Hydrobiologia 367: 163174, 1998. 163c
1998 Kluwer Academic Publishers. Printed in Belgium.
Oxygen concentration profiles in sediments of two ancient lakes: LakeBaikal (Siberia, Russia) and Lake Malawi (East Africa)Patrick Martin1, Liba Granina2, Koen Martens1 & Boudewijn Goddeeris11 Royal Belgian Institute of Natural Sciences, Freshwater Biology Section, rue Vautier 29, B-1000 Brussels,Belgium2 Limnological Institute of the Siberian Division of the Russian Academy of Sciences, Ulan-Batorskaya 3, Irkutsk664033, Russia
Received 6 August 1997; in revised form 6 February 1998; accepted 2 March 1998
Key words: oxygen microprofiles, sediment, ecological segregation, Lake Baikal, Lake Malawi
Oxygen concentration profiles have been measured with microelectrodes in sediments of two major ancient Riftlakes: Lake Baikal (Eastern Siberia) and Lake Malawi (East Africa), along depth transects in the constitutive basinsof the lakes including the deepest point, 1680 m, in Lake Baikal. Sediment oxygen penetration depths (SOPs)display very different patterns in the two lakes. In Lake Baikal, SOPs are variable, show no significant relationshipwith bathymetric depth and are surprisingly deep on Akademichesky ridge (> 50.0 mm), emphasizing the distinctivefeature of this region in the lake. While the Selenga river is an important source of eutrophication, the similarityof SOP-values in the Selenga shallow with those of most other sites suggests either a dilution of organic materialby allochthonous matter, or a strong south-to-north transport of particles. In Lake Malawi, available oxygen isrestricted to a maximum of three millimetres of the sediment, and there is a negative relationship with bathymetricdepth, as a result of a steady decline of oxygen concentration with depth through the water column. Amongst thefew parameters known to affect SOPs, the oxygen consumption by the sediment seems the most significant in bothlakes. SOP-values furthermore confirm differences in the trophic status of Baikal and Malawi, respectively. Theimportance of oxygen as a factor likely to create ecological segregation for benthic organisms is discussed. LakeMalawi offers possibilities of bathymetric segregation but no vertical segregation in the sediment. In contrast, nobathymetric segregation related to oxygen is possible in Lake Baikal, but vertical segregation in the sediment isvery likely.
Lake Baikal, situated in the Great Eastern Siberian Riftand Lake Malawi, lying in the western arm of the EastAfrican Rift Valley, have many features in common asa result of their tectonic origin. Both are very large anddeep (31 500 km2 and 30 800 km2, maximal depth of1 637 m and 785 m, respectively; Kozhov, 1963; Mar-tin, 1994; Ribbink, 1994) and their location in a stillactive graben trough has permitted their preservationover time despite continuous infilling by sediments.Both lakes are truly ancient or long-lived (Gorthner,1994) (2530 Ma and > 2 Ma for Baikal and Malawi,
respectively; Martin, 1994; Ribbink, 1994) and theyhold very diverse endemic faunas.
Due to their own respective position, the twolakes are subject to totally different climatic regimeswhich strongly affect their physics and chemistry. LakeMalawi is meromictic; there is a density discontinuityaround 230 m depth, mainly determined by salinity,sufficient to maintain a perennially deep, oxygenlesshypolimnion (Beauchamp, 1953; Patterson & Kachin-jika, 1995). In Lake Baikal, there is deep-water renewalbelow the dimictic upper 250 m layer (Kozhov, 1963;Weiss et al., 1991; Shimaraev et al., 1993; Hohmannet al., 1997a, b, Killworth et al., 1997). Water circu-lation carries oxygen to the deepest point and makes
Figure 1. A Map showing the location of Lake Baikal and the stations sampled in 1991 (dots), 1992 (triangles) and in 1994 (stars). B Mapshowing the location of Lake Malawi in the southern hemisphere, the two transects sampled in 1994 and the approximate extent of anoxic,hypolimnetic water (shaded area).
the abyssal area of the lake habitable for metazoanorganisms.
