Home > Research > NSF Proposals > 1996 Proposal

COMPARATIVE STUDY OF A SUITE OF LAKES IN WISCONSIN

a proposal to the National Science Foundation

Division of Experimental Biology Long-Term Studies Program

from the

North Temperate Lakes Long-Term Ecological Research Program

John J. Magnuson

Principal Investigator and Program Director

Center for Limnology

University of Wisconsin-Madison

Period: November 1, 1996 - October 31, 2002

Table of Contents

SECTION 1. PROJECT SUMMARY

SECTION 2. RESULTS OF PRIOR SUPPORT

• RESEARCH HIGHLIGHTS

SECTION 3. PROPOSED RESEARCH

• CONCEPTUAL FRAMEWORK
• PERCEPTION OF LONG-TERM CHANGE
• IN-LAKE DYNAMICS
• THE ORGANIZATION OF LAKE DISTRICTS
• CLIMATIC EFFECTS ON LAKES IN THE LANDSCAPE
• HUMAN - LAKE - LANDSCAPE INTERACTIONS
• SYNTHESIS AND SIGNIFICANCE

SECTION 4. LITERATURE CITED

SECTION 5. PROJECT MANAGEMENT

SECTION 6. DATA AND INFORMATION MANAGEMENT

SECTION 7. OUTREACH

SECTION 8. CURRICULUM VITAES AND COLLABORATORS

Section 1. Project Summary

Lakes are central to the vitality of landscapes and society. As collectors of water, energy, solutes, and pollutants from the landscape and atmosphere, as habitats for aquatic biota, and as attractors of human activities, lakes affect and are affected by natural and human-induced changes in the local and regional landscape and atmosphere. The North Temperate Lakes Long-Term Ecological Research program seeks to understand the long-term ecology of lakes and their interactions with a range of relevant landscape, atmospheric, and human processes. Our program has the following five, interrelated goals.

1. Perceive long-term changes in the physical, chemical, and biological properties of lake ecosystems.
2. Understand interactions among physical, chemical, and biological processes within lakes and their influences on lake characteristics and long-term dynamics.
3. Develop a regional understanding of lake ecosystems through an analysis of the patterns and processes organizing lake districts.
4. Develop a regional understanding of lake ecosystems through integration of atmospheric, hydrologic and biotic processes.
5. Understand the way human, hydrologic, and biogeochemical processes interact within the terrestrial landscape to affect lakes and the way lakes, in turn, influence these interactions.

We examine patterns, processes, and interactions of lakes and their surroundings at a nested set of spatial and temporal scales. Spatially, we focus on four scales: individual lakes, multiple neighboring lakes, entire lake districts, and the Upper Great Lakes region. Temporally, we consider scales of within-year, among-years, decades, and centuries. We use multiple approaches, including long-term observations, comparative studies, experimental manipulations, and process modeling. Our research group includes ecologists, geologists, chemists, demographers, historians, rural sociologists, climatologists, and remote sensing and data management specialists.

We expect our research to produce new conceptualizations of lake dynamics at local to regional scales. These conceptualizations will include understanding the importance of spatial positioning of lakes in a landscape, the feedback between human and lake processes, and the influence of climate and land-use change on lakes. Collectively, the understanding of landscape-lake-human interactions developed through our LTER research program will have direct relevance to development of policies affecting the future of the Upper Great Lakes Region and enhancement of the quality of life of its residents.

Section 2. Results of Prior Support

Comparative Studies of a Suite of Lakes in Wisconsin

Grant # DEB9011660 Funding (1991-1996) = $4,190,193

The North Temperate Lakes Long-Term Ecological Research (NTL-LTER) site was established in 1981. Over the past 15 years we have designed and implemented a comprehensive study of seven lakes in a forested landscape within the Northern Highland Lake District of northern Wisconsin. In November 1994, we used NSF's augmentation of our site and additional state of Wisconsin matching funds to add four study lakes in agricultural and urban catchments in southern Wisconsin, increase our breadth by adding social scientists to our research group, and increase the regional significance of our findings by supporting extensive interactions with researchers at two Canadian lake research sites.

As evidenced by the 161 peer-reviewed publications produced during 1990-1995, we have made significant advances in the understanding of lake ecology (Figure 1 and attached publication list). The listed publications include papers and theses by LTER investigators and other investigators who have used our data, and synthetic as well as policy-oriented publications generated from interactions involving our LTER scientists.

Prior results have a similar breadth of perspective as does our proposed research. Using information from our long-term database, we studied the ecology of lakes by considering in-lake processes, interactions between lakes and their surrounding environment, lake-landscape relationships within regions, and human-lake interactions. For brevity, only a few, representative examples are covered here. Others are shown in proposed research and our publication list.

We collect and manage high-quality, comprehensive, long-term datasets on the physical, chemical, and biological properties and processes of the LTER lakes and the surrounding landscape (Magnuson and Bowser 1990, Kratz et al. 1986). All of these data are available electronically to LTER investigators and collaborators and subsets are published on the World Wide Web (Table 1). The number of data requests and access history of our web page are summarized in Table 2. Data management at our site is described in more detail in Section 5: Data and Information Management.

• Invasion of Exotics: Two notable exotics, rusty crayfish and rainbow smelt, have invaded the northern Wisconsin LTER lakes since our study began. Crayfish greatly modified the macrophyte community and altered the functioning of the littoral zone (Lodge 1991). Smelt invaded the pelagic zones and through predation have caused the extirpation of a native planktivore (cisco) in one lake (McLain 1991), and through competition have greatly reduced the growth of yellow perch in another lake (Hrabik 1995) (Figure 2).

• Algal Population Regulation: We studied factors regulating aspects of algal dynamics in northern and southern Wisconsin lakes. In the north we hypothesized that dinoflagellate population differences in two LTER bog lakes (Crystal Bog and Trout Bog) could be attributed to population growth or loss processes. Unexpectedly, the interplay between basin morphology and the emergence of resting stages from sediments explained the difference (Sanderson and Frost in press). In the south, we analyzed blue green algal dynamics at two time scales in Lake Mendota. Daily concentrations of blue-green algae during summer stratification were best predicted by rainfall, irradiance and wind velocity (Soranno 1995). A stochastic model that predicts annual frequency distributions of blue-green algae from spring total P was developed using data for 1976-1995 (Stow et al. 1996).

