Water level history for Lake Turkana, Kenya in the past 15,000 years and a variable transition from the African Humid Period to Holocene aridity

Global and Planetary Change 132 (2015) 64–76

Contents lists available at ScienceDirect

Global and Planetary Change journal homepage: www.elsevier.com/locate/gloplacha

Water level history for Lake Turkana, Kenya in the past 15,000 years and a variable transition from the African Humid Period to Holocene aridity C. Bloszies a, S.L. Forman b,⁎, D.K. Wright c a b c

Department of Earth and Environmental Sciences, University of Illinois at Chicago, Chicago, IL 60607, United States Dept. of Geology, Baylor University, Waco, TX 76798, United States Department of Archaeology and Art History, Seoul National University, 1 Gwanak-ro, Seoul, 151-742, South Korea

a r t i c l e

i n f o

Article history: Received 25 November 2014 Received in revised form 17 June 2015 Accepted 19 June 2015 Available online 26 June 2015 Keywords: Lake Turkana water level Relict beaches African Humid Period Holocene aridity Monsoon variability

a b s t r a c t Relict beaches adjacent to Lake Turkana, Kenya provide a record of water level variability for the Late Quaternary. This study focused on deciphering the geomorphology, sedimentology, stratigraphy and 14C chronology of strand plain sequences in the Kalokol and Lothagam areas, at the western margin of the Lake. Nine N30 m oscillations in water level were documented between ca. 15 and 4 ka. The earliest lake level oscillation between ca. 14.5 and 13 ka is not well constrained with water level to at least 70 m above the present surface and subsequently fell to at least 50 m. Lake level increased to at least 90 m between ca. 11.2 and 10.4 ka, post Younger Dryas cooling. Water level fell by N 30 m by 10.2 ka, with another potential rise at ca. 8.5 ka to N 70 m above current level. Lake level regressed by N 40 m at 8.2 ka coincident with cooling in the equatorial eastern Atlantic Ocean. Two major N 70 m lake level oscillations centered at 6.6 and 5.2 ka may reflect enhanced convection with warmer sea surface temperatures in the western Indian Ocean. The termination of the African Humid Period occurred from ca. 8.0 to 4.5 ka and is characterized by highly variable lake level (±N40 m), rather than one monotonic fall in water level. This lake level variability reflects a complex response to variations in the extent and intensity of the East and West African Monsoons near geographic and topographic limits within the catchment of Lake Turkana. Also, for this closed lake basin excesses and deficits in water input are amplified with a cascading lake effect in the East African Rift Valley and through the Chew Bahir Basin. The final regression from a high stand of N90 m above the present lake began at 5.2 ka and water level was below 20 m by 4.5 ka; and for the remainder of the Holocene. This sustained low stand is associated with weakening of the West African Monsoon, a shift of the mean position of Congo Air Boundary west of the Lake Turkana catchment and with meter-scale variability in lake level linked to Walker circulation across the Indian Ocean. © 2015 Elsevier B.V. All rights reserved.

1. Introduction The mechanisms of Late Quaternary climatic transitions in the tropics are complex, non-linear and spatially variable. The hydrologic variability of such systems often reflects internal (e.g. sea surface temperatures) and external (e.g. insolation) climate drivers which can impact the nature of monsoonal variability, particularly on a warming planet from glacial to interglacial conditions and with human-induced warming (e.g. Toreti et al., 2013; Weldeab et al., 2014). Rainfall proxies for eastern Africa and the Greater Horn of Africa (GHA) (Fig. 1) indicate a dramatic shift from predominantly wet conditions, known as the African Humid Period (AHP) to sustained dry conditions by ca. 5 ka (e.g. Garcin et al., 2009; Tierney et al., 2011a; Berke et al., 2012b; Foerster et al., 2012; Tierney and deMenocal, 2013). The AHP resulted in increased vegetation cover (Castañeda et al., 2007; Damsté et al., 2011), surface ⁎ Corresponding author. E-mail address: [email protected] (S.L. Forman).

http://dx.doi.org/10.1016/j.gloplacha.2015.06.006 0921-8181/© 2015 Elsevier B.V. All rights reserved.

water availability (Gasse et al., 2008; Verschuren et al., 2009; Blanchet et al., 2013), lake high stands (Ritchie et al., 1985; Street-Perrott et al., 1989; Kuper and Kröpelin, 2006), and hydrologic connection of presently endorheic lakes in eastern Africa (Nyamweru, 1989; Garcin et al., 2009; Johnson and Malala, 2009). Previous studies indicate significant hydrologic variability during the AHP in equatorial Africa associated with large freshwater contributions into the North Atlantic, such as during the Younger Dryas (YD) and the 8.2 ka event (Weldeab et al., 2005; Garcin et al., 2007; Junginger et al., 2014; Costa et al., 2014). Reduced monsoonal precipitation during the YD may be linked to suppressed sea surface temperatures (SSTs) in the eastern Atlantic Ocean (Weldeab et al., 2005, 2007) and possible weakening of the East African Jet (Weldeab et al., 2014). Drying across equatorial Africa is inferred during the YD (Schefuß et al., 2005; Weldeab et al., 2005; Garcin et al., 2007; Gasse et al., 2008; Foerster et al., 2012) and linked to a limited westward migration of the West African Monsoon (Gasse et al., 2008; Tierney et al., 2011b; Costa et al., 2014) and possible weakening of the East African Monsoon (Garcin et al., 2007). This

C. Bloszies et al. / Global and Planetary Change 132 (2015) 64–76

Fig. 1. Central and East Africa and seasonal atmospheric convergence zones (dashed lines). Intertropical Convergence Zone (ITCZ) and Congo Air Boundary (CAB) for boreal winter (gray) and summer (black) (from Costa et al., 2014). The gray shaded areas represents East African topography N1500 m.a.s.l. (SRTM data). Locations of paleoclimate proxies (yellow dots) discussed in the text: (1) Gulf of Guinea SSTs (Weldeab et al., 2007, 2011), (2) Congo Basin δDwax, lakes (3) Albert; (4) Tanganyika; (5) Victoria; (6) Masoko; (7) Malawi; (8) Tana; (9) Ziway-Shala; (10) Abaya-Chamo; (11) Chew-Bahir; (12) Turkana, (13) Suguta, (14) Baringo and Bogoria, (15) Nakuru, Elmenteita and Naivasha, (16) Magadi and Natron, (17) Mt. Kilimanjaro ice core (Thompson et al., 2002), lakes (18) Challa, and (19) Abhe, (20) Gulf of Aden δDwax, (21) Mogadishu dunes, (22–23) western Indian Ocean and Straits of Madagascar SSTs (Bard et al., 1997).

dry period is broadly concordant with low stands post ca. 12.5 ka for lakes Magadi–Natron (Roberts et al., 1993), Ziway and Shala (Gillespie et al., 1983), Turkana (Morrissey and Scholz, 2014), and

65

Suguta (Junginger, 2011). Various hydroclimatic records indicate pronounced drying in the GHA and eastern Africa between ca. 8.5 and 7.8 ka (Gasse, 2000; Junginger, 2011; Costa et al., 2014) and were possibly coeval with the 8.2 ka meltwater event in the North Atlantic Ocean (Alley et al., 1997). A terminal shift from humid to arid conditions across northern and eastern Africa occurred during the Middle Holocene. Analyses of δDwax from sediment cores for eastern African lakes (Fig. 1) indicate appreciable drying between 6 and 4.5 ka (e.g. Marshall et al., 2009; Tierney et al., 2010, 2011b). Probabilistic age models for δDwax records from lakes Challa and Tanganyika and the Gulf of Aden refine the end of the AHP to 4960 ± 70 cal. yr B.P., with the transition to aridity occurring within 280–460 years (Tierney and deMenocal, 2013). However, δDwax records from the Ethiopian Highlands suggest that drying may be earlier for higher elevations sites, between 7 and 8 ka, possibly linked to the reduced eastward flux of Atlantic-derived moisture (Costa et al., 2014). Also, water levels for Lake Turkana appear to be highly variable between 8.5 and 4.5 ka, with at least three inferred N60 m oscillations (Forman et al., 2014). The final drop from a N80 m high stand above current lake level started sometime between 5.4 and 5.0 ka, with lake level falling below 20 m at 4.6 ka, though this final regression is not well constrained chronologically (Forman et al., 2014). Critical records to assess hydrologic variability in the past 15 ka in East Africa are lake level reconstructions derived from the study of well-preserved relict beaches that surround Rift Valley lakes. The relict beaches adjacent to Lake Turkana, Kenya are particularly prominent, with minimal vegetation cover and occur up to ~100 m above the current shore (Fig. 2). Radiocarbon ages on shells from the highest beaches indicate that a substantially larger lake occurred multiple times between 15 and 5 ka (e.g. Butzer et al., 1972; Butzer, 1980; Harvey and Grove, 1982; Owen et al., 1982; Brown and Fuller, 2008; Garcin et al., 2012; Forman et al., 2014). Unfortunately, there are inconsistencies amongst the reconstructions of water levels for Lake Turkana as there is often insufficient age and elevational control to assess submillennial-

Fig. 2. Topography of the Lake Turkana catchment and lake dimensions during the AHP associated with water level ~100 m above current lake level. (a) The current Lake Turkana Basin (solid red line), with Omo, Turkwel and Kerio river sub-basins (dotted red line). Maximum high stands for lakes (light blue), cascading basins (dashed red lines) and locations of basin sills (yellow arrows) from SRTM data and previous studies (Butzer et al., 1972; Grove et al., 1975; Garcin et al., 2009; Foerster et al., 2012). Adapted from Junginger and Trauth (2013). (b) Littoral deposits, major faults and sample locations for the western strand plain. Sample locations (red dots) with numbers correspond to Table 1. (c) The Kalokol beach ridge sequence. (d) The Lothagam area with elevation points are from GPS measurements. Satellite imagery from Google Earth.

