FORAMINIFERAL DISSOLUTION AT SHALLOW DEPTHS OF THE WALVIS RIDGE AND RIO-GRANDE RISE DURING THE LATEST CRETACEOUS - INFERENCES FOR DEEP-WATER CIRCULATION IN THE SOUTH-ATLANTIC

A high-resolution (10-20 kyr) record of variations in CaCO3 content an d dissolution was established for latest Cretaceous (last 0.7 Myr) dee p-sea sediments from the South Atlantic Ocean(DSDP Site 516 from the R io Grande Rise, and sites 525 and 527 from the Walvis Ridge). The degr ee of fragmentation of planktonic foraminifera (DFP) was used as a mea sure of calcite dissolution. High negative correlations between DFP an d other independent measures of carbonate dissolution (percentage of s and fraction, absolute abundance of planktonic foraminifera. and plank tonic/benthic foraminiferal ratio) validate its use as a sensitive ind ex of calcite dissolution in upper Maastrichtian deep-sea sediments. V ery high DFP and a significant negative correlation between DFP and Ca CO3 content suggest that Site 516 was located below the foraminiferal lysocline during the entire interval studied. Such a shallow position of the lysocline (paleodepth of Site 516 was 1.2 km) may be explained by ''upwelling'' of corrosive deep waters along the southern margin of the Rio Grande Rise. Sites 525 and 527 were located above the foramin iferal lysocline; however, three short periods of enhanced dissolution were recognised al Site 525 (paleodepth 1 km) and one interval of str ong dissolution was identified at Site 527 (paleodepth 2.7 km). The la ck of correspondence between the dissolution regimes al sires from the Walvis Ridge suggests limited deep-water communication across this ph ysiographic barrier. Two of the dissolution maxima recognised at Site 525 correspond to carbonate maxima at Site 527. Variations in ''upwell ing'' intensity along the Walvis Ridge, resulting in fluctuations in p rimary productivity in this area, may be the proximal cause of both ca rbonate cycles at Site 527 and dissolution cycles at Site 525. We sugg est that development of the bottom Ekman layer between a hypothetical westward geostrophic current and the topographical height of the Rio G rande Rise-Walvis Ridge system may be a plausible hydrodynamical expla nation for the proposed ''upwelling'' along the southern margin of thi s topographical structure.