Table 2. --Form of analyses of variance for quantitative genetics 



Source of variance Expected mean squares 



Blocks 



w + 





^2 





Populations 





^1 





' ''3 °F/P 



Families in populations 









- ^2 °F/P 



Experimental error' 











Within 











Contains all sources of variance involving interaction of blocks 

 where : 





Hoi comb 



Idaho City 



Boulder 



Creek 



Jack's Creek 





1.83 



1 .90 



1 , 



.89 



1 .56 





11.36 



16. 04 



16. 



,29 



4.70 



H '- 



12.01 



16.83 



16. 



.75 



5.27 





74.50 



109.04 



110. 



,47 



30.45 



h = 



334.56 



377.89 



397. 



,03 



177.01 



RESULTS AND DISCUSSION 

 Population Differentiation 



Analyses of population differentiation readily detected mean differences among populations 

 across all planting sites (table 1). But, main effects of sites were also significant, and 

 populations tended to perform differently at the various planting sites. The interaction 

 apparently arises from expression of differential adaptations of populations at planting sites 

 that encompass a large elevational (550 m) and latitudinal (> 1 degree) range. 



Least square mean heights and physiognomic criteria of the seed sources are presented in 

 table 3. These means are representative of mean performance across all four test sites, and, 

 therefore, they are confounded by interactions between genotype and environment. But, simple 

 correlation coefficients (r) of mean height for populations at each site and least square 

 means ranged from 0.64 to 0.85. The high values of these correlation coefficients suggest 

 that confounding of means by genotype-environment interactions is small. Least square means 

 apparently reflect an average performance suitable for assessing differentiation in relation 

 to physiognomic criteria of the seed source. But, because the correlations were not perfect, 

 differentiation also was assessed from performance at each site. 



3 



