Lea Tuvikene

THETHE INFLUENCE EFFECT OF OF NATURAL GROWTH VARIABILITY CONDITIONS ON ON PHYSICO-MECHANICALTHE ASSESSMENT OF ECOLOGICAL PROPERTIES STATUS OF SCOTS OF PINESHALLOW (Pinus sylvestris LAKES L.) WOOD IN ESTONIA

LOODUSLIKUKASVUTINGIMUSTE MUUTLIKKUSE MÕJU MÕJU HARILIKU MADALATE MÄNNI JÄRVEDE (PinusÖKOSEISUNDI sylvestris L.) PUIDU HINDAMISELE FÜÜSIKALIS- MEHAANILISTELE OMADUSTELE EESTIS

LEAREGINO TUVIKENE KASK

AA Thesisthesis for applying for thefor degreeapplying of forDoctor the degree of Philosophy of in Forestry Doctor of Philosophy in Applied Biology

Väitekiri Filosoofiadoktorifilosoofiadoktori kraadi kraadi taotlemiseks taotlemiseks rakendusbioloogia metsanduse erialal erialal

Tartu 20152018 Eesti Maaülikooli doktoritööd

Doctoral Theses of the Estonian University of Life Sciences

THE EFFECT OF NATURAL VARIABILITY ON THE ASSESSMENT OF ECOLOGICAL STATUS OF SHALLOW LAKES

LOODUSLIKU MUUTLIKKUSE MÕJU MADALATE JÄRVEDE ÖKOSEISUNDI HINDAMISELE

LEA TUVIKENE

A thesis for applying for the degree of Doctor of Philosophy in Applied Biology

Väitekiri Filosoofiadoktori kraadi taotlemiseks rakendusbioloogia erialal

Tartu 2018 Institute of Agricultural and Environmental Sciences Estonian University of Life Sciences

According to the verdict 6-14/13-9 of December 12, 2017, the Doctoral Committee for Environmental Sciences and Applied Biology of the Estonian University of Life Sciences has accepted the thesis for the defence of the degree Doctor of Philosophy in Applied Biology.

Opponent: Prof. Brigitte Nixdorf, Dr. rer. nat. habil. Brandenburg University of Technology, Germany

Pre-opponent: Fabio Ercoli, PhD Estonian University of Life Sciences, Estonia University of Jyväskylä, Finland

Supervisor: Peeter Nõges, PhD Estonian University of Life Sciences, Estonia

Defence of the thesis: Estonian University of Life Sciences, Karl Ernst von Baer House, Veski 4, Tartu, on 16th of February 2018, at 13.15.

The English and Estonian was edited by Tiina Nõges.

Publication of this thesis is supported by the Estonian University of Life Sciences.

LIST OF ORIGINAL PUBLICATIONS...... 7 ABBREVIATIONS...... 9 GRAPHICAL ABSTRACT...... 11 1. INTRODUCTION...... 12 2. AIMS AND HYPOTHESES OF THE STUDY...... 14 3. REVIEW OF THE LITERATURE...... 17 3.1. Principles of ecological status assessment in lakes...... 17 3.2. Confidence of status classes and risk of misclassification due to natural variability of indicators ...... 19 3.3. Types of natural variability ...... 20 3.3.1. Spatial variation ...... 20 3.3.2. Temporal variation ...... 22 3.3.2.1. Daily variation...... 22 3.3.2.2. Seasonal variation...... 23 3.3.2.3. Year-to-year variation...... 24 3.3.2.4. Multi-annual periodicity...... 24 3.3.2.5. Evolutionary and climate-related trends ...... 25 3.3.2.6. System shifts...... 28 3.3.3. Combining data with different variability ...... 30 3.4. Anthropogenic influence and its distinction from natural variability ...... 31 3.4.1. Challenges...... 32 3.4.1.1. Weak trends and the question of causality...... 32 3.4.1.2. Common chains of influence...... 32 3.4.1.3. Lag periods between impact and response ...... 34 3.4.1.4. Climate sensitivity of lake type parameters and status indicators...... 36 3.4.2. Ways forward...... 39 3.4.2.1. Monitoring of reference sites...... 39 3.4.2.2. Coherence studies ...... 39 3.4.2.3. Comparison of contrasting years...... 42 3.4.2.4. Detrending of long-term data for known sources of variability...... 43

5 3.4.2.5. Using multivariate statistics ...... 44 3.4.2.6. Model calculations...... 45 4. MATERIAL AND METHODS...... 47 5. RESULTS...... 49 5.1. Effect of water level and seasonality on water quality parameters in Võrtsjärv (I) ...... 49 5.2. Spacial and annual variability of environmental and phytoplankton indicators of Võrtsjärv (II) ...... 50 5.3. Contemporary climatic, hydrological and loading trends in Võrtsjärv and Peipsi (III)...... 52 5.4. Relationships between DOC and other water and environmental variables (IV)...... 53 5.5. Phytoplankton productivity in shallow lakes (V)...... 54 5.6. Effects of coloured dissolved organic matter on the attenuation of photosynthetically active radiation in Lake Peipsi (VI)...... 54 6. DISCUSSION AND SYNTHESIS OF ORIGINAL PUBLICATIONS...... 55 6.1. Distinction of effects of natural factors on common water quality indicators in shallow lakes...... 55 6.1.1. Influence of water level and seasonality I( )...... 55 6.1.2 Spatial and temporal variability (II)...... 58 6.2. Contemporary climatic, hydrological and loading trends at shallow lakes Peipsi and Võrtsjärv (III)...... 60 6.3. Underwater light field and its determinants in shallow lakes (IV, V, VI)...... 63 7. CONCLUSIONS ...... 67 REFERENCES ...... 69 SUMMARY IN ESTONIAN...... 102 ACKNOWLEDGEMENTS...... 107 ORIGINAL PUBLICATIONS...... 109 CURRICULUM VITAE...... 213 ELULOOKIRJELDUS...... 220

6 LIST OF ORIGINAL PUBLICATIONS

I. Tuvikene, L., Nõges, T. & Nõges, P. 2011. Why do phytoplankton species composition and “traditional” water quality parameters indicate different ecological status of a large shallow lake? Hydrobiologia, 660(1): 3–15, doi: 10.1007/s10750-010-0414-5.

II. Nõges, P. & Tuvikene, L. 2012. Spatial and annual variability of environmental and phytoplankton indicators in Lake Võrtsjärv: implications for water quality monitoring. Estonian Journal of Ecology, 61(4): 227–246, doi: 10.3176/eco.2012.4.

III. Nõges, T., Tuvikene, L. & Nõges, P. 2010. Contemporary trends of temperature, nutrient loading and water quality in large lakes Peipsi and Võrtsjärv, Estonia. Journal of Aquatic Ecosystem Health and Management, 13(2): 143–153, doi: 10.1080/14634981003788987

IV. Toming, K., Kutser, T., Tuvikene, L., Viik, M. & Nõges, T. 2016. Dissolved organic carbon and its potential predictors in eutrophic lakes. Water Research, 102: 32−40, doi: 10.1016/j. watres.2016.06.012.

V. Kauer, T., Arst, H., Nõges, T. & Tuvikene, L. 2009. Estimation of the phytoplankton productivity in three Estonian lakes. Estonian Journal of Ecology, 58(4): 297–312, doi: 10.3176/eco.2009.4.05.

VI. Reinart, A., Paavel, B. & Tuvikene, L. 2004. Effects of coloured dissolved organic matter on the attenuation of photosynthetically active radiation in Lake Peipsi. Proceedings of the Estonian Academy of Sciences. Biology, Ecology, 53(2): 88–105.

7 Author’s contribution to the papers:

Paper I II III IV V VI TK, PN, KT; LT, TN, HA, AR, Concept LT, TN, PN PN LT, BP, LT TN TKu TN TN, TK, Design PN PN KT AR PN LT AR, Data LT, LT, LT, LT, TK, LT, collection PN, * PN, * PN, * MV, * LT BP Data LT, PN, TN, KT TK AR analyses PN LT PN TK, PN, TN, KT, AR, Manuscript PN, HA, LT, PN, LT, LT, preparation LT LT, TN LT TN, + BP TN

LT – Lea Tuvikene; TN – Tiina Nõges; PN – Peeter Nõges; KT – Kaire Toming; TKu – Tiit Kutser; MV – Malle Viik; TK – Tuuli Kauer; HA – Helgi Arst; AR – Anu Reinart; BP – Birgot Paavel; + – other authors; * – co-workers.

8 ABBREVIATIONS

AMO Atlantic Multidecadal Oscillation AT air temperature ays absorption coefficient for yellow substance BM phytoplankton biomass

BOD7 biochemical oxygen demand BRT predictive modelling technique called Boosted Regression Trees CDOM coloured dissolved organic matter („Yellow substance“) Chl a chlorophyll a concentration Cl- chloride ion

CODMn chemical oxygen demand by permanganate

ColourPt-Co water colour by Platinum-Cobalt scale Cond specific conductivity

CO2 carbon dioxide C. Lim. the long-term permanent sampling site at Lake Võrtsjärv

Cvar variation coefficient DO dissolved oxygen DOC dissolved organic carbon DOM dissolved organic matter EA East Atlantic pattern (atmospheric circulation model for the Mediterranean) EIONET European Environment Information and Observation Network ENSO El Nino Southern Oscillation EQS Ecological Quality Standard EWI early warning indicator Hard total hardness - HCO3 carbonate alkalinity

H2S hydrogen sulphide K+ potassium ion

Kd diffuse attenuation coefficient λ wavelength

9 LTER Long-Term Ecosystem Research NAO North Atlantic Oscillation

NH4-N ammonium nitrogen

NO2-N nitrite nitrogen

NO3-N nitrate nitrogen Nsp number of species OAS optically active substance P primary production PAR photosynthetically active radiation pH acid-alkaline balance

PO4-P phosphate phosphorus PR precipitation q underwater planar quantum irradiance Q riverine discharge S Secchi depth Si silicon 2- SO4 sulphate ion SWT surface water temperature TFe total iron TN total nitrogen TP total phosphorus TSM total suspended matter TSS total suspended solids VRD Veepoliitika Raamdirektiiv WFD the European Water Framework Directive WFD CIS Common Implementation Strategy for the WFD WL water level WT water temperature z water depth

10 GRAPHICAL ABSTRACT

This work is a contribution to the improvement of large lake monitoring, demonstrating that besides following the class boundaries of ecological status established by law, reliable status estimates of such lakes require profound expert knowledge on ecosystem functioning, and considering the share of natural and human induced factors influencing the ecosystem.

11 1. INTRODUCTION

The European Water Framework Directive (WFD), adopted in 2000, aimed at setting water quality standards and minimising anthropogenic pressures on surface water quality in EU countries. As the first milestone, all aquatic ecosystems in EU member states were required to obtain at least good ecological quality by the year 2015. Still, e.g., according to the Estonian Ministry of the Environment, for 38% of surface water bodies in Estonia this goal was impossible to reach (Ida-Eesti vesikonna veemajanduskava, 2017), and for these water bodies a prolonged objective was set for the next river basin management period. Even more, also the next WFD goal – to reach good ecological quality of all surface waters in Europe by 2027 – has been rated to be over-ambitious (Hering et al., 2010). In order to follow the progress towards the goals set by the WFD and, if necessary, apply appropriate restoration measures, a tested and reliable monitoring methodology is of vital importance. The 16-year experience and the huge monitoring data bank generated by now for WFD targets, are valuable prerequisites for this.

Since coming of the WFD into force, numerous more or less complex assessment schemes of the status of water bodies have been developed (e.g., WFD CIS, 2003; Søndergaard et al., 2005; Ellis & Adriaenssens, 2006; Hering et al., 2006; Nõges & Nõges, 2006; Carstensen, 2007; Alahuhta et al., 2009; Kagalou & Leonardos, 2009; Clarke, 2013), and thousands of publications analysed their efficiency and reasonability, giving suggestions for further improvement of the assessment and monitoring methods for lakes and rivers (e.g., Moss et al., 2003; Sandin, 2005; Moss, 2008; Nõges et al., 2009c; Peeters et al., 2009; Hering et al., 2010; Rask et al., 2011; Caroni et al., 2013; Lyche-Solheim et al., 2013; Søndergaard et al., 2016). The primary goal of water quality monitoring is not to determine the status of a water body at a certain moment but to detect indications on its worsening (Anttila et al., 2008). The ultimate goal is to estimate an ecological quality of water bodies cost-effectively but sufficiently accurately in order to plan further activities. For planning of restoration measures, explicit knowledge at the ecosystem level is needed at highest certainty grade (Peeters et al., 2009; Søndergaard et al., 2016).

12 Any environmental metric included in a monitoring programme of water bodies is subject to four general types of variation: spatial, temporal, their interaction (meaning that a particular temporal effect operates differently in some locations than others), and variation caused by sampling and measurement error (Ellis & Adriaenssens, 2006). Any of those types of variation operate at different scales, e.g. temporal differences include diurnal and seasonal dynamics (both regular and stochastic), year-to-year differences and multiannual cycles; spatial differences involve horizontal and vertical differences, differences between different parts within a water body and differences between water bodies within a smaller or larger area. This classification of variability, however, does not say anything about the causes of spatial and temporal variability that is the main focus of any monitoring and assessment system. By introducing causality into the system, the part of the variability related to the cause of interest becomes the signal, while the rest forms the noise. Depending on the raised question(s), opposite factors may appear as noise, e.g., for WFD purposes, local human impact is the signal and the more general natural variability is the noise while in the context of climate change studies the positions of signal and noise are the other way round.

13 2. AIMS AND HYPOTHESES OF THE STUDY

Shallow lakes are largely controlled by climatic factors, either directly by increased water temperature or indirectly through the fluctuations of non-regulated water levels. Regular monitoring of the Estonian large lakes Võrtsjärv and Peipsi started for scientific purposes in the early 1960s and has continued until nowadays, with at least monthly frequency at Peipsi from May to October and at Võrtsjärv perennially that considerably exceeds the frequencies suggested by the WFD (EC, 2000) for surveillance monitoring. The research questions for both lakes have changed over time or evolved into new questions – a feature that is typical of adaptive monitoring (Lindenmayer & Likens, 2009). From mainly producing background information for the fisheries research in the 1960s and 1970s, over decades the focus has shifted more to support studies on trophic relations in the planktonic food-chain (e.g. Nõges et al., 1998; Blank et al., 2010; Agasild et al., 2014), trend detection in ecological status (Nõges et al., 1997; Nõges & Laugaste, 1998), as well as climate change impact assessment (Nõges et al., 2007c; Nõges et al., 2010a; Nõges & Nõges, 2014). Changing research needs through successive projects have added new variables, changed analytical approaches and created a high complexity of the monitoring programme, although, partly at the expense of increasing costs and compromised clarity of the overall objective. Still, the high complexity and the maintenance of the integrity of the long time-series (another principle of adaptive monitoring) has proven to be the main strength of the monitoring programmes that has enabled a process based approach in the ecological research, fulfils various requirements of the WFD, and has started to pay back also in the climate change impact studies. Due to their long-term data, Võrtsjärv and Peipsi have been included as model lakes in various EU financed projects. In such way, these lakes are good examples of long-term synergy between scientific research and environmental monitoring.

The papers included in this dissertation are based mainly on the long- term databases of lakes Võrtsjärv (I, II, III, IV, V) and Peipsi (III, V, VI), but also Lake Harku (V) and have a common aim – to disentangle the role of different environmental factors in determining ecological status of shallow lakes and affecting its assessment. Paper I focuses on the effect of natural variability factors such as water level fluctuation and

14 seasonal changes on traditional water quality parameters. Paper II adds a spatial aspect to large lake studies and paper III analyses the partly confounding effects of increasing temperature and nutrient loadings. The aim of paper IV was to clarify in which relative proportions different water and environmental variables determine the variability of dissolved organic carbon (DOC) as the main source of energy for microbial metabolism in Võrtsjärv. Paper V follows diurnal variation of the underwater quantum irradiance and the corresponding primary production estimated by model calculations, and paper VI concentrates on methodological aspects in clarifying the interdependencies between coloured dissolved organic matter (CDOM) and the underwater light field.

This study addressed the following research objectives and working hypotheses (H):

1) How to assess ecological status of shallow lakes in which (H1) the effect of natural factors such as water level fluctuation and seasonality assumingly exceed the effects of human induced pressures on such common water quality indicators as concentrations of nutrients and chlorophyll a, phytoplankton biomass, and Secchi depth (I)?

2) We hypothesised that (H2) in Võrtsjärv spatial differences in variables indicating ecological status of the pelagic area are smaller than their year-to-year variability, and (H3) the sampling station at the deepest site is sufficient to assess ecological quality of the main pelagic area while the narrow southern part should be monitored and its status evaluated separately (II). To prove this, we assessed the spatial inhomogeneity of Võrtsjärv by:

a. delineating the more homogeneous pelagic area and parts of the lake that significantly differ from it by hydrochemical, hydrooptical, and phytoplankton parameters, b. analysing the partitioning of variance between spatial (sampling stations) and temporal (years) components within the lake proper for different variables, c. assessing the effect of temperature and water-level changes on the variability of the monitored parameters,

15 d. evaluating the representativeness of the main monitoring station for characterizing the pelagic area of Võrtsjärv and the necessity of continuing whole-lake surveys. 3) To find out, if H4( ) the changes in shallow lake ecosystems reveal signals from both catchment management and climate change, we assessed the contemporary climatic, hydrological and loading trends in lakes Võrtsjärv, Peipsi, and their catchments, and evaluated their impact on water quality in these large and shallow lakes (III).

4) We evaluated the role of coloured dissolved organic matter (CDOM) for the underwater light field that determines primary production in large shallow lakes and assumed that (H5) CDOM is also a good proxy to predict the total dissolved organic matter (DOC) concentration (IV, V, VI).

16 3. REVIEW OF THE LITERATURE

3.1. Principles of ecological status assessment in lakes

The principal legislative tool in the field of water policy in Europe, the Water Framework Directive (WFD; Directive, 2000) defines the status of water bodies by the extent of anthropogenically derived deviation from the so-called “reference conditions”, i.e. conditions that should occur at sites of any particular type of water bodies in the absence of human impact. Five quality classes are defined for aquatic ecosystems: high, good, moderate, poor and bad. High quality means minimal human influence and good quality differs only slightly from that. This as well as the other narrative normative definitions of water quality classes in the WFD, and even the whole concept have been criticized (Borja & Rodriguez, 2010; Nõges et al., 2009c; Josefsson & Baaner, 2011), and probably more specified definitions will be presented in the future versions of guiding documents. For management strategies of the water bodies, the border line between good and bad ecological status is of decisive importance since the ultimate goal posed in the WFD is to achieve at least good chemical and ecological status of surface water bodies. For the lakes for which this goal was not achieved during the first period of River Basin Management Plans (2009-2015), extended objectives were set up for the second period (2015-2021).

According to WFD, particular groups of hydromorphological and physicochemical variables and groups of biota, the so-called “quality elements”, form a framework which metrics are required to estimate the ecological status of water bodies (Fig. 1). Employment of biological quality elements (BQEs) in water quality assessment and rating these over the chemical quality elements has been welcomed as the innovativeness of the WFD (Moss, 2007; Hering et al., 2010). Ideal biological metrics should be sensitive to anthropogenic pressures showing a clear and specific pressure-response relationship while being insensitive to a broad range of natural variability. The main challenge for building assessment systems is to find metrics that most resemble this ideal.

For monitoring purposes, the location of sampling sites and sampling frequency should be such to provide the best signal to noise ratio for the parameters of interest. Cavanagh et al. (1998) formulated the principles

17 for site selection in environmental monitoring. For trend monitoring, the location of sites should be such as to provide an indication of a change in the water quality within the basin as a whole. A good site for this purpose is the deepest point of the lake where different inputs to the lake are integrated. The same is valid for survey monitoring where the purpose is to establish the general overall condition of the water body. For impact assessment monitoring, on the contrary, sites should be established near the shore adjacent to the development whose impact is studied or at the mouths of tributaries and at the outlet if the impact of sub-catchments is addressed. In this case, it is also recommended that one site should be located at the deepest point of the lake to monitor overall conditions. The number of monitoring sites needed per lake depends on various factors including the purpose of sampling, lake size, stratification, morphometry, flow regime, and tributaries (EPA, 1997), which together determine the strongly lake specific area for which a station can be considered representative. Different factors of anthropogenic stress such as eutrophication or organic pollution are often correlated and therefore can be detected using the same indicators

Do the estimated values Do the physico- Do the hydro- Yes Yes Yes Classify as for the biological quality chemical conditions morphological elements meet reference meet high status? conditions meet high high status conditions? status?

No No No

Do the estimated values Do the physico-chemical for the biological quality Yes conditions (a) ensure Yes elements deviate only ecosystem functioning Classify as slightly from reference and (b) meet the EQSs good status condition values? for specific pollutants?

No No

Classify on the basis of Yes the biological deviation Is the deviation Classify as from reference moderate? moderate status conditions Greater Yes Is the deviation Classify as major? poor status Greater

Classify as bad status

Figure 1. Indication of the relative roles of biological, hydromorphological and physico-chemical quality elements in ecological status classification according to the normative definitions of the WFD (reproduced from WFD CIS (2003)). EQS– Ecological quality standard set at EU level (WFD CIS (2011)).

18 (Hering et al., 2006). The key problem, especially regarding shallow lakes, is to distinguish human-induced alterations from the natural ones which both affect the majority of indicators commonly used for status assessment of water bodies.

3.2. Confidence of status classes and risk of misclassification due to natural variability of indicators

Since every sampling, measurement and determination is influenced by several sources of errors, estimation of uncertainty is an essential component in quality assessment systems (Carstensen, 2007; Hering et al., 2010; Anttila, 2013). Therefore, terms of probabilities should always be given together with ecological status classification and estimates, nevertheless, as pointed out by Hering et al. (2010), only a small proportion of assessment systems have put this into practice. Taking variability in ecological classification into consideration helps lake managers to set up more realistic expectations and to design more effective monitoring programmes to detect changes (Anttila, 2013; Carpenter & Lathrop, 2014). In the WFD normative terms, uncertainty is referred to as “level of confidence” or “precision”.

Depending on the purpose of monitoring, the intrinsic variability of metrics can be successfully exploited in some cases as a signal to discover certain impacts on the system or relationships within the system, but in other cases, it acts as a noise complicating the detection of the signal of interest. A typical case to exemplify this dilemma is the monitoring of water bodies with two different aims: detecting human impacts with the effect of natural variability minimized or climate change impacts for which anthropogenic changes represent confounding factors. In both cases, the success of monitoring rests on the researcher’s ability to select good specific indicators that are sensitive to the underlying conditions of interest but insensitive to other factors (Patil, 1991).

It is also important to know how the established class limits must be interpreted. Estimating spatial, temporal and spatial-temporal variability of metrics is crucial for the confidence of assigned status class (Ellis & Adriaenssens, 2006; Carstensen, 2007; Hering et al., 2010). Sampling error and analytical error have to be considered as well. Large random variation can be compensated by increasing the number of measurements (Carstensen, 2007). As another extreme, on the basis of

19 fish monitoring data analysis, Ellis & Adriaenssens (2006) demonstrated that for producing a representative picture of high confidence of status class it was enough to visit each site of a river just once a year (avoiding in this way seasonal variation), whereby expanding the time window introduced additional components of variation and increased risk of misclassification until enough samples were taken. They also showed that the risk of misclassification was substantially higher, even up to 70%, when the observed value was very close to a quality class border. This is critically important to consider for the executive authorities in the phase of planning of management.

The natural variability that in principle should belong to reference conditions exceeds often the variability caused by anthropogenic factors. As the target variables of both types of variability largely overlap in natural waters, it becomes difficult to disentangle their effects that add a large uncertainty to the status assessment. E.g., seasonal and inter-annual water level (WL) fluctuations exceeding 3 m and modifying the intensity of sediment resuspension, strongly influence all water quality parameters in Estonian large and shallow Lake Võrtsjärv (Nõges & Järvet, 1995; Nõges & Nõges, 1998, 1999). Though large inter-annual variability in water quality indicators is commonly associated with eutrophic conditions, natural variability should be considered irrespectively of eutrophication level (Søndergaard et al., 2016).

3.3. Types of natural variability

The sensitivity of lakes to natural variability factors depends strongly on their morphometry. The role of physical drivers such as wind and water level in controlling ecosystem processes increases with increasing lake area and decreasing depth (Nõges, 2009a). While large and shallow polymictic lakes are extremely sensitive to natural physical drivers (Scheffer, 2004; Nõges et al., 2007a, b; Scheffer & van Nes, 2007), in deep lakes the in-lake biological and chemical factors prevail (Tilzer & Serruya, 1990). The following natural factors may have a remarkable influence on parameters commonly used to assess ecological status.

3.3.1. Spatial variation

Spatial variation in indicators of water quality involves horizontal and vertical differences, differences between parts of a water body, 20 or differences between water bodies within some certain area. Spatial variability of substances in water bodies is caused by an interaction of physical, chemical and biological processes (e.g. George & Edwards, 1976; Fergus et al., 2016), and produces inaccuracy in monitoring results of water bodies (Anttila & Kairesalo, 2010). According to WFD, a water body can be assigned to only one quality class, thus, in practice, bigger water bodies are often divided into more than one water bodies because of spatial differences on their water quality. Spatial variation plays a big role at different spatial scales.

Even at the scale of one lake, spatial horizontal and vertical variability exists for all variables. A specific phenomenon, called plankton patchiness (Levin & Segel, 1976; Rossa et al., 2013) can be seen, for example, in chlorophyll a distribution (Vilar et al., 2003; Wang & Liu, 2004; Anttila et al., 2008). Patchiness is a major challenge for traditional water quality monitoring as it is not detectable by sampling in only one or a few sites in a lake (Anttila et al., 2008). To minimise spatial variability, the monitoring stations need to represent satisfactorily different regions of a lake (Carstensen, 2007), and it may be an issue even in case of small lakes as it was illustrated by Søndergaard et al. (2016). A conventional sampling strategy involves only one mid-lake sampling site or the deepest point of a lake (Göransson et al., 2004). Still, in-lake horizontal and vertical spatial variability among biological, physical and chemical indicators can be considerable (Wetzel, 2001), and depend on several factors such as lake morphometry, surface and groundwater inputs, and biological activities (Göransson et al., 2004; Morozov & Kuzenkov, 2016). For physical and chemical variables Göransson et al. (2004) concluded a mid-lake sample to be representative. To reduce horizontal and vertical variation, volume- weighed (composite) sampling has been suggested (e.g. Blomqvist, 2001), for which several samples from different locations or depths in a lake are thoroughly mixed.

To assess the representativeness of conventional discrete sampling used in water quality monitoring, Anttila et al. (2008) proposed a geostatistical method where the semivariance of each observation pair is plotted against the distance which separates the observations in each pair. They showed that point source samples for chlorophyll a were representative only over small areas and concluded that ignoring the spatial variation of monitoring results may lead to wrong overall decisions. Still, the authors admitted that considering the patchiness in water quality parameters

21 is laborious and costly. In addition to changeful patchiness, stationary patterns in water quality exist deriving from systematic point sources of nutrients and suspended solids from inflows or from diffuse sources. Such kind of spatial distribution is characteristic mainly to river mouth and shoreline areas and can be easily detected by remote sensing (Anttila & Kairesalo, 2010).

Studies on macroscale spatial trends are especially important to understand global change at broad scales (Fergus et al., 2016). It should be kept in mind that when studying spatial variation of indicators in many lakes within a large geographical range, there may appear spatial autocorrelation, i.e., lakes closer to each-other have more similar patterns of variables than more distant lakes (ibid). For over 800 north temperate lakes in the USA, the authors identified lake-to-lake differences in effects of lake water chemistry on chlorophyll. Within 1,000 km2 to 10,000 km2 ecological drainage units, lakes of a region had more similar TP effects on chlorophyll a compared to lakes from other regions (Wagner et al., 2011). In a smaller regional scale, the amount of coloured dissolved organic matter (CDOM) was highest in the coastal/onshore areas of Estonian standing water bodies – Pärnu Bay and lakes Peipsi and Võrtsjärv, whereas the proportion of the allochthonous component in CDOM decreased towards offshore direction (Toming et al., 2009).

3.3.2. Temporal variation

The scales of temporal variations are important to consider when analysing processes in aquatic ecosystems with the aim to evaluate their ecological status. Though measuring process rates such as production, would better reflect changes in ecosystem functions, it is more common to measure the standing stocks in monitoring programmes as an easier option (Rocha et al., 2011; Cavalcanti et al., 2016). Automated water quality measurements are a good and low-cost technical approach to get enough good data to reveal temporal variation in most common indicators of water quality (Anttila et al., 2012).

3.3.2.1. Daily variation

Diurnal changes occur in physical, chemical and biological variables, some of which are regular due to a daily cycle of photosynthesis and respiration, which, in turn, modify pH, gas regime, and alkalinity. In 22 shallow wind-exposed lakes, temporal variability of characteristics such as resuspension or spatial distribution of phytoplankton can be observed even at hourly scale (Søndergaard et al., 1992, 2003; Søndergaard et al., 2016). For example, the profiles and diurnal sums of phytoplankton primary production has a pronounced diel pattern in aquatic environments depending first of all on incoming solar irradiance, light attenuation and chlorophyll a content of water (Paavel et al., 2007; Nõges et al., 2011b; Staehr et al., 2016), but also on lake morphometry, catchment geology, and availability of nutrients (Staehr et al., 2016).

To avoid the effect of diurnal differences in the data, monitoring programmes usually determine a certain sampling time for a water body, still there remains a stochastic component deriving from meteorology (Loftis et al., 1991) but this can be diminished using progressive methods. For example, the effect of cloudiness on primary production can be considered by using autonomous high-frequency in-lake measurements and measurements of meteorological properties (Staehr et al., 2016). Also, remote sensing is a good method for describing the horizontal variability of several indicators in more detail (Gitelson et al., 2008; Bolpagni et al., 2014; Wang et al., 2015; Søndergaard et al., 2016).

3.3.2.2. Seasonal variation

Seasonal changes in lakes take place with well-known regularity from year to year with randomness added to them due to yearly differences in weather conditions causing deviations in seasonality and phenology. To remove seasonality from data, several methods of time series analysis such as seasonal decomposition (Cleveland & Tiao, 1976), seasonal smoothing (Gardner, 1985) and removing seasonal components – deseasonalizing (Pouzols & Lendasse, 2010) have been used. In practice, however, large amounts of monitoring data are omitted due to seasonality problems and the status assessment of water bodies is often based on data of a single season only, mostly summer (Ellis & Adriaenssens, 2006; Søndergaard et al., 2016). Using seasonal averages requires regular and comparable data coverage for all years. If sampling takes place in a restricted seasonal window of the year, e.g. spring-summer, seasonal effect is less pronounced and the total variability becomes considerably less than in case of 12-months sampling, in which other site-specific or waterbody-specific variation own greater roles (Ellis & Adriaenssens, 2006). However, Carstensten (2007) showed that using annual means

23 with seasonal adjustment instead of seasonal means enables to reduce bias and to improve the precision of status estimates.

3.3.2.3. Year-to-year variation

Large inter-annual variability in lake status variables are generally associated with eutrophic conditions (Soininen et al., 2005), but can be important in shallow lakes at intermediate nutrient concentrations as well (Søndergaard et al., 2016). Besides, high inter-annual variability may occur independently from a trophic situation, e.g. in phytoplankton indices due to an influence of hydromorphological or climatic conditions. In Danish shallow lakes, considerable year-to-year variations occurred in nutrient concentrations even in nutrient-poor lakes which shows that high natural variability can appear irrespectively of the state of eutrophication (ibid). Still, they identified extra high year-to-year variability in eutrophic lakes and in shallow lakes with nutrient concentrations close to good– moderate status boundary and therefore often shifting between a clear and a turbid state.

3.3.2.4. Multi-annual periodicity

Prolonged changes in atmospheric circulation patterns such as the NAO (North Atlantic Oscillation) (Hurrell, 1995; Hurrell et al., 2003) or ENSO (El Nino Southern Oscillation) (Philander, 1990) affect, through modifying global scale heat and moisture transport, physical, chemical and biological properties of water bodies. Both NAO and ENSO have shown to cause large fluctuations in affluence and lake water levels in Eurasia (Rodó et al., 1997), but not so clearly in the south of the Alps (Salmaso & Cerasino, 2012) or North America (Baker et al., 2010). Still, similarly to NAO and ENSO, winter fluctuations of the East Atlantic pattern (EA), the specific atmospheric circulation model for the Mediterranean, were shown to influence Lake Garda mixing and nutrient dynamics. During its negative phase, the lake was mixed well and the long-term increase of phosphorus concentrations concurrently with fluctuating year-to- year surface phosphorus supply favoured cyanobacteria. In EA positive phase, indications of eutrophication weakened (Salmaso & Cerasino, 2012). Clearly, the decadal scale periodicities caused by atmospheric circulation patterns in lake parameters can be revealed only in really long-term monitoring data.

24 Multi-annual periodicities in lake processes can be detected analysing sediment cores. For example, by spectral analysis of two independent rock-magnetic datasets of lake sediments from the Alpine region for last 2500 years, Lanci & Hirt (2015) found a coherent statistically significant periodicity of 50-60 years. Centennial- to decadal scale shifts in the Lake Pannon biotic environment due to past (Upper Miocene) lake level rise events were revealed by a detailed ultra-high-resolution analysis of a sediment core whereat the general trends were shown to be interrupted by smaller scale cycles (Harzhauser et al., 2008). A millennial-length proxy record of estuarine fossil pigments in the North Atlantic climate region (Hubeny et al., 2006) exhibited significant cyclic components that could be related to two of the dominant internal modes of climate variability in the region: the NAO and the Atlantic Multidecadal Oscillation (AMO). In long-term data, linear trends come to light in spite the comprising oscillatory dynamics (Sharma & Magnuson, 2014), but looking at a short data series of a long-period fluctuation, a phase of its increase or decrease may seem as a trend, and bring to misleading interpretation of results if taken for a long-term trend (Tomé & Miranda, 2004).

3.3.2.5. Evolutionary and climate-related trends

Trends in lake parameters can be caused also by slow geological and climatic changes, as well as due to the evolution of lakes. According to a common ecological concept, lakes pass through a natural aging process in the course of which they get more productive (Burkholder, 2000) and fill up with sediments. Knowing the mechanisms that have driven the evolution of lakes in the far past is useful to understand contemporary processes in lakes. As a good example, Engstrom et al. (2000) reconstructed the limnological trends in Alaskan lakes of different age using chronosequence and sediment diatom studies. They concluded that local hydrological factors control initial starting point, rate and extent of evolutionary changes, and that loading of minerals and dissolved organic carbon into lakes is related to weathering of soluble minerals from upper soil horizons, a succession of vegetation and development of soil after ice recession. Profiles of chemical substances in lake sediments are a valuable source to follow evolutionary trends in lakes. Using P-fractions stratigraphy, four main historical periods were recently distinguished for Lake Peipsi, showing evident anthropogenic impacts over the last 2.3 thousand years and rapid eutrophication during the last 50-60 years (Kisand et al., 2017). Analysis of a sediment core

25 from Lake Võrtsjärv showed increasing inorganic carbon content and decreasing organic carbon since the mid-1950s (Ehapalu et al., 2017).

It is also important to understand how long-term exposure to chemical pollution, acidification and other evolutionary trends in the environment affect biota (Rogalski 2016; Derry et al., 2010). Ecological tolerance of species to substances accumulating in the environment may change through time. For example, due to secular acid deposition, acid tolerance of zooplankton species was shown to be shifted since the late 1800s (Derry et al., 2010). Shifts in ecological tolerances of species can potentially have effects also at population and ecosystem levels (Fussmann et al., 2007).

Widely recognised climate change besides its creeping nature entails more and more dramatic events having part with all sectors of modern societies including water use (Sigmund, 2006; BACC 2008; IPCC, 2013). Climatic variability affects aquatic biota, e.g., through episodic excursions in surface water chemistry during droughts or extreme wet years (Strok et al., 2016), changes in thermal stratification in deep lakes leading to shifts in plankton community structure and depth distributions (Hampton et al., 2014). In Lake Võrtsjärv, an influence of climate change manifested in milder winters, earlier and warmer springs, increasing precipitation and riverine discharge, lead to increased primary production, mineralisation rates of organic carbon, greenhouse gas emissions, and accumulation of inorganic carbon in sediments (Ehapalu et al., 2017).

One of the most pronounced effects of climate change on lakes is a substantial increase in DOC concentration, referred to as brownification or browning (Graneli, 2012; Kritzberg et al., 2014; Rasconi et al., 2015; Strock et al., 2016) which is a result of increased precipitation causing more intensive flow of dissolved substances from catchment to lakes. Trend of brownification has been observed across various lake types in many regions of Northern Hemisphere (Weyhenmeyer et al., 2004a; Evans et al., 2005; Roulet & Moore, 2006; Monteith et al., 2007; Nõges et al., 2011a; Jennings et al. 2012; Weyhenmeyer et al., 2016), and in addition to natural climatic fluctuations, it can be partly attributed to human influence. Brownification reduces drastically the penetration of solar radiation into the water column, leading to a decrease in primary production and secondary production, e.g., of benthic invertebrates and fish (Karlsson et al., 2009; Hedström et al., 2017). Brownification

26 also intensifies oxygen consumption and CO2 production by enhanced bacterial growth due to better availability of organic carbon leading to higher CO2 emissions to the atmosphere from freshwater ecosystems (Graneli, 2012). From another side, low concentrations of DOC stimulate primary production by releasing nutrients through photolysis (Seekell et al., 2015). Lapierre et al. (2015) suggested that key climatic signals are transmitted to lakes through the transport of terrestrial DOC, which can be considered not only as a driver of C cycling in lakes but even as a proxy for other matters of terrestrial origin including plant nutrients. Carbon cycle in biosphere gains more and more attention in the context of global warming, and it has been proved recently in many publications that the role of lakes in this is greater than it has been used to think (e.g., Tranvik, 1988; Bastviken et al., 2004; Baker et al., 2010; Buffam et al., 2011; Cole et al., 2011; Nõges et al., 2011b; Raymond et al., 2013; Seekell et al., 2014; Ehapalu et al., 2017).

Water temperature and ice cover are most directly affected by climate forcing. The existence of consistent trends has been demonstrated for water temperatures in rivers (Hari et al., 2006) and lakes (Livingstone, 2003; Straile et al., 2003; Arhonditsis et al., 2004; Coats et al., 2006; Dokulil et al., 2006; Hampton et al., 2014) and for ice phenology (Weyhenmeyer et al., 2005). For example, in the large and shallow lakes Peipsi and Võrtsjärv, daily surface water temperature was strongly dependent on daily air temperature during ice-free period – the average annual air temperature increased by 0.4 °C per decade during 1961-2004, resulting in surface water temperature rise in Peipsi and Võrtsjärv by 0.37-0.75 and 0.32-0.42 °C per decade in April and August, respectively (Nõges, 2009b). The projected climate change will, besides warming, increase nutrient and water input to lakes favouring lake productivity and increasing mineralisation of organic carbon. This will enhance inorganic carbon accumulation in sediments and contribute to greenhouse gas emission to the atmosphere (Ehapalu et al., 2017). These results are supported by large-scale analysis of data from 239 temperate and boreal lakes covering large environmental and geographic gradients across seven regions (Lapierre et al., 2015).

As shown recently by Woolway et al. (2017), decreasing wind speeds in some regions over recent decades, termed atmospheric stilling, may have even larger effect on lake ecosystems than increasing air temperatures. They showed that in large shallow lakes like Võrtsjärv which usually

27 has been considered well mixed (Frisk et al., 1999), decreasing surface wind speed has increased the number of days with occurring thermal stratification during spring and summer – in case of Võrtsjärv at a rate of two days per decade.

3.3.2.6. System shifts

Ecological stability of lakes has some elasticity or resilience which sustains pressures only up to a certain extent. Extrinsic and intrinsic drivers such as extreme weather events or trophic interactions may cause sudden (lasting less than a year) but durable changes in lake ecosystems (Scheffer & Carpenter, 2003; Scheffer & Jeppesen, 2007; Rogalski, 2016; Spears et al., 2016). A large shift to another stable state can be triggered by even a small incremental change when a critical threshold in the complex ecological situation has been achieved (Scheffer & Carpenter, 2003; Gsell et al., 2016).

For shallow lakes, the clear-water, macrophyte-dominated state is considered to show the pristine condition (Scheffer, 2004). The resilience of such lakes is usually undermined by eutrophication (Scheffer & Carpenter, 2003). Macrophytes are good indicators of the ecological state of shallow lakes (Jeppesen et al., 1998). In lakes where nutrient concentrations are low, usually relatively small plants dominate. With growing nutrient loading, abundance and coverage of aquatic macrophytes first increase. Typically, when nutrient loading has caused the water to turn turbid, submerged plants largely disappear because of shade created by abundant phytoplankton. By the state-of-art theory, there are two alternative stable states for shallow lakes (Scheffer, 2004; Scheffer & Carpenter, 2003) – the vegetation-dominated clear state and the phytoplankton-dominated turbid state. The system shift can be abrupt and the ecosystem response to eutrophication follows a sigmoidal hysteresis curve (Scheffer, 2004). In the initial phase of growing nutrient loadings, phytoplankton biomass and turbidity remain low in macrophyte-dominated lakes since macrophytes prevent wave- induced sediment resuspension, affect nutrient availability in the water column, harbour high densities of phytoplankton-eating cladocerans and influence algal growth through allelopathic exudates (Zingel at al., 2006). When the threshold endurance of these stabilizing mechanisms to the pressures gets surpassed, the ecosystem switches to a new stable state with turbid water and phytoplankton dominance. The new state is

28 stabilised by phytoplankton inhibiting macrophyte growth by shading and light limitation. In the absence of macrophytes, wave-induced sediment resuspension further increases turbidity and the macrophytes stay suppressed. A switch back to the clear water state can take place only at significantly lower nutrient concentrations than those occurring when switching to the turbid state happened (Scheffer & Carpenter, 2003; Scheffer, 2004). Recovery of the ecological status following nutrient load abatement may take long time from years up to centuries (Spears et al., 2016). For example, according to re-oligotrophication studies of 35 lakes, Jeppesen et al. (2005) found that in most lakes a new equilibrium between P in the sediment and water column was reached after 10–15 years. The complexity of the clear-water phase formation was well illustrated by a study of some Estonian shallow lakes which demonstrated that herbivorous ciliate grazing on phytoplankton in summer was probably the main factor sustaining macrophyte dominance in some of these lakes (Zingel et al., 2006). This is another of the numerous proofs (e.g., Nixdorf et al., 2005; Jeppesen et al., 2011; Jensen et al., 2013; Søndergaard et al., 2016) showing that meta- and protozooplankton are important players in shaping the status of lakes and therefore should be included as quality indicators into the WFD- related monitoring programs. Even more, zooplankton as an indicator of a trophic structure may appear to be a key factor in the practical management of lakes through biomanipulation (Anttila et al., 2013).

Increase in variability of an indicator is generally considered as one of the early warning signals of a regime shift (Scheffer et al., 2009), and this is not necessarily related to nutrient availability, especially in shallow lakes with intermediate nutrient concentrations (Blindow et al., 1993; Søndergaard et al., 2016). Shifts between clear and turbid water states cause changes in biological indicators used to characterise ecological quality of lakes (Søndergaard et al., 2002; Gsell et al., 2016), and in such intermediate systems, most probably larger inter-annual variations occur than in more stable lakes (Søndergaard et al., 2016).

In lakes, transitions to an alternative stable state can be observed repeatedly, and even when the thresholds may be discovered empirically, detecting early warning signals of transition in advance with aid of reliable statistical procedures makes possible to prevent or to prepare for it (Scheffer et al., 2009). The newest tools for early detection of ecological instability include (1) detecting of critical thresholds from multi-lake

29 data, (2) using palaeolimnological and long-term contemporary data to analyse the forms, causes and change rates of past events, (3) predicting transitions using high-frequency monitoring data (Spears et al., 2016). In a case study of five lake ecosystems for identifying the critical transition- generating mechanisms, Gsell et al. (2016) used four statistics of early warning indicators (EWIs) explicated by Scheffer et al. (2009): increase in autocorrelation, increasing trend of standard deviation (buildup of variance), skewed distribution of the state variable, and increasing density ratio. Based on high-quality empirical time series combined with ecological understanding and standardized methods, they found that EWIs can precede critical transitions even several years before a shift. Still, the authors concluded that relatively low detectability of EWIs before the documented transitions restrict their potential application to only well-understood ecosystems, and this opinion is supported by Spears et al. (2016). The latter authors state that ecological instability in lakes would be much better predictable if more practical proofs on ecosystem changes leading to ecological transition would be publicly available. This includes long-term data, development of analysis of ecological community behaviour before and after transitions, and irrefutable demonstration with aid of field experiments.

Adequate typology considering the system shifts to alternative stable states is needed in WFD context which is one of the challenges related to WFD implementation as lakes sustaining the similar nutrient loading may be with clear or turbid water, dominated by phytoplankton or macrophytes, and the concentrations of chlorophyll a may have really large variation. Should such lakes be treated as belonging to different types, and different criteria used to assess their ecological status, or would it be reasonable to work out schemes for evaluating their ecological status based on the abundance and composition of the dominating group of primary producers at the moment? Probably the latter solution may be better since reference conditions for both types of lakes are already determined for WFD purposes.

3.3.3. Combining data with different variability

Modelling enables to combine data with different variability. For example, several bio-optical models have been elaborated to estimate primary production from light intensity and abundance of phytoplankton pigments (e.g Bannister, 1974; Smith et al., 1989), or using daily averages

30 of photosynthetically active radiation (PAR) and the assimilation number (Sathyendranath et al., 1989). Arst et al. (2008) elaborated two versions, a spectral and an integral, for modelling integrated primary production over the water column in turbid waters. Since among the initial parameters of their model, only variation of incident irradiance was large and often irregular during a day while other parameters (Chl a, diffuse attenuation coefficient) changed slowly over time, daily data on incoming PAR irradiance were combined with episodic measurements of chlorophyll a concentration and the diffuse attenuation coefficient in the water. Applying the integral version of the model on Võrtsjärv, Nõges et al. (2011b) showed that for short to medium term studies, the modelled primary production (P) was more reliable than the sparse measurements while for longer time span, bio-optical modelling can be too much influenced by adaptive responses of phytoplankton changing their cellular chlorophyll content according to light limitation resulting from changes in water colour or water level.

3.4. Anthropogenic influence and its distinction from natural variability

WFD defines the status of water bodies by the extent of anthropogenic deviation from the reference conditions, i.e. conditions that should occur at sites of any particular type in the absence of human impact. Natural variability described in the previous chapter that in principle should belong to reference conditions often exceeds the anthropogenic variability. As the target variables of both types of variability largely overlap in natural waters, it becomes difficult to identify their specific effects that add a large uncertainty to the status assessment. Human activity exerts pressure on water bodies in several ways and extent, the best-known effects being accelerated eutrophication, organic and toxic pollution, hydromorphological degradation, acidification, and an introduction of alien invasive species. Most of the methods used for assessing the ecological status of lakes concentrate on eutrophication- related pressures (Poikāne et al., 2010; Brucet et al., 2013).

Different methods have been used to disentangle the effects of natural and anthropogenic variability in long-term data ranging from simple detrending (George et al., 2004a), applying coefficients for residual adjustment (Reist, 1986), statistical partialling of the effects of other factors and variance decomposition (Rodríguez & Magnan, 1995), to

31 linear (Carstensen & Henriksen, 2009) and nonlinear regression models (Massol et al., 2007).

3.4.1. Challenges

3.4.1.1. Weak trends and the question of causality

Scientists meet often difficulties when searching for causal relationships between the factors driving complex natural processes. Anthropogenic trends in lake quality indicators may be difficult to identify because of high natural variability, interacting effects of different driving forces and indirect relationships causing common trends in a number of variables that may be misinterpreted as causal relationships. The effects of eutrophication are often difficult to differentiate from climate change influence both inducing similar responses in lake ecosystems. For example, regardless of significant trends in seasonal air and water temperatures found in lakes Võrtsjärv and Peipsi, trends in ice phenology were absent or weak, which indicates that ice phenology is governed by more complex processes than only lake surface temperature (Nõges & Nõges, 2014). In this case, despite wide-spread notion of linear relationship between air temperature and ice freeze-up and breakup dates (e.g. Palecki & Barry, 1986; Assel & Robertson, 1995; Williams et al., 2004), nonlinear relationship deriving from sinusoidal pattern of seasonal air temperature (e.g., Weyhenmeyer et al., 2004b; Livingstone et al., 2010b) was confirmed. Changes in snowfall, cloudiness, water turbidity, and wind may further blur the relationship.

The causes of certain effects can be deduced by using statistical methods such as multivariate analysis (Rodríguez & Magnan, 1995).

3.4.1.2. Common chains of influence

Natural eutrophication of lakes is remarkably accelerated due to human activities and is the most widespread pressure affecting the whole biota through food web interactions. The main driving force for eutrophication is excess nutrient loading from the catchment, but also internal loading from sediments. Nutrients affect the food webs, primarily causing accelerated algal growth and blooms (Elliot et al., 2006; Blottière et al., 2017) which lead to destabilised oxygen conditions and

32 acid-alkaline balance (pH), such way worsening living conditions of fish and bottom animals (Jansson et al., 2015). Generally, with eutrophication piscivorous fish become scarcer, abundant omnivorous fish eat most of the zooplankton grazers, and algal growth increases (Jeppesen et al., 2010; Moss et al., 2011). The impetuous growth of algae leads to worsening of light conditions in the water column and shade-tolerant or scum forming cyanobacteria become a prevailing plankton group. During first phases of eutrophication, macrophytes play an important role in regulating light conditions in lakes by shading and reducing internal phosphorus loading through stabilising sediments in a littoral zone; their abundant presence suppresses algal growth increasing in this way water transparency (Wetzel, 2001; Cooke et al., 2005). Macrophyte zone is also a refuge for zooplankters and small fish (Timms & Moss, 1984; Meerhoff et al., 2007).

All symptoms of eutrophication are amplified by climate warming that affects nutrient dynamics and favours omnivorous fishes and cyanobacteria. As the effects of both pressures, eutrophication and climate change, to lakes can be similar, their relative roles are often difficult to distinguish (Elliot et al., 2006; Wilhelm & Adrian, 2008; Moss et al., 2011). Increasing temperature raises mineralisation in catchments and the nutrient loading into lakes (Rustad et al., 2001) as well as import of organic matter, including carbon (Cooke et al., 2005; Larsen et al., 2011). Increasing thermal stratification and growing respiration deepen the oxygen deficiency at the bottom thus increasing the internal load of nutrients (Jensen & Andersen, 1995). Due to warming, sediments are less effective to store carbon and release more of the stored methane (Walter et al., 2006). Also, due to different optimum temperatures for phyto- and zooplankton, changes in temperature regime may alter the synchronization of grazers with edible phytoplankton hence decreasing the effect of grazing in a food chain (George et al., 1990; Winder & Schindler, 2004; Morales-Baquero et al., 2006).

Differences in chains of influence exist depending on the latitude of lakes. For example, because of the prevalence of omnivorous fish, mean size of Cladocera is decreasing between 60º and 20º N (Gillooly & Dodson, 2000), and especially Daphnia is rare in low-latitude lakes. Therefore, the efficiency of zooplankton to control algal growth is decreasing from north to south. Also, pristine cold water lakes are clear and dominated by submerged plants while pristine warm-water

33 lakes are naturally turbid because of algae and covered with floating plants. In the latter case, it is not easy to decide when the natural state becomes associated with eutrophication. With increasing temperatures, cyanobacterial dominance and predominance of floating plants together with the possible disappearance of underwater vegetation happen at lower nutrient concentrations (Kosten et al., 2012). Climate warming may turn the northern lakes more similar to southern lakes (Moss et al., 2011).

