Copyright by the Ecological Society of America RAINFALL-CONTINGENT DETECTION OF FIRE IMPACTS: RESPONSES OF BEETLES TO EXPERIMENTAL FIRE REGIMES

Fire is a valuable tool for maintaining biodiversity in many ecosystems throughout the world, particularly tropical savannas. An understanding of the impact of different fire management regimes is essential if desired outcomes are to be achieved. Numerous studies have shown patchy and inconsistent responses of insects to fire. This may be because detection of responses to fire was contingent upon environmental conditions such as rainfall. Here we examine how ground-active beetles respond to savanna fires and how detection of these responses is contingent upon the amount of rain occurring prior to sampling. We also test the extent to which family-level responses are ‘‘taxonomically sufficient’’ to reveal responses detected at the species level. Our study was part of a landscape-scale, replicated field experiment at Kapalga Research Station, Kakadu National Park, in the wet–dry tropics of northern Australia. Over a 7-yr period (1988–1994), ground-active beetles were monitored using pitfall traps. In 1988 and 1989 fires were excluded from all of the 15–20-km2 experimental compartments. From 1990 to 1994, three experimental fire regimes were applied: burning annually early in the dry season (May/June), burning annually toward the end of the dry season (September/ October), and exclusion of fire. Each treatment was replicated three times. Pitfall trapping began at the start of the wet season in 1988 and continued twice a year, in the middle of the dry season (July/August) and early in the wet season (November/December), until the conclusion of the study at the end of 1994. Beetle abundance in dry-season samples was too low and variable for any meaningful statistical analyses. For wet-season samples, repeated-measures analysis of variance revealed that fires late in the dry season significantly reduced the number of ground-active beetle individuals, families, and species. However, detecting this decline was contingent upon substantial amounts of rain falling prior to sampling. In addition, canonical correspondence analysis (CCA) showed that fire intensity and amount of rainfall prior to sampling interacted to change abundance of several common beetle families and species. Fires late in the dry season appeared to disadvantage some common wet-adapted taxa, while favoring other common groups that were dry-adapted. A Mantel test, comparing dissimilarity matrices based on abundances of common families and species, confirmed that analysis at the family level was taxonomically sufficient to reveal most responses found by analysis at the species level. The strong influence of rainfall on detection of beetle responses to fire poses a challenge to managers using such insects to monitor fire-prone ecosystems. Unless monitoring strategies include rainfall in the sampling design, burning prescriptions may not appear to produce desired or consistent outcomes; fire impacts may be there, but go undetected.