26
Basin Research (1997) 9, 107–132 New palaeogeographic and lake-level reconstructions of Lake Tanganyika: implications for tectonic, climatic and biological evolution in a rift lake A. S. Cohen,* K.-E. Lezzar,² J.-J. Tiercelin² and M. Soreghan‡ *Department of Geosciences, University of Arizona, Tucson, AZ 85721, USA ²CNRS URA 1278 ‘Domaines Oce ´ aniques’, Groupe Riftoge ` nese Est-Afrique, Universite ´ de Bretagne Occidentale, B. P. 809, 29285 Brest Cedex, France ‡School of Geology and Geophysics, University of Oklahoma, Norman, OK 73019, USA ABSTRACT Palaeogeographic and lake-level reconstructions provide powerful tools for evaluating competing scenarios of biotic, climatic and geological evolution within a lake basin. Here we present new reconstructions for the northern Lake Tanganyika subbasins, based on reflection seismic, core and outcrop data. Reflection seismic radiocarbon method (RSRM) age estimates provide a chronological model for these reconstructions, against which yet to be obtained age dates based on core samples can be compared. A complex history of hydrological connections and changes in shoreline configuration in northern Lake Tanganyika has resulted from a combination of volcanic doming, border fault evolution and climatically induced lake-level fluctuations. The stratigraphic expression of lake-level highstands and lowstands in Lake Tanganyika is predictable and cyclic (referred to here as Capart Cycles), but in a pattern that diers profoundly from the classic Van Houten cycles of some Newark Supergroup rift basins. This dierence results from the extraordinary topographic relief of the Western Rift lakes, coupled with the rapidity of large-scale lake-level fluctuations. Major unconformity surfaces associated with Lake Tanganyika lowstands may have corresponded with high-latitude glacial maxima throughout much of the mid- to late Pleistocene. Rocky shorelines along the eastern side of the present-day Ubwari Peninsula (Zaire) appear to have had a much more continuous existence as littoral rock habitats than similar areas along the north-western coastline of the lake (adjacent to the Uvira Border Fault System), which in turn are older than the rocky shorelines of the north-east coast of Burundi. This model of palaeogeographic history will be of great help to biologists trying to clarify the evolution of endemic invertebrates and fish in the northern basin of Lake Tanganyika. Tanganyika is the largest, and probably the oldest extant INTRODUCTION lake of the East African Rift System (Fig. 1A). The Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika proven to be immensely valuable tools in visualizing lake presented here is based on several prior reflection seismic histories and evaluating alternative hypotheses of palaeo- surveys and coring studies (Tiercelin & Mondeguer, hydrology and palaeoclimate (e.g. Teller, 1985; Morrison, 1991; Bouroullec et al., 1991; Cohen et al., 1993a), also 1991). Because of the important role that many large, discussed in detail in the companion paper to this article tectonically formed lakes have for biological evolution, (Lezzar et al., 1996), as well as on outcrop and oil wells such reconstructions have considerable potential to studies from the region (Attou, 1988). resolve important debates among evolutionary biologists as well (e.g. Owen et al., 1990). However, the use of METHODS lacustrine reconstructions in this application has been much more limited. In this paper we present a series of Seismic stratigraphy and age estimations palaeogeographic reconstructions for Northern Lake The data and interpretations that we present here are Tanganyika with important implications for the tectonic, climatic and biological evolution of this rift basin. Lake based primarily on results from Project CASIMIR’s 1992 © 1997 Blackwell Science Ltd 107

Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

Basin Research (1997) 9, 107–132

New palaeogeographic and lake-level reconstructionsof Lake Tanganyika: implications for tectonic, climaticand biological evolution in a rift lakeA. S. Cohen,* K.-E. Lezzar,† J.-J. Tiercelin† and M. Soreghan‡*Department of Geosciences, University of Arizona, Tucson,AZ 85721, USA†CNRS URA 1278 ‘Domaines Oceaniques’, GroupeRiftogenese Est-Afrique, Universite de BretagneOccidentale, B. P. 809, 29285 Brest Cedex, France‡School of Geology and Geophysics, University ofOklahoma, Norman, OK 73019, USA

ABSTRACT

Palaeogeographic and lake-level reconstructions provide powerful tools for evaluatingcompeting scenarios of biotic, climatic and geological evolution within a lake basin. Here wepresent new reconstructions for the northern Lake Tanganyika subbasins, based on reflectionseismic, core and outcrop data. Reflection seismic radiocarbon method (RSRM) age estimatesprovide a chronological model for these reconstructions, against which yet to be obtained agedates based on core samples can be compared. A complex history of hydrological connectionsand changes in shoreline configuration in northern Lake Tanganyika has resulted from acombination of volcanic doming, border fault evolution and climatically induced lake-levelfluctuations. The stratigraphic expression of lake-level highstands and lowstands in LakeTanganyika is predictable and cyclic (referred to here as Capart Cycles), but in a pattern thatdi�ers profoundly from the classic Van Houten cycles of some Newark Supergroup rift basins.This di�erence results from the extraordinary topographic relief of the Western Rift lakes,coupled with the rapidity of large-scale lake-level fluctuations. Major unconformity surfacesassociated with Lake Tanganyika lowstands may have corresponded with high-latitude glacialmaxima throughout much of the mid- to late Pleistocene.

Rocky shorelines along the eastern side of the present-day Ubwari Peninsula (Zaire) appearto have had a much more continuous existence as littoral rock habitats than similar areas alongthe north-western coastline of the lake (adjacent to the Uvira Border Fault System), which inturn are older than the rocky shorelines of the north-east coast of Burundi. This model ofpalaeogeographic history will be of great help to biologists trying to clarify the evolution ofendemic invertebrates and fish in the northern basin of Lake Tanganyika.

Tanganyika is the largest, and probably the oldest extantINTRODUCTIONlake of the East African Rift System (Fig. 1A). The

Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyikaproven to be immensely valuable tools in visualizing lake presented here is based on several prior reflection seismichistories and evaluating alternative hypotheses of palaeo- surveys and coring studies (Tiercelin & Mondeguer,hydrology and palaeoclimate (e.g. Teller, 1985; Morrison, 1991; Bouroullec et al., 1991; Cohen et al., 1993a), also1991). Because of the important role that many large, discussed in detail in the companion paper to this articletectonically formed lakes have for biological evolution, (Lezzar et al., 1996), as well as on outcrop and oil wellssuch reconstructions have considerable potential to studies from the region (Attou, 1988).resolve important debates among evolutionary biologistsas well (e.g. Owen et al., 1990). However, the use of

METHODSlacustrine reconstructions in this application has beenmuch more limited. In this paper we present a series of Seismic stratigraphy and age estimationspalaeogeographic reconstructions for Northern Lake

The data and interpretations that we present here areTanganyika with important implications for the tectonic,climatic and biological evolution of this rift basin. Lake based primarily on results from Project CASIMIR’s 1992

© 1997 Blackwell Science Ltd 107

Page 2: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

A. S. Cohen et al.

Fig. 1. A. General sketch map of the East African Rift geometry and location of the great lakes. B. Simplified tectonic andshoreline substrate map of the northern Lake Tanganyika basin, with place names, reflection seismic lines and core/well localitiesdiscussed in the text. Fault name abbreviations as follows: UBFS, Uvira Border Fault System; BBFS, Bujumbura Border FaultSystem; WUF, West Ubwari Fault; EUF, East Ubwari Fault; KFZ, Kaboge Fault Zone; CBF, Cape Banza Fault; CMF, CapeMagara Fault; KAF, Kabezi Fault.

seismic reflection survey, which employed a 300-joule 16) used for the chronological and sequence stratigraphicinterpretations proposed in this paper. Other seismic lineCENTIPEDE Sparker system as a seismic source. This

system (developed at the Renard Centre of Marine locations and technical specifications for this survey, aswell as detailed discussions of seismic facies observations,Geology, University of Ghent, Belgium) allows an acous-

tic penetration of #300 ms (twtt), an excellent scale for are given in a companion paper by Lezzar et al. (1996).Sparker system data were supplemented in this study bythe interpretation of modern rift–lacustrine stratigraphic

sequences. Nine seismic lines, totalling 250 km, were data from earlier seismic (multichannel and subbottomechosounding) and piston coring studies by Projectsshot in the North Tanganyika Basin during the 1992

CASIMIR survey. Figure 1B shows the key lines PROBE and GEORIFT (Rosendahl et al., 1988; Tiercelinet al., 1989; Tiercelin & Mondeguer, 1991; Bouroullec(Sparker Line S

2and Project PROBE Multichannel Line

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132108

Page 3: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

Palaeogeography of Lake Tanganyika

et al., 1991), by comparison with exploration well data tural, stratigraphic and geomorphic trends to the southonly as far as the extent of major faults or other structuralfrom the Rusizi River Delta (Attou, 1988) and our

personal observations of the limited number of outcrops features that are imaged by sparker seismic lines. Thiscorresponds with the location of the East and Westexposed around the north end of the lake.

In the absence of dated core material older than Ubwari Faults, on both sides of the Ubwari Peninsula atthe south end of Burton’s Bay. For these reasons we#35 ka from Lake Tanganyika, age estimates for events

inferred from seismic data can only be generated through caution the reader against attributing more precision thanis warranted to shoreline locations.sediment accumulation rate extrapolations. Such esti-

mates serve an important function in developing hypoth- Our facies interpretations are based on correlationsbetween near-surface high-resolution profiles and litho-eses about rates and controls on geological, climatic

or evolutionary events, or corroborating independently stratigraphic data from cores, discussed more fully inBouroullec et al. (1991, 1992) and Tiercelin et al. (1992,developed chronologies. They are also falsifiable hypoth-

eses, in the sense that multiply derived estimates of the 1994). Facies interpretations for the more recent B andA seismic sequences are based on the facies maps ofsame ‘event’ should yield approximately the same value.

However, they must be clearly di�erentiated from radio- Bouroullec et al. (1992). The estimation of lake highstandsurface elevations is problematic in Lake Tanganyikametrically constrained age dates, since they depend on

large extrapolations. In order to estimate sequence bound- because of the absence of widespread highstand deposits,particularly above modern lake level. However, given theary ages we have used a modification of the reflection

seismic – radiocarbon method (RSRM), developed pre- extremely steep-sided shoreline of Lake Tanganyika, anerror in the estimate of lake highstand surface elevationviously to estimate the age of the oldest synrift sediments

in Lake Tanganyika (Cohen et al., 1993a). This method does not translate into an unacceptably large error inshoreline longitude/latitude position. For example,can be used to generate age estimates for sedimentary

horizons in stratigraphic packages with high degrees of assuming a mean topography similar to that of thepresent, a 100-m error in lake surface elevation wouldcompleteness. The sumps of hydrologically closed basins

in deep half-graben lakes have been shown elsewhere only introduce an error of 3.1 km (n=38) in shorelineposition and a 30% error in lake area. Also the completeoften to display these types of highly predictable accumu-

lation rates, with rate ‘constancy’ at the 104–105-year absence of Lake Tanganyika deposits outside of theconfines of the current structural basin, or even for thatresolution, measured over 106–107-year time scales

(Sadler, 1981; Williams, 1993; Reynolds, 1994; Moutoux, matter higher than 55 m above present lake level (thehighest reported undeformed lake bed outcrops), con-1995; Olsen et al., 1996). We used 14C-derived mean

sediment accumulation rates for late Pleistocene uncon- strains the maximal highstand extent of the lake consider-ably (Ilunga, 1984). Finally, two oil exploration wellssolidated sediments from four coring sites (Fig. 1B) to

infer the chronology of decompacted stratigraphic drilled by AMOCO during the mid-1980s on the RusiziRiver plain (north end of Lake Tanganyika) provide somesequences from the same localities, and deposited under

similar tectonostratigraphic conditions (proximity and control on the age and duration of flooding events forthe lake outside the confines of the modern lake areaactivity of adjacent faults). In a modification of earlier

RSRM methods, rate estimates varied between sites and (Attou, 1988). It seems likely that the spillway elevationfor the lake has been close to its modern elevation forover time at sites, depending on the tectonostratigraphic

setting existent at each site at the time. Details of the much of the lake’s Late Tertiary/Quaternary history andwe have used this assumption in delimiting highstandRSRM interpretive methodology and phase terminology

are presented in Lezzar et al. (1996). lake margins.We inferred palaeoshoreline substrates from a number

of clues. First, if basement uplifts are associated withPalaeogeographic reconstructions

lake margin border faults we assumed that the shorelinewas rocky. This assumption holds true today along allOur primary method for making the palaeogeographic

sketch maps presented here was to trace lowstand lake- northern Lake Tanganyika shorelines, in the sense thateither rock or mixed rock/sand coastlines are present inlevel estimates based on onlapping relationships observed

in reflection seismic profiles (Projects PROBE, these areas (Soreghan & Cohen, 1996). This assumptionwould only likely be violated in the case of antecedentGEORIFT and CASIMIR) and, to the extent possible,

relate these to the existing onshore exploration well data drainages that cross-cut major rift-bounding structures.In Lake Tanganyika, the Malagarasi River inflow –and the modern geomorphology of surrounding water-

sheds. The clear correlations between erosional truncation Lukuga River outlet system is the only strong candidatefor a river system entering and leaving the lake that issurfaces evident in three structurally distinct northern

subbasins along seismic line S2

formed the basis for antecedent to the formation of the rift (Coulter, 1991).In the case of subsequent drainages, their associatedsubsequent extrapolations of the extent of the lake(s) at

various time intervals (Lezzar et al., 1996, figs 3 and 10). sandy beaches are mostly portions of small-scale (1 km2

or less) fan deltas (Soreghan & Cohen, 1996).Given the limited number of high-resolution seismiclines in the northern basin, we have extrapolated struc- Conversely, we have assumed that all shoaling margins

