High Diversity of Waters, Communities and Stressors—Design of Croatian Fish Index for Rivers (CFIR)

Abstract

Determination of the ecological state of any water body is crucial for its adequate conservation and is, moreover, required by the Water Framework Directive. Employment of multimetric indices, which integrate various anthropogenic threats to water systems with the responses of different elements of fish communities, improve our understanding and allow us to monitor the ecological state of freshwater bodies. However, design of multimetric indices that describe the ecological state of water bodies based on fish communities has already proved difficult for the Mediterranean region for several reasons, including the frequent presence of species-poor and ecologically tolerant fish communities, high endemicity and a serious lack of localities with undisturbed fish communities. The development of an ecological state index for natural water bodies in Croatia based on fish as a biological element which we present in this paper was also hindered by similar obstacles, particularly due to exceptionally high endemicity present in rivers of the Adriatic watershed, the great number of distinct water types, and significant anthropogenic pressures. Nevertheless, based on comprehensive sampling of fish communities throughout Croatia and following appropriate statistical procedures, we were able to identify stressors acting on certain water bodies, as well as responses of fish communities to them, and, finally, describe the ecological state of natural water bodies throughout Croatia. We also propose measures that will most likely help in achieving improvement and/or maintenance of the ecological states of these water bodies.

1. Introduction

The Water Framework Directive (WFD; Directive 2000/60/EC of the European Parliament and of the Council of 23^rd October 2000 establishing a framework for Community action in the field of water policy), probably the most important tool aimed at preserving European inland, transitional, coastal and ground waters and the biodiversity within them, posed new, important tasks for European Union members in achieving WFD objectives. Those tasks have turned out to be very challenging due to several reasons: high diversity and great number of water types and fish communities harbored within them; significant anthropogenic pressures acting on almost all rivers and streams; high endangerment status of many fish species, but also restricted possibilities of conservational efforts due to high endemicity and small population sizes of many species. According to the WFD, member states shall protect, enhance and restore all bodies of surface waters, as well as all artificial and heavily modified water bodies. Furthermore, member states are required to establish programs for the monitoring of the ecological status/potential of water bodies. Following the WFD objectives, member states have cleaved to an important task of design of the ecological state/potential classification systems. Although the WFD has systematized and streamlined efforts to achieve good ecological conditions in natural water bodies, the first monitoring efforts were begun in the early 20th Century [1]. With the growth of knowledge about the state of biological communities and their responses to anthropogenically induced pressures, and particularly with the WFD implementation, design of biological indices has significantly improved over the last century, together with techniques of environmental assessment [2,3]. Determination of the ecological state of certain water bodies based on biological elements is fundamental for conservation of European freshwaters, as requested by the WFD [4]. Employment of multimetric indices, instead of the older attempts at classification of water quality based on a single metric, usually anthropogenic pressure, has yielded great advances in describing and monitoring the ecological state of freshwater bodies. Multimetric indices integrate different stressors, as well as various components of the community. Furthermore, they can be adapted to the specific conditions of a river type in a way that takes into account the most relevant stressors and specific characteristics of the biological element. Moreover, they are easy to interpret [5]. The multimetric index concept was first developed as the index of biotic integrity (IBI), first used in small warmwater streams in the USA for estimation of the status of fish communities, as an indicator of the health of water bodies [6]. The original version comprised 12 metrics describing species richness, abundance and composition, number and abundance of indicator species, trophic functions, reproductive behavior and condition of fish [6]. The IBI has become very popular and widely applied in the USA, Canada, Australia, South Africa and throughout Europe for assessment of the ecological state of various water types (summarized in [7]). Consequently, multimetric indices have been applied in many countries to achieve the WFD objectives. Furthermore, European-scale intercalibration has been conducted for fish indices for various water types. Ecological indices have turned out to be highly useful for communication among scientists, stakeholders and public [8], since they are numerical values representing modification of fish communities as a response to anthropogenic threats and, thereby, indicating the state of the water ecosystem and quality of water [5,6,8]. Unfortunately, modification of the natural hydrological parameters, excessive use of pesticides and fertilizers, river canalization, deterioration of natural floodplains, introductions of non-native fish species, etc., have caused significant changes in natural fish communities [9,10] and observed reductions of the ecological state of many waters compared to natural conditions. Precise, sound and objective description of the ecological state of any water body based on biological communities living within it is the most important prerequisite for its adequate conservation.

Following the WFD, two animal groups are used to monitor the ecological state of water bodies; benthic invertebrates [11,12] and fish [13,14]. Besides these animal taxa, macrophytes are also used as a biological quality element [15,16]. Biological elements in use for ecological state estimation must fulfil several requirements: broad ecological tolerance, quick response to environmental changes, well-resolved taxonomy, well-described distribution and ecological traits for most species, and predictable responses to stress [4]. In terms of the abovementioned requirements, fish turn out to be excellent bioindicators which also have high conservational value, are highly visible and valuable to the wider public, and are able to integrate information from lower trophic levels in addition to the chemical and physical histories of water [4,7]. Furthermore, fish possess additional attributes that make them very useful in the assessments of ecological status: they generally remain in the same area throughout the year, communities are persistent, they have large ranges and are less affected by natural microhabitat differences than smaller organisms, and most have relatively long life spans which enable them to reflect current as well as long-term water quality [7].

Based on the WFD requirements, biological indices have to rely on clear pressure–response relationships between anthropogenic pressures and changes in the status of populations of biological elements. Furthermore, they should indicate the ecological state of any water body expressed as its deviation from the reference conditions [5,8]. To assess the ecological status of a water body, it is important to take into consideration the taxonomic composition, abundance, and ratio of taxa sensitive to disturbances, as well as biological indicators, and compare those values to reference conditions [5]. Moreover, within a multimetric index, each metric is correlated with specific anthropogenically induced pressure [5].

Due to its exceptionally rich ichthyodiversity, high number of fish species, and particularly high number of endemic species harbored on a relatively small area, Croatia can be considered as one of the ichthyological ‘hot spots’ of Europe. In particular, waters of the Dinaric karst system in Croatia and neighboring countries have already been designated as European regions where an extraordinarily high number of fish species and a great proportion of endemics can be found in a relatively small area [17,18,19]. Reasons for such a high ichthyodiversity are various, and its impacts on the ecological state estimation are complex. First, division of Croatian waters into two watersheds with distinct hydrological, physico-chemical and ecological conditions provided a basis for development of diverse ichthyofauna. Rivers of the Black Sea watershed (the Danube River basin) are mostly long, with many tributaries which are all interconnected (with the exception of several small streams in the Gorski Kotar and Lika region). On the other hand, the Adriatic basin is characterized by shorter, isolated rivers, poor in tributaries but with rich and complex underground water networks. However, this distinction of Croatian waters into two watersheds, and the related speciation provoked by the isolation of the Adriatic rivers, cannot fully account for such an exceptional ichthyodiversity. Yet, the final explanation is hidden in the past, in the very complex geological history of the area, yielding the biodiversity at ecosystem, species and genetic levels that we recognize today. Although great parts of the geological history of the Dinaric karst system and surrounding area have not been described yet, or only local and scarce data exist, previous investigations [18,19,20,21,22,23,24,25,26] indicated several geological events which were particularly important for shaping fish diversity: upwelling of the Dinarid mountains, strong tectonic activities in the Dinaric karst system, evolution of the middle Paratethys basin, evolution of the Dinaric Lake System, the Messinian Salinity Crisis (which probably had significantly lower impact than previously proposed) and glaciation events [27,28,29].

Difficulties in preparing and implementing multimetric indices for ecological state estimation of freshwater bodies have already been described for Mediterranean streams in several previous investigations. Freshwater systems in the Mediterranean region are considered unpredictable, which has led to mostly species-poor and ecologically tolerant fish populations [30,31,32]. As a result, it is hard to select typical metrics of fish populations and it is extremely difficult to find undisturbed sites with fish assemblages in a natural condition [4]. It has also been noted previously that in areas with high endemicity, such as the Mediterranean region, biogeographical differences among ecoregions present an obstacle to development of broadly applicable bioassessment indices [33]. Rivers of the Mediterranean area usually have highly diverse fish assemblages with great portion of endemic species [34,35], but with strong anthropogenic pressures acting on them [36]. Furthermore, scarce knowledge of the ecological traits of many endemic species is also a problem, e.g., [22,37]. It is no surprise, therefore, that fish were previously only rarely used in water quality assessments in the Mediterranean, particularly the Eastern Mediterranean, and that freshwater systems in the Mediterranean basin were less studied than in the remaining European watersheds [33]. Using fish communities for estimating the ecological state of freshwaters started in 2009 in Greece [38]. Similar problems (high number of fish species with great portion of endemic species, lack of knowledge about certain fish species, particularly endemic ones, absence of reference condition sites, as well as high diversity of freshwater types) also complicated development of a fish index for Turkish freshwaters [39].

