Sedimentary Structures Generated by Hippopotamus amphibius in a Lake-margin Wetland, Ngorongoro Crater, Tanzania

Abstract

Large mammals, especially Hippopotamus amphibius, have created a distinctive set of traces in a spring-fed freshwater wetland on the margin of the saline-alkaline crater lake in the Ngorongoro Crater, Tanzania. It is comprised of a ∼30 m diameter zone of deep (<2m) bioturbation due to hippo wallowing, surrounded by dendritic to radial hippo trails 1–5 m wide and <0.5 m deep, infilled with organic-rich mud. These trails narrow and thin as they grade into trackways on the lake flat. Tracks and trackways of more terrestrial mammals, such as bovids and equids, are found in the lake flat muds surrounding the hippo-dominated area. Associations of sedimentary structures such as these are important indicators of paleoenvironmental conditions where they are preserved in the sedimentary record, due to the strong affinity of Hippopotamidae for freshwater environments.

Full text

212 RESEARCH LETTERS
Copyright Q2002, SEPM (Society for Sedimentary Geology) 0883-1351/01/0017-0212/$3.00
Sedimentary Structures Generated
by Hippopotamus amphibius in a
Lake-margin Wetland, Ngorongoro
Crater, Tanzania
DANIEL M. DEOCAMPO*
Department of Geological Sciences, Rutgers University, Piscataway,
NJ 08854-8066
PALAIOS, 2002, V. 17, p. 212–217
Large mammals, especially Hippopotamus amphibius,
have created a distinctive set of traces in a spring-fed fresh-
water wetland on the margin of the saline-alkaline crater
lake in the Ngorongoro Crater, Tanzania. It is comprised of
a
;
30 m diameter zone of deep (
,
2m) bioturbation due to
hippo wallowing, surrounded by dendritic to radial hippo
trails 1–5 m wide and
,
0.5 m deep, infilled with organic-
rich mud. These trails narrow and thin as they grade into
trackways on the lake flat. Tracks and trackways of more
terrestrial mammals, such as bovids and equids, are found
in the lake flat muds surrounding the hippo-dominated
area. Associations of sedimentary structures such as these
are important indicators of paleoenvironmentalconditions
where they are preserved in the sedimentary record, due to
the strong affinity of Hippopotamidae for freshwater envi-
ronments.
INTRODUCTION
Background
Large vertebrates long have been recognized as impor-
tant agents in shaping the geomorphology of their terres-
trial environments and the character of associated sedi-
ments (Laporte and Behrensmeyer, 1980). One of the dif-
ficulties in understanding the nature of large mammal-
substrate interactions in the geologic past is the general
lack of extant species similar in size to the megafauna of
the ancient past. Therefore, inferences based on modern-
process observations have been limited to those faunas,
typically in Africa, that can be observed today in natural
settings (Cohen et al., 1993).
Of the extant large mammals, perhaps the most nearly
infaunal is Hippopotamus amphibius, which is known for
conducting wholesale excavation of sediment in a variety
of environmental settings (McCarthy et al., 1998). These
organismsare particularly important inclosed-basinlake-
margin settings, where low topographic gradients and
strong geochemical gradients allow mammal activities to
affect the local hydrology, geochemistry, and vegetation
(Deocampo and Ashley, 1997; Deocampo, 2001). In thedry
lands of East Africa, Hippopotamusamphibius tends to fo-
cus activity in fresh water bodies and wetlands (Kingdon,
1979). Understanding the nature of the interactions be-
* Present address: Department of Mineral Sciences, National Museum
of Natural History, Smithsonian Institution, MRC 119, Washington,
DC 20560-0119
tween hippos and their substrates, and the resultingsedi-
mentary records, can provide important information on
the paleoenvironments and paleoecology of analogous an-
cient settings.
The purpose of this paper is to describe the set of sedi-
mentary traces associated with the activities of modern
vertebrates, especially Hippopotamus amphibius,onthe
margin of Lake Makat, in the Ngorongoro Crater, Tanza-
nia. The low-gradient lake margin is in a protected natu-
ral setting with a population of hippopotamus; hence, this
providesa basis for understanding what waslikelyamuch
more widespread phenomenon in the Cenozoic.
