Petrology of the Namib Sand Sea: Long-distance transport and compositional variability in the wind-displaced Orange Delta

Abstract

Sourced as the Nile in distant basaltic rift highlands, the Orange River is the predominant ultimate source of sand for the Namib desert dunes, as proved independently by bulk-petrography, heavy-mineral, pyroxene-chemistry, and U/Pb zircon-age datasets. Additional local entry points of sand do exist at the edges of the desert, and were quantified by comparison with detrital modes and heavy-mineral suites of hinterland-river sediments.

After long-distance fluvial transport, Orange sand is washed by ocean waves and dragged northward by vigorous longshore currents. Under the incessant action of southerly winds, sand is blown inland and carried farther north to accumulate in the Namib erg, a peculiar wind-dominated sediment sink displaced hundreds of kilometres away from the river mouth. And yet changes in sand mineralogy along the way are minor. After a multistep journey of cumulative 3000 km from their source in Lesotho, volcanic rock fragments and pyroxene are found in unchanged abundance as far as the northern edge of the desert. Only locally is volcanic detritus slightly depleted and minor but regular enrichment in quartz and garnet observed, the sole potential effect of prolonged transport or recycling of Tertiary eolianites. Selective comminution of fragile minerals is thus proved unable to substantially modify sand composition in fluvial, coastal, or eolian settings. Mechanical processes have a much greater effect on the morphology of detrital grains, which in Namib dunes appear commonly shaped into nearly perfect spheres. Eolian sorting concentrates denser minerals locally in placer lags, but such effects can be identified and compensated for. This study demonstrates that mechanical breakdown is unable to markedly affect provenance signatures even during long-distance and prolonged multistep transport in high-energy settings. In arid climates, where chemical processes are negligible, high-resolution bulk-petrography and heavy-mineral analyses are thus powerful techniques to quantitatively reconstruct provenance, and to trace sediment sources and dispersal paths over distances up to thousands of kilometres.

Full text

Author's personal copy
Petrology of the Namib Sand Sea: Long-distance transport and compositional
variability in the wind-displaced Orange Delta
Eduardo Garzanti
a,
⁎, Sergio Andò
a,1
, Giovanni Vezzoli
a,1
, Michele Lustrino
b,2
,
Maria Boni
c,3
, Pieter Vermeesch
d,4
a
Laboratorio di Petrografia del Sedimentario, Dipartimento di Scienze Geologiche e Geotecnologie, Università Milano-Bicocca, 20126 Milano, Italy
b
Dipartimento di Scienze della Terra and Istituto di Geologia Ambientale e Geoingegneria (IGAG, CNR), Università di Roma La Sapienza, 00185 Roma, Italy
c
Dipartimento di Scienze della Terra, Università di Napoli, 80134 Napoli, Italy
d
School of Earth Sciences, Birkbeck, University of London, London WC1E 7HX, UK
abstractarticle info
Article history:
Received 8 April 2011
Accepted 20 February 2012
Available online 3 March 2012
Keywords:
Provenance analysis
Mechanical abrasion
Grain roundness
Aeolian sorting
Pyroxene chemistry
U–Pb zircon ages
Sourced as the Nile in distant basaltic rift highlands, the Orange River is the predominant ultimate source of
sand for the Namib Desert dunes, as proved independently by bulk-petrography, heavy-mineral, pyroxene-
chemistry, and U/Pb zircon-age datasets. Additional local entry points of sand do exist at the edges of the
desert, and were quantified by comparison with detrital modes and heavy-mineral suites of hinterland-
river sediments.
After long-distance fluvial transport, Orange sand is washed by ocean waves and dragged northwards by vig-
orous longshore currents. Under the incessant action of southerly winds, sand is blown inland and carried far-
ther north to accumulate in the Namib erg, a peculiar wind-dominated sediment sink displaced hundreds of
kilometres away from the river mouth. And yet changes in sand mineralogy along the way are minor. After a
multistep journey of cumulative 3000 km from their source in Lesotho, volcanic rock fragments and pyroxene
are found in unchanged abundance as far as the northern edge of the desert. Only locally is volcanic detritus
slightly depleted and minor but regular enrichment in quartz and garnet is observed, the sole potential effect
of prolonged transport or recycling of Tertiary aeolianites. Selective comminution of fragile minerals is thus
proved unable to substantially modify sand composition in fluvial, coastal, or aeolian settings. Mechanical
processes have a much greater effect on the morphology of detrital grains, which in Namib dunes appear
commonly shaped into nearly perfect spheres. Aeolian sorting concentrates denser minerals locally in placer
lags, but such effects can be identified and compensated for. This study demonstrates that mechanical break-
down is unable to markedly affect provenance signatures even during long-distance and prolonged multistep
transport in high-energy settings. In arid climates, where chemical processes are negligible, high-resolution
bulk-petrography and heavy-mineral analyses are thus powerful techniques to quantitatively reconstruct
provenance, and to trace sediment sources and dispersal paths over distances up to thousands of kilometres.
© 2012 Elsevier B.V. All rights reserved.
Contents
1. Introduction .............................................................. 174
2. The Orange River and the Namib Sand Sea ............................................... 174
2.1. The Orange River ........................................................ 174
2.2. Northward sediment dispersal .................................................. 176
2.3. The Namib Sand Sea ...................................................... 176
2.4. Geology of southwestern Namibia ................................................ 176
Earth-Science Reviews 112 (2012) 173–189
⁎Corresponding author. Tel.: +39 02 64482088; fax: + 39 02 64482073.
E-mail addresses: eduardo.garzanti@unimib.it (E. Garzanti), sergio.ando@unimib.it (S. Andò), giovanni.vezzoli@unimib.it (G. Vezzoli), michele.lustrino@uniroma1.it
(M. Lustrino), boni@unina.it (M. Boni), p.vermeesch@ucl.ac.uk (P. Vermeesch).
1
Tel.: +39 02 64482088; fax: +39 02 64482073.
2
Tel.: +39 06 49914158; fax: +39 06 4454729.
3
Tel.: +39 081 253 5068; fax: + 39 081 2535070.
4
Tel.: +44 020 76792418; fax: + 44 020 76792867.
0012-8252/$ –see front matter © 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.earscirev.2012.02.008
Contents lists available at SciVerse ScienceDirect
Earth-Science Reviews
journal homepage: www.elsevier.com/locate/earscirev

Author's personal copy
3. Sand petrology and mineralogy ..................................................... 177
3.1. Sampling and analytical methods ................................................. 177
3.2. Orange River .......................................................... 177
3.3. Hinterland rivers ........................................................ 178
3.4. Namib dunes .......................................................... 180
4. Provenance of Namib dune sand .................................................... 182
4.1. Contributions from hinterland rivers ............................................... 183
4.2. Recycling from Tertiary aeolianites ................................................ 183
5. Physical processes controlling compositional variability of dune sand ................................... 183
5.1. Aeolian sorting and placer formation ............................................... 184
5.2. Mechanical durability of diverse detrital components ....................................... 184
5.3. Rounding by aeolian abrasion .................................................. 185
6. Sediment budget ............................................................ 186
6.1. Quantitative provenance assessment ............................................... 186
7. Conclusion ............................................................... 187
Acknowledgments .............................................................. 187
Appendix A. Supplementary material ................................................... 187
References ................................................................. 187
“The river”, he reminded her. “Tell me about the river”, and listened
avidly as she went on. “The river gathers them up, from every little
pocket and crevice along its course. It picks up those that were
flung into the air during the volcanic eruptions at the beginning of
the continent's existence. For hundreds of million years it has been
carrying the diamonds toward the coast. Those beaches are so rich
in diamonds that they are the forbidden area, the Spieregebied.”
Wilbur Smith, Power of the sword, p.78
1. Introduction
The Orange has been a powerful river possibly since the Jurassic
(Cox, 1989). Or at least it was, until man tamed it with dams to ex-
ploit its waters for agricultural and industrial purposes. And extreme-
ly energetic has long been the microtidal South Atlantic shore where
it debouches, subjected to vigorous waves and northward longshore
drift fuelled by persistent southerly winds (Bluck et al., 2007). As
well as an open-air diamond mine so rich that it is forbidden land
(Jacob et al., 2006), the Orange Delta is thus widely recognised as a
wave-dominated end-member (Spaggiari et al., 2006). But there is
more. Incessantly dragged northwards by longshore currents and
eventually blown inland under the long-lived wind regime, Orange
Sand has been accumulating since the Miocene at least in the Namib
Sand Sea, a sediment sink detached from the river mouth and dis-
placed on land several hundreds of kilometres away (Fig. 1;Rogers,
1977; Vermeesch et al., 2010).
In such a geological setting, changes in sediment composition and
grain morphology caused by multistep transport can be monitored
over a cumulative distance summing up to 3000 km. Whereas abra-
sion and mechanical breakdown have long been established to be
scarcely effective during long-distance fluvial transport by both stud-
ies of natural river systems and laboratory experiments (Russell and
Taylor, 1937; Kuenen, 1959), their incidence during wind transport
has been debated for a full century, and still remains controversial
(Twenhofel, 1945; Goudie and Watson, 1981). Laboratory experi-
ments have shown that aeolian abrasion is 100–1000 times more
effective than fluvial abrasion (Kuenen, 1960), and the generation of
quartz-rich sand through selective breakdown of feldspar by aeolian
impacts has received theoretical support (Dutta et al., 1993). Studies
in natural environments, however, have produced ambiguous results
(Johnsson, 1993), because detrital sources are generally multiple and
undetermined, and dune sand commonly consists of quartz whose
rounded shape may have resulted from aeolian abrasion as well as
from recycling of rounded grains from older sandstones (Garzanti
et al., 2003; Muhs, 2004; Mehring and McBride, 2007).
Understanding aeolian effects is crucial to correctly interpret
desert environments of the past, which is in turn essential for accurate
palaeogeographic and palaeoclimatic reconstructions (Dott, 2003;
Avigad et al., 2005). The peculiarity of the Namib Sand Sea, supplied
from fundamentally one single entry point with a wide spectrum of
detrital species including pyroxene and volcanic rock fragments, offers
a unique opportunity to solve such a thorny petrological problem,
provided we succeed in quantifying subsidiary detrital sources
(hinterland rivers, recycled Tertiary sediments, deflation areas along
the coast; Lancaster and Ollier, 1983; Besler, 1984)andlocaleffects
of wind-induced sorting.
The aims of the present study are to illustrate in detail the composi-
tional variability of Namib dune sands through high-resolution bulk-
petrography and heavy-mineral analysis, to pin-point all sediment
sources and quantify their relative contributions to various parts of the
erg, to evaluate wind-induced concentrations of heavy minerals, and to
assess morphological and compositional changes caused by mechanical
abrasion and selective breakdown of detrital grains with variable
durability during aeolian transport. Quantitative provenance analysis
represents an effective way to identify long-term transport paths and
sediment modifications in modern deserts, and represents a key require-
ment for reconstructing palaeowind patterns and palaeoclimate changes
during the geological evolution of such dynamic geomorphic systems.
2. The Orange River and the Namib Sand Sea
2.1. The Orange River
The Orange is one of the largest African rivers, with a drainage
area of ~970,000 km
2
(Fig. 2). Sourced not far from the Indian
Ocean in the Drakensberg mountains of Lesotho (maximum elevation
3482 m a.s.l.), it flows westwards for ~ 2200 km across an increasingly
arid interior plateau towards the Atlantic Ocean (Moore et al., 2009).
The catchment receives primarily summer rainfall (October to April),
varying from 1200 mm/yr in relatively cold Lesotho highlands (mean
temperatures −5 °C in winter and 16 °C in summer) to 40 mm/yr
near the mouth (Swanevelder, 1981). Rainfall and snowmelt in
Lesotho, occupying only 3% of the catchment, contributes 47% of
total water flow (Makhoalibe, 1999).
This ancient drainage system, originated in the Early Jurassic as a
result of ~2 km surface uplift associated with emplacement of
Drakensberg lavas during rifting from Antarctica, cuts antecedently
in its lowermost tract across the regional uplift formed during
Lower Cretaceous rifting of the South Atlantic (Cox, 1989; Moore
and Blenkinsop, 2002). The oldest basement rocks are widely exposed
north of the Vaal River (Neoarchean metavolcanic Ventersdorp
174 E. Garzanti et al. / Earth-Science Reviews 112 (2012) 173–189

