INTRODUCTION 



The basis for a practical tree improvement program of ponderosa pine in southern Idaho is 

 contained in 16-year old family evaluation tests. As detailed in preceding publications (Wang 

 1967; Wang and Patee 1974, 1976; Rehfeldt 1980), primary objectives of the tests are: to 

 develop seed zones and seed transfer guidelines that account for genetic adaptations to 

 natural environments, to identify outstanding individual trees and populations from which 

 wind-pollinated seeds will provide genetic gains to artificial reforestation, and to provide 

 materials and information prerequisite to future breeding programs. 



The present analyses of 16-year height utilize experimental materials more fully described 

 in a series of papers that have assessed performance at ages 3, 8, and 11. Test plantations 

 currently require thinning; therefore, it is appropriate to consider alternative plans for the 

 future research program and for development of seed orchards. 



MATERIALS AND METHODS 



Wind-pollinated cones had been collected from 1 to 10 trees in each of 37 populations. 

 Populations were separated by about 1 000 m in elevation and 2 degrees in latitude and longi- 

 tude; they also represented habitat types that ranged from the relatively dry Pinus 

 ponderosa/Purshia tridentata to the relatively moist Abies grandis/Vaccinium globulare . 

 Thereby, much of the geographic distribution and ecological amplitude of the species in south- 

 west Idaho was represented. Seedlings from 268 families were planted in four-tree row plots 

 in each of 10 blocks at four planting sites. About 1.5 m separated trees within rows; 3 m 

 separated rows. Test sites included Idaho City (1 200 m elevation), Holcomb (1 100 m) , in the 

 southern portion of the region; and Boulder Creek (1 450 m) and Jack's Creek (1 650 m) in the 

 north. Since the last report of this series (Rehfeldt 1980), the plantation at Boulder Creek 

 was thinned: the two shortest trees in each family plot were removed, and, at Jack's Creek, 

 three replicates of exceptionally low survival were destroyed. 



Statistical analyses for assessing population differentiation and quantitative genetics 

 were made on the 16-year height of individual trees. An appropriate analysis would involve 

 calculation of main effects for sites and populations; nested effects of blocks in sites and 

 families in populations; four interactive effects; and variance within plots. As explained 

 previously, however, imbalances in the original design were accentuated by survival percentages 

 that ranged from 22 to 80 percent at the four test sites. Because missing data were excessive, 

 least squares analyses were required for accurate estimation of variance components. Yet, the 

 large size of the matrix needed for a single analysis precluded inversion. Consequently, 

 statistical analyses were completed in separate stages. 



Population differentiation was assessed from a least square analysis of variance (tablfe 

 1) . One population, represented by a single family, was excluded from analyses. In addition 

 to assessing differentiation, this analysis allowed calculation of least square means for 

 populations within and among planting sites. These means were used in all subsequent compar- 

 isons of population performance; they reflect the height expected if an equal number of 

 observations had been available for each population in each block at all planting sites. 



1 



