Late Quaternary speleogenesis and landscape evolution in a tropical carbonate island: Pango la Kuumbi (Kuumbi Cave), Zanzibar

Abstract

Kuumbi Cave is one of a group of caves that underlie a flight of marine terraces in Pleistocene limestone in eastern Zanzibar (Indian Ocean). Drawing on the findings of geoarchaeological field survey and archaeological excavation, we discuss the formation and evolution of Kuumbi Cave and its wider littoral landscape. In the later part of the Quaternary (last ca. 250,000 years?), speleogenesis and terrace formation were driven by the interplay between glacioeustatic sea level change and crustal uplift at rates of ca. 0.10-0.20 mm/yr. Two units of backreef/reef limestone were deposited during ‘optimal’ (highest) highstands, tentatively correlated with MIS 7 and 5; (mainly) erosive marine terraces formed in these limestones in ‘suboptimal’ highstands. Kuumbi and other sub-terrace caves developed as flank margin caves, in the seaward portion of freshwater lenses during such ‘suboptimal’ highstands. Glacioeustacy-induced fluctuations of the groundwater table may have resulted in shifts from vadose (with deposition of well-developed speleothems) to phreatic/epiphreatic conditions in these caves. At Kuumbi, Late Pleistocene (pre-20,000 cal. BP) ceiling collapse initiated colluvial deposition near-entrance and opened the cave to large plants and animals, including humans. A phase of terminal Pleistocene human occupation ca. 18,500-17,000 cal. BP resulted in the deposition of a dense assemblage of Achatina spp. landsnails, alongside marine molluscs and mammal remains (including zebra, buffalo and other taxa now extinct on Zanzibar). The Holocene part of the cave stratigraphy near-entrance records phases of abandonment and intensified late Holocene human use.

Full text

Available online at scholarcommons.usf.edu/ijs
International Journal of Speleology
Off icial Journal of Union Internationale de Spéléologie
*nkouramp@exseed.ed.ac.uk
INTRODUCTION
Many islands contain extensive swathes of
geologically young, eogenetic limestones that have
not undergone burial diagenesis. Karst in such
limestones often differs in many aspects from that in
older, telogenetic limestones (e.g. Vacher & Mylroie,
2002; Ginés & Ginés, 2007; Mylroie, 2013). The
Citation:
Keywords:
Abstract: Kuumbi Cave is one of a group of caves that underlie a flight of marine terraces in Pleistocene
limestone in eastern Zanzibar (Indian Ocean). Drawing on the findings of geoarchaeological
field survey and archaeological excavation, we discuss the formation and evolution of
Kuumbi Cave and its wider littoral landscape. In the later part of the Quaternary (last ca.
250,000 years?), speleogenesis and terrace formation were driven by the interplay between
glacioeustatic sea level change and crustal uplift at rates of ca. 0.10-0.20 mm/yr. Two units
of backreef/reef limestone were deposited during ‘optimal’ (highest) highstands, tentatively
correlated with MIS 7 and 5; (mainly) erosive marine terraces formed in these limestones
in ‘suboptimal’ highstands. Kuumbi and other sub-terrace caves developed as flank margin
caves, in the seaward portion of freshwater lenses during such ‘suboptimal’ highstands.
Glacioeustacy-induced fluctuations of the groundwater table may have resulted in shifts from
vadose (with deposition of well-developed speleothems) to phreatic/epiphreatic conditions
in these caves. At Kuumbi, Late Pleistocene (pre-20,000 cal. BP) ceiling collapse initiated
colluvial deposition near-entrance and opened the cave to large plants and animals, including
humans. A phase of terminal Pleistocene human occupation ca. 18,500-17,000 cal. BP
resulted in the deposition of a dense assemblage of Achatina spp. landsnails, alongside
marine molluscs and mammal remains (including zebra, buffalo and other taxa now extinct
on Zanzibar). The Holocene part of the cave stratigraphy near-entrance records phases of
abandonment and intensified late Holocene human use.
Carbonate Island Karst; cave deposits; archaeology; Zanzibar; East Africa
Received 15 April 2015; Revised 22 August 2015; Accepted 23 August 2015
Kourampas N., Shipton C., Mills W., Tibesasa R., Horton H., Horton M., Prendergast M.,
Crowther A., Douka K., Faulkner P., Picornell L. and Boivin N., 2015. Late Quaternary
speleogenesis and landscape evolution in a tropical carbonate island: Pango la Kuumbi
(Kuumbi Cave), Zanzibar. International Journal of Speleology, 44 (3), 293-314. Tampa, FL
(USA) ISSN 0392-6672
http://dx.doi.org/10.5038/1827-806X.44.3.7
Late Quaternary speleogenesis and landscape evolution
in a tropical carbonate island: Pango la Kuumbi
(Kuumbi Cave), Zanzibar
Nikos Kourampas
1*
, Ceri Shipton
2
, William Mills
3
, Ruth Tibesasa
4
, Henrietta Horton
5
,
Mark Horton
6
, Mary Prendergast
7
, Alison Crowther
8
, Katerina Douka
3
, Patrick Faulkner
9
,
Llorenç Picornell
10
, and Nicole Boivin
3
1
Biological and Environmental Sciences, University of Stirling, & Office of Lifelong Learning, University of Edinburgh, Scotland, UK
2
British Institute in Eastern Africa, Nairobi, & McDonald Institute for Archaeological Research, University of Cambridge, UK
3
School of Archaeology, University of Oxford, UK
4
History Department, Kyambogo University, Uganda
5
The MENTOR Initiative, Monrovia, Liberia
6
School of Archaeology, University of Bristol, UK
7
Department of Sociology and Anthropology, St. Louis University in Madrid, Spain
8
School of Social Science, The University of Queensland, Australia
9
Department of Archaeology, School of Philosophical and Historical Inquiry, The University of Sydney, Australia
10
Department of Historical Sciences and Theory of Art, University of the Balearic Islands, Spain
International Journal of Speleology 44 (3) 293-314 Tampa, FL (USA) September 2015
proliferation of island cave studies in recent years
reflects a recognition of the genetic distinctiveness and
wider scientific significance of these landforms. The
concept of flank margin speleogenesis and its more
comprehensive derivative, the Carbonate Island Karst
Model (Mylroie & Carew, 1990; Mylroie et al., 1995;
2008; Mylroie & Vacher, 1999; Mylroie & Mylroie,
2007; Mylroie, 2013), have further highlighted this

294 Kourampas et al.
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
distinctiveness by emphasising the crucial role of
carbonate dissolution in the distal, seaward part of
the freshwater lens underneath carbonate islands
(and, also, continental littorals). In this speleogenetic
realm, discrete from those of epigenic and hypogenic
speleogenesis, critical domains of carbonate
dissolution (groundwater table; fresh/saltwater
mixing zone) are highly responsive to relative sea level
change, in turn driven by eustatic sea level change (at
several timescales), and isostatic or tectonic crustal
uplift or subsidence (Ginés & Ginés, 2007; Mylroie
& Mylroie 2007; Fratesi, 2013; Ginés et al., 2014).
Carbonate island caves could thus be regarded as
nodes where several lines of enquiry on the workings
of the earth system come together.
Related to the fundamental question of carbonate
island speleogenesis is the research theme of island
cave deposits: speleothems and, increasingly, detrital
sediments (Sasowsky & Mylroie, 2004; Fornós et
al., 2014). As potential links between offshore and
continental records, island cave deposits may be of
critical import for regional and global stratigraphic
Fig. 1. Zanzibar, and places mentioned in the text. a) Location; b) regional tectonic setting, with dominant
directions of plate motion (thick arrows) and clockwise rotation of the Rowuma Plate (from Nicholas et al.,
2007, simplified); c) geology (from United Nations, 1987 and Bron Sikat, 2011, simplified).
correlations of the Quaternary System. Island cave
deposits contain proxies for the long-term history
of large-scale, ocean basin-wide or global climatic
systems (e.g. Asian Monsoon, El Niño/La Niña), sea
level change, surface denudation, etc., and, also, the
ecological (including human-ecological) histories of
island landscapes (c.f. De Waele, 2009; Lace & Mylroie,
2013). Such records may be crucial for understanding
and modelling feedbacks between the atmosphere,
the global ocean, earth surface processes, island
biomes and human societies at timescales of <10
2
to
10
6
years. Interpretation of island cave sediment
proxies is underpinned by an understanding of how
the caves that contain them, and the island landscapes
of which caves are a part, form and evolve (De Waele,
2009; Bover et al., 2014; Fornós et al., 2014).
Here we attempt to reconstruct the formation
and evolution of Kuumbi Cave (Pango la Kuumbi in
Zanzibar’s Kiswahili language), and its surrounding
littoral landscape in the island of Zanzibar, equatorial
Indian Ocean (Fig. 1). We do this by drawing on
geological, geomorphological, stratigraphic and

295Speleogenesis and landscape evolution: Kuumbi Cave, Zanzibar
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
archaeological evidence collected as part of the multi-
disciplinary Sealinks Project (funded by the European
Research Council). Sealinks has undertaken field
investigations in eastern Africa in order to examine
the emergence and impact of early trading networks
across the Indian Ocean (Fuller & Boivin 2009; Fuller
et al., 2011, 2014; Boivin et al., 2012, 2013, 2014;
Helm et al. 2012; Shipton et al., 2013; Crowther et
al., 2014a,b).
Kuumbi Cave is well suited for studying island
speleogenesis since, a) its evolution can be constrained
plausibly in a late Quaternary timeframe and,
b) archaeological stratigraphy provides a means for
dating some of the geomorphic events that reshaped
the cave and its wider landscape.
Discovery, archaeological significance and history
of research
Kuumbi Cave (S 6°21’40’’; E 39°32’33’’) is situated
in the Jambiani district, on Zanzibar’s eastern coast.
The cave was first reported as an archaeological site
in 2004, by Dar es Salaam archaeologist Felix Chami,
who was guided there by Jambiani residents (Chami,
2009). Successive archaeological excavations, by
the universities of Dar es Salaam, Uppsala and, in
2010-2012, Oxford, have demonstrated a long history
of human presence from the Late Pleistocene to the
present (Sinclair et al., 2006; Chami, 2009; Shipton
et al., in press).
Kuumbi’s archaeological record may be crucial for
documenting socioeconomic transitions in the East
African littoral in early, pre-Islamic times. For this
reason, Kuumbi Cave was one of the main foci of the
Sealinks Project. Alongside archaeological excavation,
the project undertook topographical/ geomorphological
mapping and geoarchaeological fieldwork, with the
following objectives: (1) to reconstruct the geomorphic
evolution of the cave and its wider landscape;
(2) to identify processes of sediment deposition,
especially during periods of human presence, and
postdepositional change that may have affected the
integrity of the archaeological record; and, (3) to
sample cave sediments for geoarchaeological and
other palaeoenvironmental analyses.
GEOLOGICAL AND GEOMORPHIC SETTING
The island of Zanzibar, separated from the East
African mainland by a narrow, fault-controlled
shallow strait (ca. -30 m in its shallower parts),
originated from block faulting and differential uplift of
the Neogene Ruvu-Rufuzi Delta, perhaps the largest
deltaic depocenter in Neogene East Africa (Kent et al.,
1971; Fig. 1). The axial parts of the island consist of
Lower Miocene siliciclastics (channel conglomerates
and sandstones; interchannel muds and marls),
interdigitated and fringed with various Miocene
limestones (grainstones, framestones). Underlying
these there must be earlier, Mesozoic and Palaeozoic
(Cretaceous, Jurassic, Karoo) sediments, similar
to those cropping out in the mainland littoral, and,
even deeper, Precambrian crystalline basement
(cf. Mpanda, 1997).
Miocene sediments are overlain by extensive
Pleistocene limestones (“Azania Series”: Stockley,
1928) and Late Pleistocene-Holocene siliciclastics
(aeolian sands, fluvial sands and gravel, colluvia
from the reworking of Miocene sediments, etc.) and
red latosols. Zanzibar is girdled by a living coral reef
and shallow back-reef, with lagoons, platforms, sandy
beaches and mangroves (Arthurton, 2003; Punwong
et al., 2013a,b).
Pleistocene limestone of the ‘Azania Series’ extends
over the entire region of central-eastern Zanzibar,
from ca. +30 m to below present sea level (Fig. 1).
This heterogeneous lithostratigraphic unit comprises
several limestone facies, deposited in various inner-
shelf environments (reef framestones; reef slope
breccia; backreef packstones/wackestones; shoal
and beach grainstones, etc.). Azania Series limestone
hosts numerous caves, solution pipes and collapse
dolines (Fig. 1).
Miocene and Pleistocene sediments are cut by
kilometre-long, shoreward-dipping normal faults
parallel to the island’s morphological strike (N-S to
NNW-SSE: Fig. 1). Faulting and jointing of Pleistocene
(including inferred Late Pleistocene) limestone
suggests that Zanzibar (and its adjacent islands and
continental coast) remains tectonically active.
The cave in its landscape
Kuumbi Cave is situated about 3 km from the
Jambiani shore, on Zanzibar’s eastern coast (Figs.
1,2). It is one of several caves underneath a marine
terrace at +24-27 m (here termed the ’25 m terrace’)
– the highest of a flight of (five?) marine terraces on
Pleistocene Azania Limestone (Fig. 2). Limestone
exposed on the Kuumbi Cave walls and the overcave
terrace is mainly bivalve-gastropod packstone/
grainstone with high mouldic porosity (Fig. 3). This
facies probably originated in a back-reef setting.
Kuumbi opens to the surface via a large collapse
doline and several metre-sized, vertical gaping chasms,
some with solution-smoothened inner surfaces (Fig.
3). Around the cave entrance, the landscape is a
sacred grove: Kuumbi is a sacred site for the local
community, and a strong taboo prohibits tree felling
around it (Chami, 2009). Further away, overgrown
fields, seemingly abandoned drystone walls and a few
fields under cultivation evidence that the overcave
terrace is part of an old agricultural landscape.
At field observation scale, relief on the overcave
surface does not exceed ca. 1.5-2.5 m. Large parts
of this surface are devoid of soil cover; elsewhere it
is covered with loose, angular regolith or, where
woodland is present (e.g. grove around the cave
entrances, overgrown fields), by a thin (≤ 60 cm),
dark, redzina-type soil (Fig. 3).
Vestiges of an earlier, cemented weathering mantle
are also present on the 25 m terrace, in the form of
metre-sized, heavily karstified bodies of reddish,
clast- to matrix-supported breccia (Fig. 3). This
deposit testifies to the development of regolith/soil
cover under quite different past (Late Pleistocene?)
climatic conditions, and the subsequent denudation
of the terrace.

