Keywords

4.1 Introduction

Lake Ngami and the Mababe Depression are structural basins that are periodically filled by overspill from the Okavango Delta in northern Botswana. Drainage into these basins, both now and in the Pleistocene, has varied commensurately with flow variations into the Delta, which itself varies with rainfall over the central African interior (Fig. 4.1). The volume of drainage into the Okavango Rift Zone (ORZ) has also been subject to periodic and sequential reduction resulting mainly from tectonic factors related to the East African Rift (EAR) (Modisi et al. 2000). Climatic variability has doubtless also played a key role. Significantly, both the Ngami and Mababe basins occur at opposite ends of the north-west to south-east trending fault system, which controls the distal terminus of the Okavango Delta. The basins themselves form structurally controlled depressions, which are currently partially ringed by former shorelines as a result of occupancy or over-topping by numerous palaeo-lakes (Shaw 1985b). Palaeo-lake chronology is, therefore, key to understanding the geomorphological evolution of the Ngami and Mababe basins. This chapter looks into the geological and hydrological origins of the basins through time by way of explaining the characteristic features of the geomorphological landscape. The terminology of ‘Lake Ngami’ and the ‘Mababe Depression’ is entrenched in the literature and stems from the fact that the Ngami basin has been recognised as long containing lake-water while much of the Mababe basin has remained dry in historic and recent times.

Fig. 4.1
figure 1

Source Okavango Research Institute

Drainage map showing the extent of Zambezi, Cuando-Chobe and Okavango (Cuito, Cubango and Kavango) headstreams over the Angolan plateau. MAP = Mean Annual Precipitation.

Full size image

This geomorphological review of Lake Ngami and the Mababe Depression is tied up with the palaeo-lakes, which have developed in them despite the inherent semi-aridity and overall recurrence of arid events in the region (e.g. Stokes et al. 1997). Palaeo-lake incursions are controlled by the basins’ locations whereby they receive inflow from the endoreic Okavango Delta distributary system and to a lesser extent from the Cuando (aka Kwando) River (Figs. 4.1 and 4.2). The extent and duration of flow through the Okavango system are controlled by climatic factors influencing rainfall over the Angolan and Zambezian Highlands. In terms of overall climate, the entire area lies within the southern African summer rainfall zone where under the influence of the Inter-Tropical Convergence Zone (ITCZ) migration, 80% of precipitation occurs between October and March (McCarthy et al. 2000) and is interspersed by a long dry season. The distance from the upper Cubango River to the terminus of the Okavango Delta extends over 1860 km with a drop in altitude from 1780 to 900 m, providing extremely low gradients (Mendelsohn and El Obeid 2004). These low gradients result in a delay of 5–6 months between rainfall events over the Angolan Plateau and flooding downstream such that the Okavango Delta and the hydrologically linked Lake Ngami and Mababe Depression receive irregular inflow during the dry winter months, notably in June and July. When operative, this causes a greening effect and water availability downstream for numerous cattle and abundant wildlife including fishing, cattle and tourism industries in Lake Ngami (Fig. 4.3a, b). Floodwater inflow is currently less in the Mababe Depression where infrequent partial filling has led to widespread re-vegetation. Nonetheless, the depression is also a magnet for herbivores (Fig. 4.3c) and predators. In terms of relative human impacts, the total population around Lake Ngami is 6589 (Kurugundla et al. 2018) whereas the total population around the Mababe Depression is negligible and confined mainly to Mababe village in the south. Hence, human impacts and resource exploitation around the two basins are variable. More active farming and fishing activities occur around Lake Ngami while tourism and limited hunting are prevalent around the Mababe Depression, which lies mostly within Chobe National Park.

4.2 Geology, Tectonics and Hydrological Change

Much of the earlier geological history of the area continues to be significant as the present-day ORZ is still tectonically active with landscape features resulting from the re-activation of older bedrock trends (e.g. Kinabo et al. 2008). The ORZ overlies much of the Damaran mobile belt, which mainly comprises folded and faulted Neoproterozoic and Mesoproterozoic lithologies on north-east trends (Key and Ayres 2000). These and overlying Karoo sediments and volcanics underlie the elongate basins comprising Lake Ngami and the Mababe Depression. The normal faults defining the elongate basins form sub-graben, which are smaller segments of the larger ORZ structural network. Much of this network was reactivated through earth movements along with a south-west extension of the East African Rift (EAR) (Modisi et al. 2000). Specifically, Lake Ngami lies within a downfaulted sub-graben between a relatively steep horst block to the south (separated by the Kunyere fault) and the less well-developed Lecha fault to the north. Structurally, the Kunyere fault continues along the southern Okavango axis to the north-east (Fig. 4.4). A north–north-east extension referred to as the Mababe fault defines the eastern margin of the Mababe Depression. This normal fault displaced the eastern margin downwards and partially parallels a second displaced fault scarp on the western margin forming the Mababe Depression (Kinabo et al. 2008). The Depression is also bounded to the north by the more frequently active Chobe fault and related horst (Kinabo et al. 2008). The Chobe fault also forms the southern escarpment of the elongate Linyanti swamp-trough, through which a link is provided via the Savuti River between the Mababe Depression to the south and Cuando and Zambezi drainage further north (Fig. 4.4). The land surface of the Linyanti trough tilts towards the northeast, thereby facilitating mainly Cuando flow towards the Chobe River (Fig. 4.5). Major displacements have also taken place along the southern boundary of the Okavango graben, for instance along the Kunyere and Thamalakane/Mababe faults forming in places minor fault-scarps along the southern margin of the Okavango Delta. As indicated above, further reactivation on any of these faults could lead to radical landscape change, which is considered possible as the area is prone to medium–high earthquake activity (Scholz et al. 1976). Movement along the Kunyere, Mababe and Chobe faults (Fig. 4.4), which are associated with the Ngami and Mababe basins, has likely played an important role in influencing both the volume and direction of inflow over time.

