Distal tephras of the eastern Lake Victoria basin, equatorial East Africa: Correlations, chronology and a context for early modern humans

Abstract

The tephrostratigraphic framework for Pliocene and Early Pleistocene paleoanthropological sites in East Africa has been well established through nearly 50 years of research, but a similarly comprehensive framework is lacking for the Middle and particularly the Late Pleistocene. We provide the first detailed regional record of Late Pleistocene tephra deposits associated with artifacts or fossils from the Lake Victoria basin of western Kenya. Correlations of Late Pleistocene distal tephra deposits from the Wasiriya beds on Rusinga Island, the Waware beds on Mfangano Island and deposits near Karungu, mainland Kenya, are based on field stratigraphy coupled with 916 electron microprobe analyses of eleven major and minor element oxides from 50 samples. At least eight distinct distal tephra deposits are distinguished, four of which are found at multiple localities spanning >60 km over an approximately north to south transect. New optically stimulated luminescence dates help to constrain the Late Pleistocene depositional ages of these deposits. Our correlation and characterization of volcaniclastic deposits expand and refine the current stratigraphy of the eastern Lake Victoria basin. This provides the basis for relating fossil- and artifact-bearing sediments and a framework for ongoing geological, archaeological and paleontological studies of Late Pleistocene East Africa, a crucial time period for human evolution and dispersal within and out of Africa.

Full text

Distal tephras of the eastern Lake Victoria basin, equatorial East Africa:
correlations, chronology and a context for early modern humans
Nick Blegen
a
,
*
, Christian A. Tryon
b
, J. Tyler Faith
c
, Daniel J. Peppe
d
, Emily J. Beverly
d
,
Bo Li
e
, Zenobia Jacobs
e
a
Department of Anthropology, University of Connecticut, Storrs, CT 06269, USA
b
Department of Anthropology, Harvard University, Peabody Museum, 11 Divinity Ave, Cambridge, MA 02138, USA
c
School of Social Science, University of Queensland, Brisbane, QLD 4072, Australia
d
Terrestrial Paleoclimatology Research Group, Department of Geology, Baylor University, Waco, TX 76798, USA
e
Centre for Archaeological Science, School of Earth &Environmental Sciences, University of Wollongong, Wollongong, NSW 2522, Australia
article info
Article history:
Received 2 October 2014
Received in revised form
29 April 2015
Accepted 30 April 2015
Available online xxx
Keywords:
Tephrostratigraphy
East Africa
Middle Stone Age
Human evolution
abstract
The tephrostratigraphic framework for Pliocene and Early Pleistocene paleoanthropological sites in East
Africa has been well established through nearly 50 years of research, but a similarly comprehensive
framework is lacking for the Middle and particularly the Late Pleistocene. We provide the first detailed
regional record of Late Pleistocene tephra deposits associated with artifacts or fossils from the Lake
Victoria basin of western Kenya. Correlations of Late Pleistocene distal tephra deposits from the Wasiriya
beds on Rusinga Island, the Waware beds on Mfangano Island and deposits near Karungu, mainland
Kenya, are based on field stratigraphy coupled with 916 electron microprobe analyses of eleven major
and minor element oxides from 50 samples. At least eight distinct distal tephra deposits are distin-
guished, four of which are found at multiple localities spanning >60 km over an approximately north to
south transect. New optically stimulated luminescence dates help to constrain the Late Pleistocene
depositional ages of these deposits. Our correlation and characterization of volcaniclastic deposits
expand and refine the current stratigraphy of the eastern Lake Victoria basin. This provides the basis for
relating fossil- and artifact-bearing sediments and a framework for ongoing geological, archaeological
and paleontological studies of Late Pleistocene East Africa, a crucial time period for human evolution and
dispersal within and out of Africa.
©2015 Elsevier Ltd. All rights reserved.
1. Introduction
Fossil evidence suggests the earliest members of our species,
Homo sapiens, first appeared in equatorial East Africa by 195 ka
(McDougall et al., 2005; Brown et al., 2012). This area likely served
as one point of departure for subsequent Late Pleistocene hominin
dispersals across and out of Africa (Rose et al., 2011; Soares et al.,
2012; Rito et al., 2013). The archaeological record and environ-
mental context of these early H. sapiens populations are essential
data for understanding the evolutionary success of our species (e.g.,
Tryon and Faith, 2013). Continental and regional syntheses of the
Pleistocene African archaeological and environmental records are
often characterized by mismatches in temporal and spatial scales
(reviewed in Blome et al., 2012; Tryon and Faith, 2013). Deep cave
sequences in northern and particularly southern Africa have pro-
vided finely resolved, rich archaeological and environmental re-
cords (e.g., Deacon, 1979; Singer and Wymer, 1982; Avery et al.,
1997; Marean et al., 2000; Henshilwood et al., 2002; Jacobs et al.,
2006, 2008; Wadley and Jacobs, 2006; Marean et al., 2007;
Garcea, 2010; Clark-Balzan et al., 2012). However, deeply strati-
fied cave sequences are largely lacking in East Africa, where the
archaeological record consists of generally low-density open-air
sites (Tryon and Faith, 2013). Demonstrating stratigraphic equiva-
lence among these open-air sites via the correlation of tephra
provides the means to assess landscape-scale spatial variation in
past environments and human behaviors.
East Africa has the potential to demonstrate the equivalence
among sites via tephrostratigraphy, the geochemical and lithos-
tratigraphic correlation of tephra as widespread markers in the
geological record (Lowe, 2011; Brown and Nash, 2014). Rifting
along the East African Rift System (EARS; Chorowicz, 2005)
*Corresponding author. Tel.: þ1 847 436 3854.
E-mail addresses: nick.blegen@gmail.com,nicholas.blegen@uconn.edu
(N. Blegen).
Contents lists available at ScienceDirect
Quaternary Science Reviews
journal homepage: www.elsevier.com/locate/quascirev
http://dx.doi.org/10.1016/j.quascirev.2015.04.024
0277-3791/©2015 Elsevier Ltd. All rights reserved.
Quaternary Science Reviews 122 (2015) 89e111

provides the mechanisms for volcanic eruptions, rapid sedimen-
tation and burial of archaeological and paleontological sites, as well
as their subsequent exposure through continued faulting. Although
there is a well-established Pliocene and Early Pleistocene teph-
rostratigraphic framework for paleoanthropological sites in Kenya,
Ethiopia, Tanzania and Uganda that is the outcome of nearly 50
years of research (e.g., Brown et al., 1992, 2006; Feibel, 1999; Hay,
1976; McHenry et al., 2008; Pickford et al., 1991; Woldegabriel
et al., 1999, 2005, 2013), comparatively few data are available for
these areas during the portions of the Middle and Late Pleistocene
that saw the origin and dispersal of H. sapiens (for published ex-
ceptions, see Brown et al., 2012; Morgan and Renne, 2008; Sahle
et al., 2014; Tryon and McBrearty, 2002, 2006; Tryon et al., 2008,
2010).
We address this problem by focusing on archaeological and
paleontological sites associated with tephra deposits in and around
the eastern Lake Victoria basin, and develop a Late Pleistocene
tephrostratigraphic and chronometric framework for the region. At
66,400 km
2
, Lake Victoria is the largest lake in Africa by surface area
(Adams, 1996). The habitats surrounding this lake have undergone
substantial climate-driven changes throughout the Quaternary
(Nicholson, 1998; Bootsma and Hecky, 2003), likely with profound
impacts on human and other animal communities (e.g., Faith et al.,
2011, in press; Tryon et al., 2010, 2012). However, until recently, an
understanding of environmental variation prior to the Last Glacial
Maximum has been poorly constrained, and the nature of spatial
variation in environments and human behavior obscured.
We present the results of 916 electron microprobe analyses to
geochemically characterize and correlate 50 distal tephra deposits
from 32 measured sections across Rusinga Island, Mfangano Island,
and Karungu on the Kenyan mainland (Fig. 1a, b), spanning a
roughly north-south oriented transect over 60 km in the eastern
Lake Victoria basin (eLVB). Four tuffs: the Wakondo Tuff, the Nya-
mita Tuff, the Nyamsingula Tuff, and the Bimodal Trachyphonolitic
Tuff are sufficiently distinct and widespread lithostratigraphic
markers for correlation within and between discontinuous out-
crops at distantly located paleontological and archaeological lo-
calities. Optically stimulated luminescence (OSL),
UraniumeThorium disequilibrium (UeTh) and AMS
14
C dates
constrain the depositional history and ages of these tuffs to >33 to
~100 ka. This stratigraphic sequence of tuffs, radiometric dates, and
intercalated fossil- and artifact-bearing sediments provides the
fundamental framework to assess paleoecological and archaeo-
logical variation across time and space in the eastern portion of the
Lake Victoria basin.
2. Pleistocene distal tephra deposits, fossils, and artifacts in
the eastern Lake Victoria basin
The Lake Victoria basin formed in the depression between the
eastern and western branches of the EARS, probably within the last
few million years (reviewed in Danley et al., 2012). The north-
eastern part of the Lake Victoria basin (Fig. 1c) is unlikely to have
been volcanically active since the cessation of rifting and volcanism
associated with the failed Nyanza Rift in the Early Miocene (e.g.,
Van Couvering, 1972; Peppe et al., 2009). During the Pleistocene,
the eLVB formed a repository for sediments, including volcani-
clastic deposits from eruptions originating from sources outside of
the basin. Tephra input appears to be from multiple sources of the
central and southern Kenyan Rift in the eastern branch of the EARS
(Fig. 1b), demonstrated by tephra deposits from Kenyan and Tan-
zanian volcanoes mapped in areas as far west as ~35

10
0
E (see
Fig. 1b; Dawson, 2008; Peters et al., 2008; Tryon et al., 2010;
Williams, 1991). Pleistocene tephra in the eLVB west of 35

10
0
E
have been previously reported from Rusinga Island (Tryon et al.,
2010; Van Plantinga, 2011; Garret et al., in press) and from sedi-
ment cores near Buvuma Island (Kendall, 1969: 139) within Lake
Victoria, but are poorly documented due to a focus on economic
geology or igneous and metamorphic suites in the region (e.g., Le
Bas, 1977).
Localities with Pleistocene tephra, fossils and artifacts attributed
to the Early Stone Age (ESA), Middle Stone Age (MSA), and Later
Stone Age (LSA) archaeological technocomplexes are known from
the eLVB in Kenya as discussed below (Tryon et al., 2010, 2012,
2014; Faith et al., 2015), as well as from Loiyangalani (HcJd-1) in
the ash-rich Serengeti Plain of the eLVB in Tanzania (Fig. 1b,
Anderson and Talbot, 1965; Bower et al., 1981, 1985; Pickering,
1959; Thompson, 2005). A number of these western Kenyan sites
occur ~80e100 km east of Lake Victoria's Winam Gulf, and are
associated with at least two widespread marker tuffs considered
useful for local field correlation. Pickford (1982, 1984) termed these
the “Nyando Ash”or “Nyando Ashes”(Fig. 1). From the area of
mapped exposures of the “Nyando Ashes,”McBrearty (1981)
excavated lithic artifacts and fossils from within reworked Pleis-
tocene tephra at Songhor near the head of the Nyando River
(Fig. 1b, c). McBrearty (1991, 1992) also described a sequence of
twelve tuffaceous deposits at the site of Simbi in an area mapped as
Nyando Ashes (Fig. 1b, c) and reported a preliminary
40
Ar/
39
Ar age
range of ~50e200 ka (McBrearty, 1992). These preliminary dates
appear to confirm the Pleistocene age of the Nyando Ashes, but the
number of tuffaceous deposits documented at Songhor, Simbi and
Rusinga (McBrearty, 1991, 1992; Tryon et al., 2010; Van Plantinga,
2011; Garret et al., in press) indicates that there are more than
the two tephra deposits originally suggested by Pickford (1982,
1984).
Here we couple the first geochemical investigation of the
“Nyando Ashes”with geological, archeological, and paleontological
data developed during our field program investigating Pleistocene
tephra, fossils, and artifacts along the margins of Lake Victoria on
Rusinga Island, Mfangano Island and Karungu on the Kenyan
mainland (Tryon et al., 2010, 2014, 2014; Van Plantinga, 2011;
Beverly et al., in press; Garrett et al., in press).
3. The Wasiriya beds of Rusinga Island, the Waware beds of
Mfangano Island and the Pleistocene exposures of Karungu
3.1. The Wasiriya beds
The Wasiriya beds are exposed over an area of approximately
10 km
2
on the hill slopes around Rusinga Island (Fig. 2,Tryon et al.,
2010). These sediments were informally named by Pickford (1984,
1986) based on previous mapping and descriptions (Kent, 1942;
Van Couvering, 1972). The first measured sections, sedimentary
lithological descriptions and geochemical characterizations and
correlation of tephras from the Wakondo and Nyamita localities on
Rusinga Island were reported by Tryon et al. (2010). These are
shown in Fig. 2, supplemented by new data from the Nyamsingula
locality.
The Wasiriya beds are exposed in sections that are 15 m at
their thickest points, and are comprised of three primary recog-
nized lithologies: 1) poorly sorted coarse sand and gravel channels
cemented by carbonate representing episodic channel erosion and
deposition, 2) fine grained mudstone, siltstone, and silty sand-
stones preserving evidence of incipient soil development indicating
a slightly more stable landscape, and 3) tephra that has undergone
varying amounts of reworking and incipient pedogenesis (Tryon
et al., 2010, 2012). AMS
14
C dates of gastropod shells primarily
from tuffaceous sediments at the Nyamita 2 and Nyamita 3 local-
ities (Tryon et al., 2010) indicate a minimum age of 33 ka for these
and underlying deposits (Tryon et al., 2010, 2012). These dated
N. Blegen et al. / Quaternary Science Reviews 122 (2015) 89e11190

Fig. 1. Map of relevant eastern African Late Pleistocene archeological sites and major rift volcanoes. A. Outline of continent of Africa with box showing area of outset ‘B’. B. Map of
Lake Victoria region showing archaeological and paleontological localities discussed in the text and major volcanoes active during the Late Pleistocene. The dashed line indicates
approximate mapped westward known extent of Rift Valley tephra. C. Map of approximate area of the eastern Lake Victoria basin (eLVB) referenced in this study. This includes
Rusinga Island, Mfangano Island, and Karungu and other Winam Gulf tephra and archeological sites.

snails (Limicolaria cf. martensiana) only occur in sediments close to
or at the modern surface, and are found in different types of sedi-
ments at the same elevation and position relative to the modern
soil surface. These specimens also primarily occur in life position
suggesting they apparently died during aestivation. Based on these
observations, our interpretation is that the snails burrowed into the
Wasiriya beds following deposition of the sediment, but prior to
lithification, and thus they provide a minimum age for the deposits.
U-series dates of 94.0 ±3.3 ka and 111.4 ±4.2 ka on tufa at the base
of the Wasiriya beds sequence exposed at Nyamita (Fig. 2) provide a
maximum age for the overlying sediments (Beverly et al., in press).
Stone artifacts from the Wasiriya beds include MSA Levallois cores
and Levallois points as well as bifacial and unifacial trimmed points
(Tryon et al., 2010). Fauna are abundant and include both extinct
and extant taxa (Pickford and Thomas, 1984; Tryon et al., 2010;
Faith et al., 2011). Water-dependent taxa are present (e.g., Hippo-
potamus), but the majority of specimens indicate open, semi-arid
grasslands distinct from the evergreen bushlands, woodlands,
and forests historically found in the region (Garrett et al., 2015;
Tryon et al., 2010, 2014).
3.2. The Waware beds
The Waware beds of Mfangano Island, Lake Victoria, Kenya are
relatively poorly exposed over an area of ~7 km
2
on the north-
eastern periphery of the island (Fig. 2;Tryon et al., 2012). Pickford
(1984, 1986) informally named the Waware beds revising previous
work (Whitworth, 1961). These deposits, which are all 5 m thick,
preserve fluvial sediments with incipient soil development and
reworked tuffaceous deposits that unconformably overlie Miocene
sediments (Tryon et al., 2012). The fossil and artifact bearing por-
tions of the Waware beds deposits are comprised of fine-grained
sandy mudstone and siltstone beds with evidence for varying de-
grees of pedogenesis interbedded with occasional coarser grained
sandstone and conglomerate channel deposits (Tryon et al., 2012).
Based on perceived similarity of Lake Victoria's base level at the
times these various beds were deposited, Pickford (1984, 105)
inferred that the Waware beds on Mfangano, the Wasiriya beds on
Rusinga, and the Apoko Formation of the Homa Peninsula were
deposited during the same time period in the Late Pleistocene. The
only radiometric ages for the Pleistocene Waware beds on
Fig. 2. Top Center: Map of Rusinga Island and Mfangano Island showing the extent of Pleistocene outcrops of the Wasiriya beds, the Waware beds as well as archaeological and
paleontological localities discussed in the text (after Tryon et al., 2010, 2012). Below: Stratigraphic columns of measured and sampled sections, arranged west (on left) to east (right).
Lithologies are indicated for all units. Tuffs with electron microprobe determined chemical composition are color-coded to compositional group and labeled with sample number
(ex: CAT09-21). Tuffaceous units not chemically characterized and assigned are shown in grey. Dotted lines represent tuff units that can be traced laterally in the field between two
or more measured sections.
N. Blegen et al. / Quaternary Science Reviews 122 (2015) 89e11192