Author's personal copy
Supergroup and Palaeoproterozoic metasedimentary Transvaal
Supergroup; de Villiers et al., 2000). The lower Orange downstream
of the Vaal confluence cuts across Mesoproterozoic metasedimentary
and metavolcanic rocks of the Namaqua Belt (Becker et al., 2006),
whereas the upper Orange and its two main branches (Caledon
and Senqu Rivers) drain the up to 10 km thick Karoo Supergroup
exclusively.
The Karoo succession includes Upper Carboniferous to lowermost
Permian feldspathoquartzose glacial sediments (Dwyka Group), over-
lain by shale, turbiditic to fluviodeltaic sandstone and coal deposited
at wetter lower latitudes (Ecca Group), and by uppermost Permian
to Anisian floodplain deposits accumulated in warmer and drier
climates (Beaufort Group). A retroarc foreland-basin setting is indi-
cated by Ecca and lower Beaufort quartzofeldspatholithic volcaniclas-
tic sandstones, fed from a partly dissected continental arc located in
the south, and overlain by feldspathoquartzolithic to feldspatholitho-
quartzose sandstones of the upper Beaufort Group (Johnson, 1991).
Above a major unconformity, lithoquartzose braidplain sandstones
were deposited in the Carnian–Norian (Molteno Fm.), followed by
red floodplain mudstones (Elliot Fm.), and by Lower Jurassic aeolian
feldspathoquartzose sandstones (Clarens Fm.; Johnson et al., 1996;
Catuneanu et al., 2005). The succession is capped by up to 1.8 km
thick Drakensberg continental flood basalts (183 ±2 Ma; Jourdan et
al., 2005). A vast network of dolerite dykes and sills suggests that
tholeiitic lavas originally covered an area of ~2.5 10
6
km
2
. Also wide-
spread are pipe-like bodies, including diamond-bearing kimberlites,
intruded mainly during Jurassic to Cretaceous break-up of Gondwana
(Moore et al., 2008).
Most of the sediment flux is generated by high orographic rainfall
and topographic relief in Lesotho highlands. Suspended load is mainly
supplied by erodible upper Karoo mudrocks, as reflected by the
chemical composition of river muds (Bremner et al., 1990 p. 251;
Compton and Maake, 2007 p. 344–345), whereas Drakensberg basalts
supply little silt (Haskins and Bell, 1995). Suspended-sediment
yields can thus be locally as high as 2000 tons/km
2
·yr in the upper
Caledon catchment largely draining the Elliot Formation, but as
low as 10–70 tons/km
2
·yr in the upper Senqu catchment draining
basaltic rocks (Makhoalibe, 1984). Various tributaries draining
Karoo sediments owe their name to their turbid coloured waters
(e.g., in Afrikaans Vaal = pale, grey; Modder = mud), but not the
Orange itself, named in honour of William of Orange.
After construction of major dams on the Vaal (Vaal Dam in 1938;
Bloemhof Dam in 1970) and upper Orange (Gariep Dam in 1972;
Fig. 2. Geological sketch map of Namibia, South Africa and Lesotho (compiled after Schlüter, 2006, and other sources cited in text). Inset shows the location of studied aeolian dune
and fluvial samples around the Namib erg. Major dams are also indicated: B = Bloemhof, G = Gariep, H = Hardap, K = Katse, Va = Vaal, Vk = Vanderkloof, W = Welbedacht.
(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 1. The Namib Sand Sea: the displaced sediment sink of Orange River sediments.
175E. Garzanti et al. / Earth-Science Reviews 112 (2012) 173–189

Author's personal copy
Vanderkloof Dam in 1977), total Orange suspended-load flux at the
mouth has progressively dropped from ~90·10
6
tons/yr to the
present b17·10
6
tons/yr (Rooseboom and Harmse, 1979). Annual
water discharge dropped from ~11.5 km
3
to ~5.7 km
3
, but peaks of
26 km
3
are still reached in flood years (maximum discharge rate
8072 m
3
/s, 2nd March 1988; Bremner et al., 1990 p. 262). Peak
suspended-sediment concentrations were drastically reduced from
extreme values in early times (up to 63 g/l for the Caledon and
40 g/l for the Orange, second only to the Yellow River in China) to
7.4 g/l in the 1988 flood, when sediment load remained significant
(81·10
6
tons in March to May 1988) but dominantly derived from
bank erosion and river-bed scour downstream of major dams
(Bremner et al., 1990 p. 247, 285). Suspended load consisted of
~20% sand, 60% silt and 20% clay in pre-1970 times, and of ~15%
sand, 30% silt and 55% clay (25±13% montmorillonite, 63± 13% illite,
12± 3% kaolinite) during the 1988 flood (Bremner et al., 1990 p. 273).
Storage loss due to rapid sedimentation in the Welbedacht and
Gariep reservoirs (Jordaan and Clark, 1988; Sawadogo, 2008) points
to total sediment yields up to 1000–1100 tons/km
2
·yr in both
Caledon and Senqu catchments, corresponding to erosion rates up
to 0.4 mm/yr. Total annual load of the upper Orange could thus
reach 60·10
6
tons. If 49± 7 10
6
tons are accounted for by suspended
load (Rooseboom and Harmse, 1979), then bedload represents ≥10%
of total load in Lesotho, significantly more than the ~5% estimated for
the whole catchment (Bremner et al., 1990 p. 260).
Sediment yields of several hundred tons/km
2
·yr are reported
even from lower-relief subcatchments draining Karoo sedimentary
bedrock, where development of badlands and gullies was enhanced
by extensive grazing activities in the last century (Rooseboom and
Lotriet, 1992). Sediment yields are estimated to be significantly
lower for the low-relief Vaal catchment (~150 tons/km
2
·yr in
1932–1939), and very minor for the arid lower Orange catchment
downstream of the Vaal confluence (Rooseboom and Harmse, 1979
p. 463–462).
Cosmogenic
10
Be and
26
Al measurements in quartz carried by the
Orange River point to catchment-wide erosion rates of only 0.004 ±
0.001 mm/yr averaged over millennial time scales (Vermeesch et al.,
2010). Assuming an average surface rock density of 2.7 g/cm
3
, this
corresponds to sediment fluxes of only 11 ± 2 · 10
6
tons/yr and sedi-
ment yields of 11 ±2 tons/km
2
·yr for the entire Orange catchment,
much lower than recorded through the last century. As for the Nile
(Garzanti et al., 2006), this may largely reflect human-induced
accelerated erosion caused by intensive land use.
2.2. Northward sediment dispersal
The Orange River mouth is an extreme example of a wave-
dominated delta. Sand and gravel delivered to the South Atlantic
coast are carried northwards by a powerful longshore drift under pre-
vailing southerly winds. High-energy conditions persist throughout
most of the year along the coast and adjacent shelf (wave heights
90% in the 0.75–3.25 m range, average 1.5 m for summer and
1.75 m for winter). Tidal range is 1.8 m along this microtidal coast,
where the Orange River is the sole significant source of sediment
(Rogers and Rau, 2006).
Whereas mud moves largely north and west out to the continental
shelf edge, where it forms extensive mud sheets (Spaggiari et al.,
2006), much of the sand is retained within the breaker zone and
moves alongshore in a ≤3 km wide belt. Gravel spits and barrier bea-
ches predominate in proximal settings, passing northwards to linear
beaches and finally to pocket beaches. Gravel beaches, found as far
as 300 km north of the Orange mouth, include the richest diamond
placers known (Corbett and Burrell, 2001; Jacob et al., 2006).
At a number of localities sand is blown off the beaches onto the
land, largely bypasses the Sperrgebiet deflation area (Corbett, 1993),
and accumulates north of Lüderitz, where coastline orientation
changes from northwest to north. Offshore transport of Orange
River sand at subtidal to intertidal depths continues beyond Walvis
Bay (Bluck et al., 2007). Dune sand in the northwestern corner of
the Sand Sea may thus have been blown inland directly from the
Atlantic coast. Sands further to the east have travelled overland for
longer distances. Wind energy and sand-transport rates decrease
landwards (Lancaster, 1985).
2.3. The Namib Sand Sea
The Namib Desert, part of the passive margin originated during
Early Cretaceous rifting of the South Atlantic (Goudie and Eckardt,
1999), stretches for nearly 600 km along the SW African coast
between Lüderitz (~27°S) and the Kuiseb River (23°S), and for
100–150 km inland to the base of the Great Escarpment at the
1000 m contour. Southerly to southwesterly trade winds dominate
the atmospheric circulation in all but the northeasternmost parts of
the desert, although northeasterly winds are also common in winter
(Lancaster, 1985). Climate is hyperarid, with precipitation ranging
from b50 mm/yr and mostly in the form of fog near the coast, to
~100 mm/yr at the foot of the Great Escarpment (Lancaster, 2002).
Ephemeral rivers draining the Escarpment and the semiarid edge of
the southern African plateau (rainfall 200–450 mm/yr) penetrate
≤50 km into the erg, to empty their sediment-laden waters in flat
interdune playas named “vleis”. Floods may last a few weeks, after
which the river beds are dry for the rest of the year (Jacobson et al.,
1995). Only exceptionally high floods of the Kuiseb River reach the
sea at the northern edge of the Namib (16 times since 1837; Morin
et al., 2009).
The sand sea (~ 34,000 km
2
) is dominated by large linear dunes,
with areas of star-shaped dunes on its eastern margin and a belt of
simple and compound transverse and barchanoid dunes along the
coast. Optical dating in the northern Namib indicates that linear
dunes are younger (~ 5.7 ka; Bristow et al., 2007) than big star
dunes (43±10 ka; C. Bristow pers. comm. 2009). Considering that
the present estimated rate of sand input to the Namib Sand Sea is
400,000 m
3
/a (Lancaster, 1989), ≥1 Ma would be required to form
the modern erg (estimated at 375 to 1020 km
3
). This is consistent
with an average residence time ≥1 Ma for sand grains in the coastal
Namib, as determined by cosmogenic-nuclide measurements
(Vermeesch et al., 2010).
2.4. Geology of southwestern Namibia
Namibia is crossed by the Mesoproterozoic Namaqua (“Kibaran”)
and Neoproterozoic Damara (“Pan-African”) orogenic belts, welding
the Archean Kalahari and Congo Cratons (Fig. 2;Jacobs et al., 2008;
Miller, 2008). Exposed along the lower Orange to the southern
margin of the Namib erg (Aus–Lüderitz area) is the Namaqua Meta-
morphic Complex, including medium to high-grade metasediments
interpreted as originally deposited on a passive margin (Becker et
al., 2006). Exposed east of the Namib erg is the Mesoproterozoic
Sinclair Group, largely overprinted by Damaran structures in the
Rehoboth area. These low-grade volcanic and subordinate sedimenta-
ry rocks rest disconformably on medium-grade metasedimentary and
metavolcanic rocks intruded by 1.37 Ga tonalite, and may have
represented the active margin of the Kalahari Craton facing an east-
directed oceanic subduction zone. The widespread intrusion of
granitoid batholiths was followed by outpouring of continental
tholeiites at ~1.1 Ga (Becker et al., 2006).
The Damara Belt exposed at the northern margin of the Namib erg
includes 2.0 to 1.2 Ga basement gneisses overlain by Neoproterozoic
metasediments. Peak metamorphic conditions, reaching granulite-
facies and partial melting north of Walvis Bay, were attained at
525–504 Ma. Widespread are 560–470 Ma old granites and subordi-
nate diorites/granodiorites (Miller, 1983; Jung and Mezger, 2003).
176 E. Garzanti et al. / Earth-Science Reviews 112 (2012) 173–189

