ISSN:
1573-5095
Keywords:
geographic variation
;
multivariate analysis
;
trend-surface analysis
;
transfer risk
Source:
Springer Online Journal Archives 1860-2000
Topics:
Agriculture, Forestry, Horticulture, Fishery, Domestic Science, Nutrition
Notes:
Abstract Early studies of allozyme variation in plant populations suggested that allelic frequencies in some loci vary by geography. Since then, the expectation that allozymes might be useful in describing geographic patterns has generally not been borne out by single locus analyses, except on the broadest scale. Multi-locus analyses reveal the converse: canonical correlation analysis of individual, uniformly-spaced genotypes describe statistically-significant, complex patterns with geography. Multi-locus scores in four major species, Abies concolor, Pinus lambertiana, P. ponderosa, and Pseudotsuga menziesii, of the mixed conifer forest in the Sierra Nevada correlate 0.40 or greater with the first canonical vector of a geographical trend surface equation. The different species follow similar patterns by latitude and elevation. In contrast with patterns in the Sierra Nevada, large-scale differentiation is weak (R 2 〈 0.20) among populations of Pseudotsuga menziesii in the Coast Ranges and Siskyou Mountains of northern California and southern Oregon, where differentiation may be local. For the purpose of forming zones, we subdivided scores of the first two to four canonical vectors into groups and plotted them as multidimensional contour intervals. Reclassification by discriminant analysis serves as an approximate guide to transfer risks within and among these groups.
Type of Medium:
Electronic Resource
URL:
http://dx.doi.org/10.1007/BF00120649
Permalink