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

11.1 INTRODUCTION

Impetus for international collaboration in long-term ecological research (LTER) presently comes from two forces. The United States National Science Foundation has funded big-team research using the so-called 'ecosystem' approach through the 1980s. These large projects in the LTER program are intended to be continuing, and not closed after any particular time. Comparable programs have begun more recently in the Federal Republic of Germany, under the Man and Biosphere program. These big-team projects already benefit by networking among each other. Extension of the network to international collaboration would no doubt bring further benefits.

The other impetus for international collaboration comes from a new awareness among scientists that some ecological problems are global, and from a new idealism in tackling those problems. During the next century human life will be drastically affected by global atmospheric and climate change, and by alterations in land use which will stem from both further changes in agricultural politics and technology and from the atmospheric change. As scientists, we have faith that understanding these processes will improve human response to them. Many researchers now believe that if even quite simple processes can be measured simultaneously all over the world, new insights into global processes will emerge.

Australian ecologists certainly wish to make their contribution to the emerging global effort. Australia is one of the major landmasses in the southern hemisphere, and important to global processes for that reason alone. In addition, Australian soils and landforms are quite similar to those of important parts of the developing world, such as Brazil and Africa south of the Sahara. Much of Australia's ecological knowledge is more relevant to these tropical and subtropical areas than the ecological knowledge gained in the USA, Europe, or the USSR. At the September 1988 annual conference of the Ecological Society of Australia a resolution was passed which concluded: 'In particular, the Australian government is urged to support the establishment of a system of long-term study and monitoring sites throughout Australia. Such a system of sites is likely to be proposed within the International Geosphere-Biosphere Programme as one approach to the forecast of global change.'

At present any potential Australian contribution is stillborn for lack of resources. Although Australian science, including ecology, has an excellent international reputation and contributes substantially to the international peer-reviewed literature, it is funded at less than half the level, as a percentage of gross national product (GNP), of science in the United States or Federal Republic of Germany. Total expenditure by the Australian Research Council (ARC) on all aspects of ecology is in the range of one to two million dollars per year. Within present resources, most current Australian ecological research would have to be closed down in order to institute a single long-term site on the model of the United States or Federal Republic of Germany.

This chapter has four sections. First, the administrative situation for support of long-term ecological research in Australia is summarized. Second, a sample of Australian studies is described. Third, possible objectives of long-term ecological research are considered. Finally, possible scenarios for an Australian long-term ecological research program are considered in light of these objectives. Readers interested mainly in the science policy issues could begin with the final two sections.

11.2 ADMINISTRATIVE SITUATION

First, consider research in the sense of work aimed primarily at generating publishable advances in knowledge. In Australia there is no systematic support for long-term ecological research in this sense. Rather, long-term projects have been associated with individual scientists who have worked consistently at a particular site. Many studies have lasted for 10 to 30 years, in the hands of a single scientist; a very few have passed to a second generation. Koonamore (see below) has passed through several academic generations over 63 years; so far as I know, it is unique in this respect.

When Dr R. W. Johnson (now Director, Queensland Herbarium) was President of the Ecological Society of Australia during 1985 to 1986, he established a register of continuing ecological research sites, which he still holds. Forms giving basic information for this register were returned for 78 sites, 24 of these being forest growth plots of CSIRO Division of Forest Industry. The register is very far from giving complete coverage of continuing research sites in Australia, but nevertheless is a useful resource.

The CSIRO Division of Wildlife and Ecology held a workshop in November 1986 to work towards networks of research sites with a view to global change research. Different authors discussed possible sites and measurements in each of 12 major biomes (Shaughnessy et al., 1988). No further action has been taken. It seems unlikely CSIRO will be able to establish substantial new programs, considering it continues to suffer annual funding cuts, and no funds have yet been allocated for global change research. During 1989 the Australian government has made a special allocation of A$8M to CSIRO for new global change research. Most of this went to atmospheric modeling. In terms of enhanced effort in the terrestrial ecology of global change, the infusion of funds is roughly equivalent to three scientists.

The Australian Research Grants Scheme (ARGS) has been the government's agency for competitive, peer-reviewed science funding in most areas of science, including ecology. ARGS did not actively encourage work on particular topics. It received proposals, had them reviewed, ranked them, and funded some. ARGS attempted to assure 3 years for funded proposals, but it was not uncommon for funding to be taken from a project after or 2 years because new proposals had achieved higher ranking. ARGS did not sequester any funds for long-term ecology, and so far as I can discover never considered the possibility of doing so.

During 1988 ARGS was incorporated into a new Australian Research Council (ARC). The government intends the new ARC to guide research effort more actively into areas of national need. It is possible, therefore, that future Australian ecological research will have a larger element of central control. ARC's formulation of research priorities is still evolving.

