Transactions on Ecology and the Environment vol 65, © 2003 WIT Press, www.witpress.com, ISSN 1743-3541

Definition of seasonal phytoplankton events for

analysis of long term data from coastal waters of the southern with respect to the requirements of the European Water

Framework Directive

T. Rieling, S. Sagert, M. Balmwart, U. Selig & H. Schubert Institute of Aquatic Ecology,

University of , Germany.

Abstract

The key feature of the European Water Framework Directive (EU-WFD) is to measure water quality with respect to ecological categories, i.e. in terms of the community structure and functioning of natural ecosystems. For the use of phytoplankton data in comparative ecosystem assessment it is important to compare similar seasonal phytoplankton stages since phytoplankton runs through different succession stages throughout the year, all characterised by differences in biomass and taxonomical composition. In this paper calculable seasonal markers are suggested from the analysis of a long term data set indicating important successional stages at different types of coastal waters in the southern Baltic Sea. In order to test the applicability of the temporal markers for water quality assessment, phytoplankton biomass was calculated for the defined seasonal events and compared to annual mean biomass calculations. While the calculation of the annual mean failed to indicate different trophic states, the event based biomass calculation showed clear differences between sampling sites, reflecting the pronounced differences between the stations concerning their nutrient regime. Thus, the proposed phytoplankton event definition is a recommendable tool for the application of biological metrics in evaluating coastal ecosystems with respect to the needs of the EU-WFD.

Transactions on Ecology and the Environment vol 65, © 2003 WIT Press, www.witpress.com, ISSN 1743-3541

104 Water Pollution \//I: Modclling, Mca~uringand Prediction

1 Introduction

The EU Water Framework Directive (WFD) aims to establish a framework for the protection of inland, transitional and coastal waters with significant emphasis on the ecological status of aquatic ecosystems. The key feature of the directive is to measure quality with respect to ecological categories, i.e. in terms of the community structure and functioning of natural ecosystems. For this purpose the directive demands considering phytobenthos, zoobenthos, fish and phytoplankton for the assessment of the ecological state. Despite a long tradition in ecological studies at the southern Baltic coast [l, 2, 3, 4, 51 no biological evaluation system exists for the indication of the ecological state, so far. This lack is mainly caused by the extreme variability of hydrological and geochemical parameters in the coastal area (salinity, water exchange, resuspension events, nutrient loading fiom the sediment), making it difficult to assign especially specific phytoplankton pattern to certain load conditions. Thus, a typology of coastal waters into distinct ecoregions in the sense of the EU-WFD is necessary for the development of standardised evaluation procedures. A second problem in the biological assessment of ecosystem, again especially for the use of phytoplankton as indicative biocoenoses for ecosystem evaluation, is a high annual variability of phytoplankton biomass and taxonomic composition. Due to high reproduction rates phytoplankton runs through different succession stages throughout the year all characterised by differences in biomass and taxonomical composition [6, 71. Therefore, it is important to define these succession stages for analysis of long term data, to guarantee the analysis of comparable seasonal phytoplankton stages (e. g. spring bloom, summer maximum). Moreover, the specific hydrodynamic features (salinity gradient, dimictic shallow waters) of the Southern Baltic coast make it difficult to apply well known seasonality models for phytoplankton cycles in lakes and marine waters to predict the occurrence of seasonal stages in time and space. Thus, there is a need for the development of reliable procedures to define seasonality in various coastal waters. Potentially, indicators for seasonality can be derived from the analysis of long term data sets. In this work, phytoplankton data fiom 23 coastal stations at the southern Baltic Sea were analysed, to search for specific markers of seasonal stages.

2 Material and methods

Altogether 2985 data sets from 23 sampling sites were used for the analysis. Samplings were made from 1988-1999 in irregular intervals of 2-8 weeks by the Landesamt aNaturschutz und Geologie of -Vorpommern, LUNG

(Germany). The sampling area covered a broad spectrum of oligo- to mesohaline sites along the coast of Mecklenburg-Vorpornmern, which represents an almost representative set of sites along the different coastal types of the southern Baltic Sea (Fig. 1). Phytoplankton was sampled from surface waters with a Niskin bottle sampler, fixed with formaldehyde and counted using an inverse

