Welcome to the <em>Exxon Valdez</em> Oil Spill Trustee Council Home Oil Spill Facts Habitat Protection Restoration Projects Recovery Since 89
Restoration Projects

Funding Recommendations

Science Panel Recommendation: Fund

Science Panel Comments: The Herring Program team clearly gave careful thought to how modeling should be done and who should do it. Their choice and recruitment of Trevor Branch at UW is superb. This is a young rising star in fisheries dynamics modeling, who has many experienced colleagues with whom to interact. His proposal represents a good guideline for the modeling work he will begin, identifying some key processes of high value to the herring program. We expect to see evolution of the modeling as the project develops and see Branch as a leader who will make adaptive additions and modifications as new issues arise. We would like to have seen a more overt mention of how competing drivers of herring mortality will be tested against one another – physiological stress, starvation, top-down predation, and disease. These are clearly embedded in the life history modeling, but model fits to choose the factor or combinations of factors that best fit observed abundance changes would be welcome. Comments from Agency Staff (8/31/2011): Overall The proponent is a great choice for this work, and having this as a doctoral project is a cost-effective way to get some very good work done. The project description is light on details, and that is acceptable to a limited extent, given that the work includes an investigation of what has been done and the available data (via the management strategy evaluation), and that it is important to be flexible in model development. It would be helpful to have more details on the “holistic” model. For example, the Hulson et al. age structured analysis is referenced in relation to the management strategy evaluation, but there is no clear description of how the proposed holistic life-stage model relates to or builds off of the ASA, i.e., what the structure of the “holistic” model will be. Another concern is that is not clear if or how the “holistic” model will be used to aid in identifying the limiting factors in herring recruitment and recovery. That could be an important aspect of the overall herring program. The disclaimer in the second paragraph of the “Statement of the Problem” is disconcerting given the intellectual effort that the proposal aims to expend on model development: “While we do not anticipate that there will be a major change in our modeling ability in the next five years, we expect that the combination of monitoring and focused process studies will provide incremental changes over the next twenty years and result in a much better understanding of herring populations by the end of the program.” Perhaps the proponent could offer a more detailed, though conditional description of what the expected benefits might be. Other items The order of the three tasks is a bit confusing. The tasks given in Methods (p. 3-4) are: 1. Management strategy evaluation to identify most informative datasets – 2. Predict future levels of recruitment – a meta-analysis of time series for other herring and clupeid stocks. 3. Holistic model of herring dynamics – life stage model (age based), tasks conducted by UW students and faculty with access to Hilborn, Punt, and Essington. The expected order of completion of these tasks as given under Milestones (p.7) is 1. model (by 9/14), 2. MSE (by 9/15), and 3. predict recruitment (by 9/16) It is not clear why a model will be developed first, and then a different model (ASA) used in the management strategy evaluation. Also, the work to predict future recruitment, as described, appears correlational and doesn’t appear to involve the “holistic” model or a mechanistic understanding of herring dynamics, yet the timeline has this work occurring after initial model development. How would this work be related to the “holistic” model? Timeline (p. 7) FY12 dates are given as beginning October 1, 2013. Should that be 2011? The budget includes research assistant-ship and tuition for a Ph.D. student – essentially a half time position dedicated to this research. This is a cost efficient use of funds.


Science Coordinator Recommendation: Fund

Science Coordinator Comments: I concur with the Science Panel's comments. The PI's identified are skilled and well-respected in their field and will bring valuable experience to this complex project.


Public Advisory Committee (PAC) Recommendation: Fund

Public Advisory Committee (PAC) Comments: The PAC concurs with the Science Panel recommendation to fund the Branch modeling project. There were no objections.


Executive Director Recommendation: Fund

Executive Director Comments: (Not Available)