66

C. Bloszies et al. / Global and Planetary Change 132 (2015) 64–76

scale variability in water level (Fig. 3). This study presents new geomorphic, sedimentologic and stratigraphic data and associated 14C ages on mollusks from Late Quaternary relict beaches along the western margin of Lake Turkana. These data are integrated with prior lake level reconstructions from the eastern margin of Lake Turkana (Forman et al., 2014), South Island (Garcin et al., 2012) and pioneering studies of lake level variability (Butzer, 1980; Owen et al., 1982; Brown and Fuller, 2008). The objective of this study is to further document water level variability for Lake Turkana during the past 15 ka, particularly with the transition from the AHP, ca. 8 to 4 ka and relate to changes monsoonal circulation. 2. Climate and hydrology for the Lake Turkana basin Currently, a significant but secondary moisture source to Lake Turkana occurs with the expansion of the West African Monsoon in summer which transports Atlantic-derived and Congo Basin recycled moisture eastward (Nicholson, 2000b; Williams et al., 2012; Costa et al., 2014). Monsoonal rainfall over the western margin of Lake Turkana Basin usually occurs at the Congo Air Boundary (CAB), which is the convergence of air masses derived from the Indian and Atlantic oceans (Fig. 1), and often coincides with a precipitation maximum over Central Africa (McGregor and Nieuwolt, 1998; Nicholson, 2000a). Sea surface temperature variability (SST;~24 to 28 °C) in the equatorial eastern Atlantic is linked to hydrologic variability in equatorial Africa, with elevated SSTs coeval with a zonally expanded West African Monsoon (e.g. Tierney et al., 2011b) and lower SSTs is associated with Holocene drought (Weldeab et al., 2005, 2011). A sustained eastward extension of the CAB is implicated for higher lake levels during the AHP in eastern Africa (Tierney et al., 2011b; Costa et al., 2014; Junginger et al., 2014; Morrissey and Scholz, 2014).

Anomalous SSTs in the western Indian Ocean in the 20th and 21st centuries can modulate atmospheric convection and the availability of precipitable water for eastern Africa, and affect the strength and the duration of passage of the East African Monsoon (Goddard and Graham, 1999; Black et al., 2003; Saji and Yamagata, 2003). Warm SSTs (~ 26 to 29 °C) adjacent to eastern Africa enhance atmospheric convergence and result in increased boreal fall precipitation across Kenya, Ethiopia and Somalia (Ummenhofer et al., 2009; Levin et al., 2009; Becker et al., 2010; Bloszies and Forman, 2015). On balance, cooler SSTs (~ 24 to 25 °C) result in a strengthened dry southeasterly Turkana Jet (Kinuthia and Asnani, 1982; Nicholson, 1996), associated with appreciably less vapor transport, often below the threshold for precipitable water (cf. Nicholson, 1996, 2000b; Marchant et al., 2007). The current Lake Turkana level at 362 m.a.s.l. is the zero datum for discussing changes in water level in the Late Quaternary (cf. Velpuri et al., 2012). The inferred outlet elevation for Lake Turkana in the past ca. 15 ka has been estimated to be 95 to 100 m (Butzer et al., 1972; Owen et al., 1982; Garcin et al., 2012; Forman et al., 2014). The presumed outlet is located east of Lotikipi Swamp in the Ilemi Triangle with drainage into the White Nile catchment (Butzer et al., 1972; Owen et al., 1982; Fig. 2a). High stands for Lake Turkana in the past 15 ka reflect elevated monsoonal precipitation (Junginger and Trauth, 2013; Forman et al., 2014), amplified by water input from cascading lake systems to the south and to the northwest (Fig. 2a). In the East African Rift Valley at least five, lower-elevation, endorheic lakes may have overflowed sequentially and numerous times during the AHP, with final discharge into Lake Turkana (Fig. 2a). This interconnected hydrologic system added a 62,000 km2 area to the Lake Turkana catchment. The highest lakes in this system are Nakuru and Elmenteita, which coalesced to form a larger lake with a maximum high stand at

Fig. 3. Previous water level reconstructions for Lake Turkana: (a) the Kibish strand plain and associated beach features from the Omo Valley (Butzer, 1980), (b) the Koobi Fora area (Owen et al., 1982), (c) additional sampling from the Kibish area, and other areas around the lake (Brown and Fuller, 2008), (d) South Island (Garcin et al., 2012). (e) Mt. Porr radiocarbon and optical ages (Forman et al., 2014).

C. Bloszies et al. / Global and Planetary Change 132 (2015) 64–76

1940 m.a.s.l. Outflow from this larger lake cascaded northward into lakes Baringo and Bogoria (Butzer et al., 1972; Dühnforth et al., 2006). Lakes Baringo and Bogoria overflowed when water level rose to 996 and 991 m.a.s.l., respectively, and this discharge then flowed into the ephemeral Lake Logipi via the Suguta River (Junginger and Trauth, 2013). Lake Logipi overflowed at 581 m.a.s.l. into the Kerio River, which discharged into Lake Turkana (Garcin et al., 2009; Junginger and Trauth, 2013). Overflow from the Chew Bahir Basin to the north in the Late Quaternary has also contributed to high stands for Lake Turkana (Fig. 2a). This expanded lake system added ~39,000 km2 to the Lake Turkana watershed from the central Ethiopian Highlands. The start of this cascading hydrologic system was Lake Abaya, with outflow into Lake Chamo that discharged into the Segen River and this river flowed finally into the Chew Bahir Basin (Grove et al., 1975; Foerster et al., 2012). The rise in paleo-Lake Chew Bahir to 544 m.a.s.l. resulted in overflow into Lake Turkana (Foerster et al., 2012). 3. Previous studies of Lake Turkana water level Several studies in the past 45 years have focused on deciphering the ages of relict strand plains surrounding Lake Turkana and the associated archaeological record (Fig. 3). The earliest studies used 15 conventional 14 C ages on mixed shell collections to reconstruct lake level at ~70 and 90 m between ca. 11.5 and 4 ka, with low stands prior to 11.5 ka and at ca. 10.8, 9.8, 8.0, 4.4 and post 3.5 ka (Butzer and Thurber, 1969; Butzer et al., 1972; Butzer, 1980) (Fig. 3a). Owen et al. (1982) constrained lake level from 14C ages of shell, bone, and charcoal near the Koobi Fora spit. This study inferred a sustained lake level fluctuating between ~ 80 and 50 m from 11.5 to 3.8 ka, which was interrupted by two low stands at ca. 8.0 and 4.5 ka (Fig. 3b). Lake level was constrained to below ~ 15 m post ca. 2.5 ka based on archaeological evidence. A study of the Late Quaternary stratigraphy near Kibish incorporated previously reported ages from Koobi Fora, Kibish, and Lothagam areas (Brown and Fuller, 2008), and depicts lake levels higher than 50 m between ca. 13 and 4 ka interrupted by regressions to b30 m at ca. 11.4, 10.6 and 7.9 ka (Fig. 3c; Brown and Fuller, 2008). Inferred high stands are brief, possibly reaching the spill elevation at ca. 9.4, 9.0, and 6.9 ka with sustained high water levels of N 75 m until ca. 4 ka (Brown and Fuller, 2008). A study of relict beaches of South Island constrained past lake level by AMS 14C ages on single mollusk valves or gastropod tests and integrated these results with ages from the previous studies cited above. The elevations of relict beaches were adjusted for an inferred subsidence rate of ~2 mm/yr, calculated from an assumed elevation of 455 m.a.s.l. for the high stand limit. This study refined the timing of a low stand to ~50 m sometime between 12.6 and 11.7 ka, possibly coincident with the YD chronozone (Fig. 3d; Garcin et al., 2012). Lake level is inferred to drop by ~ 60 m at 5 ka to historic levels or below and remained low for the remainder of the Holocene. A recent reconstruction of Lake Turkana water level for the past ca. 8.5 ka used AMS 14C ages on shells and optically stimulated luminescence (OSL) dating of littoral sediments from relict beaches in the Mt. Porr area on the east side of Lake Turkana (Forman et al., 2014). Three water level oscillations of N70 m are depicted at ca. 7.0, 6.4, and 5.2 ka, with a low stand (b20 m) post 4.6 ka (Fig. 3e; Forman et al., 2014). This study also integrates previous 14C ages but ranked the accuracy of ages based on the shell material dated, the uncertainty in elevation, and sedimentologic relation to a relict shoreline. Finally, seismic stratigraphy for Lake Turkana and AMS 14C datedsediment core infers lower-than-present lake levels from the Last Glacial Maximum (LGM) until 11 ka (Morrissey and Scholz, 2014). A significant seismic unconformity is inferred as evidence for a lowerthan-present lake level from ca. 4 ka until 2 ka (Morrissey and Scholz, 2014). Recent studies underscore that the rate and magnitude of lake level variations exceeds the tempo of orbital precession, thus changes

67

in the position of the CAB and associated monsoon-derived moisture may more satisfactorily explain variability in lake level since ca. 15 ka (Brown and Fuller, 2008; Garcin et al, 2012; Forman et al., 2014; Morrissey and Scholz, 2014). 4. Materials and methods 4.1. Field research area This study deciphered the relict strand plain along the southwestern margin of Lake Turkana (Fig. 2b). The stratigraphy and sedimentology of littoral, sublittoral and deeper water deposits within relict beaches was studied and in situ fresh-water shells were retrieved for 14C dating, providing ages on former lake levels. Relict beaches between 15 and 40 m were targeted because there is limited age control for these surfaces and additional data is needed to test previously inferred low stands (Forman et al., 2014). Lacustrine sequences between 80 and 105 m were also examined to better understand the timing of overflow intervals for Lake Turkana. Elevations were determined with a dual frequency Trimble XM GPS receiver using Terrasync software. Data were postprocessed using Trimble's Pathfinder Office 3.10 software with RINEX data obtained from the UNAVCO-operated base station at Arba Minch University, Ethiopia. Vertical errors at 1 − σ are 2.0 to 2.5 m when satellite geometry PDOP values were ≤2.5. 4.2. Radiocarbon dating of mollusks from lacustrine sediments Previous studies have documented that relict beaches and deeper water deposits occur up to ~ 100 m and often contain well-preserved freshwater mollusks (cf. Owen and Renaut, 1986; Garcin et al., 2012; Forman et al., 2014). The two most common species are Melanoides tuberculata and Corbicula fluminalis africana, which live in maximum water depths of 2 and 5 m, respectively (Cohen, 1986; Leng et al., 1999; Genner and Michel, 2003; De Kock and Wolmarans, 2007). C. africana is a small (~ 2 cm) bivalve found in shallow, turbid, water bodies across Africa including fluvial channels (Korniushin, 2004; De Kock and Wolmarans, 2007). M. tuberculata is a gastropod which feeds on muddy detritus at depth or in backshore lagoons in depths b2 m (Genner and Michel, 2003). The other species recovered from Lake Turkana sediments are larger bivalves, the N15-cm-long Etheria elliptica and the smaller Unionidae nitia chefneuxi. E. elliptica is commonly found in lacustrine muds between 10 and 30 m water depths (Van Bocxlaer and Van Damme, 2009), whereas U. chefneuxi shells typically occur at depths of b 2 m (Abdallah and Barton, 2003). Radiocarbon ages on shells (Table 1) were determined by accelerator mass spectrometry (AMS). Samples were submitted to the AMS Laboratory at Seoul National University, South Korea (SNU #), the Illinois State Geological Survey, Urbana, Illinois, U.S.A. (ISGS #) or to the Arizona AMS Laboratory at the University of Arizona (AA #). Most 14C ages have been calendar corrected and designated as “cal. ka” (Fairbanks et al., 2005), except one age that is infinite. The association of 14C ages on mollusks to past lake level is dependent on the depositional environment of the surrounding sediments and if the shells are in situ or transported a short distance. We confined sampling for 14C dating to highly visible concentrations (coquinas) of genera such as Corbicula and Melanoides imbricated within littoral and sublittoral deposits, which indicate transportation en masse from deeper water with increased fetch, possibly during storm events. Thus, there should be little to no time (subdecadal to multidecadal) between death of the organism and deposition into beach sediments. Smaller (2–4 mm in diameter), individual gastropod shells or valves with well-preserved ornamentation and periostracum were submitted for AMS 14C dating to quantify the time since deposition of the thanatocoenosis (Table 1). These small, thin, and delicate juvenile shells with ornamentation would not survive in this pristine state upon reworking. The suitability of 14C dating of small single tests or valves

