Relationship between Morphological Characteristics and Quality of Aquatic Habitat in Mountain Streams of Slovakia

: This study evaluated the relationship between abiotic ﬂow characteristics and habitat quality. Habitat quality was assessed using the Instream Flow Incremental Methodology (IFIM), which uses bioindication. Brown trout was selected as a bioindicator because of its sensitivity to morphological changes and its occurrence in sufﬁcient reference reaches. The correlation between the morphological characteristics of the stream and the area-weighted suitability (AWS), which represents habitat quality, was evaluated. Fifty-nine reference reaches of ﬁfty-two mountain and piedmont streams in Slovakia were analysed. The correlation analysis demonstrated the strongest relationship between the AWS and the stream depth and width. The relationship between the water surface area and the AWS indicated that, for mountain streams, there is a signiﬁcantly increasing trend of the AWS value with increasing surface area. Considering piedmont streams, the AWS variation with a change in the water-surface area was minimal. These results can form the basis for deriving regression equations to determine habitat quality. Such a procedure can signiﬁcantly simplify the evaluation of the quality of aquatic habitat, making it much more accessible for design practice.


Introduction
The need to address fundamental research questions leading to advances in related science and key management issues has led to the establishment of the scientific discipline ecohydraulics [1] that brings together biologists, ecologists, fluvial geomorphologists, sedimentologists, hydrologists, hydraulic and river engineers, and water resource managers.
A practical example of alternative urban stream channel designs influencing ecohydraulic conditions has been described by Anim et al. [2]. Design scenarios of rehabilitation of the channel morphology were compared against a reference "natural" scenario using ecologically relevant hydraulic metrics. The results showed that: (i) with the addition of natural oscillations to an increasing number of individual topographic variables in a degraded channel, the ecohydraulic conditions were incrementally improved and (ii) the channel reconfiguration reduced the excessive frequency of bed mobility, loss of habitat, and hydraulic diversity, particularly as more topographic variables were added.
The diversity of a physical habitat in rivers and channel connectivity is a necessary requirement for high species diversity and high biotic production [3]. River regulation aimed at flood protection primarily changes the morphology of the riverbed. The diverse natural riverbed has a small capacity; in the case of mountain streams in the Carpathian system, it ranges from Q 1 to Q 5 (1-year to 5-year floods). Most streams are regulated along with the water surface level and measured flow. The biotic area is represented by the HSCs of individual indicator species. Such modelling requires the interdisciplinary cooperation of experts in the field of hydraulics, hydrology, and ichthyology (assuming the bioindicator is a fish). This makes it time-consuming and practically inaccessible to water management practice. Therefore, it is necessary to look for new procedures that provide equivalent data to the mentioned models. Our prior research [24][25][26] suggests that the demanding hydroecological modelling could be replaced by regression equations based on the relationship between abiotic instream characteristics and habitat quality represented by fish as a bioindicator. Therefore, the basic aim of this study was to define the relationship between the morphological parameters of natural mountain and piedmont streams and the quality of aquatic habitats.

Materials and Methods
To achieve the basic aim of the study, the following methodology was chosen: • Selection of the reference stream reaches; • Topographic and hydrometric measurements of the reference reaches; • Ichthyological surveys in the reference reaches of the streams; and • Correlation and regression analyses of the instream characteristics and their influence on the quality of aquatic habitats.

Selection of Reference Stream Reaches
The degree of reliability of the evaluation of abiotic and biotic characteristics of the aquatic habitat is determined primarily by the size of the file and the homogeneity of the data. A total of 59 suitable localities were selected in mountain and piedmont streams in Slovakia ( Figure 1) for the following reasons:

•
The negative impact of stream regulation on aquatic biota has been proven predominantly in mountain and piedmont streams. • These streams are susceptible to morphological changes; therefore, the negative responses of their regulation are common in many areas, and it can be expected that the obtained results can be generalised.

•
Mountain and piedmont streams are located in the upper parts of the river basin and are relatively short; therefore, the pollution load of the stream is usually low, and the water quality does not overshadow the influence of riverbed morphology on the quality of the aquatic habitat.
Photos from the reaches taken during field surveys to illustrate the character of the reaches are given in Supplementary Figures S1-S5.

