Science in Africa
Emerging Water Management Issues

When I began to study African inland waters the world seemed a much more stable and reliable place than it does today. We thought that the climate of historic times was the norm. The Quaternary ice ages seemed an anomaly, a temporary aberration when aquatic systems were not in normal equilibrium with their environment. The anomaly demanded explanation because it was anomalous. We sought a cause for the ice ages.

Today geologists have a very different worldview. Not even the position of the continents is fixed, although Africa is less skittishly inclined to skate around than most. We realize that the highest mountains and the deepest seas change their architecture at rates of centimeters per year. We know that for most of the last million years ice-age conditions have been the norm, that the relatively warm and ice-free present is the anomaly. At no time in the accessible history of the earth have conditions of climate been so stable as they seemed 50 years ago.

Until recently this instability posed no real and present danger. Radiocarbon dating had put a figure of 10 or 15 thousand years on the length of post-glacial time. Pollen analysis suggested that we had passed the peak of an interglacial period, and that the world was sliding majestically back into another ice age. Perhaps in five thousand years we might be threatened by a new deep-freeze, but in terms of a human lifetime there were more pressing worries.

We were mistaken. Glaciers and forests cannot respond quickly to climatic change; they take centuries or millennia to grow and spread. They cannot respond instantly to even an instantaneous change in climate, but rapid changes in climate do occur, and we need fast responding indicators to detect them.

The first warning of rapid climatic change came from fossil beetles in Europe (Coope, 1977) and the isotopic composition of glacier ice in Greenland (Dansgaard et al, 1971 and 1989). Beetles showed that a surprisingly large part of the temperature change separating an ice age from an interglacial occurs within a century; ice cores showed that it occurs within 20 years. A substantial part of the total temperature change during the last oscillation occurred in jumps of about five years (Taylor et al, 1997). A change that takes a hundred years won't compete for human attention with school fees, war, or economic collapse, but it can concern a grandparent -- or a water manager. A change that occurs in five years is fast enough to worry anyone but a teenager or a politician.

Frequency analysis of long records from marine sediments suggests that there are, and have long been, variations in temperature with periods of about 20,000, 40,000 and 100,000 years (Imbrie and Imbrie, 1980). These periods correspond very suggestively to periods of irregularity in the earth's motion around the sun. The 100,000 year period, which has been dominant for the last 700,000 years, corresponds even more closely to the period with which the plane of the earth's orbit moves back and forth through a disk-shaped cloud of circumsolar dust (Muller and MacDonald, 1997).

That dust cloud might drive ice ages by shading all parts of the world from the sun's warming rays, but the other irregularities in earth's orbit do not change the total amount of radiation that reaches our planet; they change only its seasonal distribution over the earth's surface. In particular, they are likely to alter the circulation of the ocean to modulate partitioning of carbon dioxide between ocean and atmosphere. Carbon dioxide in air bubbles trapped in glacier ice shows that the atmospheric content of this greenhouse gas has changed enormously over the last glacial-interglacial cycle, and this change can explain much of the temperature fluctuation of ice ages (Broccoli and Manabe, 1987).

If the drivers of climatic change are so ponderously slow -- 20,000 to 100,000 years -- why are the temperature changes sometimes so fast? Probably because the ice-ocean-atmosphere system has certain preferred states, and the slow accumulation of astronomical change finally triggers a rapid shift from one of these states, more or less glacial, to another, more or less interglacial (Broecker, 1997).

Detailed analysis of ice and ocean records suggests that there are other periodicities of climate, not so regular as the glacial ones, and of smaller amplitude and higher frequency, on the order of 1,000 to 2,000 years. The driver for these changes has not been identified, but they might be caused by a free oscillation of the ocean-atmosphere-cryosphere system, or they might be driven by a quasi-periodic variation in the radiant energy output of the sun. This quasi-millennial periodicity was recognized first in the glacial part of the paleoclimatic record (Bond and Lotti, 1995), and was plausibly attributed to an auto-oscillation of the North American ice sheet. Closer investigation has revealed the same periodicity in the post-glacial part of the record as well (Kerr, 1996b), when surviving remnants of the ice sheet were too small to have had such global consequences. Some periodicity seems to prevail even in tropical Africa (Stager et al, 1997).

I favor the hypothesis of solar variation (Kerr, 1996a) but, whatever the cause of these quasi-millennial variations, they are most pronounced during the transition between glacial and interglacial conditions. During the last such transition, the quasi-millennial effect was sufficient to switch the earth back and forth, so that instead of a simple end to the last glaciation we had a temporary end, followed by a temporary reinstatement of glacial conditions, before entering the interglacial period that has now persisted for about 10,000 years.

