1. Introduction
The land use of a catchment area is one of the key factors affecting water quality [
1]. Water collected in small reservoirs located in rural areas may be polluted, especially in terms of phosphate or ammonium nitrogen concentrations and salinity indices. The amount and type of substances in surface water varies greatly and depends on the conditions in the catchment area [
2]. If these substances occur in excess, they deteriorate the quality and utility value of water, and thus become pollutants whose sources can be divided into natural and anthropogenic [
2,
3]. The reasons for the first of them are rock aeration, the erosion and leaching of soils, clean precipitation, or the decomposition of dying plant and animal organisms [
4].
The reason for our study is the presence, in agricultural areas of forests, of mulch in which humus soils occur, which may cause a greater pollution of water with mineral-organic substances than the extensively used (meadows and arable land) mineral soils [
5]. The results are strongly inscribed in water-related policies and European Union legislation aimed at rational management of water resources, including socio-economic and climate change [
6]. Anthropogenic pollution, which is generally considered to be more dangerous to the aquatic environment, is linked to all human activities, in particular industrial and agricultural production, and the progression of urbanization. A significant share of the degradation of water is from rural areas, being used for agricultural purposes; through the chemicalization of agriculture and accelerated soil erosion; improper use of plant protection products and incorrect storage; and the use of sewage or waste from livestock farming [
7,
8].
Rural settlements also contribute to water pollution, which are connected with diversified population density and land development, non-agricultural economic activity, run-off of polluted water from properties, and communication areas [
9,
10]. The source of hazards may also be deposited gases and dust emitted to the atmosphere from households, which, in dissolved form, reach the surface of the Earth, along with precipitation [
11]. A major problem in Poland is also the poor quality of water resources, which are increasingly subject to anthropogenic influences limiting the availability of good quality water [
12,
13]. Therefore, the objectives of the work were divided, and an attempt was made to evaluate the following research problems: (i) physiographic characterization of the upland flysch catchment dominated by agricultural land; (ii) evaluation of the quality indices using multivariate statistics methods; (iii) assessment of anthropogenic pressures and interactions of various factors on heavy metal concentrations; and (iv) estimation of the degree of risk and intensity of water erosion, and the impact of land use on surface water quality.
This study on the hydrochemical assessment of surface water quality has a cognitive and practical aspect, which is suitable for evaluating and issuing expert opinions on the conditions of the location of small retention reservoirs and systematic water quality testing. Moreover, a detailed description of the natural environmental conditions as well as the condition of use and management of the investigated elementary catchments are presented. Qualitative studies, as shown in this paper, are necessary in order to take measures to protect water retained in water reservoirs from pollution.
2. Material and Methods
2.1. Study Area
The investigated catchment of the Korzeń stream and its tributaries is located in the Ciężkowice Foothills in the eastern part of the Małopolskie Voivodeship, in the southern part of Poland. The catchment is dominated by arable land and plant production. It is under pressure from natural factors, mainly water erosion and leaching of soils, as well as anthropogenic factors, mainly related to agricultural activities and rural settlement, including problems with water and sewage management and traffic loads. The stocking density of livestock during the study period was quite low. The total density of the asphalt and paved roads, as well as agricultural and forest roads, is 8.64 km·km
−2. The study area is situated between 21°3.17″ and 21°12.25″ east longitude, and between 49°50.35″ and 49°58.57″ north latitude. According to the Kondracki division [
14], the catchment is located in the province of the Western Carpathians and the subprovince of the Outer Western Carpathians.
