Steady rifting in northern Kenya inferred from deformed Holocene lake shorelines of the Suguta and Turkana basins

Earth and Planetary Science Letters 331–332 (2012) 335–346

Contents lists available at SciVerse ScienceDirect

Earth and Planetary Science Letters journal homepage: www.elsevier.com/locate/epsl

Steady rifting in northern Kenya inferred from deformed Holocene lake shorelines of the Suguta and Turkana basins Daniel Melnick a, b,⁎, Yannick Garcin a, b, Javier Quinteros c, Manfred R. Strecker a, b, Daniel Olago d, Jean-Jacques Tiercelin e a

Universität Potsdam, Institut für Erd- und Umweltwissenschaften, 14476 Potsdam, Germany DFG Leibniz Center for Surface Process and Climate Studies, 14476 Potsdam, Germany Deutsches GeoForschungsZentrum Potsdam, 14473 Potsdam, Germany d University of Nairobi, Department of Geology, 30197-00100 Nairobi, Kenya e UMR CNRS 6118 Géosciences Rennes, Université de Rennes 1, Rennes, France b c

a r t i c l e

i n f o

Article history: Accepted 2 March 2012 Available online 1 April 2012 Editor: P. DeMenocal Keywords: continental rifting East Africa lake shorelines Holocene extension isostatic rebound

a b s t r a c t A comparison of deformation rates in active rifts over different temporal scales may help to decipher variations in their structural evolution, controlling mechanisms, and evolution of sedimentary environments through time. Here we use deformed lake shorelines in the Suguta and Turkana basins in northern Kenya as strain markers to estimate deformation rates at the 10 3–104 yr time scale and compare them with rates spanning 101–10 7 yr. Both basins are internally drained today, but until 7 to 5 kyr lake levels were 300 and 100 m higher, respectively, maintained by the elevation of overflow sills connecting them with the Nile drainage. Protracted high lake levels resulted in formation of a maximum highstand shoreline — a distinct geomorphic feature virtually continuous for several tens of kilometers. We surveyed the elevation of this geomorphic marker at 45 sites along > 100 km of the rift, and use the overflow sills as vertical datum. Thin-shell elastic and thermomechanical models for this region predict up to ~ 10 m of rapid isostatic rebound associated with lake-level falls lasting until ~ 2 kyr ago. Holocene cumulative throw rates along four rift-normal profiles are 6.8–8.5 mm/yr, or 7.5–9.6 mm/yr if isostatic rebound is considered. Assuming fault dips of 55–65°, inferred from seismic reflection profiles, we obtained extension rates of 3.2–6 mm/yr (including uncertainties in field measurements, fault dips, and ages), or 3.5–6.7 mm/yr considering rebound. Our estimates are consistent, within uncertainties, with extension rates of 4–5.1 mm/yr predicted by a modern platekinematic model and plate reconstructions since 3.2 Myr. The Holocene strain rate of 10 − 15 s− 1 is similar to estimates on the ~ 10 6 yr scale, but over an order of magnitude higher than on the ~ 107 yr scale. This is coherent with continuous localization and narrowing of the plate boundary, implying that the lithospheric blocks limiting the Kenya Rift are relatively rigid. Increasing strain rate under steady extension rate suggests that, as the magnitude of extension and crustal thinning increases, the role of regional processes such as weakening by volcanism becomes dominant over far-field plate tectonics controlling the breakup process and the transition from continental rifting to oceanic spreading. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Deformation rates of the Earth's crust constitute an observational basis for physical models of various processes associated with plate motions, including mountain building, the earthquake cycle, and rifting of continental crust. At divergent plate boundaries, the width of the deforming region may vary as well as the rates at which it stretches (e.g., Molnar, 1988). Continental rifts represent the initiation of divergent plate boundaries, and may evolve into continental break-up, ultimately resulting in new ocean basins (e.g., McKenzie

⁎ Corresponding author. Tel.: + 49 331 977 6252; fax: + 49 331 977 5700. E-mail address: [email protected] (D. Melnick). 0012-821X/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.epsl.2012.03.007

et al., 1970). A crucial aspect to understanding the continental break-up process is the dynamic link between far-field forces exerted by plate tectonics, regional stresses that may arise from mantle plumes, and local mechanical aspects such as fault weakening, inherited rheological discontinuities, and volcanism (e.g., Buck, 2004). This link is likely to dictate the rate and style of rifting. The role of magmatic versus tectonic processes in continental rifting has been the matter of debate. Recent volcano-tectonic episodes associated with rifting in regions with thin (Abdallah et al., 1979; Wright et al., 2006) as well as thick (Calais et al., 2008) lithosphere suggest that dyke intrusions accounted for most of the strain. Conversely, numerical modeling indicates that continental lithosphere may be broken without volcanism, controlled by shear heating (Regenauer-Lieb et al., 2008). In the East African Rift System (EARS)

336

D. Melnick et al. / Earth and Planetary Science Letters 331–332 (2012) 335–346

frequent, low-magnitude seismicity has been associated with dyke intrusions (e.g., Tongue et al., 1994), but fault ruptures during large earthquakes have occurred without coupled volcanism (e.g., Ambraseys, 1991; Biggs et al., 2010; Zielke and Strecker, 2009). Thus, despite of its overall magmatic character, strain release in the EARS is not necessarily associated with dyke intrusions and magmatism. Magma-assisted rifting in the advanced northernmost EARS appears to be efficient (e.g., Wright et al., 2006), but it is not clear how the less advanced sectors farther south behave, and if rates have changed over time. While sea-floor magnetic anomalies reveal that oceanic spreading rates may be sustained over ~10 7 yr (e.g., Demets et al., 1990), our understanding of deformation rates during continental rifting has remained limited, mainly because of the lack of suitable strain markers at various temporal scales. In an attempt to quantify extension rates and gain insight into strain localization processes at a nascent divergent plate boundary, we studied Holocene deformation within the eastern branch of the EARS in northern Kenya (Fig. 1). Independent estimates of extension in this region have been obtained at the 10 6–10 7 yr scale from balanced cross sections (Hendrie et al., 1994), as well as reconstruction of oceanic magnetic anomalies and transform azimuths (Chu and Gordon, 1999), and at the 10 1 yr scale from geodesy and seismology (Calais et al., 2006). Recently, Stamps et al. (2008) quantified regional deformation rates for East Africa by jointly inverting 3.2-Myr average spreading patterns, GPS velocities, and earthquake slip vectors, finding that extension rates are consistent over 10 1–10 6 yr scales. In order to bridge the gap between extension on 10 1 yr and 10 6 yr scales in the northern Kenya Rift, we estimated extension rates at the 10 3–10 4 yr scale by taking advantage of easily identifiable lake shorelines and using them as strain markers. These shorelines formed as a result of protracted lake highstands in the Suguta and Turkana basins of northern Kenya during the African Humid Period (e.g., deMenocal et al., 2000), between the early and middle Holocene. Based on these deformed geomorphic markers, we estimated extension rates that agree with regional modern (Stamps et al., 2008) and longterm (Horner-Johnson et al., 2005) models, suggesting that steady

Fig. 1. Plate tectonic setting. Arrows show relative plate motion with respect to a stable Nubian reference frame, calculated with Euler poles from the modern plate-kinematic model of Stamps et al. (2008). Plates: SO, Somalia; VI, Victoria; NU, Nubia. WAR and EAR: western and eastern branches of the African Rift. Vectors in dark blue represent the model that includes the three plates. White dots are M > 5 earthquakes from the Centennial Catalogue (Engdahl and Villaseñor, 2002).

extension has been a hallmark on the million-year scale. Furthermore, the tectonic evolution of the Suguta–Turkana region records a progressive localization of extensional deformation, now focused along the ~ 35-km-wide inner rift trough. Whereas Miocene extension was broadly distributed (e.g., Morley et al., 1992), the consistency of late Pliocene to present-day rates and the combination of localized Quaternary deformation, shallow seismicity, and volcanism suggest that strain has become focused along the rift axis. Sustained constant extension rates and strain localization lend support to previous hypotheses that favored the role of local processes such as plume volcanism and magma-assisted deformation (e.g., Kendall et al., 2005), over far-field plate-tectonic forces, in the evolution from continental rifting to oceanic spreading.

