Multivariate patterns of adaptive genetic variation and seed source transfer in Picea mariana

A short-term retrospective test trial was carried out using 90 open-pollinated families representing 30 provenances of black spruce (Picea mariana (Mill.) BSP) from Quebec. Seedlings were transplanted on three sites along a latitudinal gradient, and eleven growth and phenological traits were measured during the second and the third growing seasons. Analyses of variance indicated for most of the traits significant differences among provenances and families-within-provenances. Principal component analysis was used to summarize the variation observed among provenances into two principal components, which accounted for 79% of the total variation for all traits. Regression models developed to relate each trait and the principal component scores to geoclimatic variables explained between 55% and 86% of the variation observed among provenances. Variation in growth traits and phenological traits appeared to be related to geoclimatic factors. The models were validated using data from a range-wide provenance test, and relative risks associated with seed source transfer were estimated. The R2 values between the transfer risk and the provenance heights ranged from 0.02 to 0.58, whereas they were slightly lower for diameters. On average, the relative risks varied from 36% to 67%. Individual provenance values ranged from 4% to 94%. A geographic information system tool was designed to assist the forest managers in making seed transfer decisions.