68

C. Bloszies et al. / Global and Planetary Change 132 (2015) 64–76

Table 1 Radiocarbon ages on shell, fish bones and beach rock from relict beaches, western shore, Lake Turkana, Kenya. Field #

Laboratory #

Genera

13

LT12-02 LT12-04 LT12-06 LT12-08 LT12-09 LT12-11 LT12-14 LT12-17 LT12-18 LT12-36 LT12-40 LT12-41 LT12-42a LT12-42b LT12-43

SNU12-589 SNU12-590 ISGS-A2781 ISGS-A2781 SNU12-591 AA100110 SNU12-592 SNU12-593 ISGS-A2783 SNU12-594 SNU12-595 SNU12-596 SNU12-597 SNU12-598 SNU12-599

Melanoides Melanoides Melanoides Melanoides Melanoides Melanoides Melanoides Melanoides Unionida Melanoides Melanoides Etheria Melanoides Corbicula Melanoides

−12.60 −2.69 −3.4 −1.2 −5.22 −1.4 −9.55 −2.12 −1.2 −1.02 −1.96 1.66 −2.07 −20.22 1.26

a b

C (‰)

Laboratory age (yr)

Calendar corrected age (yr)a

Depositional environment

Collection elevation (m)b

Inferred paleo lake level

4510 ± 60 6100 ± 50 12,585 ± 40 4710 ± 20 4980 ± 50 8008 ± 51 6970 ± 70 11,830 ± 70 N51,000 9740 ± 70 9390 ± 70 9530 ± 60 9430 ± 70 9540 ± 100 9230 ± 60

5180 ± 120 6960 ± 70 14,645 ± 95 5430 ± 30 5700 ± 65 8920 ± 100 7800 ± 80 13,680 ± 70 N51,000 11,175 ± 75 10,610 ± 95 10,830 ± 140 10,670 ± 110 10,850 ± 190 10,395 ± 100

Upper foreshore Upper foreshore Upper sublittoral Upper sublittoral Upper foreshore Upper sublittoral Upper foreshore Upper foreshore Lower foreshore Upper estuarine Upper foreshore lacustrine Upper foreshore Upper foreshore Sublittoral

89.0 ± 2.3 38.0 ± 3.4 24.8 ± 2.2 81.0 ± 2.0 37.9 ± 2.3 39.5 ± 2 36.7 ± 2.6 45.2 ± 2 43 ± 2 93.1 ± 2.7 91.3 ± 2.3 84.3 ± 3 91.3 ± 2.6 91.3 ± 2.6 92.0 ± 2.2

90 ± 2 38 ± 3 28 ± 2 83 ± 3 38 ± 2 42 ± 3 38 ± 3 45 ± 2 43 ± 2 94 ± 3 91 ± 2 90–100 91 ± 3 91 ± 3 95 ± 2

Calendar corrected from http://radiocarbon.ldeo.columbia.edu/cgi-bin/radcarbcal; Fairbanks et al. (2005). Elevation from current lake surface elevation of 362 m.a.s.l. (Velpuri et al., 2012).

was tested in the Mt. Porr area, located on the east side of the lake from this study's sites, and paired shell 14C and quartz OSL ages are concordant at one sigma errors (Forman et al., 2014). The efficaciousness of this dating approach was also tested at the Lothagam site with 14C dating of a paired, in-situ shell of E. elliptica with periostracum from a deeper water lacustrine facies with an inferred lake level of 90–100 m and compared to 14C ages on single tests/valves of M. tuberculata and C. fluminalis africana from correlative littoral facies with an estimated lake level of 93 to 95 m above current lake level. As expected, the Melanoidies spp. and Corbicula sp. shells yielded concordant ages (at two sigma errors) of 10,670 ± 110 (SNU12-597) 10,395 ± 100 (SNU12-599) and 10,850 ± 190 (SNU12-598) cal. yr BP, respectively with an age of 10,830 ± 140 cal. yr. BP (SNU12-596) on an in situ E. elliptica shell. These tests indicate that the small shells dated are not reworked from older sediments and are penecontemporaneous with the enclosing littoral sediments. In arid ecosystems in volcanic settings, a common problem in 14C dating is the presence of a reservoir effect of recycled “old” carbon into the materials being dated (e.g., Junginger et al., 2014). However, there is no evidence for a 14C reservoir affect for Lake Turkana with modern M. tuberculata shells yielding post-atomic bomb 14C levels (Garcin et al., 2012). No demonstrable 14C reservoir effect is also inferred for large (N150 μm) ostracode carapaces from a Lake Turkana sediment core which returned stratigraphically consistent ages and core top ages of 60 to 1200 yr B.P. consistent with variable core recovery (Halfman et al., 1994). In turn, concordant OSL and 14C ages from the Mt. Porr strand plain on the east side of Lake Turkana adds further support for a small (decadal to century), but indeterminate reservoir effect for much of the Holocene.

5. Results 5.1. Geomorphology of the Kalokol strand plain The relict strand plain near the town of Kalokol (Fig. 2c) flanks the western highlands of the Gregory Rift and trends broadly parallel to the dominant axis of rifting (~ N 5°E; Vétel et al., 2004; McDougall and Brown, 2009). The Lothidok highlands (Fig. 2b) reflect rift extension, with movement of the Lothidok Fault and a single fault block is inferred to underlie the western strand plain (Vétel et al., 2004; McDougall and Brown, 2009). The youngest offset deposit is N100 ka in the lower Kibish Formation (Brown and Fuller, 2008; Gathogo et al., 2008; Brown and McDougall, 2011). Careful scrutiny of remotely sensed images and geomorphic and stratigraphic assessments in the vicinity of

these faults document no perceptible vertical or horizontal offsets of relict beach-ridges. The highest beach ridges studied in the Kalokol sequence occur to 90 to 95 m high. A 1-meter high erosional scarp (see Fig. 4a) at 95 ± 2 m is interpreted as wave-cut notch associated with a high stand. There is a distinctive sequence of relict beaches between 40 and 50 m composed of parallel strandlines with broad crests (100 to 200 m) trending north to south. These strandlines are surprisingly uniform with regular horizontal separation (10 to 20 m), a common gradient (9 m/km) and may reflect a monotonic fall in lake level. In contrast, between 30 and 40 m there is a noticeable decrease in gradient to 6 m/km with indistinct strandlines covered by eolian sand, associated with a decreased rate of lake level regression or stasis. The slope decreases further below 30 m to ~ 3 m/km with common large coppice dunes and a sand sheet deposit. Also, there is a prominent beach ridge at 13 m that is b 50 m wide and ~ 2 m high, anchored by large doum palms (Hyphaene compressa).

5.2. Sedimentology, stratigraphy and geochronology of the Kalokol strand plain The sequence of Late Pleistocene to Holocene lacustrine sediments is part of the Galana Boi Formation, which has variable preservation and stratigraphic continuity around Lake Turkana (Owen et al., 1982; Owen and Renaut, 1986). The sedimentology of beach ridges for the Kalokol strand plain is well exposed in a series excavations and natural stratigraphic sections (Fig. 4). Many sites show a silt-rich, deep-water facies often underlying littoral and sublittoral deposits, which reflects lake transgression. Deeper water lacustrine deposits are commonly a silty, very fine sand to a silty clay and occur in millimeter-scale planar beds (Fig. 4b). Sublittoral deposits are typically a very well sorted fine to medium sand with weak or absent bedding, which reflects bioturbation or dewatering (Fig. 4c). Lower shoreface littoral deposits are comprised of well sorted fine to medium sands with millimeter-scale planar to sub-planar beds that dip broadly lakeward (Fig. 4e). The most common depositional facies is a well sorted, medium to coarse sand that occurs in centimeter to decimeter thick beds, with lenses of 1-to-5 mm diameter pebbles (Fig. 4d). These beds are inclined b10°, include shell concentrations of M. tuberculata and C. fluminalis africana, and indicate upper foreshore environments (Reading, 2009:175–177). Sediments interpreted to reflect backshore facies (Fig. 4f) are distinguished by pebble-tocobble-rich, moderately well sorted coarse sand, probably deposited with wash over during storm events.

C. Bloszies et al. / Global and Planetary Change 132 (2015) 64–76

Fig. 4. Depositional facies for stratigraphic sections in the western strand plain and the Lothagam tombolo. (a) Wave-cut notches at ~93 m elevation. (b) Lacustrine clays and silts, with thin (mm-scale) planar bedding. (c) Sublittoral medium to coarse sands. Note dewatering feature apparent in bedding. (d) Upper shoreface medium to coarse sands with granular and gravel interbeds. (e) Well sorted lower shoreface sands. Interbeds are poorly sorted storm deposits. (f) Backshore deposits. (g) Carbonate rhizoliths at Lothagam. (h) M. tuberculata coquina at Lothagam. (i) Rounded pebbles of Lothagam tombolo surface.