Transactions on Ecology and the Environment vol 65, © 2003 WIT Press, www.witpress.com, ISSN 1743-3541

microscope according to UTERMOHL [g]. Phytoplankton biovolume was calculated approximating the cell shape to simple geometrical solids according to ROTT [93. Taxonomic resolution was performed on the species level as far as possible. For the analysis of seasonal patterns single species were aggregated to higher taxonomical classes (Cyanobacteria, Chlorophyta, Cryptophyta, Bacillariophycea, Dinophycae). Due to the high year-to-year shifts of phytoplankton biomass at each station it was not possible to define seasonal events from threshold values of indicative phytoplankton groups. Also the frequently used stacked ordination of the relative proportion of group-specific biomass to total phytoplankton biomass did not show clear seasonal pattern due to high variability and loss of information from less abundant taxa. Though, phytoplankton biomass of a specific taxon was set in relation to total biomass

(biomassi * biomass total-'), and the occurrence of a specific seasonal event (eventi) was defined as the situation when the relative portion of a defined taxon exceeded its annual mean value by 1 standard deviation (c)(equation 1).

N biomass,

biomass, 4 biomasslo,o, + biomassi >( ) G event, equation (1) biomassIolnl bi~mass,,,,~

From this calculation a binary monthly matrix of events over the time of investigation was obtained, indicating a temporal window when the relative proportion of a single group (Cyanobacteria, Chlorophyta, Cryptophyta, Bacillariophycea, Dinophycae) exceeded the average annual contribution of this group. For comparison between sampling sites, site specific event patterns were compared pair wise for occurrence of common events during the time of investigations and related to the number of months sampled at the same time.

The calculated similarity matrix served as the basis for cluster analysis using NCSS. For cluster analysis similarity values were transformed to dissimilarities (dissimilarity = 1 - similarity value) and clustered using Complete Linkage procedure to maximize selectivity between the stations.

To gather information about the power of the event based analysis to indicate a good ecological state, 5 stations different in their degree of nutrient loading from the river Odra were analysed for their degree of eutrophication. For this purpose, mean phytoplankton biomass at a defined taxonomic seasonal event (diatom event) and the mean biomass over the year were calculated for each station over the entire investigation period. In order to test the applicability of the obtained values for water quality assessment, calculated biomass was compared to a biomass based system of classification suggested by HEINONEN[l01

Transactions on Ecology and the Environment vol 65, © 2003 WIT Press, www.witpress.com, ISSN 1743-3541

106 Water Pollution \//I: Modclling, Mca~uringand Prediction

Longitude (E)

Figure 1: Sampling area and location of sampling stations.

3 Results

The analysis of the occurrence of different seasonal taxonomic events revealed clearly different temporal and spatial patterns at different sampling sites of the Southern Baltic coast. Considering the whole data pool, the number of Cyanobacteria and Chlorphyta events ranged from 0-3 events a-' while the number of Bacillariophycae, Dinophycea and Cryptophycea events ranged from

0-4 events a 'I. No negative correlation was found for the number of missing data values and the number of events at a sampling site, indicating the screening frequency to be high enough to identify the maximum number of events. In order to estimate the spatial heterogeneity between the sampling sites, stations were clustered by their similarity in the distribution of seasonal events in the entire sampling period. Figure 2 shows a dendrogram of the cluster analysis on the basis of transformed percentual similarity between individual station pairs. Setting a threshold value of 0.35, stations divided into three groups with different seasonal event patterns, being well related to their geographical exposition to freshwater influence from the main rivers. Cluster 1 covers sampling sites UW1-UW6 located in the Warnow estuary. Cluster 2 includes stations strongly influenced by the hydrodynamic regime of the river Odra (OB1- 0B4) and the eutrophicated stations S23, S66 and GB 19, located in the transition zone between the inner and the outer coastal waters of the investigation area. The third cluster mainly consists of Baltic Sea sites (05, 09, 011, 022). Furthermore, the Baltic Sea influenced sites in semi enclosed waterbodies of the transition zone (WB3, KB90, DB6, P42) clustered with the stations of cluster 3.

