Paleomagnetism of The Paleozoic Glacial Sediments of Northern Ethiopia; A Contribution Towards African Permian Paleogeography.

Full text

Geophysical Journal International
Geophys. J. Int. (2013) doi: 10.1093/gji/ggt336
GJI Geomagnetism, rock magnetism and palaeomagnetism
Palaeomagnetism of Palaeozoic glacial sediments of Northern
Ethiopia: a contribution towards African Permian palaeogeography
Tesfaye Kidane,1,2 Valerian Bachtadse,2Mulugeta Alene1and Uwe Kirscher2
1School of Earth Science, College of Natural Science, Addis Ababa University, P.O. Box 1176, AA, Ethiopia. E-mail: tesfayek@yahoo.com
2Geophysics Section, Department of Earth and Environmental Sciences, LMU, Germany
Accepted 2013 August 20. Received 2013 August 16; in original form 2013 April 3
SUMMARY
One hundred fourteen oriented palaeomagnetic core samples were collected from 13 palaeo-
magnetic sites on subhorizontal to tilted glacial sediments at five localities of Northern
Ethiopia. Combined alternating field (AF) and stepwise thermal demagnetization techniques
were successfully applied to resolve the complete directional spectr um. A viscous remagne-
tization (VRM) and one stable component of magnetization were identified in most of the
specimens. The VRM is removed between a temperature range of 120–350 ◦CandAFof
up to 30 mT. Further heating until ∼650 ◦C results in smooth decay of the natural remanent
magnetization (NRM) intensity to about 50 per cent and the rest of the NRM is efficiently
removed by heating to 690 ◦C, while only 30–50 per cent of NRM is removed by the maximum
AF available suggesting haematite as remanence carrier. Results of the magnetization decay
curve plots and rock magnetic analyses using the variable field translation balance indicated
the presence of magnetite with minor goethite, pyrrhotite as well. The high stability component
defining a straightline segment, starting 350 ◦C and/or 30 mT is mostly directed towards the
origin and interpreted as the characteristic remanent magnetizations (ChRMs). The direction
of magnetization is determined both by best-fitting line using the least-square technique of
Kirschvink and remagnetization circles of Halls for few unresolved overlapping components.
The site mean directions of the sediments from two sites are normal polarity and are close to
present-day field directions at the sample site. The site mean directions from 11 sites, on the
other hand, are reversed in polarity with better grouping in the tilt-corrected coordinate and
pass the McFadden fold test. This overall site mean direction is Dec =143.4o,Inc=58.8o
(N=11, α95 =9.7o) with a corresponding mean pole position of Lat =26.0o,Lon=249.5o
(N=11, A95 =13.1o). This geomagnetic pole position is later rotated into West Africa coor-
dinates to allow for extensional rifting in the Benue Trough about an Euler pole position, at
19.2oN, 352.6oE through an angle −6.3o(clockwise). The resulting pole position is located at
φs=246.6oE, λs=31.8oS(N=11, A95 =13.1o), this pole with its 95 per cent confidence
circle intersects the 270–310 Ma, segment of the APW path for West Africa consistent with
ages of between late Carboniferous and early Permian. The result also implies that the Late
Carboniferous Dwyka land ice sheet had probably extended more than 1000 km further north
to Ethiopia than previously known.
Key words: Plate motions; Palaeomagnetism applied to tectonics; Rock and mineral
magnetism; Africa.
1 INTRODUCTION
There is an ongoing debate about palaeogeographic configurations
of Pangea in the Late Palaeozoic Era (e.g. Van der Voo 1993;
McElhinny & McFadden 2000; Muttoni et al.2003;Domeieret al.
2012). A comprehensive and detailed historical development of
Pangea reconstruction and the opposing arguments in the Palaeo-
zoic Era has recently been given elsewhere (Domeier et al. 2012)
and briefly presented below.
At least two competing palaeogeographic reconstructions models
have been proposed over the years; Pangaea A and Pangaea B. Re-
construction type A is a slight modification of the original Wegener’s
(1915) reconstruction. It has widely been accepted as the likely
configuration in the Early Jurassic just prior to the opening of the
C
⃝The Authors 2013. Published by Oxford University Press on behalf of The Royal Astronomical Society. 1
Geophysical Journal International Advance Access published October 3, 2013
at University of Minnesota on April 17, 2015http://gji.oxfordjournals.org/Downloaded from

