Size-fractionated chlorophyll a and phycocyanin temporal variations in a highly eutrophic lake and its isolated karstic springs

Phytoplankton constitutes a major group of primary producers in lakes and is often dominant in production, especially in eutrophic lakes (Smith et al. 1999). Three major groups of phytoplankton can be distinguished based on its size: pico- (0.2–2 μm), nano- (2–20 μm) and microphytoplankton (20–200 μm). Given that phytoplankton is the basis of the food web, its distribution to the above size fractions is of great ecological importance, because it determines the flow of matter and energy in the ecosystem (Fenchel 1988).

Chlorophyll a, present in all different representatives of phytoplankton, is commonly used as a proxy for its biomass and the trophic state of ecosystems (Hadjisolomou et al. 2016). Previous studies conducted in ocean or freshwater ecosystems have shown that the size distribution of phytoplankton varies with nutrient enrichment. In general, it has been shown that the microphytoplankton biomass increases and dominates over the smaller fractions (e.g. Watson & McCauley 1988; Kormas et al. 2002). With regard to the nano- and picophytoplankton, the findings are contrasting, because both the decrease (e.g. Bell & Kalff 2001; Søndergaard 1997; Stockner & Shortreed 1991; Szelag-Wasielewska 1997; Takamura & Nojiri 1994) and the increase (e.g. Agawin et al. 2000; Bell & Kalff 2001; Chisholm 1992; Kormas et al. 2002; Sprules & Munawar 1986; Watson & McCauley 1988) in their biomass have been reported after the ecosystem transition to the eutrophic state. In addition to nutrient enrichment, the distribution of chlorophyll a in different size classes reflects the response of phytoplankton to top-down control, because large-sized phytoplankton is well protected against grazers (Moustaka-Gouni et al. 2006).

The objective of the present study was to measure monthly variations of chlorophyll a and its size fractions as well as total phycocyanin in Lake Pamvotis and its springs in Amfithea (NW Greece). Similarly to chlorophyll a, it has been suggested that phycocyanin is a useful proxy for the actual concentration or biovolume of cyanobacteria cells (Loisa et al. 2015). Blooms of cyanobacteria are common in eutrophic lakes. In Pamvotis, cyanobacterial blooms occur throughout the year, though on a smaller scale in winter (Vardaka et al. 2005). Moreover, significant amounts of microcystins have been found in the lake, posing health hazards to fish and humans (Papadimitrou et al. 2012).

Lake Pamvotis is located in northwestern Greece, 470 m above sea level, with a mean depth of 4 m, a maximum depth of 11 m and a surface area of 22.8 km² (Kagalou et al. 2008a). In the past, its waters were entirely renewed through karstic springs, but this has changed in the last few decades due to the construction of an embankment (see Papastergiadou et al. 2010 and references contained therein on the history of the lake and human impact). At present, there is a water ski lake with an area of about 1 km² situated behind the embankment, northeast of the main lake. The drainage of water from Pamvotis takes place through a system of underground reservoirs that transfer water from the lake into the Arachthos, Louros and Kalamas rivers (Kagalou et al. 2008a). It is generally described as a shallow polymictic lake with unstable stratification in the summer season and hypoxia events (Kagalou et al. 2001). It is classified as an ecosystem of global importance (Krystufek & Reed 2004), which supports local agriculture, tourism and various fishing activities (Kagalou et al. 2008a). For the past 40 years, however, it has been a typical example of a eutrophic lake as a direct result of anthropogenic interference, especially irrigation, domestic sewage and the use of fertilizers and agrochemicals (Albanis et al. 1986; Kagalou et al. 2001; Kagalou et al. 2008a; Kotti et al. 2000; Stalikas et al. 1994).

As mentioned above, shifts in the dominance of different size fractions of chlorophyll a could be related to nutrient availability and may indicate vulnerability/resistance to grazing but also flushing and settlement processes in the lake (Moustaka-Gouni et al. 2006). For example, the increase in the internal P load in Pamvotis in summer and the decrease of N:P ratios (Kagalou et al. 2003) may lead to a change in the dominance of the largest size fraction (Masson et al. 2000). Similarly, we would expect a decrease in small size fractions in summer/autumn due to the dominance of microphagous herbivores i.e. Rotifera and Cladocera (Romero et al. 2002), which are unable to use colonial cyanobacteria. Thus, photosynthetic pigments and their size fractions could be considered as alternative (Gaedke 1995; Mouillot et al. 2006) but also complementary to phytoplankton taxonomic descriptors used previously to study Lake Pamvotis (e.g. Kagalou et al. 2008b; Vardaka et al. 2005). Considering the above, fluctuations in photosynthetic pigments could give interesting insights into ecosystem functioning. Within this framework, the objective of our study was to investigate monthly variations in (a) the distribution of chlorophyll a in different size classes and (b) phycocyanin in eutrophic Lake Pamvotis and the adjacent newly constructed water ski lake as well as to investigate possible interactions with other biotic or abiotic parameters of the ecosystem.

