• ROMANIA
  • PHYTOBENTHOS
  • LAKES

Method name

National (Romanian) Assessment Method for Natural Lakes Ecological Status based on Phytobenthos (Diatoms)

Acronym

RO-AMLP

Scientific development and confirmation of the national method for ecological status assessment of waterbodies (rivers, lakes) based on diatoms (phytobenthos), and completion of the intercalibration exercise.

1: lakes

Martyn Kelly,

Bowburn Consultancy, 11 Monteigne Drive, Bowburn, Durham DH6 5QB

Contents

1.Introduction...... 4

2.Description of the national assessment method...... 4

3.National data set...... 4

4.Typology...... 5

5.Pressure-impact relationships...... 7

6.Reference/benchmark conditions...... 12

7.Boundary setting procedure...... 16

8.Checking of WFD compliance and feasibility...... 17

8.1 Compliance checking...... 17

8.2 Feasibility checking...... 18

9.IC procedure...... 19

9.1 ROLN08 - “upland” lakes...... 19

9.2 “lowland lakes”...... 20

10. Normalisation of boundaries ...... 30

11.Conclusions 1

12. References...... 33

Appendix 1: Rationale for excluding ROLN06T from intercalibration

A1. Introduction

A2. Descriptive statistics

Conductivity

Magnesium

Total hardness

Total dissolved solids (TDS)

A3. Cluster analysis

A4. Classification analysis...... 42

A5. Composition of diatom assemblages from therapeutic lakes

Appendix 2: Procedure for computation of ecological status for Romanian lakesusing phytobenthos ......

1.Introduction

Intercalibration of phytobenthos-based methods of ecological assessment of lakes was completed as part of the second phase of intecalibration. The exercise was conducted across all GIGs, with 11 Member States participating. Romania was not part of this exercise; however, sufficient data have now been accumulated to allow options for status assessment using phytobenthos to be considered.

This report describes the Romanian phytobenthos assessment system for lakes and shows how this is compliant with the WFD Normative Definitions and that the class boundaries correspond with those agreed by the completed intercalibration exercise. However, the report goes on to show that lowland lakes in Romania behave atypically, when compared to lakes in other countries that have already been intercalibrated. There are two (possibly linked) reasons: factors other than nutrients (BOD in particular) appears to be having a strong influence on the diatom assemblages in these lakes and a number of taxa that are abundant in Romanian lakes are not represented in Rott’sTrophic Index, the intercalibration metric. Although it is not possible to intercalibrate the method, we believe that it gives an accurate picture of the state of Romanian lakes and will use it for ecological assessment in order to fulfil our obligations under the WFD. Work will continue to develop a practical approach to assessing lake status using phytobenthos.

2.Descriptionof the national assessment method

The national assessment method for phytobenthos uses diatoms as proxies for the entire phytobenthos community. Samples are collected two or three times a year from submerged stems of emergent plants or hard substrates,by brushing with a toothbrush. These are then digested using hydrogen peroxide and permanent slides are prepared. These are analysed in the laboratory and at least 300 diatoms (usually at least 400) are identified to species level and the number of each species counted. The main identification literature used is Krammer and Lange-Bertalot (1986-2004).

At the start of this exercise, no national metric had been proposed; section 5 of this report evaluates common diatom metrics used elsewhere in Europe for their suitability for use in Romania, and recommends the use of Rott’strophieindex (TI: Rott et al. 1999) as an appropriate measure for lake ecological assessment in this country.

3.National data set

Diatom count data and associated environmental information are available from 367 samples representing 59 lakes in Romania. These data were collected between 2010 and 2013 and are summarised in Table 1.

Table 1. Summary of data available for Romanian lake phytobenthosintercalibration exercise. See Table 2 for a description of lake types

Type / Number of samples
All / Benchmark / reference
Lakes / Samples / Lakes / Samples
ROLN01 / 22 / 127 / 4 / 16
ROLN02 / 10 / 10 / 1 / 9
ROLN03 / 1 / 8 / 1 / 8
ROLN04 / 3 / 20 / 3 / 20
ROLN05 / 5 / 44 / 1 / 8
ROLN06T / 6 / 40 / 0 / 0
ROLN07 / 3 / 12 / 1 / 4
ROLN08 / 8 / 28 / 1 / 9
ROLNCAPMLA01 / 1 / 8 / 0 / 0
Total / 59 / 297 / 12 / 74

4.Typology

The current Romanian typology is summarised in Table 2. The phytobenthosintercalibration exercisewas a cross-GIG exercise, in which lakes were divided into three “supertypes”: “low alkalinity”, “moderate alkalinity” and “high alkalinity”. This is possible because benthic diatoms are sampled from the shallow littoral zone of lakes and, therefore, major factors structuring other communities within lakes are less relevant as the focus is on a single habitat. This enabled methods from across the EU to be compared.

