Analysing Ecological Data (Statistics for Biology and Health) [Hardcover]

Statistics are a tool of increasing importance in ecological research, and consequently there are already a wide range of books available on this subject. However, in my view "Analysing Ecological Data" by Alain Zuur and collegues is a novel and much needed contribution to the existing literature. The authors bridge the gap between books on statistical theory and books on applied statistics for ecologist. While the former are often difficult to understand without a mathematical background, and rarely include examples that deal with the type of data that we face in ecology (non-normal distributions, auto-correlations, etc.), the latter tend to provide few details on the underlying maths. Featuring both a theoretical and an applied case study part, "Analysing Ecological Data" is a uniquely well-balanced mixture between the two.

The book is well written throughout. The statistical theory is presented in such an accessible way that it makes thirst for more. For a large number of techniques the reader is provided with sufficient information to calculate the statistics by hand, and sufficient enthusiasm to even do so! The eighteen case studies cover a wide range of statistical problems and are based on real data. Guiding the reader step by step through the processes of data exploration, model selection, validation and interpretation, these practical examples constitute extremely useful templates for data analysis.

Consequently, I would highly recommend "Analysing Ecological Data" to anybody who does not like to work with "black boxes", and who seeks to employ a thorough and well structured approach to data analysis. The book is almost perfect. There are only two things missing: First, the R code for the practical examples (which however might be uploaded on the authors' website soon); and second, more of the same!

Finally a book that explains statistics at an understandable level! It is a great book for biologists, ecologists, etc. The book also contains 17 case studies (all about 20 pages), using real data. These are very useful.

`Lies, damn lies and statistics' is the oft-used quote and in this context it should read: ecology, ecological statistics and appropriate ecological statistics. Ecologists are notoriously passionate about their subject and most, if not all, use statistics to support their field and laboratory studies. Very few ecologists are statisticians and although many of us have had some statistical training as part of our first and higher degrees, few can claim a high level of statistical expertise with a great deal of confidence. As a result there is a fair number of books written to help ecologists (and biologists) select and use appropriate statistical techniques. These range from books examining the basis of experimental design (e.g. Underwood) to those that are written to support particular statistical programmes (e.g. SPSS, or Canoco). This book `Analysing Ecological Data' is a new addition to what is a fairly competitive market place. The book has been written by Alain Zuur (and his colleagues) at Highland Statistics is a guide to using `appropriate' ecological statistics: those that start with your research question and guide you through data interrogation, normalization (or not), tool selection and interpretation.

Although written with a view to supporting Brodgar (their own statistical programme) the authors stress that any of the techniques discussed in the book and all the accompanying example datasets (housed on a website) can be analysed using R a freely available and extremely powerful, if not a little intimidating, statistical library and scripting language. This is certainly true but researchers with only limited to intermediate statistical knowledge would certainly benefit from using Brodgar itself as the windows (menu driven) approach with its hard-coded R scripts (plus a range of programmed techniques) makes for a much easier introduction to the topic.

The book has thirty-seven chapters loosely structured into three main areas: the first few chapters (four in total) provide introductory material on statistics, data management and exploration. These are essential first steps for any researcher which should be completed before selecting which analytical techniques are available and appropriate for their field or experimental data. Are your data linear? Are they normally distributed? Are there any outliers? These are important questions and have implications for the sorts of techniques you can legitimately select for the data crunching later on. The chapter on exploration (Chapter 4) illustrates a range visualisation tools that are available within the R statistical library and it is especially useful. My experience is that this is an often overlooked step in the research process so it is very positive to see it so heavily emphasized in this book.

The second section introduces readers to the multitude of techniques available to the willing ecologist, starting with Linear Regression developing through to complex Generalised Linear Models (GLMs) and their non-parametric cousins General Additive Models (GAMs) and concluding with Mixed Models. Thereafter, ordination techniques (e.g PCA, CA, CCA etc), time series analysis and spatial analyses are discussed in a similarly easy to grasp manner.

There is a strong sense of development throughout the book and the third element of the mix introduces the reader to a sequence of case studies of the use of many of the aforementioned techniques. The datasets themselves are available for download and experimentation on the website that supports the Brodgar programme (and now the book). The case studies are an excellent addition and reinforce the preceding chapters that guide readers in their selection of appropriate ways to analyze datasets from a variety of sample designs with markedly different structures.

This is first statistical text I have read (I have many on my shelves) that takes the reader from the starting point of sample design, through data visualisation to technique selection, all supported by case studies using real field data. It does this in a clear and systematic manner and is well-written throughout. These early chapters are excellent and having read them I felt much more secure and confident about the quality and structure of my own datasets - it was like having a light switched on in a dark room. These opening chapters are a real strength to the book making it a `must read'. The other major strength is the ground that the book covers - it is very wide-ranging and draws a much broader range of univariate and multivariate techniques than most of its competitor texts. If you want more confidence in your own analyses or just a deeper understanding of the statistical techniques you routinely use (and others you could use) this book is for you. I cannot praise it enough; copies are doing the rounds in my research team as I type.