• Nutrient Recycling and Availability: Based on sedimentation data, recycling within the water column serves as the primary source of N and P for annual production in several of our study lakes (Soranno 1995, Poister et al. 1994). Among-lake patterns in seasonality of production are influenced strongly by population processes of large siliceous algae that sink and efficiently remove nutrients from the water column (Poister et al. 1994).

• Acid Stress: The Little Rock Lake Experimental Acidification Project provides an example of collaborative research substantially facilitated but not directly supported by NTL-LTER. The lake's treatment basin was acidified to pH 4.7 in three stages (Figure 3). NTL-LTER data were combined with information from Little Rock Lake's reference basin to evaluate responses to acidification (Carpenter et al. 1991, Brezonik et al. 1993) which were subsequently compared with other large-scale acidification experiments (Schindler et al. 1991). Populations of individual zooplankton species responded dramatically to acidification (Frost et al. 1992); many responses could not have been predicted from laboratory bioassays (Eaton et al. 1992, Gonzalez and Frost 1994). Compared with species composition, total biomass of zooplankton responded slowly (Figure 4) indicating functional compensation in the assemblage (Frost et al. 1995). Following acidification we observed a substantial delay between fast chemical (Figure 3) and slow biological recovery (Figure 4).

• Grazing and Phosphorus Effects on Phytoplankton: Analysis of Lake Mendota Secchi disk transparencies, P levels, and food web structure from 1900-1993 revealed strong, additive effects of both grazing and P (Lathrop et al. 1996). Summer transparency improves from 1 m to 3 m when small daphnids are replaced by large ones and P loads are reduced by below-average rainfall. Synthesis of long-term records from Lake Mendota shows that planktivory by cisco can shift the daphnid assemblage from large-bodied to small-bodied species (Johnson 1995). However, the lake's two other major planktivores, yellow perch and white bass, are not capable of causing a shift in daphnid size.

• Groundwater/Lake Interactions: We used stable isotopes and numerical modelling to quantify groundwater input to lakes in the Trout Lake area (Krabbenhoft et al. 1990a,b, Bowser 1992, Kenoyer and Bowser 1992a,b, Anderson and Cheng 1993, Cheng and Anderson 1993, Krabbenhoft et al. 1994). Our lakes, characteristic of the region, receive 0 to 40 % of their water from groundwater and the remainder from direct precipitation (Ackerman 1993, Michaels 1995)

• Ice Phenology and Climate Change: Lake ice phenologies provide the longest continuous record linking limnology and climate. Lake ice cover data were analyzed as indicators of past and future climate at local to regional scales, including large portions of the Laurentian Shield (Robertson et al. 1992, Wynne and Lillesand 1992, Wynne and Lillesand 1993, Assel and Robertson 1995, Anderson et al. 1996, Wynne et al. 1996)( Figure 5). Using records for the Lake Mendota area, we found that winters have warmed by 2.5(C since 1843 and that ice duration is strongly influenced by El Ni-o events.

• Thermal Habitat and Climate Change: The consequences of climatic change for temperate lake physics, chemistry, and ecology are potentially significant (DeStasio et al. 1996, Magnuson and DeStasio 1996, Magnuson et al. 1995, Magnuson et al. 1996a,b, Webster et al. 1996, Vavrus et al. 1996). In collaboration with Australian limnologists, we used output of General Circulation Models based on 2X CO2 scenarios as input to lake physical limnological models (Dynamic Reservoir Simulation Model - DYRESEM), and used the altered lake physics to simulate changes in lake ecology. With the exception of coldwater fish in the smallest lakes, all thermal guilds of fishes had an increase in thermal habitat with warming scenarios (Figure 6).

• CO2 Dynamics: We identified and roughly quantified the role that surface waters play as conduits for terrestrially fixed carbon to the atmosphere (Cole et al. 1995). In most lakes we documented annual net movement of carbon from the lake to the atmosphere. This was especially important in dystrophic lakes (Hope et al. in review).

• Paleolimnology of Silica: We developed a new paleolimnological technique to evaluate a lake's historic silica concentrations using sponge spicules (Kratz et al. 1991a). Lakes in the Northern Highlands Lake District underwent substantial declines in silica concentrations over the last 10,000 years (Figure 7) perhaps because of changes in weathering rates caused by depletion of easily-weathered, fine particles.

• Organization of Lake Districts: We have begun to develop a conceptual framework for lakes that combines aspects of the river continuum (Vannote et al. 1980) and stream order concepts to explain spatial heterogeneity of lakes across a landscape (research area 3b, below). Groundwater flow is a key factor linking landscape position and lake structure and function (Magnuson et al. 1990, 1991, Kratz et al. 1991b, 1994, 1995, in review).

• Species Dynamics of Lakes as Islands: We have begun to elucidate the dynamics of species richness and assemblage structure in inland lakes (McLain and Magnuson 1988, Tonn et al. 1990, Magnuson et al. 1994, Arnott et al. MS). We find that lakes appear more open to invasion and seem better described by metapopulation processes than their island appearance would suggest (research area 3c, below). Among-lake heterogeneity in fish assemblages is only slightly greater than within-lake heterogeneity (Benson and Magnuson 1992)

• Advances in the Use of Remote Sensing: We developed approaches and protocols for using remote sensing to map land and lake cover on a statewide basis for Wisconsin (Bolstad and Lillesand 1992a,b,c, Lillesand 1993a,b, Benson and MacKenzie 1995) as well as ice-off dates for wide regions (Wynne and Lillesand 1993, Wynne et al. 1996).

• Nutrient Load Modelling: A model to predict annual phosphorus loads to Lake Mendota from land use in the watershed has been developed, validated, and used to explore the consequences of progressive urbanization (Soranno et al. 1996). P input rates were compared in detail with internal recycling rates (Soranno 1995). A model to predict the frequency distribution of P loading events from land use and precipitation data is partially completed (Figure 8). This model will be used to explore scenarios of climate change interacting with land-use change (research area 5, below).

