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

3.1  INTRODUCTION

In the UK consideration is being given to establishing a network of long-term ecological references sites. Why? Questions are asked with increasing frequency about changes in the environment (levels of pollutants, climate, species introductions, habitat gains and losses, shifts in land use, etc.). Answers to the questions are usually cautious because we lack information on the past, we cannot distinguish trends induced by man from natural variations or attribute cause to any observed change, or because the available information is not appropriate to the particular conditions in question. 

Instead of setting up new studies to answer each question as it arises, the approach being considered in the UK is to make use of well-established sites with existing data to establish a network of long-term ecological reference sites, with ongoing studies to help answer questions in the future. How suitable are the present sites? The sites have been maintained by different organizations which use them for their own purposes, and many of them are not known to others. However, when combined, will these sites constitute a network which adequately represents the range of environmental and management conditions in the UK?

Is the information available from these sites useful in answering questions about environmental change? These are the questions being asked and discussed by a working party established by the Natural Environment Research Council.

Although based on the working party's considerations, this chapter represents a personal view of the general issues. Selected examples are presented, in this chapter, and some other examples are given in Chapter 6 of this volume.

The main rationale for proposing a network of sites is as follows:

1. Study sites are an essential component of ecological research. To answer questions on changes in the environment, we need sites which represent the main environmental, ecological, and management variations in the UK. While studies of some individual topics will require other sites with particular characteristics, a 'core' network will provide the opportunity to use existing information on the related topics and facilities.
2. Long-term studies are required to monitor changes external to the system which take place gradually or at infrequent intervals. Responses to those changes may occur through species or processes which have a slow turnover time or through a series of linked short-term events, the results of which are only apparent in a long-term study (Likens, 1989).
3. In addition to delayed and serial responses, it is also necessary to distinguish between the different factors which cause, or interact to cause, change. For these reasons it is important to have sites with integrated or multi-media monitoring, and to carry out both observational and experimental research.
4. The scientific case for a network of long-term study sites in the UK is strong. Information on environmental changes and on their consequences is a serious need in government. By concentrating on established sites, the cost of creating such a network can be kept to a minimum.

This chapter considers three aspects of the role of study sites in long-term research and monitoring: (1). the ecological questions to be addressed, (2). the separation of natural and man-induced phenomena (noise from signal), and (3). strategies in  the selection of sites.

3.2 QUESTIONS TO BE ADDRESSED

Questions asked by those concerned with environmental policy or management test the scientists' ability to detect changes in the state of the physical, chemical, and biological environment, and to distinguish cause from effect. The basic questions ask what is happening, where it is happening, what the consequences are, whether there are thresholds, and whether the thresholds are reversible. Apart from responding that more research is needed or that it all depends on the conditions, most ecologists will turn to their favorite site and, using their favorite methods, will measure as intensively as labor allows. While many of the questions can be explored in the laboratory, we must use field observation and experiment to evaluate extrapolation from particular laboratory conditions to the complexity of the field, and also to study phenomena which cannot be examined in the laboratory, test predictions of responses to control measures, and provide early warning of new changes. It follows that the sites and conditions in which measurements are made must be selected in relation to the questions asked. Here lies a major problem. We know what today's questions are, but what are the questions of the future towards which long-term research and monitoring should be directed? Although we cannot specifically predict future environmental concerns, we do recognize the four generic subjects of climate change, chemical pollutants, management effects, and invasions or extinctions. Each of these makes particular demands upon a site, and in a study will require particular types of site, measurements, and experiments.

3.2.1 CLIMATIC CHANGE

Monitoring the physical climate is reasonably straightforward, and is systematically undertaken by the Meteorological Office. Prediction of climatic change is much more difficult, especially at the regional level. Ecological responses to general trends in climate and to irregular extreme events are complex and difficult to assess. Despite this, and despite the importance of climatic factors to ecology and crop production, there is a surprising lack of systematic monitoring and analysis of species and system response to climate change. The Rothamsted Insect Survey is one comprehensive study in which population changes can be related to climatic variation, and there are some relevant species studies on individual sites (Taylor, 1989). However, it would be sensible to establish, now, situations in which to detect degrees and types of biological and ecological response. The questions asked in these situations are likely to focus on changes in the distribution of species and in the composition of communities of nature conservation or of amenity interest, in the production of agricultural and forest crops, in the occurrence of crop pests, and in land-use patterns. While the current concern is the consequences of long-term projected trends in climate resulting from the greenhouse effect, the biota will respond to the frequency and intensity of irregular extreme events (for example, dry summers or mild winters) as much as to general trends in mean values of temperature or precipitation.

