The resulting 16 habitat type groups are shown in table 2. Analyses used 

 to develop the groups are found in appendix A. The groups combine habitat 

 types within climax series that are similar during the regeneration phase of 

 stand development. Habitat type groups would likely be different for other 

 purposes — diameter growth, mortality, canopy cover, and so on. 



Western Spruce The influence of western spruce budworm on regeneration success was a 



Budworm major focus of the studies conducted in Montana and Idaho during 1979-82. 



Defoliation Defoliation histories were collected by stand for the period 5 years before 



History harvest to the year prior to field sampling. 



In Montana, past budworm activity was indexed by analysis of increment 

 cores using the procedure of Carlson and McCaughey (1982). In Idaho, his- 

 torical aerial defoliation maps were used to determine which years, if any, 

 a stand was within a defoliated area. Using aerial defohation maps, it was 

 also possible to determine defoliation histories for stands in the 1975-76 

 study. The vast majority of stands from the 1975-76 study were not in areas 

 defoliated by budworm. 



The historical aerial defoliation maps outline areas of low, moderate, and 

 heavy defoliation as seen by an entomologist from an airplane. Boundaries 

 between levels of budworm defohation are approximate. There is also varia- 

 tion in defoliation within outlined levels of defoliation. We felt the appropri- 

 ate way to use budworm histories in developing a regeneration model was 

 as a dichotomous variable. For each year, the stand either was in a defoli- 

 ated area or it was not. This approach also allowed consistent use of data 

 collected in Montana and Idaho. 



Planting Both planted and unplanted stands were sampled. Sampling only naturally 



regenerated stands may have biased the sample, because stands that were 

 difficult to regenerate were more likely to have been planted. During the 

 first few years of data collection, crews attempted to identify planted trees on 

 plots, but this was not easy to determine retrospectively. Plots fi"om planted 

 stands were coded for analyses as a class (dummy) variable. Coefficients 

 estimated by regression analysis for planted stands are not used in the model, 

 so that predictions are only for natural regeneration. Model users then specify 

 planting for the stand, which is added to the natural regeneration predicted 

 by the model. They select the species, density, year of planting, and survival 

 of planted trees. 



Variables in the The variables used in the regeneration model are those generally recorded in 



Model forest inventories in the Northern Rocky Mountains. This means that most 



inventories are already designed to coUect the information needed to use the 



regeneration model, or could easily do so. 

 Table 3 provides definitions for variables used in the regeneration model. 



These names for variables are used in tables of regression coefficients 



(appendix B). 



During equation development, goodness of fit was evaluated at the 0.05 

 significance level. Nonsignificant independent variables were eliminated 

 fi'om equations except those we considered to be part of groups. An example 

 of a group is class variables for site preparations. The entire group was 

 kept in the equation if one of the coefficients was significant. 



8 



