tion of the seed source, and second on geographic pat- 

 terns of variation that were independent of elevation. 

 Adequacy of a model was judged according to the good- 

 ness of fit (R-). residual variance (s^,.^), and geographic or 

 ecologic patterns displayed by residuals (Draper and 

 Smith 1966). 



Elevational models considered both a linear and quad- 

 ratic relationship between performance and elevation of 

 the seed source. Deviations from the best fitting eleva- 

 tional regressions were then used as dependent variables 

 for considering geographic patterns of variation that 

 were independent of elevation. The geographic model 

 included four independent variables, plus their squares: 

 latitude, longitude, northwest departure, and northeast 

 departure. The latter two variables were obtained by 

 rotating the grid of latitude and longitude by 45°. Thus, 

 eight independent variables were included in the model, 

 which was fit with a stepwise regression program for 

 maximizing R- (Barr and others 1979). Finally, for those 

 variables in which both the elevational and geographic 

 models had been significant, elevation of the seed source 

 was added to the independent variables of the best fit- 

 ting geographic model in order to estimate the joint 

 determination of performance by geography and 

 elevation. 



RESULTS 



Because each study was autonomous, the results of 

 each are presented separately but are considered jointly 

 in establishing patterns of genetic variation. 



Growth and Development 



The effects of planting environment on growth and 

 development of seedlings were pronounced for all traits 

 (table 1). Compared to trees growing at high elevation, 

 trees at low elevation were much taller and had longer 

 leaves. In addition, late growth, defined as the propor- 

 tion of the predetermined shoot that elongated after the 

 fourth week, was measured approximately 5 weeks 

 earlier at low elevation than at high and therefore was 

 affected tremendously by the environment. Adjusted 

 heights show that even if all trees had been the same 

 height at age 2, those growing at low elevation would 

 still have been tallest after 3 years. 



For nearly all variables, the range in population mean 

 values was considerably different at the two planting 

 environments (table 1). For these ranges to differ greatly 

 illustrates the heterogeneous variances that required 

 adjustment before statistical analyses were made. 



By accounting for more than 20 percent of the total 

 variance, effects of populations were pronounced for 

 3-year height and injury from a spring frost (table 2). 

 These strong effects are illustrated by mean differences 

 between populations of as much as 0.8 foot in height 

 and 50 percent in trees injured by frost. On the average, 

 trees damaged by frost were 3.2 inches shorter after 

 3 years than uninjured trees from the same population. 

 Relatively weak effects for late growth are likely due to 

 the relatively late stage of elongation at which measure- 

 ments were begun: the average tree produced only 10 

 percent of the 3-year shoot as late growth. Consequently, 



Table 1. — Mean values according to test environment and range of population 

 means in each environment 



Variable 





Mean values 



Range in population means 



2,200 ft 



1,500 ft 



2,200 ft 



5,000 ft 



Late growth 



(inches) 



1.1 



0.7 



0.8 



0.4 



Leaf length 



(inches) 



2.9 



2.7 



1.1 



0.8 



Height 



(inches) 



17.1 



13.9 



12.2 



5.8 



Adjusted height 



(inches) 



16.2 



14.7 



5.7 



1.8 



Frost injury 



(percent) 



22 





56 





Table 2. — Results of analyses of variance for growth and development 

 presented as intraclass correlations, the ratio of individual 

 variance components to the sum of all components 



Variable 



Source of 



Late 



Leaf 





Adjusted 



Frost 



variance 



growth 



length 



Height 



height 



injury 



Environments 























Blocks in 













environment 



0.03** 







0.02** 











Populations 



.06** 



0.08** 



.36** 



0.02* 



0.20** 



Environment x 













population 







.02 



.01 



.09** 





Experimental 



.07** 



.07 



.03** 



.04** 



.79 



error 













Within plots 



.84 



.83 



.57 



.84 





*Statistical significance of the F-value at the 5 percent level of probability. 

 **Statistical significance of the F-value at the 1 percent level of probability. 



3 



