A landscape model quantifies error in reconstructing fire history from scars

Abstract

Reconstructing fire regimes from fire scars is widely used in fire ecology to understand the role of fire and to determine prescriptions for restoring fire, but systematic analyses of the accuracy of fire-regime reconstruction have never been done. Errors in reconstruction may lead to a misunderstanding of the role of fire or incorrect restoration prescriptions. Here, a stochastic landscape model is used to quantitatively assess the accuracy of a commonly used statistic, the composite fire interval (CFI), as an estimator of the population mean fire interval or fire rotation, which are the central parameters useful in comparing fire regimes. Seven factors, that may affect accuracy, are explored. Two control the fire regime, one controls the scarring process, and the other four define the sampling and analysis procedure. Results show that: (1) CFI can be from 0.15 to 5.0 times the population mean fire interval, a range of more than 30 times; (2) all factors, except population mean fire interval, significantly affect accuracy; (3) accuracy shows predictable patterns that could be useful in designing a sound sampling scheme and choosing a proper analysis method. However, the complexity of effects of these factors and the need for some prior knowledge of them make it difficult to design sampling and analysis procedures to accurately estimate the population mean fire interval.