Pattern and process: Fire-initiated grass invasion at Amazon transitional forest edges

Abstract

Anthropogenic fire and grass propagules from land use change may prompt an unintended, intractable grass-fire cycle at tropical forest edges. Yet, in the Amazon, the actual extent and mechanisms that drive grass invasion into closed-canopy forests is largely unknown. The Amazon transitional forest, situated between more humid forests to the north and savanna to the south, is particularly vulnerable due to an expanding agri-business frontier. Therefore, in this formation, we documented the probability and extent of grass invasion in three adjacent 50 ha forest plots that were burned annually for three consecutive years (B3), once (B1), and not at all (B0). We found that single and repeat fires prompted grass invasion, compared with the control. The extent and probability of grass invasion increased with repeat fires and time since fire. By three years after the initial burn the probability of grass invasion was 0.76 and 0.58 in Bl and B3 plots vs. 0.02 in the control. After three years, the mean incursion distance of grasses was 10 m in the Bl plot and 8 m in the B3 plot. Potentially interacting and non-exclusive mechanisms could explain this grass invasion, including (i) fire’s role in changing resource availability and supply rates, thereby shifting competitive advantages to favor grasses; (ii) fire’s density-independent effects that preferentially select for grasses due to growth form and life history strategy; and (iii) a fire-initiated positive feedback that maintains selection for grasses and thereby increases future fire probability and intensity by building up fine fuel loads and drying the surrounding microenvironment. Given the lattice-like pattern of degraded, burned forest edges left by rapid agricultural expansion in Amazon transitional forests, grass invasion could be an extensive—yet currently undetected—form of forest impoverishment.