I. INTRODUCTION

We propose to continue our long-term ecological research on temperate lakes and their surrounding landscape in the Northern Highland Lake District of Wisconsin (Magnuson et al. 1984) as part of the national, Long-Term Ecological Research (LTER) network (Callahan 1984, Brenneman 1989, Magnuson and Bowser 1990, Franklin et al. 1990).

Our research expands upon the traditional boundaries of ecosystem studies to encompass multiple temporal and spatial scales. We consider a suite of adjacent lakes that share a common climate but differ dramatically in their physical, chemical, and biological characteristics. We employ a long-term perspective that permits us to place analyses of seasonal and annual patterns into the broader context of year-to-year variability and to evaluate the implications of such variability for community and ecosystem processes. We use a nested series of spatial scales ranging from within individual lakes to the entire Northern Highland Lake District (Appendix Fig. 1, p. 68). This permits us to consider how processes occurring in a lake are related to factors in adjacent systems and in the general landscape that surrounds it. Our broad temporal and spatial scales lend themselves to two expansions that we propose here; evaluations of the potential effects of global change and assessments of how processes discerned on smaller spatial scales are operating within the region in general.

Lakes provide ideal systems for long-term ecological research. Their boundaries are relatively distinct, and adjacent lakes, although sharing a common setting, may differ greatly in their fundamental properties. Thus, comparisons of lakes can be used to isolate at least partially important control factors for lake processes. Their distinct boundaries make lakes particularly useful for analyses of landscape-scale patterns. Likewise, many in-lake processes operate on shorter time scales than their analogs in terrestrial habitats (e.g., population growth and succession). This facilitates the observation of repeated sequences of processes within a few years as well as experimental manipulations. These advantages are coupled with a long tradition of limnological work which has laid a groundwork for our research. A substantial amount of this work was conducted on our LTER lakes by Birge, Juday and their colleagues in the first half of this century (Frey 1963, Beckel 1987).

In addition, because lake ecosystems superficially appear quite different from the principal ecosystems at the other, predominantly terrestrial LTER sites, the inclusion of the NTL site within the LTER network fosters the development of ecological theory at a more general level. Concepts that apply to both aquatic and terrestrial ecosystems must inherently have a broader applicability than those generated for a more restricted set of systems.

Within the synthetic goal of understanding the ecological complexity generated by multiple processes acting over many temporal and spatial scales, we aim to develop a series of broad-scale evaluations of factors controlling lake processes. These evaluations are interrelated but can also be considered independently. They can be classed generally into five major objectives:

• A. To perceive long-term trends in physical, chemical, and biological properties of lake ecosystems
• B. To understand the dynamics of internal and external processes affecting lake ecosystems
• C. To analyze the temporal responses of lake ecosystems to disturbance and stress
• D. To evaluate the interaction between spatial heterogeneity and temporal variability of lake ecosystems
• E. To expand our understanding of lake-ecosystem properties to a broader, regional context.

We elaborate on these objectives separately in the next five sections of the proposal. The first four objectives have played a key role in our research over the past five years. In discussing them, we highlight some of our major results and then outline the continuations that we propose. The last objective is new and reflects an expanded interest in seeking generality at both the site and the intersite level. Conceptual extensions within the first four objectives include 1) a focus on global climate change, 2) expanded evaluations of microbial processes and 3) more detailed assessments of land-water interactions.

II. SECTION 1 - RESEARCH PROGRESS AND PROPOSED RESEARCH

The following section describes our past progress and proposed research organized according to the five major research objectives listed above.

A. PERCEPTION OF LONG-TERM TRENDS

1. THE NORTH TEMPERATE LAKES DATA SYSTEM

Introduction. Our goal is and has been to develop a set of ecological measurements that will allow investigators quantitatively to observe and analyze patterns of long-term change in the physical, chemical, and biological features of lake ecosystems. We also wanted to make available the rich base of historic data collected earlier by E.A. Birge, C. Juday, and colleagues (Frey 1963). Finally, we wanted to establish an effective data management system to make the modern and historic data easily available to researchers.

Our choices of lakes and measurements were and continue to be guided by a desire to address important ecological and natural resource questions about lakes in respect to long-term (Likens 1983, LeCren 1984) and landscape level (Naveh and Lieberman 1984, Risser et al. 1984, Turner 1989) phenomena. These long-term measurements at various ecological levels and several temporal and spatial scales should be able to capture the essential structure and function of lake ecosystems and enable analyses of interactions among the principal ecosystem components.

Results. We have focused on a suite of lakes and surrounding terrestrial areas linked through a common groundwater and surface water flow system that share common climatic, edaphic, and biogeographic features (Fig. 1; Appendix Fig. 2, p. 69). The lake set includes oligotrophic, dystrophic, and mesotrophic lakes (Appendix Fig. 3, p. 70; Appendix Fig. 4, p. 71). Our seven primary lakes, located within 5.3 km of the Trout Lake Station in north central Wisconsin, were chosen to represent marked differences in size, morphometry and habitat diversity, in thermal and chemical features, in species richness and assemblies, and in biological productivity (Table 1, p. 3). The choice of primary lakes makes groundwater one major focus of our project, because of its importance in regulating differences in the chemical composition of lakes and in linking terrestrial and lake ecosystems (Likens et al. 1977, Winter 1978, Frape et al. 1984, Crowe and Schwartz 1981a, b).

We also have secondary lakes for which less complete data are collected. The choice of secondary lakes and types of measurements may change with time, but these lakes are studied with long-term research goals in mind. They serve for comparison with the primary lakes on specific research questions of individual investigators. For example Lake Mary, a meromictic lake, has the most temporally stable deep water system and is useful to compare with the other lakes which are influenced more by year-to-year changes in weather. Presently, our secondary lakes include: Clear, Escanaba, Fallison, Firefly (Weber), Little Rock, Mary, Mendota, Mystery, Nebish, and Pallette. Lake Mendota, our only study lake located outside the Northern Highland area, is included because the extensive and historical data available for Mendota are extremely valuable, one example being the use of ice cover duration to analyze climatic patterns (Robertson 1989, Magnuson 1990).

We spent considerable effort designing and implementing a balanced and integrated data collection program (Kratz et al. 1986). Our sampling (Appendix Table 2, p.83) allows comparisons of parameters among seasons, years, and lakes. We sample most major physical, chemical and biological parameters. On each lake we established a central station where related parameters are measured concurrently.