The geomorphological development of the Okavango Delta and related basins, therefore, hinges on the volume and duration of drainage via rivers heading south-east off the Angolan-Zambian highlands into the ORZ (Fig. 4.1). Moore et al. (2012) and McCarthy et al. (2013) describe tectonic events leading to the long-term evolution of this drainage network across the Kalahari Plateau (Table 4.1). Drainage evolution occurred since the late Palaeogene/Miocene and has involved major rivers such as the Cuando and Zambezi and their tributaries in mid-stream lake formation and sequential river capture. Over time, these events have caused drainage diversions into and away from the ORZ, instigating and sustaining palaeo-lake development, augmented by climatic feedback mechanisms. The dating of these diversions has been achieved in terms of minimum dates derived from a combination of archaeological and geomorphological evidence and the phylogenetic records of aquatic vertebrates (Moore et al. 2012, their Fig. 7 ABCD). The authors have dated the occurrence of palaeo-lakes in the ORZ basin from Early Stone Age and early Middle Stone Age artefacts as indicated in, for instance, Robbins and Murphy (1998) and McFarlane and Segadika (2001). These results have led to a controversial re-evaluation of the antiquity of the major palaeo-lakes (see below). The re-evaluated early inundation events are directly linked to upstream drainage diversions involving upstream lake formation. The downstream consequences in Lake Ngami and the Mababe Depression are typified by sequential palaeo-lakes, which were initially extensive and then progressively smaller in size as discharge from the input rivers and lakes diminished (Moore et al. 2012).

Table 4.1 Drainage evolution and the development of ORZ palaeo-lakes (modified after Moore et al. 2012 and McCarthy 2013 and references therein)
Full size table

As indicated, the development of the early ORZ palaeo-lakes coincides with drainage development in the Zambezi-Cuando-Okavango rivers as these became re-oriented as a result of tectonic events. Uplift and river capture including north-east to south-west faulting initiated basin development along with Lake Ngami and Mababe Depression trends during the Early Pleistocene, alongside the formation of the Okavango graben (Table 4.1). Early river capture diverted water into the southern basins, initially causing extensive flooding in the ORZ referred to as Palaeo-Lake Deception (PLD) (McFarlane and Eckardt 2008). Around this time, the early Okavango River disgorged sediment into the graben, which extended further north-east and south-west than the present delta. This early flow formed extensive fans referred to as the Groote Laagter fan and Linyanti fan (Fig. 4.4). These early fans provided a backdrop into which the faulted Ngami and Mababe basins became nested (McCarthy 2013).

Meanwhile, continuing Early to Mid-Pleistocene drainage diversions involving the Cuando and Zambezi rivers and their tributaries led to continuing but diminished inundation within the ORZ (Moore et al. 2012). This led to the contraction of the earlier PLD to a later, slightly smaller Palaeo-Lake Makgadikgadi (PLM) (Table 4.1 and Fig. 4.4). Traces of PLM shorelines at ca. 945 m are found intermittently along the margins of the Ngami and Mababe basins (cf. Moore et al. 2012, their Fig. 3B). SRTM data imply that at the maximum PLM extent, the Mababe Depression was almost completely submerged at the 945 m level leaving the central Magikwe ridge (ca. 945 m) exposed as a series of islands (Fig. 4.3d and 4). During PLM time, the Lake Ngami area was also practically submerged. The north-west margin of the PLM extended into the present-day Linyanti trough to the north of the Chobe fault (cf. Thomas and Shaw 1991).

Later in the Early-Mid Pleistocene, the link between the Kafue and Upper Zambezi was severed leading to further declining inflow to the ORZ and the contraction of the former PLM to the 936 m level (Table 4.1). The 936 m level is referred as Palaeo-Lake Thamalakane (PLT), which has now been dated using a range of evidence including ESA and early MSA tools to at least 200–500 ka (e.g. Moore et al. 2012; McFarlane and Segadika 2001). This ca. 936 m level is prevalent as a series of dominant palaeo-shorelines, which characterise both the Ngami and Mababe basins. PLT was an extensive, mid-level palaeo-lake that ran north-east to south-west along the low-lying line of the Thamalakane and Kunyere faults and extended southwards into the Makgadikgadi basin (Ringrose et al. 2005). The northern extent of the PLT is unclear but based on mapped contours (in McCarthy 2006), land heights to the west of the Mababe Depression appear too high (940 m) to facilitate a northern extension of the PLT into the Linyanti trough, but this requires further work.

Later in the Mid-Pleistocene, further diversions reduced inflow into the ORZ, which led to the contraction of the PLT to later 920 m and 912 m palaeo-shorelines. This is attributed to the link between the Upper Cuando and Zambezi being severed, leading to the re-connection of the Upper and Mid-Zambezi (Moore et al. 2012). The re-connection of the Upper and Mid-Zambezi had the effect of directing more flow away from the ORZ. Hence as the present-day drainage evolved (Fig. 4.1), a number of extensive palaeo-lakes are believed to have developed in the ORZ mostly during the Early-Mid Pleistocene (Table 4.1). However, other younger dates for ostensibly the same palaeo-lakes (PLM, PLT, etc.), but resulting from more direct climatic and feedback factors, have been indicated in Thomas and Shaw (1991) and later by authors such as Burrough and Thomas (2008). Hence, both perspectives are considered later in this chapter.