Knowledge of the distribution of oxygen in lakesBaikal and Malawi takes on a particular importancegiven the exceptional biological context of these waterbodies. In deep, ancient lakes, oxygen is thought to cre-ate possibilities for ecological segregation, both bathy-metrically and within the sediment. Particularly in thelatter case, a sediment-depth segregation can createpopulations of benthic organisms with a dumb-bellstructure and this may lead to parapatric speciation(Martens et al., 1994).
To date, the distribution of oxygen in the watercolumns of lakes Baikal and Malawi is well document-
ed (Baikal: Tolmachev, 1957; Kozhov, 1963; Votintsev,1990; Weiss et al., 1991; Liebezeit, 1992; Shimaraevet al., 1996; Hohmann et al., 1997a, b; Killworth etal., 1997; Malawi: Beauchamp, 1953; Eccles, 1974;Patterson & Kachinjika, 1995), but data on the distribu-tion of oxygen in the sediment are scarce for both lakes(Martin et al., 1993a, b), because oxygen microelec-trodes were only recently introduced into fresh waterresearch (Sweerts et al., 1986; Sweerts, 1990).
Oxygen measurements in sediments of Lake Baikaland Lake Malawi were first published by Martin et al.(1993a, b). Since then, several new expeditions to theselakes enabled us to add to our data, so that a survey ofthe distribution of oxygen in the sediment can now be
presented on a lake-wide scale, i.e. from the shallowestto the deepest oxygenated areas (including the deepestpoint in Lake Baikal). The possible influence of oxy-gen on the segregation of organisms in these lakes isdiscussed.
Material and methods
Three transects were sampled during the August 1992expedition of the R/V Vereschagin on Lake Baikal,ranging from 23 to 820 m depth, these being repre-sentative of the three constitutive basins of the lake(Martin, 1994; Figure 1A). In 1994, an additional tran-sect was sampled in the widest part of the lake, in thecentral basin, from Barguzin Gulf to the deepest point(1 680 m, as measured by the echosounder), includingfour stations in, or on the slopes of, the subaqueous ele-vation called Akademichesky ridge, which marks theboundary between the central and northern basins (Fig-ure 1A). These data supplement the original databasebuilt from sampling three transects in the Selenga shal-low during 1991 (Martin et al., 1993a), a submergedplateau which separates the southern basin from thecentral one (Figure 1A).
In Lake Malawi, two transects, stretching out fromthe Malawi shore to the centre of the basin, were inves-tigated (Figure 1B). Samples were taken from depthsranging from 0.5 m to 140 m (transect 1: a few milessouth of Kambiri Point) and from 0.5 m to 300 m(transect 2: directly off Nkhotakota), hence below theanoxic permanent stratification in the latter case. Ear-lier measurements from the extreme north of the lakehave already been published (Martin et al., 1993b).
Sediment samples were taken with a light Reineckbox corer on Lake Malawi (c. 100 kg) while in LakeBaikal, a heavier oceanographic version (c. 1000 kg)was deployed. Subsamples were extracted by meansof a Perspex tube for oxygen measurements (46 mminner diameter, 120 mm long). Tubes were tightlysealed under water in order to avoid trapping air bub-bles inside, and were either immediately immersed intoa thermostatised water bath (Lake Baikal) or kept indark conditions, at ambient temperature, awaiting sub-sequent measurements (1 to 5 hr later; Lake Malawi).
The stopper was removed just before measurements,after a period of stabilisation in the bath.
Oxygen microelectrode measurements
Oxygen microprofiles were measured as described byMartin et al. (1993a). For the second set of measure-ments (1992 and 1994), however, a Clark-style micro-electrode with a guard cathode was used. The absenceof drift, resulting from this improvement (Revsbech,1989), allowed easy identification of the 0% response(representing the oxic-anoxic interface) because thesignal remained at the same stable value when theanoxic sediment was reached, at whatever depth thesensor was sunk.
