Effect of land-use pattern on the physicochemical and genotoxic properties of water in a low-order stream in Central Brazil

Low-order streams located near urban areas usually receive domestic, industrial, and agricultural wastewaters that negatively affect river water quality. Additionally, water pollution is associated with land-use variations around the river, which is characterized by unplanned urbanization, intense agricultural activities, and deforestation. This work correlated land-use patterns with physicochemical quality and genotoxic potential of water at four points (P1 to P4) along the Extrema River, located in an industrial and agricultural area of Central Brazil. Physicochemical analyses indicated that the water collected from the Extrema River is inappropriate for human consumption. Using the Allium cepa model, no evidence of cytotoxicity was observed at any point; in contrast, the genotoxic potential of these water samples was observed. The correlation of these results with land use showed that the water collected at P3 was the most contaminated; this is probably due to the inflow of wastewater from municipal, industrial, and agriculture activities. Different results were observed for P2 and P4, where land-use analysis attributed the water quality to forest burns. We concluded that differential use of the land changed the characteristics of the associated river water, and A. cepa parameters were more related to land-use characteristics than to physicochemical parameters. This study highlighted the importance of associating land use with the cyto genotoxic potential of water.


INTRODUCTION
Surface waters are the most important natural and economic resource for agricultural, industrial, and anthropogenic activities. Water quality plays a pivotal role not only in habitat protection but also in public health and socio-economic development (Xiao et al.,2016). Point sources of anthropogenic activities, such as sewage treatment discharge, and nonpoint sources, such as overland runoff from agricultural areas, can pollute surface water. Pollution from nonpoint sources is more difficult to identify than that from point sources because of complex biotic and abiotic interactions between runoff and the landscape (Sliva and Williams, 2001;Giri and Qiu, 2016). Therefore, variation in land use is a principal factor affecting non-point source pollution (Cheng et al., 2018). Land surface characteristics are generally modified by deforestation, agricultural activities, and urbanization; these activities alter runoff volume and water temperature, cause pollution, increase algal production, and decrease the concentration of dissolved oxygen in water bodies (Ding et al., 2015).
Thus, the majority of water pollution problems are caused by changes in land-use patterns within a given basin, as population and economic activity increase (Lee and Bastemeijer, 1991). Unplanned urbanization, intense agricultural activities, and deforestation are positively correlated with stream pollutants such as carbon, nitrogen, and phosphorous; forests and grasslands, which are less influenced by anthropogenic activities, have negative correlations (Bu et al., 2014;Teixeira et al., 2014;Su et al., 2015;Ullah et al., 2018).
In general, agricultural land use has a strong influence on nutrient parameters in river water such as nitrogen and phosphorus content, fecal coliform bacteria, and trace metal elements (Huang et al., 2013;Cao et al., 2013;Woli et al., 2004;Kang et al., 2010). Fertilizers used in agricultural activities have different chemicals such as nitrogen and phosphorus, which can potentially run off into nearby water sources and damage the nutrient content within the water 3 Effect of land-use pattern on … Rev. Ambient. Água vol. 15 n. 3, e2486-Taubaté 2020 body, affecting its overall quality (Permatasari et al., 2017). Industrial and urban land areas are associated with organic pollution, trace metals, nutrients, and fecal coliform bacteria (Li et al., 2009;Kang et al., 2010;Bu et al., 2014).
Because of the close interconnection with land use, low-order streams-which constitute over 50% of the total stream length-are highly vulnerable to land-disturbing activities such as deforestation, agriculture, and urbanization (Freeman et al., 2007;Ding et al., 2016). These streams transport water, nutrients, organic materials, and sediments from terrestrial uplands to downstream systems. Therefore, anthropogenic activities in low-order streams can increase nutrient loading and reduce nutrient retention, which can result in the eutrophication and hypoxia of distant downstream ecosystems (Dodds and Oakes, 2008;Ding et al., 2016). Despite their important role in maintaining river health, most low-order streams are not routinely monitored for water quality. Therefore, investigating the impacts of land use on water quality in low order streams is crucial. A possible strategy for identifying the principal sources of pollution is the use of multivariate statistical methods on large matrix data (Khan et al., 2016a(Khan et al., , 2016b(Khan et al., , 2017. This method has been successfully used to identify the anthropogenic and natural sources of pollution in surface water bodies (Khan et al., 2016a). The results will be effective not only for protecting local streams, but also for improving downstream water quality (Ding et al., 2016;Norton and Fisher, 2000;Brion et al., 2011).
In this work, we used multivariable analysis to correlate land-use patterns with the physicochemical quality and genotoxic potential of the Extrema River, a low-order stream located in an industrial and agricultural area of Central Brazil. While most of the previous studies correlated land-use patterns with the physicochemical parameters of water, in the present work we also added the genotoxic potential of water. Genotoxic compounds, both known and unknown, are the components of complex environmental mixtures that can have adverse health effects on humans. Therefore, genotoxicity assays should be included as additional parameters in water-quality monitoring programs (Ohe et al., 2004). For the genotoxic analysis performed in the present study, we chose the Allium cepa test because of its simplicity, relatively low cost, high sensitivity, and good correlation when compared with other test systems (Fedel-Miyasato et al., 2014). There is a correlation of 82% between the A. cepa test and the carcinogenicity test in rodents, and when compared with the prokaryotic model, A. cepa is more sensitive than the Ames test (Rank and Nielsen, 1994). The A. cepa test has been validated by the United Nations Environmental Program (UNEP), the World Health Organization (WHO) and the US Environmental Protection Agency (USEPA) as an excellent genetic model for the detection of genotoxicants and is frequently used in environmental monitoring studies (Leme and Marin-Morales, 2009;Sousa et al., 2017;Düsman et al., 2014;Silva et al., 2018;Bollani et al., 2018).

