LIKELIHOOD
METHODS AND MODELS IN ECOLOGY
April 2
6, 2012
Cary Institute
of Ecosystem Studies
Millbrook, NY
INSTRUCTOR:
Dr. Charles D. Canham, email: canhamc@caryinstitute.org
Day 1
8:30 9:30 Lecture
1: An Introduction to Likelihood for
Estimation and Inference [Simple Example in
Excel]
9:30 10:15 Case Study 1: Distribution and Abundance of Tree Species
Along Climate Gradients [Canham
and Thomas 2010]
10:15
10:30 Break
10:30
12:00 Lab
1: Calculating Likelihood,
Likelihood Surfaces, and Likelihood Profiles [R Code Section
1] [R Code Section
2]
12:00 1:00 Lunch
1:00 2:00 Lecture
2: Parameter Estimation and Evaluation
of Support
2:00
5:00 Lab 2: Global
Optimization using Simulated Annealing and Genetic Algorithms
[R code
anneal] [R code
genoud] [R code optim]
[BC Sapling
Growth Data Excel Version] [BC Sapling
Growth Data.txt (right click and Save as
)]
Recommended
Reading for Day 2:
Day 2
8:30 9:30 Lecture
3: Hypothesis Testing and Statistical
Inference Using Likelihood: The Central Role of Models
9:30 10:15
Case Study 2:
Neighborhood Models Of The Allelopathic Effects Of An Invasive Tree Species
10:15
10:30 Break
10:30
12:00 Lab 3:
Likelihood Functions for Continuous and Count Data: A Plethora of PDFs
[McLaughin
1993 Compendium of Common Probability Distributions]
[Normal PDF
with non-homogeneous variances]
[Gamma and
Lognormal PDFs] [Exponential PDF] [Beta PDF]
[Poisson,
Negative Binomial, and Zero-Inflated PDFs]
12:00
1:00 Lunch
1:00 2:00 Lecture 4: Model Selection: AIC and Akaike Weights,
Multi-model Inference
2:00 5:00 Lab 4:
Model Comparison and Hypothesis Testing in a Likelihood Framework
Recommended
Reading for Day 3: (either of the 2 listed below)
Day 3
8:30 9:30 Lecture
5: Model Evaluation: How good is the best model?
9:30 10:15 Case
Study 3: Neighborhood Models of Tree
Competition
10:15
10:30 Break
10:30
12:00 Lab 5: Model Evaluation [Sample
R Code for Model Evaluation]
12:00 1:00 Lunch
1:00
2:00 Lecture
6: Analysis of Categorical and Ordinal
Data: Binomial and Logistic Regression
2:00 5:00 Lab 6: Developing your own binomial and
logistic regression models in R
[R
code Logistic regression of windthrow data] [Damagedata.Rdata] [R code binomial
regression]
Recommended
Reading for Day 4
Day 4
9:30 10:15 Discussion:
Presenting and explaining your results in the language of likelihood
10:30 12:00 Lab 7:
Detection Probability [Models
for Estimating Site Occupancy when Detection Probability is < 1 (doc
file)] [R
Code]
1:00 5:00 Lab Project: Fundamental vs. Realized Niches of North
American Tree Species Along Climate Gradients
[See the Course
Materials page for datasets, R scripts, and documentation for the project]
8:30 12:00 Lab Projects, continued
12:00 1:00 Lunch
1:00 3:00 Presentations
of Results from Lab Projects
3:00
5:00 Wrap-up Discussion: Developing your own statistical tool-kit and
philosophy