A previous study of four dominant species from a sagebrush steppe used historical data sets and multispecies population models to demonstrate weak interspecific interactions. The models predict little competitive release following species removals. We conducted a removal experiment to test this prediction. We established new quadrats in the same location where the historical data were collected. We assigned half the new quadrats to a perennial grass removal treatment and half to a treatment in which we removed the dominant shrub, Artemisia tripartita. We modeled survival, growth and recruitment as functions of local neighborhood species composition along with an indicator variable for removal treatment. If our "baseline" model, which accounts for local plant-plant interactions, explains responses to removal, then the fitted removal indicator effect should be non-signficant.
For survival and recruitment, the removal treatment effects were never significantly positive, indicating that competitive release was not underestimated by the interspecific effects included in the baseline model. For Poa secunda recruitment, the removal treatment was actually negative, indicating that our baseline model overestimated competitive release. For growth, the removal treatment effect was significant and positive for two species, Poa secunda and Pseudoroegneria spicata, indicating that the baseline model underestimated competitive release. However, including information about the location of these grass individuals with respect to removed A. tripartita failed to improve the growth regressions, raising questions about the mechanisms driving the positive response to removal. For three species, individual based models and integral projections models based on the vital rate regressions showed that removal treatment effects had little impact on population growth rates or simulated equilibrium cover following a species removal. For P. spicata, the population models that included effects of A. tripartita removal projected greater competitive release than the baseline models. Although models based on observational data did not perfectly predict species responses to removal, the experimental results increase our confidence that competitive release is in fact small for at least three of our four study species. The implication is that interspecific resource competition is not a primary factor determining the dynamics of this community. The success of the models also supports the use of multispecies models, based on long-term observational studies, to characterize competitive interactions in a community.