296 Kourampas et al.
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
Fig. 2. a) Kuumbi and other caves of the Jambiani coast; b) Schematic geological section from Kuumbi Cave to the shore (not in scale);
c) Interpretation of observed lithologies and landforms as resulting from at least two glacioeustatic sea level cycles (see text).
Collapse chasms and caves in Kuumbi’s immediate
vicinity (Fig. 3) and the presence of caves at lower
terrace levels (e.g. Kikuaju Cave Springs: two caves
flooded by brackish water; Fig. 4) suggest that Kuumbi
Cave is part of a host of karstic cavities that perforate
Azania Limestone from (below ?) present sea level to
ca. +25 m.
The Jambiani terrace ight
Like other limestone terraces on the East African
coast, from Somalia to Mozambique, the Azania
Limestone and its terraces are thought to be Mid- to
Late Pleistocene in age. Earlier workers (Arthurton et
al., 1999; Arthurton, 2003) have correlated the Azania
limestone with one or more substages of the Last
Interglacial (MIS 5). To our knowledge, nonetheless,
no absolute dates for this, or other upper Quaternary
limestones in Zanzibar or the Tanzanian mainland,
are available. Potentially correlative upper Quaternary
limestones in coastal Kenya have yielded dates from
240,000 (+70,000 / -40,000) BP (
230
Th /
234
U date on
coral: Battistini, 1977) to ca. 21-40,000 BP (
14
C dates
on coral: Oosterom, 1988).
A date of ca. 44,000 +1900/−1500 BP (
14
C -
uncalibrated) from siliciclastic sediments underlying
the 25 m terrace in Lindi Bay, southern Tanzania
may suggest that parts of the Tanzanian coast have
undergone very rapid late Quaternary uplift (Reuter
et al., 2010). This date, however, is close to the range
limit of radiocarbon dating. In addition, lithological
and tectonic differences between the Jambiani and
the Lindi Bay coasts (the latter is situated on a major

297Speleogenesis and landscape evolution: Kuumbi Cave, Zanzibar
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
Fig. 3. Landscape, bedrock lithologies and palaeosols around Kuumbi Cave. a) On approach to Kuumbi Cave, looking west (inland): Small plots
enclosed with drystone walls. The cave is situated in the grove in the background; b) A collapse opening on the 25 m terrace provides a window
to another underground chamber, a few metres south of Kuumbi; c) ‘Older Azania Limestone’ underlying the 25 m terrace: porous packstone/
grainstone with moulds of bivalves and gastropods. Inset: detail of freshly exposed rock surface. Hammer: 33 cm; d) The same limestone on the
Kuumbi Cave wall. Note selective dissolution of marine molluscs and other skeletal grains (mouldic porosity). The green colour is due to microbial/
algal colonisation of the cave walls. Wristwatch diameter: 33 mm; e) Vestiges of karstified regolith on the 25 m terrace, about 500 m NNE of the
cave entrance. Inset: Freshly exposed surface of the latter, showing poor sorting of angular, corroded limestone clasts; f) In situ coral reef in
‘Younger Azania Limestone’, about 1.5 km NNE of Kuumbi Cave. Inset: Small polyp stony coral colony on the same reef; g) NNW-SSE striking joint
(parallel to hammer handle) cutting ‘Younger Azania Limestone’ (Last Interglacial?) on the lower, ≤ 12 m terraces.
Fig. 4. Caves of the Jambiani karst. a) Looking down the vegetated talus slope of the NE entrance to Kuumbi Cave; b) Kuumbi Cave entrance, with
speleothem column, truncated stalactites and roof fall blocks. The concrete pedestals mark earlier excavations by the University of Dar es Salaam.
Looking out from Chamber A; c) Low-ceilinged alcove in Kuumbi Cave. Note cusps on the walls and speleothem rubble on the floor; d) Entrance to
Kikuaju A cave spring, on the ≤12 m terrace(s); e) Inside Kikuaju A cave spring: cusps and bedrock pendants on the ceiling and speleothems on the
sloping floor (right hand side); f) A weathered stalagmite, partly submerged in brackish water in Kikuaju A (torch reflection marks the water surface,
which corresponds to the tidal sea level. Note the development of a thin ledge around the stalagmite.

298 Kourampas et al.
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
fracture zone within the Rowuma Microplate: Nicholas
et al., 2007; Reuter et al., 2010), render correlation of
the 25 m terrace in Jambiani with this relatively late
date untenable.
Evidence from a limited survey from ca. 500 m west
of Kuumbi Cave to the shore is summarised in Fig. 2.
This evidence is inconclusive: very limited drainage
incision across the terrace flight, dense, often
impenetrable vegetation, and the ubiquitous drystone
walls around fields cultivated and abandoned
prevented the discovery of tell-tale outcrops. The
following considerations, however, lend some tenuous
support to our interpretation of the Azania Limestone
as comprising at least two distinct lithostratigraphic
units (Fig. 2):
1) The 25 m terrace preserves remains of regolith/
palaeosol (Fig. 3). No similar palaeosol was identified
on the lower terrace(s).
2) Overall, limestone of the lower (≤12 m) terraces
appears less intensively karstified and denuded than
that of the 25 m terrace (e.g. caves under the lower
terraces have fewer/smaller openings to the surface).
Effects of lithological composition notwithstanding
(lower limestones contain reef and reef slope facies
with very porous coral fragments and in situ coral
colonies: Figs. 3, 4), it is possible that differences in
intensity of karstification and surface denudation
reflect the older age of the 25 m terrace limestone.
3) Two ‘sets’ (lithostratigraphic units) of ‘Azanian
Series’ coral limestone, one at ca. 12 m and one at
ca. 7.5 m were distinguished by Stockley (1928) in
neighbouring Pemba Island (Fig. 1) – an assertion
confirmed by later geologists (Caswell, 1956). It is
likely that these two limestone units correlate with
principal terrace levels (25 and ≤12 m) in eastern
Zanzibar. Altitudinal differences between limestone
units across the two islands are small enough to be
attributable to field measurement uncertainties and/
or neotectonic faulting or differential uplift.
4) Terraced limestones in coastal Kenya (Fig. 1)
are also resolved into at least two stratigraphic units
(Braithwaite, 1984; Oosterom, 1988; Abuodha, 2004).
The earlier/higher of these units, associated with the
ca. 30 m terrace (Abuodha, 2004), is mainly a ‘back reef’
facies (Braithwaite, 1984), broadly comparable with the
limestone of the 25 m terrace in eastern Zanzibar.
We thus interpret the observed surface geology as a
terrace flight eroded on at least two stratigraphically
distinct limestone units (sequences), each correlated
with a sea level highstand: the ‘Older Azania Limestone’
hosting Kuumbi and other caves underneath the
25 m terrace; the ‘Younger Azania Limestone’, hosting
the two Kikuaju cave springs and, reportedly, other
caves underneath the ≤12 m terraces. These sequences
are separated by (inferred) erosional unconformities/
disconformities, corresponding to marine regressions
and sea level lowstands.
In this scenario, each of the two sequences
corresponds to orbital-scale sea level cycles,
correlative with main MIS stages. The time of
deposition of the Older Azania Limestone is unknown:
it may date from the penultimate interglacial, MIS
7, to earlier Pleistocene, or (less likely) even earlier,
Neogene highstands. Chronological uncertainties
notwithstanding, our working hypothesis correlates
the Older and Younger Azania limestones with
the (composite) MIS 7 and MIS 5, respectively. The
highest/earliest terraces on each limestone sequence
date from their depositional highstands: MIS 7 for the
25 m terrace; MIS 5 (5e?) for the 10-12 m terrace,
respectively (Fig. 2). It is probable that higher-order
sea level cyclicity during each of these highstands
resulted in erosion of marine terraces, deposition of
higher-order carbonate units and/or reoccupation
of earlier terrace levels. Although these events are
unresolved, some of the relatively closely-spaced and
poorly defined terraces below ca. 12 m may thus have
resulted from erosion in MIS 5 substages.
Assuming that present terrace surfaces are a (very)
rough approximation of sea level at the time of their
formation (inner shelf relief is no more than ca. 10 m
in present day Zanzibar: cf. Arthurton, 2003), taking
-15 to -5 m and + 4 to + 6 m as the highest eustatic
sea levels in MIS 7 and MIS 5 (5e), respectively (cf.
Siddall et al., 2006), and factoring in about 8.75 and
16.8 m of denudation for the lower and higher terraces,
respectively (based on the indicative denudation rate of
eogenetic limestones: ca. 70mm/1000 yrs – Jennings,
1985 – and consistent with evidence from other
tropical carbonate islands: e.g. Miklavic et al., 2012),
this correlation suggests late Quaternary crustal
uplift at the rates of 0.10 to 0.20 mm/yr. These –
poorly constrained and purely indicative – uplift rates
are of a similar order of magnitude with the rate of
0.1 mm/yr calculated independently for parts of
coastal (Oosterom, 1988) and inland Kenya (Veldkamp
et al., 2007). The later part of the Quaternary, perhaps
since 250,000 BP, may, therefore, be a plausible
timeframe for the formation of Jambiani terraces and
their underlying caves.
CAVE MORPHOLOGY
Kuumbi Cave comprises two large chambers and
several smaller side alcoves, anastomosing and/or
appending into progressively narrowing, low-ceilinged
cavities, often developed around bedrock pillars. In
plan, morphology is that of a spongework maze (sensu
Palmer, 2011: 9). Some of the peripheral cavities, too
narrow to explore, may link Kuumbi Cave with other
karstic chambers (Figs. 5, 6).
The cave is accessed from two entrances: a large
collapse doline (surface extent ca. ½ of Kuumbi’s
largest underground chamber) in the NE, and a
smaller collapse doline in the SW. Four other circular
openings, resulting from collapse of the solutionally
undermined cave ceiling, also link the cave with the
surface.
The two main chambers are oriented NW-SE and NE-
SW (Fig. 5), perpendicular and parallel, respectively,
to the 25 m terrace cliff (about 60 m east of the cave
entrance: Fig. 1). Smaller alcoves and tubes, although
more variable in orientation, also conform with these
two principal strikes. These directions coincide with
the strikes of – rare – joints in the cave walls and in
Pleistocene limestone outcrops further from the cave

299Speleogenesis and landscape evolution: Kuumbi Cave, Zanzibar
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
Fig. 5. Geomorphological plan (a), and profiles (b) of Kuumbi Cave.

