Ó Springer 2005

Hydrobiologia (2005) 544:279–288 DOI 10.1007/s10750-005-0926-6

Primary Research Paper

Depleted methane-derived carbon in waters of Lake Baikal, Siberia Alexander A. Prokopenko* & Douglas F. Williams Department of Geological Sciences, University of South Carolina, Columbia, SC, 29208, USA (*Author for correspondence: Tel.: +1 803 777-1697; E-mail: [email protected]) Received 8 July 2002; accepted 18 January 2005

Key words: Siberia, methane depletion, carbon dioxide

Abstract Results of hydrochemical and stable isotope measurements during the ice-breaking period on Lake Baikal indicate an apparent lack of relationship between measured d13C of dissolved inorganic carbon (DIC) and phytoplankton below the trophogenic layer. While planktonic values of )31.7 to )33.5& are within a typical lacustrine range, the d13C values of DIC turned out to be very negative, from )28.9 to )35.6&. These isotopic values of DIC appear to be associated with oxidation of methane that accumulated during winter ice cover period. At the time of sampling, however, the observed depletion did not affect the phytoplankton/DIC fractionation relationship, because the difference between phytoplankton and DIC ()20 to )22& in surface waters) lies within the expected range of the fractionation coefficient. By analogy with small lakes, we explain this lack of relationship by the time lag between peak productivity and peak methane oxidation. Our interpretation of the Baikal DIC isotopic signature is consistent with methanogenesis in bottom sediments and with the known presence of widespread unstable gas hydrates and active methane seeps on the lake floor. Our findings suggest that methane is an important component of the Baikal carbon cycle, that late winter concentrations of methane in Baikal under ice may be 3–4 orders of magnitude higher than previously reported values for summer, and that the lake may be emitting a significant amount of methane to the atmosphere. Introduction: sampling during the spring diatom bloom in Lake Baikal Lake Baikal is the world’s deepest and most voluminous lake. In the annual carbon balance in Lake Baikal, 92% is attributed to the input of autochthonous organic carbon (Votintzev et al., 1975). Among autochthonous sources, phytoplankton plays the dominant role and diatoms account for as much as 90% of phytoplankton biomass (Vykhristyuk, 1980). A very characteristic feature of Baikal is the peak of primary production in spring during a diatom bloom under the ice, when the dominant large-celled species Aulacoseira baicalensis accounts for a biomass over 1000 mg m)3 during a productive year and below 500 mg m)3 during an unproductive year. By comparison, the autumn diatom bloom is 1–2 orders of magnitude less sig-

nificant (Popovskaya, 1967). The major spring bloom starts in February–March under the ice and continues through May–June. During the ice-out period, warming of surface waters progressively deepens the thermocline and the inverse winter stratification ends with the establishment of homothermy in the upper 150–300 m layer (Shimaraev et al., 1994). This thermal reorganization ends the Baikal spring diatom bloom, leading to the descent of masses of diatoms into deep waters (Votintzev & Popovskaya, 1964). Peak productivity in the spring makes this period a critical time for studying the relationship between components of the Lake Baikal carbon cycle. In this work, we report the results of hydrochemical and isotopic measurements during this important period of Baikal’s diatom bloom. Vertical profiles at two stations (Fig. 1), one in the

280

b Figure 1. The map of Lake Baikal and the location of water sampling stations during ice-out period with shading indicating the extent of ice cover at the time of sampling. Isobath contours are plotted at 100, 300, 400, 500, 800, 1200, and 1500 m. The mapped extent of gas hydrate layer (hatching) is based on Kuzmin et al. (2000).

56 N

CONTINUOUS ICE COVER DISCONTINUOUS ICE COVER

St. 5a

54 N

109 E

55 N

110 E

St. 6

Materials and methods

SAMPLING SITES

108 E

STUDIED METHANE SEEPS

106 E

107 E

53 N

MAPPED GAS HYDRATE LAYER

52 N

St. 13

104 E

Northern Basin (930 m water depth, several days after ice-breaking) and one in the Southern Basin (1300 m water depth, 2–3 weeks after ice-breaking) captured the range of variability of several quantities including pH, alkalinity, and stable carbon isotope composition of phytoplankton, particulate organic matter (POM) and dissolved inorganic carbon (DIC). These profiles reveal the dynamic nature of the spring bloom by providing ‘snapshots’ of early and late spring conditions during May 1994, and they provide new evidence for sources of depleted carbon in Lake Baikal.

