The dynamics of early infestations are of critical importance in understanding successful large-scale invasions. We report on the expansion of created patches of Microstegium vimineum (Japanese stiltgrass) over three years and the implications this has for spread. Microstegium is a nonnative weed of particular concern in forest ecosystems, where its rapid invasion threatens native diversity and interferes with forest regeneration. It has invaded entire forest networks in Pennsylvania and Maryland, USA in a time span of less than ten years. Thirty patches were established in 2003 in different habitat types (roadside, intact forest, disturbed forest and wet meadow areas); subsequent population growth and spatial expansion were monitored until plots were terminated in 2006. We developed a spatial model of patch growth, using maximum likelihood techniques to estimate dispersal and population growth parameters.

Results/Conclusions
The observed natural expansion of the patches is inadequate to account for observed large-scale spread rates. The patches expanded surprisingly slowly for a species considered to invade rapidly: the furthest distance new seedlings were found beyond the original patch area was 4 m in the four years of the experiment. At the end of the experiment, the majority of new seedlings were still within 1 m of the original patch.  There was considerable variation in patch performance within a habitat; some patches did expand rapidly while others actually decreased in spatial extent. While patches created in roadside and wet meadow habitats tended to have farther dispersal and greater reproductive ratios than those in forested communities, there were often outlier patches in each environment that raised questions about the relevance of classifying habitat. The long-term projections of the models suggest much slower spread than has typically been observed for Microstegium. The relatively small scale of natural dispersal suggests that human-mediated dispersal, likely influenced by forest road management, is responsible for the rapid spread of this invasive species. We present a new framework to quantify the spread of invasive species undergoing both natural and human mediated dispersal.