-I 



Methods 



In 1969, measurements were taken on 264 lodgepole pine trees growing within 88 

 permanent sample plots established between 1957 and 1960. The -plots represented a wide 

 range of growing conditions in unmanaged, even-aged lodgepole pine stands over an 

 extensive geographic area from northeastern Utah to northwestern Montana (fig. 1). 



Three trees of differing crown class were randomly selected from each plot for 

 individual -tree measurements. Measurements included total height, height, to crown base, 

 d.b.h., age at breast height, and 10-year radial increment. Tlie crown base was "visu- 

 ally balanced" to determine measuring point, and radial increment was determined as the 

 averaf.e of two increment core measurements taken at breast height. Cores were extracted 

 at 180° to each other, on the average diameter as located by tree calipers. 



lUameter-class tallies of the permanent plots furnished stand data at the time of 

 establishment and again in 1969. Stand values, such as number of trees per acre and 

 basal area, were linearly interpolated or extrapolated to 1959, so they would corres- 

 pond to the beginning of the growth period measured on increment cores. 



Increment of individual trees was analyzed both as radial increment, inside bark, 

 and as increment in basal area, inside bark. In addition, a logarithmic transformation 

 of each was used. To determine the most appropriate form of the dependent variable, a 

 large set of independent data from management planning inventories and Forest Survey 

 locations was utilized, employing Furnival's (1961) "Index of Fit" criterion. On this 

 basis, the logarithm of basal area increment was found to be the superior form of the 

 dependent variable. 



Linear equations of the form Y = + biXi + b2X2 + ••• b^X^^ were then fitted to 



our data and subjected to screening to determine the equations of best fit, wliere Y 

 represented the natural logarithm of basal area increment in square inches. Data were 

 analyzed using Grosenbaugh ' s (1967) Rex-Fortran IV system for combinatorial screening 

 and analysis of multivariate regressions. 



2 