2.2. Geometry and Morphometry of the Study Area
The catchment area is approximately 10 km
2 and lies at an altitude of 231.3–396.8 m above sea level (
Figure 1)—the weighted average height is 295.3 m above sea level. The catchment of the Korzeń stream has a highly concentrated area, as evidenced by the value of the calculated Gravelius index (K) −1.34. It is also characterized by a shape similar to a square (form index: CF = 0.45) and circle (circularity index: CK = 0.55). Hydrological calculations show that each km
2 of the catchment area in the Outer Western Carpathians has a significant outflow of water: SSQ = 0.008 m
3·s
−1·km
−2, Q
50% = 0.0631 m
3·s
−1·km
−2, and Q
1% = 3.109 m
3·s
−1·km
−2. Other important physiographic parameters of the catchment are presented in
Table 1. This indicates the advisability of building small retention reservoirs in such catchments, which can reduce the risk of flooding and reduce the effects of droughts by storing water.
In the Korzeń stream catchment, the weighted average land gradient is 11%. Slopes of 5–10% and 10–18% dominate, which occur on 33.5% and 37.5% of the area, respectively. The slopes above 27% cover only 7.4 ha (0.8%) of the area, the share of slopes between 18% and 27% is 11.9%, and the slopes smaller than 5% occupy 16.3% of the area (
Figure 2). In the Korzeń stream catchment area, the majority are areas with northern, north-eastern, and north-western exposures, which together, cover 49.5% of the area. There is also a significant share of the eastern and western exposure areas, whose share in the catchment area is 17.6% and 11.4%, respectively. The areas with southern exposures that are the most favorable from an agricultural point of view cover 21.5% of the area.
In spite of such a diversified relief, which is characteristic for submontane areas, the catchment is well exposed to sunlight, which positively influences the development of agriculture (
Figure 3).
2.3. Soils and Land Use
The northern part of the Ciężkowice Foothills, in which the catchment is located, is situated within the Silesian Nappe, which belongs to the Outer Western Carpathians, also referred to as the flysch Carpathians, because of their geological structure. The Silesian Nappe is the largest unit of the Outer Western Carpathians. Its western part is built of black shale and limestone, coarse sandstone, shale with siderite, siliceous sandstone, and lower Cretaceous shale, as well as a several thousand meters-thick sandstone complex originating from the upper Cretaceous with a small share of shale, referred to as the Godulski Strata and Istebna Strata. The lower Cretaceous formations are strongly corrugated and form several slices at the northern edge of the nappe, while the upper Cretaceous and Palaeogene complexes show a fairly regular fold structure. Along the northern fringe of the Silesian nappe there is a narrow strip of the Subsilesian series. In the section from Cieszyn to Brzesko, this series creates narrow, strongly disturbed scales, overlapping, along with the Silesian series, the deposits of the Miocene. To the southeast of Brzesko and Tarnów, the Silesian Nappe, together with the Subsilesian Nappe, overlap the formations of the younger Palaeogene of the Skole Nappe [
14].
The catchment of the Korzeń stream is characterized by a small typological diversity of soils. The dominant soils are Luvisols, Hamplic Cambisols, and Stagnic Cambisols, which have a 51.6%, 37.5%, and 9.2% share, respectively, in the catchment area (
Figure 4). In 92.1% of the area, the topsoil genetic layers are made up of silt (Si), and there are also silt loam (SiL), loam (L), and sandy clay loam (SCL). In the largest area, Luvisols are found on the tops and slopes, in all parts of the catchment area of the Korzeń stream. These soils are mostly made up of silt (Si) lined with silt loam (SiL). In the southern part of the catchment, mostly under forests, there are Hamplic Cambisols formed from sedimentary rocks with a non-carbonate binder. They are made up of silt (Si) and loam (L), passing at a depth of 25–50 cm into weekly skeletal sandy loam (SL) deposited on rubble or rocky soils. In the valleys of the Korzeń stream and its tributaries, there are almost exclusively Stagnic Cambisols (
Figure 5). In various parts of the catchment (
Figure 5), they developed from silt loam (SiL), silt (Si), and loam (L).
