Scope 47- Long-term Ecological Research, An International Perspective   

5.1 INTRODUCTION

The mandate of the Institute of Ecosystem Studies (IES) is to study the disturbance and recovery of north temperate ecosystems. Although 'ecosystem' is the overarching label, our interests include populations, communities, and landscapes as well as resources and stress factors. In this chapter, the term 'system' is used to encompass this multiplicity of subjects. One of the strategies for fulfilling the mandate of the IES is to employ long-term studies (LTS). Because LTS can be a major commitment of personnel, land, facilities, and money, the IES chose first to study the characteristics of long-term studies. We hope this effort will help us to design and conduct better LTS than we otherwise might. This chapter shares the experience we have gained. Much of the background comes from the IES second Occasional Publication (Strayer et al., 1986), which laid out the kinds of LTS, and sought to determine what contributed to their success. The chapter also examines alternatives to LTS which were explored in much greater depth in the 1987 Cary Conference, entitled Long-Term Studies in Ecology: Approaches and Alternatives . That conference, which drew on all aspects of ecology from populations through communities to ecosystems, had an international scope, permitting a thorough analysis of the features of long-term studies (Likens, 1989). In addition, my comments will be based to some extent on my experience in contributing to a 31-year study of plant succession.

This chapter also illustrates the nature of successful LTS, and explores the motivations of such studies. Whether the success of LTS can be unambiguously founded on rigorous a priori arguments, or whether it is an empirical matter, will also be addressed. The kinds of important questions remaining for LTS, especially those that are only apparent from LTS, will be presented. Finally, the advice of the participants of the 1987 Cary Conference on methods will be summarized.

Strayer et al. (1986) sought input on LTS from practicing ecologists for a survey that was initially intended to be a rigorous, perhaps statistical, study of LTS. However, it became clear that the difficulty of ferreting out a representative sample of LTS, including those that had failed, would make the original intent impossible. Therefore, the study evolved into a survey of existing LTS, using questionnaires and interviews with the scientists conducting them. The first insight that emerged was a catalog of the kinds of phenomena amenable to LTS. These included slow, rare, subtle, and complex phenomena. The nature of LTS that emerged from the survey is summarized in this chapter.

5.2 THE SUBJECTS OF LONG-TERM STUDIES

Long-term studies are designed to capture the effects of the environment and biotic interactions in ongoing ecological processes. Such conditions and interactions are likely to change through time. The past is composed of the actual prior environmental conditions affecting a system. A documented past is labelled 'history', but even when it is undocumented, the past may still have an influence on present and future states and trajectories of ecological systems. The environmental conditions that affect a system through time can be divided exhaustively into the following three categories (Pickett, 1989). The first two apply to all systems, while the third applies to systems having a clear beginning.

1. Constraining or enabling conditions, external to the system, can affect its performance. These can be called boundary conditions.
2. The particular order of events over time can influence the subsequent trajectory of a system.
3. Conditions that hold at the beginning of a system, e.g. a succession or anthropogenic impact, can affect the subsequent trajectory of the system. Such conditions can be labeled 'initial conditions'. Initial conditions apply to systems that have a demarcated start.

The different ways that conditions can affect a system over time can be summarized diagrammatically (Figure 5.1). To emphasize that the effects of a past environment may not be obvious in a study of conditions that are part of the current environment of a system, the concept of 'echoes of the past' serves as a useful flag (Garcia Novo, personal communication).

The goal of LTS is to document changing environmental influences and system states before they become lost to the historical record. The following four kinds of phenomena are profitably addressed by LTS.  

Figure 5.1 Schematic representation of the kinds of influences on an ecological process or system through time. Initial and boundary conditions and priority effects are defined in the text. Initial and boundary conditions are separated in cases where a process has a demarcated start. The large horizontal arrow represents the course of a system through time, and the small boxes enclosed represent specific system states whose order can influence the outcome of the entire process. The vertical line separating the head from the body of the arrow represents the present, with the past and future to the left and right respectively. Long-term studies can capture the environmental conditions and system states determining the trajectory of a process before they are lost to an unrecorded past.

5.2.1 SLOW PHENOMENA

Slow phenomena are those that take a long time to occur. The practical definition of long-term will be persistence beyond the usual limits of funding cycles, completion of a graduate degree, or the length of time 'hot' ideas remain fashionable. LTS are defined as those lasting more than 5 years. Community and ecosystem succession is the paradigm of a slow process. Space-for-time substitution, or chronosequence, is the commonest tactic for studying succession. However, chronosequences have often proven to be misleading, and long-term studies were recognized early to be necessary for the understanding of succession (Clements, 1916). In salt marsh succession, permanent plots and longer-term sediment cores have documented that the expected autogenic and deterministic patterns were not found (Clark, 1986; Niering, 1987). On stripmined sites in Pennsylvania, the long-accepted  chronosequence was found to reflect, in part, different initial conditions. Sites used to construct the chronosequence were found to differ in initial acidity of the soil as a result of the recent mining of deeper, less weathered strata (R.S. Hedin, in Pickett, 1989).