In this paper, we present the developmental procedure of the ecological state indices for natural water bodies in Croatia based on fish as a biological element, identify stressors acting on certain water bodies, as well as responses of fish communities to them, and, finally, describe the estimated ecological state of natural water bodies throughout Croatia and propose measures that will most likely help in achieving improvement and/or maintenance of their ecological states.

2. Materials and Methods

2.1. Typology and the Most Important Characteristics of Croatian Natural Running Waters

Croatia, besides being rich in freshwater systems and harboring diverse fish communities, also has a complex national typology of natural water bodies. Croatian rivers and streams belong to two different watersheds—the Black Sea (Danube basin) watershed comprises waters from northern and central part of Croatia, whereas southern river basins belong to the Adriatic watershed. The Croatian national typology of natural rivers divides Croatian rivers into three ecoregions: rivers of the Black Sea watershed mostly belong to the Pannonian ecoregion (national types HR-R_1, HR-R_2, HR-R_3, HR-3_4 and HR-R_5, with subtypes), whereas rivers of the Adriatic watershed mostly belong to the Dinaric coastal ecoregion (national types HR-R_11, HR-R_12, HR-R_13, HR-R_14, HR-R_15, HR-R_16, HR-R_17, HR-R_18, HR-R_19, some of which comprise subtypes). The Pannonian ecoregion comprises rivers draining into the Black Sea watershed, so it falls well under the Danube intercalibration (IC) CROSS GIG group, whereas rivers of the Dinaric coastal ecoregion, draining exclusively into the Adriatic basin, belong to the Mediterranean IC group. The third ecoregion based on the national typology, the Dinaric continental ecoregion (national types HR-R_6, HR-R_7, HR-R_8, HR-R_9 and HR-R_10, some with subtypes), comprises rivers from the central, mostly mountainous part of Croatia that belong to both watersheds (Black Sea and Adriatic). Note that the described national typology is at the same time the European typology for Croatian running waters. Based on the geographical location and ecological characters of those river types, they are included into the Danube group for intercalibration of national classification systems for ecological state estimation. The Pannonian ecoregion comprises small mountain, mid-altitude and lowland rivers, medium, large and alluvial lowland rivers, as well as large rivers (not included in the classification systems presented in this paper). In the Dinaric continental region, rivers and streams belong to small, medium and large mountain and mid-altitude rivers, medium and large lowland rivers, rivers in karstic fields and temporary rivers. The Dinaric coastal ecoregion contains lowland and mid-altitude small rivers, lowland rivers with slopes greater than 5‰, mid-altitude and lowland rivers, rivers in karstic fields, and temporary rivers, as well as lowland and mid-altitude small, medium and temporary rivers in Istria.

In the forthcoming text, we present our developmental procedure and the results of the fish-based index for ecological state estimation for all abovementioned national types (Supplementary materials, Table S1); the exception is HR-R_5 which concerns very large rivers, and its ecological state classification system is under development separately.

2.2. Fish Fauna Sampling

Fish sampling was conducted in accordance with HRN EN 14962:2007, ‘Water quality—Guidance on the scope and selection of fish sampling methods’, and ERN EN 14011:2003 ‘Water quality—Sampling of fish with electricity’ (EN 14011:2003). Sampling was conducted in the period from spring to early autumn (from April to September) in the period 2016–2019. Electrofishing was chosen as the preferred sampling method, as it represents a standard method for fish sampling in rivers. It enables reliable estimation of population density, species abundance, fish numbers and biomass, age structure and mutual relationships of fish at the chosen locality, and it also represents the least harmful method for the fish population.

Sampling locations were distributed throughout Croatia (Table S1, Figure 1) on natural streams and rivers. Sampling sites (localities) were identified so that the sample covered the diversity of all natural and man-induced microhabitats. The localities were large enough to include living areas of dominant species and all characteristic river habitats (e.g., faster and slower parts, sidearms, oxbows, etc.). During the sampling campaign at each locality, we made sure to cover microhabitat diversity by sampling at adequate river stretches: 1 km for small running waters (catchment area size less than 100 km²), 5 km for medium-sized running waters (catchment area size 100–1000 km²), and 10 km for large running waters (catchment area size greater than 1000 km²). Depending on the size of the river, as well as microhabitat diversity, the electric fishing generator was used to catch fish in three ways: wading in the river, electrofishing from the river bank and electrofishing from a boat. Wadable running waters (shallow watercourses up to 15 m width) were sampled in their entire width using a backpack electrofishing device, while block nets were cast along the borders of the sampling stretch, in order to prevent fish from escaping. In larger running waters with depths greater than 0.7 m and/or habitat diversity preventing efficient sampling from the bank or by wading in the riverbed, a special electric fishing boat was used and electric generators with different power ratings (2.5, 5 and 7.5 kW for small, medium and large running waters, respectively). Also depending on the size of running waters, sampling was either conducted with a handheld anode or with a boom anode. Fishing was done downstream with the boat moving along the bank, covering as many habitats as possible, especially places where fish might be hiding.

All caught fish were identified based on morphological characteristics using determination keys when necessary [40,41,42]. Since sampling was conducted by experts in ichthyology, determination was conducted in the field. Total body length (TL) of all individuals was measured from the beginning of the head to the tip of the caudal fin using an ichthyometer. During measurements, eventual visual body anomalies (e.g., skin damage, subcutaneous or fin damage, parasites, deformations, tumors, lesions) were noted. Handling of the fish was conducted quickly and carefully, by expert ichthyologists, and individual fish were soon returned to the water at their sampling localities. Mortality of fish during sampling and fish handling was less than 1%.

2.3. Obtaining of the Environmental Data and Pressure Proxies

A total of 21 parameters describing habitat conditions and anthropogenic pressures were assessed, including hydromorphological, morphological and physico-chemical parameters. Environmental parameters describing each sampling site were collected directly in the field, in the same time period as fish sampling (April–September) by an authorized agency using standardized procedures (Croatian Waters). The following parameters were measured: alkalinity, conductivity, pH, transparency, temperature, concentrations of ammonia, organic carbon, molecular ammonium, nitrates, nitrites, nitrogen, phosphorous, total organic carbon, dissolved organic carbon, dissolved orthophosphates, dissolved oxygen, oxygen saturation, and biological and chemical oxygen consumption. Average values of all physico-chemical parameters considering the warmer part of the year (April–September) were included in further analyses. Furthermore, representation of unnatural, modified shores (NNLC) was estimated using ArcGIS 10.2, as a percentage of modified shores, and hydrological components (hydrological regime, longitudinal continuity and morphological conditions) were also assessed.

2.4. Fish Fauna Metrics Used to Describe Fish Communities in Croatian Rivers

All sampled fish species were classified in groups according to their preferences for reproductive substrate (lithophilic, LITH; phytophilic, PHYT; phyto-lithophilic, PHLI; pelagophilic, PEL; psammophilic, PSAM; ostracophilic, OSTR; species that spawn in the sea, SEA), feeding preferences (herbivores, HERB; insectivores, INV; omnivores, OMNI; piscivores, PISC; and detritivores, DETR) and habitat preferences (benthopelagic, WCOL and benthic, BENT). After field investigation, identification and measurement of all individuals, we have prepared a total of 103 metrics that describe fish communities (

Table 1. Overview of fish community metrics included in the analyses (abbreviations are in brackets).