300 Kourampas et al.
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
Fig. 6. Solutional forms and speleothems in Kuumbi Cave. a) Chamber A, viewed from the Well. Stepped
floor in the foreground was formed on indurated sediment; b) Chamber A, looking towards the Well
(behind people on the right). Triangles mark notch at ca. 170 cm above present floor level. Another
notch is present a few centimetres above; c) Side alcove off Chamber B. Detrital sediments bury earlier
speleothems and erosional relief (bedrock pendants). The floor rubble is mainly fragmented carbonate
crust, possibly correlative with a later (late Holocene?) phase of CaCO
3
deposition (see text). Note the
apse and the horizontal ceiling on the left side, seemingly correlative with the notch in Fig. 6b; d) Small
cupola and cusps on the cave ceiling. The rusty-brown spots are speleothems, probably of high humic
acid content; e) Narrow vertical cavity off Chamber B, developed along N-S-trending joint. Human-made
engravings (not visible here) are present on the iron oxide-stained cavity walls; f) Heavily weathered
stalagmite; Chamber A; g) Corrosion of speleothems (above hammer), probably due to guano from a
bat-roosting cupola overhead. SW cave entrance.
(Figs. 3, 5). The coincidence between joint strikes and
the orientation of regional tectonic structures (NNW-
SSE NNE-SSW faults that control continental shelf
morphology across the Zanzibar Archipelago: Mpanda,
1997; Chorowicz, 2005; Nicholas et al., 2007; Fig. 1),
suggest that these joints are of neotectonic origin.
In cross section, the exposed part of Kuumbi Cave
is shaped like an open U. Cave profile is modified
by steep rock fall/talus cones at the two cave
entrances (Fig. 5) and by finer grained inner-cave
sediments. These obscure the geometry of the lower
parts of the cave.
Ledges and notches (from 0.4 to 1.7 m above cave
floor) and a shallow, partly buried rockshelter about
2 m below the overcave terrace (Fig. 5) manifest
a succession of karstic dissolution levels. Metre-
sized cavities under floor sediment (Fig. 5) suggest
that karstic dissolution extended below the detrital
sediment/bedrock interface.
In Chamber A, the cave floor dips from each entrance
to the ‘cave well’, a ca. 2.5 m-deep depression where
water pools in the rainy season, probably as a result
of vadose water perching. At the time of visit (late
August - early September 2012) the well was dry,
but the muddy sediment of its floor was moist. The
‘cave well’ appears to have been enlarged artificially,
probably to facilitate procurement of drinking water.
Entrances and ceiling openings permit sunlight
and air circulation in the main cave chambers and
provide entry points for rainwater, surface runoff and

301Speleogenesis and landscape evolution: Kuumbi Cave, Zanzibar
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
et al., 2011 for overviews). Proposed speleogenetic
contexts include phreatic dissolution (Sweeting, 1972;
Trudgill, 1985; Sancho et al., 2004; Piccini et al., 2007),
possibly at the site of local convection cells in conditions
of “sluggish rising forced flow” (with “less dense and
more aggressive water” located in the upper parts of
these cells: Klimchouk, 2009); pressure increase and
dissolution above pooled groundwater (Lismonde,
2000), to vadose condensation corrosion, mediated by
roosting bats (Lundberg & McFarlane, 2009).
At Kuumbi, where the groundwater table is shallow
and bats abound, cupolas may have resulted from
any of these processes. The close spatial association
of cupolas with other phreatic dissolution landforms
(e.g. bedrock pendants, cusps), nonetheless, suggests
that these features had a phreatic (to epiphreatic)
origin. Bat roosting probably modified them, but it
appears unlikely to have formed them.
Cupolas are deep enough to undermine the cave
ceiling, especially in parts of the cave where they
cluster together. Circular openings through the cave
ceiling to the surface probably resulted from ceiling
collapse following denudation of the overcave surface
(c.f. Birmingham et al., 2011). These openings probably
concentrated at the location of cupola clusters.
Bedrock pendants
Sharp-edged pendants of bedrock limestone –
remnants of bedrock that escaped dissolution – are
present in side alcoves (Fig. 6). Bedrock pendants
are thought to result from pervasive, mainly upward-
directed dissolution in phreatic (Sweeting, 1972;
Klimchouk, 2009) and/or paragenetic (when much of
the cave chamber was filled with sediment) conditions
(Farrant & Smart, 2011).
In Kuumbi Cave, there is no (direct) evidence for
former occlusion of chambers by sediment, whereas
bedrock pendants are associated with a plethora
of other, evidently phreatic meso/microforms. A
paragenetic interpretation of bedrock pendants
would produce an unduly complex narrative of
cave evolution. We thus interpret these as phreatic
landforms, resulting from upward dissolution by
slowly circulating aggressive water.
Wall notches
One or two poorly expressed notch(es) are present
on the southeastern wall of Chamber A, between
1.70-2.20 m above the sloping (present) cave floor
(Figs. 5, 6). Parts of these discontinuous notch(es)
present as a singular embayment; other parts present
as a series of closely spaced, decimetre-sized cavities
concentrated at the same level.
These notches appear to correspond approximately
to the level of a flowstone shelf around a column
underneath the present cave entrance, and, also, to the
upper level of mammillate (subaqueous?) speleothems
in Chamber B. These notches may mark one (or more)
former groundwater table stillstand(s), but whether
wall notching and speleothem deposition took place
concurrently or, and more likely, in different phases
of cave flooding by groundwater, cannot be confirmed
on current evidence.
sediment washed in from the surface. These openings
thus impart an environmentally liminal character,
akin to that of a cave entrance, or a large rockshelter.
These ‘open cave’ conditions contrast with those in
smaller, low-ceilinged alcoves (Fig. 6). The latter, which
seem to host most of the cave’s resident bats, were
perceptibly more humid and warmer at the time of
visit, with a somewhat stagnant atmosphere, probably
enriched in CO
2
from decomposition of organic matter
(mainly bat guano) on their floor, and condensation
droplets on the ceiling. It was in these alcoves that
some active water seepage was noted, but there was
little correlation between this and speleothems: most
of the well-developed stalactites appeared inactive.
Small/medium-scale forms
Cusps
Decimetre-scale, shallow, concave cavities are
ubiquitous on the walls and ceilings of those parts
of the cave that have not been modified by later
collapse. Cusps are present as juxtaposed and cross-
cutting clusters of cavities, distributed over the
entire exposed limestone surface (Fig. 6), locally in
association with bedrock pendants and pillars (see
below). They resemble scallops, but appear to be non-
directional: elliptical rather than shallowing-outwards
in cross section.
At Kuumbi Cave, cusps and alcoves appear to be
outcomes of the same dissolutional process: cusps
appear to have formed at the dissolution front as low-
ceiling alcoves became enlarged. We thus interpret
cusps in Kuumbi as “primary dissolution features”
(sensu Frank et al., 1998). Cusps have been reported
from many tropical and temperate flank margin caves
(Frank et al., 1998; Mylroie et al., 2001, 2008; Myrloie
& Mylroie, 2009); there they have been attributed to
slow phreatic dissolution.
Links and tapering tubules
Oval-shaped, generally smooth-walled cavities
(<10-100 cm), that often follow dominant orientations
of larger cave passages and joints, link side alcoves
of the cave, lead to other, currently inaccessible
chambers, or appear to continue for some length in
the bedrock (Fig. 5). These tubes are interpreted as
phreatic in origin, suggesting that Kuumbi Cave was
water-filled in earlier stages of its evolution.
Solution cavities (‘cupolas’)
Large (up to 1.5 m maximum diameter; up to
2.5-3 m deep), cylindrical to gently tapering up,
vertical cavities are very frequent on the cave ceiling
(Figs. 5, 6). Cupolas (also termed ‘bell holes’) are
often concentrated in linear clusters, or as groups
of randomly juxtaposed, cross-cutting pockets, often
in the axial part of much broader, domed sections of
the ceiling (Fig. 5). These landforms are favourite bat-
roosting sites.
Their vertical development and narrowing-up cross-
section suggest that cupolas were formed from upward
dissolution. This process may have operated in a wide
range of underground conditions, so cupola genesis
is subject to debate (c.f. Osborne, 2004; Birmingham

