The origins of herding practices in southern Africa remain controversial. The first appearance of domesticated caprines in the subcontinent is thought to be c. 2000 years BP; however, the origin of this cultural development is still widely debated. Recent genetic analyses support the long-standing hypothesis of herder migration from the north, while other researchers have argued for a cultural diffusion hypothesis where the spread of herding practices took place without necessarily implicating simultaneous and large population movements. Here we document the Later Stone Age (LSA) site of Leopard Cave (Erongo, Namibia), which contains confirmed caprine remains, from which we infer that domesticates were present in the southern African region as early as the end of the first millennium BC. These remains predate the first evidence of domesticates previously recorded for the subcontinent. This discovery sheds new light on the emergence of herding practices in southern Africa, and also on the possible southward routes used by caprines along the western Atlantic coast.

The advent of domestication is a major step that transformed the subsistence strategies of past human societies. In Africa, domestic caprines (sheep and goat) were introduced in the northeastern part of the continent from the Near East more than 9000 years ago. However, their diffusion southwards was slow. They are thought to have made their first appearance in the southern part of the continent ca. 2000 years ago, at a few Later Stone Age sites, including Leopard Cave (Erongo region, Namibia), which provided the oldest directly dated remains assigned to sheep or goat on the basis of morphology of bones and teeth. However, similarities in morphology, not only between these two domesticated caprine species, but also between them and the small wild antelopes, raised questions about the morphological species attribution of these remains. Additionally, the high fragmentation of the site's osteological remains makes it difficult to achieve species-level taxonomic identification by comparative anatomy. In this paper, we report molecular species identification of the Leopard Cave remains using palaeoproteomics, a method that uses protein markers in bone and tooth collagen to achieve taxonomic identification of archaeological remains. We also report new direct radiocarbon dates. Wild antelope remains from museum collections were used to enrich the available protein record and propose de novo type I collagen sequences. Our results demonstrate that the remains morphologically described as domesticates actually belong to a wild antelope species and that domestic caprines first appeared at Leopard Cave 1500 years later than previously thought. This study illustrates that the use of palaeoproteomics coupled with direct radiocarbon dates is particularly suited to complement classic zooarchaeological studies, in this case concerning the arrival of the first herding practices in arid environments. Understanding how past human populations interacted with their environment, and particularly with other animals, allows understanding large parts of societies, their organisation and their economy. Climatic change is one of the factors that may have pressured human populations to adapt their subsistence strategy and diet. With Saharan and Sahelian aridification starting at the end of the African Humid Period (ca. 14,800-5500 years before present [BP] 1), Africa experienced its last major climatic change 2 and humans had to adapt to new environments 3. Thus, the inception of domestication may have been constrained by fewer food resources and their increased unpredictability 4. Given that they have no wild representatives on the continent, it is now commonly accepted that domestic sheep (Ovis aries) and goat (Capra hircus) were introduced from the Near East 5,6. The first archaeozoological evidence of these species dates from the end of the 9th and the beginning of the 8th millennium BP, in the eastern part of the Sahara 7-13. Yet, their diffusion across the continent was slow: the first

In southern Africa, Middle Stone Age sites with long sequences have been the focus of intense international and interdisciplinary research over the past decade (cf. Wadley 2015). Two techno-complexes of the Middle Stone Age—the Still Bay and Howiesons Poort—have been associated with many technological and behavioural innovations of Homo sapiens. The classic model argues that these two techno-complexes are temporally separated ‘horizons’ with homogenous material culture (Jacobs et al.2008), reflecting demographic pulses and supporting large subcontinental networks. This model was developed on the basis of evidence from southern African sites regarded as centres of subcontinental developments.