In spite of the failings of chronosequences, their value is clear if certain limits are kept in mind. In the case of oldfield succession, space-for-time substitution has documented general trends. For instance, trends in life history types, the order of dominant species, convenient 'stages', and regional differences have emerged from chronosequences. Many of these insights could just as well have come from LTS, of course. However, the understanding that has emerged from the few LTS of oldfield succession is of a different sort. LTS are beginning to expose the nature of transitions, the role of year-specific conditions, the problems with end points, and the role of newly invading exotic species in succession (Pickett, 1989). Too few cases of yearly variation have been studied at this time to produce empirical principles to understand them.

Functional parameters ( e.g. nutrient availability, plant-animal interaction) have been conspicuously underrepresented in studies employing space-for-time substitution (Vitousek, 1977; Thorne and Hamburg, 1985). Chronosequences have been most often employed to assess structural or compositional aspects of ecological systems. Perhaps functional parameters are difficult to assess other than by LTS. The role of debris dams and organic storage in streams is a novel concern that has emerged only as long-term records of disturbance and sediment trapping have accumulated. The cascading effects of trophic interactions on productivity in lakes (Carpenter et al., 1985; Carpenter, 1988; Mills and Forney, 1988) have appeared in LTS. The mechanism has been elucidated using models and retrospective studies. Other indirect effects (Pace and Cole, 1989) have appeared as a result of LTS. The persistent effect of salt marsh wrack, potentially including biogeochemistry (Valiela, 1989), is another example. Because indirect effects often involve organisms with different longevities, long-lasting ecosystem components, or persistent site conditions, LTS have been important in their elucidation. A potentially important, although neglected, area is the long-term changes in the genetics of populations. Such changes may well affect other ecological processes (Pace and Cole, 1989). The reliable aging of many long-lived animals requires LTS (Taylor, 1989; McAninch and Strayer, 1989). Here the slow dynamics of cohorts is the limiting factor.

Transient phenomena can be missed or misinterpreted by short-term studies. Tilman (1989) found that 70% of ecological experiments last less than 1 year. Such limited duration might miss important results, or worse, cause the results to be misinterpreted (Carpenter, 1988). Tilman (1989) discusses the results of Tourney and Keinholz's (1931) famous trenching experiments in forest understory with a study of the plots 21 years later (Lutz, 1945). While the basic conclusion of the importance of root competition held, the role of light and the 'winning' species had to be re-evaluated. Transient effects are likely to appear in many systems as they equilibrate with new experimentally imposed environmental regimes.

Interactions between two species or components of a system that are mediated by a third party or component are called 'indirect effects'. Indirect effects are those incidental to a direct ecological interaction. Such indirect effects might not appear in the short term. For example, herbivores may have effects on an ecosystem other than the consumption of plants (Shachak et al., 1987; Naiman, 1988). Brown et al. (1986) discovered unsuspected indirect interactions among unrelated taxa in the seed-based food web of Arizona deserts using experimental LTS. It took approximately a decade to expose the indirect effects (Brown et al., 1986) because they are slow to resonate through complex trophic webs. As more information becomes available and indirect effects are more widely appreciated, perhaps they will be easier to address in short-term studies.

A more obvious problem than slow and transient phenomena in the short duration of most ecological experiments is the variation in outcome resulting from environmental variability. Experiments in different years often yield different results (Tilman, 1989). Differences in initial conditions which prevail when the experiment is established and changing boundary conditions during the experiment (Figure 5.1) (Pickett, 1989) can be important factors. Some variability may be accommodated by repeating studies at different times if no persistent priority effects (Paine, 1987) operate in the system. It is clear from the specific cases presented, and the classification of kinds of histories and indirect effects that might impinge on ecological systems, that LTS are the only sure way to determine slow processes.

5.2.2 RARE EVENTS

Rare events are the second major type of process amenable to LTS. Such events are those that occur infrequently, and may be either periodic and predictable or unpredictable. An extreme type of rare event is one that is unique, that is, unprecedented and unrepeated. If one is lucky, rare events that are predictable to some degree can be captured in extensive spatial networks of short-term studies. Phenomena such as periodic recruitment, periodic mortality, or small gap disturbance are amenable to extensive short-term sampling (Franklin, 1989).

Unique events, such as the invasion of an exotic species or a disease, are only available to direct long-term study (Taylor, 1989). The anthropogenic displacement of ecological systems to new states is a unique event. It is not likely that concepts of resistance and resilience will be completely successful in anticipating unique events, although measures of resistance and resilience may help to explain the response of different systems to extreme or unique events. Unfortunately, only LTS can capture unique events.

Other forms of rare events have been uniquely illustrated by LTS. Description of disturbance regimes provides an example of the relationship of LTS to rare events. The documentation of single- or few-treefall gaps by extensive short-term surveys has been successful (Runkle, 1982), and assuming equilibrium, the small-gap aspect of the disturbance regime can be characterized. However, testing the assumption of equilibrium in the disturbance regime, and determination of the spatial and temporal distribution of large gaps or catastrophes, has yielded only to direct LTS (Falinski, 1977; Runkle and Yetter, 1987). 

