RESPONSE OF NORTH-AMERICAN ECOSYSTEM MODELS TO MULTIANNUAL PERIODICITIES IN TEMPERATURE AND PRECIPITATION

Ecosystem models typically use input temperature and precipitation dat a generated stochastically from weather station means and variances. A lthough the weather station data are based on measurements taken over a few decades, model simulations are usually on the order of centuries . Consequently, observed periodicities in temperature and precipitatio n at the continental scale that have been correlated with large-scale forcings, such as ocean-atmosphere dynamics and lunar and sunspot cycl es, are ignored. We investigated how these natural climatic fluctuatio ns affect aboveground biomass in ecosystem models by incorporating som e of the more pronounced continental-scale cycles in temperature (4, 1 1, 80, 180 year periods) and precipitation (11 and 19 year periods) in to models of three North American forests (using LINKAGES) and one Nor th American grassland (using STEPPE). Even without inclusion of period icities in climate, long-term dynamics of these models were characteri zed by internal frequencies resulting from vegetation birth, growth an d death processes. Our results indicate that long-term temperature cyc les result in significantly lower predictions of forest biomass than o bserved in the control case for a forest on a biome transition (northe rn hardwoods/boreal forest). Lower-frequency, higher-amplitude temprat ure oscillation caused amplification of forest biomass response in for ests containing hardwood species. Shortgrass prairie and boreal ecosys tems, dominated by species with broad stress tolerance ranges, were re latively insensitive to climatic oscillations. Our results suggest per iodicities in climate should be incorporated within long-term simulati ons of ecosystems with strong internal frequencies, particularly for s ystems on biome transitions.