DTU Aqua researcher behind a revolutionary statistic approach, TMB, now presented in a new book, Spatio-Temporal Models for Ecologists. The model can be used to implement a wide range of real-world ecological analysis – and at a much higher speed than ever.
The TMB which is short for Template Model Builder has taken modelling of ecological dynamics to the next level. When TMB hit the field of ecological dynamics modeling it was like providing a Ferrari to those who before were used to riding a bike.
Now, a new book Spatio-Temporal Models for Ecologists which is out today, 26 February, offers the opportunity to learn much more about TMB and how it can be used in marine science.
"As a result, it is important for modelers to have a tool available that allows for such level of complexity while maintaining computational efficiency. TMB could therefore be that tool. "
Kasper Kristensen, Senior researcher, DTU Aqua
“For example, in chapter 10, we show how to estimate movement and habitat preference of a population from either observed counts or tagging data. These models can be used for any type of population e.g. fish, birds, humans, etc.,” says the researcher behind TMB, Kasper Kristensen.
Kasper Kristensen is Senior researcher at DTU Aqua, Section for Marine Living Resources, and he has worked out the book together with co-author and Statistical Ecologist James T. Thorson from the Alaska Fisheries Science Center.
Complex mathematical models in minutes
First of all, TMB is a gamechanger for ecologists in terms of speed. Ecological dynamics are extremely complicated, and according to the book’s preface, ecologists have often simplified analysis by describing changes in density of individuals across a landscape. Traditionally, working with ecological theory and practice and model building has been very time absorbing.
But TMB is using a minimal set of principles and numerical techniques and using this digital tool is enabling modelers to run complex mathematical models in minutes that would previously take days.
When asked about the process of developing such a tool as the Template Model Builder, Kasper Kristensen tells:
“I've been working for a long time with spatio-temporal models where everything had to be implemented from scratch, because the existing state-of-the-art tool, ADMB, Autodiff Model builder, wasn't fast enough. Developing such complex models from scratch could easily take a year.
TMB arose as a way to automate the whole development process by including all the computational tricks I had learned over the years. TMB turned out to be useful far beyond spatio-temporal models.”
A basic tool
The Template Model Builder can be used to implement a wide range of real-world ecological analysis regarding animal movement, population dynamics, community composition, causal distribution, and spatial dynamics.
The R-package TMB developed by Kasper Kristensen has seen increased use throughout ecology.
TMB is the computational backend for R-package glmmTMB which has been cited over 3.000 times from 2017-2022.
In fact, TMB is a basic model that is being used all over not only in relation to modeling of marine ecosystem dynamics and fishery.
In the book, the authors provide a step-by-step illustration of techniques and analyses conducted using a minimal toolbox and fast computation using Template Model Builder.
How will TMB change the future of working with ecological dynamics, not in the least if you think of how society can benefit from its power?
“Statistical models of real-world phenomena use available data to extract useful and interpretable information with realistic uncertainties,” says Kaser Kristensen and looks ahead:
“In the future, these models will become more realistic and therefore more complex. As a result, it is important for modelers to have a tool available that allows for such level of complexity while maintaining computational efficiency. TMB could therefore be that tool. Beyond fisheries and aquatic sciences, TMB has been used in many other fields including health science, pharma, astronomy, wind energy prediction, etc.” Kasper Kristensen concludes.
Spatio-Temporal Models for Ecologists introduces TMB through 273 pages and 12 chapters of showcasing techniques using real-world data and providing a toolset for hierarchical modeling of spatio-temporal processes. According to the preface, the book is meant for graduate level students and applied and academic ecologists.
Illustration: three times cover of the book