From sequential to integrated Bayesian analyses: Exploring the continuum with a Pacific salmon spawner-recruit model

Staton, Benjamin A.; Catalano, Matthew J.; Fleischman, Steven J.

doi:10.1016/j.fishres.2016.09.001

Cited by 10 publications

(11 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The model is structured in a hierarchical Bayesian state‐space framework that integrates both process and observation errors. Observation errors on returns and catches are integrated through a sequential approach, similar to the one developed in stock assessment models for Atlantic salmon in the Baltic (Michielsens et al., ) and Chinook salmon in Alaska (Staton et al., ). Probability distributions on returns and catches at sea were derived separately from the life cycle model and then used to approximate likelihoods (Michielsens et al., ).…”

Section: Discussionmentioning

confidence: 99%

“…() who developed a fully integrated observation model for returns. The choice of using a sequential or an integrated approach represents a trade‐off between model realism and computational efficiency (Maunder & Punt, ; Staton et al., ). A fully integrated model would provide a more transparent view of how the data are incorporated in the entire process being modelled.…”

Section: Discussionmentioning

confidence: 99%

“…Building an integrated model (Maunder & Punt, ; Rivot et al., ; Schaub & Abadi, ) that explicitly integrates complicated observation models would dramatically increase the complexity of the full model. Therefore, a sequential approach (Michielsens et al., ; Staton, Catalano, & Fleischman, ) is used that consists of (i) processing observation models separately to reconstruct probability distributions that synthesize observation uncertainty around the time series of catches and returns for the 13 SUs and (ii) using those distributions as likelihood approximations in the population dynamics state‐space model. Probability distributions for returns and catches are derived from a variety of raw data and observation models, specific to each SU (except for the mixed‐stock fisheries at sea) as originally developed by ICES to provide input for PFA models (see ICES, and Supporting Information S1 for further details).…”

Section: Methodsmentioning

confidence: 99%

See 2 more Smart Citations

Evidence for spatial coherence in time trends of marine life history traits of Atlantic salmon in the North Atlantic

et al. 2018

View full text Add to dashboard Cite

A hierarchical Bayesian life cycle model is presented that considers spatial covariation of marine life history traits of Atlantic salmon (Salmo salar) populations in the North Atlantic. The model is based on a collective analysis of the dynamics of 13 stock units (SUs) from two continental stock groups (CSGs) in North America and Southern Europe in a single hierarchical model over the period 1971–2014. The model sets up a new assessment framework for Atlantic salmon stocks. It also provides a framework to investigate the drivers of changes in Atlantic salmon population dynamics including disentangling the effects of fisheries from those of environmental factors in a hierarchy of spatial scales. It is used to test the hypothesis of a strong spatial synchrony in marine life history dynamics of Atlantic salmon populations. The trends in two key parameters associated with the early marine phase of the life cycle are estimated: (i) the marine survival during the first summer–autumn spent at sea and (ii) the proportion of fish maturing after the first winter at sea. The results provide evidence of a decline in the marine survival together with an increase in the proportion of fish that mature after the first winter at sea, common to all SUs. Our results show an increased coherence in the covariations of trends in these two marine life history traits related to geographic proximity of SUs which support the hypothesis of a coherent response of geographically proximate Atlantic salmon populations that likely share similar migration routes.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Evidence for spatial coherence in time trends of marine life history traits of Atlantic salmon in the North Atlantic

et al. 2018

View full text Add to dashboard Cite

show abstract

“…Integrated analysis of different data sources as a tool to model the dynamics of exploited populations has been used for decades in fisheries research (reviewed by Maunder and Punt 2013). More recently, IPMs have been applied to harvested populations in terrestrial ecosystems (Gauthier et al 2007, Conn et al 2009, Fieberg et al 2010, Péron et al 2012, Lee et al 2015, Staton et al 2017, Arnold et al 2018). Importantly, IPMs built on exploited populations often integrate age‐at‐harvest data and capture–mark–recapture–recovery (CMRR) data into age‐structured population models (Methot Jr and Wetzel 2013, Arnold et al 2018, Scheuerell et al 2019).…”

