This project is no longer current

Impact of Exotic Invertebrate Invaders on Food Web Structure and Function in the Great Lakes: a Network Analysis Approach

Primary Investigator:

Doran Mason - NOAA GLERL food web schematic

Co-Investigators:

Ann E. Krause, William Taylor, Ken Frank, Andrea Jaeger - Michigan State University
Robert E. Ulanowicz, Ursula Scharler - Chesapeake Biological Laboratory
Anne Clites, Tim Hunter - NOAA GLERL
Stefano Allesina - University of Parma, Italy

Sponsor:

Great Lakes Fishery Commission

Overview

The Great Lakes have recently undergone a second wave of species invasions dominated by exotic invertebrates- Zebra mussels (Dreissena polymorpha), Quagga mussels (D . bugensis), Bythotrephes cederstroemi and Cercopagis pengoi. Unlike previous fish invasions (e.g., sea lamprey and alewife), these invertebrates inserted themselves in the lower trophic levels and thus disruption percolates up through the food web with potential serious consequences to fish communities. This bottom-up effect on the food web eliminates the potential application and modification of traditional fisheries models to quantify and predict direction and magnitude of disruption. What is required is an ecosystem level approach to quantify the disruption and magnitude of disruption on food webs, and the differential effect on various regions of a lake and between lakes.

Early in the 20th century, the fish communities of the Laurentian Great Lakes were profoundly disrupted, and permanently changed, by a wave of vertebrate invaders: sea lamprey, alewife, and rainbow smelt (Eshenroder and Burnham-Curtis 1999). These disruptions were largely focused at the upper end of food webs, and, hence, traditional fishery models could be adapted to quantify the associated impacts. During the last 15 years, the Great Lakes have suffered a second wave of invasions featuring invertebrates: two species of dreissenids-Zebra mussels (Dreissena polymorpha) and Quagga mussels (D. bugensis) -and two predatory cladocerans - Bythotrephes cederstroemi and Cercopagis pengoi (MacIsaac 1999; MacIsaac et al. 1999). These four invertebrates are expected to disrupt, or have already disrupted, the fish communities in the Great Lakes in ways quite different from the earlier wave of vertebrate invaders. In contrast to the vertebrate invasion, invertebrate disruptions start at lower trophic levels and percolate up through the food web with potentially serious consequences for fisheries (Dermott et al. 1999; Ryan et al. 1999; Johannsson et al. 2000; Vanderploeg et al., 2002). The bottom-up effect on the food web eliminates the potential application and modification of traditional fisheries models to quantify and predict direction and magnitude of disruption.

Shuter and Mason (2000) argue that an ecosystem level approach, which incorporates field studies and modeling, is necessary to quantify and eventually predict the impacts of these recent invertebrate invaders. Moreover, factors such as lake morphology, trophic status and the temporal sequencing of invasions, all of which differ among lakes, likely modify the magnitude of the invertebrate impact and the current state of food web change. There are many areas of research currently underway that emphasize field and dynamic modeling approaches. However, many of these studies focus only on a subset of the system and none of them evaluates and quantifies the current state of the entire food web before and after invertebrate invasion. What is required are techniques that synoptically evaluate and quantify the structure and flows in food webs across temporal (seasons and years) and spatial (trophic status, nearshore-offshore) gradients, and how changes in the food web, from invasive invertebrates, disrupt and change these structures and flows.

Network analysis is a technique that allows one to quantify the structure and function of ecosystems by evaluating biomasses and energy flow in a food web. Efficiency with which energy and material is transferred, assimilated, and dissipated conveys significant information about the structure and function of food webs (Ulanowicz and Platt 1985; Baird and Ulanowicz 1989 and 1993; Baird et al. 1991; Ulanowicz and Wulff 1991). Network analysis evaluates these components within a food web context using input/output analysis, trophic and cycle analysis, and information theory to calculate ecosystem properties (see below for details). Thus, changes in fish communities can be linked directly to changes occurring within an ecosystem. Network analysis has been used to compare ecosystems of different size, geographical location, hydrological characteristics, and trophic status (Baird et al. 1991; Ulanowicz and Wulff 1991; Baird and Ulanowicz 1993; Monaco and Ulanowicz 1997). Most recently, arguments have been made for the use of network analysis for quantifying the health and integrity of ecosystems (Ulanowicz 2000) and evaluating the magnitude of stress imposed on an ecosystem (Ulanowicz 1995; Mageau et al. 1998). Related to all of these examples, but absent from the list of applications, is quantifying the ecosystem level impact of exotic invaders and how this impact manifests itself in fish communities.