2T. Kidane et al.
Atlantic Ocean (Van der Voo 1993); because of the perfect matching
of conjugate seafloor magnetic anomalies and marine fracture zones
across the Atlantic (e.g. Klitgord & Schouten 1986) and supported
by palaeomagnetic data for Early Jurassic (Irving 1977). However,
its validity was later disputed (Irving 1977; Kanasewich et al.1978;
Morel & Irving 1981). Irving (1977) has demonstrated the exis-
tence of significant difference between the reference latitudes of
Laurasia and Gondwana for pre-Jurassic time with a profound im-
plication that Gondwana must have been farther north, relative to
North America, than its position in the Pangea A configuration, dur-
ing the Permian and the Triassic. Consequently, significant crustal
misfit and overlap (>1000 km) between Laurasia and Gondwana
results; to eliminate this overlap, a relative eastward translation of
Gondwana of ∼3500 km was unavoidable.
In order to resolve this paradox, a new palaeogeographic recon-
struction, named as type B, was introduced by Irving (1977) and
with further additional palaeomagnetic data, the necessity of type B
reconstruction hypothesis and its validity for Carboniferous to Tri-
assic (Kanasewich et al. 1978; Morel & Irving 1981) was asserted.
Reconstruction type B, however, could not gain wider acceptance
by the geosciences community, owing to the fact that available geo-
logical and geophysical data overwhelmingly assert that the Atlantic
Ocean opened from Pangea A position. This would require Pangea
B to transform to Pangea A in the Permian to Late Triassic (Van der
Voo 1993; McElhinny & McFadden 2000; Muttoni et al. 2009). The
skepticism on the Pangea B configurations by many geoscientists
lies on the integration of suspicious poor-quality palaeomagnetic
data with poor age control from red beds of South America and
Africa and potential inclination error from inclination shallowing
in sediments potentially produce latitudinal artefacts (Rochette &
Vandamme 2001; Muttoni et al. 2003).
2 GEOLOGICAL SETTING OF THE
PALAEOZOIC DIAMICTITE OF TIGRAI
Patchy exposures of diamictite consisting of lithified, poorly sorted
terrigenous glacial sediments, are found in different parts of the
Tigrai region (Northern Ethiopia). This diamictite constitutes a
distinctive succession of strata between the underlying a late
Proterozoic, low-grade basement rocks of the Arabian Nubian
Shield (ANS) and overlying Mesozoic sedimentary sequences.
The glaciogenic sediments, initially identified as bluish–pink shale
that outcrop west of Adigrat town (Blanford 1869, 1870), were
regarded as part of the Mesozoic sedimentary sequences (Mohr
1965). Merla & Minucci (1938) had described them as red brick
to black lenses of shale outcrops near Adigrat and south of
Hauzien; also noted occurrences of beds of conglomerate with big
pebbles.
Dow et al. (1971) were the first to identify the glacial origin
of these rocks, with detailed lithological descriptions and regional
correlations recognized it as tillite. They differentiated glacigenic
sandstone and tillite facies, which are named as ‘Enticho Sandstone’
and ‘Edaga Arbi Glacials’, respectively (Beyth 1972a,b, 1973). The
tillites make up 10 per cent of the whole Palaeozoic sequence and
were laid down in N–S trending furrows (Beyth 1972b). The age
of these sediments has so far not been accurately determined. A
precise stratigraphic assignment is not possible because of the lack
of age diagnostic fossils, and hence their regional stratigraphic po-
sition remains ambiguous. Consequently, they have been attributed
either to the extensive Ordovician glaciation centred in southern
Algeria (Beuf et al. 1971; Vaslet 1990; Semtner & Klitzsch 1994;
Clark-Lowes 2005; Moreau et al. 2005), or to the early Permian—
late Carboniferous glaciation centred in southern Africa (Karroo,
Du Toit 1953; Furon 1963). Dow et al. (1971) and Saxena & As-
sefa (1983), based on rare fossil Siphonoromid impressions within
the Edaga Arbi Glacials, provided the first supporting evidence
for an Ordovician age. This Palaeontological argument was later
questioned (Sacchi et al. 2007), who stated that an alternative in-
terpretation of these findings was equally compatible with a late
Permian age. I n addition, Beyth (1972a) re ferring to a palyno-
logic analysis of the Edaga Arbi Glacials suggested ‘not older than
Devonian’.
In an attempt to determine the age of the Edaga Arbi Glacials
and the Enticho Sandstone palaeomagnetically, Shackleton & Lo-
max (1974) published results from two non-procedural hand sam-
ples from Enticho and Fincha, respectively, yielding palaeomag-
netic pole positions at 293◦E longitude, and 32◦N, latitude with
A95 =27.0◦for Enticho and 250◦E longitude and 35◦N latitude
with A95 =3.0◦,respectively.Theseresultswerecorrelatedwiththe
Tanzanian red beds of early Permian age (McElhinny et al. 1968).
This report was based on non-demagnetized data and not fulfilling
any of the palaeomagnetic quality criteria such as the ones proposed
by Van der Voo (1993). Their obtained pole therefore is of limited
use. Bussert & Schrank (2007) based on Palynomorphs identified
and extracted from the glacigenic sediments assigned an age of lat-
est Carboniferous—early Permian period to the Tillite and upper
part of Enticho Sandstone. However, they concluded that the lower
part of the Enticho Sandstone is an older sedimentary unit with an
age of upper Ordovician equivalent to those in Eritrea.
Sacchi et al. (2007) discovered volcanic pebbles in the diamictite
sequences in Hauzien—Megab (HM), Mai Kenetal—Edaga Arbi
(MK), which they considered as a supporting evidence for a Permian
age on the grounds of regional geology. However, their assumption
is inconsistent with regional/global tectonics as the mid-Permian to
Middle Triassic was a tectonically stable interval (e.g. Hallam 1983;
Smith & Livermore 1991).
The purpose of this palaeomagnetic study is to determine the
age of deposition of these sediments and to distinguish which one
of the two possible ages described earlier is right. This will be
done by comparing the results of the palaeomagnetic pole position
obtained from the study with the African apparent polar wander
path (AAPWP) curve (Besse and Courtillot 2002; McElhinny et al.
2003; Torsvik et al. 2008; Torsvik et al. 2012). For this purpose,
we have collected samples from Negash, Hauzien–Megab–Koraro
(HMK), Enticho and Edaga Arbi areas (EDAa) of Tigrai region in
Northern Ethiopia.
3 PALAEOMAGNETISM
3.1 Sampling
During two sampling seasons, 114 palaeomagnetic core samples
from 13 different palaeomagnetic sites covering five geographic lo-
cations (Enticho, HMK, Edaga–Arbi type area, Negash and its sur-
roundings, Fig. 1) were collected using a portable gasoline powered
drill. During the first sampling season, samples were collected from
Enticho which are coded as Enticho Palaeozoic sandstone (EPSST)
and NPSST); while during the second season samples were col-
lected from Negash village in the Belessa area, EDA and HMK area
coded, respectively, as Negash–Belessa sandstone (NBLS), EDA
and HMK area. All samples were oriented with a standard magnetic
compass. Each palaeomagnetic site covers a different stratigraphic
at University of Minnesota on April 17, 2015http://gji.oxfordjournals.org/Downloaded from

African Permian palaeogeography 3
Figure 1. (a) Digital elevation map (DEM) of Ethiopia and surrounding regions, white rectangle shows locations of the studied area in Tigrai region.
(b) A blow-up DEM showing details of white rectangle, that is, geographic locations of study sites. (c) A representative picture to show outcrop patterns
of the Palaeozoic and Mesozoic Sediments.
level within the two glaciogenic sediments. In the Negash area,
exposures of distinctive greyish coloured and thinly stratified dip-
ping beds (strike/dip—130o/27oSW) of the Edaga Arbi Glacials
are exposed just below the steep slope forming Triassic aged ‘Adi-
grat Sandstone’. A prominent E–W trending fault upthrown the
underlying basement rocks to the north. A valley that developed
later between the basement (north) and sediments (south) along
this E–W running fault line exposes a cross-bedded (with orienta-
tion of cross-bed, 120o/12oNE) medium to coarse-grained whitish
coloured tilted (095o/16oSW) sandstone beds adjacent to the Edaga
Arbi Glacial rocks at Negash Belessa site. A total of six sites, four
from the Edaga–Arbi Glacial strata and two from Enticho Sandstone
beds were collected. In the HMK, along a spectacular geomorpho-
logic outcrop of the steep slope forming Adigrat Sandstone and
at University of Minnesota on April 17, 2015http://gji.oxfordjournals.org/Downloaded from

