Effects of Artificial Thermocline Deepening on Sedimentation Rates

Hydrobiologia (2017) 799:65–81 DOI 10.1007/s10750-017-3202-7

PRIMARY RESEARCH PAPER

Effects of artiﬁcial thermocline deepening on sedimentation rates and microbial processes in the sediment

Andrea Fuchs . Julia Klier . Federica Pinto . Ge´za B. Selmeczy . Bea´ta Szabo´ . Judit Padisa´k . Klaus Ju¨rgens . Peter Casper

Received: 19 December 2016 / Revised: 22 March 2017 / Accepted: 15 April 2017 / Published online: 25 April 2017 Ó Springer International Publishing Switzerland 2017

Abstract Global warming affects the thermal strat- chlorophytes and hypolimnetic diatoms, while the iﬁcation of freshwater lakes, and thus can lead to dominant cryptophytes and cyanobacteria remained thermocline deepening. To follow the direct and unaffected by the treatment. Sedimentation rates, indirect effects of thermocline shifts on processes at ranging between 0.1 and 1.9 g m-2 day-1 per enclo- the sediment–water interface, the thermoclines in six sure, correlated strongly with the selected phytoplank- of twelve lake enclosures were deepened by pumping ton groups, but were not affected by the treatment. warm surface water below the thermocline. Hypolim- Concentrations of elements (Ba, Mn, S, Sr), nutrients ? 3- netic temperatures increased and oxygen concentra- (NH4 ,PO4 , Si), and greenhouse gases (CO2,CH4, tions decreased due to the treatment. Path modeling N2O) in the upper centimeter of the sediment were not suggests a correlation of the treatment to epilimnetic affected by the hypolimnetic changes. Methane and carbon dioxide emission from the enclosures to the atmosphere were negligible. Bacterial and viral abun- Handling editor: Luigi Naselli-Flores dance in sediments were not affected by the treatment. Electronic supplementary material The online version of Results suggest that the benthic ecosystem remains this article (doi:10.1007/s10750-017-3202-7) contains supple- unaffected by slight changes in the pelagial. Further, mentary material, which is available to authorized users.

A. Fuchs Á F. Pinto Á G. B. Selmeczy Á P. Casper (&) G. B. Selmeczy Á B. Szabo´ Á J. Padisa´k Leibniz-Institute for Freshwater Ecology and Inland Department of Limnology, University of Pannonia, Fisheries, 16775 Stechlin, Germany Egyetem u 10, Veszpre´m 8200, Hungary e-mail: [email protected] G. B. Selmeczy A. Fuchs Department of Biological and Environmental Sciences Carl von Ossietzky University Oldenburg, and Technologies, University of Salento, 73100 Lecce, 26129 Oldenburg, Germany Italy

J. Klier Á K. Ju¨rgens B. Szabo´ Á J. Padisa´k Leibniz-Institute for Baltic Sea Research Warnemu¨nde, MTA-PE Limnoecology Research Group, Egyetem u 10, 18119 Rostock, Germany Veszpre´m 8200, Hungary

F. Pinto Centre for Integrative Biology, University of Trento, 38123 Trento, Italy

123 66 Hydrobiologia (2017) 799:65–81 we conclude that short-term warming does not experiments demonstrated greater phytoplankton pro- increase greenhouse gas emissions from temperate, ductivity in the epilimnion during the stratified season deep, oligo-mesotrophic lakes. (Cantin et al., 2011) and an increase of total zooplankton biomass along with a shift to smaller- Keywords Climate change Á Lake metabolism Á bodied zooplankton species (Gauthier et al., 2014; Phytoplankton biomass Á Sediment microbes Á Sastri et al., 2014) due to a deepened thermocline. Greenhouse gases Á Mesocosms Such changes in the plankton community alter sedimentation rates, and thus connect the pelagic with the benthic ecosystem. Organic matter input into the sediment as well as its microbial turnover and release Introduction of nutrients into the water column are vital to supply both environments with nutrient and energy sources. Thermocline formation is the result of heat fluxes Lake sediments are affected indirectly by a deep- within lakes (Schindler, 1997) and stratifies the water ened thermocline, via the sedimentation of organic column of dimictic lakes into nutrient-depleted sur- matter and changing physical and chemical conditions face waters (epilimnion) and nutrient-rich deep waters in the hypolimnion. Published information about the (hypolimnion) in summer. Moreover, the thermocline consequences of a deepened thermocline on microbial forms a density boundary that limits the passage of processes at the sediment–water interface (SWI) is fresh organic matter from the epilimnion to profundal scarce. Thermocline deepening is correlated with sediments and of recycled nutrients from the sediment hypolimnion warming (Sastri et al., 2014), thus to the uptaking organisms in the epilimnion. Thereby, hypolimnetic respiration rates and microbial produc- the thermocline determines the growth and loss of tivity could increase and reduce the oxygen concen- phytoplankton (Sommer et al., 1986; Diehl, 2002) and trations at the SWI. Additionally, enhanced release of zooplankton (Sastri et al., 2014) populations and the nutrients to the water column would be channeled into sedimentation of organic matter. Onset, duration, and the microbial loop (Azam et al., 1983) and may depth of the thermocline are influenced by water stimulate zooplankton productivity (Cantin et al., transparency (Fee et al., 1996; Snucins & Gunn, 2000; 2011; Gauthier et al., 2014). Further, SWI anoxia Houser, 2006), lake morphometry, climate, and salin- could increase P-mobilization and increase anaerobic ity (Sterner, 1990; Archer, 1995; Imboden & Wuest, decomposition processes such as methanogenesis. 1995), and global warming is supposed to intensify the Methane is one of the main products of anaerobic stratification and the suppression of nutrient fluxes carbon mineralization during organic matter decom- from the hypolimnion to the epilimnion by increasing position (Conrad et al., 2010), with 20–60% of the density gradients (Verburg et al., 2003; Behrenfeld sedimenting carbon being converted to CH4 (Rudd & et al., 2006; Schindler, 2009). However, the predicted Hamilton, 1978; Fallon et al., 1980; Wetzel, 2001). In consequences vary among lakes, depending mainly on addition, methanogenesis is positively correlated to the morphometry. Lake thermoclines can either temperature (Zeikus & Winfrey, 1976; Kelly & deepen, due to wind stress and mixing (France, Chynoweth, 1981; Schulz et al., 1997; Glissmann 1997; Schindler, 2009), or due to lower precipitation et al., 2004; Duc et al., 2010). In mesocosm experi- frequencies (IPCC, 2007) resulting in clearer waters ments, Flury et al. (2010) did not find increased (Schindler, 2001), or become shallower, due to higher methane emission rates at higher temperatures, while precipitation frequencies resulting in increased influx Yvon-Durocher et al. (2014) observed that warming of of colored, terrestrial DOC (Snucins & Gunn, 2000). 4°C increased the emission of CH4 from carbon Thermal stratification fundamentally affects the remineralization by 20%. hydrophysical and chemical conditions in the lake, and A number of studies have focused on the effects of thus also plankton community compositions and their global warming on plankton and fish dynamics in sedimentation to the sediment. In a mesocosm exper- lakes, but to our knowledge there has not been any iment, a deeper thermocline resulted in decreased targeted study to assess the impacts on sediments. It is spring phytoplankton biomass and Daphnia density not clear so far if and how sediment microbes are peaks (Berger et al., 2014), while whole lake affected by stratification patterns in the water column, 123 Hydrobiologia (2017) 799:65–81 67 as the effects will mainly be indirect. Thermocline H2 A deepened thermocline and increased phyto- shifts increase microbial respiration and/or extend the plankton biomass will increase sedimentation rates to duration of summer stratification, resulting in an the sediment. increase of anoxic conditions at the SWI, which might H3 An increase of temperature and sedimentation be followed by severe consequences affecting the rates at the SWI will enhance microbial processes, whole lake ecosystem. The objectives of this study thereby increasing the abundance of sediment bacteria were to investigate direct and indirect effects of global as well as the products of organic matter decomposi- warming, in the form of thermocline deepening, on tion, such as NO -,NO -,NH ?, and H S, PO 3-.An lake sediments. We simulated thermocline deepening 3 2 4 2 4 increase of bacterial abundance will in turn increase in the so-called LakeLab (huge enclosures in Lake viral abundance. Stechlin; Bauchrowitz, 2012) and measured the effects on sedimentation rates and microbial responses in H4 Increasing temperatures and decreasing oxygen sediments two months after the treatment. The concentrations above the sediment will increase the following chain of hypotheses was tested (see Fig. 1): release of porewater elements. H1 A deepened thermocline leads to an increase of H5 An increase of temperature, sedimentation rates, hypolimnetic temperature, and thus of microbial and a decrease of oxygen concentrations will increase activity, and to a decrease of the oxygen concentra- the concentrations of the greenhouse gases CO2 and tions above the sediment.

