112 interchange among African, North American and European scientists. Although I was somewhat disappointed to see the reduced number of African authors in this book as compared to the first IDEAL volume, it is clear that this decade-long consortium has resulted in significant advances not only in the science of East African lakes research but also in the international recognition of the quality scientists of East Africa. In fact, this is the first of the three volumes that is edited solely by African scientists. Although some of the chapters in this book are regionally specific and their findings of limited scope, many of the chapters draw conclusions that are relevant and important to understanding modern lake variability and response to disturbance in general and to the broader role of the tropics in the global climate system. Limnogeologists, climatologists and biologists alike will find this book a useful and important general contribution to global change research. For those working along the Europe–Africa Pole– Equator–Pole (PEP III) transect, it is essential.

KRISTINA BEUNING Department of Biology University of Wisconsin – Eau Claire Eau Claire, WI 54701, USA (e-mail: [email protected])

Time-Series Analysis and Cyclostratigraphy, by G. Weedon. 2003. 259 pp. Cambridge University Press, Cambridge. ISBN: 0-521-62001-5. Price: USD 70.00; GBP 50.00. Time-series analysis has became a widely accepted tool for studying quantitative records of environmental changes from both ocean and terrestrial sediments, ice cores, speleothems, growthbands from corals, trees and molluscs. The investigation of such environmental cycles from sedimentary sections has been called cyclostratigraphy. During the last couple of years, many papers have been published dealing with the theory of time-series analysis methods and their application to paleoenvironmental data. Also a number of books appeared introducing and comparing these tools and describing theoretical and real-data examples (e.g., Schwarzacher 1993, Muller and MacDonald 2000). Moreover, a large number of commercial (e.g., Matlab, Origin, SPSS) and free software packages (Arand Software, AnalySeries, Spectrum)

including comprehensive user instructions and tutorials are available. The author of the new book Time-Series Analysis and Cyclostratigraphy, Graham P. Weedon received a D.Phil. from Oxford University in 1987 and is now Senior Lecturer in Geology at the University of Luton. His research involves studying Silurian to Recent sedimentary cycles from various paleoclimate archives. In the past, he has participated in several Ocean Drilling Program (ODP) cruises, off Oman (1987), Brazil (1994) and New Zealand (1998). On ODP sediment cores the emphasis of his work was on orbital-climatic forcing, whereas over the last few years he also worked on El Ni~no/Southern Oscillation-type cyclicities, millennial-scale cycles and even tidal cycles. The emphasis of the book according to Graham Weedon’s preface is to explain concepts, procedures and problems of time-series analysis and cyclostratigraphy, not the details of the mathematics. The author has decided to avoid equations and derivations and employs simple diagrams in the explanations. Personally I totally agree with this concept. However, in some cases the author exactly describes an equation in the text (e.g., pp. 65 and 90). Without having the equation, this is extremely hard to read and rather confusing. On p. 154, Weedon breaks with his own rules and provides a simple routine to establish how long a time series should for detecting long cycles. I believe that the readership of earth scientists would accept a few more of such routines in the text. The diagrams are excellent in most cases although the large fonts used to label the graphs often reduces clearity (e.g., on pp. 56, 66 and 67). The author used a similar layout for all graphs with stratigraphic position or time running up the page. All diagrams have extensive captions that allow them to almost stand alone. In some cases, this concept causes rather large pieces of text attached to a relatively small figure (e.g., pp. 56 and 86). However, figures and captions indeed guide the reader very well through the book. I have found only minor errors (e.g., Figure 4.4 is very confusing because of the thin vertical lines on the right suggesting a high-frequency limit of the filters) or discrepancies between graph and text (e.g., the numbers in caption of Figure 1.4 compared to the main text). The book has six chapters focussing on a theoretical background of the various time-series analysis methods and practical applications in cyclostratigraphy. The book starts by introducing basic concepts in