We propose to construct network models of food webs for the Bay of Quinte Lake Ontario, Oneida Lake, and Lake Michigan across seasons (when data are available), years (pre-invertebrate invasion and post invasion), and trophic status (e.g., Bay of Quinte - western, central, eastern basins; Lake Michigan- nearshore, offshore) to quantify how these invertebrate invasions disrupted lake wide and regional food webs. In additional, our analysis and comparisons within and among seasons, years, regions and lakes should also allow us: (1) to explore various operating hypotheses, acknowledged throughout the Great Lakes Basin, concerning how invaders have changed the ecosystem (e.g., Zebra mussels changing systems to benthic dominated systems), (2) to potential tease apart the effects of nutrient reduction (indirectly determined through changes in primary productivity) and invertebrate invasions on ecosystem structure and function, and (3) to provide a synthesis that may easily be extended to other lakes to understand changes in the ecosystem or to predict potential future changes in other Great Lakes (e.g., Lake Superior).

To achieve our goals and objectives we must upgrade the old DOS code used for network analysis. NETWRK was developed by R.E. Ulanowicz (Ulanowicz and Kay 1991) in the late 1980s (some algorithms where updated in 1999) and is a DOS-based package written in ANSI Standard FORTRAN IV. NETWRK has non-intuitive input structures, data are tedious to enter, programs are command line driven, and results are output as a large report text file. Much of the analysis then comes from extracting the information from the output file, manipulating and finally summarizing the information. The complicated procedures for using the program have restricted its use by others. Heymans and Baird (2000) comment that upgrading NETWRK would “…...be of enormous value to ecosystem analysis′”. Thus, as part of this project, we will upgrade and enhance the NETWRK software to include a graphical user interface, easy and intuitive data input, graphics and analysis, and formatted output. In addition, we will add new features to the program to extend it capabilities and applications. These capabilities may include: uncertainty analysis, improved food web balancing algorithms, system dynamics for evolving systems and potential future states, etc. The product will be a software package available to researchers and resource managers for quantifying the flows and structures in food webs.

Objective

Our overall objective is to quantify the change in food web structures and flows on fish communities in Great Lakes because of invasions by invertebrate species (e.g., Zebra mussels, Quagga mussels, Bythotrephes, and Cercopagis). Specifically, our objectives are:

To develop the windows version of the network analysis program
To construct networks for the three basins of Bay of Quinte, and for Oneida Lake, before and after Zebra mussel invasion.
To construct a network for nearshore and offshore Lake Michigan as a comparison to the Bay of Quinte and Oneida Lake network.
To quantify and contrast trends in ecosystem structure and function in the Bay of Quinte, Oneida Lake, and Lake Michigan with respect to the various phases of invasion from expansion to accommodation.
To determine the differential impact of exotic invertebrate invasion on fish communities relative to the trophic status and nearshore/offshore processes of an ecosystem.

Lake Michigan Food Web Diagram

2005 Plans

Complete Oneida Lake food web, pre- and post-Zebra mussel invasion. Data for pre and post Zebra mussel invasion have been acquired from collaborators at the Cornell University Biological Laboratory on Oneida Lake. These data include species list (phytoplankton, zooplankton, benthos, fish), abundance estimates, and site-specific and speciesspecific estimates for consumption and production. All of these data are required to “wire-up” the food web to perform the network analysis. Once the food web is constructed and balanced, we will evaluate the ecosystem level changes following Zebra mussel invasion of Oneida Lake using network tools (see below).
Begin construction of the Bay of Quinte food web, pre- and post-Zebra mussel invasion. We will acquire the necessary data and estimates for food web members in the upper Bay of Quinte (species list, consumption, production). These data will be available through collaborators construction similar, but less detailed food webs, for this site. Collaborators include individuals from the Ontario Ministry of Natural Resources (Picton, Ontario, Canada) and Dept. Fisheries and Oceans Canada (Burlington, Ontario, Canada). Upon acquiring the necessary data, food webs will be constructed for pre- and post-Zebra mussel invasion in the Bay of Quinte, Ontario. Food web will be of the same species specific detail as that being developed for Oneida Lake for later comparisons.
Make EcoNetwrk available to the scientific community via the World Wide Web. EcoNetwrk is the software we are currently developing that will be made available to all interested in doing ecological network analysis. We have been developing this software in collaborations with Dr. Robert Ulanowicz (original developer of ecological network analysis) over the last couple of years. We will be making the beta version available over the GLERL web site for final testing and use. We have acquired a global list of potential users that are interested in the software. They will be made aware of the available of the software once it is published to the web.
Add additional tools to the EcoNetwrk software The EcoNetwrk software will be made available during the first part of CY05, however, there is still much work required in the software development. The next two components of the software that will require extensive effort deal with the output structure and format of the results. Currently, results are output in a text format where users must extract the important information themselves. For CY05, we will develop a output formats that provide direct access to the results in the form of summaries and graphics.

