Response of the East African Climate to Orbital Forcing During the Last Interglacial (130–117 Ka) and the Early Last Glacial (117–60 Ka)

Response of the East African climate to orbital forcing during the last interglacial (130–117 ka) and the early last glacial (117–60 ka)

Martin H. Trauth* Institut fu¨r Geowissenschaften, Universita¨t Potsdam, Postfach 601553, Potsdam, Germany Alan Deino Berkeley Geochronology Center, 2455 Ridge Road, Berkeley, California 94709, USA Manfred R. Strecker Institut fu¨r Geowissenschaften, Universita¨t Potsdam, Postfach 601553, Potsdam, Germany

ABSTRACT Variations in the temporal and spatial distribution of solar radiation caused by changes in Earth's orbit provide a partial explanation for observed long-term ¯uctuations in Af- rican lake levels. The understanding of causal links between insolation changes and lake- level ¯uctuations is essential for the design of models predicting future changes in the hydrological budgets and water supply in Africa. Here we present a record of climate change in East Africa between 175 and 60 ka. This time span includes the last interglacial (the Eemian, 130±117 ka), which may provide the closest analogue to the present inter- glacial. Assessments of the nature and timing of East African climate changes are based on lake-level ¯uctuations of Lake Naivasha (Kenya) inferred from sediment characteris- tics, diatom assemblages, and 40Ar/39Ar dating. Our results show dramatic alternation between deep, freshwater and shallow, highly alkaline lake conditions. The Lake Naivasha record demonstrates that periods of increased humidity in East Africa mainly follow pre- cessional insolation forcing in spring, causing more intense April±May rains every 23 k.y.

Keywords: East Africa, lake sediments, Milankovitch, climate.

INTRODUCTION forcing (Street and Grove, 1979; Gasse et al., vide a detailed chronology of lake-level Present African climate is mainly controlled 1989; Bonneville et al., 1990). A series of changes and East African climate between 175 by the strength of the African-Asian mon- abrupt events of increased precipitation and and 60 ka (Fig. 1) (Trauth and Strecker, 1996). soonal circulation and the position and sea- temperature may re¯ect complex interactions Together with the two most relevant African sonal migration of the Intertropical Conver- between orbital forcing, atmosphere, ocean, records from Lake Abhe in northern Ethiopia gence Zone (Nicholson, 1996). Whereas most and land-surface conditions (Gasse, 2000). (ca. 75 ka to present; Gasse, 1977) and the of the northern and southern part of the con- Marine records (e.g., Rossignol-Strick, 1983; Pretoria Saltpan in South Africa (ca. 200 ka tinent is characterized by monsoonal climates Clemens and Prell, 1990; deMenocal et al., to present; Partridge et al., 1997), the Naiva- with summer rains and winter droughts, the 1993; Zahn and Pederson, 1991) as well as sha lake history for the ®rst time provides pa- main source of rainfall in equatorial East Af- continental records (Gasse, 1977; Partidge et leoclimate data in East Africa extending this rica is the Intertropical Convergence Zone al., 1997) suggest antiphase changes in sum- far back in time. In addition, this record pro- (McGregor and Nieuwolt, 1998). The zone of mer insolation and enhanced rainfall in the maximum rainfall follows the latitudinal po- northern and southern monsoonal belts of Af- sition of the overhead sun with a time lag of rica as predicted by orbital precession geom- ϳ4±6 weeks resulting in two rainy seasons, etry (Berger, 1978). Detailed and well-dated the long rains around April±May and the short records from equatorial Africa documenting rains around October±November. pre±23 ka ¯uctuations of the intertropical con- During the late Pleistocene, African climat- vergence and convective rainfall on the equa- ic changes on time scales of thousands of tor are rare. These records suggest a period of years apparently were paced by periodic (23± 19 k.y.) variations in insolation resulting from high lake levels and increased humidity ca. Earth's orbital precession (Rossignol-Strick, 135 ka, well before the onset of the monsoon 1983; Kutzbach and Street-Perrott, 1985; system (Butzer et al., 1969; Sturchio et al., Clemens et al., 1991). Low lake levels in 1993; deMenocal et al., 1993; Zahn and Ped- northern and eastern Africa during the last gla- erson, 1991). cial maximum (23±18 ka) that followed high- New high-resolution paleohydrological and radiometric age data from lake deposits and stands ca. 15 ka are consistent with orbital Figure 1. Map of Kenya Rift showing loca- pyroclastic intercalations in the closed Nai- tions of Lake Turkana, paleo±Lake Suguta, *E-mail: [email protected]. vasha basin (central Kenya Rift Valley) pro- Lake Naivasha, and Lake Magadi.