4T. Kidane et al.
(c)
Figure 1. (Continued.)
the gentler slope forming Edaga Arbi Glacials (Fig. 1c), two sites
were collected from the Edaga Arbi Glacial facies with flat lying
attitudes. In the Enticho-type area, palaeomagnetic sampling was
carried out at the bottom of the section from two stratigraphic levels,
where finer-grained beds were identified. Finally, three sites from
the Edaga Arbi facies near the Edaga Arbi village (Figs 1a–c) were
collected.
3.2 Laboratory analyses
A total of 190 specimens out of the 228 specimens prepared from
the 114 samples were measured. Both thermal (TH) or alternating
field (AF) demagnetization techniques were used in order to resolve
the directional spectrum. In addition, representative samples were
used to determine the rock-magnetic properties. All palaeomagnetic
and rock-magnetic experiments were carried in the Palaeomagnetic
laboratory facility at Ludwig-Maximilians-Universit¨
at M¨
unchen.
Pilot samples were demagnetized thermally in 22 steps from room
temperature up to 700 ◦C as well as in AFs from 0 to 100 mT in 13
steps. Demagnetization steps during routine laboratory procedures
were reduced to 11–15 steps.
3.2.1 Rock magnetic properties
Six representative specimens from rocks of the glaciogenic and
sandstone facies were subjected to isothermal remanent magnetiza-
tion (IRM) experiments (Fig. 2a). An initial steep rise in IRM up to
an applied field of ∼300 mT is obser ved in all samples. A further
steep (e.g. NPSST2-6B, Type B) or less steep (e.g. NPSST3-7B,
Type C) increase in intensity was observed in some samples in
fields up to ∼500 mT. None of the samples studied reaches satura-
tion at maximum fields of 2250 mT. The rapid increase in intensity
in magnetizing fields of up to 300 mT is diagnostic for magnetite,
whereas the fact that no saturation is achieved at external fields
of 2250 mT points towards the presence of a second, high coer-
cive magnetic phase very likely to be haematite and/or goethite AF
demagnetization of the IRM supports this interpretation. Depend-
ing on the type of IRM acquisition (rapid gain in intensity at low
external fields versus significant gain in intensity at higher exter-
nal fields), AF demagnetization is successful in removing between
10 and 70 per cent of the total IRM. This generally indicates
that most of the specimens are characterized by a high hcmag-
netic mineral(s) and probably significant contribution of haematite
(Fig. 2b).
3.3 Microscopic properties
Petr ographic analysis was carried out using refl ecting light mi cro-
scope on selected polished surfaces of cylindrical samples, and oil
immersion for the highest magnification. Observation under micro-
scope reveals that the opaque minerals make up less than 5 per cent
of the total material and in almost all cases identical optical prop-
erties are observed. Magnetic grains have lamellae with clear band-
ing of haematite (white—thicker stripe) and ilmenite (dark grey in
at University of Minnesota on April 17, 2015http://gji.oxfordjournals.org/Downloaded from

African Permian palaeogeography 5
Figure 2. (a) IRM acquisition experiment for representative specimens from Edaga Arbi Glacial and Enticho Sandstone facies. (b) AF demagnetization curve
of the IRM experiment in (a), the corresponding specimens names are given for each colour codes.
Figs 3a–c) as the dominant magnetic mineralogy. The size of these
grains is mostly lower than 2 µm, but can reach up to 2.5 µm (Fig. 3).
The second case is where magnetite (brighter whitish lamellae) and
hemo-ilmenite (grey lamellae) in which both show birefringent fine
alteration products (possibly calcite and sericite) and also a yel-
low bright surface possibly sulphide alterations indicting possible
existence of iron sulphides as well (Fig. 3d). The magnetic grain
size ranges (Fig. 3, compare using given scale) are consistent with
at University of Minnesota on April 17, 2015http://gji.oxfordjournals.org/Downloaded from

6T. Kidane et al.
Figure 3. Observation under the highest magnification (50×) objectives with oil immersion: (a) grain showing lamellae of haematite and ilmenite as bands,
with fine haematites spots; (b) haematite and ilmenite lamellae forming bands; (c) irregular haematite (grey colour); (d) hemo-ilmenite and magnetite (light
grey) with bright non-opaque materials (alteration products), some have yellow fresh surface (probably sulphides).
Figure 4. Hysteresis cycles with effect of paramagnetism corrected for representative specimens, all measured specimens show narrow loop with typical low
remanent saturation magnetization.
single-domain (SD) haematite and its alterations and multidomain
(MD) grains of magnetite.
3.4 Hysteresis and thermomagnetic curves
The hysteresis cycles and thermomagnetic curves were measured
using the variable field translation balance (Krasa et al. 2007) in
order to better distinguish between the coexisting magnetic materi-
als. The hysteretic cycles of all measured specimens show typical
loops with rapid rise until applied fields of 200 mT followed by a
gradual increase up to external fields of 300 mT. At applied fields
of higher than 300 mT, most of the specimens reached their sat-
uration (Fig. 4). The behaviours of all samples are identical and
characterized by a distinctive narrow loop with very small amounts
of remanent saturation, which is indicative of the presence of MD
magnetic grains. The thermomagnetic experiment was carried out
for the same number of representative samples. The samples were
heated to 700 oC under field-free conditions. In these samples, the
at University of Minnesota on April 17, 2015http://gji.oxfordjournals.org/Downloaded from