In the Korzeń stream catchment area, because of the average land gradients and fertile soils, there is 60.7% of arable land in the area. The arable land is evenly distributed throughout the object and is cultivated mainly with wheat, potatoes, and fodder beet. Grasslands, amounting to 7.7% of the catchment area, occur almost exclusively in the river valleys. The forestation of the catchment area of the Korzeń stream is 25.9%. Larger forest areas, mainly with beech, oak, and fir stand, are located in the central and eastern part of the catchment, while minor wooded and shrubby enclaves occur along all of the watercourses. In addition, there is only a small percentage (2.1%) of uncultivated agricultural land in the catchment, and 23.1 ha (2.4%) of the areas are with scattered development—they are located in the central and southern part of the catchment, unfortunately very often close to watercourses.
2.4. Topography and Hydrological Geometry
Apart from the principal part of the research works related to the determination of physical and chemical properties of water, natural environmental factors as well as the current state of use and management of the analyzed catchment were determined using cartographic materials. The physiographic elaboration of the research catchment was done by the authors.
The following geometrical parameters of the catchment were determined: surface area (A) in km2 calculated by planimetering the area between the topographical boundary of the watershed; average length of the catchment (Lz) in km, determined in a straight line from the furthest point on the watershed to the closing section; and average width (Bz) in km, calculated from the quotient of the catchment surface area and its average length. The following indices were also calculated: form (CF), resulting from the comparison of catchment shape to a square; elongation (CW), expressing the ratio of the diameter of a circle of area equal to the catchment area to the catchment length; circularity (CK), the ratio of the catchment area to the area of a circle of circumference equal to the circumference of the catchment; and watershed development (Gravelius, K), expressing the ratio of circumference of the catchment to the circumference of a circle of area equal to its surface. Moreover, all of the watercourses were digitized, which made it possible to calculate their length (L), density of the river network (Gs), and average slope of the main stream Korzeń (Jc). The following streamflow characteristics were also calculated: annual average flow (SSQ), average minimum flow (SNQ), and minimum annual flow (NNQ), as well as probable outflows, Q50% and Q1%.
The shape of the catchment area is shown on a hypsometric map and land slope map. The minimum, maximum, and weighted average absolute height of the catchment were determined. On the basis of the geospatial analysis, the share of individual classes of slopes and the exposure of slopes in the total area of the catchment were determined, and the weighted average slope of the area (J) was calculated. The current use of the catchment is shown on a map prepared with the use of Geographic Information System (GIS) tools on the basis of 1:2000 scale cadastral maps and orthophotomaps, as well as information obtained from the direct inventory of the catchment. As part of the field research, a survey was carried out covering the existing and potential sources of water quality hazard in the catchment. The surveys specified the number and type of households, the way they were supplied with water and the methods of discharging domestic and livestock sewage, the number of inhabitants, the number and density of livestock converted into livestock units (LSU), and the km2 of the catchment area.
The substrate of the catchment was characterized on the basis of digital soil and agricultural maps made available by the Marshal’s Office of the Małopolska Voivodeship. On their basis, a map was generated with the help of GIS software, on which the soil types were marked with appropriate colors. The text contains information on the percentage share of the particular types and textures of soil [
15] in relation to the total catchment area.