᭧ 2001 Geological Society of America. For permission to copy, contact Copyright Clearance Center at www.copyright.com or (978) 750-8400. Geology; June 2001; v. 29; no. 6; p. 499±502; 3 ®gures. 499 Bulk mineralogy was obtained from X-ray dif- shallow freshwater lake. The diatomite bed is fraction using standard techniques. Anortho- overlain by several beds of unaltered pyro- clase and sanidine phenocryst concentrates clastic material, 107 Ϯ 7 ka, re¯ecting sus- from several tuff beds (ignimbrites, air-fall tained freshwater conditions. However, in the tuffs, and reworked tuffs) and one lava ¯ow upper part of this unit, perlitic cracks in glass were dated by the laser total-fusion (Deino shards, abundant phytoliths, and erosional un- and Potts, 1990; Deino et al., 1998) and laser conformities herald a major lake regression 40 39 (CO2) incremental-heating Ar/ Ar dating and a return to more alkaline conditions. Sub- methods (Sharp and Deino, 1996). Silicic py- aerial conditions appear to have been attained roclastic deposits in the pro®le allow an eval- at least along the southern periphery of the uation of lake alkalinity by means of the state basin before an ignimbrite dated as 106 Ϯ 4 of alteration of volcanic glass and the presence ka covered this unit. Waterlaid and partly re- of authigenic silicates (Utada, 1966; Surdam worked pyroclastic deposits containing vol- and Sheppard, 1978). Authigenic silicates canic glass shards with perlitic cracks and such as zeolites and potassium feldspar are montmorillonite rims indicate deposition in a common in ephemeral water bodies in arid or freshwater lake with a pH between 8 and 9, semiarid regions, where high pH conditions equated with lake-level highstand III. The age result from concentration of sodium carbon- of this highstand is de®ned by three interca- ate-bicarbonate due to evapotranspiration lated and relatively unaltered tuffs dated as 92 (Surdam and Sheppard, 1978). Vertical varia- Ϯ 3, 90 Ϯ 2, and 88 Ϯ 4 ka. A subsequent tions of these authigenic mineral phases in the lake-level lowstand and short periods of inter- Naivasha sediment sequence thus de®ne lake- vening subaerial conditions are indicated by level ¯uctuations and alkalinity changes mud cracks and impact marks of air-fall pum- through time. Freshwater phases are charac- ice lapilli in overlying strata, 91 Ϯ 5 ka. To- terized by diatomite, unaltered volcanic glass, ward the top, three diatomite beds, to 16 cm and absence of authigenic silicates. In con- thick and intercalated by 80 Ϯ 4 ka pyroclas- trast, silicic glass with perlitic cracks, glass tic material, document highstand II. The dia- shards with montmorillonite rims, occasional tom assemblages are characterized by Fragi- chabazite, and phillipsite represent the transi- laria construens, F. brevistriata, and F. tion to alkaline conditions with a pH of ϳ9. pinnata, re¯ecting shallow freshwater condi- Even higher alkalinity resulted in the forma- tions, but with a possible connection to a tion of clinoptilolite, and the most alkaline deeper lake in the north. However, the occur- pore waters led to the precipitation of rence of Thalassiosira faurii, Mastogloia smi- analcime. thi, M. elliptica, and Anomoenoneis sphaero- Figure 2. Sediment pro®le of Ol Njorowa phora contrasts with the older diatomites, Gorge showing major lithologic units and