Two Antarctic desert lakes

TWO

ANTARCTIC

Charles R. Goldman,” Dcpartmcnt

David

of Zoology,

DESERT

LAKES’

T. Mason,” and John E. Hobbie4

University

of California,

Davis

9~5616

The optical and biological propertics of ice-covcrcd Lakes Vanda and Bonncy, Antarctica, were observed during two austral summers. The lakes arc among the clearest known, with extinction coefficients as low as 0.031 for blue light in Lake Vanda and 0.069 for green light in Lake Bonney. Temperature and conductivity measurements in a shallow lake on Cape Evans, Antarctica, suggest a simple freezing-out process as the possible origin of the monimolimnionic salt concentration in Lake Vnnda. The lakes have a variety of living organisms but low standing crops and production rates. Many of the organisms arc facultatively heterotrophic and may use dissolved organic substances for maintenance in the dark half of the year. INTRODUCTION

On the west side of McMurdo Sound there is a scrics of deep valleys left free of ice by the recession of the Antarctic continental ice mass. Because of the lack of ice and snow cover they are called “dry” valleys. On the floors of these valleys are several lakes replenished annually by small amounts of water from direct precipitation and from sporadic summer melting of the glaciers on adjacent highlands. During the austral summers of 1961-1962 and 1962-1963, we made a number of trips to two of these lakes: Lake Vanda (77” 32’ S lat, 161” 33’ E long) in Wright Valley, and Lake Bonney (28 km southeast of Lake 1 WC wish to acknowlcdgc the National Science Foundation U.S. Antarctic Rcscarch Program Grants G-18020 and G-23868. We are grateful to K. C. Green and Drs. Brian J. B. Wood and M. S. Goldman for their assistance and comments in preparation of the data and manuscript. Dr. R. W. Bachmann provided valuable counsel and a laboratory callibration of the unclcrwatcr light mctcr. Drs. R. A. Ragotzkie and G. E. Likens collcctcd the water used in the spectrophotometric analysis, lent the Scott albcdomctcr, and exchanged data with us. Prof. H. Skuja of the Institute of Botany, Uppsala, Swcdcn, kindly identified some of the algal material. WC also wish to thank mcmbcrs of U.S. Navy Task Force 43, who assisted in the logistics of this work. 2 Prcscn t address: Institute of Ecology, University of California, Davis 05616. 3 Prcscnt address: Dcpartmcnt of Biology, Western Washington State College, Bellingham 98225. 4 Present address : Dcpartmcnt of Zoology, North Carolina Stato University, Raleigh 27607.

Vanda ) in Taylor Valley. Both lakes arc permanently frozen except for occasional melting around their margins during midsummer. Lake Bonney differs from Lake Vanda in several morphometric parameters ( Table 1) and has an important tributary glacier at its west end. The purpose of the field study was to measure primary productivity and ancillary biological and chemical conditions in these lakes, and to determine the character of the solar radiation penetrating the ice. Armitage and House ( 1962) visited Lakes Vanda and Bonney in the summer of 1960-1961 and made preliminary limnological observations. Their report of unique thermal and chemical profiles led several investigators to study these lakes further (Wilson and Wellman 1962; Angino and Armitage 1963; Angino, Armitage, and Tash 1963, 1964, 1965; Goldman 1964; Hoare et al. 1964; Ragotzkie and Likens 1964). The geology of this region has been discussed by McKelvcy and Webb ( 1961) and Nichols ( 1962). A nearby saline lake in the upper Wright Valley was examined by Meyer et al. ( 1962), and by Tcdrow, Ugolini, and Janetschck ( 1963). The organisms living in the dry valley lakes have received little attention. Meyer et al. ( 1962) reported that they isolated no microorganisms from water from a depth of 60 m in Lake Vanda, and only four bacteria, a yeast, and a yeastlikc organism were obtained from littoral meltwaters of Lakes Vanda and Bonncy. The littoral regions of

295

296

CIIARLES

R. GOLDMAN,

DAVID

TABLE 1. Morphometric table based on a preprint of U.S. Geological Survey maps of the icefree valley arecc. (Scale 1 : 100,000)

T. MASON,

AND

JOIIN

E. HOBBIE

Goldman (1963, 1964). A lo-cm-diam Sipre corer was used to penetrate the 34-m ice cover, and a 5cm-diam Plexiglas sampler of the Van Dorn type was constructed Lake Bonney to collect the water samples. Buffer caTaken in three steps: from southMax pacity titrations were performed (Becklength: western corner of western half, to man model N pII meter ) in the field soon 7.10 km the eastern side of western half; after sample collection. through the channel along the northwestern shore of the eastern Phytoplankton counts were made with half; and to the mouth of the small an inverted microscope on samples fixed stream entering the eastern half with Lugol’s acetic acid solution. from the east. Thermal coefficients of electrical conArca : Western half 1.09 kma ductivity were determined for 13 depths in Eastern half 24.08 2.99 km* Lake Vanda and for three in Lake Bonney. km” Simultaneous readings of conductivity ( InMax breadth, perpendicular to a length axis: dustrial Instruments model RC 16 B 2 con0.85 km (both eastern and western) ductivity bridge, KCI-calibrated) and temz (Calc. from Angino, Armitagc, and Tash perature (standard mercury thermometer 1965) = 18.7 m with 0.1-C divisions) were made as the Shoreline length: 16.5 km samples were slowly warmed from 5 to 23C. The thermal coefficient of conducLake Vanda tivity was taken as the slope o,f the straight Max Taken from inflows at the eastern line visually fitted to the plot of conductivlength: and western ends with three 5.65 km straight lines rcflcxing at the southity vs. tempcraturc. The average values of ern peninsula and the northwestern 0.026/C for Lake Vanda and 0.025/C for deltaic fan. Lake Bonney ( omitting the high 5-m value ) Area: 5.21 km2 were used for conversion of the field Max breadth, perpendicular to a length axis: measurements data. Field conductivity 1.4 km were made with a KCl-calibrated undcrShoreline length: 13.9 km water probe and oscillator (Whitney Instruments ) . Because of the low water temperatures, the conductivities were corboth lakes show occasional patches of a rected to OC. No corrections were made cyanophycean matted growth closely re- for the small pressure effects in the lakes. sembling Phormidium sp., and several The irradiance photometer (Whitney Inmicrometazoans have been isolated and struments) had surface and underwater identified from these patches (Dougherty units, each with a Weston Photronic photoand Harris 1963). Armitage and House cell shielded by a perforated brass disk ( 1962) found the ciliate protozoan Nass~Zu over a milk-glass diffusing plate. Thcsc sp., the rotifer Philodina sp., and a tardishields reduced the light intensity incident grade in the meltwater at the edge of Lake on the photocells so that the electrical outVanda, only the rotifer at Lake Bonney, put was proportional to light. Red, green, and a ChZoreZZa-like organism from one of and blue filters (Schott RG 1, VG 9, and the lakes. Algae were isolated from the BG 12,2-mm thick) were also used on both littoral regions of the dry valley lakes by photocells. For the lowest light intensities, IIolm-Hansen ( 1964). Likens ( 1964) has the instrument had an amplifier (linearity described an interesting algal growth pat- was calibrated at low temperatures). When using this amplifier and the red filter, a tern in the littoral zones of Lake Vanda. small amount of energy was still detectable METHODS between 8 and 26 m in Lake Vanda. Because little red light penetrated the ice, Most of the physicochemical and biological methods have been reported by because it was likely that the red filter