Our roots are in limnology which is traditionally interdisciplinary across the natural sciences, i.e., physics, chemistry, and biology. Our analyses link processes across these disciplines (research area 1 - 5, below). We are an interactive group working at and among levels of ecological organization from autecology to landscape ecology. We study trophic ecology, biogeochemistry, population biology, determinants of biodiversity, landscape ecology, and mixes of these. We use multiple approaches including modeling, experimentation, description, and comparative ecology. We began incorporating social sciences with the 1994 augmentation. A highly successful cross disciplinary workshop was held in November 1994 and was followed by a seminar in Fall 1995 entitled Lakes and Society led by a limnologist and a rural sociologist. Our present proposal has a strong human component (research area 5).

We initiate and participate in research projects and the network data management process at an intersite level. We organized an intersite analysis of variability of North American Ecosystems that included data from 12 LTER sites (Magnuson et al. 1991, Kratz et al. 1995). Regardless of biome, spatial variability was greater than interyear variability and plant or animal parameters were more variable in space and time than chemical or physical variables (Figure 9). We expanded this approach to compare spatial heterogeneity at the scale of full-LANDSAT scenes for 14 scenes that include LTER sites (Riera et al. in prep.) (Figure 10).

Our LTER site has been successful in catalyzing LTER-related research by others as well as by our own researchers. The potential for interacting with LTER researchers and our databases has been key in generating this interest. These other projects are funded from a range of federal (NSF, EPA, DOE, USGS, USDA-SCS, NASA), state (Wisconsin DNR, University of Wisconsin) and private (endowments, Pew, and EPRI) sources. From 1990-1995 an average of $1,624,000 per year of additional research has been associated with the LTER site; of this an average of $279,000 per year is from other NSF funding and $1,324,000 is from non-NSF sources. During the same six years our LTER base plus the 1994 augmentation and other supplements (GIS, Technical, REU) averaged $653,000.

We have contributed significantly to education, 16 MS and 8 Ph.D. theses related to LTER research were produced over the six years. Education of undergraduates and outreach are covered in Section 6: Outreach. Internationally, we have hosted visiting investigators, and post-doctoral, and graduate students, including long-term collaborations with researchers from Germany, China, Spain, Scotland, Honduras, and Venezuela and shorter-term visits from Hungarian, Czechoslovakian, and Russian scientists.

Section 3. Proposed Research

Lakes are conspicuous, valuable and vulnerable landscape features throughout the world. Small, inland lakes are particularly prominent throughout the Upper Great Lakes region of North America. From the fertile, loess-capped soils of the north-central U.S. to the Precambrian outcrops of the Canadian Shield, the thousands of inland lakes play a central role in regional hydrologic and biogeochemical cycles, in biological processes influencing the area's diversity of aquatic and terrestrial life, and in a wide range of human activities. Over the past two centuries, deforestation, fire suppression, agriculture, industrialization, and urbanization have transformed landscapes within the region and fundamentally altered the relationships of lakes to their surroundings. Patterns of change in lakes and surrounding landscapes have been influenced by the availability of lakes for irrigation, industry, human waste, transportation, fishing, and recreation. For the next century and beyond, the quality of life and the economies of the region will depend upon the quality of the lakes.

Our research approach considers lakes as interactive components of their environment. As collectors of water, energy, solutes, and pollutants from the landscape and atmosphere, as habitats for aquatic biota, and as attractors of human activities, lakes affect and are affected by natural and human-induced changes in the local and regional landscape and atmosphere. The North Temperate Lakes Long-Term Ecological Research (NTL-LTER) program seeks to understand the long-term ecology of lakes and their interactions with a range of important landscape, atmospheric, and human processes. We use a nested set of spatial scales including individual lakes, multiple neighboring lakes, entire lake districts, and the Upper Great Lakes region (Figure 11). This multiscale approach distinguishes our efforts from most previous limnological research and provides an opportunity to identify controlling factors operating at scales ranging from individual lakes to regions. Within this conceptual framework, our program has the following five interrelated goals.

(1) Perceive long-term changes in the physical, chemical, and biological properties of lake ecosystems.

This goal includes the collection and management of our core datasets. Most of the research we describe depends on these continually evolving, long-term datasets. We propose to add two new long-term core datasets; one on coarse woody debris, an important structural feature of the littoral zone of lakes, the other on stable isotopes of hydrogen and oxygen, important hydrologic tracers.

(2) Understand interactions among physical, chemical, and biological processes within lakes and their influences on lake characteristics and long-term dynamics.

This goal is focused on the within-lake scale. We will examine such processes as the interaction of nutrient and food-web effects on phytoplankton; the influence of bacterial processes on nutrient availability and phytoplankton growth; the effects of lake chemistry, morphometry and foodweb structure on the fate of nutrients; the influence of functional compensation among species on ecosystem resilience to environmental change; and the way exotic species affect lakes.

(3) Develop a regional understanding of lake ecosystems through an analysis of the patterns and processes organizing lake districts.

Here, we address patterns and processes important at lake-district to regional scales. We propose to extend analyses performed previously on a subset of lakes in northern Wisconsin to the entire lake-district, and assess the generality of our results using comparisons with patterns observed at other lake districts within the region. We propose to examine the degree to which lakes exhibit synchronous dynamics, the role spatial positioning of lakes plays in controlling lake attributes and dynamics, and how patterns of connectivity among lakes can be important to regional biodiversity.

(4) Develop a regional understanding of lake ecosystems through integration of atmospheric, hydrologic and biotic processes.

This goal takes a process-modeling approach to understanding atmospheric, terrestrial, hydrologic, and aquatic linkages at lake-district to regional scales. We propose linking several process models to allow spatially-explicit tracking of water movement through the hydrosphere. We ask how changes in climate affect lakes both directly by altering lake thermal dynamics, and indirectly through changes in hydrologic processes in the terrestrial environment.

(5) Understand the way human, hydrologic, and biogeochemical processes interact within the terrestrial landscape to affect lakes and the way lakes, in turn, influence these interactions.

In this final goal, we focus on the feedback loop between terrestrial processes and lake processes. Lakes are affected directly by landscape processes, but also participate in feedbacks that alter these processes, largely through changes in human perceptions and actions. We focus on interdisciplinary analyses of land-use change and lake response.

In addressing these five goals we use a variety of approaches including long-term observations, small- and large-scale experiments, comparative studies, and process modeling. In the sections that follow, we elaborate on each of these five goals.

