NOAA – National Marine Fisheries Service Fisheries and the Environment (FATE) FY 2009 Annual Report September 2009
Executive Summary
The Fisheries and the Environment (FATE) program is a research program that develops and evaluates ecological and oceanographic indicators to be used to advance an ecosystem approach to management by improving fishery stock assessments and integrated ecosystem assessments. This information is necessary to effectively adapt management to mitigate the ecological, social, and economic impacts of major shifts in the productivity of living marine resources.
Sufficient understanding of the influence of species interactions and their responses to environmental change is required to develop forecasts and assess the long-term impact of climate and fishing on marine ecosystems. The goal of the FATE program is to provide the information necessary to effectively forecast these changes in order to evaluate management strategies needed to sustain fisheries, while preserving ecosystem structure and function. In support of this goal, the FATE program was designed to accelerate the development of next generation decision analysis and forecasting tools. FATE is providing leading indicators of ecological and oceanographic change by supporting research on the functional relationships between environmental forcing, competition for prey, or predation on the growth, distribution or reproductive success of managed species. The ultimate measure of FATE success is the incorporation of indices and functional relationships in population dynamics models, ecosystem assessments, and evaluation of fishing and climate forcing on marine resources that can be used to inform mangers of the implications of their actions.
FATE supports applied science with a focus on ecosystem indicators and forecasting tools that can be utilized to proved advice to managers. FATE accomplishes this through a core team of scientists distributed among NMFS research centers and projects selected from an annual request for proposals. This report compiles the individual annual reports of the three full time FATE scientists and the nine projects that received funding in FY 2008 and conducted research through FY 2009. In addition, eleven new projects received funding starting in FY 2009 as a result of the 2009 request for proposals. These eleven projects did not submit an annual report for 2009, as they have just been initiated, but are listed at the back of this report. Finally, this report summarizes how FATE research products are being utilized to aid living marine resource management.
More details on FATE and its products are available online at http://fate.nmfs.noaa.gov/ Table of Contents
Page
PIFSC FTE annual report – Evan Howell (NMFS-PIFSC) ...... 4
SWFSC-ERD FTE annual report – Andrew Leising (NMFS-SWFSC) ...... 7
SWFSC-FRD FTE annual report – Sam McClatchie (NMFS-SWFSC) ...... 12
07-01 Developing zooplankton and larval indices for use in Atlantic herring (Clupea harengus) stock assessments - Jon Hare, William Overholtz, and David Richardson (NMFS-NEFSC), Kenneth Able (Rutgers Univ.) ...... 14
07-03 Decadal and basin scale oceanographic variability and pelagic fishery production in the Gulf of Mexico - William Richards and John Lamkin (NMFS-SEFSC), Andrew Bakun (Univ. Miami), Barbara Muhling (Murdoch Univ.) ...... 18
07-05 Developing recruitment forecasts for age structured flatfish stock assessments in the eastern Bering Sea based on models of larval dispersal - William Stockhausen, Thomas Wilderbuer and Janet Duffy-Anderson (NMFS-AKFSC), Albert Hermann (OAR-PMEL-JISAO) ...... 26
08-01 Development of biological and physical indices for stock evaluation in the Dry Tortugas pink shrimp fishery - Joan Browder (NMFS-SEFSC), Maria Criales, Claire Paris and Laurent Cherubin (Univ. Miami) ...... 47
08-02 Predictors for larval transport success and recruitment for cod in the Gulf of Maine - James Churchill and Loretta O’Brien (WHOI), Jeffery Runge (Univ. Maine), Jim Manning (NMFS-NEFSC) ...... 65
08-03 Developing ecological indicators for spatial fisheries management using ocean observatory define habitat characteristics in the Mid-Atlantic Bigth - John Manderson (NMFS-NEFSC), Josh Kohut (Rutgers Univ.), Matthew Oliver (Univ. Deleware) ...... 70
08-04 An integrated assessment of the physical forcing and biological response of the pelagic ecosystem of the northern California Current, focused on coastal pelagics and salmon - Bill Peterson and Robert Emmett (NMFS-NWFSC) ...... 89
08-06 Developing statistically robust IPCC climate model products for estuarine- dependent and anadromous fish stock assessments - Frank Schwing, Steve Lindley, Roy Mendelssohn and Steve Ralston (NMFS-SWFSC), Eric Danner and Bruno Sanso (Univ. California Santa Cruz), Jim Overland (OAR-PMEL), Muyin Wang (Univ. Washington- JISAO) ...... 95
2 08-07 Integrating indicators: co-variation, periodicity and amplitude of key biological signals from central Oregon and northern California - William Sydeman, Alec MacCall, Kyra Mills and Julie Thayer (Farallon Inst.), Steven Bograd and Frank Schwing (NMFS- SWFSC), Chet Grosch (Old Dominion Univ.) ...... 100
FATE projects initiated in FY 2009...... 101
Summary of FATE products ...... 102
3 Project Title: PIFSC FATE FTE
Principal Investigator: Evan Howell
RECENT PROJECTS, PRODUCTS AND INDICATORS:
Oceanographic effects on dive behavior of loggerhead turtles ascertained from satellite telemetry results: Data from 17 satellite telemetry tags affixed to juvenile loggerhead turtles were analyzed to assess the oceanographic effects on the dive behavior of these animals. Multivariate classifications of dive data revealed four major dive types, three associated with deeper, longer dives and one with shallow dives shorter in duration. Turtles exhibited seasonal and regional variability in these dive types. The results of this study are contained in a manuscript that is currently in internal NMFS review and will be submitted to the Journal of Animal Ecology.
Results from these dive data for the central North Pacific during the first quarter of the year were also incorporated into the NOAA TurtleWatch product. The expected number of dives > 15 m were estimated for discrete sea surface temperature (SST) bins. These were combined with historic longline gear locations to model the potential number of dives to depths where longline gear may reside. The hypothesis was that these results would in explaining more of the variability in juvenile loggerhead bycatch occurred and be used to refine further the EOD TurtleWatch product. To test this hypothesis, a joint probability model was constructed using available fishery dependent (logbook) and independent (telemetry) data and compared with reported loggerhead interaction rates from 2005-2008.
EOD TurtleWatch product (FY 2009 update): Background: The PIFSC TurtleWatch product is a map providing up-to-date information about the thermal habitat of loggerhead sea turtles in the Pacific Ocean north of the Hawaiian Islands. Distributed daily both electronically and in paper format to individuals and entities in the Hawaii Longline Fishing industry. This product is aimed to minimize bycatch of loggerhead turtles by the longline fishery by indicating the SST values where the highest probability of a loggerhead turtle interaction may take place.
Revision: The results from the satellite telemetry-based dive data were used to model the expected number of loggerhead interactions. The dive results showed that turtles have a fairly uniform distribution in regard to dive behavior across the North Pacific during the first quarter of the year. Due to the distribution of the data, the addition of the dive data did not greatly enhance the ability of the TurtleWatch model to predict the areas of loggerhead bycatch. Additional work using in-situ data from research cruises to this area in the first quarter will be combined with remotely-sensed environmental data to continue to understand the mechanisms for why bycatch may occur in this research and continually attempt to refine and enhance the TurtleWatch model.
The TurtleWatch model was continually distributed to the industry and managers throughout the fishing season by the PIFSC as well as to fishers by the private company GeoEye. This product has been mentioned by PIFSC stakeholders as a flagship ecosystem indicator product produced by the center.
4
Modeling of the potential vulnerability of bigeye tuna in the Hawaii-based longline fishing grounds (Reported FY2008, Update FY2009): The manuscript detailing these results were accepted with minor revision by Progress in Oceanography as part of a larger body of work on dive behavior of bigeye tuna. The results of this work included the modeling of potential vulnerability of bigeye tuna to longline gear based on surface temperature, time and space. The results from this project were recently used by managers in addition to other factors to attempt to explain the decrease in catch rates in bigeye by the Hawaii-based pelagic longline fishery.
Changes among the upper trophic levels in the central North Pacific subtropical gyre ecosystem, 1996-2006: This was collaborative work with Jeffrey Polovina and others within the center to research changes in the catch rates of the Hawaii-based longline fishery over the last 10 years. This study looked at time series analysis results for the 13 most abundant species caught in the deep-set Hawaii-based longline fishery over the past decade. These results provided evidence of a change at the top of the North Pacific subtropical ecosystem. Catch rates for apex predators such as blue shark, bigeye and albacore tunas, shortbill spearfish, and striped marlin declined from 3 to 10% per year while catch rates for 4 mid-trophic species, mahimahi, sickle pomfret, escolar, and snake mackerel increased by 10 to 18% per year. The mean trophic level of the top 13 species in the catch declined by 5% from 3.85 to 3.66, and the mean production to biomass (P/B) value increased by 21%, from 0.80 to 0.97.
These results and methodologies are to be used as the foundation for a currently funded FATE research project that will incorporate these fisheries data with an updated ecosystem model of the central North Pacific Ocean to construct ecosystem indicators based on the time series of abundance and size of these fish species important to this North Pacific system.
FY2009 EDUCATION: Evan Howell received his Ph.D. in fisheries science from Hokkaido University in June 2009. His dissertation was entitled Satellite-based horizontal and vertical habitat estimation for loggerhead turtles (Caretta caretta) and bigeye tuna (Thunnus obesus) in the North Pacific Ocean.
FY2009 PUBLICATIONS:
Howell EA, DR Hawn, and JJ Polovina. In Press 2009. Spatiotemporal variability in bigeye tuna (Thunnus obesus) dive behavior in the central North Pacific Ocean. Prog. in Oceanogr. (CLIOTOP special issue).
Howell, EA. 2009. Satellite-based horizontal and vertical habitat estimation for loggerhead turtles (Caretta caretta) and bigeye tuna (Thunnus obesus) in the North Pacific Ocean. Dissertation. 156 pp.
Howell EA, JJ Polovina, GH Balazs, DM Parker, GH Balazs, H Bailey and PH Dutton. In NMFS internal review, to be submitted July 2009. Dive behavior of juvenile loggerhead turtles (Caretta caretta) in the North Pacific Ocean: responses to regional and seasonal changes in oceanography.
5 Polovina, JJ, M. Abecassis, EA Howell and P Woodworth. Accepted June 2009. Changes among the upper trophic levels in the central North Pacific subtropical gyre ecosystem, 1996-2006. Fishery Bulletin.
FY2009 PRESENTATIONS:
July 2008. Pelagic Fisheries Oceanography at the Pacific Islands Fisheries Science Center. PIFSC external review, Honolulu, HI. Oral presentation
December 2008. North Pacific circulation, productivity and migration. NOAA marine debris workshop, Honolulu, HI. Oral presentation
October 2008. TurtleWatch: a tool to aid in the bycatch reduction of loggerhead turtles Caretta caretta in the Hawaii-based pelagic longline fishery. PICES annual meeting, Dalian, China. Oral presentation
October 2008. Application of RS/GIS in Marine Protected Area delineations and tracking of marine organisms. World Fisheries Conference GIS training course, Tokyo, Japan. Oral presentation
September 2008. Small turtles, large journeys: Tracking juvenile loggerhead migrations across the North Pacific. Hanauma Bay lecture series, Honolulu, HI. Oral presentation
6 Project Title: SWFSC-ERD FATE-FTE
Principal Investigator: Andrew Leising, NOAA-SWFSC-ERD
FY2009 FATE Projects:
1) Copepod dormancy and population dynamics within the CCS and NWA 2) Development of copepod egg production index for the Southern California Current 3) Further development of BEUTI (Biologically Effective Upwelling and Transport Index) 4) The swimming and feeding response of copepods to microscale structure
Goals:
Project 1) The overall goal of this project is to understand how dormancy in Calanus spp. is controlled, particularly with relevance to environmental variables. Once this is understood, modeling can be used to make improved hindcasts and predictions of copepod abundance and productivity. This information is critical for understanding how climate influences the productivity of copepods, and hence their availability as prey for higher trophic levels, such as small pelagics and the larvae of larger fish species.
Project 2) The primary goal of this project is to develop an index of potential copepod reproductive success for the Southern California Current. Specifically, the project is centered on Calanus pacificus, which is the dominant calanoid copepod within the CALCOFI region. An overall goal is to understand how vertical water column structure affects the reproductive success of copepods over short time periods, and the resulting availability of the copepods as prey to surface feeding fish species.
Project 3) The overall goal of this project is to extend the utility of the upwelling index into a more biologically relevant framework. This new index may then be used for coupling with other models or in ways similar to current uses of the upwelling index, although with the added utility that this new index is more indicative of primary production.
Project 4) The overall goal of this project is to understand how microscale gradients in bio- physical water column structure (i.e. chemical gradients, shear, etc.) affect the short term feeding and swimming responses of copepods over brief (hours) timescales, thereby affecting their availability as prey to surface feeding planktivorous fish species.
Approach:
Project 1) During this final “pan-regional-synthesis” phase of the project, the objectives are to A) compare the dormancy responses of C. pacificus and C. marshallae from the Pacific with C. finmarchicus and C. helgolandicus from the North Atlantic, B) Compare the dormancy and population dynamics of C. finmarchicus and C. helgolandicus between the northwestern Atlantic and the eastern Atlantic, C) Further develop the IBM for inclusion of C. helgolandicus and thus
7 allow for modeling comparisons of all 4 species, and D) make projections of how all 4 species might respond to future climate variability.
Project 2) The first objective of this study is to develop an individual-based model of C. pacificus swimming and egg production in response to water column temperature and food levels, and compare the output of the model with shipboard experimental data on C. pacificus egg production. The second objective is to use the developed IBM to hindcast copepod reproductive success for the CALCOFI time series (1984-present) and then use this output to explore how regional scale and long-term changes in water column structure may affect the reproductive success and prey availability of copepods to higher trophic levels throughout the CCS. The third objective of this project is to compare copepod productive “hot spots” as identified within objective 2, with data on larval fish abundance and small pelagic egg production (such as CUFES data).
Project 3) The initial objective is to develop a simple model for predicting sea surface phytoplankton (~Chl-a levels) based on the upwelling index, for specific regions. This includes an analysis of the correlation between the “raw” upwelling index versus Chl-a for comparison with the simple model index. The second objective is to compare the best-fit parameters for different regions (i.e. differing upwelling regimes) for insight as to the mechanisms controlling primary production. The third objective is to expand the model to include additional environmental parameters, such as light levels and water column temperature, and to assess the improvement to the model versus the added complexity.
Project 4) The objectives of this study are: first, to observe the responses of target copepod species to microscale gradients of biophysical properties, second, to develop models of these behaviors, and third, to use models to compare predicted and observed field distributions of copepods. Further, the model will be used to predict reproductive success, and spatial location of the copepods, with relevance to their proximity to the vertical habitat partitioning of larval and small pelagic fish. Parts of this modeling effort will be combined with Project 2. Certain aspects of both of these models will also be used to improve the short time-scale portions of the full population dynamics IBM developed for Project 1.
Work Completed:
Project 1) As part of the previous GLOBEC-funded work, dormancy climatologies for C. marshallae and C. pacificus at four locations from the CCS, and C. finmarchicus from 5 locations across the Gulf of Maine/Scotian shelf have been compiled, along with the relevant physical (temperature and water column structure) and biological (Chl-a) parameters for the chosen target locations. The IBM has now been parameterized for all four species, except we are still waiting for more data on growth rates of C. helgolandicus from our European collaborators. Modeling runs for matching with observed climatologies have been conducted for most stations. Runs for two stations from the NW Atlantic using IPCC estimates of climate warming over the next 50 years have been conducted. From this analysis, a major finding has been the identification of the importance of a fall recruitment period to the dormant pool by C. finmarchicus in setting up the available females for the more major springtime production event. Under the climate-warming scenario, we found a large impact on this fall event, which in turn
8 has large impacts on springtime productivity. Extending this analysis, given the expected warming beyond the 50 yr time horizon, we speculate that potentially the fall recruitment to the dormant pool may fail, and thus cause a population crash for springtime C. finmarchicus production across Georges Bank. Potentially, the niche gap that this might create could be filled by the “warmer” congener, C. helgolandicus.
Project 2) The IBM model of C. pacificus swimming, feeding, and egg production as a function of food and temperature has been developed. The model has been run for all CALCOFI stations from 1984-2007 (7647 runs). Output has been compared with shipboard experimental data; the results show that the model output of egg production rate is significantly correlated with the experimental data, thus validating the model. Hot spots of copepod egg production have been identified as occurring near Pt. Conception, and down to the northern Channel Islands, and occasionally both near and southwest of the southern Channel Islands. Comparing CALCOFI lines 77, 80, and 90, line 77 has the highest yearly potential egg production. Examining the hindcast of egg production, Lines 77 and 80 show a marked switch to higher values after recovery from the El Nino of 1997 (summer of 1999), as seen by examining the long term (1984- 2007) EPR anomaly, which coincides with the switch of the PDO anomaly from positive to negative at this same time. Line 90 shows a switch to higher EPR after 2005/2006, which coincides with a switch from positive to negative anomaly for the NPGO, however, there was also an increase in chi from negative to positive anomaly at this same time, which likely explains the changes along line 90. Analysis of predation potential indicates that when the subsurface Chl maximum is shallower than the MLD, mortality losses are likely 5-6 fold higher, than when the Chl max is below the MLD. However, EPR is higher when the SCM is at or above the base of the mixed layer.
Project 3) The initial simple model has been created and optimized for the Oregon Upwelling Zone. The best-fit correlation between the SEAWIFS Chl-a and the “raw” upwelling index was found via lagging the upwelling index by 8.7 days versus the log of the Chl-a cycle, however, the correlation between the two time series is weak. Alternatively, the simple model produces a Chl index with a highly significant correlation to the observed SEAWIFS data. Additionally, the variance of the index is more similar to the observed data, and the index is able to reproduce many salient features of the observed data, particularly year-to-year variance in the timing of the spring bloom, and the length of the productive season. The next step is to optimize the model for additional regions to assess its general applicability.
Project 4) The general IBM model of copepod swimming and response to chemical cues, physical gradients etc., has been developed. The model can be run in both 2D and 3D modes, and swimming responses can be triggered by multiple cues. Experiments to examine copepod response to stimuli are underway (at G. Tech). Initial model runs have been compared to the statistics of copepod swimming determined from past experiments based on Acartia tonsa, thus partially validating the model. The model is also able to statistically reproduce the results from an earlier 2D model, but now within a fully 3D context. Additionally, this same group of collaborators has submitted a proposal to add an examination of the effects of turbulence on copepod vertical distribution. If accepted, the work from that project will be integrated with the current one, allowing the further creation of an index of copepod availability as a prey source, versus turbulent mixing levels.
9
Applications: All of the above projects have potential use for any stock assessment model or other higher trophic level models (e.g. ECOSIM, etc.) that use a planktonic prey source as input. The utility is increased if the stock concerned is known to directly prey upon calanoid copepods, however, the indices of EPR and copepod abundance may also be used somewhat as “ecosystem” indicators of the general planktonic productivity of the system. The results from project 1 have produced several useful results that may indirectly be useful for management decisions, i.e. the importance of the fall recruitment period for copepods on Georges Bank, and the predictive model for copepod dormancy timing for the Oregon upwelling zone. The real utility of these models, however, lies in their use for IEA purposes; understanding the responses of plankton to the current climate, in the context of past and future climate variability, is a key element for describing the status of the ecosystem. As calanoid copepods are the dominant mesozooplankton within the CCS, and a key prey species for many managed fisheries species, the above models can all significantly contribute to the construction of the CCS IEA; developing indices for the use in IEA construction is a major goal of the FATE program.
Publications/Presentations/Webpages: Project 1) Two manuscripts detailing this project are currently undergoing internal review by NOAA staff the NOAA-external collaborators before submission for publication:
Effects of climate variability on dormancy and population dynamics of Calanus pacificus and Calanus marshallae within the California current, I: An inter-regional data comparison. Andrew W. Leising, Cindy Bessey, Catherine Johnson, Jeffrey Runge, William Peterson, Moira Galbraith, and Dave Mackas
Effects of climate variability on dormancy and population dynamics of Calanus pacificus and Calanus marshallae within the California current, II: application of an individual-based model. Andrew W. Leising, Catherine Johnson, and Jeffrey Runge.
Website: http://www.usglobec.org/workshops/team_meetings/PRS_CLH/PRS_CLH.html has been created to document the work on this project and for sharing results with other researchers from the GLOBEC pan-synthesis phase.
Oral Presentation: Why doesn’t Calanus marshallae live in the Atlantic: a comparison across the copepod genus Calanus’ physiological rates with implications for mortality rates under climate variability. Leising, A., Pierson, J., Runge, J., Johnson, C., and Maps, F. GLOBEC Open Sciences Meeting, Victoria, BC Canada, June 2009.
Oral Presentation: Population responses to environmentally forced shifts in timing of diapause in Calanus finmarchicus in the Gulf of Maine. Runge, J., Leising, A., Johnson, C., and Maps, F.. GLOBEC Open Sciences Meeting, Victoria, BC Canada, June 2009.
Project 2) A manuscript entitled, “Climate-driven changes in copepod potential egg production rates within the southern California Current “ is nearing completion for submission for publication by Sept 30, 2009.
10
Oral Presentation: Consequences of Changes in Mixed-Layer Structure on Copepod Foraging and Production. Leising, A., Pierson, J., and Frost, B. EPOC conference. Stanford Center at Fallen Leaf Lake, Sept, 2008.
Poster Presentation: Foray foraging behavior of copepods: An empirical and numerical study. Pierson, J. J., Frost, B.W., Leising, A.W. ASLO Aquatic Sciences Meeting, Nice, France. Jan. 2009
Related publication (includes background work conducted to provide behavioral estimates for models): Pierson, J. J., B. W. Frost, D. Thoreson, A. W. Leising, J. R. Postel, & M. Nuwer. 2009. Trapping migrating zooplankton. Limnology and Oceanography Methods 7(5): 334-346.
Project 3) Results of this work will be presented at the 2009 FATE science meeting.
Project 4) This project is in progress.
11 Project Title: SWFSC-FRD FATE FTE
Principal Investigator: Sam McClatchie
Work Completed: Work has focused on 6 areas:
• Statistical characterization of sardine and anchovy habitat (refs. 5,3,1).
• Re-evaluation of the environmental index for sardine assessment (ref. 2).
• Effects of oxygen minima on fisheries and organisms (refs. 9,4)
• Annual preparation of the state of the California Current paper (refs. 7,8)
• Revamping of web site and development of new data portal for ichthyoplankton data (11,12)
• Completion of the Cruise report for the NMFS 2008 California current ecosystem Survey (ref. 10)
Applications: The most directly applicable work to stock assessment is a reassessment of the Scripps pier temperature index for sardine assessment (ref. 2). This showed that of the two fundamental relationships underpinning the original model (stock-recruit and temperature-recruit), the temperature –recruit relationship fails on the addition of more recent data, calling into question the continued utility of the environmental index for sardine assessment.
Publications/ presentation/ web pages:
In prep.
(1) Takasuka, A., McClatchie, S., Yoshioki Oozeki, Y. Spatial and environmental characteristics of spawning habitat of Japanese anchovy and sardine in the western North Pacific.
Submitted:
(2) McClatchie, S., Goericke, R., Auad, G. and Hill, K. Reassessing the temperature index for Pacific sardine (Sardinops sagax) stock assessment. Canadian Journal of Fisheries and Aquatic Sciences.
(3) Weber, E.D. and McClatchie, S.. Predictive Models of Northern Anchovy Engraulis mordax and Pacific Sardine Sardinops sagax Spawning Habitat in the California Current. Marine Ecology Progress series.
(4) McClatchie, S., Goericke, R. Rich Cosgrove, R. and Vetter, R. Oxygen in the Southern California Bight: multidecadal trends, impact of El Niño, and implications for demersal
12 fisheries. Limnology & Oceanography.
(5) McClatchie, S., Charter, R. and Ferguson, M.C. Predicting Pacific sardine (Sardinops sagax) spawning habitat in the California Current ecosystem. Canadian Journal of Fisheries and Aquatic Sciences.
In press:
(6) Weber, E.D. and McClatchie, S. rcalcofi: Analysis and visualization of CalCOFI data in R. CalCOFI Reports.
(7) McClatchie, S., Goericke, R, Koslow, A.J. and other authors. The state of the California Current, 2008–2009: Cold conditions drive regional differences in coastal production. CalCOFI Reports.
Published:
(8) McClatchie, S., Goericke, R, Koslow, A.J. and 24 other authors. 2008. The state of the California Current, 2007–2008: La Niña conditions and their effects on the ecosystem. CalCOFI Reports 49: 39-76.
(9) Vetter, R., Kohin, S., Preti, A., McClatchie, S. and Dewar, H. 2008. Predatory interactions and niche overlap between mako shark, Isurus oxyrinchus, and jumbo squid, Dosidicus gigas, in the California Current. CalCOFI Report 49, 2008: 142-156.