TWO

ANTARCTIC

DESERT

297

LAKES

TABLE 2. Extinction co@ficients of Lake Bonney and the mixed layer of Lake Vanda for two filters and no filter, and per cent of surface light reaching the water through the ice. Note: solar rwon is 1315 hours local time Filter Lake

Date

Solar

time

Bonney Vanda

14 5 25 14

Jan Jan Jan Feb

1545 1515 1115 1410

Bonney Vanda

14 5 25 14

Jan Jan Jan Feb

1545 1515 1115 1125 1410 1915 0945

26 Feb

also passed some green light (Tyler 1959)) and because the extinction coefficient was similar to that measured for green light, the red readings were discarded. Albedo values were dctermincd with a Scott albcdometer (Scott 1964) and with an Eppley pyranomcter. A spectrophotometer ( Beckman DU ) with a photomultiplicr tube and a IO-cm cuvette was used to measure the absorbance of unfiltered water using doubly glassdistilled water as a blank. The readings of the underwater thcrmistor unit (Whitney Instruments) could bc estimated to1 O.OlC, and were probably precise in measuring temperature differences at this level, although the accuracy of temperature values was nearer 0.05. The thermistor unit was calibrated at the McMurdo laboratory against a National Bureau of Standards certified mercury thermometer. Photosynthesis was mcasurcd by the 14C method as modified by Goldman ( 1963). Light and dark bottles (125 ml) were incubated in situ after 4.05 &i Naa14COR had been added to each sample bottle, Piltercd controls were included to detect any inorganic precipitation or filter adsorption of “C (Goldman and Mason 1962).

None

type

Green

Blue

Extinction coefficient of water 0.141 0.069 0.049 0.055~ 0.042 0.060 0.041 0.058

21 14 20 19 18 16 18

% penetrating 4 18 24

Ice

(cm)

0.079 0.038 0.035 0.031

ice 5 27 32

21

40

26

41

ca. 400

346 346, 346 308

Sample collection and filtration were performed at the lake in a darkened tent. With similar methods, heterotrophic activity was mcasurcd by the addition of uniformly 14C-labeled acetate and n-glucose ( New England Nuclear) to sample bottles. Five &I: of the acetate and 2..5 ,&i of glucose were added to 125-ml Pyrex bottles of lake water in the expcrimcnts. The radioactivity of the filters (2.5-cm HA Millipore@) was counted to 5,000 coants with a thin (Micromil) window, gas-flow counter ( Nuclear-Chicago ) . To eliminate problems of self-absorption and sample gcomctry and to avoid the errors associated with the USCof BaC03 standards, both the counting equipment efficiency and the absolute activity of the isotopes were determined by gas-phase counting. A National Bureau of Standards Na2,COi7 source was used to standardize the gasphase determinations. IXESULTS

AND

DISCUSSION

Optical conditions of ice and water The albedo of the ice of Lakes Bonney and Vanda changed throughout the summer as the ice surfaces were altered by insolation and ablation. In late November

298

CHARLES

R. GOLDMAN,

DAVID

T. MASON,

AND

JOIIN

E. HOBBIE

% SURFACE LIGHT

8 0 8 0” 8 0” m

0 0 0 0

0

0”

0”

0

0 0

0”

l

0

n

a

l .

0 q

0 NO FILTER

Cl cl

0 BLUE 40

0 0 08

. GREEN

0

0” l

q

OO. 0” 0”

0

n

0

l

0

0

50

0

.

0

VANDA

0

l

0

0

0 0

m

0

0

0

0

0 O

60 I

0.2

I

0

o

0

oooo

I111111

0.3 0.4

I

0.6

5 Jan

0 0

I

I

2

1

34

.

I

1963

IIll

6

I

IO

I

20

I

3c

as per cent of surface light, Light penetration in Lakes Bonney and Vanda, Antarctica, FIG. 1. Schott BG 12 (420 m,u) and VG 8 (540 mp) were used measured with an underwater photometer. for the filtered light mcasuremeits.