Ecological Setting
The Ngorongoro Crater is the ;19 km diameter caldera
ofa large volcano that collapsedin the Crater Highlands of
northern Tanzania about 2 million years ago (Hay, 1976).
The internally drained modern caldera surface (Fig. 1) is
characterized by a wide range of environmental settings
anda diverse floraland faunalassemblage (Herlocker and
Dirschl, 1972; Anderson and Herlocker, 1973; Estes and
Small, 1981; Kabigumila, 1993). Ngorongoro is well
known for its large populations of non-migratory large
mammals,including plains animals such as zebraand wil-
debeest, forest dwellers such as waterbuck, and notable
carnivores such as lion and leopard. Vertebrate activity
tendsto follow seasonal patterns,with largemammalpop-
ulations dispersed during wet times, and concentrated
aroundpastures and water sourcesduring the dryseasons
(Estes and Small, 1981; Deocampo et al., 1998).
Although it is relatively small, the Mti Moja spring, lo-
cated on the northeast shore of the saline-alkaline Lake
Makat (Fig. 1; UTM coordinates 0783064E, 9648057N), is
an important focus of large-mammal activity in the area
(Table 1). The spring is a major source of potable water in
thecenter of the Crater,farfromthe springs, streams,and
wetlands near the Crater wall (Deocampo and Ashley,
1999). Groundwater erupts at the base of an eroding cliff
and flows for about 100 m toward the lake in a restricted 2
m wide, ,50 cm deep channel (Fig. 2). Once reaching the
lower-gradient lake flat, water spreads out, feeding a com-
plex of lightly vegetated marsh-and-hippopotamus habi-
tat. The most distal spring waters, about 600–800 m away
from the source, are highly saline brines that do not sup-
port any rooted vegetation (Deocampo, 1997; Deocampo
and Ashley, 1999).
Methods
Mti Moja was visited repeatedly during July and Au-
gust, 1995–2000. When lake waters were low (1995–1997,
2000), maps of spring-water distribution and geomorphic
features were created by a combination of electronic total
station, differentially-corrected global positioning system,
and tape-and-compass. Daytime hippo activity was ob-
served on foot from distances ranging from 20–1000 m, for
timespans up to several hours; nocturnal observations
were logistically not possible. Because hippo activity out-
side pools generally is limited to the night, the actual cre-
ation of structures was not observed directly. However,
hippo structures clearly were associated with occupied
hippo pools and identifiable hippo tracks. Conductivity

NGORONGORO HIPPO TRAILS 213
FIGURE 1—Location of the study area.
TABLE 1—Large vertebrate animals observed within 100 m of Mti Moja spring, northeast shore of Lake Makat, NgorongoroCrater, Tanzania,
1996–1997.
Common name Species name
Herbivores Buffalo (Cape)
Gazelle (Grant’s)
Gazelle (Thompson’s)
Hartebeest
Hippopotamus
Syncerus caffer
Gazella granti
Gazella thomsoni
Alcelaphus buselaphus cokii
Hippopotamus amphibius
Ostrich
Rhinoceros (Black)
Warthod
Wildebeest
Zebra (Common)
African Elephant (faeces)
Struthio camelus
Diceros bicornis
Phacochoerus aethiopicus
Connochaetes taurinus
Equus burchelli boehmii
Loxodonta africana
Carnivores Cheetah
Fox (Bat-Eared)
Hyena (Spotted)
Jackal
Lion
Acinonyx jubatus
Otocyon megalotis
Crocuta crocuta
Canis mesomelas
Panthera leo
and pH of surface waters were measured in situ by elec-
tronic meter. Organic carbon content was estimated by
loss-on-ignition for one hour at 550 C (Lewis and Mc-
Conchie,1994), well below the experimentally-determined
threshold for significant structural water loss for Ngoron-
goro clays (Deocampo, 1997). Associated geomorphology,
sedimentology, and aqueous geochemistry of the wetland
will be reported elsewhere (Deocampo, in press).
