In addition, for each reproductive plant we recorded the number 

 of inflorescences and the number of fruits matured (see Lesica 

 1987 for further details) . 



A plants 's demographic properties are often more closely 

 correlated with size and life-history stage rather than age 

 (Werner and Caswell 1977, Caswell 1989), although both may be 

 important in predicting an individual's fate (Young 1985). We 

 chose these classes because they are correlated with age as well 

 as size and because they also represent a reasonable compromise 

 between having many categories with too few observations each and 

 few categories with many observations (Vandermeer 1978) . 



In 1989-92 we collected one fruit from the middle of the 

 inflorescence of each of 25 randomly chosen plants growing near 

 the transects at each site. We counted the number of mature or 

 nearly mature seeds in each fruit to obtain an estimate of 

 seeds/fruit for each site. 



We estimated canopy cover of all vascular plants as well as 

 cover of rock, bare soil and basal vegetation in each plot 

 (Daubenmire 1959) . We estimated cover to the nearest 5% with two 

 additional classes, 1-3% and 0-1%. 



Data analysis 



Population growth for year t is the percent change in the 

 size of the sample population between year t-1 and year t. It is 

 calculated by PG = N^ - N^.^/N^.,. Mortality is defined as the 

 ratio of the number of plants dying between years t-1 and t to 

 the number surviving in the same period. Recruitment rate is 

 defined as the ratio of new plants observed in year t to the 

 number of plants surviving from year t-1 to year t. 



We compared survival rates o'f uneven-age cohorts present at 

 the start of the study among the sites using the nonparametric 

 logrank test (Pyke and Thompson 1986, Hutchings et al. 1991). 

 Survivorship curves were constructed following methods outlined 

 in Hutchings et al. (1991) . Probability values were not adjusted 

 for multiple tests. 



For our study, fecundity is defined as the number of fruits 

 per plant. The effects of site and year on fecundity and number 

 of seeds/fruit were analyzed using analysis of variance (ANOVA) . 

 We used ANOVA followed by a contrast test to determine the effect 

 of site on bare soil in the plots. Plots within transect cannot 

 be considered independent because they are contiguous. Thus, we 

 used the eight transects as sampling units for the test. 



We used correlation and multiple regression analyses to 

 explore relationships between weather variables and Arabis 

 fecunda mortality, recruitment and fecundity. Weather data are 



