Mediterranean GIG River BQEs Intercalibration Exercise

Mediterranean GIG River BQEs Intercalibration Exercise
Macroinvertebrates, diatoms and macrophytes General Coordinator: Teresa Ferreira Coordinator benthic macroinvertebrates: Maria João Feio Coordinator benthic microflora: Salomé Almeida Coordinator macrophytes: Francisca Aguiar Coordinator Fish: Pedro Segurado WFD CIS Ecostat Meeting, Brussels, October 2011

Content GIG organization and meetings
BQE macroinvertebrates, phytobenthos and macrophytes Common intercalibration types and national methods Database and IC types Steps for IC procedures Boundary comparison and harmonisation Conclusions and further needs

7 countries participating: Portugal, France, Spain, Cyprus, Slovenia, Greece (only macrophytes) and Italy 42 experts and country representatives 21 international meetings, of which 7 General Med GIG meetings

Overview of Methods to be intercalibrated
Benthic macroinvertebrates Member State Method Status Reported in Wiser Portugal Rivers Biological Quality Assessment Method-Benthic Invertebrates (IPtIN, IPtIS) 1 Yes Spain 1 Iberian Biological Monitoring Working Party (IBMWP) Spain 2 Iberian Mediterranean Multimetric Index—using quantitative data (IMMi-T) France Global biological normalized index (IBGN) Cyprus STAR Intercalibration Common Metric Index (STAR-ICMi) Slovenia Slovenian Ecological Status assessment system for rivers using benthic invertebrates 1,2 Italy Based on STAR_ICM index calculation (MacrOper)

National method descriptions

Compliance checking of national methods

Common intercalibration types
Common IC type Type characteristics MS sharing IC common type RM1 catchment <100 km2; mixed geology (except non-siliceous); highly seasonal France, Italy, Portugal, Slovenia, Spain RM2 catchment km2 ; mixed geology (except non-siliceous); highly seasonal France, Greece, Italy, Portugal, Slovenia, Spain RM4 non-siliceous streams; highly seasonal Cyprus, France, Greece, Italy, Spain RM5 temporary rivers Cyprus, Italy, Portugal, Slovenia, Spain RM3 - insufficient reference sites for intercalibration purposes, and some methods were not comparable

Common intercalibration types
and TYPE 5 n-MDS ANOSIM testing THE SAME TYPOLOGICAL SIMILARITY WAS FOUND FOR ALL BQE

Biological concordance for the 2 types used
RM5 RM1,2,4 REFERENCE SITES NON-REFERENCE SITES

Macrophytes MS Method Abb. Appropriate to types Cyprus Multimetric Macrophyte Index MMI RM5 Biological Macrophytes Index for Rivers IBMR RM4 France RM1, 2, 4 Greece RM2 Italy RM1, 2, 4 and RM5 Portugal RM1,2,4 and RM5 Slovenia River Macrophyte Index RMI Spain RM1, 2, 4 and RM5 Multimetric Macrophyte Index Metrics: Number of nitrophyllous species, log (No of Gramineae), Number of Helophytes_herb, species richness, mean cover of green algae (categorical) Papastergiadou E., P. Manolaki MedGIG Report: R-M5 IC river types of Cyprus. Developing an Assessment system of R-M5 river types for Cyprus. Patras University, Greece, 21 pp. (not published) River Macrophyte Index (RMI) Described in: Kuhar, U., Germ, M., Gaberščik,A., Urbanič, G Development of a River Macrophyte Index (RMI) for assessing river ecological status. Limnologica 41: Web page describing the national method: a/ RMI was calculated according to the following equation: The RMI was calculated using the following equation: where QAi = abundance of the taxa i from the group A, QABi = abundance of the taxa i from the group AB, QBCi = abundance of the taxa i from the group BC, QCi = abundance of the taxa i from the group C, QSi = abundance of taxa i from all groups (group A, AB, B, BC, C; taxa from the group ABC are not considered), nA = total number of taxa in group A, nAB = total number of taxa in group AB, nBC = total number of taxa in group BC, nC = total number of taxa in group C, nS = total number of taxa in all groups (group A, AB, B, BC, C; taxa from the group ABC are not considered). Kuhar, U., Germ, M., Gaberščik,A., Urbanič, G Development of a River Macrophyte Index (RMI) for assessing river ecological status. Limnologica 41:

Diatoms Member State Method Status Reported in Wiser Cyprus IPS (Coste in Cemagref, 1982) 1, 2 Yes France IBD 2007 (Coste et al, Ecol. Ind. 2009) (AFNOR NF-T , December 2007) Italy ICMi (Intercalibration Common Metric) Index (Mancini & Sollazzo, 2009) No Portugal Slovenia Slovenian Ecological Status assessment system for rivers using phytobenthos based on the Saprobic index (Rott et al. 1997) and the Trophic index (Rott et al. 1999) Spain River Macrophyte Index (RMI) Described in: Kuhar, U., Germ, M., Gaberščik,A., Urbanič, G Development of a River Macrophyte Index (RMI) for assessing river ecological status. Limnologica 41: Web page describing the national method: a/ RMI was calculated according to the following equation: The RMI was calculated using the following equation: where QAi = abundance of the taxa i from the group A, QABi = abundance of the taxa i from the group AB, QBCi = abundance of the taxa i from the group BC, QCi = abundance of the taxa i from the group C, QSi = abundance of taxa i from all groups (group A, AB, B, BC, C; taxa from the group ABC are not considered), nA = total number of taxa in group A, nAB = total number of taxa in group AB, nBC = total number of taxa in group BC, nC = total number of taxa in group C, nS = total number of taxa in all groups (group A, AB, B, BC, C; taxa from the group ABC are not considered). Ecological status assessment system for rivers using phytobenthos Uradni list Republike Slovenije stran (pp) 832, št. (no) 10, Metrics: Saprobic index (SI = Sum of (Indicator Taxa Abundance * Saprobic value * Indicator weight) / Indicator Taxa Abundance* Indicator weight), Trophic index (TI = Sum of (Indicator Taxa Abundance * Trophic value* Indicator weight) / Indicator Taxa Abundance* Indicator weight).

Benthic algae MS Taxonomic composition Abundance Undesirable disturbances Bacterial tufts Combination rule CY Diatoms only Relative abundance Not included in national metric Not defined yet FR IT Included in macrophyte assessment tool (IBMR) has also macroalgae PT SI Diatoms only (other groups sampled but not included in the assessment SP Diatoms only (algae included in macrophyte sampling) Contrarily to macroalgae diatom taxonomy is well known Diatoms is considered a good proxy for the phytobenthos Bacterial tuffs and US do not affect upper quality classes Contrarily to macroalgae (namely in Mediterranean regions), diatom taxonomy is well known and diatom species have well established responses to several pressures. More basic research has to be conducted in Mediterranean regions in order to have fully operational methods including other algae and the knowledge of reliable responses to pressure. For the moment, diatoms should be considered a good proxy for the phytobenthos sub-element1. All methods being used include the taxonomic composition at species level and the relative abundance, and responses to pressure using relative abundance are well documented Europe-wide. The improvement related to the use of absolute abundances when compared to relative, has not been demonstrated, and proved irrelevant in some non-Med MS (Appendix 1 of the Rivers/X-GIG Milestone 5 report). There are also technical difficulties not yet solved when sampling and interpreting absolute abundance values. Bacterial tufts are not considered to influence classification because they only occur when other elements already classify quality as poor or bad, so the H/G and G/M boundaries are unlikely to be affected. Relationship between the four components of the normative definitions is yet to be examined. “Undesirable disturbances” is a subjective expression and hard to translate into quality assessment metrics, and it should be defined specifically for phytobenthos. UD are not considered to influence classification because they only occur when other elements already classify ecological quality as poor or bad, so the H/G and G/M boundaries are unlikely to be affected. In any case what can be considered as undesirable disturbances (e.g. lack of oxygen) is covered by the present assessment and also by other quality elements. The assessment methods of the MS were considered compliant and ready for intercalibration.

The assessment methods of the MS were considered compliant and ready for intercalibration

Collection of IC dataset
Total = 1384 samples MACROINVERTEBRATES

Collection of IC dataset
Distribution of samples by quality class (example from invertebrates) Do we have a good expression of the disturbance gradient?

Reference conditions approach (Med GIG)
The MedGIG has developed an approach for common reference conditions. The same and common thresholds were used for the BQEs macroinvertebrates, phytobenthos and macrophytes. Reference sites originally provided by MS were screened according to the proposed approach and common thresholds.