Fig. 1 Hypotheses interaction scheme 123 68 Hydrobiologia (2017) 799:65–81

- CH4 at the sediment surface. Simultaneously, NO3 STAMOID 4800, 12 m-sediment: 1 mm thickness, 2- and SO4 can inhibit the production of CH4. STAMOID 1250) reaching down until 1 m into the sediment (Fig. 2). Enclosure pairs were E1 with E13, To test these hypotheses we applied path analysis. E3 with E15, E4 with E16, E5 with E17, E7 with E20, A path model takes into account the combination of and E8 with E19, based on similar depths and direct and indirect effects that would not be suffi- ecological conditions. Pairs of enclosures were chosen ciently covered by multivariate statistics or linear randomly for treatment or control. Thus, E1, E4, E8, models. In this path model, different blocks of E15, E17, and E20 were chosen for treatment variables, each of them representing one step within (Table 1). Prior to the manipulation, water was the carbon cycle, were defined by latent variables that exchanged between the enclosures and the lake by significantly influenced the linear relations among the pumping for two weeks, and between paired enclo- blocks (Sanchez, 2013), independent of their abun- sures for another two weeks, to equalize the starting dance or significance for the ecosystem in general. conditions among enclosures. In each enclosure, underwater pumps (SUPS 4-12- 5, SPECK Pumpen Verkaufsgesellschaft GmbH, Materials and methods Neunkirchen am Sand, Germany) transported 5.5 m3 h-1 of warm epilimnetic water via aluminum Enclosures facility and treatment release rings into deeper water layers (Fig. 2). The release rings (ring diameter: 7 m, tube diameter: The study was performed in 12 enclosures (diameter: 44.3 mm), containing holes (diameter: 3 mm) every 9 m, depth ca. 20 m, Table 1) which are part of a 50 cm, were placed into the depth of the lake 24-enclosures LakeLab facility (Kwok, 2013; Giling thermocline in control enclosures, and 2 m lower in et al., 2016; Fig. 2) in Lake Stechlin (Brandenburg, the treated enclosures; thus all enclosures were equally Germany). Lake Stechlin is an oligo-mesotrophic, affected by pumping activities. As the thermocline in temperate lake with a mean depth of 23.3 m, maxi- the lake moved due to temperature changes, the mum depth of 69.5 m, and a surface area of 4.23 km2 release rings in the enclosures were moved 1 m down (Koschel & Adams, 2003). The lakes’ shoreline is each on 24 July, 25 July, and 04 September 2013. covered to *96% by trees (Casper, 1985). Parameters During the first month, only a weak pre-treatment (8 h influencing the thermal stratification of the lake during of pumping per day) was applied, to enable organisms the last 55 years in the form of artificial or global to adapt to the pumping gently, and the pumping temperature increases were reported by Kirillin et al. rhythm was then increased (12 h of pumping per day) (2013). The enclosures consist of floating aluminum in the second month to force the thermocline into a rings with curtains (0–11 m depth: 5 mm thickness,

Table 1 List of enclosures Treatment Enclosure Depth (m) Release ring depth (m) Volume (E) Volume (H) used in this study, showing treatments, water depths, Control E3 20.5 5–8 318/509 986/795 and the depth of the release Control E13 20.8 5–8 318/509 1,005/814 rings, with (E) = epilimnion; (H) = Control E5 19.2 5–8 318/509 903/713 hypolimnion Control E16 19.9 5–8 318/509 948/757 Control E7 17.9 5–8 318/509 821/630 Control E19 17.7 5–8 318/509 808/617 Treated E15 20.0 7–10 445/636 827/636 Treated E1 19.9 7–10 445/636 821/630 Treated E17 18.9 7–10 445/636 757/566 Each treated enclosure was paired with a control Treated E4 19.5 7–10 445/636 795/604 enclosure (E3–E15, E13– Treated E20 17.1 7–10 445/636 643/452 E1, E5–E17, E16–E4, E7– Treated E8 17.2 7–10 445/636 649/458 E20, E19–E8) 123 Hydrobiologia (2017) 799:65–81 69

Fig. 2 LakeLab facility in Lake Stechlin with control (gray) and treated (red) enclosures and a scheme of a single enclosure including sampling devices (photo: Martin Oczipka, IGB/HTW Dresden) deepened position. Pumping was only applied during horizon of each enclosure. Values at the SWI were daytime to allow natural cooling over night. measured in the last meter above the sediment.

Water column Phytoplankton biomass

Physical and chemical parameters Integrated water samples (volume: 5 l each, IWS 2, Hydro-Bios, Kiel, Germany) were taken from the The water column was sampled every 2 weeks, on 25 epilimnion and hypolimnion and preserved in Lugol’s June, 10 July, 23 July, 06 August, 20 August, and 11 solution. At minimum 400 settling units (cells, September 2013. Temperatures and oxygen concen- ﬁlaments, or colonies) were counted in each sample trations in the water column were measured in half- using an inverted microscope (Zeiss Axiovert 35, meter increments (from the surface to a minimum of Oberkochen, Germany). Phytoplankton numbers were 0.5 m above the sediment) and at 1 h intervals by determined with the methodology by Lund et al. automated multiprobes (YSI 6600, YSI Inc., Yellow (1958) and biomass was calculated volumetrically Springs Instrument, OH, USA) installed in each assisted by the OptiCount software (SequentiX, Klein enclosure. Average temperatures and oxygen concen- Raden, Germany). trations in the epilimnion and hypolimnion were calculated from the average temperature and oxygen Sedimentation rates concentration per volume of each enclosure on each sampling day, considering the varying volumes of Sedimentation traps (UwitecÒ, Mondsee, Austria) each depth horizon over time due to thermocline were deployed in the hypolimnion 1 m above ground deepening. Due to the depth differences among for three months, tied with a rope centrally to the enclosures (Table 1), we calculated the averages in bridge leading across the enclosures, and a weight at the hypolimnion only from values until 16 m depth in the bottom to keep the trap in position. Traps consisted each of them. In addition, we calculated the total of two tubes (diameter: 9 cm, height: 100 cm) pushed amount of oxygen and temperatures within each into two wide-neck PVC collection bottles (2 l) ﬁlled

123 70 Hydrobiologia (2017) 799:65–81 prior to deployment with hypolimnetic water. Just one Porewater chemistry and sediment characteristics collection bottle per enclosure was used to determine sedimentation rates. As a fixative, 100 ml formalde- Rhizons (length: 5 cm, diameter: 2.5 mm, pore size: hyde (final concentration 1.85%) was added to each 0.1 lm, Rhizosphere, Wageningen, Netherlands) bottle prior to deployment to inhibit further decom- were used to extract a maximum of 10 ml porewater position and zooplankton feeding (Fuchs et al., per depth, while remaining sediments were dried over 2016a). Traps were deployed from 27 June to 19 night at 105°C and weighed before and after, to September and were exchanged in monthly intervals. analyze the dry weight content. Of each porewater Samples were stored at 4°C until processing. sample, 2 ml was distributed in Eppendorf tubes for Trapped material was allowed to settle and the total the analyses of nutrients (phosphate, nitrate, nitrite, 2- volume was decanted to 500 ml. The sample was then ammonium), sulfide, silicate (SiO4 ), and dissolved homogenized and sieved through 280 lmgauzeto metals, respectively. Tubes for nutrients and silicate remove zooplankton that tended to actively swim into were washed with Milli-Q water in advance, and traps instead of sinking in. Sieved material was filtrated samples were measured using an autoanalyzer (CFA over big (diameter: 45 mm, 50 ml of homogenized QuAAtro, SEAL, Norderstedt, Germany). Tubes for sample filtrated) and small (diameter: 25 mm, 10–15 ml inductively coupled plasma optical emission spec- of homogenized sample filtrated) glass fiber filters trometry (ICP-OES) were treated with 65% v/v (GFF) in three replicates, each. GFF were combusted at suprapure HNO3 prior to sampling, and porewater 500°C for 4 h before filtration and dried at 105°Cover samples were acidified with 1% v/v suprapure HNO3. night after filtration. Big GFF were used for dry weight Samples were stored at 4°C until the measurement by analyses, while two of the three small GFF were ICP-OES (iCAP 6300 Duo, Thermo Fisher Scientific, analyzed for carbon and nitrogen (Vario El analyzer, Waltham, MA, USA). H2S was measured spectropho- Elementar Analysensysteme, Hanau). The third small tometrically (SPEKOL 1100, Analytik Jena, Jena, GFFs were used for analyses of total phosphorus (TP) Germany) according to the method of Cline (1969). and total nitrogen (TN) after extraction with potassium peroxodisulfate (Ebina et al., 1983) and subsequent Bacterial and viral abundance photometric analysis using flow injection analysis. From each sediment core, 0.5 ml of wet sediment was Sediments mixed with 0.5 ml of sodium pyrophosphate (final concentration 5 mM) and 4 ml of virus-free sterile

