SAFS hosts weekly lunch-time seminars, where students and faculty share findings from their current research. Read through our past seminars to get an idea of topics covered and be sure to check out our events calendar to download upcoming seminars on your calendar.

Any recordings will be uploaded below and on the SAFS YouTube.

Spring 2026 Quantitative Seminars

Fridays at 12:30-1:30 pm in FSH 203.

 

Option to join the seminars remotely via Zoom is available.

4/10/26

Matt Falcy (USGS, Idaho Cooperative Fish & Wildlife Research Unit)

TITLE: The Ecology of Uncertainty: Models, Decisions, and Management

ABSTRACT: Measuring uncertainty is a critical element of science, but using such measures in fisheries decision-making can be challenging. I will present two papers that illustrate how to use uncertainty in management. In the first paper, I suggest that widely used calculations of pHOS are not commensurate with the genetic theory that motivates the use of pHOS as an index of risk to salmonids from domestication selection in hatcheries. I derive an alternative metric of pHOS that is commensurate with the spatiotemporal variability of natural- and hatchery- origin fish and the genetic theory used to index risk. I will present variance estimators of pHOS and a Bayesian belief network that propagates uncertainty in pHOS into hatchery management decisions. In the second paper, I derive novel probability density functions of utility using the transformation of variable technique, and I show how the same data can lead to different decisions if the decision is based on p-values versus expected utility. This presentation features 26 equations and 0 pictures of nature to illustrate abstract and philosophical aspects of data-driven fisheries decision-making.

04/17/26

Sandra O’Neill & John Best (Washington Department of Fish and Wildlife)

TITLE: Modeling population-level effects of PCB exposure in Puget Sound Chinook salmon

ABSTRACT: Polychlorinated biphenyls (PCBs) are legacy toxic organic contaminants that can continue to persist in the environment for decades after they were banned. There has been substantial work to determine the effects of these contaminants on fish at the cellular and organism level, but these effects have seldom been extended to populations while accounting for the distribution of exposure levels within the population. Using exposure observations, the literature on organism-level effects, and existing life cycle models, we project the effects of PCB exposure on juvenile Chinook salmon to the population level in the Puyallup and Stillaguamish Rivers.

04/24/26

Mark Sorel (Washington Department of Fish and Wildlife)

TITLE: Regularized time-varying regression for Columbia River salmon forecasting

ABSTRACT: Forecasting salmon returns to set harvest limits requires navigating a fundamental tension: dozens of environmental covariates could potentially improve predictions, but return time series rarely exceed 50 years, making overfitting a serious risk. Compounding this, relationships between covariates and returns may be nonstationary — the effect of ocean conditions or cohort ratios today may differ from what they were 30 years ago. Standard approaches address these challenges separately: variable selection, model averaging, or regularization for the dimensionality problem, and time-varying parameter models for nonstationary. I am exploring a unified approach that handles both simultaneously — a dynamic linear model with an R2D2-inspired variance-budgeting shrinkage prior that penalizes both the overall importance and the degree of time-variation for each predictor. I will describe this framework, its application to Columbia River salmon, and an evaluation of forecast performance compared to alternative models.

05/01/26

Brice Semmens (University of California, San Diego)

TITLE: AI, lasers, and other sweet stuff: I’m doing fisheries science using sweet stuff like lasers. Jealous?

ABSTRACT: Low-cost imagery and citizen science offer the potential for generating valuable data for fisheries management. The challenge, however, is to develop the supporting imaging tools — such as lasers to assist depth-of-field calculations — and AI pipelines capable of distilling rich video data into a usable subset for fisheries analysis. I will describe two separate projects with partially shared AI pipelines: SMILE (Size Matters: Innovative Length Estimation) and Grouper Spotter. Both projects must solve the “many-to-one” problem, where high-resolution video produces hundreds of frames of the same individual across varying orientations. I will detail workflow designed to segment sparse instances from high-frame-rate video and select the “optimal” frame for either laser-calibrated length measurement or individual pattern recognition. By leveraging these automated pipelines and low-cost subsea imaging, we demonstrate how longitudinal citizen science efforts can be transformed into rigorous datasets for fisheries management.

05/08/26

Roger Peters (United States Fish and Wildlife Service)

TITLE:  TBD

ABSTRACT:  TBD

05/15/26

TBD

TITLE:  TBD

ABSTRACT:  TBD

05/29/26

TBD

TITLE:  TBD

ABSTRACT:  TBD

06/05/26

Kathleen Durkin

TITLE:  TBD

ABSTRACT:  TBD

Archived Quantitative Seminars

Past seminar recordings can be found on our YouTube channel Quantitative Seminar playlist

Winter 2026 Seminars
Autumn 2025 Seminars
Spring 2025 Seminars
Winter 2025 Seminars
Autumn 2024 Seminars
Spring 2024 Seminars
Winter 2024 Seminars
Autumn 2023 Seminars
Spring 2023 Seminars
Winter 2023 Seminars
Autumn 2022 Seminars
Spring 2022 Seminars
Winter 2022 Seminars
Autumn 2021 Seminars
Spring 2021 Seminars
Winter 2021 Seminars
Autumn 2020 Seminars
Spring 2020 Seminars
Spring 2015 Seminars