113 analysing cyclic sediments followed by a chapter explaining the construction of time series in cyclostratigraphy. Chapters 3 and 4 provide a comprehensive explanation of the most relevant and widely used methods of spectral estimation and other tools for time-series analysis and signal processing. For newcomers in the field, the first two chapter might be hard to read because many of the topics explained in the following parts of the books are actually required to understand phenomena such as the meaning of the Nyquist frequency and aliasing. This problem is also reflected in several cross references with Chapter 3. Unfortunately the book contains more examples of a confusing organization. Chapter 4 is entitled Additional Methods of Time Series Analysis, containing a mixture of spectral estimation methods I would shift into Chapter 3 (e.g., cross-spectral analysis and evolutionary spectra), since they use the same mathematics. The chapter also introduces alternative linear and nonlinear techniques of time-series analysis (e.g., wavelets, Lyapnov exponents). Finally, Chapter 4 gives an extremely brief introduction to filtering, which is not a method of time-series analysis at all, but a signalprocessing technique. Interestingly, this seven-page Chapter 4.3 Filtering as part of a 259-page book on cyclostratigraphy illustrates very much the imbalance in environmental research between the application of time-series analysis and signal-processing techniques. Whereas sophisticated methods of spectral estimation are now widely used to detect natural cyclicities, earth scientists are often not familiar with methods for deconvolving signal distortions and disturbancies. Chapters 5 and 6 contain an extensive compilation of examples and practical considerations. These chapters certainly represent the strongest part of the book by G. Weedon. An enormous review of the relevant literature, numerous examples from Weedon’s own projects as well as classic references from other workers illustrate all aspects of time-series analysis in environmental sciences as well as an compilation of the problems and pitfalls in the interpretation of cyclicities in the stratigraphic record. The Appendix Published algorithms for time-series analysis contains a list of available commercial and internet-derived time-series software. The book Time-Series Analysis and Cyclostratigraphy by Graham Weedon is a general reference for those using time-series analysis, and suitable for graduate courses in environmental sciences, paleoceanography and geology. I would not recommend it to

undergraduates and newcomers to the field of cyclostratigraphy since the introductory chapters are not very well organized, extremely compact and full of jargon. The most important point for not giving it to beginners in the field of time-series analysis, however, is the fact that it does not provide any practical exercises. In my opinion, a textbook on time-series analysis and cyclostratigraphy should contain a CD with sample data and working programs for processing the data. The strength of the classic book Numerical Recipes by Press et al. (1st edn. 1985, 2nd edn. 1992) containing all kinds of statistical and numerical tools is the FORTRAN or C code in addition to a theoretical introduction to all methods. The books by G.V. Middleton (2000) and J.C. Davis (2002) on statistics and data analysis in earth sciences also provide a floppy disk or a WWW link to access sample data, exercises and programs; Swan and Sandilands (1995) provide numerous examples and sample data separated from the main text or listed in the appendix illustrating the use of statistical methods in geosciences. The students mainly learn from practicals and projects and not so much from lectures on data analysis in earth sciences. References Davis J.C. 2002. Statistics and Data Analysis in Geology. John Wiley & Sons, New York. Middleton G.V. 2000. Data Analysis in the Earth Sciences using Matlab. Prentice-Hall, New Jersey. Muller R.A. and MacDonald G.J. 2000. Ice Ages and Astronomical Causes. Data, Spectral Analysis and Mechanisms. Springer, Heidelberg. Press W.H., Teukolsky S.A., Vetterling W.T. and Flannery B.P. 1992. Numerical Recipes – The Art of Scientific Computing. 2nd edn. Cambridge University Press. Schwarzacher W. 1993. Cyclostratigraphy and the Milankovitch Theory. Elsevier, Amsterdam. Swan A.R.H. and Sandilands M. 1995. Introduction to Geological Data Analysis. Blackwell, Cambridge, Massachusetts.

Software References Matlab1 is a high-performance language for technical computing. The Signal Processing Toolbox is a collection of tools built on the Matlab1 numeric computation environment supporting a wide range of signal processing and spectral analysis operations (www.mathworks.com). OriginLab produces graphing and data analysis software for scientists and engineers including time-series analysis tools (www. originlab.com). SPSS (Statistical Package for the Social Sciences) is one of the most widely available statistical software packages. SPSS also includes various tools for time-series analysis (www.spss.com).