In all these questions, effects are likely to be greatest, and most detectable, at the climatically determined margins of distribution of a species or production system. Thus the questions can be targeted quite specifically by selection of species and sites according to their biology and distribution. For example, repeated annual sampling showed that population regulation of the homopteran Neophilaenus lineatus was density dependent at Wytham Woods, a low-altitude site, but density independent and climatically determined at Moor House, a more northerly high-altitude site (Figure 3.1) (Whittaker, 1971). Population response of N. lineatus to climate change would, therefore, be most readily detected at Moor House, and it is a suitable indicator species because its density can be monitored by counting numbers of readily observed spittle. Similarly, the annual distribution and success of insects along altitudinal rather than geographical gradients of climate can be directly related to climatic variation (Coulson and Whittaker, 1978; Whittaker, 1985). The same principles apply for plants in the selection of species, sites, and parameters to detect response to climate change. For example, in the uplands of Britain, Calluna vulgaris and Rubus chamaemorus are at the cold and warm ends, respectively, of their climatically determined distribution. The upper and lower distribution of these species can be readily identified and marked to define the extent to which hypothesized or observed future changes in climate affect vegetation. Such basic monitoring can be enhanced by associated research, as in the case of Calluna and Rubus, in which physiological studies have defined their photosynthetic responses to climatic variables (Figure 3.2). From these response surfaces, Grace and Marks (1978) showed a differential response of the two species to warm years, and how growth of Calluna could affect Rubus through light interception. This type of information allows prediction of both short- and long-term responses to particular climatic variables, with potential validation by field measurement.

Figure 3.1 The relation between the logarithm of numbers of instar 2 in each generation and the subsequent 'generation' mortality (K) in Neophilaenus lineatus (from Whittaker, 1971), (a) at the Juncus site, Moore House, (b) at Upper Seeds, Wytham Woods. ¢: including parasitoid;  p: excluding parasitoid (Coulson and Whittaker, 1978)

The examples from insects and plants indicate how studies can be focused on questions of which species will be affected, and where and why. When expanded to broader questions of land use, the analysis of Parry and Carter (1985) is particularly relevant. They showed that historically, the frequency of crop failure was related to the occurrence of climatic extremes. During periods of climatic change the distribution of agricultural use expanded or contracted depending on the frequency of failure. As a result, change in land use in upland areas was concentrated in a distinct geographical zone at the margin of particular crop production systems. This is illustrated in Figure 3.3 for a particular valley in which improvement of moorland for agriculture, and abandonment of farmland to moor, has been restricted for over 200 years to a narrow fringe between stable areas of moorland and farmland.

Figure 3.2 (a) Photosynthesis of Rubus chamaemorus, aged 11 weeks, as a function of temperature and irradiance (380-720 nm) (Grace and Marks, 1978); (b) photosynthesis of Calluna vulgaris shoots, 35 days after budbreak, with no flowers, and with a temperature treatment of 10°C (Grace and Marks, 1978)

The growth of species or crops within the main part of their range will be affected by climate, but other factors will also limit their success. By definition, climate will have its greatest impact where it is the main controlling variable, i.e. at the climatically determined margins of distribution. The implications for long-term monitoring and research of the examples quoted are (1) sites should be large enough to include climatic gradients or be selected to cover a gradient, (2) species sensitive to climatic change can be identified through their distribution patterns and population dynamics, and (3) relatively simple measures, such as phenological observations, distribution limits, the current year's shoot increment, and population success, can, given a knowledge of the species based on detailed research, provide measures of short- and long-term response.

Figure 3.3 The distribution of land which has been reclaimed from or reverted to moorland in Bransdale in the North York Moors

3.2.2 CHEMICAL POLLUTANTS

Experience has shown that the range of pollutants and their effects are extremely variable and complex, yet there is frequent pressure to define cause and effect at short notice. However, there are general lessons to be learned, particularly in relation to the selection of sites for long-term monitoring and/or research. In considering pollutant problems such as acid deposition (N, S, O₃), heavy metals (Pb, Zn, Cd), fertilizers (N, P), pesticides (DDT, 2,4-D), or radionuclides (¹³⁴Cs, ¹³⁷Cs, ²³⁹Pu, ²⁴⁰Pu), the common questions concern the rates of deposition, the transformation and retention of the element or compound within an ecosystem, its transfer through biological pathways, and its toxicology. In other words, questions of element dynamics which require detailed understanding of ecosystem processes. Unless there are overriding site-specific issues, it is sensible, therefore, to associate studies on new pollutants with sites for which the dynamics of other elements are known.