2004 Accomplishments

Objective 1: To develop the windows version of the network analysis program..EcoNetwk is currently in the Beta testing stage and we expect to have the software available over the web by the end of 2004.
Objective 2: To construct networks for the three basins of Bay of Quinte, and for Oneida Lake, before and after Zebra mussel invasion. Networks for Oneida Lake will be constructed this summer (2004). We expect that the Oneida Lake networks will be completed by the end of the summer. Construction of the Bay of Quinte networks will follow the completion of the Oneida Lake networks. Also, we have learned that sufficient data to construct the networks is available only for the upper basin of the Bay of Quinte. So, we will only be developing networks for pre- and post-dreissenid invasions for the upper basin.
Objective 3: To construct a network for nearshore and offshore Lake Michigan as a comparison to the Bay of Quinte and Oneida Lake network. The Lake Michigan component of the project is nearing completion. The network to represent the food web is based on several major long-term monitoring programs supplemented with other databases appropriate for comparisons. Much of the information used to build the network has the associated literature and other sources referenced and the reference database contains 165 references. The major groups included in the network are phytoplankton, zooplankton, benthic invertebrates, small, primarily benthic-oriented fish, and large, commercially important fish. GLNPO-EPA (Rick Barbiero) has provided their long-term monitoring data on phytoplankton for both time periods. GLSC-USGS (Chuck Madenjian) has provided their long-term data from their bottom trawls for fish assessment. NOAA GLERL (Tom Nalepa) has provided their long-term assessments of benthic invertebrates. NOAA GLERL (Hank Vanderploeg) has also provided zooplankton data for the 1995-99 time period. Because GLERL does not have all of the data needed for the 80-84 time period, Hank Vanderploeg suggested that we use data collected for the Cook Power Plant in the early 80s which he believes is a good comparison. Marlene Evans, the scientist who collected the data, has provided the information we need. GLERL (Steve Pothoven) has also provided data on Mysis for the latter time period. Other sources of information have been from MSU (Jim Bence – chinook, coho, lake trout) and MI-DNR (Phil Schneeberger). We acknowledge that the network does not contain all groups found within the food web (e.g., bacteria, rotifers, ciliates, etc.) nor do we have representation of all species within the groups (e.g., suckers, etc.). These “missing” organisms have not been adequately sampled to provide a long-term comparison of their populations. With this analysis, we will provide a good example of how long-term data sets can be linked together to provide a more holistic understanding of the biotic component of an ecosystem.

Overall, we have about 200 taxa in our network with approximately 3,000 interactions, where an interaction is defined as a predator taxon consuming a prey taxon. Both the taxa and the interactions are weighted. The taxa are weighted by their relative biomass (mg dry weight/m2) for each time period. The interactions are weighted using an index based on the general selectivity of the predator taxon on the prey taxon (high, medium, low), the general overlap in horizontal distribution of the two interacting taxa for both spring and summer, and the general overlap in vertical distribution of the two interacting taxa for both spring and summer and day and night. This index is also weighted by the seasonal presence of each taxon as well. This weighting scheme is a balance between weighting by energy flow, which requires an immense amount of information that is often not available for most taxa, and not using weights at all, which would provide little information on the changes that have occurred in the network structure between the two time periods.

The analysis of this network consists of two components. First, the network will be tested for compartments as demonstrated in the second chapter of Ann Krause’s dissertation (Krause et al. Nature 426:282-285). If we find the network to be significantly compartmentalized, we will then test for the impact that Bythotrephes and Zebra mussels have on the network with the help of Dr. Ken Frank, MSU. Second, we will be assessing the sensitivity of the evenness of the weights on the interactions and taxa before and after the two invasions. We expect to find that the weights on interactions and taxa are sensitive to the addition of these two species and the compartments that contain these two species should show more sensitivity than those that do not contain these species. In addition, we expect that evenness in interactions and taxa will decrease which indicates that a few taxa have become more dominant within the network in the second time period. While we cannot say that the changes in evenness in biomass are directly a result of these invasions, we can at least say what the changes in evenness are relative to the invasions. Since sensitivity is a measure of resistance, we can also examine the patterns of resistance in the network structure relative to the disturbance of these invasions. Finally, we will have quantitative measures of how the network structure has changed over time.

