The Mwesia Beds of northern Malawi in relation to the Tanganyika Problem Open Access

Lake Malawi, reaching 700 m in depth, is the second deepest lake of the East African Rift System, after Lake Tanganyika. These lakes are also remarkably speciose, reflecting geographical differentiation and speciation. In the late nineteenth and early twentieth centuries, differences in species composition and species diversity were explained in part by an assumed Jurassic marine incursion because of fossils found in northern Malawi and presumed to be marine relicts surviving to the present day in Lake Tanganyika. This notion was known as the ‘Tanganyika Problem’. Here we provide a summary of the geological context of northern Malawi, relocate the original nineteenth century locality, document the occurrence of a chronologically significant Permian fossil from the site, and briefly comment on the Tanganyika Problem in the context of palaeogeography and history. In effect, this is a narrative of geoheritage that takes the observations and interpretations of rocks and fossils as they were in colonial times and places them within the context of current geological understanding and the cultural setting of Malawi today.

The formation of the East African Rift System is associated with uplift of the African Superswell. The Lake Malawi Rift lies on the western branch of the East African Rift System. Roberts et al. (2012) determined that initiation of rifting in the western branch of the East African Rift in Tanzania occurred about 26–25 Myr ago. They considered that the initiation of the northern Lake Malawi Rift had a similar history.

Structurally, Lake Malawi is divided into three half-graben. The hinge zone of the North Basin lies west of the lake and the border fault lies near the North Basin's greatest depth (Kolawole et al. 2018). Basement-controlled relief on the lake bottom between the three Lake Malawi basins likely influenced species distribution and the evolution of endemic diversity within the lake (Scholz et al. 2020). Fossils relevant to human evolution are found in the Chiwondo Beds along the shore (Bromage et al. 1995; Lüdecke et al. 2016). Shillington et al. (2020) noted that intrarift faults of the North Basin offset a 75 kyr-old horizon and should be considered for seismic hazards.

Tectonic fabrics in underlying Precambrian crystalline basement influence younger fault patterns in and adjacent to the Lake Malawi Rift. Paleozoic and Mesozoic faults along the rift flank formed asymmetrical graben containing Permian Karoo deposits and Cretaceous Dinosaur Beds. These fault-bounded basins in northern Malawi preserve terrestrial tetrapods including therapsids from the Permian Chiweta Beds (Haughton 1926; Dixey 1937; Jacobs et al. 2005; Botha and Angielczyk 2007; Kruger et al. 2015) and Cretaceous turtles, crocodiles and dinosaurs from the Dinosaur Beds (Dixey 1928; Haughton 1928; Clark et al. 1989; Jacobs et al. 1990, 1992, 1993; Jacobs 1993; Gomani 1997, 2005; Andrzejewski et al. 2019).

Yemane et al. (1989, 1996) and Yemane and Kelts (1996) studied the clay minerals, calcite concretions and oxygen isotopes from a large, shallow, perennial, Late Permian, Karoo lake complex in northern Malawi near Mount Chombe. Lake sediments were immature, derived from a metamorphic source, and showed little reworking. The clay mineral assemblage supports a temperate seasonal climate and oxygen isotopes suggest a Late Permian mean annual temperature of 10°C. Currently located at 9°S, 32°E, the palaeolatitude at the time of deposition was approximately 55°S. The fossiliferous Late Permian Chiweta Beds lie on the west shore of Lake Malawi, north of the Permian Karoo lake complex studied by Yemane, near the accommodation zone between the North and Central basins of Lake Malawi. Further north in Malawi, fluviolacustrine Karoo deposits called Mwesia Beds occupy small fault basins (Fig. 1). The Mwesia Beds were originally called Drummond's Beds, which played a large role in the Tanganyika Problem.

In the late nineteenth century, in northern Malawi, Stevenson Road was being laid out between Karonga on Lake Malawi to Lake Tanganyika. The descent from the plateau between the two lakes traversed countryside riddled with faults. As a result of those faults, structural basins expose sedimentary rocks of Permian age (Fig. 1).

