A study of the phenotypic and genotypic variation of 545 single tree progenies of 38 provenances of the 1970 I.U.F.R.O. Sitka spruce (picea sitchensis (bong) carr) collection

During the 1970 fall, the I.U.F.R.O. Section 22 "Working group on procurement of seed for provenance research" organized an expedition to collect Sitka spruce cones from British Columbia and Alaska. The locations of the 39 provenances range from 48.38 ° to 58.37° latitude N and from 121.93° to 134.58° longitude W. The elevation varies from 0 to 2, 200 feet above sea level. In most cases, the collections were made from 15 trees in each location. The single tree progeny collection of 557 trees representing 39 locations constitute the material of this thesis. In a first part, seed and cone morphology were studied on a single tree basis. Ten cones per progeny were randomly selected and the length of each cone measured to the nearest mm. Five randomly selected seeds from each tree were mounted on a special sheet, and seed length, seed width, wing length and wing width were measured to the nearest 0. 01 mm. Nested analyses of variance and DUNCAN's multiple range tests for all the characteristics studied have been performed using five subregions. No definite classification of the provenances was possibly using univariate anova procedures. A simple correlation matrix has been calculated between all the traits studied and longitude, latitude and altitude of the place of origin of the provenances, using the provenance means. Multiple regression analyses have been used for investigating this correlation matrix. The percentage of variation accounted for by the geographical co-ordinates varies between 10. 2% and 43. 6%. Using the seed and cone traits studied, a comparison of several multivariate statistical analyses which could be used for classification purposes has been attempted. The so-called canonical analysis, discriminant function analysis and principal component analysis have been compared and applied for classifying the provenances. The sub-regions already used were analyzed separately. Dendrograms were also constructed and analyzed. Advantages and disadvantages of each multivariate method have been discussed. It was found that the discriminant function analysis, its associated generalized distances of MAHALANOBIS and dendrograms provided the most rational classification of the provenances. In a second part, the genetic variability of 545 Sitka spruce single tree progenie s was studied in a nursery test during 1 971 andl972. A total of 545 single tree progenies grouped into 38 provenances was sown in April, 1971, using a randomized complete block design with four replications and 24 seedlings per replication or 96 seedlings per progeny. The seeds were placed in the cavities of styro-blocks using the method developed by the Pacific Forest Research Centre in co-operation with the B. C. Forest Service and they have been treated by the most recent nursery methods, in the new B. C. F. S. nursery at Surrey (B.C.). Germination rate, bud set, length of the epicotyl and survival after the first growing season were assessed in 1971. The "seedlings were translpnated in plain soil seedbeds in May, 1972, to a distance of 6" to 6", each progeny being kept separate while respecting the same statistical design as in 1971. Bud burst, bud set, colour of the needles and total height after the second growing season, were assessed in 1972. There was a clinal variation in bud burst, bud set, colour of the needles and epicotyl length. Bud burst was negatively correlated with longitude (r = -0. 50) and positively correlated with altitude (r = 0.42). Bud set appeared under strict genetic control as indicated by the second estimation of this trait, at the end of the second growing season (with latitude: r = 0. 88). Latitude and altitude of the seed sources explained 65% of the total variation in epicotyl length. Total height after the second growing season showed the same relationships as epicotyl length. General equations for components of variance for unbalanced data were originally calculated for a nested-crossed model. Components of variance and their standard error were calculated for epicotyl length and total height after the second growing season. Depending on the sub-regions, the genetic variance among provenances is generally larger than the tree to tree genetic variation. The narrow sense heritability, on an individual basis, and its standard error, for total height after the second growing season, were estimatedon a subregion basis. Heritability was found generally to be close to 0. 10, indicating low general combining ability. The relationships between the seed, cone and seedling traits measured were studied. Multiple regression analysis showed that a higher proportion of the variation of the seedling traits was accounted for by the geographical coordinates of the provenances than by the cone and seed traits studied. Variation in foliar macro- and micro- nutrients of 10 Sitka spruce provenances was studied, but no geographical pattern of variation detected in K, Ca, Mg, Fe, Mn, Zn, P, and N needle contents. Only K showed some provenance to provenance variation. Possible physiological explanations for this absence of variation are discussed.Forestry, Faculty ofGraduat