2.5. Determination of Surface Water Quality Indices
Hydrochemical tests were carried out at six measurement and check-control spots in the years 2007–2018. Moreover, the concentrations of six heavy metals and the values of two microbiological indices were examined once a quarter. In total, 864 water samples were collected in the research period. During the sampling, for analysis purposes, the pH, using the CP-104 pH meter (ELMETRON, Zabrze, Poland); specific electrical conductivity (EC), using the CC-102 conductivity meter (ELMETRON, Zabrze, Poland); and the temperature and dissolved oxygen content (DO), using the CO-411 oxygen meter (ELMETRON, Zabrze, Poland) were measured in situ in the waters flowing through the streams under investigation. The following were determined in laboratory: using the gravimetric method, the concentration of the total suspended solids (TSS) and total dissolved substances (TDS), and the concentration of calcium (Ca2+), sodium (Na+), potassium (K+), magnesium (Mg+), manganese (Mn2+), total iron (FeTot), and chromium (Cr6+) were calculated; zinc (Zn2+), cadmium (Cd2+), copper (Cu2+), nickel (Ni2+), and lead (Pb2+) were determined by atomic absorption spectrometry on UNICAM SOLAR 969 spectrometer (unicam Software GmbH, Georgensgmünd, Germany); the concentration of nitrogen in ammoniacal (N-NH4+), nitrite (N-NO2−), and nitrate (N-NO3−) were calculated; the total phosphorus (TP), phosphates (PO43−), and chlorides (Cl−) were determined by flow colorimetric analysis using the FIAstar 5000 instrument (FOSS, Hillerød, Denmark); the concentration of sulphates (SO43−) was determined by the precipitation-gravimetric method; the five-day biochemical oxygen demand (BOD5) was calculated using the Winkler titration method; the chemical oxygen demand (COD-Mn) using boiling titration with KMnO4; and the number of coliforms and fecal coliforms were determined by membrane filtration on lactose media, after incubation at 37 and 44 °C, with a tolerance of 0.5 °C. The concentrations of the ammonium (NH4+), nitrite (NO2−), and nitrate (NO3−) ions were calculated based on the results of the N-NH4+, N–NO2−, and N-NO3− determinations. The heavy metals and microbiological indices were determined on a quarterly basis, while the other indices were determined once a month.
2.6. Erosion Model
The method proposed by Józefaciuk and Józefaciuk (1999) was implemented using spatial information systems tools (GIS), the basis of which formed a study on the risk of surface water erosion for particular catchment areas was developed. In order to draw up the risk map, the following maps were used: slope maps generated from the numerical terrain model (NMT), soil–agricultural and land use maps, and information about precipitation in the studied region. The minimum area of the region to be analyzed, with an NMT resolution of 20 m per pixel, was 4 a. The map was filled with information on the percentage share of the areas with a certain degree of erosion risk in the total catchment area. The following computer programs were used to draw up the physiographic features and erosion hazard studies, as well as the mathematical calculations and statistical analyses: MapInfo Professional® 9.0 (Pitney Bowes Inc, Stamford, CT, USA), Surfer® 8.0 (Golden Software, Golden, CO, USA), CorelDRAW Graphics Suite X4 (Corel, Ottawa, ON, Canada), and ArcGIS 10 (Esri, Redlands, CA, USA).
2.7. Water Quality Indices
The degree of water pollution in the investigated streams was assessed using the Burchard and Dubaniewicz formula [
16,
17]:
where Wz is the impurity coefficient, SD
t is the permissible dissolved oxygen concentration and fecal coliforms for class I waters (the limit values according to the regulations of 2016 were used for the calculations), SW
t is the average (over the test period) dissolved oxygen concentration and fecal coliforms, SW
n−2 is the average (over the test period) concentration of the other indices used in the calculations, SD
n−2 is the permissible concentrations of other indices assumed for calculations for the class 1 water quality (limit values according to the regulations [
18] were used for the calculations), and n is the number of indices taken into account.
In accordance with the pragmatics of the method, water is considered to be clean when the impurity coefficient, Wz, is not greater than 0.75. For the values of the coefficient, 0.76–1.00, 1.01–1.50, 1.51–2.00, and above 2.00 water were considered to be slightly polluted, clearly polluted, heavily polluted, and waste, respectively.
On the basis of the hydraulic load (O
hzb), taking into account the physical parameters of the water reservoir and the calculated annual load of the total phosphorus, a projection of the trophicity of the planned water body was prepared. The hydraulic load of the reservoir (in m·year
−1) was calculated from the following formula [
19]:
where z is the average reservoir depth (m) and t is the average water retention time (years).
