This is an informal place for me to list my research projects. It is not meant to be complete or even informative. Poke around and contact me with any questions.

Broadly speaking I'm interested in developing and applying quantitative techniques to manage marine resources efficiently and sustainably. Below is a collection of research projects I’ve been involved in, with newer projects at the top. This is an evolving list so check back later for more.

Bayesian algorithms for fisheries models.

Testing the efficiency of gradient-based MCMC algorithms

Bayesian techniques have many advantages in fisheries science, but many models can often be prohibitively slow. However, a class of algorithms, called Hamiltonian Monte Carlo, has been gaining traction in the statistical community, and has high potential to substantial reduce run times.

These gradient-based MCMC algorithms are not widely used in the fisheries community, likely for several reasons. First, the software typically used for modeling (ADMB and TMB) do not have these algorithms available (at least in a usable form). Second, they are more difficult to understand and tune than traditional samplers (although new versions are self-tuning). Consequently, most fisheries analysts resort to avoiding Hamiltonian algorithms, or using them with default, suboptimal performance. Therefore research is needed to test the utility, and demonstrate the implementation of Hamiltonian algorithms in order to advance their inclusion in the toolbox of quantitative fisheries scientists. I have worked to add beta versions of these algorithms to TMB, and have been comparing their performance against JAGS. You can see the code here and I encourage you to try them out and let me know how it works.

I gave a seminar on these methods where I tried to explain them as best I could. Have a look if you're curious: PPT slides.

Standardizing commercial catch rates for Pacific halibut.

The effect of hook spacing on catch rates

The International Pacific Halibut Commission (IPHC) is responsible for managing halibut in the North Pacific. While they have reliable fishery independent data, the IPHC also uses commercial logbook data to develop a catch per unit effort (CPUE) index for use in their stock assessment. There are two steps in creating an index: (1) determine a reliable unit of effort, and (2) standardizing the data with statistical techniques used to remove factors influencing CPUE trends other than changes in abundance.

The nominal unit of effort for halibut long line gear is a hook, but the efficiency of hooks varies as a function of how close they are. To address this, I am working with IPHC staff to reanalyze data collected in an experiment from the 1970s to establish an “effective hook” relationship. This relationship can also be elucidated from the raw catch data themselves and used to corroborate the results from the empirical analysis. This new effective hook definition leads directly into the second step, standardizing the catch data.

Simulation testing of fisheries stock assessments

An R package that facilitates reproducible, flexible, and rapid end-to-end simulation testing with stock synthesis

Fisheries stock assessment models (used to management fisheries in the US and around the world) are complex and often unpredictable. One approach to help understand their statistical behavior is to use simulation testing. With this in mind a group of students and postdocs, based mainly at the UW, developed a software platform called ss3sim (code and paper) which facilitates this type of simulation testing. So far we have used our package in 5 studies:

  • The effect of length bin width. Read more
  • Using empirical weights instead of somatic growth functions. Read more
  • Retrospective patterns. Read more
  • Time-varying natural mortality. Read more
  • The importance of length and age composition data. Read more

  • Several other studies are ongoing so check back later for more.

    California blue whale assessment.

    Do fatal ship strikes threaten the recovery of this endangered population?

    Many populations of whales were hunted to near extinction, and some have shown signs of recovery while others remain critically endangered. The blue whales off California were no exception, and the lack of evidence for an increase in their population size left many concerned about ship strikes. The goal of my M.S. thesis was to investigate the past, present, and future population trends of this population, and try to quantify the risk of ship strikes.

    The first step (and my first thesis chapter) was to estimate how many whales of this population were caught by whalers, separating those out from other blue whale populations in the North Pacific. I did this by looking at the spatial patterns in acoustic calls, which are unique by population. The resulting estimates of catches from that chapter fed directly into my second chapter, which was to estimate their past, current and future status, as well as the impact of potential threats (e.g. ship strikes) moving forward. We found that the whales had likely already recovered, and the lack of increase was due to environmental limitations. This positive story went viral in the news (see here), although some scientists remain skeptical (which is good for science!).

    Trevor and I have applied for more funding to continue studying blue whales in the North Pacific, so perhaps there will be more on that later...

    Random side projects

    Some fun, non-dissertation distractions...

    I utilize the open source software ADMB a lot for research. For part of my research I needed to have a better understanding of how it works internally, so I dug through the source code to figure it out. As a way to give back I wrote up some informal documentation for other users that hopefully will be useful. Check out some details about covariance calculations and an MCMC guide.

    I spent some time practicing my visualization skills by making an animation of the history of whaling. Check it out here.

    I worked with some fellow QERM students to put together a network visualization of our program. It's still in beta but looks pretty cool so far.

    My friend Sean and I did some work for Dr. Ray Hilborn making some maps of the RAM and FAO data. Check them out as they scroll by in the background here