Background/Question/Methods Pitcher's thistle (Cirsium pitcheri), a short-lived, monocarpic herb endemic to western Great Lakes sand dunes, colonizes successional habitats and requires frequent cohort replacement to maintain populations. This federally threatened plant became extinct in Illinois before 1920. As part of federal and state recovery planning, reintroduction into former habitat at Illinois Beach State Park began in 1991.  Greenhouse-propagated plants were outplanted annually between 1991 and 2000 into 27 transplant units ranging from 1 to 92 transplanted individuals   Transplant units located south of the Dead River were less disturbed but exposed to greater invasion of Austrian Pine (Pinus nigra) and Lombardy Poplar (Populas nigra var. italica) than north of the Dead River. Natural recruitment began in the 5^th year of reintroduction, and natural recruit number surpassed transplant number  in 2001, the 11^th year of the reintroduction. The last transplant died in 2005.  Mean population size from 1991 to 2007 was 170 ± 17.2 SE and population size in 2007 was 146. The reintroduction stochastic growth rate (λ_s) is 1.025 (95% CI, 1.006, 1.038).  Because persistence and longevity of reintroduction events varied, we used logistic regression and survival analysis to investigate factors associated with whether or not transplant units persisted and the number of years they persisted.  Because transplant units began in different years, persistence at seven years, the shortest time since reintroduction, was used to avoid the confounding effect of time since transplanting on persistence.
Results/Conclusions Logistic regression analysis indicated that reintroduction persistence probability increased significantly with transplant number (P = 0.007, R_N² = 0.643).   Persistence was also significantly higher on dune habitats compared to interdunes (P = 0.018, R_N² = 0.297) and the less disturbed habitat south of the Dead River (P = 0.016, R_N² = 0.308).  Persistence probability significantly increased with decreasing transplant year drought severity (P = 0.037, R_N² = 0.553).  Survival analysis indicated that time to extinction significantly decreased with increasing transplant number (P = 0.023) and reintroductions persisted longer in dune habitats (P= 0.010) and in the less disturbed habitat south of the Dead River (P = 0.025).  Reintroductions persisted longer with decreasing transplant year drought severity (P = 0.065).  These results suggest that reintroduction success requires 27 (95% CI 15, 139) transplants in undisturbed dunes habitat and 76 (95% CI 27, 854) in disturbed interdune habitat for a 95% persistence probability, with supplemental watering during drought years.