Twigs were stored overnight at 3°C. The following morning, one set of twigs 

 from each population was frozen at a rate of 5°C/h to one of six test temperatures. 

 The six test temperatures for a given date spanned a temperature range of 10° to 

 12°C. Each test temperature was maintained for 1/2 hour; twigs were thawed for 24 

 hours at 2°C, and basal portions of twigs were placed in water within plastic drinking 

 cups. After 5 to 7 days, freezing injury to each twig was scored according to 

 discoloration of leaves. For each population, the number of twigs exhibiting leaf 

 injury was recorded for each test temperature. 



Tolerance of populations to freezing was assessed by regression analyses that 

 were made for each population according to the logistic model. This model, suitable 

 for proportional data between values of zero and one (Johnson and Kotz 1970) , is of 

 the form: 



Y.. = 1 



1 . be-^'^j 



which is represented by the linear model: 



In 1) = -rX .+ In b 



where : 



Y.. = proportion of twigs injured for population i in treatment severity j'^ 



X. - index of freezing severity = percentage of twigs from all populations 

 J 



injured in freezing treatment j. 



b = 1 where y is the predicted damage if no freezing treatment is applied 



^0 ° 



V = rate of increase in damage associated with an increase in severity index. 



Separate analyses v\?ere made for each population on data obtained before the 

 first frost (phase one of cold acclimation) and after frost (phase two) . Since 

 there were only two sampling dates before the first frost, regressions for phase one 

 were based on 12 observations, and those for phase two were based on 24 observations. 



Tolerance of each population to freezing during both phases of cold acclimation 

 was expressed as the injury predicted from the logistic model when freezing severity 

 (Z .) was 50 percent (50 percent of the twigs from all populations exhibited injury). 

 In'^fact, throughout this paper, these estimates of injury are used as expressions of 

 the hardiness of populations during cold acclimation. 



Differentiation of populations in relation to latitude, longitude, elevation, 

 and habitat type was assessed by multiple regression analyses. These analyses were 

 made to relate differentiation to geographic and ecologic variables of the seed 

 source so that hardiness could be used as a variable in delineation of seed zones 

 for northern Idaho and northeastern Washington. However, the tests included populations 

 from the Okanogan Mountains in north central Washington and from the Blue Mountains 

 in northeastern Oregon. Since these populations could represent adaptive provinces 

 that differ from the northern Idaho province in manners similar to the differentiation 

 of the latter province from those of western Montana (Rehfeldt 1979) and southern 



3 