Issues concerning which species fill gaps, and whether advanced regeneration of new invaders, sprouts, or seedlings succeed in different kinds of gaps, have been addressed definitively by LTS, but only hypothetically by short-term studies. The question of whether entire landscapes are in equilibrium under the influence of large-patch disturbances is unanswerable by chronosequence. One may assume equilibrium in a sufficiently large area, but the assumption must be tested. If catastrophic or large disturbances are clustered in time or space, only direct LTS or, in some cases, historical or paleoecological reconstruction is productive (Romme, 1982). The question of whether (or what) ecological systems are in equilibrium is a crucial one (Remmert, 1985; Shugart, 1989), and the prevalence of climate change over centuries and millennia (Davis, 1983) casts doubt on the efficacy of short-term alternatives to LTS to answer the question.

5.2.3 SUBTLE PROCESSES

Subtle processes are embedded in a variable matrix and, therefore, cannot be extracted without a long record (Strayer et al., 1986). While an overall or net trend may exist, the temporal variance will obscure the trend in a short-term study. The long-term record of CO₂ in Hawaii is a case in point. Systems or processes strongly influenced by climate are likely to exhibit subtle behavior (Franklin, 1989). For instance, what constitutes a normal deviation in a successional trajectory? What is a catastrophic yearly decline in fish stocks in a lake? Or, relevant to public concern over the drought of 1988 in North America, what differentiates a normal extreme year from an entirely new trajectory? If such questions about climate are troubling, how much worse is the uncertainty over normal (but periodic) tree mortality as opposed to irretrievable forest dieback? As in the case of slow processes, a sound underlying mechanistic understanding of the process can help extract trends in subtle processes from the temporal variance. However, LTS are still likely to be critical in generating or testing such understanding. The ability to discriminate normal from unusual ecological events, as meteorologists do, would be a powerful tool for management (Pace and Cole, 1989).

5.2.4 COMPLEX PROCESSES

Complex processes are those that have multiple causes. How can causality be evaluated in large systems or areas that cannot be manipulated? This question can be answered by observing the system over a time long enough to encompass periods when different causes dominate its structure or function. For systems that are replicated adequately in space, a comparative approach may answer the question of control by different causes. Population regulation is perhaps the paradigm of a multivariate problem. Populations can be controlled by production, predation and herbivory, competition, or dispersal, among other factors. The abiding controversies about population regulation (Andrewartha and Birch, 1984) attest to the complexity of the situation. Taylor (1989) notes that long runs of animal population data have been very helpful in disentangling the roles of various controlling factors. The interaction between wolves and moose on Isle Royale in Lake Superior is a sterling example. Likewise, the widespread declines or extirpations of migrant birds from their breeding ranges has been exposed by LTS, the spatial extent of the samples being an important feature of the data (Leck et al., 1981). The reliable aging of many long-lived animals would also be impossible without LTS (Taylor, 1989; McAninch and Strayer, 1989). It is of interest to note that many such studies of animal populations were begun out of curiosity rather than any theoretical motivations (Taylor, 1989).

5.3 CORRELATES OF SUCCESS

Strayer et al. (1986) extracted two clear administrative correlates of success from their survey of LTS. First, all successful LTS were associated with at least one dedicated leader who apparently felt personally responsible for the project. Second, the successful studies were simple and accommodating in design, which meant they were relatively easy to run, their output could be adapted to various uses, and  ancillary studies could be associated with them.

Long-term studies also exhibit a variety of other features. However, not all of these are associated with all successful studies, nor were they mentioned as critical features by all the scientists who were contacted. These secondary features include experimentation as a part of the LTS, clearly defined objectives, protected sites, archiving of samples for future analysis, short-term justification by published productivity or societal import, and synthesis and modeling during the course of the study.

The third aspect of success, identification of particular ecological processes successfully addressed by LTS, is a difficult matter. In order to determine what processes are most amenable to LTS, the universe of comparison would have to include failed studies. However, only the successful studies are available for consideration, and these cover a broad range which includes population dynamics, community dynamics, and ecosystem processes. The survey conducted by Strayer et al. (1986) indicates that substantial effort has been expended on LTS in all three of these broad areas, and examples of unexpected and important results can be cited for each area (Strayer et al., 1986; Likens, 1989). Specific LTS in each of the broad areas usually address, in order, population regulation, patterns and causes of community dynamics, and regulation of nutrient flow. Emerging interests in LTS include third-party mediation of interactions (i.e. indirect effects), herbivory , and disturbance. Notably, the study of several levels of organization and potentially interacting phenomena at the same site are new trends in LTS that are quite promising. Many insights have emerged from individual studies conceived and conducted by one or a few ecologists, and no productive and fundamental area of ecology has failed to benefit from LTS.

5.4 MOTIVATIONS OF LONG-TERM STUDIES

One of the mandates of the workshop which generated this book was a rigorous examination of the motivations of LTS. The question that must be asked is, ' Are LTS unambiguously necessary on scientific grounds?' This question might be answered unequivocally by falsifying the conditions which are recognized as conducive to LTS. The strategy of attempting to falsify propositions about LTS was proposed by the meeting organizers. In attempting such a strategy, I have accepted the classification by Strayer et al. (1986) of phenomena appropriate to LTS, as it covers most of the six conditions for LTS used in this workshop, and as it has been used by others (Franklin, 1989). Obviously, ecologists have been motivated to perform LTS by the notion that the phenomenon of interest falls into one of the categories of slow, rare, subtle, or complex. However, there are three other ways to conceive a motivation for LTS, and these are theoretical, empirical, and political motivations.

