The late Quaternary landscape at Sehonghong in the Lesotho highlands, southern Africa PETER J. MITCHELL* In the rough and rugged country of the Lesotho highlands, rock-paintings and archaeological deposits in the rock-shelters record hunter-gatherer life-ways; at Sehonghong, a long sequence runs from recent times to and through the Last Glacial Maximum. Survey of the region’s Middle and Later Stone Age sites shows a pattern of concentrations that likely applies to other parts of the Lesotho highlands. Interest in landscape archaeology, particularly issues of subsistence strategies, resource exploi- tation and the spatial patterning of sites and artefacts, is longstanding in southern African Later Stone Age (LSA) studies; the last decade has witnessed an increasing concern with so- cial, rather than purely ecological, explanatory models (Maze1 1989). Not surprisingly, the ways in which prehistoric people made use of the land has been central to most archaeological research in Lesotho, encouraged by the marked spatial patterning in resource availability that characterizes it and adjacent parts of South Africa (Mitchell 1992). The first systematic archaeological research programme in Lesotho (FIGURE 1) explicitly lo- cated itself within Clarke’s (1972) environmental paradigm, emphasizing the seasonal movement of people between ecologically complementary areas (Carter 1970; 1978). Occupation of Leso- tho’s southeastern highlands, an area of rug- ged and heavily dissected topography (FIGURE 2), was argued to have been a summer event associated with an upsurge in painting and related rituals by larger social groups making use of rock-shelters. The lower-lying thornveld of KwaZulu-Natal would have been occupied in winter by smaller social groups, living mainly in open sites. Congruent with 19th-century observations on the territories of individual Bushman groups (Vinnicombe 1976), the Senqu (Orange) River Valley was envisaged as a sepa- rate exploitation system, focused around win- ter and summer occupation of two large rock- shelters (Melikane and Sehonghong). Cable (1984) developed these ideas in a study of the recent Holocene LSA of southern KwaZulu- Natal that provided additional detail on the seasonal availabilities of specific plant and animal resources. Both Carter (1978) and Ca- ble (1984) suggested that spatial patterning in lithic assemblage composition was activity-, and thus seasonally, related; they argued for changes in LSA resource exploitation strategies around the 3rd century AD, in response to the settle- ment of Iron Age agropastoralists in the KwaZulu-Natal lowlands. Another element of chronological variability was introduced with the proposition that under stadia1 conditions the Senqu Valley and highland sourveld of southeastern Lesotho were used only in sum- mer (Carter 1976). The focus of analysis and explanation in these studies was located firmly at the regional level; comparatively little attention was paid to patterning within ecological zones, or close to excavated sites. With a pressing need for a chronostratigraphic framework, excavation concentrated on large shelters, rather than smaller shelters and open-air sites. One of these sites, Sehonghong Shelter (FIGURE 3) has been shown to have a long sequence of occupation, associated with high-quality organic preserva- tion, extending well back into the Upper Pleisto- * Pitt Rivers Museum, University of Oxford, 64 Banbury Road, Oxford ox2 ~PN, England. Received 22 February 1996, accepted 19 April 1996, revised 6 May 1996. ANTIQUITY 70 (1996): 623-38