My African friends may reasonably ask, what have these ice ages to do with us? Of course, they produced a few high altitude lakes on Ruwenzori, Mt. Kenya, and Kilimanjaro, but there are more serious reasons for African concern, especially for concern about African water resources.

We know from historical investigation of African lakes that they were not immune to the climatic changes that produced ice ages. They are as sensitive as glaciers to vagaries of climate. Even during a short historic period there have been great changes in the water level of the African Great Lakes, and greater changes in the level of smaller bodies of water such as Lake Naivasha (Nicholson, 1997). Lake Chilwa has been known to dry up completely in modern times. Harbor works on Albert, Victoria, and Tanganyika were seriously damaged by the wet years of the early 1960s. Lake Malawi rose to damage its principal fisheries laboratory at the same time. During dry episodes, the Volta River fails to sustain the reservoir at Akosombo in Ghana, or the Shire to turn the turbines that power Malawi.

In more remote times hydrologic changes were much greater. Lake Victoria seems to have been completely dry during a period that did not end before 12,500 years ago (Kendall, 1969; Stager et al, 1986; Johnson et al, 1996). The deepest part of Lake Naivasha in Kenya seems to have been dry during a brief period about 3,000 years ago (Richardson and Richardson, 1972). The presence of raised strand lines high above the water level of present lakes demonstrates clearly that there have been times when water was much more abundant than it is today (Washbourn-Kamau, 1970). To find evidence of times when water was scarcer one must core through sediments that are presently covered by water. In most lakes that have been cored there is compelling evidence of times much dryer than even the most severe known drought.

In general, the wettest periods in equatorial Africa came during early post-glacial time, and the driest during severe phases of the last ice age. That is, global climatic linkages tie the broad features of our climate to the rest of the world, with a tendency for Africa to be wet when temperate and arctic regions were warm -- but only a tendency. Nothing in glacial geology suggests that Lake Naivasha would be dry 3,000 years ago.

The other reason for African concern is that artificial increases in the greenhouse gases of the atmosphere have introduced a new complication into what was already a rather murky picture. By burning fossil fuel and converting forest land to cultivation we are increasing the carbon dioxide concentration of the atmosphere very rapidly beyond anything that prevailed during at least the last 150,000 years.

If there were a simple one-to-one relationship between atmospheric carbon dioxide and humidity, we might be entering a time of somewhat more abundant water. But carbon dioxide is only one part of the climatic system. Past changes in carbon dioxide were accompanied by changes in the seasonal distribution of solar energy and in oceanic circulation that do not prevail now.

We must have enough understanding of the processes that control climate to predict the likely consequences of increasing greenhouse gases without changing the global pattern of solar radiation. At present global circulation models are our best hope. These models are limited by the size and speed of computers to a resolution in pixels for areas that are some hundreds of kilometers on a side. Within the limits of their resolution they are very impressive. They rest on the known thermodynamic properties of the atmosphere and plausible boundary conditions, such as estimated sea surface temperature and solar radiant energy. They generate a pattern of global climate that agrees rather nicely with that based on instrumental records.

The models have some serious shortcomings. Their spatial resolution is too low to incorporate individual storms or clouds, or to take into account local variations of topography. They include such major features of the earth's shape as the Himalayas or the Andes, but African mountains are beneath their notice. If we use them to predict the water level in Lake Jipe or Amboseli, and they are not aware of Kilimanjaro, the prediction will not be reliable.

The models may be deficient in some even more serious respects, for they do not reconcile the apparent ice-age temperature of equatorial Africa with the apparent ice-age temperature of the tropical oceans around it.

Paleoceanographers believe that the sea surface temperature at high latitudes was much cooler during each ice age than it is now. At the height of the last glaciation, for example, the Atlantic ice pack, which presently lies north of Iceland, descended to the latitude of Portugal. At low latitudes, however, sea surface temperature seems to have changed very little.

Most paleothermometric evidence for the equatorial ocean is obtained in two ways. In one method, the assemblage of planktonic foraminifera in different parts of the ocean is compared with the sea-surface temperature to generate a transfer function relating the composition of the fauna to the temperature. This function is then applied to fossil assemblages down-core to estimate past sea surface temperature. It indicates temperature changes of only a degree or two in the tropical ocean on either side of Africa. There are possible problems with the argument. For example, periodic cooling may prevent the evolution of foraminiferal species requiring the warmth of the present tropical ocean, so the end-member of the series relating foraminiferal communities to temperature may be missing. Most oceanographers, however, have found past temperature estimates based on foraminiferal assemblages to be quite convincing.