It is evident that effects of climate change fasten eutrophication but contrariwise, eutrophication promotes also climate change. Eutrophication decreases the efficiency of sediments to store carbon and disposes the balance between external carbon and autotrophic carbon fixation in favour of the latter (Walter et al., 2006). Eutrophic lakes as metabolizers of carbon dioxide (Cole et al., 1994) lead to increasing production and respiration and greater volatilization of methane (Bastviken et al., 2011; Laas et al., 2012) which together with nitrous oxide segregated by denitrification are more effective greenhouse gases than carbon dioxide (Moss et al., 2011). Climate warming can foster distribution of productive cyanobacterial species to more north (Wiedner et al., 2007) where they may get supremacy in the conditions of high nutrient loadings (Moss et al., 2011).

3.4.1.3. Lag periods between impact and response

The ecological concepts derived from the theory of ecosystem resilience and alternative stable states affect also the selection of metrics that should respond to degradative as well to restorative anthropogenic action, and their calibration against pressure gradients (Nõges et al., 2009c). This task is complicated by different lags of biotic metrics in relation to changes in pressures.

Due to system resilience, ecosystems exhibit some degree of resistance to external forces. It deteriorates the performance of ecological indicators for both increasing and decreasing pressure intensities. It also causes hysteresis in case of switching between alternative stable states (described in Chapter 3.2.2.6). Usually, it requires more time and bigger effort to turn natural systems back to the previous better state. According to Jeppesen et al. (2005), a new equilibrium for total phosphorus resulting from reduced external phosphorus loading in lakes

34 was established within 10 to 15 years, while the recovery of biological variables took much longer. In a broad overview, Jones and Schmitz (2009) also found that the mean recovery time for water bodies was 10-20 years. Dokulil & Teubner (2005) described a five-year lag between the reduction of total phosphorus content and a decrease in phytoplankton biomass, with different response time for different species. The goal to revert the previous condition by removing pressures may also fail as Duarte et al. (2009) showed on an example of coastal ecosystems. The authors supposed that the ecosystem dynamics had undergone irreversible changes due to the human impact that could even question the achievability of the targets set by WFD for natural water bodies.

Some processes may act as “time bombs” requiring certain triggering conditions. Sulphates deposited in the upper <10 cm oxic sediment layer in freshwater bodies may be reduced to toxic hydrogen sulphide

(H2S) when anoxic conditions develop at the lake bottom. This may lead to precipitation of insoluble iron sulphide reducing the phosphate binding capacity of iron oxides and result in phosphorus release from sediments (Holmer & Storkholm, 2001). Another example of triggered chain reaction is the transformation of ammonium ions into toxic unionised ammonia at high pH (around 9.5 and above) (Randall & Tsui, 2002; Wicks et al., 2002). Together with too high water temperature, this was assumed to be the reason for a fish-kill in Võrtsjärv in summer 2013 (Kangur et al., 2016). In northern countries, the amount of winter precipitation may be discharged gradually to water bodies or form a powerful spring flood depending on air temperature that has the trigger function.

Other seasonal lags appear, e.g., when the nutrients deposited in the bottom layer of stratified lakes are released during autumn turnover (Sommer et al., 1986; Yang et al., 2016) or due to heterogeneity of warming (Winder & Schindler, 2004; Straile et al., 2015) and cause algal blooms. As different development stages of organisms have different pressure sensitivity, the impact depends largely on pressure timing. Resting stages of organisms may be protected against pressures but appear injured when dormancy is terminated (Aránguiz-Acuña & Serra, 2016). Low water levels at fish spawning time may affect their spawning success with consequences revealing themselves only in next generations (Järvalt & Pihu, 2002). For some biological pressures such as pathogens, parasites (Bauer, 1991) and alien species (Kreps et al., 2012),

35 the response lag depends on the latent period, i.e. the time necessary to achieve a critical mass.

Asynchronous changes in phenology of different trophic levels may cause mismatches in food chains (Straile et al., 2015). According to the above cited examples, the main reasons for response lags are (1) general inertia and resilience of the system, (2) need for a trigger factor, (3) physical barriers (e.g. thermal stratification, ice cover) postponing the effect, (4) different sensibility of development stages of organisms, and (5) incubation time needed.

3.4.1.4. Climate sensitivity of lake type parameters and status indicators

The WFD makes a clear distinction between (1) the landscape-based lake type parameters such as size, mean water depth, water level fluctuation, acid neutralising capacity, and background nutrient status (European Parliament, Directive 2000/60/CE) that are considered stable and not markedly affected by man, and (2) the status indicators (the “quality elements” according to WFD) such as biotic parameters and some chemical and hydromorphological features that are expected to reflect anthropogenic impacts by their deviation from values that they would have in unimpacted (“reference”) conditions. The concept of the WFD was developed in the 1990s when climate change was not understood as a global universal phenomenon and its impact on water bodies was totally ignored by this document (the 72-pages document does not include the word “climate”). Scientific research, however, continued compiling evidence on climate sensitivity of both status indicators and even the conservative lake type parameters (e.g. Eisenreich et al., 2005).

It has happened that, resulting from climate change, water bodies changed their type, especially the lakes located around the type boundaries (Nõges et al., 2007a). Due to climate-related reasons, the mixing type of lakes can change in warmer climate (Sorvari, 2001; Adrian et al., 2016) as well as in northern regions (Croley et al., 1998). According to Wheeler et al., (2003) and Richardson et al. (2004), increase in annual mean temperature by 1 ºC answer to about 100-200 kilometers of a latitudinal habitat shift. The same temperature change corresponds to approximately 200 meters altitude change (Körner, 1999). Lake morphometry as one of the most conservative parameters may undergo modifications due to climate

36 change as well. For example, since 1973, many lakes in fast warming Siberia were drained into subsurface because of breaching permafrost (Smith et al., 2005) and the future scenarios for warm regions foresee a shift from permanent to temporary water bodies due to increasing summer temperatures and decreasing precipitation (Nõges et al., 2007a).

Compared to quite persistent typology parameters, the labile by their nature water quality parameters within the types are easily influenced by climate change. For instance, in Lake Võrtsjärv, chlorophyll specific production decreased due to deterioration of light conditions following a water level increase in 1978/1979 which led to the change of dominating phytoplankton species Planktolyngbya limnetica (Lemm.) J. Komárková- Legnerová with Limnothrix redekei and launched adaptive responses of increasing chlorophyll content in phytoplankton cells (Nõges & Järvet, 1995; Nõges et al., 2010a; Nõges et al., 2011a). According to the analysis for the 44-year data base, another successful phytoplankton species, Limnothrix planktonica was favoured by increasing summer and autumn temperatures (Nõges et al., 2010a). As another example, increasing water temperature in spring in Estonian inland waters shifted the spawning time of bream (Abramis brama) to 10 days earlier, while no shift in the spawning time of roach (Rutilus rutilus) has been observed. As a consequence, the difference between spawning times of roach and bream decreased from 22 to 13 days and the difference in average temperatures at the onset of spawning by about 3 °C (Nõges & Järvet, 2005). Kangur et al. (2005) described a negative effect of high temperature on the abundance of the smelt stock in Lake Peipsi.

Also, type-specific reference conditions which should equal natural or almost natural conditions, cannot be regarded to be constant, they change as well due to climate change impacts on the chemical and physical environment of water bodies (Nõges et al., 2007a). The years of WFD implementation have shown that reference conditions defined for broad common lake types within a region have often brought about too large uncertainties which can be avoided by defining narrower and even site-specific reference conditions (Nõges et al., 2009c). Duarte et al. (2009) raised a question whether instead of static reference conditions, shifting baselines could be accepted for providing realistic perspective and public motivation for restoration and ecosystem recovery. An unreachable good status of a water body can be caused by irreversible environmental changes, therefore reference conditions should be more

37 flexible to respond natural changes (Nõges et al., 2007a; Battarbee et al., 2012). For this, based on periodical monitoring of reference sites, the values of reference conditions could be adjusted according to become known effects of long-term natural processes (de Wilde et al., 2002) which would be in accordance with the concept of adaptive management (Nõges et al., 2007a). Another possibility to achieve more flexible reference conditions is to relate them to natural causes of variability. In this case, single reference condition values would be replaced by a table or nomogram linking the value of a quality parameter to most significant natural factors ibid( ). For example, in Lake Võrtsjärv, depending on largely fluctuating water level, the proportion of aphotic bottom area may change from less than 5% to 90% of the total lake area, therefore reference conditions for macrophyte abundance in this lake should differ for low-water and high-water years (Nõges et al., 2007a).

Metabolism-linked variables are affected by temperature rise exponentially and may cause shifts in global carbon cycling. According to large-scale analysis (temperature data on 271 lakes and data on metabolic processes on 64 lakes around the World, time span 1970-2010), climate warming has remarkably accelerated metabolism in lakes (Kraemer et al., 2017). The accelerated metabolism has likely influenced food web dynamics, animal behaviour, life histories, species interactions and carbon cycling.

When it was realised that ignoring climate change by the WFD may have strong implications for the typology and quality assessment systems used for water bodies (Nõges et al., 2007a) and practically paralyse the concept, a Strategic Steering Group on Climate Change and Water was organised at European Commission with the aim to guide river basin management in a changing climate and a corresponding guidance was published in 2009 (WFD CIS, 2009).

The above mentioned guidance admits that although climate change is not explicitly included in the text of the WFD, the cyclical approach of the river basin management planning allows adaptively manage climate change impacts, meaning that we can revisit plans to scale up or down our response to climate change in accordance to monitored data. Surveillance monitoring should be able to detect long-term changes in natural conditions of water bodies including climate driven changes to avoid attributing their effects to other pressures that would lead to an errant assessment of the ecological status of water bodies.

38 3.4.2. Ways forward

3.4.2.1. Monitoring of reference sites

According to WFD, type-specific reference conditions in unimpacted sites have been set as a basis for evaluating the ecological status of water bodies (Directive 2000). The aim of environmental monitoring is to find out anthropogenic deviations of water body conditions from the reference. Changes independent from human impact including alterations associated with natural ageing of lakes and also partly climate-related changes can be identified through periodical monitoring of reference sites. Still, there are some weak points in this approach. First, it is hard to find sites with the total absence of any anthropogenic influence. In practice, reference conditions are often set according to measurements at “best available” sites that compromises the concept (Nõges et al., 2009c). Secondly, also climate change is partly caused by human activities but its consequences cannot be remediated at a river basin scale. Thirdly, investigating the climate impact in reference conditions to exclude noise from human impact, we are not able to estimate the effect of climate to impacted (e.g., eutrophied) lakes (Battarbee et al., 2012).

European-wide networks of surveillance monitoring stations such as European Environment Information and Observation Network (EIONET) or Long-Term Ecosystem Research (LTER) including high status reference sites as well as a quantity of degraded and restored sites, are valuable sources to detect and analyse long-term trends and effectiveness of measures taken to improve the status of water bodies (Hering et al., 2010).

3.4.2.2. Coherence studies

In addition to local driving forces, lakes respond to regional influences (Livingstone, 2008). Mostly climate forcing and catchment processes are considered as drivers for the coherence of lakes within the same region, and in practice, we may remove the coherent changes from the total variability in order to better follow local impacts. Climate-induced spatial and temporal coherence in its diverse forms of appearance is shown to exceed the effects of local driving forces on lakes (Livingstone et al., 2010a). The authors exemplify the statement with short-term spatially coherent regional air temperature that, if used in empirical 39 models, provides a good estimate of short-term fluctuations in summer lake surface water temperature, even within a large, topographically heterogeneous region. Analysis of long-term regional coherence representing inter-annual to interdecadal time scales showed that the physical properties of lakes vary generally in a more coherent way than chemical and biological properties (Kratz et al., 1998). As an expression of the long-term coherence, ice phenology in fresh water bodies was shown where North Atlantic Oscillation (NAO) as well as latitude as a proxy for annual mean temperature have to be considered (Hurrell, 1995; Blenckner et al., 2004; Magnuson et al., 2004; Nõges & Nõges, 2014). The same applies to river and lake surface water temperatures, and even to deep-water lake temperatures though with some deviations which requires cautiousness in generalising since deep-water temperatures reflect rather meteorological conditions of spring turnover than the annual climate (Livingstone 1997; Dokulil et al., 2006; Livingstone et al., 2010a).

Hydrological variables are governed mainly by spatially heterogeneous precipitation, nevertheless, they also show spatial coherence, like it was explained by Livingstone et al. (2010a) in case of north-eastern European lakes Võrtsjärv and Peipsi. Long-term coherent fluctuations have a remarkable influence on ecosystems of these lakes whereby shallower lakes are more sensible to water level fluctuations than deeper ones. Also, NAO as the driving force behind the winter precipitation should be considered in coherence studies, so in the same study, the effect of winter NAO was shown to induce long-term coherent water level fluctuations of lakes Ladoga, Peipsi and Võrtsjärv. Coherent long- term periodicity has been described also for lakes in northern Germany (Behrendt & Stellmacher, 1987) and Russia (Doganovsky & Myakisheva, 2000).

Investigating nutrient concentrations in lakes, regional coherence has to be kept in mind as well. Numerous publications have found regional coherence for different chemical constituents in lakes related with NAO driven winter temperatures, e.g., for reactive silica and pH (Weyhenmeyer, 2004), conductivity, calcium and alkalinity (Järvinen et al., 2002), total organic carbon (Arvola et al., 2004), oxygen (Livingstone, 1997; Straile et al., 2003), and soluble reactive phosphorus (George et al., 2004b). Nitrate concentrations can be positively correlated with mean air temperature during the previous winter in Fennoscandia (George et

40 al., 2004a; Weyhenmeyer, 2004), or inversely with mean temperature of the same winter in Central and western Europe (George, 2000; Monteith et al., 2000; George et al., 2004a,b; Davies et al., 2005; Jankowski et al., 2005), additionally, long-term regional coherence of nitrates in fresh water bodies is influenced by several more factors such as lake size, depth and other geomorphological features (Jankowski et al., 2005; Blenckner et al., 2007), changes in nitrogen deposition (Weyhenmeyer et al., 2007), use of fertilisers in the catchment (George et al., 2000).

Though coherence in biological variables is less common than in physical and chemical time-series, it can occur over large spatial scales mostly due to temperature-related or rate-limiting coherent physical forcing (Livingstone et al., 2010a). Temperature dependent coherence of biological variables has been found more often during winter and spring while in summer rather nutrient limitation or biotic interactions influence population dynamics (Adrian et al., 2006; Blenckner et al., 2007). Resulting from ice-off timing and the NAO, phytoplankton phenology can be coherent over large areas (Adrian et al., 1999; Weyhenmeyer et al., 1999; Gerten & Adrian, 2000). e.g., spring phytoplankton bloom and nutrient depletion in Swedish large lakes were shifted up to one month earlier (Weyhenmeyer et al., 1999; Weyhenmeyer, 2001), phytoplankton community structure in German lakes of different geographical regions responded similarly to seasonal meteorological forcing (Anneville et al., 2004, 2005). Winter NAO dependent coherence was found for zooplankton indices (Straile & Adrian, 2000; Scheffer et al., 2001; Anneville et al., 2002; Straile, 2002; Wagner & Benndorf, 2007) and for aquatic invertebrates and fish (Elliott et al., 2000; George, 2000). In lakes Võrtsjärv and Peipsi, long-term increase in water temperature lead to a coherent 10 days earlier spawning of bream (Abramis brama) (Nõges & Järvet, 2005).

Climate driven spatial coherence has been found also at a long-term supra-regional scale, reaching thousands of kilometres. Winter NAO has been shown to be a major driving force for it in northern and western Europe (Blenckner et al., 2007), but its negative or positive relation with physical, chemical and biological indicators depends more on local effects. Large-scale temporal and regional coherence appears also in carbon cycling in lakes (Lapierre et al., 2015).

41 3.4.2.3. Comparison of contrasting years

As an alternative tool to long-term data analysis, comparison of years with contrasting weather conditions is used to follow climate change impact on water bodies and/or to predict processes in ecosystems in a warmer climate (Nõges et al., 2011a). For example, the various impacts of the extraordinary heat wave in central Europe during summer 2003 were analysed in thousands of papers. Also, extreme winter conditions have been shown to have long-range effects on biota of water bodies. Extreme meteorological conditions impact the annual heat and matter cycling in lakes generally, but also timing and consecution of the events may be significant for shaping the conditions in lakes (Straile et al., 2010).

The main weakness of the “contrasting years” method is that many organisms survive the short-term (e.g. one year) extreme conditions while the prolonged deviation could be lethal for them. Therefore, one should be careful making far-reaching conclusions based on a comparison of one-year extremes but rather regard these as natural whole lake experiments (Nõges et al., 2011a).

The method is more effective when based on a long-term dataset of multiple contrasting years. Comparing contrasting years, e.g., high water vs low water years enables to spotlight the differences in ecological responses to extreme hydrological situations, and to find out how water quality indices behave in these critical situations. For instance, inter- annual water level fluctuations in Võrtsjärv exceeding 3 m modify the intensity of sediment resuspension and strongly influence all water quality parameters (Nõges & Järvet, 1995; Nõges & Nõges, 1998, 1999; Nõges et al., 2011a; Toming et al., 2013). The phytoplankton biomass in Võrtsjärv was significantly lower in years of high water level (Nõges et al., 2003b) being light-limited, and was controlled by nutrient availability at low water levels (Nõges et al., 2010b), whereas also species composition was shown to be influenced by water level (Nõges et al., 2003b). The growth of bacterioplankton in Võrtsjärv appeared to be favoured by high water level (Kisand & Nõges, 2004). Extremely low water levels (like in Võrtsjärv in 1995/1996) decrease drastically the mean depth and volume of shallow lakes, increasing concentrations of ions and phytoplankton biovolume, strengthening sediment resuspension, and, potentially, causing winter fish kills due to depletion of dissolved oxygen. Further effects such as changes in phytoplankton composition can

42 follow (Nõges & Nõges, 1999; Nõges et al., 2003b). In Lake Peipsi, low water levels coinciding with elevated water temperatures have repeatedly caused cyanobacterial blooms leading to summer fish-kills (Kangur et al., 2005; Kangur et al., 2013). For these large and shallow lakes, years of extremely low water level are most critical for the ecological status (Milius et al., 2005; Haldna et al., 2008; Nõges et al., 2010b).

3.4.2.4. Detrending of long-term data for known sources of variability

In real world, different factors act in concert. Depending on the question of interest, we can decide what should be the signal and remove part of variability attributable to known factors and term it as noise. For climate change impact studies, local anthropogenic effects form the noise that should be preferably removed from the data to study the climate change signal. In the WFD approach it is the opposite – local anthropogenic impacts are in the main focus while coherent climate induced changes form the noise that should preferably be removed from the data. Removing an aspect from the data that supposedly causes some kind of distortion or noise is called detrending, while the residual variability after this operation forms the signal (Pouzols & Lendasse, 2010). Often this simple method enables to disentangle the effects of natural and anthropogenic variability in long-term data. George et al., (2004a) used detrending to detect the influence of NAO on the physical, chemical and biological characteristics of lakes in the English Lake District between 1961 and 1997. In this analysis, for example, the increasing load of nitrates from the catchment was the dominant source of variation in the raw time series, but after detrending, the residuals correlated strongly with inter-annual variations of winter weather. According to Pouzlos and Lendasse (2010), detrending is a necessary preprocessing step for statistical analysis in many fields. The method has been helpful in studies of e.g, climate-phenology relationships (Iler et al., 2017), biological turnover of phytoplankton (Cabecinha et al., 2009), and palaeolimnological analysis of lake sediments (Lanci & Hirt, 2015). Still, Pouzlos and Lendasse (2010) suggest to use detrending with care since sometimes it may be ineffective or give only a small effect or be even harmful to further analysis. As time series are often nonstationary, the nonlinear detrending method may result in better performance than linear detrending (ibid). Deep understanding of scale-specific processes and relations is needed to avoid misinterpretation, especially in resilience

43 studies of ecological associations when using detrended time-series in mathematical models (Baho et al., 2015). Detrending presumes a simple additive effect of different factors that may not be the case on many occasions. In case of multiple stress, various stress factors are often interacting in a synergistic or antagonistic manner, i.e. strengthening or weakening each-other effects.

3.4.2.5. Using multivariate statistics

As a rule, several biotic and abiotic factors, part of them interacting with each other, act concurrently contributing to the total variability of any variable of interest. Multivariate analysis methods are developed to consider these multiple factors and their interactions and to avoid overemphasizing or misinterpreting their effects (Rodríguez & Magnan, 1995). Statistical tools for disentangling of causal webs can be divided into partialling out and partitioning tools (Cohen et al., 2003). Partialling tools enable to sequester one factor, statistically holding constant or controlling other factors. To evaluate to which extent the observed effect of a variable can be assigned uniquely to that variable and which is the share of other factors, variance partitioning is the next suitable statistical method. A prerequisite for multivariate analyses is a large number of individual sampling units which broadens the ranges of measured valuables, increases the probability of identifying the existing effects and nonlinear relationships between dependent and independent variables (Rodríguez & Magnan, 1995). There are many applications of multivariate methods for data organisation (reduction, simplification, sorting, grouping), investigation of the dependence among variables, prediction, construction and testing of hypothesis (Johnson & Wichern, 2007): multivariate linear regression models, principal component analysis, canonical correlation analysis, discriminant analysis, classification, and clustering.

Multivariate analyses have been applied for various steps relevant for ecological status assessment of lakes such as lake typology, subdividing large lakes into water bodies (areas with significant differences in characteristics in the WFD context), pressure analysis, and analysing factors determining structure of biotic communities indicating water quality (“biological quality elements” in the WFD jargon). Below are given some typical examples of these tasks. The principal component analysis was applied for creating the typology of Greek lakes according

44 to WFD requirements and the respective lake classification (Kagalou & Leonardos, 2009). Cluster analysis was used to group the sampling points at Lake Balaton according to the strength of physical, chemical and biological parameter effects (Hatvani et al., 2011), and to find out the number of reasonable sub-areas in Lake Balaton to retain a representative number of sampling locations (Kovács et al., 2012). Multiple regression analysis showed that the rate of fertilization was the most important factor determining the nitrogen runoff from the catchment of Lake Võrtsjärv (Järvet et al., 2002). Since the end of the 1980s, TP concentrations in the main inflow of Võrtsjärv and TP loadings showed a decreasing trend while the loadings of TN increased in the 2000s, resulting in an increased of the TN/TP loading ratio.

Natural populations and communities are always influenced by more than one factor. In ecological studies of aquatic environments, most commonly multivariate analyses are used to identify which factors are driving spatial and temporal patterns of the structure of communities (e.g., Cabecinha et al., 2009; Izaguirre et al., 2015; Strock et al., 2016).

3.4.2.6. Model calculations

Numerical modelling is broadly applied for exploring complex systems and testing hypotheses regarding the effect of various driving forces. In lake models, complex interactions between physical, chemical and biological processes affecting the water quality of a lake are described by a series of mathematical equations. The modelling process starts from building a conceptual model followed by describing the various steps in mathematical terms. Process based models use formalised laws of nature as the basis whereas in empirical models statistical relationships can be used. Often models apply both. Model verification must ensure that the conceptual model is reflected accurately in the computerized representation. In the calibration phase, particular features of the modelled lake are introduced to the model structure and adjusted in a way to reach the best fit between model output and real measurements. In a validation phase, the model performance is tested on a data set different from that used for calibration. A validated and well-performing model is a useful tool for testing hypotheses or effects of various management options on the lake.

45 In a study by Woolway et al. (2017), a process-based model MyLake (Saloranta & Andersen 2007) was applied to large polymictic Lake Võrtsjärv to quantify the effects of decreasing wind speeds and increasing air temperature on the frequency and duration of thermal stratification episodes. The model was validated for a 3-year period using high-frequency wind speed and water temperature data and run for a 28 year period using long-term daily meteorological data. The model simulated well long-term in situ water temperatures. Sensitivity analysis demonstrates that decreasing wind speed was the major factor causing substantial changes in stratification dynamics since 1982 while increasing air temperature had a negligible effect.

Long-term datasets containing thousands of records often appear to be fragmentary and therefore difficult to analyse. Modelling is a good tool for getting the maximum benefit from such data. For example, Haldna et al. (2013) used covariance structures provided by SAS software to improve the quality of the predictions of total phosphorus concentration from the year, the day of the year and geographical coordinates in a predictive model for Lake Peipsi.

Spatial autocorrelation may exist between lake status indicators over large geographical scales, e.g. lakes located closer to each other have more similar nutrient water colour relationships compared to lakes located farther (Fergus et al., 2016). Authors used this spatial dependency in models to improve chlorophyll predictions through its relationship with total phosphorus over water colour. They used a Bayesian framework and spatially-varying coefficient model which allowed the coefficients to vary over continuous space instead among the discrete ecological regions. Such way, spatial scales were captured separately for total phosphorus and water colour.

46 4. MATERIAL AND METHODS

For the present thesis, mostly long-term datasets for Estonian large lakes Võrtsjärv (Papers I, II, III, IV) and Peipsi (Paper III) were used. For paper V, measurements were made during summers 2007 and 2008 also in Lake Harku, and the Peipsi data for paper VI originate from 2003. Descriptions of the study lakes are given in original papers: for Võrtsjärv most comprehensively in paper II, for Peipsi in paper III and for Harku in paper V.

The methods are described explicitly in the papers, below only the basic methods are listed.

• To reconstruct the trophic state history of Lake Võrtsjärv (paper I), we used long-term data on most common trophic state parameters for lakes such as phytoplankton biomass (BM), chlorophyll a (Chl a), total nitrogen (TN), total phosphorus (TP), and Secchi depth (S) measured at the main monitoring station in Võrtsjärv from May to October. Time series of BM and S started from the year 1965, that of Chl a from 1982 and those of nutrients from 1983. Data were standardized by correcting the data for the effects of changing water level and removing the seasonality from the data. • To analyse spatial and annual variability of environmental and phytoplankton indicators in Võrtsjärv (paper II), we used lake monitoring data on pelagic parameters such as water temperature, Secchi depth, hydrochemistry and chlorophyll a (Chl a) collected from 10 sampling points during 11 lake surveys in August 2001- 2011. Data on phytoplankton composition were collected during 2008-2011. Tree Clustering method and the Variance Components analysis of STATISTICA 8.0 (StatSoft, Inc. 1984-2007) was used for data analysis. • Long-term trends in water temperature, nutrient loading, and water quality in Peipsi and Võrtsjärv (paper III) were analysed using data provided by Estonian Environment Agency for two meteorological stations, Tiirikoja and Tartu-Tõravere and hydrological stations in the rivers Väike Emajõgi, Emajõgi, and Narva. Trends and step changes in air temperature (AT) and surface water temperature (SWT) were studied in the lakes, while water discharge (Q), concentrations of

47 total nitrogen (TN), total phosphorus (TP), and the TN/TP ratio were studied in the rivers. A non-parametric Mann-Kendall test for gradual trends and a Cumulative deviation test for step changes were applied (Trend C1.0.2 Catchment Modelling toolkit, CRC for Catchment Hydrology). These methods are described in detail by Kundzewicz and Robson (2004). • A 7-year (from 2008 to 2014) database was used to study the dependence of dissolved organic carbon (DOC) on other water and environmental variables in Võrtsjärv (paper IV). Pearson correlations, regression models and a novel predictive modelling technique called Boosted Regression Trees (BRT, Elith et al., 2008) were used to analyse the determination power of different water and environmental variables for DOC. • Data for modelling primary production of three lakes were collected during summers of 2007 and 2008 (paper V). The dataset included incident solar irradiance for the photosynthetically active radiation

(PAR) region (qPAR), spectral values of underwater planar quantum

irradiance (q(λ , z)), spectra of diffuse attenuation coefficient K ( d (Δλ )), the Secchi depth (S, m), and 25 profiles of measured primary production P(z,meas) from May to end of September. Spectral values were determined in two ways: (1) using spectral values from three channels of the spectroradiometer BIC-2104, and (2) using beam attenuation coefficient spectra in filtered and unfiltered water samples over the wavelength range of 350-700 nm measured by a Hitachi U3010 laboratory spectrophotometer. Vertical profiles of primary production (P(z,calc) were simulated using the spectral model by Arst et al. (2008). • Field campaigns to measure optical properties in Lake Peipsi were carried out in 2003 (VI). Underwater irradiance was measured using two instruments: a GER 1500 (capable of measuring in the upward and downward directions, spectral range 400-900 nm, resolution 2 nm) and a profiling radiometer BIC-2104 (with bands of 412, 555, 665 nm, an underwater and a surface PAR sensor). Surface water samples were collected for spectrophotometrical pigment analysis and light absorption measurements. The concentration of suspended matter was measured gravimetrically. To compare the results from Peipsi with other lakes, we used the database of small Estonian and Finnish lakes.

48 5. RESULTS

5.1. Effect of water level and seasonality on water quality parameters in Võrtsjärv (I)

The values of all selected trophic state indicators, except TN, were significantly related to the water level (WL) during several months of the vegetation period (Table 1, I), the relationships being strongest for the Secchi depth (S). All significant relationships were negative, i.e. the lake looked significantly more eutrophic at lower WLs. In May, none of the parameters had a significant relationship with WL. In addition, the regression with WL was non-significant for LnChl a in June and September, and for LnTP in June and July. On average, the relationships between monthly values of trophic indicators and their seasonal averages (Table 2, I) were strongest in August and September and weakest in May.

After correction of the trophic state indicators for WL changes and seasonality, the mean values of the full time series did not change significantly P ( of t-test for means between 0.3 and 0.9 for different variables). The correction for WL did not significantly change the total variability of the time series (P of t-test for variance between 0.1 and 0.9 for different variables). The correction for seasonality diminished considerably the variability ranges of all variables (Fig. 3, I). The total variance decreased fivefold for LnBM, LnChl a, and LnTN, nine-fold for LnTP, and 21-fold for Ln1/S. Despite the unchanged long-term averages, both corrections affected substantially the monthly values (Fig. 4, I) and the seasonal averages (Fig. 5, I). The correction for the effects of WL changes followed closely the dynamics of the mean WL for May–October (Fig. 5, I) correcting the BM down by up to 23% in low-water years and up to 41% up in high-water years. The seasonality removal had even stronger effect ranging from 28% down to 64% up. In nearly 70% of the cases, the seasonality removal corrected the data to the same direction with the correction for WL giving a summary effect between 41% down and 65% up. For the whole 44-year period analysed, there was no significant trend in the WL (Fig. 5, I); however, there was a stepwise jump between 1977 and 1978 (Worsley test, P < 0.01). Since 1978, the WL had a slight decreasing trend (P < 0.05). The corrected biomass series (Fig. 6, I) indicated an abrupt decrease from 1978 to 1979 followed by a significant P( < 0.01) increasing trend since that.

49 The corrected Secchi depth had an increasing trend from 1981 to 1992 and decreased since that (P < 0.01). The time series of total nutrients corrected for the WL effects did not show any significant trend.

The CODMn levels measured in the period 1968-1977 (10.6 ± 3.2 mg O l-1) were significantly P( < 0.01) lower compared with those in 1998-2008

(13.0 ± 1.5). Within both periods, CODMn had a highly significant P( < 0.01) increasing trend.

The average trophic state proxy index calculated by applying the class boundaries of trophic state indicators (Table 3, I) to the long-term data (Fig. 7, I) showed two distinctive periods in the changes of the ecological status of the lake: the initial fast eutrophication in the 1970s (assessed only on the basis of BM and S) followed by a temporary improvement in 1979-1980 and a worsening trend afterwards. The latter trend for the whole period was clearly seen also without using the additional data (TP, TN, Chl a) for posterior years, and was not caused by the difference in data availability. Since 2001, the data show an accelerated deterioration of the lake’s status, although also the uncertainty of the estimate has increased in this period.

The German PTSI index (Mischke et al., 2008); however, showed different results for the period since 1979 (Fig. 8, I) when due to the change in dominant species (earlier Planktolyngbya limnetica (Lemm.) was replaced by Limnothrix redekei (Goor) Meffert and L. planktonica (Wołosz.) Meffert.). The taxonomic index revealed an irreversible drop in the ecological status of the lake.

5.2. Spacial and annual variability of environmental and phytoplankton indicators of Võrtsjärv (II)

Among physical and chemical variables measured at ten sampling stations in Võrtsjärv, the inorganic forms of nutrients (NO2-N, NO3-N, NH4-N,

Si, and PO4-P), total iron (TFe), and total suspended solids (TSS) were relatively most variable with variation coefficient (Cvar) values equal or exceeding 50%. Low variability was characteristic of pH, major cations, water temperature, dissolved oxygen (DO), and water level. General phytoplankton characteristics, such as a number of species (Nsp), Chl a, and total biomass (BM) were less variable than the biomasses of single algal classes. The long-term permanent sampling site, C. Lim. (Fig. 1,

50 II), located in the deepest part of the lake had the smallest average Cvar

(34%) and the southernmost station Riiska the largest average Cvar (56%) among stations.

The average Cvar for phytoplankton variables was twice as high as that for physical and chemical variables and decreased towards the central part of the lake (Fig. 2, II). As a characteristic pattern, Cvar for phytoplankton variables was elevated at the stations affected by major inflows. No such effect was observed for the mean Cvar for physical and chemical variables, which was rather equal for all stations.

Two distinct clusters of stations were distinguished in Võrtsjärv both by physical/chemical and by phytoplankton variables (Fig. 3, II). The two southernmost stations under the influence of the Väike Emajõgi River deviated clearly from the rather homogeneous group of the other eight stations in the main pelagic area which we termed as the ‘Võrtsjärv Proper’. If to draw the provisional boundary between these two clusters at the latitude of the mouth of the Rõngu River, the Võrtsjärv Proper occupies 98% of the lake area.

Among stations of the Võrtsjärv Proper, the year-to-year variability dominated strongly over the spatial variability, the latter being almost negligible for most of the variables (Fig. 4, II). For variables such as - water surface temperature (SWT), carbonate alkalinity (HCO3 ), total - + hardness (Hard), Cl , NO2-N, ColourPt-Co, and K , the inter-annual 2- variability accounted for more than 80% and for SO4 and Si even 90% of the total. The highest contributions of spatial variability reaching 14-15% of the total, was found for Ca2+, specific conductivity (Cond),

DO, and BOD7. The variability explained by the combined effect of ‘Year’ and ‘Station’ (and including also the error term) ranged between 6 and 86% of the total variability among the eight stations.

The two major proxies for the climatic (natural) variability – the surface water temperature and water level – explained 1-74% of the total variability in the monitored variables (Table 4, II). Most affected by the natural variability were variables such as TSS, TP, Chl a, Secchi, and TN, commonly used in the ecological status assessment of lakes. Although the biomass of three phytoplankton groups – cyanobacteria, diatoms, and chrysophytes (Bcya, Bdia and Bchr, respectively) – correlated

51 significantly both with Temp and WL, the multiple regression model relating these groups with the climate proxies remained non-significant.

Confirming the results of the cluster analysis, also the Student’s t-test showed that the mean values of most of the parameters monitored at the long-term permanent monitoring station ‘C. Lim.’ differed significantly from those at the two southernmost stations, ‘Riiska’ and ‘Pähksaar’ (Fig. 5, II). Only for conductivity and DO, the area deviating from the main monitoring station extended also to ‘Sula Kuru’ and ‘Õhne’ stations. None of the mean values of the remaining 31 variables measured at C. Lim. differed significantly from those measured at any of the other 7 stations of the Võrtsjärv Proper.

5.3. Contemporary climatic, hydrological and loading trends in Võrtsjärv and Peipsi (III)

At Tartu and Tiirikoja stations, the average annual air temeperature (AT) increased by 0.4 ºC per decade in 1961–2004 (Nõges, 2009b), and the increasing trend of AT (Fig. 2A, III) had a step change in 1987 (Table 1, III).

The average annual amount of precipitation (PR) had significant increasing trends at Tartu and Tiirikoja in 1961–2004 (Fig. 2C, III) with a step change in 1976 in Tartu and in 1977 in Tiirikoja Table 1, III). The average annual discharge (Q) in the River Väike Emajõgi was significantly positively related to the average annual PR in Tartu, and the average annual Q in this river had a significant increasing trend (Fig. 3A, III).

In the River Väike Emajõgi, the yearly average concentrations of TN and TP and the TN/TP ratio decreased significantly since 1986 (Fig. 3B, C, D, III). In the Emajõgi and Narva Rivers as well as in the lakes, the yearly average TN and TP concentrations and the TN/TP ratio did not reveal any significant trend (Table 2, III). In the 1980s, the riverine loading of nutrients into Lakes Peipsi and Võrtsjärv increased drastically; while in the early 1990s, a sharp decrease occurred, primarily in TN loadings. Since the end of the 1980s, TP concentrations in the main inflow of Lake Võrtsjärv and TP loadings showed a decreasing trend (Table 1, Fig. 4B, III) while the loadings of TN increased in the 2000s resulting in increasing TN/TP loading ratio.

52 5.4. Relationships between DOC and other water and environmental variables (IV)

Seasonally, coloured dissolved organic matter (CDOM), water colour by

Platinum-Cobalt scale (colourPt-Co), dissolved oxygen (DO), Secchi depth (S), water level (WL) and inflowing riverine discharges (Q) were usually somewhat higher from December to March, contrary to Chl a, total suspended matter (TSS, denoted as TSM in papers IV and VI), water temperature (WT), pH, precipitation (PR) which showed higher values from May to October (Fig. 2, IV). The concentrations of DOC and chemical oxygen demand by permanganate (CODMn) were quite high in January and February decreasing to the spring as were to other water colour variables (CDOM, colourPt-Co), but in reverse from others, they increased again after the spring low (Fig. 2a, IV). Inter-annually, years with lower WL, I and PR corresponded to higher Chl a, TSS, pH and

DO (Fig. 3b, c, IV) while the variables connected to the water colour and transparency (CDOM, colourPt-Co, CODMn and S) were usually lower in those years (Fig. 3a, IV). The latter was, however, not valid for DOC. Based on the annual mean data, DOC concentration was statistically significantly correlated only with S (Table 3, Fig. 4,IV ).

According to the applied BRT analysis, the annual mean values of CDOM, Chl a, TSS, S, and PR explained 1.52%, 0.50%, 0.45%, 97.3% and 0.21% of the DOC variance, respectively. As Secchi depth described the most significant proportion (97.3%) of the variability, it appeared to be the most powerful predictor for the annual DOC concentration in Võrtsjärv. Since correlations based on monthly data showed very high collinearity between some of the variables under study, it was not possible to use the BRT analyses or multiple regressions to predict the DOC on ice-free periods in Lake Võrtsjärv.

The results from large-scale analyses showed that values of CDOM and Chl a were both significantly positively correlated with concentrations of DOC among the analysed lakes (accordingly r = 0.85 and r = 0.45) and linear regressions for predicting DOC via CDOM or Chl a were statistically significant (Fig. 5a, b,IV ).

53 5.5. Phytoplankton productivity in shallow lakes (V)

The concordance of measured and calculated primary production was satisfactory for lakes Võrtsjärv and Peipsi, the determination coefficients were 0.873 and 0.879, respectively. The calculated diurnal sum of integral primary production for Lake Peipsi was 711 mgC m-2 day-1 on 27 June 2008, for Võrtsjärv 1410 mgC m-2 day-1 on 17 July 2007, and for Lake Harku 2237 mgC m-2 day-1 on 23 July 2008. Marked influence of cloudiness (Fig. 4 and 5, V) on the underwater light field as well as on primary production was seen in the calculated vertical profiles.

5.6. Effects of coloured dissolved organic matter on the attenuation of photosynthetically active radiation in Lake Peipsi (VI)

The highest value of the absorption coefficient for yellow substance

(ays, 380) in Lake Peipsi was measured in the point closest to the Suur- Emajõgi River mouth (Fig. 2, VI) and close to the shore or Piirissaar Island. The near-shore (up to 1.9 km) concentrations of CDOM in -1 Peipsi are higher (ays(380) = 7.1-16.7 m ) than in open area (ays(380) = 4.1-5.6 m-1) whereby large spatial and temporal variability is characteristic for this lake. The used exponential model for calculating the absorption spectra for the yellow substance gave less than 1% higher values than measured in the region 380-480 nm, but in other spectral regions, the values calculated by the model were underestimated by up to 40%. In Peipsi, the integral attenuation coefficient over photosynthetically active -1 radiation (Kd,PAR) varied from 0.89 to 3.76 m showing the thickness of the euphotic zone (the depth to which 1% of the subsurface irradiance reaches) to be 1.2-5.2 m and indicating good illumination of the water column in this on average 7 m deep lake. According to statistical analysis of data from Estonian and Finnish lakes for the years 1994-2000, the values of Chl a varied from 1.28 to 87.33 mg m-3, TSS was 0.5-20.80 -3 -1 g m , and ays(380) 0.68-14.24 m . In conformity with these results, the main factor influencing the light attenuation in the violet and blue parts of the spectrum is the yellow substance (at 400 nm R2 = 0.897) and in the orange and red parts the suspended matter including Chl a. In Peipsi yellow substance causes 34-80% of the daylight attenuation in the blue region and the lowest, 2-10% in the 665 nm band.

54 6. DISCUSSION AND SYNTHESIS OF ORIGINAL PUBLICATIONS

6.1. Distinction of effects of natural factors on common water quality indicators in shallow lakes

6.1.1. Influence of water level and seasonality (I)

The ultimate goal of lake monitoring should be the establishment of a coherent and comprehensive overview of the ecological and chemical status of lakes in a nearly real-time regime to enable water managers to take measures if the conditions deteriorate as a result of human impact. Selecting those among the multitude of measurable physical, chemical and biological parameters that reliably reflect the effects of human activities, but remain insensitive to extraneous conditions is a step of extraordinary importance (Karr & Chu, 1997). Several methods exist to distinguish the effects of natural and anthropogenic variability in long- term data sets (see Chapter 3.4).

For simplicity and transparency reasons, we selected the linear adjustment method to statistically account for the effects of the changing water level (WL) on the common trophic state variables in the long-term data of Lake Võrtsjärv (Fig. 1, I). In this way, the direct effect caused by WLs deviating from the long-term monthly averages could be eliminated while retaining the meaningfulness and original dimensions of the variables. The considerable correction of single seasonal mean values by more than ±40% shows the high importance of WL changes in shaping these variables. The correction dampened efficiently the effect of the record low WL in 1996 when most of the common water quality indicators were far out of range indicating strong hypertrophy (Nõges & Nõges, 1999).

The method we choose for seasonality removal allows using data from all months in which the variable has a significant relationship with the seasonal mean value. This way of standardizing has three main advantages: (1) every single measurement can be equally used for the assessment purposes, (2) more correct seasonal averages can be calculated for years

55 with data gaps, and (3) the obtained values are ecologically meaningful as estimates for the vegetation period mean value.

The calculated summary index (Table 3, I) revealed two distinct periods in the ecological status of Võrtsjärv: fast eutrophication of the lake in the 1970s followed by a temporary improvement after the high water year of 1978 (Fig. 6, I) and a slow continuous deterioration trend. The German PTSI index characterizing phytoplankton species composition (Mischke et al., 2008) distinguished even more clearly the same periods (Fig. 7, I). Paradoxically, the taxonomic index indicated a drop of the ecological status from ‘‘good’’ to ‘‘moderate’’ to ‘‘bad’’ where the average trophic index indicated an improvement. The thorough analysis of interactions between changing nutrient conditions and phytoplankton composition in connection with water level changes gave an explanation for this discrepancy between the two indices. There was a lag period between the impact of increasing nutrient concentrations and response of phytoplankton whereat the sharply increased water level had a triggering role for a shift of system to another steady state. The jump in the German index can be explained by the lake having reached a high enough trophic state where all three species could coexist, and a strengthening of light limitation at which the more specifically adapted Limnothrix species outcompeted the more eurytrophic P. limnetica. No reversal has still happened as this would require a reduction in both light limitation and trophic state. Though the WL changed as much from 1996 to 1998 as in 1977–1978 (Fig. 5, I), the very low WL in 1996 did not change permanently the species composition (Nõges & Nõges, 1999). This indicates also that the dynamics of cyanobacteria is not well understood. The German index is calculated as biomass weighed average product of trophic scores showing the average trophic preferences of indicator species, and stenoecy (=specificity) factors showing the indicator power of the species (i.e. how specifically they indicate the given trophic state). A comparison of the values of these parameters applied in the German index for the dominating cyanobacteria species in Võrtsjärv shows that they indicate nearly the same trophic status but the stenoecy factors differ substantially (Table 4, I). The PTSI index indicated a sharp deterioration of the status resulting from the switch of the dominants to more shade tolerant species. On the other hand, this change indicates to an aggravation of the situation, as the domination of Limnothrix species means a switching of the system to a steady state. This steady state presents a self-induced habitat, in which competitors

56 fail because of low-light conditions are reproduced by the dominants based on the efficient exploitation of nutrient resources (Mischke & Nixdorf, 2003). Still, it may be not so clear for Lake Võrtsjärv where the influence of resuspension and humic substances on light conditions is also remarkable and the dominance of Limnothrix species is strongly influenced by external conditions as precipitation, temperature and WL.

Both of the used indices have their advantages and disadvantages. Due to the strong resilience of the phytoplankton community, the taxonomy based index did not almost change during the fast eutrophication in the beginning of the 1970s, neither demonstrated clear patterns in the period after the change of the dominants. The big change occurring in 1979 was brought upon not so much by a change in the trophic state but expressed the tipping point evoked by a disturbance – the sudden increase of the WL. Such behaviour of the index throws doubt upon the popular belief that phytoplankton provides a good indication of lake trophic state and respond quickly and predictably to changes in nutrient status (e.g. Murphy et al., 2002). In several shallow lakes, reduction of nutrient loads has not led to discernible recovery. Internal loading and the mechanism of hysteresis, i.e. less nutrient concentrations are needed for recovering the previous better equilibrium state than it was at the time of its decline (Scheffer et al., 1993, 1997; Jeppesen et al., 2007), offer an explanation for the resistance of cyanobacteria dominance in shallow lakes to restoration efforts by means of nutrient load reduction. In that way, phytoplankton indices really reflect the ecological effect of human impact, which can last much longer than the direct impact itself. In addition, usually several other factors such as weather conditions or spatial heterogeneity, and even the cyanobacterial dominance itself, play role in developing an alternative regime (Scheffer & van Nes, 2007).

The index based on traditional trophic state variables and corrected for the simultaneous WL changes demonstrated an evolution of the trophic state more consistent with the expert opinion. However, it did not capture the tipping point occurring in phytoplankton, one of the biggest changes ever observed in the plankton community of Võrtsjärv, and without phytoplankton composition data the actual status would have been misinterpreted.

From the point of view of the WFD, which gives a priority to biological indicators, the situation is clear: preceding nutrient loadings caused a

57 pressure on the ecosystem resulting in a regime shift when a sudden disturbance (high WL) broke the resilience of the system. The new degraded stable state has demonstrated strong resistance to remediation measures.

However, this interpretation has been put together after a critical comparison of both indices and the initial monitoring data which shows the importance of harnessing the expert analysis for interpreting the monitoring data. More consistent assessment of the ecological status of the lake presumes including other biological elements such as fish, macrophytes and macrozoobenthos in the assessment as suggested by the WFD.

6.1.2 Spatial and temporal variability (II)

Specific features of large shallow Lake Võrtsjärv pose to its monitoring programme certain requirements to meet the principles of adaptive monitoring (Lindenmayer & Likens, 2009). According to these authors, successful and effective monitoring programmes should (1) address well-defined questions that are specified before the commencement of a monitoring programme, i.e. be driven by a human need, which assigns them a management relevance, (2) be underpinned by rigorous statistical design, and (3) be based on a conceptual model of how the components of an ecosystem that are targeted for monitoring might function. Common reasons for monitoring include baseline setting, trend detection, status survey, compliance checking, and impact analysis (Cavanagh et al., 1998), each of them implying a different sampling design and set of target variables.

In addition to defining the monitoring objectives, the location of permanent sampling sites is the next most critical issue in the design of a monitoring programme. For trend and survey monitoring, a station located at the deepest point of the lake is expected to provide the most integrated picture of all changes least affected by local impacts from point sources (Cavanagh et al., 1998; Göransson et al., 2004). Our study showed that the deep sampling site at Võrtsjärv (‘C. Lim.’, Fig. 1, II), characterized by the lowest year-to-year variability of physical, chemical and phytoplankton variables (Table 3, Fig. 2, II) and being representative for more than 90% of the lake aquatory (Fig. 3, II), possesses indeed the necessary qualities required for a permanent surveillance monitoring

58 station. Among the stations of the pelagic area of the lake, the year- to-year variability dominated strongly over the spatial variability, the latter being almost negligible for most of the variables (Fig. 4, II). The revealed homogeneity of the pelagic area of Võrtsjärv was so high (Fig. 5, II) that it may raise the question of redundancy of monitoring the other seven pelagic stations within the area. The highest spatial variability, reaching 14-15% of the total for Ca2+, conductivity, dissolved oxygen, and biochemical oxygen demand, was still marginal compared to the year-to-year variability and could be attributed to the effect of the ‘Sula Kuru’ and ‘Õhne’ stations characterized by a higher submerged macrophyte cover. The narrow southernmost end of the lake has been described in several studies (Nõges et al., 2004; Laas et al., 2012; Idrizaj et al., 2016) as a part clearly differing from the remaining homogeneous part of the lake while the deepest site has been considered generally representative of lake-wide conditions for a number of variables such as phytoplankton abundance and composition (Nõges et al., 2004), bacterioplankton numbers (Tammert & Kisand, 2004), and dissolved oxygen concentrations (Toming et al., 2009).

Ecological status indicators should reflect the effect of the pressure to be assessed, but they must not be affected by natural variability factors (Hering et al., 2006). Selection of right variables to monitor and their interpretation is thus of great importance. The high variability of the common water quality parameters, such as total suspended solids, total phosphorus, chlorophyll a, Secchi depth, and total nitrogen in Võrtsjärv and, especially, their strong dependence on climatic forcing, creates problems for their direct use in water quality and ecological status monitoring. Detrending as one way to overcome the high natural variation of the common water quality parameters for the effect of WL in Võrtsjärv caused a more than ± 40% change of some seasonal mean values, showing the high importance of WL changes in shaping these variables (I). On the other hand, the impact of anthropogenic pressures may be amplified or dampened due to climatic conditions. In our study, surface water temperature and water level explained approximately half of the total variability in parameters commonly used in the ecological status assessment of lakes (Table 4, II), illustrating the need to develop more universal biological metrics for assessing global change impacts on ecosystems.

59 In our data, the variance of phytoplankton parameters was nearly twice as high as that of the chemical and physical parameters (Table 3, II). As biological parameters often include larger methodological errors compared to chemical and physical parameters, it has been common to consider biological parameters too variable to monitor. Indeed, as Karr & Chu (1997) showed, some biological attributes, in particular abundance, density, and production, vary too much even at low levels of human influence, and only a few biological attributes provide reliable signals about biological conditions. In fact, the increase of variance of some biological indices itself can reflect anthropogenic stress owing to the fact that biological systems subjected to high human disturbance are less resilient to environmental change (Fore et al., 1994). Although the plankton variables used in Võrtsjärv monitoring (except the number of species) belong to the criticized above biological attributes, their increased variability near the river mouths showed their high sensitivity to detect changes.

