Environmental diagnosis of water quality in situ and ex situ in streams in Upper Paraná River Basin, Brazil

The aim of this study was to (i) evaluate water quality using limnological parameters in three streams belonging to the Ivinhema River basin, (ii) investigate the composition and structure of the landscape around the locations sampled and their possible contribution to changes in water quality, (iii) investigate genotoxicity in sh in situ as environmental bioindicators, and (iv) evaluate the cytotoxic and genotoxic potential of water through bioassays in animal and plant models. A total of seven sh species were sampled in situ as bioindicators of environmental toxicogenetic damage: Astyanax fasciatus, Astyanax lacustris, Cyphocharax modestus, Moenkhausia forestii, Piabina argentea, Psellogrammus kennedyi, and Hypostomus ancistroides. Four types of nuclear and one cytoplasmic alteration were observed, with nuclear invagination being the most frequent. Animal bioassays showed ve types of nuclear and one cytoplasmic alteration. In the plant bioassay, a reduction in germination and cell division was observed, indicating the cytotoxic and genotoxic effects of the water for Allium cepa cells. Thus, genotoxic alterations in sh and plant bioassays may reect the different types of land use and cover around streams and the conversion of forest fragments into agricultural areas, mainly due to the expansion of sugarcane crops. The evaluation of the toxicogenetic status of the three streams is essential to emphasize the need to create projects to recover the native vegetation cover, which would improve the quality of the habitat and ensure the permanence of native species.


Introduction
In aquatic ecosystems, anthropic activities cause changes in the structure of local vegetation, compromising the quality of water resources and the collapse of the aquatic biota (Carminatto et al. 2020). The reduction of riparian vegetation cover directly affects water quality, because, among several functions, riparian vegetation acts as a natural lter that dampens the transport of pollutants through the river and regulates limnological parameters (Ribeiro Junior et al. 2018). The changes in the landscape around rivers and streams lead to the formation of homogeneous and unhealthy vegetation, which directly affects and reduces local biodiversity that depend on good environmental quality for its maintenance (Roa-Fuenteset al. 2019;Zeni et al. 2019). Being that, habitat fragmentation is considered one of the main threats to biodiversity Gonçalves et al. 2019).
In Brazil, to increase agricultural productivity, a wide variety of pesticides, herbicides, and fertilizers are used in soil preparation and crop maintenance. However, these agrochemicals can be released gradually into both the water table and water bodies, where they are conducted through the course of rivers and transported through hydrological periods, causing severe damage to the health of aquatic biota, especially the sh . Thus, aquatic environments are the initial or nal destination of contaminants (Dusman et al. 2014).
In this sense, research related to the toxicological status of aquatic environments is increasingly relevant to understand the biological conditions of organisms that survive in these areas and the genetic damage associated with the presence of contaminants in the water (dos Santos et al. 2020; Riveros et al. 2021). Therefore, the use of environmental bioindicators organisms has been an important diagnostic and monitoring tool for environmental impacts in aquatic ecosystems (Bianchi et al. 2019). Because organisms respond to the effects of certain environmental impacts in the form of macrolesions on the DNA molecule resulting in genotoxic damage (Batista et al. 2016). The genotoxic effects are changes in the nucleus or in the cellular nuclear morphology resulting from potentially toxic contaminants in the aquatic environment, which is interfering in the cell division process, re ecting in damages in the genetic material of the aquatic organisms (Pereira da Silva et al. 2020).
Considering the ecological and environmental importance of the streams of the Ivinhema River basin and the lack of information on its environmental health and toxicogenetic status, an environmental monitoring study was carried out to evaluate the impact of agricultural activities in the area. Therefore, the aim of this study was to (i) evaluate water quality using limnological parameters in three streams belonging to the Ivinhema River basin, (ii) investigate the composition and structure of the landscape around the locations sampled and their possible contribution to changes in water quality, (iii) investigate genotoxicity in sh in situ as environmental bioindicators, and (iv) evaluate the cytotoxic and genotoxic potential of water through bioassays in animal and plant models.