African Permian palaeogeography 7
Figure 5. (a) Thermomagnetic curves for representative samples both with irreversible curves. Specimens EDA3-6A1 & HMK1-2A1 show their cooling curve
(blue) having higher and lower saturation magnetization, respectively, than the initial magnetizations. Both also show consistent two inflections, one between
300 and 400 oCandthesecondoneat580oC possibly corresponding to pyrrhotite and magnetite magnetic mineralogy. (b) Left column shows thermomagnetic
curves for representative samples with low applied field (0.2 A) at low temperatures till 630 oC for the upper row, and till 550 oC for the lower row; and higher
fields of 5 A for both at temperature higher than the given and during cooling process. The right column shows a blow-up of the left columns for low saturation
magnetization values to see details. Upper column shows four distinct Curie temperatures, 120, 350, 580 and 680 oC, while the lower column is consistent with
three Curie temperatures, 350, 580 and 680 oC.
at University of Minnesota on April 17, 2015http://gji.oxfordjournals.org/Downloaded from

8T. Kidane et al.
heating and cooling curves are not reversible with two clear in-
flections, one around 350 oCandasecondoneataround580oC
(Fig. 5a). The curve is irreversible in some cases with the cooling
curve ending at higher (specimen EDA3-6A1, Fig. 5a) or lower
(specimen HMK1-2A1, Fig. 5b) saturation magnetizations (js) than
the starting magnetization. These inflection points correspond with
Curie temperature estimates of goethite and pyrrhotite converting to
haematite or magnetite after heating to 700 oC. However, in all cases,
the magnetizations dropped to zero at 600 oC of the heating cycle
suggesting dominance of magnetite and scarcity of haematite in the
thermomagnetic experiments. This observation is inconsistent with
the predominance of haematite in the IRM experiments and in NRM
intensity decay curves. Contribution of magnetite in the remanence
results was observed, only in the Enticho sandstones samples at Ne-
gash that recorded secondary magnetizations. In order to understand
the reasons why magnetite dominant in the thermomagnetic experi-
ments and haematite is nearly absent, supplementary measurements
were made on more specimens. The specimens were subjected to
a weak field of 7.4 mT during the heating cycle up to 550–630 oC
and a stronger field of 148 mT at higher temperatures; during the
cooling cycle, the same strong field was maintained throughout the
process. Results indicate three to four inflections with Curie points
at 120o(only for EDA3–6A22, Fig. 5b), 350, 580 and 680 oC. These
inflection points are consistent with Curie temperatures of goethite,
pyrrhotite, magnetite and haematite. This indicates that the jsdur-
ing thermomagnetic experiments may have been swamped by MD
magnetite (strong—js), which does not appear in remanence record
and concealed the haematite (small—js) signal, a major remanence
contributor of the studied rocks. In most samples, the remanence
carrier haematite has a Curie temperature of around 685 oC indi-
cating pure haematite (Fig. 5b—specimen EDA3–6A22) but in few
cases, it shows a gradual decrease in magnetization between 600
and 700 oC (Fig. 5b—specimen HMK1–2A2).
4 PALAEOMAGNETIC DIRECTIONS
Routine palaeomagnetic directional measurements for most spec-
imens treated by both TH and AF techniques resulted in a
clear and well-defined single component of magnetization with
a secondary component whose blocking temperature spectra are
partly overlapping (Figs 6a–c). Representative Zijderveld diagrams
(Figs 6a–c) are given from the two glaciogenic facies at three differ-
ent geographic locations, Edaga Arbi, Enticho and from HMK areas,
Figure 6. Examples of Zijderveld diagrams for twin specimens treated by AF and TH from representative geographic locations and Glacial facies (a) from
Edaba Arbi Glacials at Negash, (b) from Enticho sandstone at Enticho and (c) from Edaga Arbi Glacials at Hauzien–Megab Koraro sampling areas. In all the
studied specimens, the magnetic polarity is down and pointing south consistent with reversed configuration.
at University of Minnesota on April 17, 2015http://gji.oxfordjournals.org/Downloaded from

African Permian palaeogeography 9
Figure 6. (Continued.)
respectively. The first and low stability component which does not
necessarily define directions is usually removed by heating to ∼300
oC or subjecting to AF of <20 mT. However, for some samples,
this component resists until higher temperatures and AF fields. The
second and high stability component starts to be isolated by heating
above 350 oCorAFof∼30 mT. During thermal demagnetization,
only ∼30 per cent of magnetization is removed by heating up to 600
oC, whereas the remaining 70 per cent is removed between steps at
temperatures of up to 690 oC. Only ∼50 per cent of NRM is removed
by AF of 100 mT (the maximum demagnetizing field available in
the laboratory). However, the direction of magnetizations at high
temperature and high demagnetizing fields is similar to the direc-
tion isolated at lower steps of demagnetizations greater than 35mT
and above 350 oC. The high unblocking temperatures as well as
the high-coercivity components have similar directions and define
linear segment directed towards the origin, which are interpreted
as characteristic remanent magnetization (ChRM). In cases, the
high stability component has unresolved overlapping spectra and
unblocking temperatures and no stable linear segments were ob-
tained. The direction of magnetization for specimens with multiple
components have overlapping spectra and unblocking temperatures
is determined by remagnetization circles of Halls (1976, 1978),
while for those with stable straight line segments is determined by
the best-fitting line using the least-square technique of Kirschvink
(1980).
4.1 Site mean directions
The directions of magnetizations from sample to sample, site to site
and one location to the other shows some differences probably very
likely to reflect secular variation. The directions of magnetizations
determined by either principal component analysis or remagnetiza-
tion circles techniques are used to determine the site mean
Site mean directions at site level and overall mean direction are
calculated by using Fischer (1953) statistics for those having sta-
ble endpoints and McFadden & McElhinny (1988) statistics for
combined stable endpoints and great circles using palaeomac soft-
ware (Cogn ´
e 2003). The site mean directions of 11 sites in in situ
coordinates are given with two sites in grey colour from the En-
ticho Sandstones at Negash area, which are judged remagnetized
and excluded from further analyses (Fig. 7). The red coloured stars
stand for the overall mean directions in the in situ (Fig. 7a) and
tilt-corrected coordinates (Fig. 7b).
The resulting site mean directions from the 11 sites from the four
areas studied are exclusivelyof reversed polarity, which corresponds
to the Kiaman-reversed period and only two sites from the Enticho
sandstone in Negash area are of normal. However, since the in situ
mean directions of the two normal sites are close to the direction of
the present-day field in the sampling area, while the tilt-corrected co-
ordinate mean moves away from it to unrelated position. Due to this,
they are considered to be results of recent remagnetizations. The re-
maining 11 sites (Table 1) have been averaged and an overall in situ
at University of Minnesota on April 17, 2015http://gji.oxfordjournals.org/Downloaded from