Section: Introductionmentioning

confidence: 99%

Efficient use of harvest data: a size‐class‐structured integrated population model for exploited populations

et al. 2021

View full text Add to dashboard Cite

Many animal populations are subject to hunting or fishing in the wild. Detailed knowledge of demographic parameters (e.g. survival, reproduction) and temporal dynamics of such populations is crucial for sustainable management. Despite their relevance for management decisions, structure and size of exploited populations are often not known, and data limited. Recently, joint analysis of different types of demographic data, such as population counts, reproductive data and capture–mark–recapture data, within integrated population models (IPMs) has gained much popularity as it may allow estimating population size and structure, as well as key demographic rates, while fully accounting for uncertainty. IPMs built so far for exploited populations have typically been built as age‐structured population models. However, the age of harvested individuals is usually difficult and/or costly to assess and therefore often not available. Here, we introduce an IPM structured by body size classes, which allows making efficient use of data commonly available in exploited populations for which accurate information on age is often missing. The model jointly analyzes size‐at‐harvest data, capture–mark–recapture–recovery data and reproduction data from necropsies, and we illustrate its applicability in a case study involving heavily hunted wild boar. This species has increased in abundance over the last decades despite intense harvest, and the IPM analysis provides insights into the roles of natural mortality, body growth, maturation schedules and reproductive output in compensating for the loss of individuals to hunting. Early maturation and high reproductive output contributed to wild boar population persistence despite a strong hunting pressure. We thus demonstrate the potential of size‐class‐structured IPMs as tools to investigate the dynamics of exploited populations with limited information on age, and highlight both the applicability of this framework to other species and its potential for follow‐up analyses highly relevant to management.

show abstract

“…These issues warrant exploration of alterations to spawner-recruit models that accommodate time trends in demography and heterogeneous reproductive output of different spawners, as suggested by an expert panel on declines of Chinook salmon in Alaska (Schindler et al 2013). Variability in age composition has been incorporated into spawnerrecruit models for Chinook salmon in Alaska as random fluctuations (e.g., Hamazaki et al 2012;Staton et al 2017) or with estimated time trends (Fleischman and McKinley 2013;McKinley and Fleischman 2013;Reimer and DeCovich 2020), but only 4 D r a f t as a means to explain variability in the data and not as an explicit link to escapement quality or productivity. Size-based escapement goals have been implemented for Chinook salmon in the Kenai River (Fleischman and Reimer 2017) to address assessment limitations (i.e., uncertainties in size-based sonar species apportionment) and in southeast Alaska (Heinl et al 2014), but in general, escapement quality concerns have rarely been translated to changes in Pacific salmon management.…”

mentioning

confidence: 99%

Incorporating demographic information into spawner–recruit analyses alters biological reference point estimates for a western Alaska salmon population

Staton

Catalano

Fleischman

et al. 2021

Can. J. Fish. Aquat. Sci.

Self Cite

View full text Add to dashboard Cite

Changes over time in age, sex, and length-at-age of returning Pacific salmon have been widely observed, suggesting concurrent declines in per capita reproductive output. Thus, assessment models assuming stationary reproductive output may inaccurately estimate biological reference points that inform harvest policies. We extended age-structured state-space spawner-recruit models to accommodate demographic time trends and fishery selectivity to investigate temporal changes in reference points using Kuskokwim River Chinook salmon (<i>Oncorhynchus tshawytscha</i>). We illustrate that observed demographic changes have likely reduced per capita reproductive output in an additive manner, for example, models including changes in both length-at-age or age composition showed larger declines than models incorporating only one time trend. Translated into biological reference points using a yield-per-recruit algorithm, we found escapement needed for maximum sustained catch has likely increased over time, but the magnitude further depended on size-selective harvest (i.e., larger changes for reference points based on larger mesh gillnets). Compared to traditional salmon assessments, our approach that acknowledges demographic time trends allows more complete use of available data and facilitates evaluating trade-offs among gear-specific harvest policies.

show abstract

From sequential to integrated Bayesian analyses: Exploring the continuum with a Pacific salmon spawner-recruit model

Cited by 10 publications

References 21 publications

Evidence for spatial coherence in time trends of marine life history traits of Atlantic salmon in the North Atlantic

Evidence for spatial coherence in time trends of marine life history traits of Atlantic salmon in the North Atlantic

Efficient use of harvest data: a size‐class‐structured integrated population model for exploited populations

Incorporating demographic information into spawner–recruit analyses alters biological reference point estimates for a western Alaska salmon population

Contact Info

Product

Resources

About