ELSEVIER
Tectonophysics 236 (1994) 3-22
Stratigraphic and structural constraints on mechanisms of active rifting in the Gregory Rift, Kenya Martin Smith
’
International Division, British Geologicat Survey, Keyworth, Nottingham, NGI2 SGG, UK
Received 29 October 1992; revised version accepted 5 February 1993
Abstract A review of the stratigraphic and structural events associated with the early evolution of the Qregory Rift indicates that they are consistent with the active model of rifting. At present, there is no geochemical evidence to indicate unequiv~lly the presence of magma derived from asthenospheric material. However, the evidence for limited pre-rift uplift and the mo~hologic and geochemi~l features of the early basalts and phonolite; flood Iavas support the intrusion of an asthenospheric plume into the Iithospheric upper mantle during mid-Miocene times. The scale of uplift f < 1 km) and the limited volumes of lava produced indicate that a small convective cell, with an initial diameter of perhaps only 100-150 km, rather than a major asthenospheric plume underlies the Gregory Rift. The subsequent rise of this small plume through the lithosphere to the base of the crust is documented by Late Miocene and Pliocene magmatic and rifting events. The presence of a relatively thick mechanical boundary layer beneath the Gregory Rift prior to rifting, indicated by the general absence of tholeiitic volcanism and low extension rates, is supported by geophysical and geochemical evidence for igneous underplating and significant magma ffactionation within the lithosphere. There is a strong coincidence of location of magmatic activity and rifting with pre-existing zones of crustal weakness beneath the Gregory Rift. The rising plume was focused beneath a weak zone harking the contact between the reworked and buried margin of the Tanzanian craton and the adjacent Proterozoic mobile belt. Within this zone, major crustal-scale shear zones and thrusts accommodated limited Iithospheric stretching and influenced the location and development of half-graben basins and transfer zones. Early basins were nucleated on NW-SE- and N-S-trending weaknesses. As the plume ascended and spread laterally, rifting propagated away from this zone and younger basins developed across both mobile belt and cratonic crust.
1. Introduction It has long been suspected that much of the African continent is underlain by anomalous crust and/or mantle (e.g., Girdler et al., 1969; Burke
’ Present address: British Geological Survey, Murchison House, West Mains Road, Edinburgh, EH9 3LA, UK.
and Whiteman, 1973). Recent seismia tomography data (Anderson et al., 1992) support this view and lend credence to the theory that upwelling of mantle material is important in the evolution of the African lithosphere. Implicit in many of the key papers which consider the evolution of the Kenya Rift is that thermal processes within the lithosphere
are
implant
Baker and Wohlenberg,
0040-1951/94/%07.00 0 1994 Ekevier Science B.V. All rights reserved SSDI 0040-1951(94)00020-A
during
rifting
(e.g.,
1971; Baker et al., 1972;
Area
1 Approximate outcrop of Oligocene, tholeitic basal% and rhyolites
0
Phonolite/trachyte centre
eruptive
Area 2 Outcrop of Archaean basement
I+
+-+ +
+ I”
3
f +
Fig. 1. Outline and location of rifting and volcanism within the Kenya Rift Zone. Major rift faults on downthrow side.
M. Smith / Tectonophysics236 (1994) 3-22
Logatchev, 1974; King, 1978; Williams, 1978a). However, at the time of these early studies, the deep crustal structure was largely unknown and geophysical models were based on coarse resolution gravity and seismic data. In 1981, at the conference on the Processes of Planetary Rifting (Morgan and Baker, 19831, rift development was considered in terms of two end-member models. In the active model, asthenospheric upwelling or diapirism provides the driving force for lithospheric uplift and extension, whereas the passive model relies on mechanical dilation or lithospheric stretching in response to regional plate stresses. In recent years a large number of geophysical investigations in the Kenya Rift have been directed mainly to elucidate the lithospheric structure and to provide constraints on these models. The current three-dimensional geophysical data support the active model and suggest that the location and pattern of rifting in Kenya is largely a response to the interaction of a thermal pe~urbation or plume at the base of a lithosphere undergoing pure shear extension (Henry et al., 1990; Davis, 1991; Green et al., 1991). Stratigraphic studies supported by geochrono-. logical and geochemical data, permit the subdivision of pre-10 Ma volcanism and rifting in Kenya into three distinct areas (Fig. 1). Although volcanism in all three areas may be broadly related to the upwelling of hot mantle material, each area shows a distinct evolutionary history. Thus, in the north, volcanism within the Turkana rifts was initiated in Oligocene times with the eruption of voluminous flood tholeiitic basalts and rhyolites dated at 35-22 Ma (Morley et al., 1992). These infilled a shallow sag basin in the Lotikipi area and extended as far south as southern Lake Turkana and were followed by the formation of a series of N-S- to NNW-SSE-trending asymmetric rift basins (Morley et al., 1992). In western Kenya and eastern Uganda a second area is characterised by strongly alkaline carbonatitic volcanism (King et al., 1972; Le Bas, 1987). Activity commenced with the intrusion of a suite of small carbonatitic plugs dated at 31-25 Ma and was followed in Miocene times by the eruption of a series of large central volcanoes, dated at 22-12 Ma. This area of volcanism, located on the mar-
5
gin of the Tanzanian craton, is not associated with regional uplift nor, with the exception of those in the Nyanza Rift, are the eruptive centres obviously related to any rifting events. The third area, and the main focus of this rewiew, is the central part of the Kenyan Rift, otherwise known as the Gregory Rift, and is defined as the zone of rifting between southern Lake Turkana (2”3O’N) and northern Tanzania (2”4O’S) (Fig, 1). In contrast to the two previous areas it is associated with crustal upwa~ing and is characte~sed by weakly alkaline to transitional basal@, phonolites and trachytes. Volcanism here was initiated around 20 Ma and rifting commenced post-12 Ma (Baker and Wohlenberg, 1971; Baker et al., 1972). As volcanism m the Gregory Rift preceded rift development by up to 10 Ma, it seems unlikely that it represents melt generation induced by decompression in response to regional crustal extension (e.g., White and McKenzie, 1989). Instead, the spatial and temporal relationship between pre-Miocene crustal uplift and ithe Miocene basalt and phonolite flood-style volcanism support an active model of rifting and point to the presence of a mantle plume beneath the Gregory Rift. A plume model is also supported by current geochemical modelling which requires that the volumes of basalt extrusives within the Kenya Rift as a whole, are associated with considerable volumes of partial melt within the lithosphere (Latin et al., 1993). Partial melts from this plume are thought to have risen up to the base of the crust and may be imaged by recent seismic refraction studies (e.g., Green et al., 1991). The scale of crustal uplift and the volumes of lava present in the Gregory Rift are several orders of magnitude smaller than those associated with Continental Flood Basalt provinces and active hot spots (e.g., White and McKenzie, 1989; Campbell and Griffiths, 1990). This suggests that the underlying thermal anomaly is likely to be a small convective cell within the mantle rather than a deep starting plume with a head diameter in excess of 1000 km. Nevertheless, if the above model is correct and rifting occurred in response to the upwelling of anomalously hot mantle material then a characteristic sequence of tectonic and volcanic events should result (e.g., White and
6
M. Smith / 7’ectonophysics236 (1994) 3-22
McKenzie, 1989; Hill, 1991; Hill et al., 1992). This paper summarises and highlights the key stratigraphic and structural features of the early development of the Gregory Rift which are pertinent to such a model.