The magnitude and timing of lake level changes are derived from stratigraphic observations (Fig. 5) and associated 14C ages (Table 1) for relict beaches of the Kalokol strand plain. Littoral sediments are documented at the Ekai Site (89 m) with a basal, very dark grayish-brown (10YR 3/2), well sorted, medium to coarse sand with scattered 1-to5 cm diameter rounded pebbles, which reflects an upper foreshore or backshore environment. This littoral deposit appears to lap onto basaltic basement. Capping the gravel is ~1 m thick, moderately sorted, brown (10YR 6/6), medium sand with lens of small pebbles and granules of basalt interpreted as a locally derived colluvium. A single test of M. tuberculata from the littoral sediment gave a 14C age of 5.18 ± 0.12 cal. ka (SNU12-589). The Glynnis Site exposes a ~ 3.5 m section of littoral deposits of various ages within the 81 m beach ridge (Fig. 5). The basal unit (1) is an extensively pedogenically-modified, brown (7.5YR 5/4), moderately sorted clay loam interpreted as a truncated Btkb horizon with visible argillans and stage 1 filamentous carbonate. Unit 2 is a very well sorted, grayish brown (10YR 5/2) fine sand with discontinuous centimeter-scale horizontal stratification; small sub-millimeter carbonate concretions are common. The top unit (3) is a ~ 1-m thick, pale

69

brown (10YR 6/3) medium to coarse sand in centimeter-scale beds deposited in a lower shore face environment; a single test of M. tuberculata from this deposit returned a 14C age of 5.43 ± 0.03 cal. ka (ISGS-A2782). The Kalokol strand plain is dissected by multiple ephemeral streams, which provide exposures of relict beaches between 30 m and 50 m. The Big Ridge Site is a ~2.5 m vertical section cut into the 45 m beach ridge (Fig. 5). The basal unit (1) is a poorly sorted, coarse sand, and crudely stratified, with abundant pebbles and cobbles, which reflects upper shore-face deposition. Abundant shells occur in lenses of bedded, medium-to-coarse sands. Overlying unit 2 is a concentration of well rounded pebbles and cobbles, probably lagged from subjacent unit 1. The surface unit (3) is a stratified, moderately sorted, and medium to very coarse sand, with abundant pebbles and a few cobbles reflecting an upper shore-face environment. A valve of Unionidae sp. from unit 1 yielded a 14C age of N 51 ka (ISGS-A2783), whereas a test of M. tuberculata from unit 3 returned a 14C age of 13.68 ± 0.07 cal. ka (SNU12-593). The Lunch Break Site is a vertical section exposing a variety of deposits in a stream cut bank ~ 0.5 km to the east of the Big Ridge Site (Fig. 5). The basal unit 1 is a pale brown (10YR 6/3), massive, slightly silty, well sorted fine sand with scattered small pebbles reflecting sublittoral deposition. The overlying unit 2 is a fining-upward sequence with stratified pebbly sands in the lower 20 cm and light yellowishbrown (10YR 6/4) very well sorted, fine to medium sand in the upper 0.5 m and reflects shore face to back shore deposition. The top of unit 2 was modified pedogenically with a 20-cm thick brown (10YR 5/3) cambic horizon, indicating a period of non-deposition and subaerial exposure. A massive, very well sorted, fine sand (unit 3) of eolian origin overlies this buried soil. This sedimentologic sequence reflects lake transgression and regression, with final burial by an eolian sand sheet. A single test of M. tuberculata from unit 2, a backshore deposit, yielded a 14C age of 7.8 ± 0.08 cal. ka (SNU12-592). An excavation into a relict beach at 40 m at the N 'chok site exposes a sequence of littoral deposits (Fig. 5). The lowermost unit (1) is a brown (10YR5/3), very well sorted, medium-to-fine sand with centimeterscale subhorizontal bedding and dewatering structures. This shoreface sand exhibits sub-planar bedding, dipping towards 10°E and the present lakeshore. Overlying this littoral sand is a massive, very well sorted, fine sand, similar to unit 3 at the Lunch Break Site, which is probably an eolian cover sand. A single M. tuberculata shell from ~ 0.5 m below the beach ridge surface yielded a 14C age of 8.92 ± 0.1 cal. ka (AA-100110). The Kalokol #2 site also exposes a littoral sequence with abundant shells. This site is characterized by a coarsening upward sequence, with stratified, poorly sorted coarse sand with abundant pebbles transitioning to a moderately well sorted medium to coarse sand reflecting lower shoreface to backshore deposition. A single M. tuberculata shell from the lower shoreface sediments gave a 14C age of 6.96 ± 0.07 cal. ka (SNU12-590). At the south margin of the Kalokol strand plain, near Eliye Springs, the Galana Boi site at 38 m featured steep badlands topography, dissected fluvially and with little vegetative cover. A ~7 m section exposes a variety of lacustrine sediments (Fig. 5). The bottom unit (1) is a poorly sorted gravel with an overlying ~ 6-m-thick, mica-rich, stratified silt and fine sand (unit 2). The section is capped by a 20-cm thick, poorly sorted, gravelly sand with abundant shells and in places subsequently buried by a very well sorted eolian fine sand. This sedimentologic sequence reflects transgression and regression of the Lake with the littoral gravels (units 1 and 3) bounding a deep water stratified silt and sand (unit 2). A M. tuberculata shell from unit 2 gave a 14C age of 5.7 ± 0.07 cal. ka (SNU12-591) and constrains the timing of the last transgression at this site. The surface expression of relict beaches between 10 and 30 m elevation is muted with burial by an ubiquitous eolian sand, though littoral deposits are well exposed in stream bank sections. The Goat Site at 25 m is a ~ 5-m high, south-facing exposure eroded at the banks of a seasonal river (Fig. 5). The basal unit is a ~ 2 m thick,

70

C. Bloszies et al. / Global and Planetary Change 132 (2015) 64–76

Fig. 5. Stratigraphy and sedimentology and associated calendar corrected 14C ages (Fairbanks et al., 2005) for exposures for Kalokol strand plain and Lothagam sites at 83 to 93 m (in box). All elevations are relative to 362 m.a.s.l., the current water level for Lake Turkana (Velpuri et al., 2012).

gray (Gley 1 5/10Y), very well sorted medium sand with pebble lenses which reflects sub-littoral deposition. This unit also contains common carbonate nodules, 1-to-4 cm in diameter; the top contact with unit 2 is abrupt. Unit 2 is a very well sorted, medium to fine sand with scattered small rounded pebbles, and 1 to 5 millimeterscale beds with distinctive variations in color from a very dark gray (10YR 3/1) to a light yellowish brown (10YR 6/4). This unit also contains lenses of coarse sand and pebbles with well preserved shells, which is indicative of lower foreshore to upper sublittoral

environments. Unit 3 is composed of medium to coarse sands, with 10 to 20 cm thick beds of very coarse sands with abundant pebbles and is an upper shoreface deposit. Unit 4 is a fining-upward sequence from a very coarse, poorly sorted sand with abundant pebbles to a loose medium sand and reflects backshore deposition with possible bioturbation in the top 60 cm. The capping unit is a massive, very fine to fine sand with translocated silt. A single M. tuberculata shell from unit 2 in the sublittoral facies yielded a 14C age of 14.6 ± 0.10 cal. ka (ISGS-A2781; Fig. 5).

C. Bloszies et al. / Global and Planetary Change 132 (2015) 64–76

5.3. Geomorphology of the Lothagam tombolo

6. Discussion

Littoral deposits at Lothagam trend broadly east to west (~ N80°W) and form a prominent relict tombolo (Feibel, 2003). The tombolo is approximately 600 m long and 180 m wide, and the eastern margin widens to N300 m with attachment to the basaltic bedrock below Lothagam Hill (Fig. 2d). In contrast, the western margin is dissected by ephemeral streams, which expose the relict tombolo stratigraphy. The southern and northern slope gradients of this relict tombolo are 11 m/km and 36 m/km respectively, the latter reflecting probable erosion with regression. The surface elevation of the Lothagam tombolo is consistently 92 ± 2 m and rises to 96 ± 2 m with attachment to the eastern Lothagam ridge. A layer of rounded basaltic pebbles (2 to 5 cm diameter; Fig. 4i) covers the upper surface of the tombolo, reflecting wave washing. These rounded pebbles were observed up to 102 ± 3 m on the talus slope of the eastern ridge of the Lothagam Highlands, which reflects a maximum elevational limit for wave washing.

6.1. Inferred Lake Turkana submillennial-scale water level variations

5.4. Lothagam sedimentology, stratigraphy and radiocarbon ages The stratigraphy of the Lothagam tombolo provides insight into sedimentation near the transgressive limit for Lake Turkana (Fig. 5). The depositional sequence within the relict tombolo surface is characterized by a basal, well-laminated lacustrine silt and very fine sand with a succession to a bedded, sublittoral, fine to medium sand and capped by a ~ 5 m thick littoral coarse sand and gravel. The lacustrine sediments are typically a silt to very fine sand and interbedded with centimeter-scale beds of granules. These facies contain numerous in situ paired E. elliptica with periostracum, and many in vertical position, which indicates a biocoenosis. A single E. elliptica valve from this facies yielded the 14 C age of 10.8 ± 0.14 cal. ka (SNU12-596). Based on the habitat requirement of this taxon (Van Bocxlaer and Van Damme, 2009) and associated sedimentologic features, deposition occurred in a deep-water environment (N10 m), with an associated lake level N90 m. At the Tombolo West Site a well sorted fine to medium sublittoral sand, rich in mafic minerals, overly lacustrine sediments. In the upper 60 cm of these sands there are centimeter-scale interbeds of granules with pebbles, which may reflect storm deposits (Fig. 5). There are prominent carbonate rhizoliths in the top 20 to 40 cm of littoral deposits (Fig. 4g) in exposures on the north side of the tombolo, which suggests a period of subaerial exposure with woody vegetation, post regression. Single M. tuberculata and C. fluminalis africana shells taken from the sublittoral sand and below the interbedded layers returned 14 C ages of 10.67 ± 0.11 cal. ka (SNU12-597) and 10.85 ± 0.19 cal. ka (SNU12-598), respectively. A M. tuberculata shell taken from a granular interbed at the adjacent Loco Site (92 m) returned a calibrated 14C age of 10.4 ± 0.1 ka (SNU12-599). At the southerly face of the tombolo, a partially lithified ~ 2 to 5 cm thick coquina (Fig. 4h) crops out at 92 m, above sublittoral sands, approximately 1 to 2 m below the surface of the tombolo. The coquina may reflect a low energy backshore environment on the lee side of the tombolo during a lake highstand. A M. tuberculata shell from the coquina produced a 14C age of 11.18 ± 0.08 cal. ka (SNU12-594). The uppermost deposits across the relict tombolo are moderately well sorted medium to coarse stratified sands with varying quantities of pebbles and cobbles (Fig. 5) and reflect shore face environments. Capping the upper shoreface deposit is a layer of rounded 2-to-5-cm diameter pebbles, which may reflect a backshore, shingle beach deposit (Fig. 4i). A single M. tuberculata shell taken from a gravel layer, ~ 20 cm below the tombolo surface, at the Talus Site (91 m) gave a 14C age of 10.61 ± 0.1 cal. ka (SNU12-595).