To get an overview about temporal and spatial differences in the occurrence of the defined taxonornical events, number of events during a specific month were normalised to the number of months with investigations at a specific site

Transactions on Ecology and the Environment vol 65, © 2003 WIT Press, www.witpress.com, ISSN 1743-3541

(Fig. 3 a-e). Distribution of the specific event frequencies over the set of stations and over the year showed significant seasonal differences between the different clusters of stations. Cyanobacteria events were most frequently found in cluster 3 (Baltic Sea) as well as in cluster 2 (transition zone) during late summer from July to October. Inside the Baltic Sea cluster, sampling sites influenced by freshwater (P42) as well as the semi enclosed sampling site KB90 showed a second peak of Cyanobacteria events in late autudearly winter. In contrast to the event distribution in clusters 2 and 3, a clear increase of Cyanobacteria event frequency appeared in the Warnow estuary cluster during spring. The separation between the Baltic Sea influenced and Warnow estuary sites was also indicated by the repeating development of Cryptophyte events. While at the Baltic Sea influenced sites as well as in the transition zone Cryptophyte events were registered in spring as well as in autumn, no distinct seasonal pattern of Cryptophyte events was found at the estuarine Warnow sites. Diatom events were commonly found in spring between February and April over the complete set of stations. Especially, the river influenced sites OB1-0B4 of the transition zone were characterised by high frequencies of Diatom events. Event frequencies at these stations as well as at the Baltic Sea stations 011 and 09 reached values between 0.8 and 1.O, showing the occurrence of Diatom events as highly indicative for the spring season. Occurrence of Dinophyte events indicated distinct differences between cluster 1, 2 and 3. At the estuarine sites the majority of Dinoflagellate events clustered in spring around March, while at the Baltic Sea influenced sites a clear Dinoflagellate signal was found in September. Ths accumulation of Dinophyte events could not be found in the transition area, where events covered the whole late autudearly winter period. A similar situation was found for the distribution of Chlorophyta events. Neither for the Baltic Sea or the transition zone cluster nor for the estuarine cluster apparent temporal windows were found, assigning the Chlorophyta events to specific phytoplankton succession stages. Due to their wide distribution over the months and stations Chlorophyta events could not serve as a distinct successional marker.

Summarising the results presented above, the following seasonal markers are suggested to indicate important successional stages for the analysed types of coastal waters in the southern Baltic Sea:

Cryptophyte events - markers for the winter state of phytoplankton succession in the Baltic Sea as well as in the transition zone. Diatom events - markers for the early spring bloom of the phytoplankton succession for all types of coastal waters at the southern

Baltic coast Cyanobacteria events - markers for a summer aspect of phytoplankton succession in the Baltic and in the transition zone, and early spring indicator at the estuarine sites.

Dinophyte events - markers for an autumn aspect inside the Baltic cluster of sites and marker for a spring event at the estuarine sites

Transactions on Ecology and the Environment vol 65, © 2003 WIT Press, www.witpress.com, ISSN 1743-3541

108 Water Pollution \//I: Modclling, Mca~uringand Prediction

Estuanne sites

Trans~t~on zone

Baltic influenced sites

IIIIIIIIIIIlr 1.00 0,75 0.50

distance

Figure 2: Dendrogram (furthest neighbour) of calculated similarities W] on the basis of the common occurrence of all calculated taxonomic events (Cyanobacteria, Chlorophyta, Cryptophyta, Bacillariophycea and Dinophycae).

Transactions on Ecology and the Environment vol 65, © 2003 WIT Press, www.witpress.com, ISSN 1743-3541

eyanobaderia events

cceaeevents

hClyptophyta events

Figure 3 a-e: Frequency [month-l] of the occurrence of specific taxonomical

events throughout the year. The entire investigation period served as basis for the frequency calculation.

Transactions on Ecology and the Environment vol 65, © 2003 WIT Press, www.witpress.com, ISSN 1743-3541