Reports:
(10) McClatchie, S. (Ed.) 2009. Report on the NMFS California Current Ecosystem Survey (CCES) (April and July-August 2008). NOAA Tech. Memo. 98 pp. NOAA-TM- NMFS-SWFSC-438.
http://swfsc.noaa.gov/textblock.aspx?Division=FRD&id=14568&ParentMenuId=610
Web sites and data portals:
(11) New Southwest Fisheries Science Center web pages for fisheries oceanography
http://swfsc.noaa.gov/textblock.aspx?Division=FRD&id=14560&ParentMenuId=610
(12) New web based GUI to access CalCOFI ichthyoplankton data (IchthyoDB) constructed in collaboration with Scripps Institution of Oceanography data informatics group
http://swfsc.noaa.gov/textblock.aspx?Division=FRD&id=15096&ParentMenuId=610
13 07-01 Project Title: Developing zooplankton and larval indices for use in Atlantic herring (Clupea harengus) stock assessments
1 2 3 Principal Investigators: Jonathan Hare , William Overholtz , Kenneth Able , and David Richardson4
1 – NEFSC, Ecosystem Processes Division, Oceanography Branch 2 – NEFSC, Resource Evaluation and Assessment Division, Population Dynamics Branch 3 – Rutgers University, Institute of Marine and Coastal Sciences, Rutgers University Marine Field Station 4 – joined project as NRC Post-doctoral Research Associate
Goals: The three primary goals of the project were: 1) To develop an index of larval abundance that could be used to calibrate the herring stock assessment 2) To quantify changes in the spatial patterns of herring spawning over time. 3) To determine the physical (e.g. salinity, temperature) and biological factors (e.g., zooplankton, egg predation) associated with year to year variability in herring recruitment, and multi-year fluctuations in herring populations Our 2008 report addressed work done towards accomplishing the first two goals. In this report we will provide a brief update of work on the first two goals, but will focus the report on the third goal.
Approach: Goal 1- We developed a non-linear least squares approach to generate our larval index for the Georges Bank spawning component of Atlantic herring. This method uses the age structure and abundance of larvae on multiple cruises to develop an annual index of abundance. One key question with this approach is the sensitivity of the index to inter-annual variability in the spawning season of Atlantic herring and the mortality rate of their larvae. We developed two different analyses to address this question. The first was a standard sensitivity analysis in which the larval index was recalculated with set changes in mortality and the seasonality of spawning. The second was a bootstrapping analysis, in which the larval mortality and spawning seasonality parameters were calculated from 5-year subsets of data
Goal 2- One question that remains is the relative contribution of spawning from Georges Bank (for which our larval index was developed) versus the Gulf of Maine, and the degree to which there is a coherence in larval abundance trends in both areas. Unfortunately the ichthyoplankton sampling in the Gulf of Maine has occurred less consistently and with stations spaced further apart than the sampling on Georges Bank. We are currently evaluating index development methods for the Gulf of Maine, as well as sampling strategies that may result in a better resolved Gulf of Maine index.
Goal 3- We chose to address the low frequency (>20 year) cycles associated with the collapse and recovery of Atlantic herring populations. Specifically, we are addressing the hypothesis that
14 the interaction of fishing mortality and egg predation drives the collapse of Atlantic herring populations. To quantify egg predation we made three assumptions: 1) haddock is the dominant egg predator in the region with other species being of minor importance, or varying little among years in their predatory impact, 2) haddock exhibit a type III functional feeding response when preying on herring eggs, and 3) the inter-annual variability in early stage larval herring abundance is primarily a function of herring spawning stock biomass and the intensity of haddock predation on herring eggs. A haddock predation intensity index was developed based on the haddock stock assessment abundance at age estimates, trawl survey weight at age data, and a function describing daily ration versus weight. Given the three assumptions we made, it is possible to develop a three parameter model (fitting predicted to observed larval abundance data) that can be used to describe interannual variability in Atlantic herring egg survival as a function of Atlantic herring spawning stock biomass and haddock predation intensity. Our final step in this analysis was to incorporate the egg survival equation into the stock recruitment relationship of Atlantic herring, and to subsequently run an age structured population model. Our approach makes the simplifying assumption that density dependent mortality occurs after the egg stage, and thus a decrease in egg survival has the same effect on recruitment as a comparable decrease in spawning stock biomass. Furthermore, for other parameters (e.g., growth, maturity, natural mortality) we choose values currently in use in the stock assessment. The primary objective of this population model is to explore the effects of haddock egg predation on the longer term cycles of herring populations, rather than the year to year variability in recruitment. Furthermore it allows us to look at the interaction of egg predation and fishing mortality on the stock.
Work Completed: Work officially started on the project in January 2008 when David Richardson started on an NRC post-doc. His funding extends until January 2010. Reports on this project will also be made in 2010.
Goal 1- The larval index methodology was submitted to ICES Journal of Marine Science. We are currently revising this manuscript. The index is now readily updated as new ichthyoplankton data is input into the database. One notable aspect of the larval index is the presence of two year to year drops of more than an order of magnitude. These occurred in 1975-6 and 2003-4. We are currently evaluating the cause of these drops (see Goal 3).
Goal 2- We are in the preliminary stages of evaluating herring larval abundance data in the Gulf of Maine, and the potential to develop a meaningful index from this data. Our preliminary analysis indicates larval abundance trends in the Gulf of Maine differ from those on Georges Bank.
Goal 3 - When the parameters of the predation model were solved for there was a very good fit between our predicted and observed larval abundance data. Moreover the predation model was able to capture two different order of magnitude declines in the abundance of larval herring on Georges Bank. The exclusion of 5-year continuous portions of data indicated that the model had good out of sample predictive capability. The results of this model indicate that there is extensive variability in herring egg survival rates, with high values around 70% in the early
15 1990’s (when haddock populations were low) and low values around 2% in 2004-2005 (when haddock populations were reaching very high levels). The age-structured population model for Atlantic herring, incorporating the egg survival equation, had a number of interesting characteristics. Most notable among these was the presence of an unstable equilibrium population level for Atlantic herring. A decline to a low stable equilibrium population level (i.e. a population collapse) is predicted when herring populations drop below this unstable equilibrium point, and an increase to an upper stable equilibrium is predicted when the population increases above it. The herring population level at which this unstable equilibrium point occurs increases as haddock predation intensity increases. This population model predicts that fishing mortality will interact with haddock egg predation to contribute to a stock collapse. That is, at a given level of haddock predation intensity and herring spawning stock biomass, the population may be predicted to increase when fishing mortality is low and collapse when fishing mortality is high. It is thus possible with this model to predict, on an annual basis, a threshold fishing mortality above which a population collapse is predicted. The model indicates that the increase in haddock population levels in recent years (due to the 2003 haddock year class) has resulted in this threshold being crossed (despite a low and constant F), with the resulting prediction that the Georges Bank Atlantic herring spawning stock will collapse. An evaluation of herring population trends using an average standardized anomaly of 14 different fisheries-independent time series indicates a downward trend in herring populations starting in 2006. These results match the predictions of our model.
Applications: The larval index developed as a part of this project was presented at the 2009 stock assessment update for Atlantic herring. We have been in discussions with the stock assessment scientist as to the best way to directly incorporate the index into the upcoming full assessment for Atlantic herring. The work on the egg predation hypothesis and the modified age-structured population model were presented at a meeting of the Ecosystem Assessment Program group in the NEFSC. Due to the important ecological role of herring, the results of this project should aid in a broader interpretation of long-term changes in the ecosystem. Finally, the results of this project have major implications for the management of Atlantic herring. Specifically, the modified population model questions the sole use of fixed Fmsy and F0.1 levels in managing and assessing Atlantic herring populations. Moreover the model can be used to calculate, on an annual basis, a threshold level below which fishing mortality should be maintained.
Publications Richardson, D.E., J.A. Hare, W.J. Overholtz, D.L. Johnson. In review. Development of long- term larval indices for Atlantic herring (Clupea harengus) on the northeast U.S. continental shelf. ICES Journal of Marine Science.
Presentations Richardson, D.E., Hare, J.A. (July 2009) Does haddock egg predation decouple the abundance of Atlantic herring larvae from spawning stock biomass on Georges Bank? Larval Fish Conference 2008-Portland Oregon
16 Richardson, D.E., Hare, J.A., Overholtz, W.J., Johnson, D.L. (June 2009) Development of long- term larval indices for Atlantic herring (Clupea harengus) on the northeast U.S. continental shelf. Transboundary Resource Assessment Committee Meeting –St. Andrews Canada
Richardson, D.E., Hare, J.A., Overholtz, W.J. (Aug 2008) Development of long-term larval indices for Atlantic herring (Clupea harengus) in the northwestern Atlantic Ocean. FATE Annual PI Meeting 2008 – La Jolla, California
Richardson, D.E., Hare, J.A., Overholtz, W.J. (Sept 2008) Development of long-term larval indices for Atlantic herring (Clupea harengus) in the northwestern Atlantic Ocean. ICES Annual Science Conference 2008 – Halifax-Canada
17 07-03 Project Title: Decadal and basin scale oceanographic variability and pelagic fishery production in the Gulf of Mexico – a synthesis of 30-years of data for management applications
Principal Investigators: William Richards, Barbara Muhling, John Lamkin, Andrew Bakun
Goals: Overarching goals and objectives of the project. The overall goal of this project was to incorporate environmental variability into existing larval fish indices, especially those for Atlantic bluefin tuna. Larval fish indices are currently used to refine stock assessment models, and are often the only fishery-independent indices available. We aimed to define habitat preferences of bluefin larvae within the Gulf of Mexico using archived CTD data, and to then construct a simple model to predict occurrence of bluefin larvae. Variability in the spatial and temporal extent of favorable conditions each year will be used to improve the existing larval bluefin tuna abundance index, and thus stock assessments. This method can be extended to other species, both exploited and unexploited. In addition, we also aimed to use larval fish assemblage data to quantify the influence of decadal-scale oceanographic processes on suites of larval fish assemblages and on the Gulf of Mexico ecosystem as a whole.
Approach: Description of the work that was performed
1) Habitat model for larval bluefin tuna Archived data from 24 years of SEAMAP spring plankton surveys in the northern Gulf of Mexico were used to create a model predicting suitable habitat for bluefin tuna larvae. Eleven environmental variables were initially chosen for examination as potential predictors of larval occurrence. Seven variables (temperature and salinity at the surface, 100m depth, and 200m depth, and standardized settled plankton volumes) were derived from archival CTD data, and three variables (water depth, longitude and latitude) related to position in the Gulf of Mexico. The larvae of other Atlantic tunas (Auxis rochei, Katsuwonus pelamis, Euthynnus alletteratus and other Thunnus species (which were likely a mix of yellowfin tuna (T. albacares) and blackfin tuna (T. atlanticus), but were distinct from bluefin tuna) were frequently collected in the SEAMAP surveys. The presence of any other tuna larvae in a sample was included as a categorical variable (presence or absence only).
Relationships between bluefin tuna larvae occurrences were initially defined using a simple preference model approach. Each variable was divided into between 10 and 20 bins, depending on the distribution of the data, and the proportion of stations within each bin which contained bluefin tuna larvae was calculated, and plotted. Where a trend was apparent, the significance of the relationship was tested using linear regression (if the trend was linear), or 2nd order polynomial regression (if the trend was curved). Cross-correlations between variables were calculated, using Pearson product moment correlations, and where two variables were highly correlated (R2>0.8), the weaker predictor was discarded from the model.
These relationships were used as a guide as to which variables to include in a classification tree model, using DTREG software. Classification tree modeling splits a dataset into increasingly
18 small and homogeneous subsets, with each split made using the variable which provides the greatest improvement in the homogeneity of the two resulting groups. In this study, the aim of this procedure was to define habitat conditions which had the highest likelihood of occurrence of bluefin tuna larvae. V-fold cross-validation, with 10 partitions, was used to validate the classification tree model. The accuracy of the resulting model was assessed by calculating the probability of larval bluefin tuna occurrence at each station, and then comparing these probabilities to the actual observed larval data. Data from all years between 1982 and 2006 were used to construct the classification tree model. As a visual aid, the generated probabilities for four example years (1983, 1990, 1997 and 2001) were contoured, and the bluefin tuna larvae catch locations for each of these years were overlaid.
2) Larval fish assemblage structure in the northern Gulf of Mexico Variability in larval fish assemblages has been defined for all stations sampled across the Gulf of Mexico since the early 1980s, using multivariate statistical software. Assemblage structure was examined between different zones in the Gulf of Mexico, and also through time. Abundances of all species, and of key families, were also plotted through time, and changes correlated to environmental variables.
Work Completed: Summary of progress, including results obtained / products developed, and their relationship to the goals of the project. If problems developed which resulted in unexpected results, they should be discussed.
1) Habitat model for larval bluefin tuna
Bluefin tuna larvae were most likely to be collected where sea surface temperatures were between 25 and 28°C, and where temperatures at 100m depth were between 18 and 23°C, although the trend for temperature at 100m was non-significant (Figure 1). In contrast, bluefin larvae showed a strong, linear preference for lower temperatures at 200m depth. Larvae were proportionally more abundant where salinities at the surface, and at 100m depth, were higher. Bluefin larvae appeared to prefer salinities at 200m to be around 36-36.5, however a strong outlier resulted in no significant trend being found for this variable. Stations with intermediate plankton volumes were most likely to support bluefin larvae, as were stations of intermediate depth. Bluefin tuna larvae were also more likely to be found at lower longitudes (stations farther west). In total, seven of the continuous variables showed statistically significant relationships to larval bluefin tuna occurrences: surface temperature, temperature at 200m depth, surface salinity, salinity at 100m depth, standardized plankton volumes, water depth, longitude. When added to the presence or absence of other tuna larvae, this gave eight variables as input parameters for classification tree analysis.
19 Sea surface temperature Surface salinity Plankton volume (log) 0.4 0.25 SST 0.3 SSS 0.2 0.3 0.25 Poly. (SST) Poly. (SSS) 0.2 0.15 Plankton Vol 0.2 0.15 0.1 0.1 Poly. (Plankton 0.1 0.05 0.05 Vol) 0 0 0 18 23 28 29 34 39 1 1.5 2 2.5 3
Te mpe r at ur e at 100m Salinity at 100m Depth 0.25 0.7 S_100 0.4 T_100 0.6 0.35 occupied 0.2 Linear 0.5 0.3 (S_100) 0.25 0.15 0.4 Water Depth 0.2 0.3 stations 0.1 0.15 Poly. (Water of 0.2 0.1 0.05 Depth) 0.1 0.05 y = 0.2252x ‐ 8.0826 0 0 R² = 0.8669 0 15 20 25 30 35 36 37 38 39 ‐4000 ‐3000 ‐2000 ‐1000 0 Proportion
Te mpe r at ur e at 200m Salinity at 200m Longitude 0.4 0.3 T_200 0.5 0.35 0.25 0.3 0.4 0.2 Linear Longitude (T_200) 0.25 0.3 0.15 0.2 S_200 0.2 0.1 0.15 Poly. 0.1 (Longitude) 0.05 0.1 0.05 0 0 0 10 15 20 25 30 35 35.5 36 36.5 37 37.5 38 38.5 ‐98 ‐93 ‐88 ‐83 ‐78
Latitude 0.3
0.25
0.2
0.15 Latitude 0.1
0.05
0 22 27 32
Figure 1: Modeling the response of bluefin tuna larvae to ten environmental variables, for all stations sampled 1982-2006. The probability of finding at least one bluefin tuna larvae is shown for each bin of the predictor variable. Where a significant linear, or 2nd order polynomial relationship was present, a line of best fit is shown.
With the exception of salinity at 100m depth, the strongest preferences within each variable were generally only around 30-40%. This suggested that even after a multivariate model had been formulated, a considerable proportion of favorable habitat was likely to be unoccupied by larval bluefin tuna. Throughout the larval surveys, bluefin larvae were most commonly found in numbers of <5 individuals per station, suggesting a sparse distribution, and a high likelihood that net samples may have missed larvae present in suitable habitat. We therefore aimed to use classification tree analysis to define the range of potentially suitable larval bluefin tuna habitat, rather than to define all negative stations as being in unsuitable habitat. To achieve this, the misclassification cost in DTREG was set to make false positives seven times more costly than false negatives.
The classification tree model was initially split by temperature at 200m (Figure 2), with higher temperatures less likely to be favorable for larval bluefin tuna. The next split showed that low sea surface temperatures (<23.4°C), were also unsuitable for bluefin tuna larvae. Subsequent splits showed that bluefin larvae were more likely to be found at lower (more western)
20 longitudes, and higher surface salinities. Plankton volumes and salinity at 100m depth contributed only slightly to the model.
Figure 2: Preferences of bluefin tuna, other Thunnus spp and Cubiceps spp. for sea surface temperature and temperature at 200m depth conditions over all stations sampled between 1982 and 2004. Note that temperature at 200m depth and salinity at 100m depth were 4th root transformed before analysis, plankton volumes were log transformed, and surface salinities were transformed to the power of 4.
The model was 88.1% accurate for stations containing bluefin larval, and 54.5% accurate for negative stations, reflecting the higher misclassification costs assigned to false positive results. To examine the results of the model within specific years, the generated probabilities for four years (Leg 2 surveys from 1983, 1990, 1997 and 2001) were contoured, and plotted, with locations of bluefin tuna catches overlaid (Figure 3). Generally, habitat within the Loop Current in the south-eastern Gulf of Mexico was less suitable, as were warm-core rings, and cooler water on the continental shelf. The location and size of suitable habitat was highly variable between years, which was reflected in the locations of larval bluefin tuna catches. However, as expected, a substantial proportion of theoretically suitable habitat was negative for larval bluefin tuna. Nevertheless, bluefin tuna larvae were found at 33% of “suitable” stations.
21 1983 1990
Probability of larval bluefin occurrence (%)
1997 2001 Larval bluefin catch location Sampled station
Figure 3: Probability of larval bluefin tuna occurrence from the classification tree model, plotted for the second cruise leg of four separate years. Sampled stations are shown by crosses, larval bluefin tuna catch locations are shown by circles.
2) Larval fish assemblage structure in the northern Gulf of Mexico
Larval fish assemblage data were available for all SEAMAP stations in the northern Gulf of Mexico, for each year between 1982 and 2007, with the exception of 1985. We employed multivariate statistical analyses to examine spatial and temporal trends in family-level assemblages. The Gulf of Mexico was split into three zones, reflecting the predominant oceanographic conditions found: the South zone, the West zone and the North zone (Figure 4). Larval fish assemblages were distinct between each zone, when data from all sampled years was included. A directional change in assemblages through time was also noted, as was a general increase in the mean number of larval fish collected per sample (Figure 5). This increase was strongest for bathypelagic families, such as Myctophidae and Gonostomatidae, but not evident in demersal families. An associated temporal change in surface temperatures in the Gulf of Mexico (from buoy data), or in settled plankton volumes, was not evident. This suggested that populations of bathypelagic fishes were increasing in response to some ecological driver in the Gulf of Mexico that we did not have the data to detect. The strongest increases were observed in continental shelf waters, and we hypothesize that coastal eutrophication may have been a contributing process. Our results show the promise of unexploited, but abundant, bathypelagic larval fish as indicators of ecosystem change.
22 Te mpe r at ur e (°C) 30
26.5 28 26 26 25.5
24 25
22 Sampled station 24.5
20 24
18 23.5 Plankton volume (cm3 100m‐3) Latitude 30
80 28
72 26
64 24
North 56 22 South West 20 48
18 40 Longitude Figure 4: Sampling effort and biophysical parameters in the Gulf of Mexico SEAMAP program, 1982-2007. Top: All stations sampled 1982-2007, overlaid on mean sea surface temperature from CTD (all years). Bottom: 32 stations chosen for assemblage analyses, nominally divided into three zones, overlaid on mean plankton volume per station.
23 1600 Mean Larval Concn per Sample ) 2 ‐ 10m
(no. 1200
sample
per
800 larvae
of
400 concentration
Mean
0 1980 1985 1990 1995 2000 2005 2010
2D Stress: 0.14 1997 2D Stress: 0.2 1991 1997 1984 1995 1989 1994 1993 1999 1992 1987 2002 1999 1983 2005 1996 1990 2000 2002 1998 1992 2000 1983 1987 1996 2006 1991 2006 2005 1986 1986 1994 1993 1995 1989 1984 1998 1990 2007 2007 All samples North
1984 2D Stress: 0.19 1983 2D Stress: 0.12 2006
2005 1994 1986 1999 1990 1995 2000 19841995 1998 1989 2007 1993 1990 1998 1987 2002 1997 1986 1987 1996 1992 2005 2006 1993 1992 1994 2000 1996 1991 1999 1989 1997 Decade 2002 80s 1991 1983 90s South West 00s Figure 5: Top: Mean concentration of larval fish per sample (no. 10m-2) per year from 1983 to 2007 in the Gulf of Mexico, with standard errors shown. Bottom: Multidimensional Scaling (MDS) ordination of assemblage data for each zone of the Gulf of Mexico, averaged across latitude and longitude for each year, and then by year (1983 to 2007). For clarity, points are coded to show the 1980s (1983-1989), 1990s (1990-1999) and 2000s (2000-2007).
24 Applications: Describe how results are being used in stock assessments or otherwise benefiting living marine resource management. We are working with stock assessment scientists from the Southeast Fisheries Science Centers in Miami and Pascagoula to incorporate the results from our model into bluefin tuna stock assessments. Annual catches of larval bluefin tuna can be weighted based on the nature of the habitat that was sampled, and the percentage occupancy of suitable habitat can be calculated, and compared between years. This will give stock assessment scientists a better annual indicator than the current larval index, and will help to lower to variance of the index. We are also currently discussing the results of the larval fish assemblage work with physical oceanographers and other scientists working in the Gulf of Mexico. We aim to compare assemblage patterns to existing physical datasets, and explore the possibilities of using assemblages as a multivariate indicator of the planktonic environment in the Gulf of Mexico.
Publications Muhling, BA, Lamkin, JT, Richards, WJ (in final prep) Decadal-scale patterns and changes in larval fish assemblages across the northern Gulf of Mexico: 1983-2007, Marine Ecology Progress Series. Muhling, BA, Lamkin, JT, Roffer, M (in final prep) Modeling suitable habitat for bluefin tuna larvae in the northern Gulf of Mexico using archival data, Marine Ecology Progress Series.
Presentations Muhling, BA, Lamkin, JT, Richards, WJ, Bakun, A (2007) “Bluefin Tuna larvae in the Gulf of Mexico: Analysis of twenty nine years of ichthyoplankton collections (1977- 2005)”, 1st international CLIOTOP conference - Climate Impacts on Oceanic Top Predators, La Paz, Mexico, 3-7 December 2007. Muhling, BA, Lamkin, JT, Richards, WJ (2008) Temporal and spatial patterns of larval Atlantic Bluefin Tuna in the Gulf of Mexico: Synthesis of thirty years of data, 32nd Annual Larval Fish Conference, Kiel, Germany, 4-7 August 2008. Muhling, BA, Lamkin, JT, Richards, WJ, Bakun, A (2008) Decadal and basin scale oceanographic variability and bluefin tuna fishery production in the Gulf of Mexico, Fisheries and the Environment (FATE) annual meeting, La Jolla, California, 13-14 August 2008. Muhling, BA, Lamkin, JT, Richards, WJ, Bakun, A (2008) Larval bluefin tuna in the Gulf of Mexico: 1982-2007, Bluefin Tuna research meeting, Miami, Florida, December 2008. Muhling, BA, Lamkin, JT, Richards, WJ, Bakun, A (2008) Larval bluefin tuna in the Gulf of Mexico: 1982-2007, Pre-cruise research meeting, Pascagoula, MS, March 2009.
25 07-05 Project Title: Developing recruitment forecasts for age structured flatfish stock assessments in the eastern Bering Sea based on models of larval dispersal
Principal Investigators: William Stockhausen, Thomas Wilderbuer, Janet Duffy-Anderson, Albert Hermann
Background: For the past decade, a simple 2-dimensional, wind-driven, surface current only model (OSCURS, the Ocean Surface Current Simulator) has provided a qualitative index of annual recruitment for winter-spawning flatfish on the eastern Bering Sea shelf (Fig.s 1-2; Wilderbuer et al., 2002; NMFS, 2007). The recruitment index is based on the trajectory over 90 days of a single simulated drifter released at the same time (April 1) from the same location (165° W, 56° N) each year. OSCURS calculates daily surface currents on a fairly coarse 90 km-scale grid by converting daily gridded surface pressure fields to wind-driven surface currents and combining them with long term mean geostrophic surface currents. The surface currents are used, in turn, to project the position of a simulated passive surface drifter over daily time steps. To obtain a qualitative estimate of annual recruitment strength for winter-spawning flatfish, the resulting trajectory is visually classified into one of three categories: "on shelf", "off shelf", and "mid shelf". For northern rock sole, expected recruitment is highest for trajectories (years) classified as "on shelf", while it is lowest for those classified as "off shelf" (Fig. 2). For Alaska plaice, the relationship is not as strong, with mean recruitment almost the same under the three classifications (Fig. 3). While OSCURS appears to give a satisfactory qualitative index of recruitment, allowing an analyst to determine which of three production regimes the ecosystem might be in for a given year, it was felt that OSCURS could be improved upon in a number of areas, including: 1) the physics incorporated in the currents model, 2) the biological realism incorporated in the drifter model, and 3) the nature of the recruitment index (i.e., quantitative rather than qualitative).