Topographic Survey and Hydrometric Measurement of the Reference Reaches
The hydraulic model was created for all reference reaches of the streams. Therefore, all reaches were surveyed geodetically and characterised by a set of cross sections. The velocity field was modelled directly in the SEFA software [23] (version 1.5, Aquatic Habitat Analysts Inc., Arcata, CA, United States). In the case of a significantly fragmented riverbed (the majority case), the hydraulics was modelled separately in the 1-D HEC-RAS model [27]. Overgrown, hard-to-reach riverbeds were surveyed by cross sections using levelling (Leica Sprinter 150M level machine (Leica Geosystems, Heerbrugg, Switzerland) with an accuracy of ±1.5 mm/km). For more available riverbeds, a Leica FlexLine TS02 total station (Leica Geosystems, Heerbrugg, Switzerland) with an angular accuracy of 3" (1 mgon) and a length accuracy of 1.5 mm + 2 ppm was used. To verify the hydraulic model, fixed geodetic points were established, which allowed us to accurately target the water-level regime at different water conditions by levelling. Measurements were performed in the summer at low discharge. water-level regime at different water conditions by levelling. Measurements were performed in the summer at low discharge.   The discharge determination was performed simultaneously with the ichthyological survey. The hydrometric measurements were performed in accordance with ISO 748 [28]. At the beginning of each measurement, suitable cross sections for hydrometry were selected. The bottom of the riverbed in the measured cross section was regular without large stones or other flow obstacles in the stream, consistent with Herschy [29]. A set of three propellers mounted on one rod was used to measure the flow velocities. All propellers were calibrated according to ISO 3455 [30].

Ichthyological Survey in the Reference Reaches of the Streams
The ichthyological survey aimed to determine the preference for the flow velocities and water depths by individual fish species. Data were obtained by electro-fishing, similar to that reported by Lamouroux et al. [31]. To gather the fish samples, an electro-fishing unit (Hans-Grassl ELT62IIHI (HANS GRASSL GmbH, Schönau am Königssee, Germany)) with the option of a continuous choice of electrical parameters was used. Microhabitat characteristics, water depth, and mean vertical velocity were recorded at the point of capture of each fish. The mean vertical velocity was derived from the measurements recorded by a system of three hydrometric propellers placed on one rod. The location of the hydrometric propellers was standard in the following multiples of water depths (d): Fish are good bioindicators that are sensitive to changes in riverbed morphology [32][33][34][35][36] and temperature [37]. For quality assessment and changes in aquatic habitat, one fish species, the adult brown trout (Salmo trutta m. fario), was selected using statistical methods. Previous studies [24] stated that trout can sensitively indicate changes in the morphology of the riverbed. The results show that there is a relationship between the morphological parameters of the watercourse and the characteristics of brown trout as a bioindicator of the quality of the aquatic habitat. Brown trout were also present in sufficient quantities in the reference reaches.
From the ichthyological survey, HSCs for individual fish species were derived for each stream. The set of all measured suitability curves of brown trout was generalised for four categories of mountain streams ( Figure 2). The generalisation procedure is described in more detail by Macura et al. [25]. The discharge determination was performed simultaneously with the ichthyological survey. The hydrometric measurements were performed in accordance with ISO 748 [28]. At the beginning of each measurement, suitable cross sections for hydrometry were selected. The bottom of the riverbed in the measured cross section was regular without large stones or other flow obstacles in the stream, consistent with Herschy [29]. A set of three propellers mounted on one rod was used to measure the flow velocities. All propellers were calibrated according to ISO 3455 [30].