In 1883, Henry Drummond (1851–97) was sent to conduct a natural history survey between Lake Malawi and Lake Tanganyika (Stewart and Coles 1880; Stewart 1881; Thompson 1989). Along the Stevenson Road route, Drummond found a small slab of rock with fossil fish bones (Smith 1898). That led to the discovery of more fish fragments and bivalve molluscs, which he proclaimed were ‘the only fossils that have ever been found in Central Africa’ (Drummond 1888, p. 192). The site (Fig. 1, inset) was in Mpata, an area of approximately 40 km² along Stevenson Road (Drummond 1884; Fotheringham 1891; Andrew and Bailey 1910). Drummond (1888, pp. 191–192) wrote ‘… after following the Rukuru river through a defile of granite rocks, I came, to my great surprise, upon a well-marked series of stratified beds’. That locality lies within the Mwesia fault basin, named for the Mwesia River, a north-flowing tributary to the Rukuru River. No other fossils were discovered in the Mwesia Basin until 2016 when a curious herdsman discovered a fossil that greatly improved the palaeontological significance of the Mwesia Basin and further clarified an historical conundrum.

Specifically, the fish and bivalves Drummond found at Mpata were instrumental in assessing a late nineteenth and early twentieth century hypothesis to explain the odd composition and great biological diversity of modern Lake Tanganyika. Lake Malawi lies 350 km SE of Lake Tanganyika. As close as Lake Tanganyika is to Lake Malawi, they lie in entirely separate basins and their faunas are different. An argument was made that Lake Tanganyika had been an arm of the sea in the geological past and that it had retained relict marine species acquired at that time and remaining until the present day (Moore 1903).

The Tanganyika Problem derives from the morphology of modern gastropod shells collected in 1858 by John Hanning Speke (1827–64) along the shore of Lake Tanganyika (Fryer 2000). Those shells showed similarities in thickness to marine gastropods (Woodward 1859), suggesting a biogeographical connection between Lake Tanganyika and the sea at some time in the past. Later, Günther (1893, 1894) named a modern jellyfish medusa (Limnocnida) from the lake and considered that its presence also supported an earlier connection to the sea.

J.E.S. Moore (1870–1947) led two collecting expeditions to gather data relevant to the geological setting and the hypothesized marine influence on Lake Tanganyika (Moore 1898a, b) and published a book, The Tanganyika Problem (1903), expanding on his studies. Moore considered his book to contain ‘… by far the most extended and continuous geological studies of the African interior which have hitherto been made …’ (Moore 1903, p. 55). The Tanganyika Problem includes his geological maps of the Lake Malawi and Lake Tanganyika regions (made with his surveyor; Fergusson 1901). They show an area in northern Malawi mapped as what he considered Drummond's Beds, now referred to as the Mwesia Beds. Traquair (in Drummond 1888) took the age of Drummond's beds to be Triassic; however, Moore considered that because modern Lake Tanganyika gastropods to him resembled Jurassic fossils, a Jurassic connection to the sea may have existed. Currently, no Jurassic rocks, or marine rocks of any age, are recognized in the area mapped by Moore.

William Thomas Blanford (1832–1905) (Blanford 1903a, b) reviewed Moore's book in terms of its geological conclusions on the one hand, and its conclusions dealing with the reported marine fauna in Lake Tanganyika on the other. Blanford (1903a) considered Drummond's Beds to be equivalent to the Karoo (in part) of South Africa, he stated that there was no proof that they were marine, he chided Moore for not mentioning coal in the section or plant fragments found with the bones, and he argued that Drummond's Beds were of freshwater origin. Drummond himself (1884) stated he thought them to be of likely lacustrine origin. Blanford (1903b) further elaborated that Drummond's Beds, like that part of the Karoo to which Blanford was referring, were of freshwater origin and likely fluviatile. He stated that the fish and bivalves from Drummond's beds did not require marine conditions, consistent with the conclusions of Jones (1890) and Amalitsky (1895), who also suggested Karoo affinities for fossils from Drummond's beds. Zoogeographical issues concerning Moore's formulation of the Tanganyika Problem based on the medusae and on the resemblance of Lake Tanganyika gastropods to Jurassic fossils were rejected by Cunnington (1920). This was because the medusae were found to have a broader freshwater distribution across Africa, and because, while Moore had considered that Tanganyika was originally salt water, the abundance of modern species in the lake were related to freshwater forms elsewhere.