1962, the surface ice of Lake Vanda had a gently rolling relief of several decimeters, with a superimposed microrelicf in the form of small cusps several centimeters high. The ice was blue, highly polished, and had an albedo of between 20 and 40% (November albedo observations by Ragotzkie and Likens, personal communication). Signs of modification by so,lar radiation were small strings of bubbles and small platelets in the top decimeter. The next phase of the ice cycle occurred in December and January. The surface became highly irregular, with pockets of meltwatcr up to 1 m deep where the insolation was absorbed by concentrations of wind-blown dirt. At this time much of the ice was porous, and interstitial water filled

any hole drilled into it. The albedo rose to 57% as the top few decimeters of ice became white. Finally, during February the ice surface began to return to the condition observed in the spring and had an albcdo of 3234%. This was caused by the continued ablation of the surface at the same time that the decrease in insolation an d air temperature prevented new white ice from forming. Lake Bonney showed similar seasonal changes in ice surface, but the ice was whiter and the relief more abrupt than at Lake Vanda. The surfaces of both lakes arc typical of old ice sheets of the arctic. Anquissaq Lake, a permanently frozen lake in Greenland (Barnes 1960)) has a similar cover of eroded ice crystals and rolling

TWO

LAKE

0.8 -

ANTAHCTIC

VANDA “9 m ‘50

'55 060

I-+ 0,6\ n0 $0.4II

v

0.1 -O\ 0"305m

LAKE BONNE Y

O\-

300

400

500

600

700

800

mv FIG. 2. Absorbance from 300 to 800 mp of Lakes Vanda and Bonney water taken at different depths, as measured with a Beckman DU sncctrophotometer using lo-cm cells.

topography. Apart from the surface layers, the ice of both lakes was free from air bubbles; hence a large fraction of the incident surface light penetrated the extremely clear ice and entered the water (Table 2). The light transmission of the ice approaches that of distilled water, a condition which, although apparently rare in nature ( Lyons and Stoiber 1959), is discussed by Ruttncr (1963,) in relation to the work of Saubcrer. In Lake Vanda, the shifts in albedo increase the fraction of surface light entering the water during October and November and again in March and April, thus reducing the amplitude of seasoaal variation of light in the lakes. On 14 February 1963,, unfiltcrcd light beneath the ice was also measured at different times of the day ( Table 2) ; there was a decrcasc at 1915 hours that may be explained by the changes in albedo with the changing altitude of the sun. At 1135 hours, the albcdo was 3740%, at 1345 it was 38%, and at 1845 it was 56%. After passing through the ice, the remaining light penetrated to great depth (Fig. 1). Unfiltered light was measurable to the bottom of Lake Vanda, some 61 m below the ice surface. The light penetration

DESERT

LAKES

299

curves reflect the physical division obvious from tempcraturc data, showing a wellmixed intermediate layer and a strongly stratified bottom layer. The change in slope of the curves near the bottom indicates that the deep layers absorbed or scattered the unfiltered and blue light more, and the green somewhat less, than did the upper layers. When compared with distilled water, the Lake Vanda water had an absorbance close to zero for all depths above 55 m ( Fig. 2) ; however, at 55 and 60 m the absorbance at all wavelengths increased, especially in the deep blue. The upper levels of Lake Vanda arc thus among the clearest waters in the world. This results from the absence of inflows of colloidal material or humic substances, from the low biological productivity, and from the lack of wind action, Utterback, Phifcr, and Robinson ( 1942) summarize extinction coefficient values for Crater Lake, Oregon as 0,033 in the blue (Schott BG 12) and 0.060 in the green ( Wratten 61). Lake Vanda has coefficients of 0.031 and 0.058 for the same colors. The measurements from Lake Bonney show that close to 1% of the surface light reached the bottom (Fig, 1). When Lake Bonney water was compared with distilled water (Fig. 2)) the absorbance was close to zero in the upper layers. However, the water from 30 m absorbed strongly in the lower wavelengths, indicating some color. This could be the result of concentration of an earlier lake, or o,f accumulations of detritus and dissolved organic matter in the bottom waters. Glacier meltwater entering Lake Bonncy may also reduce light penetration somewhat. In interpreting the light penetration data from the field, it is pertinent to consider that the angular distribution of light under a perfectly diffusing ice cover will approach

where iZo is the downward-directed radiance at a point x meters below the ice, in water having an attenuation per meter of a. The angle from the zenith is 8, while I0 is the intensity of light emitted in any downward direction by the ice, The shape of

300

CIIARIXS

R. GOLDMAN,

DAVID

PER CENT OF SURFACE ENERGY 0.3 0.4 I

I

0.6 0.8 1 I,,,,

2 I

,I

4 I

6 IIIII

8

10%

M 0 1963

zo-

C 14 FEB II 26FEB

-20

FIG.. 3. Per cent of surface energy vs. depth in Lake Van& and Lake Bonncy ( 14 Jan 1963). Our extrapolation based on 1963 photocell mcnsurements and 1961 bolometer measurements of Wilson and Wcllman ( 1962).

this ideal distribution is closely approximated by that of an overcast sky (Tyler and Preiscndorfer 1962)) and probably still more closely by the sort of ice cover present in midsummer at Lake Vanda. Our irradiance measurements yield diffuse attcnuation functions for this particular distribution of light energy and therefore may not bc strictly comparable to the vertical extinction coefficients found in open waters. The small seasonal changes in light cxtinction coefficients ( Table 2) illustrate the almost complete isolation of Lake Vanda water from its surroundings. Temperatures and water chemistry also changed little throughout the year. Consequently, the effects of solar radiation upon heating must be small, while heat flux From the bottom, a small amount of air-borne dust, and the occasional but rare inflow from the Onyx River comprise the remaining outside influcnces upon Lake Vanda. The percentages of surface energy reaching various depths in Lake Vanda ( Fig. 3) were calculated by the method of Bachmann

T. MASON,

AND

JOHN

E. IIOBBIE

and Goldman ( 1965). Curves other than for 5 January 1963 are based on a computation of the percentage of energy at 10 m and the assumption of a seasonally stable energy extinction, The changes in the 1963 readings ( Fig. 3, curves A-D) are a result of the gradual decrcasc of albcdo previously discussed. The. extinction coefficient of the total energy (bolometer measurements of Wilson and Wellman 1962) is close to that of the blue and green wavelengths ( photometer readings ) , indicating that essentially all wavelengths except blue and green are absorbed by the ice. Intensive work on Lake Vanda has produced two theories for the origin of the unique temperature distribution originally described by Armitage and House ( 1962). One theory, that in which Wilson and Wellman ( 1962) postulated solar radiation as the sole heat source, must now be considered incomplete for the following reasons : First, solar radiation would give a different heat distribution from that observed in the lake, as shown for Lake Bonney by the detailed investigation of Hoare et al. ( 1964). Second, the layer of deeply colored water lying in the depths of I,ake Vanda has such a high absorption of light that much less than 1 cal cm-2 year-l is available at the bottom of the lake, where the highest temperatures were fo,und. Finally, the heat flow from the water to the sediments suggcstcd by Wilson and Wellman was not confirmed by Ragotzkie and Likens ( 1964). On the contrary, they reported higher sediment than water temperatures, and also measured an outflow oE heat from the dcepwater sediments with a flux plate. They also measured a strong circulation in the isothermal zone (17 to 37 m ) . This circulation could help explain the upward transfer oE a large quantity of heat within the lake. Our results indicate that heating by solar radiation is important only in the upper layers of the lake and that another source of heat is necessary to explain the warm water near the bottom. Thus, in agreement with the findings of Ragotzkic and Likens ( 1964)) WC conclude that geothermal heating is acting to produce the tcmpcrature distribution in Lake Vanda