Mfangano Island are AMS
14
C dates of 35e42 ka on gastropod shells
from the Kakrigu locality (Fig. 2,Tryon et al., 2012). As is the case in
the Wasiriya beds on Rusinga Island, the snails (Limicolaria cf.
martensiana) postdate deposition of the Waware beds, and there-
fore provide minimum age constraints. Based on the similar
radiocarbon ages from the snails and the fact that the lithofacies of
the Waware and Wasiriya beds are identical, we have also sug-
gested that the Waware and Wasiriya beds are most likely
contemporaneous (e.g. Tryon et al., 2012; Garrett et al., 2015). The
archaeology of Mfangano Island as described from surface collected
stone artifacts also consists of Levallois technology and bifacial
points (Tryon et al., 2012). The fauna are similar to those from
Rusinga Island and also indicate an open and arid environment
relative to the present (Tryon et al., 2012; Garrett et al., 2015).
3.3. Karungu
The Pleistocene beds of Karungu, Kenya, crop out around the
town of Sori on the Kenyan mainland (Fig. 3). These beds are
discontinuously exposed over an area of ~40 km
2
and are up to
10.5 m in thickness, overlying eroded topography of Miocene
bedrock (Pickford, 1984). The Pleistocene deposits are best exposed
at the localities of Kisaaka, Aringo, Aoch Nyasaya and Obware
(Figs. 4e6;Faith et al., 2014), with the most extensive and thickest
exposures at Kisaaka (Figs. 3 and 4)(Beverly et al., in press; Faith
et al., 2015). The deposits at Karungu are comprised of fine
grained siltstone and mudstone beds that have been pedogenically
modified into paleo-Vertisols and paleo-Inceptisols, conglomeratic
beds that represent fluvial channels, variably reworked tephra, and
tufas deposited by local springs (Beverly et al., in press; Faith et al.,
2015). On the basis of field characteristics of two widespread tuffs
at Karungu, Pickford (1984) proposed correlations among expo-
sures at Karungu, and between Karungu and other western Kenyan
Pleistocene localities where the “Nyando Ashes”were also present.
Like the Wasiriya beds and Waware beds described above, all in
situ and most surface collected stone artifacts documented during
fieldwork at Karungu from 2010 to 2013 are characteristically MSA,
including Levallois cores and bifacially worked MSA points. The
fauna from Karungu resembles that of the Wasiriya and Waware
beds in displaying both extinct and extant taxa, the majority of
which indicate an open and semi-arid grassland environment
(Faith et al., 2014, 2015).
4. Tephra correlation materials and methods
4.1. Materials
All 50 tuff samples analyzed for this study were collected from
or can be stratigraphically linked to a series of 32 sections <0.50 m
to >10 m thick measured from Pleistocene outcrops on Rusinga
Island, Mfangano Island and Karungu between 2009 and 2013.
Geographic locations, lithologies, and the location of archaeological
and paleontological sites are shown in Figs. 2e6with geographic
coordinates for each stratigraphic section provided in Table 2.
Whenever possible, tuffs were sampled from sections with multi-
ple tephra deposits exposed in stratigraphic succession. Field cor-
relations were made by walking exposures and by using a Jacob's
staff and Abney level to establish the stratigraphic equivalence
between exposed tuffs. Both field and laboratory methods of cor-
relation are necessary as exposures are discontinuous and tephra
deposits in the eLVB vary widely in their thickness, amount of
subsequent soil development, and amount and/or size of natural
glass.
4.2. Methods
All samples were examined in hand specimen, with low power
(10) magnification in the field, and in thin section from selected
specimens using a petrographic microscope at 40e100 magnifi-
cations. However, like the vast majority of tephra deposits in East
Africa (Brown and McDougall, 2011), the lithological characteristics
of the tuffs in outcrop and under magnification are generic and
insufficient for correlation due to aeolian fractionation (density-
driven separation of the vitric, crystal and lithic phases with
increased distance from the source) and varied syn- and post-
depositional environments. Although crystal composition can be
useful in some settings where glass is not preserved (e.g., McHenry
et al., 2008; Smith et al., 2011; McHenry, 2012), we base our cor-
relations on chemical analysis of volcanic glass shards, character-
ized by electron probe microanalysis of eleven major element oxide
proportions (Brown and Nash, 2014). Glass composition provides
the most diagnostic ‘fingerprint’for correlation purposes (Lowe,
2011). Correlation identifies deposits that derive from the same
eruption, but because tephra can be reworked during and following
initial sedimentation (see Orton, 1996 for an extensive review), the
presence of the same tephra in multiple outcrops does not neces-
sarily define an isochron or time-plane. Following the terminology
of Feibel et al. (1989), all deposits of correlated tephra will share the
same eruptive age, but the depositional age will vary according to
local conditions. Ideally, dates from multiple outcrops would be
used to assess local variance in depositional age, but this is not
always feasible. We thus assume only general age equivalence for
correlated deposits. However, in some cases at Karungu, Rusinga,
and Mfangano additional geological evidence, such as pristine glass
shards and evidence that the tephra are airfall deposits (e.g., tephra
blanketing and uniformly filling paleotopography) indicates that
there has been little to no reworking of some the tuffs, suggesting
that depositional ages likely approximate eruptive ages. Impor-
tantly, at Karungu, Rusinga, and Mfangano, we document a strati-
graphic sequence comprised of chemically distinct and correlated
tuffs that is repeated at several localities. This sequence of correl-
ative tephra thus serves as a relative dating tool for interpreting the
Fig. 3. Map of Karungu area around the town of Sori showing localities of Pleistocene
exposure: Aringo, Aoch Nyasaya, Kisaaka and Obware.
N. Blegen et al. / Quaternary Science Reviews 122 (2015) 89e111 93

Fig. 4. Above: Map of Kisaaka showing extent of Pleistocene exposure (after Beverly et al., in press) and locations of measured and sampled sections. Below: Stratigraphic columns
of measured and sampled sections at Kisaaka, arranged southwest (on left) to northeast (right). Lithologies indicated for all units. Tuffs with electron microprobe determined
chemical composition are color-coded to compositional group and labeled with sample number. Tuffaceous units not chemically characterized or assigned are shown in grey. Dotted
lines represent tuff units that can be traced laterally in the field between two or more measured sections.

Fig. 5. Right: Map of Aringo locality showing extent of Pleistocene exposure (after Beverly et al., in press) and locations of measured and sampled sections. Left: Stratigraphic
columns of measured sections at Aringo, arranged south (on left) to north (right). Lithologies indicated for all units. Tuffs with electron microprobe determined chemical
composition are color-coded to compositional group and labeled with sample number.
Fig. 6. Center: Map of Aoch Nyasaya and Obware showing extent of Pleistocene exposure and locations of measured sections (after Beverly et al., in press). Left: Stratigraphic
columns of measured sections at Aoch Nyasaya, arranged northwest (left) to southeast (right). Right: Measured section at Obware. Lithologies are indicated for all units. Tuffs with
electron microprobe determined chemical composition are color-coded to compositional group and labeled with sample number. Dotted lines represent tuff units that can be traced
laterally in the field between two or more measured sections. Sections were not measured for samples KRU2012-7 at Aoch Nyasaya or LVP2013-14 at Obware.
N. Blegen et al. / Quaternary Science Reviews 122 (2015) 89e111 95

age relationships of interbedded sediments and fossil- and artifact-
bearing sites in the eLVB.
4.3. Preparation
All preparation protocols were adapted from the University of
Utah Electron Microprobe lab recommendations (see Nash, 1992;
Brown and Fuller, 2008). Bulk samples of tuff (10e30 g) were
prepared by disaggregation with pestle and mortar and sieved
through 250 and 125
m
m mesh screens, retaining the fraction be-
tween. Samples were then washed repeatedly with deionized wa-
ter and the suspended clay fraction was decanted until the effluent
was clear. Cleaned tephra was then treated with 10% nitric acid in
sonic bath for five minutes to remove carbonates. Tephra samples
were subsequently treated for five minutes with 5% hydrofluoric
acid in a sonic bath to remove metal salts and clays potentially
adhering to the surface of the glass shards. Samples were then
rewashed in deionized water until the effluent was clear, and dried
in an oven at 90

C for at least six hours or until all visible moisture
was removed. Dried samples were magnetically separated on a
Franz isodynamic magnetic separator in two successive runs, the
first at low (0.1e0.3) amperage to separate the strongly magnetic
mineral components such as olivine, augite and opaque minerals,
and the second run at higher amperage (~0.9e1.0) in order to
separate weakly magnetic natural glass from nonmagnetic feld-
spars and quartz. Glass separates were mounted in twelve-well
epoxy grain mounts at the University of Utah. Standard mount
sizes are 1“(25 mm) round mounts with maximum height of 1”.
The University of Utah electron microprobe lab provided carbon
coating of samples with a Denton Benchtop Turbo IV high vacuum
evaporator. Each mount contained an MM3 standard obsidian
(Brown and Fuller, 2008) so that the samples and standard have the
identical thickness of carbon coating. Samples described in Tryon
et al. (2010) and Van Plantinga (2011) were previously analyzed
in the microprobe as resin-impregnated polished thin sections
prepared by Spectrum Petrographics, Inc. To reduce inter-analysis
variation resulting from the use of different instrumentation and
analytical protocols (cf. Kuehn et al., 2011) that may confound
correlation efforts, the current dataset includes new preparation
and analysis of all samples from Rusinga Island previously analyzed
by Tryon et al. (2010), and sample AV1006T5A from the Nyamita
Valley (Rusinga Island) studied by Van Plantinga (2011). New and
published results show good correspondence particularly for ele-
ments other than those known to be highly mobile (e.g., SiO
2
,Na
2
O,
K
2
O) and poorly suited for correlation.
4.4. Analysis
The relations among the vitric, crystal, and lithic phases of the
tephra deposits were examined in thin section using plane and
polarized light, backscattered electron imagery, and energy
dispersive electron probe microanalyses (EPMA). Geochemical
characterization of the vitric (glass) phase for thin sections and
grain mounts via EPMA used a Cameca SX-50 in the Department of
Geology and Geophysics at the University of Utah, USA. Analyses
were conducted using PC1, TAP, PET and LiF crystals on four
wavelength-dispersive spectrometers, with an accelerating voltage
of 15 keV, a beam current of 25 nA, and a spot size of 10
m
m. The
analytical routine for glass included Si, Ti, Zr, Al, Fe, Mn, Mg, Ca, Na,
K, O, F, and Cl. A natural obsidian standard (MM3) was used for
calibration of O-K
a
, Si-K
a
, Al-K
a
,K-K
a
. Mineral standards include
fluorite (F-K
a
), tugtupite (Cl-K
a
), albite (Na-K
a
), diopside (Ca-K
a
,
Mg-K
a
), hematite (Fe-K
a
), rutile (Ti-K
a
), rhodonite (Mn-K
a
), and
cubic zirconia (Zr- L
a
). Rounds of three standard analyses bracketed
rounds of four sample unknowns where 15e20 shots were taken
per sample (Nash,1992). Accidental mineral analyses of feldspars or
quartz or analyses with aberrantly low totals (<90%) were excluded
from this study. Oxygen was measured directly allowing for an
estimate of the water contents of the shards. This provides a
measure of the quality of the analysis (Nash, 1992). Na was
measured first on the TAP crystal with an analysis time of four
seconds on the analytical peak and two seconds on background on
either side of the peak, in order to minimize Na loss under the
electron beam. On-peak and background measurement times are as
follows (Pk/Bg sec): Si (15/15), Ti (25/25), Zr (37/22), Al (15/15), Fe
(25/25), Mn (25/25), Mg (40/40), Ca (20/16), Na (4/4), K (20/16), O
(20/20), F (20/20), Cl (20/20). Concentrations are calculated using
the PAP matrix correction procedure of Pouchou and Pichoir (1991).
Correction for “excess”F by interference of the Fe L
a
peak with F K
a
peak was accomplished by measuring a F-free Fe-bearing standard
(hematite) to yield a correction factor of 0.031. Background in-
tensities are measured on both sides of the analytical peak for all
elements but F on the PC1 crystal, where off-peak background is
measured to one side, and on-peak background intensity is inter-
polated using the estimated slope of the continuum (Pouchou and
Pichoir, 1991).
By choosing analytical conditions identical to those widely used
by the Department of Geology and Geophysics at the University of
Utah, we are able to directly integrate our data from the eLVB to a
large body of published data on extra-basinal tephra that may be
potential correlates (e.g., Brown and Fuller, 2008; Brown et al.,
2012, 2013). The selected beam size and current can cause un-
derestimates of volatile element abundance (especially Na) and
over-representation of Si, leading to higher than expected totals (cf.
Morgan and London, 1996; Hunt and Hill, 2001; Hayward, 2012),
and it is for these reasons that we use the restricted element list
defined below for our correlations. However, inter-laboratory
comparisons confirm that the equipment and protocols used by
the Utah laboratory (lab 5 in Kuehn et al., 2011) work exceptionally
well for tephras of a wide range of compositions.
Fig. 7. Total-alkali Silica graph (after Le Bas et al., 1986) of type samples of all eight
distinct tuffs discussed in this study along with obsidian samples from the Kenya Rift.
All obsidian data are from Brown et al. (2013).
N. Blegen et al. / Quaternary Science Reviews 122 (2015) 89e11196

4.5. Interpretation
Correlation of two or more tephra deposits is best viewed as a
hypothesis with different methods of distinguishing tephras
providing independent tests of any hypothesized correlation
(Feibel, 1999). Failure to distinguish tephras from different samples
by means of stratigraphy, lithology, petrography and element oxide
compositions measured with an electron microprobe constitutes
robust evidence for correlation (Tryon et al., 2008, 2010). Reported
major element oxides are not normalized because element oxide
totals including estimated water content were high and exploratory
data analysis using normalization did not alter interpretation.
However, samples plotted on the total alkali-silica (TAS) diagram
(Fig. 7) necessarily have totals normalized to 100% for comparison
with whole-rock samples (Le Bas et al., 1986).
Our samples include both fresh vitric tephra deposits, as well as
those subsequently reworked by fluvial processes or overprinted by
pedogenesis. We define fresh vitric tephra as poorly consolidated,
well-sorted ash-sized (2 mm) vitric ashes following the termi-
nology of Schmid (1981: 42). We indicate where tephras of this
lithology were sampled in Figs. 2 and 4. The remaining samples of
tuffs are more lithified and show incipient pedogenesis or other
evidence for bioturbation. The glass shards in these samples are set
in a predominantly silt to fine-sand groundmass, and grain
boundaries on the glass shards are sharp, indicating little me-
chanical abrasion through grain to grain collisions through
reworking (Tryon et al., 2010). Lithic and crystal phases from the
eruption cannot always be reliably distinguished from those found
in detritral sediments, and thus our correlations rely on chemical
composition of the glass component for all samples.
In the majority of samples, all analyzed glass shards clustered
around a discrete mean and represent a unimodal glass composi-
tion (Table 1;Figs. 8 and 9), further indicating minimal reworking
even in those ash beds visibly altered by fluvial or pedogenic pro-
cesses during or after initial deposition. However, some samples
displayed bimodal compositions. Bimodal compositions within a
Fig. 8. Schematic stratigraphic sections for all analyzed samples from the eastern Lake Victoria basin. Composite section of Pleistocene tephra with unique colors corresponding to
unique tephras named on the left of the figure. Individual tuff samples shown as rectangles with sample numbers for each sample at each locality listed on bottom right. Discrete
chemical modes of glass found in a single sample shown as unique colors within a single rectangle. Tuffs with glass shards of multiple compositions due to mechanical mixing have
colors separated with a vertical line. Tuffs with distinct modes of glass as a product of magmatic processes during eruption have colors representing modes separated by a diagonal
line. Grey rectangles are tuff samples lacking good chemical analysis. Dotted lines represent tuff units that can be traced laterally in the field between two measured sections. White
areas represent non-volcanoclastic sediments not used for correlation. Total-alkali Silica graph (after Le Bas et al., 1986) of type samples of all eight distinct tuffs discussed in this
study along with obsidian samples from the Kenya Rift. All obsidian data are from Brown et al. (2013). (For interpretation of the references to colour in this figure legend, the reader
is referred to the web version of this article.)
N. Blegen et al. / Quaternary Science Reviews 122 (2015) 89e111 97