In assessing the consequences of pollutants, there are three main recurrent questions concerned with spatial variation, infrequent events, and interaction effects:

1. Which areas can be identified as 'hotspots' associated with particular deposition or retention characteristics? For example, concentrations of  ¹³⁷Cs following the Chernobyl accident showed high within and between site variation in the UK, resulting not only from the rainfall pattern but also from small-scale variation in surface water movement and varying ion exchange characteristics or organic and mineral soils (Figure 3.4). The decline of ¹³⁷Cs in vegetation predicted from studies of predominantly mineral soils did not correspond well with observed time trends on peaty soils, emphasizing again the effect of varying site conditions (Figure 3.5). Experience in acid deposition research has also shown the importance of climatic and topographic conditions, vegetation cover, and soil properties in determining variation within and between sites in cause and effect interactions. In particular, the combination of factors has been important in answering the question of which areas constitute hotspots which are most sensitive to acid deposition.
2. Which climatic conditions are particularly important in causing pollution problems? The climatically induced episodes of deposition and of release of acidity in snow melt, and of periods of high O₃ concentrations and flushes of NO₃ associated with drought conditions (Roberts et al., 1989), all emphasize the significance of short-term fluctuations within the context of long-term measurements.
3. To what extent can observed effects be attributed to specific pollutants? Even the immediate direct effects of a pollutant on tree growth are confused by the interaction or combined effects of a pollution cocktail or by interaction with other environmental factors, as in the case of increased sensitivity of red Appalachian spruce to frost associated with atmospheric deposition (Fowler et al., 1989). More indirect effects are well known, such as the release of toxic Al from weakly buffered acid soils as a result of shifts in ionic balance following acid deposition (Ulrich, 1987), or the concentration and transfer of some pesticides, heavy metals, and radionuclides along food chains.

Figure  3.4  ¹³⁷Cs in UK grassland vegetation following the Chernobyl accident (Bq/m²).    

Figure 3.5  Predicted (---) and observed(^__) concentration of  ¹³⁷Cs following the Chernobyl accident in UK upland vegetation on peaty soils.

The implications of these questions for further site monitoring and the assessment of pollutant cause and effect are that sampling must enable:

1. Detection of small within-site hotspots, particularly of aerial pollutants;
2. Sites to cover a range of environmental variation to allow identification of areas sensitive to mobilization or concentration of pollutants;
3. Measurement of a range of variables other than the pollutant necessary to detect interactions and indirect effects; and
4. Measurements on time scales frequent enough to detect climatic episodes, which are at least as important as long-term mean characteristics.

3.2.3 MANAGEMENT AND LAND USE

Some questions concerned with such specific management practices as fertilizer or pesticide application are considered under the heading of pollution, but there are also repeated questions concerning the amount and distribution of changes in management systems and land use, and of the consequential alteration in flora and fauna. The more obvious changes in land cover (for example, hedgerow removal or loss of moorland through agricultural improvement (Figure 3.3)), are readily determined given adequate sampling design and consistency of recording. In contrast, detection of the ecological consequences of such changes, and of more extensive management changes such as modified grazing type and intensity, are more challenging (Usher and Thompson, 1988). These consequences are particularly relevant to long-term research because the response times are often measured in decades. Three topics, not altogether new, on which questions are focused are succession, soil processes, and land cover mosaics. These are discussed below.

The limited number of detailed long-term studies of the successional sequence of vegetation following changes in management is remarkable (Miles, 1979). Those detailed long-term studies that exist, classically Watt (1960) and the Park Grass Experiment (Taylor, 1989; Thurston et al., 1976), show marked short-term variations in species composition which could be misinterpreted as long-term trends if the studies were restricted in time (Figure 3.6). In these studies, as in others on vegetation and fauna (Marrs et al., 1988; Willis, 1988; Peterkin and Jones, 1989) which constitute long-term measurement of responses to changes in land management, site specificity limits the ability to make predictions of the consequences of land use change and to assess management options. This limitation can be overcome by integrating results from isolated studies into descriptive general hypotheses of succession (see Figure 3.7) (Miles, 1988), or by more mechanistic hypotheses based on growth strategies (Grime, 1987; Noble and Slatyer, 1980). These site-specific and theoretical approaches need to be complemented by extensive surveys undertaken at infrequent intervals, as in the National Ecological Survey of Britain carried out by the Institute of Terrestrial Ecology in 1978, which involved a stratified sample of 256 sites. The survey provides a baseline definition of species composition of open areas, and linear features. It was repeated in 1990 to determine the degree of vegetation change associated with measured changes in land use.

