Table 6.— Slopes of lines regressing relative vegetation cover on number of passes 

 for each habitat type^ 



Intercept Slope Signif- 



Habitat type (a) (b) SEE icance 



Festuca scabrella- 



F. idahoensis 95.2 -4.3a 0.15 8.9 <0.001 

 Abies lasiocarpa/ 



Xerophyllum tenax ,99.3 -18.7b .40 19.8 < .001 

 Abies lasiocarpa/ 



Vaccinium caespitosum 98.4 -23.5c .57 17.7 < .001 

 Pseudotsuga menziesii/ 



Symplioricarpos albus 93.2 -24.5c .73 12.8 < .001 

 Abies lasiocarpa/ 

 Clintonia uniflora- 



Vaccinium caespitosum phase 102.7 -27.2cd .83 10.6 < .001 

 Abies lasiocarpa/ 



Clintonia uni flora 98.2 -28.8cl .86 10.0 < .001 



^The exact form of tfie regression equation used was Y = a + b log X, where Y is relative 

 cover (inverse sine transformed) and X is number of passes. Thie most resistant habitat types 

 have the least negative slope values. Two slope values followed by the same letter are not 

 significantly different {a = 0.05). SEE is the standard error of the estimate. 



show the high resistance of FESC-FEID (fig. 12a, b) and 

 the relatively high resistance of ABLA/XETE. The other 

 four types are similar enough in response to make 

 differentiation difficult. Of these four, ABLA/VACA 

 seems to be somewhat more resistant than the rest and 

 ABLA/CLUN seems to be more fragile, but the differ- 

 ences are too slight to have any real significance to 

 management. 



One question of considerable interest is whether differ- 

 ences in habitat type or differences in amount of tramp- 

 ling are more important determinants of amount of 

 vegetation loss. Clearly the answer would depend upon 

 the habitat tj^^es being compared and the range of tramp- 

 ling pressure applied. If similar vegetation types are 

 subjected to a wide range of use, amount of use will be 

 the more important factor. In contrast, if quite different 

 vegetation types are compared across a narrow range of 

 use— such as 100 versus 200 passes— habitat type is 

 likely to make more difference. The importance of 

 amount of trampling is also highly dependent upon the 

 number of passes. For example, in most habitat types 

 differences in amount of trampling would be much more 

 important when comparing the effects of five and 200 

 passes than when comparing the effects of 600 and 800 

 passes. 



Across the range of trampling from zero to 1,600 

 passes, the difference between the grassland and any of 

 the forested types explained more of the variation in 

 relative cover than differences in amount of trampling. 

 That is, in a multiple regression with pairs of habitat 

 types and number of passes as independent variables 

 and relative cover as the dependent variable, the r^ con- 

 tribution of habitat type was higher (0.31 to 0.63 

 depending upon which forested type the grassland was 

 compared to) than the r^ contribution of number of 

 passes (0.14 to 0.19). 



Amount of trampling was more important than differ- 

 ences between any of the forested types across this wide 



range of trampling intensity. If narrower ranges of tram- 

 pling are compared, however, some forested habitat type 

 differences become more important. From figures 10a 

 and lOd, it can be seen that between 300 and 1,600 

 passes, the difference in relative cover related to a differ- 

 ence in number of passes is 11 percent in the ABLAVCLUN 

 type and 17 percent in the ABLA/XETE type. The 

 difference in relative cover between these two types 

 varies from 26 to 60 percent, between 300 and 1,600 

 passes. This suggests that whether use occtirs in 

 ABLA/CLUN or ABLA/XETE has a more profound in- 

 fluence on amount of cover loss than whether amount of 

 trampling is 300 or 1,600 passes. Numerous similar 

 interpretations can be made from these data. 



Effect of Local Variations in Species 

 Composition 



As mentioned previously, local variations in the initial 

 species composition of different treatment lanes on the 

 same plot appear to have a significant effect on cover 

 loss. Unfortunately, this variability could not be con- 

 trolled and instead appears as "noise," making it more 

 difficult to interpret the effects of the controlled vari- 

 ables. Nevertheless, the nature and magnitude of this 

 variability is of considerable interest itself. 



Let us assume that any time relative cover increases 

 with an incremental increase in number of passes (refer 

 to fig. 10), this increase— the opposite of the general 

 trend— is primarily a result of proportionally more resis- 

 tant species on the more heavily trampled lane. This is a 

 reasonable assumption given that the major variables af- 

 fecting cover loss should be amount of trampling and 

 the resistance of the plants being trampled. Moreover, 

 most of the treatments with unexpectedly high cover 

 values did have large percentages of resistant species. 

 For example, an analysis of covariance for the ABLA/ 

 XETE type showed the amount of variation in relative 



18 



