James T. Thorson¹, Andrew O. Shelton¹, Eric J. Ward¹, and Hans Skaug<sup>2

1</sup>Northwest Fisheries Science Center and ²University of Bergen
October 14, 2014

Indices of abundance are the bedrock for stock assessments or empirical management procedures used to manage fishery catches for fish populations worldwide, and are generally obtained by processing catch-rate data. Recent research suggests that geostatistical models can explain a substantial portion of variability in catch rates via the location of samples (i.e., whether located in high- or low-density habitats), and thus use available catch-rate data more efficiently than conventional “design-based” or stratified estimators. However, the generality of this conclusion is currently unknown because geostatistical models are computationally challenging to simulation-test and have not previously been evaluated using multiple species. We develop a new maximum likelihood estimator for geostatistical index standardization, which uses recent improvements in estimation for random fields. We apply the model to data for 28 groundfish species off the U.S. West Coast and compare results to a previous “stratified” index standardization model, which accounts for spatial variation using post-stratification of available data. This demonstrates that the stratified model generates a relative index with 60% larger estimation intervals than the geostatistical model. We also apply both models to simulated data and demonstrate (1) that the geostatistical model has well-calibrated confidence intervals (they include the “true” value at approximately the nominal rate), (2) that neither model on average under- or overestimates changes in abundance, and (3) that the geostatistical model has on average 20% lower estimation errors than a stratified model. We therefore conclude that the geostatistical model uses survey data more efficiently than the stratified model, and therefore provides a more cost-efficient treatment for historical and ongoing fish sampling data.

Eric Ward¹, Mark Scheuerell¹, and Steve Katz^2
1Northwest Fisheries Science Center and ²Washington State University
May 21, 2014

Kelp forest ecosystems (KFEs) occur globally along temperate-zone coast lines, and are some of the most diverse and productive ecosystems of the world. They also provide a wide array of ecosystem services that have sustained coastal human communities for millenia. However, harvest of apex predators and important grazers has had profound effects on KFE structure and services provided. Unfortunately, quantitative tools to quantify the impact of these stressors on ecosystem dynamics have been lacking. Here we present a novel method for analyzing temporal changes in (1) the strength of intra- and inter-species interactions within an ecosystem, and (2) the stability of the ecosystem itself. We applied a Bayesian time-varying vector autoregressive state-space (TVVARSS) model to a 32-year time series of KFE structure at San Nicolas Island, California (USA). Although sea otters were initially absent from the island, they were reintroduced in the late 1980s. We found that the resilience of the KFE has decreased over time concomitant with increased otter density, and that the KFE has also become more reactive to external perturbations such as stron El Niño events.

Vladlena Gertseva
Northwest Fisheries Science Center
April 23, 2014

The second dorsal spine has historically been used for age determination in the spiny dogfish shark. The dorsal spines are located on the external surface of the body and are subjected to natural wear and breakage. Two methods have been developed to account for the worn portion of the spine and extrapolate the lost annuli. We compared the performance of these methods using a large data collection assembled from multiple sources, and evaluated their utility for stock assessment and management of the spiny dogfish shark Squalus suckleyi in the Northeast Pacific Ocean. Our results showed that the two methods produced very different age

Kirsten Holsman
Alaska Fisheries Science Center
March 12, 2014

Climate change is expected to impact marine ecosystems globally, with largest changes anticipated for arctic and sub-arctic ecosystems. The 2°C projected increase in mean summer sea-surface temperature for Alaskan marine ecosystems may alter trophic demand, predator and prey distributions, and overall system productivity. We use multi-species food-web and assessment models to link climate-driven changes in physical and trophodynamic conditions to recruitment and survival to help distinguish fishery impacts from large-scale climate pressures. We first used a Regional Ocean Modeling System (ROMS) model coupled to a Nutrient-Phytoplankton-Zooplankton (NPZ) model to produce detailed hindcasts for the period 1970-2012. These results modulate species interactions in a climate-driven Multispecies Statistical Model (MSM) for three groundfish species from the Bering Sea (walleye pollock, Pacific cod, arrowtooth flounder). We first fit the model to hindcast-extracted time series then used downscaled IPCC scenario-driven ROMS/NPZ model estimates of temperature, circulation, and zooplankton abundance to project MSM forward to 2040. Evaluation of differing climate effects on population projections and predator-prey interactions under various IPCC forecast scenarios, recruitment models, and management strategies will be discussed.

