* Corresponding author. E-mail address: j.dearing@liv.ac.uk (J.A. Dearing). Quaternary International 76/77 (2001) 43}56 Palaeoclimate records from OIS 8.0}5.4 recorded in loess}palaeosol sequences on the Matmata Plateau, southern Tunisia, based on mineral magnetism and new luminescence dating J.A. Dearing*, I.P. Livingstone, M.D. Bateman, K. White Environmental Magnetism Laboratory, Department of Geography, University of Liverpool, Liverpool L69 7ZT, UK School of Environmental Science, University College Northampton, Northampton NN2 7AL, UK Shezeld Centre for International Drylands Research, University of Shezeld, Winter St., Shezeld S10 2TN, UK Department of Geography, University of Reading, Whiteknights, Reading RG6 6AB, UK Abstract Mineral magnetic studies of loess}palaeosol sequences on the Matmata Plateau, southern Tunisia, coupled with a new chronology based on luminescence dating, con"rm the presence of at least four phases of pedogenesis during the period 100}250ka. Inter-site correlations between the reddened fersiallitic palaeosols con"rms that, despite modern gullying processes, the records are regional and repeatable. The palaeosol magnetic signal is controlled by the formation of secondary ferrimagnetic minerals, which may be easily detected by magnetic susceptibility and frequency-dependent susceptibility measurements. Comparison of the magnetic record with global proxy climate records shows a correlation with loess}palaeosol sequences in China and the marine oxygen isotope (OI) record during stages 8.0}5.4. Preliminary attempts to infer palaeoprecipitation levels from modern analogues of soil magnetism-climate associations suggests that during the periods 100}120ka and & 200ka precipitation was '400mm a, compared with modern precipitation ( 150mm a.  2001 Elsevier Science Ltd and INQUA. All rights reserved. 1. Introduction Loess}palaeosol sequences provide the longest terres- trial records of global environmental change. In particu- lar, sequences from the Central Loess Plateau in China span the past 2.5 Ma and many studies have shown that they provide proxy records of glacial}interglacial cli- matic cycles (e.g. Heller and Liu, 1982; Kukla et al., 1988, 1990; Hunt et al., 1995). Mineral magnetic records have helped to correlate the timing of pedogenesis in inter- glacial periods (e.g. Maher and Thompson, 1991; Florindo et al., 1999) to stages in the oxygen isotope record from deep-sea cores (Shackleton et al., 1983). They have also been used successfully to reconstruct palaeo- precipitation levels by comparison with modern soil mag- netism-climate associations (Maher et al., 1994; Liu et al., 1995; Thompson and Maher, 1995). Porter and An (1995) have claimed that the particle-size properties of loess} palaeosol sequences may hold records of wind strength and direction related to Heinrich ice-melting events. However, climate reconstructions for North Africa have not fully utilised these environmental archives. More e!ort has been placed in the study of lake-level changes (e.g. Gasse et al., 1987), marine palaeoecological records (e.g. Brun, 1979; Dupont, 1993) and geomorphic form, such as #uvial deposits (e.g. Rognon, 1987). These di!erent lines of evidence are not always easy to recon- cile. Errors in dating and the precise nature of climate controls on the observed phenomena have led to much debate about the nature of late-Quaternary climates (Rognon, 1987; Fontes and Gasse, 1989), speci"cally, the timing of humid phases during the early and mid-Holo- cene and late-Pleistocene. Palaeoclimate records from North Africa are highly desirable because of the likely sensitivity of the location to N}S movements of the polar front and the relative positioning of high latitude and sub-tropical high pressure centres. More detailed palaeoclimate reconstructions from North Africa poten- tially o!er a means for extending and corroborating links between ice and deep-sea palaeoclimate records from high latitudes and palaeoclimate records from European loess-palaeosol sequences (e.g. Singhvi et al., 1989) and lake sediments (e.g. Thouveny et al., 1994) in low/middle latitudes. 1040-6182/01/$20.00  2001 Elsevier Science Ltd and INQUA. All rights reserved. PII:S1040-6182(00)00088-4