Incorporating Environmental and Ecological Variables to Improve Stock Assessments: An Application to Northern Shrimp in the Gulf of Maine

Yong Chen and Jie Cao School of Marine Sciences, University of Maine


Northern shrimp (Pandalus borealis) population on the Gulf of Maine (GOM) is considered a single stock unin, and the southwestern Gulf of Maine (GOM) is the southern end of the species range. Abiotic factors, particularly temperature and predation, appear to influence a number of aspects of shrimp life history. The timing of egg development, egg survival, embryonic development, and larval size are linked to incubation temperature, and in recent years the start of the hatch period in the GOM has become earlier as temperatures have warmed. Predation is another factor thought to strongly influence shrimp population dynamics. In a meta-analysis of 9 populations of shrimp, Worm and Myers (2003) found evidence of top-down control by Atlantic cod predation in 8 of the 9 populations. The exception was the GOM, a system where a number of other predators are important in addition to Atlantic cod. These results suggest that incorporating temperature and predation in the assessment model is critically important.

The GOM shrimp has been assessed using the Collie-Sissenwine (catch-survey, CSA) model since 1997. CSA has been criticized for being too simple to quantify the complex shrimp life history and demography, especially the sex/size structure, which can be considerably altered by selective harvesting of females. To address these shortcomings, a new length-structured model (LSM) has been developed for shrimp in the GOM. The LSM was presented as the base model in the shrimp benchmark stock assessment in 2014. However, the results derived from all the stock assessment models including CSA and LSM were rejected by the review panel because of significant residual patterns in model fitting. The review panel recommended further development of LSM with incorporation of environmental and ecological variables to improve the model fit.

This project addresses the need for explicit consideration of environmental and ecological variables in the shrimp stock assessment. We have the following 3 goals for this proposed FATE project: (1) develop and evaluate approaches for explicit incorporation of temperature in modeling recruitment dynamics in the length-structured model; (2) evaluate various functions to link temporal variability of shrimp predator abundance with natural mortality; and (3) evaluate impacts of temperature on the growth of shrimp and develop a temperature-dependent growth transition matrix for the length- structured model.

Building on the preliminary study we have done so far and using the data collected by the state and federal surveys, we will complete the following tasks: (1) continue analyzing existing information to examine the relationships between temperature and recruitment; (2) quantify the relationships between temperature and growth; (3) develop and evaluate different functional relationships to link predator abundance with shrimp natural mortality; (4) develop and incorporate likelihood functions based on the models developed above in the length-structured model; and (5) evaluate impacts of incorporating temperature and predators on the quality of the stock assessment in both simulations and the assessment (e.g., retrospective errors and uncertainty). We plan to complete the project in two years. We will finish the first four tasks (1-4 listed above) in the first year and task (5) in the 2nd year. In the 2nd year, we will also prepare all the software and data to fully incorporate the results derived in this study in the next shrimp stock assessment.

Given substantial changes already observed in the GOM ecosystem, this project will fill a critical need by developing an assessment model that specifically includes environmental and ecological drivers. This study will not only improve our understanding of shrimp-habitat-ecosystem dynamics in the GOM, resulting in improved forecasting of recruitment and future productivity, but, will be applicable to other fishery resources in this rapidly changing environment. The successful completion of this project by the end of 2016 will be very timely as the moderate 2013 shrimp year class will be entering the fishery in 2017 (the fishery will likely be closed until 2017).