Acta Botanica 1-2017 - za web.indd 80 ACTA BOT. CROAT. 76 (1), 2017 Acta Bot. Croat. 76 (1), 80–90, 2017 CODEN: ABCRA 25 DOI: 10.1515/botcro-2016-0051 ISSN 0365-0588 eISSN 1847-8476 Do benthic diatom assemblages refl ect abiotic typology: a case study of Croatian streams and rivers Koraljka Kralj Borojević1, Marija Gligora Udovič1*, Petar Žutinić1, Gábor Várbíró2, Anđelka Plenković-Moraj1 1 University of Zagreb, Faculty of Science, Department of Biology, Rooseveltov trg 6, HR-10000 Zagreb, Croatia 2 MTA Centre for Ecological Research, Danube Research Institute, Department of Tisza River Research, Bem sqr. 18/c, Debrecen, H-4026, Hungary Abstract – Benthic diatoms are widely used in Europe and worldwide to access ecological status of running waters. One of key goals of Water Framework Directive is to classify rivers and streams using biological quality elements and type specifi c reference conditions. According to system B which incorporates additional abiotic descriptors, there are 24 water types in Croatia. For biological analyses 92 rivers and streams with 140 sampling points were chosen and sampled for benthic diatoms and water chemistry simultaneously. Self orga- nizing map (SOM) analysis was used to defi ne biotypes from species composition and abundance of benthic diatoms. Grouping of samples in SOM resulted in 10 distinctive groups. Based on their geographical position and site characteristics, groups represent sites with similar properties (as waterbed, catchment size, altitude, size of stream) belonging to different ecoregions in Croatia. Analysis of variance revealed statistically signifi - cant differences (p<0.05) among SOM groups concerning ammonia, nitrates and total phosphorus. Indicator species analysis (IndVal) singled out species that were signifi cantly characteristic (p<0.05) for SOM and abi- otic types. Compared to abiotic groups, in which 7 out of 24 have no indicator species, all SOM groups have one or several characteristic diatom species, thus indicating diatom assemblages as valuable site descriptors. Canonical analysis of principal coordinates analysis also indicated that SOM grouping of samples is statisti- cally reliable. Grouping of similar sites, although placed into different abiotic types, makes SOM groups with its corresponding representative species an easy tool for water quality assessment and description of reference assemblage. Keywords: benthic diatoms, self organizing map, Water Framework Directive, water typology. Abbreviations: BMU – best matching unit, SOM – self organizing map, * Corresponding author, e-mail: marija.gligora.udovic@biol.pmf.hr Introduction According to Water Framework Directive (WFD), Eu- ropean Union member states have to achieve a good eco- logical status for all streams whose catchment area exceeds 10 km2. Since the assessment is to be performed by observ- ing deviations from the reference conditions, one of the key goals of the Water Framework Directive was to classify riv- ers and streams using biological quality parameters and to describe hydromorphological, physicochemical and bio- logical type specifi c reference conditions (EC 2000). That should be determined using fi ve biological quality ele- ments: phytoplankton, phytobenthos, macrophytes, benthic invertebrates and fi sh. As commonly dominant phytoben- thic representatives, benthic diatoms are used worldwide to assess the ecological status of running waters (Reid et al. 1995, Ács et al. 2005, Kireta et al. 2012, Martin and Reyes Fernandez 2012, Kahlert et al. 2016). They are widespread and can be found in almost any type of running water which, combined with short generation time and ability to clearly defi ne nutrient status of their habitat, makes them great indicators of water quality. Also, they are relatively easy to sample and their ecological valences and habitats have been known for more than 100 years (Kolkwitz and Marsson 1908). During the past decades numerous indices, mostly based on trophic and saprobic status, have been used specifi cally for that purpose (Sládeček 1986, Kelly and Whit- ton 1995). Unlike the previously set index class boundaries, water quality assessment according to WFD is performed by comparison to reference conditions. As different habitats DO BENTHIC DIATOM ASSEMBLAGES REFLECT ABIOTIC TYPOLOGY ACTA BOT. CROAT. 76 (1), 2017 81 in their undisturbed stages support different assemblages, it is essential to describe reference condition for every type of habitat (Bennion et al. 2014). As a starting point, all water bodies should be classifi ed according to WFD hierarchical water body typology. It is organized by fi rstly placing the surface water bodies into the broad categories such as rivers, lakes, transitional/coast- al bodies, etc. Two typologies were acknowledged within those categories: “System A” fi xed typology, and “System B” alternative typology, which comprises a mixture of obligatory and optional factors (like catchment size, eleva- tion, fl ow velocity, granulometric properties of the water- bed, tufa forming conditions, permanence of fl ow, etc.). Abiotic typology is then used as a base for describing type specifi c reference conditions. Eventually, water quality is determined by a comparison of the actual conditions to the ones that have been referenced. Using water type system B, 24 different water types were discerned in Croatia (Mihaljević et al. 2011). Types were fi rst separated by region (continental, pannonian and coastal), then by catchment size, type of waterbed, altitude of stream fl ow and stream fl ow discharge which is deter- mined from the size, altitude and slope (Tab. 1). Such a high number of types is a refl ection of geographical posi- tion and shape of the country of which two thirds belong to karst infl uenced by high mountains separating the coastal area from the continent, along with a rather fl at Pannonian area which is extensively