6.2. Contemporary climatic, hydrological and loading trends at shallow lakes Peipsi and Võrtsjärv (III)

Water temperature and ice cover of water bodies are most directly affected by climate forcing. The increased by 0.4 ºC per decade in 1961–2004 (Nõges, 2009b) average annual air temperature (AT) in the region of Lakes Peipsi and Võrtsjärv (Table 1, III) caused significant increase in daily surface water temperature (SWT) in these large and shallow lakes in April and August, by 0.37–0.75 and 0.32–0.42 degrees per decade, respectively (Nõges, 2009b). Also, the average annual precipitation (PR) had significant increasing trends during the same period (Fig. 2C, III). The average annual discharge in the River Väike Emajõgi was significantly positively related to the average annual PR in Tartu (Fig. 2D, III). The average annual water discharge (Q) in this river had a significant increasing trend (Fig. 3A, III). In the River Väike Emajõgi the yearly average concentrations of total nitrogen (TN), total phosphorus (TP) and the TN/TP ratio decreased significantly since 1986 (Fig. 3B, C, D, III). A step change of TN and TN/TP in 1991 (Table 1, III) was caused by a substantial decrease in the application of fertilizers in the catchment after the breakdown of the Soviet Union. Compared to the levels at the end of the 1980s, only 5-10% of N-, P- and K-mineral fertilizers and 30% of the manure were applied to the fields at the end of the 1990s (Järvet et al., 2002). The decrease in TN

60 and TP concentrations continued in the River Väike Emajõgi since 1992 but in the Emajõgi and Narva Rivers as well as in the lakes, the yearly average TN and TP concentrations and the TN/TP ratio did not reveal any significant trend (Table 2;III ). The mean nutrient concentrations in the lakes and their outflows were well coupled.

The riverine loading of nutrients into Lakes Peipsi and Võrtsjärv increased drastically in the 1980s while in the early 1990s a sharp decrease occurred. Due to the faster decrease of TN loading than TP loading, the TN/TP ratio in the loadings decreased (Nõges et al., 2003a, 2005, 2007b). Multiple regression analysis showed that the rate of fertilization was the most important factor determining the nitrogen runoff from the catchment of Lake Võrtsjärv (Järvet et al., 2002). Since the end of the 1980s, TP concentrations in the main inflow of Lake Võrtsjärv and TP loadings showed a decreasing trend (Table 1, Fig. 4B, III) while the loadings of TN increased in the 2000s (Fig. 4A, III) resulting in an increased of the TN/TP loading ratio (Fig. 4C, III).

In Peipsi, the total loading of TP in 2001 was almost equal to the high loadings at the beginning of the 1980s and TP loadings from Russia in mid-1990s even exceeded those of the 1980s. In the last years, TP concentration started to decrease in Peipsi, indicating some improvement of the situation (Blank et al., 2017). This could partly be explained by smaller loads observed in dry years (Nõges P. et al., 2007). In Võrtsjärv the dynamics of both N and P concentrations (Fig. 5G, H, III) were consistent to the loading history (Fig. 4A, B, III), showing the high sensitivity of large shallow lakes to changes in human activity in the catchment.

Eutrophication of Lake Peipsi escalated in the 1970s–1980s (Heinsalu et al., 2007) and the same was found for Lake Võrtsjärv (Nõges P. & Järvalt, 2004). In the first half of the 20th century, cyanobacterial blooms, a common phenomenon in both lakes, were caused predominantly by mesoeutrophic species: Gloeotrichia echinulata in Peipsi and Anabaena lemmermannii in Võrtsjärv (Mühlen and Schneider, 1920; Laugaste et al., 2001). Blooms ceased in the 1980s due to heavy nitrogen loading and intensified again in Peipsi in late 1990s. Reappearing cyanobacterial blooms in Peipsi caused serious summer fish-kills (Kangur et al., 2005) which seem to be a straightforward consequence of reduced nitrogen

61 levels at remaining high phosphorus levels and, thus, the changed N/P ratio (Nõges et al., 2007b).

In Võrtsjärv where recurrent winter fish-kills have been observed, the climatic factors affecting the water level are more crucial (Nõges et al.,

2003b). In Peipsi, the N2-fixing Gloeotrichia echinulata, Aphanizomenon flos-aquae and Anabaena species prevail in summer phytoplankton. In Võrtsjärv the dominant cyanobacteria Limnothrix planktonica, L. redekei and Planktolyngbya limnetica are not capable to fix N2, and the main

N2-fixing taxa Aphanizomenon skujae and Anabaena spp. commonly do not dominate. Also, the TN/TP mass ratio in Võrtsjärv is about two times higher than in Peipsi (Fig. 5C, F, I, Table 2, III). The critical TN/

TP mass ratio in May-October, below which the N2-fixing cyanobacteria achieved high biomasses, was around 40 in Võrtsjärv and around 30 in Peipsi (Nõges et al., 2008). In Võrtsjärv the TN/TP ratio is above this critical value both in the inflows and in the lake while in Peipsi it often drops below this threshold.

Recently the TN/TP ratio has shown an increasing tendency in both lakes (Fig. 5C, F, I, III). Increased TN loading (Fig. 4A, III) and in-lake concentrations (Fig. 5A, D, G, III), and decreased or stabilised TP loading (Fig. 4B, III) and in-lake concentrations (Fig. 5B, E, H, III), suggest that a reduction of water blooms could be expected in Lake Peipsi. Indeed, since the last cyanobacterial bloom in 2010, no severe bloom was recorded in the lake. However, both lakes are large and shallow and largely controlled by climatic factors, either directly by increased water temperature or indirectly through the fluctuations of non-regulated water level (Nõges et al., 2003b). Temperature dependence of cyanobacteria development and N2 fixation should be taken into account in the context of global warming, as higher water temperatures support both cyanobacterial growth (Jöhnk et al., 2008; Paerl & Huisman, 2008) and P recycling from sediments (Genkai-Kato & Carpenter, 2005). According to Nõges et al. (2008), the temperature dependence of N2-fixers in Lake Peipsi is the main factor determining the increasing potential share of

N2-fixing species in the phytoplankton community along with increasing water temperature.

62 6.3. Underwater light field and its determinants in shallow lakes (IV, V, VI)

The underwater light field greatly determines the ecological conditions in water: primary production (Krause-Jensen & Sand-Jensen, 1998), species composition of the phytoplankton communities (Schanz, 1985), depth distribution of submerged macrophytes (Duarte, 1991), and heat budget of the water body (Fedorov & Ginsburg, 1992) (VI).

A vertical decrease of irradiance in the water, characterized by the diffuse attenuation coefficient (Kd), depends on surface and illumination conditions and on concentrations of optically active substances (OAS) in water (Dera, 1992). This parameter is widely used in practice to characterize the light field and propagation of photosynthetically active radiation (PAR, spectral range 400-700 nm) inside water (Kirk, 1994; Mobley, 1994). The water in Lake Peipsi is rich in all OAS: phytoplankton, suspended mineral particles and dissolved organic matter (DOM). Its chromophore-containing compounds are called as coloured dissolved organic matter (Hoge et al., 1993; Kallio, 1999) or yellow substance – CDOM – contribute significantly to the total light absorption in the water and is one of the optically active substances in water competing with phytoplankton and other aquatic plants for the capture of available light energy (IV, VI). CDOM refers to any type of DOM, regardless of its origin, which has an optical effect (Bricaud et al., 1981) and can be measured by optical methods (Lindell et al., 1999).

Statistical analysis using the spectral values of Kd and concentrations of OAS from Estonian and Finnish small lakes showed that yellow substance was the main factor influencing the light attenuation in the violet and blue parts of the spectrum while in the orange and red parts of the Kd spectra, suspended particles containing chlorophyll a had the major influence VI( ). Higher values for Secchi depth corresponded to lower attenuation. Yellow substance is the most powerful light absorbing component in many Estonian (Arst et al., 1997; Reinart et a1., 2003) as well as in Finnish and Swedish lakes (VI). In Lake Peipsi, 34-80% of the attenuation of daylight in the blue region is caused by the yellow substance. Large variation of OAS all over Peipsi causes also large variability in water transparency and shows that CDOM cannot be considered similarly over the whole aquatory of Peipsi. In Swedish large lakes Vättern and Vänern the influence of CDOM in the blue region of

63 the spectrum was higher (~90% and 86%, respectively, whereat variation of the CDOM effect was less than in L. Peipsi) (VI). The concentration and properties of allochthonous CDOM brought into the lake by rivers and from coastal soils depend very much on the annual water balance. Since seasonal fluctuations of the water level of Peipsi are more than 1.5 m, the inflow to the lake being usually 5-6 times higher in April and May than in June, and very low around midsummer (Jaani & Kullus, 2001), there appear large differences in CDOM concentrations as well. According to the state monitoring data, in the open area of Peipsi the highest values of CDOM were measured in June, but in the river mouth in November (in 2001) or August (in 2002). Also, our results showed large spatial and temporal variability caused by the complex effect of DOM on the light climate in Peipsi (VI).

Light provides the energy necessary for the transformation of inorganic matter into organic matter by the planktonic algae and all other photoautotrophic plants. Because of changing light conditions, primary production has a pronounced diel pattern, and to acquire integrated results over longer time periods (days, months, years), many consecutive measurements of the instantaneous rate of photosynthesis should be done and integrated. Generally, in situ measurements (Joniak et al., 2003; Yoshida et al., 2003; Forget et al., 2007) give reliable results only in clear waters, while in highly productive waters incubation cannot be performed during a long period (e.g. from morning to evening) as part of the 14C-label gets lost from the cells during long-term incubation due to respiration of photosynthetic products (Lancelot & Mathot, 1986) and release of extracellular products (Møller Jensen, 1985). For reliable description of primary production profiles, it is reasonable to use a ‘spectral approach’, in which the model is based on spectral data of underwater quantum irradiance and absorption coefficients of phytoplankton (Sathyendranath et al., 1989; Smith et al., 1989; Schofield et al., 1990; Kyewalyanga et al., 1992; Kirk, 1994; Sosik, 1996; Arst et al., 2006, 2008). In many studies the daily/monthly sums of primary production and their seasonal variation was estimated according to data acquired form episodic measurements in marine and lake environments (Fee, 1980; Tillmann et al., 2000; Joniak et al., 2003; Yoshida et al., 2003; Davies et al., 2004; Nõges & Kangro, 2005; Demidov, 2008). Robson (2005) studied the effects of diurnal variations in light on primary production at a seasonal time-scale. The variation of phytoplankton production is associated mainly with an availability of nutrients (Schindler,

64 1978), chlorophyll concentration (Tillmann et al., 2000; Demidov, 2008), and light limitation (Kyewalyanga et al., 1992; Tillmann et al., 2000). The most general outcome from these studies – that incoming irradiance, chlorophyll content, and attenuation coefficient of light in the water are the main factors forming the profiles and diurnal sums of the primary production – was confirmed by our studies in turbid-water lakes Peipsi, Võrtsjärv and Harku, showing satisfactory concordance between the measured and modelled production profiles (Fig. 2 and 3, V). A strong dependence of irradiance and primary production profiles on cloudiness was demonstrated, causing their irregular diurnal variation, sometimes with several highs and lows. The possibility of calculating the diurnal sums of integral primary production was demonstrated for all three lakes.

Dissolved organic carbon (DOC) as a part of DOM is a crucial component in lakes, determining their role in the global carbon cycle. A considerable portion of the total DOC pool in large eutrophic lakes with long residence time like Võrtsjärv is originating from autochthonous production (Toming et al., 2013). The concentrations of DOC and CDOM in Lake Võrtsjärv were almost two to three times higher than in other eutrophic lakes (Table 5, IV). The overall DOC concentration in Võrtsjärv is apparently regulated by an additional stabilizing mechanism of the specific seasonal pattern where the lowest level of CDOM coincides with the highest level of phytoplankton-derived DOC. As the proportions of allochthonous and autochthonous material in DOC of Lake Võrtsjärv vary seasonally and inter-annually in large scale (Toming et al., 2013), the nature of the relationships between DOC and other variables also vary in different seasons and years. During the vegetation period, DOC concentration was strongly related to phytoplankton development and variables such as chlorophyll a, total suspended matter, Secchi depth, water pH, inflowing riverine discharges and precipitation. The three first named variables can be considered as the proxies for remote sensing of DOC seasonality in Lake Võrtsjärv. The proxies for estimating DOC concentrations in long-term studies should reflect both the variability of phytoplankton and CDOM, therefore it is better to use transparency-related variables (e.g. Secchi depth).

CDOM which is usually strongly correlated with DOC in humic lakes and fluctuating synchronously (Tranvik, 1990; Molot & Dillon, 1997; Kallio, 1999; Yacobi et al., 2003; Zhang et al., 2007; Erlandsson et al.,

65 2012; Faithfull et al., 2015), did not appear to be a good predictor of the seasonality of DOC concentration in mostly non-humic Lake Võrtsjärv where seasonal variation was in the CDOM – DOC coupling. Also Brezonik et al. (2015) concluded that in optically complex waters the coloured fraction of DOC varies widely and the prediction of DOC concentrations from CDOM levels alone is subject to a substantial uncertainty.

In the analysis among different eutrophic lakes in the world (Table 5, IV), CDOM appeared to be a very good predictor for DOC concentration (Fig. 5a) and thus the remote sensing of CDOM can be applied as a proxy to estimate large-scale patterns of DOC concentrations in eutrophic lakes. As Pace and Cole (2002) concluded, the among-lake variation of DOC and CDOM primarily reflects differences in watershed size and structure that are connected to loading, lake size and shape that influence in-lake residence time and degradation. At the within-lake scale, watershed and lake structure are fixed and the principal source of variation is the influence of climatic processes such as ice-out timing, precipitation, and drought, which in turn affect loading, washout, and in-lake production and degradation.

66 7. CONCLUSIONS

1. Our analysis corroborated hypothesis 1 showing that the effects of seasonality and naturally changing water level in Võrtsjärv on traditional water quality indicators such as concentrations of nutrients and chlorophyll a, phytoplankton biomass, and Secchi depth overshadow the effects of anthropogenic stress on these indicators. Correcting of the metric values used in status assessment for the effects of natural variability factors is a necessary step in order to increase the signal/noise ratio and decrease the uncertainty of the estimate for strongly, physically driven systems such as shallow lakes where the large variation of climate-dependent driving factors, may mask the effect of human pressures (I).

2. In Võrtsjärv, surface water temperature and water level explained approximately half of the total variability in parameters commonly used in ecological status assessment of lakes. The selection of variables monitored in Võrtsjärv is excellent for climate change impact research as many of them show strong relationships with changes in water level and temperature, which can be considered good proxies for precipitation and thermal variability for this lake (II).

3. Based on chemical, physical and phytoplankton data from 10 monitoring stations, Võrtsjärv can be divided into two distinct parts: the narrow southern-most part under the influence of the main inflowing River Väike Emajõgi represented by two stations and forming only 2% of the lake’s total area, and the homogeneous pelagic area of the lake represented by eight stations. Within the main pelagic area, the inter-annual variability markedly exceeded the spatial variability for most variables monitored, the latter being almost negligible for most of the variables (II). This corroborates our hypothesis 2.

4. The permanent sampling station at the deepest site in Võrtsjärv, characterized by the lowest average variability of the parameters measured and being representative for more than 90% of the lake aquatory, possesses the necessary qualities required of a permanent surveillance monitoring station. This corroborates our hypothesis 3. The location of the station is excellent both regarding logistics

67 (close to the lab) and its statistical properties (smallest variability of measured parameters, good representativeness for the pelagic part of the lake, unaffectedness from local factors or inflows) II( ).

5. Large shallow lakes Peipsi and Võrtsjärv are largely controlled by climatic factors; either directly by increased water temperature or indirectly through the fluctuations of non-regulated water levels. Nutrient dynamics in the lakes follow the changes in loadings, showing the ability of shallow lake ecosystems to react sensitively to changes in catchment management (III). These conclusions corroborate our hypothesis 4.

6. Our analysis rejected hypothesis 5, showing that CDOM is not a good predictor of the seasonality of DOC concentration in eutrophic Lake Võrtsjärv since the CDOM – DOC coupling varies seasonally. Still, it is a powerful predictor to be applied for remote sensing to assess large-scale patterns of DOC concentrations in eutrophic lakes (IV).

7. Bio-optical model calculations provide an alternative to the time consuming 14C method for primary production calculations. A model using daily variation of incoming irradiance in the PAR region and episodic measurements of chlorophyll a concentration and diffuse attenuation coefficient in the water as input data, is suitable for estimating variation of primary production in lakes (V).

8. In Lake Peipsi, large spatial and temporal variation of optically active substances causes large variability of light absorption in different spectral regions. The concentration and properties of allochthonous CDOM depend on large seasonal changes of riverine transport and water level in the lake (VI).

Based on these findings, further upgrade of the evaluation tools for lake monitoring is needed.

68 REFERENCES

Adrian, R., Hessen, D. O., Blenckner, T., Hillebrand, H., Hilt, S., Jeppesen, E., Livingstone, D. M. & Trolle, D. 2016. Environmental Impacts—Lake Ecosystems. In: M. Quante & F. Colijn (Eds), North Sea Region Climate Change Assessment, Regional Climate Studies. doi: 10.1007/978-3-319-39745-0_10. Adrian, R., Walz, N., Hintze, T., Hoeg, S. & Rusche, R. 1999. Effects of ice duration on the plankton succession during spring in a shallow polymictic lake. Freshwater Biology, 41: 621–623. Adrian, R., Wilhelm, S. & Gerten, D. 2006. Life-history traits of lake plankton species may govern their phenological response to climate warming. Global Change Biology, 12: 652–661. doi: 10.1111/j.1365- 2486.2006.01125.x. Agasild, H., Zingel, P., Tuvikene, L., Tuvikene, A., Timm, H., Feldmann, T., Salujõe, J., Toming, K., Jones, R.I. & Nõges, T. 2014. Biogenic methane contributes to the food web of a large, shallow lake. Freshwater Biology, 59(2), 272−285. doi: 10.1111/fwb.12263. Alahuhta, J., Vuori, K.-M., Hellsten, S., Järvinen, M., Olin, M., Rask, M. & Palomäki, A. 2009. Defining the ecological status of small forest lakes using multiple biological quality elements and paleolimnological analysis. Fundamental and Applied Limnology, 175(3): 203–216. Anneville, O., Gammeter, S. & Straile, D. 2005. Phosphorus decrease and climate variability: mediators of synchrony in phytoplankton changes among European peri-alpine lakes. Freshwater Biology, 50: 1731–1745. Anneville, O., Soussi, S., Gammeter, S. & Straile, D. 2004. Seasonal and inter-annual scales of variability in phytoplankton assemblages: comparison of phytoplankton dynamics in three peri-alpine lakes over a period of 28 years. Freshwater Biology, 49: 98–115. Anneville, O., Soussi, S., Ibañez, F., Ginot, V., Druart, J.-C. & Angeli, N. 2002. Temporal mapping of phytoplankton assemblages in Lake Geneva: annual and inter-annual changes in their patterns of succession. Limnology and Oceanography, 47: 1355–1366.

69 Anttila, S. 2013. Applicability of characterized variance and ecosystem interactions in water quality monitoring. Doctoral dissertation. University of Helsinki, 35 pp. Anttila, S. & Kairesalo, T. 2010. Mean and variance estimations with different pixel sizes: case study in a small water quality monitoring area in southern Finland. Boreal Environment Research, 15(3): 335– 346. Anttila, S., Kairesalo, T. & Pellikka, P. 2008. A feasible method to assess inaccuracy caused by patchiness in water quality monitoring. Environmental Monitoring Assessment, 142: 11–22. doi 10.1007/ s10661-007-9904-y. Anttila, S., Ketola, M., Vakkilainen, K. & Kairesalo, T. 2012. Assessing temporal representativeness of water quality monitoring data. Journal of Environmental Monitoring, 14: 589–595. Anttila, S., Ketola, M., Kuoppamäki, K., & Kairesalo, T. 2013. Identification of biomanipulation-driven regime shift in Lake Vesijärvi: implications for lake management. Freshwater Biology, 58(7): 1494–1502. Aránguiz-Acuña, A. & Serra, M. 2016. Diapause as escape strategy to exposure to toxicants: response of Brachionus calyciforus to arsenic. Ecotoxicology, 25: 708–719. doi 10.1007/s10646-016-1629-7. Arhonditsis, G. B., Brett, M. T., DeGasperi, C. L. & Schindler, D. E. 2004. Effects of climatic variability on the thermal properties of Lake Washington. Limnology and Oceanography, 49: 256–270. Arst, H., Erm, A., Kangro, K., Nõges, T. & Nõges, P. 2006. Comparison of spectral and broad-band models for computing photosynthetically absorbed radiation in turbid waters. Boreal Environment Research, 11: 55–65. Arst, H., Mäekivi, S., Lukk, T. & Herlevi, A. 1997. Calculating irradiance penetration into water bodies from the measured beam attenuation coefficient. Limnology and Oceanography, 42(20): 379–385. Arst, H., Nõges, T., Nõges, P. & Paavel, B. 2008. In situ measurements and model calculations of primary production in turbid waters. Aquatic Biology, 3: 19–30.

70 Arvola, L., Räike, A., Kortelainen, P. & Järvinen, M. 2004. The effect of climate and landuse on TOC concentrations and loads in Finnish rivers. Boreal Environment Research, 9: 381–387. Assel, R. & Robertson, D. M. 1995. Changes in winter air temperatures near Lake Michigan, 1851–1993, as determined from regional lake- ice records. Limnology and Oceanography, 40: 165–176. BACC Author Group 2008. Assessment of climate change for the Baltic Sea basin. Springer, Berlin. Baho, D.L., Futter, M.N., Johnson, R.K. & Angeler, D.G. 2015. Assessing temporal scales and patterns in time series: Comparing methods based on redundancy analysis. Ecological Complexity, 22: 162–168. Baker, I. T., Denning, A. S. & Stöckli, R. 2010. North American gross primary productivity: regional characterization and inter-annual variability. Tellus, 62B: 533–549. Bannister, T. T. 1974. Production equations in terms of chlorophyll concentration, quantum yield, and upper limit to production. Limnology and Oceanography, 19: 1–12. Bastviken, D., J. Cole, M. Pace & Tranvik, L. 2004. Methane emissions from lakes: Dependence of lake characteristics, two regional assessments, and a global estimate. Global Biogeochemical Cycles, 18, GB4009, doi:10.1029/2004GB002238. Bastviken D., Tranvik, L. J., Downing, J.A., Crill, P. M. & Enrich-Prast, A. 2011. Freshwater methane emissions offset the continental carbon sink. Science, 331: 50–50. Battarbee, R. W., Anderson, N. J., Bennion, H. & Simpson, G. L. 2012. Combining limnological and palaeolimnological data to disentangle the effects of nutrient pollution and climate change on lake ecosystems: problems and potential. Freshwater Biology, 57: 2091–2106. Bauer, O. N. 1991. Spread of parasites and diseases of aquatic organisms by acclimatization: a short review. Journal of Fish Biology, 39(5): 679–686. Blottière, L., Jaffar-Bandjee, M., Jacquet, S. Millot, A & Hulot, F. D. 2017. Effects of mixing on the pelagic food web in shallow lakes. Freshwater Biology, 62: 161–177. Bolpagni, R., Bresciani, M., Laini, A., Pinardi, M., Matta, E., Ampe, E.M., Giardino, C., Viaroli, P. & Bartoli, M. 2014. Remote sensing of

71 phytoplankton–macrophyte coexistence in shallow hypereutrophic fluvial lakes. Hydrobiologia, 737: 67–76. Borja, Á. & Rodríguez, J. G. 2010. Problems associated with the ‘one- out, all-out’ principle, when using multiple ecosystem components in assessing the ecological status of marine waters. Marine Pollution Bulletin, 60: 1143–1146. Brezonik, P. L., Olmanson, L. G., Finlay, J. C. & Bauer, M. E. 2015. Factors affecting the measurement of CDOM by remote sensing of optically complex inland waters. Remote Sensing of Environment, 157: 199–215. Bricaud, A., Morel, A. & Prieur, L. 198l. Absorption by dissolved organic matter in the sea (yellow substance) in the UV and visible domain. Limnology and Oceanography, 26: 43–53. Brucet, S., Poikane, S., Lyche-Solheim & Birk, S. 2013. Biological assessment of European lakes: ecological rationale and human impacts. Freshwater Biology, 58: 1106–1115. Buffam, I., Turner, M. G., Desai, A. R., Hanson, P. C., Rusak, J. A., Lottig, N. R., Stanley, E. H. & Carpenter, S. R. 2011. Integrating aquatic and terrestrial components to construct a complete carbon budget for a north temperate lake district. Global Change Biology, 17: 1193–1211. doi:10.1111/j.1365-2486.2010.02313.x. Burkholder, J. M. 2000. Eutrophication and oligotrophication. In: S. Levin (Ed), Encyclopedia of Biodiversity, Volume 2. Academic Press, New York. 649–670. Cabecinha, E., Van den Brink, P. J., Cabral, J. A., Cortes, R., Lourenço, M. & Pardal, M. Â. 2009. Ecological relationships between phytoplankton communities and different spatial scales in European reservoirs: implications at catchment level monitoring programmes. Hydrobiologia, 628: 27–45. doi 10.1007/s10750-009-9731-y. Caroni, R., van de Bund, W., Clarke, R. T. & Johnson, R. K. 2013. Combination of multiple biological quality elements into waterbody assessment of surface waters. Hydrobiologia, 704: 437–451. doi 10.1007/s10750-012-1274-y. Carpenter, S., & Lathrop, R. 2014. Phosphorus loading, transport and concentrations in a lake chain: a probabilistic model to compare management options. Aquatic Sciences, 76: 145–154.

72 Carstensen, J. 2007. Statistical principles for ecological status classification of Water Framework Directive monitoring data. Marine Pollution Bulletin, 55: 3–15. Carstensen, J. & Henriksen, P. 2009. Phytoplankton biomass response to nitrogen inputs: a method for WFD boundary setting applied to Danish coastal waters. Hydrobiologia, 633: 137–149. Cavalcanti, J. R., da Motta-Marques, D. & Fragoso C. R. Jr. 2016. Process-based modeling of shallow lake metabolism: Spatio-temporal variability and relative importance of individual processes. Ecological Modelling, 323: 28–40. Cavanagh, N., Nordin, R. N., Pommen, L. W. & Swain, L. G. 1998. Guidelines for Designing and Implementing a Water Quality Monitoring Program in British Columbia. Ministry of Environment, Lands and Parks. Government Publication No. 7680000554. http:// www.ilmb.gov.bc.ca/risc/pubs/aquatic/ Clarke, R. T. 2013. Estimating confidence of European WFD ecological status class and WISER Bioassessment Uncertainty Guidance Software (WISERBUGS). Hydrobiologia, 704: 39–56. Cleveland, W. P. & Tiao, G. C. 1976. Decomposition of seasonal time series: a model for the Census X-11 program. Journal of the American Statistical Association, 71: 581–587. Cohen, J., Cohen, P., West, S. G. & Aiken L. S. 2003. Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences. 3rd ed. Lawrence Erlbaum Associates, Inc., Publishers, 691 pp. Coats, R., Perez-Losada, J., Schladow, G., Richards, R. & Goldman, C., 2006. The warming of Lake Tahoe. Climatic Change, 76: 121–148. Cole, J.J., Caraco, N.F., Kling, G.W. & Kratz, T.K. 1994. Carbon dioxide supersaturation in the surface waters of lakes. Science, 265: 1568– 1570. Cole, J. J., Carpenter, S. R., Kitchell, J., Pace, M. L., Solomon, C. T. & Weidel, B. 2011. Strong evidence for terrestrial support of zooplankton in small lakes based on stable isotopes of carbon, nitrogen, and hydrogen. Proceedings of the National Academy of Sciences. USA, 108(5): 1975–1980.

73 Cooke, G. D., Welch, E. B., Peterson, S. A. & Nichols, S. A. (Eds) 2005. Restoration and management of lakes and reservoirs. 3rd ed. CRC Press Taylor & Francis Group. 591 pp. Croley, T. E., Quinn F. H., Kunkel K. E. & Changnon, S. A. 1998. Great lakes hydrology under transposed climates. Climatic Change, 38: 405–433. Davies, J. J. L., Jenkins, A., Monteith, D. T., Evans, C. D. & Cooper, D. M. 2005. Trends in surface water chemistry of acidified UK Freshwaters, 1988–2002. Environmental Pollution, 137: 27–39. Davies, J.-M., Nowlin, W. H. & Mazumder, A. 2004. Variation in temporal [14C] plankton photosynthesis among warm monomictic lakes of coastal British Columbia. Journal of Plankton Research, 26(7): 763–778. Demidov, A. B. 2008. Seasonal dynamics and estimation of the annual primary production of phytoplankton in the Black Sea. Oceanology, 48: 664–678. Dera, J. 1992. Marine Physics. PWN, Warszawa and Elsevier, Amsterdam, 516 pp. Derry, A. M., Arnott, S. E. & Boag, P. T. 2010. Evolutionary shifts in copepod acid tolerance in an acid-recovering lake indicated by resurrected resting eggs. Evolutionary Ecology, 24(1): 133–145. de Wilde, A. J., Knoben, R. A. E. & van Poppel, J. W. P. M. 2002. Setting class boundaries for the classification of rivers and lakes in Europe. Final Report Project number 39534, Client Ministerie van Verkeer en Waterstaat. DG RWS—RIZA, 1–25. Directive, 2000. Directive 2000/60/EC of the European Parliament and of the council of 23 October 2000 establishing a framework for community action in the field of water policy. Official Journal of the European Communities L327: 1–72. Doganovsky, A. M. & Myakisheva, N. V. 2000. The water level in Lake Ladoga in different frequency ranges. In: A. Peltonen, E. Grönlund & Viljanen, M. (Eds), Proceedings of the Third International Lake Ladoga Symposium 1999, University of Joensuu, Publications of Karelian Institute No. 129, Joensuu, Finland, pp. 374–377. Dokulil, M. T, Jagsch, A., George, G. D., Anneville, O., Jankowski, T., Wahl, B., Lenhart, B., Blenckner, T. & Teubner, K. 2006. Twenty

74 years of spatially coherent deepwater warming in lakes across Europe related to the North Atlantic Oscillation. Limnology and Oceanography, 51(6): 2787–2793. Dokulil, M. T. & Teubner, K. 2005. Do phytoplankton communities correctly track trophic changes? An assessment using directly measured and palaeolimnological data. Freshwater Biology, 50: 1589–1593. Duarte, C. M. 1991. Seagrass depth limits. Aquatic Botany, 40: 363–377. Duarte, C. M., Conley, D. J., Carstensen J. & Sánchez-Camacho, M. 2009. Return to Neverland: Shifting Baselines Affect Eutrophication Restoration Targets. Estuaries and Coasts, 32: 29–36. doi 10.1007/ S12237-008-9111-2. Ehapalu, K., Tõnno, I., Reitalu, T., Alliksaar, T., Liiv, M., Freiberg, R. & Nõges, P. 2017. Sedimentary carbon forms in relation to climate and phytoplankton biomass in a large, shallow, hard-water boreal lake. Journal of Paleolimnology, 57: 81–93. doi 10.1007/s10933-016- 9931-1. Eisenreich, S. (Ed) 2005. Climate Change and the European Water Dimension. European Commission Joint Research Centre, EUR 21553, Ispra, Italy, 253 pp. Elith, J., Leathwick, J. R. & Hastie, T. 2008. A working guide to boosted regression trees. Journal of Animal Ecology, 77: 802–813. doi: 10.1111/j.1365-2656.2008.01390.x. Elliott, J. M., Hurley, M. H. & Maberly, S. C. 2000. The emergence period of sea-trout fry in a Lake District stream correlates with the North Atlantic Oscillation. Journal of Fish Biology, 56: 208–210. Elliot, J. A., Jones, I. D. & Thackeray, S. J. 2006. Testing the sensitivity of phytoplankton communities to changes in water temperature and nutrient load, in a temperate lake. Hydrobiologia 559: 401–411. doi: 10.1007/s10750-005-1233-y. Ellis, J. & Adriaenssens, V. 2006. Uncertainty estimation for monitoring results by the WFD biological classification tools. Bristol, Environment Agency, 26 pp. Engstrom, D. R., Fritz, S. C., Almendinger, J. E. & Juggins, S. 2000. Chemical and biological trends during lake evolution in recently deglaciated terrain. Nature, 408: 161–166.

75 EPA. 1997. Guidelines for Preparation of the Comprehensive State Water Quality Assessments (305(b) Reports) and Electronic Updates: Supplement. EPA-841-B-97-002B. U.S. Environmental Protection Agency, Washington, DC. Erlandsson, M., Futter, M. N., Kothawala, D. N. & Köhler, S. J. 2012. Variability in spectral absorbance metrics across boreal lake waters. Journal of Environmental Monitoring, 14: 2643–2652. Evans, C. D., Monteith, D. T. & Cooper, D. M. 2005. Long-term increases in surface water dissolved organic carbon: Observations, possible causes and environmental impacts. Environmental Pollution, 137(1): 55–71. Faithfull, C. L., Mathisen, P., Wenzel, A., Bergström, A. K. & Vrede, T. 2015. Food web efficiency differs between humic and clear water lake communities in response to nutrients and light. Oecologia, 177: 823–835. doi 10.1007/s00442-014-3132-2. Fedorov, K. N. & Ginsburg. A. I. 1992. The Near-Surface Layer of the Ocean. VSP. Utrecht, The Netherlands. 259 pp. Fee, E. J. 1980. Important factors for estimating annual phytoplankton production in the Experimental Lakes Area. Canadian Journal of Fisheries and Aquatic Sciences, 37: 513–522. Fergus, C. E., Finley, A. O., Soranno, P. A. & Wagner, T. 2016. Spatial variation in nutrient and water color effects on lake chlorophyll at macroscales. PLoS ONE, 11(10): e0164592. doi:10.1371/journal. pone.0164592. Fore, L. S., Karr, R. & Conquest, L. L. 1994. Statistical properties of an index of biotic integrity used to evaluate water resources. Canadian Journal of Fisheries and Aquatic Sciences, 51: 1077–1087. Forget, M.-H., Sathyendranath, S., Platt, T., Pommier, J., Vis, C., Kyewalyanga, M. & Hudson, C. 2007. Extraction of photosynthesis- irradiance parameters from phytoplankton production data: demonstration in various aquatic systems. Journal of Plankton Research, 29(3): 249–262. Frisk, T., Bilaletdin, Ä., Kaipainen, H., Malve, O., & Möls, M. 1999. Modelling phytoplankton dynamics of the eutrophic Lake Võrtsjärv, Estonia. Hydrobiologia, 414: 59–69.

76 Fussmann, G. F., Loreau, M. & Abrams, P. A. 2007. Eco-evolutionary dynamics of communities and ecosystems. Functional Ecology, 21: 465–477. doi:10.1111/j.1365-2435.2007.01275.x Gardner, E. S. Jr. 1985. Exponential smoothing: the state of the art. Journal of Forecasting 4: 1–28. Genkai-Kato, M. & Carpenter, S. R. 2005. Eutrophication due to phosphorus recycling in relation to lake morphometry, temperature, and macrophytes. Ecology, 86: 210–219. George, D. G. 2000. Using climate indicators to monitor patterns of change in freshwater and reservoirs, In: J. A. A. Jones, K. Gilman, A. Jigorel & Griffin, J. H. (Eds), Water in the Celtic world: Managing Resources for the 21st Century, British Hydrological Society, Occasional Paper 11: 93–102. George, D. G., Järvinen, M. & Arvola, L. 2004b. The influence of the North Atlantic Oscillation on the winter characteristics of Windermere (UK) and Pääjärvi (Finland). Boreal Environment Research, 9: 389–399. George, D. & Edwards, R. 1976. The effect of wind on the distribution of chlorophyll a and crustacean plankton in a shallow eutrophic reservoir. Journal of Applied Ecology, 13(3): 667–690. George, D., G., Hewitt, D. P., Lund, J. W. G. & Smyly, W. J. P. 1990. The relative effects of enrichment and climate change on the long- term dynamics of Daphnia in Esthwaite Water, Cumbria. Freshwater Biology, 23, 55–70. George, D. G., Maberly, S. C. & Hewitt, D. P. 2004a. The influence of the North Atlantic Oscillation on the physical, chemical and biological characteristics of four lakes in the English Lake District. Freshwater Biology, 49: 760–774. George, D. G., Talling, J. F. & Rigg, E. 2000. Factors influencing the temporal coherence of five lakes in the English Lake District. Freshwater Biology, 43: 449–461. Gerten, D. & Adrian, R. 2000. Climate-driven changes in spring plankton dynamics and the sensitivity of shallow polymictic lakes to the North Atlantic Oscillation. Limnology and Oceanography, 45: 1058–1066.

77 Gillooly, J. F. & Dodson, S. I. 2000. Latitudinal patterns in the size distribution and seasonal dynamics of new world freshwater cladocerans. Limnology and Oceanography, 45: 22–30. Gitelson, A. A., Dall’Olmo, G., Moses, W., Rundquist, D. C., Barrow, T., Fisher, T. R., Gurlin, D. & Holz, J. 2008. A simple semi-analytical model for remote estimation of chlorophyll in turbid waters: validation. Remote Sensing of Environment, 112: 3582–3593. Granéli, W. 2012. Brownification of Lakes. In: Bengtsson, L., Herschy, R. W. & Fairbridge, R. W. (Eds) Encyclopedia of Lakes and Reservoirs. doi:10.1007/978-1-4020-4410-6, Springer Science and Business Media B.V. Gsell, A. S., Scharfenberger, U., Özkundakci D., Walters, A., Hansson, L.-A., Janssen, A. B. G., Nõges, P., Reid, P. C., Schindler, D. E., Van Donk, E., Dakos, V. & Adrian, R. 2016. Evaluating early-warning indicators of critical transitions in natural aquatic ecosystems. PNAS PLUS, www.pnas.org/cgi/doi/10.1073/pnas.1608242113 Göransson, E., Johnson, R. K. & Wilander, A. 2004. Representativity of a mid-lake surface water chemistry sample. Environmental Monitoring and Assessment, 95: 221–238. Haldna, M., Milius, A., Laugaste, R. & Kangur, K. 2008. Nutrients and phytoplankton in Lake Peipsi during two periods that differed in water level and temperature. Hydrobiologia 599: 3–11. doi:10.1007/ s10750-007-9208-9. Haldna, M., Möls, T., Buhvestova, O. & Kangur, K. 2013. Predictive model for phosphorus in large shallow Lake Peipsi: Approach based on covariance structures. Aquatic Ecosystem Health & Management, 16(2): 222–226. doi:10.1080/14634988.2013.789766. Hampton, S. E., Gray, D. K., Izmest’eva, L. R., Moore, M. V. & Ozersky, T. 2014. The Rise and Fall of Plankton: Long-Term Changes in the Vertical Distribution of Algae and Grazers in Lake Baikal, Siberia. PLoS ONE, 9(2): e88920. doi:10.1371/journal.pone.0088920. Hari, R. E., Livingstone, D. M., Siber, R., Burkhardt-Holm, P. & Güttinger, H. 2006. Consequences of climatic change for water temperature and brown trout populations in Alpine rivers and streams. Global Change Biology, 12: 10–26. Harzhauser, M., Kern, A., Soliman, A., Minati, K., Piller, W. E., Danielopole, D. E. & Zuschin, M. 2008. Centennial- to decadal scale

78 environmental shifts in and around Lake Pannon (Vienna Basin) related to a major Late Miocene lake level rise. Palaeogeography, Palaeoclimatology, Palaeoecology, 270: 102–115. Hatvani, I. G., Kovács, J., Kovács, I. S., Jakusch, P. & Korponai, J. 2011. Analysis of long-term water quality changes in the Kis-Balaton Water Protection System with time series-, cluster analysis and Wilks’ lambda distribution. Ecological Engineering, 37: 629–635. Hedström, P., Bystedt, D., Karlsson, J., Bokma, F. & Byström, P. 2017. Brownification increases winter mortality in fish. Oecologia, 183: 587–595. Heinsalu, A., Alliksaar, T., Leeben, A. & Nõges, T. 2007. Sediment diatom assemblages and composition of pore-water dissolved organic matter reflect recent eutrophication history of Lake Peipsi (Estonia/Russia). Hydrobiologia, 584: 133–143. Hering, D., Borja, A., Cartensen, J., Carvalho, L., Elliott, M., Feld, C, Heiskanen, A.-S., Johnson, R., Moe, J., Pont, D., Solheim A. L. & van de Bund, W. 2010. The European Water Framework Directive at the age of 10: a critical review of the achievements with recommendations for the future. Science of the Total Environment, 408: 4007–4019. Hering, D., Johnson, R. K. & Buffagni, A. 2006. Linking organism groups—major results and conclusions from the STAR project. Hydrobiologia, 566: 109–13. Hoge, F. E., Vodacek, A. & Blough, N. V. 1993. Inherent optical properties of the ocean: retrieval of the absorption coefficient of chromophoric dissolved organic matter from fluorescence measurements. Limnology and Oceanography, 38: 1394–1402. Holmer, M. & Storkholm, P. 2001. Sulphate reduction and sulphur cycling in lake sediments: a review. Freshwater Biology, 46: 431–451. Hubeny, J. B., King, J. W. & Santos, A. 2006. Subdecadal to multidecadal cycles of Late Holocene North Atlantic climate variability preserved by estuarine fossil pigments. Geology, 34(7): 569–572. Hurrell, J. W. 1995. Decadal trends in the North Atlantic Oscillation regional temperatures and precipitation. Science, 269: 676–679. Hurrell, J. W., Kushnir, Y., Ottersen, G. & Visbeck, M. 2003. The North Atlantic Oscillation. Climatic Significance and Environmental

79 Impact. Geophysical Monograph 134. American Geophysical Union, Washington. Ida-Eesti vesikonna veemajanduskava. 2016. Kinnitatud Vabariigi Valitsuse poolt 7. jaanuaril 2016.a., Keskkonnaministeerium, 334 lk. (East-Estonia River Basin management plan. 2016. Republic of Estonia, Ministry of the Environment, 334 pp.). Idrizaj, A., Laas, A., Anijalg, U. & Nõges, P. 2016. Horizontal differences in ecosystem metabolism of a large shallow lake. Journal of Hydrology, 535: 93–100. Iler, A. M., Inouye, D. W., Schmidt, N. M. & Høye T. T. 2017. Detrending phenological time series improves climate–phenology analyses and reveals evidence of plasticity. Ecology, 98(3): 647–655. IPCC 2013. Climate change 2013: the physical science basis. In: Stocker, T. F., Qin, D., Plattner, G.-K., Tignor, M., Allen, S. K., Boschung, J., Nauels, A., Xia, Y., Bex, V. & Midgley, P. M. (Eds), Contribution of working group I to the fifth assessment report of the intergovernmental panel on climate change. Cambridge University Press, Cambridge. Izaguirre, I., Sánchez, M. L., Schiaffino, M. R., O’Farrell, I., Huber, P., Ferrer, N., Zunino, J., Lagomarsino, L. & Mancini, M. 2015. Which environmental factors trigger the dominance of phytoplankton species across a moisture gradient of shallow lakes? Hydrobiologia, 752: 47–64. Jaani, A. & Kullus, L.-P. 2001. Hydroloqical regime and water balance. In: Lake Peipsi. Meteorology, Hydrologv, Hydrochemistry, Nõges, T. (Ed), pp. 65–72. Sulemees, Tartu. Jankowski, T., Livingstone, D. M., Forster, R. & Bührer, H. 2005. Long-term nitrate concentrations in five perialpine lakes: regional coherence, Verhandlungen der Internationalen Vereinigung für Theoretische und Angewandte Limnologie 29(2): 927–931. Jansson, A., Norkko, J., Dupont, S. & Norkko, A. 2015. Growth and survival in a changing environment: Combined effects of moderate hypoxia and low pH on juvenile bivalve Macoma balthica. Journal of Sea Research, 102: 41–47. Jennings, E., Jones, S., Arvola, L., Staehr, P. A., Gaiser, E., Jones, I. D., Weathers, K. C., Weyhenmeyer, G. A., Chiu, C. & De Eyto, E. 2012.

80 Effects of weather-related episodic events in lakes: an analysis based on high-frequency data. Freshwater Biology, 57: 589–601. Jensen, H. S. & Andersen, F. Ø. 1995. Importance of temperature, nitrate and pH for phosphorus release from aerobic sediments of four shallow, eutrophic lakes. Limnology and Oceanography, 37: 577–589. Jensen, T. C., Dimante-Deimantovica, I., Schartau, A. K. & Walseng, B. 2013. Cladocerans respond to differences in trophic state in deeper nutrient poor lakes from Southern Norway. Hydrobiologia, 715: 101– 112. Jeppesen, E., Meerhoff, M., Holmgren, K., González-Bergonzoni, I., Mello, F., Declerck, S., De Meester, L., Søndergaard, M., Lauridsen, T., Bjerring, R., Conde-Porcuna, J., Mazzeo, N., Iglesias, C., Reizenstein, M., Malmquist, H., Liu, Z., Balayla, D. & Lazzaro, X. 2010. Impacts of climate warming on lake fish community structure and potential effects on ecosystem function. Hydrobiologia, 646(1): 73–90. Jeppesen, E., Nõges, P., Davidson, T. A., Haberman, J., Nõges, T., Blank, K., Lauridsen, T. L., Søndergaard, M., Sayer, C., Laugaste, R., Johansson, L. S., Bjerring, R & Amsinck S. L. 2011. Zooplankton as indicators in lakes: a scientific-based plea for including zooplankton in the ecological quality assessment of lakes according to the European Water Framework Directive (WFD). Hydrobiologia, 676: 279–297. Jeppesen, E., Søndergaard, M., Jensen, J., Lauridsen, T., Havens, K., Anneville, O., Gerdeaux, D., Carvalho, L., Coveney, M., Deneke, R., Dokulil, M., Foy, B., Hampton, S., Winder, M., Hilt, S., Köhler, J., Kangur, K., Lammens, E., Portielje, R., Manca, M., Miracle, M., Romo, S., Moss, B., Nõges, P., Persson, G., Willén, E., Phillips, G., Schelske, C., Straile, D. & Tatrai, I. 2005. Lake responses to reduced nutrient loading - An analysis of contemporary long-term data from 35 case studies. Freshwater Biology, 50: 1747–1771. Jeppesen, E., Søndergaard, M., Meerhoff, M., Lauridsen, T. L. & Jensen, J. P. 2007. Shallow lake restoration by nutrient loading reduction – some recent findings and challenges ahead. Hydrobiologia, 584: 239– 252. Jeppesen, E., Søndergaard, M., Søndergaard, M. & Christoffersen, K. (Eds) 1998. The Structuring Role of Submerged Macrophytes in Lakes. Ecological Studies, 131. Springer, New York.

81 Johnson, R. A & Wichern D. W. 2007. Applied multivariate statistical analysis (6th ed). Prentice Hall, 773 pp. Jones, H. P. & Schmitz, O. J. 2009. Rapid Recovery of Damaged Ecosystems. PLoS ONE, 4(5): e5653. doi:10.1371/journal. pone.0005653. Joniak, T., Gołdyn, R. & Kozak, A. 2003. The primary production of phytoplankton in the restored Maltañski Reservoir in Poland. Hydrobiologia, 506–509, 311–316. Josefsson, H. & Baaner, L. 2011. Water Framework Directive – A Directive for the Twenty-First Century? Journal of Environmental Law, 23(3): 463–486. Järvalt, A. & Pihu, E. 2002. Influence of water level on fish stocks and catches in Lake Võrtsjärv. Proceedings of the Estonian Academy of Sciences. Biology, Ecology, 51(1): 74–84. Järvet, A., Mander, Ü., Kull, A. & Kuusemets, V. 2002. Nutrient runoff change in a rural catchment in South Estonia. Large Rivers, 13 (Archiv für Hydrobiologie Supplement, 141/3–4): 305−319. Järvinen, M., Rask, M., Ruuhijärvi, J. & Arvola, L. 2002. Temporal coherence in water temperature and chemistry under the ice of boreal lakes (Finland). Water Research, 36: 3949–3956. Jöhnk, K. D., Huisman, J., Sommeijer, B., Sharples, J, Visser, P. M. & Stroom, J. 2008. Summer heatwaves promote blooms of harmful cyanobacteria. Global Change Biology, 14: 495–512. Kagalou, I. & Leonardos, I. 2009. Typology, classification and management issues of Greek lakes: implication of the Water Framework Directive (2000/60/EC). Environmental Monitoring Assessment, 150: 469–484. Kallio, K. 1999. Absorption properties of dissolved organic matter in Finnish lakes. Proceedings of the Estonian Academy of Sciences. Biology, Ecology, 48: 75–83. Kangur, K., Ginter, K., Kangur, P., Kangur, A., Nõges, P. & Laas, A. 2016. Changes in water temperature and chemistry preceding a massive kill of bottom-dwelling fish: an analysis of high-frequency buoy data of shallow Lake Võrtsjärv (Estonia). Inland Waters, 6(4): 535−542.

82 Kangur, K., Kangur, A., Kangur, P. & Laugaste, R. 2005. Fish kill in Lake Peipsi in summer 2002 as a synergistic effect of a cyanobacterial bloom, high temperature, and low water level. Proceedings of the Estonian Academy of Sciences. Biology, Ecology, 54(1): 67–80. Kangur, K., Kangur, P., Ginter, K., Orru, K., Haldna, M. & Möls, T. 2013. Long-term effects of extreme weather events and eutrophication on the fish community of shallow Lake Peipsi (Estonia/Russia). Journal of Limnology, 72(2): 376–387. doi: 10.4081/jlimnol.2013.e30. Karlsson, J., Byström, B., Ask, J., Ask, P., Persson, L. & Jansson, M. 2009. Light limitation of nutrient-poor lake ecosystems. Nature, 460: 506–509. Karr, J. R. & Chu, E. W. 1997. Biological monitoring and assessment: using multimetric indexes effectively. EPA 235-R07-001, University of Washington, Seattle, WA. 149 pp. Kirk, J. T. O. 1994. Light and Photosynthesis in Aquatic Ecosystems. Cambridge University Press, 662 pp. Kisand, A., Kirsi, A.-L., Ehapalu, K., Alliksaar, T., Heinsalu, A., Tõnno, I., Leeben, A. & Nõges, P. 2017. Development of large shallow Lake Peipsi (North-Eastern Europe) over the Holocene based on the stratigraphy of phosphorus fractions. Journal of Paleolimnology, doi: 10.1007/s10933-017-9954-2. Kisand, V. & Nõges. T. 2004. Abiotic and biotic factors regulating dynamics of bacterioplankton in a large shallow lake. FEMS Microbiology Ecology, 50: 51–62. Kosten, S., Huszar, V., Bécares, E., Costa, S. L.,Van Donk, E., Hansson, L. A., Jeppesen, E., Kruk, C., Lacerot, G., Mazzeo, N., De Meester, L., Moss, B., Lürling, M., Nõges, T., Romo, S., & Scheffer, M. 2012. Warmer climates boost cyanobacterial dominance in shallow lakes. Global Change Biology, 18(1), 118−126. 10.1111/j.1365- 2486.2011.02488.x. Kovács, J., Nagy, M., Czauner, B., Kovács, I. S., Borsodi, A. K. & Hatvani, I. G. 2012. Delimiting sub-areas in water bodies using multivariate data analysis on the example of Lake Balaton (W Hungary). Journal of Environmental Management, 110: 151–158. Kraemer, B. M., Chandra, S., Dell, A. I., Dix, M., Kuusisto, E., Livingstone, D. M., Schladow, S. G., Silow, E., Sitoki, L. M., Tamatamah, R. & McIntyre, P. B. 2017. Global patterns in lake ecosystem responses

83 to warming based on the temperature dependence of metabolism. Global Change Biology, 23: 1881–1890. doi: 10.1111/gcb.13459 Kratz, T. K., Soranno, P. A., Baines, S. B., Benson, B. J., Magnuson, J. J., Frost, T.M. & Lathrop, R. C. 1998. Inter-annual synchronous dynamics in north temperate lakes in Wisconsin, USA. In: George, D. G., Jones, J. C., Puncochar, P., Reynolds, D. S. & Sutcliffe D.W. (Eds), Management of lakes and reservoirs during global climate change, Kluwer Academic Publishers, The Netherlands, pp. 273–287. Krause-Jensen, D., & Sand-Jensen, K. 1998. Light attenuation and photosynthesis of aquatic plant communities. Limnology and Oceanography, 43: 396–407. Kreps, T. A., Baldridge, A. K. & Lodge, D. M. 2012. The impact of an invasive predator (Orconectes rusticus) on freshwater snail communities: insights on habitat-specific effects from a multilake long-term study. Canadian Journal of Fisheries and Aquatic Sciences, 69(7): 1164– 1173, 10.1139/f2012-052 Kritzberg, E. S., Granéli, W., Björk, J., Brönmark, C., Hallgren, P., Nicolle, A., Persson, A., Hansson, L. A. 2014. Warming and browning of lakes: consequences for pelagic carbon metabolism and sediment delivery. Freshwater Biology, 59(2): 325–336. Kundzewicz, Z. W. & Robson, A. J. 2004. Change detection in hydrological records—a review of the methodology. Hydrological Sciences Journal, 49: 7–19. Körner, C. 1999. Alpine Plant Life. Springer-Verlag, Berlin, Heidelberg. Kyewalyanga, M., Platt, T. & Sathyendranath, S. 1992. Ocean primary production calculated by spectral and broad-band models. Marine Ecology Progress Series, 85: 171–185. Laas, A., Nõges, P., Kõiv, T. & Nõges, T. 2012. High frequency metabolism study in a large and shallow temperate lake revealed seasonal switching between net autotrophy and net heterotrophy. Hydrobiologia, 694(1): 57−74, 10.1007/s10750-012-1131-z. Lancelot, C. & Mathot, S. 1986. Biochemical fractionation of primary production by phytoplankton in Belgian coastal waters during short- and long-term incubations with 14C-bicarbonate. Marine Biology, 86(3): 219–226.