Sampling frequency is tuned to the dynamics of individual parameters. We sample most intensively at four key times of the year: spring overturn, maximum stratification in summer, fall overturn, and winter stratification. Chemically, these periods are important because differences between spring and fall overturns indicate a net gain or removal of chemical species from the water column. At periods of maximum stratification, conditions are most different from mixis, and depletion of epilimnetic nutrients and hypolimnetic oxygen can cause severe stresses on biological components. Complete cation-anion balances are computed during these four periods. Nutrients, pH, inorganic and organic carbon are sampled monthly. Temperature, dissolved oxygen, chlorophyll a, primary productivity, and zooplankton abundance are measured every two weeks during the open water season and every 5 weeks under ice cover. Parameters that vary over longer time scales are measured annually in August. These include macrophyte distribution, fishes (abundance, biomass, and community structure) and benthic invertebrate abundance. Typical sampling sites, from two of our LTER lakes, are shown in Fig. 2. In addition, groundwater levels in selected wells are measured monthly and groundwater chemistry from these wells is measured quarterly.

We maintain an automated land-based weather station 10 km from the Trout Lake Station. Parameters measured include air and soil temperature; precipitation; longwave, shortwave, and photosynthetically active radiation; wind speed and direction; and relative humidity. Our raft-based station on Sparkling Lake records air and water temperature, wind speed at three elevations, and relative humidity.

On occasion we have modified or developed measurement techniques. For example, we have developed state-of-the-art remote sensing for making acoustic estimates of distribution, abundance, and body sizes of pelagic fishes in conjunction with research sponsored by the Office of Naval Research (Clay 1983, Rudstam et al. 1987, Jacobson et al. 1989).

Proposed Research. We propose to continue the base sampling and data collection (Appendix Table 2, p. 83) started in the first 10 years (Magnuson et al. 1984, Kratz et al. 1986).

2. LONG-TERM DATA AND USE OF HISTORIC DATA

Introduction. We have access to three kinds of "long-term" data in our LTER project: data preserved in historical documents, data preserved in lake sediments, and data collected by the LTER program since 1981. Here, we provide examples of the types of analyses of long-term data that we have done using all three of these sources.

Comparing Recent Data with Historic Measurements. We used historic data spanning 1852 to present to analyze the link between ice cover of Lake Mendota and interannual variability in weather conditions including El Nino and climate change events. Unusual climatic conditions in areas adjacent to the equatorial Pacific Ocean have been shown to be directly related to El Nino events. Our goal was to determine whether El Nino events have influenced the climate of a more distant location, Wisconsin. We analyzed meteorological parameters and the long-term ice records with respect to the Southern Oscillation Index (SOI), a monthly correlate of El Nino intensity (Robertson 1989). Fourier Analysis, performed on the coinciding time series, demonstrated a relationship between ice duration on Lake Mendota and the SOI. After 1940, El Nino events are associated with consistent climatic anomalies (Fig. 3, p.6), warmer than normal December and March air temperatures with less than normal snowfall, resulting in late ice formation, early thaw and shorter than normal ice duration. Prior to 1940, El Nino events were associated with more variable climatic conditions. The ice conditions of Shell Lake, located in northern Wisconsin, were unrelated to El Nino events, possibly because Lake Superior's influence on local climate concealed signals from El Nino events.

We also related changes in lake temperature and ice cover to changes in mean air temperature associated with known and predicted changes in climate (Fig. 4). A step change in the duration of ice cover coincided with the end of the little ice age. We also forecast that if a 2XCO2 condition develops in the next 50 years due to Greenhouse Warming, Lake Mendota will be ice free 1 out of every 30 years by 2050 to 2060 (Robertson 1989).

Using the Sediment Record. We investigated the sedimentary record of pigments in short cores taken from Crystal, Sparkling, and Trout Lakes. Although diagenesis of pigments within the water column and sediments is significant, we were able to detect non-diagenetic trends (Hurley and Armstrong 1990b). Changes in the proportion of phaeophorbide suggest a period of reduced grazing starting about 1930. The concentration of diatoxanthin, a pigment associated with diatoms, increased sharply between 1900 and 1935 (Fig. 5) as did total phorbin and carotenoid pigment concentrations. This period corresponds to a time of clearcutting and may reflect response of Trout Lake to increased nutrient loading.

We discovered a relationship between a lake's dissolved silica concentration and the width of spicules growing in live sponges. Because sponge spicules are siliceous and well preserved in lake sediments we were able to use this relationship to reconstruct past silica concentrations in several lakes. We obtained complete sediment cores down to glacial material in three clearwater and six dystrophic lakes. In each lake there is evidence of a reduction in silica concentration over the holocene. Results from two dystrophic lakes are shown in Fig. 6. Groundwater is the major external source of silica to lakes in the region. The major sink is permanent burial in the sediments in the form of diatom frustules. Trends in silica suggest long-term changes in either groundwater inputs or internal cycling of silica, or both (p. 13 ).

Data Collected by the LTER Project. The LTER project has now collected nine years of data for many parameters. Here, we show three of the many examples of interesting patterns in this long term data set. In two of the examples, chloride concentration in Sparkling Lake (Fig. 7) and water level from three lakes (Fig. 8, p. 9) we have a reasonably detailed understanding of the underlying causes of the patterns. The chloride increase is associated with road salt entering the groundwater and entering Sparkling Lake (p. 15). The decrease in water level associated with lower than normal precipitation in the past few years is a function of the hydrologic setting of the lake and gives an example of how different lakes respond to the same climatic forcing. The third example, a decrease in silica concentration in Sparkling Lake (Fig. 9, p. 9), is an example of a trend that we do not understand sufficiently. Trends such as these, discovered serendipitously, are among the most intriguing and scientifically exciting.

Proposed Research. We will continue using historical archives, the sediment record, and the growing time series of LTER data to stimulate new questions and to ask and answer the questions developed in our major objectives. These data will be the basis for answering many of the research questions described in the next four major sections.

B. DYNAMICS OF EXTERNAL AND INTERNAL PROCESSES

Long-term patterns in lakes are generated by a complex interplay between external and internal processes. Evaluations of long-term patterns must be linked with an understanding of these processes. Because internal and external processes operate on a variety of time scales, interactions among them lead to complex patterns, often with time lags between cause and event. Understanding how lakes are affected by the interaction of process occurring on different time scales, particularly when there are time lags, is a major goal of our LTER program.

External processes influencing lakes include climatic and hydrologic factors combined with land-water interactions. Climate exhibits substantial year-to-year variation with potentially major effects on lake conditions (Madden 1977). Climate also exhibits distinct long-term cyclic behavior (Jones et al. 1982). Hydrologic differences among the NTL-LTER lakes are associated primarily with differences in groundwater inputs. These differences exert important influences on both water and chemical budgets. Variability in the groundwater system operates over a longer time frame than other components of the hydrologic budget such as precipitation and surface runoff. Land-water interactions involve linkages between terrestrial and aquatic systems in which numerous processes operate at fundamentally different time scales (contrast the growth and decomposition of trees versus phytoplankton).

Internal processes involve a diversity of physical, chemical, and biological factors. Again, the interplay of processes acting over different time scales leads to complex behavior. For example, a lake's primary production during mid-summer could be controlled directly by the extent of turnover during the previous spring and by the amount of grazing exhibited by herbivorous zooplankton during that period. Each of these processes in turn can be linked with longer-term factors such external nutrient loading and the year-class strengths of zooplanktivorous fishes. Consequently, primary production in one summer may be partly determined not only by present conditions, but also by year class formation of fish three years prior and the length of the turnover period in the previous spring.