RESULTS
Hippopotamus traces
Hippo activity is centered within pools for most of the
daytime hours (Fig. 3A). Hippos lie partially submerged
within these shallow pools (5–15 m across, ,2 m deep),
wallowing and interacting with others in their group
(Kingdon,1979). Although these activities do involve some
movement and churning of the substrate (Wolanski and
Gereta, 1999), more significant substrate reworking oc-
curs as hippos move in and out of the pools at night. Dur-
ing these movements, the hippo bodies easily plow
through saturated muddy sediment. Repeated nocturnal
forays by hippos from their daytime wallows to evening
grazing pastures have created a ;100-m-long dendritic
network of trails originating in the wallowing pools (Fig.
3B-D). These trails, created as the hippos forcibly move
through saturated mud, measure 1–5 m across and ;0.5
m deep, and are infilled with organic-rich mud. The trails
commonly have elevated mud levees on their margins,
which were squeezed above the lakeflat surface by the
passage of the hippos, and then dried in place (Fig. 3E).
The largest trails are those closest to the pools, reflecting
the repeated usage of these over time; the more distal
trails may be used less frequently, depending on the des-
tination of the hippos, which may change with time. The
trails become exposed and mudcracked as the surface wa-
ter coverage shrinks during the dry season, and the abun-
dance of organic carbon content is reduced, although it is
still much higher than that of surrounding mudflats (Ta-
ble 2).
As the trails emerge onto drier land, they become shal-
lower and grade into recognizable tracks and trackways
(Fig. 3D). Whereas some of the trails adjacent to the pools
are a few meters across, the distal ends of the trails are
commonly the width of a single hippo as they emerge onto
dry land. Spring water at Mti Moja flows into the hippo
pools and then continues onto the mudflats, following the
hippo trails. The distal ends of the Mti Moja trails com-
monly contain highly saline and alkaline waters (Fig. 2),
and efflorescent crusts of evaporite minerals may form in
thesesettings (Deocampo and Ashley,1999). A mapof sur-
face water coverage during the dry seasons of 1995–2000
showsthat this distribution maychangeovertime (Fig.2).
Where these trails occur in a more stable setting, such as
adjacent to a large pool, long-term repeated usage contin-
ues to enlarge the trails, and they can be visible from long
distances (Fig. 3F).

214 DEOCAMPO
FIGURE 2—Mti Moja spring surface water distribution and geochemistry. (A) Local map showing the distribution of surface water and major
vertebrate traces at the lake-margin Mti Moja and spring-fed wetland. The central area of hippo pools grades into surrounding areasof hippo
trails, which then grade into the trackfields created by a diversity of vertebrates on the mudflats. The area was flooded by lake water during
1998–1999 and was accessible from the east again in 2000. (B) Graph showing steep geochemical gradient produced by downstreamevap-
oration of springwaters during 1995. The freshest waters are adjacent to the springhead and in the area of hippo pools, while themost saline
and alkaline waters are found in the distal ends of the hippo trails and in puddles on the mudflats (Deocampo and Ashley, 1999).
Other Vertebrate Traces
Other large mammals have a significant impact on the
Mti Moja system. Most notable aside from hippotamus is
the Cape buffalo (Syncerus caffer), which also createswal-
lows. These tend to be smaller than hippo wallows (2–3 m
wide, ,0.5 m deep), and they lack the kind of trail thatare
created by the hippopotamus. The buffalo also seem to
preferthe moist ground slightlyupslope fromthe fully sat-
urated lake-flat muds, although this may be due to hippo
defense of their wallowing pools.
The many bovids and equids that visit Mti Moja spring
also generate traces. These are generally trackfields; in-
dividual trackways are indistinguishable due to their
great number. These are best seen in the slightly saturat-
ed mudflats that the animals cross to access the freshwa-
ter-spring source. Although many of these animals roll on
their backs both for cleaning and dusting, such behavior is
generallyon dry mudflats;hence, theydo notproduce wal-
lowing structures. Where they traverse marshland, how-
ever, they may bioturbate sediment with their legs up to
;1 m deep, and they can produce a step up to 50 cm tall
marking the boundary between a dry, resistant ledge and
wet marshy sediment. Such activity is restrictedgenerally
to areas such as the first 100 m of the Mti Moja channel
(Fig. 2A), where the saturated area may be crossed easily.
