Rock varnish record of the African Humid Period in the Lake Turkana basin of East Africa

Abstract

Rock varnish is a manganiferous dark coating accreted on subaerially exposed rocks in drylands. It often contains a layered microstratigraphy that records past wetness variations. Varnish samples from latest Pleistocene and Holocene geomorphic features in the Lake Turkana basin, East Africa display a regionally replicable microstratigraphy record of Holocene millennial-scale wetness variability and a broad interval of wetter conditions during the African Humid Period (AHP). Three major wet pulses in the varnish record occurred during the generally wet interval of the early Holocene (11.5–8.5 ka) when the lake attained its maximum high stand (MHS) at 455–460 m. A >23 m drop from the MHS occurred between 8.5 and 8 ka. Subsequently two additional wet pulses occurred during the early to middle Holocene (8–5 ka) when the lake occupied its secondary high stand at 445 m. Collectively, these five wet phases represent an extended wet interval coincident with the AHP in the region. One moderate wet phase occurred during the subsequent climatic transition from the humid to arid regime (5–4.3 ka) after the lake level dropped rapidly from 445 m to <405 m. Five minor wet phases took place during the overall arid period of the late Holocene (4.3–0 ka) when the lake level oscillated below 405 m. These findings indicate that the AHP terminated rapidly around 5 ka in the Turkana basin in terms of lake level drop, but the regional shift in relative humidity from the AHP mode to its present-day condition lagged for about 700 years until 4.3 ka, hinting at a gradual phasing out in terms of moisture condition. These findings further suggest that Lake Turkana overflowed intermittently into the Nile drainage system through its topographic sill at 455–460 m during the early Holocene and has become a closed-basin lake thereafter for the past 8 ky.

Keywords

African monsoon desert varnish Holocene climate variability microstratigraphy millennial-scale wet event shoreline fluctuation

Introduction

North and East Africa experienced the “African Humid Period” (AHP) during the early to middle Holocene, followed by the continuing late-Holocene dry period (DeMenocal and Tierney, 2012; DeMenocal et al., 2000; Tierney and DeMenocal, 2013). The AHP was largely caused by dramatic increases in summer precipitation triggered by orbital forcing of African monsoonal climate and amplified by oceanic and terrestrial feedbacks (Claussen and Gayler, 1997; Kutzbach and Otto-Bliesner, 1982). Paleoclimate proxy records such as relict shorelines, pollen, potassium content, carbonate δ¹⁸O, and leaf wax δD retrieved from marine and terrestrial sediments all show a prolonged wet period between ca. 11.5–4.8 ka (1000 years before present) in tropical Africa north of 10° S latitude (Armitage et al., 2015; Collins et al., 2017; Foerster et al., 2012; Garcin et al., 2012; Gasse, 2000; Junginger et al., 2014; Khalidi et al., 2020; Tierney et al., 2008). Several centennial-scale dry spells interrupted this overall wet period, and some of them are coeval with millennial-scale cooling events such as those at 8.2 and 4.2 ka in the North Atlantic (Bloszies et al., 2015; Bond et al., 1997; Garcin et al., 2012; Gasse, 2000; Junginger et al., 2014). However, the timing and mode of the AHP termination (rapid or gradual) have been intensively studied and hotly debated, as reflected in the paleoclimatic records from lacustrine deposits of African lakes and marine sediments (Armitage et al., 2015; Costa et al., 2014; DeMenocal et al., 2000; Jaeschke et al., 2020; Kröpelin et al., 2008; Shanahan et al., 2015; Tierney and DeMenocal, 2013; Van der Lubbe et al., 2017). For example, leaf wax δD and dust proxy records in marine and lacustrine sediment cores from East and West Africa show a nearly synchronous AHP termination within centuries around 5 ka (DeMenocal et al., 2000; McGee et al., 2013; Tierney and DeMenocal, 2013), whereas proxy records from numerous lake and coastal sites in North Africa suggest a locally rapid but spatially time-transgressive termination, starting in the north and moving progressively later at lower latitudes between 6 and 3 ka (Shanahan et al., 2015, and references therein). A new transient comprehensive Earth system model simulation of the last 8000 years across North Africa largely reproduces the time-transgressive termination of the AHP evident in proxy data, with an earlier end in the north than in the south and in the east than in the west, and attributes the causes to the regionally varying dynamical controls (such as orography, rainfall seasonality and local circulation regime) on precipitation (Dallmeyer et al., 2020). Moreover, paleotemperature data from Lake Turkana in East Africa reveal a thermal anomaly of 2–4°C over the two millennia spanning the end of the AHP that may be linked to local insolation during September-November (Berke et al., 2012). Previously reconstructed Holocene water level curves of Lake Turkana (e.g. Bloszies et al., 2015; Brown and Fuller, 2008; Butzer, 1980; Forman et al., 2014; Garcin et al., 2012; Owen et al., 1982) also show diverse hydroclimatic responses during the AHP and its termination. Here we report new rock varnish evidence for the hydroclimate variability within the AHP and associated water level variations in the Lake Turkana basin. Results from this study shed new light on the termination of the AHP in the region and also help constrain the Holocene shoreline fluctuations of Lake Turkana.

Rock varnish as a desert wetness recorder

Rock varnish is a slow-accreting (1–40 μm/ky) manganese- and iron-rich dark coating on subaerially exposed rock surfaces, mostly seen in the desert regions of the world (Dorn and Oberlander, 1981; Liu and Broecker, 2000; Perry and Adams, 1978). Because of its sedimentary origin, varnish often contains a layered microstratigraphy that records past climate changes (Broecker and Liu, 2001; Liu et al., 2000). In the Great Basin of the western US, Mn- and Ba-poor varnish layers, which are orange/yellow in ultra-thin sections (5–10 μm thick) under transmitted polarized light, were formed during the overall dry period of the Holocene; Mn- and Ba-rich black layers were deposited during the last glacial interval when the Great Basin was much wetter than at present (Broecker and Liu, 2001; Liu and Dorn, 1996). This observation suggests that, in varnish microstratigraphy, Mn- and Ba-rich black layers are diagnostic of relatively humid climate, while Mn- and Ba-poor orange/yellow layers are indicative of relatively dry climate. Previous studies show that wet events represented by the glacial-age black layers in the western US varnish correlate in time with cold episodes of the Younger Dryas (YD) and Heinrich Events (H1–H6) in the North Atlantic (Liu, 2003; Liu et al., 2000). Following studies (Liu and Broecker, 2007, 2008, 2013; Liu et al., 2013) further demonstrate that fast-accumulating varnish recorded last glacial and Holocene millennial-scale wet events synchronous with millennial-scale cooling events in the North Atlantic (Bond et al., 1997). Hence, rock varnish can be used as a unique wetness recorder to study past climate changes, especially wetness variations in the world’s deserts (e.g. Baied and Somonte, 2013; Cremaschi, 1996; Dorn, 2009; Goldsmith, 2011; Lee and Bland, 2003; Zerboni, 2008).

