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

We have just enough time left in this century to achieve a major new synthesis and understanding of the Earth System.... (NASA, 1988)

 9.1 INTRODUCTION

The earth, with its global problems of overpopulation, over-use and abuse of fossil fuel and nuclear energy, and production of toxic wastes, has often been compared to a sick patient. Illness is recognized as a significant deviation from known, long- term trends. Long-term monitoring represents a minimal activity for responsible individuals and agencies interested in placing current environmental problems into perspective. Long-term measurements are directed at questions involving phenomena not interpretable or perhaps not useful when viewed over short (annual or less) time scales, but are related to the long-term 'health' or functioning of the system. At a minimum, the Long-term Ecological Research (LTER) data therefore provide the context in which short-term observational or experimental results can be interpreted (Magnuson, 1990). A much more interesting, albeit potentially less relevant, use of LTER data involves the study of a set of complex questions that cannot be resolved with short-term studies (Franklin, 1989; Tilman, 1989). The juxtaposition of basic and applied science within the context of a single research effort is a strength of the LTER program.

This chapter attempts to identify a set of long-term ecological questions that are useful to a national or international network of research sites. While there exists a nearly infinite list of interesting questions that could be addressed with long-term studies, a realistic and goal-oriented list of measurements is presented. The criteria for selecting these questions involved identifying variables that (1) are useful for intersite comparisons, (2) are not strongly biased by spatial scaling factors, and (3) can provide the necessary linkages between atmospheric/climatological variables and biological measurements. 'Focused studies of the interactions between the atmosphere and the biosphere that regulate trace gases can improve both our understanding of terrestrial ecosystems and our ability to predict regional- and global-scale changes in atmospheric chemistry' (Mooney et al., 1987). The list of proposed variables for study was developed from the 'core LTER measurements', a guideline used since the inception of the LTER effort (Callahan, 1984) from recommendations suggested in Earth System Science (NASA, 1988), and from practical experience with the recent NASA-ISLSCP (International Surface Land Climatology Project) conducted on the Konza Prairie LTER site (Sellers et al., 1988). While appropriate examples are taken from many systems, particular emphasis has been given to questions that have interested researchers studying grasslands. We build on the work of Strayer et al. (1986). Their extensive overview of long-term studies provided useful definitions of research productivity, of what constitutes 'long-term research', and reasons for the 'successes' of previous and existing long-term research efforts. Their findings emphasized that individual scientists and not specific research protocols or experimental designs were largely responsible for successful long-term research efforts. Here, however, we suggest that certain constraints on research designs are important if a goal of the research is to benefit directly a regional or global network.

9.2 APPROPRIATE OBJECTS FOR LONG-TERM NETWORK MEASUREMENTS

The five core areas of the LTER include studies of the following topics (Callahan, 1984):

1. Spatial and temporal distributions of populations;
2. Patterns and frequency of disturbance;
3. Pattern and control of primary production;
4. Pattern and control of organic matter accumulation; and
5. Patterns of inorganic input and movements through soils.

While excellent research has been done on some or all of these topics at one or more of the LTER sites, current efforts of linking sites in regional or global networks suggest that certain measurements are likely to be more useful than others.

Many of the most interesting and useful empirical studies of individual species have been long term in nature (Iker, 1983; Strayer et al., 1986). For example, Weaver (1954) documented the response of the North American prairie species to climate and grazing intensities observed over a 40-year interval in the first half of the twentieth century. The rainfall and temperature conditions under which these studies were made are exemplified by data obtained for the Manhattan, Kansas, area (Figure 9.1 ). These data show that the great drought in North America during the 1930s was accompanied by (and contributed to) relatively high ambient temperatures. During the drought, Weaver documented the eastward advance and expansion of xeric, shortgrass species at the expense of mesic, tallgrass vegetation. The return of relatively wet years in the 1940s reversed this trend. The annual values of temperature and precipitation shown in Figure 9.1 indicate that 'average' conditions for the prairie cannot be expected without an approximately 20-year record. Even then, the factors that govern patterns of species composition and abundance may be misinterpreted or overlooked. Weaver ( 1954 ), for example, did not appreciate or acknowledge the role of fire in suppressing the invasion of woody species onto the tallgrass prairie, nor did he notice that the productivity of the dominant species was often enhanced by frequent fires.

Studies of within- and between-habitat species diversity remain of keen interest to many ecologists. Nonetheless, we suggest that individual species, species lists, or indices derived from species lists make poor primary intersite comparison measurements. Many species are not found across large environmental gradients. Those species that do cover regional areas are not physiologically identical across these regions. The relevant units to address intersite comparisons should confer equivalency across sites, and these units should aggregate into meaningful values at different spatial scales. Energy and mass (including elements, trace gases, etc.) are obvious candidates for study. Biologists must still focus on the biota as cause and effect participants in energy and mass transformations, but both the forcing functions and the response variables must employ units common to all sites. Eventually, life history characteristics and physiological responses of the individual species will provide a mechanistic interpretation of site-specific responses. Even then, however, these responses will be governed by spatial patterns not often measured in population studies (Huston et al., 1988).

