Do pairwise competition outcomes predict multispecies coexistence?
In plain words
One could try to predict which species survive together by testing them two at a time and applying simple rules. If interactions involving three or more species at once matter, these two-at-a-time tests will fail for large communities.
Precise statement
For a pool of S species, assembly rules based only on pairwise competition outcomes (a species pair coexists or one excludes the other) predict which subsets coexist. Determine the fraction of multispecies outcomes correctly predicted as S grows and whether failures are caused by higher-order interactions (non-additive terms in per-capita growth rates) or by multistability of pairwise models. An answer is the prediction accuracy versus S for S up to tens of species, with the cause of failures identified.
What would settle it
Complete pairwise and multispecies competition experiments for pools of increasing S, with direct measurement of higher-order interaction terms.
Status in the literature
2017 microcosm experiments with 8 soil bacteria predicted about 90 percent of three-species outcomes from pairwise competitions, and also tested 7- and 8-species mixtures; a systematic measurement of accuracy versus S for tens of species is lacking.