2. Regional tectonic setting The EARS extends over more than 3000 km and constitutes the boundary between the African or Nubia and Somalia plates (Fig. 1) (e.g., McKenzie et al., 1970). Rifting in the EARS started during the early-middle Paleogene in southern Ethiopia, northern and central Kenya, and propagated northward and southward establishing the eastern branch of the EARS in early Miocene time (e.g., Baker and Wohlenberg, 1971; Ebinger and Sleep, 1998; Ebinger et al., 2000; Tiercelin and Lezzar, 2004; Wichura et al., 2010). During the early stages of rifting in northern Kenya, extension was distributed over a ~150-km-wide region creating several half-graben basins with up to 7 km fill (Dunkelman et al., 1989; Hautot et al., 2000; Hendrie et al., 1994; Morley et al., 1992). Tectonic activity along those basins ceased during the early Pliocene, and subsequently become localized in the present Rift Valley, resulting in the formation of the inner trough (e.g., Morley et al., 1999; Truckle, 1976). Between ~8°S and ~ 3.5°N, the EARS is structured into western and eastern branches (Fig. 1), which straddle a central cratonic region, termed the Victoria microplate (Calais et al., 2006). Fig. 1 shows plate-boundary slip vectors calculated using the Euler poles of the two- and three-plate models from Stamps et al. (2008), the latter including the Victoria microplate. Slip vectors from the three-plate model tend to agree better with the regional rift geometry, and predict rates of extension that along the EARS between southern Ethiopia and central Kenya decrease southward from ~7 to 1.4 mm/yr (Fig. 1). This pattern is mimicked by a southward increase in crustal thickness estimated from seismic-refraction profiles (KRISP, 1991), compatible with a decrease in the magnitude of finite extension. Shallow earthquakes of M > 5 have occurred along the EARS during the past century, with only a few large events associated with surface ruptures (Abdallah et al., 1979; Ambraseys, 1991; Parsons and Thompson, 1991). It is unlikely that seismic slip on crustal faults accounts for all the extension across the rift, unless very large events with long recurrence times are missing from historical catalogs (e.g., Zielke and Strecker, 2009). In fact, both the 2005–2009 Afar (Grandin et al., 2010; Wright et al., 2006) and 2007 Natron (Calais et al., 2008) rifting episodes were associated with dyke intrusions and aseismic fault slip, which accounted for most of the strain, suggesting that part of the extension is coupled with volcano-tectonic processes. In the Suguta and southern Turkana regions, clusters of shallow microseismicity recorded by a temporary local seismological network have occurred along the rift axis, where volcanic centers are aligned along steep faults (Pointing et al., 1985). This is similar to frequent microseismicity along the axis of the Baringo basin to the south inferred to be related to dyking (Tongue et al., 1994). Tectonic activity in Baringo and within the Central Kenya Rift has also migrated into the present-day rift center, and apparently postdates a change in the regional extension direction from east–west to northeast–southwest (Strecker et al., 1990).

D. Melnick et al. / Earth and Planetary Science Letters 331–332 (2012) 335–346

3. Geomorphic and structural settings of the Suguta and Turkana basins Our study focused on the Suguta Basin and southern part of the Turkana Basin of northern Kenya, two tectonic depressions occupying the axis of the EARS with more than 1 km of cross-sectional relief (Fig. 2). Lake Turkana is one of Africa's largest lakes, with a length of more than 250 km, a surface area of ~7000 km 2, and a catchment area of 148,000 km 2. Immediately to the south, the Suguta Valley is a smaller basin, partially covered by the shallow ephemeral Lake Logipi. The Suguta Valley is ~ 100 km long and 15–20 km wide, its catchment area is 13,300 km 2, and it is separated from the Baringo and Turkana depressions by volcanic edifices. Four major volcanic complexes are aligned along the Suguta Valley and southern Lake Turkana: South Island, The Barrier, Namarunu, and Emuruangogolak (Fig. 3). These polygenetic edifices and shield volcanoes record protracted effusion of bimodal volcanic products over the past ~ 3 Myr (Dunkley et al., 1993). Lake Turkana is internally drained, fed principally by the Omo River that drains the Ethiopian Highlands (Fig. 2). These conditions were apparently established 2.2–2 Myr ago, when volcanism along the southeastern basin margin impounded the east-directed drainage of the paleo-Omo river (Bruhn et al., 2011). Hydrologic isolation of the Suguta and Turkana basins was subsequently established at ~ 0.2 Myr as a result of growth of The Barrier complex that today separates both basins (Dunkley et al., 1993; Truckle, 1976).

337

The shallow crustal structure of the Turkana region has been revealed by on- and offshore seismic-reflection profiles. Onshore profiles across the Lokichar Basin, west of the present rift (Fig. 2), image east-dipping normal faults with variations in dip angle between 20° and 60° (Morley, 1999). In turn, offshore profiles across the rift axis imaged sequences of east-dipping faults with steeper dips of 55–65° that have controlled deposition of Miocene to recent sediments (Dunkelman et al., 1988). These structures are exposed along the southern margin of Lake Turkana, where they offset late Pleistocene– Holocene lava flows, monogenetic cones, and Quaternary alluvium. The EARS in the Lake Turkana region is characterized by marked variations in geometry and structure (Fig. 2). In the south, where this study focuses on, deformation is restricted to a relatively narrow zone, 25 to 30 km wide, accommodated by NNE-striking faults parallel to the rift trend. To the north of South Island, the Rift Valley transitions into an oblique transfer zone (Dunkelman et al., 1988), and deformation is partitioned between the main rift along Lake Turkana and a parallel system of normal faults to the east (Fig. 2). Apparently, this parallel system has accommodated very minor extension at ~0.1 mm/yr (Vetel et al., 2005). Various structural models have been proposed for the Suguta Valley. Based on field observations in the northeastern sector of the valley, Bosworth and Maurin (1993) proposed a principal westdipping rift-bounding fault along the eastern rift flank. Conversely, Dunkley et al. (1993) inferred a symmetrical graben. The Suguta depression is delimited in the west by a remarkably linear normal

Fig. 2. Regional topography of the Suguta and Turkana regions, drainage systems, and uplift resulting from isostatic rebound. Rebound calculated using the thin-plate method (Ventsel and Krauthammer, 2001) and the effective elastic thickness from Pérez-Gussinyé et al. (2009), including loads associated with Suguta and Turkana lake-level changes, and sediment deposition in Lake Turkana, see text for details and Fig. S3. Triangles denote active volcanoes. Inset shows distribution of elastic thickness.

338

D. Melnick et al. / Earth and Planetary Science Letters 331–332 (2012) 335–346

Fig. 3. Topography, structures, and studied sites. Thick faults are main regional structures; thin lines are local faults identified in the field and mapped using SPOT images. White boxes indicate location of swath profiles in Fig. 6. Inset shows histograms with fault strikes and maximum highstand shoreline elevations below the respective basin sill. Volcanic centers: EM—Emuruangogolak; NA—Namarunu; TB—The Barrier (includes Kalolenyang, Kakorinya, and Teliki); SI—South Island.

fault system with constant NNE strike for over ~ 100 km, whereas in the east it is bounded by a series of smaller and less continuous fault systems (Fig. 3). Both fault arrays cut Holocene alluvium. Active faulting also occurs along the rift axis, associated with monogenetic volcanism and geothermal activity. The up to 20-km-wide monocline described by Bosworth and Maurin (1993) along the eastern Suguta margin is a distinct feature of this region. The monocline bends and its dip steepens east of Namarunu volcano, which has resulted in gravitational collapse of the rift flank and the formation of a large landslip. Both first and second-order faults along the Suguta and southern Turkana basins strike parallel to the rift orientation (inset in Fig. 3), suggesting that rift-normal extension dominates. From the interpretation of satellite imagery, Vetel and Le Gall (2006) inferred that

oblique-striking transfer zones cut the Suguta and southern Turkana regions, but neither the topography nor our field mapping (Fig. 3) supports the existence of such structures. 3.1. Late Pleistocene–Holocene lake highstands of the Suguta and Turkana basins At the end of the last glacial period, the generally dry climate of tropical Africa was superseded by a transient wet phase, also known as the African Humid Period (deMenocal et al., 2000), which prevailed between ~12 and ~ 5 kyr (e.g., deMenocal et al., 2000; Ritchie et al., 1985). During this wet period lake levels were high in several rift basins (e.g., Bergner et al., 2009; Garcin et al., 2009; Gasse,

D. Melnick et al. / Earth and Planetary Science Letters 331–332 (2012) 335–346

339

Fig. 4. Radiocarbon ages and inferred lake level histories for the Suguta and Turkana basins. Makers denote radiocarbon ages calibrated with OxCal 4.1 (Bronk Ramsey, 2001) and IntCal09 (Reimer et al., 2009) and 3σ error bars. Thick stippled lines indicate inferred main lake levels. For references and details on the ages see Garcin et al. (this volume) for Turkana, and Garcin et al. (2009) and Junginger (2011) for Suguta. Dark gray rectangles show elevation of overflow sills and light gray inferred timing of lake level drops and abandonment of the maximum highstand shorelines. Suguta ages from Junginger (2011) older than 7 ka are from near-shore and offshore lacustrine sediments, providing minimum water levels; ages from Garcin et al. (2009) are from lacustrine shells immediately below shorelines.