5.4.1 THEORETICAL MOTIVATIONS

In a discussion of the theories that have demanded LTS, succession is a topic that is proposed as having clear theoretical motivation (Odum, 1969; Jackson, 1981; Pickett et al., 1987). Long-term studies have been instrumental in rejecting tenets of the earlier versions of succession theory and in refining the theory. Some specific, theoretically motivated questions have addressed the dominance of facilitation as a mode of species turnover, the existence of equilibrium communities terminating succession (Remmert, 1985), and the role of seed rain versus seed banks. Indeed, the approximately 30-year-old Buell Succession Study was motivated in part by the desire to test Egler's (1954) idea of initial floristic composition (H. Buell, personal communication). The classic experiments at Rothamsted were motivated by a controversy over fertilization (Johnston, 1989) that may be considered theoretical in a broad sense. LTS on population regulation have been stimulated by the persistent debate over density-dependent versus density-independent population regulation (Taylor, 1989). The 1987 Cary Conference did not identify other theories that stimulated initiation of LTS. Perhaps this is because most ecological theories are explicitly equilibrial (Valiela, 1989) and do not address slow processes or lags, indirect effects, or rare events.

5.4.2 EMPIRICAL MOTIVATIONS

The strongest empirical motivation for LTS is the accumulating experience with rare events or continuous change (Pickett and White, 1985; Weatherhead, 1986). Such experience is contrary to much classical thinking in ecology (Valiela, 1989), and has helped to shape a revolution in the discipline. One of the most impressive bodies of data demonstrating the ubiquity and magnitude of long-term dynamics in ecology is the paleoecological record. Species have migrated individualistically since deglaciation (Davis, 1983; Jacobson et al., 1987), and climatic fluctuation has been common for even longer intervals (COHMAP Members, 1988). The various experiments at Rothamsted, including both the classic and modern LTS, illustrate the empirical value of LTS in general. The original goals of several of the Rothamsted experiments have long since been fulfilled (Taylor, 1989; Johnston, 1989); however, they continue to be of value as demonstrations and as sources of continuing insight into agricultural practice (Johnston, 1989). Unexpected environmental factors were found to be important during the course of the studies. Such insight is purely empirical, since it was not anticipated by theory or a priori hypothesis.

The unexpected empirical discovery of major anthropogenic changes is so striking that it continues to motivate ecologists and governments (Franklin, 1989) to pursue LTS. The long-term weather and stream gauging records are clear cases. Networks must be established or improved to capture anthropogenic change and to relate it explicitly to potential change in ecological interactions and systems. Networks should include both modified and still unmodified systems, for even in modified systems synergisms and new stresses may appear, and systems as yet unmodified are unlikely to remain so as population pressures grow and global climate change proceeds. The strength of the empirical motivation of LTS is reflected in the finding of Strayer et al. (1986) that no LTS has been abandoned voluntarily by any of the approximately 100 scientists they surveyed.

5.4.3 POLITICAL MOTIVATION

The need to convince policy or decision makers that some environmental changes such as abatement, mitigation, and prevention are worth attention is a potentially powerful justification of LTS. This motive is, however, difficult for an ecologist to evaluate, and political scientists or historians are perhaps in a better position to determine whether data from LTS have in fact motivated legislative or executive decisions. The reaction of the Reagan administration against the 'let burn' policy of the US Park Service, as a result of the large number of intense fires in the severe drought year of 1988, is an indication that factors other than ecological knowledge (some of which is based on LTS) may interfere with evaluation of any sort of ecological study (Findlay and Jones, 1989).

5.4.4 EXISTING LONG-TERM STUDIES AS A DATA BASE

The desire to evaluate the motivations and conditions under which LTS are unquestionably justified assumes that the data base on LTS is unbiased. Are the LTS surveyed by Strayer et al. (1986) and the contributors to Likens (1989) indeed an unbiased sample? Possibly not. Failed LTS have not been adequately sampled (Strayer et al., 1986). More fundamentally, in spite of any perceived or documented values of LTS, concerns with career 'fitness' and the nature of funding schedules may have limited the number and nature of LTS in the past. For instance, Strayer et al. (1986) note that most LTS are less than 20 years old, and their number has been increasing since the late 1960s.

A second major problem in using the existing body of LTS to guide design of new studies is the assumption that the strategies that worked in the past are valid now and for the future. The social and political environment for research may have changed. The correlation of simplicity and a dedicated single investigator with existing studies may reflect the limits of funding and the past reward system of the ecological community as much as desirable design features. Funding and scientific recognition of LTS, and appreciation of multi-investigator projects, have all increased. Furthermore, society may recognize the benefits of ecological LTS and decide to institutionalize them. Under such circumstances, the attributes of simplicity and a single dedicated leader may not be required for success. While the formulas for success based on existing LTS clearly embody much wisdom, they should not be followed blindly. Rather, the extent to which features of existing LTS reflect a vanished research environment, and the environment that new studies will be likely to encounter, must be evaluated.

 5.5 THE 'NATURAL HISTORY' OF LONG-TERM STUDIES

The value of LTS seems unarguable when based on the sources I have abstracted. The phenomena and motivations listed earlier are all represented among the ranks of successful LTS. Important insights have resulted from LTS where short-term studies had been equivocal, silent, or even wrong (Pickett, 1989; Tilman, 1989; Taylor, 1989).