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132 109

Page 4: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

A. S. Cohen et al.

of half-grabens or structural platform shorelines were 1980; Kampunzu et al., 1983; Pasteels et al., 1989). Thisinitiation of tectonic activity in the northern Lakemade up of soft substrates. Given the size and nearshore

wave energy around a large lake such as Tanganyika, this Tanganyika region may be coincident with the start ofsubsidence in northern Lake Malawi (the deposition ofwould be winnowed to a sandy shoreline in all but the

most protected areas. Again this assumption would be the Nyasa–Baobab Sequence (estimated at #8.6 to#2.5 Ma) (Scholz et al., 1989; Ebinger et al., 1993)).correct today, with the exception of carbonate-cemented

beachrock outcrops and a few isolated and small rocky Project PROBE profiles 83-06A and 84-228 (Rosendahlet al., 1986; Morley, 1988) suggest that there was littleoutcrops along the central Burundi (Resha area) and

northern Zaire coasts. Finally, we assume that significant or no activity along the Bujumbura Border Fault System(i.e. the eastern side of the North Rusizi half-graben) atmarshlands existed around all major active axial drainage

deltas, as is the case today in most rift lakes regardless this time. A topographic high appears to have existedalong the nascent Uvira Border Fault System, delimitingof lake stand condition (Butzer, 1971; Cohen, 1982;

Tiercelin et al., 1992). sediment transport towards the south into a majordepocentre.For purposes of clarity we have put the locations of

all the major structural elements discussed below (major Further south, in the Kigoma region, Versfelt (1988)and Rosendahl et al. (1988) have suggested that the early,faults, geomorphic features and half-graben structures)

only on Fig. 1B. post-Nyanja event history of the lake was characterizedby the presence of a broad lake, with little or no activityalong the major, modern border faults (‘a water-covered

PA LAEOGEOGRAPHIC cratonic basin developed on a peneplain (Nyanja Event)’according to Rosendahl et al., 1988, p. 26).RECONSTRUCTIONS

During the RBM Phase, the West Ubwari FaultInitial synrift infills – Phase RBM (RSRMbecame tectonically active long after the deposition ofestimated age #7.4 Ma to #1.1 Ma)these initial post-Nyanja Event sediments. RSRM esti-(Figs 2A–C & Fig. 3) mates suggest this occurred #4.9 Ma. A major depocen-tre developed at this time in the hangingwall of the WestThe oldest sediments of the Northern Tanganyika Basin

were deposited in the South Rusizi half-graben over a Ubwari Fault in the subsiding South Rusizi half-graben(Fig. 2B). The Ubwari horst began to develop as a majorprerift surface (the Nyanja Event), which appears to have

been very broad and flat (Rosendahl et al., 1988). RSRM structural feature at this time, along with the two trans-verse zones: the Rusizi accommodation zone and theage estimates for this event from both Cohen et al.

(1993a) and Lezzar et al. (1996) suggest these sediments Kaboge Fault Zone. By this time the lake had expandedslightly in area, extending further to the east (as indicatedare #7.4 Ma. Structural highs developed in response to

footwall uplift of the West Ubwari Fault, defining the by sedimentation over the nascent western Ubwari horst)than in the immediate post-Nyanja Event period (foreastern limit of the present basin. The South Rusizi half-

graben was the major depocentre of the North Basin cross-sections in the areas described see figs 3 and 4 inLezzar et al., 1996).during this period, with major sediment accumulation

along the western side of the active West Ubwari Fault Apparently the Ubwari/North Kigoma structural areaswere high and underwent continued erosion or nondepos-(central segment north of Cape Banza) (Fig. 1B and 2A).

The principal sediment source for this basin appears to ition during the early history of subsidence in the SouthRusizi half-graben. At the north end of the modern lake,have been from an axial, ‘proto-Rusizi’ drainage to the

north, lateral drainages from the western shoaling margin subsidence probably accelerated along the Uvira BorderFault System, resulting in a nascent North Rusizi half-of the basin, based on stratal geometries imaged in

PROBE lines 84–200 and 14 (Rosendahl et al., 1988), graben that underwent rapid sedimentary infilling, mostlikely from sources to the north (Fig. 2B, C). RSRMand probably from lateral drainages from the West

Ubwari Fault escarpment. A lake extended from the estimates suggest this occurred between #4.9 Ma and#3.6 Ma.northern part of the nascent North Rusizi half-graben

almost as far south as modern Cape Banza. The northern Subsequently, activity along the Uvira Border FaultSystem expanded considerably to the south along withextent of the lake is delimited by the presence of late

Miocene marginal lacustrine deposits in the Buringa I subsidence within the North Rusizi half-graben.Figure 2C illustrates the state of the basin at #3.6 Maand Rusizi I wells (Attou, 1988).

At the northern end of the North Rusizi half-graben, (RSRM estimate). Sediment accumulation within theNorth Rusizi half-graben was concentrated along itstectonic activity began at this time along a nascent Uvira

Border Fault System. This may be related to the initiation rapidly subsiding western margin, with some accumu-lation also in the modern Rusizi delta region. Much ofof volcanic doming (the ‘Kivu–Rusizi Local Volcanic

Dome’) that began to develop in the Kivu and Rusizi the sediment discharged from the axial ‘proto-Rusizi’River appears to have also been ponded against the north-Basins at about this time (Ebinger, 1989a,b; Coussement,

1995), corresponding to the first stage (#7.8 to #5 Ma) west side of the West Ubwari Fault. An accommodationzone linking the North Rusizi and South Rusizi half-of the second cycle of doming activity (Bellon & Pouclet,

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132110

Page 5: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

Palaeogeography of Lake Tanganyika

Fig. 2. (A to N) Palaeogeographic/palaeotectonic sketch maps of the northern Lake Tanganyika region from the early NyanjaEvent to the present. Maps are based on a combination of seismic, outcrop, well and geomorphic data. Lake boundaryplacements are approximate, guided by lapout boundaries and the position of major structures known to be active during eachtime interval. Fault name abbreviations are the same as for Fig. 1B. The southern limit of the mapped area represents thefurthest extent of structures that cross-cut seismic lines for which we have chronological control data. bpll=elevation of lakesurface below present lake level. RSRM=reflection seismic radiocarbon method.

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132 111

Page 6: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

A. S. Cohen et al.

grabens was well developed by this time. Activity had Rusizi plain indicates the presence of a pronouncedalso commenced by this time along the nascent East angular unconformity in this area. This may reflectUbwari Fault. Simultaneously the lake had expanded ongoing tectonism during this low-lake period. Underconsiderably to the south and east, as is evidenced by this scenario, low lake stands and desiccation would havethe resumption of sedimentation across the eastern persisted in the northern Lake Tanganyika region for amargin of the future North Kigoma half-graben. considerable part of the latest Pliocene and early

Pleistocene. Rosendahl et al. (1988) have independentlysuggested that the deep scour evident in some areas onOnset of regional erosion and focusedthe KMSB surface argues for a very long interval ofrifting – Phase F–E (RSRM estimated ageerosion. deMenocal (1995) has suggested that increased#1.1 to #0.39 Ma) (Figs 2D–G & 3)aridity occurred in Africa at about 1 Ma, based on dust

A major change occurred in the northern end of the variability records.Lake Tanganyika basin with the onset of focused rifting It is clear that lake level increased in the basin at someand the development of associated regional erosion sur- time during this interval, although the precise timing offaces (Fig. 2D). We estimate this occurred about 1.1 Ma, this event, as well as its magnitude, is not well con-corresponding to the initiation of Phase F-E of Lezzar strained. Shortly after the lowstand event that we estimateet al. (1996). This surface, variously referred to as the occurred #1.1 Ma, sedimentation first resumed in theRusizi Sequence Boundary 1 (RSB

1), the Banza Sequence actively subsiding South Rusizi half-graben, but it was

Boundary 1 (B1) or the Kigoma–Makara Sequence not until much later (RSRM estimate #670 ka) that

Boundary (KMSB), depending on location, resulted from sedimentation resumed on the eastern North Kigomaa combination of a major relative lake level decline half-graben margin.(#650–700 m below present lake level, bpll, based on By the middle of Phase F–E, at around 670 ka byseismic reflection geometries), as well as extensive uplift RSRM estimate (Fig. 2E), lake levels had risen substan-along the developing East and West Ubwari Faults and tially (though the magnitude of this rise is unknown).the Uvira Border Fault System (Morley, 1988; Rosendahl However, the Ubwari horst continued to separate theet al., 1988; Scholz & Rosendahl, 1988; Lezzar et al., northern and southern terminal lakes (Lezzar et al.,1996). Rosendahl et al. (1988) showed that after the 1996). Motion along the strike-slip Kaboge Fault Zoneformation of the KMSB, rifting within the Kigoma (KFZ) appears to have occurred at this time as indicatedProvince became focused, and its major border faults and by the faulted, dome-shaped morphology of the F and Emodern basin asymmetry developed. Deposition of the

Sequence reflectors (fig. 7A in Lezzar et al., 1996).early Phase F-E lacustrine sediments (Lezzar et al., 1996)

Erosion occurred throughout most of the North Rusiziwas restricted at this time to small and probably saline

half-graben at this time. Subsequently, the deposition ofterminal lakes in the depocentres west of the West Ubwari

Phase F–E sediments across the Ubwari horst indicatesFault and east of the East Ubwari Fault. Activity on thethat a single lake overtopped this feature. The state ofKivu–Rusizi Local Volcanic Dome had resumed by aboutthe basin at #550 ka (RSRM estimate) is depicted in1.9 Ma, the second stage of the second cycle of domingFig. 2F. Based on their seismic reflection characteristics,of Bellon & Pouclet (1980), Kampunzu et al. (1983) andSequence F deposits on the Ubwari horst appear toPasteels et al. (1989), and significant relief persisted incomprise stratified, coarse- to medium-grained detritalthe northern North Rusizi Basin, which was well abovedeposits that were derived axially from either the northlake level at this time, even along the very active Uviraor south ends of the Rusizi half-grabens. These depositsBorder Fault System. Major activity on the East andprogressively filled the subsiding Magara slope and onlapWest Ubwari Faults at this time created a separationonto the eastern, gently dipping flank of the Banza slope,between the South Rusizi half-graben and the Northindicating the occurrence of minor uplift of the MagaraKigoma half-graben, referred to as the higher Ubwarishoal and continuous normal faulting along the Easthorst (Fig. 2D).Ubwari Fault (Banza Shoal).Di�erentiating climatic from tectonic causes for the

Much of the North Rusizi half-graben had also floodedearly Pleistocene contraction of lakes in the Tanganyikaby this time, with a major depocentre forming in thebasins is di�cult, if not impossible given our presentdeep basin east of the Ubwari Border Fault System.state of knowledge. It is possible that the reduction inDuring this time, motion along the strike-slip Kabogelake area began at #2.5–2 Ma as a response to the lateFault Zone appears to have also accelerated, whereasPliocene regional aridification in East and Central Africasubsidence in the South Rusizi half-graben, west of the(de Heinzelin, 1955; Bonnefille, 1976; Cerling & Hay,West Ubwari Fault, appears to have slowed. Sed-1986; Cerling et al., 1988; Pickford et al., 1991;imentation in this same region also slowed at this timedeMenocal, 1995). This is also consistent with a signifi-as a consequence of sediment trapping to the north. Nocant change in lithologies in the Rusizi Plain wells, fromstructural or seismic data are available to constrain thePliocene lacustrine deposits to Pleistocene fluvial depositstiming of motion along the Bujumbura Border Fault(probably separated by an unconformity) (Attou, 1988).

Unpublished seismic data acquired by AMOCO over the System (Fig. 1B), or whether its formation is associated

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132112

Page 7: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

Palaeogeography of Lake Tanganyika

Fig. 2. (continued).

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132 113

Page 8: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

A. S. Cohen et al.

with the ancient West Ubwari Fault or the more recent or cessation of subsidence in the South Rusizi half-graben. RSRM estimates suggest this occurred betweenCape Magara Fault.

