“Evaluating the drivers of post-wildfire successional dynamics across large-scale environmental gradients”

ABSTRACT
Ecologists have long sought a better understanding of post-disturbance vegetation community dynamics. This understanding is increasingly important, given recent predictions that changes in global climate will result in shifting disturbance regimes.  Along with shifts in the nature and frequency of disturbances, there will certainly be changes in the repercussions of these disturbances on natural systems. It is therefore important to determine how large-scale environmental factors, such as climate, impact the rates and trajectories of post-disturbance vegetation dynamics.

This research addresses one aspect of this discussion by investigating differences in post-wildfire vegetation dynamics across the Western United States. An understanding of post-disturbance vegetation dynamics across a diverse set of environmental conditions will be valuable in predicting and planning for novel combinations of these factors in the face of ongoing global change. To this end, locations of wildfires for the year 2000 were used to investigate changes in post-wildfire primary productivity from 2001 to 2006. Fire locations were determined using thermal anomaly data derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) satellite, available from the USGS. Net primary productivity (NPP) estimates derived from MODIS satellite data were obtained from the Numerical Terradynamic Simulation Group (NTSG) at the University of Montana. We hypothesized that the rate of change of post-wildfire NPP varies spatially at a broad scale across the Western United States, as a function of large scale environmental factors. To test this hypothesis, rates of change in NPP as a function of time since burn (ΔNPP) for each site were compared to large scale environmental drivers, including annual precipitation, average temperature maxima and minima, and the percentage of total precipitation which occurred in the winter months (% Winter).

As a first test, we wanted to know whether or not the ΔNPP data were autocorrelated, which would indicate that areas which were near to one another, and therefore influenced by similar large scale processes, had similar post-disturbance vegetation dynamics. Using Moran’s I statistic to test for spatio-autocorrelation, we determined that the ΔNPP data were spatially autocorrelated (p <.001), with the highest degree of spatio-autocorrelation at around 200km.

Second, we wanted to determine which of the potential large scale drivers accounted for most of the variation in ΔNPP across the Western US. Using stepwise linear regression, we determined that total precipitation and % winter precipitation accounted for 32% of the variation, after adjusting for multiple comparisons.

The relationships observed between rates of vegetation change since a wildfire event and large scale climatic drivers indicate that these forces are important to consider when evaluating the potential for post-wildfire ecosystem recovery. Understanding these relationships should add valuable information to the prediction of post-disturbance ecosystem dynamics in the face of future changes in broad scale climate regimes.