The existing body of LTS seems to have much in common with the products of biological evolution. Its products are diverse, the features of specific studies are contingent upon the situation from which they arose, and their structure is opportunistic. Diversity of LTS is illustrated by the fact that some are manipulative, while others are observational; some are focused on populations, while others focus on ecosystems; and some document patterns, while others examine mechanisms. Long-term studies are contingent upon context in that some are performed by 'lone wolves' on shoe-string budgets, measuring a modest number of parameters, while others are multi-investigator and multi-site studies which have cost 'megabucks'. Long-term studies are opportunistic in that they take advantage of various combinations of research strategies, including retrospective studies, short-term experiments, and modeling; they may also incorporate historical data and data from other disciplines when it is appropriate and profitable.

This richness of long-term studies was the most striking feature of the survey conducted by Strayer et al. (1986) and it was confirmed by the analyses at the 1987 Cary Conference (Likens, 1989). Indeed, it would appear that the past success of LTS as a class has something to do with its richness and flexibility. Some philosophers and historians argue that the success of science as a whole results from similar flexibility and opportunism (Cohen, 1985; Lloyd, 1987). The flexibility of LTS permits new concerns to be addressed, and the simplicity enhances comparability in time and space. But it is important to recognize the evolving needs for LTS. Geographical breadth and comparability are important features which will be required in the future. A network of planned LTS may differ in important ways from existing LTS, most of which (Strayer et al., 1986) were not originally intended to be long term. For this reason, while still retaining the benefits of the evolved nature of most extant LTS, networks of LTS may require a more systematic approach. It is possible that the opportunistic, ad hoc nature of most existing LTS might, for example, hinder cross-site comparison.

5.6 RECOMMENDATIONS

The participants at the 1987 Cary Conference met in various discussion groups to evaluate LTS and make recommendations for planning and executing future LTS. The recommendations are abstracted here, with citations of the authors of the meeting group reports. The number of participants in the discussions was large, and attempts were made to reach consensus or to note the more important alternatives. The names of all contributors to each discussion are found in the papers cited below (Likens, 1989).

5.6.1 CONCEPTUAL MODELS

Begin with a conceptual model. Although many studies surveyed by Strayer et al. (1986) violated this requirement, some of the most productive and ultimately largest LTS (e.g. those at Hubbard Brook and Coweeta Hydrologic Laboratory) had a motivating conceptual model (Caraco and Lovett, 1989). Eaton and McDonnell (1989) suggested that any experimental LTS must have a conceptual model. I emphasize 'conceptual model' rather than simple hypothesis because it is important to spell out the assumptions and logical claims of argument in order to evaluate the tests of hypotheses. If it is impossible to construct a tentative conceptual model, then experimentation is premature. In considering the establishment of a network of LTS, it would seem even more critical to have a conceptual model. Such a model will guide the selection of sites and will suggest the manipulations to be performed (if it suggests any at all) and the parameters to be measured. A model will also assist in determining what pre-treatment parameters should be measured (Berkowitz et al., 1989). Some situations may suggest alternative models; the rationale for specifying them is the same as that outlined above.

5.6.2 DATA HANDLING

One striking problem with LTS is the overwhelming body of data that can be accumulated. There is already a wealth of relatively unexploited data from existing LTS. Franklin ( 1989) suggested that resources for analysis and synthesis were needed. At the very least, more funding should be made available to permit productive use of existing, high-quality data sets. Even where there may be some problems with the data from existing studies, timely analysis may be the best way to learn what those problems are and to reduce the likelihood of repeating or perpetuating the errors. Most of the LTS on the list accumulated via questionnaire in the Permanent Plotter Newsletter (G.G. Parker, unpublished manuscript) have yet to produce available output. Without timely analysis of the potentially rich resource of existing long-term data, we may be reinventing wheels.

5.6.3 APPLICATION TO EXPERIMENTS

Long-term work is required even in field experimental work (Tilman, 1989). To avoid the conclusion that transient dynamics are the ultimate outcome of an experiment, they must be followed for longer periods; to this end, ecological research that is not necessarily designed to address one of the traditional areas of concern of LTS (e.g. succession, population dynamics, stress effects) should be conducted for longer periods than is now common. This may require protection, and commitment of sites, and funding for longer periods than was common in the past. To summarize this point, ecology as a whole can benefit from extending its temporal scope.

5.6.4 SPATIAL EXTENT

A broader spatial extent is required in LTS as it is in all ecological research. A high degree of generality of some patterns and phenomena is now emerging, for example, for acid rain and natural disturbance. However, most ecological research, whether short term, long term, observational, or experimental, is conducted at single sites.

The lack of comparison saps the ability to evaluate the degree of generality of the results (Bradshaw, 1987). The power of the large networks of dated paleoecological cores is a convincing argument for comparison among LTS (COHMAP Members, 1988). National parks would make ideal components for such networks (Parsons, 1989).

