Application of transfer functions to Indian Ocean planktonic Foraminifera

Abstract

The distribution and abundance of planktonic Foraminifera from the Indian Ocean are used to illustrate geographic variations in faunal assemblages in the plankton and on the seabed caused by sedimentary and postdepositional processes and to analyze the effect of these variations on paleoecological reconstruction. Principal components analysis of these data describes the composition and distribution of faunal assemblages in plankton-tow samples, low-dissolution core-top samples, and high-dissolution core-top samples. Factor-comparison analysis describes the relationships among these three sets of assemblages: The species composition of low-dissolution faunal assemblages may be accurately described as a simple linear mixing of plankton assemblages. The geographical distributions of the faunal assemblages in the sediments, however, are often displaced equatorward of their counterparts in the plankton. Dissolution causes complex changes in the composition of faunal assemblages and produces an equatorward displacement of several high-dissolution assemblages relative to their counterparts in low-dissolution sediments. Three transfer functions, or equations, are derived using plankton, low-dissolution, and high-dissolution data. Numerical experiments indicate that transfer functions lose accuracy when applied to discordant data sets: The plankton transfer function often underestimates temperatures in core-top sediments, and the low-dissolution transfer function underestimates temperatures in high-dissolution sediments. These systematic differences in temperature estimates are illustrated by applying the three transfer functions to downcore samples representing conditions 18,000 years ago. Other experiments indicate that these distortions can be reduced by using larger size fractions and calibrating transfer functions with both low- and high-dissolution core-top samples.