1 10 Water Pollution \//I: Modclling, Mca~uringand Prediction

4 Discussion

From the analysis of the LUNG long term data set 4 internal seasonal markers are suggested to define site specific successional stages in 3 different coastal zone types at the Southern Baltic: winter Cryptophyte event, spring Diatom event, summer Cyanobacteria event and spring or autumn Dinophyte event. The occurrence of a winter Cryptophyte bloom was also demonstrated for coastal zones of the southern Baltic by WASMUNDet al. [l11 who supposed Cryptophytes to be limited by low light intensities and low temperatures. Besides their photochemical acclimation the main representative of the Cryptophyte community (Rhodomonas salina) has a high potential for mixotrophic nutrition, allowing a positive net growth rate even at low irradiances [12]. The regular occurrence of Diatom events in the early spring is resulting fiom the fast growth of diatoms caused by high nutrient input due to enhanced river runoff and turbulent mixing. Cyanobacteria as well as Dinophyte events in late sumrnerlearly autumn indicated a later state of succession, consisting of more grazing resistant taxa generally prevailing in meso- to hypertrophc waters [3, 61. The EU-WFD prescribes the evaluation of anthropogenic impact on European waters on the basis of biological criteria. For this purpose different phytoplankton community parameters are suggested to give information about the degree of anthropogenic impact which mainly results hom eutrophication processes. For example, the relationship between the abundance and biomass of specific algae groups was formalized in indices for the assessment of water quality in lakes [13]. Furthermore, TIKKANEN & WILLEN [l41 summarised the occurrence of indicator species in nutrient rich and nutrient poor lakes to serve as a tool for the evaluation of the loading conditions in Finnish lakes. Besides the taxonomic composition of phytoplankton communities, total phytoplankton biomass is used to give information about the trophic state in different aquatic ecosystems. HETNONEN[l01 suggested 6 treshold values of phytoplankton biornass (0.2, 0.5, 1.0, 2.5, 10.0, >10.0 mm3 I-') for the classification into different trophic states (ultraoligotrophic, oligotrophic, mesotrophic-l, mesotrophic-2, eutrophic and hypertrophic). Investigating Norwegian lakes BRETTUM[l51 recommended a similar classification system with a higher resolution of 7 classes. For marine or brackish systems of the southern Baltic Sea WASMUNDet al. [5] suggested an annual mean biomass of 2500 mg mm3to discriminate between mesotrophic and eutrophic coastal waters of the southern Baltic referring to the classification system of HEINONEN[10]. However, as also shown here, phytoplankton biomass passes through different successional stages over a year. In late winterlearly spring small flagellates develop even beneath an ice cover [7]. Following this event, diatom biomass increases forming a spring bloom in limnic as well as marine waters. After this bloom a clearwater phase develops, reflecting the grazing pressure fiom different zooplankton groups. Later in the year seasonal progression of the phytoplankton community runs through intermediate stages reaching a summer state with the dominance of larger, K-selected species. Ths progressive succession pattern does not only lead to differences in the taxonomic

Transactions on Ecology and the Environment vol 65, © 2003 WIT Press, www.witpress.com, ISSN 1743-3541

composition of the phytoplankton community, but also to high seasonal differences concerning total phytoplankton biomass. Thus, a simple and rigid summary of phytoplankton biomass data over defined temporal intervals (growth season or yearly mean values) is problematic, since it does not consider the

important seasonal variability. Because of this temporal variability, application of phytoplankton data as metrics for water quality assessment need a definition of seasonal events to obtain a basis for comparable investigations of different sampling sites and periods. For this purpose BEUKEMA& CADEE[l61 suggested

from analysis of long term data increasing cell concentrations during Phaeocystis pouchetii bloom events to be indicative for increasing eutrophication processes. Furthermore, the massive development of Cyanobacteria blooms under eutrophicated conditions in lakes is well known as an indicator of nutrient loading [6].

The analysis of the phytoplankton data of the LUNG-dataset focussing on the seasonal occurrence of specific taxonomic events showed clear advantages compared to analysis based on yearly mean values or mean values over the growth season. Calculation of phytoplankton biomass for stations of the

transition area (cluster 2) on the basis of mean yearly biomass only led to slight differences between the sampling sites. This result was contrary to marked differences between the sampling sites concerning their influence of freshwater by the heavily eutrophicated river Odra (Tab. 1). Thus, the annual mean of total

phytoplankton biomass failed to indicate the different trophic states. However, restricting phytoplankton biomass calculations to that of Diatom events, pronounced differences could be found between the sampling sites. At the spring stage highest mean value for the Diatom biovolurne concentration (5.8 mm3 l") was found at station OB1 in the Swina river mouth, which delivers nutrient rich

water of the inner Odra estuary to the Baltic Sea. In contrast, low biovolume concentrations were found at sites with lower nutrient input from the river Odra, and this clearly proved the applicability of the event-based analysis for ecosystem assessment. Using the trophic scale of HEINONEN[10], Odra-

influenced sites (OB1-0B4) covered a spectrum of eutrophic to hypertrophic conditions. In correspondence to their lower riverine nutrient input, with the event-based approach the sampling sites S23 and S66 were classified as mesotrophic locations. At the current stage of analysing long-term datasets, the proposed phytoplankton event definition is a recornmendable tool for the

application of biological metrics in evaluating coastal ecosystems with respect to the needs of the EU-WFD.

Transactions on Ecology and the Environment vol 65, © 2003 WIT Press, www.witpress.com, ISSN 1743-3541

1 12 Water Pollution \//I: Modclling, Mca~uringand Prediction

Table 1: Mean yearly phytoplankton biovolume (mm3 1-l) of the transition zone calculated fiom the growth season interval or from the time when a Diatom event occurred.