Figure 3.6 Changes in species composition of a grassland inside and outside a rabbit-proof enclosure in the English Breckland (from Watt, 1960, p. 220, courtesy of Journal of Ecology)

Figure 3.7 Changes in heather-dominated dwarf shrub heaths as a result of variations in intensity of grazing and burning. Present vegetation is shown in double circles: probable and possible stages in the  vegetation succession are shown by single and broken circles respectively. Rate of change where known is shown above the arrows: sheep numbers are per hectare.  Differing soil conditions are noted in the left-hand margin (after ITE, 1978)

Studies of the response of soil processes to land management are essentially long term, given the nature of pedogenetic processes. There are many pragmatic principles of soil protection, particularly those to minimize erosion and maximize fertilizer use efficiency, which are amenable to short-term studies. However, recurrent questions of sustained fertility, of soil improvement or degradation by trees, and of the effects of pesticides demand long-term field observations. The Rothamsted Experiments, described by Johnston in Chapter 6 of this volume, are unique in their time scale. They are also important in demonstrating the value of including management variations, even with inadequate statistical design (Taylor, 1989). One key feature in the determination of long-term trends in soil properties is the definition of within-site heterogeneity; geostatistics are a valuable aid in this respect. Retrospective studies (Davis, 1989) and space-for-time substitution (Pickett, 1989) provide valuable complementary approaches to long-term studies on sites which provide the only means of controlled manipulation.

Questions of the effects of loss or gain of habitats are being expressed with increasing frequency in terms of fragmentation, isolation, connectivity, or unit size and shape, i.e. in the terms of theories of island biogeography. This emphasis is related to recent legislation which provides financial support for management practices to sustain or enhance wildlife in environmentally sensitive areas, and to support farm woodlands and set-aside of agricultural land. The response of invertebrates and flora, as mentioned above, is long term, involving questions of dispersal, colonization, and establishment. Such research is appropriate in scale to most study sites, but observation of existing patterns is unlikely to provide sufficient variation to give satisfactory answers. The subject is very amenable to experimental manipulation yet, surprisingly, is virtually unexplored. On a much larger scale, studies of the effect of structural characteristics of woodlands or of habitat mosaics on large mammals and avian predators have very particular site requirements which are not compatible with most other topics.

3.2.4 SPECIES INTRODUCTIONS

In The Ecology of Invasions by Animals and Plants, Elton (1958) documented and analyzed many cases of the population expansion of an introduced species. Interest in the subject has been sustained in the UK by population expansion of introductions such as coypu, mink, sika and muntjak deer, and collared dove. More recently, the subject has been stimulated by questions concerning the environmental consequence of the accidental or intentional introduction of genetically manipulated organisms (Kornberg and Williamson, 1987).

Following the course of accidental introductions, whether genetically manipulated or not, may require long-term studies, but obviously these will have to be done at the sites of occurrence. The role of established sites will be to provide an opportunity to examine experimentally the response of the introduced species to a range of environmental conditions and communities. The potential role of a network of study sites for systematic testing of introductions has apparently not been considered.

The above discussion of the four general subjects to which environmental questions are addressed has identified a number of factors which must be considered when planning long-term research and defining the role of study or reference sites. The main considerations are that:

1. Monitoring is likely to be most effective when it is integrated with field experiments and with research which can help to assess causal and consequential relationships. Study sites must, therefore, be large enough to allow a range of activities.
2. Research and monitoring of different subjects may require very different approaches in terms of site selection. For example, some questions on effects of climate change or species introductions require analysis of population distributions which are not defined by sites. However, selection of study sites along major environmental gradients will provide a suitable focus for many such analyses.
3. Most long-term studies are site-specific. This restricts the ability to define the limits of the problem, and is, therefore, a danger in concentration of effort. The need to define both within- and between-site variation argues for the provision of a gradient of study sites.
4. Occasional extremes in climate, or other short-term episodes, are important determinants of long-term change. The need for long-term observations to distinguish short-term fluctuations from longer trends has been emphasized in the work of Likens (1985) and Tilman (1989). The case is well illustrated in Figure 3.6. In the rabbit-proof enclosure and in the Park Grass plot receiving  mineral fertilizer there were periods in which changes in species composition over a number of years moved contrary to the long-term trend. In both cases the successional trend in response to an abrupt change in management took decades to emerge.
5. Short-term 'natural' variation (for example, year-to-year changes in species composition) needs to be separated from long-term trends. This is discussed more fully in the next section.

