Pergamon Quaternary Science Reviews, Vol. 16, pp. 541-563, 1997. 0 1997 Elsevier Science Ltd. PII: zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA SO277-3791(%)00081-9 All rights reserved. Printed in Great Britain. 0271-3791l97 $32.00 DIATOM-INFERRED SALINITY IN PALAEOLAKES: AN INDIRECT TRACER OF CLIMATE CHANGE FRANCOISE GASSE,” PHILIP BARKER,? PETER A. GELL,$ SHERILYN C. FRITZ3 and FRANCOISE CHALIl?*** *URA 723 CNRS-Laboratoire d’Hydrologie et Geochimie Isotopique, Bat. 504, Univ. Paris Sud, F. 91 405 Orsay Cedex, France (E-mail: fchalie@geol.u-psud.frj TDepartment of Geography, Lancaster University, Lancaster LA1 44B, U.K. SDepartment of Geography and Environmental Sciences, Monash University, Clayton Melbourne, Victoria 3168, Australia §Department of Earth and Environmental Sciences, Lehigh University, 31 Williams Drive, Bethlehem, PA 18015. U.S.A. Abstract - Lakes in arid and semi-arid regions respond to climatic change through shifts in lake water volume and ionic concentration. Because diatom distribution is highly correlated with lake hydrochemistry, diatoms can be used to infer changes in salinity and brine composition and thus to infer past climates. Here we critically examine the use of diatom-inferred salinity as a climate proxy, with examples taken from both modern waterbodies and sedimentary profiles. Sediment records may contain assemblages mixed from periods or sites of differing hydrochemistries because of the high degree of spatial and temporal heterogeneity in saline systems. Dissolution and diagenesis in saline brines may further complicate interpretation of sedimentary assemblages. Furthermore threshold effects, salinity regulation via groundwater seepage, antecedent conditions, and other aspects of local hydrology may modify the relationship between salinity change and climatic forcing. These complexities necessitate critical examination of the ecology and taphonomy of sedimentary assemblages, as well as evaluation of potential non-linearities in the salinity/climate relationship through comparison of diatom-inferred salinity with other proxy records of hydrochemical change, historical documents where they exist, and with other regional lake systems. 0 1997 Elsevier Science Ltd. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA QSR INTRODUCTION The contribution of diatom analysis to the under- standing of both palaeo- and contemporary eimnological processes has increased considerably over the last two decades. This is largely due to the development of statistical methods for inferring lake ecologically im- portant hydrochemical variables (e.g. ter Braak and van Dam, 1989; Birks et al., 1990; Juggins and ter Braak, 1992). Diatoms are potential climate tracers in arid and semi-arid regions where lakes may respond to changes in the precipitation-evaporation (P-E) balance by large variations in water volume and ionic concentration. Although the relationships that link lake salinity with climate are complex, shifts in total dissolved solids (TDS) and hydrochemical facies may be regarded as surrogate climate indicators. Currently, international efforts are being undertaken in the CASPIA program (Juggins et al., 1994) to improve estimates of diatom-inferred salinity for palaeoclimate reconstruction. An important challenge is **Corresponding author. to identify and quantify directional changes at time scales of 10’ to lo’ years, which can be interpreted in terms of climate fluctuations. Unambiguous correlations between modern diatom communities and ionic concentration are well documen- ted from different parts of the world (e.g. Gasse et al., 1983; Ben Khelifa, 1989; Servant-Vildary and Roux, 1990; Fritz et al., 1993; Wilson et al., 1994; Gel1 and Gasse, 1994). For example, the squared correlation coefficients between diatom assemblages of surface sediments and salinity from 55 lakes of the North American Great Plains (Fritz et al., 1991) and between living communities and conductivity from 282 samples from Northern Africa (Gasse et al., 1995) are rz = 0.82 and r2 = 0.87, respectively. Predictive models (transfer functions) are now developed at large regional scales for pH, salinity (S) or conductivity (C), and ionic composi- tion (Gasse and Tekaia, 1983; Roux et al., 1991; Fritz et al., 1991; Cumming and Smol, 1993; Gasse et al., 1995). Merging of regional data sets is in progress, e.g. the North American Great Plains (Fritz et al., 1993) and the African (Gasse et al., 1995) data sets, which may help to 547