limnological observations on lake kariba during 1967 with emphasis
Transcription
limnological observations on lake kariba during 1967 with emphasis
LIMNOLOGICAL OBSERVATIONS ON LAKE KARIBA DURING 1967 WITH EMPHASIS ON SOME SPECIAL FEATURES’ G. W. Begg Lake Kariba Fisheries Research Wild Life Institute, Department of National Management, Rhodesia Parks and ABSTRACTI’ Results of limnological recordings made during 1967 on Lake Kariba are described. Kariba is monomictic, mesotrophic, and has five defined basins, each of which exhibits its own individuality. The two upstream basins are riverine; they are flushed out in May by the Zambezi River floods and thereby assume turnover characteristics earlier than the other three basins, which are truly lacustrine with temperature-induced turnover. Great amplitude of variation in water chemistry exists relating to the basin locality, biotope (river, estuary, cleared area, open water), time of year (turnover winter/summer ), and depth. Generally, values of dissolved oxygen, conductivity, alkalinity, ancl pH fall from surface to bottom. When H2S is present, alkalinity and conductivity values increase. A Occasionally density currents occur at chemocline follows the profile of the thermocline. depth, particularly after the first seasonal floods. Transparency decreases toward the head of the lake and in all areas decreases after turnover. The source of H&3 is the Sal&&z-infested rivers. Dissolved oxygen in the hypolimnion The hydrobiological effects of Salvinia are disis depleted 4 to 5’ months after tumovcr. cussed, and the influence of the bottom topography is described. INTRODUCITON Lake Kariba lies in the geologically leached out Gwembe Rift Valley on the Zambezi River at an altitude of 484 m. It was formed by construction of a dam, primarily desigpcd to produce electricity. The dam wall, 128 m high, was begun in 1956 and completed in 1960; impoundment of water began in December 1958 and, in this phase, it rose as fast as 5 m/24 hr. The lake had filled to 487.8 m above sea level by 1963. The discharge through flood gates and turbines results in considerable variation in water level, which should become less when additional turbines are put in operation by 1973. The lake has five well-defined basins. The axis of the lake lies along a SW-NE line, between 16’28’ S and 18’6’ S lat and 26”40’ E and 29’3’ E long. The lake extends for 280 km, is 40 km across at its widest part, and covers an area of 5,250 l This paper was presented at the Symposium on Standing Waters of the Limnological Society of Southern Africa at Rhodes University, Grahamstown, in July 1968. sq km. The maximum depth is 120 m, but 30% of the arca is shallower than 17 m. Previous hydrological studies of Lake Kariba have been made by the Joint Fisheries Organization from 1960-1964 under D. Harding ( 1962, 1964). Only the work of Cochc (1965) covered the total expanse of the lake. According to these surveys, the lake has exhibited a mono,mictic, stratified nature since as early in its formation as November 1959, with variation of the pattern only in the timing, depending on and previous meteorological conditions winter tcmpcratures ( Harding 1966). The temperature of the hylDo,limnion dictates the following season’s turnover tempcrature. In fact the uniqueness of conditions at Kariba lies in its novel hydrological conditions-the changing nature along its axis ( Fig. 1). It is misleading to discuss conditions in “the lake,” as is done by Harding, by referring to results obtained at only two localities (Sampa Karuma and the vicinity of Kariba Gorge ) , since conditions in these areas are atypical. Throughout 1967, limnological observations were continued by the Lake Kariba Fisheries Research Institute. I am sincerely 776 LIMNOLOGY LAKE OF LAKE KARIBA 777 KARIBA Legend: B3 Basin fj’)‘:> Bush-cleared a II DEKA R GiAAlR ; , European Basin areas settlement division MLIBl21 II “‘I,.~..... Old Zambezi FIG. 1. M~IZI of Lake Kariba desia, and general information. showing division of lake grateful for the valuable guidance of Chief Fisheries Research Officer Mr. M. I. van der Lingen and the criticism of Officer in Charge Mr. F. J. R. Junor, and to Dr. J. F. Talling for reviewing the manuscript. I would like to thank my colleagues in the research team at Lake Kariba Fisheries Research Institute for their interest, encouragement, and unfailing help. Also I would like to, acknowledge the help of laboratory assistants, Titus Matare and Geoffrey Marowa, Oswald Nyatitos2 ( field assistant ) , and Afiki Shayibu ( coxswain, MV Sampa). I am indebted to the Power Station Superintcndent, Central Africa.n Po’wer Corporation, for meteorological and hydrological data. METHODS The following observations were carried out at each station using standard methods. 