Also of use for assessing long-term variability in the natural climate system are biological records. Coral colonies have been used as a living measure of the natural variability of the tropical surface ocean [ Shen et al., 1992]. Growth rate data from a colony off the Galapagos Islands over a 47-year period indicates that growth, and thus sea surface temperature, has varied dramatically over the annual, biennial, and ENSO frequency bands [ Shen et al., 1992]. It remains to be seen whether a longer record will reveal decadal patterns.
Tree ring width is also a good proxy for climate: in one study variability in annual growth was correlated with summer temperature and winter precipitation [ Graumlich, 1991]. Tree ring records are also of value because they are of global extent and reveal oscillation patterns that differ from region to region [e.g., D'Arrigo and Jacoby, 1993]. A 2,290-year tree ring record from Tasmania documents relatively stable oscillatory modes with mean periods of 31, 56, 79, and 204 years [ Cook et al., in press]. Treeline location and the pollen record [e.g., Newnham, 1992] are also proxies for past climate. Pollen, diatoms, and sediment geochemistry from a site in central Canada reflect shifts from tundra to forest-tundra environmental conditions and back over 150 years [ MacDonald et al., 1993].
Multi-year biological observations are used in conjunction with models of ecosystem dynamics to explore ecosystem sensitivity to climate. Multi-decadal site records are few, however, even in countries with a history of ecosystem science. Measures of above ground biomass from 11 grassland sites distributed globally, ranging from Kenya to Kansas to Kazakhstan, were assembled and used to validate Century, a model that simulates plant and soil organic matter dynamics [ Parton et al., 1993]. Substantial interannual and interdecadal differences in productivity occurred between dry and wet periods, with additional modulation by temperature and internal dynamics [ Parton et al., 1993]. An ongoing effort to monitor climate, plant production, and soil carbon is the U.S. Long-Term Ecological Research Program (LTER), which is in the process of building long records of climate and biological variability from 19 sites (including Puerto Rico and Antarctica) [ Greenland and Swift, 1990]. Recently, satellite images of vegetation (normalized difference vegetation index: NDVI) have been used by Malo and Nicholson [1990] to reveal rainfall fluctuations over the Sahel region of Africa. Their data indicate that NDVI, which is correlated with the fraction of photosynthetically active radiation utilized by vegetation, also serves as a proxy for rainfall and available soil moisture [ Malo and Nicholson, 1990; see also Schimel et al., 1991, for similar results from the semi-arid U.S.]. The international Global Terrestrial Observing System (GTOS) [ Heal et al., 1993] is proposed as a means of amassing long-term biospheric data on a global basis.