if strong enough to alter sea-ice distributions, may also be self-sustaining in terms of radiative balance. Global warming of near 6°C could be achieved by about 50% increases in poleward ocean heat transport.
A greater role for the oceanic thermohaline circulation may be a plausible mechanism for an increased role of the oceans, because of their heat capacity. Today, deep water forms in geographically restricted regions, for which a density contrast from the main ocean occurs through interaction with the atmosphere. The source of deep water depends on the buoyancy flux (Brass et al., 1982), which is a function of the density contrast and the volume flux. Brass et al. (1982) demonstrate the plausibility of a thermohaline circulation very different from today, including the potential for subtropical deep water formation. Interestingly, ocean GCM simulations completed for several periods during the Cenozoic predict warm saline deep water formed within the subtropics during the Eocene (Barron and Peterson, 1990). Modeling work by Ogelsby and Saltzman (1990) gives added support to the concept of warm salty bottom-water formation. Given the large volume fluxes of the deep circulation in the ocean and the evidence for the possibility of warm saline deep water formation, the thermohaline circulation is a plausible candidate for an increased role by the oceans in poleward heat transport. Much additional research is required to examine the potential role of changes in the thermohaline circulation.
In summary, the calculation of Horrell (1990) may render unlikely scenarios in which tropical ocean sea-surface temperatures are as low as 15°C. This conclusion is also supported by the recent analysis of Crowley (1991). However, decreases in tropical sea-surface temperatures of 3 to 5°C are plausible.
Third, Matthews and Poore (1980) argue, in part based on the conclusions of Newell and others described above, that the low latitude surface ocean had a stable temperature very close to present-day values. The difference between the isotopic temperature and the actual temperature is explained by changes in the oxygen isotopic composition of the oceans due to the storage of isotopically light waters as snow and ice in ice caps. The argument of Matthews and Poore (1980) essentially proposes extensive ice on Antarctica during much of the Cenozoic. The occurrence of ice on Antarctica is a matter of substantial debate, however, and Shackleton (1984) takes exception to the interpretation that the difference in tropical values is due to ice volume. Shackleton points out that such an interpretation would require more ice on Antarctica during the warm middle Miocene interval than today and would also require very large fluctuations in ice volume in the Early Miocene. Such a result appears unlikely and is unsubstantiated.
The arguments for Cenozoic tropical sea-surface temperatures near 15°C have also been challenged by Adams et al. (1990), based on biotic evidence. Adams and others note that the existence of Eocene mangroves, corals with zooxanthellae, and larger foraminifera preclude such low temperatures. These forms have minimum temperature limits closer to 18 to 20°C. If the temperature tolerances estimated for these organisms are accurate, then very cool tropical temperatures can be rejected. However, temperature variations within 5°C of modern values cannot be eliminated by these data.
In conclusion, none of the discussions presented above provides convincing arguments for tropical temperature stability, only limits to temperature variation. Tropical temperatures near 15°C can probably be rejected based on heat transport arguments and the biotic composition within the tropics during the Cenozoic. However, tropical temperature variations within 3 to 5°C of present-day values are not eliminated by any of the physical or biological arguments proposed to date.
Interestingly, much of the emphasis on the interpretation of tropical temperatures and the evaluation of estimates using geologic data described above has focused on cases in which tropical temperatures may have been lower than at present. Unfortunately, there is a notable failure of atmospheric GCMs to simulate tropical climates with lower temperatures than the present day (Barron, 1987). If the conclusions from the experiments of Covey and Thompson (1989) are correct, and an increased role by the ocean in poleward heat transport is required to achieve cooler tropical temperatures, then this problem is explained by the lack of an explicit ocean formulation in current atmospheric climate models. At present, the debate over tropical cooler temperatures cannot be addressed explicitly by current climate models. The results from ocean GCM experiments (Barron and Peterson, 1991) for an ocean driven by an Eocene atmospheric simulation in an uncoupled mode, which produced deep water within the subtropics, are suggestive of a different role for the oceans. However, to date, climate models have not simulated reduced tropical sea-surface temperatures based solely on physical processes incorporated within the model. The conditions for reduced tropical temperatures in models remain problematic.
The prospect of higher tropical ocean surface temperatures during warm climates, perhaps the most interesting case for future global change projections, has received much less attention than the "cool" tropics cases. Barron and Washington (1985) noted that higher carbon dioxide climates proposed for the mid-Cretaceous might result in