302 Kourampas et al.
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
Kuumbi’s speleothems are grouped together in Unit
St. These include various forms, from floor deposits,
with some very large stalagmites and columns and
a flowstone cascade (near the SW cave entrance),
to straw and conical stalactites and various types
of globular speleothem. Mammillate speleothems
that may have been deposited underwater or from
supersaturated films(?) are also present locally,
largely buried under later detrital sediment (Fig. 9).
The rusty brown colour of some speleothems on the
cave ceiling (e.g. Fig. 6) is probably due to their high
content in soil-derived humic acids (van Beynen et
al., 2001). Porous flowstone of algal-microbial origin
is also present on roof-fall blocks near the SW cave
entrance. This deposit evidently post-dates ceiling
collapse and the opening of the entrance (Fig. 9).
Speleothem deposition was manifestly diachronous
and multiphase, as suggested by the juxtaposition
of several speleothem forms, and the interdigitation
of flowstone and detrital sediment locally. The bulk
of speleothem deposition appears to have predated
detrital sedimentation, but there is evidence for late(?)
Holocene resumption of carbonate deposition in the
cave (see below). Chronological resolution of speleothem
deposition requires further dating and field evidence.
Cave entrance
Metre-sized rounded boulders of bedrock limestone
and speleothems by the NE cave entrance (Unit III
TR10
)
may signal ceiling collapse linked to the formation
of this entrance, probably in Late Pleistocene times.
These boulders, evidently deposited as open-work
breccia, were buried in reddish yellow pebbly loam
(Unit IV
TR10
: contexts 1026-1025; Fig. 7) with Achatina
spp. landsnails, sporadic fragments of marine
molluscs, leopard, and relatively abundant small
bovids and bat remains. Charcoal from this unit
was dated to around 20,000 cal. BP (Fig. 8). This,
and one earlier radiocarbon date of landsnail shell
(21695±300 BP: Sinclair et al., 2006, which calibrates
to around 25-26,000 cal. BP), suggest that this unit
was deposited on the eve of the Last Glacial Maximum
(LGM). This interboulder loam was probably deposited
in a talus cone, with sediment supplied from the
reworking of earlier soils/sediments on the overcave
surface, perhaps in conditions of sparser vegetation
and/or more pronounced rainfall seasonality on the
approach to the LGM.
Unit V
TR10
(contexts 1024-1018), above the latter
(Fig. 7), is a dense accumulation of Achatina spp.
landsnails, mingled with diverse mammal remains
(predominantly small bovids; also bushpig, hyrax,
small primates, and, significantly, larger taxa now
extinct from Zanzibar: zebra, buffalo, waterbuck,
reedbuck, bushbuck, bush duiker and possible
Thomson’s gazelle), small amounts of fish bone,
marine molluscs (predominantly Nerita spp. and
Lunella coronata), charcoal and lithics. Up to 11% of
bones are burnt; remains of monkey, dwarf bovid and
larger bovis are cutmarked. This deposit, dated to
around 18,500-17,000 cal. BP (Fig. 8), is the earliest
unequivocal indication of regular human presence in
the cave (Shipton et al., in press).
Floor cavities
A low-ceilinged chamber (≥2 m) extends under
the cave floor near the cave well. This chamber was
dissolved in bedrock and indurated floor sediment,
and is partly filled with organic mud similar to that
deposited at the well (Fig. 5). Although the chamber
dimensions and morphology could not be assessed
(its opening is too narrow to enter), its presence
hints at a complex succession of detrital deposition,
cementation and dissolution phases.
A few metres from that chamber, also by the cave
well, indurated detrital sediment is penetrated by a
network of <20 to 60 cm-deep solutional cavities and
numerous smaller-scale pores, occluded by flowstone
and later detrital deposits.
Dissolution of speleothems
At Trench KC12, underneath a cupola cluster, a
carbonate floor crust is perforated by a network of
decimetre-scale, irregular solutional cavities filled
with organics and phosphate-rich, probably guano-
derived mud (Fig. 7). Accelerated dissolution of the
floor crust there was probably due to the low pH
effluents generated by guano diagenesis. Guano-
induced pitting of speleothems, and their diagenetic
alteration (to as yet unidentified phosphates?) is
present on some stalagmites (Fig. 6).
Speleothem dissolution microforms occur elsewhere
in the cave: cm-scale solution features resulted from
aggressive water trickling over ledges solutional
microrelief cuts across stalagmite lamina locally;
larger, m-sized cavities and windows through
stalactites and draperies may have resulted from
condensation corrosion.
KUUMBI CAVE SEDIMENTS:
STRATIGRAPHY AND FACIES
A stratigraphic synthesis of Kuumbi Cave, compiled
from the three excavated trenches and several sediment
exposures, is shown in Fig. 7. The cave stratigraphy,
of thickness ranging from ca. 2.5 to 1.5m, is resolved
into nine depositional units of detrital deposits and
speleothems. Most of these units group together several
stratigraphic contexts (sensu Harris, 1989) identified
on excavation, and can be correlated across trenches,
either within each chamber or cave-wide. These units
are enumerated with Roman numerals (with a trench
identifier where appropriate). Two depositional units
which were not exposed in the excavated trenches, and
whose stratigraphic position was inferred tentatively
from lateral relationships, are listed with descriptive
abbreviations (units St: multiphase speleothems and
IL: indurated loams). Radiocarbon and OSL dates
on charcoal, bone and ceramics collected during the
SEALINKS excavation from the near-entrance Trench
KC10 are shown in Fig. 8 (see Shipton et al. (in
press) for a discussion of dating methodologies and
chronostratigraphic interpretation).
The basal sediments of Kuumbi Cave include a
red siltstone of unknown age (Unit I) and altered
carbonate crusts (Unit II), locally with extensive iron
mottling (e.g. Trench KC11).

303Speleogenesis and landscape evolution: Kuumbi Cave, Zanzibar
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
Fig. 7. Cave floor stratigraphy in near-entrance (Trench 10: top) and inner cave settings (Trench 11: middle, Trench 12: bottom; Insert: schematic
stratigraphic synthesis). I to IX: stratigraphic units (see text). Rose diagrams: clast orientation. Top: a) Heavily weathered, crumply carbonate crust
(Unit II) underneath Unit III collapse boulders; b) Unit III boulder surface with flowstone encrustation; c) Landsnail (Achatina spp.), bone and ash
accumulation marking the earliest unequivocal phase of human occupation: Unit V: ca. 18,500-17,000 cal. BP. Middle: a) Carbonate crust (Unit II)
with Fe/Mn oxide intercalatations and mottling, consistent with waterlogging; b) Concentration of landsnail shell fragments in upper parts of Unit VII;
c) Late(?) Holocene flowstone (Unit VIII) between detrital Unit VII and subrecent floor deposits (IX). This flowstone intercalation can be traced in
the stub stalagmite at the background. Bottom: High organic content in lower Unit VI and neoformed (phosphatic?) nodules reflect bat guano inputs
from an overhanging cupola.

304 Kourampas et al.
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
Fig. 8. AMS radiocarbon (SHCal13 calibration curve; 95.4% probability) and OSL dates from talus
slope sediments near the entrance of Kuumbi Cave (Trench 10). Dates cluster in three time frames:
Late Pleistocene (20,500 to 17,000 cal. BP, with unequivocal occupation deposits between 18,500-
17,000 cal. BP: Unit VT
R10
); mid-Holocene (6200-4800 cal. BP); and late Holocene (2000 to 600 cal.
BP) – both in Unit VII
TR10
. The youngest date cluster (on ceramics as well as on charcoal), records a
phase of renewed late Holocene occupation. Mid-HoIocene charcoal (of uncertain provenance) may
have originated from fires on the overcave surface: evidence for increased fire frequency during a
possible Mid-Holocene arid climatic phase has been inferred from elsewhere in Zanzibar (Unguja
Ukuu: Pungwong et al., 2013b). In the Holocene part of stratigraphy (Unit VII
TR10
), incongruent dates
from the same stratigraphic context and chronological inversions may have resulted from reworking of
older, mid-Holocene charcoal into late Holocene occupation deposits by slope processes, bioturbation
and human activity (for full discussion see Shipton et al. in press).
Very heterogeneous, poorly sorted loams with
diverse human inputs (ash, charcoal, burned bone
and shell, and sparse but diverse material culture,
including lithics, worked bone and, in higher
layers, ceramics) and occasional hearths and other
palaeofloor deposits constitute the upper part of
the entrance talus (Unit VII
TR10
: Fig. 7). Mammalian
remains are abundant, with burned and cutmarked
bones indicating anthropogenic origin of the faunal
assemblage. Taxonomic diversity is high, with various
marine molluscs, fish and (rare) turtle and sea urchin
remains alongside various land mammals, reflecting
the broad foraging spectrum of the cave’s human
occupants. Extinct taxa are present in the lower parts
of the unit but decrease in relative frequency up-
sequence, disappearing entirely in the higher layers.
Fauna still found on Zanzibar (dwarf bovids – suni and
blue duiker, suids, monkeys, hyrax, giant rat) are also
common. Smaller taxa, perhaps unrelated to human
occupation (bushbabies, bats, small rodents, reptiles),
and carnivores (leopard, mongoose, civet) occur at low
frequencies. Indications that carnivores had any role
in the accumulation of the faunal assemblage are,
nonetheless, lacking.
Radiocarbon (on charcoal and bone) and OSL (on
ceramics) dates range from ca. 6000 to 600 BP (Fig. 8),
perhaps indicating a long occupation hiatus between
terminal Pleistocene and late Holocene (although
people may have still inhabited other parts of the
landscape around Kuumbi: Shipton et al., in press).

305Speleogenesis and landscape evolution: Kuumbi Cave, Zanzibar
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
Fig. 9. Kuumbi Cave speleothems. a) Column at the NE entrance, with a shelfstone ledge, encrusted by later dripstone/flowstone,
in the middle. The ledge, which differs from typical phreatic overgrowth speleothems (c.f. Tuccimei et al., 2010), indicates that vadose
speleothem deposition was interrupted by at least one phase of partial submergence of the cave floor; b) Flowstone crust cascade,
dipping towards the cave interior. SW entrance; c) Detail of the eroded surface of the same crust (frame in previous), and stubby
stalagmite; d) Mammillate wall crusts. Alcove off Chamber B. Note postdepositional weathering (below pen); e) Renewed stalactite
growth on weathered and broken, earlier generation stalactite. Alcove off Chamber A; f) Columns and stalagmites in Chamber A (point
C in Fig. 5a). Variation in column thickness along-axis reflects changes in past drip rates; g) Detail of degrading stalagmite (frame in
previous). Weathering and mineral alteration (mineralogy unknown) may be due to reaction of speleothem CaCO
3
with by-products of
bat guano diagenesis; h) Flowstone, possibly of late Holocene age (?), encrusts detrital floor sediment and landsnail shells (by pencil).
Alcove off Chamber A; i) Porous carbonate crust on roof fall block at the SW entrance of the cave (c.f. Fig. 5), deposited by blue-
green algae after ceiling collapse and opening/enlargement of the entrance. This deposit is currently being eroded; j) Stalagmites and
columns in alcove off Chamber A. Erosion has exposed speleothem lamination at the base of the column.
Sediment deposition was complex, with various
colluvia and, in the upper part of the unit, habitation
floor deposits. Reworking of human habitation debris
(some of which may have originated further upslope)
by surface wash and runnels, bioturbation (see
below) and, possibly, human activity, may account for
chronological inversions and incongruent radiocarbon
dates from the same context (Fig. 8).
Inner cave
In Chamber B, by contrast, the basal siltstone and
carbonate crust – here intensely weathered – are
succeeded by about 1.2 to 1.7 m of unconformable
loams and organic/phosphate-rich mud (Unit VI
TR11,12
,
located under a cupola cluster; Fig. 7). Mud deposition
may have taken place in a pool, or on seasonally
flooded floor. Above this come poorly sorted loams,
locally rich in landsnails and vertebrate bones but
devoid of artefacts or other (identifiable) evidence of
human presence.
A flowstone crust, locally linking with small, stubby
stalagmites, and correlative CaCO
3
-cemented loams
(Unit VIII) can be traced across much of the inner
cave floor, on the surface or in shallow subsurface
levels (Figs. 7, 9). This late(?) Holocene flowstone
signals a renewed phase of calcium carbonate
deposition, perhaps as a result of regional climatic
change or change in human use of the overcave
landscape. A shallow cemented layer interbedded
with archaeological deposits near the cave entrance
may be correlative with this crust, but this remains
to be confirmed.

