Special Workshop on Salmon Escapement Goal Science
November 18-19, 2010
ABSTRACTS
(In order of presentation)
CLICK ON TITLES TO LINK TO PRESENTATIONS
Why new approaches are needed for salmon escapement
Jack Williams, Senior Scientist, Trout Unlimited
A manager’s view of the needs for improved escapement goal science
Kit Rawson, Tulalip Tribes
Escapement was traditionally defined as the portion of a stock that gets past the fishery to spawning areas. However, today managers consider all stages of the salmon life cycle, not just spawning, all parameters that describe a viable salmonid population, not just abundance, and all activities that affect population status, not just harvest. A rethinking of escapement goals for the 21st century, therefore, means expressing population objectives in terms of all viable salmonid population parameters (abundance, productivity, spatial distribution, and diversity) at all stages of the salmon life cycle. These become the goals of recovery plans that coordinate harvest, hatcheries, habitat restoration and habitat protection and the basis for strategic monitoring to track progress towards recovery. Managers are attempting to incorporate these concepts into implementation of the recovery plan for Puget Sound Chinook salmon.
Emerging Science for Estimating Salmonid Natural Production Potential
Eric Knudsen, Consulting Fisheries Scientist, Mt. Vernon, WA, and Randy Ericksen, State of the Salmon, Portland, OR
Pacific salmon have been diminished because of habitat degradation, dams, pollution, hatchery management, and overfishing. The combination of these assaults has resulted in “lost baselines” regarding expectations for the number, productivity, diversity, and distribution of spawners that should be filling available habitats, many of which have been lost or degraded, and some of which are new being restored. Conceptual and quantitative models hold the best promise for estimating the abundance and diversity of salmon that should be refilling the available habitats. In this talk, we review past, current, and developing computational and simulation models designed to assist managers in estimating salmon escapement goals. We suggest that the next generation of salmon escapement goal modeling science must fully integrate habitat, life-history, biological diversity, nutrient cycling, species interactions, and environmental variation into estimates of natural production potential of Pacific salmon. The advanced models must also account for the effects of hatchery, harvest, and hydropower management on the natural production potential. As one step toward evaluating the future of modeling the production potential, we will set the stage for the remainder of the workshop by providing a framework for evaluating the relative utility of various models for the future. From this evaluation, workshop attendees can make recommendations for emphasis of emerging modeling research and development.
Problems and solutions in escapement goal management of upper Cook Inlet salmon fisheries
Ray Beamesderfer, Cramer Fish Sciences, 600 NW Fariss Road, Gresham OR 97030, (503)491-9577, beamesderfer [at] fishsciences.net
Alaska’s escapement-based management system has produced the largest and most valuable wild salmon fishery the in world. However, no management system designed by humans and implemented by the government can be presumed to be without it’s imperfections. This presentation critically examines three limitations of escapement-based management including: 1) unpredictability introduced by normal environmental variation, 2) inaccurate escapement goals due to unrepresentative or non-stationary stock-recruitment data, and 3) fishery impacts on mixed stocks or substocks of differing productivity. Potential remedies include: a) increased use of risk-based analysis of fishery strategies that explicitly consider normal variability in salmon dynamics, b) precautionary application of escapement goals when based on limited information, and c) management safety factors to protect species and stock diversity. Issues are illustrated with examples from current Upper Cook Inlet fishery management.