Various agencies responsible for land management have monitoring schemes. By monitoring I mean the information collected is not intended mainly for publication, but to be used directly in management. Forest management agencies have forest growth plots. In recent years legislators have asked for the impact of livestock grazing to be monitored. (Comparisons between places thought to have different grazing histories have not produced a very clear consensus about grazing impact, so it has proved difficult for government agencies to demonstrate with certainty that individual leaseholders or landowners are managing badly enough for action to be taken against them.) The Western Australian Rangeland Monitoring System (WARMS) (Holm et al., 1987) included 867 monitoring sites on 92 stations by 1984 (Holm, 1986). Of these, 254 included a photopoint, a belt transect for shrub numbers and dimensions, and observations of soil surface features. The large number of sites is necessary because of the aim of collecting significant amounts of evidence about individual stations. It is intended that sites should be remeasured at 4- to 5-year intervals, but it remains to be seen whether the Western Australia Department of Agriculture can muster the resources to achieve this for such a large number of sites. In general, agencies responsible for land management operate at the level of States or Territories, so there are no Australia-wide schemes of this sort.

11.3 SOME AUSTRALIAN STUDIES

Five Australian long-term sites (Figure 11.1) have been chosen to describe briefly here. They are rather varied, and serve to illustrate different virtues, weaknesses, and difficulties of long-term research. 

Figure 11.1 Locations of the five long-term study areas briefly described here.

11.3.1 THE T.G.B. OSBORN VEGETATION RESERVE AT KOONAMORE

The reserve of 390 ha was established on a pastoral station in South Australia by Osborn in 1925, and has always been operated by the Botany Department of the University of Adelaide (Sinclair, 1986). The area is arid, with mean annual rainfall about 180 mm. The station had been heavily grazed by sheep for about 50 years of European occupation, and the intention was to follow processes of vegetation recovery. The reserve was fenced against sheep and rabbits. There are permanent quadrats, charted from time to time, and photopoints.

Research at Koonamore has fallen roughly into four phases. During the first 10 years much intensive work was done, reported by Osborn et al. (1935) and Wood (1936). There followed a long period when no research was reported. Intermittent recording was maintained, reputedly due to the personal commitment of Constance Eardley. Continued regeneration to 1962 was reported by Hall et al. (1964). Then during the 1970s two students of R.T. Lange made creative use of the long-run records. Noble found ways to estimate biomasses from photos of vegetation, and modeled germination and growth of both ephemerals and perennials in the chenopod shrublands (Noble, 1977; Noble and Crisp, 1980). Crisp examined demography of woody perennials from the records and from size frequency distributions (Crisp and Lange, 1976; Crisp, 1978), showing inter alia that sheep and rabbits had prevented establishment of Acacia aneura for nearly a century in this part of Australia. While short-term experiments can show such a grazing effect, they could not have indicated that it had applied to all germination episodes over an extended time.

Sinclair is currently in charge of the reserve. He is analyzing events of the past 15 years, particularly the consequences of the heavy summer rains of the mid 1970s, which led to widespread establishment of woody plants. It should be noted that Graetz, in the CSIRO draft report on long-term sites referred to earlier, regards Koonamore as not representative of the generality of either saltbush or bluebush types in Australia.

11.3.2 MARIA ISLAND

At Maria Island, off the south-east coast of Tasmania, surface water has been collected fortnightly since 1945. Physical and chemical characteristics of the water have been measured. Maria Island is now part of a network of coastal water sampling stations, managed by Dr Denis Mackey of CSIRO Division of Oceanography.

This long run of simple data showing between-year variation in seasonal cycles at one place has become valuable in interpreting wide-area processes. Satellite imagery and hydrographic cruise information have shown that the dominant feature of south-east Australian waters is the subtropical convergence (STC), running across from the southern end of Tasmania to New Zealand (Harris et al., 1987). The broad zone where subtropical and subantarctic water masses meet has deep mixing, high phytoplankton productivity, and concentrated carbon export to deep water. Water temperature patterns and the timing of spring bloom at Maria Island are sensitive to penetration of subantarctic water up the east coast of Tasmania. This penetration in turn is driven by the zonal westerly winds at Tasmanian latitudes. These winds are cyclical with a mean periodicity of 11 years (Harris et al., 1988). These phenomena in turn are correlated, positively or negatively, with atmospheric pressure at Darwin in the tropics and at Macquarie Island in the subantarctic, with rainfall, lake levels, and trout catches in the Tasmanian highlands, with seal abundance at Macquarie Island, and with catches of spiny lobsters not only off Tasmania but also off New South Wales and New Zealand. The correlations with lobster catches show an immediate effect and also an effect time-lagged by 5 to 7 years, consistent with known life history (Harris et al., 1988). However, the biological basis of any effect on lobster recruitment is not known.

This is an example where correlations among different variables at one location showed up few relationships beyond those that were already well understood. It was the intercorrelations across space that led to substantial advances in understanding.

11.3.3 CAPE BANKS

Cape Banks is the north headland of Botany Bay, within the Sydney metropolitan area. Being so close to the universities in Sydney it has been used intermittently for research for a long time. Since 1972, under the leadership of A.J. Underwood of the University of Sydney, it has become a major center for the experimental ecology of rocky shores. For example, it has been shown that the boundary between the upper limit of algal turf and the lower limit of the barnacle and limpet zone is not determined by predation on molluscan herbivores, as had been expected on the basis of North American work. Rather, limpets are unable to establish lower on the shore where algal productivity is greater, and herbivores are unable to keep plant biomass low (Underwood and Jernakoff, 1981). Since 1972 a total of 98 publications and 32 theses have emerged from Cape Banks.

