Quaternary alluvial paleosols of the Atbara River, eastern Sudan: description and paleoenvironments [original]

JOURNAL OF QUATERNARY SCIENCE (2024) 39(1) 102–118 ISSN 0267-8179. DOI: 10.1002/jqs.3574

Quaternary alluvial paleosols of the Atbara River, eastern Sudan:

description and paleoenvironments

M. MOHAMMEDNOOR,

1,2,3

* F. BIBI,

A. EISAWI,

3,4

S. TSUKAMOTO

and R. BUSSERT

Technische Universität Berlin, Institute of Applied Geosciences, Berlin, Germany

Museum für Naturkunde, Leibniz Institute for Evolution and Biodiversity Science, Berlin, Germany

Department of Geology, University of Khartoum, Khartoum, Sudan

Al Neelain University, Faculty of Petroleum and Minerals, Khartoum, Sudan

Leibniz Institute for Applied Geophysics, Hannover, Germany

Received 12 May 2023; Revised 15 September 2023; Accepted 16 October 2023

ABSTRACT: Quaternary climatic changes in the Nile Basin and their effects on the evolution of African mammals and

vegetation are poorly understood, particularly for the last 1 Ma. Pleistocene (~230 to <17 ka) alluvial sediments exposed

along the middle stretches of the Atbara River in eastern Sudan are rich in fossil vertebrates and are ideal for

paleoenvironmental reconstruction during this time interval. We performed petrographic, mineralogical and

geochemical analyses on the middle Atbara paleosols to reconstruct the paleoclimate and paleolandscape. We

describe Aridisols/Calcisols characterized by calcretes and containing gypsum and halite, and Vertisols with pedogenic

slickensides and a relatively large amount of smectite. The paleosols indicate that the study area transitioned from an arid

to semi‐arid climate during Marine Isotope Stage (MIS)7/6, to a more humid climate during MIS5, and then a return to

more arid conditions during MIS2. The studied paleosols likely supported a range of grassland and wooded grassland

savanna habitats. Our study confirms that the Atbara River Valley provided favorable living conditions for Pleistocene

large mammal communities including Homo, potentially facilitating dispersals out of Africa through the Nile corridor.

KEYWORDS: Atbara River; calcrete; paleosols; Pleistocene; Sudan

Introduction

The East African region witnessed major climatic changes

during the Quaternary. Due to regional factors such as tectonic

activity in the East African Rift Valley, and global factors such

as the establishment of tropical Walker Circulation and

intensification of northern hemisphere glaciations (Maslin &

Trauth, 2009; Trauth et al., 2021), Plio‐Pleistocene African

environments shifted towards more open conditions with

grassland habitats expanding at the expense of woody cover

(Cerling et al., 2015). Over the last 8 Ma, the currently hyper‐

arid Saharan region has experienced numerous humid

episodes, with some 230 humid intervals identified in oceanic

dust records (Larrasoaña et al., 2013; Skonieczny et al., 2019),

the most recent of which occurred between 11 and 5 ka

(Tierney et al., 2017). During these ‘green Sahara’events,

grassland and shrubland were established across areas that are

today desert (Tierney et al., 2017; Blanchet et al., 2021).

An important region in this regard is the Sahel, which

encompasses Senegal, Mauritania, Mali, Burkina Faso, Niger,

Nigeria, Chad, Sudan and Eritrea. Today this area straddles the

transitional zone between the humid savannas and forests of

central Africa to the south, and the arid savannas and Sahara

Desert to the north. Despite stretching over 5900 km from east to

west, very little information is present on Pleistocene environ-

mental conditions within this large area from terrestrial records.

Furthermore, in the eastern Sahel, the Nile Valley is a unique

hydrographical feature that connects equatorial eastern Africa

with the Mediterranean, forming an important dispersal corridor

for vegetation and animals through the Sahara. The Nile has been

in existence in some form since the Oligocene (Fielding

et al., 2018; Faccenna et al., 2019) but studies revealing

information on paleoenvironmental changes from the terrestrial

sedimentary record within the Nile Basin are scarce, in particular

regarding the last 1 Ma. To close this information gap, it is

important to investigate sediment archives that document long

periods of the Pleistocene and contain fossils and archeological

finds. Such sediment archives exist in eastern Sudan along the

Atbara River, which today has its headwaters in the northern

Ethiopian Plateau and is the last major tributary of the Nile before

it flows through the Sahara. Pleistocene deposits along the Atbara

have produced mammalian fossils and stone tools from the

Acheulean and Middle Stone Age, which indicate that the Atbara

River or its precursors and the surrounding floodplains provided

suitable habitats for terrestrial mammal communities including

hominins (Abbate et al., 2010; Masojćet al., 2019, 2021; Ehlert

et al., 2022). Additionally, these sediments host fossil soils

(paleosols) that provide information on climatic and environ-

mental conditions during deposition.