In the other method, the ratio of ¹⁸O to ¹⁶O in planktonic foraminifera is used to estimate past temperature. The relation between ambient temperature and isotopic selectivity by foraminifera is well established, but the isotopic composition of sea water changes as more or less of the lighter isotope is locked up in glacier ice on land. A reasonable bound can be placed on the part of the isotopic change attributable to changes in the isotopic composition of sea water by measuring the composition of benthic species of foraminifera. The tropical deep sea is very close to freezing now, so it can never have been much colder, but bottom-dwelling foraminifera change almost as much in isotopic composition as surface dwellers. Apparently most of the change in oxygen isotopes of planktonic foraminifera can be attributed to the isotopic composition of sea water, and the remaining small part implies temperature change of a few degrees or less.

This argument rests implicitly on a very simple model of the worldwide isotopic composition of sea water. A somewhat more realistic model, taking account of plausible local differences in evapo-precipitation, weakens the argument and would reconcile the isotopic record with a change in sea-surface temperature of five degrees centigrade (Norton et al, 1997).

The alkenones of marine phytoplankton are believed to depend on the temperature at which the plankton grows, and down-core studies of changes in alkenones also indicate that surface temperatures in the tropical ocean did not change more than a degree or two (Lyle, 1992). On the other hand, both the composition of coral skeletons (Guilderson et al, 1994) and the distribution of hydroids in the western tropical and subtropical Atlantic (Deevey, 1950) seem more consistent with a temperature change of five degrees. Perhaps the sea surface temperature difference between an ice age and an interglacial period is greater in the shallow near-shore ocean than in the open water of the deep sea.

There are many indications that the temperature of equatorial Africa fell by at least five to six degrees during the last ice age. Snow-line lowering on Kilimanjaro, Mt. Kenya, and Ruwenzori would require a temperature that was some five degrees lower if there were no change in precipitation (Osmaston, 1975 and 1989); lake evidence showing much lower precipitation would indicate temperatures seven to eight degrees lower (Livingstone, 1993).

Most tropical African pollen analysts believe that their data demand a temperature lowering of five to six degrees. The presence of C-3 species of grasses near sea level in Ghana during the last ice age (Talbot et al, 1984) is difficult to understand if the temperature were not seven to nine degrees centigrade cooler than it is today (Livingstone, 1993). The similarity of the grasses growing at intermediate altitudes in the highland areas of Cameroon and East Africa is understandable if the temperature were 10 degrees lower over the intervening lowlands. Grasses are relatively mobile organisms, and the period or periods of such low temperatures might be quite brief, perhaps corresponding to troughs in the quasi-millennial periodicity. By contrast, the periods of cooling do not seem to have lasted long enough for the much slower-moving flightless insects to travel freely even between the separate mountains of East Africa (Bruehl, 1997).

If the open ocean conclusion that temperatures fell by only one or two degrees is valid, and if the shallow ocean and terrestrial conclusion that they fell by five to ten degrees is valid, there must be something wrong with the global circulation models, which will not reconcile a discrepancy of more than three degrees in the cooling of ocean and land. It is never wise to be dogmatic, especially where the evidence is mixed, but the difference in the ice-age cooling of land and water seems to have been greater than the general circulation models will accept. The most likely way out of the impasse is that the models are deficient, perhaps because of the way they parameterize the cloudiness of the ice-age world. In short, global circulation models are the best tool we have for predicting the effect of global warming on African water supplies, but they cannot yet be used with confidence to do so. In particular, they do not provide solid assurance that global warming will make Africa generally wetter. If the wetness of African interglacial periods is a fairly simple function of tropical sea-surface temperatures that may happen, but if it depends more on latitudinal temperature gradients -- for example, the gradients that drive the African and Asian monsoons -- those will not be sharpened by simply raising the temperature everywhere.

What significant consistencies emerge from historical study of African lakes? First, they are extremely variable. Lakes everywhere are notoriously evanescent, because geological processes tend to obliterate them with sediment or drain them by cutting through the sill that holds the water in place. Although some African volcanic lakes are of quite respectable age (indeed some tectonic lakes of the Rift, particularly Tanganyika and Malawi, are among the oldest in the world), many (like the glacial lakes of the high mountains, river flood plain lakes, and lakes formed by recent volcanism) are no older than the numerous glacial and thaw lakes of higher latitudes.