Most of this work has been experimentation lasting 1 to 3 years (reviewed in Underwood, in press). However, long-term monitoring (mostly not yet published) has been an important complement. For example, severe storms in 1983 stripped the tubeworm Galeolaria off rocks, which were then recolonized by algae and tunicates, and this change of. state appears persistent. Another example is that during 1980 to 1984 the barnacle Chamaesipho columna declined due to a failure of recruitment. This failure was general throughout the 1000 km of the NSW coast.

We need to know not only how frequent extreme events are, but also how widespread. It will be recalled that by the late 1970s, much emphasis was being put on the role of stochastic effects in recruitment in controlling the structure of assemblages. An important recent trend in Australian ecology is to go beyond marveling at the variability of phenomena, and to use nested ANOVA designs to estimate percentages of the variance attributable to different scales in space and time. The first example of this dealt with barnacle recruitment, and was centered at Cape Banks (Caffey, 1985).

At present a program investigating succession in artificially-created tidepools is under way. A peculiarity of this site is that it lies above mean low water, but below mean high water. In consequence it falls into a legislative vacuum, is not anyone's property, and apparently no agency has the power to prevent the public from visiting the site as they please.

11.3.4 MUNMARLARY

The Munmarlary site was established by R.J. Hooper in 1972, and is operated by the Conservation Commission of the Northern Territory .It is in the monsoonal tropics, and the vegetation has a grassy understory beneath a variable canopy of eucalypts. Four different fire regimes have been applied, each to three replicate sites 100 x 100 m, this design being repeated in both forest (conventionally defined in Australia as projective foliage cover of trees 30-70%) and woodland (tree cover 10-30%). Both these vegetation types would be called savanna in tropical Africa or America.

The first 12 years of this study have recently been reported by Bowman et al. (1988). While 12 years may not seem very long term, in this environment it represents 12 fire-cycles of the prevailing fire regime. The four treatments have been yearly fires in early and in late dry season, fires every two years in early dry, and no fires.

To understand the significance of the work it is necessary to appreciate that also present in these landscapes are closed forests or monsoon rainforests (canopy cover > 70% ). These are dominated by species with rainforest rather than sclerophyll affinities. They are sensitive to fire, also less flammable, and at present are patchily distributed in fire-shadow locations. Some experimentation in Africa has indicated that if savannas are protected against fire they tend to be invaded by fire-sensitive species of closed forest (Rose-Innes, 1972). On the strength of this, Stocker and Mott (1981) suggested that much of the monsoonal north Australian landscape was occupied by monsoonal rainforest until an increase in fire frequency about 40 000 years BP, when people are thought to have arrived. A vegetation shift somewhat analogous can be seen in the 190 000-year pollen core from Lynch's Crater in north Queensland (Kershaw, 1986).

On the other hand, other evidence from Africa (Geldenhuys, 1977) and America (Kellman and Miyanishi, 1982; Sarmiento and Monasterio, 1975) suggests changes in fire regime affect vegetation structure considerably, but floristics are mainly controlled by soils.

The Munmarlary experiment has not shown any evidence that substantial shifts in floristics follow fire exclusion (Bowman et al., 1988). However, a considerable woody understory develops, from species that were present before, but with suppressed height development. There are substantial differences between replicates, possibly attributable to soils effects.

Munmarlary shows the importance of manipulated treatments, with proper replication, in long-term studies. It also exposes the limitations of studies concentrated at one place in a landscape. The evidence now available from three continents strongly suggests that fire exclusion can allow closed forest to develop in some locations, but not in others. What we need now is to learn rules to identify parts of landscapes which can potentially become closed forest if protected against fire, and studies at single sites will not achieve this.

11.3.5 MYALL LAKES

In Myall Lakes National Park coastal sclerophyll forest and heath is found on a sand mass created about 6000 years BP. The coastal dunes are mined for heavy minerals. Sand-mining involves bulldozing away the vegetation and stockpiling the topsoil. The dune is then excavated to a depth of tens of meters, making a pond. A slurry of sand and water is sucked from one edge of the pond, and the small percentage of rutile grains is separated by centrifugal force. The remainder of the sand is then pumped back onto the other side of the pond. In this way the pond moves progressively along the dune, creating a 'mining path' from which the vegetation has been removed, and the substrate reconstructed to considerable depth. Under current regeneration practices, the dune is first restored to approximately its original shape. The topsoil, including soil seed reserves, is replaced. The dune is stabilized with a crop of hybrid sorghum during the first year after mining, and natural vegetation develops over ensuing years. In consequence, a sequence of vegetation ages is found along each mining path.

Barry and Marilyn Fox recognized the potential of this situation for studying succession while still research students at Macquarie University, and established permanent plots during 1975 to 1976 on sites then aged 5 to 11 years after mining. They have studied successional processes in vegetation, small mammals, and ant assemblages. This research gains special strength from the complementary use of four types of information: comparisons among sites of different ages, longitudinal studies at each site, comparisons between regeneration after mining and regeneration after fire (permanent sites for fire studies were also set up during 1974 to 1977 (Fox, 1988)), and experimental manipulations of interactions between species.

