Fire and population dynamics of woody plants in a neotropical savanna: matrix model projections

INTRODUCTION

Humans have had a large impact on fire regimes of tropical savannas. At present, the moist savannas of Australia, Africa, Asia, and South America commonly burn at intervals of 1-3 yr (Eiten and Goodland 1979, Lacey et al. 1982, Trollope 1984, Coutinho 1990, Stott 1990, Menaut et al. 1991, Russell-Smith et al. 1997), primarily due to anthropogenic causes. There is little doubt that burning would be less frequent in the absence of humans, although it is uncertain what range of fire frequencies are natural for such savannas. In the savannas of the Brazilian cerrado, it has been estimated that burning intervals ranged from 3-10 yr under indigenous occupation (Ratter et al. 1973, Eiten 1975, Eiten and Goodland 1979).

Predicting the long-term effects of humans on savanna diversity and structure will require an understanding of the effects of fire on woody plants. Woody cover plays an important role in nutrient cycling (Kellman 1979, Belsky et al. 1989, Isichei and Muoghalu 1992, Mordelet et al. 1993, Archer 1995), soil carbon storage (Belsky et al. 1989, Weltzin and Coughenour 1990, Dunham 1991, Garcia-Miragaya et al. 1994, Scholes and Archer 1997), seedling establishment (Kellman and Miyanishi 1982, Kellman 1985, Bowman and Panton 1993, Hoffmann 1996a), and the distribution of animals (Dall'Aglio 1992, Alho 1994), so changes in woody plant density can have widespread effects on savanna ecology. However, predicting the effects of fire on woody-plant dynamics is complicated by the need to consider interspecific variability in fire sensitivity, as well as temporal variation in fire frequency.

The difficulty in predicting the effects of burning on populations is largely due to the complexity of fire effects. Fire simultaneously affects sexual reproduction, vegetative reproduction, seedling establishment, individual size, growth, and mortality. These effects interact to alter the growth rate of a population, complicating attempts to determine which plant traits are most important for determining success or failure under a particular fire regime. Matrix population models are a valuable tool for predicting population responses to burning and for determining which demographic traits are critical for long-term success in frequently burned savannas (Silva et al. 1991, Canales et al. 1994, Caswell and Kaye 1996). In this paper, I use matrix population models to address the following questions for five species of woody plants of the savannas of the Brazilian cerrado:

How does fire frequency affect the population growth rate of woody plants of the cerrado? - The relationship between fire frequency and population growth rate will determine whether a species will decline toward extinction or become abundant under a particular fire regime. The range of fire frequencies that permit positive population growth is particularly important, as this will determine how well a species can cope with long-term changes in fire frequency. If each species is able to persist only under a narrow range of fire frequencies, we would expect large shifts in species composition in response to changes in fire frequency. However, if many cerrado species are able to maintain relatively stable populations under a wide range of fire frequencies, we should expect high ecosystem resilience (sensu Holling 1973).

How does fire frequency affect the mean individual size? - Within a species, individual size may span several orders of magnitude. Even if burning were to have little effect on population size, shifts in the size distribution can have large effects on the physical structure of cerrado vegetation. Understanding the long-term effects of burning upon woody cover may be just as valuable as understanding the effects upon numbers of individuals.

Is there a trade-off between fire tolerance and success in unburned conditions? - Species in fire-prone ecosystems exhibit a diversity of adaptations to burning, such as large carbohydrate reserves (Miyanishi and Kellman 1986, Bowen and Pate 1993, Bell et al, 1996) and thick bark (Landers 1991). Such traits require resources that could otherwise be allocated for reproduction or resource capture, so they may limit population growth in the absence of fire. I hypothesized that there is a negative relationship between success under frequent burning and success under protection from fire.

Which demographic parameters contribute most to the overall effect of burning on population growth? - Fire has an impact on all components of the life cycle of the study species, thus changing the population growth rate. We do not know which demographic parameters contribute most to this change in the population growth rate. A large effect of fire upon a particular demographic parameter cannot be taken as proof that the population growth rate will be greatly affected. The impact upon population growth rate also depends upon the sensitivity of the population growth rate to changes in the particular demographic parameter.