STUDY AREA
The study was conducted at four points along the Extrema River, in the city of Anápolis, located in the mesoregion of the Goiano Center of Brazil. The area is between 16°5'00" S and 16°31'24" S latitude and 48°42'31" W and 49°14'12" W longitude at an approximate altitude of 1017 m. According to the Köppen and Geiger classification, Anápolis presents an Aw tropical climate. The mean annual temperature is 22.2 ± 3.5 o C and the mean annual precipitation is 1,441 ± 238 mm (Climate-Data, 2020). The Extrema River is part of the Antas River watershed and extends for 22,852 km, comprising the regional planation surface IIA (Ferreira et al., 2014). The study area runs from its source to its mouth, where it encounters the Antas River ( Figure 1). The Extrema River experiences an inflow of effluents produced by the Industrial District of Anápolis (DAIA) after treatment at the Anápolis effluent treatment station. These effluents are produced by the chemical, pharmaceutical, automotive, food, and agricultural industries (CODEGO, 2018). According to the Department of Environment of DAIA, the load Elisa Flávia Luiz Cardoso Bailão et al.
of the raw sewage is approximately 3,750 Kg BOD/day.

Water sampling and physicochemical analytical methods
The Extrema River has lentic characteristics and is stratified primarily during the dry season. The low levels of sewer coverage impair the river by increasing the pollutant load, especially during the dry season, and reduce water quality. Consequently, the samples were collected from four locations (P1-P4) along the Extrema River in August 2017, a dry period in the area ( Figure 1). Surface water was collected in three different polyethylene bottles (± 500 mL) from each sampling point. The containers were adequately labeled and kept under refrigeration (-4°C) for further analysis. During sample collection, water temperature (ºC), electrical conductivity (μs/cm), dissolved oxygen concentration (mg/L O2), turbidity (NTU), and pH were obtained using a multiparameter probe (Digimed). Total ammoniacal nitrogen and soluble reactive phosphorus were quantified according to the methods explained by Solórzano (1969) and Murphy and Riley (1962), respectively, in each sample collected from the four points.
An atomic absorption spectrometer (AAnalyst 400, Perkin Elmer, Singapore) equipped with deuterium background correction technology was used to measure the concentration of environmentally hazardous metals through flame atomic absorption spectrometry. A hollowcathode lamp (Perkin Elmer, Singapore) was used as the radiation source. An air/acetylene flame supplied at flow rates of 13.5 and 2.0 L/min, respectively, was used. First, the water samples were acidified. They were then subjected to acid digestion; the instrument was operated under the conditions recommended by the manufacturer. The pH of the solutions was adjusted with the addition of 0.3 mol/L NaOH or HNO3 (Merck, Germany) and by using a pH meter Effect of land-use pattern on … Rev. Ambient. Água vol. 15 n. 3, e2486-Taubaté 2020 equipped with a combined glass electrode. The contents of Fe, Zn, Cu, Cr and Pb were analyzed. The metal quantifications were performed by comparing the samples with dilutions of individual analytical grade standard solutions (1000 mg/L).