For example, consider small mammals. Mus musculus abundance peaks at sites mined 3 to 5 years previously, while Pseudomys novaehollandiae abundance overtakes it at sites mined 5 to 7 years previously (Fox and Fox, 1984). The same pattern was present both when the mine path was sampled in 1982, and when it was sampled in 1987 (Twigg et al., 1989). The actual sites representing a given time since mining were, of course, different in the two sampling years. On sites regenerating after fire, the first 5 to 6 years of the succession after mining are compressed into a single year, and P. novaehollandiae becomes more abundant during the second year (Fox and McKay, 1981; Fox and Fox, 1984). Probably this is because many plant species regenerate vegetatively after fire, whereas their rhizomes and lignotubers are destroyed during mining, so that understory vegetation develops much more quickly after fire. P. novaehollandiae abundance is correlated with vegetation cover below 50 cm (Fox and Fox, 1978). Indirect evidence indicated M. musculus abundance declines because they behaviorally avoid P. novaehollandiae. This hypothesis has been confirmed by experimental removals of P. novaehollandiae, and also by experimental additions (Fox and Pople, 1984), which are less often carried out.

A comparable situation is found among ants. lridomyrmex species C dominates for the first 8 to 9 years after mining, and is then replaced by lridomyrmex species A (Fox and Fox, 1982). Within the two-species mosaic of colonies found around the time one species replaces the other, experimental deletion of colonies of either species produced increases in the other species (Fox et al., 1985; Haering and Fox, 1987), with the relationship mediated by sharp behavioral competition at territory boundaries. Because lridomyrmex species are so active and behaviorally dominant (Greenslade, 1976), different assemblages of other ants are associated with dominance by the two different lridomyrmex species (Fox and Fox, 1982).

11.4 OBJECTIVES OF LONG-TERM ECOLOGICAL RESEARCH

This section discusses what sorts of objectives might be achievable by long-term research but not by short-term research. The discussion does not specify variables to be measured. In any research, the variables worth measuring depend on which questions are thought interesting. The intention here is to consider the virtues of long-term ecological research generally, across the wide variety of possible questions and variables.

For the purpose of discussion, objectives of long-term measurement, or networking across space, or both, can be split into five categories:

1. Enough replicate years to allow statistical analysis (e.g. correlation) of between-year variation;
2. Perspective on rare events, estimating their return times;
3. Studying second-phase effects of experimental manipulations. By second- phase effects are meant those which involve species arriving subsequently, or effects via consequences for slow variables; in contrast to first-phase effects, which involve those on species present at the outset of the experiments and on interactions among them;
4. Estimating the proportions of variation in an ecological measurement which are attributable to differences between years or between places, or to place-year interactions, such as patterns which occur every year, but are shifted in space.
5. Measuring changes over time in ecological variables, on the premise that we need to begin collecting a coherent record now, in order to have data against which to assess any understanding or models of global change we may develop in 10, 30, or 50 years' time.

11.4.1 PREVIEW

Briefly, the opinion we will put forward is that the highest priority for Australia is to contribute to global networks of many sites, making simple measurements, with a view to objectives 4 and (especially) 5. Extending intensive research to 10- to 50-year runs at a few individual sites, with a view to objectives 1-3, would not be such a high priority. Within a limited budget, there will be a conflict between having many sites and studying fewer sites more intensively. It will be important to restrain the natural instinct to measure more variables more often, lest spatial coverage be endangered.

This opinion is based both on a modest assessment of what can be achieved with respect to objectives 1 to 3 by runs of data 10 to 50 years long at any one site, and on a very high assessment of the importance of objective 5.

11.4.2 MODEST VALUE OF DATA RUNS OF 10 TO 50 YEARS  FOR OBJECTIVES 1 TO 3

11.4.2.1 Objective 1:Years as replicates for analyzing correlation between two variables

Suppose two variables are in fact related by an ecological process. For example, the earliness of the wet season (A) might affect the peak population density achieved by a herbivorous insect (B). Now suppose A is capable of accounting for 20% of the variance in B (r²=0.20); this is a decidedly strong relationship by ecological standards. Fifty replicate years would be needed to have a 90% chance of detecting such a relationship at P <0.05 (Figure 11.2). Thus runs of data in the 10- to 50-year range can only be relied on to detect very strong relationships. In order to reliably detect relationships with r² in the range 0.01 to 0.10 (where many interesting and important ecological relationships lie), data runs of 100 to 1000 years are needed. This suggests that monitoring ecological variables in one place for 10- to 50-year periods, while by no means valueless, is a relatively weak approach to detecting such correlations.

Two suggestions can be made for stronger approaches to detecting or assessing relationships between ecological variables at one place. First, variation in driving variables should be created experimentally where possible. Second, we should be turning stronger attention to the possibilities for obtaining long data runs from records of the past, whether historical or paleoecological. New technologies are opening up many possibilities for acquiring high-resolution environmental records from the past. There are many situations where only funding is restricting the availability of valuable data (Chappell 1988; Wasson, 1988). However, this should not be seen as a recommendation to simply hand funds to paleobiologists. Funding should be made conditional on paleobiologists collaborating closely with theoretical and experimental ecologists in defining questions and data collection designs which will allow the questions to be answered.

