Rampant rainforests

: an assessment of rainforest boundary dynamics in Kakadu National Park based on aerial photography, field surveys and multi-scale modelling

Abstract

Understanding the extent and causes of savanna-forest dynamics in tropical regions is vital as small but widespread changes to tropical forests can have a major impact on global climate, biodiversity and human well-being. Changes to the boundaries of 50 monsoon rainforest patches in Kakadu National Park were assessed using temporal sequences of digitised aerial photography between 1964 and 2004. Wet and dry rainforest boundaries were manually classified for each year using a 20 x 20 m point lattice in a GIS. This revealed an overall expansion of rainforest patches by an average of 28.8%. A reassessment of the rate of change using bootstrapping of all available aerial photography supported the existence of an overall expansion of rainforest, but suggested that the rate of change was overestimated for dry rainforest. Floristic, structural, environmental and disturbance attributes of the changes were investigated by sampling 588 survey plots across 30 rainforest patches. The abundance of rainforest trees and grass were consistent with the trend of rainforest expansion. The view that the rainforest boundaries had been highly dynamic at a decadal scale was supported by (i) the overall floristic composition of newly established rainforest and stable rainforest being similar, and (ii) the rapid rates of change observed for a subset of nine rainforest patches with detailed aerial photographic histories. Generalized linear modelling supported an effect of late dry season fire frequency, historical buffalo impact and rainforest type in mediating the rate of change. The effect of fire in driving the changes was also supported by an effect of topographic fire protection and flammable weeds in GIS mapping of three rainforest case studies. However disturbance factors were unable to account for the overall expansion of rainforest. We conclude that the boundary expansion is likely to have been primarily driven by factors that have shown similar increases during the study period, such as annual rainfall and atmospheric CO2. Nonetheless, fire and buffalo impact have mediated the rate of change. This study provides a unique historical perspective on rainforest dynamics and will contribute to ‘adaptive management’ programs, particularly with respect to fire management.

Date of Award	2007
Original language	English
Supervisor	David Bowman (Supervisor)