10.2478/v10060-008-0026-1

Annals of Warsaw University of Life Sciences – SGGW Land Reclamation No 38, 2007: 95–104 (Ann. Warsaw Univ. of Life Sci. – SGGW, Land Reclam. 38, 2007)

Possible effects of climate change on ecological functioning of shallow lakes, Lake Loenderveen as a case study SEBASTIAAN A. SCHEP1, GERARD N.J. TER HEERDT2, JAN H. JANSE3, MAARTEN OUBOTER2 1Witteveen+Bos, The Netherlands 2Waternet, The Netherlands 3MNP/NLB, The Netherlands

Abstract: Possible effects of climate change on INTRODUCTION ecological functioning of shallow lakes, Lake Loenderveen as a case study. The European The European Water Framework Water Framework Directive (WFD) requires all inland and coastal waters to reach “good Directive (WFD) requires all inland and ecological status” by 2015. The good ecological coastal waters to reach “good ecological status of shallow lakes can be characterised by status” ultimately in 2027. In general, clear water dominated by submerged vegetation. a shallow lake in good ecological The ecological response of shallow lakes on condition can be characterised by clear nutrients largely depends on morphological and hydrological features, such as water depth, water, small amounts of , retention time, water level fl uctuations, bottom vegetation dominated by submerged type, fetch etc. These features determine the vegetation and rich (macro) invertebrate “critical nutrient load” of a lake. When the actual and fi sh communities (Scheffer, 2001). nutrient load of a lake is higher than the critical The actual ecological state (clear nutrient load, the ecological quality of this lake will deteriorate, resulting in a turbid state dominated or turbid) of shallow lakes largely by algae. Climate change might lead to changes depends on the actual nutrient load and in both environmental factors and ecosystem the “critical nutrient load”. The critical response. This certainly will have an effect on the nutrient load is the nutrient load needed to ecological status. As an illustration the results of make a clear lake become turbid and vice a multidiscipline study of a shallow peaty lake versa. Because of hysteresis processes (Loenderveen) are presented, including hydrology, geochemistry and ecology. Ground- and surface two critical nutrient loads (CNL) can be water fl ows, nutrient dynamics and ecosystem defi ned (Scheffer, 2001). The shift from functioning have been studied culminating in an clear to turbid (eutophication, CNLeu) application of the ecological model of the lake often occurs at a much higher nutrient (PCLake). Future scenarios were implemented load than the switch back from turbid through changing precipitation, evaporation and temperature. Climate change will lead to higher to clear (oligotrophication, CNLoligo). nutrient loads and lower critical nutrient loads. As When the nutrient load is below CNLoligo a consequence lakes shift easier from clear water the lake is stably clear, when the load is to a turbid state. above CNLeu the lake is stably turbid (Fig. Key words: shallow lakes, climatic change, 1). In the intermediate range, both states PCLake model, nutrient load. may exist and switches between the two 96 S.A. Schep et al.

FIGURE 1. Critical nutrient loads of shallow lakes related to submerged vegetation states are possible (multiple stable state Chemical and bacteriological processes (Scheffer et al., 1993; Scheffer, 2001)). and the growth rates of phytoplankton The critical nutrient load depends on and submerged vegetation increase when morphological features (water depth, soil temperature increases (Mooij et al., 2005; type and lake size), hydrological features Mooij et al., 2007). (ground- and surface water fl ow, retention Each individual lake will be differently time, water level fl uctuations) and climate affected by climate change. In general, conditions (water temperature). The the critical nutrient loads, CNLoligo actual nutrient load depends on climate and CNLeu might likely decrease with conditions (precipitation, evaporation) increasing temperature and precipitation and environmental conditions (water (Mooij et al., 2005). The decrease of quality). CNLeu combined with an increase of Most shallow lakes in the Netherlands the nutrient load might cause a clear are in the turbid, eutrophic state (Gulati lake to become turbid, while a decreased and van Donk, 2002). Because of CNLoligo will hamper the return towards mostly food-web related self-stabilizing the clear state. Therefore, to determine mechanisms (see e.g. Scheffer, 1998) realistic and sustainable goals and restoration measures mainly aimed at measures for the WFD, water quality reducing the nutrient load did not succeed. managers should take climate change Therefore other approaches are needed. into account. Instead of reducing the nutrient load, the The main goal of this study is to critical nutrient load can be increased by relate effects of climate change to changing system characteristics or when the ecological functioning of Lake the nutrient load is in the intermediate Loenderveen. Further a methodology range biomanipulation can be a preferable to study the potential effect of climate measure (Gulati and Van Donk, 2002). change on lake ecosystems is developed Climate change might lead to and tested. changes in both environmental factors Ground- and surface water fl ows, and ecosystem response. Hydraulic and nutrient dynamics and ecosystem nutrient load will increase with increasing functioning have been studied culminating precipitation (Mooij et al., 2005). in an application of the ecological model Possible effects of climate change on ecological functioning... 97