105E

Phytoplankton were collected by net towing. The tow suspensions were concentrated on pre-combusted glass fiber filters (mesh 0.7 lm) and dried at 50 °C. Water and particulate organic matter (POM) were sampled from the same volume of water from a 10 l Niskin bottle. POM was collected on glass fiber filters after pre-filtering water through three screens with mesh sizes 325, 250 and 125 lm and dried at 50 °C. Bubble-free water samples for dissolved inorganic carbon (DIC) isotopic analyses were collected in HCl pre-washed 20 ml scintillation vials and poisoned with 2–3 drops of 15% mercuric chloride solution. Vials were tightly sealed, wrapped in foil and refrigerated. The stable isotopic composition of DIC was measured using an Optima Isotopic Ratio Mass Spectrometer with 0.3& (1%) precision. For CO2 extraction, a 5 ml aliquot of water sample was injected with a syringe through a rubber septa into a reaction vessel under vacuum with phosphoric acid, continuously agitated with a magnetic stirrer. Reaction was allowed to take place for 40 min. At first, CO2 and water vapor were frozen with liquid nitrogen in a trap while the non-condensable phases (NC) were evacuated. Pure CO2 gas was then collected in a cold finger with liquid nitrogen and flame-sealed in a glass tube for later intro-

281 duction into the mass spectrometer. The yields of both NC and CO2 for each of the samples were monitored using a thermocouple gauge. Profiles of temperature, light absorbance and oxygen were obtained using a SeaCat CTD instrument deployed with a wire winch to depths of 300 m at St. 6 and 1000 m at St. 13. The oxygen detector was calibrated prior to the expedition using the Winkler titration method (Votintzev, 1961). The pH was measured on board on 70–100 ml samples using a HACH ONE standard pH-meter with free flowing junction electrode. Every measurement of the fresh sample was followed by control measurements of 4.0 and 6.96 standard pH buffers and then by the second measurement of the sample. The pH values were corrected correspondingly to buffer offsets, which were on average ±0.05 (0.8%). Due to low ionic strength of Baikal water (Votintzev, 1961), it took up to 10 min to reach stable readings (1.5–2 min for buffers). The first pH reading appeared to be more reliable because the second measurement, especially for the high pH surface layer samples, was usually lower, perhaps due to sample warming and exchange with atmosphere. Alkalinity was measured by titration of 50-ml water samples with 0.1 M H2SO4 using the HACH ONE Digital Titrator. The standard error of alkalinity measurements was 0–3%; however, for some samples it was as high as 6–10%. In cases when reproducibility was a problem, three or more measurements were taken to calculate the average values.

Results Boundaries between three layers, most evident in temperature and dissolved oxygen profiles, are clearly observed within the water column at both stations 6 and 13, accented by shading in Figures 2 and 3. The boundary between the upper mixed layer and the underlying ‘intermediate’ layer corresponds to the position of the thermocline. The lower boundary of the ‘intermediate’ layer (light shading in Fig. 2) lies at ca. 220 m water depth and 3.4 °C in the Northern Basin and at ca. 320 m water depth and 3.6 °C in the Southern Basin. Thus, the observed shifts in water chemistry between the ‘intermediate’ and deep waters occur

at temperatures close to those of the mesothermal maximum layer. This layer forms at different levels between 150 m water depth at 3.7 °C and 250 m depth at 3.4 °C (depending on the extent of autumn mixing) and serves as an impermeable density boundary limiting free convection in the epilimnion (Shimaraev & Granin, 1991). Vertical profiles at St. 6 in the Northern Basin represent early spring conditions. The thermocline is shallow (60 m) and the inverse winter stratification is still strong (Fig. 2). Phytoplankton are concentrated in the trophogenic layer, as shown by the relative changes in light absorbance and by high dissolved oxygen concentrations. Isotopic values of particulate organic matter (POM) above the thermocline at St. 6 ()32.8 to )33.9&) correspond well to the d13C of phytoplankton sampled at a neighboring Northern Basin St. 5a ()32.8&). Below the thermocline at St. 6, the d13C of POM becomes distinctly heavier (more positive) due to organic matter decomposition. The pH profile at St. 6 reveals the effect of CO2 uptake by phytoplankton on the carbonate system by showing significant departures towards the higher end of the range reported to be typical for Baikal (7.5–8) (Votintzev, 1961) and from the pH values observed below the thermocline (Fig. 2). Alkalinity measurements, although suffering from a lack of reproducibility at some depth levels, generally lie close to the Baikal summer range of 1.08–1.1 leq l)1 (Falkner et al., 1991). At or below the thermocline at St. 6, the measured d13C values of dissolved inorganic carbon (DIC) are quite unusual in that they are more negative than POM and even as negative as the phytoplankton measured at St. 5a (Fig. 2). For instance, only in the uppermost layer at St. 6, and in one bottom water sample measured at 900 m at St. 5a, are the DIC values of )21.7 to )23.3& more positive than POM values. At thermocline depths, by contrast, the measured DIC is as negative as )30.4 to )33.6& (Fig. 2). Vertical profiles at St. 13 in the Southern Basin represent the transition from late spring to summer conditions. The temperature profile shows that the waters are close to homothermy. The thermocline, which is quite deep at ca. 120 m, is actually better expressed in the profiles of dissolved oxygen and light absorbance than in the temperature profile itself (Fig. 3). Phytoplankton and POM at St. 13 are practically invariable in composition ()28.4 and