All Romanian lake types with the exception of ROLN08 fitted the criteria for the “high alkalinity” supertype. ROLN08 covered a range of alkalinity values, but most corresponded to the “moderate alkalinity” supertype. ROLN06T, on the other hand, has very high alkalinity and conductivity, associated with mineral rich springs, and this is distinguished as a separate type. As both water chemistry and the diatom assemblage show brackish influences (see Appendix 1), this type is not included in the intercalibration. ROLN09 (shallow, silicious, temporary lakes) has also been omitted as there are no similar lakes amongst those currently intercalibrated against which they can be compared.

In practice, ROLN01, ROLN02, ROLN03, ROLN04 and ROLN05 can all be considered as variants of a basic “lowland high alkalinity” type for the purpose of this intercalibration. In addition, ROLN07, representing a transitional type between the “lowland” and “highland” lakes has been included with this lowland type on the basis of the relatively high alkalinity ofthe lakes in this type and the general habitat of the lake littoral.

Table 2. The Romanian national lake typology.

Typology / General description / Ecoregion / Altitude (m) / Average depth (m)
ROLN01 / Lowland, very shallow, silicious, very small, small and medium size. / 12, 16 / <200 / <3
ROLN02 / Lowland, very shallow, calcareous, small, medium and very large size. / 12 / <200 / <3
ROLN03 / Lowland, very shallow, calcareous, very large size. / 12 / <200 / <3
ROLN04 / Lowland, very shallow and shallow, peat, small, medium and large size. / 12 / <200 / <3
ROLN05 / Lowland, shallow, silicious/calcareous, medium size. / 12 / <200 / 3-15
ROLN06T / Lowland, very shallow and shallow, silicious/calcareous.(therapeutic). / 12 / <200 / <3
ROLN07 / Hills and tableland, very shallow and shallow, silicious, very small size. / 10, 16 / 200-800 / < 15
ROLN08 / Highland, very shallow and shallow, silicious, very small size. / 10 / >800 / <-15
ROLN09 / Lowland - temporary lakes, very shallow, silicious, small and medium size - Not validated. / 12 / <200 / <3

Fig. 1. Range of alkalinity values associated with Romanian national lake types, based on data collected between 2010 and 2013. Horizontal lines show limits between types and supertypes: blue line: 0.2 meq L-1 (threshold between low and moderate alkalinity supertypes); red line: 1 meq L-1 (threshold between moderate and high alkalinity supertypes). Note that the scale on the Y axis is logarithmic.

5.Pressure-impact relationships

The phytobenthosintercalibration exercise addresses the effect of eutrophication on benthic algal assemblages. Although, in theory, any nutrient may be limiting, in practice attention focuses on the role of phosphorus. Preliminary analyses (Figs 1 & 2) indicated that total nitrogen is unlikely to be the limiting nutrient except in a few already eutrophic instances in Romania. Lakes with high values of TN tend to have elevated TP as well, so it should be possible to use TP to indicate the primary “stress” gradients within the dataset.

As most of the lakes in Romania are shallow, additional consideration needs to be given to the nature of “pressure-impact” relationships in situations where cause-effect relationships are known to be non-linear (Moss, 2010). The status of the phytobenthos needs to be considered alongside that of other BQEs in order to understand the potential impacts of both “top down” and “bottom up” influences.

Fig. 2. Relationship between Total Phosphorus (TP) and Total Nitrogen (TN) in Romanian lakes expressed as concentrations (left) and as the ratio between the two (right). Red line shows TN:TP = 16. All axes are shown logarithmically transformed.

Preliminary analyses established the performance of various widely-used diatom metrics available in the Omnidia package (version 5.3; Lecointe et al., 1993: Fig. 3). Note that the version of the TDI in this package is not the same as the metric currently used by UK and Ireland for assessing ecological status based on lake phytobenthos composition (Bennion et al., 2014). Linear regressions between the diatom metric and log10 TP and log10 Alkalinity indicate the ability of the metric to capture the pressure gradient, and the scale of interference from type / geological factors (Table 3). In all cases, both “pressure” and “type” were significant, though R2 was very low in the case of Sladacek’s index (INDSLA). The greatest contribution of alkalinity was seen in the case of the IBD, a “general degradation“ rather than a specific “eutrophication” index. The IPS, Rott’s TI and Rott’s SI all performed similarly with respect to both parameters.