All LTER sites have been charged with studying 'disturbance' as a core measurement. Our own experience with this topic has suggested serious problems, associated with the concept that may prevent 'disturbance ecology' from becoming a major tenet of ecological theory (Evans et al., 1989). One problem has been the popularity of the topic, and the inevitable misuse of the term that comes with popularity. 'Disturbance' is used simultaneously to describe a system input (e.g. a storm) and system output (e.g. species die-off) (Rykiel, 1985). Obviously, the latter is the interaction of the system with an input, and is therefore very much a characteristic of the system while the former is uncontrolled by the state of the system. A second problem with disturbance theory is that identical inputs can produce very different outputs depending upon the initial state of the system and the scales at which the output is measured. For example, fire adversely affects a number of populations of plants and animals in the tallgrass prairie. Nonetheless, certain species are benefited and periodic fires are required for the perpetuation of the system. Is fire or the absence of fire the disturbance in this system? Can systems lacking stable equilibria be disturbed? System-level properties of resistance and resilience to disturbances can be viewed more logically and mechanistically as consequences of structural and life-history characteristics of biological systems. Different disturbances (fire, drought, grazing, etc.) produce very different species and system responses. Konza Prairie researchers found the discussions about species and ecosystem responses to 'disturbances' to be largely an exercise in after-the-fact descriptive ecology and a topic not conducive to the development of predictive models. A much more productive approach to generic disturbance-type questions involves explicit identification of forcing functions and the responses of the system  at specific levels of resolution. In other words, we believe that the LTER core area involving disturbance can be adequately addressed within the context of studies focused on the other core areas. This is certainly true in grasslands and agroecosystems where studies use fire, grazing, or tillage practices as experimental manipulations. 

Figure 9.1 Deviations from average annual precipitation (a) and mean maximum temperatures (b) for an area near Manhattan, Kansas and the Konza Prairie LTER site. The data are for the period 1891-1987. Note that the temperature maxima corresponded with low precipitation values during the 1930s.

The remaining three core areas of the LTER program (net primary production, organic matter, and nutrient dynamics) provide a logical, unified focus for regional and global networks. These core areas employ units that are constants and provide the direct links between biotic and atmospheric processes. A combination of relatively new, spatially explicit measurements, in conjunction with traditional methodologies, will allow ecologists to study biota-climate interactions while concurrently focusing on questions of local interest.

 9.2.1 PRIMARY PRODUCTIVITY

Forested sites have considerable potential to demonstrate the linkages among net primary productivity, trace gases, and climatic changes. Dendrochronology studies have used annual woody growth increments to reconstruct recent past climates. Other studies have combined paleobotany, records of lake ash deposition, and dendrochonology to reconstruct forest species composition, fire frequency, and growth relationships. Clark (1988) demonstrated the relationship between climate and fire frequency which, together, shaped the species composition and   productivity of the north temperate forests. Of particular interest has been the work of LaMarche et al. (1984) which suggests that subalpine forests in western North America began to alter their growth patterns with respect to climatic variables some time in the 1960s. Those authors suggested CO₂ enrichment as a possible factor. Anthropogenic sources of nutrients in bulk precipitation could, perhaps, be an alternative hypothesis. Regardless, the measurement of woody growth and, therefore, a record of the past productivity is possible at many sites, and is a reasonable, partial index of above ground net primary productivity. Such data are particularly desirable since (1) sampling can be accomplished on a very infrequent, year-to-decades basis, (2) large sample sizes can be obtained and potentially interacting variables (soils, species, etc.) can be evaluated, and (3) the samples can be easily archived so that future analyses or reanalysis of the same, original data set are possible. To complete the story of above ground productivity, foliage production should be measured. Litterfall or needle production measurements and procedures are common, but should be supplemented, if possible, with satellite-derived digital images. These images can provide a spatial perspective not possible with microplot measurements, and the types and uses of currently available satellite images are discussed below. 

Retrospective analyses of grassland productivity cannot be as easily accomplished as forest studies. Sedimentation rates of glacial lakes, in conjunction with pollen analyses, may provide some useful historical data. Also, carbon isotope studies of sediments, soils (including paleosols), and groundwaters in conjunction with these or other research may also provide an interesting story, particularly with respect to changes in the composition of C₃ and C₄ plants (O'Leary, 1988).

Table 9.1 Characteristics of current satellite imagery^a

More recent retrospective analyses of indices of grassland productivity can also be conducted using the satellite imagery. Researchers and sites should move quickly to secure these images lest useful information be lost by agencies not funded as data archives. A listing of potential data sources (Table 9.1 ) indicates the resolution and information available from each type of satellite. Investigators need to be aware of the various trade-offs involved in using these various types of data, and some important considerations are outlined in Sellers et al. (1988). In general, we believe that the high spatial resolution (small pixel size) of the Landsat TM or SPOT satellites is extremely useful in evaluating within-site topoedaphic or experimental (fire or grazing) effects. However, a seasonal time-series of these types of data is expensive or simply unlikely to be obtained due to relatively infrequent overflights in conjunction with moderate to high probabilities of cloud cover. In contrast, the NOAA-AVHRR satellite provides relatively low spatial resolution (large pixel size) but high temporal resolution, making cross-site, cross-year, and seasonal comparisons possible. 