Goals: The principal goal of this project was to develop quantitative recruitment indices and forecasts for two of these flatfish, northern rock sole (Lepidopsetta polyxystra) and Alaska plaice (Pleuronectes quadrituberculatus), based on more detailed models for behavior of early life stages and patterns of 3-dimensional advective transport in the eastern Bering Sea. A secondary goal was to demonstrate the utility of these indices/forecasts in age-structured stock assessment models for these species. The quantitative recruitment indices were expected to provide an alternative to, and improvement on, the current qualitative indices.
Approach: Our approach to achieve the principal goal of the project was to use a coupled biophysical individual-based model (DisMELS: the Dispersal Model for Early Life Stages; Fig. 4) that incorporates ontogenetic changes in early life parameters and simulates egg and larval dispersal under 3-dimensional oceanographic currents to predict relative fluctuations in annual recruitment due to density-independent, advective processes on the eastern Bering Sea shelf. Simulated eggs or larvae (in the case species with of demersal eggs, such as northern rock sole) are released along the bottom throughout a pre-defined “spawning area” during the spawning season.
26 Individuals are tracked through time and space using Lagrangian particle tracking “with behavior”; i.e., individuals are allowed to undergo vertical migration: they actively swim up or down in the water column until they enter a “preferred” depth range (different ranges can be defined for daytime and nighttime, allowing diel migration to be incorporated). Individual growth and mortality can also be tracked. Ontogenetic changes in vertical migration behavior, as well as growth or mortality rates, can be incorporated in the model by defining a series of life history sub-stages that individuals progress through: different model parameters for different sub-stages yield ontogenetic changes in behavior, growth, and/or mortality. Individuals reaching the final larval sub-stage metamorphose into a “settlement-competent” stage that can “settle” to the benthos upon reaching pre-defined juvenile “nursery” habitat. Individuals in the settlement stage that do not reach suitable nursery habitat before the stage terminates (a pre-set amount of time) are regarded as having died. A relative quantitative index of annual recruitment can be generated by simulating thousands of individuals over one model year and calculating the fraction that are transported from the spawning areas to the nursery areas and successfully settle in that year. A time series can be generated by running the model for multiple years. DisMELS uses stored output of 3-dimensional oceanographic currents, temperature and salinity fields at daily time steps from a Regional Ocean Modeling Systems (ROMS) oceanographic model for the northeast Pacific Ocean (NEP4) to simulate advective transport of pelagic eggs and larvae. ROMS is a free-surface, hydrostatic, primitive equation ocean model that employs orthogonal curvilinear coordinates in the horizontal and stretched, terrain-following ("sigma") coordinates in the vertical (Haidvogel et al., 2008). The algorithms that comprise the ROMS computational nonlinear kernel are described in detail in Shchepetkin and McWilliams (2003, 2005). The NEP4 ROMS model is the regional, intermediate resolution (~10 km horizontal resolution, 42 stretched-coordinate layers in the vertical) component of a suite of nested computational domains extending from the basin scale to coastal regions (Fig. 5, Curchitser et al., 2005; Hermann et al., 2009). A coarse-grained basin-scale ROMS model for the North Pacific (NPac), with a nominal resolution of 0.4o and 30 vertical layers, was used to generate initial and boundary conditions for the NEP4 model (Curchitser et al., 2005). The two models were run for concurrent time periods. Daily surface forcing (winds and surface heat flux) was obtained from the NCEP/NCAR Reanalysis Project (Kalnay et al., 1996). Sea ice generation and dissolution was included in the model, but tidal forcing was not. To incorporate the resulting indices within the appropriate stock assessments, our approach was to modify the existing stock assessment models for Alaska plaice and northern rock sole to accommodate the annual recruitment indices as, effectively, recruitment surveys. The assessment models were to be modified to incorporate an additional likelihood component that allowed the assessment model to “fit” the recruitment indices. As will be shown below, the DisMELS recruitment indices for both northern rock sole and Alaska plaice exhibited poor agreement with estimates from the associated age-structured models used for stock assessment. As such, rather than integrating the DisMELS-generated recruitment estimates into the assessment models as we had proposed, we devoted our effort toward investigating why the DisMELS-generated indices showed such poor agreement with those from the assessment models.
Work Completed: Work was completed in the following project areas: acquiring ROMS oceanographic model output for the eastern Bering Sea, processing previously-collected egg and larval samples,
27 improving DisMELS model capabilities, generating initial recruitment indices for Alaska plaice and northern rock sole for the time period 1979-2004, performing sensitivity analyses with regard to the recruitment indices, and head-to-head comparisons between OSCURS model predictions and DisMELS.
ROMS output acquisition 3-dimensional oceanographic model results were acquired for the eastern Bering Sea at ~10 km horizontal resolution, 42 levels (stretched coordinates) in the vertical, and a daily time step for the time period 1979-2004 from prior runs of the NEP4 ROMS model at the Alaska Region Supercomputing Center. The resulting dataset is ~300 GB. The NEP4 ROMS model is part of a nested suite of models. Boundary conditions and forcing were provided by the basin-scale NPac ROMS model within which the NEP4 ROMS model was nested. The NPac ROMS model was forced with daily atmospheric variables from the NCEP/NCAR Reanalysis Project (http://www.cdc.noaa.gov/data/reanalysis/reanalysis.shtml) using bulk flux formulae, as well as freshwater inflow along the Alaskan coast modeled as point and line sources. Formation and melting of sea ice was included in the model dynamics, but tides were not. Although ROMS can be run in a sophisticated "adjoint" mode or in a simpler "nudging" mode, no such attempts to constrain the model's evolution with observed data were made during these model runs. Utilities were also developed to extract ROMS model output and visualize it in ESRI’s ArcGIS. Graphical user interfaces allow the user to extract a time series of horizontal slices of scalar (e.g., temperature) and vector (u,v currents) quantities as shapefiles, suitable for plotting on maps in ArcGIS. Displaying properly transformed vector quantities in ArcGIS necessitated development of a “custom renderer”, as well.
Egg/larval sample processing and data analysis All data for Alaska plaice egg, larval, and juvenile distributions collected under the EcoFOCI program at the Alaska Fisheries Science Center were examined, analyzed statistically, and graphically plotted. Maps of distribution were generated, and growth rates were estimated using back-calculations based on lengths. Data were summarized seasonally and interannually, and were provided for modeling efforts. Data included: vertical distribution, growth rate, time in plankton, seasonality of occurrence, purported spawning areas, presumed drift trajectories, and presumed juvenile habitat areas. A manuscript has been prepared (Duffy-Anderson et al., accepted) that describes and summarizes the results of the field investigations on Alaska plaice. Data on northern rock sole larvae and juveniles recently collected under the EcoFOCI were also analyzed to support modeling efforts. Data included seasonality of occurrence of pelagic larvae and juveniles (eggs are demersal), growth rate, vertical distribution, horizontal distribution, and juvenile (age-0) collections. These data were used, in part, to determine values for DisMELS parameters such as preferred diel depth ranges, stage durations, spawning areas and nursery habitats for species-specific model simulations (Table 1, Fig.s 6-8).
DisMELS model development The DisMELS modeling framework underwent substantial revision and improvement to facilitate multiple-year simulations, calculation of spatial patterns of connectivity between spawning and nursery areas, and visualization of results. Model classes representing adult and settlement-competent life stages were developed and incorporated within the modeling
28 framework. Creation of the “adult” class simplified the process of “injecting” simulated pelagic eggs or larvae into the model throughout extended spawning seasons and during model runs that extend over multiple model years. Creation of the “settlement-competent” class allowed more biological realism to be incorporated within the model with regard to the transition from the pelagic environment to the benthos for juvenile flatfish. "Settlement-competent" individuals have a time interval (a competency period or "window of opportunity") during which they can actively move from the pelagic environment to the benthos upon finding suitable nursery habitat; individuals that find suitable nursery habitat during the competency period are counted as successful recruits, individuals that do not find suitable habitat in time are counted as dead.
Model results We completed a suite of model runs to: (1) generate recruitment indices covering 1979-2004 for both northern rock sole and Alaska plaice and (2) explore the sensitivity of model results to underlying parameter values. The baseline model runs to generate the recruitment indices time series were run in similar fashion for both species. Based on field data, Alaska plaice eggs are found in the water column in the eastern Bering Sea starting at the beginning of April and for several weeks thereafter. This is also the time frame within which the smallest sizes of pre-flexion northern rock sole larvae are found in the water column; northern rock sole eggs are demersally attached, and hence non- dispersive. Consequently, we defined a similar “adult” life stage for each species to “inject” simulated pelagic stage individuals (eggs or larvae) into the model over the same extended “spawning” period of 35 days starting on April 15 of each year. For Alaska plaice, we assumed (with no evidence to the contrary) that pre- and post-flexion larvae had similar depth ranges and did not undergo diel vertical migration. Thus, we defined three ontogenetic pelagic stages for Alaska plaice: “egg”, “larva”, and “settler” (Table 1). For northern rock sole, it is clear that pre- and post-flexion larvae have different vertical behavior (Fig. 2), with post-flexion larvae undergoing diel vertical migration. Consequently, we defined three different pelagic stages for northern rock sole: “early larvae” (pre-flexion), “late larvae” (post-flexion), and “settler” (Table 1). In the eastern Bering Sea, juveniles of both species are generally found in presumed nursery areas inside the 50 m isobath; thus, we assigned the suitable depth range for settlement to be 0-50 m for both species. We allowed settlement competent individuals (“settlers”) 10 days from transformation in which to be transported within the 50 m isobath; “settlers” that did not reach suitable nursery habitat before the 10 days expired were then regarded as having perished. In each model year, approximately one thousand simulated adults were “injected” into the model on April 15 at uniform density across species-specific spawning areas (Fig. 7). The spawning areas for northern rock sole were based on an analysis of spatial fishing patterns from 1986-2004 during the late winter when a roe fishery for rock sole is active in the eastern Bering Sea, as well as an hypothesized area in the south of Unimak Pass (area 6 in Fig. 7). The presumed spawning area for Alaska plaice was a “best guess” as this stock is not targeted by a roe fishery. Each adult released a single egg (Alaska plaice) or pre-flexion larva (northern rock sole) at seven day intervals over the 35-day spawning season, for a total of about five thousand pelagic-stage individuals released over the course of the spawning season. We adopted this approach to average over environmental variability within the time period of spawning (or hatching, more correctly, for northern rock sole). The model was run for 120 days after the annual start date to allow all simulated individuals to complete the pelagic portion of their life cycle by reaching shallow nursery areas (Fig. 8) or perishing. For each model year, a
29 (quantitative) index of recruitment was calculated as the fraction of individuals that successfully “settled” in that year. Intermediate results from the baseline northern rock sole model runs are shown in Fig.s 9-11. Animations are available within model years for tracks of individuals as well as across model years for final locations classified by spawning area (as in Fig. 10) and starting locations classified by settlement area (as in Fig. 11). Animations are also available showing the change in the connectivity matrix (Fig. 12) between spawning areas and settlement areas with model year. The recruitment indices generated from the baseline DisMELS model runs exhibit substantially different patterns for northern rock sole and Alaska plaice both spatially and temporally (Fig.s 13-14). Model results suggest settlement for Alaska plaice is concentrated in a counterclockwise arc from the eastern part of the Alaskan Peninsula, into and through Bristol Bay, and up to Cape Newenham (Fig. 13). Settlement for northern rock sole, in contrast, is concentrated somewhat further north, with less settlement in Bristol Bay and more in areas (settlement areas 5, 18) straddling Cape Newenham, as well as some occurring near Nunivak Island (areas 15, 16). The model results also suggest that northern rock sole settlement may occur further to the west along the Alaskan Peninsula (in area 1) than does Alaska plaice settlement. The model results also suggest different temporal characteristics between the two species in total settlement (Fig. 14), with peaks and troughs in settlement not occurring simultaneously and with some suggestion that northern rock sole exhibits high frequency interannual variability while Alaska plaice exhibits somewhat more “red-shifted” variability. To test the sensitivity of these results to the specific model parameterizations we chose, we conducted a set of numerical experiments by varying aspects of the baseline biological parameterization we had chosen for northern rock sole to address a series of hypotheses. The first hypothesis we considered was that the differences between the baseline results we obtained for the two species was due primarily to differences in the assumed spawning locations we used. To address this hypothesis, we reran the northern rock sole model using the spawning areas for Alaska plaice and compared the results with the two baseline model runs (Fig. 15). The average fraction recruiting is higher in the “hybrid” model than the two baseline models while the CV (coefficient of variation) is lower, but both are closer to the Alaska plaice model results than to the baseline northern rock sole results. The temporal variation in total settlement for the hybrid model is more highly correlated with that for the Alaska plaice model (ρ=0.75) than with the baseline northern rock sole model (ρ=0.40), and the autocorrelation structures are more similar (Fig. 15, bottom right plot). The spatial patterns of settlement for the hybrid model also appear to be more similar to that for the Alaska plaice model (Fig. 15, lower right plot), in that most settlement occurs in the region of Bristol Bay; although settlement in the new model tends to be concentrated somewhat more northerly (areas 3, 4) than in the Alaska plaice model (areas 2, 3). Thus, the difference in spawning areas appears to account for much of the difference between the two baseline models. We also considered two other variations on the baseline northern rock sole model as additional sensitivity tests: the first extended the time period for settlement from 10 days to 30 days and the second increased the depth range regarded as suitable settlement habitat from 0-50 m to 0-60 m. A third variation focused more on the stability of the settlement statistic: we increased the number of individuals released per model run by a factor of six. Extending the time period from 10 days to 30 days had little impact on the resulting recruitment indices: the overall mean settlement fraction increased by 10% while the CV decreased by 12% relative to the baseline case. The correlation between the two time series was
30 extremely high (ρ=0.98), indicating almost no change in the pattern of temporal variation. Extending the maximum depth for suitable settlement habitat from 50 to 60 m substantially increased the mean settlement fraction by 250%, but did not substantially affect the pattern of temporal variability in the recruitment time series (ρ=0.76). Increasing the number of individuals used in the simulation by a factor of six also had little effect on the resulting recruitment index time series. The mean fraction recruiting decreased somewhat (12%), but the temporal pattern remained very similar to the baseline (ρ=0.93). Although these sensitivity tests were not exhaustive, they gave us some confidence that the DisMELS model results would be reasonably robust to errors in the biological parameters adopted for the two flatfish species. However, comparison of the two species-specific recruitment indices with estimates of recruitment from the corresponding stock assessment models was downright unfavorable (Fig.s 16). The DisMELS recruitment index is the fraction of individuals successfully reaching suitable nursery areas within an allotted time frame, and hence represents a measure of survival (mortality) that is density-independent. In contrast, recruitment estimates from the stock assessment models include density-dependent mortality through use of a stock-recruit relationship as a central tendency. Consequently, before comparing the DisMELS recruitment indices with the assessment model estimates, we first calculated the deviations from best-fit Ricker stock-recruit curves for the assessment model results for both stocks. The resulting deviations and the DisMELS results were then natural-log transformed, standardized as z-scores, and then compared (Fig. 16). The comparison for both stocks is poor: the correlation coefficients between the assessment model estimates and the DisMELS indices are negative (!), indicating that (to put a “positive” spin on it) the best thing to do with the DisMELS index is to believe the opposite of what it indicates. This poor agreement between the DisMELS indices and the stock assessment results was really quite surprising, given the apparent utility of the qualitative OSCURS results and the presumption that the NEP4 ROMS model provided a more accurate description of the physical environment than OSCURS, given the former’s incorporation of more realistic physics. Because our sensitivity checks regarding the biological model indicated that model should be reasonably robust to errors in its parameterization, we concluded that at least part of the problem had to lie with the NEP4 ROMS oceanographic model input to DisMELS. In order to address this concern, we first tested whether DisMELS, using the NEP4 ROMS model output, could reproduce the qualitative OSCURS indices. To do this, we ran DisMELS in “OSCURS mode”: we released a single simulated passive surface drifter on April 1 each model year at 165° W, 56° N and tracked it for 90 days (see examples in Fig. 17). We visually assessed whether the DisMELS track was similar to the corresponding OSCUR track using a three-category scale: (1) very similar, (0) inconclusive, and (-1) very dissimilar (Fig. 18). On the whole, most of the DisMELS tracks were qualitatively similar to the OSCURS tracks, but a few were very dissimilar. We also classified the DisMELS tracks according to the OSCURS categories (“on shelf”, “off shelf” or “mid shelf” drift) and compared them with the OSCURS classifications (Fig. 19). The agreement in classification was somewhat poorer than the qualitative agreement between the individual tracks might indicate. This occurred because, even when the DisMELS track “looked” similar to the corresponding OSCURS track, there was a tendency for the DisMELS track to be further offshore than the OSCURS track, such that several DisMELS tracks were classified as “mid shelf” or even “off shelf” when the OSCURS track was “on shelf.” Differences in the results between the two models probably reflect differences in spatial resolution, underlying physics, and the use of different data for atmospheric forcing. In
31 particular, the higher spatial resolution in the ROMS model yields mesoscale eddy structures not present in the OSCURS model. Because the NEP4 ROMS model was not run in an adjoint fashion (i.e., it was not constrained by in situ observations), these eddies are realistic in a spatiotemporally statistical sense, but may also be regarded as computational anomalies. Given the Lagrangian nature of the integration involved, they may have substantial effects on the trajectories of simulated drifters. Another factor arises because the ROMS model uses a “terrain- following” vertical coordinate system: the model has difficulty in producing cross-isobath flow-- the exact type of flow needed to produce trajectories that move from the mid-shelf to the inner shelf. Thus, the ROMS model underestimates the strength of cross-shelf flow. Also, the better resolution (and physics) in the ROMS model allows it to resolve the so-called “inner front”, a known oceanographic feature that approximately follows the 50 m isobath on the eastern Bering Sea shelf, whereas the OSCURS model can not resolve this feature. This front provides a barrier to passive transfer from the mid shelf to the inner shelf. Flatfish larvae and/or early benthic juveniles must pass this barrier to successfully recruit from mid-shelf spawning areas to inshore nursery areas, but the mechanism by which they accomplish this in reality is unknown (and thus not included in the DisMELS biological model). The incorporation of tides in a follow-on ROMS model (NEP5) may improve the model’s ability to simulate cross-shelf flow, as well as provide a potential mechanism allowing simulated larvae or juveniles to move across the inner front using selective tidal stream transport. In contrast to NEP4, though, OSCURS’ poorer resolution and less detailed physical model may actually allow it to provide a “better picture” of flatfish recruitment success because it glosses over some of the difficult details. “Bigger” is not necessarily always “better.”
Literature cited Curchitser, E.N., D.B. Haidvogel, A.J. Hermann, E.L. Dobbins, T.M. Powell and A. Kaplan. 2005. Multi-scale modeling of the North Pacific Ocean: Assessment and analysis of simulated basin-scale variability (1996-2003). Journal of Geophysical Research 110 C11021, doi:10.1029/2005JC002902. Duffy-Anderson, JT, MJ Doyle, KL Mier, PJ Stabeno and TK Wilderbuer. Accepted. Early life ecology of Alaska plaice (Pleuronectes quadrituberculatus) in the eastern Bering Sea: seasonality, distribution, and dispersal. Journal of Sea Research. Haidvogel, D. B., H. Arango, W. P. Budgell, B. D. Cornuelle, E. Curchitser, E. Di Lorenzo, K. Fennel, W. R. Geyer, A. J. Hermann, L. Lanerolle, J. Levin, J. C. McWilliams, A. J. Miller, A. M. Moore,T. M. Powell, A. F. Shchepetkin, C. R. Sherwood, R. P. Signell, John C. Warner, J. Wilkin. 2008. Regional Ocean Forecasting in Terrain-following Coordinates: Model Formulation and Skill Assessment. J. Comput. Phys. 227: 3595- 3624. Hermann, A.J. and C. W. Moore. 2009. Visualization in fisheries oceanography: new approaches for the rapid exploration of coastal ecosystems. In: B.A. Megrey and E. Moksness (eds.), Computers in Fisheries Research, 2nd ed., DOI 10.1007/978-1-4020-8636-6_10, Springer Science and Business Media B.V. Kalnay, E., M. Kanamitsu and 20 other authors. 1996. The NCEP/NCAR 40-Year Reanalysis Project. Bulletin of the American Meteorological Society 77:437-471. NMFS. 2007. Appendix C: Ecosystem Considerations for 2008. In: 2007 North Pacific Groundfish Stock Assessment and Fishery Evaluation Reports for 2008. North Pacific
32 Fishery Management Council, 605 W. 4th Avenue, Suite 306, Anchorage, AK 99501. 236 pp. Shchepetkin, A.F. and J.C. McWilliams. 2003. A method for computing horizontal pressure- gradient force in an oceanic model with a non-aligned vertical coordinate. Journal of Geophysical Research 108 (C3), 3090. Shchepetkin, A.F. and J.C. McWilliams. 2005. The regional oceanic modeling system (ROMS): a split-explicit, free-surface, topography-following-coordinate oceanic model. Ocean Modelling 9: 347-404. Wilderbuer, T.K., A.B. Hollowed, W.J. Ingraham, Jr., P.D. Spencer, M.E. Conners, N.A. Bond, and G.E. Walters. 2002. Flatfish recruitment response to decadal climatic variability and ocean conditions in the eastern Bering Sea. Progress in Oceanography 55:235-247.
Applications: None. At this point, we regard the DisMELS recruitment indices (see above) as problematic. Thus, we have not provided them to the appropriate stock assessment authors as an alternative to the OSCURS-based qualitative recruitment indices. However, interested parties are encouraged to explore the use of DisMELS to characterize dispersal in their own systems. We emphasize the need, though, for appropriate validation studies before model results are used for management purposes. On the other hand, DisMELS may appropriately be used for conducting process-oriented numerical studies of dispersal, e.g. comparing dispersal envelopes under identical environmental conditions but different vertical behavior by larvae.
Products: Software 1. DisMELS: The Dispersal Model for Early Life Stages. Executable, code, and user manual available. a. Description: DisMELS is an individual-based biophysical model that incorporates ontogenetic changes in early life history parameters and simulates egg and larval dispersal under 3-dimensional oceanographic currents to predict relative fluctuations in annual recruitment due to density-independent, advective processes in the eastern Bering Sea (EBS). DisMELS uses stored output of 3-dimensional oceanographic currents, temperature and salinity fields at daily time steps from a Regional Ocean Modeling System (ROMS) oceanographic model to simulate advective transport of pelagic eggs and larvae. Simulated individuals (eggs or larvae) are released along the bottom throughout a pre-defined “spawning area” during the spawning season. Individuals are tracked through time and space using Lagrangian particle tracking “with behavior”; i.e., individuals are allowed to undergo vertical migration: they actively swim up or down in the water column until they enter a “preferred” depth range (different ranges can be defined for daytime and nighttime, allowing diel migration to be incorporated). Individual growth and mortality can also be tracked. Ontogenetic changes in vertical migration behavior, as well as growth or mortality rates, can be incorporated in the model by defining a series of sub- stages that individuals progress through: different model parameters for different sub-stages yield ontogenetic changes in behavior, growth, and/or mortality. Individuals reaching the final larval sub-stage metamorphose into a “settlement-competent” stage that settles to the benthos upon reaching pre-defined juvenile nursery habitat. Individuals in the settlement stage that do not reach suitable nursery habitat before the stage terminates (a pre-set amount of time) are
33 regarded as dead. A relative index of annual recruitment is generated by simulating thousands of individuals over one model year and calculating the fraction that are transported from the spawning areas to the nursery areas and successfully settle for that year. A time series for this index is generated by running the model for multiple years. b. Use: Until suitable ROMS model output becomes available, DisMELS will not be used to generate annual recruitment indices for flatfish in the EBS. However, interested parties are encouraged to explore the use of DisMELS to characterize dispersal in their own systems. We emphasize the need, though, for appropriate validation studies before model results are used for management purposes. On the other hand, DisMELS may appropriately be used for conducting process-oriented numerical studies of dispersal, e.g. comparing dispersal envelopes under identical environmental conditions but different vertical behavior by larvae. c. Maintenance and Updates: The DisMELS software is maintained and updated by W. Stockhausen ([email protected]).
2. SimpleVectorCustomRenderer: A java-based "custom renderer" extension to ESRI's ArcGIS (version 9.3.1 and higher). a. Description: This extension allows the user to plot 2-dimensional vectors in ArcGIS while accounting for differences between the spatial references of the data layer and the map layer. b. Maintenance and Updates: The SimpleVectorCustomRenderer software is maintained and updated by W. Stockhausen ([email protected]).
