Context dependence can impede accurate prediction in ecology, including plant colonisation. Although is well established that ecology is complicated, most studies in ecology focus on one or two factors at a time and most experiments run for less than three years. As such, most studies only provide a partial picture, and sometimes those pictures can appear contradictory, leading researchers and practitioners to question what factors are important and when. Using a 12-year multifactorial grassland seed addition experiment, we tested whether we could accurately predict drivers of sown colonist abundance based on the species, sites and conditions in question.

Up to 15 grassland species, divided into resource-acquisitive (“fast”) or resource-conservative (“slow”) life history groups, were added at two different seed rates, with and without the other colonist group, into grassland plots in Minnesota, US. The plots represented early, mid and late successional communities, and half of the plots were regularly disturbed. We surveyed sown “colonist” cover six times over 12 years.

None of the five factors examined (colonist life history group, colonist group composition, seed rate, community successional age, disturbance – plus time) had a consistent relationship with colonist abundance but general trends did emerge.

Resource-acquisitive colonists with fast life histories were most successful soon after seed addition while resource-conservative slow colonists reached peak abundances in later years. Colonist cover was higher in early successional than late successional communities, reflecting soil N limitation in late succession.

Physical disturbance facilitated colonisation when soil N was limiting, especially for resource acquisitive colonists, but disturbance otherwise impeded colonisation as it exacerbated effects of low soil moisture and reduced microhabitat suitability. When sown together into communities not already limited by resource competition, there was some evidence that fast colonists impeded slow colonists through niche pre-emption.

Context dependence can sometimes seem overwhelming in ecology, but – in this study – trends were largely predictable when we considered how causal drivers and limiting factors likely vary across different ecological conditions. Our study highlights the value of long-term multifactorial experiments and suggests that the interplay of limiting factors can help predict mechanistic context dependence in plant colonisation.