Monitoring fire effects for managed burns and wildfires: Coming to terms with pseudoreplication

Collecting unbiased monitoring data on fire effects is often problematic. S amples collected for assessing the effects of managed (prescribed) fires an d wildfires are often "pseudoreplicated" because it is impossible to replic ate the disturbance event. Furthermore, monitoring data for managed fires a nd wildfires may be confounded because it is difficult to randomize the eff ects of fires not under strict experimental control. It is not possible to replicate or randomize large-scale events such as wildfires and many prescr ibed fires, yet there are techniques that can account for some of the bias introduced by these problems. Since monitoring usually involves repeated ob servations; this paper discusses simple time-series analysis, along with tw o common modifications: impact/reference designs and before/after compariso ns. While there are many possible monitoring strategies, most monitoring ef forts are covered by these broad categories. In this paper we attempt to ou tline the assumptions, strengths, and limitations of these methods. We reco mmend four primary strategies to improve the confidence of findings when as sessing fire effects: (1) acknowledge pseudoreplication in the data when it exists; (2) expand the use of managed fire and wildfire data for quantifyi ng fire effects; (3) increase the use of unburned reference sites to improv e the confidence of analyses of fire effects, and (3) in some instances, co nsider treating data taken from multiple fires as independent replicates. T he concepts discussed in this paper are illustrated by examples taken from data sets for prescribed fire effects in Sequoia and Kings Canyon National Parks, California, USA.