Deposition during the latter part of Phase F–E #360 and 290 ka. Burton’s Bay continued to develop,extending further south as fault activity expanded on the(Fig. 2G) occurred throughout almost the entire extent

of the modern northern Lake Tanganyika (except for Cape Banza Fault and the southern part of the UviraBorder Fault System. The Baraka furrow acted as aBurton’s Bay and the extreme north-west). Rapid subsid-

ence continued in the North Rusizi half-graben and very major conduit for sediments shed o� the western side ofthe developing Burton’s Bay, although axial sources fromslow subsidence continued in the South Rusizi half-

graben. Motion slowed on the West Ubwari Fault at this further south in the developing Burton’s Bay also contrib-uted sediments to this region. These basin fill sedimentstime, but began along the Cape Banza Fault (CBF), on

the west side of the modern Ubwari Peninsula (Figs 1B are expressed seismically as chaotic to weakly stratifiedunit boundary reflectors (Fig. 4), with medium-amplitudeand 2G).reflectors displaying a hyperbolic character (figs 3A and7A in Lezzar et al., 1996). They infill irregular, erosionalSedimentation in a mature rift basin and thetopographic lows on the prior (d) lowstand surface. Wedevelopment of the modern northerninfer that they comprise coarse-grained alluvial and col-Tanganyikan subbasins – Phases D to Aluvial deposits. Later, this style of sedimentation gave(RSRM estimated age #390 Ka to present)way to the extensive deposition of a very thick sequence(Figs 2H–N & 3)of sheet-drape deposits. Seismically, these sheet-drapedeposits comprise laterally extensive bundles of parallelSedimentation during this major phase of northern Lake

Tanganyika history has been modulated by a series of to subparallel, high-amplitude reflectors, interlayeredwith thin chaotic to subtransparent layers (Fig. 4). Welarge, but declining-amplitude lake-level fluctuations. A

detailed discussion of these deposits and their seismic infer these deposits to comprise predominantly pelagicoozes and hemipelagic muds (figs 7A and 9B in Lezzarfacies characteristics is presented in the companion paper

to this paper (Lezzar et al., 1996). The frequency and et al., 1996). In the north-eastern part of the lake,subsidence began along the Cape Magara Fault. Strike-stratigraphic regularity of these fluctuations, in compari-

son with earlier ones, strongly suggests that they are slip movement is evident at this time along the KabogeFault Zone, as indicated by the smooth dome shapeclimatically driven, although there is clearly a strong

tectonic overprint to the morphology of the lake through- (positive flower structure) of Sequence D reflectors inthe South Rusizi half-graben.out this time interval. Accompanying these lake-level

fluctuations are cyclical patterns of sedimentation (dis- Several major sediment sources that generated deep-water lacustrine fans were active at this time in the north;cussed more fully in the ‘Facies expression of lake-level

fluctuations’ section). the Proto-Rusizi drainage (providing sediments derivedfrom the upper Rusizi Valley), lateral drainages from theAt the end of Phase F–E renewed lake lowstand

conditions, represented by the (d) erosional period, nascent Cape Magara Fault region and lateral drainageson the north-west margin of the modern Burton’s Bay.resulted in a major reduction in the area of Lake

Tanganyika (Fig. 2H). Reconstruction of the (d) bound- We recognize these fans seismically by their chaoticinternal structure and lens-shaped geometry, extendingary morphology suggests that lake level declined to

#350 m bpll (RSRM age estimate #390–360 Ka). Local over kilometres to tens of kilometres. Upper SequenceD sediments on the Ubwari horst and in the Northuplift of the northern part of the North Rusizi half-

graben suggests that renewed activity on the Kivu–Rusizi Kigoma half-graben consist of two seismic facies: domi-nant, chaotic units that are thin and continuous, and thatVolcanic Dome occurred at about this time, or possibly

slightly earlier (Pouclet, 1976; Bellon & Pouclet, 1980; we interpret as gravity flow deposits (presumed to bepredominantly distal turbidites), and subsidiary, thickerEbinger, 1989a). This doming may have contributed to

the erosion on the (d) sequence boundary at the north but more localized, chaotic units, that we also interpretas gravity flow deposits (presumably proximal turbidites).end of the study area. Significant horizontal and vertical

motion along the Kaboge Fault Zone and both the East The proximal turbidites deposited on the buried westernside of the horst were probably derived from the incipientand the buried West Ubwari Faults (Figs 1B and 2H),

along with subsequent erosion, resulted in considerable Burton’s Bay (see rivers in Fig. 1B) and the easternRuzibazi River, near Cape Magara. Thinner, proximaltopographic irregularity on the (d) sequence boundary

(figs 3 and 7A in Lezzar et al., 1996). Relative fault turbidites in the Kigoma half-graben were derived fromlateral drainages of the eastern rift escarpment (mainlymotions between the buried Magara shoal and the Kaboge

Dome resulted in the formation of the incipient Baraka the Ruzibazi and Murembwe Rivers). The distal tur-bidites in both the Ubwari horst and the North Kigomafurrow. This, along with the extension of the Cape Banza

Fault to the south, initiated the formation of the present- half-graben areas probably originated from a northernaxial source. The Cape Magara Fault appears to haveday Burton’s Bay.

The return of lake highstand conditions represented acted to divert the growth of a northerly sourced axialfan towards the south-west. Elsewhere throughout theby Sequence D (Fig. 2I) was accompanied by a slowing

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132114

Page 9: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

Palaeogeography of Lake Tanganyika

Fig. 2. (continued).

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132 115

Page 10: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

A. S. Cohen et al.

Fig. 2. (continued).

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132116

Page 11: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

Palaeogeography of Lake Tanganyika

Fig. 3. Sequence correlation chart andpalaeolimnological history of thenorthern Lake Tanganyika basin.

lake, sedimentation was dominated by sheet-drape Fault. Simultaneously the eastern side of the NKHG wassubaerially exposed and undergoing active erosion.deposits.

Unconformity (c) marks a renewal of lake lowstand Lake level rise during Lezzar et al.’s Phase C (Fig. 2K)resulted in the flooding of essentially the entire modernconditions within the North Tanganyika Basin

(#350 m bpll) (Fig. 2J). RSRM age estimates suggest northern part of Lake Tanganyika, with the possibleexception of the southernmost portion of Burton’s Bay,this occurred #290–260 ka. At various times during this

interval, a northern, slightly deeper lake along the Uvira which may not have formed by this time. RSRM ageestimates suggest this occurred between #260 andBorder Fault System may have periodically been separ-

ated from the main lake across a very shallow region at 190 ka. In the early part of the transgression, basin-filldeposits infilled topographic irregularities on the (c)the south end of the South Rusizi half-graben. The lake

floor in the South Rusizi area appears to have been erosion surface in the nascent, present-day Bujumburasubbasin. These sediments may have been derived fromrelatively flat following the end of active subsidence,

presaging the morphology of the modern Bujumbura either northern (proto-Rusizi) or southern (proto-Kasandjala and Mutembala) axial drainages (Figs 1B andsubbasin. Emergent conditions (dry or swampy areas)

probably existed throughout much of the subbasin, with 2K). As the transgression proceeded, these gave way tothe deposition of extensive deep-water lacustrine fanshallow lake conditions (#25 m maximum water depth)

developed within a nascent Magara–Banza depression deposits that developed near the major sediment dis-charge point. A deep lacustrine fan also developed at theperched on top of the Ubwari horst. Continuous subsid-

ence of this depression resulted in the collapse of the outlet of the Baraka River. Sheet drape deposits, compris-ing fine-grained detrital clays and organic oozes, overliecentral part of the Ubwari horst (along the active Cape

Banza Fault), which in turn promoted the formation of the fan deposits. Most of the clastic material in thesewell-stratified deposits probably was derived from athe Capart Channel on the buried Magara shoal. A

submerged rocky shoal may have crossed the lake between northern (proto-Rusizi) axial drainage. In the CapartChannel, Sequence C sediments appear to be dominatedmodern Cape Banza and the Kabezi Fault on the east

side of the lake at this time. by coarser detrital sediments (based on their seismiccharacter). In both the Magara–Banza depression (on theMinor but widespread erosion of Sequence D sedi-

ments occurred during the (c) period on the gentle slope Ubwari horst) and the North Kigoma half-graben,Sequence C comprises only sheet drape deposits. A fan-forming the Kigoma half-graben. Lake level reconstruc-

tions for the North Kigoma half-graben (NKHG) suggest shaped geometry within Sequence C along the westernside of the Magara–Banza depression suggests continuousa maximum water depth of #30 m along the East Ubwari

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132 117

Page 12: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

A. S. Cohen et al.

Fig. 4. Idealized Capart Cycles from the northern Lake Tanganyika basin. Facies interpretations relating core and high-resolutionseismic data, average cycle and facies thicknesses are based on Tiercelin & Mondeguer (1991), Baltzer (1991), Bouroullec et al.(1991, 1992) and Lezzar et al. (1996). The seismic line blowups illustrating the seismic facies characteristics for each part of thecycle are from CASIMIR Sparker Line S

2, in the Magara–Banza depression (Ubwari horst and the southern Bujumbura

subbasin–South Rusizi half-graben (see Lezzar et al., 1996, fig. 7A).

subsidence along the Cape Banza Fault whereas, on its continued subsidence of the Kigoma half-graben wascoupled with the uplift of the Banza shoal and probableeastern side, a dome-like shape suggests active strike-slip

motion along the Kabezi Fault (figs 3 and 7 in Lezzar rollover of the basin margin, represented by the shallowRumonge platform (see Lezzar et al., 1996, figs 7C andet al., 1996).

At the end of Phase C, renewed regression resulted in 10, evolution of Rumonge platform). These rolloverstructures are comparable with those described in Lakethe formation of the (b) unconformity and erosion surface

of Lezzar et al. (1996) (Fig. 2L). RSRM age estimates Malawi by Ebinger et al. (1993). The lower part of theslope probably subsided as a consequence of movementsuggest this occurred between #190 and 170 ka. We

interpret discontinuity (b) as the result of a lake level fall along the N-20°-trending fault, antithetic to the mainEast Ubwari Fault. This subsidence also resulted in theto #250 m bpll. This value is close to the #300 m fall

previously proposed for the (A∞) and (W) discontinuities deepening of a small graben that subsequently becamethe nascent, present-day Rumonge Channel (#150 mby various authors (Tiercelin et al., 1989; Mondeguer,

1991; Tiercelin & Mondeguer, 1991; Bouroullec et al., maximum water depth) during the (b) lowstand, with aneastern shoreline closely coinciding with the shelf break.1992). The reconstruction of the palaeotopography at the

(b) lowstand suggests the existence of a flat basin floor Sequence B was deposited under a renewed episodeof transgression (Fig. 2M). RSRM age estimates suggestin the Bujumbura subbasin, and more pronounced relief

in the subsiding Magara–Banza depression (see Lezzar this occurred between #170 and 40 ka. The earlieststages of lake level rise were associated with extensiveet al., 1996, fig. 10C). We estimate a maximum water

depth of #75–100 m in both of these areas. A large part erosion, particularly in the Capart and Rumonge chan-nels. The Capart Channel was probably fluvially incisedof Burton’s Bay was probably emergent and/or occupied

by swampy areas, an idea consistent with earlier obser- by an axial drainage flowing northward from the emergentBurton’s Bay. Erosion in the Rumonge channel wasvations of Tiercelin & Mondeguer (1991). To the east,

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132118

Page 13: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

Palaeogeography of Lake Tanganyika

directly related to coarse detrital inputs from the lateral insu�cient to divide the lake into separate water bodies,but did result in the near-total desiccation of Burton’sRuzibazi River, which was diverted southward at this

time by the very active Cape Magara Fault (Figs 1B Bay, and probably the Rumonge platform and north-eastLake Tanganyika. The (a) sequence boundary can beand 2M).

Deposition during Phase B in the Bujumbura subbasin correlated to the (A) boundary defined in the SouthernMpulungu subbasin (Tiercelin et al., 1989; Mondeguer,began with detrital inputs, and coarse-grained basin fill

(Fig. 4; fig. 7A in Lezzar et al., 1996). Above this unit, a 1991). The #160 m lake level fall estimate for thenorthern lake is consistent with a #150 m estimate forcoarse clastic lens developed on the relatively flat bottom

of the axial part of the subbasin. This lens-shaped unit the Southern Mpulungu subbasin. The basin floor top-ography reconstructed at the time of erosional event (a)is interpreted as a deep lacustrine fan, formed by coarse-

grained gravity flows at the outlet of the lateral Baraka appears to be nearly identical to the present morphology,except for the Banza sublacustrine shoal, which wasRiver in Burton’s Bay. Above and on both sides of this

fan, the upper part of Sequence B is characterized by probably somewhat lower than at present (fig. 10 inLezzar et al., 1996). We estimate a maximum water depthsheet drape sedimentation during highstand conditions.

This same sedimentary succession is duplicated in the of #220 m to #130 m for the subbasins crossed bysparker line S

2. Two subsequent subsequence boundaries,Magara–Banza depression (Fig. 4; fig. 7A in Lezzar et al.,

1996). Coarse deposits mainly accumulated on the west- termed (x) and (y) by Bouroullec et al. (1991), have alsobeen observed in both seismic and short core data. Theseern flank of the depression as a large lens that pinches

out towards the east, probably as a consequence of a boundaries correspond to lake level falls dated radio-metrically at #23 ka and #18 ka, respectively. Thesereduction in subsidence rates on the Magara slope and

continuous uplift of the Banza slope. surfaces are not figured here, since they were primarilydriven by changes in the lake’s water budget, and by thisIn the Capart channel, the base of Sequence B com-

prises a narrow (1 km wide) coarse-grained, lens-shaped time all of the major modern tectonic elements of themodern lake were in place. Thus after the formation ofbody that probably resulted from detrital inputs from

the axial Kasandjala and Mutembala Rivers at the sou- the (a) boundary lake-level changes more or less havefollowed modern bathymetry.thern end of Burton’s Bay. These deposits are overlain

in the channel axis by a unit of alternating autochthonousorganic sediments and thin detrital units.