Marine flooding of the Congo Basin was a pervasive consideration in the Tanganyika Problem. Indeed, a marine incursion into the Congo from the East was generally accepted for much of the twentieth century and into the twenty-first century based upon a thin, fish-bearing, presumably marine Lime Fine bed found low in the Upper Jurassic Stanleyville Group of the eastern Congo Basin (Cahen 1954, 1983; Saint-Seine 1955; Saint-Seine and Casier 1962). Subsequently, the presence of marine fish was questioned, the marine origin of the Lime Fine was repudiated, sedimentological analysis concluded marine beds were not present in the Stanleyville Group, and palaeosol analysis indicated continental conditions (Myers et al. 2011; Taverne 2011; Agyemang et al. 2016; Caillaud et al. 2017). In contrast, palaeogeographical modelling maintains a marine incursion into the eastern Congo (Scotese 2021).

The fossils found by Drummond were matched by fossils found by Andrew and Bailey (1910) in similar beds of the Ngana Basin, along the northern border of Malawi with Tanzania, about 40 km north of the Mwesia Basin (Fig. 1). The fossiliferous sediments in the Mwesia and Ngana basins were mapped as Karoo by Andrew and Bailey (1910) and are consistent with the distribution of Drummond's Beds on Moore's geological map in his book. The fossil fish from the Ngana Basin were studied by Traquair (1910), who had previously studied Drummond's original fossils (Drummond 1888). Dixey (1926) continued the geological study of the Malawi Karoo and discovered fossils of terrestrial vertebrates at Chiweta near Mount Chombe (Haughton 1926). The fossils from Chiweta, which lies some 90 km south of the Mwesia Basin, were considered younger than those of Mwesia and Ngana based on lithological correlation (Dixey 1926). Subsequently, mapping and evaluation of coal resources by the Geological Survey of Nyasaland and then the Malawi Geological Survey Department, especially in the Ngana Basin, clarified Karoo stratigraphical nomenclature in Malawi and applied the name Mwesia Beds to the fossiliferous deposits in the Mwesia and Ngana basins (Bloomfield 1957; Thatcher 1974; Kemp 1975; Ray 1975).

No other fossils from either basin were reported until 2016 when Benjamini Sikapite, while tending livestock in the Mwesia Basin (Fig. 2a, b), picked up a fist-sized stone and recognized a skeleton visible on the surface. The stone containing the skeleton was a calcareous concretion (Fig. 3a) not unlike the bone-containing concretions of the South African Karoo. Sikapite was employed by Harrison H. Simfukwe, Director of the Cultural and Museum Centre Karonga, Malawi. The museum displays fossils as well as historical artefacts, and it is the public face of geohistory in Malawi. The fossil Sikapite found was a nicely preserved skeleton of Eunotosaurus (Cultural and Museum Centre Karonga, Mpata Karoo, MK 16-1), an iconic, easily identified taxon, which until Sikapite's discovery was only known from the Karoo Basin, South Africa, some 2400 km away. Simfukwe obtained GPS coordinates of the locality and undertook a five-day reconnaissance in which he discovered an additional five prospects with fossil bone in the Mwesia Basin. Precise locality data are on file at the Cultural and Museum Centre Karonga.

The most immediately recognizable feature of Eunotosaurus is the flattened, enlarged, easily recognized ribs, suggestive of a primitive turtle carapace (Fig. 3a, b). The characteristics of the ribs and skull in particular have been used to infer a basal phylogenetic position for Eunotosaurus in the evolution of turtles (Lyson et al. 2013, 2016; Bever et al. 2015, 2016). However, this evolutionary position for Eunotosaurus is not universally accepted (Abel and Werneburg 2021; Lichtig and Lucas 2021). Regardless of its evolutionary position, all known South African specimens of Eunotosaurus are included in a single species, Eunotosaurus africanus. The orbit of MK 16-1 seems relatively large and round (Fig. 3c) compared with E. africanus, a possible juvenile ontogenetic feature, but it otherwise appears to exhibit features of that species.