306 Kourampas et al.
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
probably excavated by small to medium-sized
vertebrates (Fig. 7), to numerous tree root channels
and terrestrial arthropod galleries (from termites to
crabs). Depositional boundaries are, nonetheless,
traceable across the exposed profiles and the cave’s
sediment fill appears to retain its stratigraphic
integrity overall.
LATE QUATERNARY SPELEOGENESIS
AND LANDSCAPE EVOLUTION
Cross-cutting and cut-and-fill relationships
between erosional landforms, speleothems and
detrital sediments permit the establishment of a
provisional morphostratigraphy of Kuumbi Cave: a
succession of – for the most part undated – phases
of wall modification and sediment deposition
(Fig. 10). This morphostratigraphy provides the basis
for a preliminary reconstruction of cave evolution.
Speleogenetic interpretation of Kuumbi Cave is
impeded by a number of lacunae. Our current
chronological framework, based on extrapolation
from (inferred) marine terrace chronologies and
best-fit geological scenarios, is tentative at best.
Exploration and survey of karstic landforms, both
surface and underground, are limited, and hydraulic
and sedimentary links between these landforms are
unexplored. Subsurface geology – especially the lithology
and hydraulic behaviour of rocks underlying the cave-
hosting limestone, is also inferred from extrapolation.
Fig. 10. Morphostratigraphic synthesis of Kuumbi Cave, showing landforms and sediments in inferred temporal order. Italics: landforms/sediments
of uncertain morphostratigraphic position.
Recent deposits (Unit IX: colluvia, winnowed
lags, organic mud in the cave well, and various
kinds of human-deposited debris in a matrix of
floury carbonate(?) dust) reflect the variegated
microenvironments of Kuumbi Cave. Human inputs
within these are reported to result predominantly
from ritual activity.
Postdepositional change
Sediment diagenesis is spatially variable, evidently
reflecting local (cm- to m-scale) hydrological and
geochemical conditions across the variegated cave floor.
Carbonate cementation is widespread in sediments
deposited in inner Chamber B (trenches KC11, KC12)
and much less pronounced in the near-entrance talus
(trench KC10). Mineral neoformation (with its corollary
of possible shell and bone dissolution and potential
loss of archaeological evidence) appears to be highly
localised. Neomorphs include suspected phosphates/
nitrates(?) in the guano-rich, wet setting of trench
KC12 (Fig. 7), and a few calcitic nodules, resulting from
recrystallisation of ash deposits, in the entrance talus
(trench KC10). Spatially variable diagenesis is common
in cave sediments, especially in caves where bat
guano deposition is highly localised (cf. Weiner et al.,
2002; Shahack-Gross et al., 2004). Higher-resolution
(micromorphological and geochemical) analyses are
expected to refine this preliminary, field-based picture.
Detrital deposits were affected by bioturbation at
various scales, from up to ca. 50 cm unfilled burrows,

307Speleogenesis and landscape evolution: Kuumbi Cave, Zanzibar
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
associated with former sea level stillstands (Mylroie &
Carew, 1990; Mylroie at al., 2001; Mylroie, 2013 and
references therein).
The ‘spongework maze’ plan of Kuumbi Cave,
probably evolved from original porosity that comprised
interconnected pores rather than joints (Klimchouk,
2009), is consistent with the overall architecture
of flank margin caves in other geologically young
carbonate islands (e.g. Guam, the Bahamas, Puerto
Rico, Frank et al., 1998; Mylroie et al., 2001).
Centimetre-scale elliptical pores distributed at distinct
horizontal level(s) in the cave (parallel to, and in part
constitutive of the wall ‘notch/-es’) may exemplify the
dimensions and geometry of early karstic porosity,
before large chambers had developed.
Cupolas, hemispherical holes and other upward-
directed solution landforms, abundant in Kuumbi Cave,
are common in flank margin caves, as much as they are
in hypogene caves formed in confined aquifers (Mylroie
et al., 1995; Osborne, 2004; Klimchouk, 2009; Mylroie
& Mylroie, 2009; Palmer, 2011). In the envisaged coastal
setting of early Kuumbi Cave, tidal pumping may have
been a critical driver of aggressive flows some distance
from the shore: borehole evidence suggests that tidal
brackish groundwater extends for ca. 2-3 km inland
from the Jambiani shore, under the ≤ 12 m terraces
(United Nations, 1987; Bron Sikat, 2011).
Figure 11 summarises our proposed reconstruction,
based on what evidence is available at present. This
reconstruction is in effect a set of hypotheses to be
further refined and tested through geochronology,
fieldwork and sediment analyses. We interpret
Kuumbi and other Jambiani caves as successive
generations of flank margin caves, initially formed at
around groundwater table level, underneath newly
emergent marine terraces. Terrace and cave formation
were thus largely concurrent, diachronous processes,
driven by the interplay between glacio-eustatic sea
level change and crustal uplift of Zanzibar, and
extending over several sea level cycles.
Formation of the 25 m terrace and its underlying
caves: later Mid-Pleistocene (MIS 7 to 6)?
Deposition of the “Older Azania Limestone”
The cave’s host rock, bivalve-gastropod packstone/
grainstone of the Older Azania Limestone, was
deposited in the warm shallow sea that flooded
much of the island during a sea level highstand that
predated the Last Interglacial. Its depositional setting
was probably that of a relatively low-energy back-reef,
similar to the shallow lagoons and platforms that
fringe present-day Zanzibar.
As discussed, the age of the Older Azania Limestone
is unknown. Notwithstanding this uncertainty, and in
analogy with inferences made for other East African
Pleistocene limestones at similar altitudes (see above),
we propose, as a working hypothesis, the correlation
of the Older Azania Limestone with the penultimate
interglacial, MIS 7. During the three ‘optimal’
highstands of MIS 7 (MIS 7e, 7c, 7a; the former lasting
for ca. 5,000 years; the latter two for ca. 8,000 years
each), eustatic sea level is estimated to have ranged
from -15 and to -5 m below present (Siddall et al.,
These limitations notwithstanding, genetic
interpretation of Kuumbi Cave (and its wider landscape)
is constrained by the following considerations:
• A late Quaternary time frame. As discussed, a
tenuous chronology for the formation of the
overcave terrace is the later part of the Mid-
Pleistocene (tentatively MIS 7: sometime between
240,000 and 190,000 BP).
• Scarcity of surface drainage and stream incision
across the Jambiani terraces. This probably
suggests that much of the rainwater falling on
the surface of this – presumably very porous –
sediment recharges the carbonate aquifer at the
expense of surface runoff. Development of kastic,
alongside primary, porosity to take up this
recharge must have occurred relatively rapidly,
early post-emergence: rapidly enough to prevent
the incision of surface drainage.
• Multi-storey, step-like regional karst, that overall
‘mirrors’ the terraced surface morphology. Each of
the two principal terrace levels (25 and <12 m) is
underlain by caves.
• Underground karst is not intercepted by surface
erosional landforms. None of the (few) visited
caves are cut by terrace cliffs or escarpments.
Kuumbi and other caves open to the surface via
collapse openings and (fewer) solution pipes.
• Abundance of cupolas, cusps and other phreatic
dissolution features.
• Well developed speleothems, the bulk of which
seems to predate most of the detrital sediment fill.
Flank margin speleogenesis
The congruence between surface terraces and under-
terrace cave levels at Jambiani probably reflects an
intimate, syngenetic relationship between sea level
stands and littoral speleogenesis. Such a link is
becoming increasingly recognised in carbonate islands
and coastal regions throughout the world. The model
of ‘flank margin speleogenesis’, originally proposed
to interpret cave formation in relatively young (Late
Cainozoic), eogenetic limestones, and its subsequent
evolution into the more generic and nuanced Carbonate
Island Karst model emphasise the critical speleogenetic
role of the freshwater lens – a feature highly responsive
to sea level change (Mylroie & Carew, 1990; Frank et
al., 1998; Mylroie et al., 2001; Fratesi, 2013; Mylroie,
2013). Various processes operating at the seaward
(halocline: fresh and seawater mixing zone) and upper
(groundwater table) boundaries of this lens promote
undersaturation of circulating aqueous solutions
in CaCO
3
, limestone dissolution and the removal of
dissolution products. These processes include mixing
corrosion (c.f. Bögli, 1980), convection generated by
temperature, salinity and (other) density differences
and, also, by the pumping effect of tidal flows (potentially
a very significant process in Zanzibar’s macrotidal
coasts), and accumulation of organic debris around
groundwater table levels, where percolating meteoric
fluids meet slow moving, phreatic water. Accelerated
dissolution at the (sea level-controlled) distal margin
of the freshwater lens results in formation of flank
margin caves of characteristic morphology, at levels

308 Kourampas et al.
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
Fig. 11. Inferred late Quaternary evolution of Kuumbi Cave and its wider landscape (see text).

309Speleogenesis and landscape evolution: Kuumbi Cave, Zanzibar
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
from later reflooding or freshwater pooling in the cave:
current evidence is too sketchy to permit inferences to
be drawn with any confidence).
Draining of the cave
Eustatic sea level fell dramatically on the wane of
the last MIS 7 highstand (MIS 7a). In terminal MIS 7
and MIS 6 (ca. 180,000-135,000 BP), it ranged from
-60 to -120 m below present (Siddall et al., 2006; Fig.
11c). This (composite) sea level lowstand resulted in
prolonged emergence of the Zanzibar shelf, seaward
migration of the littoral (perhaps near the shelf-break
off the eastern Zanzibar coast), and corresponding
drop of the groundwater table.
Groundwater table drop probably drained Kuumbi
(and other caves under the 25 m terrace) for several
tens of thousands(?) of years. Water pooling on the
cave floor, if present, was probably short-lived, during
periods of high rainfall.
Pedogenesis on the overcave terrace, now a low relief
surface some distance from the shore, resulted in the
development of regolith and soil. Vestiges of cemented
regolith on the 25 m terrace may thus date from this
period (or from later Pleistocene times). Fines infiltrated
through this inferred soil mantle may have fed some
of the earliest(?) detrital deposits in Kuumbi Cave,
Unit I red siltstone (Fig. 7). Also, some of the earliest
subaerial speleothems in the cave may have formed
during this period of prolonged vadose conditions.
Flank margin speleogenesis may have progressed in
the seaward parts of the exposed limestone platform
during this prolonged sea level lowstand. The form
and extent of caves dating to this period is unknown,
since such caves, if extant, are now undersea.
Sea level drop must have induced a steep topographic
gradient in eastern Zanzibar. The absence of canyons
incised through the 25 m terrace possibly suggests
that drainage of (the inner parts of) the exposed shelf
was taken up by already well developed underground
cavities, but direct evidence for turbulent stream flow
(e.g. wall scalloping) is absent. Near-surface cavities
may have linked up with even deeper karst though
a process of karstic undercutting, but evidence for
this is currently lacking, as the lower reaches of the
Jambiani karst are inaccessible or unexplored.
Formation of the ≤12 m terrace(s) and caves: Late
Pleistocene (MIS 5)?
‘Younger Azania Limestone’ and terraces
Eustatic sea level rose again and the warm shallow
sea transgressed the eastern Zanzibar shelf. The
shallow-marine carbonates of the ‘Younger Azania
Limestone’ (coral bioherms and coral-algal rudstones
on and around reefs, grainstones in beaches and
backreef shoals), were deposited as a result. In the
stratigraphic scheme proposed here, Younger Azania
Limestone deposition is attributed to one (or more?)
highstand(s) of the composite MIS 5 (ca. 130-70,000
BP) – perhaps around MIS 5e (ca. 128-115,000 BP),
when sea level was between 0 and + 6 m (Siddal et
al., 2006). Several closely spaced, mainly(?) erosive
marine terraces from +12 m to present sea level may
have resulted from coastal erosion during shorter-
2006 and references therein). MIS 7 thus afforded
about 21,000 years of relatively elevated sea level.
Erosion of the terrace platform and deposition of one
(or more) carbonate unit(s) took place during one (or
more) of these highstands.
Formation of Kuumbi’s proto-caverns
Late Quaternary highstands included long periods
during which sea level was high, yet lower than the
maximum levels attained during each highstand (Fig.
11b). If the correlation of the Old Azania Limestone
with MIS 7 is valid, ‘suboptimal’ highstand conditions,
during which sea level was between -30 and -20 m
below present, extended for ca. 25,000 years. Moderate
drop of eustatic sea level (and, perhaps, the filling of
accommodation space by the aggrading/prograding
carbonate platform), resulted in repeated(?) emergence
and subaerial exposure of parts of the carbonate
platform and submersion of the limestone aquifer
in a freshwater/brackish water lens. The onset of
flank margin speleogenesis, therefore, may have been
broadly contemporaneous with (phases of) limestone
deposition and terrace formation. Even if the tenuous
correlation of the Older Azania Limestone with MIS
7 proves invalid, this reasoning may be applicable to
other Pleistocene highstands.
Pulses of crustal uplift may have also induced or
accentuated relative sea level drop during highstands.
Our current understanding of late Quaternary
tectonics of Zanzibar, however, is too sketchy to permit
evocation of tectonic factors at these timescales.
During such ‘suboptimal’ highstands, groundwater
that permeated the porous, newly emergent
limestone (probably largely uncemented lime sand/
ooze immediately post-emergence) circulated mainly
by diffuse, ‘Darcynian’ flow in an initial network of
interparticle and incipient dissolution pores (unlike
the fissures and conduits of cemented, telogenetic
limestones: c.f. Mylroie & Carew, 1990; Vacher &
Mylroie, 2002). These pores provided early foci for
karstic dissolution: progressively enlarging porosity
from which larger cavities were to evolve. Much of this
solutional porosity may have emerged fairly rapidly,
from preferential dissolution of aragonitic skeletal
grains (calcareous green algae plates, whole mollusc
shells, scleractinean corals, etc.). Aragonitic grains,
unstable in meteoric and mixed water diagenetic
environments (Choquette & Pray, 1970; Longman,
1980), probably abounded in Older Azania Limestone.
Mollusc-mouldic pores, like those ubiquitous on the
present-day cave walls (Fig. 3d), may thus provide an
analogue for incipient karstic porosity.
Incipient flank margin caves, produced by such
(largely) even, ‘uncompetitive’ cavity enlargement,
were aligned with the – then – watertable level. The
latter is currently poorly constrained, but appears to
have stood only a few metres below the 25 m terrace.
Higher-order fluctuations of the groundwater table
during the formative highstand (cf. Esat et al., 1999)
may have resulted in repeated draining and flooding
of proto-Kuumbi Cave (and other flank margin caves).
Notches and subaqueous speleothems in Kuumbi
Cave may, therefore, date from these early stages (or