used for agriculture. Croatia has scarce and non-continuous data on attached diatoms (Plenković-Moraj 1995) with parts of the country, especially poorly populated areas, that have never been ex- plored. This study yielded the fi rst comprehensive diatom dataset on most waterbodies in Croatia. The aim of this paper is to investigate relationship between previously determined 24 types of rivers and streams in Croatia and potential bio- types that are derived from benthic diatom assemblage data. Considering the number of types in neighbouring countries and previous data, it is expected that diatom assemblages will show regional differences along with some differences from other factors, but will altogether group certain abiotic types and thus reduce them. Materials and methods Sampling design Data on diatom species composition, in the form of pro- portion of total diatom valves in each sample, has been gathered within the two projects for nation-wide survey covering all relevant important types of running waters along with concurrent and relevant environmental variables. The fi nal goal of those projects was to describe water qual- ity elements for determination of water class category ac- Tab. 1. Abiotic water types (codes from AB1-AB24) in Croatia according to System B with factor values. Abiotic type Altitude / m Catchment / km2 Discharge (Q / m3 s–1) Region Waterbed AB1 >600 10–100 Q<2 Pannonian silicate AB2 200–600 10–100 Q<2 Pannonian carbonate-silicate AB3 <200 10–100 Q<2 Pannonian silicate AB4 <200 100–1000 220 Pannonian silicate AB6 <200 1000–10000 Q>20 Pannonian silicate AB7 <200 1000–10000 Q>20 Pannonian carbonate AB8 <200 1000–10000 Q>20 Pannonian silicate AB9 >600 10–100 Q<2 Continental carbonate AB10 >600 10–100 Q<2 Continental carbonate AB11 200–600 10–100 Q<2 Continental carbonate AB12 200–600 100–1000 220 Continental carbonate-silicate AB16 200–600 10–100 Q<2 Coastal carbonate AB17 <200 10–100 Q<2 Coastal carbonate AB18 200–600 10–100 220 Coastal carbonate AB21 <200 1000–10000 Q>20 Coastal carbonate-silicate AB22 200–600 100–1000 210000 km2 – very large rivers (4); type of waterbed with silicates (1), carbon- ates (2), organic (3) and mixed (4); altitude of stream’s fl ow – lowland streams <200 m (1), submountain streams 200– 600 m (2) and mountain streams >600 m (3); stream fl ow discharge (determined from the size, altitude and slope) as 3 groups: (1) low (Q<2 m3 s–1), (2) medium (2 m3 s–120 m3 s–1). The assemblage data for each taxon were normalized before the rescaling process by logarithmic transformation [log (x=1)]. The SOM consists of two layers: an input layer containing the samples and its variables, and an output layer containing so-called virtual units (VU). A VU can be considered a virtual sample unit with its virtual set of variables. In selection phase, weights of output layer were randomly assigned. Then a sample unit was chosen randomly and a best matching unit (BMU) was selected by calculating the Euclidean distance between the weights of the input layer and the weights of the output lay- er. It is important to note that the selection of the BMUs was based exclusively on diatom species abundances and environmental variables were masked out. This was done by applying a mask function that assigns a weight of 1 to the biological variables used for the selection of the SOMs VU and a weight of 0 to environmental variables. In the learning phase, the BMU weights (variables) in the output layer defi ned in the selection phase were updated with the weights of the input unit. The update process included both biological and environmental weights. The BMU was not the only grid unit updated: a neighborhood was defi ned around the BMU and all units within this neighborhood were also updated. In this way the values of environmental variables can also be visualized on an SOM that was previ- ously trained, but only with the biological variables (Céré- ghino and Park 2009). The resulting hexagon map with its weights was visualized using the SOM toolbox as compo- nent planes. Each component plane represents variables that the SOM algorithm has learned. The SOM toolbox (http://www.cis.hut.fi /projects/somtoolbox) was used to im- plement the SOM in a MATLAB™ environment. The de- tailed algorithm of the SOM can be found in Kohonen (2001) and Lek and Guégan (1999) for ecological applica- tions. To divide the output SOM into different diatoms groups according to their similarity, a hierarchical cluster analysis of the Ward linkage method with Euclidean dis- tances was used (Park et al. 2004, 2006). After clustering the SOM the BMUs of the original data matrix were calculated, thus we could obtain in which clus- ter a given sampling event would occur. Species which were important and stable for a given cluster were obtained using the indicator value index (IndVal) according to Du- frene and Legendre (1997) based on the original data ma- trix. Statistical analysis To test the differences among assigned groups (both bi- otic and abiotic), one-way analyses of variance (ANOVA) with abiotic and SOM groups as grouping factors, followed by post-hoc Tukey honest signifi cant difference (HSD) tests were employed with a statistically signifi cant value of p<0.05. ANOVAs were performed using statistical program Statistica 9.0. Primer 6 (Clarke and Warwick 2001) was used to allocate physical and chemical parameters that had the greatest impact on the assemblage. Firstly, the Bio-Env procedure which allocates the combination of environmen- tal variables explaining the highest proportion of variance within the assemblage was used. For the analysis both Bray-Curtis similarity matrix of biotic data and of normal- ized environmental data were used. Afterwards, canonical analysis of principal coordinates (CAP) was performed. Its purpose was to fi nd axes through multivariate cloud of points that either: (i) are the best at discriminating among a priori groups or (ii) have the strongest correlation with some other set of variables. Discriminant analysis of CAP DO BENTHIC DIATOM ASSEMBLAGES REFLECT ABIOTIC TYPOLOGY ACTA BOT. CROAT. 