Testing disturbance theory with long-term data: Alternative life-history solutions to the distribution of events

A model of disturbance effects on fire-dependent tree populations is developed, parameterized, and tested using long-term data from northwestern Minnesota to determine the extent to which disturbance controls species composition. The model assumes fires ate necessary for recruitment, and they may cause mortality, depending on species, Reproductive success is estimated as an integral equation (analogous to Lotka's equation for R(0)) taking into account fecundity schedules, mortality schedules, and age-specific disturbance effects, Life histories are parameterized for the three dominant fire-dependent tree taxa in the region, Pinus resinosa, Betula papyrifera, and Populus. Long-term disturbance regimes are summarized as density functions of intervals between fires parameterized from fire scars on trees and in sediment charcoal records. Changes in the density of fire intervals in the past are the basis for predictions that different taxa would dominate, depending on their life histories, Fossil pollen data indicate changing abundances of tree taxa and are used to test predictions of the model of reproductive success. Comparisons of model predictions with changing abundances of pollen indicate that the density of fire regimes is one of the important controls on composition. Pinus resinosa is limited by the high frequency of fire, but longevity is an advantage in the region for this species; Betula is limited by frequent fire, and longevity is of little advantage; and Populus is least sensitive to fire regime. Results show that knowing simply the frequency of fire can lead to naive interpretations of fire effects (or lack thereof). The higher moments of disturbance densities can be critical for understanding responses of species that require extended intervals to achieve resistance to fire. These extended intervals may be race, in which case they are better described by the higher moments of the density than they are by the mean interval.