Deforestation, Fire Susceptibility, and Potential Tree Responses to Fire in the Eastern Amazon

Abstract

In the state of Para, Brazil, in the eastern Amazon, we studied the potential for sustained fire events within four dominant vegetation cover types (undisturbed rain forest, selectivity logged forest, second—growth forest, and open pasture), by measuring fuel availability, microclimate, and rates of fuel moisture loss. We also estimated the potential tree mortality that might result from a wide—scale Amazon forest fire by measuring the thermal properties of bark for all trees in a 5—ha stand of mature forest, followed by measurements of heat flux through bark during simulated fires. Partial logging resulted in dramatic increases in downed woody debris. Total fuel mass was significantly greater in the logged forest (180 Mg/ha) compared to the other cover types (30—60 Mg/ha). However, the readily combustible fine—fuels (e.g., grasses, litter, herbs) were significantly greater in pastures (° 11 Mg/ha) than in all other cover types (° 6 Mg/ha). Anthropogenic disturbance altered microclimate, which in turn affected rates of fuel moisture loss and the dynamic equilibrium of fuel moisture contents. In pastures of average midday temperature was almost 10 o C greater, and the average midday relative humidity was 30% lower, than in primary forest. There was a sixfold difference in average vapor pressure deficit between the primary—forest and the open—pasture cover types. Given the relatively steep gradient between the vapor pressure of the fuel particles and the surrounding atmosphere in the disturbed communities, fuel moisture loss was more rapid and equilibrium moisture contents were lower than in primary forest. Based on the changes in fuels and microclimate, we determined that cattle pastures were the most fire—prone ecosystem. During much of the 6—mo "dry" season (total rainfall: 200—400 mm), sustained combustion was possible in this community within 24 h following rainfall events. Openings in the selectively logged forest would burn after 5—6 rainless days and in the second—growth forest after 8—10 d. In contrast, sustained combustion was not possible in the primary forest even after prolonged rainless periods (e.g., >30 d). Through an examination of bark tissues and simulated fires in primary forest we found that only a small percentage of the standing vegetation would likely survive even a low—intensity, surface fire. Mean (@+ SE) bark thickness for trees > 20 cm in diameter was 7.3 @+ 0.14 mm (n = 699) with values ranging from 1.5 to 28.9 mm. We found a significant relationship (r 2 = 0.77) between bark thickness and maximum cambium temperatures during fire simulations, and thereby estimated that in the event of a surface fire in the primary forest, 98% of all stems ° 1 cm diameter at breast height would be killed. Even though the autogenic factors in primary forest of the eastern Amazon create a microclimate that virtually eliminates the probability of fire, they are currently a common event in disturbed areas of Amazonia. As many as 8 x 10 6 ha burned in the Amazon Basin of Brazil in 1987 alone. In terms of current land—use patterns, altered microclimates, and fuel mass, there are also striking similarities between the eastern Amazon and East Kalimantan, Indonesia (the site of recent rain forest wildfires that burned 3.5 x 10 6 ha).