Acknowledgments

This work was financially supported by the BMBF (Bundesministerium fir Bildung und Forschung, Germany, "ELBO", Az0330014), by the LUNG

(Landesamt fur Umwelt, Naturschutz und Geologie Mecklenburg-Vorpommem) and by a European Community Grand ("CHARM", EVK3-CT-2001-00065).

References

[l] Trahrns, K., Zur Kenntnis der Salzverhaltnisse und des Phytoplanktons der Hiddenseer und dex Riigenschen Boddengewasser. Arch. Hydrobiol. 32, pp.

75-90, 1937. [2] Wasmund, N., Characteristics of phytoplankton in brackish waters of different trophic levels. Limnologica 20 (l),pp. 47-51, 1989. [3] Wasmund, N. & Schlewer, U., lherblick zur Okologie und Produktions-

biologie des Phytoplanktons der Darss Zingster Boddenkette (siidliche Ostsee). Rostock. M~eresbiolog.Beitr. 2, pp. 41-60, 1994. [4] Bahnwart, M., Hiibener, T. & Schubert, H., Downstream changes in phytoplankton composition and biomass in a lowland river-lake system (Warnow River, Germany). Hydrobiologia 391, pp. 99-11 1, 1999.

[S] Wasmund, N., Nausch, G., Postel, L., Witek Z., Zalewski, M., Grornisz, S., Lysiak-Patuszak, E., Olenina, I., Kavolyte, R., Jasinskaite, A., Miiller- Karulis, B., Ikauniece, A., Andrushaitis, A., Ojaveer, H., Kallaste, K., & Jaanus, A., Tropic status of coastal and open areas of the south-eastem

Baltic Sea based on nutrient phytoplancton data fiom 1993-1997. Meereswiss. Ber., Warnemiinde, 38, pp. 1-86,2000. [6] Lampert, W. and Sommer U., Lirnnookologie, Thieme Verlag: Stuttgart and New York, Pp 1-440, 1993. [7] Stewart, A., J. & Wetzel, R.,G., Cryptophytes and other microflagellates as

couplers in planktonic community dynamics. Arch. Hydrobiol. 106 (l), 1- 19, 1986.

Transactions on Ecology and the Environment vol 65, © 2003 WIT Press, www.witpress.com, ISSN 1743-3541

[8] Utermohl, H., Zur Vervollkornmnung der quantitativen Phytoplankton- Methodik Mitt. Int. Ver. Limnol. 9, pp. 1-38, 1958. [9] Rott, E., Primary Productive and Activity Coefficients of the Phytoplankton of a Mesotrophic Soft-Water Lake (Piburger See, Tirol, Australia). lnt. Rev.

Gesamt. Hydrobiol. 66 (l),pp. 1-27, 1981. [ 101Heinonen, P., Quantity and composition of phytoplankton in Finnish inland waters, Publications of the Water Research Institute (Helsinki) 37, 1-91, 1980.

[ll] Wasmund, N., Nausch, G., & Matthaeus, W., Phytoplankton spring blooms in the southern Baltic Sea-spatio-temporal development and long-term trends. J. Plankton Res. 20 (6), pp. 1099-1117, 1998. [12]Hamrner, A., Schurnann, R., & Schubert, H., Light and temperature

acclimation of Rhodomonas salina (Cryptophyceae): Photosynthetic performance. Aquat. Microb. Ecol., 29, pp. 287-296,2002. [13]Willen, E., Phytoplankton in Water Quality assessment-an indikator concept, Hydrological and lirnnological Aspects of Lake Monitoring, Heinonen, P., Ziglio, G., van der Beken, A., John Wiley & Sons, 57-80.

[14]Tikkanen, T. & Willen, E., Vuxtplanktonfiva, Naturvardsverket: Stockholm, Sweden, 1992. [15]Brettum, P., Alger som Indikator pa Vannkvalitet I Norske Innsjoer: Planteplankton, Norsk Institutt for Vannforskning (NIVA)-Report 0-86116

pp. 2344,1989. [l6]Beukema, JJ. & Cadee, GC, Growth rates of the bivalve Macoma balthica in the Wadden Sea during a period of eutrophication: Relationships with concentrations of pelagic diatoms and flagellates. Mar. Ecol. Prog. Ser., 68 (3), pp. 249-256, 1990.

Transactions on Ecology and the Environment vol 65, © 2003 WIT Press, www.witpress.com, ISSN 1743-3541