10 T. Kidane et al.
Figure 6. (Continued.)
mean direction of Dg=130.0o,Ig=57.1o(α95 =13.3o,k=12.8,
N=11) and that for tectonic-corrected coordinates is Ds=143.4o,
Is=58.8o(α95 =9.7o,k=23.0, N=11) are obtained (red star
symbol in Fig. 7, Table 1). The Fischer (1953) precision parameter
(k) increases from 12.8oto 23.0o,whilethe95percentconfidence
interval decreases from 13.3oto 9.7oin tectonic-corrected coor-
dinates. Seven sites have horizontal bedding attitudes, while four
sites have gently tilting attitudes, as a result we have applied the
McFadden (1990) fold test. The test results in statistical parameter
ξ2of 6.99 in geographic coordinates that is greater than the critical
value of the statistical parameter ξcof 3.865 and 5.378 at 95 and
99 per cent confidence level, respectively. This result implies that
the in situ average is correlated to bedding and the palaeomagnetic
directions were acquired shortly after deposition.
Besides for consistency purposes, we have made comparison of
our tilt-corrected pole position with previous results from Africa
such as data from Dwyka System from South Africa (Opdyke et al.
2001), Jebel Nehoud ring complex in Kordofan, central Sudan
(Bachtadse et al.2002)andwiththeGondwanapolesfor260–
320 Ma ages window (McElhinny et al. 2003). Fig. 8 shows details
of this comparison; our data are consistent with all of the above data
and particularly strongly coincident with the palaeomagnetic pole
of Dwyka System of South Africa, suggesting that they are most
likely contemporary.
Virtual geomagnetic poles (VGPs) for the 11 sites were then
calculated (Fig. 9) and an overall mean in situ palaeomagnetic
pole position of Long=257.5oE, Latgs=20.2oS(A95 =16.5o,
N=11), and that for tectonic-corrected coordinate Lons=249.5oE,
Lats=26.0oS(A95 =13.1o,N=11; Fig. 9, Table 2). When
these both poles are transferred to West African coordinate,
they move towards the given APWP curve becoming more
compatible.
5 DISCUSSION AND INTERPRETATION
5.1 Age estimation
The age of the Edaga Arbi Glacials and Enticho sandstones in
Northern Ethiopia has been ambiguous. The absence of age di-
agnostic fossils and the presence of two possible distinct glacial
events in the region preclude the distinction between Ordovician
or Permo/Carboniferous age. Comparison of obtained palaeomag-
netic pole position with the APWP curve for Africa is one way of
at University of Minnesota on April 17, 2015http://gji.oxfordjournals.org/Downloaded from

African Permian palaeogeography 11
Figure 7. Stereographic projection showing site mean directions for 13 sites analysed. Two sites from coarse Enticho sandstones at Negash locality having
different directions are shown in grey colours and the rest 11 sites are shown in black full circles. The overall mean directions for in situ coordinates (a) and
tilt-corrected coordinates (b) are given. In both cases, red star symbol represents overall mean directions.
Tab l e 1 . Palaeomagnetic geographic site mean directions for the 11 averages together with over all mean directions for the
Palaeozoic rocks of Ethiopia are given. Mean 1 and mean 2, respectively, indicate overall mean directions including and
excluding two geographic sites where strata were tilted.
Sample name Strike/dip ND
gIgDsIsKα95
NBLS1 150/20 11 122.1 71.4 183.2 70.3 55.0 6.2
NBLS2 145/27 15 113.4 47.7 147.1 55.0 39.5 6.2
NBLS3 125/25 10 81.9 34.0 100.4 48.2 176.6 3.6
NPSST1 130/27 23 91.3 34.9 112.9 47.9 76.8 3.5
EPSST1 0/0 25 162.9 43.7 162.9 43.7 35.1 5.0
EPSST2 0/0 18 156.2 50.7 156.2 50.7 15.4 9.2
EDA1 0/0 13 149.9 61.1 149.9 61.1 134.4 3.6
EDA2 0/0 12 168.6 47.5 168.6 47.5 83.4 4.8
EDA3 0/0 11 149.3 56.7 149.3 56.7 164.6 3.6
HMK1 0/0 13 112.3 63.7 112.3 63.7 23.5 8.7
HMK2 0/0 13 140.9 72.1 140.9 72.1 116.8 3.9
Overall mean 11 130.0 57.1 143.4 58.5 23.0 9.7
Notes: Sample name; strike and dip, N, number of specimens; Dgand Ig, declination and inclination in geographic coordinates;
Dsand Is, declination and inclination in stratigraphic coordinates; K, Fisher precision parameter; α95, 95 per cent confidence
interval.
at University of Minnesota on April 17, 2015http://gji.oxfordjournals.org/Downloaded from

