The samples were stored in the iab at -25° C. Each of the 

 first two phloem samples were separated from the bark and 

 ground in a Wiley grinder at 20-mesh size by freezing the 

 sample in liquid nitrogen and by passing large amounts of dry 

 ice through the grinder to keep the grinder cold. 



Soluble nitrogen and soluble sugars were extracted with 80 

 percent ethanol. Insoluble products underwent chemical di- 

 gestion in order to convert them into a soluble form that could be 

 analyzed. Insoluble nitrogen in the sample was converted to 

 ammonia by repeated digestion with a 20 percent sulfuric acid 

 and cleaned with hydrogen peroxide (Hodges and others 1 968). 



To analyze terpenes, 0.2 to 0.3 g of the ground phloem was 

 placed in a vial with 2 ml of isopropyl ether (free of alcohols, 

 chromatoquality reagent) in a sealed vial and shaken for at least 

 2 days. We then put 10 microliters of this solution in a Varian 

 Aerograph series 1 700 gas chromatograph with a flame ioniza- 

 tion detector. The identification and quantity of each component 

 was determined by running dilute standards of the pure compo- 

 nents. The peaks were cut out, and the quantity of each compo- 

 nent determined from its peak weight. We used a 1 .83-m col- 

 umn packed with Porapack Q because the water in the sample 

 from the phloem did not affect this column packing. The injector 

 temperature was 275° C, detector temperature 250° C, carrier 

 gas (high purity helium) 40 psi, and column temperature was 

 programmed from 50° to 250° C at 10° per minute. Ultra high 

 purity hydrogen and air were used for hydrogen detection. 



Laboratory analysis was focused on monoterpenes, soluble 

 nitrogen, total nitrogen, reducing sugars, starches, pentoses, 

 and hexoses. Nitrogen was analyzed by the colorimetric Mes- 

 sier Method (Jacobs 1965). Insoluble nitrogen was determined 

 as the difference between total nitrogen and soluble nitrogen. 

 Sugars, hexoses, and pentoses were determined at the same 

 time with the cysteine and sulfuric acid general reaction on 

 carbohydrates (Dische 1955). Their absorption spectra were 

 then read at 320 mu and 405 mu, which allows the determina- 

 tion of both sugars. Reducing sugars were determined by 

 methods discussed in Dische (1955). Starches were hydro- 

 lyzed and then determined by the same procedure as the 

 sugars. 



DATA ANALYSIS 



Presented in table 1 are average percentages of the phloem 

 (by weight) and associated standard deviations found in dry 

 matter, sugars, starch, nitrogen, and monoterpenes for the 

 trees sampled. Means for June 6 were compared statistically 

 with comparable means at (a) July 10 and (b) July 31. 



To establish possible links between tree characteristics and 

 phloem constituents, the latter were fitted as linear functions of 

 all combinations of six pertinent tree characteristics: d.b.h., 

 percent crown length of total tree height, height, phloem thick- 

 ness, average radial growth for the 5 years prior to sampling, 

 and age. 



Results of the regression screen are summarized in table 2 

 and show that rather weak regression information (R^) was 

 developed throughout. The July 31 monoterpenes were, 

 however, most strongly related to the tree characteristics evalu- 

 ated. While even the strongest of these, phloem depth and 

 growth, seem of marginal strength (0.14 R^ « 0.34), they do 

 confirm the presence of associated linear, positive increases in 

 monoterpenes. The results provide an information base neces- 



Table 1.— Selected lodgepole pine phloem constituents, percent 

 by weight. 



June 6 July 10 July 31 

 Constituent x s 7 s x s 



Percent of total phloem weight 



Dry matter 



54.6 



12.7 



45.2 



4.1 



41.4 



4.3 





-Percent of phloem 



dry matter weight- 



Soluble pentoses 



2.0 



.56 



6.0 



1.16 



4.7 



1.02 



Soluble hexoses 



1.5 



.43 



4.3 



.61 



6.3 



1 .18 



Total 



3.5 



.96 



10.3 



1 .61 



1 1 .0 



1 .74 



All pentoses 



3.7 



.98 



11. 4 



2.72 



8.2 



1 .44 



All hexoses 



2.7 



.81 



7.8 



1 .60 



1 1 .6 



2.97 



Total 



6.4 



1.62 



19.2 



3.57 



19.8 



3.30 



Soluble reduced sugars 



3.3 



1.70 



2.2 



.82 



1.8 



.82 



Starch 









O. 1 O 



CS.Ci 



O QT 



iLM 1 



Insouble nitrogen 



.13 



.04 



.11 



.02 



.11 



.02 



Total nitrogen 



.18 



.07 



.13 



.03 



.12 



.03 



Monoterpenes 















a-pinene 



.052 



.058 



.039 



.033 



.030 



.026 



p-phellandrene 



.203 



.277 



.140 



.124 



.144 



.127 



3-terpenes 



.120 



.192 



.077 



.067 



.064 



.054 



(3-carene + 















myrcene + 















a-pinene) 















Total 



.375 



.447 



.256 



.203 



.238 



.191 



sary to the development of more advanced hypotheses, to be 

 evaluated with new data when available. In this case, an in- 

 teractive hypothesis was developed from the July 31 data. We 

 used two of the variables exhibiting the strongest linear effects 

 (phloem thickness and growrth) and one weak variable (d.b.h.) 

 that proved reasonably strong in past MPB dynamics models. 



Here, "total terpene" data for July 31 were partitioned over 

 the ranges of phloem thickness, tree growth, and tree d.b.h. and 

 were explored graphically for interactive effects. The data 

 appeared to support a three-way interaction characterized by: 

 positive, shallow concave-upward effects for phloem thickness 

 and growth; a more-or-less bell-shaped effect for d.b.h., max- 

 imizing at about 10.5 inches (26.67 cm); and convergence to 

 zero with low growth and phloem thickness. The d.b.h. effect is 

 not oriented at zero but is not meaningful at zero anyway. These 

 effects were in general accord with the mountain pine beetle 

 preference for larger, more vigorous trees, although the rather 

 strong negative trend in terpene content for larger trees — 

 d.b.h. > 10.5 inches (26.67 cm) — was not. Nevertheless, 

 d.b.h. was retained in the model and the resulting four- 

 dimensional relation was formulated mathematically using the 

 techniques specified by Jensen (1 973, 1 976, 1 979) and Jensen 

 and Homeyer (1 970, 1 971 ), and was refitted to the data set from 

 which it was partially derived, by weighted^ least squares. The 

 final hypothesized form (R^ = 0.39, Sy x = 0.15) is shown 

 graphically in figure 1 and mathematically in appendix table 7. 



Variance about the initial model Y was expressed as a function of Y. The 

 inverse of this, 1/Y^^, was used as the fitting weight. 



2 