5.6.5 QUESTIONS MANDATING LONG-TERM STUDIES

Discussion at the 1987 Cary Conference attempted to identify classes of questions that absolutely require LTS. Two conditions were suggested as logically demanding LTS (Pace and Cole, 1989): unanticipated changes, and situations where no surrogate method is available. Transient dynamics or indirect effects are likely to be common among unanticipated changes. Unexpected changes that catastrophically shift systems to new domains are also possible, and are important motivations for LTS (Garcia Novo, 1977). Because such catastrophic shifts have been neglected in ecology, there is neither theory nor empirical generalization to guide the search for these phenomena. Theory and generalization may better develop to incorporate transient and indirect effects, but some effects are simply unprecedented and cannot be anticipated. Unanticipated changes are common among the catalog of anthropogenic effects, and are often startlingly novel. Witness the question of ozone depletion in the stratosphere. The most widespread and significant changes now facing ecology and society are anthropogenic ones. The spread of the megalopolis throughout the world suggests a strategy for establishing networks to capture the changes that are almost certain to result in natural populations, biotic communities, ecosystems, and landscapes. Both terrestrial and aquatic realms are involved, and networks will be needed to answer questions about unexpected anthropogenic changes in all ecological systems. A hub and spoke design to assess gradients of anthropogenic effects on forests centered around a major city is presently being planned (McDonnell and Round tree, unpublished manuscript).

Changes in global climate are the other major class of environmental changes that are likely to be unanticipated. The magnitude and significance of such changes are only now beginning to be appreciated. The role of LTS in documenting such changes and the ecological responses to them is clear (Pace and Cole, 1989). LTS are also required to validate mechanistic and predictive models of global change and ecological results. LTS are mandated when no surrogate method is available (Pace and Cole, 1989). Various contributors to the 1987 Cary Conference (Likens, 1989) evaluated surrogate methods. The evaluations suggest that chronosequences are not uniformly good substitutes for LTS (Pickett, 1989), and retrospective studies cannot forecast unforeseen future changes, although they can give necessary perspective to the type and distribution of rare or slow events in the past (Davis, 1989). Modeling can give insight into suspected trends that may result from future periodic or unique events (Shugart, 1989), but not all societally impactful changes are likely to be suspected by ecologists. Likewise, the applicability of microcosms may be limited by problems of scaling (Shugart, 1989). It would seem unwise to rely upon surrogate methods entirely in the future, since they have not compensated completely for LTS in the past.

Having considered the circumstances in which LTS are called for, several broad questions that require LTS can be presented. Although the number of specific questions that might arise in various specialties is too large to list here, it is possible to indicate the kind of pure ecological question that might be most profitably addressed through LTS. Such questions will deal, in all likelihood, with the degree to which various ecological systems are organized and the role of internal versus external organizing influences. This issue is stated here in an admittedly abstract form, but it can be made operational in various specific ecological contexts (Pickett et al., 1989). The degree of organization will be reflected in the persistence of functions and their attendant structures, in resistance to change, and in resilience after disturbance. Because the temporal dimension is implicit in each of these questions, the use of LTS will be critical.

5.6.6 METHODS

Methods were addressed by many contributors to the 1987 Cary Conference. The resulting recommendations should be especially useful in planning networks of LTS. They should apply to both large multi-investigator projects and the work of a single investigator. Samples should be archived for future analyses not now possible or considered important (Lewis et al., 1984), and for calibration with new techniques that may become available (Strayer et al., 1986; Caraco and Lovett, 1989). LTS should be combined with other methods (Caraco and Lovett, 1989). While direct LTS will usually be the backbone of the effort, new questions may constantly arise (Johnston, 1989). Such questions may often be about mechanisms or details of pattern, and can be profitably addressed with shorter-term experiments associated with a monitoring effort (Pace and Cole, 1989). In general, a combination of methods has been used in LTS studies addressing processes at all levels of organization (Caraco and Lovett, 1989). Variance of forecasts from LTS may be reduced if they are based on a variety of methods (Parker, 1989). Likewise, extrapolation of one LTS to other sites requires a variety of safeguards not yet adequately explored by ecologists (Berkowitz et al., 1989).

Long-term studies must accommodate different scales of variability (Eaton and McDonnell, 1989). Because the scales of events likely to be encountered over a long time span probably differ greatly, the design of LTS should be capable of capturing events on different scales. Attention should be paid to sampling regimes that can be aggregated or disaggregated to fit the scale of a process or structure of interest. The problem of scale deserves increased attention, especially since unanticipated changes may involve scale shifts. For example, not even long-term succession studies have effectively accommodated the changing scales of various growth forms. The parameters measured throughout a study must be selected to be relevant over the range of phenomena expected to impinge on the system. For example, 'standard' plant community parameters and plot sizes are not designed to address non-equilibrium processes and patchiness; nor are they sensitive to architectural changes in the community. The value of LTS will be increased when they overlap paleoecological records (Davis, 1989).

These various recommendations result from the combined experience of many ecologists responsible for the design, execution, and interpretation of existing LTS. Because of the diversity of approaches, levels of focus, and questions addressed by these LTS (Strayer et al., 1986), it may be that the recommendations encapsulate the vast majority of problems likely to exist in LTS. The recommendations should be of great value in avoiding problems in work that is explicitly designed to contribute over the long term.

5.7 ACKNOWLEDGMENTS

I appreciate the comments of Dave Strayer, Gene Likens, Johnny Johnston, Paul Risser, and Mark McDonnell. This paper is a contribution to the program of the Institute of Ecosystem Studies with financial support from the Mary Flagler Cary Charitable Trust.