Composition/ Abundance Metrics	Richness/ Diversity Metrics	Sensitivity/ Tolerance Metrics	Functional Metrics	Other Metrics
Proportion of native species (pSn) Proportion of non-native species (pSa) Proportion of lithophilic species (pLITH) Proportion of phytophilic species (pPHYT) Proportion of phyto-lithophilic species (pPHLI) Proportion of pelagophilic species (pPEL) Proportion of psammophilic species (pPSAM) Proportion of ostracophilic species (pOSTR) Proportion of species spawning in the sea (pSEA) Proportion of herbivorous species (pHERB) Proportion of invertivorous species (pINV) Proportion of omnivorous species p(OMNI) Proportion of piscivorous species (pPISC) Proportion of detritivorous species (pDETR) pPISC/pINV Proportion of benthopelagic species (pWCOL) Proportion of benthic species (pBENT)	Total number of species (S) Number of native species (Sn) Number of non-native species (Sa) Proportion of Salmoniform species (pSALM) Proportion of Cypriniform species (pCYPR) pSALM/pCYPR pPERC (proportion of Perciform species)/pCYPR Shannon index (H) Reciprocal Simpson index (1/S) Margalef index (Ml) Alpha index (A) Berger-Parker index (d) Shannon index based on native species (Hnat) Reciprocal Simpson index for native species (1/S) Margalef index for native species (Mlnat) Alpha index for native species (Anat) Berger–Parker index for native species (dnat) Hnat-H (Hdif) 1/Snat-1/S (1/Sdif) Mlnat-Ml (Mldif) Anat-A (Adif) dnat-d (ddif) Hnat/H (Hrat) 1/Snat/1/S (1/Srat) Mlnat/Ml (Mlrat) Anat/A (Arat) dnat/d (drat)	Proportion of native individuals (uSn) Proportion of non-native individuals (uSa) Proportion of lithophilic individuals (uLITH) Proportion of phytophilic individuals (uPHYT) Proportion of phyto-lithophilic individuals (uPHLI) Proportion of pelagophilic individuals (uPEL) Proportion of psammophilic individuals (uPSAM) Proportion of ostracophilic individuals (pOSTR) Proportion of individuals spawning in the sea (uSEA) Proportion of herbivorous individuals (uHERB) Proportion of invertivorous individuals (uINV) Proportion of omnivorous individuals (uOMNI) Proportion of piscivorous individuals (uPISC) Proportion of detritivorous individuals (uDETR) uPISC/uINV Proportion of benthopelagic individuals (uWCOL) Proportion of benthic individuals (uBENT) Proportion of Salmoniform individuals (uSALM) Proportion of Cypriniform individuals (uCYPR) uSALM/uCYPR uPERC (proportion of Perciform individuals)/uCYPR Proportion of native individuals’ biomass (bnat) Proportion of non-native individuals’ biomass (balo)	Number of lithophilic species (LITH) Number of phytophilic species (PHYT) Number of phyto-lithophilic species (PHLI) Number of pelagophilic species (PEL) Number of psammophilic species (PSAM) Number of ostracophilic species (OSTR) Number of species spawning in the sea (SEA) Number of herbivorous species (HERB) Number of invertivorous species (INV) Number of omnivorous species (OMNI) Number of piscivorous species (PISC) Number of detritivorous species (DETR) Number of benthopelagic species (WCOL) Number of benthic species (BENT) Proportion of phytophilic species biomass (bPHYT) Proportion of phyto-lithophilic species biomass (bPHLI) Proportion of pelagophilic species biomass (bPEL) Proportion of psammophilic species biomass (bPSAM) Proportion of ostracophilic species biomass (bOSTR) Proportion of biomass of species spawning in the sea (bSEA) Proportion of herbivorous species biomass (bHERB) Proportion of invertivorous species biomass (bINV) Proportion of omnivorous species biomass (bOMNI) Proportion of piscivorous species biomass (bPISC) Proportion of detritivorous species biomass (bDETR) bPISC/bINV Proportion of benthopelagic species biomass (bWCOL) Proportion of benthic species biomass (bBENT) Proportion of Salmoniform species biomass (bSALM) Proportion of Cypriniform species biomass (bCYPR) bSALM/bCYPR	Total biomass (B) Biomass of native individuals (Bnat) Biomass of non-native individuals (Balo) Total length of the most abundant species based on the number of individuals (TLmaxn) Total length of the most abundant species based on the biomass (TLmaxb)

). Metrics belonging to four metric types have been prepared (following [43]), but also several additional metrics which consider specificities of the Croatian watersheds. Note that collocation of certain fish metrics under metric types (as defined by [43]) is somewhat arbitrary, because the same metric can sometimes be designated to more than one type. For example, proportion of individuals and biomass of species belonging to a certain feeding or habitat preference type can be addressed as functional metrics, because they correspond with ecological functions of taxa, but also as sensitivity/tolerance metrics, since they will change in response to certain stressors. All metric types are well-represented in the metrics that describe fish communities of Croatian flowing waters. The pertinence of the fish species inhabiting Croatian freshwaters to the groups described is presented in Table S2.

It is important to notice that many of the calculated metrics are intercorrelated and that they describe the same or similar parts of fish communities. However, since it was not possible to foresee which of the intercorrelated metrics would be the most sensitive to certain anthropogenic pressures, all of them were calculated in the first step of our analyses, but no intercorrelated metrics were eventually included in the ecological state indices, as explained in the forthcoming paragraphs.

2.5. Statistical Analyses for Metric Selection

As described above, two sets of parameters were prepared (one describing fish communities and the second one concerning environmental parameters and pressure proxies). Metrics in both sets were subjected to similar statistical procedures in order to choose the ones that are not intercorrelated, that have normal distribution and for which a clear pressure–response relationship can be confirmed. Parameters in both data sets were first standardized. Log-transformation was used for standardization of count measures, and the logistic model was used for standardization of proportions, whereas diversity indices and metrics derived from them were considered as already standardized measures. After standardization, Pearson’s correlation coefficient was calculated for all the metrics inside each data set, and in cases where the coefficient was higher than 0.7, one or more metrics were excluded and the one with better ecological interpretation was retained for further analyses. In cases where it was not clear which metric has stronger ecological interpretation, both were included in the next step and the one with no or lower pressure–response relationship was excluded later.

Responses of fish fauna metrics to the environmental parameters and pressure proxies were analyzed by stepwise linear regression. Metrics that were significantly correlated with at least one pressure (R² > 0.4 and significance level, p < 0.05) were checked for compliance with linear regression assumptions (normal distribution, linearity and absence of multicollinearity). In rare cases when no significant response had R² > 0.4, responses with R² > 0.2 and statistical significance were taken into consideration. If both conditions were met (significant correlation with at least one pressure and linear assumptions), those metrics were considered for index development. Again, correlation coefficients were calculated among metrics of both data sets and, finally, in cases of significant correlations, metrics with better pressure–response relationships were included in the index calculation.

The statistical procedure described above was conducted for each river type. However, sometimes it was reasonable (due to similarity of fish communities in different river types and/or low number of localities inside certain river types) to combine two or more river types together and calculate pressure–response correlations for groups of types, instead of for a single type. If such combinations resulted in better pressure–response relationships than single river types, the same index of ecological state was applied to all river types within a group.

Statistical analyses were performed using Microsoft Excel and Statistica 14.0.0.15 software.

2.6. Upper and Lower Anchors for Ecological Quality Ratios

Reference conditions were determined for each river type/group of types. Since some river types comprised only a small number of locations which are influenced by various anthropogenic pressures, it was impossible to identify a satisfactory number of watercourses in which the fish community is in the optimal condition in order to determine the reference sites for all metrics of the fish communities included in the index calculation, and for all national river types. Moreover, there is a lack of estimated reference conditions for some of the physico-chemical parameters, preventing extrapolation of reference conditions of fish metrics in cases where they show significant response to certain environmental parameters. Therefore, for determining the Ecological Quality Ratios (EQRs), we applied the approach proposed by [43] for cases where it is not possible to determine reference locations—that is, we set the upper and lower anchors for individual metrics of fish communities (which had shown pressure response and, therefore, were chosen for inclusion in the index). The upper anchor was defined as a value observed or proposed for a natural, stable community (reference condition for a certain river type), whereas the lower anchor is considered to be the worst possible value obtained for severely altered communities or, in case of the absence of bad conditions in watersheds of a certain type, as the expert assessment of the worst possible conditions (e.g., for metrics considering the proportion of non-native species, the worst possible situation is that only non-native species have been recorded in a locality).