A preliminary analysis of the network was present at the Quantitative Ecosystem Indicators for Fisheries Management International Symposium in Paris, France, March 31-April 3, 2004. In this analysis, the sensitivity of the evenness in interactions was simulated with the hypothetical removal of lake trout, chinook, lake whitefish, and yellow perch. As expected, the compartment that each species was a member was more sensitive to its removal than the other compartments. In addition, it was found that the two species, lake whitefish and yellow perch, that were more central taxa within their compartment (i.e. strongly interacting with many others within their compartment) had a larger impact on the network than the two species, lake trout and chinook, that were more peripheral (not as connected) within their compartment. This analysis was presented as a poster and was nominated for one of the best poster awards at the symposium.
Objective 4. To quantify and contrast trends in ecosystem structure and function in the Bay of Quinte, Oneida Lake, and Lake Michigan with respect to the various phases of invasion from expansion to accommodation. No progress to date. This objective will be achieved following the construction and the analysis of the Lake Michigan, Oneida Lake, and Bay of Quinte networks.
Objective 5. To determine the differential impact of exotic invertebrate invasion on fish communities relative to the trophic status and nearshore/offshore processes of an ecosystem. No progress to date. This objective will be achieved following the construction and the analysis of the Lake Michigan, Oneida Lake, and Bay of Quinte networks.

2003 Accomplishments

We participated in the fall 2002 workshop of the Bay of Quinte and Oneida Lake group (also funded by the GLFC) held at the Cornell Biological Field Station to determine the availability of data for constructing the food webs. A second workshop was held this past spring (2003) at the Canadian Center for Inland Waters (Burlington, ONT), but we were unable to attend due to the SARS outbreak in Toronto and the corresponding government travel restrictions. A full-scale effort on this objective will likely begin in the fall of 2003 when the graduate student on this project starts at Michigan State University.

The development and parameterization of the two preliminary food webs for Lake Michigan, pre and post-Zebra mussel invasion, have been constructed using the tools available through ECOPATH (www.ecopath.org). We presented our preliminary results at two scientific conferences last summer (2002). The first presentation was entitled, “Disruption of an Ecosystem: Changes in ecosystem properties following the establishment of an exotic mussel in Lake Michigan,” and was presented at the American Society of Limnology and Oceanography, Victoria, BC. The second one entitled, “Disruption of Lake Michigan’s ecosystem processes by the invertebrate community: implications for the fish community,” was presented at the American Fisheries Society Conference, Baltimore, MD. In addition, DMM was invited to give a seminar on invasive species and food web disruption at the University of Florida. In preparing the material for these presentations, we identified inherent biases in the food web balancing algorithm that estimates respiration rates for species. (Note that in ECOPATH, respiration can only be estimated from the balancing algorithm and not manually entered). For example, the ratio of respiration to consumption (R/C) cannot exceed the value of 1.0 based on the principles of thermodynamics (i.e., you cannot respire more energy than you can consume in a balanced food web). However, on several occasions the balancing algorithm estimated the R/C > 1.0. Because estimates of respiration are critical to the analysis and that the estimates derived from ECOPATH are suspect, we consider the two Lake Michigan food webs preliminary. The problem associated with ECOPATH is rectified in the new software that we are developing to perform the network analysis.

Despite the prerequisites of having the food webs developed from objectives 1 and 2 prior to working on this objective, we have made theoretical progress. We have used a new technique from the social sciences towards understanding community structure in complex food webs. The technique “Cohesion Analysis” identifies compartments (communities) where interactions within compartments are stronger than interactions between compartments. The technique has four properties that are crucial towards us achieving our objectives of quantifying structural changes in the food web. First, taxa are assigned to non-overlapping compartments that maximize the concentration of interactions within all of the compartments of the food web. Second, the algorithm generally requires no a priori specification of subjective parameters (including number of compartments) and thus removing arbitrary decisions of who belongs where. Third, the algorithm has been applied to extensive simulated data with known compartment assignments allowing one to calibrate the performance of algorithm. Fourth, compartment boundaries can be embedded in a graphical representation of the food web, thus facilitating interpretation. We will use this new technique on the various food webs to explore and quantify how invasive invertebrates have restructured community dynamics within complex food webs of the Great Lakes.