2 Deceasccl. number into basins, main affluent rivers river from bed Rho- Temperature against depth-by using a bathythermograph [Wallace and Tiernan recorder, Model FA 190012, O-200 ft (O-61 m) range]. Light penetration-by Secchi disc (30 cm diam) and photometer (O.R.E. Model 504). Dissolved oxygen-by Winklcr determination. Detection of H&3 by smell. Conductivity (expressed as pmhos/ cm at 2X)-by portable meter (Beckman K25 So\lubridge Type RBs with dip cell). pH-electrometrically (Beckman Model N2 meter). Total alkalinity (as ppm CaC03)by titration using BDH 4.5 indicator and 50 N H2S04. Water samples from any desired depth by means of a. messenger-closed Ruttner bottle. Analyses were done in the field. OBSERVATIONS Meteorology Table 1 shows the monthly mean air temperatures recorded at Kariba during 1967. 778 TABLE from G. W. 1. Mean air temperntures (“C) in 1967, Central African Power Corporation meteorological records, Ku&a Township ~-__--- Month 2. Thermocline depth basins l-5 Nov 19G7 Min 1 2 ii 5 19-32 15-30 16-25 13-18 7-12 Max Min 31.17 31.75 31.70 22.47 22.78 22.22 34.6’9 35.0 34.24 21.11 21.11 20.0 31.97 22.96 34.44 18.78 29.14 27.67 25.96 29.09 18.58 17.45 15.15 17.32 30.81 29.79 29.14 32.38 16.16 15.56 11.45 13.58 32.22 35.45 35.40 32.2 20.25 24.54 24.69 22.22 37.78 41.17 43.34 37.42 16.36 21.16 21.21 19.75 Max TABLE Hasin Extremes Mean REGG (in meters) in Jan 1968 31-36 20-30 19-25 15-20 10-17 Of this capacity, 9 x lo6 m3 are lost by evaporation each year, and 40 x 10° m3 pass through the flood gates and turbines, depending on requirements. Only a third of the lake volume is replaced each year. PHYSICOCHEMICAL CONDI’SIONS Figure 2 shows the thermal situation in basin 5. Surface temperatures rose to 32X The no,rmal rainfall in Kariba area in February, while hypolimnetic tempera(basins 3, 4, and 5) is 610-800 mm.; in tures generally remained at 22C-reflecting Binga area (basins 1 and 2) it is 400-610 the previous winter turnover temperatures. mm. The bulk of rain falls in December Clearly, except for July and August, the and January. Rainfall for the 196&1967 lake remained thermally stratified. By Deseason was moderate, totaling 560 mm; the cember 1967 the thermocline was well 1967-1968 seasonal rains were late and established and stable. As the year propoor, totaling 430 mm, follo’wed by an gressed, and the epilimnion cooled, it sank unusually cold winter. During the rainy to 3540 m, at the same time becoming season, light rains may fall at night and sharper and more abrupt. Winter circulathunderstorms develop quickly during the tion occurred in July and oxygen pcneday-followed by a rapid drop in tempcratrated to the bottom. As Table 2 suggests, ture and rise in relative humidity. The the thermocline occurs shallower in basin winds generally blolw from the northcast, 5 than in basin 1, creating the progressive although in basins 4 and 5 southwesterlies phases of winter circulation (referred to blow frequently in the latter half of the below ) . Thermally, basins 3 and 4 behave year. rather like. basin 5. Deviations from this normal situation ocHydrology cur, such as the appearance of secondary Lake level fluctuation over 1967 was bethermoclines, either at the surface during twcen 482 and 484 m above sea level ( asl). the hottest part of summer days, or deeper, Coulter (1967) estimated the area oE due to colder inflows fro,m rivers. Variation Lake Kariba to be 1,718 sq miles (4,448 in the extent of the thermocline (or metasq km), but this could only be when the limnion) according to the thermal drop lake was at 475 111:asl-9 m below the per meter ranges from, 0.3-2.5 degrees C/ present level. At exactly 484 m asl, the m, largely determined by the degree of lake area is 5,250 sq km. At its peak height exposure to wind action of the particular in 1963 of 487 m asl,” the lake covered locality. 