TWO

ANTARCTIC

TAULE 3. Chemical analyses of new ice and the underlying water from two lakes on Cape Evans, Ross Islard, Antarctica. The samples were taken on 1 G February 1962 --.-.-____ ---~ --

Skua Lake New ice Water Alga Lake New ice Water

Chloride ( IJlJ*‘~)

Divalent cations (as ppm Mg)

285 1,160

48.8 155

370 1,910

54.2 266

~-

DEXXT

TADLE 4. Chemical analyses of ice from a small lake frozen to the bottom; Cape Evans, Ross Mad, Antarctica. The samples were taken on 25 November 1961 Chloride (mm)

Divnlent cations (as pprn Mg)

IIardness (mm CaCO,)

Toy ice (0.0 to 0.2 m)

283

53.9

21.3

Mid-ice (0.2 to 1.0 m)

326

80.0

14.0

IIardness (mm QCO,)

13.8 45.8 5.1 20.3

with a small seasonal effect of solar heating. On the other hand, IIoarc et al. ( 1964) concluded from their detailed studies that insolation is the only source of heat in Lake Bonney. Origin of Lake Vanda zoatef The peculiar thermal propcrtics of Lake Vanda are dependent upon the interesting chemical structure of this meromictic lake. Wilson ( 1964) has argued from a model diffusional system that the bulk of the salt load existed at o,ne time in some small volume and was covered by fresh infIowing water, He suggests that a relatively sharp and definite climatic change occurred about 1,200 years ago and brought in meltwater from the Wilson Piedmont glacier via the Onyx River. Ragotzkie and Friedman (1965) prcscnted evidence of a relatively low dcuterium content in Lake Vanda (compared with seawater) and also suggest the Onyx River sources for the water of the lake. The monimolimnic salt concentration can bc explained either by the above hypothesis of the flooding of an originally more concentrated source, or by another mechanism that would force salts downward out of solution. Such a mechanism is different from the usually cited causes of meromixis in that it does not involve the introduction of material to the system, but rather of energy to separate spatially the salts into the Iowcr monimolimnion. The most obvious source of this energy is from the system itself as it cools in a thermal gradient. Cryogenic meromixis, as we shall call it, affects

301

LAKES

Bottom ( below -~

brine 1.0 m )

6,932

419

198

monimohmnctic saIt concentration by a simple freezing-out process. Several shallow lakes on nearby Cape Evans clearly exemplify this freezing-out mechanism. There, in the irregular terrain underlain by permanently frozen ground, small lakes have formed from melting ice and snow. Their ionic content is enriched by blown salt spray from the precipitation, nearby sea, leaching from the volcanic rocks of the area, and biological agents such as the skua gull (Cath.aracta skua) which bathe in some of the lakes in great numbers during summer. Depending upon the morphometry of the lake and its exposure to the wind, the circulation may be complete or only partial. At the onset of freezing, the new ice on two of these lakes contained fewer ions than the water from which it was formed ( Table 3). As freezing continued downward, an ice layer was formed in which the salinity incrcascd toward the bottom ( Table 4). A comparable situation has been recorded by David and Priestley (1914) at Green Lake on Cape Royds. In Iakcs where freezing occurs to the bottom, the ice over the deepest portion of the lake frequently buckles up to compensate for the expansion of the last water to, freeze. These upthrusts on the Iakc surface arc called “hydrolacoliths” by permafrost geologists and provide some indication of the morphometry of a frozen Iakc. During the early spring, warming of the dark rocks (Llano 1959; KcIly and Zumbergc 1961) permits a small amount of melting to occur under the ice along the margins

302

CIIARLES TABLE

~--___-

5.

11. GOLDMAN,

Laboratory

-

~--

Sample depth ( n-l 1

DAVID

measurements

T. MASON,

of conductivity

Temp (“Cl

365 395 415 441 452 476 495 540

5.0 6.6 9.4 12.6 15.0 16.9

413 429 466 509 537 565

5.1 8.3 11.7 13.9 16.1 21.1

490 544 595 631 667 745

15

4.9 9.7 14.6 18.6 21.2

857 970 1,105 1,208 1,276

20

4.9 8.8 12.8 16.8 23.2

1,274 1,420 1,578 1,733 1,997

39

5.1 8.6 11.6 15.6 18.4 21.5

1,517 1,663 1,795 1,978 2,124 2,270

42

5.9 7.9 11.0 13.8 17.3 23.1

1,978 2,082 2,261 2,426 2,628 3,005

45

5.2 7.7 11.3 14.2 18.0 19.8

2,854 3,052 3,325 3,589 3,947 4,088

5.0 8.1 --

3,730 4,051

47 .~---.