Study area

Lake Turkana is located at the latitudes of 3°–6° north of the equator within the eastern branch of the East African Rift System in Kenya (Figure 1b). It is about 250 km long and 30 km wide, with a maximum water depth of 120 m and average depth of 35 m (Johnson and Malala, 2009). Lake Turkana is presently a closed-basin lake and its water level is therefore very sensitive to changes in regional precipitation. It receives about 90% of water input from the Omo river that drains the Ethiopian Highlands to the north (equivalent to 2.3 ± 0.6 m in lake level), and derives the remaining 10% of water flow primarily from the Turkwel and Kerio rivers to the south (equivalent to 0.13 m in lake level) (Avery, 2010). The climate of the Turkana basin is hot and arid, with extended periods of unusually intense diurnal winds (Johnson and Malala, 2009). Mean annual maximum and minimum air temperatures are 32.5°C and 26°C, respectively (Ferguson and Harbott, 1982). Mean annual rainfall is about 200 mm/year (Nicholson et al., 1988), while mean annual evaporation is about 2300 mm/year, which translates to a lake level drop of 2.33 m/year during the dry season (Ferguson and Harbott, 1982). The rains over the watershed of Lake Turkana are highly seasonal and about >80% of the rainfall occurs between March and November with the biannual passage of the East and West African monsoons (Avery, 2010). This reflects the dynamics of the Intertropical Convergence Zone (ITCZ), which migrates to the north over the Ethiopian Highlands during boreal summer and to the south of Lake Malawi during boreal winter (Johnson and Malala, 2009).

Figure 1.

Location maps (a, b) of the Lake Turkana basin in East Africa. The blue line in (b) delineates the lake’s maximum high stand (MHS) at 455–460 m attained during the early Holocene (Brown and Fuller, 2008; Butzer, 1980; Garcin et al., 2012; Owen et al., 1982). The terrain map was adapted from Garcin et al. (2012) with modifications. Google Earth image of the Lothagam basaltic hills (c) shows the locations of varnish sampling sites.

Like many rift valley lakes in East Africa, Lake Turkana expanded aerially and volumetrically during the AHP when the ITCZ and associated monsoonal rain belt migrated to a more northern position than today (Braconnot et al., 1999; Kutzbach and Liu, 1997). Holocene lacustrine deposits are found around the margins of the Turkana basin, almost 95 m above the current water surface of about 360 m (meters above mean sea level) (Brown and Fuller, 2008; Butzer, 1980; Owen et al., 1982). Holocene shorelines at elevations of 455–460 m have been interpreted as the maximum high stand (MHS) zone for Lake Turkana during the AHP (Brown and Fuller, 2008; Butzer, 1980; Garcin et al., 2012; Owen et al., 1982) (Figures 1 –3). For instance, in his pioneer geomorphological investigation of the region, Butzer (1980) obtained an elevation of 460 m for the highest lake stand using multiple altimeters to cross check while a control aneroid was read at regular intervals in the base camp (with reference to the 376 m level of Lake Turkana in 1968). Regional geological survey by Brown and Fuller (2008) yielded an elevation estimate of 455–460 m for the lake overflow level on the topographic map of the Turkana basin, which was compiled from the Shuttle Radar Topographic Mission data (90-m posting). Recent kinematic differential GPS survey of the paleoshorelines by Garcin et al. (2012) documented the present-day maximum elevation of the lake overflow at 460 m. The lake level was relatively low, around 440–410 m, during the YD arid period (13–11.5 ka) (Bloszies et al., 2015; Forman et al., 2014; Garcin et al., 2012). During the Holocene, the lake spilled over into the Nile drainage basin through a northwest topographic low at 455–460 m and also made hydrological connections to the south with water bodies from the Suguta Depression and to the north with water bodies from the Chew Bahir Basin overflowing into the Turkana basin (Bloszies et al., 2015; Brown and Fuller, 2008; Butzer, 1980; Garcin et al., 2012; Harvey and Grove, 1982; Junginger et al., 2014; Owen et al., 1982; Van der Lubbe et al., 2017) (Figure 1b). Sedimentary sequences emplaced during the onset of the AHP are stratigraphically thin, laterally discontinuous, and not ubiquitously distributed around the Turkana basin (Butzer et al., 1972; Owen et al., 1982). Coring of sediments under the modern lake’s waters has been difficult because of high winds affecting the stability of watercraft and a rapid sedimentation rate making it difficult to obtain a core that goes beyond the last few 1000 years (Halfman et al., 1992; Morrissey and Scholz, 2014). Hard water reservoir effects and carbonate fractions of lake sediments may pose some challenges with radiocarbon dating of outcrop deposits (Butzer et al., 1972; Halfman et al., 1994; Owen et al., 1982). Therefore, new datasets and alternative approaches are needed to understand the Holocene hydroclimatic history of Lake Turkana.

Figure 2.

Field photos showing the geomorphic settings of the varnish sampling sites. Numbers in parentheses are the elevations (meters above mean sea level; ±3 m) of the sampling spots obtained from the Google Earth imagery and checked in the field with GPS and altimeters: (a) Turkwel river south site (View S). (b) Varnish coated pebble (LT-1). (c) Tombolo site (View W). (d) Lothagam north pillar stones site (View W). (e) Lothagam west pillar stones site (View S). (f) Field camp site (View S).

Figure 3.

Oblique Google Earth image (a) shows the Holocene high stands (HS) and recessional shorelines of Lake Turkana on the east-facing basaltic hillslopes at Lothagam. Field photo (b) presents a ground view of part of the Holocene high stands and recessional shorelines in (a). The maximum high stand (MHS) of lake phase 1 (ca. 11.5–8.5 ka) truncated older debris-flow fan and bedrock surfaces at 455–460 m, while the high stand of lake phase 2 (ca. 8–5 ka) carved onto relatively younger debris-flow fan surfaces at ~445 m. No shorelines are visible on the younger debris-flow fan surfaces between 445 and 460 m, indicating that the lake has not transgressed to its MHS since ~8 ka. The younger debris-flow fan deposits extend downslope from 460 to 437 m, further suggesting a lake level drop of at least 23 m at ca. 8.5–8 ka during the rapid regression of lake phase 1. The archaeological site of the Lothagam North Pillar Stones is situated on the shoreline bench at 437 m. Radiocarbon dating by Hildebrand and Grillo (2012) indicates the erection of these pillar stones by humans around 5–4.8 ka, yielding a minimum-limiting age constraint of ~5 ka for the lake regression from the 445 m high stand during the AHP termination. Varnish sample (LT-4) was collected from the 455 m wave-abraded bedrock face at this locality for age calibration (ca. 8.5–8 ka) of the layering sequence shown in Figures 4d and 5.