DISCUSSION
Impacts of Hippopotamus amphibius
The flow of water from the Mti Moja spring onto the
Lake Makat mudflats is affected strongly by hippopota-
mus trails, in large part due to the low hydraulic conduc-
tivity of the clay-rich substrate. Changes over time in the
distribution of the trails may be related to changes in pre-
ferred upland grazing pastures. This flow of spring waters
through hippopotamus trails is similar to the effect ob-
served in the Okavango Delta, where channel location is
influenced strongly by hippos (McCarthy et al., 1992,
1998). Such a pattern is likely to be seen in any setting
with a substrate soft enough to allow deformation by hip-
popotamus torsos.
A zoning in traces is evident at Mti Moja, withhippopot-
amus dominating the environment. At the core of the sys-
tem is an area with hippo pools characterized by the deep-
est bioturbation profiles, freshest water, and organic-rich
sediment.This reflects the persistent concentration of hip-
po activity within the pools throughout the day, and per-
haps also for much of the night. Immediately adjacent to
the pools is the area of dense trails, some with dendritic
patterns, created by movements in and out of the pools.
These are widest adjacent to the pools, where the trails
converge, and they spread out in the direction of travel to
grazing pastures. At Mti Moja, these directions are gener-
ally shore-parallel, as may be expected because the saline-
alkaline lake is not preferred hippopotamus habitat. The
distal ends of these trails narrow, to perhaps one hippo-
potamus width, and they shallow, before emerging onto
dry land. Surrounding the area of hippo trails is an
ephemerally wet area characterized by buffalo wallows
and ungulate trackfields. The depth of bioturbation in
theseareas is less thantheothers, with a greater diversity
of evident traces.
PaleoenvironmentalImplications
Some general characteristics of the Mti Moja lake-mar-
gin spring wetland can be applied to paleoenvironmental

NGORONGORO HIPPO TRAILS 215
FIGURE 3—Examples of large sedimentary structures created by
Hippopotamus amphibius
. (A) Active hippo pool occupied by two individuals,
with a trail in the foreground and surrounding trackfields on the mudflats in the background. (B) A dendritic trail originating in thepool atupper
left, with branches diverging off the trunk in both directions. (C) Hippo’s view of a dried trail burrowed into the mudflat. Geological hammerfor
scale at right. (D) A hippo trail as it emerges onto what was drier land at the time of the trail’s creation. (E) Close-up of mud levee on the
margin of the hippo trail shown in C. Geological hammer for scale at bottom. (F) View from the Crater rim (;600 m above the floor) of a hippo
trail emerging from the Lerai Hippo Pool, on the south edge of Ngorongoro Crater. Black dots at the bottom of the picture are wildebeestand
zebra.
problems. The first one is the hippopotamus trail itself—
this structure has not been recognized widely, but it may
be a common feature of lake-margin deposits, such as in
thePlio-Pleistocene Olduvai basin (Deocampoand Ashley,
1997; Ashley et al., 2000). Seen in outcrop, certain U-
shapedstructures that are about1m deep and 1–5 mwide
are possibly hippopotamus trails. These may be infilled
with either organic-rich sediment, as at Mti Moja, or pos-
sibly by lacustrine muds if the wetland is flooded by lacus-
trinetransgression. A biological origin forthese structures
is likely when evidence such as massive infill texture, fine
grainsize, a lack ofscourortool marks, and perhaps agra-

216 DEOCAMPO
TABLE 2—Estimates of organic carbon content in Mti Moja lake-flat
wetlands based on loss-on-ignition analysis.
Sample site Sample weight
(g) Organic carbon
(wt %)
Hippo pool margin
Hippo pool margin
Hippo trail
Hippo trail
Mudflat
Mudflat
Mudflat
14.24
13.20
11.51
12.02
12.27
7.82
6.51
12.5
11.7
17.2
16.0
7.7
8.0
7.5
dational contact argue against fluvial incision and chan-
nelization. In high-energy settings such as the Okavango,
the distinction may be more difficult to make because hip-
po trails later may become fluvial channels (McCarthy et
al., 1992; 1998). As in modern settings, the depth and
width of excavation of these trails may be reduced with
distance from the core of the hippopotamus habitat.
The second aspect of hippopotamus bioturbation that
may be useful in paleoenvironmental interpretation isthe
zoning of traces that is apparent in the modern setting.