114 The Arand software package is a freeware time series and spectral analysis application for Apple Macintosh computers written by Philip Howell at Brown University (www.ngdc.noaa.gov/paleo/ softlib/arand/arand.html). AnalySeries is an Apple Macintosh application for spectral analysis. The full descriptive reference is available online from AGU: Paillard D., Labeyrie L. and Yiou P. 1996. Macintosh Program Performs Time-Series Analysis. Eos, Transactions, American Geophysical Union, 77, 379 (www.ngdc.noaa.gov/paleo/ softlib.html). Spectrum is a menu-driven PC program that allows the analysis of unevenly spaced time series in the frequency domain. The software has been written by Michael Schulz at Bremen University (www.palmod.uni-bremen.de/mschulz).

MARTIN H. TRAUTH Institut f € ur Geowissenschaften Universit€ at Potsdam Postfach 601553 14415 Potsdam Germany (e-mail: [email protected])

Statistics with Applications in Biology and Geology, by P. Blsild and J. Granfeldt. 2003. 555 pp. Chapman and Hall/CRC, Boca Raton. Price: USD 59.95; GBP 29.99. ISBN: 1-58488-309-X. This book is, according to the authors, based on a longstanding introductory statistics course and practicals for biology and geology students at the University of Aarhus, Denmark. Prior to this course, the students have taken a course in basic probability theory and in mathematics. The book has an introductory chapter on statistical analysis, model specification and checking, and statistical inference. This is followed by chapters on preliminary investigations (‘exploratory data analysis’), normal data (one, two or more samples, linear regression), linear normal models (analysis of variance, comparison of regression lines, two-way analysis of variance), power analysis and experimental design, correlation, the binomial and multinomial distributions, the Poisson distribution, generalized linear models including Poisson regression and dose–response models, analysis of directional data, the likelihood method, and selected non-parametric tests. Each chapter includes worked examples using SAS and concludes with a range of examples. The data sets and SAS macros can be downloaded from the authors’ course web site. The book is designed to provide ‘a firm basic in models, the likelihood method, and numeracy’ and ‘a practical introduction to using fundamental para-

metric statistical methods frequently applied to data analysis in biology and geology’. Given these aims, I am surprised that there is absolutely no mention of any multivariate data analysis and essential techniques such as principal components analysis and multivariate analysis of variance, of randomisation or permutation tests, of cross-validation in model selection, of Bayes theorem and Bayesian frameworks for data analysis, and of temporal and spatial auto-correlation. All these topics, plus the battery of multivariate dataanalytical approaches presented by, for example, Jongman et al. (1995), Legendre and Legendre (1998), and Davis (2002) are an important part of the modern data-analytical paradigm and repertoire that biology and geology students should be aware of. The theoretical level presented here as an introductory course for biological and geological students is very high, much higher than in the standard textbook (‘bible’) of Sokal and Rohlf (1995) on biometry. I have rarely, if ever, come across undergraduate biology or geology students in Cambridge, Minneapolis, London, or Bergen who would feel comfortable with the theoretical level presented here and I would wonder how many such students would go on and apply numerical thinking and numerical techniques in their research if their introductory undergraduate course was at the level required here. Given the title of the book, I was disappointed that it was so theoretically based, that it was limited in the topics it covered, and that it contained little of everyday use for quantitative palaeolimnologists. As a book it is more suited to post-graduate biology and geology students who, after having acquired a feel for statistical data analysis, want to improve their theoretical understanding of, for example, linear regression. Here they will find 42 pages devoted to this. They will, however, still have to turn to Legendre and Legendre (1998) to find out about model II simple linear regression and the essential distinctions between models I and II regressions in a biological context. Despite the availability of this interesting book, I will continue to use books like Sokal and Rohlf (1995) and Legendre and Legendre (1998) in my teaching of undergraduate and post-graduate biologists and environmental scientists in Bergen and London.

References Davis J.C. 2002. Statistics and Data Analysis in Geology. Wiley and Sons, New York, 638 pp.