Fitness and niche differences are both important in explaining responses of plant diversity to nutrient addition.

Plant species loss due to eutrophication is a common phenomenon in temperate perennial grasslands. It occurs in a nonrandom fashion and is usually explained by increased competitive size asymmetry between the co-occurring winner (tall species with optima in productive habitats) and loser species (small-statured plants typical for unproductive habitats). It remains unclear why nutrient addition decreases diversity in communities consisting of losers only, whereas it has little effect on winner-only communities. Here, I used the framework of modern coexistence theory to explore fertilization-driven changes in fitness and niche differences between different combinations of field-identified winner (W) and loser (L) species. I experimentally estimated competition parameters for plant species pairs constructed from a pool of eight species, including pairs of species from the same (WW, LL) and different species categories (LW) grown for approximately 2 years in control and fertilized conditions. Concurrently, I also followed plant species diversity in mesocosm communities constructed from the same species pool (four-species communities including winners, losers, or both) exposed to control and nutrient addition. I found that nutrient addition can reduce but, unexpectedly, also promote species coexistence depending on the type of species pairs. Whereas nutrient addition eroded the coexistence of losers with winners, but also with other losers, treatment had the opposite effect on the persistence of winner species. Fertilization induced large fitness differences between species in loser-winner and loser-loser combinations, but had little effect on the fitness differences of species within the winner-winner combination. In addition, the persistence of winner pairs was promoted by larger niche differences compared to loser species, irrespective of soil nutrients. The differences in how nutrient addition modified coexistence at the pairwise level were reflected by differences in the evenness of multispecies communities assembled from the corresponding species categories. These results suggest that the effect of eutrophication on plant species richness cannot simply be explained by an increased competitive asymmetry. To fully understand the effect of fertilization on the diversity of temperate grasslands, interspecific and intraspecific interactions should be explored while considering differences in species' ecological optima.