Sample collection and methods

In this study, we documented Holocene millennial-scale wetness variations in the Lake Turkana basin using varnish microstratigraphic records from geomorphic features such as desert pavements, relict shorelines, hillslope deposits, and fluvial terraces at Lothagam and the surrounding regions (Figures 1 –3; Table 1), where radiocarbon dates on lake sediments and raised shorelines have been previously reported (Bloszies et al., 2015; Brown and Fuller, 2008; Butzer, 1980; Forman et al., 2014; Garcin et al., 2012; Owen et al., 1982; Robbins, 1972; Robbins and Lynch, 1978) and thus can be used to radiometrically calibrate the varnish wetness record. Manganiferous dark varnish is well developed and preserved on surficial rock clasts of various lithologies (such as basalt and quartzite) in the study area (Figure 2). Given the time-transgressive nature of varnish accretion on subaerially exposed rocks (Liu and Broecker, 2000), we collected only the largest possible in-situ varnished clasts from stable geomorphic surfaces where no post-depositional surface erosion by fluvial activities or bioturbation was observed at the sampling spots (Figure 2). This sampling strategy assures that the oldest possible varnish was collected from each of the sampled geomorphic surfaces for the retrieval and age calibration of the varnish wetness record.

Table 1.

Rock varnish samples collected from the Lake Turkana basin, East Africa.

Sample location	Sample label	Latitude (N) and longitude (E)	Altitude (m)	Age control (ka)	Geomorphic context and varnish development
Turkwel River South	LT-1	03°05′02.3″/35°50′30.5″	500	>11.5^a	Well-varnished quartzite pebble on pre-Holocene desert pavement
Koobi Fora North	LT-2	04°08′35.5″/36°15′36.9″	461	~11.5^b	Well-varnished basaltic boulder on hillslope cut by the MHS
Tombolo Site	LT-3	02°55′09.7″/36°03′08.7″	461	~11.5^b	Well-varnished basaltic boulder on hillslope cut by the MHS
Tombolo Site	LT-5	02°55′01.1″/36°03′18.9″	451	8.5–8^c	Well-varnished basaltic boulder on the 450–455 m tombolo surface
Tombolo Site	LT-6	02°55′01.1″/36°03′18.9″	451	8.5–8^c	Well-varnished basaltic boulder on the 450–455 m tombolo surface
Lothagam North Pillar Stones	LT-4	02°55′54.7″/36°03′10.8″	455	8.5–8^c	Well-varnished basaltic bedrock cut by the 455 m HS shoreline
Lothagam West Pillar Stones	LT-7	02°55′46.4″/36°02′51.3″	445	5–4.8^d	Well-varnished basaltic boulder on the 445 m HS shoreline
Field Camp Site	LT-8	02°56′13.0"/36°03′34.0″	405	<5^e	Well-varnished basaltic boulder on abandoned strath terrace

Morphostratigraphic age based on a well-developed pre-Holocene desert pavement surface above the 11.5–8.5 ka maximum high stand (MHS) at 455–460 m.

Morphostratigraphic age based on relict hillslope deposits truncated by the 11.5–8.5 ka MHS (Butzer et al., 1972; Garcin et al., 2012; Owen et al., 1982).

Radiometric ages based on the dated lake regression from the 450-455 m high stands (HS) (Butzer et al., 1972; Garcin et al., 2012; Owen et al., 1982).

Radiocarbon ages from the Lothagam West Pillar Stones site occupied at 5-4.8 ka after the lake level drop from the 445 m HS (Hildebrand and Grillo, 2012).

Morphostratigraphic age based on a strath terrace formed after the rapid lake level drop below 405 m around 5 ka (Forman et al., 2014; Garcin et al., 2012).

Varnish samples were collected from six selected locations in the study area (Figures 1 –3; Table 1). At these sampling sites, rock clasts on desert pavements or patches of stabilized hillslopes higher than 460 m were never covered by lake water, providing a varnish record of at least the past 11.5 ky (1000 years) on samples (LT-1, 2, and 3). The basalt gravel shorelines between 460 and 450 m were formed during the first half of the AHP (ca. 11.5–8.5 ka; lake phase 1) (Brown and Fuller, 2008; Butzer et al., 1972; Garcin et al., 2012; Owen et al., 1982), hosting a varnish record of the past ~8.5 ky on samples (LT-4, 5, and 6) after the lake level dropped from these elevations and varnish started to accrete on the sampled rocks. The basalt gravel shorelines at or below 445 m were formed during the second half of the AHP as well as its termination (ca. 8–5 ka; lake phase 2) (Forman et al., 2014; Garcin et al., 2012; Nutz and Schuster, 2016; Owen et al., 1982), yielding a varnish record since ~5 ka on samples (LT-7 and 8). Small debris-flow fans were developed on the hillslopes of Lothagam between 460 and 437 m following the ca. 8.5–8 ka lake regression (Bloszies et al., 2015; Butzer, 1980; Forman et al., 2014; Garcin et al., 2012; Owen et al., 1982), indicating a lake level drop of at least 23 m (Figure 3). Radiocarbon-dated archaeological sites of early to late Holocene ages are scattered on those abandoned shorelines (Beyin, 2011; Butzer et al., 1972; Hildebrand and Grillo, 2012; Robbins, 1969, 1972; Wright and Forman, 2011), offering opportunities for elevation-specific age calibration of the varnish wetness record (Figure 2d and e).

In the lab, ultra-thin sections of varnish samples were prepared using a special method detailed elsewhere (cf. Goldsmith, 2011; Liu and Dorn, 1996), which reduces failure rates from 80% to <5% and permits rapid preparation and hence inter-comparison of many sections. The conventional way of making varnish thin sections cannot be employed because the resulting thin sections are too thick (25–30 μm) and do not reveal the microstratigraphy, which is opaque in normal geological thin sections. Because varnish also accretes in a time-transgressive fashion within millimeter-sized microbasins on subaerially exposed rock faces, only well-developed and preserved varnish microstratigraphies within shallow microbasins (with a depth/size diameter ratio of about 1–3/10) were selected for ultra-thin sectioning, which would maximize the chance of getting the oldest possible varnish records close to the surface exposure of the dated geomorphic features (Liu and Broecker, 2008, 2013) (Figures 2b and 4). Varnish ultra-thin sections were then photographed under a Leica DMLB polarized light microscope, equipped with a Ximea-brand xiD digital camera. The color pictures obtained provide high-resolution (~1 µm) images of varnish microstratigraphy for layering pattern analysis. Over hundreds of varnish microstratigraphies from 53 ultra-thin sections (each hosts multiple varnish-filled microbasins) were examined to generalize a regionally replicable Holocene layering sequence. Varnish ultra-thin sections displaying unambiguous and reproducible layering patterns were selected for further microprobe analysis. Chemical depth profiling of varnish microstratigraphy in ultra-thin sections was undertaken on a fully automated five-spectrometer CAMECA SX100 electron probe, with the use of quantitative mode of wavelength dispersive X-ray spectrometry. The use of a focused probe beam ensured high-resolution (~2 µm) detection of Mn, Ba, and other elemental fluctuations in varnish microstratigraphy. Both microscopic images and microprobe chemical data were used to assist in identification and interpretation of varnish layering patterns.