4.3 Landforms and Geomorphological Evolution

4.3.1 Lake Ngami

Lake Ngami in 2018–2019 formed a picturesque water-body lying at the heart of fishing and farming communities (Figs. 4.3a, b and 4.6) but was desiccated in 2020. A useable lake was probably intermittently prevalent on the landscape at least since the Lower Stone Age (Robbins et al. 2009) and was first noted and mapped by early European explorers such as Passage and Livingstone in the 1850s (Shaw 1985a). From the various descriptions, the lake level has been calculated at this time as ca. 930 m (Shaw et al. 2003). The nineteenth century lake was characterised by a series of inflow channels including the anastomosing branches of a former Okavango distributary called the Thaoge River (Fig. 4.6), which is thought to have been blocked ca. 1880 as a result of excessive Papyrus growth (Shaw 1985a). Currently, much of the Lake Ngami inflow comes from west and central Okavango distributaries (Fig. 4.2), which are highly variable in terms of their proportional flow volume (Wolski and Murray-Hudson 2008). The western distributaries include the Xudum and Matsibe rivers, which coalesce north of the Kunyere River forming a channel, which flows south-westwards along the fault line towards Lake Ngami. Inflow also originates from the more central Okavango distributaries including the Gomoti, Santantadibe and Boro rivers, which flow south-eastwards until reaching the Thamalakane fault, where they converge into the Thamalakane River (Fig. 4.2). Much of the Thamalakane flow is diverted south-eastwards down through the Boteti River towards the Makgadikgadi Pans (Shaw 1984). The remaining flow continues south-west as the Nchabe (or Lake) River, which continues along with the trend of the Thamalakane fault to Lake Ngami. The Kunyere and Nchabe rivers converge near Toteng and flow into the lake.

Fig. 4.2
figure 2

Source Google Earth, image date 31 December 2014

Relationships between Lake Ngami, the Mababe Depression and inflow rivers derived mainly from the Okavango Delta.

Full size image
Fig. 4.3
figure 3

Photos depicting aspects of Lake Ngami and the Mababe Depression. a Livelihoods around Lake Ngami include cattle raising—note Lake Ngami in background (Photo December 2018, W. Matheson) b Birdlife around Lake Ngami, which attracts tourists to the area (Photo December 2018, W. Matheson) c Savuti marsh in the Mababe Depression is frequently dry and attracts herbivores (Photo January 2017, W. Matheson) d The Magikwe Ridge south of Savuti camp showing exposed aeolian sand towards the ridge surface (Photo January 2017, W. Matheson) e Flooding along the Mababe River and recent road bridge re-construction (Photo December 2018, W. Matheson) f One of the Gubatsa hills that mark the periphery of the Mababe Depression in the vicinity of Sauvti gThe Goha hills occur immediately north of the Mababe Depression h Part of lower Ngami trench showing sandy silt overlain by white, diatomaceous silt at the 350 cm hiatus. Upper bed comprises organic-rich silt, hammer 35 cm (Photo P. Huntsman, 2003) i Part of the upper Ngami trench showing burnt peat over diatomaceous silt. Lower bed comprises silty sand with calcareous nodules, hammer 35 cm (Photo P. Huntsman, 2003)

Full size image
Fig. 4.4
figure 4

SRTM imagery showing key structural features within the Okavango Rift Zone, fans of the early Okavango and inundation extent of palaeo-lake Makgadikgadi (PLM). PLM shown as blue gradations ca. 945–950 m. G = Gumare fault, K = Kunyere fault, T = Thamalakane fault, M = Mababe fault, C = Chobe fault (SRTM via C. Hartnady)

Full size image
Fig. 4.5
figure 5

Source Google Earth image date 31 December 2014

The north-western extent of the Mababe Depression and the Savuti River inflow from the Linyanti swamps. The Linyanti swamps receive inflow from the Cuando river (see Fig. 4.1) but also from the Okavango River via the Selinda spillway. Bold arrow indicates the direction of flow from early Okavango Linyanti fan. Stars depict areas of reworked sediment to the west of the Magikwe ridge.

Full size image
Fig. 4.6
figure 6

Main geomorphological features of the Ngami basin, described in text. (Map modified after Thomas and Shaw 1991)

Full size image

A recent history of lake levels in Lake Ngami and their variability, using a combination of historic records and inflow data from the upper Okavango, has been compiled by Kurugundla et al. (2018). Their work indicates that the lake derives more than 90% of its water from Okavango drainage and the remaining ca. 10% from local precipitation. The floodwaters typically arrive between June and July, peaking in late August. The water then recedes steadily between October and May of the following year. Multiple observations indicate that the behaviour of the lake does not directly reflect the impact of historic climate changes in the local area, but variation in the Okavango inflow, which is dependent on periodic rainfall fluctuations in the headwater catchments in Angola (Kurugundla et al. 2018). After documenting over a century of observations and measurements, the authors concluded that Lake Ngami has mainly functioned since historical times (1850s) as a shallow wet basin with intermediate (and sometimes extensive) dry periods (Table 4.2). Water depths measured in September 2012 showed that the average lake depth was ca. 3.5 m. In September 2012, the volume of water determined for the lake was 974 Mm3 occupying 277.7 km2, with an average water level contour of 923.3 m (cf. 923.5 m, Shaw et al. 2003). At this level, the lake functions as a closed basin, whereas lake levels > 930 m infer a through-flow system to the Boteti River (Shaw 1985b).

Table 4.2 Conditions reflecting lake levels in Lake Ngami between 1850 and 2016 (after Kurugundla et al. 2018)
Full size table

The physical landscape of much of the present-day Ngami basin is determined by the Kunyere fault scarp with its north-west facing down-slip and the termination of former inflow channels of the Thaoge River, giving the basin its west-southwest by east-south east alignment (Fig. 4.6). The outer basin is approximately 70 km long and 38 km wide, representing a surface area of about 2 660 km2, while the flat inner lake bed is 45 km long and 8 km wide (or about 360 km2). Hence ca. 88% of the entire Ngami basin comprises outer stepped palaeo-shorelines and offshore bars, in addition to pans, back swamps and lacustrine foreshore areas. Most geomorphological descriptions emphasise the multiple palaeo-shorelines, first described by Grove (1969) and later by Cooke (1980), Mallick et al. (1981) and Shaw (1985b). Palaeo-shorelines from ca. 928 m to 940 m are also discussed in Thomas and Shaw (1991) and Burrough et al. (2007, 2009b). The predominant palaeo-shorelines comprise the western Magotlawanen ridge (934–936 m) and the Dautsa ridge (928–936 m) complexes, which fringe the inner basin. The Dautsa ridge is incised at the Phatane Gap (Fig. 4.6) and is regarded as an offshore barrier beach, capped by summit dunes formed of reworked aeolian sand (Shaw 1985b). The offshore bars were formed by wind-generated wave action on the leeward side of the palaeo-lakes that once occupied the basin (Grove 1969; Shaw 1988). In addition, towards the western periphery of the basin, the possibly older Kareng ridge stands between 938 and 940 m. A series of former back-swamps such as the Dautsa flats occur between the Kareng and Dautsa ridges (Fig. 4.6). Measured average heights from prominent ridges indicate that the Lake Ngami basin has experienced minor downward tectonic displacement (<5 m) with no tilting since the Mid-Pleistocene (Gumbricht et al. 2001; McCarthy 2013), hence bifurcation on the northern Dautsa ridge may be related to tectonic controls (Huntsman-Mapila et al. 2006).