RESULTS suggesting slightly alkaline conditions with a inferred lake-level ¯uctuations based on pH just above 8. The deposits of highstand II diagenetic mineral phases and diatomites. Five lake-level highstands and intermittent lowstands can be distinguished in the pro®le, are overlain by several waterlaid tuff layers, above silicic lava dated as 320 Ϯ 20 ka and dated as 73 Ϯ 3 and 72.8 Ϯ 1.8 ka, that con- vides data for a longitudinal transect of late ca. 400 ka deposits of Longonot volcano tain various amounts of zeolites indicating Pleistocene climate-proxy records, now per- (Scott, 1980; Clarke et al., 1990), which higher alkalinity and lower lake levels. A re- mitting evaluation of the timing of tropical cli- formed the southern boundary of the Naivasha turn to prevailing freshwater conditions is mate response to orbitally induced insolation basin. The ®rst manifestation of a larger lake documented by a succession of waterlaid py- forcing across the equator. (highstand V) in this basin is indicated by 145 roclastic deposits containing unaltered glass Ϯ 2 ka ¯uvial deposits that grade into a 3.3- shards, indicating a pH Ͻ 9. These units are METHODS m-thick diatomite bed. A tuff within this di- interpreted to correspond to highstand I. Over- Situated at an elevation of 1890 m, Lake atomite yields an age of 140 Ϯ 3 ka. Predom- lying gray air-fall tuffs indicate subsequent Naivasha is the highest freshwater lake of the inant diatom taxa as Aulacoseira granulata subaerial conditions at 59 Ϯ 4 ka. An increas- rift (0Њ55ЈS, 36Њ20ЈE). Annual precipitation in and A. goetzeana suggest a large lake with pH ing yellow coloration toward the top and the basin is 750 mm/yr, whereas potential Ͻ 8. A lake-level drop is indicated in the up- smaller amounts of analcime in the highest evapotranspiration reaches 1600 mm/yr. The per part of the diatomite bed by superseding parts of these strata as well as in the overlying lake is fed by rivers draining the as much as ¯uvial deposits, rare sponge spicules, phyto- lake beds and 60 Ϯ 2 ka yellow tuffs suggest 4000-m-high eastern rift shoulder, where rain- liths (Cyperaceae), and diatom fragments, a ®nal short lake transgression characterized fall exceeds 1200 mm/yr (Clarke et al., 1990). which are topped by a 112 Ϯ 2 ka tuff. The by high alkalinity before the lake disappeared. We analyzed a 60-m-thick sedimentary se- inference of a major drop in lake level coupled quence in the Ol Njorowa Gorge at the margin with increasing alkalinity is substantiated by DISCUSSION AND CONCLUSIONS of the Naivasha basin, which was incised and the presence of corresponding authigenic sil- Because the Naivasha basin was not affect- exposed after the last lake-level highstand ca. icates. This lowstand is superseded by high- ed by important tectonic movements or vol- 10±9 ka (Washbourn-Kamau, 1977). The pa- stand IV, indicated by the deposition of a canic activity after volcanic closure of the ba- leoecology of the basin was reconstructed us- prominent 1.6-m-thick laminated diatomite sin ca. 320 ka (Clarke et al., 1990; Strecker et ing sedimentary characteristics and silica- bed. Diatom species such as Gomphonema an- al., 1990), the water-level ¯uctuations record- algae (diatom) assemblages (Gasse, 1986). gustatum and G. gracile indicate a large, but ed in the Ol Njorowa Gorge sediments (Fig.