Table 1
Mean major and minor element oxides by weight percent. Sample listed on left (No ¼number of analyses and M,N ¼mode and numberof modes forthe sample).One standard
deviation from the mean listed below each element oxide mean.
No M,N SiO
2
TiO
2
ZrO
2
Al
2
O
3
FeO MnO MgO CaO Na
2
OK
2
O F Cl Sum Less O Sum less O H
2
O Total
Wakondo Tuff
CAT09-05 17 1,1 61.67 0.54 0.07 15.51 7.97 0.36 0.32 1.02 9.01 4.75 0.46 0.31 102.00 0.26 101.73 0.03 100.87
1.39 0.09 0.04 0.28 0.19 0.02 0.02 0.03 0.54 0.51 0.19 0.01 2.36 0.08 2.41 0.88 1.88
CAT09-01 20 1,1 60.56 0.51 0.08 15.65 8.11 0.36 0.33 1.04 9.18 4.79 0.38 0.31 101.32 0.23 101.09 0.62 100.81
1.03 0.09 0.05 0.15 0.14 0.03 0.02 0.04 0.24 0.26 0.06 0.02 1.15 0.03 1.14 0.88 0.47
CAT09-22 11 1,1 58.73 0.56 0.07 15.28 7.41 0.31 0.33 1.02 9.12 4.61 0.25 0.28 97.97 0.17 97.80 2.27 99.24
0.81 0.09 0.07 0.28 0.42 0.04 0.02 0.03 1.08 0.50 0.11 0.02 1.14 0.05 1.11 0.80 0.79
CAT10-01 18 1,1 61.53 0.56 0.09 15.34 7.94 0.36 0.32 1.02 8.73 4.50 0.27 0.31 100.97 0.18 100.78 0.30 99.60
2.26 0.06 0.04 0.34 0.22 0.03 0.02 0.06 0.76 0.72 0.04 0.02 3.71 0.02 3.72 1.31 2.81
KRU2012-15 18 1,1 57.23 0.52 0.08 15.41 7.78 0.35 0.33 0.98 8.34 4.30 0.44 0.31 96.07 0.25 95.81 5.63 100.58
2.15 0.06 0.04 0.38 0.22 0.02 0.02 0.04 0.88 0.84 0.11 0.01 4.13 0.05 4.14 1.60 2.72
CRJ11-27 8 1,1 59.50 0.59 0.12 16.04 8.05 0.35 0.33 1.09 5.96 1.76 0.39 0.32 94.51 0.23 94.27 3.87 97.25
1.30 0.08 0.02 0.21 0.19 0.03 0.02 0.05 0.59 1.02 0.17 0.01 2.15 0.07 2.14 1.71 0.88
CRJ11-28 8 1,1 59.39 0.62 0.11 15.77 8.03 0.33 0.34 1.05 5.53 1.88 0.40 0.29 93.73 0.23 93.49 3.78 96.39
0.73 0.09 0.03 0.21 0.17 0.03 0.02 0.04 1.00 1.22 0.23 0.03 2.83 0.10 2.83 1.57 1.55
LVP2013-16 10 1,1 62.07 0.55 0.10 15.61 8.34 0.36 0.35 1.10 8.67 4.93 0.52 0.28 102.89 0.28 102.61 0.24 101.91
1.51 0.06 0.05 0.56 0.50 0.02 0.03 0.10 1.08 0.31 0.25 0.02 3.47 0.10 3.55 1.50 2.10
Nyamita Tuff
CAT09-21 21 1,1 64.38 0.59 0.05 15.62 6.76 0.29 0.36 1.13 7.73 5.02 0.26 0.17 102.36 0.15 102.21 0.66 100.80
1.16 0.08 0.04 0.14 0.13 0.03 0.02 0.04 0.28 0.42 0.15 0.01 1.51 0.06 1.49 1.09 0.53
CAT09-02b 12 1,2 62.03 0.58 0.04 15.55 6.78 0.29 0.36 1.13 7.58 4.83 0.13 0.18 99.47 0.09 99.38 1.46 100.08
1.54 0.08 0.05 0.16 0.17 0.03 0.01 0.03 0.39 0.39 0.05 0.02 2.13 0.02 2.14 0.86 1.55
CAT09-03 27 1,1 60.47 0.58 0.07 15.87 6.86 0.29 0.36 1.20 7.14 4.54 0.27 0.17 97.81 0.15 97.66 1.41 98.31
1.31 0.09 0.04 0.49 0.10 0.03 0.02 0.05 0.69 1.24 0.10 0.02 2.71 0.04 2.71 1.16 2.15
CAT 11-01 17 1,1 59.99 0.60 0.06 15.32 6.62 0.27 0.36 1.10 7.72 4.79 0.09 0.18 97.09 0.08 97.01 3.13 99.40
1.88 0.10 0.04 0.37 0.16 0.03 0.01 0.05 0.56 0.37 0.04 0.02 2.91 0.02 2.91 1.50 1.70
CAT11-02a 19 1,2 60.04 0.55 0.06 15.47 6.58 0.28 0.35 1.11 7.69 5.06 0.27 0.16 97.62 0.15 97.47 3.31 100.05
1.42 0.08 0.04 0.36 0.55 0.05 0.04 0.07 0.33 0.58 0.13 0.02 2.13 0.06 2.12 1.38 1.13
CAT11-05 16 1,1 59.37 0.58 0.08 16.03 6.93 0.37 0.37 1.18 5.32 1.72 0.24 0.18 92.38 0.14 92.23 3.42 94.88
1.82 0.08 0.05 0.48 0.27 0.11 0.02 0.06 1.13 0.92 0.20 0.01 3.55 0.08 3.52 1.43 2.38
KRU2012-01 19 1,1 61.28 0.57 0.05 15.50 6.48 0.28 0.34 1.11 7.68 4.99 0.08 0.16 98.51 0.07 98.44 2.44 100.16
1.25 0.10 0.05 0.28 0.57 0.03 0.04 0.08 0.39 0.26 0.06 0.02 1.53 0.03 1.53 1.07 1.07
KRU2012-02 20 1,1 60.68 0.60 0.08 15.29 6.55 0.26 0.33 1.06 7.41 4.80 0.08 0.16 97.30 0.07 97.23 2.81 99.31
0.98 0.11 0.04 0.40 0.54 0.06 0.07 0.23 1.28 1.41 0.05 0.04 2.27 0.03 2.26 1.55 1.07
KRU2012-03 19 1,1 54.76 0.63 0.05 15.84 6.79 0.28 0.37 1.10 7.67 4.57 0.30 0.16 92.54 0.16 92.38 9.13 100.75
2.15 0.08 0.04 0.13 0.15 0.03 0.02 0.04 0.33 0.56 0.13 0.01 2.64 0.06 2.65 1.33 1.47
KRU2012-06 20 1,1 56.31 0.61 0.04 15.47 6.68 0.27 0.37 1.05 7.40 4.36 0.27 0.18 93.00 0.15 92.84 7.83 99.93
2.42 0.07 0.04 0.64 0.20 0.03 0.03 0.09 0.81 0.75 0.10 0.03 4.38 0.04 4.39 1.33 3.42
KRU2012-07 14 1,1 60.14 0.61 0.04 15.42 6.95 0.27 0.36 1.12 7.83 4.98 0.12 0.17 98.01 0.09 97.92 1.97 99.12
1.29 0.06 0.04 0.38 0.17 0.03 0.03 0.04 0.50 0.33 0.05 0.01 2.03 0.02 2.02 1.20 1.97
KRU2012-10 16 1,1 59.08 0.56 0.02 15.89 6.84 0.28 0.36 1.12 7.63 4.76 0.27 0.16 96.97 0.15 96.82 5.83 101.48
1.20 0.06 0.05 0.30 0.12 0.03 0.01 0.04 0.47 0.64 0.05 0.01 2.18 0.02 2.19 1.10 1.58
KRU2012-11 18 1,1 58.79 0.60 0.05 15.72 6.77 0.27 0.36 1.08 7.59 4.16 0.29 0.17 95.53 0.16 95.37 7.15 102.12
1.18 0.06 0.05 0.19 0.12 0.03 0.01 0.06 0.53 0.63 0.07 0.02 2.07 0.03 2.06 0.94 1.68
KRU2012-13 18 1,1 57.90 0.56 0.05 15.61 6.75 0.30 0.36 1.09 7.36 4.23 0.35 0.17 94.72 0.18 94.54 7.00 100.79
1.78 0.08 0.04 0.39 0.15 0.03 0.03 0.05 0.53 0.65 0.21 0.02 2.96 0.09 2.96 1.14 2.34
KRU2012-16 18 1,1 59.56 0.57 0.04 15.82 6.67 0.28 0.35 1.09 7.90 4.86 0.31 0.18 97.63 0.17 97.46 4.73 101.45
1.02 0.08 0.08 0.19 0.22 0.03 0.04 0.04 0.42 0.41 0.11 0.06 1.50 0.06 1.48 1.22 0.65
KRU2012-17 17 1,1 59.14 0.58 0.04 15.75 6.63 0.28 0.36 1.08 7.62 4.11 0.37 0.17 96.13 0.19 95.94 4.82 100.02
1.49 0.07 0.05 0.16 0.13 0.03 0.02 0.03 0.37 0.48 0.17 0.01 2.05 0.07 2.07 1.38 1.04
KRU2012-18 10 1,1 61.47 0.56 0.08 15.40 6.55 0.28 0.38 1.08 7.37 4.77 0.47 0.17 98.59 0.24 98.35 2.76 100.37
1.92 0.09 0.04 0.51 0.42 0.03 0.11 0.07 0.72 0.27 0.17 0.02 3.53 0.07 3.53 1.68 2.19
LVP2013-01 20 1,1 62.18 0.59 0.08 15.92 6.67 0.30 0.35 1.13 6.98 5.11 0.44 0.17 99.93 0.22 99.71 2.28 101.24
0.95 0.04 0.04 0.27 0.22 0.02 0.01 0.05 0.36 0.22 0.17 0.01 1.67 0.07 1.73 1.04 1.63
LVP2013-02 19 1,1 62.13 0.58 0.10 15.89 6.71 0.28 0.33 1.10 7.09 5.01 0.44 0.17 99.82 0.22 99.60 2.70 101.55
0.75 0.05 0.04 0.14 0.08 0.03 0.04 0.09 0.50 0.21 0.08 0.02 1.35 0.04 1.35 1.32 0.58
LVP2013-03a 3 1,2 62.48 0.59 0.06 15.97 6.73 0.27 0.34 1.12 6.99 5.24 0.42 0.16 100.38 0.21 100.17 2.11 101.53
0.36 0.03 0.03 0.10 0.10 0.01 0.01 0.04 0.23 0.14 0.02 0.00 0.68 0.01 0.69 0.92 0.24
LVP2013-04 20 1,1 61.41 0.60 0.08 15.56 6.57 0.29 0.36 1.11 6.90 5.06 0.46 0.16 98.56 0.23 98.33 2.65 100.25
0.80 0.04 0.04 0.25 0.10 0.03 0.02 0.05 0.35 0.22 0.06 0.01 1.38 0.02 1.38 1.15 1.02
LVP2013-06 20 1,1 61.75 0.59 0.09 15.82 6.61 0.29 0.36 1.10 6.92 5.15 0.44 0.16 99.28 0.22 99.06 2.34 100.67
1.01 0.03 0.06 0.24 0.14 0.03 0.02 0.04 0.26 0.20 0.05 0.01 1.62 0.02 1.63 1.32 1.05
LVP2013-07a 14 1,2 62.25 0.60 0.10 15.98 6.55 0.28 0.36 1.08 6.83 5.12 0.49 0.16 99.80 0.24 99.56 1.77 100.60
1.32 0.05 0.04 0.33 0.14 0.02 0.01 0.06 0.38 0.23 0.17 0.02 2.14 0.07 2.19 1.50 1.14
LVP2013-11 16 1,1 63.25 0.59 0.10 16.28 6.56 0.31 0.36 1.15 7.12 5.19 0.45 0.16 101.52 0.23 101.29 1.23 101.79
0.74 0.05 0.05 0.21 0.39 0.04 0.02 0.06 0.37 0.24 0.10 0.01 1.37 0.04 1.39 1.69 0.90
12KIS26 16 1,1 62.96 0.58 0.09 16.27 6.44 0.29 0.36 1.15 6.87 5.08 0.46 0.17 100.73 0.23 100.49 2.35 102.12
1.08 0.03 0.04 0.22 0.10 0.06 0.02 0.07 0.35 0.23 0.06 0.02 1.62 0.03 1.62 1.48 0.64
12KIS28 13 1,1 62.82 0.61 0.08 16.13 6.56 0.30 0.37 1.11 7.04 5.01 0.53 0.17 100.72 0.26 100.45 0.02 99.75
1.72 0.04 0.04 0.36 0.15 0.03 0.02 0.05 0.45 0.26 0.12 0.01 2.78 0.05 2.80 0.01 2.79
Nyamsingula Tuff
CAT10-03 15 1,1 58.16 0.45 0.18 16.22 6.73 0.30 0.27 0.99 9.24 5.00 0.61 0.40 98.57 0.35 98.22 3.54 101.01
1.16 0.06 0.04 0.29 0.34 0.03 0.04 0.06 0.44 0.40 0.12 0.06 1.49 0.06 1.48 0.87 0.79
CAT11-02b 6 2,2 59.88 0.41 0.18 15.83 7.18 0.33 0.19 1.03 8.97 4.73 0.44 0.38 99.55 0.27 99.28 1.35 99.83
N. Blegen et al. / Quaternary Science Reviews 122 (2015) 89e11198

