Hydrobiologia 506–509: 265–272, 2003. © 2003 Kluwer Academic Publishers. Printed in the Netherlands.
265
Phytoplankton response to changed nutrient level in Lake Peipsi (Estonia) in 1992–2001 Külli Kangur, Tõnu Möls, Anu Milius & Reet Laugaste Institute of Zoology and Botany of the Estonian Agricultural University, Võrtsjärv Limnological Station, 61101 Rannu, Estonia E-mail:
[email protected] Key words: shallow lake, recovery, nutrients, phytoplankton, water level
Abstract In the 1990s, the nutrient load from the catchment area, carried into the large shallow eutrophic Lake Peipsi (area 3555 km2 ), diminished in comparison with the 1980s. The aim of the investigation was to analyse statistically biotic response to alterations in nutrient level in the lake following the reduction in external loading. In-lake changes of the relationship between phosphorus and nitrogen compounds and phytoplankton were analysed for 10 years (1992–2001). The content of biogenic elements (total phosphorus, orthophosphate, total nitrogen and ammonia) indicated a decrease and thereafter a stabilization of the parameter values during the observation period. An opposite trend was observed for the nitrite and nitrate ions. The lowest Ntot:Ptot ratio (about 11–13) was registered in 1995–96. However, despite a definite decrease of the nutrient level, the biomass of phytoplankton (particularly cyanobacteria) displayed an increasing trend. Strong and long-lasting (up to October–November) algal blooms were noted in recent years. The structure of the phytoplankton community changed considerably: the share of cyanobacteria (particularly, heterocystous forms) demanding less nutrients increased while the contribution of chlorophytes decreased. The amount of nutrients per biomass unit of different phytoplankton groups decreased. Water level appears to be a significant factor affecting the content of dissolved inorganic nutrients. In particular, NO3 N increased and PO4 P decreased at higher water level. Large inter-annual water level fluctuations will decrease stability of the ecosystem.
Introduction The state and evolution of aquatic ecosystems are affected by various biotic and abiotic factors, as well as by natural and man-induced processes at different levels and with a different duration (Punning et al., 1999). The eutrophication process has been studied extensively, comparatively fewer studies exist on the response of lakes to reduced loading (Jeppesen et al., 2002). Various hypotheses of biological response to re-oligotrophication are proposed. Biological changes are expected, simultaneous or delayed, in relation to nutrient changes, and both chemical and biological reversibility has been assumed (Wilander & Persson, 2001). However, there exists growing awareness that the effects of eutrophication can rarely be reversed by nutrient reduction from the catchment area alone
(Madgwick, 1999). Within a certain trophic level the internal mechanisms of the circulation of matter can compensate to some extent for variations in the external load (Punning et al., 1999). The intensive anthropogenic eutrophication of L. Peipsi started in the 1970s (Starast et al., 2001). The pollution load from the catchment area to L. Peipsi was the highest in the 1970s and 1980s (Loigu et al., 1999). Since the early 1990s, the nutrient content in small rivers has continuously decreased owing to lower agricultural production and smaller amounts of waste water discharged into the rivers (Leisk & Loigu, 2001). The aim of the present investigation was to analyse statistically the biotic response to the decreased nutrient level in L. Peipsi following reduction in external loading. For 10 years we examined in-lake
266
Figure 1. Yearly changes in nutrient content and Ntot :Ptot ratio in L. Peipsi s.s. The estimated values (continuous line) and 95% confidence limits (dashed lines) were calculated for 19 July using the model fitted to the whole dataset.
changes in the concentration of phosphorus and nitrogen compounds as well as phytoplankton as a variable of response. We analysed the relationships between nutrients and phytoplankton biomass as well as the relative abundance of different groups in phytoplankton. Water level as an important factor in the dynamics of the ecosystem of L. Peipsi was also included in analysis.
