PARENT SESSION
Poster Session 11: Disturbance Ecology
Tuesday, August 9, 5:00 PM - 6:30 PM, Exhibit Hall 220 A-E, Level 2, Palais des congrès de Montréal
Do results from small scale experiments help in understanding landscape patterns of boreal tree recruitment within a natural wildfire?
Johnstone, Jill*,1, 2, 3, Chapin, F. Stuart3, Kasischke, Erik3, 4, 1 Department of Geography, Ottawa, ON, Canada2 Arts and Science Division, Whitehorse, YT, Canada3 Institute of Arctic Biology, Fairbanks, AK, USA4 Department of Geography, Baltimore, MD, USA
ABSTRACT- Biologically important wildfires occur at scales of tens to thousands of hectares and this is often the scale at which we seek to understand and predict fire effects. However, landscape experiments are difficult to implement and we must often rely on small scale experiments to test hypotheses related to fire effects. As ecologists, we are challenged to reconcile the gap of scales to generate meaningful interpretations and predictions of landscape responses to fire disturbance. This study draws on data from two small scale experiments designed to test independent effects of burn severity and aspen competition on rates of tree recruitment after fire. These experiments were carried out 1-3 years after fire in small, replicated plots (plot size = 1.5 m2) in regenerating forest stands in Alaska and Yukon Territory. Response models derived from these experiments were compared with observations of stand-level recruitment patterns (plot size = 2500 m2) measured 8 years after fire within a 10,000 ha burn in interior Alaska. Fitted regression models were used to test whether observations from the natural burn were consistent with predictions from the experimental models. Informative priors within a Bayesian statistical framework were used to generate an analysis of recruitment patterns in the natural burn that were informed by the experimental results. Models without prior information provided a baseline to evaluate how knowledge from the experiments affected our interpretation of landscape patterns. Results from the comparisons indicated that the experiments provided critical information needed to interpret natural variations in tree recruitment. Nevertheless, models based on single-factor experiments had little predictive ability because of interacting effects and variation in the importance of different factors in driving stand-level responses. We conclude that learning about fire effects in boreal forests requires a multi-scale approach where experiments and observations both contribute to interpretation and prediction.
Key words: fire, boreal forest, spatial scale, recruitment