Cave Life of Oklahoma and Arkansas: Exploration and Conservation of Subterranean Biodiversity. G. O. Graening, Danté B. Fenolio, and Michael E. Slay. University of Oklahoma Press (Animal Natural History Series), Norman; 2011. ISBN 978-0-8061-4223-4. 6 by 9 inches, 226 pages, hardbound. $59.95.
This small but expensive book is sort of a hybrid between an introduction to cave biology and its conservation in the area and a formal contract report for the Subterranean Biodiversity Project. A casual reader can get a pretty good notion about the principles of cave biology from parts of the text and the color photos, but he'll have to put up with an awful lot of pedantry and pseudo-science along the way, because the book is very heavily biased toward the report aspect. The authors have compiled an extensive record of animals seen in caves in Oklahoma and Arkansas, with 1355 taxa listed, 690 to the species level, in Appendix A. Much of the data resulted from generally brief visits to a large number of caves, where eyeball searches were used. But a considerable amount was obtained from extensive surveys of literature, from scientific papers to caving-club magazines. The authors recognize that this has resulted in a rather unsystematic database of a pretty random collection of observations, but that doesn't discourage them from applying lots of statistics. The actual scientific value of the book is the list of fauna and the caves in which they were observed, which in principle makes it possible to at least create distribution maps. However, that won't be easy in practice, because they've elected to put the distribution data in Appendix B, which is the list of caves and the serial numbers of the taxa in Appendix A that were seen in each of them. That means that to find out where a given species has been found one must search for its number throughout that fifteen-page Appendix B. It would have been a whole lot better to number the caves, not the taxa, and list the cave numbers for each taxon in Appendix A, with just the names (or, often, just cave-survey numbers) of the caves in numerical order in Appendix B.
The authors seem to think they were being paid by the number of literature citations they could cram into the text, and so the innocent reader is subjected, for example, to numerous citations for things that are common knowledge about biospeleology and can be found in any introduction to the subject. It's a rare paragraph that doesn't have several intrusive citations. Some pedantry, such as a half-page list of the collecting permits the project had, is easy to skip over, but then there are things like the information that they used "Access 2007 (Microsoft Corp., Redmond, Washington)." Who cares what database they used? Who else makes Access? How many Microsofts are there?
The pseudo-science comes in when the authors apply statistical techniques to their data, despite its acknowledged limitation and biases. For example, for each site they recorded qualitative data such as how extensively it is visited, lightly, moderately, or heavily. Then they applied a statistical test to see whether this "affects" species richness. In this case, they find that the most heavily visited caves have the greatest biological diversity, to their surprise, but this is just because cavers prefer to visit longer caves. Correlation is not causation. They fit curves to scatter plots of things like site richness versus site length, even though there is no theoretical reason to expect the data to fit that particular form of equation. In one case, they fit both linear and exponential functions to the same data, displaying the best-fit coefficients to four allegedly significant digits with no confidence intervals. Both fits give p < .0001. What p is that? I doubt the authors know; it just fell out of the software. The mathematical qualifications of the authors may be judged by the statement that the number of taxa found at a site tends to increase exponentially with the number of specimens collected.
In truth, there is a good bit of useful information buried in this book, and I suppose even a lay reader who is not as easily annoyed as I am could learn some things from it. But I shudder to think of the graduate students who will accept this book as a good model for their theses and dissertations. It is an excellent example of what happens when somebody carelessly leaves statistics software lying around where anybody can get at it.--Bill Mixon