Rationale. One of the basic goals of the North Temperate Lake LTER program is and has been the collection and management of ecologically important data that allow investigators to observe and analyze long-term changes in physical, chemical, and biological features of lakes. Long-term observations and analyses are crucial to understanding lake ecology. Natural phenomena, such as strong year-classes of long-lived predators or a series of drought years, can cause multi-year to decadal effects in lakes, often with substantial time lags between cause and effect (Magnuson et al. 1990, Carpenter and Leavitt 1991). In addition, many human-induced pressures influencing lakes, such as invasions of exotic species, eutrophication, and climate change, operate over time scales of years to decades or longer. Holling (1995) points out that the accumulation of these effects can cause changes which are difficult to understand without a long-term context. To provide such a context for our research, we collect and maintain a series of 'core' databases (Table 1, Table 3). These datasets provide the basis for addressing most of our research questions.

Background. Over the past 15 years we have designed and implemented a balanced and integrated data collection program (Kratz et al. 1986, Magnuson and Bowser 1990). Our choices of lakes and measurements have been guided by a desire to address important interdisciplinary questions regarding the ecology and management of lakes from a long-term perspective at individual lake, multiple lake, lake-district and regional scales.

We focused our data collection on two sets of lakes and their surrounding landscapes. One set is in the forested and tourism-dominated Northern Highland Lake District in northern Wisconsin, the other is in the agricultural- and urban-dominated landscape in and near Madison in southern Wisconsin (Figure 12). Both regions have a substantial history of limnological research dating back to 1900 (Frey 1963).

In addition, we have working relationships with two Canadian groups with similar data on two other lake districts, the Experimental Lakes Area in western Ontario and the Dorset Research Centre in eastern Ontario (see Section 4: Project Management). Collectively, the data and research programs at these four focal lake districts afford a unique opportunity for regional analyses.

In northern Wisconsin, beginning in 1981, we focused on a suite of lakes and surrounding terrestrial areas linked through a common groundwater and surface water flow system and sharing a common climatic, edaphic, and biogeographic regime. The lake set includes oligotrophic, dystrophic, and mesotrophic lakes (Table 4, Figure 12). Our seven primary study lakes, located within 5 km of the Trout Lake Station, were chosen to represent marked differences in size, morphometry and habitat diversity, in thermal and chemical features, in species richness and assemblies, and in position in the groundwater flow system. In addition to the seven primary lakes we also have a set of secondary lakes for which less complete information is collected. The choice of secondary lakes and types of measurements change over time. These lakes are used for comparisons with primary lakes on specific research questions.

With the augmentation of our LTER project in November 1994, we added four primary study lakes in southern Wisconsin (Table 4). These four eutrophic lakes were chosen in a 2x2 design of urban vs agricultural setting and headwater vs lower in the landscape. Substantial historical data are available on these lakes. Lake Mendota has been studied since 1888 (Kitchell 1992, Brock 1985) and had been an LTER secondary lake prior to the 1994 augmentation. The Wisconsin DNR has maintained a high-quality database for Lakes Mendota and Monona since 1976 (Lathrop et al. 1992, Lathrop and Carpenter 1992a,b) which is now being integrated with the NTL-LTER database. Lake Wingra was studied intensively during the IBP program in the early 1970's (Watson and Loucks 1979). Fish Lake is part of a Wisconsin DNR network of sentinel lakes. WDNR and the Center for Limnology have studied Lakes Fish and Wingra since 1991 through a collaborative project on macrophyte-fish interactions (Trebitz et al. 1993, Carpenter et al. 1995a).

Our sampling program allows comparisons of parameters and processes among seasons, years, lakes, and lake districts. We sample most major physical, chemical and biological parameters (Table 3) with sampling frequencies 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 severely stress or limit biological components. Complete cation-anion balances are computed for each period. Nutrients, pH, inorganic and organic carbon are sampled every two or four weeks, depending on the lake and the nutrient. Temperature, dissolved oxygen, chlorophyll a, light penetration, and zooplankton abundance are measured every two weeks during the open-water season and every five weeks under ice cover. Samples for phytoplankton community composition are collected and primary production rates are measured every two weeks from selected lakes. 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. Groundwater levels in selected wells are measured monthly and groundwater chemistry from a subset of these wells is measured annually. In addition, we maintain an automated land-based weather station at the local airport 10 km from Trout Lake; a raft on Sparkling Lake for measurements of evaporation, wind stress, and high resolution thermal structure; and have access to National Weather Service data from the Madison airport.

In addition to providing comprehensive limnological data, this sampling program positions us to detect invading exotic species in our primary lakes. Potential new invaders include many European species (Mills et al. 1993) that have reached the Laurentian Great Lakes. These large lakes now act as a nearby source (mainland in an island biogeographic sense) of colonists including fishes (ruffe, rainbow smelt, rudd, round goby, etc.), zooplankton (Bythotrephes cederstroemi, Eurytemora affinis, etc.), molluscs (zebra mussels, fingernail clams, and a variety of snails), and a macrophyte (Eurasian watermilfoil, which occupies the Madison area lakes and could invade the Trout Lake area (Nichols and Lathrop 1994, Nichols and Yandell 1995)). We have designed our sampling so that introductions of these or other invading species will be discovered early and we can implement specific research activities to understand consequences of these introductions.

To provide basic information about the terrestrial landscapes surrounding our study lakes, we have developed a geographic information system that includes data layers on land cover (from presettlement to present), soils, roads, and geological substrate. We have acquired multiple Landsat Thematic Mapper and SPOT scenes for both the Trout Lake and Madison areas (Table 3). Members of the LTER research group have been leaders in the research and development of the Wisconsin Initiative for Statewide Cooperation on Landscape Analysis and Data (WISCLAND), a statewide land-cover mapping program cooperatively sponsored by federal, state, and local agencies. When fully implemented WISCLAND will provide statewide land cover mapping for Wisconsin at 30 meter resolution. WISCLAND incorporates numerous state-of-the-art image processing procedures to ensure accuracy and categorical detail of the classification. These procedures include: dual date TM data to use phenology to aid vegetation classification; an ecoregion-by-ecoregion study area stratification; separate classifications of rural vs. urban areas and wetlands vs. uplands; applications of hybrid "guided" clustering and maximum likelihood classification; and employment of stratified, systematic, non-aligned statistically-based sampling for collection of training and accuracy assessment data. With over 40,000 field verified sample polygons distributed across all of Wisconsin, WISCLAND is one of the largest and most technologically complex land cover mapping programs in the world. The WISCLAND land cover classification system and image processing protocol have been adopted for use throughout Minnesota and Michigan as part of the Upper Midwest GAP Analysis Program (Lillesand 1996). NTL is ideally positioned in the center of the regional land cover database resulting from these efforts.