Sediment cores H2O. After an incubation of 10 min at room temperature, samples were gently sonicated in a water bath Two sediment cores (diameter: 6 cm) per enclosure sonicator (Transsonic Digital, Elma, Singen, Ger- were collected before the start of the experiment many) three times for 20 s interrupting for 10 s every (26–28 June) and after two months of thermocline time during which the sample was shaken manually. deepening (28–30 August), using a gravity corer Then a centrifugation step at 8009g at 15°C was (UwitecÒ, Mondsee, Austria). Cores were taken in the applied for 1 min, and 1 ml of the supernatant was center of each enclosure, carefully choosing a new withdrawn in duplicate and mixed with virus-free spot for every core. Sediments were pushed to the top distilled water in order to achieve the desired dilution. of the core and sliced into the layers 0–1, 1–2, 2–4, The extracted sample (1 ml) was first treated with 4–6, 6–8, 8–10, 10–12, 12–14, 14–16, 16–18, and DNase (final concentration 1 U ml-1) and then fil- 18–20 cm. We retrieved only 0–8 cm of sediment tered through 0.02 lm ultra-fine pore size filter from cores of E8, E19, and E20 in June, and of E19 and (Anodisc, Whatman, Little Chalfont, UK). As the E20 in August due to low water content in deeper filter dried, a drop (10 ll) of SYBR Green I (2.59 final layers. While sediments of one core were analyzed for concentration diluted in Milli-Q water, Invitrogen, sediment characteristics and porewater chemistry in Waltham, USA) was laid on the sample and kept in the each layer, sediments of the second core were dark for 15 min. Filters were then washed with virus- analyzed for greenhouse gas concentrations and the free lake water and put on a slide. Slides were stored at abundance of bacteria and viruses at the SWI. -20°C until counted using an epifluorescence 123 Hydrobiologia (2017) 799:65–81 71 microscope (magnification 1.0009, blue light excita- resulted in changing concentrations of many param- tion filter cube L5, Leica DM5500B, Leitz, Wetzlar, eters over time. Thus, we calculated for each param- Germany). 10 to 20 fields were viewed until a total of eter in each enclosure the delta values, as T1 - T0. 200–300 virus-like particles (VLP) or bacteria were Boxplots and statistical analyses were performed counted, respectively. using these delta values, excluding values from enclosures 5 (control) and 8 (treated) due to missing data. Greenhouse gases The direct and indirect effects of the treatment on sedimentation rates and sediments were tested by Concentrations of carbon dioxide, methane, and Partial Least Squares Path Modeling using the pack- nitrous oxide were measured from 2 ml of wet age ‘‘plspm’’ (Sanchez, 2013). PLS path modeling has sediment per layer added to 4 ml distilled water in rarely been used for ecological approaches (Fu et al., 10 ml glass vials. Vials were sealed gas-tightly by 2015; Wang et al., 2016); however, it is advantageous butyl caps. HgCl (final concentration 200 mg l-1) 2 to other models when analyzing complex networks of was added for fixation to avoid microbial production or variables. In a path model, the correlation between consumption of gases. Vials were stored upside down at latent variables is analyzed, taking into account direct 4°C in the dark until measurement. Gas measurements as well as indirect effects. Each latent variable is were performed using a gas chromatograph (GC-14B, defined by various measured manifest variables. In Shimadzu Europa GmbH, Duisburg, Germany) this study, the inner model of latent variables was equipped with TCD (CO analysis), FID (CH analysis), 2 4 constructed based on our hypotheses. We defined the and ECD (N O analysis) and an autosampler AOC 5000 2 treatment by the changes of temperatures and oxygen (CTC Analytics AG, Zwingen, Switzerland). CO and 2 concentrations in the epilimnion and hypolimnion, CH concentrations were calculated per dry weight 4 assuming that water column parameters (such as (DW) sediment based on Henry’s Law solubility con- plankton biomass and sedimentation rates) would be stants (k : 0.00117192 mol kg-1 bar-1, k : H;CH4 H;CO2 affected only by the epilimnetic treatment, while the 0.034 mol kg-1 bar-1) retrieved from NIST. sediment parameters (such as microbial abundances, Greenhouse gas fluxes at the air–water boundary porewater elements, and greenhouse gas concentra- were measured by an aluminum flux chamber (1 m2 tions) would be affected by the hypolimnetic treatment area, 240 l volume), shaped as a truncated pyramid and sedimentation. The PLS path model was consid- and floating on an inflated rubber rectangle, in the ered acceptable when the goodness-of-fit (GoF) esti- center of each enclosure. Each flux chamber possessed mate reached[0.7. a connection valve with a gas-tight septum. Chambers Additionally, all variables were tested blockwise on were placed on the enclosures at the end of the significant differences between control and treated experiment, one day after the pumping stopped, and enclosures using multivariate statistics (vegan-pack- gas samples were taken immediately (t0), and after 6 h age, Oksanen et al., 2016). For this, all measured (t1), 24 h (t2), and 96 h (t3). 20 ml vials were capped parameters were included, independent of their use in gas-tightly and pre-evacuated by a vacuum pump for the PLS path model. 90 s prior to sampling and connected to the flux chamber valves via cannulas for 90 s. Gas samples were measured immediately after sampling. Results

Statistics The path model reached a goodness-of-ﬁt estimate of GoF = 0.701. The inner model summary shows the All statistical analyses and graphs were performed relevance of the latent variables to the path model using RStudio, Version 3.1.2 (R Studio, Inc. 2013, (Table 2), with the elements having the highest R2 Boston, MA, USA). For several parameters, the (0.837) and nutrients having the lowest R2 (0.145). variability between replicate enclosures was high The hypolimnion treatment, porewater elements, and (raw data given in Supplement Table 1). In addition, greenhouse gases were highly important for the inner the natural development of summer stratiﬁcation model, reaching R2 values[0.7, while phytoplankton 123 72 Hydrobiologia (2017) 799:65–81

Table 2 Summary of the Block Type R2 Block communality Mean redundancy inner path model, with Treat = treatment, Phyto = Treat (E) Exogenous 0.00 0.75 0.00 phytoplankton, Traps = Treat (H) Endogenous 0.72 0.70 0.50 sedimentation traps, (E) = epilimnion, and (H) = Phyto Endogenous 0.49 0.71 0.35 hypolimnion Traps Endogenous 0.69 0.95 0.66 R2 values above 0.7 (in Nutrients Endogenous 0.14 0.96 0.14 bold) indicate signiﬁcant Elements Endogenous 0.84 0.84 0.70 response variable blocks Gases Endogenous 0.70 0.77 0.54 within the model and nutrients had a R2 \ 0.5, showing no signiﬁcant decreasing irradiation cooled the surface water and relevance to the path analysis. facilitated a homogenously mixed epilimnion.

