Bulgaria Report on fitting a new classification method to the results of the completed intercalibration of the Fish Cross GIG (Danubian Group)

BQE: Fish

Ministry of Environment and Water of Consortium DICON – UBA Apostolos Apostolou (IBER-BAS | DICON) Luchezar Pehlivanov (IBER-BAS | DICON) Michael Schabuss (ProFisch | UBA) Horst Zornig (ProFisch | UBA) Georg Wolfram (DWS Hydro-Ökologie | UBA) Robert Konecny (UBA) Emil Todorov (DICON) Doychin Todorov (DICON) Didier Pont (IRSTEA)

Version: 6.2 (Submitted)

2 \ 28 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish

Vs. Date Comments

1.0 25 Jan 16 1st draft by Apostolos Apostolou, distributed via e-mail to DICON, UBA and fish experts

2.0 2 Feb 16 2nd draft, including remarks from Michael Schabuß and Georg Wolfram, submitted to the fish experts as well as DICON / UBA by Georg

3.0 8 Feb 16 2nd draft, after fish group meeting in Sofia, submitted to the fish experts as well as DICON / UBA

4.0 11 Feb 16 2nd draft by Apostolos Apostolou, distributed via e-mail to DICON, UBA and fish experts, based on draft 3.0 and additional comments. 4.1 12 Feb 16 Minor comments and amendments by Georg W, distributed via e-mail to DICON, UBA and fish experts 5.0 25 May 16 Near-final draft, with results based on the statistical calculations of Didier Pont, Georg Wolfram and Michael Schabuß; distributed via e-mail to DICON, UBA and fish experts 5.1 31 May 16 Near-final draft after minor corrections from Michael; sent to Didier by Michael 5.2 13 Jun 16 Near-final draft, sent to Didier Pont 5.3 16 Jun 16 9:00 Near-final draft, including minor remarks from Michael and Apostolos. 6 21 Jun 16 Comments on method description. Changed legends of fig. 3,4,5 6.1 21 Jun 16 Minor changes in chapter 8 (“ad 3”) 6.2 21 Jun 16 Final draft, accepted by the Ministry of environment, submitted to JRC

June 2016 Apostolos et al. (Consortium DICON – UBA) Fit-in procedure: CrossGIG (Danubian Group) in Bulgaria, BQE: Fish 3 \ 28

CONTENT

1 Introduction ...... 6 2 Typology ...... 6 3 Description of the national assessment method ...... 9 3.1 Concept ...... 9 3.2 Sampling and laboratory calculations ...... 10 3.3 Overview of the assessment method ...... 10 4 National data set ...... 11 5 Pressure – impact relationships ...... 12 6 Reference/benchmark setting process and boundary setting procedure ...... 15 6.1 Rationale of reference conditions for the TsBRI parameters ...... 15 6.2 Reference conditions in Bulgarian waters ...... 15 6.3 Minimally disturbed sites at the European level ...... 16 6.4 Final list of reference (benchmark) sites ...... 17 6.5 Description of boundary setting procedure ...... 18 7 Checking of WFD compliance and evaluation of the assessment concept ...... 19 8 IC Procedure ...... 20 9 References ...... 26 10 Annex ...... 27

June 2016 Apostolos et al. (Consortium DICON – UBA) 4 \ 28 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish

TABLE OF CONTENTS

Table 1: General IC river types identified in Bulgaria, included in the Cross GIG for fish fauna in rivers, Danube group ...... 7 Table 2: Characteristics of national types and sites included in IC Cross GIG for fish fauna in rivers of the Danubian group (subset for the final intercalibration, n = 77) ...... 9 Table 3: Metrics used in the TsBRI and correspondence to the WFD obligatory criteria 10 Table 4: Brief description of the algorithm of the respective metrics used in TsBRI ...... 11 Table 5: Correlation analysis between pressures and the national assessment method TsBRI ...... 12 Table 6: River types, fish zones and reference values for the TsBRI. For each type/zone, the number of dominant, subdominant (= accompanying) and rare species as well as the total species richness under reference conditions is given ...... 16 Table 7: Criteria used for undisturbed sites selection. Only sites characterized by no or low pressure intensity (depending of the considered pressure) are selected (red modalities) ...... 17 Table 8: Intercalibration reference sites (IRS), derived from the criteria on the European level (NDS) and national criteria (NRS). The last columns shows the EQR values for different site-years. For the BSP, the dates marked with an asterisk were chosen .. 18 Table 9: Class boundaries for the EQR and the TsBRI values in Bulgarian rivers of the Danubian Group. Abbreviations see Table 6 ...... 18 Table 10: Compliance criteria for the assessment method based on fish in Bulgarian rivers in the Danubian Group ...... 19 Table 11: Number of reference (= benchmark) sites in the three countries within the Danubian Group, which have finalized the intercalibration ...... 21 Table 12: Values of the common metrics expressed in EQR for the reference sites of the different Danubian countries ( with 2 methods for salmonid and cyprinid rivers) ...... 22 Table 13: Translation of the reference and boundary positions of the national method on the basis of OLS regression (see Figure 6) into ICM ...... 24 Table 14: Summary of boundary values ...... 25

June 2016 Apostolos et al. (Consortium DICON – UBA) Fit-in procedure: CrossGIG (Danubian Group) in Bulgaria, BQE: Fish 5 \ 28

FIGURE OF CONTENTS

Figure 1: PCA analysis based on the metric abiotic characteristics of monitoring sites from the four river types R2, R4, R7 and R8 (in different colours; n = 147)...... 8 Figure 2: Metric hydro-morphological parameters for national river types included in the IC of the Cross GIG (Danubian Group; n = 147)...... 8 Figure 3: Box-plots for selected pressures against the TsBRI (n = 147)...... 13 Figure 4: Box-plots for selected pressures against the TsBRI (n = 147)...... 14 Figure 5: Relationship between the pressure index and the EQR (TsBRI) in rivers of the Danubian Group (national river types R2, R4, R7 and R8)...... 14 Figure 6: OLS regression to establish the relationship between the classification of the national methods (x) and the common metric Fish.Index (y) in the Danubian Group of rivers...... 23 Figure 7: Class boundaries ±0.25 class width of the national methods on the common metric scale compared with the global mean of the finalized method (Romania with 2 methods for salmonid and cyprinid rivers)...... 25

