FishStats

Class lectures, length-comp data, and post-doc opportunities

2018-05-24T00:00:00+00:00

A few updates since my last email in March:

The UW class “Spatio-temporal models for ecologists” is wrapping up. All presentations are online, and recordings are also available here: https://github.com/James-Thorson/2018_FSH556. Its a good resource to learn about the statistical basis for spatio-temporal models.

A new paper shows benefits to developing age/length composition data for stock assessments using spatio-temporal models: http://www.nrcresearchpress.com/doi/abs/10.1139/cjfas-2018-0015#.WwbehPkvyUk. I still compile all reports and papers that use FishStats (https://github.com/nwfsc-assess/geostatistical_delta-GLMM/wiki/Applications), so please email me if you have one to add.

Colleagues and I have three spatio-temporal post-docs to announce:

Model testing for new size-structured snow crab population, based in Seattle and funded for 1 year (potential to extend based on performance). Please email Jim Thorson (James.Thorson@noaa.gov) for details.
Model development for combining multiple data sources in the Eastern Bering Sea to predict environmental impacts on distribution, based in Seattle and funded for 2 years. Please email Stan Kotwicki (stan.kotwicki@noaa.gov) and Jim Thorson (James.Thorson@noaa.gov) for details.
Analysis of species interactions using bird communities, based in Finland and funded for 2.5 years. Please see https://drive.google.com/file/d/1mKDRxyRxjv8g7Z-5FYjS46V0phAdxXRl/view?usp=sharing for how to apply. All positions are suitable for motivated Ph.D. graduates with experience in R, strong statistical background, and experience in applied ecology. We can develop skills in TMB and spatio-temporal methods during the position.

CAPAM workshop, CJFAS issue, and more updates

2017-11-29T00:00:00+00:00

Hi all,

A few updates since my last email (back in May!):

The CAPAM workshop on spatio-temporal analysis of fishery catch and effort data, Feb. 26-March 2 in La Jolla is now accepting abstracts (http://www.capamresearch.org/sites/default/files/CAPAM_spatio-temporal_modelling_workshop_Anouncement_2.pdf). It should be a great meeting! We’re looking for case-studies on spatio-temporal models and fishery CPUE analysis and hope people present.
I’ll be teaching a 1-week workshop, Jan. 29-Feb 1 in Seattle, on using VAST for generating indices from trawl and acoustic data using AFSC surveys. Please add your name online (https://docs.google.com/document/d/1z1MSlE0pvj6HrhNTRd5zD5L1wxBLLi1l4Zu9Hv7hxV0/edit?usp=sharing) if you want to receive information to remotely connect
The CJFAS special issue on spatial models is now out (http://www.nrcresearchpress.com/toc/cjfas/74/11). It has many examples of spatially-structured population models, plus how to integrate with spatio-temporal delta-models.
A new “Poisson-link delta-model” is now available in VAST, using ObsModel[2]=1, which approximates a Tweedie distribution and is more interpretable than the conventional delta-model (http://www.nrcresearchpress.com/doi/abs/10.1139/cjfas-2017-0266#.Wh7VUIanFaQ). Comparison using >100 stocks shows that it often fits better than the conventional delta-model.
I’m maintaining a list of stock assessments, reports, and journal articles using VAST or its precursors (https://github.com/nwfsc-assess/geostatistical_delta-GLMM/wiki/Applications). Please email me to add things, and its a good place to check to keep up with developments and applications.

Jim

Two post-doc positions now available

2017-05-16T00:00:00+00:00

Colleagues and I are currently looking for two post-doctoral candidates to collaborate on spatio-temporal models:

2.5 years of funding to understand environmental drivers of pelagic fish distribution near Vancounver Island BC, including impacts on predator-prey dynamics and natural mortality (details at https://tinyurl.com/lrpqmmg). The candidate must be located in Nanaimo for the majority of work (but funding includes periodic travel to Seattle), and must have received a Ph.D. within five years. Interested applicants should email: 1) CV; and 2) cover letter outlining the experience and skills they bring to the project to Jennifer Boldt (Jennifer.Boldt@dfo-mpo.gc.ca), Jim Thorson (James.Thorson@noaa.gov), and Robyn Forrest (Robyn.Forrest@dfo-mpo. gc.ca).
1 year of funding to understand how best to estimate length-structure for tropic tuna species, in preparation for an international CAPAM meeting on the topic in March 2018 (details at http://tinyurl.com/l4vgaev). The candidate must be located in La Jolla, and will be invited to contribute to the planned CAPAM meeting. Applicants should send a CV plus an cover letter explaining their interest to Brice Semmens (semmens@ucsd.edu), Jim Thorson (James.Thorson@noaa.gov), and Mark Maunder (mmaunder@iattc.org).

