Traits linked with species invasiveness and community invasibility vary with time, stage and indicator of invasion in a long-term grassland experiment.

Much uncertainty remains about traits linked with successful invasion - the establishment and spread of non-resident species into existing communities. Using a 20-year experiment, where 50 non-resident (but mostly native) grassland plant species were sown into savannah plots, we ask how traits linked with invasion depend on invasion stage (establishment, spread), indicator of invasion success (occupancy, relative abundance), time, environmental conditions, propagule rain, and traits of invaders and invaded communities. Trait data for 164 taxa showed that invader occupancy was primarily associated with traits of invaders, traits of recipient communities, and invader-community interactions. Invader abundance was more strongly associated with community traits (e.g. proportion legume) and trait differences between invaders and the most similar resident species. Annuals and invaders with high-specific leaf area were only successful early in stand development, whereas invaders with conservative carbon capture strategies persisted long-term. Our results indicate that invasion is context-dependent and long-term experiments are required to comprehensively understand invasions.