Products

Publications

Krause, A.E, K.A. Frank, D.M. Mason, R.E. Ulanowicz, and W.W. Taylor. 2003. Compartments revealed in food web structure. Nature 426:282-285.

Mason, D.M. 2003. Quantifying the impact of exotic invertebrate invaders on food web structure and function in the Great Lakes: A network analysis approach. Interim Progress Report to the Great Lakes Fisheries Commission- yr 1. 3pp.

Lake Michigan Food Web Poster

Software

EcoNetwrk: A Windows-compatible tool to analyze ecological flow networks

EcoNetwrk is based on the text-based application Netwrk 4.2 by Robert E. Ulanowicz. EcoNetwrk performs all of the analysis of Netwrk 4.2 but in a windows friendly environment. Netwrk 4.2 is copyrighted (1982, 1987, 1998, 1999) and was used with the author’s permission and guidance.

Presentations

Mason, D.M., A. Jaeger, A.E. Krause. Food web disruption: Comparison of Lakes Michigan and Huron. Salmonid Communities in the Great Lakes: Special session- What does the Future Hold for the Laurentian Great Lakes. 134th Annual Meeting of the American Fisheries Society. Madison WI. August 21-26, 2004.

Krause, A.E., K.A. Frank, D.M. Mason, and W.W. Taylor. Changes in the connectivity pattern of a food-web network after biological invasions. Special Session- Integrating Approaches to Connectivity: Landscapes, Patches & Networks. 89th ESA Annual Meeting of the Ecological Society of America, Portland, Oregon. August 1-6, 2004.

Krause, Ann E., W. W. Taylor, and D. M. Mason. 2004. Assessing the potential impact of fishing on compartments in the food web of southeastern Lake Michigan. Quantitative Ecosystem Indicators for Fisheries Management. International Symposium. March 31-April 3, 2004. Paris France.

Mason, D.M. Structure and function in aquatic ecosystems: from food webs to habitat. Florida Atlantic University. March 1, 2004.

Mason, D.M. Application of network analysis to the Great Lakes. National Science Foundation Long Term Ecological Research (LTER)-All Scientist Meeting. Seattle, WA. September 19-21, 2003. (Invited)

Krause, A.E., and D.M. Mason. Invasive species and food web disruption: Example from the Laurentian Great Lakes. 17th Biannual conference of the Estuarine Research Federation. Seattle, WA. September 14-18, 2003. (Invited)

Mason, D.M. Invasive species and food web disruption. University of Florida. February 29, 2003. (Invited)

Krause, A.E., D.M. Mason, K. Franks, and R.E. Ulanowicz. Intuitive compartments: the other half of trophic structure in food webs. 88th Annual Meeting of the Ecological Society of American. Savannah, GA., August 3-8, 2003.

Mason, D.M., and A.E. Krause. Food web structure and energy flow: implications for lake whitefish production in Lake Michigan. Great Lakes Whitefish-Diporeia Workshop. Ann Arbor, MI February 26-27, 2002.

Mason, D.M. Disruption of Food Webs in the Great Lakes. The Great Lakes: Current Issues and Future Challenges. Annual Great Lakes Conference, Agriculture and Natural Resources Week, Michigan State University. March 7, 2002.

Krause, A.E., and D.M. Mason. Disruption of an ecosystem: Changes in ecosystem properties following the establishment of an exotic mussel in Lake Michigan. 2002 summer meeting of the American Association of Limnology and Oceanography, June 10-14, 2002 · Victoria, British Columbia, Canada.

Krause, A.E. and D.M. Mason. Disruption of Lake Michigan’s ecosystem processes by the invertebrate community: implications for the fish community. 132nd AFS Annual Meeting, Baltimore, MD, August 18-22, 2002.

References

Baird, D., J.M. Glade, and R.E. Ulanowicz. 1991. The comparative ecology of six marine ecosystems. Philos. Trans. R. Soc. Lond. 333:15-29.

Baird, D., and R.E. Ulanowicz. 1989. The seasonal dynamics of the Chesapeake Bay ecosystem. Ecol. Monographs 59:329-364.

