Gavin Fay

From Santa Fe Institute Events Wiki

View of Seattle and the Space Needle from Kerry Park. May 2009.

About Me

G’day! I am currently employed as a postdoctoral research fellow in fisheries modelling at CSIRO Marine and Atmospheric Research in Hobart Australia, and am also in the process of completing my PhD in Aquatic & Fishery Sciences at the University of Washington, Seattle. I am originally from England, and made the move from Seattle to Tasmania about 18 months ago. When I am not modelling fisheries I enjoy travelling, hiking, playing soccer, listening to live music, camping, kayaking, watching movies, and drinking good beer, the latter preferably shared with good company over great food.

e-mail: gavin DOT fay AT csiro DOT au

Research Interests

My general research interests centre on the application of quantitative approaches for providing scientific advice for the management of marine populations and ecosystems.
A blue sky type question could be “How should we manage marine systems?”

My research focuses on developing and testing tools for stock assessment of marine fisheries and marine mammals, and using simulation modelling tools to evaluate the performance of management strategies designed to satisfy fisheries management objectives. More specifically, I am interested in methods for characterising and accounting for uncertainty within exploited marine ecosystems when taking action to satisfy fisheries management objectives, and evaluating how the performance of different management actions is impacted by this uncertainty.

My PhD research at UW examined the impacts of uncertainties associated with spatial structure when modelling and managing marine metapopulations.

My postdoctoral research project at CMAR is addressing the implications of accounting for non-stationary dynamics for stock assessment and management of marine fisheries within Australia. I am asking questions such as:

  • “How do our management systems perform when the things assessment scientists routinely assume to be constant (e.g. growth, productivity, predation mortality) are actually changing?”, and
  • “Can we identify candidate management strategies that are robust to such changes?”.

This involves a lot of simulation modelling, of course based around and guided by good old-fashioned data.

Besides my research projects, I also conduct stock assessments for several species (most recently for Patagonian toothfish at Macquarie Island), test harvest strategies and data collection schemes, and communicate the results to industry and managers.


My quantitative background definitely lies in the realm of fitting ecological models to data, with most expertise in the statistical modelling of population dynamics, both in a maximum likelihood and Bayesian framework. State-space models, hierarchical models, integrated analysis, random effects: as Julie Andrews would say, these are a few of my favourite things.

Using the results of these types of models, I have experience with methodologies such as Risk Analysis and Policy Evaluation. I spend a lot of time comparing the performance and likely consequences of assessment methods and alternative management actions, using what in fisheries has become known as the ‘Management Strategy Evaluation’ (MSE) approach (alternatively, simulation modelling with operational feedback control).

A few years worth of participating in stock assessment meetings, communicating assessment results to fisheries managers and industry members, has given me a perspective on how the management of a complex multispecies fishery actually works (and some of what doesn’t).

For statistical programming and other analyses, I use R extensively, as well as AD Model Builder (which uses C++, sort of), and BUGS. Occasionally (or more so), I’ll also write programs in FORTRAN.


I am increasingly interested in ‘whole-of-system’ approaches to understanding marine ecosystem dynamics and the communities (including but not limited to, fishing fleets) that interact with them. Recognizing ‘process error’ when assessing populations can reflect variability in a range of factors such as: population dynamics, climate impacts, ecological interactions among species, the effects of fishing, all of which operate at a suite of spatial and temporal scales. Methods to account for possible changing behaviour of groups and individuals are important in being able to quantify possible system responses to management decisions. For example, an important fishery management question such as: “How do fishermen make decisions about where and when to allocate effort?” in a multiple species fishery implies knowledge and/or understanding about behavioural response to management actions, economic drivers, and the spatial and temporal dynamics in availability and abundance of the fish resource.

Simple, transparent approaches to resource management are good (stakeholders can understand them), but the simple methods need to be tested against hypotheses that reflect an understanding of system behaviour. That’s where, for me, CSSS comes in.

I am looking forward to learning about a lot of new methods, and hope to develop my thinking on how these can be applied within my field. However, I do not necessarily want restrict myself to analyses of marine systems, or indeed population dynamics. I am excited to meet, collaborate with, and learn from the other participants in the course, all of whom I am immensely impressed with after reading some of these pages. I look forward to lots of great multi-disciplinary ideas and discussions over the three weeks and beyond.

Project Ideas

I've added a few ideas to the projects page. These are probably way over-convoluted right now, but represent some of my current thinking....