REPORTS 27. E. Canales, E. J. Corey, J. Am. Chem. Soc. 129, 12686–12687 31. D. J. Robins, D. S. Rycroft, Magn. Reson. Chem. 30, Supplementary Materials (2007). 1125–1127 (1992). www.sciencemag.org/content/342/6160/840/suppl/DC1 28. P. Karrer, F. Canal, K. Zohner, R. Widmer, Helv. Chim. Acta 32. D. Gray, T. Gallagher, Angew. Chem. Int. Ed. 45, Supplementary Text 11, 1062–1084 (1928). 2419–2423 (2006). Figs. S1 to S8 29. M. Hajri, C. Blondelle, A. Martinez, J.-L. Vasse, NMR Spectra J. Szymoniak, Tetrahedron Lett. 54, 1029–1031 Acknowledgments: This work was supported by the Deutsche HPLC Spectra (2013). Forschungsgemeinschaft (Graduiertenkolleg GRK 1626 References (33–42) 30. G. Adamson, A. L. J. Beckwith, M. Kaufmann, A. C. Willis, Chemical Photocatalysis) and by the Fonds der Chemischen J. Chem. Soc. Chem. Commun. 1995, 1783–1784 Industrie (scholarship to R.B.). We thank O. Ackermann for 16 August 2013; accepted 16 October 2013 (1995). help in conducting HPLC analyses. 10.1126/science.1244809 rate of 32 cm per thousand years (supplementary Abrupt Shifts in Horn of Africa materials). We used the hydrogen isotopic composition d Hydroclimate Since the Last of leaf waxes ( Dwax) as a proxy for aridity and, more generally speaking, hydroclimate, includ- ing precipitation/evaporation balance and changes Glacial Maximum d in regional convection. Dwax has been widely used in African paleoclimate and is an effective Jessica E. Tierney1* and Peter B. deMenocal2 indicator of changes in the isotopic composition of precipitation (dDP) and aridity, with enriched The timing and abruptness of the initiation and termination of the Early Holocene African isotopic values corresponding to drier conditions Humid Period are subjects of ongoing debate, with direct consequences for our understanding and depleted values to wetter conditions (10, 22). of abrupt climate change, paleoenvironments, and early human cultural development. Here, More generally, tropical water isotopes are good we provide proxy evidence from the Horn of Africa region that documents abrupt transitions into tracers of large-scale changes in atmospheric cir- and out of the African Humid Period in northeast Africa. Similar and generally synchronous culation (18, 23) and therefore reflect regional, abrupt transitions at other East African sites suggest that rapid shifts in hydroclimate are a rather than local, shifts in the hydrological cycle. regionally coherent feature. Our analysis suggests that the termination of the African Humid Because Congo basin moisture is effectively Period in the Horn of Africa occurred within centuries, underscoring the nonlinearity of the blocked by the Ethiopian highlands and the region’s hydroclimate. Horn of Africa receives the majority of its rain- fall from the Indian Ocean (24), we interpret the uring the Early Holocene epoch between be smoothed because of mixed contributions dDwax values to primarily represent changes in roughly 11 to 5 thousand years ago (ka), from distal terrains (2). Alternatively, there may western Indian Ocean hydroclimate. d Dthe presently hyperarid Saharan desert be regional heterogeneity in both the timing and The Dwax record from the Gulf of Aden in- was dotted with large and small lakes, savannah duration of the AHP termination, reflecting the dicates that Horn of Africa hydroclimate has grasslands, and in some regions, humid tropical variable sensitivity of different regions to certain changed dramatically during the past 40,000 years forests and shrubs (1, 2). This “African Humid feedback mechanisms (in particular, vegetation (Fig. 2). After the arid conditions of the Last Period” (AHP) was a unique hydrological regime feedbacks) (3, 4, 6, 15, 16). Glacial Maximum (LGM) (26 to 19 ka), the Horn and has been a focal point of African paleoclimate East Africa and the Arabian Peninsula also region experienced a severe dry period coincident studies, both for its climatological implications experienced humid conditions during the Early with the North Atlantic cooling event, Heinrich (3, 4) and its influence on the emergence of phar- Holocene (17, 18). Speleothem d18O data from Event 1 (H1) (Fig. 2), which is consistent with aonic civilization along the Nile (5, 6). The fun- southern Oman (Qunf Cave) and dust strontium previous proxy (25) and model (23) evidence damental cause of the AHP—dramatic increases in isotopes off of Somalia suggest a gradual attenua- summer precipitation triggered by orbital forcing tion of humid conditions during the Holocene, of African monsoonal climate and amplified by much like the eastern Saharan pollen data (19, 20). oceanic and terrestrial feedbacks—is well under- These observations have led to the suggestion that stood (7, 8). However, the abruptness with which the eastern Sahara and northeast Africa experi- the AHP began and, most particularly, ended is enced a gradual end to the AHP (3, 4, 12, 15, 21) still debated. Dust proxy data from the west coast and that abrupt responses were therefore limited of Africa indicate a rapid, century-scale termina- to the western