A major goal of our LTER program is to develop a basic understanding of the roles of external and internal processes in influencing lake ecosystem conditions. Our primary efforts in this area include: 1) identification of major linkages between specific external or internal processes and limnological conditions, 2) evaluation of the interactions that occur among major external and internal processes, and 3) assessment of how long-term variability in specific processes influences long-term patterns in lake ecosystem parameters. Below we highlight separately some of our major research efforts on external and internal lake processes. This separation is not absolute but rather one of emphasis. Our goal has been to integrate both external and internal processes to generate a basic understanding of lake ecosystem processes.

1. External Processes

Our investigations of external lake processes have emphasized considerations of climate, groundwater hydrology, and land-water interactions.

a. Climate

Introduction. Climatic forcing controls the thermal environment of north temperate lakes by determining the timing of freeze and thaw, the length of the growing season, the depth of the mixed layer, the intensity of stratification, and the temperature of the various water layers. Thermal conditions form the setting for a lake's chemical and biological processes, many of which have rates dependent on water temperature. The LTER database is ideal for investigating the role of climatic variability in generating long-term trends in limnological processes (Robertson 1989) and in exploring relationships between climatic and non-climatic factors in lakes (Magnuson et al. 1990d). The threat of drastic climate change in the next century increases our incentive to learn about how lakes respond to climatic forcing. We will use the LTER database to develop and verify models of thermal, chemical, and biological conditions in the lakes relevant to global climate change.

Results. One emphasis of our climate research has focused on the effects of climatic factors on thermal conditions within lakes. Robertson (1989) used a long time series of ice cover records for Lake Mendota to reveal climatic variation on several time scales (Fig. 3, p. 6; Fig. 4, p. 7).

Robertson (1989) also demonstrated the utility of statistical and functional models in predicting thermal properties of lakes based on climatic variables. An empirical model employing meteorological conditions and water temperature successfully predicted the times of ice-on and ice-off for Lake Mendota. Climatic variables likewise showed the highest degree of temporal coherence in long-term comparisons across the LTER primary study lakes (Magnuson et al. 1990d).

Proposed Research. One of our long-range goals is to predict changes in limnological processes due to global climate change. This involves a three step process in which we 1) develop models of the influence of climate on thermal conditions within lakes, 2) predict the impacts of warming scenarios on lake conditions, and 3) examine the effects of predicted thermal conditions on internal lake processes.

What is the relation between climatic factors and thermal conditions in lakes? We will explore the relationship between climatic forcing and thermal response of lakes using functional and statistical models (Magnuson et al. 1990c). Modeling the annual thermal structure of the lakes requires predictions of the timing of ice formation and breakup, evaluation of heat exchange between the lake and atmosphere, and analysis of internal heat redistribution processes. For statistical models, we will employ the methods used to predict ice cover on the Lake Mendota (Robertson 1989) to develop similar models for other LTER lakes, using the historical meteorological database and ice cover records. We will also expand on existing functional models (e.g., Imberger and Patterson 1981) to predict the development of thermal conditions within lakes during the ice-free season. These models include explicit formulations of heat exchange at the air-water interface and between layers within the lake and predict thermocline depth and seasonal changes in epilimnetic and hypolimnetic temperatures. We intend to implement such models for several NTL LTER lakes.

How will thermal structure of lakes change following global climate warming? Using the models described above, we can create scenarios of the thermal structure of our lakes following global climate change using scenarios output from GCM's (e.g., Manabe and Stouffer 1980, Hansen et al. 1988). Transient climate warming scenarios are available representing the evolution of the global climate from 1XCO2 to 2XCO2 conditions. We will use these transient warming scenarios to predict the timing of thermal changes in our lakes.

How will lake processes respond to global climate warming? The chemical and biological responses of lakes to warming scenarios can be evaluated using models with varying levels of sophistication. Some models examine the influence of temperature on mass transport and kinetic interactions of biota and nutrients (Park et al. 1974, Blumberg and DiToro 1990). A relatively simple version of such a model might consider the dynamics of chlorophyll-a, carbon associated with zooplankton, phosphorus, nitrogen, silica, and dissolved oxygen. Such a model would be lake-specific and could be calibrated with LTER lake data. More detailed models could include the effects of individual species responses to changing thermal conditions. Several models incorporating these factors have been generated for lakes similar to those at our LTER site (Carpenter and Kitchell 1987).

The effects of global climate warming can be expected to be particularly strong on fish assemblages in north temperate lakes. The occurrence of fish assemblages in LTER lakes has been directly linked to the availability of suitable temperatures (McLain and Magnuson 1988). Bioenergetics models (Kitchell et al. 1977) can be used to produce scenarios of fish growth following climate warming (Hill and Magnuson 1990). Thermal niche models (Magnuson et al. 1990e) can provide estimates of changes in water volumes with temperatures suitable for different thermal guilds of fish (Magnuson et al. 1979). We will use changes in thermal niche space to forecast potential species extinctions or invasions in the lakes.

b. Hydrology

Introduction. Lakes within the NTL-LTER site vary markedly in the contributions of precipitation, groundwater, and surface water to the input portion of their hydrologic budgets. A strong gradient exists among our primary LTER study lakes that reflects increasing proportions of groundwater inputs (Okwueze 1983). This gradient has served as a major organizing theme in our program. We have examined this gradient's influence by 1) quantifying groundwater inputs, 2) determining the effect of groundwater on lake conditions, and 3) examining the role of groundwater in the transport of solutes, particularly nutrients and contaminants. We have emphasized studies of individual lakes. Success at this level permits us to expand our proposed efforts to a more regional scale.

Results. Quantitative hydrologic studies were initiated on Crystal Lake, which represents one extreme in the groundwater input gradient, with a marked predominance of precipitation inputs (Kenoyer 1986). Groundwater accounted for only 5% of input water to Crystal Lake, yet, because groundwater is more concentrated chemically than precipitation, groundwater accounted for substantial amounts of incoming solutes. For example, about 50% of the silica budget comes into the lake via groundwater, exerting a major influence on primary production in Crystal Lake (Table 2, p. 13) (Hurley et al. 1985). Detailed evaluations of groundwater flow paths confirmed that silicate hydrolysis in the glacial till surrounding the LTER lakes led to increasing concentrations of minerals with increasing time of groundwater contact with till (Fig. 10, Kenoyer et al. in press). Groundwater inputs to lakes like Crystal have the potential to vary substantially from year to year and provide a mechanism to control year-to-year differences in other lake parameters such as primary production or chlorophyll.

