PROJECT OVERVIEW

This Bonneville Power Administration (BPA) funded project was initiated to develop the statistical theory, methods, and statistical software to design and analyze PIT-tag survival studies.  This project developed the initial study designs for the National Marine Fisheries Service (NMFS) /University of Washington (UW) Snake River survival studies of 1993–present.  Since then, survival investigations have been extended to the use of balloon, radio, and acoustic tags.  Survival investigations now include route-specific survival at hydro projects, partitioning project survival into pool and dam components, relating smolt survival to river conditions and dam operations, and the estimation of upriver adult survival.  This project continues to respond to the changing needs of the scientific community in the Pacific Northwest as they face new challenges to extract life-history data from an increasing variety of fish-tagging studies.  The project's mission is to help assure tagging studies are designed and analyzed from the onset to extract the best available information using state-of-the-art statistical methods.  In so doing, investigators can focus on the management implications of their findings and not become distracted by concerns whether the studies design and analyses are correct or not.

To accomplish this project's mission, the project provides quantitative support to the fisheries community from study design through data analysis and interpretation.  Starting from first principles, statistical methods are tailored specifically for the intent of the tagging study objectives.  From the onset, investigators therefore know exactly what information can and cannot be extracted and what model assumptions must be met for a valid study.  By tailoring the statistical models to the specific purpose of the study, maximum precision is also obtained.

The project also develops interactive statistical software to allow investigators to easily and competently analyze tagging data.  Program SURPH was developed to analyze juvenile PIT-tag data; Program USER, to analyze radio- and acoustic-tag studies; Program ROSTER, to analyze joint juvenile and adult PIT-tag detections; and Program PitPro, to easily and correctly extract detection histories from the PTAGIS data base for the analysis of adult and juvenile survival and transportation studies.  Technical support is provided to investigators through instruction, consultation, and assistance in the design and analysis of complex datasets.  Program SampleSize was developed to assist investigators in the design of single-release, paired-release, transport-inriver, and balloon-tag survival studies.  Each year, this project assists investigators in the design and analysis of over 20 major tagging studies conducted by state and federal agencies, public utility districts, and tribal organizations in the Columbia Basin.

Tagging studies within the Columbia Basin have become extensive, with tens of millions of CWT, over a million PIT tags, tens of thousands of radio tags, and thousands of acoustic tags used annually to obtain life-history information on salmonid stocks.  Yet, this BPA-funded project is the only project dedicated to providing statistical support for their design and analysis.  This project helps assure cost-effective tagging studies, valid analysis of the data, and proper interpretation of results to best manage water and salmonid resources.  Because evaluation of mitigation projects and system recovery rely on performance measures from tagging studies, reliable study designs and data analysis are crucial to the Fish and Wildlife Program (FWP).


PROJECT GOAL

Ensure that salmonid tagging studies in the Columbia Basin are conducted with the best available design and analysis, including sample size guidance, state-of-the-art statistical software, and consultation in order to provide cost-effective and precise research, monitoring, and evaluation studies.


PROJECT OBJECTIVES

     1. Provide consistent application of statistical methodologies for survival estimation across all salmon life-cycle stages to assure comparable performance measures and assessment of results through time, to maximize learning and adaptive management opportunities, and to improve and maintain the ability to responsibly evaluate the success of implemented Columbia River FWP salmonid mitigation programs and identify future mitigation options.

     2. Improve analytical capabilities to conduct research on survival processes of wild and hatchery Chinook salmon and steelhead during smolt outmigration, to improve monitoring and evaluation capabilities and assist in-season river management to optimize operational and fish passage strategies to maximize survival.

     3. Provide statistically valid estimates of downriver juvenile survival, ocean survival, upriver adult survival, smolt-to-adult ratios, transport-inriver ratios, delayed mortality (i.e., D), transport fractions, age-at-return compositions, and associated standard errors.  Other parameters include adult fallback rates at hydro projects and adult straying rates that are important in understanding the effects of the hydro system on salmonid recovery.