Materials and methods Study site Lake Peipsi, the fourth largest lake in Europe, is located on the border of Estonia and Russia. The total area of the lake is 3555 km2 , mean depth is 7.1 m, and maximum depth is 15.3 m (Jaani & Raukas, 1999). L. Peipsi consists of three parts: the largest and deepest northern L. Peipsi s.s. (2611 km2 , mean depth 8.3 m, maximum depth 12.9 m), the southern part L. Pihkva (708 km2, mean depth 3.8 m, maximum depth 5.3 m), and the narrow strait-like L. Lämmijärv (236 km2 , mean depth 2.5 m, maximum depth 15.3 m) connecting them. The volume of the whole lake is 25.07 km3 and water residence time is about 2 years. The main inflows are the Velikaya River in the south and the Emajõgi River in the west. The only outflow, the Narva River, connects the lake with the Gulf of Finland. Average water level is 30 m above sea level but water level fluctuations in L. Peipsi are large, 3.04 m for the last 80 years (Jaani, 1996). The average
annual range of water level fluctuations is 1.15 m. The lake is well mixed by the waves and currents. The ice cover lasts from December to April. Lake water is the warmest (21–22 ◦ C in open water) in July– August. The lake is clearly heterogeneous in physical, chemical, biological, and other parameters. In L. Peipsi, humic substances (0.05–2.00 mg l−1 ) make the water brown. The total content of organic matter is characterized by a chemical oxygen demand (CODCr ). In 1992–2001, CODCr was mostly 14–59, mean 29 mg O2 l−1 . Since the 1990s, a clear decrease in CODCr has been registered in the northern part of the lake. In 1992–2001, water transparency by Secchi disk was mostly 0.9–3.3 with an overall mean of 1.7 m. According to hydrochemical classification (Alekin, 1953), the water of L. Peipsi belongs to the calcium (Ca) group of the hydrogen carbonate class. The ionic composition of water is dominated by bicarbonate (HCO3 − ) followed by sulphate (SO4 2− ) and chloride (Cl− ) ions. The Ca2+ is the most abundant cation, followed by magnesium (Mg2+ ), sodium and potassium (Starast et al., 2001). Average total alkalinity of L. Peipsi was 2.43 mmol l−1 in 1992–2001. Strong anthropogenic impact on the lake, reflected by the increase in SO4 2− and Cl− ions from 1950s up to the late 1980s (Starast et al., 2001), was diminishing in all lake parts during the 1990s. In 1992–2001, mean concentrations of SO4 2− and Cl− were 0.16 and 0.21 mmol l−1 , respectively, and dropped to the level of the late 1960s. The content of Ca2+ and Mg2+ varied relatively little, being 1.02 and 0.47 mmol l−1 ,
267
Figure 2. Yearly changes in the biomass of phytoplankton (FBM), cyanobacteria (CY), cryptophytes (CRYP) and diatoms (BAC) in L. Peipsi s.s. The estimated values (continuous line) and 95% confidence limits (dashed lines) were calculated for 19 July using the model fitted to the whole dataset.
respectively. The water of L. Peipsi is rich in dissolved oxygen. In 1992–2001, O2 content was usually 7.8– 14 mg l−1 , oxygen saturation 80–125%; the overall mean 10.6 mg l−1 (100%). The water of L. Peipsi is slightly alkaline: mean pH was 8.42 (95% tolerance limits 7.7–9.1).
et al., 1996). Phytoplankton samples were preserved with Lugol’s solution. Fixed materials were identified and counted using the Utermöhl (1958) technique. The composition of the phytoplankton community was studied mainly at the division level: cyanobacteria (cyanoprokaryotes), chlorophytes, diatoms, chrysophytes, dinoflagellates and cryptophytes.