5,500 sq km. The thermal situation at Kariba Gorge The capacity at 484 m as1 (mean retennear the dam wall differs significantly from tion level.) is 160 X 1ctGm3. the normal due to the flow through the gates and turbines. The opening of the 3 The lake subsequently reached this level again gates had the .transient, temporary effect in 1969. FIG. 2. 25 Oc showing 3m ZD- - o- 2 20 *C F 25 30 changes in the relative 60 60 - profiles, 30 60 Lc temperature 20 40 I 40 20 Monthly 3m 0 position N during 7- J of the thermocline bo bo 20 1967 in the center of basin 5, Lake Kariba. 3- A 3 ~ 780 G. W. Jan Feb BEGG Mar 0 I- Gat’e opened on 20th FIG. 3. Showing the effect on the isotherms in the vicinity of Kariba Gorge the suction created by the opening of a flood gate on 20 February 1967. of drawing the thermocline deeper, but shattering it, so it is more gradual and extensive. Figure 3 indicates that the opening of two flood gates on 20 February coincided with the ascent of the 22.0, 22.5, and 23.OC isotherms. This confirms Cache’s (1968) observations in 1965. Turnover did not occur in Kariba Gorge until the night of 28 July, -shortly before surface temperatures began to rise in August. This delay is related to the fact that the gorge is the deepest part of the lake. Had turnover not occurred by August, the stagnant water of the hypolimnion would, in all probability, have remained at this locality for over a year.4 * Since this paper was written a phenomenon arose in Kariba Gorge where the winter in 1969 was not sufficiently severe to bring about turnover, so that this area was characterized by a double stratification, and hydrogen sulfide would have remained in the hypolimnion below the secondary thermocline for an entire season. Cochc ( 1968) refers to areas in basin 3 where dcoxygenation persisted even after turnover. (near the dam wall) of During the early stages of restratification, a thermocline may be found within the cleared areas and shallower zones osf the lake, which are normally homothermal. During midwinter, temperatures in the shallow margins of the lake may fall to 1lC at night. Inflowing river water, which is generally cooler than the lake water, is most apparent where the Zambezi enters the lake at Devil’s Gorge, particularly in July, when the Zambezi River water has a temperature of 17C. This is the coldest water to enter the lake, and thus the annual thermal range is greatest in basin 1. From Fig. 4, it is apparent that there is no conventional turnover in basins 1 and 2. As the Zambezi enters, the temperature of the entire water mass changes in these basins. At Devil’s Gorge in March, the temperature of the entire water mass was 29C. In July the water mass temperature was 17C-a difference of 12 degrees within 4 months. This indicates that when peak LIMNOLOGY OF LAKE KAIUBA 781 782 G. W. flow of the Zambczi (resulting from the Barotseland floods) occurs in Devil’s Gorge in May, the entire water mass in basins 1 and 2 is flushed out, The process is normally complete by the end of May. This could bc interpreted as turnover because of the following characteristic consequences: chemical homogeneity from top to bottom; abscncc of HZS; dccreasc in normal physicochemical values of surface water, transparency, and plankton density. Because these turnover characteristics occurred in May, when lake sulfate tempcratures in basins 3, 4, and 5 wcrc at 2X’, it is clear they were not induced by the sinking of cool surface lake water during the coIdest month, but arose as a result of a flushing-out process by the Zambczi River itself, which provides 70% of the total inflow into the lake. During May, “entirety shifts,” although most marked in Devil’s Gorge, wcrc noticeable throughout basin 1, through Sebungwe Narrows and as far as Binga, at the Logola River entrance. This flushing process continued so that by May, while the rest of the lake was stratified, basins 1 and 2 assumed turnover characteristics. The noticeable rise in chemical values after passing from basin 2, through Chctc Gorge, into basin 3, is due to the fact that the nutrient-poor Zambezi water is seldom apparent beyond Chcte. Here it sinks and may be found along the bottom of basin 3, as an underflow, tractable as far as the Chisangasanga Strait. The influx of relatively sterile Zambezi water at peak Elow not only noticeably altered the tempera,turcs, but also the water chemistry of basins 1 and 2. This was reflected by the low alkalinity, pH, and conductivity, the scarce plankton, the increase