Trustee Council Decision: Fund

Trustee Council Comments: The PI has satisfactorily addressed the questions raised by both the Council and ADF&G. This project is approved for funding. Responses: “PWS herring research and monitoring program – modeling the population dynamics of PWS herring” response to ADF&G agency staff comments on proposal Exxon Valdez Oil Spill Trustee Council, project number 12120111-Q Trevor A. Branch, tbranch@uw.edu This document summarizes changes to the proposal made in response to comments from ADF&G staff. Many valuable comments and clarifications were made, pointing out some parts of the proposal that, in hindsight should have been expanded or simply written better. Throughout, though, the intent was to retain flexibility in the modeling approach that could be used, under the assumption that different types of models would be employed to examine different explanatory factors for the decline of PWS herring. Substantive comments are interspersed with responses. “we would have liked to have seen a more overt mention of how competing drivers of herring mortality will be tested against one another” As suggested by the Science Panel, we intend to compare different drivers of herring mortality (physiological stress, disease, starvation, top-down predation, disease, competition with hatchery salmon, EVOS, etc.) using model fits to the time series of data. Models will be fit in either a likelihood or Bayesian approach, depending on model complexity, allowing for comparison of competing hypothesis with likelihood ratio tests for nested models, or AIC for non-nested models. “It would be helpful to have more details on the “holistic” model… no clear description of how the proposed holistic life-stage model relates to or builds off of the ASA” A flexible modeling approach is envisaged, including a range of possible models from modifications to the existing ASA model, to a mechanistic model of each age class, to individual-based modeling, to area-based modeling. Mechanistic models would delve into the particular factors influencing mortality of each age class, to provide predictions that are age-specific and can be compared to observed data on age composition. Individual-based modeling using super-individuals and incorporating detailed space-based information, and variation among individuals in disease-resistance, could provide insight into why spawning has shifted from one area to another. Much of the change in the population consists of the shift in spawning areas from the Northeast region (1970s) to the North (1980s) followed by a collapse in the North and Northeast spawning areas (1990s) and replacement by spawning mainly around Montague Island and the Southeast (Pearson et al. in press). At a coarser scale, aggregating data by these larger areas would allow for intermediate modeling between individual-based and the current ASA model, by accounting for area-specific differences in biological characteristics, disease prevalence, oil spill magnitude, and other factors. It is a near-certainty that the modeling results will feed back into continued development and expansion of the ASA model to include additional relevant factors. “The disclaimer in the second paragraph of the Statement of the Problem is disconcerting: While we do not anticipate that there will be a major change in our modeling ability in the next five years, we expect that the combination of monitoring and focused process studies will provide incremental changes over the next twenty years and result in a much better understanding of herring populations by the end of the program.” This statement should be removed. It was part of the original overall project proposal, before the modeling component was added to address these exact concerns. The proposed modeling, management strategy evaluation and the meta-analysis of global herring populations will add substantially to model development and understanding of PWS herring dynamics. “Perhaps the proponent could offer a more detailed, though conditional, description of what the expected benefits might be.” 1.Modeling: a variety of models will be used to explore which hypotheses best explain the decline and subsequent failure to recover, of PWS herring. Including area-based modeling, mechanistic modeling, and possibly individual based modeling, together with additional modeling approaches to be determined, will greatly expand the range of modeling tools used. 2.Management strategy evaluation: this will be used to determine which data time series contain the most useful information for tracking trends in abundance, and thus provide information about the relative quality of different types of data. Tests include excluding one dataset at a time to see how each dataset influences the ability of management rules to ensure rebuilding while minimizing foregone catch. 3.Management strategy evaluation: the second component of this modeling will be to test and develop alternatives to the current management rules used to manage the fishery. 4.Meta-analysis of herring populations: this will result in estimates of the average duration of fisheries collapse for different herring population, the expected autocorrelation in recruitment, and hence a quantitative estimate of the probability of recovery of PWS herring based solely on other herring populations. “The order of the tasks is a bit confusing” The description of the tasks should have been arranged in the same order as the timeline, but was erroneously not. Clearly model development and enhancement should be first, since this may inform or provide alternatives to the ASA model. Management strategy evaluation should follow this modeling process, since MSE requires an underlying model to test datasets and management rules. The meta-analysis to predict recruitment could be conducted at any point, either before, after, or in parallel with the other two tasks. It is meant to provide a predictive assessment of the likelihood of recovery independent of data from PWS herring. Thus the order could be model, MSE, meta-analysis, or it could be meta-analysis, model, MSE. “It is not clear why a model will be developed first, and then a different model (ASA) used in the MSE” As indicated above, model development is first since this may suggest modifications to the ASA model used to manage the fishery at present. MSE will still use the ASA (together with alternative models) since this is the most realistic way to model actual management decisions, which are based on ASA. “The work to predict future recruitment, as described, appears correlational and doesn’t appear to involve the “holistic” model or a mechanistic understanding of herring dynamics” This meta-analysis is deliberately designed to borrow data from other herring stocks. Numerous herring stocks around the world have collapsed in analogous ways to the PWS herring stock, while some have also recovered. The RAM Legacy database (Ricard et al. 2011 in press) which I have made extensive use of, and contributed to (Branch et al. 2010, Branch et al. 2011), contains 23 stock assessments for Clupea pallasii or Clupea harengus. This database will be expanded to include more herring stocks, which will be used to answer the questions “What is the average duration of low biomass for herring stocks”, “How many years should we expect low recruitment to continue for?” and “What is the typical autocorrelation in recruitment among years”. This section is deliberately intended to be of global importance and independent of factors involved in the collapse of any particular herring stock such as the PWS stock. Another way of thinking about this is that this meta-analysis provides a “control”, thus if the average duration of collapse in other herring stocks is 5 years, the PWS collapse is certainly exceptional and strongly suggests some factor that is specific to PWS; but if the average duration of collapse in other herring stocks is 30 years, then the PWS collapse could be viewed as “typical” of herring stocks and requiring no particular explanation. “FY12 dates are given as beginning October 1, 2013” This is a typo and should have been October 1, 2011. Additional information and clarification is welcomed. References Branch, T. A., O. P. Jensen, D. Ricard, Y. Ye, and R. Hilborn. 2011. Contrasting global trends in marine fishery status obtained from catches and from stock assessments. Conservation Biology 25:777-786. Branch, T. A., R. Watson, E. A. Fulton, S. Jennings, C. R. McGilliard, G. T. Pablico, D. Ricard, and S. R. Tracey. 2010. The trophic fingerprint of marine fisheries. Nature 468:431-435. Pearson, W. H., R. B. Deriso, R. A. Elston, S. E. Hook, K. R. Parker, and J. W. Anderson. in press. Hypotheses concerning the decline and poor recovery of Pacific herring in Prince William Sound, Alaska. Reviews in Fish Biology and Fisheries DOI 10.1007/s11160-011-9225-7. Ricard, D., C. Minto, J. K. Baum, and O. P. Jensen. 2011 in press. Examining the knowledge base and status of commercially exploited marine species with the RAM Legacy Stock Assessment Database. Fish and Fisheries DOI: 10.1111/j.1467-2979.2011.00435.