The escapement goal/recruitment modeling knot versus an alternative: the Red Queen paradigm
Mark Chilcote, NOAA Fisheries
Population recruitment models represent one of the traditional approaches used by fish managers to identify and set spawner escapement goals. Such models depend on the assumption that a single predictor variable (spawner abundance) has a strong relationship with the number of subsequent recruits produced. However, in an examination of data for 121 populations of steelhead, Chinook, and coho salmon I found the detection of a clear relationship between spawners and recruits to be rare, as evidenced by a median R2 value for fitted models of 0.04. In addition, the pattern of model residuals for most populations was serially correlated suggesting that important elements controlling the recruitment process had not been accounted for. However, by adding an environmental variable to these models I found that the ability to account for the observed annual variation in recruitment was greatly improved, with the median R2 value for the fitted models increasing to 0.44. In achieving this result I considered four indices for the environmental variable including: PDO, Columbia River flow, air temperature, and mountain snow depth. At least two of the indices, temperature and snow, are sensitive to climate change and I use this link to hypothesize possible scenarios for the modeled populations in terms of future abundance and persistence. I concluded that for many of these populations future survival will depend on the ability to rapidly adapt to the forthcoming environmental change. I use these results as another piece of evidence that a new fish management paradigm that focuses on evolutionary speed rather than meeting simple escapement goals is needed. I explore the practical elements of this proposal drawing on the idea that species are in a evolutionary race with a changing environment and must constantly ‘run’ just to stay in one place; a paradigm that evolutionary biologist Leigh Van Valen described in 1973 as the “Red Queen” hypothesis.
Greg Blair (presenter), Larry Lestelle, Lars Mobrand, Jesse Schwartz and Chip McConnaha, ICF International, 206 463 6022, gblair [at] icfi.com
We present results of simulation modeling of salmon (Oncorhynchus sp.) life history diversity and its potential to affect population productivity and abundance. Cumulative productivity and capacity for a particular life history pathway is affected by the quantity of key habitat encountered, its quality, and the amount of exposure to these factors. Juvenile Chinook salmon typically express multiple life history pathways from emergence to ocean entry. Each pathway is a unique history of exposure to habitat conditions in time and place. The Ecosystem Diagnosis and Treatment model was used to simulate the Beverton-Holt productivity and capacity parameters across a variety of life histories. The model was used to explore the consequence of life history diversity and habitat conditions on population productivity and abundance of fall Chinook originating from streams in the lower Columbia and Puget Sound. Results show how priorities for habitat restoration can differ considerably for different life histories. Simulation modeling of historic conditions and life histories are used to inform the potential for life history diversity. Furthermore, we show how life history diversity should be a consideration when in discussing escapement management.
A Habitat And Life-History Based Approach To Estimating Basin Carrying Capacity For Chinook Salmon
Steven P. Cramer, Mark Teply, and Randolph Ericksen, Cramer Fish Sciences, 600 NW Fariss Road, Gresham OR 97030, (503)491-9577
We developed a model to predict the freshwater carrying capacity for rearing Chinook salmon with both ocean-type and stream-type life histories. Variable and sometimes large proportions of juvenile Chinook salmon are known to migrate from their natal area, and rear for varying durations in downstream areas. This behavior has complicated attempts to estimate carrying capacity based on habitat availability for juvenile Chinook. We developed a habitat-based model that addressed this challenge by accounting separately for six life history strategies, incorporating changing habitat preferences as fish grow, using observed patterns of migration timing to account for the average timing and duration of fish use in each stream reach, and applying seasonal survival rates to account for reduced numbers of fish over time and migration distance. The habitat capacity in each stream reach was estimated by the Unit Characteristic Method, including a large-channel scalar to account for decreasing fish use with distance from shore. Migration between stream reaches was accounted for via bi-weekly time steps. We completed a demonstration of the model for all spring Chinook populations in the Willamette Basin, which included 8 subbasins, and 859 stream reaches. Several datasets were available to inform the framework with input values on extent of spring Chinook fish use, habitat quality preferred by juvenile spring Chinook, choice of juvenile life-history pathways, timing of emergence, and timing of migration either in-stream or to sea. We found meaningful differences in production potential and life history composition between subbasins. Subyearling life-histories extensively used non-natal reaches, while less than 10% fish remaining through the summer were in non-natal reaches. Model predictions of adult equivalent carrying capacity were congruent with the observed distribution of spring Chinook production in the basin. This method of estimating carrying capacity has the advantages of linking local habitat manipulations to consequences at the population level, and for establishing spawner escapement goals where the time series of estimated adult escapement is limited.