FAC IES EXPRESSION OF LAKE-LEVELIn the deepest part of the Rumonge subbasin, Sequence FLUCTUATIONS

B comprises a thick deposit forming the primary infill ofthe Rumonge channel (Fig. 4) (fig. 7A in Lezzar et al., In addition to the marked low lake stand erosional

surfaces evident in the northern Lake Tanganyika reflec-1996). The base of this deposit consists of a relativelyflat and thin, apparently coarse-grained detrital layer. tion seismic records, seismic facies analysis allows us to

infer some characteristics of the vertical stacking ofAbove this unit is a wide, lens-shaped deposit that weinfer to be coarse grained, which probably corresponds sediments within each stratigraphic sequence (Bouroullec

et al., 1991; Lezzar et al., 1996). These sequencesto a deep lacustrine fan formed at the outlet of theRuzibazi River. Overlying this are small lenses of appar- themselves appear to be strongly cyclical, an inference

that receives support from the uppermost ‘A’ sequence,ently coarse-grained material, with cut-and-fill channelfeatures. These channels are interbedded with thin strati- which has been partially to completely penetrated by

gravity, Mackereth and piston cores (Tiercelin et al.,fied layers that we interpret as distal turbidites.On the Rumonge slope and platform, Sequence B 1988, 1989; Baltzer, 1991; Tiercelin & Mondeguer, 1991).

We refer to these cycles as Capart Cycles (Fig. 4), inconsists of a thin, coarse-grained layer overlain by a unitof prograding clinoforms and aggrading strata. The clino- honour of Andre Capart, who first recognized from

echosounder profiles the probability that Lakeforms are probably associated with a nearby river delta(C. A. Scholz, personal communication, 1995), possibly Tanganyika had undergone significant fluctuations in

water level through the Pleistocene (Capart, 1952).the lateral Murembwe River (see Lezzar et al., 1996,fig. 7C). Pleistocene highstand terraces in the Rusizi Reflection seismic profiles and seismic facies interpret-

ations for these cycles are also shown in Fig. 4 and inRiver Plain located at 830 m asl (i.e. 55 m above presentlake level) probably formed during this period, suggesting Lezzar et al. (1996, figs 7 and 8). Capart Cycles consist

of a basal erosional surface, overlain by a variable thick-that the lake was somewhat more extensive (particularlyto the north and in Burton’s Bay) than it is today. ness (0 to several metres) of coarse-grained siliciclastic

deposits (gravels and mud balls) with very localizedHowever, no Pleistocene lacustrine deposits occur ineither of the AMOCO wells, suggesting that when distributions. These ‘basin fill’ units are expressed seis-

mically by chaotic or weakly stratified/hyperbolic reflec-transgressions reached beyond the modern lake bound-aries, they did so only briefly, or were poorly preserved. tors of medium amplitude and with hyperbolic sequential

surfaces. These presumed coarse clastics both infill andOur youngest reconstruction, Fig. 2N, is for the (a)unconformity. The RSRM age estimate for this boundary overtop localized depressions on the underlying erosion

surfaces and appear to have formed under relatively lowis #40–35 ka. This lake level fall (#160 m bpll) was

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132 119

Page 14: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

A. S. Cohen et al.

lake stand conditions, at the beginning of a lake trans- longer-duration cycles. These longer-duration cyclesprobably approximate Capart Cycles in timing, but never-gressive phase. The ‘basin fill’ units are typically overlain

by the deposits of ‘deep lacustrine fans’, lens-shaped theless retain the facies elements unique to Van HoutenCycles (i.e. the proportions of facies vary but not thebodies, typically with chaotic internal structure and

hyperbolic sequential surfaces. They are interpreted to facies suite), rather than taking on characteristics similarto Capart Cycles. The second explanation for di�erencesrepresent stacks of metres-thick packages of high-energy,

coarse-grained gravity flow deposits, generated by suc- in cyclicity seems more likely. Lake Tanganyika is (andapparently has been) an extremely steep-sided and verycessive floods of lateral or axial drainages, and similar in

organization to those described from the Early Cretaceous deep lake. Lake-level fall under such conditions leads tovery little lateral facies migration where slopes are steep,Reconcavo Basin rift–lacustrine turbidites (Medeiros &

Ponte, 1981; Magnavita & Da Silva, 1995). In total these but potentially extensive erosion. Conversely, the NewarkBasin has been suggested to have been a relatively broad‘basin fill’ stacks average between 20 and 40 m thick in

each cycle. The upper half to two-thirds of each cycle topographic basin with little topographic relief (P. Olsen,personal communication, 1989). Additionally, the Newark(20–40 m) consists of parallel to subparallel, high-

amplitude reflectors that alternate with thin, chaotic or Basin lakes were probably shallow and relatively low-energy environments (D. Reynolds, personal communi-subtransparent layers. These ‘sheet drape’ units are

interpreted to be the rising to lake highstand mudrocks cation, 1996). Under such conditions extensive palaeosolsmight be developed during low lake stands without thethat were deposited contemporaneously over vast areas

of the lake floor. In sequence A, where they are penetrated pervasive deep erosional surfaces typical of LakesTanganyika and Malawi (Scholz & Rosendahl, 1988;by cores, they are dominantly laminated to massive

diatomaceous silts (‘oozes’). Bouroullec et al., 1991). Field observations and computermodelling experiments on the Miocene Rubielos de MoraCyclical patterns of deposition have been described

previously from other rift lake basins but at a scale and (RDM) Basin, Spain, support this conclusion (Anadonet al., 1988, 1991; Cohen et al., 1993b). Van Houtenwith lithofacies packages that are very di�erent from

what we observe in the Capart Cycles of Lake cycles occur in the western portion of the RDM basin,which has been interpreted to have been the low-relief,Tanganyika. Most notably, the Van Houten Cycles of the

Newark Supergroup basins (Olsen, 1986, 1990) also low-subsidence portion of this half-graben lake. This VanHouten type cyclicity disappears in the eastern portionreflect repeated lake-level fluctuations. Van Houten and

Capart Cycles di�er significantly in their overall thickness of the RDM palaeolake, which appears to have been amuch steeper sided and rapidly subsiding portion of the(#10 m for a typical Van Houten Cycle, vs. 50–100 m

for a typical Capart Cycle), and probably their duration. lake basin. Facies patterns in the eastern part of the RDMpalaeolake are similar but not identical to the CapartAdditionally, the strongly developed and laterally con-

tinuous lake-lowstand, fine-grained palaeosols do not Cycles of northern Lake Tanganyika.have a parallel in the Capart Cycles, at least for thelowstand portions observed in cores to date. Instead

REGIONAL CLIMATIC A NDlowstands are expressed as profound erosion surfaces TECTONOVOLCANIC IMPLICATIONSoverlain by relatively coarse-grained deposits. Although OF THIS STUDYpalaeosols almost certainly developed on exposed highsand interfluves during Lake Tanganyika lowstands, they Reflection seismic radiocarbon method (RSRM) age esti-

mates of lake-level fluctuations provide intriguing hintsapparently were subject to subsequent erosion and there-fore much lower preservation potential than in the case that lake-level lowstands in Lake Tanganyika may have

corresponded with high-latitude glacial maxima through-of Van Houten Cycle palaeosols. In some respects theCapart Cycles resemble the deposits of the more humid, out much of the Quaternary. Abundant evidence exists

that correlates the most recent Late Pleistocene, lake-southern basins of the Newark Supergroup, the‘Richmond-Type Facies Complexes’ (Olsen, 1990), level lowstands in Africa with high-latitude glacial con-

ditions during 18O stage 2 (e.g. Street-Perrott & Harrison,although the thickness of the Capart Cycles appears tobe greater than those typical of the Richmond Basin. 1985), although the causal relationship (or lack thereof )

between the two events has been controversialThe di�erences between how major lake-level fluctuationsare expressed in palaeolakes that generate Van Houten (deMenocal et al., 1993, and references therein;

deMenocal, 1995). However, there have been fewCycles (e.g. the Newark and Hartford Basins) (Olsen,1986) vs. those that produce Capart Cycles may result attempts to estimate directly the timing of lacustrine

highstands and lowstands in Africa during thefrom two factors: (1) di�erences in temporal scale (weinfer that Capart Cycles are of considerably longer Quaternary, prior to #40 ka, from the lakes themselves.

A considerable body of evidence suggests that Africanduration than Van Houten Cycles) or (2) di�erences intopographic relief (and, secondarily, relative subsid- climate change during the Quaternary has been strongly

cyclical, and some of this evidence relates indirectly toence:sediment accumulation rate) between the two typesof lakes. We consider the former explanation unlikely, lake-level fluctuations. For example, deMenocal et al.

(1993) have shown that the period of freshwater Melosirasince Van Houten Cycles are themselves bundled into

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132120

Page 15: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

Palaeogeography of Lake Tanganyika

diatom valves deflated from dry lake beds and blown as (d), (c), (b), (a), (x) and (y), respectively) provides strongevidence that these events are the consequence of a dropaeolian dust into the deep sea shows a strong 23–19-ka

cyclicity at ODP core site 663, o� the west coast of in the lake’s surface elevation. The concordance of RSRMage estimates for each of these boundaries between severalAfrica, whereas aeolian dust and phytolith records from

the same core are strongly cyclical at 100- and 41-ka subbasins (table 1 in Lezzar et al., 1996) is also consistentwith this hypothesis. Furthermore the (x) and (y) eventsintervals. Evidence from the Arabian Sea also shows

strong Milankovitch-band cyclicity, with dominant fre- can be correlated with similar surfaces at the south endof Lake Tanganyika (Tiercelin & Mondeguer, 1991).quencies varying between onshore and o�shore coring

localities (Clemens et al., 1991; Prell et al., 1992; Two plausible explanations can be o�ered for theselake-level declines. First, they may have resulted fromAnderson & Prell, 1993). Modelling results suggest that

eastern Africa would be expected to respond more episodes of stream capture for major influent drainages,caused by such factors as fault growth, migration orstrongly to northern European ice cover, whereas western

Africa would respond more directly to North Atlantic propagation, stream diversion related to seismic events,volcanic doming or damming or competitive fluvial aggra-sea surface temperatures (de Menocal & Rind, 1993).

Lake Tanganyika, lying as it does on the boundary dation (Leeder & Jackson, 1993). This type of lake-levelfluctuation control has been documented for other largebetween these climatic domains, is well situated for

examining their interactions in regulating central fault-bounded lakes, for example Walker Lake, Nevada(Bradbury et al., 1989). Second, they may have resultedAfrican climate.

In Fig. 3 we summarize our recently developed from climate change associated with relative drops inP/E ratios within the Lake Tanganyika watershed. Ofsequence boundary chronology for the northern basin of

Lake Tanganyika, along with our RSRM age hypotheses these two possibilities we consider the second much morelikely. Judging from the relatively small thicknesses offor these boundaries, and our interpretation of the prob-

able causal mechanism driving each major erosional event. the deeper basinal sequences correlative with the uncon-formities, it does not appear that the unconformitiesThe distinction between early (pre #0.4 Ma) tectonic

causes and more recent climatic causes of major uncon- represent very long intervals of time. A tectonic modelimplying ‘stream piracy’ events therefore would have toformities must be interpreted cautiously, given the better

seismic resolution of events in the upper portion of the invoke repeated episodes of watershed discharge reversal,with the Tanganyika discharge always being of longerlake’s sediment fill. Tectonovolcanic events are almost

certainly responsible for many of the major events duration than the drainage to outside the lake. Thisscenario is highly improbable. Furthermore, the threerecorded in the Miocene to mid-Pleistocene stratigraphic

record of northern Lake Tanganyika. It is conceivable most recent of the unconformities can be correlated withwidespread lake-level lowstands known from elsewherethat this distinction is simply an artefact of the penetra-

tion capabilities of the sparker system, and that improved in intertropical Africa during the late Pleistocene (seeRegional and global climatic events section below).seismic resolution at depth would reveal a pattern similar

to that observed in the D–A Sequences. However, wethink this is probably not the case. The profound strati-

Regional tectonovolcanic eventsgraphic heterogeneity between basins prior to SequenceD and the similarities after Sequence D argue for earlier The timing of Neogene tectonism in the northern Lake

Tanganyika area is consistent with other tectonic eventstectonovolcanic controls as opposed to later climatic ones.Furthermore, Project PROBE seismic line 16 (which occurring throughout the western rift region (Fig. 1A).