Baird, D., and R.E. Ulanowicz. 1993. Comparative study on the trophic structure, cycling and ecosystem properties of four tidal estuaries. Mar. Ecol. Prog. Ser. 99:221-237.

Dermott, R,M., M. Munawar, L.Witzel, and P. Ryan. 1999. An assessment of food-web changes in eastern: impact of Dreissena spp. and phosphorous management on rainbow smelt, Osmerus mordax. Pages 367-386 In: State of Lake Erie - Past, Present and Future. M. Munawar, T. Edsall, and I.F. Munawar (eds.) Backhuys, Leiden.

Eshenroder, R.L., and M.K. Burnham-Curtis. 1999. Species succession and sustainability of the Great Lakes fish community. Pages 145-184 In: Great Lakes Fisheries Policy and Management, W.W. Taylor and C.P. Ferreri, eds. Michigan State University Press, East Lansing, Michigan. Heymans, J.J., and D. Baird. 2000. Network anlaysis of the northern Benguela ecosystem by means of NETWRK and ECOPATH. Ecol. Mod. 131:97-119.

Johannsson, O.E., R. Dermott, D.M. Graham, J.A. Dahl, E.S. Millard, D.D. Myles, and J. LeBlanc. 2000. Benthic and pelagic secondary production in Lake Erie after the Invasion of Dreissena spp. with implications for fish production. J. Great Lakes Res. 26:31-54.

MacIsaac, H.J. 1999. Biological invasions in Lake Erie: past, present and future. Pages 305-322 In: State of Lake Erie - Past, Present and Future. M. Munawar, T. Edsall, and I.F. Munawar (eds.) Backhuys, Leiden.

MacIsaac, H.J., I.A. Grigorovich, J.A. Hoyle, N.S. Yan and V.E. Panov. 1999. Invasion of Lake Ontario by the Ponto-Caspian predatory cladoceran Cercopagis pengoi. Can. J. Fish. Aquat. Sci. 56:1-5.

Mageau, M.T., R. Constanza, and R.E. Ulanowicz. 1998. Quantifying the trends expected in developing ecosystems. Ecol. Modelling 112:1-22.

Monaco, M.E., and R.E. Ulanowicz. 1997. Comparative ecosystem trophic structure of three U.S. mid-Atlantic estuaries. Mar. Ecol. Prog. Ser. 161:239-254.

Ryan, P.A., L.D. Witzel, J.R. Paine, M.J. Freeman, M. Hardy, and K.L. Sztramko. 1999. Recent trends in eastern Lake Erie fish stocks within a changing trophic state and food web (1980-1994). Pages 241-290 In: State of Lake Erie - Past, Present and Future. M. Munawar, T. Edsall, and I.F. Munawar (eds.) Backhuys, Leiden.

Shuter, B., and D.M. Mason. 2001. Exotic invertebrates, food-web disruption, and lost fish production: understanding impacts of dreissenid and cladocerans invaders on lower-lakes fish community and forecasting invasion impacts on upper-lakes fish communities. Report to the Great Lakes Fishery Commission. 16pp.

Ulanowicz, R.R. 1995. Trophic flow networks as indicators of ecosystem stress. Pp. 358-368. In: G.A. Polis and K.O. Winemiller (eds), Food webs: integration of patterns and dynamics. Chapman and Hall, NY.

Ulanowicz, R.E. 2000. Toward the measurement of ecological integrity. In: Pimentel, D., L. Westra, and R.F. Noss (eds.) Ecological Integrity. Island Press, Washington:102-113.

Ulanowicz, R.E., and J.J. Kay. 1991. A package for the analysis of ecosystem flow networks. Env Software 6:131-142.

Ulanowicz, R.E., and Platt. 1985. Ecosystem theory for biological oceanography. Can Bull. Fish. Aquatic Sci. 213:1-260.

Ulanowicz, R.E., and Wulff. 1991. Comparing ecosystem structures: the Chesapeake Bay and the Baltic Sea. In: J. Cole, G. Lovett, and S. Findlay (eds.) Comparative analysis of ecosystems, pattern, mechanism, and theories. Springer-Verlag, New York.

Vanderploeg, H.A., T.F. Nalepa, D.J. Jude, E.L. Mills, K.T. Holeck, J.R. Liebig, I.A. Grigorovich,and H. Ojaveer. 2002. Dispersal and emerging ecological impacts of Ponto-Caspian species in the Laurentian Great Lakes. Can. J. Fish. Aquat. Sci 59:1209-1228.