Results

Optical examinations of ultra-thin sections revealed stratigraphic layering in varnish samples from the Lake Turkana basin. It comprises of black and orange/yellow layers similar to the ones observed in the western US varnish (Liu and Broecker, 2007, 2008) (Figure 4). Varnish microstratigraphies that contain similar layering sequences were observed in samples from the same location (Figure 4e and f) and from different locations (Figure 4b, c, g, and h), yielding convincing evidence that the formation of microlaminations in varnish microstratigraphy reflects environmental fluctuations in the study area. Microprobe chemical analyses along depth profiles indicate a systematic association of Mn, Ba, and to a lesser degree, Si with visual microlaminae (Figure 4). Black layers are generally enriched in Mn (15–40%) and Ba (0.5–4%) but depleted in Si (20–25%), while yellow/orange layers are depleted in Mn (<15–20%) and Ba (<0.5–1%) but enriched in Si (25–50%) (note that, unless specified otherwise, elemental concentrations in this paper are reported in oxide wt.%). The concentrations of Al, Fe, Ca, and P fluctuate at the levels of 10–25%, 5–15%, 1–7%, and 0.5–6%, respectively, and do not display any systematic association between black or orange/yellow layers (Figure 4h). These data show that visual varnish microstratigraphy in the Turkana basin mainly reflects elemental fluctuations of Mn and Ba contents, thus indicative of wetness variations over time (Liu and Broecker, 2007, 2008, 2013). On the other hand, the elemental concentration levels of Mn and Ba in varnish layers may vary from sample to sample (Figure 4), likely reflecting local variations of source material and differential enhancement or post-depositional chemical leaching of these elements in varnish microbasins (Liu, 2003).

Figure 4.

Composite microscopic images of regionally replicable microstratigraphies in varnish samples from the latest Pleistocene geomorphic surfaces, Holocene high stands, and recessional shorelines of Lake Turkana in the study area (see Figure 5 and the text for terminology and layering pattern interpretation). The light blue and green curves depict the microprobe depth-profiles of Mn and Ba concentrations, respectively, in the varnish microstratigraphies. Also shown in Figure 4h are the microprobe depth-profiles of elemental concentrations for Al, Si, Fe, Ca, and P. Mn- and Ba-rich dark layers represent periods of wet climate during the early to middle Holocene, while Mn- and Ba-poor orange/yellow layers represent periods of the late-Holocene dry climate. The rapid transition between the two climatic regimes at ~4.3 ka is visually striking and marked by a yellow layer (WH5+) in the layering sequence. Yellow layers (WH8+ and WH6+) in Figure 4a, d represent two short-lived periods of extremely dry climate at ca. 8.5–8 and 5 ka, respectively, coeval with the timing of large water level drops (>50 m) in the Turkana basin; and the latter also marks the AHP termination in the region (Forman et al., 2014; Garcin et al., 2012). Note that dark layers (WH11–WH6) in some fine-grained and fast-growing varnish samples (Figure 4a, b, d, and g) contain two to five minor dark bands, indicative of centennial-scale wet events. Also note that post-depositional erosion often leads to an unconformity in varnish microstratigraphy (Figure 4c) and post-depositional chemical leaching turns a dark layer into an Mn- and Ba-poor yellow layer (Figure 4g and h). Numbers in parentheses next to sample labels are elevations of the sampling spots. LU = layering unit; YD = Younger Dryas; WH = wet period in Holocene.

Layering sequences observed in varnish microstratigraphy provide a regionally replicable record of Holocene wetness variations in the study area. The aggregate Holocene varnish layering sequence (Figure 5) contains eleven dark layers intercalated with eleven orange/yellow layers. Six approximately evenly-spaced, Mn- and Ba-rich (i.e. 25–40% Mn and 1–4% Ba) dark layers (WH11–WH6) appear in the lower to middle portion of the layering sequence, representing six wet pulses during the generally wet early to middle Holocene. Five approximately evenly-spaced dark layers (WH5–WH1) that are less enriched in Mn (15–20%) and Ba (0.5–1%) occur in the upper portion of the layering sequence, diagnostic of relatively weak wet phases in the generally dry late Holocene. Within the darker (wetter) early to middle Holocene interval, several orange/yellow layers containing <15–20% Mn and <0.5–1% Ba indicate the occurrence of short-lived, centennial-scale dry phases punctuating the AHP. Among them, orange/yellow layers denoted as WH10+, WH8+, WH6+ and WH5+ are visually striking in most of the layering sequences examined in this study, representing severe regional dry spells during the AHP (Figures 4 and 5).

Figure 5.

Holocene millennial-scale wetness variations recorded in microstratigraphy of rock varnish from the Lake Turkana basin and their comparison with Holocene wet periods, especially the AHP seen in the reconstructed Holocene water level, leaf wax δD, and fossil diatom records of Lakes Turkana, Abhe and Tanganyika in East Africa (Butzer, 1980; Forman et al., 2014; Garcin et al., 2012; Gasse, 2000; Halfman et al., 1992; Khalidi et al., 2020; Owen et al., 1982; Tierney et al., 2008). Mn- and Ba-rich dark/gray layers in the generalized varnish layering sequence represent periods of relatively wet climate, and Mn- and Ba-poor orange/yellow layers represent periods of relatively dry climate (Liu and Broecker, 2013). Age assignments are based on preliminary radiometric age calibration of basal layers in varnish microstratigraphy from the radiocarbon-dated high stands of Lake Turkana (Forman et al., 2014; Garcin et al., 2012) and archaeological features on the shoreline surfaces at Lothagam (Hildebrand and Grillo, 2012). The vertical orange dashed lines depict the highest stands reached during lake phases 1 and 2 and the late Holocene after ~5 ka based on this study. The red dots denote the elevations and ages of the varnish samples collected in this study with respect to the reconstructed Holocene water level curves of Lake Turkana. The horizontal yellow bands represent periods of short-lived arid phases inferred from these reconstructed lake level curves in East Africa that likely correlate in time with those dry periods signified by narrow orange/yellow layers (WH10+ to WH5+) in the varnish wetness record. The radiocarbon age scale of the diatom curve is anchored at ~4.3 ka on the calendar age scale of this figure for the rapid decrease in the relative abundance of freshwater genera (Melosira) in the lake sediments (Halfman et al., 1992; Johnson and Malala, 2009). The lake curves of Butzer (1980) and Owen et al. (1982) with calendar age scales were adapted from Bloszies et al. (2015). Note that the overall timing and duration of the AHP seen in the varnish wetness record largely align with those seen in the hydroclimate records from both Lake Abhe and Lake Tanganyika. Also note that four centennial-scale dry phases in the varnish record (i.e. those represented by orange/yellow layers WH7+ to WH10+) are not seen in the Lake Tanganyika record, likely reflecting region-specific climate variations during the AHP, but all the records show a lengthy period of dry climate interrupted by several short-lived moist pulses after the AHP. LU = layering unit; YD = Younger Dryas; WH = wet period in Holocene.

Radiocarbon-dated lake sediments and archaeological features from previous studies in the Turkana basin provide preliminary age constraints on the varnish wetness record. Garcin et al. (2012) documented that, between ca. 8.5 and 8 ka in South Island of the Turkana basin, the water level of the lake dropped below 410–360 m from its MHS. Three varnish samples (LT-4, 5, and 6) on shoreline boulders and basalt bedrock between 451 and 455 m from Lothagam display similar layering sequences of LU-1 (WH8+) with a basal yellow layer (WH8+) (Figure 4d–f). Assuming a lag time of 100 years or less for varnish accretion on subaerially exposed stable rock surfaces in deserts (Liu, 2018), the deposition of WH8+ on those sampled rocks should have occurred shortly after ca. 8.5–8 ka. Two archaeological features, namely Lothagam West and North Pillar Stones, are found on the shoreline surfaces at 445 and 437 m along the west and east flanks of the Lothagam hills, respectively (Figures 1c, 2d, and e). Radiocarbon dating of charcoal and ostrich eggshell beads by Hildebrand and Grillo (2012) yielded the oldest possible age of 5–4.8 ka for the construction of the megalithic pillars, which postdates the lake regression from these elevations during the AHP termination. Two varnish samples, one from the Lothagam West Pillar site (LT-7) (Figure 2e) and the other (LT-8) from a 1.5 m high strath terrace next to an ephemeral stream channel at 405 m nearby our field camp site (Figure 2f), display similar layering sequences of LU-1 (WH6+) with a basal yellow layer (WH6+) (Figure 4g and h). Again given a lag time of 100 years or less, the deposition of WH6+ should have taken place at or slightly before ~5 ka when the sampled rocks became subaerially exposed. Studies by Garcin et al. (2012) and Beck et al. (2019) further indicate that Lake Turkana transgressed to the MHS from its YD water level at 410–445 m around ~11.5 ka. This implies that hillslope deposits in the Turkana basin that are truncated by the lake’s MHS at 455–460 m must have been emplaced there at or before ~11.5 ka. Two varnish samples on hillslope deposits at 461 m from Lothagam (LT-3) and Koobi Fora (LT-2) on the east side of the lake (Figures 1 and 2c) display similar layering sequences of LU-1 (WH11) with a basal dark layer (WH11), suggesting the initial deposition of WH11 likely around ~11.5 ka (Figure 4b and c). The oldest varnish sample (LT-1) in this study was collected from an old desert pavement surface at an elevation of 500 m, about 7 km south of Turkwel River (or 28 km northwest of Lothagam) (Figures 1, 2b, and c). The varnish microstratigraphy in this sample contains a complete Holocene layering sequence of dark layers WH11–WH1 underlain by a basal orange layer (WH11+) (Figure 4a), indicative of a dry phase immediately before the onset of the Holocene that is most likely contemporaneous with the YD arid period in the region (Bloszies et al., 2015; Foerster et al., 2012; Garcin et al., 2012; Junginger et al., 2014). Figure 5 depicts the age calibration of the aggregate Holocene layering sequence and compares the calibrated varnish wetness record with paleoclimate proxy records from previously reconstructed shoreline fluctuations as well as leaf wax δD and diatoms of Lakes Turkana, Abhe, and Tanganyika in East Africa (Butzer, 1980; Forman et al., 2014; Garcin et al., 2012; Gasse, 2000; Khalidi et al., 2020; Owen et al., 1982; Tierney et al., 2008).

Discussion

The varnish wetness record obtained from this study is closely tied to the reconstructed shoreline fluctuations of Lake Turkana (Figure 5). Based on our preliminary age calibration of the varnish wetness record, dark layers (WH11–WH9) represent three major wet phases during the early Holocene (ca. 11.5–8.5 ka) when the lake attained its MHS at 455–460 m (Butzer, 1980; Garcin et al., 2012; Owen et al., 1982) (Figure 5). Dark layers (WH8 and WH7) represent two major wet phases during the early to middle Holocene (ca. 8–5 ka) when the lake rose to its secondary high stand at 445 m after a >23 m drop from the MHS between ca. 8.5 and 8 ka (Forman et al., 2014; Garcin et al., 2012; Owen et al., 1982) (Figures 3 and 5). Collectively, these five wet phases constitute an extended wet interval coincident with the AHP in the region (Garcin et al., 2012; Junginger et al., 2014; Van der Lubbe et al., 2017). Dark layer (WH6), although rich in Mn and Ba as those other dark layers (WH11–WH7) (Figure 4), represents an intermediate wet phase associated with a much lower water level (<405 m) of Lake Turkana following the termination of the AHP around ~5 ka. Dark layers (WH5–WH1) that are less enriched in Mn and Ba signify five minor wet phases during the overall arid period of the late Holocene when the lake level oscillated below 405 m and most likely around 360 ± 30 m (Bloszies et al., 2015; Forman et al., 2014; Garcin et al., 2012; Morrissey and Scholz, 2014). The rapid transition from an early to middle Holocene humid period to a late-Holocene arid period is marked by a narrow yellow layer (WH5+) in the varnish layering sequence (Figures 4 and 5). This post-5 ka yellow layer is Mn- and Ba-poor and thus diagnostic of an extremely dry phase that most likely took place around ~4.3 ka when Lake Turkana experienced a rapid fresh-to-saline water transition (Halfman et al., 1992; Johnson and Malala, 2009) (Figure 5) during the 4.2 ka cooling event in the North Atlantic (Bond et al., 1997). Similarly, Mn- and Ba-poor narrow yellow layers (WH8+ and WH6+) in the varnish layering sequence are indicative of two short-lived dry episodes coeval with the two extremely low water levels of Lake Turkana at ca. 8.5–8 ka and 5 ka (Forman et al., 2014; Garcin et al., 2012). The dry phase at ca. 8.5–8 ka is broadly contemporaneous with the 8.2 ka cooling event in the North Atlantic (Bloszies et al., 2015; Bond et al., 1997; Garcin et al., 2012). This 8.2 ka dry phase has also been registered elsewhere in northeastern Africa and the Sahara (Gatto and Zerboni, 2015, and references therein). Moreover, Mn- and Ba-poor narrow orange layers (WH10+, WH9+, and WH7+) document the occurrence of three short-lived dry spells. We presume these orange layers formed during periods of low water levels at ca. 10.4, 9.4, and 6.6 ka, respectively, inferred from those previously reconstructed lake level curves in East Africa (Butzer, 1980; Forman et al., 2014; Garcin et al., 2012; Gasse, 2000; Owen et al., 1982) (Figure 5). However, additional radiometric age calibration of the varnish layering sequence is needed to confirm this speculation.