The significance of Sikapite's find in the context of the Malawi Karoo is that Eunotosaurus is a Permian genus with a reasonably constrained chronostratigraphic position in the South African Karoo Basin. Figure 4 is a stratigraphic and geochronological chart showing South African Karoo lithostratigraphy, biostratigraphy and radiometric dates (modified from Rubidge et al. 2013; Day et al. 2015), with the stratigraphic range of Eunotosaurus highlighted (from Day et al. 2013; Groenewald et al. 2019) and the Mwesia Beds and Chiweta Beds added. The Mwesia Beds correlate with the Tapinocephalus and Pristerognathus zones because that is the range of Eunotosaurus in South Africa. The top of the Pristerognathus Zone could be as young as 258 Ma, taking uncertainty into account, which is the youngest the Mwesia Beds are likely to be. The top of the Tapinocephalus zone is close to 260 Ma and the age of the base is >264 Ma (Groenewald et al. 2019; Day et al. 2022). The Chiweta Beds of Malawi are correlated with the Cistecephalus Zone of South Africa based on the fauna (Jacobs et al. 2005; Kruger et al. 2015), which lasted approximately two million years (roughly 257–255 Ma).

Today, Moore's Tanganyika Problem, especially his interpretation of Drummond's Beds and its fossils, is an anachronism, notwithstanding its occasional mention in more recent literature (e.g. Wilson et al. 2008). While his geological approach with respect to Drummond's Beds is untenable, it remains that the evolutionary transition of clades from salt water to freshwater habitats undoubtedly occurred on many occasions through geological time. Cladistic analyses and ancestral state reconstructions suggest that the evolutionary dispatch from marine to freshwater is the more common direction (Schultz and Boush 2022). Further, Schultz and Boush (2022) discuss which physiographic changes facilitate transitions, foremost of which is the establishment of estuarine gateways during maximum marine flooding events. Estuaries are characterized by intermingling variable halohabitats that change at various timescales. Thus, an understanding of changing coastlines and palaeogeography provides an approach to identifying likely time intervals in which opportunities for marine to freshwater transitions are maximized.

Scotese (2021) – see also Kocsis and Scotese (2021) – provides a series of palaeogeographical maps representing 5 Myr intervals through the Phanerozoic. The map for 160 Ma (Oxfordian, Fig. 5) shows a maximum incursion of the sea in the Late Jurassic. This time predates by over 100 Myr the uplift of the African Superswell and the formation of the East African Rift System and its lakes. The incursion spread westward across Africa, to the north of what is now Malawi, to eastern Congo. Moreover, while an incursion into the Congo Basin was rejected by Caillaud et al. (2017), Taverne (2019) described a new genus and species of horned pycnodontiform fish, Congopycnodus cornutus, from the Stanleyville Group of the eastern Congo.

Pycnodontiform fish are considered to have a Late Triassic marine origin. They comprise a diverse clade of predominantly marine species, which went extinct in the Paleogene (Cawley et al. 2021). Nevertheless, pycnodontiforms are present in Cretaceous freshwater deposits of Hungary (Kocsis et al. 2009; Szabó et al. 2016), presumably having secondarily adapted to a freshwater habitat. If the habitat of Congopycnodus was freshwater, it too may have belonged to a secondarily adapted clade, facilitated in its transition to freshwater by the 160 Ma transgression shown by Scotese (2021). If it was in fact marine, its presence is explained by the same flooding event. No other relevant transgressions are known in Africa between the Late Triassic, when pycnodontiforms originated, and the Late Jurassic occurrence of Congopycnodus.

Complementing time-calibrated inundations and fossils, time-calibrated genomic techniques can unravel complex problems of phylogenetics, define monophyletic groups, and then be applied to biogeography, evolutionary origins and distribution patterns for extant taxa. Lavoué (2020), for instance, determined that ten African fish lineages out of 37 monophyletic groups originated in shifts from marine to freshwater habitats. Wilson et al. (2008) derived the freshwater herring of Lake Tanganyika from a Late Cretaceous–Paleogene West African incursion of the sea, which formed the Trans-Saharan Seaway, and Bragança and Costa (2019) proposed that African lampeyes originated in the retreat of the same seaway. Weiss et al. (2015) concluded that the cichlids of Lake Tanganyika derived from several cichlid lineages that diverged prior to the formation of the lake based on older genomically generated dates than the geologically determined age of the lake. With respect to gastropods, a group that was central to Moore's Tanganyika Problem, Wilson et al. (2004) proposed that four lineages of Lake Tanganyika gastropods pre-date the lake's formation.

