Fossil trees from the basal Triassic Lebung Group at the
Makgaba site, west of Mokubilo, Botswana
M. De Wit1, M. Bamford2& C. Van Waarden3
1Tsodilo Resources Ltd and University of Pretoria
2Evolutionary Studies Institute and School of Geosciences, University of the Witwatersrand, Johannesburg
3Marope Research, Francistown, Botswana
Received 18 July 2017. Accepted 19 February 2018
In 2016, the declared palaeontological heritage site on
the southern edge of Sua Pan and below the Mosu Escarp-
ment was visited to collect fossil wood samples. This
Makgaba (Simanentsa) fossil tree site 16-A4-3 (Table 1) is
just north of the Moriti wa Selemo bush camp (Fig. 1). This
camp is 7 km east of the Tlalamabele veterinary gate on
the A30 and some 95 km east of Orapa (Fig. 2). Because the
site is a protected monument, permission was obtained
from the Botswana National Museum in August 2016
(Ref: NM 6/1/1 II (100)) to remove samples for identifica-
tion and age determination.
The objective was to place these Makgaba fossils in their
original stratigraphic context by field investigation of the
local geology and species identification of wood samples.
Presently, the National Monument sign at the Makgaba
site refers to it as ‘these rare fossils are remnants of ancient
trees that turned into solid rock some 50 million years
ago.’ Such mis-information is misleading and in an
attempt to improve local tourism the signage will be
updated and information supplied to the local tourism
MATERIALS AND METHODS
Description of Makgaba fossil tree site
The Makgaba site is roughly 3.5 km north of the Moriti
wa Selemo camp (Fig. 2). At the site there are four places
where the fossil wood occurs (Table 1). However, the
manager of the Moriti wa Selemo Camp has located more
fragments of fossil trees on the Mosu scarp to the south
and west of the Makgaba site. Coordinates of each wood
occurrence were taken with a Garmin 62 handheld GPS
using the UTM WGS84 (Zone 35K) datum (Table 1).
The fossils at the Makgaba site listed in Table 1 are
located on the slopes of a small hill within an embayment
of the pan and is surrounded by higher ground to the
west, south and east. A scarp, which separates the higher
ground from this embayment is some 50 m in height, and
is the easterly extension of the Mosu Escarpment which
reaches its maximum elevation of some 91 m east of Mosu.
The escarpment and the hill are composed of Karoo
194 ISSN 2410-4418 Palaeont. afr. (2018) 52: 194–200
Fossil wood samples were collected from an area underlain by Karoo Supergroup rocks along the southern edge of Sua Pan in east cen-
tral Botswana. From the local stratigraphy it suggests that these fossils have been derived from the Mosu sandstones that occurs at the
base of the Mosolotsane Formation and which is time-equivalent to the Molteno Formation in South Africa that is of Triassic age. Based
on the arrangement of tracheid pits the fossil wood has been identified as Agathoxylon, and most likely Agathoxylon africanum. This
species has a Permian to Triassic time range in southern Africa and probably is the first published record of Agathoxylon africanum in
Keywords:Agathoxylon, Molteno flora.
*Author for correspondence. E-mail: firstname.lastname@example.org
Palaeontologia africana 52: 194–200 — ISSN 2410-4418 [Palaeontol. afr.] Online only
Permanently archived on the 22nd of March 2018 at the University of the Witwatersrand, Johannesburg, South Africa.
The article is permanently archived at: https://hdl.handle.net/10539/24149
Palaeontologia africana 2018. ©2018 M. De Wit, M. Bamford & C. Van Waarden. This is an open-access article published under the Creative Commons Attribution 4.0
Unported License (CC BY4.0). To view a copy of the license, please visit http://creativecommons.org/licenses/by/4.0/. This license permits unrestricted use, distribution,
and reproduction in any medium, provided the original author and source are credited.
The article is permanently archived at: https://hdl.handle.net/10539/24149
Table 1. Position of fossil tree samples at the Makgaba site (using the UTM WGS84 (Zone 35K) datum).
Location Eastern Northern Dimensions Comment
(L, W, H) in cm
Tree 1 – T1 0426747 7647494 55 × 50 × 40 Loose boulder
Tree 2 – T2 0426785 7647494 30 × 45 × 25 Loose boulder
Tree 3 – T3 0426784 7647629 200 × 50 × 30 ‘Near’ In situ
Tree 4 – T18 0426788 7647691 Many fragments Loose pebbles
Supergrouprocks (Stansfield1973) thatare partlycovered
by scree material.