David Sampson
Oregon State University
November 27, 2013

Most population models and stock assessment approaches assume that fish and fishing are uniform in their spatial distribution. In this Think Tank talk we will examine some of the effects that spatial structure can have on selectivity, which refers to the phenomena by which age- or size-classes of fish are not equally vulnerable to fishing. Fishery selection (selectivity) controls how fishing affects the agestructure of a fish stock, which in turn influences a stock’s productivity and ability to sustain harvests. It is possible to analyze selectivity at multiple spatial scales and selectivity can vary at these different scales. For example, the shape of the selection curve at the level of the population will be dome-shaped under a wide range of conditions, even though the gear-selection curve operating in common across all the regions is asymptotic. During this talk we will explore the factors that influence the shape of areas-asfleets selection curves, which are the form of selectivity that results from applying separate regional selection coefficients to the entire stock rather than to the regional sub-populations. Areas-as-fleets selectivity provides a mechanism for modeling spatial catch-at-age in an assessment model that is not spatially explicit. We use a simplified set of survival equations from Sampson and Scott (Can. J. Fish. Aquat. Sci. 68, 1077-1086. 2011), in which there are three distinct spatial regions and the same asymptotic selection curve operating in each region, and show that the overall population-selectivity often will be dome-shaped, but the corresponding areas-as-fleets selection curves will have shapes that differ from each other and from the population-selection and gear-selection curves. Factors that influence how much the selectivity curves differ from each other include the magnitude of the differences in the regional fishing mortality rates, the rates of fish movement and whether the movements are diffusive or directional, and the regional distribution of the recruits. Based on ad hoc explorations with the model it appears that areas-as-fleets selectivity will generally differ from the underlying gear-selectivity, except under the extraordinary conditions of equal recruitment and equal fishing mortality across all spatial regions. The talk will also explore the temporal variations that changes in the regional recruitment or the rates of fishing mortality will induce in population-selection and areas-as-fleets selection curves.

Ivonne Ortiz^1,2, and Kerim Aydin^2
1UW School of Aquatic & Fishery Sciences
²Alaska Fisheries Science Center
April 19, 2011

The Bering Sea Integrated Research Program (BSIERP) is a 5 year program designed to test hypotheses regarding the ecosystem’s response to climate change. Both historical and BSIERP field data are used to put together a vertically integrated model that includes 5 modules: i) Climate (specific to the Bering Sea), ii) ROMS (Regional Ocean Modeling System) , iii) NPZ-benthos (Nutrient-Phytoplankton-Zooplankton-benthos), iv) FEAST (Forage/Euphausiid Abundance in Space and Time), and v) Economics (fleet movement model). The vertically integrated model in turn, will be used as the “real world” model in a Management Strategy Evaluation for the Bering Sea pollock, and Pacific cod fisheries. We will start with a presentation of the vertically integrated model structure and feedbacks, summarizing some of the challenges of working across multiple scales, units and objectives. We will then discuss the development, ongoing tuning and validation of the fish module FEAST, highlighting transitions from original to current formulations of bioenergetics, movement and reproduction through several rounds of meetings with field biologists. Finally, we’ll detail what processes in the model are expected to respond to climate change, where we expect them to be buffered, and where they are non-sensitive. FEAST runs on a grid of ~10 km resolution and models size-based, two way interactions between 13 fish species (walleye pollock, Pacific cod, arrowtooth flounder, salmon, capelin, herring, eulachon, sandlance, myctophids, squids, shrimp, crabs and other), and the 7 zooplankton groups in the NPZ model (small/large microzooplankton, small/large copepods, euphausiids, jellyfish, and benthic infauna). Both temperature and advection from ROMS are used in the bioenergetics and movement components. The operating hypothesis in FEAST is that forage fish and macrozooplankton are tightly coupled in a two-way interaction, and the dynamics of this interaction under different climate scenarios is a strong structuring element for the ecosystem as a whole.

Melissa Haltuch
Northwest Fisheries Science Center
March 14, 2011

Petrale sole are a commercially important flatfish that migrate seasonally between feeding and spawning grounds, and have recently been declared overfished. The summer trawl survey shows a decline in petrale sole abundance since 2005 similar to the unstandardized summer catch-per-unit-effort (CPUE) from the fishery. However, many stakeholders disagree that petrale sole abundance has been declining, instead choosing to focus on the unstandardized winter CPUE that shows a strong increase beginning in 2000. The assessment attributes the increasing trend in winter CPUE to management actions that forced the fleet to 1) increase fishing effort during the winter, and 2) conduct winter fishing in locations with high historical catch rates. Standardized fishery CPUE was not used in the assessment due to changing management regulations beginning in the late 1990s and the high likelihood of a winter CPUE index showing hyper-stability due to the fishery focusing on the aggregated spawning stock. Given the potential discrepancy between the assessment results and the experience of the groundfish fleet, particularly during the winter fishing season, and the limited conclusions that can be drawn from unstandardized CPUE, this work explores the utility of the summer and winter fishery CPUE series as indices of abundance for the petrale sole stock assessment. The ultimate goals are to determine if an adequate index of abundance can be created using fishery CPUE, and to address the uncertainty due to the discrepancy between the fishery independent and fishery dependent data sources and therefore the perceived stock assessment uncertainty.