2. Stratigraphical and volcanoiogicsl considerations Within the Gregory Rift, significant preserved sequences of sedimentary strata are rare and the existence of deep sedimentary basins remains unproven. Volcanic rocks predominate with estimated volumes in excess of l~,~ km3 (King, 1978) and their event stratig~phy is constrained by some 500 published and unpublished radiometric dates, many of which are contained in the theses and maps of the East African Geological Research Unit. In this section the evidence for pre-rift crustal uplift and the stratigraphic and volcanic features of two important phases of Miocene and Pliocene flood-style volcanism are considered. 2.1. The lyenya l)ome Laboratory studies of fluid dynamic analogs of mantle plumes (Campbell and Griffiths, 1990; Hill, 1991) and thermal modelling (White and McKenzie, 1989) predict that the ascent of hot and buoyant asthenospheric material will dynamically uplift the overlying lithosphere. The amount of crustal uplift is dependent upon a number of variables including the size, temperature and density contrasts between the plume, the surrounding asthenosphere, and the overlying lithosphere. Within the Gregory Rift the focus of Neogene volcanism, rifting, the regional Bouguer gravity anomaly (Bechtel et al., 19871, and the zone of anomalous mantle material (Painting and Maguire, 1990) are coincident with an area of crustal upwarping commonly termed the Kenya Dome. Gravity and isostatic studies (Bechtel et al., 1987) indicate that it is associated with only minor amounts of uplift and is in isostatic equilibrium, supported by the loading of anomalous mantle within the underlying lithosphere. It
should be noted that the term “Kenya Dome” is also used by some workers to refer to the topography of the volcanic a~cumuiations in the central part of the rift (e.g., Williams, f978a; King, 1978). The recognition of crustal uplift has previously relied upon the assumption that erosion surfaces in Kenya can be traced seaward into unconformities within the marine (post-Cretaceous) beds on the downwarped continental margin in eastern Kenya. The morphologic and altimetric correlation of weathered and often disparate topographic (or sub-volcanic) surfaces was used by Saggerson and Baker (1965) to document the gradual upwarping of the crust in the vicinity of the Gregory Rift. However, the simple connection between inland denudational uplift with coastal downwarp and the preservation of continental-wide erosion surfaces has been questioned by Summerfield (1985) and in Kenya is seriously undermined by recent apatite fission-track thermochronology (Foster and Gleadow, 1992). These data indicate that in the flank mountains northeast of the rift, up to 1.9 km of uplift occurred during Late Cretaceous times and that the Cretaceous erosion surface of Saggerson and Baker was deeply buried at this time. Foster and Gleadow further suggest that many of the reiict surfaces to the east of the Gregory Rift represent the upturned edges of tilted fauit blocks adjacent to a large buried Mesozoic rift system which underlies much of northeastern Kenya. Elsewhere, apatite fission-track studies (Wagner et al., 1992) suggest that crustal uplift adjacent to the Gregory Rift is variable and in many places insignificant as much of the basement that went through the apatite anneating zone has not yet reached the surface. Along the western margin of the rift, however, in the Loita and Cherangani areas, the total post-Cretaceous uplift is estimated to be in excess of 1.5 km (Wagner et al., 1992). Uplift attributable purely to Tertiary rifting events is everywhere difficult to determine. An alternative approach to quantifying pre-rift uplift is to consider the early patterns of sedimentation and volcanism. Indirect evidence for postCretaceous uplift in central and western Kenya is provided by the patterns of sedimentation and volcanism in northern Kenya. In the Turkana
M. Smith / Tectonophysics 236 (1994) 3-22
area, half-graben structures of Late Oligocene to Early Miocene age were progressively infilled by sandstones and grits derived from areas of uplifted Pre~~brian basement to the south and southwest (Morley et al., 1992). How much of this uplift is related specifically to the Kenya Dome or to more localised uplift across the adjacent Tanzanian craton is unknown. Pre-rift uplift is also implied by the palaeo-drainage reconstructions of King et al. (1972) and King (1978, fig. 3.2) who believed that the Kenya Rift was developed along an ancient watershed, largely uplifted during Palaeogene times. This is supported by the dispersion patterns and thickness variations within the phonolite lavas of the Gregory Rift (Fig. 2). Striking examples of topographic controi on these flows are shown by the Yatta phonolite which travelled down a palaeo-valley, southeast of the rift, for 250 km (Lippard, 1973a), and also by the phonolite flows which travelled westward, around the Tinderet-Timboroa volcanic complex, to infill the eastern part of a gentle downwarp (the proto-Nyanza Rift), and by those which flowed southwest across the Mara plains to the Isuria Plateau (Williams, 1964). When faulting and tilting effects are removed, the present-day outcrop of these lavas defines a low-relief, elongate shield 400 km in length, 200 km wide and with a total area of ca. 80,000 km2 (Fig. 2). In summary, the available evidence supports the presence of an elongate area of limited (< 1 km) crustal uplift which formed prior to the initiation of volcanism in the Gregory Rift. In the absence of onlapping relationships and/or prerift erosional unconformities the exact timing and amount of this uplift is difficult to establish. The current geological constraints indicate that the uplift is post-~retaceous in age and had more or less ceased by Early to mid-Mi~ene times. Within the rift there is no evidence to support significant amounts of Pliocene uplift and the fission-track data suggest that the thermal event associated with the Tertiary rifting is weak. Accepting the above, we can use the area1 extent of the uplift and initial volcanism to place first order constraints on the size of the mantle thermal anomaly prior to spreading (e.g., Campbell and Griffiths, 1990). Assuming widths of 200
I
km for the early volcanic province (Fig. 2) and a maximum of 300 km for the area of uplift, a diameter of loo-150 km for the initial plume beneath the Gregory Rift is indicated, The models of Campbell and Griffiths (1990) and others, postulate that crustal uplift (in excess of 500 m> should commence when a plume head reaches depths of one to two diameters below ‘the surface and this should occur lo-20 Ma befone the onset of volcanism. Within the Gregory Rifl initial volcanic activity is mainly dated between 16 and 11 Ma, although older dates of around 20 Ma also occur (Baker et al., 1971; Chapman and Brook, 1978). Thus the initiation and main phase of uplift should have occurred in the period 40-21 Ma. 2.2. Flood volcanism in the Gregory R@ As a plume buoyantly ascends, it will spread laterally and melt and entrain lower lithospheric material. Withdrawal of magma from the plume head and the penetration of the upper lithosphere by partial melt is indicated by the onset of sub-aerial fissure or flood-style volcanism (C~pbell and Griffiths, 1990; Hill, 1991). Depending on the pre-existing thermal state of the upper mantle and overlying crust into which the plume is ascending, this volcanism may be associated with subsidence and crustal extension. The Miocene Samburu Basalts and Plateau Phonolites, and the Pliocene ignimbrlte province of the central rift sector, represent the most volumetrically significant periods of extrusive activity in the Gregory Rift. The loci of eruptive centres associated with these rocks are broadly coincident with the axis of the Kenya Dome, and their distribution, internal stratigraphies and geochemical signatures provide important contlrols on processes taking place within the ascending plume and sub-rift lithosphere. The Samburu Basalts The Samburu Basalts (20-11 Ma) and their equivalents including the Elgeyo Basalts (Lippard, 19721, the Kapcherat Formation (McClenaghan, 1971; Webb, 1971) and the Nachola Formation (Tatsumi and Kimura, 1991) rest with
M. Smith / Tectonophysics 236 (1994) 3-22
Ll
I
I
+
*
N
Archaean basement
-3”N Basalt centre
50kn
i
Phonolite centre
.&
,Nachoia Fm.
