1. Development of diatom/variable model
1.1 According to the comparison of the two regions (Sweden and Hungary) the Swedish streams with low alkalinity, pH and nutrient levels were segregated. Their characteristic species are widely distributed in Sweden (e.g. Cymbella excisiformis, Fragilaria gracilis) and are known to be acidophilus (e.g. Brachysira neoexilis, Eunotia implicata and Tabellaria flocculosa; VAN DAM et al., 1994). The Hungarian streams were also grouped. Planothidium lanceolatum, Navicula lanceolata, Gomphonema olivaceum and Nitzschia dissipata are typical examples of the most common species. This group included some Swedish streams from regions with calcareous hydrogeochemistry and higher nutrient concentrations.
1.2 The CCA analysis aimed to select the principal environmental parameters responsible for the observed diatom patterns. The correlation between diatom and environmental variables for axis CCA axis 1 (r = 0.966) and axis 2 (r = 0.875) were high, denoting a strong relationship between diatoms and the measured physical and chemical parameters. Ca2+, alkalinity, Mg2+, and pH were the most significant factors to axis 1. These variables are associated primarily with buffering conditions. In addition, there was close correlation with TP.
1.3 The pH and TP models were developed using weighted average method and can be applied in different European phycogeographical regions. In the pH model, correlation was 0.97 and the RMSEP 0.3, the test set fit the model. The combined model covered the pH scale from 4.5 to 9.5. In the TP calibration the best results were obtained with a restricted set (n = 56) using weighted averaging with tolerance weighting with inverse regression. The correlation was high (0.96) and the prediction error was 0.193 mg L-l on the scale from 0 to 0.755 mg L-1.
1.4 The pH optima and tolerances were calculated for 243 taxa using WA method. The optimum of the pH varied between 5.3 (Eunotia naegelii) and 8.1 (Navicula lanceolata) (Table 3). Species indicating acidic waters are members of the Eunotia genus. There are some typical species indicating circumneutral pH (e.g. Achnanthidium minutissimum).
At pH values >7, many species with different optima were found (e.g. Fragilaria arcus, Gomphonema olivaceum). The calculated optima for some species were different from data suggested in the literature: Psammothidium subatomoides, Brachysira neoexilis, Craticula riparia. Such findings may indicate that different ecotypes of the same species do exist within Europe and provide substantial data for regional re-setting sensitivity and indicator values of widely used indices.
1.5 The TP optima and tolerances were also calculated for 211 taxa. Species distribution patterns did not show clear relationships with the concentrations, therefore we selected only the most indicative taxa for eutrophication levels from the training set. Nitzschia capitellata indicated unequivocally the hypereutrophic state of the stream. The eutrophic state was clearly indicated by Gyrosigma acuminatum, Cyclotella meneghiniana, Cocconeis placentula var. euglypta. Some species with broad tolerance range (e.g Navicula cryptotenella) can be found from oligotrophic to eutrophic waters.
2. Development of trophic diatom index for Hungarian lakes
2.1 Using the WA technique, a robust and accurate TP transfer function was developed spanning from 0.01 to 5.72 mg L-1 TP. In our Hungarian sample set, diatom assemblages occurred along wider TP gradient than the range of other diatom-based models (ANDERSON et. al., 1993: 25-800 µg L-1; HALL andSMOL, 1992: 5-28 µg L-1). In the TP model the correlation was strong (r2 = 0.96, RMSE = 0.17) using weighted averaging without tolerance downweighting regression (WAtol) with inverse deshrinking. The error of the model cannot be considerably reduced as a consequence of the complexity of eutrophic lakes, particulary if they are shallow and highly productive (BENNION, 1994).
2.2 The TP model was suitable for calculating the optima and tolerances of the species and establishing the indicator values of the species at different TP levels. Indicator values of 127 species provide the basis of the developed TDIL index. These species are the most frequent ones in Hungarian lakes. Cymbella helvetica, Gomphonema angustum and Diatoma moniliformis were sensitive and characteristic species for low trophic state. On the other edge of the trophic spectrum there were several tolerant, eutrophic taxa e.g. Nitzschia communis, Amphora veneta, Craticula cuspidata.
