Harvard Forest recently hosted a 5-day workshop for faculty and graduate students interested in learning to work with generalized linear mixed models, a statistical approach for estimating different kinds of variation in ecological processes. Mixed models are particularly useful for structured data, such as blocked experiments with low replication, or monitoring data, in which researchers want to separate observation error from environmental variation.
Workshop participants included eight professors from the University of Massachusetts, Colby College, Clayton State University, University of South Carolina, University of Minnesota, Wilkes-Barre University and Adam Mickiewiscz University (in Poland,) and six Ph.D. students from Harvard University, University of Massachusetts, University of Pittsburgh, University of Washington, and University of Bergen in Norway.
Many of the participants were studying effects of ungulate browsing, small mammal granivory (seed predation), and the impacts of climate factors such as temperature and light on forest establishment. Another set of participants was interested in the specific problem of separating observation error from environmental stochasticity (randomness) in demographic time series. The workshop was led by Harvard Forest ecologist Elizabeth Crone, with help from postdoctoral researchers Josh Rapp and Leone Brown.
We encourage LTER researchers interested in learning about mixed models in ecology to watch for an announcement for the 2014 course in September, or to contact Dr. Crone for more information.