76 (1), 2017 83 procedure was used to identify proportion of the samples that were correctly allocated to both SOM and abiotic types used as nominal factors of initial matrix. The other way of using CAP is to analyze how well multivariate data predict positions of samples along a continuous or quantitative gra- dient (Anderson et al. 2008). For discriminant analysis only Bray-Curtis similarity matrix of biotic data was used, and for canonical analysis, a matrix with normalized environ- mental data was used additionally. Results Typology Grouping of samples in SOM (Online suppl. Fig. 1) re- sulted in 10 distinctive biotypes. Based on their geographi- cal position and site characteristics, groups represent sites with similar properties (such as waterbed, catchment size, altitude, size of stream) belonging to different ecoregions of Croatia (Fig. 1). Most of the samples of group D1 are small to medium sized streams on higher altitudes on carbonate waterbed in continental region. Groups D2, D3 and D4 combine small to medium streams on lower (group D3) and higher altitudes (groups D2 and D4) on silicate (group D2) or carbonate waterbed (groups D3 and D4) mostly in coast- al region. Groups D5, D6, D8 and D10 combine streams on silica waterbed in continental and pannonian region (group D5, D6, D8, D10) – they differ in size of stream and its alti- tude. Sites from groups D7 and D9 belong to coastal region and combine medium to large lowland rivers (group D7) or small streams in higher altitudes (group D9) on karstic car- bonate waterbed (Online Suppl. Tab. 1). Environmental variables One-way ANOVA revealed that SOM groups statisti- cally signifi cantly differ (F=2.16, p<0.001) concerning com- bination of all environmental variables. To single out vari- ables that contributed mostly towards statistical signifi cance, post-hoc Tukey HSD test was performed. It revealed that group 6 is statistically different (p<0.05) from all other groups concerning concentrations of ammonia, nitrates, or- thophosphates and total phosphorus. Infl uence of nominal variables that were the base for abiotic typology can be seen in component planes of the SOM for the typological attribute categories (Fig. 2). Fig. 1. Map of Croatian investigation sites divided into diatom-based self organizing map (SOM) groups. SOM cluster groups are as- signed with codes from D1–D10. Lines denote different ecoregions of Croatia. KRALJ BOROJEVIĆ K., GLIGORA UDOVIČ M., ŽUTINIĆ P., VÁRBÍRÓ G., PLENKOVIĆ-MORAJ A. 84 ACTA BOT. CROAT. 76 (1), 2017 Among abiotic groups ANOVA also revealed statisti- cally signifi cant differences (F=1.59, p<0.001), but post-hoc Tukey HSD test did not single out any particular parameter for any of the groups to be statistically signifi cant (thresh- old: p<0.05). Biotic data One-way ANOVA on diatom abundance data set re- vealed that for both typologies their corresponding groups statistically signifi cantly differ one from another (F=138.08, p<0.001 and F=2.4, p<0.001, for SOM and abiotic groups, respectively). IndVal analysis was employed to discover which of the variables species have signifi cant indicator value for each of the groups. The analysis confi rmed spe- cies with highest abundance and/or frequency within each SOM group as signifi cant representatives of each group (Tab. 2), which can also be seen from SOM component planes (Fig. 3). Unlike SOM groups, some of the abiotic groups (groups AB4, AB8, AB9, AB12, AB15, AB18 and AB19) were not represented by any of the species (Tab. 3). Tab. 2. Self organizing map (SOM) groups with indicator species according to indicator species analysis (IndVal). Abundance is repre- sented as average percentage (%) of total number of valves of particular species for each site belonging to that group ± standard deviation. Cluster CODE Species name Abundance IndVal p-value D1 ADBI Achnanthidium biasolettianum (Grunow) Bukhtiyarova 74.42±4.09 0.64 <0.001 D2 ACAF Achnanthidium affi ne (Grunow) Czarnecki 7.36±3.18 0.41 <0.001 CPLA Cocconeis placentula Ehrenberg 23.51±5.37 0.34 <0.001 D3 CAFF Cymbella affi nis Kützing 3.55±0.99 0.23 0.01 CALS Caloneis sp. 0.09±0.06 0.13 0.05 CYMS Cymbella sp. 1.32±0.65 0.27 0.01 NPAL Nitzschia palea (Kützing) W.Smith 2.45±0.71 0.23 0.02 NRHY Navicula rhynchocephala Kützing 0.22±0.12 0.19 0.01 SPHO Stauroneis phoenicenteron (Nitzsch) Ehrenberg 0.04±0.03 0.13 0.05 SRMA Staurosira martyi (Heribaud) Lange-Bertalot 0.56±0.31 0.25 0.01 SSMU Staurosira mutabilis (W.Smith) Pfi tzer 1.67±1.02 0.25 0.01 D4 ENCM Encyonopsis microcephala (Grunow) Krammer 6.8±1.93 0.42 <0.001 D5 CMTS Cymatopleura sp. 0.04±0.03 0.20 <0.001 GYAC Gyrosigma acuminatum (Kützing) Rabenhorst 1.54±0.78 0.23 <0.001 GYRS Gyrosigma sp. 0.17±0.11 0.19 <0.001 NAMP Nitzschia amphibia Grunow f. amphibia Grunow 0.36±0.27 0.19 <0.001 OROE Orthoseira roeseana (Rabenhorst) O’Meara 2.19±1.99 0.20 <0.001 PTDS Planothidium sp. 0.76±0.46 0.20 <0.001 D6 AUGR Aulacoseira granulata (Ehrenberg) Simonsen 2.75±1.19 0.30 <0.001 CYLS Cyclotella sp. 7.68±4.58 0.51 <0.001 FPYG Fallacia pygmaea (Kützing) Stickle & Mann ssp. pygmaea Lange-Bertalot 7.04±2.69 0.56 <0.001 GYAT Gyrosigma attenuatum (Kützing) Rabenhorst 1.02±0.61 0.40 <0.001 HHUN Hippodonta hungarica (Grunow) Lange-Bertalot, Metzeltin & Witkowski 0.56±0.39 0.34 <0.001 NACI Nitzschia acicularis (Kützing) W.Smith 15.23±4.60 0.53 <0.001 NVLC Nitzschia valdecostata Lange-Bertalot et Simonsen 2.14±1.05 0.40 <0.001 SHAN Stephanodiscus hantzschii Grunow 10.7±6.59 0.62 <0.001 D7 ADMI Achnanthidium minutissimum (Kütz.) Czarnecki 59.95±7.32 0.31 <0.001 ECAE Encyonema caespitosum Kützing 4.82±2.57 0.53 <0.001 ESLE Encyonema silesiacum (Bleisch) D.G.Mann 2.55±1.58 0.30 <0.001 Fig. 2. Component planes of the self organizing map (SOM) for the typological attribute categories. Darker cells mean higher val- ues of the given attributes. DO BENTHIC DIATOM ASSEMBLAGES REFLECT ABIOTIC TYPOLOGY ACTA BOT. CROAT. 