June 2016 Apostolos et al. (Consortium DICON – UBA) 6 \ 28 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish

1 Introduction

The official intercalibration of the Bulgarian fish-based method of ecological status assessment in the river types of the Danubian Group was not finalized due to the fact that such an assessment system was not fully developed and officially approved until 2012. The objective of this report is i) to present the finalized national method, ii) to prove that the present Bulgarian classification method based on fish for Eastern Continental rivers is compliant with the WFD normative definitions and its class boundaries are in line with the results of the completed intercalibration exercise of the Fish Cross GIG (Danubian Group). In particular, the classification method was verified for WFD compliance and IC feasibility and the pressure – response relationship was tested. Class boundaries were compared with agreed boundaries from the Cross GIG intercalibration exercise following the instructions of the CIS Guidance Document n°30: “Procedure to fit new or updated classification methods to the results of a completed intercalibration exercise”.

2 Typology

The Fish Cross GIG did not use the common intercalibration typology in its IC work (Eastern Continental GIG, Alpine GIG etc.). Instead, it was agreed to define the following regional groups: Nordic Group Lowland-Midland Group Alpine-type Mountains Group Mediterranean South-Atlantic Danubian Group Within each regional group, all the river sites were considered together for the IC exercise. Hence, a feasibility check in terms of typology is not relevant (Pont et al. 2011). Bulgaria has rivers which belong to the Mediterranean South-Atlantic Group (EABD and WABD) and to the Danubian Group (DRBD). The rivers which drain to the Black Sea (BSBD) are included in the Danubian Group as well. The Danube is not considered here, but will be treated within the XGIG for Very Large Rivers. Bulgaria is included in the Danube group of the Cross GIG for fish-based assessment methods in rivers along with the Czech Republic, Slovakia and Romania. In view of the assembled type of the GIG, common IC river types are not defined. In the frame of each IC group all monitoring sites are considered as a whole during the IC process. In connection with the GIG agreed common data base included in the IC process, Bulgarian data are not divided according to their typological characteristics. In the common IC database there will be included the following Bulgarian river types:

June 2016 Apostolos et al. (Consortium DICON – UBA) Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish 7 \ 28

 R2 Mountain rivers in ER12 – general type R-E1b;  R4 Semi-mountain rivers in ER12 – general type R-E4;  R7 Large Danube tributaries – general type R-E2;  R8 Medium and small rivers Danube – general type R-E3. In the Cross GIG (Danubian Group) Bulgaria participates with four national types which may be referred to the relevant common European rivers types (Table 1).

Table 1: General IC river types identified in Bulgaria, included in the Cross GIG for fish fauna in rivers, Danube group

Characterization Catchment Altitude Type Ecoregion Geology Substratum of the rivers area(km2) (m) Mountain rivers in 500 – R-E1a/ R2 10 10 – 1000 mixed the Ecoregion 12 800 Medium and small R-E2/ R8 rivers, Danube and 11 & 12 100 – 1000 < 200 mixed sand and silt Black Sea Large Danube sand, silt and R-E3/ R7 11 & 12 > 1000 < 200 mixed tributaries gravel Semi-mountain 200 – sand and R-E4/ R4 rivers in the 11 & 12 100 – 1000 mixed 500 gravel Ecoregion 12

The national types included in the Cross GIG (Danubian Group) correspond with the common river types, in view of their hydro-morphological characteristics. The total number of monitoring sites included in the IC program is 147. Among these, 77 were selected for the final comparison based on the following criteria:  Only sites with wetted river width ≤12 m;  Only data with at least 30 individuals (except national river R2);  Only data with at least 5 species (national river type R7) or 4 species (national river type R4 and R8); no restriction for national river type R2;  Only 1 sampling date per site (in most cases, the data from 2015 were used, since 2014 high water had hampered optimal sampling).

The main metric hydro-morphological variables included in the IC exercise concerning national river types are:  Altitude of the water body;  Size of catchment area;  Distance from the spring;  River slope. The ordination plot in Figure 1 and the box plots in Figure 2 show the variability of the four types in terms of morphological characteristics. Table 2 gives an overview of the abiotic characteristics for the subset of 77 sites.

June 2016 Apostolou et al. (Consortium DICON – UBA) 8 \ 28 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish

Figure 1: PCA analysis based on the metric abiotic characteristics of monitoring sites from the four river types R2, R4, R7 and R8 (in different colours; n = 147)

Figure 2: Metric hydro-morphological parameters for national river types included in the IC of the Cross GIG (Danubian Group; n = 147).

Apostolou et al. (Consortium DICON – UBA) June 2016 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish 9 \ 28

Table 2: Characteristics of national types and sites included in IC Cross GIG for fish fauna in rivers of the Danubian group (subset for the final intercalibration, n = 77)

Average size of Average Number Type Characteristics the watershed Altitude Geology Substratum of sites (kм2) (m) Mountain rivers in boulders and R2 24 96 561 mixed the Ecoregion 12 stones Semi-mountain big and small R4 rivers in the 31 622 299 mixed gravel, sand Ecoregion 12 Large Danube sand, silt, clay, R7 6 3316 50 mixed tributaries loam, gravel Medium and small gravel, sand, R8 16 437 91 mixed rivers Danube silt, clay, loess

