The Leg 165 assemblages show very poor preservation, and hence low diversity in the sediments, relative to previously reported assemblages of equivalent age from DSDP Legs 4 and 15 from the Caribbean and DSDP Leg 10 from the Gulf of Mexico. DSDP Leg 4 collected useful Caribbean middle Eocene sequences with radiolarians at Site 29 (Riedel and Sanfilippo, 1970) and Leg 15 provided Caribbean equivalents of the sediment sequences in the Gulf of Mexico (Riedel and Sanfilippo, 1973) which permitted correlation with calcareous microfossils. Leg 10 sampled a similar sequence (Foreman, 1973; Sanfilippo and Riedel, 1973) in the Gulf of Mexico. All of these legs lacked continuous coring from duplicate sites. Subsequent DSDP/ODP Legs (76-78, 96, 100-102, and 110) in the Caribbean region recovered only younger than middle Eocene radiolarian-bearing sediments. Although more numerous Paleocene and lower Eocene samples have been obtained from different parts of the world ocean, the recovery is still intermittent and radiolarian preservation commonly not adequate for detailed biostratigraphic work (Sanfilippo and Nigrini, 1998b). Other studies from this region have reported both very poor (Sanfilippo and Hull, in press) and relatively good (Florez Abin, 1983) preservation in Cuban land-based sections. Good preservation of Eocene material from Barbados was reported by Saunders et al. (1985), but preservation in other sections from Barbados (Sanfilippo, unpubl. data) is inconsistent and generally deteriorates with age.
Sporadically occurring well-preserved radiolarians in the Leg 165 sites (e.g., 165-999B-28R-1, 0-2 cm) represent assemblages characteristic of open-ocean conditions, but with notably less diversity. However, the following genera, which are of particular stratigraphic importance in low latitudes, are notably sparse or absent from the three sites investigated herein: Podocyrtis, Thyrsocyrtis, and Theocotyle.
Not only do we see anomalous occurrences of well-preserved and poorly preserved assemblages within the region, we have also noted that, within a given site, and even with a given sample, the state of preservation is variable (Pl. 2, figs. 14-17; Pl. 3, figs. 11-13). For example, in Hole 999B the upper Eocene and Oligocene assemblages are composed of common, very poorly preserved radiolarians together with time-equivalent very rare but well-preserved forms, whereas the middle Eocene assemblage is characterized by abundant, very poorly preserved radiolarians that show evidence of strong dissolution. The upper Paleocene and lower Eocene assemblages are mostly barren of radiolarians except for sporadic occurrences of abundant, very poorly preserved forms. This observation suggests that the absence of siliceous microfossils in sediments from this region is a result of dissolution within the sediment rather than the absence of specimens in the surface water.
The majority of our samples contain dissolved (Pl. 1, figs. 3, 9) and/or infilled radiolarian fragments or molds (Pl. 1, fig. 4). Frequently, diagnostic morphological features are missing (Pl. 1, figs. 5-7, 11), making identifications uncertain or impossible. In many of the nassellarians, dissolution is concentrated in the collar region (Pl. 2, fig. 11), which often results in the loss of the cephalis (Pl. 1, fig. 12). Even when specimens are sufficiently complete for identification, terminal segments, ornamentation, and spines are commonly missing or poorly preserved (Pl. 3). Fortunately, some forms have such a characteristic outline that they can be identified despite advanced dissolution or partial fragmentation (Pl. 1, figs. 8, 16; Pl. 2, figs. 3, 10). We have also observed that the refractive index of the silica in a few samples is lower (Pl. 1, fig. 10) than normal for Paleocene assemblages.