Date of Award


Level of Access Assigned by Author

Campus-Only Thesis

Degree Name

Master of Science (MS)


Marine Biology


Richard Wahle

Second Committee Member

Carla Guenther

Third Committee Member

Yong Chen

Additional Committee Members

James Wilson


The creation of predictive models based on demographic and environmental information is one of the most important responsibilities of fisheries science. However, in general, such models have had mixed success in the past, and despite increased complexity and computational sophistication, remain highly controversial. Paradoxically, with this increased complexity comes increased uncertainty due to the additive nature of statistical error and the difficulty with which complex environmental and demographic data are acquired. Enhanced fisheries management outcomes may be realized if effective models are created that accurately quantify error and its related problems, and fishing industry stakeholders may benefit from increased capacity for advanced decision making. In the highly valuable American lobster (Homarus americanus) fishery, early indications of declines in juvenile abundance over the past several years and overdependence on the fishery by coastal communities in the Gulf of Maine necessitate the development of early-warning systems, such as forecast models, to allow fishermen and fishery managers to make good decisions in anticipation of fluctuations in landings. Such beneficial outcomes will be more likely if scientists use appropriate statistical methodologies and develop effective communication strategies, including the establishment of trust with stakeholders, throughout the model development process.

Chapter 1 of this thesis describes the development of quantitative approaches to forecasting in fisheries over the past several decades. It discusses the controversy surrounding forecast models throughout multiple stages of development and levels of sophistication. Although ecosystem-based fisheries management approaches include the incorporation of environmental and demographic variables into forecasts, many studies cast serious doubt on the efficacy of forecast models in general and recruitment forecasts in particular. In this chapter, three case studies that illustrate the increased sophistication of forecasting models over the past 60 years are presented and discussed. Although many methodological issues in forecasting exist, steps can be taken to mitigate each one, and when used, enable forecast models to be used as effective management tools.

Chapter 2 presents and discusses a series of new fishery recruitment forecast models for 12 American lobster fishery statistical reporting areas in the Gulf of Maine and Southern New England. A series of predictive models with iteratively increasing complexity were used to investigate the utility of environmentally-based models designed to predict fishery recruitment. The most sophisticated models use a Monte Carlo simulation procedure to create forecasts, as well as several novel approaches to quantifying lobster growth rates, and commercial landings data are used for validation purposes. Forecasts were statistically significantly correlated with landings in 10 of 12 study areas for which models were created. Declines in landings are predicted in all Gulf of Maine study areas, and stable but historically low landings are predicted for southern New England study areas. The implications for the fishery and the uncertainty surrounding process variables used in the model are discussed in detail.

Chapter 3 investigates the issues surrounding industry stakeholders’ perception and value of different sources of knowledge pertaining to the American lobster fishery in Maine. Lobster fishermen in Maine may be able to benefit substantially from scientific information, but significant institutional barriers between science and industry exist that challenge the establishment of a greater trust between them. A printed survey, which was presented to fishermen attending industry meetings in Western, Midcoast, and Downeast regions of Maine, was used to assess fishermen’s use of fishery-independent and fishery- dependent information and their perceptions of interactions with scientists the scientific institutions. Analytical results from the survey were used to create a series of recommendations to scientists in order to enhance the utility of scientific information to industry stakeholders and to work towards improving trust between scientific and fishing industry institutions.