E. HOBBIE

as a function

___

of temperature ~---

Conductivity (hmhos/cm)

4.8 6.9 9.2 11.2 12.8 14.6 16.4 20.2

12

JOHN

.-n-x---

Sample

;lepth

__--_

..-

--_-

Conductivity t umhos /cm

m

Lake Vanda

4

AND

11.6 15.8 20.9

4,427 4,927 5,482

48

5.1 10.6 14.5 17.3 21.1

5,303 6,142 6,764 7,235 7,866

50

5.2 9.9 13.0 16.2 20.4

14,840 16,670 17,940 19,310 21,100

55

4.6 8.8 12.5 14.8 19.5

44,930 50,110 53,980 58,220 64,430

60

4.6 8.5 10.6 12.5 15.6 21.6

67,350 74,420 77,240 80,070 88,550 96,080

Lake Bonney

5

5.0 9.0 11.0 12.6 14.8 16.5 17.8 18.5 19.4 20.0

648 756 800 817 920 970 1,003 1,022 1,041 1,062

10

4.0 8.0 12.3 14.6 16.2 16.7 17.9 19.0 20.0

12,950 13,490 14,980 16,110 16,390 17,520 17,990 18,460 18,840

4-z 10:5 13.5 16.0 18.5 20.0

74,420 79,130 83,840 92,320 94,200 98,910 102,700

15

_

)

TWO ANTAHCTIC

DESEllT LAKES

303

FIG. 4. h4clting stages of a shallow lake at Cape Evans, Antarctica, in December 1961. A, Lake B. Melting of highly salinc ice along the bottom with temperatures completely frozen to the bottom. and conclu&ivities indicated. C. Continued melting thaws highest salinity ice along the bottom of the lake. D, Lake completely thawed with warm monimolimnion (see Fig. 5).

of the lakes before the air temperature is above freezing ( Fig. 413) . In early Dcccmber 1961, we frequently noted melt crosssectional profiles of this type. In several lakes at Cape Evans, the general rise in temperature next caused melting of the highly saline ice at the bottom and produced the profile shown in Fig. 4C. Continued insolation and melting in a sheltered location may permit the devclopment of the strong thermal gradient and the compensating salinity increase shown in Fig. 5. This lake was 1 m deep, about 20 m long, 10 m wide, and located in the lee of a large bank of drifted snow near the southern shore of the cape, which effectively shielded it from the prevailing southeast winds. A thin layer of recently formed ice (less than 5 mm thick) covered this lake when we visited it on 16 January 1962. Insolation on the cape during this month was about 1.0 ly/min during midday, and the dark lava substrate of the lake presumably absorbed most of the energy reaching it. There was a striking temperature rise with depth (Fig. 5); the water tempcraturc incrcascd from 4.4C just below the surface to 16.OC at 1 m. A concomitant increase in conductivity indicated that the thermal structure was stabilized by in-

creases in salinity with depth, and this was confirmed by the calculated density. The principal density gradient occurred between 0.6 and 0.9 m, while the change of tempcraturc was relatively small in this region. A mixolimnion may have existed down to 0.6 m at one time, but, if so, diffusion apparently obscured its lower boundary. This cryogenic mechanism of monimolimnion formation is probably confined to the coldest regions of the earth. The warm monimolimnion found in Lake Vanda may have been formed by this process as suggestcd by Armitagc and House (1962). However, the geomorphologic evidence for a relatively recent catastrophic flood ( Smith 1965) in the upper Wright Valley and the arguments based on salt diffusion (Wilson 1964 > and on dcuterium content (Ragotzkie and Friedman 1965) lend support to the suggestion od catastrophic origin of the prcscnt Lake Vanda water. Ch~mistf+y and biology Because increasing attention is being paid to the temperature coefficient of elcctrical conductivity ( Smith 196.2; Weyl1964), WC have presented our basic data in full ( Table 5). The coefficients have been

304

CHARLES

R. GOLDMAN,

DAVID

T. MASON,

AND

JOHN

6. Summary of Lake Vanda plankton counts made at 400x magnification on acid-Lug01 preserzjed water samples by inverted microscope techniques. All values are 106/liter ~~____~___ _ -~--_.._

TABLE

Date (1963)

“““p m

Coccoid blue-green algae (2-8 p)

midium (100 p)

TEMPERAT,URE("C)

0°1

LY;f&;yPhytoflagellates

E. HOBBIE

Heliozozms

5 Jan

7 9 14 24 34

0.6 1.0 0.6 0.4

1.1 0.2 0.3 0.4 0.3

-

+ -

14 Feb

40 45 50 55 60

0.3 0.3 0.3 0.2 1.0

0.2 0.1 0.03 0.2 -

0.1 0.1 0.1 0.1 0.1

3 + -

plotted as functions of depth in Fig. 6. This pattern is not simple, and undoubtedly reflects changing values of both ionic ratios and ionic concentrations. Since the molar thermal coefficient of electrical conductivity for potassium ions, d C,+/d& is 1.32 times that of sodium, potassium ions are significantly more important than sodium ions in determining the coefficient. A proportionality between the molar K : Na ratio and the thermal coefficient of conductivity would be expcctcd if these ions arc predominant in influencing conductivity. On the basis of the three values determined for Lake Bonney (Fig. S), the thermal coefficient of conductivity appears to increase sharply with depth and salt concentration in this lake. Acid titrations of Lake Vanda water (Fig. 7) show the three-layered chemical nature of the lake. Poorly buffered waters of low salinity are found in the uppermost part. A saline circulating zone with high buffer capacity constitutes the second layer; below this, the buffer capacity increases strongly and pII drops with the increasing salinity. The plateaus at ~1-16.6 suggest bicarbonate buffering, with no evidence of EI$ or organic acid buffer components. There were numerous planktonic forms, less than 20 p in size, throughout the water column of Lake Vanda. For convenience, algal counts have been grouped into broad categories in Table 6. The coccoid blue-

i.OI 0 I 1.0000

I

I

I

CONDUCTIVITY(~TI~~*&I-I x 10' at 25OC) I

I 1.0020

I

1 40 I 1.0040

DENSITY FIG. 5,. Profiles of temperature, conduction, and density in a small pond near the south short of Cape Evans, Antarctica (77” 38’ 30” S lat, 166” 24’ 00” E long). The in situ conductivity measurements were corrected to 25C using a coefficient of 0.025. The density was calculated from conductivity using the specific ion values suggested by the American Public Health Association (1960) and the assumption that the salts were predominantly sodium chloride.

green algae, which appear to be in the genus Synechocystis of Drouet and Daily ( 1956), are considered as one category. All colored flagellates-predominantly Ochromonas sp., but with numerous Chlamydomonas spp.-are grouped in a second category. The third group is composed of 100-p units of Phormidium ( Lyngbya? ) . A heliozoan ( Actinophrys? ) was present in three samples, but in numbers too low to allow statistically significant counts. No zones of high bacterial concentration were observed in Lake Vanda samples, so no attempt was made to count bacteria. Data for Lake Bonney microorganisms arc similarly condensed in Table 7. In this lake, bacteria were concentrated just below the region of maximum temperature and