76 (1), 2017 85 Cluster CODE Species name Abundance IndVal p-value D7 FRAS Fragilaria sp. 0.29±0.15 0.13 0.05 FSAP Fistulifera saprophila (Lange-Bertalot & Bonik) Lange-Bertalot 0.37±0.17 0.43 <0.001 GTRU Gomphonema truncatum Ehrenberg 1.13±0.96 0.37 <0.001 D8 AFOR Asterionella formosa Hassall 14.19±4.61 0.69 <0.001 APEL Amphipleura pelucida Kützing 1.06±0.32 0.29 <0.001 ATRI Achnanthes trinodis (W.Smith) Grunow 1.68±0.74 0.32 <0.001 CCMP Cymbella compacta Østrup 0.82±0.44 0.27 <0.001 CLBE Cymbella lange-bertalotii Krammer 3.05±1.15 0.38 <0.001 DOBL Diploneis oblongella (Nägeli ex Kützing) Cleve-Euler 7.74±2.53 0.65 <0.001 DVCA Diatoma vulgaris var. capitulatum Grunow 14.07±3.91 0.63 <0.001 EPRO Encyonema prostratum (Berkeley) Kützing 5±1.55 0.53 <0.001 ETUR Epithemia turgida Kützing 0.68±0.31 0.21 <0.001 FARC Fragilaria arcus (Ehrenberg) Cleve 2.23±1.81 0.34 <0.001 GDEC Geissleria decussis (Østrup) Lange-Bert. & Metzeltin 1.49±0.57 0.43 <0.001 SSPE Staurosira sp. 0.7±0.27 0.35 <0.001 D9 CPED Cocconeis pediculus Ehrenberg 6.41±1.82 0.21 0.04 ENVE Encyonema ventricosum (C.Agardh) Grunow 1.45±0.42 0.19 0.03 D10 ADSU Achnanthidium subatomus (Hustedt) Lange-Bertalot 11.01±5.38 0.19 0.01 CRBU Craticula buderi (Hustedt) Lange-Bertalot 3.88±2.30 0.22 0.03 FVUL Frustulia vulgaris (Thwaites) De Toni 0.12±0.09 0.15 0.03 NMEN Navicula menisculus Schumann var. menisculus 5.98±2.52 0.25 <0.001 NUMB Nitzschia umbonata (Ehrenberg) Lange-Bertalot 1.27±0.45 0.38 <0.001 SBRE Surirella brebissonii Krammer & Lange-Bertalot var. brebissonii 1.81±0.96 0.20 0.02 STAN Stauroneis anceps Ehrenberg 0.05±0.04 0.15 0.03 Tab. 2. – continued Tab. 3. Abiotic groups with indicator species according to indicator species analysis (IndVal). Average represents percentage (%) of total number of valves of particular species for each site belonging to that group. STD – standard deviation. Abiotic group Species Average STD p-value AB1 Achnanthidium affi ne (Grunow) Czarnecki 8.4 0.101 <0.01 Cocconeis placentula Ehrenberg 30.0 0.202 <0.05 Craticula cuspidata (Kützing) Mann 5.8 0.115 <0.05 Cymbella tumida (Brébisson) Van Heurck 3.6 0.072 <0.05 Gomphonema gracile Ehrenberg 3.0 0.060 <0.05 AB2 Diatoma tenuis Agardh 0.5 0.012 <0.05 Navicula trophicatrix Lange-Bertalot <0.1 0.001 <0.05 Sellaphora pupula (Kützing) Mereschkowksy 4.8 0.081 <0.05 AB3 Achnanthidium subatomus (Hustedt) Lange-Bertalot 18.5 0.227 <0.05 AB5 Aulacoseira granulata (Ehrenberg) Simonsen 2.3 0.033 <0.05 Gyrosigma sp. 0.2 0.003 <0.05 AB6 Asterionella formosa Hassall 35.3 0.483 <0.05 Cymbella stuxbergii (Cleve) Cleve 1.2 0.017 <0.01 Fallacia pygmaea (Kützing) Stickle & Mann 12.3 0.173 <0.05 AB7 Asterionella formosa Hassall 32.7 0.463 <0.01 AB10 Encyonema ventricosum (Agardh) Grunow 5.9 <0.05 Nitzschia dissipata (Kützing) Grunow 4.8 <0.05 Nitzschia sp. 1.2 <0.01 AB11 Achnanthidium biasolettianum (Grunow) Bukhtiyarova 62.2 0.400 <0.05 Aulacoseira ambigua (Grunow) Simonsen 0.2 0.004 <0.05 Cymbella subhelvetica Krammer 0.1 0.003 <0.05 Nitzschia linearis W.Smith 0.2 0.004 <0.05 KRALJ BOROJEVIĆ K., GLIGORA UDOVIČ M., ŽUTINIĆ P., VÁRBÍRÓ G., PLENKOVIĆ-MORAJ A. 86 ACTA BOT. CROAT. 76 (1), 2017 Bio-Env procedure showed that the main environmental variables describing the data set were nitrogen fractions NH4+ and NO3– and total P (Rho=0.235; signifi cance level of sam- ple statistic=0.1%, number of permutations performed 999). CAP procedure in Primer that corresponds to canonical analysis of principal coordinates (to test correlation to en- vironmental variables) did not completely confi rm those results (Fig. 4). The results show strong and signifi cant cor- relations between the diatom abundance and the en vi ron- mental variables (p=0.001). First canonical correlation was 0.74 and the second one was 0.67. CAP axis 1 was corre- lated with oxygen saturation (r=–0.924) and CAP axis 2 with total P (r=0.571) and NO3– (r=0.490). Discriminant CAP analysis showed that 10 SOM groups are indeed distinguishable one from another. First two ca- nonical correlations are quite high (δ1=0.90 and δ2=0.76). Diagnostic showed that the choice of 3 PCA axes includes 34.29% of total variation, i.e. that many samples were cor- rectly classifi ed, with values of Q0m’HQ0m=3.17 (p=0.001) and δ12=0.81 (p=0.001). Discriminant CAP on abiotic groups showed that abiot- ic typology is also relevant – with fi rst two canonical corre- lations being 0.91 and 0.74 (δ1 and δ2, respectively). Al- though permutation test revealed high statistical signifi cance (Q0m’HQ0m=4.85 (p=0.001) and δ12=0.84 (p=0.001), abiotic grouping correctly allocated only 19.29% of samples. Tab. 3. – continued Abiotic group Species Average STD p-value AB13 Didymosphenia geminata (Lyngbye) Mart.Schmidt 0.1 0.002 <0.05 Encyonema ventricosum (Agardh) Grunow 2.7 0.028 <0.05 Gomphonema olivaceum (Hornemann) Brébisson 3.9 0.059 <0.05 Navicula slesvicensis Grunow 0.5 0.010 <0.05 Nitzschia palea (Kützing) W.Smith 7.5 0.110 <0.05 Surirella ovalis Brébisson 0.1 0.001 <0.05 AB14 Aulacoseira italica (Ehrenberg) Simonsen 7.4 0.127 <0.01 Campylodiscus noricus Ehrenberg 0.1 0.003 <0.01 Fragilaria crotonensis Kitton 6.1 0.106 <0.01 Ulnaria ulna var. acus (Kützing) Lange-Bertalot 7.9 0.137 <0.01 AB16 Gomphonema acuminatum Ehrenberg 0.2 0.005 <0.05 AB17 Cymbella compacta Østrup 2.6 0.046 <0.01 Denticula tenuis Kützing 8.4 0.135 <0.05 Epithemia argus (Ehrenberg) Kützing 0.1 0.001 <0.01 Gomphonema pala Reichardt 0.2 0.003 <0.01 Gomphonema subclavatum Grunow 0.2 0.004 <0.05 Navicula hofmanniae Lange-Bertalot 6.0 0.056 <0.05 AB20 Cyclotella ocellata Pantocsek 5.6 0.083 <0.05 Navicula cryptotenelloides Lange-Bertalot 2.9 0.051 <0.01 AB21 Cyclotella ocellata Pantocsek 4.0 0.080 <0.05 Encyonema caespitosum Kützing 5.9 0.119 <0.05 Melosira arenaria Moore ex Ralfs 12.5 0.250 <0.05 Urosolenia eriensis (H.L.Smith) Round & R.M.Crawford 0.2 0.004 <0.05 AB22 Encyonopsis microcephala (Grunow) Krammer 27.8 <0.01 Sellaphora minima (Grunow) Mann 1.2 <0.01 AB23 Epithemia sp. 0.6 0.012 <0.05 Frustulia creuzburgensis (Krasske) Hustedt 1.7 0.034 <0.05 Nitzschia palea (Kützing) W.Smith 5.4 0.063 <0.05 Stauroneis smithii Grunow 0.3 0.007 <0.05 AB24 Navicula antonii Lange-Bertalot 2.4 0.041 <0.01 Navicula weinzierlii Schimanski 1.0 0.018 <0.05 Reimeria sinuata (Gregory) Kociolek & Stoermer 0.4 0.008 <0.05 Surirella angusta Kützing 0.3 0.005 <0.05 Surirella linearis W.Smith 0.1 0.001 <0.05 Surirella ovalis Brébisson 0.1 0.001 <0.05 Surirella sp. 4.1 0.071 <0.05 DO BENTHIC DIATOM ASSEMBLAGES REFLECT ABIOTIC TYPOLOGY ACTA BOT. CROAT. 