2000; Owen et al., 1982; Truckle, 1976). Lake levels in the Suguta and Turkana basins were at that time 300 m and 100 m higher than today, respectively (Fig. 4). Both highstands were limited by the elevation of the overflow sills of each basin, the topographic thresholds that restricted fluvial connectivity with the White Nile (Owen et al., 1982) (Fig. 2). The Suguta sill is located on the western basin margin at an elevation of 580.15 m, as measured by dGPS (Garcin et al., 2009), providing fluvial connectivity into Lake Turkana. The Turkana sill is located in the Elemi triangle region west of the lake (Fig. 2) to which access is restricted due to political unrest; Shuttle Radar Topography Mission (SRTM) data constrains its elevation to between 457 and 460 m (Brown and Fuller, 2008). As a result of protracted periods of lake-overflow water-levels lasting several thousand years, a maximum highstand shoreline (MHS) was formed by wave erosion in each basin. The MHS is clearly exposed on the surroundings of the Suguta and Turkana basins where it forms distinctive features in the landscape that can be followed continuously for several tens of kilometers. Importantly, due to their erosive origin and its longevity (Garcin et al., 2009), the MHS is much wider and better developed than the multiple regressive shorelines at successively lower elevations (Fig. 5a). The chronology of Suguta and Turkana lake levels has been deduced from radiocarbon dating of gastropod and bivalve shells from freshwater organisms, bones, and charcoal in lacustrine sediments (Garcin et al., 2009; Garcin et al., this volume; Junginger, 2011). Radiocarbon ages from archeology have been also used to constrain the level of Lake Turkana (e.g., Owen et al., 1982). For the Suguta basin, based on dating of paired samples of charcoal and lacustrine carbonate material, Junginger (2011) determined an average lake-carbon reservoir age of 1.9 kyr for the paleo-lake. Considering this reservoir, 44 radiocarbon ages suggest a protracted highstand between ~14 and 7 kyr, followed by a rapid level fall between ~7 and 5 kyr (Fig. 4). Short transient periods of lower levels may have occurred during this highstand, but their exact timing and magnitudes remain unresolved. For the Turkana basin, radiocarbon dating of modern freshwater snail shells excludes a carbon reservoir effect (Garcin et al., this volume). High lake levels starting at ~12 kyr falling abruptly at 5.3± 0.3 kyr are inferred from 102 radiocarbon ages (Fig. 4). Detailed descriptions of the settings of dated samples and their interpretation in light of lake-level change and paleoclimatic context are presented in a companion paper (Garcin et al., this volume).

4. Materials and methods 4.1. Differential GPS survey of maximum highstand shorelines We surveyed MHS elevations for the Suguta and Turkana basins at 45 sites along 135 km of the rift (Fig. 3) using a differential global positioning system (dGPS; Leica system 1200) aided by a helicopter. The base-rover distance ranged from 47 to 75 km resulting in a height quality of 12 cm (standard deviation of over 200,000 points). These analytical errors are well below the elevation range of lacustrine geomorphic markers (~0.3–1.5 m). In order to match our measurements with the SRTM reference frame, we added the height separations (EGM96 geoid–WGS84 ellipsoid) to the dGPS elevations. Elevation data acquired previously was found consistent with the SRTM data (Garcin et al., 2009), suggesting dGPS sites could be compared with each other and the global reference frame. Depending on bedrock lithology and local slope, the MHS is typically associated with a distinct wave-cut notch, a terrace–cliff pair with up to 100-m-wide threads, a ridge containing well-rounded beach gravels, or a marked abrasion platform carved into resistant bedrock with a sharp shoreline angle (Fig. 5; further views in Supplementary material). The method used to estimate the elevation of the MHS and its associated uncertainty differs depending on the geomorphic expression. The most straightforward markers of paleo-lake levels are wave-cut notches; such features were found at sites associated with vesicular basalts and welded tuffs. In these cases, the elevation of the MHS was determined from averaging static points. At sites characterized by a wide abrasion platform and paleo-cliff, the shoreline angle was used to determine the MHS elevation. In the few cases where the shoreline angle was covered by younger colluvium, we extrapolated the exposed portions of the platform and cliff on profiles and determined the elevation by line intersection, as commonly done for analog features of marine origin (e.g., Lajoie, 1986). 4.2. Modeling isostatic rebound associated with lake level changes The Earth's crust behaves as a stiff, elastic solid when subjected to changes in surface loads (e.g., Turcotte and Schubert, 1982). It is therefore expected that changing water levels within rift basins will cause isostatic rebound, as reported for other large lakes (e.g., Adams et al., 1999; Bills et al., 1994). Unraveling the temporal

340

D. Melnick et al. / Earth and Planetary Science Letters 331–332 (2012) 335–346

a

b

c

d

e

f

Fig. 5. Field views of survey sites. Maximum highstand shoreline denoted by white arrows. White dots indicate site locations. a) Aerial view to the west of site LORI. Note the marked geomorphic expression and continuity of the MHS compared to lower regressive shorelines. b) Aerial view to the north of Namurinyang cone and sites NAMU, NAMC, NAM2, and NAM3. East-dipping normal faults controlling tilted blocks offset the MHS. c) View to the north of Losetum flank cone and sites LOSE and LOSC. Subvertical normal fault offsets the MHS by 15 m. d) Aerial view to the south of Andrews cone with sites SANW, SAND, and SAN3. Normal faults offset young lavas from The Barrier volcanic complex as well as the MHS. The Suguta Valley can be seen in the background. e) Aerial view to the south of Abili Agituk cone and sites TUAA and TUA2. Teliki's volcano (TE) and historical lava flows can be seen in the background. f) Aerial view toward east along the southern shore of Lake Turkana, and sites TUAB, TUAA, and SETU. Additional field views can be found in the Supplementary material (Fig. S5).

evolution of rebounds is crucial to estimate the magnitude of uplift and assess the role of rebound on deformation rates. In addition, since isostatically driven warping influenced vertical movements on the periphery of the basins it may have affected the degree of fluvial connectivity at the overflow locations. The finite amplitude and wavelength of this rebound will depend on the elastic rheology of the lithosphere, which can be described by its effective elastic thickness (Te), Young's modulus, and Poisson ratio. The temporal evolution of rebound, in turn, will depend on the viscosity of the lithosphere, a function of its composition and thermal state. In order to assess both the magnitude and temporal evolution of rebound associated with lake-level changes, we performed three experiments using different modeling strategies. We constructed a 1000-km-long reference profile normal to the rift across the northern Suguta Valley (Fig. 2), where lake-level changes were most pronounced (300 m). The load considered the difference between MHS elevations and present topography, and a water density of 998 kg/ m 3 (modern Lake Turkana with temperature of 27.5 °C and salinity of 2500 mg/L; Johnson and Malala, 2009). In order to consider marked changes in lithospheric rheology across an active rift, we

performed a 3D experiment using the thin-shell method (Ventsel and Krauthammer, 2001), a regional Te map (Pérez-Gussinyé et al., 2009), and volumetric loads. We fixed Young's modulus to 83 GPa and the Poisson ratio to 0.25, building on a previous modeling study in the region (Golke and Mechie, 1994). To obtain the temporal evolution of rebound we performed a set of simulations by means of a thermo-mechanical (TM) model (Popov and Sobolev, 2008); details of the model setup can be found in the Supplementary material (Figs. S1 and S2). Fortunately, both detailed gravity (Mariita and Keller, 2007) and seismic refraction (KRISP, 1991) data are available for the region, providing key estimates of crustal thickness and density distributions across the reference profile. We adopted variable lithospheric thicknesses from regional seismic tomography studies (Artemieva, 2006; Conrad and Lithgow-Bertelloni, 2006), and the geothermal gradient estimated from mantle xenoliths found in recent lavas (Henjes-Kunst and Altherr, 1992) and its relationship to seismic velocities (Mechie et al., 1994). The lithospheric thickness and geothermal gradient are probably the least well-constrained input parameters due to the limited resolution of regional tomography and restricted spatial