in turbidity, and the flushing out of the S&in&choked Zambezi system. While temperatures fell in July, basin 3 tended to set up a back-pressure in response to a lcsscning of flow, so the above chemical characteristics of basins 1 and 2 rose slightly and continued to rise. Subsequently flow was so slight that a thermocline appeared in Devil’s Gorge by September and BEGG TABLE 3. Secchi disc depths (aueruged and in cm) clt a-month intervals along axis of Lake Karibn ~___Basin NO. 1 2 3 4 5 MU May 84 286 470 550 325 131 240 4768 580 433 Jnl Sep Nov Jan 248 300 412 524 407 320 325 426 506 410 317 462 486 533 455 145 385 520 593 424 became fully cstablishcd by November, by which time basin 3 water completely modificd the surface of basins 1 and 2. Basins 1 and 2 altered from riverine conditions to assume Iakclike characteristics-constant hypolimnctic temperatures, stratificd water body, and increased values of conductivity, alkalinity, and pH. During this phase II$ appeared briefly from river sources, such as the Senkwe and Sebungwe, and penctrated the deoxygenated hypolimnion. Thus for 5 months of the year (September to January) basins 1 and 2 are lacustrine, while for the remaining months of the year they become riverine in nature. When lacustrine basins 3, 4, and 5 have turned over in July, basins 1 and 2 have rcstratified. Although the surface water over the lake is at its coldest, it is prcvented from sinking in basins 1 and 2, due to the very cold Zambezi underflow. The back-prcssurc from basin 3 assists in causing it to remain at the surface, creating a primary thermocline, while the Zambczi water flows beneath a decpcr secondary thermocline. Cache ( 1968) terms this a “density current.” Water transparency Secchi disc transparency varied over the lake from 10 cm to 10 m. In 1965 Cache ( 1968) observed Secchi values of 12 m. Relative light intensity ( RLI ) measurements indicated that an RLI of 1% seldom lies deeper than 20 m, and 50% of the light absorption takes place within the first 3 m of water. This confirms Cache’s ( 1968) observations conduclcd in 1965. LIMNOLOGY TAIHX 4. Secchi disc depths DigGG;nJof LL . I- Jill1 Kariba sub-basin ( Rcdcliff ) Sanynti sub-basin 580 424 (averaged FCh 593 417 OF LAKE and in cm), recorded Mar Apr May JUll 593 325 640 407 806 433 880 533 Transparency decreased in the areas cleared of original bush due to the richer planktonic life and sediments brought up into suspension by wave and wind action. Water transparency generally decreases from basin 4 to basin 1. Basin 4 has the clcarcst water ( see Table 3). Particularly in basins 1 and 2, the cIcar;est water conditions coincide with decreased river flow. The decline in transparency in bmasin 5 is due to the greater productivity of basin 5 and to inflow of the Sanyati River. Dccrease in transparency from basin 4 to, 1 is associated with change in water color: from a deep blue-green, to green, to olivcgreen, to brown. Estuarine and riverine localities have due to silt, richer lower transparency, plankton, and suspended organic matter. SaZvinia may tend to increase the transparcncy of river water owing to the flo’w restriction created by the mat (causing suspcndcd matter to scttlc out), nutrient extraction, and the inhibition of pIankton grolwth. Table 4 indicates the great differcnce in the optical pro’pcrties of the water at the Kariba and Sanyati sub-basins (exccpt during turnover), The former has higher Secchi values because river inflow is Iacking. Most of the rivers carry heavy silt loads at peak flo,w (e.g., the Sanyati in February and March). The Zambezi does not normally carry a heavy silt load unless its tributarics, such as the Deka (in January 1968) or Gwaai (March 1967), arc in spate. Water chemistry The overall chemical nature of the lake is related essentially to bottom topography an d river inflow. 