Organic Geochemistry and Carbon Cycle
This course will consist of 8 sessions where topics in organic geochemistry and the carbon cycle will be discussed. Sessions will include an hour of lecture and 40 minutes of discussion of a paper on the topic. Lecture topics will include:
Global carbon cycle; Dissolved organic carbon; Particulate organic carbon and carbon export; Decomposition of organic matter; Preservation of organic matter
Alkenones as biomarkers; Lignins as biomarkers; Black carbon; Other topics may be substituted depending on student interest.
Discussions of papers from the literature will cover the scientific questions being asked in the paper, how the authors approached the problem, whether the results are reasonable, what further research could be done to answer unfinished questions, etc.
The final exam will be a short oral presentation by each student on a topic of their choice.
Statistics and Experimental Design
Progress in quantitative understanding of scientific processes requires three components: (1) quantitative statements of the set of hypotheses that are to be explored (“models”); data sets that are sufficiently complete that they can be used to determine which of the set of models is “best supported” by data; and (3) a systematic way (“statistics”) to allow the models to be compared to one another given the data. Statistics is a powerful tool for deriving maximum information from data, for assessing which competing models best explain observed data, and for summarizing and presenting the results obtained. But to take advantage of this power, data must be gathered in such a way that measurements are “independent” and as free as possible from “confounding factors.”
This course has three major goals: 1) to develop an appreciation for sampling strategies and experimental designs that promote the statistical independence of data and avoid confounding factors; 2) to review the concepts of “standard” statistics: means, variances, least-squares methods (both linear and nonlinear); various statistical distributions (Bernoulli, binomial, Poisson, Normal, lognormal) and their normalization constants; 3) then proceed to more modern statistics based on Bayes’ theorem and likelihood, and requiring computer programming. Examples will include both theoretical exercises and “real world” examples where likelihood methods have been applied to oceanic problems, emphasizing work by the instructor. Methods for plotting data (most notably box plots and nonparametric tests to distinguish them) will also be discussed.
Requirements: class participation; homeworks, where students will be asked to present their results to their fellow students in class to encourage discussion and exchange of ideas; and a final exam.
Hilborn, R., Mangel, M. 1997. The ecological detective. Confronting models with data. Princeton Univ. Press.
A very understandable exploration of fundamental concepts of statistics as a powerful method for connecting scientific ideas to data. Approx. $50.