3.3 SEPARATION OF 'NOISE' FROM 'SIGNAL'

To identify environmental change and its consequences we have to determine the existing or normal fluctuations (short-term noise) and trends (long-term noise) in order to distinguish new variations (signal) induced by extrinsic factors.

Monitoring by itself can be of value in some circumstances, particularly following the introduction of a new element whether it be a chemical (e.g. pesticide or ¹³⁷Cs), a species (e.g. coypu), a crop (e.g. Sitka spruce), or even a gene. However, in detecting changes in existing elements (e.g. nitrate concentration in water, atmospheric SO₂, or sward species composition) and in system responses (e.g. plant growth or soil organic matter accumulation or frequency of flooding), monitoring is of limited value. In these cases monitoring data do not allow distinction between the observed and the expected. It is not possible to determine whether the nitrate concentration is high because of increased fertilizer use or because of particular climate conditions; whether the decline in the bird population is part of a cyclical change or caused by increased predation; or whether the expansion of regeneration at the tree line is due to increased temperature, reduced grazing pressure, or a good seed-production year. These questions can only be answered if monitoring is an integral part of research, and particularly where it is used to test predictions, i.e. to separate noise from signal.

In selecting sites for long-term monitoring and research it is important to understand the fluctuations and trends that might be expected. Trends, whether they are determined by management or occur naturally, are often induced in response to events occurring years or decades previously, and knowledge of site history is particularly important. The state of the system will influence, for example, the deposition and retention of pollutants or nutrients through variation in vegetation cover and soil organic matter content; the plant growth response to climate through variation in species composition or stage of development; and the success of introduced species through variation in sites for colonization or in competition.

Although there are many variations on the theme, five general patterns of long-term trends are recognized - constant, cyclical, directional, episodic, and catastrophic. These are illustrated hypothetically in Figure 3.8, with short-term fluctuations superimposed. The y axis may represent many different parameters (p) for example, plant production, soil nitrogen or organic matter concentration, species diversity, and soil water table or moisture content. The trend is determined by the type of system. Thus, an arable system with consistent short-term management would be expected to show only annual variations around constant values of  p (Figure 3.8(a)) in contrast to repeated management or natural cycles such as heather burning and regeneration (Figure 3.8(b)) and natural succession or afforestation with a long-term directional trend (Figure 3.8(c)). The pattern induced by episodes such as drought or defoliation which allow recovery (Figure 3.8(d)) contrasts with catastrophic situations (Figure 3.8(e)) which result in a long-term change from one state to another, as in the conversion of grassland to arable. It can be argued that the constant, cyclical, and directional patterns simply represent expansion of time-scale with annual arable cropping, 20-year burning, and 50-year forest rotation. Similarly, the episodic and catastrophic patterns simply represent extreme variations in the rate or degree of recovery (e.g. from defoliation). Note also that in episodic or catastrophic situations, while some attributes such as standing crop or species diversity will change rapidly, others such as soil organic matter may continue to respond over decades: the echo from the past.

Figure 3.8 Main types of long-term trends in ecosystems in response to management or to natural succession, with short-term fluctuations superimposed (see text).

The real world is much more complex than the simple representations presented here. However, the key points arising from this discussion are that our ability to detect changes caused by particular human activities can be confused by the uncertainty of whether the changes result from previous actions or from natural trends such as pedogenesis or vegetation succession. In monitoring change, the greatest success is likely to be achieved when the monitoring is combined with research which includes prediction of fluctuations and trends against which data from monitoring can be tested. There are advantages in the selection of sites with a known history of management and which are sensitive to the targeted change. In the latter case, the site may be marginal to a species or crop production system, and change will be readily apparent, although the fact that fluctuations will also be greatest at the margin may cause a problem. 