783 KARIBA of basin in two divisions Jul 393 407 (Turnover) Dissolved Aug scp 806 530 673 410 5 during Ott 440 317 oxygen and hydrogen 2067 Nov DW 473’ 455 680 481 sulfide Dissolved oxygen values of the surface water in the open lake arc generally bctween 6 and 7 ppm, rising in the shallow cleared areas, whcrc greater water movemcnt and increase. in photosynthetic activity by plankton and aquatic weeds may be found. River water is usually low in oxygen as a result of the barrier created by Sa~lvinda mats at the air-water interface, the lack of photosynthesis beneath the mat, and the uptake of oxygen by the suspended organic matter in the river water. The highest dissolved olxygcn values rccorded were from the cold water of the Zambczi in July, when these values rose to 10 ppm, rcndcring the water supersaturated ( 115% saturation). After turnover and circulation, oxygen penctratcd to the bottom, of the lake and initially was distributed evenly from top to bottom. With rcstratification, oxygen dcplction began in the hypolimnion and continued until a sharp oxyclinc dcvclopcd at the thermocline. Cache (1968) remarked on the same phenomenon encountered during his studies. In November, oxygen depletion in the hypolimnion is more advanced in basins 1 and 2 than in basin 5--another indication of the progressive phases of rcstratification. Table 5 clearly indicates that surface oxygen values decrcasc at turnover as a result of mixing. IIigh oxygen tensions rarely persist in the hypolimnion for more than 4 months. This is related to the rate of restratification, whereby water exchang,es between hypolimnion and epilimnion arc rcduccd. The bo,ttom water gradually bccomcs co;Mpletcly deoaygenated, finally bcco’ming utterly reduced and 112s bearing within a 784 G. W. TAESLE 5. Rate of oxygen depletion 1967 0 20 40 60 80 100 Thermocline depth (In) Jun Jul 6.6 6.0 IIZS H,S HzS Has 5.1 5.1 5.1 4.8 4.8 4.9 (turnover) 32 BEGG in Kariba Gorge (concentrations in ppm) hs Sep Ott Nov Dee 1968 Jan 6.7 ::: 6.6 5”*; ii-2 4:9 4:2 4.2 4.2 7.4 5.8 5.0 3.1 3.6 2.9 7.5 3.7 3.6 2.9 2.0 1.8 6.9 3.6 3.5 3.3 1.4 1.4 6.7 2.6 2.2, 2.0 0.4 0.7 Nil 5 7 S-10 7-11 9-13 few months after commplete dcoxygenation. In the S&&&-infested river localities, the water toward the bottom rapidly dcoxygcnates and here H$ first appears. The worst Salvinin-infested rivers were the first to have I-I& at the bottom of their old channels. As if in contradiction, it is commonly found in the relatively weed-free Sanyati River-where it may seriously affect movements of the anadromous fish spetics (Cache 1968). Thus, while the lake water is still oxygenated from the top to bottom and although oxygen depletion is in process, the river systems rapidly become deoxygenatcd and the majority of them H$ bearing. As oxygen depletion draws to completion in the lake deeps, the rains arrive, the rivcrs begin to flow, and the dense H2S-laden water is flushed into’ the lake where it scttles below the thermocline. The rivers now become fully oxygenated water bodies, while in the lake the reverse applies. The II$ phase at any locality therefore varies, depending on the time the rivers flow and on the magnitude of the flood. Because of the weak flow from the rivers after the poor 1967-1968 rains, H2S did not penetrate basin 5 farther than the sub-basin south of the Long Island-Redcliff ridge. Thus H,S did not appear in the Kariba Gorge vicinity in 1968, where the previous year it was found for 7 months. This, in itself, is a bottom topography influence on the Harding ( 1966) stated water chemistry. that the H$ period was growing shorter as the rotting vegetation on the lake floor diminished; by 1963 this period in Kariba Gorge existed only for 3 months. Cache ( 1968) reinforced this opinion. By the end of Cache’s work in 1965, I&S could be detected at only three localities in the hypolimnion. Today, the II&S period at any locality is a variable phenomenon. Some rivers will carry it during one season and not the next, and the period could vary from 2-4 months, depending on the degree of Sulvinia infestation and the arrival of floods. Hydrogen sulfide periods farther up the lake in basins 1 and 2 may also be short-lived as a result of the flushing process in May. The presence of H& in the water has several important effects. As a reducing agent, it assists in mobilizing certain ions, thus causing alkalinity and conductivity to increase in II&S-laden water. An increase in alkalinity and conductivity renders this water more dense; it therefore sinks and remains within deep depressions on the lake floor. Low pH values are often obtained where the I-1$-laden water is in contact with the oxygenated epilimnion (over the chemocline), This could bc due to the vertical distribution of carbon dioxide and bicarbonate o’r to the formation of unstable sulfurous acid. From a practical viewpoint, I-1$ renders the chlorine injections in the turbine