Dominant increase
Models revealed several changes in species interactions consistent with observed declines in evenness. First, competitive effects on the dominant species declined in all plots where Deschampsiaincreased over time (SNW, SW, and NW) (Fig 3, Table S3). In SNW and NW plots, this was driven primarily by reduced interspecific competition, while in SW plots this was driven primarily by reduced intraspecific competition of the dominant species on itself (Fig 3, Table S3). The dominant species increased its intraspecific competition in plots with added N (SNW, NW), consistent with the positive effect of N addition on its density-independent growth rates (Fig 2). Furthermore, net competitive effects increased in SNW and SW plots for subdominant and (to a lesser extent) moderate species, primarily driven by increased interspecific competition with each other, reflecting higher-order interactions that benefit the dominant species, and contribute to their decline in these treatments over time.
In the NW treatment, competitive effects declined for all species groups (except rare) (Fig 3, Table S3). Reduced competition, in combination with density-independent patterns observed, help explain the lower magnitude of moderate and subdominant species declines in NW compared to SW and SNW treatments. However, this pattern was more pronounced for moderate than subdominant species, likely due to a strong reduction in the competitive effect of the dominant on moderate species (Fig S8c). In addition, predictive steady-state distributions revealed a non-linear (left-skewed) distribution of subdominant species across the observed N gradient in NW plots, suggesting that subdominant species only benefits from competitive release at low N levels, after which the dominant takes over (Fig 4).