310 Kourampas et al.
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
lived, intra-MIS 5 highstands (Lambeck & Chappell,
2001; Siddall et al., 2006; Fig. 2).
Formation of Kikuaju and neighbouring caves
Large parts of the Last Interglacial carbonate shelf
became emergent repeatedly during suboptimal intra-
MIS 5 highstands. Recapitulation of the diagenetic
and cave-forming processes that had previously
affected the ‘Older Azania Limestone’ resulted in a
younger generation of – apparently smaller – flank
margin caves under the ≤12 m terraces. Kikuaju A
and B (and, perhaps, other caves under the Jambiani
coastal plain) probably formed in this period (Fig. 11e).
Last Interglacial speleothems?
Kuumbi Cave probably remained emergent, or only
partly submerged (in optimal highstands, and/or due
to water pooling during most of the Last Interglacial.
Regional climate was probably warm and moist, with
at least ca. 10,000 years of monsoon (Yuan et al.,
2004; Chiang, 2009). We hypothesise that much of
the speleothem deposition in Kuumbi Cave (Units
St, II) dates from the Last Interglacial. High rainfall
and prolific vegetation growth and pedogenesis on
the overcave terrace may have enhanced limestone
dissolution in the epikarst and deposition of vadose
speleothems in the cave. Possible subaqueous
speleothems (mammillates, shelfstones) may also
date from Last Integlacial flooding of the cave floor,
but their relative dating is uncertain.
Vadose to open cave: Late Pleistocene (MIS 4 to 2)
Drained caves and terminal Pleistocene speleothems
Base level fell on the wane of the Last Interglacial,
from -60 m in MIS 3 times to -120 m in the LGM (Siddall
et al., 2006). This prolonged Late Pleistocene lowstand
resulted in groundwater table drop and emergence
of under-terrace caves, as coastal speleogenetic
environments migrated seawards (perhaps to the
shelf break/shelf slope east of the Jambiani coast).
It is possible that underground karst formed during
this prolonged but multi-phase lowstand, but its
configuration and hydraulic connectedness with
highstand caves, such as Kuumbi and Kikuaju, are
unknown. Flank margin caves resulting from lower-
than-present sea level highstands during MIS 3
(-50 to -60 m) may lie submerged offshore; however,
their presence and distribution are unconfirmed.
Subaerial speleothems were probably deposited in
both Kuumbi and Kikuaju caves in this later part of
the Pleistocene. Much of the speleothem deposition
may have occurred during warm and moist phases
(e.g. MIS 3), and, also, in terminal Pleistocene-
beginning of the Holocene, before rising sea level
partially submerged caves under the ≤12 m terrace(s)
(e.g. Kikuaju cave ‘springs’) in brackish water.
Opening of Kuumbi Cave
Drained of groundwater that had provided
hydrostatic support, and undermined by earlier
dissolution, Kuumbi Cave was structurally unstable
and prone to collapse (cf. Gillieson, 1996; Osborne,
2002). Collapse events may have occurred several
times in the history of Kuumbi and other caves in the
region (as manifest by the numerous collapse dolines
that dot the Jambiani terraces).
The earliest documented collapse event in Kuumbi
Cave (manifest by the Unit III
TR10
collapse boulders)
probably dates to the Late Pleistocene, before ca.
25,000 cal. BP (the earliest radiocarbon dates,
from landsnail in interboulder colluvial fill space:
c.f. Sinclair et al., 2006). This event transformed a
substantial part the cave into a collapse doline.
Collapse blocks remained exposed to the elements
for a long period before their burial by red colluvium
(Unit IV
TR10
), as episodic sheetwash, debris flows and
small runnels scoured the slope and overcave surface.
Much of this colluvium appears to have been derived
from erosion of a red soil. Soil erosion may have
been enhanced by low/sparse vegetation – perhaps
shrubland or open woodland. These entrance colluvia
possibly reflect arid climatic conditions, consistent
with those prevailing over much of equatorial East
Africa during MIS 2, when monsoonal circulation was
severely weakened (Yuan et al., 2004; Kiage & Liu,
2006; Barker, 2007; Chiang, 2009).
Opening of the cave amounted to a radical
refashioning of its physical environment (temperature,
P
CO2
, air circulation, humidity) and ecology. These
changes must have impacted on carbonate equilibria
dynamics in recharging solutions, and thus on the
rate of speleothem deposition. Air circulation through
the enlarged entrance(s) also caused localised erosion
of speleothems at this and later times. From this
point onwards, Kuumbi Cave has been accessible to
large plants (e.g. tree roots) and animals, including
humans. Significantly, the earliest shells of Achatina
landsnails occur in interstices between Unit III
TR10
collapse boulders.
Kuumbi’s first humans
It is in this entrance talus that the earliest definitive
evidence for human presence in and around Kuumbi
Cave was discovered by earlier researchers (Sinclair et al.,
2006; Chami, 2008) and confirmed by our excavations:
a dense accumulation of Achatina landsnails comingled
with marine molluscs, mammalian remains, charcoal
and lithics, dated to ca. 18,500-17,000 cal. BP. Scarce
charcoal and marine molluscs in underlying sediments
raise the possibility of even earlier sporadic human
presence. The terminal Pleistocene foragers who
inhabited Kuumbi Cave preyed upon a diverse array of
mammals, including zebra, buffalo and several other
taxa currently extinct in Zanzibar.
Abandonment, re-occupation and landscape
change: Holocene (MIS 1)
Colluvial deposition continued on the entrance
talus and elsewhere in the cave in Holocene times.
A paucity of radiocarbon dates between the terminal
Pleistocene and late Holocene may indicate that the
cave (and its surroundings?) became depopulated in
early Holocene times. Holocene sea level rise resulted
in rapid transformation of Zanzibar into an island
ca. 11,000-10,000 cal. BP (Ruby, pers. comm. 2015).
Links between the rapidly changing island landscape

311Speleogenesis and landscape evolution: Kuumbi Cave, Zanzibar
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
interpreted as broadly contemporaneous (i.e. within
the same Milankovitch-scale glacioeustatic cycle),
shaped by the interplay between glacioeustatic sea
level change, karstic dissolution around highstand
groundwater tables, and ceiling collapse, in a coast
undergoing crustal uplift, perhaps at the order of ca.
0.10-0.20 mm/yr.
Kuumbi Cave formed as a flank margin cave during
‘suboptimal’ highstands, by dissolution around
the seaward parts of the freshwater lens when sea
level/groundwater table was a few metres under the
emergent carbonate shelf (25 m terrace). Phreatic
to epiphreatic speleogenesis in these early phases
resulted in the formation of a spongework maze, with
cusps and upward-directed solution cavities (cupolas).
Early-stage speleogenesis progressed fast enough for
nascent underground porosity to take up much of the
drainage of the overcave terrace. Rapid dissolution
may have been favoured by high primary porosity and
the unstable, aragonitic composition of skeletal grains
in the limestone aquifer. A proposed period of early
speleogenesis is the Penultimate Interglacial (MIS 7).
Lowering of the groundwater table in sea level
lowstands (intra-MIS 7 lowstands?; MIS 6?) resulted
in repeated draining of Kuumbi Cave. The earliest
detrital sediments and carbonate speleothems may
date from these times (especially from MIS 6), but
geochronological confirmation is wanting. Possible
subaqueous speleothems record episodes of cave floor
flooding, perhaps triggered by high-order sea level
cyclicity and/or water pooling.
Marine transgression of the Zanzibar littoral in the
Last Interglacial resulted in limestone deposition
(probably in MIS 5e) and terrace formation (≤12 m
terraces). Flank margin caves formed under these
terraces during suboptimal, intra-MIS 5 highstands.
These caves were drained and transformed to vadose
chambers in the course of Late Pleistocene (MIS 4
to 2) sea level/groundwater table fall. High rainfall,
lush vegetation and well developed soils in the warm
and humid Last Interglacial may have promoted
deposition of vadose speleothems in caves under the
higher, 25 m terrace. We hypothesise that many of
Kuumbi Cave’s speleothems date from this period.
Late Pleistocene (but pre-ca. 22,000 cal. BP)
collapse of part of the Kuumbi Cave ceiling, already
undermined by earlier phreatic dissolution, amounted
to major reorganisation of the cave environment.
From this point onwards, Kuumbi Cave received large
quantities of detrital sediment from the overcave
surface and was accessible to large surface biota.
Near-entrance sediments record the evolution of an
entrance talus from this terminal Pleistocene collapse
to the present. Colluvia burying collapse boulders
resulted from erosion of a seemingly drier, more
sparsely vegetated terminal Pleistocene (ca. 22,000
cal. BP) landscape. Later talus sediments include
terminal Pleistocene (18,500-17,000 cal. BP) deposits
of Achatina spp. landsnails and other fauna, associated
with human occupation and, after a prolonged
habitation hiatus, later Holocene colluvia and floor
deposits with diverse human inputs: terrestrial
snails and marine mollsucs, vertebrate bones, ash
and the demographic and ecological responses
of Zanzibar’s human and (other) mammalian
inhabitants are being investigated through ongoing
faunal analyses, palaeogeographical modelling and
environmental reconstruction.
Later Holocene near-entrance sediments at Kuumbi
Cave largely consist of human-deposited debris with
terrestrial and marine animal remains, LSA lithics
and, in the upper parts, Iron Age pottery (Unit VII
TR10
),
indicating intensification of human habitation in
and around the cave. Cave users practiced a broad
spectrum foraging economy that showed considerable
continuity over time (Shipton et al., in press).
Ubiquitous marine shell and small amounts of fish
bone demonstrate that Kuumbi Cave was part of an
inhabited landscape that extended to the ocean shore
and the shallow shelf beyond.
The presence of a shallow, presumably Holocene,
carbonate crust and correlative CaCO
3
-cemented
floor deposits in the inner cave floor manifests a
shift in cave depositional dynamics: accelerated
rates of carbonate deposition may reflect changes
in the hydrologic regime and/or sediment dynamics
in the cave catchment, the 25 m terrace. These
changes may have been driven by Holocene climate
change, changing human practices of land use, or the
combined effects of both.
Ongoing reorganisation of the cave’s environment,
triggered by further, localised ceiling collapse,
probably resulted in an overall drier cave atmosphere,
and, also, in insolation of a large part of the cave.
Photosynthetic algal communities (that cover much
the present cave walls) may have been a major agent
of bioerosion, contributing copious quantities of fine
carbonate matrix (cf. Northup & Lavole, 2001).
The last few centuries
Human presence in Kuumbi Cave appears to have
been sporadic in these later times, as indicated by
the lower frequency of material culture and other
human-induced deposits in late Holocene Unit IX.
Oral tradition, as recorded by Chami (2008), hints at
a lull in human use and, even, knowledge of Kuumbi
Cave: reportedly, the cave was rediscovered by a
couple searching for a place ‘to consummate their
relationship’. This tradition also records the presence
of a water pool in the rear part of the cave, presumably
caused by a somewhat elevated watertable.
Human visitation in the very recent past may have
been mainly for drawing water from the cave well,
conducting spirit worship rituals (c.f. Chami, 2008),
and, in the last decade, tourism.
CONCLUSIONS
Kuumbi Cave, on the eastern Zanzibar littoral,
is one of several caves under a flight of Pleistocene
marine terraces. Terraces and caves formed in two
porous limestone units, an older one, tenuously
correlated with the Penultimate Interglacial highstand
(MIS 7) – but which could also be much older, and
a younger one, correlated with the Last Interglacial
(MIS 5). Speleogenesis and terrace formation are