3.4 SELECTION OF SITES

Environmental problems are complex. and understanding the cause-effect relationships requires in-depth study involving a number of disciplines which must focus, sooner or later, on the field situation. For this, there must be a stage of site selection. The normal process is to identify sites which are well known, then see how they fit together to cover environmental variation which at best has been broadly defined. During the International Biological Program sites were selected by participant countries for research on production processes and then combined to form international Biome networks. Analysis of the Tundra Biome, constituted in this way, showed that the sites covered a wide range of variation along environmental gradients of temperature, moisture, soil organic matter, acidity, and nutrient availability (Figure 3.9), against which biological processes could be related (French, 1981). This approach of selecting then classifying sites resulted in important gaps, although flexibility in the Canadian program did allow the late selection of a missing high arctic site at Devon Island. The classification also showed that local site conditions (for example. the wet and dry subsites on Devon Island) had more in common with geographically distant subsites (Finland, Antarctica, Ireland) than with their immediate neighbors. This 'accidental' inclusion of local variation greatly enhanced the potential of between-site comparison and synthesis for determination of global controls of production and decomposition processes (Bliss et al., 1981).

The method of site selection in the International Biological Program had certain drawbacks. An alternative would have been to determine the environmental variation to be covered, stratify, and then select sites within the strata. An example of this approach was an ecological survey of the UK which was subsequently used in assessing change in land cover and use (Bunce and Heal, 1984). From a hierarchical classification based on climatic, topographic, and geological attributes, 32 land classes were used as strata from which 256 sample sites (1 km²) were randomly selected (eight squares in each stratum) for survey and monitoring. This approach allows ease of regional and national quantification and of extension of sampling for particular purposes. It does not select sites for which information is readily available, but it provides a framework within which these sites can be placed. Thus, the two approaches of select and classify or classify and select can be complementary rather than exclusive.

Figure 3.9 (opposite) Abiotic analysis of tundra sites (from French. 1981), showing the distribution of sites along components I and II, indicating primary clusters. The analysis used seven climatic variables, four soil temperature variables, two soil moisture variables and eight chemical variables. Arrows show the nearest linkages of 'outlier' sites. Main site codes: G: Glenamoy, Ireland; MH: Moor House, UK; H: Hardangervidda, Norway; K: Kevo, Finland; A: Abisko, Sweden; D: Devon Island, Canada; B: Point Barrow, Alaska. USA; T: Tareya, Taimyr, USSR; M: Macquarie Island, Australia; SG: South Georgia, Antarctica; S: Signy Island, Antarctica; DK: Disko Island, Greenland; N: Niwot Ridge, Colorado, USA. Other letters and numbers in codes refer to sub-sites.

As already indicated, sites which are at the margin of the distribution of a species or land use will be more sensitive to change in environmental conditions than those towards the center of the range. Thus, these latter sites, which are considered to be typical or representative and are selected for characteristics close to the mean values will, for some purposes, be of limited value. The tree line at high latitudes or elevation provides an obvious example, and one which corresponds with the ecotone concept (di Castri et al., 1988). However, an important component of the ecotone concept is zones of transition in time. In such situations a small change in a particular variable such as temperature or element concentration can cause a relatively large change in the state of the species or system, i.e. it is at or near a threshold. di Castri et al. (1988) tended to focus on spatially defined boundaries or zones of transition, but some boundaries are temporally defined, particularly by the state of soil conditions. This is best shown by the following three soil examples.

Miles and Young (1980) showed that soil improvement by birch tended to occur on brown podzolic soils in Scotland and northern England. They selected sites with different ages of birch (chronosequences) and found that the ability of birch to reduce acidity and increase nutrient availability was determined by the initial state of the soil; some sites changed while others did not.

In a second example, J. D. Ovington, W. H. Pearsall, and others hypothesized that some soils were particularly susceptible to the effect of coniferous planting, i.e. they were on a threshold between soil types. They selected a site at Gisburn in north-west England on which Pinus sylvestris, Picea abies, Alnus glutinosa , and Quercus petraea were planted in both pure and mixed stands. After 30 years, distinction between the effects of conifer and deciduous trees on soil conditions were apparent, with some unexpected responses (Brown and Harrison, 1983).

In the third example, Hornung et al. (1989) formalized the theoretical and practical information on the sensitivity of waters to acid deposition and afforestation. Based on climatic, geological, and pedological data, they mapped the distribution of areas in Wales in which water acidification (and Al toxicity) would be most likely to occur through afforestation because of the soil buffering characteristics and the increased deposition in forests (Fowler et al., 1989) (Figure 3.10). The acid soils with Al buffering were sensitive in the short term, and those which have little residual Ca buffering were found to be sensitive in the medium term.

