Lakeside View: Sociocultural Responses to Changing Water Levels of Lake Turkana, Kenya

Hydrology and Climatology

Lake Turkana is a closed basin with an estimated annual evaporation rate that translates to an average fall in lake level of 2.3 m/year in the absence of hydrological inputs (Ferguson and Harbott 1982, p. 32). The Omo River delivers about 90 % of the water input for Lake Turkana with the remaining discharge deriving from the Turkwel and Kerio rivers and ephemeral streams that surround the lake (Yuretich 1979). The mean annual precipitation within most of the catchment is <250 mm/year (Källqvist et al. 1988; Nicholson 1996).

Lake Turkana is currently a closed basin and is therefore a sensitive proxy for precipitation in northern and eastern Africa. Residing at the lowest elevational limit of a series of cascading, endoheric lakes within the East African Rift Valley, runoff from this catchment zone ultimately discharges into Lake Turkana until the water level rises to about 100 m above its current level. Once this threshold is achieved, Lake Turkana spills over into the White Nile River catchment. There are three main mechanisms for precipitation on the Ethiopian Plateau: (1) the summer monsoons (June–September), (2) anticyclonic tropical depressions from the southern Indian Ocean associated with the winter monsoons, and (3) convergence from the Red Sea which mainly affects the northern reaches of Ethiopia (Conway 2000). The East African Monsoon brings the majority of rain to the northeast quarter of the African continent and occurs twice per year with the migration of the Intertropical Convergence Zone (ITCZ).

Increased precipitation in eastern Africa and high lake levels between 14,000 and 4,000 years BP are inferred to reflect a more easterly position of the Congo Air Boundary (CAB; Bergner and Trauth 2004; Tierney et al. 2011, 2013). The CAB is a zone of convergence between Atlantic and Indian Ocean moisture sources where there is a precipitation maximum inferred to be affected by Northern Hemisphere precessional-forced insolation changes (Kizza et al. 2009; Nicholson 1996). Fluctuations in lake levels in eastern Africa have been connected to enhanced East African Monsoon activity with eastward migration of the CAB, though Indian Ocean sources for precipitation cannot be discounted (Bloszies et al. in press; Forman et al. 2014; Junginger et al. 2014; Levin et al. 2009; Morrissey and Scholz 2014; Nicholson 1996; Tierney et al. 2011).

Other Controls on Lake Turkana Hydrology

Lake Turkana lies within the East African Rift of the Great Rift Valley, which extends from the Red Sea to southern Africa. The Turkana Basin is comprised of a series of half-grabens and accommodation zones which change in orientation along the axis of the rift (Dunkelman et al. 1988). The accommodation zones are characterized by structural highs and zones of oblique slip-faulting (Dunkelman et al. 1988). In addition, transverse and rift parallel faults have been detected, especially prominent along the western edges of the lake (Vétel et al. 2004). High-angle growth faults detected in seismic reflection profiles of the sediments in the lake indicate that the basin has remained kinematically active throughout the Holocene (Johnson et al. 1987; see also Melnick et al. 2012; Morrissey and Scholz 2014).

Postdepositional changes in the elevation of relict beaches have been attributed to rifting processes ongoing throughout the Holocene and isostatic processes associated with changes in water volume of the lake (Garcin et al. 2012; Melnick et al. 2012). From a study on South Island, Melnick et al. (2012) estimate that ground surface uplift during the Holocene occurs at 7.5–9.6 mm/year when tectonic activity and isostatic rebound are placed within elastic and thermomechanical models. The aggregate effect of these processes resulted in ~20 to 30 m of absolute elevation change in South Island throughout the Holocene relative to an arbitrary 0 datum (Melnick et al. 2012). However, South Island is located in the center of the rift zone and occupies one of the most kinematically active areas of the region (Dunkelman et al. 1988; Le Gall et al. 2005) and does not appear to be representative of other relict shorelines around Lake Turkana (Bloszies et al. in press; Forman et al. 2014).

Paleoclimate of the Turkana Catchment

There has been significant variability in climate and hydrology of Lake Turkana during the Holocene (Fig. 2(a)) and the understanding of the timing and magnitude of these shifts has evolved over the last three decades. There is evidence for sustained water levels <300 m.a.s.l. from 22,000 until 12,000 years BP in Lake Turkana as recorded in seismic profiles of the lake (Johnson et al. 1987; Morrissey and Scholz 2014) and relict beach ridge sequences up to ~100 m above present lake level (Brown and Fuller 2008). This period correlates to generally cooler- and drier-than-present conditions across equatorial and northern Africa associated with deglaciation of the northern latitudes and cooling in the North Atlantic Ocean (Berke et al. 2012b; Beuning et al. 1997; Braconnot et al. 2000; Gasse 2000; Gasse et al. 2008; Schefuß et al. 2011; Stager and Mayewski 1997; Tiercelin et al. 1988; Tierney et al. 2011).

By 12,000 years BP, a change in the hydrology of Lake Turkana occurs and numerous proxy studies show that the lake level remained above its present threshold for almost 7,500 years, although its relative position was variable therein. Analysis of diatomaceous silts from the Koobi Fora Formation indicates Early Holocene (ca. <12,000 to 8,000 years BP) lake stands were +75 to 80 m above the 1976 lake level (Owen et al. 1982). Based on his research in the lower Omo River Valley, Butzer (1980) argues that Lake Turkana receded to its present elevation at 6,600 years BP, although Owen et al. (1982) question this assertion. Owen et al. (1982) argue that there were three phases of lake transgression with two high stands in the Middle to Late Holocene. Profound lake level drops are posited at ~7,000 and 4,000 ¹⁴C years BP (7,800 and 4,500 cal. years BP) and a more gradual, but ultimately more severe drop in lake levels after 3,300 cal. years BP. However, the age models used in both Butzer (1980) and Owen et al. (1982) have been refined (Bloszies et al. in press; Forman et al. 2014; Garcin et al. 2012). Overflows of Lake Turkana into the Nile River may have occurred between 11,500 and 7,800 years BP and 7,400 and 4,300 years BP (Johnson and Malala 2009; see also Morrissey and Scholz 2014 in which Total Organic Carbon (TOC) derived from lake core sediments indicates overflow conditions between 10,000 and 6,000 years BP).