3 Description of the national assessment method

3.1 Concept The multi-metric Fish Based Index (BRI) follows the classical approach of Karr (1981, 1986). The first conception was reported by S. Mihov (2010). Later versions resulted in a group work, described in Belkinova et al. (2013). More than two years of improvement – like calibration with the benthic invertebrates’ assessment system, completing the sensitive species tables and eliminating expert judgement – led to a final version of the Bulgarian Fish Index, approved officially by the Ministry of the Environment by the end of 2012. For the current IC exercise the method was slightly adapted on the basis of a new dataset (harmonized sampling method), new type specific reference communities/values and new class boundaries. It is also calibrated for fish assessment of the trout zone as well as for small rivers with a river width smaller than 6 m (restriction for the older version of the index). The current version is renamed as Type Specific Bulgarian Fish Index (TsBRI), in order to be distinguished from previous versions. It is further assumed that TsBRI can be applied also to rivers with a wetted width >12 m. In this case, however, more than two back-pack electro-fishing devices or (if river depth exceeds 0.7 m) an electro-fishing boat is required. During the last 2 sampling campaigns (2014 and 2015), a suitable electro-fishing boat was not available. The calculation revealed that the assessment may give unreliable and implausible results, if the number of fish individuals or the number of species is too low. Therefore, the following restrictions are recommended  In rivers belonging to the national river types R4, R7 and R8, the number of individuals in the sample should be at least 30. For R2 rivers, there is no restriction;  In rivers belonging to the national river types R4, R7 and R8, the species richness per sample should be at least 4 (R4 and R8) or 5 (R7). For R2 rivers, there is no restriction.

June 2016 Apostolou et al. (Consortium DICON – UBA) 10 \ 28 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish

3.2 Sampling and laboratory calculations The assessment method is adapted to fish samples collected by electric current, according to European standard EN 14011(1) and the following remarks:  sample collection during dry season;  fishing area/site selection: multi-habitat sampling. Sampling is performed once annually in a specified time window each year (summer, early autumn when hydrological conditions are stable). It is usually done by wading in the river, according to standard EN 14011 (Water Quality: Sampling of Fish with Electricity). The sampling strategy is focused on all accessible habitats per site (multi- habitat sampling). Sample processing: all specimens per species are counted and measured (total body length). Total biomass per species is weighted in situ; then all specimens are released in the river. Laboratory calculations: total and individual species abundance (Ind./ha) and biomass (kg) per ha are calculated. Status of the registered species is clarified (sensitive, alien, predatory, migratory etc.), dominance is also calculated. The age structure is formulated as bad, average and good in accordance with the registered length classes.

3.3 Overview of the assessment method Following the requirements defined in the CIS Guidance No 7 – Monitoring (WG 2.7), the following indicative parameters have to be included for fish in rivers:  Taxonomic composition;  Abundance;  Disturbance sensitive taxa;  Age structure. The Type Specific Bulgarian Fish Index (TsBRI) is a multi-metric index, which uses eight metrics for the calculations. All WFD obligatory criteria indicative of the BQE are covered (Table 3).

Table 3: Metrics used in the TsBRI and correspondence to the WFD obligatory criteria Metrics in TsBRI WFD Criteria, referred to the parameter Total species number Taxonomic composition Total abundance/biomass Abundance Type-specific species Age structure + Disturbance Predatory species (only for some national types) Age structure + Disturbance Invasive species Age structure + Disturbance Migratory species Age structure + Disturbance Sensitive species Age structure + Disturbance/Sensitive taxa Dominant species Age structure + Disturbance/Sensitive taxa

Apostolou et al. (Consortium DICON – UBA) June 2016 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish 11 \ 28

total sum of sampling scores The index is calculated as: total sum of reference scores

The key elements of the assessment method are described in detail in Mihov (2010) and Belkinova et al. (2012). Seven metrics are used, each of them with points from 0 to a maximum of 10, 15 or 20 (depending on the metric). The Type Specific Bulgarian Fish Index (TsBRI) is calculated as the sum of the points (Table 4).

Table 4: Brief description of the algorithm of the respective metrics used in TsBRI

Categories Metrics Max. points per metric 1. Community structure of the indicative species/Community structure of 10 3 the predatory species (only for some river types) 2. Community structure of the migratory species 20 5 3. Community structure and comparative abundance of the sensitive 20 5 species 4. Tolerance of the dominant species 20 4 5. Total abundance and biomass per unit area 0 6 6. Comparative abundance of alien species 15 5 7. Total number of species 15 25 Total points = TsBRI 100

For the final assessment, the EQR is calculated based on type-specific reference values for the TsBRI. These were derived from reference coenoses adapted for each river type and fish zone (see Annex). Finally, the EQR is converted to normalized EQR values (nEQR) to get class boundaries at 0.2, 0.4, 0.6 and 0.8. According to the obtained results, the status is classified in one to five classes, as required by the WFD:

Status Class EQR nEQR High ≥0.86 ≥0.80 Good 0.65–0.85 0.60–0.79 Moderate 0.43–0.64 0.40–0.59 Poor 0.22–0.42 0.20–0.39 Bad <0.22 <0.20

4 National data set

The 77 sampling sites are spread over whole Bulgaria (range latitude: 42.03334 – 43.92621, range longitude: 22.45623 – 27.77138) and belong to four different national types (R2, R4, R7 and R8). The range of ecological status covers four classes:

June 2016 Apostolou et al. (Consortium DICON – UBA) 12 \ 28 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish

 High: 10;  Good: 39;  Moderate: 23;  Poor: 5. Although no sites in bad status are available, the gradient is considered as sufficient for intercalibration of the national method. The gradient of pressure is shown in the following section on the pressure – impact relationship.

5 Pressure – impact relationships

A correlation analysis revealed a highly significant correlation for the most important pressures and for the sum of all pressure values (Table 5). Also the correlation between the sum of all pressures and the TsBRI is highly significant (p<0.001).