71

The elevation of paleo-lake levels and associated 14C ages from the western strand plain are combined with previously reported 14C ages for lacustrine deposits from other relict beaches (Table 1; Appendix 1). Similar to a previous analysis (Forman et al., 2014) we rank the reliability of 14C ages from 1 to 3 with a rank of 1 for the most reliable. A ranking of 1 is usually assigned to an AMS 14C age on a single gastropod or a single valve from a paired occurrence, associated with a direct elevation measurement and a well-documented sedimentology to infer paleo-water depth. A ranking of 2 indicates a conventional 14C age on the larger bivalve of Etheria sp. or Uniondae sp. These large bivalves probably provided the most secure 14C ages prior to the advent of AMS dating because there was sufficient mass to date a single valve or at the most 2 to 4 valves. However, it is difficult to discern if the shells were associated with deep-water facies, sublittoral or reworked into littoral sediments and thus, uncertainty (± 10 to 30 m) remains on the inferred water level. Lastly, a ranking of 3 indicates a conventional 14 C on bulk shells with species unidentified or from a mixed species collection with less accurate location and uncertainties in elevation. Early studies recognized that 14C ages on bulk shell collections can yield spurious ages (Owen et al., 1982). In light of this uncertainty Owen et al. (1982) inferred significant oscillations in lake level between 50 and 80 m for the Holocene. The goal of this assessment of age veracity is to include the earliest 14C ages from pioneering research in the Turkana Basin (Grove et al., 1975; Butzer, 1980; Owen et al., 1982) but weighted in context to evolving approaches in quantifying submillennial oscillations for Lake Turkana (cf. Forman et al., 2014; Morrissey and Scholz, 2014). This reconstruction of water level includes previously published data from strand plains at Mt. Porr (Forman et al., 2014), South Island (Garcin et al., 2012), Koobi Fora (Owen et al., 1982), and Kibish (Butzer, 1980; Brown and Fuller, 2008). Also included in this reconstruction are optical ages from the Mt. Porr strand plain (Forman et al., 2014) and 14C ages on archeological material from prior studies (Appendices 1 and 2). The resulting integrated water level reconstruction for Lake Turkana (Fig. 6) reveals nine centennial to millennial scale oscillations in water level between 15 and 4.5 ka. Six radiocarbon ages constrain lake level between 15.0 and 11.5 cal. ka (Fig. 6). Lake level rose from ~30 m at ca. 14.6 ka to at least ~80 m as assessed at the Big Ridge Site (Fig. 5). An AMS 14C age on a single bivalve constrains the transgression limit to at least 75 m at ca. 12.4 ka (Brown and Fuller, 2008) with a subsequent fall in lake level to at least 50 m by 11.5 ka. Water level rose to about 95 m by ca. 11.2 ka constrained by four shell ages from Lothagam, Kibish and South Island. A low stand is inferred between ca. 11.2 and 10.8 ka, though constrained by a 14C age on an Unionidae sp. shell (Butzer, 1980); this age has a large error (N5%) and the elevation of this sample was decreased by 62 m by Brown and Fuller (2008). Lake level may have fallen by at least 15 m by ca. 11.2 ka in context of archaeological evidence for human habitation at 84 m (Beyin, 2011). Another high stand at N 90 m between 10.8 and 10.3 ka is constrained by five 14C ages from Lothagam (Table 1; Fig. 5). Water level was at 91 m and up to N95 m by ca. 10.8 ka supported by three 14 C ages from sublittoral sand and lacustrine sediment (Fig. 5). A brief still stand at ~ 91 m is indicated by a 14C age of 10.61 ± 0.1 ka on a shell from upper foreshore sands. Water level may have transgressed slightly to 95 m by ca. 10.4 ka as inferred from sublittoral sands at the Loco Site. A 14C age of ca. 10.3 ± 0.2 ka on E. elliptica shell constrains a subsequent regression to below 75 m. Four lake level oscillations of N 20 m are inferred between 9.3 and 6.5 ka. A low stand below 65 m occurred post ca. 10.3 ka and possibly until ca. 9.3 ka, based on a 14C age on an E. elliptica shell from 63 m (Brown and Fuller, 2008). An AMS 14C age of 9.1 ± 0.1 ka on a single test of M. tuberculata from South Island indicates a possible high stand

72

C. Bloszies et al. / Global and Planetary Change 132 (2015) 64–76

Fig. 6. Reconstruction of Lake Turkana water levels for the past 15 cal. ka.

to N 90 m, and is associated with spillover at the outlet (Garcin et al., 2012). A regression to b 42 m is constrained by an AMS 14C age of 8.9 ± 0.1 ka on a single gastropod shell from an upper sublittoral deposit (Table 1). An OSL age of 8.3 ± 0.6 ka on quartz grains from a littoral deposit indicates a ~ 35 m transgression from this low. A conventional 14C age of 8.4 ± 0.1 ka on unidentified shells at 68 m at Lothagam is consistent with the timing of this transgression (Butzer et al., 1972). A low stand to at least 38 m is constrained by a sole 14C age of 7.8 ± 0.2 ka on a M. tuberculata test from an upper shoreface deposit (Table 1). In turn, water level may have regressed by as much as 70 m, suggested by a conventional 14C age of 7.8 ± 0.1 ka on mixed shells from a relict beach near Kibish at 16 m (Butzer, 1980), and subsequently adjusted to 4 m elevation by Brown and Fuller (2008). Water level rose to N60 m between 7.8 and 6.6 ka, with two oscillations at ca. 7.6 and 7.0 ka, which may have occurred within b500 years. The oldest of the two transgressions is constrained by 14C ages of 7.6 ka and 7.5 ka on a single gastropod from 58 m at Mt. Porr (Forman et al., 2014) and a snail shell at 60 m at Eliye Springs, respectively (Garcin et al., 2012). Subsequently, lake level fell to at least 27 m by 7.5 ± 0.1 ka, based on a 14C age on a gastropod from Mt. Porr (Forman et al., 2014). A rise in lake level to at least 72 m is constrained by a 14C age of 7.0 ± 0.1 ka on a Unionidae sp. shell from a relict beach at 82 m (Butzer, 1980), and adjusted to 72 m by Brown and Fuller (2008). Also, a 14C age of 7.0 ± 0.1 ka on mixed shells from 84 m suggests a higher lake level (Fig. 6). A low stand to ~35 m from ca. 6.9 to 6.5 ka is constrained by three AMS 14C ages on M. tuberculata shells of 6.96 ± 0.07 ka, 6.75 ± 0.025 ka and 6.67 ± 0.07 ka, of which the latter age is from 30 m in the Mt. Porr area (Forman et al., 2014). A prominent rise in lake level to at least 95 m at ca. 6.3 ka is depicted in previous reconstruction of water level (Forman et al., 2014), and is now constrained by multiple 14C ages. This transgression is supported by a conventional 14C age of 6.6 ± 0.1 ka from an oyster-type shell at 72 m near Kataboi (Garcin et al., 2012). Peak water levels of N 85 m are reconstructed from an AMS 14C age of 6.4 ± 0.1 ka on a single valve of E. elliptica from an in situ occurrence in lacustrine sediments; an optical age of 6.4 ± 0.5 ka on these sediments provides further confirmation of age (Forman et al., 2014). The subsequent regression post 6.3 ka appears rapid, within 100 to 400 years, and is constrain by numerous 14C ages. A drop in water level to b 28 m is constrained by ages of 6.3 ± 0.1 ka at 36 m on a M. tuberculata shell and 6.3 ± 0.1 ka at 28 m on a C. fluminalis africana shell (Forman et al., 2014). This fall in water level may have continued to be below 12 m, indicated by two AMS 14C ages of ca. 6.0 ka on a C. fluminalis africana shell at 12 m and a M. tuberculata shell from 7 m,

and adjacent upper sublittoral deposits at 14 m yielded an optical age of 5.8 ± 0.4 ka (Forman et al., 2014). A minor oscillation of uncertain amplitude may have occurred between 5.9 and 5.5 ka, constrained by less certain ages, with elevation assessed by digital elevation models (±16 m error; Table 2). Charcoal from an archaeological site at Lothagam (N 83 m) returned a 14C age of 5.8 ± 0.2 ka, placing the coeval water surface below this site elevation, but how low is uncertain (Butzer et al., 1972). This transgression occurred to at least ~65 m based on two 14C ages of 5.7 ± 0.05 ka and 5.7 ± 0.26 ka on M. tuberculata shell (Garcin et al., 2012) and an E. elliptica shells (Owen et al., 1982), respectively. The last inferred high lake stand to N85 m occurred between ca. 5.4 and 5.1 ka. This transgression is constrained by three AMS 14C ages from the South Island, which infer a rapid rise from 31 m at 5.4 ± 0.1 ka to 76 m at 5.4 ± 0.1 ka (Garcin et al., 2012). A high stand of N90 m is also indicated by backshore deposits at the Ekai site associated with an age of 5.18 ± 0.12 ka on a M. tuberculata shell. The subsequent regression from this peak level is constrained by a conventional 14C age of 4.99 ± 0.28 ka on E. elliptica shells at 67 m from the Koobi Fora area (Owen et al., 1982), and two AMS 14C ages of 4.92 and 4.87 ka at 47 and 65 m from South Island, respectively (Garcin et al., 2012). Low water levels below 13 m are inferred for a regressional beach sequence near Mt. Porr with an OSL age of 4.6 ± 0.3 ka (Forman et al., 2014). OSL and 14C ages from archaeological sites indicates that lake level didn't surpass 20 m in the past ca. 3 ka (e.g. Wright and Forman, 2011; Forman et al., 2014; Morrissey and Scholz, 2014). 6.2. What caused millennial-scale water level variability for Lake Turkana between 15 and 5 ka? We report up to nine N 30 m oscillations in water level for Lake Turkana between 14 and 4.5 ka (Fig. 6). Previous records of hydroclimatology from East African lakes (e.g. Berke et al., 2012a,b; Tierney et al., 2011a,b) and Nile River discharge (Weldeab et al., 2014) do not reveal similar millennial to submillennial scale variability. Lake Turkana appears to amplify hydroclimatic variability during the AHP with repeated overflow from the Suguta (Garcin et al., 2009; Junginger et al., 2014) (Fig. 7c) and the Chew Bahir basins (Grove et al., 1975; Foerster et al., 2012). A decline of water level below a 95 to 100 m threshold for Lake Turkana results in a closed basin, with dominant evaporative losses and a rapid (centennial to decadal scale) decline in lake level (Bloszies and Forman, 2015). These basins during the AHP received precipitation from air masses conditioned over the equatorial Atlantic and Indian oceans associated with varying intensity of the West and the East African monsoons. An eastward extension of

