The NOAA Central Library seminar program provides an educational forum for the presentation of ideas, research updates and news in support of NOAA’s mission: to understand and predict changes in climate, weather, oceans and coasts; to share that knowledge and information with others; and to conserve and manage coastal and marine ecosystems and resources. The National Stock Assessment program has partnered with the NOAA Central Library to host a National Stock Assessment Science Seminar Series. Below are links to past seminars on a range of stock assessment methods, models, and programs.
Introducing Casal2 for Assessments
Ian Doonan, PhD, Population Modelling Group, National Institute of Water and Atmosphere Research (NIWA), New Zealand
Abstract: Casal2 is replacing CASAL, New Zealand’s stock assessment package written in 2001. CASAL implements integrated stock assessments, but it has hard-wired fisheries structures that makes introducing newer concepts impossible. Casal2 was written so that it is easily maintained, and we can implement new concepts. Currently, Casal2 can replicate most features of CASAL, apart from little used ones, and it gives the same parameter estimates as CASAL over a suite of test assessments. Following a review of Casal2’s capabilities, an assessment in Casal2 will be demonstrated.
Cross-comparison study of AD software in the application of population dynamic models
Andrea Havron, PhD, NOAA Fisheries Office of Science and Technology National Stock Assessment Program Modeling Team
Abstract: Motivated by the Fisheries Integrated Modeling Systems initiative, this study aims to compare automatic differentiation software platforms used in population dynamic modeling. Pathways for both frequentist and Bayesian inference are outlined for state-space and explicit spatial models developed and implemented across the following software platforms: ADMB, Julia, Stan, and TMB. Software features of each platform are detailed along with benchmark results on performance and speed from simulations. Features considered include statistical and computational efficiency, accuracy, accessibility to the developer community, and user friendliness.
Close-kin Genetic Methods for Estimating Census Size and Effective Population Size
Abstract: It is now possible to genetically identify close relatives in wild populations, and this information can be used in a mark-recapture framework (CKMR) to estimate abundance (N). Successful CKMR applications require information about vital rates and other life history traits, and the same genetic and demographic data can be used to estimate effective population size (Ne), the evolutionary analogue to N. I discuss how life history information differently affects CKMR estimates of N and genetic estimates of Ne, and how combined datasets can be leveraged to better understand both the ecological and evolutionary correlates of abundance.
SSMSE: A Tool for Management Strategy Evaluation Using Stock Synthesis Operating Models
Kathryn Doering and Nathan Vaughan
Abstract: While Management Strategy Evaluation (MSE) is becoming a routine task, creating realistic operating models (OMs) for use in MSE and running MSE analyses relatively quickly is a challenge. We will describe and demonstrate SSMSE, an R package for direct use of Stock Synthesis (SS) models as OMs in MSE. We will discuss the features of the package and demonstrate its capabilities using a worked example.
State Space Stock Assessment Models: Previous Applications, Future Potential
Abstract: State-space modeling (SSM), in which the process variation is estimated independently from variation due to observation error, has been employed in fisheries stock assessment for decades. Software such as Template Model Builder (TMB) sparked new interest in SSM because theoretical advantages such as estimating error variances could be realized thanks to the computationally efficient integration of process error from the marginal likelihood. I use my own research in rewriting existing age-based assessment models of Lake Whitefish with SSM elements (e.g., multiple time-varying processes) to make broader conclusions about the applicability of SSM.
Bringing the State of the World’s Fisheries Assessment into the 21st Century: What Is Needed to Improve Our Global Coverage, and How to Make the Assessment Accessible to Interested Parties Globally
Rishi Sharma and Arni Magnusson
Abstract: We present the current workflow in developing a Global Fish Stock Assessment for SOFIA and identify issues in estimating this globally important metric indicating status of stocks into three categories: i) overfished, ii) fully sustainably fished, and iii) underfished. We highlight how a new assessment package SRAplus can be used to get better estimates of regional and country estimates of overfishing categories using effort data and depletion priors based on external data. Transparency issues are highlighted and a transparent assessment framework (TAF) is used to implement a modular workflow for SOFIA (TSAF).
Developing an Expert System to Construct an Ensemble of Models for Fisheries Stock Assessment
Abstract: The increasing demand for quality stock assessments to provide fisheries management advice requires a more efficient approach. This is amplified by the call for using an ensemble of models to improve management advice and to better represent uncertainty. The availability of flexible general models, such as Stock Synthesis, has greatly facilitated the development and implementation of stock assessments. However, this is only one component of the Expert System that is needed to fulfill the need under limited financial and human resources. I describe the components of a fisheries stock assessment expert system and the approach to develop it.
