ORIGIN OF STRATIFICATION IN AN AFRICAN RIFT LAKE J. F. Tailing Freshwater Biological Association, Amblcside, England ABSTRACT Physical, chemical, and algal features of Lake Albert (East Africa) are described, chicfly from three longitudinal surveys in 1980-61. The major ionic composition of the water is largely determined by another lake (Lake Edward) higher in the drainage system, but abiogenic sources are possibly responsible for a great enrichment of phosphate and depletion of silica. Contrary to most earlier accounts, a well-marked thermal and chemical stratification can develop in the lake. The thermal stratification is probably related to profile-bound density currents of cooler water, which flow from shallow areas of the lake and have varying effects in deeper water. Their interpretation is supported by examples of chemical and algal stratification, -particularly of dissolved oxygen, silica, and a species of the diatom Nitxschiu. form of thermal stratification in the lake, and hence on the dependent distribution of various chemical constituents and of planktonic algae. These physical, chemical, and algological observations are described here, chiefly in their relation to the conditions of stratification; features otherwise distinctive of the lake are also summarized. My thanks are due to the Director, the late Dr V. D. van Somercn, and staff of the East African Fisheries Research Organization, Jinja, for their co-operation and laboratory facilities, and to the Uganda Fish and Game Department for making available a boat and crew. Mr Eric Hamblyn gave much valuable assistance during the cruises, as did my wife who was responsible for the analyses of total phosphorus, phosphate, nitrate, silica, and sulphate. Spectrophotometric work was made possible by a grant from the Royal Society. Other financial assistance was received from the Colonial Office, London. Dr J. W. G. Lund and Mr R. S. A. Beauchamp gave helpful criticism on the manuscript. INTRODUCTION A pronounced stratification in many tropical lakes is controlled by the density changes associated with unusually small differences of temperature. This feature is due particularly to the rapid change of water density with temperature, which exists at the high temperatures usual in thcsc lakes. Ruttner (1931, 1937) has drawn attention to the unexpectedly high stability of stratification which can result. However, if the water density alters rapidly with temperature, it is likely to be unusually responsive to small exchanges of heat. This sensitivity is clearly seen in the superficial stratification which develops diurnally in many tropical waters ( e.g., Talling 1957a), and is usually ended by nocturnal cooling and mixing. Apart from such superficial changes there can be distinguished (cf. Worthington and Beadle 1933) the more persistent stratification of some tropical lakes, where a deeper discontinuity in temperature and density exists. The mode of origin of the discontinuity, and its relation to the surface exchanges of heat, are little known. This follows particularly from the rarity of studies in tropical lakes on the seasonal variation of thermal structure, and METHODS Most made during of the lake, between Butiaba-Ntoroko-Butiaba, on 2427 November 1960, 16-18 March 1961, and 7-9 August 1961. Some additional samples of surface water, obtained on 7 February and 20 June 1961, were also analyzed. Water samples from various depths were three of its horizontal distribution within the lake b asin. Thcsc aspects were studied in an East African Rift lake, L,ake Albert, on several occasions during 1960-61. The results throw some light on the origin of an unexpected 68 observations cruises along were the length ORlGlN OF STRATIFICATION IN AN AFRICAN RIFT LAKE 69 was measured once during each cruise, collected by a Ruttner or a van Darn <1956) sampler, with capacities of 1 L and using a selenium photo-cell with violet, 5 L respectively. Temperature was meas- blue, green, and red color filters. This optiured by a mercury-in-glass thermometer of cal work will be described more fully elsethe Ruttner sampler (read to 0.05”C) and, where; in connection with measurements of photosynthesis by