282 STATION 6, NORTHERN BAIKAL, several days after ice-breakup o

Temperature, C 3.4

2.4

13

pH

1.4

7.4

7.8

C, %°PDB

-20

8.2

-24

-28

Carbon dioxide yield, relative units

-32

-36

250

450

650

0 20 40 60 80

phytoplankton (St. 5a) Oxygen

400

Light absorbance

120 200

Temperature

water depth, m

100

600 800

8.5

9.5

DIC POM

10.5

(St.5a)

Oxygen, mg.L-1 1000 10 15 20 25 30

Light absorbance, %

1.0

1.2

1.4

Alkalinity, µeq.L-1

100

140

180

non-condensible phase, relative units

Figure 2. Vertical water column profiles at station 6 in the Northern Basin of Lake Baikal supplemented with phytoplankton and bottom-water DIC measurements from the neighboring station 5a. Horizontal error bars show standard deviation of replicate measurements. In pH profiles solid and open triangles indicate the first and second measurements correspondingly. The yields of carbon dioxide and non-condensible phases are given in relative units corresponding to the vacuum gauge readings during sample extraction. Note that the temperature scale is reversed with cold to the right.

)31.3& correspondingly) and are uniformly distributed throughout the upper mixed layer. Below the thermocline, the d13C ratios of POM and abundant phytoplankton (collected by net towing at depths down to 350 m) show little departure from the average values of )28.7 and )31.3& accordingly (Fig. 3). Unlike at St. 6, the pH profile at St. 13 does not show high-amplitude changes indicative of rapid CO2 uptake in the uppermost 15 m of the trophogenic layer. Instead, pH values in the entire mixed layer above the thermocline are shifted by 0.2–0.3 towards the higher end of the typical Baikal pH range (Votintzev, 1961) (Fig. 3). Alkalinity values also fall within the reported range (Falkner et al., 1991) and in deep waters at St. 13 they were more reproducible than at St. 6. Notable features of the pH profile at St. 13 are the negative shifts observed with the transition from the upper mixed layer above thermocline to the underlying ‘intermediate’ layer and once again with transition from ‘intermediate’ to deep waters (shading in Fig. 3). The distinct negative DIC compositions noted at St. 6 (Fig. 2) are even more

pronounced at St. 13, where measured DIC values reach )35.6, 4–7& more negative than POM and even phytoplankton values (Fig. 3). Interestingly, the d13C profile of measured DIC appears closely related with stratification: DIC d13C shifts to heavier values below 300 m water depth, with the transition from ‘intermediate’ layer to deep waters (Fig. 3). Visual correspondence between the CO2yield and DIC-d13C profiles (Fig. 3) clearly indicates the presence of ‘additional’ CO2 in the isotopically depleted samples.

Discussion The relationship between [CO2]aq and the d13C of DIC, phytoplankton and POM The nature of our springtime data set forces us to divide the discussion of the isotope composition of the main components of the Baikal’s carbon cycle into two parts and discuss the upper 25-m interval of the water column constituting the trophogenic

283

STATION 13, SOUTHERN BAIKAL, two-three weeks after ice-breakup Temperature, oC 3.0

4.0

13

pH

5.0

7.4

C, %°PDB

7.6 7.8

-20

8.0

-24

-28

-32

Carbon dioxide yield, relative units -36

250

450

650

850

0 20 40 60 80

100

200 200

ALK

pH

Light absorbance

Oxygen

150

Temperature

water depth, m

100

phytoplankton DIC POM

400 600 800 1000 1200 1400

7.5

8.5

9.5

Oxygen,mg.L-1

10

15

20

Light absorbance, %

1.0

1.1

100

1.2

Alkalinity, µeq.L-1

150

200

non-condensible phase, relative units

Figure 3. Vertical water column profiles at station 13 in the Southern Basin of Lake Baikal. For explanation of special symbols please see Figure 2.