Methodology Study area
Three sampling sites were selected at Vitória and Piravevê small basins, lower portion of Ivinhema River Basin, Upper Paraná basin (Fig. 1). The Ivinhema River basin is located on the right bank of the Paraná River in Mato Grosso do Sul, Brazil and is about 600 km long with an area of 45,000 km² (Súarez et al. 2011).
Rosário stream are a rst order stream, with approximately 1.5m width and 0.75m depth and water velocity near to 0.69 m.s − 1 . Riparian vegetation are scarce with predominance of grasslands and highly anthropized forest and sugarcane at stream border. Vitória Stream are a fourth order stream with mean width of 5.5m, depth varying from 0.9m and water velocity 0.84 m.s − 1 , riparian cover are also grasslands and pasture in sampled portion of the basin. Piravevê stream are more voluminous site ( fth order), with width varying from 7 to 15m and mean depth of 1.4m and water velocity of 0.3m.s − 1 , they present secondary vegetation and pasture predominance at riparian vegetation. Considering these sites, Rosário and Vitória streams are located in area that recently (last ve years) present higher increase of sugarcane cultivation, replacing livestock and Piravevê stream still present native (highly anthropized) vegetation and pasture. All portions are located to near 5km long to urban areas and did not receive urban sewage, however can be indirectly affected by improper disposal of domestic sewage from small farms surrounding streams.

Sampling
Water and sh were collected in November 2016 and March 2017 using rectangular sieve (0.8x1.2m) and dragging nets (1.5x5m) in non standardized sampling effort in a portion of approximately 100 long in all streams, near to highways that facilitated access to the river Data collection and analysis Limnological parameters: To characterize the limnological conditions of the studied areas, the following parameters were obtained in situ using a previously calibrated Horiba multi-parameter probe: dissolved oxygen (mg·L − 1 ), pH, electrical conductivity of water (µS·cm − 1 ), and temperature (°C).

