A growing literature has emphasized the impacts of wildfire on riparian environments. It is difficult to assess the overall ecological and management importance of riparian fire, however, because the literature is largely anecdotal, made up of individual studies that document the impact of specific fires in specific locales. We seek a contextual understanding of how frequently riparian zones actually do burn, and of how that frequency varies across a variety of climatic settings.
In this paper we use data for 21 years of wildfires to examine the occurrence of riparian wildfire across the state of California. We used the Monitoring Trends in Burn Severity and LANDFIRE databases to identify fires that had burned in areas classified as having riparian vegetation, and the Fire and Resource Assessment Program database to determine the date of occurrence of each. From 1990 through 2010, an average of 1197 ha of riparian vegetation burned per year, which extrapolates to a fire return interval of 843 years. The statewide totals are misleading, however, because there is substantial geographic variance in the occurrence of riparian fire. In different ecoregions return intervals range from as less than 75 years to more than 1000 years. There is also substantial geographic variation in the season of riparian fire, and in the relationship between fire occurrence and annual precipitation.
Wildfire serves as an important agent of disturbance across a range of ecosystems, and riparian environments are no exception. A growing body of literature suggests that wildfire does indeed have quite significant impacts on riparian ecosystems, especially in regions with distinct dry seasons, such as those with Mediterranean-type climates. However, most of the research on riparian fire has been anecdotal – that is, it has focused on the impacts of individual fires, rather than on the frequency with which they occur. The exceptions to this tendency are a few local studies from coniferous forests in western North America where dendroecological evidence has been used to reconstruct histories of riparian fire. There are no regional-scale studies available thus far that actually quantify the frequency with which riparian fire occurs. In this paper, we use remotely sensed data to determine the frequency with which riparian environments burn in the state of California. We focus on California because it is a region where wildfire is known to be of great importance, importance that is likely to increase with global climate change. We include the entire state because we recognize the likelihood that riparian fire frequency will vary in different eco-regions, and we seek to capture that variation. Specifically, our goals are to determine the frequency with which riparian environments in burn, to determine the variation in that frequency among ecoregions within the state, to compare the frequency of riparian fire with that of surrounding uplands, to describe the variation in seasonality of fire occurrence among ecoregions, and to explore whether variation in riparian fire can be related to basic climatological variables.
We used the Monitoring Trends in Burn Severity (MTBS) database to determine the area of individual fires throughout California. MTBS data are derived from comparison of pre- and post-fire Landsat Thematic Mapper and Enhanced Thematic Mapper Plus images at a 30 m resolution, and provide perimeter and burn severity information for fires greater than 4 km2 in. Our interest was specifically in the riparian area burned, so we overlaid the MTBS data on the LANDFIRE database, which (among other things) delineates vegetation typ. We classified overlapping areas of fire and riparian vegetation (i.e. patches that had been burned at low, medium or high severity according to the MTBS data, and were also riparian according to the LANDFIRE data) as burned riparian. To determine the starting dates of those fires that had included riparian areas, we identified each within the Fire Perimeter Database developed by California’s Fire and Resource Assessment Program (FRAP), which includes the requisite ignition dates. In data for years prior to 1990, fire perimeters in the two databases (MTBS and FRAP) did not consistently align, suggesting that there might be some reliability concerns with one or both. We therefore limited our analysis to fires that occurred from 1990 through 2010.
To examine the influence of climate on riparian fire, we used considered annual precipitation, seasonal temperature maxima, and drought severity. We used the spatial datasets of precipitation and temperature developed by the PRISM Climate Group (http://www.prism.oregonstate.edu/); these data are at a 30 second resolution. For precipitation, we used annual totals for the water year (October 1 to September 31), averaged across each Ecoregion. For temperature, we focused on the seasonal high temperatures likely to affect fire occurrence and spread, calculating average maximum temperatures for Spring (March-May), Summer (June-August) and Autumn (September-November), again averaged across each Ecoregion. Because our concern was to determine the impact of precipitation and temperature, rather than the effect of specific conditions, we normalized each of these variables, using z-scores for our analyses.
For drought severity, we used Palmer Drought Index (PDI) data from the National Climate Data Center (http://www.ncdc.noaa.gov/temp-and-precip/drought/historical-palmers.php). The NCDC data are monthly values, mapped by climate zone. We first calculated an annual mean for each climate zone.
Then, because the NCDC climate zones do not closely match the USEPA Ecoregions, we calculated spatially weighted averages for each Ecoregion.
Overall, an average of 1197 ha of riparian vegetation burned per year, approximately .12% of the total area mapped as riparian vegetation. This extrapolates to a return interval of 843 years. By comparison, ~.32% of upland vegetation burned per year, for a recurrence interval of 318 years. Those totals conceal substantial geographic variation, however, as ecoregions were quite variable in their riparian fire frequency.
It is notable that while riparian fire is almost non-existent in some ecoregions, in others it is very prominent. Indeed, within the Southern California Mountains, riparian fire burns with an extrapolated recurrence interval of 74 years – that is more frequent than the fires in some upland conifer forests that are considered to be exemplars of fire-adapted ecosystems.
The season of burning also varied, with most but not all ecoregions experiencing the most riparian fire in summer. In Southern California, however, more riparian fires began in autumn, presumably reflecting the importance of seasonal “Santa Ana” foehn winds in that area. Relationships between climate and riparian fire were also geographically contingent. In some ecoregions, such as the Southern California Mountains, riparian fire peaked during years of below average precipitation, presumably reflecting the tendency of riparian fuels to dry as discharge declines. But in other settings, the opposite was true: in the Mojave Basin and Range, for example, riparian (and upland) fire accompanied anomalously high precipitation, which allowed for the growth of extensive herbaceous cover in an environment where fuels are normally too discontinuous to carry fire. Riparian fire showed very little relationship to variation in temperature or PDI.
Our results suggest that in at least some regions riparian fire is sufficiently common that its ecological and biogeomorphic impacts should be considered normal rather than exceptional influences. However, the frequency of riparian fire and its relationship to climatic variables are geographically contingent, a factor that must be considered in both ecological interpretation and in planning applications.