9 

 calculations were performed on a microcomputer using SYSTAT 

 (Wilkinson 1986) . Eigenvalues and confidence intervals were 

 obtained using GAUSS (Anon. 1988) . 



Results 



Before the start of the experiment, cover of Centaurea 

 maculosa was ca. 30% at Birch Creek and 25% at Charley's Gulch, 

 and the difference between treatment and control plots was not 

 significant (Table 2) . After one year, C^ maculosa cover was 

 significantly reduced relative to controls at Charley's Gulch, 

 but not at Birch Creek (Table 2). The resiliency of the C^ 

 maculosa population during the first year was due mainly to 

 recruitment from the seed bank. In spite of this recruitment, 

 removal of C_j_ maculosa probably reduced competitive effects on A^. 

 fecunda . After two years, C\_ maculosa cover was significantly 

 reduced at both sites (Table 2) . 



Summary demographic statistics for both sites are presented 

 in Table 3 . At Birch Creek there were fewer plants in the 

 treatment plots compared to controls at the beginning of the 

 experiment, but more than controls after one year. There were 

 nearly equal numbers of fruiting plants in treatment and control 

 plots during the first year of the experiment, but more in 

 treatment plots in the last two years. Treatment plots had a 

 higher recruitment rate during the first two years, but often 