PCLake. A nutrient balance is used to on the water balance of the adjacent lake predict effects of climate change on the Terra Nova, Terra Nova (water surface nutrient load. PCLake is used to predict 89.5 ha, average depth 1.43 m, residence effects of climate change on the critical time 0.5 year) is included in the water and nutrient load. nutrient balance. The studied period is from 1970 to 2004, because of important historical nutrient loads. METHODS The hydrology of both lakes is simple. Precipitation and evaporation dominate This study is focused on the isolated and the water balance, because of the isolated mainly rainwater fed Lake Loenderveen, position of the lakes. Groundwater part of the Loosdrecht Lakes system infl uence is almost negligible. During in the Netherlands (Fig. 2). The lake is water shortage of Lake Loenderveen or oval shaped. Some characteristics: the Terra Nova, water is supplied from the size is 237 ha, the fetch is 1540 m, the adjacent Lake Loosdrecht via an inlet mean depth is 2.31 m and the residence culvert in the south. When the water time is > 1.5 years. The bottom is peaty; level is high in either Lake Loenderveen with 10% dry matter, containing 58% or Terra Nova, water is pumped to the organic matter and 12% lutum. Organic river Vecht through a pumping station. matter is mainly detritus (90%). Until Since 1973 Lake Loenderveen and Terra 1984 the water was clear and large areas Nova are separated through a weir. Water were covered with submerged vegetation passes the weir during high water levels (Best et al., 1994). In 1987 the water in Lake Loenderveen or water shortage had become turbid (Dekker et al., 1992) in Terra Nova. and all submerged vegetation had been Precipitation and evaporation gone. data were collected from the Royal The ecological status of the lake is Netherlands Meteorological Institute, predicted by the ecological lake model weather station De Bilt. Water levels of PCLake. The framework of the model is Lake Loenderveen, Terra Nova and its defi ned by an accurate water and nutrient adjacent lakes were measured. Seepage balance. Because inlet and outlet of water and infi ltration in both Terra Nova and in Lake Loenderveen is partly depended Lake Loenderveen were modelled with

Terra Nova (outlet) weir pumping station Lake Loenderveen

inlet culvert FIGURE 2. Lake Loenderveen (Google Earth 2005) 98 S.A. Schep et al. the Multi Layer Analytical Element estimated as 0.02 mg·l–1 and 0.1 mg·l–1 Model MLAEM (de Lange, 1991) and respectively. PCLake calculates P release appeared to be a linear function of the from the sediment. Nitrogen fl uxes are difference between the water levels of less important in Lake Loenderveen. the adjacent lakes and Lake Loenderveen Based on measurements, the NP-ratio in and Terra Nova. Discharges through PCLake was set to 34. weir, culvert and outlet were estimated PCLake is a mathematical, deter- as the water surplus and water shortage ministic, zero dimensional, process respectively, and compared to the based, integrated ecosystem lake model measured in and outlet. (Fig. 3). The model has been calibrated The water balance was validated on more than 40 mainly Dutch lakes and with the chloride balance. Chloride a systematic sensitivity and uncertainty concentrations are measured in inlet analysis has been performed (Aldenberg water and lake water and estimated in et al., 1995; Janse, 2005). precipitation and seepage water as 6 The aim of PCLake is to analyze mg·l–1 and 25 mg·l–1 respectively. the probability of a transition from a The nutrient balance is mainly clear water (dominated by submerged focused on phosphorus (P), as on average vegetation) to a turbid water state P is the limiting nutrient for algae and (dominated by phytoplankton), or vice water plant production. Concentrations versa, as a function of the external of total P in inlet water and lake water nutrient loading and other factors (Janse, were measured. Concentrations of P in 2005). With PCLake the critical nutrient precipitation and seepage water were loads, CNLeu and CNLoligo, are estimated,