Despite the ESA-early MSA, Mid-Pleistocene ages (500–200 ka) discussed earlier for the PLT ridges (e.g. Moore et al. 2012) OSL dates from the Kareng, Dautsa and Magotlwanen palaeo-shorelines (936–934 m) suggest that these ridges only date back to ca. 90–100 ka with younger ridges dated at 3–4 ka (Shaw et al. 2003). A perspective that has been endorsed by later MSA dates on lithic tool sites (Nash et al. 2013). The PLT palaeo-shorelines are regarded in Shaw et al. (2003) as being older landforms subsequently modified by Holocene fluctuations in lake level and subsequent aeolian reworking. Holocene reworking included water levels peaking at 936 m between 4 and 3 ka and returning to 934 m at 2.8–2 ka, before falling rapidly to 932 m and later historical levels. According to Shaw et al. (2003), the 936 m level, in particular, is correlated with increased through flow in the Okavango Delta, hence increased rainfall in the catchment. The subsequent decline is attributed to episodic closure and rerouting of Okavango distributaries, in particular the blocking of the Thaoge River.

Later Burrough et al. (2007) obtained 74 OSL dates from Lake Ngami palaeo-shoreline samples from depths of 3 to 6 m. The OSL results are interpreted as being derived from sediments deposited as sequential stacked palaeo-shorelines resulting from ‘multiple lake highstands’ at ca. 935 m over a period from ca. 140 ka to 0.3 ka. Burrough et al. (2007) indicate that the highest lake levels were recorded from the 938 m Kareng ridge (Fig. 4.6), which resulted in palaeo-lake events recurring at 140 ± 11 ka, 124 ± 8 ka, 115 ± 8 ka, 87 ± 7 ka, 59 ± 4 ka, 32 ± 2 ka and 12 ± 1 ka. Other dates were obtained from the lakeward flanks of the Dautsa ridge (936–934 m), which record palaeo-lake deposition at 15.2 ± 1.8 ka, 30 ± 3 ka, 32 ± 3 ka, 33 ± 3 ka, 38 ± 3 ka, 47 ± 4 ka and 56 ± 5 ka. Burrough et al. (2007 and references therein) discuss these highstand features in terms of the regional climatic controls, increased local water availability, the feedback effects of a large body of water on local rainfall and the relationships between dated Holocene wet periods, which show local precipitation increases and flow through the upper Okavango. Burrough et al. (2007) conclude that the Dautsa 934–936 m ridge complex began to be constructed during the Late Pleistocene (ca.18 ka), with further palaeo-shoreline activity until ca. 13.5 ka. This work emphasises the frequency of extensive later PLT palaeo-lakes at the 934–936 m level in the Ngami basin and beyond during the Late Pleistocene (Burrough et al. 2007). Furthermore, Shaw et al. (2003) suggest that equally extensive later PLT palaeo-lakes continued in the basin up to 3–4 ka during Holocene times.

4.3.2 The Mababe Depression

There is a relative absence of historical accounts pertaining to water bodies in the relatively featureless Mababe Depression. Travellers like Passarge (1896–1898) who encountered Lake Ngami mapped it as the ‘Ngami See’ as it was full of water at the time (Table 4.2), whereas a general area to the east of the Okavango is entitled ‘Mababe Sumpt’ on 1896–98 maps due to marshes in the Depression (Vanderpost and Hancock 2018). The present-day lake bed is dry with minor ephemeral wetlands, specifically the Savuti marsh to the north-west (Fig. 4.3c) and the Mababe marsh to the south of the Depression (Fig. 4.7). Currently, drainage into the Mababe basin mainly comprises overflow from the Okavango Delta in addition to some flow from the Cuando River (Fig. 4.1). South-eastern Okavango drainage enters the Mababe Depression at its southern tip and is derived mostly from the Khwai distributary through the Machaba and Kudumane rivers (Shaw 1984). Historical evidence suggests that these rivers (Fig. 4.7) fed a substantial swamp around 1910. Since then the southern basin has contained water on seven occasions derived from Khwai inflow (Shaw 1985b). The Khwai River also flowed variably between 2009 and 2015, causing flooding in and around Mababe village. The floods damaged the Mababe road bridge (Fig. 4.3e) and filled the southernmost part of the Depression. The second southern inflow is currently a fossil back-swamp, which runs north-east along the Thamalakane fault line called the Mokhokhelo River (Fig. 4.7). Furthermore, north inflow to the Mababe Depression takes place through the misfit Savuti River, which crosses over the Linyanti fault before flowing eastwards along with the Linyanti fan. The river cuts through the Magikwe ridge into the northern section of the Mababe depression, forming the Savuti swamp (Fig. 4.5). Historically, the Savuti River fed the Savuti marsh regularly until the late 1880s, after which it dried up completely until flow resumed in 1958 (Shaw 1985b). Following a long dry spell, flow again resumed in 2009–2015.