Land use
The map of land use of the Piravevê, Vitória, and Rosário streams was made from high-resolution aerial images obtained from Google Earth Pro® software (years 2014 and 2015), with a resolution of 1 m (Digital Globe). For study limits, buffers with a radius of 1.5 km were generated around each sampling point and the categories of land use were classi ed as agriculture, fragmented vegetation, wet areas, water bodies, or anthropic occupation (IBGE, 2013). For image interpretation, an unsupervised classi cation (Clustering) was used, using the tools from the ArcGIS® program, Trial version 10.3 (ESRI 2015), calculating the areas and percentages of each category of land occupation based on buffer areas.
Collection and genotoxicity of sh in situ: For the collection of sh species in the streams, a rectangular metal frame sieve measuring 0.8 m ×1.2 m with a 2-mm mesh was used. After catching the sh were identi ed according to Graça and Pavanelli (2007). Then, larger-sized species above of standard length (30.00 mm), were selected, being (A. fasciatus, A. lacustris, Cyphocharax modestus, Moenkhausia forestii, Piabina argentea, Psellogrammus kennedyi, and Hypostomus ancistroides), to obtain blood samples. Shortly after, biometric data were measured: standard length (mm) and total weight (g). Then, the sh were immersed in cold water in situ to reduce their activity to obtain blood samples through a cut in the caudal peduncle. For each sh sample, two blood smear slides were made that were air dried for 15 min and xed in absolute alcohol for another 15 min (Schmid 1975;Jesus et al. 2016;De Souza et al. 2019). Then, the slides were hydrolyzed in hydrochloric acid (HCl; 1N) for 10 min at 60°C, washed in distilled water, stained with reactive Schiff overnight, and counter-stained with Fast green. After preparation, 2000 cells per slide were analyzed, resulting in a total of 4000 erythrocytes per individual, using a Nikon optical microscope (Eclipse, E200), with a magni cation of 1000× for micronuclei counting and identi cation of nuclear and cytoplasmic alterations. To calculate the genotoxicity index, all nuclear alterations identi ed in the study were grouped, the ratio between the total number of altered cells and the total number of cells observed calculated, and this ratio multiplied by 100. For the micronucleus test and nuclear alterations, we followed the protocol described by Schmid (1975) and Souza et al. (2019).
Genotoxicity bioassay with Astyanax lacustris: Water samples collected of each sites, was stored in polyethylene containers of 20 L, taken to the laboratory, and placed in independent glass aquariums (40 cm × 30 cm × 20 cm) (ABNT 2016). For the negative control (NC), groundwater was placed in other aquarium. We selected the sh species A. lacustris, as it is widely used as a model organism, for performing the toxicogenic effects tests Nascimento et al. 2020). Ten A. lacustris were placed in each aquarium, provided by a commercial sh farm (Douradense farm), with standard length (mean ± standard deviation: 68.75 ± 6.22 mm) and weight (mean ± standard deviation: 72.00 ± 32.03 g).
The sh were well fed throughout the exposure period, and the temperature (26°C), pH (7.15), and dissolved oxygen (7.5 mg/L) were kept within the appropriate limits for the animals. After a period of 72 h, blood samples were obtained from the sh by caudal puncture with heparinized syringes. For each specimen, two thin layer slides were prepared using a drop of blood. Subsequently, the same steps described in the genotoxicity item of sh in situ were performed.
Allium cepa test: Seeds of A. cepa were sown in plates with approximately 3 mL of water sampled, previously collected in sterile plastic containers, from the investigated streams. These seeds were left to germinate for a period of 96 h at a temperature of 23 ± 3°C, according to the procedure adapted from Leme and Marin-Morales (2009). After germination, the seeds were counted manually and their roots measured with digital calipers (Digmess). Later, the roots were xed in Carnoy's solution (3:1 Absolute ethanol:glacial acetic acid), subjected to hydrolysis with 1 mol·L − 1 HCl, washed with distilled water, and stained with Schiff's reagent. For each treatment, ten slides of the root meristem were made and a total of 500 cells per slide were analyzed under a Nikon microscope (Eclipse, E200) at a magni cation of 400×.
Positive control (PC) seeds were exposed to the herbicide tri uralin (0.84 ppm), and distilled water was used as a NC. Cytotoxicity, genotoxicity, and mutagenicity parameters, mitotic index (MI), chromosomal alteration index (CAI), mutagenicity index (MTI), and cell death rate (CDR), were calculated according to Leme and Marin-Morales (2009) and Francisco et al. (2018). Statistical analysis Before the analyses, the data were subjected to the Shapiro Wilk test to verify their normality. Analysis of variance (ANOVA) (α = 0.05) was performed to evaluate the genotoxicity indices in sh. To compare the frequency of nuclear and cytoplasmic alterations in the erythrocytes of sh species both in situ and in the bioassay analyzed in the different streams, the Kruskal Wallis nonparametric test was applied with Dunn's posteriori (α = 0.05). To evaluate the cytotoxic damage in a plant model, ANOVA (α = 0.05) was performed. The tests were conducted using the R platform (R Core Team 2019).

Limnological parameters
Among the physicochemical parameters, dissolved oxygen in all sampled streams was in accordance with the Brazilian legislation CONAMA (357/2005) for freshwater class 2 ( Table 1). The pH in the Vitória and Rosário streams was lower than that established according to the legislation (Table 1). For water temperature and electrical conductivity, there is no maximum limit established by Brazilian legislation.

Land Use
Regarding land use and cover in the three streams, we observed the predominance of anthropic activities. Piravevê had 56.76% of its area represented by the expansion of pasture, and Vitória and Rosário had a predominance of agricultural areas with 57% and 73.4%, respectively. In these agricultural areas, monoculture sugarcane was observed, and consequently attributed to the reduction of vegetation cover around the three streams (Fig. 2).  For genotoxic alterations in the sh species, we found four types of nuclear and one cytoplasmic alteration, the most frequent alteration being nuclear invagination for all species (Table 3). Different letters represent signi cant differences (p < 0.05).

Genotoxicity of sh in situ
Nuclear invagination in H. ancistroides and P. argentea did not differ from each other (p > 0.05) but differed from other species. Nuclear budding and lobulated nucleus in C. modestus and H. ancistroides did not differ but showed differences in relation to the other species. Regarding micronuclei, only C. modestus showed more genotoxic damage. There was no signi cant difference between species regarding cytoplasmic invagination (p > 0.05) ( Table 3).

