three were from the lower Snake River basin of central 

 Idaho; and two were from a small disjunct population 

 from Deadline Ridge in south-central Idaho (fig. 1). Seed- 

 lings from the 60 populations were used in separate 

 studies of (1) growth and development in field environ- 

 ments, (2) periodicity of shoot elongation in a green- 

 house, and (3) cold hardiness in the laboratory. Because 

 cone collections, experimental procedures, and statistical 

 analyses are detailed in the first paper of this series, 

 which involved populations from the Wasatch and Uinta 

 Mountains of Utah (Rehfeldt 1985a), only an outline of 

 these procedures follows. 



Growth and Development 



One set of seedlings from each of the 60 populations 

 grew for 6 months in plastic containers (10 in-^) in a 

 shadehouse at Moscow, ID (lat. 48.5° N., long. 116.7° 

 W.). In the fall, seedlings were transplanted into two 

 environments, 2,200 and 5,000 feet elevation, at the 

 Priest River Experimental Forest, 150 miles north of 

 Moscow. Eight seedlings, spaced at 1.0 foot and 0.5 foot 

 at low and high elevation sites, respectively, were 

 planted in row plots, separated by 1.5 feet and 1.0 foot 

 at the respective sites. Three blocks were estabUshed at 

 each site, and both plantings were maintained under 

 intensive culture. 



The performance of each seedling was described by 

 four variables for which differentiation of Utah popula- 

 tions had been pronounced: (1) height— seedling height 

 after 3 years; (2) late growth— amount of the 3-year 

 predetermined shoot that elongated after the fourth 

 week of elongation in the respective environments; 



(3) leaf length— the length of a leaf near the center of the 

 3-year shoot; and (4) adjusted height— 3-year height 

 adjusted by regression on 2-year height. By representing 

 the increment from a common 2-year height, adjusted 

 height is relatively independent of previous environmen- 

 tal (such as transplanting shock) and genetic effects and 

 thereby is capable of reflecting adaptation of populations 

 to a particular environment for a single growing season. 



In addition, scores were made for the presence or 

 absence of injury from a freeze (16 °F) that occurred in 

 mid-May at the low-elevation site. Injuries ranged from 

 death of the developing shoot to reductions in internode 

 lengths near the shoot tip. 



To account for heterogeneous variances at the two test 

 environments, data were transformed to standard normal 

 deviates (Steel and Torrie 1960) for each test site and 

 the deviates were analyzed according to a model of ran- 

 dom effects. The model estimated: (1) main effects for 

 test environments, blocks within environments, and 

 populations; (2) the interaction of populations X environ- 

 ments; (3) an experimental error, composed of the inter- 

 action of populations X blocks within environments; and 



(4) a residual error. As a result of the transformations, 

 mean squares associated with the main effects of test 

 environments were zero. A harmonic mean of 7.20 

 reflected the number of seedlings representing each 

 population in each block. 



Injury from the spring frost was analyzed according 

 to a model of random effects, which estimated main 

 effects for blocks and populations, plus an error. 



Periodicity of Shoot Elongation 



Another set of seedlings grew for 1 year in plastic con- 

 tainers (45 in^) in a shadehouse at Moscow. Nine seed- 

 lings from each population grew in each of three blocks. 

 In early March of the second growing season, seedlings 

 were moved into a greenhouse before shoot elongation 

 had begun. Greenhouses were maintained under natural 

 lighting; temperatures were about 75 °F during the day 

 and 55 °F at night. All seedhngs were measured three 

 times each week until elongation of the preformed bud 

 had ceased. 



As described by Rehfeldt and Wykoff (1981), shoot 

 elongation of individual trees was expressed mathemati- 

 cally by a logistic function that included a hyperbolic 

 time term. Regression statistics allowed calculation of 

 the following variables for describing periodicity of shoot 

 elongation of individual seedlings: (1) initiation of 

 growth— the day on which 0.1 inch of cumulative growth 

 had occurred; (2) cessation of growth— the day on which 

 all but 0.1 inch of growth had occurred; (3) duration of 

 growth— the number of days between initiation and ces- 

 sation; (4) rate of growth— elongation per day during the 

 period of maximum elongation; and (5) total elongation. 



Population differentiation was assessed according to 

 least squares analyses of random effects that estimated 

 main effects for populations and blocks, the interaction 

 of populations X blocks, and an error. A harmonic mean 

 of 8.81 reflected the number of seedlings representing 

 each population in each block. 



Cold Hardiness 



To estimate the relative cold hardiness of populations 

 in early autumn, laboratory tests of freezing tolerance 

 were made according to the general procedures of Levitt 

 (1972). Mature leaves were collected in late September 

 from near the center of the 3-year shoot of trees planted 

 at 5,000 feet elevation in the study of growth and 

 development. 



Four sets of eight leaves were collected from each 

 eight-tree plot; each set contained one leaf from each 

 tree. One set of leaves from each plot was exposed to 

 one of four test temperatures (—4 °F, — 6 °F, — 10 °F, 

 and- 12 °F) by cooling at the rate of 10 °F/h. Injured 

 leaves were identified visually by the presence of dis- 

 colored and flaccid tissue (Rehfeldt 1980). Population 

 differentiation for the proportion of injured leaves from 

 each plot was assessed from a model of random effects 

 that allowed estimation of main effects for populations, 

 blocks, and test temperatures; three two-way interac- 

 tions; and a residual. 



Patterns of Variation 



Geographic patterns of genetic variation were assessed 

 from a series of regression analyses that fit independent 

 variables describing the origin of populations to the 

 array of population means. Because the primary purpose 

 of these analyses was to describe genetic patterns of 

 variation for the upper Snake River Basin, regression 

 analyses considered only the 51 representative popula- 

 tions. The sequence of regression analyses concentrated 

 first on the relationship between performance and eleva- 



2 



