14 mg/L glyphosate ae and 4.59 mg/L imazapyr ae). The hypothetical concentrations of glyphosate ae (1.8 mg/L) and imazapyr ae (1.66 mg/L) might initiate lethality in the larvae of A. altiparanae. It is of interest that Babalola and van Wyk (2021) reported that a concentration similar to the hypothetical concentration of imazapyr from Arsenal® NA is thyroid active in X. laevis. The hypothetical concentrations of glyphosate ae and imazapyr ae may be lower than the values obtained in this investigation for total inhibition of sperm motility from A. altiparanae. However, the lower imazapyr ae (34.5 mg/L) used reduced sperm motility by approximately 30%, and the hypothetical concentration of imazapyr might affect sperm motility. Reglone® was the least toxic herbicide for A. altiparanae in this study. Considering this worst-case scenario, Reglone® may serve as an alternative herbicide for aquatic vegetation, in addition to the Sonar AQ NA allowed for this purpose in Brazil. However, the toxicity of the hypothetical diquat ai (0.467 mg/L) needs to be considered for all aquatic organisms. In addition to these herbicides not being allowed by Brazilian law for aquatic use, the hypothetical concentrations of these herbicides might make the active ingredients reach levels in water higher than allowed Brazilian MPL for glyphosate. Souza et al. (2020) used experimentally Roundup Transorb® (the same formulation used herein) to control aquatic plants, applying 7.0 L per ha or 3.36 kg of acid equivalent per ha in reservoirs with and without Eichhornia crassipes. Data demonstrated that the half-life of glyphosate in water from reservoirs with plants was 11 days and without plants 21 days. The glyphosate levels were >50 µg/L in the reservoirs with plants and >250 µg/L without plants for the 4 th and 16 th days after applications, respectively. In both cases, the levels of glyphosate were lower than LC 50 obtained for adult and embryo as well as inhibition of sperm motility of A. altiparanae. However, Souza et al. (2020) suggested a low environmental impact of glyphosate use in the control of Eichhornia crassipes in reservoirs and that glyphosate can be recommended for use in continuous-flow aquatic environments due to herbicide dissipation and degradation in the environment.
With sperm motility, the most toxic herbicide was Arsenal® NA with 137 mg/L imazapyr ae for total inhibition of sperm motility, corresponded to the concentration of approximately half of glyphosate ae (Roundup Transorb®) and 82-fold lower than diquat ai (Reglone®). Our results differ from previous data on A. lacustris (Gonçalves et al. 2018). In that study, concentrations as low as 50 µg/L glyphosate from formulation-derived glyphosate (below the Brazilian MPL) inhibited sperm motility. However, our results (Table 4) were similar to Nerozzi et al. (2020) data found in pig semen, in which total inhibition of sperm motility was achieved at 360 mg/L glyphosate ae (Roundup), with 5 mg/L glyphosate significantly reducing sperm motility after incubation period of 1-3 hr. Nerozzi et al. (2020) verified that Roundup is more toxic to sperm than its main component, glyphosate. Torres-Badia et al. (2021) suggested that the detrimental effects of Roundup on sperm function may be attributed to its surfactant. Human semen exposed to Roundup at 0.36 mg/L glyphosate ae reduced 11.6% of sperm motility after 1 hr incubation and 6.3% after 3 hr relative to control (Anifandis et al. 2018). Exposure of adult Danio rerio to pure glyphosate at concentrations from 5 to 10 mg/L lowered sperm functionality and duration of sperm motility resulting in a decrease in adult male fertility (Lopes et al. 2014).
The selected herbicides used in this study significantly reduced sperm motility at high concentrations compared with environmentally relevant concentrations. In light of these results, the levels obtained to interfere with sperm motility are greater than those allowed by MPL of 500 µg/L glyphosate in water, as established in current Brazilian regulations.