84 Lanci, L. & Hirt, A. M. 2015. Evidence of Atlantic Multidecadal Oscillation in the magnetic properties of Alpine lakes during the last 2500 years. Palaeogeography, Palaeoclimatology, Palaeoecology 440: 47–52. Lapierre, J.-F., Seekell, D. A. & Del Giorgio, P. A. 2015. Climate and landscape influence on indicators of lake carbon cycling through spatial patterns in dissolved organic carbon. Global Change Biology, 21: 4425–4435, doi: 10.1111/gcb.13031. Larsen, S., Andersen, T. & Hessen, D. O. 2011. Climate change predicted to cause severe increase of organic carbon in lakes. Global Change Biology, 17: 1186–192. Laugaste, R., Nõges, T., Nõges, P., Jastremskij, V. V., Milius A. & Ott, I. 2001. Algae. In: Pihu, E. & Haberman, J. (Eds), Lake Peipsi. Flora and Fauna, pp. 31–49. Sulemees Publishers, Tallinn, Estonia. Levin, S. & Segel, L. 1976. Hypothesis for origin of planktonic patchiness. Nature, 259: 659. Lindell, T., Pierson, D., Premazzi, G. & Ziliolli, E. 1999. Manual of Monitoring European Lakes Using Remote Sensing Techniques. Office for Official Publications of the European Communities, Luxembourg. Lindenmayer, D. B. & Likens, G. E. 2009. Adaptive monitoring: a new paradigm for long-term research and monitoring. Trends in Ecology and Evolution, 24: 482–486. Livingstone, D. M. 2003. Impact of secular climate change on the thermal structure of a large temperate central European lake. Climatic Change, 57: 205–225. Livingstone, D. M. 2008. A change of climate provokes a change of paradigm: taking leave of two tacit assumptions about physical lake forcing. International Review of Hydrobiology, 93(4–5): 404–414. Livingstone, D. M. 1997. An example of the simultaneous occurrence of climate-driven “sawtooth” deep-water warming/cooling episodes in several Swiss lakes. Verhandlungen der Internationalen Vereinigung für Theoretische und Angewandte Limnologie, 26(2): 822–826. Livingstone, D. M., Adrian, R., Arvola, L., Blenckner, T., Dokulil, M. T., Hari, R. E., George, G., Jankowski, T., Järvinen, M., Jennings, E., Nõges, P., Nõges, T., Straile, D. & Weyhenmeyer G. A. 2010a.

85 Chapter 17: Regional and Supra-Regional Coherence in Limnological Variables. In: George, G. (Ed), The Impact of Climate Change on European Lakes Aquatic Ecology Series 4. Springer, Dordrecht: 311–337. Livingstone, D. M., Adrian, R., Blenckner, T., George, G. & Weyhenmeyer, G. A. 2010b. Lake ice phenology. In: George, D. G. (Ed), The Impact of Climate Change on European Lakes, Aquatic Ecology Series 4. Springer, Dordrecht: 51–61. Loftis, J. C., McBride, G. B. & Ellis, J. C. 1991. Considerations of scale in water quality monitoring and data analysis. Journal of the American Water Resources Association, 27: 255–264. Lyche-Solheim, A., Feld, C. K., Birk, S., Phillips, G., Carvalho, L., Morabito, G., Mischke, U., Willby, N., Søndergaard, M., Hellsten, S., Kolada, A., Mjelde, M., Böhmer, J., Miler, O., Pusch, M. T., Argillier, C., Jeppesen, E., Lauridsen, T. L. & Poikane, S. 2013. Ecological status assessment of European lakes: a comparison of metrics for phytoplankton, macrophytes, benthic invertebrates and fish. Hydrobiologia, 704: 57–74. doi 10.1007/s10750-012-1436-y Magnuson, J. J., Benson, B. J. & Kratz, T. K. 2004. Patterns of coherent dynamics within and between lake districts at local to intercontinental scales, Boreal Environment Research 9: 359–369. Massol, F., David, P., Gerdeaux, D. & Jarne, P. 2007. The influence of trophic status and large-scale climatic change on the structure of fish communities in Perialpine lakes. Journal of Animal Ecology, 76: 538–551. Meerhoff, M., Iglesias, C., De Mello, F. T, Clemente, J. M., Jensen, E., Lauridsen, T. L. & Jeppesen, E. 2007. Effects of habitat complexity on community structure and predator avoidance behaviour of littoral zooplankton in temperate versus subtropical shallow lakes. Freshwater Biology, 52:1009–1021. Milius, A., Laugaste, R., Möls, T., Haldna, M. & Kangur. K. 2005. Water level and water temperature as factors determining phytoplankton biomass and nutrient content in Lake Peipsi. Proceedings of the Estonian Academy of Sciences. Biology, Ecology, 54(1): 5–17. Mischke, U. & Nixdorf, B. 2003. Equilibrium phase conditions in shallow German lakes: how Cyanoprokaryota species establish a steady state phase in late summer. Hydrobiologia, 502: 123–132.

86 Mischke, U., Riedmüller, U., Hoehn, E., Schönfelder, I. & Nixdorf, B. 2008. Description of the German system for phytoplankton-based assessment of lakes for implementation of the EU Water Framework Directive (WFD). In: Mischke, U. & Nixdorf, B. (Eds), Gewäserreport (Nr. 10), Brandenburg Technical University of Cottbus, Cottbus. ISBN 978-3-940471-06-2, BTUC-AR 2/2008: 117–146. Mobley, C. D. 1994. Light in Water-Radiative Transfer in Natural Waters. Academic Press, San Diego. Molot, L. A. & Dillon, P. J. 1997. Colour - mass balances and colour- dissolved organic carbon relationships in lakes and streams in central Ontario. Canadian Journal of Fisheries and Aquatic Sciences, 54(12): 2789–2795. Monteith, D. T., Evans, C. D. & Reynolds, B. 2000. Are temporal variations in the nitrate content of UK upland freshwaters linked to the North Atlantic Oscillation? Hydrological Processes, 14: 1745–1749. Monteith, D. T., Stoddard, J. L., Evans, C. D., De Wit, H. A., Forsius, M., Høgåsen, T., Wilander, A., Skjelkvåle, B. L., Jeffries, D.S, Vuorenmaa, J., Keller, B., Kopácek, J. & Vesely, J. 2007. Dissolved organic carbon trends resulting from changes in atmospheric deposition chemistry. Nature, 450: 537–541. Morales-Baquero, R., Carrillo, P., Barea-Arco, J., Pérez-Martínez, C. & Villar-Argaiz, M. 2006. Climate-driven changes on phytoplankton- zooplankton coupling and nutrient availability in high mountain lakes of Southern Europe. Freshwater Biology, 51(6): 989–998. Morozov, A. Y. & Kuzenkov, O. A. 2016. Towards developing a general framework for modelling vertical migration in zooplankton. Journal of Theoretical Biology, 405: 17–28. Moss B. 2007. Shallow lakes, the water framework directive and life. What should it all be about? Hydrobiologia, 584: 381–94. Moss, B. 2008. The Water Framework Directive: total environment or political compromise? Science Total Environment, 400(1–3): 32–41. Moss, B., Stephen, D., Alvarez, C., Becares, E., van de Bund, W., Collings, S. E., Van Donk, E., De Eyto, E., Feldmann, T., Fernandez-Alaez, C., Fernandez-Alaez, M., Franken, R. J. M., Garcıa-Criado, F., Gross, E. M., Gyllström, M., Hansson, L. A., Irvine, K., Järvalt, A., Jensen, J. P., Jeppesen, E., Kairesalo, T., Kornijow, R., Krause, T., Künnap, H., Laas, A., Lill, E., Lorens, B., Luup, H., Miracle, M. R., Nõges, P.,

87 Nõges, T., Nykänen, M., Ott, I., Peczula, W., Peeters, E. T. H. M., Phillips, G., Romo, S., Russell, V., Salujõe, J., Scheffer, M., Siewertsen, K., Smal, H., Tesch, C., Timm, H., Tuvikene, L., Tõnno, I., Virro, T., Vicente, E. & Wilson, D. 2003. The determination of ecological status in shallow lakes – a tested system (ECOFRAME) for implementation of the European Water Framework Directive. Aquatic Conservation: Marine Freshwater Ecosystems, 13: 507–549. Moss, B., Kosten, S., Meerhoff, M., Battarbee, R. W., Jeppesen, E., Mazzeo, N., Havens, K., Lacerot, G., Liu, Z., De Meester, L., Paerl, H. & Scheffer, M. 2011. Allied attack: climate change and eutrophication. Inland Waters, 1: 101–105. Murphy, K. J., Kennedy, M. P., McCarthy, V., Ó’Hare, M. T., Irvine K. & Adams, C. 2002. A Review of Ecology Based Classification Systems for Standing Freshwaters. SNIFFER Project Number W(99)65. Environment Agency R&D Technical Report: E1-091/TR. Møller Jensen, L. 1985. 14C-labelling patterns of phytoplankton: specific activity of different product pools. Journal of Plankton Research, 7(5): 643–652. Mühlen, M. von zur & Schneider, G. 1920. Der See Wirzjerw in Livland. (Lake Võrtsjärv in Livland. In German). Archiv für die Naturkunde des Ostbaltikums 14, 1–156. Nixdorf, B., Lessmann, D. & Deneke, R. 2005. Mining lakes in a disturbed landscape: Application of the EC Water Framework Directive and future management strategies. Ecological Engineering, 24: 67–73. Nõges, P., van de Bund, W., Cardoso, A. C., Solimini, A. G. & Heiskanen, A.-S. 2009c. Assessment of the ecological status of European surface waters: a work in progress. Hydrobiologia, 633 (1), 197−211. Nõges, P., van de Bund, W., Cardoso, C. & Heiskanen, A. S. 2007a. Impact of climatic variability on parameters used in typology and ecological quality assessment of surface waters – implications on the Water Framework Directive. Hydrobiologia, 584: 373–379. Nõges, P. & Järvalt, A. 2004. Long-term changes in the ecosystem of Lake Võrtsjärv and their causes. In: Haberman, J., Pihu, E. & Raukas, A. (Eds), Lake Võrtsjärv, pp. 347–355. Estonian Encyclopaedia Publishers, Tallinn, Estonia.

88 Nõges, P. & Järvet, A. 1995. Water level control over light conditions in shallow lakes. Report Series in Geophysics. Report No 32. University of Helsinki, Helsinki: 81–92. Nõges, P. & Järvet, A. 2005. Climate driven changes in the spawning of roach (Rutilus rutilus (L.)) and bream (Abramis brama (L.)) in the Estonian part of the Narva River basin. Boreal Environment Research, 10: 45–55. Nõges, P., Kägu, M. & Nõges, T. 2007c. Role of climate and agricultural practice in determining the matter discharge into large shallow Lake Võrtsjärv, Estonia. Hydrobiologia, 581: 125–134. Nõges, P. & Laugaste, R. 1998. Seasonal and long-term changes of phytoplankton in L. Võrtsjärv. Limnologica, 28(1): 21–28. Nõges, P., Laugaste, R. & Nõges, T. 2004. Phytoplankton. In: Haberman, J., Pihu, E. & Raukas, A. (Eds), Lake Võrtsjärv, pp. 217–231. Estonian Encyclopaedia Publishers, Tallinn. Nõges, P., Mischke, U., Laugaste, R. & Solimini, A. G. 2010a. Analysis of changes over 44 years in the phytoplankton of Lake Võrtsjärv (Estonia): the effect of nutrients, climate and the investigator on phytoplankton-based water quality indices. Hydrobiologia, 646: 33–48. Nõges, P. & Nõges, T. 1998. The effect of fluctuating water level on the ecosystem of Lake Võrtsjärv, Central Estonia. Proceedings of the Estonian Academy of Sciences. Biology, Ecology, 47: 98–113. Nõges, P., Nõges, T., Haberman, J., Laugaste, R. & Kisand, V. 1997. Tendencies and relations in the plankton community and pelagic environment of Lake Võrtsjärv during three decades. Proceedings of the Estonian Academy of Sciences. Biology, Ecology, 46: 40–57. Nõges, P., Nõges, T. & Laas, A. 2010b. Climate-related changes of phytoplankton seasonality in large shallow Lake Võrtsjärv, Estonia. Aquatic Ecosystem Health & Management, 13(2): 154–163, doi: 10.1080/14634981003788953. Nõges, P. & Nõges, T. 2006. Indicators and criteria to assess ecological status of the large shallow temperate polymictic lakes Peipsi (Estonia/ Russia) and Võrtsjarv (Estonia). Boreal Environmental Research, 11: 67–80.

89 Nõges, P. & Nõges, T. 2014. Weak trends in ice phenology of Estonian large lakes despite significant warming trends. Hydrobiologia, 731: 5–18. Nõges, P., Nõges, T., Ghiani, M., Paracchini, B., Grande, J. P. & Sena, F. 2011a. Morphometry and trophic state modify the thermal response of lakes to meteorological forcing. Hydrobiologia, 667: 241–254. doi 10.1007/s10750-011-0691-7. Nõges, T. 2009a. Relationships between lake morphometry, geographic location and water quality parameters of European lakes. Hydrobiologia, 633: 33–43. Nõges, T. 2009b. Trends of air and water temperature in Estonia and in large lakes Peipsi and Võrtsjärv, and possible consequences on water quality. Stuttgart Verhandlungen der Internationalen Vereinigung für Theoretische und Angewandte Limnologie, 30(7): 997–999. Nõges, T., Arst, H., Laas, A., Kauer, T., Nõges, P. & Toming, K. 2011b. Reconstructed long-term time series of phytoplankton primary production of a large shallow temperate lake: the basis to assess the carbon balance and its climate sensitivity. Hydrobiologia, 667(1): 205−222, s10750-011-0647-y. Nõges, T., Järvet, A., Kisand, A., Laugaste, R., Loigu, E., Skakalski, B. & Nõges, P. 2007b. Reaction of large and shallow lakes Peipsi and Võrtsjärv to the changes of nutrient loading. Hydrobiologia, 584: 253–264. Nõges, T., Järvet, A., Laugaste, R., Loigu, E., Leisk, Ü., Tõnno, I. & Nõges, P. 2005. Consequences of catchment processes and climate changes on the ecological status of large shallow temperate lakes. In: Ramachandra, T. V., Ahalya, N. & Murthy, C. R. (Eds), Aquatic ecosystems, conservation, restoration and management, pp. 88–99. Capital Publishing Company, New Dehli, Kolkata, Bangalore. Nõges, T., Järvet, A., Laugaste, R., Loigu, E., Tõnno, I. & Nõges, P. 2003a. Current State and Historical Changes of Nutrient Loading and Ecological Status of Large Lakes Peipsi and Võrtsjärv, pp. 199– 203. In: Simola, H., Terzhevik, A. Yu, Viljanen, M. & Holopainen, I. K. (Eds), Proceedings of 4th International Lake Ladoga Symposium 2002 September 2-6 Russia University of Joensuu, Finland. Nõges, T. & Kangro, K. 2005. Primary production of phytoplankton in a strongly stratified temperate lake. Hydrobiologia, 547: 105–122.

90 Nõges, T., Kisand, V., Nõges, P., Põllumäe, A., Tuvikene, L. & Zingel, P. 1998. Plankton seasonal dynamics and its controlling factors in shallow polymictic eutrophic Lake Võrtsjärv, Estonia. Internationale Revue der gesamten Hydrobiologie und Hydrographie, 83(4), 279– 296. Nõges, T., Laugaste, R., Nõges, P. & Tõnno, I. 2008. Critical N:P ratio for cyanobacteria and N2-fixing species in the large shallow temperate lakes Peipsi and Võrtsjärv, North-East Europe. Hydrobiologia, 599: 77–86. Nõges, T. & Nõges, P. 1999. The effect of extreme water level decrease on hydrochemistry and phytoplankton in a shallow eutrophic lake. Hydrobiologia, 409: 277–283. Nõges, T., Nõges, P. & Laugaste, R. 2003b. Water level as the mediator between climate change and phytoplankton composition in a large shallow temperate lake. Hydrobiologia, 506–509: 257–263. Paavel, B., Arst, H., & Herlevi, A. 2007. Dependence of spectral distribution of inherent optical properties of lake waters on the concentrations of different water constituents. Nordic Hydrology, 38(3): 265−285. Pace, M. L. & Cole, J. J. 2002. Synchronous variation of dissolved organic carbon and color in lakes. Limnology and Oceanography, 47(2): 333– 342. Paerl, H. W. & Huisman, J. 2008. Blooms like it hot. Science, 320: 57–58. Palecki, M. A. & Barry, R. G. 1986. Freeze-up and break-up of lakes as an index of temperature changes during the transition seasons: a case study for Finland. Journal of Climate and Applied Meteorology, 25: 893–902. Patil, G. P. 1991. Encountered data, statistical ecology, environmental statistics, and weighted distribution methods. Environmetrics, 2: 377–423. Peeters, E. T. H. M., Franken, R. J. M., Jeppesen, E., Moss. B., Bécares, E., Hansson, L.-A., Romo, S., Kairesalo, T., Gross, E. M., vanDonk, E., Nõges, T., Irvine, K., Kornijów, R. & Scheffer, M. 2009. Assessing ecological quality of shallow lakes: Does knowledge of transparency suffice? Basic and Applied Ecology, 10(1): 89–96.

91 Philander, S. G. 1990. El Niño, La Niña and the Southern Oscillation. Academic Press, San Diego: 293 pp. Poikāne, S., Alves, M. H., Argillier, C., van den Berg, M., Buzzi, F., Hoehn, E., de Hoyos, C., Karottki, I., Laplace-Treyture, C., Solheim, A. L., Ortiz-Casas, J., Ott, I., Phillips, G., Pilke, A., Pádua, J., Remec-Rekar, S., Riedmüller, U., Schaumburg, J., Serrano, M. L., Soszka, H., Tierney, D., Urbanic, G. & Wolfram, G. 2010. Defining Chlorophyll-a Reference Conditions in European Lakes. Environmental Management, 45(6): 1286−1298, 10.1007/s00267-010-9484-4. Pouzols, F. M. & Lendasse, A. 2010. Effect of Different Detrending Approaches on Computational Intelligence Models of Time Series. WCCI 2010 IEEE World Congress on Computational Intelligence July, 18-23, 2010 - CCIB, Barcelona, Spain. Randall, D. J. & Tsui T. K. N. 2002. Ammonia toxicity in fish. Marine Pollution Bulletin, 45: 17–23. Rask, M., Vuori, K. M., Hämäläinen, H, Järvinen, M., Hellsten, S., Mykrä, H., Arvola, L., Ruuhijärvi, J., Jyväsjärvi, J., Kolari, I., Olin, M., Salonen, E. & Valkeajärvi, P. 2011. Ecological classification of large lakes in Finland: comparison of classification approaches using multiple quality elements. Hydrobiologia, 660(1): 37–47. Rasconi, S., Gall, A., Winter, K. & Kainz, M. J. 2015. Increasing Water Temperature Triggers Dominance of Small Freshwater Plankton. PLoS ONE 10(10): e0140449. doi:10.1371/journal.pone.0140449 Raymond, P. A., Hartmann, J., Lauerwald, R., Sobek, S., McDonald, C., Hoover, M., Butman, D., Striegl, R., Mayorga, E., Humborg, C., Kortelainen, P., Dürr, H., Meybeck, M., Ciais, P. & Guth, P. 2013, Global carbon dioxide emissions from inland waters. Nature, 503: 355–359, doi:10.1038/nature12760. Reinart, A., Herlevi, A., Arst, H. & Sipelgas, L. 2003. Preliminary optical classification of lakes and coastal waters in Estonia and South- Finland. Journal of Sea Research, 49: 357–366. Reist, J. D. 1986. An empirical evaluation of coefficients used in residual and allometric adjustment of size covariation. Canadian Journal of Zoology, 64: 1363–1368. Richardson, A. D., Lee X. & Friedland, A. J. 2004. Microclimatology of treeline spruce-fir forests in mountains of the northeastern United States. Agricultural and Forest Meteorology, 125: 53–66.

92 Rocha, M. R., Vasseur, D. A., Hayn, M., Holschneider, M. & Gaedke, U. 2011. Variability patterns differ between standing stock and process rates. Oikos, 120: 17–25. doi: 10.1111/j.1600-0706.2010.18786.x. Robson, B. J. 2005. Representing the effects of diurnal variations in light on primary production on a seasonal time-scale. Ecological Modelling, 186: 358–365. Rodó, X., Baert, E. & Comin, F. A. 1997. Variations in seasonal rainfall in Southern Europe during the present century: relationships with the North Atlantic Oscillation and the El Niño-Southern Oscillation. Climate Dynamics, 13: 275–284. Rodríguez, M. A. & Magnan, P. 1995. Application of multivariate analyses in studies of the organization and structure of fish and invertebrate communities. Aquatic Sciences –Research Across Boundaries, 57: 199–216. Rogalski, M. A. 2016. Ecological and evolutionary responses of zooplankton communities to changes in lake chemical environments. Networked Digital Library of Theses & Dissertations, Ipswich, MA. Rossa, F., Fasani, S. & Rinaldi, S. 2013. Conditions for patchiness in plankton models. Theoretical Population Biology, 83: 95–100. Roulet, N. & Moore, T. R. 2006. Browning the waters. Nature, 444: 283– 284. Rustad, L. E., Campbell, J. L., Marion, G. M., Norby, R. J., Mitchell, M., Hartley, A. E., Cornellissen, J. H. C. & Gurevitch, J. 2001. A meta-analysis of the response of soil respiration, net nitrogen mineralization, and aboveground plant growth to experimental ecosystem warming. Oecologia, 126: 543–562. Salmaso, N. & Cerasino, L. 2012. Long-term trends and fine year-to-year tuning of phytoplankton in large lakes are ruled by eutrophication and atmospheric modes of variability. Hydrobiologia, 698: 17–28. doi: 10.1007/s10750-012-1068-2. Saloranta, T. M. & Andersen, T. 2007. MyLake—a multi-year lake simulation model code suitable for uncertainty and sensitivity analysis simulations. Ecological Modelling, 207: 45–60. Sandin, L. 2005. Testing the EC Water Framework Directive ‘‘one-out, all-out’’ rule –simulating different levels of assessment errors along a pollution gradient in Swedish streams. Internationale Vereinigung

93 für Theoretische und Angewandte Limnologie: Verhandlungen, 29: 334–336. Sathyendranath, S., Platt, T., Caverhill, C. M., Warnock, R.E. & Lewis M.R. 1989. Remote sensing of oceanic primary production: computations using a spectral model. Deep-Sea Research, 36: 431–453. Schanz, F. 1985. Vertical light attenuation and phytoplankton development in Lake Zürich. Limnology and Oceanography, 30(2): 299–310. Scheffer, M. 2004. Ecology of Shallow Lakes. Kluwer Academic Publishers, London: 374 pp. Scheffer, M., Bascompte, J., Brock, W. A., Brovkin, W., Carpenter, S. R., Dakos, V., Held, H., van Nes, E. H., Rietkerk, M. & Sugihara, G. 2009. Early-warning signals for critical transitions. Nature, 461: 53–59. doi:10.1038/nature08227. Scheffer, M. & Carpenter, S. R. 2003. Catastrophic regime shifts in ecosystems: linking theory to observation. Trends in Ecology & Evolution, 18(12): 648–656. Scheffer, M., Hosper, S. H., Meijer, M. L., Moss, B. & Jeppesen, E. 1993. Alternative equilibria in shallow lakes. Trends in Ecology & Evolution, 8: 275–279. Scheffer, M. & Jeppesen, E. 2007. Regime shifts in shallow lakes. Ecosystems, 10: 1–3. Scheffer, M., Rinaldi, S., Gragnani, A., Mur, L. R. & van Nes, E. H. 1997. On the dominance of filamentous cyanobacteria in shallow, turbid lakes. Ecology, 78: 272–282. Scheffer, M. & van Nes, E. H. 2007. Shallow lakes theory revisited: various alternative regimes driven by climate, nutrients, depth and lake size. Hydrobiologia, 584: 455–466. Scheffer, M., Straile, D., van Nes, E. H. & Hosper, H. 2001. Climatic warming causes regime shifts in lake food webs. Limnology and Oceanography, 46: 1780–1783. Schindler, D. W. 1978. Factors regulating phytoplankton production and standing crop in the World’s freshwaters. Limnology and Oceanography, 23: 478–486. Schofield, O., Bidigare, R. R. & Prézelin, B. B. 1990. Spectral photosynthesis, quantum yield and blue-green light enhancement

94 of productivity rates in the diatom Chaetoceros gracile and the prymnesiophyte Emiliania huxleyi. Marine Ecology Progress Series, 64: 175–186. Seekell, D. A., Carr, J. A., Gudasz, C. & Karlsson, J. 2014. Upscaling carbon dioxide emissions from lakes. Geophysical Research Letters, 41(21): 7555–7559. https://doi.org/10.1002/2014GL061824. Seekell, D. A., Lapierre, J.-F. & Karlsson, J. 2015. Trade-offs between light and nutrient availability across gradients of dissolved organic carbon concentration in Swedish lakes: implications for patterns in primary production. Canadian Journal of Fisheries and Aquatic Sciences, 72: 1663–1671. dx.doi.org/10.1139/cjfas-2015-0187. Sharma, S. & Magnuson, J. J. 2014. Oscillatory dynamics do not mask linear trends in the timing of ice breakup for Northern Hemisphere lakes from 1855 to 2004. Climatic Change, 124: 835–847. doi: 10.1007/s10584-014-1125-0 Sigmund, A.-M. 2006. Opinion of the European Economic and Social Committee on Meeting the challenges of climate change — The role of civil society. Official Journal C 318, 23/12/2006 P. 0102 – 0108. http://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CEL EX:52006IE1160&from=EN (10.04.2017) Smith, L. C., Sheng Y., MacDonald G. M. & Hinzman, L. D. 2005. Disappearing Arctic Lakes. Science, 308(5727): 1429. Smith, R. C, Prézelin, B. B., Bidigare, R. R. & Baker, K. S. 1989. Biooptical modeling of photosynthetic production in coastal waters. Limnology and Oceanography, 34: 1524–1544. Soininen, J., Tallberg, P. & Horppila, J. 2005. Phytoplankton community assembly in a large boreal lake – deterministic pathways or chaotic fluctuations? Freshwater Biology, 50: 2076–2086. Sommer, U., Gliwicz, Z. M., Lampert, W. & Duncan, A. 1986. The PEG- model of seasonal succession of planktonic events in fresh waters. Archiv für Hydrobiologie, 106: 433–471. Sorvari, S. 2001. Climate Impacts on Remote Subarctic Lakes in Finnish Lapland: Limnological and Palaeolimnological Assessment with a Particular Focus on Diatoms and Lake Saanajaärvi. Academic dissertation in Hydrobiology University of Helsinki, Helsinki.

95 Sosik, H. M. 1996. Bio-optical modelling of primary production: consequences of variability in quantum yield and specific absorption. Marine Ecology Progress Series, 143: 225–238. Søndergaard, M., Jensen, J. P. & Jeppesen, E. 2003. Role of sediment and internal loading of phosphorus in shallow lakes. Hydrobiologia, 506: 135–145. Søndergaard, M., Jensen, J. P., Jeppesen, E. & Hald Møller, P. 2002. Seasonal dynamics in the concentrations and retention of phosphorus in shallow Danish lakes after reduced loading. Aquatic Ecosystem Health and Management, 5: 19–29. Søndergaard, M., Jeppesen, E., Jenser, J. P. & Amsink, S. L. 2005. Water Framework Directive: ecological classification of Danish lakes. Journal of Applied Ecology, 42: 616–629. Søndergaard, M., Kristensen, P. & Jeppesen, E. 1992. Phosphorus release from resuspended sediment in the shallow and wind-exposed Lake Arresø, Denmark. Hydrobiologia, 228: 91–99. Søndergaard, M., Larsen, S. E., Johansson, L. S., Lauridsen, T. L. & Jeppesen, E. 2016. Ecological classification of lakes: Uncertainty and the influence of year-to-year variability. Ecological Indicators, 61: 248–257. Spears, B. M., Carvalho, L., Futter, M. N., May, L., Thackeray, S. J., Adrian, R., Angeler, D. G., Burthe, S. J., Davidson, T. A., Daunt, F., Gsell, A. S., Hessen, D. O., Moorhouse, H., Huser, B., Ives, S. C., Janssen, A. B. G., Mackay, E. B., Søndergaard, M. & Jeppesen, E. 2016. Ecological Instability in Lakes: A Predictable Condition? Environmental Science & Technology, 50: 3285−3286. Staehr, P. A., Brighenti, L. S., Honti, M., Christensen, J. & Rose, K. C. 2016. Global patterns of light saturation and photoinhibition of lake primary production. Inland Waters, 6: 593–607. Straile, D. 2002. North Atlantic Oscillation synchronizes food-web interactions in central European lakes. Proceedings of the Royal Society of London, B 269: 391–395. Straile, D. & Adrian, R. 2000. The North Atlantic Oscillation and plankton dynamics in two European lakes – two variations on a general theme. Global Change Biology, 6: 663–670.

96 Straile, D., Jöhnk, K. & Rossknecht, H. 2003. Complex effects of winter warming on the physicochemical characteristics of a deep lake. Limnology and Oceanography, 48: 1432–1438. Straile, D., Kerimoglu, O., Peeters, F., Jochimsen, M. C., Kümmerlin, R., Rinke K. & Rothhaupt, K.-O. 2010. Effects of a half a millennium winter on a deep lake – a shape of things to come? Global Change Biology, 16: 2844–2856. Straile, D., Kerimoglu, O. & Peeters F. 2015. Trophic mismatch requires seasonal heterogeneity of warming. Ecology, 96(10): 2794–2805. Strock, K. E., Saros, J. E., Nelson, S. J., Birkel, S. D., Kahl, J. S. & McDowell W.H. 2016. Extreme weather years drive episodic changes in lake chemistry: implications for recovery from sulfate deposition and long-term trends in dissolved organic carbon. Biogeochemistry, 127: 353–365. Tammert, H. & Kisand, V. 2004. Bacterioplankton. In: Haberman, J., Pihu, E. & Raukas, A. (Eds), Lake Võrtsjärv, pp. 207–215. Estonian Encyclopaedia Publishers, Tallinn. Tillmann, U., Hesse, K.-H. & Coljin, F. 2000. Planktonic primary production in the German Wadden Sea. Journal of Plankton Research, 22(7), 1253–1276. Tilzer, M. M. & Serruya C. (Eds) 1990. Large lakes: Ecological Structure and Function. Springer-Verlag, Berlin. Timms, R. M. & Moss, B. 1984. Prevention of growth of potentially dense phytoplankton populations by zooplankton grazing in the presence of zooplanktivorous fish, in a shallow wetland ecosystem. Limnology and Oceanography, 29: 472–486. Tomé, A. R. & Miranda, P. M. A. 2004. Piecewise linear fitting and trend changing points of climate parameters. Geophysical Research Letters, 31. doi:10.1029/2003GL019100. Toming, K., Arst, H., Paavel, B., Laas, A. & Nõges, T. 2009. Spatial and temporal variations in coloured dissolved organic matter in large and shallow Estonian waterbodies. Boreal Environment Research, 14(6): 959−970. Toming, K., Nõges, P., Arst, H., Kõiv, T. & Nõges, T. 2013. Reconstruction of long-term changes of the underwater light field in large shallow lakes Peipsi and Võrtsjärv, North-East Europe. Proceedings of

97 the Estonian Academy of Sciences, 62(3): 202–212. doi: 10.3176/ proc.2013.3.06. Tranvik, L. 1988. Availability of Dissolved Organic Carbon for Planktonic Bacteria in Oligotrophic Lakes of Differing Humic Content. Microbial Ecology, 16(3): 311–322. Retrieved from http:// www.jstor.org/stable/4251025 Tranvik, L. J. 1990. Bacterioplankton growth on fractions of dissolved organic carbon of different molecular weights from humic and clear waters. Applied and Environmental Microbiology, 56: 1672–1677. Vilar, J., Solé, R. & Rubí, J. 2003. On the origin of plankton patchiness. Physica A: Statistical Mechanics and its Applications, 317: 239–246. Wagner, A. & Benndorf, J. 2007. Climate-driven warming during spring destabilises a Daphnia population: a mechanistic food web approach. Oecologia, 151: 351–364. doi:10.1007/s00442-006-0554-5. Wagner, T., Soranno, P. A., Webster, K. E. & Cheruvelil, K.S. 2011. Landscape drivers of regional variation in the relationship between total phosphorus and chlorophyll in lakes. Freshwater Biology, 56(9): 1811–1824. doi:10.1111/j.1365-2427.2011.02621.x. Walter, K. M., Zimov, S. A., Chanton, J. P., Verbyla, D. & Chapin, F. S. III. 2006. Methane bubbling from Siberian thaw lakes as a positive feedback to climate warming. Nature, 443: 71–75. Wang, X. J. & Liu, R. M. 2004. Spatial analysis and eutrophication assessment for chlorophyll a in Taihu Lake. Environmental Monitoring and Assessment, 101(1–3): 167–174. Wang, J., Zhang, Y., Yang, F., Cao, X., Bai, Z., Zhu, J., Chen, E., Li, Y. & Ran, Y. 2015. Spatial and temporal variations of chlorophyll-a concentration from 2009 to 2012 in Poyang Lake, China. Environmental Earth Sciences, 73: 4063–4075. Wetzel, R. G. 2001. Limnology: lake and river ecosystems. 3rd ed. Academic Press, London, 1006 pp. Weyhenmeyer, G. A. 2001. Warmer winters: are planktonic algal populations in Sweden’s largest lakes affected? Ambio, 30: 565–571. Weyhenmeyer, G. A. 2004. Synchrony in relationships between the North Atlantic Oscillation and water chemistry among Sweden’s largest lakes. Limnology and Oceanography, 49(4): 1191–1201.

98 Weyhenmeyer, G. A., Blenckner, T. & Pettersson, K. 1999. Changes of the plankton spring outburst related to the North Atlantic Oscillation. Limnology and Oceanography, 44: 1788–1792. Weyhenmeyer, G. A., Jeppesen, E., Adrian, R., Arvola, L., Blenckner, T., Jankowski, T., Jennings, E., Jensen, J. P., Nõges, P., Nõges, T. & Straile, D. 2007. Nitrogen-depleted conditions on the increase in shallow northern European lakes. Limnology and Oceanography, 52(4): 1346–1353. Weyhenmeyer, G. A., Meili M. & Livingstone, D. M. 2004b. Nonlinear temperature response of lake ice breakup. Geophysical Research Letters, 31: L07203. Weyhenmeyer, G. A., Meili, M. & Livingstone, D. M. 2005. Systematic differences in the trend towards earlier ice-out on Swedish lakes long a latitudinal temperature gradient. Verhandlungen des Internationalen Verein Limnologie, 29: 257–260. Weyhenmeyer, G.A., Müller, R.A., Norman, M. & Tranvik, L. J. 2016. Sensitivity of freshwaters to browning in response to future climate change. Climatic Change, 134: 225–239. doi:10.1007/s10584-015- 1514-z. Weyhenmeyer G. A., Willen, E. & Sonesten, L. 2004a. Effects of an extreme precipitation event on water chemistry and phytoplankton in the Swedish Lake Malaren. Boreal Environment Research, 9: 409– 420. WFD CIS 2003. Common Implementation Strategy for the Water Framework Directive (2000/60/EC). Guidance Document No 5. Transitional and Coastal Waters – Typology, Reference Conditions and Classification Systems. Office for Official Publications ofthe European Communities, Luxembourg, 107 pp. WFD CIS 2005. Common Implementation Strategy for the Water Framework Directive (2000/60/EC). Guidance Document No 13. Overall Approach to the Classification of Ecological Status and Ecological Potential. Office for Official Publications of the European Communities, Luxembourg: 47 pp. WFD CIS 2009. Common Implementation Strategy for the Water Framework Directive (2000/60/EC). Guidance Document No 24. River basin management in a changing climate. European Communities, 134 pp.

99 WFD CIS 2011. Common Implementation Strategy for the Water Framework Directive (2000/60/EC). Guidance Document No. 27. Technical Guidance For Deriving Environmental Quality Standards. Office for Official Publications of the European Communities, Luxembourg, 2011, 204 pp. Wheeler, P. A., Huyer A. & Fleischbein, J. 2003. Cold halocline, increased nutrients and higher chlorophyll off Oregon in 2002. Geophysical Research Letters, 30(15): 8021, doi:10.1029/2003GL017395. Wicks, B. J., Joensen, R., Tang, Q. & Randall D. J. 2002. Swimming and ammonia toxicity in salmonids: the effect of sub lethal ammonia exposure on the swimming performance of coho salmon and the acute toxicity of ammonia in swimming and resting rainbow trout. Aquatic Toxicology, 59: 55–69. Wiedner, C., Rücker, J., Brüggemann, R. & Nixdorf, B. 2007. Climate change affects timing and size of populations of an invasive cyanobacterium in temperate regions. Oecologia, 152: 473–84. Wilhelm, S. & Adrian, R. 2008. Impact of summer warming on the thermal characteristics of a polymictic lake and consequences for oxygen, nutrients and phytoplankton. Freshwater Biology, 53: 226– 237. Williams, G., Layman, K. L. & Stefan, H. G. 2004. Dependence of lake ice covers on climatic, geographic and bathymetric variables. Cold Regions Science and Technology, 40: 145–164. Winder M. & Schindler D. E. 2004 Climate change uncouples trophic interactions in an aquatic ecosystem. Ecology, 85: 2100–2106. Woolway, R. I., Meinson, P., Nõges, P., Jones, I. D. & Laas, A. 2017. Atmospheric stilling leads to prolonged thermal stratification in a large shallow polymictic lake. Climatic Change, doi:10.1007/s10584- 017-1909-0. Yacobi, Y.Z. 2006. Temporal and vertical variation of chlorophyll a concentration, phytoplankton photosynthetic activity and light attenuation in Lake Kinneret: possibilities and limitations for simulation by remote sensing. Journal of Plankton Research, 28: 725–736. Yang, Y., Pettersson, K. & Padisak J. 2016. Repetitive baselines of phytoplankton succession in an unstably stratified temperate lake

100 (Lake Erken, Sweden): a long-term analysis. Hydrobiologia, 764: 211–227, doi: 10.1007/s10750-015-2314-1 Yoshida, T., Sekino, T., Genkai-Kato, M., Logacheva, N. P., Bondarenko, N. A., Kawabata, Z., Khodzher, T. V., Melnik, N. G., Hino, S., Nozaki, K., Nishimura, Y., Nagata, T., Higashi, M. & Nakanishi, M. 2003. Seasonal dynamics of primary production in the pelagic zone of southern Lake Baikal. Limnology (The Japanese Society of Limnology), 4: 53–62. Zhang, Y., Qin, B., Zhu, G., Zhang, L. & Yang, L. 2007. Chromophoric dissolved organic matter (CDOM) absorption characteristics in relation to fluorescence in Lake Taihu, China, a large shallow subtropical lake. Hydrobiologia, 581: 43–52. Zingel, P., Nõges, P., Tuvikene, L., Feldmann, T., Järvalt, A., Tõnno, I., Agasild, H., Tammert, H., Luup, H., Salujõe, J. & Nõges, T. 2006. Ecological processes in macrophyte- and phytoplankton-dominated shallow lakes. Proceedings of the Estonian Academy of Sciences. Biology, Ecology, 55(4): 280−307.

101 SUMMARY IN ESTONIAN

LOODUSLIKU MUUTLIKKUSE MÕJU MADALATE JÄRVEDE ÖKOSEISUNDI HINDAMISELE

21. sajandi algusest on nii Euroopas kui suuremas osas maailmast hakatud rohkem tähelepanu pöörama vee ökosüsteemide talitlusele tervikuna kui nende „tervise“ mõõdupuule ning nende seisundi hindamisel on füüsikaliste ja keemiliste kriteeriumide kõrval saanud määrava tähtsuse ökoloogilised kriteeriumid. Euroopa Parlamendi ja Euroopa Nõukogu direktiiv 2000/60/EC (Directive, 2000), tuntud kui Veepoliitika Raamdirektiiv (VRD), on alusdokument, mis korraldab veega seotud problemaatikat Euroopas. Pinnavete osas seab VRD põhieesmärgiks saavutada kõigi pinnaveekogude hea ökoloogiline seisund. Veekogude puhul, kus seda ei õnnestunud saavutada seatud tähtajaks, 2015. aastaks, on püstitatud uueks tähtajaks 2027. aasta.

1960ndate aastate algul teaduslikel eesmärkidel alanud Võrtsjärve ja Peipsi seisundi uuringud jätkuvad seniajani kuise sammuga – Peipsil maist oktoobrini ja Võrtsjärvel aastaringselt. Lisandunud eesmärkide ja uurituse taseme kasvu tõttu on mõlema järve pikaajaline andmebaas muutunud järjest komplekssemaks. Kui 1960ndatel ja 1970ndatel aastatel oli vaja peamiselt taustateavet kalandusuuringutele, siis aastakümnete jooksul on jõutud toitumissuhete selgitamiseni (nt Nõges et al., 1998; Blank et al., 2010; Agasild et al., 2014), ökoloogilise seisundi trendide tuvastamiseni (Nõges et al., 1997; Nõges & Laugaste, 1998) ja kliimamuutuste mõjude analüüsimiseni (Nõges et al., 2007c; Nõges et al., 2010a; Nõges & Nõges, 2014). Uuringute edenedes on seireprogrammi lisandunud uusi analüütilisi meetodeid ja näitajaid, mis on kasvatanud seirekulusid ja on isegi osaliselt muutnud tulemuste tõlgendamise raskemaks. Sellele vaatamata on seire suur komplekssus ja pikkade aegridade jätkamine kui kohanduva seire tunnusjoon seireprogrammide peamiseks tugevuseks, võimaldades protsessipõhist lähenemist ökoloogilistes uuringutes, samaaegselt täites VRD erinevaid eesmärke ja pakkudes tänuväärt materjali kliimauuringuteks.

Veekvaliteedi seire esmane eesmärk on tuvastada veekogude seisundi halvenemist näitavad märgid (Anttila et al., 2008) ja tema lõppeesmärk on meetmete kavandamine veekogude vähemalt hea seisundi säilitamiseks

102 ja sellest halvema seisundi parandamiseks, milleks on vaja põhjalikke teadmisi ökosüsteemi talitlusest (Peeters et al., 2009; Søndergaard et al., 2016).

Iga veekeskkonda iseloomustavat keemilist, füüsikalist ja bioloogilist näitajat iseloomustab nelja liiki muutlikkus – ruumiline muutlikkus, ajaline muutlikkus, nende koosmõju, mis avaldub kindla ajalise mõju erinevas toimes sõltuvalt tema toimumiskohast, ja mõõtemääramatusest põhjustatud muutlikkus (Ellis & Adriaenssens, 2006). Igal nimetatud muutlikkuse liigil on omakorda erinevad skaalad, näiteks ajalisi erinevusi võib vaadelda ööpäevases, aastaajalises, aastatevahelises või pikaajalises lõikes; ruumilised erinevused võivad ilmneda veekogu erinevate osade vahel või sügavuti, veesamba piires. Erinevused ilmnevad ka suurema või väiksema piirkonna erinevate veekogude vahel. Kirjeldatud skaalad ei ütle aga midagi ajalise ja ruumilise muutlikkuse põhjuste kohta, mis on seire peamine tähelepanu keskpunkt. Põhjuslikkuse sissetoomine seiresse ja keskkonnamõjude hindamisse omistab osale muutlikkusest „signaali“ tähenduse ja jätab ülejäänu „müraks“. Olenevalt probleemipüstitusest võivad „müra“ ja „signaal“ isegi oma kohad vahetada. Näiteks VRD mõistes on „signaaliks“ kohalikud inimtegevuse mõjud ja „müraks“ üldisem looduslik muutlikkus, kuid kliimamuutuste uurimise puhul on tähendused vastupidised. VRD mõistes on tähtis leida arvukate füüsikaliste, keemiliste ja bioloogiliste näitajate hulgast sellised, mis on tundlikud inimmõju suhtes, kuid looduslike tegurite poolt vähemõjutatavad. Raske on leida hästitoimivaid indikaatoreid selliste veeökosüsteemide seisundi hindamiseks, mida inimtegevuse kõrval mõjutavad üheaegselt mitmed looduslikud tegurid nagu suured veetaseme kõikumised ja sesoonsus. Näiteks Võrtsjärves näitavad kõik tavapärased veekvaliteedi mõõdikud madala vee perioodidel halvemat olukorda ka muutumatu inimmõju tingimustes (Nõges et al., 2003b) ja nii on raske vahet teha inimtekkeliste ja looduslike muutuste vahel. Samuti tuleb arvestada, et muutunud keskkonnatingimuste mõju ei tarvitse avalduda kohe, vaid alles järgmisel aastal või mõnel juhul aastate pärast.

Peamiselt Võrtsjärve ja Peipsi pikaajaliste andmeridade analüüsile tuginedes on doktoritöös analüüsitud loodusliku muutlikkuse mõju suurte madalate järvede ökoloogilise seisundi hindamisele. Töö eesmärgid ja hüpoteesid olid:

103 1) Selgitada, kuidas hinnata madalate järvede ökoloogilist seisundit, kui (H1) veetaseme kõikumise ja sesoonsuse kui looduslike tegurite mõju traditsioonilistele veekvaliteedi näitajatele nagu taimtoitainete ja klorofüll a kontsentratsioonid, fütoplanktoni biomass ja vee läbipaistvus, ületab inimmõjust tingitud muutlikkuse (I).

2) Meie hüpoteesid olid, et (H2) Võrtsjärve avaveelise osa ökoloogilise seisundi näitajates on piirkondadevahelised erinevused väiksemad kui aastatevahelised erinevused (II) ja (H3) Võrtsjärve sügavas kohas asuv peamine seirekoht on piisav hindamaks järve kogu avaveelise osa ökoloogist seisundit, kuid järve kitsast lõunaosa tuleks seirata ja selle seisundit hinnata eraldi (II). Hüpoteeside tõestamiseks hindasime seisundinäitajate ruumilist muutlikkust Võrtsjärve erinevate piirkondade vahel. Selleks:

a) piiritlesime järve peamise, avaveelise osa ja sellest hüdrokeemiliste, hüdrooptiliste ja fütoplanktoni näitajate poolest oluliselt erinevad järveosad, b) analüüsisime muutlikkuse jagunemist järve avaveelises osas ruumilise (proovikohad) ja ajalise (aastad) komponendi vahel, c) hindasime temperatuuri ja veetaseme kõikumise mõju seiratavate näitajate varieeruvusele, d) hindasime Võrtsjärve peamise seirekoha esinduslikkust järve avaveelise osa iseloomustamisel ja kogu järve akvatooriumi haaravate uuringute jätkamise põhjendatust.

3) Selgitamaks, kas (H4) madalate järvede ökosüsteemides peegelduvad nii valgla majandamisega kui kliimamuutustega seotud märgid, analüüsisime Võrtsjärve ja Peipsi klimaatilisi, hüdroloogilisi ja toiteainete koormuse trende ning hindasime nende mõju nende suurte ja madalate järvede veekvaliteedile (III).

4) Analüüsisime veealuse valgusvälja ja seda kujundavate komponentide, eriti kollase aine ehk värvunud lahustunud orgaanilise aine (CDOM) mõju suurte madalate järvede primaarproduktsioonile ja oletasime, et (H5) lahustunud orgaanilise aine (DOC) kontsentratsiooni suurtes madalates järvedes saab hästi hinnata tema värvunud komponendi (CDOM) alusel (IV, V, VI).

104 Töö tulemusena leiti järgmist:

1. Meie analüüs kinnitas hüpoteesi nr.1, näidates, et Võrtsjärves on veetaseme loodusliku kõikumise ja sesoonsuse mõju veekvaliteedi tavanäitajatele nagu taimtoitainete ja klorofüll a kontsentratsioonid, fütoplanktoni biomass ja vee läbipaistvus suurem kui inimtegevusest põhjustatud mõjud neile näitajatele. Füüsikaliselt tugevalt mõjutatud veekogudes nagu seda on madalad järved, võib klimaatiliste tegurite toime varjutada inimmõju. Selliste järvede seisundinäitajate väärtusi on vaja korrigeerida looduslikku varieeruvust põhjustavate tegurite suhtes „signaali“/“müra“ suhte suurendamiseks ja antava seisundihinnangu usaldusväärsuse tõstmiseks (I). 2. Vee pinnatemperatuuri ja veetaseme muutlikkus selgitavad ligikaudu poole Võrtsjärves seiratavate ökoloogilise seisundi näitajate koguvarieeruvusest. Võrtsjärvel kasutatav näitajate valik sobib hästi kliimamõjude uurimiseks, näidates tugevaid seoseid veetaseme ja -temperatuuriga, mis iseloomustavad sademete- ja soojusrežiimi varieeruvust (II). 3. Võrtsjärve 10 seirekoha andmete analüüs näitas, et järve saab jagada kaheks erinevaks osaks: 1) kahe proovikohaga iseloomustatav kitsas lõunaosa, mis moodustab ainult 2% järve kogupindalast ja on tugevalt Väikese Emajõe mõju all ja 2) homogeenne avaveeline osa, mida iseloomustavad 8 seirekohta. Avaveelises osas ületas ökoloogilise seisundi näitajate aastatevaheline varieeruvus märkimisväärselt seirekohtadevahelisi erinevusi, mis olid enamiku näitajate osas ebaolulised (II). Tulemused kinnitavad hüpoteesi nr 2. 4. Võrtsjärve avaveelise osa 8 seirekoha võrdluses on sügavas kohas asuva peamise seirekoha näitajate varieeruvus väikseim ja see koht iseloomustab hästi kogu avaveelist osa ehk rohkem kui 90% järve akvatooriumist. Sellega leidis meie hüpotees nr 3 kinnitust. Lisaks mõõdetavate parameetrite vähimale varieeruvusele ja heale esinduslikkusele, mille tagavad sissevoolude ning kohalike mõjufaktorite puudumine, on seirepunkti vaieldamatuks vooruseks tema soodne asukoht laborite lähedal (II). 5. Suurte madalate järvede Võrtsjärve ja Peipsi vee kvaliteet on suurel määral mõjutatud kliimast – kas otseselt veetemperatuuri kaudu või kaudselt, läbi reguleerimata veetaseme loodusliku kõikumise. Toiteainete dünaamika madalates järvedes muutub vastavalt nende

105 sissekandele, näidates ökosüsteemi võimet reageerida tundlikult valgla majandamisele (III). Need järeldused kinnitavad hüpoteesi nr 4. 6. Meie analüüs ei kinnitanud hüpoteesi nr 5, näidates, et DOC ja CDOM seose sesoonse muutlikkuse tõttu ei saa lahustunud orgaanilise aine koguhulka tema värvunud komponendi alusel Võrtsjärves kuigi hästi hinnata. Analüüsi põhjal osutus CDOM siiski heaks lähendiks eutroofsete järvede DOC mustrite hindamiseks suuremate alade kaugseirel (IV). 7. Bio-optilised mudelarvutused on ajakulukale 14C meetodile heaks alternatiiviks primaarproduktsiooni hindamisel. Mudel, kus sisendandmetena kasutatakse pidevalt mõõdetud veepinnale langevat fotosünteetiliselt aktiiivset kiirgust ning episoodiliselt mõõdetud klorofüll a ja spektraalse vertikaalse nõrgenemiskoefitsiendi andmeid, võimaldab hästi hinnata primaarproduktsiooni muutlikkust järvedes (V). 8. Peipsi järves on optiliselt aktiivsete ainete suurest ruumilisest ja ajalisest varieeruvusest tulenevalt suured erinevused valguse neeldumises erinevates spektripiirkondades. Allohtoonse värvunud lahustunud orgaanilise aine kontsentratsioon ja omadused sõltuvad sissevoolavate jõgede vooluhulkade ja järve veetaseme sesoonsetest muutustest (VI).

