Space, time, and uncertainty: Detecting and characterizing boundaries in forest fire disturbances

Emulating characteristics of natural disturbances is an improved approach to forest management: maintaining forest ecosystem characteristics within their historic range of variation is believed to contribute to ecosystem integrity. While various characteristics of fire regimes have been investigated ( e.g., shape, size, frequency and legacies of pattern, process, composition), no study has focused on the localization of fire disturbance boundaries. Boundaries, as landscape elements, have properties that can be manipulated to restore and maintain ecological processes. To examine forest fire disturbance boundaries and their characteristics with respect to space, time, and uncertainty, (1) I developed new spatial statistical methods for detecting spatial location of persistent boundaries; (2) I constructed the State Transition Index to quantify phase transitions of boundaries for consecutive fire events and explored temporal patterns of boundaries and broad-scale top-down controls ( i.e., climate); and (3) I determined uncertainty in boundary location and applied fuzzy methods to quantify fire size using uncertainty. Low-severity fire scars derived from dendrochronological data at two study sites in the British Columbia Stein River watershed were used to test these novel spatial methods. In the middle Stein, 33 one-hectare plots from a 1512-hectare study area were analyzed for the years 1785 to 1937, while in the lower Stein, 35 one-hectare plots from a 412-hectare study area were analyzed for the years 1879 to 1947. Significant spatial relationships between fire boundaries and bottom-up topographic control (such as riparian zones) were found. Significant boundary transitions (such as oscillation) based on the Index were identified. Location uncertainty methods showed that boundary membership values were high near riparian zones. These results contribute to knowledge of historic ranges in variability in low-severity forest fire. Importantly, they provide spatial information of historical fire breaks to improve fire and forest management ("fire-smart" management). Moreover, the results provide information of boundary transitions over time, suggesting boundary responses under climate change. Uncertainty quantification improves understanding of boundary zones and fire sizes and is applicable to refugia management. In management of forest landscapes, I recommend manipulating forest boundaries according to the historical locations of natural disturbance boundaries to restore or maintain natural patterns and process.