NEFSC Seminar Series
Speaker: Sarah Gaichas
Title: "Fragile ecosystems, robust assessments? (What I did with my summer vacation)”
Mathematical models have two primary and related uses in marine natural resource management: estimation of the current state of a system, and projection of the consequences of alternative future management strategies on that system. Models used in fisheries management have a wide range of complexity, and can be applied at multiple levels from single stocks to multispecies communities. However, formal skill assessment is not routinely applied to these models, primarily because observations of modeled quantities can be highly uncertain, and skill assessment evaluates how well a given model reproduces the truth. Here, we demonstrate methods for using ecosystem models as simulators to provide both a true system state for skill assessment, and datasets for input into fisheries stock assessment models with realistic observation and process uncertainty.
The California Current and Nordic/Barents Seas are experiencing rapid global change, fundamentally altering the productivity of ecosystems and fish stocks. To facilitate the development of fishery management advice that is robust to climate change, we tested the performance of stock assessment modeling approaches under simulated climate scenarios across these two ecosystems. As ‘operating models’ or virtual testbeds, we applied Atlantis ecosystem models, configured for the California Current and Nordic/Barents Seas, and forced with scenarios for warming ocean temperature. These scenarios project conditions in the 2060’s, with associated impacts on biomass dynamics via changes in fish growth, trophic interactions, and predation mortality. These ecosystem models are spatially explicit and include biological groups ranging from primary producers to top predators; here we focus primarily on key harvested species including Pacific sardine (Sardinops sagax) and Pacific hake (Merluccius productus), and Northeast Arctic cod (Gadus morhua) and Norwegian spring spawning herring (Clupea harengus). Using a new R package, AtlantisOM, we simulate survey and fishery sampling from Atlantis output, including the uncertainty and bias associated with survey and catch observations, before passing data to a Stock Synthesis assessment. Within Stock Synthesis, we will evaluate the efficacy of different modeling assumptions (e.g., time-varying, empirical, or constant) on growth and natural mortality parameters to account for changing productivity driven by climate change. We will evaluate stock assessment performance by quantifying the bias and precision of derived quantities related to population size, fishing intensity, and depletion, and by evaluating management performance on forward projections in which fishing rates were set based on reference points estimated in the assessment.