More recent sedimentologic analyses have employed new technologies and techniques to reconstruct prehistoric lake levels adjacent to Lake Turkana. Using a more comprehensive age model coupled with a detailed dGPS survey of shoreline deformation features, Garcin et al. (2012) argue for rapid drops in lake levels ca. 8,300 and 5,200 years BP with spillover, from the basin sill into the Nile River basin occurring ca. 11,000, 10,000–8,300, and possibly ca. 6,500 years BP. More recently, a detailed geomorphic and sedimentologic study of relict beach sequence in the Mt. Porr and Lothagam areas infer up to ten 20+ m century-scale oscillations in the level of Lake Turkana between ca. 14,000 and 4,000 years ago (Fig. 2(a); Bloszies et al. in press; Forman et al. 2014). In this reconstruction, generally pluvial conditions are inferred from 11,500 to 10,000 years BP followed by significant Middle Holocene lake level variability (Bloszies et al. in press). At least three spillover or near-spillover events are inferred dating to ca. 9,000, 6,200, and 5,000 years BP, each of which is bracketed by 30+ m transgressive-regressive phases in the lake elevation that occur on subcentennial timescales (Bloszies et al. in press). The final fall in lake level occurred ca. 5,000 years BP, when the lake began to lose significant amounts of water. By 4,500 years BP, lake levels had dropped to within 20 m of its current elevation and did not rise above this level based on the presence of archaeological sites above +20 m from the modern lake level (Bloszies et al. in press; Forman et al. 2014; Garcin et al. 2012). Smaller-scale oscillations in lake level are recorded in the past century as lake levels are ~375 m.a.s.l. at 100 years BP and fall 10+ m to present-day levels by about 50 years BP (Avery 2010; Källqvist et al. 1988).

Paleoclimate of Adjacent Basins

Geochronologies derived outside the Turkana Basin suggest enhanced monsoon activity during the Early Holocene across the western Indian Ocean, with significant spatial variability in monsoon precipitation beginning in the Middle Holocene. Presently, high monsoon activity in the western Indian Ocean is associated with warmer sea-surface temperatures (SSTs) and increased precipitation for the Turkana Basin (Bloszies and Forman 2015; Ummenhofer et al. 2009; Velpuri et al. 2012). Holocene climates were also modulated by the variable distribution of SSTs as well as zonal (Walker Cell) climate patterns. In the Gulf of Aden (Fig. 1(a)), deuterium isotopic values from terrigenous leaf waxes, an indicator precipitation within the Congo Basin (δD_wax) from an ocean sediment core, indicate variable but strong monsoon activity until ca. 5,000 years BP with an inferred weakening of monsoon-derived precipitation at ca. 4,500 years BP (Fig. 2(k); Tierney and deMenocal 2013). Off the Somali coast (Fig. 1(b)), reconstructed SSTs using alkenones are high in the Early Holocene and lower throughout the Middle Holocene, and express peak values between 5,500 and 3,800 years BP (Bard et al. 1997) although the data are sparse. Although Indian Ocean SSTs tied to monsoon strength are presently the main drivers of precipitation variability in the Turkana Basin as well as the rest of eastern Africa, the correlation between high SSTs and amplified monsoon activity is not linear, and secondary sources of moisture tied to zonal climate phenomena such as the Indian Ocean Dipole (IOD) can affect the level of Lake Turkana, but on meter scale (Bloszies and Forman 2015).

Open image in new window

Open image in new window

Lake Turkana lies within an extensive rift zone and is adjacent to a series of cascading basins that have significantly smaller catchments than Lake Turkana (Olaka et al. 2010). Relict shoreline analysis of the Suguta Valley (Fig. 1(c)), located to the immediate south with a transient connection to the Turkana Basin, detects the presence of a 300-m deep lake (390 km³) periodically overflowing into Lake Turkana from 14,500 to 6,500 years BP (Fig. 2(b); Garcin et al. 2009; Junginger et al. 2014). Another transiently connected catchment to the north, Lake Chew Bahir (Fig. 1(d)) shows regressions in water levels from 12,800 to 11,600, 10,800 to 10,500, and 9,800 to 9,100 years BP (Fig. 2(c); Foerster et al. 2012). Gradual, but progressively lower lake levels ensue from 8,000 to 7,500 years BP followed by an abrupt and severe regression associated with the termination of the African Humid Period (AHP) and onset of aridity after 7,000 years BP (Fig. 2(c); Foerster et al. 2012). Brief pluvial events are recorded at 3,000 and ~2,000–1,300 years BP (Fig. 2(c); Foerster et al. 2012).

