In this Article, a processing error led to the wrong versions of Fig. 3 and Extended Data Fig. 4 being published. Figure 3e did not include the entirety of the eastern Africa soil carbonate δ13C database as compiled by ref. 13. Fig. 3 of the original Article has been corrected, and Fig. 1 of this Amendment shows the original and corrected Fig. 3 side by side, for transparency. In the Methods section of the original Article, there are further details about how this record has been produced. The last paragraph of the Methods has been corrected; the original text was: “On the basis of ref. 13, time series of δ13C values from soil carbonate were combined for the Omo-Turkana Basin and the southern Kenyan-Tanzanian sites using their medians, and interquartile ranges using six-data-point bins.” Furthermore, the original version of Extended Data Fig. 4 did not display data from eastern African hominin site Afar; the figure and caption have been updated accordingly, and the original and corrected versions are shown as Fig. 2 to this Amendment. These changes do not alter any inferences drawn from the data. These errors have been corrected in the online version of the Article.

@en

Today, the eastern African hydroclimate is tightly linked to fluctuations in the zonal atmospheric Walker circulation^1,2. A growing body of evidence indicates that this circulation shaped hydroclimatic conditions in the Indian Ocean region also on much longer, glacial–interglacial timescales^3–5, following the development of Pacific Walker circulation around 2.2–2.0 million years ago (Ma)^6,7. However, continuous long-term records to determine the timing and mechanisms of Pacific-influenced climate transitions in the Indian Ocean have been unavailable. Here we present a seven-million-year-long record of wind-driven circulation of the tropical Indian Ocean, as recorded in Mozambique Channel Throughflow (MCT) flow-speed variations. We show that the MCT flow speed was relatively weak and steady until 2.1 ± 0.1 Ma, when it began to increase, coincident with the intensification of the Pacific Walker circulation^6,7. Strong increases during glacial periods, which reached maxima after the Mid-Pleistocene Transition (0.9–0.64 Ma; ref. ⁸), were punctuated by weak flow speeds during interglacial periods. We provide a mechanism explaining that increasing MCT flow speeds reflect synchronous development of the Indo-Pacific Walker cells that promote aridification in Africa. Our results suggest that after about 2.1 Ma, the increasing aridification is punctuated by pronounced humid interglacial periods. This record will facilitate testing of hypotheses of climate–environmental drivers for hominin evolution and dispersal.

@en

The manufacture of geometric engravings is generally interpreted as indicative of modern cognition and behaviour. Key questions in the debate on the origin of such behaviour are whether this innovation is restricted to Homo sapiens, and whether it has a uniquely African origin1. Here we report on a fossil freshwater shell assemblage from the Hauptknochenschicht (‘main bone layer’) of Trinil (Java, Indonesia), the type locality of Homo erectus discovered by Eugène Dubois in 1891. In the Dubois collection (in the Naturalis museum, Leiden, The Netherlands) we found evidence for freshwater shellfish consumption by hominins, one unambiguous shell tool, and a shell with a geometric engraving. We dated sediment contained in the shells with 40Ar/39Ar and luminescence dating methods, obtaining a maximum age of 0.54 ± 0.10 million years and a minimum age of 0.43 ± 0.05 million years. This implies that the Trinil Hauptknochenschicht is younger than previously estimated. Together, our data indicate that the engraving was made by Homo erectus, and that it is considerably older than the oldest geometric engravings described so far. Although it is at present not possible to assess the function or meaning of the engraved shell, this discovery suggests that engraving abstract patterns was in the realm of Asian Homo erectus cognition and neuromotor control.@en