single sample were separated into ‘a’and ‘b’compositions and
treated as potentially different tephras for analysis (Fig. 9). Identi-
fying the consistent presence of a bimodal composition is an
important step in distinguishing tephra deposits that contain two
modes of glass as a product of magmatic processes during eruptions
(e.g., due to a differentiated magma chamber or sampling host rock
during magma ascent), and therefore represent the same eruptive
event, from those that contain two modes as a product of post-
depositional mixing. Bivariate plots of Cl versus TiO
2
visually
display the discrete modes of the tuffs discussed in this study
(Fig. 9). We chose this combination because it most clearly displays
similarities and differences within and between all correlative
groups of tuffs on a single plot (Fig. 9). Both TiO
2
and Cl are
immobile elements that are unaffected by hydration and analytical
variations between samples.
Postdepositional hydration, ion exchange and migration of the
alkalis in analysis are known to affect silica (Si), sodium (Na), po-
tassium (K) and possibly fluorine (F) content (Cerling et al., 1985;
Hunt and Hill, 2001). These diagenetic and analytical conditions
that selectively affect measurement and calculation of SiO
2
,Na
2
O,
and K
2
O have prompted us to follow Brown et al. (2012) in
excluding these element oxides for correlation purposes. We use a
restricted list of seven element oxides, TiO
2
,Al
2
O
3
, FeO, MnO, MgO,
CaO, and Cl in computation of similarity coefficients and other
statistical analyses. We depart slightly from Brown et al. (2012) by
including Al
2
O
3
. This element oxide is important in distinguishing
discrete chemical modes within samples attributed to a widespread
bimodal deposit (the BTPT, described below).
All correlations are supported by similarity coefficients (SCs) to
quantify similarity between means of glass analyses from samples
after which a tephra is named, the ‘type sample’, and the mean
value of all other modes in our dataset (Table 2,Brown et al., 2012;
Tryon et al., 2008). For any two-tephra comparisons, SCs are the
mean of the ratios obtained by dividing pairs of sample means
(with the larger value of the two samples always the denominator
such that the ratio is always 1) element by element as defined in
Table 1 (continued )
No M,N SiO
2
TiO
2
ZrO
2
Al
2
O
3
FeO MnO MgO CaO Na
2
OK
2
O F Cl Sum Less O Sum less O H
2
O Total
0.62 0.06 0.03 0.20 0.27 0.05 0.02 0.06 0.44 0.20 0.07 0.05 1.23 0.04 1.23 1.06 0.50
DP10-16 20 1,1 58.84 0.44 0.19 16.31 6.60 0.30 0.30 1.02 8.99 5.05 0.58 0.39 99.02 0.33 98.69 2.56 100.51
1.11 0.04 0.06 0.31 0.40 0.03 0.04 0.08 0.46 0.47 0.09 0.07 1.42 0.05 1.40 0.92 0.73
DP10-17 15 1,1 59.14 0.43 0.18 16.30 6.45 0.29 0.28 1.02 9.29 4.82 0.59 0.40 99.19 0.34 98.85 2.73 100.87
0.81 0.08 0.05 0.27 0.18 0.04 0.02 0.04 0.44 0.32 0.10 0.04 1.49 0.05 1.47 0.62 1.19
DP10-18 17 1,1 59.41 0.45 0.19 16.31 6.34 0.28 0.29 1.03 9.04 4.98 0.58 0.38 99.27 0.33 98.94 2.58 100.82
0.84 0.08 0.04 0.17 0.12 0.03 0.04 0.03 0.63 0.25 0.17 0.04 1.47 0.08 1.44 1.02 0.89
LVP2013-3b 31 2,2 60.94 0.44 0.22 16.17 7.16 0.33 0.19 1.01 8.06 4.89 0.81 0.38 100.59 0.43 100.16 1.96 101.32
0.98 0.04 0.06 0.23 0.23 0.03 0.02 0.08 0.45 0.28 0.13 0.05 1.51 0.06 1.53 1.18 1.06
LVP2013-7b 6 2,2 61.17 0.44 0.21 16.32 7.14 0.32 0.19 1.00 7.93 4.88 0.78 0.40 100.78 0.42 100.36 1.98 101.54
0.40 0.05 0.04 0.14 0.11 0.02 0.02 0.07 0.54 0.13 0.09 0.05 1.07 0.05 1.09 0.95 0.49
LVP2013-8 13 1,1 60.99 0.45 0.21 16.16 7.04 0.32 0.20 1.02 7.62 4.82 0.75 0.36 99.93 0.40 99.53 1.80 100.55
2.00 0.05 0.04 0.61 0.27 0.03 0.03 0.08 0.70 0.28 0.07 0.04 3.56 0.03 3.58 1.31 2.68
LVP2013-13 12 1,1 61.39 0.46 0.24 16.99 6.41 0.30 0.31 1.03 7.57 5.14 0.78 0.36 100.98 0.41 100.57 2.17 102.02
1.56 0.05 0.06 0.42 0.31 0.02 0.04 0.07 0.80 0.29 0.10 0.07 2.35 0.06 2.38 1.35 1.48
12KIS34 19 1,1 60.32 0.43 0.22 16.15 7.01 0.31 0.21 1.03 7.89 4.79 0.89 0.39 99.65 0.46 99.19 2.78 101.18
1.65 0.10 0.07 0.33 0.30 0.03 0.08 0.15 0.70 0.32 0.27 0.11 2.44 0.13 2.49 1.47 1.69
BTPT
CAT11-07a 7 1,2 62.25 0.66 0.10 12.72 9.14 0.44 0.20 0.99 8.20 4.76 0.38 0.26 100.11 0.22 99.89 0.72 99.60
0.52 0.11 0.07 0.14 0.18 0.04 0.01 0.05 0.25 0.17 0.12 0.02 0.82 0.05 0.83 0.58 0.35
CAT11-07b 13 2,2 61.92 0.81 0.02 15.22 7.07 0.33 0.43 1.40 7.25 5.46 0.20 0.11 100.21 0.11 100.10 1.14 100.46
0.86 0.11 0.04 0.30 0.39 0.04 0.08 0.11 0.46 0.17 0.20 0.02 1.28 0.09 1.22 0.71 0.79
CAT10-05a 9 1,2 62.24 0.69 0.12 12.64 9.01 0.44 0.21 0.97 7.08 4.44 0.51 0.26 98.61 0.27 98.34 0.53 97.86
0.43 0.03 0.03 0.14 0.19 0.03 0.02 0.04 0.51 0.16 0.06 0.02 1.05 0.03 1.04 0.67 0.77
CAT10-05b 3 2,2 61.28 0.79 0.04 15.23 6.94 0.36 0.43 1.37 6.21 5.43 0.32 0.11 98.51 0.16 98.35 0.10 97.68
1.73 0.07 0.02 0.50 0.24 0.03 0.02 0.07 0.27 0.23 0.02 0.03 3.06 0.01 3.07 1.14 1.93
LVP2013-09a 15 1,2 64.03 0.68 0.13 13.33 8.97 0.44 0.21 1.05 7.45 4.64 0.62 0.24 101.79 0.32 101.47 0.25 100.73
0.85 0.06 0.04 0.47 0.49 0.04 0.03 0.11 0.40 0.15 0.12 0.02 1.28 0.05 1.26 0.79 1.23
LVP2013-09b 12 2,2 62.97 0.78 0.06 15.74 7.00 0.35 0.42 1.40 6.50 5.36 0.43 0.10 101.10 0.20 100.90 1.12 101.24
0.86 0.06 0.05 0.35 0.41 0.04 0.06 0.08 0.34 0.13 0.06 0.02 1.15 0.03 1.14 1.44 1.35
LVP2014-10a 14 1,2 62.66 0.66 0.12 12.78 8.86 0.43 0.20 0.97 7.37 4.63 0.54 0.26 99.49 0.29 99.20 0.21 98.00
1.49 0.07 0.04 0.40 0.50 0.05 0.04 0.09 0.68 0.16 0.06 0.02 2.23 0.03 2.22 1.22 2.15
LVP2014-10b 6 2,2 62.05 0.78 0.02 15.50 6.84 0.34 0.41 1.38 6.52 5.37 0.33 0.10 99.64 0.16 99.48 0.20 98.92
0.47 0.07 0.03 0.37 0.24 0.04 0.07 0.08 0.43 0.18 0.04 0.02 0.90 0.02 0.89 1.10 0.62
LVP2014-14a 7 1,2 63.81 0.65 0.13 12.57 8.24 0.34 0.13 0.98 6.39 4.61 0.47 0.25 98.57 0.25 98.32 1.22 98.62
1.04 0.04 0.06 0.47 0.37 0.03 0.02 0.08 0.26 0.22 0.06 0.04 1.54 0.03 1.53 1.16 0.83
LVP2014-14b 13 2,2 62.10 0.77 0.05 14.73 6.78 0.31 0.37 1.45 5.79 5.30 0.32 0.11 98.07 0.16 97.91 1.20 98.35
1.77 0.10 0.05 0.55 0.44 0.04 0.13 0.13 0.48 0.35 0.07 0.03 2.46 0.03 2.47 1.78 1.24
LVP2013-15a 5 1,2 65.25 0.63 0.12 13.25 8.73 0.40 0.18 1.09 6.81 4.66 0.54 0.24 101.89 0.28 101.61 0.38 101.01
1.72 0.03 0.08 0.62 0.54 0.04 0.13 0.17 1.01 0.23 0.08 0.03 1.70 0.03 1.68 1.65 0.39
LVP2013-15b 21 2,2 63.20 0.77 0.05 15.33 6.82 0.31 0.41 1.52 6.20 5.42 0.37 0.10 100.50 0.18 100.32 2.49 102.05
1.27 0.04 0.04 0.47 0.36 0.04 0.10 0.09 0.26 0.16 0.05 0.02 1.42 0.02 1.41 1.48 0.54
Unique Tuffs
AV1004T5A 15 1,1 63.83 0.52 0.20 10.80 8.05 0.35 0.14 0.66 7.16 3.74 0.65 0.31 96.41 0.34 96.06 1.57 96.73
2.08 0.07 0.04 0.16 0.14 0.03 0.03 0.02 0.83 0.52 0.17 0.02 2.90 0.07 2.90 1.46 1.88
CAT09-02a 5 1,2 64.80 0.62 0.17 10.51 9.39 0.37 0.13 0.68 7.54 3.31 0.56 0.35 98.44 0.32 98.12 1.23 98.30
1.00 0.08 0.07 0.12 0.22 0.02 0.01 0.03 0.72 0.79 0.04 0.01 2.43 0.02 2.42 1.30 1.25
X-5A-3 14 1,1 57.90 0.63 0.15 13.75 9.20 0.43 0.37 1.22 8.39 4.81 0.53 0.21 97.62 0.27 97.34 0.48 96.80
1.46 0.07 0.06 0.14 0.19 0.03 0.01 0.04 0.38 0.17 0.09 0.02 1.49 0.04 1.49 0.77 0.81
LVP2013-05 14 1,1 71.21 0.13 0.41 10.30 3.96 0.06 0.02 0.10 4.75 4.14 1.18 0.44 96.70 0.60 96.11 4.62 100.29
0.95 0.03 0.07 0.43 0.19 0.02 0.01 0.04 0.22 0.13 0.11 0.04 1.15 0.05 1.17 0.61 1.05
N. Blegen et al. / Quaternary Science Reviews 122 (2015) 89e111 99

Borchardt et al. (1972: 302). In this study we have restricted SC
analysis to the seven element oxides noted above (see Brown et al.,
2012). Resulting SCs range from 0 (complete dissimilarity) to 1
(perfect similarity).
Previous studies have proposed arbitrary cutoffs for interpreting
SCs in terms of potential correlation. For example, Kuehn and Foit
(2006) propose a value of 0.95 for definitive correlation,
whereas Froggatt (1992) recognizes that values 0.92 are typically
accepted for correlations. In this study, we implement randomi-
zation procedures to develop empirically informed SC cutoffs for
accepting or rejecting potential correlations. For each of the tephra
type or representative samples, we use the mean and standard
deviation of each element oxide to generate 5000 random
normally-distributed samples using the R statistical package (R
Development Core Team, 2014); this effectively represents 5000
replicates of the type tephra. We then calculate SCs between each of
the 5000 replicates and the type sample, which is used to generate
a frequency distribution of expected SC values when comparing
two samples of the same tuff. From this distribution, we determine
the lower SC limit that encompasses the upper 95% of observations.
We use this value as the cutoff for rejecting potential correlations.
For example, an SC value between an unknown tuff and a type
sample that falls below this cutoff is excluded for consideration as a
potential correlate. To increase the stringency of our protocol, we
also require that the seven element oxides of the unknown tuff
considered in our analysis overlap within two standard deviations
of the mean of the type sample. This is because an unknown sample
that is very similar in composition for most element oxides (e.g., 6
of 7) to a type sample will record a relatively high SC value, even if
one oxide is distinct and outside the range of expected values. The
SCs included in this analysis were used as a data exploration and
confirmation technique. All correlations were investigated in more
detail utilizing the known stratigraphy of a site and visual inspec-
tion of the tephra datasets.
5. Radiometric dating
There are a number of available methods that focus on the vitric
or crystal phases of a tephra to determine its eruption age (Feibel
et al., 1989) including fission track, thermoluminescence, and
40
Ar/
39
Ar methods; other approaches such as UePb dating of zir-
cons more accurately dates crystal formation rather than eruption
(e.g., Simon et al., 2008). We have not been able to apply any of
these methods to directly estimate the eruption ages of any of the
eLVB tephra deposits because of iron-oxide mineral inclusions in
the glass shards and because datable minerals are either too fine-
grained or sparse to be dated.
The alternative approach used here is to determine the depo-
sitional age of the tephra (Feibel et al., 1989). Differences in the time
between eruption and initial deposition (sedimentation) of fall-out
deposits can range from minutes to hours for areas proximal to the
source volcano, to several years for finer particles that enter the
stratosphere. Once deposited, tephra can be remobilized and
reworked through normal sedimentary processes, and thus the
timing of re-deposition may differ substantially from the initial
depositional age.
We use two methods to estimate the depositional age of the
eLVB tuffs. The first, already mentioned, is the direct AMS
14
C dating
of gastropod shells found in life position within tuffs from Rusinga
Table 2
Representative or type samples of the named tuffs and discrete chemical modes discussed in this study. GPS coordinates for each type or representative sample or mode
provided. SCs ¼similarity coefficients defined using TiO
2
,Al
2
O
3
, FeO, MnO, MgO, CaO, Cl (see text for details). Results of randomization of type sample or mode means and
standard deviations to determine mean similarity coefficient values and lower 95% similarity coefficient confidence limits.
Tuff Type sample GPS coords (WGS 1984) SCs
Mean Lower 95% Confidence Limit
Wakondo CAT09-05 S 00

25.565
0
0.95 0.93
E 034

10.347
0
Nyamita CAT09-21 S 00

25.246
0
0.95 0.93
E 034

09.599
0
Nyamsingula CAT10-03 S 00

24.706
0
0.93 0.89
E 034

11.126
0
BTPT Mode A CAT11-07a S 00

48.342
0
0.95 0.92
E 034

08.216
0
BTPT Mode B CAT11-07b S 00

48.342
0
0.90 0.85
E 034

08.216
0
Nyamita Valley Trachytic Tuff AV1004T5A S 00

25.584
0
0.95 0.92
E034

09.554
0
Nyamita Valley Trachytic mode CAT09e02a S 00

25.369
0
0.96 0.93
E 034

09.615
0
Songhor X-5A 3S00

03.324
0
0.96 0.94
E 035

13.084
0
Rhyolite LVP2013-05 S 00

48.278
0
0.85 0.76
E 034

08.274
0
Fig. 9. Bivariate plot of Cl against TiO
2
for means of all samples discussed in this study.
N. Blegen et al. / Quaternary Science Reviews 122 (2015) 89e111100

Island and Mfangano Island. As the snails burrowed into the pre-
vious deposited sediments, they post-date deposition, but appar-
ently pre-date lithification, providing a minimum age for the
deposits. The second method was to utilize optically stimulated
luminescence dating to determine sediment burial ages.
5.1. Optical dating
We used optical dating of sediments to constrain the age of
tephra deposition. All dates derive from localities on Rusinga Island.
We collected two sediment samples (RUP-1 and RUP-2) respec-
tively above and below the type sample of the Nyamita Tuff at
section Nyamita 2 (Figs. 2 and 10;Li et al., 2015). As discussed in
detail below, textural, microscopic, and geochemical data strongly
suggest that this deposit underwent little to no reworking
following deposition, and thus depositional age likely approxi-
mates eruptive age. At Wakondo we collected three samples (RUP-
3, RUP-4 and RUP-5) from the same sedimentary unit in a channel
complex ~1 m above the Wakondo Tuff. The age of the channel
complex provides a minimum age for the deposition of the
Wakondo Tuff. For all samples, we extracted 180e212
m
min
diameter potassium-rich (K) feldspar grains, using standard pro-
cedures for dating; no quartz grains were present in the sediment
samples. We measured individual grains of K-feldspar from each of
Fig. 10. (A) Photograph of the Wakondo Tuff at Wakondo, Rusinga Island. (B) Note the fresh grey vitric character of the tuff in close-up view of the outcrop. (C) Photograph of the
type section Nyamita 2 from the Nyamita Valley, Rusinga Island. The type sample of the Nyamita Tuff, CAT09-21, taken from the fresh, vitric tephra deposit at the base of the tuff in
this section is indicated. OSL samples RUS-1 and RUS-2 bracketing CAT09-21 also indicated. (D) Photograph of sample CAT10-03, type sample of the Nyamsingula Tuff, from section
DP10.14-DP10.15 at the locality of Nyamsingula, Rusinga Island. Note the fresh grey, ashy fine texture of the fresh, vitric tephra deposit and the contrast with the overlying soil. (E)
Panoramic photograph of the section Kisaaka Main and the surrounding exposure showing (from bottom to top) the Nyamita Tuff (LVP2013-01, 02), the Nyamsingula Tuff (LVP2013-
08) and the representative sample (CAT11-07) of the BTPT in stratigraphic sequence. Also note the lateral extent of the Nyamita Tuff at Kisaaka.
N. Blegen et al. / Quaternary Science Reviews 122 (2015) 89e111 101

Table 3
Similarity coefficients for all distinct modes of samples analyzed in this study based on the restricted seven-element list. Samples listed vertically (left) and compared with the
type samples of every chemically unique tuff or mode listed horizontally (top). Lower 95% confidence limit for each type sample or mode listed in parentheses. Colored squares
are samples or modes with similarity coefficient the 95% lower confidence limit determined by randomization when compared to the type sample and overlap at two
standard deviations for the seven element oxides used for correlation when compared to the type sample. Black squares are the type sample or mode compared to itself.
N. Blegen et al. / Quaternary Science Reviews 122 (2015) 89e111102

the samples for equivalent dose (D
e
) determination, using a two-
step post-IR IRSL measurement procedure (Thomsen et al., 2008),
in which a prior-IR IRSL stimulation at 200