Table 5: Correlation analysis between pressures and the national assessment method TsBRI Pressure intensity - number of sites TsBRI Pressure No (0) Low (1) Medium (2) High (3) Spearman correlation VegRiparian_site 44 56 111 39 p < 0.001 VegRiparian_seg 52 63 96 39 p < 0.001 HabAlter_seg 61 104 65 20 p < 0.001 Dykes 173 14 28 35 p < 0.001 HabAlter_site 101 88 43 18 p < 0.001 Abstraction 119 84 46 1 p < 0.05 Channelization 188 16 25 21 p < 0.001 Dams_seg 133 92 0 25 p < 0.001 Migr_Barriers 163 54 0 33 ns Impoundment 103 119 0 28 ns Migr_Barriers_catch_down 202 21 0 27 ns Migr_Barriers_catch_up 185 40 18 7 ns Dams_site 213 14 0 23 p < 0.05 Migr_Barriers_down 218 10 7 15 ns Migr_Barriers_up 222 22 2 4 ns Hydropeaking 217 28 0 5 ns TempMod 240 5 0 5 ns Toxic 250 0 0 0 ns Acidification 248 2 0 0 ns Navigation 250 0 0 0 ns Recreational 250 0 0 0 ns

Apostolou et al. (Consortium DICON – UBA) June 2016 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish 13 \ 28

Figure 3: Box-plots for selected pressures against the TsBRI (n = 147)

June 2016 Apostolou et al. (Consortium DICON – UBA) 14 \ 28 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish

Figure 4: Box-plots for selected pressures against the TsBRI (n = 147)

The final pressure-response relationship is shown in Figure 5. The pressure index included the following parameters (weights in brackets):

Pressure index = migration + hydro-peaking + water abstraction + temperature alteration + channelization * 2 + riparian vegetation alteration + habitat alteration * 2 + dykes * 2 + toxicity * 10 + acidification + CLC urban / 5 + CLC intensive agriculture / 10 (The reason for dividing the CLC by 5 and 10, respectively, was to bring the number approximately to the same scale: hydro-morphological pressures range from 0 to 3 or 4, CLC ranges theoretically from 0 to 100).

EQR(TsBRI)

Figure 5: Relationship between the pressure index and the EQR (TsBRI) in rivers of the Danubian Group (national river types R2, R4, R7 and R8)

Apostolou et al. (Consortium DICON – UBA) June 2016 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish 15 \ 28

6 Reference/benchmark setting process and boundary setting procedure

6.1 Rationale of reference conditions for the TsBRI parameters The Type Specific Bulgarian Fish Index (TsBRI) assesses the difference between the actual fish community and the reference fish community. In Bulgarian rivers the reference fish community shows a great species diversity. Sensitive species predominate in number and biomass, while the tolerant species play a minor role, though they still form a part of the community. Migrating species can freely reach spawning and feeding grounds. All ecological niches are occupied, being in dynamic equilibrium, thus invasive species are usually less successful than in deteriorated systems. In areas where the rivers are large enough to allow the establishment of a great number of ecological niches, the trophic structure of fish communities is composed of all trophic levels; even predatory species are represented by several species with good age structure. All species find appropriate conditions for feeding and reproduction and show a good age structure. Hybrids, sick, individuals infested with parasites are rarely met.

6.2 Reference conditions in Bulgarian waters Due to the fact that most water bodies in Bulgaria are affected by various anthropogenic factors, it is difficult to establish reference conditions and define exact reference values for all parameters used in the TsBRI. Accordingly, the least-disturbed Bulgarian sites are characterized by near-reference conditions. Reference values were derived from the combination of:  Best available (= near-reference) sites: Index values preferably higher than 0.85 for the last sampling years;  Historical data: species community, mainly presence/absence (partly dominance);  Expert judgment, based on the ecological requirements of the species and the abiotic characteristics in the national river types and fish regions. Type specific communities were determined for all national river types, which are covered in this report (see chapter 2) as well as all fish zones occurring in these types. The fish taxa per river type / fish zone lists are provided in Annex II. The classification of fish species according to their dominance in different national types and fish regions follows the approach of , Germany and . The reference TsBRI values for each river type / fish zone were derived as follow: - the reference coenosis was defined (Annex II). Taxa were classified as dominant, accompanying or rare species; - The relative proportion was converted to abundance under the following assumptions: o total abundance = 10,000 ind./ha; o relative proportion of dominant species = 75%;

June 2016 Apostolou et al. (Consortium DICON – UBA) 16 \ 28 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish

o relative proportion accompanying species = 20%; o relative proportion of rare species = 5%; o All species within each dominance class were evenly distributed, but finally evaluated by exert judgment depending on ecology. - Biomass was roughly derived by assuming a mean weight per individual and fish; - Age class was estimated as follows: o 1 = natural, all age classes present; o 2 = some age classes missing or only with a few specimens; o 3 = several age classes missing; o Under reference conditions, all taxa should have age class 1, but given a defined sampling method, rare species will hardly be caught in all size classes (e.g. predatory species). Hence, even under reference conditions, only dominant species should have age class 1, while accompanying species may have 1 or 2 and rare species 2 or 3. The TsBRI reference values for each river type / fish zone are listed in Table 6.

Table 6: River types, fish zones and reference values for the TsBRI. For each type/zone, the number of dominant, subdominant (= accompanying) and rare species as well as the total species richness under reference conditions is given

Ecoregion 12 7 Water shed DR BS WA EA River type R2 R2 R4 R4 R6 R7 R8 R2 R4 R4 R14 R14 R14 Fish zone ER MR MR HR EP EP HR/EP MR MR HR MR HR HR dominant 1 2 2 3 7 4 4 1 1 4 2 2 3 accompanying 1 3 4 4 16 5 4 4 4 2 2 4 5 rare 0 5 7 10 33 26 13 3 3 7 2 5 2 total 2 10 13 17 56 35 21 8 8 13 6 11 10 TsBRI reference value 59 69 65 79 – 78 74 70 72 82 80 64 80 Ecoregion 12 = Pontic province (Danube river basin), 7 = Eastern Balkan (Mediterranean catchment). Water sheds: DR = Danube River, BS = Black Sea, WA/EA = Western / Eastern Aegean. Types see chapter 2. The Danube River itself (R6) is no covered by the method in this report. ER = epirhithral, MR = metarhithral, HR = hyporhithral, EP = epipotamal.

6.3 Minimally disturbed sites at the European level At the European level, all reference sites have to be qualified as minimally disturbed sites at the local scale, using a set of common variables describing the intensity of different types of pressures (water quality, hydro-morphological pressures, connectivity alteration…). These sites are mentioned as “undisturbed sites” or “non-disturbed sites” (NDS).