Figure 11.2 Number of replicate years required to detect with 90% probability (power = 0.90), at 5% significance, a correlation (between variables, across years) accounting for a given percentage of variation. Derived from a table in Cohen ( 1969)

11.4.2.2 Objective 2:Perspective on the incidence of unusual but biologically important events 

The domination of ecology by studies of 1 to 3-year duration has led to many misinterpretations of the incidence of unusual events (Weatherhead, 1986). However, even 20-year studies are only capable of assessing the return times of events which occur about 10 times per century or more often. Further, these return times are changing, so that past return times cannot be simply extrapolated to the future.

Further still, it should be considered that if unusual events are biologically important, merely monitoring them will not advance our knowledge fast enough. We need to find ways to organize science funding such that hypotheses are tested experimentally during unusual events. Some rare events can be anticipated, e.g. the 18.6-year tidal cycle which Clark (1986) hypothesized reset Gramineae and Cyperaceae populations in a Long Island high marsh to a new exponential growth phase. Other experiments could in principle be designed in advance for implementation at an unknown future time (Westoby et al., 1989). For example, a widespread rangelands problem is establishment of shrubs in semi-arid grasslands after they have germinated during two successive years of heavy summer rains. It is possible that destocking during a critical period would allow grass growth to suppress shrub seedling establishment. Experiments to test this hypothesis need to be funded in such a way that they can be implemented at whatever future time mass shrub germination comes about.

While a long study is always better than a short one, this should not lead us to concentrate our efforts into a relatively few research sites studied at length. On the contrary, my own experience has been that problems in generalizing from one or a few studies arise at least as often because they are unrepresentative in space as because they are unrepresentative in time.

At present, short-term studies at a single site often report that some event they think unusual had a significant effect on the outcome of their study (Weatherhead, 1986) (for example, mass kill of shrubs by drought or of echinoids by disease). It would be very valuable to have more perspective on how widespread such events are in space. In a reasonable proportion of cases some evidence (e.g. dead shrubs) is to be seen for a period after the event, so that a single visit to a site can assess whether the event happened there. It would be cost effective for agencies to make modest funds available so that when an unusual event is encountered during a 3- to 6-year study, funding can quickly be sought for travel, research assistance, satellite imagery , or whatever was necessary to allow a rapid wide-area assessment.

11.4.2.3 Objective 3: Second-phase effects of experimental manipulations 

It is undoubtedly very important that second-phase effects should be recognized and studied in experimental ecology. This objective justifies some 10- to 50-year studies. However, it would be impossibly expensive to extend all field experiments for 10 to 50 years on the grounds that second-phase effects needed to be studied. Other approaches to second-phase effects should also be considered. If the nature of the effects can be anticipated, they can be factored into the initial design of experiments. For example, if effects involving species not yet present are anticipated, treatments in which they are introduced could be included in the design. For effects involving slow litter accumulation, litter could be added; and so forth.

Another possibility is to make better use of the many sites where intensive experimental studies have been undertaken over 3 to 6 years. Suppose investigators were to submit proposals to revisit their previously funded experimental sites from time to time. In effect, they would be saying they had no particular hypothesis, but would like to keep a watching brief on their sites to see if anything interesting happened. Under the criteria used by most competitive research funding agencies at present, such proposals would be ranked very low. However, there is a defensible case that by making modest funding (e.g. a week's fieldwork every 2 years) available for such proposals, funding agencies could procure insight into second-phase effects more cheaply than by any other means.

11.4.3 HIGH VALUE OF DATA RUNS OF 10 TO 50 YEARS FOR OBJECTIVES 4 AND 5

11.4.3.1 Objective 4:Assessing variation across space in relation to variation through time 

Most of our present understanding of ecosystems is based on information which is restricted in space (often from plots of a hectare or less) as well as in time (studies extending over 3 years or less). This is especially true of that part of our understanding which is underpinned by field experimentation. Arguably the most important problem for ecological science over the next few decades is scaling up - sorting out situations where this narrow-scale understanding can be simply extrapolated to wider areas and longer times from situations where new concepts and approaches are needed.

Two of the Australian studies summarized above illustrate the strong relationship between wide spatial coverage and long-term research. Work centered on Cape Banks has shown how formal ANOVA design can be used to assess the relative strengths of variation at different scales in space and time. Such information allows more intensive narrow-scale studies to be put in context, and also permits estimates of the space and time scales over which aggregated variables become relatively predictable because narrow-scale variation is averaged out. The study centered on Maria Island detected a variety of interesting correlations between variables measured at sites hundreds or even thousands of kilometers apart. Given <50 year- replicates, only correlations with r² at least 0.20 can reliably be detected (Figure 11.2). Correlations this strong arising from processes at one place will often prove to be well understood already. For example, a relationship between surface water temperature and phytoplankton production at the Maria Island site, mediated by nutrients in surface water, would have been no surprise to marine ecologists. But correlations between spiny lobster catches off Tasmania and New Zealand and (in the opposite direction) off New South Wales would have been much more surprising. Such wide-area correlations may contribute strongly to understanding relationships between ground-truth ecological measurements and regional or global stale climatology.

In summary , much is to be gained in a research sense from having networks of many sites, even if only a few simple measurements are taken at each. The arrangement of sampling across space and through time should be designed with a view to partitioning variance into components attributable to different scales in space and time, and to interactions between space and time.