C(Li and Li, 2012) and a
post-IR IRSL stimulation at 275

C were adopted, to overcome
possible age underestimation caused by anomalous fading (Wintle,
1973; Huntley and Lamothe, 2001). The dose rates were estimated
using a combination of laboratory-based beta counting, field-based
gamma spectrometry measurements, together with calculations of
the cosmic ray and internal beta dose rates. The applicability of the
pIRIR method for the samples in this study and the reliability of the
measurements were determined using a suite of different tests. A
description of the method and D
e
and dose rate measurement
N. Blegen et al. / Quaternary Science Reviews 122 (2015) 89e111 103

procedures and the test results are presented in detail in Supple-
mentary Information and in Li et al. (2015).
6. Results
6.1. Tephra correlation
Element oxide wt. % abundances, and comparisons using SCs
indicate the presence of eight distinct distal tephra deposits of nine
chemical compositions (Fig. 9,Tables 1, 3 and 4). Seven of these
tephra deposits occur across Rusinga Island, Mfangano Island, and
Karungu, four of which are sufficiently widespread to merit names
and type or representative localities. These four units in strati-
graphic order from the base upwards are the Wakondo Tuff, the
Nyamita Tuff, the Nyamsingula Tuff, and the Bimodal Trachypho-
nolitic Tuff (BTPT). The first three of these deposits include fresh
vitric tephra and are named on the basis of a geographic type lo-
cality following the North American Stratigraphic code (North
American Commission on Stratigraphic Nomenclature, 2005). No
such deposits of the BTPT have been found in the eLVB, and we thus
provide a provisional name based on its characteristic bimodal
chemical composition. Fig. 8 shows these deposits in stratigraphic
order along with the three unique tuffs from the eLVB: two
compositionally similar trachytes from Rusinga Island's Nyamita
Valley, and a single rhyolite from the Kisaaka locality at Karungu.
The eighth distinct tuff, from Songhor (Fig. 1b, c), is outside the area
of the eLVB that is the primary focus of this study. Thus this tuff is
not shown in Fig. 8, but is important in that it suggests a still un-
tapped potential for tephra correlation in and around the eLVB.
6.1.1. The Wakondo Tuff
The Wakondo Tuff is named after the Wakondo locality on
Rusinga Island where it was first recognized by Tryon et al. (2010).
Because of its fresh lithology, abundant glass and documented
stratigraphic position at Wakondo (Fig. 2;Tryon et al., 2010: 5),
sample CAT09-05 serves as the type sample of this tephra (Fig. 10a,
b).
6.1.1.1. Composition of the Wakondo Tuff. The type sample (CAT09-
05) is a ~20-cm-thick, massively bedded, weakly consolidated,
grey-green fresh, vitric ash composed of microscopic pumices and
pumice fragments (50e200
m
m) and sinuous glass shards generally
<5
m
me~100
m
m with oval and linearly stretched vesicles. At the
type section, the Wakondo Tuff is interbedded within ~4 m of
siltstone (Fig. 2,Tryon et al., 2010). The Wakondo Tuff is a phonolite
(Fig. 7;Tryon et al., 2010; Van Plantinga, 2011). All samples
attributed to it are unimodal (Fig. 9;Table 1). There is minimal
evidence for reworking of the Wakondo Tuff at its type locality.
6.1.1.2. Similarity coefficients of the Wakondo Tuff. SC values calcu-
lated between CAT09-05 and the randomized replicates of this
sample produced an average SC value of 0.95 and a lower 95%
confidence limit of 0.93 (Table 2). The seven samples correlated to
the Wakondo Tuff have a SC of 0.95 or higher when compared to the
type sample CAT09-05 (Table 3). All other tephra samples have SCs
of 0.89 or lower when compared with samples of the WakondoTuff
(Table 3).
6.1.1.3. Additional exposures of the Wakondo Tuff. In the Wasiriya
beds of Rusinga Island, samples attributed to the Wakondo Tuff
occur at the Wakondo locality (type sample CAT09-05 and two
other minimally reworked deposits, CAT10-01 and LVP2013-16
(Fig. 2). At the Nyamita locality, the Wakondo Tuff is present as
sample CAT09-01, a fresh, vitric tephra deposit at the base of the
section at Nyamita 1 and a reworked and redeposited bed up-
section sampled as CAT09-22 (Fig. 2). The Wakondo Tuff has not
been found on Mfangano Island. At Karungu, the Wakondo Tuff is
found at the northernmost section (Kisaaka North) sampled as
KRU2012-15 (Fig. 4).
6.1.2. The Nyamita Tuff
The Nyamita Tuff is named after locally extensive but discon-
tinuous ~1 km exposures of this deposit along the Nyamita Valley of
Rusinga Island (Tryon et al., 2010; Van Plantinga, 2011; Garrett
et al., 2015). We follow Tryon et al. (2010) in using sample
CAT09-21 from section Nyamita 2 in the Nyamita Valley as the type
sample for this tuff (Figs. 2 and 10c; Tryon et al., 2010).
6.1.2.1. Composition of the Nyamita Tuff. The type sample CAT09-21
is a 0.33-m-thick fresh, vitric deposit of unconsolidated grey ash,
found at the base of a ~3.5-m-thick massive deposit of the Nyamita
Tuff (Fig. 10c; Tryon et al., 2010). The type sample contains abun-
dant grey glass shards generally 5e100
m
m. The Nyamita Tuff is a
trachyphonolite (Fig. 7), with samples distinguished by their
consistent and low average Cl content of 0.16 ±0.02 weight percent
(Table 1,Fig. 9). There is no evidence for reworking of the type
deposit of the Nyamita Tuff.
6.1.2.2. Similarity coefficients of the Nyamita Tuff. SC values calcu-
lated between CAT09-21 and the randomized replicates of this
sample produced an average of 0.95 and a lower 95% confidence
limit of 0.93 (Table 2). Samples attributed to the Nyamita Tuff have
SCs between 0.94 and 0.99 compared with the type sample CAT09-
21 (Table 3). Other samples compared with CAT09-21 have SCs of
0.89.
6.1.2.3. Additional exposures of the Nyamita Tuff. Field observations
suggest that the Nyamita Tuff is the most common and widespread
tuff in the eLVB, a hypothesis subsequently confirmed by chemical
correlation. CAT09-21 samples the freshest vitric tephra deposit
from Rusinga Island's Nyamita Valley, but this tuff can be physically
traced laterally to section Nyamita 3, visually correlated using a
Jacobs staff and Abney level between all localities in Nyamita Val-
ley, and chemically correlated throughout the Nyamita Valley over
a north-south transect for ~1 km (Garrett et al., 2015; Tryon et al.,
2010; Van Plantinga, 2011:20). The present study demonstrates
Table 4
Dose rate data, De values and pIRIR ages for the 2 samples from Nyamita and 3 samples from Wakondo.
Sample name Moisture content (%) Dose rates (Gy/ka) Total dose rate (Gy/ka) D
e
(Gy) pIRIR age (ka)
Beta Gamma Cosmic Internal
Nyamita RUP-1 11.5 1.73 ±0.09 0.83 ±0.04 0.18 ±0.03 0.84 ±0.07 3.57 ±0.12 166 ±11 46.4 ±3.6
RUP-2 4.5 1.71 ±0.08 1.01 ±0.04 0.16 ±0.02 0.84 ±0.07 3.72 ±0.11 185 ±11 49.7 ±3.5
Wakondo RUP-3 4.0 2.17 ±0.10 1.37 ±0.06 0.21 ±0.03 0.84 ±0.07 4.59 ±0.14 38 ±6 73.8 ±6.3
RUP-4 5.7 2.18 ±0.11 1.29 ±0.06 0.21 ±0.03 0.84 ±0.07 4.46 ±0.14 49 ±7 67.5 ±6.3
RUP-5 7.5 2.28 ±0.12 1.41 ±0.07 0.21 ±0.03 0.84 ±0.07 4.75 ±0.16 23 ±8 63.6 ±6.0
N. Blegen et al. / Quaternary Science Reviews 122 (2015) 89e111104

near ubiquity of the Nyamita Tuff at 21 of 32 measured sections in
the eLVB, including 13 of 15 measured sections of Pleistocene ex-
posures at Karungu (Figs. 4e6, 8). Although locally widespread on
Rusinga Island, the Nyamita Tuff is absent from our studied expo-
sures on Mfangano Island. At Karungu, the Nyamita Tuff occurs as a
fresh, vitric tephra deposit in three samples (12KIS28, LVP2013-01
and LVP2013-04) in three measured sections at Kisaaka (KIS-10,
Kisaaka Main and Kisaaka North respectively; Fig. 4). The bottom
~10e15 cm of the ~75-cm-thick tuffs from which these three
samples were taken contain similar grey ash-sized grains, are
powdery to the touch in hand samples, display abundant
~50e100
m
m angular glass shards visible at 40e100 magnification
using a petrographic microscope, and are most commonly found in
micro-lows on the landscape indicating that the basal portions of
the tuffs were deposited by airfall and at most minimally reworked.
The upper ~60 cm of these tuffs at sections KIS-10, Kisaaka Main
and Kisaaka North are slightly more reworked deposits grey-brown
to tan color, with a more silty texture in hand sample, small ~1 cm
Mn-stained root casts and more detrital grains visible in thin sec-
tion. Field observations show the Nyamita Tuff is laterally contin-
uous for ~1 km at the locality of Kisaaka (Figs. 4 and 10e) and at this
specific locality chemical correlation confirms this lithostrati-
graphic correlation. Chemical correlations presented here show all
samples from the laterally continuous units (KRU2012-10,
KRU2012-11, KRU2012-13, 12KIS26, 12KIS28, LVP2013-01, LVP2013-
02, CAT11-05, LVP2013-06, LVP2013-04, KRU2012-06) are chemi-
cally homogenous, unimodal and indistinguishable therefore
constituting robust evidence for correlation. This laterally contin-
uous exposure of the Nyamita Tuff at Kisaaka also drapes the gilgai
topography of the paleosols on which is sits (Fig. 11). This demon-
strates laterally extensive evidence for airfall deposition, burial and
preservation of the Nyamita Tuff at Kisaaka.
The presence of fresh, vitric tephra deposits of the Nyamita Tuff
in the Nyamita Valley and at Kisaaka ~50 km to the south, the
consistent chemical homogeneity of Nyamita Tuff samples and
their stratigraphic position in relation to other marker beds make
the Nyamita Tuff a locally useful lithostratigraphic marker (Fig. 8).
In addition to being widespread, Nyamita Tuff outcrops are the
thickest of any tephra deposit in our sample, with a maximum
observed thickness of ~4 m at Nyamita and >2 m in and around the
Kisaaka North section of Karungu (Figs. 2, 4 and 10c). The extent to
which these thicknesses represent post-depositional sedimentary
admixture is as yet undetermined. The Nyamita Tuff may represent
either deposits of the largest eruption preserved inthe eLVB,and/or
a period of major changes in the local erosional and depositional
regime.
6.1.3. The Nyamsingula Tuff
The Nyamsingula Tuff was first recognized at the Pleistocene
exposures of the western Nyamsingula locality on Rusinga Island
(Figs. 2 and 7). Nyamsingula contains four laterally traceable out-
crops of fresh, grey, vitric ash (CAT10-03, DP10-16, DP10-17, DP10-
18) with indistinguishable compositions all attributed to the
Nyamsingula Tuff (Fig. 2). We use a particularly fresh, vitric tephra
deposit, sampled as CAT10-03 from near the base of the 15-m-thick
section DP10.14-DP10.15, as the type sample for the Nyamsingula
Tuff (Figs. 2 and 10d).
6.1.3.1. Composition of the Nyamsingula Tuff. In outcrop, CAT10-03
is a 20 cm-thick grey, vitric ash (Fig. 10d) of abundant grey glass
with microscopic pumices, pumices fragments, some 50e200
m
m,
round to subround and slightly oval often overlapping vesicles. The
Nyamsingula Tuff is a phonolite (Fig. 7) and distinguished by its
high average aluminum (Al
2
O
3
>16.0 wt. %), low average titanium
(TiO
2
~0.45 wt. %) and high average chlorine content (Cl ~ 0.40 wt.
%) (Fig. 9,Table 1).
6.1.3.2. Similarity coefficients of the Nyamsingula Tuff. SC values
calculated between CAT10-03 and the randomized replicates of this
sample produced an average SC value of 0.93 and a lower 95%
confidence limit of 0.89 (Table 2). All samples attributed to the
Nyamsingula Tuff have a SC of 0.91 or greater compared to the type
sample CAT10-03, and no other sample has a SC of over 0.87
compared to CAT10-03 (Table 3). Additionally, samples of Nyam-
singula Tuff are consistently found stratigraphically above the
Nyamita Tuff and below the Bimodal Trachyphonolitic Tuff.
6.1.3.3. Additional exposures of the Nyamsingula Tuff. On Rusinga
Island the Nyamsingula Tuff is known only from the type locality,
and it is not present on Mfangano Island. At the Karungu exposures
this tuff is found at Kisaaka, Aringo and Obware. At the Kisaaka
Main and KIS-10 sections, the Nyamsingula Tuff overlies the Nya-
mita Tuff (Fig. 4). At the RCS and ZTG sections at Kisaaka, modes ‘b’
of samples LVP2013-03 and LVP2013-07 are attributed to the
Nyamsingula Tuff, admixed with glass from underlying deposits of
the Nyamita Tuff, indicating local syn- or post-depositional
reworking. Similar admixture of the glass of the Nyamsingula Tuff
and Nyamita Tuff is seen at Aringo Section A (CAT11-02; Fig. 5). The
Nyamsingula Tuff is also present at Obware at the WPT 212-214
section (Fig. 6). At Kisaaka Main and at Obware, the Nyamsingula
Tuff is overlain by the Bimodal Trachyphonolitic Tuff (Figs. 4, 6 and
8).
6.1.4. The Bimodal Trachyphonolitic Tuff (BTPT)
The BTPT is named for its distinctive composition. The repre-
sentative sample chosen for this tuff is CAT11-07 from section
Kisaaka Main at the locality of Kisaaka, Karungu because it contains
abundant glass and the sample occurs in stratigraphic sequence
with other named tuffs (Figs. 4, 8 and 10e). Unlike the other three
named tuffs discussed in this study, no example of fresh, vitric
Fig. 11. Photographs of the Nyamita Tuff exposed throughout the locality of Kisaaka
(modified from Beverly et al., submitted for publication). The Nyamita Tuff drapes
gilgai topography of the paleo-Vertisol. This topography is formed when the smectitic
clays of Vertisols shrink and swell with the wet and dry seasons. The rapid airfall
deposition of the tuff preserved these micro-highs and lows on the landscape from
fluvial truncation.
N. Blegen et al. / Quaternary Science Reviews 122 (2015) 89e111 105

