Understanding natural disturbance boundary formation using spatial data and path analysis

Natural disturbance boundaries have fundamental influences on landscape function; however, quantitative examination of the factors influencing their formation has been limited. Boundary formation is the final component of a disturbance event trio (initiation, propagation, and cessation) and is least well understood. I examined the influences of five stand-level variables (tree age, tree density, basal area of susceptible trees, slope aspect, and slope angle) and space as a surrogate for the unmeasurable factors on standardized tree mortality (basal area dead/total basal area) using multiple causal path hypotheses with data from seven wildfire and 11 mountain pine beetle (MPB) boundaries in British Columbia, Canada. One hundred percent of fire sites and 91% of MPB sites showed strong support for a positive effect of the basal area of susceptible trees (as defined by their propensity to burn or to be attacked by MPB) on tree mortality (r = 0.21-0.81). In addition, seven of 11 MPB boundaries had mortality patterns that were affected only indirectly by the spatial location across the boundary. This indirect effect of space suggests that other unmeasured spatial factors were not responsible for the pattern of mortality. Thus, MPB dispersal limitation did not appear to be an important factor stopping the outbreaks at these seven sites. A population based approach would be insufficient for understanding the location of the boundaries. On the other hand, mortality in all fire boundaries was directly affected by the spatial location across the boundary. In the context of multiple hypothesis testing of path models, this indicates that there was an important influence of spatially varying but unmodeled factors, such as weather and surface vegetation. Furthermore, path models that included other factors explicitly examined in this study (tree density, tree age, aspect, and slope) were rarely supported by the data, suggesting little contribution to boundary formation. These results demonstrate that the combined use of spatial plots, path analysis, and a priori multiple hypotheses can give important insight into the functioning of ecological processes.