Sparkling Lake, lying at an intermediate elevation and with a presumed, higher proportional contribution of groundwater to its hydrologic inputs, was the next focus for quantitative evaluation. The more complex flow regime for Sparkling Lake, with substantial groundwater inputs and outputs led to the development of new techniques for quantifying hydrologic budgets. Stable isotope analyses of oxygen and hydrogen were coupled with more traditional, well measurements to evaluate the importance of groundwater in the lake's hydrologic budget (Krabbenhoft et al. in press a, b). We discovered that local groundwater had an isotopic signature distinct from mean annual precipitation values (Fig. 11). These differences exist because most inputs to the groundwater flow system occur during winter when the isotopic composition of precipitation differs from other seasons. During summer, precipitation is transpired and groundwater recharge is minimal. Models are being developed to examine the relationship between lake isotopic composition and the proportional contribution of groundwater to hydrologic inputs for lakes in general (McGrath et al. 1988).

Bog lakes at our LTER site are isolated to a large extent from surrounding groundwater flow systems. However, our studies have revealed that bog lakes exhibit a similar pattern, across landscapes, to that exhibited by the LTER lakes in general. The proportional input of groundwater to the bogs' hydrologic inputs increases with decreasing elevation in the landscape (Fig. 12, p. 15; Kratz and Medland in press). More detailed studies of Crystal Bog reveal a complex interplay between conditions in the lake and in the surrounding wetland due to the groundwater flow regime and resulting chemical transfers (Marin et al. in press).

The groundwater flow system can serve as either a barrier or a conduit to the transport of anthropogenic contaminants. The rate at which acid deposition falling within a region is transported to a lake varies substantially with hydrologic regime (Anderson and Bowser 1986). Because of buffering during groundwater flow, only lakes with a predominance of precipitation input would be expected to respond to acid deposition. In contrast, another contaminant, road salt, is transported fairly efficiently (Krabbenhoft and Bowser ms). This pathway has lead to a substantial systematic increase in the chloride concentration of Sparkling Lake since 1982. Even in the case of road salt, however, groundwater flow does not serve as a simple conduit. Increases in Na, the predominant cation paired with Cl in road salt, do not match those for Cl indicating that soil and aquifer ion-exchange processes are impeding its transport.

Proposed Research. Our goal is to integrate hydrologic studies with overall evaluations of lake conditions. This will involve two primary areas: continuing evaluations of the hydrologic budgets of individual lakes and expanding our hydrologic perspective to include a regional analysis of our entire study site.

What are the contributions of groundwater, precipitation, and surface flow to hydrologic and chemical budgets of Trout Lake? Our analyses of hydrologic inputs have moved from simpler to more complex systems. Trout Lake, the lowest elevation lake in our landscape, is the next logical choice for detailed study. The significant contribution of surface flow into Trout Lake will require an expansion of the methods that we have used for seepage lakes to include measurements of surface water flow. We will undertake these efforts in cooperation with the Wisconsin District of the U.S. Geological Survey.

Can stable isotope techniques be used to evaluate hydrologic processes for lakes and watersheds? Our work on Sparkling Lake has indicated the utility of stable isotope techniques for individual lakes, but the general applicability of these methods to larger scale systems remains to be explored. The stable isotope signature of oxygen and hydrogen of a water body at a particular time reflects an interplay between hydrologic processes that alter the proportions of stable isotopes and those that do not cause fractionation. Stable isotopes ratios are affected by evaporation but not by flow processes or transpiration. The degree of isotopic fractionation during evaporation varies as a function of temperature, relative humidity and the magnitude of evaporation itself. The isotopic signature of precipitation, in turn, varies with season and among storm events reflecting, in part, differences in evaporative processes.

The development of a general methodology for the use of stable isotopes in evaluating lake hydrology will require the cross-calibration of isotopic techniques with other more traditional measures of hydrologic parameters. We are expanding our measurements of such parameters for our LTER study lakes through the use of a combination of land-based and floating meteorological stations and expanded well networks. These measurements will be combined with previous hydrologic studies conducted near our LTER site (e.g., Little Rock Lake).

Stable isotope analyses will also provide insight into watershed processes. Summer precipitation appears to be largely captured and transpired by terrestrial vegetation, whereas winter precipitation is recharged into the groundwater system. Thus, with proper calibration, isotopic contrasts between local groundwater and rain should provide an estimate of long-term rates of terrestrial transpiration. This work will play an important role in our regionalization efforts (p. 42) and in our assessments of effects of global climate change (p. 10).

How are lake chemical conditions influenced by variability in groundwater inputs? Our work on lake/groundwater interactions has focussed on the chemistry of major ions and silica. We will expand these efforts to include assessments of additional elements, particularly N and P. By combining chemical analyses of groundwater, streamflow, and precipitation with the quantitative measures of these flows described above, we will be able to make direct assessments of their relative contributions to lake chemical budgets for the full range of lakes in our flow system.

We also will expand our work on the transport of contaminants by groundwater. The trends in Sparkling Lake's chemistry provide an ideal opportunity to examine these processes. The seasonally pulsed nature of road salt use may provide a particularly useful marker for examining trends in groundwater flow and in the transport of contaminants. This work will involve more detailed monitoring of wells near Sparkling Lake coupled with groundwater modeling.

c. Interaction between Terrestrial and Aquatic Ecosystems

Introduction. The NTL-LTER site is a mosaic of terrestrial and aquatic ecosystems (Appendix Fig. 1, p. 68) where terrestrial systems have the potential to exert major influences on lake conditions. Such terrestrial influences operate directly through the input of allochthonous materials and indirectly through an influence on hydrologic inputs. Characteristics of terrestrial ecosystems surrounding lakes, and their potential influence on lakes, vary as a function of soil substrate and natural and/or human-mediated disturbance regimes. Therefore, landscape position and disturbance regimes of terrestrial ecosystems must be considered in evaluating the factors influencing long-term patterns in aquatic ecosystems.

Results. Landform and soils vary greatly within the NTL-LTER study site reflecting past glacial history. Soils range from coarse sands with low water holding capacity and nutrient availability to fine loams with moderate to high water and nutrient availability (Appendix Fig. 5, p. 72). Several studies have demonstrated the relationship between soil characteristics and vegetation in the region (Curtis 1959, Kotar et al. 1988). Studies in the Pacific Northwest have demonstrated strong correlations between soil water availability and vegetational characteristics such as leaf area index (Grier and Running 1977, Gholz 1982) and net above-ground primary production (Gholz 1982). Vegetation is further influenced by disturbance and management regimes; these factors and their interaction have led to substantial differences in the vegetation surrounding the primary LTER study lakes (Appendix Fig. 5, p. 72) with important consequences for land-water interactions.

Allochthonous inputs add significant but varied quantities of carbon and nutrients to our study lakes (Appendix Fig. 6, p. 73). Perry and co-workers estimated that annual leaf litter inputs range from 400-1720 g/m of shoreline; annual phosphorus inputs associated with the fine litter ranged from 300-1470 mg/m of shoreline. The importance of leaf inputs varies among the LTER lakes owing to differences in landscape position, forest species composition along shorelines, and shoreline length relative to lake volume. Decay rates for allochthonous inputs also varied among lakes and with the species of source material.