-Limit of Flood (Plateau) phonolites
Fig. 2. Distribution of known eruptive centres for the Samburu Besalts and the Plateau Phonolites. Question marks represent possible eruptive centres for the Kapiti phonolite which flows westward from the rift to the Nairobi area. Compiled from various sources given in text.
M. Smith / Tectonophysics236 (1994) 3-22
marked unconformity on Precambrian rocks and locally on lacustrine-fluvial sediments which infill depressions in the underling to~graphy. Outcrops are confined mainly to the rift margins of the northern graben sector, extending as far south as the Lake Bogoria-Marmanet region (Fig. 2). They have a cumulative thickness in excess of 1150 m and their sheet-like form and association with dyke swarms and intrusive plugs indicate they are the products of fissure-type flood eruptions (Griffiths, 1977; Golden, 1979). Source areas for the Samburu Basalts are uncertain. Within the Baringo-Nakuru sector of the rift, the eroded remnants of a series of en-echelon displaced basalt shield volcanoes are associated with NNESSW-trending dyke swarms and fissures (Griffiths, 1977; Golden, 1979). Intraformational unconformities, pyroclastic deposits and lateral thickness variations also suggest a source in the Tiati area (Webb, 1971). Further south on the eastern margin of the central part of the rift, the Samburu Basalts also outcrop adjacent to the Sattima fault (McCall, 1967) and are probably contemporaneous with the Simbara Basalts (Shackleton, 1945) in the Aberdare Mts. (Fig. 2). The Pliocene age (5.5 f 1.0 Ma) usually quoted for the Aberdare volcanics (Baker et al., 1971) is based on a sample of basalt lava in the “Sagana river, south of Nyeri”. However, this locality lies within the Sirrima (or Laikipian) Basalts which are part of the Thomson’s Falls Phonolite Formation (McCall, 1967; Hackman, 1988) and the claim by Baker et al. (1971) that “the Aberdare volcanics overlie the Rumuruti and Thomson’s Falls Phonolites” cannot therefore be substantiated. The Aberdare Mts. are likely to be of Miocene age, and if the Sattima Phonolites are contemporaneous with the Plateau Phonolites (see below) then the Simbara Basalts must be older than 12 Ma. The Plateau Phonolites The eruption of the Plateau Phonolites represents the largest phase of extrusive volcanism in the Gregory Rift. Eruptions began in the early Middle Miocene (16 Ma) and continued into the Late Miocene (8 Ma) with the bulk of the lavas erupted between 13 and 11 Ma. Estimated total
9
volumes for the Plateau Phonolites range from 25,000 to 50,000 km3 (Williams, 197% Lippard, 1973a) with indi~du~ flows up to 2’70 m thick erupted over relatively short periods of time (Lippard, 1973a). The sheet-like rno~hol~~ and extent of individual flows suggest high extrusion rates of low-viscosity magma. A series of deeply eroded, low-relief phonolitic shields, intrusive plugs and domes located along and within the margins of the present-day rift have been identified as possible sources (Lippard, 1973b; Williams and Chapman, 1986) (Fig. 2). Further south the undated Mau Phonolite, up to 300 m’thick (Williams, 1990, and the Sattima Serieson the Aberdare Mts., indicate that the central sector of the rift was also covered by phonolites at this time. Subsidence and eruptive centres Previous reviewers envisaged that the voluminous basalt and phonolitic lavas progressively infilled an early rift depression. This is largely based on the stratigraphy exposed within the Elgeyo escarpment (Lippard, 1972) whiqh indicates the formation of a shallow basin, initiially infilled with fluvio-lacustrine sediments and Iater by the Uasin Gishu Phonolite. Later, the phonolite overstepped the depression before the main movements on the Elgeyo fault (post-12 Ma). In the Laikipia region an unconformity between the Samburu Basalts and the Plateau Phonolites (McCall, 1967; Golden, 1979) indicates downwarping and faulting in the interval 14-12 Ma. Thus, in the Baringo-Elgeyo area subsidence was broadly contemporaneous with volcanism. The distribution of known eruptive centres (Fig. 2) associated with this depression indicates a number of features. Firstly, that the ‘Initial eruptions were mainly focused in the Baringo-Elgeyo area, extending as far north as Tiati qnd perhaps as far south as Naivasha. Secondly, although the sources of the early lavas remain speculative (Williams and Chapman, 1986), the preserved eruptive centres are located primarily on the eastern margins of the present-day rift and were controlled by N-S- and NW-SE-trending fractures. Thirdly, no obvious pattern of eruptjve chronology has been determined. However, if the lavas of the Aberdare and Mau escarpments prove to be
of Mid-Late Miocene age (i.e. < 12 Ma) then a southerly progression is indicated as phonolitic volcanism did not commence in the southern sector (the Olenguluo Phonolites) until 7-8 Ma (Crossley and Knight, 1981). Chemistry Detailed geochemical data on the early volcanic sequences in the Gregory Rift are surprisingly sparse. The published data indicate that the eariy activity is atypical of plume-related flood volcanism containing a wide spectrum of rock types including, in order of importance, phonoIites, strongly alkaline basalts, picrites, ankaramites, melabasanites and subordinate amaunts of more evolved trachytes and mugearites (Baker, 1987). In Continental Flood Basalt provinces the production of voluminous tholeiitic basalts is thought to result from the entrainment of the surrounding mantle by the rising plume (Campbell and Griffiths, 1990). These basalts characteristically have isotopic signatures typical of Oceanic Island Basalts and are often associated with picrites. In the Gregory Rift, tholeiitic basalts are rare and no rocks with com~sitio~s unequivo~lly those of a primary magma have been discovered (Macdonald, 1994). Although the Samburu Basalts have initial Sr ratios similar to those of Ocean Island Basalts, they do not match these in detail (Macdonald, 1994). Instead the published traceelement and isotopic data indicate that both the basalts and phonolites were probably derived by fraetionation of an alkali basalt source (Lippard, 1973b; Gales, 1976; Hay and Wendlandt, 1992). The source area for this basalt is unknown and may lie within the lithosphere (Macdonald, 19941, within the asthenosphere (Latin et al., 19931, or both. The relative importance of lithospheric and crustal contamination and volatiles, especially CO, during the melting processes also remain poorly understood. Quoting experimental results from modal peridotite systems Wendlandt and Morgan (1982) suggest source depths of Xl-70 km for the Samburu Basalts and 40-50 km for the Flood Phonolites. Using settling times for phenocrysts and volume/ density estimates Gales (1976) con-
cluded that the phonolites were erupted shortly after formation and that a large volume (up to 500 x lo3 km31 of residual ultrabasic materia1 must exist within the lithosphere. Applying the volume approach of Karson and Curtis (1989) for extrusive volumes of 50,000 km3, a somewhat lower figure of 223 x lo3 km” of gabbro cumulate is indicated. These considerable volumes of ultrabasic material are presumed to occur as a series of reservoirs located at depths between 10 and 24 km beneath the rift (Gales, 1976; Hay and Wendlandt, 1992) and have been imaged by recent P-wave velocity seismic data as a series of sill-like intrusions (Henry et al., 1990). Pliocene ignimbrites Renewed faulting and subsidence during the Pliocene (5.3-1.6 Ma) coincided with a second phase of widespread trachyte and basalt fissure eruption. The marked contrasts in this volcanicity between the northern and central sectors of the Gregory Rift have been described by Smith and Mosley (1993). Within the central sector the exposed sections are dominated by explosive (phonolitic and trachytic) volcanic deposits. Thick (100-300 m) sequences of welded and non-welded ash-flow tuffs, and air-fall tuffs extend from the Equator as far south as Nairobi (Fig. 3) and broad correlations suggest that these deposits may have covered an area of at least 29,000 km2. The eruptive sources are unknown, but lateral facies and thickness variations suggest that the ash-flow tuffs were erupted within the Nakuru-NaivashaSuswa area (McCall, 1967; Baker et al., 1988; Williams, 1991). Individual flows, lo-50 m thick, travelled up to 70 km away from the central part of the rift. Assuming an average thickness of 100 m for the deposits outside the rift and a minimum rift infill of 3900 km3, total aggregate volumes in the order of 6000-7000 km3 and possibly up to 10,000 km3 are indicated. The age of these pyroclastic deposits is not known. LithologicalIy similar tuffs in the Equator area have yielded dates in the range 6.4-4.2 Ma (Jones and Lippard, 1979) and suggest that volcanism may have been initiated in the north. But most of the deposits, in the Mau, Bahati, Nairobi and Thika areas, correlate with the Kinangop
hf. Smith / Tect~~pkysics
234 (1994) 3-22
r 5.
