Patterns of Genetic Variation 



Multiple regression models were used to relate genetic 

 variation between populations to the elevation and geo- 

 graphic location of the seed source. Independent variables 

 included elevation, latitude, longitude, northwest depar- 

 ture, southwest departure, and their squares. Squared 

 values were used to accommodate the possibility of non- 

 linear patterns of variation. Northwest and southwest 

 departures were derived by rotating the grid of latitude 

 and longitude by 45°. Elevation was considered without 

 geographic interaction because preliminary analyses sug- 

 gested that the relationship between performance and 

 elevation was similar throughout the region of study. 

 Thus, 10 independent variables were included in a stepwise 

 regression for maximizing i?^ (SAS 1982) according to the 

 general model: 



J 3 



where is the performance of population i ; Xy is in- 

 dependent variable j for population i; and ^o. /^i. and P2 are 

 regression coefficients, j = 1 ... 5. 



Adequacy of a model was judged according to the good- 

 ness of fit (R^), residual variance (Sy.^), patterns displayed 

 by residuals (Draper and Smith 1981), and the degree by 

 which results were biologically plausible. 



RESULTS 



The analyses detected differences between populations 

 for numerous traits involving (1) growth and development 

 and (2) the periodicity of shoot elongation. Genetic varia- 

 tion in these traits was related to the elevation and 

 geographic origin of the seed. 



Growth and Development 



Planting sites had a tremendous effect on the growth 

 and development of seedlings. Under mesic cultural condi- 

 tions, the average 3-year-old tree was 46 cm tall with 

 leaves 14 cm long, but under xeric conditions, the average 



tree was only 23 cm tall with leaves 7 cm long. Thus, xeric 

 culture reduced the growth potential expressed under op- 

 timal growing conditions by an average of 23 cm. Values 

 of adjusted height show that if trees on both sites had 

 been the same height at age 2, those growing under the 

 mesic regime would still have been 11 cm taller at age 3 

 than trees on the dry site even though 3-year height ac- 

 counted for 69 percent of the variance in 2-year height. 



Because of these differences associated with cultural 

 conditions, main effects of test sites accounted for most 

 (68 to 85 percent) of the variance for those variables 

 measured on both sites and, therefore, dominate analyses 

 of variance (table 1). Nevertheless, differences between 

 populations were detected for all variables measured on 

 both sites even though the main effects of populations 

 accounted for less than 5 percent of the total variance 

 (table 1). 



The significant interactions (table 1) for height and ad- 

 justed height indicate that the differences between popula- 

 tions on one site were not the same as the differences 

 between the same populations on the other site. Because 

 these interactions accounted for nearly a third as much 

 variance as the main effects of populations, subsequent 

 analyses involving population means will consider height 

 and adjusted height from each site as separate variables. 

 By contrast, the lack of a significant interaction for leaf 

 length shows that the relative performance of populations 

 was similar at each site even though leaves were much 

 shorter under xeric conditions. 



Populations also differed significantly in the reduction in 

 height attributable to xeric culture. The absolute amount 

 of reduction ranged from 28 cm (56 percent) for one 

 population to 14 cm (43 percent) for another. But popula- 

 tions did not differ in mortality under xeric conditions 

 because drought mortality reached only 16 percent and 

 was confined to essentially one block. 



Periodicity of Shoot Elongation 



Shoot elongation of individual seedlings was completed 

 between 31 and 53 days after being placed in the green- 

 house. This means that 11 to 17 observations were 



Table 1 — Results of analyses of variance of growth and development in field environments presented 

 as intraclass correlations, the ratio of a variance component to the sum of all components 



Variable 



Source of 



3-year 



Adjusted 



Leaf 



Xeric height 



Drought 



variance 



height 



height^ 



length 



reduction^ 



mortality 



Sites 



0.850** 



0.689** 



0.850** 







Blocks in site 



.000 



.002** 



.008** 



0.006** 



0.977** 



Populations 



.028** 



.021** 



.004* * 



.317** 



.012 



Sites X populations 



.010** 



.009** 



.002 







Experimental error 



.007** 



.047** 



.020** 



.103** 



.010 



Within plots 



.106 



.232 



.116 



.574 





Error mean square^ 



107.433 



72.969 



5.824 



3.119 



.027 



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

 ^3-year height adjusted by regression on 2-year height. 



^Difference between xeric height and the growth potential expressed by mesic height. 



^Absolute value of the interaction mean square that was used to test for differences between populations. 



3 



