Resume-Nous presentons 30 ages de traces de fission sur apatite et 22 mesuresde longueur de trace de fission pour des echantillons du socle dans les flancs desrifts de Malawi et Rukwa (rift est-africain), dans Ie but de reconstituer I'histoirethermo-tectonique des flancs de ces rifts. Les relativement courtes longueursmoyennes (11.0-13.2 Jim) et larges distributions (1.3-2.3 Jim) des traces de fissionsuggerent un refroidissement prolonqe pour la region, allant du Permien (Karoo) a laperiode recente. La reconstruction de I'histoire thermique par modelisation inversede la distribution des longueurs de trace de fission suqqere que des phases repeteesde refroidissement rapide et de denudation des flancs de rift se sont succedees a250-200 Ma, -150 Ma et s40-50 Ma. II apparait qu'elles sont reliees a differentes
5Present address: Laboratoire de Geodvnarnique des Chaines Alpines, InstitutDolomieu, 15 rue Maurice Gignaux, 38031 Grenoble cedex, France
Journal of African Earth Sciences 363
P. VAN DER SEEK et al.
phases de rittoqenese connues dans la region, et qu' elles peuvent etre correleesavec Ie depot de ditferentes unites sedimentaires dans les bassins. L' erosion et Iereajusternent isostatique qui sen suivit ont modifie la topographie induite par latectonique autour du rift: l'elevation du compartiment inferieur (footwall flank) est
(Received 25 November 1996: revised version received 28 August 1997)
INTRODUCTIONThe East African Rift System has, throughoutthis century, been considered as a classical areafor the study of continental extension. TheWestern Branch of the East African Rift System,or Western Rift, has been the topic of numerousstudies over the last decade alone, whichprovided many details about the stratigraphy,kinematics and chronology of rifting (e.g.Rosendahl, 1987; Tiercelin et et., 1988;Chorowicz, 1990; Ebinger et al., 1987, 1989,1993; Delvaux et al., 1992; Ring et al., 1992;Rosendahl et et., 1992; Wheeler and Karson,1989). In contrast, the uplift history of the riftflanks has so far been less constrained, mainlybecause of the lack of stratigraphical markers;the flanks expose Precambrian basement rocks.The uplift and denudation history of the regionhas traditionally been studied in a framework ofcyclical geomorphic evolution, with muchemphasis being put on the recognition of erosionsurfaces of regional to continental extent (e.g.Dixey, 1937, 1947; King, 1963). Although theimportance of denudation in creating the presentday morphology of the East African Rift flankswas recognised by early workers (e.g. Dixey,1947), the elevation of the flanks has, in general,been directly tied to tectonic uplift during LateCenozoic rifting (e.g. Ebingeretal., 1991, 1993).
During the last decade, apatite fission track(AFT) thermochronology has emerged as apowerful technique to unravel the denudationhistory of regionally elevated basement regions,because of its ability to constrain the lowtemperature « 120°C) cooling history of rocksamples (e.g. Foster and Gleadow, 1993;Fitzgerald, 1994; Brown et el., 1994). In thisstudy, AFT analyses of 30 samples collected inrecent years from the flanks of the Malawi(Nvasa) and Rukwa Rifts are presented in order
364 Journal ofAfrican Earth Sciences
to assess the denudation history of these regions.The AFT data reveal distinct periods of cooling,coincident with extensional phases within theWestern Rift. The data also show that thetectonic topography induced by Cenozoic riftinghas been modified by erosion. Finally, this paperwill show that the summits of the LivingstoneMountains and the eastern flank of the RukwaRift define the same structural level, but theirdenudation history is inconsistent with theJurassic age that was previously assigned tothis surface.
PHYSIOGRAPHIC AND TECTONIC SETTINGOF THE WESTERN RIFT
The Western Rift is part of the East African RiftSystem (EARS), which runs from the Red Sea inthe north down to the Zambezi River mouth inMozambique to the south (Fig. 1, inset). TheWestern Rift (or Western Branch of the EARS)is segmented into a number of deep and narrowhalf-graben, often with alternating polarity(Bosworth, 1985; Rosendahl, 1987; Ebinger etel., 1987; Ebinger, 1989). The extensional basinsof the Western Rift show distinct and highlyelevated rift flanks. Each half-graben isassociated with a clearly defined depocentre anda highly uplifted footwall flank; the ramping sideof the basins show less topographic elevation.The topography generally rises to over 2 km a.s.1.on the footwall side of the basins adjacent tothe border faults, whereas the elevationdifference on the ramping sides amounts to somehundreds of metres, enhancing basin asymmetry(Ebinger, 1989).
Although the Malawi Rift lies south of theMbeya region, where the Eastern and WesternBranches of the EARS intersect (Figs 1 and 21.it is usually considered as part of the Western
Denudation history of the Malawi and Rukwa Rift flanks from apatite fission track thermochronology
o 600 1200 1800
Elevation (m)
2400 3000
Figure 1. Topographic map of the Malawi-Rukwa Rift region, constructed from the USGS gtopo-30 database(30" resolution; here reproduced at 1 'L showing the distribution of apatite fission track samples for this study.Samples are indicated by their AFT age; .: analysed by van der Beek (1995); 0: samples from Mbede et al.(1993). Inset shows location of the study area On the African continent; strings of lakes outline the position ofthe East African Rift branches: AB: Albert; KI: Kivu; TA: Tanganyika; RU: Rukwa; MA: Malawi (Western Branch);ET: Ethiopian; TU: Turkana; KE: Kenya (Eastern Branch).
Rift because of its structural similarities with theother rift basins in this branch (e.g. Rosendahl,1987; Rosendahl et al., 1992). The eastern(footwall) flank of the northernmost part of theMalawi Rift rises up to 2.0 km above the present
lake level, forming the Livingstone Mountains,whereas the elevation of the ramping side ofthe basin is approximately 1.5 km above thepresent lake level (Fig 1). The LivingstoneMountains are bounded by the Livingstone
Figure 2. Simplified geological map of the study region. A·A' is the profile across the MalawiRift depicted in Figs 9 and 10; B·B' is the profile across the Rukwa Rift shown in Fig. 11.
Mountains Border Fault (LMBF) but theLivingstone Mountains Escarpment appears tobe at least partly erosional in origin (Dixey,1947). The summit of the mountains defines anerosional plateau, thought to be a remnant ofthe Jurassic "Gondwana surface" (King, 1963;Delvaux and Wopfner, 1992). The morphologyof the Rukwa Rift provides an exception to thegeneral pattern; here the flank adjacent to thelupa Border Fault is elevated by less than 400m with respect to lake level, whereas theopposite flank (the Ufipa Plateau) rises more than1000 m above the surrounding topography(Mbede, 1993).
The oldest rock units flanking the Western Riftwithin the Rukwa-Malawi zone (Fig. 2) are thebasement rocks of the Ubendian system,composing granitic gneisses, paragneisses,migmatites, metadolerites and amphibolites
(Harkin, 1955). The age of metamorphism of theUbendian rocks is placed between 2100-1700Ma (Lenoir et al., 1994); the last Precambrianthermal activity recorded in the region is the olderPan-African tectonothermal event, at ca 750 Ma(Theunissen et al., 1992; U-Pb on zircon).
Three Phanerozoic episodes of extension andbasin filling are recognised in the Western Rift:the Karoo rifting event (Permo-Triassic), a LateMesozoic-Cenozoic phase, and Late CenozoicRecent rifting. During these periods, rifting wasconcentrated in coinciding zones, which werestrongly controlled by the pre-existing basementstructures (Daly et al., 1989; Versfelt andRosendahl, 1989; Theunissen et al., 1996). Theborder faults of the Rukwa and Malawi Rifts aresuperimposed on major Pan-African shear zoneswhich separate the Tanzanian Craton from theZambian and central African Cratons.
366 Journal of African Earth Sciences
Denudation history of the Malawi and Rukwa Rift flanks from apatite fission track thermochronology
60 ~.2(tJ>-<(])
90 ~(])......o
20 %(;Jc,
10050 75FT age (Ma)
Palaeotemperature CC)120 90 6030
4 • FTage-O---rnean length
Mean length (urn)8 12 16
• 0'\"-0.\'\
Mean length (urn) ~"\
8 12 16 ~~,,-v.
• FTage-O---rnean length
1 30 9'!2
(])
60>-<;::1
..c:: ........... (tJ
0.. >-<(])
(]) PAZ 0..°3 90 5
(])
f-<
4 a 120
25 50 75 100FT age (Ma)
c
25 50 75 100FT age (Ma)
Figure 3. (a) Variation of fission track age and mean track length with temperature, calculated using the annealing model ofLaslett et al. (1987), for a 100 Ma stable thermal regime. A depth scale is added assuming a 30°C km:' geotherm and OOCsurface temperature. Both fission track age and mean track length decrease rapidly between -60-120°C in the partialannealing zone (PAZI. (b) Pattern of fission track ages and mean track lengths with elevation after 5 km of uplift anddenudation from 25 Ma onward. The base of the fossil PAZ is characterised by a break in slope in the age-elevation plot andrecords the amount of exhumation as well as the timing of its onset. (c) Resulting age versus length plot; shaded arearepresents mean length ±standard deviation for samples from different palaeotemperatures (scale on top). Samples from thebase of the fossil PAZ produce the low-age peak in the diagram and date the onset of exhumation. Insets show modelledfission track length distributions for rocks originating (from left to right) from below, within and above the PAZ. Modified fromvan der Beek et al. (1996).