The termination of the AHP marks a period of progressive decline in rainfall across northeastern Africa; however, temporal and geographic expression of this phenomenon was variable. The diatom and oxygen isotope record extracted from Lake Ashenge (Fig. 1(e)), northern Ethiopia, shows overflow conditions from 7,600 to 5,600 years BP followed by an abrupt transition to arid conditions (Marshall et al. 2009). A sediment core from Lake Tilo (Fig. 1(f)) in central Ethiopia records short periods of drying at ca. 8,400–8,800 and 6,700 years BP, with a permanent shift to aridity ca. 4,700 years BP (Fig. 2(d, e); Lamb et al. 2000). Less than 50 km to the north, two lakes with transient hydrological connections, lakes Ziway and Shala (Fig. 1(g)), show low stands sometime between 5,000 and 2,000 years BP (Fig. 2(f); Gillespie et al. 1983). Droughts are inferred from magnetic and geochemical data from cores extracted from Lake Tana (Fig. 1(h)) in the northern Ethiopian Highlands, centered around 8,400 and 7,500 years BP with progressive weakening of monsoon rains after 6,800 years BP (Fig. 2(g–i); Marshall et al. 2011). In another isotopic measure of moisture (δD_wax) from the same lake core, the Early Holocene pluvial period ended ca. 8,500 years BP as conditions transitioned to a permanent xeric state (Costa et al. 2014). Maximum aridity is recorded ca. 4,200 years BP and a permanent lowstand at Lake Abbe (a.k.a. Abhé; Fig. 1(i)) is documented after 4,400 years BP (Fig. 2(j); Gasse and Van Campo 1994). These data correlate well with deep-water sediment cores from the Arabian Sea and Gulf of Aden showing progressive declines in monsoon strength after 5,500 years BP (Fig. 2(k); Overpeck et al. 1996; Tierney and deMenocal 2013), but are not completely synchronous.

The northern Ethiopian Highlands drain into the Blue Nile River lying to the immediate north of the Turkana Basin. High sedimentation rates and low flows in the Blue Nile (Fig. 1(j)) are documented from 5,650 to 4,750 and 2,050 to 1,000 years BP (Krom et al. 2002). An analysis of oxygen and carbon isotopes in carbonates and diatom assemblages from a sediment core from Lake Hayq (Fig. 1(k)), which is close to the western edge of the Rift Valley and immediately east of the Blue Nile drainage basin, detects wetter conditions between 700 and 1,300 years BP, separated by drier episodes (Lamb et al. 2007). However, this lake has remained relatively full throughout the last 2,000 years (Lamb et al. 2007) indicating varying hydrologic thresholds for lacustrine systems in northeastern Africa.

Fluctuating precipitation regimes across space and time in the paleoclimatic proxy record on the Ethiopian Plateau have been attributed to the eastern limit and precipitation delivery rates of the CAB during the Holocene (Bloszies et al. in press; Costa et al. 2014; Forman et al. 2014; Levin et al. 2009; Liu et al. 2003; Tierney and deMenocal 2013; Tierney et al. 2010, 2011). During the Early Holocene, the CAB has been inferred (Bloszies et al. in press; Costa et al. 2014; Forman et al. 2014; Junginger et al. 2014; Levin et al. 2009; Tierney and deMenocal 2013) to extend east of the Rift Valley from present-day Branched and Isoprenoidal Tetraether (BIT) isotopic data, providing a significant amount of summer rainfall (Tierney et al. 2011). Enhanced monsoon strength in the Gulf of Guinea (West Africa) and positive state of the IOD are attributed to a more easterly positioning of the CAB and higher rainfall across Lake Turkana, specifically (Morrissey and Scholz 2014), and eastern Africa, more generally (Tierney et al. 2011). Between 8,500 and 7,800 years BP, there are numerous proxy records across northern and equatorial Africa indicating low lake levels; however, most of the lakes are recharged at the end of this event (Gasse 2000). The exception to this appears to be in the northern portion of the Horn of Africa, where the termination of the AHP is earlier than in the southerly locations (Costa et al. 2014). This is likely the result of the CAB gradually migrating further west (Costa et al. 2014) which correlated with the progressively more southerly displacement of the ITCZ in the Middle Holocene (Barker et al. 2002; Thevenon et al. 2002). By 4,500 years BP, the final transition to more arid conditions was underway in the northern Rift (including the Turkana region) with periods of intensified monsoon activity providing exceptions to this trend and temporarily recharging lakes under more pluvial conditions (Berke et al. 2012a; Bloszies et al. in press; Costa et al. 2014; Forman et al. 2014; Levin et al. 2009; Tierney and deMenocal 2013; Tierney et al. 2010, 2011).

Early Holocene, 12,000–9,000 Years BP

Human settlement patterns in the Lake Turkana Basin are relatively well documented during the EH (Fig. 3a). This period correlates to a pluvial period distributed broadly across the northern two-thirds of the African continent (Castañeda et al. 2009; Gasse 2000). Similar to other locations throughout northern Africa (Hoelzmann et al. 2001; Kuper and Kröpelin 2006; Sutton 1977; Yellen 1998), settlement of the Lake Turkana region probably included low residential mobility and was focused on exploiting abundant aquatic resources (Barthelme 1985; Butzer 1980; Phillipson 1977; Robbins 1974). Archaeological surveys of the Lake Turkana region suggest that there were at least seven major loci of EH settlement: (1) the Koobi Fora region in the northeast (Barthelme 1985), (2) Lowasera in the southeast (Phillipson 1977), (3) the Lothagam-Napadet Range (Robbins 1972, 1974), (5) the Kerio-Natome basin (personal communication [AB] with Robert Foley and Marta Lahr), (6) the Kalodir-Kapua Basin (Beyin 2011), and (7) the Omo River Delta/Kibish region in the northwest (Brown 1975; Butzer 1980). Riparian foragers with a more generalized subsistence and higher residential mobility patterns have also been identified from surrounding regions (Barthelme 1985, p. 277–278).