Author's personal copy
Precambrian fold belts are unconformably overlain by the 3 km-thick
shallow-marine to fluvial Nama Group, a Neoproterozoic to Cambrian
succession including quartzose to micaceous sandstones, mudrocks,
and interbedded limestones deposited in the foreland basin of the
Damara orogen (Saylor et al., 1995; Geyer, 2005; Blanco et al.,
2009). Deformation increases towards the Damara belt, and very-
low metamorphic grade is reached in slates and phyllites of the
Naukluft mountains to the north (Ahrendt et al., 1978). Neoprotero-
zoic successions locally hosting volcanic rocks are overprinted by up
to lower amphibolite-facies metamorphism in the Gariep Belt to the
south (Frimmel and Frank, 1998).
Younger rocks include Upper Palaeozoic Karoo terrigenous
sediments, the Lower Jurassic Kalkrand basalts, and Upper Cretaceous
kimberlite pipes (Smith et al., 1993; Stollhofen et al., 2000; Davies
et al., 2001).
3. Sand petrology and mineralogy
3.1. Sampling and analytical methods
Twelve, fine to medium-grained dune-sand samples of 1 kg each
were collected in summer 2007 along three E–W transects, five
along the northern edge of the Namib Desert south of Kuiseb River
canyon (NAM 1–5), three in the Sossusvlei area (NAM 6–8), and
four along the southern edge of the desert between Aus and Lüderitz
(NAM 9–12; Fig. 2). Because dark-red to black patches where heavy
minerals (pyroxene, garnet, magnetite) are concentrated by wind
turbulence commonly occur on dune flanks, all samples were collect-
ed on the crest of the largest dunes. Four Orange River sands, sampled
from the mouth (NAM 13), to ~80 (S4037) and ~100 km upstream
(NAM 14), include one Quaternary terrace (S4038). In order to assess
the petrographic and mineralogical signatures of all potential sources
of Namib dunes, sand samples were also collected from eleven hinter-
land rivers across SW Namibia in July 2009.
Quartered aliquots of bulk-sand samples were impregnated with
Araldite, cut into standard thin sections, stained with alizarine red
to distinguish dolomite and calcite, and analysed by counting 400
points under the petrographic microscope (Gazzi–Dickinson method;
Ingersoll et al., 1984). From the 32–500 μm fraction of each sample,
treated with sodium ditionite–citrate–bicarbonate to remove Fe-
oxide coatings, heavy minerals were separated by centrifuging in so-
dium metatungstate (density ~2.90 g/cm
3
), and recovered by partial
freezing with liquid nitrogen; 200 to 250 transparent heavy minerals
were counted in grain mounts (“area method”;Mange and Maurer,
1992). The reasons for choosing a 4ϕ-wide size-window for heavy-
mineral analysis are discussed in Appendix A1.
In order to determine their provenance, WDS analyses of 134
pyroxene grains from five selected samples of fluvial (Orange and
Fish Rivers) and Namib dune sands were carried out with a Cameca
SX-50 microprobe at the CNR-IGAG Laboratories (method illustrated
in Lustrino et al., 2005). Data on pyroxene chemistry are given in
full and discussed in detail in Appendix A2.
In order to assess the large-scale pattern of sediment transport
and sand residence time in the Namib Sand Sea, the U–Pb ages of
~100 zircon grains were determined for Orange River sample NAM
13 and for each of the twelve Namib dune samples. Cosmogenic
10
Be and
26
Al measurements on detrital quartz were carried out
along a longitudinal transect from the mouth of the Orange River to
the northern edge of the Namib erg. Details on the adopted methods,
followed strategy, and obtained results are discussed in Appendix A3
and illustrated in Vermeesch et al. (2010).
Sands were classified according to their main components
exceeding10%QFL (QFL = quartz+ feldspars+ lithic fragments), listed
in order of abundance (e.g., in a quartzolithic sand L> Q > 10%QFL>F,
in a lithofeldspathoquartzose sand Q> F > L > 10%QFL); an adjective
reflecting the most common rock-fragment type may be added
(e.g., sedimentaclastic, volcaniclastic, plutoniclastic, metamorphiclas-
tic). Metamorphic grains were classified according to protolith com-
position and metamorphic rank. Average rank for each sample is
expressed by the Metamorphic Index (MI; Garzanti and Vezzoli,
2003), which varies from 0 in detritus from unmetamorphosed cover
rocks to 500 in detritus from granulite-facies basement rocks.
Heavy Mineral Concentration indices express the volume percent-
age of transparent (tHMC) and total (i.e., including opaque and turbid
grains; HMC) heavy minerals in each sample; the SRD (Source Rock
Density) index is the weighted average density of extrabasinal terrige-
nous grains (Garzanti and Andò, 2007a). The Hornblende Colour
Index (HCI) estimates the average metamorphic grade of amphibolite-
to granulite-facies source rocks (Garzanti and Andò, 2007b). Transpar-
ent heavy-mineral suites are described as “poor”(0.5≤tHMC b1),
“moderately poor”(1≤tHMCb2), “moderately rich”(2≤tHMC b5),
“rich”(5≤tHMCb10), “very-rich”(10≤tHMCb20), or “extremely
rich”(20≤tHMCb50). Minerals representing ≥10% of the transparent
heavy-mineral suite are listed in order of abundance; rarer but signifi-
cant species are also mentioned. The roundness of counted detrital min-
erals was evaluated by visual comparison with home-made standard
images (nomenclature after Powers, 1953).
Major and trace element concentrations of the sand fraction
(63–2000 μm) of Orange sample NAM 14 were measured by ICP-
AES and ICP-MS at the ACME Laboratories, Vancouver (group 4A–
4B; for detailed information see http://acmelab.com/). Weathering
indices were calculated using molecular proportions and correcting
for CaO in phosphates and carbonates; extensive alteration is indicated
by high CIA (Chemical Index of Alteration; Nesbitt and Young, 1982)
and low WIP (Weathering Index; Parker, 1970).
3.2. Orange River
Orange sand samples range from very-fine to medium-grained,
from moderately-poor to moderately-well sorted, and from feld-
spatholithoquartzose to lithofeldspathoquartzose (Fig. 3A). Plagioclase
prevails over K-feldspar. Rock fragments include equally abundant
volcanic/subvolcanic (largely mafic lathwork types) and sedimentary
grains (shale, quartzose to feldspathoquartzose siltstone/sandstone,
limestone, dolostone, hybrid carbonate), along with common granit-
oid, metasedimentary (slate, metasandstone, quartz–sericite, quartz–
mica, schist, marble) and subordinate metaigneous (porphyroid,
chloritoschist, amphibolite) grains (Fig. 3B). Biotite is abundant in
finer-grained samples. Heavy-mineral assemblages are rich and
clinopyroxene-dominated, with subordinate opaque Fe–Ti–Cr oxides,
epidote, amphibole, and garnet (Fig. 4). Olivine was never recorded.
Compositional variability is particularly marked for the pyroxene/
epidote ratio, which varies from 2 to 32 in modern sands and reaches
37 in the Quaternary sand (Table 1). Hardly ascribed to grain-size ef-
fects and hydraulic sorting because pyroxene and epidote do not have
markedly dissimilar density and shape, such strong variability is pos-
sibly enhanced by anthropogenic effects (irregular sediment flux after
extensive dam regulation of the upper Orange and Vaal, and exten-
sive mining along the lower Orange). If this is true, the pyroxene-
rich Quaternary terrace may reflect the original sediment composi-
tion more faithfully than modern sand, locally epidote-rich because
of reduced supply from the headwaters and consequently greater
relative contribution from greenschist-facies rocks exposed along
the lowermost course (Frimmel and Frank, 1998).
Although the Orange catchment is huge and characterised by a
quite varied geology, the chemical composition of detrital pyroxenes
is remarkably homogeneous, suggesting a similar origin for most ana-
lysed grains. Low-Ti augite is dominant, commonly relatively rich in
Cr
2
O
3
or Al
2
O
3
, and with Mg# 73 ± 9, suggesting equilibrium with
basaltic to basaltic andesite melts. Pigeonite is rare. The composition
of both augites and pigeonites resembles that of clinopyroxenes
commonly found in continental flood basalts, and is virtually
177E. Garzanti et al. / Earth-Science Reviews 112 (2012) 173–189

Author's personal copy
indistinguishable from available analyses of pyroxenes in Karoo
basalts (Fig. 5;Sweeney et al., 1994; Melluso et al., 2008; Galerne
et al., 2010).
Relative to the estimated average concentration of chemical ele-
ments in the upper continental crust (UCC; Taylor and McLennan,
1995; McLennan, 2001; Hu and Gao, 2008), Orange sand contains
half or less Na, Cs, Be, Sr, U, Nb, Ta, Mo, Ni, Sn, Pb, Sb, Bi, and is richer
only in Si and particularly Cr. The REE pattern is similar to the UCC but
with more gently sloping LREE (La
N
/Sm
N
2.9, Gd
N
/Lu
N
1.4) and less
negative Eu anomaly (Eu/Eu⁎0.83 versus Eu/Eu⁎
UCC
0.65), reflecting
significant basaltic sources. Weathering indices (CIA 46, WIP 50),
lower than in upper Orange suspended load (CIA 60 ± 2, WIP
38±2; data after Compton and Maake, 2007), indicate minor
weathering and limited depletion in alkali and alkaline-earth metals
in semiarid subtropical climate (Fig. 6).
The distribution of U/Pb zircon ages in Orange River sample
NAM 13 has a log-normal appearance mostly ranging between 2000
and 500 Ma, with a peak at 1000 Ma and a few grains at 300 Ma
(Fig. 7).
3.3. Hinterland rivers
River sands of SW Namibia are derived from Proterozoic to Cam-
brian sedimentary/metasedimentary, volcanic/metavolcanic, plutonic
and metamorphic rocks in various proportions (for geological details
of catchment areas see Becker et al., 2006). The Swakop River, largely
Fig. 3. Comparison between fluvial Orange and aeolian Namib sands. (A, B, C) Orange sand contains volcanic lithics (Lv) and angular clinopyroxene (p) from Drakensberg basalts,
and siltstone grains (Lt) from Karoo sedimentary rocks. (D, E, F) Namib dunes have similar composition. Heavy minerals (o = opaque Fe–Ti–Cr oxide; g = garnet) are locally con-
centrated in placer lags (black arrows point at Fe-oxide coatings) and markedly rounded by wind action (white arrows). All photos but E with crossed polars; blue bar= 250 μm.
(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
178 E. Garzanti et al. / Earth-Science Reviews 112 (2012) 173–189

Author's personal copy
draining Damaran granites and amphibolite-facies metasediments
of the Nosib and Swakop Groups (Jung and Mezger, 2003), carries
feldspathoquartzose sand (MI 440); the moderately-rich amphibole–
garnet–clinopyroxene–staurolite heavy-mineral suite includes tourma-
line, apatite, titanite, and minor kyanite and fibrolitic sillimanite
(Table 1). The Kuiseb River, largely draining lower-amphibolite-facies
quartzites to micaceous schists (Khomas Subgroup; Miller, 1983; de
Kock, 2001), carriesquartzose to feldspathoquartzose metamorphiclas-
tic sands with abundant biotite (MI 396; Fig. 8A); the moderately-rich
amphibole–epidote–garnet suite includes apatite, tourmaline, clinopyr-
oxene, and rare zircon, rutile, titanite, staurolite, kyanite and monazite.
The Gaub River, a Kuiseb tributary largely draining lower-amphibolite-
facies rocks of the Rehoboth Terrane, carries feldspathoquartzose
metamorphiclastic sand with biotite, muscovite, a few amphibolite
and calcsilicate grains (MI 385), and a rich epidote–amphibole suite.
Along its course at the northern boundary of the Namib erg, Kuiseb
sand is slightly enriched in clinopyroxene, revealing mixing with dune
sand.
The TsondabRiver, largely drainingsedimentary and metasedimen-
tary rocks of the Nama Group and Naukluft Nappe Complex, carries
feldspathoquartzolithic sand with limestone, dolostone, hybrid carbon-
ate, sandstone, metasandstone, shale, slate, phyllite and schist grains
(MI 157; Fig. 8B), and a moderately-rich epidote–amphibole–clinopyr-
oxene heavy-mineral suite. The Zebra/Tsauchab River, largely
draining Nama Group sedimentary rocks, carries quartzolithic sedi-
mentaclastic sand dominated by shale/slate, limestone, quartzose
to feldspathoquartzose sandstone and hybrid carbonate grains
(MI 30±7; Fig. 8C); the poor clinopyroxene–hornblende–epidote
suite includes garnet, titanite, Ti-oxides, minor apatite, tourmaline,
and trace staurolite. After entering the Namib erg towards Sossusvlei,
Fig. 4. Petrography and mineralogy of coastal Namib dunes document dominant provenance from the Orange River. Orange sand contains more sedimentary and metamorphic
lithics, which are selectively destroyed during high-energy longshore transport. Instead, neither volcanic lithics (which maintain their relative proportion) nor pyroxene are deplet-
ed in coastal Namib dunes. Progressive eastward enrichment in quartz, K-feldspar, epidote and amphibole relative to volcanic lithics and pyroxene documents hinterland contri-
butions by the Kuiseb and Koichab Rivers in the NE and SE Namib, respectively. Conversely, Kuiseb and Tsauchab sands are enriched downstream in volcanic lithics and pyroxene
due to aeolian mixing with dune sand. Note compositional similarity between Orange and Nile sands (data after Garzanti et al., 2006), both largely derived from rift-related basaltic
highlands. Q = quartz; F = feldspar; L = aphanite lithics (Lm = metamorphic; Lv = volcanic; Ls = sedimentary).
179E. Garzanti et al. / Earth-Science Reviews 112 (2012) 173–189

