The multiple regression analyses pertaining to hardiness during phase one have 

 direct application in delineating seed zones that reflect adaptive differentiation in 

 the cold hardiness of populations. Methods described by Morgenstern and Roche (1969) 

 were used to obtain estimates of selection intensity (i) for latitude (0.65), longitude 

 (0.14), and elevation (0.21). The resultant percentages of similarity among populations 

 at various units of distance involving one degree latitude, one degree longitude, or 

 100 meters elevation are: 



Units of 



geographic Percent similarity for 

 distance Latitude Longitude Elevation 



0.5 82 98 96 



1 63 95 90 



2 28 88 75 



3 8 75 60 



4 -- 56 50 



Morgenstern and Roche (1969) show that their assumptions (/z^ = 1 and ^y.^ ~ cj^) 



introduce bias that maximizes estimates of the selection intensity (i) . Accordingly, 

 the percentages of similarity presented above are minimal estimates. 



DISCUSSION 



Freezing tests showed variation in hardiness among populations of Douglas-fir 

 during the two phases of cold acclimation. Yet hardiness of populations during the 

 first phase (autumnal dormancy) was not related (r = -0.05) to hardiness during the 

 second phase, which is characterized by low levels of cold hardiness. Moreover, 

 regression models accounted for significant proportions of the variance in hardiness 

 for only phase one. Consequently, it appears that selective differentiation of 

 populations is systematically related to geography and ecology for physiological 

 conditioning to the first frost (phase one) . Little variance differentiates populations 

 in cold acclimation after the first frost (phase two) , and most of that variance 

 seems to be random. As such, the variation could be related to random events within 

 the history of the population rather than to environmental conditions systematically 

 related to geography and ecology. 



The latitude and elevation of the seed source seemed to control differentiation 

 of populations for hardiness during phase one. Longitude and habitat type had little 

 effect. For habitat types to have little influence on differentiation is contradictory 

 to previous results (Rehfeldt 1974, 1979) that related population differentiation to 

 Abies tasiooanpa habitat types. However, one of the populations in the present study 

 represented an Abies lasiocavpa habitat type from a small frost pocket at a relatively 

 low elevation (980 m) . Contrary to a previous report (Rehfeldt 1974), levels of 

 hardiness for this population corresponded to those of populations from similar 

 latitudes and elevations rather than to those from subalpine environments at relatively 

 high elevations. Finally, that none of the regression coefficients for constant 

 terms (table 1) were statistically significant also suggests a lack of effects of 

 habitat types on genetic differentiation in cold acclimation. 



9 