tephra for the BTPT have yet been found, and thus we do not
designate a formal name and type locality. Samples attributed to
the BTPT are always found as a ~10e50-cm-thick bed with varying
degrees of pedogenic development, near the modern surface and
often incorporated into paleosols. Bed thickness ranges from 10 to
55-cm, commonly with small (<5 mm) Mn-stained root casts.
6.1.4.1. Composition of the BTPT. Despite their weathered appear-
ance in outcrop and hand sample, all samples of the BTPT preserve
fresh glass of a distinctive grey-brown, green-brown or light brown
color under plane-polarized light. Glass occurring as microscopic
pumices and pumice shards, sometimes over 200
m
m, with round
to stretched vesicles is common in most samples of this tuff, but
smaller shards 5e25
m
m occur. As the name suggests, this tuff is a
trachyphonolite (Fig. 7), with glass compositions divided into two
chemically distinct modes (Fig. 9,Table 1). One mode (CAT11-07a)
is distinguished by low aluminum content (Al
2
O
3
~ 12.5 wt. %) and
high iron content (FeO ~ 8e9 wt %), and the second mode (CAT11-
07b) is distinguished by its higher aluminum content
(Al
2
O
3
~ 15.25 wt %) and lower iron content (FeO ~ 6.5e6.75 wt%).
The bimodal nature of the BTPT is illustrated in Fig. 9. For clarity we
label the low aluminum mode ‘a’and the higher aluminum mode
‘b’in all samples attributed to the BTPT (Fig. 9,Table 1). Each of
these two compositional modes is distinct from each other and
from all other tephra samples found in the eLVB (Fig. 9). The two
compositional modes found in all samples of the BTPT are petro-
graphically distinct from all other glass shards in this study, but
indistinguishable from one another. These two modes exhibit glass
chemistry indicative of density dependent zonation of a magma
chamber (Macdonald et al., 1994) and are always found together
whereas neither of the two modes is found in samples attributed to
other tuffs from the eLVB. For these reasons we interpret the
bimodal composition of these samples as the result of magmatic
processes during eruption (cf. Shane et al., 2008) rather than post-
depositional mixing.
6.1.4.2. Similarity coefficients of the BTPT. SC values calculated be-
tween CAT11-07a and the 5000 randomized replicates of this
sample produced an average SC value of 0.95 and a lower 95%
confidence limit of 0.92. The same procedure calculated for CAT11-
07b produced an average SC value of 0.90 and a lower 95% confi-
dence limit of 0.85 (Table 2). The lower SC values generated here
reflect the relatively high variability (standard deviations) of Al, Fe,
and Ca oxides measured in glasses from mode CAT11-07b. Previous
studies note the difficulty in using similarity coefficients with
heterogeneous tephra deposits (Riehle et al., 2008). However, all
samples attributed to the BTPT share a mode ‘a’similarity coeffi-
cient of 0.92 or higher with the type sample CAT11-07a. No other
sample has a SC of over 0.86 compared to CAT11-07a. All samples
attributed to the BTPT also share a SC of 0.94 with the ‘b’mode of
the type sample CAT11-07b (Table 3). No other sample has a SC of
over 0.84 compared to CAT11-07b. The agreement of the SCs from
both modes of the BTPT (Table 3), the distinctive brown color of
BTPT glass and consistent stratigraphic observations in the field
(Figs. 4 and 6) make a strong case for correlation of these five tuffs
attributed to the BTPT.
6.1.4.3. Additional exposures of the BTPT. The BTPT is not currently
known from Rusinga Island. It is present on Mfangano Island, at the
Walangani locality sampled as CAT10-05. It is also present at Kar-
ungu at Aoch Nyasaya and Obware, in addition to the representa-
tive locality at Kisaaka (Figs. 5 and 6). A single sample from Obware,
LVP2013-14, was found at the same stratigraphic level as LVP2013-
15 (Fig. 6), and based on our field lithostratigraphic correlation this
tephra deposit likely also correlates with the BTPT. While LVP2013-
14b shares a SC of 0.94 with CAT11-07b, LVP2013-14a shares a SC of
0.89 with CAT11-07a, below the 95% lower confidence limit
(Table 3). For this reason we refrain from confidently designating
LVP2013-14 to the BTPT here, but note that the similarities of the
glass color, stratigraphic position and the bimodal nature of the tuff
suggest that it is a diagenetically altered unit of the BTPT.
6.1.5. Unnamed trachytic tuffs of the Nyamita Valley
Two unnamed trachytic tuffs occur in the Nyamita Valley. The
first was originally identified as a distinct mode of glass (CAT09-
02a) from the upper-most tuff at section Nyamita 1 (Fig. 2,Tryon
et al., 2010). Glass shards attributed to CAT09-02a are distinct
based on major element oxide composition but are morphologi-
cally indistinguishable from shards attributed to CAT09-02b from
the same tuff (Tryon et al., 2010). The CAT09-02a trachytic mode
has lower magnesium and calcium contents than seen in any of the
other trachytic tuffs in the eLVB (Table 1). We refrain from naming a
tuff based on this mode, but we present SCs to show that it is
distinct from all other tephras in this study. Randomization of mean
element oxides from CAT09-02a produced an average SC value of
0.95 and a lower 95% confidence limit of 0.93 (Table 2).
A similar trachytic tuff was also identified in isolation
(AV1004T5A) at another section, AV1004, further north in the
Nyamita Valley (Van Plantinga, 2011,Fig. 2). At section AV1004 this
tuff is stratigraphically below samples of the Nyamita Tuff (Van
Plantinga, 2011,Figs. 2 and 7). SC values calculated between
AV1004T5A and the 5000 randomized replicates of this sample
produced an average SC value of 0.95 and a lower 95% confidence
limit of 0.92 (Table 2).
The SC between sample AV1004T5A and CAT09-02a is just
outside of the 95% confidence limit at 0.91 while SCs between
either of these samples and every other tuff reported here is less
than 0.86. CAT09-02a has a mean FeO of 9.39 wt. %, which is greater
than 2 standard deviations above the mean FeO content of
AV1004T5A (8.05 ±0.14). Because of these differences, we refrain
from diagnosing correlation here, but we note the overall chemical
similarity of these tuffs and suggest they may represent tephras
from a similar source, or perhaps one or more phases of the
eruption leading to the Nyamita Tuff not found elsewhere. If more
samples of similar trachytic glass are found, these samples may also
prove to be end-members of a tuff with variable iron content, as is
the case with the Nyamsingula Tuff.
6.1.6. Unnamed rhyolitic tuff
A rhyolitic tuff is known from a single sample, LVP2013-05, from
near the base of section ZTG at Kisaaka, Karungu (Fig. 4). At this
section LVP2013-05 forms a ~4.5-m-long, 0.17-m-thick lens of fresh,
light-grey, vitric ash ~1 m above the base of section and ~1.75 m
below the Nyamita Tuff. Sample LVP2013-05 is a fresh, vitric tephra
deposit consisting of fine silt to clay-sized sediment dominated by
abundant gray glass shards of <5
m
me20
m
m. This is the only
rhyolitic tephra deposit in our sample, categorically distinct from all
other trachytic and phonolitic samples in this study (Fig. 7). SC
values calculated between LVP2013-05 and the 5000 randomized
replicates of this sample produced an average SC value of 0.85 and a
lower 95% confidence limit of 0.76 (Table 2).
Rhyolitic sample LVP2013-05 from Kisaaka is compositionally
similar to a sample of an obsidian source, MER 10, from Ololbutot 2
Oserion Farm, south of Lake Naivasha (SC ¼0.83 with LVP2013-05;
see Table 3;Fig. 1;Brown et al., 2013). The observed SC, although
low because of high variation in rhyolites, is within the range of
expected values for replicates of LVP2013-05 (Table 3) and all
means of all major element oxides overlap at ±1 standard deviation
providing confident correlation (Brown and Nash, 2014). Analysis of
the MER 10 obsidian sample, like all tuff samples analyzed for this
N. Blegen et al. / Quaternary Science Reviews 122 (2015) 89e111106

study including LVP2013-05, were done on the same microprobe at
the University of Utah using the same standards and protocols
(Brown et al., 2013). As obsidians are effusive volcanics found close
to their vent source, this correlation suggests that at least some of
the tephras found at or near the base of the eLVB section originated
from sources near Lake Naivasha in the central Kenyan rift.
6.1.7. A “Nyando Ash”from Songhor
We analyzed a single sample of the “Nyando Ashes”(Pickford,
1984) from the Pleistocene sediments of Songhor (X-5A-3) origi-
nally collected by McBrearty (1981) during her excavations there
(Table 1,Fig. 1bec). In thin section, X-5A-3 is gray silt-sized sedi-
ment dominated by abundant <5
m
me100
m
m glass shards with
acutely angled margins and round to slightly oval vesicles. This tuff
is a phonolite (Fig. 9), similar to the Wakondo and Nyamsingula
Tuffs, but distinguishable from these other phonolites by lower
Al
2
O
3
(~13.75 wt%) and higher FeO (9.2 wt%). This sample does not
correlate with any other tuff known from the eLVB. SC values
calculated between X-5A-3 and the 5000 randomized replicates of
this sample produced average SC value of 0.96 and a lower 95%
confidence SC limit of 0.93 (Table 2). No other sample analyzed here
produced a SC value greater than 0.89 when compared to X-5A-3
(Table 3).
6.2. Optical dating results
The optical dating results for all five samples are presented in
Table 4 and the D
e
distributions are presented in Fig. S3 as radial
plots. Ages of 46 ±4 (RUP-1) and 50 ±4 ka (RUP-2) were calculated
for the samples collected above and below sample CAT09-21
(Fig. 2), providing age constraints for the deposition of the type
sample of the Nyamita Tuff. Both ages are statistically consistent
with each other at 1
s
, and represent our best constraints on the
depositional age of the Nyamita Tuff. The type sample, CAT09-21,
around which the optical dating samples were collected, is a
fresh vitric deposit indicating airfall deposition (Fig. 12) Thus, the
depositional age likely closely approximates the eruptive age of the
Nyamita Tuff.
For the samples from Wakondo, Bovid Hill locality, the ages
range from 64 ±6 (RUP-5) to 74 ±6 ka (RUP-3) for samples from a
channel complex that occurs ~1 m stratigraphically above the type
sample of the WakondoTuff (CAT09-05) (Fig. 2;Jenkins et al., 2012).
All three optical dating ages are statistically consistent with each
other at 1
s
and there is no stratigraphic evidence to suggest that
the sediments from which the samples were collected were
deposited at different times. It is, therefore, our best estimate to
obtain a weighted mean age of 68 ±5 ka for the depositional age of
the Bovid Hill locality channel complex from Wakondo. These age
constraints provide a minimum age of 68 ±5 ka for the deposition
of the underlying Wakondo Tuff.
7. Discussion: succession, age, and transport of the eLVB
tephras
We recognize a single sequence of tephras among Pleistocene
outcrops from Rusinga Island, Mfangano Island, and Karungu,
exposed along a north-south transect of >60 km. From bottom to
top, the sequence includes the WakondoTuff, the Nyamita Tuff, the
Nyamsingula Tuff, and the Bimodal Trachyphonolitic Tuff (BTPT),
with an additional three compositionally unique tephras. Multiple
radiometric dates bound the deposition of these tephras and
intercalated sediments to between ~100 ka and ~35 ka. Although
tuffs may be locally correlated in the field where exposure is good,
geochemical compositional data are required for accurate correla-
tion between discontinuous outcrops and over longer distances
between localities. No sample from Rusinga Island, Mfangano Is-
land, or Karungu correlates with the “Nyando Ashes”from Songhor.
Tryon et al. (2010) hypothesized an origin in the Kenyan Rift
(particularly Longonot and Suswa volcanoes, Fig. 1) for eLVB tephras
based on the absence of known Pleistocene-age volcanoes in Lake
Victoria basin, the fine grain size of the ashes, and the restricted
distribution of trachyte- and particularly phonolite-producing
eruptions in East Africa during the Pleistocene. This hypothesis is
empirically supported by our correlation of the distal rhyolitic ash
sample LVP2013-05 to obsidian vent sources at Oserian Farm south
of Lake Naivasha in the Kenyan Rift (Fig. 1), 250 km east. These
distant sources have important implications for understanding the
depositional mode of the eLVB tephra and interpreting both their
stratigraphic position and age.
The topography of the Mau escarpment on the western edge of
the Kenyan Rift prevents rivers from running east-west between
Fig. 12. Photomicrographs of resin-impregnated thin sections taken in plane light from
samples CAT09-21 (bottom) and CAT09-04 (top) collected from the Nyamita 2 strati-
graphic section in the Nyamita Valley, Rusinga Island, shown in stratigraphic order.
Stratigraphic positions of samples are shown in Figs. 2 and 10c. Basal sample CAT09-21
is fresh vitric tephra composed almost entirely of volcanic glass shards (dark spots on
photomicrograph are voids or bubbles formed during sample preparation) and some
secondary calcite is present. The OSL samples RUS 1 is from directly above CAT09-21
and dated 46 ±4 ka and RUS 2 from directly below CAT09-21 and is dated
50 ±4 ka. This sample is interpreted as a primary fallout tephra that has been, at most,
minimally reworked. Overlying sample CAT09-04 is a reworked and bioturbated
portion of the same deposit, with sparse glass shards set in a silteclay matrix with
epiclastic minerals, lithic fragments and some secondary calcite present. All glass
shards in these samples are attributed to the Nyamita Tuff on the basis of geochemical
composition (CAT09-04 was analyzed on the microprobe as ‘KRU2012-18’).
N. Blegen et al. / Quaternary Science Reviews 122 (2015) 89e111 107

the eLVB and the Rift Valley volcanoes, eliminating the possibility
for long-distance fluvial transport of the tephras, as is the case in
the Lake Turkana Basin (Feibel, 2011). As suggested elsewhere
(Tryon et al., 2010, 2012, 2014; Faith et al., 2015), the Pleistocene
sediments on Rusinga Island, Mfangano Island, and Karungu
formed when Lake Victoria was absent or substantially diminished,
excluding deposition or reworking by large scale lacustrine pro-
cesses. Airfall deposition is thus the most likely method of transit
for these tephras from the Kenyan Rift (or elsewhere) into the eLVB.
Direct evidence for airfall deposition can be seen at Kisaaka, where
the Nyamita Tuff drapes and preserves gilgai topography over
~1 km of lateral exposure (Fig. 11). In the rock record, gilgai
topography is rarely preserved because the granular peds of the A
horizon in Vertisols are easily transported and are often eroded
during the next depositional event, truncating the characteristic
undulations (Caudill et al., 1996; Driese and Mora, 1999; Driese
et al., 2000, 2003). The preservation of the gilgai topography at
Kisaaka demonstrates that airfall deposition of this tephra is pre-
served for at least some of the exposures of the Nyamita Tuff in the
eLVB (Fig. 11).
Although there is evidence to indicate that at least some of the
tephra are airfall deposits, the eLVB tephra were deposited on a
landscape that included small, probably seasonally active channels
(Tryon et al., 2010, 2012, 2014) and springs (Beverly et al., in press).
Some deposits show clear evidence for reworking, including at least
three deposits of the Nyamita Tuff in the Nyamita Valley (Fig. 2),
three deposits of Nyamita Tuff at Kisaaka (Fig. 4) and a single de-
posit at Aringo Section A (Fig. 5). Such deposits are recognized on
the basis of chemically distinct, multimodal populations of glass
shards, a mixture of detrital clasts and volcanic glass, and/or evi-
dence of pedogenesis, as is likely the case for the upper portions of
the Nyamita 2 section (Fig. 2). These processes indicate that the age
of the deposition of some tuffs may differ substantially from the age
of the eruption that produced it. Where reworking occurs, it is most
commonly between a chemically unimodal and lithologically
coherent unit of the Nyamita Tuff and some smaller, strati-
graphically superjacent tuff deposit. These redeposited units are
easily recognized in the field based on their lithology and evidence
for pedogenesis. Chemically, the reworked tephra are multimodal
with one mode belonging to shards of the reworked Nyamita Tuff
and the other modes from the superjacent tuff deposit, usually the
Nyamsingula Tuff (see Figs. 4, 5). At section WPT 210, Aoch Nyasaya,
the BTPT is redeposited ~20 cm up-section in clumped clasts
(Fig. 6).
While some syn- and post-depositional reworking of the ash
occurred, such events are distinguishable from primary deposition
of the tuffs under consideration. For most of the tuffs in this study,
glass populations are chemically homogenous and fresh indicating
rapid burial in an environment where paleosol formation is a
recurrent feature (e.g. Van Plantinga, 2011; Beverly et al., in press).
The documented airfall deposition of tuff units such as those at
Kisaaka, as well as the fairly rapid burial of such tuff units, facilitates
lithostratigraphic correlation between identified stratigraphic units
at temporal scales (~10
1
e10
4
years), the time scales widely
employed in paleoecological and archaeological studies of ancient
landscapes (e.g., Potts et al., 1999; Behrensmeyer et al., 2000).
OSL dates bracketing the fresh vitric ash of the type sample of
the Nyamita Tuff in Rusinga Island's Nyamita Valley indicate an age
of ~49 ka for its deposition there (see Table 5 and supplementary
information). Sedimentary features (e.g., the sample consists
almost entirely of glass shards) suggest that the dated deposit
(CAT09-21) is a primary fall-out tephra deposit that underwent
minimal reworking following deposition, and thus its depositional
age likely approximates eruption age (Fig. 12). We propose that this
49 ka date provides the earliest age of deposition of the Nyamita
Tuff in the Nyamita Valley of Rusinga Island. It also likely reflects an
approximate age for the continuous lateral deposits of the Nyamita
Tuff at Kisaaka, which drape the gilgai topography indicating they
are airfall deposits (Figs. 4 and 11). Additionally, the 49 ka data for
the Nyamita Tuff in the Nyamita Valley also provides a maximum
age for the other deposits of the Nyamita Tuff that were not un-
equivocally deposited via airfall.
Glass shards from the ~2 m of overlying tuffaceous sediment
(sampled in thin section as CAT09-04 and analyzed as KRU2012-18)
overlying the dated type sample (CAT09-21) at Nyamita 2 are also
chemically attributable to the Nyamita Tuff. However, glass from
the upper 2 m at Nyamita 2 occurs in a deposit that has undergone
substantial post-depositional turbation (Fig. 12) indicating a
considerably different depositional history than the underlying ~49
ka deposit sampled as CAT09-21. Gastropods occur in the upper
~2 m of the deposit, and one gastropod from the deposit was dated
using AMS radiocarbon to ~40.5 ka. Based on our field observations,
the gastropods at Nyamita 2 most likely burrowed into the tuff after
deposition, but before lithification. Thus, the minimum age for re-
deposition of the reworked Nyamita Tuff at Nyamita 2 is ~40.5 ka,
or <9 kyr after deposition of the vitric, airfall deposit at the base of
Nyamita 2. Radiocarbon dates on intrusive snail shells at adjacent
Nyamita Tuff outcrops from the Nyamita Valley (Fig. 2) range from
33 to 45 ka (Table 5) indicating reworking of the Nyamita Tuff
throughout the Nyamita Valley occurred before 33 ka.
We interpret the widespread Nyamita Tuff as the most useful
marker bed in the eLVB. Its position near the middle of the teph-
rosequence, its distinctive and relatively homogenous chemical
composition and its OSL age estimate of ~49 ka for its initial
deposition make it well suited as an informal boundary between
upper and lower portions of the eLVB sedimentary sequence
(Fig. 8). However, the age of the sediments above the Nyamita Tuff,
including the Nyamsingula Tuff and BTPT, are poorly constrained.
They postdate the ~49 ka deposition of the Nyamita Tuff, and may
fall into the 33e45 ka range of dates suggested by the gastropod
shells, when snails were actively burrowing into the Nyamita Tuff,
which formed the land surface (or near subsurface) at the time.
Archaeological evidence provides further support for this inferred
age, as all of the tuffaceous eLVB sediments contain only Middle
Stone Age (MSA) artifacts, and no Later Stone Age (LSA) material
has been found (Tryon et al., 2010, 2012, 2014; Faith et al., 2015).
Elsewhere in East Africa, MSA technologies are replaced by LSA
ones during the same 33e45 ka interval (Tryon and Faith, 2013),
and if the sediments were much younger than this we would expect
to have recovered an LSA archaeological component above the
Nyamita Tuff in sediment interbedded with the Nyamsingula Tuff
and BTPT.
Below the Nyamita Tuff, OSL dates of ~68 ka from the Wakondo
locality, collected from sediments above the Wakondo Tuff, provide
a minimum age for the deposition of the Wakondo Tuff (Tables 4
and 5). U-series dates of 94.0 ±3.3 ka and 111.4 ±4.2 ka from
tufa deposits at the base of the sequence at Nyamita (Beverly et al.,
in press;Fig. 2,Table 5) provide a maximum age for the deposition
of the Wakondo Tuff, as well as for the entire sedimentary
sequence. Compared to sediments above the Nyamita Tuff, which
were likely deposited over 16 kyr (i.e. 33e49 ka), the lower
portion of the tephrosequence appears to span a considerably
longer interval, from ~49 kae100 ka.
The tephrostratigraphy presented here coupled with accompa-
nying chronometric dates provides a robust and testable hypothesis
for the depositional age of and correlation between Late Pleistocene
fossil- and artifact-bearing deposits on Rusinga and Mfangano
Islands and near Karungu. The tephrostratigraphic framework
supports our lithostratigraphic correlations between stratigraphic
sections on Rusinga and at Karungu. Additionally, the
N. Blegen et al. / Quaternary Science Reviews 122 (2015) 89e111108

