Nordic river fish intercalibration recent advances

Transcription

1 Nordic river fish intercalibration recent advances Teppo Vehanen, FI (coordinator), Magnus Dahlberg (& Ulrika Beier) SE, Fiona Kelly & Ronan Matson IR, Trygve Hesthagen & Jo Halvaard Halleraker NO, Liam O'Connor NI, Willie Duncan & Alistair Duguid SC

2 Short river fish intercalibration history PILOT EXERCICE (May 2006 June 2007 ) : 1st Intercalibration on Fish for Rivers Meeting, May 2006, Paris, followed by Rotterdam (2006) and Ispra (2007) Management by JRC (Niels Jepsen) and Cemagref (Didier Pont) 21 MS : Austria, Belgium, Czech Republic, Denmark, England, Finland, France, Germany, Hungary, Ireland, Italy, Lithuania, Luxemburg, Netherland, Poland, Portugal, Romania, Sweden, Slovakia, Slovenia, Spain

3 Short intercalibration history pilot exercise Objectives of the pilot exercise: Comparison of reference conditions Preliminary description and comparison between national methods Testing common metrics (European Fish Index, EFI) at European level Results: Feasibility of the intercalibration exercise using common metrics at the European level, but Common metrics not calibrated for particular situations Lack of well-defined common criteria for reference sites

4 Short intercalibration history - INTERCALIBRATION EXERCICE : 4th meeting in Ranco 2008, 5th Scharfling 2008, 6th Dublin 2009, 7th Edinburgh (Oct. 2009), Düsseldorf 2010? 26 MS : Austria, Belgium-Wallonia, Belgium-Flanders, Czech Republic, Denmark, England-Wales, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Latvia, Lithuania, Luxemburg, Netherland, Northern Ireland, Norway, Portugal, Romania, Scotland, Slovakia, Slovenia, Spain, Sweden

5 Short intercalibration history Objectives: Defining common reference criteria's All data centralized in one common database - Pre classification of pressure - Environmental variables Evaluation and testing of a new set of common metrics (EFI+ project) Comparison between National Methods within Regional groups Harmonizing H/G and G/M class boundaries between MS.

6 Organization of the work Since the beginning of 2009, the leader of the whole group is Didier Pont (Cemagref, France) The five regional groups are coordinated by: Nordic Group: Teppo Vehanen Lowland-Midland Group: Tom Buijse and Cornelia Schuetz (and with Marco Beers) Alpine-type Mountains Group: Mickael Schabuss Mediterranean South-Atlantic: Teresa Ferrera (with Pedro Segurado) Danubian Group: Vladimir Kovac Cemagref (France) is in charge of the development of the common

7 INTERCALIBRATION Organization of the work Common database - Environm. variables - Pressure types - Fish data 2 3 Common approach (ICM, Option 2) -Ref. conditions between Countries / Reg. Groups - Common metrics computation -High/Good & Good/Moderate boundaries comparisons - Associated uncertainties -- Common metrics -- National methods Discussion Agreements Specificities (Environment, pressures) - Fish as indicator? - Data collection - National methods - Option 3 General meetings Nordic FI IR SC N-IR SE NO Alpine-type Mountains AT DE FR SL ES Lowland Midland NL DE DN BE-F BE-W LUX FR LT LV EST ENG-W Mediterranean South-Atlantic FR ES PT GR Danubian RO CZ SVK HU AT SV Regional meetings The large rivers (mainly large floodplain rivers) are not considered in this second round.

8 Estimated timetable January-April 2009: Data collection / Data checking. Common database May 2009: Preliminary analysis (common metrics and regional groups) May : 6th IC Meeting. Ireland. First results and discussion. Harmonisation of statistical methods Table of contents of the final report June-August 2009: Data analysis. Exchange between partners. Final results for the Milestone report.. September 2009: Report to ECOSTAT October 2009: ECOSTAT Meeting. Recommendations. October 2009: 7th Fish IC meeting. Results and discussion about H/G and G/M classes boundaries February 2010: 8th Fish IC meeting. Discussion about H/G and G/M classes boundaries October 2010: Final report: methodology, results, boundaries between class 1/2/3 Questions, & environmental situations remaining open. Future tasks : Large rivers,

9 EUROPEAN COMMISSION DIRECTORATE GENERAL JRC JOINT RESEARCH CENTRE Institute of Environment and Sustainability River Fish Intercalibration Group WFD Intercalibration Phase 2 : Milestone 1 report September 2009 Pont D., M. Beers, T. Buijse, O. Delaigue, T. Ferrera, N. Jepsen, V. Kovac, M. Schabuss, P. Segurado, C. Schuetz & T. Vehanen Methods to be intercalibrated Checking of compliance of national assessment methods with the WFD requirements Progress on feasibility checking: method acceptance criteria Progress on collection of IC dataset and design the work for IC procedure Progress on reference conditions/benchmarking

10 Nordic Group -Progress on collection of IC dataset and design the work for IC procedure Database collection completed Reference conditions identified Nordic Group database Finland Ireland N-Ireland Norway Scotland Sweden Sum. No. Sites Reference Sites

11 Methods to be intercalibrated - FIFI Type specific reference condition approach (typology based on geology, and catchment area) Nordic methods VIX Site specific approach No typology (expected values) Each type has its own reference conditions Each metric evaluated according to response model (degree of disturbance vs. metrics), cumulative frequency curves set the value (0-1) for each metrics) Five metrics compose the fish index Index is the mean out of five metrics Border High Good : 75% references = High status Rest of classes determined in equal intervals towards zero For all sites -Expected values for each metric calculated by multimetric regression functions (catchment size, altitude, wetted width, slope,..). Reference sites were identified using maximum values The distance from the reference conditions sets the value for the index Six metrics compose the index Index = mean of Standardized residuals Probability values (expected response) The border between high and good status - probability of classifying an unimpacted site as impacted is less than 5%. The level between good and moderate status is set so that the probability to misclassify impacted (classes 3-5) and unimpacted (1-2) sites is equal.