None of the fossil tree specimens appear to be embed-
ded in the Karoo Supergroup strata and are not found in
their living positions. The largest however, the T3 fossil,
occurs as a horizontally lying trunk broken into several
boulder size fragments. It is almost 2 m in length (Fig. 1)
and totally unabraded. This suggests that this specimen,
after been exhumed from Karoo sediments at the site, has
experienced little or no post-exhumation movement. We
are therefore confident that these can be referred to as
‘nearly’ in situ (Fig. 1). The T1 and T2 fossils are abraded
ISSN 2410-4418 Palaeont. afr. (2018) 52: 194–200 195
Figure 1. Site ‘Fossil Tree 3’ of broken up fossil tree trunk. Hammer 30 cm in length for scale.
Figure2. Localgeology (From:Quarter DegreeSheet 2126A,Tlalamabele, GeologicalSurvey of Botswana,scale 1:125,000,withbrief descriptionof the
geology by G. Stansfield, 1973). Pale yellow – Ntane Sandstone Formation; pale blue – Tlhabala Formation and Tlapana Formation, stippled where
carbonaceous ; green – dolerite sills; thick black lines – faults; red lines with dots – probable dyke; purple lines – linear photo feature; red star – small
hill with fossil tree occurrences. Thick red line is approximate position of section in Fig. 5.
andfound atlower levelswhile theT18 siteis composedof
many small fragments that have been collected and
dumped at this site.
According to the geological map of the quarter degree
sheet Tlalamabele 2126A (Fig. 2) the fossil trees occur close
to the contact of Upper Ecca Group Tlapana mudstones,
andthe overlyingsandstones ofthe LebungGroup (Fig.2)
Stansfield(1973) describesthe mudstonesof theTlapana
Formation (Fig. 2), the uppermost formation of the Ecca
Group in the Tlalamabele area, as generally being
non-carbonaceous.These arelight-grey tobluish incolour
and weather to light-yellow on surface. According to
Smith (1984) the lower part of the Tlapana Formation
is carbonaceous and contains coal seams. The non-
carbonaceous upper part of the Tlapana mudstones has
been referred to by Smith (1984) as the Tlhabala Forma-
tion (Table 2). The non-carbonaceous grey to brown
mudstones of the uppermost part of this formation con-
tainlimestone bandsand septariannodules, andis known
as the Kautse Member (Smith 1984). These have been
logged in borehole N4/1 approximately 8 km southeast of
Mosu and some 23 km west of the fossil tree site
(Stansfield 1973). Smith (1984) suggests that the Kautse
Member is limited in extent and displays some rapid
lateral facies changes. In another borehole N1/3 (Smith
1984), estimated to be some 10 km to the southeast of the
fossil site, the lowest part of the Tlhabala Formation is
overlain directly by the Ntane Sandstone Formation and
there are no Kautse Member rocks present. According to
Stansfield (1973) the upper Tlhabala Formation, in places
lies directly and conformably on the ‘non-carbonaceous’
mudstones of the upper Tlapana Formation. The overly-
ing Lebung Group is subdivided into the Mosolotsane
and Ntane Formations (Table 2). On the scarp near Mosu
the proximal sandy facies of the Mosolotsane Formation
comprising immature, coarse-grained, cross-bedded
sandstones with isolated pebbles that is weakly cemented
by ‘iron ores’ (Stansfield 1973). Smith (1984) refers to these
facies as the Mosu Member of the Mosolotsane Formation
that occurs between the Tlhabala and Ntane Sandstone
Formations. Both the Kautse and Mosu Members do not
continue eastwards in the Makgaba area according to
Stansfield (1973). Stansfield (1973) suggested that this was
due to a period of gentle tilting and flexing of the pre-
Lebung Group sediments followed by a period of erosion
prior to the deposition of the Ntane Sandstone Formation
and that this was most pronounced in the east of the area.
Hence, the Mosu Sandstone, the Kautse beds and the
upper part of the Tlapana Mudstone in the eastern part of
the Tlalamabele area were removed by erosion prior the
deposition of the Ntane sandstone. The Mosolotsane
Formation appears to be absent to the north where the
Ntane Sandstone Formation overlies the older Karoo
formations unconformable (Smith 1984).