D. Melnick et al. / Earth and Planetary Science Letters 331–332 (2012) 335–346

distribution of xenoliths, respectively. To account for uncertainties, we performed a sensitivity analysis by changing the lithospheric thickness by 25 km and the geothermal gradient by ±50 °C. After a stabilization period of 300 kyr, we removed the loads corresponding to the mass of water involved in the Suguta lake-level change episode, and 2 kyr later for Lake Turkana. Sediment cores from Lake Turkana have revealed relatively rapid sedimentation rates of 2–3.5 mm/yr over the past 3.5–4 kyr (Halfman et al., 1994). We estimated that a mean of 10 m of sediments is likely to have been deposited since 5 kyr over an area of 8100 km 2. Using the measured sediment density (Barton and Torgersen, 1988) we estimated the flexural subsidence associated with this load and subtracted it from the unloading response (Fig. S3). As a result of unloading, uplift at the center of the rift is greater than on the flanks where the overflow sills are located. Tectonically driven subsidence and extension may therefore be underestimated if the effects of isostatic uplift are not included. Considering such issues is important because both the Suguta and Turkana sills are at different locations with respect to the center of unloading, and because the orientation of the rift changes between these regions. The rebound at

341

each MHS site is thus estimated as the deflection from the thin-shell model minus the deflection of the respective sill (Table 1). 4.3. Estimating extension rates from deformed Holocene shorelines We determined extension in the uppermost crust by summing fault heave under the assumption that planar faults mapped in the field account for all the strain. Because of the difficulty in measuring fault dips in the field due to limited deep vertical exposed sections, it is more accurate to determine fault throws and estimate heave for a range of dips (e.g., Bell et al., 2011; Lamarche et al., 2006). We calculated the amount of extension along four profiles on South Turkana, Logipi, Namarunu, and Baragoi (Figs. 3 and 6). Tectonic subsidence was first estimated by calculating the cumulative vertical fault throw for groups of faults that offset the MHS, both for the original elevations and those corrected for isostatic rebound (Fig. 6, Tables S1 and S2). For the faults located beyond the extent of the MHS, throw was estimated with respect to the elevation of the overflow sills. Fault throw was converted to heave by assuming fault dips between 55° and 65°, the range estimated using seismic reflection profiles along the southern

Table 1 Maximum highstand shoreline sites. Elevation at SLTU from SRTM data. Site code

Site locality

Longitude (°)

Latitude (°)

Elevation (m)

Error (m)

Deflectiona (m)

NAMN NAMW NAME CTTR LORE STT2 EAND WLOG EALO BAHS AROL KAIN NARW EMU1 EMU2 MURU LOSE LOS2 NAMR KAMU NAMS NAKI KALO SAND NAMU LORI STTR TRTR SILL LOS1 LOSC SANW SAN3 NAMC NAM1 NAM2 NAM3 NETU ILTU JJTU SWTU TUAB TUAA SETU TUA2 SLTU

Namarunu north Namarunu west rhyolite Namarunu east Tir–Tir plateau center Lorenkipi road Tir–Tir plateau south Andrews cone east Logipi west Alowalan east Baragoi River Aroliao Kainagokin Namruy west Emuruangogolak 1 Emuruangogolak 2 Muruangikokolak Losotem East Losetum East Namruy sediments Kamuge river Namarunu south Nakitoekirion Kalolenyang south Andrews cone center Namurinyang west Loriu plateau east Tir–Tir south Tir–Tir landslip Sill Suguta Losetum west Losotem center Andrews cone west Andrews cone east Namurinyang center Namurinyang east 1 Namurinyang east 2 Namurinyang east 3 El Molo South Island Luguruguru north Luguruguru south Neangoil Abili Agituk east Nakujakabon Abili Agituk west Sill Turkana

36.3922 36.4097 36.4507 36.5251 36.2612 36.5153 36.5882 36.445 36.6674 36.4604 36.3161 36.2924 36.2173 36.3342 36.3454 36.402 36.4025 36.3738 36.2214 36.1791 36.4152 36.3775 36.5141 36.5766 36.6112 36.3985 36.5104 36.5435 36.0884 36.3719 36.4013 36.5752 36.579 36.6121 36.6146 36.6155 36.6167 36.7127 36.5871 36.5141 36.5223 36.5642 36.5869 36.7056 36.5861 35.21075

2.08503 1.98324 1.97959 1.92662 1.71407 1.78976 2.26859 2.25166 2.19337 1.61757 1.87475 1.81441 1.58116 1.56085 1.56487 1.56672 1.57872 1.57264 1.55131 1.72246 1.95587 1.94573 2.28294 2.27299 2.24826 2.15995 1.75922 1.99773 1.81675 1.57189 1.57797 2.27597 2.27255 2.24808 2.24775 2.24693 2.24737 2.67459 2.66132 2.55886 2.44133 2.39224 2.40949 2.45176 2.4035 4.95952

570.08 573.24 558.17 568.50 573.74 566.78 559.18 570.74 572.01 571.86 572.53 569.97 557.23 565.65 565.48 558.54 566.43 557.06 563.00 574.36 577.41 570.70 568.70 558.86 568.76 572.59 566.70 533.95 580.15 560.94 553.21 560.45 559.20 569.90 569.44 570.58 570.42 444.12 437.54 439.56 442.57 441.48 441.22 446.50 439.42 458.50

0.25 0.97 0.11 0.47 0.13 0.54 0.61 0.20 0.10 0.33 0.30 0.81 2.00 0.36 0.07 0.54 0.37 0.28 0.80 1.53 0.67 0.48 1.00 0.40 0.14 0.90 1.70 0.66 0.83 1.20 0.50 0.24 0.10 0.10 0.34 0.87 0.61 0.18 0.56 0.10 0.60 0.32 0.20 0.17 0.38 1.50

2.21 1.89 1.94 1.71 0.31 1.05 2.73 2.77 2.28 0.12 1.21 0.87 − 0.47 − 0.31 − 0.27 − 0.19 − 0.12 − 0.19 − 0.61 0.05 1.80 1.68 2.87 2.77 2.62 2.45 0.89 2.00 0.00 − 0.19 − 0.13 2.78 2.76 2.62 2.61 2.61 2.60 4.33 5.12 5.20 4.96 4.80 4.77 4.26 4.76 0.00

a

Deflection at each site minus deflection of the respective sill.

342

D. Melnick et al. / Earth and Planetary Science Letters 331–332 (2012) 335–346

Fig. 6. Maximum highstand shorelines along rift-normal profiles. Swath areas are shown in Fig. 3. Projected shoreline sites and second-order faults, blue faults are east down, red west down. Fault numbers as in Table S1, f—fault, b—block. Site LORE has been shifted to the west to account for a change in the strike of fault f2. Stippled blue lines indicate interpreted horizon used to estimate fault-group throw. Sites affected by inferred doming along the Baragoi profile have not been considered when estimating deformation. See Section 5.2 for details on the interpretation.

shore of Lake Turkana (Dunkelman et al., 1989; Morley et al., 1992). Accordingly, the extension rate across the rift was calculated as the sum of fault-heave values divided by the timing of MHS abandonment, defined as 7.0 ± 0.5 kyr and 6.0 ± 0.5 kyr for the Suguta and Turkana basins, respectively (Fig. 4). Using the line-length change of each profile we also estimated the total strain and strain rate (Table S2). In some regions we observed block tilting in the field (Fig. 5b), and also deduced it from the linear relation of MHS elevations with respect to fault positions and their correlation to the topography, such as in the western sectors of the South Turkana and Namarunu profiles (Fig. 6). Extension associated with block tilting was estimated following the domino-block model (Wernicke and Burchfiel, 1982), neglecting internal faulting within the blocks.

5. Results: using deformed shorelines to infer deformation rates 5.1. Flexural rebound associated with lake-level changes The temporal and spatial evolution of unloading associated with lake-level changes was estimated along the reference profile (Figs. 7 and 8). Despite introducing variations in the initial lithospheric thickness or geothermal gradient, the TM experiments converge toward stability after ~ 0.1 Myr (Fig. 8a). Changes in these initial conditions only affect the absolute elevation at the rift axis, and despite variations in the input parameters the same overall evolution is observed. Therefore, our results for the relative elevation change associated with local unloading will not be strongly influenced by uncertainties

D. Melnick et al. / Earth and Planetary Science Letters 331–332 (2012) 335–346

a

b

Fig. 7. Isostatic rebound modeling reference profile. a) Topography and effective elastic thickness (Te) from a 100-km-wide swath (location of profile in Fig. 2). Te from PérezGussinyé et al. (2009). b) Flexural isostatic response to unloading associated with lake-level falls in the Suguta and Turkana basins, for different modeling approaches. TM—thermomechanical model, see text for details.

in the input parameters. In particular, the response to unloading is almost identical for models with different geothermal gradients, both in amplitude and temporal evolution, and ~ 1 m lower for the model with thicker lithosphere (Fig. 8b). After unloading, all models predict an initial phase of rapid uplift at 2–4 mm/yr lasting 2–4 kyr, followed by a phase of protracted, slow uplift (Fig. 8b). The uplift rate starts to decrease ~1 kyr after the Suguta unloading episode, but accelerates