Apostolou et al. (Consortium DICON – UBA) June 2016 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish 17 \ 28

Undisturbed sites were defined using the criteria listed in Table 7. No information was available for the criteria “water quality index”, “water quality alteration” and “fish farms / ponds”. However, the sites selected based on the other criteria are known by expert judgment not to be affected by these three criteria. The 28 selected sites could thus be accepted as NDS.

Table 7: Criteria used for undisturbed sites selection. Only sites characterized by no or low pressure intensity (depending of the considered pressure) are selected (red modalities)

Pressure type Scale Pressure intensity Nb of modalities Presence of downstream artifical barriers on the catchment scale catchment no low high 3 Artificial barriers upstream from the site segment no low medium high 4 Artificial barriers downstream from the site segment no low medium high 4 Impoundment site no low high 3 Hydropeaking site no low high 3 Water abstraction site no low medium high 4 Colinear connected reservoir (fish farms, fish ponds ...) segment no high 2 Upstream dams influence site no low high 3 Water temperature modification (excuding dam effect) site no high 2 Channelisation / Cross section alteration (segment scale) segment no low medium high 4 Riparian vegetation site no low medium high 4 Local Habitat alteration site no low medium high 4 Dykes (flood protection) segment no low medium high 4 Toxic Risk. Priority substances list segment no low high 3 Water acidification segment no low high 3 National water quality index (segment scale) segment no low medium high 4 Water quality alteration (local scale) site no low medium high 4 Navigation segment no high 2 Recreational use with high intensity (angling, boating,..) site no high 2 impairment of indigenous species segment no high 2 heavy predation site no high 2 major effect on indigenous populations by stocking activities segment no high 2

6.4 Final list of reference (benchmark) sites According to the CrossGIG report, a site will be selected as a reference site at the European level when the site is both an undisturbed site (European level) and a national reference site. In that way, it is assumed that the reference sites selected within all Europe reflect comparable reference condition (in relation with the considered pressures) and also the regional specificity of rivers at the regional/national scale. Non disturbed sites (NDS) are classified undisturbed (“yes”) or not (“no”) regarding to the common criteria listed above. The National reference sites (NRS) are the sites classified as reference sites by Bulgaria regarding to the national criteria (EQR for TsBRI ≥0.85). The Exceptional reference sites (ERS) are the sites classified as a reference sites, while it is considered as a disturbed site (a good justification is given). The Intercalibration reference sites corresponds to all sites, which are both classified as non disturbed (NDS = “yes”) and as a national reference site (NRS = “yes”) as well as Exceptional disturbed sites accepted as reference sites (ERS = “yes”).

June 2016 Apostolou et al. (Consortium DICON – UBA) 18 \ 28 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish

The total number of sites selected as Intercalibration reference sites in Bulgaria is 5 sites and 9 site-years (some sites were sampled in two subsequent years). Table 8: Intercalibration reference sites (IRS), derived from the criteria on the European level (NDS) and national criteria (NRS). The last columns shows the EQR values for different site- years. For the BSP, the dates marked with an asterisk were chosen ICSites_SiteCode River Site Date EQR BG1OG00699MS120 Botunya above the town of Varshets /district Zanozhene/ 17.06.15 0.93* BG1OG00699MS120 Botunya above the town of Varshets /district Zanozhene/ 30.08.14 0.90 BG1ER00099MS010 Erma near Strezimirovtsi village 20.06.15 0.86* BG1ER00099MS010 Erma near Strezimirovtsi village 01.09.14 0.86 BG1ER00033MS020 Erma near the town of Tran 20.06.15 0.88* BG1ER00033MS020 Erma near the town of Tran 01.09.14 0.93 BG1IS00078MS210 Shipochnitsa before influx in Iskar dam 30.09.14 0.91 BG1IS00078MS210 Shipochnitsa before influx in Iskar dam 02.09.15 0.98* BG2VE08633MS425 Aydere bridge to Stoilovo 10.08.11 0.90 BG2VE08633MS425 Aydere bridge to Stoilovo 16.07.13 0.82*

6.5 Description of boundary setting procedure Neither any discontinuity in the relationship between pressure and the impact nor a paired metric analysis (both approaches according to Guidance Document No. 14) was helpful to set the boundaries for the status classes. Following Step 8 of the Boundary Setting Protocol (CIS Guidance Document No. 14 (EC 2011), page 65), the continuum of impact was divided into equal width classes. In a first step, the H/G boundary was defined as the 25th percentile of the EQR values of the reference sites (IRS), which is 0.86. Then, the remaining gradient was divided to get equal distance class boundaries. The EQR boundaries are: H/G boundary 0.86 G/M boundary 0.65 M/P boundary 0.43 P/B boundary 0.22

Table 9: Class boundaries for the EQR and the TsBRI values in Bulgarian rivers of the Danubian Group. Abbreviations see Table 6 Ecoregion 12 7

Water shed DR BS WA EA

River type R2 R2 R4 R4 R6 R7 R8 R2 R4 R4 R14 R14 R14

Fish zone ER MR MR HR EP EP HR/EP MR MR HR MR HR HR

EQR Reference value 1.00 59 69 65 79 – 78 74 70 72 82 80 64 80 High status ≥0.86 >50 >59 >55 >67 >67 >63 >60 >61 >70 >68 >55 >68 Good status ≥0.65 >38 >44 >42 >51 >50 >48 >45 >46 >53 ≥52 >41 ≥51 Moderate status ≥0.43 >25 >29 >27 >33 >33 >31 >30 >30 >35 >34 >27 >34 Poor status ≥0.22 >12 >15 >14 >17 >17 >16 >15 >15 >18 >17 >14 >17

Apostolou et al. (Consortium DICON – UBA) June 2016 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish 19 \ 28

7 Checking of WFD compliance and evaluation of the assessment concept

WFD Compliance According to EC (2011) only assessment methods meeting the requirements of the WFD can be intercalibrated. An important step in the intercalibration procedure is the checking of the national methods considering various WFD compliance criteria. The WFD compliance criteria are specified in the reporting template for milestone reports (Annex VI of EC 2011). We referred to this template to document the compliance of the national assessment methods in the following. The compliance check showed that the Bulgarian method fulfils the requirements of the WFD (Table 10).