I’m also happy to point out several new papers demonstrating spatio-temporal analysis using FishStats models including:

Multispecies models for interactions using package MIST: http://onlinelibrary.wiley.com/doi/10.1002/ecy.1760/full
Showing improvement in assessment models given spatio-temporal index standardization using package SpatialDeltaGLMM: http://www.nrcresearchpress.com/doi/abs/10.1139/cjfas-2016-0137#.WRxln_nytaQ
Attributing distribution shift to climate and age-structured dynamics using package VAST: http://onlinelibrary.wiley.com/doi/10.1111/faf.12225/full
Synthesizing 10+ data sets to estimate distribution for input to ecosystem models using package SpatialDeltaGLMM: http://www.sciencedirect.com/science/article/pii/S016578361730111X
Standardizing fishery-dependent data while accounting for fisher targetting using package VAST: http://www.nrcresearchpress.com/doi/abs/10.1139/cjfas-2015-0598#.WRxnQvnytaQ

and the many exciting papers in the forthcoming CJFAS special issue on spatial stock assessment models.

Finally, Mark Maunder, Brice Semmens and I are organizing a CAPAM meeting in La Jolla for Feb. 26-March 2, 2018 on spatio-temporal models for analyzing compositional and fishery data. We look forward to hearing about the diverse work in this topic, and will organizing a special issue (maybe in ICES JMS).

cheers, Jim

Multispecies Interactions Spatio-Temporal (MIST) model now public

2017-02-01T00:00:00+00:00

I have two updates to regarding FishStats:

A new R package for a Multispecies Interactions Spatio-temporal (MIST) model is now public at www.FishStats.org, and is described in Thorson Munch Swain (2017) Ecology, abstract here. This biomass-dynamic model allows estimates of (1) local asymmetric impacts of each species on per-capita productivity of other species, (2) symmetric covariance to shared environmental drivers, and (3) fishing impacts using spatial harvest information. It has identical input/output, plotting, and Rmarkdown tutorial as other FishStats packages.
The first FishStats training workshop is next week. We have a full room, but people can attend remotely using details below. Please contact Sarah Stein with any connection questions. I’ll post recordings of the workshop and presentation slides later this month.

Webinar Information:

To officially register for the webinar, please follow the webex links below for each day’s workshop registration. Upon completing the online form for each day, you will receive a confirmation email with additional information including operating system requirements. The daily schedule is 8:30am - 4:30pm Pacific Time Wednesday, Feb. 8th through Friday, Feb. 10th. NMFS OCIO invites you to attend a Web seminar using WebEx. This event requires registration.

Day 1:

Topic: AFSC Training Workshop

Host: NMFS OCIO

Date and Time: Wednesday, February 8, 2017 11:00 am, Eastern Standard Time (New York, GMT-05:00)

To register for the online event ——————————————————-

Go to https://noaaevents2.webex.com/noaaevents2/onstage/g.php?MTID=e2e5d9b203b660ef59f8f53b1219f1e41
Click “Register”.
On the registration form, enter your information and then click “Submit”.

Day 2:

Topic: AFSC Training Workshop (Day 2) Host: NMFS OCIO Date and Time: Thursday, February 9, 2017 11:00 am, Eastern Standard Time (New York, GMT-05:00) ——————————————————- To register for the online event ——————————————————-

Go to https://noaaevents2.webex.com/noaaevents2/onstage/g.php?MTID=e3e750409c9c3aa7eb4e1d0f5d4098eb2
Click “Register”.
On the registration form, enter your information and then click “Submit”.

