Background/Question/Methods Small, spatially isolated plant populations or those with altered disturbance regimes are at higher risk of extinction because of inbreeding depression, loss of genetic heterogeneity, and reduced dispersal rates.  However, interpretation of extinction risk can be influenced by the choice of model used to characterize population states.  We have access to a thrice replicated population census (1991, 1999, and 2008) and associated fire histories for seven state or federally listed rare vascular plant species endemic to the sandhills region of North Carolina.  These data provide a unique opportunity to examine the role of both environmental and spatial factors on assessment of extinction risk.  For each species, we visited all previously known subpopulations on the Fort Bragg Military Reservation, recorded presence or absence, determined number of individuals, and measured spatial extent.  We also collected these data at newly identified subpopulations.  After completing the three censuses, we had identified a total of 469 subpopulations.  We applied five population state models to these data to classify each subpopulation as extinct, persistent, or colonized.  We then used logistic regression and structural equation modeling to explain the variation in population state as a function of multiple spatial and environmental variables.
Results/Conclusions Preliminary results suggest that previous subpopulation density, subpopulation spatial extent, mean fire frequency, and time since last fire are important predictors of species persistence and extinction risk.  However, results vary depending on which population state model is used to determine whether a subpopulation is persistent or has gone locally extinct.  For example, a two-step model (1991 present -1999 absent) identified 21 subpopulations as extinct that would have been identified as persistent in a three-step model (1991 present -1999 absent - 2008 present).  In a comparison of multiple three-step models, we found that 74 out of 469 subpopulations exhibited variation in population state.  This finding underscores the important role of population state models in assessing extinction risk.  Our results can help make predictions about the consequences of rarity or increased habitat loss/fragmentation on rare plant species conservation, assist in defining rare plant conservation success, and provide direction for making management decisions.