the phytoplankton. jn more detail, by a thermistor thermometer Counts of planktonic algae were made described by Mortimcr (read to 0.02OC ) as using iodine sedimentation and the inverted and Moore ( 1953). Other analyses included microscope (Lund, Kipling, and Le Cren dissolved oxygen by the Winkler method, without modification for reducing sub- 1958 ) , stances; pII, calorimetrically using thymol GENERAL FEATURES AND MORPHOMETRY blue indicator and comparison against colored glass standards with the Lovibond Lake Albert is the northernmost lake in “Nessleriser”; silica, by reduction to molybthe west limb of the Great Rift Valley (Fig. denum blue and spectrophotometric estima1). Apart from direct rainfall, the lake retion ( Mullin and Riley 1955) ; soluble phos- ceives water principally through the Semliki phatc-phosphorus, by Deniges’ method folRiver, which flows from Lake Edward, and lowed by visual estimation in the Lovibond intermediate streams draining the Ruwen“Ncssleriser”; total phosphorus by the same zori massif. A number of small streams fall method after oxidation with 20% perchloric directly into the lake from the surrounding acid, but with final spectrophotometric esti- Rift escarpment, and several hot springs mation; nitrate-nitrogen by reaction with and salt wells near the shore probably conphenol disulphonic acid, decolorization with tribute salts to the lake. The Victoria Nile aluminum hydroxide, and final spectroenters its northcast corner, but almost imphotometric estimation; alkalinity due to mediately turns away bearing a variable bicarbonate and carbonate, by titration to contribution of lake effluent (IIurst 1925; pH 4.5 with centinormal hydrochloric acid; Beauchamp 1956; Tailing 1957b ) . calcium and magnesium, by complcximetric The lake lies at an altitude of 616 m titration with versenate (Heron and Mac( 2,050’ ft ) , is 145 km long, 35-46 km broad, kereth 1955 ) ; sodium and potassium, on has an area of about 5,600 km2, maximum one sample only, by flame photometry; depth of approximately 58 m and an averchloride, by titration with silver nitrate age depth of approximately 25 m. The eastusing chromate indicator; sulphatc, by pre- ern shoreline shows a fine succession of cipitation as barium sulphate in acid solu- curved spits, which may reflect wave and tion and estimation of turbidity with the current motion parallel to the shore generand electrical conducspectrophotometer; ated by the predominantly southerly winds tivity ( k20) by the portable “Dionic” con- (Worthington 1929a, p. 48; 192910,p. 118). ductivity meter. Determinations made in Deep sediments have been deposited in the the field shortly after sampling included lake basin since its origin; they include the pH, phosphate-phosphorus, the evaporation famous Kaiso fossil beds. Sedimentation Cd loo-ml samples for later nitrate analysis, near the Semliki discharge has led to a and silica estimation by Atkins’ modification shallow underwater bar projecting into the cd Dienert and Wandenbulcke’s m&hod. lake ( Fig. 1 ), dividing the main deep-water The values finally used for silica were those area near the western escarpment from a obtained spectrophotometrically after stor- subsidiary deep groove or zcanyon” near age of the samples in the dark for l-5 days. the eastern (Uganda) shore. The lake All chemical determinations were calibrated water is often rather turbid with fine, silty using standard solutions, and, except for material in suspension, besides varying denoxygen measurements, all samples were col- sities of phytoplankton; the euphotic (and lected in polyethylene bottles. Light pene- photosynthetic) zone was measured as 6tration at an arca midway along the lake 11 m deep. Suspended silt was often con- J. F. TALLING b-IG. 1. Bathymetric insct shows its relation map of Lake Albert, to other African lakes. with spicuous in the lowermost 5-10 m of the water column at the deeper offshore stations, suggesting that water movements were appreciable in this deep water. Further sources of relevant limnological information include the surveys of Worthington ( 1929a, b, 1930), and by a Belgian expedition published partly in Verbeke (1957) and van der Ben (1959). A general compilation is given by van Meel ( 1953), but contains errors in two tables of chemikal constituents (pp. 131, 234). WATER CHIEMISTRY The highly distinctive chemistry of the lake water is shown in Table 1, which sum- isobaths in mctcrs, from Vcrbckc (19rj7). The marizes the most complete analyses available. The data of Elskens have not yet been fully published and are taken from the works of Verbeke ( 1957) and van der Ben ( 1959). Less detailed analyses have been given by Worthington ( 1929a, 1930), Fish ( 1952a, 1952b, 1953)) and Talling ( 1957b). The moderately high salt content in Lake Albert is reflected in the values for conductivity and alkalinity. These indicated no pronounced variation with depth or along most of the lake’s axis, though small inflow streams caused local dilution, as at Ntoroko. The high proportion of potassium to sodium, and of magnesium to calcium, are unusual in African and indeed in most lake waters; ORIGIN TrADLE 1. Chemical OF STRATIFICATION composition IN AN AFRICAN of the sur(ace water RIFT 71 LAKE of lakes AZbert and Edward Lake Elskcns” Fcb ‘53 - Conductivity ( k20, pmho ) p 11 Na (mg/L) :a [m$k{ * Station No. Talling Nav 740 9.12 97 9.3 66 Mg (Zg,L) 31.5 HCOa + CO, (meq/L) 7.8 Cl (mg/L) 32 SOi (mg/L) 25 SO2 (mg/L) 3.4 l%*P (pg/L) 130 Total P ( ,ug/L ) N0a.N b-M4 9 Sum of cations (mcq/L) 8.95 Sum of anions ( meq/L ) 9.22 1149 in Lake Albert, Lake Albert ‘60 Feb ‘61 Mar June ---- ‘61 Aug ‘61 735 91 9.8 65 735 9.0 10.8 - 780 9.9 - 730 8.9 9.0 - 31.2 7.30 32.9 27 0.1-0.8 120-170 4.5-7.5 31.2 7.33 33 32 0.09-0.9 200 8.74 8.93 32.1 7.27 33.7 43 0.04-1.1 31.9 7.30 38.5 34 0.33 168 9.09 31.5 7.25 45 o-4-0.9 120-150 133-164 10-33 - No. 683 in Lake Edward. they also occur in other lakes of the western Rift such as Tanganyika, Kivu, and Edward. Comparison with the analyses for Lake Edward suggests that the major ionic composition of Lake Albert water is determined by inflow from the region of the upper lake, with a slight dilution, mainly due to run-off from the Ruwenzori mountain mass ( Beauchamp 1956). Beauchamp also believed that the sulphate content of Lake Albert water varied considerably in relation to corresponding changes in the surface water of Llake Edward. Though the published data are scanty, this view gains some slight support from the apparent variation of sulphatc shown by the present analyses from Lake Albert. Features more peculiar to Lake Albert a:re a remarkably high concentration of phosphate (120-200 pug PO4 P/L) and a generally low concentration of dissolved silicon (here expressed as silica ), The source of the phosphate is not established, but hot springs near the lake and beside the Scmliki River inflow may be suspected, or possibly changes in other forms of phosphorus prcsent in Lake Edward water ( cf. Table 1). The total phosphorus content usually agrees, within experimental error, with the soluble p hosphatc-phosphorus, so that any uptake by phytoplankton is obscured by the large l Tailing Elskcns* ‘61 720 8.9 10.0 - 8.79 Edward 120-170 13 9.12 Conductivity corrcctcd from J;m ‘54 DCX ‘60 June ‘61 935 8.89 112 9.7 79 880 100 12.5 82 925 9.1 110 12.4 90 44.5 10.1 27 35 2.0 30 23 11.0 11.6 50.5 9.22 35 32 5.7 - 47.8 9.85 38 31 6.5 18 127 24 11.6 11.5 values 1G 10.9 at 25°C. excess of fret phosphate. The same is largely true of the release of phosphate from the bottom deposits. Concentrations of silica are more variable, both with respect of depth ( discussed later) and to distance along the axis of the lake. The very low values often found (Figs. 2-4) differ so much from those expected from the Semliki inflow, where 3.8 mg/L was found by Elskens (van dcr Ben 1959, p. SO), that some active mechanism for silica removal must exist in the lake. The lowest values, often less than 0.5 mg/L, are usually found at stations most distant from the Semliki inflow, and values above 1 mg/L (including at least one by Elskens) are from the area adjacent to that inflow. The planktonic diatoms common in the lake must contribute to silica depletion, but the large quantities which are probably involved suggest the occurrence of some non-biological precipitation or polymerization. Such a