A Multivariate Study Of Geographic Variation In The Cones Of Pinus Contorta In Relation To Environmental Variables

A multivariate analysis was made of geographic variation in the cones of Pinus contorta as related to environmental factors. Variation between four previously defined subspecies was compared to variation in geographic location, elevation, topographic situation, climate and soil. The allocation of measurement effort was optimized through preliminary studies which objectively selected a set of characters for study, and determined the number of cones to measure to represent each site.;After choosing a resemblance function with desirable properties, data structure was investigated using three-dimensional stereograms of principal components, accompanied by a demonstration that almost no distortion resulted from this procedure. A discontinuous data structure was revealed, with groups corresponding to the subspecies previously recognized, except for an extension of the range of the population previously described as restricted to the neighbourhood of Mendocino, California. A wide range of clustering algorithms confirmed the robustness of this group structure. The ranking algorithm used for character selection identified characters having the greatest ability to distinguish between the groups. The most important extrinsic factors related to group structure were geographic region and available soil phosphorus. The groups were shown to be essentially homogeneous except for a subdivision of the Rocky Mountain group associated with cone serotiny, and the divergence of several sites which could be explained on the basis of extreme conditions associated with those sites. Trends were found within the groups associated with variation in soil phosphorus and elevation