For the common approach at the European level using common metrics, France (Cemagref) was mainly in charge of the constitution of the common database and of the data analysis. JRC was mainly in charge of organization of the meetings, the writing and distribution of minutes, the collection of all the national datasets on the CIRCA web-site and the synthesis of regional group results.
Contents of the fish intercalibration report
This report is mainly based on results and descriptions submitted from each of the 7 IC-groups. We (Jepsen & Pont) have tried to draw some main conclusions and recommendations based on the work done by all the people involved and also some editing of the reports and other documents to make the information more accessible and clearer for the reader, however the contend is unaltered and expresses the experience and views from the groups.
Each group report is placed as an annex. To limit the variation, these group reports are organised in a similar way, first answering a set of questions, then describing in more detail the process and the results, then descriptions of the national methods used and finally the criteria used for setting reference conditions. For countries involved in several groups (like Germany and France) the description of method and reference setting is only given once!
General results and conclusions
Since the first meeting one year ago in May 2006, much effort has been put into this pilot exercise by a number of people from more than 20 MS. We have been working with 3 categories: 1) MS with an accepted/approved national method, 2) MS with methods under development/approval and 3) MS without any national method (planning to use common metrics or other countries methods). This intercalibration work included 12 national methods being compared in 7 regional groups. Some of the methods showed very similar response and here it would be possible to set boundaries, provided that results from more sites (especially reference sites) were available.
Conclusions from the groups (Exempts from the group reports):
Mediterranean: All the applied methods in the Mediterranean region must be related to the local fish fauna (WFD: Water district concept). At least at this moment, there are no possibilities to intercalibrate methods or to search for common metrics. The proposal is to work on impact evaluating criteria (preclassification). The method must be used as benchmark (as diagnostic tool) and the final judgement should be issued by experts.
Alpine: Bilateral data exchange was problematic due to the short timeframe and the limitation in resources. For Germany the federal structure seems to create some organizational problems concerning comprehensive availability and support of suitable data sets. The common method (EFI) does not react to hydromorphological pressures, which are clearly dominating in the concerned countries. There is no correlation with national methods. As a result boundaries could not be set.
Comparisons between the different datasets with all applicable methods showed that correlations between Austrian (FIA) and German (fiBS) methods are relatively good, correlations of EFI with FIA and fiBS do not work well at all, as does FBI compared to FIA or fiBS. Since FBI does not have a Danube fish region it obviously can not be expected to work with our data. On the other hand, the French data would work with Austrian and German methods, so this would have been an interesting comparison.
Carpathian: The (IC-) work is at its beginning and national methods are being developed now. Only Czech Republic has a national method, Romania is expected to have a method ready this year (2007).
Nordic: Good correlation between Finnish and Swedish methods. The FIFI method appears to consistently class sites higher than the VIX method. There are considerably more sites classified as high or good status using FIFI than using VIX. Furthermore, the outcome in the FIFI and VIX indices appear to be clearer regarding chemical impact, especially impacts from increased nutrient input compared to hydrological or morphological impact. However, it is still apparent that FIFI generally gives a higher index value compared to VIX.
Midland: Due to a very short timetable, the full intercalibration process couldn’t be achieved for the midlands. Some indexes are still under development, do not exist or are applicable only to some of river types or region. There was a big difference in the number of data provided by the countries which limit the conclusions. The provisional results are interesting and show that the three indexes gave different classification status using the same data. Considering the eference data set, the three indexes (EFI, IBIP and FBI) identify in a comparable way the reference sites as good or high quality status, but there are strong differences between indexes concerning the HG boundary. For the moment it seems that criteria for selecting reference sites are not fully comparable. Moreover, indexes developed at a regional scale are not expected to be applied elsewhere, this extrapolation could explain a part of the divergences observed. For the moment, the EFI does not constitute a highly consistent intercalibration tool, since it does not exhibit better results than direct national indices comparison ; but in the perspective of joining lowland and midland, EFI and a fortiori the EFI+ version could be really helpful.
Lowland: It was possible to build a large dataset for comparison within a short time. An intercalibration exercise within the participating countries of the lowland and midland group will be possible. But there is a lot of work to do, several crucial issues (e.g. reference sites) still have to be addressed and the development of different assessment methods is still going on. For the 47 sites that could be calculated by all methods the German method assessed the sites as being the worst. Then the Netherlands method was most severe. EFI and IBI gave comparable average values. Regarding correlations between the results the EFI, FIBS and Dutch Index were best comparable. The Flanders IBI did not correlate with the other methods. So we conclude that the intercalibration for fish in Low- and Midland Rivers will be possible by June 2009.
Atlantic: The preliminary results indicate that the indication provided by the EFI in relation with pressures should be improved, at least for the Spanish dataset. Moreover, even if the response to pressures provided by EFI in the French dataset seems better in comparison with the Spanish one, the comparison between the French national system and the EFI shows a weak relationship (r2=0.4).