C. Bloszies et al. / Global and Planetary Change 132 (2015) 64–76

73

Fig. 7. Water level chronologies and constraining ages for lakes adjacent to Lake Turkana. (a) Lake Ziway-Shala (Gillespie et al., 1983). (b) Lake Turkana water levels for comparison (this study). Note solid black arrow, which reflects possible spillover from Lake Chew Bahir. (c) Lake Suguta water levels, with 14C ages corrected for a ~1.9 ka carbon reservoir effect (Garcin et al., 2009; Junginger et al., 2014). (d) Radiocarbon ages on shells from Lake Baringo (Williams and Johnson, 1976). (e) Lake Nakuru water level chronology based on diatom assemblages (Richardson and Dussinger, 1987). Blue shaded areas reflect periods of spillover. All ages calibrated by Fairbanks et al. (2005).

the CAB during much of the AHP is implicated for sustained high lake levels in East Africa (e.g., Tierney et al., 2011b; Costa et al., 2014; Junginger et al., 2014), though there appears to be insufficient resolution to relate to specific high water stands for Lake Turkana (cf., Morrissey and Scholz, 2014). There is evidence from marine sediment cores that varying SSTs in the equatorial eastern Atlantic Ocean may have modulated the intensity of the West African Monsoon over the Kenyan Highlands for much of the Late Quaternary (Weldeab et al., 2005, 2007, 2014), and may be an important driver for water level variability for Lake Turkana (cf. Tierney et al., 2011a,b). However, the role of SST variability in the western Indian Ocean on the synoptic conditions conducive for rainfall over the Kenyan and Ethiopian plateaus on Late Pleistocene to Holocene timescales is unclear. Variability in precipitation within the Turkana Basin associated with the Indian Ocean Dipole in the 20th and 21st centuries has been implicated in meter- to submeter-scale variations in water level (Bloszies and Forman, 2015. Isotopic studies of moisture and synoptic-scale back trajectories in the 20th century indicate an appreciable moisture sources from the southeast for Kenya and southeastern Ethiopia during rainfall events associated with the East African Monsoon (cf. Levin et al., 2009). Other studies have underscored that a rise in SSTs to 26° to 28 °C adjacent to equatorial East Africa in the Late Quaternary enhance deep atmospheric convection and are important non-linear drivers of precipitation variability over the Ethiopian Highlands (e.g., Berke et al., 2012a,b; Tierney and deMenocal, 2013; Forman et al., 2014). There is limited evidence from Lake Turkana strandplains for an initial transgression to at least 70 m for Lake Turkana between 14.4 and 13.0 ka. Proxies for surface runoff for many sites in East Africa also

indicate wetter conditions ca. 14 to 13 ka (Fig. 9g–i; Verschuren et al., 2009; Marshall et al., 2011; Foerster et al., 2012). These wetter conditions are associated with a strengthened East African monsoons inferred from an abrupt depletion in δDwax post ca. 14.5 ka for sediment cores from the Congo River delta (Schefuß et al., 2005) and the Gulf of Aden (Tierney and deMenocal, 2013). In contrast, multiple lacustrine and oceanic proxies indicate that the Younger Dryas period (12.8 and 11.5 ka) was associated with pronounced drying across equatorial Africa (cf. Schefuß et al., 2005; Berke et al., 2012a; Tierney and deMenocal, 2013) and for the Ethiopian and the Kenyan highlands (e.g. Verschuren et al., 2009; Marshall et al., 2011; Foerster et al., 2012). Lake Turkana water level fell by at least 25 m during the YD period (Fig. 8). This broad-scale drying across equatorial East Africa appears coincident with cooling of the Gulf of Guinea, potentially suppressing the East African Monsoon (Weldeab et al., 2007). There is compelling evidence for a rise in Lake Turkana post the YD period and sustained high water level near the outlet elevation (90 to 100 m) between 11.5 and 10.5 ka (Fig. 6). This high stand is coincident with overflow from the Suguta Basin and possibly from the Chew Bahir Basin (Fig. 7c). An elevated BIT index for a sediment core from Lake Challa (Fig. 8e), a proxy for surface run off, indicates increased precipitation in the lake's catchment post the YD period. Sediment cores from lakes Tanganyika, Tana, Victoria and Challa at ca. 11 ka show a marked decrease in δDwax (Fig. 8), which may signify an increased eastward flux of moisture from Atlantic-derived sources (Tierney et al., 2011b; Fig. 8f). Multiple water level oscillations of N50 m for Lake Turkana between ca. 9.5 and 4.5 ka is consistent with variable moisture flux from Atlantic-

74

C. Bloszies et al. / Global and Planetary Change 132 (2015) 64–76

Fig. 8. Paleoclimate chronologies from equatorial Africa. Oceanic cored δDwax records of terrigenous runoff: (a) Gulf of Aden core (Tierney and deMenocal, 2013). (b) Congo River delta core (Schefuß et al., 2005). Lacustrine δDwax records (c) Lake Tana (Costa et al., 2014). (d) Lake Victoria (Berke et al., 2012a,b). (e) Lake Challa (Tierney et al., 2011a,b). (f) Lake Tanganyika (Tierney et al., 2010). Various terrestrial runoff proxies for East African lakes. (g) Lake Tana titanium concentration (Marshall et al., 2011). (h) Chew Bahir potassium record (Foerster et al., 2012). (I) Gulf of Guinea SSTs, from Mg/Ca ratios on planktonic forams (Weldeab et al., 2007). (k) western Indian Ocean SST record, based on alkenones (Bard et al., 1997). (l) eastern Indian Ocean SST record, from Mg/Ca ratios onplanktonic forams (Mohtadi et al., 2010).

derived sources for lacustrine systems across East Africa during the Early to Middle Holocene (Fig. 8). Although, there may be insufficient temporal resolution and proxy insensitivity to resolve submillennial

scale variability as revealed in the lake level record. The likely cause of these lake level changes is linked to variability of the East and West African monsoons. Surface ocean temperatures in the eastern equatorial Atlantic peaked (~28.5 °C) at about 9 ka and variably decreased for the remainder of the Holocene, except for the sharp drop in SSTs associated with the 8.2-ka event (Figs. 8j), which may correspond to a low lake level. In contrast, western Indian Ocean SSTs rise post 9 ka with apparent peak values at 7.5 and 6 ka (Fig. 8k; Bard et al., 1997) which may be associated with an enhanced East African Monsoon. A sharp increase in δDwax for a sediment core from Lake Tana at ca. 8 ka and sustained for the remainder of the Holocene (Fig. 8c) indicates diminished contributions of Atlantic-derived moisture to the Ethiopian Highlands. In contrast, the record from Lake Victoria shows a steady increase in δDwax between 9 and 5 ka, indicating a more gradual decrease in Atlanticderived moisture and potentially increasing contributions from the East African Monsoon (cf. Berke et al., 2012a,b) (Fig. 8d). Lake level fell precipitously by over 80 m reaching a record low stand by ca. 6 ka (Fig. 6). This distinctive drop in water level is broadly coincident with desiccation of Lake Suguta (Fig. 7c) and drier conditions in Chew Bahir Basin (Fig. 8h), though age control is limited. This dual association of drying indicates a substantial decrease of precipitation in the Kenyan and the Ethiopian highlands and, thus weakening of the East and West African monsoons. The last high stand for Lake Turkana that may have reached the basin sill elevation (~95 to 100 m) occurred at ca. 5.2 ka. Available 14C ages indicate that Lake Suguta was falling by 5 ka (Fig. 7c) and thus, outflow from this basin is unlikely associated with this high stand. A noticeable drop in potassium for sediment core from the Chew Bahir Basin (Fig. 8h) and a sustained high stand for Lake Ziway-Shala (Fig. 7a) both at ca 5.2 ka indicate that outflow from the Chew Bahir Basin may have contributed to latest high stand. Thus, precipitation from the Ethopian Highlands is implicated in this latest highstand. This b 500-year high stand appears coincident with elevated SSTs in the western Indian Ocean (Fig. 8k) and may reflect an intensified East African Monsoon, though submillennial-scale resolution is limited for this marine record. The high variability in lake levels from ca. 13 to 4.6 ka identified from the relict beach record may reflect modest changes in latitudinal extent of the ITCZ and related longitudinal extent of the CAB (cf. Tierney et al., 2011a,b; Costa et al., 2014; Junginger et al., 2014; Morrissey and Scholz, 2014). During periods of enhanced monsoon activity the northern extent of the ITCZ is inferred to be ≥ 5° further north than at present (Tierney et al., 2011a,b, 2013; Junginger et al., 2014). An expanded ITCZ enhanced advection of moisture from the Congo Basin and supplied the cascading eastern Rift Valley lake system with additional precipitation, with final spillover raising lake level in last lake of the system, Turkana. The eastern Rift Valley catchment occurs near the eastern limit of the CAB and as a closed system it is sensitive to varying hydrological inputs, thus oscillations in lake level may reflect longitudinal variability in the position and strength of the CAB with the summer monsoon and variations in moisture advected the intensity and westward penetration of the Winter Monsoon. Lake level rapidly dropped N 80 m post 4.9 ka and water level remains below 20 m since 4.6 ka (cf. Forman et al., 2014). Other East African lakes like Tanganyika, Victoria and Challa reached low stands between 5 and 4 ka (Fig. 8). Similarly, metrics of rainfall for the Ethiopian Highlands indicate a profound drying at ca. 4.9 ka (Marshall et al., 2011; Foerster et al., 2012) with desiccation of Ethiopian lakes Ziway, Shala and Abhe, though not well dated (Gasse and Van Campo, 1994) (Fig. 7a). When did the African Humid Period end? Probabilistic age models for δDwax records from lakes Challa and Tanganyika and the Gulf of Aden indicate that termination of the African Humid Period commenced at 4960 ± 70 cal. yr B.P. with the transition to aridity occurring within 280–460 years (Tierney and deMenocal, 2013). However, dating of relict beaches surrounding Lake Turkana indicates that the transition from high to low lake level is highly variable post 8 ka, with a well

