Both surface water temperatures and the intensity of thermal stratification have increased recently in large lakes throughout the world. Such physical changes can be accompanied by shifts in plankton community structure, including changes in relative abundances and depth distributions. Here we analyzed 45 years of data from Lake Baikal, the world's oldest, deepest, and most voluminous lake, to assess long-term trends in the depth distribution of pelagic phytoplankton and zooplankton. Surface water temperatures in Lake Baikal increased steadily between 1955 and 2000, resulting in a stronger thermal gradient within the top 50 m of the water column. In conjunction with these physical changes our analyses reveal significant shifts in the daytime depth distribution of important phytoplankton and zooplankton groups. The relatively heavy diatoms, which often rely on mixing to remain suspended in the photic zone, shifted downward in the water column by 1.90 m y-1, while the depths of other phytoplankton groups did not change significantly. Over the same time span the density-weighted average depth of most major zooplankton groups, including cladocerans, rotifers, and immature copepods, exhibited rapid shifts toward shallower positions (0.57–0.75 m y−1). As a result of these depth changes the vertical overlap between herbivorous copepods (Epischura baikalensis) and their algal food appears to have increased through time while that for cladocerans decreased. We hypothesize that warming surface waters and reduced mixing caused these ecological changes. Future studies should examine how changes in the vertical distribution of plankton might impact energy flow in this lake and others.

Citation: Hampton SE, Gray DK, Izmest'eva LR, Moore MV, 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. https://doi.org/10.1371/journal.pone.0088920

Editor: Adrianna Ianora, Stazione Zoologica, Italy

Received: August 21, 2013; Accepted: January 16, 2014; Published: February 25, 2014

Copyright: © 2014 Hampton et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: The work was supported by the National Science Foundation Dimensions of Biodiversity program (DEB-1136637; http://www.nsf.gov/) and the Brachman Hoffman Fund at Wellesley College. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Climate change is significantly impacting freshwater ecosystems worldwide. Recent studies indicate that many lakes are experiencing physical changes that include warmer surface water temperatures, altered water levels and wind patterns, longer ice-free periods, altered thermal stratification, and changes in water transparency [1]–[3]. Ecologists are beginning to understand the direct and indirect effects of these physical changes on biological communities [4]. Some of the documented responses of plankton to climate change include changes in abundance, phenology, body size, community structure, life history parameters, and vertical migration patterns (reviewed in [4]).

Altered thermal stratification is one of the most consequential indirect pathways through which climate affects plankton. Stratification not only provides vertical thermal structure, but it also alters the distribution of nutrients and plankton [5]–[7]. During periods of summer stratification lakes are often separated into a warm, shallow, well-lit epilimnion and a deep, cool hypolimnion that receives less solar energy. As time passes after the onset of stratification, nutrient availability can become reduced in the upper stratum due to the lack of vertical mixing that brings nutrients up from the hypolimnion [8]–[10]. Heavier plankton and those without buoyancy or mobility mechanisms may sink away from the upper waters where light is most readily available [5], [11].

The effects of climate change on summer stratification can be highly system-specific, complicating ecological predictions [12]. However, two decades of modeling studies and empirical observations of deep northern temperate lakes indicate that climate change is altering stratification [13]. In general, the length of the stratification period and thermal stability has increased through time [12], [13]. These changes have been linked to observed shifts in plankton communities [13]. For example, in Lake Tahoe surface waters the algal community has shifted toward small, slow-sinking taxa as turbulent mixing has decreased over the past 23 years, with a downward shift of relatively heavy diatoms [11].

While previous studies have explored the effects of increased thermal stability on the depth distribution of phytoplankton species, few have explored how zooplankton might respond to changes in stratification (reviewed by [4]). The vertical position of zooplankton in the water column frequently exhibits a diurnal pattern whereby individuals are found in deeper waters during the day but migrate closer to the surface at night [14], [15]. This vertical migration is thought to be a behavioral adaptation that balances the risk of predation from visually orienting predators with the potential benefits of inhabiting the epilimnion, such as access to food and the metabolic benefits of warmer ambient temperature.