Allium cepa bioassay
Regarding the bioassay using A. cepa, it was observed that the water samples from the three streams signi cantly inhibited seed germination compared to the NC (p < 0.05) ( Table 4). The germination percentage (%GR) of the water samples from the Rosário stream differed signi cantly from Piravevê and Vitória, presenting a higher number of germinated seeds. When the relative growth (RRG) of the root was analyzed, no signi cant difference between the streams and the NC was observed, only between the streams Piravevê, Vitória, and Rosário. There was a signi cant difference in MI between the NC and the streams, demonstrating that the water from these streams caused a decrease in cell division in the meristem cells of A. cepa roots. No difference was found between the CDR of the streams, demonstrating a low frequency of cell death (Table 4). In terms of CAI, NC showed a signi cant difference only in relation to Vitória. However, no signi cant difference was observed (p > 0.05) in MTI (Table 4).

Bioassay using A. lacustris
Regarding the bioassay using A. lacustris, we observed ve types of nuclear and one cytoplasmic alteration (Table 5). The most frequent nuclear alteration in the erythrocytes of A. lacustris was nuclear invagination, presenting a signi cant difference between the negative control and the Vitória and Rosário streams. Lobulated nuclei also showed a signi cant difference between the Vitória and Rosário streams (Table 5).
Regarding nuclear budding, binucleated cells, micronuclei, and cytoplasmic invagination, no signi cant differences were found between NC and the streams (p > 0.05) ( Table 5).
Regarding the genotoxicity index of the bioassay with A. lacustris, there was a difference between the NC and Vitória and Rosário streams (p < 0.05) (Fig. 4). Conversely, among the three streams, there were no signi cant differences (p < 0.05) (Fig. 4).