FIGURE 3. PCLake Model Structure Possible effects of climate change on ecological functioning... 99 depending on among others water depth, both a macrophyte dominated clear and hydraulic loading rate, fetch, sediment a phytoplankton dominated turbid state. type, marsh size and water temperature A clear state is defi ned as chlorophyll-a (which is related to climate scenarios). < 50 μg/l, a turbid state as chlorophyll-a Output variables are e.g. concentrations ≥ 50 μg/l. of total N, P and chlorophyll-a, and The long term ecological changes coverage of submerged plants (Janse, were simulated by PCLake based on the 1996; 2005). results of the water and nutrient balance Climate scenarios (year 2050) are (1970–2004). The water and nutrient derived from the WB21 scenarios of balance are adjusted to evaporation and the Royal Netherlands Meteorological precipitation according to the climate Institute, based on the Second Assessment scenarios. Report of the IPCC (Kors et al., 2000; Können, 2001). In Table 1 weather data is given for different climate scenarios. RESULTS In the meantime new scenarios have been published (KNMI, 2006; IPCC, Calculated and measured discharges 2007): the expected temperature increase through inlet weir and outlet culvert is even higher. are comparable (Fig. 4). Simulated and measured chloride concentrations TABLE 1. Climate scenarios for 2050 compared to present (based on Kors et al., 2000). T – air temperature, Prec. – precipitation, Ev. – evaporation, S – summer, W – winter Scenario T Prec. S Prec. W Ev. S Ev. W Low estimate wet +0.5 % +0.5 % +3 % +4 % +4 % Central wet +1 % +1 % +6 % +4 % +4 % High estimate wet +2 % +2 % +12 % +8 % +8 % High estimate dry +2 % –10 % –10 % +8 % +8 %

Two types of simulations have been follow the same pattern as chloride done: one to determine critical loads and concentrations (Fig. 5). We conclude that one to simulate long term ecological the water balance and the nutrient balance changes. The critical nutrient loads are are suffi ciently accurate for modelling. determined by running PCLake with The water balance (Fig. 6) is dominated an adjusted water temperature for each by precipitation and evaporation. In dry climate scenario. The water temperature years the water composition can differ, is calculated with air temperature, because of inlet from Lake Loosdrecht. radiation, water depth, wind velocity and The phosphorus load (Fig. 7) is humidity. To determine critical load, the determined mainly by water inlet from simulation period was set to 35 years. Lake Loosdrecht. In 1976 the external Phosphorus load was varied within a P-load was very high. From 1984 range of 0.01 to 0.17 mg/m2·d in 13 the external P-load dropped due to steps for each scenario, starting from management measures. 100 S.A. Schep et al.

FIGURE 4. Measured and simulated cumulative fl ows (mm) of Lake Loenderveen

FIGURE 5. Measured and simulated chloride (mg/l) of Lake Loenderveen (1984–2004)

FIGURE 6. Water balance of Lake Loenderveen (1970–2004)

FIGURE 7. Phosphorus balance and phosphorus load of Lake Loenderveen (1970–2004) Possible effects of climate change on ecological functioning... 101

Critical P-loads: at present, the shift from the clear to the turbid state critical P-loads of Lake Loenderveen was simulated fi ve years earlier than the as estimated by PCLake are: 0.028 and observed shift according to chlorophyll. 2 0.065 mg/m ·d for CNLoligo and CNLeu The delayed response of the observed respectively. These values are very low chlorophyll-a compared to the P-load compared to those for ‘average’ or ‘most can be explained by retention processes common’ Dutch lakes as presented by in the sediment. These temporary Janse (2005) and Mooij et al. (2007): processes are to complex to simulate with 2 CNLoligo 0.6–1.0 mg/m ·d and CNLeu PCLake. Beside that simulation results 2–5 mg P /m2·d. This can be explained are comparable with observed changes. by the low hydraulic loading rate, the Thus, the methodology is useful for long high fetch, water depth and the high NP- term climate scenario studies. ratio (Janse, 2005). As an effect of the different climate Long term simulation: according scenarios CNLeu decreases while to chlorophyll (Fig. 8) three phases are CNLoligo remains nearly constant (Fig. distinguished: a clear water state before 9). This means that the ecological status the high nutrient load (1970–1976), a will shift easier from the clear water state shift (1977–1982) and a turbid state to the turbid state. Because temperature (1982–2004). PCLake simulated both increase according to recent scenarios clear and turbid state very well, so the (KNMI, 2006; IPCC, 2007) is even critical P-loads are accurate enough for higher, critical P-loads will decrease climate scenarios studies. However, the even more.

FIGURE 8. Measured (from 1977) and simulated (PCLake) chlorophyll-a (and ecological state)

FIGURE 9. Critical nutrient (P) loads CNLoligo and CNLeu (present and climate scenarios) 102 S.A. Schep et al.