Treatment Temperatures and oxygen concentrations (H1)

During the ﬁrst month of pre-treatment, the thermo- At the SWI, mean temperatures were 5.5°C in the cline was not deepened in order to allow organisms to control and treated enclosures, both in June (control: establish a community under the new treatment-like ±0.2°C SD, treated: ±0.3°C SD) and August (control: conditions. From 26 July onwards pumping intensity ±0.1°C SD, treated: ±0.2°C SD), while mean oxygen was increased and the thermocline was deepened by concentrations decreased from 10.2 ± 1.4 mg l-1 2 m (Fig. 3). However, we did not achieve a complete (control) and 10.5 ± 1.1 mg l-1 (treated) in June to mixing of the epilimnion between 25 July and 14 7.4 ± 1.6 mg l-1 (control) and 6.1 ± 1.1 mg l-1 August. During this time, a 3–5 m thick density (treated) in August. However, considering the delta gradient formed above the thermocline with 21–22°C values per enclosure in the hypolimnion, we observed in the control and 15–18°C in the treated enclosures an increase of the temperature (Fig. 3) as well as a (Fig. S1—Supplementary Material). After this period, decrease of oxygen concentrations (Fig. 4). Path

Fig. 3 Temperature proﬁles (mean ± SE) in the water column during the experiment in the control (gray) and treated (red) enclosures 123 Hydrobiologia (2017) 799:65–81 73

Fig. 4 Delta values of the parameters at the WSI that were used and August in control (white) and treated (gray) enclosures. in the path analysis, showing an increase (positive values)or Boxes include the median (solid lines), minimum and maximum decrease (negative values) of the concentrations between June values (whiskers), and outliers (circles) analysis confirmed a positive correlation between Table 3 Results of PERMANOVA analysis (method = Eu- clidean), with Treat = treatment, Phyto = phytoplankton, Traps epilimnetic and hypolimnetic treatment effects. Mul- = sedimentation traps, (E) = epilimnion, (H) = hypolimnion, tivariate statistics (Table 3) confirmed that the treat- and (SWI) = sediment-water interface ment affected average temperatures and oxygen Df SS FR2 P value concentration in both the epilimnion (PERMANOVA, R2 = 0.405, P = 0.018) and the hypolimnion (PER- Treat (E) 1 7.292 5.448 0.405 0.020 MANOVA, R2 = 0.552, P = 0.01), as well as total Treat (H) 1 9.936 9.857 0.552 0.013 temperatures and oxygen concentrations in the epil- Treat (SWI) 1 4.693 2.821 0.261 0.052 2 imnion (PERMANOVA, R = 0.634, P = 0.011) and Phyto (E) 1 5.577 0.777 0.088 0.726 2 the hypolimnion (PERMANOVA, R = 0.319, Phyto (H) 1 4.593 0.629 0.073 0.775 P = 0.027), but not at the SWI (PERMANOVA, Traps 1 3.039 1.015 0.112 0.382 2 R = 0.261, P = 0.079). The permanent profiling did Nutrients 1 6 1.000 0.111 0.424 not reveal anoxic conditions in any of the enclosures. Elements 1 11.331 0.938 0.105 0.448 Gases 1 2.108 1.061 0.117 0.427 Phytoplankton biomass and sedimentation rates (H2) Df degrees of freedom, SS sums of squares, F F-model, P values based on 999 permutations Total phytoplankton biomass was not affected by the Statistical significance at P \ 0.05 is indicated in bold treatment and showed contrasting developments in epilimnion and hypolimnion: between July and dominance of cyanobacteria (main species: Plank- August, the mean overall phytoplankton biomass in tothrix rubescens (De Candolle ex Gomont) Anagnos- the epilimnion remained similar (525–564 lgl-1), but tidis & Koma´rek) in August, forming a deep increased in the hypolimnion (517–668 lgl-1). The chlorophyll maximum in the hypolimnion (see data phytoplankton community was initially dominated by in Selmeczy et al., 2016). These changes in phyto- epilimnetic cryptophytes and shifted to an overall plankton community composition were not reflected in 123 74 Hydrobiologia (2017) 799:65–81

Fig. 5 Path analysis plot indicating positive (green) and phytoplankton biomass (phyto), sedimentation rates (traps), negative (red) correlations of the inner model. The inner model sediment nutrients (nutrients), porewater chemistry (elements), consists of seven latent variables representing epilimnetic and greenhouse gases (gases) treatment (treatE), hypolimnetic treatment (treatH), sedimentation rates, which decreased from an average The parameter ‘‘traps’’ was defined by dry weight of 686.6 ± 197.8 mg m-2 day-1 in July to an average of sedimented matter and its carbon content. Path 495.7 ± 498.1 mg m-2 day-1 in August. In August, analysis shows that the sedimentation rate was sedimentation rates showed very high variation strongly influenced by the selected phytoplankton between enclosures, with dry weights ranging from groups and not affected by the treatment. However, 129.2 mg m-2 day-1 in E17 (treated) to despite a small increase of the diatom biomass, 1,890.5 mg m-2 day-1 in E20 (treated). However, sedimentation rates and their carbon content decreased the change in the phytoplankton community compo- over time. Phytoplankton biomass and sedimentation sition was reflected by the composition of the sedi- rates were not significantly different between control mented material. The average C:N ratio of the and treated enclosures (PERMANOVA, Table 3). sedimented material decreased from 10.6 ± 0.9 in July to 8.2 ± 1.8 in August, due to increases in the Abundances of bacteria and viruses (H3) percentage of nitrogen. While E20 showed the highest sedimentation rate and total phytoplankton biomass in Mean viral abundance doubled between June July and moderately high values in August, the C:N (2.4 9 1010 ± 1.4 9 1010 VLP g-1) and August ratio of the sedimented material reached the lowest rate (5.5 9 1010 ± 2.5 9 1010 VLP g-1) and was one in July and the highest in August. Although cyanobac- order of magnitude higher than mean bacterial abun- teria and cryptophytes dominated the phytoplankton dance (3.5 9 109 ± 3.2 9 109 cells g-1 in June, community, they did not show any relevance for the 2.4 9 109 ± 2.3 9 109 cells g-1 in August) at the path model and were thus removed from it. This is SWI. Neither bacterial nor viral abundance were justified by their known marginal contribution to affected by thermocline deepening, and additionally sedimentation. Instead, epilimnetic chlorophytes and did not contribute with any impact to the path model. hypolimnetic diatoms (Fig. 4) defined the parameter - - ? 3- ‘‘phyto,’’ despite their little contribution to the phyto- Nutrients (NO3 ,NO2 ,NH4 ,PO4 ), H2S, plankton biomass. Path analysis indicates a strong and Si (H3) negative correlation between the treatment parameters in the epilimnion and a weak positive correlation of the Nutrients-, H2S-, and dissolved Si concentrations in hypolimnetic treatment parameters on these phyto- the porewater showed no significant difference plankton groups (Fig. 5). between control and treated enclosures (Table 3).

123 Hydrobiologia (2017) 799:65–81 75

Nitrate, nitrite, and sulfide showed no relevance to the not differ between control and treated enclosures path model variable block ‘‘nutrients.’’ In contrast, (Table 3). In contrast to CH4, mean CO2 concentra- ? -1 mean concentrations of NH4 (60.4 ± 40.7 lmol l tions decreased over time, but less in the treated than in -1 3- in June, 65.4 ± 46.3 lmol l in August), PO4 the control enclosures (Fig. 4). Path analysis (Fig. 5) (7.1 ± 5.0 lmol l-1 in June, 5.0 ± 4.9 lmol l-1 in revealed a strong positive correlation of nutrients and a August), and also Si (95.0 ± 52.0 lmol l-1 in June, strong negative correlation of sedimentation rates to 97.0 ± 65.2 lmol l-1 in August) contributed to the the greenhouse gases, while the porewater chemistry path model and were strongly negatively correlated to as well as hypolimnetic temperatures and oxygen the hypolimnion treatment parameters (Fig. 5). Mean concentration showed no influence. values (Fig. 4) of all three variables were higher in the Greenhouse gas emissions at the air–water bound- ? treated than in the control enclosures (DNH4 : ary, as measured by flux chambers, were below the 91.9 ± 47.5 lmol l-1 in control, 98.1 ± 61.6 detection limit during the first 24 h. After 96 h, -1 3- -1 lmol l in treated; DPO4 : -3.7 ± 7.3 lmol l methane fluxes were detected in seven enclosures in in control, -0.5 ± 5.7 lmol l-1 in treated, DSi: amounts below 50 lmol m-2 h-1, only exceeded by -31.7 ± 65.1 lmol l-1 in control, 35.7 ± 62.5 emissions in E16 (157.2 lmol m-2 h-1) and E15 lmol l-1 in treated). Amount and carbon content of (444.9 lmol m-2 h-1), but did not differ between sedimented materials were weakly correlated to the control and treated enclosures. Fluxes in E1, E8, and nutrient parameters. E13 had slopes with an R2 \ 0.6 and were not considered for further calculations. Fluxes of carbon Porewater elements: Ba, Ca, K, Li, Mg, Mn, Na, S, dioxide and nitrous oxide could not be detected at any and Sr (H4) time.