process would be expcctcd to be favored by the high pH of the water, and possibly also by the presence of calcium which may be readily precipitated. TIIERMAL STRATIPICATION In the most detailed accounts of the lake, by Worthington ( 1929a, 1930), Verbeke ( 1957), and van der Ben ( 1957)) thermal 72 J. F. TALLING stratification is said to be slight, with windinduced mixing usually extending to all the depths sampled. However, Fish ( 1952b) found that an appreciable depletion of dissolved oxygen could develop in the deeper water. Elsewhere (1956) he suggestedthough without direct evidence-that a discrete and deoxygenated bottom layer might originate from the inflow of colder Semliki River water, and cause the mass mortalities of Nile Perch ( Lutes nlbertianus ) which arc occasionally reported. The present work shows that a discrete bottom layer of colder water can develop, but an alternative explanation of its origin appears more likely. Figure 2 shows depth profiles and longitudinal sections of thermal structure in the lake during November 1960. Thermal discontinuities with depth appear partly in relation to a superficial diurnal heating, and partly bounding a lowermost layer of water below 27.1”C. The latter discontinuity is most obvious at stations 5, 6, and 8, but the chemical evidence ( discussed later) shows the 27.1” isotherm to mark an equivalent boundary to vertical circulation at stations 0 and 10. Water below 27.l”C is also present in the shallow southern end of the lake, so that this isotherm, with others, tilts strongly along the axis of the lake. Similar tilting of the isotherms is obvious in the March and August 1961 data (Figs. 2 and 4)) but there is little indication of a deep thermal discontinuity in March. In August it has reappeared about the 26.O”C isotherm in the deeper area adjacent to the southern shallows, but does not extend to the more fall northerly stations. The tempcraturc about the deep discontinuity is always small, not exceeding 0.7”C and usually much less. Three possible explanations of these conditions can be considered. 1) A direct origin of the lower layer from cold Semliki River water is improbable, as a perceptible change with depth of the major ionic constituents was not found at any station. This applied even to the quantities most precisely measured such as alkalinity, I+ l%, calcium plus magnesium, I+ I%, and conductivity, + 3%. The composition of the Semliki inflow water, however, shows comparatively large seasonal variations. Beauchamp ( 1956, Fig, 1) recorded changes of conductivity, from a station near the lake, between 360 and 810 pmho. 2) The stress of southwest winds on the lake surface might lead to an upwelling of deeper and cooler water at the windward end, in the manner illustrated by Mortimer ( 1952). Although some of the higher isotherms shown here may be tilted in this way, it can scarcely account for the close connection between the lower colder stratum and the basin profile. 3) Given this connection and an intralacustrine origin, the bottom layer is most easily seen as a profile-bound density current descending from the shallow end of the lake, where surface heat exchange has produced an area of slightly colder and denser water. Nocturnal cooling, for example, can be very pronounced in shallow African waters (Talling 1957a). At 0830 on 8 August 1961 the O-5 m water in the shallow Buhuka Bay was at a temperature (25.75”C) lower than any measured above 20 m at the adjacent station of the offshore transect line. A convective descent of cooler water from such shallows can therefore be visualized. Since the largest shallow area is in the southern end of the basin, a longitudinal section of the lake would show the observed asymmetry of thermal stratification. Other and smaller areas of shallow water may also contribute to the effect. This suggestion can be tested further by reference to examples of stratification in chemical constituents and phytoplankton. CHEMICAL STRATIFICATXON Chemical stratification is shown in measurements of dissolved oxygen, nitrate, silica, pH, phosphate, total manganese, and total iron. The first three of these are the most useful indicators. The pH changes are small due to the high buffering of the lake, and phosphate changes are obscured by the large excess of this ion. The distribution of total manganese and iron is insufficiently known; small amounts, such as 15-30 pg/L of manganese and 45-80 pg/L of iron, were mcasurcd in the deeper water. Three types of oxygen-depth distribution ORIGIN OF STRATLZ1‘ICATION /’ IN AN ,& B AFRICAN (station RIFT nos.) 73 LAKE ,/ 3 _..