We investigated how ecological realism might impact the outcome of three experimental manipulations of species richness to determine whether the patterns and the mechanisms underlying richness–variability relationships differ as ecological communities are increasingly exposed to external forces that may drive richness–variability patterns in nature. To test for such an effect, we conducted experiments using rock pool meio-invertebrate communities housed in three experimental venues: controlled laboratory microcosms, artificially constructed rock pools in the field, and naturally occurring rock pools in the field. Our results showed that experimental venue can have a strong effect on the outcome of richness manipulation experiments. As ecological realism increased, the strength of the relationship between species richness and community variability declined from 32.9% in the laboratory microcosms to 16.8% in the artificial pools to no effect of species richness on community variability in the natural rock pools. The determinants of community variability also differed as ecological realism increased. In laboratory microcosms, community variability was driven solely by mechanisms related to increasing species richness. In artificial rock pools, community variability was driven by a combination of direct and indirect environmental factors as well as mechanisms related to increasing species richness. In the natural rock pools community variability was independent of species richness and was only related to environmental factors. In summary, we found that stabilizing mechanisms associated with species interactions were influential in establishing species richness–variability relations only in the less realistic experimental venues (the laboratory microcosms and the artificial rock pools in the field), and that these mechanisms diminished in importance as ecological realism and complexity of the experimental venue increased. Our results suggest that the effects of diversity might be more difficult to detect in natural systems due to the combined effects of biotic and abiotic forcing, which can mask our ability to detect richness effects.