Figure 1. Historical trends of total phosphorus load and phytoplankton density in the western basin of Lake Erie (from Nichols and Hopkins, 1993).
by
Joseph V. DePinto1,2
Victor J. Bierman, Jr.2
Timothy J. Feist2
Jagjit Kaur1
1 Great Lakes Program
Department of Civil, Structural and Environmental Engineering
University at Buffalo
202 Jarvis Hall
Buffalo, NY 14260-4400
2 Limno-Tech, Inc.
501 Avis Drive
Ann Arbor, MI 48108.
May 9, 1999
Abstract
Historically, mathematical modeling of aquatic resources within the Great Lakes has focused on assessment and evaluation of management strategies for individual management issues (e.g., eutrophication, fisheries, toxic chemical exposure). With the advent of the "Ecosystem Approach" for governing and managing the Great Lakes, we have begun to observe and recognize that actions directed toward one management area can impact other problem areas. This realization has led us to a vision for the next generation of aquatic resource models, which incorporates these ecosystem linkages by coupling models of heretofore separate problem domains. In this paper we will present the conceptual framework for a Lake Erie Aquatic Ecosystem Management Model that can address important Lake Erie management issues. The conceptual model contains the aquatic biotic and abiotic components that are necessary to investigate some of the important ecosystem linkages between nutrient dynamics, phytoplankton functional groups, zooplankton, benthic populations (including zebra mussels), forage fish, top predator fish, and bioaccumulative chemicals of concern (such as PCBS). Progress toward this vision is exemplified by results of three ongoing projects: investigation of the effect of nutrient loadings and zebra mussel functioning on phytoplankton dynamics in Saginaw Bay; application of a screening-level model of the potential impact of zebra mussels on cycling and potential for bioaccumulation of PCBS in Lake Erie; and conceptualization of a Lake Michigan coupled pelagic-benthic food web model as part of the Lake Michigan Mass Balance Study. Additional model development, process research and field data acquisition is needed in many areas before this framework can be applied for supporting management decisions in Lake Erie. Some broad areas for research include: upper food web predator-prey interactions, population dynamics and coupling with lower food web; determination of organic carbon flow pathways through the microbial food chain, benthic primary and secondary production and coupling with pelagic food web; dynamic effects of trophic structure and function on contaminant bioaccumulation; dreissenid population dynamics and processing of nutrients and contaminants; and the impact of fine-scale physical processes on ecosystem-level biological interactions in the system. Also, a coherent field program that includes measurement of all relevant stressors (e.g., nutrient and contaminant loads, zebra mussel density distribution), all important all system response variable, and process rates where possible would be very valuable in the site-specific calibration of this ecosystem model to Lake Erie.
Introduction and Problem Statement
Lake Erie has undergone tremendous changes over the past 15-20 years. Most of those changes can be attributed to significant phosphorus loading control measures implemented in the basin. However, some of the recent changes may be the result of the invasion of Dreissenids (zebra and quagga mussels) to the lake. In any event there is considerable interest in developing an understanding of these ecosystem changes and how they are related to management actions on Lake Erie (e.g., nutrient control, toxics load reduction, fish consumption advisories, fish management programs). This interest is evidenced by publications such as the Journal of Great Lakes Research special issue on "Evidence for the Restoration of Lake Erie," 19(2), 1993 and by workshops such as "The Changing Face of the Lower Great Lakes Ecosystems," co-hosted by the New York Sea Grant Institute and the Great Lakes Program at the University at Buffalo (February 5, 1994). Also, the stakeholders within the Lake Erie basin are currently in the process of developing and implementing a Lake Erie Lakewide Management Plan (LaMP), which has the task of identifying beneficial use impairments in Lake Erie as a whole and developing and implementing a management plan for eliminating those impairments. The LaMP process requires the use of an Ecosystem Approach for managing the lake and, therefore, requires a quantitative understanding of the Lake Erie ecosystem structure, function, and response to multiple stressors acting in concert. In particular, the response of Lake Erie ecosystem to changes in loads of nutrients and bioaccumulative contaminant of concern - both manageable stressors - needs to be understood and quantified.
The need for an Ecosystem Approach to managing Lake Erie has led to a vision for the next generation of aquatic resource models, which incorporates important ecosystem linkages by coupling models of heretofore separate issues. The conceptual ecosystem model presented in this paper contains the aquatic biotic and abiotic components that are necessary to investigate some of the important ecosystem linkages between nutrient dynamics, phytoplankton functional groups, zooplankton, benthic populations (including zebra mussels), prey fish, sport fish, and bioaccumulative chemicals of concern (such as PCBS). Progress toward realization of the vision is exemplified by results of three ongoing aquatic ecosystem modeling projects: investigation of the effect of nutrient loadings and zebra mussel functioning on phytoplankton dynamics in Saginaw Bay; application of a screening-level model of the potential impact of zebra mussels in Lake Erie on cycling and potential bioaccumulation of PCBS; and conceptualization of a Lake Michigan Ecosystem Model as part of the Lake Michigan Mass Balance Study.