Publications/Presentations/Webpages: Publications Haidvogel, D. B., H. Arango, W. P. Budgell, B. D. Cornuelle, E. Curchitser, E. Di Lorenzo, K. Fennel, W. R. Geyer, A. J. Hermann, L. Lanerolle, J. Levin, J. C. McWilliams, A. J. Miller, A. M. Moore,T. M. Powell, A. F. Shchepetkin, C. R. Sherwood, R. P. Signell, John C. Warner, J. Wilkin. 2008. Regional Ocean Forecasting in Terrain-following Coordinates: Model Formulation and Skill Assessment. J. Comput. Phys. 227: 3595-3624.
Hermann, A.J. and C. W. Moore. 2009 Visualization in fisheries oceanography: new approaches for the rapid exploration of coastal ecosystems. In: B.A. Megrey and E. Moksness (eds.), Computers in Fisheries Research, 2nd ed., DOI 10.1007/978-1-4020-8636-6_10, Springer Science and Business Media B.V.
Duffy-Anderson, JT, MJ Doyle, KL Mier, PJ Stabeno and TK Wilderbuer. Accepted. Early life ecology of Alaska plaice (Pleuronectes quadrituberculatus) in the eastern Bering Sea: seasonality, distribution, and dispersal. Journal of Sea Research.
Stockhausen, WT, JT Duffy-Anderson, TK Wilderbuer and AJ Hermann. In prep. DisMELS: an Extensible Individual-based Framework for Modeling Dispersion of Early Marine Life Stages For submission to Ecological Modeling.
Stockhausen, WT, JT Duffy-Anderson, TK Wilderbuer and AJ Hermann. In prep. Predicting recruitment for flatfish in the EBS: Can we improve on OSCURS? For submission to Fishery Bulletin.
34
Presentations 2009 Stockhausen, W., J. Duffy-Anderson, T. Wilderbuer and A. Hermann. 2009. Predicting recruitment for flatfish in the EBS: Can we improve on OSCURS? Pacific Marine Environmental Laboratory Seminar Series; Seattle, WA; March 2009.
Stockhausen, W., J. Duffy-Anderson, T. Wilderbuer and A. Hermann. 2009. Predicting recruitment for flatfish in the EBS: Can we improve on OSCURS? PCCRC Spatial Modeling Workshop; Seattle, WA; June 2009.
Cooper, D., J. Duffy-Anderson, W. Stockhausen, P. Stabeno and C. Jump. 2009. Northern rock sole (Lepidopsetta polyxystra) connectivity between spawning and settling areas along the Alaska Peninsula in relation to currents and hydrography. 33rd Annual Larval Fish Conference; Portland, OR; July 2009.
2008 Duffy-Anderson, J.T., Mier, K.L., Stabeno, P., Doyle, M., Stockhausen, W., and Hermann, A. Distribution and dispersal of Alaska plaice on the eastern Bering Sea shelf. Contributed presentation. 8th International Flatfish Symposium; Sesimbra, Portugal; November 2008.
Hermann, A., K. Aydin, N. Bond, W. Cheng, E. Curchitser, G. Gibson, Katherine S. Hedstrom, and M. Wang. Efficient Strategies for Climate Downscaling in the Bering Sea. Contributed presentation. Alaska Marine Sciences Symposium; Anchorage, AK; January 2008.
Hinckley, S., C. Parada, B. Ernst, L. Orensanz, D. Armstrong and A. Hermann. Temperature Effects on Larval Snow Crab Transport and Source-Sink Relationships in the Bering Sea: How Will Larval Settlement Success be affected by Climate Change? Contributed presentation. Alaska Marine Sciences Symposium; Anchorage, AK; January 2008.
Stockhausen, W., J. Duffy-Anderson, T. Wilderbuer and A. Hermann. Can different early life history strategies explain different recruitment patterns for two flatfish species in the eastern Bering Sea? Contributed presentation. 2008 Larval Biology Symposium; Lisbon, Potugal; July 2008.
35 Tables
Table 1. DisMELS parameters for pelagic life stages for Alaska plaice and northern rock sole. Alaska plaice Northern rock sole characteristic egg larva settler early larva late larva settler stage duration (d) 21 40 10 18 40 10 diurnal migration? no no no no yes no daytime depth 10-100 10-20 10-50 20-30 10-30 10-50 range (m) nighttime depth 10-100 10-20 10-50 20-30 0-10 10-50 range (m) vertical swimming 011 1 1 1 speed (cm/s) settlement depth - - 0-50 - - 0-50 (m)
36 Figures
Figure 1. OSCURS simulated annual surface drifter trajectories by decade (left: 1980s, right: 1990's; individual years are not identified). Each trajectory represents the track of a single simulated drifter released at 165° W, 56° N on April 1 of each year and tracked for 90 days. The track for each year is visually classified as "on shelf", "off shelf", or "mid shelf" to provide a qualitative index of potential recruitment success for winter-spawning flatfish (Wilderbeuer et al., 2002; NMFS 2008).
6 1.0
Northern rock sole "on shelf" wind drift "on-shelf" wind drift "off shelf" wind drift "mid-shelf" wind drift 5 "off-shelf" wind drift "mid shelf" wind drift 0.8
4 0.6
3 0.4 Probability (billions)
Recruitment Recruitment 2 0.2 1
0.0 0 0123456 1980 1985 1990 1995 2000 Recruitment (billions) Figure 2. Left: estimates of annual recruitment for northern rock sole from its stock assessment model. The bars are color-coded according to the on shelf/off shelf/mid shelf classification of the OSCURS simulated drifter track for the same year. Right: Estimated means (dotted lines) and probability distributions (solid lines) for northern rock sole recruitment under "off shelf" (blue), "mid shelf" (yellow), and "on shelf" (red) conditions (as classified by the OSCURS simulations, assuming a lognormal distribution for recruitment under each condition). Higher recruitment is expected in years classified as having "on shelf" conditions.
3.5 Alaska plaice "on shelf" wind drift 1.2 "on-shelf" wind drift "mid-shelf" wind drift 3.0 "off shelf" wind drift "mid shelf" wind drift "off-shelf" wind drift 1.0 2.5 0.8 2.0 0.6 1.5 Probability (millions)
Recruitment Recruitment 0.4 1.0 0.2 0.5 0.0 0.0 0123456 1980 1985 1990 1995 2000 Recruitment (millions) Figure 3. As in Fig. 2, but for Alaska plaice.
37 egg egg larva larva settler benthic stage 1 stage N stage 1 stage M stage juvenile adult 1
egg egg larva larva settler benthic stage 1 stage N stage 1 stage M stage juvenile
egg egg larva larva settler benthic stage 1 stage N stage 1 stage M stage juvenile adult L
egg egg larva larva settler benthic stage 1 stage N stage 1 stage M stage juvenile
Pelagic Life Stages Figure 4. Cartoons of: 1) the linkage between DisMELS' environmental and biological sub-models (above left) and 2) the life stage components in DisMELS' biological sub-model (above right).
Figure 5. ROMS model grid coverages for the north Pacific Ocean, including the coarse-resolution NPac model grid (~0.4° horizontal resolution; in red) used to provide initial and boundary conditions to the medium-resolution NEP4 model grid (~10 km horizontal resolution; in green). The latter has 42 vertical layers. Daily model output from the NEP4 model from 1979-2004 is used in this project.
38
Northern rock sole Alaska plaice
All larvae Eggs
0-10 m 10-20 m 20-30 m 30-40 m 40-50 m 50-60 m below 60 m 4MF03 5MF05
0 5 10 15 20 25 0 5 10 15 20 25
Catch 10 m-2
Post-flexion larvae Preflexion larvae
Figure 6. Results based on depth-discrete sampling (MOCNESS tows) for northern rock sole (left column) and Alaska plaice (right column). Northern rock sole eggs are attached demersally; pre-flexion larvae are concentrated between 10-30m depth in the water column (upper left graph); post-flexion larvae appear to undergo diel vertical migration (lower left graph; white bars: daytime distribution, dark bars: nighttime distribution). Alaska plaice eggs are pelagic and distributed through the upper water column; preflexion larvae occur somewhat below the surface, but higher in the water column than northern rock sole larvae.
39 Alaska
Bering Sea
Gulf of Alaska
Figure 7. Presumed spawning areas for Alaska plaice (left) and northern rock sole (right). The large contiguous areas were divided into smaller areas to look at connectivity between spawning areas and nursery areas on somewhat finer spatial scales. The spawning areas in the Bering Sea for northern rock sole (map at right, areas 1-5) were based on the mean spatial pattern of fishing activity during late winter (Mar-Apr) by the commercial roe fishery for 1986- 2004.
Figure 8. Presumed settlement areas for Alaska plaice and northern rock sole. Both species use relatively shallow, nearshore areas as nurseries. We assigned all areas where the total depth was less than 50 m as potential settlement habitat.
40
Figure 9. Tracks of 10,000 simulated northern rock sole larvae over a 120 day period for model year 1982. Tracks are color-coded by sub-stage: preflexion larvae--gold; post-flexion larvae--red; competent-to-settle juveniles--cyan. Final locations are indicated by the dark blue dots.
Figure 10. Map of the final locations of the individuals shown in Fig. 1, color-coded by the area in which each was “spawned”. Spawning areas are denoted colored polygons numbered 1-6. Thus, e.g., individuals that were spawned in area 6 (blue polygon) are also colored blue.
41
Figure 11. Map of the starting locations of the individuals shown in Fig. 1, color-coded by the area in which each found suitable nursery habitat (depth shallower than 50m, polygons 1-19) and “settled”. Thus, e.g., individuals that settled in area 18 (pink polygon) are also colored pink. Black dots represent the starting location of individuals that never reached suitable nursery habitat.
Figure 12. Plot of the connectivity matrix between the spawning areas (“source” polygons) and nursery areas (“sink” polygons) from Fig. 1. Colors represent the fraction of individuals originating from a “source” polygon that settled in a “sink” polygon.
42
Figure 13. Plots from the baseline northern rock sole (left plot) and Alaska plaice (right plot) models for the fraction of individuals from all spawning locations by nursery area (“sink” polygon) for all model years. The plots indicate substantial interannual variability in both settlement area and strength for a give species, as well as different patterns between species.
0.4
0.3
0.2
recruitment index recruitment 0.1
northern rock sole Alaska plaice 0.0 1980 1985 1990 1995 2000 2005
Figure 14. Baseline recruitment indices by model year for northern rock sole (blue line) and Alaska plaice (green line).
43 0.5 0.6 AP baseline NRS baseline NRS baseline AP baseline 0.4 NRS with AP spawning areas 0.5 NRS using AP spawning areas
0.4 0.3
0.3 Value 0.2
0.2
0.1 recruitment index 0.1
0.0 0.0 mean cv 1980 1985 1990 1995 2000 2005 Statistic
1.2
AP baseline 1.0 NRS baseline NRS with AP spawning area 0.8
0.6
0.4
Autocorrelation 0.2
0.0
-0.2 01234 Lag (years) Figure 15. Model results using northern rock sole (NRS) baseline parameters but Alaska plaice (AP) spawning areas. Top left: Comparison of means and CVs (coefficient of variation) for the various models. Top right: Comparison of the time series of recruitment indices from this “alternative” model (cyan) with those from the baseline models (dark blue: northern rock sole; dark green: Alaska plaice). Bottom left: Comparison of the time series autocorrelation structure of the recruitment indices for the various models. Bottom right: Spatiotemporal variation in the settlement areas for the hybrid model (compare with plots in Fig. 13).
44 Northern rock sole Alaska plaice 3 3
2 2
1 1
0 0 z-score z-score
-1 -1
-2 -2 Assessment model Assessment model DisMELS ρ = -0.27 DisMELS ρ = -0.20 -3 -3 1980 1985 1990 1995 2000 2005 1980 1985 1990 1995 2000 2005 Year Year Figure 16. Comparison between recruitment indices from the associated stock assessment model and the baseline DisMELS model for northern rock sole (left graph) and Alaska plaice (right graph). Indices from the stock assessment models are standardized deviations from best-fit Ricker stock-recruit curves for each stock. DisMELS recruitment indices have been natural log-transformed and standardized, as well. Cross-correlation coefficients (ρ) are also shown.
Figure 17. Comparison of tracks of passive surface drifters from DisMELS run in “OSCURS mode” (yellow line) with the corresponding OSCURS tracks (orange line). The left column illustrates years with “very dissimilar” tracks whereas the right column illustrates years with “very similar” tracks.
45
1.5
similar 1
0.5
0
Agreement 79 81 83 85 87 89 91 93 95 97 99 01 03 0 0 -0.519 19 19 19 19 19 19 19 19 19 19 2 2 Mean=0.73 dissimilar -1 -1.5 Figure 18. Visual similarity scoring by model year for DisMELS tracks run in “OSCURS mode” compared to the corresponding OSCURS tracks. The mean similarity was based on assigning 1, 0 , or -1 to the classifications “similar”, “inconclusive”, and “dissimilar”, respectively.
1.5 OSCURS DisMELS on-shelf 1
0.5
0
Recruitment 979 981 983 985 987 993 -0.51 1 1 1 1 1989 1991 1 1995 1997 1999 2001 2003 off-shelf -1 ρ=0.67 -1.5
Figure 19. Drift classification by model year for OSCURS tracks (blue bars) and DisMELS tracks run in “OSCURS mode” (purple bars). The correlation coefficient () was based on assigning 1, 0 , or -1 to the classifications “on shelf”, “mid shelf”, and “off shelf”, respectively.
46 08-01 Project Title: Development of Biological and Physical Indices for Stock Evaluation in the Dry Tortugas Pink Shrimp Fishery
Principal Investigators: Joan A. Browder - NOAA/NMFS Southeast Fisheries Science Center Maria M. Criales, Claire.B. Paris, and L. Cherubin - University of Miami, RSMAS, Cooperative Institute for Marine and Atmospheric Sciences
A. Introduction With support from the FATE program we are developing a biophysical Lagrangian model for the pink shrimp (Farfantepenaeus duorarum) in south Florida to simulate the transport of larval pink shrimp from offshore spawning grounds on the outer southwest Florida shelf between the Dry Tortugas and Marquesas Islands to nursery grounds in Florida Bay and young adult shrimp from Florida Bay to their spawning grounds. Our model will employ existing data and new knowledge about the behavior of south Florida pink shrimp in relation to their cross-shelf migrations (Criales et al. 2005, 2006, 2007) to develop indices of recruitment that relate to environmental variation and, potentially, to climate changes.
We will examine the interconnections between behavior and local, mesoscale, and regional scale processes in the migration of pink shrimp from the early larval stage to settlement-stage postlarvae to answer questions that will help refine stock assessments by improving predictions of recruitment to the fishery. Some questions that will be addressed are as follows: What are the survival implications of alternative spawning locations? What is the behavior and transport during the first development stages? Is flood tidal transport the dominant onshore transport mechanism of this species in south Florida? Are there other relevant migration paths and transport mechanisms than the ones we have identified? What is the origin of anomalous water column conditions found at that Marquesas that appeared to promote larval recruitment, how wide-spread is their presence, and is the occurrence of these conditions predictable?
Model results will help refine assessments of the pink shrimp stock in south Florida to incorporate the influence of environmental variation. Stock assessments that consider environmental factors will better inform fishery managers on the causes of recruitment variation and lead to better fishery management. Knowledge about the biological effects of coastal processes driven by freshwater inflows (e.g., baroclinic circulation and salinity gradients) will help inform water managers about the biological impacts of currently planned large-scale water management changes that are expected to substantially affect the quantity, timing, and distribution of flows to Florida Bay and the lower southwest Florida coast.
B. Background
B.1. The study area The study area is the Florida shelf at the southern tip of the Florida peninsula and Florida Bay. The outer perimeter of this area is bounded by the Florida shelf break and the Loop-Current- Florida Straits-Gulf Stream. The inner perimeter is bordered by the Florida mainland. The Florida Keys and reef tract occupy narrow bands that fit inside and are contoured with the shelf break on the southeastern side of the Florida shelf. The bathymetry of the southwest Florida
47 shelf is characterized by low relief algal plain with sandy shoals that are culminated in the low, flat islands of the Dry Tortugas at the southwestern edge and the Marquesas Islands centrally located between the Dry Tortugas and Key West. Florida Bay is a labyrinth of expansive banks dotted with islands and encircling shallow basins. Water flow is through broad to narrow channels that bisect the banks as well as across the banks (Lee et al. 2006) Physical processes that dominate the shelf and oceanic areas on both sides of the Florida Keys, as well as Florida Bay, have been described by Weisberg et al. (1996), Lee et al. (2002), He and Weisberg (2002), among others. The circulation on the SWF shelf results from several factors. Because the shelf is so broad, its inner shelf is most affected by local wind forcing, while its middle and outer shelf regions receive the direct influence of the Loop Current that enters the Gulf of Mexico (Weisberg et al. 2000). A model of the southwestern Florida shelf by Weisberg (2000)extends to cover this area. Smith (1997) has described the spatial and temporal dynamics of Florida Bay tides. Tides are primarily semi-diurnal in Florida Bay and on the southwest shelf, whereas they are diurnal on the Atlantic coast side of the Keys. Tides in the eastern part of the bay are affected by the mixing of Gulf of Mexico and Atlantic Ocean waters and the interaction of the two tides (Smith 2000). Kelble et al. (2007) described salinity patterns in Florida Bay.
B.2. The Simulation Period The simulation period covers 11 years, from 1995 through 2005. We selected this period because of the greater availability of boundary data, calibration data, and data on pink shrimp on both the fishing ground and the nursery grounds. This 11-yr period embodies the range and temporal variability in fishery CPUE, a rough index of stock abundance.
B.3. The Tortugas Pink Shrimp Stock and Fishery Pink shrimp habitat alternates between offshore spawning grounds such as the Tortugas grounds and nursery grounds in Florida Bay and other estuaries. Development from the egg to ready-to- settle postlarval stage takes about 30 days, and young pink shrimp travel from the spawning grounds to the nursery grounds within that period. After about 2 to 5 months on their nursery grounds young adult pink shrimp return offshore to start their life as spawners (Costello and Allen 1966). They are subjected to fishing pressure primarily on the spawning grounds, although recreational fishing and small commercial bait and food shrimp fishing take place along the Florida Keys and in Biscayne Bay. A pink shrimp fishery operates along the entire west Florida coast and merges with fisheries for other penaeid species across the U.S. coast to Texas and along the Mexican coast.
Historically, the Tortugas grounds have been the most important fishing grounds for pink shrimp in Florida in terms of both landings and value. Hart (2008) summarized the history of the Tortugas fishery. The Tortugas grounds have been active since the late 1940s, and fishery statistics (catch and effort) have been collected since 1960. Fishing effort, measured in thousands of vessel-24-hr-days, was almost constant from 1960 through 1980, but declined in the 1980s, rising again in the mid 1990s and falling in recent years. The pattern of annual effort was roughly reflected in the total catch-per-unit-effort (CPUE), except from 2004 through 2006, when effort fell to the lowest level in the fishery record, and CPUE was about as high as ever in its recorded history (639, 736, and 613 pounds of all-sized shrimp per vessel-day, respectively). CPUE had previously been over 700 pounds only in 1960, 1965, and 1981 (725, 730, and 788 pounds per vessel-day, respectively). Total CPUE averaged 598 pounds per vessel day from
48 1960 through 1985. The lowest total shrimp CPUE in the history of the fishery was 349 pounds per vessel-day in 1999.
The Tortugas pink shrimp fishery has not received exclusive stock assessment attention in recent years. The stock assessment of Nichols (1984) covered the entire pink shrimp fishery from Brownsville, TX, to Key West as one unit. The NMFS stock assessment of Nance (2008) was applied to all Gulf of Mexico penaeids as one unit. Analyses specific to the Tortugas fishery were conducted by Klima and Patella (1985) and Klima et al. (1986) and addressed the impact of the Tortugas Sanctuary, a nearshore spawning area recently closed at the time of their study to protect young spawners. Kutkuhn (1966) conducted a stock assessment specific to the Tortugas fishery.
Ehrhardt and Legault (1999) conducted a cohort analysis of the Tortugas pink shrimp population from 1965 through 1995. They noted changes in the seasonal pattern of recruitment from one period of years to another. While recruitment occurred all year, the seasonal pattern consisted of one or more recruitment peaks that differed in magnitude and monthly position from year to year. The March-April recruitment peak showed an increasing trend from 1965 through 1981, followed by a decreasing trend in the 1980s and early 1990s. The October-November recruitment peak decreased consistently over the same years. Recruitment during the period 1982-1993 was at the lowest level historically. Examining the relationship between recruitment and spawning stock, Ehrhardt and Legault (1999) foud the best relationship with a 5-month lag between parents and recruits, but the models were weak, and they concluded that recruitment was influenced more by environmental factors than the size of the spawning stock. Except for 1995, the Ehrhardt and Legault (1999) analysis did not cover the period of the fishery that will be examined in the present study. C. Biophysical Modeling C.1. Dispersal Model Description Numerical modeling has become a popular approach in population connectivity and recruitment variability studies (Werner et al., 2007). In this approach, the early life of the target species and their interaction with the environment is typically simulated by coupling an Individual-Based Model (IBM, Grimm et al., 2006) with discretized (x,y,z,t) information from physical ocean circulation models (Vikebo et al., 2007). The larval connectivity model used is described in Paris et al. (2007) and is essentially composed of four modules that are coupled through a parallel distributed computing framework: • Ocean Circulation Module (ROMS) • Benthic Seascape Module (spawning and nursery grounds) • Biological Module (life history traits and larval behavior) • Individual-Based Module (IBM - track individual larva or active particles)
Substantial progress has been made in adapting the ROMS to the southwest Florida shelf, Florida Bay, and our use. Other modules are currently under development. Here we focus on describing progress in the ROMS and describing the IBM coupling with initial output of ROMS to estimate Lagrangian trajectories (passive) from the spawning sites. C.2. Ocean Circulation Module
49 The Regional Ocean Modeling System (ROMS, Shchepetkin and McWilliams 2004) is used to study the transport of pink shrimp larvae. This version (ROMS-AGRIF) performs local refinement via nested grids (Adaptive Grid Refinement in Fortran; Blayo and Debreu 1999) and has the ability to manage an arbitrary number of embedding levels as well as to do solution adaptive grid-refinement. Namely, several numerical simulations using different grid resolutions and embedded in each other are run simultaneously. A 2-grid level model is being developed for the FL Keys region. First, a parent simulation at 2.8 km horizontal resolution was set-up (Fig. 1) and then a 0.7 km horizontal resolution simulation is being nested in the parent model. For now the grid has 366 points on the x-axis, 366 on the y- axis and 25 sigma (terrain following) layers. This could change because we are still searching for the right positioning of the model boundaries as explained in the following. Currently, the model grid extends to the south to Cuba, and Belize on the Yucatan Peninsula and to the Great Bahamas Bank to the east. The rationale for using ROMS and background on other models important to this project are given in Appendix 1.
Figure 1: Current bathymetry and grid of the parent model
C.2.1. Simulation configuration
C.2.1.1. Bathymetry The Florida Keys consists of the juxtaposition of very shallow banks (4-5 m deep), Bay of Florida to the west and north, connected to the Florida Current through cuts between small
50 islands. Only the General Bathymetric Chart of the Oceans (GEBCO – http://www.gebco.net/), a global 30 arc-second grid is accurate enough over the entire grid domain rather than only the FL Keys. The 90m horizontal resolution bathymetry is available at the NGDC database (http://www.ngdc.noaa.gov/mgg/bathymetry/relief.html)
C.2.1.2. Atmospheric forcing The second step toward building the regional simulation for the FL Keys region consisted of building the atmospheric forcing. The most adequate, high resolution atmospheric fields that we found for the region, which consists of a 30-year reanalysis (2009-present), and which is focus on North America is provided by the National Center for Environmental Prediction (NCEP, http://www.cdc.noaa.gov/data/gridded/data.narr.html) and is called the North American Regional Reanalysis. It resolution is about 0.3° and the forcing frequency is 8 times daily. We are still in the process of evaluating the wind by comparison with locally observed winds at buoys and towers in the FL Keys. Although the spatial resolution is not better than satellite QuikSCAT winds that we used for other simulations, the presence of land is accounted for in the reanalysis and not in satellite winds.
The response of the model's ocean to the atmospheric forcing is calculated by a method, known as the bulk formulation, which consists of having ROMS calculating its own heat fluxes using its sea surface temperature (SST) and salinity (SSS). We will use comparison of model with observed SSTs to assess the accuracy of the formulation that could be tuned to improve the model surface fields if necessary.
Required atmospheric data for the bulk formulation in ROMS are: - Air temperature at 2m - Relative air humidity - Precipitation rate - Outgoing longwave radiation - Shortwave radiation - Zonal wind speed - Meridional wind speed
Ideally one would assume that there is an interaction between the ocean surface and the lower atmosphere, such as the feedback of oceanic changes on atmospheric forcing. Unfortunately, this is not possible since the atmospheric model is not coupled to our ocean model. However, we can assume that the ocean model that was used along with the atmospheric model to produce the forcing would simulate similar ocean conditions as in our ocean model. Fields on the surface forcing are shown below for different days in January 1994.
51
52
Figure 2: From left to right and top to bottom: surface air temperature, precipitation, relative humidity, wind speed, out longwave radiation, shortwave radiation, meridional wind and zonal wind.