Sampling and analyses Statistical methods Changes in nutrient concentrations and phytoplankton biomass were estimated during the growth season starting from 1992. Seasonal observations (3 times) from May to November 1992–1996 were carried out at 7–16 stations. At 4–5 of these stations the observations were continued in 1997–2001 using monthly sampling between April–May and November. Additional sampling was performed in March 1994 and 1999. All stations were located in the Estonian part of the lake (except for 9 stations in the Russian side of the lake sampled in 1992 and in October 2001). Integrated water samples for phytoplankton analysis were taken from the surface and with 1-m intervals through the water column; samples for chemical analysis were taken from a depth of 0.1–1.0 m, both with a Ruttner sampler. The chemical composition of water (total phosphorus – Ptot , total nitrogen – Ntot , orthophosphate ion – PO4 P, nitrate ion – NO3 N, nitrite ion – NO2 N, ammonium ion – NH4 N) was analysed at Tartu Environmental Researchers Ltd, Estonia. Methods of chemical analysis have been published previously (Möls
All chemical, physical and plankton variables, except for pH and temperature, were log2 -transformed to improve their statistical properties. Prior to this, zero values were replaced by appropriate small numbers to enable the calculation of logarithms. The whole data array was carefully filtered by inspecting studentized residuals. The means of logarithm-transformed variables are the geometric means. As in our previous study (Möls et al., 1996), we employed the technique of general linear models, provided by the SAS System, Release 8.1 (SAS Institute Inc., 1996), especially GLM and MIXED procedures. Two large regression models were used throughout the study. The first model describes each water variable separately. In the second model, the dependent variables (Ptot, Ntot etc) were packed together to a single variable. Each value of this formal variable means the value of a chemical or biological variable and was treated as the repeated measurement carried out in a given water sample. In the corresponding statistical repeated measures analysis, the
268 within-subject covariance structure was specified as UN (‘unstructured’) in the sense of SAS terminology. On the basis of these models, the water variables and their ratios were predicted for each study year (given the 200th day of the year and the sampling site in Peipsi s.s.). For predicting and calculating the 95% confidence limits, specific tailored ESTIMATE statements of the SAS GLM or MIXED procedure were used. These predictions were thereafter used for making graphs of long-term changes with the corresponding confidence limits. The dependence of water variables on the mean water level as well as on minimum water level was estimated analogously. To eliminate the effect of seasonality, the variables were adjusted to 19 July date. The mean water level and minimum water level were calculated for the quarter of year when the water sample was taken. To test if a variable (or the ratio of variables) was increasing or decreasing in a particular year, special ESTIMATE statements corresponding to the partial derivatives of the model regression function were tested for zero. If the estimate of the derivative for a given year was significantly (on the level of α=0.05) different from zero, then the variable (or the ratio of variables) was considered to be increasing or decreasing in that year, depending on the sign of the estimate.
Results Changes in nutrient level In 1992–2001, the concentration of nutrients varied in a broad range (Table 1). The Ptot content for L. Peipsi s.s. decreased significantly during 1995–1999 (Fig. 1). The course of PO4 P too was clearly decreasing during 1993–1999 (Fig. 1). A decreasing trend in Ntot content was observed from the beginning of the observation period up to 1995 (Fig. 1). The NH4 N content was clearly decreasing in L. Peipsi s.s during 1992–1995. Contrary to NH4 N, NO3 N and NO2 N contents in L. Peipsi s.s. increased until the mid-1990s. The Ntot :Ptot ratio decreased significantly in 1992–1994 (Fig. 1). In the following years the changes were not significant (Table 2). The lowest Ntot :Ptot ratio (about 11–13) was registered in 1995–1996.