Proposed New Core Datasets. Collecting and maintaining the core datasets listed in Table 3 provide the basis for addressing the research questions presented in this proposal. We will continue to collect and maintain these datasets. In addition we will add two new datasets, one on coarse woody debris, the other on stable isotopes of oxygen and hydrogen in lakes, groundwaters, and precipitation.

Coarse woody debris (CWD) provides habitat and is a critical ecosystem component in streams and rivers (Harmon et al. 1986, Andrus et al. 1988, Bilby and Ward 1991), but its role in lake ecology is poorly understood. In lakes, CWD is likely to be important as habitat for periphyton, invertebrates, and fish. In lakes, as in rivers, human activities significantly alter CWD inputs and standing stocks (Spies et al. 1988, Maser and Sedell 1994). For example, cottage development is associated with a dramatic reduction in CWD (Christensen et al. 1996). Depletion of CWD could affect lakes for centuries (Andrus et al. 1988, Maser and Sedell 1994). Yet the dynamics of CWD in lakes is unknown. We will initiate a program to document rates of introduction and removal of CWD in littoral zones with different amounts of cottage development and contrasting structure of riparian forests. CWD in each primary NTL-LTER study lake will be mapped annually to ca. 1 meter accuracy using a global positioning system. We will be able to estimate annual input and removal rates as a function of cottage density, shoreline use, and riparian vegetation. The database will also provide a long-term context for graduate student and REU research projects to study influences of CWD by periphyton, invertebrates and fishes.

Measurement of stable isotopes will allow us to document variation in the annual fluxes of water to lakes through groundwater and direct precipitation, key links to external fluxes of nutrient and major ions to lakes. Krabbenhoft et al. (1990b) and Gat et al. (1994) have emphasized the use of stable isotopes of oxygen and hydrogen to quantify hydrologic budgets of seepage lakes and the dominant role of groundwater in the flux of major ions and nutrients to lakes. We will sample stable isotopes quarterly in LTER lakes and local precipitation and calculate year to year variation in groundwater fluxes to lakes. Combined with core groundwater chemical data, annual fluxes of nutrients and other solutes to lakes and their interannual variation will be estimated. Results will provide a valuable link between external and internal controls on in-lake processes.

(2) Understand interactions among physical, chemical, and biological processes within lakes and their influences on lake characteristics and long-term dynamics.

Rationale and Background. Understanding the fundamental processes operating within lakes is a focus of the NTL-LTER program. Internal physical, chemical, and biological processes filter the effects of external factors on the long-term properties of lake ecosystems (e.g., Reynolds 1984, Likens 1985, Carpenter 1988). Evaluations of external factors (research areas 4 and 5) controlling broader-scale patterns in the properties of lakes within a landscape (research area 3) draw on a substantial knowledge of mechanisms controlling the properties of individual lakes.

Our past work on key linkages between in-lake processes and lake-ecosystem features helped us identify five research questions. These efforts, ranging from intensive experimental and observational studies on single lakes to multi-lake comparisons, center on processes occurring within lakes. A challenge in ecology is to evaluate interactions among patterns and processes that fluctuate at different time scales (O'Neill et al. 1986; Carpenter 1988, Holling 1995). Our diverse approaches combining mechanism with context help clarify linkages among phenomena that span a wide range of spatial and temporal scales.

Proposed Research.

a) How do nutrient availability and food-web structure interact in lakes of differing trophic status to influence phytoplankton dynamics?

Phytoplankton biomass in lakes can be controlled by nutrients and food-web processes, but the interactions of these factors are not yet predictable (Reynolds 1994). Abundant evidence links phosphorus inputs to primary production (Schindler 1977, 1978). Food-web processes can affect phytoplankton through grazing (Hrbacek et al. 1961, Gulati et al. 1990, Carpenter and Kitchell 1993) or recycling of nutrients (Carpenter et al. 1992, Elser and Hassett 1994). However, food-web manipulations of lakes yield variable results (Reynolds 1994). We expect that a lake's response to shifts in nutrients or food-web processes will depend on its trophic condition.

We will examine the role of trophic condition by comparing nutrient and food web effects on phytoplankton in lakes of contrasting trophic state. We have observed substantial inter-annual changes in the water clarity of eutrophic Lake Mendota and oligotrophic Crystal Lake (Figure 13). In both lakes, shifts in the abundance of a dominant planktivorous fish (yellow perch in Crystal, cisco in Mendota) correlate with changes in water clarity (Magnuson 1990; Rudstam et al. 1993, Johnson 1995, Lathrop et al. 1996). We hypothesize that both nutrient and food-web dynamics drive year-to-year changes in water clarity, but that the mechanisms by which these processes influence producers differ. Specifically, we predict that, in the oligotrophic system, nutrient effects are mediated by grazers' excretions, whereas, in the eutrophic system, phytoplankton are controlled by an interaction of grazing and nutrient loading from runoff and hypolimnetic entrainment.

Long-term data will be combined with rate estimates from the literature and short-term experiments (e.g. Lehman and Sandgren 1985, Elser and Goldman 1991, Elser and Hassett 1994), to assess the relative impacts of grazing, nutrient recycling by grazers, and phosphorus loading on changes in phytoplankton abundance. Necessary data on nutrients, zooplankton, and phytoplankton are in hand. Time-series methods (Box et al. 1994) will be used to estimate the effects of changes in nutrients and grazers on phytoplankton using long-term data.

While grazer effects on phytoplankton have been measured in lakes across a range of phosphorus levels (Jeppesen et al. 1990, Reynolds 1994, Carpenter et al. 1996), the general role of trophic status is still not clear. In our study, the context provided by a long-term perspective and strongly contrasting lakes will provide insight into the role of trophic conditions on the interaction of zooplankton, nutrients and phytoplankton. A replacement of yellow perch in Crystal Lake by invading rainbow smelt (Question e, below) provides an additional set of conditions within which nutrient/food-web interactions can be evaluated.

b) How does the bacterial community influence nutrient availability and phytoplankton growth?