C.2.1.3. Ocean boundaries forcing Numerical model variables of the ocean state (temperature, salinity, currents) at the open ocean boundaries will be relaxed at different time periods depending on the model used at the boundaries. For ECCO, the time interval is 10 day, ecco2, 30 days and TOPAZ 7 days.
ECCO simulation assimilates altimetric sea level measurements from TOPEX/POSEIDON. Altimetric data provide the most extensive spatial and temporal sampling of the ocean. Other data being used in our studies include XBTs, current meter measurements, drifters, floats, CTDs, satellite sea surface temperatures, tide gauges, bottom pressure measurements, and air-sea flux estimates from the NCEP reanalysis project. Ecco2 assimilates the same data. However the higher resolution allows for meso-scale eddies to be simulated, yielding a realistic Loop Current. TOPAZ develops advanced data assimilation systems for a coupled ocean circulation and marine ecosystem model for the North Atlantic and the Arctic Ocean with enhanced resolution in the coastal zones, assimilating satellite and in-situ data available for real time operational use. The observations used are satellite observed Sea Level Anomaly (SLA), Sea Surface Temperature (SST), Sea-ice concentrations from AMSR-E, sea-ice drift products from CERSAT and Coriolis in-situ Temperature and Salinity profiles.
Tides in the ROMS model were set at the boundary by the TPXO7 global tide model1 which provides sea surface height and barotropic currents of the tide at any location on the globe using the TOPEX/POSEIDON altimeter. Once initialized tides cycles are calculated by ROMS.
1 http://www.esr.org/polar_tide_models/Model_TPXO71.html
53 C.2.2. ROMS Model Results We are currently in a test phase because the grid choice has not been producing stable simulations. Indeed, the Yucatan Channel is a very dynamical region with very strong currents surrounded by very steep topography, where the Loop Current enters the Gulf of Mexico. As can be seen in Figure 1, the southern boundary is the one that faces the strong currents from the outer model (ECCO, ecco2 or TOPAZ) and therefore its location is critical for the model stability. We initially started with a 2 km resolution model that has 430x460x25 grid points. This grid extended only to the north of Cuba. At this resolution and in that region, the model would resolve the spinoff eddies off the coast of Cuba and along the FL Keys. However we never managed to finish the month of January because of stability issues due to very strong current parallel to the boundary in the south west corner of the domain. Multiple choices of model parameters have been explored in this configuration without success. We are currently exploring new boundaries that were pushed further south to Belize latitude as shown in Figure 1. In that region, the current is less energetic, and the model should better handle its stability in that region. The new simulations are being prepared and a stable configuration well be achieved soon. In order to keep the computer time reasonable for this new grid the resolution was degraded to 2.8km. However the zoom will remain at 0.7 km resolution.
C.3. Seascape Module The preferred spawning habitat of the pink shrimp, Farfantepenaeus duorarum, is low profile, hard, coarse, calcareous bottom sediment, typical to the sandy areas from the Dry Tortugas east to the Marquesas. We are using the Coral Reef Millennium data to identify these areas and build the source polygons for the matrix model. The nursery grounds (or settling polygons) of postlarvae are within Florida Bay.
C.4. Biological Module This module is being built using lab and field observations from PI Criales (Univ. Miami). Further laboratory and field observation in related studies will determine the preference of early settlement stages to specific seagrass and environmental traits in Florida Bay, the environmental cues associated with the Selective tidal stream transport (STST), and patterns of postlarval supply and settlement in relation to the tide and moon phases.
C.4.1 Spawning The pink shrimp, F. duorarum, spawn year-round but primarily between June and August with a large spawning peak in late spring and summer (Cummings 1961, Iversen et al. 1960). Shrimp females spawn northeast of the Dry Tortugas. Spawning activity is correlated with water temperature (Kennedy and Barber 1981). Spawning occurs at bottom temperatures of 22°-29°C between 3-16 m depth. The eggs are free and not positively buoyant as larvae are found predominantly within the 30-40 m depth strata in the Florida Current (Jones et al. 1968). There is evidence that females move into shallow waters to spawn during summer months, but the proximity to the coast of the summer migration path has not been well established (Jones et al. 1970, Munro et al. 1968). The possible onshore migration during summer may be an avoidance response of spawners to cool bottom water temperatures associated with the vertical
54 stratification. This hypothesis is supported by the high concentration of early protozoea found in vertically stratified water near Marquesas (Criales et al. 2007). Alternatively, young adults migrating out of Florida Bay may pause to spawn in the shallow waters of the near shelf.
C.4.2. Competency period Larval development is rapid, passing through five nauplii, three protozoeae, and three myses in about 15 days, and three to six additional planktonic postlarval stages in other 15 days (Ewald 1965). The pelagic larval duration of approximately 30 days is variable, depending on pelagic conditions (Criales et al. 2003).
C.4.3 Swimming The swimming behavioral model will use selective tidal stream transport (STST), a vertical migratory behavior that has been observed in pink shrimp postlarvae and myses (see Criales et al., in CIMAS report 2007). With this behavior postlarvae ascend into the water column during the flood tide to take advantage of the current flow in the direction to the nursery grounds of Florida Bay, and rest near the bottom during the ebb tide. This behavior could be the result of an endogenous activity rhythm, or could result from a combination of responses to environmental cues such as light, salinity, and turbulence. As part of a related study, postlarvae presently are being evaluated for the presence of an endogenous rhythm in activity and responses to salinity and turbulence under controlled laboratory conditions.
C.4.4. Simulated traits and behavior of larval F. duorarum Based on the above observations we selected the following characterization of larval behavior: 1. Production: larval density from plankton sampling will serve as a proxy for production throughout the year. Production will simulate peak spawning seasons. 2. Spawning: Particle will be released during 8 days centered around the full moon from March to September. 3. Swimming: Present knowledge of diel vertical migration and selective tidal stream transport will be refined by ongoing laboratory experiments in a related study. 4. Competency: on average, larvae are competent around 30 days. We will carry out sensitivity analyses on the effect of decreasing or extending the competency period based on water temperature from the ROMS. 5. Survival: a daily mortality rate will be assigned to the number of particles released on the basis of differential survival with increased planktonic duration.
C.5. IBM Module The larvae, represented as Individual particles, (larvae) are moved by the 3-D current advection from ROMS, local turbulence, and a specific behavior, including survivorship. The base of the IBM tracking is a Lagrangian Stochastic Model (LSM) with a 4th order Runge-Kutta stepping scheme to resolve mesoscale currents and a random flight with a 2nd order Markov process to parameterize sub-grid-scale dispersive processes
55 (Dutkiewicz et al. 1993, Griffa 1996, Paris et al. 2002). Sub-grid-scale turbulence is needed not only to help support dispersion but also to maintain fidelity with Eulerian predictions and is strongly tied to the eddy kinetic energy of the region (Okubo 1971, North et al. 2008). Because the circulation in the Florida Keys is complex and characterized by both mesoscale and small-scale eddies translating along the Loop Current and the Florida Current, the Lagrangian statistics used in the stochastic particle- tracking scheme will be spatially-explicit and estimated from the Eulerian fields of the regional circulation model (Paris et al. 2007). The Lagragian parameters are: (m2 s-1 ), the eddy diffusivity, or variance of the velocity field; TL (s), the Lagrangian time scale over which particle velocity is correlated; =
D. Bio-physical Coupled Simulations with Passive Transport Particles were moved passively in the flow to test the fidelity of the trajectories with the Eulerian predictions given the parameterization of the LSM as described above. A series of monthly releases of particles were tracked for 30 d from the Tortugas spawning ground area. These preliminary releases used the HYCOM-FLK, which has no tidal current; therefore the STST could not be simulated. Therefore these simulations provide a glimpse of what transport of larval pink shrimp from the Tortugas grounds might be in the absence of STST behavior.
The passive transport simulations can best be viewed in relation to the two main pathways hypothesized for rapidly developing planktonic stages of pink shrimp larvae migrating from Dry Tortugas to Florida Bay: (i) larvae drift south-southeast downstream with the Florida Current front and enter Florida Bay through the tidal inlets of the middle Florida Keys (Munro et al. 1968, Criales et al. 2003); (ii) larvae move northeast across the SWF shelf and enter Florida Bay at its northwestern boundary (Jones et al. 1970, Criales et al. 2006).
During the first days of development larvae may drift with the alongshore current. There is evidence that protozoae migrate vertically with a diel vertical migration (DVM), but a cross-shelf transport with this DVM behavior has not yet been demonstrated (Jones et al. 1970; Criales et al., 2007). Later in development, late myses and postlarvae have been observed performing rhythmic vertical migrations synchronized with the semidiurnal tide near Marquesas at about 20 m of depth (Criales et al., 2007). Postlarvae experienced a considerable onshore flux with this behavior. This tidal behavior was observed under conditions of strong density gradients and shear due to the presence of non-linear tides.
56
Figure 3: Sample Lagrangian-particle trajectories from multiple sites: One hundred particles were released monthly. The lines represent 30-day trajectories; the yellow stars indicate the release sties and the red dots the location 30 days after release. Passive trajectories show the presence of a mesoscale eddy in the Dry Tortugas, and smaller-scale looping trajectories are also evident on its eastern edge.
57
From the preliminary passive scenarios, we see those two major routes, in particular during early spring of 2004, and spring and summer of 2007 (Fig. 3). However, passive advection does not allow the arrival of the particles (red dots in Fig. 3) at the nursery grounds of Florida Bay, especially from the SWF pathway (Fig. 4) that is presumably the most important pathway (Criales et al. 2006). Therefore, we conclude that both tidal flow and larval behavior must have a large influence in the transport over the shelf and are critical to the successful transport of the pink shrimp larvae to their nursery grounds.
Figure 4. Lagrangian probability density function (PDF) from the Dry Tortugas (yellow stars) for the advection time of 30 days. The Lagrangian PDFs are computed from all the lagrangian trajectories over 3 months. Color indicate the density of particles.
We have recent evidence of wind-driven transport in postlarvae collected in Florida Bay (Criales et al. in review). Most postlarval maxima occurred during periods of strong southeasterly winds generated mainly by the passage of hurricanes. Persistent southeasterly winds drive onshore surface Ekman flow in the upper layer on the SWF shelf, carrying larvae to the coastline of Florida Bay. These data support the idea that estuarine ingress is a 2-stage migration process (e.g. Miller and Shanks, 2004), indicating that both wind-driven and tidal transport mechanisms (including non-linear tides) contribute to the shoreward transport and estuarine ingress of postlarvae if they match with larval behavior. Although Ekman flow and associated strong southeasterly wind events may transport larvae from offshore waters into the bay, tides seem to be the most important transport mechanism at the inner shelf (Criales et al. 2007).
Postlarvae may enter Florida Bay through channels of the Florida Keys by the countercurrent generated by eddies or by favorable onshore winds (Munro et al., 1968; Criales et al., 2003). However, the percentage of postlarvae entering through the Upper and Middle Florida Keys was small compared to the percentage entering through the SWF shelf during monthly collections in a 4-yr study (Criales et al., 2006). The transport across the channels of the Lower Florida Keys and Key West will be explored with this biophysical model.
58
E. Analysis of Interrelationships of Pink Shrimp Life Stages Information about the abundance of various life stages of pink shrimp in southwest shelf waters and in Florida Bay will be used in several ways to support the project. Trajectories simulated by the model will be compared to reported locations of postlarval collections. Variation in the abundance of recruits and spawners developed from Tortugas fisheries data will be examined in relation to variation in postlarvae concentrations in passes leading to Florida Bay and densities of juvenile pink shrimp in western Florida Bay. The available data on various life stages are summarized in Table 1.
F. Future Work We will couple the bio-physical model to the high-resolution ROMS with tidal currents in order to implement both diel vertical migration and STST. We will examine the interconnections between behavior and local, mesoscale, and regional scale processes in the migration of pink shrimp from larvae to answer questions that will help refine stock assessments by improving predictions of recruitment to the fishery.
G. References Cited Costello, T. J., Allen, D.M.. 1966. Migrations and geographic distribution of pink shrimp, Penaeus duorarum, of the Tortugas and Sanibel grounds, Florida. Fishery Bulletin 65:449-569). Criales, M.M., Browder, J.A., Mooers, C., Robblee, M.B., Jackson, T.L. 2007. Cross-shelf transport of pink shrimp larvae: interactions of tidal currents, larval vertical migrations, and internal tides. Marine Ecology Progress Series 345:167-184. Criales, M. M., Wang, J., Browder, J.A., Robblee, M.B., Jackson, T.L., Hittle, C. 2006. Cross- shelf transport of pink shrimp postlarvae into Florida Bay via the Florida shelf. Fishery Bulletin 104:60-74. Criales, M. M., Wang, J., Browder, J.A., Robblee, M.B. 2005. Tidal and seasonal effects on transport of pink shrimp postlarvae. Marine Ecology Progress Series 286:231-238. Criales MM, Yeung C, Jones D, Jackson TL, Richards WJ (2003) Variation of oceanographic processes affecting the size of pink shrimp (Farfantepenaeus duorarum) postlarvae and their supply to Florida Bay. Estuar Coast Shelf Sci 57 (3):457–468 Cummings DC (1961) Maturation and spawning of pink shrimp, Penaeus duorarum Burkenroad. Trans Amer Fish Soc 90:462–468 Dutkiewicz S, Griffa A, Olson DB (1993) Particle diffusion in a meandering jet. J Geophys Res 98: 16487-16500 Ehrhardt, N., Legault, C. 1999. Pink shrimp, Farfantepenaeus duorarum, recruitment variability as an indicator of Florida Bay dynamics. Estuaries 22(2B):471-483. Ewald JJ (1965) The laboratory rearing of pink shrimp, Penaeus duorarum Burkenroad. Bull Mar Sci 15:436–449
59 Griffa A. (1996) Applications of stochastic particle models to oceanographic problems. P. 114- 140, In: Adler R.J., Muller R., Rozovskii B.L. [eds.] Stochatic Modelling in Physical Oceanography, Birkhauser. Grimm V, Berger U, Bastiansen F, Eliassen S, Ginot V et al. (2006) A standard protocol for describing individual-based and agent-based models. Ecological Modelling 198:115-126. Hart RA (2008) A biological review of the Tortugas pink shrimp fishery 1960 through 2007. NOAA Technical Memorandum NMFS-SEFSC-573. 28 pp. He R, Weisberg RH (2002) West Florida shelf circulation and temperature budget for the 1999 spring transition. Cont Shelf Res 22:719–748 Iversen ES, Jones AE, Idyll CP (1960) Size distribution of pink shrimp, Penaeus duorarum, and fleet concentrations on the Tortugas fishery grounds. US Fish Wild Ser Spec Sci Rep 356:1–62. Jones, A.C., Dimitriou, D.E., Ewald, J.J., Tweedy, J.H.. 1970. Distribution of early developmental stages of pink shrimp, Penaeus duorarum, in Florida waters. Bulletin of Marine Science 20:634-661. Kelble, CR, Johns, EM, Nuttle, WK, Lee, TN, Smith, RH, Ortner, PB. 2007. Salinity patterns in Florida Bay. Estuarine Coastal and Shelf Science 71:318-334. Kennedy FS and Barber DG (1981) Spawning and recruitment of pink shrimp, Penaeus duorarum, off eastern Florida. J Crust Biol: 14-474-485 Klima, E.F., Patella, F.J.. 1985. A synopsis of the Tortugas pink shrimp, Penaeus duorarum, fishery, 1981-84, and the impact of the Tortugas Sanctuary. Marine Fisheries Review 47(4):11-18. Klima EF, Matthews GA and Patella FJ (1986) Synopsis of the Tortugas pink shrimp fishery, 1960-1983, and the impact of the Tortugas Sanctuary. North Amer J Fish Manag 6, 301- 310 Kutkuhn, J.H. 1966. Dynamics of a penaeid shrimp population and management implications. Fishery Bulletin 65:313-338. Lee TN, Johns E, Williams E, Johns E Wilson, D Smith NP (2002) Transport processes linking South Florida coastal ecosystems. In: Porter JW, Porter KG (eds) The Everglades, Florida Bay, and Coral Reefs of the Florida Keys. Boca Raton, Fl, CRC Press, p 309–342 Lee TN, Johns E, Melo N, Smith RH, Ortner P, Smith D (2006) On Florida Bay hypersalinity and water exchange. Bull Mar Sci 79:301–327 Miller JA, Shanks AL (2004) Ocean-estuary coupling in the Oregon upwelling region: the abundance and transport of juvenile fish and larval invertebrates. Mar Ecol Prog Ser 271:267–279 Munro JL, Jones AC, Dimitriou D (1968) Abundance and distribution of the larvae of the pink shrimp (Penaeus duorarum) on the Tortugas Shelf of Florida, August 1962-October 1964. Fish Bull US 67:165–181 Nance, J., W. Keithy, Jr, C. Caillouet, Jr., J. Cole, W. Galdry, B. Gallaway, W. Griffin, R. Hart, and M. Travis. 2008. Estimation of effort, maximum sustainable yield, and maximum
60 economic yield in the shrimp fishery of the Gulf of Mexico. NOAA Tech Mem, NMFS- SEFSC-570. 71 pp. Nichols S (1984) Updated assessments of brown, white, and pink shrimp in the U.S. Gulf of Mexico. Paper presented at the SEFC Stock Assessment Workshop. Miami, Florida, May 1984. North EW, Gallego A, Petitgas P (2009) Manual of recommended practices for modelling physical-biological interactions during fish early life, ICES RRC 295, 112 pp Okubo A, Levin SA (1989) Diffusion and Ecological Problems: Modern Perspectives, Springer NY, 469 pp. Paris CB, Cherubin LM, Cowen RK (2007) Surfing, spinning, or diving from reef to reef: effects on population connectivity Mar Ecol Progr Ser 347: 285-300. Paris CB, Cowen RK, Lwiza KMM, Wang DP, Olson DB (2002) Objective analysis of three- dimensional circulation in the vicinity of Barbados, West Indies: Implication for larval transport. Deep Sea Research.49: 1363-1386. Smith, N.P. 1997. An introduction to the tides of Florida Bay. Florida Scientist 60:53-67. Smith NP (2000) Transport across the western boundary of Florida Bay. Bull. Mar. Sci. 66: 291-304. Vikebø F, Jørgensen C, Kristiansen T, Fiksen Ø (2007) Drift, growth, and survival of larval Northeast Arctic cod with simple rules of behaviour. Mar Ecol Prog Ser 347:207–219. Weisberg RH, Black BD, Yang H (1996) Seasonal modulation of the West Florida continental shelf circulation. J Geophys Res 23:2247–2250. Weisberg RH, Black BD and Li Z (2000) An upwelling case study on Florida’s west coast. J Geophys Res 105:11,459-469. Werner FE, Cowen RK, Paris CB (2007) Coupled biophysical models: Present capabilities and necessary developments for future studies of population connectivity, Oceanography, 20(3):54-69.
.
61 Table 1. Summary of data available on relative abundance of various early life stages of pink shrimp in south Florida.
Stages Dates Sampling Gear Location Reference
bimonthly -stratified Gulf V plankton 1962-1964 sampling nets Dry Tortugas, Marquesas, Key West Munro et al. (1968)
Larvae & 1959-1962 several cruises plankton nets between DT and FLB across Jones et al. (1970) postlarvae Discovery nets the SWF shelf
Criales and Lee May-June 1991 SEFCAR cruise LH3 MOCNEES Dry Tortugas and Florida Straits (1995)
hourly sampling during 3 Jul-04 days Tucker net SWF shelf- Dry Tortugas, Marquesas Criales et al. (2007) and inshore st.
Oct/97-Jun/99 monthly collections at night channel nets 1 st. in Middle Fl K, 1 sts in Upper FLK Criales et al. (2003) during new moon
2 sts in the western border of Florida 2000-2003 monthly collections at night channel nets Bay Criales et al. (2006) 2 in the bay side of Middel Florida Postlarvae during new moon Keys
Jul/2004 to 1 st in the western border and 5 Criales et al. (in Nov/2004 monthly collections at night channel nets interior review) Jun/2005 to Oct/2005 during new moon sts in northern FLB
nightly collections on Jul-Aug/2006-07 different channel nets 2 interior sts FLB moon phases
Early nigthly and daily collections settlement Jul-Sep/2008 in Renfro net 2 interior sts FLB stages northern-central FLB
62 Roblee et al. (in Juveniles 1984- 1987 collections every 6 weeks Throw-trap Western Florida Bay, 36 samples/coll prep.) 1994-2005
63 Appendix 1.
Relatively high resolution numerical simulations (~6-14 km) for various ocean basins and the global ocean are now accessible directly through the Web. They can be used and analyzed directly on-line through various protocols such as OpenDAP (http://www.opendap.org/faq/whatClients.html), THREDDS and LAS (http://ferret.pmel.noaa.gov/LAS/). These simulations are made with different ocean models (e.g MITgcm for the ECCO2 simulation, HYCOM3, NCOM4), which are operated in a reanalysis and/or an operational forecast mode. Therefore they use data assimilation techniques that constrain ocean models with in-situ observations and remotely sensed observations. These simulations show a great degree of accuracy in the open ocean where most observations are available, however in coastal waters discrepancies are noticeable. One of the main factors is the bathymetry resolution and accuracy, which is inappropriate to resolve coastal circulation at the scale for instance of the Florida Keys. In most of the large scale simulations, the archipelago is represented as a mask and tide is not simulated. Therefore in order to simulate the circulation of the Florida Keys, it was necessary to build a high resolution simulation centered on the region encompassing the most accurate bathymetry and tides. This simulation time span is a decade starting in January 1995 and ending in December 2005. Few larger scale simulations of the global ocean or of the Atlantic Ocean basin, at a high enough resolution are available currently. HYCOM consortium provides the highest horizontal resolution outputs, though the first run with assimilated in-situ data starts in June 2003. Ocean models that were started earlier, aimed at reproducing the climate variability over the past twenty years, provide outputs at a much lower resolution, which does not allow for the Loop Current to be resolve in the Gulf of Mexico, for instance. This is the case for ECCO, which was started in 1992 and has 1/3° horizontal resolution close to the equator, and has daily outputs. Ecco25 provides only monthly outputs for the 3D fields (daily for the 2D fields) at 1/8° horizontal resolution, which contain the Loop Current. As shown in numerous studies, the Loop Current has a strong influence on the western Florida shelf. Therefore its presence, in terms of the conduct of remote changes in the Caribbean and the Atlantic Ocean, is a pre-requisite for accurate simulation of the decadal circulation in the Florida Keys. The TOPAZ6 project has recently started a twenty year reanalysis of the North Atlantic. Their outputs have a 1/12° horizontal resolution, the same as the HYCOM consortium’s, and they assimilate the same type of data as for the HYCOM consortium. They also run HYCOM and their simulation was started in 1980. The first outputs have been made available to our project. However, the ECCO and ecco2 simulation have been validated whereas the current TOPAZ outputs have not been completely evaluated yet. In order to test the effect of lack of resolution and the accuracy of the state of the TOPAZ field, we ran a one month experiment with each fields that will be compared to estimate the bias and trends of their differences.
2 http://ecco.jpl.nasa.gov/external/ 3 http://www.hycom.org/dataserver/ 4 http://www7320.nrlssc.navy.mil/global_ncom/ 5 http://ecco2.jpl.nasa.gov/products/output/cube/ 6 http://topaz.nersc.no/ 08-02 Project Title: Predictors for Larval Transport Success and Recruitment for Cod in the Gulf of Maine
Principal Investigators: James Churchill, Jeffrey Runge, Jim Manning and Loretta O’Brien
Goals: • Assess larval transport success from identified spawning regions in the Gulf of Maine to suitable settlement areas in the gulf, • Relate the year-to-year variation in transport success for each region to environment conditions (particularly mean wind velocity and water temperature), • Assess the viability of utilizing larval transport simulations for indicating larval recruitment success by comparing, over several years, the results of the simulations to NMFS survey data in the Gulf of Maine, • Execute transfer of our computer code so that it can be routinely used by NMFS personnel to predict cod recruitment success.
Approach: In addressing these objectives, we use velocity fields generated by a hydrodynamic model of the Gulf of Maine/Georges Bank region to simulate the movement of cod eggs and larvae and to determine the likelihood that the developing cod arrive at areas suitable for settlement as young juveniles. The model velocities employed thus far were supplied by the Marine Ecosystem Dynamics Modeling group at U. Mass. Dartmouth (courtesy of C. Chen) and were generated by the FVCOM Gulf of Maine/Georges Bank model (http://fvcom.smast.umassd.edu/FVCOM/index.html) (Chen et al., 2006a,b). These are high resolution velocity fields both in time (1 hour interval) and space (order 1 km cell size in the coastal zone). So that our ability to predict larval cod transport in future years will not be compromised should this set of model results become unavailable, we are also developing code to carry out the larval transport simulations using velocity fields generated by other Gulf of Maine models. Currently, our focus is on the velocity fields generated by the Gulf of Maine Ocean Observing System (GoMOOS) nowcast/forecast model (Xue et al., 2000, 2005, 2008). These velocity fields are updated in near real time and are publically available online.