12 T. Kidane et al.
Figure 8. Comparison of our pole rotated in western coordinates (solid
red star symbol) with some other poles from the region. The solid red
cross symbol represents data from Dwyka system (Opdyke et al. 2001),
the solid diamond symbol is from Jebel Nehoud ring complex in Kordofan,
Sudan (Bachtadse et al. 2002), and also shown in McElhinny et al. (2003)
Gondwana APWP data for time interval between 260 and 320Ma.
testing or determining past relative positions of continents.
Fig. 10(a) shows spherical projection with the APWP curve of
Africa in north western African coordinates for the last 310 Ma as
compiled by Besse & Courtillot (1991, 2003) and Cogn´
e(2003)and
the obtained palaeomagnetic pole position for the Palaeozoic rocks
of this study. Red coloured open diamond and star symbols represent
the obtained palaeomagnetic pole in the in situ and tilt-corrected
coordinate, respectively, both in Northeast African coordinate. Red
coloured full diamond and star symbols represent pole in the in
situ and tilt-corrected coordinate, respectively, both transferred in
West African coordinates. The obtained in situ pole position from
Northern Ethiopia is shown in open red diamond symbol (Fig. 10a,
Table 2), while the tilt-corrected pole position is shown in open
red star symbol. These pole positions are rotated to co-ordinates
of West Africa (full red diamond and star symbols) to allow for
extensional rift system from the Benue Trough about an Euler pole
at 19.2oN, 352.6oE through an angle −6.3o(clockwise) (Lottes &
Rowley 1990; McElhinny et al. 2003). The transferred in situ pole
position is located at Long=255.2oE, Latg=26.1oS(A95 =16.5o),
and that for tectonic corrected is Lons=246.6oE, Lats=31.8oS
(A95 =13.1o) (full red diamond and star symbols, respectively,
Fig. 10a). The transferred tectonic-corrected pole position is the
most consistent with the APWP curve and with its 95 per cent con-
fidence circles intersects the path in the 270–310 Ma segment and
hence is compatible with an age between late Carboniferous and
early Permian. In addition, our obtained tilt-corrected pole posi-
tion is transferred in South African coordinates (Lons=249.2oE,
Lats=26.4oS, A95 =13.1o,N=11) about an Euler pole position,
at 40.5oE, 298.6oS, through an angle 0.7o(Torsvik & Cocks 2004;
Torsvik et al.2008)andcomparedwithAPWPcurveofGondwana
(Torsvik et al. 2012). This comparison considered the 95 per cent
confidence circles slightly overlaps with APWP corresponding to
ages of between 260 and 280 Ma, while it bears no similarity with
the APW path before 290 Ma (Fig. 10b).
Both these results indicate that the Palaeozoic glacial sediments
of Ethiopia have an age between 270 and 310 Ma, which is compati-
ble with the assigned age of latest Carboniferous—early Permian pe-
riod based on Palynomorphs from the glacigenic sediments (Bussert
& Schrank 2007). However, the claim by Bussert & Schrank (2007)
that the lower part of the Enticho sandstone is Ordovician in age
is refuted by this study as palaeomagnetic directions from both the
Edaga Arbi Tillite and Enticho Sandstone are not statistically dif-
ferent and that any existing age difference is only stratigraphic. Be-
cause our obtained pole position is strongly coincident with palaeo-
magnetic pole of Dwyka System from South Africa (Opdyke et al.
2001), the glaciogenic sediments in Northern Ethiopia are equiv-
alent to the Dwyka sediments of Southern Africa, also consistent
with south–north striations widespread in these areas. This implies
that the late Carboniferous Dwyka land ice sheet had probably ex-
tended further north than previously known and that the age of
the sediments is probably older than 300 Ma, consistent with vari-
ous age estimates for Dwyka group sediments, formed during the
Karoo Ice ages, in the Karoo region of South Africa (Hambrey
& Harland 1981; Deynoux et al. 1994; Rubidge 2005; Selden &
Nudds 2011).
6 CONCLUSIONS
Rock magnetic, microscopic studies and demagnetization behaviour
of the glaciogenic sediments of Northern Ethiopia vindicated that
they retained original magnetization of high quality, carried domi-
nantly by detrital haematite. When the overall in situ and tectonic-
corrected coordinates are determined; the precision parameter (k)
increases, while the 95 per cent confidence interval decreases in bed-
ding corrected coordinates yielding a positive fold test (McFadden
1990).
The palaeomagnetic pole position from the glacial sediment in
this study is remarkably coincident with a palaeomagnetic pole of
the Dwyka System from south Africa (Opdyke et al. 2001), with
the Gondwana APWP curve with ages of between 260 and 320Ma
(Besse & Courtillot 1991, 2002; McElhinny et al. 2003), and also
agree with the palaeomagnetic pole position obtained from igneous
rocks of the Jebel Nehoud ring Complexes of Kordofan, South
Sudan (Bachtadse et al.2002).Comparisonofthispalaeomag-
netic pole position with the current APWP curve for West Africa
gives ages ranges of between ≈270 and 310Ma, assigning ages
of Early Permian and Late Carboniferous. The result also implies
that the Late Carboniferous Dwyka land ice sheet had probably ex-
tended more than 1000 km further north than previously known and
agrees with the widespread south–north oriented basins and also fur-
row structures reported in the Palaeozoic rocks outcrops, Northern
Ethiopia.
The data being better quality and coming from North Eastern
Africa where no data were previously reported potentially contribute
solving earlier ambiguities on early Permian—late Carboniferous
palaeomagnetic data of Gondwana and potentially contribute to-
wards the new road of reconciliations as more data are produced
from this part of the world.
ACKNOWLEDGEMENTS
We are very grateful for the fellowship support from Alexander
von Humboldt Fellowship that made this investigation possible.
We also thank the Research and Publication office of the Addis
Ababa University for funding the field work. We are very grateful
to Andy Biggin and two anonymous reviewers for their constructive
comments that improved the manuscript.
at University of Minnesota on April 17, 2015http://gji.oxfordjournals.org/Downloaded from

African Permian palaeogeography 13
Figure 9. Stereographic projections (projected on to vertical plane) of the virtual geomagnetic pole (VGP) of all the 13 site mean directions are given. The
VGPs of the two sites from coarse Enticho sandstones at Negash locality having different directions are shown in grey colours. The VGP for the rest 11 sites is
shown in black full circles for which overall mean palaeomagnetic pole positions were calculated (red star symbol); (a) In situ coordinate and (b) tilt-corrected
coordinate.
Tab l e 2 . Site mean VGPs for the 11 averages together with the overall mean pole position for the Palaeozoic rocks of Ethiopia are
given. In addition, obtained pole was transferred to the West Africa coordinates and values are given as ROTPOLE.
Sample name Strike/dip NφgλgφsλsK(dp)α95 (dm)
NBLS1 150/20 11 247.9 5.2 217.6 21.7 9.3 10.7
NBLS2 145/27 15 275.2 12.9 250.9 31.6 6.2 8.7
NBLS3 125/25 10 293.1 -11.9 278.8 2.1 3.1 4.8
NPSST1 130/27 23 290.6 -3.3 275.2 12.4 3.0 4.5
EPSST1 0/0 25 241.9 46.9 241.9 46.9 3.9 6.2
EPSST2 0/0 18 245.4 38.9 245.4 38.9 8.3 12.4
EDA1 0/0 13 243.8 27.3 243.8 27.3 4.2 5.5
EDA2 0/0 12 233.5 45.8 233.5 45.8 4.0 6.2
EDA3 0/0 11 247.4 31.1 247.4 31.1 3.7 5.2
HMK1 0/0 13 260.1 5.1 260.1 5.1 11.0 13.9
HMK2 0/0 13 239.8 12.0 239.8 12.0 6.0 6.8
Overall mean 11 257.5 20.2 249.5 26.0 13.1 13.1
Notes: Sample name; strike/dip; N, number of samples; φgand λg, pole longitude and latitude in geographic coordinates; φsand λspole
longitude and latitude in stratigraphic coordinates; K(dp), Fisher precision parameter; α95, 95 per cent confidence interval. The overall
mean VGPs directions are calculated for sites (N=11).
at University of Minnesota on April 17, 2015http://gji.oxfordjournals.org/Downloaded from

