587 Reports Ecology, 86(3), 2005, pp. 587–593 2005 by the Ecological Society of America FIRE GENERATES SPATIAL GRADIENTS IN HERBIVORY: AN EXAMPLE FROM A FLORIDA SANDHILL ECOSYSTEM TIFFANY M. KNIGHT 1 AND ROBERT D. HOLT University of Florida, Department of Zoology, 223 Bartram Hall, P.O. Box 118525, Gainesville, Florida 32611-8525 USA Abstract. Disturbance can directly affect the interactions among species at different trophic levels. Because disturbances are typically localized, and many consumers are con- strained in their mobility, disturbances can generate spatial variability in trophic interac- tions. Here, we consider how fire alters plant–insect interactions in a longleaf pine ecosystem in central Florida, USA. We hypothesized that fire in the short term would directly depress herbivore abundance, and that the rate of re-colonization by herbivores would depend upon distance from the edge of the recently burned area. Thus, fire should generate spatially varying herbivory. Orthopterans are dominant insect consumers in our system, and many species are flightless, either as juveniles or adults; these species are particularly likely to show constrained post-fire mobility. We quantified insect abundance and herbivory levels on eight common understory plants in the edge and interior of three recently burned sites. As predicted, insect abundance at the edge of burned areas was much higher than in the interior. In addition, all plant species experienced at least twice the level of herbivory in the edge than in the interior of burned sites. This demonstrates that disturbance can create strong spatial variation in the magnitude of trophic interactions. We suggest that larger burns may aid in plant management by reducing herbivory, whereas the presence of un- burned refugia may be critical to insect conservation. Key words: disturbance ecology; fire; Florida (USA) pine ecosystem; herbivory; Orthoptera; prescribed burn; sandhill ecosystem; spatial gradient; trophic interactions. INTRODUCTION Disturbances influence the structure and function of many communities (Petraitis et al. 1989, Shea et al. 2004). Indeed, disturbances are often defined as dis- crete events that disrupt ecosystem, community, or population structure and change resource, substrate availability, or the physical environment (sensu White and Pickett 1985). Because disturbances are almost al- ways spatially localized in extent and intensity (Turner et al. 2001), they create spatial structure in a landscape; areas disturbed by some event are typically adjacent to areas less disturbed by that same event. Moreover, dis- turbances may differentially affect the abundance of species in different trophic levels (Power et al. 1996, Wootton 1998, Moore et al. 2004), and can therefore alter the intensity of trophic interactions and alter spa- tial patterns of species diversity and composition through disturbance-induced edge-effects (Fagan et al. 1999, Cantrell et al. 2001). Disturbances that are large in spatial extent may have qualitatively different effects on species abundances than smaller disturbances (Romme et al. 1998). For example, gaps created by a single treefall are generally filled by understory trees present before the distur- Manuscript received 2 July 2004; revised 19 August 2004; accepted 27 August 2004. Corresponding Editor: D. P. C. Peters. 1 Present address: Washington University, Department of Biology, Campus Box 1137, Saint Louis, Missouri 63130 USA. E-mail: tknight@biology2.wustl.edu bance, whereas gaps created by multiple treefalls pro- vide adequate light to allow recruitment of new indi- viduals from seed (e.g., Lorimer 1989). Here we sug- gest that large disturbances may also allow for edge effects, since displaced organisms recolonize the edge first, and then the interior. For example, if predators are displaced by disturbance and then recolonize, prey will experience higher levels of predation near the edge than near the interior of the disturbed site. This would create spatial heterogeneity in trophic interactions, the duration of which will depend on both the size of the disturbance and the dispersal ability of the displaced species. In this study we examine how fire influences the spatial pattern of trophic interactions between herbiv- orous insects and plants in a longleaf pine ecosystem, which was historically maintained by periodic fires (Platt et al. 1988). Effects of fire on plant diversity and community composition are well documented, as are the effects of temporal burn patterns (e.g., Menges and Hawkes 1998). However, less is known about how fire influences other trophic levels, such as herbivorous in- sects (Swengel 2001). Even less is known about how trophic interactions vary spatially within burned sites (but see Vickery 2002). Fire may directly and indirectly affect the abundance and diversity of herbivorous insects (Swengel 2001). In the short term, fire negatively affects insect numbers; many are killed by fire, or die of starvation before their host plants reemerge or germinate post-fire. Less mo-