Localities with undisturbed or almost undisturbed fish communities exist inside some national river types, and they present national reference conditions for those river types. In those communities, chosen fish metrics have the best possible or almost the best possible values, yielding Ecological Quality Ratios (as described later) of 1 or nearly 1. Note that reference communities, in cases where they could be identified, are referent considering their structure. That is, they exhibit reference values of those fish community metrics that were, following the described procedure, chosen for inclusion in the Croatian fish index for rivers. Understandably, this does not mean that all fish communities belonging to a certain national river type should have, under natural type-specific conditions, the same composition or even comprise the same number of species. On the contrary, fish communities belonging to a certain river type are very diverse concerning species composition and richness, which is a natural consequence of ecological and evolutionary differences.

2.7. Calculation of Ecological Quality Ratios

Ecological quality ratios (EQRs) were calculated separately for all chosen metrics of fish communities inside each type/group of types. For that purpose, the formula used in [43] was adopted:

EQRmetric = (Metric value − Lower limit)/(Upper limit − Lower limit), for metrics decreasing with increasing intensity of a pressure

or

EQRmetric = 1 − (Metric value − Lower limit)/(Upper limit − Lower limit), for metrics increasing with increasing intensity of a pressure.

As already explained, upper and lower anchors for each metric were determined as the best or the worst possible conditions, either observed based on reference localities, or estimated as reference values.

2.8. Generation of the Croatian Index for Natural Running Waters

After identification of metrics suitable for inclusion in the multimetric index and calculation of EQRs, the Croatian fish index for rivers (CFIR) was calculated separately for each river type/group of types, as the sum of EQRs determined for a certain river type/group of types divided by the number of ecological quality ratios included.

CFIR = (EQR1+EQR2+⋯+EQRn)/n

2.9. National Boundary Setting

Ecological status classes were defined by dividing the EQR values into a five-class equidistant scale according to the prescriptions of the WFD. Specifically, CFIR expresses the relationships of the observed metrics in certain fish communities with the same metrics under reference conditions, and the ecological status of a certain locality can be designated as belonging to one out of five classes. CFIR values close to 1 imply undisturbed or only slightly disturbed fish communities, whereas values closer to 0 represent communities that are significantly modified as a consequence of anthropogenic pressures. The EQRs, and therefore also CFIR, were developed in comparison with reference conditions, so class boundaries were arranged equidistantly from 1 (reference conditions) to 0 for five ordinal rating categories for assessment of impairment in accordance with WFD (0.8–1, high/very good Ecological Quality State (EQS); 0.6–0.79, good EQS; 0.4–0.59, moderate EQS; 0.21–0.39, poor EQS; 0.0–0.2, bad EQS).

Finally, it is important to mention that intercalibration of the developed indices at the European level has been completed, and no changes in moderate/good nor good/very good borderlines were required, so here we present the finalized classification system.

3. Results

3.1. Description of the Pressure–Response Relationships

Clear pressure–response relationships have been established for all investigated river types or combinations of river types (

Table 2. Fish fauna metrics that showed significant responses to certain pressures in a certain river type/group of types and that also show normal distribution and satisfy presumptions of linearity.

Ecoregion	River Types	Description of the River Types	Pressure Response	R2	p
Pannonian	HR-R_1, HR-R_2	Small mountain, mid-altitude and lowland rivers	Number of native species (Sn) shows response to phosphorous concentration	0.250	0.00024
			Proportion of psammophilic species (pPSAM) shows response to dissolved oxygen concentration (O2)	0.264	0.0001
			Difference between Shannon indices based on native species and the whole community (Hdif) shows response to dissolved nitrogen concentration (N)	0.220	0.00051
	HR-R_3, HR-R_4A	Medium, large and alluvial lowland rivers	Number of non-native species (Sa) shows response to water temperature (temp)	0.545	0.00000
			Ratio between Shannon indices based on the native species and the whole community (Hrat) shows response to dissolved oxygen concentration (O2)	0.256	0.00098
Dinaric continental	HR-R_6, HR-R_7, HR-R_8	Small, medium and large mountain and mid-altitude rivers, as well as medium and large lowland rivers	Proportion of invertivorous individuals (uINV) shows response to dissolved ammonia	0.530	0.00011
			Proportion of omnivorous individuals (uOMNI) shows response to dissolved ammonia	0.503	0.00019
			Proportion of benthopelagic species (pWCOL) shows response to suspended particle concentration	0.361	0.00237
	HR-R_9, HR_R_10	Rivers in karstic fields and temporary rivers	Proportion of non-native species (pSa) shows response to water temperature	0.584	0.04737
			Proportion of native species’ individuals (uSn) shows response to nitrates concentration in water	0.614	0.04033
			Proportion of piscivorous individuals (uPISC) shows response to dissolved oxygen concentration (O2)	0.593	0.04515
Dinaric coastal	HR-R_11, HR-R_14	Lowland and mid-altitude small rivers and lowland short rivers with >5‰ slope	Proportion of omnivorous species (pOMNI) shows response to dissolved oxygen concentration (O2)	0.620	0.01220
			Ratio between proportion of piscivorous and invertivore individuals (uPISC/uINV) shows response to chemical oxygen consumption (KPK)	0.506	0.02890
			Proportion of piscivorous individuals (uPSIC) shows response to ammonia concentration	0.477	0.03490
			Shannon index based on native species (Hnat) shows response to nitrates concentration	0.606	0.01400
	HR-R_12, HR-R_13, HR-R_15	Mid-altitude and lowland rivers, and rivers in karstic fields	Proportion of lithophilic species (pLITH) shows response to phosphorous concentration (P)	0.472	0.00140
			Proportion of phytophilic species (pFITO) shows response to phosphorous concentration (P)	0.640	0.00000
			Proportion of individuals belonging to native species (uSn) shows response to the proportion of non-native species (pSa)	0.392	0.00425
			Number of invertivore species (INV) shows response to conductivity	0.493	0.00100
	HR-R_16	Temporary rivers	Proportion of individuals belonging to Cypriniformes (uCYPR) shows response to suspended particles concentration	0.429	0.04651
			Proportion of piscivorous individuals (uPISC) shows response to ammonia concentration	0.458	0.03900
			Ratio between Shannon indices based on native species and the whole community (Hrat) shows response to nitrates concentration	0.465	0.03737
	HR-R_17. HR-R_18, HR-R_19	Lowland and mid-altitude small, medium and temporary rivers in Istria	Proportion of phytophilic species (pFITO) shows response to phosphorous concentration (P)	0.489	0.01000
			Proportion of invertivorous species (pINV) shows response to alkalinity	0.419	0.01860
			Alpha index based on native species shows response to conductivity	0.678	0.00113
			Ratio between Shannon indices based on native species and the whole community (Hrat) shows response to nitrates concentrations	0.378	0.02600

). Metrics of fish communities that showed significant responses to certain pressures were included in the index calculation.

For small rivers in the Pannonian ecoregion, those metrics were number of native species, proportion of psammophilic species, and difference between Shannon’s index based on the native species and all recorded species. In the lowland rivers of the Pannonian ecoregion, the number of non-native species was significantly higher as a response to pressures, which also affected the Hrat metric.

In the majority of rivers belonging to the Dinaric continental ecoregion (small, medium and large mountain, mid-altitude and lowland rivers), the proportion of individuals of invertivorous and omnivorous species, as well as the number of benthopelagic species (species feeding in the water column) turned out to best reflect anthropogenic threats and were, therefore, included in the index of ecological state. The proportions of both invertivorous and omnivorous individuals in fish communities showed significant response to the concentration of ammonia in the water. Those two metrics naturally interchange in watersheds, since the number of invertivores is higher in the upper parts of rivers and streams, while in downstream river stretches, a higher proportion of omnivores is present and both are sensitive to elevated levels of ammonia in the water column. On the other hand, the most sensitive metrics to environmental threats in rivers in karst fields and temporary rivers in the Dinaric continental ecoregion were the proportion of non-native species, the proportion of individuals of native species and the proportion of individuals of piscivore species, which vary in response to water temperature rise, elevated levels of nitrates and decreased concentration of dissolved oxygen in water, respectively.

