PALEOCEANOGRAPHY, VOL. 8, NO. 4, PAGES 527-547, AUGUST 1993 ON STABLE ISOTOPIC VARIATION AND EARLIEST PALEOCENE PLANKTONIC FORAMINIFERA Steven D'Hondt Graduate School of Oceanography, University of Rhode Island, Narragansett James C. Zachos Earth and Marine Sciences, Universityof California,Santa Cruz Abstract. Extantplanktonic foraminifera display positive covariance between •513C signals and test size.Asdocumented byother studies, primary causes of increased •513C values with increased test size may include increased reliance on ambient CO2 for calcification at larger test sizes, decreased kinetic fractionation during calcification at larger test sizes, and increased photosymbiotic activity in larger symbiont-bearing planktonic foraminifera. Planktonic foraminiferal •5180 values also oftencovary with testsize,although thedirection of this covariance is taxon dependent. Possible explanations for relationships between •5180 signals and test size include changing habitat depth overontogeny, correlations between adulttest size and environmental conditions, and changing isotopic disequilibrium with size,ontogenetic stage, or photosymbiont density.In order to assess themagnitude and implications of similar size dependence in earliest Paleocene planktonic foraminifera, we measured thestable isotopic signals of multiplesizefractions of 10 earliest Paleocene species. All of these taxaexhibit a strong positive correlation between •513C and test size. Theslope and magnitude of this trend varies between species, with Woodringina claytonensis displaying the largest shift (1.1%0 over a 130 gm range in mean sieve size)andGuembelitria cretacea displaying the smallest (0.2 %0 over a 38 gm range). By analogy with modern planktonic foraminifera, thisgeneral relationship between •513C and size probably resulted from increased reliance on ambient CO2 for calcification at larger testsizes. The high magnitude of this shiftin some taxamay reflect either photosymbiotic enhancement of thegeneral trend or relativelygreater changes in the proportions of metabolic and ambient CO2 used for calcification at different testsizes. Copyright 1993 by the American Geophysical Union. Paper number 93PA00952. 0883-8305/93/93 PA-00952510.00 Failureto account for relationships between testsizeand •513C signals can lead tounderestimation of early Paleocene surface ocean •513C values by 1%0 ormore. These size-related •513C effects provide an alternative explanation for decreases in whole-rock •513C values and some decreases in planktonic-to- benthic foraminiferal •513C gradients documented atmarine K/T boundary sequences. At all sizefractions, the 10 Paleocene taxadisplay a very limitedinterspecies range of •5180 derived paleotemperatures. Despite this limited range, paleobiogeographic patterns and •5180 signals appear to provide realistic estimates of relative paleodepth andseasonal affinities of earliest Paleocene planktonic foraminiferal species. Earliest Paleocene •5180 and biogeographic data are consistent with a general trend of surface-to-deep diversification of microperforate planktonic foraminifera following the K/T boundary. Sucha trendmay simplyresult from exploitation of a near-surface open-ocean habitat by the epicontinental K/T survivor G. cretacea. INTRODUCTION Stable isotopic ratios provide a powerful and widelyused tool for reconstructing fossil paleoceanographic and paleoclimatic conditions from fossil marine carbonates. In particular, oxygen isotopic ratios of coccolithophorid and planktonic foraminiferal calcite arecommonly used to estimate sea surface paleotemperature and paleosalinity, whilecarbon isotopic ratios have been used to estimate carbon fluxes between different major carbon reservoirs. For example, oxygen isotopic ratios of marine carbonates have been used to estimate changing sea surface and deepwater paleotemperature across theCretaceous/Tertiary (K/T) boundary [Douglas and Savin, 1973; Boersma and Shackleton,1981; Hsii et al., 1982; Perch-Nielsen et al., 1982; Shackletonet al., 1984a; Williams et al., 1985; Zachos et al., 1985, 1989a, b; Zachos and Arthur, 1986; Oberhansli, 1986; D'Hondt andLindinger, 1988; Keller 527