     4. Develop statistical methods for survival estimation for all potential users and make this information available through peer-reviewed publications, statistical software, and technology transfers to organizations such as NMFS, the Fish Passage Center, US Fish and Wildlife Service (USFWS), US Geological Survey (USGS), US Army Corps of Engineers (USACE), First Nations, Public Utility Districts (PUDs), the Independent Scientific Advisory Board (ISAB), and other members of the Northwest fisheries community.

     5. Provide and maintain statistical software for tag analysis and provide user support.

     6. Provide improvements in statistical theory and software as requested by user groups.

These improvements include extending software capabilities to address new research issues, adapting tagging techniques to new study designs, and extending the analysis capabilities to new technologies such as radio-tags and acoustic-tags.


PROJECT ACCOMPLISHMENTS

Integration, Coordination, and Information Transfer

This project is integrated and coordinated with Columbia River fisheries agencies and Tribes, the Northwest Power Planning Council, and other scientific expert panels investigating survival processes.  The transfer and coordination of information and products produced by this project has occurred throughout the history of the project.  Information transfer is accomplished by printed technical reports, scientific publications, public presentations, electronic internet-based products, quarterly newsletter (Salmon Insider), and one-on-one statistical consulting.  During this last year, this project provided technology transfer to a variety of organizations, including:  (a) NMFS, (b) USGS laboratory at Cook, Washington, (c) the ACOE, (d) the Nez Perce Indian Nation,  (e)  Grant County PUD, (f) Chelan County PUD, (g) Douglas County PUD, (h) Idaho Fish and Game (IDFG), (i) Washington Department of Fish and Wildlife (WDFW), and (j) Battelle Pacific National Laboratory.  


Biological/Ecological Results to Date

The project spearheaded the survival studies currently being conducted on the Snake/Columbia Rivers (1993–2007).  Monte Carlo investigations identified sensitivities of PIT-tag studies to violations of some assumptions and robustness to other assumptions paving the way for field trials.  In conjunction with NMFS, this study has helped to generate new biological understandings of the dynamics of smolt outmigration.  These findings include:

       1. Information on smolt travel time – survival relationships.
       2. Survival rates for subyearling and yearling Chinook salmon and steelhead smolts throughout the Columbia Basin.
       3. Information on river flow/temperature – survival relationships.
       4. Comparisons of hatchery and wild Chinook salmon and steelhead smolt survival.
       5. Comparison of smolt survival across 12 years and 5–8 river reaches.
       6. Baseline survival data for comparison with potential mitigation practices in years to come.
       7. Information on fish guidance efficiency (FGE) and spill effectiveness (SE) relationships at Snake and Columbia River dams.
       8. Season-wide estimates of smolt survival.
       9. Route-specific survival and passage probabilities at hydroelectric projects.
     10. Comparison of turbine passage survival and turbine operating efficiency.
     11. Information on transportation effects (i.e. transport/inriver ratios), and adult returns (SARs), and delayed mortality (D).
     12. Detection rates of juvenile and adult PIT-tag facilities.
     13. Adult fall-back rates.
     14. Estimates of Bonneville-to-Bonneville survival.
     15. Comparison of transported and nontransported fish.  

These data have influenced the choices of current mitigation practices and strategies for the future.

In 1998, PIT-tag survival studies were extended to the Mid-Columbia where technical developments from this project were used to design and analyze studies conducted by NMFS and PUDs.  In 1999–2006, smolt survival methods were extended to the use of paired release-recapture studies using PIT-tagged, radio-tagged, and acoustic-tagged releases to estimate survivals at five Mid-Columbia projects.  This BPA program was the inspiration for those investigations based on the 1998 BPA technical report by Skalski et al. (1998).  The findings from these studies include:

     1. Information on river flow/temperature – survival relationships within season.
     2. Comparison of survival rates of hatchery and run-of-the-river juvenile steelhead and yearling Chinook salmon.
     3. Partitioning of project survival into dam and pool components.
     4. Comparison of the outmigration dynamics of PIT-tagged, radio-tagged, and acoustic-tagged steelhead and Chinook salmon.
     5. Estimates of route-specific survival through spillways, sluiceways, surface collectors, and powerhouses.
     6. Assessment of smolt survival with Habitat Conservation Plan (HCP) standards.
     7. Technical information on how to best implement single-release, paired-release, and multi-release study designs to estimate smolt survival using radio-tag and acoustic-tag studies.