water algicidally ineffective; those turbines drawing on HzS-laden water have overheating problems due to algal accumulation. Density currents Apart from those formed by the entry of Zambezi water, density currents from the Sanyati River were aIso apparent. As Fig. 5 indicates, the water fro,m upstream LIMNOLOGY 60 - 70 - OF enters the lake as a tongue or wedge at depth- In JanuV 1968 such an oxygenated body of water was noticed at 54 m. Above this, at 30-36 m, lay an H&S-laden wedge, In January 1967 an oxygenated wedge temporarily appeared at 30 m, bearing the silt-laden surface water characteristic of upstream areas. In the Gwaai River a deoxygenated wedge occurred at 10 m in January 1968. TL influence of bottom topography Dissolved oxygen values decline in the immediate vicinity of affluent rivers, as river water generally maintains lower dissolved oxygen values. The river water tends to remain confined to the old river channels beneath the lake and to follow their original courses to some extent. This is apparent when viewed from the air. In basin 5, the isopleths indicating lomw surfact dissolved oxygen show no,ticeable intrusions opposite the inflow of the Gachc Gache and Sanyati Rivers. The water in the Kariba sub-basin (north of the Long Island-Redcliff ridge) clearly has different characteristics, relating particularly to transparency (see Table 4)) caused by the lack of river inflows. Maps of surface dissolved oxygen reveal a pattern of isopleths that conform to the LAKE 785 KA-RIISA main features of bottom topography. The cleared areas of the Gache Gache, Sanyati East, and Sanyati West normally have high dissolved oxygen values and decreased water transparency because of their shallow nature. Over the basin 5 deeps, surface oxygen values decline. This pattern may alter slightly during the year according to the prevailing wind direction and flow from the affluent rivers. Depth is always significant. In basin 5, should oxygen values at 30 m be 2.0 ppm over deep water, at a locality only 30 m deep (c.g., west of Sanyati River) the water is completely deoxygenated at 30 m. The nearness of the bottom increases the sharpness of the chemical gradients within the water body, particularly in the last few meters above the lake floor. The shallow cleared areas of basin 5 may even show steep oxygen (7.8-U ppm,) and ~11 (9:07.2) gradients over 10 m. This is understandable while there is a thermocline in the shallow margins of the lake, but it d oes indicate a. high degree of biological activity on and within the bottom muds. . ..PH The epilimnetic waters are normally distinctly alkaline, with pH between 7.8-8.5 and occasionally rising to 9.0. During advanced stages of stratification, the water of the hypolimnion ( generally 7.3-7.0) may become acid and pH may fall as low as 6.3. Surface water pH values during 1967 were considerably higher than 7.2-7.5 as found by Harding ( 1966). These values decrease as one passes from, basin 5 to basin 1, but most noticeably after Chete Gorge. These findings are in close agreement to those of Cache ( 1968)) indicating a similarity between the situation in 1967 and that 2 years before. Total alkalinity Alkalinity of the surface water (as ppm CaC03) in basin 5 in 1967 varied within 41-43 ppm. Again these results are closely related to those of Cache ( 1968). In pro-, ceeding from basin 5 to 1, alkalinity values, 786 G. W. fall into the 20-30 ppm range, indicating the poorness of the Zambczi water, which complctcly modifies basins 1 and 2. The highest alkalinity values ( lo& ppm ) were recorded from the Sanyati River in flood in January 1968. From the same locality, Cache recorded values of 196 ppm in August 1965, when the river could not have been flowing. Generally alkalinity falls with increase in depth-but may rise owing to the prescncc of Has-laden water or the nearness of the bottom. In IIZSladen water, the alkalinity may rise to 56 ppm. In basins 1 and 2, alkalinity is lowest in May ( 19 ppm ) during the period of peak Zambezi flow. As these arcas assume lakelikc characteristics under the influence of less flow from the Zambczi and modification from the basin 3 back-pressure, alkalinity values rise to 35 ppm by November. Total hardness of the surface water near the dam wall, as shown by Harding (1966>), fell from 49 ppm CaCOR in 195S to 35 ppm by 1964. Cache ( 1968) generally records slightly