Model Misspecification or Data Bias: The Assessment of Prince Rupert Herring
- Price: $0.75 Sale: $0.00
Pacific herring stock assessments have been conducted with two analytical models since the early 1980s. Correspondence between the estimates of abundance provided by these independent assessments of stock size has been good for four of the five major stock groups. In the remaining area, the Prince Rupert assessment region, estimates of abundance from these models have diverged widely, particularly in recent years. An investigation of the biological sampling data for this area revealed limited information on age structure from the fall food and bait fisheries of the late 1970s and early 1980s. In addition, there were shifts in the pattern of sample collection throughout the region with time which could be responsible for a bias in the estimates from the age-structured or catch-age model. Deletion of the fall samples from the input data resulted in an improvement in model fit. An investigation of possible bias in the spawn index, used to tune the age-structured model, due to changes in survey methodology yielded minor improvements in model fit and may be artifactual rather than real. Analyses based on the reduced data series described here appear to provide a more realistic reconstruction of stock size and trajectory than in the past and should be used in future assessments of the herring population. This paper provides an example of the benefits of exploring the value and consistency of data inputs or sources.
- Item number: AK-SG-98-01u
- Year: 1998
- DOI: https://doi.org/10.4027/fsam.1998.21