Response of phytoplankton to changes in nutrient level The two main phytoplankton groups in L. Peipsi are diatoms (BAC) and cyanobacteria (CY, Table 1). The character species of the lake is Aulacoseira islandica (O. Müller) Sim. which may account for up to 96% of biomass in spring and autumn. A. granulata (Her.) Sim. and Stephanodiscus binderanus (Kütz.) Krieger are prevalent in the warm period among BAC. CY can yield maximum biomasses in summer and autumn. Gloeotrichia echinulata (Smith) Richter and Anabaena and Microcystis species are abundant in summer, while Aphanizomenon flos-aquae (L.) Ralfs is numerous mostly in autumn. Chlorophytes (CHL) had an essential share in phytoplankton biomass. Some other groups as cryptophytes (CRYP) and chrysophytes (CHR) exist in plankton in vegetation period, sometimes very abundantly, but do not reach large biomasses (Table 1). Dinophytes (DINO) occupy a modest place in phytoplankton biomass. Phytoplankton biomass (FBM), especially the biomass of CY, did not follow the dynamics of nutrients but showed an increasing trend (Fig. 2). In L. Peipsi s.s., a significant increase in FBM was observed in the years 1994–1998, while the biomass of CY increased in 1994–1999. The biomass of CRYP increased in L. Peipsi s.s. in 1993–1995; BAC increased in 1994– 1997. On the contrary, the biomass of CHR and CHL decreased in the early 1990s. Changes in DINO were not significant. In L. Peipsi, strong and long-lasting algal blooms have been observed in recent years, despite a definite decline in the nutrient content of surface water. The dynamics of FBM depends on some large dominant species, as the filiform cyanobacteria A. flos-aquae, which usually dominates in August–September. In the warm summer of 1999, the most intensive algal bloom occurred in July and August, whereas in 2000 and 2001 such a bloom was shifted to the autumn months (September–November) (Fig. 3). Relations between nutrients and different groups of phytoplankton The Ptot :FBM ratio as well as PO4 P:FBM decreased clearly in the studied years (Table 2, Fig. 4). The PO4 P as well as Ptot amount per FBM unit decreased (Fig. 4). As the trends of Ptot and PO4 P were decreasing, changes in the PO4 P:Ptot ratio were not significant. The Ntot:FBM ratio decreased signific-
269 Table 1. Content of nutrient elements and phytoplankton parameters in Lake Peipsi and in its parts in 1992–2001. Tolerance limits bound approximately 95% of all observed values Variable
Unit
Ptot PO4 P Ntot NO3 N NO2 N NH4 N FBM BAC CY CHL CRYP CHR DINO
mg P m−3 mg P m−3 mg N m−3 mg N m−3 mg N m−3 mg N m−3 g m−3 g m−3 g m−3 g m−3 g m−3 g m−3 g m−3
Geom. mean 46 11 698 72 1.9 24 5.0 2.38 0.69 0.14 0.12 0.007 0.004
95% tolerance limits 18 2 270 11 0.3 2 0.7 0.18 0.02 0.02 0.01 0.000 0.000
Mean for L. Peipsi s.s.
113 56 1830 465 12.4 250 37.5 30.88 29.93 0.99 1.78 0.30 3.62
41 10 620 58 1.6 18 4.2 2.02 0.54 0.13 0.12 0.006 0.007
Mean for L. Lämmijärv 56 12 803 89 1.7 31 6.7 3.15 0.90 0.15 0.10 0.007 0.001
Mean for L. Pihkva 72 15 930 89 1.8 55 7.3 4.66 0.83 0.21 0.13 0.023 0.002
of chlorophytes in FBM decreased (Fig. 4, Table 2). The prevalence of cyanobacteria over chlorophytes increased (Table 2). Response of nutrients and phytoplankton on changes in water level
Figure 3. Dynamics of Aphanizomenon flos-aquae in L. Peipsi s.s.
antly during the years 1994–1998 (Table 2, Fig. 4). A significant decrease in the ratio of dissolved inorganic N compounds (NH4 N+NO3 N+NO2 N=DIN) to FBM was observed in 1994–1998 too. There occurred no significant change in the DIN: Ntot ratio nor DIN:PO4 P ratio during the study years. The ratios of Ptot , Ntot , PO4 P and DIN to different groups of phytoplankton (CY, BAC, CHL, CRYP, DINO) decreased for several years, except for PO4 P:CY and PO4 P:CHL which increased at the beginning of the study period (Table 2). Thus for different phytoplankton groups the amount of nutrients per biomass unit decreased. The relationships between the algal groups changed significantly during the last decade: cyanobacteria became more dominating in phytoplankton, whereas the share
Water level was represented by the two characteristics: mean level and minimum level for the quarter of observation. Among nutrients, the compounds of dissolved inorganic nitrogen were influenced by water level most significantly (P < 0.0001). The NO3 N concentration was higher at higher water level, while PO4 P content increased at lower water level (Fig. 5). The Ntot :Ptot ratio was clearly (P