Charlotte Boyd and Mathieu Woillez
UW School of Aquatic & Fishery Sciences
February 28, 2011

Geostatistical conditional simulations are a useful tool for capturing the full range of variability in spatial data. The distribution of schooling fish is inherently patchy – this patchiness may be lost in kriging which presents a best local estimate at each point and tends to smooth over local variability. In contrast, stochastic geostatistical simulations reproduce the variability of a variogram model. Simulations can be conditioned on the observed data values, and, on average, reproduce the statistical properties and spatial pattern of the sample data. Multiple simulations can therefore be used as the basis for estimating uncertainty at given locations. Geostatistical simulation is thus relevant to analyses of spatial sample data in which measurement error, local variability, or sampling uncertainty are important (such as risk assessment and decision analysis). In the Gaussian case, geostatistical conditional simulation can be achieved relatively easily by adding a simulated error term to the kriging. However, in the case of acoustic surveys of schooling fish, where the acoustic backscatter is often characterized by a high proportion of zeroes and skewed positive values, some transformations are necessary. Here, we present and discuss two contrasting approaches to perform geostatistical conditional simulation in this context – one based on transformed Gaussian simulations and on a Gibbs sampler to handle the numerous zero acoustic values within a classical geostatistical framework, and the other based on a binomial/ lognormal hurdle model within a generalized linear mixed modeling framework.

Bill Clark¹, and Rick Methot^2
1International Pacific Halibut Commission and UW School of Aquatic & Fishery Sciences
²Northwest Fisheries Science Center
February 7, 2011

Current efforts to conserve Pacific salmon (Oncorhynchus spp.) rely on a variety of information sources including empirical observations, expert opinion, and models. I will outline a framework for incorporating detailed information on density-dependent population growth, habitat attributes, hatchery operations, and harvest management into conservation planning in a time-varying, spatially explicit manner. The model relies on a multi-stage Beverton-Holt model to describe the production of salmon from one life stage to the next. We used information from the literature to construct relationships between the physical environment and the necessary productivity and capacity parameters for the model. As an example of how policy makers can use the model in recovery planning, we applied the model to a threatened population of Chinook salmon (O. tshawytscha) in the Snohomish River basin in Puget Sound, Washington, USA. By incorporating additional data on hatchery operations and harvest management for Snohomish River basin stocks, we show how proposed actions to improve physical habitat throughout the basin translate into projected improvements in four important population attributes: abundance, productivity, spatial structure, and life history diversity. I will also describe how to adapt the model to a variety of other management applications.

Judith E. Zeh
UW Department of Statistics
October 05, 2005

The International Whaling Commission (IWC) was established in 1946 by the International Convention for the Regulation of Whaling, signed by 14 whaling nations, “to provide for the proper conservation of whale stocks and thus make possible the orderly development of the whaling industry”. Part of the Convention is a Schedule that contains the actual regulations regarding species and numbers of great whales that can be caught, times and places in which whaling is allowed, etc. Amendments to the Schedule, which require a 3/4 majority vote for adoption, must be “based on scientific findings”. Thus, since its inception, the intent of the IWC has been to base management on science, and one of its standing committees has been the Scientific Committee (SC). The SC meets annually, just before the Commission meets, and the Chair of the SC presents SC findings to the Commission. I will talk about successes and failures of this management process before, during, and since my 1999-2002 term as SC Chair. Successes have come when the Commission obtained and followed good scientific advice. Failures have sometimes occurred because of inadequate scientific advice, but more often because economics or politics got in the way of following good advice. Both successes and failures occurred in the 1960s, when a committee of three scientists appointed by the Commission recommended immediate protection of Antarctic humpback and blue whales from whaling and drastic reductions in fin whale catches. The Commission did protect humpback and blue whales, but delayed reductions in fin whale catches because of pressure from whaling nations. Eventually greater reductions in fin whale catches had to be made to allow the stock to recover. The management procedure developed by the SC during the 1970s proved unworkable because it required classifying whale stocks on the basis of quantities that were difficult to estimate. Meanwhile, some whaling nations stopped whaling and other nations joined the IWC. It now has 66 members, the majority of which are non-whaling nations and many of which could be characterized as anti-whaling nations. This adds a complicating dimension to the “science and policy interface”. During the 1980s, the Commission imposed a moratorium on commercial whaling that is still in effect. However, the Convention allows whaling in spite of the moratorium by nations that objected to its adoption and by any nation under Special Permits for scientific research. Meanwhile, the SC has developed a revised management procedure (RMP) that requires only regular estimates of abundance of a stock and the known catch history. The RMP was tested by simulations of 100 years of catches using it. These simulations took into account uncertainties in a wide range of factors. In my view, whales and whalers would be better protected by use of the RMP to manage whaling than by the moratorium. The SC currently provides advice on aboriginal subsistence catch limits for bowhead whales using a similar management procedure.