Open image in new window

Stable highstands during the EH occurred against the backdrop of the evolution of specialized subsistence groups. A low-alkaline Lake Turkana and adjacent lakes (such as paleolakes Chalbi, Suguta, and Chew Bahir) were likely rich in aquatic resources, which people exploited (Stewart 1989). EH sites adjacent to lakes Ziway and Shala in the central Ethiopian Rift also include high proportions of fish in the faunal assemblages (Ménard et al. 2014). Fishing is presumed to have taken place using barbed bone points and cooking done in ceramic vessels (Fig. 4). High local precipitation levels also promoted denser stands of vegetation as evidenced by the recovery of closed-forest species in the archaeological faunal assemblages (Barthelme 1985; Phillipson 1977; Robbins 1974; Stewart 1989, p. 170).

Open image in new window

Direct cultural affinities are difficult to establish between the inhabitants of the Turkana Basin and those of other “Aqualithic” communities in northern Africa. Obsidian analysis of artifacts from sites with barbed bone points (GaJj1 and Lowasera) indicates that raw materials for tools have a local provenance and are not coming from distant sources (Nash et al. 2011). The distribution of fishing communities in what is now the Sahara and Sahel using barbed bone points and wavy-line pottery suggests at least minimal transmission of subsistence technologies over vast areas (Stewart 1989; Sutton 1974, 1977; Yellen 1998). Such networks laid the foundation for what was to later transform the economy of sub-Saharan Africa when the ecosystem would no longer support fishing. However, regional technological variants abound such as at lakes Ziway-Shala, where the production of the “Ethiopian Blade Tool tradition” diverge, from known EH lithic traditions elsewhere in Africa (Ménard et al. 2014), suggesting localized adaptations to pluvial conditions.

Early Middle Holocene, 9,000–4,500 Years BP

This period begins with lake regressions in all local basins (Turkana, Chalbi, Suguta, Chew Bahir) potentially associated with cooler SSTs in the western Indian Ocean and weakening of the East African Monsoon (Bard et al. 1997; Tierney and deMenocal 2013). The later phases of the EMH are marked by pronounced variability in the level of Lake Turkana above 370 m.a.s.l., although the potential for regressions below the present lake elevation is plausible (Fig. 2). Paleolakes Chalbi (Nyamweru 1989), Chew Bahir (Foerster et al. 2012; Golubtsov and Habteselassie 2010; Grove et al. 1975; Umer et al. 2004), and Suguta (Garcin et al. 2009; Garcin et al. 2012; Junginger et al. 2014) show concurrent desiccation, reflecting a change in local climate and a reduction in Omo River inflows. Later transgression phases inferred for Lake Turkana may have been part of a regional pluvial phase based on recharge of smaller water bodies in the southern Ethiopian Plateau (Lamb et al. 2000; Umer et al. 2004) and the northern Kenyan Highlands (Barker et al. 2001; Butzer et al. 1972; Hamilton 1982; Olago et al. 2000).

The variable location of the CAB and irregular distribution of monsoon rainfall across the Turkana Basin catchment during the EMH would have enhanced interannual uncertainty regarding the distribution of resources. Domesticated plants and animals appear in the Fayum region of northern Egypt by 7,400 years BP (Linseele et al. 2014) and domesticated cereals had been introduced from the Near East by 7,000 years BP in northern Sudan (Madella et al. 2014). Long-distance exchange networks across northeastern Africa from this period, accompanied by diverse subsistence strategies and nascent adoption of domesticates, suggest that cultural solutions to Middle Holocene climatic variability involved migrating between resource patches. Boattini et al. (2013; see also Fernández et al. 2007) argue that the mitochondrial DNA (mtDNA) of modern Afro-Asiatic (Cushitic) populations from eastern Africa supports the linguistic interpretation that ancestral population movements into eastern Africa from the Ethiopian Plateau began ca. 7,000 years BP (Ehret 1979). Likewise, Y-chromosome data support an expansion of Afro-Asiatic speakers out of their northeast African homelands 8,000 ± 2,500 years BP with subsequent expansion across the northern periphery of central Africa and southward into equatorial Africa (Černý et al. 2009). Population movements associated with new pastoral economies were tethered to the edges of shrinking water bodies in the Sahara (Hoelzmann et al. 2001; Kuper and Kröpelin 2006) or, much later in time, following elevation gradients in the Ethiopian Highlands (Brandt and Carder 1987; Lesur 2007; Lesur et al. 2014) to secure fodder for their animals. Throughout the Horn of Africa, the adoption of domesticates appears to have occurred slowly and conservatively, with livestock being incorporated into local subsistence economies without radically transforming the fundamental technological or settlement structure of the region until much later in the Holocene (Lesur et al. 2014).

