Study 236102 - 1
ATTACHMENT A
Michigan Department of Natural Resources Renewed: 2009-2010
Study 236102New Study: 2002-03
Name of Study:Using population and community dynamic models and quantitative fisheries analysis to promote improved fisheries management in the Great Lakes.
- Problem:The Great Lakes fisheries are, in part, supported by release of hatchery-reared fish, and are influenced by fishing regulations and abiotic factors.Fishery managers are faced with decisions regarding total numbers, species composition and locations to plant fish, and the means and extent to which harvest needs to be limited.Quantitative information on how these decisions are likely to influence fishing success, including information on uncertainty and uncontrollable variability, along with and other measures of ecosystem performance, would allow improved management decisions.
- Objective:Develop a research, education, and outreach/service program aimed at improving quantitative fishery stock assessments and analysis of population/community dynamics with a view toward an enhanced understanding of the dynamics of the Great Lakes system.
- Justification:Fishery management questions are often, in their essence, quantitative ones.These include questions such as “how many fish should be harvested?”, “what sizes should be harvested?”, “what is the probability of a stock collapse?”, “how many years will it take to rebuild this population?”, and so on.Answering these kinds of question usually requires the formulation of a mathematical model, estimation of the parameters of the model, and use of the model to project potential consequences. While increases in computer capacity and improvements in software make increasingly more complex assessment models possible, our understanding of how the available quantitative assessment methods perform as well as the quantitative expertise to apply these methods and make use of the results has not kept apace.Enhancement of the state-of-the-art in our understanding of how these quantitative methods perform and in the level of expertise available to apply the methods would provide substantial fishery benefits. In some cases, identifying and collecting data that fill a key gap in our understanding is a key part of this process, which can provide tremendous benefits.
D.Status:Michigan’s extensive aquatic resources provide sport and commercial fishing opportunities and form an important basis for outdoor recreation in the state. Recreational angling, in particular, has a significant economic impact.With nearly 29 million days spent by recreational anglers in the state during 1996, Michigan ranked fifth nationally in recreational fishing activity (U.S. Department of Interior, Fish and Wildlife Service and U.S. Department of Commerce, Bureau of the Census 1997).One out of 12 residents owns a boat and 52% use their boats for fishing (Garling et al. 1995), and total annual expenditures associated with angling exceed $1.5 billion (U.S. Department of Interior, Fish and Wildlife Service and U.S. Department of Commerce, Bureau of the Census 1997).
Fishing opportunities on the Great Lakes are critical.Over 21% of recreational fishing in Michigan takes place on the Great Lakes.The 38,075 square miles of Great Lakes waters in Michigan is 43% of the total and more anglers fish on Michigan’s waters of the Great Lakes than on those of any other state. With 12 state-operated and 58 municipal harbors on the Great Lakes, Great Lakes boat anglers are never more than 15 shoreline miles from a harbor in Michigan waters (Garling et al. 1995).There are also many public access boat launching sites and marinas.There are approximately 500 charter fishing operators working in Michigan’s waters.Commercial fishing on Michigan’s waters of the Great Lakes produced approximately 15 million pounds with a dockside value of nearly $9 million during the 1990s.An additional $9 million was generated by activities associated with marketing and processing.Fish produced by commercial fishing provide an important connection to the Great Lakes for many consumers and is part of many participants’ cultural heritage.In addition this commercial harvest has exceeded the amount of sport harvest (Garling et al. 1995).The rights of Native Americans to harvest fish from the Great Lakes are protected by treaty and supported by court settlements and a substantial Native American commercial harvest (approximately equal to sport harvest combined with state licensed commercial harvest) seems likely to continue in the future.
Michigan’s Great Lakes fishery resources are intensively managed and future benefits depend upon this management.The current management arena has resulted from substantial perturbations in the past (e.g., Bence and Smith 1999).Lake trout, once the dominant predator in four of the Great Lakes, was driven extinct or nearly extinct in these lakes through the combined actions of commercial fishing and sea lamprey.Walleye was the top predator in Lake Erie and in SaginawBay and its populations declined both from the effects of fishing and from habitat loss associated with pollution.The absence of top predators may have allowed the exotic alewife to invade the system.This small planktivorous fish has adversely impacted many native fish species through predation on eggs and larvae and competition for food.In addition, this species is prone to periodic die-offs and large die-offs can litter beaches with dead fish. A method to successfully control sea lamprey was first implemented in the late 1950s.This method relies on periodic chemical treatment of streams that are nursery habitat for larval sea lamprey.The sea lamprey control program still relies on chemical control although alternative control methods are being explored.With a method to control sea lamprey in place, fishery agencies restocked the Great Lakes with salmon and trout.These predators have reduced alewife populations, and attract much sport fishing effort.However, to a large extent existing fish stocks depend upon ongoing stocking.Only in Lake Superior have large self-sustaining populations of lake trout recovered.In Lakes Michigan and Huron almost all the lake trout continue to be supplied by hatcheries.Although there is significant recruitment of wild born Chinook salmon and rainbow trout, stocking still more than matches their wild production.