106 ACKNOWLEDGEMENTS

From the bottom of my heart, I thank my supervisor Dr Peeter Nõges – I am grateful for your valuable time and wisdom, and for your long friendship. My greatest thanks go to my boss and good friend Prof. Tiina Nõges for her versatile support. My heartfelt gratitude to dear co-authors of my publications! Sincere thanks to all my dear colleagues – my teachers and co-wanderers during all these years. I am indebted to my pre-opponent Dr Fabio Ercoli for his valuable comments on my thesis! Special thanks to my friend Edgar Karofeld – I am grateful for your persistent belief in me.

Very special thanks go to my family: to my dear husband Arvo and our by now grown-up children Mari, Mikk, Tõnn and Kaur – I consider you as my first PhD which I am really proud of. And I always experienced unconditional support from my mother and father – thank you!

This research was supported by Estonian Target Financed Projects SF03962480s03, SF0712699s05, SF0170011s08 and SF0180009s11 of the Estonian Ministry of Education and Research, by the Estonian Science Foundation grants 5738, 7156, 7600, 8729 and 9102, and by the Institutional research grant IUT21-2 from Estonian Science Agency. Co-financing from the Swedish National Space Board Grant 132/03, the EU Framework programme projects CLIME, REFRESH, WISER and MARS, Estonian Environmental Agency through state monitoring programme and the Swiss Grant for Programme „Enhancing public environmental monitoring capacities“ is greatly appreciated.

107

ORIGINAL PUBLICATIONS Tuvikene, L., Nõges, T. & Nõges, P. 2011. Why do phytoplankton species composition and “traditional” water quality parameters indicate different ecological status of a large shallow lake? Hydrobiologia, 660(1): 3–15, doi: 10.1007/s10750-010-0414-5. Why do phytoplankton species composition and “traditional” water quality parameters indicate different ecological status of a large shallow lake?

Hydrobiologia The International Journal of Aquatic Sciences

ISSN 0018-8158 Volume 660 Number 1

Hydrobiologia (2010) 660:3-15 DOI 10.1007/ s10750-010-0414-5

1 23

111 Your article is protected by copyright and all rights are held exclusively by Springer Science+Business Media B.V.. This e-offprint is for personal use only and shall not be self- archived in electronic repositories. If you wish to self-archive your work, please use the accepted author’s version for posting to your own website or your institution’s repository. You may further deposit the accepted author’s version on a funder’s repository at a funder’s request, provided it is not made publicly available until 12 months after publication.

1 23

112 Author's personal copy

Hydrobiologia (2011) 660:3–15 DOI 10.1007/s10750-010-0414-5

EUROPEAN LARGE LAKES II

Why do phytoplankton species composition and ‘‘traditional’’ water quality parameters indicate different ecological status of a large shallow lake?

Lea Tuvikene • Tiina No˜ges • Peeter No˜ges

Received: 4 November 2009 / Revised: 31 July 2010 / Accepted: 10 August 2010 / Published online: 24 August 2010 Ó Springer Science+Business Media B.V. 2010

Abstract Long-term data on phytoplankton species status, distinguished a period of fast eutrophication in composition in large and shallow Lake Vo˜rtsja¨rv the ﬁrst half of the 1970s (not captured by the indicated a sharp deterioration of the ecological status phytoplankton species index), a fast improvement at at the end of the 1970s. The more traditional water the end of the 1970s (when the species index showed quality indicators, such as the concentrations of deterioration) followed by a continuous deterioration nutrients and chlorophyll a, phytoplankton biomass, trend (when the species index remained rather and Secchi depth, failed to capture this tipping point constant). The causes of this inconsistency are or even showed an improvement of the status at that discussed in the light of the alternative stable states time. As the shift coincided with a large increase of theory and the priority of biotic indicators stipulated the lake’s water level (WL), we hypothesized that by the EU Water Framework Directive. direct effect of the changing WL on traditional water quality indicators might have blurred the picture. We Keywords Water Framework Directive Á removed statistically the direct effect of the WL and Phytoplankton taxonomic index Trophic state the seasonality from the traditional water quality indicators Long-term data HighÁ natural variability Á Á Á indicators in order to minimize the effects of natural Alternative stable states variability. The average of the standardised water quality indicators, used as a proxy for the ecological

Guest editors: T. Blenckner, T. No¨ges, L. Tranvik, Introduction K. Pettersson, R. Naddafi / European Large Lakes II. Vulnerability of large lake ecosystems - Monitoring, management and measures The principal legislative tool in the field of water policy in Europe, the Water Framework Directive L. Tuvikene (&) T. No˜ges P. No˜ges (WFD; Directive, 2000) defines the status of water Á Á Centre for Limnology, Institute of Agricultural bodies by the extent of anthropogenically derived and Environmental Sciences, Estonian University of Life Sciences, 61117 Rannu, Tartumaa, Estonia deviation from the reference conditions, i.e. condi- e-mail: [email protected] tions that should occur at sites of any particular type in the absence of human impact. Still, the latter is T. No˜ges often overshadowed by the natural variability appear- European Commission, Joint Research Centre, Institute for Environment and Sustainability, Via Enrico Fermi ing at longer or shorter time scales (No˜ges et al., 2749, 21027 Ispra, Italy 2007a, b). The following natural factors may have

123

113 Author's personal copy

4 Hydrobiologia (2011) 660:3–15 remarkable influence on parameters commonly used 1990). Natural variability that in principle should to assess ecological status: belong to reference conditions exceeds often the variability caused by anthropogenic factors. As the (1) Diurnal changes in the physical, chemical and target variables of both the types of variability largely biological variables, some of which are regular overlap in natural waters, it becomes difficult to due to daily cycle (e.g. photosynthesis and disentangle their effects that add a large uncertainty respiration), still comprise a stochastic compo- to the status assessment. To illustrate this problem, nent deriving from meteorology. we have chosen as an example the large and shallow (2) Seasonal changes take place with well-known Lake Vo˜rtsja¨rv (Estonia, 270 km2, average depth regularity from year to year. The randomness is 2.8 m), famous by its huge natural variability caused added to them due to differences in the mete- by fluctuating WLs. orological conditions between years, causing Seasonal and inter-annual WL fluctuations exceed- deviations in seasonality and phenology. ing 3 m and modifying the intensity of sediment (3) The prolonged changes in atmospheric circula- resuspension, strongly influence all water quality tion patterns such as the North Atlantic Oscil- parameters in Vo˜rtsja¨rv (No˜ges & Ja¨rvet, 1995; lation (NAO; Hurrell, 1995) or El Nino No˜ges & No˜ges, 1998, 1999). The lake has been Southern Oscillation (ENSO; Philander, 1990) identified as an individual type in the Estonian affect through different mechanisms physical, classification of lake status for state monitoring. In a chemical and biological properties of water recent study where four phytoplankton taxonomic bodies. Both have shown to cause large fluctu- indices were tested on the long 44-year time series of ations in affluence and lake water levels WLs phytoplankton data from this lake (No˜ges et al., (Rodo´ et al., 1997). 2010a), all indices showed a unidirectional deteriora- To avoid the effect of diurnal differences in the tion of the lake’s ecological status with a major data, the monitoring programmes usually determine stepwise change occurring in 1979. To some extent, it a certain sampling time for a water body (Loftis was a surprise for us as the nutrient loadings had a et al., 1991). To remove seasonality from data, decreasing trend since the end of 1980s (No˜ges et al., several methods of time series analysis such as 2010b), and we expected to see an improvement also seasonal decomposition (Cleveland & Tiao, 1976) in the biotic indices. The traditionally monitored and seasonal smoothing (Gardner, 1985) exist. water quality indicators, such as the concentrations of However, in practice, large amounts of monitoring total phosphorus, total nitrogen, chlorophyll a, phy- data are omitted due to seasonality problems and the toplankton biomass, and Secchi depth, failed to status assessment of water bodies is often based on capture the sudden deterioration at the end of the data of a single season only, mostly summer. Using 1970s or even showed an improvement of the status at seasonal averages requires regular and comparable that time. As all these indicators are strongly data coverage for all years. The decadal scale influenced by changes in WL, we hypothesized that periodicities caused by atmospheric circulation pat- this factor could blur the picture and cause the terns can be revealed only in really long-term contradictory results not allowing a consistent esti- monitoring data and still no standard solution exists mation of the ecological status of this lake. to eliminate them. To clarify the trophic state history of the lake, we The sensitivity of lakes to natural variability applied the traditional water quality indicators in factors depends strongly on their morphometry, and which we statistically removed the direct effect of the the role of physical drivers like wind and WL in changing WL and the seasonality in order to mini- controlling the ecosystem processes increases with mize the effects of natural variability. In this article, increasing lake area and decreasing depth (No˜ges, we analyse what caused the inconsistency in the 2009). While large and shallow polymictic lakes are status assessments based on traditional water quality extremely sensitive to natural physical drivers indicators, on one hand, and on phytoplankton (Scheffer, 2004; No˜ges et al., 2007a, b; Scheffer & species index, on the other hand, and whether it van Nes, 2007), in deep lakes, the in-lake biological was caused by ignoring the dynamic reference caused and chemical factors prevail (Tilzer & Serruya, by natural variability in this lake.

123

114 Author's personal copy

Hydrobiologia (2011) 660:3–15 5

Materials and methods the deviation of the WL from the monthly mean value (DWL). We applied this correction step We used long-term data on most common trophic only for those variables and months for which state parameters for lakes such as phytoplankton the relationship with WL was statistically biomass (BM), chlorophyll a (Chl), total nitrogen significant, otherwise the variables remained (TN), total phosphorus (TP), and Secchi depth unchanged. (S) measured at the main monitoring station in Lake (3) To remove the seasonality from the data series, Vo˜rtsja¨rv from May to October. Time series of BM we calculated first the monthly and the seasonal and S started from the year 1965, that of Chl from (May–October) averages of all corrected vari- 1982 and those of nutrients from 1983. The assess- ables for all years. We excluded those years ment system was created by the following steps: from the seasonal average calculations for which less than four of the six monthly values (1) Data standardization. Concentrations (TN, TP, were available. Second, we calculated regres- Chl) and biomass were Ln-transformed in order sions to derive the seasonal average from single to achieve normal distribution of the variables. monthly values. Non-significant regressions In the case of Secchi depth, we used the were omitted. Finally, a new seasonal average reciprocal (1/S) to make it increase with the was calculated based on those derived from trophic state like the other variables, and used monthly values. In this way, more correct then the natural logarithm values as the best seasonal averages could be calculated for those approximation to normal distribution. Further, years for which only few measurements were the transformed variables were used as normal available. As a result, the new corrected data ones in all steps of the analysis (regressions, consisted of approximations of seasonal average averaging) and the exponents (i.e. the geometric values corrected for the effect of WL changes. means of the initial variables) were calculated (4) To define reference conditions and set the quality only at the end to indicate the quality class class boundaries for individual variables, we boundaries in their original units. supposed, based on historical data and expert (2) To correct the data for the effects of changing opinions (No˜ges et al., 2001; No˜ges, 2003; WL, we used monthly linear regressions No˜ges & No˜ges, 2006), that for most of the between Ln-transformed variables and the WL period studied, the lake has been deviating only of the sampling day (Fig. 1). We corrected the slightly from the reference conditions described variable values by adding DY corresponding to in the second decade of the twentieth century (Mu¨hlen, 1918; von zur Mu¨hlen & Schneider, 1920), i.e., has been in ‘‘good’’ status according to the WFD (Directive, 2000). Hence, we considered that at least one-half of the parameter values (25th to 75th percentile) should indicate ‘‘good’’ status. In line with the guidance document on reference conditions (CIS, 2003), we supposed that the median of the ‘‘high’’ class values (values below the 25th percentile) should describe the site-specific reference conditions for the lake. The upper 10% was considered to characterize the ‘‘poor status’’ (Fig. 2). The analysis of historical biotic changes (No˜ges et al., 2001) showed that the lake has never fallen to ‘‘bad’’ status, which, according to WFD (Directive, 2000), is defined by the ‘‘ absence Fig. 1 Principal scheme of correcting water quality parame- … ters for water level changes based on linear regression. Grey of large portions of the relevant biological area marks the data distribution communities normally associated with the

123

115 Author's personal copy

6 Hydrobiologia (2011) 660:3–15

Table 1 Characteristics of monthly regressions between logarithmic values of trophic state indicators (Y) and water level (cm) Y Month rP Intercept Slope

LnBM 5 -0.262 0.051 n.s. 2.895 -0.0040 Fig. 2 Classification criteria based on the statistical distribu- LnBM 6 -0.437 0.001 3.279 -0.0064 tion of geometric mean values of the water quality metrics in Lake Vo˜rtsja¨rv for the period studied. It was supposed that half LnBM 7 -0.522 0.000 3.318 -0.0077 of the measurements (25th–75th percentile) should indicate the LnBM 8 -0.409 0.002 3.222 -0.0065 ‘‘good’’ quality class mostly identified by previous studies. LnBM 9 -0.383 0.003 3.210 -0.0043 Reference conditions were defined as the median of the ‘‘high’’ LnBM 10 -0.330 0.014 3.238 -0.0037 class values LnChl 5 -0.249 0.085 n.s. 3.617 -0.0020 LnChl 6 -0.111 0.463 n.s. 3.540 -0.0015 surface water body type under undisturbed LnChl 7 -0.414 0.004 3.630 -0.0041 conditions’’. LnChl 8 -0.362 0.008 3.714 -0.0040 (5) For the final assessment, values from 1 to 4 LnChl 9 -0.161 0.286 n.s. 3.786 -0.0016 were given, correspondingly, to the classes LnChl 10 -0.308 0.037 3.900 -0.0025 ‘‘high’’ (H), ‘‘good’’ (G), ‘‘moderate’’ (M), and LnTP 5 0.124 0.427 n.s. 3.658 0.0010 ‘‘poor’’ (P). These quality scores of all variables LnTP 6 -0.223 0.157 n.s. 3.868 -0.0024 were averaged for the final assessment where LnTP 7 -0.108 0.496 n.s. 3.954 -0.0012 the values of 1.5, 2.5, and 3.5 served as the H/G, LnTP 8 -0.317 0.036 3.967 -0.0052 G/M, and M/P boundaries. Given that the LnTP 9 -0.455 0.002 4.042 -0.0053 variables were not independent, averaging of LnTP 10 -0.301 0.050 4.096 -0.0031 them was considered as a pragmatic step to get a LnTN 5 0.234 0.152 n.s. 0.083 0.0019 summarising status estimate. We expressed the LnTN 6 0.296 0.063 n.s. -0.154 0.0024 uncertainty of the final estimate as the standard deviation (STDev) of the quality class number. LnTN 7 0.107 0.530 n.s. -0.161 0.0015 LnTN 8 -0.071 0.666 n.s. -0.013 -0.0008 To characterize the phytoplankton species compo- LnTN 9 -0.271 0.075 n.s. 0.104 -0.0022 sition, the PTSI index (Mischke et al., 2008) was LnTN 10 -0.023 0.889 n.s. 0.128 -0.0002 calculated for all samples from the study period. Ln1/S 5 -0.184 0.115 n.s. 0.091 -0.0010 We used the chemical oxygen demand (CODMn) Ln1/S 6 -0.405 0.001 0.444 -0.0028 measured in two periods, 1968–1977 and 1998–2008, Ln1/S 7 -0.454 0.000 0.425 -0.0026 as a rough overarching proxy for water colour to find Ln1/S 8 -0.593 0.000 0.442 -0.0035 out possible long-term impacts on light conditions. Ln1/S 9 -0.612 0.000 0.496 -0.0047 We used the Mann–Kendall test (Kendall, 1938) Ln1/S 10 -0.507 0.000 0.405 -0.0034 for trend analysis and the Worsley likelihood ratio test (Worsley, 1979) to find step-changes in the n.s. Non-significant relationships series. Units of initial measurements: phytoplankton biomass (BM) and total nitrogen (TN), mg l-1; total phosphorus (TP) and chlorophyll (Chl), lgl-1; Secchi depth (S), m

Results significant relationship with WL; in addition, the The values of all selected trophic state indicators regression was non-significant for LnChl in June and except TN were significantly related to the WL during September, and for LnTP in June and July. In cases several months of the vegetation period (Table 1), when the relationship was non-significant, we used while the relationships were the strongest for the the measured values in the following steps, otherwise Secchi depth. All significant relationships were neg- the measured values were corrected according to the ative, i.e. the lake looked significantly more eutrophic regression to correspond to the long-term average WL at lower WLs. In May, none of the parameters had a of the month.

123

116 Author's personal copy

Hydrobiologia (2011) 660:3–15 7

On average, the relationships between monthly ranges of all variables (Fig. 3). The total variance values of trophic indicators and their seasonal decreased fivefold for LnBM, LnChl, and LnTN, averages (Table 2) were strongest in August and nine-fold for LnTP, and 21-fold for Ln1/S. September and weakest in May. The relationship was Despite the unchanged long-term averages, both non-significant for TP in July and for Secchi depth in corrections affected substantially the monthly values May. (Fig. 4) and the seasonal averages (Fig. 5) as exem- As a result of the two corrections/transformations, plified with phytoplankton biomass. the mean values of the full time series did not change The correction for the effects of WL changes significantly (P of t test for means between 0.3 and followed closely the dynamics of the mean WL for 0.9 for different variables). The correction for WL May–October (Fig. 5) correcting the BM by up to did not significantly change the total variability of the 23% down in low-water years and up to 41% up in time series (P of t test for variance between 0.1 and high-water years. The seasonality removal had even 0.9 for different variables). The correction for stronger effect ranging from 28% down to 64% up. In seasonality diminished considerably the variability nearly 70% of the cases, the seasonality removal

Table 2 Characteristics of XY rP Intercept Slope monthly regressions between logarithmic values LnBM May LnBM May–October 0.703 0.0000 1.832 0.438 of trophic state indicators (variable names as in LnBM June LnBM May–October 0.589 0.0001 1.516 0.512 Table 1) corrected for water LnBM July LnBM May–October 0.736 0.0000 1.091 0.635 level changes where LnBM August LnBM May–October 0.824 0.0000 1.334 0.485 appropriate (X) and their LnBM September LnBM May–October 0.784 0.0000 1.045 0.602 seasonal mean values (Y) LnBM October LnBM May–October 0.779 0.0000 0.941 0.626 LnChl May LnChla May–October 0.699 0.0001 1.175 0.702 LnChl June LnChla May–October 0.703 0.0001 2.415 0.344 LnChl July LnChla May–October 0.840 0.0000 1.198 0.710 LnChl August LnChla May–October 0.646 0.0003 1.601 0.556 LnChl September LnChla May–October 0.773 0.0000 1.545 0.540 LnChl October LnChla May–October 0.781 0.0000 1.306 0.593 LnTP May LnTP May–October 0.481 0.0174 2.301 0.404 LnTP June LnTP May–October 0.711 0.0001 2.458 0.372 LnTP July LnTP May–October 0.292 0.1657 n.s. 3.156 0.166 LnTP August LnTP May–October 0.648 0.0005 2.395 0.381 LnTP September LnTP May–October 0.805 0.0000 1.915 0.491 LnTP October LnTP May–October 0.604 0.0014 2.493 0.326 LnTN May LnTN May–October 0.409 0.0426 -0.029 0.316 LnTN June LnTN May–October 0.600 0.0012 0.063 0.454 LnTN July LnTN May–October 0.786 0.0000 0.070 0.397 LnTN August LnTN May–October 0.873 0.0000 0.093 0.563 LnTN September LnTN May–October 0.600 0.0012 0.052 0.426 LnTN October LnTN May–October 0.708 0.0001 -0.014 0.532 Ln1/S May Ln1/S May–October 0.287 0.0802 n.s. 0.251 0.141 Ln1/S June Ln1/S May–October 0.585 0.0001 0.187 0.279 Ln1/S July Ln1/S May–October 0.458 0.0084 0.174 0.287 Ln1/S August Ln1/S May–October 0.543 0.0007 0.137 0.386 Ln1/S September Ln1/S May–October 0.659 0.0000 0.141 0.330 n.s. Non-signiﬁcant Ln1/S October Ln1/S May–October 0.580 0.0003 0.162 0.244 relationships

123

117 Author's personal copy

8 Hydrobiologia (2011) 660:3–15

Fig. 3 The effect of corrections on the mean values and variability ranges of different trophic state variables. A uncorrected variables, B variables corrected for the effect of water level changes (was not done for TN because of missing relationship), C variables corrected both for water level changes and seasonality. Variable names as in Table 1

corrected the data to the same direction with the slight decreasing trend (P \ 0.05). The corrected correction for WL giving a summary effect between biomass series (Fig. 6) indicated an abrupt decrease 41% down and 65% up. from 1978 to 1979 followed by a significant For the whole 44-year period analyzed, there was (P \ 0.01) increasing trend since that. The decrease no significant trend in the WL (Fig. 5); however, in phytoplankton biomass was reflected also in there was a stepwise jump between 1977 to 1978 improved Secchi transparency (corrected series) (Worsley test, P \ 0.01). Since 1978, the WL had a although the water turned brown in this period

123

118 Author's personal copy

Hydrobiologia (2011) 660:3–15 9 obviously due to humic substances carried into the nutrients corrected for the WL effects did not show lake during the rainy 1978. The corrected Secchi any significant trend. depth had an increasing trend from 1981 to 1992 and The CODMn levels measured in the period decreased since that (P \ 0.01; note that in the Fig. 6 1968–1977 (10.6 ± 3.2 mg O l-1) were significantly there is shown 1/S to make S increase with the trophic (P \ 0.01) lower compared with those in 1998–2008 state like other variables). The time series of total (13.0 ± 1.5). Within both periods, CODMn had a highly significant (P \ 0.01) increasing trend. The average trophic state proxy index calculated by applying the class boundaries of trophic state indicators (Table 3) to the long-term data (Fig. 7) showed two distinctive periods in the changes of the ecological status of the lake: the initial fast eutrophication in the 1970s (assessed only on the basis of BM and S) followed by a temporary improvement in 1979–1980 and a worsening trend afterwards. The latter trend for the whole period was clearly seen also Fig. 4 The effect of corrections on monthly mean values of without using the additional data (TP, TN, Chl) for measured phytoplankton biomass posterior years, and was not caused by the difference

Fig. 5 The effect of corrections on seasonal mean values of measured phytoplankton biomass and the changes in May– October mean water level

Fig. 6 Long-term changes of the May–October mean values of the common trophic state variables corrected for the changes in the water level in Lake Vo˜rtsja¨rv. Variable names as in Table 1

123

119 Author's personal copy

10 Hydrobiologia (2011) 660:3–15

Table 3 Reference conditions and class boundaries for the in data availability. Since 2001, the data show an geometric mean values of trophic state parameters (variable accelerated deterioration of the status, although also names as in Table 1) corrected for seasonal variability and the the uncertainty of the estimate has increased in this effect of water level changes in Lake Vo˜rtsja¨rv period. Percentile RC 25% 75% 90% The German PTSI index (Mischke et al., 2008); Class boundary H/G G/M M/P however, showed different results for the period since BM, g m-3 13.4 14.5 23.2 27.2 1979 (Fig. 8) when due to the change in dominant Chl, lgl-1 30.4 32.3 43.6 48.8 species (earlier Planktolyngbya limnetica (Lemm.) TP, lgl-1 39.3 42.8 50.2 54.2 was replaced by Limnothrix redekei (Goor) Meffert TN, mg l-1 0.9 1.0 1.2 1.3 and L. planktonica (Wołosz.) Meffert.). The taxo- S, m 0.90 0.82 0.74 0.70 nomic index revealed an irreversible drop of the ecological status of the lake. Hence, our hypothesis Reference conditions were calculated as the median value of the ‘‘high’’ class that the inconsistency of the assessment results based

Fig. 7 Long-term changes in the ecological status of Lake Vo˜rtsja¨rv based on trophic state parameters and corrected for seasonality and water level changes (A), and the standard deviation of the ﬁnal evaluation (B). Variable names as in Table 1

Fig. 8 Long-term changes in the average status index based on traditional trophic state parameters (grey line) and the phytoplankton taxonomy-based PTSI index (box and whiskers) in Lake Vo˜rtsja¨rv

123

120 Author's personal copy

Hydrobiologia (2011) 660:3–15 11 on phytoplankton taxa and traditional trophic state carried out for years with data gaps for relevant indicators was caused by the effect of changing WLs, months while, on the other hand, part of the seasonal was disproved. data is omitted from the analysis. In principle, if seasonal monitoring is carried out, the calculated indicator should make use of all available data. Discussion Reducing of datasets is acceptable only if proven that a more precise indicator can be obtained from a The ultimate goal of lake monitoring should be the subset of data (Carstensen, 2007). The method we establishment of a coherent and comprehensive choose for seasonality removal allows using data overview of the ecological and chemical status of from all months in which the variable has a signif- lakes in nearly real-time regime to enable water icant relationship with the seasonal mean value. This managers to take measures if the conditions deteri- way of standardizing has three main advantages: orate as a result of human impact. Selecting those (i) each single measurement can be equally used for among the multitude of measurable physical, chem- the assessment purposes, (ii) more correct seasonal ical and biological parameters that reliably reflect the averages can be calculated for years with data gaps, effects of human activities, but remain insensitive to and (iii) the obtained values are ecologically mean- extraneous conditions is a step of extraordinary ingful as estimates for the vegetation period mean importance (Karr & Chu, 1997). Given the often value. strong effects to multiple causal factors operating The calculated summary index revealed two simultaneously, it is unlikely to find such suitable distinct periods in the ecological status of Vo˜rtsja¨rv: metrics in ecosystems strongly physically controlled the initial fast eutrophication of the lake in the 1970s by natural factors. Different ways have been used to followed by a temporary improvement after the high disentangle the effects of natural and anthropogenic water year of 1978 (Fig. 6) and a slow continuous variability in long-term data ranging from simple de- deterioration trend. The German PTSI index (Mischke trending (George et al., 2004), applying coefficients et al., 2008) distinguished even more clearly the same for residual adjustment (Reist, 1986), statistical periods (Fig. 7). Paradoxically, the taxonomic index partialling of the effects of other factors and variance indicated a drop of the ecological status from ‘‘good’’ decomposition (Rodrı´guez & Magnan, 1995), to to ‘‘moderate’’ to ‘‘bad’’ where the average trophic linear (Carstensen & Henriksen, 2009) and nonlinear index indicated an improvement. Although the scale regression models (Massol et al., 2007). of the PTSI index was not adapted for Vo˜rtsja¨rv and For simplicity and transparency reasons, we the ‘‘bad’’ status was obviously exaggerated, the selected the linear adjustment method to statistically divergence of the two indices remains a fact. Indeed, account for the effects of the changing WL on the during the fast nutrient enrichment in the first half of common trophic state variables. In this way, the the 1970s, phytoplankton biomass increased, but no direct effect caused by WLs deviating from the long- remarkable changes were observed in the species term monthly averages could be eliminated while composition. The increase in the WL by more than retaining the meaningfulness and original dimensions 1 m in 1978 changed the conditions considerably. Due of the variables. The considerable correction of single to less resuspension, the nutrient concentrations and seasonal mean values by more than ±40% shows the phytoplankton biomass decreased and water became high importance of WL changes in shaping these less turbid. However, the nearly 50% increase in the variables. The correction dampened efficiently the mixing depth due to higher water probably strength- effect of the record low WL in 1996 when most of the ened light limitation and created favourable condi- common water quality indicators were far out of tions for two highly shade tolerant Limnothrix species, range indicating strong hypertrophy (No˜ges & No˜ges, which replaced the previous dominating cyanobacte- 1999). rium P. limnetica. The significant difference between

To cope with the strong seasonality of trophic state the CODMn values measured in the 1970s and in the variables, often values of only some months, seasonal later period suggests a possible increase in water averages or seasonal maxima are used for the status colour that could be an additional supporting factor assessment. In this way, correct assessment cannot be explaining the success of Limnothrix. Oscillatoriales,

123

121 Author's personal copy

12 Hydrobiologia (2011) 660:3–15 either by Planktothrix or Limnothrix species dominate with which a taxon can be detected within its proposed in several very shallow polytrophic lakes creating preference range, is a common practice in bioindica- steady-state communities, while a lower phosphorus tion (e.g., Zelinka & Marvan, 1961; Gervais et al., requirements and a lower light tolerance are the 1999; Fre´de´ric & Luc, 2005). This factor gives high possible advantages for Limnothrix over Planktothrix weights to the specific indicator species which some- (Ru¨cker et al., 1997). Though the WL changed as times may be not numerous, like it is often the case with much from 1996 to 1998 as in 1977–1978 (Fig. 5), the character species used in phytocoenology. The weight- very low level in 1996 did not change permanently the ing factors, however, overemphasize the importance of species composition (No˜ges & No˜ges, 1999). This dominant species. In our case the PTSI index indicated indicates that the dynamics of cyanobacteria is not a sharp deterioration of the status resulting from the well understood. switch of the dominants to more shade tolerant species. The German index is calculated as biomass On the other hand, this change indicates to an weighed average product of trophic scores showing aggravation of the situation, as the domination of the average trophic preferences of indicator species, Limnothrix species means a switching of the system to and stenoecy factors showing the indicator power of a steady state. This steady state presents a self-induced the species (i.e. how specifically they indicate the habitat, in which competitors fail because of low-light given trophic state). A comparison of the values of conditions are reproduced by the dominants based on these parameters applied in the German index for the efficient exploitation of nutrient resources (Mischke & dominating cyanobacteria species in Vo˜rtsja¨rv shows Nixdorf, 2003). Still, it may be not so clear for Lake that they indicate nearly the same trophic status but Vo˜rtsja¨rv where the influence of resuspension and the stenoecy factors differ substantially (Table 4). humic substances on light conditions is also remark- Consequently, the jump in the index can be explained able and the dominance of Limnothrix species is by the lake having reached a high enough trophic strongly influenced by external conditions as precip- state where all three species could coexist, and a itation, temperature and WL. strengthening of light limitation at which the more There remains the question, should we trust one of specifically adapted Limnothrix species outcompeted these indices to correctly interpret the lake monitor- the more eurytrophic P. limnetica. No reversal has ing results or further research is needed to develop a still happened as this would require a reduction in suitable assessment system for this complicated both light limitation and trophic state. As at the physically driven lake. Both of the indices have their moment of the change in dominants triggered by light advantages and disadvantages. Due to strong resil- limitation, no further increase in trophic status was ience of the phytoplankton community, the taxonomy required (the trophic score of L. redekei is even lower based index did not almost change during the fast than that of P. limnetica), the traditional trophic state eutrophication in the beginning of the 1970s, neither indicators did not reflect it. On the contrary, they demonstrated clear patterns in the period after the showed an improvement caused by the dilution of change of the dominants. The big change occurring in substances and weaker sediment resuspension caused 1979 was brought upon not so much by a change in by the higher WL. the trophic state but expressed the tipping point The use of a stenoecy (=specificity) factor, i.e., a evoked by a disturbance—the sudden increase of the weighting factor that describes the degree of constancy WL. Such behaviour of the index throws doubt upon the popular belief that phytoplankton provides a good Table 4 Trophic scores and stenoecy factors of the cyano- indication of lake trophic state and respond quickly bacteria species dominating in Lake Vo˜rtsja¨rv as applied in the and predictably to changes in nutrient status (e.g. German phytoplankton index for polymictic lowland lakes Murphy et al., 2002). Also Kaiblinger et al. (2009) (Mischke et al., 2008) who tested phytoplankton indices on three large peri- Species Trophic score Stenoecy factor alpine lakes to analyze their suitability for trophic classification, concluded that the indices were only Planktolyngbya limnetica 5.18 1 appropriate to roughly distinguish lakes of different Limnothrix redekei 4.68 2 water quality but were not sensitive enough to track Limnothrix planktonica 5.40 4 changes that occur within a lake. In several shallow

123

122 Author's personal copy

Hydrobiologia (2011) 660:3–15 13 lakes reduction of nutrient loads has not led to increase the signal/noise ratio and decrease the discernible recovery. The main causes of delay are uncertainty of the estimate. This step is of utmost phosphorus storage and its subsequent release from importance for strongly, physically driven systems sediments (Van Liere & Gulati, 1992). Internal such as shallow lakes where the large variation of loading and the mechanism of hysteresis, i.e. less driving factors may not only mask the effect of nutrient concentrations are needed for recovering the human pressures but also the effect of restoration previous better equilibrium state than it was at the measures. time of its decline (Scheffer et al., 1993, 1997; In cases of high uncertainty of the status estimate Jeppesen et al., 2007), offer an explanation for the (different metrics or different quality elements give resistance of cyanobacteria dominance in shallow controversial results), the causes of the controversy lakes to restoration efforts by means of nutrient load should be analysed and the more appropriate metrics reduction. In that way, phytoplankton indices really and elements selected before averaging the results or reflect the ecological effect of human impact, which applying the ‘‘one out – all out’’ principle. can last much longer than the direct impact itself. In As alternative stable states may exist in water addition, usually several other factors like weather bodies making some of the biological response conditions or spatial heterogeneity, and even the indicators highly resilient, pressure indicators (e.g. cyanobacterial dominance itself, play role in devel- nutrient concentrations) could be used in parallel to oping an alternative regime (Scheffer & van Nes, reflect the trends and get a more awarding system for 2007). assessing the managerial efforts. The index based on traditional trophic state variables and corrected for the simultaneous WL Acknowledgments The study was supported by Estonian changes demonstrated an evolution of the trophic target funding project SF 0170011508, by grant 7600 from Estonian Science Foundation, and RE 201—the Estonian state more consistent with the expert opinion. Environmental Monitoring Programme. However, it did not capture the tipping point occurring in phytoplankton, one of the biggest changes ever observed in the plankton community of Vo˜rtsja¨rv, and without phytoplankton composition References data the actual status would have been misinterpreted. From the point of view of WFD, which gives Carstensen, J., 2007. Statistical principles for ecological status a priority to biological indicators, the situation is classification of Water Framework Directive monitoring data. Marine Pollution Bulletin 55: 3–15. clear: preceding nutrient loadings caused a pressure Carstensen, J. & P. Henriksen, 2009. Phytoplankton biomass on the ecosystem resulting in a regime shift when a response to nitrogen inputs: a method for WFD boundary sudden disturbance (high WL) broke the resilience of setting applied to Danish coastal waters. Hydrobiologia the system. The new degraded stable state has 633: 137–149. CIS, 2003. River and Lakes—Typology, Reference Conditions demonstrated strong resistance to remediation mea- and Classification Systems. Common Implementation sures. However, this interpretation has been put Strategy for the Water Framework Directive (2000/60/ together after a critical comparison of both indices EC). Guidance document 10, European Commission: and the initial monitoring data. For a more consistent 86 pp [available on internet at http://circa.europa.eu]. Cleveland, W. P. & G. C. Tiao, 1976. Decomposition of sea- assessment of the ecological status of the lake, other sonal time series: a model for the Census X-11 program. biological elements such as fish, macrophytes and Journal of the American Statistical Association 71: macrozoobenthos should be included in the assess- 581–587. ment as suggested by the WFD. Directive, 2000. Directive 2000/60/EC of the European Par- liament and of the council of 23 October 2000 establishing a framework for community action in the field of water policy. Official Journal of the European Commu- Conclusions nities L327: 1–72. Fre´de´ric, R. & E. Luc, 2005. Role of diatoms in the application of the Water Framework Directive in Europe: recent devel- We suggest that correcting of the metric values used opments in France. Diatomededelingen 28–29: 31–35 in status assessment for the effects of natural [available on internet at http://membres.multimania.fr/ variability factors is a necessary step in order to rimetfrederic/Rimet-diatomededelingen-2005.pdf].

123

123 Author's personal copy

14 Hydrobiologia (2011) 660:3–15

Gardner, E. S. Jr., 1985. Exponential smoothing: the state of No˜ges, P. & A. Ja¨rvet, 1995. Water level control over light the art. Journal of Forecasting 4: 1–28. conditions in shallow lakes. Report Series in Geophys- George, D. G., S. C. Maberly & D. P. Hewitt, 2004. The ics. Report No 32. University of Helsinki, Helsinki: influence of the North Atlantic Oscillation on the physical, 81–92. chemical and biological characteristics of four lakes in the No˜ges, P. & T. No˜ges, 1998. The effect of fluctuating water English Lake District. Freshwater Biology 49: 760–774. level on the ecosystem of Lake Vo˜rtsja¨rv, Central Estonia. Gervais, F., S. Berger, I. Scho¨nfelder & R. Rusche, 1999. Basic In Proceedings of the Estonian Academy of Sciences. limnological characteristics of the shallow eutrophic lake Biology, Ecology 47: 98-113. Grimnitzsee (Brandenburg, Germany) Limnologica. No˜ges, T. & P. No˜ges, 1999. The effect of extreme water level Ecology and Management of Inland Waters 29: 105–119. decrease on hydrochemistry and phytoplankton in a Hurrell, J. W., 1995. Decadal trends in the North Atlantic shallow eutrophic lake. Hydrobiologia 409: 277–283. Oscillation regional temperatures and precipitation. No˜ges, P. & T. No˜ges, 2006. Indicators and criteria to assess Science 269: 676–679. ecological status of the large shallow temperate polymictic Jeppesen, E., M. Søndergaard, M. Meerhoff, T. L. Lauridsen & lakes Peipsi (Estonia/Russia) and Vo˜rtsja¨rv (Estonia). J. P. Jensen, 2007. Shallow lake restoration by nutrient Boreal Environment Research 11: 67–80. loading reduction—some recent findings and challenges No˜ges, P., T. Feldmann, J. Haberman, A. Ja¨rvalt, A. Kangur, ahead. Hydrobiologia 584: 239–252. K. Kangur, H. Timm, T. Timm, A. Tuvikene & P. Zingel, Kaiblinger, C., O. Anneville, R. Tadonleke, F. Rimet, J. 2001. Deviation of Lake Vo˜rtsja¨rv from its pristine status C. Druart, J. Guillard & M. T. Dokulil, 2009. Central- documented 90 years ago. In: Proceedings of the 9th European water quality indices applied to long-term data International Conference on the Conservation and Man- from peri-alpine lakes: test and possible improvements. agement of Lakes, Session 5: 221–224. Hydrobiologia 633: 67–74. No˜ges, P., W. Van de Bund, A. C. Cardoso & A. S. Heiskanen, Karr, J. R. & E. W. Chu, 1997. Biological Monitoring and 2007a. Impact of climatic variability on parameters used Assessment: Using Multimetric Indexes Effectively. EPA in typology and ecological quality assessment of surface 235-R07-001. University of Washington, Seattle: 149 pp. waters–implications on the Water Framework Directive. Kendall, M. G., 1938. A new measure of rank correlation. Hydrobiologia 584: 373–379. Biometrika 30: 81–93. No˜ges, T., A. Ja¨rvet, A. Kisand, R. Laugaste, E. Loigu, B. Loftis, J. C., G. B. McBride & J. C. Ellis, 1991. Considerations Skakalski & P. No˜ges, 2007b. Reaction of large and of scale in water quality monitoring and data analysis. shallow lakes Peipsi and Vo˜rtsja¨rv to the changes of Journal of the American Water Resources Association 27: nutrient loading. Hydrobiologia 584: 253–264. 255–264. No˜ges, P., U. Mischke, R. Laugaste & A. G. Solimini, 2010a. Massol, F., P. David, D. Gerdeaux & P. Jarne, 2007. The Analysis of changes over 44 years in the phytoplankton of influence of trophic status and large-scale climatic change Lake Vo˜rtsja¨rv (Estonia): the effect of nutrients, climate on the structure of fish communities in Perialpine lakes. and the investigator on phytoplankton-based water quality Journal of Animal Ecology 76: 538–551. indices. Hydrobiologia 646: 33–48. Mischke, U. & B. Nixdorf, 2003. Equilibrium phase conditions No˜ges, T., L. Tuvikene & P. No˜ges, 2010b. Contemporary in shallow German lakes: how Cyanoprokaryota species trends of temperature, nutrient loading and water quality establish a steady state phase in late summer. Hydrobio- in large lakes Peipsi and Vo˜rtsja¨rv, Estonia. Aquatic logia 502: 123–132. Ecosystem Health and Management 13: 143–153. Mischke, U., U. Riedmu¨ller, E. Hoehn, I. Scho¨nfelder & Philander, S. G., 1990. El Ninõ, La Ninã and the Southern B. Nixdorf, 2008. Description of the German system for Oscillation. Academic Press, San Diego: 293 pp. phytoplankton-based assessment of lakes for implemen- Reist, J. D., 1986. An empirical evaluation of coefficients used tation of the EU Water Framework Directive (WFD). In in residual and allometric adjustment of size covariation. Mischke, U. & B. Nixdorf (eds), Gewa¨serreport (Nr. 10), Canadian Journal of Zoology 64: 1363–1368. Brandenburg Technical University of Cottbus, Cottbus. Rodo´, X., E. Baert & F. A. Comin, 1997. Variations in seasonal ISBN 978-3-940471-06-2, BTUC-AR 2/2008: 117–146. rainfall in Southern Europe during the present century: Murphy, K. J., M. P. Kennedy, V. McCarthy, M. T. O´ ’Hare, K. relationships with the North Atlantic Oscillation and the Irvine & C. Adams, 2002. A Review of Ecology Based El Ninõ-Southern Oscillation. Climate Dynamics 13: Classification Systems for Standing Freshwaters. SNIF- 275–284. FER Project Number W(99)65. Environment Agency Rodrı´guez, M. A. & P. Magnan, 1995. Application of multi- R&D Technical Report: E1-091/TR. variate analyses in studies of the organization and struc- No˜ges, P. 2003. Milliseks hinnata Vo˜rtsja¨rve praegust o¨ko- ture of fish and invertebrate communities. Aquatic loogilist seisundit fu¨toplanktonis 90 aasta jooksul toimu- Sciences—Research Across Boundaries 57: 199–216. nud muutuste po˜hjal? (How to evaluate the ecological Ru¨cker, J., C. Wiedner & P. Zippel, 1997. Factors controlling the quality of Lake Vo˜rtsja¨rv on the basis of phytoplankton dominance of Planktothrix agardhii and Limnothrix rede- changes during 90 years?) In Mo¨ls, T., J. Haberman, L. kei in eutrophic shallow lakes. Hydrobiologia 342(343): Kongo, E. Kukk. & E. Mo¨ls (eds), Eesti LUS-i Aast- 107–115. araamat 81, Nordon, Tartu: 60–81. Scheffer, M., 2004. Ecology of Shallow Lakes. Kluwer Aca- No˜ges, T., 2009. Relationships between lake morphometry, demic publishers, London: 374 pp. geographic location and water quality parameters of Scheffer, M. & E. H. van Nes, 2007. Shallow lakes theory European lakes. Hydrobiologia 633: 33–43. revisited: various alternative regimes driven by climate,

123

124 Author's personal copy

Hydrobiologia (2011) 660:3–15 15

nutrients, depth and lake size. Hydrobiologia 584: von zur Mu¨hlen, L. 1918. Zur Geologie und Hu¨drologie des 455–466. Wirtsjerwsees. Abhandlungen der Ko¨niglichen Preussis- Scheffer, M., S. H. Hosper, M. L. Meijer, B. Moss & chen Geologischen Landesanstalt, Neue Folge, Berlin: E. Jeppesen, 1993. Alternative equilibria in shallow lakes. 83. Trends in Ecology & Evolution 8: 275–279. von zur Mu¨hlen, M. & G. Schneider, 1920. Der See Wirzjerw Scheffer, M., S. Rinaldi, A. Gragnani, L. R. Mur & E. H. van in Livland. Archiv fu¨r die Naturkunde des Ostbaltikums Nes, 1997. On the dominance of filamentous cyanobac- 14: 1–156. teria in shallow, turbid lakes. Ecology 78: 272–282. Worsley, K. J., 1979. On the likelihood ratio test for a shift in Tilzer, M. M. & C. Serruya (eds), 1990. Large lakes: location of normal populations. Journal of the American Ecological Structure and Function. Springer-Verlag, Berlin. Statistical Association 74: 365–367. Van Liere, L. & R. D. Gulati, 1992. Restoration and recovery Zelinka, M. & P. Marvan, 1961. Zur Pra¨zisierung der biolog- of shallow eutrophic lake ecosystems in The Netherlands: ischen Klassifikation der Reinheit fliessender Gewa¨sser. epilogue. Hydrobiologia 233: 283–287. Archiv fu¨r Hydrobiologie 57: 389–407.

123

125

II Nõges, P. & Tuvikene, L. 2012. Spatial and annual variability of environmental and phytoplankton indicators in Lake Võrtsjärv: implications for water quality monitoring. Estonian Journal of Ecology, 61(4): 227–246, doi: 10.3176/eco.2012.4.          

                

    

                      

           

                                                                                                                                                                                                                                                    

             



                                                                                     



129     

                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                         



130       

                                                                                                                                                                                           

    

                                                                                                                                         

                                                                                                                                    



131     

           

                                                                                                                                                                                             



132       

        

                                    

 ′″      ′″  ′″      ′″  ′″      ′″  ′″        ′″   ′″      ′″  ′″      ′″   ′″       ′″  ′″      ′″  ′″        ′″    ′″       ′″



                                                                                                                         



133     

         

                                                                                                                                                                                                                                                                                                       



134       

 

                                                                                                                                 

                                                                                                                                                                                                                                                                                                                                                                                      



135     

                                           



                                                                                                                                                                                                                                                                                                                                                                                                                                 



136       

                                       

               

   

                                                                                                                                                                                                                                                                                                                                                                                                                                         



137     

                                      

                                           

                                   



138       

                             

                                   β                                                              β  β  

         

         

               

                

                                                                         

       

         



139     

         β  β  

       

        

            

       

       

       

       

       

       

       

        

              

      

       

                                                                                          



                                                                                                                     



140       

                                       



141     

    

                                                                                                                                                                                                                                                                                                                                                                          

      

                                                                                



142       

                                                                                                                                                                                                                        

  

                                                                                                                                                                            

       ±                                                                                           



143     

                                                                                                                                                                                                                                                                        



•                                                                                     •                                                          •                   



144       

                                        •                                                  •                                                                                                          



                                                               



                                                                                                                                                   



145     

                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                     



146       

                                                                                                                                                                                                                                                                                                                                             

            

    

                                                                            



147     

                                                                                                 



148 III Nõges, T., Tuvikene, L. & Nõges, P. 2010. Contemporary trends of temperature, nutrient loading and water quality in large lakes Peipsi and Võrtsjärv, Estonia. Journal of Aquatic Ecosystem Health and Management, 13(2): 143–153, doi: 10.1080/14634981003788987 Contemporary trends of temperature, nutrient loading, and water quality in large Lakes Peipsi and Vortsj˜ arv,¨ Estonia 1,2 1 1,2 Tiina Noges,˜ ∗ Lea Tuvikene, and Peeter Noges˜ 1Centre for Limnology, Institute of Agricultural and Environmental Sciences, Estonian University of Life Sciences, 61120 Rannu, Tartumaa, Estonia 2Rural, Water and Ecosystem Resources Unit, TP-483, Institute for Environment and Sustainability, EC Joint Research Centre, Via E. Fermi 2749m 21027 Ispra (VA), Italy ∗Corresponding author: [email protected]

From 1961–2004, surface water temperature in large and shallow Lakes Peipsi and Vortsj˜ arv¨ in Estonia increased significantly in April and August; respectively 0.37–0.75 and 0.32–0.42 degrees per decade reflecting the changes in air temperature. The average annual amount of precipitation in the catchment increased significantly. Reflecting practices in agriculture and wastewater treatment, nutrient loadings to the lakes increased rapidly in the 1980s and decreased again in the early 1990s. As total nitrogen (TN) loading decreased faster than total phosphorus (TP) loading, the TN/TP ratio in the loadings decreased. Both the increased temperature and low TN/TP ratio favoured the development of cyanobacteria blooms in Lake Peipsi. In Vortsj˜ arv,¨ where the TN/TP mass ratio is about two times higher than in Peipsi, blooms did not occur. Recently, the TN/TP ratio has shown a tendency of increase in both lakes suggesting a certain reduction of blooms to be expected also in Lake Peipsi. Nutrient dynamics in the lakes followed the changes in loadings, showing the ability of shallow lake ecosystems to react sensitively to changes in catchment management as well as in climate.

Keywords: surface water temperature, nutrients, longterm trends

Introduction tributing 41% of its water discharge (Noges˜ P.et al., 2008); the River Emajogi˜ is the outflow of Vortsj˜ arv¨ Lake Peipsi (3,555 km2, mean depth 7.1 m) lo- and the largest river discharging into Peipsi from the cated on the Estonian-Russian border is the largest Estonian part of its watershed (Noges˜ T.et al., 2005). international lake in Europe. The volume of wa- The majority of phosphorus and nitrogen com- ter in Peipsi is 25 km3and its mean residence time pounds (>80%) are carried into Lake Peipsi by the is about 2 years (Jaani, 2001). The River Narva Rivers Velikaya and Emajogi,˜ the first carrying bi- contains the outflow of Lake Peipsi and runs into ologically treated sewage from the Russian town of the Gulf of Finland. Lake Vortsj˜ arv¨ (270 km2, Pskov ( 200,000 inhabitants) and the latter receiv- mean depth 2.8 m) is the largest lake belonging en- ing the∼ effluent of the wastewater treatment plant tirely to Estonia (Figure 1). The watershed of Lake of the Estonian town of Tartu ( 100,000 inhabi- Vortsj˜ arv¨ (3,104 km2) lies within the Lake Peipsi tants). The treatment plant has been∼ in operation in catchment (47,800 km2). Riverine transport is the Tartu since the end of 1998; previously the untreated major pathway for nutrient input into both lakes. wastewater from Tartu was discharged directly to the Vaike¨ Emajogi˜ is the largest inflow of Vortsj˜ arv¨ con- river.

143

C Aquatic Ecosystem Health & Management, 13(2):143–153, 2010. Copyright � 2010 AEHMS. ISSN: 1463-4988 print / 1539-4077 online DOI: 10.1080/14634981003788987

151 144 Noges˜ et al. / Aquatic Ecosystem Health and Management 13 (2010) 143–153

Figure 1. Location map of sampling stations in Lakes Peipsi and Vortsj˜ arv¨ and their catchment area.