Distinction of the thickness of oxidized layer
In Lake Baikal, there is a clear colour zonation ofundisturbed sediment cores. The depth of the orange-brown to grey change in the sediment was used as thedepth of the oxidized layer. Jrgensen & Revsbech(1989) showed that such a procedure is valid in mod-erately reducing sediments where Eh is of limited use.In Lake Baikal, the two parameters are clearly related(Leybovich, 1983). In Lake Malawi, a visual distinc-tion of an oxidized layer was not possible.
Factors governing penetration depths of oxygen intosediment
The characteristic penetration depth of oxygen intosediments by molecular-scale diffusion is:
h = 2Ds
where h is the depth of penetration of oxygen intothe sediment (= SOP), D
the whole diffusion coeffi-cient in the sediment, C
the oxygen concentration atthe sediment surface, the porosity and J the oxy-gen consumption per unit area of the sediment surface(Revsbech et al., 1980a; Revsbech & Jrgensen, 1986).
In practice, the depth of the oxic zone dependslargely on the total rate of oxygen uptake of the sedi-ment and thus on the sedimentation of organic matter(Revsbech et al., 1980a; Reimers & Smith, 1986b;Jrgensen & Revsbech, 1989).
Figure 2. Oxygen concentration profiles measured in various depthsin Lake Baikal. A northern basin, B Akademichesky ridge, C dimictic zone of central basin (selected profiles), D abyssal zone ofcentral basin (selected profiles), E southern basin (as for Selengadelta, see Martin et al., 1993a).
Table 1. Gazetteer and characteristics of the stations investigated in Lake Baikal during the 1991, 1992 and 1994expeditions: sediment oxygen penetration depths (SOP), thickness of oxidized layer (OL) and temperature of thesediment; concentration of oxygen in near-bottom water (25 mm above the sediment).
Station Date Co-ordinates Depth SOP OL Tsed O2(m) (mm) (mm) (C) (mol l1)
Northern basin1 08/08/92 545903000 N 1094000500 E 30 9.5 10 6.6 3662 08/08/92 545903000 N 1093903000 E 70 8.1 43 3.9 3903 08/08/92 545901800 N 1093900300 E 105 4.6 12 3.9 3274 08/08/92 545902400 N 1093800000 E 225 5.5 21 3.7 3165 07/08/92 545801500 N 1093502400 E 485 17.0 29 3.7 3766 07/08/92 544400000 N 1091800000 E 820 > 50.0 70 3.6 306Akademichesky ridge7 11/08/94 533105600 N 1075502900 E 275 >44.5 >340 3.7 4218 09/08/94 533104300 N 1085405800 E 310 >42.5 3.6 4389 11/08/94 532801200 N 1075205400 E 415 >38.0 180 3.5 41310 11/08/94 533504200 N 1075202200 E 580 >49.5 130 3.8 391Central basin11 11/08/92 524202400 N 1073300000 E 33 9.0 33 7.5 30912 11/08/92 524204500 N 1073303000 E 65 5.0 20 6.4 25413 11/08/92 524400000 N 1074000000 E 110 14.5 32 3.6 32814 10/08/92 524701500 N 1074800000 E 240 3.0 23 4.2 29515 10/08/92 524600000 N 1074700000 E 420 13.8 23 3.9 31316 10/08/92 524604800 N 1074203000 E 820 6.1 7 3.5 26617 07/08/94 532600500 N 1084202100 E 200 1.1 1 29618 07/08/94 531803600 N 1082405500 E 810 3.6 6 3.6 36419 08/08/94 532303400 N 1083603000 E 1250 1.1 1 4.4 25320 12/08/94 532504600 N 1080504400 E 1365 4.8 19 3.6 38521 08/08/94 532502200 N 1083103400 E 1410 17.3 29 3.6 37622 06/08/94 530500200 N 1072805300 E 1665 4.6 5 3.3 33023 12/08/94 530901200 N 1074801500 E 1680 7.4 46 3.3 394Selenga delta24 06/09/91 521800800 N 1061205600 E 18 5.5 18 13.7 31325 05/09/91 521903700 N 1060900300 E 79 16.5 45 6.2 33126 04/09/91 520603500 N 1060903600 E 22 6.0 15 13.9 29327 04/09/91 520600700 