and nutrients are the important attributes for vegetation 

 (Daubenmire 1970; Loucks 1962). None of these are 

 adequately assessed by current soil taxonomic systems. 



McCune and Allen (1985) were unable to statistically 

 relate site characteristics to climax tree species along the 

 eastern front of the Bitterroot Range in western Montana. 

 They attributed only 10 percent of the compositional vari- 

 ation to measured site factors, assigning the rest of the 

 variation mostly to historical factors. 



Hann (1982) described three site types for both a for- 

 ested and nonforested habitat type in western Montana. 

 Although all soils classified to two closely associated fami- 

 lies, Hann stated that considerable variation was found 

 between sites. He qualitatively describes a number of 

 soil-parent material -environmental conditions which, in 

 his study area, relate very well to differing successional 

 communities and specific habitat types. 



In classifying sagebrush-grass habitat types of southern 

 Idaho, Hironaka and others (1983) conducted a more 

 intensive but similar qualitative analysis of the 

 vegetation-soil relationship. Where soil-series-level 

 classifications were available, correlation between the soil 

 series or series-phase and habitat type was discussed. 

 Statistical analysis of the physical and chemical data 

 collected during this study would have greatly increased 

 the knowledge of individual and combined soil character- 

 istics relative to the vegetation being supported. Even 

 without this further analysis, this study is the most inten- 

 sive of regional plant communities and soil relationships 

 thus far published for the Western United States. 



In a study of the major plant communities of the 

 Guadalupe Mountains of Texas and New Mexico, Bunting 

 (1978) conducted an extensive analysis of topoedaphic 

 variables as predictors of potential natural vegetation 

 groups. In addition to physical site and soil descriptions, 

 samples were analyzed for organic matter, pH, N0 3 ", P 2 6 , 

 KjO, Mg**, Na*, CaO, total soluble salts, and carbonate 

 reaction. Discriminant function classification of stands 

 achieved 90-95 percent accuracy by using a combination 

 of topographic and edaphic variables. 



Tisdale and Bramble-Brodahl (1983) conducted a statis- 

 tically based, intensive study of vegetation communities 

 and soil along the Salmon and Snake Rivers. On their 

 study area, much reduced in geographic scale compared to 

 either the Hironaka and others (1983) or Bunting (1978) 

 studies, they concluded the currently available vegetation 

 and soil classification systems are not compatible, possi- 

 bly due to a relative difference in scale. Soil units are 

 divided much more finely than vegetational units. A 

 second part of the Tisdale and Bramble-Brodahl study 

 analyzed 16 individual site and soil factors as independ- 

 ent variables for modeling vegetation-site relationships. 

 Discriminant function classification accuracy ranged from 

 85 to 100 percent. Of the leading six factors, the most 

 important (elevation and radiation index) were site loca- 

 tion and orientation dependent. The other four factors 

 were soil related. They concluded that a satisfactory set 

 of soil-site variables could be developed to identify the 

 habitat type of a site, even though only serai vegetation 

 might be present. 



In Major's (1951) factorial approach to plant ecology, 

 the same five functional factors that Jenny applied to soil 



formation were used as independent formative factors in a 

 vegetation equation. Major concluded . .there are no 

 universal correlations between vegetation and soil;. . .soil 

 is not determined by vegetation, vegetation is not deter- 

 mined by soil; vegetation and soil develop concomitantly." 

 I hasten to submit at this point that even though no uni- 

 versal relationships appear to exist between soil and vege- 

 tation, it is exactly this concomitant development in a 

 localized area that should provide quantifiable character- 

 istics by which we can understand the plant community 

 and soil-forming processes. 



The objectives of this study were: to investigate numeri- 

 cal taxonomic techniques for analysis of patterns of physi- 

 cal soil characteristics; to investigate the relationship 

 between known habitat types and soil units created by 

 numerical taxonomy; and to develop the ability to predict 

 habitat type using physical soil characteristics. Due to an 

 acknowledged incompatibility of classification systems, 

 this study, unlike those of Daubenmire and Hironaka and 

 others, did not dwell on attempts to correlate habitat 

 types and soil family or series units. With knowledge of 

 the correlations between climax vegetation, soil, and site 

 characteristics within a specific geographic region, we 

 should be able to more accurately classify any given site 

 within that region to habitat type and phase. This will 

 also improve the ability to identify highly disturbed serai 

 vegetation stages to habitat type and phase and more 

 accurately position them within their successional devel- 

 opment pathway. 



THE STUDY AREA 



The study area comprised northern Idaho from the 

 Salmon River to the Canadian border (fig. 1). Sampling 

 was done on five National Forests (Kaniksu, 



Boise 



Figure 1 — Study area comprised Idaho 

 panhandle north of the Salmon River. 



2 