layer of Lake Baikal (Popovskaya, 1967) separately from the deeper layers. Within the upper 25m layer of the Baikal water column, the relationship between [CO2]aq and the d13C of DIC, phytoplankton and POM appears quite reasonable. For instance, carbonate equilibrium calculations suggest that [CO2]aq could drop 1.4–2.2 times as a result of uptake by phytoplankton, because the pH shift of 0.2–0.4 is observed at an alkalinity value of ca. 1.1 leq l)1 in the trophogenic layer (Figs. 2 and 3). From this calculation, using the relationship ep ¼ 11.64(log[CO2]aq))3.56 between fractionation coefficient ep and [CO2]aq (Hollander & McKenzie, 1991), we may expect ep to decrease 1.1–1.2 times. For instance, if ep of 15& could be expected under slow cell growth and plentiful supply of carbon dioxide at [CO2]aq concentrations of 40 lmol l)1, ep could decrease by 3& during a phytoplankton bloom. The observed d13C difference between phytoplankton and DIC in the Bai-

kal trophogenic layer (10.8& within 0–25 m in the Northern Basin and 10.7& within 0–15 m in the Southern Basin) therefore appears quite reasonable. Apparently, the ‘normal’ DIC carbon isotope composition in Lake Baikal is around )21&, as seen from the matching values between the uppermost layer ()22.8& in Northern Basin, )21.0& in Southern Basin) and the deep waters ()21.8& in Northern Basin, )20.2& in Southern Basin). Below the trophogenic layer, however, measured DIC values at St. 6 and 13 are too negative to be explained by the expected [CO2]aq / ep relationship alone. The DIC d13C profile is related to stratification as follows from an evident similarity between isotopic, temperature and dissolved oxygen profiles (Fig. 3) and therefore is unlikely to be due to analytical error. The departure from the expected ‘normal’ relationship between DIC and phytoplankton apparently reflects contribution of

284 ‘additional’ carbon dioxide characterized by depleted (negative) carbon isotope composition. The presence of this ‘additional’ carbon dioxide is also indicated by high CO2 yield in the unusual isotopically-depleted samples, 1.2–1.7 times higher than in surface and in deep waters (Fig. 3). The likely explanation for the major negative excursion of the d13C in DIC in the intermediate waters is the contribution of depleted carbon from oxidation of dissolved methane. Methane accumulation under the ice during winter stratification Methane is known to accumulate in ice-covered lakes during winters and subsequently escape to the atmosphere during ice break-up (Phelps et al., 1998; Striegl et al., 2001). In their review of potential methane emissions in 19 north-temperate lakes, Michmerhuizen et al. (1996) found that a pulse in methane release at the time of ice melt may make up as much as 40% of annual emission of methane from a lake to the atmosphere. The authors concluded that because lakes store methane under ice cover during winter, CH4 concentrations and emission rates measured during the open-water period may substantially underestimate the true methane source strength of a given lake. Methane may co-exist with dissolved oxygen and carbon dioxide for several months under the ice with little oxidation. For example, as shown by the methane pumping experiments in Methane Lake, methane oxidized at a slow rate of 0.32 lmol l)1Æday)1 when oxygen concentrations are 11–13 mg l)1, and produced no discernible effect on water chemistry (Welch et al., 1980). As discussed below, slow to undetectable methane oxidation in the hypolimnion at high oxygen concentrations is due to nitrogen limitation of methanotrophic bacteria (Rudd et al., 1976; Rudd & Hamilton, 1978). During spring turnover in small north-temperate lakes oxidative losses of methane are small and therefore strong CH4 fluxes to the atmosphere have been observed immediately following the ice melt (Michmerhuizen et al., 1996; Striegl & Michmerhuizen, 1998). Although generally supportive of our hypothesis of significant methane accumulation in Lake Baikal under the ice, the implications from the above studies of small lakes are not fully applicable to the current Baikal data set. Firstly, we did