These three examples emphasize the threshold or poised nature of some soils that will determine the response of the plant community and chemical condition to changes in management or pollution. Thus, selection of sites will determine whether or not a response is detectable. The first two examples also illustrate alternative approaches in research design, i.e. selection of a series of sites which represent stages in time as distinct from long-term analysis of an experimental manipulation. The third example indicates the possibility of selecting sites to test defined hypotheses. In all three examples, the concept of a threshold of response is explicit (though not quantified) in the selection of sites. Reversibility is another question, but it is one that Miles (1988) is examining by following the observational chronosequence approach with experimental planting of Calluna on cleared birch plots to complement validation experimental planting of birch on Calluna sites.    

Figure 3.10  Probable occurrence of acid waters in Wales predicted from soils, geology and land use.

The identification of constant, successional, or recurrent management as an important criterion in site selection has already been made, and is illustrated in Figure 3.8. Having identified the management and environmental conditions, the question then is the selection of intensive sites which combine into a network. However, an important variation on the theme is the organization of a network combining intensive and extensive sites. The concept was clearly planned in the US Grassland Biome of IBP (van Dyne, 1972), with detailed process-related research at the intensive site designed to predict responses over a wider range of conditions, the predictions being tested at the extensive series of sites. Variations on this approach, using a number of sites on which limited observations are made to complement an intensive site, are distinct from a network of many sites with the same measurements, and can provide a powerful tool in combining detailed understanding with widespread application. 

3.5 CONCLUSIONS

Many of the points made in this chapter are obvious, but the experience of the current discussion in the UK indicates that they are worth re-examining. The main lessons seem to be:

1. Defining the problem is critically important in selection of sites for long-term ecological research and monitoring. The problems of the future may be unforeseen, but are likely to belong to four generic types which have particular requirements (climate change, pollution, land use, species introductions).
2. Definition of inherent short-term fluctuations and longer-term trends will be important in detecting changes attributable to more recent activities of man. Because of this, monitoring is likely to be most useful when combined with research, and when based on sites which have a good knowledge of management history and understanding of processes.
3. For many problems the influence of man is most likely to be important and detectable under conditions which are marginal to the species or system, or which are transitional between states. Thus the selection of ecotone or threshold sites can be as important as selection of sites which represent more average conditions, depending on the question.
4. Selection of sites should include careful analysis of the range of variation to be covered and can usefully incorporate and build on a planned network of intensive and extensive sites.

Throughout this chapter questions of spatial and temporal scale have been implicit rather than explicit. The questions are appropriate to site selection and to the measurements to be made on the sites, from soil crumb to catchment. Many of the issues are identified in Risser (1986) with the clear message that the scales adopted, and the interrelationships between the hierarchy of scales, must  be carefully identified in relation to the problem being considered. The challenge in long-term studies is in economy of effort to answer the problem. As identified by participants in the International Biological Program: 'We underestimated our ability to collect data and overestimated our ability to do something with them.'

3.6 REFERENCES

Ball, D. F., Dale, J., Sheail, J. and Heal, O. W. (1982). Vegetation Change in Upland Landscapes. Institute of Terrestrial Ecology, Cambridge, UK.

Bliss, L.C., Heal, O. W. and Moore, J. J. (1981). Tundra Ecosystems: a Comparative Analysis. Cambridge University Press, Cambridge, UK.

Brown, A.H. F. and Harrison, A. F. ( 1983). Effects of tree mixtures on earthworm populations and nitrogen and phosphorus status in Norway spruce (Picea abies) stands. In Lebrun et al. (Eds) New Trends in Soil Biology. Dieu-Brichart, Louvain-la-Neuve, 101-108.

Bunce, R.G.H. and Heal, O.W. (1984). Landscape evaluation and the impact of changing  land use on the rural environments: the problem and an approach. In Roberts, R.D. and Roberts, T. M. (Eds) Planning and Ecology, Chapman & Hall, London, 164-188.

Coulson, J.C. and Whittaker, J. B. (1978). Ecology of moorland animals. In Heal, O. W. and Perkins, D. F. (Eds) Production Ecology of British Moors and Montane Grasslands. Springer-Verlag, Berlin, 52-93.

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

di Castri, F., Hansen, A. J. and Holland, M. M. (1988). A new look at ecotones. Emerging International Projects on Landscape Boundaries. Biology International, Special Issue 17.

Elton, C. (1958). The Ecology of Invasions by Animals and Plants. Methuen, London. 

Fowler, D., Cape, J. N., Deans, J.D., Leith, J.D., Murray, M. B., Smith, R.I., Sheppard, L. J.  and Unsworth, M. H. (1989). Effects of acid mist on the frost hardiness of red spruce seedlings. New Phytologist, 113, 321-335.

