The ratio/temperature variations in different zonal intervals among the holes (Fig. 9, Fig. 10, Fig. 11; Appendix) require examination. In the foregoing description of each hole, we took the conventional approach of assuming that fluctuating ratios were caused entirely by changing surface-water temperatures. It is possible, however, that although surface-water temperatures are controlling the overall long-term trends, regional variation is being overprinted by the short-term trends of other, unknown influences (nutrient supply? salinity? turbidity?). This may explain the variation in ratio patterns seen among widely separated holes from the Tyrrhenian, Balearic, and Alboran Seas.
Several overall trends do emerge from the data. All holes show a dramatic reduction in D. brouweri beginning at least by the early part of Zone NN18. This might be interpreted simply as the declining abundance of a taxon nearing its extinction (at the end of Zone NN18). In most temperate to subtropical sites, however, D. brouweri occurs in significant numbers throughout Zone NN18, and we believe cooling waters were the true cause of this marked decline in abundance in Zone NN18 in the Mediterranean. Evidence for the timing of the onset of Northern Hemisphere glaciation (between 2.8 and 2.5 Ma) has been summarized by King (1996). The lower boundary of Zone NN18 occurs at about 2.5 Ma (Fig. 2). Wei et al. (1988) also found that the D. brouweri-C. pelagicus ratio decreased sharply after about 2.5 Ma on the Galicia Margin. Another predominantly cool-water interval also occurs in Zone NN13 in Leg 161 holes.
Rio et al. (1990b) presented a paleoclimatic synthesis of ODP Site 653 (Tyrrhenian Sea) based on micropaleontologic and oxygen isotope data. They also found that a major cooling began at 2.4 Ma (using a different time scale). Their dates for cool-water intervals in the early Pliocene (estimated to be in Zones NN14-NN15) are somewhat later than the corresponding interval suggested here (late Zone NN13). Their time scale is presented in terms of years, rather than biozones, and at least part of the discrepancy may result from the different scales used to date the intervals. Rio et al. (1990b) identified cooling at about 3.1 Ma (which on our scale would be in Subzone NN16A), based on an increase in cool-water C. pelagicus. We also found an overall increase in C. pelagicus abundance during this interval, but a correspondingly much larger increase in warm-water D. brouweri, which suggests a time of generally warm waters in late Subzone NN16A. Overall warm-water intervals occur in most holes in late Zone NN12-early NN13 and in Zones NN15, NN16A, and NN16B.
The warmest surface waters in the Mediterranean during the Pliocene occur during a brief interval centered on Zone NN16B. This "warmest" peak was found at Sites 974B, 977A, 978A, and 979A; a warm peak also occurs in Subzone NN16B at Site 975B, although not the warmest interval at that site. Zones NN13 to NN17 are missing at Site 976B. In several holes, this time of warm waters began in late Zone NN16A and extended into Zone NN17, but the peak warm period is in the middle Pliocene from Subzones NN16A to latest NN16B (about 3.0 to 2.6 Ma).
Crowley (1996), Dowsett et al. (1996), Raymo et al. (1996), and others have stated that the mid-Pliocene was the last time when global average temperatures were greater than temperatures of today. The time of this warm period is estimated to be around 3.0 Ma (Raymo et al., 1996). Crowley (1996) and Raymo et al. (1996) suggested that this Pliocene warmth may have been caused by higher atmospheric CO2 levels, quoting data showing that mid-Pliocene CO2 levels were about 98-100 ppm higher than today's levels, or by increased oceanic heat transport. Whatever the cause of the mid-Pliocene climatic peak, data from the Mediterranean presented here provide supporting evidence for the warm interval and further constrain the time of its occurrence.