For anyone who has lamented the low numbers of large-scale, long-term, experimental studies of habitat fragmentation, this book provides an excellent example of what can indeed be accomplished. The Tumut Fragmentation Study, which began in 1995, includes observational studies that have run for over a decade as well as experimental manipulations, integrated demographic and genetic research, and simulation modeling. Set in southeastern Australia, the investigators took advantage of a landscape that included large pine plantations, and both remnant and intact eucalypt forests. A total of 166 sites were used in the study, including 86 eucalypt remnants surrounded by Radiata Pine, 40 sites dominated by Radiata Pine, and 40 sites in large contiguous areas of native eucalypt forest. The remnant patches ranged from 10 to 60 years in time since isolation and 0.1–124 ha in size. In contrast to some of the earlier experimental habitat fragmentation studies that focused on speciesarea relationships, the primary goal of the Tumut project was to quantify the effects of landscape context. Another novel aspect of this project was that it incorporated a comparison of unfragmented ‘‘control’’ areas, a missing aspect of many fragmentation studies. The book begins with a review of the major conceptual models of landscape cover relative to biodiversity conservation. Lindenmayer compares and contrasts alternative themes used in studies of impacts of landscape change on biodiversity, starting with non-human versus human centered approaches and then moving into landcover and connectivity approaches. He reviews three pattern-based landscape models, including the island model, the patch-matrixcorridor model, and the variegation model. These models run the gamut of perspectives in assessing species responses to discrete patches versus more gradient-based models. His argument is that although the island and patch-matrix-corridor models have been widely adopted, they do not generally deal well with spatial continua and they incorrectly assume that a single classification of landscape pattern can work for all species. One of the important messages of Lindenmayer’s work is that we should not assume that all organisms perceive landscapes in the same way humans do. As such, patterns of landscape cover classified from the human perspective may not always provide a useful framework for interpreting biotic responses. A second important message is that human-defined patches may not correspond to habitat for groups of species. Thus, Lindenmayer argues that in some cases it is reasonable to aggregate species responses and in other cases responses may need to be assessed at the individual species level. He advocates for a landscape contour model where each species is assumed to have its own patterns of D. M. Debinski (&) Department of Ecology, Evolution, and Organismal Biology, Iowa State University, 253 Bessey Hall, Ames, IA 50011, USA e-mail: Debinski@iastate.edu