1. Lakes Northern GIG Phytoplankton (comp) / Eutrophication
FI: Ansa Pilke and Liisa Lepistö, Finnish Environment Institute NO: Dag Rosland, Norwegian National Pollution Control Authority Robert Ptacnik, NIVA, Anne Lyche Solheim, NIVA/JRC SE: Mikaela Gönzci, Swedish EPA and Eva Willèn, SLU UK: Geoff Phillips and Sian Davies, Environmental Agency for England and Wales IE: Deirdre Thierney, and Wayne Trodd, Irish EPA

2. Common metric:
% Cyanobacteria, defined as % of total phytoplankton biovolume:
All Cyanobacteria, excluding Chroococcales, but including Microcystis.
The following genera are included:
Achroonema, Anabaena, Aphanizomenon, Cylindrospermopsis, Gloeotrichia, Limnothrix, Lynbya, Oscillatoria, Phormidium, Planktolyngbya, Planktothrix, Pseudanabaena, Tychonema, Microcystis, Woronichinia.
Only late summer samples used, max 4 obs./lake

3. No difference between clearwater types
No difference between humic types
proportion Cyanobacteria
proportion Cyanobacteria
But clearwater types different from humic types

4. Types aggregated to two major types:
Type description
Countries participating,
LN1
Lowland, mod. alk., clear, shallow
NO, UK, IE
LN2a
Lowland, low alk, clear, shallow
NO, SE, FI, UK, IE
LN2b
Lowland, low alk, clear, deep
NO, UK
LN3a
Lowland, low alk. mesohumic, shallow
LN5a
Boreal, low alk., clear, shallow
(may also include high latitude lakes)
NO, SE,
LN6a
Boreal, low alk., mesohumic, shallow
LN8a
Lowland, mod alk., mesohumic, shallow
NO, SE (?), FI, UK, IE
Clear
Humic
Latitude is maybe equally important in the Scandinavian countries
LN8, not enough data

5. Setting reference conditions
Using median of values from ref. lakes
170 ref.lakes from clearwater types
40 ref.lakes from humic lake types
Ref. values:
Clearwater lakes: 1% Cyanobacteria
Humic lakes: 2% Cyanobacteria
These values are also consistent with response curves for these two major types

6. Setting boundaries – starting point
Could we use the response curves and agreed chla boundaries directly?
Response curves not useful because:
If using the agreed G/M chlorophyll boundaries, the corresponding % Cyanobacteria was so low (<5% for all types) that this would not represent any real change in the taxonomic composition of the phytoplankton community, and thus not be compliant with the normative definitions.
Also the differences between the ref. value, H/G and G/M boundaries would be so small (1, 2 and 5% for Clearwater lakes, and virtually no difference for humic lakes due to a flat reponse curve, see Annex C), that it would be impossible to distinguish the different classes due to the uncertainty of analyses.
Thus a different approach was developed

7. Setting boundaries – new probabilistic approach
Divided all late summer samples (July – Sept.) into two groups:
reference lakes with chla lower than the mean H/G boundary (< 4 µg/L in clear lakes and < 5 µg/L humic lakes)
impacted lakes (from moderate to bad status) with chla higher than the mean G/M boundaries (> 7 µg/L in clear lakes and > 9 µg/L humic lakes)
Box-plots used to show the statistical distribution of samples (proportion of observations) exceeding different values of % Cyanobacteria.
Such box-plots were made for ref. lakes and for impact lakes for each major lake type.
Different values of % Cyano were tested to find which ones that best would separate the reference samples and impact samples for the two major lake types.

8. clearwater Ref lakes = REF Impacted lakes = no-R humic Probability
Probability
Ref lakes = REF
Impacted lakes = no-R
humic
Probability

9. Setting G/M boundaries
Decided which value of % Cyanobactreria that could be used as the G/M boundary for each major lake type. Three criteria were used to make this decision:
the mean probability of observations exceeding a certain value of % Cyanobacteria had to be close to zero for reference samples. This is based on the need for managers to be able to distinguish reference sites from clearly impacted sites (< good status) with a very high probability.
the mean probability of observations exceeding a certain value of % Cyanobacteria had to be significantly different between reference samples and impact samples.
the boundary value should be high enough to be compliant with the normative definitions, e.g. the % Cyanobacteria in the impacted sites should represent a real change in the taxonomic composition of the phytoplankton, and also represent a real risk for undesirable secondary impacts, such as Cyanotoxins. There was a general expert agreement within the NGIG group that this value should be at least 20% Cyanobacteria.

10. clearwater Ref lakes = REF Impacted lakes = no-R humic Probability
Probability
Ref lakes = REF
Impacted lakes = no-R
humic
Probability

11. Setting boundaries – H/G boundaries and EQRs
H/G boundary was judged from the need to distinguish ref. sites from impacted sites with an uncertainty in phytoplankton composition analyses that is at least 20%. The difference between the G/M and H/G boundary thus must be at least 20%.
The final step was to calculate the EQRs. To avoid too low EQRs we normalized the ratio, using the following formula:
EQR = (1- boundary value) / (1-ref.value).

12. Results (preliminary)
Boundary Clearwater lakes Humic lakes
Ref value 1% 2%
H/G value % %
H/G EQR
G/M value % %
G/M EQR

13. Next steps untill July
Testing common metric vs. national metrics for SE (ready now) and UK (expected ready in early May)
New NGIG meeting in Oslo 25th May to discuss results of the tests and accept, adjust or reject the common metric and the preliminary boundaries
Revise the milestone report before ECOSTAT in July