We depict in Fig. 6 a scheme of the vertical stratification of physical, chemical and biological components in two saline lakes Shira and Shunet. Notably, patterns of physico-chemical stratification of the two lakes are quite different mainly because both mean depth and maximum depth of lakes differ as well as their salinity levels differ too. Of these 2 lakes, Lake Shira is deeper and less saline has a chemocline with variable depth. Depending on the weather conditions, the chemocline can sink to deeper waters (16 m) or ascend to shallower depth (11 m). However, we did not observe mixing of entire water column, nor is it reported in literature. Considering that the mixolimnion of Lake Shira is relatively deep and that it is surrounded by hills, the mixolimnion is also thermally stratified during the summer period. The thermocline is formed at the depth of 3–4 m, usually in the middle of June. In summer, the surface waters become warmer and the thermocline sinks downwards to 8–9 m depth by the end of August. Thus, in the summer time, the three layers are an upper warm epilimnion, the metalimnion (where the temperature drop is >1°C per m) and an oxic hypolimnion, which is separated from the anoxic monimolimnion by the chemocline.
This type of physico-chemical stratification in L. Shira leads to stratification of biological components. The chemocline of the lake, as in many meromictic lakes, is inhabited by bacterial community consisting from purple sulphur and heterotrophic bacteria. However, because the depth of the chemocline is variable, the bacterial community does not reach high densities. When the chemocline shifts downwards or upwards the bacterial community is “diluted”. The thermally stratified in summer, the mixolimnion also creates different habitats for various species. The distribution of phytoplankton is non-uniform with the peak of biomass in the metalimnion. The distribution of zooplankton is also non-uniform with rotifers and juvenile copepods in warm epilimnion and older copepods in the cold oxic hypolimnion (Zadereev and Tolomeyev 2007). Zooplankton comprised of the calanoid Arctodiaptomus sp. and rotifers feed mostly on phytoplankton. Moreover, we consider the bacterial community in the chemocline as a not-essential part of the trophic chain in the mixolimnion. The amphipods which can be assigned to the higher trophic link in the lake’s ecosystem are also distributed non-uniformly, with their peak densities generally associated with the thermocline. They stay in pelagic region, feed on seston and probably also on zooplankton. However, studies of fatty acid composition of Gammarus reveal no traces of any zooplankton prey in its gut.
The chemocline in Lake Shunet is located at 5 m, and unlike for Lake Shira, this depth varies only narrowly, ±20 cm, which is certainly due to a very sharp salinity gradient between mixolimnion (17–20 g l−1) and monimolimnion (up to 66 g l−1). As Lake Shunet has much lower mean depth and the mixolimnion in Lake Shunet is not deep enough, like in Lake Shira, its thermal stratification in summer is not stable. Because the depth of the chemocline in Lake Shunet is stable and the mixolimnion is relatively shallow, the chemocline is inhabited by the extremely dense bacterial community, a population of Cryptomonas sp. and ciliate community comprising several species. In contrast, in Lake Shira the densities of both Cryptomonas sp. and ciliates are low and more evenly distributed in the water column. As the mixolimnion of Lake Shunet is not thermally stratified for a long period, the vertical distribution of phytoplankton and zooplankton too is not vertically stratified. However, gammarids tend to concentrate in narrow layer located 1–2 m above the chemocline. The trophic interactions in Lake Shunet appear to partially differ from those in Lake Shira, where bacterial community is much denser and Cryptomonas sp. and several ciliate species inhabit the chemocline. Both ciliates and zooplankton feed on bacterial community in the chemocline. Both the ciliates and calanoid, A.salinus, also feed on Cryptomonas sp. which are most probably mixotrophic in the chemocline and consume bacteria.