Metrics of fish communities in lowland and mid-altitude small rivers in the Dinaric coastal ecoregion that showed clear, statistically significant responses to anthropogenic pressures mostly concern feeding preferences of fish species: the proportion of omnivorous species, and the ratio between proportion of piscivorous and invertivore individuals and proportion of piscivorous individuals and their modification in the community structure is connected with oxygen levels (the answer was attributed to dissolved oxygen concentration and chemical oxygen consumption) and ammonia concentration. Furthermore, the Shannon index based on native species corresponds to nitrates concentration in lowland and mid-altitude small rivers in the Dinaric coastal ecoregion. In the mid-altitude and lowland rivers of the Dinaric coastal ecoregion, as well as in rivers in karstic fields, the most problematic threats seem to be phosphorous concentration (provoking two responses in fish communities’ structures—proportion of lithophilic and proportion of phytophilic species), the proportion of non-native species (inducing expected changes in the proportion of native species), but also conductivity which affects the number of invertivore species. Two metrics connected with the reproductive mode of fish show different responses to elevated phosphorous levels—the proportion of lithophilic species is reduced, whereas the proportion of phytophilic species is increased with elevated phosphorous levels. Such a result probably reflects growth of vegetation due to eutrophic conditions. The metrics of fish communities in temporary rivers that are the most sensitive to anthropogenic threats are the proportion of individuals belonging to the order Cypriniformes, the proportion of individuals belonging to piscivorous species, and the ratio between Shannon indices based on native species and on the whole community. Those metrics respond to suspended particles, ammonia and nitrates concentrations, respectively. Waters of the Istrian Peninsula, that are separated into distinct river types following national typology, are particularly impacted by increased phosphorous and nitrates concentrations, as well as alkalinity and conductivity, provoking changes in the proportion of phytophilic species, the ratio between Shannon indices based on native species and on the whole community, the proportion of invertivorous species, and the alpha index based on native species. Temporary rivers in the Dinaric coastal ecoregion are characterized by particularly harsh environmental conditions. Nevertheless, they were not spared from significant anthropogenic influence as well. Major threats in this river type are elevated concentrations of suspended particles in water, ammonia and nitrates, and the fish community metrics that are the most sensitive to these threats are the proportion of species belonging to the order Cypriniformes, the proportion of individuals belonging to piscivore species, and the ratio of Shannon index based on native species and on the whole community.

3.2. Reference Localities for Croatian Freshwaters

Reference conditions were estimated following the procedure described in the Materials and Methods section for each river type/group of types, for which response to pressures was established. The most appropriate method for reference condition estimation is designation of reference localities, namely localities that still support undisturbed or minimally disturbed fish communities which remain in a natural or near-natural structural state. Unfortunately, due to the low number of rivers and streams within some river types, as well as the presence of anthropogenically induced pressures on the majority of localities, it was possible to identify adequate localities with fish communities in optimal states and which could serve as reference localities for classification systems only in eight out of 20 water types/groups of water types for which different indices of ecological states were developed (

Table 3. Reference localities at the national level for fish as a biotic element for certain river types/groups of types.

National River Types	Locality Code	Locality Name (Coordinates x, y)	Structure of the Ichthyological Community
HR-R_1	21114	Ivanečka Železnica, influence (476725, 5120041)	Natural and rich fish community, comprising nine native fish species.
HR-R_2A and HR-R_2B	There is not a single fish community in natural or slightly disturbed state (all communities are severely altered). Therefore, it was not possible to identify referent localities, so reference conditions were estimated based on expert judgement.
HR-R_3A, HR-R_3B, HR-R_3C and HR-R_3D	21052	Boščak II (507426, 5143287)	Natural fish community, comprising six typical, native species.
	21049	Bistrec–Rakovnica I (514270, 5136728)	Natural fish community, comprising five typical, native species.
	21050	Bistrec–Rakovnica II (523693, 5133252)	Natural and rich fish community, comprising nine native fish species.
HR-R_4A, HR-R_4B and HR-R_4C	18005	Sutla, Luke Poljanske (431485, 5113190)	Natural fish community, comprising seven typical, native species.
HR-R_6	30018	Curak, before its influence on Kupica (374243, 5033181)	Natural fish community, comprising four typical, native species.
HR-R_7	Neither locality can be designated as reference, although several fish communities have high or even maximal ecological quality ratios for all metrics included in the index. However, at the localities analyzed under this water type, fish communities are peculiar and naturally species-poor, or we found non-native species that have similar ecological characteristics as native ones, enabling retention of the ecological state on a high level, but still presenting severely altered fish communities.
HR-R_8A and HR-R_8B	There is not a single fish community in natural or slightly disturbed state (all communities are severely altered). Therefore, it was not possible to identify referent localities, so reference conditions were estimated based on expert judgement. It is important to mention that these water types include only two localities that could be analyzed.
HR-R_9	There is not a single fish community in natural or slightly disturbed state (all communities are severely altered). Therefore, it was not possible to identify referent localities, so reference conditions were estimated based on expert judgement. It is important to mention that this water type includes only two localities that could be analyzed.
HR-R_10A and HR-R_10B	There is not a single fish community in natural or slightly disturbed state (all communities are severely altered). Therefore, it was not possible to identify referent localities, so reference conditions were estimated based on expert judgement.
HR-R_11A and HR-R_11B	There is not a single fish community in natural or slightly disturbed state (all communities are severely altered). Therefore, it was not possible to identify referent localities, so reference conditions were estimated based on expert judgement.
HR-R_12	14006	Una, by Loskun spring (456858, 4950894)	Natural fish community, comprising four typical, native species.
HR-R_13 and HR-R_13A	40204	Zrmanja, Berberov buk (442116, 4895311)	Natural fish community, comprising four typical, native species.
HR-R_14A, HR-R_14B and HR-R_14C	None of the localities can be considered reference, based on the fish communities, because only three localities belonging to these river types were included in analyses, and those localities are very different based on their ichthyocenoses.
HR-R_15A and HR-R_15B	Only one locality belonging to these types was included in the analyses, with a moderately disturbed fish community.
HR-R_16A and HR-R_16B	There is not a single fish community in natural or slightly disturbed state (all communities are severely altered). Therefore, it was not possible to identify referent localities, so reference conditions were estimated based on expert judgement.
HR-R_17	Only one locality belonging to these types was included in the analyses, with a moderately disturbed fish community.
HR-R_18	There is not a single fish community in natural or slightly disturbed state (all communities are severely altered). Therefore, it was not possible to identify referent localities, so reference conditions were estimated based on expert judgement.
HR-R_19	There is not a single fish community in natural or slightly disturbed state (all communities are severely altered). Therefore, it was not possible to identify referent localities, so reference conditions were estimated based on expert judgement.

). For the remaining 12 water types/groups of types, upper and lower anchors of fish metrics were estimated. In localities that can be designated as reference localities for certain water types/groups of types (Ivanečka Železnica for HR-R_1; Boščak II, Bistrec-Rakovnica I and Bistrec-Rakovnica II for HR-R_3 (all subtypes); Sutla in Luke Poljanske for HR-R_4 (all subtypes); Curak before its influence into the Kupica River for HR-R_6; Una by Loskun spring for HR-R_12; and Zrmanja, Berberov buk for HR-R_13 and HR-R_13A), we found natural, undisturbed communities, often rich with species, comprising four to nine native fish species.

3.3. Ecological Indices for Croatian Running Waters

Ecological indices based on fish as a biotic element were successfully developed for all types/group of types of running waters in Croatia (

Table 4. Formulas for calculation of indices of ecological state based on fish in each of the running water types in Croatia.