312 Kourampas et al.
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
Barker P.A., 2007 – The Monsoon’s past. Science,
316: 1295-1296.
http://dx.doi.org/10.1126/science.1143451
Battistini R., 1977 – Ages absolus Th230/Ur234 de
depots marins Pleistocenes à Madagascar et dans
les fles voisines. Madagascar Revue de Geographie,
31: 73-86.
Birmingham A.B., Mylroie J.R., Mylroie J.E. & Lace M.J.,
2011 – Bell hole origin: constraints on developmental
mechanisms, Crooked Island, Bahamas. Speleogenesis
and Evolution of Karst Aquifers, 10: 70-79.
Bögli A., 1980 – Karst hydrology and physical speleology.
Springer, Berlin, 284 p.
http://dx.doi.org/10.1007/978-3-642-67669-7
Boivin N., Fuller D.Q. & Crowther A., 2012 – Old world
globalization and the Columbian exchange: Comparison
and contrast. World Archaeology, 44: 452-469.
http://dx.doi.org/10.1080/00438243.2012.729404
Boivin N., Crowther A., Helm R. & Fuller D., 2013 – East
Africa and Madagascar in the Indian Ocean World.
Journal of World Prehistory, 26: 213-281.
http://dx.doi.org/10.1007/s10963-013-9067-4
Boivin N., Crowther A., Prendergast M. & Fuller D.Q.,
2014 – Indian Ocean food globalisation and Africa.
African Archaeological Review, 31: 547-581.
http://dx.doi.org/10.1007/s10437-014-9173-4
Bover P., Valenzuela A., Guerra C., Rofes J., Alcover J.A.,
Ginés J., Fornés J.J., Cuenca-Bescós G. & Merino A.,
2014 – The Cova des Pas de Vallgornera (Llucmajor,
Mallorca): a singular deposit bearing an exceptional
well preserved Early Pleistocene vertebrate fauna.
International Journal of Speleology, 43: 175-192.
http://dx.doi.org/10.5038/1827-806X.43.2.6
Braithwaite C.J.R., 1984 – Depositional history of the late
Pleistocene limestones of the Kenya coast. Journal of
the Geological Society, 141: 685-699.
http://dx.doi.org/10.1144/gsjgs.141.4.0685
Bron Sikat L., 2011 – Assessing the spatial and temporal
characteristics of groundwater recharge in Zanzibar:
towards the optimal management of groundwater
resources. Unpublished MSc thesis, University of
Twente, 43 p.
Caswell P.V., 1956 – Geology of the Kilifi-Mazeras area.
Geological Survey of Kenya, Nairobi, 54 p.
Chami F.A., 2008 – Diffusion in the studies of the African
past: reflections from new archaeological findings.
African Archaeological Review, 24: 1-14.
http://dx.doi.org/10.1007/s10437-007-9012-y
Chami F.A., 2009 – Excavation of Kuumbi Cave, Zanzibar.
In: Chami F.A. (Ed.), Zanzibar and the Swahili Coast
from c.30,000 years ago. E&D Vision Publishing, Dar
es Salaam, p. 41-78.
Chiang J.C.H., 2009 – The tropics in paleoclimate. Annual
Reviews in Earth and Planetary Sciences, 37: 263-297.
http://dx.doi.org/10.1146/annurev.earth.031208.100217
Choquette P.W. & Pray L.C., 1970 – Geologic nomenclature
and classification of porosity in sedimentary carbonates.
American Association of Petroleum Geologists Bulletin,
54: 207-250.
Chorowicz J., 2005 – The East African rift system. Journal
of African Earth Sciences, 43: 379-410.
http://dx.doi.org/10.1016/j.jafrearsci.2005.07.019
Crowther A., Horton M., Kotarba-Morley A., Prendergast
M., Quintana Morales E., Wood M., Shipton C., Fuller
D.Q., Tibesasa R., Mills W. & Boivin N., 2014a – Iron Age
agriculture, fishing and trade in the Mafia archipelago,
Tanzania: new evidence from Ukunju Cave. Azania,
49: 21-44.
http://dx.doi.org/10.1080/0067270X.2013.878104
and charcoal, lithics, and, in shallower strata, Iron
Age ceramics. Erosional surfaces and palaeofloors
suggest that sediment deposition was discontinuous,
with pulses of sediment flux, separated by periods of
little deposition and/or erosion.
Inner cave deposits include the filling of floor
depressions by silt and bat guano and later floor wash
deposits with no apparent human inputs.
A shallow flowstone crust and correlative CaCO
3
-
cemented floor, may record a late(?) Holocene shift
in cave depositional dynamics. This may have been
caused by regional climate change (increase in rainfall
rates), and/or human-mediated vegetation change on
the overcave terrace.
In recent times Kuumbi Cave has been a ritual
space, seasonal water resource, marker of Swahili
heritage and identity, and a tourist destination of
growing importance. The cave and its wider landscape
are highly significant – and contested – places for
members of the Jambiani community. Future research
on these landforms should engage actively with their
local users and custodians.
ACKNOWLEDGEMENTS
The Sealinks Project is funded by a European Research
Council (ERC) grant to N. Boivin (Starter Grant 206148,
‘SEALINKS), under the ‘Ideas’ specific Programme of
the 7th Framework Programme (FP7). Ceri Shipton and
Alison Crowther were funded by postdoctoral fellowships
from the University of Queensland and the British
Academy, respectively. Llorenç Picornell was funded by
a postdoctoral fellowship by the Conselleria d’Educació,
Cultura i Universitats (Government of the Balearic
Islands) and the European Social Fund. Fieldwork at
Kuumbi Cave was carried out under a research permit
issued by the Office of Chief Government Statistician,
Zanzibar Research Committee, and an excavation
license issued by the Zanzibar Department of Museums
and Antiquities. We gratefully acknowledge the support
of the Zanzibar Department of Antiquities, particularly
Amina Issa and Abdallah Khamis. The work was also
undertaken with the permission and collaboration of
the cave’s local custodians, particularly: Bandari Ame
Haji, Asha Ali Makame, Meja Haji Nyonje, Nyonje Pandu
Nyonje and Hassan Ali. We are indebted to Joaquin
Ginés, Bogdan P. Onac and two anonymous reviewers
for their thorough and constructive reviews that greatly
improved earlier drafts of this paper.
REFERENCES
Abuodha J.O.Z., 2004 – Geomorphological evolution of
the southern coastal zone of Kenya. Journal of African
Earth Sciences, 39: 517-525.
http://dx.doi.org/10.1016/j.jafrearsci.2004.07.012
Arthurton R., 2003 – The fringing reef coasts of Eastern Africa
– Present processes in their long-term context. Western
Indian Ocean Journal of Marine Science, 2: 1-13.
Arthurton R.S., Brampton A.H., Kaaya C.Z. & Mohamed
S.K., 1999 – Late Quaternary coastal stratigraphy on
a platform-fringed tropical coast: a case study from
Zanzibar, Tanzania. Journal of Coastal Research,
15: 635-644.

313Speleogenesis and landscape evolution: Kuumbi Cave, Zanzibar
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
Crowther A., Veall M-A., Boivin N., Horton M., Kotarba-
Morley A., Fuller D.Q., Fenn T., Haji O. & Matheson
C.D., 2014b. – Use of Zanzibar copal (Hymenaea
verrucosa Gaertn.) as incense at Unguja Ukuu, Tanzania
in the 7–8
th
century CE: chemical insights into trade and
Indian Ocean interactions. Journal of Archaeological
Science, 53: 374–390.
http://dx.doi.org/10.1016/j.jas.2014.10.008
De Waele J., 2009 – Evaluating human disturbance on
Mediterranean karst areas: the example of Sardinia,
Italy. Environmental Geology, 58: 239-255.
http://dx.doi.org/10.1007/s00254-008-1600-x
Esat T.M., McCulloch M.T., Chappell J., Pillans B.
& Omura A., 1999 – Rapid fluctuations in sea level
recorded at Huon Peninsula during the penultimate
deglaciation. Science, 283: 197-201.
http://dx.doi.org/10.1126/science.283.5399.197
Farrant A.R. & Smart P.L., 2011 – Role of sediment
in speleogenesis; sedimentation and paragenesis.
Geomorphology, 134: 79-93.
http://dx.doi.org/10.1016/j.geomorph.2011.06.006
Fornós J.J., Ginés J., Gràcia F., Merino A., Juncadella
A.M., Gómez-Pujol L. & Bover P., 2014 – Cave
deposits and sedimentary processes in Cova des Pas
de Vallgornera (Mallorca, Western Mediterranean).
International Journal of Speleology, 43: 159-174.
http://dx.doi.org/10.5038/1827-806X.43.2.5
Frank E.F., Mylroie J., Troester J., Calvin Alexander
Jr. E. & Carew J.L., 1998 – Karst development and
speleogenesis, Isla de Mona, Puerto Rico. Journal of
Cave and Karst Studies, 60: 73-83.
Fratesi B., 2013 – Hydrology and geochemistry of the
freshwater lens in coastal karst. In: Lace M.J. & Mylroie
J.E. (Eds.), Coastal karst landforms. Springer, Berlin,
p. 59-75.
http://dx.doi.org/10.1007/978-94-007-5016-6_3
Fuller D.Q. & Boivin N., 2009 – Crops, cattle and
commensals across the Indian Ocean: current and
potential archaeobiological evidence. Études Ocean
Indien, 42/43: 13-46.
Fuller D.Q., Boivin N., Castillo C., Hoogervorst T. &
Allaby R.G., 2014 – The archaeobiology of Indian Ocean
translocations: current outlines of cultural exchanges by
proto-historic seafarers. In: Tripati S. (Ed.), Maritime
contacts of the past: deciphering connections amongst
communities. Delta Book World, New Delhi, p. 1-23.
Fuller D.Q., Boivin N. & Hoogervorst T., 2011 – Across
the Indian Ocean: the prehistoric movement of plants
and animals. Antiquity, 85: 544-558.
http://dx.doi.org/10.1017/S0003598X00067934
Gillieson D., 1996 – Caves: processes, development,
management. Blackwell, Oxford, 324 p.
http://dx.doi.org/10.1002/9781444313680
Ginés A. & Ginés J., 2007 – Eogenetic karst, glacioeustatic
cave pools and anchialine environments on Mallorca
Island: a discussion of coastal speleogenesis.
International Journal of Speleology, 39: 57-67.
http://dx.doi.org/10.5038/1827-806X.36.2.1
Ginés J., Fornós J.J., Ginés A., Merino A. & Gràcia F.,
2014 – Geologic constraints and speleogenesis of Cova
des Pas de Vallgornera, a complex coastal cave from
Mallorca Island (Western Mediterranean). International
Journal of Speleology, 43: 105-124.
http://dx.doi.org/10.5038/1827-806X.43.2.2
Harris E.C., 1989 – Principles of archaeological
stratigraphy. Academic Press, London, 169 p.
Helm R., Crowther A., Shipton C., Tengeza A., Fuller D.Q.
& Boivin N., 2012 – Exploring agriculture, interaction
and trade on the eastern African littoral: preliminary
results from Kenya. Azania, 47: 39-63.
http://dx.doi.org/10.1080/0067270X.2011.647947
Jennings J.M., 1985 – Karst geomorphology. Blackwell,
Oxford, 293 p.
Kent P., Hunt J. & Johnstone D., 1971 – The geology and
geophysics of coastal Tanzania. Institute of Geological
Sciences, London, Special Paper, 101 p.
Kiage L.M. & Liu K., 2006 – Late Quaternary
paleoenvironmental changes in East Africa: a review of
multiproxy evidence from palynology, lake sediments,
and associated records. Progress in Physical Geography,
30: 633-658.
http://dx.doi.org/10.1177/0309133306071146
Klimchouk A., 2009 – Morphogenesis of hypogenic caves.
Geomorphology, 106: 100-117.
http://dx.doi.org/10.1016/j.geomorph.2008.09.013
Lace M.J. & Mylroie J.E., 2013 – The biological and
archaeological significance of coastal cave and karts
features. In: Lace M.J. & Mylroie J.E. (Eds.), Coastal
karst landforms. Springer, Berlin, p. 111-126.
http://dx.doi.org/10.1007/978-94-007-5016-6_5
Lambeck K. & Chappell J., 2001 – Sea level change
through the last glacial cycle. Science, 292: 679-686.
http://dx.doi.org/10.1126/science.1059549
Lismonde B., 2000 – Corrosion des coupoles de plafond
par les fluctuations de pression de l’air emprisonné.
Karstologia, 35: 39-46.
Longman M.W., 1980 – Carbonate diagenetic textures
from nearsurface diagenetic environments. American
Association of Petroleum Geologists Bulletin, 64:
461-487.
Lundberg J. & McFarlane D.A., 2009 – Bats and bell
holes: the microclimatic impact of bat roosting, using
a case study from Runaway Bay Caves, Jamaica.
Geomorphology, 106: 78-85.
http://dx.doi.org/10.1016/j.geomorph.2008.09.022
Miklavic B., Mylroie J.E., Jenson J.W., Randall R.H.,
Banner J.L. & Partin J.W., 2012 – Interglacial limestone
and its geomorphic features on Guam: implications
for relative sea-level change and flank margin cave
formation. In: Gamble D.W. & Kindler P. (Eds.),
Proceedings of the 15
th
Symposium on the Geology of
the Bahamas and other Carbonate Regions. Gerace
Research Center, San Salvador, Bahamas, p. 107-115.
Mpanda S., 1997 – Geological development of the East
African coastal basin of Tanzania. Almqvist & Wiksell
International, Stockholm, 121 p.
Mylroie J.E., 2013 – Coastal karst development in
carbonate rocks. In: Lace M.J. & Mylroie J.E. (Eds.),
Coastal karst landforms. Springer, Berlin, p. 77-109.
http://dx.doi.org/10.1007/978-94-007-5016-6_4
Mylroie J.E. & Carew J.L., 1990 – The flank margin model
for dissolution cave development in carbonate platforms.
Earth Surface Processes and Landforms, 15: 413-424.
http://dx.doi.org/10.1002/esp.3290150505
Mylroie J.E., Carew J.L. & Vacher H.L., 1995 – Karst
development in the Bahamas and Bermuda. In: Curran
H.A. & White B. (Eds.), Terrestrial and shallow marine
geology of the Bahamas and Bermuda. Geological
Society of America, Boulder, CO, Special Paper,
p. 251-267.
http://dx.doi.org/10.1130/0-8137-2300-0.251
Mylroie J.E., Jenson J.W., Taboroshi D., Jocson J.M.U.,
Vann D.T. & Wexel C., 2001 – Karst features of Guam
in terms of a general model for carbonate island karst.
Journal of Cave and Karst Studies, 63: 9-22.
Mylroie J.E. & Mylroie J.R., 2007 – Development of the
Carbonate Island Karst Model. Journal of Cave and
Karst Studies, 69: 59-75.