Historic Perspective - Need for a Lake Erie Aquatic Ecosystem Management Model
In order to develop a quantitative understanding of how management actions affect the structure and function of the Lake Erie ecosystem, it is important to review the history of the lake's responses to changes in both manageable and unmanageable stressors. Because of its morphology and hydrology, Lake Erie is the most susceptible of the Laurentian Great Lakes to cultural eutrophication. The history and description of past and current problems of Lake Erie is quite nicely related in a monograph published by the Ohio Sea Grant College Program (1987). Changes through the 1970s have been well documented in a special issue of the J. Great Lakes Research (Boyce, et al. 1987), and additional changes through the 1980s are reported in another J. Great Lakes Research special issue (Makarewicz and Bertram, 1993). Briefly, beginning with human settlement in the early 1800's, draining of vast coastal wetlands and clear-cutting of forests in rich uplands greatly increased the loading of nutrient-rich sediments to the lake and, in the process, accelerated eutrophication and destroyed fish habitat. The tremendous population and industrial boom during the first half of the twentieth century (population within the basin increased from about 4 million in 1900 to about 14 million in 1980) caused additional stress from municipal and industrial inputs of nutrients and toxic substances. Finally, extensive development of agricultural lands within the basin (approximately 67% of the current land use in the basin) resulted in large pesticide loadings and additional nutrient inputs to the lake. By the 1960s large mats of blue-green algae covered much of the western basin and southern shore of the lake and the central basin hypolimnion exhibited a large area of anoxia as summer progressed.
By the mid to late 1960s the total phosphorus loading to Lake Erie was over 20,000 MT (metric tonnes)/yr. The scientific community had come to a consensus that phosphorus load reduction was the only valid solution to the cultural eutrophication problem in the Great Lakes. With the signing of the Great Lakes Water Quality Agreement in 1972 and its revision in 1978, the governments in the U.S. and Canada implemented a program of phosphorus load reduction that was unprecedented in any region of the world (DePinto, et al. 1986a). Through the insights gained by development and application of nutrient-eutrophication models, a program of total phosphorus load reduction was established for each Great Lake. Target total phosphorus loads were established for each lake (11,000 MT(metric tons)/y was the target load for Lake Erie) on the basis of modeling results to achieve certain water quality goals (7 µg/L and 5 µg/L chlorophyll a for the western and central/eastern basins of Lake Erie, respectively) (Task Group III, 1978). The major load reductions were achieved through phosphate detergent bans and municipal point source controls, which were largely achieved in the Lake Erie basin by the early 1980's. However, it had been determined that the target load for Lake Erie could not be achieved without an additional 30% reduction in nonpoint sources. Best management practices were implemented on agricultural lands within the basin (DePinto, et al. 1986a). By 1992, 34% of the Ohio Lake Erie basin land used for corn and soybeans was being farmed using conservation tillage practices (Ohio Lake Erie Office, 1993). Dolan (1993) reported the IJC estimates of total phosphorus to Lake Erie from 1986-90. He found that point source inputs remained fairly constant between ~2200-2500 MT/y (corresponding to very close to an average of 1 mg/L effluent concentration), but he noted that the nonpoint tributary loading varied from a low of 3837 MT/y in 1988 to a high of 9063 MT/y in 1990. The tributary loads were closely related to the hydrologic runoff for a given year and were almost solely responsible for the ± 3000 MT/y variation around the Lake Erie target load.
In response to the phosphorus load reduction program in the Great Lakes, Lake Erie phosphorus levels and phytoplankton biomass had dropped considerably by the mid-1980s. Bertram (1993) noted that spring isothermal total phosphorus levels in the central basin had dropped from close to 20 µg/L in the 1970s to the target of 10 µg/L by 1987. Makarewicz (1993) concluded that a significant reduction in phytoplankton biomass had occurred in all three basins of Lake Erie between 1970 and the mid-1980's. During his five year study from 1983-87, he found average phytoplankton biomass values of 1.88±0.12 g/m3 (dry wt.), 1.04±0.075 g/m3, and 0.63±0.071 g/m3 for the western, central, and eastern basins, respectively. These values represented a 52-89% reduction in mean basin-weighted algal biomass from 1970 values measured by Munawar and Munawar (1976). Also, slight but delayed improvements in the degree of summer anoxia in the central basin have been observed (Bertram, 1993; Charlton, et al. 1993). El-Shaarawi (1987) confirmed that a statistically significant reduction in chlorophyll a had taken place between 1968 and 1980. He also developed a statistical model that demonstrated a significant relationship between total phosphorus and dissolved oxygen depletion rate, so long as lake level is also included in the regression.
The Lake Erie responses to phosphorus load reductions are very close to those predicted by DiToro and Connolly (1980) in their modeling work used to establish the target load in 1978. Their post-audit examination of the model's long-term predictability indicated that it had reasonably well predicted the lake's response to the phosphorus load reductions through the 1980's (DiToro, et al. 1987). They were remarkably accurate in their prediction of changes in phosphorus and chlorophyll a levels in all three basins and in hypolimnetic oxygen depletion rates in the central basin. This success should give us confidence that we understand the causal chain between phosphorus loading to Lake Erie and it phytoplankton biomass, so that we can potentially identify perturbations to this relationship that might be imposed by the insertion of zebra mussels into the trophic structure.