Porewater element concentrations did not differ between control and treated enclosures (Table 3). We Discussion tested barium, calcium, potassium, lithium, magnesium, manganese, sodium, sulfur, and strontium to fit into the Artificial thermocline deepening in lake enclosures led path analysis, but only barium, manganese, sulfur, and to increased temperatures and decreased oxygen strontium correlated with the model. Furthermore, path concentrations in the hypolimnion, but did not result analysis (Fig. 5) showed a strong correlation between in any direct or indirect effect on sedimentation rates porewater nutrients and elements, but only weak and or on the microbial abundance, greenhouse gas negative influences of the treatment and the sedimented concentrations, and porewater chemistry at the SWI. material on element concentrations. Concentrations The formation and deepening of the thermocline in a ranged between 26.1 ± 5.7 lgl-1 (June) and 24.0 ± lake during summer stratification is supposed to have 3.6 lgl-1 (August) for barium, 122.1 ± 52.1 lgl-1 dramatic effects on all physical, chemical, and (June) and 120.7 ± 48.7 lgl-1 (August) for man- biological processes; therefore, the lack of responses ganese, 9.1 ± 1.4 mg l-1 (June) and 10.1 ± to such a treatment emphasizes the complexity of the 1.5 mg l-1 (August) for sulfur, and 76.4 ± 1.4 lgl-1 processes as well as the autonomy of the sediment (June) and 77.3 ± 3.8 lgl-1 (August) for strontium community, as already indicated by distinct microbial (delta values given in Fig. 4). communities in waters and sediments in various environments (Lozupone & Knight, 2007; Cole

Greenhouse gases: CO2,CH4, and N2O (H5) et al., 2012). In contrast to extreme events like storms and droughts, the impact of thermocline deepening Average greenhouse gas concentrations at the SWI might be visible only in the long term, as slight were between 19.2 ± 5.6 lmol g-1 DW (June) and changes in the SWI temperature and oxygen regime at 12.8 ± 6.2 lmol g-1 DW (August) for carbon diox- the end of summer might be carried over to the ide, and between 5.5 ± 4.0 lmol g-1 DW (June) and following winter and spring season and add up to more 7.4 ± 9.8 lmol g-1 DW (August) for methane. Con- severe changes after several years. However, our centrations of nitrous oxide were below detection experiments shed more light on short-term effects on limit. Statistically, CO2 and CH4 concentrations did the chemical and biological interactions at the SWI. 123 76 Hydrobiologia (2017) 799:65–81

Hypothesis 1 A deepened thermocline leads to an where the thermocline was deepened, as reported by increase in hypolimnetic temperature and thus in Cantin et al. (2011). In our study, however, at the microbial activity, leading to a decrease of the oxygen beginning of the experiment the phytoplankton com- concentrations above the sediment. munity was dominated by cryptophytes, a motile group that dominates in Lake Stechlin during the non- The statistical approaches used (path analysis stratified period and prevails in the upper hypolimnion model and multivariate statistics) confirmed that a during summer (Padisa´k et al., 1998). Thus, their deepening of the thermocline in *20 m deep lake initial dominance in the enclosures’ epilimnia was enclosures results in an increase of temperature and a probably related to the previous homogenized mixing decrease of oxygen concentrations in the hypolimnion. (Phillips & Fawley, 2002; Teubner et al., 2003; Ongun This was observed for average as well as total values in Sevindik et al., 2015), followed by strong biomass relation to the water volume, although not at the SWI losses in the epilimnion when the mixing ended and directly. The differences in temperature in the the thermal stratification was established, but without hypolimnion were mainly found in the water layers increasing sedimentation rates. Presumably, the algae below the thermocline, and decreased in the deeper were consumed by grazers or microbially decomposed layers. In contrast, differences in oxygen concentra- within the water column, although it may be possible tions increased with depth. However, oxygen concen- that they do not contribute enough weight to signif- trations at the SWI varied by up to 4.3 mg l-1 among icantly alter sedimentation rates. Simultaneously, replicate enclosures, mainly due to differences in phytoplankton biomass in the hypolimnion, domi- depth. The results imply that changes in the stratifi- nated by the cyanobacterium Planktothrix rubescens, cation depth can affect pelagic organisms, but have no increased during the experimental period, and resulted immediate direct effect on the biogeochemistry and in an increase of the total phytoplankton biomass microbial abundance at the sediment surface. Never- between June and August, despite the steady total theless, the trend towards higher temperatures and biomass in the epilimnion. P. rubescens becomes lower oxygen concentrations in the hypolimnion abundant in Lake Stechlin after very cold winters indicates that more severe consequences could occur (Padisa´k et al., 2010), and was involved in the in shallow or eutrophic lakes in the long term. formation of a deep chlorophyll maximum (DCM) in Moreover, the results suggest that sediments at the Lake Stechlin in the same year (Selmeczy et al., 2016). depth level below the thermocline will experience Another cyanobacterium species, Dolichospermum temperature increases, but probably no changes flos-aquae (Bre´bisson ex Bornet & Flahault) Wacklin, regarding the oxygen regime. However, due to Hoffmann & Koma´rek, increased biomass 2.5 weeks seiches, sediments at the thermocline depth level after artificial strong mixing in a different LakeLab experience changing epilimnetic and hypolimnetic experiment (Giling et al., 2016). As reported previ- conditions regularly (Kirillin et al., 2009), without ously, such events can lead to massive calcite precip- dramatic effects on the biogeochemistry or microbial itation (Kasprzak et al., 2017) followed by massive activity (Frindte et al., 2013). sedimentation (Fuchs et al., 2016a), but such conse- Hypothesis 2 A deepened thermocline and quences were not observed in our study since increased phytoplankton biomass will increase sedi- thermocline deepening showed no effect on the mass mentation rates of the sediment. of sedimenting particles reaching the sediment. In enclosures, sedimentation rates are biased by the lack Phytoplankton biomass and sedimentation rates of currents and by an altered impact of wind force, but were not directly affected by thermocline deepening. sedimentation rates measured close to the LakeLab The path model indicated a strong negative correlation facility in the previous year (Fuchs et al., 2016a) of the epilimnetic treatment with the biomass of exhibited similar values in July (568.3 mg m-2 epilimnetic chlorophytes and hypolimnetic diatoms, day-1) and 1.8 times higher rates in August as well as a strong positive correlation of these (942.8 mg m-2 day-1). The increasing percentage of phytoplankton groups with sedimentation rates. nitrogen in the sedimented material between July and Chlorophytes also contributed to increased phyto- August could occur as a result of N -fixation by plankton production in the epilimnion of lake basins 2 cyanobacteria (Dokulil & Teubner, 2000) or fecal 123 Hydrobiologia (2017) 799:65–81 77 pellets deposition, as in parallel the phytoplankton into nitrite, nitrous oxide, or molecular nitrogen, and community shifted to a dominance of cyanobacteria in sulfide. As mentioned above, the treatment had no

August, but N2-fixing species did not reach prominent direct effect on the biogeochemistry at the SWI, thus biomasses within this group. Neither the nitrogen also the concentrations of nitrate, nitrite, and sulfide content of the sedimented material nor the cyanobac- remained unaffected. Interestingly, the shift of phyto- teria biomass were correlated to the treatment or plankton composition and sedimenting C:N ratio also showed any relevance to the path model. did not correlate to the nitrate concentrations at the SWI. Instead, ammonium, phosphate, and silicate Hypothesis 3 An increase of temperature and sed- contributed to the ‘‘nutrients’’ part of the path model imentation rates at the SWI will enhance microbial and were strongly correlated to the ion and greenhouse processes, thereby increasing the abundance of bacte- gas concentrations in the sediment. Ammonium and ria as well as the products of organic matter decom- phosphate are products of microbially mediated position, such as NO -,NO -,NH ?,H S, or PO 3-. 3 2 4 2 4 organic matter decomposition and have been shown An increase of bacterial abundance will in turn to increase under anoxic condition (Frindte et al., increase viral abundance. 2013), indicating an increase of microbial activity and Bacterial abundance was affected neither by the a shift to anaerobic processes towards the end of treatment nor by the temporally decreasing sedimen- summer. Additionally, phosphate can be released from tation rate and its changing C:N composition. Simi- the sediment by altering redox conditions (Hupfer & larly, Goedkoop et al. (1997) demonstrated that a Lewandowski, 2008) and compete with Si, which is difference in organic matter supply, in the form of derived from sedimented diatoms, for sorption sites diatom addition, to incubated Lake Erken sediments (Gonsiorczyk et al., 2003). Nevertheless, the path did not affect bacterial abundance, although it model showed no significant correlation of these increased bacterial production and growth rates. nutrients to hypolimnetic temperatures and oxygen Despite previous indications that all these parameters concentrations. (bacterial production, abundance, and growth rates) Hypothesis 4 Increasing temperatures and decreas- are more strongly affected by temperature than ing oxygen concentrations above the sediment will substrates (Bostro¨m et al., 1989; Shiah & Ducklow, increase the release of porewater elements. 1994), bacterial abundances correlated neither to the treatment nor to the sedimentation rate in the path Porewater elements are part of the sediment model. geochemistry, being dissolved into the porewater by While the deepened thermocline did not result in an rock weathering or introduced by rainfall or anthro- increase of viral abundance at the SWI, we found an pogenic activities. Temperature increase and oxygen increase of viral abundance over time. It has been decrease can affect the binding complexes of these shown that viral abundance is highly correlated to elements, due to shifts in pH values and shifts in the oxygen (Ricciardi-Rigault et al., 2000), organic matter availability of electron acceptors and carbonates as (Maranger & Bird, 1996), and bacterial abundance binding partners. We did not observe a significant (Proctor et al., 1993). In contrast, we observed an correlation of the treatment or the sedimented material increase of viral abundance towards the end of on the concentrations of manganese, strontium, sulfur, summer stratification, where oxygen concentrations and barium; however, the nutrients phosphate and and sedimentation rates were the lowest, while the ammonium, and silicate, were very highly correlated bacterial abundance remained unaffected. As Pinto to these elements, indicating simultaneous et al. (2013) discussed, it is still not clear which are the mobilization. drivers of viral abundance variability. Local condi- Hypothesis 5 An increase of temperature, sedimen- tions at the end of the summer (low oxygen concen- tation rates, and a decrease of oxygen concentrations tration) might have induced the release of lysogenic will increase the concentrations of the greenhouse viruses, and thus resulted in a higher viral abundance. gases CO and CH at the sediment surface. Simul- When oxygen is depleted in the upper few 2 4 taneously, NO - and SO 2- can outcompete the centimeters of the sediment, nitrate and sulfate are 3 4 production of CH the main electron acceptors transformed by microbes 4. 123 78 Hydrobiologia (2017) 799:65–81