lower values (32.0-34.6 ppm) for the outflowing water in 1965. Conductivity Conductivity values drop in proceeding from basin 5 (9g105 pmhos/cm) to basin 1 (55-75 pmhos/cm). Cochc (1968) noted an increase in conductivity from basin 5 to 1 in January 1966. Harding (1966) recorded values of specific conductivity in 1964 of 76 pmhos/cm, having decreased steadily from 121 pmhos/cm in 1958. These values have increased to over 90 pmhos/cm by 1965 ( Cache 1968) in the surface waters of basin 5, and further in 1967 to values betwccn 95 and 100 pmhos/cm. The lowest value obtained was 49 pmhos/cm, from the Zambczi at peak flow in May; .thc highcst, 240 pmhos/cm, was from the Mlibizi River. Related to stratification, conductivity in the lake water normally decreases with increase in depth but may rise near the bottom, either because of the prcscnce of I&S or bccausc of the nearness of the bottom sediments. BEGG Mineral springs are common over the lake floor, particularly in basins 1 and 2 ( Mauffc 1933). A sample taken from the vicinity of a mineral spring, in Devil’s Gorge, had a conductivity of well over 1,000 pmhos/cm and pH of 3.5. In the Mulolo River (on the north bank of the lake) Cache (1968) recorded a mineral spring with water of a total hardness of 623 ppm CaCO:,. The conductivity of the affluent water is higher than that of the lake water due to its silt- and nutrient-laden state. Conductivity increases in rivers, such as the Sanyati, when they are flo,wing, so that values may rise to 205 pmhos/cm as in January 1968. Cache has recorded values of more than 300 pmhos/cm from the Sanyati River in flood. Salvinia The area of Salvinia in June 1967 was 82,000 ha, covering 15% of the lake surface ( Mitchell, personal communication), and was restricted entirely to rivers, sheltered bays, and uncleared areas. In such confines, a complex of plants grow as sudd on top of the Snlviniu mats (mainly comprised of scirpzls cubenasis) . Recently Lemna and Pistia stratioltes have appeared in the upper reaches of the old Sanyati River area, together with a dense Salvinia mat. The Eern Sdvinia auriculuta indirectly and directly plays an important role in many hydrobiological aspects of tho lake. Nutrient absorption and subsequent rclcasc during decomposition results in retention of nutrients within the lake. Further, dcnsc Salvinia tends to cause up to 25% dcoxygenation by decreasing photosynthetic activity and preventing direct atmospheric reoxygenation of surface water. Records oE water surface temperatures taken dircc tly beneath Salvinin: mats indicate so,me degree of insula .tion. A permanent stable mat may lower surface temperatures appreciably in certain localities, and, likewise, the water tempcraturcs beneath Salviniu may remain at a higher temperature than the. cooler lake water that is free of Salviniu. Salviniu mats increase water transparency by restricting water movement and by preventing plankton development. In addition, exclusion of light by permanent Salviniu prevents development of rooted vascular plant cofmmunities ( Lagarosiphon, Vallisneriu, Potamogeton, Ceratophyllum ) . It seems that the production of H$ is related to the quantity of rotting Salvinia present in the old riverbeds, forming a substrate on which I12S-forming bacteria thrive, Naturally, there is still a certain amount of H$ formation in the lake deeps, caused by rotting and decay of submerged trees-this is hoswever no, longer the primary source of this gas. Certain fish species (Tilapia mossambica, Tilapia melanopleura, Hemihaplochromis philander, Al&es imberi, Distichodus mossambicus, Labeo coqpro, Clarias gawkpinus, Gnuthonemus mucrokpidotus, and Cyphomyrus discorhynchus) have been known to ingest Salviniu, cithcr inadvertently in quest of the periphyton or to feed directly on various parts of the plant. In completely dcaquaria, T. melanopleura vour the roots and leaves of SaZvinia. As indicated by 13.G. Donnelly (personal communication) , the protection afforded by shoreline Salvinia mats to juvenile cichlids as primary nursery areas is of great importance. Equally important from the hydrobiological aspect is the community living within Salvinia (Molluscs, Crustacea, Annelida, Nemertca, Protozoa, Hydrachnida, Insecta, Algae, and diatoms ) . Biologically, the presence of impcnetrable mats of Salvinia and sudd renders the river systems less hospitable for