Despite our several studies (e.g. Degermendzhy et al. 2002; Gaevsky et al. 2002; Rogozin et al. 2005; Tolomeyev et al. 2006; Zadereev and Tolomeyev 2007) on the vertical structure of L. Shira and L. Shunet, our knowledge on the food web interactions and mass and energy flows is rather scanty. We still do not have a good grasp of the impact of microbial loop and bacterial and ciliates communities in the chemocline on the macrobial food web and energy budget in the mixolimnion. Moreover, we still do not quite understand the effect of Gammarus population (Zadereev et al. 2010) on the pelagic food web. The studies of trophic interactions are the subject for futures studies. Also, a missing link in our research, as in many research studies on lakes, is the interactions between littoral and pelagial lake zones. Because gammarids are quite abundant in lake’s littoral as well as its pelagial and perform horizontal migrations between these two regions (Yemelyanova et al. 2002), the work on Gammarus needs to be integrated more coherently with lakes’ other food chain studies.
Both L. Shira and L. Shunet, with their complex stratified vertical structures appear to be very sensitive to external changes especially to climate change. First, meromixis is sensitive to lake water level, which depends on the input–output balance of water, depending mainly on annual precipitation and evaporation. The water levels of our study lakes are quite variable, as evident from great changes witnessed over last 150 years. The increase in water level though increased input of rain water or anthropogenic run-off leads to a decrease in salinity and consequently to less stable stratification. Thus, with the consistent rise of water level, meromixis can be eventually destroyed. The water level of both lakes increased during last one decade. We have, however, no information if this increase was due to climate change or just a local, temporal trend. The effect of climate changes on the local conditions is difficult to assess. However, what we at this stage can model and predict is the effect of lake water level increases on the changes in stability of the stratification of water column. Such an estimate will enable us to monitor and account for possible changes in lake properties in future.
The stratification of physicochemical and biological parameters in the ecosystems of lakes Shira and Shunet, and the mathematical analysis shows that there are no simple and uniform mechanisms to explain how stratification is formed (Fig. 7). The depth of the chemocline in L. Shira is determined by meteorological conditions (air temperature and wind) rather than by microbial activity. Also, in Lake Shira, where stratification is less stable, the development of the bacterial community is primarily limited by the fluctuating depth of the chemocline On the other hand, in L. Shunet, where the salinity gradient is more marked and the stratification more stable, also the biological bacterial activity can determine the depth of the chemocline.
The phytoplankton is non-uniformly distributed with depth in L. Shira. The vertical distribution of many phytoplankton species is determined together by both physico-chemical and biological processes. For example, sensitivity analysis of the mathematical model of the Lake Shira ecosystem (Prokopkin et al. 2010) shows that the depth of the biomass maximum of green algae is strongly affected by maximum sedimentation rate. The vertical distribution of green algae is also influenced by trophic pressure of zooplankton (Fig. 4b) and others factors (Fig. 7).
Cryptomonas sp. and several ciliate species show distinct preference for a certain depth; they establish dense community in the chemocline of Lake Shunet. Phytoflagellates (Cryptomonas sp.) may inhabit the stable and permanent chemocline because of the availability there of nutrients and organic substrate. We know that many genera of phytoflagellates can switch to mixotrophic feeding in fresh and marine aquatic ecosystems (Porter 1988; Sanders 1991). Another possible mechanism of formation of deep phytoflagellate maximum is the absence of large predators, too sensitive to withstand high content of H2S. Ciliate density maximum above the chemocline may be due to both the high content of nutrients and the high tolerance of ciliates to H2S (Fig. 7).
Presence of Gammarus lacustris, traditionally regarded as a benthic-littoral species in the lake’s pelagic area, raises certain questions as to how these animals adapt to such a habitat. G. lacustris not only inhabits the pelagic zone, but also forms a strongly marked non-homogeneous distribution with its maximum in the thermocline zone. Again, the role of hydrophysical and biological factors in this distribution is not well understood. On the one hand, the average depth where the population occurs is consistently associated with the thermocline (Zadereev et al. 2010). The thermocline of these 2 saline lakes is the depth zone where the amphipods are maximally buoyant. Also in the thermocline zone, the conditions are optimal for growth and development of the amphipods, i.e. low temperatures, high oxygen concentrations and high seston concentrations (Fig. 7).