TWO

thermal 0.022

0,

ANTARCTIC

DESERT

305

LAKES

coefficient

of conductivity

0.024

0.026

0.038 AT 0.035

IO-

8 : 20a t ii 30:

3 4o

0

z .-9

v) $

-o- Lake Vanda Vanda molar ratios (Angino et al)

50 I

8

Lake

Bonney

%i 60 E : 0.06

0.08

0.10

0.12

molar

0.14

ratio

Thermal coefficients of conductivity FIG. 6. K : Na data of Lake Vanda recalculated from

Green coccoid forms -- Phytoflagellates 1OP 5fi

14 Jan

5 10 15 20 25 30

1.6 0.5 0.3 0.4

0.7 1.0 2.5 0.6 0.2 0.8

0.4 2.0 0.07 0.07 0.5 1.3

0.20

for Lakes Vanda and Bonncy Angino et al. ( 1065).

TADLE 7. Summary of Lake Bonney plunkton counts made at 400~ magnification on acid-Lug01 preserved water samples by inverted microscope techniques. All values are lO’/liter

~

0.18

K: No

below the zones of highest concentrations of photosynthetic organisms. Photosynthetic flagcllatcs are distinguished from the two of coccoid forms. Again, size-groups Ochromonas sp. and Chlamydomonas spp. were the flagellates observed, while the coccoid forms have not yet been identified.

Date (1963)

0.16

Lkteria

0.9 1.2 Few 120 Few Few

plotted

along with

molar

Microscopic examination of sedimentary material from Lake Vanda revealed few idcntifiablc organic remains with the exception of small cystlikc walls apparently associatcd with the coccoid blue-green algal cells. A red-orange, filamentous aggregation adhcrcd to sampling lines left suspended for several days in Lake Vanda. This material has been found to contain an atypical form of Lyngbya musicola, two species of Chlorella (one similar to C. w&ark, one very small), an cncystcd Chlamydomonas sp., and GZoeotricha contracta. These algae, probably of benthic origin, accumulated only at the level o,f the upper part of the monimolimnion. Perhaps they were prevented from sinking by the high density water below. The dense zone of bacterial growth at 20 m in Lake Bonney and the low bottom temperatures in the lake suggest that most

306

CHARLES

Ii. GOLDMAN,

DAVID

T. MASON,

AND

JOHN

E. HOBBIE

LAKE VANDA BUFFER

CURVES

FIG. 7. Titration curves of Lake Vanda water using 0.01 N sulfuric acid; data of several titrations from various depths in the 0-10-m region ( 0 ) and the 20-45-m region (x) have each been summarized by a single line.

oE the production by phytoplankton occurs under the favorable light and temperature conditions of the upper 15 m. Below this zone, heterotrophic activity probably dominates as sinking plankters arc subjected to bacterial attack. The prescncc of from 10’) to over 10” photosynthetic organisms per liter in both lakes is comparable to arctic lakes (Rodhe 1955; Barnes 1960). Flagellated forms were relatively abundant, and either flagella or very small size are probably necessary to prevent plankton from sinking to the bottom because there is no wind-induced turbulcnce in these lakes. The suspension of organisms on pycnoclines may compensate to some extent for the lack of wind mixing in keeping planktonic forms suspended. Whether the heliozoans found in Lake Vanda have a significant grazing effect is

not known, but there appears to be a slightly lower concentration of photosynthetic organisms in Lake Vanda than in Lake Bonncy (Tables 6 and 7). Many more plankton counts would be necessary to resolve this question, however, and more intensive sampling may eventually reveal a planktonic consumer in Lake Bonney. Primary productivity in Lake Vanda was extremely low on the basis of both volume and surface units. On 14 February 1963, in situ measurements were made to a depth of 60 m ( Fig. 8). When corrected to full-day photosynthesis, they indicated a rate of carbon fixation in the light of 29 mg C m-2 day-l. Dark fixation (including adsorption) amounted to 15 mg C m-2 day-l, so that uptake attributable to photosynthesis was only 14 mg C m-2 day-l. On the basis of a 60-m euphotic zone, this is only

TWO

ANTARCTlC

0.24 mg C m-” day- I. A considerably higher rate of light fixation was recorded in late December 1961 ( Goldman 1964 ) . The lack of significant organic sedimcntation in the presence of particulate production of this order suggests three possibilities: Sinking of dead plankton may be prevented by the high density of the lower waters; mineralization of organic material may be occurring at a rapid rate; and, an important part of the fixed carbon may be released in soluble form in the course of the summer by photosynthesizing organisms of both phytoplankton and periphyton. Fixation of carbon by both bacteria and phytoplankton at the expense of an organic energy source may serve to maintain this unusual community during the months of winter darkncss. Such a process has been postulated for arctic lakes by Rodhc ( 1955), but remains to be proved. The quantitative importancc of production of extracellular products has recently been demonstrated by Fogg, Nalewajko, and Watt ( 1965) in Lake Windcrmere, and by Goldman in Castle Lake, California ( unpublished data). Once in January and once in February, dark-bottle carbon fixation exceeded that in the light by 21%. However, a significant amount of nonbiological carbon uptake was evident in comparable expcriments, where adsorption by a control sample to which formalin was added amounted to 71% of the dark assimilation in littoral water (28 December 1961, Lake Vanda ) , and to 70% of the dark assimilation in pelagic water on the following day. In both cases, the adsorption was 28% of the carbon uptake in the light. The remarkably increased level of photosynthesis ( Fig. 8) in the warm bottom waters of Lake Vanda implies that one of the primary limitations to organic production is temperature. This heightened activity is consistent with findings of an extremely high Qlo for photosynthesis in other Antarctic phytoplankton populations (Goldman, Mason, and Wood 1963). Other factors in this warm region may contribute to the high photosynthetic rate: The ionic content is much greater than, and qualitatively different from, that of the upper

DESEIIT

307

LAKES mg C mM3 doy -I

I 5

0

Id00

I 15

I IO

3600

I

20 5600

I

25 *c pm ho/cm

at

O°C

Depth profiles of conductivity, tempcraturc, and primary productivity beneath the ice in Lake Van& on 14 Feb 1863. Carbon assimilation was measured in situ with “C and is reported was also deterin mg C mW3 day-l. Conductivity mined by in situ mcasurcmcnt and is rcportecl in pmho/cm adjusted to OC. FIG.

