These three kinds of reactions were similar to those de- 

 scribed by Hoff (1986). They also appear to be of the same 

 type reported by Allen and others (1988) for resistance in 

 lodgepole pine to western gall rust. Two other resistance 

 reactions, namely rate of gall appearance and variation of 

 rust-killed trees, were highly variable among provenances 

 but were not statistically different. 



Much of the variation among populations appeared 

 to be random. But, for the degree of infection, variation 

 among populations could be partly explained by geogra- 

 phy. The R 2 for elevation was 0.15. Generally, as eleva- 

 tion of populations increased so did the level of infection; 

 however, the relationship was curvilinear, with the level 

 of infection leveling off at about 4,000 feet. That low- 

 elevation sources are less susceptible is not surprising 

 because weather conditions appear to be ideal for infec- 

 tion, with the probable result of more intense selection 

 for resistance. 



The multiregression that gave the best fit was a combi- 

 nation of elevation, azimuth, distance from inoculum 

 source, and southwest departure. These accounted for 26 

 percent of the variation in infection. Most surprising was 

 that the predicted level of infection was higher near the 

 inoculum source, and then decreased in all directions with 

 distance from that point. If resistance was closely tuned 

 to the genetic variation of the fungus, the opposite pattern 

 would be expected. Possibly variation within the rust is 

 absent or very small and the differences among the popu- 

 lations, as indicated by the model, represent the interac- 

 tion of the environmental conditions and genetics of the 

 host. On the other hand, Powers and Matthews (1980) 

 felt that the fungus has adapted to the host and conse- 

 quently has produced races that are more virulent in the 

 local sites. I find this explanation easier to accept with a 

 system like loblolly pine; namely a fusiform rust that has 

 been excessively disturbed, than with a system like pon- 

 derosa pine; characterized by western gall rust, which has 

 had comparatively little disturbance. 



The small but significant negative correlation of the 

 level of infection and 3-year height is consistent with the 

 association of growth and elevation, namely low-elevation 

 populations grow faster and are less susceptible to west- 

 ern gall rust and high-elevation populations grow slower 

 and are more susceptible. The clines in Rehfeldt's (1986) 

 figure 2 and figure 4 in this paper show these relation- 

 ships. But the association between infection and geo- 

 graphic pattern as predicted by the model and growth 

 was not consistent because the fastest growing popula- 

 tions were from the west-central portion of the sample 

 area (Rehfeldt 1986). These populations were the most 

 susceptible, based on the predicted value, and from here 

 growth and of susceptibility both decrease outwards. 

 Rehfeldt's figure 3 on the geographic growth pattern and 

 the geographic pattern of rust susceptibility (fig. 4) are 

 almost identical. Thus there are two contrasting clines: 

 (1) With geographic area, susceptibility decreasing with 

 slower growth outwards in all directions from the inocu- 

 lum source and (2) susceptibility increasing with slower 

 growth due to an increase of elevation. I cannot explain 

 this difference; however, the higher elevation effect, as 

 shown by the much larger simple correlations, possibly 



explains why susceptibility and growth are negatively 

 correlated. And because infection is correlated with 

 growth when elevation is set to the mean of the popula- 

 tions, growth per se is not a causal factor in the differ- 

 ences in these patterns and is therefore possibly due 

 mostly to environmental patterns. 



Likewise it is hard to explain the association of long 

 needles with mortality of the fungus and mortality of 

 rust-killed trees. Rehfeldt (1986) found that the popula- 

 tions in the northwest portion of the sample area had the 

 shortest needles and that the length increased in a south- 

 easterly direction. Possibly these relationships are spuri- 

 ous. If real, more detailed work will have to be done to 

 explain them. 



The negative correlation of the trees that died of 

 drought with stem reactions (either a gall or bark reac- 

 tion) is also hard to explain. In other words, populations 

 that have a higher inherent resistance to drought were 

 more infected, that is more easily penetrated by the fun- 

 gus, whether the penetration produced a gall or was re- 

 pulsed by a bark reaction. This could possibly be due to 

 chemical aspects of the stem, fatty acids, or higher mois- 

 ture content of drought resistant cells — the fungus has 

 only to penetrate the cuticle to get into a "good" environ- 

 ment. On the other hand, the root structure of drought- 

 resistant provenances may produce moisture conditions 

 in the stem that are best suited for penetration by the 

 fungus. Answers will also have to wait for more detailed 

 analyses of the penetration process and the environ- 

 mental requirements of the process. 



A major difficulty with artificial inoculation is that the 

 trees are inoculated at one point in time. This coupled 

 with the variation of growth initiation and rate and 

 amount of growth among populations could lead to sus- 

 ceptibility patterns related to periodicity instead of inher- 

 ent resistance of each population. But correlations of the 

 growth data, either from the field or greenhouse, with 

 total stem symptoms (galls plus bark reactions) were all 

 very low and nonsignificant, indicating no relationship of 

 initial infection level with growth characters for popula- 

 tions. I assumed therefore that all the populations had 

 an equal chance of becoming infected during the period 

 of inoculation and that differences would be due to the 

 inherent genetic nature of the population in terms of sus- 

 ceptibility to gall rust and not to its periodicity. 



MANAGEMENT IMPLICATIONS 



In terms of tree improvement, there appears to be little 

 danger in moving seed within the middle Columbia River 

 system unless a particularly susceptible stand unknow- 

 ingly is chosen for the source of seed. Thus, it is impor- 

 tant for forest managers to closely inspect stands prior to 

 seed collection to determine the level of susceptibility to 

 western gall rust. Further, because some individuals are 

 highly susceptible, care must also be taken in selecting 

 seed trees for natural regeneration, or in selecting trees 

 to include in tree improvement programs. 



6 