Conditions of acceptance The selection of IC reference sites was done through a 3 steps procedure. Steps 1 and 2, up to the establishment of reference thresholds, were performed with the information for MS reference sites provided to the MedGIG database. We used sites supplied to the MedGIG for invertebrates, diatoms and macrophytes databases. The global database was composed of a total of 919 member states reference sites distributed through the 4 IC river types (RM1, RM2, RM4, RM5) and 7 MS (CY, FR, GR, IT, PT, SI, SP). Step 1. Original national reference sites are selected only if all their categorical variables have class 1, no impact or minimal impact. Step 2. Reference thresholds are calculated for numerical pressure variables using the common data base, based on reference sites selected in Step 1 and for each IC type. Extreme values for each pressure variable and IC type were previously excluded after histograms and boxplots inspection. These observed ranges characterize the pressure levels existent in the minimally impacted sites, for each IC type, in the Mediterranean region. A unique value for each pressure variable (Reference thresholds) was afterwards calculated for all IC types, corresponding to the maximum pressure acceptable overall Mediterranean types, in order to reach a common tolerance level. However, for RM5, the temporary rivers, different ranges for water oxygenation were established for low water periods. Step 3. Final abiotic screening of reference sites. Potential reference sites, left out in Step I, are rescreened and those with categorical variables in class 2 (any number) but simultaneously with numerical variables values within the thresholds defined in Step 2 are chosen and added to the reference set of Step 1. After this screening, we checked if for the same site some samples passed step 3 and others not. We re-included in the IC benchmarks all samples belonging to sites for which the mean values passed the thresholds. Finally, for MS with a low cover of reference sites for a given type, we looked at those sites supplied as disturbed (at the national level) and if they passed the established quantitative thresholds they were also included also in the set of benchmarks. This happened however, only for Slovenia for types RM1 and RM2, for the invertebrate’s dataset. Step 1- Only sites with 1 categorical pressure were sorted from the 919 RS Step 2- Reference thresholds were calculated Step 3- Final screening and values

Checking Med reference sites (macroinvertebrates) ICMi outliers were eliminated

Checking Med reference sites (macroinvertebrates) The IC benchmark sites include sites with small changes at the connectivity level and land use, which was the accepted level for common thresholds.

Checking Med reference sites (diatoms) Samples with ICM values below 0.6 (circled in red) were considered outliers/extremes and removed from subsequent analyses

Checking Med reference sites (macrophytes) Number of observations of Med GIG reference sites of types IC RM1,2,4 along the individual pressure gradients

Checking Med reference sites (macroinvertebrates) Seasonality influence on EQR values: MedGIG program In general, the ICMi values of reference sites obtained in different seasons cannot be distinguished and also that in within seasons there is a high variability of values. However in RM1 (smaller rivers) the values for autumn and winter samples seem to be generally lower that those of spring and summer even though there are also some equivalent lower values within that season.

Collection of IC reference dataset: macrophytes
432 sites were provided by MS (25 to 108 at country level) Reference data set was subjected to screening using Mediterranean reference conditions River types RM1, RM2 and RM4 will be gathered for the intercalibration, and a sufficient number of reference sites are available to make a statistically reliable estimate. For RM5 the number of available reference sites can be insufficient (8 sites for the overall database). Member State RM1 RM2 RM4 RM5 TOTALl Cyprus - 3 100% 5 63% 8 73% France 6 60% 10 59% 16 Greece 9 69% 1 71% Italy 4 80% 11 92% 15 88% Portugal 50% 33% 55% Slovenia Spain 0% 57% 21 54% Sum 23 66% 14 58% 41 67% 47% 86 For RM5 the number of available reference sites can be insufficient (8 sites for the overall database).

Response to pressures: macrophytes
Method Pressure RMI (Slovenia) Catchment land use, Eutrophication IBMR (Cyprus, Greece, France, Italy, Portugal, Spain) Cyprus –hydrological, morphological, land-use and physico-chemical degradation. Greece - hydrological, morphological, land use and physico-chemical degradation. France - Eutrophication, General degradation, Hydromorphological degradation; Italy –Eutrophication, General degradation, Pollution by organic matter; Portugal – Eutrophication, General degradation. Spain - Eutrophication, General degradation. MMI (Cyprus) Cyprus – hydrological, morphological, land use and physico-chemical degradation. River Macrophyte Index (RMI) Described in: Kuhar, U., Germ, M., Gaberščik,A., Urbanič, G Development of a River Macrophyte Index (RMI) for assessing river ecological status. Limnologica 41: Web page describing the national method: a/ RMI was calculated according to the following equation: The RMI was calculated using the following equation: where QAi = abundance of the taxa i from the group A, QABi = abundance of the taxa i from the group AB, QBCi = abundance of the taxa i from the group BC, QCi = abundance of the taxa i from the group C, QSi = abundance of taxa i from all groups (group A, AB, B, BC, C; taxa from the group ABC are not considered), nA = total number of taxa in group A, nAB = total number of taxa in group AB, nBC = total number of taxa in group BC, nC = total number of taxa in group C, nS = total number of taxa in all groups (group A, AB, B, BC, C; taxa from the group ABC are not considered). the Mediterranean GIG database was used to test the response of each metric to: individual types of pressures and to a general degradation gradient (obtained from PCA axes scores)