N 1060505400 E 40 6.5 20 9.7 35428 04/09/91 520602500 N 1060404900 E 60 23.0 50 5.2 42329 04/09/91 520603600 N 1060304800 E 84 10.5 40 5.0 34330 12/08/92 522401800 N 1063003000 E 23 5.0 14 8.5 27431 05/09/91 522400800 N 1063005000 E 25 0.6 1 13.0 6532 05/09/91 522403200 N 1062900500 E 39 4.5 18 8.9 33033 05/09/91 522501000 N 1062604600 E 65 11.0 27 6.9 34534 05/09/91 522500700 N 1062500400 E 75 15.0 30 7.1 346Southern basin35 15/08/92 520502900 N 1055501500 E 245 4.7 7 5.7 28436 15/08/92 520605000 N 1054903600 E 430 2.0 30 3.7 23437 13/08/92 521001800 N 1055000000 E 810 21.3 28 3.7 335
Figure 3. Relationship between sediment oxygen penetration depthand thickness of the oxidized layer of the sediment. The open sym-bols refer to stations where the anoxic layer was not reached (seetext) and which were omitted for computing the correlation coeffi-cient (r = 0.669, P 0.05, n = 32; Pearson PMC). The straight lineindicates the significant linear regression (y = 1.54x + 9.01; P> 0.8),extrapolated for x-values out of the scatter plot and flanked by its95% confidence intervals.
Most oxygen microprofiles in Lake Baikal (Figure 2)have a sigmoid shape, which is typical for sedimentsoutside of photosynthetic influence (Revsbech et al.,1980b; Revsbech & Jrgensen, 1986; Sweerts, 1990),except for profiles measured in Akademichesky ridgestations, which show much variability in oxygen pen-etration depths. SOP-values in this lake (Table 1) varygreatly: they range from 0.6 mm (Selenga delta) tomore than 50.0 mm (Akademichesky ridge and north-ern basin). In addition, oxygen profiles at Akademich-esky ridge sites can have peculiar shapes, as sud-den increases in oxygen concentrations are noticedat depths where values would have been expected todecrease (stations at 415 m and 580 m, near 35 mm;Figure 2).
Overlying waters of Lake Baikal are constantlywell oxygenated: oxygen concentrations measured at25 mm above the sediment are never lower than230 mol l1 (390440 mol l1 on Akademicheskyridge) (Figure 2, Table 1), except for one station near
the Selenga delta (65 mol l1, 25 m). The latteranomaly, caused by local enrichment of sediment withorganic matter, was discussed by Martin et al. (1993a).
As a rule, there is no relationship between SOPand bathymetric depth (r
0, P 0.05, n = 32, Pear-son Product Moment Correlation). In contrast, SOPis positively correlated with the thickness of the oxi-dized layer (Figure 3). Figure 3 and Figure 4 show aclear distinction between Akademichesky ridge sitesand the station at 820 m in the northern basin, and allother stations. They fully confirm the special natureof these sites, as further suggested by the pattern ofoxygen profiles.
There is a statistically significant difference in meanvalues among each basin, Akademichesky ridge andSelenga delta (P< 0.0001; ANOVA), and a multiplecomparison procedure (Newman-Keuls test) allows tostatistically isolate the Akademichesky ridge from allother sites (P< 0.05; Figure 4).
While most oxygen profiles in Lake Malawi are typ-ically sigmoid (Figure 5A, B), there is a rapid lineardecrease of oxygen concentrations in overlying watersat 200, 260 and 300 m (Figure 5B). As these sta-tions are located below the oxic-anoxic boundary, zero-values of oxygen are expected. It is not to be exclud-ed that the pronounced imbalance of oxygen betweenthe atmosphere and the samples caused c...