Although not a paleoecological parameter per se, careful examination of preservation is important to evaluate the bias in any paleoecological signal, and given that preservation is correlated to productivity, changes in preservation may provide important paleoceanographic information. Preservation can be analyzed based on fragmentation, dissolution, and recrystallization. Whereas recrystallization is not of major concern in the material (no chert was observed above Core 183-1138A-41R; 383 mbsf), fragmentation and dissolution of tests seriously affect the overall preservation of the faunas. As tectonic strain is virtually absent, the observed fragmentation can be used as a proxy of opal dissolution. The traditional classification of preservation into poor, moderate, and good applied to ODP micropaleontological samples thus may be translated into strongly dissolved/fragmented, moderately dissolved/fragmented, and weakly dissolved/fragmented faunas.
We tentatively classified preservation by visual estimation (Table T3) as done in most other ODP reports, but additionally we suggest that a more rigorous classification of dissolution can be applied. Only with a more quantitative measure of preservation can we hope to statistically test correlations between shell dissolution and productivity. As a first approximation, we calculated the percentage of Siphocampe and Artostrobus in our samples (Table T3). These genera are among the most robust taxa in our material and were observed in a fairly high absolute abundance throughout (Pl. P9, figs. 3-18; Pl. P10, figs. 1-8). Although the percentage of Siphocampe and Artostrobus may be controlled by additional factors and there are occasional other robust taxa in the samples, it is thought to be an independent quantitative proxy of fragmentation/dissolution. The higher the proportion of Siphocampe and Artostrobus, the higher dissolution is thought to be. Because Siphocampe is by far the more abundant of the two genera in our samples, we define the name "Siphocampe index" for the cumulative percentage of Siphocampe and Artostrobus in a sample. The significant correlation between the qualitative preservation evaluation and our Siphocampe index (Fig. F7) supports the suggestion that the latter may represent a proxy for radiolarian faunal preservation in the Oligocene. Both the qualitative preservation index and the Siphocampe proxy indicate a significant upward increase of preservation in the studied interval. The percentage of Siphocampe and Artostrobus declines from >80% in the lowest three samples to <10% in the upper part. The first well-preserved faunas are recorded in interval 183-1138A-35R-2, 105-107 cm, according to qualitative studies, whereas the Siphocampe index suggests that dissolution/fragmentation is reduced already in interval 183-1138A-35R-5, 20-22 cm.
Radiolarian abundance ranges from ~2,500 individuals per gram to >150,000 individuals per gram of dry sediment. Abundance increases upcore as significantly as preservation. Core 183-1138A-36R consistently yields abundances of <10,000 individuals per gram; abundance in Core 35R varies between ~12,000 and ~80,000 individuals per gram; and Core 34R always shows values >50,000 individuals per gram. Both the Siphocampe proxy and the qualitative estimate of preservation are significantly correlated with radiolarian abundance (Fig. F7). Following common arguments (Baldauf and Barron, 1990), both opal preservation and abundance of siliceous plankton groups are thought to co-vary with productivity, although the relationship is not simple (Nelson et al., 1995). Lazarus and Pallant (1989) have shown that Oligocene radiolarian abundance in the Labrador Sea is very well correlated to total organic carbon content and other independent proxy indicators of productivity. This, however, is not true for diatoms that are generally thought to be even better productivity proxies (Ragueneau et al., 2000).
All diversity indices are strongly correlated with each other and with our preservational proxies (Figs. F7, F8). This observation poses problems for the paleoecological interpretation of our data. All diversity indices and the abundance data are largely explained by fluctuations in preservation. Based on R2 values, up to 90% of the variation in diversity can be explained by variations in preservation. Important exceptions are evenness and the percentage of diatoms, which are only weakly correlated with our qualitative measure of preservation but still exhibit very high correlations with our Siphocampe proxy. The weakest correlation in the whole data set is between the percentage of diatoms and radiolarian abundance. This may point to an independence of diatom and radiolarian abundance. If so, simple measurements of opal flux are probably not sufficient to characterize the productivity and interpret the signal, a point already emphasized by Diester-Haass (1995) and Ragueneau et al. (2000).
Even a principal component approach (Varimax rotation) does not help greatly to constrain primary patterns. Only one factor has an eigenvalue of >1, explaining 77% of the total variance in the data set. Three factors explain 91% of the total variance. The first two factors are best interpreted as preservation. It is only the third factor that has the highest loadings on the percentage of diatoms in the assemblage, again indicating a somewhat decoupled pattern of diatom and radiolarian abundances.
When comparing only diversity patterns of equally well-preserved faunas in the section, no significant trend through the lower Oligocene is evident. The maximum diversity (Shannon and Margalef indices and species richness) is reached in Sample 183-1138A-34R-3, 105-107 cm, well below the top of the investigated time interval.