The lack of archaeological data in the Turkana region until after the EMH high lake stands suggests that two primary cultural phenomena were occurring. First, human settlement was likely closely restricted to the lake margins. Archaeological settlements are documented using charcoal or shell ages from the late regressive phase of Lake Turkana (ca. 5,000–4,500 years BP) in localities 102 and 103 in the Koobi Fora formation (Barthelme 1985, p. 22). There are few other securely dated EMH habitation sites situated on the shore margins based on depositional facies. GaJi4 is an early pastoral site with a recently generated radiocarbon age on charcoal of 4,180 ± 40 ¹⁴C years BP (4,580–4,840 cal. years BP, Beta-252053, Ashley et al. 2011), but is coeval with the final regression of Lake Turkana. Importantly, a charcoal age associated with an eroded human burial and barbed bone points at the “Lothagam Fishing Site” is 6,300 ± 800 ¹⁴C years BP (5,320–8,970 cal. years BP, UCLA-2124A, Robbins and Lynch 1978) and indicates that Aqualithic fishing traditions persisted in the region late into the EMH. In 2012, a radiocarbon age was analyzed on fish bones from the same depositional unit from which Robbins and Lynch’s (1978) charcoal age was obtained and yielded an age of 5,515 ± 30 ¹⁴C yr BP (6,280–6,400 cal. years BP, ISGS A-2348). Radiocarbon dates from sites within the Lothagam Hill area occur in the form of a charcoal age at Zu-10 with a dating to 6,200 ± 130 ¹⁴C years BP (6,790–7,415 cal. years BP, N-812). These are reportedly associated with LSA microliths, a shell age from Zu-6 with decorated pottery and LSA microliths dating to 7,960 ± 140 ¹⁴C years BP (8,450–9,240 cal. years BP, N-813), and burned clay from Bb-14 located in the Lorungelup area with pottery and LSA microliths dating to 5,020 ± 220 (5,310–6,285 cal. years BP, N-814; Robbins 1972; Yamasaki et al. 1972, p. 237–238). Lawrence Robbins indicates in the research notes in the Radiocarbon article (Yamasaki et al. 1972) that N-813 stratigraphically agrees with a previously generated shell age of 7,560 ± 1,000 (6,300–11,070 cal. years BP, UCLA-1247E) from a “nearby” location (Yamasaki et al. 1972, p. 237–238). Site “P8” contains a fragmented human burial and hippopotamus remains with an associated shell age of 4,250 ± 100 ¹⁴C years BP (4,450–5,250 cal. years BP, L-1203-I), with eroded microliths and ceramic sherds that have no clear stratigraphic association with the in situ remains (Butzer 1980). Butzer (1980) and Owen et al. (1982) added a presumed reservoir effect to all of their radiocarbon ages of +400 years. Based on current available information, the presumed reservoir effect has been “uncorrected.” Prior to the final transgression of the lake, near-shore settlement patterns left few traces in the archaeological record because sites were destroyed or deeply buried by littoral sediments by the oscillating lake level.

The second cultural phenomenon hypothesized from the available data is that residential mobility was probably relatively low and subsistence was focused on lacustrine resources. According to the results reported from Lothagam Hill (Robbins 1974; Yamasaki et al. 1972) and Kibish Plain (Butzer 1980), archaeological occurrences during the EMH were located on lacustrine or littoral deposits and appear to have been situated when the lake was close to the sites. The lack of evidence for EMH settlements outside the strand plain suggests continuity with the preceding EH settlement scheme of intensively occupied sites located adjacent to a water body, although there was likely a turnover in available fish species as lake salinity increased (Cerling 1979) and new interregional exchange networks evolved (Stewart 1989, p. 168–169).

Analysis of ichthyofaunal assemblages from EH sites in the basin shows that during the EH high stand, people focused on large fish such as Nile perch and cichlids (Stewart 1989). Stewart (1989) finds that with the retreat of the lake in the EMH, the diversity of exploited aquatic species increased and small fish dominated the human diet. On the basis of this observation, the initial use of pottery in eastern Africa may have arisen to cook small fish during the time when most large fish became scarce in the lake.

The ceramic data provide compelling evidence for the persistence or evolution of a regional cultural exchange network. The presence of dotted-wavy line pottery around the Turkana Basin in association with barbed bone points, formerly called “harpoons,” has been published from EH–EMH deposits in Koobi Fora (Barthelme 1985) and wavy-line pottery at Lowasera (Phillipson 1977). Whether or not these finds are, in fact, dotted-wavy line/wavy-line traditions is highly speculative. A detailed, unpublished report made by Alan Jacobs in 1972 of dotted-wavy line and wavy-line ceramics from a surface collection made in the Eliye Springs area describes structural, decorative, and compositional features of sherds that associate them with the Khartoum Mesolithic (see also Fig. 4), but all of these identifications and conclusions require careful scrutiny. If confirmed, this is a likely extension of the bone-point (Aqualithic) traditions distributed across Saharan Africa (Sutton 1974, 1977), which included dotted-wavy line and wavy-line ceramics and had more restricted residential mobility patterns (Hoelzmann et al. 2001; Kuper and Kröpelin 2006; Kuper and Riemer 2013). Dotted-wavy line pottery and barbed bone points have been previously associated with Nilo-Saharan (Nilotic) speakers (Ehret 1974, 1993; Haaland 1992, 1993), who maintained broad, regional exchange networks across northern Africa throughout the EMH (Barich 2002; Mohammed-Ali and Khabir 2003). By the end of the EMH, members of the Afro-Asiatic (Cushitic) language family had expanded throughout northern Africa (Ehret 1974, 1998; Haaland 1992) replacing broad-spectrum foraging/fishing/animal husbandry with specialized forms of pastoralism in the Sahara (Barich 2002). However, in the Turkana Basin, specialized pastoralism did not emerge until much later (Barthelme 1985; Marshall et al. 1984) despite linguistic evidence suggesting that Cushitic speakers had arrived in the region (Ehret 1974, 1998). Thus, based on the scant available evidence during the early portions of the EMH, the Turkana Basin may have been an area where syncretic blending of Nilo-Saharan, Cushitic, and Khoi subsistence (and probably material/social) cultures occurred. As in the northern portions of the Horn (Lesur et al. 2014), there is no evidence to suggest that there was a rapid shift in the subsistence economy or radical change in the technology until the latest phase of the EMH in the Turkana Basin.