Through the accelerated eutrophication process in Lake Erie, its fish community of the lake suffered considerably. Since the late 1800's, the combined effect of stream obstruction, wetland draining, extreme pollution, heavy siltation, increased flooding, over fishing and introduction of exotic species has led to the extinction or virtual elimination of several important sport and commercial fish species, including sturgeon, cisco, whitefish, and blue pike (Arnold, 1969; Beeton, 1969). By the 1960s, the walleye population had plummeted and yellow perch produced the major commercial catch in the lake (White, 1987). Also, the rainbow smelt, introduced accidentally in the 1930s, was becoming a major planktivorous species in the lake. However, the major water pollution control efforts in the Great Lakes through the 1970s and early '80s, along with commercial catch restrictions, seems to have led to a rebound in the walleye population in the lake. According to White (1987), only slightly more than 100,000 walleye were caught by Ohio anglers in 1975, but almost 4.5 million were taken in 1986. Today walleye and yellow perch are the dominant game fish in the system, but there is a lake trout restoration program being undertaken in the eastern basin (Coldwater Task Group, 1994). This recovery of walleye and the relative success of the lake trout program is attributed in large part to the major improvement in Lake Erie water quality over the past twenty years. Yellow perch are not doing as well in Lake Erie, with the predominant reason being given as competition from white perch.
Just when the Lake Erie ecosystem seemed to have recovered to its healthiest state in many years, it was hit with the invasion of zebra mussels. Zebra mussels were first discovered in the Great Lakes in Lake St. Clair in June, 1988 (Hebert, et al. 1989); judging from the shell size it was theorized that the introduction had taken place some time in 1986. The first confirmed sighting in Lake Erie was in July, 1988 (O'Neill and MacNeill, 1991; Leach, 1993). By the summer of 1989 extensive colonies of up to 30,000 to 40,000 individuals per square meter were reported in the western basin (OMNR, 1990; Wu and Culver, 1991). It is now known that virtually the entire lake has been infested; even the deeper waters of the eastern basin has not been spared. In 1991, a second species of Dreissena was discovered in Lake Ontario waters (May and Marsden, 1992); originally identified as the "quagga" mussel, it has recently been given the taxonomic identification Dreissena bugensis (Spidle, et al. 1994; Rosenberg and Ludyanskiy, 1994). The quagga has become the dominant dreissenid in the deeper waters of Lake Erie, especially in the eastern basin where it outnumbers the zebra mussel by 14 to 1 at water column depths greater than 20 meters (Mills, et al. 1993).
The literature is replete with impacts associated with zebra mussels clogging water intake pipes in the Great Lakes (LePage, 1993; Kovalak, et al. 1993). Much of the early research on the problem of zebra mussels in the Great Lakes was devoted to controlling their propensity to clog water intake pipes and foul the hulls of vessels. However, more recent emphasis has been placed on the ecological impacts of this invasive genus. Two very good overviews of the zebra mussel problem in North America are presented in a Sea Grant coastal resources fact sheet (O'Neill and MacNeill, 1991) and a NOAA/GLERL report (GLERL and CILER, 1994). A very good collection of the early research findings on the biology, impacts and control of zebra mussels in North America is presented in a book edited by Nalepa and Schloesser (1993).
Many of the impacts of zebra mussels in the Great Lakes result not only from its ability to colonize on hard surfaces but from the mussel's role as a suspension feeder capable of filtering all particles down to a size of 1 um from the water column (Sprung and Rose, 1988). Given a typical filtration capacity of 1 liter/day-individual and typical (for suitable substrate) densities of 103-104/m2, mussels can filter 1-10 meters of water column per day free of phytoplankton and similar sized particles. This filtered material is either assimilated (converted to zebra mussel biomass) or deposited to the bottom substrate as feces or pseudofeces.
The water quality impacts of zebra mussels in the Great Lakes have been well documented. Holland (1993) demonstrated a 100% increase in transparency accompanying an 82-92% decrease in planktonic diatoms after establishment of zebra mussels in western Lake Erie. This greatly increased light penetration could have a significant impact on growth of submerged aquatic vegetation in bays and nearshore waters of the lakes. Nicholls and Hopkins (1993) reported a >90% decrease in phytoplankton densities along the North Shore of Lake Erie over the same period. But the recent changes in phytoplankton density cannot be directly attributed to phosphorus loading reductions. As Figure 1 shows, Nichols and Hopkins (1993) demonstrated that the relationship between phosphorus load reductions and phytoplankton response (as predicted by eutrophication models (DiToro et al., 1987)) that existed in western Lake Erie for the period 1974-1987 no longer applied to data collected subsequent to the zebra mussel invasion. Beginning in 1988 and especially in 1989-90 there was an additional decrease in phytoplankton biomass, with no decrease in phosphorus load. The only logical hypothesis is that the additional loss of algal biomass is due to feeding of zebra mussels on algae. This is another hypothesis that can be tested by an appropriately designed Lake Erie Aquatic Ecosystem Management Model.
In addition to affecting phytoplankton density by filter feeding, zebra mussels may pose an indirect effect on algal density and seasonal patterns. Bierman, et al. (1998) have demonstrated that renewed blue-green algal blooms in Saginaw Bay subsequent to the zebra mussel invasion there are not only the result of selective rejection of blue-greens by zebra mussels but are enhanced by increased sediment-water phosphorus fluxes in late summer. They postulate that an indirect effect of the large zebra mussel induced flux of algae and associated nutrients in spring and early summer is appropriate conditions for a late summer pulse of nutrients at a time when temperatures are favorable for blue-green algal growth. Effler and Siegfried (1998) confirmed that zebra mussels contribute to an alteration in phosphorus dynamics in a system by noting that the zebra mussel invasion of the Oswego River had led to a significant enhancement of soluble reactive phosphorus in the water column without a significant change in the total phosphorus concentration. DePinto, et al. (1986b) demonstrated that phytoplankton growth in Lake Erie in the late summer was largely controlled by recycle of phosphorus in the water column and from the sediments. If dreissenids are altering the spatial and temporal nature of this important process in Lake Erie, then it is crucial to understand the mechanisms involved and to incorporate them into our modeling framework.
