DISCUSSION

At Site 975, the amplitude of the ¹⁸O oscillations increased by steps from the uppermost Pliocene up to Holocene, from about 1.0 prior to 1.7 Ma, to 1.6 between 1.7 Ma and 0.9 Ma, to 1.9 between 0.9 Ma and 0.42 Ma, to reach 2.9 during the upper Pleistocene. Synchronous shifts, but of less amplitudes, were also observed at Sites 976 and 977 in the Alboran Basin (von Grafenstein et al., Chap. 37, this volume) and at Site 653 in the Tyrrhenian Basin (Vergnaud-Grazzini et al., 1990). Outside the Mediterranean, the ¹⁸O variation by about 1.2 records the cyclic changing of ice volume in the Northern Hemisphere during glacial and interglacial periods (Fairbanks, 1989). Superimposed on this global ice effect, the additional ¹⁸O changes in the foraminifer records are a result of either thermal effects with 1°C cooling resulting in a ¹⁸O increase by 0.24 (Craig, 1965) or to local variations in the freshwater budget. For instance, in the modern Mediterranean, an increase by 1 of the surface water ¹⁸O values corresponds to a 4 increase of the salinity (Pierre, in press).

In the West Balearic Basin, the global increase of the amplitude of the ¹⁸O oscillations from the uppermost Pliocene up to the Holocene was caused by a multi-step drift by 1.7 of the glacial ¹⁸O values, that registers the effects of major climatic or hydrological changes (Fig. 1A, Fig. 1B). In the other parts of the western Mediterranean, a similar drift is observed in the glacial ¹⁸O values but it is only of 1.0 in the Alboran Basin (von Grafenstein et al., Chap. 37, this volume) and in the Tyrrhenian Basin (Vergnaud-Grazzini et al., 1990). This indicates that the climatic and hydrological changes during the Pleistocene were not exactly similar within the western Mediterranean.

At about 1.7 Ma, the glacial ¹⁸O values increased by 0.3 and the interglacial ¹⁸O values decreased by 0.3 at Site 975. These variations are thought to register weak climatic changes toward dryer/cooler glacial stages, and wetter/warmer interglacial stages at the onset of Pleistocene.

Near 0.9 Ma, the glacial ¹⁸O values were again shifted by 0.4 while the interglacial values remained at similar levels. This shift is also recorded in the Alboran Basin at Sites 976 and 977 with a similar amplitude, while it appears weaker at Site 653 in the Tyrrhenian Basin. Converting the ¹⁸O variation to temperature variation indicates that, in the period 0.42-0.9 Ma, the surface temperatures during glacial stages were lower by 2°C compared to those during the glacial stages before 0.9 Ma. Alternatively, this variation might be caused by higher rates of evaporation, in turn a result of increasing dryness in the glacial Mediterranean area.

After 0.42 Ma, the ¹⁸O values of G. bulloides at Site 975 were similar to those measured at Site 653 in the Tyrrhenian Basin (Vergnaud-Grazzini et al., 1990), except for the fact that they were more positive by about 0.5 than at Sites 976 and 977 in the Alboran Sea (von Grafenstein et al., Chap. 37, this volume). This may be fully explained by the ¹⁸O variability of the Mediterranean surface waters where G. bulloides precipitated their shells. In the modern Mediterranean, the ¹⁸O values of the surface waters in the Balearic Basin are enriched by about 0.4 to 0.5 relative to those from the Alboran Basin because of the increasing evaporation over the Mediterranean, an effect that is also responsible for the ~1.5 increase in surface-water salinity between the Alboran Basin and the Balearic Basin (Pierre, in press). The fact that the ¹⁸O gradient was similar to the present day during the upper Pleistocene means that at this time the surface salinity gradient between these parts of the western Mediterranean was not significantly different from today's salinity gradient. During this period, the glacial ¹⁸O values increased by 1.0 at Site 975, a value higher by 0.5 than at Sites 976 and 977 in the Alboran Basin (von Grafenstein et al., Chap. 37, this volume) and at Site 653 in the Tyrrhenian Basin (Vergnaud-Grazzini et al., 1990). The change by 0.5 may be attributed to cooler glacial conditions in the surface waters during the upper Pleistocene than before; a temperature decrease by about 2°C may thus be estimated between the middle and the upper Pleistocene. If we assume that the thermal response to the glacial conditions was rather uniform in the surface waters of the western Mediterranean, then the remaining 0.5 ¹⁸O variation at Site 975 has to be attributed to a local effect. It might have corresponded to an increased freshwater deficit in the West Balearic Basin during glacial times of the upper Pleistocene. Assuming that the slope between ¹⁸O and salinity was not different from the modern value of 0.25 (Pierre, in press), salinities of surface waters in the West Balearic Basin were thus higher by 2 during the glacial stages of the upper Pleistocene, compared to those during the glacial stages of the middle to lower Pleistocene. An alternative should be that before 0.42 Ma, the salinity gradient in the western Mediterranean was similar to the modern one but the glacial surface water temperatures were higher by 2°C in the West Balearic Basin compared to those of the Alboran and Tyrrhenian Basins.