Ichthyological Survey in The Reference Reaches of The Streams
The ichthyological survey aimed to determine the preference for the flow velocities and water depths by individual fish species. Data were obtained by electro-fishing, similar to that reported by Lamouroux et al. [31]. To gather the fish samples, an electro-fishing unit (Hans-Grassl ELT62IIHI (HANS GRASSL GmbH, Schönau am Königssee, Germany)) with the option of a continuous choice of electrical parameters was used. Microhabitat characteristics, water depth, and mean vertical velocity were recorded at the point of capture of each fish. The mean vertical velocity was derived from the measurements recorded by a system of three hydrometric propellers placed on one rod. The location of the hydrometric propellers was standard in the following multiples of water depths (d): 0.2 × d, 0.4 × d, and 0.8 × d.
Fish are good bioindicators that are sensitive to changes in riverbed morphology [32][33][34][35][36] and temperature [37]. For quality assessment and changes in aquatic habitat, one fish species, the adult brown trout (Salmo trutta m. fario), was selected using statistical methods. Previous studies [24] stated that trout can sensitively indicate changes in the morphology of the riverbed. The results show that there is a relationship between the morphological parameters of the watercourse and the characteristics of brown trout as a bioindicator of the quality of the aquatic habitat. Brown trout were also present in sufficient quantities in the reference reaches.
From the ichthyological survey, HSCs for individual fish species were derived for each stream. The set of all measured suitability curves of brown trout was generalised for four categories of mountain streams ( Figure 2). The generalisation procedure is described in more detail by Macura et al. [25].

Correlation and Regression Analysis of Instream Characteristics and Their Influence on the Quality of Aquatic Habitat
The main goal was to characterise the interaction of stream characteristics on the quality of the aquatic habitat of mountain and piedmont streams. Table 1 describes the basic statistics on the set of characteristics of the 59 reference reaches. Complete characteristics are given in Supplementary Table S1. The Pearson product-moment correlation coefficient (r) [38] was evaluated between these data. Table 1. Basic statistical characteristics of the data set: 364-day discharge (Q 364d ), catchment area to the last profile of the reference reach (A p ), maximum value of the water depth in the reference reach (d max ), average maximum depth from all cross sections (d am ), average longitudinal slope of the water level (i p ), average number of fish covers (microhabitats) per 100 m of reach (n 100 ), fish cover length in the flow direction (L Cov ), fish cover width (W Cov ), average value of the area-weighted suitability for the monitored fish covers (AWS Cov ), area-weighted suitability for the monitored reach (AWS Rch ), minimum values of monitored parameters (Min), variability of the interquartile range using the value of the first quartile (1QR), middle value of the range of monitored data (Median), average values of the monitored parameter (Mean), variability of the interquartile range using the value of the third quartile (3QR), and maximum values of monitored parameters (Max). M-day discharge (Q Md ) is the average daily discharge reached or exceeded by M days a year. The values of Q 364d for each reference reach were determined in cooperation with the Slovak Hydrometeorological Institute. Actual hydrometric, topographic, and ichthyological measurements were performed in the summer at low water levels ranging from Q 90d to Q 355d .
In ecohydraulics, a habitat can be numerically expressed by a special index named the area-weighted suitability (AWS) in units of m 2 ·m −1 and can be calculated as the combined habitat suitability index (CSI) weighted by the area of the water level [23]. The average value of the AWS for the monitored fish covers (AWS Cov ) was determined according to Equation (1): where A Cov = area of the fish cover (m 2 ), P d = depth suitability determined from the HSC, and L Cov is the fish cover length in the flow direction The resulting value determines the area (m 2 ) usable for the monitored bioindicator (brown trout) that falls within a microhabitat length of one meter. The AWS Cov value describes the quality of the aquatic habitat at the microhabitat level.
On the other hand, AWS Rch is the value of the area-weighted suitability for the stream within the whole reference reach. This area was determined according to Equation (2): where L Rch = reach length in the flow direction and A Rch = total area of the reference reach (m 2 ).