The base of the Lebung Group is marked by an uncon-
formity which, just east of Mosu, comprises a pebble
conglomerate up to 25 cm thick with well-rounded clasts
up to cobble size composed of sandstone (Stansfield 1973).
The base of the Ntane Sandstone has been intruded by
Green (1965) regarded the Kautse beds as most probably
‘Beaufort Series’. Smith (1984) allocated the whole
Tlhabala Formation as the Beaufort Group equivalent.
The contact of the Tlhabala Formation, and the overly-
ing Lebung Group, either by the Mosolotsane Formation
or the younger Ntane Sandstone Formation, represents a
major unconformity separating the ‘upper’ Karoo from
the ‘lower’ Karoo Supergroup (Smith 1984, Bordy et al.
2010). It should also be emphasized that the base-Molteno
angular unconformity is well-developed in many of the
Karoo basins in south-central Africa (Catuneanu et al.
The small hill on which the uppermost occurring tree
fossils (T3) are found has grey mudstones at the base and
is overlain by medium grained cross-bedded sandstones
that are pebbly in places. The fossil site occurs some 3 m
above this contact.
The grey mudstones are non-carbonaceous and have
some interbedded limestone lenses. The mudstones are
196 ISSN 2410-4418 Palaeont. afr. (2018) 52: 194–200
Table 2. Correlation of the Karoo lithostratigraphic units of the main Karoo Basin and the Kalahari Basin in northeast Botswana.
highly weathered and partially covered by scree material.
The carbonates are between 0.2 and 0.4 m thick, and are
restricted in lateral extent but can be followed for up to 20 m.
They contain septarian nodules that can be up to 15 cm in
diameter. These are almost round, light grey in colour and
are cut by carbonate veins. All around the hill and below
the mudstone/sandstone contact septarian nodules are
found that have weathered out of the limestones.
The sandstones directly above the mudstones are
medium-grained with low angle planar cross-beds. The
grains are sub-angular and composed of quartz, feldspar
and some micas. Well-rounded small pebbles of mainly
vein quartz are present in some of the beds. The sand-
stones are highly weathered. Palaeo-flow direction from
only four readings suggest a flow was to the southeast
(144° Mn). At the base of the sandstones is a conglomerate
that directly overlies the grey mudstones at sites T3, T11
and between T16 and T19 with an erosional contact
(Fig. 3). The conglomerate comprises a single bed of
well-rounded medium to large pebbles composed of
mainly of quartz-schist and quartzite that have a
ferruginized cement along the basal erosional contact
(Fig.3). Directlybelow themudstone/sandstone contactat
T3 and some 3 m below the fossil tree T3, similar pebbles
are found with clasts up to 12 cm in diameter. The clasts
are mainly composed of clean and grey quartzite, bedded
dark quartzite, quartz-schist, quartz porphery and vein
quartz. The clasts are all well-rounded and polished with
no obvious percussion marks. Some are facetted and a
large component has tabular shapes.
The interpretation is that the mudstones belong to the
Kautse Member of the Thlabala Formation based on the
lithological descriptions provided by Smith (1984) of the
Kautse Member near Mosu. The limestone lenses and the
presence of septarian nodules clearly equates these sedi-
mentswith theKautse Member.The overlyingsandstones
with its basal conglomerate is believed to be an easterly
extension of the Mosu Member of the Mosolotsane
Formation. Both members were not mapped as far east as
the Makgaba site but based on the lithological similarities
with the descriptions of Stansfield (1973) and Smith (1984)
it is suggested that the basal conglomerate, mapped in the
field just below the T3 fossil site, is the unconformity
between the Kautse Member of the Tlhabala Formation
(Beaufort Group) and the Mosu Member of the
Mosolotsane Formation (Lebung Group).
The Lebung Group above the basal conglomerate
comprises coarse sandstones, grits and conglomerate
lenses of up to 10 cm thick that are seen at T11 and T19
(Fig. 5). The small- to medium-sized pebbles are
sub-roundedto sub-angularin shapeand theclasts within
these conglomerates are composed mainly of vein quartz
and some chert are found higher up within the
Mosolotsane Formation (Fig. 4). The sandstones are
cross-bedded and contain vertical ‘burrow’ structures
(?Scolithos) on some bedding planes (Fig. 4).