In carrying out the transport simulations, the newly spawned cod eggs are assumed to be buoyant and to reside in the near surface layer, subject to direct forcing by the surface wind stress. The manner in which larval transport is impacted by diel vertical migration of larvae is also examined. The age of settlement capability is assumed to be in the range of 45-60 d (e.g. all cod are assumed to settle by an age of 60 days). Based on the distribution of age-0 juvenile cod reported by Howe (2002), the western Gulf of Maine region suitable as an early stage juvenile habitat is taken as the coastal area with depth of < 30 m.
In initiating the transport simulations, ensembles of cod eggs distributed over a given spawning region are released into the modeled flow field at intervals of 3 days. Each developing cod egg/larvae particle is tracked for 60 days. From the tracks, we determine the ensemble likelihood that the drifting cod arrive at a settlement suitable area at an age when settlement capable, a quantity defined as transport success by Huret et al. (2007). This quantity is taken as
65 the percent of time that particles (larvae) of the ensemble are over depths of less than 30 m (i.e. in a region suitable for settlement) during the last 15 days of their 60 day drift (i.e. the assumed age range of settlement capability). The transport success is averaged over ensembles giving an expected success of transport to settlement areas for releases (spawns) over a given time period.
Transport simulations were carried out over 11 spring and winter spawning events, from 1995 to 2005, and the resulting year-to-year variation in computed transport success was compared with the wind, measured at the NOAA 44013 weather buoy in Massachusetts Bay.
Work Completed: Although cod spawning may occur intermittently in the western Gulf of Maine throughout the year, the appearance of large spawning aggregations in the western gulf is confined to winter and spring “spawning events”. Winter spawning typically extends from November through February and is broadly distributed over the western gulf. Spawning in the spring is concentrated in the area of Ipswich Bay and near Cape Ann. The most intensive spawning of the spring event extends over April through June, with peak activity tending to occur in May.
The results of our analysis indicate that the conditions during spring often favor the retention of cod eggs and larvae within the western Gulf of Maine. The tendency for retention is most strongly tied with the alongshore wind. Cod eggs released from spawning sites during a time of predominately upwelling favorable winds tend to be flushed out of the western Gulf of Maine, as the upwelling circulation carries the developing eggs (presumed to reside in the surface mixed layer) to the Western Maine Coastal Current. Despite its name, this current is typically centered near the 100-m isobath and bypasses cod spawning and nursery areas in Ipswich and Massachusetts Bays. Cod eggs and larvae that are transported to the Western Maine Coastal Current by upwelling circulation are rapidly transported out of the western Gulf of Maine. This is in contrast with the fate of cod released from coastal spawning areas during a time of predominately downwelling favorable along-shore winds. These cod tend to be carried onshore and successfully transported to juvenile habitats in Massachusetts and Ipswich Bay.
The tendency for downwelling circulation to favor retention of cod spawned in the wGoM during spring is reflected in the correlation of the wind with yearly values of recruitment success to the age-1 cod population in the Gulf of Maine, compiled by the National Marine Fisheries Service since 1982. Recruitment success is strongly correlated with mean along-shore wind of May, such that the success is higher when the mean May wind is more strongly downwelling favorable. These findings are summarized by Churchill et al. (submitted; available at www.whoi.edu/scientist/jchurchill/FATE/).
The results of our simulations also indicate that cod eggs spawned during winter may tend to be dispersed much more broadly than cod eggs spawned during spring, a consequence of the strong upwelling winds prevalent during the winter spawning period. This is consistent with recent analysis of genetic markers by Wirgin et al. (2007), Kovach (2009) and Kovach et al. (submitted), which indicates that the cod that spawn in Ipswich Bay during spring are genetically distinct from other regional cod stocks, whereas cod spawning in the western Gulf of Maine in winter are genetically similar to cod of other regions, particularly in Nantucket Shoals and the eastern New York Bight. Based on this finding, and recent analysis of cod tagging data (Howell
66 et al., 2008; Tallack, 2009), we have formulated to a two-population hypothesis to explain the fine-scale genetic heterogeneity of the western Gulf of Maine cod stock. One population is comprised of the spring spawning cod, which are hypothesized to have limited migration and a strong tendency for natal homing to historical spring spawning areas. This second population is comprised of the winter spawning cod. Relative to the spring spawning sub-stock, these are hypothesized to be more broadly dispersed in the larval stage and be more active migrators as adults. The possibility that two genetically different fish populations, with markedly different life histories, may be contained with a single stock assessment unit clearly has important implications for fisheries management. To better understand the drivers and consequences of this observed genetic heterogeneity in the western Gulf of Maine cod stock, we (Runge and Churchill) have joined with colleagues in submitting a multidisciplinary proposal to the National Science Foundation to study all life stages, from early planktonic to adult, of the two genetically different cod populations in the western Gulf of Maine.
Applications: Our analysis has shown that the mean alongshore wind velocity measured in the Massachusetts Bay during May is a robust indicator for the recruitment success of age-1 cod to the Gulf of Maine cod stock, roughly 90 % of which is contained in the western Gulf of Maine. In particular, a mean downwelling favorable wind during May gives a strong indication of high recruitment success. The five years of highest recruitment success of Gulf of Maine cod, as measured by the recruits/spawning stock biomass ratio, occurred during the five years when the mean May winds in Massachusetts Bay were most downwelling favorable (from the NW).
As detailed above, this result is consistent with the findings of our larval transport simulations. So that these simulations can be used as a tool to investigate the details of cod larval transport and retention in the western Gulf of Maine, we are working to make the simulation code easily usable by NMFS personnel and capable of utilizing velocities from a number of hydrodynamic models.
Publications/Presentations/Webpages:
Publications: Churchill, J.H. and J. Runge (2009) What Maintains the Western Gulf of Maine Cod Stock? Proceedings of a workshop on Exploring Fine-scale Ecology for Groundfish in the Gulf of Maine and Georges Bank, 2-3 April 2009, York, Maine, 8 pp. [http://www.gmri.org/community/seastate/Churchill_Jim/ex_abstract_churchill_revised.pdf]
Churchill, J.H. J. Runge and C. Chen, Processes controlling retention of spring-spawned Atlantic cod in the western Gulf of Maine and their relationship to an index of recruitment success. submitted to Fisheries Oceanography.
Presentations: Churchill, J.H and J.A. Runge, What Maintains the Western Gulf of Maine Cod Stock? Workshop on Exploring Fine-scale Ecology for Groundfish in the Gulf of Maine and Georges Bank, York, Maine, 2-3 April 2009.
67 Churchill, J.H., Modeling larval cod transport and recruitment. Coastal Ocean Fluid Dynamics Laboratory seminar, 22 May 2009.
Runge, J.A., A. Kovach, R. Jones, S. Tallack, J. Churchill, C. Chen, G. Sherwood, H. Howell, J. Grabowski and D. Berlinsky, Understanding climate impacts on the spatial dynamics of Atlantic cod in coastal waters of the Gulf of Maine. ICES Cod and Climate Workshop at the GLOBEC Open Sciences Meeting, Victoria, BC, June 2009.
Webpage: http://www.whoi.edu/scientist/jchurchill/FATE/
Literature Cited in Report: Chen, C., G. Cowles, and R. C. Beardsley. 2006a. An unstructured grid, finite-volume coastal ocean model: FVCOM User Manual, 315 pp. SMAST/UMASSD Chen, C., R. C. Beardsley, and G. Cowles. 2006b. An unstructured grid, finite-volume coastal ocean model (FVCOM) system. Special Issue entitled “Advance in Computational Oceanography”, Oceanography, 19, 78-89. Churchill, J.H. J. Runge and C. Chen, Processes controlling retention of spring-spawned Atlantic cod in the western Gulf of Maine and their relationship to an index of recruitment success. submitted to Fisheries Oceanography. Howe, A., S. Correia, T. Currier, J. King and R. Johnston (2002). Spatial distribution of ages 0 and 1 Atlantic cod (Gadus morhua) off the eastern Massachusetts coast, 1978–1999, relative to ‘Habitat Area of Special Concern’. Technical Report 12, Massachusetts Division of Marine Fisheries, Pocasset, MA. Howell, W.H., M. Morin, N. Rennels and D. Goethel (2008) Residency of adult Atlantic cod (Gadus morhua) in the western Gulf of Maine. Fisheries Research, 91, 123-132. Huret, M., J.A. Runge, C. Chen, G. Cowles, Q. Xu, and J.M. Pringle (2007). Dispersal modeling of fish early life stages: sensitivity with application to Atlantic cod in the western Gulf of Maine. Marine Ecology Progress Series, 347, 261-274. Kovach, A.I., T.S. Breton, D.L. Berlinsky, L. Maceda, and I.Wirgin. Submitted. Fine-scale spatial and temporal genetic structure of Atlantic cod in U.S. coastal waters. Marine Ecology Progress Series Kovach, A. I. 2009. Genetic insights into the stock structure of cod in US waters. Exploring Fine-scale Ecology for Groundfish in the Gulf of Maine and Georges Bank. Available from
68 Xue, H., L. S. Incze, D. Xu, N. Wolff, and N. R. Pettigrew. 2008. Connectivity of lobster populations in the coastal Gulf of Maine Part I: Circulation and larval transport potential. Ecological Modelling, 210, 193-211. Xue, H., L. Shi, S. Cousins, and N. R. Pettigrew. 2005. The GoMOOS Nowcast/Forecast System. Continental Shelf Research, 25, 2122-2146. Xue, H., F. Chai, and N. R. Pettigrew. 2000. A model study of seasonal circulation in the Gulf of Maine. Journal of Physical Oceanography, 30, 1111-1135.
69 08-03 Project Title: Developing Ecological Indicators for Spatial Fisheries Management using Ocean Observatory Defined Habitat Characteristics in the Mid-Atlantic Bight
Principal Investigators: John Manderson1, Josh Kohut2, Matthew Oliver3 Graduate Students: Laura Palamara2, Steven Gray2 Research associate: John Goff4 1. Ecosystems Processes Division, NEFSC/NMFS/NOAA, James J. Howard Marine Highlands, NJ. 2. Institute of Marine and Coastal Science, Rutgers University, New Brunswick, NJ. 3. College of Marine and Earth Studies, University of Delaware, Lewes, DE. 4. University of Texas Institute for Geophysics, Austin Texas
Goals: Our objective is to develop coastal ocean habitat indicators for important fishery resource species by applying modern statistical habitat modeling techniques to data collected by the North East Fisheries Science Center (NEFSC) and the Mid-Atlantic Regional Coastal Ocean Observing System (MARCOOS). Specifically we will identify benthic and pelagic habitat features related to the main variations in fish community structure, and to variations in the abundance of select resource species in the region. We intend to produce results useful for constructing dynamic habitat maps that can be considered in fisheries management plans. To achieve this objective we have set the following 4 goals:
1) Merge NEFSC bottom trawl and hydrographic data, NOAA bottom habitat maps, with MARCOOS remotely sensed ocean observatory data describing demersal fishery resources and potentially important benthic and pelagic habitat features in the mid- Atlantic Bight coastal ocean.
2) Use multivariate analyses of merged data to quantify fish and macro-invertebrate community variation explained by pelagic and benthic habitat characteristics, and describe the spatial dynamics of important environmental gradients affecting fish and invertebrate distributions.
3) Informed by goal # 2 analysis, train and test single species habitat models for several federally managed species including longfin squid and summer flounder.
4) Develop outreach strategies and products to effectively communicate results of #2 and #3 to important stakeholder groups including state and federal fisheries managers.
Approach: We have adopted the following approaches to meet these 4 goals:
1) Identify the spatial and temporal grain and extent of data we can use to train statistical habitat models. Acquire and collate available data, and derive additional, potentially important habitat variables. Identify and acquire independent datasets we will use to quantify model uncertainty and predictive power.
70
2) Use advanced multivariate techniques to identify subsets of non-collinear habitat variables strongly related to fish community variation measured with trawl surveys. Perform partial analysis to quantify the relative “explanatory” power of benthic and pelagic habitat characteristics measured by NOAA ships and MARCOOS remote sensing technologies. These partial analyses will also be used to select a reduced habitat dataset composed of non redundant variables for the final analysis. Perform final multivariate analysis to identify the main environmental gradients associated with fish community variation in the MAB. Quantify and visualize the spatial dynamics of derived gradients.
3) Use generalized additive modeling (GAM) to develop parsimonious and ecologically sensible statistical habitat models for target species guided by results of #2. In an effort to ensure final GAMs include habitat covariates likely to reflect ecological effects, our model building approach blends objective variable selection with final variable elimination based on an ecological knowledge base developed from peer reviewed literature. We will assess uncertainty and predictive power of final GAMs by comparing predictions with data not used in model training.
4) Identify science and policy issues facing federal marine fishery managers and scientists in the Mid-Atlantic to develop a project framework which is supportive of fishery management under the Magnuson Stevens Act. By developing outreach strategies such as (a) online surveys to fishery managers and scientists, (b) presentations to fishery management councils, and (c) individual consultations with the council’s Science and Statistical Committee (SSC) and technical staff, we will tailor our products to meet the specific needs of Atlantic Coast fishery professionals. This approach increases the likelihood our research will be integrated into the management decision-making process.
Work Completed: We started work on the project in September 2008 when Laura Palamara, who is focused on statistical habitat modeling, began as a Masters candidate at the Institute of Marine and Coastal Sciences, Rutgers University. Steven Gray, an ecology PhD candidate at Rutgers, also joined the team in September to evaluate the implications of our research to fisheries management and to implement an outreach plan. In addition, Dr. John Goff, an expert in sub-marine geology, joined the team as an associate in the fall of 2008. Our team has completed the following tasks pursuant to our 4 goals:
Goal #1) Based on an examination of archived bottom trawl collections, and benthic and pelagic habitat data available from NEFSC,MARCOOS and the usSEABED database (Reid et al. 2005), we determined that model building could be performed with data collected from February 2003 through October 2007, between latitudes 37.14 & 40.85 N and Longitudes -70.83 & - 75.16W (Fig. 1). The performance of models developed with 2003-2007 data will be evaluated using 2008 data.
Species data: Ground fish and invertebrates collections were made by the National Marine Fisheries Service, Northeast Fisheries Science Center’s (NMFS-NEFSC) autumn, winter, and spring bottom trawl surveys. The survey design and trawl characteristics are described in Azarovitz (1981). Winter cruises occurred in February (year day 39-57), spring cruises between
71 March and the beginning of May (year day 63-123), and autumn cruises were made from the beginning of September through late October (year day 243-313). An average of 101 stations was sampled in the study area during the spring and autumn cruises. An average of 70 stations was sampled during the winter. Survey tows were made with a # 36 Yankee trawl equipped with rollers and a 10.4 m wide x 3.2 m high opening. The net was constructed with a 12.7 cm stretched mesh at the opening, an 11.4 cm stretched mesh cod end, and 1.25 cm stretched mesh lining in the codend and upper belly. The net was towed at ~3.5 knots over the bottom for 30 minutes. Distances nets were towed over the bottom averaged 1.9 km (95% Confidence limits 1.75-2.01 km). Trawl tows were made throughout the 24 hour day.
Benthic habitat data: We computed topographic bottom habitat characteristics from the 3-arc- second NGDC Coastal Relief Model ( http://www.ngdc.noaa.gov/mgg/coastal/coastal.html; 93m cell size; Fig. 1). We used circular moving window analysis in GRASS GIS software to calculate median and standard deviations of indices of bottom texture including depth, aspect, slope, profile and tangential curvature from the relief model (Neteler and Mitasova 2008). We selected a window diameter of 1950 meters since this spatial grain was similar to the median length of NEFSC trawl tows (1910 meters). Profile and tangential curvature statistics measured the concavity (=valleys: - values) and convexity of the surface (=ridges: +values) parallel and tangential to major axes of slope, respectively (Neteler and Mitasova 2008). Dr. John Goff, who has published several articles describing the sub-marine geology of the mid-Atlantic Bight (Goff et al. 2005, Goff et al. 2008), provided a map of sediment grain size derived by interpolation of records in the usSEABED data base (Reid et al., 2005). The sediment map was constructed with a spatial grain of 2000 meters using sampling bias correction, maximum-likelihood resampling, and a spline-in-tension algorithm described in Goff et al. (2005 & 2008).
NEFSC hydrographic data: We used Conductivity, Temperature and Depth (CTD) profiles measured during NEFSC trawl surveys to describe several potentially important pelagic characteristics. These characteristics included surface and bottom temperatures and salinities, as well as derived variables describing water column structure and stability. The derived variables were a stratification index calculated as the density difference between 50m and the surface, the “mixed layer” depth at which density was 0.125kg m-3 higher than at the surface layer (Levitus, 1982), and Simpson’s potential energy anomaly (PE; Simpson, 1981). Since Simpson PE was positively correlated with bottom depth, we limited the calculation of the PE anomaly to the upper 30 meters of the water column. Maureen Taylor of NEFSC graciously provided us with this data.
MARCOOS remote sensing data: We used several variables derived from ocean surface current data collected with High Frequency (HF) radar in our analysis. HF radar systems, typically deployed along the coast, use Bragg peaks within a signal (3 ~ 30 MHz) scattered off the ocean surface to calculate radial components of the total surface velocity at a given location (Barrick et. al., 1977). At a frequency of 5 MHz, the radial current vectors from the coastal sites are geometrically combined to produce total vector surface current maps each hour with a resolution of 6 km across the entire shelf. Rutgers University, along with its regional MACOORA partners, maintains an extensive HF radar network between Cape Hatteras, NC and Cape Cod, MA. For our analysis the entire record at each grid point was detided using a least-squares fit of the five strongest constituents
72 (M2, S2, N2, K1, and O1) to the raw time series data. The sub-tidal data were then low pass filtered with a cutoff period of 30 hours. The surface data used in this study only included grid points that had at least 25% return over the annual record. Surface divergence and surface vorticity, normalized by the local Coriolis parameter, of the lowpass filtered fields were calculated using finite difference. We considered one and eight day average low-pass filtered cross-shore and along-shore velocity, divergence, vorticity, and the variance of the raw fields within 10 km of each trawl sample in our modeling. Seasonal trends in divergence and vorticity were also calculated. All instantaneous divergence (represented as vertical velocity) values were assigned a new value of -1 (if below a threshold of -0.1 m/day), 0 (if between -0.1 and +0.1 m/day), or +1 (if above +0.1 m/day). For each point on the grid, these new values were averaged over each season (both as individual seasons, i.e. spring 2004, and as total seasons, i.e. spring average for 2003-2007). The same processing was performed to generate vorticity trends using threshold values of ±0.02. We also considered sea surface temperature and ocean color measured by satellites in our models. We used NOAA Advanced Very-High Resolution Radiometer (AVHRR) day and night- time passes at 1km spatial resolution for sea surface temperature. Individual passes of sea surface temperature were de-clouded by comparing passes to monthly climatological averages from PODAAC and by comparing each individual pass to all passes within 24 hrs to detect and remove quickly moving clouds. We used the Moderate Resolution Imaging Spectrometer (MODIS-Aqua) for ocean color. Data downloaded from the ocean color website (oceancolor.gsfc.nasa.gov) was binned to 1km resolution with the standard data quality flags using Seadas v5.3. We considered measurements of chlorophyll (mg m-3), and normalized water leaving radiance (W m-2 st-1 um-1) at 412, 443, 488, 531, 551, and 667nm. We used SST and ocean color values located within 10 km of each trawl sample We are still in the process of deriving statistically significant water masses using the ensemble clustering approach of Oliver et al. (2004) and Oliver and Irwin (2008). The approach classifies water masses based on normalized water leaving radiance at 443nm and 551nm and SST and has been verified in this region using available in situ observations. This analysis will provide distances of each trawl sample to water mass boundaries as well as the strength of the boundary. The processing of water mass classification is computer intensive and ongoing.
Goal #2) We have performed multivariate analysis using the vegan library in R software (Oksanen et al. 2009, R Core Team 2009). We limited the analysis to the 65 fish and invertebrate species occurring in at least 10 trawl samples (Table 1). Initial canonical correspondence analyses (CCA) were performed independently on bottom, shipboard pelagic, and IOOS pelagic habitat data sets to select non-collinear variables with explanatory power. For each habitat set, variables were removed by backward selection until all variables included in the model had significant marginal effects in Monte Carlo permutation tests at an alpha level of 0.01. In analysis of IOOS data, eight-day averaged satellite and HF radar data were selected instead of one-day averages to increase sample sizes. Seasonal divergence and vorticity trends for individual years were used instead of seasonal composites so as to consider interannual variability. To eliminate redundancy between several environmental variables remaining in the data sets, GAMs were fit to highly correlated variables and residuals of one variable in each pair were used in the final models. This was performed with slope and depth, bottom salinity and depth, and 551 nm irradiance and 488 nm irradiance. We used residuals of the first variable and
73 raw values for the second variable listed in each pair in CCA analyses. The habitat variables selected using this approach were:
Benthic habitat: natural log of depth, slope residuals, profile curvature, and sediment grain size. Pelagic habitat measured with ships: bottom temperature, bottom salt residuals, mixed layer depth, and Simpson’s potential energy. Pelagic habitat measured with IOOS: 488 nm irradiance, 551 nm irradiance residuals, sea surface temperature, cross-shore velocity, cross-shore variance in velocity, and divergence trend.
Using these habitat data matrices, we have performed a partial CCA to quantify the proportion of species variation measured in trawl surveys related to benthic habitat, and pelagic habitat measured with ships and IOOS using the method of Borcard and Legendre (1994; see also Cushman and McGarigal, 2002) . A total of 25.0% of the species variance was explained by the 14 environmental variables included. Shipboard pelagic habitat data had the highest explanatory power, followed by IOOS pelagic habitat data and then the benthic habitat data (Fig. 2). Overlap (= data redundancy) was especially high for IOOS and shipboard habitat data (~25% of species variance explained), indicating that remote sensing can provide a relatively good representation of some pelagic habitat characteristics measured with ship surveys. In addition, ~ 17% of “explained” species variation was described by IOOS remotely sensed data alone. We have performed a full CCA using variables above to define environmental gradients associated with variations in fish community structure in the MAB. We are currently interpreting and evaluating the results and our methods in preparation for a final multivariate analysis that will consider water mass type and proximity to water mass boundaries classified using satellite data.
Goal #3) We have nearly completed training a single species GAM habitat model for longfin squid (Loligo pealeii; Fig. 3). All GAM modeling has been performed using the mgcv package in R software (Wood 2006). Examination of squid catch data indicated that it was best modeled using an over dispersed Poisson distribution. Trawl tow distance was used as an offset throughout the modeling process. Our objective model selection used a backward stepwise approach that minimized the model generalized cross validation (GCV) statistic (Wood, 2006). We set gamma to 1.4 which inflated the penalty for degrees of freedom to 1.4 in the GCV calculation. This reduced the chances of over fitting models. In the initial GAM we included all covariates considered in the multivariate community analysis. We used a spline smoother to model single term covariates which we eliminated beginning with those with the lowest significance values. Covariates were eliminated from the model until GCV increased. We then examined all possible combinations of first order interactions among selected covariates, retaining those that caused a further decrease in model GCV. Interactions were modeled using tensor product smooths which are appropriate when interacting covariates are measured on different scales. The model built using this stepwise approach was further reduced by eliminating covariates for which 2 standard error confidence bands of species responses included 0 throughout the covariate domain in deviance plots. Finally we examined the objective model in the light of a literature review of the species. Several covariates were removed from the objective model to produce a final ecologically defensible model. The habitat model we produced indicated that squid abundance was strongly related to variations in surface current divergence trends, water leaving radiance of 488 nm wavelength,
74 sediment grain size, bottom temperature, surface current velocity, and an interaction between sea surface temperature and depth (Table 2). Together these habitat covariates accounted for ~ 85% of the squid abundance variance. Few squid were collected at bottom water temperatures less than ~ 7.5ºC and depth distributions of squid were seasonally dependent as indicated by the significant interaction between station depth and sea surface temperature (Fig. 4 a,b). During fall surveys when sea surface temperatures were warm, large catches of squid were made where depths were < 100 m. During the winter and spring when sea surface temperatures were cold, squid were primarily collected on the outer continental shelf and slope where bottom depths ranged from 100 to 250 meters (Fig. 4b). The animals were abundant over bottoms with fine gravel to coarse sand sediment and in areas where surface current velocities were relatively high (Fig 4c,d). Finally, squid were abundant in all seasons at locations with a high probability of surface current divergence (i.e. upwelling) and moderate amounts of colored organic matter as estimated by satellites (488nm =0.72-0.9 W m-2 st-1 um-1) (Fig. 4f, g). These areas were associated with the Hudson Shelf Valley and the shelf slope sea. Several large squid collections were also made in areas where convergence and frontal formation were likely (Fig. 4f). Habitat associations of longfin squid identified with our GAM are consistent with those identified for the species in the North West Atlantic coastal ocean and summarized in Jacobson (2005). We are in the process of acquiring and collating the 2008 NEFSC and MARCOOS data we will use to assess the performance of the squid habitat model. We have also begun to develop a single species GAM habitat model for summer flounder using the same approach described above. We are considering changing our third focal species from spiny dogfish to butterfish, an important by-catch species in squid fisheries, to develop habitat modeling techniques useful for addressing by-catch issues.