Table 10: Compliance criteria for the assessment method based on fish in Bulgarian rivers in the Danubian Group

Compliance checking Compliance criteria conclusions 1. Ecological status is classified by one of five classes (high, good, Yes (see chapter 3.3) moderate, poor and bad). 2. High, good and moderate ecological status are set in line with the Yes (see chapter 6.5) WFD’s normative definitions (Boundary setting procedure) 3. All relevant parameters indicative of the biological quality element Yes (see chapter 3.3) are covered (see Table 1 in the IC Guidance). A combination rule to combine parameter assessment into BQE assessment has to be defined. If parameters are missing, Member States need to demonstrate that the method is sufficiently indicative of the status of the QE as a whole. 4. Assessment is adapted to intercalibration common types that are Yes (see chapter 2) defined in line with the typological requirements of the WFD Annex II and approved by WG ECOSTAT 5. The water body is assessed against type-specific near-natural Yes (see chapter 6.1) reference conditions 6. Assessment results are expressed as EQRs Yes (see chapter 3.3) 7. Sampling procedure allows for representative information about Yes (see chapter 3.2) water body quality/ ecological status in space and time 8. All data relevant for assessing the biological parameters specified Yes (see chapter 3.2 and 3.3) in the WFD’s normative definitions are covered by the sampling procedure 9. Selected taxonomic level achieves adequate confidence and Yes (see chapter 6.2) precision in classification

June 2016 Apostolou et al. (Consortium DICON – UBA) 20 \ 28 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish

Evaluation of the assessment concept The concept of TsBRI focuses on different community characteristics. The assessment method uses metrics as structural (total number of species, abundance, and biomass) along with functional (sensitive, migratory, dominant, predatory, alien species) metrics. Thus it is comparable in this respect with the methods of the finalized IC exercise. All assessment methods included in the finalized IC exercise as well as the Bulgarian national sampling method acquire the fish community data by following the principles of European standards. Sampling strategy is similar, according to EN 14011 protocol: multi-habitat sampling. Captured fish are identified to the lowest level possible (species level), aggregated as ecologic groups or traits and evaluated by their relative pro- portions in the sample. Hence, the Bulgarian method can be accepted and the IC exercise using the fit-in procedure is considered as feasible in terms of assessment concepts.

8 IC Procedure

Process of fit the BG method to the completed IC exercise Following Fig. 1 in the CIS Guidance No. 30 (European Union 2015), case A1 will be applied for the assessment method. The IC Manual lists the following steps for this case: 1. Calculate the common metric (CM) on the national dataset. 2. Use the associated pressure data to identify sites in the national dataset that meet the criteria established by the GIG for the selection of benchmark or reference sites. 3. Standardise the common metric (CM_bm) against the benchmark according to the approach used in the completed exercise. If benchmark standardisation was concluded not to be required in the completed exercise the mean CM value of the joining method’s benchmark sites must lie inside the range of mean values of the benchmark sites of the methods already intercalibrated for this conclusion to remain applicable. If the joining method’s benchmark sites lie outside of this range the joining method must benchmark standardise its sites relative to the global mean CM value of the benchmark sites included in the completed exercise. These scenarios are illustrated in Table 1 and 2 of the IC Manual. 4. Use OLS regression to establish the relationship between CM_bm (y) and the EQR of the joining method (x). A specialist case is that when a joining method relies exclusively on the common metric developed in the completed exercise for its classification rather than devising an original method (then being more like Option 1). In such cases a regression would be meaningless as y is directly dependent on x. The goal for an MS choosing to use the CM as the basis for their method is simple – after any benchmarking their boundaries must simply lie within one quarter of class of the global mean view.

Apostolou et al. (Consortium DICON – UBA) June 2016 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish 21 \ 28

5. Predict the position of the national class boundaries (MP, GM, HG and reference) on the CM_bm scale. 6. Apply the comparability criteria as summarised in Chapter 6 of the IC Manual. ad 1. Calculate the common metric (CM) on the national dataset The final report of the CrossGIG includes two versions of common metrics. In this report we apply the second version, which is applicable in the Lowland, Nordic and Danubian Group (see chapter 7.2.2 in European Union 2015). The common metric (Fish.Index) includes two metrics for all sites: Fish.Index = (Ni.O2.Intol + Ric.RHt.Par) / 2 Ni.O2.Intol =Density (number of individuals per 100 m2 in the first run of a sample site) of species intolerant to oxygen depletion, always more than 6 mg/l O2 in water Ric.RHt.Par = Richness (number fo species in the 1st rund of a sample site) of species requiring a rheophilic reproduction habitat, i.e. preference to spawn in running waters The common metric was calculated in the same way as in the finalized IC exercise. The model calculation was performed in R by Didier Pont. ad 2. Identify benchmark sites in the national dataset See above, chapter 6.4, Table 8. ad 3. Benchmark standardization According to the CrossGIG report, no benchmark standardisation was carried out. Hence, it has just to be guaranteed that the mean common metric value of the BG benchmark sites (CM_bench_BG) lies inside the range of mean values of the benchmark sites of the methods already intercalibrated (avgCM_bench_GIG) (Table 11). As this is the case (Table 12), no benchmark standardization is needed.

Table 11: Number of reference (= benchmark) sites in the three countries within the Danubian Group, which have finalized the intercalibration

MS Sites NDS NRS ERS IRS CZ 93 14 14 0 14 RO 143 34 14 0 14 SK 76 37 37 0 37 BG 137 15 23 0 5 NDS = non-disturbed sites, NRS = national reference sites, ERS = exceptional reference sites, IRS = intercalibration reference sites.