In 2001, this project started to provide statistical support to the smolt survival studies conducted by USGS for USACE at Lower Columbia River dams.  This project provided the statistical models and software to estimate dam passage and route-specific survival estimates.  In 2004, this project worked with USGS to implement a triple-release design at The Dalles Dam to improve estimates of dam passage survival.  This project also assisted USGS in sample size calculations and study design for the 2004 summer spill study at Bonneville Dam.  The findings include:

     1. Estimates of route-specific survival through spillways, sluiceways, and powerhouses.
     2. Comparison of spill and powerhouse passage survival under different operating conditions.
     3. Information on lower river flow/temperature-survival relationships within season.
     4. Baseline survival data for comparison with potential mitigation practices in the future.

Recent efforts in 2005 on behalf of NMFS, USACE, the Mid-Columbia PUDs, and First Nations fisheries include new applications of radio tags and acoustic tags for the purpose of estimating juvenile passage survival.  The project is also assisting NMFS, ACOE, and the Mid-Columbia PUDs with the evaluation of its adult PIT-tag detection facilities at adult ladders throughout the Columbia Basin.  

In 2006, this project began a comprehensive effort of estimating ocean and upriver survival of returning adult salmonids using the software program ROSTER.  Program ROSTER software analyzes juvenile and adult PIT-data and also extracts estimates of SARs, transport/inriver ratios, and tests for continued effects of smolt transportation on adult upriver migration success.  In 2007, an updated version of Program ROSTER 2.0 was publicly released and a report detailing the tagging results was submitted for public review.


Nonbiological Results to Date

This project almost single-handedly has infused statistical defensibility and reliability into fish tagging studies in the Columbia Basin.  This project did the prerequisite statistical research to launch the first PIT-tag juvenile tagging studies in the Basin in collaboration with NMFS.  The statistical models and software for juvenile PIT-tag survival studies have made those investigations the "gold standard" of survival studies.  Subsequently, the project has developed the statistical models and software to conduct balloon-tag, radio-tag, and acoustic-tag smolt survival investigations used throughout the Basin.  More recently, models to analyze the complex adult PIT-tag (i.e., Program ROSTER) and radio-tag data have been developed.  Throughout this entire process, project personnel have worked with investigators to transfer this technology to the field staff responsible for collecting much of the monitoring and evaluation data collected in the Basin.

These ongoing efforts have produced statistical software programs for the analysis of salmonid survival studies (e.g., SURPH 2.2b, USER 3.1, TurbSurv, ROSTER 1.3a) that are accessible to the Columbia Basin fisheries community via internet http://www.cbr.washington.edu/analysis.html).  Additional products include comprehensive user’s manuals for SURPH 2.2 and USER 3.1, software for sample size calculations (i.e., SampleSize 1.3), and software to convert PTAGIS databases to databases ready for statistical survival and transportation effects estimation (i.e., PitPro).  In 2005, Program ROSTER 1.0 was made publicly available to analyze joint juvenile and adult PIT-tag data to estimate transport-inriver ratios, SARs,  and delayed mortality (D).  In 2007, an updated version of Program ROSTER 2.0 was released as well as an updated version of PitPro (4.0) that accounts for largely unpublished anomalies associated with the transport program (e.g., transport bypass events).

Specific accomplishments by year follow:

1989:  Developed theory to assess survival effects that result from ambient river conditions.

1990:  Began software development for statistical analyses, and began statistical theory to assess individual covariate effects on survival.

1991:  Demonstrated ability to simultaneously assess ambient effects and individual covariate effects.  

1992:  Extended computer software to include analysis of group and individual covariate effects.  Proposed "strawman" design for development of PIT-tag facilities on Snake/Columbia River.  Developed study plan for a Snake River survival study evaluation.

1993:  Completed statistical software development of analysis package--final debugging of computer program, helped facilitate Snake River survival study, and conducted analysis of hatchery survival studies.