Long-term dynamics in nutrient and organic by milder and more economic practices in the early matter loadings reflect changes in agricultural 1990s. The knowledge about the nature of climate practices and climate. In the past 40–50 years, change and its impacts is far from being complete. European lakes have been subjected to eutrophica- The response of lake ecosystems to climate change tion caused by the increased supply of nutrients. is complex and often unpredictable as different lake As a result of decreased loadings since the late types and ecosystem components may react in indi- 1980s or early 1990s, the ecological status of many vidual ways. For instance, increasing water temper- lakes has improved considerably (Jeppesen et al., ature in spring in Estonian inland waters shifted the 2005b). The structure of changes in the nutrient sup- spawning time of bream (Abramis brama) to 10 days ply has been rather different in the Western Euro- earlier, while no shift in the spawning time of roach pean countries compared to the countries of the for- (Rutilus rutilus) has been observed. As a conse- mer Soviet Union. In the western countries (e.g. in quence, the difference between spawning times of Denmark), extensive efforts have been made to im- roach and bream decreased from 22 to 13 days and prove wastewater purification and to reduce phos- the difference in average temperatures at the onset phorus concentration in the effluent (Kronvang of spawning by about 3◦C (Noges˜ P. and Jarvet,¨ et al., 2005; Jeppesen et al., 2005a). In Eastern 2005). Kangur et al. (2007) described a negative ef- Europe (e.g. in the Baltic States), a major nitro- fect of high temperature on the abundance of the gen reduction occurred as a result of the collapse of smelt stock in Lake Peipsi. Climate change may the Soviet-type agriculture (Jarvet,¨ 2001; Blinova, either diminish or magnify the effects of eutrophi- 2001; Juhna and Klavins, 2001). In Estonia the op- cation. ulent use of fertilizers including swine slurry in the In the present paper we review the contemporary 1970s and 1980s, which lead to high phosphate, am- climatic, hydrological and loading trends observed monium, and organic matter loadings, was replaced in Lakes Vortsj˜ arv,¨ Peipsi and their catchments, and

152 Noges˜ et al. / Aquatic Ecosystem Health and Management 13 (2010) 143–153 145 their impact on water quality in these large and shal- Kolmogorov-Smirnov test, all indicated data were low lakes. normally distributed (p > 0.2). Changes in a time series can occur either gradually or abruptly (a step Material and methods change). A non-parametric Mann-Kendall test for gradual trends and a Cumulative deviation test for We analysed long-term trends of air tempera- step changes were applied (Trend C1.0.2 Catch- ture (AT) and precipitation (PR) at two meteoro- ment Modelling toolkit, CRC for Catchment Hy- logical stations, Tiirikoja and Tartu, located within drology), the details of these methods are described the catchment of Lake Peipsi (Figure 1). The trend by Kundzewicz and Robson (2004). of surface water temperature (SWT) was studied in both Lakes Vortsj˜ arv¨ and Peipsi; the trends of water Results and Discussion discharge (Q), the concentrations of total nitrogen (TN) and total phosphorus (TP), and the TN/TP ratio Trends of air temperature and lake surface were studied in the rivers Vaike¨ Emajogi,˜ Emajogi,˜ water temperature and Narva. AT, SWT and Q were measured by the Estonian At Tartu and Tiirikoja stations the average an- Institute of Hydrology and Meteorology according nual AT increased by 0.4◦ per decade in 1961–2004 to the guidelines of World Meteorological Organisa- (Noges˜ T., 2009), and the increasing trend of AT tion (WMO, 1994; 2008). TN and TP were analysed (Figure 2A) had a step change in 1987 (Table 1). at the accredited laboratory Tartu Environmental Water temperature and ice cover are most directly Research Centre Ltd according to the methods de- affected by climate forcing. The existence of consis- scribed by Grasshoff et al. (1983) and corresponding tent trends has been demonstrated for water temper- ISO standards. The data were gathered in the state atures in rivers (Hari et al., 2006) and lakes (Living- monitoring program of Estonian surface water qual- stone, 2003; Straile et al., 2003; Arhonditsis et al., ity and obtained through the Centre of Information 2004; Coats et al., 2006; Dokulil et al., 2006) and and Technology at Estonian Ministry of Environ- for ice phenology (Weyhenmeyer et al., 2005). In ment. the large and shallow Lakes Peipsi and Vortsj˜ ar¨ v, We analysed the data from the river stations for daily SWT was largely dependent on daily AT dur- which the longest time series of nutrient concentra- ing ice-free period (R2 0.86, n > 3000). From tions were available (see Tables 1 and 2 for the length 1961–2004, SWT in Peipsi= and Vortsj˜ arv¨ increased of time series). In the River Vaike¨ Emajogi˜ the sta- significantly in April and August, by 0.37–0.75 and tion V.Emajogi035˜ is situated downstream from the 0.32–0.42 degrees per decade, respectively (Noges˜ town of Valga, the main polluter of this river. In the T., 2009). In Vortsj˜ arv¨ and in the southern part of River Emajogi˜ the station Emajogi101˜ is situated at Peipsi at Mehikoorma, the increasing trend of SWT the outflow from Lake Vortsj˜ arv¨ and was assumed to in August (Figure 2B) had a significant step change characterise the integrated water quality in this lake; in 1989 (Table 1) which did not occur in northern the station Emajogi016˜ is situated downstream from Peipsi at Mustvee (Noges˜ T., 2009). the town Tartu, the main polluter of this river. In the River Narva, the station Narva077 is situated at the Trends and relationships of precipitations outflow from Lake Peipsi and was assumed to char- and river hydrology acterise the integrated water quality in this lake. To characterise nutrient concentrations in Lake Peipsi The average annual PR had significant increas- main basin (Peipsi s.s.; Figure 1) we averaged the ing trends at Tartu and Tiirikoja in 1961–2004 data from three pelagic sampling stations; for Lake (Figure 2C) with a step change in 1976 in Tartu and Vortsj˜ arv¨ the data from one pelagic sampling station in 1977 in Tiirikoja (Table 1). On a monthly basis, was used (see Figure 1 for location of the sampling PR increased significantly for January-February stations). (p < 0.01), March (p < 0.1), and June (p < 0.05). To reveal long-term trends, we analysed the The average annual discharge in the River Vaike¨ monthly and yearly average values of AT, Q and Emajogi˜ was significantly positively related to the SWT calculated from the daily measurements, and average annual PR in Tartu (Figure 2D). From yearly average values of TN and TP calculated 1961–2004, the average annual Q in this river had from the monthly measurements. Based on the a significant increasing trend (Figure 3A) with a step

153 Mean Min Max STD 1987–2004 5.321987–2004 2.56 5.82 6.67 3.20 1.01 7.17 0.98 1977–2004 1.751976–2004 1.23 1.76 2.13 1.25 0.26 2.30 0.31 1991–2007 25.1 16.2 46.5 6.6 1989–2004 18.91989–2004 15.2 18.9 22.2 16.7 1.9 21.0 1.2 1977–2007 8.801991–2007 4.52 1.681991–2007 15.05 1.29 0.071 1.95 0.056 2.50 0.089 0.31 0.009 < 0.01 1961–1986< 0.01 4.29 1961–1986 2.75< 0.01 4.72 1961–1976 6.18 3.32 1.44 1.01 6.60 0.98 0.94 1.77 0.22 < 0.01 1961–1975 1.45 0.98 1.96 0.25 < 0.05 1961–1988 17.9 15.8 19.8 1.1 < 0.01 1961–1976 6.74< 0.01 1986–1990 4.66 3.66 13.49 2.68< 0.01 2.07 1986–1990 5.43 54.8 1.09 47.3 68.9 8.3 Step < 0.01 1987 < 0.01 1987 < 0.01 1977 < 0.01 1976 < 0.05< 0.01 1989 NS 1961–1988 17.9 15.2 20.4 1.3 < 0.01 1977 < 0.01 1991 < 0.01< 0.05 1991 NS 1986–1990 0.091 0.065 0.150 0.034 ¨ arv and Peipsi and their watershed. ˜ ortsj ˜ ˜ ˜ ˜ ogi035 1961–2007 increasing ogi035 1986–2007 decreasing ogi035 1986–2007 decreasing ogi035 1986–2007 decreasing ¨ arv 1961–2004 increasing Site Period Trend Trend p year Step p ˜ ortsj V.Emaj V.Emaj V.Emaj − 1 − 1 − 1 C PeipsiCV Mustvee 1961–2004 increasing s − 1 − 1 ◦ ◦ − 1 CC Tiirikoja Tartu 1961–2004 increasing 1961–2004 increasing 3 ◦ ◦ Trends of average annual air temperature (AT), precipitations (PR), discharge (Q), concentration of total nitrogen (TN) and phosphorus (TP) and TN/ TP ratio, mm day mm day mg mg Table 1. and surface water temperature (SWT) in August in Lakes V Parameter ATyear, PRyear, Tiirikoja 1961–2004 increasing PRyear,SWTaug, Tartu 1961–2004 increasing SWTaug, Qyear, m ATyear, TNyear, mg l TPyear, mg l TN/TPyear, V.Emaj 146

154 Noges˜ et al. / Aquatic Ecosystem Health and Management 13 (2010) 143–153 147

Table 2. River and lake water chemistry in 1992–2007 in the watershed of Vortsj˜ arv¨ and Peipsi.

n Mean Min Max STD M-K trend, p

V. Emajogi035˜ 1 TN, mg l− 189 1.62 0.50 4.60 0.74 decreasing, <0.1 1 TP, mg l− 189 0.072 0.032 0.160 0.022 decreasing, <0.01 1 TN/TP, mg mg− 189 23.7 3.6 75.0 11.4 NS Vortsj˜ arv¨ 1 TN, mg l− 315 1.40 0.34 3.70 0.60 NS 1 TP, mg l− 317 0.050 0.010 0.330 0.026 NS 1 TN/TP, mg mg− 315 34.3 4.4 142.9 22.4 NS Emajogi101˜ 1 TN, mg l− 155 1.31 0.46 5.00 0.63 NS 1 TP, mg l− 155 0.046 0.010 0.190 0.027 NS 1 TN/TP, mg mg− 155 37.4 2.6 120.0 25.7 NS Emajogi016˜ 1 TN, mg l− 192 2.08 0.55 5.60 0.94 NS 1 TP, mg l− 192 0.077 0.028 0.220 0.027 decreasing, <0.1 1 TN/TP, mg mg− 192 30.6 6.9 132.4 19.1 NS Peipsi s.s. 1 TN, mg l− 274 0.65 0.25 2.10 0.25 NS 1 TP, mg l− 274 0.042 0.008 0.240 0.021 NS 1 TN/TP, mg mg− 274 18.2 3.2 51.8 9.0 NS Narva077 1 TN, mg l− 153 0.62 0.07 1.40 0.26 NS 1 TP, mg l− 153 0.038 0.010 0.140 0.024 NS 1 TN/TP, mg mg− 152 22.3 2.4 138.0 18.5 NS change in 1977 (Table 1). On a monthly basis, Q had lakes, the yearly average TN and TP concentrations significant increasing trends for January (p < 0.05), and the TN/TP ratio did not reveal any significant February (p < 0.01), March (p < 0.01), and June trend (Table 2). The mean nutrient concentrations in (p < 0.05). the lakes and their outflows were well coupled.

Trends of river water chemistry Trends of riverine nutrient loads In the River Vaike¨ Emajogi˜ the yearly average In the 1980s, the riverine loading of nutrients into concentrations of TN and TP and the TN/TP ratio Lakes Peipsi and Vortsj˜ arv¨ increased drastically; decreased significantly since 1986 (Figure 3B, C, while in the early 1990s a sharp decrease occurred, D). A step change of TN and TN/TP took place primarily in TN loadings. As TN loading decreased in 1991 (Table 1) at the time of the breakdown of faster than TP loading, the TN/TP ratio in the load- the Soviet Union. The application of large amounts ings decreased (Noges˜ T. et al., 2003a, 2005, 2007). of fertilizers in the 1980s was often accompanied Multiple regression analysis showed that the rate by substantial nutrient leakage into water bodies. of fertilization was the most important factor de- Compared to the levels at the end of the 1980s, only termining the nitrogen runoff from the catchment of 5–10% of N-, P- and K-mineral fertilizers and 30% Lake Vortsj˜ arv¨ (Jarvet¨ et al., 2002). Since the end of of the manure were applied to the fields at the end the 1980s, TP concentrations in the main inflow of of the 1990s (Jarvet¨ et al., 2002). In the post-Soviet Lake Vortsj˜ arv¨ and TP loadings showed a decreas- period, since 1992, the decrease in TN and TP con- ing trend (Table 1, Figure 4B) while the loadings of centrations continued in the River Vaike¨ Emajogi.˜ TN increased in the 2000s (Figure 4A) resulting in In the Emajogi˜ and Narva Rivers as well as in the an increased of the TN/TP loading ratio (Figure 4C).

155 148 Noges˜ et al. / Aquatic Ecosystem Health and Management 13 (2010) 143–153

8 23 Tiirikoja: r = 0.45, p = 0.002 A Peipsi Mustvee: r = 0.31, p = 0.038 B C 22

o Tartu: r = 0.51, p = 0.0004 Võrtsjärv:r = 0.44, p = 0.003 7 C

o 21

6 20

19 5 18

4 17 Tiirikoja 16 Tar tu August, in SWT Average Peipsi Mustvee 3 Võrtsjärv

Average annual air temperature, temperature, air annual Average 15

2 14 1955 1965 1975 1985 1995 2005 1955 1965 1975 1985 1995 2005 1960 1970 1980 1990 2000 2010 1960 1970 1980 1990 2000 2010

2.4 Tiirikoja: r = 0.42, p = 0.004 16

-1 Tartu: r = 0.50, p = 0.0006 C D

2.2 -1 r = 0.69, p < 0.0001 s 14 3 2.0 12 1.8

1.6 10

1.4 8

1.2 Tiirikoja 6 1.0 Tar tu m discharge, annual Average

Average annual precipitation, mm day precipitation, annual Average 4 0.8 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 1955 1965 1975 1985 1995 2005 1960 1970 1980 1990 2000 2010 Average annual precipitation, mm day -1 Figure 2. Linear trends of average annual air temperature (A) and precipitations (C) in the meteorological stations Tartu and Tiirikoja, surface water temperature (SWT) in Lakes Vortsj˜ arv¨ and Peipsi (B) in 1961–2004 (n 44), and the relationship between = precipitations in Tartu and the discharge in the River Vaike¨ Emajogi.˜ See site locations on Figure 1 and the results of Mann-Kendall test for trends and a Cumulative deviation test for step changes in Table 1.

In Peipsi, it is more difﬁcult to get an overview of but as part of this loading reaches the main basin the total nutrient loadings as the data is scarce from through Lake Lammij¨ arv,¨ we can indirectly judge a large part of the watershed belonging to Russia. To the dynamics of Russian loadings based on the sit- provide insight, we combined two sets of published uation in Lammij¨ arv.¨ TP concentration in this basin data (Figure 4D–F). Noges˜ T. et al. (2005) analysed increased until 2003 (Figure 5E) and a correspond- all available data on loadings from both the Estonian ing growth of phosphorus loading from the Russian and the Russian subcatchments and showed that the subcatchment could be supposed. It could be due total loading of TP in 2001 was almost equal to to the lack of maintenance of efﬁcient wastewa- the high loadings at the beginning of the 1980s and ter treatments in Pskov and other settlements after that TP loadings from Russia in mid 1990s even the collapse of the Soviet Union (Noges˜ T. et al., exceeded those of the 1980s. Loigu and Leisk (1996) 2007). In the last years, TP concentration started reported that in each of 1985–1989 the annual TN to decrease both in Lammij¨ arv¨ and Peipsi, indicat- and TP loads into Lake Peipsi were 55,350 and 1,163 ing some improvement of the situation. This could tonnes, respectively. According to Noges˜ T. et al. partly be explained by smaller loads observed in dry (2003a), the respective values in 1998 were 23,800 years. As our recent studies showed (Noges˜ P.et al., and 1,300 tonnes, and 60% of the total TP loading 2007), both components forming the load (water dis- entered the southernmost basin (Lake Pihkva) from charge and concentrations of substances) decreased the River Velikaya. Loigu et al. (2008) reported that in dry years. In Vortsj˜ arv¨ the dynamics of both N Peipsi received 15,650 tonnes of TN (Figure 4D) and P concentrations (Figure 5G, H) were consistent and 712 tonnes of TP (Figure 4E) in 2001–2005. to the loading history (Figure 4A, B), showing the The information on the loadings from the Rus- high sensitivity of large shallow lakes to changes in sian basin of Peipsi is quite scarce and inaccessible, human activity in the catchment.

156 Noges˜ et al. / Aquatic Ecosystem Health and Management 13 (2010) 143–153 149

16 12 A B

-1 14 10 s 3 Mean Min-Max 12 8 -1

6 10 r = 0.27, p = 0.06 TN, mg l 4 8

2 6 Average annual discharge, m discharge, annual Average

0 4 1986 1990 1994 1998 2002 2006 1960 1970 1980 1990 2000 2010 1988 1992 1996 2000 2004

0.30 200 0.28 C 180 D 0.26 0.24 Mean Min-Max 160 Mean Min-Max 0.22 140

0.20 -1

-1 0.18 120 0.16 100 0.14

T P, m g l 0.12 80

0.10 mg TN/TP, 60 0.08 0.06 40 0.04 20 0.02 0.00 0 1986 1990 1994 1998 2002 2006 1986 1990 1994 1998 2002 2006 1988 1992 1996 2000 2004 1988 1992 1996 2000 2004 Figure 3. Linear trend of the annual average discharge (A) in 1961–2007 (n 47) and long-term dynamics of the concentrations of = total nitrogen (TN) and phosphorus (TP), and TN/TP ratio in 1986–2007 in the River Vaike¨ Emajogi.˜ The results of Mann-Kendall test for trends and a Cumulative deviation test for step changes are presented in Tables 1 and 2.

Figure 4. Long-term dynamics of the loading of total nitrogen (TN) and phosphorus (TP) and TN/TP ratio in the loading of Lake Vortsj˜ arv¨ (A, B, C) and into Lake Peipsi (data from Noges˜ T. et al., 2005; Loigu et al., 2008).

157 F I C Võrtsjärv s.s. Lämmijärv Peipsi Peipsi 1993 1995 1997 1999 2001 2003 2005 2007 1993 1995 1997 1999 2001 2003 2005 2007 1993 1995 1997 1999 2001 2003 2005 2007 5 0 5 0 0 40 35 30 25 20 15 10 50 20 40 35 30 25 20 15 10 90 80 70 60 40 30 10 -1 -1 TN/TP, mg mg TN/TP, mg mg TN/TP, mg mg-1 H B E 2007 2007 2007 2005 2005 2005 2003 2003 2003 2001 2001 2001 ± Standard error Mean ± Standard deviation 1999 1999 1999 1997 1997 1997 1995 1995 1995 1993 1993 1993 Võrtsjärv Peipsi s.s. Peipsi Lämmijärv 0.10 0.08 0.06 0.04 0.02 0.00 0.20 0.18 0.16 0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.00 0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.00 TP, mg l-1 TP, mg l-1 TP, mg l-1 G D A ¨ arv (G, H, I) in post-Soviet period 1993–2007. ˜ ortsj Lämmijärv ¨ arv (D, E, F) and V . s.s ¨ ammij Changes of total nitrogen (TN) and phosphorus (TP) concentration and TN/TP ratio in Lakes Peipsi (A, B, C; average of three sampling stations shown on Peipsi Peipsi 1993 1995 1997 1999 2001 2003 2005 2007 1993 1995 1997 1999 2001 2003 2005 2007 1993 1995 1997 1999 2001 2003 2005 2007 Võrtsjärv 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 2.8 2.6 2.4 2.2 2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 Figure 5. Figure 1), L 150 TN, mg l-1 TN, mg l-1 TN, mg l-1

158 Noges˜ et al. / Aquatic Ecosystem Health and Management 13 (2010) 143–153 151

Trends of the lake water quality fluctuations on the ecosystem of Lake Vortsj˜ arv¨ are discussed in detail by P. Noges˜ et al., in this issue. According to paleolimnological evidence, the eu- Temperature dependence of cyanobacteria develop- trophication of Lake Peipsi escalated in the 1970s– ment and N2 fixation should be taken into account 1980s (Heinsalu et al., 2007). The eutrophication in the context of global warming, as higher wa- history was similar for Lake Vortsj˜ arv¨ (Noges˜ P. ter temperatures support both cyanobacterial growth and Jarvalt,¨ 2004). Cyanobacterial blooms, a com- (Johnk¨ et al., 2008; Paerl and Huisman, 2008) and mon phenomenon in both lakes in the first half of the P recycling from sediments (Genkai-Kato and Car- 20th century, were caused predominantly by meso- penter, 2005). According to Noges˜ T. et al. (2008) eutrophic species: Gloeotrichia echinulata in Peipsi the temperature dependence of N2 fixers in Lake and Anabaena lemmermannii in Vortsj˜ arv¨ (Muhlen¨ Peipsi is the main factor determining the increasing and Schneider, 1920; Laugaste et al., 2001). Blooms potential share of N2 fixing species in the phyto- ceased in the 1980s due to heavy nitrogen loading plankton community along with increasing water and intensified again in Peipsi in late 1990s, to- temperature. gether with the decrease of the N/P ratio (Noges˜ et al., 2007). Reappearing cyanobacteria blooms in Peipsi have caused serious summer fish-kills in recent years (Kangur et al., 2005). These fish- Conclusions kills seem to be a straightforward consequence of reduced nitrogen levels at remaining high phos- Studied lakes Peipsi and Vortsj˜ arv¨ are largely phorus levels and, thus, the changed N/P ratio. controlled by climatic factors; either directly by in- In Vortsj˜ arv¨ where recurrent winter fish-kills have creased water temperature or indirectly through the been observed, the climatic factors affecting the wa- fluctuations of non-regulated water levels. Changes ter level are more crucial (Noges˜ T. et al., 2003b). In in air temperature, surface water temperature in Peipsi the N2-fixing Gloeotrichia echinulata, Apha- both lakes have increased significantly during 1961– nizomenon flos-aquae and Anabaena species prevail 2004. Practices in agriculture and wastewater treat- in summer phytoplankton. In Vortsj˜ arv¨ the domi- ments are reflected through nutrient loadings to nant cyanobacteria Limnothrix planktonica, L. re- the lakes, which increased rapidly in the 1980s dekei and Planktolyngbya limnetica are not capa- and decreased again in the early 1990s. As total ble to fix N2, and the main N2-fixing taxa Aph- nitrogen (TN) loading decreased faster than total anizomenon skujae and Anabaena spp. commonly phosphorus (TP) loading, the TN/TP ratio in the do not dominate. TN/TP mass ratio in Vortsj˜ arv¨ is loadings decreased. Both the increased temperature about two times higher than in Peipsi (Figure 5C, and low TN/TP ratio favoured the development of F, I, Table 2). The critical TN/TP mass ratio in cyanobacteria blooms in Peipsi. In Vortsj˜ arv,¨ where May–October, below which the N2-fixing cyanobac- the TN/TP mass ratio is about two times higher teria achieved high biomasses, was around 40 in than that in Peipsi, blooms did not occur. Nutrient Vortsj˜ arv¨ and around 30 in Peipsi (Noges˜ T. et al., dynamics in the lakes followed the changes in load- 2008). In Vortsj˜ arv¨ the TN/TP ratio is above this ings, showing the ability of shallow lake ecosystems critical value both in the inflows and in the lake to react sensitively to changes in catchment man- while in Peipsi it often drops below this threshold. agement. Recently, the TN/TP ratio has shown a Recently the TN/TP ratio has shown an increasing tendency to increase, suggesting a certain reduction tendency in both lakes (Figure 5C, F, I). Increased in blooms to be expected. TN loading (Figure 4A) and in-lake concentrations (Figure 5A, D, G), and decreased or stabilised TP loading (Figure 4B) and in-lake concentrations (Fig- Acknowledgements ure 5B, E, H), suggest that a reduction of water- blooms could be expected in Lake Peipsi. This study was supported by SF 0170011508 However, both lakes are large and shallow and from the Estonian Ministry of Education and Re- largely controlled by climatic factors, either directly search, by grant 7600 from Estonian Science Foun- by increased water temperature or indirectly through dation and by EU FP7 grant 226273 (WISER). Data the fluctuations of non-regulated water level (Noges˜ collection in state monitoring program was sup- T.et al., 2003b). The consequences of the water level ported by the Estonian Ministry of the Environment.

159 152 Noges˜ et al. / Aquatic Ecosystem Health and Management 13 (2010) 143–153

References of harmful cyanobacteria. Global Change Biology 14, 495– 512. Arhonditsis, G. B., Brett, M. T., DeGasperi, C. L., Schindler, Juhna, T., Klavins, M., 2001. Water-Quality Changes in Latvia D. E., 2004. Effects of climatic variability on the thermal and Riga 1980–2000: Possibilities and Problems. Ambio 30, properties of Lake Washington. Limnology and Oceanogra- 306–314. phy 49, 256–270. Kangur, K., Kangur, A., Kangur, P., Laugaste, R., 2005. Fish kill Blinova, I., 2001. Water quality and pollution load of the rivers of in Lake Peipsi in summer 2002 as a synergistic effect of a L. Peipsi basin. Riverine load into the L. Peipsi. In: T. Noges,˜ cyanobacterial bloom, high temperature and low water level. (Ed.), Lake Peipsi. Meteorology, Hydrology, Hydrochemistry, Proceedings of the Estonian Academy of Sciences. Biology. pp. 94–96. Sulemees Publishers, Tallinn, Estonia. Ecology 54, 67–80. Coats, R., Perez-Losada, J., Schladow, G., Richards, R., Kangur, A., Kangur, P., Kangur, K., Mols,¨ T., 2007. The role of Goldman, C., 2006. The warming of Lake Tahoe. Climatic temperature in the population dynamics of smelt Osmerus Change 76, 121–148. eperlanus eperlanus m. spirinchus Pallas in Lake Peipsi Dokulil, M. T., Jagsch, A., George, G. D., Anneville, O., (Estonia/Russia). Hydrobiologia 584(1), 433– Jankowski, T., Wahl, B., Lenhart, B., Blenckner, T., 441. Teubner, K., 2006. Twenty years of spatially coherent deep- Kronvang, B., Jeppesen, E., Conley, D. J., Søndergaard, M., water warming in lakes across Europe related to the North At- Larsen, S. E., Ovesen, N. B., Carstensenc, J., 2005. Nutrient lantic Oscillation, Limnology and Oceanography 51, 2787– pressures and ecological responses to nutrient loading reduc- 2793. tions in Danish streams, lakes and coastal waters. Journal of Genkai-Kato, M., Carpenter, S. R., 2005. Eutrophication due Hydrology 304, 274–288. to phosphorus recycling in relation to lake morphometry, Kundzewicz, Z. W., Robson, A. J., 2004. Change detection in temperature, and macrophytes. Ecology 86, 210–219. hydrological records—a review of the methodology. Hydro- Grasshoff, K., Ehrhardt, M., Kremling, K. (Eds), 1983. Methods logical Sciences Journal 49, 7–19. of Seawater Analysis. Verlag Chemie, Weinheim. Laugaste, R., Noges,˜ T., Noges,˜ P., Jastremskij, V. V., Milius A., Hari, R. E., Livingstone, D. M., Siber, R., Burkhardt-Holm, P., Ott, I., 2001. Algae. In: E. Pihu, J. Haberman, (Eds.), Lake Guttinger,¨ H., 2006. Consequences of climatic change for Peipsi. Flora and Fauna, pp. 31–49. Sulemees Publishers, water temperature and brown trout populations in Alpine Tallinn, Estonia. rivers and streams. Global Change Biology 12, 10–26. Livingstone, D. M., 2003. Impact of secular climate change on Heinsalu, A., Alliksaar, T.,Leeben, A., Noges,˜ T.,2007. Sediment the thermal structure of a large temperate central European diatom assemblages and composition of pore-water dissolved lake. Climatic Change 57, 205–225. organic matter reﬂect recent eutrophication history of Lake Loigu, E., Leisk, U.,¨ 1996. Water quality of rivers in the drainage Peipsi (Estonia/Russia). Hydrobiologia 584, 133–143. basin of Lake Peipsi. Hydrobiologia 338, 25–35. Jaani, A., 2001. The location, size and general characterization Loigu, E., Leisk, U.,¨ Iital, A., Pachel, K., 2008. Peipsi jarve¨ valgla of Lake Peipsi. In: T. Noges˜ (Ed.), Lake Peipsi. Meteorology, reostuskoormus ja jogede˜ veekvaliteet (Pollution load of the Hydrology, Hydrochemistry, pp. 10–17. Sulemees Publishers, Lake Peipsi catchment and river water quality. In Estonian.) Tallinn, Estonia. In: J. Haberman, T. Timm, A. Raukas (Eds.), Peipsi, pp. 179– Jarvet,¨ A., 2001. Long-term changes of nutrient runoff in a Lake 199. Eesti Loodusfoto, Tartu, Estonia. Vortsj˜ arv¨ catchment area. In: M. Seepold˜ (Ed.), Environ- Muhlen,¨ M. von zur, Schneider, G., 1920. Der See Wirzjerw in mental Impact and Water Management in a Catchment Area Livland. (Lake Vortsj˜ arv¨ in Livland. In German). Archiv fur¨ Perspective, pp. 54–62. Tallinn Technical University, Tallinn, die Naturkunde des Ostbaltikums 14, 1–156. Estonia. Noges,˜ T., 2009. Trends of air and water temperature in Estonia Jarvet,¨ A., Mander, U.,¨ Kull, A., Kuusemets, V., 2002. Nutrient and in large lakes Peipsi and Vortsj˜ arv,¨ and possible con- runoff change in a rural catchment in South Estonia. Large sequences on water quality. Stuttgart Verhandlungen der In- Rivers. Archiv fur¨ Hydrobiology, Supplement 141, 305–319. ternationalen Vereinigung fur¨ Theoretische und Angewandte Jeppesen, E., Jensen, J.-P., Søndergaard, M., Lauridsen, T. L., Limnologie 30(7), 997–999. 2005a. Response of ﬁsh and plankton to nutrient loading Noges,˜ P.,Jarvalt,¨ A., 2004. Long-term changes in the ecosystem reduction in eight shallow Danish lakes with special emphasis of Lake Vortsj˜ arv¨ and their causes. In: J. Haberman, E. Pihu, on seasonal dynamics. Freshwater Biology 50, 1616–1627. A. Raukas (Eds.), Lake Vortsj˜ arv,¨ pp. 347–355. Estonian Jeppesen, E., Søndergaard, M., Jensen, J.-P., Havens, K. E., Encyclopaedia Publishers, Tallinn, Estonia. Anneville, O., Carvalho, L., Coveney, M. F., Deneke, R., Noges,˜ P.,Jarvet,¨ A., 2005. Climate driven changes in the spawn- Dokulil, M. T., Foy, B., Gerdeaux, D., Hampton, S. E., ing of roach (Rutilus rutilus (L.)) and bream (Abramis brama Kangur, K., Kohler,¨ J., Hilt, S., Lammens, E. H. H. R., (L.)) in the Estonian part of the Narva River basin. Boreal Lauridsen, T.L., Manca, M., Miracle, M. R., Moss, B., Noges,˜ Environmenatl Research 10, 45–55. P.,Persson, G., Phillips, G., Portielje, R., Romo, S., Schelske, Noges,˜ T., Jarvet,¨ A., Laugaste, R., Loigu, E., Tonno,˜ I., Noges,˜ C. L., Straile, D., Tatrai, I., Willen, E., Winder, M., 2005b. P., 2003a. Current State and Historical Changes of Nutrient Lake responses to reduced nutrient loading—an analysis of Loading and Ecological Status of Large Lakes Peipsi and contemporary long-term data from 35 case studies. Freshwa- Vortsj˜ arv,¨ pp. 199–203. In: H. Simola, A. Yu Terzhevik, M. ter Biology 50, 1747–1771. Viljanen, I. K. Holopainen (Eds.), Proceedings of 4th Interna- Johnk,¨ K. D., Huisman, J., Sommeijer, B., Sharples, J, Visser, tional Lake Ladoga Symposium 2002 September 2–6 Russia P. M., Stroom, J., 2008. Summer heatwaves promote blooms University of Joensuu, Finland.

160 Noges˜ et al. / Aquatic Ecosystem Health and Management 13 (2010) 143–153 153

Noges,˜ T., Noges,˜ P., Laugaste, R., 2003b. Water level as the shallow temperate lakes Peipsi and Vortsj˜ arv,¨ North-East Eu- mediator between climate change and phytoplankton compo- rope. Hydrobiologia 599, 77–86. sition in a large shallow temperate lake Hydrobiologia 506(1), Noges,˜ P., Noges,˜ T., Adrian, R., Weyhenmeyer, G. A., 2008. 257–263. Silicon load and the development of diatoms in three river- Noges˜ T., Jarvet¨ A., Laugaste R., Loigu E., Leisk U.,¨ lake systems in countries surrounding the Baltic Sea. Hydro- Tonno˜ I., Noges˜ P., 2005. Consequences of catchment biologia 599, 67–76. processes and climate changes on the ecological status Paerl, H. W., Huisman, J., 2008. Blooms like it hot. Science 320, of large shallow temperate lakes. In: Ramachandra, T. 57–58. V., Ahalya, N., Murthy, C. R. (Eds.), Aquatic ecosys- Straile, D., Johnk,¨ K., Rossknecht, H., 2003. Complex effects of tems, conservation, restoration and management, pp. 88– winter warming on the physicochemical characteristics of a 99. Capital Publishing Company, New Dehli, Kolkata, deep lake. Limnology and Oceanography 48, 1432–1438. Bangalore. Weyhenmeyer, G. A., Meili, M., Livingstone, D. M., 2005. Sys- Noges,˜ P., Kagu,¨ M., Noges,˜ T., 2007. Role of climate and agri- tematic differences in the trend towards earlier ice-out on cultural practice in determining the matter discharge into Swedish lakes long a latitudinal temperature gradient. Ver- large shallow Lake Vortsj˜ arv,¨ Estonia. Hydrobiologia 581, handlungen des Internationalen Verein Limnologie 29, 257– 125–134. 260. Noges,˜ T., Jarvet,¨ A., Kisand, A., Laugaste, R., Loigu, E., World Meteorological Organisation (WMO), 1994. Guide to Skakalski, B., Noges,˜ P.,2007. Reaction of large and shallow Hydrological Practices. Fifth edition, WMO Publication lakes Peipsi and Vortsj˜ arv¨ to the changes of nutrient loading. No. 168, World Meteorological Organization, Geneva. Hydrobiologia 584, 253–264. WMO, 2008. Guide to Meteorological Instruments and Methods Noges,˜ T., Laugaste, R., Noges,˜ P., Tonno,˜ I., 2008. Critical of Observation. Seventh edition, WMO-No. 8, World Mete- N:P ratio for cyanobacteria and N2-ﬁxing species in the large orological Organization, Geneva.

161

IV Toming, K., Kutser, T., Tuvikene, L., Viik, M. & Nõges, T. 2016. Dissolved organic carbon and its potential predictors in eutrophic lakes. Water Research, 102: 32−40, doi: 10.1016/j. watres.2016.06.012. Water Research 102 (2016) 32e40

Contents lists available at ScienceDirect

Water Research

journal homepage: www.elsevier.com/locate/watres

Dissolved organic carbon and its potential predictors in eutrophic lakes

Kaire Toming a, b, *, Tiit Kutser a, Lea Tuvikene b, Malle Viik b, Tiina Noges~ b a Estonian Marine Institute, University of Tartu, Maealuse€ 14, Tallinn 12618, Estonia b Centre for Limnology, Institute of Agricultural and Environmental Sciences, Estonian University of Life Sciences, Kreutzwaldi 5, Tartu 51014, Estonia article info abstract

Article history: Understanding of the true role of lakes in the global carbon cycle requires reliable estimates of dissolved Received 16 December 2015 organic carbon (DOC) and there is a strong need to develop remote sensing methods for mapping lake Received in revised form carbon content at larger regional and global scales. Part of DOC is optically inactive. Therefore, lake DOC 1 June 2016 content cannot be mapped directly. The objectives of the current study were to estimate the relation- Accepted 3 June 2016 ships of DOC and other water and environmental variables in order to ﬁnd the best proxy for remote Available online 6 June 2016 sensing mapping of lake DOC. The Boosted Regression Trees approach was used to clarify in which relative proportions different water and environmental variables determine DOC. In a studied large and Keywords: Dissolved organic carbon shallow eutrophic lake the concentrations of DOC and coloured dissolved organic matter (CDOM) were Coloured dissolved organic matter rather high while the seasonal and interannual variability of DOC concentrations was small. The re- Eutrophic lakes lationships between DOC and other water and environmental variables varied seasonally and inter- Remote sensing annually and it was challenging to ﬁnd proxies for describing seasonal cycle of DOC. Chlorophyll a (Chl a), Boosted Regression Trees total suspended matter and Secchi depth were correlated with DOC and therefore are possible proxies for remote sensing of seasonal changes of DOC in ice free period, while for long term interannual changes transparency-related variables are relevant as DOC proxies. CDOM did not appear to be a good predictor of the seasonality of DOC concentration in Lake Vortsj~ arv€ since the CDOMeDOC coupling varied seasonally. However, combining the data from Vortsj~ arv€ with the published data from six other eutrophic lakes in the world showed that CDOM was the most powerful predictor of DOC and can be used in remote sensing of DOC concentrations in eutrophic lakes. © 2016 Elsevier Ltd. All rights reserved.

1. Introduction airborne sensors must affect the optical properties of water (e.g. reflectance), or correlate directly with water characteristics that Understanding of the true role of lakes in the global carbon cycle affect the optical water properties (Kutser et al., 2015a). Coloured requires reliable estimates of dissolved organic carbon, DOC, as dissolved organic matter, CDOM, absorbs light and there is typically 90e95% of organic carbon in lakes is in the dissolved form (Wetzel, a strong correlation between DOC and CDOM in humic lakes where 2001). Monitoring of DOC in lakes is expensive and the extent of both parameters are fluctuating synchronously (Tranvik, 1990; field sampling is limited, both spatially and temporally. Therefore, Molot and Dillon, 1997; Kallio, 1999; Yacobi et al., 2003; Zhang there is a strong need to develop remote sensing methods for et al., 2007; Erlandsson et al., 2012). Therefore, CDOM is often mapping lake carbon content at larger regional and global scales used as a proxy in mapping lake DOC content (Kutser et al., 2005; (Kutser et al., 2015a). Del Castillo and Miller, 2008; Kutser et al., 2009). In non-humic Only the visible part of electromagnetic radiation can penetrate lakes where DOC and CDOM do not vary synchronously the situa- the water surface and give us information about water properties. tion is much more complicated than in humic lakes. Molot and Consequently, the parameter we want to measure from space or Dillon (1997) stated that if optical parameters are used as surro- gates for all or some fraction of DOC, then the mathematical relationship between these parameters and the DOC fraction must be time invariant. There is a lack of data about the CDOMeDOC rela- * Corresponding author. Estonian Marine Institute, University of Tartu, Maealuse€ tionship in eutrophic lakes, where autochthonous DOC may form a 14, Tallinn 12618, Estonia. E-mail address: [email protected] (K. Toming). considerable portion of the total DOC pool compared to http://dx.doi.org/10.1016/j.watres.2016.06.012 0043-1354/© 2016 Elsevier Ltd. All rights reserved.

165 K. Toming et al. / Water Research 102 (2016) 32e40 33 allochthonous DOC (Toming et al., 2013). Autochthonous DOC is performance (e.g. Elith et al., 2006, 2008; Leathwick et al., 2006, produced inside the lake by phytoplankton and other photosyn- 2008). The objectives of the current study are: (1) to estimate the thetic organisms. It does not absorb light and consists mainly of correlative relationships of DOC and other water and environ- non-humic substances (Bertillson and Jones, 2003) that are labile mental variables, and using BRT analyses (2) to clarify in which and easily utilized or degraded by microorganisms (Thurman, relative proportions different water and environmental variables 1985). Allochthonous DOC originates primarily from vascular determine DOC in Lake Vortsj~ arv.€ Furthermore, DOC, CDOM and Chl plants and soil organic matter of the catchment area. It consists a data from Lake Vortsj~ arv€ were analysed together with published mainly of humic substances, is refractory to decomposition, absorbs data of eutrophic lakes Balaton (Hungary), Taihu (China), Miastro light and is coloured brownish (Thurman, 1985). Thus, allochtho- (Belarus), Batorino (Belarus), Mendota (USA) and Kinneret (Israel) nous DOC can be considered mainly as coloured (CDOM) and for large scale comparisons. We hypothesize that CDOM alone is not autochthonous mainly as non-coloured DOC due to their proper- a good predictor for DOC in large, shallow and eutrophic lakes and ties. Furthermore, the spatio-temporal dynamics of DOC and CDOM that DOC correlations with other water parameters change in time might differ largely due to their different sources, chemical and in the gradient of environment properties. composition, degradation processes (photochemical and microbial), discharge from rivers and other factors. Due to the high share of 2. Material and methods autochthonous DOC in eutrophic waters, CDOM might not be the best predictor for DOC. 2.1. Study site Increases in DOC concentrations have often been detected in rivers and lakes (Evans et al., 2005; Clark et al., 2010; Filella and Data were analysed from a large, shallow, eutrophic Lake Rodríguez-Murillo, 2014) over the past decades. DOC plays a sig- Vortsj~ arv€ (57�500e58�300N and 25�350e26�400E), Estonia (Fig. 1). nificant role in the carbon and energy cycle of lakes and as the main The lake area is 270 km2, volume 0.75 km3, mean depth 2.8 m, source of energy for microbial metabolism it can have a broad effect maximum depth 6 m, and catchment area is 3104 km2. The water on food chains and on the proportions of auto- and heterotrophic column is well mixed by surface waves and currents. The renewal of processes (Tranvik, 1992). DOC also has an influence on nutrient water takes 240e384 days and can differ markedly between dry retention and release and on the mobility of metals (De Haan, and rainy years (Jaani, 1990). A specific feature of Vortsj~ arv€ is the 1992). Moreover, the high concentration of organic acids in DOC large natural climate-related variability of water level, which cau- gives a naturally low pH to the water of humic lakes (Kortelainen, ses up to a 3-fold difference in its water volume (Noges~ et al., 2010). 1999) and the photochemical degradation of DOC decreases oxy- The lake is covered by ice for an average 184 days. The ice-free gen concentration (Lindell and Rai, 1994). High levels of DOC must period lasts from April to October and the ice period from be removed from drinking water where it is to be disinfected using November to March. The flow regimes of the inflowing rivers are chlorination (Eikebrokk et al., 2004). Further, the coloured natural, and discharges usually peak in April. The lake has 4 main component of DOC e CDOM is one of the optically active substances inflows (the Rivers Vaike€ Emajogi,~ Ohne,~ Tarvastu, and Tanassilma)€ in water competing with phytoplankton and other aquatic plants and one outflow (the River Emajogi).~ for the capture of available light energy. At the same time, CDOM protects aquatic organisms against harmful UV radiation (Kirk, 1980; Jones and Arvola, 1984; Davies-Colley and Vant, 1987; 2.2. Data collection and analysis Arvola et al., 1999). Thus, DOC is a very important parameter to monitor in water We have a unique 7-year (from 2008 to 2014) database on DOC bodies and mapping both CDOM and DOC with remote sensing is and other water and environmental variables in Lake Vortsj~ arv:€ an important task. On the other hand, the reasons described above CDOM, chemical oxygen demand by permanganate (CODMn), water indicate that other satellite products e.g. total suspended matter, colour by Platinum-Cobalt scale (colourPt-Co), chlorophyll a (Chl a), turbidity, and transparency, could be used to predict lake carbon total suspended matter (TSM), water temperature (WT), dissolved fl levels. Besides, CDOM may be one of the most difficult water quality oxygen (O2), water pH, Secchi depth (S), water level (WL), in owing variables to map with remote sensing (Brezonik et al., 2015). For riverine discharges (I) and precipitation (PR). The data, aggregated providing efficient decision-making tools for lake managers and with monthly time step were used in analysis. For those indices, 3 1 relevant input information for carbon cycle and climate models, it is which were measured daily (WL, cm; I, m s� ; PR, mm) monthly important to understand all possible relationships of DOC with averages were calculated prior to the analysis together with the 1 different water and environmental characteristics. indices which were measured once per month (DOC, mg C l� ; 1 1 1 We have a unique 7-year database on DOC and other water and CDOM, mg l� ; CODMn,mgOl� ; colourPt-Co, mg Pt l� ; Chl a, 1 1 1 environmental variables in large, shallow and eutrophic Lake mg l� ; TSM, mg l� ; WT, t�; O, mg l� ; pH; S, m). WL, I, PR, CODMn, Vortsj~ arv.€ This database makes it possible to explore the suitable colourPt-Co, Chl a, TSM were measured as part of the state moni- predictors for DOC. Moreover, DOC, CDOM and Chl a data from Lake toring programme. These data were obtained from Estonian Envi- Vortsj~ arv€ were analysed together with published data of the ronment Agency. eutrophic lakes Balaton (Hungary), Taihu (China), Miastro (Belarus), For determination of DOC concentrations, water was passed Batorino (Belarus), Mendota (USA) and Kinneret (Israel) for large through pre combusted (3 h at 500 �C) Whatman GF/F glass mi- scale comparison. crofiber filters and the carbon content of the filtrate was measured Usually, regression models are used for quantifying the rela- according to Toming et al. (2013). tionship between a dependent variable and the others on which it The amount of CDOM was characterized by its concentration 1 depends. In current study the traditional Pearson correlations, (mg l� ) calculated from equation below (Eq. (1))(Højerslev, 1980; regression models and a novel predictive modelling technique Sipelgas et al., 2003): called Boosted Regression Trees (BRT) were used in the analysis of * the data. The BRT approach differs fundamentally from traditional c f l C ð Þ (1) regression methods that produce a single ‘best’ model, instead CDOM ¼ exp S l l a* l ð� ð � 0ÞÞ CDOMð 0Þ using the technique of boosting to combine large numbers of relatively simple tree models adaptively, to optimize predictive where a*CDOM(l0) is the specific absorption coefficient of DOM,

166 34 K. Toming et al. / Water Research 102 (2016) 32e40

1 1 which numerical value at l0 380 nm was 0.565 L m� mg� in which individual terms are simple trees, fitted in a forward, stage ¼ 1 (Højerslev, 1980), S is the slope parameter equal to 0.017 nm� wise fashion (Elith et al., 2008). Boosted Regression Trees incor- (Sipelgas et al., 2003), and c*f(l) was taken from spectrometric porate important advantages of tree-based methods, handling reading at l 380 nm. different types of predictor variables and accommodating missing ¼ 3 For Chl a (mg m� ), 0.1e1 L of water was passed through data. They have no need for prior data transformation or elimina- Whatman GF/F glass microfiber filter and concentrations were tion of outliers, can fit complex nonlinear relationships, and auto- measured spectrophotometrically (Edler, 1979) at a wavelength of matically handle interaction effects between predictors. Fitting 665 nm from 96% ethanol extracts of the filters. multiple trees in BRT overcomes the biggest drawback of single tree WT and O were both measured in situ using a portable dissolved models: their relatively poor predictive performance (Elith et al., oxygen meter Marvet Junior 2000 (Elke Sensor, Estonia), and pH 2008). Although BRT models are complex, they can be summa- with a portable device ProfiLine pH 3210 (Wissenschaftlich-Tech- rized in ways that give powerful ecological insight, and their pre- nische Werkstatten€ GmbH), since 2012 all the three indicators were dictive performance is superior to most traditional modelling measured with a portable YSI Professional Plus instrument (YSI methods (Elith et al., 2008). For BRT modelling the predictive var- Incorporated). S was measured with a 30 cm diameter Secchi disk. iables were chosen. Since there should be no collinearity between predictive variables, the results of Pearson correlation and variance inflation factor (VIF) was used to estimate the collinearity between 2.3. Statistical analysis variables (CDOM, CODMn, colourPt-Co, Chl a, TSM, WT, O2, pH, S, WL, I, PR). KolmogoroveSmirnov two sample test was used to find out if The Pearson correlations, regression models and a novel pre- there are statistically significiant differences in DOC and in water dictive modelling technique called Boosted Regression Trees (BRT) and environmental variables from ice period and ice-free period were used in the analysis of the data in Lake Vortsj~ arv.€ The deter- (April to October). The significance level to indicate differences and mination power of different water and environmental variables for relationships was set at p < 0.05. the DOC value in Lake Vortsj~ arv€ was explored using the Boosted Regression Trees, BRT (Elith et al., 2008) in R software (R 3.2.2. for Windows). BRT is an ensemble method for fitting statistical models 2.4. Data for large-scale comparison that differ fundamentally from conventional techniques that aim to fit a single parsimonious model (Elith et al., 2008). BRT combine the For large scale comparisons we analysed DOC, CDOM and Chl a strengths of two algorithms: regression trees and boosting. The data from Lake Vortsj~ arv€ together with published data for the final BRT model can be understood as an additive regression model eutrophic lakes Balaton (Hungary), Taihu (China), Miastro (Belarus),

Fig. 1. Study site.