Fig. 4.7
figure 7

Main geomorphological features of the Mababe Depression, described in text. (Map modified after Thomas and Shaw 1991)

Full size image

Interestingly, the Savuti River inflow arises from both the Okavango Delta and the Linyanti swamps, which are fed by the Cuando River. Some Cuando-Linyanti outflow in turn drains circuitously towards the Zambezi River via the Chobe River backflow. The upper Savuti River drains from the Linyanti swamp across from where the Selinda spillway (an Okavango distributary) accesses the Linyanti swamp. This co-incidence of inflow and outflow drainage suggests that when Okavango flooding takes place through the Selinda spillway, this initially fills the southern Linyanti swamp, which increases flow down through the Savuti River (Fig. 4.5). Hence, it is currently difficult to trace the origins of northern Mababe drainage as this may originate as Cuando inflow through the Linyanti swamp (which is mainly tilted NE) or as Okavango inflow diverted through the Selinda spillway (Fig. 4.2). Furthermore, movement on the relatively active Linyanti or Chobe faults (Kinabo et al. 2008) could readily divert inflow to the Mababe Depression in favour of either the Cuando River or Selinda spillway (Fig. 4.5).

In terms of dimensions, using the outer circumscribing boundaries, the Mababe Depression is approximately 92 km long and 64 km wide (or 5 888 km2), while the inner featureless lake bed is 51 km long and 20 km wide (or 1 020 km2). Hence, the depression is significantly larger than in the Ngami basin. About 85% of the Mababe Depression is made up of palaeo-shorelines, foreshore plains, estuaries and deltas (Fig. 4.7). The basin is subject to relatively high levels of seismic activity, with a higher rate of earthquake incidence than Lake Ngami. This occurs mainly along the southern margin of the Cuando-Linyanti trough and the west-central part of the Mababe Depression (Scholz et al. 1976). Gumbricht et al. (2001) indicated that the main Mababe palaeo-shoreline, the Magikwe ridge (Figs. 4.3d and 4.7), varies in elevation from 930 m in the south to 945 m in the north, suggesting that the entire basin is tilting southwards. Given the tilting, the minimal elevation (sump) level occurs towards the southern basin at ca. 920 m.

The geomorphology of the Mababe basin is determined mainly by the uplifted Mababe fault scarp to the south-east, with its north-west facing down-slip and the Magikwe ridge which stands 15–20 m above the lake bed along the western margin (Figs. 4.3f and 4.7). Due to structural controls, the Mababe basin is aligned north–north-east to south–south-west. The Mababe Depression also comprises a number of unusual rocky outcrops towards the north of the main Magikwe ridge. These mainly comprise six small hills, which occur around the Savuti River inflow to the Savuti marsh including the Gubatsa Hills (Fig. 4.3f). The larger Goha Hill (Fig. 4.3g) arises beyond the northern edge of the basin. Extensive palaeo-shorelines, estuarine deposits and deltas in the Depression have been described by Cooke (1980), Mallick et al. (1981), Shaw (1985b), Thomas and Shaw (1991) and Burrough and Thomas (2008). The main Magikwe ridge runs along the entire western side of the Depression as an extensive palaeo-shoreline rising between 936 and 950 m (Fig. 4.3d). The ridge height has led to considerations of its formation by wind-generated wave action since it occurs on the leeward side of the Mababe palaeo-lakes (Grove 1969; Shaw 1988). The Magikwe ridge is capped by aeolian deposits (to 950 m), with sand blow-outs interspersed along the ridge crest (Fig. 4.3d) (e.g. Thomas and Shaw 1991). The ridge bifurcates northwards near the Savuti channel where ‘tombolo like features’ (Shaw 1985) anchor segments of the ridge to the Gubatsa Hills. The gentle west-facing back-slope of the ridge consists of ‘reworked Kalahari sand’ (Thomas and Shaw 1991), which may reflect foreshore reworking during the waning phase of PLM during the Early to Mid-Pleistocene (Figs. 4.5 and 4.7). The reworked sand area comprises linear ‘lagoonal’ ridges not unlike those identified on the backslope of the Gidikwe ridge of the Makgadikgadi system (e.g. Cooke and Verstappen 1984; McFarlane and Eckhardt 2008). Further fossil drainage on the west side of the Mababe Depression includes a south-east trending branch of the Savuti River, which cuts through the Magikwe ridge at the Tsatsarra Gap (Fig. 4.7). The area north of the eastern fault scarp is breached at two locations by the misfit Ngwezumba River and its tributary the Kashaba River, which drain intermittently off the eastern plateau. The Ghautumbi River to the south is less distinctive with no known flow (Shaw 1985b). The entry of these rivers into the Mababe basin has led to the formation of deltas and offshore bars attributed to palaeo-lakes standing between 924 and 936 m (Fig. 4.7) (Thomas and Shaw 1991).

Comparing the maps of Lake Ngami (Fig. 4.6) and the Mababe Depression (Fig. 4.7), it becomes apparent that palaeo-lake levels in the two basins show a measure of consistency, although the altitudinal rise on the Magikwe ridge (from ca. 936 to 950 m) detracts from this. While Moore et al. (2012) describe a single-event 936 m palaeo-shoreline (PLT) in both basins dating back to ca. 200–500 ka, Burrough and Thomas (2008) had previously identified and OSL dated apparently much younger palaeo-lake levels along the same Magikwe ridge at the PLT 936 m level. Their first palaeo-lake event is regarded as having taken place during Late Pleistocene time between 38 ± 6 ka and 35–36 ± 5 ka. This event is regarded as being consistent with the formation of the Dautsa Ridge in Lake Ngami between 38 ± 3 ka and 30 ± 3 ka and the Kareng ridge at ca. 32 ± 2 ka (Burrough and Thomas 2008). The second palaeo-lake event to attain the 936 m height is regarded as having occurred in both basins between 18 and 13 ka. This appears to be augmented by evidence of delta and terrace formation associated within the Ngwezumba and Savuti rivers (Fig. 4.7). The third 936 m palaeo-lake event that occurred in both basins is regarded as having taken place during the Early-Mid Holocene in the Mababe Depression at 8.4 ± 1.1 ka, 6.6–6.3 ± 1.1 ka, 5.5 –5.2 ± 0.6 ka. Burrough and Thomas (2008) suggest that these extensive inter-basin palaeo-lakes (as the PLT extended from Ngami to the Mababe basins, along the Thamalakane fault and southwards) resulted from positive hydrological budgets in the catchment area and a considerable variability in inflow through the Okavango Delta. The debate regarding the relative feasibility of one single Mid-Pleistocene palaeo-lake PLT at 936 m (Moore et al. 2012) vis a vis the three or four Late Pleistocene-Holocene palaeo-lakes at ca. PLT levels (Burrough and Thomas, 2008) hinges partly on a reasonable source for the volume of water required to feed and sustain such extensive water bodies (>40 000 km2, Moore et al. 2012). The occurrence in the Ngami and Mababe basins of such diverse events is considered again below.