June 2016 Apostolou et al. (Consortium DICON – UBA) 22 \ 28 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish

Table 12: Values of the common metrics expressed in EQR for the reference sites of the different Danubian countries (Romania with 2 methods for salmonid and cyprinid rivers)

MS Min. 1st Qu. Median Mean 3rd Qu. Max. CZ 0.4269 0.9060 1.0770 0.9958 1.1520 1.2010 RO cypr 0.7445 0.8283 0.8922 0.9044 0.9684 1.0890 RO salm 0.7916 0.9418 0.9917 1.0220 1.1510 1.2130 SK 0.7270 0.9626 1.0880 1.0500 1.1370 1.2330 BG 0.8859 0.9499 0.9943 0.9728 1.0070 1.0270 ad 4. OLS regression The regression between the EQR values from the national assessment method TsBRI and the common metric Fish.Index is highly significant (Figure 6). The results of the regression model are given below. The slope and the correlation coefficients are above 0.5, the regression model is highly significant both for the intercept and the slope. x<-a4[a4$Country_code=="BG",] ## 77 96 77 lines summary(lm(x$IndexAll_Mean~x$BG_index)) regression common metrics against BG index

Residuals: Min 1Q Median 3Q Max -0.34395 -0.04616 0.00082 0.08026 0.32652

Coeff. Estimate Std. Error t value Pr(>|t|) Intercept 0.30075 0.06616 4.546 2.06e-05 *** Slope 0.70139 0.08241 8.511 1.27e-12 ***

Residual standard error: 0.1293 on 75 degrees of freedom Multiple R-squared: 0.4913, Adjusted R-squared: 0.4845 F-statistic: 72.43 on 1 and 75 DF, p-value: 1.271e-12 sqrt(0.4913) ## 0.700928 Correlation coefficient

Apostolou et al. (Consortium DICON – UBA) June 2016 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish 23 \ 28

Bulgaria Czech Republic

Romania Slovakia

Figure 6: OLS regression to establish the relationship between the classification of the national methods (x) and the common metric Fish.Index (y) in the Danubian Group of rivers

June 2016 Apostolou et al. (Consortium DICON – UBA) 24 \ 28 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish ad 5. Position of the national class boundaries on the common metric scale National class boundaries on the common metric scale were derived from the OLS regression above and are presented in Table 13.

Table 13: Translation of the reference and boundary positions of the national method on the basis of OLS regression (see Figure 6) into ICM

Method_country Type HG raw HG fit HG lwr HG upr GM raw GM fit GM lwr GM upr BG_index no 0.860 0.986 0.944 1.014 0.650 0.819 0.775 0.861 RO_EFIplus_index cypr_wading 0.939 0.937 0.906 0.951 0.800 0.813 0.733 0.844 RO_EFIplus_index salmonid 0.911 0.895 0.861 0.914 0.755 0.759 0.704 0.793 SK_FIS_index no 0.710 0.943 0.910 1.012 0.570 0.811 0.778 0.844 CZ_index no 0.780 0.928 0.894 0.967 0.585 0.791 0.757 0.826 Mean no – 0.926 – – – 0.794 – – Median no – 0.933 – – – 0.801 – – ad 6. Application of the comparability criteria High-good boundary bias calculation The H/G boundary of BG method translated into common metric scale (0.986) lies above the global mean view of the H/G boundary of all MS from the completed exercise (0.926). Therefore, there is no need to change the boundary. The deviation of the BG method minus 0.25 class width (0.944) and the global mean is low (0.018). The BG method for the H/G boundary is thus slightly stricter than the methods of the other countries of the Danubian group. The range of the BG method (translated into common metric scale) ±0.25 the class width overlaps with the ranges of all other methods except the Romanian method for salmonid rivers (Figure 7). There is no need to change the H/G boundary of the BG method.

Good-moderate boundary bias calculation The G/M boundary of BG method translated into common metric scale (0.819) lies slightly above the global mean view of the G/M boundary of all MS from the completed exercise (0.794). The range of the BG method ±0.25 class width, however, includes the global mean. There is no need to change the G/M boundary of the BG method.

Apostolou et al. (Consortium DICON – UBA) June 2016 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish 25 \ 28

Common metrics

H_G_fit

SK_FIS_index

RO_EFIplus_index

RO_EFIplus_index

CZ_index

BG_index

G_M_fit SK_FIS_index Figure 7: Class boundaries ±0.25 class RO_EFIplus_index width of the national methods on the RO_EFIplus_index common metric scale compared with

CZ_index the global mean of the finalized method (Romania with 2 methods for salmonid BG_index and cyprinid rivers) 0.0 0.5 1.0 1.5

The following Table 14 summarizes the results of the fit-in procedure for the national assessment method of Bulgaria using fish in rivers of the national river types R2, R4, R7 and R8 (IC Danubian group).

Table 14: Summary of boundary values

Boundary nEQR nEQR on cm scale Reference 1.00 – H / G 0.86 0.986 G / M 0.65 0.819 M / P 0.43 – P / B 0.22 –

Conclusions This report documents the fitting procedure of the Bulgarian (BG) fish-based assessment method for the national and common river types R2, R4, R7 and R8 to the results of the completed Cross GIG (Danubian Group) intercalibration exercise. IC feasibility and compliance of the presented assessment method and reported sufficient pressure-response relationships are documented and described. The sampling campaigns of the joint BG-AT project have enabled establishing an adequate dataset for the national IC types belonging in Danube basin. Based on the relationships and following the CIS Guidance Documents, reference values and class boundaries were

June 2016 Apostolou et al. (Consortium DICON – UBA) 26 \ 28 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish defined for the TsBRI, which is used for assessing four river types in the Danubian basin (belonging to Cross GIG – Danubian Group) using the BQE fish. Based on the new data, the national assessment method was compared with the finalized IC exercise of the Cross GIG (Danubian Group) following the fit-in procedure of CIS Guidance Document n° 30 (085/2015). The R scripts of the finalized IC exercise were used to perform the calculations of the common metric as well as all comparability criteria (carried out by the former GIG lead Didier Pont). No adjustment of the class boundaries was necessary. Following the criteria defined by the CIS Guidance Document n°30 (085/2015), the national assessment method of Bulgaria is considered comparable with the existing methods of the Danubian group. It is recommended to submit the method to the ECOSTAT group for official approval.

