Books on ecological modeling/statistics/R, hopefully with comments and ratings (???) from a broad spectrum of readers/users at varying levels of expertise …

[TWS] These comments were provided by Tyler Smith. Quick sketch of my background - postdoc, working in community ecology, PhD in systematics. Grad-level training in multivariate stats, a couple of undergrad courses in 'biometry'. Current interests include maximum likelihood techniques and mathematical modelling, but I'm far from an expert in these topics.

[ID] These comments were made by Ian Dworkin. I am an assistant prof at Michigan State University, and I teach a graduate level course in statistical modeling for ecology and Evolution using EMD and some other texts, in addition to numerous papers from the literature.


  • Gotelli and Ellison (A Primer of Ecological Statistics)
    • [TWS] An excellent overview of the most common statistical approaches used by ecologists. I wrote a full review when I was a grad student.
    • [ID]While I agree with the above statements, Gotelli and Ellison is mostly useful in teaching statistical concepts (and it does a very good job with respect to thinking about inference). However, it is less useful for either the mechanics or computational side of basic statistics. It is very useful in combination with Dalgaard (described below) as an introduction.
  • Dalgaard (Intro Stats with R)
    • [TWS] A very nice first book to start learning R with. Doesn't assume much computing experience, and walks you through using R to do the sorts of analyses covered in undergrad 'stats for biology' courses. Probably best as a companion to an intro stats textbook, but it does explain the statistics implemented as well as the code.
  • McCarthy 2007 (Bayesian methods for Ecology)
    • [BMB] This is a really clear introduction to Bayes and WinBUGS. I don't agree with all of it, and a lot of the things he discusses can actually be done with maximum likelihood as well, but as a basic intro it's really good. Interestingly, McCarthy primarily uses Bayes/WinBUGS as (1) a way to model ecological systems flexibly and (2) a way to incorporate prior data into conservation biology problems with extremely sparse data. He barely touches the reasons that many ecologists are using Bayesian methods (multi-level/mixed/hierarchical/state-space models that are too difficult to estimate in other ways).
  • Crawley ? 2005 (Statistics: an intro using R)
  • Hilborn and Mangel (Ecological Detective)
    • [TWS] Covers much the same material as EMD, but with more emphasis on big-picture concepts, less emphasis on how you actually implement the analyses. Having read this first, I find EMD is the lab manual that fills in a lot of the blanks from Ecological Detective. In other words, ED gives tells you why, EMD tells you how. I've posted R code for most of chapters 3 -7 on the R wiki
  • Verzani?


  • Faraway regression books? (I don't know these very well: Linear Models with R, Extending linear models)
  • Gelman and Hill (multilevel modeling)
    • [BMB] very clear, practical advice. Bayesian focus. Since their subject areas are more in politics and public policy the data tend to be large and normally distributed, but there are some categorical-data examples as well.
  • Quinn and Keough
    • [BMB] very clear and thorough, with broad and sensible coverage, but a bit more of an "ANOVA table" approach to ecological statistics than I might like
  • Crawley 2002 (Statistical computing)
  • Clark 2007 Models for Ecological Data
    • [BMB] thorough introduction to Bayesian methods, much more coverage of population dynamic models than I give in EMD, but so comprehensive as to be a little daunting for first-time readers
  • Zuur et al books from Highland Statistics


  • Harrell "Regression modeling strategies"
    • [BMB] biomedically oriented. Very rigorous but still practical advice on parameter reduction, etc.
  • Gelman et al Bayesian Data Analysis
  • Venables and Ripley MASS
    • [TWS] This is by far the most consistently recommended book on r-help. It is very terse, and I think primarily aimed at people who already know the technique they want to use, and just need to see how it's done in R (kind of like r-help that way). So, if you are already have a sound understanding of the stats you're using, this might save you some time getting R to do what you want. If you need more background on the stats, you'll need to look somewhere else.
  • Pinheiro and Bates 2000
    • [BMB] until such time as Bates writes a new book on multi-level models that covers lme4, this is the book on mixed models in R. No GLMMs, though — only linear and nonlinear (i.e., all normal) mixed models.

special topics

  • graphics
    • Murrell (R Graphics)
    • Sarkar (Lattice)
  • multivariate analysis:
    • Everitt 2005
    • Manly
  • survival analysis:
    • Therneau and Grambsch
  • GAM: Wood
  • time series: Chatfield, Diggle
  • spatial: Cressie (general), Diggle (point processes), Diggle et al. (model-based geostatistics), Christakos (geostats)

and what about …

  • Maindonald and Braun?
  • Royle?
  • Spector (data manipulation)?
Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License