Multiscale 107 



Lack of readily discernable patterns between captures and selected 

 habitat variables could be a reflection of the variables we selected and lack of dif- 

 ferences between spatial scales for some variables. There are at least three 

 potential reasons for the lack of a readily discernable pattern between capture 

 and no-capture sites. First, low statistical power might have resulted in the lack 

 of differences observed in our study. However, because of the large amount of 

 variation observed for all means and the finding of significant differences for the 

 larger scales, the lack of differences is most likely not due to low statistical 

 power. Second, captures of P. leucopus are often related to factors other than 

 habitat. The species is known to respond to new objects placed within a famil- 

 iar area (Lackey et al. 1985), and densities have been shown to correlate to food 

 distribution (Getz 1961). Third, the spatial scales selected for study might have 

 been of an incorrect size for ascertaining capture patterns. However, the lack of 

 patterns in our study does not necessarily indicate scale is not important in asso- 

 ciating captures of P. leucopus with habitat, only that a different scale may be 

 warranted for future studies. 



Our study shows at least two potential means by which selection of 

 scale could influence results of a study warranting investigation of potential pat- 

 terns at each scale. First, significant differences found for some variables sug- 

 gest difference at the 10-m 2 scale, but not at smaller scales. Second, the differ- 

 ences in the loadings of variables on the principal components axes, differences 

 in variables selected for use in discriminant analysis, and the decrease in outliers 

 as scale increases suggest differences between all of the scales. These differ- 

 ences between scales may potentially have a large effect on the conclusions (see 

 Schneider and Piat 1986, Woodby 1984). For example, our study had inconse- 

 quential findings at the 1-m 2 scale; but significant results at the larger scales. 



Our data show that, even when using the small scales that we selected, 

 differences in habitat affinities for capture can occur around the same trap site. 

 Because our data were collected in a homogenous habitat over a short period of 

 time, the differences observed in our study concerning correlations between 

 habitat and capture can be attributed to the different scales. While a species may 

 appear to be a habitat generalist with an affinity toward a variety of habitats (e.g. 

 P. leucopus), studies incorporating a multiscale approach may indicate a narrow- 

 er range of optimal habitat affinities. Thus, studies assessing habitat use should 

 incorporate analyses at multiple scales. It appears this may be achieved by the 

 incorporation of at least three scales of assessment allowing for comparison at 

 different scales in the same habitat. 



More study is needed in the selection of scale to be measured. Selec- 

 tion of scale is difficult to evaluate due to differences in habitat at each study site. 

 However, we feel choice of scale should be selected based on at least the fol- 

 lowing factors. Of primary concern should be the habitat in which the study is 

 conducted. More homogeneous habitats may require a larger number of scales 