The potential for using these images as analogs of regional productivity and for estimating trace gas interactions and energy exchange is just beginning to be developed. Recent improvements of algorithms, particularly those employing the vegetation index (Tucker et al., 1985; Goward et al., 1986) or some combination of the vegetation index in conjunction with thermal measurements (Forrest Hall et al., unpublished results), can demonstrate both seasonal and long-term trends in plant biomass and plant vigor. We expect that the more sophisticated, high-resolution imaging spectrometers scheduled for space orbit in the near future will provide more useful data for measuring both biomass and plant productivity at moderate scales. This enhancement begins with the anticipated launch of Landsat 6 with the Enhanced Thematic Mapper (ETM) on board. Eight bands of spectral information are planned, four in the visible (one being a 15 m pixel panchromatic), two in the near-infrared, and two in thermal portions of the spectrum. This system is reported to be very sensitive to surface temperature changes and should, therefore, be very useful in relating vegetation dynamics to energy dynamics. Subsequent satellite equipment scheduled for the Earth Observing System (EOS) program will make considerable advancements in the spectral resolution of these digital images. These standard products could also be supplemented with aerial photography, including standard panchromatic, color, and color IR images. Photographic records are proving useful for a variety of retrospective analyses.

9.2.2 THE INTERACTION BETWEEN PRODUCTIVITY AND SURFACE CLIMATE

A conceptual model developed by Shugart (1986) (Figure 9.2) suggests how we might think about the relationship of LTER measurements to studies involved in trace gas fluxes. The latter measurements are, by necessity, made on a scale that detects strong diurnal and seasonal fluctuations. In contrast, LTER measurements of NPP, organic matter, or elements have a much coarser temporal scale. However, as suggested by the model, these long-term ecological processes function as constraints on short-term physiological processes, and therefore mediate the response of vegetation to climate. We present an example of this phenomenon to emphasize the need to recognize that changes in ecological constraints such as fire frequency, herbivory, or nutrient availability may temporarily overshadow direct changes in temperature or rainfall.

Figure 9.2 Conceptual model by H.H. Shugart suggesting the relationships between LTER- type measurements (right side of figure) and those variables strongly influenced by diurnal variations (left side of figure).

Figure 9.3 Time series of maximum foliage production on annually burned and unburned prairie. Year-to-year climatic fluctuations affect the vegetation response to treatment.

Figure 9.4 Weekly mean minimum-maximum soil temperatures in summer 1987 for burned (solid lines) and unburned (dashed lines) tallgrass prairie at 2 cm and 10 cm soil depths. Note that temperatures are relatively cooler on the burned site at 2 cm but are relatively warmer at 10 cm.

Our data on temperate grassland plant productivity demonstrate a strong relationship between the type of management treatment and productivity (Figure 9.3). The tallgrass prairie requires periodic fires to maintain its species composition and productivity (Knapp and Seastedt, 1986). In average or wet years, annual burning in late spring benefits the C₄ grasses. However, some but not all drier than average years result in more productivity by the combination of C₄ and C₃ grasses, forbs, and woody species found in the unburned prairie. Following a fire, the blackened soil surface of burned prairie is exposed to direct solar radiation and converts much of this energy into sensible heat absorbed by the soil (Figure 9.4). However, by midsummer, the re-establishment of the canopy, in conjunction with greater rates of evapotranspiration, results in a cooler soil surface. This pattern is reversed at the 10 cm depth, where the drier soils on the burned sites lack the thermal inertia of generally moistened, litter-covered soils of the unburned sites. The greater rates of evaporation coupled, perhaps, with higher rates of reflected infrared radiation keep burned areas cooler in midsummer than unburned areas (Figure 9.5). The thermal (channel 6) Landsat TM image in Figure 9.5 shows that the radiometric brightness on burned watersheds is, on average, less than that measured for adjacent unburned areas (Figure 9.6) (Asrar et al., 1988). These data demonstrate that the ecological constraints operating on the vegetation (here, a spring fire) influence both the hydrologic and energy budget. These changes are detectable at both a micro- and macro-scale level. Obviously, a change in the fire frequency of relatively large tracts of grassland could have an impact on the regional climate.

