RADIOLARIAN SEDIMENTARY IMPRINT IN ATLANTIC EQUATORIAL SEDIMENTS - COMPARISON WITH THE YEARLY FLUX AT 853 M

Radiolarian specific compositions in a series of 20 sediment trap samples covering an entire year (1.3.1989 to 16.3.1990, collected at 853 m) were compared with bottom (0-1 cm) materials from the same site (eastern equatorial Atlantic: 01 degrees 47.5'N, 11 degrees 07.6'W). Data on mean sediment accumulation rates at the site of the mooring (1.59 g/cm(2)/kyr), mean radiolarian flux at 853 m (28,446 shells/m(2)/day), and abundance in the 0-1 cm bottom layer (48,258 shells/g) suggest that approximately 95% of the radiolarians produced are lost to the fossil record. Sediment trap sample-to-sample correlations (based on relative abundances of 40 radiolarian species present at levels greater than or equal to 1% in at least one sample, mean value, r=0.886) did not differ significantly from correlations between each water column sample and surface sediments (mean r=0.878). Similarities between the flux and the sediments were not associated with time of year and with periods of enhanced radiolarian output. Two taxa had lower, and nine tars had higher percentage contributions in the sediments than in any one sediment trap sample, and a few of the abundant species had averages up to 7 times higher in either the water column or the sediments. These dissimilar percentage loadings are attributed to selective dissolution, lateral subsurface and deep advection of shells from higher-latitude areas, and identification biases. As opposed to species-level inventories, family-level databases (including shells identified to family only) differed significantly between the water column and the sediments. Spumellaria (especially Spongodiscidae) were more abundant in the sediments (35%) than in the water column (19%), while Nassellaria showed the opposite trend (64% and 80%, respectively). It is suggested that ease of identification of spongodiscid fragments and fragility of juvenile nassellarians are responsible for these differences.