The concentration of herbicides reported in environmental water bodies is lower than the quantity necessary to produce acute toxicity in A. altiparanae, or inhibit sperm motility. The highest glyphosate levels found in surface water from areas of extensive agriculture in the Paraná basin ranged from 100 to 700 µg/L (Peruzzo, Porta, and Ronco 2008). These glyphosate levels are lower than those related to adult 96 hr LC 50 (3.14 mg/L glyphosate ae), early embryonic development 17 hr LC 50 (16.515 mg/L for glyphosate ae), as well as lower glyphosate concentration used in sperm motility inhibition experiment (100 mg/L glyphosate ae). The level of imazapyr found in the water was 3.4 mg/L (Patten 2003) and 268.37 ± 464.29 µg/ kg (Peng et al. 2022) after direct herbicide application to surface water, followed by an exponential decay in both studies. The data demonstrated that environmental imazapyr levels were lower compared to adults 96 hr LC 50 (4.59 mg/L imazapyr ae), early embryonic development 17 hr LC 50 (9.327 mg/L for imazapyr ae), and lower concentration used for inhibition of sperm motility (137 mg/L imazapyr ae). Imazapyr dissipation in environmental waters was shown to be rapid such that imazapyr levels in the water in 1 day fell dramatically. In Brazil, imazapyr was detected in the waters of the Paraná basin at concentrations from 0.3 to 0.5 µg/L (Santos 2013). The diquat concentrations reported in water after herbicide exposure were found to be 0.81 mg/L (Grzenda, Nicholson, and Cox 1966), 0.06 mg/L (Yeo 1967), 0.08 mg/L (Sewell 1970), 0.5 µg/L to 10.3 µg/L (Serdar 1998), and 0.18 to 0.4 mg/L (Siemering, Hayworth, and Greenfield 2008). Diquat levels found were lower compared to adult 96 hr LC 50 (>28 mg/L diquat ai), early embryonic development 17 hr LC 50 (1084 mg/L diquat ai), and lower level that inhibits sperm motility (1,510 mg/L diquat ai).

Conclusions
Adult A. altiparanae was more sensitive to the herbicide formulations than to the initial embryonic stage, based upon 96 hr LC 50 and 17 hr LC 50 , respectively. High levels of herbicides were necessary for total inhibition of sperm motility. Assays of acute toxicity to herbicides in the larval stages of A. altiparanae need to be performed to search for the life stage most sensitive to herbicides. The stage more sensitive is appropriate to evaluate herbicide toxicity to assess the environmental impact of herbicide formulations.
The concentration of glyphosate permitted by Brazilian law (500 µg/L) and the environmental concentrations reported for used herbicides in this investigation did not affect adults, initial embryonic development, or sperm of A. altiparanae. In addition to surfactants being degraded quickly and adsorbed in particles and sediments, residual surfactants in the environment need to be evaluated together with herbicides to establish the relationship between herbicides/surfactants, aiming to adjust to the herbicide concentration permitted by law. This study verified that different herbicide formulations might be more toxic for a specific stage of development. Arsenal® NA was found to be more toxic to initial embryonic development and sperm motility, while Roundup Transorb® was for adults of A. altiparanae. Reglone® showed low toxicity for all experiments related to Arsernal® NA and Roundup Transorb®, suggesting that different herbicide formulations might be indicated for applications in aquatic environments (if permitted by law) or terrestrial crops near watercourses, and yet, in a specific season, such as the reproductive period of reproduction in fish. The indication of different herbicide formulations might help reduce the environmental impact on aquatic organisms and promote species conservation in agricultural regions. Toxicity tests demonstrated that LC 50 for herbicide formulations varied for different fish species. Thus, using an animal model representative of a specific region is essential to evaluate the toxicity of pesticides. It is suggested that A. altiparanae be considered as an emerging candidate for a fish model and sentinel for toxicity tests for the Upper Paraná River Basin, which may be expanded to the Neotropical region. The data obtained here may be applied to both basic and applied sciences. Further studies are needed to assess the toxic properties of surfactants and pure herbicides in A. altiparanae to understand the individual actions of the herbicide formulations.