The actual phosphorus load (Fig. the Botshol Nature Reserve, for example 10) is much higher than both critical phosphorus runoff from the adjacent phosphorus loads. This explains why grasslands will increase signifi cantly as the lake remained turbid until 2004. All a result of climate change (Rip et al., climate scenarios for 2050 will lead to an 2007). increase of the phosphorus load. Because the critical phosphorus load decrease, the turbid state becomes even more stable. CONCLUSIONS At the end of the simulation period all scenarios result in a turbid lake with a Climate change lead to changes in both short clear water period in early spring. environmental factors and ecosystem Compared to the present scenario, response. Hydraulic and nutrient load all scenarios for 2050 will lead to a increase through increasing precipitation decrease of the ecological status, as and evaporation. Because of water the chlorophyll-a concentration will be temperature related changing chemical higher (Fig. 11). and bacteriological processes and The relatively small infl uence increasing growth rates of phytoplankton of climate change on the ecological and submerged vegetation critical functioning of Lake Loenderveen can be nutrient loads decrease. explained by the hydrological isolated The decrease of critical nutrient loads position and the impact of the very high combined with an increase of the nutrient P-load in 1976. In less isolated lakes, like load might cause a clear lake to become

FIGURE 10. P-loads (cumulative) of present and climate scenarios compared to present CNL

FIGURE 11. Simulated chlorophyll-a concentrations (present and climate scenario) Possible effects of climate change on ecological functioning... 103 turbid and hamper a return towards the status and recent restoration measures. clear state. Hydrobiologia, 191, 173–188. In Lake Loenderveen the result of GULATI R.D., van DONK E. 2002: climate change is a turbid state that is Lakes in the Netherlands, their origin, and restoration: state- even more stable than the actual turbid of-the-art review. Hydrobiologia, 478, state. The results of this study show that 73–106. in the future more effort might be needed IPCC 2007: Climate Change 2007: The to reach the same ecological goals as a Physical Science Basis. Contribution of result of climate change. Working Group I to the Fourth Assessment The multidisciplinary approach Report of the Intergovernmental Panel on presented in this study provides a useful Climate Change [Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. and cost-effective method to study the Averyt, M. Tignor and H.L. Miller (eds.)]. potential effect of climate change on Cambridge University Press, Cambridge, lake ecosystems. It’s recommend that the United Kingdom and New York, NY, effects of climate changes are included USA, 996 pp. in the WFD processes to investigate the JANSE J.H. 1996: A model of nutrient sustainability of the expensive measures dynamics in shallow lakes in relation (particularly nutrient load reduction) that to multiple stable states. Hydrobiologia will be taken the in next years. 342–343 1–8. JANSE J.H. 2005: Model studies on the eutrophication of shallow lakes and ditches. PhD thesis, Wageningen REFERENCES University. KNMI 2006: KNMI Climate Change ALDENBERG T., JANSE J.H., KRAMER Scenarios 2006 for the Netherlands. P.R.G. 1995: Fitting the dynamic lake KNMI Scientifi c Report, WR 2006–01. model PCLake to a multi-lake survey KORS A.G., CLAESSEN F.A.M., through Bayesian statistics. Ecological WESSELING J.W., Können G.P. 2000: Modelling, 78, 83–99. Scenario’s externe krachten voor WB21. BEST E.P.H., de VRIES D., REINS A. 1984: RIZA/WL and KNMI publication. The macrophytes in the Loosdrecht KÖNNEN G.P. 2001: Climate scenarios Lakes: a story of their decline in the for impact studies in The Netherlands. course of eutrophication. Verhandlungen KNMI, De Bilt. der Internationale Vereinigung für MOOIJ W.M., HÜLSMANN S., DE Limnologie, 22, 868–875. SENERPOORT DOMIS L.N., NOLET De LANGE W.J. 1991: A groundwater model B.A., BODELIER P.L.E., BOERS for the Netherlands. RIZA, Lelystad. P.C.M., DIONISIO PIRES M.L., GONS DEKKER A.G., MALTHUS T.J., WIJNEN H.J., IBELINGS B.W., NOORDHUID M.M., SEYHAN E. 1992: Remote sens- R., PORTIELJE R., WOLFSTEIN ing as a tool for assessing water quality in K., LAMMENS E.H.R.R. 2005: The Loosdrecht Lakes. Hydrobiologia, 233, impact of climate change on lakes in the 137–159. Netherlands: a review. Aquatic Ecology, GOOGLE EARTH 2005: http://earth.google. 39, 381–400. com. MOOIJ W.M., JANSE J.H., DE GLATI R.D. 1990: structure in SENERPONT DOMIS L.N., HÜLSMAN the Loosdrecht lakes in relation to trophic S., IBELINGS B.W. 2007: Predicting the effect of climate change on temperate 104 S.A. Schep et al.