Ecoregion	Water Body Type Code	Water Body Type Name	Formula for the Index Calculation
Pannonian	HR-R_1, HR-R_2A, HR-3_2B	Small mountain and mid-altitude rivers	$CFIR={\displaystyle \frac{EQR\left(Sn\right)+EQR\left(pPSAM\right)+EQR\left(Hdif\right)}{3}}$
	HR-R_3A, HR-R_3B, HR-R_3C, HR-R_3D HR-R_4A, HR-R_4B, HR-R_4C	Medium, large and alluvial lowland rivers	$CFIR={\displaystyle \frac{EQR\left(Sa\right)+EQR\left(Hrat\right)}{2}}$
Dinaric continental	HR-R_6, HR-R_7, HR-R_8A, HR-R_8B	Small, medium and large mountain and mid-altitude rivers, as well as medium and large lowland rivers	$CFIR={\displaystyle \frac{EQR\left(uINV\right)+EQR\left(uOMNI\right)+EQR\left(pWCOL\right)}{3}}$
	HR-R_9, HR-R_10A, HR-R_10B	Rivers in karstic fields and temporary rivers	$CFIR={\displaystyle \frac{EQR\left(pSa\right)+EQR\left(uSn\right)+EQR\left(uPISC\right)}{3}}$
Dinaric coastal	HR-R_11A, HR-R_11B	Lowland and mid-altitude small rivers	$CFIR={\displaystyle \frac{EQR\left(pOMNI\right)+EQR\left(uPISC/uINV\right)+EQR\left(uPISC\right)+EQR\left(Hnat\right)}{4}}$
	HR-R_12, HR-R_13, HR-R_13A	Mid-altitude and lowland rivers	$CFIR={\displaystyle \frac{EQR\left(pLITH\right)+EQR\left(pFITO\right)+EQR\left(uSn\right)+EQR\left(INV\right)}{4}}$
	HR-R_14A, HR-R_14B, HR-R_14C	Lowland short rivers with >5‰ slope	$CFIR={\displaystyle \frac{EQR\left(pOMNI\right)+EQR\left(uPISC/uINV\right)+EQR\left(uPISC\right)+EQR\left(Hnat\right)}{4}}$
	HR-R_15A, HR-R_15B	Rivers in karstic fields	$CFIR={\displaystyle \frac{EQR\left(pLITH\right)+EQR\left(pFITO\right)+EQR\left(uSn\right)+EQR\left(INV\right)}{4}}$
	HR-R_16A, HR-R_16B	Temporary rivers	$CFIR={\displaystyle \frac{EQR\left(uCYPR\right)+EQR\left(uPISC\right)+EQR\left(Hrat\right)}{3}}$
	HR-R_17, HR-R_18, HR-R_19	Lowland and mid-altitude small, medium and temporary rivers in Istria	$CFIR={\displaystyle \frac{EQR\left(pFITO\right)+EQR\left(pINV\right)+EQR\left(Anat\right)+EQR\left(Hrat\right)}{4}}$

). Indices are based on two to four Ecological Quality Ratios, so they mostly comprise the responses of fish communities to the two to four most problematic stressors in a certain water type/group of types. Metrics of all types turned out to show responses to certain pressures and were, therefore, included in index calculations. However, only one functional metric (INV) was used in indices for five river types. By contrast, seven community composition metrics (pSa, pINV, pOMNI, pLITO, pFITO, pPSAM and pWCOL) are important for index calculations in 36 river types, six richness/diversity metrics (Sn, Sa, Hdif, Anat, Hnat and Hrat) are included in indices for 29 river types, and six sensitivity/tolerance metrics (uSn, uCYPR, uINV, uOMNI, uPISC and uPISC/uINV) are used for index calculations for 28 river types in Croatia.

3.4. Ecological State of Croatian Running Waters

Following the newly developed classification scheme (described in this paper), we have estimated the ecological states of running waters throughout Croatia and revealed great variations—from undisturbed localities with natural fish communities that are fortunately still present inside some river types to localities with highly altered and damaged fish communities due to pronounced anthropogenic threats (Figure 2). Unfortunately, more than one stressor is present in most localities, with more or less severe effects on water systems and provoking alterations of fish communities. Generally, alterations are less pronounced in the upper parts of the river basins, and they become more obvious and problematic in the middle and, particularly, downstream areas.

Concordant with the conclusion that threats, and also alterations, are less present in the upstream rivers and streams, the ecological state of the small mountain and mid-altitude rivers and streams in the Pannonian ecoregion is mostly good, with the majority of localities having good or very good ecological state. However, even within this river type, there are localities with damaged fish communities, where elevated levels of phosphorous seem to be the most problematic threat to fish communities. We can assume eutrophication is occurring at those localities, since elevated levels of nitrogen have been recorded as well, so further deterioration of the water quality can be expected, followed by further reduction of the ecological state of those watersheds. This can only be prevented by taking urgent measures focusing on pollution prevention. Within this river type, non-native species have been recorded at two localities (Bednja, Stažnjevec and Kosteljina, Vrh Pregradski). The ecological state index is 0.61 in the Bednja River, indicating an urgent need for conservational measures directed at non-native species removal, in order to prevent further reduction of its ecological state below the moderate/good boundary. A second problem in the Bednja River is reduced oxygen concentration, which is probably a consequence of significant hydromorphological alteration of this river, but to a certain extent also eutrophication and pollution. Multiple stressors also cause great problems for fish communities in small lowland rivers, but in rivers of these types threats are even more pronounced. Mitigation measures are rarely present and fish communities are also composed of mostly sensitive, stenovalent species. As a consequence, rivers of this type are among the worst in terms of their ecological state in Croatia. Only one locality in this type has been assessed as having very good ecological state, 22 localities are good, in five localities the ecological state is moderate, eight localities have bad ecological state and two are in a very bad state. Again, in most localities with reduced ecological states and altered fish communities, more than a single stressor is acting. The most problematic threats for fish communities in this water type are reduced concentration of oxygen in the water, in some cases coupled with high levels of nitrogen. In addition, non-native species are present in many localities, causing additional problems for native species through competition, predation and/or alteration of habitat conditions. Despite their location in the downstream parts of river basins and the various threats affecting them, alluvial lowland rivers and medium and large lowland rivers mostly have fish communities less damaged then those in small lowland rivers. The probable reason for this result lies in the natural structure of fish communities—populations in small rivers are naturally less abundant than in medium-sized and large rivers, where species are also more tolerant to ecological fluctuations and there is more space for fish to avoid pollution and/or other negative effects. Out of 12 localities belonging to lowland alluvial rivers, seven are in very good ecological state based on fish communities, three are in good state, and two are moderate. The greatest problems for these localities are elevation of temperatures and reduction of oxygen level, as well as the presence of non-native species which have been recorded in some localities. Similarly, the majority of localities belonging to medium and large lowland rivers have very good or good ecological states (nine and ten localities, respectively), and for five localities we have designated moderate ecological state. Most medium and large lowland rivers harbor rich fish communities (with the number of species reaching 17). However, in most localities non-native species are also present. A second threat to these localities is reduced oxygen level, coupled and probably provoked by elevated temperatures.

A similar pattern regarding variations of the ecological states in various water types can be observed in the Dinaric continental region—ecological states are worse in types where multiple anthropogenically induced stressors act simultaneously and/or fish communities comprise sensitive, intolerant species. Most localities representing small, medium and large mountain and mid-altitude rivers, as well as medium and large lowland rivers in the Dinaric continental ecoregion express very good ecological state, with fish communities mostly composed of native species. Even in localities where non-native species are present, they do not dominate communities. Ichthyocenoses are stable and balanced, although not rich with species, but that is a natural condition for those watersheds. Invasive species can be identified as the most serious threat, with other stressors affecting fish communities in certain locations. The ecological state of rivers in karstic fields and temporary rivers in the Dinaric continental ecoregion is much worse. Ecological state has been estimated for two rivers in karstic fields: the ecological state of the Lika River at the Kosinj bridge locality is bad, which is no surprise, because all species found are non-native; the ecological state of the Gacka River is good for now, because 50% of species found are native. However, lowering of the ecological state of the Gacka River could be expected in the coming years, if the lack of invasive species removal and other conservational measures continues. Among the temporary rivers, one locality had only slightly disturbed fish community (Potkoš), showing an otherwise very good ecological state. On the other hand, the ecological state of the Lika River at the Bilaj locality and Ličanka is very bad, expressing well the observed conditions—reduced oxygen level, elevated temperatures and fish communities comprising exclusively non-native species.