Identical thin bands and pockets of rounded quartz
pebbles have been described in sandstones that form the
Mosu escarpment some 30 km to the west (Stansfield,
1973). These coarse clastics are therefore correlated with
the Mosu Member and it is estimated that this member is
about 20 m thick.
The Ntane Sandstone Formation (Fig. 2), based on
descriptions by Smith (1984), are more uniform fine-
grained and have been mapped from T8 and upwards
(Fig. 5). The Ntane sandstones have been intruded by
dolerite sills of the Stormberg Lava Group that are seen at
T6, T8 and T10. The contact between the sills and the
cross-bedded Ntane Sandstone was mapped at T8. The
contact between the Ntane Sandstone and Mosolotsane
Formations is somewhere between sites T10 and T11 but
was not pinpointed due to the extensive scree develop-
The fossil trees have therefore been placed at the base of
Mosu Sandstone of the Mosolotsane Formation which
forms part of the lower Lebung Group and just above the
Tlhabala/Mosolotsane unconformity. It can be correlated
with the Molteno Formation in South Africa litho-
stratigraphically (Table 2) (Bordy 2010).
Two samples, 3A (BP/16/1956) and 3B (BP/16/1957), were
ISSN 2410-4418 Palaeont. afr. (2018) 52: 194–200 197
Figure 3. Contact between the Tlhabala Formation and overlying sandstones of the Mosolotsane Formation (left) at T11. Note isolated pebbles occur
at this contact and the ferruginous nature of the contact. Well-rounded pebbles derived from the base of the Mosolotsane Formation (right).
both taken from fossil T3. The wood is completely
replaced by silica and three petrographic thin sections
were cut of each sample along the x,yand zaxes.
BP/16/1956 is highly compacted and the tracheids have
been compressed. The wood has a zig-zag appearance in
transverse section (Fig. 6) but non-compacted tracheids
would normally have a rounded to squarish outline and
be aligned in more or less straight rows. No growth rings
could be detected. In longitudinal section medullary rays
are uniseriate and ghosts of bordered pits on the tracheid
198 ISSN 2410-4418 Palaeont. afr. (2018) 52: 194–200
Figure 5. North–south section from the fossil tree site (T1, T3 and T18) to the main road Orapa-Francistown road (T4). The horizontal dashed line on
the section marks the unconformity between the Tlhabala Formation and the overlying Mosolotsane Formation. Contact between Mosolotsane and
Ntane Sst Formations was not seen. Horizontal purple zones are Karoo dolerite sills c. 182 Ma. Approximate positionsof fossil treesare shown inred
in the section.
Figure 4. Mosolotsane coarse conglomerates (left) with mainly quartz pebbles and coarse-grained cross-bedded sandstones with an upper
bioturbated (?) zone (right) at T19.
ISSN 2410-4418 Palaeont. afr. (2018) 52: 194–200 199
walls can be seen in a biseriate, compressed, alternate
arrangement. No cross-field pits are visible. It is possible
that this sample represents a twisted branch or root of the
Sample BP-16-1956 (Fig. 6) is highly compressed but is
an Agathoxylon sp. Identification to species level was not
Sample BP/16/1957 (Fig. 7) from the same tree is much
better preserved and not distorted. It has a clear trans-
verse section with regular tracheids, narrow latewood
and also some false growth rings. In longitudinal section
rays are uniseriate and the same ghosts of biseriate,
alternate tracheid pits can be seen but not clearly (Fig. 7).
Based on the arrangement of tracheid pits, i.e. araucarian,
it is clear that sample BP-16-1957 is Agathoxylon, and most
likely Agathoxylon africanum because the tracheid pits are
biseriate. This species has a Permian to Triassic time range
in southern Africa (Bamford 1999).
Other samples of Permian-Triassic woods have been
recorded from Botswana but their taxonomy needs to be
updated. From the Boteti River near Motopi Village, three
samples were described (Bamford, 1997) but they differ
from this sample. The wood from the southwestern edge
of the Sua pan, near Tshaitshe (Bamford 1997) was
described as Dadoxylon parenchymosum but this genus is
invalid (Philippe 1993; Bamford & Philippe 2000; Philippe
& Bamford 2008; Rossler et al. 2014) and because of its
mixed tracheid pitting should be called something else,
for example Metapodocarpoxylon or Brachyoxylon. There are
a number of options but without cross-field pits it is not
possible to select the genus.
