DISCUSSION

Table 7 summarizes four proposed mechanisms for sapropel formation, the geochemical signatures associated with each mechanism, and the evidence for and against each proposed mechanism. Based on this information and on data collected in this study, we conclude that no single mechanism can explain all the sapropel-related isotopic trends in the Mediterranean Sea, or even those in the eastern basin of the Mediterranean. Instead, we believe that a combination of localized mechanisms, perhaps triggered by a single global or regional event, produced the observed trends.

The positive ¹³C excursions associated with sapropel formation at Sites 969 and 974 are inconsistent with the addition of significant amounts of either terrigenous organic matter or terrigenous DIC, both of which are isotopically lighter than marine organic carbon. Fontugne and Calvert (1992) cite terrigenous DIC input as a possible cause of the negative ¹³C excursion during sapropel formation near the mouth of the Nile (Site MD 84641), but data from Site 969 on the Mediterranean Rise constrain the spatial extent of this mechanism: most Site 969 sapropels are more enriched in ¹³C than surrounding sediments, though a few (S1 and S6-S8) from the late Pleistocene have roughly the same isotopic composition as surrounding sediment or are depleted by up to 1. This variability could reflect the changing intensity of the Nile floods from sapropel to sapropel, and thus the changing extent of their oceanographic influence. The larger, positive ¹³C excursions (up to 4.4) from the early Pleistocene suggest that terrigenous input from the Nile did not reach the Mediterranean Ridge at this time.

The lack of a positive ¹³C gradient away from the mouth of the Nile is also inconsistent with a widespread influence of terrestrial input. Although ¹³C would be expected to increase as the importance of the isotopically light terrestrial source decreased, sediment ¹³C actually decreases west of the Nile (Table 5).

During glacial sea-level drawdown, the sill at Gibraltar would have become very shallow (<200 m deep), thus restricting deep-water flow from the Mediterranean to the Atlantic. Consequently, deep waters would have been enriched in isotopically light carbon and nitrogen, as well as other remineralization products, such as silica and phosphate. During the transition to interglacial conditions, eastern Mediterranean upwelling (driven by westward surface flow and estuarine circulation) could bring this water to the surface, creating high-nutrient conditions, an isotopically lighter DIC pool, and, since increases in productivity caused by upwelled nutrients could outstrip the supply of upwelled DIC, a heavier ¹³C_POC pool. Bidigare et al. (1997) observed these patterns in the Peru upwelling zone, where ¹³C_DIC is 1 lighter, but ¹³C_POC is 4 heavier than at less productive Pacific sites.

These predictions match the ¹³C_DIC patterns found in the eastern Mediterranean (Vergnaud-Grazzini et al., 1986) and the Tyrrhenian Sea (Thunell et al., 1990) and are also consistent with the negative ¹⁵N excursions associated with the sapropels. Three additional lines of empirical evidence suggest that upwelling and enhanced productivity contributed to sapropel formation. First, Boyle and Lea (1989) found a five-fold increase in Cd/Ca ratios in the most recent sapropel, suggesting an enrichment in surface-water phosphate concentration at that time. Second, Calvert (1983) noted that barium is twice as abundant in five eastern Mediterranean sapropels as in the associated marls, and subsequent work demonstrates that the barium concentration in foraminifers correlates positively with silica abundance and with primary productivity rates (Boyle and Lea, 1989; Bishop, 1988). Finally, the high diatom concentrations found in sapropel S5 at four sites south and east of Crete suggests that upwelling of silica-rich waters might have occurred in this region (Schrader and Matherne, 1981). This upwelling could explain the isotope trends at Site 969, which lies within the estimated range of the diatom enrichment.

Assuming an estuarine-type circulation during episodes of sapropel formation, surface currents in the Eastern Mediterranean would flow westward and upwelling at Site 969 would not affect Site MD 84641, which shows an isotope excursion in the opposite direction. Instead, the negative ¹³C excursion at Site MD 84641 could reflect local terrigenous input of DIC or organic matter.

Westward surface flow would carry upwelled nutrients across the Ionian Sea and possibly into the western Mediterranean and the Tyrrhenian Sea. Although this scenario is generally consistent with isotope trends at Site 974, it presents two problems. First, nutrients in the upwelled water parcel would be consumed as they moved the substantial distance westward toward the Tyrrhenian Sea, which should lead to a smaller nitrogen isotopic depletion in sapropels at Site 974; in fact, the observed ¹³C and ¹⁵N are not significantly different for the two sites. Second, this interpretation assumes that the Tyrrhenian Sea sapropels formed contemporaneously with those in the eastern Mediterranean, but no correlation has yet been demonstrated. Alternatively, a separate upwelling system in the western Mediterranean could independently explain the isotope shifts at Site 974.

Sachs (1996) found negative ¹⁵N excursions in six eastern Mediterranean sapropels, consistent with our findings and those of Calvert et al. (1992), but he disputed the interpretation of Calvert et al. that these excursions reflect increases in primary productivity. Instead, Sachs argued that they represent a diagenetic signal and that surface ocean processes did not change significantly during episodes of sapropel formation. According to this argument, diagenesis leads to a 2 to 8 enrichment in ¹⁵N of sediments relative to sinking particles and a roughly 0.5 to 2.5 depletion in ¹³C (Sachs, 1996, and references therein; Fontugne and Calvert, 1992, and references therein; McArthur et al., 1992). Thus, the higher ¹⁵N values in nonsapropel sediments resulted from diagenetic alteration of organic matter while the lower ¹⁵N values within sapropels indicate a decrease in diagenesis during periods of anoxia and enhanced preservation or organic matter.

Although isotope data from Sites 969 and 974 are consistent with this scenario, many of Sachs's assumptions are open to challenge. For example, Altabet (1988) and Saino and Hattori (1987) have found that most alteration in ¹⁵N of sinking particulate organic matter occurs in the upper few hundred meters of the water column, and thus would occur regardless of deep-water oxygen concentrations. François et al. (1992) argue that diagenesis has only a minor influence on the ¹⁵N of sedimentary organic matter. Furthermore, Sachs' argument relies on the assumption that virtually no diagenesis occurred under anoxic conditions (or, at least, no diagenesis that affected the isotope record), but post-depositional breakdown is known to continue in anoxic sediments, albeit at a slower rate (Pedersen and Calvert, 1990). Sachs' scenario is also inconsistent with the considerable evidence supporting altered surface ocean processes during sapropel formation.

If deep-water anoxia has a small diagenetic effect on isotope ratios (as most studies suggest), then our data set provides little information about whether deep-water anoxia occurred during sapropel formation. However, enhanced preservation is not inconsistent with the observed isotopic trends, and circumstantial evidence appears to support some role for enhanced preservation. For example, in an estuarine circulation regime (hypothesized to occur during sapropel formation), the eastern Mediterranean would contain less oxygen than the western basin since altered deep Atlantic water would progressively lose its oxygen as it flowed eastward. This scenario is consistent with the smaller enrichments in organic carbon in the Tyrrhenian Sea sapropels relative to those in the eastern Mediterranean. In addition, deep-water oxygen depletion occurs whenever an increased downward flux of organic carbon is not offset by increased deep-water oxygen supply, thus suggesting that any mechanisms that generate sapropels through increased surface productivity would have the secondary effect of facilitating deep-water anoxia (Emeis et al., 1991) and enhanced preservation of organic matter.