Eldama
6. Kipting
Ravine Beds
Tuffs (0.3km).
Quaternary
Elementeita
KINANGOP
4.$Ma
(0.2km)
volcano
PLATEAU
Fig. 3. Dis~bution of Pliocene ignimbrites and air-fall tuffs in the central sector of the Gregory Rift. Possible eruptive centres indicated by hachured areas. Compifed from various sources including reports of the Geological Survey of Kenya abd theses of the East African Geological Research Unit. Cross-section along line A-A’; vertical exaggeration x2; Pliocene tuffsi shown by solid
I2
M. Smith / Tectonophysics 236 (1994) 3-22
Tuff, obsidian from which has yielded reliable K-Ar dates of 3.7-3.4 Ma (Baker et al., 1988) indicating a Late Pliocene event. If the above estimates are correct, then the central graben sector of the Gregory Rift must contain a significant volume of low-density pyroclastic deposits associated with large infilled calderacs) and/or multiple linear fissure systems. Structures on a scale comparable to the Toba caldera complex in northern Sumatra (e.g., Rose and Chesner, 1987) would not seem unreasonable. Large (35 km wide) tectono-magmatic structures have been proposed by Williams (1978b), Crossley (1980) and Baker et al. (1988) in the Kinangop (Gilgil) and Suswa areas. Supporting evidence is given by the regional gravity data; a large negative anomaly (- 230 mGa1) east of Lake Nakuru (MOE, 1987) may be interpreted as basalt or trachyte lavas overlain by 2-3 km of lower-density volcanic deposits including pumice and lapilli tuffs (Fig. 3, cross-section). Further south, the arcuate arrangement of faults in the Kijabe-Kedong area and the association of anomalously low-velocity delay times and high Bouguer gravity values in the Suswa area have been considered to indicate the remains of a large magma chamber at depth (Savage and Long, 1985). The proposed volumes also require either crustal melting or fractionation on a large scale. For a small plume, such as that postulated beneath the Gregory Rift, to produce these large volumes of trachyte by anatexis requires that the intruding asthenospheric mantle has an excessively high temperature and was able to reside at lower-crustal depths for a considerable period of time. At present, the available geochemical data do not support this hypothesis. Alternatively, if the magmas were derived by fractionation processes then significant volumes of residual ultrabasic cumulate should exist within the crust. Replenishment of the mid-crustal magma bodies beneath the central sector of the rift by fresh asthenospheric melt from the rising plume, and subsequent fractionation could account for the Pliocene trachytes and phonolites. But additional fractionation and/ or contamination at higher crustal levels cannot be excluded. In summary, the distribution of mid-Miocene
volcanic centres and lavas indicate eruption onto a low-relief topographic dome which had been uplifting throughout the Palaeogene. The area1 extent and amount of the uplift, and the volumes of the early flood lavas, are relatively small, suggesting that a small mantle plume underlies the Gregory Rift. The coincidence of volcanism and subsidence with the centre of the Kenya Dome suggests that initially, the axis of this plume was located below the central part of the rift where the E-W-trending Nyanza Rift forms a third arm to a tri-radial junction (Fig. 1). By mid-Miocene times as the plume spread laterally, thermal erosion and weakening of the overlying lithospheric mantle allowed decompression melting of the asthenosphere to take place. If the “dry solidus” for peridotite controls this melting (e.g., Latin et al., 1990) then the asthenosphere must have ascended to within 50-70 km of the surface. Small volumes of melts separated and rose up to near the base of the crust where they accumulated and interacted with the surrounding lithosphere. The presence of a high-velocity seismic layer (7.1 km s-‘> within the lower crust of the central rift segment (Green et al., 1991) may represent this zone of mafic magma accumulation. Magmas from this zone then rose up to mid-crustal levels to pond and fractionate in a series of magma bodies from which the first basalt and phonolite lavas were derived. The subsequent penetration of the asthenospheric diapir and the zone of partial melting into the lower crust is marked by widespread rifting and the second phase of flood volcanism during the Pliocene. In the central sector of the rift this culminated in a major ignimbritic event.
3. Structural considerations The mechanics of rifting in the Gregory Rift have been explained by two contrasting models. In the model of Bosworth (19891, extension is accommodated along major listric detachments of lithospheric extent, which alternate in polarity along the length of the rift. In the second model, rifting and crustal thinning occur in response to the forcible injection of asthenospheric material
M. Smith / Tectonophysics 236 (199413-22
(e.g., Mohr, 1987; Davis, 1991). The latter is supported by geophysicai data and current models argue for a combination of pure shear deformation in the lower crust and lithospheric mantle with mechanical stretching and faulting in the upper crust in response to external forces (e.g., Green et al., 1991). Whilst it is unlikely, from thermal and mechanical considerations, that the mantle plume beneath the Gregory Rift alone could drive rifting, the horizontal deviatoric stress generated by uplift, and the conductive transfer of heat above the plume head will signi~cantly affect the rheoiogy of the overlying lithosphere. This may serve to activate rifting or transfer the locus of extension to the plume centre. If a structurally weak zone exists in the pre-rift lithosphere above the plume head, then it may become a site of enhanced thermal erosion and extension, and aid the ascent of asthenospheric melts into the crust. In this section the evidence for pre-existing weaknesses in the lithosphere beneath the Gregory Rift is summarised and their influence on upper crustal structure briefiy considered. 3.1. Influence of lithospheric heterogeneity and pre-rift structure Some African Archaean cratons are known to contain a deep ( > 250 km) Iithospheric root comprising layered, high-velocity and chemically distinct mantle material. For example, ages of 3.3 Ga at depths of 250 km (Boyd et al., 1985) from isotopic data on garnet inclusions in diamonds indicate that the lithosphere beneath the Kaapvaal craton is old, thick and resistant to thermal erosion. In East Africa, low regional heat flow values (Nyblade et al., 1990) and the presence of diamondiferous kimberlites in northern and central Tanzania suggest that a similar deep and thermally resistant root may underlie parts of the Tanzanian craton. In contrast, based on presentday geothermal gradients, the surrounding mobile belt terrains are thought to be underlain by continental lithosphere, 150-100 km thick (e.g., Black and Liegeois, 1993). This lithosphere will have been modified by subsequent Phanerozoic orogenie and extensional events. The Gregory Rift is
13