It was assumed that the annual phosphorus load of the reservoir corresponds approximately to the phosphorus load transported in the stream at the average annual flow. The criteria according to which an oligotrophic reservoir is a body loaded with phosphorus load of L
Ppow (in grams P, per 1 m
2 area, and in one year), such that L
Ppow < 0.01 (O
hzb + 10), were used for the assessment. A reservoir is assumed to be eutrophic when L
Ppow > 0.03 (O
hzb + 10) and mesotrophic when 0.01 (O
hzb + 10) < L
Ppow < 0.03 (O
hzb + 10) [
18].
The objective of the environmental risk assessment (ERA) is to assess the trend and extent of the adverse effects that may occur or do occur as a result of exposure to environmental stressors. In this study, an environmental risk assessment of the flysch stream was carried out, assigning a weight of w
i = 1 ÷ n
i to each sample, where n
i is the number of samples in the i
th range of the numerical values to which the data are sorted in the statistical analysis. In addition, the hazardous concentration at the fifth centile (HC
05) has been calculated. The Hazard quotient index (HQ) was obtained by dividing the measured concentration (MEC) at each sampling point by HC
05, according to the following formula:
The idea of this method assumes that with values of HQ ≥ 10 there is a high ecological risk for aquatic organisms, values between 1.0 ≤ HQ < 10 indicate a moderate risk, values between 0.1 ≤ HQ < 1.0 indicate a low risk, and with HQ < 0.1 the risk is reduced [
19].
The heavy metal pollution index (HPI) was determined by assigning a rating or weight (W
i) for each selected parameter. Weight is any value in the range of 0 to 1, and its choice reflects the relative individual importance for the quality aspects. It can be defined as inversely proportional to the standard limit value. The assessment of heavy metal pollution in the surface waters of the study area was carried out on the basis of a calculated index [
20,
21], as follows:
where Q
i is the subindex of the next i
th parameter, W
i is the unit weight assigned to i
th parameter, and n is the number of parameters considered.
The subindex (Q
i) is calculated from following equation:
where M
i is the monitored values of the heavy metal of the i
th parameter, I
i is the ideal value of the i
th parameter, and S
i is the maximum allowed or recommended value of the i
th parameter.
The unit weight (Wi) of the parameter is determined as follows:
where k is a constant of proportionality.
Heavy metal contamination is an issue of serious concern. The heavy metal evaluation index (HEI) gives general information on water quality in relation to the concentrations of heavy metals, and is expressed by the following formula:
The Mi and Si parameters are monitored and the maximum admissible concentration (MAC) of a given parameter, respectively. The higher the concentration of the metal compared to its respective normative value, the worse the water quality.
2.8. Statistical Analysis and Data Processing
A Principal Component Analysis (PCA) is a linear combination of the output variables and is used to, among other things, reduce the number of variables. The purpose of the PCA is to generate the main components that are constructed as a linear combination of the output variables. Each of the generated main components should be treated as a weighted combination of the original variables. Those components that explain the variability in a significant way were selected for analysis—the limit value is usually assumed to be 75%. A graphical illustration of the PCA analysis is the projection of data, usually in two-dimensional space (PC1 and PC2). The PCA analysis was performed in PQStat (PQStat Software, Poznań, Poland) statistical program in version 1.66.
Based on the obtained results of the hydrochemical tests, an assessment of the degree of differentiation of the concentration of the tested water quality indices was carried out by means of a regression for each variable. Multivariate adaptive regression splines (MARS) were used for the projections. MARS is a non-parametric statistical procedure, particularly useful for a larger number of variables [
22]. It does not require assumptions of a functional relationship between the dependent and independent variables. The applied method additionally takes into account the interactions between the explanatory variables. The general form of the weighted sum of the base functions has the following formula [
23]:
where y is a function of the predictive variables of X (including interaction), β
0 is the starting ordinate, β
m is the weighted ordinate (by weights), and h
m(X) is the sum of one or more base functions.