The Great Rift Valley in all its reaches is an immense source of geoheritage. We have focused on the Mwesia Beds in the fault basins of northern Malawi, and that portion referred to originally as Drummond's Beds, including the citizen science that allowed resolution of their age. Although the Mwesia Beds have nothing at all to do with Lake Tanganyika, the Tanganyika Problem and the Mwesia Beds provide an interesting saga in the history of geology: they relate to the understanding of the African Permian, and they highlight geology as a significant factor in deciphering the biodiversity and biogeography of African fishes and aquatic invertebrates, including molluscs. These topics have been of great interest since the ‘discovery’ of the African Great Lakes. Moreover, there is a resurgence in the study of Karoo basins distributed across the countries neighbouring Malawi (Wopfner 2002; Catuneanu et al. 2005; Simon et al. 2010; Angielczyk et al. 2014a, b; Sidor et al. 2014; Cox and Angielczyk 2015; Nesbitt et al. 2017; Araújo et al. 2020). Only recently has a terrestrial vertebrate fossil, not a fish or bivalve mollusc, been recovered from the Mwesia Beds, that being the Eunotosaurus skeleton found by Sikapite.

Thus, over a century after Moore's (1903) The Tanganyika Problem was published, significant issues relevant to unravelling geology, palaeoecology and evolution are being pursued with new tools and techniques in northern Malawi and surrounding areas. Indeed, a recent paper addressed the question, ‘how convergent are Lake Tanganyika's gastropods to marine ones?’ (Vermeij 2019, p. 508). Of broader significance, and of particular relevance, to this discussion, Estes and Vermeij (2022) address the issue of ‘why future progress in ecology demands a view of the past’. The answer to that question seems apparent in the context of this discussion. Estes and Vermeij (2022, p. 7) state

with the current genomics revolution, we stand at what may well be the opening of a new frontier in our ability to understand mechanism, timing, and the historical abundances of species.

We would add that timing, historical abundances and, for many issues, mechanisms are also in the realm of Earth sciences.

The fact that the Eunotosaurus specimen from the Mwesia Beds was found by Benjamini Sikapite and taken to the local Cultural and Museum Centre Karonga is also significant. Several studies have documented the cultural biases inherent in colonial exploration (Monarrez et al. 2022; Raja et al. 2022). Here, Benjamini Sikapite's discovery and his subsequent actions necessitate a rethinking of Malawi's geoheritage, demonstrating that geoheritage is not simply the rocks we see. It is an evolving view of geology shaped by the people who see it. The Monuments and Relics Act (Laws of Malawi Chapter 29:01) provides for the conservation and preservation of, among other things, sites and objects of geological, anthropological and palaeontological interest. The Mwesia Beds qualify for this legal protection.

Special thanks to Benjamini Sikapite, who discovered the fossil and took it upon himself to take the rock he recognized as special to the Cultural and Museum Centre Karonga. We thank Zerina Johanson and Emma Bernard of the Natural History Museum, London, Nicholas Fraser of the National Museums Scotland, and Joy Wheeler of the Royal Geographic Society. We thank Christopher R. Scotese for providing us with the raw data for his 160 Ma palaeogeographical map showing the westward incursion of the sea to the Congo. Aaron Pan provided information about modern ecology. We are grateful to Sherene James-Williamson, Roger Scoon, and the editors for their constructive reviews and comments.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

EGC: conceptualization (equal), writing – review & editing (equal); LLJ: conceptualization (equal), writing – original draft (lead); YMJ: conceptualization (equal), writing – review & editing (equal); MLP: visualization (equal), writing – review & editing (equal); MJP: data curation (lead), software (lead), visualization (lead), writing – review & editing (equal); HHS: data curation (equal), investigation (lead), writing – review & editing (equal); DPV: visualization (equal), writing – review & editing (equal); DAW: writing – review & editing (equal).

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Inquiries concerning raw CT data prior to full publication should be addressed to MJP.