Grazing by cattle also had a measurable affect on canopy temperatures as measured by the radiometric brightness of the TM image (Figures 9.5 and 9.6). Grazed areas were cooler in August, presumably because the grazed vegetation was physiologically more active than a similar amount of ungrazed vegetation and was transpiring relatively greater volumes of water. Consumers affect both the amounts and physiology of the vegetation and thereby can greatly alter vegetation-climate interactions, particularly in grasslands. Investigators should also be aware that interactions between energy and nutrients may affect consumers, so that consumers become important transient controlling factors on net primary productivity (White, 1984). These controls can operate directly via consumption of plant parts or indirectly by controlling plant species composition (Schowalter, 1981). Thus, knowledge of consumer populations may contribute to an understanding of vegetation-climate interactions. This observation also has particular relevance in agroecosystems, where biotic mechanisms of consumer regulation have been severely altered. 

Figure 9.5 A Landsat TM thennal (channel 6) photo of Konza Prairie. Watershed boundaries have been superimposed over the image. Burned watersheds or grazed pastures are distinguishable by the darker pixel values.

Figure 9.6 Means and standard deviations of pixel values of radiometric canopy brightness obtained from Figure 9.5. All treatments exhibit statistically different values.

9.2.3 NUTRIENTS

Virtually all LTER sites measure nutrient inputs, standing crops, and outputs. The input data may be restricted to analyses of wetfall, often associated with the National Atmospheric Deposition Program (NADP). This measurement is often inadequate because dryfall deposition or deposition associated with dew can be considerable (Lindberg et al., 1986). Most sites obtain pH measurements in conjunction with the inputs of nitrate, ammonia, sulfate, and the major cations. Measurements of the standing crops of the major elements in vegetation were initiated at many sites during the International Biological Program. It is to be hoped that such data have been archived for future analyses or as baselines for future comparisons. Our site archives plant and soil samples along with the numerical data, and this procedure has received endorsement by other research groups (Pace and Cole, 1989). To our knowledge, no LTER site has engaged in long-term monitoring of net inputs or outputs of trace gases (CO₂, NO_x, NH₃, H₂S, or SO₂). However, with the advent of large path-length infrared spectroscopy (Gosz et al., 1988), and procedures to estimate fluxes, this deficiency should be resolved, at least at a few sites. Moreover, as mentioned above, the trace gas fluxes are diurnal phenomena operating under the ecological constraints being studied by the LTER. Empirical results and modeling efforts currently underway as part of FIFE (First ISLSCP Field Experiment) at Konza Prairie should be able to tell us the relationships and sensitivity of measurements such as productivity to short-term and seasonal estimates of gas flux. 

Nutrients become constraints on plant growth during periods when energy and water are not limiting, i.e. under conditions otherwise favorable for plant growth. An obvious question of interest to those involved with climatic change studies is the extent that nutrient limitations may mediate vegetation responses to enhanced CO₂ (Tissue and Oechel, 1987). If plant growth is nutrient as opposed to energy limited, then carbon dioxide enrichment and/or increased temperatures should not immediately affect productivity. In tallgrass prairie, an improved energy environment (created by fire) results in a higher nitrogen use efficiency (NUE) of the vegetation (Ojima, 1987). With this greater production, however, comes increased detritus build-up and nutrient immobilization. In several biomes, including the taiga (van Cleve et al., 1983) and tallgrass prairie (Knapp and Seastedt, 1986), plant litter has a direct negative physical effect on energy availability to plants. Detritus production could, therefore, affect productivity both by affecting usable energy inputs and by influencing nutrient availability. Seasonal shifts in energy, nutrient, and water limitations, in conjunction with negative feedbacks resulting from biomass production, prevent these ecosystems from maximizing their production responses. 

The need for long-term nutrient measurements relates to the fact that climate can influence both amounts and availability of nutrients. For example, the rainfall data shown in Figure 9.1 indicate that precipitation at the Konza Prairie LTER site was above normal for the period 1981 to 1987. Accordingly, plant productivity was above the long-term average during this interval. Since inorganic nitrogen availability in soils is inversely related to the amount of  'new'  fixed carbon present, the organic matter build-up during this interval undoubtedly adversely affected inorganic nitrogen availability to plants. Generalizations about nitrogen availability and cycling and its importance to vegetation made during this wet interval are therefore biased and potentially incorrect, in spite of a seven-year data base.

Agroecosystems have additional nutrient inputs and outputs not found or not important in natural systems. Nutrient supplements from fertilizers and outputs in the form of harvested plant parts tend to create an artificially dynamic system. Areas employing irrigation also have potential additional exports of trace gases or leaching losses, and certain agricultural practices, such as conversion of largely aerobic, vegetated sites to largely anaerobic rice fields, are probably having a large effect on trace gas dynamics (Mooney et al., 1987). A detailed accounting of these nutrients is warranted given the progressive enrichment of groundwaters with undesirable organics and nitrates.

Moreover, the tillage of the soil, the artificial, excessive harvesting of plant nutrients, in conjunction with applications of supplemental water and fertilizers, have created unique situations of nutrient limitation, soil acidification and aluminium toxicity problems for agricultural systems (Adams, 1984). Indeed, many sites have been so totally altered by intensive agricultural practices that moderate changes in temperature, rainfall, rainfall chemistry, or rainfall pH would appear of secondary consideration relative to the direct human manipulations. The relevant emphasis from a network standpoint is, therefore, not how these systems are affected by climate change scenarios, but rather how the systems are affecting regional energy and trace gas dynamics. The ecological constraints of agroecosystems are the crop and tillage manipulations. These, like fire and grazing in the prairie, control the system interactions and responses to climatic inputs. 