500 GEOLOGY, June 2001 Such results would provide important con- straints for future modeling and prediction of African climatic conditions and hydrological budgets.

ACKNOWLEDGMENTS This project was funded by two grants to Streck- er and Trauth by the German Research Foundation. We thank the government of Kenya and the Kenya Wildlife Service for research permits. We also thank the editor D. Fastovsky and two anonymous review- ers for their helpful comments on the manuscript. We gratefully acknowledge F. Gasse for help with diatom taxonomy and interpretation as well as for many inspiring discussions. We also thank A. Berg- ner, S. Clemens, B. Dong, J. Kutzbach, S. Laue, G. Muchemi, P. deMenocal, F. Sirocko, W. Smykatz- Kloss, and R. Zahn for discussions.

REFERENCES CITED Berger, A., 1978, Long-term variations of daily in- solation and Quaternary climatic changes: Figure 3. Correlation of East African high lake levels and equatorial insolation changes Journal of Atmospheric Science, v. 35, during late Pleistocene. Data sources: (1) Hillaire-Marcel et al. (1986), (2) this work, (3) p. 2362±2367. Sturchio et al. (1993), (4) Butzer et al. (1969). Insolation data are from Berger (1978). Bonne®lle, R., Roeland, J.C., and Guiot, J., 1990, Temperature and rainfall estimates for the past 40 000 years in equatorial Africa: Nature, v. 346, p. 347±348. 3) clearly re¯ect changes in the precipitation November rains controlled by variations in Butzer, K.W., Brown, F.W., and Thruber, D.L., 1969, Horizontal sediments of the lower Omo val- and evaporation balance. The ®rst highstand equatorial insolation during fall. ley; the Kibish formation: Quaternaria, v. 11, of Lake Naivasha, between 145 and 120 ka Summarizing the ®rst continous and pre- p. 15±29. with a maximum ca. 135 ka, correlates with cisely dated East African climate record for Clarke, M.C.G., Woodhall, D.G., Allen, D., and highstands in other rift-valley lakes between the last interglacial and the ®rst half of the last Darling, G., 1990, Geological, volcanological and hydrogeological controls on the occur- 140 and 120 ka, such as paleo±Lake Suguta glacial, the observed correlation of water-level rence of geothermal activity in the area sur- (Sturchio et al., 1993), 135 ka high shorelines ¯uctuations and insolation changes appears rounding Lake Naivasha, Kenya: Nairobi, in the Magadi-Natron basin (Hillaire-Marcel compatible with orbital forcing theory (Ber- Kenya, Ministry of Energy, 138 p. et al., 1986), and a highstand of Lake Turkana ger, 1978). As predicted by orbital precession Clemens, S., and Prell, W., 1990, Late Pleisto- (Butzer et al., 1969). However, paleoecologic geometry, East African humidity signi®cantly cene variability of Arabian Sea summer monsoon winds and continental aridity: Eo- information and age control of associated de- increased ϳ8 k.y. before the onset of the Af- lian records from the lithogenic component posits are rather limited. The younger periods rican-Asian monsoon system as reported from of deep-sea sediments: Paleoceanography, of increased humidity at ca. 110, 90, 80 and marine records (e.g., Rossignol-Strick, 1983; v. 5, p. 109±146. 66 ka as recorded in the Naivasha sediments Clemens and Prell, 1990; deMenocal et al., Clemens, S., Prell, W., Murray, D., Shimmield, G., and Weedon, G., 1991, Forcing mechanisms have not been reported from other East Afri- 1993). The most relevant continental record of the Indian Ocean monsoon: Nature, v. 353, can lake basins. There is no evidence for high- for North African climate from Lake AbheÂin p. 720±725. er lake levels between 66 ka and a Holocene northern Ethiopia (Gasse, 1977) only extends Deino, A., and Potts, R., 1990, Single-crystal 40Ar/ 39 highstand of Lake Naivasha between ca. 10 back to ca. 70 ka, and there is good age con- Ar dating of the Olorgesailie Formation, Southern Kenya Rift: Journal of Geophysical and 9 ka (Washbourn-Kamau, 1977; Richard- trol only back to ca. 40 ka. The Pretoria Salt- Research, v. 95, no. B6, p. 8453±8470. son and Richardson, 1972). pan time series of summer precipitation in Deino, A.L., Renne, P.R., and Swisher, C.C., 1998, High lake levels at 135, 110, 90, and ca. 66 South Africa covers the past 200 k.y. (Par- 40Ar/39Ar dating in paleoanthropology and ar- ka precisely match maximum spring insola- tridge et al., 1997), but provides radiocarbon chaeology: Evolutionary Anthropology, v. 6, p. 63±75. tion at the equator without any signi®cant lag ages only back to ca. 43 ka. The lower part deMenocal, P., Ruddiman, W.F., and Pokras, E.M., (Berger, 1978). The magnitude of the high- of that record was tuned using a ®ssion-track 1993, In¯uences of high- and low-latitude pro- stands correlates with the amplitude of inso- age for the formation of the Saltpan basin of cesses on African terrestrial climate: Pleisto- lation curve. This suggests that orbitally in- ca. 220 Ϯ 52 ka. Both records are therefore cene eolian records from equatorial Atlantic duced changes in summer solar radiation dif®cult to use for a correlation with the Nai- Ocean Drilling Program Site 663: Paleocean- ography, v. 8, p. 209±242. certainly account for a large part of hydrolog- vasha record. Gasse, F., 1977, Evolution of Lake Abhe (Ethiopia ical changes in East Africa. Such radiation There are still large gaps in climate infor- and TFAI), from 70 000 B.P.: Nature, v. 265, changes could cause a relatively large increase mation for Africa for the time before the last p. 47±45. in spring temperatures over land and therefore glacial maximum. Taking into account the re- Gasse, F., 1986, East African diatoms: Taxonomy, ecological distribution: Berlin, J. Cramer, an intensi®cation of the intertropical conver- ported inconsistencies in the timing of hydro- 201 p. gence and convective rainfall in this region. logical changes explained by radiation chang- Gasse, F., 2000, Hydrological changes in the Af- The long duration of the earliest wet period es, ¯uctuations in sea-surface conditions, and rican tropics since the last glacial maxi- between 145 and 120 ka as well as a lake- ocean-atmosphere interactions, further long mum: Quaternary Science Reviews, v. 19, p. 189±211. level highstand ca. 80 ka may have been high-resolution climate studies are needed for Gasse, F., LeÂdeÂe, V., Massault, M., and Fontes, J.C., caused by interference between the weakening elucidating large-scale atmospheric dynamics 1989, Water-level ¯uctuations of Lake Tan- April±May rains and more intense October± during the last two glacial-interglacial cycles. ganyika in phase with oceanic changes during