From the mid- to late Miocene, volcanic eruptionspenetrated to basement across nearly the same transectas CASIMIR Sparker Line S

2) (Fig. 1B) clearly shows a occurred in the Kivu–Rusizi region north of Lake

Tanganyika (Bellon & Pouclet, 1980; Kampunzu et al.,distinction between a lower package (prior to our Dsequence) of reflectors, with an overall fan-shaped mor- 1983; Pasteels & Boven, 1989; Pasteels et al., 1989). This

period corresponds to the initiation of what would becomephology, separated from an upper (D–A sequence) pack-age of comparatively flat-lying reflectors (figs 2 and 3 in the Kivu–Rusizi local volcanic dome (Pouclet, 1976;

Ebinger, 1989a,b; Coussement, 1995). Simultaneously, inLezzar et al., 1996). These lines of evidence, of course,do not eliminate the possibility that climatically controlled Northern Lake Malawi, faulting and volcanism resulted

in a broad asymmetric lake basin (Ebinger et al., 1993).lake-level fluctuations occurred prior to the deposition ofthe D Sequence, but only that if such events did occur, These ages coincide with the minimum RSRM age

estimate of #7.4 Ma for the initiation of the Northerntheir stratigraphic signature was completely over-shadowed by major tectonovolcanic events. Lake Tanganyika Basin (Cohen et al., 1993a; Lezzar

et al., 1996). The initiation of rifting and formation ofLake-level forcing mechanisms, related to the lake’shydrological budget, are probably responsible for the the Lake Kivu Basin and the nearby Rusizi Basin

occurred between #7.5 and #4 Ma, with a culminationpronounced cyclicity of the stratigraphic record since thedeposition of the D Sequence (which we estimate as the of volcanic activity centred between 6 and 5.5 Ma

(Ebinger, 1989a; Pasteels et al., 1989). These eventspast #400 ka). The widespread nature of the four majorand two minor unconformities that we observe (labelled probably correspond to the beginning of the RBM Phase

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132 121

Page 16: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

A. S. Cohen et al.

(RSRM estimate #7.4–1.1 Ma) in the northern Lake & Webb, 1993). Simultaneous with this late Plioceneglobal cooling, regional aridification occurred in North,Tanganyika Basin. An increase in subsidence rates and a

reduction in volcanic activity occurred in the south Kivu East and Central Africa (de Heinzelin, 1955; Bonnefille,1976; Cerling & Hay, 1986; Cerling et al., 1988; Pickfordregion between #4 and #2 Ma, simultaneous with the

active development of the North Rusizi Basin (Ebinger, et al., 1991; Suc et al., 1993; de Menocal, 1995). InNorth Africa, aridification appears to have begun inter-1989a). During this same period, but further south, in

Lake Malawi, horizontal and vertical crustal movements mittently, by #3.2 Ma in the north-west Sahara (LeRoy& Dupont, 1993) and by #2.6 Ma in the eastern Saharaenhanced basinal asymmetry. From #2 Ma to the pre-

sent, a major episode of rifting has been associated with (Suc et al., 1993). Aeolian dust accumulation, indicativeof regional aridification, also increases in the ocean basinsrenewed, intense volcanism in the Kivu province (Bellon

& Pouclet, 1980; Pasteels et al., 1989) and, in northern surrounding Africa starting at about 2.8 Ma (deMenocal,1995). A cooler and drier climate is evidenced from theLake Malawi, with motions along border faults, intrabasi-

nal faults and uplift of the rift flanks that have resulted Ethiopian Rift uplands at 2.5–2.35 Ma (Bonnefille, 1983)and rainfall decreased dramatically in the Turkana Riftin the formation of the modern basin topography (Ebinger

et al., 1993). Unconformities RSB2

and RSB3, identified at 2.0–1.8 Ma (Cerling et al., 1977). For the period

between 1.9 and 1 Ma, core data o� the coast of weston Project PROBE multichannel seismic lines (Rosendahlet al., 1988) within the RBM Phase seismic sequences, Africa indicate relatively constant climatic conditions

(Stein & Sarnthein, 1984). Given these major global andare probably related to these major regional tectono-volcanic events. regional climatic changes throughout the late Neogene,

some of the main unconformities observed in the RBMPhase deposits in the Lake Tanganyika infill (RosendahlRegional and global climatic eventset al., 1988; Lezzar et al., 1996; RSRM estimated agerange #7.4–1.1 Ma) could be climatically driven, evenChanges in lake levels in Lake Tanganyika during the

Neogene may be related to regional and even global if tectonovolcanic events were of primary importance.Phase F–E of the Tanganyika basin evolution (RSRMclimatic events. From #7 Ma and #5 Ma, a sharp

decline occurred in global temperature, along with a estimated age #1.1 Ma to #0.39 Ma) corresponded toa period of significant tectonism (Lezzar et al., 1996),pronounced lowering of sea level. These events a�ected

Africa profoundly, and resulted in major changes in and the dominant signature in the stratigraphic recordof this period that can be resolved using thecontinental biota associated with cooler temperatures

(Van Zinderen Bakker & Mercer, 1986). This cool period CENTIPEDE sparker system appears to be tectonic.From Phase D to the present, the four most recent ofwas followed by rapid warming after #5 Ma and a global

rise in sea level (Van Zinderen Bakker & Mercer, 1986; the unconformities, (d) to (a), may be correlated withevidence for climatic shifts known from elsewhere inKrantz, 1991; Webb & Harwood, 1991; Harwood &

Webb, 1993). An expansion of proto-Lake Tanganyika intertropical Africa during the late Pleistocene. Regionallycool climates and a lowering of sea level from #450 tooccurred at about 3.6 Ma, either as a result of increased

prerift downwarping or alternatively as a consequence of 350 ka (Van Zinderen Bakker & Mercer, 1986) may becorrelated with the (d) Tanganyikan low lake standwetter climatic conditions. A short-lived cold episode has

been identified worldwide at about 3.6 Ma, followed in (Fig. 2H). Between #350 and 300 ka, a long-termclimatic warming in Africa has been identified fromthe southern hemisphere by warmer conditions which

prevailed until 2.6 Ma (Shackleton & Kennett, 1975; sediment cores of the Zaire deep-sea fan ( Jansen et al.,1984; Gasse et al., 1989). This event may correspond toMercer, 1983, 1984). In the East African Rift, a major

expansion of lacustrine conditions in the Malawi Basin the D sequence highstand in Lake Tanganyika (Fig. 2I).At #270 ka, a major dry episode has been defined inis recorded from the Chiwondo beds during the mid-

Pliocene, somewhat prior to 4.0 Ma, based on biostrati- the Congo Basin (Gasse et al., 1989), which may correlatewith the (c) low lake level (Fig. 2J). Pronounced increasesgraphic correlations with radiometrically dated, fossilifer-

ous sediments in East Africa (Bromage, 1995). Evidence in arid-climate indicators (phytolith concentrations, aeol-ian dust and diatoms from deflated lake beds) occur infor lake expansion and wetter climates elsewhere in

Central and East Africa during the mid-Pliocene also deep-sea sediments o� the west coast of Africa between300 and 260 ka (deMenocal et al., 1993). In nearshoreexists (Gautier, 1970; Cerling et al., 1977; Pickford, 1990;

Verniers & de Heinzelin, 1990) and Gasse (1990) has core records from the Arabian Sea, evidence for a weakAsian monsoon also occurs between 300 and 250 kasuggested that a wetter climate may have been responsible

for the expansion and freshening of lakes in the Afar (Anderson & Prell, 1993). At #250 ka, high lake standsdescribed in the Lake Magadi–Natron basin of theregion (Djibouti) during the same period. Evidence also

exists for increased runo� and precipitation in the Sahara southern Kenya Rift (Hillaire-Marcel et al., 1986;Sturchio et al., 1993) may correlate with the beginningbetween #3.7 and 3.5 Ma (LeRoy & Dupont, 1993).

Between 2.6 and 1.8 Ma, temperature dropped world- of the transgressive C Phase in Lake Tanganyika(Fig. 2K). Between #250 and 180 ka deep-sea evidencewide, associated with the reglaciation of both the

Northern and Southern hemisphere ice sheets (Harwood from both the East and West African coasts indicates

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132122

Page 17: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

Palaeogeography of Lake Tanganyika

increases in intensification of the Asian monsoon, an regressive periods have been identified within the mostrecent phase (Phase A) in Northern Lake Tanganyika’sincrease in precipitation and probably higher lake levels

(based on fewer deflated freshwater diatoms) (Anderson history, by our estimates from 40 ka to the present.These are variously estimated (RSRM) or dated radio-& Prell, 1993; de Menocal et al., 1993). At 180 ka, Gasse

et al. (1989) have identified another dry episode from the metrically from #40–35 ka to #23 ka, from #23 ka to#18 ka and from #18 ka to the present (Tiercelin &Congo Basin that can be correlated with the (b) low lake

level in Lake Tanganyika (Fig. 2L). Deep-sea evidence Mondeguer, 1991; Bouroullec et al., 1991; Lezzar et al.,1996). This pattern of lake-level fluctuation is similar tofrom both coasts of Africa support generally more arid

conditions between #200 and 140 ka (Anderson & Prell, that identified for much of intertropical Africa (Street &Grove, 1979; Van Zinderen Bakker, 1982; Bonnefille &1993; deMenocal et al., 1993).

Highstand conditions characterize Phase B in northern Riollet, 1988; Tiercelin et al., 1988; Gasse et al., 1989;Street-Perrot et al., 1989; Vincens, 1989, 1993; VincensLake Tanganyika, which we estimate occurred between

#170 and 40 ka (Fig. 2M). At #145–120 ka, highstand et al., 1993; Chalie, 1995), although a major exception tothis regional picture is known from Lake Malawi, whereconditions also prevailed in Lake Turkana in the northern

Kenya Rift (Butzer et al., 1969, 1972), in the Magadi– early Holocene major lake lowstands date from 10 to 6ka ( Johnson & Davis, 1989; Finney & Johnson, 1991;Natron basin in the southern Kenya Rift (Hillaire-Marcel

et al., 1986) and in the northern Kenya Rift (Sturchio Finney et al., 1996).The RSRM age estimates for the (d), (c), (b) and (a)et al., 1993). Phytolith abundance data from the West

African deep-sea record suggest a strong shift in the unconformities (Lezzar et al., 1996) bear a strikingrelationship to the timing of marine oxygen isotope stagesposition of the African tall-grass savannah between 140

and 40 ka (deMenocal et al., 1993). In the Arabian Sea, 10, 8, 6 and 2, respectively (Pisias & Leinen, 1984)(Fig. 5). Previous workers have demonstrated a corre-a renewed intensification of the Asian monsoon is indi-

cated starting at about 150 ka, and continuing until spondence between low lake levels in the tropics/subtrop-ics of East and North Africa, and high-latitude, late#40 ka (Anderson & Prell, 1993). Lezine & Casanova

(1991) have argued for a series of humid phases in Pleistocene glacial events (late Weischelian/Wisconsin,oxygen isotope stage 2) (Street & Grove, 1979; Flohn &Africa at 140–118 ka, 105–96 ka, 92–73 ka, 52–44 ka

and 12–2 ka, based on pollen and dinocyst records Nicholson, 1980; Littman, 1989). Gasse et al. (1990) havefurther argued that transitions from arid to humid con-from the eastern Atlantic.

At Lake Malawi the record is more complex. The baseof the Songwe Sequence (a major lowstand indicator incentral Lake Malawi) has been assigned highly variableage estimates at di�erent locations, between 150 and 44ka (Scholz & Finney, 1994). This sequence bearsintriguing similarities to the combined B–A Sequence ofLake Tanganyika, both in terms of similar seismic faciescharacteristics and thicknesses at comparable waterdepths (Scholz & Finney, 1994; Lezzar et al., 1996; DeBatist et al., 1996), but the age uncertainties make anycorrelations premature at this time.