Response to pressures: macrophytes
the Mediterranean GIG database was used to test the response of each metric to: individual types of pressures and to a general degradation gradient (obtained from PCA axes scores) River Macrophyte Index (RMI) Described in: Kuhar, U., Germ, M., Gaberščik,A., Urbanič, G Development of a River Macrophyte Index (RMI) for assessing river ecological status. Limnologica 41: Web page describing the national method: a/ RMI was calculated according to the following equation: The RMI was calculated using the following equation: where QAi = abundance of the taxa i from the group A, QABi = abundance of the taxa i from the group AB, QBCi = abundance of the taxa i from the group BC, QCi = abundance of the taxa i from the group C, QSi = abundance of taxa i from all groups (group A, AB, B, BC, C; taxa from the group ABC are not considered), nA = total number of taxa in group A, nAB = total number of taxa in group AB, nBC = total number of taxa in group BC, nC = total number of taxa in group C, nS = total number of taxa in all groups (group A, AB, B, BC, C; taxa from the group ABC are not considered). Response of assessment methods (mean IBMR ERQ calculated using Med GIG references, for IC type RM1,2,4) to individual pressures

Reference and Disturbed sites ICM
The Spearman correlations of the common metric to the individual pressures (based on reference and disturbed samples dataset) are: The above results show that the ICM index responds well to nutrients and especially to phosphates and ammonium. Response to pressures: diatoms Reference and Disturbed sites ICM (Spearman rho, p, n) General Morphology (class) -0.112, p<0.0004, 1007 General Hydrology (class) -0.123, p< , 1014 Riparian Vegetation (class) -0.323, p< , 979 O2 (% Sat.) 0.100, p<0.0023, 916 N-NH4+ (mg/l) -0.519, p< , 829 N-NO3- (mg/l) -0.254, p< , 858 P-Total (mg/l) -0.428, p< , 377 P-PO4 (mg/l) -0.608, p< , 803 BOD5 (mg/l) -0.221, p< , 391 Land Use (%) -0.220, p< , 1042 Agriculture (%) -0.242, p< , 1042 Urbanisation (%) -0.005, p>0.5, 1033

Response to pressures: macroinvertebrates
General degradation

Biological communities at reference sites: benthic diatoms
The SIMPER analysis (Bray-Curtis similarity; Primer 6.0) was used to determine taxa contributing the most (up to 90% of cumulative contribution) to the average similarity within type. ADMI – Achnanthidium minutissimum ADBI – Achnanthidium biasolettianum NCTE – Navicula cryptotenella GPUM – Gomphonema pumilum CAFF – Cymbella affinis ENCM – Encyonopsis microcephala ENMI – Encyonema minutum CPLA – Cocconeis placentula CPLI - Cocconeis placentula var. lineata ACLI – Achnanthidium lineare FGRA – Fragilaria gracilis CEUG – Cocconeis euglypta FBCP – Fragilaria biceps Representative taxa RM1, RM2, RM4 RM5 Average abundance ADMI 31.57 28,59 ADBI 7.84 CPLI 10.51 NCTE 1.48 ACLI 8.87 GPUM 2.59 1.80 CAFF 2.28 FGRA 2.45 ENCM 1.76 CEUG 2.06 ENMI 1.71 4.14 CPLA 1.99 FBCP 3.88 2.30

Methods intercalibration feasibility check
Intercalibration is feasible in terms of response to pressures and taxa specificity

HARMONIZATION IN MED GID
MACROPHYTES INTERCALIBRATING FOR THE FIRST TIME

IC MedGIG macrophytes steps
Whenever possible, River Macrophyte Index, RMI (Slovenian National Method) was computed in the entire database. The IBMR was also computed for the Slovenian sites. The IBMR was regressed against RMI in order to check the relatedness of indices and to convert RMI boundaries into the IBMR scale. Only sites from Greece, Italy, Portugal and Slovenia were used in the regression (Spain, Cyprus and France were not considered because the RMI was computed for too small number of sites).

Results RM1,2,4 6 countries – national method IBMR (same data acquisition, method and calculation) 1 country – Slovenia RMI 1) IC Option 1 – 6 countries/ all dataset 2) IC Option 3 – Slovenia r=0,677 OK p= OK slope of regression OK

RM1,2,4 – preliminary steps Calculate the EQR site values for Mediterranean GIG (MedGIG EQR), using the IBMR absolute values divided by the median of the IBMR values of MedGIG reference sites (previously screened with MedGIG criteria). Convert boundaries to IC EQR using regression analyses per MS: MedGIG EQR vs. National EQR IBMR reference sites. MS Regression equations CY y = x FR y = x GR y = x IT y = x PT y = x - 4E-05 SP y = x CY FR GR IT PT SP H/G 0.9031 0.9177 0.9861 0.8505 0.9966 0.9077 G/M 0.7981 0.8066 0.8661 0.7516 0.7474 0.6770 M/P 0.6931 0.7034 0.7461 0.6033 0.4983 0.4463 P/B 0.5881 0.5922 0.6324 0.4550 0.2491 0.2251

RM1,2,4 – Option 1 Boundary harmonization using IC Option 1: all countries except Slovenia (national method IBMR) The boundary bias criterion was used: boundary bias should be less than a quarter of the width of a class. We used the median of the boundaries to compute the boundary bias. We changed iteratively the values of the boundaries that did not meet the criteria until the median of the boundaries was included within the quarter of the class (High or Good).