The varnish wetness record sheds new light on the termination of the AHP in the Turkana basin. Stranded basalt gravel shorelines between 460 and 435 m at Lothagam recorded two major lake-level rises corresponding to two lengthy wet periods represented by dark layers (WH11–WH9) (lake phase 1) and by dark layers (WH8–WH7) (lake phase 2) (Figures 3 –5). Lake phase 2 had a slightly reduced water level at 445 m, about 15 m lower than the MHS of lake phase 1, suggesting that the AHP continued in the Turkana basin until ~5 ka. However, the termination of the AHP appears to have been quite rapid around ~5 ka, as evidenced by a large drop in water level of more than 40 m (i.e. from 445 to <405 m) within a few 100 years (Garcin et al., 2012; Nutz and Schuster, 2016) (Figures 3 and 5). The abrupt AHP termination at ~5 ka in terms of lake level drop reflects a dramatic reduction in monsoonal rainfall that is generally attributed to a nonlinear response of the African monsoon system to the slowly evolving insolation forcing, consistent with a nearly synchronous termination of the AHP around 4.9 ± 0.2 ka in northwest Africa and 4.96 ± 0.07 ka in northeast Africa (McGee et al., 2013; Tierney and DeMenocal, 2013). On the other hand, the moisture condition in the Turkana basin shifted from the prevalent humid regime to a present-day arid regime only after 4.3 ka, as indicated by the change from a set of Mn- and Ba-rich dark layers (WH11–WH6) to a set of dark layers (WH5–WH1) that are less enriched in Mn and Ba in the varnish layering sequence (Figures 4 and 5). A fossil diatom record from Lake Turkana (Halfman et al., 1992) shows a large drop in relative abundance of freshwater diatom genera and an increase of brackish-tolerant taxa at 4.3 ka, in accord with the varnish wetness record (Figure 5). Such a shift in regional moisture condition (especially relative humidity) lagged for about 700 years after the >40 m lake level drop in the Turkana basin, hinting at a gradual phasing out of the AHP regime. The gradual phasing out likely reflects the deferred responses of regional vegetation shift and reduction in vegetation cover and soil moisture to the insolation forcing-induced southward retreat of the African monsoonal rain belt, in line with a time-transgressive termination of the AHP around 3–4 ka at the 3–6.5° N latitudes (DeMenocal, 2015; Shanahan et al., 2015). Clearly, further research in regions with distinct rainfall regimes (i.e. the northwest Sahara, eastern Sahara and southern Sahara/Sahel; cf. Dallmeyer et al., 2020) is needed to see if the above two seemingly contradicting ways of the AHP termination are registered in the varnish wetness records from the drylands of North Africa. It is noteworthy that, in Koobi Fora, Ashley et al. (2017) documented a lake level rise of about 45 m above the current water surface around 4.7–4.8 ka based on radiocarbon dating of charcoal from lake sediments, coincident with both the wet period represented by dark layer (WH6) in the varnish record and the diatom record (Figure 5). In central Cameroon, Vincens et al. (2010) reported pollen sequences from Lake Mbalang showing an episode of slight forest regeneration between 5.2 and 4.2 ka in the Adamawa Plateau. Taken together, these data yield convincing evidence for a possible regional wet pulse with much reduced rainfall around 4.7–4.8 ka in northern tropical Africa.

The varnish wetness record further helps constrain the Holocene water level fluctuations of Lake Turkana. Previous studies (Butzer, 1980; Forman et al., 2014; Garcin et al., 2012; Owen et al., 1982) demonstrated that Lake Turkana retained less variable water levels around 455–460 m during lake phase 1, but experienced several large fluctuations around or below 445 m during lake phase 2 (Figure 5). The lake also likely reached its MHS of 455–460 m at ca. 7 and 4 ka (Brown and Fuller, 2008; Butzer, 1980; Owen et al., 1982) (Figure 5) or at ca. 6.5 and 5 ka (Bloszies et al., 2015). Our varnish data generally corroborate these observations, but with exceptions for those transgressions above 445 m since lake phase 2. Varnish samples (LT-4 to LT-6) from the 455 to 451 m high stand contain three dark layers (WH8, WH7, and WH6), while those (LT-7 and LT-8) from the 445 m high stand or below contain only one dark layer (WH6) (Figures 4 and 5). These data indicate that, during lake phase 2 and the time thereafter, Lake Turkana has not transgressed to the 455–451 m high stand. Otherwise, varnish samples from the 455 to 451 m shorelines at Lothagam would have been submerged in water during such lake transgression, which would reset the varnish clock, such that it would display younger layering sequences similar to those in samples from the 445 m or lower shorelines. These data further suggest that Lake Turkana intermittently overflowed into the Nile drainage basin through its sill at 455–460 m during lake phase I and became a closed-basin lake thereafter for the past ~8 ky. Such an attestation is in disagreement with claims from some earlier studies (Brown and Fuller, 2008; Butzer, 1980; Harvey and Grove, 1982; Johnson and Malala, 2009; Owen et al., 1982), but is consistent with recent observations by others (Forman et al., 2014; Garcin et al., 2012) and also backed up by the geomorphic evidence from this study (Figure 3).

Conclusion and implication

Rock varnish in the Lake Turkana basin recorded millennial-scale wet phases during the early to middle Holocene that are closely tied to an overall high stand stage of the lake during the AHP. Three major wet phases occurred between 11.5 and 8.5 ka when the lake attained its MHS at 455–460 m, and two major wet phases appeared between 8 and 5 ka when the lake reached its secondary high stand at 445 m. One moderate wet phase took place between 5 and 4.3 ka after the lake level dropped from 445 to <405 m during the climatic transition from the AHP mode to the present-day arid mode. The varnish wetness record indicates that the AHP terminated rapidly at ~5 ka in the Turkana basin in terms of lake level drop, or phased out gradually from 5 to 4.3 ka in terms of regional moisture condition. These findings have important paleoclimatic and geochronological implications. Since the AHP-related wetness variations in the varnish record are mainly modulated by the African monsoons and associated precipitation events, it may be feasible now to map the spatiotemporal distribution of the northward incursion of the early to middle Holocene African monsoonal moisture over the present-day Sahara Desert and the Sinai/Arabian Peninsulas (e.g. Enzel et al., 2015; Fleitmann et al., 2007; Pachur and Hoelzmann, 1991; Ritchie et al., 1985; Tierney et al., 2017) with varnish microstratigraphy. Because climatic signals archived in varnish are regionally synchronous on a millennial to centennial time scale, once fully calibrated by connecting more dated geomorphic and archeological features to the layering patterns in the future, the Holocene varnish layering sequence documented in this study may also be used as a correlative age determination tool to date surface geoarchaeological features such as petroglyphs and stone artifacts in the Lake Turkana basin and the surrounding regions (cf. Cremaschi, 1996) where other conventional radiometric means (e.g. AMS radiocarbon and cosmogenic surface exposure dating) are difficult to apply.

Footnotes

Acknowledgements

We would like to thank the National Commission for Science Technology and Innovation and the government of Kenya for issuing permits NACOSTI/P/15/0767/6515 and NCST/RRI/12/1/BS011/54 and allowing us to conduct the research. We also thank the Turkana Basin Institute for logistic support during our fieldwork and Adrian Fiege at the American Museum of Natural History in New York City for assistance in microprobe analyses. We are grateful to the associate editor Sandra Passchier for handling of the manuscript and Mathieu Schuster and Andrea Zerboni for their critical reviews and constructive suggestions, which greatly improved the quality of the manuscript. This is Lamont Doherty Earth Observatory contribution number 8496.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by a Comer Science and Education Foundation grant (PG006397).