located across a contact zone which separates these two iithospheric types. On the basis of morphology, structure and volcanic signature, Smith and Mosley (1993) subdivide the Gregory Rift into three riI?t segments which are separated by major ~-SE-trending zones of ductile shear and thrusting in the underlying Precambrian basement (Fig. 4). This segmentation is supported by the distribution of strongly alkaline carbonatitic volcanism which displays a characteristic time-independent zonation around the craton margin (Fig. $I (Le Bas, 1987 and refs. therein). In East Africa, olivinepoor nephelinitic and carbonatitic volcanism is largely restricted to areas of exposed Archaean rocks and here, the lithosphere beneath may be at least 90-100 km thick, as sugge&ted by the generation depths for carbonatites (e.g., Wendlandt and Morgan, 1982). In the Gregany Rift the presence of craton-type lithosphere beneath the southern segment is indicated by widespread nephelinitic-carbonatitic volcanism south of Lake Natron. Thus, the margin to the Tanzanian Archaean craton may extend under the southern Gregory Rift and its northern limit is marked by a major ductile zone of shearing and thrusting (the Aswa-Nandi-Loita shear zone) [Smith and Mosley, 1993). In the adjacent Loita-Mara area, the Archaean basement is thought to be effectively buried by gravitationally collapsed nappe structures. In contrast, the northern segment of the Gregory Rift north of Lake Baring0 has formed within mobile belt (Mozambique Belt) type crust. OIivine-rich nephe~initic volcanism is associated with large shield volcanoes and high-volume eruptions of alkali and sub-tholeiitic basalts containing mantle xenoliths. This indicates that here the underlying lithosphere is thinner and more easily eroded and penetrated by rising magmas. In northern and northeastern Kenya this lithosphere had previously undergone a significant phase of stretching and warming associated with Cretaceous and Palaeogene rifting events. Between these two crustal types /the central segment of the Gregory Rift overlies bn orogenitally thickened crust which marks t reworked and structurally dismembered f the Tan-
14
M. Smith / Tectonophysics 236 (1994) 3-22
zanian craton. This margin broadly trends NWSE and is bound by a linked system of ductile shear zones and thrusts, cut by later ductile/ brittle fractures and shears (Fig. 4). Thin-skinned thrust tectonics dominate and imbricate thrust stacks contain slices of Precambrian greenstone and ironstone strata; in addition, gneissic xenoliths found in the overlying Tertiary lavas have yielded Archaean isotopic ages (Smith and Mosley, 1993). These data suggest that the craton margin is largely underlain by Archaean crust. On a regional scale this margin closely correlates with a transitional or overlap zone, 75-100 km wide, containing both olivine-rich and olivinepoor nepheliniti~ volcanism (Fig. 5). In the Gregory Rift, the coincidence of flood volcanism and pre-rift uplift show that the axis of the rising mantle plume was focused under this marginal zone. Recent teleseismic and tomographic data from the central part of the rift (Davis, 1991; Achauer, 1992) have imaged an as~metri~ asthenosphere-lithosphere boundary with a steep westerly to southwesterly dipping contact under the western margin, extending down to depths of at least 200 km. At shallower crustal levels the preexisting framework of basement thrusts and shear zones is likely to have influenced rift development. Reactivation of the steep west-verging, NW-SE-, NNW-SSE- and N-S-trending ductile and brittle shear zones and thrusts along the craton margin, under suitable deviatoric stress conditions, may control the location and orientation of rift basins. This reactivation would have been particularly important during the early stages of rifting when the thermal effects of the rising plume were limited and the ductile/ brittle transition boundary was deep. 3.2. Half-graben basins and transfer tones The sinuous course of the Gregory Rift (Fig. 4) is often attributed to the en-echelon ar~nge.m~~t of faults defining a series of asymmetric halfgraben (Bosworth et al., 1986; Bosworth, 1989). Because the constraints on the development and timing of the early faults and graben structures
are few, this model remains controversial (Mohr, 1987; Morley, 1988). In fact, the asymmetric graben model can only be applied with confidence in the Elgeyo-Baring0 area where deep dissection and good stratigraphic and geophysical control allows a number of basins, ranging in age from Miocene to Pleistocene, to be distinguished. But even here the depth of basin infili and interpretation of the Bouguer gravity anomalies remain contentious (e.g., Swain, 1992). Further south, previous workers also considered that the early half-graben structures formed on the western margin of the rift (e.g., Chapman et al., 1978; Jones and Lippard, 1979; Crossley, 1979; Baker et al., 1988). Within the central sector, the relative importance of the Mau and the Sattima boundary faults is unresolved. The Sattima fault scarp cannot be traced south of the Kipipiri volcano and shows no evidence of displacement pre4 Ma. The Mau escarpment has been suggested by several workers to be the major boundary fault, but unlike other half-graben boundary faults, it appears to arc in the opposite direction to that expected, and is not as clearly defined, being composed of a series of short en-echelon faults. In addition, the available gravity data do not indicate the existence of buried basins in front of the Mau escarpment, although data resolution is hampered by a low station density, particularly on the rift margin, and the effects of shallow intrusions. In extensional terrains, adjacent rift basins and their bounding faults often terminate or are deflected into parallelism with discrete transverse zones characterised by en-echelon overlapping faults with oblique-slip and/ or strike-slip motions. These transfer or accommodation zones usually lie parallel to major weaknesses in the pre-rift basement. Transfer zones within the Gregory Rift have yet to be described in detail. The original zones postulated by Bosworth (Fig. 6A) have littIe surface expression and some trend at a high angle to known basement structures. Alternative transfer zones proposed by Smith and Mosley (1993) (Fig. 6B) emphasise the importance of NW-SE-trending lineaments on rift geometry and are distinguished by pronounced deflections and offsets in the regional trends of the
15
M. Smith / Tectonophysics 236 (1994) 3-22
Proterozoic mobile belt
A
thrust or shear with thrust sense
I
major rift fault
Fig. 4. Tect on& setting of the Gregory Rift. Rift segments and faults shown in relation to major basement. Iiimplified from Smith and Mosley (1993, Fig. 2).
StnMUreS
in 1the
rift margin faults (e.g., at Bogoria, Kedong and the northern end of the Nguruman fault). The fault patterns and transfer zones shown in Fig. 6B suggest that the Gregory Rift is characterised by small rift basins with lengths of be-
tween 30 and 50 km and length/width ratios of 4-5. The largest of these, the Kerio Basin. from gravity modefling and fault displacement data Ilippard, 19721, is 3-4 km deep and 40-50 km in length. This contrasts with other rift settings, for
Otivine-rich nephelinitic volcanic centrss
I
A
Mixed neph~linitic volcanic centres Oiivine-poor nephelinitic volcanic centres Zone of overtap of magmatic provinces
0
lOOkIs I
1
MT. KENYA
N
t
A CHYULUS OL DOINYO LE
lake
Evasi
2
Fig. 5. Distribution of nephelinitic and carbonatitic volcanism along the northeastern margin of the Tanzanian craton. Diagram modified from Le Bas (1987, Fig. 3). Age (Ma) of volcanic centres, where known, indicated by numbers in brackets. Outline of Gregory Rift superimposed and indicated by fault pattern.