RM1,2,4 – before harmonisation
CY FR GR IT PT SP Max 1.011 1.393 1.144 1.226 1.443 1.386 MedGIG_H/G 0.903 0.918 0.986 0.851 0.997 0.908 MedGIG_G/M 0.798 0.807 0.866 0.752 0.747 0.677 MedGIG_M/P 0.693 0.703 0.746 0.603 0.498 0.446 MedGIG_P/B 0.588 0.592 0.632 0.455 0.249 0.225 H width to Max 0.11 0.48 0.16 0.38 0.45 G width 0.10 0.12 0.25 0.23 M width 0.15 H/G bias -0.01 0.00 0.07 -0.06 0.08 G/M bias 0.02 0.03 0.09 -0.02 -0.03 -0.10 H/G bias_CW -0.09 0.04 0.61 -0.17 0.34 G/M bias_CW 0.22 0.31 0.76 -0.23 -0.11 -0.42 Median MedGIG H/G 0.9127 G/M 0.7749

RM1,2,4m – after harmonisation
CY FR GR IT PT SP Max 1.011 1.393 1.144 1.226 1.443 1.386 MedGIG_H/G 0.903 0.918 0.928 0.851 0.967 0.908 MedGIG_G/M 0.774 0.771 0.757 0.752 0.747 0.709 MedGIG_M/P 0.693 0.703 0.746 0.603 0.498 0.446 MedGIG_P/B 0.588 0.592 0.632 0.455 0.249 0.225 H width to Max 0.108 0.476 0.216 0.376 0.478 G width 0.129 0.147 0.171 0.099 0.220 0.199 M width 0.081 0.068 0.011 0.148 0.263 H/G bias -0.010 0.005 0.015 -0.062 0.054 -0.005 G/M bias 0.020 0.017 0.003 -0.003 -0.007 -0.045 H/G bias_CW -0.09 0.03 0.09 -0.17 0.25 -0.01 G/M bias_CW 0.24 -0.03 -0.23 Median MedGIG H/G 0.9127 G/M 0.7543

RM1,2,4 Boundary harmonization (SI) using Option 3 was done after conversion to the IC EQR IBMR scale using the regression equation. The boundary bias of RMI was computed using the median values after option 1 boundary harmonization. In case the RMI boundaries did not meet the criteria (boundary bias lower than a quarter of a class), these were moved until the median of the boundaries was included within a quarter of a class.

Boundary comparison SI
RMI_BS : benchmark standardized - RMI boundary values calculated with the median values of the MedGIG reference data. RMI_reg IBMR are the boundary values converted to the IBMR scale. RMI RMI_BS RMI_reg IBMR Max 1.1925 H/G 1.4454 0.8814 G/M 0.9412 0.7363 M/P 0.7059 0.5910 P/B 0.4706 0.4459 Min 0.2353 0.3007 SI Max 1.1925 MedGIG_H/G 0.8814 MedGIG_G/M 0.7363 MedGIG_M/P 0.5910 MedGIG_P/B 0.4459 H width to Max 0.311 G width 0.145 M width H/G bias -0.031 G/M bias -0.018 H/G bias_CW -0.101 G/M bias_CW -0.125 The harmonized median value H/G and G/M was fixed to intercalibrate RMI.

RM1,2,4 Class agreement was computed for all pair combination of National methods. Only sites that were classified as H, G or M according to both National Methods were used to compute class agreement. First, a piecewise transformation of the IC EQR IBMR values was performed using the formula: MinT – ((X – Min)*0.2) / (Max – Min) where MinT – Minimum of the new transformed class (0.6 for G and 0.8 for H), X – index value, Min –theoretical index minimum, Max – theoretical index maximum. Class agreement was computed as the mean absolute difference between the index values after piecewise transformation divided by 0.2 (the width of each class after piecewise transformation). Class agreement should be less than 1 (meaning that the mean differences should be less than the width of 1 class).

