Polygonella myriophylla (Polygonaceae) is a clonal shrub listed as endangered and narrowly endemic to pyrogenic scrub ecosystems in central Florida. It is almost restricted to gaps within the matrix of shrubs in the scrub but also occurs along adjacent road-side habitats. I hypothesize that persistent disturbed microhabitats and more dynamic sand accretion in roadsides will increase rooting probabilities compared to more stable scrub habitats, affecting survival, growth and reproduction. In April 2004- March 2006, I compared plant (genet) and basal branch (ramet) performance between experimentally manipulated plants in native scrub and roadside habitats at two locations within the Lake Wales Ridge State Forest in Polk County: LC01 and Old School. We completed a total of 6 evaluations in each site (April, July and November 2004, May and November 2005, and March 2006). Fifteen plants per replicated habitat in LCO1 and Old School were selected based on presence of four unrooted branches. Each unrooted branch within a plant randomly received one of four possible treatments: forced branch burial, branch lifting, procedural control, and no manipulation (total N= 60 genets and 240 ramets). Forced burial was implemented to mimic sand burial and evaluate rooting probability and performance in both habitats. Branch lifting was applied to prevent sand burial and evaluate demography of unrooted branches in both habitats. The procedural control served to evaluate wire effects on ramet demography. The control provided vital and rooting rates of branches in natural conditions. Road populations exhibited larger crown area and higher monthly diameter (controlled by initial diameter) and higher monthly length growth rates compared to scrub populations. Rooting probability was only affected by treatment one (buried wire) not habitat or site. Forced sand burial increased rooting (67 % after forced contact vs. 20-30 % for other treatments). Rooted branches did not exhibit variation in survival, growth, or fecundity compared to unrooted branches. Old School populations exhibited larger crown area, higher monthly diameter and monthly length growth rates compared to LC01 populations. Prescribed fires killed several plants explaining significantly higher branch survival at the unburned LC01 (66.1%) compared to recently burned Old School (36.2 %). LC01 populations exhibited higher fecundity and ramet survival compared to Old School populations. In February - December 2006, I describe the reproductive schedule at (LC01) in 10 road and 10 scrub plants. Monthly, I counted number of inflorescences and flowers per inflorescence (one inflorescence per plant) for each plant. Number of inflorescences per plant was highest between May and September and higher in road than in scrub. Our results indicate significant different demographic performance of P. myriophylla at plant and branch level between road and scrub habitats. A longer term study is needed to determine if the persistence of P. myriophylla is threatened by increasing roadside populations.