M. Smith / Tectonophysics 236 (I 994) 3-22
example, the Turkana rifts (Morley et al., 1992) and the Western Rift, where rift basins are 50-100 km in length and have depths of 5-10 km. When combined with the evidence for low rates of extension (i.e., < 10 km), (King, 1978; Henry et al., 1990) this suggests that half-graben structures in the Gregory Rift are shallow crustal features and are not associated with significant amounts of lithospheric stretching. To summarise, tectonism in the Gregory Rift is restricted to a narrow volcanic and sediment-filled trough which formed atop a broad crustal arch in mid-Miocene times. This restriction of volcanism and tectonism to a narrow zone is consistent with the dyke injection model. During the Miocene, as the mantle diapir released partial melts into the lithosphere, fault slip above and in front of the rising wedge led to the formation of an axial trough at the surface. By Late Miocene times the upper lithosphere had become sufficiently weak-
Fig. 6. (thrusts Mosley outline
17
ened to respond to the regional stress field and rifting commenced along a series of major bounding faults in the period 10-8 Ma. The consistent initiation of these early faults along the western margin of the trough has been related to the reactivation of weaknesses in the Precambrian crust under an E-W-oriented stress field (Smith and Mosley, 1993). During the Pliocene, as the lithosphere progressively weakened and stretched, second order transfer zones formed within the rift and probably controlled the formation and propagation histories of border faults, As the rift developed, subsidence and graben formation propagated southwards into northern Tanzania and northwards to Lake Turkana. The younger Pleistocene-Recent basins show less influence of basement control and were developed across a more heterogeneous crust that was being warmed and intruded by significant volumes of magma. Overall, the locus of rifting has naarowed with
Pattern of linked half-graben structures and transfer zones in relation to craton margin (stippled) and and shears) in the Precambrian basement (symbols as for Fig. 4); (A) after Bosworth et al., (19861, (B) (1993). KBM = Kerio-Bogoria-Marmanet transfer zone, EML = Engorika-Magadi-Lembolos transfer of Kenya Dome indicated by broken line.
18
M. Smith / Tectonophysics
time and migrated inwards from the rift margins to the floor of the inner trough and is accompanied by an increase in fault density and a decrease in fault length and displacement. This inward contraction with time differs from the apparent eastward migration of rift basins observed in the Turkana area (Morley et al., 1992; Morley, 1994). 4. Discussion The active model of rifting provides a useful explanatory framework for the evolution of the Gregory Rift. Extension was probably initiated by the dynamic uplift associated with the emplacement of a small mantle plume at the base of a heterogeneous lithosphere subject to a weak regional stress. The limited amounts of uplift associated with the Kenya Dome, the early rift subsidence and the limited heating inferred from the apatite fission-track data are consistent with the underlying asthenosphere having temperatures of about 100°C above normal (138O”Cl (White and McKenzie, 1989). The pre-rift history of the lithosphere plays an important role in locating magmatic and tectonic activity. It has been shown that within the Gregory Rift the coincidence of location of magmatic activity with a pre-existing crustal weakness is strong. Rifting was initiated above a basement shear system marking the contact between the Archaean craton and the Proterozoic mobile belt and the location of the early rift basins and their linking transfer zones was influenced by a framework of large ~-SE-trending shear zones. The lateral spreading of the rising plume under, and into, the adjacent lithospheric types is indicated by the diachronous development of rift basins and by the timing of the initiation of volcanicity at the surface. The mechanics and style of this propagation provide a focus for future studies. The clearest evidence for a plume is generally the presence of magma derived from anomalously hot mantle, but unfortunately in the Gregory Rift relevant data are currently unavailable. Petrochemical data, including rare earth and isotopic analyses are required for the Samburu Basalts and the Plateau Phonolites to place meaningful
236 (19941 3-22
constraints on the source, nature and interactjon of the ascending plume with the sub-rift lithosphere. In particular, the juxtaposition of compositionally and mechanically contrasting lithospheric types suggests that there may be significant differences in the magmatic processes, rates of melt production, and the ease with which asthenospheric melts can penetrate the lithosphere, along the length of the rift. The degree of extension and thickness of the pre-rift lithosphere or mechanical boundary layer, will influence the amount of melting within the plume head. In the Gregory Rift, extension and lithospheric stretching appear to have been limited. This, plus a thick mechanical boundary layer, could account for the general absence of tholeiitic volcanism. Continental flood tholeiites may be produced when the lithosphere is thinned by a factor > 2, (White and McKenzie, 19891. The presence of lavas approaching tholeiitic compositions in the northern Turkana Rift and in southern Ethiopia suggests that a thinner mechanical boundary layer, disturbed by previous extensional events, underlies these rifts. This is supported by the higher amounts of total extension (35-40 km) (Morley et al., 1992) and present-day thin (20 km) crust. Thus, slow extension of relatively thick lithosphere beneath the Gregory Rift accumulated (or underplated) partial melts within the lithosphere and contamination and fractionation processes dominate the subsequent early extrusive rocks, whereas high extension rates and thin lithosphere in the Turkana area favoured larger volumes of high-temperature partial melts which were able to ascend rapidly through the crust with limited fractionation. In other rifts (e.g., North Sea) rates of extension and stretching histories are well constrained and can be used to estimate the amount of partial melt generated during rifting. However, in the Gregory Rift the application of depth-todetachment calculations and section balancing techniques is severely restricted by the large volumes of magma, in the form of both dykes and sills, present at all levels in the crust. Rates of extension remain problematic, and partial melt volumes should be better predicted from petrological data.
19
M. Smith / Tectonophysics 236 (1994) 3-22
w
E axial dyke swarm
Nyambeni Mts.
0
50 Archaean crust 5
lal
E
1 150 0
Partial melt (asthenospheric)
200 0
Mt Kenya
depleted
*
magma body
shear zone
mantle root
250
Fig. 7. A 4-stage evolutionary cartoon for the Gregory Rift. See text for explanation of crustal structure beneath draton margin. Solid arrows indicate regional extension, fine solid lines indicate circulating mantle.
h4. Smith / Tectonophysics
The main features of the active model of rifting pursued in this paper are summarised in Fig. 7 and outlined below. (I) Pre-Miocene. A mantle thermal perturbation rises up the side of the lithospheric root beneath the Tanzanian craton. The source area of this plume is unknown; it may lie beneath the Tanzanian craton or, alternatively, the timetransgressive nature of the initiation of volcanism and rifting from southern Ethiopia (45-35 Ma) through to northern Tanzania (< 5 Ma) suggests that its source could lie to the north in Ethiopia. When this thermal anomaly or plume reaches depths of less than 150 km, dynamic uplift commences at the surface to form the Kenya Dome. The nature and configuration of the pre-rift lithosphere shown in Fig. 7-l are speculative. Ashwal and Burke (1989) have suggested that during Tibetan-style collisional orogenesis tectonically thickened mantle lithosphere may be delaminated or eroded and replaced by more “fertile” asthenospheric mantle, isotopically enriched and less depleted in crustal or basaltic components. When this “fertile” mantle is heated, it readily melts and may provide a source of enriched magma during Cenozoic rifting. It should be noted, however, that the Ashwal and Burke model is only one of many (cf. Black and Liegeois, 1993) and that current petrological models of basalt generation in the Gregory Rift (e.g., Macdonald, 1994) do not support the presence of enriched asthenospheric mantle sources at depth. (2) Miocene. The asthenospheric plume intrudes into the upper levels of the mantle lithosphere. This coincides with the culmination of uplift as the plume spreads and partial melts separate out and accumulate below the base of the crust. Magmas from this zone of underplating rise up into the crust where they form mid-crustal reservoirs which fractionate to produce basalt and phonolite flood volcanism at the surface. Initially, an axial trough forms in response to the rising asthenospheric wedge, collapsing later in the period 12-7 Ma to form an asymmetric rift structure. (3) Pliocene. Asthenospheric melts reach the base of crust (35-40 km), and dyke injection and lateral propagation initiate a new phase of vol-
236 (1994) 3-22
canism and faulting. Replenishment of midcrustal magma bodies produces a new phase of fissure eruptions marked in the central graben sector by extensive pyroclastic eruptions. Extension and basin subsidence within the rift is accomplished by a combination of magma injection and lithospheric stretching in response to the regional stress field. As the plume spreads laterally away from the craton margin any inherent weaknesses or thin zones within the mobile belt lithosphere are likely to be exploited and will provide fresh sites for thermal erosion and renewed melting. The presence of off-axis volcanism, represented by Mt. Kenya and the Pliocene-Quaternary basalt shields of northern Kenya, recent microseismic activity, and anomalous seismic velocities indicating partial melt at depth, point to such a zone some 100 km east of the Gregory Rift. (4) Pleistocene to Recent. Partial melt from the plume head reaches depths of ca. 30 km and is marked by the inward focusing of volcanism and rifting into the inner trough. Dyke injection to shallow crustal levels (l-5 km) accommodates most of the extension leading to inward collapse and formation of the inner trough. Acknowledgements
I thank Peter Mosley for many hours of fruitful discussion, Peter Dunkley, Malcom Howells, Ray Macdonald and Manfred Strecker who provided useful comments on an early draft, and Caron Simpson who drafted the figures. The manuscript also benefited from reviews by two anonymous referees. This paper was written whilst the author was involved in a programme of exploration for geothermal energy in Kenya supported by the Overseas Development Administration of the United Kingdom. Publication is by permission of the Director, British Geological Survey (NERC). References Achauer, U., 1992. A study of the Kenya rift using delay-time tomographic analysis and gravity modelling. Tectonophysics, 209: 197-207.
