Changes in some combination of niche availability, niche overlap and the strength of interspecific interactions are thought to drive changes in plant composition along resource gradients. However, because these processes are difficult to measure in the field, their relative importance in driving compositional change in plant communities remains unclear. In an Australian temperate grassland, we added seeds of three native and three exotic grasses to 1875 experimental plots in a way that allowed us to simultaneously estimate niche availability, niche overlap and the strength of pairwise interspecific interactions along a gradient of nutrient availability, obtained by adding 0, 5 or 20 g m−2 each of nitrogen, phosphorous and potassium jointly to plots. Niche availability (the proportion of microsites suitable for establishment and growth) was generally low and did not vary in response to nutrient addition. Most species co-occurred along the nutrient gradient by partitioning the available niche space. Where species interacted due to niche overlap, the abundance of one species, the native Chloris truncata, was usually facilitated by other species, with each of the five other species increasing the niche availability to C. truncata under at least one nutrient treatment. Chloris truncata also competitively excluded two species from some but not all sites they could otherwise have occupied. These outcomes did not clearly differ across nutrient treatments. Our results show that fine-scale spatial heterogeneity in establishment microsites can enable species to co-occur via niche partitioning, and competitive exclusion is rare. This finding contributes to an emerging picture that niche partitioning is common and frequently a stronger influence on recruitment outcomes than interspecific competition. The importance of competition in structuring plant communities may be overestimated if recruitment processes are overlooked.