Anoxia is one of the most severe environmental Stechlin have shown high methane production rates impacts, restraining the life of benthic eukaryotes and in deeper sediment layers ([20 cm, Casper, 1992; larval stages and shifting the composition and activ- Casper et al., 2005; Fuchs et al., 2016b) as well as ities of microbial sediment communities towards methane production in the water column (Grossart anaerobic processes. The final step of organic matter et al., 2011; Tang et al., 2014), but fluxes of methane decomposition is the production of the greenhouse gas from the enclosures to the atmosphere were very low. methane. Methanogenesis occurs anaerobically, Thermoclines act as a physical barrier for the passage nonetheless, methanogens are present in all layers of gases, thereby additionally enabling methane down to 25 cm depth in Lake Stechlin sediments oxidation at the barrier. Up to 99% of the methane (Fuchs et al., 2016b) and have been shown to produce released from sediments can be oxidized before methane even at the sediment surface when exposed to reaching the atmosphere, with highest methane oxi- favorable conditions (Dos Santos Furtado & Casper, dation rates at the oxic/anoxic interface (Bastviken 2000; Conrad et al., 2007). Previous incubation et al., 2002). experiments with sediments of Lake Stechlin have shown that elevated temperatures decreased methane concentrations (Fuchs et al., 2016b), but shifts from Conclusions oxic to anoxic conditions weakly enhanced the functional gene expression of methanogens and The impact of global warming-induced thermocline methanotrophs (Frindte et al., 2015). Studies in other deepening on a cascade of variables affecting the lake and river sediments have also shown that methane sediment biogeochemistry was analyzed in a deep production as well as methane oxidation are both stratified lake. We found that a 2 m deepened elevated at higher temperatures and can outbalance thermocline in lake enclosures affected neither sedi- each other (Kelly & Chynoweth, 1981; Shelley et al., mentation rates, nor microbial and viral abundances in 2015). In this study, we did not observe any impact of the sediment, nor greenhouse gas concentrations at the the treatment on methane concentrations. However, SWI or emissions to the atmosphere. Thus, we reject the treatment effects were not as strong as in the our Hypotheses 2–5. Nevertheless, our results con- previous sediment incubation experiments by Frindte firmed Hypothesis 1, indicating a high potential of et al. (2015) and Fuchs et al. (2016b). Additionally, reaching or increasing anoxia due to a deepened methanogenesis could have been outcompeted by thermocline. Stronger effects could be expected energetically more favorable sulfate reducers, but H2S especially in lakes that already experience low oxygen concentrations at the SWI did not indicate an increase concentrations at the end of summer stratification. of sulfate reduction rates in response to decreasing Results showed that the deepened thermocline was oxygen concentrations or changes in the composition strongly correlated to specific phytoplankton groups, of sedimenting organic material. and has the potential to indirectly affect sedimentation Recent research indicates that global warming will rates and concentrations of nutrients in sediment substantially impact methane emission rates, mainly porewater, which were strongly correlated to green- due to methane release from permafrost areas (Anisi- house gas concentrations in the sediments. However, mov, 2007), thawing lakes (Walter et al., 2006), and our results do not indicate a biomass increase of from ice-bubble storage after melting in spring phytoplankton groups leading to mass sedimentation (Sepulveda-Jauregi et al., 2015), but also due to events. Moreover, the path model demonstrates that increased gas transport through emergent plants the dominant phytoplankton groups, as well as related to sediment temperature (Kankaala et al., elements related to the N-cycle and microbial abun- 2004). In contrast, global warming does not seem to dances at the SWI, did not explain the correlations increase methane emission from deep temperate lakes. between the blocks of variables considering the impact We did not observe an increase or decrease of methane of the treatment on the C-cycle. Furthermore, nutri- emission due to a lowered thermocline. Likewise, ents, elements, and greenhouse gas concentrations at Flury et al. (2010) did not find any effect of warming the SWI seemed to be more affected by the interac- on greenhouse gas emission from mesocosms in Lake tions among each other and the organic matter input Hallwil (Switzerland). Previous studies in Lake than by temperatures and the availability of electron 123 Hydrobiologia (2017) 799:65–81 79 acceptors. For the future, more work is needed community in surface sediments of a shallow, eutrophic focusing on the impacts of global warming on the lake. Aquatic Sciences 51: 153–178. Cantin, A., B. E. Beisner, J. M. Gunn, Y. T. Prairie & J. sediment. Certainly the repetition of such experiments G. Winter, 2011. Effects of thermocline deepening on lake over several years, and the inclusion of different plankton communities. Canadian Journal of Fisheries and plankton communities, would allow us to see remark- Aquatic Sciences 68: 260–276. able trends in the response of the ecosystem. Casper, S. J. (ed.), 1985. Lake Stechlin. A Temperate Olig- otrophic Lake. Dr. W. Junk Publishers, Dordrecht. Casper, P., 1992. Methane metabolism in Baltic lakes of dif- Acknowledgements We deeply thank A. Penske, M. ferent trophic state. Limnologica 22: 121–128. Sachtleben, O. M. Schlockermann, H. Volkmann, and R. Casper, P., D. D. Adams, A. L. S. Furtado, O. C. Chan, T. Yague¨ for technical support and the preparation of the Gonsciorczyk & R. Koschel, 2005. Greenhouse gas cycling experiment; M. Brehm, G. D. Flores, T. Gonsciorczyk, V. in aquatic ecosystems – methane in temperate lakes across Kasalicky´, C. Kasprzak, and T. Weber for their assistance in an environmental gradient in northeast Germany. Ver- preparation and sampling; U. Beyer, C. Burmeister, O. Dellwig, handlungen der Internationalen Vereinigung fu¨r G. Idoate, M. Lentz, F. Okagawa, and G. Siegert for chemical theroretische und angewandte Limnologie 29: 564–566. analyses; D. Giling, F. Ho¨lker, T. Petzold, and G. Singer for Cline, J. D., 1969. Spectrophotometric determination of advice on statistics and modeling in R. We thank the entire team hydrogen sulfide in natural waters. Limnology and of the TemBi project for the planning, preparation, and Oceanography 14: 454–458. conduction of this experiment, including C. Engelhardt, M. Cole, J. K., J. P. Peacock, J. A. Dodsworth, A. J. Williams, D. O. Gessner, H.-P. Grossart, T. Hornick, J. Hu¨peden, E. Huth, P. B. Thompson, H. Dong, G. Wu, B. P. Hedlund, 2012. Kasprzak, G. Kirillin, L. Krienitz, E. Mach, U. Mallok, G. Mohr, Sediment microbial communities in great boiling spring M. Monaghan, R. Rossberg, M. Sachtleben, J. Sareyka, M. are controlled by temperature and distinct from water Soeter, and C. Wurzbacher. Furthermore, we thank the communities. The ISME Journal 7(4): 718–729. anonymous reviewers who helped to improve the manuscript. Conrad, R., O.-C. Chan, P. Claus & P. Casper, 2007. Charac- Funding for this study was provided by the Leibniz Association terization of methanogenic archaea and stable isotope (project ‘‘Climate-driven changes in biodiversity of microbiota- fractionation during methane production in the profundal TemBi’’ SAW-2011-IGB-2). The construction of the enclosure sediment of an oligotrophic lake (Lake Stechlin, Ger- facility was funded by the German Federal Ministry of many). Limnology and Oceanography 52: 1393–1406. Education and Research (BMBF, No. 033L041B). Conrad, R., P. Claus & P. Casper, 2010. Stable isotope fractionation during the methanogenic degradation of organic matter in the sediment of an acidic bog lake, Lake Grosse Fuchskuhle. Limnology and Oceanography 55: References 1932–1942. Diehl, S., 2002. Phytoplankton, light, and nutrients in a gradient Anisimov, O. A., 2007. Potential feedback of thawing per- of mixing depths: theory. Ecology 83: 386–398. mafrost to the global climate system through methane Dokulil, M. T. & K. Teubner, 2000. Cyanobacterial dominance emission. Environmental Research Letters 2: 045016. in lakes. Hydrobiologia 438: 1–12. Archer, D., 1995. Upper ocean physics as relevant to ecosystem Dos Santos Furtado, A. L. & P. Casper, 2000. Factors influ- dynamics: a tutorial. Ecological Applications 5: 727–739. encing methane production in an oligotrophic and in a Azam, F., T. Fenchel, J. G. Field, J. S. Gray, L. A. Meyer-Reil & eutrophic German lake. Verhandlungen der Internationalen F. Thingstad, 1983. The ecological role of water-column Vereinigung fu¨r theoretische und angewandte Limnologie microbes in the sea. Marine Ecology Progress Series 10: 27: 1–5. 257–263. Duc, N. T., P. Crill & D. Bastviken, 2010. Implications of Bastviken, D., J. Ejlertsson & L. Tranvik, 2002. Measurement of temperature and sediment characteristics on methane for- methane oxidation in lakes: a comparison of methods. mation and oxidation in lake sediments. Biogeochemistry Environmental Science and Technology 36: 3354–3361. 100: 185–196. Bauchrowitz, M., 2012. The LakeLab – a new experimental Ebina, L., T. Tsutsui & T. Shirai, 1983. Simultaneous deter- platform to study impacts of global climate change on mination of total nitrogen and total phosphorus in water lakes. SIL News 60: 10–12. using peroxodisulfate oxidation. Water Research 17: Behrenfeld, M. J., R. T. O’Malley, D. A. Siegel, C. R. McClain, 1721–1726. J. L. Sarmiento, G. C. Feldman, A. J. Milligan, P. Fallon, R. D., S. Harrits, R. S. Hanson & T. D. Brock, 1980. The G. Falkowski, R. M. Letelier & E. S. Boss, 2006. Climate- role of methane in internal carbon cycling in Lake Mendota driven trends in contemporary ocean productivity. Nature during summer stratification. Limnology and Oceanogra- Letters 444: 752–755. phy 25: 357–360. Berger, S. A., S. Diehl, H. Stibor, P. Sebastian & A. Scherz, Fee, E. J., R. E. Hecky, S. E. M. Kasian & D. R. Cruikshank, 2014. Separating effects of climatic drivers and biotic 1996. Effects of lake size, water clarity, and climatic feedbacks on seasonal plankton dynamics: no sign of variability on mixing depths in Canadian Shield lakes. trophic mismatch. Freshwater Biology 59: 2204–2220. Limnology and Oceanography 41: 912–920. Bostro¨m, B., A.-K. Pettersson & I. Ahlgren, 1989. Seasonal Flury, S., D. F. McGinnis & M. O. Gessner, 2010. Methane dynamics of a cyanobacteria-dominated microbial emission from a freshwater marsh in response to 123 80 Hydrobiologia (2017) 799:65–81