There is now a large body of literature examining the factors driving daily variation in zooplankton depths (i.e. the factors responsible for vertical migration), but less attention has been given to vertical positioning across seasons or years. The few studies that have examined seasonal differences in zooplankton vertical distribution suggest that stratification plays an important role. Thackeray et al. [16], [17] found that the onset of stratification and the vertical position of the thermocline were both related to zooplankton depth distributions. Several other seasonal studies reported that zooplankton tend to shift to shallower positions when summer stratification sets in [18]–[20]. In a more direct test of the impact of thermal structure, Marcogliese and Esch [21] demonstrated that artificially deepening the epilimnion caused simultaneous changes in the depth distribution of zooplankton. Taken together, the previous work indicates that alterations in stratification due to climate change may have strong effects on the depth distribution of zooplankton.

Subarctic Lake Baikal may be especially sensitive to changes in mixing patterns. The lake is covered with ice for almost half the year, from January to May, with stratification occurring weakly for about 6 to 8 weeks in August and September [22], [23] and also under the ice in winter [24]. Density gradients are relatively low at Lake Baikal's low water temperatures, and summer stratification is readily broken down by upwellings, storms, and wind events [25], [26]. Thus the dominant plankton are well adapted to mixed, dynamic environmental conditions. However, like many other lakes worldwide, Lake Baikal has experienced dramatic warming. The ice-covered period is shorter and ice thickness has decreased [27]. Warming has been strongest in the summertime, and in the upper stratum [28], [29]. Warming is not yet manifest in deeper waters (>50m), implying that summer stratification should be stronger, and thus may have the potential to last longer [28]. These ongoing changes in surface temperatures and thermal stratification are expected to lead to a shift in pelagic phytoplankton communities away from one dominated by the coldwater diatoms Aulacoseira baicalensis and Cyclotella minuta to one dominated by green and cyanobacteria picoplankton [30], [31].

In this study we use 45 yr of data from Lake Baikal to examine how the depth distribution of major zooplankton and phytoplankton groups has changed through time. In addition, we explore the implications that changes in depth distributions may have for interactions between phytoplankton and their zooplankton grazers. Our results provide further evidence that significant long-term changes are occurring in Lake Baikal's plankton community and that these changes are likely driven by climate.

Data used in the study are part of a historic Russian data set, registered with the Russian government (No. 2005620028). No endangered, protected, or vertebrate species were targeted in those sampling efforts. No contemporary data were collected for this study.

Since 1945 researchers from Irkutsk State University (ISU) have collected daytime plankton, temperature and Secchi depth data at least monthly, usually every 7–10 days, in depth profiles from the surface to at least 250 m at a single main station approximately 2.7 km offshore from Bol'shie Koty in the Southern Basin (Fig. 1). This station is not influenced by discharge from the Baikalsk pulp mill, more than 80 km to the south [22], [32]. While limitations are presented by analyzing data from a single station, trends in plankton abundance at this station are similar to those reported for a second location in the Southern basin [33]. Sampling did not occur during crepuscular hours, as diel vertical migrations are well known for many Baikal plankton. We have focused our analyses on the summer months in which stratification most frequently occurs – July, August, and September.

https://doi.org/10.1371/journal.pone.0088920.g001

Temperature was measured with a mercury thermometer in water collected at discrete depths by a 10 L Van Dorn bottle; those measurements used here were from depths of 0, 10, 50, 100, and 200 m. Phytoplankton samples obtained at these same depths with the Van Dorn bottle were preserved before settling in Utermöhl chambers. A change in phytoplankton preservation, from the use of formalin to a Lugol's solution in 1973, complicated our analysis, so unless otherwise stated our analyses include only phytoplankton data from 1975 forward, allowing a conservative buffer for the adjustment to the new protocol. There are no obvious effects of the preservation change on diatom data, so we have examined diatom records beginning in 1964 when sampling became consistent across depths and over time.

