Patterns of invasion of an urban remnant of a species‐rich grassland in southeastern Australia by non‐native plant species

Abstract. The invasion by non‐native plant species of an urban remnant of a species‐rich Themeda triandra grassland in southeastern Australia was quantified and related to abiotic influences. Richness and cover of non‐native species were highest at the edges of the remnant and declined to relatively uniform levels within the remnant. Native species richness and cover were lowest at the edge adjoining a roadside but then showed little relation to distance from edge. Roadside edge quadrats were floristically distinct from most other quadrats when ordinated by Detrended Correspondence Analysis. Soil phosphorus was significantly higher at the roadside edge but did not vary within the remnant itself. All other abiotic factors measured (NH 4 , NO 3 , S, pH and % organic carbon) showed little variation across the remnant. Non‐native species richness and cover were strongly correlated with soil phosphorus levels. Native species were negatively correlated with soil phosphorus levels. Canonical Correspondence Analysis identified the perennial non‐native grasses of high biomass as species most dependent on high soil nutrient levels. Such species may be resource‐limited in undisturbed soils. Three classes of non‐native plants have invaded this species‐rich grassland: (1) generalist species (> 50 % frequency), mostly therophytes with non‐specialized habitat or germination requirements; (2) resource‐limited species comprising perennial species of high biomass that are dependent on nutrient increases and/or soil disturbances before they can invade the community and; (3) species of intermediate frequency (1–30 %), of low to high biomass potential, that appear to have non‐specialized habitat requirements but are currently limited by seed dispersal, seedling establishment or the current site management. Native species richness and cover are most negatively affected by increases in non‐native cover. Declines are largely evident once the non‐native cover exceeds 40 %. Widespread, generalist non‐native species are numerous in intact sites and will have to be considered a permanent part of the flora of remnant grasslands. Management must aim to minimize increases in cover of any non‐native species or the disturbances that favour the establishment of competitive non‐native grasses if the native grassland flora is to be conserved in small, fragmented remnants.