Multiple stressors are acting on many localities in the Dinaric coastal ecoregion, and their effects are well reflected in the reduced ecological state indices of the water bodies. Lowland and mid-altitude small rivers in this ecoregion mostly show good ecological state, based on their fish communities. The exception is the Orašnica River, before its influence into the Krka River, which has a bad ecological state. Very low oxygen level, elevated chemical oxygen consumption and an extremely high ammonia level are all indicators of the eutrophication and pollution occurring at that locality and provoking pronounced changes in the fish community. Lowland short rivers with >5‰ slope are variable regarding their ecological states, depending on the threats acting at certain localities. Very good ecological state has been estimated for the Žrnovnica and Krupa Rivers, which are inhabited by natural or near-natural fish communities. In the downstream part of the Jadro River, the ecological state is moderate and elevated levels of nitrates indicates eutrophication. Ecological states of mid-altitude and lowland rivers, and rivers in karstic fields is mostly good and very good, and the fish community comprises mostly or exclusively native fish species. Nevertheless, the presence of non-native species in several localities (Krka downstream of Knin, Cetina in Trilj, Zrmanja below the Muškovci dam, Vrljika) is a reason for strong concern and urgent actions. In the temporary rivers and streams of the Dinaric coastal ecoregion, indices of the ecological state vary greatly among localities, but several localities do not show a satisfactory ecological state. The most problematic threat is the presence of non-native species, which are even dominant in several localities. Other threats causing negative effects for native fish communities and responsible for reduced ecological indices are elevated levels of suspended particles, nitrates concentrations, and ammonia concentrations. In the rivers and streams of the Istrian Peninsula, ecological state based on fish communities is good in the majority of localities. However, the presence of non-native species has again been observed in many localities (Pazinčica, Raša, Mirna, Stara Mirna, Boljunčica and Dragonja). Non-native species are not dominant in these fish communities at the moment. Nevertheless, their negative effects on the native populations are probably occurring already and might become more significant in the future. Greater abundances of non-native species were observed in two localities, which also express moderate ecological state (Boljunčica at Kožljak and the influence of the Dragonja River).

4. Discussion

4.1. The Strongest Pressures on Viability of Fish Communities in Croatian Rivers

Various threats and stressors negatively affect fish communities worldwide and are a reason for the strong decline of many populations, and even local extinctions. The most problematic threats reported worldwide are modification of habitats and natural flow regimes, deterioration of natural floodplains, various types of pollution and invasive fish species [9,10]. The same threats have been noticed in this investigation, provoking negative effects in fish communities throughout Croatia, even though the threats that have the most significant effects are not the same in all parts of Croatia, nor in all river types.

In the small rivers of the Pannonian ecoregion, the greatest problems for fish communities are posed by elevated phosphorous levels, elevated levels of dissolved nitrogen and reduced concentrations of dissolved oxygen, whereas in lowland alluvial, medium and large rivers of the Pannonian ecoregion, the strongest observed responses of fish communities have been to temperature and dissolved oxygen levels. Excessive phosphorous in watercourses appears as a result of pollution and eutrophication and is considered one of the best indicators, but also one of the strongest causes of eutrophication [44,45]. As expected, its effect is the strongest in small watercourses, where water is very clean and the natural concentration of dissolved phosphorous low. Since fish communities of small watercourses are adapted to these conditions and are mostly composed of stenovalent, sensitive species, a slight change in conditions is enough to provoke significant change in ichthyocenosis. Similarly to phosphorus, nitrogen is considered an important cause and indicator of eutrophication. It usually reaches watercourses via rinsing of agricultural fields where nitrogen-rich fertilizers have been used, but can also come from other pollutants. Another indicator of pollution and eutrophication is reduced oxygen level, which provokes a response from sensitive fish species inhabiting smaller watercourses. Dissolved oxygen concentration also turned out to be a serious threat for fish communities in lowland Pannonian rivers, together with temperature. The increase in the water temperature is usually a consequence of hydrological and morphological changes in watercourses, e.g., slowing down the flow with dams and barriers which leads to an increase in temperature and is often present in Croatian rivers, as well as throughout the Danube basin in Europe. Increased temperature is usually suboptimal or completely unsuitable for native species, especially sensitive ones, while invasive species have wider ecological valency and can tolerate more pronounced temperature fluctuations.

A high concentration of ammonia in the water, together with elevated levels of suspended particles, turned out to be the major threat to fish communities in mountain, mid-altitude and lowland rivers of the Dinaric continental ecoregion in Croatia. Increased levels of ammonium and ammonium ions in watercourses are usually a consequence of intensive use of artificial fertilizers and leaching from agricultural land, animal waste, industrial and urban wastewater and bacterial activity [46]. Also, large quantities of foreign species in watercourses can lead to an increase of ammonium and ammonium ions from excretion and microbial decomposition of dead fish. Ammonium ions dissolved in the water can lead to ionic imbalance in fish blood [47], and the molecular form can easily enter the fish body and shift to the ionic form, which is significantly more toxic and causes cell damage. Concentration of suspended particles in the water column, which yielded an obvious response in benthopelagic fish, can be increased by different kinds of water pollution. It is clear that it affects species that feed in the water column, e.g., reduced visibility leads to reduced feeding, abrasions on the skin can provoke infections, and also clogging of the gills and blood vessels of fish. Water temperature, nitrates and concentration of dissolved oxygen were denoted as major threats for fish communities in rivers in karst fields and temporary rivers of the Dinaric continental ecoregion. As already mentioned, native species adapted to specific conditions (particularly in karst field streams and temporary watersheds, where fish communities are highly unique, comprising endemic fish species with particular adaptations and specializations; see for example [24]) often cannot tolerate significant changes in temperature. On the other hand, invasive species (which are generalists and opportunists by nature) take advantage of conditions unfavorable to the indigenous community and create stable populations, causing additional negative impacts on the native fish species by predation, competition for resources or other negative interactions. Furthermore, an increase of nitrates in water and a decrease in the dissolved oxygen concentration are also noticed as threats causing significant and strong responses in fish communities in karst fields and temporary rivers.

Watersheds in the Dinaric coastal ecoregion, hosting unique fish communities with particularly high portions of endemic fish species, are influenced by various stressors, among which reduced concentration of dissolved oxygen, increased levels of ammonia, nitrates, phosphorous concentration, suspended particles, alkalinity, connectivity, and also abundant populations of invasive species have been denoted as the most problematic ones for the native fish communities. The greatest threats to the native fish communities in small mid-altitude and lowland rivers in the Dinaric coastal ecoregion are reduced levels of oxygen concentration and its greater consumption, as well as concentrations of ammonia and nitrates. The response of sensitive components of the fish community to changes in the oxygen concentration was recorded also in watercourses of the Pannonian ecoregion, and it is clear that stable oxygen concentration is crucial for preserving native ichthyofauna of small streams. Furthermore, oxygen consumption is an indicator of eutrophication, particularly in combination with elevated levels of ammonia and nitrates concentration, that are all denoted as threats causing changes in ichthyofauna of small and mid-altitude rivers in the Dinaric coastal ecoregion. Our results imply that piscivore species, which are native to these types of watercourses, are especially sensitive to increases in ammonia, and that their proportion in fish communities reduces with higher ammonia concentrations. The greatest pressures for ichthyofauna in mid-altitude and lowland medium and large rivers and rivers in karst fields in the Dinaric coastal ecoregion are concentration of phosphorous (again indicating eutrophication), but also conductivity and the proportion of non-native species. Electrical conductivity increases in cases of inflow of untreated wastewater (from industries, but also from cities and villages) into watercourses, given that wastewaters have a high conductivity. Conductivity depends on water temperature and it increases with higher temperatures [48,49]. Freshwater fish species generally do not well tolerate changes of salt concentration in the water. Increased salinity affects their osmoregulation, and they consume significantly more energy for osmoregulation or cannot perform it at all. In many investigations of pressures in water ecosystems, the appearance of non-native species has been considered as a consequence of other threats (habitat degradation, pollution, etc.) because they have occurred as a consequence of other anthropogenic activities. However, all non-native fish species in watercourses of the Dinaric coastal ecoregion in Croatia were introduced as a result of anthropogenic activities and not as a consequence of some other pressure, and surely they can be considered as a pressure for native fish communities. Their negative influence on native species is even more pronounced than physico-chemical, morphological or hydrological changes. Temporary rivers of the Dinaric coastal ecoregion, which in their natural state are characterized by harsh environmental conditions, specifically large fluctuations of environmental conditions and water level, are also under significant human influence. Suspended particles of organic and inorganic origin end up in water from different pollution sources—household wastewater, illegal landfills near or in watercourses, industrial wastewater, leaching of shores from which natural vegetation has been removed, etc. On the other hand, increased concentration of suspended particles in water negatively affects fish. Regardless of the type of particles involved, suspended particles reduce visibility and thus the efficiency of feeding in a watercourse. The presence of organic particles leads to increased microbial degradation, which provokes higher oxygen consumption and reduced concentration of dissolved oxygen in the water column. Inorganic particles cause even bigger problems—they often lead to clogging of the gills or blood vessels and to the death of fish. Elevated concentrations of ammonia and nitrates are, along with suspended particles, a major threat in the temporary rivers of the Dinaric coastal ecoregion and their effects have already been described.