8.

regions (Angino et al. 1965); the plankton population is different; and the organic content, as evidenced by the brown color of the water, may supply heterotrophic requirements for COa assimilation, The primary productivity measurements in Lake Bonney reported by Goldman (1964) yielded a photosynthetic rate of 31 mg C m-3 day-l to a depth of 10 m on 29 November 1961. A maximum rate of 8 mg C m-n day-l was found in the warmer zone 5 m bcncath the ice-water interface. It is probable that temperature, nutrient, and osmotic effects on photosynthesis are of much greater importance than the small difference in light pcnctration of the two lakes. From measurements using littoral water, the productivities of the shallow areas of

308

CHARLES

R. GOLDMAN,

DAVID

TABLE 8. Lake Vanda water (5 Jan 1963) from two depths with additions of uniformly labeled D-&COSe and unlabeled D-ghCOSe. The final radioactive glucose concentrations were 60 mg/ liter and 10 &i/liter; the ‘“COa was at a concentration of 50 mg/liter, and the WOs was added at 32.4 &i/liter. Incubation in situ, between 1530 and 2210 hours (local time). The initial filtration using radioactive glucose attempts to give a control on adsorption phenomena on filters and particulate matter. Values are counts per second ___.~.-. D-glucoseJ4C ___.

filtration

-

3 m

10

m

Initial

filtration

Light Dark

6.15 4.40*

6.31 5.97

After

incubation

Light Dark

5.98 6.13

9.69 8.58

Difference

Light Dark

-0.17 -t

Control

Light Dark

0.48 0.20

With nonradioactive D-glucose

Light Dark

0.66 0.71

Difference

Light Dark _.--__-~~

0.18 0.51

Inorganic

‘“CO,

uptake

---

3 in

~* Filter was only t Not significant.

3.38 2.61

wetted

with

culture

solution.

the two lakes appeared to be rather similar. The differences between the two probably arise chiefly from the extent of melting around their margins and light conditions during the incubation. Values ranging from 0.5 to IO mg C m-3 day-l were measured in 1,ake Vanda, whereas the Lake Bonncy littoral photosynthesis averaged 18 mg C m-3 day-l. Important amounts of fixation may occur in benthic producers in the marginal regions oE the lakes. As considerable light reaches the cntirc bottom, the whole floor area is probably available to photosynthetic organisms. The filamentous algal material found on lines at the top of the monimolimnion in Lake Vanda suggests an abundant benthic population at or above this depth. During the 1962-1963 season, both acetate-‘JC and glucose-l”C were used in an attempt to demonstrate hctcrotrophic activity in Lake Vanda. There was no measurable uptake of labeled acetate in these

T.

MASON,

AND

JOHN

E. IIOBBIE

TABLE 9. Effect of nitrate addition on radioactive carbon fixation of Lake Vanda water. Values are counts per second. Experiment performed 30 Nm-2 Dee 1962 __--.-_ ~. ----Littoral water (16.3 hr)

___-

Pelagic 2-m water (13.3 hr) ~__-

1 ppm NO:,as NaNOs

Light Dark

65.4 1.12

3.52 -

Control

Light Dark

46.4 4.15

1.34 -

experiments, and strong adsorption of glucase-l”C to IIA Milliporc@ filters was observed. This was corrected for in the control, which was filtered immediately after the addition of glucose to the lake water. The in situ experiments showed a significant biological uptake of glucose at 10 m, with the light-bottle values slightly higher than the dark ( Table 8). At the 3-m level, those with glucose added did not The work of differ from the controls. Wright and IIobbie (1965) implicates bacteria as agents in this uptake. It has previously been reported that the addition of nitrate stimulated carbon fixation in Lake Vanda littoral water (Goldman 1964)) but phosphate and a mixture of minor elements did not. These experiments were repeated at Lake Vanda between 30 Novcmbcr and 2 December 1962 (Table 9), using both littoral and pelagic waters. The reason for the apparent depressive effect of nitrate on dark uptake of carbon in the littoral water is not clear. Nevertheless, photosynthetic carbon assimilation in the littoral water to which 1 ppm of nitratenitrogen was added was 41% over the control, while the nitrate-enriched pelagic water was 163% higher than the control. These ice-covered desert lakes of Antarctica probably approach the ideal microcosms more closely than any other lakes on this planet. As such, they provide an unusual opportunity for the study of a surprising and select variety of organisms adapted to colonize and survive this severe polar environment.

TWO

ANTARCTIC

REFERENCES

AMERICAN PUIELIC HEALTII ASSOCIATION. 1960. Standard methods for the examination of water and wastewater, 11th cd. APHA, New York, N.Y. 626 p. ANGINO, E. E., AND K. B. ARMITAGE. 1963. A geochemical study of Lakes Bonney and Vanda, Victoria Land, Antarctica. J. Geol.,