Bacteria may have both positive and negative effects on the availability of nutrients to phytoplankton through remineralization and uptake. In some situations, bacterial growth appears to be phosphorus limited (Morris and Lewis 1992, Coveney and Wetzel 1995). At the same time, bacterial growth may also be influenced by organics released by algae (Cole et al. 1988). Predicting phytoplankton-bacteria interactions is complex. We will measure seasonal and intra-lake differences in growth limitation of bacteria and phytoplankton to determine the nature of their interactions for resources.

Mesocosm experiments will be used to determine nutrient limitation of bacterial and algal growth by carbon, nitrogen, and phosphorus. A dilution method, similar to that of Morris and Lewis (1992), will be used to minimize the influence of grazers in nutrient limitation experiments. We will test bacterial influences on algal growth using bacterial growth inhibitors and bacterial additions in short-term experiments. All experiments will be conducted across NTL lakes of contrasting trophic status, and during different seasons, to determine hydrologic, geologic and seasonal influences.

We predict that the influence of bacteria on algal growth will decrease during periods when each group is limited by a different nutrient. Limiting nutrients are expected to change between lakes and seasons as indicated by different ratios of available nutrients (Le et al. 1994). We predict that bacteria will compete for phosphorus and limit algal growth in the oligotrophic, phosphorus-poor lakes of the Trout Lake region. We anticipate that bacterial growth is more likely to be limited by carbon, and algal growth by phosphorus and/or nitrogen in the nutrient-rich southern Wisconsin lakes. Bacteria should influence algal growth less in these eutrophic lakes. Because bacterial growth can be strongly temperature limited in cold waters (White et al. 1991, Shiah and Ducklow 1994), we expect bacteria to influence algal growth less during spring and fall.

Understanding the relationship between nutrient availability and algal growth is central to lake management. Our study will demarcate the conditions under which bacteria influence the availability of nutrients to phytoplankton. Knowledge of intra-lake and seasonal differences in bacterial-algal interactions will increase our understanding of how phytoplankton dynamics in lake ecosystems are controlled.

c) How do chemistry, morphometry and food-web structure of lakes interact to influence the availability of spring nutrients to summer communities?

Spring is a critical time when weather and a lake's community structure can affect nutrient availability during later seasons. A combination of increasing insolation, high nutrient concentrations and low herbivory often lead to a spring build up of phytoplankton biomass (Sommer et al. 1986). The fraction of this biomass that sinks out of recycling zones helps determine total annual production (Reynolds 1984; Harris 1986, Guy et al. 1994, Poister et al. 1994).

We propose that chemistry, morphometry and food-web structure determine the fraction of spring P that becomes unavailable for later use. Si:P ratios and morphometry are related to net total phosphorus (TP) dynamics in NTL-LTER lakes. We will make use of strong gradients in Si, P, and morphometry among the lakes near Trout Lake and Madison to test four predictions. 1) Lakes with high initial Si:TP ratios will lose more of their spring TP to sinking because diatoms will dominate their spring phytoplankton communities. 2) Among lakes with low Si:P, larger populations of crustacean zooplankton will be associated with greater losses of spring P. 3) Lakes with littoral zones that are large relative to their mixed-layer volumes will retain more P within the water column. 4) Interannual variability in the ratio of spring TP: summer TP will be greatest in larger lakes where the depth and timing of stratification varies most from year to year. Our work will draw upon previous NTL-LTER studies linking lake-silica dynamics with ground-water inputs (Hurley et al. 1985, Krabbenhoft 1988) and tying patterns of ecosystem variability to landscape position (Kratz et al. 1991b). We will compare sinking fluxes and their standard deviations measured by inverse modeling (Jackson and Matsu'ura 1985, Vezina 1989, Vezina and Pace 1994).

Comparative studies of flow networks have proven powerful in marine ecology (Wulff et al. 1989), stream ecology (DeAngelis et al. 1989), and community ecology (Pimm 1991, DeAngelis 1992). Our inverse modeling is the basis for a comparable synthetic approach to P cycles of diverse lakes.

d) What is the role of functional compensation in ecosystem resilience to environmental change?

Ecological responses to stress typically involve dramatic changes at the population level, but major disruptions in ecosystem processes are less frequent (Schindler 1987, Schindler et al. 1991, Frost et al. 1995, Tilman in press). Previous studies suggest that compensatory dynamics among functionally-similar species play a major role in determining ecosystem resilience to stress; however, this phenomenon is poorly understood (Ives 1995, Frost et al. 1992, Lawton and Brown 1993, Walker 1995). Understanding the factors that determine functional compensation in the dynamics of stressed communities will be critical to predicting ecological responses in a changing environment. We propose to investigate the interplay between functional compensation and ecosystem resilience in unperturbed and manipulated lakes.

Using the NTL-LTER zooplankton dataset, we have begun to assess the responses of individual species and groups of species in the same trophic guild to naturally-occurring environmental fluctuations. Preliminary results indicate that the summed biomass of species in the same trophic guild is less variable than biomass of individual species, suggesting that functional compensation is prevalent in the dynamics of zooplankton communities (Carpenter et al. 1993). To investigate links between diversity and community-level stability, we plan to compare the degree of functional compensation among lakes with a gradient of zooplankton species richness. We will use known allometric relationships to explore the implications of compensatory changes in zooplankton community structure for grazing potential. To assess the relationship between functional compensation and resilience in stressed systems, we will build upon our past experience and compare responses of different plankton assemblages to acidification in replicated mesocosms.

Our study of functional compensation in ecosystem response to environmental change has important implications for understanding ecological resilience, designing monitoring programs, and planning for conservation. Resilience is a core concept in ecology and ecosystem management, yet its underlying mechanisms remain poorly understood (Pimm 1991, DeAngelis et al. 1989, Gunderson et al. 1995). Ideal environmental indicators foreshadow any loss of ecosystem resilience (Gunderson et al. 1995). Conservation planning is increasingly moving toward an integration of population, community, and ecosystem processes (Franklin 1994, Jones and Lawton 1995); that will require explicit consideration of functional compensation.

