Emanuel, Shugart, and Stevenson have used an interesting and potentially instructive approach to the assessment of the impact of CO2 on climate. Particularly striking is the depiction in map form, using the Holdridge Life-zone Classification, of the expected changes in vegetation. It is therefore unfortunate that there appear to be two serious flaws in the construction of the charts for the 2 x CO2 case in high latitudes.
The first and less serious is the use of the annual mean temperature increases from Manabe and Stouffer (1980) rather than seasonally varying values. Figure 23 of Manabe and Stouffer shows that over land in northern high latitudes, the annual mean changes are typically 5-8K north of about 55 ~ N, much greater than those for summer, which range from 1.25 -6.25K (mostly less than 5K). Since the growing season or, in terms of the Holdridge Classification, the period with temperatures above 0 ~ is concentrated in the summer months at these latitudes, these smaller values are the ones to use. The authors make no reference to the effect of their use of annual values in high latitudes where it can more than double the change.
The second and more serious error which is apparent from comparison of the maps is that the annual mean temperature adjustments have been applied to the 'biotemperatures' for the normal CO2 case. The correct approach would have been to recalculate the biotemperature s for 2 x CO2. The results so obtained differ because, as described by the authors, 'Average biotemperature is the average temperature over a year with the unit temperature values (daily, weekly or monthly temperatures) that are used in computing the average set to 0~ if they are less than or equal to 0~ '. (This method has clearly been used by the authors in the normal case). The effect of this error is that at stations in high latitudes, where mean temperatures are below freezing for all but 3-6 months of the year, the changes are 2 to 4 times too large.
Combining the two errors it is clear that the changes from normal to doubled CO2 Can be grossly exaggerated. I have made some calculations from observed data for eastern Canada where the Manabe-Stouffer estimates of temperature change are below 5K for 4 x COs in summer and 7 to 9K in the annual mean. In this area Emanuel et al. show changes in Holdridge Classification from 'polar' to 'boreal forest' and from 'tundra' to 'boreal forest' or even 'cool temperate forest or steppe' due to the 3 to 5K changes (C-A in Table I) in biotemperature they apply. Calculations avoiding the errors give changes (B-A in Table I) of only about 1K. The changes in Holdridge Classification are much reduced. In particular the polar and tundra types, eliminated in the published map, still occupy large areas.