Up to 3 km of Karoo sediments were depositedin half-graben bounded by the same faults asthose which were active during Late Cenozoicrifting (Morley et et., 1992; Kilembe andRosendahl, 1992; Mbede, 1993). Karoo riftingalso took place along northeast-southwestorientated zones such as the Ruhuhu, Luangwaand Usangu Basins (Wopfner, 1990; Fig. 2), butonly some of the northeast trending Karoo basinshave been moderately reactivated during thePleistocene (Delvaux et al., 1992).Unconformably overlying the Karoo Supergroupis the Red Sandstone Group (or "Dinosaur Beds";
Dixey, 1928), deposited during a phase ofextension that took place between the Karooand Late Cenozoic events. The age of this eventremains controversial however, with recentestimates ranging from Late Jurassic-EarlyCretaceous to Miocene (e.g. Jacobs et et., 1989;Dypvik et et., 1990; Wescott et st., 1991;Damblon et et., 1998).
The latest stage of rifting in the Rukwa-Malawiregion commenced at around 8 Ma, as shownby 4°Arf39Ar dating of rift-related volcanics andfield relationships (Ebinger et el., 1989, 1993);the oldest rift-related deposits contain Upper
Journal of African Earth Sciences 367
P. VAN DER SEEK et al.
Miocene palynomorphs and mammalian fossils(Wescott et el., 1991). This phase is still activetoday, as indicated by historical volcanism andinstrumental seismicity (e.g. Delvaux and Hanon,1993; Jackson and Blenkinsop, 1993;Camelbeeck and Iranga, 1996). The Cenozoicextension direction of basins within the WesternRift is controversial and has been variouslyinferred as either purely extensional (northeastsouthwest; Ebinger et al., 1987; Morley et al.,1992), strike-slip (northwest-southeast; Tiercelinet al., 1988; Wheeler and Karson, 1989; 1994)or initially radial changing to pure extension(Delvaux et el., 1992; Ring et et., 1992). Thepresent-day stress regime from earthquake focalmechanisms appears to be northeast-southwest(Jackson and Blenkinsop, 1993; Camelbeeck andIranga, 1996).
APATITE FISSION TRACKTHERMOCHRONOLOGY
BackgroundOver the past decade, apatite fission track (AFT)thermochronology has become an extremelyvaluable technique to constrain the lowtemperature « 120°C) thermal histories ofexhumed basement blocks (ct. Brown et et., 1994for a review). Tracks are produced continuouslyover geological time by spontaneous fission of238U and the accumulation of fission tracks inminerals such as apatite thus provides a measureof sample age. The tracks are not stable,however, but anneal at strongly temperaturedependent rates. At temperatures between -60and 120°C the amount of annealing in apatiteincreases rapidly towards total; this temperaturerange has been termed the partial annealing zone(pAZ; Wagner, 1979). The temperaturedependence of annealing rates is nowquantitatively understood (Laslett et al., 1987;Green et al., 1989b) and, under tectonicallystable conditions, AFT ages and mean tracklengths decrease predictably with increasingtemperature and hence depth (Fig. 3a). Themain control on the decrease of AFT ages andtrack lengths with depth is exerted by thegeothermal gradient; secondary controls areexerted by the duration of tectonically stableconditions and chemical composition, with Frich apatites annealing more quickly than CIrich apatites (Green et al., 1989a; Crowley etal., 1991).
When a tectonic block cools as a result ofdenudation, remnants of the characteristicAFT age-depth pattern may be retained,
368 Journal of African Earth Sciences
producing a trend of increasing AFT ages withelevation (Fig. 3b). The base of the exhumed(or "fossil") PAZ, when exposed, will producea characteristic break in slope in the ageelevation plot, the age of which approximatesto the initiation of cooling and denudation(Gleadow and Fitzgerald, 1987). Samples frombelow the break in slope were exhumed fromtemperatures ~120°C and contain only tracksformed during and after cooling. In contrast,samples above the break in slope contain twogenerations of tracks, one from before andone from after the onset of cooling. The actualshape of the age-elevation plot, and the trendof track length distributions with elevation,will depend on the amount and rate ofexhumation (Brown et el., 1994).
The dependence of track length distributions(Gleadow et al., 1986) on thermal history canalso be used to constrain the amount and timingof rock cooling. A plot of fission track ageagainst mean track length of an exhumed terrainwill show a characteristic 'boomerang' shape(Fig. 3c), the length of the peak at young agescorresponding to samples which were exhumedfrom the base of the PAZ and thus retains onlytracks formed after the onset of cooling (e.g.Omar et et., 1989). These samples exhibitnarrow, negatively skewed track lengthdistributions with relatively long mean lengths.In contrast, samples which resided within thePAZ before the onset of denudation will typicallyshow wide, bi-modal track length distributions(Fig. 3c). Because the kinetics of annealing inapatite are now quantitatively understood, theobserved track length distribution can be usedto reconstruct the cooling trajectory (T-t path)below 120°C of a sample through inversemodelling (e.g. Lutz and Omar, 1991;Gallagher, 1995).
It should be noted that fission trackthermochronology strictly records the cooling ofa sample only. Where fission track data showthe characteristic relationship to elevationdescribed above, or when additional geologicaldata are available (e.g. the correspondence ofcooling intervals with periods of sedimentdeposition in nearby sedimentary basins or thepresence of erosion surfaces of correspondingage), cooling can be interpreted to result fromdenudation. The amount of denudation can bequantified if the geothermal gradient is known.Inferring uplift from AFT data involves anotherinterpretative step and can only be done if theinitial elevation of the region and the degree ofisostatic compensation of denudation are known
Denudation history of the Malawi and Rukwa Rift flanks from apatite fission track thermochronology
or inferred (e.g. Brown et a/., 1994; van derBeek et a/., 1994).
Previous fission track studies in East AfricaPrevious AFT studies around the EARS includereconnaissance studies by van den Haute (1984)in Rwanda and Burundi and by Wagner et al.(1992) in Kenya, a detailed study of fourmountain ranges surrounding the central KenyaRift (Foster and Gleadow, 1992, 1996) and aregional study in Tanzania (Noble, 1997; Nobleet sl., 1997). Both van den Haute (1984) andWagner et at. (1992) found that their AFT dataindicated slow continuous cooling of basementrocks throughout the last 300-400 Ma. Onlysamples from very close to the Tanganyika-Kivuand Kenya Rifts showed AFT ages and tracklength distributions indicative of Tertiary cooling,either through moderate denudation (van denHaute, 1984) or after reheating (Wagner et et.,1992) of the flanks.
Foster and Gleadow (1992, 1996) constructeda composite age-elevation profile from fourregions surrounding the central Kenya Rift andwere able to distinguish three episodes of rapiddenudation during Meso-Cenozoic times: at~220, 140-120, and 60-70 Ma. Of these, theEarly Tertiary (60-70 Ma) event appears themost important, with up to 2.5 km of denudationrecorded; the other two events are associatedwith -0.5 km of denudation each. Inversemodelling of track length distributions from theeastern flank of the Kenya Rift indicated -40°Cof Miocene-Recent cooling, probably associatedwith Late Cenozoic rifting. Foster and Gleadow(1993) related the episodes of rapid denudationto intracontinental tectonic phases, correlatedto changes in plate motion, which gave rise toblock faulting and local uplift. They also showedthat the denudation history inferred from theirAFT data is inconsistent with classicalcorrelations of regional erosion surfaces,because samples from inferred Mesozoic erosionsurfaces remained at depths ~2 km until after60 Ma.
In Tanzania, Noble et el. (1997) found thatmost of their samples record protracted slowcooling histories throughout the Meso-Cenozoic,interrupted by episodes of faster cooling ataround 110 Ma (in eastern Tanzania only) and-65 Ma (throughout eastern and southernTanzania). They interpret the significance ofthese accelerated cooling events, similar toFoster and Gleadow (1993), as caused by blockfaulting during periods of intracontinentaltectonics.
APATITE FISSION TRACK DATA FROM THEMALAWI-RUKWA RIFTS
Sampling and proceduresSamples for AFT analysis were collected fromboth flanks of the Malawi and Rukwa Rifts(Fig. 1). All samples were collected fromUbendian basement rocks; sampled lithologiesinclude granites, ortho- and para-gneisses,amphibolites and anorthosites. In theLivingstone Mountains, a profile (eightsamples) spanning an elevation from lake level(400 m) to the summit plateau at 2200 m wascollected, as well as a transect (four additionalsamples) from the top of the escarpment upto 70 km north-northeast of it. In the Kyelaregion, west of the Malawi Rift, six sampleswere collected from areas within the rift andup to 50 km west of it, spanning an elevationrange from 600 to 2020 m. Two samples werecollected from basement underlying Karoosediments in the Songwe district, in thenorthwestern sector of the Malawi Rift. Threesamples were collected from basement inliersin the Mbeya region, in the intersectionbetween the Malawi, Rukwa and UsanguBasins. AFT data are also reported for ninesamples from the Rukwa Rift flanks, whichwere collected by Mbede et al. (1993) fromelevations between 910 and 2185 m, closeto the eastern and western border faults ofthe rift. The latter samples were processedby the London Fission Track Research Group(Mbede et al., 1993), while those from theMalawi Rift were processed at the VrijeUniversiteit, Amsterdam (van der 8eek, 1995).Sample preparation and processing techniquesfollowed the recommendations of Hurford(1990); details can be found in the reportsmentioned above.