In addition to their impact on primary production, zebra mussels can affect secondary production. Griffiths (1993) attributed increases in benthic fauna (amphipods, flatworms, snails, worms) in Lake St. Clair following the zebra mussel infestation to a combination of alteration of benthic habitat structure and deposition of feces and pseudofeces. On the other hand, Nalepa and co-workers found significant reductions in the abundance of native North American freshwater mussels (Unionidae) in both Lake St. Clair (GLERL and CILER, 1994) and western Lake Erie (Nalepa, et al. 1993). Significant impacts on the biological structure and functioning of Saginaw Bay as a result of the zebra mussel invasion were noted in a special section of the Journal of Great Lakes Research (Nalepa and Fahnenstiel (eds.), 1995).
It is evident that the invasion of Dreissenids in Lake Erie has led to an alteration of energy (and carbon) flow through the system and an alteration of nutrient cycling within the system. This has led not only to a change in both pelagic and benthic community structure and function but also to a change in the nutrient-productivity relationships relative to pre-invasion conditions. Only an ecosystem modeling framework that integrates all of the various process interactions and system feedback mechanisms can forecast the quantitative response of Lake Erie to management actions relative to nutrient loads and fisheries.
The Dreissenid-induced alteration of carbon flow through the Lake Erie system may also be having an impact on bioaccumulation of contaminants like PCBS in top predator fish such as walleye. Again, this impact must be viewed in the context of how the lake behaved prior to the Dreissenid invasion. Like the other Great Lakes, the loadings of persistent, bioaccumulative chemicals (like PCBS) to Lake Erie peaked in the early 1970s and dropped off in response to use bans and source controls implemented through the 70s and early 80s. Lake Erie, however, did not appear to suffer nearly as much as Lake Ontario in terms of bioaccumulation of persistent chemicals like PCBS in top predator fish. Fish consumption advisories were not nearly as restrictive in Lake Erie. Although the loadings of PCBS to Lake Erie were comparable to Lake Ontario on a volumetric basis, the top predator fish did not appear to accumulate PCB nearly as much. Rathke and McRae (1989) used 1985 data from Lake Erie to demonstrate a typical PCB biomagnification from approximately 0.015 ppm in net plankton to 0.2 ppm in smelt to 1.5 ppm wet weight in walleye. At the same time, PCBS in lake trout in Lake Ontario - PCB loading to the two lakes in the mid-1980s was about the same - was in the range of 4-6 ppm wet weight (DeVault, et al. 1996; Heustis, et al. 1996).
It has been hypothesized that the higher levels of suspended solids in Lake Erie served, through adsorption and settling, as a mechanism for reducing bioaccumulation through the grazing food chain. If this is true, then it is possible that the Dreissenid invasion in the late 1980s may have led to an alteration of PCB bioaccumulation in Lake Erie. Indeed, the trend data of
PCB levels in Lake Erie walleye (Figure 2 (after Devault, et al. 1996)) show a significant break in the first-order trend observed prior to 1986 relative to the post-Dreissenid invasion period. It is possible that, due to the decrease in water column suspended solids concentrations caused by Dreissenids, there has been a change in the phase distribution of the PCBS remaining in the water column toward a higher fraction of the dissolved (and therefore bioavailable) fraction. This altered cycling and phase distribution of PCBS in Lake Erie may be responsible for a change in the quantitative relationship between PCB loading to Lake Erie and observed bioaccumulation of PCBS in the food chain. We know from analyses such as the one conducted by Rowan and Rasmussen (1992), who compiled PCB and DDT body burden data from all of the Great Lakes (including the three basins of Lake Erie), that exposure concentrations alone cannot explain most of the between-basin variability. In conducting a multiple regression analysis of the relationship between these data an basin-specific properties, they concluded that ecological attributes such as fish lipid content, organism trophic level, and structure of the food chain had to be considered. Findings such as these also point to the potential importance of zebra mussels in PCB cycling and bioaccumulation in Lake Erie. Testing this hypothesis is another important goal of the proposed Lake Erie Aquatic Ecosystem Management Model and process research necessary for its development.
Management Questions for Lake Erie
Presented above is some evidence that recent ecological changes in Lake Erie are the result of a combination of phosphorus load reductions, the dreissenid invasion, and loss of fish fhabitat. Among the most significant potential impacts are:
There are certainly other intermediate effects, but these are the important end-points linked to phosphorus loading, PCB inputs, and the dreissenid invasion. In fact, a key overall question for the Lake Erie ecosystem might be "What would be the state of Lake Erie today if there had not been a dreissenid invasion?" In light of the happenings in the lake since the late 1980s, this question can only be answered by constructing an ecosystem model which can numerically remove dreissenids from the system and simulate its progression from 1985-1999 as if the invasion had not taken place.
Given the understanding of how dreissenids have impacted the Lake Erie ecosystem, managers can then choose from a number of possible management actions that are available for addressing the ecological impacts of concern. Among the actions that can be implemented either alone or in combination are:
Each of these management actions has potential impacts (both positive and negative) throughout the ecosystem. It is the goal of the Lake Erie Aquatic Ecosystem Management Model to identify and quantify those impacts.