The appearance of monumental mortuary features known as namoratunga appear to have no direct cultural precedent elsewhere in Africa, but are widely distributed across the Turkana Basin at the end of the EMH (Grillo and Hildebrand 2013; Hildebrand and Grillo 2012; Hildebrand et al. 2011; Lynch and Robbins 1978, 1979; Nelson 1995). Insomuch as archaeological traditions can show diachronic culture changes, these sites appear qualitatively distinct from their known antecedents. Namoratunga are comprised of compound human burials of which many individuals possess ornamental grave goods made from ostrich eggshell and worked bone, semiprecious stones such as amazonite and carnelian, and ceramic zoomorphic figurines (Grillo and Hildebrand 2013; Hildebrand et al. 2014). Found on both east and west sides of the lake, namoratunga are interpreted as the remains of a single cultural group attesting to the presence of well-developed transportation and exchange networks at this time (Grillo and Hildebrand 2013). Grillo and Hildebrand (2013) argue that namoratunga sites are associated with the first herders in the basin, but the seemingly rapid appearance of both domesticated animals and elaborate mortuary features may reflect a preservation bias against sites that predate them due to the transgression of the lake between 6,000 and 5,000 years BP. In light of the profound plunge in the level of the lake between 5,000 and 4,000 years BP, namoratunga may represent a cultural response to maintain social solidarity through an ecologically challenging circumstance.

Late Middle Holocene, 4,500–2,500 Years BP

This period is marked low lake levels, not only in Lake Turkana, but throughout northern and equatorial Africa associated with the termination of the AHP (Fig. 2; Berke et al. 2012a; Gasse 2000; Kröpelin et al. 2008; Mayewski et al. 2004). In the Turkana region, the termination was abrupt, but the magnitude of the drop was within the scope of fluctuating lake levels that occurred during the EMH (Bloszies et al. in press; Forman et al. 2014). The difference between LMH and EMH conditions is that the mean position of the CAB made a final westward shift between 5,000 and 4,000 years BP, which resulted in diminished rainfall adjacent to the lake margins. Smaller adjacent lakes such as Chew Bahir (Foerster et al. 2012) and Suguta (Junginger et al. 2014) reflect the reduced localized rainfall as the region transitioned to more permanently xeric conditions.

Animal husbandry was present in the Lake Turkana region by at least 4,500 years BP by people making jars with elaborate impressions called Nderit Ware (Fig. 5; Barthelme 1985; Marshall et al. 1984). The construction of namoratunga ceased by this time (Grillo and Hildebrand 2013) and the Lake Turkana region’s archaeological assemblages became diverse (Marshall et al. 1984, p. 126). The archaeological culture of these first herders is called the Pastoral Neolithic (PN), comprised of Afro-Asiatic speakers who kept domesticated animals and produced and used microlithic tools as well as the aforementioned ceramics (Bower 1991; Bower and Nelson 1978; Bower et al. 1977).

Open image in new window

Initially, livestock was incorporated as an additional subsistence component within the broad-spectrum, riparian foraging regime that characterized EH and EMH settlements. One facet of the subsistence regime that clearly changed between the EH and the LMH is the near disappearance of barbed bone-point fishing techniques and exploitation of Nile perch. However, early PN settlements appear to have been restricted to the margins of the lake with no known archaeological settlements documented in the foothills or adjacent areas away from permanent water sources. The majority of artifact assemblages at LMH sites postdate 3,000 years BP and demonstrate that generalized subsistence strategies were practiced, including hunting and fishing, long after the appearance of specialized forms of pastoralism elsewhere in eastern Africa (Barthelme 1985; Robbins 1984; Wright and Forman 2011).

Another means that humans used to buffer themselves against environmental uncertainty was to develop cultural exchange networks. There are two primary lines of evidence to suggest that there was a robust interbasin cultural exchange system in operation from at least the LMH with potential deep roots in the EMH or EH. First, there are widely distributed ceramic traditions that suggest ongoing social interactions. Nderit tradition ceramics are virtually ubiquitous on PN sites from this time period (Wandibba 1981). Other vessel forms such as a distinct impressed coil jar with an orangish fabric found along the southeastern and western shores of Turkana (Fig. 6) are poorly dated, but their association with microliths on the strand plain that postdates the final regression of the lake strongly suggests a LMH provenience. Turkwel tradition ceramics occur during the late LMH and are also found across primarily on the western shoreline of the lake and finds of these artifacts extend into eastern Uganda (Fig. 7; Lynch and Robbins 1979; Robbins 1984). Second, analysis of obsidian analysis shows sustained trade in artifacts and raw materials across the lake, probably by the use of watercraft (Nash et al. 2011; Ndiema et al. 2011). The obsidian analysis shows the lack of cultural barriers inhibiting the movement of specific materials across ethnic or language divides. Lake Turkana was the incubator of early livestock herding cultures of equatorial and southern Africa (Bower 1991, 1996; Gifford-Gonzalez 1998, 2000; Wright 2011), and the archaeological and social histories of the region attest to a long legacy of sustained cultural syncretism.

Open image in new window

Open image in new window

The spread of domesticated animals did not advance in a significant way south of the Turkana Basin until after 3,000 years BP (Bower 1991; Gifford-Gonzalez 1998; Marshall 2000). The spatial distribution of obsidian artifacts identified from the region indicate that although there is evidence for wide ranging local exchange networks, there is virtually no evidence for the introduction of raw materials from the East African Highlands to the south or the Ethiopian Highlands to the north, where high quality and quantity obsidian sources are abundant (Nash et al. 2011). As the geographic expansion of food production stalled in Lake Turkana, the frontier entered a “consolidation phase.” During consolidation phases of human frontier expansion, intensification and specialization of agricultural production techniques occurs (Alexander 1980; Lane 2004). This process was apparently slow in the Turkana Basin (lasting at least 1,500 years) and involved sustained intercommunity cultural exchanges. The consolidation of animal husbandry techniques occurred against the backdrop of xeric conditions across eastern Africa.