ResultsAFT results are presented in Table 1; theirgeographical distribution is shown in Fig. 1. Allages are reponed as central ages (Galbraith andLaslett, 1993), with a ± 10 error. AFT agesrange from 30 ± 15 to 296 ± 10 Ma, spanning arange in age from the Karoo to Late Cenozoicrifting events. All AFT ages are much youngerthan the latest Pan-African tectonothermalevent. They are interpreted as representing thelong-term cooling history of basement rocksresulting from regional denudation. The sampleshave relatively short mean track lengths (MTL),within the range of 11.4 to 13.2 ut«, and broadtrack length distributions (FTLD) with 0= 1.32.0 jJm (Fig. 4).
Journal of African Earll) Sciences 369
Table 1. Fission track analytical data
'" Sample Elevation Number Ps (Ns) Pi (Ni) Pd (Nd) Pill age ± 1c D MTL std. No. of"0~ (m) of grains (x 10 6 cm 2
Notations: p.: spontaneous track density; p.: induced track density (includes 0.5 geometry factor); P,: density of tracks in the glass dosimeter; N., N" N,: number of tracks actually counted todetermine the reported track densities. All ages are reported as central ages (Galbraith and Laslett, 1993). Employed ~-factors for the samples processed at the VU Amsterdam (all samples labelledDD and T/92) are 113.6 ± 3.8 for glass dosimeter CN2 (p, = 2.551 x 10' em") and 11493 ± 522 for NBS963 (p, = 0.251 x 10' em'); the ~-factor for the samples processed by the London FissionTrack Research Group (samples labelled T/89 and EAR797) is 347 ± 5 Iglass dosimeter CN5). Pix'): Chi-squared probability that the single grain ages represent one population; D: age dispersion. IfPiX') < 5 and/or D > 15 the single grain ages represent a multiple age population (ct. Galbraith, 1990; Galbraith and l.aslett. 1993).
Denudation history of the Malawi and Rukwa Rift flanks from apatite fission track thermochronology
00115a 400
80 80 160 240 320
00480age (Ma)
120 8f
2 50 ±6Ma
• 60 0= 0.11
• •• •0:~.
• 40V
10
~20 80 160 240 320
age (Ma)240
200
o!----"'----;;;w----=--
4 8 12 16length (11m)
1l.011m(J = 1.811mN =24
n ~ hl
20
12.511m
30 (J = 1.511mN= 160
20
10
~ rh4 8 12 16
length (11m)
12.211m30 (J = 1.511m
N=64
20
10 J n4 8 12 16
length (11m)
30
160 240 320age (Ma)
217±15Ma0= 0.05
178 ± 14 Ma0= 0.09
160
240
200
240
160
•
320
..
•
. ..
80
.•••
00512
00122
2
2
-2
-2
-2
4 8 12 16length (11m)
160 240 320age (Ma)
8080
32f1
• • 135 ±lOMa 12.411m
• 160 0= 0.03 30 (J = 1.511mN = 100
• 201,. ••
• • 120... . • 10
00510
O~----=--::-c::"--------i
2
-2
4 8 12 16length (11m)
12.011m(J = 1.811mN=33
10
20
30
80 160 240 320age (Ma)
188 ± 29Ma0=0.09
200
240
- 160
642
•:.. .•• •
o
2
01!---..:-T-=---------1
-2
Figure 4. AFT ages and track length distributions for representative samples. Single-grain age distributions areplotted in radial plots (left) and as age histograms and probability curves (centre). Radial plots show the precisionof individual track counts on the x-axis and their deviation with respect to the central age on the y-axis (ct.Galbraith, 1990). Histograms of confined track length distributions are shown on the right. Sample DD 115a isfrom the Livingstone Mountains summit; DD122 from the Livingstone Mountains Escarpment at an elevation of-1500 m; DD480 is from the foot of the Livingstone Mountains Escarpment; DD512 is from the Kyela region justwest of the rift; DD510 is from -50 km west of the Malawi Rift.
Journal of African Earth Sciences 371
P. VAN DER SEEK et al.
Livingstone MountainsSamples from the Livingstone Mountains showa clear variation of AFT age and MTL withelevation (Fig. 5a). Sample ages from the summitplateau at elevations ~1890 m (00013, 486,115a, 478) are concordant between 214 ± 22and 246 ± 31 Ma. These samples also have verysimilar track length distributions, with MTL =12.3-12.6jim and a=1.3-1.6jim. One samplefrom the summit plateau to the south of theprofile (Kalunduru granite), dated by the LondonFission Track Research Group (Mbede et al.,1993) has a slightly older but overlapping ageof 266 ± 27 Ma. Below 1850 rn, AFT agesincrease with increasing elevation, from 50 ± 6Ma at 675 m (00480) to 192 ± 15 Ma at 1840m (00298). MTL for these samples are short,ranging from 11.0 to 12.2 jim, and FTLO arewide, with a = 1.5-2.0 jim. There are two outlierswithin this general pattern: sample 00483records an anomalously low AFT age of 30 ± 15Ma, with an extreme age dispersion (145%);most of the grains in this sample have zero ages,the rest record a spread in ages between 65and 180 Ma (Fig. 6). Sample 00485 from thefoot of the scarp has a much older age (128 ± 20Ma) than the two samples immediately aboveit, but does fall on a linear age-elevation trendline (-0.02 km Ma') with the higher (> 1 kma.s.l.) samples. Of the two intermediatesamples, 00480 has a well defined age of 50 ± 6Ma (0 = 11 %), whereas the age dispersion of00484 (0 = 25 %) indicates multiple agecomponents; most of the grains in this samplehave ages between 40-80 Ma. but some aresignificantly older.
Western flank of the Malawi Rift (Kyela region)The western flank of the Malawi Rift istopographically less prominent compared to thesharp scarp of the Livingstone Mountains.Samples from the Kyela region show an unusualinverse age-elevation relationship (Fig. 5b). Thedata do show a clear geographical spreadhowever (Fig. 1). Samples 00329a and 00514afrom basement outcrops in and around theSongwe coal mine, still within the Karoo Basin,have AFT ages of 186 ± 33 and 214 ± 37 Ma,respectively. These ages are comparable withthose of samples from the summit of theLivingstone Mountains. Both samples from theSongwe mine contained insufficient tracks forFTLO measurements. Immediately to the westof the rift, a group of three samples (00508,509,512) from elevations between 1 and 2 kmhave younger ages; between 100 ± 12 and
372 Journal ofAfrican Earth Sciences
137 ± 13 Ma, with MTL between 11.4 and 12.4jim and o > 1.5-1.6 jim. The westernmostsample (00510) again has an age of 188 ± 29Ma, with MTL = 12.0 jim and a = 1.8 jim. Thissample resembles those from the Songwe coalmine and from the Livingstone Mountainssummit.
Mbeya regionThe three samples from the northeasternmostpart of the Malawi Rift all have very young AFTages. Two samples (T92/22a and b) were takenfrom the northern extension of the LivingstoneMountains, close to the LMBF. T92/22a has anage of 33 ± 8 Ma; this represents a single agepopulation (age dispersion 0 = 3%); T92/22b,on the other hand, has a large age dispersion(0 = 70%) with a central age of 66 ± 16 Ma andsingle grain ages between 6 and 150 Ma (Fig.61. The third sample from this region (00503)was collected from a small basement inlieradjacent to the Mbaka Fault (cf. Oelvaux andHanon, 1993; Ebinger et al., 1993); this samplehas an age of 35 ± 6 Ma with a dispersion of0=36%, also suggesting multiple agepopulations. None of these samples hadsufficient tracks for track length measurements.All samples from the Mbeya region werecollected close to the Rungwe volcanic centre,one of the main centres of recent volcanicactivity in the EARS. It is likely that the volcanicssignificantly perturbed the geothermal gradientin this area, leading to recent reheating of thesamples.
Rukwa RiftFive samples (T89/25a, band T89/26a, b, c)from the eastern flank of the Rukwa Rift (theLupa Plateau) have ages between 202 ± 16 and264 ± 26 Ma, with MTL between 11.7 and 12.4jim and a = 1.7-2.3 jim. These ages and FTLOare comparable to those from the LivingstoneMountains summit plateau. In contrast, two ofthe samples from the western flank of the RukwaRift (the Ufipa Plateau) have the oldest ages inthis study (271 ± 21 - 296 ± 10 Mal, whereasthe other two have ages that are comparable tosamples from the Kyela district: 148 ± 11 and172 ± 12 Ma. The latter two samples have thehighest MTL (13.0-13.2 jim) of all samples inthis study, with a relatively narrow FTLO (a =1.3 jim). The nine samples from the Rukwa Riftflanks do not show an obvious age-elevationtrend (Mbede et al., 1993; Fig. 5cl, probablybecause all samples from the Lupa Plateau weretaken at comparable elevations (-1 km a.s.I.).
Denudation history of the Malawi and Rukwa Rift flanks from apatite fission track thermochronology
6 mean length (urn) mean length (urn) mean length (urn)10 11 12 13 14 15 10 11 12 13 14 15 10 11 12 13 14 15
2500 2500 2500a) Livingstone Mts. b) Kyela region c) Rukwa rift
Figure 5. Age-elevation and mean track length-elevation profiles for fission track samples from fa) the Livingstone Mountains tothe east of the Malawi Rift; Ib! the Kyela profile on the western flank of the Malawi Rift; and tel the flanks of the Rukwa Rift.