shallow lakes with the niem jego stanu ekologicznego. W momencie, PCLake. Hydrobiologia, 584, 443–454. kiedy rzeczywisty ładunek substancji biogennych RIP W.J., OUBOTER M., LOS H.J. 2007: znajdujących się w wodach jeziora przekracza Impact of climatic fl uctuations on wartość ładunku krytycznego, jakość ekologicz- Characeae biomass in a shallow, restored na takiego jeziora maleje, prowadząc w rezultacie lake in the Netherlands. Hydrobiologia, do zmętnienia wody oraz jej zdominowania przez 584, 415–424. glony. Zmiany klimatyczne mogą prowadzić do RYDING, S.O., FORSBERG C. 1977: zmian zarówno w odniesieniu do czynników śro- Sediments as a nutrient source in shallow dowiskowych, jak i reakcji ekosystemów. Zaist- polluted lakes. In: H. Golterman (ed.), nienie takiego typu zmian będzie miało, z pew- nością, wpływ na stan ekologiczny. Do ilustracji Interaction between sediments and fresh możliwych efektów takich zmian wykorzystano water. Dr. W. Junk, Den Haag, 227–235. i przedstawiono w niniejszym opracowaniu wyni- SCHEFFER M. 1998: Ecology of shallow ki wielodyscyplinarnych badań, obejmujących ba- lakes. Chapman & Hall. dania hydrologiczne, geochemiczne i geologicz- SCHEFFER M. 2001: Ecology of shallow ne, płytkiego jeziora pochodzenia torfowiskowe- lakes. Kluwer Academic Publishers, go (Loenderveen). Przeprowadzono identyfi kację Dordrecht, Boston, London. oraz analizę ruchów (kierunki i przepływy) wód SCHEFFER M., HOSPER S.H., MEIJER M.L., gruntowych oraz powierzchniowych, dynamiki MOSS B., JEPPESEN E. 1993: Alternative zmian wielkości ładunków substancji biogennych Equilibria in Shallow Lakes. Trends in oraz funkcjonowania ekosystemu, które zostały Ecology & Evolution, 8, 27–279. następnie wykorzystane i zastosowane w ekolo- Van LIERE L., JANSE J.H. 1992: Restoration gicznym modelu jeziora (PCLake). Implementa- and resilience to recovery of the Lake cja scenariuszy biorących pod uwagę potencjal- Loosdrecht ecosystem in relation to its ne zmiany klimatyczne przeprowadzona została phosphorus fl ow. Hydrobiologia, 233, przez wprowadzenie zmian w wielkościach opa- 95–104. dów, ewapotranspiracji oraz temperatury. Zmia- VERAART J., WESTEIN E. 2005: De ny klimatyczne będą prowadzić do zwiększenia klimaatdimensie van de Kaderrchtlijn ładunków emitowanych substancji biogennych, Water. Wageningen, Alterra. przy jednoczesnym obniżeniu krytycznego ła- dunku substancji biogennych dla jezior. W kon- sekwencji jeziora będą łatwiej przechodziły ze Streszczenie: Możliwe skutki wpływu zmian kli- stanu charakteryzującego się wodami przezroczy- matycznych na ekologiczne funkcjonowanie płyt- stymi do stanów, gdzie wody będą mętne czyli kich jezior, na przykładzie jeziora Loenderveen, łatwiej będzie następowało pogorszenie ich stanu Holandia. Ramowa Dyrektywa Wodna Unii Eu- ekologicznego. ropejskiej wymaga, aby wszystkie wody (części wód) śródlądowe i przybrzeżne osiągnęły co MS. received December 2007 najmniej dobry stan ekologiczny do roku 2015. Dobry stan ekologiczny w przypadku płytkich je- zior może być scharakteryzowany poprzez wodę o dużej przezroczystości i wysokiej jakości, w Authors’ addresses: której dominuje roślinność zanurzona. Charakter Sebastiaan A. Schep ekologicznej reakcji płytkich jezior na dopływ Witteveen+Bos, P.O. Box 233, 7400 AE Deventer, substancji biogennych w przeważającym stopniu The Netherlands, [email protected] zależy od cech morfologicznych i hydrologicz- Gerard N.J. ter Heerdt, Maarten Ouboter nych tych jezior, takich jak: głębokość wody, czas Waternet, P.O. Box 8169, 1005 AD Amsterdam, retencji wody, zmiany stanów wody, typ dna itp. The Netherlands, [email protected] Te cechy określają wielkość krytycznego ładun- Jan H. Janse ku substancji biogennych, jaki może przyjąć dane MNP/NLB (postbak 45), P.O. Box 303, 3720 AH jezioro bez konsekwencji związanych z obniże- Bilthoven, The Netherlands [email protected]