Late Holocene, 2,500–200 Years BP

There are few data to constrain lake level between 4,000 and 200 years ago besides a sediment core extracted from the southeastern portion of the lake and seismic data (Halfman et al. 1992; Halfman and Johnson 1988; Morrissey and Scholz 2014). Seismic unconformities dated to approximately 2,300 years BP indicate that the lake may have been 40 m below its current level (Morrissey and Scholz 2014). The terrigenous pollen record from the lake core infers a local arid interval centered between 2,500 and 1,600 years BP (Mohammed et al. 1995). Inversely, more pluvial adapted taxa are abundant during the subsequent period spanning from 1,600 to 600 years BP (Mohammed et al. 1995), which dates to the inferred transgression of the lake based on TOC and Total Inorganic Carbon (TIC) assayed from the core (Morrissey and Scholz 2014). Charcoal from the archaeological site of Apaget 1 located at 9 m above the present lake level dates to 1,800 ± 300 ¹⁴C years BP (1,070–2,430 cal. years BP, UCLA-2124K) with ages of 870 ± 80 ¹⁴C years BP (680–930 cal. years BP, UCLA-2124G), 950 ± 80 ¹⁴C years BP (690–1,050 cal. years BP, UCLA-2124J), 1,100 ± 80 ¹⁴C years BP (800–1,240 cal. years BP, UCLA-2124H), and 1,375 ± 125 ¹⁴C years BP (1,000–1,540 cal. years BP, GX-5041) is reported from the site of Lopoy, which is 18 m above the lake level (Lynch and Robbins 1979; Robbins 1980, 1984). These archaeological data are the terminus post quem for the transgression of the lake during the LH.

Population movements and the advent of specialized food producers, some of whom relied on highly mobile transhumant pastoralism, are documented from this period. The site of Namoratunga I with a bone collagen date of 1,200 ± 100 ¹⁴C years BP (940–1,290 cal. years BP, UCLA-21240; Soper 1982) is one such site interpreted as showing the presence of specialized pastoralists (Robbins 1984, p. 207). The archaeological evidence correlates to a loosely dated period when migrations of eastern Afro-Asiatic and eastern and western Nilo-Saharan speakers occurred into the basin from the north and east (Ehret 2011, p. 117–119; Schlee 1985; Spear and Waller 1993). MtDNA genetic analysis of 287 individuals from across Kenya, including Turkana, shows extensive admixture between East African communities sufficient to mask “genetic founders” of distinct populations (Castrì et al. 2008; see also Batai et al. 2013). Population movements and interregional exchange networks evolve in the context of the development of highly itinerant pastoralism in eastern Africa (Marshall 1990, 2000).

LH inhabitants of Turkana continued to utilize stone tools and ceramics, but archaeological sites after 3,000 years BP show diversity in settlement patterns. Robbins (1980, 1984) defines five primary late prehistoric subsistence economies from the west side of Lake Turkana: (1) specialized cattle herders who did not fish nor hunt wild animals, (2) lakeside fishers using thin-sided Akira pottery who probably kept domesticated livestock but did not hunt wild animals, (3) lakeside hunter/pastoralists who did not fish, (4) relatively sedentary lakeside fishers using thick-walled/grooved Turkwel pottery who kept livestock and did little hunting, and (5) sorghum cultivating agropastoralists who inhabited the Turkwel River riparian zone and exploited the annual lake level fluctuations. Robbins (1984) points out that these archaeological signatures are not necessarily inclusive of the total range of cultural diversity and subsistence options available or exploited by the late prehistoric inhabitants of the region (see also Mutundu 2010 for an ethnoarchaeological perspective on site faunal patterning in eastern Africa).

The lakeside site of Lopoy (radiocarbon dates provided above) is probably the best-studied LH site in the Turkana Basin and shows the resilience and flexible nature of settlement through the various phases of occupation. People continued to produce and heavily rely on stone and bone tools long into the East African Iron Age and subsistence was focused on shoreline resources (Robbins 1980). Two groups are presumed to have occupied the site: one with Akira ceramics in which people were primarily hunting and the other using Turkwel ceramics in which people were primarily subsisting by fishing (Robbins 1980). Midden deposits appear to show continuous, year-round occupation of the site for hundreds of years and indicate low residential mobility (Robbins 1980).

Our view is that these subsistence categories might reflect the general archaeological site types found across the basin, but subsistence is a flexible concept that people change when the environmental or cultural situation warrants it. Many modern pastoralists also raise crops to supplement their diets and many so-called farmers keep small numbers of livestock. Mace (1993) argues that agropastoral households throughout sub-Saharan Africa adopt different subsistence strategies based on yearly subsistence needs. When resources are scarce, households will concentrate on the most viable means of subsistence available to them at the time. Because every ecosystem is different, there is no consistent “fallback” that most households will adopt. Chance plays a large role in determining household “wealth,” but this chance can be minimized if flexible farming and pastoral strategies are adopted (Mace 1993, p. 365). In areas that frequently experience rapid climate fluctuations, this can be an important strategy to minimize the risk of complete crop failure or herd decimation.