not directly measure CH4 concentrations in Lake Baikal water and therefore we cannot discuss methane emissions to the atmosphere during the ice melt. Secondly, unlike in small lakes studied by Michmerhuizen et al. (1996), we do observe significant methane oxidation as suggested by the extremely negative d13C of the DIC and by the presence of ‘additional’ CO2 in the intermediate waters (Figs. 2 and 3). In order to numerically constrain the production of carbon dioxide from methane oxidation, we use the relationship d13CCH4
[CCH4] + d13CDICbefore
[CDICbefore] ¼ d13CDICafter
[CDICafter], thus calculating the amount of CH4-derived depleted carbon necessary to shift the isotopic composition of DIC (Kennett et al., 2000). Using this numerical relation, we can estimate that it would take an addition of 162 to 253 lmol l)1 CH4 to exchange with ca. 50 mg l)1 of DIC (at pH  7.8 and ALK 1.1 leq l)1) in order to shift the isotopic composition from the initial )21& to the resultant )30 to )35&. If we factor in the 1.5 increase in total CO2 yield (DIC) observed during vacuum extraction (Fig. 3), the maximum estimate of CH4 addition necessary to shift the DIC composition to )35& during the winter-spring period in Lake Baikal would fall within the range of 433 to 568 lmol l)1, translating into 6.1 to 12.7 ml l)1 of dissolved CH4. Although these values are 3–4 orders of magnitude higher than previously reported CH4 measurements in Baikal (Namsaraev et al., 1995), our estimates are not unreasonable. For instance, the measured methane efflux from high-latitude lakes during a 10day period during spring ice melt was as high as 1.5– 2.07 g CH4 )m)2 when 780 lmol l)1 (17.5 ml l)1) of dissolved CH4 accumulated under ice without significant oxidation (Phelps et al., 1998). Even in small lakes of the Alaskan Arctic, dissolved CH4 levels could reach as high as 165 lmol l)1 (3.7 ml l)1) (Kling et al., 1992). In 102 small Finnish lakes surveyed for CO2 and CH4 emissions, the average methane concentration was 10.8 lmol l)1 with a maximum of 280 lmol l)1 (Striegl et al., 2001). Thus, the results of our measurements of spring DIC compositions in Baikal indicate significant accumulation of methane in lake waters during the winter ice-cover period. The fact that estimates of Baikal methane concentrations in spring are towards the high end of the range measured in other

285 high-latitude lakes should be attributed to several potential sources of methane as described below. Sedimentary sources of methane in Lake Baikal Unlike in small lakes, where CH4 is produced in hypoxic or anoxic hypolimnion from the primary produced organic matter, the Baikal hypolimnion is oxygenated throughout the year at above 70–75% saturation (Votintzev, 1990), and thus methane sources are apparently sedimentary. Active methanogenesis is typical of Baikal sediments: active methanogenesis was detected in 49 samples out of 50 analyzed by Namsaraev et al. (1995), and according to their estimates of the rates of methanogenesis vary from 0.01 to 534.7 ll kg)1 day)1. As much as 97.6 to 99.9% of methane in Baikal deepwater sediments is generated from CO2 and H2 (Namsaraev et al., 1995). In the Northern Basin of Lake Baikal methane serves as an important carbon source in microbial mats, where d13C compositions of sedimentary organic matter vary from )44.2– )49.5& (Namsaraev et al., 1994). In addition, thick strata of Lake Baikal sediments contain widespread layer of gas hydrates throughout Southern and Central Basins, with thicknesses from 34 to 450 m, on average 260 m (Kuzmin et al., 2000). The isotopic composition of sedimentary methane sampled in BDP-97 drill core was )63 ± 4& (Kuzmin et al., 2000). Baikal gas hydrates show evidence of destabilization as a result of an upward heat advection by hydrothermal fluids (DeBatist et al., 2001). This destabilization results in active venting, e.g., formation of mud volcanoes on the lake’s bottom (Van Rensbergen et al., 2001). Modern benthic organisms in the vicinity of active methane seeps in Lake Baikal have isotopic compositions as low as )60 to )72& and the apparent radiocarbon ages of 6860– 10 200 14C years (Grachev et al., 1996). While our DIC measurements during one season may not be entirely representative of the average annual methane efflux from Lake Baikal sediments, they at the very least suggest that methane deserves much more attention as an important component in Baikal carbon cycle than it has been given. For instance, in Lake Washington it was found that methane represents 20% of upward flux of organic carbon decomposed within the sediments and that

