Distance sampling has become the method of choice for shipboard surveys of cetaceans and aerial surveys of large herbivores, thanks largely to this dedicated team of authors. This book surveys developments to 2004.
The Introduction by Buckland and Anderson includes a brief statement of the history and merits of distance sampling. In their enthusiasm, they resort to a dubious characterization of mark–recapture estimates of abundance, which we are told (p. 2) ‘commonly require more than 20 parameters’! Chapter 2 (Borchers and Burnham) is a lucid overview of models to appear later. Chapter 3 (F. F. C Marques and Buckland) summarizes methods for using covariates to improve models of detection probability, especially when stratum-specific estimates are needed (see Marques et al. Reference Marques, Thomas, Fancy and Buckland2007 for a more accessible treatment).
Explicit modelling of density is avoided in the general distance framework, which relies on rigorous sampling design to obtain an unbiased estimate of density in each stratum and overall. Chapter 4 on spatial distance sampling models (Hedley, Buckland and Borchers) diverges from the general framework to develop methods for fitting a density surface to line transect data. An illuminating example concerns decreasing density of minke whales with distance from the coast of Antarctica.
Perhaps the most useful chapter has little to do with distance sampling. In Chapter 5, Thomas, Burnham and Buckland provide an overview of temporal inference (population trend analysis) and associated power analyses. The data are any periodic estimates of abundance. Distance models could include time as a covariate of density, as with space (Chapter 4), but such models are missing. The contours of required sample size for detecting trend in Fig. 5.6 are inaccurate; readers should rely on the equations or their own simulations. Sensible advice is offered on the design and maintenance of long-term studies.
Chapter 6 (Laake and Borchers) has the innocuous title ‘Methods for incomplete detection at distance zero’. The authors dissect the problem energetically, leading to ‘a maze of possible methods’, most of which use multiple observers and ‘mark-recapture-distance-sampling’ (MRDS). At over 80 pages, this chapter dominates the book, but the topic is important and the coverage excellent. MRDS raises the secondary problem of correlated detection between observers, but the method of ‘point independence’ developed by these authors provides some relief.
In Chapter 7, Strindberg, Buckland and Thomas address the placement of sampling effort within the survey region. They stress the advantages of automated design algorithms using geographic information systems. Perhaps more importantly, they explain the issues (edge effects, overlapping plots, uneven coverage and probability sampling) that should be understood to use the algorithms effectively. Adaptive sampling (Chapter 8, Pollard and Buckland) provides modest gains in precision when objects are distributed very patchily. The development of adaptive methods for distance sampling is welcome, but uptake is likely to be limited.
Trapping webs and related passive distance sampling methods (Chapter 9, Lukacs, Franklin and Anderson) remain controversial, and the treatment here lacks the authority of other chapters. The ‘no movement’ assumption of distance sampling is mis-stated as ‘no directional movement’. Passive methods integrate over time, during which animals move; detection models that allow for movement would seem to be necessary for unbiased estimation.
Chapter 10 (Fewster and Buckland) revisits the logic of distance sampling and advocates a thoughtful and structured approach to simulation studies of estimator performance.
The ‘further topics’ in a final chapter (Burnham, Buckland, Laake, Borchers, Marques, Bishop and Thomas) span a wide range (sampling in three dimensions, full likelihood, random line length, search process models, combining mark-recapture and removal methods with distance sampling, point transect sampling of cues, migration counts, measurement errors, animal sign and goodness-of-fit tests). Curiously, the main selling point of distance methods, robustness to heterogeneous detection given rather weak assumptions, is saved for the very end.
In general, this is a compilation of strong research that will be mined for its many technical insights. The chapters on trend and spatial sampling are of broad relevance to field ecologists. Errors are few. Themes are tossed back and forth between authors. The standard and robust approach to spatial variation in density (design-based estimation) seems to be under challenge from model-based approaches that can increase precision (Chapters 4, 7 and 10); it will be interesting to see which prevails.
