MSA from Karungu, Kenya and the patchy landscape of Late Pleistocene Africa

J. Tyler Faith and colleagues report in the current Journal of Human Evolution on their work understanding the context of the Middle Stone Age archaeological deposits from Karungu, Kenya. Karungu is an area on the eastern shore of Lake Victoria, with many Pleistocene exposures that contian archaeological evidence. Faith and colleagues set out to reinvestigate archaeological sites that were first explored during the 1930s and then later in the 1980s, to hopefully determine where they fit in time and context of other East African MSA occurrences.

The article presents paleoenvironmental data including the faunal list, species abundances of different faunal elements, and some stable isotope data on the fauna, all of which reflect an arid grassland paleoenvironment. That’s a contrast to today’s local environment in the Lake Victoria basin, which is woodier, including a range of bushlands and forests. The team were able to narrow down the timing of the archaeological deposits to some extent, placing them between approximately 92,000 and 45,000 years ago, the later part of the MSA in this part of Africa.

There are two really interesting aspects to Faith and colleagues’ presentation of the data, that together add up to a hypothesis about the movement and contacts of people during late MSA times in eastern Africa. First is a series of observations that suggest that Lake Victoria was much reduced during the time of the Karungu deposits, with lots of herbivores adapted to an arid grassland environment. Second is a statistical link between the archaeological assemblages at Karungu and other sites further to the north. When combined, Faith and colleagues argue that these data suggest north-to-south dispersal of archaeological patterns along with the spread of arid grasslands during the Late Pleistocene. The timing is very interesting: At or shortly after modern humans had established populations in West Asia, some populations of MSA people may have been expanding to the south through East Africa.


The Karungu sites present abundant evidence of now-extinct grassland-adapted species. These include extinct relatives of modern blesbok, impala, wildebeest and oryx. As Faith and colleagues discuss, several of these species were thought to have become extinct more than 400,000 years earlier:

Previous research on late Quaternary fossil assemblages from East Africa indicated that—with a few minor exceptions (Marean and Gifford-Gonzalez, 1991 and Marean, 1992)—an essentially modern faunal community was in place by ∼400 ka (Potts et al., 1988 and Potts and Deino, 1995). However, the emerging evidence from the Lake Victoria Basin reveals the long-term survival of archaic lineages thought to have disappeared >500 ka, including Kolpochoerus, together with the many extinct bovids characterized by large body mass or exceptional hypsodonty, including S. antiquus, Megalotragus, D. hypsodon, R. atopocranion, and Aepyceros sp. nov. ( Faith et al., 2011, Faith et al., 2012, Faith et al., 2014 and Faith, 2014). Of these, only S. antiquus and D. hypsodon are known from other Late Pleistocene faunal assemblages in East Africa ( Marean and Gifford-Gonzalez, 1991, Marean, 1992, Faith et al., 2012 and Rowan et al., 2015).

In addition to these extinct species, the faunal community is further dominated by grassland species that still exist today, including some that have geographic ranges fairly far from the Lake Victoria region. They further note that Mfangano Island has Pleistocene fauna that may date to 80,000 years ago that similarly suggest an open grassland and a connection to the mainland, which would have required at least a 25 m reduction in the lake level. Put together, these pieces of evidence suggest that sometime between 100,000 and 50,000 years ago, the Lake Victoria region was a grassland mecca hosting a mix of species no longer found anywhere in Africa.

This raises a question: Was the Lake Victoria area a refuge for these grassland species into the later Middle and Late Pleistocene? Or have the extinct species been missed at other sites because of small faunal samples? Faith and colleagues do not exclude either of these possibilities. Considering the evidence for substantial fluctuations in African rainfall and lake levels, it seems unlikely to me that the Lake Victoria region was a long-term refugium; those animals must have been living somewhere else during Pleistocene times that rainfall and lake levels were higher.

It is a sobering message that the sampling of African paleoecology during the Middle Pleistocene is so sparse that we have missed many large herbivores in the period following 500,000 years ago. Even more sobering is the implication that previous paleoecologists and archaeologists have promoted the idea of a large-scale faunal turnover which in reality merely marks local environments and sparse sampling.

MSA archaeology

Ancient humans seem to have been effective at dispersing within this arid grassland. Faith and colleagues compared the archaeological assemblage from Karungu with other sites across eastern Africa, keeping track of sites both above 5 degrees North latitude and at the equator and further south. They find that these sites contrast in the pattern of assemblage composition in a way that reflects latitude, with the more northern sites contrasting with southern sites. The Lake Victoria sites, Karungu and Rusinga, are outliers in their similarity to sites further to the north.