Summary of Boundary adjustements
RM1,2,4 Original Adjusted Final agreed national boundaries H/G G/M Cyprus Boundary 0.795 0.596 0.550 ↓ Bias CW -0.09 0.22 0.24 France 0.93 0.79 0.745 0.79* 0.04 0.31 0.25 Greece 0.75 0.56 0.66 0.39 0.61 0.76 0.09 Italy 0.90 0.80 -0.17 -0.23 Portugal 0.92 0.69 0.89 0.34 -0.11 Spain 0.95 0.71 0.74 ↑ 0.74* -0.01 -0.42 Slovenia 0.60 -0.101 -0.125 CY, FR, GR, PT could be more relaxed BUT THEY CHOSE TO MAINTAIN THE NATIONAL BOUNDARIES SP INCREASED G/M BORDER ALL CLASS AGREEMENT VALUES <1 *waiting for final official agreement, but agreed by macrophyte experts and national representatives

Ecological characterization between G and M ecological status
Major changes between G and M associated with: ↓species richness (median in G sites =12; M sites=7), ↑ cover and frequency of pondweed taxa, such as Potamogeton pectinatus and P. nodosus, macroalgae (e.g. Enteromorpha sp., Cladophora sp.), and other hydrophyes (Lemna gibba, Nuphar lutea ), and of some emergent species, such as Schoenoplectus lacustris.

Ecological characterization
loss and/or ↓ cover of bryophytes, mainly Rhynchostegium riparioides, Fontinalis antipyretica, Fissidens crassipes, Eurhynchium praelongum, Lunularia cruciata and Amblistegium riparium, and of some amphibious and hygrophyte species (Lotus pedunculatus, Carex elata, C. pendula). Some infrequent species in the data base, bryophytes (Bryum sp.) , isoetids, Juncus sp., Myosotis sp. were only observed in sites classified in Good ecological status. Some alien invasive species raise their abundance due the raise of pressures, this is the case of Azolla sp.

RM5 – relatedness checking
Sites MedGIG references CY 43 5 IT 13 - PT 30 3 SP 4 SI

RM5 – Option 1 (IT and PT) Boundaries Original Original MedGIG IT PT Max 1.1240 1.1538 1.1405 1.1535 H/G 0.90 0.93 0.946 0.930 G/M 0.80 0.70 0.859 0.700 M/P 0.65 0.46 0.729 0.460 P/B 0.50 0.23 0.599 0.230 Regression analyses per MS used to convert the national boundaries into comparable boundaries. IT: Y=0.8674x ; PT: Original Harmonized IT PT Max 1.1404 1.1535 MedGIG_H/G 0.946 0.930 MedGIG_G/M 0.859 0.700 0.726 0.728 MedGIG_M/P 0.729 0.460 MedGIG_P/B 0.599 0.230 Median H/G MedGIG 0.938 Median G/M MedGIG 0.780 0.727 H width to Max 0.194 0.224 G width 0.087 0.220 0.202 M width 0.130 0.240 -0.003 0.268 H/G bias 0.008 -0.008 G/M bias 0.080 -0.080 -0.001 0.001 H/G bias_CW 0.092 -0.036 0.036 G/M bias_CW 0.612 -0.346 -0.005 0.004

Final agreed national boundaries
Summary of Boundary adjustements/ RM5 RM5 Original Adjusted Final agreed national boundaries H/G G/M Italy Boundary 0,946 0,859 0.726 ↓ Not accepted due to the small database Bias CW 0.092 0.612 -0.005 Portugal 0,930 0,700 0.728 ↑ -0.036 -0.346 0.004 IT PT Original boundaries (national scale) H/G 0.90 0.93 G/M 0.80 0.70 Harmonized boundaries (national scale) 0.65 0.73 BUT: Only 3 reference sites all together for the type Are the IC2 results reliable in this case? ALL MS PARTICIPATING FEEL METHODS SHOULD BE MORE TESTED IN TEMPORARY RIVERS

BENTHIC MACROINVERTEBRATES

Total = 1384 samples Number of reference sites finally selected MedGIG criteria for macroinvertebrates, by MS and type. Spring-Summer samples are within parentheses. RM1 RM2 RM4 RM124 RM5 Total CY - 25(14) 12(9) 37 (23) FR 53(37) 6(5) 94(65) 153(107) 153 (107) IT 8(2) 7(6) 6(2) 21(10) 5(3) 26 (13) PT 8(8) 5(5) 13(13) 3(3) 16 (16) SI 6(4) 4-1*(3) 11(9) 1(0) 12 (9) SP 18(18) 1(1) 25(25) 44(44) 25-2*(23) 69 (69) 93 (69) 25 (23) 150 (106) 268(198) 46 (40) 314 (240) *sample eliminated due to exceptionally low ICMi value.

