Separating the mechanisms that influence ecosystem functioning has been a goal of recent high profile experiments. Integrating the various experimental and analytical methods that attempt this goal across terrestrial and aquatic ecosystems, as well as careful definition of ‘complementarity’, produces novel insights and valuable lessons about new directions for research. (1) Experimental designs differ in temporal scale and whether standing stock or another ecosystem process was the response variable. (2) Mathematically identical variables in different designs have contrasting ecological interpretations. For example, different sets of ecological processes can contribute to different variables in different experimental designs. (3) The frequency of transgressive overyielding of standing stock (e.g. total above ground biomass) in polycultures implies little about the prevalence of transgressive overyielding in other ecosystem processes. (4) Measuring the contribution to ecosystem functioning of individual species, rather than just total ecosystem functioning of a polyculture, is not essential for estimating effects of complementarity. (5) Further research will profit from distinguishing standing stock from all other ecosystem functions. (6) None of the analytic methods can distinguish the effects of individual processes or mechanisms such as resource use differentiation, facilitation, or allelopathy, for which additional experimental treatments are required.