In Lake Abhe, in the Afar region of north-east Africa,Gasse (1977) and Gasse & Street (1978) defined twotransgressive phases between #100 ka and #40 ka thatare contemporaneous with the B Phase in LakeTanganyika. Prior to 30 ka, cool and dry conditionscharacterize the central African highlands (Bonnefille &Riollet, 1988; Bonnefille et al., 1990). Around 40–31 ka,a regression occurred in Lake Abhe, following a decreasein temperature (Gasse, 1977; Gasse & Street, 1978). This

Fig. 5. RSRM age-estimated lake-level chronology for Lakecooling event may be correlated with the North LakeTanganyika superimposed over the marine oxygen isotopeTanganyika (a) low lake period which we estimatechronology for the north-western Pacific (Pisias & Leinen,occurred between #40 and 35 ka (Fig. 2N). This may1984). The highstand elevations for Lake Tanganyika arealso correspond to the #250 m lake level drop recordedconstrained only by the fact that deposits in exploration wells

for Lake Malawi between >40 and 28 ka (Scholz & on the Rusizi plain, near the present-day lake margin, areFinney, 1994), although Finney et al. (1996) have sug- almost entirely composed of either fluvial or marginal lakegested that lake-level fluctuations in Lakes Tanganyika deposits. The ‘B’ Sequence is thought to have been depositedand Malawi may be out of phase with one another during when lake levels were #55 m higher than present, based on athe Late Pleistocene. probable correlation with nearshore lacustrine deposits at this

elevation in the Rusizi River valley.Following the (a) low lake level, several transgressive–

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132 123

Page 18: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

A. S. Cohen et al.

ditions are synchronous across a broad swath of both the subtle sedimentological or palaeontological evidence thatmust be deduced from cores.Saharan and the Sahelian regions of Africa. This corre-

spondence has been interpreted in various ways(deMenocal et al., 1993). Numerous authors have stated IMPLICATIONS FOR THE EVOLUTIONthat there are strong interconnections between glacial AND BIOGEOGRAPHY OF LAKEconditions at high latitudes (resulting in global cooling TANGANYIKA ENDEMIC SPECIESof sea surface temperatures) and an intensification of

The history of Lake Tanganyika fragmentation and cre-coastal upwelling along the west coast of Africa, with aation of patches of littoral rocky substrates along majorcontemporaneous weakening or elimination of monsoonalborder faults has important implications for understand-circulation across equatorial Africa (Parkin & Shackleton,ing the evolution of endemic organisms in the lake1973; Muller & Suess, 1979; Flohn & Nicholson, 1980;(primarily cichlid fish, ostracods and gastropods), manyLittman, 1989; Gasse et al., 1990). During the lateof which are restricted to these habitats (Coulter, 1991;Pleistocene and Holocene, the intensification of theMichel et al., 1992; Sturmbauer & Meyer, 1993).African monsoon may have been forced by the strength-

Similarly, the origin of major regions of soft-bottomened northern hemisphere summer insolation that occurssubstrates within the littoral zone can be traced to theirevery 23–19 ka as a result of the Earth’s orbital precessiontectonostratigraphic setting in the lake and the lake-level(Hillaire-Marcel et al., 1986; Prell & Kutzbach, 1987;conditions prevailing at the time. These soft-bottomLezine & Casanova, 1991).benthic habitats house their own endemics and act asThe dominant periodicities of climate fluctuationspotential substrate barriers to dispersal for rock-substrateduring the late Neogene in Africa may have varied overdwellers. Our palaeogeographic reconstructions indicatetime. Using deep-sea dust records, deMenocal (1995) hasrepeated episodes of desiccation for the northernmostinferred that 23- and 19-ka (orbital precession) cyclespart of Lake Tanganyika (Fig. 2A–N). Small and probablywere dominant during the period from 4.5 to 2.8 Ma.closed basin lakes occupied the topographic sumps of theBetween 2.8 and 1.0 Ma, 41-ka cycles dominate the dustevolving structural subbasins in this region, in locationsrecord, coinciding with the onset of similar durationthat are not directly predictable from modern bathymetry.cycles at high latitudes. Since 1.0 Ma, however,As indicated by our palaeoshoreline reconstructionsdeMenocal shows that 100-ka cycles have dominated the(Fig. 6A–N), the northern shoreline of Lake TanganyikaAfrican dust record, corresponding to the large-amplitudecan be subdivided into a series of discrete segments ofglacial cycles of high latitudes. However, the absence ofrocky or sandy substrate, each with a separate history.

extensively and continuously dated lacustrine sequencesThese include the following regions, from south to north.

of middle Pleistocene age in Africa has prevented palaeo-limnologists from evaluating this chronology directly.

Ubwari Peninsula, East Coast. This rocky coast today isOur RSRM age estimate results, tied to the geographi- formed by the East Ubwari Fault, the western escarpment

cally discontinuous but better dated evidence from else- margin of the North Kigoma half-graben (Fig. 1B). Thewhere in Africa, support the notion that cyclicity at the East Ubwari Fault has been active since synrift deposition#100-ka frequency has in fact been in place for at least began in this area (by our estimate, about 3.6 Ma in thethe past 0.4 Ma. It is also consistent with the temporal north Kigoma region). Since the (f ) lowstand (by ourarguments for orbital forcing mechanisms controlling estimate #1.1 Ma) the northern part of the East Ubwaritropical lake levels at an orbital eccentricity time scale Fault appears to have been inundated along the coastline(Olsen, 1990), though no shorter duration cyclicity is (Figs 2C,D and 6C,D). Thus rocky habitat has persistedevident from our data. Elsewhere, Reynolds (1992, 1994) in this area for at least this long.has argued on theoretical grounds and through thegeneration of synthetic seismic models that a 100-ka Burton’s Bay. The present-day Burton’s Bay formedMilankovitch eccentricity cyclicity should be identifiable starting at the time of the (d) sequence boundary (RSRMin reflection seismic profiles (depending on seismic source estimate #390–360 ka) following the formation of thesignature) from an equatorial lake like Tanganyika. Our Baraka furrow, through southwards-directed rift propa-RSRM age estimate results are consistent with this gation between the southern extension of the Uviraprediction. It is important to note, however, that our Border Fault System (UBFS) and the Cape Banza Faultresults do not provide a priori evidence against higher (CBF) (Figs 2H,I and 6H,I), resulting in progressivelyfrequency cyclicity in lake-level fluctuations, simply younger maximum ages of the bay toward the southbecause such cycles are not evident in a seismic strati- (Lezzar et al., 1996). The bay was probably completelygraphic record. Major lake-level excursions that occur dry during the (a) boundary lake-level lowstand (RSRMover brief intervals of time may go unreflected in the estimate #40–35 ka) (Fig. 6N). The east side of Burton’srecord of major unconformities and sequence geometry Bay comprises a rocky coastline that follows the Capein large lakes, or may be unresolvable given the seismic Banza Fault, and would have formed a steep but dryacquisition parameters we have used. Evidence for these valley margin following the Ubwari Peninsula during late

Pleistocene periods of lake-level lowstand. This area oftypes of events may only leave their mark through more

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132124

Page 19: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

Palaeogeography of Lake Tanganyika

Burton’s Bay could have been recolonized during the South Rusizi half-graben formed a separate lake(Fig. 6L).past #35 kyr, presumably from the east side of the

Ubwari Peninsula.The south side of Burton’s Bay forms the southern North Burundi rocky coast. This coast represents the

north-eastern escarpment margin of the South Rusiziaxial margin of the South Rusizi half- graben. By analogywith present-day conditions around most major axial half-graben, close to the present-day Cape Magara Fault.

This sublacustrine high and fault system may havedrainage deltas, this area has probably been a zone ofeither marshland or valley cutting during lowstand con- formed as a consequence of uplift/subsidence motion

along the Ubwari horst and the South Rusizi half grabenditions, with intermittent sandy coasts, for its entirehistory. (Fig. 6C,I), which have probably been intermittently

active throughout the history of rifting in the area.However, these faults appear to have undergone a periodZaire coast north of the Baraka River. This is the western

shoaling margin of the South Rusizi half-graben of quiescence prior to the formation of the (d) boundary(RSRM estimate #390–360 ka), during which time they(Fig. 1B). It appears to have been a sandy/muddy coast

for most of its history. During the early, active subsidence may not have formed a rocky coast (Fig. 6A–H).Resumption of faulting would have renewed the existencephase of the South Rusizi half-graben (Fig. 2A–H) it

probably sloped steeply o�shore, but the o�shore gradient of rocky areas during or slightly before Phase C (RSRMestimate #260–190 ka). However, following the (c) andbecame flatter following the end of active subsidence at

the (d) boundary (RSRM estimate about 390–360 ka). (b) lake-level lowstands (Fig. 6J, RSRM estimate #290to #260 ka, and Fig. 6L, RSRM estimate #190 to#170 ka, respectively) the area of the Cape Magara FaultNorth Zaire coast along the Uvira Border Fault System

(UBFS). This coast is formed by the UBFS, the western would have been above lake level. Thus environmentssuitable for rock-dwelling endemics have probably beenescarpment margin of the North Rusizi half-graben. It

has probably been an area of rocky outcrops for most of in continuous existence along the north-eastern coast ofLake Tanganyika since some time after #170 ka (i.e.the duration of the North Rusizi half-graben, starting at

6–5 Ma (Ebinger et al., 1989a,b; Coussement, 1995). since the resumption of high lake stand conditions duringSequence B). Regardless of the accuracy of the RSRMThe region was completely above lake level during the

(c) boundary low lake stand event (RSRM estimate estimates, the relative sequence chronology demonstratesthat this is a shorter continuous duration of rocky habitat#290–260 ka, Figs 2J and 6J). Thus its current phase

of colonization by lacustrine species must postdate that than has existed along the other northern lake rockycoastlines. Rocky corridors connecting Cape Magarainterval. Since that time it has probably been in continu-

ous existence as a rocky habitat. (Burundi) to Cape Banza (Zaire) may have been devel-oped briefly during lake level fall at the end of Phase Band again early in Phase A (Fig. 6M,N), although thisThe Rusizi River deltaic environment is located at the

extreme end of the North Tanganyika Basin. The thick would probably have only lasted a few thousand years.(700–800 m) fluviodeltaic sequence present in wellsdrilled by AMOCO on the modern Rusizi River delta Central Burundi sandy coast. This coast formed as the

shoaling margin of the North Kigoma half-grabensuggests it has been in continuous existence as an axialdrainage throughout the history of Lake Tanganyika, (Fig. 1B). Sandy corridors have existed along this coast

continuously since the (f ) limit for the North Kigomaalthough unconformities in its record suggest its dis-charge may have been low and/or its discharge point half-graben (which we estimate to be at least #550 ka),

although parts of the coastline have been reduced substan-avulsing at various times (Attou, 1988). The delta fronthas probably been marshy for much of its history. tially during lake-level falls. Rocky habitats may have

existed sporadically along this coast, particularly duringConsequently, the Rusizi represents one of the mostpermanent, although not particularly wide, barriers periods of activity along the normal or strike-slip faults

on the Rumonge Platform (Figs 1B and 2N).between rocky substrates along the lake’s northerncoastline.

South Burundi rocky coast. The present-day SouthBurundi rocky coast is a northern extension of the EastNorth Burundi sandy coast. This coast, part of the

Bujumbura platform (Fig. 1B), developed on the rollover Kigoma Border Fault System. Project PROBE data(Rosendahl et al., 1988) suggests that this fault systemzone of the shoaling margin of the active North Rusizi

half-graben. This coastline has migrated on and o�shore has been active periodically since the Nyanja Event,although the precise timing of activity is unknown. Itswith changes in lake level and, prior to the middle

Pleistocene, with changes in local volcanic doming activity history cannot be deduced from the CASIMIR data set,except to the extent that this region would have formed(Kivu–Rusizi local dome) (Figs 2A and 6A) and it has

periodically been connected to the central/southern a sandy shoreline during extreme lake-level lowstands,when the border fault system in this area would haveBurundi sandy coasts (most recently during the (b) lake

lowstand – RSRM estimate #190–170 ka), when the been exposed on land.

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132 125

Page 20: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

A. S. Cohen et al.

Fig

.6.(

Ato

N)P

alae

osho

relin

ere

cons

truc

tions

for

nort

hern

Lak

eT

anga

nyik

afo

rth

etim

efr

ames

/int

erva

lsdi

scus

sed

inth

ete

xt(s

eeal

soF

ig.2

A–N

).Pa

laeo

shor

elin

esu

bstr

ate

desi

gnat

ions

are

inte

nded

tosh

owth

edo

min

ants

ubst

rate

type

soc

curr

ing

atth

etim

ein

the

near

shor

een

viro

nmen

t.Fa

ultn

ame

abbr

evia

tions

are

the

sam

eas

for

Fig

.1B

.The

wid

thof

the

subs

trat

eba

ndis

not

sign

ifica

nt.b

pll=

elev

atio

nof

lake

surf

ace

belo

wpr

esen

tlak

ele

vel.

RSR

M=r

eflec

tion

seis

mic

radi

ocar

bon

met

hod.

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132126

Page 21: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

Palaeogeography of Lake Tanganyika

Fig

.6.(

cont

inue

d).

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132 127

Page 22: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

A. S. Cohen et al.

The lake level and palaeogeographic history that we Marc De Batist and the RCMG Team (Renard Centreof Marine Geology, University of Ghent, Belgium) forpropose predicts that the modern fauna of northern Lake

Tanganyika must have recolonized this portion of the the sparker data acquisition and processing. Thanksto Herve Bellon for his great help in constructing thelake on repeated occasions from refugia to the south

(including the Cape Banza region). Furthermore, the tectonovolcanic correlation sections, and to Bill Wescottand Denise Stone, and AMOCO, for making well logs,three major zones of rocky habitat at the north end of

the lake can now be ordered in terms of the duration of samples and unpublished data available to us in thepreparation of this paper. Special thanks to Gashagazacontinuous habitability (since they were last desiccated),

with Cape Banza being the oldest (at the time of post- Masta Mukwaya, Gaspard Ntakimazi, Laurent Ntahugaand Pontien Ndabaneze for facilitating this research, toNyanja Event flooding, by our estimates >#3.6 Ma,

Fig. 6C), northern Zaire, much younger but intermediate Bernadette Coleno for assistance with illustrations, andto Michael Talbot, Chris Scholz, Dave Reynolds and(following the (c) lowstand, estimated at #290–#260

ka, Fig. 6J), and northern Burundi the youngest (follow- Kathleen Nicoll for their many useful suggestions onways to improve this paper. This is publication :50 ofing the (b) lowstand, postdating #170 ka, Fig. 6M).