As the first record of woods of Molteno age in Botswana,
the occurrence of Agathoxylon sp. indicates that further
research should be done to find other elements of this
flora, for example the iconic Dicroidium leaf flora of the
upper Karoo sequence. This would improve the record of
fossils in southern Africa but also has implications for
understanding the stratigraphy (potentially for economic
applications) and past climate, biodiversity and local
environmental settings. The Mosolotsane Formation is
the place to begin to look for more woods and other plants
The fossil trees, found on the southern edge of Sua Pan,
have been sourced from the base of the Mosu Member
sandstones which forms the lower Mosolotsane Forma-
tion. This puts it somewhere at the base of the Molteno
equivalent above the regional unconformity. The fossil
wood, which is the first published record of Agathoxylon
africanum in Botswana, from a biostratigraphic point of
view, has been placed in the Triassic. Although the age
rangeof thespecies islonger,lithostratigraphically theage
of the Makgaba fossils would be constrained to between
240 and 250 Ma.
Andries Kruger of the Moriti wa Selemo camp is thanked for his hospitality during
the visits to the area and Prosper Bande (Evolutionary Studies Institute, University
ofthe Witwatersrand) for makingthe petrographicslides of thesilicified wood.The
authors also wish to thank Drs Roger Smith and Bastien Linol for their input and
suggestions which has greatly improved this manuscript.
BORDY, E.M., SEGWABE, T. & MAKUKE, B. 2010. Sedimentology of the
Upper Triassic-Lower Jurassic (?) Mosolotsane Formation (Karoo
Supergroup), Kalahari Karoo Basin, Botswana. Journal of African Earth
Sciences 58, 127–140.
Figure 7. Photomicrographs of sample BP-16-1957. Top , a transverse sec-
tion (ts) showing a growth ring with narrow latewood (thick-walled
cells). Bottom, BP-16-1957 a radial longitudinal section (rls) with ghosts
of biseriate, alternate bordered pits.
Figure6. Photomicrographof atransverse sectionof sampleBP-16-1956.
BAMFORD, M.K. 1999. Permo-Triassic fossil woods from the South Afri-
can Karoo Basin. Palaeontologia africana 35, 25–40.
BAMFORD, M.K. 1997. Fossil wood from the Boteti River and Tshaitshe,
Botswana. Botswana Notes and Records 29, 1–8.
BAMFORD, M.K. 2016. Fossil wood and leaves from Mike de Wit August
2017. Note for the record, 1 p.
BAMFORD, M.K. & PHILIPPE, M. 2001. Jurassic – Early Cretaceous
Gondwanan homoxylous woods: a nomenclatural revision of the genera
with taxonomic notes. Review of Palaeobotany and Palynology 113,
CATUNEANU, O., WOPFNER, H., ERIKSSON, P.G., CAIRNCROSS, B.,
RUBIDGE, B.S., SMITH, R.M.H. & HANCOX, P.J. 2005. The Karoo b as-
ins of south-central Africa. Journal of African Earth Sciences 43, 211–253.
GREEN, D. 1965. The Kautse Beds and the sandstones of the Kedia area.
Geological Survey of Botswana, unpublished report.
JOHNSON, M.R., van VUUREN, C.J., HEGENBERGER, W.F., KEY, R. &
SHOKO, U. 1996. Stratigraphy of the Karoo Supergroup in southern
Africa: an overview. Journal of African Earth Sciences 23, 3–15.
PHILIPPE, M. & BAMFORD, M.K. 2008. A key to morphogenera used for
Mesozoic conifer-like woods. Review of Palaeobotany and Palynology 148,
RÖßLER, R.,PHILIPPE, M.,van KONIJNENBURG-van CITTERT, J.H.A.,
McLOUGHLIN, S., SAKALA, J., ZIJLSTRA, G. et al. 2014. Which
name(s) should be used for Araucaria-like fossil wood? – Results of a
poll. Taxon 63(1), 177–184.
SMITH, R.A. 1984. The Lithostratigraphy of the Karoo Supergroup in
Botswana. Geological Survey of Botswana, Bulletin 26, 239 pp.
STANSFIELD, G. 1973. The geology of the area around Dukwe and
Tlalamabele, Central Districts, Botswana. Geological Survey of Botswana,
District Memoir 1, 74 pp.
200 ISSN 2410-4418 Palaeont. afr. (2018) 52: 194–200