- TEMPERATURE PR( 3FILES DEPTH -I 27 0 Ii I.5 28 27L 28’ .. 29’ 27I 28, 29, zp 27 2,8 - 27 28 29 270 -27 28 -- 27 28 tbT-----l 2.3 29 Oc . I .* \ .i made during 2427 Novcmbcr 1960, at stations indicated on the map above. -. FIG. 2. Observations ‘l’cmpcrature measurements arc shown on a longitudinal section and in individual depth profiles; chemical and algal data arc given as depth profiles at the stations numbered. 74 J. 17. TALLING KILOME TRES 7 27 28 27 ’ 20 ’ 29 ’ ;7L;Q 27 .. 1 216 1 0 FIG. 3. Observations made during 16-18 March 1961, shown i” 20 as in Figure 40 2. 100 500 ORIGIN I I OF S’IltATWICATION I I I I IN I I1 AN AFRICAN 120 L3 ; y LAKE I 0 O RIFT KILOME TRE 5 _L -7 I2 PROFILES I 30 45 24 25 LY 25 0 26 2r-Tb 2zpy, 25 26 25 ~zpzi-- 25 - 26 27 26 27 26 d7 . . . . 1 r I3 IS .. "K" !* . 30 .* d 2s - 2276 PO,-P I 45 L0 I 0 (I@) 200 7 .. 8 9 .a . . J ” FIG. 4. i 13 . Observations made during ‘7-9 August 1961, shown as in Figure 2. IO 76 J. F. TALLING were found ( Figs. 2-4). In November 1960 the water below the deep thermal discontinuity was poor in oxygen. Concentrations increased rapidly across the discontinuity and values close to saturation (about 7.2 mg/L) were found between 0 and 30 m. The concentrations in the cooler water below 27.l”C were highest in the shallows (station 3), less in the adjacent deep water ( station 6)) and lowest in the more distant northern stations (0 and lo). This sequence is in agreement with the expected “aging” of the postulated downflow of cooler water. In March 1961 the oxygen concentrations decreased rather continuously with depth at all stations. The absence of a sharp discontinuity in the profiles accords with the non-development of the deep thermal discontinuity. However, the oxygen deficit in the deeper water columns is still large, implying active consumption at least in the lower layers. During August 1961 the oxygen stratification is practically abolished. Though this condition corresponds with the more nearly isothermal state of most of the lake, it also extends to the shallow southern end, where a bottorn layer of colder water is recorded from stations along an “underwater valley.” This layer must therefore be of local and recent origin, Its spread to the more distant sampled arcas of the lake floor is apparently prevented by more intense vertical mixing. As before, a substantial oxygen deficit exists in the deeper water columns, and even the surface water is appreciably undersaturated. The vertical distribution of nitratenitrogen suggests a production in the lower layers or mud surface, and consumption in the upper layers. A strong vertical mixing in August is indicated, when the highest surface concentrations were recorded. Similar sites of production and depletion are likely from the depth profiles for dissolved silica. In November 1960 the widespread bottom layer of cooler water contains high concentrations, sharply distinguishcd from the low values above. More gradual vertical changes exist in March 1961, as in the corresponding tcm- perature and oxygen profiles; in August 1961 the vertical differences are almost eliminated. In these features silica enrichment generally follows oxygen depletion, and a similar hydrological background can be invoked. However, an additional source of dissolved silica probably exists in the Semliki River inflow, as mentioned above in relation to the higher silica concentrations usually found in the shallow southern end of the lake. If water from this area descends to northerly stations, a part of the deeper silica accumulation there may be rccruitcd from the southern shallows. PEIYTOPLANKTON Two diatoms, a Nitzschia sp. (probably N. bacata Hust. ) and Stephanodiscus astraea ( Ehr. ) Grun. with its variety minutula (Kiitz.) Grun., normally