methane oxidation consumes 7–10% of total oxygen flux into the sediments (Kuivila et al., 1988). Apparently, due to hydrothermal gas hydrate destabilization and the presence of active vents, a much higher upward methane flux occurs in Lake Baikal and large part of this flux escapes into lake waters. Due to in situ oxidation, this methane may strongly affect the composition of the DIC in the intermediate waters in Lake Baikal as suggested by our d13C measurements. It is interesting to note from our data set, however, that these depletions did not propagate into the pool of photosynthetically-fixed carbon, because the isotopic composition of phytoplankton is decoupled from strongly negative d13C of the DIC (Figs 2 and 3). We address this seeming contradiction in the following two sections. Comparison of our observations with annual cycles of methane accumulation and oxidation studied in detail in a series of small North American lakes allows better understanding of the nature of the processes possibly taking place in Lake Baikal. The mechanism of methane oxidation in the Lake Baikal water column Methane produced in sediments of north-temperate lakes moves upward through the water column where it is oxidized by methanotrophic bacteria or emitted to the atmosphere (Striegl & Michmerhuizen, 1998 and references therein). In their culture experiments, Rudd et al. (1976) have shown that methane-oxidizing bacteria, although usually inhibited by high oxygen concentrations (above 1 mg l)1), may become oxygen-insensitive when concentrations of dissolved inorganic nitrogen increase. For lakes this implies that by itself, oxygen is not lethal to these bacteria; instead, the lack of methane oxidation is due to nitrogen limitation. As a result, whenever nitrogen loading is increased during the overturn (fall overturn in their study), a rapid oxidation of methane may take place despite high oxygen concentrations (Rudd et al., 1976). With the progression of the overturn, methane is rapidly oxidized in larger and larger proportions of the water column, resulting in as much as 90% of annual methane oxidation during and immediately following the overturn (Rudd & Hamilton, 1978). It is quite likely that the extreme DIC

286 depletions measured in Lake Baikal within the intermediate water layer (Fig. 3) have captured a similar process of rapidly progressing methane oxidation. During the spring overturn, higher nitrogen availability in intermediate waters due to enhanced water exchange, and also from the rain of decomposing POM from the Baikal early spring diatom bloom, could have enabled methane oxidizing bacteria to become ‘oxygen-insensitive’, thereby anomalously increasing oxidation of the methane, which accumulated in lake waters during the prior period of winter stratification. This rapid oxidation may then be expected to produce ‘additional’ carbon dioxide in the intermediate waters thereby affecting the stable isotopic composition of DIC. This phenomenon appears to be talking place in Lake Baikal as best observed in the profiles of CO2-yield and d13C of DIC at St. 13 (Fig. 3). Methane oxidation as a source of photosynthetically-fixed carbon dioxide The observed lack of relationship between ‘normal’ d13C of phytoplankton/POM and extremely depleted DIC in the intermediate waters in Lake Baikal (Figs. 2 and 3) does not appear contradictory when compared with the dynamics of carbon and methane cycles in small lakes. For instance, Rudd & Hamilton (1978) concluded that methane oxidation was not a major source of carbon dioxide utilized in primary production because of the time lag between the primary productivity peak (occurring when lakes are stratified) and the methane oxidation peak (occurring during and after the overturn, when primary productivity is low). At times of stratification, methane oxidation is not a major source of carbon dioxide for primary production because oxidation takes place within a narrow depth range at the thermocline and vertical diffusion of hypolimnetic methane across the thermocline does not exceed 7–9% of the total methane storage. In addition, only half of the oxidized methane is converted to CO2, whereas the other half is converted into methanotrophic bacterial cell material (Rudd & Hamilton, 1978). During overturn, when anomalously rapid methane oxidation takes place, photosynthesis is generally low, and as a result, the internal carbon dioxide production from methane oxidation does

not constitute a major source of carbon dioxide for primary producers either (Rudd & Hamilton, 1978). If applicable to Lake Baikal, this mechanism would suggest that at the time of peak productivity (occurring prior and during the ice-out period, when inverse winter stratification exists in Lake Baikal), the effect of methane oxidation on the isotopic composition of DIC utilized by primary producers is not very significant due to limited vertical diffusion of methane. Later, during spring overturn when homothermy is established in the upper 150–300 m of the Baikal water column, primary productivity is no longer at it’s peak and therefore potential change in DIC due to anomalously rapid methane oxidation would fail to leave an isotopic imprint on the primary produced organic matter. These assumptions appear consistent with our d13C analysis of different components of Lake Baikal carbon cycle in this study (Figs. 2 and 3).