Measurement of nutrient outputs from ecosystems has proved to be an extremely relevant and useful long-term index of integrated system behavior. Likens et al. (1977), Likens (1983) and Driscoll et al. (1989) have provided ample examples of these measurements. Their 25+ year effort on the relationships between nitric and sulfuric acid rain inputs and stream pH and stream nutrient responses comprises some of the most relevant and important ecological research of this century. Ironically, this work began with some focused, short-term experimental studies, but the utility of these measurements for questions requiring a longer study period became obvious shortly after the initiation of the experiments. Stream chemical analyses have provided a measurement of the integrated ecosystem response to changes in atmospheric inputs or to those induced by within-system manipulations. In similar fashion, the new generation of remote sensing equipment scheduled for earth orbit within the next 10 years should provide equivalent information for terrestrial systems. Multispectral scanner, high-resolution sensors will provide a spatially explicit measurement of the integrated landscape response to changes in atmospheric inputs and landscape manipulations. Certain chemical properties of vegetation such as water status and nitrogen content can already be measured to some extent with current satellite data (Rock et al., 1986; Waring et al., 1986). 

9.2.4 ORGANIC MATTER

Plant detritus and soil organic matter provide the major reservoir of nutrients in most terrestrial ecosystems. This storage component provides the 'resistance' of the system to changes caused by the destruction of the vegetation. The tropics-to-taiga gradient in organic matter is an example of the interaction between net primary productivity, decomposers, and climate (Swift et al., 1979). Any brief interpretation of this pattern is an oversimplification. Nonetheless, plants appear to have dealt better with climatic restraints than have the decomposers. In the United States the east-to-west gradient in soil organic matter observed across the prairie is largely controlled by moisture (Jenny, 1930). Prediction of changes in the organic reservoir, therefore, potentially depends upon the interaction of temperature and moisture, and the net effects that these variables have on production and decomposition (Hunt et al., 1988). Soil organic matter measurements tend to be rather insensitive to short-term manipulations of productivity and decomposition, but should be useful monitors of long-term changes (Jenny, 1980; Ojima, 1987; Ojima et al., 1990). Moreover, such data are generally available on a regional basis, and have been modeled very successfully using climate and management constraints as forcing functions (Parton et al. 1987a,b).

Investigators need to recognize that edaphic factors and climatic variables produce interaction effects that add to the complexity of regional patterns. A recent example is from Sala et al. (1988). That study found that sites with coarse, sandy soils were relatively more productive than fine, clay soils under below-average rainfall, while clay soils were relatively more productive under average or above-average rainfall. The coarse soils lacked the fertility of fine soils, but tended to allow water to penetrate below the zone of evaporation. Hence, the variability in productivity in coarse soils was reduced (i.e. productivity in wet years was diminished while the consequences of drought years were less severe). This phenomenon, when linked to a climatic gradient, produces a complex pattern (Figure 9.7). The relevance of these findings to models linking ecosystems to global climatic models should be particularly obvious.

Figure 9.7 An 18-year record of maximum monthly stream flow from two US Geological Survey Benchmark watersheds (data courtesy USGS)

9.2.5 SCALING AND SAMPLING CONSIDERATIONS

The problem of how to integrate point measurements so that these data can be useful and accurate estimates of regional dynamics remains unresolved. A large literature on scaling is developing (Allen et al., 1984; Urban et al., 1987). By far the most productive approach we have seen involves the use of explicit spatial models to aggregate ecosystem processes (Huston et al., 1988). Successful large-scale regional models of net primary production, nutrient cycling, and organic matter dynamics have to date employed a coarser approach based on the ecological constraints of climate and soils (Parton et al., 1987a,b). However, plans are underway to interface the fine-scale, spatially explicit models as inputs to the larger scaled models (Shugart et al., Chapter 12, this volume). We believe that a minimum of a two-step approach (organismic to ecosystem process level phenomena and ecosystem to global climatic models) will be required. Successful models will include those that adequately portray the operation of temporal and spatial ecological constraints on biotic processes. 

Inputs required for global scale models may require large spatial resolution but fine temporal resolution. As discussed above, such data will probably use satellite data and algorithms developed from FIFE-type projects (Sellers et al., 1988). Our own work with that project has convinced us that certain characteristics measured at small plot scales can be directly related to larger scale measurements (Figures 9.3 , 9.4 and  9.5). These measurements can be scaled up to function in input-output relationships with large-scale climate models. However, large errors will be introduced if the ecological constraints (i.e. land management) contributions are not included. In our region, changes in the ecological constraints to net primary productivity, i.e. changes in fire frequency or grazing intensity in prairie or in cropping and tillage practices in agroecosystems, will alter these algorithms. 

