Compounded Perturbations Yield Ecological Surprises Robert T. Paine,1* Mia J. Tegner,2 and Edward A. Johnson3 1Department of Zoology, University of Washington, Seattle, Washington 98195, USA 2Scripps Institution of Oceanography, University of California, San Diego, La Jolla, California 92093-0201, USA and 3Department of Biological Sciences and Kananaskis Field Station, University of Calgary, Calgary, Alberta, Canada T2N 1N4. ABSTRACT All species have evolved in the presence of distur- bance, and thus are in a sense matched to the recurrence pattern of the perturbations. Conse- quently, disturbances within the typical range, even at the extreme of that range as defined by large, infrequent disturbances (LIDs), usually result in little long-term change to the system���s fundamental character. We argue that more serious ecological consequences result from compounded perturba- tions within the normative recovery time of the community in question. We consider both physi- cally based disturbance (for example, storm, volca- nic eruption, and forest fire) and biologically based disturbance of populations, such as overharvesting, invasion, and disease, and their interactions. Dis- persal capability and measures of generation time or age to first reproduction of the species of interest seem to be the important metrics for scaling the size and frequency of disturbances among different types of ecosystems. We develop six scenarios that de- scribe communities that have been subjected to multiple perturbations, either simultaneously or at a rate faster than the rate of recovery, and appear to have entered new domains or ������ecological surprises.������ In some cases, three or more disturbances seem to have been required to initiate the changed state. We argue that in a world of ever-more-pervasive anthro- pogenic impacts on natural communities coupled with the increasing certainty of global change, compounded perturbations and ecological surprises will become more common. Understanding these ecological synergisms will be basic to environmental management decisions of the 21st century. Key words: altered community states dispersal multiple disturbances recovery intervals scaling disturbances. INTRODUCTION All natural assemblages are perturbed by both physi- cal and biological forces. These agents of change occur with different intensities, frequencies, and spatial distributions. Some essentially scour the landscape, resetting the successional clock to time zero. More commonly, disturbances leave a residual assemblage that provides a legacy on which subse- quent patterns build. We consider the range of single perturbations, from small-scale/frequent dis- turbances to large/infrequent catastrophes, to be central to much traditional ecology such directional or cyclical changes stimulated the development of ecology���s first paradigm, succession (Cowles 1899 Clements 1905, 1916). A century of accumulated detail on the interplay between pattern and process has provided descriptors for the nature of succes- sional change and system-dependent rates of recov- ery. There are few surprises embedded here: depend- ing on the time frame of interest, species arrive and depart, canopies or other structures develop, and the system ������recovers,������ converging on the predistur- bance state at a rate reflecting the spatial extent and intensity of the disruptive forces. Such patterns have been extensively reviewed (Pickett and White 1985), and variation in recovery dynamics can be attributed to different processes acting indepen- dently or in concert (Drury and Nisbet 1973 Con- Received 14 July 1998 accepted 18 September 1998. *Corresponding author e-mail: painert@zoology.washington.edu Ecosystems (1998) 1: 535���545 ECOSYSTEMS r 1998 Springer-Verlag 535

nell and Slatyer 1977). Even large, infrequent distur- bances (LIDs) do not appear to override the biotic mechanisms that structure eventual recovery. For example, the 1988 Yellowstone National Park fire, which burned 36% of the park and was an order of magnitude larger than comparable large, infrequent fires, has to date generated no ecological surprises: ������the postfire ecosystems are shaping up to be essen- tially the same as those that prospered before the flames������ (Stone 1998: 1527). We argue that cycles of disruption and recovery are the usual state of affairs and submit that rapidly compounded perturbations have more serious implications for long-term alter- ations of community state, occasionally or even often generating a different assemblage of species. Physical agents of change are well documented and described by such terms as windstorm, land- slide, forest fire, flood, hurricane, and volcanic eruption. Many of these are primarily of terrestrial importance and leave their signature on landscapes as sites with recognizable boundaries and measur- able shapes and areas. Biologically based counter- parts���clear-cutting of terrestrial forests and trawl- ing on the ocean floor���generate similar map properties. Populations are also subject to biological disturbances that vary from slight to catastrophic. Although these may lack spatially discrete bound- aries, their implications for community structure can be at least as profound. Here we combine, when appropriate, biological disturbances like pestilence, population eruptions, invasions, and overharvest- ing with the more traditional physical forms of disturbance. In so doing, we add an animal and therefore a trophic dimension to a subject tradition- ally dominated by plant ecologists. Figure 1 is a heuristic portrayal of our approach. In the top panel, a single large disturbance is followed by eventual return to some baseline condi- tion at which point the assemblage can be consid- ered ������recovered.������ The diverse literature on succes- sion is primarily concerned with this pattern and its underlying mechanisms. The following panels iden- tify our focus. In the middle panel, two large disturbances are shown to occur nearly simulta- neously or in close progression. We believe that recovery, if possible, is often substantially delayed under such conditions, and we provide examples below. In the bottom panel, a major disturbance is superimposed on an assemblage already maintained in an altered state, usually by anthropogenic pro- cesses. Current examples could include populations depressed by persistent overfishing, whole systems altered by chronic pollutants, or the developing impacts of climate change, such as the apparent increases in frequency and intensity of major storms and other climate extremes with increasing tempera- tures [for example, see Flavin (1996)]. Jansson and Velner (1995: 332), for example, suggest that in the Baltic Sea, a brackish body of water with minimal connection to the North Sea, which has been heavily impacted by eutrophication and toxic inputs, ������pollu- tion has reached the point where damage may be irreversible.������ The scenarios discussed next include systems that appear, albeit temporarily, to have entered a new ecological domain that is, they have not recovered. They share two features in common. First, all have been subjected to large (based on duration or spatial Figure 1. Schematic representation of the effects of large, infrequent disturbances (LIDs) on community state. Top, A normal community is subjected to a single LID and subsequently recovers. Middle, A normal community undergoes a second (or multiple) disturbance(s) before recovery from the first is completed the combined effects lead to long-term alteration in community state. Bottom, A major disturbance is superimposed on an assemblage already already altered by anthropogenic processes or disease again the combination of stresses leads to long- term alteration of comunity state. Arrowheads mark the disturbances. 536 R. T. Paine and others