4.3.3 Basin Lake Floor Sediments

Work on lake floor sediments in both the Ngami and Mababe basins began in 2002 and has resulted in trench data being independently reported by Huntsman-Mapila et al. (2006) and Gamrod (2009). Here a comparison of the two trench data sets is presented for the first time to invoke further discussion on Late Pleistocene-Holocene events in the basins. The data sets were derived from open trenches dug by mechanical excavators such that 3D bedding relationships and hiatuses in the sediment pile were readily identifiable. Trench-derived stratigraphic records from Lake Ngami were excavated at 922 m asl, close to the local palaeo-lake sump at 921 m, while the Mababe Depression excavation was taken from 920 m, close to the 919 m sump (Figs. 4.6 and 4.7). As basin depths may be as great as 800 m in Mababe and 400 m in Ngami (Kinabo et al. 2008), the trench data are evidently restricted to a fraction of the entire sequence. Sediment descriptions from the 460 cm deep Ngami trench and the 560 cm deep Mababe trench are shown in Fig. 4.8. These are augmented by comparative dates from both profiles (Table 4.3), diatom habitat interpretations from the Lake Ngami profile (Table 4.4) and particle size data that are used to help characterise the Mababe Depression profile (Fig. 4.9).

Fig. 4.8
figure 8

Stratigraphic profiles from the Ngami and Mababe trenches indicating four depositional phases described in text. Late Pleistocene-Holocene dates and depositional hiatuses are shown for each profile. (Individual profiles modified after Huntsman-Mapila et al. 2006 and Gamrod 2009)

Full size image
Table 4.3 Comparison of age dates from different dating techniques in the Ngami and Mababe sub-basins (after Huntsman-Mapila et al. 2006 and Gamrod 2008)
Full size table
Table 4.4 Interpretation of diatom genera, Ngami profile with examples of dominant taxa (original data in Huntsman-Mapila et al. 2006)
Full size table
Fig. 4.9
figure 9

Source Gamrod (2009)

Particle size data from Mababe trench showing dated depositional discontinuities at three hiatus level. 480 cm hiatus dated ca. 64.8 ka (OSL), 380 cm hiatus dated ca.41.3 ka and 150 cm hiatus ca. 27.0 ka.

Full size image
Table 4.5 Summary of Ngami (LN) and Mababe (MD) selected major and trace metals (original data from Huntsman-Mapila et al. (2006; Gamrod 2008)
Full size table

Dates for the two profiles were obtained using OSL (Mababe Depression) and TL with 14C (Lake Ngami) methods (Table 4.3). These are plotted on the profiles (Fig. 4.8) and indicate how the Mababe deposits appear generally older than those in the Ngami trench. To illustrate the relative antiquity of the two profiles, lines were drawn to identify similar dates from each. The area between the semi-isochronous lines has led to the identification of four depositional phases. For instance, the first line is drawn between the location of the sample used for the 40.5 ka date in Ngami and the 41.3 ka date in the Mababe profile. Hence, Phase 1 extends from an unknown time at depth to ca. 40 ka in both profiles. A second line is drawn to represent ca. 27 ka in both profiles, whereas the third line represents ca. 11.5 ka. Interestingly, depositional hiatuses representing non-deposition and/or drying events were also identified (Figs. 4.8 and 4.9). In most cases, the hiatuses occur at or near the dates described above. Hence, the profile data are subdivided into what becomes four depositional phases mostly separated by near-contemporary hiatuses and based on dates.

The earliest Phase 1 depositional period comprises initially deeper water lacustrine sediment at the base of the Mababe trench followed by organic clayey silt, indicative of swamp conditions at ca. 65.8 ka. This changes vertically into more shallow water sediments, which terminate at the second hiatus dated at ca. 41.3 ka. This compares with sediment at the base of the Lake Ngami profile, which comprises silty sand with calcrete nodules, which may also be indicative of an increasingly shallow lacustrine environment terminating with possible sub-aerial exposure around 40 ka. This non-depositional interval appears to be in some agreement with a drying event (41–46 ka) indicated in Stokes et al. (1997), but is at variance with the high palaeo-lake levels at 38 ± 6 ka from both basins in Burrough and Thomas (2008). A ca. 40 ka date is also at variance with data from Ringrose et al. (2008) in terms of floodplain formation in the Okavango Delta at this time. Phase 2 extends over ca. 13 000 years, (Fig. 4.8) when 230 cm of sediment was deposited in the Mababe Depression contrasting with only 90 cm in Lake Ngami. Phase 2 deposits in the Ngami profile comprise a relatively thin pale yellow sandy-silt possibly attributable to fluvial or through-flow deposition (Fig. 4.3h). The thicker sediment in the Mababe trench commences with deeper water silts and a sandy-silt, similar to that found in the Ngami trench (Fig. 4.8). However, most of the profile in Phase 2 comprises thick white diatomaceous silts indicative of shallow lacustrine conditions, capped by deeper water silt and clay. Hence, a contrast exists between thin, fluvial sediments in Lake Ngami and the thicker lacustrine sediments of the Mababe Depression between ca. 40 ka and 27 ka. The 27 ka marker is again represented by a hiatus, which is coincident with a drying interval between 20 and 26 ka (Stokes et al. 1997) and a non-palaeo-lake period in Burrough and Thomas (2008).