Using Landscape-Based Modeling to Predict Salmonid Habitat Capacity and Productivity Potential at the Watershed Scale
Jody Lando, Frank Ligon, Bill Dietrich, Peter Baker, Colin Bode
Studying all aspects of an ecosystem to develop suitable escapement goals is either daunting in its complexity or necessarily generalized to non-mechanistic indices of ecosystem health, which may or may not have relevance for salmon populations. Instead we consider the benefits of RIPPLE, a digital terrain?based model that predicts the distribution of fish habitat conditions throughout a watershed and simulates salmon population dynamics. It is a powerful tool for evaluating the effectiveness of restoration and recovery planning strategies. Developed in collaboration between Stillwater Sciences and UC Berkeley, RIPPLE characterizes the geomorphic and ecological processes that create and maintain freshwater salmon habitat.
One of the guiding principles of RIPPLE is the assumption that physical processes and the resulting environment—specifically topography, geology, climate, drainage area, channel gradient, channel longitudinal profile—are essentially time?invariant compared with ecosystems and the animal and plant populations supported by these ecosystems. This assumption enables us to construct a model that establishes a physical template composed of information such as topographic data, channel networks, and geology.
RIPPLE thus uses geomorphic characteristics and physical habitat characteristics, combined with density and suitability criteria by species and life stage, to predict reach?specific historical, current, and potential future salmon habitat conditions. RIPPLE then employs a multi?stage, stock?production model to predict long?term average abundance for each life stage. The model can also produce a single number, such as ?long?term average escapement? for the purposes of scenario comparisons or hypothesis testing. Operationally, these elements are evaluated through three sub?models that are run sequentially: (1) a physical model (“GEO”), executed through ESRI’s ArcGIS Desktop, (2) a habitat carrying capacity model (“HAB”), and (3) a population dynamics model (“POP”).
RIPPLE is a flexible, scientifically rigorous tool designed for salmon conservation and management. It can be used with limited data and still produce credible, preliminary results to guide further hypothesis testing; it can be customized with watershed specific data to generate progressively more refined results; and it can be used to assess the value of collecting additional field data.
Winter OBAN (Oncorhynchus Bayesian Analysis), a Statistical Life-Cycle Model for Winter-Run Chinook
Noble Hendrix, R2 Resource Consultants, Inc., Redmond, WA, Robert Lessard, University of Washington, Seattle, WA, Ray Hilborn, University of Washington, Seattle, WA
Salmonid research in the California Bay – Delta has tended to be focused on the controllable freshwater factors that affect salmon run variability, such as flows and diversions, but there has been less emphasis on other sources of variability such as the ocean. We constructed a life-cycle model of winter-run Chinook in the Sacramento River, California that accounts for mortality during all phases of the life-history, estimates model coefficients in a statistical framework, evaluates covariates that may explain dynamic vital rates, and incorporates uncertainty in the model structure and model coefficients through Bayesian estimation. The model is entitled Oncorhynchus Bayesian Analysis (OBAN). Evaluation of multiple anthropogenic and environmental driver variables indicated that temperature in the spawning reaches, minimum flows in the fry rearing reaches, access to Yolo bypass and water exports in the Delta, upwelling dynamics in the Gulf of Farallones, and ocean harvest were able to explain variability in the winter-run Chinook population dynamics. Evaluation of the impact of the effects indicated that the winter-run abundance is most sensitive to temperatures in the spawning reaches and flows in the fry rearing stage. The sensitivity of the model is somewhat dependent upon the abundance data to which the model was statistically fit. Thus, adult escapement and juvenile counts at Red Bluff Diversion Dam (RBDD) can be used to evaluate the upstream portion of the life-history, whereas the impact of factors occurring after RBDD are less easily identified due to correlation of survival rates in the Delta, Gulf of Farallones, and Ocean stages. Such life history models are critical for understanding the factors that have been associated with changes in population abundance historically, and thus form a short list of actions that may facilitate population recovery in the future.