Discussion
The limnological parameters sampled in the three streams and dissolved oxygen was within the criteria established by CONAMA (357/2005). There are no values stipulated by the legislation regarding temperature. In the Vitória and Rosário streams, the pH presented values below those established by the legislation, which may be associated with the highest proportions of sugarcane monoculture around these streams (Fig. 2). According to Nhiwatiwa et al. (2017), the low pH in the waters of rivers and streams near sugarcane plantations can be attributed to the extensive application of fertilizers, in addition to the fertigation process. In addition, other chemicals are also used in agricultural activities, which can change the pH of water . Another aggravating factor in environmental imbalance is the reduction of riparian forests and vegetation surrounding the aquatic environment, reducing the area of protection of these environments against contaminants ). The low pH in these streams indicates environmental imbalance, because acidic water promotes metabolic alterations in sh representing a threat to species conservation.
This pH value can also be a response to the reduction of vegetation cover, which causes an imbalance in the physicochemical parameters of water, such as rising temperature and drop in pH. In aquatic environments, the latter can inhibit the growth of detritivorous bacteria and alter the development of zooplankton and phytoplankton, causing an imbalance (Birungi et al. 2007). Furthermore, low pH is associated with increased organic matter and decreased dissolved oxygen concentration (Bueno et al. 2005). Algae proliferation, due to nutrient enrichment from urban e uents, without proper treatment in aquatic environments, can result in decomposition and also reduce water pH (Marotta et al. 2008).
The areas near the streams showed a predominance of agricultural areas and a reduction in vegetation cover. However, Piravevê had a lower proportion of agriculture and a higher proportion of pasture areas in relation to the other streams. The replacement of vegetation cover around aquatic environments by agricultural areas is a major environmental problem, mainly because water bodies are unprotected from pollutants (Gonçalves et al. 2019). Dense vegetation cover has a barrier function, dampening the transport of sediments and pollutants into the stream during periods of increased rainfall, thus avoiding silting and imbalance in the structure and functioning of the ecosystem (Castro and Souza 2013;dos Santos et al. 2020). In addition, such coverage provides a reduction in eutrophication, promotes the conservation of fauna and ora, and facilitates the in ltration of rainwater into the soil (Tundisi and Tundisi 2010). In this sense, the fragility of the vegetation around the streams leaves the environment more exposed and vulnerable to contaminants, compromising the health of aquatic biota, especially the sh. Several studies have reported the serious consequences of reducing plant cover for the health of aquatic organisms, which may cause a reduction in species due to the degradation of environmental quality Ribeiro Junior et al. 2018;Gonçalves et al. 2019).
When the genotoxic index was analyzed, C. modestus and H. ancistroides suffered greater damage to their genetic material in relation to the other species, which may be related to the feeding habits (detritivory) of these two species. Viana et al. (2018) reported that H. ancistroides presented a greater number of genotoxic alterations in streams near urban and rural areas, compared to other species that did not have a detritivorous diet. This is because many contaminants are suspended for a period in the water, but later settle with the sediments of rivers and streams. Thus, the sh foraging at the bottom of aquatic environments end up biomagnifying higher loads of contaminants (Cardone 2006;Labarrère et al. 2012). Another reason for the greater genetic damage in H. ancistroides is that because of their benthopelagic behavior, these sh accumulate more pollutants, which consequently re ect as genetic damage (Labarrère et al. 2012). In this context, the literature suggests that the evaluation of environmental genotoxicity in different species of sh in situ generates more consistent results due to their different sensitivities to the aquatic environment (Leung et al. 2014;Silva et al. 2016). Zhang et al. (2019) highlighted that evaluating native sh species in situ in aquatic environments not only re ects toxic impacts, but also reveal their ability to adapt to unfavorable conditions in the environment. Several anthropic activities alter land use near aquatic environments, reducing the connection between riparian vegetation cover and the aquatic environment (Gonçalves et al. 2019).
The analysis of cytotoxicity, genotoxicity, and mutagenicity of the water samples from the streams resulting from the bioassay with A. cepa indicated that all samples signi cantly affected seed germination. However, considering relative root growth, no signi cant difference was observed between the NC and the samples from the streams, as presented in Table 4. According to the MI results, there was a reduction in cell division in all sites in relation to the NC. We also veri ed a signi cant increase in CAI in the Vitória stream compared to the NC, indicating that the waters of these three streams are cytotoxic and genotoxic to the cells of A. cepa. However, none of the water samples showed mutagenic effects on the cells.
In the present study, signi cant changes were observed in seed germination and cell division of A. cepa exposed to water samples from the evaluated streams in relation to the NC, as shown in Table 4. The inhibition of germination, as well as the decrease in MI values and increase in CAI, indicated the presence of substances in the water that promoted cytotoxicity and genotoxicity in A. cepa cells. The MI serves as an indicator for the analysis of the cytotoxic potential of various substances that may inhibit or increase cell proliferation (Leme and Marin-Morales 2009;Francisco et al. 2018;). On the other hand, the chromosomal alterations measured by CAI indicated the presence of clastogenic and aneugenic agents in the environment, which allowed de ning the genotoxic effect (Athanásio et al. 2014).