Single zooplankton samples were collected with a closing plankton net (37.5 cm diameter, 100 µm mesh) from depth layers of 0–10, 10–25, 25–50, 50–100, 100–150, 150–250, and 250–500 m. Samples from the 25–50 and 250–500 m depth layers were excluded from our analyses because sampling frequency was least consistent at these depth layers across the time series. The 100 µm mesh may not sample smaller individuals such as some rotifer species and age classes, so these results should be interpreted cautiously. Zooplankton samples were fixed in formalin throughout the duration of the long-term monitoring program with greatest consistency of temporal and spatial sampling occurring from 1955 forward, the years included in these analyses. Both phytoplankton and zooplankton were identified and counted at the species level, and copepods were enumerated by age class, following a subsampling protocol that was consistent since the inception of ISU's research program [22] in which subsamples are examined until at least 100 individuals of each species or age group are observed. The zooplankton community in the open water is dominated by the herbivorous copepod Epischura baikalensis, comprising approximately 90% of zooplankton biomass [34].

Vertical resolution of the temperature data did not allow determination of the thermocline depth or a quantitative evaluation of stratification. As an alternative metric for the conditions under which stratification likely occurred, we calculated relative thermal resistance to mixing (RTRM; [35]) based on the density (D) of water at 0 m and 50 m temperatures (T), where .

Secchi depth (m) was used to estimate the depth of the photic zone (PZ), the depth to which 0.1% surface light penetrates, using the classic relationship described in Cole [36] that has been used in previous Baikal research [22], [37]. The light extinction coefficient is calculated as and the depth of the photic zone is then calculated as .

To describe the average light environment experienced by each phytoplankton group, we calculated a density-weighted exposure to light (DWL) for each phytoplankton taxon i with abundance n at time t using the percentage of light (l) at each depth: .

The average depth of each taxonomic group was calculated as a density-weighted average depth [38] where x is the abundance of each taxon i at depth z on a given date t. For zooplankton, the depth at the midpoint of the vertical tow was used as zi. Dates on which samples were not collected at all five depth intervals were excluded from analyses.

We examined the trajectory of DWA through time for five zooplankton taxonomic/lifestage groups and seven phytoplankton groups (Table 1). A general linear model with density-weighted depth as the response variable and year and taxa as fixed factors was implemented in R using the lm{stats} function. Zooplankton and phytoplankton were analyzed separately (i.e. two models were used). To test whether trends differed from zero we performed general linear hypothesis tests using the glht{multcomp} function that corrects p-values for multiple comparisons [39]. Durbin-Watson tests conducted on individual least squares model fits for each taxon/lifestage group suggested that there was significant temporal autocorrelation in the residuals for adult copepods, cyanobacteria, cryptophytes, and green algae. Generalized linear models that incorporated an autoregressive parameter were also implemented for these taxa, but the statistical conclusions did not differ from the standard linear model (results not presented).

https://doi.org/10.1371/journal.pone.0088920.t001

To evaluate if and how the spatial overlap of zooplankton and phytoplankton changed, we calculated the difference between the DWA of zooplankton and that of phytoplankton through time. For this analysis the DWA for each of the five zooplankton taxon/life stage groups (Table 1) was compared with the overall DWA for all phytoplankton groups combined. The significance of the trends in the difference between phytoplankton and zooplankton DWA through time was explored using the methods described above for DWA – a general linear model combined with general linear hypothesis tests. Other modes of calculating overlap (e.g. [40]) yielded similar results (unpublished results) but were considered less appropriate because of the difference in sample collection of zooplankton (stratum sampled by closing net) and phytoplankton (discrete depths sampled by bottles).

For plankton groups that exhibited significant changes in DWA we also analyzed trends in abundance by depth interval. To do this we calculated the annual mean abundance of each group in summer (July, August, September) at each depth. Trends were examined with a general linear model combined with general linear hypothesis tests, following methods described above for DWA.

Temperature profiles suggested that summer stratification changed in Lake Baikal, with the temperature gradient between the surface and 50 m becoming stronger through time, resulting in significantly increasing relative thermal resistance to mixing (Fig. 2). The average summer surface temperature during this time period was 10.7° C (with maxima sometimes reaching 19° C), while the average summer temperature at 50 m was 5.5° C. Among the phytoplankton, cyanobacteria numerically dominated (Figs. 3 and 4); however, it is important to recognize that small picoplankton (