Author's personal copy
Tsauchab sand is progressively enriched in subrounded to rounded
quartz, feldspars, volcanic lithic fragments and clinopyroxene, revealing
mixing with dune sand (Fig. 4). Tsaris sand is feldspatholithoquartzose
sedimentaclastic (MI 36) with a moderately-poor amphibole–epidote–
clinopyroxene suite, reflecting mixed provenance from Nama sedimen-
tary and Sinclair magmatic rocks.
The Konkiep River, sourced in the very low-grade volcanic and
subordinately sedimentary rocks of the Sinclair Group with associat-
ed plutonic intrusions, carries feldspatholithoquartzose sand with
volcanic/metavolcanic, subvolcanic, plutonic, sandstone/metasandstone,
and shale/slate grains (MI 64); the moderately-rich epidote-
dominated suite includes amphibole and minor clinopyroxene.
Lovedale sand, entirely derived from such granitoids, is quartzofelds-
pathic, with a moderately-poor epidote–amphibole–clinopyroxene
suite (Fig. 8D).
Similarlyplutoniclastic is the lithofeldspathoquartzose Koichab sand
derived from the NamaquaMetamorphic Complex, which contains sub-
ordinate sedimentary (limestone, hybrid limestone, sandstone) and a
few amphibolite and felsic volcanic grains (MI 103; Fig. 8E); the
moderately-rich amphibole–epidote–clinopyroxene–garnet suite in-
cludes common green–brown hornblende (HCI 25), reflecting erosion
of amphibolite-facies to granulite-facies rocks exposed in the Aus area.
The Fish River, a tributary of the Orange draining the Nama Group
and the overlying Karoo sedimentary and basaltic rocks (Kalkrand
basalts of central Namibia), carries feldspatholithoquartzose sand with
abundant quartzose or feldspathoquartzose sandstone and shale/slate
grains, common volcanic lithic fragments (MI 16; Fig. 8F), and a
moderately-rich clinopyroxene-dominated suite. Pyroxene grains dis-
play quite similar chemical composition as those carried by the Orange
River (Fig. 5), indicating provenance dominantly from the Kalkrand
continental flood basalts widely exposed in the upper course.
3.4. Namib dunes
The analysed coastal Namib dune sands are mainly lower
medium-grained, well-sorted, symmetrical and mesokurtic (1.9±
Table 1
Bulk-petrography and heavy-mineral modes for Orange River, Namib dune, and hinterland-river sands (mean in bold, standard deviation in italics). No. = number of samples; Q =
quartz (RDN = rounded; sRDN = subrounded); F = feldspar; L = aphanite lithics (Lv = volcanic; Lc = carbonate; Lt = siltstone/sandstone; Lm = metamorphic). MI = Metamor-
phic Index. HMC = Heavy Mineral Concentration. Op = opaque Fe–Ti–Cr oxides; Px = pyroxene; Amp = amphibole; Ep = epidote; Grt = garnet; St = staurolite; Zrn = zircon; Tur
= tourmaline; TiOx = Ti oxides; Ttn = titanite; Ap = apatite; & = others (andalusite, sillimanite, kyanite, monazite). HCI = hornblende-colour index.
No. Q F Lv Lc Lt Lm MI Q
RDN
Q
sRDN
HMC Op Px Amp Ep Grt St Zrn Tur TiOx Ttn Ap & HCI
R. Orange 4 57 21 8 2 7 4 100.0 56 3% 42% 6.6 12 62 7 12 3 0 0 1 3 1 1 0 100.0 6
975362 10 3%11%1.9 6 22 2 12 2 0 0 1 3 1 1 0 4
SW Namib 2 66 23 9 0 1 1 100.0 32 7% 41% 10.1 9 76 4 3 6 0 1 0 1 0 0 0 100.0 11
650010 26 6%2%2.0 4 12 1 1 7 0 1 0 0 0 0 0 2
NW Namib 3 68 20 10 0 1 1 100.0 38 10% 37% 7.2 11 72 2 4 9 1 0 0 0 1 0 0 100.0 7
213000 17 2%10%1.6 2 3 2 0 1 0 0 0 0 1 0 0
SE Namib 2 70 26 3 0 1 0 100.0 31 8% 53% 7.2 34 31 9 6 15 1 2 0 0 2 0 0 100.0 16
640001 11 1%5%2.4 23 25 6 0 6 0 1 0 0 0 0 0 11
CE Namib 3 73 19 6 0 1 1 100.0 52 10% 55% 10.4 27 50 6 5 8 0 1 0 1 0 0 0 100.0 6
240011 56 2%17%12.3 18 19 5 1 4 0 0 0 1 0 0 0 2
NE Namib 2 79 16 3 0 1 1 100.0 48 7% 42% 1.3 28 13 9 26 17 1 0 1 2 3 0 1 100.0 6
121000 14 4%9%0.6 12 13 2 3 6 1 0 0 0 1 0 0 2
Swakop 1 66 33 0 1 0 1 100.0 338 0% 26% 3.0 31 9 21 4 13 6 0 4 3 3 4 2 100.0 3
Kuiseb 2 85 9 0 2 0 5 100.0 396 0% 45% 3.2 39 3 18 15 7 1 2 5 3 2 5 0 100.0 1
Gaub 1 73 20 0 1 0 6 100.0 385 0% 40% 7.8 12 2 38 38 1 0 0 1 2 4 3 0 100.0 0
Tsondab 1 37 24 0 17 5 16 100.0 157 0% 64% 4.4 44 6 13 23 2 0 4 1 2 1 2 0 100.0 6
Tsauchab 2 14 2 1 28 44 12 100.0 27 0% 0% 1.0 15 30 25 18 4 0 0 1 2 4 1 0 100.0 2
Tsaris 1 44 25 4 15 10 2 100.0 36 0% 40% 1.3 37 14 19 18 1 0 2 1 2 3 2 0 100.0 6
Lovedale 1 48 49 1 1 0 1 100.0 60 1% 84% 2.0 58 4 8 20 2 0 5 0 1 2 0 0 100.0 6
Koichab 1 43 41 2 10 4 1 100.0 103 0% 31% 3.6 52 6 23 7 6 0 3 0 1 1 1 0 100.0 25
Konkiep 1 44 28 18 0 4 6 100.0 64 1% 69% 4.3 18 7 16 56 1 0 1 0 0 1 1 1 100.0 15
Fish 1 60 14 5 1 18 3 100.0 16 1% 80% 2.7 12 73 3 6 3 0 1 1 1 0 0 0 100.0 n.d.
Fig. 5. Clinopyroxene grains in Orange sand, Fish sand, and Namib dunes (N = NAM 3, NW = NAM 1, SW = NAM 12). Pyroxene grains display the same range of compositions as in
various types of Karoo lavas (data after Sweeney et al., 1994; Melluso et al., 2008; Galerne et al., 2010), indicating common provenance, dominantly from Karoo continental flood
basalts. Pyroxene quadrilateral diagram after Morimoto et al. (1988).
180 E. Garzanti et al. / Earth-Science Reviews 112 (2012) 173–189

Author's personal copy
0.2ϕ,σ0.46±0.11ϕ, Sk 0.0 ± 0.1, Ku 2.6 ±0.0). As observed by
Lancaster and Ollier (1983 p. 72;Lancaster, 1989), grain size tends
to decrease, and sorting to increase in eastern Namib dunes (2.1±
0.2ϕ,σ0.37±0.02ϕ, Sk 0.0 ± 0.0, Ku 2.7 ± 0.0). Sand colour varies
across the sand sea from light yellowish brown in coastal areas
(10 YR 6/4), to reddish yellow in central areas (7.5 YR 6/6), to yellow-
ish red in eastern areas (5 YR 5/8). The colour primarily reflects Fe-
oxide coatings on detrital grains (White et al., 2007), which are
visibly thicker in the eastern Namib sands, reflecting both more
ancient origin of the dunes and greater moisture content (Fig. 3E;
Folk, 1976; Walden and White, 1997). Weathering of heavy minerals
is minor. Pyroxene is occasionally corroded, rarely etched, and
exceptionally skeletal in Orange sand; etched pyroxene grains with
rounded or subrounded outline occur in Namib dunes. Amphibole
and epidote grains may be corroded, but rarely etched.
Dune sands in the coastal Namib are invariably lithofeldspatho-
quartzose volcaniclastic, with homogeneous composition from
Lüderitz to Walvis Bay (Fig. 4). Plagioclase prevails over K-feldspar.
Mainly mafic volcanic rock fragments predominate over granitoid,
sedimentary (quartzose to feldspathoquartzose siltstone/sandstone,
shale, minor limestone and hybrid carbonate) and metamorphic
(quartz–sericite, quartz–mica, quartz–epidote, amphibolite) grains
(Fig. 3D). Micas are absent. Rich heavy-mineral assemblages are
clinopyroxene-dominated, with subordinate opaque Fe–Ti–Cr oxides,
garnet, epidote, and amphibole.
Eastern Namib dunes are feldspathoquartzose, with less volcanic
rock fragments. K-feldspar prevails over plagioclase. Micas are
absent. Heavy-mineral assemblages, generally moderately-rich and
clinopyroxene-dominated with Fe–Ti–Cr oxides, amphibole, garnet
and epidote, are locally rich (NAM 9) or even extremely rich (NAM
Fig. 6. Sand chemistry in big African rivers (elements in UCC-normalised multielement diagram arranged following the periodic table group by group). Composition of Orange sand
~100 km upstream of the mouth (NAM 14) compares closely to that of Nile sand upstream Lake Nasser (Padoan et al., 2011). For both passive-margin river systems, formed long
ago by rift-related domal uplift, Cr enrichment and Eu anomaly less negative than UCC reflect focused erosion of distant basaltic highlands. The Cr peak is controlled chiefly by au-
gitic pyroxene, and subordinately by basaltic rock fragments and Fe–Ti–Cr oxides. Only moderate depletion in alkali and alkaline-earth metals suggests limited weathering in semi-
arid tropical climate. In contrast, weathered sand of the equatorial Congo River is turned into a quartz residue, as shown by strong depletion in all elements but Si and by pattern
similar to detrital quartz. In Orange sand, depletion in REE and transition metals, removed from primary minerals and involved in the formation of oxyhydroxides (Quinn et al.,
2006), is largely ascribed to fractionation of tiny phyllosilicates in suspended load. Soluble Mo is strongly depleted. All three analysed sand samples have normal grain density
(SRD 2.71±0.02) and are not markedly affected by selective-entrainment effects. Data for Upper Orange suspended load after Compton and Maake (2007). Quartz separates
obtained from 4 central African and Arabian sand samples.
Fig. 7. Detrital zircon U–Pb age spectra for the Orange River and Namib Sand Sea (modified after Vermeesch et al., 2010). Age spectra display a pronounced east–west dichotomy,
with coastal samples appearing nearly identical to the Orange sample, whereas samples near the edges of the desert show evidence for additional local sources.
181E. Garzanti et al. / Earth-Science Reviews 112 (2012) 173–189

Author's personal copy
8; Fig. 3E), and display a sharp increase in ultradense Fe–Ti–Cr
oxides and garnet at the expense of less dense pyroxene. In the NE
corner of the Namib, instead, assemblages are moderately poor
and include epidote, Fe–Ti–Cr oxides, garnet, clinopyroxene, amphi-
bole, minor titanite, tourmaline, rutile, staurolite, andalusite, and
apatite (NAM 4–5). Pyroxene grains from dune sands at the SW,
NW and N edges of the sand sea display very similar chemistry as
those in Orange River sands (Fig. 5).
U–Pb age distributions of detrital zircons remain virtually identical
to the Orange sample all along the coastal Namib, whereas the
distinct age peak at 1000 Ma becomes less pronounced inland.
The similarity of age populations is greater for samples along longitu-
dinal profiles than along latitudinal profiles, indicating northward
sand migration, parallel to the coast and to the linear-dune system
that dominates the desert. Deviations from the Orange-sand
spectrum are most marked at the northeastern edge of the erg,
indicating mixing with locally sourced populations of zircon grains
(Fig. 7).
4. Provenance of Namib dune sand
The proposed sources of Namib sand beside the Orange River
include ephemeral hinterland rivers, whose detritus may be blown
to less peripheral parts of the sand sea by the northeasterly winter
winds of the NE Namib, recycled ancestral fluvial or aeolian deposits,
and deflation areas along the coast (e.g., Sperrgebiet; Besler, 1984;
Lancaster, 1985).
The strict compositional resemblance between Orange sand and
most desert dunes proves that most of the Namib sand is ultimately
derived from the Orange River (Fig. 9). Specifically, overall abundance
and relative percentages of feldspar, volcanic-rock-fragment, and
heavy-mineral species including clinopyroxene are virtually identical
Fig. 8. Petrography of hinterland-river sands. Sharply distinct from both Orange and Namib dune sand, composition ranges from quartzose metamorphiclastic (A), to quartzolithic
sedimentaclastic (B, C), and quartzofeldspathic plutoniclastic (D, E). Only the Fish River, an Orange tributary, similarly carries mafic volcanic and sedimentary detritus from Kalk-
rand basalts and underlying Karoo and Nama sedimentary rocks (F). K = K-feldspar; P = plagioclase; m = mica; L = lithics (Lv = volcanic; Lc = carbonate; Lt = siltstone; Lms =
metasandstone; Lm = metamorphic). All photos with crossed polars; blue bar= 250 μm.
182 E. Garzanti et al. / Earth-Science Reviews 112 (2012) 173–189