DD483 100 80 60/ / •
2 40 30 ± 15 Ma• D = 1.45•
0 . 10M"...-2 •
• 0 n40 80 120 160
T/92/22aage (Ma)
2 33± 8Ma40 D=0.03
0 I.•
-2/" ~
20 40 80 120 160
T/92/22bage (Ma)
100
2 80 64± 16Ma
• D= 0.70.-• · 64Ma0 - 60
-2••
-,
20 40 40 80 120 1601 age (Ma)
0 2 4 6 (J
Figure 6. Single grain age distributions for the youngest samples in this study,plottedin radial plots (left) and as age histograms andprobability curves fright).
Journal of African Earth Sciences 373
P. VAN DER SEEK et al.
14
S 13 - +-+ .1J d-~..c Lto r+- I ::
.~JTc
+-+-~
~~12 - T 0+--u
ell I.... +....c tellQ)
~ 11-
10 I I
50 100 150 200 250 300
FT age (Ma)
Figure 7. Plot of mean track length against AFT age for samples from this study and fromMbede et al. (1993).
Interpretation: thermotectonic evolution of theMalawi-Rukwa regionThe relatively short mean track lengths (MTL)and broad track length distributions (FTLD) ofmost samples suggest a protracted coolinghistory spanning the entire Mesozoic andCenozoic eras (e.g. Gleadow et sl., 1986; Greenet et., 1989a). The trend of the age-elevationprofile from the Livingstone Mountains suggestslong-term denudation rates from this region inthe order of 20 m Mao', assuming that thethermal structure of the Livingstone Mountainsblock has remained stable throughout this time.However, the track length data indicate that suchan interpretation is overly simple.
A plot of AFT age versus MTL (Fig. 7) suggeststhat the long-term thermotectonic history of theregion was punctuated by several phases ofaccelerated cooling. Samples from the summitsof the Livingstone Mountains and the LupaPlateau plot to the right in this diagram, beingcharacterised by ages between -200-300 Maand MTL between -12.2-12.8 pi». FTLD of these
samples are narrower than of most other samplesin this study (a",1 .3-1 .8 pm) and are generallyuni-modal. These samples record cooling throughthe PAZ (ie. from> 120°C to -60°C) at around200-250 Ma.
A peak in MTL (12.4-13.2pm) is apparent inFig. 7 around 140-170 Ma. Samples with thesecharacteristics derive from the western flanksof the Rukwa and Malawi Rifts. These sampleshave narrow (a = 1 .3-1.5 pm) and uni-modalFTLD, suggesting that they cooled relatively
rapidly through the PAZ at around 150 ± 20 Ma.Finally, samples from the Livingstone
Mountains Escarpment form a band ofdecreasing MTL with decreasing age. Theyoungest samples in this suite also have thelowest MTL « 12.0 pm) with wide (a= 1.5-2.0pm) negatively skewed or bi-modal FTLD. Thesesamples remained at temperatures ~120oC untilLate Mesozoic-Early Cenozoic times (60-70 Ma).The abundance of short tracks in these samplesindicate that final cooling to surfacetemperatures took place relatively recently.
Figure 8. Modelled thermal histories for representative samples from the Malawi Rift flanks. Thermal histories within theshaded bands fit the observed AFT ages within 10' error and pass the Kolmogorov-Smirnov test for track length distributionat the 95% confidence level (ef. van der Beek, 1995). Insets show AFT ages and FTLO used as input for the modelling. (a)Livingstone Mountains summit (combined inversion of samples 00013, 00486 and DO 115a, 56 single grain ages and 360track length measurements in total). The light shaded band labelled MOurango" is for the inversion adopting the Leslett et al.(1987) parameterisation for Durango apatite; the dark band labelled MF-apatite" adopts the Crowley et al. (1991) parameterisationfor F apatites (see text for discussion). (b) Livingstone Mountains Escarpment, > 1 km e.s.l. (samples DO 122 and 00482,total of 35 single grain ages and 92 track length measurements). (c) Base of the Livingstone Mountains Escarpment (samples00484 and 00480, total of 31 single grain ages and 86 track length measurements). (d) Kyela region (western flank of theMalawi Rift; samples 00508 and 00512, total of 40 single grain ages and 160 track length measurements).
374 Journal ofAfrican Earth Sciences
a) Livingstone Mts. Summit b) Livingstone Mts. Escarpment (> 1 km)
lj I 200 150 100 50 200 150 100 50atime (Ma) time (Ma)<oJ
'J
'"
P. VAN DER SEEK et al.
In order to assess the cooling histories of thedifferent subregions within the study area morequantitatively, the thermal histories have beenreconstructed from the track length distributionsof a number of representative samples, usingthe inverse modelling approach of van der Seek(1995). The inversion is based on the notionthat each track, depending upon its formationduring the cooling history of the sample, willrecord a specific part of this history through areduction in its length (e.g. Green et ei., 1989b).Thus, a given track length distribution can berelated quantitatively to a thermal history, usinga mathematical description of the annealingprocess (Lutz and Omar, 1991 ; Gallagher, 1995).The thermal histories obtained from the inversemodelling of track length distributions are shownin Fig. 8.
Livingstone MountainsThe fact that all samples from the summitplateau of the Livingstone Mountains haveconcordant AFT ages and similar FTLO isconsistent with the interpretation of thisplateau as a continuous erosion surface (e.g.Oelvaux and Wopfner, 1992). AFT ages andFTLO from the Livingstone Mountains summitplateau are also concordant with those fromthe Lupa Plateau east of the Rukwa Rift,supporting the interpretation of this surfaceas a continuous planation level covering a largeregion. The elevation of the Lupa Plateau is-1 km lower than that of the LivingstoneMountains summit, suggesting that differentialvertical motions have affected the surfaceafter its exhumation.
Figure 8a shows the inferred cooling historyfor the summit of the Livingstone Mountains,calculated using the combined age and tracklength data from samples 00013, 00486 and00115a. The modelling suggests a two-stagecooling history, with cooling from 110-120°Cto -60°C taking place between -250-200 Ma,followed by a period of relative stability between-200 and -40 Ma. Final cooling to surfacetemperatures took place from -40 Ma onwards.This cooling history is based on the annealingparameterisation of Laslett et al. (1987) for theOurango apatite, with [Cn/[CI + F[~0.2, whereasmost samples from Ubendian basement rocksappear to be nearly pure F apatites (Noble,1997). In order to test the sensitivity of theinferred cooling history on the chemicalcomposition of the apatites, the inversion wasredone using the parameterisation of Crowleyet al. (1991) for F apatite. This model tends to
376 Journal ofAfrican Earth Sciences
predict faster annealing rates compared to theLaslett et al. (1987) model at low (::S:60°C)temperatures, but slower annealing at high(2:90 0C) temperatures (Corrigan, 1993; van derSeek, 1995). As a result, the two episodes ofrapid cooling are less distinct for this model,being smeared out between -250-150 Ma and-50-0 Ma, respectively, but they are still clearlydistinguishable (Fig. 8a).
The cooling episode between -200-250 Marecorded by these samples can be correlated toa widely recognised Late Triassic to EarlyJurassic erosion event, which marked thetermination of Karoo activity in eastern andsouthern Africa (Wopfner, 1990, 1993). TheAFT data from the summit plateau do not,however, support the inferred Jurassic age ofthis surface (correlated to the "Gondwanasurface" by King, 1963); the scarcity of long(> 13.5 Jim) tracks in the samples require thatthey did not cool to surface temperatures duringthe late Karoo erosional event but only duringthe Tertiary. A late cooling stage is often inferredfrom inverse calculations of thermal histories andhas been attributed to model artefacts (e.g.Corrigan, 1993). However, because bothannealing models employed require late cooling,this study indicates that the late cooling event isreal. It is therefore suggested that the maximumexposure age of the Livingstone Mountains summitplateau is Palaeogene (-40 Ma).
The general increase in AFT age with elevationat the Livingstone Mountains Escarpment isconsistent with samples from lower elevationshaving been exhumed from progressively greaterdepths. There are some irregularities in thepattern however, with sample 00483 havingan anomalously young AFT age (30 ± 15 Maland 00485a being anomalously old (128 ± 20Ma). Noble (1997) analysed a suite of samplesfrom an elevation profile spanning from -500 to800 m in the Matema region at the northeastcorner of the lake, approximately 100 kmnorthwest from the profile studied here. He foundsimilar irregularities, with AFT ages rangingbetween 20 and 100 Ma and large agedispersions in the majority of his samples. Thereare three possible explanations for the irregularAFT age pattern at elevations < 1400 m a.s.1.Firstly, the samples may have varying CIcontents, which will affect their annealingcharacteristics (e.g. Green et st., 1989a; Crowleyet al., 1991). At this stage there are no chemicalanalyses of the samples from this study toinvestigate this effect. However, inspection ofetchpit diameters suggests roughly similar [F11
Denudation history of the Malawi and Rukwa Rift flanks from apatite fission track thermochronology
{CI] ratios for the samples 00480 to 00485a.Samples analysed by Foster and Gleadow (1996)and Noble (1997) from high-grade basementrocks in East Africa also showed little variationin apatite composition, with most grains beingnearly pure F apatites. Alternative explanationsfor the AFT age distribution become apparent ifit is considered that the base of the LivingstoneMountains Escarpment coincides with the LMBF,a long-lived crustal-scale shear zone (Wheelerand Karson, 1989; Theunissen et el., 1996).Differential movements of discrete blocks withinthis fault zone would lead to disruption of thepattern of monotonously increasing AFT ageswith elevation (e.g. Fitzgerald, 1994; Foster andGleadow, 1992, 1996). Samples with youngAFT ages would thus define blocks exhumedfrom structurally deeper levels within the faultzone. Finally, the LMBF has provided a pathwayfor pervasive fluid flow throughout its history,as indicated by the deposition of a large varietyof hydrothermal minerals within the fault zone(e.g. Wheeler and Karson, 1989). Fluid flowactivity would have led to leaching, dissolutionand re-precipitation of apatite within the LMBF(Noble, 1997). Van der Beek et al. (1996)suggested a similar mechanism to account forspurious AFT ages from samples within thePrymorsky Border Fault Zone of the Baikal Rift,a structure very similar to the LMBF. Theabundance of zero grain ages in sample 00483suggests that fluid flow through the fault zoneis still active today, as is also indicated by activehydrothermal centres and Quaternary travertinedeposits in the Mbeya region, both clearlyassociated with active faults (Harkin, 1960;Oelvaux and Hanon, 1993).