343

again as a result of unloading in Lake Turkana, and finally stabilizes after ~3 kyr. All models suggest that the rift has been relatively stable over the past ~2 kyr, causing local uplift at a subdued rate of ~0.1 mm/yr. Rebound along the reference profile reached between 7 and 9 m, depending on the method (Fig. 7). The most robust estimates are likely those from the TM model, as it included realistic lithospheric geometries as well as physical and thermal properties. However, the estimates of the thin-shell model along the reference profile are within the range predicted by the TM model, both at the center of the rift and at the western flank where both overflow sills are located. The TM model predicts larger subsidence on the eastern rift flank, an effect associated with the different lithospheric thicknesses of both rift-bounding blocks. This effect is related to the better resolution of the TM model and processes that the thin-shell model cannot simulate. We unfortunately lack subsurface data to construct a regional 3D TM model. However, because the amplitude of rebound exhibits pronounced latitudinal variations (Fig. 2), and given that our TM model predicts that the rapid phase of rebound-induced uplift finished ~ 2 kyr ago (Fig. 8b), the use of the thin-shell model is justified. The Suguta sill is located within the zone of rebound and thus was uplifted 4.5 m, while the Turkana sill is located on the western periphery and thus may have experienced only at most 2.7 m of uplift. In order to account for these differences in sill uplift the MHS elevations corrected for rebound were estimated by adding the difference between uplift at each MHS site and at the respective basin sill (Table 1). 5.2. Interpretation of deformed maximum highstand shorelines

a

b

Fig. 8. Temporal evolution of isostatic rebound and model thermomechanical sensitivity tests. a) Uplift predicted by the thermo-mechanical model along the axis of the rift for scenarios with different geothermal gradients and lithospheric thicknesses. Note that all models stabilize after ~0.15 Myr. Unloading is imposed at 0.3 Myr. b) Evolution of unloading in response to lake-level falls in the Suguta and Turkana basins, corresponding to 300 m at 7 kyr and 100 m at 5 kyr, respectively. All models indicate that rebound stabilized at ~2 kyr.

Our survey of the MHS at 45 sites shows that their elevations are between 2.7 m and 46.2 m below the respective sill (inset in Fig. 3), as a result of subsidence of the inner rift. We analyzed MHS distributions and quantified deformation rates along four 50-km-long swath profiles normal to the orientation of the rift (Figs. 3 and 6). All profiles show a consistent pattern with higher MHS elevations on the rift flanks and lower elevations toward the rift center, mimicking the local topography. Marked changes in MHS elevations across individual normal faults can be observed in all profiles as well as in the field (Fig. 5b–e). Many of these faults also offset post-MHS alluvial fans and Holocene sediments. However, in order to obtain deformation rates certain interpretations and assumptions need to be made, which are detailed below. The highest MHS sites along the Turkana profile occur on the eastern side of the lake (SETU). The westernmost site (JJTU) is at a lower position than the site immediately to the east (SWTU), an effect that we attribute to westward block tilting controlled by the western riftbounding fault system (Fig. 6). Footwall uplift and westward tilting of blocks with lengths greater than 5 km are evident from the topography and imaged in seismic profiles immediately to the north (Dunkelman et al., 1989). The lowest MHS site (ILTU) is at South Island, ~15 km from the profile. Because correlating active faults over such a distance is tenuous due to possible structural complexity and along-strike changes in slip, we discard this site from the estimate of deformation rate. Along the Logipi profile, the highest MHS sites are located on both rift flanks and have similar heights. Most of the faults along this profile dip east, except the eastern rift-bounding fault (f20 in Fig. 6) and minor faults between Andrews cone (SAND) and the Namurinyang eruptive center (NAMU). MHS elevations decrease asymmetrically toward the rift axis following fault-dip direction. Minor westward block tilting occurs at Namurinyang (Fig. 5b), an effect that we account for by including a small block (b15) between faults 15 and 16. The Namarunu profile has the greatest range of MHS elevations, reaching 44 m between NAMS and TRTR. However, the steep eastern rift flank suffered from large-scale gravitational collapse resulting in

344

D. Melnick et al. / Earth and Planetary Science Letters 331–332 (2012) 335–346

an extensive landslip (Fig. 3) (Bosworth and Maurin, 1993; Dunkley et al., 1993), which most likely accounts for the low elevation of TRTR. In order to avoid such bias we excluded TRTR from the estimates of deformation rates. The highest site along this profile (NAMS) is associated with westward block tilting along the western rift-bounding fault system. Footwall uplift and block tilting in this region are also evident from the broad-scale topography and from field observations; we include a block (b4) between faults 4 and 5. Along the Baragoi profile, MHS sites at both rift footwalls are at higher positions than their counterparts on the valley floor. However, at Losetum on the rift axis the MHS is offset west-down by 15 m across a normal fault (Fig. 5c), and farther west MHS elevations increase continuously by ~14 m over a horizontal distance of 7 km (Fig. 6). This bell-shaped pattern is limited to the rift floor and occurs on the northern flank of the Emuruangogolak shield volcano and caldera complex; we associate it with domal uplift resulting from protracted magmatic inflation. Transient uplift with a similar wavelength has also been documented at other shield volcanoes in Kenya (Biggs et al., 2009). In order to avoid such bias, we excluded these sites (LOS1, LOS2, EMU1, EMU2) from our estimates of deformation rates. The 15 m offset of the MHS across a fault at Losetum (Fig. 5c) is probably also associated with volcanic activity. This fault is very well exposed, it has a subvertical dip and open fissures that expel volcanic gases. Thus, slip on this fault does not directly contribute to tectonic opening of the rift; this fault possibly represents the surface expression of a dyke intrusion, which may contribute to opening of the rift by magmatic addition, but that cannot be quantified using the MHS. 5.3. Extension rates across the EARS in the Suguta and South Turkana regions The plate-kinematic model of Stamps et al. (2008) predicts extension rates decreasing linearly from 4.3–5.1 mm/yr at south Turkana to 4.0–4.8 mm/yr at Baragoi (Fig. 9). Overall extension rates determined from the MHS range from 3.2 to 6.0 mm/yr, or 3.5 to 6.7 mm/yr after correcting for rebound, but with no clear pattern of variation along the rift (Fig. 9). The most important source of uncertainty in estimating extension rates arises from the assumed range of fault dips. The effect of rebound is greatest on the Turkana profile, where differences between uncorrected and corrected extension rates reach 0.9 mm/yr; southward, this effect decreases to almost zero at Baragoi. The highest extension rates of 3.7–5.6, or 4.5–6.7 mm/yr considering

Fig. 9. Modern and Holocene extension rates along the northern Kenya Rift. Presentday rates from plate-kinematic model of Stamps et al. (2008). Holocene rates shown are uncorrected and corrected for flexural rebound induced by lake-level falls. Cumulative throw from our four profiles (Fig. 6) converted to heave using fault dips of 55° to 65°, inferred from seismic profiles collated at the south of Lake Turkana (Dunkelman et al., 1989). The adopted range of fault dips introduces the large apparent uncertainties in extension rates.