Option 2, minor differences in field data acquisition, sampling protocols and area sampled, and the way to express qualitative/quantitatively the data. The Slovenian method uses a different taxonomic level and cannot be applied to the datasets of the other MS. Common metric: ICMi EXCLUDED

IC type-specific biological communities representing the borderline between G and M status SIMPER analysis (no transformation; Bray-Curtis coefficient; up to 90% of contribution to av. Similarity; Primer 6) More chironomidae, baetidae, oligochaeta, caenidae and hydrobiidae in moderate sites

National EQR Original National EQR Harmonized Type H/G G/M H/G PT-Type 1 0.870 0.650 PT-Type 2 0.830 0.610 PT-Type 3 0.850 0.590 PT-Type 4 0.880 0.660 PT-Type 5 0.950 0.700 0.963 0.724 PT-Type 6 SP1-Type 1 0.780 SP1-Type 2 SP1-Type 3 0.810 SP2-Type 1 0.800 0.600 0.612 SP2-Type 2 0.740 0.550 0.632 SP2-Type 3 0.640 SP1-Type 4 1.019 0.931 1.019* 0.931* SP2-Type 4 0.620 FR-Type 1 0.940 IT-Type 1 0.970 0.720 IT-Type 2 IT-Type 3 IT-Type 4 0.730 SL-Type 1 SL-Type 2 SL-Type 3 CY-Type 1 0.972 0.729 CY-Type 2 0.982 0.737 ONLY SPAIN AND PORTUGAL HAD TO CHANGE THE NATIONAL BOUNDARIES

RM1,2,4 Class agreement was computed for all pair combination of National methods. Only sites that were classified as H, G or M according to both National Methods were used to compute class agreement. First, a piecewise transformation of the IC EQR IBMR values was performed using the formula: MinT – ((X – Min)*0.2) / (Max – Min) where MinT – Minimum of the new transformed class (0.6 for G and 0.8 for H), X – index value, Min –theoretical index minimum, Max – theoretical index maximum. Class agreement was computed as the mean absolute difference between the index values after piecewise transformation divided by 0.2 (the width of each class after piecewise transformation). Class agreement should be less than 1 (meaning that the mean differences should be less than the width of 1 class).

BQE MACROINVERTEBRATES All class agreement values lower than 1
PT vs. SP1 0,175 PT vs. SP2 0,260 PT vs. FR 0,234 PT vs. IT 0,551 PT vs. SL 0,240 0,299 PT vs. CY 0,336 SP1 vs. PT 0,172 SP1 vs. SP2 0,082 SP1 vs. FR 0,378 SP1 vs. IT 0,553 SP1 vs. SL 0,338 0.307 SP1 vs. CY 0,321 SP2 vs. PT 0,253 SP2 vs. SP1 0,094 SP2 vs. FR 0,329 SP2 vs. IT 0,621 SP2 vs. SL 0,289 0.319 SP2 vs. CY 0,331 FR vs. PT 0,236 FR vs. SP1 0,397 FR vs. SP2 0,325 FR vs. IT 0,663 0.360 FR vs. SL 0,210 IT vs. PT 0,559 IT vs. SP1 0,560 IT vs. SP2 0,626 IT vs. FR IT vs. SL 0,554 0,573 IT vs. CY 0,475 SL vs. PT 0,237 SL vs. SP1 0,344 SL vs. SP2 SL vs. FR 0,212 SL vs. IT 0,548 0.341 SL vs. CY 0,406 CY vs. PT CY vs. SP1 0,323 CY vs. SP2 0,328 CY vs. IT 0,473 CY vs. SL 0,415 0.381 % Class agreement BQE MACROINVERTEBRATES All class agreement values lower than 1 AVERAGE CLASS AGREEMENT

BENTHIC DIATOMS

Statistics of linear regression and Pearson correlation for MS methods EQR and the Intercalibration Common Metric (ICM). MS Method Type Linear regression Pearson r; p Cyprus/IPS RM4 ICM = IPS – r=0.97; p<<0.001 RM5 ICM = IPS r=0.91; p<<0.001 France/IBD RM1, RM2, RM4 ICM = IBD r=0.93; p<<0.001 Italy/ICMi ICM = ICMi – 0.005 r=0.82; p<<0.001 ICM = ICMi r=0.98; p<<0.001 Slovenia/ SESAR RM1, RM2 ICM = SESAR r=0.86; p<<0.001 ICM = SESAR r=0.96; p<<0.001 Portugal/IPS ICM = IPS – 0.066 ICM = IPS – 0.063 Spain/IPS ICM = IPS r=0.92; p<<0.001 ICM = IPS – 0.013 MS sampled differently (Slovenia – multi-habitat sampling). Not all MS sampled exclusively in riffle area (IT, SI) and sampling took place in several seasons of the year in some cases (CY, IT, SP). Option 2 for intercalibration has been chosen because we have different data acquisition and each MS has a different assessment method. Intercalibration Common Metric (ICM) is composed of IPS and TI