Here, we report the results of mineralogical, petrographic and

geochemical investigations of the middle Atbara paleosols in

order to reconstruct the Pleistocene paleoenvironments of this

transitional zone between the humid and arid Sahelian belt.

Geological background

The study area is located near Khashm El Girba, at the northern

rim of the Gedaref Basin in eastern Sudan (Fig. 1). The Gedaref

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any

medium, provided the original work is properly cited.

Correspondence: M. Mohammednoor, as above.

E‐mail: [email protected]

Basin is surrounded by Precambrian metavolcanics and metasedi-

ments (Fig. 1b) and is filled with fluvial to deltaic sediments of the

Late Cretaceous Gedaref Formation (Eisawi and Schrank, 2009).

These are partially overlain by Cenozoic basalts (Vail, 1988;

Fig. 1b) and intruded by basaltic sills and dykes.

Quaternary alluvial sediments exposed along the middle

Atbara River comprise sands and gravels representing channel

deposits alternating with silt and clay floodplain deposits. The

sediments contain abundant vertebrate remains and stone

tools. The Pleistocene alluvial sediments were studied by

Abbate et al. (2010) over a 20‐km stretch north of Khashm El

Girba, around Butana Bridge. The authors identified and

named two major sedimentary units: the older Butana Bridge

Synthem (BBS) and the younger Khashm El Girba Synthem

(KGS), separated by a major unconformity. Each of these units

was further divided into three intervals (BBS1–3) or sub-

synthems (KGS1–3) separated by minor unconformities

(Fig. 2a). The BBS1 interval consists of amalgamated gravel

channel deposits, BBS2 mainly of pedogenized silt with

manganese (Mn) and calcrete nodules, and BBS3 of fluvial

sand and pedogenized sandy silt. KGS1 is gravelly, but

predominantly comprises sandy channel fills and floodplain

deposits. KGS2 is very coarse basally, with pebbly sand

overlain by trough cross‐bedded coarse sandy channel fills, the

upper parts of which are composed of ripple cross‐bedded

fine‐grained sand. These sandy channels are interbedded with

brownish fine sandy to silty floodplain paleosols characterized

by slickensides, calcretes, and Fe and Mn nodules. The lower

part of the KGS3 is composed of stacked channel fills of

conglomerate overlain by crudely cross‐bedded, poorly sorted

sand, while the upper part is made up of highly calcareous fine

sand to silty floodplain paleosols.

The sandy bodies in the lower part of the KGS2 and KGS3

are covered by river oyster (Etheria) patch reefs with

stromatolitic coatings. U–Th dating of two Etheria shells by

Abbate et al. (2010) gave ages of 126.1 ±1.0 and 92.2 ±0.7 ka

for the bases of the KGS2 and KGS3, respectively, while

mammalian biochronology in combination with magnetostra-

tigraphy indicated a late Early to early Middle Pleistocene age

for the BBS1 interval (i.e. around or after 800 ka), with the

BBS2–3 spanning most of the Middle Pleistocene, and a

temporal gap of around 100 ka between the BBS and KGS

(Abbate et al., 2010). However, a recent study including

luminescence, radiocarbon and

230

Th/U dating revealed ages

of ~220–160, ~160–130, ~130–30, and ~30 to <17 ka for the

BBS, KGS1, KGS2 and KGS3, respectively (Fig. 2b; Tsukamoto

et al., 2022), which indicates relatively continuous deposition

with no major temporal gaps, and no sediments older than the

late Middle Pleistocene in the BBS–KGS sequence.

Materials and methods

Field methods

Paleosols intercalated with Pleistocene alluvial sediments

(Figs. 2a and 3) along the middle Atbara River were studied

and sampled in the vicinity of Butana Bridge near the village of

Al Sharafa (15°03′00″N, 35°55′59″E; Fig. 1b). From each

sedimentary unit –except the BBS1 interval which lacked

paleosols –all paleosols were noted in the field, and the

paleosol profiles which contained the most developed

pedogenic features were selected for study and sampling.

Detailed paleosol sections were measured after digging

trenches to access sediments unaffected by modern weath-

ering. Paleosols were described following Tabor et al. (2017).