Habitat Quality
During low-flow periods, the ichthyofauna is concentrated in fish covers; therefore, the number, characteristics, and distribution of fish covers are decisive for evaluation of the quality of the aquatic habitat. As an example, we present the results from the reference reach of the Drietomica stream. Nineteen fish covers were identified within a length of 240 m. At the minimum flow Q = 0.03 m 3 ·s −1 , the total flooded area of the reach was 268.3 m 2 , of which the fish cover area was 86.21 m 2 (32%). The water surface area outside the fish covers was designated as free water. As the flow increased, the total area also increased; at flow Q = 0.08 m 3 ·s −1 , the percentage of the fish cover area remained at approximately 30% of the total surface area. The ichthyological survey, which was carried out at two different flow rates (Q = 0.55 m 3 ·s −1 and Q = 1.48 m 3 ·s −1 ), shows that 93% of brown trout individuals were captured in fish covers [26]. There is a logical assumption that, during periods of low flow, trout prefers fish covers and that the rest of the stream is used for migration. Therefore, it is useful to evaluate the quality of fish covers. The evaluation of the free water is only significant in terms of the possibility of migration (whether the fish are able to migrate through the free water area). Evaluation of 59 reaches shows that the average fish cover area is 36.31% of the total water surface. Figure 3 describes the development of the number of streams with the ratio of microhabitat areas to the total flooded area. The reference reaches of the five streams with low percentages of fish cover areas are predominantly riffle types. where LRch = reach length in the flow direction and ARch = total area of the reference reach (m 2 )

Habitat Quality
During low-flow periods, the ichthyofauna is concentrated in fish covers; therefore, the number, characteristics, and distribution of fish covers are decisive for evaluation of the quality of the aquatic habitat. As an example, we present the results from the reference reach of the Drietomica stream. Nineteen fish covers were identified within a length of 240 m. At the minimum flow Q = 0.03 m 3 ·s −1 , the total flooded area of the reach was 268.3 m 2 , of which the fish cover area was 86.21 m 2 (32%). The water surface area outside the fish covers was designated as free water. As the flow increased, the total area also increased; at flow Q = 0.08 m 3 ·s −1 , the percentage of the fish cover area remained at approximately 30% of the total surface area. The ichthyological survey, which was carried out at two different flow rates (Q = 0.55 m 3 ·s −1 and Q = 1.48 m 3 ·s −1 ), shows that 93% of brown trout individuals were captured in fish covers [26]. There is a logical assumption that, during periods of low flow, trout prefers fish covers and that the rest of the stream is used for migration. Therefore, it is useful to evaluate the quality of fish covers. The evaluation of the free water is only significant in terms of the possibility of migration (whether the fish are able to migrate through the free water area). Evaluation of 59 reaches shows that the average fish cover area is 36.31% of the total water surface. Figure 3 describes the development of the number of streams with the ratio of microhabitat areas to the total flooded area. The reference reaches of the five streams with low percentages of fish cover areas are predominantly riffle types.  Figure 4 shows 10 variables, of which AWSCov and AWSRch have been determined to be dependent on other parameters. On the diagonal of the image, there is a name of the variable along with a thumbnail of its histogram. Above the diagonal, there are Pearson product-moment correlation coefficients (r) [38] distinguished by colour as negative or positive, and the size of the correlation is emphasized by the font size. Below the diagonal, there are thumbnails of scatter plots with correlation lines. The corresponding correlation coefficient or scatter plot between the two variables can be found at the intersection of the horizontal and vertical from these variables from its box on the diagonal.  Figure 4 shows 10 variables, of which AWS Cov and AWS Rch have been determined to be dependent on other parameters. On the diagonal of the image, there is a name of the variable along with a thumbnail of its histogram. Above the diagonal, there are Pearson product-moment correlation coefficients (r) [38] distinguished by colour as negative or positive, and the size of the correlation is emphasized by the font size. Below the diagonal, there are thumbnails of scatter plots with correlation lines. The corresponding correlation coefficient or scatter plot between the two variables can be found at the intersection of the horizontal and vertical from these variables from its box on the diagonal. Figure 4 shows that AWS Cov has the closest dependence on average width of the fish cover (W Cov ) and average maximum depth (d am ). The most significant dependencies are shown in Figure 5 in more detail. Figure 4 shows that AWSCov has the closest dependence on average width of the fish cover (WCov) and average maximum depth (dam). The most significant dependencies are shown in Figure 5 in more detail.    Table 1. Each scale corresponds to a parameter range from Table 1.  Table 2 describes the Pearson product-moment correlation coefficient (r) [38] between AWSCov and the individual stream characteristics, together with the p-value for the level of 5% and the upper and lower 95% confidence intervals of the value of r. The p-  Table 2 describes the Pearson product-moment correlation coefficient (r) [38] between AWS Cov and the individual stream characteristics, together with the p-value for the level of 5% and the upper and lower 95% confidence intervals of the value of r. The p-value is the probability of obtaining test results at least as extreme as the observed results. The table shows that the closest relationships are between AWS Cov , and W Cov and d am , respectively. Analysis of the relationship between AWS Cov and d am shows that water depth has a substantial influence on the development of the aquatic habitat quality. This means that conditions for rheophilic species (brown trout and similar) improve as discharge increases [39]. This fact can be used in habitat modelling using IFIM, which mainly uses depth HSCs [40].