The choice of Rott’s TI (Rott et al., 1999) as the basis for ecological status assessments in Romania is made for the following reasons:

  • Although Rott’s TI was not designed as a metric for assessing lakes, it performed well when applied to datasets from standing waters and was chosen as the “Intercalibration CommonMetric” for the cross-GIG phytoplankton intercalibration exercise (Kelly et al., 2014);
  • Rott’s TI was, moreover, designed as a “trophic”, rather than a “saprobic” or “general degradation” metric, as is the case for the IPS and Rott’s SI. It should, in theory be most sensitive over the range where phosphorus exerts a causal effect on the phytobenthos, rather than merely correlate with high phosphorus concentrations at a part of the gradient where other factors (e.g. low oxygen concentration) are exerting a stronger effect on community composition;
  • As Rott’s TI was used as the Intercalibration Common Metric,the intercalibration process will be more straightforward, as there will be no additional errors introduced by regressingthe national metric against the ICM in order to convert Romanian data to the common scale.

However, as the work continued, limitations of the TI when applied to Romanian lakes became apparent. This will be discussed in more detail below (see 9.2) but the outcome was that lowland lakes in Romania are also subject to relatively high levels of organic pollution (manifest as high BOD levels, in particular) which complicates a comparison with data from other Member States where the primary stress gradient is due to inorganic nutrients. The possibility of using the IPS as an alternative national metric were explored; however, anybenefits of using the IPS are cancelled out as this metric has to be converted back to TI in order to compare the boundary positions with those of other MS.

Table 3. Regression parameters between diatom index values and TP alone and in combination with alkalinity. Both TP and alkalinity were significant in all cases; “difference” indicates the increase in R2 through addition of Alkalinity to the equation.

Index / TP / TP + Alkalinity
R2 / Slope / Significance / R2 / Difference
IBD / 0.223 / -3.223 / *** / 0.341 / 0.118
TDI_20 / 0.152 / -2.509 / *** / 0.173 / 0.021
IPS / 0.242 / -3.038 / *** / 0.286 / 0.044
Rott TI / 0.221 / -2.816 / *** / 0.270 / 0.049
Rott SI / 0.263 / -2.534 / *** / 0.297 / 0.034
EPI-D / 0.203 / -2.430 / *** / 0.239 / 0.036
INDSLA / 0.039 / -1.162 / *** / 0.054 / 0.015

Fig. 3. Scatterplots showing the relationship between candidate diatom metrics and total phosphorus (TP, left) and alkalinity (right). Both X axes are displayed on a logarithmic scale.

Fig. 3 (continued)

6.Reference/benchmark conditions

Table 4 summarizes the criteria used to select reference sites. Although many lakes corresponding to ROLN08 have low levels of human impact, only one can be defined as a ”reference lake” following all the criteria listed in Table 4(Lake Bucura, ROLN08). Other lakes may be added in the near future, once screening is complete. ”Benchmark” lakes, representing least disturbed conditions for this type, identified during the development of the macrophyte method.

The following criteria were used to identify benchmarks forlakes corresponding to E-C1:

  • no major point sources in catchment,complete zonation of the macrophytes in the littoral zone,
  • no (or insignificant) artificial modifications of the shore line,
  • no mass recreation (camping, swimming, rowing)
  • low/moderate fishing (fish standing stock <50kgha–1)
  • Based on TP, TN, COD values and intensity of fishing a combined stressor was developed. The stressor ranges from 0–4. Lakes considered as alternative benchmark sites have a combined stressor value<1.5.This means that:

Fishing is low(fish stock <50kgha–1)

Vegetation period mean TP <115 µgl-1

Vegetation period mean TN <1550 µgl-1

Eleven lakes fulfilled these alternative benchmark criteria, representing all lowland lake types except ROLN06T.

Benchmark sites from lowland lakes had significantly lower values of TP and TN than non-benchmark sites, but also had lower values of variables reflecting geological type (alkalinity and conductivity) (Fig. 4). There was no significant difference in TI or IPS values between benchmark and non-benchmark sites for lowland lakes (Fig. 5). These benchmark lakes were not used to set class boundaries; however, they were used to validate the boundaries, with an expectation that the average condition of these lakes should be ”good ecological status”.