Toxicity test using Allium cepa assays
Bulbs of common onion (Allium cepa L.) were purchased from a local supermarket, and 14 bulbs were prepared for each sample. The bulbs were grown in distilled water at room temperature for three days, after which the newly-formed root tips were cut from each bulb and examined morphologically for any visible abnormalities. The bulbs were then treated with surface water collected from the Extrema River. Another set of onion bulbs was placed in distilled water to be a negative control. After 48 h in the test solutions, seven bulbs were used for macroscopic analysis of toxicity and the root tips of five bulbs were used for microscopic analysis. During macroscopic analysis, each root was measured in centimeters with a ruler. During microscopic analysis, the root tips of each bulb were collected and fixed in a Carnoy's fixative solution (1 acetic acid: 3 alcohol) for 24 h. They were then stored in 70% alcohol. For analysis, the root tips were rinsed a few times with distilled water and hydrolyzed with 5 M HCl solution at room temperature for 1 min. Following hydrolysis, the roots were squashed in acetic acid (45%, v/v), and a coverslip was placed on each slide. The coverslips were removed after freezing in liquid nitrogen, and the roots were stained with Giemsa (5%) for 5 min. The colored slides were evaluated using a Zeiss Optical Microscope (Zeiss, Oberkochen, Germany) with 40x and 100x objectives. In total, 1,000 meristematic cells per slide and 5,000 cells per treatment were analyzed. The cytotoxic potential was calculated through determination of the mitotic index (MI). The MI was calculated for each treatment using the ratio of the number of dividing cells to the total number of cells (Seth et al., 2008). The genotoxic potential was estimated using the number of chromosomal and nuclear aberrations (CNA), and the mutagenic potential was estimated using the number of micronuclei (MN).

Spatial analysis of use and land cover
Landsat-8 Operational Land Imager (OLI) images were chosen for the extraction of information on land use. To match the water quality data, we used the image with path/row number 222/71, which was obtained on September 11, 2017. Further image processing analysis was carried out using ENVI 5.0. According to the methodology described by Santos et al. (2017), we snipped a slit in the area of interest and analyzed the color composition of that area by stacking spectral bands 6 (1.560-1.660 µm), 5 (0.845-0.885 µm), and 4 (0.630-0.680 µm). Highlights and contrast were used to facilitate the interpretation of the targets and terrestrial features in the image. For the land-use and ground-cover survey, the supervised maximum likelihood classification method was used-MAXVER (Santos and Lima, 2018). Land use, including that of natural and anthropic areas, was classified into seven types: a) Agriculture, which includes areas occupied by annual and perennial crops; b) Water, which includes lakes, rivers, and streams; c) Urban area, which encompasses roads, residences, and industries; d) Pasture, which comprises pasture areas intended for cattle raising; e) Woodland, which consists of forests with vegetation areas from native Cerrado; f) Burn scar, which includes regions where burnings or forest fires have occurred; and g) Exposed soil, which includes areas without vegetation cover. The overall accuracy of this land-use classification was 96.6%, and the Kappa coefficient was 0.96, showing high classification accuracy and indicating that the land-use map could be employed in the subsequent analysis. The spatial distribution of land-use and watermonitoring sites in the study area is shown in Figure 1. The proportions of land-use types in the four zones were also computed using ArcGIS 10.2. Buffers of 500 m, 1000 m, and 2000 m radii were created around each water collection point. Geoprocessing tools were used to analyze the total area occupied by each of the land-use types at the different collection sites. The attribute tables containing information on each buffer were exported from ArcGIS in an Excel-compatible format and were used in the total and percentage area calculations by mapping class.