The case of Ele Bor near the eastern edge of the Turkana Basin provides such an example of shifting resource utilization strategies in the face of different environmental pressures. Holocene occupations of the Ele Bor Rockshelter extend from the EMH to the LH period (Phillipson 1984). Archaeofauna show the shifts in foraging and pastoral strategies from large-bodied to small-bodied species, reflecting the available biomass and reduced precipitation from 5,000 years ago to recent times (Gifford-Gonzalez 2003). The data do not show specialized subsistence pursuits, but reflect flexible resource acquisition as a means of adapting to different climate and vegetation pressures.

Historic Period, 200–0 Years BP

European exploration of the Lake Turkana area began in earnest when Count Samuel Teleki and Ludwig von Höhnel arrived at the lake in 1888 and named it after their benefactor, Austrian Crown Prince Rudolf (von Höhnel 1894). However, the “northern frontier” was encircled by European interests and proxy claims for decades preceding the physical arrival of explorers to the region (Parkingham 1991). The region was noted for its inhospitable climate, rugged terrain, and fierce populations and was never fully incorporated into either the British colonial or Kenyan postcolonial governance systems. Christian missionaries, the United Nations, nongovernmental organizations (NGOs), and the Kenyan government have constructed the most enduring infrastructure around the lake margins in the form of schools, churches, and aid distribution centers and have become critical fallback centers for the local populations during periods of resource stress.

The primary factor separating the HP from the preceding periods is the presence of a 100+ year lake level curve based on gauges and altimeters. Critically, the data show the scale of lake level oscillations measured from ca. AD 1900 to 1955, and the lake dropped 20 m in absolute elevation (Butzer 1971, p. 123; Källqvist et al. 1988). Unfortunately, complementary rainfall data for the southern Ethiopian Highlands for the same period is unavailable and the scant records that exist are difficult to calibrate (Conway et al. 2004). In piecing together the fragmentary precipitation data that do exist using a singular value decomposition analysis to gridded rainfall data from southern Ethiopia, there does not appear to be a linear decline in precipitation that the lake level data would suggest (Jury 2010) indicating that Turkana lake levels are tied to more factors than southern Ethiopian precipitation alone (see Bloszies and Forman 2015 for a full discussion).

The Turkana word for the lake is anam ka’alakol or “lake of many fish,” suggesting that aquatic resources have been important food resources for many centuries. However, the degree to which this resource was dependable certainly varied on water levels, which control salinity in the lake (Ferguson and Harbott 1982; Källqvist et al. 1988; Yuretich 1979; Yuretich and Cerling 1983). Large-bodied species, such as Nile perch, are high trophic organisms that are saline-tolerant, but they migrate into deeper waters following lower-chain species that are more sensitive to slight changes to the water’s chemistry (Källqvist et al. 1988).

The diets of the modern inhabitants of the Lake Turkana region are specialized, reflecting intercommunity differences in settlement and subsistence practices. Stable isotope analyses conducted on modern populations living adjacent to Lake Turkana show clear divergences in dietary habits between ethnic communities (Kiura 2005). The three groups of people studied (Dassanetch, Elmolo, Gabbra) are clearly separated in their δ¹³C and δ¹⁵N values, reflecting their drastically different diets (Fig. 8). However, LH subsistence diversification is also recorded among HP “pastoralists” when factors such as distance to markets, gender, family wealth, and local ecology are accounted (Little et al. 2001). The differences between how subsistence specialization takes form in the Turkana Basin between the LH and HP are more reflective of the synchronic nature of cultural anthropological data and diachronic nature of archaeological data than of true behavioral differences in people from the two periods.

Open image in new window

Early ethnographic studies within the Turkana region suggest that clear patterning of subsistence according to ethnic affiliation was deeply ingrained in the cultural fabric of the region (Fig. 9; Gulliver 1951, 1955). Despite sharing a common language, settlement, and social structure, the Rendille and Ariaal pastoralists who inhabit contiguous territories in northern Kenya maintain distinct ethnic identities based on different modes of livestock production and ecological niche exploitation (Fratkin 1986). However, pastoralists like the Dassanetch and Turkana have traditionally been adept at absorbing immigrants into their communities and maintaining robust interethnic alliances so that they can access resource patches during ecologically challenging periods (de Vries et al. 2006; Sobania 1988, 2011). Schlee (1991) documents ritualized movements of nomadic pastoralists in the circum-Turkana region based on clan affiliations and scheduled circumcision rituals. An ethnoarchaeological site documented on the Mount Porr strand plain (GeJk18) was reported as the location of one such annual circumcision and marriage festival (Wright and Forman 2011). These movements reinforce social cohesion of the groups and maintain territorial boundaries based on competition for resources and ethnic affiliation.

Open image in new window

Over the last 50 years, the Turkana Basin has been pulled into the global economy and indigenous social networks have been refocused to control access to this new resource base. Oil has been discovered and is being drilled in select areas of the region. Prior to that, Lake Turkana was a focus of humanitarian efforts led by the United Nations, NGOs, Christian missionaries, and the Kenyan government, all of which had little understanding of the local economy (Fratkin 1991; McPeak and Barrett 2001; Middleton and O'Keefe 1997). With so many centers where nonlocal resources are infiltrating the region, there has been a substantial shift in the settlement ecology and nature of interactions between people living in the region. Presently, many pastoralists use the villages as child-rearing centers while males who are able to shepherd animals move from pasture to pasture, returning home occasionally to sell animals or resupply their families (de Vries et al. 2006; Fratkin 2001; Schlee 1991; Sun 2005). Conflicts over resource patches have also developed. In the past, interdependent social networks evolved over vast areas to mitigate risk associated with the changing climate. However, as the local economy is increasingly dependent on outside resources, there are fewer incentives to cooperate within the neighborhood and more incentives to fight a zero-sum game for primacy in controlling access to those resources.