Modeling Extinction. Newman, M. E. J., and R. G. Palmer. 2003. Santa Fe Institute Studies in the Sciences of Complexity. Oxford University Press, Oxford, United Kingdom. 114 pp. (102 + xii). $25.00 (paperback). ISBN 0–19–515946–2.
For the past 20 years or so, the paleontological study of extinction has focused on three general issues: (1) causes and biological consequences of mass-extinction episodes such as the end-Cretaceous event 65 million years ago; (2) selectivity in extinction—for example, the preferential survival of geographically widespread genera; and (3) secular patterns and other large-scale statistical properties of the historical record of extinction intensity. In this last area, paleontologists have been concerned largely with why average extinction intensity has declined over the course of the Phanerozoic Eon (roughly the past half billion years); whether mass-extinction episodes are periodically spaced in time, and if so, why; and the extent to which mass extinctions represent a simple extension of normal or “background” extinction, as opposed to a separate statistical class of events with distinct causes.
Since about 1990, a number of scientists from outside paleontology, many of them physicists and many associated with the Santa Fe Institute, have also contributed to the study of large-scale statistical properties of the extinction record. To some extent this work involves data analysis, but it is mainly focused on erecting mathematical models that attempt to explain a range of complex observations with a few simple rules. The book by Newman and Palmer, who are trained in theoretical physics and have taken an interest in biological extinction, provides a clear and concise synopsis of this large body of work—in general much more lucid than the original studies themselves.
The observation that most models have tried to replicate is the supposed tendency for the magnitude of extinction events to follow a power-law distribution, in which the frequency of events of a given magnitude is inversely proportional to the magnitude raised to some exponent. Put another way, a log-log plot of frequency against magnitude is said to yield a linear relationship whose slope is the characteristic exponent.
Competing models that have attempted to explain the power-law relationship differ in the extent to which they incorporate biotic interaction, adaptive evolution, and environmental stress, and in whether they imply self-organized criticality—a property that leads to catastrophes even in the absence of perturbations. Self-organization in this context produces a power-law distribution, but Newman himself has shown that such a distribution can result without self-organization.
Given that the models reviewed in this book concern questions of great importance to paleontology, it is natural to ask why they have had very little tangible influence upon paleontological research. I think one can reject the obvious possibility that paleontology is somehow immune to contributions by outsiders. Take the Alvarez hypothesis, for example. Although the idea that an asteroid impact triggered the end-Cretaceous mass extinction met some resistance when it was first proposed—and still does—the general importance of impacts and other extraterrestrial phenomena is now part of the paleontological consciousness.
One can also rule out the notion that the models are too unorthodox. For one thing, many of the models involve rather conventional assumptions (for example, that species evolve in response to selection pressure). More to the point, paleontology and evolutionary biology have in fact been shaped by unorthodox models—for example the classic MBL work of the early 1970s, which explored the possibility that evolutionary trends, differential diversification, and other aspects of macroevolution may result from stochastic noise superimposed on a simple branching process of evolution.
Instead, I see three principal reasons that the work reviewed by Newman and Palmer has not significantly affected paleontology or any other mainstream field of biology or geology. First, it is not at all clear that the fossil record in fact documents the phenomena that the models are supposed to explain. As Newman and Palmer show in their first chapter, extinction data may be consistent with a power-law distribution, but there are substantial deviations between model and data, and other distributions may fit even better. The model-data agreement has often been assessed subjectively, and fairly loose standards have been used to determine a good fit. By contrast, before the hypothesis of periodic extinction was taken seriously, rigorous standards of what constitutes an extinction maximum and a periodic signal were set forth. Until similarly tight standards are used in assessing other statistical properties of extinction, it is unlikely that the associated models will be influential. It is also conspicuous that most of the work reviewed by Newman and Palmer has ignored a number of important technical problems: for example, how to measure extinction intensity, how to correct for paleontological incompleteness, whether to detrend the data to factor out the long-term secular decline in rates, and so on.
Second, as Newman and Palmer state, details of extinction mechanisms are not always clear in the models. It is one thing to talk about a “coevolutionary avalanche,” for example; it is quite another to state what is actually happening to biological populations. It is of particular relevance to conservation biologists that the models generally ignore geography.
Last and I think most important, the models discussed in this book are, almost without exception, aimed at explaining an existing observation rather than generating new, testable predictions. The importance of novel predictions is especially critical because, as Newman and Palmer emphasize, many of the models, although quite different in their assumptions, agree with existing data equally well (or poorly). Thus there is little statistical (let alone biological) basis on which to distinguish them.
Modeling Extinction is worth reading as a clear overview of one approach to understanding extinction in Earth history. The authors are in a rare position: they are familiar with this approach from the inside and are sympathetic to its aims. At the same time, they are critical of its shortcomings. Finally, unlike many of the contributors to this body of work, they seem to understand paleontological data. This all makes for a book that has real potential to move the modeling effort forward.