Statistical analysis
In the A. cepa analysis, the onions were divided for five treatments, each containing five replications. The values obtained from toxicity, genotoxicity, and mutagenic analyses were evaluated through analysis of variance (ANOVA) and compared through Tukey post-hoc tests. The obtained p < 0.05 value was considered as indicative of significance.
Cell parameters such as root length, MI, CNA, and the incidence of MN were considered response variables for the set of sampled points. Furthermore, all relevant water quality attributes, such as water temperature, electrical conductivity, dissolved oxygen content, turbidity, pH, trace metals, ammoniacal nitrogen, and soluble reactive phosphorus content, were used as environmental parameters. For the characterization of water quality and determination of the interaction between biological parameters, the corresponding canonical analysis (CCA) was applied. The parameters identified for land-use characterization at different distances in the buffers were also analyzed using the same test. Quantitative biological parameters were compared with water-and land-use characteristics using simple polynomial regressions; the statistical significance and regression coefficient were derived.

Physicochemical characteristics of the water collected along Extrema River
The results of physical analyses (temperature, turbidity, electrical conductivity, dissolved oxygen content, and pH) of the samples collected from the Extrema River (P1-P4) were compared with the standards established by the Brazilian law for Class 2 water resources (Table 1), which includes water for human consumption, agriculture, recreation, and fishing activities (CONAMA, 2005).
The samples from P1, P2, and P4 showed results that were within the standards established by Brazilian law. The sample from P3 presented low levels of dissolved oxygen and low pH. It is important to highlight that although Resolution 357 of CONAMA does not establish limits for electrical conductivity, according to the Environmental Company of the State of São Paulo (CETESB, 2015), levels above 100 μs/cm are indicative of impacted environments. That is the case at P1 and P3. Thus, with respect to physical parameters, only water from P2 and P4 appears to be appropriate for human consumption. Water collected from the sampling points was analyzed for N, P, and trace metal contents (Table 2). Water collected at P3 presented the highest N, P, and Fe contents. The Fe and Cr contents were above the limits established by Brazilian law at all sampling points. Therefore, 7 Effect of land-use pattern on … Rev. Ambient. Água vol. 15 n. 3, e2486-Taubaté 2020 with respect to chemical content, water from the Extrema River is not appropriate for human consumption.

Cytogenotoxicity of water collected from the Extrema River
Water collected from the Extrema River (P1-P4) was evaluated using the A. cepa model to verify its cyto genotoxic potential (Table 3). Water collected at P1 and P2 was found to increase A. cepa root growth, suggesting the presence of substances with mitogenic activity. In agreement with this result, the MI of P1 was higher than the MI of the NC. However, the MIs of the other sample points (P2, P3, and P4) were similar to that of the negative control. In contrast, the number of CNAs was higher in A. cepa meristematic cells treated with water collected at P3 than in the negative control, indicating the genotoxic effect of the water collected from P3. The most common CNAs observed were chromosome bridge, lagging chromosome, chromosome stickiness, chromosome loss, chromosome break, C-mitosis, binucleated cell, and nuclear bud (Figure 2). Water samples collected at P2, P3, and P4 were seen to induce MN formation in A. cepa meristematic cells (Table 3 and Figure 2) when compared with the negative control, indicating its mutagenic potential.