Our analysis of the Late Pleistocene MSA record from East Africa reveals previously unrecognized north–south variation in assemblage composition (Fig. 10), paralleling the geographic patterns observed in the genetic records of ungulates (Lorenzen et al., 2012) and other vertebrates (Dehghani et al., 2008 and Miller et al., 2011). We interpret these differences, reflecting to some extent the variable occurrence of bipolar cores, anvils, large bifaces, and Levallois points or point cores (Table 5), as potentially indicating the development of regionally distinct behavioral patterns during past episodes of population fragmentation, for which the equatorial dispersal barrier is a probable driver. In agreement with Cowling et al. (2008), this implies that the potential for north–south human dispersals across East Africa would have been maximized during climate phases that promoted an expansion of grassland cover.
The grassland-associated MSA assemblages from Karungu and nearby Rusinga Island are characterized by a combination of artifact types that is more typical of sites found further north of the equator. This may reflect the southward dispersal of northern behavioral repertoires during a grassy phase that facilitated dispersals, including those dispersals of northern ungulates such as Grevy's zebra and white rhinoceros (Fig. 8).

The “equatorial dispersal barrier” here refers to the forest corridor that expands across the African equatorial region during wetter periods of time. Although this forest serves as a major dispersal corridor for species that depend upon it, including the great apes, it has been conceived as a barrier for grassland-adapted species.

Faith and colleagues help to show that different patterns of material culture existed within an area where they had not previously been noted in this way. In other parts of Africa, long-term cultural differences during the MSA have long been apparent. These are the traces of ancient populations that were to some extent biologically differentiated, but of whom we have very little evidence from physical appearance. We have some evidence from the genetics of living Africans that highly-differentiated populations once coexisted, and later mixed. It is not evident from genetics whether those populations existed over large parts of the continent or whether they were more localized. Nor is it apparent how many such populations there may have been.

It is therefore very significant that Faith and colleagues are able to show long-distance connections across the grassland environment of East Africa. The connections between the Lake Victoria sites and other MSA sites to the north are one piece of evidence for such connections. Another is the presence of exotic obsidian in the archaeological assemblage from Karungu, which they show is most likely to have come from the Rift Valley 250 km to the east. Another suggestion of contacts with the Rift Valley comes from a third outlier in the north-south dichotomy of archaeological pattern, the site of Prolonged Drift, which lies near Lake Nakuru in the Kenyan rift. Humans were able to exploit a changing climate, even as it became more arid. They maintained long-distance trade and exchange of ideas across distances of hundreds of kilometers.

Maybe this population was mixing more with people who had formerly been more isolated and differentiated from each other.

Or maybe this apparent expansion of northern technical patterns merely represents one layer among a series of flows of populations through the region during the Late Pleistocene. I am hesitant to accept the notion of an “equatorial dispersal barrier” as applied to human populations generally. Even as applied to herbivores, the concept has problems, as the interruption in gene flow caused by spreading forests is relatively short compared to the time that grazing species have existed. With humans, the problem is that dispersal is not limited to grasslands. Some human cultures elsewhere seem to have been very effective at dispersing along ecotones, edge habitats between savanna and woodland.

As an example, today’s central African Pygmy peoples descend from ancient populations that had already become established by 50,000 years ago. Many have speculated that the ancestors of today’s Pygmy peoples, established by 50,000 years ago, were forest-adapted people from their earliest times. If so, the habitation of some of the forested parts of central Africa may have been well underway by the time of the Karungu sites. A continuous forest may be a barrier to the dispersal of people with a grassland cultural commitment, but its edges and even its center may have been prime dispersal corridors for other cultural groups.

Culture itself may have been the more formidable barrier to human dispersal. In earlier phases of human evolution, it is possible to imagine that the store of cultural knowledge within any given group would have been fairly low. A population may have expanded to exploit a new habitat without abandoning much of an investment in the old habitat, and the ability of a group to expand or displace other people within a habitat may have relied only in minor ways upon their inherited cultural knowledge. By contrast, the MSA shows signs of greater differences between groups in material culture, suggesting a deeper store of cultural knowledge. Human groups entering new environments without existing human populations would have been able to expand rapidly, but groups entering a previously-occupied habitat niche would face a cultural deficit that would be hard to overcome.

If the dispersal of human populations during the MSA appears to have followed habitat gradients, it may be ultimately for cultural reasons. That scenario is not one in which a mobile grassland population would necessarily have connected other populations that had previously been more isolated. It is a scenario in which an unstable balance of culture areas may have been fluctuating as climate oscillations caused some ecologies to expand and others to contract continent-wide. And again, the sobering reality is the proportion of Africa that does not contribute to our current knowledge of this MSA cultural variability.


Faith JT et al. (2015) Paleoenvironmental context of the Middle Stone Age record from Karungu, Lake Victoria Basin, Kenya, and its implications for human and faunal dispersals in East Africa. Journal of Human Evolution 83:28-45. doi:10.1016/j.jhevol.2015.03.004