M. Smith/ Tectoncphysics236 (199413-22 ~der~n, D.L., Toshiro, T. and Zhang, Y., 1992. PIate tectonics and hot spots: The third dimension. Nature, 256: 1645-1651. Ashwal, L.D. and Burke, K., 1989. African iithospheric structure, volcanism and topography. Earth Planet. Sci. Lett., 96: S-14. Baker, B.H., 1987. Outline of the petrology of the Kenya rift aikaline province. In: 3.G. Fitton and B.G.J. Upton CEdito&, Alkaline Igneous Rocks. Geol. Sot. London, Spec. Publ., 30: 293-311. Baker, B.H. and Wohlenberg, J., 1971. Structure and evolution of the Kenya Rift Valley. Nature, 229: 538-542. Baker, B.H., Wi~iiams, L&J., Miller, J.A. and Fitch, F.L., 1971. Sequence and geochronoIogy of the Kenya rift volcanics. Tectonophysics, 11: 191-215. Baker, B.H., Mohr, P.A. and Williams, L,A.J., 1972. Geology of the eastern rift system of Africa. Geol. Sot. Am., Spec. Pap., 136: l-67. Baker, B.H., Mitchell, J.G. and Williams, L.A.J., 1988. Stratigraphy, geochronology and volcano-tectonic evolution of the Kedong-Naivasha-~n~gop region, Gregory rift vatfey, Kenya. J. Geof. See. London, 145: 107-116. Bechtel, T.D., Forsyth, D.W. and Swain, C.J., 1987. Mechanisms of isostatic compensation in the vicinity of the East African Rift, Kenya. Geophys. J.R. Astron. Sot., 90: 445465. Black, R. and Liegeois, 3.-P.> 1993. Cratons, mobile belts, alkaline rocks and ~ontinenta1 lithosphere mantle: the Pan-African testimony. J. Geol. Sot. London, 150: 89-98. Bosworth, W., 1989. Basin and range style tectonics in East Africa. J. Afr. Earth Sci., 8: 191-201. Bosworth, W., Lamb&e, 3.J. and Keisler, R., 1986. A new look at Gregory’s rifb the structural style of continental rifting. EOS, Trans. Am. Geophys. Union, 67: 577, 582, 583. Boyd, F.R., Gurney, J.3. and Richardson, S.H., 1985. Evidence for a 150-200 km thick Archaean lithosphere from diamond inclusion thermobarometry. Nature, 315: 387389. Burke, K. and Whiteman, A.J., 1973. Uplift, rifting and the break-up of Africa. fn: D.H. Tarhng and SK. Runcorn ~~ito~~, Implications of Continental Drift to the Earth Sciences, 2: 735-755. Campbell, I.H. and Griffiths, R.W., 1990. Implications of mantle plume structure for the evolution of flood basalts. Barth Planet. Sci. Lett., 99: 79-93. Chapman, G.R. and Brook, M., 1978. Chronostratigraphy of the Baring5 basin, Kenya. In: W.W. Bishop @ditor), Geological Background to Fossil Man, Geological Society of London, pp. 207-223 Chapman, G.R., Lippard, S. and Martyn, J.E., 1978. The stratigraphy and structure of the Kamasia range, Kenya Rift valley. J. Geol. Sot. London, 135: 265-281. Crossley, R., 1979. The Cenozoic strat~~phy and structure of the western part of the rift valley in southern Kenya. 3, Geof. Sot. London, 136: 393-405. Crossley, R., 1980. Structure and volcanism in the S. Kenya
21
rift. In: Ge~~amic Evolution of the Afro-Arabic Rift System. Acad. Nazionale dei Lincei, Rome+ pp. 89-98. Crossley, R. and Knight, R.M., 1981. Volcani&n in the western part of the rift valley in southern Kenya. Bull. Volcanal., 44: 117-128 Davis, P.M., 1991. Continental rift structures and dynamics with reference to teleseismic studies of the Rio Grande and East African rifts. Tectonophysic~ 197’ 309-325, Foster, D.A. and Gleadow, A.J.W., 1992. The ~orphotectonic evolution of rift-margin mountains in centjal Kenya: constraints from apatite fission-track thermochronology. Earth Planet. Sci. Lett., 113: 157-171. Gird&, R.W., Fairhead, SD., Searle, R.C. aind ~werbut~, W.T.C., 1969. Evolution of rifting in Afrio!a. Nature, 224: 1178-1182. Golden, M., 1979. The Geology of the Area,East of Silale, Rift Valley Province, Kenya. PhD. thesis, Univ. of Iondon. Gales, G.G., 1976. Some constraints on the origin of phonolites from the Gregory rift, Kenya, and inferences concerning basahic magmas in the rift system. Limos, 9: l-8. Green, W.V., Achauer, U. and Meyer, R.P., 1991. A three-dimensional seismic image of the crust and upper mantle beneath the Kenya Rift. Nature, 354: 1991203. Griffiths, P.S., 1977. The Geology of the Are11 Around Lake Hannington and the Perkerra River, Rift Valley Province, Kenya. PhD. thesis, Univ. of London. Hackman, B.D., 1988. Geology of the Bar~ngo~-Laikipia Area. Report 104, Mines and Geology Department, Nairobi, 79 PP. Hay, DE. and Wendlandt, R.F., 1992. experimental results bearing on the origin of Kenyan rift flood phonolites. EC& Trans. Am Geophys. Union, 73(14): 337. Henry, W.J., Mechie, J., Maguire, P.K.H.:, Khan, M.A., Prodehl, C., Keller, G.R. and Patel, J., 1990. A seismic investigation of the Kenya rift valley. Geophys. J. Int., 100: 107-130. Hill, R.I., 1991. Starting plumes and continental break-up. Earth Planet. Sci. Lett., 104: 398-416. Hill, RI,, Campbell, I.H., Davies, G.F. and Griffiths, R.W., 1992. Mantie plumes and continental tedtonics. Science, 256: 186-193. Jones, W.B. and Lippard, S.J., 1979. New age determinations and the geology of the Kenya rift-Kavirondo rift junction, W. Kenya. 3. Geol. Sot. London, 136: 693*-704. Karsan, J.A. and Curtis, PC., 1989. TectoniC: and magmatic processes in the eastern branch of the east,African rift and implications for magmatically active Continental rifts. J. Afr. Earth Sci,, X: 431-453. King, B.C., 1978. Structural evolution of the Gregory rift valley. In: W.W. Bishop (Editor), Geologtcal Background to Fossil Man. Geological Society of London, pp. 29-54. King, B.C., Le Bas, M.J. and Sutherland, D.S., 1972. The history of the alkaline volcanoes and intmsive complexes of eastern Uganda and western Kenya_ J. Geol. Sot. London, 128: 173-205. Latin, D.M., Dixon, J.E., Fitton, J.G. and White, N., 1990.