9 References

Белкинова Д., Гечева Г., Чешмеджиев С., Димитрова-Дюлгерова И., Младенов Р., Маринов М., Тенева И., Стоянов, П., Иванов П., Михов С., Пехливанов Л., Варадинова Е., Карагьозова Ц., Василев М., Апостолу А., Велков Б., Павлова М., 2013. Биологичен анализ и екологичната оценка на типовете повърхностни води в България. Университетско издателство „Паисий Хилендарски “ 235pp. ISBN 978-954-423-824-7. European Union (2015). Guidance Document No. 30. Procedure to fit new or updated classification methods to the results of a completed intercalibration exercise. Technical report – 2015 – 085. EC (2011). Guidance document on the intercalibration process 2008–2011. Guidance Document No. 14. Implementation strategy for the Water Framework Directive (2000/60/EC). European Commission, Technical report – 2011 – 045 European Union (2013). Commission decision of 20 September 2013 establishing, pursuant to Directive 2000/60/EC of the European Parliament and of the Council, the values of the Member State monitoring system classifications as a result of the intercalibration exercise and repealing Decision 2008/915/EC. Mihov S. 2010. Development of fish based index for assessing ecological status of Bulgarian Rivers (BRI). Biotechnol. & Biotechnol. Eq. 24/2010/Se, Special Edition/On-Line.

Apostolou et al. (Consortium DICON – UBA) June 2016 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish 27 \ 28

10 Annex

Annex I. Type-specific abiotic characteristics for the national river types R2, R4, R7, R8: ER = epirhithral, MR = metarhithral, HR = Hyporhithral, EP = Epipotamal.

Ecoregion EC 12 - Danube watershed EC 12 - Black Sea National Type R2 R2 + R4 R4 R8 R7 R2 R4 Biocoenotic ER MR HR HR/EP EP MR HR region 200- 150- Altitude (m) 600-1200 400-750 50-200 20-150 300-500 50-400 500(600) 300 10-11 Temperature, avg 6-9 °C 8-10 °C 10-11 °C 10-12 °C 10-12 °C 8-10 °C 10-13 °C °C 8- 8- 10- Slope 12-50‰ (1)2-12‰ 2-8(15)‰ 0.5-5‰ 20(40) 2-10‰ 20(40)‰ 20‰ ‰ Catchment area 10- 50- 100- (20)100- 1000- (50)100- 10-100 10-100 (km2) 100(400) 400(1000) 1000(5000) 1000 10000 1000 River width <10 m <20 m 5-50 m <25 m 10-100 m <10 m <10 m 5-20 m Substrate, main gravel gravel gravel sand sand gravel gravel gravel

Annex II. Type-specific communities for the national river types R2, R4, R7, R8: ER = epirhithral, MR = metarhithral, HR = hyporhithral, EP = epipotamal. The presumed dominance in classified in three classes: +++ ... dominant (“Leitart”), ++ ... typical accompanying species, + ... rare species.

Ecoregion 12 Watershed DR BS Type R2 R4 R6 R7 R8 R2 R4 Fish zone ER MR MR HR EP EP HR/EP MR MR HR +++ 1 2 2 3 7 4 4 1 1 4 ++ 1 3 4 4 16 5 4 4 4 2 + 0 5 7 10 33 26 13 3 3 7 sum 2 10 13 17 56 35 21 8 8 13 Scientific names

Abramis ballerus ++ +

Abramis brama +++ + +

Abramis sapa +++

Acipenser gueldenstaedtii +

Acipenser nudiventris +

Acipenser ruthenus +

Acipenser stellatus +

Alburnoides bipunctatus ++ ++ +++ + ++ ++ ++ +++

Alburnus alburnus ++ +++ +++ +++ +

Alburnus chalcoides +

Alosa immaculata +

Alosa tanaica +

Aspius aspius ++ +

Barbatula barbatula + + + + + + + +

Barbus barbus ++ ++ +++ ++

Barbus bergi ++ +++ +++

Barbus petenyi +++ +++ +++ ++ +++

Benthophilus nudus +

Blicca bjoerkna +++ + +

June 2016 Apostolou et al. (Consortium DICON – UBA) 28 \ 28 Fit-in procedure: Cross GIG (Danubian Group) in Bulgaria, BQE: Fish

Carassius carassius +

Carassius gibelio + +

Chondrostoma nasus + + ++ +

Cobitis elongata + + +

Cobitis elongatoides + + + ++ ++ ++

Cobitis pontica + + ++

Cottus gobio ++ + +

Cyprinus carpio ++

Esox lucius + + + ++ + +

Eudontomyzon mariae + +

Gasterosteus aculeatus +

Gobio gobio ++ ++ ++ ++ ++ + + +

Gobio kovatchevi ++ ++ ++

Gymnocephalus baloni +

Gymnocephalus cernuus + +

Gymnocephalus schraetser ++

Huso huso +

Leucaspius delineatus +

Leuciscus idus ++ +

Lota lota + +

Misgurnus fossilis + +

Neogobius fluviatilis +++ +

Neogobius gymnotrachelus + +

Neogobius kessleri + +

Neogobius melanostomus + +

Pelecus cultratus ++

Perca fluviatilis + ++ + + +

Perccottus glenii +

Petroleuciscus boristhenicus +

Phoxinus phoxinus + ++ ++ + +++ +++ +

Proterorhinus semilunaris ++ + + +

Pungitius platygaster +

Rhodeus amarus + ++ ++ +++ +++ + + +++

Romanogobio kessleri + +

Romanogobio uranoscopus + +

Romanogobio vladykovi +++ + +

Rutilus frisii +

Rutilus rutilus +++ +

Sabanejewia balcanica ++ ++ + + +

Sabanejewia bulgarica +

Salmo labrax + +

Salmo trutta +++ + +

Sander lucioperca ++ + +

Sander volgensis +

Scardinius erythrophthalmus + +

Silurus glanis ++ + +

Squalius cephalus +++ +++ +++ + +++ +++ ++ ++ +++

Syngnathus abaster +

Tinca tinca + +

Umbra krameri +

Vimba tenella +

Vimba vimba + ++ ++ +

Zingel streber + +

Zingel zingel ++ +

Apostolou et al. (Consortium DICON – UBA) June 2016