The projection was based on the statistical analyses selected in the previous procedure, which reflected the variability of the water quality indices, on the accompanying or disturbing factors, that is the slope or slope exposure, as well as on the soil type. MARS is a method for selecting the most important features (predictors) and reducing (removing) the least important base functions. On the basis of the measurements carried out, a model was created for all of the test points (based on equal cases investigated). Erosion processes in unfavorable habitat conditions may lead to the destruction of the entire soil profile, over time leading to the formation of an erosion landscape. Land at such risk requires comprehensive protective measures, with part of the arable area to be used for anti-erosion equipment. In order to implement the research methodology related to the selection of the significant predictors in the applied model of the explanatory variables, the least squares method was used to estimate the parameters. The generalized cross validation (GCV) was calculated according to the following formula [
23]:
where y
i is the i
th observed response value; f
M(x
1) is the fitted response value obtained for the i
th observed predictor vector; n is the number of cases (observations); M is the maximum number of base functions selected for the model; and P(M) represents the effective number of parameters, which is a penalty measure for the complexity.
GCV determines the error in fitting the model to the real data and takes into account both the residual error and the complexity of the model. Therefore, the applied statistical measures allow for identifying significant changes in the research area. Both the hydrochemical and physiographic parameters were compared. The results of the MARS model were elaborated in the statistical program Statistica 12.5 (StatSoft, Tulsa, OK, USA), and the selection of the functions to the model was made using the Gnu Regression, Econometrics and Time-series Library (GRETL) calculation program (General Public License—open source).
The significance of the dynamics of the changes in the values of the water quality indices was assessed using the Permutational multivariate analysis of variance (PERMANOVA) two-factor test. The homogeneity of variance was tested using the Levene test. To assess the differences between the land use forms, an analysis of the inequalities of the variance was carried out with the F-Welch test. The calculations were performed with the use of PAST software version 3.14 (Natural History Museum, University of Oslo, Oslo, Norway).
5. Conclusions
The pollution of water reservoirs in rural and urban areas is proof of the threat to surface waters posed by area and point sources. In the assessment of the chemical state of the flysch stream, rural areas are often classified as heavily polluted. The obtained results indicate the influence of an anthropogenic factor on water quality. When assessing water quality, the EC and TDS parameters turned out to be the determining factor, according to the multidimensional analysis. Our studies confirmed that the physicochemical composition of the flysch stream waters is modified by the anthropogenic factor, especially in the case of water bodies located on slopes with a higher gradient. In some cases, the average values of the water quality indices did not show dependence on the anthropogenic factors. In assessing the impact of the anthropogenic factors, other aspects, such as the proximity of households to watercourses, should also be taken into account, which in the case of disorderly water and sewage management, may have a significant negative impact on water quality. The assessment of water pollution by the Burchard and Dubaniewicz method showed that the tested waters were heavily contaminated, mainly with suspended solids, saline ions, and organic substances, which were influenced by a significant share of arable land, moderate forest cover in the catchment, and a significant risk of water erosion. Because of the low concentrations of nutrients, the waters flowing out of the catchment were not at risk of eutrophication during the study period. For the same reasons, it is predicted that a small retention reservoir may be of a mesotrophic type. In addition, the MARS model has shown that EC and TDS can also have an impact on water quality—the HQ index increased with their values. In order to maintain good water conditions, it will be necessary to monitor water quality in order to be able to react in the event of adverse changes within their catchment areas. Our multiannual results show that no significant changes in the amount of dissolved nitrogen were detected in the examined watercourses, because only its form changed. The obtained results indicate that the use of the catchment area should be treated as a factor modifying the concentration of the tested water quality parameters. Such a disturbance should not be referred to as strong dependencies, but rather it is possible to indicate only certain, usually weak tendencies. The results of the water quality studies showed that in the flysch catchment, measures should be taken to reduce soil erosion, and the inflow of point and area pollution from settlement areas and agriculturally used areas.