76 (1), 2017 87 Finally, all investigated sites plotted according to envi- ronmental variables and biotic data were effi ciently associ- ated with both biotic (SOM) and abiotic groups (Tab. 4). Discussion Determination of water quality according to WFD great- ly depends on reliable typifi cation of waterbodies, which includes entire range of hydrological, geographical, physi- cal, chemical and biological characteristics of sampling sites. Since the status itself is derived by comparison to ref- erence conditions, it is essential to have all sites and their corresponding reference assemblages described and grouped into usable groups. As abiotic typology in Croatia resulted in 24 different types, which in relation to surface area seem rather high, especially in comparison to Germany (20 types, Schmedtje et al. 2001), Hungary (26 types, Van Dam et al. 2007), Great Britain (14 types, Davy-Bowker et al. 2006) or Sweden (11 types, Davy-Bowker et al. 2006), it was ex- pected for some types to group together. SOM analysis re- vealed 10 groups according to diatom assemblages, which is a number of types comparable to other countries. As shown in SOM component planes, type of waterbed, altitude, size of catchment area and type of stream (its size) were variables infl uencing the development of diatom as- semblage, as also shown in the other studies (Biggs 1995, Pan et al. 2000, Potapova and Charles 2003). All of those variables are in a direct relation to measured physical and chemical parameters of an investigated site, although there Tab. 4. Relationship among benthic diatom-based SOM (self organizing map) groups and abiotic typology for Croatian rivers and streams. Sum represents a number of sites belonging to the group/site. Croatian abiotic type sumAB 1 AB 2 AB 3 AB 4 AB 5 AB 6 AB 7 AB 8 AB 9 AB 10 AB 11 AB 12 AB 13 AB 14 AB 15 AB 16 AB 17 AB 18 AB 19 AB 20 AB 21 AB 22 AB 23 AB 24 SO M ty pe D1 1 1 1 1 2 3 1 1 1 1 13 D2 3 1 2 1 2 9 D3 1 1 1 1 1 4 1 2 3 1 16 D4 1 1 2 2 2 2 1 5 1 2 1 1 21 D5 2 1 2 1 1 1 1 1 10 D6 1 1 3 1 3 9 D7 1 1 1 1 1 1 1 1 1 9 D8 3 2 1 1 9 1 1 1 1 20 D9 1 1 1 1 1 2 1 1 1 2 1 2 1 1 2 1 20 D10 2 6 2 1 1 1 13 sum 4 6 12 10 11 2 2 17 10 1 5 3 4 3 11 4 3 11 6 3 4 1 4 3 140 Fig. 3. Component planes of the self organizing map (SOM) for characteristic diatom species. SOM cluster groups are assigned with codes from D1–D10. Darker cells mean relative higher abun- dance of the given diatom species. Abbreviations of diatom names are the same as in Tables 2 and 3. Fig. 4. Canonical analysis of principal coordinates (CAP) ordina- tion plot relating benthic diatom assemblages to environmental variables. KRALJ BOROJEVIĆ K., GLIGORA UDOVIČ M., ŽUTINIĆ P., VÁRBÍRÓ G., PLENKOVIĆ-MORAJ A. 88 ACTA BOT. CROAT. 76 (1), 2017 are other variables, such as type of vegetation, size of sub- strate or exposure, that also greatly infl uence the nutrient availability. Altitude is usually related to fl ow velocity which also greatly infl uences the development of the as- semblage (Primc-Habdija et al. 2001). Size of the catch- ment area and size of the stream, in a way, refl ect the num- ber of tributaries which bring additional nutrients to the stream. Subsequently, it also extends the possible presence of settlements such as industrial zones and agricultural es- tates which greatly infl uence nutrient infl ow. Increasing phosphorus levels can be mainly related to those variables and not exclusively to the natural sources phosphorus lev- els. As published in many other studies (Bothwell 1989, 1988, Stanley et al. 1990) phosphorus, both as orthophos- phate and total phosphorus, was shown to be an important factor in the development of benthic diatom assemblage. Phosphorus and nitrogen fractions have been generally con- sidered to be most critical variables (Hutchinson 1957), as shown in Bio-env and CAP procedures. Composition of diatom assemblages and the character- istic (indicator) species as Achnanthidium, Cymbella, Ency- onopsis or Nitzschia species, from this set of Croatian sam- ples closely corresponded to those observed in other geographical areas (Tison et al. 2005, Park et al. 2006, Tornés et al. 2007, Dohet et al. 2008, Tornés et al. 2012). Diatom assemblages are infl uenced by many variables in- cluding those that are site specifi c at various temporal and special scales (DeNicola et al. 2004, Pan et al. 2004), as well as those that refl ect human interventions in the envi- ronment. Statistically, both types of typology proved to be signifi - cant, but abiotic typology resulted in several groups being without one or several discriminant species. On the other hand, SOM groups had several characteristic species for each of the groups. Typical species for SOM group D1, small to medium, mountain, continental streams on carbonate waterbed, Ach- nanthidium biasolettianum (Grunow) Bukhtiyarova is a species reported from upstream, low human impact sites at siliceous streams (Soininen et al. 2004). Presence of A. bi- asolettianum in carbonate streams can be explained by the fact that all streams from