5.8 REFERENCES

Andrewartha, H.G. and Birch, L.C. (1984). The Ecological Web: More on the Distribution and Abundance of Animals. University of Chicago Press, Chicago.

Berkowitz, A.R., Kolasa, J., Peters, R.H. and Pickett, S T.A.. (1989). How far in space and time can the results from a single long-term study be extrapolated? In Likens, G.E. (Ed.) Long-Term Studies in Ecology: Approaches and Alternatives. Springer-Verlag, New York, 192-198.

Bradshaw, A.D. (1987). Comparison - its scope and limits. New Phytologist, 106 (supplement), 3-21.

Caraco, N.M. and Lovett, G.M. (1989). How can the various approaches to studying long-term ecological phenomena be integrated to maximize understanding? In Likens, G.E. (Ed.) Long-Term Studies in Ecology: Approaches and Alternatives. Springer-Verlag, New York, 186 -188.

Carpenter, S.R., Kitchell, J.F. and Hodgson, J.R. (1985). Cascading trophic interactions and   lake productivity. BioScience, 35, 634-639.

Carpenter, S.R. 1988. Transmission of variance through lake food webs. In Carpenter, S.R. (Ed.) Complex Interactions in Lake Communities. Springer-Verlag, New York, 119-135.

Christensen, N.L. and Peet, R.K. (1981). Secondary forest succession on the North Carolina Piedmont. In West, D.C., Shugart, H.H. and Botkin, D.B. (Eds) Forest Succession: Concept and Application. Springer-Verlag, New York, 230-245.

Clark, J.S. (1986). Late-Holocene vegetation and coastal processes at a Long Island tidal marsh. Journal of Ecology, 74,561-578.

Clements, F,E. (1916). Plant Succession. Camegie Institution of Washington, Publication No.242. Washington, DC.

Cohen, I.B. (1985). Revolution in Science. Harvard University Press, Cambridge, Massachusetts.

COHMAP Members (1988). Climatic changes of the last 18,000 years: observations and model simulation. Science, 241, 1043-1052.

Davis, M.B. (1983). Quaternary history of deciduous forest of eastern North America and Europe. Annals of the Missouri Botanical Garden, 70, 550-563.

Davis, M.B. (1989). Retrospective studies. In Likens, G.E. (Ed.) Long-Term Studies in Ecology: Approaches and Alternatives. Springer- Verlag, New York, 71-89.

Eaton, J.S. and McDonnell, M.J. (1989). What are the difficulties in establishing and interpreting the results from a long-term manipulation? In Likens, G.E. (Ed.) Long-Term Studies in Ecology: Approaches and Alternatives. Springer-Verlag, New York, 189-191.

Egler, F.E. (1954). Vegetation science concepts. I. Initial floristic composition, a factor in old field vegetational development. Vegetatio, 4, 412-417.

Elliott, J.M. (1985). The choice of a stock-recruitment model for migratory trout, Salmo trutta, in an English Lake District stream. Arch. Hydrobiol., 104, 145-168.

Falinski, J.B. (1977). Research conducted in vegetation and plant population dynamics conducted in Bialowieza Geobotanical Station of Warsaw University in the Bielowieza Primaeval Forest. Phytocoenosis, 6, 5-148.

Findlay, S.E.G. and Jones, C.G. (1989). How can we improve the reception of long-term studies in ecology? In Likens, G.E. (Ed.) Long-Term Studies in Ecology: Approaches and Alternatives. Springer-Verlag, New York, 201-202.

Franklin, J.F. (1989). Importance and justification of long-term studies in ecology. In Likens, G.E. (Ed.) Long-Term Studies in Ecology: Approaches and Alternatives. Springer-Verlag, New York, 3-19.

Garcia Novo, F. (1977). The ecology of vegetation of the dunes in Doñana National Park (South-west Spain). In Jefferies, R.L. and Davy, A.J. (Eds) Ecological Processes in Coastal Environments. Blackwell, Oxford, 571-592.

Hamburg, S.P. and Cogbill, C.V. (1988). Historical decline of red spruce populations and climatic warming. Nature, 331, 428-431.

Jackson, J.B.C. (1981). Interspecific competition and species distribution: the ghosts of theories and data past. American Zoologist, 21, 889-901.

Jacobson, G.L., Webb, III, T. and Grimm, E.C. (1987). Patterns and rates of vegetation change during the deglaciation of eastern North America. In Ruddiman, W.F. and Wright, E.H. (Eds) North America and Adjacent Oceans During the Last Deglaciation. Geological Society of America, Boulder, Colorado, 277-288.

Johnston, A.E. (1989). The value of long-term experiments - a personal view. In Likens, G.E. (Ed.) Long-Term Studies in Ecology: Approaches and Alternatives. Springer-Verlag, New York, 175-179.

Leck, C.F., Murray, B.G. Jr. and Swinebroad, J. (1981). Changes in breeding bird populations at Hutcheson Memorial Forest since 1958. William L. Hutcheson Memorial Forest Bulletin, 6, 8-15.

Lewis, R.A., Stein, N. and Lewis, C.W. (Eds) (1984). Environmental Specimen Banking and Monitoring as Related to Banking. Nijhof, Boston.

