It is supposed, and oft stated, that the increase in the rate of global warming due to anthropogenic influences may be so great that terrestrial ecosystems will not be able to adapt, and catastrophic change or collapse will ensue. The IPCC impacts assessment group stated (IPCC, 1990b):

Such warming would not only be greater than recent natural fluctuation, but it would occur 15 to 40 times faster than past natural changes. Such a rate of change may exceed the ability of many species to adapt or disperse into more favorable regions and many plant and animal species may become extinct.

This statement was based on Schneider (1989), which states that the global warming typically projected is "10 to 60 times as fast as the natural average rate of temperature change that occurred from the end of the last Ice Age to the present warm period (that is, 2°C to 6°C warming in a century from human activities compared to an average natural warming of 1°C to 2°C per millennium from the waning of the Ice Age to establishment of the present interglacial epoch)."2

Ecologists and foresters are properly concerned that the predicted changes might produce catastrophic ecological disruption. In this context, it is appropriate to examine the climatic record to determine the magnitudes of past climatic fluctuations, in particular, temperature variability on various time and space scales. Since our object of concern is ecological systems, we need to find what conditions plants and animals have been exposed to in the course of their existence. This will not tell us how any particular plant, species, or ecosystem will respond, but it will provide a useful background against which we can assess or predict response to global warming.

The objectives of the analysis below are threefold: (1) to analyze records of past climates in order to determine actual rates of warming so these rates can be compared with model predictions; (2) to investigate the variability of climate and microclimate on time scales ranging from a few minutes to a half-year; and (3) to take a brief look at the spatial variation of microclimates—the climate that plants actually live in—over space scales ranging from a meter or so to about one kilometer. I will concentrate on temperature, even though it is only one of many factors affecting plant behavior, because it is the most readily available type of observational data.


Climatic modeling has emphasized global mean temperature and its generalized latitudinal variation (IPCC, 1990a; hereinafter referred to as IPCC-I). Similarly, I intend to focus on global mean temperature, with only a cursory examination of the Northern Hemisphere and the temperate latitudes, even though according to IPCC-I "land surfaces warm more rapidly than the ocean, and high northern latitudes warm more than the global mean in winter."

Only in the past 150 years or so have the instrumental observations of temperature been sufficiently good to permit the calculation of global means. Nevertheless, there are many surrogate, or proxy, observations that may be suitable for estimating global mean temperatures prior to the era of instrumental observations. While there are many uncertainties as to their applicability, such surrogate observations have been widely used for that purpose (see, for example, Chapter 7 in IPCC-I). Since most of the concern over global temperature increase appears to relate to the rate of increase rather than to absolute magnitude, I will examine records of historical climate for examples of such rates, and attempt to clarify the related uncertainties.

The IPCC Scientific Assessment Panel (IPCC-I) has made an evaluation of global temperature since 1861. For the period from 1861 to 1989, they relied on the data of Jones (1988), which show a value of 0.45°C per hundred years, or 0.58°C for the 128 years of record. Given the scantiness of the earlier observations, a more reliable estimate might be made using data from 1881 to date. Such an analysis yields a value of 0.53°C per century, or 0.57°C for the 108 years of record (see Figure 1). (Note that the abscissa is for the appropriate time interval, Δy, not for time from the present.) Since my investigation spans several orders of magnitude in the time scales and temperature scales, it is appropriate to plot the data on a log/log scale.

Balling (1992), using data limited to a more recent interval, has calculated a linearized increase in global mean temperature of 0.45°C, slightly less than the 0.53°C of the IPCC-I report. Using the relationship between the global-temperature anomalies of Jones et al. (1986) and a stratospheric dust index (estimated from solar radiation records from a high-elevation station in Austria; see Wu et al., 1990), Balling estimated the effect of dust loading on global temperature. He found that one-third of the global-temperature trend of the past 100 years disappeared, bringing that portion of the global-temperature rise that might be attributed to atmospheric carbon dioxide increase to about 0.30°C. Although it is perhaps not directly comparable, Wu et al. (1990) have estimated the increase in global night marine air temperature to be 0.49°C for the period 18561988, and 0.29°C for the period 1888-1988. When the variability caused by two additional factors, solar irradiance


It should be noted that the ratio of rates of warming depends on the rate chosen for the denominator. If one chooses a period of very small temperature increase, the ratio can be made very large, producing an obviously absurd comparison.

The National Academies of Sciences, Engineering, and Medicine
500 Fifth St. N.W. | Washington, D.C. 20001

Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement