Statistics on thin ice

An interdisciplinary course on Statistics, Arctic Ecology and Statistical Ecology

Greenland, 1-16 May 2023.

It is both exciting and daunting to understand and predict the ongoing ecosystem changes in the Arctic. Exciting because huge data sets are now collected by modern empirical methods, including detailed information on behavior of marine mammals together with extensive measurements on local characteristics. Information we could only dream of just a few years back. Daunting because over the past 30yrs sea ice and climate has changed rapidly in the Arctic generating unpredictable effects that are cascading through the marine food chains even affecting the top predators.

Course content

In this course, we study recent discoveries on Arctic ecology, climate changes and marine life in the Arctic, as well as modern concepts and theory behind statistical ecology and methods for doing statistical inference from biologging and environmental data. Biologging data collected from freely moving wild animals are giving rise to huge and complex data sets that require sophisticated statistical models and methodology to analyze. Combining the analysis with environmental data is essential for understanding the complex system and how it changes in the Arctic.

The course is interdisciplinary and aimed at PhD students in either statistics/data science with a keen interest in Arctic ecology and climate or in biology/geophysics with a keen interest in quantitative analysis and data science. It will be based on the participants own research data and problems.

The course content consists of:

An advanced subject (selected by the participants, individual projects) in statistical modeling of a biological system
Introduction to field work for collecting data on marine life in the Arctic
Relationships between physical environment (climate, bathymetry) and ecological systems (marine life) in West Greenland
Modelling states and developments of - and relationships between - individuals, populations and ecosystems in the waters around West Greenland
Independent literature studies
Theoretical work on models and methods relevant for data analysis
Practical work on statistical modeling, implementation and/or data analysis
Organization of a larger project and writing of reports
Written presentation of methodology, data analysis and results

Recommended Academic Qualifications:

Basic knowledge of probability theory (such as densities, conditional independence, conditional distributions and the Markov property) and standard statistical tools (such as likelihood theory, regression techniques and tests).
Basic knowledge of programming in R.

Learning outcome

Knowledge:

aspects of applied statistics
the specific subject as outlined in the description of the chosen project
basic knowledge of Arctic ecology and data collection techniques
simulation algorithms and R-packages for statistical ecological models
statistical methods for parameter estimation in statistical ecological models

Skills: Ability to

perform a statistical analysis of data from the chosen project
compute parameter estimates and construct confidence intervals for these models
interpret the output from the analyses and relate this to the scientific questions
perform model diagnostics, statistical tests, model selection and model assessment for these models
use R to be able to work with the above points for practical data analysis
independently read scientific level statistics and ecology literature
present models and methods studied in a concise way
compare sampling strategies, assess their suitability, and select appropriate strategies for given natural resource contexts

Competences: Ability to

document models, methods, implementation and/or data analysis in a coherent report
be able to prioritize efforts in the studies, the practical implementations, the data analysis and the documentation so that the reader can assess and, if necessary, reproduce the results
document that the goals of the project description are accomplished
do statistical inference and simulate synthetic data sets from these models
evaluate if a given model is adequate
discuss the relevance, reliability, validity and interpretation of empirical data and results obtained in particular contexts
evaluate empirical evidence, put results into perspective and discuss consequences in relation to sustainable management

Workload

Lectures (a mixture of class room lectures, as well as a visit at Icefjods Center in Ilulissat and visit of bræ in Ilulissat): 24 hours
Preparation of short synopsis of scientific problem and data (before course start, each student has to develop an individual scientific project): 10 hours
Data science exercises: 10 hours
Project work (with active interaction and discussion with teachers): 50 hours
Presentations by students: 10 hours
Field work (Including field trip on ship Porsild and whale tagging): 10 hours
Writing of final report (to hand-in two weeks after the course, feedback will be provided): 24 hours

Total: 138 hours