Vegetational characteristics of the NTL-LTER site have undergone substantial changes over the past 125 years (Table 3). The pine forests that predominated in the region during presettlement times have been largely superseded by northern hardwoods (Morrison and Ribanszky 1989). Differences in vegetation also have the potential to influence the quantity and quality of groundwater flow within a region (as discussed above). These vegetation shifts have the potential to influence present day lake conditions with a variety of time lags. Shifts due to differences in the nature of allochthonous materials entering the lakes would occur within a few years for most materials. In contrast, there are substantial lags between the time water enters the groundwater and when it reaches a lake (Anderson and Bowser 1986).

Proposed Research. Expanded research here will focus on more detailed measures of the interaction between aquatic and terrestrial systems.

How does vegetation affect the hydrologic regimes of lakes? The absolute quantity of hydrologic inputs to lakes and the chemical characteristics of those inputs are influenced by the characteristics of terrestrial systems that surround them. For example, Swank and Douglas (1974) demonstrated that streamflow is greatly reduced by converting deciduous hardwood forests to pine due to the greater annual water losses via evapotranspiration in pine than deciduous hardwood forests. Borman and Likens (1979) observed a large increase in streamflow from a watershed that was harvested and noted that annual solute content in the stream was positively correlated to annual streamflow volume. We will develop a general model of the interplay between landform, vegetation, and hydrologic conditions for the NTL-LTER region.

Vegetational characteristics of the region will be determined through the use of remote sensing and GIS technologies (Appendix Fig. 5, p. 72). Preliminary studies suggest that the Landsat Thematic Mapper T can provide accurate hardwood vs. softwood and upland vs. lowland forest type separation (Hopkins et al. 1988, Morrison and Ribanszky 1989). Researchers in the Environmental Remote Sensing Center will continue to develop systems for analyzing TM data using GIS technologies.

Leaf Area Index has proven to be a useful ecosystem parameter for comparing energy, mass, and water exchange across diverse ecosystems (Running et al. 1986, 1989; Running and Nemani 1988, Running and Coughlan 1988), and can be quantified by remote sensing, in some systems (Running et al. 1986, Peterson et al. 1987). Leaf Area Index will be combined with information on environmental factors in an ecosystem process model (FOREST-BGC, Running and Coughlan 1988) to estimate major water fluxes in terrestrial systems. This model was developed for coniferous forests and it will be modified for use with other forest types present at our site. Forest models will be combined with ongoing projects, funded separately by NSF

(to Gower et al.) to examine above- and below-ground carbon balances for terrestrial ecosystems.

What are the relative effects of allochthonous inputs of coarse woody and fine litter material on dynamics of lakes? Recent studies have demonstrated that coarse woody detritus can equal or exceed leaf litter input in some forest ecosystems (Vogt et al. 1986, Harmon et al. 1986). We propose to establish long-term plots along the shoreline of selected LTER lakes and bogs to estimate coarse woody detritus production. In addition, we plan to initiate a long-term woody decomposition study to estimate carbon and nutrient immobilization-mineralization patterns of woody material placed in lakes and bogs. This information will be combined with the studies on carbon dynamics described above and with the fine-litter detritus study by Perry.

How do long-term changes in terrestrial vegetation influence lakes? We will extend previous analyses of long-term changes in terrestrial vegetation for the regions adjacent to the seven LTER lakes and develop a more extensive analysis for northern Wisconsin. We will rely on historic vegetation surveys (Finley 1951) to recreate past vegetation patterns. Present conditions will be assessed using remote sensing and GIS technologies in conjunction with the modeling efforts discussed above. These data will be combined with our process oriented studies of terrestrial interaction with lakes (described above) to assess the importance of shifting landscape conditions on lakes.

2. Internal Lake Processes

Evaluations of internal processes emphasize controls over the phytoplankton, in terms of chlorophyll abundance and primary production, and the abundance of aquatic species in general.

a. Seasonal Chlorophyll Dynamics

Introduction. Chlorophyll density is a conspicuous lake feature controlled by an interplay of external and internal factors. At a given time, the standing biomass of chlorophyll within a lake results from an interaction of growth and loss processes. Differences in both growth rates, sometimes termed "bottom-up" control (Dillon and Rigler 1974, Schindler 1978, and McQueen et al. 1986), and loss rates, termed "top-down" control (Edmondson and Litt 1982, Shapiro and Wright 1984, Carpenter et al. 1985, 1987) have been shown to control chlorophyll biomass under some circumstances. It is unclear how the relative importance of these factors vary under natural conditions (Crowder et al. 1988), but models considering both bottom up and top down controls perform better that those focusing on either process alone (Carpenter et al. in press).

Results. Three LTER study lakes exhibit fundamental long-term differences in their seasonal chlorophyll dynamics. In Trout Lake, for every year between 1982 and 1989, spring and fall chlorophyll maxima are routinely punctuated by summer and winter minima (Fig. 13) in a pattern that is characteristic of many temperate-zone lakes (Hutchinson 1967, Marshall and Peters 1989). Two other lakes diverge from this pattern with reduced spring peaks in Allequash Lake in most years and with reduced fall peaks in Sparkling lakes in many years (Fig. 13). Can these differences, particularly the absence of a peak in some seasons, be attributed to a predominance of differences in growth or loss processes? Information on nutrients and zooplankton biomass indicate that they can not.

Reduced availabilities of nutrients in some seasons or years, owing to differences in mixing events or loading, would be a likely mechanism to lead to reduced chlorophyll levels through shifts in growth rates. Data on nitrate, used here as a general marker of the input of nutrients, indicates substantial year-to-year differences in peak concentrations in lakes and seasons (Fig. 14) but these differences are not correlated with peak chlorophyll concentrations. Shifts in the grazing pressure exerted by herbivorous zooplankton could also reduce chlorophyll levels through shifts in loss. Zooplankton biomass varies markedly across seasons and years in both Trout Lake and Sparkling Lake (Fig. 15, p. 23) but here too there is no correlation between chlorophyll and zooplankton.

Proposed Research Can inter-lake and interannual differences in peak chlorophyll values be associated with shifts in growth processes or loss processes?