Correlation and Regression Analyses of Instream Characteristics and Their Influence on the Quality of Aquatic Habitat
For larger streams, the water depth is favourable for brown trout, even at minimum flows. For small streams, where minimum flows do not create enough water depth for brown trout, microhabitats have unsuitable conditions. Therefore, an analysis of the influence of the stream size on the ratio of AWS Cov and AWS Rch to the total flooded surface area was performed. The reaches were sorted in ascending order by d am . From Figure 6, it follows that, for torrent streams (d am from 0.08 to 0.20 m), there is a significant increasing trend of AWS Rch and AWS Cov . In the d am range from 0.21 to 0.55 m (Figure 7), we can state that the water depth does not affect the ratio of AWS to the total flooded area. AWS Cov increases slightly, while AWS Rch slightly falls.

Discussion
The results of the correlation analyses in Tables 1 and 2 show the mutual relationship between AWS Cov and W Cov , d max , Q 364d , A p , and d am . A correlation was not confirmed between the parameters AWS Cov and i p , n 100 , and L Cov . The correlation analysis of the AWS Rch variable was also performed; however, the resulting correlation values were lower than those for AWS Cov . This result implies that the quality of the aquatic habitat for brown trout is primarily determined by the cover possibilities of the stream. The rest of the stream has little effect on trout when this part of the stream does not create migration barriers. It can be stated that the impact of technical interventions, such as river regulation, can be determined according to the water depth conditions, based on stream covers with the appropriate water depth and flow velocity [26]. A bioindicator that is sensitive to morphological changes, such as trout, must be selected. Ecohydraulic research has confirmed the dominant effect of flow velocity and water depth on fish habitats [41]. This fact is also reflected in the modelling of aquatic habitat quality using the IFIM methodology. Studies have confirmed a correlation between the characteristics of the HSCs and hydraulic parameters, especially the depth and velocity [25,42]. River regulation also changes the velocity field of the stream, which has a minor effect on habitat quality during periods of minimum flow. This was also confirmed by the results of our previous research, where the optimal ratio of the influence of velocity and depth, expressed in the form of HSCs, was 2:8 [43].
Analysis of the influence of the stream size on the ratio of AWS Cov and AWS Rch to the total flooded surface area demonstrated that, in torrent streams, there was a gradual increase in the AWS ratio to the total surface area (streams up to d am = 0.20 m). The piedmont streams were not sensitive to changes in water depth, even at greater depths, and there was a reduction in the AWS share. The reason for this development may be that, in shallow streams, the area of optimal habitats (with sufficient water depth) is low. By increasing the discharge, the habitat area is higher (approximately 40% or more of the entire surface area). It can be stated that, the larger the stream, the less sensitive it is to changes in depth and that there are minimal changes in the area of suitable habitats with changes in depth. In deeper streams (in our case, 0.4-0.5 m), there may be a decrease in the AWS with increasing water depth. This trend is likely to be affected by a change in the velocity field. This means that there are also higher flow velocities at greater depths, which can have a negative effect on ichthyofauna habitat preferences. In other words, changes in the morphology of the stream bed, primarily the depth conditions induced by river regulation, also affect the velocity field of the stream. Water quality does not radically change with river regulation. It is logical that the aquatic habitat quality not only of the regulated stream but also of the restored stream should be evaluated particularly based on stream morphology and bioindicator, in this case, ichthyofauna. Such an evaluation is provided by models based on the IFIM methodology, such as SEFA. Modelling a larger set of streams using these models would be extremely challenging; in the case of classification of water bodies into five classes, as required by the Water Framework Directive [22], it would be nearly impossible. From the above statistical analysis, in order to interpret the relationships between the dependent variable AWS Rch and other variables, it follows that 5 variables that result from the stream characteristics are important. The characteristics are Q 364d (m 3 ·s −1 ), A p (km 2 ), d max (m), d am (m), and W Cov (m). Figure 8 shows the relative importance of these variables, as determined by the method published in [44]. These results are the basis for the creation of regression equations which can be used to determine the dependent variable of AWSRch. In practice, this will mean that, instead of a detailed topographic survey, hydraulic modelling, and ichthyological research, AWSRch will be set based on average values of Ap, dmax, dam, and WCov, which can be determined during field reconnaissance of the stream and the discharge to be hydrometrically measured. Of course, generalized HSCs of the representative species must be available. Such a procedure can significantly simplify evaluation of the quality of aquatic habitat, making it much more accessible for design practice.
The database for deriving regression equations was divided into calibration data (59 reaches representing the data evaluated by the authors for 22 years of field measurements) and validation data (for which other measurements are currently being performed). The regression results of the calibration file show a high value of the coefficient of determination (r 2 = 0.8452). This determines the ratio of the common variance of the independent and dependent variables, i.e., how much the change in independent variables affects the AWSRch dependent variable. These results also suggest that determination of AWSRch by regression equations is promising. These results are the basis for the creation of regression equations which can be used to determine the dependent variable of AWS Rch . In practice, this will mean that, instead of a detailed topographic survey, hydraulic modelling, and ichthyological research, AWS Rch will be set based on average values of A p , d max , d am , and W Cov , which can be determined during field reconnaissance of the stream and the discharge to be hydrometrically measured. Of course, generalized HSCs of the representative species must be available. Such a procedure can significantly simplify evaluation of the quality of aquatic habitat, making it much more accessible for design practice.
The database for deriving regression equations was divided into calibration data (59 reaches representing the data evaluated by the authors for 22 years of field measurements) and validation data (for which other measurements are currently being performed). The regression results of the calibration file show a high value of the coefficient of determi-nation (r 2 = 0.8452). This determines the ratio of the common variance of the independent and dependent variables, i.e., how much the change in independent variables affects the AWS Rch dependent variable. These results also suggest that determination of AWS Rch by regression equations is promising.