rebound, occur at the Turkana profile, whereas lower rates of 3.2–4.7, or 3.5–5.3 mm/yr after correction, occur immediately to the south on the Logipi profile (Figs. 6 and 9). The total mid-Holocene extension ranges from 22 to 39 m or 6 × 10 − 4 to 10 − 3%, at an average strain rate of 5 × 10 − 15 s − 1 (Table S2). Interestingly, if all faults are considered, there is an apparent eastward decrease in displacement rate (Fig. S4). 6. Discussion 6.1. Rift structure along the Suguta and southern Turkana troughs Various models have been proposed for the upper-crustal structure of the Suguta Valley, involving a low-angle detachment (Bosworth, 1987), a half-graben with a west-dipping border fault (Bosworth and Maurin, 1993), and a symmetric graben (Dunkley et al., 1993). However, seismic reflection profiles at the southern Lake Turkana have imaged an east-dipping master fault controlling a halfgraben (Dunkelman et al., 1988). Based on our observations, we propose that this east-dipping master fault and halfgraben extend continuously southward and define the structural setting of the entire Suguta Valley. This interpretation is based on (1) the similar asymmetric topography in all swath profiles, characterized by steep western flanks with footwall back tilt and gently sloping eastern flanks with a broad, antithetically faulted monocline (Figs. 3 and 6), (2) the remarkably linear morphology of the western rift flank and of fault traces over a distance exceeding 120 km, which contrast both with the morphology and fault geometries along the eastern flank (Fig. 3), and (3) greater fault displacement rates along the western sectors and their eastward decrease (Fig. S4). Such a continuous rift structure is essential to compare extension rates determined from our four profiles in a framework of linked fault systems. 6.2. Implications of Holocene extension rates for rift dynamics Within our study region the azimuth of extension predicted by the modern plate-kinematic model is 104.3° (Stamps et al., 2008), which is orthogonal to the average rift orientation of 14°, suggesting plane strain. Thus, any significant bias in extension estimates due to oblique rifting can be discarded. On average, our estimates of extension rates along four profiles agree with rates predicted by this model (Fig. 9). However, our estimates should be considered minimum values due to incomplete sampling of the MHS, particularly along the Namarunu and Turkana profiles, and due to the possible contribution of smallscale block rotation and small faults (e.g., Marrett and Allmendinger, 1992). Nevertheless, the agreement between both rates implies that the plate-boundary strain in this segment of the EARS is accommodated along a relatively narrow region, 30 to 40 km wide. This is also supported by the distribution of young faults observed in the field. Thus the crustal blocks separated by this plate boundary, the Victoria microplate and the Somalia Plate (Calais et al., 2006), could be considered rigid, implying that kinematic block models may be appropriate to describe regional deformation patterns. During incipient rifting in northern Kenya in the Paleogene several coeval halfgraben basins formed over a >150-km-wide region (e.g., Morley et al., 1992; Morley et al., 1999). Rifting in the Suguta Valley started at ~ 10 Myr (Saneyoshi et al., 2006), contemporaneous with distributed tectonic activity. Tectonic activity then became localized at ~3 Myr establishing the present-day rift trough (Dunkley et al., 1993). The consistent extension rates at various time scales, but continuous decrease in the width of the deformation zone implies an increase in the strain rate. In fact, the Holocene strain rates of 10 − 15 s − 1 along our profiles are similar to the rates estimated for the past ~ 3 Myr, but at least an order of magnitude higher than those estimated for the Miocene (Hendrie et al., 1994). The estimates by Hendrie et al. (1994) only represent minimum values and hence

D. Melnick et al. / Earth and Planetary Science Letters 331–332 (2012) 335–346

the increase in strain rate could be even larger. An increasing strain rate but constant extension rate could be interpreted as a result of an increase in crustal thinning and localized magmatism, leading to lithospheric weakening and localization of the deformation. 6.3. Comparison of deformation rates over various time scales At continental scale, GPS deformation rates for the Africa–Arabia– Eurasia plate system tend to agree with geologic rates inferred on ~ 10 6 yr scales for structures in a wide range of tectonic settings, suggesting similar geological processes acting over 3–5 Myr (Reilinger et al., 2006). However, up to five-fold discrepancies exist for some structures, ascribed to uncertainties in geologic fault-slip measurements, variations in fault activity, distributed deformation, and lack of dense GPS measurements across narrow deformation zones. Few estimates of deformation rates spanning different time scales have been obtained for volcanic rifts. Subsidence rates in the Asal Rift (Djibouti) estimated from a Holocene shoreline (Stein et al., 1991) apparently agree with rates deduced for the past 10 5 yr and on scales of >10 6 yr (De Chabalier and Avouac, 1994). Geodetic measurements obtained during the past 25 yr across the Asal Rift suggest a ~ 20% faster rate, interpreted as transient rapid activity following a volcanotectonic rifting episode in 1978 (Vigny et al., 2007). Thus the Asal Rift, where continental breakup is complete and incipient oceanic spreading is taking place, is characterized by steady-state deformation and complete compensation for crustal stretching by magmatism (De Chabalier and Avouac, 1994). The consistency in extension rates on 10 3–10 4 and 10 1–10 6 yr time scales in the Suguta–Turkana region suggests that steady-state rifting conditions may be reached already during the initiation of the continental breakup process, when the role of far-field forcing is overtaken by the local effect of crustal thinning and magmatic weakening. However, we can only assess the strain accounted for by faults at the surface; at deeper levels in the crust this strain may also be accommodated by dyke intrusion and magmatic addition. The sustenance of tectonic deformation rates in the Suguta– Turkana region over different time scales is interesting because other extensional regions in continental interiors do not exhibit similar behavior. For example, in the Basin and Range province of the western US extension is distributed over a broad area; GPS rates are consistent with Holocene rates, but are ~3 times faster than those estimated for time scales of 10 5–10 7 yr (Friedrich et al., 2003). This discrepancy has been interpreted as a result of changes in surface loads associated with the disappearance of a large pluvial lake and distant ice sheets (e.g., Hetzel and Hampel, 2005). Lithospheric isostatic rebound in extensional regions increases the differential stress promoting normal faulting. Numerical models with a simple layered structure and no thermal component suggest that an acceleration in fault-slip rate associated with changes in surface loads only occurs if the viscosity of the lower crust is higher than that of the underlying mantle (Hampel et al., 2010). The viscosity predicted by our TM model along the rift axis decreases continuously from 10 27 to 10 19 Pa × s between 30 and 65 km depths, implying a stronger lower crust. However, no increase in deformation rates along the Suguta– Turkana region is apparent from our data. The magnitude of the load removed by the lake-level changes may possibly not have been sufficient to trigger an increase in deformation rates observable within the uncertainties of our method. Alternatively, conditions in thermally weakened volcanic rifts, where deformation is localized, may be different from those in extensional provinces in intracontinental regions. 6.4. Isostatic rebound and tectonic control on lake-level changes Our thin-shell model predicts up to 9.7 m of uplift from isostatic rebound associated with falls in lake levels of the Suguta and Turkana

345

basins together with flexure from post-5 kyr sedimentation in Lake Turkana (Figs. 2 and S3). The effect of rebound on the lake level is determined by the position of the overflow sill, which are at ~ 35 km and ~100 km from the axis of unloading for the Suguta and Turkana basins, respectively (Fig. 2). If the lake levels begin to fall as a result of climate change, isostatic rebound will raise the rift center, potentially providing higher discharge, external connectivity, and sediment routing away from the rift. This tectonic/climatic interaction will continue until a threshold controlled by precipitation, evaporation, and rebound is reached, ultimately resulting in hydrologic isolation. However, coeval subsidence of the inner rift continues as a result of ongoing extension, thus counteracting isostatic rebound. Accordingly, the paleo-record of lake-level changes expressed by lacustrine shorelines involving amplitudes of less than ~20–30 m encapsulates tectonic, isostatic, and climatic components, a fact that need to be considered in paleo-climate reconstructions. Supplementary materials related to this article can be found online at doi:10.1016/j.epsl.2012.03.007. Acknowledgments Funding was provided by the German Research Foundation (DFG) projects GRK1364, TR419/6-1, and STR373/16-1. Y.G. was supported by an Alexander von Humboldt fellowship. We thank the Government of Kenya for research permits OP/13/001/23C 290 and MOEST 13/001/ 23C 290 and 13/001/30C 59/ 22. SPOT imagery acquired through ISIS project 156; logistic support by Wild Frontiers and Tropic Air. Photo in Fig. 5e courtesy of Ch. Strebel, Yellow Wings Air, Nairobi. We thank M. Trauth for help with the organization and an anonymous reviewer for constructive comments. References Abdallah, A., Courtillot, V., Kasser, M., Le Dain, A.-Y., Lepine, J.-C., Robineau, B., Ruegg, J.-C., Tapponnier, P., Tarantola, A., 1979. Relevance of Afar seismicity and volcanism to the mechanics of accreting plate boundaries. Nature 282, 17–23. Adams, K.D., Wesnousky, S.G., Bills, B.G., 1999. Isostatic rebound, active faulting, and potential geomorphic effects in the Lake Lahontan basin, Nevada and California. Bull. Geol. Soc. Am. 111, 1739–1756. Ambraseys, N.N., 1991. Earthquake hazard in the Kenya Rift: the Subukia earthquake 1928. Geophys. J. Int. 105, 253–269. Artemieva, I.M., 2006. Global thermal model TC1 for the continental lithosphere: implications for lithosphere secular evolution. Tectonophysics 416, 245–277. Baker, B.H., Wohlenberg, J., 1971. Structure and evolution of Kenya Rift Valley. Nature 229, 538–542. Barton, C.E., Torgersen, T., 1988. Palaeomagnetic and 210Pb estimates of sedimentation in Lake Turkana, East Africa. Palaeogeogr. Palaeoclimatol. Palaeoecol. 68, 53–59. Bell, R.E., McNeill, L.C., Henstock, T.J., Bull, J.M., 2011. Comparing extension on multiple time and depth scales in the Corinth Rift, Central Greece. Geophys. J. Int. 182, 463–470. Bergner, A.G.N., Strecker, M.R., Trauth, M.H., Deino, A., Gasse, F., Blisniuk, P., Duhnforth, M., 2009. Tectonic and climatic control on evolution of rift lakes in the Central Kenya Rift, East Africa. Quatern. Sci. Rev. 28, 2804–2816. Biggs, J., Anthony, E.Y., Ebinger, C.J., 2009. Multiple inflation and deflation events at Kenyan volcanoes, East African Rift. Geology 37, 979–982. Biggs, J., Nissen, E., Craig, T., Jackson, J., Robinson, D.P., 2010. Breaking up the hanging wall of a rift-border fault: the 2009 Karonga earthquakes, Malawi. Geophys. Res. Lett. 37, L11305. Bills, B.G., de Silva, S.L., Currey, D.R., Emenger, R.S., Lillquist, K.D., Donnellan, A., Worden, B., 1994. Hydro-isostatic deflection and tectonic tilting in the central Andes: initial results of a GPS survey of Lake Minchin shorelines. Geophys. Res. Lett. 21, 293–296. Bosworth, W., 1987. Off-axis volcanism in the Gregory rift, east Africa: implications for models of continental rifting. Geology 15, 397–400. Bosworth, W., Maurin, A., 1993. Structure, geochronology and tectonic significance of the northern Suguta Valley (Gregory rift), Kenya. J. Geol. Soc. 150, 751–762. Bronk Ramsey, C., 2001. Development of the radiocarbon calibration program OxCal. Radiocarbon 43, 355–363. Brown, F.H., Fuller, C.R., 2008. Stratigraphy and tephra of the Kibish Formation, southwestern Ethiopia. J. Hum. Evol. 55, 366–403. Bruhn, R.L., Brown, F.H., Gathogo, P.N., Haileab, B., 2011. Pliocene volcano-tectonics and paleogeography of the Turkana Basin, Kenya and Ethiopia. J. Afr. Earth Sci. 59, 295–312. Buck, W.R., 2004. Consequences of asthenospheric variability on continental rifting. In: Karner, G.D., Taylor, B., Driscoll, N.W., Kohlstedt, D.L. (Eds.), Rheology and Deformation of the Lithosphere at Continental Margins. Columbia University Press, pp. 1–30. Calais, E., Ebinger, C.J., Hartnady, C., Nocquet, J.M., 2006. Kinematics of the East African Rift from GPS and earthquake slip vector data. In: Yirgu, G., Ebinger, C.J., Maguire,