A single ”true” reference site is available for ROLN08 at present although, as Figs 6 and 7 show, neither this nor the single ”benchmark” site for this type are typical for the type as a whole. The reference site has water that is softer than most of the other lakes, whilst the water at the benchmark site is harder. There is no difference in nutrient concentrations, but note that, overall, concentrations of nutrients are much lower in ROLN08 than in the lowland lakes. The reference site has lower Ti than other reference sites used in the moderate alkalinity intercalibration (Fig. 8); however, this may also reflect the relatively low alkalinity of this site. Lacu Rosu is a barrier lake in an area of volcanic rocks in the Eastern Caparthians which fulfils the type criteria for ROLN08 in all respects, even if it is atypical for the ”moderate alkalinity” subtype.

Table 4.Criteria for establishing reference conditions of Romanian lakes.

  1. Assessment criteria
high ecological state for biological quality elements and supporting hydromorphological elements and good ecological state for supporting physico-chemical elements.
  1. Land use, agriculture, forestry
influences of urbanization, land use and forestry must be reduced as much as possible;
land use: > 85% natural (eg. “natural” forests, wetlands, marshes, meadows, pastures);
without intensive crops (including vines), in the surroundings;
≤5% urbanization and peri-urban areas in the surroundings.
  1. Pollution sources
reduced impact of wastewater from scattered households = that can not be connected to a centralized wastewater collection system (ie <10 inhabitants/km2) in the whole basin;
values of specific synthetic pollutants must be below detection limit for the most advanced analytical techniques that are available; the values for specific non-synthetic pollutants must be at most equal to the natural background value;
nodirect discharges of treated or untreated wastewater.
  1. Hydromorphological alterations
the level of direct morphological alteration (eg bank structures, river profiles and lateral connectivity) must be reduced;
only water abstractions that result in insignificant reduction of the flows that have very limited effects on the quality elements are allowed;
only flow regularizations that result in insignificant reduction of the flow that have very limited effects on the quality elementsare allowed;
absence/ minor influence of artificial barriers upstream of the section;
absence/ minor influence of artificial barriers downstream of the section;
absence of cross artificial structures that can reduce natural water flow speed;
absence of dams on short sectors (to protect against flooding).
  1. Riparian vegetation
riparian vegetation must be consistent with the type and geographical location of the river.
  1. Biological pressures
Introduction of alien species
compatible with a minor alteration of indigenous biota by introduction of plant or animal species;
without alterations caused by invasive plant or animal species.
Fisheries and aquaculture
fishing operations should allow conservation of the structure, the productivity, the function and the diversity of the ecosystem (including habitat and ecologically related dependent species) which the fishery exploitation depends on;
non indigenous fish stocks should not significantly affect the ecosystem structure and function;
without bio-manipulation.
  1. Other pressures
Recreational uses
no use of reference sections for recreational purposes (no intensive camping, swimming, boats, sailing).

Fig. 4. Variation in values of key chemical variables (a: alkalinity; b: conductivity; c: total phosphorus (TP); and, d: total nitrogen (TN)) in benchmark and other lakes for all Romanian lowland lake types (excluding soda lakes). Results of Wilcoxon test for significance of differences: alkalinity: P < 0.001; conductivity: P < 0.001; TP: P < 0.01; TN: P < 0.001.

Fig. 5. Variation in values of TI and IPS between benchmark and other lakes for all Romanian lowland lake types (excluding soda lakes). Result of Wilcoxon test: Not significant in either case.

Fig. 6. Variation in values of key chemical variables (a: alkalinity; b: conductivity; c: total phosphorus (TP); and, d: total nitrogen (TN)) in benchmark and other lakes for ROLN08 (highland lakes). Results of Kruskal-Wallis tests for significance of differences: alkalinity: P < 0.01; conductivity: P < 0.01; TP: Not significant; TN: not significant.

Fig. 7. Variation in values of TI (left) and IPS (right)between reference, benchmark and other lakes for ROLN08 (highland lakes). Result of Kruskal-Wallis test: Not significant in either case.

Fig. 8. Comparison between TI values for the ROLN08 reference site (“RO”) and other reference sites used in intercalibration of the “moderate alkalinity” supertype (“GIG”). Result of Wilcoxon test: P < 0.01.

7.Boundary setting procedure

The issue of establishing status class boundaries for Romanian lakes is complicated by the complete absence of reference sites for the “lowland” types, and a single, possibly atypical, reference site for ROLN08, the “upland” type. For this reason, the average of reference valuesused elsewhere in the cross-GIG exercise have been used as the denominator in EQR calculations. These are: 1.88 for lowland lakes and 1.38 for upland lakes, expressed as TI. These values equate to IPS values of 15.8 and 18.0 respectively (Fig. 9). These values were used to derive boundaries between status classes (Table 5). These are presented for both TI and IPS although, for reasons explained above, the IPS cannot be intercalibrated either.