Conceptualization of Lake Erie Aquatic Ecosystem Management Model
Mathematical models have the ability to synthesize and integrate information regarding the interaction among components of complex ecosystems. This capability enables resource managers to identify how decisions made in one management area will affect the system with regard to another. Of course, this goal can only be achieved by developing a modeling framework that can quantify the impacts of multiple stressors, both natural and anthropogenic, acting in concert on key ecosystem components to generate multiple response endpoints. The principal stressors and system responses to be quantitatively linked by the Lake Erie Aquatic Ecosystem Management Model are depicted in Figure 3. The system responses in represent quantitative measures of ecological impacts mentioned above that will result from a specified set of ecosystem stressors, some of which can be affected by the management actions listed in the previous section.
The general strategy for constructing the Lake Erie Aquatic Ecosystem Management Model will be to formulate coupled modules (or submodels) that can be activated as required to develop a system-specific application. Each of these modules will have "generic" components that are potentially important but may take on different attributes in different ecosystems. For example, there will be a benthic suspension feeder that may be represented by the attributes of dreissenids in Lake Erie. At the upper end of the food web, the system will contain a number of components and age classes for prey fish and predator fish; these constituents can also be parameterized to represent particular species on a site-specific application basis.
The modular structure for a Lake Erie Aquatic Ecosystem Management Model that integrates the management issues mentioned above is depicted in Figure 4. Also shown in this diagram are the various system stressors and the linkages between modules (arrows) that are needed to represent the most important ecosystem feedbacks or homeostatic processes. For example, unusually successful walleye recruitment can have a top-down effect on the lower food web; an increase in predator fish can cause a decrease in prey fish, which in turn will reduce the predation pressure on zooplankton.
It should be noted, however, that integration of single-issue models as depicted in Figure 4 is not trivial. While the Great Lakes has a wealth of experience and success in developing and applying the single-issue models, relatively little work has been conducted on coupling these models into an ecosystem analysis framework. The experience gained through our initial ecosystem modeling efforts (briefly discussed below) has made us aware of the many scale dependency and kinetic process linkage issues that are involved in formulating coupling aquatic ecosystem models by coupling what we have learned and developed through our single-issue modeling. It has also made us confident that the proposed conceptual approach to developing models of complex ecosystem interactions can be accomplished.
The first step in converting the overall framework depicted in Figure 4 into a conceptual model that can address the management issues of concern is to determine the ecosystem components (state variables) that need to be included. In establishing the key state variables in the model, one must also consider the "currency" of the model. For example, in most eutrophication models the "currency" involves a measure of the biomass in various biota compartments (i.e., chlorophyll a or dry weight for phytoplankton) as well as the nutrient concentration in various biotic and abiotic compartments that are relevant to the problem definition. In an aquatic ecosystem model such as the one conceived for Lake Erie, one needs a common currency for biomass throughout the food web in addition to tracking both the nutrient content (phosphorus, nitrogen, and silica) and the PCB content in all important biotic and abiotic compartments of the system. We believe that the best biomass measure for this model is organic carbon. It has several advantages for a single currency to be used across the entire food web. First, it relates reasonably well with bioenergetics analysis of organisms. It also can be used as an aggregate biomass measure for the lower food web (plankton), while it can be used to represent average size for individual organisms higher up the food chain; then biomass of a given species becomes the product of individual carbon level and species density. This is important because some processes impact individual size while others (like reproduction and recruitment) operate on numbers of a given species or age class within a species. Two other important reasons for using organic carbon as the common biomass currency for this model are that it greatly facilitates the mass balance modeling of hydrophobic organic compounds like PCBS and that it also facilitates cycling of nutrients through organisms based on carbon to nutrient stoichiometries.
Combining the integrated module configuration shown in Figure 4 with the carbon biomass concept, one can develop a conceptual model for carbon flow through the Lake Erie aquatic ecosystem. This carbon flow diagram is depicted in Figure 5. Each box in Figure 5 represents a functional reservoir of organic carbon within the sediment-water system. Each of these reservoirs has basic characteristics with respect to how they process carbon and how they are coupled with other carbon reservoirs within the system. As indicated in the diagram it is possible for a given functional reservoir to contain more than one species or more than one age class of a given species. In this way an ecosystem hierarchy is developed where there are certain higher-level functions that are characteristic of all compartments within a reservoir (e.g., all phytoplankton contain chlorophyll a, obtain energy and synthesize new biomass by primary production, settle through the water column, and are potentially grazed on by zooplankton) but different classes of species within the phytoplankton reservoir may have specific properties or behaviors that make differentiation necessary within a specific problem context. For example, if one wants to investigate the implications of blue-green algae not being grazed by zooplankton or zebra mussels, then the phytoplankton reservoir must comprise of at least two classes (or objects in programming terminology), one for blue-greens and one for all other phytoplankton. If concerns such as the effects of silica, nitrogen, phosphorus, light, and temperature on seasonal succession are important then a finer differentiation such as indicated in the phytoplankton reservoir in Figure 5 may be warranted. Similar arguments can be made for having multiple classes within each of the other reservoirs in Figure 5.
It should be noted that Figure 5 presents only the carbon flow diagram, however it does contain all of the functional reservoirs in the prototype Lake Erie Aquatic Ecosystem Management Model. The fate and transport of nutrients and PCBS through this system is not shown because of space limitations. However, this carbon flow conceptualization can serve as a framework for building nutrient and PCB mass balance equations into the overall framework. In fact some of this model coupling has already been accomplished in the projects discussed below.