22
M. Smith / ~ec~~~o~h~sics 236 (1994) 3-22
Mesozoic magmatic activity in the North Sea basin: imphcations for stretching history. In: R.F.P. Hardman and J. Brooks (Editors), Tectonic Events Responsible for Britain’s Oil and Gas Reserves. Geol. Sot. London, Spec. Publ., 55: 207-227. Latin, D.M., Norry, M.J. and Tarzey, R.J.E., 1993. Magmatism in the Gregory Rift, East Africa: evidence for melt generation by a plume. J. Petrol., 34: 1007-1027. Le Bas, M.J., 1987. Nephelinites and carbonatites. In: J.G. Fitton and B.G.J. Upton (Editors), Alkaline Igneous Rocks. Geol. Sot. London, Spec. Publ., 30: 53-83. Lippard, S.J., 1972. The Stratigraphy and Structure of the Elgeyo Escarpment, Southern Kamasia Hills and Adjoinin8 Region, Rift Valley Province, Kenya. PhD. thesis, Univ. of London. Lippard, S.J., 1973a. Plateau phonolite lava flows, Kenya. Geol. Mag., 110: 543-549. Lippard, S.J., 1973b. The petrology of phonolites from the Kenya rift. Lithos, 6: 217-234. Logatchev, N.A., 1974. Volcanism and tectonics of the Kenya Rift zone. Geotectonics, 3: 163-173. Macdonald, R., 1994. Petrological evidence regarding the evolution of the Kenya Rift Valley. In: C. Prodehl, G.R. Keller and M.A. Khan (Editors), Crustal and Upper Mantle Structure of the Kenya Rift. Tectonophysics, 236: 373390. McCall, G.J.H., 1967. Geology of the Nakuru-Thomson’s Falls-Lake Hannington area. Rep. Geol. Surv., Kenya, No. 78. McClenaghan, M.P., 1971. The Geology of the Ribkwo Area, Baringo District, Kenya. PhD. thesis, Univ. of London. MOE, 1987. 1.0 M Geological Map of Kenya with Bouguer Gravity Contours. Ministry of Energy, Kenya. Mohr, P., 1987. Structural style of continental rifting in Ethiopia: reverse decollements. EOS, Trans. Am. Geophys. Union, 68: 721, 729, 730. Morgan, P. and Baker, B.H., 1983. Introduction-Processes of continental rifting. Tectonophysics, 94: l-10. Morley, C.K., 1988. Comment and reply on “Off-axis volcanism in the Gregory rift, east Africa: implications for models of continental rifting”. Geology, 16: 569. Morley, CR, 1994. Interaction of deep and shallow processes in the evolution of the Kenya rift. In: C. Prodehl, G.R. Keller and M.A. Khan (Editors), Crustal and Upper Mantle Structure of the Kenya Rift. Tectonophysics, 236: 8191. Morley, C.K., Wescott, W.A., Stone, D.M., Harper, R.M., Wigger, S.T. and Karanja, F.M., 1992. Tectonic evolution of the northern Kenyan rift. J. Geol. Sot., 149: 333-348. Nyblade, A.A., Pollack, H.N., Jones. D.L., Podmore, F. and Mushayandebvu, M., 1990. Terrestrial heat flow in east and southern Africa. J. Geophys. Res., 95: 17,371-17,384. Pointing, A.J. and Maguire, P.K.H., 1990. A seismic velocity model for the upper mantle in northern Kenya derived from teleseismic earthquake data. J. Afr. Earth Sci.. 11: 391-399. Rose, WI. and Chesner, CA., 1987. Dispersal of ash in the great Toba eruption, 75 ka. Geology, 15: 913-917.
Saggerson, E.P. and Baker, B.H., 1965. Post-Jurassic erosionsurfaces in eastern Kenya and their deformation in relation to rift structure. Q.J. Geol. Sot. London, 121: 51-72. Savage, J.E.G. and Long, R.E., 1985. Lithospheric structure beneath the Kenya dome. Geophys. J.R. Astron. Sot., 82: 461-477. Shackleton, R.M., 1945. Geology of the Nyeri area. Rep. Geol. Surv., Kenya, No. 12. Smith, M. and Mosley, P., 1993. Crustal heterogeneity and basement influence on the development of the Kenya Rift, East Africa. Tectonics. 12: 591-606. Summerfield, M.A., 1985. Plate tectonics and landscape development on the African continent. In: M. Morisawa and J.T. Hack (Editors), Tectonic Geomorphology. (Proccedings of the 15th Annual Binghamton Geomorphology Symposium, September 1984). Allen and Unwin, London, pp. 281-294. Swain, C.J., 1992. The Kenya rift axial gravity high: a re-interpretation. Tectonophysics, 204: 59-70. Tatsumi, Y. and Kimura, N., 1991. Secular variation of basalt chemistry in the Kenya Rift: evidence of the pulsing of asthenospheric upwelling. Earth Planet. Sci. Lett., 104: 99-113. Wagner, M., Ahherr, R. and Van den Haute, P., 1992. Apatite fission-track analysis of Kenyan basement rocks: constraints on the thermotectonic evolution of the Kenya dome. A reconnaissance study. Tectonophysics. 204: 93110. Webb, P.K., 1971. The Geology of the Tiati Hills, Rift Valley Province, Kenya. PhD. thesis, Univ. of London. Wendlandt, RF. and Morgan, P.. 1982. Lithospheric thinning associated with rifting in East Africa. Nature, 298: 734-736. White, R. and McKenzie, D., 1989. Magmatism at rift zones: the generation of volcanic continental margins and flood basalts. J. Geophys. Res., 94: 768x5-7729. Williams, L.A.J., 1964. Geology of the Mara River-Sianna area. Rep. Geol. Surv., Kenya, No. 66. Williams, L.A.J., 1972. The Kenya rift volcanics: a note on volumes and chemical composition. Tectonophysics, IS: 83-96. Williams, L.A.J., 197&a. The volcanological development of the Kenya rift. In: E.R. Neumann and LB. Ramberg (Editors), Petrology and Geochemistry of Continental Rifts. D. Reidel, Dordrecht, pp. 101-121. Williams, L.A.J., 1978b. Character of Quaternary volcanism in the Gregory Rift Valley. In: W.W. Bishop (Editor), Geological Background to Fossil Man. Geological Society of London, pp” 55-69. Williams, L.A.J., 1991. Geology of the Mau area. Rep. Geol. Surv., Kenya, No. 96. Williams, L.A.J. and Chapman, G.R.. 1986. Relationships between major structures, salic volcanism and sedimentation in the Kenya rift from the equator northwards to Lake Turkana. In: L.E. Frostick. R.W. Renaut, 1. Reid and 3.5. Tiercelin (Editors), Sedimentation in the African Rifts. Geol. Sot. London, Spec. Publ., 25: 59-74.