This chronology can serve to constrain phylogenetic the International Decade of East African Lakes (IDEAL)programme.hypotheses about episodes of speciation and the geo-

graphical origin of lacustrine endemic fish and invert-ebrates in the northern part of Lake Tanganyika.

REFERENCES

A , P., C, L., I, M., J, R. & M, M.

(1988) The Miocene Lacustrine Basin of Rubielos de Mora.CONCLUSIONIAS/IGCP Project 219 Excursion Guidebook, 32, pp. Inst.

In this study we have outlined a series of working de Geologia ‘Jaume Almera’ CSIC, Barcelona.A , P., C, L., J, R. & M, M. (1991)hypotheses about the chronology of tectonic, geographical

Sequential arrangement and asymmetrical fill in the Mioceneand climatic events in the northern Lake TanganyikaRubielos de Mora Basin (northeast Spain). Spec. Publ. Int.region. Our conclusions are broadly consistent with whatAss. Sediment., 13, 257–275.was previously known about regional tectonic and climatic

A, D. M. & P, W. L. (1993) A 300 Kyr record ofhistory, giving us considerable confidence that RSRMupwelling o� Oman during the Late Quaternary: evidenceage estimation provides a useful approach for understand-of the Asian Southwest monsoon. Paleoceanography, 8,

ing lake basin histories. As such it may provide a model 193–208.for application in other lake basins. This study has also A, A. (1988) Evolution paleogeographique des fosse nord-produced important testable hypotheses about the timing Tanganyika et de la Rusizi (Rift-est-africain) depuis le Mioceneof events of critical importance for fields as disparate as superieur. Diplome d’Etudes Approfondies GIS Oceanologiepalaeoclimatology and evolutionary biology. Our chal- et Geodynamique, Universite de Bretagne Occidentale,

Brest, France.lenge now, and the challenge for others interested in theB, F. (1991) Late Pleistocene and Recent detrital sedi-geological history of the African Great Lakes, is to obtain

mentation in the deep parts of northern Lake Tanganyikalong records through deep drilling programmes that can(East African Rift). Spec. Publ. Int. Ass. Sediment., 13,confirm or refute the hypotheses we present here. In147–173.many ways our ideas about chronology are analogous to

B, H. & P, A. (1980) Datations K/Ar de quelquesthose derived in planetary geology from crater counting.laves du Rift-ouest de l’Afrique Centrale: implications sur

We will only know the full story when we finally get a l’evolution magmatique et structurale. Geol. Rdsch., 69,chance to visit and sample the rocks. 49–62.

B, R. (1976) Palynological evidence for an importantchange in vegetation of the Omo Basin between 2.5 and 2million years. In: Earliest Man and Environments in the LakeRudolf Basin (Ed. by Y. Coppens, F. C. Howell, G. L. IsaacACKNOWLEDGEMENTS and R. E. Leakey), pp. 421–431. University of ChicagoPress, Chicago.Research authorizations were provided by the Ministry

B, R. (1983) Evidence for a cooler and drier climateof Higher Education and Scientific Research of thein the Ethiopian uplands towards 2.5 Myr ago. Nature,Republic of Burundi, the Ministry of Energy and303, 487–491.Ministry of Scientific Research and Technology of the

B, R. & R, G. (1988) The Kashiru PollenRepublic of Zaire and the Tanzanian Commission for Sequence (Burundi). Paleoclimatic implications for the last,Science and Technology. The 1992 high-resolution seis- 40,000 year B. P. in tropical Africa. Quat. Res., 30, 19–35.mic programme in the North Tanganyika Basin was B, R., R, J. C. & G, J. (1990) Temperaturesupported by the CASIMIR Project (Belgium), Elf- and rainfall estimates for the past 40,000 years in equatorialAquitaine Production (France), the Ministry of Foreign Africa. Nature, 346, 347–349.

B, J. L., R, J. P., R, J., T, J. &A�airs (France) and INSU-CNRS (France). Thanks to

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132128

Page 23: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

Palaeogeography of Lake Tanganyika

M, A. (1991) Quaternary sedimentary processes Est-africain. Thesis, Universite de Bretagne Occidentale,Brest, France.and dynamics in the northern part of the Lake Tanganyika

D B, M., V R, P., B, S. & K, J.Trough, East African Rift System. Evidence of lacustrine(1996) Structural framework, sequence stratigraphy and lakeeustatism? Bull. Centres Rech. Explor.-Prod. Elf Aquitaine,level variations in the Livingstone Basin (northern Lake15, 343–368.Malawi). First results of a high-resolution reflection seismicB, J. L., T, C., T, J. J., R, J.,study. In: The Limnology, Climatology and PaleoclimatologyR, J. P. & M, A. (1992) Sequences sismiquesof the East African Lakes (Ed. by T. Johnson and E. Odada),haute resolution du fosse nord-Tanganyika, Rift Est-africain.pp. 509–522. Gordon & Breach Publ., Amsterdam.Implications climatiques, tectoniques et hydrothermales. C.

M, P. B. (1995) Plio-Pleistocene African climate.R. Acad. Sci. Paris, 315, 601–608.Science, 270, 53–59.B, J. P., F, R. M. & T, R. S. (1989)

M, P. B. & R, D. (1993) Sensitivity of Asian andLate Quaternary paleolimnology of Walker Lake, Nevada.African climate to variations in seasonal insolation, glacial iceJ. Paleolimnol., 1, 249–268.cover, sea surface temperature, and Asian orography.B, T. (1995) Paleobiogeography of the Malawi Rift:J. geophys. Res., 98, 7265–7287.Vertebrate and invertebrate faunal contexts of the Chiwondo

M, P. B., R, W. F. & P, E. M. (1993)Beds, northern Malawi. J. Human Evol., 28, 37–58.Influences of high- and low latitude processes on AfricanB, K. (1971) Recent history of an Ethiopian delta: The Omoterrestrial climate: Pleistocene eolian records from equatorialRiver and the level of Lake Rudolf. University of ChicagoAtlantic Ocean Drilling Program Site 663. Paleoceanography,Department of Geography Res. Paper, 136 .8, 209–242.B, K. W., B, F. W. & T, D. L. (1969)

E, C. (1989a) Geometric and kinematic development ofHorizontal sediments of the lower Omo Valley. The Kibishborder faults and accommodation zones, Kivu-Rusizi rift,Formation. Quaternaria, 11, 15–29.Africa. Tectonics, 8, 117–133.B, K., I, G. L., R, J. L. & W-

E, C. (1989b) Tectonic development of the westernK, C. (1972) Radio-carbon dating of East African lakebranch of the East African Rift System Bull. Geol. Soc. Am.,levels. Science, 175, 1069–1076.101, 885–903.C, A. (1952) Le millieu geographique et geophysique. In:

E, C., D, A. L., T, A. L., B, T. & R,Exploration Hydrobiologique du Lac Tanganyika, 1946–47,U. (1993) Tectonic controls on rift basin morphology: evol-Resultats Scientifiques (Ed. by E. Leloup), 1, pp. 3–27. Inst.ution of the Northern Malawi (Nyasa) Rift. J. geophys. Res.,Royal des Sciences Naturelles de Belgique, Bruxelles.98, 821–836.C, T. E., B, J. R. & O’N, J. R. (1988) An

F, B. P. & J, T. C. (1991) Sedimentation in Lakeisotopic study of a fluvial-lacustrine sequence: the Plio-Malawi (East Africa) during the past, 10,000 years: a continu-Pleistocene Koobi Fora Sequence, East Africa. Palaeogeog.ous paleoclimatic record from the southern tropics.Palaeoclim. Palaeoecol., 63, 335–356.Palaeogeogr. Palaeoclim. Palaeoecol., 85, 351–366.

C, T. E. & H, R. L. (1986) An isotopic study ofF, B. P., S, C. A., J, T.C., T, S.

palaeosol carbonates from Olduvai Gorge. Quat. Res., 25,& S, J. (1996) Late Quaternary lake level changes of

63–78. Lake Malawi. In: The Limnology, Climatology andC, T. E., H, R. L. & P, D. W. (1977) Isotopic Paleoclimatology of the East African Lakes (Ed. by T. Johnson

evidence for dramatic climatic changes in East Africa during and E. Odada), pp. 495–508. Gordon & Breach Publ.,the Pleistocene. Nature, 267, 137–138. Amsterdam.

C, F. (1995) Paleoclimats du bassin Tanganyika Sud au F, H. & N, S. (1980) Climatic fluctuations incours des 25 derniers milles ans: reconstitution quantitative the arid belt of the ‘Old World’ since the last glacialpar le traitement statistique de donnees polliniques. C. R. maximum; possible causes and future implications. Palaeoecol.Acad. Sci. Paris, 320, 205–210. Africa, 12, 3–21.

C, S., P, W., M, D., S, G. & G, F. (1977) Evolution of Lake Abhe (Ethiopia and TFAI),W, G. (1991) Forcing mechanisms of the Indian Ocean from 70,000 b.p. Nature, 265, 42– 45.monsoon. Nature, 353, 720–725. G, F. (1990) Tectonic and climatic controls on lake distri-

C, A. S. (1982) Ecological and paleoecological aspects of the bution and environments in Afar from Miocene to Present.Rift Valley Lakes of East Africa. PhD dissertation, University In: Lacustrine Basin Exploration-Case Studies and Modernof California-Davis, CA. Analogs (Ed. by B. Katz), Mem. Am. Ass. Petrol. Geol.,

C, A. S., C, C. G. & A , P. (1993b) Three 42, 19–42.dimensional sequence stratigraphic modelling of the Miocene G, F., L, E. V., M, M. & F, J.-C (1989)Rubielos de Mora Basin, Spain. Geol. Soc. Am. Ann. Mtg. Water level fluctuations of Lake Tanganyika in phase withAbst. w/prog., 25, 113–114. oceanic changes during the last glaciation and deglaciation.

C, A. S., S, M. R. & S, C. A. (1993a) Nature, 342, 57–59.Estimating the age of formation of lakes: an example from G, F. & S, F. A. (1978) Late Quaternary lake-levelLake Tanganyika, East African Rift System. Geology, 21, fluctuations and environments of the northern rift valley and511–514. Afar region (Ethiopia and Djibouti). Palaeogeog. Palaeoclim.

C, G. W. (1991) Lake Tanganyika and its Life. Oxford Palaeoecol., 24, 279–325.University Press, London. G, F., T, R., D, A., G, E. & F, J. C.

C, C. (1995) Structures transverses et extension intra- (1990) The arid-humid transition in the Sahara and Sahelcontinentale. Le role des zones de failles d’Assoua et Tanganyika- during the last deglaciation. Nature, 346, 141–146.

G, A. (1970) Fossil freshwater Mollusca of the LakeRukwa-Malawi dans la cinematique neogene du systeme de Rift

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132 129

Page 24: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

A. S. Cohen et al.

Albert-Edward rift (Uganda). Annales Mus. Roy. Afr. Centr., K, P. W. (1992) Large African lakes as natural laboratoriesSer. IN-80 Sci. Geol. 67:1–144. Tervuren, Belgium. for evolution: examples from the endemic gastropod fauna

H, D. & W, P.-N. (1993) Impact and evidence of of Lake Tanganyika. Mitt. Internat. Verein. Limnol., 23,Pliocene warmth on Antarctic paleoclimate and ice sheet 85–99.collapse. Am. Ass. Strat. Palynol. 26th ann. mtg. abst. M, A. (1991) Bassins sedimentaires en contexte extensifw/prog., p. 36. et decrochant: l’exemple du ‘Complexe des fosses Sud-

H, (), J. (1955) Le fosse tectonique sous le Parallele Tanganyika’. Rift Est-africain. Morphostructures et sedimen-d’Ishango. Inst. Parc. Nat. Col. tation. Thesis, Universite de Bretagne Occidentale, Brest,

H-, C., C, O. & C, J. (1986) 14C France.and Th/U dating of Pleistocene and Holocene stromatolites M, C. (1988) Variable extension in Lake Tanganyika.from East African paleolakes. Quat. Res., 25, 312–329. Tectonics, 7, 785–801.

I, L. K. (1984) Etude geomorphologique et stratigraphique M, R. B. (1991) Quaternary stratigraphic, hydrologicde la plaine de la Rusizi. Thesis, Vrije Universiteit Brussel. and climatic history of the Great Basin, with emphasis on

J, J. H. F., V , T. C. E., G, R. & V Lakes Lahontan, Bonneville, and Tecopa. In: The Geology of, J. (1984) Middle and Late Quaternary oceanography North America (Ed. by A. R. Palmer and J. O. Wheeler).and climatology of the Zaire-Congo fan and the adjacent Geol. Soc. Amer., Boulder, CO, USA.eastern Angola basin. Netherlands J. Sea Res., 17, 201–249. M, T. E. (1995) Palynological and Tephra Correlations

J, T. C. & D, T. W. (1989) High resolution seismic Among Deep Wells in the Modern Great Salt Lake, Utah, USAprofiles from Lake Malawi, Africa. J. Afr. Earth Sci., 8, – Implications for a Neogene Through Pleistocene Climatic383–392. Reconstruction. MSc thesis, University of Arizona.