Physico- chemical data
HARMONIZATION IN MED GID Member State Number of samples Biological data Physico- chemical data Pressure data Cyprus 60 France 193 Italy 102 Portugal 120 Slovenia 39 Spain 608 1222 SITES PROVIDED REFERENCE SITES USING MED GIG CRITERIA MS RM1 RM2 RM4 RM5 Total CY - 11 92% 6 75% 17 85% FR 35 69% 8 62% 39 63% 82 65% IT 3 1 50% 2 67% 4 80% 10 71% PT 5 33% 16 55% SI 60% 100% SP 33 73% 19 106% 29 66% 84 76% 36 78% 81 18 217 70%

SIMPER analysis (no transformation; Bray-Curtis coefficient; up to 90% of contribution to av. Similarity; Primer 6) was performed to determine the taxonomic specificities Taxa contributing to dissimilarity (type RM124) Good (av. Abundance) Moderate ADMI - Achnanthidium minutissimum 14,11 5,44 NINC - Nitzschia inconspicua 4,32 7,18 APED - Amphora pediculus 5,76 4,91 CPLA - Cocconeis placentula var. placentula 6,7 1,87 PLFR - Planothidium frequentissimum 2,17 5,51 MPMI - Mayamaea permitis 1,37 4,3 NPAL - Nitzschia palea 2,28 4,6 NFON - Nitzschia fonticola 2,59 2,86 CEUG - Cocconeis euglypta 1,85 1,69 ESBM - Eolimna subminuscula 1,02 3,1 FULN - Fragilaria ulna var. ulna 2,65 1,42 MVAR - Melosira varians 1,92 EOMI - Eolimna minima 1,77 1,8 NDIS - Nitzschia dissipata var. dissipata 2,23 1,61 NAMP - Nitzschia amphibia 0,98 2,48 NCTE - Navicula cryptotenella 2,29 1,63 NGRE - Navicula gregaria 1,62 1,7 Higher relative abundance of some less sensitive taxa in “Moderate” class (Gomphonema parvulum, Nitzschia palea) as opposed to the dominance of sensitive taxa in quality class “Good” (i.e Achnanthidium minutissimum).

H/G National EQR G/M National EQR Original Final PT-Type 1 0,970 0,730 PT-Type 2 0,910 0,680 PT-Type 3 PT-Type 4 SP-Type 1 0,937 0,701 0,727 SP-Type 2 0,938 0,702 SP-Type 3 0,935 FR-Type 1 0,940 0,780 FR-Type 2 FR-Type 3 IT-Type 1 0,800 0,610 IT-Type 2 IT-Type 3 SL-Type 1 0,600 SL-Type 2 CY - Type 3 0,683 ONLY SP INCREASES THE BOUNDARY ORIGINAL BOUNDARIES FOR OTHERS RM5 H/G – National EQR G/M – National EQR Original Final Original G/M Final G/M PT-Type 5 0,940 0,700 PT-Type 6 0,800 0,600 0,651 SP-Type 4 0,935 CY-Type 4 0,958 0,981 0,718 IT-Type 4 0,880 0,650 Sl-Type 4 CY AND PT have an increase of the boundary ORIGINAL BOUNDARIES FOR OTHERS All class agreement values lower than 1

IC EXERCISE IN MED GID: ACHIEVEMENTS
All Mediterranean BQE were intercalibrated: macroinvertebrates, diatoms, macrophytes and fish Seven countries participated, all but Malta A reference condition common approach was developed and reference consistency checked New methods and countries were intercalibrated, improving the previous exercise. A larger and more complete data base was gathered A ring test to check for taxonomic inconsistencies was made in diatoms A list of common macrophytes with “aquaticity features” was developed Pressure response of data was checked using quantitative criteria Influence of seasonality on macroinvertebrates in temporary rivers was addressed IC results were translated into national systems. However Temporary rivers should not be considered intercalibrated for macrophytes The combination between benthic micro and macroflora was only discussed (as also at the X-GIG level) RM3-Large rivers still could not be intercalibrated due to lack of data.

Mediterranean GIG River BQEs Intercalibration Exercise

Similar presentations

Presentation on theme: "Mediterranean GIG River BQEs Intercalibration Exercise"— Presentation transcript:

Similar presentations

About project

Feedback

Log in

Auth with social network:

Mediterranean GIG River BQEs Intercalibration Exercise

Similar presentations

Presentation on theme: "Mediterranean GIG River BQEs Intercalibration Exercise"— Presentation transcript:

Similar presentations

About project

Feedback