Our approach involves combining our continued, long-term measurements of lake chlorophyll, nutrients, and zooplankton with a series of more detailed measurements of the nutrient and zooplankton dynamics taken during periods of increasing, maximum, and declining chlorophyll concentrations. We will focus on the epilimnion as a reasonable unit within which to develop a budget of the net interplay of growth and loss processes. The importance of nutrients will be assessed through experimental manipulations of nutrient availability along with direct assays for N and P stress. The influence of zooplankton will be evaluated directly using microcosms with various levels of grazing pressure (Lehman and Sandgren 1985). The role of mixing phenomena in controlling the availability of nutrients will be assessed through continuous monitoring of thermal profiles of our study systems. Funding to support our maximum expansion in this area has been requested in a separate proposal to the NSF Ecology Program (p. 118). If we fail to obtain that funding, we will conduct a reduced series of more detailed experiments and measurements than have been routinely included in our LTER sampling. b. Controls on Primary Production

Introduction. As in the case of chlorophyll biomass, the factors controlling a lake's primary production are a fundamental concern for aquatic ecosystem ecologists. We have focused on a subset of three of our primary study lakes to examine factors influencing production. The lakes were chosen to provide generally contrasting limnological conditions.

Results. Based upon several factors including secchi depths, major differences in groundwater inflow and associated solute loadings (Kratz et al. 1986), we had anticipated substantial differences in primary production among Crystal, Trout, and Sparkling Lakes. Several years of data on average daily rates of depth-integrated primary production, however, revealed remarkable similarities among the three lakes (117, 93, and 86 g C/m2/year for Trout, Sparkling, and Crystal (Adams et al. in press)). We have been examining the factors that account for this similarity, including possible differences in light penetration, similarities in nutrient loading particularly to the photic zone, and differences in internal nutrient cycling. Sampling for primary production is shown in Appendix Fig. 7, p. 74.

Light penetration varies substantially among Crystal, Sparkling and Trout Lakes with corresponding differences in the depth of the water column over which primary production occurs (Fig. 16, p. 24). Proportionally more production occurs at greater depths in Crystal Lake compensating for higher levels in shallower regions of Sparkling and Trout Lakes. These differences are attributable not only to differences in seston among the lakes but also to major differences in dissolved organic carbon (DOC). This DOC absorbs substantial quantities of light and competes with phytoplankton for light. A simulation in which the DOC in Crystal was increased to the levels that occur in Trout resulted in a 30 % decrease in calculated production (Kratz and Meinke, in prep.). Thus, relatively small differences in DOC may affect light penetration substantially and play a major role in regulating production in lakes where nutrient levels are low and photic zone may extend below the thermocline. Phytoplankton vary in their quantum use efficiency across lakes, with depth and with season. We expect this variation in how phytoplankton use light to be related to variability in primary production across lakes.

Similarities in production may also be related to nutrient inputs. Although groundwater inflow and solute loadings increase in the order Trout >Sparkling>>Crystal, the loading of limiting nutrients may be similar among the lakes. Low phosphate concentrations and high C/P ratios in seston (Hurley 1984) indicate P limitation during much of the year in all three lakes. P levels in groundwaters are low, and atmospheric deposition may be a relatively important external source of P, resulting in relatively similar areal loadings of P among the three lakes.

Internal cycling may also result in similar limiting nutrient loadings to the photic zones of the three lakes. Significant differences have been observed in the cycling of C, N, and P among the lakes. For carbon, sediment trap and bottom sediment accumulation rate measurements (Table 4, Hurley 1984 and NTL-LTER database) indicate high (> 50%) but similar decomposition rates in the sediments. Total water column respiration rates appear to be quite similar, but substantial differences in zooplankton biomass among lakes suggest that the proportion of carbon respired by zooplankton is much lower in Crystal than in Trout or Sparkling. This suggests that microbial respiration must be substantially higher in Crystal. These evaluations are only approximate and direct assessments of microbial respiration and production are planned to further ascertain the fate of produced carbon.

Internal cycling of P and N contrasts with that of C and exhibits substantial differences among Crystal, Sparkling and Trout Lakes. Low C/P ratios in bottom sediments as compared to seston reflect selective chemical immobilization of P in sediments (Table 5 ). C/P ratios higher than a Redfield proportion in Crystal Lake but substantially lower in Sparkling and Trout indicate that P is recycled much more efficiently from bottom sediments in Crystal Lake (Hurley 1984). Within the NTL-LTER region, the recycling of inorganic P from sediments has been linked with an interplay between Fe concentrations and the extent of anoxic conditions (Williams et al. 1971). These factors appear to be operating to generate the differences observed in P cycling within the LTER lakes. Evaluations of C/N ratios indicate that N cycling shows a similar pattern to that of P with a greater recycling of N in Crystal and Sparkling Lakes than in Trout.

Proposed Research. Our results to date indicate the influence of several factors in controlling the production of our study lakes. Evaluations of the interplay of these factors remain to be completed.

What is the role of nutrient loading to the photic zone in controlling primary production among lakes differing in groundwater inflow? With a database of several years on primary production and seston deposition from the water column combined with data on groundwater inflow, atmospheric deposition, and sedimentation rates, we are in position to evaluate the budgets of C, N, and P to the photic zone. We plan to develop ecosystem level models to evaluate these budgets for Crystal, Sparkling, and Trout Lakes. Several questions will be addressed: What are the relative nutrients fluxes from groundwater, atmospheric deposition, and internal recycling? Are year-to-year variations in fluxes appreciable? Is the flux of phosphorus limiting? Is immobilization of phosphorus in sediments a major factor regulating the phosphorus supply? What is the relative importance of regeneration within the water column (zooplankton, bacteria) and at the sediment-water interface (bacterial)?

To what extent does the proportion of primary production processed by zooplankton and the microbial food web vary among lakes? Estimates to date suggest that the relative proportion of carbon processed by microbes and zooplankton varies markedly across our study lakes. Our assessments have been limited, however, by a lack of direct measures of microbial processing. We propose to make direct determinations of microbial production and couple them with more detailed assessments of zooplankton grazing. This work will also be coupled with the nutrient budgets discussed above to evaluate the influence of microbes and zooplankton on primary production.

Bacterial production and respiration will be measured directly. Ideally, this work will be done in collaboration with Cole and Pace and/or with Dodson and Graham; both groups have grants pending at NSF to support this work. If their proposals are unsuccessful, more limited assessments will be made within our LTER program. Assessments of zooplankton effects will be expanded through the use of more detailed calculations based on species and size abundance data and size/species-specific respiration rates (Downing and Rigler 1984) and grazing rates (Sierszen and Frost In press), and through measurements of the proportion of chlorophyll a converted to phaeophorbide and collected in sedimentation traps (Hurley and Armstrong 1990a, 1990b). These data will be combined with carbon budgets of the water column, and fluxes of C into and accumulating in sediments to estimate bacterial respiration in sediments.

What is the role of P immobilization by Fe in limiting production in the NTL region? Our data suggest that P immobilization in sediments is strongly influenced by Fe, however, information on the role of Fe-P interactions relative to other factors limiting production is scarce. Measurements of the sedimentary component of C, N, P budgets combined with analyses of sedimentary inorganic P will be used to assess selective immobilization of P and its influence on regeneration of P relative to N (Armstrong et al. 1987). These evaluations will be expanded to include other lakes in the region with a wider range of Fe fluxes and sedimentary accumulation rates.