Thermal histories from inverse calculations forsamples from the Livingstone MountainsEscarpment are depicted in Fig. 8b (for samples00122 and 00482, between 1-1.5 km a.s.L)and 8c (for samples 00484 and 00480, at < 1km a.s.l.I. The modelling suggests that thelowermost samples remained at temperatures~1 20°C until -60-70 Ma (Late CretaceousPalaeogene) and only cooled from -80°C downto surface temperatures in the last -20 Ma.These samples suggest that a significant amountof denudation of the Livingstone MountainsEscarpment has taken place during the Cenozoic,more than half of which took place during LateCenozoic rifting.
Kyela regionSamples from the basement in and around theSongwe coal mine have ages similar to those
from the summit plateau of the LivingstoneMountains. Although there is no supporting tracklength data, it is suggested that these sampleshave experienced a similar cooling history. Karoodeposition took place between LateCarboniferous and Triassic times in Tanzania(e.g. Wopfner, 1990); in the Songwe region onlythe lower part of the Karoo sediments have beenpreserved, the youngest unit having a probablemid-Permian age (Semkiwa, 1992). The AFTdata from this study indicate that the samplesreached maximum temperatures of ~1200C
before -240 Ma, i.e. shortly after deposition.Vitrinite reflectance (Ro= 0.5-1.0; Kreuser et al.,1988) and Rock-Eval (T
m ax= 400-450 0C; Oypvik
et al., 1990) data for Karoo coals from southernTanzania are consistent with this palaeotemperature estimate. A minimum overburdenequivalent to a -50-60°C temperaturedifference (the PAZ) was removed during LateTriassic-Early Jurassic denudation. For ageothermal gradient of 25-30 oC krn' withinthe Karoo basins (determined from vitrinitereflectance data by Kreuser et al., 1988) thiscorresponds to 2.0 ± 0.4 km of removedsection.
Samples immediately to the west of the rift(00512, 00508) record a Late Jurassic - EarlyCretaceous cooling event around -150 Ma (Fig.8d), in addition to Cenozoic ($40 Ma) cooling.Late Jurassic AFT ages are also encountered onthe western flank of the Rukwa Rift. Sample00508 from an elevation of around 1000 m a.s.1.has shorter MTL than sample 00512 from anelevation of above 2000 m (11.4 versus 12.4Jim). Modelling suggests that the lower samplehas remained 20-30 oC warmer than thetopographically higher one from Late Jurassicto Palaeogene times.
The fact that Late Jurassic-Early Cretaceouscooling is only recorded in the western flanks ofboth rifts suggests that this event was causedby more local denudation, possibly related toblock faulting around that time. Seismic datafrom the Rukwa Rift indicate that the eastern(Lupa) and western (Ufipa) border faults of therift were active intermittently during its evolution(e.g. Morley et al., 1992; Kilembe andRosendahl, 1992; Mbede, 1993). Within theMalawi Rift Basin, the possibly Upper JurassicCretaceous Red Sandstone Group ("Dinosaurbeds") crops out unconformably on top of Karoosediments, suggesting renewed rifting andsediment accumulation during that time.Sediment provenance for the Red Sandstonesappears to be mainly from the west (Dypvik et
Journal of African Earth Sciences 377
P VAN DER SEEK et al.
al., 1990), consistent with contemporaneousdenudation of the western flank. The dating ofthe Red Sandstones in the Rukwa and MalawiRifts is, however, controversial (Damblon et al.,1998) and their deposition could also be relatedto the inferred widespread Cenozoic denudationthat is suggested by the modelling.
Regional correlationsThe data of this study indicate that the flanksof the Malawi and Rukwa Rifts have experiencedaccelerated regional cooling (and, by inference,denudation) during Triassic-Early Jurassic (-250200 Mal, Late Jurassic-Early Cretaceous (-150Ma) and Tertiary (:<;;40-50 Ma) times. Samplesfrom the base of the Livingstone MountainsEscarpment indicate that more than half of theCenozoic cooling may have taken place frommid-Miocene (-20 Ma) onwards. Of thesecooling events, the Triassic-Early Jurassic andCenozoic events appear to be most widespread,being recorded by practically all samples. TheLate Jurassic-Early Cretaceous event is onlyrecorded by samples from the western flanks ofthe Malawi and Rukwa Rifts, whereas postMiocene cooling can only unambiguously bedemonstrated from the base of the Malawifootwall flank.
The Triassic-Early Jurassic cooling phase marksthe end of the Karoo sedimentary regime and iswidely recognised as a major erosional event(e.g. Wopfner, 1993). As denudation during thisphase appears to have affected both the Karoobasins and their flanks equivalently, there mayhave been very little topography left at the endof Karoo times and the erosional episode musthave been triggered externally. The Late JurassicEarly Cretaceous phase is possibly coeval withdeposition of the Red Sandstone Group, althoughthe latter may also be correlated with renewedTertiary cooling. There is thus a close correlationbetween phases of denudation of the rift flanksand the onset or termination of rifting events,as recorded by the different sedimentary unitswithin the Malawi-Rukwa Rifts.
Brown et al. (1990) reported a widespreadphase of denudation around 140 ± 10 Ma insouthern Africa, whereas Foster and Gleadow(1993) report denudation events at around 220Ma, 140-120 Ma and 60-70 Ma from centralKenya. The results of this study appear roughlyconsistent with these previous studies, suggestingthat the inferred phases of denudation are of(sub)continental significance. The largestdifference in timing appears for the Late JurassicEarly Cretaceous phase, but this is the least
378 Journal ofAfrican Earth Sciences
constrained from the samples of this study.Foster and Gleadow (1993) suggested that suchregional phases of denudation can be correlatedto periods of plate tectonic reorganisation andintracontinental deformation. For instance,Triassic-Early Jurassic denudation is coeval withthe onset of rifting between East Africa andMadagascar (e.g. Wopfner, 1993); the LateJurassic-Early Cretaceous is the time of initialrifting in the South Atlantic, and the earliestTertiary event can be correlated with a majorplate reorganisation in the Indian Ocean (e.q,Foster and Gleadow, 1993; Janssen et al.,1995). These changes in plate motion appearto be correlated with basin reactivationthroughout the African continent (Bosworth,1992; Janssen et et., 1995).
UPLIFT AND EROSION HISTORY OF THEMALAWI-RUKWA RIFT FLANKS
The morphology of the high level plateaus anduplands flanking the Western Rift system in EastAfrica have been variably explained as eitheressentially relict features, initiated by Karootectonics and only influenced by erosionalprocesses since then (e.g. Dixey, 1947) or, incontrast, as being a direct result of tectonic upliftduring Late Tertiary rifting (e.g. Ebinger et el.,1991, 1993). This study indicates that the riftflanks record a protracted denudation history,with subsequent erosional phases correlated tothe different rifting events within the MalawiRukwa Rift Zone. This section will discussattempts to quantify the amounts of denudationthat have taken place on the flanks during thedifferent erosional episodes. Subsequentattempts to constrain the relative roles of upliftand erosion in creating the present-daymorphology of the flanks through numericalmodelling of rifting and flank uplift will bediscussed.
Estimates of denudationCooling of the upper crust can, in general, becaused either by a reduction in the geothermalgradient or by denudation. In order to constrainthe amount of denudation that has occurred,some bounds need to be placed on thegeothermal gradient. Mbede (1993) calculateda present-day geothermal gradient of -40°C krn'for the Rukwa Rift from well-log data within thebasin. This intra-basinal thermal gradient cannotbe directly extrapolated to the flanking regions,however. A regional study by Nyblade et al.(1990) indicates that heat flow within the Pan-
Denudation history of the Malawi and Rukwa Rift flanks from apatite fission track thermochronology
A
SWLake Malawi LivingstoneMts
post-Karoo
A'
NE
4-
188
o 20km
217
Figure 9. Simplified structuraland topographic protile across the northern Malawi Rift (after Ebinger et al., 1993) showing thedistribution of representative AFT ages (black dots with numbers) and the reconstructed regional distribution of denudationacross the section for the different erosional phases recognised. Note that these "peteeotendsurteces" do not representpalaeo topographic elevations because they are not isostatically balanced (e.g. van der Beek et al., 1994). The structure ofthe northern Malawi Rift is from the project PHOBE seismic line 804 (e.g. Rosendahl et al., 1992). coarse stippling indicatesLate Cenozoic rift deposits; grey shading: Karoo sediments; LMBF: Livingstone Mountains Border Fault.
