Spatial Analysis: A Guide for Ecologists. Marie-Josée Fortin and Mark Dale. Cambridge University Press, Cambridge, United Kingdom, 2005. 365 pp. $60.00 (ISBN 0521009731 paper).
The biosphere is spectacularly diverse and beautiful, thanks to the fact that things are not randomly arranged. Instead, the world is “patchy,” with nonuniform spatial structures. Spatial heterogeneity, manifested in various forms of interwoven patchiness and gradients, is both a cause and a consequence of biodiversity and ecosystem functioning. Underlying this tremendous heterogeneity in nature is spatial autocorrelation: Things that are closer to one another are more similar. Ecology, the study of the relationship between organisms and their environment, is replete with phenomena in which patterns and processes are spatially autocorrelated or dependent. Thus, ecology is really the study of the interrelationship between pattern and process on different organizational levels ranging from individual organisms to the entire biosphere. Quantifying patterns, then—particularly in the spatial domain—is a critical step toward ecological understanding.
Yet ecological theory has long been dominated by nonspatial perspectives, and only in recent decades have spatially explicit views begun to take a central place in ecology. This may be seen as something of a paradigm shift, which owes much to the rapid development of landscape ecology, whose goal is to understand the relationship between spatial patterns and ecological processes on multiple scales. However, dealing with spatial patterns can be conceptually complicated and technically challenging. Although various spatial analysis methods are available to ecologists, a guide to the proper use of these methods, particularly in ecology, has been elusive. Spatial Analysis: A Guide for Ecologists, by Marie-Josée Fortin and Mark Dale, is intended to fill the gap. The two authors are outstanding Canadian ecologists with exceptional background and experience in spatial statistics. Some of their previous publications on spatial pattern analysis of vegetation are well known to plant and landscape ecologists who are keen on spatial issues.
The book consists of seven chapters: an introduction, five core chapters organized largely around data types, and a conclusion. The introductory chapter is a concise but informative overview of some fundamental concepts of spatial analysis (e.g., stationarity, spatial autocorrelation versus dependence, sampling design). Commonly used methods of spatial analysis are covered in the next two chapters: Chapter 2 describes dozens of methods for analyzing completely censused population data for point patterns (the authors recommend those based on Ripley's K function and wavelets), and chapter 3 covers methods of analyzing surface patterns that are suited for sample data, as well as maps of continuous variables, including join count statistics, spatial autocorrelation indices, fractal dimension, Mantel statistics, variograms, and various interpolation techniques. In this chapter the authors discuss at length whether and how to analyze global as opposed to local spatial statistics to deal with the problem of nonstationarity in data sets encompassing large, heterogeneous areas.
Two sets of techniques for patch identification (spatial clustering) versus boundary delineation (edge detection) are compared in chapter 4. Chapter 5 is devoted to the problems of spatial autocorrelation: its mathematical nature, underlying assumptions, statistical solutions, and ecological interpretation. Chapter 6 tackles what is arguably the most challenging topic covered by this book—spatiotemporal analysis, in which autocorrelation in both space and time must be taken into account. Finally, chapter 7 concludes by highlighting several key issues discussed in the previous chapters.
The overall organization of the book makes a lot of sense. The writing is generally good, and at times enthusiastic and engaging. Many technical books on spatial statistics seem mathematically complicated and ecologically terse (dry and boring, in plain English), but this is not the case here. The authors never forget their target audience—ecologists—in the discussion of any topic, and throughout the book their numerous examples and illustrations make connections between statistical details and ecological interpretations. I also found the concluding remarks at the end of each chapter quite useful, because they not only summarize main elements of the chapter but also contain valuable pointers and thought-provoking ideas. I particularly liked chapter 5, which is probably the most informative and useful reading on dealing with spatial autocorrelation in ecology.
There are, however, a few things that seem less impressive. For example, it is hard to find any conceptual logic in chapter 6, which is more of an agglomeration of seemingly disparate methods. The somewhat superficial discussions on cellular automata and chaos seem to detract from, rather than add to, the strength of the book. There is little discussion on landscape pattern metrics, which have been widely used in ecological studies. Also, spatial scaling is a central issue in ecology and environmental science, and it would have been interesting to include some discussion as to how spatial statistics can help revealing scaling patterns and quantifying scaling uncertainties.
Spatial Analysis is a guide, not a recipe book. In today's quest for ecological understanding of spatial patterns, recipes of methods clutter bookshelves, and an insightful guide like this one is much in need. I agree with the authors that ecologists are not adequately cognizant of the issues of scale and the consequences of spatially autocorrelated data for sampling design, data analysis, and ecological interpretations, and commend them for having done an admirable job in alleviating this situation. Without any hesitation, I highly recommend this book to anyone who is interested in spatial analysis in ecology and environmental sciences. However, although only a basic grasp of statistics is assumed, the readers should not be surprised to find out that, to fully comprehend the materials covered in this book, it will require both broad knowledge in ecology and in-depth understanding of spatial statistics.