tephrostratigraphy presented here allows the correlation of distant
exposures over ~60 km. The tephrostratigraphic and chronometric
framework presented here represents our hypothesis that dated
and non-dated localities that have been correlated lithostrati-
graphically and/or using tephrostratigraphy are approximately
contemporaneous. The hypothesis that the sedimentary deposits
are contemporaneous can be further tested by constraining the
eruptive age of tephras through a combination of correlation to
more proximal, pumiceous facies and through methods that allow
direct dating of the tephra (i.e.,
40
Ar/
39
Ar).
The tephrostratigraphic sequence presented here thus provides
the initial basis for sampling spatial and temporal variation in
paleoenvironments and hominin behaviors across ancient land-
scapes in the eLVB. Such an approach, with tephrostratigraphic
correlation as one of many integrative tools, has proven highly
successful though decades of research at Early and Middle Pleis-
tocene deposits in the Turkana Basin, at Olorgesailie in Kenya, and
at Olduvai Gorge, Tanzania (Ashleyet al.,1999; Behrensmeyer et al.,
2000; Blumenschine et al., 2008, 2008; Hay, 1976; Rogers et al.,
1994; Stern, 1994; Potts et al., 1999). Our efforts in this direction
are just beginning (Van Plantinga, 2011; Tryon et al., 2014; Faith
et al., in press; Garrett et al., in press), and the tephrostratigraphy
of the region presented here makes an important contribution to-
wards the goal of developing a detailed chronostratiraphic frame-
work of contemporaneous Late Pleistocene sites around the Lake
Victoria basin.
8. Conclusions
Analyses of distal tephras from the eLVB demonstrate the
presence of eight distinct tephras of at least nine chemical com-
positions. Chemical characterization combined with field stratig-
raphy show that four of these tephras correlate over a distance
>60 km: the Wakondo Tuff, the Nyamita Tuff, the Nyamsingula Tuff,
and the Bimodal Trachyphonolitic Tuff (BTPT). Radiometric dates
bound the tephrostratigraphic sequence. The base of the sequence
is ~100 ka, based on U-series dates from tufa deposits that underlie
the entire sedimentary sequence (Beverly et al., in press). These
dates also provide a maximum age for the deposition of the
Wakondo Tuff, and OSL dates of ~68 ka from sediments above the
Wakondo Tuff provide a minimum age for its deposition. The
Nyamita Tuff, which is bounded by bounded by OSL age estimates
of 46 ±4 (RUP-1) and 50 ±4 ka (RUP-2) was likely deposited ~49 ka,
a depositional age that may closely approximate its eruptive age.
The Nyamsingula Tuff and the BTPT, which caps the sequence, were
then deposited in sequence between ~49 and 33 ka, based on
14
C
dates on gastropod shells that post-depositionally burrowed into
the underliying Nyamita Tuff. The upper boundary of these sedi-
ments is poorly dated, but consistent with the available archaeo-
logical data. No Later Stone Age (LSA) artifacts have yet been
recovered, and elsewhere in eastern Africa LSA assemblages appear
~45e30 ka (Tryon and Faith, 2013).
Our correlations among the tephras and the age constraints
provided by a variety of geochronological and other methods,
demonstrate shared depositional sequences among disparate
Pleistocene exposures from two islands in Lake Victoria and mul-
tiple exposures on the Kenyan mainland spanning ~33e100 ka. This
study broadly confirms the initial hypothesis of Pickford (1984) of
widespread Pleistocene tephra deposits in the eLVB and a shared
depositional history for Rusinga Island, Mfangano Island, and Kar-
ungu, but differs substantially in the details, particularly in the
number of tephra present and the need for geochemical composi-
tional data for reliable correlation.
This study further provides the stratigraphic control necessary
for ongoing paleoenvironmental and human behavioral re-
constructions based on fossils, soils, and artifacts from Pleistocene
exposures in the eastern Lake Victoria basin. Furthermore, we
predict that the Wakondo Tuff, Nyamita Tuff, Nyamsingula Tuff, the
BTPT, and other eLVB distal tephra deposits will be found in Pleis-
tocene sediments in other depositional basins in East Africa, sub-
stantially expanding the scale of the work presented here.
Acknowledgments
Fieldwork at Rusinga Island and Karungu was conducted under
research permits NCST/RCD/12B/012/2 issued to NB, NCST/5/002/R/
576 issued to CAT, NCST/RCD/12B/012/31 issued to JTF, NCST/RCD/
12B/01/07 issued to DJP, and NCST/5/002/R/605 issued to EJB, and
an exploration and excavation license issued by the National Mu-
seums of Kenya (NMK). We thank Dr. Francis Brown for assistance,
hospitality and input in the process of data analysis. We would also
like to thank Kirsten Jenkins for her work in assisting with the
Wakondo Bovid Hill OSL data collection. Our fieldwork is made
possible through the support of the NMK and with financial sup-
port from the National Geographic Society Committee for Research
and Exploration (9284-13 and 8762-10), the National Science
Foundation(BCS-1013199, BCS 1013108 and BCS-084 1530), the
Leakey Foundation, the Geological Society of America, the Society
for Sedimentary Geology, the University of Queensland, Baylor
University, the Baylor University Department of Geology Dixon
Fund, New York University's Research Challenge Fund, Harvard
University, and the American School for Prehistoric Research. We
thank the editorial staff at Quaternary Science Reviews as well as
four anonymous reviewers for their time and effort critiquing this
paper. It has been greatly improved as a result.
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.quascirev.2015.04.024.
References
Adams, W.M., 1996. Lakes. In: Adams, W.M., Goudie, A.S., Orme, A.R. (Eds.), The
Physical Geography of Africa. Oxford University Press, Oxford, pp. 122e133 .
Table 5
Age estimates for tuffs discussed in this study.
Tuff Minimum
age (ka)
Method Maximum
age (ka)
Method
BTPT 33
14
C 49 Nyamita Tuff below þtephra correlations.
Nyamsingula 33 BTPT þtephra correlation 49 Nyamita Tuff þtephra correlations.
Nyamita 49 OSL 49 OSL
Nyamita Valley Trachytes 49 OSL of Nyamita Tuff þtepha correlations 94e111 UeTh
Rhyolitic tuff 49 OSL of Nyamita Tuff þtephra correlation 94e111? presumed stratigraphic position above the Wakondo Tuff?
Wakondo 68 OSL 94e111 UeTh on tufa from Nyamita Valley.
Wakondo 100 Compositional similarity to Suswa phonolites.
N. Blegen et al. / Quaternary Science Reviews 122 (2015) 89e111 109

