M ost of us these days are all too aware of the disruptive impact of hurricanes in human affairs. Yet disturbances ranging from minor local disruptions to massive large-scale catastrophes are part-and-parcel of life in most natural ecosystems (1, 2). These disturbances often provide scientific opportunities, because sometimes one learns the most about how a system functions by watching it recover after it has been kicked by a major disturbance (e.g., ref.3). Ecologists increasingly recognize that the structure of natural communities reflects the interplay of processes acting over a wide range of temporal and spatial scales (4) that are well beyond the scope of manipulative experiments. The article in this issue of PNAS by Schoener and Spiller (5) provides a deft testament to the insights that can sometimes be gleaned from ''natural'' experiments generated by large-scale disturbances, which permit an examination of system responses that could not be readily examined with manipulative experiments.
Responses to DisturbanceSchoener and Spiller (5) provide a portrayal of the impact on spider communities of a major hurricane (Floyd) that in 1998 slammed into a suite of 41 Bahamian islands, completely inundating them and driving multiple extinctions. Studies by Schoener, Spiller, and their associates before Hurricane Floyd provide a rich understanding of many aspects of this system and may indeed provide one of the better-understood terrestrial food webs. By comparing data collected from islands for several years before the hurricane, with an equal number of years after the hurricane, Schoener and Spiller (5) characterize key dimensions of community response to this disturbance. Both before and after the hurricane, some islands had lizards, and others did not. Earlier correlative and experimental studies (see references in ref. 5) found that lizards on these small islands act as effective top predators, limiting spider abundance and species richness. The spider communities on these islands also match a pervasive pattern in community ecology, the species-area relationship, which describes how species richness increases with increasing island area (6, 7). A useful statistic for describing the strength of this relationship is the z value, which is the slope of a regression of log species versus log area. The study by Schoener and Spiller (5) provides an analysis not only of different facets of community resilience to disturbance but also of how the effect of disturbance on the species-area relationship depends on trophic structure.As Schoener and Spiller (5) note, by some measures these communities appear to be highly resilient, but by other measures, they are not. Within 4 years of the hurricane, average species richness had increased to its prehurricane value, but average abundance was still substantially depressed. The former result matches the findings of the celebrated defaunation study of Simberloff and Wilson (8) in the Florida Keys; arthropod communities contain many highly vagile species, and if islands are not t...