314 Kourampas et al.
International Journal of Speleology, 44 (3), 293-314. Tampa, FL (USA) September 2015
Mylroie J.E. & Mylroie J.R., 2009 – Diagnostic features
of hypogenic karst: is confined flow nececessary? In:
Stafford K.W., Land L. & Veni G. (Eds.), Advances
in hypogene karst studies. National Cave and Karst
Research Institute Symposium 1, National Cave and
Karst Research Institute, Carlsbad, NM, p. 12-26.
Mylroie J.E., Mylroie J.R. & Nelson C.S., 2008 – Flank
margin cave development in telogenetic limestones of
New Zealand. Acta Carsologica, 37: 15-40.
Mylroie J.E. & Vacher H.L., 1999. – A conceptual view of
carbonate island karst. In: Palmer A.N., Palmer M.V.
& Sasowsky I.D. (Eds.), Karst modeling, Karst Waters
Institute, Charles Town, WV, Special Publication, 48-57.
Nicholas C.J., Pearson P.N., McMillan I.K., Ditchfield
P.W. & Singano J.M., 2007 – Structural evolution of
southern coastal Tanzania since the Jurassic. Journal
of African Earth Sciences, 48: 273–297.
http://dx.doi.org/10.1016/j.jafrearsci.2007.04.003
Nortup D.E. & Lavole K.H., 2001 – Geomicrobiology of
caves: a review. Geomicrobiology Journal, 18: 199-222.
http://dx.doi.org/10.1080/01490450152467750
Oosterom A.P., 1988 – The geomorphology of southern
Kenya. Unpublished Doctoral thesis, Agricultural
University, Wagenigen, 227 p.
Osborne R.A.L., 2002 – Cave breakdown by vadose
weathering. International Journal of Speleology, 31: 37-53.
http://dx.doi.org/10.5038/1827-806X.31.1.3
Osborne R.A.L., 2004 – The trouble with cupolas. Acta
Carsologica, 33/2: 9-36.
Palmer A.N., 2011 – Distinction between epigenic and
hypogenic maze caves. Geomorphology, 134: 9-22.
http://dx.doi.org/10.1016/j.geomorph.2011.03.014
Piccini L., Forti P., Giulivo I. & Meccia M., 2007 – The
polygenetic caves of Cuatro Ciénegas (Coachuila,
Mexico): morphology and speleogenesis. International
Journal of Speleology, 36: 83-92.
http://dx.doi.org/10.5038/1827-806X.36.2.4
Punwong P., Marchant R. & Selby K., 2013a – Holocene
mangrove dynamics in Makoba Bay, Zanzibar.
Palaeogeography, Palaeoclimatology, Palaeoecology,
379/380: 54–67.
http://dx.doi.org/10.1016/j.palaeo.2013.04.004
Punwong P., Marchant R. & Selby K., 2013b – Holocene
mangrove dynamics from Unguja Ukuu, Zanzibar.
Quaternary International, 298: 4-19.
http://dx.doi.org/10.1016/j.quaint.2013.02.007
Reuter M., Piller W.E., Harzhauser M., Berning B. & Kroh
A., 2010 – Sedimentary evolution of a late Pleistocene
wetland indicating extreme coastal uplift in southern
Tanzania. Quaternary Research, 73: 136-142.
http://dx.doi.org/10.1016/j.yqres.2009.09.004
United Nations, 1987 – Hydrogeological map of Zanzibar,
including the islands of Zanzibar and Pemba, The United
Republic of Tanzania, Scale 1:125,000, United Nations,
New York.
Sancho C., Peña J.L., Mikkan R., Osácar C. & Quinif
Y., 2004 – Morphological and speleothemic development
in Brujas Cave (Southern Andean Range, Argentine):
palaeoenvironmental significance. Geomorphology,
57: 367–384.
http://dx.doi.org/10.1016/S0169-555X(03)00166-1
Sasowsky I.D. & Mylroie J.E. (Eds.), 2004 – Studies
of cave sediments: physical and chemical records of
paleoclimate. Springer, Dordrecht, 329 p.
http://dx.doi.org/10.1007/978-1-4419-9118-8
Shahack-Gross R., Berna F., Karkanas P. & Weiner S.,
2004 – Bat guano and preservation of archaeological
remains in cave sites. Journal of Archaeological
Science, 31: 1259-1272.
http://dx.doi.org/10.1016/j.jas.2004.02.004
Shipton C., Crowther A., Prendergast M.E., Kourampas
N., Horton M., Douka K., Schwenninger J-L., Faulkner
P., Quintana-Morales E.M., Langley M., Tibesasa R.,
Picornell-Gelabert L., Doherty C., Wilmsen E.,
Petraglia
M. & Boivin N., (in press, 2015) – Reinvestigation of
Kuumbi Cave, Zanzibar, reveals stratigraphic complexity
and discontinuous human occupation spanning the Late
Pleistocene to Late Holocene. Azania.
Shipton C., Helm R., Boivin N., Crowther A., Austin P.
& Fuller D., 2013 – Intersections, networks and the
genesis of social complexity on the East African coast.
African Archaeological Review, 30: 427-453.
http://dx.doi.org/10.1007/s10437-013-9140-5
Siddall M., Chappell J. & Potter E-K., 2006 – Eustatic sea
levels during past interglacials. In: Sirocko F., Claussen
M., Litt T. & Sanchez-Goni M.F. (Eds.), The climate of
past interglacials. Elsevier, Amsterdam, p. 75 - 92.
Sinclair P., Juma A. & Chami F., 2006 – Excavations
in Kuumbi Cave in Zanzibar in 2005. Studies in the
African Past, 5: 95-106.
Stockley G.M., 1928 – Report on the geology of the Zanzibar
Protectorate. Government of Zanzibar, Zanzibar, 126 p.
Sweeting M.M., 1972 – Karst landforms. Macmillan,
London, 362 p.
Trudgill S.T., 1985 – Limestone geomorphology. Longman,
London, 196 p.
Tuccimei P., Soligo M., Ginés J., Ginés A., Fornós J.J.,
Kramers J. & Villa I.M., 2010 – Constraining Holocene
sea levels using U-Th ages of phreatic overgrowths on
speleothems from coastal caves in Mallorca (Western
Mediterranean). Earth Surface Processes and
Landforms, 35: 782-790.
http://dx.doi.org/10.1002/esp.1955
Vacher H.L. & Mylroie J.E., 2002 – Eogenetic karst
from the perspective of an equivalent porous medium.
Carbonates and Evaporites, 17: 182-196.
http://dx.doi.org/10.1007/BF03176484
Van Beynen P., Bourbonniere R., Ford D.C. & Schwarcz
H., 2001 – Causes of colour and fluorescence in
speleothems. Chemical Geology, 175: 319-341.
http://dx.doi.org/10.1016/S0009-2541(00)00343-0
Veldkamp A, Buisa E., Wijbrans J.R., Olagoc D.O.,
Boshovena E.H., Marée M. & van den Berg van
Saparoea R.M., 2007 – Late Cenozoic fluvial dynamics
of the River Tana, Kenya, an uplift dominated record.
Quaternary Science Reviews, 26: 2897–2912.
http://dx.doi.org/10.1016/j.quascirev.2007.06.033
Weiner S., Goldberg P. & Bar-Yosef O., 2002 – Three-
dimensional distribution of minerals in the sediments
of Hayonim Cave, Israel: diagenetic processes and
archaeological implications. Journal of Archaeological
Science, 31: 1289-1308.
http://dx.doi.org/10.1006/jasc.2001.0790
Yuan D., Cheng H., Edwards R.L., Dykoski C.A., Kelly
M.J., Zhang M., Qing J., Lin Y., Wang Y., Wu J.,
Dorale J.A., An Z. & Cai Y., 2004 – Timing, duration,
and transitions of the Last Interglacial Asian Monsoon.
Science, 304: 575-578.
http://dx.doi.org/10.1126/science.1091220