11.4.3.2 Objective 5:Data base on global change against which to assess future understanding 

Global change - all its aspects, not just atmospheric change  -  is plainly the most important applied problem for this and the next generation of ecologists. It is not possible to predict what sort of understanding of global change processes will have been generated by research 30 to 50 years from now. However, it is possible to predict with certainty that confidence in that future understanding will be crippled unless we have in place, at that time, a coherent data base describing the actual course of change from the 1990s forward. Initiating such a data base should be regarded as easily the most important objective of long-term ecological studies, and the work should be designed with this objective dominant.

There is one desideratum of long-term ecological study which arises from this objective. One thing an ecological data base will need to do is to connect with general circulation models (GCMs) of global climate. Current GCMs work with cells about 500 x 500 km. These will become smaller with increased computing power, but it is quite likely that cells will still be on the order of 100 x 100 km in 30 years' time. That is, models of climate-vegetation relationships to be incorporated in GCMs need to deal with variables which represent averages over at least 100 x 100 km. A minimal requirement for a data base oriented to global change research is that it should provide descriptions of vegetation and land surface characteristics averaged over no less than 100 x 100 km. At present most ecological research measures variables at much smaller scales than this. It is, of course, possible that future research will develop rules which will allow us to scale up from small-scale measurements to wide-area estimates. A great deal of research is going into self -similarity of spatial patterns at different scales, for example; and into nesting regional-scale climate models within GCMs. However, it would be wrong to rely on methods being developed in the future to scale up from local-scale measurements to averages over wider areas. The data base should be designed from the outset with a view to providing reliable averages over wide areas. This could be done either through remote sensing, or through proper procedures of stratification and randomization in choosing sites for ground measurement, depending on the variable. We should not fall into the errors of using a small number of sites, choosing them arbitrarily on grounds of convenience, and relying on future developments to allow reliable scaling up.

11.5 SCENARIOS FOR AN AUSTRALIAN EFFORT IN LONG-TERM ECOLOGICAL RESEARCH IN THE CONTEXT OF GLOBAL CHANGE

There are many complexities in assessing what form Australian efforts should best take. I will summarize the most important issues by putting forward and commenting on three scenarios. Of these, I would regard the first as a disastrous outcome (though there is a real risk it might happen), the second as having some good and some bad features, and the third scenario as the path we ought to take if the government can be persuaded.

Under scenario 1, no extra funding would flow to Australian ecology for purposes of global change research. Instead, some of the funds currently spent on research in all areas of ecology would be redirected towards long-term studies. This would be most likely to come about through ARC shifting its funding criteria to put greater emphasis on long-term studies in the context of global change, and relatively less emphasis on peer-review rating of the excellence of the proposal and the investigators.

Under this scenario Australia's existing research effort in ecology would be seriously damaged. A significant development of long-term studies would require much or all of the funds currently spent by ARC on ecology in general. The advance of our fundamental understanding of ecosystem organization would slow or stop. The long-term studies would not substitute for the previous ecology research effort in this respect, because they would not be asking questions of the some incisiveness.

Further, it needs to be considered that a policy of funding long-term projects automatically means that projects once begun must be favoured for continued funding. The most important factor in research progress is the quality of the principal investigators and the intensity with which they commit their time and energy to the research. The best method yet found for assuring this quality and commitment is free and open competition among all applicants, on the basis of peer review, at frequent intervals. The hard truth must be faced that if some projects are given continuity for periods upwards of 10 years, this will be done at least partly at the expense of other projects and other investigators which would be rated better by peer review. There is a cost of continuity, which takes the form of foregoing the opportunity to switch funds to better investigators, and which increases over time. This cost means that long-term research cannot be expected to substitute satisfactorily for open-competition research. We need long-term studies to complement research on a shorter funding cycle, not to replace it.

Under scenario 2, the Australian government would commit substantial new funds to global change research, and a reasonable proportion of it, say A$5 to 10 million per year, would be allocated to ecology. Because of the pressure of precedent from the style of long-term research adopted in the United States and Federal Republic of Germany, combined with science politics within Australia, these funds would be directed into about five big-team projects. Depending on the strength of influence from overseas, these projects might have an 'ecosystem' orientation, or might be oriented towards approaches more highly regarded in Australia, such as vegetation dynamics or experimental ecology. Under this scenario, understanding of Australian ecosystems would be advanced by a burst of fieldwork, representing at least a doubling of current ARC expenditure. This would certainly be an excellent thing. However, the data base which resulted would come from relatively few sites, perhaps five to 20. Its geographical coverage would be far too thin to give a satisfactory picture of the spatially patchy processes expected in response to global change (for example, processes of certain elements of some vegetation types colonizing into new locations, or of increased incidence of outbreaks of some herbivorous insects within some vegetation types). For these reasons, scenario 2 would not be making a very good contribution to the data base for global change, which is the most important objective for long-term ecological studies in Australia.

Under scenario 3, as under scenario 2, the Australian government would find significant new funds for global change research, and A$5 to 10 million per year of these would be spent on ecological aspects. Under scenario 3, however, these extra funds would be spent on a large number (more than 200) of sites. Relatively simple measurements would be taken at most of these sites, not necessarily every year.