Rules for minimum sample frequency and minimum sample size must be developed, based on knowledge of the underlying population variance (Wiegert, 1962; Kimmons, 1973; Greig-Smith, 1983). Investigators must decide what constitute significant shifts in their systems given the intrinsic level of variability. This analysis is relatively straight-forward for time-series measurements that are assumed to be measured without error or that have no within-site variance component. For example, given our current precipitation data base (Figure 9.1), annual rainfall at our site must deviate by about 50% of the mean before we have an 'unusual' wet or dry year (i.e. assuming a normal distribution, outside of the 95% confidence). For each measurement, investigators can calculate the necessary sample size needed to detect statistically significant differences given hypothetical differences in population means with known or estimated population variances. 

Consider the problems associated with measuring export of nutrients from two watersheds (Figure 9.7). In the Dismal River system, a few baseflow samples accompanied by a storm-event sampler should produce a very accurate measure of export. The lack of inherent variability and high degree of predictability in flow from this sandhills prairie makes detection of slight changes in mean values or in the magnitude of annual variations a potentially easy task. In contrast, the Blue Beaver Creek system in Oklahoma exhibits extreme variability and little predictability. Stream discharge appears to be largely controlled by surface runoff in this mixed-grass drainage. The ability to show some statistically significant change in export in this system as a result of changes in land management or climatological inputs would be difficult if not impossible for studies shorter than a decade in duration. Likens (1983) discussed similar problems for measuring element and ion export for northeastern United States streams.

Figure 9.8 Variation in foliage production in the central United States (mean values of good years minus mean values in bad years, divided by the overall mean). Reproduced with permission from Ecology (Sala et al., 1988) 

On a regional basis, the variability in stream flow appears to be correlated with the variability in annual foliage productivity. Figure 9.8 illustrates that the source of the Dismal River, the sandhills of Nebraska, is a relatively more stable environment for plant production than the drainage area of Blue Beaver Creek in the western plains of Oklahoma. This variability, itself, can become a long-term measurement, and analyses such as those conducted by Sala et al. (1988) should identify sites that have intrinsically lower variability. For studies interested in evaluating directional changes, low variability and high predictability appear to be a desirable characteristic. The Sala et al. data also demonstrate the need for a large number of sites to characterize the regional response to year-to-year climatic variability.

9.3 DOCUMENTATION AND DATA BASE MANAGEMENT

The value of creating permanent plots, adequately documenting procedures, and developing a user-friendly data base cannot be overemphasized. With few exceptions, data bases have not outlived the investigators who collected them (Strayer et al., 1986). Those that have survived have become ecological treasures. Tilman (1989) noted that about 90% of all field studies are three years or shorter in duration. Even these short-term studies, if adequately documented and site referenced, could be subsequently resampled for similar or other ecological questions. We believe that we have lost many thousands of dollars' worth of valuable data because a number of ecological studies with a 'short-term focus' were not well documented on our site. Since those projects were terminated, we have come up with a number of questions that could have been addressed using the original data if we could only locate the site where the original measurements were obtained. A similar argument can be made for user-friendly documentation of the data. We have found a variety of new questions for old data sets. These data can be quickly retrieved and reanalyzed, even in the absence of the individual(s) responsible for the original data set. One cannot be serious about measuring decade-to-century-level phenomena without making a serious time and financial commitment to documentation. Researchers are referred to Gurtz ( 1986) and other references in Michener ( 1986) for excellent guidelines in this area. 

9.4 CONCLUSIONS

The recent Earth System Science Report (NASA 1988), in its recommendations and review of ongoing and proposed research for the IGBP, concluded, 'The overwhelming importance of sustained, long-term measurements of global variables emerges clearly from these studies' (page 137). We contend that a subset of the LTER core measurements, NPP, nutrients, and organic matter dynamics are particularly appropriate for relating vegetation dynamics to surface climatological measurements at a regional or larger scale. Biophysical measurements obtained from small plots, measured under known ecological constraints, will scale up in a fashion conducive to the modeling approaches suggested by Urban et al. (1987) and Huston et al. (1988). In our region, these ecological constraints include the fire and grazing regimes of the grasslands, or the particular management practice imposed on agroecosystems. We do not mean to ignore biodiversity efforts, and emphasize that individual organisms are driving the spatially explicit site responses (Huston et al., 1988). Nonetheless, these effects must be translated into biophysical rather than simply biological units to be useful at the intersite level.

9.5 ACKNOWLEDGMENTS

We thank M.E. Gurtz, US Geological Survey, for the stream examples. Ideas about ecological constraints developed from discussions with D.S. Schimel. Climatic data were provided by Dr Dean Bark. We appreciate a review of an earlier draft of this manuscript by A.K. Knapp and C.L. Turner. Support for our research and the preparation of this effort was provided by the National Science Foundation (BSR- 8514327) and NASA (NAG-5-897) grants to Kansas State University.

9.6 REFERENCES

Adams, F. ( 1984 ). Soil Acidity and Liming, 2nd Edition, Amer. Soc. of Agronomy, Number  12, Soil Sci. Soc. America, Madison, WI.