Day 3

Topic: AFSC Training Workshop (Day 3) Host: NMFS OCIO Date and Time: Friday, February 10, 2017 11:00 am, Eastern Standard Time (New York, GMT-05:00)

To register for the online event ——————————————————-

Go to https://noaaevents2.webex.com/noaaevents2/onstage/g.php?MTID=ef414c4c68d644d3738cb8e1f2ddfd916
Click “Register”.
On the registration form, enter your information and then click “Submit”.

Once the host approves your registration, you will receive a confirmation email message with instructions on how to join the event.

For assistance ——————————————————- You can contact NMFS OCIO at: nmfs.servicedesk@noaa.gov

Upcoming training workshops for FishStats

2016-11-28T00:00:00+00:00

There will be three upcoming opportunities to learn about the spatio-temporal tools for estimating fish abundance, habitat associations, and species interactions using tools at www.FishStats.org:

A 2-hour hands-on tutorial at the UW Think Tank, tomorrow Nov. 29 at 9-11am PST in FSH 203 (abstract below). If you’d like to attend remotely, please go to www.GoToMeeting.com and click “Join a Meeting,” and enter using Meeting ID #242-434-181
A 1-hour QUEST overview of recent research and papers, Dec. 15 at 10-11am PST. I’ll send out offsite attendance information closer to the date.
A 1-week long training to use FishStats for index standardization, spatial dynamic factor analysis, fishery-dependent standardization with targeting, and ecosystem analysis using VAST and SpatialVAM, Feb 20-24 at AFSC. Jin Gao will be TA, and we will encourage people to attend with data sets. I’ll again send out attendance info later.

I hope to keep improving our user- and training-resources, and welcome feedback.

Think Tank Abstract:
Collaborators and I have been developing, testing, and documenting spatio-temporal models and software for single- and multi-species analysis of fishery and/or survey data. R packages for implementing these models are now publicly available and documented through www.FishStats.org, a spatio-temporal toolbox for fisheries science.

In this Think Tank, I will present a small number of slides (<15 total) to recap recent model developments and applications. I will then lead a discussion of different applications for which FishStats could be useful in 2017 stock assessments, e.g., estimating survey indices, standardizing fishery catch rates, estimating distribution shifts or range expansion/contraction, or identifying multispecies associations. With input from attendees, I will then provide a semi-structured, hands-on demonstration of how different modelling goals can be achieved.

I encourage attendees to bring a laptop and to install a recent version of R (e.g., 3.3.1) and appropriate Rtools (if using a windows OS). Attendees will install other necessary software (including INLA and TMB) as we go through tutorials together.

Introducing resources in FishStats

2016-10-28T00:00:00+00:00

FishStats compiles different tools for harmonizing data for marine fishes worldwide, analyzing data to estimate abundance, distribution shift, range expansion, and spatial overlap, and also visualize results. Current tools include:

a markdown tutorial for a single-species spatio-temporal model, SpatialDeltaGLMM. SpatialDeltaGLMM can now calculate range expansion/contraction from a new ProcB paper, as well as distribution shifts from a MEE paper, in addition to previous abundance-index estimates from an ICESJMS paper.
a markdown tutorial for a multispecies model VAST. VAST is identical to SpatialDeltaGLMM for single-species data (including indices, range shift, and range expansion metrics), but can also estimate community dynamics via spatial dynamic factor analysis from a Global Ecology Biogreography paper, or control for fisher targeting using multispecies fishery data to estimate abundance from a CJFAS paper.
FishViz.org visualizes model output from SpatialDeltaGLMM applied to survey data for >400 species in 15 regions worldwide (including all groundfishes in Alaska and West Coast regions).
FishStats now has a FishStats-listserv

Spatial statistics: An introduction

2016-05-31T00:00:00+00:00

FishStats uses Template Model Builder and modern statistical techniques to rapidly estimate biological parameters (fixed effects) and spatial correlations (via inclusion of random effects) when analyzing real-world data. The techniques are relatively new, and not well documented in any single textbook or paper. For interested students, I recommend reading class materials from a Spring 2016 UW course “Spatio-temporal mdoels for ecologists”. Materials include:

Recorded lectures
Example code in TMB and R
Homework assignments and solutions

These materials will eventually be distributed as a textbook, so please do not reproduce them without permission from James Thorson