African mobile belts (including the Karoo andrecent rifts) is -60-80 mW rn", whereas thecratons record a much lower heat flow of -35mW rn". For a thermal conductivity of crystallineupper crustal rocks of -2.5 W m' K-1
, theseestimates suggest that 25-30°C krn' is areasonable value for the present-day geothermalgradient in the Malawi-Rukwa region.
Assuming a surface temperature of 25-30oC,
the widespread Cenozoic cooling from -60°Crepresents 1.0-1.4 km of regional denudation.Because of the possible artefacts in the annealingmodels discussed previously, this amount ofcooling and denudation must be considered amaximum value. Samples from the foot of theLivingstone Mountains Escarpment recordcooling from ~120oC during Tertiary times,which would correspond to 2.7-3.8 km ofdenudation. Of this amount, 2.2±0.4 km(corresponding to cooling from -80°C down tosurface temperatures) appears to have takenplace since the Miocene (i.e. post-20 Mal, andprobably contemporaneously with the presentrifting activity. The summit of the LivingstoneMountains lies at an elevation of 1.5-2 km abovethe lake level. This elevation difference suggeststhat Cenozoic uplift and isostatic rebound of thelake shore was greater than that of theLivingstone Mountains summit, 10-20 km awayfrom the rift (Fig. 9).
Constraining the amount of denudation relatedto the earlier cooling events is more difficult
because there is no direct measure of palaeoheatflow. However, a vitrinite reflectance study byKreuser et al. (1988) suggests that thegeothermal gradient through the Karoo basinsof southern Tanzania at the time of maximumburial (i.e. -250 Ma) was around 25-30 oC krn1, similar to the present-day gradient. If it isassumed that regional heat flow remained stableduring Meso-Cenozoic times, then the 50-60°Ccooling recorded by samples from the westernflanks of the Malawi and Rukwa Rifts duringthe Late Jurassic-Early Cretaceous correspondto 2.0 ±0.4 km of local denudation. The regionaldenudation event at the end of the Karoodepositional episode would have removed asimilar amount of overburden from the entireregion (Fig. 9).
The total post-Karoo denudation of theregion thus varies between 3.2 ± 0.6 km forthe summit plateau of the LivingstoneMountains to 5.3±0.7 km for the lake shorein front of the Livingstone MountainsEscarpment. Of the three denudation episodesrecorded, the late Karoo phase appears to bethe most important, removing -2 km ofmaterial from the entire region. The LateJurassic-Early Cretaceous event caused similaramounts of denudation, but only on thewestern flanks of the rifts, whereas Cenozoicdenudation was probably ~1 krn, except atthe Livingstone Mountains Escarpment, whereit may have reached -3 km.
JournalofAfrican Earth Sciences 379
P. VAN DER SEEK et al.
Table 2. Best-fit parameters employed in thermomechanical modelling of extension and rift flank uplift
parameter
crustal thickness (km)lithospheric thickness(km)onset of rifting (Ma)
Figure 10. Predicted basin geometry and flank topography for a profile across the northern Malawi Rift, adopting a detachmentdepth of 30 km, 7.2 km extension on the Livingstone Mountains Border Fault and T =30 km, both with and without'Cenozoic erosion (amount of erosion as in Fig. 9J. The thick grey line indicates the obse:'ved topography, light grey shadingis the observed Cenozic basin fill; dark shading: Karoo sediments.
Modelling Cenozoic rift flank uplift and erosionIn order to quantitatively assess the roles oftectonic uplift, erosion and isostatic rebound incontrolling topography around the Malawi andRukwa Rifts, numerical models for continentalextension and rift flank uplift have been used.Flank uplift along continental rifts such as theWestern Rift of East Africa is best explained asa result of the flexural isostatic response of the
380 Journal 01 African Earth Sciences
lithosphere to extensional unloading during rifting(e.g. Weissel and Karner, 1989; Ebinger et et.,1991, 1993; van der Seek et et., 1994; Upcottet el., 1996).
A profile which runs across the northerntermination of Lake Malawi has been modelledusing a thermomechanical model developed byter Voorde and Cloetingh (1996). In this model,crustal extension is accommodated by slip at a
Denudation history of the Malawi and Rukwa Rift flanks from apatite fission track thermochronology
Free Air Gravity Anomaliesl00mGal
-.JtO
-100
-200€> observed gravity
--c
- -
Topography and Basement depth2km
Ufipa Plateau L. Rukwa
a
-2
-4
-6
observedno erosionincluding erosion
Lupa Plateau
aI
100 krnI
Figure 11. (lower panel) Predicted basin geometry and flank topography for a profile across the central Rukwa Rift, adoptinga detachment depth of 40 km, 4 km extension on the Lupa Fault during Karoo rifting and 6 km during Cenozoic rifting, 5 kmextension on the Ufipa Fault during Cenozoic rifting, and T = 28 km, both with and without erosion (after Mbede, 1993). Themodel is fitted to the observed basement geometry in s~ismic line TVZ07 (Morley et al., 1992; Kilembe and Rosendahl,1992; Mbede, 1993). (upper panel) Predicted Free Air gravity anomaly profile, with and without erosion. Note particularly thenegative Free Air anomalies over the eastern flank, which provide strong evidence for erosion of the flank.
finite rate along one or more crustal normalfaults, and is counterbalanced by distributedpure-shear thinning and extension below thedepth at which the fault soles out (thedetachment depth z). A similar profile wasmodelled by Ebinger et al. (1991, 1993) usingthe instantaneous extension model of Weisseland Karner (1989); these previous analyses didnot take erosion of the rift flanks into account.
The predicted rifting kinematics and isostaticrebound are mainly controlled by threeparameters: the shape of the fault, the amountof extension, and the flexural rigidity (orequivalent elastic thickness T) of the lithosphere.The border faults of the Malawi and Rukwa Riftsappear to sole out in the middle to lower crust,at a depth of 20-30 km (e.g. Jackson and
Blenkinshop, 1993; Camelbeeck and Iranga,1996). The amount of extension and Te of thelithosphere was determined by fitting thebasement depth beneath the basin, as observedon seismic reflection profiles, and the elevationof the rift flanks (van der Beek, 1995). Best-fitmodelling parameters are given in Table 2. It isassumed that, prior to Late Cenozoic extension,the region was flat and at an elevation of 1.5km a.s.l., the present-day elevation of most ofthe Tanzanian Craton.
The modelling results, presented in Fig. 10,suggest that Cenozoic erosion has been animportant factor in modifying flank topography.For a detachment depth of 30 km and Te =30krn, 7.2 km of extension on the LMBF is requiredto fit the observed basement depth. A model
Journal of African Earth Sciences 381
P VAN DER SEEK et al.
that does not take erosion of the rift flanks intoaccount does not predict sufficient topographyfor the footwall flank but over estimates theelevation at the hanging wall side of the rift. Iferosion of the flanks is incorporated, theadditional isostatic rebound provides a muchbetter fit to the observed elevation of theLivingstone Mountains, whereas the elevationof the western hanging wall flank is lowered.The amounts of Cenozoic erosion incorporatedin the modelling are those derived from the FTdata and depicted in Fig. 9. This result isconsistent with structural studies of the northernMalawi Basin (Ebinger et st., 1993), whichindicate that several hundreds of metres ofsediment have been eroded from the hangingwall side of the basin since the onset of rifting,the eroded material being redeposited in thedepocentre close to the LMBF.
Mbede (1993) used a broadly similar numericalmodel (Weissel and Karner, 1989; Ebinger eta/., 1991) to predict the present-day topographysurrounding the Rukwa Rift (Fig. 11). Sheadopted a two-stage rifting model to predict thebasin evolution from the Karoo to recent times,firstly modelling the basement subsidence andflank uplift due to Karoo rifting, then removingthe amount of material eroded since Karoo times(e.g. Fig. 9) and calculating the isostaticresponse, and finally imposing Cenozoicextension. Again, the fit to the observedtopography becomes much better when theamount of denudation that is recorded by theAFT data is incorporated in the modelling. Mbede(1993) also used the model to predict Free Airgravity anomalies over the basin, in order to havea separate control on amounts of denudation.The Free Air anomaly is normally more sensitiveto changes in crustal thickness due to denudationthan topography (ct. Ebinger et el., 1991). Acomparison of observed and predicted Free Airanomaly profiles (Fig. 11) provides independentevidence for considerable amounts of denudationsince Karoo rifting, especially on the eastern(Lupa Plateau) flank of the rift.
CONCLUSIONSThe AFT data from the flanks of the Malawi andRukwa Rifts record a protracted cooling historywith three distinct phases of accelerated coolingand denudation that are related to the differentrifting events in the region. Both rifting anddenudation appear to be ultimately controlledby continental-scale tectonic events related tochanges in plate motions.
382 Journal of African Earth Sciences
The oldest denudation event recorded by theAFT data, at around 250-200 Ma, can beascribed to late Karoo erosion that affected theentire area. Samples from the LivingstoneMountains summit and those from the easternflank of the Rukwa Rift define a similar structurallevel or erosion surface. This surface hastraditionally been correlated to the Jurassic"Gondwana surface" of eastern and southernAfrica, but the AFT data indicate that finalexposure of this surface did not take place beforethe Palaeogene (-40 Ma).
