based on Friedman rank sums were used to identify 

 which of the trampling treatments were different from 

 the control and from each other and which habitat types 

 differed in their response (Hollander and Wolfe 1973). 



For all habitat types combined, exposure increased sig- 

 nificantly (p < 0.001) as number of passes increased. 

 The treatment of 100 passes and treatments of 300 

 passes and above resulted in significantly more exposure 

 of mineral soil than on the control {a = 0.05). Differ- 

 ences in exposure were very small; this is underscored 

 by the result that none of the individual trampling treat- 

 ments differed significantly. 



For individual habitat types, mineral soil exposure in- 

 creased significantly as trampling increased on the 

 ABLA/XETE, PSME/SYAL, and FESC-FEID habitat 

 types. Although increases were as great as 13 to 15 per- 

 cent after some treatments, results were so erratic (refer 

 to high standard deviations in table 9) that there were 

 no statistically significant differences between treat- 

 ments and control or among treatments for any indi- 

 vidual habitat type. 



The resistance of these habitat types to mineral soil 

 exposure appears to be affected primarily by depth of 

 the organic horizons and the slope of the plot perpendic- 

 ular to the treatment lanes. Soil exposure was greatest 

 on the PSME/SYAL habitat type, which had a slope of 

 20 to 25 percent. All of the other habitat types were 

 almost flat. Walking perpendicular to this slope cut ter- 

 raced paths into the slope, exposing more mineral soil 

 than if there had been no slope. For the other types, 

 those with the deepest organic horizons, ABLA/CLUN 

 and ABLA/CLUN-VACA, had the least exposure; those 

 with the shallowest horizons, ABLA/XETE and FESC- 

 FEID, had the most. Despite how readily apparent these 

 differences are in the field and from the results in 

 table 9, the only statistically significant difference 

 between habitat types (a = 0.05) is that ABLA/CLUN 

 experienced less mineral soil exposure than 

 PSME/SYAL. 



To generalize, 1,600 passes are sufficient to reveal a 

 general tendency for increased trampling to lead to 

 increased exposure of mineral soil. Increases are highly 

 erratic, difficult to predict, and generally not very pro- 

 nounced. This suggests that, in contrast to vegetation 

 loss, there is considerable resistance to soil exposure and 

 that it would take either prolonged trampling or much 

 higher levels of trampling to cause pronounced increases 

 in mineral soil exposure. This corroborates a study of 

 existing campsites that found soil exposure to be one of 

 only a few types of impact that only become serious 

 after use of campsites exceeds low-use levels (Cole and 

 Fichtler 1983). 



EFFECT OF TRAMPLING ON SOIL 

 COMPACTION 



Penetration resistance, as measured with a pocket soil 

 penetrometer, was used as a measure of soil compaction. 

 Penetration resistance readings are affected by a number 

 of soil characteristics, but particularly by differences in 

 soil moisture. Because of this, a comparison of pre- and 

 posttreatment readings would include a sizable compo- 

 nent of change that was unrelated to the treatment itself. 



Therefore the measure of change used was simply the 

 penetration resistance of the treatment subplot minus 

 the mean penetration resistance of the control lane. 



A two-way analysis of variance again showed that 

 both number of passes and habitat type influenced 

 increase in penetration resistance and that there was a 

 significant interaction between these two main effects 

 (p < 0.001). In this case, variances were made more 

 homogeneous with a logarithmic (base 10) transforma- 

 tion. Figure 17 graphs the relationship between compac- 

 tion and trampling intensity for each habitat type and 

 table 10 shows the resistance of each habitat to each 

 level of trampling. 



It is clear that tramphng generally increased penetra- 

 tion resistance; only a few values were negative. More- 

 over, with the exception of the ABLA/CLUN habitat 

 type, there was a clear tendency for resistance to 

 increase as trampling increased. As with vegetation loss, 

 this relationship was a curvilinear one in which the 

 increase in compaction caused by a given increment of 

 trampling decreased as trampling increased. For most 

 habitat types, the most rapid increase occurred with the 

 first 50 to 75 passes. The rate of increase was less, but 

 still rapid, up to 400 passes. Beyond this level of tram- 

 pling the rate of increase was much less. 



Pocket soil penetrometer readings are not very precise 

 and are highly variable (Jones 1978). This variabiUty, 

 which could have been compensated for by making more 

 observations, is probably a major source of the erratic 

 nature of the increases in penetration resistance appar- 

 ent in figure 17. 



Another factor that contributed to the erratic nature 

 of these increases is variability in amount of surviving 

 vegetation on treatment lanes. Vegetation apparently 

 cushions the impact of trampling, because compaction 

 was usually less on lanes with high relative cover values. 

 For example, the unexpectedly high penetration resis- 

 tance value after 800 passes in ABLA/CLUN-VACA 

 (fig. 17b) coincides with an unusually low relative cover 

 value of 8 percent (fig. 10b). 



Compaction from trampling was least pronounced on 

 the ABLA/CLUN and ABLA/CLUN-VACA types. In 

 fact, in the ABLA/CLUN type none of the treatments 

 were significantly different (a — 0.05) from the control. 

 The least resistant types were ABLA/VACA and 

 ABLA/XETE. Generally it appears that those types 

 with the thickest organic horizons are least susceptible. 

 Theoretically this would be expected because, in most 

 situations, organic matter cushions the mineral soil from 

 compaction (Lunt 1937; Lutz 1945; Lull 1959). 



Differences between types in resistance to soil compac- 

 tion were considerably less pronounced than differences 

 in resistance to vegetation loss. Across the range of 

 trampling from zero to 1,600 passes, no differences 

 between habitat types explain as much of the variation 

 in penetration resistance increase as differences in 

 amount of trampling. In multiple regressions with pairs 

 of habitat types and number of passes as independent 

 variables and increase in penetration resistance as the 

 dependent variable, the r- contribution of number of 

 passes (0.15 to 0.50, depending upon which types were 

 compared) was always higher than the r- contribution of 

 habitat type (0.003 to 0.14). 



29 



