Aquatic barriers such as dams, weirs, levees, and culverts provide many societal services. However, they are increasingly being removed for myriad reasons such as economic or structural obsolescence, public safety or liability, prohibitive repair costs, reduced return on investment, and increasing support for river restoration and species conservation. These projects can be complex to plan and execute due to many technical, logistical, and communication challenges. This short course seeks to familiarize attendees with the context for dam removal and tools available to inform these decisions. Specifically, we seek to:
The short course covers applications of Bayesian statistics in selected environmental and ecological science fields. Case studies prepared for the short course include water quality monitoring and assessment, statistical calibration in the chemical measurement process, modeling of lake eutrophication, ecological modeling in population and community analysis, drinking water safety assessment, risk assessment of invasive species, and fishery ecology. Instead of covering statistical methods in sequence, the short course emphasizes the process of model formulation, parameter estimation, and model assessment. The focus of the short course is the process of linking ecological science and statistics for developing scientifically meaningful models. Each case study includes its scientific background and the nature of the data to articulate why the proposed model is most appropriate and what the alternative models considered. These case studies highlight some conceptual difficulties in Bayesian statistics (e.g., the prior). In addition to models for different types of data, the short course also covers modern Bayesian computation using R and Stan. Annotated computer code will be distributed during the class and posted on GitHub.
Outline
Philosophical considerations
Bayes as the hero of WWII, the cold war, and science
Introductory examples
Bayesian inference basics -- what does she mean when McGrayne says "He/She used Bayes" in her book?
Case studies
Prior as a distribution across exchangeable units
Effective visuals have become a necessity to accurately communicate scientific results. Even within scientific publications and proposals, graphs and conceptual figures are often a primary component of communicating the document’s key points. As ecologists work with increasingly large datasets, designing effective visuals to communicate our results becomes increasingly challenging. In this workshop, we will use the Stoichiometric Traits of Organisms In their Chemical Habitats (STOICH) database from aquatic ecosystems to learn new tools about designing effective graphics and visualizations for large ecological datasets. Participants will work in the R computing environment to create visualizations of this data, but are also encouraged to bring their own datasets or conceptual problems to consider if they would like. While prior experience in R will be helpful, no prior experience with R is required to attend. This workshop is geared towards early career researchers who are interested in gaining a familiarity with best data visualization practices, using the R computing environment for simple to complex data visualizations, and utilizing stoichiometric data to explore patterns across freshwater organisms and ecosystems.
Learning Objectives:
1) Using a large stoichiometric ecological dataset, attendees will use code to upload and manipulate the data frame in R
2) Attendees will learn how to produce different visualizations and discuss pros and cons to each figure type based on the data being presented
3) Attendees will then work together in groups to produce and present a novel visualization of the ratio data for given variables