ORCID iDs

Tanzhuo Liu

Christopher J Lepre

References

Armitage

Bristow

Drake

(2015) West African monsoon dynamics inferred from abrupt fluctuations of Lake Mega-Chad. Proceedings of the National Academy of Sciences 112: 8543–8548.

Ashley

Ndiema

Spencer

JQG

, et al. (2017) Paleoenvironmental reconstruction of Dongodien, Lake Turkana, Kenya and OSL dating of site occupation during late Holocene climate change. African Archaeological Review 34: 345–362.

Avery

(2010) Hydrological impacts of Ethiopia’s Omo Basin on Kenya’s Lake Turkana water levels and fisheries. Final report, African Development Bank Group, Tunis.

Baied

Somonte

(2013) Mid-Holocene geochronology, palaeoenvironments, and occupational dynamics at Quebrada de Amaicha, Tucumán, Argentina. Quaternary International 299: 80–89.

Beck

Feibel

Wright

, et al. (2019) Onset of the African Humid Period by 13.9 kyr BP at Kabua Gorge, Turkana Basin, Kenya. The Holocene 29: 1011–1019.

Berke

Johnson

Werne

, et al. (2012) A mid-Holocene thermal maximum at the end of the African Humid Period. Earth and Planetary Science Letters 351–352: 95–104.

Beyin

(2011) Recent archaeological survey and excavation around the Greater Kalokol Area, west side of Lake Turkana: Preliminary findings. Nyame Akuma 75: 40–50.

Bloszies

Forman

Wright

(2015) 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: 64–76.

Bond

Showers

Cheseby

, et al. (1997) A pervasive millennial-scale cycle in North Atlantic Holocene and glacial climates. Science 278: 1257–1266.

10.

Braconnot

Joussaume

Marti

, et al. (1999) Synergistic feedbacks from ocean and vegetation on the African monsoon response to mid-Holocene insolation. Geophysical Research Letters 26: 2481–2484.

11.

Broecker

Liu

(2001) Rock varnish: Recorder of desert wetness? GSA Today 11: 4–10.

12.

Brown

Fuller

(2008) Stratigraphy and tephra of the Kibish Formation, southwestern Ethiopia. Journal of Human Evolution 55(3): 366–403.

13.

Butzer

(1980) The Holocene lake plain of North Rudolph, East Africa. Physical Geography 1: 42–58.

14.

Butzer

Isaac

Richardson

, et al. (1972) Radiocarbon dating of East African Lake levels. Science 175: 1069–1076.

15.

Claussen

Gayler

(1997) The greening of the Sahara during the mid-Holocene: Results of an interactive atmosphere-biome model. Global Ecology and Biogeography Letters 6: 369–377.

16.

Collins

Prange

Caley

, et al. (2017) Rapid termination of the African Humid Period triggered by northern high-latitude cooling. Nature Communications 8: 1372.

17.

Costa

Russell

Konecky

, et al. (2014) Isotopic reconstruction of the African Humid Period and Congo Air Boundary migration at Lake Tana, Ethiopia. Quaternary Science Reviews 83: 58–67.

18.

Cremaschi

(1996) The rock varnish in the Messak Settafet (Fezzan, Libyan Sahara), age, archaeological context, and paleo-environmental implication. Geoarchaeology 11: 393–421.

19.

Dallmeyer

Claussen

Lorenz

, et al. (2020) The end of the African humid period as seen by a transient comprehensive Earth system model simulation of the last 8000 years. Climate of the Past 16: 117–140.

20.

DeMenocal

Ortiz

Guilderson

, et al. (2000) Abrupt onset and termination of the African Humid Period: Rapid climate responses to gradual insolation forcing. Quaternary Science Reviews 19: 347–361.

21.

DeMenocal

(2015) End of the African Humid Period. Nature Geoscience 8: 86–87.

22.

DeMenocal

Tierney

(2012) Green Sahara: African humid periods paced by Earth’s orbital changes. Nature Education Knowledge 3: 12.

23.

Dorn

(2009) Rock varnish and its use to study climatic change in geomorphic settings. In: Parsons

Abrahams

(eds) Geomorphology of Desert Environments. Dordrecht: Springer, pp.657–673.

24.

Dorn

Oberlander

(1981) Rock varnish origin, characteristics and usage. Zeitschrift fur Geomorphologie 25: 420–436.

25.

Enzel

Kushnir

Quade

(2015) The middle Holocene climatic records from Arabia: Reassessing lacustrine environments, shift of ITCZ in Arabian Sea, and impacts of the southwest Indian and African monsoons. Global and Planetary Change 129: 69–91.

26.

Ferguson

AJD

Harbott

(1982) Geographical, physical, and chemical aspects of Lake Turkana. In: Hopson

(ed.) Lake Turkana: A Report of the Findings of the Lake Turkana Project, 1972–1975. London: Overseas Development Administration, pp.1–110.

27.

Fleitmann

Burns

Mangini

, et al. (2007) Holocene ITCZ and Indian monsoon dynamics recorded in stalagmites from Oman and Yemen (Socotra). Quaternary Science Reviews 26: 170–188.

28.

Foerster

Junginger

Langkamp

, et al. (2012) Climatic change recorded in the sediments of the Chew Bahir basin, southern Ethiopia, during the last 45,000 years. Quaternary International 274: 25–37.

29.

Forman

Wright

Bloszies

(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. Quaternary Science Reviews 97: 84–101.

30.

Garcin

Melnick

Strecker

, et al. (2012) East African mid-Holocene wet–dry transition recorded in palaeo-shorelines of Lake Turkana, northern Kenya Rift. Earth and Planetary Science Letters 331–332: 322–334.

31.

Gasse

(2000) Hydrological changes in the African tropics since the Last Glacial Maximum. Quaternary Science Reviews 19: 189–211.

32.

Gatto

Zerboni

(2015) Holocene supra-regional environmental changes as trigger for major socio-cultural processes in northeastern Africa and the Sahara. African Archaeological Review 32: 301–333.

33.

Goldsmith

(2011) Characterizing rock varnish developed on earliest Holocene Negev flint artifacts as a potential paleoenvironmental or paleoclimatic indicator. Jerusalem, Israel: Geological Survey of Israel.

34.

Halfman

Jacobson

Cannella

, et al. (1992) Fossil diatoms and the mid to late Holocene paleolimnology of Lake Turkana, Kenya: A reconnaissance study. Journal of Paleolimnology 7: 23–35.

35.

Halfman

Johnson

Finney

(1994) New AMS dates, stratigraphic correlations and decadal climatic cycles for the past 4 ka at Lake Turkana, Kenya. Palaeogeography Palaeoclimatology Palaeoecology 111: 83–98.

36.

