Abstract. Using matrix modeling we assessed the current status and explored the possible future population trends of three populations of Lobelia boykinii, a rare plant species endemic to Carolina Bays. These three populations are found in the North Carolina coastal plain. We first produced a demographic model using data we collected on the species’ life history. Survival and fecundity data were obtained from marked individuals. Seed germination data were collected from experimental and control seedling plots. The resulting demographic model was used to produce a 2 X 2 matrix model examining the population dynamics at each site. The scalar eigenvalue derived from our matrices was used as our measure of population growth rate. The results indicate that in a “good” year all three populations have a population growth rate above one, although the 95% confidence intervals include values below one. Projections, including different frequencies of “good” and “bad” years based on amount of precipitation, suggest that some populations are really declining in size. Elasticity analyses revealed that new rosette production by surviving adults is the parameter that contributes most to population growth rate. Sensitivity analyses indicated that germination contributes most to the variance in population growth rate in two bays while new rosette production by surviving adults contributes most to the variance in the third bay. These results indicate that to maintain all three populations, conservation efforts must be directed toward maximizing germination and rosette survival and reducing their variability across years in the drier sites. This can be accomplished by periodically burning the sites.

Key words: demographic modeling, elasticity, Lobelia boykinii, risk assessment, sensitivity