BIODIVERSITY RESEARCH: Population expansion in an invasive grass, Microstegium vimineum : a test of the channelled diffusion model

Aim  The greatest biodiversity impact of non‐native plant species is caused by rapid expansion of colonist populations. Unfortunately, invasion has rarely been documented in real time at a population scale, and demographic mechanisms of invasion remain unclear. Our goal is to describe real‐time expansion of populations, using channelled diffusion as a null model. Location  The study examined three populations of the invasive annual grass Microstegium vimineum in mature second‐growth forests of south‐eastern Ohio and nearby West Virginia, USA. Methods  Distributions were recorded in belt transects perpendicular to population edges over a period of 3 years. A second group of belt transects documented spread along five types of potential movement corridor. Observed changes in distribution were compared with predictions from a diffusion model. A seed‐sowing experiment tested seed availability, microsite quality and proximity to potential movement corridors as factors controlling population spread. Results  Population boundaries showed little change over the study period. Colonization was limited by propagule availability over distances as little as 0.25 m, and to a lesser extent by litter cover. Populations did not advance along several potential movement corridors including unpaved roads, off‐road vehicle trails and footpaths. Advance was observed along deer trails and stream courses but did not conform to the wave‐form distribution predicted by diffusion theory. During the study, seeds were moved out of experimental plots by sheet flow and minor flooding events along small streams. Main conclusion  At a population level, invasion is driven by processes that are episodic in time and non‐random in space – probably a common condition in non‐native plant species. Spatially realistic models are likely to be more useful than diffusive models in managing invasions at these scales.