Likens, G.E. (Ed.) (1989). Long-Term Studies in Ecology: Approaches and Alternatives. Springer- Verlag, NewYork                                                                                                                                                                                                           

Lloyd, E.A. (1987) Confirmation of ecological and evolutionary models. Biology and Philosophy, 2, 227-293.

 McAninch, J.B. and Strayer, D,L. (1989). What are the tradeoffs between immediacy of management needs and the longer process of scientific discovery? In Likens, G.E. (Ed): Long-Term Studies in Ecology: Approaches and Alternatives. Springer-Verlag, New York 203-205.

 Mills, E.L. and Forney, J.L. (1988). Trophic dynamics and development of freshwater pelagic foodwebs. In Carpenter, S.R. (Ed.) Complex Interactions in Lake Communities. Springer Verlag, New York, 11-30.

Naiman, R.J. (1988). Animal influences on ecosystem dynamics. BioScience, 38, 750-752.

Niering, W.A. (1987). Vegetation dynamics (succession and climax) in relation to plant community management. Conservation Biology, 4, 287-295.

Odum, E.P. (1969). The strategy of ecosystem development. Science, 164, 262-270.

Pace, M.L. and Cole, J.J. (1989). What questions, systems, or phenomena warrant long-term ecological study? In Likens, G.E. (Ed.) Long-Term Studies in Ecology: Approaches and Alternatives. Springer-Verlag, New York, 183-185.

Paine, R.T. (1987). Controlled manipulations in the marine intertidal zone, and their contributions to ecological theory. Special Publication of the Academy of Natural Sciences of Philadelphia, 12, 245-270.

Parker, G.G. (1989). Are currently available statistical methods adequate for long-term studies? In Likens, G.E. (Ed.) Long-Term Studies in Ecology: Approaches and Alternatives . Springer- Verlag, New York, 199-200.

Parsons, D.J. (1989). Evaluating national parks as sites for long-term studies. In Likens, G.E. (Ed.) Long-Term Studies in Ecology: Approaches and Alternatives. Springer-Verlag, New York, 171-173.

Pickett, S.T.A. (1989). Space-for-time substitution as an alternative to long-term studies. In Likens, G.E. (Ed.) Long-Term Studies in Ecology: Approaches and Alternatives. Springer-Verlag, New York, 110-135.

Pickett, S.T.A., Collins, S.L. and Armesto, J.J. (1987). Models, mechanisms and pathways of succession. Botanical Review, 53, 335-371.

Pickett, S.T.A. and White, P.S. (Ed.) (1985). The Ecology of Natural Disturbance and Patch Dynamics. Academic Press, Orlando.

Pickett, S.T.A., Kolasa, J. Armesto, J.J. and Collins, S.L. (1989). The ecological concept of disturbance and its expression at various hierarchical levels. Oikos, 54, 129-136.

Remmert, H. (1985). Was geschiet im Klimax-Stadium? Naturwissenschaften, 72, 505-512. 

Romme, W.H. (1982). Fire and landscape diversity in subalpine forests of Yellowstone  National Park. Ecol. Monogr., 52, 199-221.

Runkle, J.R. (1982). Patterns of disturbance in some old-growth, mesic forests of eastern North America. Ecology, 63, 1533-1546.

Runkle, J.R. and Yetter, T.C. (1987). Treefalls revisited: gap dynamics in the southern Appalachian Mountains. Ecology, 68, 417-424.

Shachak, M., Jones, C.G. and Granot, Y. (1987). Herbivory in rocks and the weathering of a desert. Science, 236, 1098-1099.

Shugart, H.H. (1989). The role of ecological models in long-term ecological studies. In Likens, G.E. (Ed.) Long-Term Studies in Ecology: Approaches and Alternatives. Springer-Verlag, New York, 90-109.

Strayer, D., Glitzenstein, J.S., Jones, C.G., Kolasa, J., Likens, G.E., McDonnell, M.J., Parker, G.G. and Pickett, S.T.A. (1986). Long-term ecological studies: an illustrated account of their design, operation, and importance to ecology. Occasional Publication of the Institute of Ecosystem Studies, Number 2. Millbrook, NY.

Taylor, L.R. (1989). Objective and experiment in long-term research. In Likens, G.E. (Ed.) Long- Term Studies in Ecology: Approaches and Alternatives. Springer-Verlag, New York, 20-70.

Thorne, J.F. and Hamburg, S.P. (1985). Nitrification potentials of an old-field chronosequence in Compton, New Hampshire. Ecology, 66, 1333-1338.

Tilman, D. (1989). Ecological experimentation: strengths and conceptual problems. In Likens, G.E. (Ed.) Long-Term Studies in Ecology: Approaches and Alternatives. Springer-Verlag, New York, 136-157.

Valiela, I. (1989). Conditions and motivations for long-term ecological research: some notions from studies on salt marshes and elsewhere. In Likens, G.E. (Ed.) Long-Term Studies in Ecology: Approaches and Alternatives. Springer-Verlag, New York, 158-169.

Vitousek, P.M. (1977). The regulation of element concentrations in mountain streams in the northeastern United States. Ecological Monographs, 47, 65-87.

Weatherhead, P.J. (1986). How unusual are unusual events? American Naturalist, 128, 150- 154.