14 T. Kidane et al.
Figure 10. (a) Spherical projections with the major plates in their present-
day configurations and the APWP curve of Africa in West African coor-
dinates (Besse & Courtillot 1991, 2002, 2003; McElhinny et al. 2003) is
given. Our obtained pole position is given in red coloured diamond and
star symbols in north–east African and west African coordinate about Euler
pole position in black coloured cross symbol. The open diamond and star
symbols represent the data in situ and after tectonic correction, respectively.
The rotated pole is consistent with ages of 270–310 Ma. (b) Spherical pro-
jections with the major plates in their present-day configurations and the
APWP curve of Gondwana in South African coordinates (Torsvik et al.
2012) are given. Our obtained pole position for the sample site in East
Africa (black star) is shown in red diamond symbol and the red star symbol
represents the geomagnetic pole after transferred in to the south African
coordinate about Euler pole position shown in four-leg star symbol. The 95
per cent confidence circle of both poles slightly cut APWP curve at ages of
250–300 Ma.
REFERENCES
Bachtadse, V., Z¨
anglein, R., Tait, J. & Soffel, H.C., 2002. Paleomag-
netism of the permo/carbonifereous (280 Ma) Jebel Nehoud ring complex,
Kordofan, Central Sudan, J. Afr. Earth Sci., 35, 89–97.
Besse, J. & Courtillot, V., 1991. Revised and synthetic apparent polar wander
paths of the African, Eurasian, North American and Indian plates, and
true polar wander since 200Ma, J. geophys. Res., 96, 4029–4050.
Besse, J. & Courtillot, V., 2002. Apparent and true polar wander and the
geometry of the geomagnetic field over the last 200 Myr, J. geophys. Res.,
107(B11), 2300, doi:10.1029/2000JB000050.
Besse, J. & Courtillot, V., 2003. Correction to “Apparent and true polar
wander and the geometry of the geomagnetic field over the last 200 Myr,”
J. geophys. Res., 108(B10), 2469, doi:10.1029/2003JB002684.
Beuf, S., Biju-Duval, B., de Charpel, O., Rognon, P., Gariel, O. & Bennacef,
A., 1971. Les Gr`
es du Pal´
eozoique Inf´
erieur au Sahara: Sedimentation et
Discontinuities ´
Evolution Structurale d’un Craton, Technip.
Beyth, M., 1972a. To the geology of Central-Western Tigre, Dissertation,
Rheinische Friedrichs-Wilhelms-Universitat Bonn, Bonn, p. 155.
Beyth, M., 1972b. Paleozoic–Mesozoic sedimentary basin of Mekele Out-
lier, Northern Ethiopia, Bull. Am. Assoc. Petrol. Geol., 56, 2426–
2439.
Beyth, M., 1973. Correlation of Palaeozoic–Mesozoic sediments in North-
ern Yemen and Tigre, Northern Ethiopia, Bull. Am. Assoc. Petrol. Geol.,
57, 2440–2443.
Blanford, W.T., 1869. On the geology of a portion of Abyssinia, J. geol. Soc.
Lond., 25, 401–406.
Blanford, W.T., 1870. Observations on the Geology and Zoology of
Abyssinia, Mcmillan & Co., p. 487.
Bussert, R. & Schrank, E., 2007. Palynological evidence for a latest
Carboniferous-Early Permian glaciation in Northern Ethiopia, J. Af r.
Earth Sci., 49, 201–210.
Clark-Lowes, D.D., 2005. Arabian glacial deposits: recognition of palaeo-
valleys within the Upper Ordovician Sarah Formation, Al Qasim district,
Saudi Arabia, Proc. Geol. Assoc., 116, 331–347.
Cogn´
e, J.P., 2003. PaleoMac: a MacintoshTM application for treating pa-
leomagnetic data and making plate reconstruction, Geochem. Geophys.
Geosyst., 4(1), 1007.
Deynoux, M., Miller, J.M.G. & Domack, E.W., 1994. Earth’s Glacial
Records, Cambridge University Press, 266 pp.
Domeier, M., Van der Voo, R. & Torsvik, T.H., 2012. Paleomagnetism
and Pangea: the road to reconciliation, Tectonophysics, 514–517, 14–
43.
Dow, D.B., Beyth, M. & Hailu, T., 1971. Palaeozoic glacial rocks recently
discovered in northern Ethiopia, Geol. Mag., 108, 53–60.
Du Toit, A.L., 1953. The Geology of South Africa,Hafner.Pub.Co.
Fischer, R.A., 1953. Dispersion on a sphere, Proc. R. Soc. Lond., Ser. A.,
217, 295–305.
Furon, R., 1963. Geology of Africa,Hafner.Pub.Co.
Halls, H.C., 1976. A least-squares method to find a remanence direction
from converging remagnetization circles, Geophys. J. R. astr. Soc., 45,
297–304.
Halls, H.C., 1978. The use of converging remagnetization circles in paleo-
magnetism, Phys. Earth planet. Inter., 16, 1–11.
Hambrey, M.J. & Harland, W.B., 1981. Earth’s pre–Pleistocene Glacial
Record, International Geological Programme, Project 38: Pre–
Pleistocene Tillite, P76.
Hallam, A., 1983. Supposed Permo-Triassic megashear between Laurasia
and Gondwana, Nature, 301(5900), 499–502.
Irving, E., 1977. Drift of the major continental blocks since the Devonian,
Nature, 270, 304–309.
Kanasewich, E.R., Havskov, J. & Evans, M.E., 1978. Plate tectonics in the
Phanerozoic. Can. J. Earth Sci., 15(6), 919–955.
Kirschvink, J.L., 1980. The least squares line and plane and the analysis of
paleomagnetic data, Geophys. J. R. astr. Soc., 62, 699–718.
Klitgord, K.D. & Schouten, H., 1986. Plate kinematics of the Central At-
lantic, in The Geology of North America: The Western North Atlantic
Region, eds Vogt, P.R. & Tucholke, B.E., Geol. Soc. Am.
at University of Minnesota on April 17, 2015http://gji.oxfordjournals.org/Downloaded from