167 K. Toming et al. / Water Research 102 (2016) 32e40 35

Table 1 Average (Av.), minimum (Min), maximum (Max) and standard deviation (SD) of the concentrations of DOC and its possible predictors from January to December (All), ice-free period from April to October (ApreOct) and in ice-period in 2008e2014 in Lake Vortsj~ arv.€

Unit Apreoct Ice-period

Av. Min Max SD Av. Min Max SD

1 DOC mg C l� 15.7 11.0 20.8 2.85 13.8 7.83 20.8 3.83 1 CDOM mg l� 7.01 3.49 13.9 2.36 11.2 4.51 17.4 3.79 1 CODMn mg O l� 14.4 11.0 18.0 1.85 13.8 9.00 19.0 2.96 1 ColourPt-Co mg Pt l� 59.4 20.0 120 24.3 74.1 30.0 140 29.6 1 Chl a mg l� 41.7 0.55 79.6 17.7 7.96 0.33 35.5 9.00 1 TSM mg l� 16.8 3.00 32.0 6.77 5.73 2.20 12.0 3.25

WT t� 14.7 3.20 25.1 5.51 1.32 0.20 6.10 1.19 1 O2 mg l� 10.3 7.70 14.1 1.52 11.6 4.64 18.9 3.65 pH 8.56 7.70 9.18 0.25 7.85 7.31 8.38 0.29 S M 0.75 0.45 1.75 0.27 1.31 0.50 2.40 0.45 WL Cm 71.0 28.0 181 53.6 95.6 26.4 150 44.2 3 1 � � Ims� 19.5 5.43 133 20.1 34.5 8.41 134 29.9 PR Mm 78.1 17.3 204 40.5 39.5 10.8 98.5 22.0

Batorino (Belarus), Mendota (USA) and Kinneret (Israel). The In the ice-free lake (AprileOctober) DOC was significantly Pearson correlations and regression models were used for large- positively correlated with Chl a, TSM, pH and PR, and negatively scale comparison. with S and I (Table 4). Unfortunately, the correlations were always smaller than 0.4 and it was not possible to find a very good linear 3. Results regression model to predict DOC. Still, Chl a, TSM and S appeared to be slightly better predictors for DOC in ice-free period in Lake 3.1. Dynamics of dissolved organic carbon and its possible Vortsj~ arv€ than other variables (Eqs. (2)e(4)). predictors DOC 12.0 0.19*TSM (r 0.40, p < 0.05) (2) ¼ þ ¼ Average values of DOC, Chl a, TSM, WT, pH, S, PR were somewhat DOC 12.3 0.07*Chl a (r 0.39, p < 0.05) (3) higher from April to October (ice-free period) than the values of the ¼ þ ¼ same variables in the ice-period (Table 1). CDOM, CODMn, colourPt- DOC 17.8e3.39*TSM (r 0.38, p < 0.05) (4) Co,O2, WL and I were higher in ice period. ColourPt-Co, Chl a, WL, I ¼ ¼� and PR showed high annual variability. KolmogoroveSmirnov two sample test showed that there were statistically significiant dif- Under the ice DOC correlated only with ColourPt-Co and PR ferences (p < 0.05) in CDOM, Chl a, TSM, WT, pH, S, WL, I, PR be- (Table 4). tween ice period and ice-free period (April to October). There was Based on the annual mean data, DOC concentration was statistically significantly correlated only with S (Table 3, Fig. 4). not statistically significiant difference (p > 0.05) in DOC, CODMn, The annual mean data of CDOM, Chl a, TSM, S, and PR were used colourPt-Co and O2. as predictors of DOC concentration in BRT analysis. These variables Seasonally, CDOM, colourPt-Co,O2, S, WL and I were usually somewhat higher from December to March, contrary to Chl a, TSM, explained 1.52%, 0.50%, 0.45%, 97.3% and 0.21% of the variance of fi WT, pH, PR which showed higher values from May to October DOC respectively. As Secchi depth described the most signi cant proportion (97.3%) of the variability, it appeared to be the most (Fig. 2). The concentrations of DOC and CODMn were quite high in January and February decreasing to the spring as were to other powerful predictor for the annual DOC concentration in Lake Vortsj~ arv.€ water colour variables (CDOM, colourPt-Co), but in reverse from others they increased again after the spring low (Fig. 2a). Interannually, DOC was somewhat higher in 2008e2010 and in 3.3. Results of large-scale analyses 2014 (Fig. 3a). Years with lower WL, I and PR corresponded to higher Chl a, TSM, pH and O2 (Fig. 3b, c) while the variables con- The concentrations of DOC, CDOM and Chl a from 7 different nected to the water colour and transparency (CDOM, colourPt-Co, eutrophic lakes were compared (Table 5). The mean depth of the CODMn and S) were usually lower in those years (Fig. 3a). The latter lakes was 4.31 m and the mean area was 14,2 ha. The concentra- 1 was however not valid for DOC. tions of DOC varied from 3.60 to 16.8 mg C l� (mean 8.33). The 1 mean values of Chl a and CDOM were 16.4 mg l� (SD 17.1) and 1 ¼ 3.2. Relationships between dissolved organic carbon and its 4.34 mg l� (SD 2.68). Values of CDOM and Chl a were both fi ¼ possible predictors signi cantly positively correlated with concentrations of DOC among these lakes (accordingly r 0.85 and r 0.45) and linear ¼ ¼ Correlations based on monthly data showed very high collin- regressions for predicting DOC via CDOM or Chl a were statistically fi earity between some of the variables under study (Table 2). For signi cant (Fig. 5a and b). instance, CDOM had statistically highly significant correlations with 10 other variables and Chl a with 9 other variables (Table 2). 4. Discussion Therefore, it was not possible to use the BRT analyses or multiple regressions to predict the DOC on ice-free periods in Lake Vortsj~ arv.€ Large lakes with long water residence times tend to have Correlations based on annual mean data (Table 3) showed collin- smaller DOC concentrations and less coloured water because of earity only between water colour variables (CDOM, CODMn, and lower areal loading rates and higher in-lake rates of photochemical ColourPt-Co), and O2, WL and I. Chl a was statistically significantly and biological mineralization of DOC (Pace and Cole, 2002). How- correlated only with WT. ever, the concentrations of DOC and CDOM were rather high in Lake

168 36 K. Toming et al. / Water Research 102 (2016) 32e40

20 110 20 110 ), ), ), ), DOC CDOM COD Colour S*10 (a) ) Mn -1 DOC CDOM COD Colour Secchi (S)*10 Mn -1 0 l l 100

18 1 m 100 18 ( g g * ) ) ) 0 m m -1 90 -1 ( ( m 1 16 16 (

* 90 M M S S 80 , , O O ) 14 ) 14 D D -1 80 -1 l l 70 C (mg Pt l C (mg Pt l , ,

O 12 O 12 ) ) g g -1 -1 60 Pt-Co

70 Pt-Co l l m 10 m 10 ( ( C C 50 g 60 g Mn Mn m 8 m 8 Colour Colour ( ( 40 C 50 C COD 6 COD 6 O (a) O 30 D 4 40 D 4 20 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 2008 2009 2010 2011 2012 2013 2014

70 20 27 60

) Chl a TSM O pH (b) Chla TSM O2 2 60 -1 55 18 24 ), pH -1 50 21 50 16 ) ) -1 -1

18 (mg O l 2 g g

40 45 14 (mg O l μ μ 2 (l (l

15 ), O a a -1 ), O l l 30 40 12 h h 12 -1 C C 20 9 35 10

6 8 10 (mg l TSM 30 (b) 3 TSM (mg l 0 25 6 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 2008 2009 2010 2011 2012 2013 2014

160 24 120 11.0 ) ) m WL I PR WT 22 m (c) 140 (c) WL I PR WT m m 10.5

( 100 20 ( R 120 18 R P P 10.0

, 80 16 , ) ) 100 ) -1 14 -1

s 9.5 s 3 80 12 3 60 m m ( (

WT (t°) 9.0 10 WT (t° I 60 I 40 , , ) 8 )

m 8.5 40 6 m c c 20 ( 4 ( L 20 L 8.0 2 W W 0 0 0 7.5 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 2008 2009 2010 2011 2012 2013 2014 Month Year

Fig. 2. Mean (±standard error) monthly values of a: dissolved organic carbon (DOC), Fig. 3. Mean annual values of a: dissolved organic carbon (DOC), coloured dissolved coloured dissolved organic matter (CDOM), chemical oxygen demand by permanga- organic matter (CDOM), chemical oxygen demand by permanganate (CODMn), colour nate (CODMn), colour by Platinum-Cobalt scale (colourPt-Co), Secchi depth (S), b: by Platinum-Cobalt scale (colourPt-Co), Secchi depth (S), b: chlorophyll a (Chl a), total chlorophyll a (Chl a), total suspended matter (TSM), dissolved oxygen (O2) and c: water suspended matter (TSM), dissolved oxygen (O2) and c: water temperature (WT), water temperature (WT), water level (WL), inﬂowing riverine discharges (I) and pre- pH, water level (WL), inﬂowing riverine discharges (I) and precipitations (PR) in the cipitations (PR) in the Lake Vortsj~ arv€ in 2008e2014. Lake Vortsj~ arv€ in 2008e2014.

Vortsj~ arv€ being almost two to three times higher than in other During the vegetation period, DOC concentration was strongly eutrophic lakes (Table 5). CDOM discharge from inflows and related to phytoplankton development and the variables like Chl a, phytoplankton extracellular release of organic compounds strongly TSM, pH, S, I and PR could be used as DOC proxies in this period. influence the standing stock of DOC, both being remarkable sources Since it is possible to use remote sensing for estimating DOC con- of DOC also in Vortsj~ arv€ (Toming et al., 2013) keeping its concen- centrations only in ice-free period, variables like Chl a, TSM and S tration in this large shallow and eutrophic lake high and rather should be considered as the proxies for remote sensing of DOC stable both seasonally and interannually. Usually, low seasonal and seasonality in Lake Vortsj~ arv€ (Eqs. (2)e(4)). The proxies for esti- interannual variance reflects the slow dynamics of a large and mating DOC concentrations in long term studies should reflect both relatively recalcitrant pool of organic matter (Wetzel, 2001). In the variability of phytoplankton and also CDOM and therefore it is Vortsj~ arv,€ the additional stabilizing mechanism of overall DOC better to use transparency-related variables (e.g. S). On the whole, concentration is assumingly the specific seasonal pattern where the our study in a eutrophic lake showed, similarly to the results from lowest level of CDOM coincides with the highest level of the oligotrophic and mesotrophic lakes in Canada (Molot and phytoplankton-derived DOC. As the proportions of allochthonous Dillon, 1997), that CDOM alone might not be the best predictor and autochthonous material in DOC of Lake Vortsj~ arv€ vary for describing seasonal changes of DOC concentrations since the seasonally and interannually in large scale (Toming et al., 2013), the CDOMeDOC relationship varies seasonally. nature of the relationships between DOC and other variables also As well, Jiang et al. (2012) found that the direct correlation be- vary in different seasons and years, making it difficult to find some tween DOC and CDOM absorption was spectrally variable and specific proxies for DOC characterization throughout different generally poor in spring and autumn for most datasets in large, seasons and years. shallow and eutrophic Lake Taihu and the highest correlation was

169 K. Toming et al. / Water Research 102 (2016) 32e40 37

Table 2

Pearson correlation coefficients of dissolved organic carbon (DOC) and coloured dissolved organic matter (CDOM), chemical oxygen demand by permanganate (CODMn), water colour by Platinum-Cobalt scale (colourPt-Co), chlorophyll a (Chl a), total suspended matter (TSM), water temperature (WT), dissolved oxygen (O2), water pH, Secchi depth (S), water level (WL), inflowing riverine discharges (I) and precipitations (PR) in 2008e2014 in Lake Vortsj~ arv,€ based on monthly mean values. p < 0.05 are shown, ns-not significant.

DOC CDOM CODMn ColourPt-Co Chl a TSM WT O2 pH S WL I DOC CDOM ns

CODMn 0.23 0.47 ColourPt-Co ns 0.67 0.49 Chl a 0.26 0.60 ns ns À TSM ns 0.59 ns 0.37 0.76 À À WT ns 0.39 ns ns 0.56 0.31 À O ns ns 0.27 ns ns ns 0.50 2 À À pH 0.22 0.44 0.25 ns 0.76 0.55 0.70 ns À S 0.33 0.45 ns ns 0.76 0.68 0.43 ns 0.67 À À À À À WL ns 0.75 0.32 0.58 0.37 0.48 ns ns 0.22 0.32 À À À I ns 0.37 ns 0.24 0.34 ns 0.41 Ns 0.46 ns 0.49 À À À PR 0.27 0.24 0.24 ns 0.51 ns 0.48 0.22 0.51 0.45 ns ns À À À

Table 3

Pearson correlation coefficients of dissolved organic carbon (DOC) and coloured dissolved organic matter (CDOM), chemical oxygen demand by permanganate (CODMn), water colour by Platinum-Cobalt scale (colourPt-Co), chlorophyll a (Chl a), total suspended matter (TSM), water temperature (WT), dissolved oxygen (O2), water pH, Secchi depth (S), water level (WL), inflowing riverine discharges (I) and precipitations (PR) in 2008e2014 in Lake Vortsj~ arv,€ based on annual mean data. p < 0.05 are shown, ns-not significant.

DOC CDOM CODMn ColourPt-Co Chl a TSM WT O2 pH S WL I DOC CDOM ns

CODMn ns 0.90 ColourPt-Co ns 0.91 ns Chl a ns ns ns ns TSM ns ns ns ns ns WT ns ns ns ns 0.80 ns O ns 0.95 0.85 0.8 ns ns ns 2 À À À pH ns ns ns ns ns ns ns ns S 0.89 ns ns ns ns ns ns ns ns À WL ns 0.86 0.83 0.79 ns ns ns ns ns ns I ns 0.93 0.84 0.94 ns ns ns 0.81 ns ns 0.94 À PR ns ns ns ns ns ns ns ns ns ns ns ns

found to occur at lower wavelengths. However, in eutrophic Lake temperate and subtropical regions (Niu et al., 2014). Furthermore, a Taihu, CDOM optical properties and DOC concentration behaved spatial study in the northern part of Lake Taihu showed a significant conservatively and were strongly correlated in April and in August, linear relationship between CDOM absorption and DOC concen- indicating that in situ CDOM measurements could be used as a trations, especially at shorter wavelengths (Zhang et al., 2007). The proxy for DOC concentration in these periods of the year in relationships between DOC and CDOM in Lake Taihu might show that in short term (daily to monthly scale) and spatial studies it might be possible to use CDOM as proxy for DOC since the rela- Table 4 tionship between those two variables is rather stable in those cases. Pearson correlation (r) and determination (R2) coefficients with their p values Kohler€ et al. (2013) and Xiao et al. (2013) indicated that the asso- (p < 0.05 bold) between dissolved organic carbon (DOC) and coloured dissolved ciation of dissolved iron with DOC affects its colour intensity, and organic matter (CDOM), chemical oxygen demand by permanganate (CODMn), water variations in iron concentrations therefore can lead to the poor colour by Platinum-Cobalt scale (colourPt-Co), chlorophyll a (Chl a), total suspended matter (TSM), water temperature (WT), dissolved oxygen (O2), water pH, Secchi depth (S), water level (WL), inflowing riverine discharges (I) and precipitations (PR) in ice-free (ApreOct) and ice-covered Lake Vortsj~ arv€ in 2008e2014. ns-not significant. 19 18 ApreOct Ice-period ) -1 l r pr p 17 C

CDOM ns ns g 16 m CODMn ns ns ( 15 ColourPt-Co ns 0.45 0.03 C Chl a 0.39 0.01 ns O 14 DOC = 28.6 - 14.7*S D TSM 0.40 0.00 ns 13 r = -0.89, p = 0.01 WT ns ns 12 O2 ns ns pH 0.34 0.02 ns 6 7 8 9 10 11 S 0.38 0.01 ns À Secchi depth (m) WL ns ns I 0.30 0.03 ns À Fig. 4. Relationship between dissolved organic carbon (DOC) and Secchi depth (S) in PR 0.34 0.02 0.52 0.01 2008e2014 in Lake Vortsj~ arv,€ based on annual mean data.

170 38 K. Toming et al. / Water Research 102 (2016) 32e40

Table 5 Dissolved organic carbon (DOC), chlorophyll a (Chl a), coloured dissolved organic matter (CDOM) and physical characteristics of some eutrophic lakes around world.

Lake name Country Site Latitude (N) Longitude (E) Area (ha) Residence Mean Season DOC Chl a CDOM Reference 1 1 1 time (days) Depth (m) (mg C l� ) (mg l� ) (mg l� )

Balaton Hungary West lake 46� 460 17� 350 5920 730 3.20 Spring 7.90 8.30 4.60 Toth� et al., 2007 Balaton Hungary West lake 46� 460 17� 350 5920 730 3.20 Summer 9.30 13.5 4.50 Toth� et al., 2007 Balaton Hungary West lake 46� 460 17� 350 5920 730 3.20 Autumn 9.60 8.50 5.10 Toth� et al., 2007 Balaton Hungary West lake 46� 460 17� 350 5920 730 3.20 Winter 9.90 13.0 4.30 Toth� et al., 2007 Balaton Hungary East lake 46� 560 18� 010 5920 730 3.20 Spring 7.70 3.30 3.30 Toth� et al., 2007 Balaton Hungary East lake 46� 560 18� 010 5920 730 3.20 Summer 9.20 2.90 4.40 Toth� et al., 2007 Balaton Hungary East lake 46� 560 18� 010 5920 730 3.20 Autumn 8.90 3.90 4.10 Toth� et al., 2007 Balaton Hungary East lake 46� 560 18� 010 5920 730 3.20 Winter 8.80 1.10 3.80 Toth� et al., 2007 Taihu China Meiliang Bay 31� 280 120� 100 22500 264 1.90 Spring 3.70 5.67 1.60 Ye et al., 2012 Taihu China Meiliang Bay 31� 280 120� 100 22500 264 1.90 Summer 5.13 15.0 4.20 Ye et al., 2012 Taihu China Lake centre 31� 140 120� 070 22500 264 1.90 Spring 3.97 1.17 1.30 Ye et al., 2012 Taihu China Lake centre 31� 140 120� 070 22500 264 1.90 Summer 4.60 7.00 1.40 Ye et al., 2012 Taihu China Gonghu Bay 31� 230 120� 180 22500 264 1.90 Spring 4.03 1.50 1.20 Ye et al., 2012 Taihu China Gonghu Bay 31� 230 120� 180 22500 264 1.90 Summer 4.67 7.33 1.30 Ye et al., 2012 Mendota USA 43� 05’ 89� 220 3985 80 12.7 All year 4.52 50.6 e Fallon and Brock, 1979, Pedros-Alio and Brock, 1982 Miastro Belarus 54� 510 26�440 1310 913 5.40 All year 7.03 20.0 e Ostapenya et al., 2009

Batorino Belarus 54� 510 26�440 630 365 3.00 All year 9.75 50.1 e Ostapenya et al., 2009 Kinneret Israel 32� 500 35�350 17000 1825 25.6 All year 3.60 18.0 e Serruya, 1978, Yacobi, 2006 Vortsjarv Estonia 58� 170 26� 020 27000 365 2.80 Winter 15.0 16.4 10.3 Current study Vortsjarv Estonia 58� 170 26� 020 27000 365 2.80 Spring 13.0 15.6 9.98 Current study Vortsjarv Estonia 58� 170 26� 020 27000 365 2.80 Summer 16.8 47.1 6.67 Current study Vortsjarv Estonia 58� 170 26� 020 27000 365 2.80 Autumn 16.1 51.4 6.04 Current study

relationship between DOC and CDOM. On the other hand, Kutser however, that in optically complex waters the coloured fraction of et al. (2015b) found that some MERIS processors cope well with DOC varies widely and the prediction of DOC concentrations from the variable carbon-iron ratios and are suitable for predicting lake CDOM levels alone (either measured in the laboratory or by remote DOC. According to Brezonik et al. (2015) it can be concluded, sensing), is subject to a substantial uncertainty. Until reaching a better understanding of variations in DOCeCDOM relationships, ﬁeld sampling for calibration is essential for verifying DOC concentrations predicted from remotely sensed CDOM measurements. 18 DOC = 3.04 + 1.33*CDOM (a) In the analysis among different eutrophic lakes in the world ) r = 0.85, p < 0.001

-1 (Table 5), CDOM appeared to be a very good predictor for DOC l concentration (Fig. 5a) and thus the remote sensing of CDOM can be C 14

g applied as a proxy to estimate large-scale patterns of DOC con- m ( 10 centrations in eutrophic lakes. DOC correlation with Chl a was C statistically signiﬁciant, although rather weak (Fig. 5b) and CDOM O

D 6 should be preferred to Chl a to predict DOC in eutrophic lakes. In such large-scale assessments the data averaging smooths the 2 abrupt changes typical for seasonal studies (Reche and Pace, 2002). 024681012 Furthermore, Pace and Cole (2002) found in their study of 20 lakes CDOM (mg l-1) that the major variation of DOC and CDOM appears among systems and within-lake temporal variation is less pronounced. The among- lake variation of DOC and CDOM primarily reﬂects differences in 18 DOC = 6.58 + 0.11*Chl a (b) watershed size and structure that are connected to loading, lake

) r = 0.45, p = 0.04

-1 size and shape that inﬂuence in-lake residence time and degrada- l

C 14 tion. At the within-lake scale, watershed and lake structure are g ﬁxed and the principal source of variation is the inﬂuence of cli- m ( 10 matic processes such as ice-out timing, precipitation, and drought, C

O which in turn affect loading, washout, and in-lake production and

D 6 degradation. Above mentioned together with stable DOC concentrations in Lake Vortsj~ arv€ might be the reasons why CDOM was not 2 a good proxy for Vortsj~ arv€ alone, but appeared to be a very good 0 10 20 30 40 50 60 proxy in the large scale study. Chl a (μg l-1)

Fig. 5. Relationship between dissolved organic carbon (DOC) with a: coloured dis- 5. Conclusions solved organic matter (CDOM) and b: chlorophyll a (Chl a) in eutrophic lakes shown in Table 5. The values of Lake Vortsj~ arv€ (mean values of DOC, CDOM and Chl a in spring, The concentrations of DOC and CDOM were rather high in Lake summer, autumn and winter in 2008e2014) are marked with four black dots. � Vortsj~ arv€ being almost two to three times higher than in other

171 K. Toming et al. / Water Research 102 (2016) 32e40 39

analysed eutrophic lakes, but the seasonal and interannual Evans, C.D., Monteith, D.T., Cooper, D.M., 2005. Long-term increases in surface water variability of the concentrations of DOC was rather small. dissolved organic carbon: observations, possible causes and environmental impacts. Environ. Pollut. 137, 55e71. The relationships between DOC and other water and environ- Fallon, R.D., Brock, T.D., 1979. Decomposition of bluegreen algae (cyanobacterial) mental variables varied seasonally and interannually and it was blooms in lake Mendota, Wisconsin. Appl. Environ. Microbiol. 37, 820e830. difficult to find a remote sensing proxy for describing seasonal Filella, M., Rodríguez-Murillo, J.C., 2014. Long-term trends of organic carbon concentrations in freshwaters: strengths and weaknesses of existing evidence. cycle of DOC. In ice free period, Chl a, and TSM or S should be Water 6, 1360e1418. applied as proxies for remote sensing of DOC in Lake Vortsj~ arv.€ Højerslev, N.K., 1980. On the origin of yellow substance in the marine environment. In long term interannual studies transparency-related variables Oceanogr. Rep. Univ. Cph. Inst. Phys. 42, 1e35. Jaani, A., 1990. The water regime and water balance of Lake Vortsj~ arv.€ Eesti Loods 11, are relevant as DOC proxies. 743e747. CDOM did not appear to be a good predictor of the seasonality of Jiang, G., Ma, R., Loiselle, S.A., Duan, H., 2012. Optical approaches to examining the DOC concentration in eutrophic Lake Vortsj~ arv€ since the dynamics of dissolved organic carbon in optically complex inland waters. En- viron. Res. Lett. 7, 034014. CDOMeDOC coupling varies seasonally. Jones, R.I., Arvola, L., 1984. Light penetration and some related characteristics in In large scale (eutrophic lakes Balaton, Taihu, Miastro, Batorino, small forest lakes in southern Finland. Verh. Intern. Ver. Limnol. 22, 811e816. Mendota, Kinneret and Vortsj~ arv)€ assessment CDOM appeared Kallio, K., 1999. Absorption properties of dissolved organic matter in Finnish lakes. to be powerful predictor of DOC and can be applied for remote Proc. Est. Acad. Sci. Biol. Ecol. 48, 75e83. Kirk, J.T.O., 1980. Spectral absorption properties of natural waters: contribution of sensing of large-scale patterns of DOC concentrations in eutro- the soluble and particulate fractions to light absorption in some inland waters phic lakes. of south eastern Australia. Aust. J. Mar. Freshw. Res. 31, 287e296. Kohler,€ S.J., Kothawala, D., Futter, M.N., Liungman, O., Tranvik, L., 2013. In-lake processes offset increased terrestrial inputs of dissolved organic carbon and Acknowledgements color to lakes. PLoS One 8, e70598. Kortelainen, P., 1999. Occurrence of humic waters. Temporal and spatial variability. In: Keskitalo, J., Eloranta, P. (Eds.), Limnology of Humic Waters. Backhuys This work was funded by the Estonian Ministry of Education and Publishers, Leiden, The Netherlands, pp. 46e55. Research (IUT 21-02, SF0180009s11), Estonian Science Foundation Kutser, T., Alikas, K., Kothawala, D.N., Kohler,€ S.J., 2015b. Impact of iron associated to (grant 9102), MARS project (Managing Aquatic ecosystems and organic matter on remote sensing estimates of lake carbon content. Remote e water Resources under multiple Stress) funded by the European Sens. Environ. 156, 109 116. Kutser, T., Pierson, D.C., Tranvik, L., Reinart, A., Sobek, S., Kallio, K., 2005. Using Union under the 7th Framework Programme, Theme 6. (Environ- satellite remote sensing to stimate the coloured dissolved organic matter ab- ment including Climate Change), contract no. 603378, http://www. sorption coefficient in lakes. Ecosystems 8, 709e720. mars-project.eu) and Swiss Grant for Programme ‘Enhancing public Kutser, T., Tranvik, L., Pierson, D.C., 2009. Variations in coloured dissolved organic matter between boreal lakes studied by satellite remote sensing. J. Appl. environmental monitoring capacities’. The Estonian Ministry of Remote Sens. 3, 033538. Environment and Estonian Environment Agency supported data Kutser, T., Verpoorter, C., Paavel, B., Tranvik, L.J., 2015a. Estimating lake carbon collection in the state monitoring programs. We are grateful to two fractions from remote sensing data. Remote Sens. Environ. 157, 138e146. Leathwick, J.R., Elith, J., Chadderton, W.L., Rowe, D., Hastie, T., 2008. Dispersal, anonymous reviewers and the editor for their comments, which disturbance, and the contrasting biogeographies of New Zealand’s diadromous greatly improved the manuscript. and non-diadromous fish species. J. Biogeogr. 35 (8), 1481e1497. Leathwick, J.R., Elith, J., Francis, M.P., Hastie, T., Taylor, P., 2006. Variation in demersal fish species richness in the oceans surrounding New Zealand: an References analysis using boosted regression trees. Mar. Ecol. Prog. Ser. 321, 267e281. Lindell, M.J., Rai, H., 1994. Photochemical oxygen consumption in humic waters. Arvola, L., Salonen, K., Rask, M., 1999. Food webs of humic waters. Trophic in- Arch. Hydrobiol. 43, 145e155. teractions. In: Keskitalo, J., Eloranta, P. (Eds.), Limnology of Humic Waters. Molot, L.A., Dillon, P.J., 1997. Colour e mass balances and colour e dissolved organic Backhuys Publishers, Leiden, The Netherlands, pp. 265e276. carbon relationships in lakes and streams in central Ontario. Can. J. Fish. Aquat. Bertillson, S., Jones, J.B., 2003. Supply of dissolved organic matter to aquatic eco- Sci. 54, 2789e2795. systems: autochthonous sources. In: Findlay, S.E.G., Sinsabaugh, R.L. (Eds.), Niu, C., Zhang, Y., Zhou, Y., Shi, K., Liu, X., Qin, B., 2014. The potential applications of Aquatic Ecosystems: Interactivity of Dissolved Organic Matter. Academic Press, real-time monitoring of water quality in a large shallow lake (Lake Taihu, China) pp. 3e25. using a chromophoric dissolved organic matter fluorescence sensor. Sensors 14, Brezonik, P.L., Olmanson, L.G., Finlay, J.C., Bauer, M.E., 2015. Factors affecting the 11580e11594. measurement of CDOM by remote sensing of optically complex inland waters. Noges,~ T., Mischke, U., Laugaste, R., Solimini, A.G., 2010. Analysis of changes over 44 Remote Sens. Environ. 157, 199e215. years in the phytoplankton of Lake Vortsj~ arv€ (Estonia): the effect of nutrients, Clark, J.M., Bottrell, S.H., Evans, C.D., Monteith, D.T., Bartlett, R., Rose, R., Newton, R.J., climate and the investigator on phytoplankton-based water quality indices. Chapman, P.J., 2010. The importance of the relationship between scale and Hydrobiologia 646, 33e48. http://dx.doi.org/10.1007/s10750-010-0178-y. process in understanding long-term DOC dynamics. Sci. Total Environ. 408, Ostapenya, A.P., Parparov, A., Berman, T., 2009. Lability of organic carbon in lakes of 2768e2775. different trophic status. Freshw. Biol. 54, 1312e1323. Davies-Colley, R.J., Vant, W.N., 1987. Absorption of light by yellow substance in Pace, M.L., Cole, J.J., 2002. Synchronous variation of dissolved organic carbon and freshwater lakes. Limnol. Oceanogr. 32, 416e425. color in lakes. Limnol. Oceanogr. 47 (2), 333e342. De Haan, H., 1992. Impacts of environmental changes on the biogeochemistry of Pedros-Alio, C., Brock, T., 1982. Assessing biomass and production of bacteria in aquatic humic substances. Hydrobiologia 229 (1), 59e71. eutrophic Lake Mendota, Wisconsin. Appl. Environ. Microbiol. 44, 203e218. Del Castillo, C.E., Miller, R.L., 2008. On the use of ocean color remote sensing to Reche, I., Pace, M.L., 2002. Linking dynamics of dissolved organic carbon in a measure the transport of dissolved organic carbon by the Mississippi River forested lake with environmental factors. Biogeochemistry 61, 21e36. Plume. Remote Sens. Environ. 112, 836e844. Serruya, C., 1978. Lake Kinneret. Monographiae. Biologicae. Dr W. Junk, The Hague, Edler, L., 1979. Recommendations for marine biological studies in the Baltic Sea. 501 pp. Phytoplankt. Chlorophyll Baltic Mar. Biol. 5, 1e38. Sipelgas, L., Arst, H., Kallio, K., Erm, A., Oja, P., Soomere, T., 2003. Optical properties Eikebrokk, B., Vogt, R.D., Liltved, H., 2004. NOM increase in Northern European of dissolved organic matter in Finnish and Estonian lakes. Nord. Hydrol. 34, source waters: discussion of possible causes and impacts on coagulation/con- 361e386. tact filtration processes. Water Sci. Technol. Water Supply 4 (4), 47e54. Thurman, E.M., 1985. Organic Geochemistry of Natural Waters. Martinus Nijhoff/Dr Elith, J., Graham, C.H., Anderson, R.P., Elith, J., Graham, C.H., Anderson, R.P., W. Junk Publishers, Dordrecht and Boston. Dudík, M., Ferrier, S., Guisan, A., Hijmans, R.J., Huettmann, F., Leathwick, J.R., Toming, K., Tuvikene, L., Vilbaste, S., Agasild, H., Kisand, A., Viik, M., Martma, T., Lehmann, A., Li, J., Lohmann, L.G., Loiselle, B.A., Manion, G., Moritz, C., Jones, R., Noges,~ T., 2013. Contributions of autochthonous and allochthonous Nakamura, M., Nakazawa, Y., Overton, J. McC., Peterson, A.T., Phillips, S.J., sources to dissolved organic matter in a large, shallow, eutrophic lake with a Richardson, K.S., Scachetti-Pereira, R., Schapire, R.E., Soberon,� J., Williams, S., highly calcareous catchment. Limnol. Oceanogr. 58 (4), 1259e1270. Wisz, M.S., Zimmermann, N.E., 2006. Novel methods improve prediction of Toth,� N., Vor€ os,€ L., Mozes,� A., V.-Balogh, K., 2007. Biological availability and humic species’ distributions from occurrence data. Ecography 29, 129e151. properties of dissolved organic carbon in Lake Balaton (Hungary). Hydro- Elith, J., Leathwick, J.R., Hastie, T., 2008. A working guide to boosted regression biologia 592, 281e290. trees. J. Anim. Ecol. 77, 802e813. http://dx.doi.org/10.1111/j.1365- Tranvik, L.J., 1990. Bacterioplankton growth on fractions of dissolved organic carbon 2656.2008.01390. of different molecular weights from humic and clear waters. Appl. Environ. Erlandsson, M., Futter, M.N., Kothawala, D.N., Kohler,€ S.J., 2012. Variability in Microbiol. 56, 1672e1677. spectral absorbance metrics across boreal lake waters. J. Environ. Monit. 14, Tranvik, L.J., 1992. Allochthonous dissolved organic matter as an energy source for 2643e2652. pelagic bacteria and the concept of the microbial loop. Hydrobiologia 229 (1),

172 40 K. Toming et al. / Water Research 102 (2016) 32e40

107e114. phytoplankton photosynthetic activity and light attenuation in Lake Kinneret: Wetzel, R.G., 2001. Limnology: Lake and River Ecosystems, third ed. Academic Press, possibilities and limitations for simulation by remote sensing. J. Plankton Res. London. 28, 725e736. Xiao, Y.H., Sara-Aho, T., Hartikainen, H., Vah€ atalo,€ A.V., 2013. Contribution of ferric Ye, L., Shi, X., Wu, X., Kong, F., 2012. Nitrate limitation and accumulation of dissolved iron to light absorption by chromophoric dissolved organic matter. Limnol. organic carbon during a spring-summer cyanobacterial bloom in Lake Taihu Oceanogr. 58, 653e662. (China). J. Limnol. 71 (1), 67e71. http://dx.doi.org/10.4081/jlimnol.2012.e6. Yacobi, Y.Z., Alberts, J.J., Takacs, M., McElvaine, M., 2003. Absorption spectroscopy of Zhang, Y., Qin, B., Zhu, G., Zhang, L., Yang, L., 2007. Chromophoric dissolved organic colored dissolved organic carbon in Georgia (USA) rivers: the impact of mo- matter (CDOM) absorption characteristics in relation to ﬂuorescence in Lake lecular size distribution. J. Limnol. 62, 41e46. Taihu, China, a large shallow subtropical lake. Hydrobiologia 581, 43e52. Yacobi, Y.Z., 2006. Temporal and vertical variation of chlorophyll a concentration,

173

V Kauer, T., Arst, H., Nõges, T. & Tuvikene, L. 2009. Estimation of the phytoplankton productivity in three Estonian lakes. Estonian Journal of Ecology, 58(4): 297–312, doi: 10.3176/ eco.2009.4.05.          

        

        

                                  

       

                                                                                                                       

            



                                                                                                                                                                                                                    



177    

                                                                    α β                                                                                                                                                                                                                                                              

    

                                                                             ′   ′                     ′   ′                                     



178    

                                           

                                      

                 ′  ′                                 



                                 

•         = +                                                                                                                  ∆λ   

      ∆λ  = +=   ∆λ    =+    ∆λ                   •          λ   

  ∆λ  =−=+−    ∆λ    ∆λ             ∆λ                ∆λ      •     ∆λ                  

       ∆λ                                             



179    

                                 

   ∆λ                        •                                                                •                                                                                                                                                          

    

                                 

 = Ψ       

         Ψ                                                                                 

  = λ ′ λλ   ∫           

  λ              ′ λ                



180    

      ′ λ          

−λ ′ λλ=    

     ′ λ                                

                        

 =       +   

                                                                                =                                                                                                                                      =                     λ    λ                                ′ λ                                             

               

            =                            

        −=−            −=−          



181    

        

                      

               

                          =                



182    

                                                                                                                    

                                     

                                



183    

                                          =            =         

       

                                                                                                                                                                                                                                                                                                                                                      

                                            

                                     

                       



184 Estimation of phytoplankton productivity

Fig. 4. Variation of the spectral quantum irradiance q(! ,, tz ) from morning to evening in Lake Võrtsjärv.

305

185 T. Kauer et al.

Fig. 5. Variation of the spectral quantum irradiance q(! ,, tz ) from morning to evening in Lake Harku.

306

186 Estimation of phytoplankton productivity

Fig. 6. Primary production Pzt(,) in dependence of time and depth for (a) Lake Peipsi on 27 June 3 3 2008, CChl = 8.8 mg m and (b) Lake Võrtsjärv on 17 July 2007, CChl = 43 mg m . In both cases there was variable cloudiness.

307

187 T. Kauer et al.

Fig. 7. Primary production Pzt(,) in dependence of time and depth for (a) Lake Võrtsjärv on 3 17 September 2007 (cloudy weather, low illumination), CChl = 82 mg m and (b) Lake Harku on 3 23 July 2008, CChl = 155 mg m (in general sunny weather with some clouds in the afternoon).

308

188    

                         

                                                            

                  



                                                                                                                                                                                                   λ  λ                                                  



189    

                                                                      



                                                                                                                                                        



                                                



                                                                                                               



190    

                                                                                                                                                                                                                                                                                                                 ł                                                                                                                                                                                                                                                        



191    

                                                                                                                                                                                                                                                                                                                                                   

     

        

                                                                                                   



192 VI Reinart, A., Paavel, B. & Tuvikene, L. 2004. Effects of coloured dissolved organic matter on the attenuation of photosynthetically active radiation in Lake Peipsi. Proceedings of the Estonian Academy of Sciences. Biology, Ecology, 53(2): 88–105. Proc. Estonian Acad. Sci. Biol. Ecol., 2004, 53, 2, 88–105

Effect of coloured dissolved organic matter on the attenuation of photosynthetically active radiation in Lake Peipsi

Anu Reinarta*, Birgot Paavelb, and Lea Tuvikenec a Department of Limnology, Uppsala University, Norbyvägen 20, 752 36, Uppsala, Sweden b Estonian Marine Institute, University of Tartu, Mäealuse 10A, 12618 Tallinn, Estonia c Võrtsjärv Limnological Station, Institute of Zoology and Botany, Estonian Agricultural University, 61101 Rannu, Tartumaa, Estonia

Received 12 March 2004, in revised form 30 March 2004

Abstract. On the basis of underwater radiation measurements and laboratory analyses of water samples, the effect of coloured dissolved organic matter (CDOM) on the light field in the largest Estonian lake, Lake Peipsi, was investigated. CDOM variation and its optical properties were compared with two other large European lakes, Vänern and Vättern in Sweden, as well as with 41 small lakes in Estonia, Sweden, and Finland. The light absorption coefficient at 380 nm for filtered water varied from 4.1 to 16.7 m–1 with the highest values close to the inflow of the Suur-Emajõgi River. CDOM values in L. Peipsi are in the same range as measured earlier in small Estonian and Finnish lakes. Various optical water types (“brown”, “moderate”, “turbid”) were sampled over L. Peipsi. It is shown that the use of GF/F filters results in on average 4% higher absorption values than the use of filters with pore size 0.2 m, and the effect might be up to ±20% in clear waters. Comparison of spectral measurements of diffuse attenuation coefficient with three-band spectrometer’s data showed high consistence of datasets. This allows supplementing the existing database of optical properties of lakes in Estonia with new data from L. Peipsi.

Key words: light field, optical properties, dissolved organic matter.

INTRODUCTION

Lake Peipsi on the border of Estonia and Russia is a large northern lake, unique in its nature – shallow, eutrophic, biologically very productive, and bordered by many wetland areas along its coast. About 240 rivers and streams fall into L. Peipsi

* Corresponding author, [email protected]

195 and there is only one outflow, the Narva River, carrying the water from L. Peipsi to the Gulf of Finland (Nõges, 2001). The underwater light field determines components of ecological conditions in water: primary production (Krause-Jensen & Sand-Jensen, 1998), species composition of the phytoplankton communities (Schanz, 1985), depth distribution of submerged macrophytes (Duarte, 1991), and heat budget of the water body (Fedorov & Ginsburg, 1992).

The diffuse attenuation coefficient, (Kd ), characterizes the gradient of the vertical decrease of irradiance in the water. It depends on surface and illumination conditions, but also on concentrations of optically active substances (OAS) in water (Dera, 1992). This parameter is widely used in practice to characterize the light field and propagation of photosynthetically active radiation (PAR, spectral range 400–700 nm) inside water, because it is rather easily determined by standard and commercially available instruments, and also closely related to inherent optical properties of water (Kirk, 1994; Mobley, 1994). The water in Lake Peipsi is rich in all OAS: phytoplankton, suspended mineral particles, and dissolved organic matter (DOM). In natural water bodies DOM is not identifiable as a distinct molecule, but it is a rather indeterminate mixture of dissolved organic substances (Dera, 1992). Its chromophore-containing compounds contribute significantly to the total light absorption in the water. These OAS are mostly referred to as coloured dissolved organic matter (Hoge et al., 1993; Kallio, 1999) or chromophoric dissolved organic matter, CDOM (Miller et al., 2000). In northern inland waters DOM is an important variable of the water ecosystem as it affects the water colour and quality, it captures energy that could be available for photosynthesis, and influences photochemical mineralization of dissolved organic carbon (Davies-Colley & Vant, 1987; Jansson, 1998; Kallio, 1999). Concentrations of DOM are especially high in boreal and northern lakes, where it is transported to lakes from soil leaching and surface water runoff (mainly rivers). The lakes influenced by high amounts of DOM develop carbon dioxide supersaturation of the water and export of carbon dioxide to the atmosphere (Cole et al., 1994; Sobek et al., 2003). Released carbon dioxide on the other hand has an effect for climate change (Freeman et al., 2001). The concentration of DOM in freshwater environments has implications for the further transport to the marine environment and therefore has a significant influence on coastal ecosystems. CDOM (also called yellow substance and gilvin) refers to any type of DOM, regardless of its origin, which has optical effect (Bricaud et al., 1981) and can be measured by optical methods (Lindell et al., 1999). Its amount in water is commonly expressed as the absorption coefficient of filtered water at some fixed wavelength and a parameter describing spectral variation (Davies-Colley & Vant, 1987; Dera, 1992; Gege, 2000). We investigated the effect of CDOM on the light field in Lake Peipsi, relying on the data obtained in the course of field measurements of downwelling and

196 upwelling irradiances, as well as by laboratory analyses of water samples. For comparison analogous data from two other large European lakes (Vänern and Vättern in Sweden) and numerous small lakes and coastal waters in Estonia and Finland were used. As the datasets were collected using different instruments and different methods, special attention was paid to comparative analyses.

MATERIALS AND METHODS

Field campaigns to measure optical properties were carried out on 11–18 and 26 June, 18 Aug, and 4 Nov 2003 in Lake Peipsi s.s. and in narrow Lake Lämmi- järv (Fig. 1). As the third basin of Lake Peipsi – Lake Pihkva (Pskov) – belongs almost entirely to Russia it was not visited by us in 2003. Underwater irradiance measurements were made using two instruments. One of them was a GER 1500 (capable of measuring in the upward and downward directions, spectral range 400–900 nm, resolution 2 nm) mounted into a water- proof box. The measurements setup was the same as used in deep Swedish lakes

(Strömbeck, 2001). The vertical diffuse attenuation coefficient, Kd ( ), can be

59.1

58.9 L. Peipsi s.s.

58.7

58.5

58.3

Latitude, N Latitude, L. Lämmijärv

58.1 L. Pihkva

57.9

57.7 26.8 27.3 27.8 28.3 Longitude, E

Fig. 1. Scheme of L. Peipsi with measurement sites during the field campaign in 2003: dots show points for water samples, circles for radiation measurements, and crosses for spectral measurements.

197 calculated for the depth interval from the measured irradiances at two different depths (z1 and z2 ) using the equation:

1 Ed ( , z2 ) Kd ( ) ln , (1) z2 z1 Ed ( , z1)

where Ed ( , z1) and Ed ( , z2 ) are the values of irradiances at depths z1 and z2 . During our field trips it was possible to use this instrument setup only in the deepest places (7–9 m) in the centre of L. Peipsi s.s. (Fig. 1; 3 measurements), as a special ship was needed to lower the frame with all instruments (Reinart et al., 2004). Another instrument that we used for measuring underwater irradiance was a profiling radiometer BIC-2104 (Biospherical Instruments Inc.) with three bands (412, 555, 665 nm), a PAR sensor (units Einsteins m–2 s–1), and an additional surface PAR sensor. This instrument is easy to handle and suitable for measurements even in very shallow water (at least ~20 cm). Data were collected continuously during lowering and lifting the instrument and stored together with the respective depth (precision ±2 cm) and temperature data into a laptop computer. Immediately after vertical profiling dark current values were also measured during 5 s and subtracted from data according to the measurement instructions provided by the producer. Irradiance data were corrected for changes in incident light according to Virta & Blanco-Sequeiros (1995). The depth-averaged values of

Kd ( ) and Kd,PAR (respectively spectral and PAR diffuse attenuation coefficients) were determined in the following way: irradiance values down to the depth where the value is 1% of the subsurface value were fitted by least-squares to a straight line on a semilog plot, the slope of which is determined by Kd . The determination coefficient of logarithmic fit was always higher than 0.97. The measurements with the BIC-2104 were made in 11 points in L. Peipsi (Fig. 1).

The effect of a dark current on the value of Kd was investigated separately. Application of a dark current correction yielded 12.7–37.2% higher Kd in the 412 nm band; the effect in other channels was much lower, being –2.6 to +0.4% in the 555 nm band and –2.8 to +0.2% in the 665 nm band. The corrected and uncorrected results differed by –17.4 to +0.4% in the PAR band. The differences were highest in the case of low values of irradiance (cloudy day and/or late time of day). Water samples (16) were collected only from the surface layer (0.2 m). A standard water sampler was used and the samples were stored in dark and cold for less than 7 h before filtering. The relative transparency of water was measured using Secchi disc during the time of sampling. For chlorophyll concentrations

(Cchl ) 0.5–1.0 L of water was filtered through Whatman GF/F filters (pore size 0.6–0.7 m). The chlorophyll a + phaephytin a concentration was measured spectrophotometrically on ethanol extracts of filters according to the ISO 10260 standard method (Lindell et al., 1999). The concentration of suspended matter, CSM , was measured gravimetrically after filtration of the same amount of water through

198 pre-weighed and pre-combusted Whatman GF/F filters; the inorganic fraction,

CSIM , was measured after combustion at 550°C for 30 min. The light absorption coefficient of filtered water was measured spectrophotometrically using a PERKIN ELMER Lambda 40 UV/VIS spectrometer (in the range 300–900 nm) in a 10 cm cuvette as the difference between the sample and * distilled water, the result is denoted as c f . The values of the absorption coefficient for yellow substance, ays , can be estimated in the following way (Bricaud et al., 1981):

* ays ( ) c f ( ) ( ), (2) where ( ) is the correction for residual scattering:

* g ( ) c f ( R )( R ) . (3)

Usually R is taken equal to 750 nm, but for the parameter g the values 0, 1, and 2 have been proposed (Bricaud et al., 1981; Davies-Colley & Vant, 1987; Gallie, 1994; Mäekivi & Arst, 1996; Kallio, 1999; Aas, 2000; Sipelgas et al., 2003).

In the present study we use the values R 750 nm and g 1. The absorption spectra were approximated by a linear regression between the logarithm of ays ( ) and the wavelengths between 380 and 550 nm (e.g. Bricaud et al., 1981):

S ( 0 ) ays ( ) ays ( 0 )e , (4)

–1 where ays ( 0 ) is the absorption coefficient (in m ) at reference wavelength 0 (in nm), and S is the slope factor of the spectrum. We calculated the factor S over the spectral range 380–550 nm and took 0 420 nm as suitable for para- meterization of the simple bio-optical model elaborated by Pierson & Strömbeck (2001). For comparison with other studies in Estonian lakes made by Arst et al.

(1999), Sipelgas et al. (2003), and Reinart et al. (2003), ays values at both 420 and 380 nm are used in the present study. We made repeated measurements using Whatman GF/F (pore size 0.6–0.7 m) and Whatman polyethersulfone filters (pore size 0.2 m), as in many previous studies elsewhere and especially in Swedish, Estonian, and Finnish lakes (Mäekivi & Arst, 1996; Östlund et al., 2001; Strömbeck & Pierson, 2001; Sipelgas et al., 2003) filters with different pore size were used. Additionally to 16 L. Peipsi water samples, we analysed water samples collected from 20 southern Swedish lakes (26 June to 7 July 2003), 6 southern Estonian lakes (11–13 June 2003), 15 southern Finnish lakes (9–19 June 2002), 12 samples from L. Vättern (29–31 May 2002), and 21 samples from L. Vänern (21–23 May 2002). Data on OAS in all investigated water bodies (102 samples in total) are shown in Table 1. Samples from L. Vänern were filtered first through GF/F filters and only after two months analyses were repeated for 0.2 m filters.

199 Table 1. Mean values and respective standard deviations for some bio-optical parameters in lake waters obtained during measurement campaigns in 2002–03

Cchl, CSM, CSIM, ays(380), S(380–550) Secchi mg m–3 g m–3 g m–3 m–1 depth, m

L. Peipsi 12.7 ±14.8 9.5 ±7.5 4.9 ±4.1 6.93 ±3.95 0.018±0.001 1.7 ±0.9 Estonian lakes* 21.1 ±20.1 6.3 ±5.9 1.2 ±1.5 4.92 ±3.22 0.016±0.002 2.7 ±1.8 L. Vänern 3.6 ±1.5 1.2 ±0.4 0.4 ±0.3 3.63 ±1.31 0.014±0.002 4.0 ±0.6 L. Vättern 1.8±0.6 0.6 ±0.2 0.1 ±0.1 0.95 ±0.21 0.014±0.002 9.1 ±0.4 Finnish lakes* 13.7 ±14.2 5.9 ±7.3 0.3 ±0.2 7.80 ±4.26 0.016±0.004 2.4 ±1.6 Small Swedish lakes 2.18 ±2.98 3.0 ±2.6 1.1 ±2.0 12.03±7.60 0.015±0.001 2.32 ±1.26 ______* Data from small lakes database (1994–2000) are included.

For comparison of the results obtained for L. Peipsi with other lakes, we used the database of small Estonian and Finnish lakes collected by the marine optics workgroup, Estonian Marine Institute, in cooperation with the Department of Geo- physics, University of Helsinki, during the years 1994–2000 (published partly by Reinart & Herlevi, 1999; Arst, 2003; Reinart et al., 2003). We processed raw data to get values of Kd ( ) from underwater irradiance measurements with a spectroradiometer LI-1800 UW. LI-1800 UW measures in the range 300–850 nm, with a resolution of 2 nm (W m–2 nm–1), at depths 0.5–6 m, depending on water properties and light conditions. Changes in underwater irradiance due to variation of incident irradiance (time, cloud cover) were taken into account using simultaneously recorded air pyranometer data. From these spectra the averaged over depth attenuation coefficient and integral over PAR region (Kd,PAR ) were estimated also by logarithmic fit irradiance vs. depth data. Mostly the values of

Kd for the 0.5–2 m layer were used. Data with exceptionally high (or low) values were excluded as these contained more errors and noise. All together 84 spectra (400–800 nm) were analysed. This database includes also values for OAS, which are shown together with our samples in Table 1.

RESULTS AND DISCUSSION Absorption by yellow substance

Over L. Peipsi 16 points were sampled at different distances from the mouth of the Suur-Emajõgi River towards the open area of the lake (0.5–15 km), in the shallow area around Piirissaare Island, and in L. Lämmijärv. The values of ays (380) for samples that were taken at least 2.9 km from the shore were lower (4.1–5.7 m–1), being therefore slightly lower than the average for many Estonian and Finnish lakes (6.6 m–1) estimated earlier by Sipelgas et al. (2003). The

200 concentration of CDOM in the water of L. Vänern was close to the lowest limit of L. Peipsi, while L. Vättern had a really low absorption by CDOM (Table 1). As can be expected, the highest value of ays (380) in L. Peipsi was measured in the point closest to the Suur-Emajõgi River mouth (Fig. 2, at 634 m). The other points in Fig. 2 with high values of ays (380) were sampled up to 12 km from the river mouth but close to the shore or Piirissaare Island. There is no systematic variation in the slope parameter S(380–550) estimated from samples. It varies from 0.015 to 0.019 nm–1 and is in the same range as reported before for Estonian lakes (Arst, 2003; Sipelgas et al., 2003). In L. Vänern and L. Vättern S(380–550) values are slightly lower, but there is also rather large variability among samples. For L. Vättern the average value reported earlier is much higher (0.017 nm–1 by Strömbeck, 2001) than the value estimated by our latest measurements (Table 1) using the same method. However, in L. Vättern there was also large variation over a limited number of samples (15 samples and the range of S(380–550) was 0.009–0.033 nm–1, by N. Strömbeck, 2004, pers. comm.). In L. Peipsi the lowest value of S(380–550) was measured in the closest point to the shore (Fig. 2); however, by this one point we cannot conclude that there is really variation in the types of DOM. In natural waters large variation of S values from 0.004 up to 0.053 nm–1 over seasons, locations, and water types have been reported (overview in Aas, 2000). A rather thorough investigation on the optical properties of DOM was carried out by Sipelgas et al. (2003). The measurement data in the period 1994–99 for 13 Estonian and 11 Finnish small lakes (altogether 462 spectra) were used. The mathematical approximation of ays ( ) spectra as well as the values of slope factor were studied. The respective range of S for Estonian lakes (different spectral intervals and different values of g) was 0.006–0.030 (Sipelgas et al., 2003).

ays(380) S(380-550) 18 0.02

15 0.015 S (380-550), nm (380-550), -1 12

9 0.01 (380), m

ys 6 a 0.005 -1 3

0 0 353 506 634 795 1201 1938 2889 2910 2955 2987 3134 5034 6263 9263 14006 14073 Distance from shore, m

Fig. 2. Variation of ays(380) and S(380–550) in L. Peipsi obtained by measurements in 2003.