During the whole Pleistocene, the amplitudes of the ¹⁸O oscillations became, thus, more and more higher in the Mediterranean compared to those of the open ocean, a characteristic of the restricted Mediterranean basin that amplified the global oceanic signal when the terms of the evaporation-precipitation budget were modified, even weakly. During the glacial stages, the increased dryness over the Mediterranean, combined with the restriction of the Atlantic water advection at the Gibraltar strait due to the global sea level lowering, acted together to increase surface salinities to much higher levels than in the open ocean. In addition, from the lower Pleistocene to Holocene, the interglacial Mediterranean temperatures were rather stable while the glacial Mediterranean temperatures became progressively cooler by about 4°C to 5°C in response to the global climatic deterioration, as evidenced by the development of extensive ice sheets in northern Europe and in the Alps and Pyrenees.

The ¹³C variations of G. bulloides throughout the Pleistocene at Site 975 show that in the West Balearic Basin, as in the global ocean, the glacial periods were generally more productive than the interglacial periods. The detailed examination of the ¹³C curve shows that the glacial-interglacial transition is often accompanied by a rapid drop (more negative) in ¹³C values; during full interglacial stages, ¹³C oscillates between low values and high values that may reach those of glacial stages. This means that in addition to the effect of the global ocean ¹³C variation, the ¹³C record in the west Mediterranean registers also specific responses to local changes in the hydrological conditions controlling productivity, mostly the thermohaline stratification of the surface waters.

During periods of high primary production and strong stratification in the surface waters, the upper waters become enriched in ¹³C because the ¹²C-depleted CO₂ is preferentially used by phytoplankton; the resulting increased rate of remineralization causes the release of ¹²C-depleted CO₂ in the underlying waters, which are then up-mixed by homogenization of the water column during the winter convection. As a result, G. bulloides that are living in spring blooms where the upper water layers are fertilized by the recycled nutrients, would exhibit low ¹³C values.

The hypothesis of shoaling pycnocline, related either to sea-level lowering during glacial times or to decreased excess evaporation during sapropel deposition was evoked by Ganssen and Troelstra (1987) and Rohling (1994) to explain the fertilization of surface waters and the enhanced fluxes of export production that is preserved in the sediment at the time of sapropel deposition. The deepening of the pycnocline, which may occur by the decrease of the temperature and/or salinity vertical gradient, would stop the supply of deep nutrients to the surface where oligotrophic conditions prevail.

The sapropel layers were deposited during interglacial stages or warm episodes of glacial stages; a few layers occurred during glacial stages, but at times of transition from cold to warmer conditions. They are mostly characterized by low ¹³C values of G. bulloides, indicative of enhanced nutrient supply to the depth of the pycnocline and thus of high levels of primary production in the surface waters.

The deposition of sapropels at Site 975 was not a regular process since the time span between two sapropels can range between 1 k.y. and 189 k.y. (Table 1). All sapropels are correlated with the maximum values of the 65°N summer insolation curve, and each of them may be ascribed to one insolation cycle (Table 1), following the codification of Laskar (1990). Most of the sapropels identified in the West Balearic Basin are synchronous to the sapropel layers of the eastern Mediterranean; the first sapropel (536) at 1809 ka is contemporaneous of sapropel C6 defined by Verhallen (1987) and the last sapropel (501) at 123 ka is equivalent to sapropel S5 (Ryan, 1972). Sapropels S1, S2, S3, S4, and S8 are lacking at Site 975, but low ¹³C values of G. bulloides at 8 ka, 55 ka, 80 ka, 102 ka, and 217 ka, which are the times of their deposition in the eastern Mediterranean, might indicate increasing rates of organic matter remineralization at subsurface depths in the West Balearic Basin; however, deep ventilation would have been strong enough to prevent the preservation of the sinking particulate organic matter. The recent results obtained in the eastern Mediterranean cores from ODP Leg 160 will make the correlation between the stagnation events in the different parts of the Mediterranean more precise.