PARENT SESSION
Oral Session 5: Disturbance Ecology I: Scaling, Wind, and Ice.
Presiding: D McKenzie
Monday, August 2, 8:00 AM to 11:30 AM, Meeting Room B 113.
Grazing and fire as tools for managing invaded grasslands in California.
Christian, Caroline*,1, 2, Doak, Dan1, Pollock, Jacob1, Hujik, Peter2, 1 University of California, Santa Cruz, CA2 The Nature Conservancy, San Francisco, CA
ABSTRACT- Testing the impacts of large-scale disturbances remains a problem for ecologists. Many disturbances - such as wildfires and floods - are unplanned events that occur at large spatial scales, making it difficult or impossible to collect randomized, well-replicated data. The same problems exist for large-scale experiments designed to test disturbance effects. As a result, data from such experiments are often dismissed as un-analyzable or are analyzed inappropriately. One way out of this dilemma is to use methods from spatial statistics to directly include the effects of spatial autocorrelation (SAC) and environmental gradients in statistical tests for the effects of disturbances. We assessed how well this alternative to extensive randomization and replication would work for real ecological data sets. Drawing on prescribed-burn experiments in California grasslands, we compared the performance of traditional experimental analysis of our data with relatively simple analysis methods that directly account for spatial effects. We found that even when data came from randomized and well-replicated burn experiments, the usual parametric analysis methods failed to adequately account for SAC, resulting in substantial bias in the results. In contrast, even analyses that accounted for SAC in a fairly simplistic way did a much better job of revealing the actual burn effects in our study. Overall, our results demonstrate that even in the absence of well-replicated experiments, statistical power can be successfully increased for monitoring studies by using under-appreciated statistical methods that account and correct for SAC.
Key words: fire ecology, experimental design, disturbance, grasslands