What does seem to set them apart from many lakes elsewhere is that they are very sensitive to climatic change. On more humid continents most lakes receive a large excess of water that runs off to sea. In Africa, water inflows commonly do not exceed evaporation greatly, and a slight change in rainfall can have a great effect on the level of a lake. The tacit hydrological assumptions of water managers elsewhere are likely to get us into trouble in even some of the more humid parts of Africa.

Although there is a broad-brush consistency between the history of African lakes and world climate on the time scale of ice ages and interglacial periods, and some suggestion that quasi-periodic changes with lengths of one or two millennia, known from polar glacier and deep-sea records, affect African lakes, there is no simple correlation between global climatic change and some very significant excursions of African lake level during the Holocene.

Understanding of the dynamics of global climate is not good enough for us to predict the consequences for African lakes of the climatic changes we are now producing by changing the concentration of greenhouse gases in the atmosphere. Historical study of African lakes offers two challenging possibilities. First, by further study of African lacustrine records we could greatly improve our understanding of past climatic changes. Especially, we need to test present beliefs about the amount of last ice-age cooling in Africa by examining the evidence in places where it has not been consulted, and by using kinds of temperature proxies that have not yet been widely applied. It is particularly critical to extend the available proxy record back through several ice ages, because the one for which we have data may not give us a proper appreciation of the dynamic possibilities of the system. It is also critical to look in great detail at the lacustrine record of the last millennium. In many stratified lakes that record should be annually or even seasonally resolvable, and would provide a valuable extension of the instrumental record, which is everywhere too short for estimating the probability even of droughts and floods.

Estimating probabilities is much less helpful than making reliable predictions, which may not be possible. In the instrumental and the paleolimnological records suggestions of periodicity are weak. Even the quasi-millennial periodicity of polar glaciers is not so regular as to provide a reliable basis for prediction. In tropical Africa climatic variability seems much less dominated by any single period, and so is inherently even less predictable.

Year-to-year reconstructions of lake conditions, however, would be valuable in quite another way. They would provide a measure of the natural background variability against which the consequences of artificial global changes must be perceived. The consequences of increasing greenhouse gases will not be felt equally in all parts of the world, nor will they be felt regionally in proportion to how great a region's contribution to the global problem has been. It would be prudent for African water managers to keep a sharp eye on how global change is affecting their waters, so they may speak up quickly and assuredly as soon as their resource is threatened.

I have emphasized lake level change because it is much easier to estimate past changes in lake level than past changes in river discharge, and because the presence or absence of a lake seems a matter of overarching importance. That is not to imply that other biological, physical, and chemical changes are trivial. A water resource can suffer serious damage long before it evaporates completely. Evaporative concentration may render a drinking water supply undrinkable, and changes much less severe than evaporation to dryness have a profound effect on fisheries resources.

For example, changes in Lake Victoria during the last 30 years have had a profound effect on the fish yield of the lake, but also on the kinds of fish that are taken and the means that must be employed in their taking. Among the plethora of possible contributors to this result there are several that could be tested by studying the history of African lakes.

In Uganda, Lake Albert has had Nile Perch throughout its known history, and probably for 12,000 years. That voracious predator, which is commonly credited with the changes in Lake Victoria, was introduced into Lake Kioga first, then into Lake Victoria. It has also been introduced into at least one smaller Ugandan lake, but has not yet reached Lake Edward.

It would be instructive to make a comparative paleolimnological study of these various lakes to see if such changes as replacement of diatoms and green algae by blue-greens, which did affect Lake Victoria, occurred in each of the other lakes into which Nile Perch had been introduced, and in the right order. On the other hand, if the changes were a result of increased atmospheric inputs of fixed nitrogen from fires, or changes in land use around Victoria, they should show a different pattern.

In conclusion, the written and sedimentary historical records of African lakes show that they are capricious, changing rapidly in area and depth over a wide range of time scales. Their natural variability will change in unpredictable ways by current modifications of atmospheric composition and transparency. They cannot be wisely managed on the assumption that they will long remain as they are now.

References

Bond, G., and Lotti, R., 1995. Iceberg discharges into the North Atlantic on millennial times scales during the last glaciation. Science 267:1005-1010.

Broccoli, A.J., and Manabe, S., 1987. The influence of continental ice, atmospheric CO₂, and land albedo on the climate of the last glacial maximum. Climate Dynamics 1:87-99.

Broecker, W.S., 1997. Thermohaline circulation, the Achilles Heel of our climate system: will man-made CO₂ upset the current balance? Science 278:1582-1588.