C. Bloszies et al. / Global and Planetary Change 132 (2015) 64–76

documented low stand below 10 m at ca. 6.0 ka. The latest highstand for Lake Turkana places the termination of the wettest conditions at ca. 5.1 ka with the transition to aridity by ca. 4.6 ka with the fall in water level to b 20 m. 7. Conclusion This study of relict beach sequences reveals up to nine potential oscillations in water level of N30 m between ca. 15 and 4 ka for Lake Turkana. This variability in lake level reflects varying rainfall contributions from the West and East African monsoons, amplified by cascading lake systems in the eastern Rift Valley and the Ethiopian Plateau that terminate into Lake Turkana. Lake level rose to at least 70 m between Heinrich 1 event and the Younger Dryas chronozone (ca. 14.5 and 13 ka), possibly associated with a rise in SSTs in the eastern equatorial Atlantic and concomitant westward expansion of the Congo Air Boundary After the Younger Dryas, lake level rose to at least 90 m between ca. 11.2 and 10.4 ka, falling by N30 m by 10.2 ka; and with another potential rise at ca. 8.5 ka to at least 70 m with a subsequent N40 m fall of water level at 8.2 ka. There were two major N70 m lake level oscillations centered at 6.6 and 5.2 ka and may reflect enhanced convection with warmer SSTs in the western Indian Ocean, rather than far traveled moisture associated with CAB. The transition between the AHP as deciphered from relict beaches for Lake Turkana is characterized by highly variable water level (± N40 m) from 8.0 to 4.5 ka, with the final fall in lake level from N 90 m commencing at 5.2 ka. This variability in lake level reflects a complex response to variations in the extent and intensity of the East and West African Monsoons near geographic and topographic limits within the catchment of Lake Turkana. Also, basin excesses and deficits in water input are amplified in this closed lake basin with a cascading lake effect in the East African Rift Valley and through the Chew Bahir Basin. Lake level was below 20 m by 4.6 ka and was sustained below 20 m for the Late Holocene and the historic period. This low stand reflects weakening of the West African Monsoon, considerably less precipitation into the Turkana from convergence of the Congo Air Boundary, and with meter-scale fluctuations in lake level linked possibly to the Indian Ocean Dipole. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.gloplacha.2015.06.006. Acknowledgements The research was conducted with permission from the Office of the President of the Republic of Kenya (permit MOHEST 13/001/30C 220) and in collaboration with the National Museums of Kenya Department of Earth Sciences. Partial funding of the research was made by the National Research Foundation of Korea grant #2013S1A5A8021512. Special thanks go to Purity Kiura, head of the Archaeology Section, for her invaluable assistance and advice in facilitating fieldwork and exporting samples. We thank the Turkana Basin Institute in Lodwar for hosting us during fieldwork and providing logistical support. Kristina Dziedzic Wright generously hosted and provisioned the researchers in Nairobi as they moved between different field projects. The ideas, counsel and hospitality of John Shea and Lisa Hildebrandt facilitated field research. Our sincerest appreciation goes to the people of western Turkana who patiently put up with strangers showing up unannounced, digging holes, and taking samples for vague purposes. Our conversations and encounters were always positive and we hope that our research results will repay their generosity in some small measure. References Abdallah, A.M., Barton, D.R., 2003. Environmental factors controlling the distributions of benthic invertebrates on rocky shores of Lake Malawi, Africa. J. Great Lakes Res. 29 (Supplement 2), 202–215. Alley, R.B., Mayewski, P.A., Sowers, T., Stuiver, M., Taylor, K., Clark, P.U., 1997. Holocene climatic instability: a prominent, widespread event 8200 yr ago. Geology 25, 483–486.

75

Bard, E., Rostek, F., Sonzogni, C., 1997. Interhemispheric synchrony of the last deglaciation inferred from alkenone palaeothermometry. Nature 385, 707–710. Becker, M., LLovel, W., Cazenave, A., Güntner, A., Crétaux, J.-F., 2010. Recent hydrological behavior of the East African great lakes region inferred from GRACE, satellite altimetry and rainfall observations. Compt. Rendus Geosci. 342, 223–233. Berke, M.A., Johnson, T.C., Werne, J.P., Grice, K., Schouten, S., Sinninghe Damsté, J.S., 2012a. Molecular records of climate variability and vegetation response since the Late Pleistocene in the Lake Victoria basin, East Africa. Quat. Sci. Rev. 55, 59–74. Berke, M.A., Johnson, T.C., Werne, J.P., Schouten, S., Sinninghe Damsté, J.S., 2012b. A mid-Holocene thermal maximum at the end of the African Humid Period. Earth Planet. Sci. Lett. 351–352, 95–104. Beyin, A., 2011. Recent archaeological survey and excavation around the Greater Kalokol Area, west side of Lake Turkana: preliminary findings. Nyame Akuma 75, 40–50. Black, E., Slingo, J., Sperber, K.R., 2003. An observational study of the relationship between excessively strong short rains in coastal East Africa and Indian ocean SST. Mon. Weather Rev. 131, 74–94. Blanchet, C.L., Tjallingii, R., Frank, M., Lorenzen, J., Reitz, A., Brown, K., Feseker, T., Brückmann, W., 2013. High-and low-latitude forcing of the Nile River regime during the Holocene inferred from laminated sediments of the Nile deep-sea fan. Earth Planet. Sci. Lett. 364, 98–110. Bloszies, C.A., Forman, S.L., 2015. Potential relation between equatorial sea surface temperatures and historic water level variability for Lake Turkana, Kenya. J. Hydrol. 520, 489–501. Brown, F.H., Fuller, C.R., 2008. Stratigraphy and tephra of the Kibish Formation, southwestern Ethiopia. J. Hum. Evol. 55, 366–403. Brown, F.H., Mcdougall, I., 2011. Geochronology of the Turkana depression of northern Kenya and southern Ethiopia. Evol. Anthropol. Issues News Rev. 20, 217–227. Butzer, K., 1980. The Holocene lake plain of North Rudolph, East Africa. Phys. Geogr. 1, 42–58. Butzer, K.W., Thurber, D.L., 1969. Some late Cenozoic sedimentary formations of the Lower Omo Basin. Nature 222, 1138–1143. Butzer, K.W., Isaac, G.L., Richardson, J.L., Washbourn-Kamau, C., 1972. Radiocarbon dating of East African Lake levels. Science 175, 1069–1076. Castañeda, I.S., Werne, J.P., Johnson, T.C., 2007. Wet and arid phases in the southeast African tropics since the Last Glacial Maximum. Geology 35, 823–826. Cohen, A.S., 1986. Distribution and faunal associations of benthic invertebrates at Lake Turkana, Kenya. Hydrobiologia 141, 179–197. Costa, K., Russell, J., Konecky, B., Lamb, H., 2014. Isotopic reconstruction of the African Humid Period and Congo Air Boundary migration at Lake Tana, Ethiopia. Quat. Sci. Rev. 83, 58–67. Damsté, J.S.S., Verschuren, D., Ossebaar, J., Blokker, J., van Houten, R., van der Meer, M.T., Plessen, B., Schouten, S., 2011. A 25,000-year record of climate-induced changes in lowland vegetation of eastern equatorial Africa revealed by the stable carbon-isotopic composition of fossil plant leaf waxes. Earth Planet. Sci. Lett. 302, 236–246. De Kock, K., Wolmarans, C., 2007. Distribution and habitats of Corbicula fluminalis africana (Mollusca: Bivalvia) in South Africa. Water SA 33, 709–716. Dühnforth, M., Bergner, A.G., Trauth, M.H., 2006. Early Holocene water budget of the Nakuru–Elmenteita basin, central Kenya Rift. J. Paleolimnol. 36, 281–294. Fairbanks, R.G., Mortlock, R.A., Chiu, T.C., Cao, L., Kaplan, A., Guilderson, T.P., Fairbanks, T.W., Bloom, A.L., Grootes, P.M., Nadeau, M.J., 2005. Radiocarbon calibration curve spanning 0 to 50,000 years BP based on paired 230Th / 234U / 238U and 14C dates on pristine corals. Quat. Sci. Rev. 24, 1781–1796. Feibel, C.S., 2003. Stratigraphy and Depositional History of the Lothagam Sequence. Columbia University Press, New York, pp. 17–29. Foerster, V., Junginger, A., Langkamp, O., Gebru, T., Asrat, A., Umer, M., Lamb, H.F., Wennrich, V., Rethemeyer, J., Nowaczyk, N., Trauth, M.H., Schaebitz, F., 2012. Climatic change recorded in the sediments of the Chew Bahir basin, southern Ethiopia, during the last 45,000 years. Quat. Int. 274, 25–37. Forman, S.L., Wright, D.K., Bloszies, C.A., Pierson, J., Gomez-Mazzocco, J., 2014. Variations in water level for Lake Turkana in the past 8500 years near Mt. Porr, Kenya and the transition from the African Humid Period to Holocene aridity. Quat. Sci. Rev. 97, 84–101. Garcin, Y., Vincens, A., Williamson, D., Buchet, G., Guiot, J., 2007. Abrupt resumption of the African Monsoon at the Younger Dryas—Holocene climatic transition. Quat. Sci. Rev. 26, 690–704. 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. Quat. Sci. Rev. 28, 911–925. Garcin, Y., Melnick, D., Strecker, M.R., Olago, D., Tiercelin, J.-J., 2012. East African midHolocene wet–dry transition recorded in palaeo-shorelines of Lake Turkana, northern Kenya Rift. Earth Planet. Sci. Lett. 331–332, 322–334. Gasse, F., 2000. Hydrological changes in the African tropics since the Last Glacial Maximum. Quat. Sci. Rev. 19, 189–211. Gasse, F., Van Campo, E., 1994. Abrupt post-glacial climate events in West Asia and North Africa monsoon domains. Earth Planet. Sci. Lett. 126, 435–456. Gasse, F., Chalié, F., Vincens, A., Williams, M.A., Williamson, D., 2008. Climatic patterns in equatorial and southern Africa from 30,000 to 10,000 years ago reconstructed from terrestrial and near-shore proxy data. Quat. Sci. Rev. 27, 2316–2340. Gathogo, P.N., Brown, F.H., McDougall, I., 2008. Stratigraphy of the Koobi Fora Formation (Pliocene and Pleistocene) in the Loiyangalani region of northern Kenya. J. Afr. Earth Sci. 51, 277–297. Genner, M.J., Michel, E., 2003. Fine-scale habitat associations of soft-sediment gastropods at Cape Maclear, Lake Malawi. J. Molluscan Stud. 69, 325–328. Gillespie, R., Street-Perrott, F.A., Switsur, R., 1983. Post-glacial Arid Episodes in Ethiopia Have Implications for Climate Prediction.

