for sampling among the approximately 470 available: 



a) land use (from Fig. 23 of Montagne et al. 1982) - 

 quadrangles falling predominately (i.e., over 50%) 

 in agricultural land uses were excluded from fur- 

 ther consideration 



b) average annual precipitation (from Sheet 2 of Ross 

 and Hunter 1976) - three classes were subjectively 

 defined, i.e., <12 inches, 12 - 16 inches, >16 

 inches. This attribute was regarded as an indica- 

 tor of moisture regime. 



c) average length of freeze-free season (from Fig. 13 

 of Montagne et al . 1982) - three classes were 

 subjectively defined, i.e., <100 days, 100 - 120 

 days, and >120 days. This attribute was regarded 

 as a indicator of temperature regime. 



d) surficial geology (from Figs. 9, 13, 17, 21, 23, 



25, and 32 of Veseth and Montagne 1980) - the six 

 classes represented by the Veseth and Montagne 

 figures were used (Figs. 21 and 23 were subjec- 

 tively merged) . This attribute was regarded as a 

 indicator of nutrient regime. 



e) Radiation regime was not considered in this 

 process since it varies greatly at relatively fine 

 geographic scales for different slopes and 

 aspects, particularly in complex terrain. Plot 

 selection in the field attempted to include a wide 

 range of slope/aspect combinations in each samp- 

 ling area. 



A total of 175 plots v;ere targeted for sampling based on the 

 time available for this study (note: only 125 plots were ulti- 

 mately sampled) . A total of 5 plots/selected quadrangle were 

 chosen as a reasonable average to represent local-scale patterns 

 in community composition. Thus, 35 quadrangle maps were selected 

 for sampling (i.e., 5 x 35 = 175). 



After eliminating agriculturally dominated quadrangles from 

 the pool (this reduced the number of quadrangles from about 470 

 to 221) , a matrix of precipitation/freeze-free classes was 

 constructed and the number of quadrangles in each class was 

 recorded. The percentage in each class relative to the total 

 number of quadrangles (221) v;as used to determine the number of 

 quadrangles (by class) to be included in the pool to be sampled 

 (e.g., 25% in class Z x 35 sample quadrangles = 9 plots of class 

 Z in the sample pool) . 