Sahara. tion of the AHP near 5 ka (9). In contrast, isotopic We revisited the timing and abruptness of proxies from central Africa (10, 11) and pollen and transitions into and out of the AHP in northeast sedimentological data from a lake in the eastern Africa using a new record of hydroclimate from Sahara (12, 13) suggest a more gradual reduction a key, yet previously understudied, region: the in rainfall during the mid-Holocene tracking the Horn of Africa. This record is derived from a ma- orbital decline in boreal summer insolation. The rine core (P178-15P) located in the Gulf of Aden discrepancy remains unresolved. Previous studies (Fig. 1). The Gulf of Aden receives substantial have attributed the difference in climate response amounts of terrestrial material during the sum- to differing proxy sensitivities; for example, dust mer monsoon season, when prevailing south- may respond nonlinearly to a gradual drying of westerly winds transport dust from the Horn the Sahara (14), and conversely, pollen data may (Fig.1andfig.S1).Therefore,theterrestrialcom- ponents (including organic matter) in the sedi- Fig. 1. A map of East Africa. The map includes 1Woods Hole Oceanographic Institution, 266 Woods Hole Road, ments predominantly reflect conditions in the Horn topography, wind climatology for June-July-August Woods Hole, MA 02540, USA. 2Lamont Doherty Earth Observ- and Afar regions (supplementary materials). Twenty (JJA) (46), the location of the study site (Gulf of Aden atory, Palisades, NY 10964, USA. radiocarbon dates constrain the chronology of P178-15P; 11° 57.3′ N, 44° 18' E, 869 m water depth), *Corresponding author. E-mail: [email protected] P178-15P and indicate an average sedimentation and other sites mentioned in the text. www.sciencemag.org SCIENCE VOL 342 15 NOVEMBER 2013 843 REPORTS from the Afro-Asian monsoon domain. After both cases, we accepted the null hypothesis if has evolved through time. Although our site sits H1, there was a rapid transition to intermediate P >0.05. outside of the upwelling zone, it is still likely that conditions coincident with the Bölling-Allerød Several findings emerged from this analysis. the regional radiocarbon reservoir was modulated (B/A) period and then a reversal into dry condi- First, we found that the termination of H1 and by the intensity of Arabian Sea upwelling, espe- tions during the Younger Dryas (YD), another the YD are not likely synchronous at all sites. cially during the deglaciation, when large changes North Atlantic cold event (Fig. 2). Upon the ter- All age model iterations indicate an older timing mination of the YD, the Horn of Africa rapidly for the H1–B/A transition at Lake Tanganyika 475 moved into the most humid conditions of the past than at the Gulf of Aden site [Lake Challa does 15°N JJA (W/M2) 40,000 years, coincident with the AHP (Fig. 2). not have a detectable H1 termination (supplemen- 470 These conditions persisted until ~5 ka. tary materials)]. The median date of the transi- 465 460 The dDwax record from the Gulf of Aden is tion in the Gulf of Aden record [14,680 years dominated by abrupt transitions that occur more before the present (B.P.)] is close to the generally 455 rapidly than would be predicted from orbital forc- accepted timing of 14,700 years B.P. (29), where- 450 ing alone (Fig. 3). Two other dDwax records from as the transition at Lake Tanganyika occurs nearly 445 East Africa—from Lake Tanganyika (22) and 1000 years earlier (median = 15,760 years B.P.). −150 ADEN 440 Lake Challa (26)—show similar and generally For the Challa and Tanganyika sites, the termi- −145 coeval abrupt transitions at the end of H1, the YD, nation of the YD is likely synchronous (t test, −140 and the AHP, even though these sites sit ~2000 km P = 0.99) and the timing (median of both = −135 to the south of the Gulf of Aden (Figs. 1 and 3). 11,600 years B.P.) is in good agreement with −130 The overall similarity between the three records the transition (11,570 years B.P.) dated by tree- −125 attests to the ability of water isotope proxies, spe- ring chronologies (30), but the transition in the −120 d cifically Dwax, to record large-scale hydroclimatic Gulf of Aden record occurs later in all iterations −115 TANGANYIKA patterns in tropical Africa. (median = 10,850 years B.P.). −110 −130 The perceived timing and abruptness of the The observed offsets in the Tanganyika and −105 transitions in each dDwax record is subject to dating Aden dDwax data across the H1 and YD termi- −120 uncertainties as well as sedimentary properties. nations, respectively, could reflect meaningful Therefore, we used a Monte Carlo method (27)to local climatic deviations; however, because these −110 create empirical probability distributions for the millennial-scale events are remotely forced by −100 midpoint and duration of each identified transi- North Atlantic processes, it seems more likely that tion and more clearly assess the duration and they reflect unconstrained changes in site radio- −90 14 synchronicity of these key climate transitions (sup- carbon ( C) reservoirs.