Large niche differences emerge at the recruitment stage to stabilize grassland coexistence

Abstract

Niche differences and average fitness differences jointly determine coexistence. However, little empirical information about the magnitude of these two mechanisms is available. Using multispecies population models fit to long‐term demographic data for common, co‐occurring species in five grassland and shrubland plant communities in western North America, we estimated the strength of stabilizing niche differences and average fitness differences. In all five communities, both pairwise and full community comparisons showed evidence for strong stabilizing mechanisms and relatively small average fitness differences. For a total of 17 species pairs, a measure of niche differences based on simulations of invasion growth rates ranged from 0.59 to 0.93 with a mean of 0.81, where 0 indicates complete niche overlap and 1 indicates zero niche overlap. A corresponding measure of average fitness differences ranged from 1.02 to 2.54 with a mean of 1.53, where 1 indicates identical fitness and a value of 2 indicates a fourfold difference in sensitivity to competition. Comparisons of full communities displayed similar patterns: niche differences ranged from 0.58 to 0.69 with a mean of 0.64, and the average fitness differences ranged from 1.42 to 1.63 with a mean of 1.47. In almost every case, the stabilizing mechanisms were much stronger than minimally necessary to prevent competitive exclusion. Considering that all but one of the species we studied are perennial grasses, which are often grouped in the same functional type, the magnitude of these niche differences is surprising. In all five communities, differences between intra‐ and interspecific effects at the recruitment stage contributed far more to stabilization than interactions affecting growth and survival. Our results indicate that for these abundant, co‐occurring species (1) dynamics are far from neutral, with strong niche differences and weak fitness differences combining to stabilize coexistence, and (2) processes operating at early life stages account for a large proportion of the stabilizing effect. Given the limitations of our inductive approach, both these findings represent hypotheses in need of experimental testing.