346

D. Melnick et al. / Earth and Planetary Science Letters 331–332 (2012) 335–346

P.K.H. (Eds.), The Afar Volcanic Province Within the East African Rift System: Geological Society Special Publication, London, pp. 9–22. Calais, E., d'Oreye, N., Albaric, J., Deschamps, A., Delvaux, D., Deverchere, J., Ebinger, C., Ferdinand, R.W., Kervyn, F., Macheyeki, A.S., Oyen, A., Perrot, J., Saria, E., Smets, B., Stamps, D.S., Wauthier, C., 2008. Strain accommodation by slow slip and dyking in a youthful continental rift, East Africa. Nature 456, 783–788. Chu, D., Gordon, R.G., 1999. Evidence for motion between Nubia and Somalia along the Southwest Indian ridge. Nature 398, 64–67. Conrad, C.P., Lithgow-Bertelloni, C., 2006. Influence of continental roots and asthenosphere on plate–mantle coupling. Geophys. Res. Lett. 33, L05312. De Chabalier, J.-B., Avouac, J.-P., 1994. Kinematics of the Asal Rift (Djibouti) determined from the deformation of Fieale Volcano. Science 265, 1677–1681. deMenocal, P., Ortiz, J., Guilderson, T., Adkins, J., Sarnthein, M., Baker, L., Yarusinsky, M., 2000. Abrupt onset and termination of the African Humid Period: rapid climate responses to gradual insolation forcing. Quatern. Sci. Rev. 19, 347–361. Demets, C., Gordon, R.G., Argus, D.F., Stein, S., 1990. Current plate motions. Geophys. J. Int. 101, 425–478. Dunkelman, T.J., Karson, J.A., Rosendahl, B.R., 1988. Structural style of the Turkana Rift, Kenya. Geology, pp. 258–261. Dunkelman, T.J., Rosendahl, B.R., Karson, J.A., 1989. Structure and stratigraphy of the Turkana Rift from seismic-reflection data. J. Afr. Earth Sci. 489–510. Dunkley, P.N., Smith, M., Allen, D.J., Darling, W.G., 1993. The geothermal activity and geology of the northern sector of the Kenya Rift Valley. British Geological Survey. : Research Report SC/93/1. 185 pp. Ebinger, C.J., Sleep, N.H., 1998. Cenozoic magmatism throughout east Africa resulting from impact of a single plume. Nature 395, 788–791. Ebinger, C.J., Yemane, T., Harding, D.J., Tesfaye, S., Kelley, S., Rex, D.C., 2000. Rift deflection, migration, and propagation: linkage of the Ethiopian and Eastern rifts, Africa. Bull. Geol. Soc. Am. 112, 163–176. Engdahl, E.R., Villaseñor, A., 2002. Global seismicity: 1900–1999. In: Lee, W.H., Kanamori, H., Jennings, P.C., Kisslinger, C. (Eds.), International Handbook of Earthquake and Engineering Seismology. Academic Press, pp. 665–690. Friedrich, A.M., Wernicke, B.P., Niemi, N.A., Bennett, R.A., Davis, J.L., 2003. Comparison of geodetic and geologic data from the Wasatch region, Utah, and implications for the spectral character of Earth deformation at periods of 10 to 10 million years. J. Geophys. Res. 108, 2199. Garcin, Y., Junginger, A., Melnick, D., Olago, D.O., Strecker, M.R., Trauth, M.H., 2009. Late Pleistocene–Holocene rise and collapse of Lake Suguta, northern Kenya Rift. Quatern. Sci. Rev. 28, 911–925. Garcin, Y., Melnick, D., Olago, D., Strecker, M., Tiercelin, J.J., this volume. East African mid-Holocene wet–dry transition recorded in fossil shorelines of Lake Turkana, northern Kenya Rift. Earth Planet. Sci. Lett. Gasse, F., 2000. Hydrological changes in the African tropics since the last glacial maximum. Quatern. Sci. Rev. 19, 189–211. Golke, M., Mechie, J., 1994. Finite-element modelling of the structure and evolution of the South Kenya Rift, East Africa. Tectonophysics 236, 439–452. Grandin, R., Socquet, A., Doin, M.P., Jacques, E., de Chabalier, J.B., King, G.C.P., 2010. Transient rift opening in response to multiple dike injections in the Manda Hararo rift (Afar, Ethiopia) imaged by time-dependent elastic inversion of interferometric synthetic aperture radar data. J. Geophys. Res. 115, B09403. Halfman, J.D., Johnson, T.C., Finney, B.P., 1994. New AMS dates, stratigraphic correlations and decadal climatic cycles for the past 4 ka at Lake Turkana, Kenya. Palaeogeogr. Palaeoclimatol. Palaeoecol. 111, 83–98. Hampel, A., Karow, T., Maniatis, G., Hetzel, R., 2010. Slip rate variations on faults during glacial loading and post-glacial unloading: implications for the viscosity structure of the lithosphere. J. Geol. Soc. 167, 385–399. Hautot, S., Tarits, P., Whaler, K., Le Gall, B., Tiercelin, J.-J., Le Turdu, C., 2000. Deep structure of the Baringo Rift Basin (Central Kenya) from three-dimensional magnetotelluric imaging: implications for rift evolution. J. Geophys. Res. 105, 23493–23518. Hendrie, D.B., Kusznir, N.J., Morley, C.K., Ebinger, C.J., 1994. Cenozoic extension in northern Kenya: a quantitative model of rift basin development in the Turkana region. Tectonophysics 236, 409–438. Henjes-Kunst, F., Altherr, R., 1992. Metamorphic petrology of xenoliths from Kenya and Northern Tanzania and implications for geotherms and lithospheric structures. J. Petrol. 33, 1125–1156. Hetzel, R., Hampel, A., 2005. Slip rate variations on normal faults during glacial–interglacial changes in surface loads. Nature 435, 81–84. Horner-Johnson, B.C., Gordon, R.G., Cowles, S.M., Argus, D.F., 2005. The angular velocity of Nubia relative to Somalia and the location of the Nubia–Somalia–Antarctica triple junction. Geophys. J. Int. 162, 221–238. Johnson, T.C., Malala, J.O., 2009. Lake Turkana and its link to the Nile. In: Dumont, H.J. (Ed.), The Nile. : Monographiae Biologicae. Springer, pp. 287–304. Junginger, A., 2011. East African Climate Variability on Different Time Scales: The Suguta Valley in the African–Asian Monsoon Domain, PhD Thesis, University of Potsdam, Potsdam, 153 pp. Kendall, J.M., Stuart, G.W., Ebinger, C.J., Bastow, I.D., Keir, D., 2005. Magma-assisted rifting in Ethiopia. Nature 433, 146–148. KRISP, 1991. Large-scale variation in lithospheric structure along and across the Kenya Rift. Nature 354, 223–227. Lajoie, K.R., 1986. Coastal tectonics. Active Tectonics 95–124. Lamarche, G., Barnes, P.M., Bull, J.M., 2006. Faulting and extension rate over the last 20,000 years in the offshore Whakatane Graben, New Zealand continental shelf. Tectonics 25, TC4005. Mariita, N.O., Keller, G.R., 2007. An integrated geophysical study of the northern Kenya rift. J. Afr. Earth Sci. 48, 80–94.