The deepest bioturbation profile is observed in the core
pool/wallow area, with trails emerging from it dendritical-
ly or radially, depending on local hydrology. These trails
generally become narrower toward nearby pasturelands,
not downslope toward the lake. Surrounding deposits con-
tain evidence for shallower bioturbation and smaller trac-
es, such as tracks and trackways. Surrounding deposits
also have a higher diversity of traces, because hippopota-
mus activity in the core pool area can obliterate the traces
of other animals.
Indryland basins, traces such as theseare importantin-
dicators of freshwater conditions, as they represent fo-
cused biological activity dependent on freshwater in an
arid setting (Deocampo and Ashley, 1997). This provides a
basis for interpreting high-resolution aspects of analogous
paleoenvironments, such as in arid-land basins of East Af-
rica during the Pleistocene. Such structures also may be
identified in the more distant geologic past, becausemem-
bers of Hippopotamidae have preferred relatively fresh
aquatic environments since at least the late Miocene
(Kingdon, 1979). The morphology of fossil species, such as
Hippopotamus gorgops, suggests that members of this ge-
nus were even more aquatic than extant species (Coryn-
don, 1976; Stuenes, 1989). Prior to the evolution of the
Hippopotamidae, other large semi-aquatic organisms may
have occupied a similar ecological niche, and exhibited a
similar pattern of interaction with the substrate.
CONCLUSIONS
Large mammals, especially Hippopotamus amphibius,
can create a systematic pattern of traces in arid lake-flat
settings, largely controlled by the localized availability of
fresh surface water. This is characterized by a central bio-
turbation zone extending up to 2 m deep, with dendritic or
radial trails leading out of the wallows up onto drier
ground. Trails are 1–5 m wide, ,0.5 m deep, and infilled
with organic-rich sediment. They narrow and become
shalloweras they emerge ontodrier ground andgrade into
trackways. Other vertebrate trackways, such as buffalo
wallows and other ungulate trackways and trackfields,
may be recorded surrounding the area of hippo-dominated
activity. Sets of traces such as these are important indica-
tors of freshwater conditions in dryland basins due to the
strong affinity of Hippopotamus amphibius for relatively
fresh water. These features can provide important evi-
dence for paleoenvironmental reconstructions of arid
lands during the Neogene and, perhaps, also in the more
distantgeological past, for ecologicallysimilar paleofauna.
ACKNOWLEDGMENTS
Sincere thanks to the Tanzanian Commission on Sci-
ence and Technology and the Ngorongoro Conservation
Area Authority for permission to conduct this research,
and to C. Liutkus, L. Moruo, and A. Tlaa, for able field as-
sistance. Thanks to the Geological Society of America, Sig-
ma Xi- the Scientific Research Society, the Exploration
Club, and the Olduvai Landscape Paleoanthropology Pro-
ject (NSF Archaeology BNS-9000099, SBR-9601065,SBR-
9602478 and EAR-9903258) for support. This paper bene-
fited from helpful discussions with G. Ashley, R. Blumen-
schine, R. Estes, R.Hay, C. Peters, and R. Renaut.
REFERENCES
A
NDERSON
, G.D., and H
ERLOCKER
, D.J., 1973, Soil factors affecting
the distribution of the vegetation types and their utilization by
wild animals in Ngorongoro Crater, Tanzania: Journal ofEcology,
v. 61, p. 627–651.
A
SHLEY
, G.M., D
EOCAMPO
, D.M., and L
IUTKUS
, C.M., 2000, Track re-
cords of large mammals in lake-margin wetlands, L.M. BedII, Ol-
duvai Gorge, Tanzania: International Association of Sedimentolo-
gists, Regional meeting Abstracts, Dublin, p. 72ndash;73.
C
OHEN
, A.S., H
ALFPENNY
,J.,L
OCKLEY
, M., and M
ICHEL
, E., 1993,
Modern vertebrate tracks from Lake Manyara, Tanzania and
their paleobiological implications: Paleobiology, v. 19,p. 433–458.
C
ORYNDON
, S.C., 1976, Fossil Hippopotamidae from Plio-Pleistocene
successions of the Rudolf Basin: in Coppens, Y., Howell, F.C.,
Isaac, G.L., and Leakey, R.E.F., eds., Earliest man and environ-
ments in the Lake Rudolf Basin; Stratigraphy, paleoecology,and
evolution: University of Chicago Press, Chicago, p. 238–250.