Chlorophyll a is a useful indicator for assessing the trophic status and water quality of surface waters, as it can be used as a proxy for phytoplankton biomass. During the study, chlorophyll a concentration in Lake Pamvotis (Site 1) was always higher than 36 μg l^-1, reaching values > 120 μg l^-1 in summer/early autumn. Our results indicate eutrophic to hypereutrophic ecological status in Lake Pamvotis with values similar to those recorded in the late 1980s (Koussouris et al. 1991) before opening and starting the wastewater treatment plant for the city of loannina (Papastergiadou et al. 2010). These values are five to six orders of magnitude higher than those recorded for the years 1998–1999, after changing the municipal sewage system in the city (Kagalou et al. 2001). A shift to eutrophic conditions was also recorded in 2005 by Kagalou et al. (2008b). The eutrophic state of Lake Pamvotis can also be perceived macroscopically by observing phytoplankton blooms throughout the year, especially in summer when they color the water and increase its turbidity. Although sampling was not performed in a network of sites which would enable us to investigate possible spatial variability, we believe that our results are quite representative of the whole ecosystem. In fact, Papadimitriou (2010) has previously indicated the lack of statistically significant differences between coastal (2 sites) and pelagic (2 sites) sampling sites for different photosynthetic pigments in Lake Pamvotis.

Considerably lower values (in general < 7 μg l^-1) were recorded at the second sampling site (Site 2) behind the Amfithea embankment that was constructed in the 1970s, causing changes in the hydraulic connection between the lake and its karstic springs (Papastergiadou et al. 2010). In 2013, the water ski lake was constructed at this site. With the exception of September when mesotrophic conditions prevail, the above values indicate an oligotrophic state (OECD 1982) for the ecosystem adjacent to the main lake ecosystem. This is also confirmed by the large Secchi depth values. In general, a great Secchi depth corresponds to smaller values of chlorophyll or phycocyanin (Tilzer 1988). In the present study, the Secchi depth was found to increase at Site 1 over the months, which seems to be due to the decline in chlorophyll a. The statistically significant correlation between chlorophyll a and Secchi depth (r = −0.81, p < 0.05) seems to confirm this finding. A different pattern (great Secchi depth with basically no monthly fluctuations) was observed at Site 2 due to the inflow of fresh water from the surrounding mountains. It has to be noted, however, that during the sampling period, the water ski track was used only in September, and more precisely just before the sampling for this month was carried out. This probably explains the peak of chlorophyll a observed during that month (17.7 μg l^-1), which classifies the lake as mesotrophic. The discharge of the lake is interrupted during water ski races, resulting in a different flushing effect.

This study was the first attempt to investigate chlorophyll a size fractionation in Lake Pamvotis. Our findings indicate a decrease in the biomass of both nano- and microphytoplankton when total chlorophyll a decreased at eutrophic Site 1 (i.e. from August to January). This is in accordance with the previous studies that showed that both the large and small size fractions can be affected by shifts in nutrient concentrations (e.g. Agawin et al. 2000; Bell & Kalff 2001). Picophytoplankton showed small fluctuations, however, it is interesting to note that its concentration reached 4.3 μg l^-1 in October. Microphytoplankton was found to be the dominant contributor to the total chlorophyll a content in late summer (August, 68.1%), while nanophytoplankton dominated in late winter (January, 81.2%). The two fractions were more or less equally represented from September to December. The positive correlation between microphytoplankton chlorophyll a, and temperature (r = 0.94, p < 0.01) indicates that under eutrophic/hypereutrophic conditions, temperature may affect community structure. Furthermore, based on elemental ratios of N and P and divergences from the Redfield ratio, Kagalou et al. (2008b) have reported nitrogen limitation in the lake in summer/autumn. Also in another study conducted simultaneously with the current one, we found T-N > 80 μg l^-1, T-P > 700 μg l^-1 and N:P ratios << 16:1 (data not shown). Nitrogen limitation shapes phytoplankton communities favoring the occurrence of N-fixers, which in turn may promote the growth of large Microcystis colonies (Beversdorf et al. 2013). Other, experimental studies have also shown that large phytoplankton dominates in nitrogen-limited systems, while in phosphorous-limited systems, picophytoplankton abundance and biomass increases (Stockner & Shortreed 1991; Masson et al. 2000). Finally, we should mention that large-size phytoplankton is considered well protected against grazing and thus dominant in eutrophic ecosystems (Moustaka-Gouni et al. 2006). At the second sampling site (Site 2, Springs), the contribution of microphytoplankton to the total chlorophyll a only once exceeded 50%. In all other cases, considering the criteria defined by Wang et al. (2013) for the dominant size class, nanoplankton contributed the most to chlorophyll a. However, we should mention that the contribution of picoplankton exceeded 25% twice. There is evidence from previous studies that pico- and nanoplankton, with a high surface to volume ratio, are better competitors for nutrients than larger plankton under oligotrophic conditions (Andersson et al. 1994). Interestingly, the disturbances that occurred in September at Site 2 favored nanophytoplankton. This indicates a close association between this size fraction and ecosystem nutrient enrichment that may have occurred because, as mentioned before, the track is cut off from the natural outflows during water ski races and the hydrology of the system is changing. Further studies are needed to investigate how these manipulations affect flushing rates of the system and consequently its physicochemical and biological characteristics.

