In many salt lakes around the world, the relative abundance of preserved authigenic minerals is different from that predicted from solute mass balance calculations. Conventional mass balance models assume that chloride behaves conservatively over long periods of time and fail to take into account the role of diffusion, deflation and fractional crystallisation/dissolution of salts. An alternative approach is to use oxygen isotopes as these reflect directly water molecule rather than solute concentrations and have the added advantage of providing a palaeohydrological record in lacustrine carbonates. We present a steady-state, stable-isotope model, in conjunction with stable isotopic measurements of sub-surface brines, regional groundwaters and carbonate deposits from the Lake Malata-Lake Greenly plays complex in South Australia, to estimate the apparent leakage and palaeoleakage from these superficially closed playa lakes. The steady-state model calculations, using the present delta(18)O and delta(2)H compositions of the lake brines and inflowing groundwater, suggest that the apparent present-day leakage for the complex is 75 to 90% of inflow (similar to 35 times that calculated from Cl- and Br-). Under such conditions, only low magnesian calcite precipitates and the lake water experiences reduced effects of evaporation, gas and vapour exchange and, consequently, reduced isotopic and chemical enrichment. Further, our model shows that calcite becomes increasingly Mg-rich until leakage is reduced to similar to 55 to 70% of inflow - a condition favourable for dolomitisation. delta(18)O and delta(13)C of the lacustrine carbonates show excursions on the order of 5 parts per thousand over the length of a 2.3 m core, indicating thar the lake complex has varied from being throughflow dominated (presence of low Mg-calcite relatively depleted in delta(18)O and delta(13)C) to evaporation dominated thigh Mg-calcite/dolomite relatively enriched in delta(18)O and delta(13)C) throughout late Quaternary. Our estimates of leakage fractions are consistent with the observed mineralogical suite, but there remains a discrepancy between apparent closure indicated by the presence of the highly saline brine reservoir (< 1% leakage) and high rates of throughflow inferred from stable isotope data. We propose that the brine was formed by winter time re-solution of a seasonal halite crust which forms during summer dominated evaporative discharge. Recirculation of the secondary brine, and mixing with regional groundwater may decouple the solute cycle from the water cycle in many playa lakes. The end result is a partially mixed brine characterised by nearly conserved solutes but with isotopic signatures indicative of brine-rainfall-groundwater interactions. (C) 2000 Elsevier Science B.V. All rights reserved.