Relationship between land-use pattern, water quality and cellular alterations
The land-use analysis data, obtained from land-use mapping and soil cover, are presented in Table 4.
Analysis of the regression coefficient showed that the A. cepa root length was directly proportional to the pH of the water and inversely proportional to the amount of Fe and ammoniacal nitrogen in the water. In agreement with the ammoniacal nitrogen relation, the root length was inversely proportional to pasture areas with a radius of 2000 m. CNAs were positively related to Fe and to orthophosphate content. CNA showed a positive correlation with the area occupied by water with a radius of 500 m, agricultural area with a radius of 2000 m, and an area of exposed soil with a radius of 2000 m. MN frequency was inversely proportional to water turbidity and did not present any relationship with land use. Also, because MI was not statistically different among the sampled collection points, the MI did not present a specific relationship with any parameter of water quality and land use investigated in this work (Table 5).
Investigations regarding the characteristics at each sample collection point along the Extrema River revealed that P1 presented a relation with urban areas in the three analyzed buffer distances (Figure 3). This was expected because the highest values of area occupied by urban class land use were observed near P1 (Table 4). This point presented the lowest Cr content, placing P1 opposite to the Cr vector in the canonical correspondence chart (Figure 3). By extension, P1 also presented the lowest relationship with MN, CNA, and MI (Figure 3).

9
Effect of land-use pattern on … Rev. Ambient. Água vol. 15 n. 3, e2486-Taubaté 2020 Table 4. Measurement of areas occupied by the land-use classes for landuse analysis. Agri: Agriculture; Wat: Water; Urb: Urban area; Pas: Pasture; For: Forest; Burn: Burn scars; Soil: Exposed soil. 500 m, 1000 m, and 2000 m codes are relative to the buffer radius lengths presented in Figure 1. Table 5. Regression coefficient (R²) and statistical significance (p) of the relationship between parameters investigated using the Allium cepa model and those related to water quality and land use.

Allium cepa parameters Water-and land-use parameters R 2 p
Root length pH 0.98 0.01  The relationship between water quality, soil use and toxicity parameters with the water samples collected along the Extrema River. A canonical correspondence analysis showed that the water collected along the Extrema River was, in general, more related to the landscape pattern than to water parameters (physicochemical and toxicological). In agreement, Allium cepa parameters were more related to the landscape use characteristics than to the water quality parameters per se.
P2 and P4 are interrelated and related to burn scars and forest use (Figure 3) in the different distances analyzed, because these classes of land use occupy more areas around the two points (Table 4) and present land-use patterns most related to dissolved oxygen (DO), water pH, Cr content, and A. cepa parameters [root length (RL), CNA, MN, and MI]. Burned areas are related to P2 and P4 land-use patterns; thus, the effects observed in the water at these points are probably caused by this type of land use. P3 is the point most related to agriculture, pasture, and exposed soil classes. In agreement, P3 was related to NH4-N, Fe, orthophosphate, lower pH, and electrical conductivity (EC).

