Though many dramatic advances in the science of climate change over the last decade have altered our understanding of the phenomenon, three factors in particular have contributed to the emergence of a new perspective. First, modelers have developed transient ocean-atmosphere models that do a better job of capturing the dynamic relationship between oceans and the atmosphere. Second, scientists have discovered that sulfates cool the atmosphere. Third, continued climate measurements have confirmed the presence of a warming trend, but the trend is smaller than what earlier models predicted.

The transient ocean-atmosphere models, which capture the interaction between the warming of the ocean and climate, provide a more realistic tool for modeling the dynamics of greenhouse gas-climate interactions. The models predict much slower warming than the earlier equilibrium climate models. The discovery that sulfates are coolants explains an important geographic anomaly: that the cooler temperatures around the northern industrial countries are a consequence of the high quantities of sulfates found in those regions. Careful measurements of climate around the globe yield another important insight. The climate has been warming, but at a rate closer to the coupledmodel predictions than to the predictions of earlier models. The net result of all these changes is that climate scientists are more confident in their prediction that greenhouse gases will cause warming, though they have also revised downward their estimate of the range and the expected amount of warming that will occur.

Whereas the EPA report worked with changes of from 3œC to 6œC by the middle of the next century, the International Panel on Climate Change (IPCC) now predicts changes of from 11C to 3.5œC by 2100 ( Houghton et al. 1996 ). The average amount of global warming expected for the next century without any abatement program is 2œC. Given that the most dramatic warming is predicted for the poles, most places where people currently live would endure changes of less than 2œC spread over the next century. The rate of predicted warming has fallen from about 0.5œC per decade to about 0.2œC per decade. Conjoined with these small temperature changes are an average increase in precipitation, more than a doubling of carbon dioxide, a reduction in the diurnal cycle (from warming at night), and a sea-level rise of about 50 centimeters. The reductions in predicted magnitudes are very significant.

Ecosystem modeling has developed rapidly, and ecogeography models have improved their ability to generate the potential distribution of natural vegetation across the earth ( Neilson and Marks 1994 ; Neilson et al. 1992 ; Prentice et al. 1992 ; Haxeltine and Prentice 1996 ; Wood ward et al. 1995). The models can predict the areas where forests of different types will grow and the areas that will become grasslands. The models can also predict how ecosystems will shift in a new climate. Not surprisingly, the models predict that warming will cause ecosystems to shift generally to higher latitudes.

The predictions of the ecogeological models are in sharp contrast to the predictions of ecosystem collapse generated by the gap models ( Smith and Tirpak 1989 ). Rather than portraying the destruction of ecosystems around the world, the models simply display an alternative geographic pattern of ecosystems across the earth's surface. Even if many of the expanding ecosystems are considered desirable, the process of change will inevitably produce unwanted effects as well. The effects on endangered species remain largely unstudied, but they are likely to be harmful. Some shifts will be more harmful than helpful: deserts might expand, for example, or forest systems become grasslands. In presenting this complex set of changes, analysts must be careful not to focus only on harmful consequences but to give a balanced and representative view of all the myriad effects.

A second set of ecosystem models has begun to explore the effect of climate change on the metabolism or productivity of ecosystems. These ecophysiology models measure such key phenomena as photosynthetic rates and net primary productivity ( Melillo et al. 1993 ; Running and Coughland 1988 ; Running and Gower 1991 ; Parton et al. 1988 ). Their assessment of carbon fertilization has been especially important. Experiments in laboratory conditions have indicated that certain plants grow much more readily in a CO ₂ -enhanced world. Although the magnitude of the benefits may be somewhat mitigated in a natural ecosystem context, the carbon-fertilization results point to more optimistic outcomes. They were largely supported by the ecophysiology models, which suggest that net primary productivity is likely to increase with warming. That is, a warmer, wetter world, with enhanced CO ₂, is likely to be a greener world as well.

The insights of the ecophysiology and ecogeographic models together yield a new perspective on the ecosystem consequences of greenhouse gases. Tropical and temperate systems move to higher latitudes and push boreal forests into current tundra. Potential forest vegetation appears overall to increase. As ecosystems shift to new locations, animal populations will also shift. Overall productivity rises, but not uniformly.

There will be many changes associated with warming, then, and they do not readily translate into any single index. That is to say, a worldwide shift in ecosystems is likely to lead both to myriad benefits and to myriad damages. In general, the new results suggest an expansion in bioproductivity, which is a far more optimistic result than past predictions. It is important to note, however, that the available models focus on equilibrium effects. We still know little of the dynamics of ecosystem change and of how ecosystems will shift from one equilibrium to another. Thus, it is likely that the pace of change will be important to ecosystems and may introduce transient effects that the equilibrium models cannot foresee.

More optimistic scientific results also affect agronomy. The direct effect of increases in temperature on crop yields is mildly harmful in temperate climates and more severe in tropical climates ( Reilly et al. 1996 ). But that effect is counterbalanced by a strong, positive carbon-fertilization effect. The average crop is 30 percent more productive in a CO ₂ enhanced world ( Reilly et al. 1996 ). Aggregate world production will likely be robust; production increases in temperate climates ( Reilly et al. 1996 ) will offset small reductions in tropical output.

Even estimates of heat stress have changed over time. Daily mortality studies show that large increases in death among the elderly follow early summer heat waves ( Watson et al. 1996 ). The studies were used to argue that warming would increase heat-stress deaths by from 6,600 to 9,800 per year in the United States alone ( Pearce et al. 1996 ). Analyses of annual mortality rates, however, show that the elderly live longer in warmer climates ( Mendelsohn and Shaw 1998 ; Moore 1998 ). A closer examination of heatstress deaths reveals that they are higher in cold parts of the United States with high seasonal temperature variability. The death rates are relatively low in stable warm climates. Thus, heat-stress deaths appear to be caused not by warming but by temperature variability. They will grow in number not as climates warm but as the variability in climate increases. . .