The Sport of Ensemble Modeling
Abstract: Ensemble modeling has been identified as a tool to better characterize uncertainty in stock assessment, and to avoid having to identify a single “best” model when there could be more than one model that performs reasonably well. While research on ensemble modeling is growing, important questions remain about the details of implementation and how to communicate the results. This talk will highlight some of those questions with simulated case studies, and will borrow ideas for visualizing results from sports.
Science to support management of a fishery with competing interests: the Atlantic menhaden story
Amy M. Schueller
Abstract: The Atlantic menhaden fishery is the largest, by volume, on the Atlantic Coast of the United States, and Atlantic menhaden are an important forage species for predators such as striped bass and bluefish. Driven by the competing interests for this stock, the last assessment process included both single and multi-species models, which allowed for addressing multiple management objectives and for development of ecological reference points.
JABBA: An alternative to Data Moderate Stock Assessments
Felipe Carvalho and Henning Winker
Abstract: JABBA (Just Another Bayesian Biomass Assessment) is a stock assessment tool that is transparent, reproducible, and customizable for use by anyone in the world. It is an exciting example of international scientific collaboration. Since its publication in 2018, over two dozen stock assessments have used JABBA. Recently, the original model was extended to overcome many limitations common to conventional biomass dynamic models, and from this effort JABBA-SELECT was developed. In this seminar, we describe the JABBA framework, provide examples of stock assessments, and why it can be considered an alternative to data-moderate assessments.
Ecosystem-linked assessment model for Gulf of Alaska Pacific cod to assess climate change driven changes in productivity
Steven J Barbeaux
Abstract: Modern fisheries are largely managed based on assumptions of stationarity in the productivity of stocks consistent with a baseline time period. As the productivity of a stock depends on environmental conditions, the assumption of stationarity in productivity is dependent on future environmental conditions varying within the bounds of the assumed baseline conditions and productivity tending to a mean. The climate of the earth is changing due to anthropogenic forcing and although there are long-term objectives of minimizing these impacts, these impacts appear to be largely unavoidable and irreversible over the next century. Therefore it is essential that marine resource managers consider climate change in developing fisheries management plans and harvest control rules to assure long-term sustainable fisheries and global food security.
In this seminar we consider an ecosystem-linked single species assessment model in which environmental conditions drive key biological elements of productivity and projecting these models forward using forecasts generated from International Panel for Climate Change (IPCC) Representative Concentration Pathway (RCP) scenarios. Ecosystem-linked model projections assume stationarity in the relationship between environmental conditions and biological elements contributing to productivity instead of assuming stationarity in productivity itself. Development of ecosystem-linked assessment models is limited to stocks for which the relationships among environmental conditions and key biological elements of productivity have been well established and where projections of the environmental conditions are available. One such stock is the Gulf of Alaska Pacific cod (Gadus macrocephalus) for which an ecosystem-linked assessment model has been developed. In this model sea surface and bottom temperatures, and the occurrence of marine heatwaves are used to drive Pacific cod growth and recruitment. This model is then used to obtain estimates of stock productivity under three IPCC RCP scenarios using five climate models under each scenario to the end of the century. For Gulf of Alaska Pacific cod we observe significant variability in productivity among climate models projections, however under all of the IPCC scenarios productivity is reduced in comparison to status quo, more so under higher carbon concentration scenarios. Fishing under current reference points significantly increases the risk of stock collapse, which suggests that the use of more conservative reference points is warranted. There is much uncertainty in the management advice that is provided from the method presented, however the greatest uncertainty at this time is in how we will act, or fail to act, in regards to mitigating climate change impacts.
Using ecosystem-based fisheries management to address climate- related impacts to Gulf of Alaska Pacific cod
Abstract: Management of Gulf of Alaska Pacific cod was challenged by a precipitous decline in biomass closely linked with a marine heatwave during 2014-2016. This seminar will review the management response during and after this time, which has co-occurred with rapid progress in the science underpinning the decline as well as developments in ecosystem-based fisheries management. The recurrence of a marine heatwave in 2019 along with other unprecedented environmental changes underscore the need for fisheries management that can address these events in both short-term tactical decisions and longer-term strategic planning.
The Woods Hole Assessment Model (WHAM): A general state-space assessment framework
Brian Stock and Tim Miller
Abstract: We describe the Woods Hole Assessment Model (WHAM) framework and software package. WHAM can estimate time- and age-varying random effects on annual transitions in numbers at age, natural mortality (M), and selectivity, as well as fit environmental time-series with process and observation errors, missing data, and nonlinear links to recruitment and M. Including time-varying processes via random effects or environment-productivity links using WHAM may alleviate serious concerns over changing productivity and retrospective patterns for several U.S. Northeast groundfish assessments.