1994:  Completed SURPH statistical software and dissemination of a user's manual for statistical analysis of data.

1995:  Produced a PC version of SURPH software and sample size program to design tag-release studies.

1996:  Developed statistical methods for estimating season-wide survival.  Investigated nonparametric methods for improved confidence intervals.  Developed statistical models to estimate fall Chinook salmon smolt survival and residualization probabilities.

1997:  Developed shareware on the internet to convert PTAGIS databases to formats suitable for survival estimation.  Also developed batch routines to conduct large, repetitive data analyses of survival studies.

1998:  Demonstrated the feasibility of estimating smolt survival and passage probabilities using radiotelemetry data.  Developed longitudinal statistical analyses to analyze time-varying covariates in smolt survival studies.

1999:  Develop a user-specified model development tool that permits investigators to create likelihood models tailored to their evolving research needs.   The model development tool will be extremely valuable in implementing radiotelemetry survival studies where fixed detection locations no longer exist but where antenna arrays can be placed in an unending number of configurations.

2000:  Released upgraded version of Program SURPH.2.  The new program has improved modeling capabilities along with the automatic ability to test model assumptions and produce more reliable confidence interval estimates.  The new version is particularly suited to the paired release-recapture studies ongoing in the Mid-Columbia and Lower Columbia reaches.

2001:  The user's manual for Program SURPH.2 was completed.  In addition, a User Specified Estimation Route Program (USER.1) was developed that provides an extremely flexible and user-friendly environment to create statistical models for the analysis of both juvenile and adult salmonid survival studies.  Program USER.1 is extremely valuable in the design and analysis of the wide variety of radio-tag and acoustic-tag applications currently planned in the Columbia Basin.  The project also assisted NMFS and PSMFC in the evaluation of adult ladder PIT-tag detection facilities at Bonneville Dam.  The project developed statistical models for estimating detection probabilities at that site that will have wider application as additional adult facilities are installed elsewhere in the river.

2002:  The software program User Specified Estimation Route Program was expanded to USER.2.  The new program computes profile likelihood confidence intervals for complex functions of parameters.  This feature is particularly useful in estimating joint survival processes through hydro projects or estimating pool and dam passage survival.  New sample size programs were developed to design single-release, paired-release, and transport-inriver survival studies (i.e., Program SampleSize 1.0).  Program PitPro 1.0 was developed to improve capabilities to construct data files for survival analysis from PTAGIS.  New capabilities include extracting capture histories for both juvenile and adult survival studies.

2003:  Major efforts this year included the development of paper and electronic user manuals for all the major software packages (i.e., SURPH 2.1, USER 2.1, PitPro 1.0, and SampleSize 1.0). With more people using the software, error debugging became important as users tested the limits of the software capabilities.  Improved data entry capabilities were added to Program SURPH 2.1.  

2004:  Diagnostic and graphical tools were added to Program USER to improve interpretation of fitted models.  Program SampleSize was expanded to include replicate releases for the estimation of mean survival.  The most significant accomplishment was the development of a new first-generation program called ROSTER (River Ocean Survival and Transportation Estimation Routine) to provide joint juvenile-to-adult PIT-tag survival analyses.

2005:  Alternative statistical models were developed that could analyze the complex adult radiotelemetry data collected by the US Army Corps of Engineers and the University of Idaho.  The models permit estimation of upriver survival, taking into account fallbacks and straying.  These improved models should help identify more of the "unaccountable loss" and provide confidence bounds on the estimates of adult passage.  Program ROSTER will be made available by the end of 2005.  The program will provide estimates of downriver smolt survival, ocean survival, upriver adult survival, smolt-to-adult ratios, transport-inriver effects, and delayed mortality from PIT-tag releases.

2006:  Program ROSTER (River Ocean Survival and Transportation Estimation Routine) 1.3a became operational with numerous performance measures estimated from the model.  Three technical reports were issued; one on the proper deployment of hydrophone arrays at the mouth of the Columbia River to estimate movement and survival of smolts from the estuary to the continental shelf, and the other, on the estimation of transportation on fall Chinook salmon based on the analysis of PIT-tag data; and a third technical review of marking techniques for salmonid fry and release-recapture methods for estimating fry survival.

