Decadal and century-scale climate variability in tropical Africa during the past 2000 years

Abstract

Holocene climate in high-latitude regions of the world has been relatively stable compared to glacial climates. In contrast, tropical Africa and other low-latitude continental regions were marked by a succession of millennium-scale wet and dry episodes, separated by rather abrupt transitions (Gasse and Van Campo 1994; Lamb et al. 1995; Gasse 2000). These continent-wide fluctuations in the balance of rainfall and evaporation must somehow have resulted from large-scale variation in the position or intensity of large-scale tropical monsoon systems, but their relationship to Holocene climate variability in extra-tropical regions (e.g., Bond et al. (2001)) and the likely mechanisms of external climate forcing are only just beginning to be revealed (Gupta et al. 2003; Hoelzmann et al., this volume). Compared to this marked hydrological instability of African climate during the early and middle Holocene, the last 2000 years have commonly been thought of as rather stable and uneventful. This idea can be traced back to the first reviews of late-Quaternary vegetation and lake-level change in Africa (Butzer et al. 1972; Livingstone 1975; 1980; Hamilton 1982), which in their focus on the prominent late-Glacial and early-Holocene events found littleworth mentioning in the late Holocene. Lowtime resolution and poor age control meant that the last 2000 years were typically represented by just a few data points floating on an interpolated section of the time line. In addition, evidence for 20th-century landscape disturbancewas often missing because soft surface muds had not been recovered, or had been discarded. This lack of a reference frame for signatures of pre-modern human impact, together with the assumption of relative climatic stability, helped perpetuate among palaeoecologists, archaeologists, and geomorphologists the paradigm that most evidence for vegetation and landscape change in tropical Africa younger than 2000 years is due to human activity (Taylor 1990; Jolly et al. 1997; Eriksson 1998).