Allen, T.F.H., O'Neill, R.V. and Hoekstra, T.W. (1984). Interlevel relations in ecological research and management. Some working principles from Hierarchy theory. USDA Tech. Rep. RM-110, Rocky Mountain Forest and Range Experiment Station, Fort Collins, CO.

Asrar, G., Harris, T.R. and Lapitan, R.L. (1988). Radiative surface temperatures of the burned and unburned areas in a tallgrass prairie. Remote Sensing of Environment, 24, 447-457.

Callahan, J.T. (1984). Long-term ecological research. BioScience, 34, 363-367.

Clark, J.S. (1988). Effect of climate change on fire regimes in northwestern Minnesota. Nature, 334, 233-234.

 Driscoll, C.T., Likens, G.E., Hedin, L.O., Eaton, J.S. and Bormann, F.H. (1989). Changes in the chemistry of surface waters. Environmental Science and Technology, 23, 137-143.

 Evans, E.W., Finck, E.J., Briggs, J.M., Gibson, D.J., James, S.W., Kaufman D.W. and Seastedt, T.R. (1989). Is fire a disturbance in grasslands? In Bragg, T.J. and Stubbendiech, J. (Eds) Proceedings of the Eleventh North American Prairie Conference. University of Nebraska Press, 159-161.

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

Gosz, J.R., Dahm, C.N. and Risser, P.G. (1988). Long-path technology for large scale ecological studies. Ecology, 69, 1326-1330.

Goward, S.N., Tucker, C.J. and Dye, D.G. (1986). North American vegetation patterns observed with meteorological satellite data. In Dyer, M.I. and Crossley, D.A., Jr (Eds) Coupling of Ecological Studies With Remote Sensing. US Department of State Publication 9504, Bureau of Oceans and International Environmental and Scientific Affairs, Washington, DC, 96-115.

Greegor, D.H. (1986). Ecology from space. BioScience, 36, 429-432.

Greig-Smith, P. (1983). Quantitative Plant Ecology. Studies in Ecology, Vol. 9. University of California Press, Berkeley.

Gurtz, M.E. (1986). Development of a research data management system. In Michener, W.K. (Ed.) Research Data Management in the Ecological Sciences. University of South Carolina Press. Columbia, SC, 33-38.

Hardisky, M.A., Gross, M.F. and Klemas, V. (1986). Remote sensing of coastal wetlands. BioScience, 36, 453-460.

Hunt, H.W., Ingham, E.R., Coleman, D.C., Elliott, E.T. and Reid, C.P.P. (1988). Nitrogen limitation of production and decomposition in prairie, mountain meadow and pine forest. Ecology, 69, 1009-1016.

Huston, M., DeAngelis, D. and Post, W. (1988). New computer models unify ecosystem theory. BioScience, 38, 682-691.

Iker, S. (1983). A lifetime of listening. Mosaic, 14, 8-13.

Jenny, H. (1930). A study on the influence of climate upon the nitrogen and organic matter content of the soil. Res. Bull 152, Missouri Agr. Exp. Station.

Jenny, H. (1980). The soil resource. Ecological Studies 37. Springer- Verlag, New York. 

Kimmins, J.P. (1973). Some statistical aspects of sampling throughfall precipitation in  nutrient cycling studies in British Columbia coastal forests. Ecology, 54, 1008-1019. 

Knapp, A.K. and Seastedt, T .R. ( 1986). Detritus accumulation limits productivity of tallgrass prairie. BioScience, 36, 662-668.

LaMarche, V.C., Jr, Graybill, D.A., Fritts, H.C. and Rose, M.R. (1984). Increasing atmospheric carbon dioxide: Tree ring evidence for growth enhancement in natural vegetation. Science, 225, 1019-1021.

Likens, G.E. (1983). A priority for ecological research. Bull. Ecol. Soc. Amer., 64, 234-243.

Likens, G.E., Bormann, F.R., Pierce, R.S., Eaton, J.S. and Johnson, N.M. (1977). Biogeochemistry of a Forested Ecosystem. Springer-Verlag, New York.

Lindberg, S.E., Lovett, G.M., Richter, D.D. and Johnson, D.W. (1986). Atmospheric deposition and canopy interactions of major ions in a forest. Science, 231, 141-145.

Magnuson, J .J .( 1990). Long-term ecological research and the invisible present. BioScience, 40, 495-501.

Michener, W.K. (Ed.) (1986). Research data management in the ecological sciences. The Belle W. Baruch Library in Marine Sciences Number 16, University of South Carolina Press, Columbia, SC.

Mooney, H.A., Vitousek, P.M. and Matson, P.A. (1987). Exchange of materials between terrestrial ecosystems and the atmosphere. Science, 238, 926-932.

NASA (Anonymous) (1988). Earth system science: a closer view. Report of Earth System Sciences Committee, NASA Advisory Council. University Corp. for Atmospheric Research, Boulder, CO.