76

C. Bloszies et al. / Global and Planetary Change 132 (2015) 64–76

Goddard, L., Graham, N.E., 1999. Importance of the Indian Ocean for simulating rainfall. J. Geophys. Res. 104 (19,099-019,116). Grove, A.T., Street, F.A., Goudie, A.S., 1975. Former lake levels and climatic change in the Rift Valley of southern Ethiopia. Geogr. J. 141 (2), 177–194. 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. Harvey, C.P.D., Grove, A.T., 1982. A prehistoric source of the Nile. Geogr. J. 148, 327–336. Johnson, T.C., Malala, J.O., 2009. Lake Turkana and its link to the Nile. The Nilepp. 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, p. 153 (pp.). Junginger, A., Trauth, M.H., 2013. Hydrological constraints of paleo-Lake Suguta in the Northern Kenya Rift during the African Humid Period (15–5 kaBP). Glob. Planet. Chang. 111, 174–188. Junginger, A., Roller, S., Olaka, L.A., Trauth, M.H., 2014. The effects of solar irradiation changes on the migration of the Congo Air Boundary and water levels of paleoLake Suguta, Northern Kenya Rift, during the African Humid Period (15–5 ka BP). Palaeogeogr. Palaeoclimatol. Palaeoecol. 396, 1–16. Kinuthia, J., Asnani, G., 1982. A newly found jet in North Kenya (Turkana Channel). Mon. Weather Rev. 110, 1722–1728. Korniushin, A.V., 2004. A revision of some Asian and African freshwater clams assigned to Corbicula fluminalis (Müller, 1774; Mollusca: Bivalvia: Corbiculidae), with a review of anatomical characters and reproductive features based on museum collections. Hydrobiologia 529, 251–270. Kuper, R., Kröpelin, S., 2006. Climate-controlled Holocene occupation in the Sahara: motor of Africa's evolution. Science 313, 803–807. Leng, M.J., Lamb, A.L., Lamb, H.F., Telford, R.J., 1999. Palaeoclimatic implications of isotopic data from modern and early Holocene shells of the freshwater snail Melanoides tuberculata, from lakes in the Ethiopian Rift Valley. J. Paleolimnol. 21, 97–106. Levin, N., Zisper, T.E., Cerling, T.E., 2009. Isotopic composition of waters from Ethiopia and Kenya: insights into moisture sources for eastern Africa. J. Geophys. Res. 114, D23306. http://dx.doi.org/10.1029/2009JD012166. Marchant, R., Mumbi, C., Behera, S., Yamagata, T., 2007. The Indian Ocean dipole—the unsung driver of climatic variability in East Africa. Afr. J. Ecol. 45, 4–16. Marshall, M.H., Lamb, H.F., Davies, S.J., Leng, M.J., Kubsa, Z., Umer, M., Bryant, C., 2009. Climatic change in northern Ethiopia during the past 17,000 years: a diatom and stable isotope record from Lake Ashenge. Palaeogeogr. Palaeoclimatol. Palaeoecol. 279, 114–127. Marshall, M.H., Lamb, H.F., Huws, D., Davies, S.J., Bates, R., Bloemendal, J., Boyle, J., Leng, M.J., Umer, M., Bryant, C., 2011. Late Pleistocene and Holocene drought events at Lake Tana, the source of the Blue Nile. Glob. Planet. Chang. 78, 147–161. McDougall, I., Brown, F.H., 2009. Timing of volcanism and evolution of the northern Kenya Rift. Geol. Mag. 146, 34–47. McGregor, G.R., Nieuwolt, S., 1998. Tropical Climatology: An Introduction to the Climates of the Low Latitudes. John Wiley & Sons Ltd. Mohtadi, M., Lückge, A., Steinke, S., Groeneveld, J., Hebbeln, D., Westphal, N., 2010. Late Pleistocene surface and thermocline conditions of the eastern tropical Indian Ocean. Quat. Sci. Rev. 29, 887–896. Morrissey, A., Scholz, C.A., 2014. Paleohydrology of Lake Turkana and its influence on the Nile River system. Palaeogeogr. Palaeoclimatol. Palaeoecol. 403, 88–100. Nicholson, S.E., 1996. A review of climate dynamics and climate variability in eastern Africa. In: Johnson, T.C., Odada, E.O. (Eds.), The Limnology, Climatology and Paleoclimatology of the East African lakes, pp. 25–56. Nicholson, S.E., 2000a. Land surface processes and Sahel climate. Rev. Geophys. 38, 117–139. Nicholson, S.E., 2000b. The nature of rainfall variability over Africa on time scales of decades to millenia. Glob. Planet. Chang. 26, 137–158. Nyamweru, C., 1989. New evidence for the former extent of the Nile drainage system. Geogr. J. 155, 179–188. Owen, R.B., Renaut, R.W., 1986. Sedimentology, stratigraphy and palaeoenvironments of the Holocene Galana Boi Formation, NE Lake Turkana, Kenya. Geol. Soc. Lond., Spec. Publ. 25, 311–322. 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. Reading, H.G., 2009. Sedimentary Environments: Processes, Facies and Stratigraphy. John Wiley & Sons.

Richardson, J., Dussinger, R., 1987. Paleolimnology of mid-elevation lakes in the Kenya Rift Valley, Paleolimnology IV. Springer, pp. 167–174. Ritchie, J., Eyles, C., Haynes, C.V., 1985. Sediment and pollen evidence for an early to mid-Holocene humid period in the eastern Sahara. Nature 314, 352–355. Roberts, N., Taieb, M., Barker, P., Damnati, B., Icole, M., Williamson, D., 1993. Timing of the Younger Dryas Event in East Africa from Lake-level Changes. Saji, N., Yamagata, T., 2003. Possible impacts of Indian Ocean dipole mode events on global climate. Clim. Res. 25, 151–169. Schefuß, E., Schouten, S., Schneider, R.R., 2005. Climatic controls on central African hydrology during the past 20,000 years. Nature 437, 1003–1006. Street-Perrott, F., Marchand, D., Roberts, N., Harrison, S., 1989. Global Lake-level Variations from 18,000 to 0 years ago: A Palaeoclimate Analysis. Oxford Univ. (UK). Geography School. Thompson, L.G., Mosley-Thompson, E., Davis, M.E., Henderson, K.A., Brecher, H.H., Zagorodnov, V.S., Mashiotta, T.A., Lin, P.N., Mikhalenko, V.N., Hardy, D.R., Beer, J., 2002. Kilimanjaro ice core records: evidence of Holocene climate change in tropical Africa. Science 298, 589–593. Tierney, J.E., deMenocal, P.B., 2013. Abrupt shifts in horn of Africa hydroclimate since the Last Glacial Maximum. Science 342, 843–846. Tierney, J.E., Russell, J.M., Huang, Y., 2010. A molecular perspective on Late Quaternary climate and vegetation change in the Lake Tanganyika basin, East Africa. Quat. Sci. Rev. 29, 787–800. Tierney, J.E., Lewis, S.C., Cook, B.I., LeGrande, A.N., Schmidt, G.A., 2011a. Model, proxy and isotopic perspectives on the East African Humid Period. Earth Planet. Sci. Lett. 307, 103–112. Tierney, J.E., Russell, J.M., Sinninghe Damsté, J.S., Huang, Y., Verschuren, D., 2011b. Late Quaternary behavior of the East African monsoon and the importance of the Congo Air Boundary. Quat. Sci. Rev. 30, 798–807. Tierney, J.E., Smerdon, J.E., Anchukaitis, K.J., Seager, R., 2013. Multidecadal variability in East African hydroclimate controlled by the Indian Ocean. Nature 493, 389–392. Toreti, A., Naveau, P., Zampieri, M., Schindler, A., Scoccimarro, E., Xoplaki, E., Dijkstra, H.A., Gualdi, S., Luterbacher, J., 2013. Projections of global changes in precipitation extremes from Coupled Model Intercomparison Project Phase 5 models. Geophys. Res. Lett. 40, 4887–4892. Ummenhofer, C.C., Sen Gupta, A., England, M.H., Reason, C.J., 2009. Contributions of Indian Ocean sea surface temperatures to enhanced East African rainfall. J. Clim. 22, 993–1013. Van Bocxlaer, B., Van Damme, D., 2009. Palaeobiology and evolution of the Late Cenozoic freshwater molluscs of the Turkana Basin: Iridinidae Swainson, 1840 and Etheriidae Deshayes, 1830 (Bivalvia: Etherioidea). J. Syst. Palaeontol. 7, 129–161. Velpuri, N.M., Senay, G.B., Asante, K.O., 2012. A multi-source satellite data approach for modelling Lake Turkana water level: calibration and validation using satellite altimetry data. Hydrol. Earth Syst. Sci. 16, 1–18. Verschuren, D., Damsté, J.S.S., Moernaut, J., Kristen, I., Blaauw, M., Fagot, M., Haug, G.H., Van Geel, B., De Batist, M., Barker, P., 2009. Half-precessional dynamics of monsoon rainfall near the East African Equator. Nature 462, 637–641. Vétel, W., Le Gall, B., Johnson, T.C., 2004. Recent tectonics in the Turkana Rift (North Kenya): an integrated approach from drainage network, satellite imagery and reflection seismic analyses. Basin Res. 16, 165–181. Weldeab, S., Schneider, R.R., Kölling, M., Wefer, G., 2005. Holocene African droughts relate to eastern equatorial Atlantic cooling. Geology 33, 981–984. Weldeab, S., Lea, D.W., Schneider, R.R., Andersen, N., 2007. 155,000 years of West African monsoon and ocean thermal evolution. Science 316, 1303–1307. Weldeab, S., Frank, M., Stichel, T., Haley, B., Sangen, M., 2011. Spatio-temporal evolution of the West African monsoon during the last deglaciation. Geophys. Res. Lett. 38, L13703. http://dx.doi.org/10.1029/2011GL047805. Weldeab, S., Menke, V., Schmiedl, G., 2014. The pace of East African monsoon evolution during the Holocene. Geophys. Res. Lett. 41 (5), 1724–1731. Williams, R., Johnson, A., 1976. Birmingham University radiocarbon dates X. Radiocarbon 18, 249–267. Williams, A.P., Funk, C., Michaelsen, J., Rauscher, S.A., Robertson, I., Wils, T.H., Koprowski, M., Eshetu, Z., Loader, N.J., 2012. Recent summer precipitation trends in the Greater Horn of Africa and the emerging role of Indian Ocean sea surface temperature. Clim. Dyn. 1–22. Wright, D.K., Forman, S.L., 2011. Holocene occupation of the Mount Porr Strand Plain in Southern Lake Turkana, Kenya. Nyame Akuma 47–62.