Changes in Crystal Lake are likely to occur through two mechanisms. Smelt prefer smaller zooplankton prey (Hrabik 1995) and cooler temperatures than perch. They can also maintain higher consumption rates at cooler temperatures and over a longer growing season. We will test the importance of these mechanisms using small-scale experiments. We predict shifts in zooplankton community structure, abundance, and vertical distribution. By coordinating this study with analyses of plankton and water clarity change (Question a, above), we will assess effects of smelt invasions at the ecosystem level.

(3) Develop a regional understanding of lake ecosystems through an analysis of the patterns and processes organizing lake districts

Rationale and Background. Lakes in a lake district share a common climatic regime and geologic setting, but can differ markedly in physical, chemical, and biological attributes, as well as in their responses to regional phenomena such as climatic events, deposition of atmospheric pollutants, invasion by exotic species, or landscape alteration (Eilers et al. 1983, Cook and Jager 1991, Kratz et al. 1991b, Webster et al. 1996). Here we describe our efforts to develop a general conceptual framework which explains the spatial variability in properties of lakes at multiple lake, lake district and regional scales. We expect over the next six years that this research will lead to a new and powerful integration of lake-landscape interactions, prompting new ideas that we will develop and test. This framework will be analogous in scope and power to the River Continuum Concept for flowing waters (Vannote et al. 1980, Johnson et al. 1995).

We have accumulated substantial evidence for organizing patterns and processes at landscape scales within the Northern Highland Lake District. Diverse attributes, including lake area, major ion chemistry, vertical distribution of primary production, benthic trophic interactions, and species richness of fishes, have all been linked with the spatial position of lakes within local and regional hydrologic flow systems (Eilers et al. 1983, Webster et al. 1996, Kratz et al. 1991b, Kratz et al. in review, Riera et al. in prep) (Figure 14, Figure 15).

The generality of these organizing patterns and processes will be tested by extending this landscape framework from the Northern Highland to the Experimental Lakes Area, Dorset, and Madison lake districts (Figure 11). The four lake districts represent broad regional gradients, contrasting drivers of lake variability, and histories of disturbance. Most importantly, all four lake districts are sites of extensive research programs of unusual longevity. Comparisons across these four focal lake districts will be developed in two ways. First, we can compare the lake districts and test for generality of results observed at any one of them. Second, we can pool data from all lake districts to gain a regional perspective across the major ecoregions of the Upper Great Lakes region.

We expect that the relative importance of organizing factors changes as a function of spatial scale. Addressing these scale dependencies is an important goal of our proposed research.

Proposed Research. The history of research at each of the four lake districts provides us with a unique opportunity to understand patterns and processes organizing lake districts. We have identified three areas for emphasis during the next six years.

a. At what spatial and temporal scales, for which types of limnological variables, and to what extent do lakes vary synchronously?

Lakes are affected by many driving variables acting at a variety of scales. We expect the composite behavior of lakes over a large region to reflect a complex mixture of local, intermediate, and regionwide drivers. The scale at which a particular driving force is most important can be predicted by analyzing the temporal and spatial scales at which related limnological variables exhibit coherency (synchronous temporal variability, Magnuson et al. 1990). For example, if temperature and rainfall are important regional drivers on an annual time scale, we would expect lake water levels to increase regionwide in wet years, and decrease in dry years. However, at weekly scales, lakes levels may show little coherence across the entire region but high coherence within the same local drainage system.

We will use data from the four lake districts to test for coherency in a diverse set of physical, chemical, and biological variables at local, lake district, and regional spatial scales, and at a variety of time scales. We have begun these analyses for selected physical parameters in the entire Upper Great Lakes Region, and have found a high degree of coherence in lake water temperature (Benson et al. in prep). We will expand this research to the full set of physical, chemical, and biological lake attributes.

We plan to assess the generality of results from individual districts by comparing patterns of coherency among lake districts. In the Northern Highland Lake District, lakes tend to be more coherent in physical parameters, less so in chemical parameters, and least in biological parameters (Magnuson et al. 1990). This pattern may not be repeated in other lake districts; important differences may emerge in the analysis of other lake districts. For example, we expect that phosphorus loading, which shows no among-year coherence in the northern Wisconsin lakes, will be coherent in southern Wisconsin because agricultural runoff accounts for the major phosphorus input to these southern Wisconsin lakes, but is not an important driver in the north.

b. How does a lake's spatial position influence its physical, chemical, and biological properties?

A focal point for research at our site has been the development of a relationship between a lake's spatial position in the landscape and its physical, chemical, and biological properties (Magnuson et al. 1990, Kratz et al. 1991b, Krabbenhoft et al. 1994, Kratz et al. 1995, Webster et al. 1996). There are four logical extensions of this work that we propose to pursue.

First, we need to refine a simple quantitative method to locate or categorize the position of lakes in the hydrologic flow system. In lake districts dominated by surface water connections, lake order can be defined easily using criteria similar to those used to classify streams. However, in the Northern Highland Lake District, many lakes have no surface water inlets or outlets and groundwater provides the hydrologic connection between lakes. For these lakes we can quantify relative position in the landscape by estimating the percentage of the lake's hydrologic budget that comes from groundwater inputs using stable isotopes of hydrogen and oxygen as tracers (Krabbenhoft et al. 1990a,b, 1994). We will pursue linking surface- and groundwater- based definitions of lake order to develop a general quantification of landscape position that is inclusive of both seepage and drainage lakes.

Second, we propose extending our ordering of lake attributes according to position in the hydrologic flow system from the seven primary LTER lakes in northern Wisconsin to the entire Northern Highland Lake District. Although multiple historic and recent surveys of lake characteristics in the district provide a rich collection of limnological information (Juday et al. 1938, Black et al. 1963, Eilers et al. 1983, Linthurst et al. 1986, Tonn et al. 1990), these data have never been analyzed in a spatially explicit context. Initially, we will use GIS to assess relationships between a lake's landscape position and limnological parameters such as lake area, specific conductance, dissolved organic carbon, and fish species richness. Later we expect to expand this analysis to other parameters.

