Geographic Variation of Vegetative Characters of Populus deltoides

Abstract

Spatial autocorrelation analysis was used to study the patterns of geographic varia- tion in Populus deltoides (Salicaceae). Ten characters reflecting the vegetative morphology in this species were analyzed for each of 522 individuals in 302 localities scattered throughout eastern North America. Factor analysis reduced the dimensionality of the matrix. The first factor, reflecting leaf size, shows a clear clinal pattern throughout the study area, with larger leaves in the south and southeast. Other factors, reflecting twig size, leaf apical shape, and leaf base shape also exhibit spatial patterns but on a smaller geographic scale. The influence of large-scale environmental and climatic variation on these patterns was examined by regressing the factor scores against a matrix of five environmental parameters. The regression residuals were then examined for spatial pattern. In most cases the residuals retain no spatial patterning. However, the first factor, that reflecting leaf size, shows some small-scale pattern. Analysis of geographic variation patterns of populations within a species, especially one with a wide area of distribution, permits an insight into its population structure and may lead to the factors that cause the observed dif- ferentiation, cohesion, or disjunction of the gene pool. We report on a study of geographic variation of leaf and twig characters of the east- ern cottonwood Populus deltoides Bartr. ex Marsh. subsp. deltoides. The opportunity for the pres- ent study arose as a by-product of geographic variation studies of aphid species of the genus Pemphigus, which form galls on the leaves and petioles of species of Populus (Sokal et al. 1980; Sokal and Riska 1981). Collections of samples from these trees were made throughout the conterminous U.S. and portions of Canada. To be useful in the study of the aphid parasite, these Populus specimens had to be collected in the summer, which precluded obtaining flow- ers or fruits. For this reason, the collections contain only leaves and twigs. The principal method of analysis in this pa- per is spatial autocorrelation analysis. This is a recently introduced technique for the analysis of geographic variation, which has as its special goal the description of the population structure of the studied samples (Sokal 1979, 1984; Sokal and Oden 1978a, 1978b; Sokal and Wartenberg 1981, 1983). To date it has been applied to data sets on various animals including humans