Although stands were used to select study areas, we were Interested in 

 sampling microsites within stands. A series of yaoo-acre plots were sampled 

 in each stand. After obtsdning stand histories and stereo aerial photographs 

 for each selected stand, either transects or a sampling grid was used to sys- 

 tematically locate about 25 plot centers per stand. In Idaho, transects crossed 

 through areas observable on aerial photographs that varied in such things 

 as aspect, overstory density, site preparation, and topographic position. In 

 Montana, a sampling grid was used to locate plot centers. 



Each sample point was the center of a Vaoo-acre circular fixed area plot and 

 a variable radius plot for sampling the overstory (if present) with a 10 basal 

 area factor prism. Each plot was made as independent as possible by recording 

 plot attributes — ^habitat type, slope, aspect, type of site preparation, topographic 

 position, and residual overstory density by species. Systematically spacing 

 plots also helped maintain independence among plots within stands because 

 plots were dispersed throughout the stand. The year of cutting, geographic 

 location, elevation, and budworm defoliation history were stand variables 

 that did not differ among plots within a stand. 



Plots within stands could represent different "years since last disturbance" 

 due to different years of site preparation. Untreated plots were last disturbed 

 when the stand was harvested. Plots mechanically prepared or burned were 

 last disturbed when the site preparation took place — either, or both, could 

 occur after the harvest year. Because the year of site preparation can take 

 place after the year of harvest, we use "years since last disturbance," rather 

 than stand age, to predict regeneration. 



Established conifer regeneration on each plot in the Idaho studies was 

 counted by species; in the Montana study up to 10 established trees were re- 

 corded per plot. Species included in the model are shown in table 1. Mountain 

 hemlock {Tsuga mertensiana) could not be included in the model because of 

 insufficient data. Table 1 also lists the minimum establishment heights by 

 species — these are 0.5 foot for shade-tolerant species and 1.0 foot for shade- 

 intolerant species. Minimum heights roughly correspond to the height of a 

 3-year-old tree. Maximum size for all regeneration was 2.95 inches diameter at 

 breast height (d.b.h.). Tree heights were measured as of the end of the pre- 

 vious growing season to ensure that all trees sampled during one field season 

 used the same standards for establishment. 



Tree height to the nearest 0.1 foot, tree age at groundline, and tree condi- 

 tion (damage, insects, disease, and so on) were recorded on a subsample of 

 regeneration caUed 'T^est trees." Best trees were chosen by the following rules: 



1. Select the two tallest trees on each Vaoo-acre plot regardless of species. 



2. Select the one tallest tree of each additional species present on the plot. 



3. If the first two rules do not total four trees, select in order of descending 

 height fi'om any remaining trees until at least four trees, if present, are chosen. 



One site preparation treatment was recorded for each Vaoo-acre plot. It was 

 the treatment that covered the largest proportion of the plot — mechanical, 

 bum, no treatment, or road. The "road" category included all road cuts, road 

 fills, and unmaintained roadbeds. Plots that occurred on maintained road- 

 beds were not sampled. The intensity of each site preparation was not re- 

 corded because this was more difficult to determine as years since disturbance 

 increased. 



National Forests were used as independent variables in data analyses be- 

 cause they divide the study area into units that can account for variation in 

 latitude, longitude, climate, and so on. 



4 