4.2. Multiple Stressors in Croatian Running Waters and Their Effects on Fish Communities

Multiple stressors are present in all types/groups of types of running waters in Croatia, and not a single river type can be denoted where no or only a single pressure is present. Such a situation, present in many localities worldwide, is very problematic for native fish communities. Specifically, besides each stressor having its own negative effects and consequences, since more stressors are acting at the same time, their effects are usually intensifying at the expense of the populations, the structure of which becomes more and more damaged, and their abundancy reduced. Moreover, negative effects of the observed anthropogenically induced stressors are provoking further ‘internal’ stressors. Such a phenomenon, described as the extinction vortex, poses one of the greatest problems in conservation biology [50], because even when the primary, anthropogenically provoked threats become resolved and their negative effects on fish communities mitigated, recovery of the population does not occur, because internal stressors (that were first caused by external stressors) are still acting, causing further and further reduction of population size and diversity, as well as other negative effects on demographic and genetic parameters.

In previous chapters, we have already pointed out the most problematic stressors causing negative effects in Croatian freshwaters. In many localities, one stressor causes another one, but both of them separately, and even more so in combination, result in significant negative consequences for native freshwater species and communities. For example, in medium, large and alluvial lowland rivers, as well as in rivers in karstic fields and temporary rivers, water temperature rise (caused by slowing down of the river’s flow by various dams and obstacles and/or by climate change) is accompanied by a reduction in oxygen concentration; in small rivers of the Pannonian and Dinaric coastal ecoregion, temporary rivers and rivers in Istria, elevated levels of dissolved phosphorous are accompanied with elevated levels of dissolved nitrates, often also ammonia (in such cases eutrophication causes multiple stressors, all of which negatively affect native fish communities, particularly when acting at the same time). Elevated chemical oxygen consumption provokes a reduction in dissolved oxygen concentration, which again negatively affects fish communities in lowland and mid-altitude small rivers and short rivers with pronounced slope in the Dinaric coastal ecoregion.

In this investigation, we have analyzed only external stressors. Nevertheless, there is not much doubt that internal stressors have already been caused by external stressors and that they already act in many populations. Some have previously been revealed by other research; others have yet to be described, and this is a prerequisite for their mitigation. In the population of the native Salmo labrax Pallas, 1814 in the Plitvice Lakes area, as well as in the Žumberak-Samoborsko gorje Nature Park, reduction of genetic diversity and absence of gene flow was noticed as a consequence of multiple stressors [23,51], and a similar situation was observed for Barbus balcanicus Kotlík, Tsigenopoulos, Ráb & Berrebi, 2002 on several localities in the Danube basin in Croatia [25]. Population decline of Aulopyge huegelii Heckel, 1843 has also been connected with multiple anthropogenic threats [52], and low effective population size and genetic diversity was also observed within that species [22].

We can only expect that the effects of multiple stressors on fish communities and water ecosystems will be more pronounced and severe in the forthcoming years, and the only way to prevent negative consequences for fish communities and save species from the extinction vortex is the application of scientifically-based, multilevel and complex conservation plans, comprising actions focusing on all external and internal stressors simultaneously.

4.3. Mitigation Measures for Resolving Negative Effects of Multiple Stressors in Croatian Freshwaters

As already stated, due to multiple stressors acting on many localities in running waters in Croatia, only complex multilevel conservation plans can help mitigate their negative effects, assure stability of fish communities and, thereafter, bring those water bodies up to at least good and preferably very good ecological state. Those conservation plans should include activities directed at each of the observed stressor, but in many cases also directed at native fish species, because it is not reasonable to expect improvement in their structure, diversity and/or viability without active assistance.

For the localities belonging to type HR-R_1 (mountain and mid-altitude running waters in the Pannonian ecoregion) that have reduced ecological state, prevention of pollution can be pinpointed as the most important conservation measure, coupled with invasive species removal (from localities where they have been reported), but also prevention of the spread of invasive species into new watersheds. At certain localities (particularly the Bednja River), natural hydromorphological conditions should be restored, in addition to pollution prevention, because it is likely that they also cause reduction in the oxygen level and other negative effects on fish communities. Prevention of pollution and eutrophication, as well as invasive species removal, are also the two most important mitigation measures for small lowland rivers of the Pannonian ecoregion (HR-R_2A and HR-R_2B). Measures focusing on the observed stressors should be implemented by reintroductions and augmentations of native species, because some have already disappeared from certain watersheds and others have reduced population sizes and viabilities. In the lowland alluvial rivers (HR-R_3A, HR-R_3B, HR-R_3C, HR-R_3D), special attention should be devoted to non-native species and their removal should be the most important conservational measure, coupled with prevention of their further spread. Since invasive species have mostly achieved stable populations in localities with very pronounced fluctuations of physico-chemical parameters (in particular, temperature elevation and reduction of oxygen concentration), achieving oxygen levels adequate for native species and lowering temperatures would also be beneficial. Removal of the invasive species should also be a priority in large lowland rivers of the Pannonian ecoregion (HR-R_4A, HR-R_4B, HR-R_4C), because they present an important stressor in the majority of localities. This again should be coupled with measures aiming to lower water temperatures and increase oxygen levels, but also to recover and stabilize the natural fish community.

Fortunately, non-native species are not abundant in mountain, mid-altitude and lowland streams and rivers of the Dinaric continental ecoregion (HR-R_6, HR-R_7, HR-R_8A, HR-R_8B), where good and very good ecological states were estimated for the majority of localities. Therefore, only local mitigation measures are required, considering invasive species removal at localities where they have been observed, as well as pollution prevention (a particularly high concentration of ammonia was recorded at Otuča downstream of Gračac). The greatest stressors for fish communities in running waters of the karstic fields (HR-R_9) are invasive species and habitat degradation, implying invasive species removal and habitat restoration, together with measures focusing on the recovery of native fish species, should be the mitigation measures of choice for localities with estimated ecological state lower than good. Invasive species removal, coupled with an increase in oxygen concentration and lowering of the water temperature, would have a favorable impact on the fish communities and ecological states of temporary rivers of the Dinaric continental ecoregion.

In the mid-altitude and lowland small running waters in the Dinaric coastal ecoregion (HR-R_11A, HR-R_11B), ecological state is mostly good, but mitigation measures are urgently required for the locality Orašnica before its influence to Krka, and they should be directed towards halting pollution and eutrophication, which are both significant there. A similar situation can be seen in the lowland rivers with greater slope (HR-4_14A, HR-R_14B, HR-R_14C), where in certain localities (lower part of the Jadro River) measures against pollution and invasive species should be undertaken. Removal and control of the non-native species should be the primary mitigation measure in mid-altitude and lowland medium and large rivers, as well as rivers in the karstic fields in the Dinaric coastal ecoregion. Namely, in several localities belonging to those river types, non-native species are already present, and an increase in their abundance—expected in the near future if no measures are taken—would certainly cause more significant negative effects in native fish communities and lowering of the ecological states of water bodies. Invasive species are also a great threat for fish communities in temporary rivers of the Dinaric coastal ecoregion (HR-R_16A, HR-R_16B); therefore, conservation plans should include their control and removal. Moreover, prevention of pollution is a required mitigation measure in several localities. Since fish communities in several localities are severely altered, mitigation measures should also aim at recovering the structure and diversity of native fish communities. Removal of invasive species and prevention of their spread are also required in order to improve the ecological states of many Istrian running waters (HR-R_17, HR-R_18, HR-R_19).