Progress Toward an Operational Aquatic Ecosystem Model for the Great Lakes
The Principal Investigators are not only recognized leaders in the Great Lakes modeling community but are at the forefront of research in the development and application of aquatic ecosystem models. Bierman, DePinto and Feist are working on a project, funded by EPA-GLNPO, to develop a nutrient-phytoplankton-zebra mussel-PCB mass balance model for Saginaw Bay. This ecosystem model is being formulated by coupling previously developed (through EPA funding) individual-issue models: a nutrient cycling, multi-class phytoplankton model (Bierman and Dolan, 1981; Bierman, et al. 1984; Bierman, et al. 1986a, 1986b); coupling a zebra mussel bioenergetics model to the phytoplankton model (LTI 1995, 1997); and finally, with the ongoing project, coupling of a PCB mass balance model with the coupled phytoplankton-zebra mussel model. DePinto previously conducted a screening-level modeling analysis of the impact of zebra mussels in Lake Erie on the cycling and bioaccumulation potential PCBS (DePinto and Narayanan, 1997; DePinto, et al. 1997). Bierman, Feist and DePinto are also working on the Lake Michigan Mass Balance Study to develop an enhanced carbon mass balance model that can provide an accurate simulation of organic carbon dynamics in Lake Michigan for use as input to the hydrophobic organic carbon mass balance model being developed for that system. DePinto has also developed a whole-lake annual average nutrient-trophic transfer model (Jain and DePinto, 1996) to investigate the trade-offs between phosphorus control and fish stocking levels on salmonid fish production in Lake Ontario.
Research and Data Needs
Development of a modeling framework such as the one presented above to the point where it can be used confidently for supporting management decisions in Lake Erie would benefit from further research and monitoring in three broad areas: model development, process research, and field measurements. There is a symbiotic relationship among these three components that must be considered in any aquatic ecosystem assessment. Models can provide great insight and make projections, but only with the support of monitoring, which provides model inputs and credibility, and research, which provides understanding and parameterization for model development.
Modeling research takes the form of developing new modules for the proposed Lake Erie integrated framework and coupling them with existing modules at the appropriate time and space scales (see Figure 4). There are several modules and linkages that require significant development work in the construction of a Lake Erie Aquatic Ecosystem Management Model depicted in Figure 4:
Development and parameterization of the modeling framework described above would benefit from process experimental research in several areas. Among the highest priority areas are:
Last but certainly not least in importance is the collection of field data to provide inputs for models and data for comparison with model output. In general, this effort should be as coordinated and coherent as possible. It should be conducted at appropriate spatial and temporal scales relative to modeling needs; and it should be designed so that a less intensive subset of the field data can be collected on a long-term basis to provide valuable trend information. Field monitoring falls into three categories: monitoring of external stressors; monitoring of models output variable (i.e., system response variables), and field monitoring of the rate and extent of processes included in the conceptual model of the system. With regard to external stressors, such measurements as sediment, nutrients and bioaccumulative chemicals of concern loadings is essential. Other stressor related measurements include dreissenid densities, hydrometeorological conditions, and fish stocking and harvesting. Also, routine in situ monitoring of all response variables (model output) upon which management decisions are made should be conducted at space and time scales appropriate to the variable of interest. Typical response variables include: algal biomass and class composition; fish biomass, species composition, age distribution and condition; nutrient levels (including organic carbon) in water, sediments and biota; bioaccumulative chemical concentrations in water, sediments and biota; dissolved oxygen and particulate matter spatial and temporal profiles. Finally, field measurement of the rate and extent of process incorporated into the model can provide a very valuable set of additional constraints on calibration and field confirmation of the model. This component of the field program might include measurement of processes such as primary production, zooplankton production, fish growth rates, bottom sediment resuspension rates, areal hypolimnetic oxygen depletion rate, and air-water mass transfer rates.
Mathematical mass balance models provide an excellent means of synthesizing what we know about the behavior of an aquatic system. The next generation of such models should focus on integrating the components of the system that are important to the range of management areas being addressed for that system. With very important management questions regarding nutrient controls, toxic chemical exposure, exotic species invasions, and fisheries management, Lake Erie provides an excellent ecosystem within which to demonstrate the feasibility and utility of this next generation of models.
List of Figures
References
Arnold, D.E. 1969. The ecological decline of Lake Erie. New York Fish and Game Journal. 16:27-45.
Beeton, A.M. 1969. Changes in the environment of biota of the St. Lawrence Great Lakes. In Eutrophication: Causes, Consequences, Correctives. pp. 150-157.
Bertram, P. 1993. Total phosphorus and dissolved oxygen trends in the central basin of Lake Erie. J. Great Lakes Res. 19:224-236.
Bierman, V.J. and D.M. Dolan. 1981. Modeling of phytoplankton-nutrient dynamics in Saginaw Bay, Lake Huron. J. Great Lakes Res. 7(4):409-439.
Bierman, V.J. and D.M. Dolan. 1986a. Modeling of phytoplankton in Saginaw Bay: I. Calibration phase. Jour. Environ. Eng., ASCE. 112(2):400-414.
Bierman, V.J. and D.M. Dolan. 1986b. Modeling of phytoplankton in Saginaw Bay: II. Post-audit phase. Jour. Environ. Eng., ASCE. 112(2):415-429.
Bierman, V.J., Jr., D.M. Dolan, R. Kasprzyk and J.L. Clark. 1984. Retrospective analysis of the response of Saginaw Bay, Lake Huron, to reductions in phosphorus loadings. Environmental Science and Technology. 18(1): 23-31.
