that parallel a change of 80 days in the frost-free period 

 across 1,000 m elevation (Baker 1944). Likewise, the 

 geographic patterns of variation illustrated in figxire 4 

 unmistakably parallel geographic climatic patterns of 

 figure 2. As a result, geographic patterns of genetic varia- 

 tion are strongly influenced by the Bitterroot Range and 

 Salmon River Drainage. 



Thus, populations from mild environments exhibit a high 

 innate growth potential that develops from a long duration 

 of elongation. Those from severe environments exhibit the 

 low growth potential and short duration of elongation ex- 

 pected in populations adapted to a short growing season. 

 For these clines to reflect adaptive differentiation is un- 

 questionable; rejection of such indirect evidence requires 

 acceptance of an untenably improbable alternative: that 

 the systematic patterns have developed randomly. 



Although direct evidence for adaptive clines ultimately 

 resides with the survival or death of indi\iduals, the 

 demographic cycles of population establishment in P. con- 

 torta endow an adaptive value to all traits that condition 

 growth and development in a particular environment. 

 Thus, direct support of adaptive clines is provided by the 

 differential incidence of snow damage, needle cast, and 

 mites, as well as by differences in growth potential itself. 

 At high elevations, such as PREF 1,500 where snow ac- 

 cumulations commonly exceed 3 m, populations from mild 

 environments suffered considerable damage from the 

 snow. Mean differences calculated within populations 

 showed that the 7-year shoot elongation of trees damaged 

 by the snow was 22 percent less than that of trees not 

 damaged. Consequently, damaged trees had an adjusted 

 height that was 5 percent less than undamaged trees from 

 the same population. 



Similarly, needle cast is a disease most common in 

 relatively humid valleys (Krebill 1975), and therefore, 

 populations native to either high elevations or arid 

 climates are rarely exposed to the disease. The present 

 data, as well as those of Hoff (1985), show that popula- 

 tions from either high elevations or southern latitudes are 

 much more susceptible to needle cast than populations 

 from low elevations. Mean differences within populations 

 depict a tremendous reduction in 7-year height for 

 diseased trees. For each increase in the damage score of 

 one unit— an increase of approximately 25 percent in 

 infected leaves per tree— 7-year height was reduced about 

 4 percent. This means that reductions in 7-year height 

 between uninfected and severely infected trees from the 

 same population w^ould amount to nearly 17 percent. This 

 reduction accrued largely from a corresponding reduction 

 of 34 percent in shoot elongation during year 7. 



Mites receive no recognition in pest surveys for Rocky 

 Mountain conifers (Tunnock and others 1984; Gibson and 

 others 1984). But when populations are transferred large 

 geographic distances, infestations by mites increase sub- 

 stantially. Because mites generally deform terminal shoots, 

 growth and development are affected greatly. 



The components of adaptedness tend to be integrated by 

 values of adjusted height. By representing growth from a 

 common height at age 5, adjusted height was free of most 

 genetic and environmental effects that had accrued up to 

 that age. Consequently, the variable was capable of ex- 

 pressing adaptation of populations to particular environ- 



ments over a short period. At low elevation, populations 

 from mild environments had larger adjusted heights than 

 populations from severe environments. A part of this dif- 

 ference was due to the high innate growth potential of 

 populations from low elevation, and a part was due to the 

 high susceptibility to needle cast of populations from high 

 elevations. At PREF 1,500, populations from about 

 1,500 m had the largest adjusted heights; populations from 

 extremely high elevations displayed an innately low- 

 growth potential, while populations from low elevations 

 suffered snow damage. At Lost Valley, however, environ- 

 mental effects sufficient for adaptively differentiating 

 populations evidently are only beginning to be expressed. 

 Although southern populations of low growth potential 

 displayed the smallest adjusted heights, superior perfor- 

 mance of relatively local sources cannot yet be detected 

 (fig. 3). 



Adaptive differentiation is readily quantified from 

 regression statistics. Table 5 shows that elevational clines 

 are particularly steep for 7-year height, late growth, 

 needle cast, and snow damage. Populations separated by 

 1,000 m are expected to differ by 16 cm (33 percent of the 

 mean) in 7-year height largely because of a large differ- 

 ence in the amount of late growth. These same popula- 

 tions, when planted at low elevation, are expected to 

 differ by 48 percent in leaves infected with needle cast. 

 When planted at high elevation, they are expected to dif- 

 fer by 28 percent in trees damaged by the snow. Previous 

 studies (Rehfeldt 1980) also predict that populations 

 separated by 1,000 m will differ by 31 percent in trees 

 damaged by a fall frost that causes a mean injury of 

 50 percent. Elevational clines are so steep that populations 

 separated by merely 224 m tend to be genetically differen- 

 tiated (95 percent level of probability) for late growth 

 while those separated by 500 m are differentiated for 

 numerous traits (table 5). 



The elevational cline, moreover, is the dominant cline 

 (table 5). For those variables in which differentiation was 

 pronounced (height, late growth, needle cast, and snow 

 damage), the amount of differentiation associated with 

 1,000 m of elevation varies from half to twice that asso- 

 ciated with 7° latitude at a constant elevation. Conse- 

 quently, differentiation per unit distance is much greater 

 along the elevational cline than along the geographic cline. 



Nevertheless, both clines arise from environmental selec- 

 tion along climatic gradients. Because the frost-free period 

 decreases by 80 days across an elevational interval of 

 1,000 m (Baker 1944), an average difference of only 

 18 days in frost-free period seems sufficient for inducing 

 differentiation of populations for late growth, a variable 

 reflecting tolerance to early fall frosts. Consequently, the 

 steep elevational cline reflects the rapid change in frost- 

 free period associated with elevation. And the gentle 

 geographic cline arises from a gradual geographic gradient 

 in the frost-free period. 



These results have direct practical application. First, 

 quantitative estimates of differentiation along adaptive 

 clines define the risks involved with seed transfer in artifi- 

 cial reforestation. Reforestation goals involve increasing 

 productivity while maintaining adaptiveness. To accom- 

 plish this, limits of seed transfer must reflect adaptive 

 clines. The steep elevational chnes described in this study 



8 