D
EOCAMPO
, D.M., 1997, Modern sedimentation and geochemistry of
freshwater springs: Ngorongoro Crater, Tanzania: Unpublished
M.S. Thesis, Rutgers University, 140 p.
D
EOCAMPO
, D.M., 2001, Geochemistry and sedimentology of modern
East African wetlands and a Pleistocene paleo-wetland atOlduvai
Gorge, Tanzania: Unpublished Ph.D. Dissertation, Rutgers Uni-
versity, 154 p.
D
EOCAMPO
, D.M., in press, Sedimentary processes and lithofacies in
lake-margin groundwater-fed wetlands in East Africa: in Renaut,
R.W., and Ashley, G.M., eds., Sedimentation in Continental Rifts:
SEPM, Tulsa.
D
EOCAMPO
,D.M.,andA
SHLEY
,G.M., 1997, Hippo trails:A large-scale
freshwater biogenic structure in arid East Africa: Geological Soci-
ety of America, Abstracts with Programs, v. 29,1., p. 40–41.
D
EOCAMPO
, D.M., and A
SHLEY
, G.M., 1999, Siliceous islands in a car-
bonate sea: Modern and Pleistocene records of spring-fed wet-
lands in Ngorongoro Crater and Olduvai Gorge, Tanzania: Jour-
nal of Sedimentary Research, v. 69, p. 974–979.
D
EOCAMPO
, D.M., A
SHLEY
, G.M., and C
OPELAND
, S.R., 1998, Springs
andstreamsin an East African closedbasin: Controls onbiological
communities in Ngorongoro Crater, Tanzania: Geological Society
of America, Abstracts with Programs, v. 30,7, p. 120.
E
STES
, R.D., and S
MALL
, R., 1981, The large herbivore populations of
Ngorongoro Crater: African Journal of Ecology, v. 19, p. 175–185.
H
AY
, R.L., 1976, Geology of the Olduvai Gorge: University of Califor-
nia Press, Berkeley, 203 p.

NGORONGORO HIPPO TRAILS 217
H
ERLOCKER
,D.J.,and D
IRSCHL
,H.J.,1972, Vegetation oftheNgoron-
goro Conservation Area, Tanzania: Canadian Wildlife Service Re-
port Series, v. 19, p. 1–39.
K
ABIGUMILA
, J., 1993, Feeding habits of elephants in Ngorongoro
Crater, Tanzania: African Journal of Ecology, v. 31, p. 156–164.
K
INGDON
, J., 1979, Hippopotamids: in Kingdon, J., East African
Mammals, v. IIIB: Academic Press, London, p. 250–277.
L
APORTE
,L.F.,and B
EHRENSMEYER
,A.K., 1980, Tracksand substrate
reworking by terrestrial vertebrates in Quaternary sediments of
Kenya: Journal of Sedimentary Petrology, v. 50, p. 1337–1346.
L
EWIS
, D.W., and M
C
C
ONCHIE
, D., 1994, Analytical Sedimentology:
Chapman and Hall, New York, 197 p.
M
C
C
ARTHY
, T.W., E
LLERY
, W.N., and S
TANISTREET
, I.G., 1992, Avul-
sion mechanisms on the Okavango fan, Botswana: The controlof a
fluvial system by vegetation: Sedimentology, v. 39, p. 779–795.
M
C
C
ARTHY
, T.S., E
LLERY
, W.N., and B
LOEM
, A., 1998, Some obser-
vations on the geomorphological impact of hippopotamus (Hippo-
potamus amphibius L.) in the Okavango Delta,Botswana: African
Journal of Ecology, v. 36, p. 44–56.
S
TUENES
, S., 1989, Taxonomy, habits, and relationships of thesubfos-
sil Madagascan hippopotami Hippopotamus lemerlei and H. mad-
agascariensis: Journal of Vertebrate Paleontology, v. 9, p. 241–
268.
W
OLANSKI
, E., and G
ERETA
, E., 1999, Oxygen cycle in a hippo pool,
Serengeti National Park, Tanzania: African Journal of Ecology, v.
37, p. 419–423.
ACCEPTED OCTOBER 3, 2001