Samples from the western flanks of the Malawiand Rukwa Rifts record a Late Jurassic-EarlyCretaceous (-150 Ma) phase of denudation,possibly related to renewed rifting and thedeposition of the Red Sandstone Group. Finally,thermal history reconstructions from inversemodelling of track length distributions suggestthat most samples record -35°C cooling duringthe Cenozoic. Samples from the base of theLivingstone Escarpment were exhumed fromtemperatures ~120°C during Cenozoic times,with more than half the denudation taking placesince 20 Ma. Estimated amounts of denudationfor the late Karoo and Late Jurassic-EarlyCretaceous events are 2.0 ± 0.4 km each,whereas Cenozoic denudation probablyamounted to ~1.2 ±0.2 km
Modelling of Cenozoic extension and rift flankuplift indicates that rift related erosion of theflanks has modified the observed topography.The high elevation of the Livingstone Mountainsaway from the Malawi Rift Zone appears to becaused by regional isostatic response to erosionof the escarpment, whereas the elevation of thehanging wall flank may have been loweredsignificantly by erosion. Topographic and gravitydata from the Rukwa Rift suggest that especiallythe eastern flank (Lupa Plateau) of that rift hasbeen severely modified by erosion.
ACKNOWLEDGMENTSThis research formed part of the Ph. D. studiesof P. van der Beek at the Vrije UniversiteitAmsterdam, and E. Mbede at the TechnischeUniversitat Berlin. Fieldwork was carried out aspart of a UNESCO sponsored geotraverse withinthe framework of the CASIMIR (ComparativeAnalysis of Sedimentary Infill Mechanisms inRifts) project, funded by the Belgian government.The Tanzania Commission for Science andTechnology (COSTECH), the University of Dares Salaam and Madini office provided logisticalsupport. Thanks are due to R. Kajara and J.
Denudation history of the Malawi and Rukwa Rift flanks from apatite fission track thermochronology
Sarota of Madini Oodoma for assistance duringsample collection. The London Fission TrackResearch Group provided AFT dating facilitiesfor the samples from the Rukwa Rift. The othersamples were processed at the Vrije UniversiteitAmsterdam, with assistance from Lodewijk Ijlstand Tineke Vogels. The authors thank WayneNoble for stimulating discussions on theinterpretation of AFT data from Tanzania andfor providing preprints of parts of his thesis. Themanuscript was substantially improved by thecomments of C. Ebinger, H. Wopfner and ananonymous reviewer. Netherlands ResearchSchool of Sedimentary Geology Publication no271197. This is a contribution to IGCP400project "Geodynamics of Continental Rifting".
REFERENCESBosworth. W. 1985. Geometry of propagating continental
rifts. Nature 316,625-627.Bosworth, W. 1992. Mesozoic and early Tertiary rift tectonics
in East Africa. Tectonophysics 209, 115-137.Brown, R. W., Rust, D. J., Summerfield, M. A., Gleadow,
A. J. W. and de Wit, M. C. J. 1990. An Early Cretaceousphase of accelerated erosion on the south-western marginof Africa: Evidence from apatite fission track analysis andthe offshore sedimentary record. Nuclear Tracks RadiationMeasurements 17, 339-350.
Brown, R. W., Summerfield, M. A. and Gleadow, A. J. W.1994. Apatite fission-track analysis: Its potential for theestimation of denudation rates and implications for modelsof long-term landscape development. In: Process Modelsand Theoretical Geomorphology (Edited by Kirkby, M. J.)pp24-53. Wiley, New York.
Camelbeeck, T. and lranga, M. D. 1996. Deep crustalearthquakes and active faults along the Rukwa trough,Eastern Africa. Geophysical Journallntemationa/124, 612630.
Chorowicz, J. 1990. Dynamics of different basin types inthe East African rift. Journal African Earth Sciences 10,271-282.
Corrigan, J. 1993. Apatite fission-track analysis of Oligocenestrata in South Texas, U.S.A.: Testing annealing models.Chemical Geology (Isotope Geoscience Section) 104, 227249.
Crowley, K. D., Cameron, M. and Schaefer, R. L. 1991.Experimental studies of annealing of etched fission tracks influorapatite. Geochimica Cosmochimica Acta 55, 1449-1465.
Daly, M. C., Chorowicz, J. and Fairhead, J. D. 1989. Riftbasin evolution in Africa: the influence of reactivated steepbasement shear zones. In: Inversion Tectonics (Edited byCooper, M. A. and Williams, G. D.) Geological SocietySpecial Publication 44, 309-334.
Damblon, F., Gerrienne, Ph., d'Outrelepont, H., Delvaux, D.,Beekman, H. and Back, S. 1998. Identification of a fossilwood specie men in the Red Sandstone Group ofsouthwestern Tanzania: stratigraphical and tectonicimplications. In: Tectonics, sedimentation and volcanismin the East African Rift System (Edited by Delvaux, D. andKhan, M. A.I Journal African Earth Sciences 26,387-396.
Delvaux, D. and Hanon, M. 1993. Neotectonics of the Mbeyaarea, SW Tanzania. Rapport Annue/1991-1992, pp8797. Departement de Geologie et Mineraloqie, Musee Royalde I' Afrique Centrale, Tervuren, Belgium.
Delvaux, D. and Wopfner, H. 1992. Some observations onthe geomorphology and recent rift movements in theLivingstone Mountains, SW Tanzania. UNESCO, GeologyEconomic Development Newsletter 9, 171-173.
Delvaux, D., Levi, K., Kajara, R. and Sarota. J. 1992.Cenozoic paleostress and kinematic evolution of the Rukwa- North Malawi rift valley (East African rift system). BulletinCentres Recherches Exploration Production Elf-Aquitaine16,383-406.
Dixey, F. 1928. The Dinosaur beds of Lake Nyasa.Transactions Geological Society South Africa 16, 55-66.
Dixey, F. 1937. The early Cretaceous and Miocenepeneplains of Nyasaland and their bearing on the age anddevelopment of the Rift Valley. Geological Magazine 74,49-82.
Dixev, F. 1947. Erosion and tectonics in the East Africanrift system. Quarterly Journal Geological Society London102, 339-388.
Dypvik, H., Nesteby, H., Ruden, F., Aagaard, P., Johansson,T., Msindai, J. and Massay, C. 1990. Upper Paleozoicand Mesozoic sedimentation in the Rukwa-Tukuyu Region,Tanzania. Journal African Earth Sciences 11,437-456.
Ebinger, C. J. 1989. Tectonic development of the westernbranch of the East African rift system Geological SocietyAmerica Bulletin 101,885-903.
Ebinger, C. J., Rosendahl, B. R. and Reynolds, D. J. 1987.Tectonic model of the Malawi rift, Africa. Tectonophysics141,215-235.
Ebinger, C. J., Deino, A. L., Drake, R. E. and Tesha, A. L.1989. Chronology of volcanism and rift basin propagation:Rungwe volcanic province, East Africa. JournalGeophysical Research 94, 15,783-15,803.
Ebinger, C. J., Karner, G. D. and Weissel, J. K. 1991.Mechanical strength of extended continental lithosphere:constraints from the Western rift system, Africa. Tectonics10,1239-1256.
Ebinger, C. J., Deino, A. L., Tesha. A. L., Becker, T. andRing, U. 1993. Tectonic controls on rift basin morphology:Evolution of the Northern Malawi (Nyasa) Rift. JournalGeophysical Research 98, 17,821-17,836.
Fitzgerald, P. G. 1994. Thermochronologic constraints onpost-Paleozoic tectonic evolution of the centralTransantarctic Mountains, Antarctica. Tectonics 13, 818836.
Foster, D. A. and Gleadow, A. J. W. 1992. Themorphotectonic evolution of rift-margin mountains incentral Kenya: Constraints from apatite fission-trackthermochronology. Earth Planetary Science Letters 113,157-171.
Foster, D. A. and Gleadow, A. J. W. 1993. Episodicdenudation in East Africa: A legacy of intracontinentaltectonism. Geophysical Research Letters 20,2395-2398.
Foster, D. A. and Gleadow, A. J. W. 1996. Structuralframework and denudation history of the flanks of theKenya and Anza Rifts, East Africa. Tectonics 15, 258271.
Galbraith, R. F. 1990. The radial plot: Graphical assessmentof spread in ages. Nuclear Tracks Radiation Measurements17,207-214.
Galbraith, R. F. and Laslett, G. M. 1993. Statistical modelsfor mixed fission track ages. Nuclear Tracks RadiationMeasurements 21,459-470.
Gallagher, K. 1995. Evolving temperature histories fromapatite fission track data. Earth Planetary Science Letters136,421-435.
Gleadow, A. J. W. and Fitzgerald, P. G. 1987. Uplift historyand structure of the Transantarctic Mountains: newevidence from fission track dating of basement apatitesin the Dry Valleys area, southern Victoria land. EarthPlanetary Science Letters 82, 1-14.
Journal of African Earth Sciences 383
P. VAN DER SEEK et al.
Gleadow, A. J. W., Duddy, I. R., Green, P. F. and Lovering,J. F. 1986. Confined fission track lengths in apatite: adiagnostic tool for thermal history analysis. ContributionsMineralogy Petrology 94, 405-415.
Green, P. F., Duddy, I. R., Gleadow, A. J. W. and Lovering,J. F. 1989a. Apatite fission-track analysis as apaleotemperature indicator for hydrocarbon exploration.In: Thermal history of sedimentary basins (Edited byNaeser, N. D. and McCulloh, T. H.) pp181-195. SpringerVerlag, New York.
Green, P. F., Duddy, I. R., Laslett, G. M., Hegarty, K. A.,Gleadow, A. J. W. and Lovering, J. F. 1989b. Thermalannealing of fission tracks in apatite 4. Quantitativemodelling techniques and extension to geologicaltimescales. Chemical Geology (Isotope GeoscienceSection) 79, 155-182.