Goal #4: To gain an understanding of the needs of fishery decision-makers and to increase the applicability of our research, we have developed significant dialog with fisheries management institutions and other stakeholder groups. In the fall of 2008 we distributed our Fate Proposal to, and conducted an online survey of, 43 fishery professionals along the Atlantic coast to determine where our results might best fit into the management decision-making structure (Fig. 5). Using this survey we began to develop an outreach framework that included presentation of preliminary results at the Mid Atlantic Fisheries Management Council (MAFMC) meeting in New York City in June, 2009. Our intent was to more formally solicit advice on ways we might tailor final products to meet the MAFMC’s needs. Since then, we have been working closely with MAFMC, their technical staff and SSC, as well as other stakeholder groups including Pew Charitable Trusts, the Recreational Fishing Alliance, and United National Fishermen Association, to develop final products useful for fishery and habitat management in the mid-Atlantic Bight. Based on consultations we believe our results could inform management decisions in the following areas:
Modeling • Outline Essential Fish Habitat (EFH) and Habitat of Particular Concern (HPAC) for managed species • Simulate possible future population changes associated with shifts in water temperature and productivity with climate change Assessment
75 • Provide insight into ecological indicators useful as covariates in stock assessment • Provide insight into habitat conditions affecting abundance and distribution independently of fishing pressure that could be used to tighten assessments, and decrease uncertainty of reference points • Identify pelagic features not currently considered in fisheries habitat management plans Management: • Curb by-catch by developing models predicting areas and times of overlap (and non- overlap) for target and non-target species Management Measures: • Spatial Management strategies for Fishing and Non-fishing impacts • Designation of Marine Protected Areas • Temporally and spatially explicit closures
Applications: Results from our preliminary work have produced several potentially important applications for fisheries management. New directions, including federal mandates, require shifting perspectives from traditional single-species management towards an ecosystem-based approach to assessment and decision-making. In an effort to help management work toward these new objectives we are refining our research framework to try to meet management needs in the following areas:
a) Outline and articulate temporally and spatially explicit Essential Fish Habitat (EFH) as required by the Magnuson Stevens Act. b) Identify environmental data that could be used as covariates in single species stock assessment. c) Provide insight into the role changing environmental conditions may play in stock rebuilding, independently of fishing pressure. d) Provide techniques to curb by-catch in single species fisheries by developing habitat models predicting habitat overlap (e.g. butterfish by-catch in the longfin squid fishery). e) Providing a first step toward the development of spatially explicit, mechanistic habitat models estimating environmental effects on dispersal, growth, survival and reproduction of key fishery and keystone species in the ecosystem.
Publications/Presentations/Webpages: Palamara, L., J.P. Manderson, J. Kohut, M. Oliver, J. Goff, S. Gray (2009) Developing Ecological Indicators for Spatial Fisheries Management using Ocean Observatory Defined Habitat Characteristics in the Mid-Atlantic Bight. to be presented FATE Annual PI Meeting September .2009 – Seattle, Washington
Gray, S. (in press) Are robots and satellites the future of fisheries management? Fisheries.
Manderson, J.P., J. Kohut, L. Palamara and B. Phelan. 2009. Fish and Physics: Using Ocean Observing Systems to Measure effects of Seascape Dynamics of Fisheries Recruitment. White paper and Presentation to be presented September, 2009 at the ICES annual Science Conference, Berlin Germany. Contribution No. CM K:11 2009.
76
Manderson, J., Palamara, L., Kohut, J., Oliver, M., Goff, J., Gray, S. (2009) Developing Ecological Indicators for Fisheries Management using IOOS Defined Habitat Characteristics in the Mid-Atlantic Bight. Mid-Atlantic Fisheries Council, New York, NY. June 2009.
Oliver, M., A. Irwin, O. Schofield, P. Falkowksi (2008) Objective Global Ocean Biogeographic Provinces. Joint Carbon/Biodiversity Meeting, Washington DC.
Manderson, John, Josh Kohut, Matthew Oliver (2008) Developing Ecological Indicators for Fisheries Management using IOOS Defined Habitat Characteristics in the Mid- Atlantic Bight. FATE Annual PI Meeting 2008 – LaJolla, California
77 Literature Cited
Azarovitz, T. 1981. A brief historical review of the Woods Hole Laboratory trawl survey time series. Pages 62–67 in W. G. Doubleday and D. Rivard, editors. Bottom Trawl Surveys. Canadian Special Publication Fisheries and Aquatic Sciences. Vol 58 Barrick, D.E., M.W. Evens and B.L. Weber, 1977: Ocean surface currents mapped by radar. Science, 198, 138-144. Cushman, S. A. and K. McGarigal. 2002. Hierarchical, Multi-scale decomposition of species- environment relationships. Landscape Ecology 17:637-646. Goff, J. A., C. J. Jenkins, and S. J. Williams. 2008. Seabed mapping and characterization of sediment variability using the usSEABED data base. Continental Shelf Research 28:614– 633. Goff, J. A., L. A. Mayer, P. Traykovski, I. Buynevich, R. Wilkens, R. Raymond, G. Glang, R. L. Evans, H. Olson, and C. Jenkins. 2005. Detailed investigation of sorted bedforms, or "rippled scour depressions," within the Martha's Vineyard Coastal Observatory, Massachusetts. Continental Shelf Research 25:461. Jacobson, L.D. 2005. Essential fish habitat source document: Longfin Inshore Squid, Loligo pealeii, life history and habitat characteristics. NOAA Techical Memorandum NMFS- NE-193. Levitus, S. 1982. Climatological atlas of the world's oceans. NOAA Professional Paper 13:173. Neteler, M. and H. Mitasova. 2008. Open source GIS: A GRASS GIS Approach. 3 edition. Springer. Oksanen, R. Kindt, P. Legendre, B. O'Hara, G. L. Simpson, P. Solymos, M. H. H. Stevens, and H. Wagner. 2009. vegan: Community Ecology Package. Oliver, M. J., S. M. Glenn, J. T. Kohut, A. J. Irwin, O. M. Schofield, M. A. Moline, and W. P. Bissett. 2004. Bioinformatic approaches for objective detection of water masses on continental shelves. Journal of Geophysical Research 109. Oliver, M. J. and A. J. Irwin. 2008. Objective global ocean biogeographic provinces. Geophysical Research Letters 35 10.1029/2008GL034238, 032008. Reid, J. M., J. A. Reid, C. J. Jenkins, M. E. Hastings, S. J. Williams, and L. J. POPPE. 2005. usSEABED: Atlantic coast offshore surficial sediment data release. U.S. Geological Survey Data Series. Simpson, J. H. 1981. The shelf fronts: implications of their existence and behaviour. Philosophical Transactions of the Royal Society, London A 302:531–546. R Core Team 2009. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Wood, S. 2006. Generalized Additive Models: An Introduction with R. Chapman & Hall.
78 Table 1. Scientific and common names of the 65 species collected in NEFSC bottom trawl surveys from 2003 through 2007 and included in the multivariate analysis. Scientific Name Common Name Alosa aestivalis Blueback Herring Alosa pseudoharengus Alewife Alosa sapidissima American Shad Ammodytes dubius Northern Sand Lance Antigonia capros Deepbody Boarfish Argentina striata Striated Argentine Cancer borealis Jonah Crab Cancer irroratus Atlantic Rock Crab Centropristis striata Black Sea Bass Chlorophthalmus agassizi Shortnose Greeneye Citharichthys arctifrons Gulf Stream Flounder Clupea harengus Atlantic Herring Congridae Conger Eel (unclassified) Dipturus laevis Barndoor Skate Etropus microstomus Smallmouth Flounder Geryon quinquedens Red Deepsea Crab Glyptocephalus cynoglossus Witch Flounder Helicolenus dactylopterus Blackbelly Rosefish Hemitripterus americanus Sea Raven Homarus americanus American Lobster Illex illecebrosus Northern Shortfin Squid Lepophidium profundorum Fawn Cusk-Eel Leucoraja erinacea Little Skate Leucoraja garmani Rosette Skate Leucoraja ocellata Winter Skate Limanda ferruginea Yellowtail Flounder Liparis atlanticus Atlantic Seasnail Loligo pealeii Longfin Squid Lophius americanus Goosefish Macrozoarces americanus Ocean Pout Macrorhamphosus scolopax Longspine Snipefish Macrouridae Grenadier (unclassified) Majidae Spider Crab (unclassified) Melanogrammus aeglefinus Haddock Menidia menidia Atlantic Silverside Merluccius albidus Offshore Hake Merluccius bilinearis Silver Hake Mustelus canis Smooth Dogfish Myctophidae Lanternfish (unclassified) Myliobatis freminvillei Bullnose Ray Myoxocephalus octodecemspinosus Longhorn Sculpin Myxine glutinosa Atlantic Hagfish Ophichthidae Snake Eel (unclassified)
79 Scientific Name Common Name Paralichthys dentatus Summer Flounder Paralichthys oblongus Fourspot Flounder Peprilus triacanthus Butterfish Peristedion miniatum Armored Searobin Petromyzon marinus Sea Lamprey Placopecten magellanicus Sea Scallop Pomatomus saltatrix Bluefish Prionotus carolinus Northern Searobin Prionotus evolans Striped Searobin Pseudopleuronectes americanus Winter Flounder Raja eglanteria Clearnose Skate Scophthalmus aquosus Windowpane Scomber scombrus Atlantic Mackerel Scyliorhinus retifer Chain Dogfish Sepiolidae Bobtail (unclassified) Squalus acanthias Spiny Dogfish Stenotomus chrysops Scup Symphurus sp. Tonguefish (unclassified) Urophycis chuss Red Hake Urophycis regia Spotted Hake Urophycis tenuis White Hake Zenopsis conchifera Buckler Dory
80
Table 2. Final Generalized Additive Habitat model for Longfin Squid collected in NEFSC bottom trawl surveys in the study area from 2003 through 2007. s indicates spline smoother used to model individual covariates. Te indicates where tensor product smooths were used to model interactions.
Parametric coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 1.3587 0.7107 1.912 0.0568 . --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Approximate significance of smooth terms: edf Ref.df F p-value s(Sea Surface Divergence) 6.026 6.026 10.014 3.29e-10 s(Sediment grain size) 8.136 8.136 7.531 2.30e-09 s(Bottom temperature) 7.038 7.038 5.506 4.98e-06 s(488 nm reflectance 6.533 6.533 7.603 4.02e-08 te(along & cross shelf velocity) 20.595 20.595 3.647 3.14e-07 te(SST, Bottom depth) 10.903 10.903 6.184 3.25e-09
R-sq.(adj) = 0.853 Deviance explained = 80.5% GCV score = 520.02 Scale est. = 379.57 n = 395
81
Fig. 1. Locations of NEFSC bottom trawl samples collected from 2003 through 2007 and used in multivariate and generalized additive modelling overlaid on the NGDC Coastal Relief Model.
82
Fig. 2. Venn diagram derived from partial canonical correspondence analysis used to quantify the power (%) of benthic, shipboard CTD and IOOS habitat matrices to explain fish community variation in the mid-Atlantic Bight.
83 Fig. 3. Relative seasonal abundance of longfin squid in NEFSC bottom trawl samples collected between 2003 and 2007 and used to train the GAM habitat model. + symbols indicate trawl tows in which squid were absent.
84 Fig. 4. Deviance plots from the final GAM showing the additive effects of habitat covariates on squid abundance. Effects of A) bottom water temperature, B) interaction between sea surface temperature and bottom depth.
85 Fig. 4 continued. GAM deviance plots of the additive effects of C) sediment grain size, and D) surface currents measured with HF radar, on longfin squid abundance.
86 Fig. 4 continued. GAM deviance plots of the additive effects of E) seasonal surface current divergence trend s, and F) Water leaving radiance at 488 nm on longfin squid abundance.
87
Fig. 5. Schematic showing information flow in the fisheries management process under the Magnuson Stevens Fisheries Management Act. We believe results of this FATE funded project can be integrated into Observation/Modeling and Data Product areas.
88
08-04 Project Title: An integrated assessment of the physical forcing and biological response of the pelagic ecosystem of the northern California Current, focused on coastal pelagics and salmon
Principal Investigators: William T. Peterson and Robert L Emmett
Goals: In our proposal we identified six goals, as follows:
1. Provide an assessment of ecosystem baseline conditions in the pelagic environment for the coastal NCC; 2. From the baseline, calculate anomalies that will illustrate whether a set of observations in a given year are above or below the baseline threshold values; 3. Use the anomaly data sets to determine if the NCC is, in a relative sense, productive, neutral, or unproductive, for each food chain component; 4. Assessment of the stressors (causes and consequences) on the ecosystems (Drivers and Pressures) – PDO state, coastal upwelling, and seasonal cycles of plankton production 5. Provide forecasts of pelagic ecosystem status in general, and salmon production in particular, that are based on a set of physical and ecological indicators; 6. Evaluate the success of forecasts, and update states relative to baseline conditions
Approach:
We continued work begun as part of earlier FATE funding which supported development of indices and posting indices to our Center’s web page.
Work Completed:
Completed 30 research cruises along the Newport Hydrgraphic Line between Aug 2008 and July 2009. These cruises are the source of several of our key indicators of salmon survival including SST, date of biological spring transition, copepod species richness and northern copepod biomass anomalies. All zooplankton samples were enumerated and entered into our data base to facilitate calculation of the copepod indices.
All physical and biological indices were updated monthly and posted to our web-page in January 2009 (see color chart below). Forecasts of salmon returns for 2009 (coho) and 2010 (Chinook) were also posted to the web in January 2009. This effort involves updating the following tasks: averaging the PDO for December-March and May-September; monthly averaging SST from the NOAA Buoy 46050 for December-March and May-September; monthly averaging SST from our research cruises; calculation of date of physical spring transition (following Logerwell et al. 2003); upwelling index for April-May; temperature and salinity of deep water on the shelf averaged over May-September, length of the upwelling season (following Bograd et al. 2008); calculation of three copepod indicators (species richness, northern copepod anomaly and date of biological spring transition).
89 Environmental Variables 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 PDO (December-March) 11 5 2 8 4 12 7 10 9 6 3 1 PDO (May-September) 8 2 3 4 5 10 9 11 7 6 1 MEI Annual 11 1 3 5 10 9 7 8 6 4 2 MEI Jan-June 11 2 3 5 7 9 6 10 4 8 1 SST at 46050 (May-Sept) 9 2 4 5 1 7 11 8 6 10 2 SST at NH 05 (May-Sept) 8 2 1 4 7 6 11 10 5 9 3 SST winter before going to sea 12 7 5 6 4 8 11 10 9 3 1 2 Physical Spring Trans (Logerwell) 7 6 2 1 4 9 8 11 9 3 5 Upwelling (Apr-May) 7 1 11 3 6 10 9 12 7 2 4 5 Deep Temperature at NH 05 11 4 6 2 2 7 8 10 9 5 1 Deep Salinity at NH05 11 3 3 5 8 9 10 7 6 1 1 Length of upwelling season 6 3 2 9 1 10 8 11 5 4 7
Copepod richness 11 2 1 5 3 8 7 10 9 6 4 N.Copepod Anomaly 11 8 3 5 2 9 6 10 7 4 1 Biological Transition 11 6 2 4 3 9 5 10 8 7 1
Catches of salmon in surveys June-Chinook Catches 11 2 3 9 6 8 10 12 7 5 1 4 Sept-Coho Catches 9 2 1 4 3 5 10 11 7 8 6
Mean of Ranks of Environmental D 9.7 3.6 3.4 4.7 4.5 8.8 8.2 9.9 7.1 5.2 2.5 RANK of the mean rank 10 3 2 5 4 9 8 11 7 6 1
Other work completed included:
• Produced Quarterly Updates on “Climatic and Ecological Conditions in the California Current LME” in fall, winter and spring (October-December 2008; Jan-March 2009 and April-June 2009). Reports posted to the PaCOOS website
• Produced annual update of ocean conditions off Oregon including some of our indicators for the annual “State of the Ocean” report prepared by colleagues with the Canadian Department of Fisheries and Oceans. Feb 2009.
• Will soon submit our spring transition dates to a University of Washington web-page that lists dates from various investigators: http://www.cbr.washington.edu/data/trans.html
Problems along the way:
We have pretty much done everything that we set out to do except for writing up a formal “Integrated Ecosystem Assessment”. We have done an assessment of ocean conditions and the returns of coho and spring Chinook salmon to Oregon’s rivers, but we did not yet bring in an assessment of other living marine resources. We plan to post an integrated assessment which includes not only an evaluation of “ocean conditions” for a given year (as we do now), but also sockeye salmon and the fall Chinook salmon that return to both the Columbia River and the Sacramento River; we will include time series and catch data for forage fish (smelts, anchovy and herring) from our rope trawl surveys, and catch data for pelagic predators (hake and mackerels) that are landed from our rope trawl surveys. Ultimately we will bring in all resources that are assessed in the northern California Current (sablefish, hake, some of the rockfish) but we are finding that one person cannot do all of this work – that is, run the cruises, update the
90 indicators, and get together all of the fisheries data. To be successful and useful, the production of an integrated ecosystem assessment should be more of a team effort.
Our extensive krill data sets continue to baffle us and none of our attempts to develop indicators of krill biomass have worked out. The problem is that krill are extraordinarily patchy thus it is difficult to come up with reliable estimates of their biomass. We do want to include some aspect of krill ecology in our ecosystem assessments but have not yet determine what aspects of krill ecology best characterize their role in the food chain. This is a task that will be explored this year by our FATE post-doc.
We have also learned that there is a need to establish a server here at Newport upon which we can provide near real-time updates of the data and indices that characterize present ocean conditions and that forecast salmon returns. Our present way of posting updated information is extraordinarily clunky – all written text must be first reviewed by a scientific editor (in Seattle) who in turn does the mock-up of the web-page. This is then sent to an IT committee (in Seattle) who then review content; they then approve the material for posting on the web and pass the information to another person who does the posting. It can easily take a full month to accomplish this task. We are now trying to find funds to purchase a server and depending on the willingness of our Center to trust that we will not post anything sensitive, we will either operate the server through our NOAA lab here in Newport, or we will go “off-line” and operate the server through Oregon State University.
Applications
1. Contribution to the forecasting of salmon returns to the Columbia River. Indices are used by the Columbia River TAC (the folks that make management decisions). Our indices are also incorporated into modeling carried by staff at CRITFC (Columbia River Inter-Tribal Fish Commission).
2. Our indices and forecasts are also used by local watershed councils who use our ocean indicators to assess the success or failure of their habitat restoration projects. For this they need to determine the relative degree to which returns of fish to rivers were due to ocean conditions vs. actions taken to improve freshwater habitats.
3. The current state of “ocean conditions” on our web-page is also monitored by the general public and members of the press. We frequently get calls from both the Northwest and Southwest Fisheries Regional Offices to field calls from the press. Reports from the Associated Press, the Portland Oregonian and the Seattle Times call most frequently for articles that are prepared on ocean conditions with a prognosis for the future. A front page story appeared in the Oregonian that included, among other things, a color photo of the copepod, Neocalanus cristatus – surely this is the first time that an image of a copepod has appeared on the front page of a major newspaper.
4 W. Peterson also appeared in several media outreach affairs – lengthy interview by NOAA for a NOAA-Climate Change video, interviews on local NPR stations and an interview on ocean
91 conditions, climate change and salmon. A 20 minute segment will appear in early September on the Oregon Public Broadcasting radio network.
Publications
McClatchie, S. and 23 others including W.T.Peterson. 2008. The State of the California Current, 2007-2008. La Niña conditions and their effects on the ecosystem. CalCOFI Rep. 49: 39-76
Schwing, F.B., W.T. Peterson, N. Cyr, and K.E. Osgood. 2009. Future research requirements for understanding the effects of climate variability on fisheries for their management. In: The Future of Fisheries Science in North America, (R.J. Beamish and B.J. Rothschild, eds.), Fish & Fisheries Series, Springer Science, pp. 619-634.
Peterson, W. Copepod species richness as an indicator of long term changes in the coastal ecosystem of the northern California Current. CalCOFI Reports 50. Accepted.
Auth, T.D., R.D.Brodeur, H.L.Soulen, L. Cianelli and W.T.Peterson. An investigation of the response of fish larvae to decadal changes in environmental forcing factors off the Oregon coast. Fisheries Oceanography (submitted)
Keister, J.E., C.A.Morgan, N. Mariani, V. Combes, E. DiLorenzo and W.T.Peterson. Copepod species composition linked to ocean transport in the northern California Current. Geophys. Res. Lett. (submitted)
Presentations
August 2008 Newport 5-6th grade kids from a Robotics Club in Portland “Climate is what you expect; weather is what you get!”
August 2008 Santa Cruz Central Valley Salmon meeting “Ecological Indicators of Ecosystem Productivity in the Northern California Current: 1996-2008 -or- Where have all the salmon gone?”
August 2008 San Diego FATE presentation at annual science meeting
August 2008 Newport Presentation at a media workshop organized by Oregon State University: “Climate, ocean conditions and salmon”
Sept 2008 Klamath Falls Oregon Watershed Enhancement Board “Climate, ocean conditions and salmon”
Sept 2008 Sun Valley Pacific States Marine Fish Commission “Recent high-frequency variability in the PDO and ocean conditions in the northern California Current: forecasting impacts on ecosystem structure and salmon survival”
September 2008 S. Lake Tahoe Eastern Pacific Ocean Conference (EPOC)
92 “Recent high-frequency variability in the PDO and ocean conditions in the northern California Current: forecasting impacts on ecosystem structure and salmon survival”
October 2008 Salem Oregon Fish and Game Commission “Climate, ocean conditions and salmon”
October 2008 Lake Chelan Pacific Coast Shellfish Growers Association “Hypoxia in the coastal upwelling zone off Oregon
October 2008 Dalian, China PICES Meeting “Forecasting returns of coho and Chinook salmon in the n. California Current: A role for high-frequency long term observations”
November 2008 Newport Briefing over the internet to the West Coast Governor’s Staff on Hypoxia off the Oregon and Washington Coasts
November 2008 Astoria Internal NOAA Fish Ecology Division Team Meeting “What I do…with salmon forecasting”
November 2008 Corvallis Workshop at Oceanography Department, OSU “Another unusual year in the ocean…”
December 2008 Newport Hatfield Marine Science Center, Seminar “The impact of greatly improved ocean conditions in 2008 on salmon and other marine organisms”
December 2008 Eugene University of Oregon – Invited Seminar “Climate variability and change in the northern California Current: impacts on ocean conditions, zooplankton and salmon populations”
December 2008 San Diego CalFOCI Conference “The N. California Current in the 21st century: recent high frequency variability in hydrography, and copepod community structure”
December 2008 Santa Barbara National Center for Ecosystem Analysis and Synthesis. “Climate change, ocean conditions, and marine food webs”
January 2009 Portland Benson High School – Ocean Science Bowl training session. “Biological oceanography: upwelling, primary productivity, ecology”
January 2009 San Francisco Climate change, natural resources and coastal management “Climate change, ocean conditions, and marine food webs”
February 2009 Seal Rock State of Oregon, Department of Transportation, Environmental Managers, “Climate change, ocean conditions, and marine food webs”
93 February 2009 Newport Scientists and Fishermen’s Exchange; quarterly meeting of fishermen and scientists; “A biological index for the spring transition”
March 2009 Salem Briefing for State Legistlators, Staff for climate change issues, “Climate change, marine food webs and salmon in the northern California Current”
April 2009, Corvallis Oregon State University; Seminar on “Zooplankton in Eastern Boundary Current Upwelling Systems”
April 2009 Shelton American Fisheries Society, Northwest Section, “Climate change, marine food webs and salmon in the northern California Current”
April 2009 Cannon Beach Project WILD International Coordinators Conference (sponsored by the Council on Environmental Education) “Climate change, marine food webs and survival of juvenile salmon during their first summer at sea in the Northern California Current”
May 2009 Wash DC NOAA/Silver Spring Seminar on “Climate change, ocean conditions, marine food webs and salmon forecasting”
May 2009 Seattle Briefing for Jane Lubchenco on “Climate Change, Marine Food Webs, Hypoxia and Salmon”
June 2009 Olympia Meeting with Columbia River Salmon - Technical Advisory Committee (the folks that make management decisions on catches of Columbia River salmon, “Ocean Indicators: Development and Application”
July 2009 Reedsport Umpqua River Watershed Council, “Climate change, marine food webs and salmon in the northern California Current”
Webpages
1. Forecasting web-page http://www.nwfsc.noaa.gov and click on Ocean Index Tools
2. Quarterly updates of “Climatic and Ecological Conditions…” http://www.pacoos.org for
3. Canadian “State of the Ocean” http://www.dfo-mpo.gc.ca/CSAS/Csas/Publications/SAR- AS/2009/2009_030_e.pdf
4. Dates of spring and fall transition http://www.cbr.washington.edu/data/trans.html
94 08-06 Project Title: Developing Statistically Robust IPCC Climate Model Products for Estuarine- dependent and Anadromous Fish Stock Assessments
Principal Investigators: Frank Schwing, Eric Danner, Steve Lindley, Roy Mendelssohn, Jim Overland, Steve Ralston, Bruno Sanso, Muyin Wang
Goals: To develop objective methods for combining the outputs of the AR4 models at the regional scale for selected output variables to construct (probabilistic) indicators of future climate change, and to use these indicators in existing stock assessment models to project abundances for the next several decades
Approach: The project has two principal tasks: (1) develop a set of rules for objectively evaluating and averaging climate model projections, and use them to develop probabilistic estimates of the future state of the CCLME; and (2) produce leading indicators of fishery-relevant climate parameters, and incorporate these indicators into selected stock assessment models to project future abundance tendencies under realistic climate scenarios.