occupation by anadromous fish and possibly interferes with the movements of adult fish on their way upstream to spawn and juveniles rcturning to the lake before the rivers dry up. These rivers, as a result of the impounded water, have already been drastically reduced in extent, Should sudd and SaEvinia enter the main body of the lake as a result of river flow or wind action, these masses of vegetation have a beneficial effect when blown up against the lake shore where they become available as fod- der and cover. Under this, due to wave action, the water quality is not affected as drastically as in the permanently suddchoked rivers (such as the Sengwa, Bumi, Scnkwc, and Luzilukulu ), DISCUSSION Chemical analyses show variation in the vicinity of each affluent river, and values of dissolved salts tend to decline as one approaches the head of the lake where the Zambezi influence becomes most strongly felt. The fall in chemical values rcflcct the low concentrations characteristic of the Zambczi water, and the influence of the Zambezi causes turnover to) occur at the head of the lake before winter. Turnover is thus brought about by a flushing mechanism and is not temperature induced as in basins 3, 4, and 5. It is interesting to foillow the increase in dissolved solids as the lake filled ( Harding 1966). With the inundation of new land and the associated leaching and decaying processes, dissolved solids rose from 26 ppm in June 1950 to 65 ppm, in 1958. When the lake level stabilizcd in 1964, dissolved solids had fallen to 42 ppm in the vicinity of the dam site. Harding predicted that these figures would continue to decline after 1964, until the lake water took on the characteristics of the inflows, But conductivity (as an indcx of dissolved salts) has risen from 76 pmhos/cm in 1964 to 95-100 pmhos/cm in 1967. Kariba will never bccomc a water body of uniform characteristics. There arc marked changes in the vicinity of affluent rivers, and in the Zambezi-modified “river basins” 1 and 2. In this respect, as indicated by Cache ( 1968), Kariba reflects the characteristics of a typical reservoir. The gradual dcclinc from 1960-1964 in the period of deoxygenation (Harding 1966) has not continued; the length of the period is variable, depending on locality within the lake and seasonal conditions. IIydrogcn sulfide was prcscnt for 7 months of 1967 at Harding’s sampling locality and was totally absent during 1968 (but see footnote 4, p. 780). 788 G. W. BEGG Harding ( 1966)) Coulter ( 1967)) and Cache ( 1968) also realized that the affluent rivers had a marked influence on the productivity of the lake. Seiche movements are slight and insignificant. Currents exist at localities where different bodies of water flow as wedges at depth, as well as in Kariba Gorge when the flood gates are open. The Zambezi-modified water from basins 1 and 2 extends along the bottom of basin 3 as a density current. As indicated by Harding ( 1966), hydrobiological investigations should start with the process of filling any new impoundment. Cabora Bassa, another lake on the Zambezi now under construction some 400 km below Kariba, will be longer than Kariba, of greater depth, narrower, and have a steep, shelving shoreline. It will, therefore, in all probability have riverine characteristics such as those in basins 1 and 2 of Kariba. It is hoped that the lesson learnt at Kariba, that preimpoundment surveys an d a co,ntinuity of limnological research arc necessary, will be heeded in this case. REFERENCES COCEIE,A. G. 1965. Limnological research program for Lake Kariba, p, 63-65. In Record of the Kariba Res. Symp., Lake Kariba Fish. Res. Inst., June 1965. ( Mimeographed. ) 1968. Description of physico-chemical -. aspects of Lake Kariba, an impoundment in Fish. Res. Bull. Zambia Zambia-Rhodesia. 5: 200-267. 1967. What’s happening at COULTER, G. w. Kariba. New Sci. 37: 750-753. HARDING, D . 1962. Research of Kariba, p. 3240. In Joint Fish. Res. Organ. Annu. Rep. 10, 1960. Govt. Printer, Lusaka. 1964. Research on Lake Kariba, p. 25-. 50. In Joint Fish. Res. Organ. Annu. Rep. 11, 1961. Govt. Printer, Lusaka. Lake Kariba. The hydrology -. 1966. In R. and development of fisheries, p. 7-20. I-1. Lowe-McConnell [ea.], Man made lakes. Symp. Inst. Biol. 15. Academic. MAUJTE, H. B. 1933. A preliminary report on the mineral springs of Southern Rhodesia. Geol. Surv. Bull. 23, p. 19-27.