group D1 belong to continental part of Croatia that is not karstic and carbonates are prevail- ing but are not exclusive in the waterbed. Similar sized streams, but on carbonate mostly karst waterbed, belong to groups D3 and D4 whose indicative species Cymbella affi nis Kützing and Encyonopsis micro- cephala (Grunow) Krammer were reported from Mediter- ranean mineralized headwaters in Spain (Tornés et al. 2007) and Pyrenean calcareous springs (Sabater and Roca 1992). Also, Cymbella species seem to have high tendency to in- habit such calcium-rich waters, as reported from numerous streams in USA by Potapova and Charles (2003). Species typical for group D2 (mountain, siliceous head- water streams) are Achnanthidium affi ne (Grunow) Czar- necki and Cocconeis placentula Ehrenberg. Those two spe- cies are most common in the streams with higher velocities, A. affi ne as a primary colonizer and C. placentula as a tight- ly attached, fl ow resistant species (Hoagland et al. 1982, Plenković-Moraj and Jasprica 2000, Kralj et al. 2006). Groups D5, D6 and D10 represent mostly lowland, larg- er or smaller silicate rivers and streams impacted by agri- culture (predominantly in Pannonian region) or urban set- tlements as shown by typical planktonic species like Aulacoseira granulata (Ehrenberg) Simonsen and Nitzschia acicularis (Kützing) W.Smith with a benthic species Na- vicula menisculus Schumann, all favoring greater nutrient load (Ács and Kiss 1991, Ács et al. 2003). Group D8 showed statistically proven differences in many aspects from all other groups. As a group it is speci- fi ed by the lack of a common descriptive species and pres- ence of species like Achnanthes trinodis (Ralfs) Grunow in Van Heurck, Cymbella lange-bertalotii Krammer, Encyo- nema prostratum (Berkeley) Kützing or planktonic Asterio- nella formosa Hassall. Group D8 proved to be specifi c be- cause it encompasses sites in lower reaches of large rivers that can be considered almost as lentic sites. Achnanthidium minutissimum (Kützing) Czarnecki as a characteristic species in group D7, (small, mountain, karstic creeks) refl ects more the nature of sites than their nutrient load. The species is known as an early colonizer that favors high fl ow velocities (Primc-Habdija et al. 2001, Kelly 2002, Kralj et al. 2006, Stenger-Kovács et al. 2013, B-Béres et al. 2016) and is commonly reported as a dominant species of similar habitats. Group D9 collects sites from Istra, the largest Croatian peninsula that differs from the rest of Croatian coast (Pre- logović et al. 1995), consisting of upper reaches of small karstic streams and a few continental streams. Although dif- fering signifi cantly, they share a species Nitzschia palea (Kützing) W.Smith, commonly reported from sites affected by agriculture and industry (John 2002, Soininen 2002). Therefore, group D9 encompasses streams affected by high artifi cial nutrient load. Since complete data set included mostly undisturbed and slightly disturbed sites, due to war in Croatia at the be- ginning of 1990s and consequential deterioration of entire industry, all of the data, except from extremely disturbed sites were used. That approach ignores the differences in disturbance to similar sites, but general trends which can help in further research and monitoring were observed. Benthic diatom assemblages do refl ect abiotic typology by grouping similar sites and therefore reducing the number of types. That reduction implies that diatom comminutes sim- plify abiotic groups, as also shown in Hungary (Van Dam et al. 2007). Also, the fact that this study described typical as- semblage for each SOM group, in comparison to many un- defi ned abiotic groups, as well as practicality of considering only 10 in comparison to 24 groups, makes this approach valuable contribution towards description of biotypes and easier determination of water quality according to WFD. Conclusions As particular species have different ecological prefer- ences, some with narrow and some with wide ecological valences, diatom assemblage itself does not seem to show clear preference to particular type of stream. Diatom assem- blage clearly groups similar types of rivers and streams, in- dicating that for estimation of water quality purposes there DO BENTHIC DIATOM ASSEMBLAGES REFLECT ABIOTIC TYPOLOGY ACTA BOT. CROAT. 76 (1), 2017 89 References Ács, É., Kiss, K. T., 1991: Investigation of periphytic algae in the Danube at Göd (1669 river km, Hungary). Archiv für Hydro- biologie-Algological Studies 62, 47–67. Ács, É., Szabó, K., Kiss, K. T., Hindák, F., 2003: Benthic algal in- vestigations in the Danube river and some of its main tributar- ies from Germany to Hungary. Biologia 58, 545–554. Ács, É., Reskóné, M. N., Szabó, K., Taba, G., Kiss, K. T., 2005: Application of epiphytic diatoms in water quality monitoring of Lake Velence – recommendations and assignments. Acta Botanica Hungarica 3–4, 211–223. Anderson, M. J., Gorley, R. N., Clarke, K. R., 2008: PERMANO- VA for PRIMER: Guide to software and statistical methods. PRIMER-E Ltd., Plymouth. APHA, 1995: Standard methods, 19th edition. American Public Health Association, Washington, DC. B-Béres, V., Lukács, Á., Török, P., Kókai, Z., Novák, Z., T-Krasz- nai, E., Tóthmérész, B., Bácsi, I., 2016: Combined eco-mor- phological functional groups are reliable indicators of coloni- sation processes of benthic diatom assemblages in a lowland stream. Ecological Indicators 64, 31–38. Bennion, H., Kelly, M. G., Juggins, S., Yallop, M. L., Burgess, A., Jamieson, J., Krokowski, J., 2014: Assessment of ecological status in UK lakes using benthic diatoms. Freshwater Science 33, 639–654. Biggs, B. J. F., 1995: The contribution of fl ood disturbance, catch- ment geology and land use to the habitat template of periphy- ton in stream ecosystems. Freshwater Biology 33, 419–438. doi:10.1111/j.1365-2427.1995.tb00404.x Bothwell, M. L., 1988: Growth rate responses of lotic periphytic diatoms to experimental phosphorus enrichment: The infl u- ence of temperature and light. Canadian Journal of Fisheries and Aquatic Sciences 45, 261–270. Bothwell, M. L., 1989: Phosphorus–limited growth dynamics of lotic periphytic diatom communities: areal biomass and cellu- lar growth rate responses. Canadian Journal of Fisheries and Aquatic Sciences 46, 1293–1301. Céréghino R., Park Y-S., 2009: Review of the self-organizing map (SOM) approach in water resources: commentary. Environ- mental Modelling and Software 24, 945–947. Clarke, K. R., Warwick, R. M., 2001: Change in marine communi- ties: an approach to statistical analysis and interpretation, 2nd edition. Primer-E, Plymuth. Davy-Bowker, J., Clarke, R. T., Johnson, R. K., Kokes, J., Mur- phy, J. F., Zahrádková, S., 2006: A comparison of the Europe- an Water Framework Directive physical typology and RIVPACS-type models as alternative methods of establishing reference conditions for benthic macroinvertebrates. Hydro- biologia 566, 91–105. DeNicola, D. M., Eyto, E. de, Wemaere, A., Irvine, K., 2004: Us- ing epilithic algal communities to assess trophic status in Irish lakes. Journal of Phycology 40, 481–495. doi:10.1111/j.1529- 8817.2004.03147.x Dohet, A., Ector, L., Cauchie, H.-M., Hoffmann, L., 2008: Identi- fi cation of benthic invertebrate and diatom indicator taxa that distinguish different stream types as well as degraded from reference conditions in Luxembourg. Animal Biology 58, 419–472. doi:10.1163/157075608X383719 Dufrene M., Legendre P., 1997: Species assemblages and indica- tor species: the need for a fl exible asymmetrical approach. Ecological Monographs 67, 345–356. EC, 2000. Directive 2000/60/EC of the European Parliament and of the Council. Offi cial Journal of the European Communities 1–73. European Committee for Standardization, 2003: European Stan- dard EN 13946. Water Quality – Guidance standard for the routine sampling and pretreatment of benthic diatoms from rivers. CEN, Brussels. European Committee for Standardization, 2004: European Stan- dard EN 14407. Water Quality – Guidance Standard for the identifi cation, enumeration and interpretation of benthic dia- toms from running waters. CEN, Brussels. Hoagland, K. D., Roemer, S. C., Rosowski, J. R., 1982: Coloniza- tion and community structure of two periphyton assemblages, with emphasis on the diatoms (Bacillariophyceae). American Journal of Botany 69, 188. Hutchinson, G. E., 1957: A treatise on limnology. Volume 1: Ge- ography, physics, and chemistry. Wiley, New York. John, J., 2004: Diatom assemblages as indicators of wastewater discharge in a temporary stream in Western Australia. In: Pou- lin, M. (ed.), Proceedings of the 17th International Diatom Symposium 2002. Biopress Ltd., Ottawa, Canada. Kahlert, M., Ács, É., Almeida, S. F. P., Blanco, S., Dreßler, M., Ector, L., Karjalainen, S. M., Liess, A., Mertens, A., Wal, J. van der, Vilbaste, S., Werner, P., 2016: Quality assurance of diatom counts in Europe: towards harmonized datasets. Hy- drobiologia 772: 1–14. Kelly, M. G., 2002: Role of benthic diatoms in the implementation of the Urban Wastewater Treatment Directive in the River Wear, North-East England. Journal of Applied Phycology 14, 9–18. Kelly, M. G., Whitton, B. A., 1995. The Trophic Diatom Index: a new index for monitoring eutrophication in rivers. Journal of Applied Phycology 7, 433–444. Kireta, A. R., Reavie, E. D., Sgro, G. V., Angradi, T. R., Bolgrien, D. W., Hill, B. H., Jicha, T. M., 2012: Planktonic and peri- phytic diatoms as indicators of stress on great rivers of the United States: Testing water quality and disturbance models. Ecological Indicators 13, 222–231. Kohonen, T., 2001. Self-organizing maps, 3rd ed, Springer Series in Information Sciences, Vol. 30. Springer, New York. Kolkwitz, R., Marsson, M., 1908: Ökologie der pfl anzlichen Sap- robien. Berichte der Deutschen Botanischen Gesellschaft 26a, Berlin, 505–519. is a need to reduce a number of stream types that naturally occur in some area. Statistically signifi cant grouping of dia- toms into SOM groups with several characteristic species in this study has shown that diatom assemblages can be used as valuable site descriptors. Grouping of similar sites, al- though they initially, according to abiotic typology, belong to different types, makes SOM groups with its correspond- ing representative species an easy tool for determination of water quality and description of reference assemblage. Acknowledgements These results were obtained from AQUEM and ESV projects and resources provided by Croatian Ministry of Science, Education and Sports project 119-0000000-1229. Authors would like to thank Zorana Sedlar, PhD for great patience and help with a map. Also, authors would like to thank two anonymous reviewers whose helpful comments improved the quality of the paper. KRALJ BOROJEVIĆ K., GLIGORA UDOVIČ M., ŽUTINIĆ P., VÁRBÍRÓ G., PLENKOVIĆ-MORAJ A. 90 ACTA BOT. CROAT. 76 (1), 2017 Kralj, K., Plenković-Moraj, A., Gligora, M., Primc-Habdija, B., Šipoš, L., 2006: Structure of periphytic community on artifi - cial substrata: Infl uence of depth, slide orientation and coloni- zation time in karstic Lake Visovačko, Croatia. Hydrobiologia 560, 249–258. Krammer, K., Lange-Bertalot, H., 1991a: Bacillariophyceae 3. Teil: Centrales, Fragilariaceae, Eunotiaceae. In: Ettl, H., Gerl- off, J., Heynig, H., Mollenhauer, D. (eds.), Süsswasserfl ora von Mitteleuropa. Band 2/3. Gustav Fischer Verlag, Stuttgart. Krammer, K., Lange-Bertalot, H., 1991b. Bacillariophyceae 4. Teil: Achnanthaceae, Kritische Ergänzungen zu Navicula (Li- neolatae) und Gomphonema. Gesamtliteraturverzeichnis Teil 1–4. In: Ettl, H., Gärtner, G., Gerloff, J., Heynig, H., Mollen- hauer, D. (eds.), Süsswasserfl ora von Mitteleuropa. Band 2/4. Gustav Fischer Verlag, Stuttgart. Krammer, K., Lange-Bertalot, H., 1997a: Bacillariophyceae 