Author's personal copy
in Orange and coastal Namib sands. Common mafic volcanic detritus
and rich clinopyroxene-dominated suites with minor epidote and
amphibole are signatures that characterise no hinterland river drain-
ing into the Namib erg, and are even more extreme in coastal Namib
dunes than in the modern Orange itself. Such signatures best compare
with the Orange Quaternary terrace, suggesting greater contribution
from the upper Orange catchment in pre-dam times. Modern post-
dam Orange sediments, displaying detrital modes less dissimilar
from those of hinterland rivers, may be derived in greater proportion
from the lower catchment (for instance, relative to modern Orange
sand amphibole%tHM is half in coastal Namib dunes and Quaternary
terrace, but invariably twice or more in hinterland rivers).
Only minor and local supply from non-Orange sources is indepen-
dently confirmed by zircon U–Pb age spectra of Namib dune sands,
which compare closely with Orange sand (Fig. 7) rather than with
ages of Namaqua or Damara basements exposed nearby. Abundance
of zircon grains does not result to be greater in Orange sand (0.03 ±
0.03%) than in hinterland-river sands (0.05 ± 0.05%), indicating that
relative budgets based on zircon-age spectra are not significantly
biased by markedly different fertility of diverse sources of sediment
(Moecher and Samson, 2006). Zircon content in dune sands resulted
to be similar (0.04±0.04%, notably increasing to 0.2% in NAM 9 and
to 0.3% in NAM 8 due to aeolian-sorting effects, as discussed below),
proving that such estimates are not inaccurate, although scarcely
precise because of the limited number of counted zircon grains.
4.1. Contributions from hinterland rivers
Virtually unchanged detrital modes from the Orange mouth to
Walvis Bay reveal lack of additional sediment sources along the
coast, and irrelevant supply from both Sperrgebiet outcrops in the
south and Swakop mouth in the north. At the northern edge of the
desert, volcanic lithic fragments and pyroxene grains remain as abun-
dant, and epidote and amphibole as scarce, as in Orange or Lüderitz
dune sand.
Only at the northeastern edge of the desert, where relict fluvial
deposits associated with former courses of the Kuiseb and Tsondab
Rivers occur (Lancaster et al., 2000), does the mineralogy of dune
sands drastically change, revealing the existence of local entry points
of detritus. Most evident are hinterland contributions at the NE
corner of the Namib erg, where quartz is more abundant, volcanic
detritus scarce, and opaque–epidote–garnet assemblages poorer and
distinct from clinopyroxene-dominated Orange and coastal Namib
suites. Affinity with Kuiseb sand is indicated. Very minor supply
from modern or ancient hinterland-river sediments cannot be
excluded for samples NAM 1, NAM 2 and NAM 3, which include
garnet and trace staurolite.
Significant hinterland contributions are detected also at the SE
corner of the Namib. Landwards of Lüderitz (NAM 11 to NAM 9),
quartz increases at the expense of volcanic detritus, and garnet,
Fe–Ti–Cr oxides, amphibole (including green and green–brown
hornblende), epidote, titanite, staurolite and zircon become more
significant, suggesting contributions from the Koichab River and
locally exposed Namaqua basement rocks. In sample NAM 9, the
HCI index matches that of Koichab sand draining high-temperature
Namaqua rocks; concentration of garnet and Fe–Ti–Cr oxides is
instead ascribed to aeolian sorting (see below).
Quartz is relatively high and volcanic rock fragments low in the
Sossuvlei transect, where amphibole is slightly more abundant locally
(NAM 6); strong concentration of garnet, Fe–Ti–Cr oxides and epidote
in sample NAM 8 is an aeolian-sorting effect. Such changes cannot be
accounted for by contributions from the Tsauchab River, which
carries abundant shale/slate and limestone rock fragments, little
quartz, few heavy minerals, and virtually no garnet.
A quantitative estimate of hinterland contributions shall be
attempted by comparing detrital modes of dune and hinterland-
river sands after all other factors controlling compositional variability
are carefully considered.
4.2. Recycling from Tertiary aeolianites
A major potential source of recycled sand is the up to 220 m-thick
Tsondab Sandstone, underlying much of the modern sand sea (Ward,
1988). This unit displays quite similar morphology, geometry and
mineralogy of modern linear dunes, and represents a Miocene ana-
logue of the present erg (Kocurek et al., 1999; Ségalen et al., 2004).
The Tsondab aeolianites are characterised by rich and pyroxene-
dominated heavy-mineral assemblages (HMC 9± 8, Px 85±6%tHM),
but for the northern edge of the Namib erg where heavy-mineral con-
centration is much less (HMC 1.1±0.8), and common garnet is asso-
ciated either with staurolite, epidote and tourmaline (south of the
Kuiseb River downstream of Gobabeb), or with hornblende, zircon,
and rutile (Tsauchab and Tsondab valleys as far as the NAM 4 locality;
Dickinson and Ward, 1994; Besler, 1996).
The close mineralogical similarity between Tertiary aeolianites
and modern dunes suggests that the patterns of sand dispersal have
not changed much since the Miocene at least, but prevents us to
establish how much detritus is recycled from Tertiary sandstones.
Recycling may be particularly significant in the type area, where
Tsondab aeolianites are more widely exposed, and similar petro-
graphic changes are observed in Tertiary and modern dunes (Ward,
1988).
Fe–smectite and zeolite cements, as well as grain-dissolution
features, are commonly observed in Tsondab aeolianites (Dickinson
and Ward, 1994), but abundant pyroxene and high heavy-mineral
concentration (Besler, 1996) prove that the abundance of unstable
detrital species is not significantly affected by diagenesis at such
very shallow burial depths. The invariably drastic decrease of
pyroxene close to the eastern margin of the sand sea, both in modern
sands and Tertiary aeolianites (Besler, 1996), indicates that basement
rocks east of the Namib erg do not represent a major source of
detrital pyroxene (as hypothesised instead by Lancaster and Ollier,
1983 p. 81).
5. Physical processes controlling compositional variability
of dune sand
In order to be properly and quantitatively understood, the
multiple factors controlling sediment composition should be ideally
singled out and studied one by one. This is easier to attempt in
modern desert settings, where mineralogy is only affected by physical
processes, with negligible chemical weathering and no diagenesis.
Although compositional variability still reflects the superposed effects
Fig. 9. Namib dunes compare compositionally with Orange River sand, and are distinct
from hinterland-river sand. Affinity with Orange sand, greatest in the coastal Namib,
decreases inland and is least for NAM 4–5 dunes at the NE corner of the erg, indicating
mixing with local detrital sources. Cluster analysis performed with Aitchison distance
(van den Boogaart and Tolosana-Delgado, 2008) on the whole bulk-petrography and
heavy-mineral dataset.
183E. Garzanti et al. / Earth-Science Reviews 112 (2012) 173–189

Author's personal copy
of diverse factors, and ambiguous interpretation may thus ensue (e.g.,
ultradense and durable garnet may increase locally because of selec-
tive destruction of labile components, recycling of underlying sand-
stones, contribution from external sources, or aeolian sorting),
favourable conditions in this respect are offered by the Namib Sand
Sea, where mechanical abrasion and breakdown of detrital grains
with different durabilities can be monitored during long-distance
aeolian transport from the single principal entry-point in the south
to the northern edges of the desert. Once the localization and compo-
sitional signatures of potential external sources are identified, the
mechanical causes of compositional and morphological variabilities
can be quantitatively investigated in Namib dune sands, after assess-
ment and removal of aeolian-sorting effects.
5.1. Aeolian sorting and placer formation
Aeolian deposits may be enriched in dense minerals to various de-
grees by wind turbulence on dune flanks. Dense minerals are also
concentrated up the stoss side of dunes where flow velocity increases,
whereas low-density minerals are preferentially entrained over the
crest and down the slip face (Komar, 2007). Such effects abide simple
physical rules, and can thus be quantified and predicted (Garzanti et
al., 2008; 2009). For instance, epidote, titanite or diamond (densi-
ty~3.5 g/cm
3
) should be ~0.4 ϕsmaller than associated quartz, and
enriched by ~3 times in placer sand with SRD index~ 3.0 (diamond
was never found in our samples, unfortunately).
Aeolian-sorting effects can be assessed by determining the average
grain density for each sample, which in absence of environmental bias
is expected to be equal to average source-rock density (Garzanti and
Andò, 2007a). The SRD index of Orange sand is 2.72±0.04, close to
the average density of the upper crust and thus of sediments carried
by many big rivers including the Nile (Garzanti et al., 2006; 2010).
Eight out of twelve Namib dune samples have similar SRD indices
(2.71±0.02). Such remarkable consistency testifies that the care
taken to collect samples unaffected by environmental bias was largely
successful. Higher indices characterise samples NAM 9 (HMC 9, SRD
2.79) and particularly NAM 8 (HMC 25, SRD 3.03), reflecting wind-
induced concentration of ultradense detrital components (Fe–Ti–Cr
oxides 39–42%HM, garnet 11–19%HM, zircon 1–2%HM; Fig. 3E).
Strong concentration of ultradense minerals was much more fre-
quently reported by Lancaster and Ollier (1983). Possibly due to a dif-
ferent sampling strategy, they found garnet> 20%HM in 15 out of
their 26 samples, and even prevailing over clinopyroxene in 9 sam-
ples (which we only observed in samples NAM 5 and NAM 9, where
hinterland contributions are significant).
Because the degree to which any denser mineral lags behind a less
dense mineral is primarily a function of their density ratios (Trask and
Hand, 1985), detrital minerals are systematically enriched in propor-
tion to their density in samples with higher SRD. The observed com-
positions can thus be recalculated for each sample by iteratively
correcting abundances of detrital species in proportion to their densi-
ties, until the correct SRD index is restored for each (SRD correction of
Garzanti et al., 2009). This cannot be safely done for samples with
lower grain density (SRDb2.70) where either selective removal of
dense grains with anomalous physical behaviour (e.g., biotite flakes
supplied by the nearby Kuiseb River for samples NAM 4 and NAM
5) or quartz dilution (e.g., recycling of Tsondab Sandstone for samples
NAM 6 and NAM 7) might have taken place. In such cases, the SRD
correction would restore an excess of densest components.
For the other samples, aeolian-sorting effects were compensated
for by recalculating detrital modes to an SRD index of 2.715 ± 0.005.
Such corrections are minor for most samples and most minerals
(b20% even for garnet), but significant for placer sand NAM 8. The
compositional variability in homogeneous sample groups tends to de-
crease after the SRD correction, indicating that even subtle environ-
mental bias can be detected and removed. On the other hand, the
risk of amplifying analytical errors by such procedure must always
be taken into account.
Our calculations suggest that aeolian sorting accounts for ~20%
depletion in quartz, feldspar and lithic fragments in sample NAM 8,
and corresponding four-fold enrichment in heavy minerals. Opaque
Fe–Ti–Cr oxides are estimated to be enriched 10–15 times, zircon~8
times, garnet~5 times, staurolite, epidote and titanite 3–4 times, py-
roxene and amphibole~2 times. Modifications are lesser for sample
NAM 9, with ~3% depletion in quartz, feldspar and lithic fragments,
and corresponding 30% enrichment in pyroxene and amphibole;
staurolite, epidote and titanite are estimated to be enriched ~2
times, garnet~3 times, zircon ~ 4 times, and Fe–Ti–Cr oxides~ 5 times.
Besides such locally strong effects, the abundance of dense and
ultradense minerals, average grain density and grain size do not
show any systematic change with transport distance as far as the
northern edge of the erg. This lack of evidence for long-term sorting
of grains according to density or size during northward transport is
apparently in contrast with the observed decline in abundance and
size of diamond grains north of the Orange mouth (Jacob et al.,
2006 p. 496).
5.2. Mechanical durability of diverse detrital components
The mineralogy of the Lüderitz dune (NAM 12) compares closely
with Orange sand, and particularly with the Quaternary-terrace sam-
ple. The absence of micas and the only sporadic occurrence of carbon-
ate, terrigenous and metamorphic lithic fragments, common to all
Namib dunes, are the sole notable differences. Compositional modifi-
cations in high-energy beach settings thus appear to be limited to
winnowing of slow-settling micas, widely dispersed offshore
(Spaggiari et al., 2006; Bluck et al., 2007), and to selective destruction
of non-durable sedimentary/metasedimentary rock fragments and
alterites (Cameron and Blatt, 1971; McBride and Picard, 1987;
Garzanti et al., 2002). Most severely depleted are shale/slate grains,
followed by carbonate, sandstone/metasandstone, and other meta-
morphic rock fragments. All types of volcanic rock fragments, as
well as most heavy-mineral species, appear instead to healthily sur-
vive longshore transport; possible exceptions are apatite and amphi-
bole, which show slight depletion.
The composition of dune sand remains virtually identical after an-
other 600 km of transport as far as Walvis Bay at least, with the excep-
tion of a slight relativeenrichment inquartz and garnet (Fig. 10). Slight
enrichment in quartz and garnet is also observed in dune sand~ 60km
NE of Lüderitz (NAM 11), whereas compositional changes further east
(NAM 10–9) arelargely ascribed to hinterland supply and aeolian sort-
ing. Central eastern Namib sands (NAM 7–6 and NAM 8 SRD-corrected
for aeolian sorting) are slightly enriched in quartz and garnet, and de-
pleted in volcanic rock fragments and pyroxene relative to Orange
sands and the Lüderitz dune (NAM 12). Along the northern Namib tra-
verse, detrital modes of dune sand remain virtually identical for
50–100 km (NAM 1–2–3), and next change chiefly because of supply
from external sources (NAM 4–5). Micas are virtually absent in dune
sands even close to the Kuiseb River that carries abundant biotite,
suggesting efficient selective winnowing.
Comparing sand composition at the entry point (Lüderitz dune
NAM 12, where quartz is 56%, volcanic lithic fragments 9%, pyroxene
8%, and garnet 0.2% of bulk sand) with dunes showing only minor
hinterland contributions (NAM 1–2–3, 6–7–8, 10–11, where quartz
is 65±3%, volcanic lithic fragments 7± 3%, pyroxene 4 ±1%, and gar-
net 0.6± 0.2%of bulk sand) allows us to assess the maximum possible
effect of selective breakdown during prolonged aeolian transport.
Although sedimentary processes along the high-energy Namibia
coast are held to be extreme, and severe sand abrasion in the defla-
tion basin of the Namib erg can destroy naturally exhumed bone ma-
terial in a few months (Corbett and Burrell, 2001 p. 64), our data
consistently indicate that only trivial compositional changes are
184 E. Garzanti et al. / Earth-Science Reviews 112 (2012) 173–189