compose the main part of the phytoplankton. Besides the 1960-61 observations, one or both species were probably also noted in 1952 by Fish ( 1952b) and Ross (1955), and in May 1954 by Dr G. A. Prowse and myself. Bachmann ( 1933) recorded S. astraeu as the dominant alga of net samples from offshore water collected in 1928. These diatoms usually occurred throughout a wide range of depths (Figs. 2-4), and densities of the Nitxschiu sp. greater than 1,000 cells/ml were occasionally recorded, as in February and March 1961. Certain other algae were less regular components, but sometimes reached large densities in the uppermost ( O-10 m ) layer; these included the blue-green alga Anabaena flos-aquae ( Lyngb. ) B&b. in November 1960 (Fig. 2)) a species of the flagellate Gymnodinium in August 1961 (Fig, 4), and two unidentified flagellates ( one a cryptomonad ) in March 1961. Various algae, and particularly Anabaenopsis tanganyikae (G. S. West) Wolosz. et Miller and Lyngbya limnetica Lemm., arc frequent in shallow or inshore water. The stratification of the Nitzschia poplation could be clearly traced on the three main cruises, and is closely related to the varying thermal stratification. In November 1960 ( Fig. 2) a sharp drop in cell numbers often occurred at the deep thermal dis- ORIGIN OF STRATIFICATION continuity. A more gradual decline with depth was present in March 1961 ( Fig. 3), when there was no obvious discontinuity in the deeper water. The situation in August 1961 (Fig. 4) is particularly interesting, as thlp higher cell numbers were then found in the shallow southern area ( stations 3 and 10) and in the colder bottom layer of water at the nearby deeper stations. EIere the Nitzschia concentration may act as a biological “tag” to the water in the supposed profile-bound density current. The survival of apparently healthy cells in this water is another indication of its recent surface or.igin during August, although sedimentation of cells from above cannot be excluded. DISCUSSION From this work three distinctive features of limnological interest emerge. Physically, there exists a peculiarly mobile and asymmetric form of stratification; chemically, the high enrichment in phosphate combined with an often extreme impoverishment in silica are remarkable; and algologically, the phytoplankton is of a singularly restricted composition centered on the diatom genera Nitzschia and Stephanodiscus. None of th#ese three conditions seems to be known from the other large African lakes. However, if the convective theory of the origin of stratification in Lake Albert is valid, similar processes are likely to be widespread in tropical lakes which have small vertical differences of temperature, More surveys of the horizontal variation of thermal stratification in such lakes would be valuable. The most detailed yet published are of Lake Victoria by Fish (1957) and Newell ( 1960). Here the enormous lake area and complex shoreline make an interpr’etation more difficult. An annual cycle of stratification was followed during 1960-61 and will be described elsewhere. In non-tropical lakes profile-bound density currents have been occasionally discussed, particularly in relation to a thermal (convective) origin under ice ( Hutchinson 1957, p. 476; cf. Mortimer and Ma&e&h 1958) and to a solute-enrichment origin during later stratification (\ Hutchinson 1957, IN AN AFRICAN RIFT LAKE 77 p. 477). The significance of such solute additions in tropical lakes is conjectural, but tropical conditions would seem to favor a thermal origin of density currents. In such examples as Lake Albert these currents may possibly come to dominate the thermal subdivision of the lake. A full description of the annual sequence of thermal stratification in Lake Albert is lacking, but Verbeke ( 1957, Fig. 14) and van der Ben ( 1959, Fig. 9) have published a series of temperatures at three depths, 1, 20, and 40 m, taken throughout 1953 at a location probably near Kasenyi. Here the annual temperature range at 2040 m is only 26.6-27.4”C, and most of the vertical differences of temperature between 1 and 20 m are probably controlled by diurnal heating and cooling. These results, and the present