Conclusions Our data collected in Lake Baikal during the iceout period and spring overturn appear to have captured ‘snapshots’ of the components of the lake’s carbon cycle during this dynamic period. The most striking features in the depth profiles at two stations are: (1) the extremely negative measured d13C of DIC as low as )35.6& in the intermediate waters; and (2) the lack of the expected fractionation-driven relationship between d13C of phytoplankton/POC and d13C of DIC in the intermediate waters (Figs. 2 and 3). The d13C composition of phytoplankton and POC appear to be ‘normal’ relative to ca. )21& DIC in surface and bottom waters of Lake Baikal, whereas the d13C of the DIC of intermediate waters is strongly affected by the in situ oxidation of methane. By analogy with other lakes, we hypothesize that methane-driven d13C depletions did not propagate into the pool of photosynthetically-fixed carbon because of the time lag between peak primary production occurring during inverse winter stratification in Lake Baikal and peak methane oxidation, which likely occurred during the spring overturn, when increased nitrogen availability allows methanotrophic bacteria to rapidly oxidize

287 methane despite high concentrations of oxygen throughout the water column. Our new estimates of dissolved methane concentrations in Lake Baikal during the spring ice-out period (based on estimated departures of DIC d13C compositions in the intermediate waters of the lake) are 3–4 orders of magnitude higher than values previously reported for Baikal waters. Because Baikal research cruises are usually scheduled in the summer, this apparent discrepancy indicates that the importance of seasonal ice cover and winter stratification in methane accumulation in the lake may be an overlooked component of the carbon cycle of Baikal. Our dissolved methane estimates fall toward the higher end of the range measured in other high-latitude lakes, and suggest that sedimentary sources in Lake Baikal are actively emitting anaerobically-produced methane. In addition to the methanogenesis in recent sediments beneath the oxidized surface layer, there may be a significant geological source of methane from hydrothermally destabilized gas hydrate layer operating somewhat continuously in this active rift basin. The rates of methane emission from Lake Baikal sediments could have been significantly different in the past. For instance, historic data reviewed by Granin & Granina (2002) suggest that the abundance of gas steam-throughs in the ice cover of Lake Baikal and the rates of gas evasion in the 1950s were higher than today. Major kills of pelagic fishes were recorded at that time, but have not apparently happened since. In other studies, we have documented that during the past 130,000 years, major periodic gas hydrate dissociation events (driven by climatic cycles) left distinct d13C signatures in the total organic matter in Baikal sediments (Prokopenko & Williams, 2004). Overall, we conclude that Baikal may be a significant source of methane to the atmosphere and that methane is a rather important component of the Baikal carbon cycle. It would be interesting and worthwhile to monitor the annual cycle of methane emissions from this deepest and most voluminous lake in the world.

Acknowledgements We thank the crew of R/V Vereschagin and all participants of the spring 1994 cruise on Lake

Baikal including the researchers from Limnological Institute (Irkutsk, Russia) and students from the USC Baikal Undergraduate Research Group, University of South Carolina. This study was supported by the Department of Geological Sciences, University of South Carolina and by the Samuel Freeman Charitable Trust. We cordially thank H.J. Spero for help and for the access to mass spectrometry facility at the University of California, Davis, and Eric Tappa for the CTD and other hardware used in our cruise. We thank R. Striegl and G. Kling for helpful discussions. The manuscript greatly benefited from comments and suggestions of an anonymous reviewer.

References De Batist, M., J. Klerkx, P. Van Rensbergen, M. Vanneste, J. Poort, A. Y. Golmshtok, A. A. Kremlev, O. M. Khlystov & P. Krinitsky, 2002. Active hydrate destabilization in Lake Baikal, Siberia? Terra Nova 14: 436–442. Falkner, K. K., C. I. Measures , S. E. Herbelin, J. M. Edmond & R. F. Weiss, 1991. The major and minor element geochemistry of Lake Baikal. Limnology and Oceanography 36: 413–423. Grachev, M., V. Fialkov, T. Nakamura, T. Ohta & T. Kawai, 1996. Extant fauna of ancient carbon. Nature 374: 123–124. Granin, N.G. & L. Z. Granina, 2002. Gas hydrates and escapes of gases in Lake Baikal. Russian Geology and Geophysics 43: 625–633. Hollander, D. J. & J. A. McKenzie, 1991. CO2 control on carbon-isotope fractionation during aqueous photosynthesis: a paleo-pCO2 barometer. Geology 19: 929–932. Kuivila, K. M., J. W. Murray, A. H. Devol, M. E. Lidstrom & C. E. Reimers, 1988. Methane cycling in the sediments of Lake Washington. Limnology and Oceanography 33: 571–581. Kuzmin, M. I., G. V. Kalmychkov, A. D. Duchkov, V. F. Gelety, A. Y. Golmshtok, E. B. Karabanov, B. N. Khakhaev, L. A. Pevzner, N. Ioshida, N. M. Bazhin, Y. A. Dyadin, E. G. Larionov, A. Y. Manakov, M. M. Mandelbaum & I. F. Vashenko, 2000. Methane hydrates in sediments of Lake Baikal. Geologiya Rudnykh Mestorozhdenii 42: 25–37 (in Russian). Michmerhuizen, C. M., R. G. Striegl & M.E. McDonald, 1996. Potential methane emission from north-temperate lakes following ice melt. Limnology and Oceanography 41: 985–991. Namsaraev, B. B., L. E. Dulov, G. A. Dubinina, T. I. Zemskaya, L. Z. Granina & E. B. Karabanov, 1994. The role of bacteria in the processes of synthesis and destruction of organic matter in Lake Baikal microbial mats. Mikrobiologia 63: 345–351 [in Russian].