Bruehl, C.A., 1997. Flightless insects: a test case for historical relationships of African mountains. Journal of Biogeography 24:233-250.

Coope, G.R., 1977. Fossil coleopteran assemblages as sensitive indicators of climatic changes during the Devensian (Last) cold stage. Phil. Trans. R. Soc. Lond. B 280:313-340.

Dansgaard, W., Johnsen, S.J., Clausen, H.B., and Langway, Jr., C.C., 1971. Climatic record revealed by the Camp Century ice core. In The Late Cenozoic Glacial Ages, K. K. Turekian, ed. New Haven: Yale University Press.

Dansgaard, W., White, J.W.C., and Johnsen, S.J., 1989. The abrupt termination of the Younger Dryas climate event. Nature 339:532-534.

Deevey, Jr., E.S., 1950. Hydroids from Louisiana and Texas, with remarks on the Pleistocene biogeography of the western Gulf of Mexico. Ecology 31:334-367.

Guilderson, T.P., Fairbanks, R.G., and Rubenstone, J.L., 1994. Tropical temperature variations since 20,000 years ago: modulating interhemispheric climate change. Science 263:663-665.

Imbrie, I., and Imbrie, J.Z., 1980. Modeling the climatic responses to orbital variations. Science 207:943-953.

Johnson, T.C., Scholz, C.A., Talbot, M.R., Kelts, K., Ricketts, R.D., Ngobi, G., Beuning, K., Ssemanda, I., and McGill, J.W., 1996. Late Pleistocene dessication of Lake Victoria and rapid evolution of cichlid fishes. Science 273:1091-1093.

Kendall, R.L., 1969. An ecological history of the Lake Victoria basin. Ecol. Monogr. 84:227-232.

Kerr, R.A., 1996a. A new dawn for sun-climate links? Science 271:1360-1361.

----------, 1996b. Ice rhythms: core reveals a plethora of climatic cycles. Science 274:499-500.

Livingstone, D.A., 1993. Evolution of the African climate. In Biological Relationships Between Africa and South America, P. Goldblatt, ed. New Haven: Yale University Press.

Lyle, M., 1992. Molecules record sea change. Nature 356:385-386.

Muller, R.A., and MacDonald, G.F., 1997. Glacial cycles and astronomical forcing. Science 277:215-218

Nicholson, S.E., 1997. Extending the African lake record in the early 19th Century. IDEAL Bulletin, Winter 1997.

Norton, F.L., Hausman E.D., and McElroy, M.B., 1997. Hydrospheric transports, the oxygen isotope record, and tropical sea surface temperatures during the last glacial maximum. Paleoceanography 12:15-22.

Osmaston, H.A., 1975. Models for the estimation of firnlines of present and Pleistocene glaciers. In Processes in Physical and Human Geography, R. Pell, M. Chisholm, and P. Haggett, eds. London: Heinemann.

Osmaston, H.A., 1989. Glaciers, glaciations and equilibrium line altitudes on Kilimanjaro. In Quaternary and Environmental Research on East African Mountains, W.C. Mahaney, ed. Rotterdam: Balkema.

Richardson, J.L., and Richardson, A.E., 1972. The history of an African Rift lake and its climatic implications. Ecol. Monogr. 42:499-534.

Stager, J.C., Cumming, B., and Meeker, L., 1997. A high-resolution 11,400-yr diatom record from Lake Victoria, East Africa. Quaternary Research 47:81-89.

Stager, J.C., Reinthal, P.N., and Livingstone, D.A., 1986. A 25,000-year history for Lake Victoria, East Africa and some comments on its significance for the evolution of cichlid fishes. Freshwater Biology 16:15-19.

Talbot, M.R., Livingstone, D.A., Palmer, P.G., Maley, J., Melack, J.M., Delibrias, G., and Gulliksen, S., 1984. Preliminary results from sediment cores from Lake Bosumtwi, Ghana. Palaeoecol. Africa 16:173-192.

Taylor, K.C., Mayewski, P.A., Alley, R.B., Brook, E.J., Gow, A.J., Grootes, P.M., Meese, D.A., Saltzman, E.S., Severinghaus, J.P., Twickler, M.S., White, J.W.C., Whitlow, S., and Zielinski, G.A., 1997. The Holocene-Younger-Dryas transition recorded at Summit, Greenland. Science 278:825-827

Washbourn Kamau, C.K., 1970. Late Quaternary chronology of the Nakuru-Elmenteita Basin, Kenya. Nature 226:253-254.