Bierman, V.J., Jr., D.W. Dilks, T.J. Feist, J.V. DePinto, and R.G. Kreis, Jr. 1998. Coupled phytoplankton-zebra mussel model for Saginaw Bay, Lake Huron. Proceedings of Workshop on Aquatic Ecosystem Modeling and Assessment Techniques for Application within the U.S. Army Corps of Engineers, Paper No. EL-98-1, U.S. Army Engineer Waterways Experiment Station, Vicksburg, MS. pp. 43-67.
Boyce, F.M., M.N. Charlton, D. Rathke, C.H. Mortimer, and J.R. Bennett (Eds.). Lake Erie Binational Study: 1979-80. Special Issue of J. Great Lakes Res. 13(4):405-840.
Charlton, M.N., J.E. Milne, W.G. Booth, and F. Chiocchio. 1993. Lake Erie offshore in 1990: restoration and resilience in the central basin. J. Great Lakes Res. 19:291-309.
DePinto, J.V. and R. Narayanan. 1997. What other ecosystem changes have zebra mussels caused in Lake Erie: Potential bioavailability of PCB's. Great Lakes Research Review. 3(1): 1-8.
DePinto, J.V., R. Narayanan, and V.J. Bierman, Jr. "The effects of the zebra mussel invasion of Lake Erie on the Transport and Phase Distribution of PCBs." Contributed paper at 7th International Zebra Mussel and Aquatic Nuisance Species Conference, New Orleans, LA (January 28-31, 1997).
DePinto, J.V., T.C. Young, and L.M. McIlroy. 1986a. Great Lakes water quality improvement. Environ. Sci. Technol. 20(8):752-759.
DePinto, J.V., T.C. Young and D.K. Salisbury. 1986b. Impact of Phosphorus Availability on Modeling Phytoplankton Dynamics. Dutch Hydrobiological Bulletin 20(1/2):225-243.
Devault, D.S., R. Hesselberg, P.W. Rodgers, and T.J. Feist. 1996. Contaminant trends in lake trout and walleye from the Laurentian Great Lakes. J. Great Lakes Res. 22(4):884-895.
DiToro, D.M. and J.P. Connolly. 1980. Mathematical Models of Water Quality in Large Lakes, Part 2: Lake Erie. Report No. EPA-600/3-80-065, report to Large Lakes Research Station, ERL-Duluth, Grosse Ile, MI 48138.
DiToro, D.M., N.A. Thomas, C.E. Herdendorf, R.P Winfield, and J.P. Connolly. 1987. A post audit of a Lake Erie eutrophication model. J. Great Lakes Res. 13(4):801-825.
Dolan, D. 1993. Point source loadings of phosphorus to Lake Erie: 1986-1990. J. Great Lakes Res. 19(2):212-223.
Effler, S.W. and C. Siegfried. 1998. Tributary water quality feedback from the spread of zebra mussels: Oswego River, New York. J. Great Lakes Res. 24(2):453-463.
El-Shaarawi, A.H. 1987. Water quality changes in Lake Erie, 1968-1980. J. Great Lakes Res. 13(4):674-683.
Great Lakes Environmental Research Lab/NOAA and U. of Michigan Cooperative Institute for Limnology and Ecosystems Research (CILER). 1994. The Ecological Approach to the Zebra Mussel Infestation in the Great Lakes. Publications Office, NOAA, Ann Arbor, MI.
Griffiths, R. W. 1993. Effects of zebra mussels (Dreissena polymorpha) on the benthic fauna of Lake St. Clair. In Nalepa and Schloesser (eds), Zebra mussels: Biology, Impacts, and Control. Lewis Publishers, CRC Press, Inc., Boca Raton, FL. pp. 415-437.
Hebert, P.D.N., B.W. Manchester, and G.L. Mackie. 1989. Ecological and genetic studies on Dreissena polymorpha (Pallas): a new mollusc in the Great Lakes. Can. J. Fish. and Aquatic Sci. 46(9):1587-1591.
Huestis, S. Y., M. R. Servos, D. M. Whittle, and D. G. Dixon, Temporal and age-related trends in levels of polychlorinated biphenyl congeners and organochlorine contaminants in Lake Ontario lake trout, J. Great lakes Res., 22(2):310-330, 1996.
Jain, R. and J.V. DePinto. 1996. Modeling as a tool to manage ecosystems under multiple stresses: An application to Lake Ontario. Journal of Aquatic Ecosystem Health. 5: 23-40.
Kovalak, W.P., G.D. Longton, and R.D. Smithee. 1993. Infestation of power plant water systems by the zebra mussel (Dreissena polymorpha Pallas). In Nalepa and Schloesser (eds), Zebra mussels: Biology, Impacts, and Control. Lewis Publishers, CRC Press, Inc., Boca Raton, FL. pp. 359-380.
Leach, J.H. 1993. Impacts of the zebra mussel, Dreissena polymorpha, on water quality and spawning reefs in western Lake Erie. In Nalepa and Schloesser (eds), Zebra mussels: Biology, Impacts, and Control. Lewis Publishers, CRC Press, Inc., Boca Raton, FL. pp.381-398.
LePage, W.L. 1993. The impact of Dreissena polymorpha on waterworks operations at Monroe, Michigan: A Case History. In Nalepa and Schloesser (eds), Zebra mussels: Biology, Impacts, and Control. Lewis Publishers, CRC Press, Inc., Boca Raton, FL. pp.333-358.Limno-Tech, Inc. 1995. A preliminary ecosystem modeling study of zebra mussels (Dreissena polymorpha) in Saginaw Bay, Lake Huron. Report prepared for Great Lakes Program, University at Buffalo under subcontract No. S015-92 of Cooperative Agreement CR-818560 with the U.S. EPA, ERL-Duluth, LLRRB, Grosse Ile, MI.