Enhancing stock assessments for main Hawaiin islands bottomfish through inclusion of research video-camera surveys and fishing industry engagement
Brian Langseth and Benjamin Richards
Abstract: We present a summary of the collective efforts between stock assessment scientists, survey technologists, fishermen, and academics in incorporating a fishery-dependent survey into the stock assessment for deepwater bottomfish in Hawaii. These efforts were initiated in 2012 but first incorporated into stock assessment in 2018, and were awarded the NOAA Bronze medal this year. We highlight the process and discuss lessons learned for inspiring similar efforts in other data moderate fisheries.
Estimating catch misreporting in a state-space stock assessment model
Charles Perretti, Jon Deroba, and Chris Legault
Abstract: State-space stock assessment models have become increasingly common in recent years due to their ability to estimate unobserved variables and multiple sources of error. Given these features, they may be able to estimate the unobserved process of misreported fishery catch. I describe recent research examining whether the state-space assessment model SAM is able to estimate misreported catch in a simulated fishery. I present results from a factorial experiment testing three formulations of SAM, including a new approach utilizing a random walk model of misreporting, and show the impact of misreporting on important stock assessment output.
Advancing Fish Assessments to Support EBFM
Spatial Processes and Stock Assessment Methods (SPASAM): What has our group learned after four years of simulations?
Daniel Goethel, Katelyn Bosley, Aaron Berger, Dana Hanselman, Amy Schueller, Jonathan Deroba, Brian Langseth, and Kari Fenske
Abstract: Spatial stock assessment models can enhance sustainable fisheries management. Over the last four years our group has developed and applied a spatially explicit simulation-estimation framework to explore the dynamics of fish resources under different population structure and movement assumptions. Results from our work have provided insight around spatial quota allocations, movement parameterizations, tagging study designs, and the consequences associated with misaligned management and population boundaries. Listen in as we discuss the implications of our work for developing and implementing spatial stock assessment models on management advice.
The Metapopulation Assessment System (MAS): A modular stock assessment framework
Matthew Supernaw, Bai Li, and Christine Stawitz
Abstract: The Metapopulation Assessment System (MAS) is a modern, modular stock assessment software package that is designed to analyze fish populations spanning multiple areas and population segments. The software is written in C++ and consists of separable classes for life history functions, data assimilation, and likelihood functions. This paradigm facilitates rapid estimation and enables scalable and flexible incorporation of new technology approaches. In a testament to the software’s flexibility, the National Stock Assessment Program’s modeling team recently created an R interface, making available a reproducible, R-based pipeline to build a stock assessment model, run it in MAS, and view and generate output. In this presentation, we will outline the MAS architecture and development philosophy and demo the R interface. We are giving this presentation as part of the release of MAS version 1 and hope attendees will be inspired to install the software and try it for themselves.
Stock SMART: Stock Status, Management, Assessment, and Resource Trends tool
Jeffrey Vieser, Kristan Blackhart, and Abigail Furnish
Abstract: NOAA Fisheries recently launched Stock SMART, a web tool providing public access to information related to Stock Status, Management, Assessment, and Resources Trends. With Stock SMART, users can access, visualize, compare, and download thousands of stock assessment results for federally managed fish stocks dating back to 2005. Future development will add information on fisheries management and status determinations. Stock SMART increases transparency, understanding, and trust in the fisheries management decision-making process by broadening awareness of the condition of fishery resources and informing discussions about sustainable management.
A comparison of four primary age-structured stock assessment models used in the United States
Abstract: A multi-model comparison framework was developed to evaluate the reliability of four age-structured assessment packages used by NOAA Fisheries: (1) an Assessment Model for Alaska, (2) the Age Structured Assessment Program, (3) the Beaufort Assessment Model, and (4) Stock Synthesis. All four packages give similar estimated quantities of interest, although differences among packages include initial numbers-at-age computation and bias adjustment of recruitment. The findings demonstrate how to compare multiple models through code comparison and simulation-based evaluation and highlight the need to clarify terminology used in assessment reports.
Data-integrated models for life-history parameters, and suggestions for future life-history research
Abstract: The R-package FishLife integrates available life-history and stock-recruit records to estimate Bayesian priors for fish stock assessment parameters. Results show that natural mortality is predictable based on growth and timing of maturation, and steepness is predictable based on phylogeny but not life-history parameters. Recommendations for future life-history research include 1) jointly predicting life-history and ecosystem parameters like energy density and thermal-response curves; 2) expanded life-history analysis of invertebrates; 3) integrating national life-history research with climate-vulnerability and habitat prioritization efforts.
Incorporating Environmental Data into a Stock Assessment Model and Future Population Projections
Abstract: In considering climate in stock assessment models, we also must think about how climate impacts future projections of stocks. We are working to incorporate environmental covariates into the 2018 North Pacific swordfish stock assessment and future catch projections. The Southern Oscillation Index (SOI) correlates with swordfish annual recruitment and include a forecasted SOI in future projections using SSFutures to evaluate how the projections change in the near term (2-4 years).