African Permian palaeogeography 15
Krasa, D., Petersen, K. & Petersen, N., 2007. The variable field translation
balance, in Encyclopedia of Geomagnetism and paleomagnetism: Dor-
drecht, pp. 977–979, eds Gubbins, D. & Herrero-Bervera, E., Springer.
Lottes, A.L. & Rowley, D.B., 1990. Early and Late Permian reconstructions
of Pangaea, in Paleozoic Paleogeography & Biogeography, Vol. 12, pp.
383–395, eds McKerrow, W.A. & Scotese, C.R., Geol. Soc. Mem.
McElhinny, M.W. & McFadden, P.L., 2000. Paleomagnetism Continents and
Oceans, Academic Press, 386 pp.
McElhinny, M.W., Briden, J.C., Jones, D.L., Brock, A., 1968. Geological
and geophysical implications of paleomagnetic results from Africa, Rev.
Geoph., 6(2), 201–238.
McElhinny, M.W., Powell, McA, C. & Pisarevsky, S.A., 2003. Paleozoic
terranes of eastern Australia and the draft history of Gondwana, Tectono-
physics, 362, 41–65.
McFadden, P.L., 1990. A new fold test for paleomagnetic studies, Geophys.
J. In t ., 103, 163–169.
McFadden, P.L. & McElhinny, M.W., 1988. The combined analysis of re-
magnetization circles and direct observations in paleomagnetism, Earth
planet. Sci. Lett., 87, 152–160.
Merla, G. & Minucci, E., 1938. Mission Geologica Nel Tigrai, Roma Real
Accadema d’Italia.
Mohr, P.A., 1965. The Geology of Ethiopia, Poligrafico.
Moreau, J., Ghienne, J.-F., Le Heron, D.P., Rubino, J.-L. & Deynoux, M.,
2005. 440 Ma ice stream in North Africa, Geology, 33, 753–756.
Morel, P. & Irving, E., 1981. Paleomagnetism and the evolution of Pangea,
J. geophys. Res., 86(B3), 1858–1872.
Muttoni, G., Kent, D.V., Garzanti, E., Brack, P., Abrahamsen, N. & Gaetani,
M., 2003. Early Permian Pangea “B” to Late Permian Pangea “A”, Earth.
planet. Sci. Lett., 215(3–4), 379–394.
Muttoni, G. et al., 2009. Opening of the Neo-Tethys Ocean and the Pangea
B to Pangea A transformation during the Permian, GeoArabia, 14(4),
17–48.
Opdyke, N.D., Mushayandebvu, M. & de Wit, M.J., 2001. A new paleomag-
netic pole for the Dwyka System and correlative sediments in sub-Saharan
Africa, J. Afr. Earth Sci., 33, 143–153.
Rochette, P. & Vandamme, D., 2001. Pangea B; an artifact of
incorrect paleomagnetic assumptions?, Ann. Geofisica., 44, 649–
658.
Rubidge, B.S., 2005. Re–uniting lost continents – fossil reptiles from the
ancient Karoo and their wanderlust, South Afr. J. Geol., 108(1), 135–
172.
Sacchi, R., Alene, M., Barbieri, M. & Conti, A., 2007. On the Paleozoic
Tillite of the Adigrat Group (Tigrai, Ethiopia), Per. Min e ra l , 76(2–3),
241–251.
Saxena, G.N. & Assefa, G., 1983. New evidence on the age of the glacial
rocks of northern Ethiopia, Geol. Mag., 120, 549–554.
Selden, P. & Nudds, J., 2011. Karoo. Evolution of Fossil Ecosystems, 2nd
edn, Manson Publishing, pp. 104–122.
Semtner, A.K. & Klitzsch, E., 1994. Early Paleozoic paleogeography of
the northern Gondwana margin: new evidence for Ordovician-Silurian
glaciation, Geol. Rundschau, 83, 743–751.
Shackleton, R.M. & Lomax, K., 1974. Paleomagnetic evidence for the Per-
mian age of tillites in Ethiopia, Research Institute of African Geology,
University of Leeds, Annual Report,No.18,p.13.
Smith, A.G. & Livermore, R.A., 1991. Pangea in Permian to Jurassic time,
Tectonophysics, 187(1–3), 135–179.
Torsvik, T.H. & Cocks, L.R.M., 2004. Earth geography from 400 to 250 Ma:
a paleomagnetic, faunal and facies review, J. geol. Soc., 161, 555–
572.
Tor sv ik , T.H ., M ¨
uller, R.D., Van der Voo, R., Steinberger, B. & Gaina, C.,
2008. Global Plate motion frames: toward a unified model, Rev. Geophys.,
46, RG3004, doi:10.1029/2007RG000227.
Tor s v ik, T. H . et al.,2012.Phanerozoicpolarwander,paleogeographyand
dynamics, Earth Sci. Rev., 114, 325–368.
Van der Voo, R., 1993. Paleomagnetism of the Atlantic Tethys, and Iapetus
Oceans, Cambridge University Press, 411 pp.
Vaslet, D., 1990. Upper Ordovician glacial deposits in Saudi Arabia,
Episodes, 13, 147–161.
Wegener, A., 1915. Die Entstehung der Kontinente und Ozeane, Sammlung
Vieweg , 94 pp.
at University of Minnesota on April 17, 2015http://gji.oxfordjournals.org/Downloaded from