1. Teil: Naviculaceae. In: Ettl, H., Gerloff, J., Heynig, H., Mollenhau- er, D. (eds.), Süsswasserfl ora von Mitteleuropa. Band 2/1. Gustav Fischer Verlag, Jena. Krammer, K., Lange-Bertalot, H., 1997b: Bacillariophyceae 2. Teil: Bacillariaceae, Epithemiaceae, Surirellaceae. In: Ettl, H., Gerloff, J., Heynig, H., Mollenhauer, D. (eds.), Süsswasser- fl ora von Mitteleuropa. Band 2/2. Gustav Fischer Verlag, Jena. Lange-Bertalot, H., 2001: Navicula sensu stricto, 10 genera sepa- rated from Navicula sensu lato, Frustulia. In: Lange-Bertalot, H. (ed.), Diatoms of Europe, 2. A. R. G. Gantner Verlag K. G., Ruggell, Liechtenstein. Lek, S., Guégan, J. F., 1999: Artifi cial neural networks as a tool in ecological modelling, an introduction. Ecological Modelling 120, 65–73. Martin, G., Reyes Fernandez, M. de los, 2012: Diatoms as indica- tors of water quality and ecological status: sampling, analysis and some ecological remarks. In: Voudouris, K. (ed.), Ecolog- ical Water Quality – Water Treatment and Reuse. InTech, Cro- atia. Mihaljević, Z., Mrakovčić, M., Mustafi ć, P., Kerovec, M., Primc Habdija, B., Plenković-Moraj, A., Ternjej, I., Zanella, D., Gli- gora Udovič, M., Ćaleta, M., Marčić, Z., Buj, I., Kralj Bo ro- jević, K., Sertić Perić, M., Dražina, T., Žutinić, P., 2011: Bio- logical methods in assessment of the ecological status (Water Framework Directive (2000/60/EC) in the representative wa- terbodies of Pannonian and Adriatic ecoregion. Study of the Department of Biology, Faculty of Science, Zagreb (in Cro- atian). Pan, Y., Herlihy, A., Kaufmann, P., Wigington, J., van Sickle, J., Moser, T., 2004: Linkages among land-use, water quality, physical habitat conditions and lotic diatom assemblages: A multi-spatial scale assessment. Hydrobiologia 515, 59–73. Pan, Y., Stevenson, R. J., Hill, B. H., Herlihy, A. T., 2000: Ecore- gions and benthic diatom assemblages in Mid-Atlantic High- lands streams, USA. Journal of the North American Ben- thological Society 19, 518–540. Park, Y.-S., Chon, T.-S., Kwak, I.-S., Lek, S., 2004: Hierarchical community classifi cation and assessment of aquatic ecosys- tems using artifi cial neural networks. Science of the Total En- vironment 327, 105–122. Park, Y.-S., Tison, J., Lek, S., Giraudel, J.-L., Coste, M., Delmas, F., 2006: Application of a self-organizing map to select repre- sentative species in multivariate analysis: A case study deter- mining diatom distribution patterns across France. Ecological Informatics 1, 247–257. Plenković-Moraj, A., 1995: Diatoms (Bacillariophyceae) of the Cro- atian Freshwater. Acta Botanica Croatica 54, 22–33. Plenković-Moraj, A., Jasprica, N., 2000: Microphytobenthic com- munities in the fresh waterof Trsteno Arboretum (southern Croatia). Acta Botanica Croatica 59, 351–359. Potapova, M., Charles, D. F., 2003: Distribution of benthic diatoms in U.S. rivers in relation to conductivity and ionic composi- tion. Freshwater Biology 48, 1311–1328. Prelogović, E., Kuk, V., Jamičić, D., Aljinović, B., Marić, K., 1995: Seismotectonic activity of the Kvarner area. In: Vlahović, I., Velić, I., Šparica, M. (eds.), Book of abstracts of the 1st Cro- atian geological congress. Institut za geološka istraživanja, Zagreb, 487–490 (In Croatian). Primc-Habdija, B., Habdija, I., Plenković-Moraj, A., 2001: Tufa deposition and periphyton overgrowth as factors affecting the ciliate community on travertine barriers in different current velocity conditions. Hydrobiologia 457, 87–96. Reid, M. A., Tibby, J. C., Penny, D., Gell, P. A., 1995: The use of diatoms to assess past and present water quality. Australian Journal of Ecology 20, 57–64. Sabater, S., Roca, J.R., 1992: Ecological and biogeographical as- pects of diatom distribution in Pyrenean springs. British Phy- cological Journal 27, 203–213. Schmedtje, U., Sommerhaeuser, M., Braukmann, U., Briem, E., Haase, P., Hering, D., 2001: „Top-down” – Konzept einer bio- zönotisch begründeten Fliessgewässertypologie Deutchlands. In: Deutsche Gesellschaft für Limnologie (ed.), Tagungsbe- richt 2000 (Magdeburg), Tutzing, 147–151. Sládeček, V., 1986: Diatoms as indicators of organic pollution. Acta Hydrochimica et Hydrobiologica 14, 555–566. Soininen, J., 2002: Responses of epilithic diatom communities to environmental gradients in some Finnish rivers. International Review of Hydrobiology 87, 11–24. Soininen, J., Paavola, R., Muotka, T., 2004: Benthic diatom com- munities in boreal streams: community structure in relation to environmental and spatial gradients. Ecography 27, 330–342. Stanley, E. H., Short, R. A., Harrison, J. W., Hall, R., Wiedenfeld, R. C., 1990: Variation in nutrient limitation of lotic and lentic algal communities in a Texas (USA) river. Hydrobiologia 206, 61–71. Stenger-Kovács, Cs., Lengyel, E., Crossetti, L. O., Üveges, V., Padisák, J., 2013: Diatom ecological guilds as indicators of temporally changing stressors and disturbances in the small Torna-stream, Hungary. Ecological Indicators 24, 138–147. Tison, J., Park, Y.-S., Coste, M., Wasson, J. G., Ector, L., Rimet, F., Delmas, F., 2005: Typology of diatom communities and the infl uence of hydro-ecoregions: a study on the French hydro- system scale. Water Research 39, 3177–3188. Tornés Cambra, J., Gomà, J., Leira, M., Ortiz, R., Sabater, S., 2007: Indicator taxa of benthic diatom communities: a case study in Mediterranean streams. Environmental Sciences 43, 1–11. Tornés, E., Leira, M., Sabater, S., 2012: Is the biological classifi - cation of benthic diatom communities concordant with eco- types? Hydrobiologia 695, 43–55. Van Dam, H., Stenger-Kovacs, C., Ács, É., Borics, G., Buczko, K., Hajnal, E., Soroszki-Pinter, E., Varbiro, G., Tothmeresz, B., Padisák, J., 2007: Implementation of the European Water Framework Directive: development of a system for water quality assessment of Hungarian running waters with diatoms. Archiv für Hydrobiologie. Supplementband. Large rivers 17, 339–364.