Remote sensing technology will allow us to characterize the absorption and scattering properties of constituents such as chlorophyll, DOC, and suspended solids in lakes throughout a lake district. We propose to measure wavelength-specific absorption coefficients (e.g., Bolgrien et al. 1995) using spectrophotometric and radiometric data, mixture decomposition methods, and our core lake data. The removal of atmospheric effects from radiometric data, critical to bio-optical modeling, will be facilitated by our participation in the NASA/LTER sun photometer network. We hypothesize that coefficients will be related to phytoplankton community structure and landscape variables regulating allochthonous inputs. Our ultimate objective is to develop a bio-optical model for constituent concentrations (Bukata et al. 1995) enabling remote monitoring of changes in regional lake water quality.

Third, we plan to extend our conceptual framework of lake-landscape position, which was developed in a lake district structured primarily by groundwater input, to the two lake districts on the Canadian Shield, both of which are dominated by surface water flow. We will start by using this framework to analyze the response of lakes in each of the lake districts to the regionwide drought of the late 1980's. We will test the hypothesis that the response of lakes to drought depends on landscape position and water retention time. For the first time, we are in a position to comprehensively analyze lake response to regional climatic events such as drought in a landscape context.

Finally, we propose to examine spatial scale dependencies of landscape effects on the chemistry of north temperate lakes. Lake chemistry can often be predicted by characteristics of the surrounding landscape, but the spatial scales at which landscape attributes exert a detectable influence on lakes are not well understood. We will start by asking two questions. First, what attributes of the terrestrial landscape explain significant variability in chemistry among lakes, and at what spatial scales are these effects apparent? And, second, does the spatial scale of influence vary between seepage lakes, which lack surface water inlets and outlets, and drainage lakes? We will approach these questions by analyzing the relationship between lake chemistry and landscape parameters (e.g., area of watershed in peatlands, vegetation and slope of the surrounding watershed) within buffer zones of increasing width for approximately 100 lakes. After an analysis of patterns in the Northern Highland Lake District, we will expand our analysis to include lakes from the other lake districts. We expect that the spatial scale of landscape influence will be a function of lake type (seepage vs drainage), hydrologic regime (atmospheric-dominated, groundwater-dominated, or surface-water dominated) and geologic setting.

Collectively, we expect results from these four objectives will enable us to derive a conceptual framework relating lake spatial position to static physical, biological, and chemical attributes as well as the dynamic responses of small inland lakes to regional phenomena in the Upper Great Lakes region.

c. How does spatial connectivity of lakes in a region influence the relation between long-term local and regional diversity?

Communities are assembled through a variety of processes that occur over a wide range of spatial and temporal scales, including both within-lake and regional processes (Ricklefs 1987, Ricklefs and Schluter 1993). The extent of connectivity among habitats will influence the relative effects of regional processes on species composition (Figure 16). Island biogeography theory and related concepts provide an appropriate framework for evaluating connectivity within a lake district (Barbour and Brown 1974, Magnuson 1976). Studies of fish diversity in northern Wisconsin, Finland and Alberta show that both local extinction factors and regional colonization factors are important determinants (Tonn et al. 1990, Magnuson et al. MS). High estimates of species turnover for zooplankton (Arnott et al. MS) and fishes (Magnuson et al. 1994) and low inter-annual variation in species richness suggests that a balance exists between colonization and extinction processes. The importance of colonization will depend on interactions between the form and extent of connectivity and the vagility of the taxa. Within lake districts, avenues of connectivity include animal transport (including humans), streams, and wind and storm events. Plankton, fishes, and snails differ in vagility and utilize different dispersal modes. We propose to investigate the relation between lake connectivity, species vagility and species diversity in several north temperate lake districts.

By characterizing how connectivity and species vagility interact, we can assess how system openness influences long-term local and regional diversity. For zooplankton, our regional assessment of diversity will be extended to a continental scale using an extensive spatial database (Dodson 1992). In regions with isolated lakes and for species with low vagility, we expect long-term local diversity to be low relative to long-term regional diversity. With a highly connected lake district and vagil species, long-term local diversity should approximate the long-term regional diversity. We expect that with less vagil species in highly connected systems, the relation between long-term local diversity and long-term regional diversity will be affected by the taxa's dispersal mechanisms and the manner in which the system is connected.

Biodiversity studies are limited by taxonomic knowledge and resolution. Species composition may change owing to taxonomic effort and revision, rather than actual changes in community composition. We will quantify effects of taxonomic shifts by re-evaluating the taxonomy of species from archived samples.

One measure of connectivity may be the dispersion patterns of exotic species; several exotics are spreading into Wisconsin lakes. Distribution patterns and rates of colonization to new water bodies may indicate the extent of connectivity among lakes. We will reconstruct colonization patterns of exotic species (such as rainbow smelt, Eurasian watermilfoil, and rusty crayfish) to determine how lake connectivity influences the species movement through lake districts.

We will develop a spatially explicit dispersal model to generate time scenarios for distribution patterns via alternate dispersal vectors. We will do this for dispersal from those lakes first known to contain smelt and rusty crayfish in the Northern Highlands Lake District and compare that with present distributions. Future scenarios will be made for comparisons later in the long term. Considering the island nature of the lake district, the process is expected to play out over decades and centuries. These analyses will help guide natural resource agencies to interfere with dispersal where possible, and where not, to prepare for the inevitable. The later alternative will be guided, in part, from our inlake studies of the effects of invaders on lakes (research question 2c).

(4) Develop a regional understanding of lake ecosystems by integrating atmospheric, hydrologic, and terrestrial ecosystem processes.

Rationale and Background. Atmospheric and terrestrial ecosystem processes are important in determining the amount and chemical properties of water in aquatic ecosystems. We plan to examine key linkages between the atmosphere, terrestrial ecosystems, and aquatic systems for small inland lakes in the Great Lakes region. Of particular interest are mechanisms controlling how these interacting systems respond to variations in climate and changes in land cover. Specifically, we will extend our understanding of linkages between atmospheric, aquatic, and terrestrial ecosystems to the regional scale by using a combination of process-based modeling and on site and remote sensing databases.

Proposed Research.

a. How do variations in climate affect lakes?

Regional-scale variations in climate manifest themselves in many ecological phenomena, including changes in lake thermodynamics and biology. We propose new modeling and comparative studies to analyze linkages of climate and lakes at regional scales.

To study changes in lakes resulting from climate variability, we simulate lake thermodynamics with a recent version of the Dynamic Reservoir Simulat