increasing elevation of the seed source, but freezing 

 tolerance increases. The strong relationship between ele- 

 vation and adjusted height indicates that, at the low 

 elevational planting, populations from mild environments 

 still would have been tallest even if all trees had been 

 the same height at age 2. Quadratic models provided a 

 significant reduction in the residual mean square of the 

 linear model for only the cessation and duration of 

 elongation. 



The geographic model (table 5) was statistically signifi- 

 cant for only half of the variables. At most, this model 

 accounted for less than a third of the variance in the 

 dependent variables that was not explained by elevation. 

 Consequently, geography accounted for less than 25 per- 

 cent of the variance among populations. Thus, eleva- 

 tional clines tend to be steep, and geographic clines are 

 relatively gentle. Nevertheless, geographic variables plus 

 elevation (the combined model) accounted for 43 to 83 

 percent of the variance of populations. 



Geographic patterns of genetic variation that are 

 independent of elevation are presented in figure 4 for 

 several traits. These patterns were generated from 

 values predicted by the geographic model. The contour 

 interval is scaled to a value equaling half the least sig- 

 nificant difference (Steel and Torrie 1960) between popu- 

 lations at the 80 percent level of probability [V2 lsd{Q.2)]. 

 Because values of Isd were calculated from the analysis 

 of variance (table 2), contours represent about half the 

 geographical distance associated with population 

 differentiation at the 80 percent level of probability. 

 Contouring was begun with the overall mean, the zero 

 deviation from the elevational regression. 



All variables for which the geographic model was sig- 

 nificant presented similar patterns of genetic variation. 

 Those variables selected for figure 4 illustrate patterns 

 of greatest divergence. In general, populations in the 

 west-central regions are of highest growth potential and 

 lowest hardiness. From this area, growth potential 

 decreases and hardiness increases in all directions. The 

 general pattern, however, is influenced greatly by the 

 rugged Teton Range. Populations of relatively high 

 growth potential and low hardiness extend into the 



DURATION OF ELONGATION 



LEAF LENGTH 



Cl:0.9 DAYS 



Cl:0.08 INCHES 



3-YEAR HEIGHT 



FREEZING INJURY 



Cl = 0.47 INCHES 



Ci:7 % 



Figure 4.— Geographic patterns of variation 

 ttiat are independent of elevation. Shading 

 marlKS the distribution of lodgepole pine in 

 reference to the Teton Range. C.I. = the con- 

 tour interval, which is scaled to a value of 

 V2\s6(0.2). The zero contour represents the 

 mean of all populations standardized for 

 elevation. 



Table 6. — Correlation matrix relating population means for all studies'' 



Variable 



Code 



LL 



H 



AH 



Fl 



1 



D 



R 



C 



EL 



IN 



Growth and development 

























Late growth 



LG 



0.56 



0.68 



0.61 



0.34 



-0.24 



0.54 



0.61 



0.53 



0.61 



0.13 



Leaf length 



LL 





.62 



.53 



.33 



-.27 



.62 



.59 



.62 



.64 



,21 



Height 



H 







.81 



.56 



-.21 



.84 



.87 



.85 



.90 



.36 



Adjusted height (2,200 ft) 



AH 









.44 



-.32 



.71 



.71 



.70 



.75 



.20 



Frost injury 



Fl 











.08 



.42 



.43 



.43 



.44 



.27 



Periodicity of shoot elongation 

























Initiation 



1 













-.33 



-.23 



-.27 



-.31 



.10 



Duration 



D 















.84 



.99 



.93 



.27 



Rate 



R 

















.84 



.97 



.40 



Cessation 



C 



















.93 



.28 



Elongation 



EL 





















.37 



Cold hardiness 

 Injury 



IN 



''Coefficients of absolute value >0.27 are statistically significant at the 5 percent level. 



6 