2007:  Program PitPro 4.0 was released and greatly improves the joint handling of juvenile and adult PIT-tag detections.  The software also accounts for unpublished anomalies associated with the smolt transport program (e.g., transport bypass events) which could bias PIT-tag estimates if not taken into account.  Program ROSTER 2.0 was updated with additional summary statistics on transport effects and delayed mortality.  A "wiki" site was established so that PIT-tag investigators could share information on tag anomalies and better approaches to data analyses.


Program Accomplishments:  1 December 2006 – 30 November 2007

Major milestones accomplished in the last year include the following:

      1. Program ROSTER 2.0 was publicly released.  The new version includes an expanded list of performance measures that can be extracted from the juvenile–adult life-cycle PIT-tag data.  

      2.  Program PitPro 4.0 was publicly released.  The new version allows for the simultaneous analysis of both juvenile and adult records suitable for subsequent ROSTER analysis.  A substantial quality assurance effort was performed and peer-reviewed (Ben Sandford, NMFS) to make sure juvenile transport data were properly handled and accounted for unscheduled transport bypass events.  ACOE transportation records were reviewed at all transport dams and unscheduled events incorporated into the software programming for PitPro.

      3.  A wiki site was established on DART to allow PIT-tag analysts to exchange information on tag analysis issues.  (Wiki is described by Wikipedia as "a collaborative website which can be directly edited by anyone with access to it.")

      4.  Program USER (4.0) was updated with new analysis capabilities including:

          a.  The ability to estimate parameters in the combinatorial expressions in likelihood models.  This now allows analysis of fish abundance models.  
          b.  The ability to more precisely analyze complex joint likelihood models.
          c.  The ability to more readily find the maximum likelihood solution of models.  New minimization/maximization algorithms were incorporated to find solutions using less informed "seeds" from investigators, allowing less technically sophisticated users to run the program.  
          d.  The ability to compare competing models.

       5.  Developed tagging and analysis protocols to examine subyearling fall Chinook salmon survivals, movements, and residualization in mainstem reservoirs.

       6.  Project staff provided assistance to NMFS, ACOE, USGS, USFWS, Mid-Columbia PUDs, First Nations (i.e., Nez Perce), IDFG, and WDFW in the design and analysis of tagging studies.

       7.  Provided BPA with technical assistance by peer-reviewing the Comparative Survival Study's 10-Year Report for Regional Review.

       8.  Attended basin-wide meetings on the future of tagging studies.

       9.  Performed sensitivity analyses for US Fish and Wildlife Service on the bias in T/I ratios caused by undetected passage of fall Chinook salmon subyearlings at Lower Granite Dam.

     10.  Began a statistically rigorous review of the state-of-the-art for methods of estimating adult salmonid escapement to streams.

     11. The following technical reports and peer-reviewed journal articles were prepared during this last year.

           a. Buchanan, R. A., and J. R. Skalski.  2007.  A life-cycle release-recapture model for salmonid PIT-tag investigations.  Journal of Agricultural, Biological, and Environmental Statistics (in press).

           b. Buchanan, R. A.,  and J. R. Skalski. 2007.  Modeling adult salmonid movement and survival using telemetry tags in an industrialized river.  Submitted.

           c. Buchanan, R. A., J. R. Skalski, J. L. Lady, P. Westhagen, R. L. Townsend, and S. G. Smith.  2007.  Survival and transportation effects for migrating Snake River hatchery Chinook salmon and steelhead:  Historical trends from 1996-2004.  Volume XV in the BPA technical report series on Monitoring and evaluation of smolt migration in the Columbia Basin.  Bonneville Power Administration, Portland, Oregon.  

           d. Skalski, J. R., R. A. Buchanan, R. L. Townsend, T. W. Steig, and A. Grassell.  2007.  A quadruple-release model to estimate route-specific and dam passage survival at a hydroelectric project.  Journal of Agricultural, Biological, and Environmental Statistics (submitted).

     12.  The following software program user's manuals were updated:
            
              a. PitPro 4.0
              b. ROSTER 2.0
              c. USER 4.0