Anderson, G.D., Talbot, L.M., 1965. Soil factors affecting the distribution of the
grassland types and their utilization by wild animals on the Serengeti Plains,
Tanganyika. J. Ecol. 53, 33e56.
Ashley, G.M., Tactikos, J.C., Owen, R.B., 2009. Hominin use of springs and wetlands:
paleoclimate and archaeological records from Olduvai Gorge (~1.79e1.74 Ma).
Palaeogeogr. Palaeoclimatol. Palaeoecol. 272, 1e16.
Avery, G., Cruz-Uribe, K., Goldberg, P., Grine, F.E., Klein, R.G., Lenardi, M.J., ,
et al.Wilson, M.L., 1997. The 1992e1993 excavations at the Die Kelders middle
and later stone age cave site, South Africa. J. Field Archaeol. 24, 263e291.
Behrensmeyer, A.K., Kidwell, S.M., Gastaldo, R.A., 2000. Taphonomy and paleobi-
ology. In: Erwin, D.H., Wing, S.L. (Eds.), Deep Time: Paleobiology's Perspective,
Supplement to Paleobiology, vol. 26. Paleontological Society, Lawrence, KS.
Beverly, E.J., Driese, S.G., Peppe, D., Johnson, C.R., Michel, L.A., Faith, J.T., Tryon, C.A.,
Sharp, W.D., 2015a. Recurrent spring-fed rivers in a middle to Late Pleistocene
semi-arid grassland: Implications for environments of early humans in the Lake
Victoria Basin, Kenya. Sedimentology. 10.1111/sed.12199 (in press).
Beverly, E.J., Driese, S.G., Peppe, D.J., Arellano, L.N., Blegen, N., Faith, J.T., Tryon, C.A.,
2015b. Reconstruction of a semi-arid Late Pleistocene paleocatena from the
Lake Victoria region, Kenya. Quat. Res. (submitted for publication).
Blome, M.W., Cohen, A.S., Tryon, C.A., Brooks, A.S., Russel, J., 2012. The environ-
mental context for the origins of modern human diversity: a synthesis of
regional variability in African climate 150,000e30,000 years ago. J. Hum. Evol.
62, 563e592.
Blumenschine, R.J., Masao, F.T., Tactikos, J.C., Ebert, J.I., 2008. Effects of distance from
stone source on landscape-scale variation in oldowan artifact assemblages in
the paleo-olduvai Basin, Tanzania. J. Archaeol. Sci. 35, 76e86.
Bootsma, H.A., Hecky, R.E., 2003. A comparative introduction to the biology and
limnology of the African Great Lakes. J. Gt. Lakes Res. 29, 3e18.
Borchardt, G.A., Aruscavage, P.J., Millard Jr., H.T.,1972. Correlation of the Bishop Ash,
a pleistocene marker bed, using instrumental neutron activation analysis.
J. Sediment. Petrol. 42, 301e306.
Bower, J.R.F., Grogan-Porter, P., 1981. Prehistoric Cultures of the Serengeti National
Park, Tanzania. Department of Sociology and Anthropology, Iowa State Uni-
versity, Ames. Papers in Anthropology No. 3.
Bower, J.R., Gifford, D.P., Livingstone, D., 1985. Excavations at the Loiyangalani Site,
Serengeti National Park, Tanzania, pp. 41e56. National Geographic Society
Research Reports, 20.
Brown, F.H., Sarna-Wojcicki, A.M., Meyer, C.E., Haileab, B., 1992. Correlation of
pliocene and pleistocene tephra layers between the Turkana Basin of East Africa
and the Gulf of Aden. Quat. Int. 13, 55e67.
Brown, F.H., Haileab, B., McDougall, I., 2006. Sequence of tuffs between the KBS Tuff
and the Chari Tuff in the Turkana Basin, Kenya and Ethiopia. J. Geol. Soc. Lond.
163, 185e204.
Brown, F.H., Fuller, C.R., 2008. Stratigraphy and tephra of the Kibish Formation,
southwestern Ethiopia. J. Hum. Evol. 55, 366e403.
Brown, F.H., McDougall, I., 2011. Geochronology of the Turkana depression of
northern Kenya and southern Ethiopia. Evol. Anthropol. Issues News Rev. 20,
217e227.
Brown, F.H., McDougall, I., Fleagle, J.G., 2012. Correlation of the KHS Tuff of the
Kibish formation to volcanic ash layers at other sites, and the age of early Homo
sapiens (Omo I and Omo II). J. Hum. Evol. 63, 577e585.
Brown, F.H., Nash, B.P., Fernandez, D.P., Merrick, H.V., Thomas, R.J., 2013.
Geochemical composition of source obsidians from Kenya. J. Archaeol. Sci. 40,
3233e3251.
Brown, F.H., Nash, B.P., 2014. Correlation: Volcanic Ash, Obsidian In: Reference
Module in Earth Systems and Environmental Sciences Treatise on Geochem-
istry, Second ed., vol. 14, pp. 63e80
Caudill, M.R., Driese, S.G., Mora, C.I., 1996. Preservation of a paleo-Vertisol and an
estimate of Late Mississippian paleoprecipitation. J. Sediment. Res. 66, 58e70.
Cerling, T.E., Brown, F.H., Bowman, J.R., 1985. Low-temperature alteration of vol-
canic glass: hydration, Na, K,
18
O and Ar mobility. Chem. Geol. 52, 281e293.
Chorowicz, J., 2005. The east African rift system. J. Afr. Earth Sci. 43, 379e410.
Clark-Balzan, L.A., Candy, I., Schwenninger, J.L., Bouzouggar, A., Blockley, S.,
Nathan, R., Barton, R.N.E., 2012. Coupled U-series and OSL dating of a late
Pleistocene cave sediment sequence, Morocco, North Africa: significance for
constructing palaeolithic chronologies. Quat. Geochronol. 12, 53e64.
Danley, P.D., Husemann, M., Ding, B., DiPietro, L., Beverly, E., Peppe, D.J., 2012. The
impact of the geologic history and the paleoclimate of the African Great Lakes
on the diversification of East African cichlids. Int. J. Evol. Biol. 20. Article ID
574851.
Dawson, J.B., 2008. The Gregory Rift Valley and Neogene-recent Volcanoes of
Northern Tanzania. The Geological Society, London. Memoir No. 33.
Deacon, H.J., 1979. Excavations at boomplaas cave ea sequence through the upper
Pleistocene and Holocene in South Africa. World Archaeol. 10, 241e257.
Driese, S.G., Mora, C.I., Stiles, C.A., Joeckel, R.M., Nordt, L.C., 2000. Mass-balance
reconstruction of a modern Vertisol: implications for interpreting the
geochemistry and burial alteration of paleo-Vertisols. Geoderma 95,
179 e204.
Driese, S.G., Jacobs, J.R., Nordt, L.C., 2003. Comparison of modern and ancient Ver-
tisols developed on limestone in terms of their geochemistry and parent ma-
terial. Sediment. Geol. 157, 49e69.
Faith, J.T., Choiniere, J.N., Tryon, C.A., Peppe, D.J., Fox, D.L., 2011. Taxonomic status
and paleoecology of Rusingoryx atopocranion (Mammalia, Artiodactyla), an
extinct Pleistocene bovid from Rusinga Island, Kenya. Quat. Res. 75, 697e707.
Faith, J.T., Tryon, C.A., Peppe, D.J., Beverly, E.J., Blegen, N., 2014. Biogeographic and
evolutionary implications of an extinct late Pleistocene impala from the Lake
Victoria Basin. J. Mammalian Evol. 21, 213e222.
Faith, J.T., Tryon, C.A., Peppe, D.J., Beverly, E.J., Blegen, N., Blumenthal, S., Chritz, K.,
Driese, S.G., Patterson, D., 2015. Paleoenvironmental context of the middle
stone age record from Karungu, Lake Victoria Basin, Kenya, and its implications
for human and faunal dispersals in East Africa. J. Hum. Evol. 83, 28e45.
Feibel, C.S., 1999. Tephrostratigraphy and geological context in paleoanthropology.
Evol. Anthropol. 8, 87e100.
Feibel, C.S., Brown, F.H., McDougall, I., 1989. Stratigraphic context of fossil hominids
from the Omo Group deposits: northern Turkana Basin, Kenya and Ethiopia.
Am. J. Phys. Anthropol. 78, 595e622.
Feibel, C.S., 2011. A geological history of the Turkana Basin. Evol. Anthropol. Issues
News Rev. 20, 206e216.
Froggatt, P.C., 1992. Standardization of the chemical analysis of tephra deposits.
Report of the ICCT working group. Quat. Int. 13, 93e96.
Garcea, E.A., 2010. South-eastern Mediterranean Peoples between 130,000 and
10,000 Years Ago. Oxbow books.
Garrett, N., Fox, D., McNulty, K., Tryon, C., Faith, J.T., Peppe, D.J., Van Plantinga, A.,
2015. Stable isotope paleoecology of the late Pleistocene middle stone age
humans from the equatorial East Africa, Lake Victoria basin Kenya. J. Hum. Evol.
82, 1e14.
Hay, R.L., 1976. Geology of the Olduvai Gorge: a Study of Sedimentation in a
Semiarid Basin. University of California Press.
Hayward, C., 2012. High spatial resolution electron probe microanalysis of tephra
and melt inclusions without beam-induced chemical modification. Holocene
22, 119e125.
Henshilwood, C.S., d'Errico, F., Yates, R., Jacobs, Z., Tribolo, C., Duller, G.A., ,
et al.Wintle, A.G., 2002. Emergence of modern human behavior: middle stone
age engravings from South Africa. Science 295, 1278e1280.
Hunt, J.B., Hill, P.G., 2001. Tephrological implications of beam size-sample size ef-
fects in electron microprobe analysis of glass shards. J. Quat. Sci. 16, 105e117.
Huntley, D.J., Lamothe, M., 2001. Ubiquity of anomalous fading in K-feldspars and
the measurement and correction for it in optical dating. Can. J. Earth Sci. 38,
1093e1106 .
Jacobs, Z., Duller, G.A., Wintle, A.G., Henshilwood, C.S., 2006. Extending the chro-
nology of deposits at Blombos Cave, South Africa, back to 140ka using optical
dating of single and multiple grains of quartz. J. Hum. Evol. 51, 255e273.
Jacobs, Z., Roberts, R.G., Galbraith, R.F., Deacon, H.J., Grun, R., Mackay, A., Mitchell, P.,
Vogelsang, R., Wadley, L., 2008. Ages for the middle stone age of southern Af-
rica: implications for human behavior and dispersal. Science 322, 733e735.
Jenkins, K., Faith, J.T., Tryon, C., Peppe, D., Nightingale, S., Ogondo, J., Roure, C.,
Driese, S., 2012. New excavations of a late Pleistocene bone bed and associated
MSA artifacts, Rusinga Island, Kenya. In: Abstract for Paleoanthropology Society
Annual Meetings Memphis, Tennessee.
Kendall, R.L., 1969. An ecological history of the Lake Victoria Basin. Ecol. Monogr. 39,
121e176.
Kent, P.E., 1942. The country round the Kavirondo gulf of Victoria Nyanza. Geogr. J.
22e31.
Kuehn, S.C., Foit Jr., F.F., 2006. Correlation of widespread Holocene and Pleistocene
tephra layers from Newberry Volcano, Oregon, USA, using glass compositions
and numerical analysis. Quat. Int. 148, 113e137.
Kuehn, S.C., Froese, D.G., Shane, P.A.R., 2011. The INTAV intercomparison of electron-
beam microanalysis of glass by tephrochronology laboratories: results and
recommendations. Quat. Int. 246, 19e47.
Le Bas, M.J.,1977. Carbonatite-nephelinite Volcanism. John Wiley &Sons, New York.
Le Bas, M.J., Le Maitre, R.W., Streckeisen, A., Zanettin, B.,1986. IUGS Sub-commission
on the systematics of igneous rocks. A chemical classification of volcanic rocks
based on the total alkali-silica diagram. J. Petrol. 27, 745e750.
Li, B., Li, S.H., 2012. A reply to the comments by Thomsen et al. on “Luminescence
dating of K-feldspar from sediments: a protocol without anomalous fading
correction”. Quat. Geochronol. 8, 49e51.
Li, B., Roberts, R.G., Jacobs, Z., Li, S.H., Guo, Y.J., 2015. Construction of a ‘global
standardised growth curve’(gSGC) for infrared stimulated luminescence dating
of K-feldspar. Quat. Geochronol. 27, 119e130 .
Lowe, D.J., 2011. Tephrochronology and its application: a review. Quat. Geochronol.
6, 107e153.
Marean, C.W., Goldberg, P., Avery, G., Grine, F.E., Klein, R.G., 2000. Middle stone age
stratigraphy and excavations at Die Kelders cave 1 (Western Cape Province,
South Africa): the 1992, 1993, and 1995 field seasons. J. Hum. Evol. 38, 7e42.
Marean, C.W., Bar-Matthews, M., Bernatchez, J., Fisher, E., Goldberg, P., Herries, A.I.,
Jacobs, Z., Jeradino, A., Karkanas, P., Minichillo, T., Nilssen, P.J., Thompson, E.,
Watts, I., Williams, H.M., 2007. Early human use of marine resources and
pigment in South Africa during the Middle Pleistocene. Nature 449, 905e908.
Macdonald, R., Navarro, J.M., Upton, B.G.J., Davies, G.R., 1994. Strong compositional
zonation in peralkaline magma: Menengai, Kenya Rift Valley. J. Volcanol. Geo-
therm. Res. 60, 301e325.
McBrearty, S., 1981. Songhor: a middle stone age site in western Kenya. Quaternaria
23, 171e190.
McBrearty, S., 1991. Recent research in western Kenya and its implications for the
status of the Sangoan industry. In: Clark, J.D. (Ed.), Cultural Beginnings: Ap-
proaches to Understanding Early Hominid Lifeways in the African Savanna.
Romisch-Germanisches Zentralmuseum, Forschunginstitut fur Vor-und Fruh-
geschichte, Bonn, pp. 159e176. Monographien 19.
N. Blegen et al. / Quaternary Science Reviews 122 (2015) 89e111110

McBrearty, S., 1992. Sangoan technology and habitat at Simbi. Nyame Akuma 38,
34e40.
McDougall, I., Brown, F.H., Fleagle, J., 2005. Stratigraphic placement and age of
modern humans from Kibish, Ethiopia. Nature 433, 733e736.
McHenry, L.J., Mollel, G.F., Swisher III, C.C., 2008. Compositional and textural cor-
relations between Olduvai Gorge Bed I tephra and volcanic sources in the
Ngorongoro Volcanic Highlands, Tanzania. Quat. Int. 178, 306e319.
McHenry, L.J., 2012. A revised stratigraphic framework for Olduvai Gorge Bed I
based on tuff geochemistry. J. Hum. Evol. 63, 284e299.
Mora, C.I., Driese, S.G., 1999. Palaeoclimatic significance and stable carbon isotopes
of Palaeozoic red bed paleosols, Appalachian Basin, USA and Canada. In:
Thiry, M., Simon-Coinçon, R. (Eds.), Palaeoweathering, Palaeosurfaces and
Related Continental Deposits. International Association of Sedimentologists,
pp. 61e84. Special Publication No. 27.
Morgan VI, G.B., London, D., 1996. Optimizing the electron microprobe analysis of
hydrous alkali aluminosilicate glasses. Am. Mineral. 81, 1176e118 5.
Morgan, L.E., Renne, P.R., 2008. Diachronous dawn of Africa's middle stone Age:
new 40Ar/39Ar ages from the Ethiopian Rift. Geology 36, 967e970.
Nash, W.P., 1992. Analysis of oxygen with the electron microprobe: applications to
hydrated glass and minerals. Am. Mineral. 77, 453e457.
Nicholson, S.E., 1998. Historical fluctuations of Lake Victoria and other lakes in the
northern rift valley of East Africa. In: Lehman, J.T. (Ed.), Environmental Change
and Responses in East African Lakes. Kluwer Academic Publishers, Dordrecht,
pp. 7e35.
Nomenclature, S., 2005. North American stratigraphic code. AAPG Bull. 89 (11),
1547 e1591.
Orton, G.J., 1996. Volcanic environments. In: Reading, H.G. (Ed.), Sedimentary En-
vironments: Processes, Facies and Stratigraphy, Third ed. Blackwell Science,
London, pp. 485e567.
Peppe, D.J., McNulty, K., Harcourt-Smith, W., Dunsworth, H., Cote, S., 2009. Strati-
graphic interpretation of the Kulu Formation Early Miocene, Rusinga Island,
Kenya) and its implication for primate evolution. J. Hum. Evol. 56, 447e461.
Peters, C.R., Blumenschine, R.J., Hay, R.L., Livingstone, D.A., Marean, C.W.,
Harrison, T., Armour-Chelu, M., Andrews, P., Bernor, R., Bonnefille, R.,
Werdelin, L., 2008. Paleoecology of the Serengeti-Mara system. In:
Sinclair, A.R.E., Packer, C., Mduma, S.A.R., Fryxell, J.M. (Eds.), Serengeti III: Hu-
man Impacts on Ecosystem Dynamics. University of Chicago Press, Chicago,
pp. 47e94.
Pickering, R., 1959. The Serengeti Plain. Rec. Geol. Surv. Tanganyika 7, 11e13.
Pickford, M., 1982. The tectonics, volcanics and sediments of the Nyanza Rift Valley,
Kenya. Geomorphology 42, 1e33.
Pickford, M., 1984. Kenya Paleontological Gazetteer, vol. 1. National Museums of
Kenya, pp. 1e294.
Pickford, M., 1986. Sedimentation and fossil preservation in the Nyanza Rift system,
Kenya. Geol. Soc. Lond. Spec. Publ. 25, 345e362.
Pickford, M., Thomas, H., 1984. An aberrant new bovid (Mammalia) in subrecent
deposits from Rusinga island, Kenya. In: Proceedings of the Koninklke Neder-
landse Akademie van Wetenschappen. Series B. Palaeontology, Geology, Physics
and Chemistry, vol. 87, pp. 441e452.
Pickford, M., Senut, B., Poupeau, G., Brown, F.H., Haileab, B., 1991. Correlation of
tephra layers from the western Rift Valley (Uganda) to the Turkana Basin
(Ethiopia/Kenya) and the Gulf of Aden. Stratigraphy 313, 223e229.
Potts, R., Behrensmeyer, A.K., Ditchfield, P., 1999. Paleolandscape variation and Early
Pleistocene hominid activities: members 1 and 7, Olorgesailie Formation,
Kenya. J. Hum. Evol. 37, 747e788.
Pouchou, J.-L., Pichoir, F., 1991. Quantitative analysis of homogeneous or stratified
microvolumes applying the model “PAP.”In: Heinrich, K.F.J., Newbury, D.E.
(Eds.), Electron Microprobe Quantitation. Plenum Press, New York, pp. 31e75.
Riehle, J.R., Ager, T.A., Reger, R.D., Pinney, D.S., Kaufman, D.S., 2008. Stratigraphic
and compositional complexities of the late quaternary Lethe tephra in South-
central Alaska. Quat. Int. 178, 210e228.
Rito, T., Richards, M.B., Fernandes, V., Alshamali, F., Cerny, V., Pereira, L., Soares, P.,
2013. The first modern human dispersals across Africa. PLoS One 8, e80031.
Rogers, M.J., Harris, J.W., Feibel, C.S., 1994. Changing patterns of land use by Plio-
Pleistocene hominids in the Lake Turkana Basin. J. Hum. Evol. 27, 139e158.
Rose, J.I., Usik, V.I., Marks, A.E., Hilbert, Y.H., Galletti, C.S., Parton, A., Geiling, J.M.,
C
erny, V., Morley, M.W., Roberts, R.G., 2011. The Nubian complex of Dhofar,
Oman: an African middle stone age industry in southern Arabia. PLoS One 6,
e28239.
Sahle, Y., Morgan, L.E., Braun, D.R., Atnafu, B., Hutchings, W.K., 2014. Chronological
and behavioral contexts of the earliest middle stone age in the Gademotta
formation, main Ethiopian rift. Quat. Int. 331, 6e19.
Schmid, R., 1981. Descriptive nomenclature of classification of pyroclastic deposits
and fragments: recommendations of the IUGS subcommission on the system-
atics of igneous rocks. Geology 9, 41e43.
Shane, P., Nairn, I.A., Martin, S.B., Smith, V.C., 2008. Compositional heterogeneity in
tephra deposits resulting from the eruption of multiple magma bodies: impli-
cations for tephrochronology. Quat. Int. 178, 44e53.
Simon, J.I., Rene, P.R., Mundil, R., 2008. Implications of pre-eruptive magmatic
histories of zircons for U-Pb geochronology of silicic intrusitions. Earth Planet.
Sci. Lett. 266, 182e194.
Singer, R., Wymer, J., 1982. The Middle Stone Age at Klasies River Mouth in South
Africa. University of Chicago Press, Chicago.
Smith, V.C., Pearce, N.J., Matthews, N.E., Westgate, J.A., Petraglia, M.D., Haslam, M.,
Lane, C.S., Korisettar, R., Pal, J.N., 2011. Geochemical fingerprinting of the
widespread Toba tephra using biotite compositions. Quat. Int. 246 (1), 97e104.
Soares, P., Alshamali, F., Pereira, J.B., Fernandes, V., Silva, N.M., Afonso, C.,
Costa, M.D., Muslov
a, E., Macaulay, V., Richards, M.B.,

Cerný, Pereira, L., 2012.
The expansion of mtDNA haplogroup L3 within and out of Africa. Mol. Biol. Evol.
29, 915e927.
Stern, N., 1994. The implications of time-averaging for reconstructing the land-use
patterns of early tool-using hominids. J. Hum. Evol. 27, 89e
105 .
Thompson, J.C., 2005. The impact of post-depositional processes on bone surface
modification frequencies: a corrective strategy and its application to the
Loiyangalani site, Serengeti Plain, Tanzania. J. Taphon. 3, 67e89.
Thomsen, K.J., Murray, A.S., Jain, M., Botter-Jensen, L., 2008. Laboratory fading rates
of various luminescence signals from feldspar-rich sediment extracts. Radiat.
Meas. 43, 1474e1486.
Tryon, C.A., McBrearty, S., 2002. Tephrostratigraphy and the Acheulian to Middle
Stone Age transition in the Kapthurin formation, Kenya. J. Hum. Evol. 42,
211e235.
Tryon, C.A., McBrearty, S., 2006. Tephrostratigraphy of the Bedded Tuff Member
(Kapthurin Formation, Kenya) and the nature of archaeological change in the
later Middle Pleistocene. Quat. Res. 65, 492e507.
Tryon,