experimentally simulated global warming and nitrogen Houser, J. N., 2006. Water color affects the stratification, sur- enrichment. Journal of Geophysical Research 115: face temperature, heat content, and mean epilimnetic G01007. doi:10.1029/2009JG001079. irradiance of small lakes. Canadian Journal of Fisheries and France, R., 1997. Land water linkages: influences of riparian Aquatic Sciences 63: 2447–2455. deforestation on lake thermocline depth and possible con- Hupfer, M. & J. Lewandowski, 2008. Oxygen controls the sequences for cold stenotherms. Canadian Journal of phosphorus release from lake sediments – a long-lasting Fisheries and Aquatic Sciences 54: 1299–1305. paradigm in limnology. International Review of Hydrobi- Frindte, K., W. Eckert, K. Attermeyer & H.-P. Grossart, 2013. ology 93: 415–432. Internal wave-induced redox shifts affect biogeochemistry Imboden, D. M. & A. Wuest, 1995. Mixing mechanisms in and microbial activity in sediments: a simulation experi- lakes. In Lerman, A., D. M. Imboden & J. R. Gat (eds.), ment. Biogeochemistry 113: 423–434. Physics and Chemistry of Lakes. Springer, Berlin: 83–138. Frindte, K., M. Allgaier, H.-P. Grossart & W. Eckert, 2015. Intergovernmental Panel on Climate Change (IPCC), 2007. Microbial response to experimentally controlled redox Fourth Assessment Report of the Intergovernmental Panel transitions at the sediment water interface. PLoS ONE 10: on Climate Change: The Physical Science Basis. Cam- e0143428. doi:10.1371/journal.pone.0143428. bridge University Press, Cambridge/New York. Fu, C., S. Large, B. Knight, A. J. Richardson, A. Bundy, G. Kankaala, P., A. Ojala & T. Ka¨ki, 2004. Temporal and spatial Reygondeau, J. Boldt, G. I. Van Der Meeren, M. A. Torres, variation in methane emissions from a flooded transgres- I. Sobrino, A. Auber, M. Travers-Trolet, C. Piroddi, I. sion shore of a boreal lake. Biogeochemistry 68: 297–311. Diallo, D. Jouffre, H. Mendes, M. F. Borges, C. P. Lynam, Kasprzak, P., T. Shatwell, M. O. Gessner, T. Gonsiorczyk, G. M. Coll, L. J. Shannon & Y.-J. Shin, 2015. Relationships Kirillin, G. Selmeczy, J. Padisa´k & C. Engelhardt, 2017. among fisheries exploitation, environmental conditions, Extreme weather event triggers cascade towards extreme and ecological indicators across a series of marine turbidity in a clear-water lake. Ecosystems. doi:10.1007/ ecosystems. Journal of Marine Systems 148: 101–111. s10021-017-0121-4. Fuchs, A., G. B. Selmeczy, P. Kasprzak, J. Padisa´k & P. Casper, Kelly, C. A. & D. P. Chynoweth, 1981. The contributions of 2016a. Coincidence of sedimentation peaks with diatom temperature and of the input of organic matter in control- blooms, wind, and calcite precipitation measured in high ling rates of sediment methanogenesis. Limnology and resolution by a multi-trap. Hydrobiologia 763: 329–344. Oceanography 26: 891–897. Fuchs, A., E. Lyautey, B. Montuelle & P. Casper, 2016b. Effects Kirillin, G., C. Engelhardt & S. Golosov, 2009. Transient con- of increasing temperatures on methane concentrations and vection in upper lake sediments produced by internal methanogenesis during experimental incubation of sedi- seiching. Geophysical Research Letters 36: L18601. ments from oligotrophic and mesotrophic lakes. Journal of doi:10.1029/2009GL040064. Geophysical Research: Biogeosciences 121: 1394–1406. Kirillin, G., T. Shatwell & P. Kasprzak, 2013. Consequences of Gauthier, J., Y. T. Prairie & B. E. Beisner, 2014. Thermocline thermal pollution from a nuclear plant on lake temperature deepening and mixing alter zooplankton phenology, bio- and mixing regime. Journal of Hydrology 496: 47–56. mass and body size in a whole-lake experiment. Freshwater Koschel, R. & D. D. Adams, 2003. Preface: an approach to Biology 59: 998–1011. understanding a temperate oligotrophic lowland lake (Lake Giling, D. P., J. C. Nejstgaard, S. A. Berger, H.-P. Grossart, G. Stechlin, Germany). Archiv fu¨r Hydrobiologie Special Kirillin, A. Penske, M. Lentz, P. Casper, J. Sareyka & M. Issues Advances in Limnology 58: 1–9. O. Gessner, 2016. Thermocline deepening boosts ecosys- Kwok, R., 2013. The great outdoors. Nature 503: 301–303. tem metabolism: evidence from a large-scale lake enclo- Lozupone, C. A. & R. Knight, 2007. Global patterns in bacterial sure experiment simulating a summer storm. Global diversity. Proceedings of the National Academy of Sci- Change Biology. doi:10.1111/gcb.13512. ences 104(27): 11436–11440. Glissmann, K., K. J. Chin, P. Casper & R. Conrad, 2004. Lund, J. W. G., C. Kipling & E. D. Le Cren, 1958. The inverted Methanogenic pathway and archaeal community structure microscope method of estimating algal numbers and the in the sediment of eutrophic Lake Dagow: effect of tem- statistical basis of estimations by counting. Hydrobiologia perature. Microbial Ecology 48: 389–399. 11(2): 143–170. Goedkoop, W., K. R. Gullberg, R. K. Johnson & I. Ahlgren, Maranger, R. & D. F. Bird, 1996. High concentrations of viruses 1997. Microbial response of a freshwater benthic com- in the sediments of Lac Gilbert, Que´bec. Microbial Ecol- munity to a simulated diatom sedimentation event: inter- ogy 31: 141–151. active effects of a benthic fauna. Microbial Ecology 34: Oksanen, J., G. Blanchet, R. Kindt, P. Legendre, P. R. Minchin, 131–143. R. B. O’Hara, G. L. Simpson, P. Solymos, M. H. H. Stevens Gonsiorczyk, T., P. Casper & R. Koschel, 2003. Long-term & H. Wagner, 2016. Vegan: Community Ecology Package. development of the phosphorus accumulation and oxygen- R package version 2.3-5. consumption in the hypolimnion of oligotrophic Lake Ongun Sevindik, T., E. Altundal & F. Kucuk, 2015. The sea- Stechlin and seasonal variations in the pore water chem- sonal and spatial distribution of the phytoplankton of a istry of the profundal sediments. Archiv fu¨r Hydrobiologie mesotrophic lake related to certain physical and chemical Special Issues Advances in Limnology 58: 73–86. parameters. Ekoloji Dergisi 24: 14–23. Grossart, H.-P., K. Frindte, C. Dziallas, W. Eckert & K. Padisa´k, J., L. Krienitz, W. Scheffler, R. Koschel, J. Kristiansen W. Tang, 2011. Microbial methane production in oxy- & I. Grigorszky, 1998. Phytoplankton succession in the genated water column of an oligotrophic lake. PNAS 108: oligotrophic Lake Stechlin (Germany) in 1994 and 1995. 19657–19661. Hydrobiologia 369(370): 179–197. 123 Hydrobiologia (2017) 799:65–81 81