12 -Cumulative frequency distributions (a point estimate for classical probability to have Methods to be intercalibrated Fifi Nordic methods VIX

13 Methods to be intercalibrated Nordic methods metrics FIFI (Finland) VIX (Sweden) Number of fish species Proportion intolerant fish species Proportion tolerant fish species Number of age 0+ salmonids Number of Cyprinidae individuals Total abundance of salmon and trout Proportion of reproducing salmonid species Proportion of tolerant species Proportion of intolerant species Proportion of lithophilic individuals Proportion of tolerant individuals

14 Checking of compliance of national assessment methods with the WFD requirements Reference conditions 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

15 Checking of compliance of national assessment methods with the WFD requirements The monitoring method for the parameters conforms to the international standards (CEN-standard, Water quality Sampling of fish with electricity, EN 14011). Normative definitions for status classes are followed: High, Good, Moderate, (Poor, Bad). EQR (0-1) is used to represent results. Reference conditions are established. Relationship between biological metrics and anthropogenic pressure data are verified. Metrics represent the values of the quality elements for the classification of ecological status of rivers specified in Annex V of the WFD: composition, abundance, and age structure (density of age 0+-salmonids, proportion of salmonids reproducing) of fish fauna.

16 Nordic Group progress on feasibility checking problems encountered Differences rise from: - differences in river types among the reference sites (typology) Reference sites Reference sites To solve this problem overrepresentation of small siliceous rivers in the Swedish data has been diminished New reference data has been used to calculate new borders for ecological classes in the Finnish index

17 Progress on feasibility checking: method acceptance criteria Nordic Group tested different river typology options using altitude (4 classes), geology (3 classes) and catchment size (5 classes) as variables The general result was that typology (in all cases) improved the comparison of the results between the two methods compared Currently the simple typology including geology and altitude (nine river types) is used by the group FiFi- index is a strict type based method using reference conditions for each river type separately, VIX calculates the reference conditions by across all types, i.e. including all types as a reference. Typology is used in the Nordic Group comparisons only by removing sites from those types that are overrepresented in reference group in VIX calculations. Direct comparisons between types cannot be made, and therefore the use of typology is partly unfeasible

18 Progress on feasibility checking: method acceptance criteria The basic assessment concept is the same in the both approaches: index is composed of several metrics which are found effective in response to human pressures. Both indexes tend to respond to pressures similarly: there is a relatively high correlation (r = 0.713) between the indexes. This also holds when data from each countries is analyzed separately

19 Progress on feasibility checking: method acceptance criteria The basic difference in the concept is the setting of boundaries between the ecological status classes. The FiFi index assumes that most of the reference sites (75%, the good-high boundary is set here) are in high status. Rest of the status class borders are set in equal intervals towards zero value. VIX-index sets the border between good and moderate status where the probabilities of making type-i and type- II errors are equal, i.e. where the risks of classifying an impacted site as unimpacted are equal. The border between high and good status ( HG ) is chosen so that the probability of classifying an unimpacted site as impacted is less than 5%.

20 Progress on feasibility checking: method acceptance criteria Reference sites All sites All Countries Typology: Altitude*Geology VIX status (rows) by FiFi status (columns) Total Total Example: Irish sites Moderate/Good Typology: Altitude*Geology VIX status (rows) by Fifi status (columns) Total Total M/G Typology: Altitude*Geology VIX status (rows) by FiFi status (columns) Total Total

21 Progress on reference conditions/benchmarking Guidance: Similar data acquisition, but different numerical evaluation by the use of common metrics IC Options 3 supported First both national methods were applied to the dataset. Then we calculated normalised EQRs based on the national view. National values were then rescaled to a maximum of 1 and minimum of 0. Then we calculated the median value for the standardized reference dataset for both indexes and transformed the values into EQR s by dividing the standardized value by the median value. All sites that were classified as high status by the Finnish system (FiFi) were selected, and for those sites the standardised EQR values from the Swedish system we extracted, and vice versa. We then used ANOVA test the possible differences. The difference between the countries was significant (p<0.001, see the figure).

22 Progress on reference conditions/benchmarking We plotted the distribution of standardised EQR values of the reference sites together with the boundaries for the high-good and good moderate sites. FiFi index sets the boundary to the range of the 25% percentile of the EQR values of the reference sites (variation due to differences in river types). VIX-index is stricter and the high-good boundary leaves most of the reference sites (93%) outside the high status. The moderate-good status boundary in the FiFi system leaves only few reference sites to moderate status (2 %), whereas in VIX-system 21% of the reference sites achieve only moderate status.

23 Conclusions? We need to harmonise the setting of the borders for the ecological status classes!

24 Common intercalibration Didier Pont, CEMAGREF

25 Common intercalibration Didier Pont, CEMAGREF

26 Common intercalibration Didier Pont, CEMAGREF

27 Common intercalibration Didier Pont, CEMAGREF

28 Thanks for listening