Work Completed: Climate model output produced as part of the IPCC AR4 provide a basis to produce (probabilistic) indicator series that can be used in ecosystem models to examine the possible effects of climate change on marine related ecosystems. There are some 15-20 IPCC models that produce relevant output over grids of differing resolution, and the problem is how to combine the information from the different models in a probabilistic fashion, to reduce the dimensionality of the results and provide indices that are usable in ecosystem models.
State-space, hierarchical Bayesian and EOF analyses were performed on IPCC AR4 model output from 20th century runs in order to find the “best’ models or combination of models to produce {probabilistic) forecasts. These results will then be used with the results from model runs of various future climate scenarios to produce indices that can be used to project the ecosystem consequences of those climate scenarios, in particular with regard to salmon.
For a first pass at the data, the model output were converted to a resolution of 5◦ over the coordinates 22.5◦ - 62.5◦ North and 112.5◦ − 247.5◦ East, consisted with grids used in previous analyses of SST over the North Pacific ocean. Monthly decadal averages were then calculated over the 100-year period from 1900-2000 for both the model output as well as for the observed data.
The first approach to analyzing the data was to fit univariate state-space models to the model output, and then calculate common underlying structure in the results. A univariate state-space decomposition separates the observed series into a nonparametric (in time) trend term, a non- stationary seasonal term, a stochastic but stationary cycle and an error term. Subspace identification methods are then used to calculate lower-dimension “common trends”, ‘common seasonals”, and “common cycles”.
95
“Common trends” were calculated for each model and for the observed data, and then those results were combined into a joint common trend analysis (Figure 1).
Figure 1 - The first common trend from each model (colored lines), from the data (dashed black lines), and the common trend (solid black line).
It can be seen that the first common trend of some of the models are similar to that from the observed SST, but many are not. Moreover, when analyzed this way, since no extra weight is given to the observed data, all results are treated equally and the models that misfit the data pull the joint common trend away from the data common trend.
A way to remedy this is to first project the model results onto the observed data common trend spatial pattern, and then use results from subspace identification methods to calculate the system matrices of the state-space model (Figure 2). This reproduces the observed data first common trend. However, only a few of the models significantly weight in the reconstruction, and some of those weight negatively. This makes it problematic using these weightings in future projections. Moreover, there is no guarantee that runs from the future climate scenarios will have the same results. We are in the midst of analyzing the runs from several of the future projections, as well at looking at sparse versions of the subspace methods that will force some of the weights to zero.
96
Figure 2 - Reconstruction of the data common trend (bottom) from projecting the model common trends onto the data common trends. The top graph shows the relative weight given each model in the projection.
A second approach starts with calculating the EOF’s of the model output anomalies as well as for the observed data. As with the common trends, this produced inconsistent results. Instead the model output where projected onto the observed data first EOF (Figure 3). Then a hierarchical Bayesian model was fit to these results. This model assumes there is a common trend in the observed data and the projected model output, “local” time-dependent discrepancy in each model, and model specific error. The same can be done for output from future projections and produce forecasts into the future with probability distributions. This procedure produces reasonable results (Figures 4-5) and provides a basis for developing useful probabilistic indices for ecosystem model input.
Figure 3 – First two observational data EOFs.
97
Figure 4 - Common trends of the projected model anomalies EOFs and forecast into the future.
Figure 5 - Posterior mode forecast for selected time periods from the common trend model for the projected EOFs.
Applications: The results of the analyses of the climate models will be applied to stock assessment models as part of the Year 2 effort of this project.
Publications/Presentations/Webpages: "Joint Projections of North Pacific Sea Surface Temperature from Different Global Climate Models" by Francisco M Beltran, Ricardo T. Lemos, Bruno Sanso, and Roy Mendelssohn –
98 Presented at Graduate Session at UCSC.
"Joint Projections of North Pacific Sea Surface Temperature from Different Global Climate Models" by Francisco M Beltran, Ricardo T. Lemos, Bruno Sanso, and Roy Mendelssohn – Presented at “Applying IPCC-class Models of Global Warming to Fisheries Prediction” at GFDL in Princeton, NJ.
99 08-07 Project Title: Integrating Indicators: Co-variation, Periodicity and Amplitude of Key Biological Signals from Central Oregon and Northern California
Principal Investigators: William Sydeman, Alec MacCall, Kyra Mills (Farallon Inst. Advanced Ecosystem Research [FI]), Frank Schwing & Steven Bograd (SWFSC/ERD), and Chet Grosch (Old Dominion University), Julie Thayer (FI)
Goals: (1) to develop interpretable multivariate ecological indicators that reflect the complex bio-physical interactions which drive variability in key fish and wildlife populations for inclusion in developing CCLME IEA efforts, and (2) assess the co- variation in periodicities and amplitude (i.e., cross-species “signals”) in key time series datasets.
Approach: We will integrate key Oregon and northern California biological time series using EOF/CEOF, Spectral, and Wavelet analysis (as possible).
Work Completed: Due to delays in funding transaction, we are a bit behind on this project. To date, we have examined spectra of time series of copepod abundance and diversity, coho salmon survival, seabird reproductive success, and timing, and upwelling indices at 36o, 39o, and 42oN. Remarkably, most series show strong variability at the 12-14 year period. The complexity of this analysis is substantial. To obtain appropriate spectra, the annual signal must be removed, and then the series is de-trended. We have also integrated values of the NPI, NOI, SOI, MEI, and PDO with these biological measurements.
Applications: Analyses will be included in next versions of the California Current Integrated Ecosystem Assessment (IEA). The first version, Module 1, on select indicators is available (see below, Sydeman and Elliott 2008). The second version, Module 2, on trends and variability of select indicators, is under internal NOAA-NMFS and Farallon Inst. review (Sydeman and Thompson 2009).
Publications/Presentations/Webpages: Sydeman and Elliott (2008) is available online (www.faralloninstitute.org). Sydeman and Thompson (2009) should be available online within 1-2 months (by 30 September 2009). Bograd and Sydeman (in prep.) have contributed information to the PICES North Pacific Ecosystem Status Report.
100 FATE projects initiated in FY 2009
09-01 Salmon growth as an indicator of ecosystem variability - Brian Beckman (NWFSC), Joe Orsi (AKFSC)
09-02 Developing recruitment indices for West Coast rockfish using individual-based models and environmentally forced ocean circulation models - Eric Bjorkstedt, Steve Ralston, John Field, and Xi He (SWFSC), Robert VanKirk (Humboldt State Univ.)
09-03 Beyond the spring transition: winter pre-conditioning of ecosystem dynamics and implications for sentinel species and fisheries - Bryan Black (OSU), Isaac Schroeder and Steven Bograd (SWFSC), William Sydeman (Farallon Institute), Vladlena Gertseva and Peter Lawson (NWFSC)
09-04 Future conditions in the Bering Sea: applications to walleye pollock and flatfish - Nicholas Bond (UW - JISAO), Anne Hollowed and Tom Wilderbuer (AFSC), Jim Overland (PMEL)
09-05 Fish and crab larvae as indicators of climate change and variability in recruitment potential of juvenile salmon - Richard Brodeur and Bill Peterson (NWFSC), Toby Auth and Elizabeth Daly (CIMRS/OSU)
09-06 An integrated assessment of the physical forcing and biological response of the pelagic ecosystem of the eastern Bering Sea, focused on walleye pollock and Pacific cod - Lisa Eisner and Edward Farley (AFSC)
09-07 Relating environmental conditions to recruitment and growth of the Atlantic sea scallop (Placopecten magellanicus) - Kevin Friedland and Deborah Hart (NEFSC)
09-08 Ecosystem indicators for the North Pacific Subtropical Gyre marine ecosystem - Evan Howell and Jeffrey Polovina (PIFSC)
09-09 An evaluation of the significance of climate forcing on time-varying growth and fecundity of rockfish in the California Current - Susan Sogard and John Field (SWFSC), Chris Harvey (NWFSC)
09-10 Integrating ocean climate indices into Pacific Salmon stock assessments - Thomas Wainwright, Peter Lawson and Bill Peterson (NWFSC), Thomas Helser (NWRO)
09-11 Evaluating the influence of the fall phytoplankton bloom on the energy available for growth and reproduction, and subsequent recruitment of Georges Bank haddock - Mark Wuenschel, Kevin Friedland, Elizabeth Brooks, Sandy Sutherland and Richard McBride (NEFSC)
101 Summary of FATE Products
The following table represents an inventory of the products generated by the FATE program to date. The products listed demonstrate the great range of materials being developed with FATE support and document the use of many of the products by stock assessment scientists and living marine resource managers. FATE contributions support many of the goals of ecosystem approaches to management including building sustainable fisheries (e.g. improving estimates of recruitment, growth, and catchability), reducing bycatch, maintaining and recovering protected species, protecting essential fish habitat and developing ecosystem assessments.
102
FATE Products through 2009
PIFSC FATE Indices and products Status of Product/Index product Description of product Relation to stock assessment/LMR management SSH/EOF-based regional These regional products indicate ENSO/PDO-scale The central region indicator has been used to assess and indicators for North Pacific #1: physical changes in the upper layers of the water monitor temporal changes in the physical environment for Central Maintained column. research and management recommendations SSH/EOF-based regional These regional products indicate ENSO/PDO-scale The equatorial region indicator has been tied to bigeye indicators for North Pacific #2: physical changes in the upper layers of the water CPUE in southern region of the fishery. This indicator has Equatorial Maintained column. been used to predict bigeye catch rates in this region North Pacific Transition Zone This index provides the mean latitudinal position of the This indicator was the basis for the Baker et al. 2007 Chlorophyll Front winter Transition Zone Chlorophyll Front during January- juvenile monk seal survival model. This indicator is latitudinal position Maintained February in the central North Pacific currently used by managers to predict juvenile MS survival This SST and Altimetry-based product describes the dynamic region where higher loggerhead turtle This product is delivered daily to members of the Western interactions with the longline fishery are expected to Pacific Regional Fishery Management Council (WPRFMC) TurtleWatch Maintained occur. and the longline fishing industry. The time series of the monthly mean area (km2) with surface chlorophyll less than or equal to 0.07mg chl/m3 between 5° – 45° N/S latitude was developed to monitor This index is currently used in monitoring and North Pacific least productive the observed expansion of the least productive area of understanding recent ecosystem-based changes in the waters index Maintained the North Pacific. North Pacific. This study looked at time series analysis results for the 13 most abundant species caught in the deep-set Hawaii-based longline fishery over the past decade. The derived index from this study was a time series of the Trophic level percentage catch percentage of the catch of the 13 species with trophic time series Maintained level greater than or equal 4.0 This indicator is the mean altimetry along the Oregon coast to the 1000 m bathymetric contour. This was developed with Michael Schirripa of the NWFSC to Oregon coast sea level height attempt to update the tide gauge SSH index currently This product is still maintained but incorporation has been index Maintained used in the yearly Sablefish assessments. put on hold as Dr. Schirripa has relocated to the SEFSC. Research on spatiotemporal varibility of bigeye tuna dive behavior indicates that the "potential vulnerability", or amount of time that bigeye spend at depths where This relationship has been used in trying to explain Spatiotemporal variability in deep longline sets target them, is correlated with surface fluctuations in bigeye CPUE around Hawaii. This work has bigeye tuna dive behavior Utilized temperatures. been presented to the WPRFMC. The results indicated that incorporation of the dive data did not explain more bycatch than the surface-based model. Results from the research on the dive variability of Further work with this data will be incorporated with recent Dive behavior of juvenile juvenile loggerhead turtles were incorporated to attempt in-situ oceanographic data to attempt to re-model the loggerhead turtles Utilized to refine the TurtleWatch product. TurtleWatch bycatch zone. Altimetry-based eddy and Various ecosystem indices including an eddy index, a Not Currenty used, these indices were developed as part of chlorophyll-based regional coastal chlorophyll index and oceanic chlorophyll bloom a larger project developing several types of ecosystem indices for the North Pacific Published index. indicators for the North Pacific
SWFSC FATE Indices and products Status of Relation to stock assessment/LMR management/IEA Product/Index product Description of product development The technique of "crossdating" was applied to the otoliths of long-lived rockfishes. Using the dating Long-term indices of annual Published technique, connections between climate and species Importance of wintertime ocean conditions found to be growth in long lived groundfish papers sensitivity were made. critical for rockfish. Information conveyed to management. Examined the correlation between spawning location This work suggests that remote sensing could be used to (CUFES data) of sardine and anchovy within the determine spawning area for these two managed species. Hake larval abundance and Basis of current CALCOFI region and compared with remotely sensed Sardine assessment will use these relatinships in future spawning center FATE-FTE work data. assessments. Analyzed data on market squid paralervae in Aided in development of age-temperature model of squid conjunction with temperature data from CALCOFI growth. *note: squid is not a currently managed species, yet Market squid paralarve indicators surveys to connect squid population dyanmics to El nino it is the largest commercial landing for california waters, of El Nino response published cycles. and as such, may be managed in future Both of these contributions have been useful to management by shedding light on the relevant spatial scale of management units. In particular, the Field paper showed that large scale (500-1000 km) physical forcing mechanisms are responsible for recruitment variability in Scale of physical processes Showed that arge scale (500-1000 km) physical forcing commercially important rockfish stocks. The Petersen governing groundfish mechanisms are responsible for recruitment variability in paper showed that average dispersal distances during the reproductive success published commercially important rockfish stocks. larval phase of commercially important rockfish stocks is
104 ~500 km, which is quite influential in the design of marine reserve networks.
Accessed by large community, federal, academic, PFEL webpage/data server Maintained Serves FATE-relevant data management etc. Salmonid growth in California as leading index to the impacts of Quantified relationships between large-scale ocean climate change on fishery environmental indices and salmon growth along their resource productivity published north Pacific range. Used to defend regional management and study of salmon Quantifying the effect of ocean environment on the growth rates and age and length at maturity of Developed mechanistic models for quantifying effects of Used to defend reigonal management and offers predictive California Chinook salmon maintained/ local environment on Chinook salmon from CA, AK, and relationships that were instrumental in determining the population published WA. factors causing the recent CA salmon collpase.
Bbiologically-relevant indices of Developed a set of indices quantifying thermocline water column structure in the depth and magnitude of stratification. Described California current Maintained interannual variability in water-column indices. Indices have been used in California Current IEA.
Developed an index of ecosystem state that elucidated Used to defend reigonal management and offers predictive Local community production the mechanisms of environmental forcing on local relationships that were instrumental in determining the index maintained community production, with relevance to top predators. factors causing the recent CV salmon collpase. Database and toolset for summarizing freshwater and climate variables by waterwshed for all CA watersheds Anticipated use in CC IEA. Freshwater environmental Climate and Freshwater impact draining to the Pacific, with regards to anadromous variation has now been included in the starry flounder stock assessment tools maintained species. assessment. Developed a set of indices that quantify the timing, duration and intensity of coastal upwelling along the West Coast. Characterized long-term variability in these properties. Correlated these indices with a number of Upwelling phenology indices maintained biological series. Indices have been used in California Current IEA. Data analysis package for use with CALCOFI data, Rcalcofi maintained using the R statistical language publicly available
105
NWFSC FATE Indices and products Status of Relation to stock assessment/LMR management/IEA Product/Index product Description of product development
This index is based on bi-weekly sampling along the Newport Hydrographic Line, at station NH 05, five miles from shore. The index is the anomaly of the biomass of three “northern” copepod species: Calanus marshallae, Pseudocalanus mimus and Acartia longiremis. The This index is one of 15 indices used to form a composite to index tracks changes in food chain structure in coastal forecast marine survival of salmon in the northern California waters of the northern California Current; it is well Current and their return to freshwater. The indices and Northern Copepod Biomass maintained, correlated with returns of coho and fall Chinook salmon forecasts are available on the NWFSC's salmon forecasting anomaly updated monthly to rivers and streams of Washington and Oregon. web page.
Based on cluster analysis of copepod enumeration data collected biweekly from station NH 05. Analyses are done twice each year with cluster # 1 representing the This index is one of 15 indices used to form a composite to winter copepod community; and cluster # 2 is the forecast marine survival of salmon in the northern California maintained, summer community. We define the “date of biological Current and their return to freshwater. The indices and updated every spring transition” as that date when the summer forecasts are available on the NWFSC's salmon forecasting Biological spring transition spring copepod community becomes established. web page. This index is one of 15 indices used to form a composite to This is a simple indicator of biodiversity – the number of forecast marine survival of salmon in the northern California copepod species in samples collected biweekly from NH Current and their return to freshwater. The indices and maintained, 05. Species richness is highly correlated with salmon forecasts are available on the NWFSC's salmon forecasting Copepod species richness updated monthly returns. web page. Based on ordination of the copepod enumeration data from station NH 05 – we use the X-axis score of the maintained, ordination as an index of community structure – This index is among the newer indices and is planned for updated every negative score indicate a cold water community; positive incorporation with the 15 present indices used to form a Copepod community structure spring and fall scores a warm water community. composite forecast for northern California Current salmon. The approach developed by Bograd is used to This index is one of 15 indices used to form a composite to determine the date of spring transition (the date when forecast marine survival of salmon in the northern California maintained, the cumulative upwelling index at 45 N starts to become Current and their return to freshwater. The indices and updated every positive). This index produces virtually the same date forecasts are available on the NWFSC's salmon forecasting Physical spring transition index spring. as the Logerwell spring transition index. web page.
106 maintained, Spring transition based on updated Uses tide gauge data from stations in Washington and coastal sea level data. annually Oregon to calculate a sea level height index. Used in stock assessment of sablefish. This index is one of 15 indices used to form a composite to forecast marine survival of salmon in the northern California maintained, Current and their return to freshwater. The indices and updated every forecasts are available on the NWFSC's salmon forecasting Length of the upwelling season autumn We use the Bograd approach (his “LUSI”) for 45 N. web page. maintained, updated in July The CPUE of salmon catches in our May, June and This index is one of 15 indices used to form a composite to for spring September surveys are proving to be useful in forecast marine survival of salmon in the northern California Chinook and forecasting salmon returns one year in advance (for Current and their return to freshwater. The indices and October for coho) and two years in advance (spring Chinook) of forecasts are available on the NWFSC's salmon forecasting CPUE of juvenile salmon Coho their returns to their natal streams to spawn. web page. The index is the mean of the abundance of five maintained, ichthyoplankton taxa collected at stations NH 05 and 10. This is a new index that is planned for incorporation with Ichthyoplankton: abundance of updated Taxa are osmeriids, rockfish, sand lance, cottids and the 15 present indices used to form a composite forecast five dominant species annually anchovies. for northern California Current salmon. Moran’s I is a measure of the connectedness of patches maintained, of coho salmon in our June surveys. This index is highly This index is among the newer indices and is planned for updated correlated with coho salmon returns one year in incorporation with the 15 present indices used to form a Moran’s I annually advance of their return to their natal streams. composite forecast for northern California Current salmon. under This index is based on Bi et al. 2008, Fish Oceanogr. development, 17: 463-476. The index is area of “good habitat” This index is still being evaluated for possible inclusion with updated available to salmon based on ocean color data the 15 present indices used to form a composite forecast Habitat area annually (SeaWiFS). for northern California Current salmon.
AFSC FATE Indices and products Status of Relation to stock assessment/LMR management/IEA Product/Index product Description of product development no longer Regional ocean circulation models (ROMS) used to Indicators were developed for use in predicting year class Gulf of Alaska ocean circulation updated develop multi-decadal ocean circulation indicators. strength of walleye pollock in the Gulf of Alaska. Utilzed ROMS and OSCURS models to develop Predicted cross shelf tranport indices were to be utilized for Eastern Bering Sea cross shelf indicators of cross shelf transport of rock sole larvae in recruitment forecasts for age structured flatfish stock transport unsuccessful the Bering Sea assessments.
107 Ecosystem Considerations report The North Pacific Fishery Management Council and its Plan of the Stock Assessment and maintained, teams utilze the information in the SAFE report in making Fisheries Evaluation (SAFE) updated Compilation of ecosystem indices and status for the recommendations for total allowable catches of Alaskan document annually Bering Sea marine fisheries.
maintained, Index of groundfish stock survival (recruits per spawning updated biomass anomalies) to detect decadal- or regime-scale Index is incorporated into the Bering Sea ecosystem Groundfish stock survival annually changes in stock productivity. assessment. Indices of water column structure and frontal regions to explain temporal patterns of recruitment of walleye Water column structure/frontal under review, pollock and winter spawning flatfish. Water column regions in the eastern Bering updated structure may be related to pollock cannibalism, which Not yet incorporated. If successful, will be incorporated into Sea annually affects recruitment. the Bering Sea ecosystem assessment. published in Indices of size diversity in the Bering Sea, Aleutian Ecosystem Islands, and Gulf of Alaska. These indices provide Considerations measures of overall system state and information on rpt., in progress biomass, diversity, size structure, trophic level and Size diversity for journal pub. production. Is incorporated in the Bering Sea ecosystem assessment. maintained, updated every Index of groundfish condition developed from trawl Index is incorporated into the Bering Sea ecosystem Groundfish condition 2-3 years survey data. assessment. Indices of the spatial variations in water mass characteristics, nutrients, phytoplankton, and forage fish updated (juvenile salmon and young of the year pollock), during Results have been related to the annual vaiability in Water mass characteristics annually the fall in the eastern Bering Sea. juvenile sockeye salmon size and marine survival. Examined interannual differences in abundance, growth and condition of juvenile salmon and age-0 pollock in relation to their spatial distribution, diet, consumption, updated growth efficeincy, food availability and spatial-temporal Salmon and pollock survival annually patterns in environmental conditions.
Developed stock projection models that incorporated the influence of ecosystem variability on commercial fish. Demonstrated how indicators of climate change can be Management strategy evaluation integrated directly ito fisheries management advice. Gulf of Alaska walleye pollock stock projection
108 Developed stock projection models that incorporated the influence of ecosystem variability on commercial fish. Identified the level of accuracy and precision needed for indicators if their use in assessments is likely to lead to Management strategy evaluation improved management performance. Gulf of Alaska walleye pollock management strategy Examined climate models for the Bering Sea and Arctic Provide projections of future environmental conditions of to project future climate conditions and their affect on importance to long-range living marine resource Climate forecasting tools the ecosystems. management.
NEFSC FATE Indices and products Status of Product/Index product Description of product Relation to stock assessment/LMR management Described shifts in distribution of 36 fish stocks over the 2 manuscrips; 1 NE US continental shelf and developed a protocol to Distributional shifts in resource in press, 1 in ensure that changes in distributions are adequately Decision tree proposed for use by stock assessment species review accounted for in stock assessments. scientists. Included in Ecosystem Status Report. Based on mean preferred temperature of species found Community preferred in bottom trawl surveys. Results indicate a shift in temperature index developed biomass to warmer water species. Included in Ecosystem Status Report. Utilize environmental information to modify stock- Environmentally-explicit stock- under recruitment relationshops used by stock assessment recruitment relationships development scientists. Will be considered for future stock assessments. manuscrip in review, operationalized Index presented during herring assessment update, will be for herring, considered for inclusion in next benchmark. Potential for being applied to Based upon a relation between the larval index and other larval indices to be included in other stock Atlantic herring larval index other species spawning stock biomass. assessments.
SEFSC FATE Indices and products Status of Product/Index product Description of product Relation to stock assessment/LMR management Stage-specific indices of abundance for 5 species of For use in improving stock assessments by state, regional Abundance indices developed fishes in the SE. and federal management agencies in the SE U.S.
109 Developed environmentally-corrected index of abundance for post-larval gag grouper. A fishery Improved abundance estimates for post-larval gag grouper Gag grouper index developed independent index of spwning stock biomass. used to improve predictions of stock abundance. Index relates position of Loop Current to abundance and Index helps to standardize CPUE and understand Loop Current index developed CPUE of dolphin fish in the Gulf of Mexico variations in the longline fishery. under Habitat model for larval bluefin tuna in the Gulf of To be incorporated into spawning stock biomass Bluefin habitat model development Mexico assessment of bluefin tuna. under Environmental indices relating to abundance and Indices will be used to improve predictions of pink shrimp Pink shrimp indices development recruitment are being developed. recruitment and abundance.
110