Marrett, R., Allmendinger, R.W., 1992. Amount of extension on “small” faults: an example from the Viking graben. Geology 20, 47–50. McKenzie, D., Davies, D., Molnar, P., 1970. Plate tectonics of the Red Sea and East Africa. Nature 226, 243–248. Mechie, J., Fuchs, K., Altherr, R., 1994. The relationship between seismic velocity, mineral composition and temperature and pressure in the upper mantle—with an application to the Kenya Rift and its eastern flank. Tectonophysics 236, 453–464. Molnar, P., 1988. Continental tectonics in the aftermath of plate tectonics. Nature 35, 131–137. Morley, C.K., 1999. Marked along-strike variations in dip of normal faults—the Lokichar fault, N. Kenya rift: a possible cause for metamorphic core complexes. J. Struct. Geol. 21, 479–492. Morley, C.K., Wescott, W.A., Stone, D.M., Harper, R.M., Wigger, S.T., Karanja, F.M., 1992. Tectonic evolution of the northern Kenyan Rift. J. Geol. Soc. 149, 333–348. Morley, C.K., Wescott, W.A., Stone, D.M., Harper, R.M., Wigger, S.T., Day, R.A., Karanja, F.M., 1999. Geology and geophysics of the Western Turkana Basins, Kenya. In: Morley, C.K. (Ed.), Geoscience of Rift Systems-evolution of East Africa: American Association of Petroleum Geologists Studies in Geology, pp. 19–54. Owen, R.B., Barthelme, J.W., Renaut, R.W., Vincens, A., 1982. Palaeolimnology and archaeology of Holocene deposits north-east of Lake Turkana, Kenya. Nature 298, 523–529. Parsons, T., Thompson, G.A., 1991. The role of magma overpressure in suppressing earthquakes and topography: worldwide examples. Science 253, 1399–1402. Pérez-Gussinyé, M., Metois, M., Fernandez, M., Verges, J., Fullea, J., Lowry, A.R., 2009. Effective elastic thickness of Africa and its relationship to other proxies for lithospheric structure and surface tectonics. Earth Planet. Sci. Lett. 287, 152–167. Pointing, A.J., Maguire, P.K.H., Khan, M.A., Francis, D.J., Swain, C.J., Shah, E.R., Griffiths, D.H., 1985. Seismicity of the northern part of the Kenya Rift Valley. J. Geodynamics 3, 23–37. Popov, A., Sobolev, S., 2008. Slim3d: a tool for three-dimensional thermomechanical modeling of the lithospheric deformation with elasto-viscoplastic rheology. Phys. Earth Planet. Inter. 171, 55–75. Regenauer-Lieb, K., Rosenbaum, G., Weinberg, R.F., 2008. Strain localisation and weakening of the lithosphere during extension. Tectonophysics 458, 96–104. Reilinger, R., McClusky, S., Vernant, P., Lawrence, S., Ergintav, S., Cakmak, R., Ozener, H., Kadirov, F., Guliev, I., Stepanyan, R., Nadariya, M., Hahubia, G., Mahmoud, S., Sakr, K., ArRajehi, A., Paradissis, D., Al-Aydrus, A., Prilepin, M., Guseva, T., Evren, E., Dmitrotsa, A., Filikov, S.V., Gomez, F., Al-Ghazzi, R., Karam, G., 2006. GPS constraints on continental deformation in the Africa–Arabia–Eurasia continental collision zone and implications for the dynamics of plate interactions. J. Geophys. Res. 111, B05411. Reimer, P.J., Baillie, M.G.L., Bard, E., Bayliss, A., Beck, J.W., Blackwell, P.G., Bronk Ramsey, C., Buck, C.E., Burr, G.S., Edwards, R.L., Friedrich, M., Grootes, P.M., Guilderson, T.P., Hajdas, I., Heaton, T.J., Hogg, A.G., Hughen, K.A., Kaiser, K.F., Kromer, B., McCormac, F.G., Manning, S.W., Reimer, R.W., Richards, D.A., Southon, J.R., Talamo, S., Turney, C.S.M., van der Plicht, J., Weyhenmeyer, C.E., 2009. IntCal09 and Marine09 radiocarbon age calibration curves, 0–50,000 years cal BP. Radiocarbon 51, 1111–1150. Ritchie, J.C., Eyles, C.H., Haynes, C.V., 1985. Sediment and pollen evidence for an early to mid-Holocene humid period in the eastern Sahara. Nature 314, 352–355. Saneyoshi, M., Nakayama, K., Sakai, T., Sawada, Y., Ishida, H., 2006. Half graben filling processes in the early phase of continental rifting: the Miocene Namurungule Formation of the Kenya Rift. Sed. Geol. 186, 111–131. Stamps, D.S., Calais, E., Saria, E., Hartnady, C., Nocquet, J.M., Ebinger, C.J., Fernandes, R.M., 2008. A kinematic model for the East African Rift. Geophys. Res. Lett. 35, L05304. Stein, R.S., Briole, P., Ruegg, J.C., Tapponnier, P., Gasse, F., 1991. Contemporary, Holocene, and quaternary deformation of the Asal Rift, Djibouti: implications for the mechanics of slow spreading ridges. J. Geophys. Res. 96, 21789–21806. Strecker, M.R., Blisniuk, P.M., Eisbacher, G.H., 1990. Rotation of extension direction in the central Kenya Rift. Geology 18, 299–302. Tiercelin, J.J., Lezzar, K.E., 2004. A 300 million years history of rift lakes in Central and East Africa: an updated broad review. In: Odada, E., Olago, D. (Eds.), The East African Great Lakes: Limnology, Palaeolimnology and Biodiversity. : Advances in Global Change Research. Springer, pp. 3–60. Tongue, J., Maguire, P., Burton, P., 1994. An earthquake study in the Lake Baringo basin of the central Kenya Rift. Tectonophysics 236, 151–164. Truckle, P.H., 1976. Geology and late Cainozoic lake sediments of the Suguta Trough, Kenya. Nature 263, 380–383. Turcotte, D.L., Schubert, G., 1982. Geodynamics: Applications of Continuum Physics to Geological Problems. Cambridge University Press, New York. Ventsel, E., Krauthammer, T., 2001. Thin Plates and Shells. Theory, Analysis, and Applications. M Dekker, New York. 666 pp. Vetel, W., Le Gall, B., 2006. Dynamics of prolonged continental extension in magmatic rifts: the Turkana Rift case study (North Kenya). Geol. Soc., Lond., Spec. Publ. 259, 209–233. Vetel, W., Le Gall, B., Walsh, J.J., 2005. Geometry and growth of an inner rift fault pattern: the Kino Sogo Fault Belt, Turkana Rift (North Kenya). J. Struct. Geol. 27, 2204–2222. Vigny, C., de Chabalier, J.-B., Ruegg, J.-C., Huchon, P., Feigl, K.L., Cattin, R., Asfaw, L., Kanbari, K., 2007. Twenty-five years of geodetic measurements along the Tadjoura– Asal rift system, Djibouti, East Africa. J. Geophys. Res. 112, B06410. Wernicke, B., Burchfiel, B.C., 1982. Modes of extensional tectonics. J. Struct. Geol. 4, 105–115. Wichura, H., Bousquet, R., Oberhansli, R., Strecker, M.R., Trauth, M.H., 2010. Evidence for middle Miocene uplift of the East African Plateau. Geology 38, 543–546. Wright, T.J., Ebinger, C., Biggs, J., Ayele, A., Yirgu, G., Keir, D., Stork, A., 2006. Magmamaintained rift segmentation at continental rupture in the 2005 Afar dyking episode. Nature 442, 291–294. Zielke, O., Strecker, M.R., 2009. Recurrence of large earthquakes in magmatic continental rifts: insights from a paleoseismic study along the Laikipia–Marmanet Fault, Subukia Valley, Kenya Rift. Bull. Seismol. Soc. Am. 99, 61–70.