Lake ecosystems, such as Lake Pamvotis, which owe their eutrophication state to human activities tend to experience a massive increase in cyanobacteria (Moustaka-Gouni 1993; Watson et al. 1997; Moustaka-Gouni et al. 2006). In this paper, the concentration of the blue pigment, phycocyanin, was used as a proxy for cyanobacterial abundance fluctuations during the sampling period. Several papers have been published on phycocyanin extraction methods (Horváth et al. 2013; Lawrenz et al. 2011; Zhu et al. 2007; Zimba 2012). In all cases, it was suggested that there is an underestimation of its concentration due to methodological issues. In the present study, the measurement of phycocyanin was based on the protocol of Horváth et al. (2013), with changes concerning the filter type, freeze/thaw cycles and the amount of buffer solution, to create the optimal conditions for pigment extraction from our samples. The results showed that Site 1 had higher concentration of phycocyanin compared to Site 2. This also suggests that Site 2, close to the springs, is less enriched with nutrients compared to Site 1 (Li et al. 2017). However, our data probably indicate an important underestimation of the total phycocyanin, given that cyanobacterial blooms occur in Lake Pamvotis throughout the year, though on a smaller scale in winter (Vardaka et al. 2005). Long-lasting blooms of these organisms color the water and create an unpleasant odor (Lee 2008). Both features were observed in Lake Pamvotis during the study period. Furthermore, previous studies of eutrophic freshwater ecosystems (Lyu et al. 2013; Randolph et al. 2008; Simis et al. 2005; Song et al. 2012; Pyo et al. 2017; Wang et al. 2014) have shown phycocyanin concentration one order of magnitude higher compared to this study. However, most of the above studies found a high standard deviation on a spatial or temporal scale, while some authors (Randolph et al. 2008; Song et al. 2012) concluded that the technique used for the calculation of phycocyanin gave results several classes lower than expected. In our study, phycocyanin was statistically significantly correlated with total chlorophyll (r = 0.98, p < 0.01) as well as with micro- (r = 0.97, p < 0.01) and nanophytoplankton (r = 0.84, p < 0.05) fractions, indicating that cyanobacteria represent an important component of large-sized phytoplankton in Lake Pamvotis. Microcystis aeruginosa (Kützing) Kützing is the most widespread and most often dominant species of cyanobacterial blooms in Lake Pamvotis and other Greek lakes (Katsiapi et al. 2011; Vardaka et al. 2005). Moreover, Cook et al. (2004) reported that the scum in Lake Pamvotis during the summers of 1997–2000 was caused by M. aeruginosa and Dolichospermum flos-aque (Brébisson ex Bornet & Flahault) P.Wacklin, L.Hoffmann & J.Komárek (formerly Anabaena flos-aque). These two species were also identified in this work in November, when a sample check was carried out (data not shown).

An increase in the bacterial abundance was observed in late autumn (October–November) at Site 1.This is probably due to the bacterial degradation of organic matter, produced during summer by the excessive growth of phytoplankton (Agustí & Duarte 2013; Buchan et al. 2014; Ye et al. 2014) or due to bacterial degradation of microcystins (Moustaka-Gouni et al. 2006). The reduction of dissolved oxygen in October at Site 1 supports this finding, which probably indicates an increase in the activity of the microbial loop (Azam et al. 1983). At Site 1, bacterial abundance peaks again in January, probably due to the inflow of allochthonous material in the form of sewage after heavy rain fall (Ye et al. 2014). Indeed, the rainfall in January was 19.8 mm compared to < 0.2 mm from August to December, except in October when it reached 3 mm (Pedini meteorological station, loannina; http://stratus.meteo.noa.gr/front). Similarly, the increase in bacterial abundance at Site 2 in October may be related to the high chlorophyll a content measured at this site in the previous months.

To sum up, the biomass of both nano- and microphytoplankton fractions increases when the total chlorophyll a increases under eutrophic/hypereutrophic conditions of Lake Pamvotis and the alternation in the dominance of these two size fractions was observed throughout the study period. Picophytoplankton contribution was low (≤ 0.03%) but its chlorophyll a content reached 4 μg l^-1 in late autumn. At the same time, the enrichment with nutrients in the ski track adjacent to the main lake, which receives freshwater directly from the karstic springs of Mitsikeli Mountain, seems to be limited. Under these conditions, total chlorophyll a concentration is low and the nanophytoplankton fraction alone or together with picophytoplankton dominates. Interestingly, the nanophytoplankton seems to respond quickly and thrive when the ecosystem is disturbed. The above findings may have implications for the restoration of the eutrophic/hypereutrophic lake through water management and, in particular, the feeding of the lake through its natural springs.