Author's personal copy
caused by mechanical breakage during transport over a distance of
600 km and more (Fig. 10). The slight increase in durable quartz
and garnet across the sand sea cannot be ascribed to external sources
(hinterland rivers carry similar amounts or less quartz and garnet
than the Orange River), but may be partly accounted for by recycling
of underlying Tertiary aeolianites.
Particularly worthy of note is the durability of basaltic rock frag-
ments and pyroxene grains, which do not only endure more than
2000 km of transport from Lesotho highlands along the Orange
River, but survive in virtually unchanged proportions multistep trans-
port in high-energy coastal and aeolian environments for another
1000 km, prolonged over a time period as long as 1 Ma or more
(Vermeesch et al., 2010). Basaltic rock fragments and pyroxene grains
are similarly observed to remain virtually unchanged after 4000 km
of fluvial transport along the Nile from Ethiopian rift highlands to
the Mediterranean Sea (Shukri, 1950; Garzanti et al., 2006). We can
thus conclusively contradict the common belief that mafic volcanic
detritus is readily worn away within a few tens of km of its source,
and that rift-related volcanic rocks are consequently not expected to
leave significant trace in the sedimentary record (Blatt, 1978;
Ingersoll, 1984; Ingersoll et al., 1993 p. 942).
5.3. Rounding by aeolian abrasion
Although aeolian impacts do not prove to be effective enough to
change sand composition by selectively destroying labile components,
they can spectacularly modify the morphology of detrital minerals,
which may become as well rounded as perfect spheres. Evident at
first sight (Fig. 3C, F), this effect can be quantified by comparing the
Fig. 10. Long-distance transport in high-energy coastal and aeolian settings does not selectively destroy volcanic rock fragments and pyroxene grains (compositional changes given
in log ratios). Whereas sedimentary and metamorphic rock fragments are effectively reduced during longshore transport from Orange mouth (OR) to Lüderitz (LÜ), the F/R
v
ratio
remains unchanged to Walvis Bay (WB) and the Q/R
v
ratio increases only slightly. The same is true during aeolian transport eastwards across the western Namib, whereas Orange-
derived volcanic detritus decreases eastwards across the eastern Namib chiefly by dilution of hinterland-derived sand. Inset shows changing ratios between key parameters. Q =
quartz; F = feldspars (P = plagioclase; K = K-feldspar); RF = rock fragments (R
v
= volcanic; R
m
= metamorphic; R
s
= sedimentary); HMC = heavy-mineral concentration; Px =
pyroxene; Ep = epidote; Amp = amphibole; Grt = garnet; ZTR = zircon+ rutile+ tourmaline.
185E. Garzanti et al. / Earth-Science Reviews 112 (2012) 173–189

Author's personal copy
roundness of diverse minerals in Namib dunes and in Orange and
hinterland river sands.
Very few detrital grains are rounded in Orange sands (e.g.,
apatite), where quartz, titanite, amphibole, pyroxene, or more rarely
other minerals may be subrounded. The same is true for hinterland-
river sediments, where subrounded quartz and heavy minerals are
observed more frequently in the Zebra/Tsauchab and Fish Rivers,
recycling Nama Group or Karoo sedimentary rocks.
In Namib dunes, instead, sand grains are mostly subrounded
(Goudie and Watson, 1981), and commonly rounded. Only in sample
NAM 5 are grains uncommonly rounded (9%), and frequently
angular/subangular (23%). Among heavy minerals, titanite appears
as most commonly rounded (67% grains rounded, 28% grains sub-
rounded), followed by pyroxene (55% rounded, 39% subrounded),
opaque Fe–Ti–Cr oxides (49% rounded, 45% subrounded), apatite
(40% rounded, 60% subrounded), amphibole (40% rounded, 54% sub-
rounded), tourmaline (38% rounded, 50% subrounded), epidote (29%
rounded, 50% subrounded), zircon (13% rounded, 65% subrounded),
garnet (2% rounded, 79% subrounded), and rutile (8% rounded, 33%
subrounded). Positive correlation between mineral roundness and
hardness, as empirically measured by the Mohs' scale (Marsland
and Woodruff, 1937; Dietz, 1973), is thus indicated (r= 0.72; 2%
sign. lev).
Quartz appears to be least commonly rounded and subrounded at
Lüderitz, but evidence for systematic landward increase in quartz
roundness is weak along the southern and central transects. Pyroxene
tends to become more commonly rounded landwards, but other min-
erals display erratic trends. Conversely, angularity tends to increase
inland along the northern transect, confirming the influence of hin-
terland sources for dune samples NAM 4 and NAM 5.
Sand grains thus appear to be efficiently and rapidly rounded in
aeolian settings, whereas the effects of further multistep aeolian
transport are barely evident and locally dimmed by supply of more
angular detritus from hinterland rivers. Roundness may have been
acquired up to an undetermined degree during high-energy long-
shore transport before entering the sand sea, or partly inherited
from recycling of Tsondab aeolianites. Comparison with sediment
transport along the Nile coastal cell from the Delta to northern Israel
(own data) indicates that heavy-mineral roundness does increase
over a longshore distance of ~400 km, but such a change is at least
partly due to reworking of coastal dunes and aeolianites.
6. Sediment budget
Having assessed the diverse causes of compositional change, and
determined the localizations and signatures of all possible external
sources (i.e., potential contributions from the Great Escarpment via
the Koichab, Tsaris, Tsauchab, Tsondab and Kuiseb Rivers in the
southeast to northeast, and along shore from the Swakop mouth in
the northwest; Fig. 1), we can now attempt to quantitatively assess
the Orange versus hinterland contributions to diverse parts of the
Namib Sand Sea with simple statistical techniques (linear mixing
model; Weltje, 1997).
Orange sand, including sedimentary to mafic volcanic lithic frag-
ments and clinopyroxene from Carboniferous to Jurassic rocks of the
Karoo Supergroup, is readily distinguished from hinterland-river
sand, containing granitoid and metasedimentary detritus with
epidote, amphibole, garnet and locally staurolite from Proterozoic to
lowermost Palaeozoic basement rocks of the Namaqua and Damaran
Orogens. Considering that the marked compositional variability of
modern Orange sand results at least in part from the modification of
natural sediment fluxes caused by dams and mining activities, we
chose to discard modern sands of the lowermost course (S4037,
NAM 13), and thus assumed as Orange end-member the average de-
trital modes of samples NAM 14 and S4038. More robust results were
obtained when the proximal Lüderitz dune (NAM 12) was included in
the average. For all samples we used a double data string, one with
bulk-petrography modes, recalculated to 100% after eliminating
micas and labile sedimentary/metasedimentary lithic fragments but
including total heavy-mineral content (i.e., HMC index), and the
other with heavy minerals only, recalculated to 100%.
Contrary to chemical processes, which can be safely considered as
negligible in such a hyperarid environment, physical processes are
relevant, and have to be evaluated and wherever possible corrected
for. Most evident is heavy-mineral enrichment due to wind action
in samples NAM 9 and particularly NAM 8, the detrital modes of
which have been SRD-corrected to an SRD index of 2.715±0.005.
For simplicity, raw data were used for all of the other samples.
6.1. Quantitative provenance assessment
Whereas sand composition varies little from the Orange mouth to
Walvis Bay, along a distance of nearly 1000 km along the coast, all
three studied E–W transects display a systematic variability in detrital
modes, with eastward increase in quartz (and particularly of sub-
rounded to well-rounded grains) and parallel decrease in volcanic
lithic fragments and pyroxene. Such variability can be modelled as
simple mixing of Orange-delivered sand with detritus produced in
the hinterland. As a measure of statistical similarity, we used the co-
efficient of multiple correlation Rbetween observed and modelled
petrographic and heavy-mineral modes (for further methodological
details see Vezzoli and Garzanti, 2009).
Linear-mixing calculations indicate that dune sand of the coastal
Namib is virtually entirely Orange-derived, with external contribu-
tions being ≤15% even at the northernmost edge of the sand sea
(NAM 1–2–3, R=0.99). External sources become dominant only at
the northeastern edge of the erg (~75%; NAM 4, R=0.95; ≤98%,
NAM 5, R=0.97), where dune sand is largely derived from the Kuiseb
River, either directly or indirectly by recycling of relict deposits.
Quartz and heavy minerals are less commonly rounded than in
other dunes, suggesting that recycling of Tertiary aeolianites is not
extensive. Contribution from the Tsauchab River to the central east-
ern Namib is minor (≤15%; NAM 7–6 and NAM 8 SRD-corrected,
0.90bRb0.95). At the southern edge of the desert, the contribution
from the Koichab River and locally exposed Namaqua basement
rocks increases steadily eastwards from ≤15% (NAM 11, R= 0.97),
to 35–40% (NAM 10, R=0.96) and up to 70–75% (NAM 9 SRD-
corrected, R=0.88). Because samples were collected from accessible
dunes at the edge of the sand sea in areas more prone to receiving
detritus from hinterland sources, external contributions are expected
to decrease in the interior of the erg.
The similarity indices thus obtained (R= 0.95 ± 0.04) can be
improved by tentatively quantifying the progressive breakdown of
mechanically less resistant grains, with consequent enrichment in
quartz. We obtained the best overall fit(R=0.98 ±0.01) by assuming
that most carbonate lithic fragments, over two/thirds of other sedi-
mentary to low-rank metasedimentary lithic fragments, and 10–15%
of feldspar (Q/F increases from Lüderitz to Walvis Bay) are selectively
destroyed in aeolian settings, and that garnet is enriched at the
expense of cleavable amphibole and of other heavy minerals to a
lesser extent. The sediment budget thus calculated is illustrated in
Fig. 11.
If these assumptions are correct, the estimated Tsauchab contribu-
tions to the central eastern Namib may reach locally as high as
30–40% (NAM 7–6–8, R=0.98± 0.01), whereas hinterland contribu-
tions at the northwestern edge of the Namib become negligible (b5%,
NAM 1–2, R=1.00). The estimated contribution from the Koichab
River and Namaqua rocks exposed along the southern edge of the
erg (≤15%, NAM 11, R=1.00; 35–40% NAM 10, R= 0.98; ~75%
NAM 9, R=0.96), and from the Kuiseb River at the northeastern
edge (b10%, NAM 3, R=0.98; ~75%, NAM 4, R= 0.97; ≤98%, NAM
5, R=0.97) remain virtually unchanged.
186 E. Garzanti et al. / Earth-Science Reviews 112 (2012) 173–189