K, A. B., V, P.-J., C, J.-P., L, R. M, P. J. & S, E. (1979) Productivity, sedimentationT., K, M. & R, B. T. (1983) Le volcanisme rate, and sedimentary organic matter in the oceans, Part I,et l’evolution structurale du Sud-Kivu (Zaire). Un modele Organic carbon preservation. Deep Sea Res., 26, 1347–1362.d’interpretation geodynamique du volcanisme distensif intra- O, P. E. (1986) A 40 million year lake record of Earlycontinental. In: Ancient Rifts and Troughs (Ed. by M. Popo� Mesozoic climatic forcing. Science, 234, 789–912.and J.-J. Tiercelin), Bull. Centres Rech. Explor.-Prod. Elf O, P. E. (1990) Tectonic, climatic and biotic modulationAquitaine, 7, 257–271. of lacustrine ecosystems-Examples from Newark Supergroup

K, D. E. (1991) A chronology of Pliocene sea-level of Eastern North America. In: Lacustrine Basin Exploration.fluctuations, U. S. Atlantic Coastal Plain. Quat. Sci. Rev., Case Studies and Modern Analogs (Ed. by B. Katz), Mem.10, 163–174. Am. Ass. Petrol. Geol., 50, 209–224.

L, M. R. & J, J. A. (1993) The interaction between O, P. E., K, D. V., C, B., W, W. & S,normal faulting and drainage in active extensional basins, R. W. (1996) High resolution stratigraphy of the Newark riftwith examples from the western United States and central basin (early Mesozoic, eastern North America). Geol. Soc.Greece. Basin Res., 5, 79–102.

Am. Bull., 108, 40–77.L, S. & D, L. (1993) Upper Pliocene history of the

O, B., C, R., J, T. C., T, D.,Sahara by high resolution marine palynology: ODP Site 658.

K, I., D, S., E, D. H. & E, D.Amer. Assoc. Strat. Palynol. abst. w/prog., 26th ann. mtg.,E. (1990) Major low levels of Lake Malawi and implicationsp. 50.for speciation rates in cichlid fishes. Proc. Roy. Soc. Lond.L, A.-M. & C, J. (1991) Correlated oceanic andB, 240, 519–553.continental records demonstrate past climate and hydrology

P, D. W. & S, N. J. (1973) Trade wind andof North Africa (0–140 ka). Geology, 19, 307–310.temperature correlations down a deep-sea core o� the SaharanL, K. E., T, J. J., D B, M., C, A.,coast. Nature, 245, 455–457.B, T., V R, P., L T, C., M,

P, P. & B, A. (1989) Exces d’argon dans les basaltesW. & K, J. (1996) New seismic stratigraphy and Latede la zone volcanique d’Idjwi sud (Kivu, Zaıre). Mus. roy.Tertiary history of the North Tanganyika Basin, East AfricanAfr. centr. Tervuren (Belgium) Department of Geol. & Min.Rift system, deduced from multichannel and high-resolutionRapport Ann., 1987–1988, 101–107.reflections seismic data and piston core evidence. Basin Res.,

P, P., V, M., D P, P. & K, J.8, 1–28.(1989) Timing of the volcanism of the southern KivuL, T. (1989) Climatic change in Africa during the Lastprovince: implications for the evolution of the western branchGlacial: Facts and problems. Palaeoecol. Africa, 20, 163–179.of the East African Rift system. Earth planet. Sci. Lett.,M, L. P. & S, H. T. F. (1995) Rift border94, 353–363.system: The interplay between tectonics and sedimentation

P, M. (1990) Tempo and mode of molluscan evolutionin the Reconcavo Basin, northeastern Brazil. Am. Ass. Petrol.in the Pliocene of the Albertine Rift, Uganda-Zaire. C. R.Geol. Bull., 79, 1590–1607.Acad. Sci. Paris Serie II, 311, 1103–1109.M, R. A. & P, F. C. (1981) Roteiro Geologico da

P, M., S, B., V, A., van N, W.,Bacia do Reconcavo. Petrobras SEPES-DIVEN, Bahia, Brazil.S, I., B, Z. & M, E. (1991) NouvelleM, J. H. (1983) Cenozoic glaciation in the Southernbiostratigraphie du Neogene au Quaternaire de la region deHemisphere. Ann. Rev. Earth planet. Sci., 11, 99–132.Nkondo (Bassin du lac Albert, Rift occidental ougandais). C.M, J. H. (1984) Late Cainozoic glacial variations in SouthR. Acad. Sci. Paris Serie II, 312, 1667–1672.America south of the equator. In: Late Cainozoic Paleoclimates

P, N. & L, M. (1984) Milankovitch forcing of theof the Southern Hemisphere (Ed. by J. C. Vogel), pp. 45–58.oceanic system: evidence from the northwest Pacific.Balkema, Rotterdam.

M, A. E., C, A. S., W, K., J, M. R. & Milankovitch and Climate (Ed. by A. Berger, J. Imbrie,

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132130

Page 25: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

Palaeogeography of Lake Tanganyika

J. Hays, G. Kukla and B. Saltzman), NATO ASI Series C, S, N. C., D, P. N. & S, M. (1993) Climate-driven variations in geothermal activity in the northern126, Part 1, 307–330.

P, A. (1976) Volcanologie du Rift de l’Afrique centrale. Le Kenya rift valley. Nature, 362, 233–234.S, C. & M, A. (1993) Mitochondrial phylogenyNyamuragira dans les Virunga. Essai de magmatologie du rift.

Thesis, Faculte des Sciences d’Orsay, Paris. of the endemic mouthbrooding lineages of cichlid fishes fromLake Tanganyika, East Africa. Mol. Biol. Evol., 10, 751–768.P, W. L. & K, J. E. (1987) Monsoon variability

over the past 150,000 years. J. geophys. Res., 92, 8411–8425. S, J. P., D, A., B, E., B, A., Z, Z.,A, S., D, F., C-N, N., L,P, W. L., M, D. W., C, S. C. & A,

D. M. (1992) Evolution and variability of the Indian Ocean S., F, J. & D, D. (1993) Pliocene vegetation andclimate in the West European and Mediterranean regions.summer monsoon: evidence from the Western Arabian Sea

Drilling Program. Synthesis Of Results From Scientific Amer. Assoc. Strat. Palynol. 26th ann. mtg. abst. w/prog., p. 74.T, J. T. (1985) Glacial Lake Agassiz and its influence onDrilling In The Indian Ocean. Geophy. Monograph, 70,

447–469. the Great Lakes. In: Quaternary Evolution of the Great Lakes(Ed. by P. F. Carrow and P. E. Calkin), Geol. Ass. CanadaR, D. J. (1992) Seismic reflection character of lacustrine

facies: implications for chronostratigraphic applications. EOS, Spec. Paper, 30, 1–16.T, J.-J., C, A. S., S, M. & L, K.73, 289.

R, D. J. (1994) Sedimentary basin evolution: tectonic and E. (1994) Pleistocene–Modern deposits of the LakeTanganyika rift basin, East Africa: a modern analog forclimatic interactions. PhD dissertation, Columbia Univ, N. Y.

R, B. R., et al. (1988) Seismic Atlas of Lake lacustrine source rocks and reservoirs. In: LacustrineReservoirs and Depositional Systems (Ed. by A. J. Lomando,Tanganyika. Project PROBE Geophysical Atlas Series, Duke

University, Durham, NC, USA. C. Schreiber and P. M. Harris), SEPM Core Workshop,19, 37–59.R, B. R., R, D. J., L, P. M., B,

C. F., MG, J., S, D., L, J. J. & D, T, J.-J. & M, A. (1991) The geology of theTanganyika Trough. In: Lake Tanganyika and its Life (Ed.S. J. (1986) Structural expressions of rifting. Lessons from

Lake Tanganyika, Africa. In: Sedimentation in the African by G. W. Coulter), pp. 7–48. Oxford University Press,London.Rifts (Ed. by L. E. Frostick, R. W. Renaut, I. Reid and

J-J. Tiercelin), Spec. Publ. Geol. Soc. London, 25, 127–139. T, J.-J., M, A., G, F., H-M,

C., H, M., L, P., L, V., M, P.,S, P. (1981) Sediment accumulation rates and the com-pleteness of stratigraphic sections. J. Geol., 89, 569–584. R, C., R, J. F., T, N., V, A. &

W, D. (1988) 25000 ans d’histoire hydrologique etS, C. A. & F, B. (1994) Late Quaternary sequencestratigraphy of Lake Malawi (Nyasa), Africa. Sedimentology, sedimentaire du lac Tanganyika, Rift Est-Africain. C. R.

Acad. Sci. Paris, 307, 1375–1382.41, 163–179.S, C. A. & R, B. R. (1988) Low lake stands in T, J.-J., S, C. A., M, A., R,

B. R. & R, C. (1989) Discontinuites sismiques etLakes Malawi and Tanganyika, East Africa, delineated withmultifold seismic data. Science, 240, 1645–1648. sedimentaire dans la serie de remplissage du fosse du

Tanganyika, Rift Est-africain. C. R. Acad. Sci. Paris, 309,S, C. A., R, B. R., V, J. W., K,

K. J. & W, L. D. (1989) Seismic Atlas of Lake Malawi 1599–1606.T, J.-J., S, M. R., C, A. S., L, K.(Nyasa), East Africa. Project PROBE Geophysical Atlas

Series, Duke University, Durham, NC, USA. E. & B, J. L. (1992) Sedimentation in large riftlakes: examples from the Middle Pleistocene-ModernS, N. J. & K, J. P. (1975) Late Cenozoic

oxygen and carbon isotopic changes at DSDP site 284: deposits of the Tanganyika Trough, East African Rift System.Bull. Centres Rech. Explor. Prod. Elf Aquitaine, 16, 83–111.Implications for glacial history of the Northern Hemisphere

and Antarctica. In: Initial Reports of the Deep Sea Drilling V Z B, E. M. (1982) African palaeoclimates,18,000 years B. P. Palaeoecol. Africa, 15, 77–99.Project, 29 (Ed. by J. P. Kennett, et al.), pp. 801–807. US

Govt Printing O�ce, Washington, DC. V Z B, E. M. & M, J. H. (1986) MajorLate Cainozoic events and palaeoenvironmental changes inS, M. R. & C, A. S. (1996) Textural and composi-

tional variability across littoral segments of Lake Tanganyika: Africa viewed in a world wide context. Palaeogeog. Palaeoclim.Palaeoecol., 56, 217–235.the e�ect of asymmetric basin structure on sedimentation in

large rift lakes. Am. Ass. Petrol. Geol. Bull., 80, 382–409. V, J. & H, (), J. (1990) Stratigraphy andgeological history of the upper Semliki: a preliminary report.S, R. & S, M. (1984) Late Neogene events of

atmospheric and oceanic circulation o�shore Northwest Virginia Museums of Natural History Memoir, 1, 17–39.V, J. W. (1988) Relationships Between PrecambrianAfrica: high resolution record from deep-sea sediments.

Palaeoecol. Africa, 16, 9–36. Basement Structures and the Cenozoic Rift Architecture of theMalawi (Nyasa) Rift Zone, East Africa, MSc thesis, DukeS, F. A. & G, A. T. (1979) Global maps of lake-level

fluctuations since 30 000 BP. Quat. Res., 12, 83–118. University, Durham, NC, USA.V, A. (1989) Paleoenvironnements du bassin Nord-S-P, F. A. & H, S. P. (1985) Lake levels

and climate reconstruction. In: Paleoclimate Analysis and Tanganyika (Zaıre, Burundi, Tanzanie) au cours des 13derniers mille ans: apport de la palynologie. Rev. Palaeobot.Modelling (Ed. by A. D. Hecht), pp. 291–340. Wiley &

Sons, NY. Palynol., 61, 69–88.V, A. (1993) Nouvelle sequence pollinique du lacS-P, F. A., M, D. S., R, N. &

H, S. P. (1989) Global lake-level variations from, Tanganyika, 30,000 ans d’histoire botanique et climatique duBassin Nord. Rev. Palaeobot. Palynol., 78, 381–394.18,000–0 years ago: a palaeoclimatic analysis. United States

Department of Energy, Tech. Rept., TR 046. V, A., C, F., B, R., G, J. &

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132 131

Page 26: Basin Research New palaeogeographic and lake-level ...€¦ · Palaeogeographic and lake-level reconstructions have palaeogeographic synthesis of Northern Lake Tanganyika ... In this

A. S. Cohen et al.

T, J.-J (1993) Pollen-derived rainfall and temperature W, S. K. (1993) Tephrochronology and Basinal Correlationof Ash Deposits in the Bonneville Basin, Northwest Utah. MScestimates from Lake Tanganyika and their implications for

Late Pleistocene water levels. Quat. Res., 40, 343–350. thesis, University of Utah.W, P.-N. & H, D. M. (1991) Late Cenozoic glacial

history of the Ross embayment, Antarctica. Quat. Sci. Rev., Manuscript received 18 December 1995; revision accepted4 September 1996.10, 215–223.

© 1997 Blackwell Science Ltd, Basin Research, 9, 107–132132