James D. Rising. 2001. Studies in Avian Biology No. 23, Cooper Ornithological Society, Allen Press, Lawrence, Kansas. 65 pp., 11 tables, 21 figures, including 1 map and 2 appendices. ISBN 1-891278-28-X. Paper, $7.00. — James D. Rising, author of two other major papers on the morphometrics of Savannah Sparrows (Passerculus sandwichensis) (Rising 1987, 1988), coauthor of the Savannah Sparrow species account for the Birds of North America (Wheelwright and Rising 1993), and author of a field guide to the sparrows of the United States and Canada (Rising 1996), has now written a monograph on geographic variation in the size and shape of Savannah Sparrows. The work is based on the remarkable collection of more than 2,200 skeletons that he has amassed at the Royal Ontario Museum in Toronto. The 65 localities for those specimens span large areas of North America, from northern Canada and Alaska to the northeastern United States, through the central Great Plains and the highlands of the western United States, along the Pacific coast into Mexico, and south to central Mexico. The appendices give summary statistics by sex and locality for 24 skeletal measurements of birds from 45 localities. The availability of this extensive data set is an important contribution to ornithology in North America.

Rising's approach to data analysis, as in his previous papers, is almost entirely multivariate. After performing a principal-components analysis (PCA) on the correlation matrix among the measurements of the bones, he plotted the average scores for each locality in a graph, interpreting PC1 as a general size variable and PC2 as a general shape variable. That graph shows his main result. Savannah Sparrows are exceptionally large on Sable Island, Nova Scotia (the home of the subspecies P. s. princeps), on Umnak Island in Alaska, in the Aleutian Islands generally, and out on the Aleutian Peninsula. Along the Pacific Coast, the size of Savannah Sparrows increases southward. Those resident Savannah Sparrows have proportionately shorter wings than birds elsewhere. Their bills are especially large in the salt marshes of Sonora and Sinaloa in Mexico. In comparison, Savannah Sparrows across the interior of North America show very little geographic variation in size. The larger birds in Nova Scotia and the smaller ones in Utah differ by only 2 mm in average length of the tibiotarsus, a good univariate indicator of size. Using a reduced set of variables and localities, Rising performed a discriminant-function analysis of non-salt-marsh populations and a stepwise discriminant function analysis of salt-marsh populations. Then, to look for covariation between morphology and climate, geography, and the number of potentially competing species, he calculated correlations and multiple regressions. In those cases, for morphology he used the derived multivariate scores for his size axes (PC1, DF1) and shape axes (PC2, DF2). Finally, he calculated principal components for a set of environmental data for the same localities and correlated environmental and phenotypic PC scores, ran a canonical correlation between the two matrices, and performed a Procrustean analysis. He concluded that Savannah Sparrows are smallest in dry areas where maximum summer temperatures are highest (which turns out to be at high elevations in the west) and largest on cool moist islands. The birds tend to be small where more species of potentially competing sparrow-like birds are present. A further conclusion is that only P. s. sandwichensis and P. s. princeps should be retained as subspecies of non-salt-marsh populations. Evaluation of relationships among the nine subspecies of Savannah Sparrows in the western salt marshes awaits the results of ongoing biochemical studies being conducted in collaboration with Robert Zink.

Rising's general conclusions from skeletal data support previous descriptive work on the basis of study skins but not carried out at such a comprehensive scale. Had he included his data from the study skins of those same specimens, he could have added analyses of geographic variation in the external morphology of the wing, tail, and bill, each of which is likely to have important ecological relevance. Two weak aspects of this work are its omission of any discussion of allometry, the covariation of size and shape, and the neglect of the likelihood that a substantial fraction of the shape variation in a morphological data set is included in the first principal component. Also, because he uses correlations between component scores and values of the original measurements to interpret principal components, rather than the coefficients of the PC equation, Rising's interpretations of principal components reduce to univariate relationships (Rencher 2002).

After the discriminant analysis and comparisons with environmental variation, Rising interprets his results as not following Bergmann's Rule, which he defines, in the way so many others do, as a relationship between size and latitude rather than a relationship between size and gradients in the temperature-moisture regime, which may well be independent of latitude (James 2001). His multivariate results find the smallest birds to be in high dry sites in the west. Even so, others have concluded that the smallest Savannah Sparrows are along the humid coast between Vancouver Island and northwestern California (Peters and Griscom 1938), and Rising's own sample of females from Hoquiam, on the Washington coast, have the smallest femurs and tibiotarsi of any of his samples. The data from coastal prairies and marshes seem to contradict the argument, made for other populations, that size increases with cooler and more humid environments.

The general pattern of geographic variation in Savannah Sparrows rangewide looks like three separate systems of covariation: the larger-on-coolmoist-islands system, the west-coast prairies and salt marshes system, and the system across most of the North American continent. By interpreting those three, perhaps separate, phenomena in one graphic space, Rising may be confounding some interesting within-system allometric phenomena. For example, in the salt-marsh populations, which tend to have slender bills, the bills become even more slender as the birds get larger along the southwestern Mexican coast. Those shape differences are hidden within Rising's first principal components and his first discriminant functions.

This publication is an important step toward characterizing intraspecific geographic variation in a widespread species. Unfortunately, even more samples and further analyses may be necessary before geographic variation in the size and shape of the Savannah Sparrow can be fully characterized and fully reliable inferences made about its environmental correlates.