Author's personal copy
7. Conclusion
The Namib Sand Sea is chiefly fed by the Orange River, as docu-
mented by bulk petrography, heavy-mineral suites, percentages of
feldspar species and volcanic-rock-fragment types, pyroxene chemis-
try, and U/Pb age populations of detrital zircons, which all remain
remarkably uniform all along the western, coastal part of the erg.
Hinterland sediment sources do exist along the eastern margin of
the dune field, but major external contributions occur only locally at
its NE and SE corners, where sand is largely provided by the Kuiseb
and Koichab Rivers, respectively.
The peculiarity of the Namib erg, with essentially one single entry
point of detritus supplying a variety of labile rock fragments and
heavy minerals, offers a unique opportunity to verify whether and
to what extent mechanical abrasion and selective breakdown of less
durable components can effectively modify grain morphology and
sand composition. Once local provenance and aeolian-sorting effects
are identified and compensated for, the composition of Orange-
derived sand and specifically the abundance of basaltic lithic
fragments and pyroxene grains are observed to remain virtually
unchanged along the Namibia coast and as far as the northern edge
of the erg. Substantial compositional modifications due to mechanical
processes do take place before entering the erg in high-energy coastal
settings, where micas are readily winnowed by waves and deposited
offshore, and sedimentary to low-rank metasedimentary rock
fragments (shale/slate, limestone) are selectively destroyed. Within
the sand sea, the only systematic change, possibly ascribed in part
to recycling of underlying Tertiary aeolianites, is a slight northward
and eastward enrichment in quartz and garnet, and a slight depletion
in volcanic detritus in the central eastern Namib. Aeolian abrasion, a
relatively inefficient agent of compositional change, meets far greater
success in sculpting detrital grains, invariably much better rounded in
aeolian dunes than in both Orange and hinterland fluvial sands. The
roundness of quartz grains tends to improve with transport distance,
but heavy minerals appear nearly as well rounded in the Lüderitz
dune as in the rest of the erg, where a minority of subangular grains
remains. Roundness is thus acquired readily in the aeolian environ-
ment, with minor further improvement.
Beyond the origin of Namib dune sands, the new data presented
here solve issues that have remained controversial in sedimentary
petrology for long. We demonstrated the inefficacy of selective
mechanical breakdown in modifying sediment composition, even
through a sequence of extremely energetic fluvial, coastal and aeolian
environments, along a routing system of more than 3000 km, and
during time periods up to 1 Ma and more. In particular, we conclu-
sively proved the durability of basaltic rock fragments and pyroxene
grains. Rift-related volcanism thus can and does leave traces in the
sedimentary record, and represents an independent and significant
type of sediment provenance. Finally, we confirmed the efficiency of
wind-induced grain-to-grain collisions in sculpting detrital minerals,
which become rapidly rounded once passed from fluvial and marine
environments to the aeolian domain. Such newly acquired insight
on the effects of physical processes will result in higher resolution
for provenance interpretations of ancient sandstones, and thus in
more accurate reconstructions of climates and landscapes of the past.
Acknowledgments
The article benefited from comments and advice by Asish R. Basu
and an anonymous reviewer. Giles Wiggs and Charlie Bristow kindly
provided laser grain-size and geochronological data on Namib
dunes. Marta Padoan and Alberto Resentini helped in the SEM
analyses of pyroxene grains and statistical analysis.
Appendix A. Supplementary material
Supplementary material to this article, found online at doi:10.
1016/j.earscirev.2012.02.008, includes discussions on alternative
heavy-mineral counting methods (Appendix A1), on composition
and provenance of detrital pyroxene (Appendix A2) and on age-
patterns of detrital zircon (Appendix A3).
References
Ahrendt, H., Hunziker, J.C., Weber, K., 1978. Age and degree of metamorphism and time
of nappe emplacement along the southern margin of the Damara Orogen, Namibia
(SW-Africa). Geologische Rundschau 67, 719–742.
Avigad, D., Sandler, A., Kolodner, K., Stern, R.J., McWilliams, M., Miller, N., Beyth, M.,
2005. Mass-production of Cambro-Ordovician quartz-rich sandstone as a conse-
quence of chemical weathering of Pan-African terranes: environmental implica-
tions. Earth and Planetary Science Letters 240, 818–826.
Becker, T., Schreiber, U., Kampunzu, A.B., Armstrong, R., 2006. Mesoproterozoic rocks of
Namibia and their plate tectonic setting. Journal of African Earth Sciences 46,
112–140.
Besler, H., 1984. The development of the Namib dune field according to sedimentolog-
ical and geomorphological evidence. In: Vogel, J.C. (Ed.), The Late Cainozoic Palaeo-
climates of the Southern Hemisphere. Balkema, Rotterdam, pp. 445–453.
Besler, H., 1996. The Tsondab sandstone in Namibia and its significance for the Namib
erg. South African Journal of Geology 99, 77–87.
Blanco, G., Rajesh, H.M., Germs, G.J.B., Zimmermann, U., 2009. Chemical composition
and tectonic setting of chromian spinels from the Ediacaran–Early Paleozoic
Nama Group, Namibia. Journal of Geology 117, 325–341.
Blatt, H., 1978. Sediment dispersal from Vogelsberg Basalt, Hessen, West Germany.
Geologische Rundschau 67, 1009–1015.
Bluck, B.J., Ward, J.D., Cartwright, J., Swart, R., 2007. The Orange River, southern Africa:
an extreme example of a wave-dominated sediment dispersal system in the South
Atlantic Ocean. Journal of the Geological Society of London 164, 341–351.
Bremner, J.M., Rogers, J., Willis, J.P., 1990. Sedimentological aspects of the 1988 Orange
River floods. Transactions of the Royal Society of South Africa 47, 247–294.
Fig. 11. Sediment budget for the Namib Sand Sea. Estimated contributions of hinterland
sources are given for all studied samples across the erg. The coefficient of multiple cor-
relation Rmeasures the similarity between the observed and modelled detrital modes.
Main paths of longshore and aeolian sediment transport are shown (wind directions
and potential annual sand flux after Lancaster, 1985; Bristow and Lancaster, 2004).
187E. Garzanti et al. / Earth-Science Reviews 112 (2012) 173–189

Author's personal copy
Bristow, C.S., Lancaster, N., 2004. Movement of a small slipfaceless dome dune in the
Namib Sand Sea, Namibia. Geomorphology 59, 189–196.
Bristow, C.S., Duller, G.A.T., Lancaster, N., 2007. Age and dynamics of linear dunes in the
Namib desert. Geology 35, 555–558.
Cameron, K.L., Blatt, H., 1971. Durabilities of sand size schist and “volcanic”rock
fragments during fluvial transport, Elk Creek, Black Hills, South Dakota. Journal of
Sedimentary Petrology 41, 565–576.
Catuneanu, O., Wopfner, H., Eriksson, P.G., Cairncross, B., Rubidge, B.S., Smith, R.M.H.,
Hancox, P.J., 2005. The Karoo basins of south-central Africa. Journal of Asian
Earth Sciences 43, 211–253.
Compton, J.S., Maake, L., 2007. Source of the suspended load of the upper Orange River,
South Africa. South African Journal of Geology 110, 339–348.
Corbett, I., 1993. The modern and ancient pattern of sandflow through the southern
Namib deflation basin. International Association of Sedimentologists. Special
Publication 16, 45–60.
Corbett, I., Burrell, B., 2001. The earliest Pleistocene(?) Orange River fan-delta: an
example of successful exploration delivery aided by applied Quaternary research in
diamond placer sedimentology and palaeontology. Quaternary International 82, 63–73.
Cox, K.G., 1989. The role of mantle plumes in the development of continental drainage
patterns. Nature 342, 873–877.
Davies, G.R., Spriggs, A.J., Nixon, P.H., 2001. A non-cognate origin for the Gibeon
Kimberlite megacryst suite, Namibia: implications for the origin of Namibian
kimberlites. Journal of Petrology 42, 159–172.
de Kock, G., 2001. A reappraisal of the Namibian Damara stratigraphy in part of the
Southern Swakop Terrane and its implications to basin evolution. South African
Journal of Geology 104, 115–136.
de Villiers, S., Compton, J.S., Lavelle, M., 2000. The strontium isotope systematics of the
Orange River, Southern Africa. South African Journal of Geology 103, 237–248.
Dickinson, W.W., Ward, J.D., 1994. Low depositional porosity in eolian sands and sand-
stones, Namib desert. Journal of Sedimentary Research 64, 226–232.
Dietz, V., 1973. Experiments on the influence of transport on shape and roundness of
heavy minerals. Schweizerbart, Stuttgart. Contributions to Sedimentology 1, 69–102.
Dott, R.H., 2003. The importance of eolian abrasion in supermature quartz sandstones
and the paradox of weathering on vegetation-free landscapes. Journal of Geology
111, 387–405.
Dutta, P.K., Zhou, Z., dos Santos, P.R., 1993. A theoretical study of mineralogical matu-
ration of eolian sand. Geological Society of America, Special Paper 284, 203–209.
Folk, R.L., 1976. Reddening of desert sands: Simpson Desert, Northern Territory, Aus-
tralia. Journal of Sedimentary Petrology 46, 604–615.
Frimmel, H.E., Frank, W., 1998. Neoproterozoic tectono-thermal evolution of the Gariep
Belt and its basement, Namibia and South Africa. Precambrian Research 90, 1–28.
Galerne, C.Y., Neumann, E.-R., Planke, S., 2010. Magmatic differentiation processes in
saucer-shaped sills: evidences from the Golden Valley Sill in the Karoo Basin,
South Africa. Geosphere 6, 163–188.
Garzanti, E., Andò, S., 2007a. Heavy-mineral concentration in modern sands: implica-
tions for provenance interpretation. In: Mange, M.A., Wright, D.T. (Eds.), Heavy
Minerals in Use, Developments in Sedimentology Series, 58. Elsevier, Amsterdam,
pp. 517–545.
Garzanti, E., Andò, S., 2007b. Plate tectonics and heavy-mineral suites of modern sands.
In: Mange, M.A., Wright, D.T. (Eds.), Heavy Minerals in Use, Developments in
Sedimentology Series, 58. Elsevier, Amsterdam, pp. 741–763.
Garzanti, E., Vezzoli, G., 2003. A classification of metamorphic grains in sands based on
their composition and grade. Journal of Sedimentary Research 73, 830–837.
Garzanti, E., Canclini, S., Moretti Foggia, F., Petrella, N., 2002. Unraveling magmatic and
orogenic provenances in modern sands: the back-arc side of the Apennine thrust-
belt (Italy). Journal of Sedimentary Research 72, 2–17.
Garzanti, E., Andò, S., Vezzoli, G., Dell' Era, D., 2003. From rifted margins to foreland
basins: investigating provenance and sediment dispersal across desert Arabia
(Oman, UAE). Journal of Sedimentary Research 73, 572–588.
Garzanti, E., Andò, S., Vezzoli, G., Ali Abdel Megid, A., El Kammar, A., 2006. Petrology of
Nile River sands (Ethiopia and Sudan): sediment budgets and erosion patterns.
Earth and Planetary Science Letters 252, 327–341.
Garzanti, E., Andò, S., Vezzoli, G., 2008. Settling-equivalence of detrital minerals and
grain-size dependence of sediment composition. Earth and Planetary Science
Letters 273, 138–151.
Garzanti, E., Andò, S., Vezzoli, G., 2009. Grain-size dependence of sediment composi-
tion and environmental bias in provenance studies. Earth and Planetary Science
Letters 277, 422–432.
Garzanti, E., Andò, S., France-Lanord, C., Vezzoli, G., Censi, P., Galy, V., Najman, Y., 2010.
Mineralogical and chemical variability of fluvial sediments. 1. Bedload sand (Ganga-
Brahmaputra, Bangladesh). Earth and Planetary Science Letters 299, 368–381.
Geyer, G., 2005. The Fish River Subgroup in Namibia: stratigraphy, depositional envi-
ronments and the Proterozoic–Cambrian boundary problem revisited. Geological
Magazine 142, 465–498.
Goudie, A.S., Eckardt, F., 1999. The evolution of the morphological framework of the
central Namib desert, Namibia, since the Early Cretaceous. Geografiska Annaler
81A, 443–458.
Goudie, A.S., Watson, A., 1981. The shape of desert sand dune grains. Journal of Arid
Environments 4, 185–190.
Haskins, D.R., Bell, F.G., 1995. Drakensberg basalts: their alteration, breakdown and
durability. Quarterly Journal of Engineering Geology 28, 287–302.
Hu, Z., Gao, S., 2008. Upper crustal abundances of trace elements: a revision and
update. Chemical Geology 253, 205–221.
Ingersoll, R.V., 1984. Felsic sand/sandstone within mafic volcanic provinces: rift sand/
sandstone shows no basaltic provenance. Geological Society of America, Abstracts
with Programs 16 548 pp.
Ingersoll, R.V., Bullard, T.F., Ford, R.L., Grimm, J.P., Pickle, J.D., Sares, S.W., 1984. The
effect of grain size on detrital modes: a test of the Gazzi–Dickinson point-
counting method. Journal of Sedimentary Petrology 54, 103–116.
Ingersoll, R.V., Kretchmer, A.G., Valles, P.K., 1993. The effect of sampling scale on
actualistic sandstone petrofacies. Sedimentology 40, 937–953.
Jacob, J., Ward, J.D., Bluck, B.J., Scholz, R.A., Frimmel, H.E., 2006. Some observations on
diamondiferous bedrock gully trapsites on Late Cainozoic, marine-cut platforms
of the Sperrgebiet, Namibia. Ore Geology Reviews 28, 493–506.
Jacobs, J., Pisarevsky, S., Thomas, R.J., Becker, T., 2008. The Kalahari Craton during the
assembly and dispersal of Rodinia. Precambrian Research 160, 142–158.
Jacobson, P.J., Jacobson, K.M., Seely, M.K., 1995. Ephemeral Rivers and their Catch-
ments: Sustaining People and Development in Western Namibia. Desert Research
Foundation of Namibia, Windhoek. 160 pp.
Johnson, M.R., 1991. Sandstone petrography, provenance and plate tectonic setting in
Gondwana context of the southeastern Cape–Karoo Basin. South African Journal
of Geology 94, 137–154.
Johnson, M.R., van Vuuren, C.J., Hegenberger, W.F., Key, R., Shoko, U., 1996. Stratigraphy
in the Karoo Supergroup in southern Africa: an overview. Journal of African Earth
Sciences 23, 3–15.
Johnsson, M.J., 1993. The system controlling the composition of clastic sediments. In:
Johnsson, M.J., Basu, A. (Eds.), Processes Controlling the Composition of Clastic
Sediments: Geol. Soc. America, 284, pp. 1–19.
Jordaan, J.M., Clark, D., 1988. Regime changes in the Caledon River associated with
Welbedacht Barrage: physical model, prototype and theoretical correlations. In:
White, W.R. (Ed.), Int. Conf. on River Regime. Hydraulic Research Ltd., J. Wiley,
Wallingford, England, pp. 375–384. 18–20 May 2008.
Jourdan, F., Féraud, G., Bertrand, H., Kampunzu, A.B., Tshoso, G., Watkeys, M.K., Le Gall,
B., 2005. Karoo large igneous province: brevity, origin, and relation to mass extinc-
tion questioned by new
40
Ar/
39
Ar age data. Geology 33, 745–748.
Jung, S., Mezger, K., 2003. Petrology of basement-dominated terranes: I. Regional meta-
morphic T–t path from U–Pb monazite and Sm–Nd garnet geochronology (Central
Damara Orogen, Namibia). Chemical Geology 198, 223–247.
Kocurek, G., Lancaster, N., Carr, M., Frank, A., 1999. Tertiary Tsondab Sandstone Forma-
tion: preliminary bedform reconstruction and comparison to modern Namib sand
sea dunes. Journal of African Earth Sciences 29, 629–642.
Komar, P.D., 2007. The entrainment, transport and sorting of heavy minerals by
waves and currents. In: Mange, M.A., Wright, D.T. (Eds.), Heavy Minerals in Use.
Developments in Sedimentology Series, 58. Elsevier, Amsterdam, pp. 3–48.
Elsevier, Amsterdam.
Kuenen, P.H., 1959. Experimental abrasion; 3, fluviatile action on sand. American
Journal of Science 257, 172–190.
Kuenen, P.H., 1960. Experimental abrasion; 4, Eolian action. Journal of Geology 68,
427–449.
Lancaster, N., 1985. Winds and sand movement in the Namib Sand Sea. Earth Surface
Processes and Landforms 10, 607–619.
Lancaster,N., 1989. The NamibSand Sea: Dune Forms,Processes, and Sediments. Balkema,
Rotterdam. 200 pp.
Lancaster, N., 2002. How dry was dry? Late Pleistocene palaeoclimates in the Namib
desert. Quaternary Science Reviews 21, 769–782.
Lancaster, N., Ollier, C.D., 1983. Sources of sand for the Namib Sand Sea. Zeitschrift für
Geomorphologie 45, 71–83 Supplement band.
Lancaster, N., Schaber, G.G., Teller, J.T., 2000. Orbital radar studies of paleodrainages in
the Central Namib desert. Remote Sensing of Environment 71, 216–225.
Lustrino, M., Melluso, L., Brotzu, P., Gomes, C.B., Morbidelli, L., Muzio, R., Ruberti, E.,
Tassinari, C.C.G., 2005. Petrogenesis of the Early Cretaceous Valle Chico igneous
complex (SE Uruguay): relationships with Paraná–Etendeka magmatism. Lithos
82, 407–434.
Makhoalibe, S., 1984. Suspended sediment transport measurement in Lesotho.
Challenges in African Hydrology and Water Resources: IAHS Publ., 144,
pp. 313–321.
Makhoalibe, S., 1999. Management of water resources in the Maloti/Drakensberg
mountains of Lesotho. Ambio 28, 460–461.
Mange, A., Maurer, H.F.W., 1992. Heavy Minerals in Colour. Chapman and Hall, London.
147 pp.
Marsland, P.S., Woodruff, J.G., 1937. A study of the effects of wind transportation on
grains of several minerals. Journal of Sedimentary Petrology 7, 18–30.
McBride, E., Picard, D.M., 1987. Downstream changes in sand composition, roundness
and gravel size in a short-headed, high-gradient stream, Northwestern Italy.
Journal of Sedimentary Petrology 57, 1018–1026.
McLennan, S.M., 2001. Relationships between the trace element composition of
sedimentary rocks and upper continental crust. Geochemistry, Geophysics, Geo-
systems 2 2000GC000109.
Mehring, J.L., McBride, E.F., 2007. Origin of modern quartzarenite beach sands in a
temperate climate, Florida and Alabama, USA. Sedimentary Geology 201,
432–445.
Melluso, L., Cucciniello, C., Petrone, C.M., Lustrino, M., Morra, V., Tiepolo, M.,
Vasconcelos, L., 2008. Petrology of Karoo volcanic rocks in the southern Lebombo
monocline, Mozambique. Journal of African Earth Sciences 52, 139–151.
Miller, R.M., 1983. The Pan-African Damara Orogen of South West Africa/Namibia. In:
Miller, R.M. (Ed.), Evolution of the Damara orogen of South West Africa/Namibia:
Geol. Soc. South Africa, Spec. Publ., 11, pp. 431–515.
Miller, R.M., 2008. The Geology of Namibia, 3 vol. Ministry of Mines and Energy,
Windhoek.
Moecher, D.P., Samson, S.D., 2006. Differential zircon fertility of source terranes and
natural bias in the detrital zircon record: implications for sedimentary provenance
analysis. Earth and Planetary Science Letters 247, 252–266.
188 E. Garzanti et al. / Earth-Science Reviews 112 (2012) 173–189