work, suggest that the lake possibly approaches the condition of a constant temperature bath more closely than any other large tropical lake so far investigated in sufficient detail. This equable background, and the shallows placed asymmetrically in the lake, may allow the convective consequences of local cooling to be expressed in an unusually striking way. REXERENCE3 H. 1933. Phytoplankton von Victoria Nyanza-, Albert Nyanza- und Kiogasee. Bcr. Schweiz. Bot. Ges., 42: 705-717. BEAUCHAMP, R. S. A. 1956. The electrical conductivity of the headwaters of the White Nile, Nature, 178: 616-619. BEN, D. VAN DER. 1959. La &g&ation dcs rives dcs lacs Kivu, gdouard et Albert. Explor. Hydrobiol. des Lacs Kivu, adouard et Albcrt (1952-54), Vol. 4 (l), 191 pp. Bruxcllcs. DOIIN, W. G. VAN. 1956. Large volume water samplers. Trans. Amer. Geophys. Union, BACHMANN, 37: 682-684. FISH, -. -. G. R. 1952a. Appendix A (5). Zn East African l?isheries Research Organization Annual Report for 1951, p. 33. East African High Commission. Nairobi. 1952b. Appendix A, Chemical Analyses, Lake Albert. Zn East African Fisheries Rcscarch Organization Annual Report for 1952, p. 27. 1954. Appendix A2, Lake Albert. Zn East African Fisheries Research Organization Annual Report for 1953, p. 27. J. -, I?. 1956. Some aspects of the respiration of six species of fish from Uganda. J. Exp. Biol., 33: 186-195. -. 1957. A seiche movcmcnt and its cffcct on the hydrology of Lake Victoria. Fish. Publ., Lond., 10, 68 pp. HERON, J., AND F. J, MACKERETII. 1955. The estimation of calcium and magncsiui~1 in nattural waters with particular reference to those of low alkalinity. Mitt. Intern. Vcr. Limnol., 5, 7 PP. IIUItST, H. E. 1925. The Lake Plateau Basin of the Nile. Ministry of Public Works, Egypt. Physical Dept. Paper No. 21, Govt. Press, Cairo. (Section 13: Salt content of the water of the Lake Plateau, pp. 67-73. ) HUTCHINSON, G. E. 1957. A treatise on limnology. Vol. I. Geography, physics and chcmistry. New York. pp. xiv + 1015. LUND, J. W. G., C. KIPLING, AND E. D. LE CHEN. 1958. The inverted microscope method of estimating algal numbers and the statistical basis of estimations by counting. Hydrobiol., 11: 143-170. MEEL, L. VAN. 1954. Lc Phytoplancton. ExLac Tanganika 1946-47, plor. Hydrobiol. Vol. 4( 1) : 681 pp. Bruxelles. MORTIMER, C. II. 1952. Water movements in lakes during summer stratification; evidence from the distribution of temperature in Windermere. Phil. Trans. B, 236: 355-304. AND F. J. MACKERETEI. 1958. Convectidn and its conscqucnccs in ice-covered lakes. Vcr. Intern. Ver. Limnol., 13: 923-932. -, AND W. H. MOORE. 1953. The use of thermistors for the measurcmcnt of lake temperatures, Mitt. Intern, Ver. Limnol., 2: l-42. TALLING MULLIN, J. B., AND J. P. RILEY. 1955. The colorimetric determination of silicate with special rcfcrencc to sea and natural waters. Analyt. Chim. Actn, 12: 162-176. NEWELL, B. S. 1960. The hydrology of Lake Victoria. Hydrobiol., 15 : 363-383. Ross, R. 1955. The algae of the East African Great Lakes. Vcr. Intern. Vcr. Limnol., 12: 320-326. RUTTNER, F. 1931, Hydrographische und hydrochcmische Bcobachtungcn auf Java, Sumatra und Bali. Arch. Hydrobiol. (Suppl. ), 8: 197-454. -. 1937. Stabilitgt und Umschichtung in tropischen und tempericrten Seen. Arch. Hydrobiol. ( Suppl. ), 15: 178-186. TALLING, J. F. 1957a. Diurnal changes of stratification and photosynthesis in some tropical African waters. Proc. Roy. Sot. B, 14’7: -. 57-83. 1957b. The longitudinal succession of water characteristics in the White Nile. Hydrobiol., 11: 73-89. VERBEKE, J. 1957. Rechcrchcs &cologiques sur la faune dcs grands lacs de l’est du Congo Beige. Explor. Hydrobiol. Lacs Kivu, Bdouard et Albert ( 1952-54), Vol. 3( 1) : 177 pp. Bruxclles. WORTIXINGTON, E. B. 1929a. A report of the fishery survey of lakes Albert and Kioga. H.M.S.O. London. 136 pp. 1929b. The lift of Lake Albert and -. Lake Kioga. Gcograph. J., 74: 109-132. 1930. Observations on the temperature, -. hydrogen-ion concentration, and other physical conditions of the Victoria and Albert Nyanzas. Int. Rev. Hyclrobiol., 24: 328-357. -, AND L. C. BEADLE. 1933. Thcrmoclines Nature, 129 : 55-56. in tropical lakes.