Predicting extinction risk

Article metrics

Population Viability Analysis

Edited by:
University of Chicago Press: 2002. 577 pp. $95 (hbk); $35 (pbk)

Population viability analysis (PVA) is an important tool in conservation biology. It is the process of predicting the risk of extinction from the combined effects of the deterministic threats — such as habitat loss, overexploitation, pollution and species introductions — and stochastic threats, including demographic, environmental and genetic fluctuations and catastrophes. This is typically done using stochastic computer simulations. PVA is also used to compare alternative management options designed to help threatened species recover. The technique arose in the 1980s but is based on the accumulated knowledge from more than a century of research in demography, ecology and genetics. In some ways it is similar to weather forecasting and the modelling of economics and global climate.

Despite vigorous activity in this field, there have been few overviews of it. This long-awaited book, the authors of which are a veritable who's who of PVA, fills the gap. The book opens with an overview of PVA before moving on to consider the construction of PVA models, how to integrate theory and practice when using PVAs, and finally the future of PVA. The focus is primarily on animals throughout.

The book succeeds in reflecting the breadth of the field, the diversity of opinions about PVA, and its use in conservation biology. The contents are of variable quality, as is common in edited volumes. I found much to applaud and much to disagree with. The highlight for me was the chapter by Mark Shaffer and colleagues on PVA and conservation policy. It was incisive and thoughtful, and had a useful practical perspective, as befits a contribution from a founder of the discipline.

By contrast, the treatment of genetics throughout the book is sometimes dubious. However, the chapter by Fred Allendorf and Nils Ryman provides a thorough, authoritative survey of the role of genetic factors in PVA. Several chapters refer to the problems of modelling genetic factors in PVA, but these apparent difficulties are largely illusory. VORTEX software can model inbreeding depression well for juvenile survival, and well-known functions can be used for other aspects of the life cycle. Furthermore, the chapter by Sue Haig and Jon Ballou describes work that encompasses inbreeding depression, as do other published papers. There are limited data for parameterizing inbreeding depression, but even that problem is slowly being addressed.

The book arose out of an international symposium in 1999, but many authors have updated their contributions to include more recent material. For example, our subsequent paper (Nature 404, 385–387; 2000) on the predictive accuracy of PVA is referred to by several authors, although some describe its methods and contents incorrectly.

I was disappointed by the perspective provided by the book, despite there being many fine contributions. For example, the relationship of PVA with other related fields, which offer useful methods, analogies and insights, receives only cursory attention. The context of conservation biology as a crisis discipline, in which immediate decisions must be made on the basis of inadequate data, seems to have been overlooked in many contributions. And several authors refer to alternatives to PVA but fail to point out that the only realistic alternative, human judgement and intuition, is inaccurate.

Deficiencies in data are a serious problem for PVAs of endangered species, as several authors point out. They discuss various remedies, including eliminating problems in data collection, separating the effects of sampling variation and intrinsic variation, and the use of bayesian methods to incorporate uncertainties about parameter values into predictions. However, the use of data from related species receives little attention. Such data may be particularly valuable for variances of input parameters. For example, environmental variation is similar across a range of herbivore taxa (J.-M. Gaillard et al. Annu. Rev. Ecol. Syst. 31, 367–393; 2000). Furthermore, analyses of long-term data sets to estimate the frequency and severity of catastrophes and to ask whether they differ widely across taxa would be highly desirable.

The way to improve PVA is clear, but receives little attention here. It involves cycles of building models, making predictions, testing them, improving the models, and so on, as for all fields involving complex systems. The emphasis in PVA has been on building models, but the testing phase has only just begun. Much more testing is required if the field is to advance swiftly and rationally, as is happening with climate modelling.

Despite some limitations, this volume should serve as a major reference book on PVA for professional scientists, advanced undergraduates and graduate students in conservation biology.

Author information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Frankham, R. Predicting extinction risk. Nature 419, 18–19 (2002) doi:10.1038/419018a

Download citation

Further reading

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.