Harkin, D. A. 1955. The geology of the Songwe-Kiwira coalfield, Rungwe District. Geological Survey TanganyikaBulletin 27, 17p.
Harkin, D. A. 1960. The Rungwe volcanics at the northernend of Lake Nyasa. Geological Survey Tanganyika, Memoir11,172p.
Hurford, A. J. 1990. International union of geologicalsciences subcommission on geochronologyrecommendation for the standardization of fission trackdating calibration and data reporting. Nuclear TracksRadiation Measurements 17, 233-236.
Jackson, J. and Blenkinsop, T. G. 1993. The Malawiearthquake of March 10, 1989: Deep faulting withinthe East African Rift System. Tectonics 12, 11311139.
Jacobs, I.. L., Winkler, D. A., Kaufulu, Z. M. and Downs, W.R. 1989. The Dinosaur Beds of Northern Malawi, Africa.National Geographic Research 6, 196-204.
Janssen, M. E., Stephenson, R. A. and Cloetingh, S.1995. Temporal and spatial correlations betweenchanges in plate motions and the evolution of riftedbasins in Africa. Geological Society America Bulletin107,1317-1332.
Kilembe, E. and Rosendahl, B. R. 1992. Structure andstratigraphy of the Rukwa Basin. Tectonophysics 209,143-158.
King, I.. C. 1963. South African Scenery. 308p. Oliver andBoyd, Edinburgh, London.
Kreuser, T., Schradmedei, R. and Rullkotter, J. 1988. Gasprone source rocks from cratogene Karoo basins inTanzania. Journal Petroleum Geology 11, 169-184.
Laslett, G. M., Green, P. F., Duddy, I. R. and Gleadow, A. J.W. 1987. Thermal annealing of fission tracks in apatite 2.A quantitative analysis. Chemical Geology (IsotopeGeoscience Section) 65, 1-13.
Lenoir, J. 1.., Lieqois. J.-P., Theunissen, K. and Klerkx, J.1994. The Paleoproterozoic Ubendian shear belt inTanzania: geochronology and structure. Journal AfricanEarth Sciences 19, 169-1 84.
Lutz, T. M. and Omar, G. I. 1991. An inverse method ofmodeling thermal histories from apatite fission-track data.Earth Planetary Science Letters 104, 181-195.
Mbede, E. I. 1993. Tectonic development of the Rukwa RiftBasin in SW Tanzania. Berliner GeowissenschaftlicheAbhandlungen A 152, 92p.
Mbede, E. I., Hurford, A., Ebinger, C. J. and Schandelmeier,H. 1993. A constraint on the uplift history of the Rukwarift, SW Tanzania: a reconnaissance apatite fission trackanalysis. Rapport Annuel 1991-1992, pp99-1 08.Departement de Geologie et Mineraloqie, Musee Royal deI' Afrique Centrale, Tervuren, Belgium.
Morley, C. K., Cunningham, S. M., Harper, R. M. andWescott, W. A. 1992. Geology and geophysics of theRukwa rift, East Africa. Tectonics 11, 68-81.
384 Journal of African Earth Sciences
Noble, W. P. 1997. Post Pan-African tectonothermalevolution of Eastern Africa: An apatite fission track study.Ph. D. Thesis. LaTrobe University, Bundoora (Victoria,Australia) .
Noble, W. P., Foster, D. A. and Gleadow, A. J. W. 1997.The Post Pan-African tectonothermal development of theMozambique belt in eastern Tanzania. Tectonophysics 275,331-350.
Nyblade, A. A., Pollack, H. N., Jones, D. 1.., Podmore , F.and Musjayandebvu, M. 1990. Terrestrial heat flow inEast and Southern Africa. Journal Geophysical Research95,17,371-17,384.
Omar, G. I., Steckler, M. S., Buck, W. R. and Kohn, B. P.1989. Fission-track analysis of basement apatites at thewestern margin of the Gulf of Suez rift, Egypt: evidencefor synchroneity of uplift and subsidence. Earth PlanetaryScience Letters 94, 316-328.
Ring, U., Betzler, C. and Delvaux, D. 1992. Normal vs. strikeslip faulting during rift development in East Africa: theMalawi rift. Geology 20,1015-1018.
Rosendahl, B. R. 1987. Architecture of continental rifts withspecial reference to East Africa. Annual Review EarthPlanetary Sciences 15, 445-503.
Rosendahl, B. R., Kilembe, E. and Kaczmarick, K. 1992.Comparison of the Tanganyika, Malawi, Rukwa andTurkana rift zones from analyses of seismic reflection data.Tectonophysics 213,235-256.
Semkiwa, P. Z. 1992. Depositional environments and coalpetrography of Permian coal deposits of SW Tanzania.Geologisches Institut Universitiit Koln Sonderverottentlichungen 84, 184p. Cologne, Germany.
Ter Voorde, M. and Cloetingh, S. 1996. Numericalmodeling of extension in the faulted crust: effects oflocalized and regional deformation on basin stratigraphy.In: Modern Developments in Structural Interpretation,Validation, and Modelling (Edited by Buchanan, P. G.and Nieuwland, D. A.I Geological Society SpecialPublication 99, 283-296.
Theunissen, K., Lenoir, J.-I.., l.ieqeois, J.-P., Delvaux, D. etMruma, A. 1992. Empreinte mozambiquienne majeuredans la chaine ubendienne de Tanzanie sud-occidentale:geochronologie U-Pb sur zircon et contexte structural.Comptes Rendus Academie Sciences Paris, Serie II 314,1355-1362.
Theunissen, K" Klerkx, J., Melnikov, A. and Mruma, A. 1996.Mechanisms of inheritance of rift faulting in the westernbranch of the East African Rift, Tanzania. Tectonics 15,776-790.
Tiercelin, J.-J., Chorowicz, J., Bellon, H., Richert, J.-P.,Mwambene, J. T. and Walgenwitz, F. 1988. East Africanrift system: offset, age and tectonic significance of theTanganyika-Rukwa-Malawi intracontinental fault zone.Tectonophysics 148, 241-252.
Upcott, N. M., Mukasa, R. K., Ebinger, C. J. and Karner, G.D. 1996. Along-axis segmentation and isostasy in theWestern rift, East Africa. Journal Geophysical Research101,3247-3286.
Van den Haute, P. 1984. Fission-track ages of apatites fromthe Precambrian of Rwanda and Burundi: relationship toEast African rift tectonics. Earth Planetary Science Letters71,129-140.
Van der Beek, P. A. 1995. Tectonic evolution of continentalrifts: Inferences from numerical modelling and fission trackthermochronology. Ph. D. Thesis 248p. Vrije UniversiteitAmsterdam, the Netherlands.
Van der Beek, P. A., Cloetingh, S. and Andriessen, P. A. M.1994. Mechanisms of extensional basin formation andvertical motions at rift flanks: Constraints from tectonicmodelling and fission-track thermochronology. EarthPteneterv Science Letters 121, 417-433.
Denudation history of the Malawi and Rukwa Rift flanks from apatite fission track thermochronology
Van der Beek, P. A., Delvaux, D., Andriessen, P. A. M. andLevi, K. G. 1996. Early Cretaceous denudation related toconvergent tectonics in the Baikal region, SE Siberia.Journal Geological Society London 153, 515-523.
Versfelt, J. and Rosendahl, B. R. 1989. Relationshipsbetween pre-rift structure and rift architecture in LakesTanganyika and Malawi, East Africa. Nature 337, 354357.
Wagner, G. A. 1979. Correction and interpretation of fissiontrack ages. In: Lectures in Isotope Geology (Edited byJager, E. and Hunziker, J. C.) pp170-177. SpringerVerlag, Berlin.
Wagner, M., Altherr, R. and van den Haute, P. 1992. Apatitefission-track analysis of Kenyan basement rocks:constraints on the thermotectonic evolution of the Kenyadome. A reconnaissance study. Tectonophysics 204,93110.
Weissel, J. K. and Karner, G. D. 1989. Flexural uplift of riftflanks due to mechanical unloading of the lithosphereduring extension. Journal Geophysical Research 94,13919-13950.
Wescott, W. A, Krebs, W. K., Engelhardt. D. W. and Cunningham,S. M. 1991. New biostratigraphic age dates from the LakeRukwa rift basin in Western Tanzania. American AssociationPetroleum Geologists Bulletin 75, 1255-1263.
Wheeler. W. H. and Karson, J. A. 1989. Structure andkinematics of the Livingstone Mountains border fault zone,Nyasa (Malawi) Rift, southwestern Tanzania. JournalAfrican Earth Sciences 8, 393-413.
Wheeler, W. H. and Karson, J. A 1994. Extension and subsidenceadjacent to a "weak" continental transform: an example ofthe Rukwa Rift, East Africa. Geology 22, 625-628.
Wopfner, H. 1990. Rifting in Tanzanian Karoo basins and itseconomic implications In: Etudes Recentes sur la Geologiede I'Afrique, 15e colloque de Geologie Africaine, Nancy1990 (Edited by Rocci, G. and Deschamps, M.) CIFEGOccasional Publications 1990/22, 217-220.
Wopfner, H. 1993. Structural development of TanzanianKaroo basins and the break-up of Gondwana. In: GondwanaEight; Assembly, Evolution and Dispersal {Edited byFindlay, R. H., Unrug, R., Banks, M. R. and Veevers. J..L) pp531-540. Balkema, Rotterdam.