Fowler, D., Cape, J. N. and Unsworth, M. H. (1989). Deposition of atmospheric pollutants on forests. Phil. Trans. R. Soc. Lond. B., 324, 247-265.

French, D. D. (1981). Multivariate comparisons of IBP Tundra Biome site characteristics. In Bliss, L.C., Heal, O. W. and Moore, J.J. (Eds) Tundra Ecosystems, a Comparative Analysis. Cambridge University Press, Cambridge, UK, 47-75.

Grace, J. and Marks, T.C. (1978). Physiological aspects of bog production at Moor House. In Heal, O.W. and Perkins, D.F. (Eds) Production Ecology of British Moors and Montane Grasslands. Springer-Verlag, Berlin, 38-51.

Grime, J.P. (1987). Dominant and subordinate components of plant communities: implications for succession, stability and diversity. In Gray, A.J., Crawley, M. J. and Edwards, P. J. (Eds) Colonization, Succession and Stability, Blackwell Scientific, Oxford, 413-428.

Hornung, M., Le Grice, S., Brown, N. and Norris, D. (1989). The role of geology and soils in controlling surface water acidity in Wales. In Stoner, J. and Edwards, R. W. (Eds) Acid Waters in Wales. Kluwer Academic Publishers, Dordrecht, The Netherlands, 55-66.

Kornberg, H. and Williamson, M. H. ( 1987). Quantitative Aspects of the Ecology of Biological Invasions. The Royal Society, London.

Likens, G. E. ( 1985). An experimental approach for the study of ecosystems. The fifth Tansley Lecture. J. Anim. Ecol., 73, 381-396.

Likens, G. E. (Ed.) (1989) Long-Term Studies in Ecology. Springer-Verlag, New York. 

Marrs, R .H., Bravington, M. and Rawes, M. (1988). Long-term vegetation change in the Juncus sqarrosus grassland at Moor House NNR in northern England. Vegetatio, 76, 179-187.

Miles, J. (1979). Vegetation Dynamics. Chapman & Hall, London.

Miles, J. (1988). Vegetation and soil change in the uplands. In Usher, M. B. and Thompson, D. B. A. (Eds) Ecological Change in the Uplands. Blackwell Scientific, Oxford, 57-70.

Miles, J. and Young, W. (1980). The effects on heathland and moorland soils in Scotland and northern England following colonization by birch (Betula spp ). Bulletin d´ Ecologie , 11, 233-244.

Noble, I. R. and Slatyer, R. O. (1980). The use of vital attributes to predict successional changes in plant communities subject to recurrent disturbances. Vegetatio, 43, 5-21.

Parry, M. L. and Carter, T. R. (1985). The effects of climatic variations on agricultural risk. Climatic Change, 7, 95-110.

Peterkin, G. F. and Jones, E.W. (1989). Forty years of change in Lady Park Wood: the young-growth stands. J. Ecol., 77, 401-429.

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

Risser, P. G. (1986). Spatial and temporal variability of biospheric and geospheric processes. Report of a Workshop, ICSU Press, Paris.

Roberts, T. M., Skeffington, R. A. and Blank, L. W. (1989). Causes of Type 1 spruce decline in Europe. Forestry, 62, 180-222.

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

Thurston, J. M., Williams, E. D. and Johnson, A. E. (1976). Modem developments in an experiment on permanent grassland started in 1856: effects of fertilisers and lime on botanical composition and crop and soil analyses. Annales agronomiques, 27, 1043-1082. 

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

Usher, M. B. and Thompson, D.B. A. (Eds) (1988). Ecological Change in the Uplands. Blackwell Scientific, Oxford.

Van Dyne, G. M. (1972). Organization and management of an integrated ecological research programme - with special reference on systems analysis, universities and scientific co-operation. In Jeffers, J.N.R. (Ed.) Mathematical Models in Ecology. Blackwell Scientific, Oxford, 111-172.

Watt, A. ( 1960). Population changes in acidiphilous grass-heath in Breckland, 1936-57. J . Ecol., 48, 605-629.

Whittaker, J. B. (1971). Population changes in Neophilaenus lineatus (Homoptera: Cercopidae) in different parts of its range. J. Anim. Ecol., 40, 425-443.

Whittaker, J. B. (1985). Population cycles over a 16-year period on an upland race of Strophingia ericae (Homoptera: Psylloidea) on Calluna vulgaris. J. Anim. Ecol., 54, 311-321.