This would be the best of the three scenarios in that it would be directed at the most important objective for global change research, which is to set in place a data base recording the time-course of global change with complete geographical coverage and relatively high spatial resolution. In saying that it would be the best of the three scenarios, I do not at all mean that it would cover all the ecological research that needs to be done in relation to global change; much more than A$5 to 10 million would be needed for that. I mean simply that it would be better than scenarios1 and 2.

There would certainly be problems with scenario 3. Notably, means would have to be found to assure effective quality control of the data and management of the data base. Normally, researchers can be expected to pay close attention to their data because they examine and analyze them themselves and publish from them within a few years. But in the case of simple measurements collected from less than 100 locations, many people would be involved in taking the measurements, and many of the data would have no immediate use in testing hypotheses. Comparability among different sites and over time would need to be assured, and the data base would need to be organized to allow ready access by any interested parties.

Protocols for data management can, no doubt, be organized. However, the key to quality research is the sort of people who are involved. When choosing sites for a monitoring network, we face the enervating prospect of a series of workshops during which everyone advocates the sites they work at, naturally for entirely objective reasons of the site's interest, representativeness, and backlog of data. There is a risk that we could emerge from such a process without the country's best researchers involved with the selected sites. We should start out instead by selecting first-rate researchers, using the usual criteria of continuing publication in international peer-reviewed journals, etc. These researchers could then be put in a position to run networks of sites, whether in person or by subcontracting.

The use of new funds to set in place such a data base should parallel a steady enhancement of fundamental research in ecology, through the medium of short-term grants given competitively through ARC. Some minor modifications could usefully be made to current funding practices, with a view to gaining better perspective on the representativeness of results in time and space. Specifically, paleobiology could be better funded provided proposals are devised in collaboration with experimental and theoretical ecologists; means could be found to fund experiments to be implemented at unknown future times, when the relevant configuration of circumstances arises; when unusual events occur during a 3- to 6-year study at a single site, some funds could be available for rapid assessments of how widespread the events were; and after intensive studies have been completed involving experimental manipulations at a site, modest funds could be provided for proposals to revisit the site occasionally to see whether any interesting longer-term effects have become apparent.

11.6 ACKNOWLEDGMENTS

My thanks for helpful discussion, comments and contributions of information from Paul Adam, Andy Beattie, David Bowman, Barry Fox, Frank Golley, Graeme Harris, Ian Hume, Bob May, Ron Pulliam, Graham Pyke, Barbara Rice, Paul Risser, Russell Sinclair, Tony Underwood, Dedee Woodside, Paul Zedler, an anonymous referee and from many people at the Berchtesgaden workshop. The conclusions are mine.

11.7 REFERENCES

Bowman, D.M.J.S., Wilson, B.A. and Hooper, R.J. (1988). Responses of Eucalyptus forest and woodland to four fire regimes at Munmarlary, Northern Territory, Australia. Journal of Ecology, 76, 215-232.

Caffey, H.M. (1985). Spatial and temporal variation in settlement and recruitment of intertidal barnacles. Ecological Monographs, 55, 313-332.

Chappell, J. (1988). The quaternary environmental record. In Australian National Committee for the IGBP, Convenors. Global Change. Australian Academy of Science, Canberra, 30- 35.

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

Cohen, J. (1969). Statistical Power Analysis for the Behavioral Sciences. Academic Press, New York.

Crisp, M.D. ( 1978). Demography and survival under grazing of three Australian semi-desert shrubs. Oikos, 30, 520-528.

Crisp, M.D. and Lange, R.T. (1976). Age structure, distribution, and survival under grazing of the arid zone shrub, Acacia burkittii. Oikos, 27, 86-92.

Fox, B.J. (1982). Fire and mammalian secondary succession in an Australian coastal heath. Ecology, 63, 1415-1425.

Fox, M.D. (1988). Understory changes following fire at Myall Lakes, New South Wales. Cunninghamia, 2, 85-95.

Fox, B.J. and Fox, M.D. (1978). Recolonization of coastal heath by Pseudomys novaehollandiae (Muridae) following sand-mining. Australian Journal of Ecology, 3, 447-465.

Fox, M.D. and Fox, B.J. (1982). Evidence for interspecific competition influencing ant species diversity in a regenerating heathland. In Buckley, R.C. (Ed.) Ant-Plant Interactions in Australia. Dr W. Junk, The Hague, 99-110.

Fox, B.J. and Fox, M.D. (1984). Small-mammal recolonization of open-forest following sandmining. Australian Journal of Ecology, 9, 241-252.

Fox, B.J., Fox, M.D. and Archer, E. (1985). Experimental confirmation of competition between two dominant species of Iridomyrmex (Hymenoptera: Formicidae). Australian Journal of Ecology, 10, 105-110.

Fox, B.J. and McKay, G.M. (1981). Small-mammal responses to pyric successional changes in eucalypt forest. Australian Journal of Ecology, 6, 29-41.

Fox, B.J. and Pople, A.R. (1984). Experimental confirmation of interspecific competition between native and introduced mice. Australian Journal of Ecology, 9, 323-334.