Padisa´k, J., E´ . Hajnal, L. Krienitz, J. Lakner & V. U¨ veges, 2010. dioxide emissions from 40 lakes along a north-south lati- Rarity, ecological memory, rate of floral change in phyto- tudinal transect on Alaska. Biogeosciences 12: 3197–3223. plankton – and the mystery of the Red Cock. Hydrobiologia Shelley, F., F. Abdullahi, J. Grey & M. Trimmer, 2015. 653: 45–64. Microbial methane cycling in the bed of a chalk river: Phillips, K. A. & M. W. Fawley, 2002. Winter phytoplankton oxidation has the potential to match methanogenesis community structure in three shallow temperate lakes enhanced by warming. Freshwater Biology 60: 150–160. during ice cover. Hydrobiologia 470: 97–113. Shiah, F.-K. & H. W. Ducklow, 1994. Temperature and sub- Pinto, F., S. Larsen & P. Casper, 2013. Viriobenthos in aquatic strate regulation of bacterial abundance, production and sediments: variability in abundance and production and specific growth rate in Chesapeake Bay, USA. Marine impact on the C-cycle. Aquatic Sciences 75: 571–579. Ecology Progress Series 103: 297–308. Proctor, L. M., A. Okubo & J. A. Fuhrman, 1993. Calibrating Snucins, E. & J. Gunn, 2000. Interannual variation in the ther- estimates of phage-induced mortality in marine bacteria: mal structure of clear and colored lakes. Limnology and ultrastructural studies of marine bacteriophage develop- Oceanography 45: 1639–1646. ment from one-step growth experiments. Microbial Ecol- Sommer, U., Z. M. Gliwicz, W. Lampert & A. Duncan, 1986. ogy 25: 161–182. The PEG-model of seasonal succession of planktonic Ricciardi-Rigault, M., D. F. Bird & Y. T. Prairie, 2000. Changes events in fresh waters. Archiv fu¨r Hydrobiologie 106: in sediment viral and bacterial abundances with hypolim- 433–471. netic oxygen depletion in a shallow eutrophic Lac Brome Sterner, R. W., 1990. The ratio of nitrogen to phosphorus (Quebec, Canada). Canadian Journal of Fisheries and resupplied by herbivores: zooplankton and the algal com- Aquatic Sciences 56: 1284–1290. petitive arena. The American Naturalist 136: 209–229. Rudd, J. W. M. & R. D. Hamilton, 1978. Methane cycling in a Tang, K. W., D. F. McGinnis, K. Frindte, V. Bru¨chert & H.-P. eutrophic shield lake and its effects on whole lake meta- Grossart, 2014. Paradox reconsidered: methane oversatu- bolism. Limnology and Oceanography 23: 337–348. ration in well-oxygenated lake waters. Limnology and Sanchez, G., 2013. PLS Path Modeling with R. Trowchez Edi- Oceanography 59: 275–284. tions, Berkeley. http://www.gastonsanchez.com/ Teubner, K., M. Tolotti, S. Greisberger, H. Morscheid, M. PLSPathModelingwithR.pdf. T. Dokulil & H. Morscheid, 2003. Steady state phyto- Sastri, A. R., J. Gauthier, P. Juneau & B. E. Beisner, 2014. plankton in a deep pre-alpine lake: species and pigments of Biomass and productivity responses of zooplankton com- epilimnetic versus metalimnetic assemblages. Hydrobi- munities to experimental thermocline deepening. Limnol- ologia 502: 49–64. ogy and Oceanography 59: 1–16. Verburg, P., R. E. Hecky & H. Kling, 2003. Ecological conse- Schindler, D. W., 1997. Widespread effects of climatic warming quences of a century of warming in Lake Tanganyika. on freshwater ecosystems in North America. Hydrological Science 301: 505–507. Processes 11: 1043–1067. Walter, K. M., S. A. Zimov, J. P. Chanton, D. Verbyla & F. Schindler, D. W., 2001. The cumulative effects of climate S. Chapin III, 2006. Methane bubbling from Siberian thaw warming and other human stresses on Canadian freshwa- lakes as a positive feedback to climate warming. Nature ters in the new millennium. Canadian Journal of Fisheries Letters 443: 71–75. and Aquatic Sciences 58: 18–29. Wang, J., F. Pan, J. Soininen, J. Heino & J. Shen, 2016. Nutrient Schindler, D. W., 2009. Lakes as sentinels and integrators for the enrichment modifies temperature–biodiversity relation- effects of climate change on watersheds, airsheds, and ships in large-scale field experiments. Nature Communi- landscapes. Limnology and Oceanography 54: 2349–2358. cations 7: 13960. doi:10.1038/ncomms13960. Schulz, S., H. Matsuyama & R. Conrad, 1997. Temperature Wetzel, R. G., 2001. Limnology – Lake and River Ecosystems. dependence of methane production from different precur- Academic, San Diego, CA. sors in a profundal sediment (Lake Constance). FEMS Yvon-Durocher, G., A. P. Allen, D. Bastviken, R. Conrad, C. Microbiology Ecology 22: 207–213. Gudasz, A. St-Pierre, N. Thanh-Duc & P. A. del Giorgio, Selmeczy, G. B., K. Tapolczai, P. Casper, L. Krienitz & J. 2014. Methane fluxes show consistent temperature Padisa´k, 2016. Spatial- and niche segregation of DCM- dependence across microbial to ecosystem scales. Nature forming cyanobacteria in Lake Stechlin (Germany). 507: 488–495. Hydrobiologia 764: 229–240. Zeikus, J. G. & M. R. Winfrey, 1976. Temperature limitation of Sepulveda-Jauregi, A., K. M. Walter Anthony, K. Martinez- methanogenesis in aquatic sediments. Applied and Envi- Cruz, S. Greene & F. Thalasso, 2015. Methane and carbon ronmental Microbiology 31: 99–107.

123