71: 89-95. AND J. C. TASI-I. 1963. Nutrient elements in two Antarctic saline lakes. Bull. Ecol. Sot. Am., 44: 38-39. AND -. 1964. PhysicochLmica liknology of Lake Bonncy, Antarctica. Limnol. Occanog., 9: 207-217. -, AND -. 1965. A chemical an; limnological study of Lake Vanda, Vietoria Land, Antarctica. Univ. Kansas Sci. Bull., 415: 1097-1118. ARMITAGE, K. B., AND II. B. HOUSE. 1962. A limonological rcconnaissancc in the arca of McMurdo Sound, Antarctica. Limnol. Oceanog., 7: 36-41. BACHMANN, R. W., AND C. R. GOLDMAN. 1965. Hypolimnetic heating in Castle Lake, California. Limnol. Oceanog., 10: 233-239. BA~NE,S, D. F. 1960, An investigation of a perennially frozen lake, p. 134. In Air Force Surveys in Geophysics No. 129. AFCRL-TN60-660. DAVID, T. W. E., AND R. E. PRIESTLEY. 1914. Geology: Glaciology, physiography, stratigraphy, and tectonic geology of South Victoria Land. Brit. Antarctic Exped. 1907-9, Rept. Sci. Invest. Geol., 1: l-319. DOUGI-IERTY, E. C., AND L. G. HARRIS. 1963. Antarctic micromctazoa: Freshwater spccics in the McMurdo Souncl area. Scicncc, 180: 497-498. 1956. Revision DROUET, F., AND W. A. DAILY. of the coccoid Myxophyccae. Butler Univ. Bot. Studies, 12: l-218. FOGG, G. E., C. NALEWAJKO, AND W. D. WATT. 1965. Extraccllular products of phytoplankton photosynthesis. Proc. Roy. Sot. (London), Ser. B., 162: 517-534. GOLDMAN, C. R. 1963. The measurement of primary productivity and limiting factors in freshwater with carbon-14, p. 103-113. In M. S. Doty red.], Proceedings of the conference on primary productivity measurement, marine and freshwater. U.S. Atomic Energy Commission, TID-7633. -. 1964. Primary productivity studies in Antarctic lakes, p. 291-299. In R. Carrick, M. Holdgate, and J. Prevost [cds.], Biologic Antarctiquc. Compt. Rend., Premier Symp. organise par le S.C.A.R. Actualities Sci. Indus. No. 1312. Hcrmann, Paris. -, AND D. T. MASON. 1962. Inorganic precipitation of carbon in productivity expcriments utilizing carbon-14. Science, 136: 1049-1050. ->->

DESERT

LAKES

309

AND B. J. B. WOOD. 1963. Liihtinjury’ and inhibition in Antarctic freshwater phytoplankton. Limnol. Oceanog., 8: 313-322. HOARE, R. A., K. B. POPPLEWELL, D. A. IIOUSE, R. A. HENDERSON, W. M. PHEBBLE, AND A. T. WILSON. 1964. Lake Bonney, Taylor Valley, Antarctica: A natural solar energy trap. Nature, 202 : 886-888. HOLM-HANSEN, 0. 1964. Isolation and culture of terrestrial and freshwater algae of Antarctica. Phycologia, 4: 43-51. KELLY, W. C., AND J. H. ZUMI)EIXGE. 1961. Weathering of a quartz diorite at Marble Point, McMurdo Sound, Antarctica. J. Gcol., 69 : 433-446. of L1KENSy G* E* 1964. An unusual distribution algae in an antarctic lake. Bull. Torrey Bot. Club, 91: 213-217. LLANO, G. A. 1959. Antarctic plant life. Trans. Am. Geophys. Union, 40: 200-203. LYoNSy J* Iz-, ANL)R* E* s’rolBm* 1’S’- The absorptivity of ice: a critical review. Sci. Rept. No. 3, Contract AF 19: (604) 2159. Prepared for Geophysics Research Directorate, Air Force Cambridge Research Center, Bedford, Massachusetts. 13 p. MCKELVEY, B. C., AND P. N. WEBB. 1961. Gcological reconnaissance in Victoria Land, Antarctica. Nature, 189 : 545-547. MEYEH, G. H., M. B. MORROW, 0. WYSS, T. E. BERG, AND J. L. LITTLEPAGE. 1962. Antarctica : The microbiology of an unfrozen saline pond. Science, 138: 1103-1104. NICI-IOLS, R. L. 1962. Geology of Lake Vanda, Wright Valley, South Victoria Land, Antarctica. Am. Geophys. Union, Geophys. Monograph, 7: 47-52. RAGOTZKIE, R. A,, AND I. FRIEDMAN. 1965. Low dcuterium content of Lake Vanda, Antarctica. Scicncc, 14*8: 1226-1227. AND G. E. LIKENS. 1964. The heat baiancc of two Antarctic lakes. Limnol. Occanog., 9: 412-425. RODIIE, W. 1955. Can plankton production procecd during winter darkness in subarctic I;;k7~~22Vcrl~andl. Intern. Vcr. Limnol., 12 : RU~TNEQ F. ’ 1963. Fundamentals of limnology, 3rd cd. Univ. Toronto Press. 295 p. SCOTT, J. T. 1964. A comparison of the heat balance of lakes in winter. Tech. Rept. No. 13. ONR Contract No. 1202 (07) to Dept. Mcteorol., Univ. of Wisconsin. 138’ p. SMITEI, H. T. U. 1965. Anomalous erosional topography in Victoria Land, Antarctica. Science, 148 : 941-942. SMI’~II, S. H. 1962. Tcmpcrature correction in conductivity measurements. Limnol. Oceanog., 7 : 330-334. TEDROW, J. C. F., F. C. UGOLINI, AND H. JANETSCIIlx. 1963. An antarctic saline lake. New Zealand J, Sci., 6: 150-156.

310

CHARLES J. E. chromator.

Il. GOLDMAN,

DAVID

1959. Natural water as a monoLimnol Qceanog., 4: 102-105. AND R. W. PREISEND~RFER. 1962. Li&t, p. 3C3rl-451. In M. N. Hill [cd.], The sea, v. 1. Intcrscicncc, New York. UTTEIXBACK, C. L., L. D. PIIIFEH, AND R. J. ROIHNSON. 1942. Some chemical, physical and optical characteristics of Crater Lake. Ecology, 23: 97-m3. WEYL, I’. K. 1964. On the change in electrical conductance of seawater with tcmpcraturc. TYLEIX,

T. MASON,

AND

JOHN

E. HOBBIE

Limnol. Occanog., 9 : 75-78. A. T. 1964. Evidence from chemical diffusion of a climatic change in the McMurdo Dry Valley 12QQ years ago. Nnturc, 201: 176-177. AND H. W. WELLMAN. 1962. Lake V&la : An Antarctic lnkc. Nature, 196: 1171-1173. WRIGHT, R. T., AND J. E. HOBBIE. 1965. The uptake of organic solutes in lake water. Limnol. Oceanog., 10: 22-28. WILSON,