Harvey

CPD

Grove

(1982) A prehistoric source of the Nile. Geographical Journal 148: 327–336.

37.

Hildebrand

Grillo

(2012) Early herders and monumental sites in eastern Africa: Dating and interpretation. Antiquity 86: 338–352.

38.

Jaeschke

Thienemann

Schefuß

, et al. (2020) Holocene hydroclimate variability and vegetation response in the Ethiopian Highlands (Lake Dendi). Frontiers in Earth Science 8: 585770.

39.

Johnson

Malala

(2009) Lake Turkana and its link to the Nile. In: Dumont

(ed.) The Nile: Origin, Environments, Limnology and Human Use. Monographiae Biologicae, Vol. 89. Dordrecht: Springer, pp.287–304.

40.

Junginger

Roller

Olaka

, et al. (2014) The effects of solar irradiation changes on the migration of the Congo Air Boundary and water levels of paleo-Lake Suguta, Northern Kenya Rift, during the African Humid Period (15–5ka BP). Palaeogeography Palaeoclimatology Palaeoecology 396: 1–16.

41.

Khalidi

Mologni

Ménard

, et al. (2020) 9000 years of human lakeside adaptation in the Ethiopian Afar: Fisher-foragers and the first pastoralists in the Lake Abhe basin during the African Humid Period. Quaternary Science Reviews 243: 106459.

42.

Kröpelin

Verschuren

Lézine

, et al. (2008) Climate-driven ecosystem succession in the Sahara: The past 6000 years. Science 320: 765–768.

43.

Kutzbach

Liu

(1997) Response of the African monsoon to orbital forcing and ocean feedbacks in the middle Holocene. Science 278: 440–443.

44.

Kutzbach

Otto-Bliesner

(1982) The sensitivity of the African-Asian monsoonal climate to orbital parameter changes for 9000 years B.P. in a low-resolution general circulation model. Journal of the Atmospheric Sciences 39: 1177–1188.

45.

Lee

Bland

(2003) Dating climatic change in hot deserts using desert varnish on meteorite finds. Earth and Planetary Science Letters 206: 187–198.

46.

Liu

(2003) Blind testing of rock varnish microstratigraphy as a chronometric indicator: Results on late Quaternary lava flows in the Mojave Desert California. Geomorphology 53: 209–234.

47.

Liu

(2018) Dating rock and desert varnish. In: López Varela SL (ed.) The Encyclopedia of Archaeological Sciences. Hoboken, NJ: John Wiley & Sons, Inc., pp.1–7.

48.

Liu

Broecker

Bell

, et al. (2000) Terminal Pleistocene wet event recorded in rock varnish from Las Vegas Valley, southern Nevada. Palaeogeography Palaeoclimatology Palaeoecology 161: 423–433.

49.

Liu

Broecker

Stein

(2013) Rock varnish evidence for a Younger Dryas wet period in the Dead Sea basin. Geophysical Research Letters 40: 2229–2235.

50.

Liu

Broecker

(2000) How fast does rock varnish grow? Geology 28: 183–186.

51.

Liu

Broecker

(2007) Holocene rock varnish microstratigraphy and its chronometric application in the drylands of western USA. Geomorphology 84: 1–21.

52.

Liu

Broecker

(2008) Rock varnish evidence for latest Pleistocene millennial-scale wet events in the drylands of western United States. Geology 36: 403–406.

53.

Liu

Broecker

(2013) Millennial-scale varnish microlamination dating of late Pleistocene geomorphic features in the drylands of western USA. Geomorphology 187: 38–60.

54.

Liu

Dorn

(1996) Understanding spatial variability in environmental changes in drylands with rock varnish microlaminations. Annals of the Association of American Geographers 86: 187–212.

55.

McGee

DeMenocal

Winckler

, et al. (2013) The magnitude, timing and abruptness of changes in North African dust deposition over the last 20,000 yr. Earth and Planetary Science Letters 371–372: 163–176.

56.

Morrissey

Scholz

(2014) Paleohydrology of Lake Turkana and its influence on the Nile River system. Palaeogeography Palaeoclimatology Palaeoecology 403: 88–100.

57.

Nicholson

Kim

Hoopingarner

(1988) Atlas of African Rainfall and its Interannual Variability. Tallahassee, FL: Department of Meteorology, Florida State University, p.237.

58.

Nutz

Schuster

(2016) Stepwise drying of Lake Turkana at the end of the African Humid Period: A forced regression modulated by solar activity variations? Solid Earth 7: 1609–1618.

59.

Owen

Barthelme

Renaut

, et al. (1982) Palaeolimnology and archaeology of Holocene deposits north-east of Lake Turkana, Kenya. Nature 298: 523–529.

60.

Pachur

Hoelzmann

(1991) Paleoclimatic implications of late quaternary lacustrine sediments in Western Nubia, Sudan. Quaternary Research 36: 257–276.

61.

Perry

Adams

(1978) Desert varnish: Evidence for cyclic deposition of manganese. Nature 276: 489–491.

62.

Ritchie

Eyles

Haynes

(1985) Sediment and pollen evidence for an early to mid-Holocene humid period in the eastern Sahara. Nature 314: 352–355.

63.

Robbins

(1969) Lothagam: A later stone age fishing settlement in the Lake Rudolf area of the Kenya Rift System. PhD Thesis, University of California, Berkeley, CA, p.63.

64.

Robbins

(1972) Archeology in the Turkana District, Kenya. Science 176: 359–366.

65.

Robbins

Lynch

(1978) New evidence on the use of microliths from the Lake Turkana Basin, East Africa. Current Anthropology 19: 619–620.

66.

Shanahan

McKay

Hughen

, et al. (2015) The time-transgressive termination of the African Humid Period. Nature Geoscience 8: 140–144.

67.

Tierney

DeMenocal

(2013) Abrupt shifts in Horn of Africa hydroclimate since the Last Glacial Maximum. Science 342: 843–846.

68.

Tierney

Pausata

FSR

DeMenocal

(2017) Rainfall regimes of the Green Sahara. Science Advances 3: e1601503.

69.

Tierney

Russell

Huang

, et al. (2008) Northern Hemisphere controls on tropical Southeast African climate during the past 60,000 years. Science 322: 252–255.

70.

Van der Lubbe

Krause-Nehring

Junginger

, et al. (2017) Gradual or abrupt? Changes in water source of Lake Turkana (Kenya) during the African Humid Period inferred from Sr isotope ratios. Quaternary Science Reviews 174: 1–12.

71.

Vincens

Buchet

Servant

(2010) Vegetation response to the "African Humid Period" termination in Central Cameroon (7° N) – New pollen insight from Lake Mbalang. Climate of the Past 6: 281–294.

72.

Wright

Forman

(2011) Holocene occupation of the Mount Porr strand plain in southern Lake Turkana, Kenya. Nyame Akuma 76: 47–62.

73.

Zerboni

(2008) Holocene rock varnish on the Messak plateau (Libyan Sahara): Chronology of weathering processes. Geomorphology 102: 640–651.