Geldenhuys, C.J. (1977). The effect of different regimes of annual burning on two woodland communities in Kavango. South African Forestry Journal, 103, 32-42.

Greenslade, P.J.M. (1976). The meat ant Iridomyrmex purpureus (Hymenoptera: Formicidae) as a dominant member of ant communities. Journal of the Australian Entomological Society, 15, 237-240.

Haering, R. and Fox, B.J. (1987). Short-term coexistence and long-term competitive displacement of two dominant species of Iridomyrmex : the successional response of ants to regenerating habitats. Journal of Animal Ecology, 56, 495-507.

Hall, E.A.A., Specht, R.L. and Eardley, C.M. (1964). Regeneration of the vegetation on Koonamore Vegetation Reserve, 1926-1962. Australian Journal of Botany, 12, 205-264.

Harris, G.P., Davies, P., Nunez, M. and Meyers, G. (1988). Interannual variability in climate and fisheries in Tasmania. Nature, 333, 754-757.

Harris, G., Nilsson, C., Clementson, L. and Thomas, D. (1987). The water masses of the east coast of Tasmania: seasonal and interannual variability and the influence of phytoplankton biomass and productivity. Aust. J. Mar. Freshw. Res., 38, 569-590.

Holm, A.McR. ( 1986). The assessment of range trend in Western Australian pastoral lands. In Joss, P.J., Lynch, P. W. and Williams, O.B. (Eds) Rangelands: A Resource Under Siege. Proceedings 2nd International Rangeland Congress. Australian Academy of Science, Canberra.

Holm, A.McR., Burnside, D.G. and Mitchell, A.A. (1987). The development of a system for monitoring trend in range condition in the arid shrublands of Western Australia. Australian Rangeland Journal, 9, 14-20.

Kellman, M. and Miyanishi, K. (1982). Forest seedling establishment in neotropical savanna: observations and experiments in the Mountain Pine Ridge savanna, Belize. Journal of Biogeography, 9, 193-206.

Kershaw, A.P. ( 1986). Climatic change and Aboriginal burning in north-east Australia during the last two glacial/interglacial cycles. Nature, 322, 47-49.

Noble, I.R. (1977). Long-term biomass dynamics in an arid chenopod shrub community at Koonamore, South Australia. Australian Journal of Botany, 25, 639-653.

Noble, I.R. and Crisp, M.D. ( 1980). Germination and growth models of short-lived grass and forb populations based on long-term photo-point data at Koonamore, South Australia. Israel Journal of Botany, 28, 195-210.

Osborn, T.G.B., Wood, J.G. and Paltridge, T.B. (1935). On the climate and vegetation of the Koonamore Vegetation Reserve to 1931. Proceedings of the Linnean Society of New South Wales, 60, 392-427.

Rose-Innes, R. (1972). Fire in West African savanna. Proceedings of the Tall Timbers Fire Ecology Conference, 11, 147-173.

Sarmiento, G. and Monasterio, M. (1975). A critical consideration of the environmental conditions associated with the occurrence of savanna ecosystems in tropical America. In Golley, F.B. and Medina, E. (Eds) Tropical Ecosystems: Trends in Terrestrial and Aquatic Research. Springer-Verlag, New York, 223-250.

Shaughnessy, P.D., Leigh, J.H. and Walker, B.H. (1988). Preliminary proposals for a set of long-term ecological study and monitoring sites in Australia. Working Paper, available from CSIRO Division of Wildlife and Ecology, PO Box 84, Lyneham, ACT 2602 Australia.

Sinclair, R. (1986). The T.G.B. Osborn Vegetation Reserve at Koonamore: long-term observations of changes in arid vegetation. In Joss, P.J., Lynch, P.W. and Williams, O.B. (Eds) Rangelands: A Resource Under Siege. Proceedings 2nd International Rangeland Congress. Australian Academy of Science, Canberra, 67-68.

Stocker, G.C. and Mott, J.J. (1981). Fire in the tropical forests and woodlands of Northern Australia. In Gill, A.M., Groves, R.M. and Noble, I.R. (Eds) Fire and the Australian Biota. Australian Academy of Science, Canberra, 425-439.

Twigg L.E., Fox, B.J. and Luo Jia. The modified primary succession following sandmining: A validation of the use of chronosequence analysis. Australian Journal of Ecology, 14, 441-448.

Underwood, A.J. Rocky intertidal shores. In Hammond, L.S. and Synnot, R. (Eds) Australian Marine Biology. Longman-Cheshire, Melbourne. In press.

Underwood, A.J. and Jernakoff, P. (1981). Interactions between algae and grazing gastropods in the structure of a low-shore algal community. Oecologia, 48, 221-233.

Wasson, R.J. (1988). Landscape denudation in Australia in the late Holocene. In Australian National Committee for the IGBP, Convenors. Global Change. Australian Academy of Science, Canberra, 56-59.

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

Westoby, M., Walker, B.H. and Noy-Meir, I. (1989). Opportunistic management for rangelands not at equilibrium. Journal of Range Management, 42, 266-274.

Wood, J .G. ( 1936). Regeneration of the vegetation of Koonamore Vegetation Reserve, 1926-36. Transactions of the Royal Society of South Australia, 60, 96-111.