Author's personal copy
Moore, A., Blenkinsop, T., 2002. The role of mantle plumes in the development of
continental-scale drainage patterns: the southern African example revisited.
South African Journal of Geology 105, 353–360.
Moore, A., Blenkinsop, T., Cotterill, F., 2008. Controls on post-Gondwana alkaline volca-
nism in Southern Africa. Earth and Planetary Science Letters 268, 151–164.
Moore, A., Blenkinsop, T., Cotterill, F., 2009. Southern African topography and erosion
history: plumes or plate tectonics? Terra Nova 21, 310–315.
Morimoto, N., Fabries, J., Ferguson, A.K., Ginzburg, I.V., Ross, M., Seifert, F.A., Zussman, J.,
Aoki, K., Gottardi, G., 1988. Nomenclature of pyroxenes. Mineralogy and Petrology
39, 55–76.
Morin, E., Grodek, T., Dahan, O., Benito, G., Kulls, C., Jacoby, Y., Van Langenhove, G.,
Seely, M., Enzel, Y., 2009. Flood routing and alluvial aquifer recharge along the
ephemeral arid Kuiseb River, Namibia. Journal of Hydrology 368, 262–275.
Muhs, D.R., 2004. Mineralogical maturity in dunefields of North America, Africa and
Australia. Geomorphology 59, 247–269.
Nesbitt, H.W., Young, G.M., 1982. Early Proterozoic climates and plate motions inferred
from major element chemistry of lutites. Nature 299, 715–717.
Padoan, M., Garzanti, E., Harlavan, Y., Villa, I.M., 2011. Tracing Nile sediment sources by
Sr and Nd isotope signatures (Uganda, Ethiopia, Sudan). Geochimica et Cosmochi-
mica Acta 75, 3627–3644.
Parker, A., 1970. An index of weathering for silicate rocks. Geological Magazine 107,
501–504.
Powers, M.C., 1953. A new roundness scale for sedimentary particles. Journal of Sedi-
mentary Petrology 23, 117–119.
Quinn, K.A., Byrne, R.H., Schijf, J., 2006. Sorption of yttrium and rare earth elements by
amorphous ferric hydroxide: influence of pH and ionic strength. Marine Chemistry
99, 128–150.
Rogers, J., 1977. Sedimentation on the continental margin off the Orange River and the
Namib Desert. Geol. Survey/Univ. Cape Town Marine Geosci. Group Bull., 7. 162 pp.
Rogers, J., Rau, A.J., 2006. Superficial sediments of the wave-dominated Orange River
delta and the adjacent continental margin off southwestern Africa. African Journal
of Marine Science 28, 511–524.
Rooseboom, A., Lotriet, H.H., 1992. The new sediment yield map for southern Africa.
Erosion and Sediment Transport Monitoring Programmes in River Basins: IAHS
Publ., 210, pp. 527–538.
Rooseboom, A., Harmse, H.J. von M., 1979. Changes in sediment load of the Orange
River during the period 1929–1969. Hydrology of Areas of Low Precipitation:
IAHS Publ., 128, pp. 459–479.
Russell, R.D., Taylor, R.E., 1937. Roundness and shape of Mississippi River sands. Journal
of Geology 45, 225–267.
Sawadogo, O. 2008. Analysis of observed reservoir sedimentation rates in South Africa.
Thesis, University of Stellenbosch, http://www.aims.ac.za/resources/archive/2007/
ousmane.pdf.
Saylor, B.Z., Grotzinger, J.P., Germs, G.J.B., 1995. Sequence stratigraphy and sedimentol-
ogy of the Neoproterozoic Kuibis and Schwarzrand Subgroups (Nama Group),
southwestern Namibia. Precambrian Research 73, 153–171.
Schlüter, T., 2006. Geological Atlas of Africa, with Notes on Stratigraphy, Tectonics, Eco-
nomic Geology, Geohazards and Geosites of Each Country. Springer, Heidelberg.
272 pp.
Ségalen, L., Rognon, P., Pickford, M., Senut, B., Emmanuel, L., Renard, M., Ward, J., 2004.
Reconstitution of dune morphologies and palaeowind regimes in the Proto-Namib
since the Miocene. Bulletin de la Societe Géologique de France 175, 537–546.
Shukri, N.M., 1950. The mineralogy of some Nile sediments. Quarterly Journal Geolog-
ical Society of London 105, 511–534.
Smith, R.M.H., Eriksson, P.G., Botha, W.J., 1993. A review of the stratigraphy and sedi-
mentary environments of the Karoo-aged basins of Southern Africa. Journal of Af-
rican Earth Sciences 16, 143–169.
Spaggiari, R.I., Bluck, B.J., Ward, J.D., 2006. Characteristics of diamondiferous Plio-
Pleistocene littoral deposits within the palaeo-Orange River mouth, Namibia. Ore
Geology Reviews 28, 475–492.
Stollhofen, H., Stanistreet, I.G., Bangert, B., Grill, H., 2000. Tuffs, tectonism and glacially
related sea-level changes, Carboniferous–Permian, southern Namibia. Palaeogeo-
graphy, Palaeoclimatology, Palaeoecology 161, 127–150.
Swanevelder, C.J., 1981. Utilising South Africa's largest river: the physiographic back-
ground to the Orange River scheme. GeoJournal Suppl. 2, 29–40.
Sweeney, R.J., Duncan, A.R., Erlank, A.J., 1994. Geochemistry and petrogenesis of
Central Lebombo basalts of the Karoo igneous province. Journal of Petrology 35,
95–125.
Taylor, S.R., McLennan, S.M., 1995. The geochemical evolution of the continental crust.
Reviews of Geophysics 33, 241–265.
Trask, C.B., Hand, B.M., 1985. Differential transport of fall-equivalent sand grains, Lake
Ontario, New York. Journal of Sedimentary Petrology 55, 226–234.
Twenhofel, W.H., 1945. The rounding of sand grains. Journal of Sedimentary Petrology
15, 59–71.
van den Boogaart, K.G., Tolosana-Delgado, R., 2008. “Compositions”: a unified R pack-
age to analyze compositional data. Computers & Geosciences 34, 320–338.
Vermeesch, P., Fenton, C.R., Kober, F., Wiggs, G.F.S., Bristow, C.S., Xu, S., 2010. Sand res-
idence times of one million years in the Namib Sand Sea from cosmogenic nuclides.
Nature Geoscience. doi:10.1038/NGEO985.
Vezzoli, G., Garzanti, E., 2009. Tracking paleodrainage in Pleistocene foreland basins.
Journal of Geology 117, 445–454.
Walden, J., White, K., 1997. Investigation of the controls on dune colour in the Namib
sand sea using mineral magnetic analyses. Earth and Planetary Science Letters
152, 187–201.
Ward, J.D., 1988. Eolian, fluvial and pan (playa) facies of the Tertiary Tsondab Sand-
stone Formation in the central Namib Desert, Namibia. Sedimentary Geology 55,
143–162.
Weltje, G.J., 1997. End-member modelling of compositional data: numerical–statistical
algorithms for solving the explicit mixing problem. Journal of Mathematical Geol-
ogy 29, 503–549.
White, K., Walden, J., Gurney, S.D., 2007. Spectral properties, iron oxide content and
provenance of Namib dune sands. Geomorphology 86, 219–229.
189E. Garzanti et al. / Earth-Science Reviews 112 (2012) 173–189