Ojima, D.S. (1987). The short-term and long-term effects of burning on tallgrass ecosystem properties and dynamics. PhD Dissertation, Colorado State University, Fort Collins, CO. 

Ojima, D.S., Parton, W.J., Schimel, D.S. and Owensby, C.E. (1990). Simulating impacts  of annual burning on prairie ecosystems. In Collins, S.L. and Wallace, L.L. (Eds) Fire in North American Tallgrass Prairies. University of Oklahoma Press, Norman, 118- 132.

O'Leary, M. (1988). Carbon isotopes in photosynthesis. BioScience, 38, 328-336.

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

Parton, W.J., Schimel, D.S., Cole, C.V. and Ojima, D.S. (1987a). Analysis of factors controlling soil organic matter levels in Great Plains grasslands. Soil Sci. Soc. Am. J., 51,1173-1179.

Parton, W.J., Steward, J.W.B. and Cole, C.V. (1987b). Dynamics of C, N and P in grassland soils: a model. Biogeochemistry, 5, 109-131.

Rock, B.N, Vogelmann, J.E., Williams, D.L., Vogelmann, A.F. and Hoshizaki, T. (1986). Remote sensing of forest damage. BioScience, 36, 439-445.

Rosenzweig, C. and Dickinson, R. (1986). Climate-Vegetation Interactions. Rep. OIES-2, University Corp. for Atmospheric Research, Boulder, CO.

Rykiel, Jr, E.J. (1985). Toward a definition of ecological disturbance. Aust. J. Ecology, 10, 361-365.

Sala, O.E., Parton, W.J., Joyce, L.A. and Lauenroth, W.K. (1988). Primary production in the central grassland region of the United States. Ecology, 69, 40-45.

Schowalter, T.D. (1981). Insect-herbivore relationship to the state of the host plant: biotic regulation of ecosystem nutrient cycling through succession. Oikos, 37, 126-130.

Sellers, P.J., Hall, F.G., Asrar, G., Strebel, D.E. and Murphy, R.E. (1988). The first ISLSCP field experiment (FIFE). Bull. Amer. Meteor. Soc., 69, 22-27.

Shugart, H.H. (1986). Conclusions. In Rosenzweig, C. and Dickinson, R. (Eds) Climate- Vegetation Interactions. Report OIES-2, University Corp. Atmospheric Research, Boulder, CO, 152-154.

Strayer, D., Glitzenstein, J.S., Jones, C.G., Kolasa, J., Likens, G.E., McDonnell, M.J., Parker, G.G. and Pickett, S.T.A. (1986). Long-Term Ecological Studies: An Illustrated Account of Their Design, Operation, and Importance to Ecology. Occasional Publ. Inst., Ecosystem Studies Number 2. Institute of Ecosystem Studies, New York Botanical Garden, Millbrook, NY.

Swift, M.J., Heal, O.W. and Anderson, J.M. (1979). Decomposition in Terrestrial Ecosystems. Blackwell Scientific, Oxford.

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.

Tissue, D.T. and Oechel, W.C. (1987). Response of Eriophorum vaginatum to elevated CO₂ and temperature in the Alaskan tussock tundra. Ecology, 68, 401-410.

Tucker, C.J., Townshend, J.R.G. and Goff, T.E. (1985). Continental land cover classification using satellite data. Science, 227, 369-374.

Urban, D.L., O'Neill, R.V. and Shugart, H.H., Jr (1987). Landscape ecology. BioScience, 37,119-127.

Ustin, S.L., Adams, J.B., Elvidge, C.D., Rejmanek, M., Rock, B.N., Smith, M.O., Thomas, R.W. and Woodward, R.A. (1986). Thematic mapper studies of semiarid shrub communities. BioScience, 36, 446-452.

Van Cleve, K., Dryness, C.T., Viereck, L.A., Fox, J., Chapin, III, F.S. and Oechel, W. (1983). Taiga ecosystems in interior Alaska. BioScience, 33, 39-44.

Waring, R.H., Aber, J.D., Melillo, J.M. and Moore, III, B. (1986). Precursors of change in terrestrial ecosystems. BioScience, 36, 433-438.

Weaver, J.E. (1954). North American Prairie. Johnsen Publishing Co., Lincoln, Nebraska.

White, P.S. and MacKenzie, M.D. (1986). Remote sensing and landscape pattern in Great Smoky Mountains Biosphere Reserve, North Carolina and Tennessee. In Dyer, M.I. and Crossley, Jr, D. A. (Eds) Coupling of Ecological Studies With Remote Sensing. United States Department of State Publication 9504, Bureau of Oceans and International Environmental and Scientific Affairs, Washington, DC, 52-70.

White, T .C.H. ( 1984 ). The abundance of invertebrate herbivores in relation to the availability of nitrogen in stressed food plants. Oecologia, 63, 90-105.

 Wiegert, R.G. (1962). The selection of an optimum quadrat size for sampling the standing crop of grasses and forbs. Ecology, 43, 125-129.