DISCUSSION
A water quality assessment is the evaluation of the physical, chemical, and biological nature of the water. However, the majority of the studies consider only one method for monitoring water quality (Chapman, 1996). In this work, we evaluated the physical, chemical, and toxic potential of water collected from the Extrema River, a low-order stream. To complement this strategy, we also evaluated the land-use patterns in the basin and derived the relationship between land use and water quality.
Water collected at P3 was the most affected and presented low levels of dissolved oxygen, low pH, EC above 100 μs/cm, high levels of P, Fe, and Cr, and a high number of CNAs and MN. High phosphorus content could explain the low levels of dissolved oxygen. Eutrophication management has centered on phosphorus control (Dodds et al., 2009). In this scenario, orthophosphate is important because it is the principal form of phosphate assimilated by autotrophs (Correll, 1998). The nutrient input can cause an increase in algal blooms, which consume O2 and are detrimental to water quality (Dodds et al., 2009). Phosphorus levels may increase because of fertilizer use and the inflow of municipal and industrial wastewater (Conley et al., 2009). A sewage treatment plant near P3 that disposes its effluents can be considered as the phosphorus source. Moreover, the land-use study revealed that P3 was the point most related to agriculture, pasture, and exposed soil. The increase in nutrients in the soil because of agricultural fertilization makes surface water more susceptible to the inflow of these elements 11 Effect of land-use pattern on … Rev. Ambient. Água vol. 15 n. 3, e2486-Taubaté 2020 (Permatasari et al., 2017). Soil areas exposed by erosion, through which nutrient elements are carried to lower parts of the terrain (usually where the water bodies run), also contribute to the inflow of these nutrients in surface waters (Bertol et al., 2003).
Furthermore, the higher EC observed at P3 could be explained by the chemical element input into the water at this point; these chemical elements are derived from activities in the surrounding areas, such as agricultural fertilization and wastewater disposal. Similar results have been reported by Ribeiro et al. (2014). An increase in EC with an increase in the agricultural area was observed. The concentration of ions in water increased with an increase in the proportion of agricultural land, corroborating the potential for nutrient transport from soil to water (Ribeiro et al., 2014).
The high levels of CNA and MN observed in A. cepa cells treated with water collected at P3 can be explained by the high content of Fe and Cr detected. Transition metal ions, including Fe and Cr, can catalyze hydrogen peroxide decomposition with the generation of hydroxyl radicals (HO•), known as Fenton and Fenton-like reactions. Therefore, in biological systems, transition metals can undergo redox cycling reactions and produce reactive oxygen species (ROS) (Chumakov et al., 2016). ROS can promote DNA breaks and mutations responsible for chromosome aberrations and micronucleus formation (Luzhna et al., 2013). Interestingly, we observed a positive relation between CNA and Fe content. The presence and bioaccumulation of trace metals in surface water and the influence of these elements on toxicogenomic damage both in A. cepa and in the fish Oreochromis niloticus were previously investigated .
Similar to our work, previous studies have reported that some anthropogenic activities in river basins, such as urbanization and agriculture, play an important role in water quality as sources of pollution and cause significant changes in the soil surface and the runoff of nutrients (Sliva and Williams, 2001;Buck et al., 2004;Lee et al., 2009;Huang et al., 2013). This relationship highlights the urgent necessity for preserving the Extrema River spring. Similar to P3, the water collected at P1 presented an EC above 100 μs/cm and levels of Fe and Cr above the limits established by Brazilian law (CONAMA, 2005). Although P1 is considered a river spring, it is located in an area with high urbanization and pasture use, which contributes to its impacted environment.
The results obtained for P2 and P4 are interesting because both are related to burn scars and forest use and are more related to dissolved oxygen (DO), water pH, Cr content and A. cepa parameters [root length (RL), CNA, MN, and MI]. These results seem in contrast with what was previously observed, because forests are supposed to mitigate water quality degradation (Sliva and Williams 2001;Lee et al., 2009). However, it is important to highlight that the burned areas also had correlations with the same parameters as P2 and P4. Therefore, it is probable that the effects observed in the water at these points are caused primarily by the burns. Forest-land water cannot always be considered adequate for human consumption, as the nitrogen output is high in forested areas during storms, heavy rainfall, and burns (Tong and Chen, 2002;Smith et al., 2011). Following burns, increased erosion rates and changes in runoff generation and pollutant sources may greatly increase the flux of sediment, nutrients, and other water quality constituents, potentially contaminating water bodies. In the first year post-fire, suspended sediment, total N, total P, Fe, Mn, As, Cr, Al, Ba, and Pb exports commonly increase compared with unburned areas (Smith et al., 2011).

CONCLUSION
The physicochemical and genotoxic results obtained from this study showed that the surface water of the Extrema River is inappropriate for human consumption and in urgent need of preservation actions. Additionally, the A. cepa assay was shown to be a useful genotoxic hazard recognition tool that should be implemented in monitoring programs. The multivariable analysis associated the physicochemical and cytogenotoxic parameters with the different uses of the land, demonstrating that the type of land use changed the water characteristics.

ACKNOWLEDGMENTS
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The author Leovigildo Aparecido Costa Santos was supported with a scholarship from Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES); the authors Sara dos Santos Almeida and Patrícia Lima D'abadia were supported with a scholarship from Fundação de Amparo à Pesquisa do Estado de Goiás (FAPEG); the authors Elisa Flávia Luiz Cardoso Bailão, Samantha Salomão Caramori, and Luciane Madureira de Almeida were supported by Universidade Estadual de Goiás with a fellowship of the program PROBIP (Scientific Production Support Program). This study was financed in part by the CAPES -Finance Code 001 (Convênio n° 817164/2015 CAPES/PROAP) and by FAPEG.