Occupancy Estimation and Modeling.Darryl I. MacKenzie, James D. Nichols, J. Andrew Royle, Kenneth H. Pollock, Larissa L. Bailey, and James E. Hines. 2006. Elsevier-Academic, San Diego, California, xviii + 324 pp., 23 figures, appendix. ISBN 0-12-088766-5. Hardback, $64.95.-Species presence and absence (i.e., occupancy) data are increasingly being used by avian biologists to assess the status, distribution, and dynamics of bird populations (e.g., Oison et al. 2005, Tornberg et al. 2005, Karanth et al. 2006) and for developing conservation strategies (e.g., Freemark el al. 2006, Jiguet and Julliard 2006). Unfortunately, complete detection of a species is usually impossible, and the ability to detect a species is often related to species-specific traits and the physical characteristics of sample units (reviewed in Thompson 2002). Consequently, incomplete detection can bias occupancy estimates and impede the ability to make sound conservation decisions. Several methods have recently been developed to incorporate incomplete detection in occupancy models (MacKenzie et al. 2002, 2003, 2004; Dorazio and Royle 2005; MacKenzie and Royle 2005). These pioneering efforts, however, have been presented as separate works and often in a manner that was difficult for all but the most technically savvy to understand. This book is an attempt to synthesize existing ideas on occupancy estimation and modeling in a form that is understandable to biologists and ecologiste without strong statistical backgrounds.

The book is a well-organized and comprehensive treatment of occupancy estimation that entails multiple aspects including sample design, analysis, and interpretation. The first three chapters cover basic ecological and statistical background and introduce terminology and concepts that are used throughout the book. Chapter 1 is a philosophical treatment of the nature of science and management and the role of field surveys and monitoring. This philosophy is reflected in much of the material presented throughout the book. We believe that such context is important and often lacking from statistics-oriented texts. Chapter 2 provides an overview of the ecological aspects of occupancy and includes a description of metapopulation dynamics. Although the chapter is not intended to be a thorough review, the authors have done a commendable job compiling and synthesizing an abundance of information on metapopulation dynamics as it relates to occupancy estimation. Chapter 3 is an excellent and thorough review of the basic principles of statistical estimation and inference that should prove useful for professionals and for graduatelevel instruction. The chapter thoroughly details all aspects of parametric statistics: maximum likelihood estimation, hypothesis testing, goodness-of-fit, and model selection. However, none of these topics is covered in relation to Bayesian methods. As such, readers will have no basis for evaluating the goodness-of-fit, convergence, and selection of Bayesian models. Yet later chapters include computer code for fitting Bayesian occupancy models. We believe that this is a potentially hazardous combination and hope that the authors can remedy the problem in future editions.

Chapters 4-7 cover single-species occupancy estimation and gradually build from relatively simple, constant detection-probability estimation (chapter 4) to more complex, multiple-season models (chapter 7). Each chapter begins with a useful general introduction and explanation of purpose. Models are then derived in logical sequences, with sufficient mathematical details and clear, concise explanations that should satisfy and enlighten biologists, whatever their level of statistical proficiency. Each chapter contains at least two examples that are used to illustrate model fitting, parameter estimation, and the presentation and interpretation of results. The material in this section of the book is very thorough and is generally presented in a logical sequence. Chapter 6, which covers the design of single-season occupancy studies, includes a thorough evaluation and discussion of factors that are crucial for developing efficient and effective occupancy studies (e.g., study site selection, allocation of sampling effort). However, the chapter is probably of limited use for developing monitoring designs. Chapter 7 (multiple-season models) provides useful study-design guidance that is relevant to monitoring (e.g., the limitations of a rotating panel design), but lacks detail on statistical power. A general treatment of study design that included the details in both chapters 6 and 7 would have been preferable.

Chapters 8 and 9 deal with two ways to investigate multiple-species occupancy patterns: (1) interactions among a small number of species (chapter 8) and (2) changes in species richness (chapter 9). As the authors acknowledge in the introduction to chapter 8, these two chapters are not as well developed as previous sections of the book, providing few, if any, examples for each method. The lack of associated software and example code in this section, with the exception of the two-species interaction model implemented in PRESENCE, version 2 (Hines 2006), will limit the use of these methods to statistically savvy readers with knowledge of computer programming. Despite the lack of implementation detail and the paucity of examples, the authors do an excellent job, as in previous sections, in presenting the material in a logical order and clearly deriving and explaining all models in a way accessible to all biologists. Analysis of occupancy data at the community level is also a very active area of research, and we expect user-friendly software implementing many of the methods described here to become available in the near future.

It contains short sections detailing possible models and approaches (most of which are untested) for multipleoccupancy states and integration of habitat, abundance, and marked animals in occupancy estimation.

Most of the examples in the book were analyzed using PRESENCE, which is freely available via the internet and includes copies of the example data sets. However, the book is neither intended nor appropriate for instruction in using PRESENCE. Several examples also include code for fitting Bayesian models with WINBUGS, version 1.4 (Spiegelhalter et al. 2003). Their use will require familiarity with the software and Bayesian analysis methodology.