Spatial and temporal demographic variability in the endemic plant species Centaurea corymbosa (Asteraceae)

Centaurea corymbosa is an endemic plant species restricted to a 3-km2area in southern France. This species is known from only six small populations that are highly differentiated genetically. Matrix models based on eight years of data (1994-2001) were used to assess the pattern of variation in the demographic vital rates of this species, and to investigate the causes of their variation. Asymptotic growth rates λ varied widely between years and populations (0.613-1.424). Randomization tests were developed to test for spatial and temporal variation in the asymptotic growth rates. These tests rely on individual data on both survival and fecundity. As our demographic survey only allowed us to estimate average fecundities, additional fecundity data collected from 1994 to 1996 were used to assess the distribution of individual fecundity expected within populations under demographic stochasticity or sampling error. Randomization tests showed that asymptotic growth rates were significantly different between populations and between years in C. corymbosa. In contrast, log-linear analysis performed only on transition data (i.e., excluding fecundity) suggested that the observed variations in transition probabilities were mostly explained by sampling error or demographic stochasticity rather than environmental stochasticity. This suggested that variations in fecundity among populations and among years may play a key role to explain temporal and spatial differences in λ. Life-table response experiment analysis revealed that variations in fecundity and especially in the number of just-emerged seedlings per plant explained most of the observed variance in λ. Spatial and temporal variations were detected for most lower-level vital rates, but causal factors that may account for these patterns are still unknown: no effect of genetic diversity was detected on the dynamics of the species, and correlation between lower-level vital rates and climatic data did not reveal any clear trends. Our study emphasizes the need to conduct long-term demographic surveys and to collect individual fecundity data to get more insights into the causes of variation of the demographic behavior of C. corymbosa.