The influence of Fe on P release by sediments is strongly influenced by sediment anoxia. The potential for influences on the photic zone and primary production are high when P is released into the epilimnion or metalimnion. Thus, the sediment contact area in the metalimnetic zone plays an important role in seasonal internal cycling of P (Stauffer and Armstrong 1984). We will also investigate whether morphometry-related sedimentary P recycling plays an important role in the NTL region.

c. Shifts in Species Abundance

Introduction. Understanding the factors that control the presence or absence of species within an ecosystem is a fundamental goal for both ecological and evolutionary studies, particularly at a time when there is a major interest in the factors controlling the overall diversity of organisms. Moreover, studies of aquatic systems have revealed numerous cases where the abundance of particular species exerts a major influence on both community and ecosystem processes. Paine (1980) and Brooks and Dodson (1965), provided classic examples of fundamental shifts in community structure induced by single species and presaged numerous examples of the importance of such species driven processes (Stein et al. 1988). Hrbacek et al. (1961), Shapiro and Wright (1984), and Carpenter et al. (1987) have demonstrated that the effects of such species shifts can extend beyond community composition to exert a major influence over ecosystem processes.

Results. Within the LTER primary study lakes, our analyses of species abundance patterns have revealed that major species shifts are common. We have used these shifting abundance patterns, coupled with the comparative network provided by our lakes, to explore the factors controlling long-term variability in population levels. In addition, we have used these shifts to examine the impact of such changes on ecosystem processes and the importance of time lags that occur between events controlling a particular taxon and their subsequent effects on ecosystem processes.

Introductions of exotic species (e.g., rainbow smelt and rusty crayfish) into some of our LTER lakes provide, perhaps, the most dramatic cases of shifting species abundances (see Disturbance Section p. 30). In addition to these invasions, we have documented numerous instances where fish and zooplankton species have undergone shifts of greater than two orders of magnitude in abundance during a two to three year period. Crystal Lake provides a prime example. Here strong year classes of yellow perch in some years contrast with a nearly complete absence in other years. In other LTER lakes, other fishes, particularly zooplanktivorous cisco have undergone dramatic population shifts (Fig. 17). Similar patterns occur for zooplankton and benthic populations.

Situations where population levels are substantially different among years allow examination of factors controlling populations, particularly when the comparisons can be expanded to include two or more lakes. Shifts in cisco populations (Fig. 17) show the value of this approach. Autecological studies revealed that cisco require cold, oxygenated waters (Rudstam and Magnuson 1985). Such habitat conditions were met in Trout and Sparkling Lakes but were unavailable in Big Muskellunge Lake during 1982 (Fig. 18). Cisco were lost from Big Muskellunge Lake during this period but persisted in Trout and Sparkling Lakes. Thus cisco were excluded by physical and chemical conditions (McLain and Magnuson 1988). In a similar type of study, direct experimental analyses (Gonzalez 1988) showed that seasonal declines in rotifer populations were controlled by differing levels of food limitation. Broader comparative analyses have revealed the influence of varying silica levels on freshwater sponges (Frost and Elias in press) and the capacity of an animal-capturing plant to vary its investment in carnivory in direct response to differences in habitat conditions (Knight and Frost submitted).

Observations of Crystal Lake provide a strong example of the link between species shifts and ecosystem parameters. Adult yellow perch are zooplanktivores when they occur in a lake's pelagic zone. During years in which adult perch are abundant in the pelagic zone of Crystal Lake, the average concentration of the lakes dominant herbivore, a calanoid copepod, is markedly reduced (Fig. 19). Correlated with this reduction is a substantial decrease in the lake's water clarity attributable to a change in the standing biomass of primary producers. These shifts in water clarity appear to be influenced, at least in part, by events initiated two and three years prior to their impact (Magnuson 1990, Magnuson et al. 1990b). The abundance of adult yellow perch is linked directly to the success of year class recruitment, which is controlled by weather conditions during the first year of life. Thus, the shifts in water clarity observed in the lake are linked to external, climatic processes that operate with a two to three year time lag.

Proposed Research. Ultimately, a major goal of our research is to link separate studies of external and internal processes in a detailed evaluation of lake conditions. Shifts in species abundance with subsequent impacts on trophic interactions provide an organizing theme around which to build such a synthesis.

What is the relative role of external and internal factors in driving year-to-year differences in water clarity of Crystal Lake? As discussed above, we have observed substantial year-to-year differences in the secchi depth of Crystal Lake (Fig. 19). We have also identified two distinctly different mechanisms that could impart such year-to-year differences. Year classes of yellow perch can exert very different grazing pressures on the lake in different years. Likewise, year-to-year differences in groundwater inputs also have the potential to fuel substantially different levels of

primary production. We propose to integrate our measures of trophic and chemical effects on lake processes using similar models to those discussed in the Primary Production Section above. In particular we will contrast annual differences in top down control, as driven by fish, and bottom up control, as driven by differences in the quantity and quality of groundwater inputs.

C. RESPONSES TO DISTURBANCE AND STRESS

Perturbations affecting ecosystem processes operate on a broad range of time scales. On one extreme, some events, often termed disturbances (Pickett and White 1986), take place quickly but have effects that are longer lasting. In contrast, the effects of other perturbations, classified as stress, take place over an extended period. Stress and disturbance can be considered as two ends of a spectrum in which the time over which impact occurs varies. Understanding the role of both slow- and fast-acting perturbations is a fundamental goal in ecological studies (Connell 1978, Bender et al. 1984, Pickett and White 1986, Turner 1987). Our efforts at the NTL-LTER site have emphasized slower-acting perturbations, particularly acid deposition and species invasions. Our work on disturbance has considered the role of turnover events as disturbances in plankton communities and we are continuing these efforts in conjunction with our work on chlorophyll dynamics discussed above. Using a series of temporary ponds at the NTL-LTER site, we also have tested explicitly the hypothesis that the importance of biotic interactions in determining community structure increases with the extent of time between disturbance events.

1. Biological Invasions

Introduction. Lakes are island-like ecosystems (Fig. 20), isolated to varying extents from adjacent systems (Barbour and Brown 1974, Magnuson 1976). Natural immigrations of aquatic organisms are rare events, particularly where connections are limited among lakes. Such is the case at our site where five of our seven primary lakes are not connected by streams. Human-influenced species introductions, however, are generally much more common. To understand how species invasions affect lakes we need to answer two basic questions: 1) What factors control the probability of invasions into lakes? and 2) What are the effects of invasion on lake ecosystems?

The factors that control probability of the invasion of a habitat by a new species are poorly understood (Groves and Burdon 1986, Kornberg and Williamson 1986, Mooney and Drake 1986), but several studies suggest that perturbation leaves the ecosystem more vulnerable to invasion (Fox and Fox 1986, Orians 1986, Lawton and Brown 1986), through changes in resource availability or reductions in native species. We expect that climate change, a long-term perturbation, will affect the probability of species invasions in our lakes (Mandrak 1989).