201 Some authors note a general trend of the shape factor to increase with decreasing yellow substance absorption in a large range of lakes and sea areas (Davies-Colley & Vant, 1987; Aas, 2000). Rather often the slope factor is high in lakes where the amount of yellow substance is large (Herlevi et al., 1999; Arst, 2003); this is also the case in L. Peipsi (0.0176 nm–1 compared with 0.0144 nm–1 in L. Vänern; Table 1). This variation may be caused by different organic compounds forming yellow substance (Kallio, 1999; Miller et al., 2000). As shown by Sipelgas et al. (2003), the slope factor is very much affected by the spectral range for which it is calculated. The slope parameter may be ~30% lower when calculated over the spectral range 350–700 nm compared with the range 350–500 nm as attenuation at longer wavelengths is generally small. Correction for residual scattering also increases S values by 10–20% depending on the correction type (parameter g may be 0, 1, or 2). Analyses of the effect of the pore size of the filter on the absorption coefficient showed a very good correlation between data from GF/F (pore size 0.6–0.7 m) and 0.2 m filters (Fig. 3). The values estimated using GF/F filters were on average 4% larger than those obtained using filters with the small pore size; however, the correlation coefficient was very high (R2 0.99) between the two datasets

(Fig. 3a). Differences in absorption values are larger in waters whose ays (380) values are less than 2.5 m–1 (±20%) because of measurement errors. In clear- water lakes ays (380) may be even higher after the second filtering as micro bubbles cause additional scattering (Gege, 2000). In turbid waters ays (380) values are lower when smaller pore size is used as more particles are removed from the –1 water. The difference is less than 4% when ays (380) is higher than 8 m . The two months’ delay in the second filtering analyses (L. Vänern data) brought about more scattered results; however, they are still close to other data. A decrease

Fig. 3. Comparison of results obtained using filters with pore size 0.6–0.7 m and 0.20 m: (a) absorption coefficient at 420 nm and (b) slope parameter S(380–550) in lakes in Sweden, Finland, and Estonia. Samples for which filtering was delayed are not included into calculation.

202 in ays (380) values after filtering samples through 0.2 m pore filters was also found in a few samples from eutrophic lakes (Sipelgas et al., 2003). The slope parameter values estimated from both datasets (GF/F and 0.2 m) vary more irregularly (Fig. 3b). Analyses with smaller pore size filters gave a higher slope (in L. Peipsi 0.0176±0.0009) compared with larger pore size analyses (in L. Peipsi 0.0162±0.0007). Again, samples that were analysed later differed more from the other samples (standard deviation 0.002), probably because some colloidal material was formed. The small negative intercept value in the relationship between absorption values (Fig. 3a) and the positive intercept in the relationship between slope values (Fig. 3b) indicate that some small particles had really remained in water after filtering through a GF/F filter and that these were removed by second filtering. Bio-optical models by different authors use different wavelengths to para- meterize yellow substance absorption. In this study we compared the results obtained for wavelengths 420 and 380 nm. There are very good linear relationships between the absorptions measured at 420 and 380 nm. The relevant parameters are listed in Table 2 (only GF/F filter results are shown). The correlation coefficient obtained using Eq. 4 and slope parameters calculated from measured spectra was also very high. The respective standard error was 0.08 m–1. When a constant slope factor is applied (0.00154 nm–1 as average over the whole dataset), correlation will still remain high and the relative error of the estimate is the same as when the empirical linear relationship is used (0.13 m–1). This suggests that a slope value as close as possible to the value for the water body under investigation should be used in bio-optical models. However, variation in measurements still causes an error around 10% in the estimated values of ays (420). The suitability of the exponential function for modelling the yellow substance absorption coefficient in L. Peipsi was estimated by calculating the relative differences between the measured and the modelled (Eq. 4) spectra using the average slope value 0.0162 (±0.0007) nm–1, and also the slope values calculated from every single measured spectrum. In the blue region of the spectrum both results are very close to the measured values – average S results in a larger standard deviation among results, but the error does not exceed 1% (Fig. 4). In other spectral regions differences are larger (around 20%) and the use of average S increases the standard deviation among samples up to 42%. In general, the exponential

Table 2. Parameters for linear regression between measured (Y ) and calculated (X ) ays (420) values (N 102)

X Relationship R 2 SE ays (380) measured Y 0.499( 0.002)X 0.087( 0.023) 0.998 0.13 ays (420) ays (380) exp ( S 40) Y 0.880( 0.003) X 0.999 0.08 ays (420) ays (380) exp ( 0.0152 40) Y 0.918( 0.005)X 0.087( 0.023) 0.997 0.13

203 by measured S(380-550) by average S(380-550) 40

0 Relative difference, % -10 380-480 480-580 580-680 680-700 Spectral region

Fig. 4. Comparison of values of relative differences [ays (meas) ays (calc)] ays (meas) using the average slope S(380–550) and estimated from each sample of L. Peipsi. Boxes show the average value and vertical lines the standard deviation of samples.

model gives slightly higher values of absorption in the region 380–480 nm, but in other spectral regions the values are underestimated by up to 40%. When 700 nm, errors may increase many times, as absolute values become very low in all L. Peipsi samples. Analysing 462 filtered water samples from different lakes, Sipelgas et al. (2003) showed that the smallest difference between modelled and measured spectra is achieved when the slope parameter is calculated over a narrow wavelength interval (380–500 instead of 350–700 nm) despite the scattering correction method. For models over the whole PAR 350–700 region it causes underestimation of absorption in the red region of the spectrum by around ±40%, which is very similar to our results for L. Peipsi spectra. As total absorption in the red region is high because of absorption of light by water itself, the produced errors may be considered insignificant except for waters with high concentrations of CDOM –1 where ays (380) 6.3 m . Then ays (600) becomes higher than absorption by water at this wavelength, 0.2224 m–1 (Pope & Fry, 1997). Therefore these errors will be significant in bio-optical modelling of L. Peipsi coastal waters.

Spectral diffuse attenuation coefficient

All optically active components affect irradiance attenuation in water in a different manner, therefore the absolute values and shape of Kd spectra may differ significantly (Fig. 5). In L. Peipsi the spectral diffuse attenuation coefficient

204 (a) (b) 14 14 BIC2104 data L. Tuusulanjärvi 12 12 GER 1500 data L. Peipsi 10 10 L. Vänern -1 -1 8 L. Paukjärv 8 m , , m d d L. Vättern 6 6 K K

4 4

2 2

0 0 400 500 600 700 400 500 600 700 Wavelength, nm Wavelength, nm

Fig. 5. Examples of diffuse attenuation coefficient spectra: (a) average measured in different lakes; (b) measured in different sampling stations in L. Peipsi by spectral and multispectral spectrometers.

Kd ( ) was measured only in the deepest part of the lake where OAS have the –3 –3 lowest concentrations (Cchl 2.9 6.9 mg m , CSM 0.2 1.7 g m , and –1 ays (380) 4.7 4.9 m ). Therefore they represent the lowest Kd values in this lake, being still notably higher than the average spectra from lakes Vättern,

Vänern, and Paukjärv. Lake Paukjärv represents the lowest Kd values measured during the expeditions in Estonia and Finland in 1994–2000 (Fig. 5a). However,

Kd values measured with a three-channel spectrometer BIC-2104 show that attenuation in shallow areas of a lake is sometimes comparable with most turbid samples measured earlier (e.g. L. Tuusulanjärvi) (Fig. 5b). Detailed analysis of diffuse attenuation spectra in different types of lakes is presented by Reinart & Herlevi (1999). In clear lakes (like L. Vättern) a small amount of CDOM causes an increase in the attenuation in the violet and blue parts of the spectrum, but the main attenuating factor in the red part of the spectrum is water itself. For these lakes it is characteristic that the highest attenuation is in the red part of the spectrum and that the irradiance that penetrates to the deepest layers is in the range 540–550 nm. The values of Kd ( ) measured in the open area of L. Peipsi are comparable to the surface waters of L. Verevi, while in shallow areas of L. Peipsi they are similar to those in lakes Võrtsjärv and Ülemiste. In these shallow lakes rich in phytoplankton, the attenuation in the violet and blue parts of the spectrum typically exceeds the attenuation in the red part. An additional attenuation peak around 680 nm, caused by absorption in phytoplankton pigments, is remarkable (even a weak maximum ~620 nm could be seen). In these lakes irradiance is attenuated strongly over the whole spectrum, the most penetrating radiation being in the spectral range 590–650 nm.

By our earlier measurements (Reinart & Herlevi, 1999) the values of Kd,PAR range from 0.3 m–1 in L. Paukjärv to 7.2 m–1 in Nohipalu Mustjärv. In L. Peipsi

205 –1 Kd,PAR varied from 0.89 to 3.76 m , which means that the euphotic depth (the depth to which 1% of the subsurface irradiance reaches) varies there from 1.2 to 5.2 m. As the mean depth of L. Peipsi is only 7 m, large areas of bottom should be well lighted. This makes the light conditions in L. Peipsi very different from deep Swedish lakes, Vänern and Vättern, where PAR penetrates on average to depths of 5.7 m and 11.9 m, the mean depth of lakes being 27 and 40 m, respectively. To understand the effect of the different configuration of the instrumentation to measurement results, Kd,PAR data were plotted against Kd values at three wavelengths extracted from spectral Kd , according to BIC-2104 bands 412, 555, and 665 nm (Fig. 6). There is a significant correlation between Kd at these three 2 bands (R 0.74 0.95). The Kd,PAR values estimated from earlier spectral measurements in small Estonian and Finnish lakes gave the highest correlations 2 with Kd (555) (R 0.67). This band is centred in the middle of the PAR region, and obviously is less affected by different OAS in water. Lower correlation coefficients were observed for bands 412 and 665 nm (R 2 0.58 and 0.57, respectively). The BIC-2104 instrument gave quite similar results (see L. Peipsi measurements in Fig. 6) and despite the noteworthy scattering of points they are all in 95% limits estimated by previous measurements.

band 421 vs PAR band 665 vs PAR lower and upper 95% limits Peipsi, band 412 nm Peipsi, band 665 nm

14 K d(412) = 3.859*K d(PAR) - 0.102 R 2 = 0.59 12 N -1 = 84 10 (665), m

d 8 K K d(665) = 0.788*K d(PAR) + 0.243 6 2 R = 0.57 (412),

d N = 84 4 K

0 012345

-1 K d(PAR), m

Fig. 6. Linear relationships of Kd,PAR with K d for bands centred at 412 and 665 nm (thin dotted lines show lower and upper limits of 95% confidence) and results from measurements in L. Peipsi (triangles).

206 Effect of coloured dissolved organic matter on the diffuse attenuation of downwelling irradiance

A statistical analysis was performed using the spectral values of Kd and concentrations of OAS from Estonian and Finnish small lakes. The database was compiled during the years 1994–2000. The values of Cchl varied from 1.28 –3 –3 –1 to 87.33 mg m , CMS was 0.5–20.80 g m , and ays (380) 0.68–14.24 m . The spectral distributions of the correlation coefficients between Kd and other water parameters are presented in Fig. 7. According to these results the main factor influencing the light attenuation in the violet and blue parts of the spectrum is 2 yellow substance (at 400 nm R 0.897). In the orange and red parts of the Kd spectra the correlation coefficient was highest for suspended particles containing chlorophyll a. Secchi depth is correlated negatively with spectral attenuation – its higher values correspond to lower attenuation, R 2 being 0.64–0.76. These not very high values can be explained by simultaneous influence of many factors, including weather and sensitivity of human eyes.

To derive an empirical relationship between CDOM and Kd relying on BIC- 2104 data the earlier dataset was reviewed and from Kd spectra values according to three spectral bands of the BIC instrument were extracted. Best relationships with ays (380) were shown by Kd (412) and the ratio Kd (555) Kd (655) (Table 3 –1 and Fig. 8). The respective errors in the values of ays (380) were 1.28 and 1.45 m . The non-zero intercept in the relationship using bands 555 and 665 shows that a significant part of the attenuation at these bands is caused by other OAS than CDOM.

1.0 0 a ys -0.1 C 0.8 chl -0.2 C SM -0.3 R

0.6 -0.4 (Secchi) -0.5 (OAS)

R 0.4 -0.6 Secchi -0.7 0.2 -0.8 -0.9 0.0 -1 400 500 600 700 Wavelength, nm

Fig. 7. Spectral distribution of correlation coefficient R (p < 0.05) between Kd ( ) and concentrations of optically active substances (left axis) and K d,PAR and Secchi depth (right axis). The wavelength interval for each curve is 5 nm.

100

207 2 Table 3. Determination coefficients (R ) for regression between the absorption coefficient of yellow substance ays (380 ) and Kd at three BIC bands

Band 412 Band 555 Band 665 ays (380) vs. Kd 0.85 0.77 0.42 ln[ays (380)] vs. ln[Kd ] 0.91 0.74 0.44 Ratio 412/665 Ratio 412/555 Ratio 555/665 ays (380) vs. ratio 0.61 Insignificant 0.89 ln[ays (380)] vs. ln[ratio] 0.75 Insignificant 0.83

Yellow substance is the most powerful light absorbing component in many Estonian lakes (Arst et al., 1997; Reinart et al., 2003) and also in lakes investigated by us in Finland and Sweden (Reinart et al., 2004). In Lake Peipsi 34–80% of the attenuation of daylight in the blue region is caused by yellow substance. Its effect is comparable with absorption by pure water in the middle of the visible region (band centred at 555 nm), where the absorption by CDOM is 13–39%. The effect of CDOM is lowest in the 665 nm band (2–10%). This large variability in every spectral region is caused by the large variation of OAS in L. Peipsi and shows that CDOM cannot be considered similarly over the whole aquatorium. In L. Vättern the influence of CDOM in the blue region of the spectrum is higher (~90%) than in L. Peipsi, because the concentration of other OAS is very small. In L. Vänern, where also the concentration of OAS is generally smaller than in L. Peipsi (Table 1), the effect caused by CDOM varies around 80% in the blue region and 36% in the green, being slightly higher than in L. Peipsi. On the basis of this limited number of samples the variation of the CDOM effect in L. Vänern and L. Vättern was not so large as in L. Peipsi.

3 18 y = 1.145(±0.038)x y = 11.30(±0.43)x - 4.44(±0.43) R 2 = 0.91 2 15 R = 0.89 N = 95 N = 95

2 -1 12 (380)]

ys 9 a (380), m (380),

ln[ 1 ys

a 6

3 0 0 0123 0 0.5 1 1.5 2 ln[Kd(412)] K d(555)/K d(665)

Fig. 8. Relationships between ays (380 ) and Kd values at BIC-2104 bands derived from spectral measurements in Estonian and Finnish lakes (filled squares) and data obtained for L. Peipsi in 2003 (open circles).

101

208 The variations caused by spatial and seasonal variability of DOM and other substances in waters induce the different water types over the lake. In June the open area of the lake is mainly of “Moderate” type by the classification of small lakes (Reinart et al., 2003), while measurements in August showed very turbid water in the southern part of L. Peipsi s.s. and in L. Lämmijärv, and coastal waters probably belong even to “Brown” type. The concentration and properties of allochthonous CDOM brought into the lake by rivers and from coastal soils depend very much on the annual water balance. The water level and its seasonal change in L. Peipsi are highly variable (difference more than 1.5 m), the inflow to the lake is usually 5–6 times higher in April and May than in June, and very low around midsummer (Jaani & Kullus, 2001). According to regular monitoring data (T. Nõges, pers. comm., 2004) in –1 2001–02 ays (380) showed an almost twofold variation (5.0–7.9 m in the open area of the lake and 6.9–12.6 m–1 close to the Suur-Emajõgi River inflow). How- ever, for the open area the highest values of CDOM are measured in June, but in the river mouth in November (in 2001) or August (in 2002). Even these values may be slightly higher than were obtained in the present study (because no correction for residual scattering was applied), they still show large spatial and temporal variability due to the complexity of problems related to the effect of DOM to the light climate in L. Peipsi. This points out the need for more extensive and detailed investigations.

CONCLUSIONS

Results obtained by in situ measurements and analysis of water samples in 16 sampling stations in Lake Peipsi in June, August, and November 2003 describe the variability of light attenuation in water as well as the influence of dissolved organic matter on the underwater light field in a large, seasonally changeable lake. The beam absorption coefficient of filtered water at 380 nm varied from 4.1 to 16.7 m–1. In general there was no direct connection between the distance from the mouth of the Suur-Emajõgi River and the concentration of coloured dissolved organic matter, but its highest values were observed in the area of the river inflow. All water samples taken from L. Lämmijärv and closer than 1.9 km to the coast of L. Peipsi gave the values of ays (380) higher than its average over the Estonian and Finnish lakes studied earlier. The amount of CDOM in L. Peipsi is also much higher than in two other large European lakes, Vänern and Vättern in Sweden. In L. Peipsi the slope factor S(380–550) varied from 0.015 to 0.018 nm–1 when GF/F filters were used, and in the range 0.016–0.019 nm–1 in the case of filters with pore size 0.2 m. These values are within the range obtained by us and other authors for small lakes, but slightly higher than we estimated in large Swedish lakes. Comparison of the results obtained using filters with pore size 0.6–0.7 (GF/F) and 0.2 m (102 water samples from various types of lakes in Estonia, Finland,

102

209 and Sweden were analysed) showed that two datasets of ays (380) had very high determination coefficients (R 2 0.993). Using smaller pore size filters we got on –1 average 4% lower absorption values. When ays (380) was more than 8 m , the effect of smaller pore size was less than 4%. The respective values of the slope factor S(380–550) had lower values of R 2 (0.49), and an increase of about 10% was observed in the slope. This needs to be taken into account when different datasets are merged. In L. Peipsi yellow substance causes 34–80% of the attenuation of daylight in the blue region of the spectrum. CDOM absorption constitutes 13–39% of the attenuation in the middle of the visible region, and its effect there is comparable with the absorption by pure water. The measurement results of the diffuse attenuation coefficient carried out by new instrumentation (BIC-2104) are highly consistent with our earlier data (spectral measurements using a spectrophotometer LI-1800UW) in Estonian and

Finnish lakes. The values of ays (380) can be predicted by the logarithmic relationship with Kd at 412 nm or the ratio of Kd (555) to Kd (665) (the standard error is 1.3 and 1.5 m–1, respectively). For a more profound description of spatial and seasonal variation in bio-optical parameters (including CDOM) over L. Peipsi complex optical and hydro-meteorological investigations are needed.

ACKNOWLEDGEMENTS

This study was supported by EU Marie Curie Fellowship, contract No HPMF- CT-2001-01284, Swedish National Space Board grant 132/03, and Estonian Science Foundation grant 5738. We acknowledge the Estonian Maritime Administration for providing a boat for the field survey and financial co-support from the European Commissions’s Environment and Sustainable Development Programme under contract EVK1-CT-2002-00121 (CLIME). Additional data obtained in the frames of the State Monitoring Programme of the Estonian Ministry of the Environment were used.

REFERENCES

Aas, E. 2000. Spectral slope of yellow substance: problems caused by small particles. In Proc. XV Conference of Ocean Optics, Monaco, p. 8 (CD-ROM). Arst, H. 2003. Optical Properties and Remote Sensing of Multicomponental Water Bodies. Springer, Praxis Publishing, Chichester, UK. Arst, H., Mäekivi, S., Lukk, T. & Herlevi, A. 1997. Calculating irradiance penetration into water bodies from the measured beam attenuation coefficient. Limnol. Oceanogr., 42(20), 379–385. Arst, H., Erm, A., Kallaste, K., Mäekivi, S., Reinart, A., Nõges, P. & Nõges, T. 1999. Investigation of Estonian and Finnish lakes by optical measurements in 1992–1997. Proc. Estonian Acad. Sci. Biol. Ecol., 48, 5–24. Bricaud, A., Morel, A. & Prieur, L. 1981. Absorption by dissolved organic matter in the sea (yellow substance) in the UV and visible domain. Limnol. Oceanogr., 26, 43–53.

103

210 Cole, J. J., Caraco, N. F., Kling, G. W. & Kratz, T. K. 1994. Carbon dioxide supersaturation in the surface waters of lakes. Science, 265, 1568–1570. Davies-Colley, R. J. & Vant, W. N. 1987. Absorption of light by yellow substance in freshwater lakes. Limnol. Oceanogr., 32, 416–425. Dera, J. 1992. Marine Physics. PWN, Warszawa and Elsevier, Amsterdam. Duarte, C. M. 1991. Seagrass depth limits. Aquat. Bot., 40, 363–377. Fedorov, K. N. & Ginsburg, A. I. 1992. The Near-Surface Layer of the Ocean. VSP, Utrecht, The Netherlands. Freeman, C., Evans, C. D., Monteith, D. T., Reynolds, B. & Fenner, N. 2001. Export of organic carbon from peat soils. Nature, 412, 785. Gallie, E. A. 1994. Optical calibration parameters for water-color models from Swan lake, Northern Ontario. Can. J. Remote Sensing, 20,156–161. Gege, P. 2000. Gaussian model for yellow substance absorption spectra. In Proc. XV Conference of Ocean Optics, Monaco, p. 9 (CD-ROM). Herlevi, A., Virta, H., Arst, H. & Erm, A. 1999. Results of light absorption/attenuation measurements in Finnish and Estonian lakes in summer 1997. Proc. Estonian Acad. Sci. Biol. Ecol., 48, 46–62. Hoge, F. E., Vodacek, A. & Blough, N. V. 1993. Inherent optical properties of the ocean: retrieval of the absorption coefficient of chromophoric dissolved organic matter from fluorescence measurements. Limnol. Oceanogr., 38, 1394–1402. Jaani, A. & Kullus, L.-P. 2001. Hydrological regime and water balance. In Lake Peipsi. Meteorology, Hydrology, Hydrochemistry (Nõges, T., ed.), pp. 65–72. Sulemees, Tartu. Jansson, M. 1998. Nutrient limitation and bacteria–phytoplankton interactions in humic lakes. In Aquatic Humic Substances: Ecology and Biogeochemistry (Hessen, D. O. & Tranvik, L. J., eds.), pp. 177–195. Ecol. Stud., 133. Springer Verlag. Kallio, K. 1999. Absorption properties of dissolved organic matter in Finnish lakes. Proc. Estonian Acad. Sci. Biol. Ecol., 48, 75–83. Krause-Jensen, D. & Sand-Jensen, K. 1998. Light attenuation and photosynthesis of aquatic plant communities. Limnol. Oceanogr., 43, 396–407. Kirk, J. T. O. 1994. Light and Photosynthesis in Aquatic Ecosystems. Cambridge University Press. Lindell, T., Pierson, D., Premazzi, G. & Ziliolli, E. 1999. Manual for Monitoring European Lakes Using Remote Sensing Techniques. Office for Official Publications of the European Communities, Luxembourg. Mäekivi, S. & Arst, H. 1996. Estimation of the concentration of yellow substance in natural waters by beam attenuation coefficient spectra. Proc. Estonian Acad. Sci. Ecol., 6, 108–123. Miller, L. W., Johannessen, S. & Kuhn, P. 2000. The optical properties of dissolved organic matter (DOM) in coastal and open waters. In Proc. XI Conference of Ocean Optics, Monaco, p. 6 (CD-ROM). Mobley, C. D. 1994. Light in Water-Radiative Transfer in Natural Waters. Academic Press, San Diego. Nõges, T. (ed.) 2001. Lake Peipsi. Meteorology, Hydrology, Hydrochemistry. Sulemees, Tartu. Östlund, C., Flink, P., Strömbeck, N., Pierson, D. & Lindell, T. 2001. Mapping of the water quality of Lake Erken, Sweden from Imaging Spectrometer and Landsat Thematic Mapper. Sci. Total Environ., 268, 139–164. Pierson, D. C. & Strömbeck, N. 2001. Estimation of radiance reflectance and the concentrations of optically active substances in Lake Mälaren, Sweden, based on direct and inverse solutions of a simple model. Sci. Total Environ., 268, 171–188. Pope, M. & Fry, E. S. 1997. Absorption spectrum (380–700 nm) of pure water. II. Integrating cavity measurements. Appl. Optics, 36, 8710–8723. Reinart, A. & Herlevi, A. 1999. Diffuse attenuation coefficient in some Estonian and Finnish lakes. Proc. Estonian Acad. Sci. Biol. Ecol., 48, 267–283. Reinart, A., Herlevi, A., Arst, H. & Sipelgas, L. 2003. Preliminary optical classification of lakes and coastal waters in Estonia and South-Finland. J. Sea Res., 49, 357–366.

104

211 Reinart, A., Paavel, B. & Pierson, D. 2004 Comparison of optical properties of three large European lakes. Boreal Envir. Res. (submitted). Schanz, F. 1985. Vertical light attenuation and phytoplankton development in Lake Zürich. Limnol. Oceanogr., 30, 299–310. Sipelgas, L., Kallio, K., Arst, H., Erm, A., Oja, P. & Soomere, T. 2003. Optical properties of dissolved organic matter in Finnish and Estonian lakes. Nord. Hydrol., 34, 361–386. Sobek, S., Algesten, G., Bergström, A.-K., Jansson, M. & Tranvik, L. 2003. The catchment and climate regulation of pCO2 in boreal lakes. Global Change Biol., 9, 630–641. Strömbeck, N. 2001. Water quality and optical properties of Swedish lakes and coastal waters in relation to remote sensing. Comprehensive Summaries of Uppsala Dissertations from Faculty of Science and Technology, 633. Uppsala. Strömbeck, N. & Pierson, D. 2001. The effects of variability in the inherent optical properties on the estimations of the chlorophyll-a by remote sensing in Swedish freshwaters. Sci. Total Environ., 268, 123–137. Virta, J. & Blanco-Sequeiros, A. 1995. Correction of measured underwater spectral irradiance to the variation of some external effects with examples. Rep. Ser. in Geophys. University of Helsinki (Pukkinen, K., ed.), 32, 93–100.

Vees lahustunud orgaanilise aine mõju valguse nõrgenemisele Peipsi järves

Anu Reinart, Birgot Paavel ja Lea Tuvikene

Käesolevas töös on uuritud vees lahustunud värvunud orgaanilise aine (CDOM) mõju Peipsi järve veealusele valgusväljale veesiseste kiirgusmõõtmiste ja vee- proovide analüüside kaudu. Tulemusi on võrreldud suurte järvede Vänerni ja Vätterniga (Rootsi) ja 41 väikejärvega Eestis, Soomes ja Rootsis. CDOM-i hulk –1 kaldast kuni kauguseni 1,9 km on suurem (ays (380) 7,1–16,7 m ) kui avajärve –1 mõõtepunktides (ays (380) 4,1–5,6 m ). Otsest seost CDOM-i kontsentratsiooni ja kauguse vahel Suure-Emajõe suudmest ei leitud. Peipsi järvele on iseloomulik suur ajaline ja ruumiline CDOM-i hulga varieeruvus, kuid see jääb siiski sama- desse piiridesse, kui varem uuritud Eesti ja Soome väikejärvedes. Peipsi vesi sisaldab keskmiselt 2 korda rohkem lahustunud orgaanilist ainet kui Vänern ja 7 korda rohkem kui Vättern. Vee optiline tüüp Peipsi järves varieerub sõltuvalt piirkonnast. CDOM-i hulka vees on iseloomustatud filtreeritud vee neeldumis- koefitsiendi järgi fikseeritud lainepikkusel ja neeldumise eksponentsiaalset vähene- mist lainepikkusega kirjeldatava nn tõusu parameetriga. On uuritud sageli kasutata- vate erineva poorisuurusega filtrite mõju neile tulemustele (0,6–0,7 m ja 0,2 m). Osutub, et kahel viisil mõõdetuna korreleerusid CDOM-i hulga väärtused oma- vahel väga hästi (R2 = 0,99), kuid tõusu parameetrid vähem (R2 = 0,49). Varem kogutud spektraalse vertikaalse nõrgenemiskoefitsiendi andmebaasi ja uue, 3-kana- lilise spektromeetri mõõtmiste võrdlemine andis kooskõlalised tulemused. Leitud on ka esialgsed empiirilised seosed CDOM-i ja kiirguse nõrgenemiskoefitsiendi vahel Peipsi järve jaoks. Täpsemaid seoseid ning CDOM-i ruumilist ja ajalist varieeruvust Peipsi järve akvatooriumis tuleks aga edaspidi uurida kooskõlas teiste hüdrometeoroloogiliste faktoritega.

105

212 CURRICULUM VITAE

I General

Name: Lea Tuvikene

Date of birth: November 24, 1959

Citizenship: Estonian

Address: Centre for Limnology, Institute of Agricultural and Environmental Sciences, Estonian University of Life Sciences, Limnoloogia tee 1, Rannu 61117, Tartu County

Phone: +372 731 3574 e-mail: [email protected]

Education: 1975–1978 Puhja Secondary School 1978–1983 Tartu State University, biology-hydrobiology and ichthyology 1995–1995 University of Tartu, MSc in hydrobiology 1995–1999 University of Tartu, doctoral student 2017–2018 Estonian University of Life Sciences, doctoral studies (extern)

Professional employment: 1983–1995 Institute of Zoology and Botany (IZB), senior laboratory assistant 1999–present Estonian University of Life Sciences, Institute of Agricultural and Environmental Sciences, Centre for Limnology, researcher (1.00)

II Scientific activity

Research interests: Biosciences and Environment; 1.4. Ecology, Biosystematics and -physiology; CERCS SPECIALTY: B260 Hydrobiology, marine biology, aquatic ecology, limnology; 1.8. Research relating to the State of the Environment and

213 to Environmental Protection SPECIALITY: Hydrochemistry, hydrobiology, lake monitoring and restoration

Grants and projects:

2016–2017 Hydrobiological monitoring and studies of Lake Võrtsjärv. Estonian Environment Agency. Project leader. 1.01.2014−31.12.2019 IUT21-2. Lake food webs and C metabolism across gradients of catchment alkalinity and climate (Tiina Nõges). Researcher. 1.06.2012−29.02.2016 Increasing the capability of environmental monitoring on Estonian lakes and rivers. Estonian-Swiss Cooperation Programme (Külli Kangur, Lea Tuvikene). Project leader. 1.02.2012−31.03.2015 INTERREG IVC project 1237R4. Regional administration of lake restoration initiatives. Project leader of Partner 5, Estonian University of Life Sciences. 1.02.2012−31.03.2015 Regional administration of lake restoration initiatives. Environmental Investment Centre, project No 3925. Project leader. 1.02.2010−31.01.2014 EC 244121. Adaptive Strategies to Mitigate the Impacts of Climate Change on European Freshwater Ecosystems. Researcher. 1.01.2008–31.12.2013 SF0170011s08. Will climate change alter the relative importance of catchment and in-lake processes in the carbon balance of shallow lakes? (Tiina Nõges). Researcher. 2004–2015 Hydrobiological monitoring and studies of Lake Võrtsjärv. Estonian Ministry of the Environment. Project leader. 1.01.2004−31.12.2007 ETF5738. Impact of climate change on ecosystems of large shallow lakes, mediated by the altered inflow of substances (Tiina Nõges). Researcher. 2.07.2004−31.12.2007 INTERREG IIIC project 3N00531. Lakepromo – Tools for water management and

214 restoration process. Project leader of Partner 8, Estonian University of Life Sciences. 1.01.2003−31.12.2007 SF0362480s03. Impact of climatic change on shallow lake ecosystems (Tiina Nõges). Researcher. 1.01.2003−31.12.2005 EVK1-CT-2002-00121. Climate and Lakes Impact in Europe (Tiina Nõges). Researcher.

Supervisions:

2008 Tuuli Soomets, Master’s Degree, (sup) Helgi Arst; Lea Tuvikene, Measurements and modelling of underwater light field in Estonian lakes (in Estonian), Estonian University of Life Sciences, Institute of Agricultural and Environmental Sciences, Centre for Limnology. 2015 Kristjan Soo, Master’s Degree, 2015, (sup) Arvo Tuvikene; Lea Tuvikene, Lake restoration in Estonia: history and perspectives (in Estonian), Estonian University of Life Sciences, Institute of Agricultural and Environmental Sciences. 2016 Kerli Pettai, Master’s Degree, 2016, (sup) Tuuli Soomets; Lea Tuvikene, Efficiency of remote sensing to assess chlorophyll a concentrations in different Estonian lakes (in Estonian), Estonian University of Life Sciences, Institute of Agricultural and Environmental Sciences, Centre for Limnology.

Main publications:

Cremona, F., Tuvikene, L., Haberman, J., Nõges, P. & Nõges, T. 2018. Factors controlling the three-decade long rise in cyanobacteria biomass in a eutrophic shallow lake. Science of the Total Environment, 621, 352−359.10.1016/j.scitotenv.2017.11.250.

Toming, K., Kutser, T., Tuvikene, L., Viik, M. & Nõges, T. 2016. Dissolved organic carbon and its potential predictors in eutrophic lakes. Water Research, 102: 32−40, 10.1016/j.watres.2016.06.012.

Vilbaste, S., Järvalt, A., Kalpus, K., Nõges, T., Pall, P., Piirsoo, K., Tuvikene, L. & Nõges, P. 2016. Ecosystem services of Lake Võrtsjärv under multiple stress: a case study. Hydrobiologia, 780: 145−159, 10.1007/s10750-016-2871-y. 215 Mahdy, A., Hilt, S., Filiz, N., Beklioğlu, M., Hejzlar, J., Özkundakci, D., Papastergiadou, E., Scharfenberger, U., Šorf, M., Stefanidis, K., Tuvikene, L., Zingel, P., Søndergaard, M., Jeppesen, E. & Adrian, R. 2015. Effects of water temperature on summer periphyton biomass in shallow lakes: a pan-European mesocosm experiment. Aquatic Sciences, 77(3): 499−510, 10.1007/s00027-015-0394-7.

Rõõm, E.-I., Nõges, P., Feldmann, T., Tuvikene, L., Kisand, A., Teearu, H. & Nõges, T. 2014. Years are not brothers: two-year comparison of greenhouse gas fluxes in large shallow Lake Võrtsjärv, Estonia. Journal of Hydrology, 519: 1594−1606.10.1016/j.jhydrol.2014.09.011.

Agasild, H., Zingel, P., Tuvikene, L., Tuvikene, A., Timm, H., Feldmann, T., Salujõe, J., Toming, K., Jones, R. I. & Nõges, T. 2014. Biogenic methane contributes to the food web of a large, shallow lake. Freshwater Biology, 59(2): 272−285.10.1111/fwb.12263.

Landkildehus, F., Søndergaard, M., Beklioglu, M., Adrian, R., Angeler, D. G., Hejzlar, J., Papastergiadou, E., Zingel, P., İdil Çakiroğlu, A., Scharfenberger, U., Drakare, S., Nõges, T., Šorf, M., Stefanidis, K., Tavşanoğlu, Ü.,N., Trigal, C., Mahdy, A., Papadaki, C., Tuvikene, L., Larsen, S.E., Kernan, M. & Jeppesen, E. 2014. Climate change effects on shallow lakes: Design and preliminary results of a cross-European climate gradient mesocosm experiment. Estonian Journal of Ecology, 63 (2), 71−89, 10.3176/eco.2014.2.02.

Toming, K., Tuvikene, L., Vilbaste, S., Agasild, H., Kisand, A., Viik, M., Martma, T., Jones, R. & Nõges, T. 2013. Contributions of autochthonous and allochthonous sources to dissolved organic matter in a large, shallow, eutrophic lake with a highly calcareous catchment. Limnology and Oceanography, 58(4): 1259−1270.10.4319/lo.2013.58.4.1259.

Nõges, P. & Tuvikene, L. 2012. Spatial and annual variability of environmental and phytoplankton indicators in Vrtsjärv: implications for water quality monitoring. Estonian Journal of Ecology, 61(4): 227−246, 10.3176/eco.2012.4.01.

Tuvikene, L., Nõges, T. & Nõges, P. 2011. Why do phytoplankton species composition and “traditional” water quality parameters indicate

216 different ecological status of a large shallow lake? Hydrobiologia, 660 (1): 3−15, 10.1007/s10750-010-0414-5.

Kauer, T., Arst, H. & Tuvikene, L. 2010. Underwater light field and spectral distribution of attenuation depth in inland and coastal waters. Oceanologia, 52(2): 155−170.

Nõges, T., Tuvikene, L. & Nõges, P. 2010. Contemporary trends of temperature, nutrient loading and water quality in large lakes Peipsi and Võrtsjärv, Estonia. Journal of Aquatic Ecosystem Health and Management, 13(2): 143−153, 10.1080/14634981003788987.

Kauer, T., Arst, H., Nõges, T. & Tuvikene, L. 2009. Estimation of the phytoplankton productivity in three Estonian lakes. Estonian Journal of Ecology, 58(4): 297−312.

Zingel, P., Nõges, P., Tuvikene, L., Feldmann, T., Järvalt, A., Tõnno, I., Agasild, H., Tammert, H., Luup, H., Salujõe, J. & Nõges, T. 2006. Ecological processes in macrophyte- and phytoplankton-dominated shallow lakes. Proceedings of the Estonian Academy of Sciences. Biology, Ecology, 55 (4): 280−307.

Reinart, A., Paavel, B. & Tuvikene, L. 2004. Effects of coloured dissolved organic matter on the attenuation of photosynthetically active radiation in Lake Peipsi. Proceedings of the Estonian Academy of Sciences. Biology, Ecology, 53(2): 88−105.

Tuvikene, L., Kisand, A., Tõnno, I. & Nõges, P. 2004. Chemistry of lake water and bottom sediments. In: Haberman, J., Pihu, E. & Raukas, A. (eds), Lake Võrtsjärv (89−102). Estonian Encyclopaedia Publishers.

Moss, B., Stephen, D., Alvarez, C., Becares, E., Van de Bund, W., Collings, S. E., Van Donk, E., De Eyto, E., Feldmann, T., Fernandez-Alaez, C., Fernandez-Alaez, M., Franken, R.J.M., Garcia-Criado, F., Gross, E.M., Gyllstrom, M., Hansson, L.A., Irvine, K., Järvalt, A., Jensen, J.P., Jeppesen, E., Kairesalo, T., Kornijów, R., Krause, T., Künnap, H., Laas, A., Lill, E., Lorens, B., Luup, H., Miracle, M.R., Nõges, P., Nõges, T., Nykänen, M., Ott, I., Peczula, W., Peeters, E.T.H.M., Phillips, G., Romo, S., Russell, V., Salujõe, J., Scheffer, M., Siewertsen, K., Smal, H., Tesch, C., Timm, H., Tuvikene, L., Tõnno, I., Virro, T., Vicente, E. & Wilson, D. 2003. The

217 determination of ecological status in shallow lakes – a tested system (ECOFRAME) for implementation of the European Water Framework Directive. Aquatic Conservation-Marine and Freshwater Ecosystems, 13(6): 507−549, 10.1002/aqc.592.

Nõges, P., Nõges, T., Tuvikene, L., Smal, H., Ligeza, S., Kornijow, R., Peczula, W., Becares, E., Garcia-Criado, F., Alvarez-Carrera, C., Alvarez- Carrera, C., Ferriol, C., Miracle, M., Vicente, E., Romo, S., Van Donk, E., van de Bund, W., Jensen, J.P., Gross, E.M., Hansson, L.-A., Gyllström, M., Nykänen, M., de Eyto, E., Irvine, K., Stephen, D., Collings, S. & Moss, B. 2003. Factors controlling hydrochemical and trophic state variables in 86 shallow lakes in Europe. Hydrobiologia, 506: 51−58.

Nõges, P., Tuvikene, L., Feldmann, T., Tõnno, I., Künnap, H., Luup, H., Salujõe, J. & Nõges, T. 2003. The role of charophytes in increasing water transparency: a case study of two shallow lakes in Estonia. Hydrobiologia, 506: 567−573.

Kisand, V., Tuvikene, L. & Nõges, T. 2001. Role of phosphorus and nitrogen for bacteria and phytoplankton development in a large shallow lake. Hydrobiologia, 457: 187−197.

Nõges, P., Tuvikene, L., Nõges, T. & Kisand, A. 1999. Primary production, sedimentation and resuspension in large shallow Lake Võrtsjärv. Aquatic Sciences, 61(2): 168−182.

Nõges, P., Järvet, A., Tuvikene, L. & Nõges, T. 1998. The budgets of nitrogen and phosphorus in shallow eutrophic Lake Võrtsjärv (Estonia). Hydrobiologia, 363: 219−227.

Nõges, T., Kisand, V., Nõges, P., Põllumäe, A., Tuvikene, L. & Zingel, P. 1998. Plankton seasonal dynamics and its controlling factors in shallow polymictic eutrophic lake Võrtsjärv, Estonia. International Review of Hydrobiology, 83(4): 279−296.

Thesis of International conferences:

218 water optics international workshop, 7th Baltic Sea Science Congress, Tallinn, Estonia, 17-21 August, 2009. Tallinna Tehnikaülikool.

Vassiljev, A., Venohr, M. & Tuvikene, L. 2009. Comparison of the MONERIS and MESAW models for the Peipsi lake watershed. First MONERIS User Conference: Nutrient fluxes in river basins - State of play and perspectives, Berlin, November 18th-20th, 2009. Berlin, 1−2.

Tuvikene, A., Tuvikene, L. & Viik, M. 2007. Effect of winter hypoxia on fish in small lakes with different water quality. Fish Physiology, Toxicology, and Water Quality: Ninth International Symposium in fish physiology, toxicology and water quality. Capri, Italy, April 24-28, 2006. EditorsAbbr C. J. Brauner, K. Suvajdžic, G. Nilsson, D. Randall. Ecosystems Research Division, U.S. Environmental Protection Agency, 201−208.

Tuvikene, L., Nõges, P., Nõges, T. & Thurston, R. 2002. Hypoxia/anoxia in Lake Võrtsjärv, Estonia. Proceedings of the Sixth International Symposium on Fish Physiology, Toxicology, and Water Quality, La Paz B.C.S. Mexico, January 22-26, 2001. Athens, Georgia, USA: U.S. Environmental Protection Agency, Ecosystems Research Division, 163−170.

219 ELULOOKIRJELDUS

I Üldandmed

Nimi: Lea Tuvikene

Sünniaeg: 24. november 1959

Kodakondsus: Eesti

Aadress: Limnoloogiakeskus, Põllumajandus- ja keskkonnainstituut, Eesti Maaülikool, Limnoloogia tee 1, Elva vald 61117, Tartumaa

Telefon: +372 731 3574 e-post: [email protected]

Haridus: 1975–1978 Puhja Keskkool 1978–1983 Tartu Riiklik Ülikool, bioloogia-hüdrobioloogia ja ihtüoloogia; 1995–1995 Tartu Ülikool, MSc hüdrobioloogia erialal 1995–1999 Tartu Ülikool, doktorant 2017–2018 Eesti Maaülikool, doktorantuur (ekstern)

Töökogemus: 1983–1995 Eesti TA ZBI insener, vanemlaborant 1999–praeguseni Eesti Maaülikool, Põllumajandus- ja keskkonnainstituut, limnoloogiakeskus, Teadur (1,00)

Teadustegevus Peamised uurimisvaldkonnad: CERCS ERIALA: B260 Hüdro bioloogia, mere-bioloogia, veeökoloogia, limnoloogia; PÕHISUUNAD: Hüdrokeemia, hüdrobioloogia, järvede seire ja tervendamine

220 Grandid ja projektid: 2016–2017 Keskkonnaagentuur. Võrtsjärve hüdrobioloo- giline seire ja uuringud. Vastutav täitja. 1.01.2014−31.12.2019 IUT21-2. Järvede toiduahelad ja süsiniku metabolism valgala karbonaatsuse ja kliima gradien- dis (Tiina Nõges). Teadur. 1.06.2012−29.02.2016 Eesti järvede ja jõgede keskkonnaseire suutlik- kuse tõstmine. Eesti-Šveitsi koostööprogramm (Külli Kangur, Lea Tuvikene). Projektijkuht. 1.02.2012−31.03.2015 INTERREG IVC projekt 1237R4. Regional administration of lake restoration initiatives (Järvede tervendamise piirkondlik haldamine). Partneri nr 5, Eesti Maaülikool, projektikjuht. 1.02.2012−31.03.2015 KIK Veemajandusprogrammi projekt nr 3925. Järvede tervendamise piirkondlik haldamine. Projektijuht. 1.02.2010−31.01.2014 EC 244121. Adaptive Strategies to Mitigate the Impacts of Climate Change on European Fres- hwater Ecosystems. Teadur. 1.01.2008–31.12.2013 SF0170011s08. Kas kliimamuutus muudab valgala- ja järvesiseste protsesside osakaalu madala järve süsinikubilansis? (Tiina Nõges). Teadur. 2004–2015 Keskkonnaministeerium. Võrtsjärve hüdrobio- loogiline seire ja uuringud. Vastutav täitja. 1.01.2004−31.12.2007 ETF5738. Kliimamuutuste mõju suure madala järve ökosüsteemile ainete sissekande muutuste kaudu (Tiina Nõges). Teadur. 2.07.2004−31.12.2007 INTERREG IIIC projekt 3N00531. Lakepromo – Tools for water management and restoration process (Lakepromo – Veekogude majandamise ja tervendamise vahendid). Part- neri nr 8, Eesti Maaülikool, projektikjuht. 1.01.2003−31.12.2007 SF0362480s03. Kliimamuutuste mõju madalate järvede ökosüsteemile (Tiina Nõges). Teadur. 1.01.2003−31.12.2005 EVK1-CT-2002-00121. Climate and Lakes Impact in Europe (Tiina Nõges). Teadur.

221 Juhendamised:

2008 Tuuli Soomets, magistrikraad, (juh) Helgi Arst, Lea Tuvikene. Eesti järvede veealuse valgusvälja mõõtmine ja modelleerimine. Eesti Maaülikool, Põllumajandus- ja keskkonnainstituut, limnoloogiakeskus. 2015 Kristjan Soo, magistrikraad, (juh) Arvo Tuvikene, Lea Tuvikene. Järvede tervendamine Eestis: ajalugu ja arenguvõimalused. Eesti Maaülikool, Põllumajandus- ja keskkonnainstituut. 2016 Kerli Pettai, magistrikraad, (juh) Tuuli Soomets, Lea Tuvikene. Kaugseire kasutamise võimekusest klorofüll a kontsentratsioonide hindamisel erinevates Eesti järvedes. Eesti Maaülikool, Põllumajandus- ja keskkonnainstituut.

Peamiste publikatsioonide loetelu:

Toming, K., Kutser, T., Tuvikene, L., Viik, M. & Nõges, T. 2016. Dissolved organic carbon and its potential predictors in eutrophic lakes. Water Research, 102: 32−40, 10.1016/j.watres.2016.06.012.

Mahdy, A., Hilt, S., Filiz, N., Beklioğlu, M., Hejzlar, J., Özkundakci, D., Papastergiadou, E., Scharfenberger, U., Šorf, M., Stefanidis, K., Tuvikene, L., Zingel, P., Søndergaard, M., Jeppesen, E. & Adrian, R. 2015. Effects of water temperature on summer periphyton biomass in shallow lakes: a pan-European mesocosm experiment. Aquatic Sciences, 77(3): 499−510, 10.1007/s00027-015-0394-7.

Rõõm, E.-I., Nõges, P., Feldmann, T., Tuvikene, L., Kisand, A., Teearu, H. & Nõges, T. 2014. Years are not brothers: two-year comparison of

222 greenhouse gas fluxes in large shallow Lake Võrtsjärv, Estonia. Journal of Hydrology, 519: 1594−1606.10.1016/j.jhydrol.2014.09.011.

Tuvikene, L., Nõges, T. & Nõges, P. 2011. Why do phytoplankton species composition and “traditional” water quality parameters indicate different ecological status of a large shallow lake? Hydrobiologia, 660 (1): 3−15, 10.1007/s10750-010-0414-5.

Kauer, T., Arst, H. & Tuvikene, L. 2010. Underwater light field and spectral distribution of attenuation depth in inland and coastal waters. Oceanologia, 52(2): 155−170.

Nõges, T., Tuvikene, L. & Nõges, P. 2010. Contemporary trends of temperature, nutrient loading and water quality in large lakes Peipsi

223 and Võrtsjärv, Estonia. Journal of Aquatic Ecosystem Health and Management, 13(2): 143−153, 10.1080/14634981003788987.

Kauer, T., Arst, H., Nõges, T. & Tuvikene, L. 2009. Estimation of the phytoplankton productivity in three Estonian lakes. Estonian Journal of Ecology, 58(4): 297−312.

224 M., Nykänen, M., de Eyto, E., Irvine, K., Stephen, D., Collings, S. & Moss, B. 2003. Factors controlling hydrochemical and trophic state variables in 86 shallow lakes in Europe. Hydrobiologia, 506: 51−58.

Kisand, V., Tuvikene, L. & Nõges, T. 2001. Role of phosphorus and nitrogen for bacteria and phytoplankton development in a large shallow lake. Hydrobiologia, 457: 187−197.

Nõges, P., Tuvikene, L., Nõges, T. & Kisand, A. 1999. Primary production, sedimentation and resuspension in large shallow Lake Võrtsjärv. Aquatic Sciences, 61(2): 168−182.

Nõges, P., Järvet, A., Tuvikene, L. & Nõges, T. 1998. The budgets of nitrogen and phosphorus in shallow eutrophic Lake Võrtsjärv (Estonia). Hydrobiologia, 363: 219−227.

Rahvusvaheliste konverentside teesid:

Kauer, T., Arst, H. & Tuvikene, L. 2009. Underwater light field and the spectral distribution of the attenuation depth in different water bodies. Proceedings of 7th Baltic Sea Science Congress: Remote sensing and water optics international workshop, 7th Baltic Sea Science Congress, Tallinn, Estonia, 17-21 August, 2009. Tallinna Tehnikaülikool.

225 Tuvikene, A., Tuvikene, L. & Viik, M. 2007. Effect of winter hypoxia on fish in small lakes with different water quality. Fish Physiology, Toxicology, and Water Quality: Ninth International Symposium in fish physiology, toxicology and water quality. Capri, Italy, April 24-28, 2006. EditorsAbbr C. J. Brauner, K. Suvajdžic, G. Nilsson, D. Randall. Ecosystems Research Division, U.S. Environmental Protection Agency, 201−208.

226 VIIS VIIMAST KAITSMIST

ANNA PISPONEN LACTOSE CLUSTERING AND CRYSTALLIZATION - AN EXPERIMENTAL INVESTIGATION OF LACTOSE PURE SOLUTION AND RICOTTA CHEESE WAY LAKTOOSI KLASTERDUMINE JA KRISTALLISEERUMINE: PUHTA LAKTOOSI LAHUSE JA RICOTTA JUUSTU VADAKU EKSPERIMENTAALUURING Professor Avo Karus, Dr Väino Poikalainen (Teadus ja Tegu OÜ) 10. november 2017

ANNE PÕDER THE SOCIO-ECONOMIC DETERMINANTS OF ENTREPRENEURSHIP IN ESTONIAN RURAL MUNICIPALITIES ETTEVÕTLUST MÕJUTAVAD SOTSIAAL-MAJANDUSLIKUD TEGURID EESTI VALDADES Professor Rando Värnik 27. november 2017

ENE TOOMING THE SUBLETHAL EFFECTS OF NEUROTOXIC INSECTICIDES ON THE BASIC BEHAVIOURS OF AGRICULTURALLY IMPORTANT CARABID BEETLES NEUROTOKSILISTE INSEKTITSIIDIDE SUBLETAALNE TOIME PÕLLUMAJANDUSLIKULT OLULISTE JOOKSIKLASTE PÕHIKÄITUMISTELE Vanemteadur Enno Merivee, teadur Anne Must 11. detsember 2017

KADI PALMIK EFFECTS OF NATURAL AND ANTHROPOGENIC PRESSURES AND DISTURBANCES ON THE MACROPHYTES OF LAKE PEIPSI LOODUSLIKE JA INIMTEKKELISTE SURVETEGURITE MÕJU PEIPSI JÄRVE SUURTAIMESTIKULE Vanemteadur Helle Mäemets, PhD Külli Kangur 15. detsember 2017

AIMAR NAMM EXPRESSION OF BMP AND PAX PROTEINS IN THE CENTRAL NERVOUS SYSTEM OF HUMAN AND RAT EMBRYOS AT EARLY STAGES OF DEVELOPMENT BMP JA PAX SIGNAALMOLEKULIDE AVALDUMINE INIMESE JA ROTI EMBRÜOTE KESKNÄRVISÜSTEEMI VARAJASTEL ARENGUETAPPIDEL Professor Marina Aunapuu, professor Andres Arend (Tartu Ülikool) 15. detsember 2017

ISSN 2382-7076 ISBN 978-9949-629-13-8 (publication) ISBN 978-9949-629-14-5 (pdf)