Dan Rawding, Washington Department of Fish and Wildlife
The Washington Department of Fish and Wildlife developed a Statewide Steelhead Management Plan in 2008. The next phase of the plan calls the development of quantifiable population escapement objectives balancing ecosystem, demographic, genetic concerns with fishing opportunities. Lower Columbia River steelhead populations are listed under the Endangered Species Act and historic maximum sustainable yield escapement goals for these populations were based on professional opinions from US vs. Oregon Technical Advisory Committee or application of the Potential Parr Production model from the Boldt Case area. Challenges to develop biological reference points for Lower Columbia River steelhead include: 1) a ten-fold variation in smolt to return rates since the 1980’s and short data series, 2) various levels of hatchery spawning and relative reproductive success, 3) few data sets with low escapement measurement error such as census counts or precise mark-recapture estimates, and many with large measurement error due to AUC redd counts and sparse coverage of spawning areas, and 4) derived estimates of hatchery escapement. Approaches to address these challenges are to 1) standardize counts into fish per unit of area allowing the use of hierarchical modeling to avoid overfitting data with high measurement error while borrow strength from better monitoring programs, 2) analyze data using various spawner-smolt relationships to reduce order of magnitude marine survival variability that obscures true spawner to adult recruit patterns with different error structures, and 3) use estimates of relative reproductive success to the smolt stage to adjust hatchery spawners to wild equivalent spawners. This approach is stronger on biological reference points and stochasticity but weaker on the incorporation of diversity, distribution, and ecological consideration, which are measured annually but not incorporated into the model. In addition, the smolt estimates, which are required by the model, can be used to determine trends in smolt abundance or freshwater productivity and capacity from spawner recruit analysis in a before-after design to measures of the effectiveness of restoration programs. Future work includes refinement of the watershed area approach to a spawning distribution approach using GIS derived attributes such as gradient to that may more accurately reflect individual population biological reference points.
Estimating Coho Habitat Capacity Using GIS-based Variables
Mark Meleason, Pete Lawson, Dan Miller, Kelly Burnett, and Gordie Reeves
We are linking a dynamic landscape model with a habitat-based life cycle model for Oregon coastal Coho to explore scenarios of natural disturbance and forest management. The landscape model simulates both wood and sediment inputs to the stream. The Coastal Landscape Analysis and Modeling Study provides inputs to the landscape model of initial forest conditions and simulated responses to a variety of land management scenarios. Modeled wood inputs to the stream are from riparian and upslope processes (e.g., tree mortality and debris flows) and modeled sediment sources are landslides and debris flows. In-channel wood and sediment budgets are simulated for each reach. The Coho life-cycle model is based on the approach used in Nickelson and Lawson (1998). They estimated production of egg, summer parr, smolts, adults, and spawners for each modeled stream reach. Smolt capacity, which reflects habitat quality, was estimated from stream survey data using Oregon Department of Fisheries and Wildlife’s Habitat Limiting Factors Model (HLFM). In their published model application, habitat quality was either held constant or varied uniformly over the landscape. There was no explicit spatial component of either habitat quality or salmon distribution among reaches.
The research presented here describes how we are linking outputs from the dynamic landscape model to estimate habitat capacity spatially in the landscape, which is then used in the Coho life cycle model. Specifically, we are developing a method that estimates habitat capacities at various stages of the life cycle based on simulated estimates of sediment and wood loads, and physical characteristics of the channel from a 10-m digital elevation model. We are using HLFM v 7.1 to calibrate our estimates of habitat capacities for a given life stage. Once this model is complete we will be able to explore the implications of spawner distribution within a basin, and the implications of changing habitat patterns as the landscape evolves.
Rishi Sharma, Columbia River Intertribal Fisheries Commission
Essential ecological considerations for managing salmon populations
Martin Liermann, Northwest Fisheries Science Center, NOAA Fisheries
Historical Accounts: Outliers or Anchor Points?
Jay Hesse – Nez Perce Tribe