2.3 Comparing the species list of the TDIL to that of TDI Austria (ROTT et al., 1999), 31 species were absent in our material, mostly centric diatoms. The consequence of the absence of centric diatoms indicating nutrient-rich conditions is that the TDI Austria index can underestimate (optimistic) the trophic status of lakes (KITNER and POULÍĈKOVÁ, 2003). The TDI Austria developed for rivers and was used all over Europe (ROTT et al., 2003). It proved to be applicable for samples from reed stems with actual and average TP for perialpine lakes, since character of these lakes was similar to that of the alpine stream ecosystems (POULÍĈKOVÁ et al.; 2004). Furthermore, 73% of the species in our material were eutrophic or hypertrophic according to the van Dam trophic index (VAN DAM et al., 1994) which would predict unacceptably high trophic level similarly to the OECD system. Although in Czech fishponds the van Dam’s index was appropriate to recognise extremes (clear and dirty) of lake environments (KITNER
and POULÍĈKOVÁ, 2003), for the Hungarian lakes it was not suggested (VAN DAM et al., 2005). The TI (HOFMANN, 1999) was well applicable in Germany (SCHAUMBURG et al., 2004) but it includes indicator value for only 50 species of the 127 most frequent diatom species in the Hungarian shallow lakes. Therefore, the TI would carry a considerable uncertainity if applied to assess lake trophic status in Hungary.
2.4 According to the TDIL, the ecological status of 4 lakes were in excellent, 25 in good, 21 in medium, 21 in tolerable and 12 in bad status, which appears representative for the status of the Hungarian lakes. Most of the sampling sites in bad or poor status by the TDIL were shallow, saline lakes with naturally high TP content and high
2.6 According to the conductivity transfer function the correlation between the observed and the diatom inferred conductivity was high (r2= 0.95, n=67). The RMSEP value was
356 µS cm-1. The characther species of this water type were, among others, Amphora veneta, Anomoeoneis sphaerophora, Cymbella pusilla, Craticula cuspidata, Nitzschia clausii, N. supralitorea, N. vitrea and Surirella peisonis. Altogether 19 species characterized the exceptionally high conductivity and TP. The corrected TDIL index values did not indicate the bad ecological status of these lakes. Consequently, in the case of temporary, saline lakes modified indicator values have to be used to assess the realistic ecological status. After applying the modified values bad or tolerable ecological status by the simple TDIL changed: of the 30 sampling sites fall into 3 medium, 6 lakes were in good and rest 21 were in excellent ecological status.
3. Applicability of Achnanthidium minutissimum as an indicator species 3.1 Based on the data we conclude that the discharge-dependent Si and NO3
--N content of the stream water combined with the good colonization abilities (r-strategy) of the species are the major factors that influence the temporal distribution of the A.
minutissimum. It typically reaches higher dominance after flood periods and therefore it primarily indicates natural disturbances. Since such events wash a significant amount of NO3
-N from the adjacent agricultural fields it also indicates presence of diffuse pollutants.
3.2 According to most diatom sampling protocols for WFD qualification, periods after floods should be avoided. In this view we may conclude that high contribution of A.
minutissimum in the samples from alkaline/circumneutral streams may indicate a preceding flood and therefore such samples, in general, do not provide a reliable basis for ecological status assessment within the WFD.
4. The influence of damming on compositional features of attached diatom assemblages downstream
Hydro-morphological modifications can result the change in the diversity. The diversity of natural, undisturbed streams is different from the hydro-morphologically modified (dammed) water bodies according to river-types. The proportion of attached against planktonic species appeared to be a good indicator of the damming, both upstream and downstream, on two ways: (a) the relative contribution of attached species decreased in the total encountered species (attached, planktonic and unknown life forms), (b) the proportion of planktonic species increase. Furthermore, character species of the modifications were appeared.
5. Conservation biological concerns
During the ecological surface-water monitoring (ECOSURV project) according to Water Framework Directive 46 diatom species were identified in different threathened categories according to the Hungarian Algal Red List. 7 presumably endangered species were found in 17 sampling sites (e.g. Bacillaria paradoxa GMELIN, Cymbella tumidula GRUNOW, Diploneis ovalis (HILSE)CLEVE). There were 8 low risk species in 51 sites (e.g. Aulacoseira distans (EHRENBERG) SIMONSEN, Cymbella brehmii HUSTEDT). 25 vulnerable species were found in 232 samling sites (e.g. Achnanthes marginulata GRUNOW, A. petersenii HUSTEDT, Aulacoseira muzzanensis (MEISTER) KRAMMER, Cymbella hungarica (GRUNOW)PANTOCSEK, C. lacustris (AGARDH)CLEVE, Diploneis oblongella (NÄGELI)CLEVE-EULER). Fragilaria capucina DESMAZIÈRES var. austriaca (G ) L -B was the only one endangered species, and 2 species
(Gomphonema vibrio EHRENBERG, Luticola (Navicula) goeppertiana (BLEISCH) H. L.
SMITH) are known to be extinct. This list represents that the conceptions (species level identification, habitat diversity) of WFD conforms with the conservation biological purposes.