Within each studied paleosol, fresh samples were collected at

regular intervals (every 10 cm) for mineralogical, geochemical

and grain size analyses, which were carried out at the

Department of Applied Geochemistry, Technische Universität

(TU) Berlin. Paleosol color was described using the Munsell

Capsure Color Matching Tool RM200SOIL. Rhizoliths were

classified following Klappa (1980) as root casts and root

petrifications. The former are tubular voids left by decayed

roots and then filled by sediment or cement or both, while the

latter are impregnations or replacements of organic matter by

minerals without total loss of root tissues. The abundance of

calcareous rhizoliths was estimated by counting their number

Figure 1. Maps of (A) the location and (B) the geology of the study area (modified from Tsukamoto et al., 2022) and the sampling position of the

potential parent rocks. [Color figure can be viewed at wileyonlinelibrary.com]

PLEISTOCENE PALEOSOLS OF EASTERN SUDAN 103

in areas 10 cm thick and 100 cm wide, and averaged for each

paleosol horizon. The presence of more than 10 rhizoliths in

such an area was classified as many, between five and 10 as

common, and fewer than five as few. For describing the

abundance of the calcareous nodules, we followed the US Soil

Survey (Soil Survey Staff, 2017); nodules were classified as

many when they formed more than 20% of the exposed

surface, common if they formed 2–20%, and few when they

Figure 2. The BBS (Butana Bridge Synthem) and KGS (Khashm El Girba Synthem) Pleistocene alluvial sediments near the village of Al Sharafa. (A)

Stratigraphic log showing the sampling position of the studied paleosols (see Fig. 5), luminescence dating ages and the position of other paleosols. (B)

Luminescence age model (mean modeled ages with 95% confidence interval), and corresponding marine isotope stages (MIS) showing the formation

time of the studied and other paleosols (modified from Tsukamoto et al., 2022). Abbreviations: cl =clay; si =silt; fs =fine sand; ms =medium sand;

cs =coarse sand; gr =granule; pe =pebble; co =cobble; bo =boulder. [Color figure can be viewed at wileyonlinelibrary.com]

104 JOURNAL OF QUATERNARY SCIENCE

were <2%. Blocks of oriented samples were carefully

extracted from BBS2 paleosol with a trowel and hammer,

and were then wrapped with appropriate protection. In

addition to these oriented samples, calcrete samples (nodules,

crusts and rhizoliths), which were collected from the Butana

Bridge area and other locations in the study area, were

micromorphologically analyzed. To determine the source

rocks of the sediments, samples were collected from three

outcrops south of the studied sections (Fig. 1b): a Precambrian

gneiss from near Khashm El Girba, and two Cenozoic basalts

exposed in the Wadi Turk and Al Showak areas.

For paleosol classification, we used the taxonomy of the Soil

Conservation Service of the USA (Soil Survey Staff, 2022), the

World Reference Base for Soil Resources (WRB, 2022) and the

paleosol classification of Mack et al. (1993). The US soil

taxonomy system classifies soils based on their diagnostic

horizons and features and the WRB classification is based on

diagnostic horizons, properties and materials, which should be

measurable and observable in the field and which are often

similar to the diagnostic horizons and characteristics of the US

system (Blume et al., 2016). Mack et al. (1993) based their

classification on pedogenic features that have the highest

preservation potential. They classified paleosols into nine orders,

four of which (Histosol, Spodosol, Oxisol and Vertisol) were

borrowed from the US classification, with other orders excluded.

Retallack (2019) suggested that modern soil classifications should

be used because they are helpful in understanding paleosols.

For describing the paleosol horizons, we used the soil

nomenclature of the US Department of Agriculture (Soil Survey

Staff, 2022). From top to bottom a typical soil profile consists of O,

A, B and C master horizons. O is the zone where organic matter

has accumulated; A is the leaching zone, which is characterized

by the accumulation of humified organic matter mixed with the

mineral fraction; B is the zone of accumulation, which shows

enrichment in clay, carbonate and sesquioxides; and C comprises

unweathered parent material. The master horizon is described in

detail to identify specific physical or chemical characteristics.

These are denoted by lowercase suffixes: k indicates accumulation

of carbonates, o of sesquioxides, t of clay (argillic) and ss the

presence of pedogenic slickensides.