Following the dry period, Phase 3 starts after 27 ka and continues to ca. 11.7 ka (Fig. 4.8). During this time, the Mababe sequence initially depicts shallow water deposition followed by a deeper water interval, which later changed to an intermittently wet, swampy environment. Over the same time period, Lake Ngami deposits show an initial period of shallow, followed by deeper water sediments. This is followed by two totally different laminar and massive diatomaceous beds up to 140 cm in thickness, which are indicative of shallow, alkaline waters suggesting a closed basin environment (Table 4.4). The diatomaceous beds terminate in a further hiatus suggesting drying at ca. 11.7 towards the beginning of the Holocene. Hence, a contrast exists between the thicker lacustrine diatomaceous sediments of the closed Ngami basin and the thinner lacustrine-swamp sediments of the also closed Mababe basin between ca. 27 ka and 11 ka. This is co-incident in terms of timing with overall drying between 9 and 16 ka (Stokes et al. 1997) and a non-palaeo-lake period in Burrough and Thomas (2008). During Phase 4, drying swamp conditions continued to prevail in the Mababe Depression to the point of desiccation. However, the Phase 4 Lake Ngami deposits represent an extended period of shallow water depicted by calcareous sandy-silt and diatomaceous beds (Fig. 4.8). These were later capped with peat beds inferring swamp conditions (Fig. 4.3i), here dated at 3.7 ka. The swamp vegetation was later burnt (cf. Shaw et al. 2003).

Cordova et al. (2017) describe the interval ca. 16 600–12 500 cal BP as being one of fluctuating but high summer rainfall in the Ngami basin, which augmented Okavango inflow and probably contributed to the deposition of shallow, diatomaceous sediment (Fig. 4.8). This was followed by reduced rainfall at ca.11.4 ka, reflecting the dry interval implied by the hiatus at ca. 11.0 ka in the Ngami profile (Fig. 4.8). Interestingly, a date of 11.7 ka was also obtained from the middle of the upper organic-rich silty-clays in the Mababe profile and so recognising the early part of Phase 4 as a very Early Holocene drying event. In contrast, the upper beds in the Ngami profile may initially reflect increasing rainfall with shallow water conditions throughout the last 10 000 years (cf. Cordova et al. 2017).

Additional geochemical evidence was obtained to shed light on possible reasons for different conditions being prevalent in both lake basins in terms of the timing of inflow from different sources (cf. Huntsman-Mapila et al. 2005; Ringrose et al. 2009). For instance, in Table 4.5, the transition metals, V, Co, Pb and Ni, show noticeably higher abundances in Mababe sediments, whereas Cr values are higher in the Ngami deposits. The transition metals are likely derived from headstream areas characterised by the Palaeoproterozoic Congo craton of the Angolan Highlands (Fig. 4.1). V is relatively common in the schists and related meta-sediments of the Angolan highlands forming vanadiferous (with chromite and magnetite) mafic to ultra-mafic intrusions (e.g. Jelsma et al. 2018). Comparative plots were constructed of V, Co and Pb abundances within Lake Ngami and Mababe Depression profile sediments (Fig. 4.10a, b). Most data variability is depicted by the V trends, which generally decline through Phase 1 in both the Ngami and Mababe data sets. During this early time span (ca. 65 ka to 40 ka), the range of values in the Lake Ngami profile lies between 75 and 25 ppm, whereas the range in the Mababe Depression lies between 240 and 100 ppm. Since these ranges of values are so dissimilar, different inflow sources are inferred for both basins. In Phase 2, the Ngami V data in Fig. 4.10a essentially occupy a trough suggesting low sediment inputs at around 25 ppm, corresponding to sandy-silt deposition (Fig. 4.8). In contrast, Phase 2 in the Mababe data (Fig. 4.10b) is represented by fluctuating peaks and troughs ranging through 240–110 ppm, suggesting variable inflow and sediment inputs commensurate with a diatomaceous palaeo-lake. The overall much higher V content in Mababe suggests a different inflow source from that received in the Ngami basin. This changes in Phase 3 where the Ngami data (post 27 ka) now show a range of fluctuating inputs from 175 to 52 ppm whereas the Mababe data are relatively flat between 175 and 125 ppm. This may reflect a new regime when river inflow was essentially derived from a similar source. The similar though fluctuating V values continue through Phase 4 when the Ngami data show a wide range of V peaks and troughs (between 155 and 22 ppm) throughout the Mid-Late Holocene.

Fig. 4.10
figure 10

a Transition metal plot from Ngami trench showing trends through four depositional phases. Note differences between the V scale on right and Co, Pb scale on left of graph. Source P. Huntsman, 2006. b Transition metal plot from Mababe trench showing trends through four depositional phases. Source Gamrod 2009

Full size image

The relatively high abundance of V in addition to Co and Pb, in the Mababe Depression during Phases 1 and 2, may be explained by the closer proximity of Mababe to vanadiferous deposits in the headstream source area, which are more directly accessible through the Cuando River (Fig. 4.1). Conversely, Lake Ngami appears to have long obtained inflow via a longer route through the Okavango system. In addition to receiving inflow from the Okavango River, the Delta is known to be a sink for a wide range of metals (Masamba and Muzila 2005), which are retained within the sand and peat rich sediments. Hence, the Okavango Delta distributaries that flow into Lake Ngami likely became relatively depleted in transition metals (Fig. 4.2). The abrupt transition from higher V-enriched Cuando inflow into the Mababe basin between ca. 65 and 27 ka (depositional Phases 1 and 2) to Okavango inflow after 27 ka (during depositional Phases 3 and 4) may be related to local tectonic shifts (cf. McCarthy et al. 1993) particularly within the earthquake active zone towards the southern tip of the Linyanti swamp-trough (Fig. 4.5) (Scholz et al. 1976; Cruse et al. 2009; Gamrod 2009) after which both basins received inflow via Okavango Delta sources. Hence, tectonic movement along the Linyanti or Chobe faults is invoked to explain the transition metal change, which occurred only in the Mababe Depression sediments. This, in turn, suggests that the two basins operated separately in terms of inflow at least until ca. 27 ka. After this date and throughout the Holocene, the predominant separate inflow appears to have been derived from mainly decreasing Okavango River and Delta sources.