A variety of factors influence Great Lakes fish populations and many of these are outside direct control of fishery managers. Managers can, however, significantly impact Great Lakes fish populations and the ecosystem through regulations that influence harvest amount and composition, the number of hatchery-reared fish that are stocked into the lakes and control efforts for parasitic sea lamprey.Management is complicated because of the need for co-management with other jurisdictions and by the competing desires of different user groups.Because many important sport and commercial fish move extensively and fish populations do not follow political boundaries, this management involves significant interactions among the Great Lakes states, the U.S. and Canadian federal governments, and Native American tribes with treaty fishing rights. Even within Michigan, some users desire greater emphasis on lake trout restoration and native species, while others support greater emphasis on sport or commercial fishing.Effective fishery management and balancing of the many necessary trade-offs requires an ability to predict likely consequences of management choices.Population and community models provide one means to this end and can form critical components of more expansive ecosystem models.The usefulness of these models, however, depends critically upon how well they describe the system and how well their parameters (that determine fish survival and recruitment over time) are estimated.Consequently, this project proposes to develop and parameterize population and community models, and to promote such activities by others, relevant to selected Great Lakes fish populations and to explore more generally the performance of methods for fitting the models and the performance of management strategies based upon the resulting model predictions.In some cases this project may collect or direct the collection of key data needed to improve the models.The results will be useful to the Michigan Department of Natural resources, other state and federal agencies through the Great Lakes Fishery Commission mediated LakeCommittee process, and tribal fishery management agencies (e.g., the Chippewa-Ottawa Resource Authority) in their ongoing management of Great Lakes fisheries.Results will also provide general guidance for fishery managers in other regions.
The emphasis of this project is the development, parameter estimation and use of models describing the dynamics of fish population and communities.The fish community models are effectively multiple population models with the populations connected by their predator-prey interactions.The theory and basis for models of fish population dynamics has a long history.Most of the basic theory is outlined by Beverton and Holt (1957), Ricker (1975), and Quinn and Deriso (1999).Such models need to account for recruitment of young fish into the population and mortality (or emigration).For self-sustaining populations, recruitment can be modeled as a function of the adult population that produced it (allowing stochastic variation about the expected recruitment).Such stock-recruitment functions can be extended to incorporate inhibition of recruitment of one species by another or alterations related to known (i.e., measured) environmental variables (see Hilborn and Walters 1992).For a system supported by stocking of hatchery-reared fish, the stocked fish need to be accounted for as part of recruitment.Sometimes this is simply the number stocked, with an adjustment for early post-stocking mortality (e.g., Stewart et al. 1981, Stewart and Ibarra 1991, Sitar et al. 1999).
For many populations, mortality can be modeled reasonably using year and age specific instantaneous rates (Beverton and Holt 1957), although when mortality is extremely high seasonal variations become important.The total instantaneous mortality rate needs to be decomposed into its important components because management actions will typically impact one or a few components.For example, Sitar et al. (1999) and Sitar (1996) decomposed total mortality of lake trout in Lake Huron into components due to commercial fishing, recreational fishing, sea lamprey and background (natural) morality.This allowed an evaluation of how age-specific mortality rates of lake trout would change in response to particular management actions.
Predator (or parasite as in the case of sea lamprey) effects can be introduced into a fish population model in a variety of ways.One approach is to have some external measure of the predator’s attack rate so that the predator-prey interaction is not directly modeled.Sitar et al. (1999) used observed numbers of marks made by sea lamprey on lake trout following this approach (see also Bence et al. 2002).An alternative approach would be to have a measure of predator abundance and calculate the mortality rate caused by the predator via a functional response (sensu Holling 1959) that relates predator attack rate to prey abundance (see Vetter 1988 and Bence et al. 2002).This requires estimation of functional response parameters, but has the advantage of being able to predict future mortality rates in response to changes in prey and predator abundance.Such a model is still not a community model since the predator population is assumed known based on external information.A true community model would simultaneously model the population dynamics of both predator and prey.Such models have been developed for Great Lakes salmonine-prey communities (e.g., Jones et al. 1993).