Laboratory methods

Optical microscopy

The micromorphology of soils is extremely valuable in the

characterization of pedogenic processes and climatic condi-

tions during soil formation (Stoops, 2018). Twelve thin sections

of calcretes and oriented paleosol samples were prepared for

micromorphological analyses. Pedogenic features were iden-

tified according to the guidelines of Wright and Tucker (1991).

Calcrete is an accumulation of calcium carbonate near to the

surface, where the vadose and shallow phreatic groundwaters

become saturated in calcium carbonate, which occurs in soil

profiles, bedrock and sediments in various forms from

powdery to indurated (Wright & Tucker, 1991). Gile et al.

(1966) and Machette (1985) classified pedogenic calcrete

based on its development stages in the soil profile. It starts with

few filaments in the soil (stage I) to scattered nodules (stage II),

then develops over time to dense nodules in a firm to

moderately cemented matrix (stage III). In the more developed

stages (stage IV), thin (<0.2 cm) to moderately thick (1 cm)

laminae form in the upper part of the K horizon (calcium

carbonate‐rich massive layer; Soil Survey Staff, 2022). Over

time, the thickness of laminae exceeds 1 cm (stage V), and in

the late stages of development, multiple generations of

cemented laminae, breccia and pisolites are formed (stage VI).

Carbonate content and grain‐size analysis

The carbonate content was determined using the ‘Karbonat‐

Bombe’method according to Müller and Gastner (1971).

Prior to grain‐size analysis, the authigenic carbonates (cal-

cretes) in the samples were dissolved by using 20% HCl; samples

were not treated for the organic materials, because the paleosols

are poor in organic materials. The grain‐size distribution (sand,

Figure 3. The BBS (Butana Bridge Synthem) and KGS (Khashm El Girba Synthem) Pleistocene alluvial sediments with the intercalated BBS2 and

KGS2 paleosols exposed along the Middle Atbara River near Butana Bridge (bottom: 15°04′39.3″N, 035°57′29.2″E) and village of Al Sharafa (top:

15°04′57.1″N, 035°56′56.4″E). [Color figure can be viewed at wileyonlinelibrary.com]

PLEISTOCENE PALEOSOLS OF EASTERN SUDAN 105

silt, clay) was determined by sieving the samples with a 63‐µm

sieve and subsequent centrifugation and sedimentation analysis

of the silt fraction using an areometer.

Age–depth modeling

The ages of the paleosols were assumed using the quartz and

feldspar luminescence ages of Tsukamoto et al. (2022) (Fig. 2b).

An age–depth model was created for every 10 cm for the

whole 40‐m sediment sequence using the rbacon package

(Age‐Depth Modeling using Bayesian Statistics; Blaauw and

Christen, 2011) using R (R Core Team, 2022). This age

modeling divides a sediment sequence into equally sized

sections and assumes a linear accumulation within each

section. We divided the sequence into 20‐cm sections, and

added boundaries at depths of 4, 19.5 and 34.8 m (36, 20.5

and 5.2 m in the stratigraphic column in Fig. 2a), which are the

boundaries of KGS3/2, KGS2/1 and BBS3/2. However, no

boundary was set between KGS and BBS, where we see little

age difference. The code we used, plus the input and output

data can be found in the Supplementary Data 1–3. The age

range of each paleosol profile was calculated from the mean

modeled age of the upper and lower limits of the paleosols. For

the uppermost paleosol in KGS3, the upper limit age was

calculated by a linear extrapolation of the modeled curve. The

depositional ages of different units (BBS, KGS1, KGS2 and

KGS3) were also calculated using the modeled age.

X‐ray diffraction

The mineralogy of the paleosols depends strongly on the

weathering intensity, and thus on the local paleoclimate. The

most useful indicators here are pedogenic minerals such as

carbonates, evaporites and clay minerals (Beverly et al., 2018).

The bulk mineralogy of 55 paleosol samples was identified

using X‐ray diffraction (XRD) analysis. The samples were first

pulverized in an XRD‐milled McCrone mill for 15 min together

with 10 mL of 95% ethanol. The milled, randomly oriented

samples were then measured in a Bruker D2 diffractometer

from 3°to 80°2θwith a 0.02°step width and 0.5‐s

measurement time per step using Cu‐Kαradiation. The

minerals were determined from the position of the diffracted

peaks using the DIFFRAC.SUITE EVA (Bruker) software with

the PDF‐4 mineral database, in which the sample height is

corrected to the position of quartz (101) at 26.644°2θ.

The clay mineralogy of the paleosols was determined on 25

samples. Clay minerals are chemically reactive and thus

potentially of high paleoenvironmental and local paleoclimatic

significance (e.g. Singer, 1984; Chamley, 1989; Birkeland, 1989).

