Random effects in state-space stock assessment models: how to structure them, how to project them, and implications for advice setting and management

Wednesday, March 11, 2026 at 3:00pm to 4:00pm

Seminar Announcement

Department of Fisheries Oceanography

"Random effects in state-space stock assessment models: how to structure them, how to project them, and implications for advice setting and management"
Emily Liljestrand
Fish Biologist, Northeast Fisheries Science Center

Wednesday, March 11, 2026
3:00 - 4:00 pm
SMAST E 101-103 and via Zoom

Abstract:

Age-based stock assessment models estimate the abundance of cohorts through time by fitting to age-specific data from catch and surveys. In a state-space modeling framework, the data are assumed to be imperfect observations derived from unobserved “states” that progress over time. The yearly deviations in these states, either from the previous year’s value (in a random walk or autocorrelated process) or a global mean (in white noise) are random effects (RE). In stock assessment models these states can include the recruitment, abundance, selectivity, natural mortality, or other time-varying processes. In the Northeast U.S. several stocks are now managed using the state-space modeling platform Woods Hole Assessment Model (WHAM) which allows for one or multiple simultaneous RE processes. One popular feature of WHAM are the numbers-at-age random effects (NAA RE) which add stochasticity to the progression of cohorts through time, are centered around zero, can be correlated across years or ages, and has a variance estimated as a fixed effect. These NAA RE capture additional variability in NAA that can exacerbate or mitigate the effects of mortality by implicitly accounting for environmental effects on survival, predation, movement, or other processes that are not accounted for elsewhere in the model. Though their theoretical application is well established, the process of interpreting, explaining, and ultimately using random effects, particularly the NAA RE in management track assessments has proved challenging. This presentation will detail how random effects are modeled in WHAM, show results from recent stock assessment models that use random effects, and demonstrate how they affect short-term projections which can impact catch advice.

Join Meeting
https://umassd.zoom.us/j/93758230260
Note: Meeting ID and passcode required, email contact to obtain.

SMAST East 101-103 : 836 S. Rodney French Boulevard, New Bedford MA 02744
Callie Rumbut
c.rumbut@umassd.edu
https://umassd.zoom.us/j/93758230260