Perhaps the most significant challenge in developing fish population and community models is developing a plausible description with parameters that can be estimated from available data.For many populations in the ocean and in the Great Lakes, available data will consist of information on fishery harvest (amount and age-composition), fishery effort and survey indices of abundance. Analysis of such data and fitting of population models to the data is called catch-age analysis, and state-of-the-art methods use numerical maximum likelihood methods to estimate parameters and follow a Bayesian approach to allow for prior information and to better describe uncertainty (NRC 1998).This involves the simultaneous estimation of all free parameters jointly to all available data.This requires some explicit judgment regarding the relative quality of data and information from different sources, and model-building efforts need to emphasize parameter sparse constructions.For example, fishing mortality is often estimated as the product of a year effect and an age effects (e.g., Fournier and Archibald 1982, Deriso et al. 1985) to reduce the number of parameters that need to be estimated.These same methods can be applied to multi-species models (see Pope 1989, NRC 1998, Quinn and Deriso 1999).State-space applications that explicitly incorporate measurement and process error have recently become important in catch-age applications (e.g., Sullivan 1992, Schnute 1994, Richards et al. 1997).Increased emphasis is also being placed on providing realistic estimates of uncertainty for parametersand there is much to be learned about the performance of all these catch-age methods (NRC 1998).In the Great Lakes emphasis has historically centered more on describing species interactions and system limits and less on statistical rigor of parameter estimation.For example there are numerous applications of bioenergetics models (e.g., Stewart et al. 1981, Stewart and Ibarra 1991) evaluating predator demand for prey and full salmon-prey community models exist for Lake Michigan and LakeOntario.Sitar et al. (1999), however, was the first statistical catch-age application to lake trout in the Great Lakes.In contrast, most fishery management models used in marine systems do not incorporate species interactions, but the tradition of statistical approaches to model fitting is more firmly established (NRC 1998).
Development of a population/community models and estimation of parameters is only part of the fishery assessment task.In addition we need to use the model to evaluate management alternatives.In the past most analyses used deterministic models based on the best fit model and produced management criteria such as maximizing the equilibrium yield to a fishery (Ricker 1975).Rational decisions require some idea of the uncertainty associated with estimates and forecasts, as well as information on the expected magnitude of future temporal variability in the fishery (Hilborn and Walters 1992).Thus, more recently emphasis has moved to stochastic simulation models and evaluation of distributions of outcomes using risk assessment and decision analysis (NRC 1998, Peterman et al. 1998).These approaches combine information on uncertainty and temporal variability to estimate the probability that a management approach will lead to acceptable results. Advances in parameter fitting methods, such as Bayesian approaches (Gelman et al. 1995) and state-space models (Schnute 1994) are well suited for providing the information these methods need.
E.Procedure:
Job 1.Review literature and unpublished studies on the Great Lakes ecosystem, fish and community population dynamics, and fishery stock assessment including the related area of environmental statistics.Synthesize literature as needed using informal and formal (e.g., meta-analysis) methods.
Job 2.Expand the PERM research program into new areas complementary to the existing funded research programs with an emphasis on system modeling, fishery stock assessment and the population dynamics of Great Lakes fish stocks.Seek additional funding from a variety of partners to enhance the research program.
Job 3.Provide service to Michigan DNR fisheries division, other partners of the PERM program including MichiganStateUniversity, and the broader scientific community with activities such as program reviews, review of journal articles and grant proposals, statistical consulting, membership on committees, editorial activities, academic governance, and participation in activities of professional societies.
Job 4.Provide training and other educational opportunities in a variety of areas related to quantitative fisheries in the form of formal courses, short-courses, workshops, seminars, and training programs.Such activities should have pre-service, in-service, and public outreach components.
Job 5.Prepare annual report and as appropriate communicate program results in the form of peer-reviewed publications, reports, popular articles, and presentations.
F.Schedule:
Year / Work planned2007-08 / Job 1. / Review literature.
Job 2. / Expand research program
Job 3. / Provide service
Job 4. / Provide training, education and outreach
Job 5. / Prepare annual report and communicate results
2008-09 / Job 1. / Review literature.
Job 2. / Expand research program
Job 3. / Provide service
Job 4. / Provide training, education and outreach
Job 5. / Prepare annual report and communicate results
2009-10 / Job 1. / Review literature.
Job 2. / Expand research program
Job 3. / Provide service
Job 4. / Provide training, education and outreach
Job 5. / Prepare annual report and communicate results
2010-11 / Job 1. / Review literature.
Job 2. / Expand research program
Job 3. / Provide service
Job 4. / Provide training, education and outreach
Job 5. / Prepare annual report and communicate results
2011-12 / Job 1. / Review literature.
Job 2. / Expand research program
Job 3. / Provide service
Job 4. / Provide training, education and outreach
Job 5. / Prepare annual report and communicate results
G.Geographic Location:The scope of this research isstate-wide on the Great Lakes, and as such, covers mainly Lakes Michigan, Huron, Superior, and Erie.
H.Personnel:James R. Bence, Associate Professor, Dept. of Fisheries and Wildlife, MichiganStateUniversity.