The clay fraction (<2 µm) was separated from aliquots using

centrifugation. X‐ray diffractograms of smear slides were then

obtained in three runs using a Bruker D2 Phaser and Cu‐Kα

radiation between 3°and 30°2θwith 0.01°step width and 0.5‐s

measurement time per step, first after air drying, second after

ethylene‐glycol solvation in vapor mode and finally after heating at

550 °C for 2 h (Moore and Reynolds, 1997).

For calculating the bulk mineralogy, semi‐quantitative analysis

was conducted based on the pattern's relative height of identified

peaks and the values of intensity Analyte to intensity Corundum

(I/I

cor

) which are available from the PDF‐4 mineral database.

X‐ray fluorescence

Data of the major and minor element composition of the paleosol

and the basement rock samples were analyzed with a Bruker AXS

S8 TIGER X‐ray fluorescence (XRF) device. Pulverized samples

were first heated at 105 °C for 24 h to remove volatiles such as

water, and then samples were heated at 1000 °Cfor3hto

determine the loss on ignition. Then, 0.5 g of the ignited samples

were mixed with 8.5 g of lithium tetraborate as a fusion agent and

then fused at 1200 °C with a Fluxana XRF application solution;

the fused discs were then measured using XRF. Calibration was

performed using 32 standard samples and five validation

standards. The summed masses range between 100.2 and

99.4% with an average of 99.9%. The obtained major and

minor element compositions were molecular weight normalized

to calculate weathering indices and molecular ratios.

The geochemistry of paleosols can provide information on

weathering intensity. Various weathering indices (CIA, CIA‐K, WIP

and CALMAG) were calculated from geochemical analyses

(Table 1). The chemical index of alteration (CIA) of Nesbitt and

Young (1982) is a measure of the weathering of feldspars and their

degradation to clay minerals. Maynard (1992) proposed to

calculate the CIA without potash (CIA‐K) to account for the

influence of potassium metasomatism in paleosols. The weathering

index of Parker (WIP) is another index used to estimate the

weathering potential for silicate rocks where hydrolysis is the main

agent of weathering (Parker, 1970). The calcium–magnesium

weathering index (CALMAG) of Nordt and Driese (2010) is similar

to CIA but uses Mg instead of Na and K in mol% to evaluate

weathering intensity specifically in Vertisols. CIA, CIA‐KandWIP

were used to determine the weathering intensity that the studied

paleosols underwent.

Paleoprecipitation (MAP =mean annual precipitation) was

calculated for non‐calcareous (<5% CaCO

) paleosol samples.

CALMAG was used to calculate the paleoprecipitation for

Vertisols only, while CIA‐K was used for all paleosol types.

Molecular ratios such as base loss, clayeyness, leaching and

salinization can provide additional information on pedogenic

processes (Sheldon et al., 2002; Retallack, 2019). Useful

proxies for assessing the extent of hydrolysis in paleosols are

the sum of the bases (Na +Mg +K+Ca) divided by Al

(Σbases/Al) and Al/Si ‘clayeyness’, as Al accumulates in clay

minerals relative to the silicate parent material (Sheldon

et al., 2002; Retallack, 2019). Soil salinity can be assessed

based on the (Na +K/Al) ratio (Sheldon et al., 2002) and

leaching by determining base loss ratios, where the abundance

of a given base is related to the abundance of Ti (base/Ti)

(Sheldon et al., 2002). Based on the geochemical proxies or

molecular ratios, several climofunctions have been proposed

to calculate paleoprecipitation (MAP) (Table 2).

Results

Field and petrographic investigations

The BBS and KGS units in the study area contain paleosols

(Figs. 2a and 3) which are easily identified in the outcrops by

Table 1. Equations and reference sources of the weathering indices.

Weathering index Equation Reference

CIA Al/(Al +Ca +Na +K)

×100

Nesbitt and

Young, 1982

CIA‐K Al/(Al +Ca +Na) ×100 Maynard, 1992

WIP (Na/0.35 +Mg/0.9 +K/

0.25 +Ca/0.7) ×100

Parker, 1970

CALMAG Al/(Al +Ca +Mg) ×100 Nordt and

Driese, 2010

CIA, chemical index of alteration; CIA‐K, chemical index of alteration

minus potassium; WIP, weathering index of Parker; CALMAG,

calcium–magnesium weathering index.

106 JOURNAL OF QUATERNARY SCIENCE

Loading more pages...