overstory and understory trees larger than 7 inches d.b.h. 

 in small blocks less than 2 acres in size; and (4) control, 

 where no harvesting was done. Overstory was defined as 

 the canopy stratum that included dominant and codomi- 

 nant trees, whereas understory included all other trees. 

 Unit size for treatments 1 and 2 varied from 3 to 6 acres, 

 and treatments 3 and 4 were about 1.5 acres. Volumes 

 removed in blocks 1 and 2, respectively, were 4,962 and 

 3,533 fWacre for treatment 1; 2,950 and 2,570 for treat- 

 ment 2; and 7,655 and 6,241 for treatment 3 (Benson and 

 Gonsior 1981). 



Ten Vso-acre circular plots were systematically estab- 

 lished in each treatment, spaced about equally through- 

 out the area. All trees on the plots were numbered with 

 metal tags and measured in 1973, prior to treatment. We 

 remeasured the trees in summer 1985 using the 1984 

 terminal bud scar as the reference for current height. 

 Thus, the time interval for height growth was 11 years. 

 But because d.b.h. measurements necessarily included 

 1985 cambial growth, the interval for diameter growth 

 was 12 years. We sampled only the first 10 trees encoun- 

 tered on the plot, beginning at north azimuth and pro- 

 ceeding clockwise, because preliminary data indicated 

 that an adequate sample for assessing growth response 

 could be obtained in this way. Furthermore, in 1985 we 

 sampled only trees not killed or seriously damaged by the 

 logging process because Benson and Gonsior (1981) had 

 already reported on that aspect of the study. 



Variables recorded for each tree were: 



1. Species. 



2. Degree of overstory removal. Coded 1 if partial re- 

 moval; 2, if total removal on treatment sites greater than 

 3 acres; 3, if total removal on treatment sites smaller than 

 3 acres; and 4, for no overstory removal, the control. 



3. Total height, to nearest 1 foot. 



4. D.b.h., nearest 0.1 inch. 



5. Length of live crown, nearest 1 foot. 



6. Crown class (dominant, codominant, intermediate, 

 suppressed, or open grown). 



7. Mortality not caused by logging. 



8. Ten-year radial growth prior to 1973, inch. 



9. Percent of foliage removed by western spruce 

 budworm, by crown thirds, visual estimate. 



For item 8, increment cores were taken at breast height 

 from plot trees 1 inch d.b.h. and larger to determine age 

 and 10-year radial increment to the nearest 0.0004 inch 

 for the period 1964-73. Smaller trees were 

 a special problem because we felt that extracting a core 

 would alter their growth. And disks could not be taken 

 from the small permanent sample trees because that 

 would have destroyed them. Therefore, we collected disks 

 from small trees (surrogates) nearby, but not within the 

 plots, so we could develop predictive models for prerelease 

 10-year radial growth of small trees on the permanent 

 plots. Twenty trees each of Douglas-fir, Engelmann 

 spruce, and subalpine fir within each of five height classes 

 were selected without bias. Classes were to 2.0, 2.1 to 

 3.0, 3.1 to 4.0, and 4.1 to 5.0 feet. Total height and crown 

 length were recorded for each tree. Disks were cut trans- 

 versely just above the root collar of each tree, placed in 



plastic bags, and frozen at the end of each field day. Ten- 

 year radial increment was measured in the laboratory to 

 the nearest 0.0004 inch. 



DATA ANALYSIS 



We used multiple linear regression analyses to develop 

 predictive models for 11-year height and 12-year diameter 

 growth and to test statistical significance of independent 

 variables. If needed, data were transformed to meet as- 

 sumptions underlying regression analysis. All variables 

 except degree of overstory removal were continuous; 

 dummy variables were created for the categories of this 

 variable. Variables were deemed significant at P < 0.05. 

 Similarly, the influence of tree and plot variables on defo- 

 liation by budworm was tested using multiple linear re- 

 gression procedures. Multiple linear regression was per- 

 formed on the data from the small surrogate trees, using 

 10-year radial growth as the dependent variable and 

 height and crown length as independent variables. Sepa- 

 rate regressions were developed for each species. Missing 

 values for small trees on the permanent plots were then 

 computed using coefficients developed in the regressions. 

 Mortality was analyzed using a linear modeling procedure 

 for categorical data (Grizzle and others 1969). All data 

 management and analyses were done with SAS proce- 

 dures (SAS Institute 1985). 



RESULTS 

 Height Growth 



Degree of overstory removal had no effect on post-re- 

 lease height growth of advance Douglas-fir regeneration, 

 but increasing change in basal area had a positive influ- 

 ence, reflecting the release from competition. Basic statis- 

 tics for all variables are given in table 1 and regression 

 models for estimating 10-year prerelease radial growth for 

 small trees are presented in table 2. (Height and crown 

 length were significant predictors (P ^0.05) of prerelease 

 radial growth for all three species.) Tree age was not used 

 in any of the analyses because the data had been mis- 

 placed and could not be recovered. The natural log of 

 postrelease 11-year height growth was directly related to 

 10-year radial increment prior to treatment (table 3); 

 height growth increased with larger radial growth (fig. 1). 

 Defoliation by western spruce budworm tended to depress 

 height growth of Douglas-fir. Figure 2 shows the com- 

 bined effects of prerelease radial growth and defoliation 

 on postrelease height growth, and figure 3 shows the 

 influence of change in plot basal area. The height growth 

 model for Douglas-fir was not very strong; the coefficient 

 of multiple determination (CD) was only 0.20. 



Postrelease height growth of Engelmann spruce in- 

 creased with increasing prerelease radial growth (CD = 

 0.15), but none of the other variables tested exerted a 

 significant influence (table 3, fig. 1). 



2 