Previous studies also reported a decrease in MI and an increase in CAI were also observed in A. cepa cells after exposure to water samples when compared to the NC in different periods and streams (Athanásio et al. 2014;Da Silva et al. 2018). These studies associated the alterations of these parameters with the agricultural products used around the collection sites. Based on this, the results of this study may have been in uenced by the effect of the use of agricultural products around the sampling sites (Fig. 2).
The period of collection of water samples may also have in uenced the cytotoxic and genotoxic effects veri ed, because during rainy periods there is an increase in the ow of water that transports the compounds derived from leaching and percolation from diffuse sources, which can have negative impacts on the environment. Similar results to those observed in our study were also veri ed by Salles et al. (2016), who evaluated the dry and rainy seasons and obtained lower MI values and higher CAI values for A. cepa cells exposed to surface water samples from the rainy season in relation to the dry season. Considering the MI and CAI, the cells of A. cepa tested with the water sample of Vitória stream were the ones that most demonstrated inhibition of cell division and induction of chromosomal alterations in relation to the other treatments. These effects may be related to the land use around this site, since, as observed in Fig. 2, the Vitória stream, in addition to presenting conversion to agriculture land, also presented higher anthropic occupations in its surroundings compared to the others. However, further studies in these streams are necessary to identify the nature of pollutants present in the water samples that cause cytotoxicity and genotoxicity.
In the bioassay with A. lacustris, nuclear invagination was the alteration that presented the highest frequency, especially in the Vitória and Rosário streams. In general, the impacts of genotoxic events on erythrocytes before cell division can generate some nuclear changes, in addition to triggering micronuclei or even cell death (Garcia et al. 2017). Nuclear invagination is closely related to cell repair, which, when detecting a defective chromosomal area, initiates a repair process, eliminating the affected area with imperfections occurring in the nuclear membrane (Shimizu et al. 1998). In this sense, literature reports that nuclear alterations are responses to exposure to genotoxic agents (Ergene et al. 2007;Riveros et al. 2021).
As observed, the greatest genotoxic damage in sh exposed to water from the Vitória and Rosário streams may be associated with some type of contaminant in the water body, derived from the cultivation of sugarcane, which causes genetic damage to erythrocytes. Furthermore, we can predict that the genotoxic damage observed in sh may be linked to low vegetation cover due to the expansion of agricultural areas near these streams. The results of the present study corroborate those of Viana et al. (2017) and Dourado et al. (2016), who reported that sh exposed to polluted waters in this region had chromosomal alterations and DNA damage. The genotoxicity indices of the bioassay with A. lacustris observed in the Vitória and Rosário streams were signi cantly different from the negative control, con rming a higher number of genotoxic damage from the waters of these two streams. In the evaluation of sh species in situ, nuclear invagination was the most evident alteration in all sh species; this being the rst response of the organism when detecting chromosomal imperfections, initiating a process of repair and/or elimination of the affected area, resulting in imperfections in the membrane before nishing the process and characterized by nuclear anomalies (Thomé et al. 2016;Oliveira et al. 2018). Thus, different land uses associated with agricultural activities such as pastures and sugarcane cultivation may variably affect the environmental integrity of water bodies, and later the biota of streams. In this sense, even though the Piravevê stream had higher proportions associated with the pasture area, the results showed that the impacts were lower when compared to the Vitória and Rosário streams where the occupation in their surroundings is associated with agricultural areas, especially the monoculture of sugarcane.
In summary, the three streams presented low native vegetation cover due to the expansion of agricultural areas directly re ecting the loss of environmental quality. The Vitória and Rosário streams presented lower pH, which may be related to the higher proportion of agricultural areas around these sites.
Regarding in situ sh sampling, more genotoxic damage was identi ed in sh species with detritivorous feeding habits. Therefore, the combination of different tests and tools applied in this study were fundamental to assess the toxicogenetic status of these sites and emphasize the need for the recovery of riparian vegetation cover to improve the quality of the habitat and exposed organisms. In the bioassay with A. lacustris, a greater number of genotoxic alterations were observed in sh erythrocytes in the Vitória and Rosário streams. In the bioassay with A. cepa, the water samples from all streams caused a decrease in cell division and induction of chromosomal alterations in the cells; however, these changes were more pronounced in the Vitória stream, where there were higher proportions of agriculture land and little local vegetation coverage. It is worth mentioning, therefore, the need to create a reforestation project of native species throughout the study area to maintain the balance of this aquatic ecosystem.

Declarations Data availability
The data is not stored in a database, but we accept to share. Figure 1 Location of the sampling sites in streams, Piravevê, Vitória and Rosário, belonging to the Ivinhema River basin, Upper Paraná River, Mato Grosso do Sul. Land use and cover around three streams, Piravevê, Vitória and Rosário, belonging to the Ivinhema River basin, in Mato Grosso do Sul.

Figure 3
Genotoxicity index relation (mean and con dence interval), of sh species in situ in the Piravevê, Vitória and Rosário streams, belonging to the Ivinhema River basin, in Mato Grosso do Sul. Genotoxicity index (mean and con dence interval), of the bioassay with A. lacustris exposed to water samples from the streams Piravevê, Vitória and Rosário, belonging to the Ivinhema River basin, in Mato