The establishment of non-native species may have wide ranging effects (Simberloff 1981, Herbold and Moyle 1986, Moyle 1986, Vitousek 1986) . Such effects seem to be particularly likely in aquatic habitats where the substantial influence of trophic interactions is well demonstrated (Carpenter and Kitchell 1988).

Results. While there have been many species introductions within lakes at the NTL-LTER site over the past 80 years, primarily through fishery manipulations, three invading species stand out. Human activity has led to dispersal in our region of two crayfish species (Orconectes propinquus and O. rusticus) and one fish species, the rainbow smelt (Osmerus mordax). Of our primary study lakes, four have been invaded by O. propinquus (the earliest invader), two by O. rusticus, and two by smelt.

The effects of these invasions vary. We predicted that the invasion of smelt into Sparkling Lake in 1982 would result in the local extinction of cisco (Coregonus artedii), the native pelagic planktivore (Magnuson and Beckel 1985). Although extinction has not yet occurred, cisco are much less abundant and have suffered a near-total recruitment failure since smelt became abundant in the lake (Fig 21 top) (McLain and Magnuson 1988). In 1984 smelt invaded Crystal Lake, a small lake dominated by perch. So far there has been no significant change in perch abundance, and smelt have remained a small portion of the pelagic catch (Fig 21 bottom).

For the crayfishes, we predicted that the two invading species would become dominant sequentially, displacing the native species Orconectes virilis (Capelli 1982, Capelli and Magnuson 1983, Lodge et al. 1986). Typically the first invader, O propinquus, has not displaced the native species. In several of our lakes both continue to coexist. O. rusticus, which is more aggressive than the other species, overwhelmingly dominates in Sparkling Lake. In Trout Lake it has spread slowly, but where it is present it reaches abundances far higher than those of the other species (Fig. 22, p. 33). Experimental studies have shown that, at observed population levels, O. rusticus has the potential to exert a major influence over littoral zone communities (Lodge and Lorman 1987); however its impact under natural conditions remains to be demonstrated.

Proposed Research. We will continue to evaluate the effects of invading species on our study lakes. We also will focus on shifting patterns of invasion frequency and effect as they respond to environmental changes associated with climate.

How will climate associated changes in water levels and thermal regimes influence patterns of colonization and extinction? The rapid changes in climate anticipated in the next half-century (Manabe and Stouffer 1980) will likely cause a new set of stresses on the aquatic ecosystems at the NTL LTER site. These may include drops in water level both in our lakes and their associated wetlands, as well as changes in flow regimes. Increased temperatures are expected to alter thermal habitats, especially in the littoral zone and surface waters, where many organisms spend the larval life. Such additional pressures on the native communities may increase the probability of establishment of invading species and the extinction of native taxa, thus increasing the severity of the effects on the lakes.

We will approach this problem by using results from the regional hydrology (p. 42), thermal structure (p. 11), and water quality models (p. 44), along with our records of current species assemblages and habitat requirements to predict changes in associations in the LTER lakes. Ultimately we would like to generalize our findings to the entire lake district.

How will the spread of rusty crayfish affect the littoral zone communities in Trout Lake? We will continue to monitor the natural experiment of crayfish invasion and spread in Trout Lake. We have nearly annual crayfish distribution data dating back to 1970 and macrophyte and macroinvertebrate data starting in 1981. Collection of these detailed data will continue. Because the rusty crayfish has increased in abundance in the past few years, we expect to see shifts in littoral zone communities over the next several years.

What is the effect of macrophytes on their environment and other biota. Vegetation can effect the physical, chemical, and biological properties of the littoral zone (Lodge et al. 1988, Lodge and Lorman 1987). Consequentially the invading crayfish, O. rusticus by eliminating the macrophytes, can alter in significant ways the function of littoral communities. Obvious effects include loss of cover for fishes and habitats for macroinvertebrates. We will also examine the influence of the macrophytes on the oxygenation of the rooting zone (Jaynes and Carpenter 1986) and the subsequent influence of that on the fauna. A variety of differently vegetated habitats across the range of lakes will be compared, including the deep-water, poorly-illuminated submersed moss community below 10 m depth in Crystal Lake.

2. Acidic Deposition

Introduction. Acid deposition is an environmental stress of internationally recognized importance (National Academy of Sciences 1986, Altshuller and Lindhurst 1984). Lakes at the NTL-LTER site receive substantial acid loading (mean pH = 4.6 at the Trout Lake NADP Site) and many are classified as sensitive to the effects of acidification (Eilers et al. 1989, Watras and Frost 1989). Our research on acid deposition has focused on analysis of the role of groundwater buffering in mitigating acidification effects (p. 15) and on a whole-lake acidification experiment.

Results. The Little Rock Lake Experimental Acidification Project (LRL) was established to identify both the direct and indirect mechanisms by which increasing acidity affects population, community and ecosystem level conditions in lakes (Watras and Frost 1989). The program is funded by US-EPA and involves investigators from five institutions (Table 6). Little Rock Lake is a secondary LTER study lake that lies within the same groundwater system as our primary lakes. Sampling on Little Rock has been designed to parallel LTER efforts and there is close coordination between the two programs.

Following a baseline period, the two basins of Little Rock Lake were separated with a watertight curtain in fall 1984, and acid additions were begun to the north basin at ice-out in spring 1985. Our experimental design involves three, two-year acidification stages beginning at the lakes original pH of 6.1 and progressing through 5.6, 5.1 and 4.6. Baseline data collected prior to acidification indicated that the two lake basins were similar in physical, chemical, and biological conditions. Results after acidification indicated several distinct responses during the pH 5.6 stage (Table 7, p.

35) and an increased number at pH 5.1 (Table 8, p. 36). Data from the first year at pH 4.6 revealed in-lake conditions that were fundamentally different than those prior to acidification.

A key feature of the interplay between the LRL and LTER projects has been the development of general techniques for the evaluation of unreplicated, large-scale experiments. Experiments of this scale have considerable advantages (Schindler 1988) but they can only be interpreted against a background of natural variability (Frost et al. 1988). Data collected on LTER lakes in parallel with Little Rock Lake provide such critical information on natural variability. We are developing techniques to characterize this variability systematically (Kratz et al. 1987, Frost and Kratz in preparation). Much of this work has been conducted in collaboration with S. Carpenter who has been involved in whole-ecosystem manipulations on lakes nearby to our LTER site (e.g., Carpenter et al. 1989).

Proposed Research. Our efforts in this area will involve continued work on the influence of groundwater on lake acid-base conditions, the completion of the acidification stage of Little Rock Lake, and the initiation of a recovery experiment on Little Rock Lake (with non-LTER funding).

D. SPATIAL AND TEMPORAL VARIABILITY

Introduction. Traditionally, lake