The general picture provided by basin sediments in both Lake Ngami and the Mababe Depression is of two separately functioning lake basins often characterised by shallow but often fluctuating palaeo-lake levels with earlier sediments derived from different sources. Hence the overall results of palaeo-lake sediment analyses fail to endorse the existence of sequential, deep (for example ca. 15 m deep at the 936 m) and persistent palaeo-lakes within the Lake Ngami and the Mababe Depression, between ca. 65 ka and 4 ka (cf. Burrough and Thomas 2008).

4.4 Issues and Future Work

Two main controversial issues have been described during this review of the geomorphological histories of two relatively small lake basins, Lake Ngami and the Mababe Depression, which are peripheral to the Okavango Delta. The first involves the age differential indicated by different literature sources for the palaeo-shorelines, especially those of Palaeo-Lake Thamalakane (PLT). The second issue centres around the frequency or persistence of the palaeo-lakes hinging on whether these were single sequential occurrences or multiple occurring events.

Table 4.1 depicts incremental palaeo-lake incursions into the ORZ, which formed shoreline sequences in both Lake Ngami and the Mababe Depression. Moore et al. (2012) and McCarthy (2013) suggest that the palaeo-shorelines (from PLM and PLT in particular) formed from relatively ancient palaeo-lakes, which developed from ca. 1.4 Ma to 200 ka. The existence of the large palaeo-lakes is based on inflow (and later diversions) of the Cuando, Zambezi and Okavango rivers. These palaeo-lakes have been allocated minimum ages using ESA and early MSA stone tools (Robbins and Murphy 1998; McFarlane and Segadika 2001) associated with the two basins and the adjacent Makgadikgadi. Work by Moore et al. (2012) demonstrates that the existence of the early mega-lakes effectively terminated after the reconnection of the Upper and Mid-Zambezi (Table 4.1). Hence the palaeo-lakes were single ever diminishing, sequential occurrences. On the other hand, Burrough et al. (2009a) have inferred several later mega-lake phases (covering both the Ngami and Mababe basins) using numerous OSL dates, at the PLM/PLT levels during the last ca. 100 ka. Burrough et al. (2009b) argue that an augmented Okavango River provided most of the water for the later mega palaeo-lakes and suggest that cooler temperatures led to reduced evapotranspiration which in turn drove mega-lake expansion from ca. 100 to 20 ka. For the period from 19 to 15 ka, however, Burrough et al. (2009b) propose a southward shift of the ITCZ resulting in higher precipitation in the catchment area of the Okavango River which thus drove later high-level palaeo-lake events. Hence, mainly climatic factors drove these multiple occurring events to the same palaeo-lake level. While it may be argued that possible errors in these younger dates result from aeolian sediment reworking on the palaeo-shorelines (Moore et al. 2012) or bioturbation possibly via termites (McFarlane et al. 2005; Bateman et al. 2007), the discrepancies remain. Since the dates 200–500 ka versus 100–15 ka pertain to the exact same palaeo-shorelines (945 m PLM and 936 m PLT), then additional work is required to help resolve the differences presented including the palaeo-climatic implications.

Obtaining an unequivocal answer may lie in a different approach to the problem such as the initiation of an interdisciplinary deep drilling (to near bedrock) project. This could provide a series of cores to several 100 m (Kinabo et al. 2008) through both the Ngami and Mababe basins and involve the analysis and dating of as many samples as possible. There is abundant evidence in the literature to suggest that the interpretation of deep drill cores is increasingly useful (e.g. Johnson et al. 2016). The location of the Mababe-Ngami structural depressions between the Lake Malawi drill site(s) and the Pretoria Salt Pan (Partridge et al. 1997) should provide ideal locations from which to obtain conclusive palaeo-lacustrine evidence and useful additions to the palaeo-climatic record of southern Africa (A. Moore, pers.com).

4.5 Conclusions

  1. 1.

    Lake Ngami and the Mababe Depression currently receive inflow from the Okavango River via the Okavango Delta and as such form overflow basins of temporary storage from the Okavango system. Both the Ngami and Mababe basins form deep, geomorphologically diverse elongated troughs with maximum extents ca. 2 660 km2 and 5 888 km2, respectively.

  2. 2.

    Whereas the timing of their origin is largely uncertain, the basins were formed as structural depressions originating as a result of East African Rift propagation along the Kunyere and Thamalakane (Mababe) fault lines within the Okavango Rift Zone. The two basins are embedded in older Okavango Delta fans, namely the Groote Laagater and Linyanti fans (McCarthy 2013).

  3. 3.

    Recent evidence from Moore et al. (2012) suggests that the Ngami and Mababe basins were entirely submerged during the Palaeo-Lake Deception period of the Early Pleistocene, with higher points emerging as islands during the Palaeo-Lake Makgadigadi period (Early-Mid Pleistocene). Palaeo-shorelines mainly developed along the basins’ peripheries during the Palaeo-Lake Thamalakane (PLT-936 m) period possibly ca. 200–500 ka. However, Shaw et al. (2003), Burrough et al. (2007) and Burrough and Thomas (2008) suggest that the 936 m palaeo-shorelines within the ORZ were formed on numerous occasions within the last 100 ka due to mainly climatic and feedback factors.

  4. 4.

    Limited trench data from Ngami and Mababe sediments suggest that after 65 ka the two basins operated as mostly separate, often closed system alkaline lakes, with early inflow rivers from different sources.

  5. 5.

    Controversies prevail in terms of the dating of the palaeo-shorelines (500–200 ka vs. 100–4 ka) and in terms of the palaeo-climatic implications of the same large palaeo-lakes being prevalent at different times. Such divergent results may be resolved by a comparative comprehensive drilling programme with full sedimentological analyses and the dating of samples in both basins.