GEOLOGY, June 2001 501 the last glaciation and deglaciation: Nature, tory of an African rift lake and its climatic Trauth, M.H., and Strecker, M.R., 1996, Late Pleis- v. 342, p. 57±59. implications: Ecological Monographs, v. 42, tocene lake-level ¯uctuations in the Naivasha Hillarie-Marcel, C., Carro, O., and Casanova, J., p. 499±534. Basin, Kenya, in Johnson, T.C., and Odada, 1986, 14C and Th/U dating of Pleistocene Rossignol-Strick, M., 1983, African monsoons, an E.O., eds., The limnology, climatology and and Holocene stromatolites from East Afri- immediate climate response to orbital insola- paleoclimatology of the East African lakes: can paleolakes: Quaternary Research, v. 25, tion: Nature, v. 304, p. 46±49. The international decade for the East African p. 312±329. Scott, S.C., 1980, The geology of Mt. Longonot lakes (IDEAL): Newark, New Jersey, Gordon Kutzbach, J.E., and Street-Perrott, F.A., 1985, Mil- volcano, central Kenya; a question of vol- & Breach, p. 549±557. ankovitch forcing of ¯uctuations in the level umes: Royal Society of London Philosophical Utada, M., 1966, Zeolites in sedimentary rocks, of tropical lakes from 18 to 0 ka: Nature, Transactions, v. 296, p. 437±465. with reference to the depositional environ- v. 317, p. 130±134. Sharp, W., and Deino, A.L., 1996, CO2 laser heating ments and zonal distribution: Sedimentology, McGregor, G.R., and Nieuwolt, S., 1998, Tropical of samples for Ar-Ar geochronology: Eos v. 7, p. 237±257. climatology: An introduction to the climates (Transactions, American Geophysical Union), Washbourn-Kamau, C.K., 1977, The Ol Njorowa of the low latitudes: Chichester, John Wiley & v. 77, p. F773. Gorge, Lake Naivasha Basin, Kenya, in Sons, 311 p. Strecker, M.R., Blisniuk, P.M., and Eisbacher, Greer, D.C., ed., Desertic terminal lakes: Lo- Nicholson, S.E., 1996, A review of climate dy- G.H., 1990, Rotation of extension direction gan, Utah, Utah Water Research Laboratory, namics and climate variability in eastern Af- in the central Kenya Rift: Geology, v. 18, p. 297±307. rica, in Johnson, T.C., and Odada, E.O., eds., p. 299±302. Zahn, R., and Pedersen, T.F., 1991, Late Pleistocene The limnology, climatology and paleoclima- Street, F.A., and Grove, A.T., 1979, Global maps of evolution of surface and mid-depth hydrogra- tology of the East African lakes: The inter- lake-level ¯uctuations since 30 000 yr B.P.: phy at the Oman Margin: Planktonic and ben- national decade for the East African lakes Quaternary Research, v. 12, p. 83±118. thic isotope records at Site 724, in Proceedings (IDEAL): Newark, New Jersey, Gordon & Sturchio, N.C., Dunkley, P.N., and Smith, M., 1993, of the Ocean Drilling Program, Scienti®c re- Breach, p. 25±56. Climate-driven variations in geothermal activ- sults, Volume 117: College Station, Texas, Partridge, T.C., deMenocal, P.B., Lorentz, S.A., ity in the northern Kenya rift valley: Nature, Ocean Drilling Program, p. 291±308. Paiker, M.J., and Vogel, J.C., 1997, Orbital v. 362, p. 233±234. forcing of climate over South Africa: A Surdam, R.C., and Sheppard, R.A., 1978, Zeolites Manuscript received October 24, 2000 200 000-year rainfall record from the Pretoria in saline, alkaline lake deposits, in Sand, L.B., Revised manuscript received February 8, 2001 Saltpan: Quaternary Science Reviews, v. 16, and Mumpton, F.A., eds., Natural zeolites: Oc- Manuscript accepted February 19, 2001 p. 1125±1133. currence, properties, use: Oxford, Pergamon Richardson, J.L., and Richardson, A.E., 1972, His- Press, p. 145±175. Printed in USA

502 GEOLOGY, June 2001