Q: How would you describe your role at Cramer Fish Sciences?

A: Once a project's objectives are defined and the data is collected, my role bridges the gap between the raw data and the defined objectives. It sounds simple, but data is messy and rarely informative by itself. Statistical and ecological models informed by the raw data can be extremely insightful, however. Building these models and tailoring them specifically to the situation in question is truly my forte and is my primary focus here at Cramer Fish Sciences. I work with project leaders to interpret how the outcome of these models provides insight to the defined project objectives.

Q: Your training and experience combines both biology and statistics. What excites you about this convergence of disciplines?

A: The two disciplines can certainly learn a lot of from each other. It will be really exciting to see what methodologies biologists adopt from those that are on the cutting-edge of statistics. Having a good grasp of what's going on in both fields, I would definitely say that quantitative disciplines like statistics are evolving at a much quicker pace than biology. It's up to biologists to stay on the vanguard of statistics to provide better insight into questions and problems that perhaps were previously unanswerable due to limitations in the quantitative methodology.

A hybrid approach, blending biology and statistics, seems to fit me well and I get just as excited working up a big catch of fish in the field as I do writing computer code to carry out statistical calculations. Keeping up to date in both fields will always remain a welcomed challenge and Cramer Fish Sciences is a good fit for me to do this.

Q: What role does modeling and statistical analysis play in the type of projects you work on at CFS?

A: Modeling and statistical analysis is a critical component of any project and my fellow scientists at Cramer Fish sciences rely on me to deliver this analysis in a sound and meaningful manner. This is only one small component of the many steps required to bring a project from start to completion, however. CFS has an amazing crew of field technicians that are collecting the data and extremely experienced senior scientists that are tackling the big questions. I am just one player with a well-defined role in an amazing team of scientists.

Q: What trends do you see in modeling and analysis in the field of fisheries science?

A: The way things used to be done involved simply collecting some data and fitting a line through it. If the line seemed to fit the data, you called it done. This is still being done, of course, but Bayesian Analysis now means it can be done in a more sophisticated manner. For example, the Bayesian fit to the curve can additionally incorporate uncertainty based on ancillary data or prior knowledge. The Bayesian fit to the stock-recruitment curve can reflect prior knowledge of the stock-recruitment relationship, e.g., from data of the same species in a different basin that has been updated in light of current data being used to fit the curve.

Why is this a new trend? It's really only been in the last decade or so that computers have been powerful enough to do the calculations necessary to perform Bayesian analysis. There has been and will continue to be a paradigm shift from the classical to the Bayesian approach to model fitting in fisheries science.

Another type of modeling that has only recently gained attention is Individual Based Modeling (IBM). This can be compared to cohort population models (e.g. tracking age-classes of fish throughout time). In the IBM approach, modeling is carried out at the individual level, not the cohort level. If the population in question consists of 500,000 individuals, 500,000 fish are synthetically created in the computer, all with different characteristics, life-histories and rules that define how they behave within their system that may also vary by space and time.

In addition to modeling heterogeneity in the population, IBMs can be particularly insightful about ecological processes. If the modeler can define a set of rules for individuals to follow in the population and the synthetic population seems to mimic what we observe naturally, these rules provide strong support into properties that are governing the population in question. IBMs are extremely powerful tools, but as you can imagine they are computationally expensive. More powerful computers have recently made this possible and scientists have increasingly caught onto this trend.

You may contact Tommy via email by clicking here.