288 Namsaraev, B. B., L. E. Dulov, E. N. Sokolova & T. I. Zemskaya, 1995. Bacterial methane produciton in the bottom sediments of Lake Baikal. Mikrobiologia 64: 411–417 (in Russian). Phelps, A. R., K. M. Peterson & M. O. Jeffries, 1998. Methane efflux from high-latitude lakes during spring ice melt. Journal of Geophysical Research 103: 29029–29036. Popovskaya, G. I., 1967. Baikal phytoplankton and its role in the production of autochthonous organic matter. In Galazyi, G. I. (ed), Cycling of Matter and Energy in Lakes. Nauka, Moscow: 216–222 (in Russian). Prokopenko, A. A. & D. F. Williams, 2004. Deglacial methane emission signals in the carbon isotopic record of Lake Baikal. Earth and Planetary Science Letters 218: 135–147. Rudd, J. W. M., Furutani, A., Flett, R. J. & R. D. Hamilton, 1976. Factors controlling methane oxidation in shield lakes: the role of nitrogen fixation and oxygen concentration. Limnology and Oceanography 21: 357–364. Rudd, J. W. M. & R. D. Hamilton, 1978. Methane cycling in a eutrophic shield lake and its effects on whole lake metabolism. Limnology and Oceanography 21: 357–364. Shimaraev, M. N. & N. G. Granin, 1991. On stratification and convection mechanism in Baikal. Doklady AN 321: 381–385 (in Russian). Shimaraev, M. N., V. I. Verbolov, N. G. Granin & P. P. Sherstyankin, 1994. Physical Limnology of Lake Baikal: A review. Baikal International Center for Ecological Research, Irkutsk-Okayama. Striegl, R. G. & C. M. Michmerhuizen, 1998. Hydrologic influence on methane and carbon dioxide dynamics at two

north-central Minnesota lakes. Limnology and Oceanography 43: 1519–1529. Striegl, R. G., P. Kortelainen, J. P. Chanton, K. P. Wickland, G. C. Bugna & M. Rantakari, 2001. Carbon dioxide partial pressure and 13C content of north temperate and boreal lakes at spring ice melt. Limnology and Oceanography 46: 941–945. Van Rensbergen, P., M. De Batist, J. Klerx, R. Hus, J. Poort, M. Vanneste, N. Granin, O. Khlystov & P. Krinitsky, 2001. Sublacustrine mud volcanoes and methane seeps caused by dissociation of gas hydrates in Lake Baikal. Geology 30: 631–634. Votintzev, K. K., 1961. Hydrochemistry of Lake Baikal. Nauka, Moscow (in Russian). Votintzev, K. K. & G. I. Popovskaya, 1964. On the condition of Melosira baicalensis (K. Meyer) Wisl., sinking into deep waters of Lake Baikal. Doklady AN SSSR 155: 673–676 (in Russian). Votintzev, K. K., A. I. Mesheryakova & G. I. Popovskaya, 1975. Cycle of organic matter in Lake Baikal. Nauka, Novosibirsk (in Russian). Votintzev, K. K., 1990. Oxygen regime as an indicator of vertical water exchange in Lake Baikal. Doklady AN 310: 964–968 (in Russian). Vykhristyuk, L. A. 1980. Organic Matter of Bottom Sediments of Lake Baikal. Nauka, Novosibirsk (in Russian). Welch, H. F., J. W. M. Rudd & D. W. Schindler, 1980. Methane addition to an arctic lake in winter. Limnology and Oceanography 25: 100–113.