Conclusions
The results of this research in the reference reaches of mountain and piedmont streams in Slovakia indicates that the quality of instream habitats can be characterized by the relationship between the morphological characteristics of the stream and the biotic characteristics represented by the AWS. The results of the correlation analysis demonstrate the influence of individual parameters on the instream biota. Furthermore, brown trout respond sensitively to abiotic parameters, such as riverbed morphology, flow velocity, and water depth. Based on the obtained results, it is possible to determine the characteristics of the stream which have a substantial effect on the instream habitat quality. Specifically, based on the morphological changes, which are represented by habitat suitability curves for water depth, it is possible to evaluate the impact of river regulation or to predict the effect of stream restoration.

Supplementary Materials:
The following are available online at https://www.mdpi.com/2073-444 1/13/2/142/s1, Table S1: Basic characteristics of the data set, Figure S1-S5: Photos from the reaches taken during field surveys to illustrate the character of the reaches.  Institutional Review Board Statement: Following the legislation of Slovak Republic, the liability to notify the Slovak Fishery Union (SFU) as the responsible organization was fulfilled one week prior to each sampling. A representative of SFU was present at each sampling to supervise the correctness of the methods used. Each sampling was performed by a person with special permit for electrofishing issued by the Ministry of Environment of the Slovak Republic. All the fish samples collected by electrofishing were immediately counted, weighted, measured, and returned back to original locality in the stream. All sampling procedures used in this study strictly followed the ethical standards and animal welfare declared during the process of obtaining the field permit from the Ministry of Environment of the Slovak Republic. Authors declare that all customary standards concerning handling the live material applicable in the EU were complied.

Data Availability Statement:
The data presented in this study are available within the article and Supplementary Material.