Limno-Tech, Inc. 1995. A Preliminary Ecosystem Modeling Study of Zebra Mussels (Dreissena polymorpha) in Saginaw Bay. Project report for subcontract S015-92 from Great Lakes Program, University at Buffalo, under Cooperative Agreement CR-818560 with the U.S. Environmental Protection Agency, ERL-Duluth, Large Lakes Research Station, Grosse Ile, Michigan.
Limno-Tech, Inc. 1997. Application of a coupled primary production - exotic species model for Saginaw Bay, Lake Huron. Report prepared for EPA contract No. ERLD-95-0219, EPA-ORD, National Health and Environmental Effects Lab, Mid-Continent Ecology Division - Duluth, Large Lakes Research Station, Grosse Ile, MI.
Makarewicz, J.C. 1993. Phytoplankton biomass and its species composition in Lake Erie, 1970 to 1987. J. Great Lakes Res. 19(2):258-274.
Makarewicz, J.C. and P.E. Bertram (Eds.). 1993. Evidence for the Restoration of Lake Erie. Special Issue of J. Great Lakes Res. 19(2):197-309.
May, B. and J.E. Marsden. 1992. Genetic identification and implicationsof another invasive species of dreissenid mussel in the Great Lakes. Can. J. Fish. Aquat. Sci. 49:1501-1506.
Mills, E.L., R.M. Dermott, E.F. Roseman, D. Dustin, E. Melina, D.B. Conn, and A.P. Spidle. 1993. Colonization, ecology, and population structure of the "quagga" mussel (Bivalvia: Dressenidae) in the lower Great Lakes. Can. J. Fish. Aquat. Sci. 50:2305-2314.
Nalepa, T.F., B.A. Manny, J.C. Roth, S.C. Mozley, and D.W. Schloesser. 1993. Long-term decline in freshwater mussels (Bivalvia: Unionidae) of the western basin of Lake Erie. Journal of Great Lakes Research. 17(2):214-219.
Nalepa, T.F. and D.W. Schloesser (Eds.). 1993. Zebra mussels: Biology, Impacts, and Control. Lewis Publishers, CRC Press, Inc., Boca Raton, FL.
Nalepa, T.F. and G.L. Fahnenstiel. 1995. Dreissena polymorpha in the Saginaw Bay, Lake Huron Ecosystem: Overview and Perspective. Journal of Great Lakes Research. 21(4):411-416. (Preface to special section of that issue on pages 411-573)
Nicholls, K.H. and Hopkins, G.J., 1993. "Recent Changes in Lake Erie (North Shore) Phytoplankton: Cumulative Impacts of Phosphorus Loading Reductions and the Zebra Mussel Introduction," Journal of Great Lakes Research, Vol. 19, No. 4, pp. 637-647.
Ohio Lake Erie Office. 1993. State of the Lake - 1992 Governor's Report on Lake Erie. Ohio Lake Erie Commission, Columbus, OH.
Ohio Sea Grant College Program. 1987, reprinted in 1993. The Great Lake Erie, R.A.Fortner and V.J. Mayer (eds), The Ohio State University Research Foundation , Columbus, OH.
O'Neill, C.R. and D.B. MacNeill. 1991. The Zebra Mussel (Dreissena Polymorpha): An Unwelcome North American Invader. New York Sea Grant/Cornell Cooperative Extension/State University of New York, Coastal Resources Fact Sheet, New York Sea Grant Office, SUNY-College at Brockport, Brockport, NY.
Rathke, D.E. and G. McRae. 1989. 1987 Report on Great Lakes Water Quality - Appendix B: Great Lakes Surveillance, Vol. I. Great Lakes Water Quality Board, IJC, Windsor, Ontario.
Rosenberg, G. and M.L. Ludyanskiy. 1994. A nomenclature review of Dreissena (Bivalvia: Dreissendae), with identification of thequagga mussel as Dreissena bugensis. Can. J. Fish. Aquat. Sci. 51:1474-1484.
Rowan, D.J. and J.B. Rasmussen. 1992. J.Great Lakes Res. 18(4):724-741.
Spidle, A.P., J.E. Marsden, and B. May. 1994. Identification of the Great Lakes quagga mussel as Dreissena bugensis from the Dneiper River, Ukraine on the basis of allozyme variation. Can. J. Fish. Aquat. Sci. 51:1485-1489.
Sprung, M. and Rose, U., 1988. "Influence of Food Size and Food Quantity on the Feeding of the Mussel Dreissena polymorpha," Oecologia, Vol. 77, pp. 526-532.
Task Group III (J.R. Vallentyne and N.A. Thomas, co-chairs). 1978. Fifth Year Review of Canada-U.S. Great Lakes Water Quality Agreement. IJC-Regional Office, Windsor, Ontario.
White, A.M. 1987. History of changes in the Lake Erie Fishery. In The Great Lake Erie, R.A.Fortner and V.J. Mayer (eds), Ohio Sea Grant College Program, Columbus, OH. pp. 116-126.
Wu, L. and Culver, D.A., 1991. "Zooplankton Grazing and Phytoplankton Abundance: An Assessment Before and After Invasion of Dreissena polymorpha," Journal of Great Lakes Research, Vol. 17, No. 4, pp. 425-436.