Plant assemblage composition explains and predicts how biodiversity affects salt marsh functioning

Knowing that diverse plantings enhanced biomass and nitrogen (N) accumulation in a restored California salt marsh, we asked if the biodiversity effect'' was due to species selection or complementarity. In a two-year greenhouse experiment, we found positive biodiversity effects on total, root, and shoot biomass, total and root N crop, and on biomass and N allocation; negative effects on root and shoot N concentration; and no effect on shoot N crop. Overyielding among trios and sextets was supported by significant deviations in observed yield from that expected relative to solo yields (D-T). However, both trios and sextets strongly underyielded relative to the highest yielding solo in the assemblage (D-max) in all attributes, and to the dominant species in the assemblage (D-dom) in most attributes. When we decomposed biodiversity effects on shoot characteristics, selection effects primarily drove over- and underyielding. The only complementarity effect was underyielding of sextet shoot biomass. These analyses were possible because we replicated assemblages and evaluated 11 response variables. One species (Salicornia virginica) dominated functioning when present; when absent, another dominated (e.g., Frankenia salina). Effects varied with the response tested, however. For both shoot biomass and N crop, S. virginica was the dominant overyielding species (based on D-ij and comparisons of trios +/- target species). For shoot N concentration, however, the dominant was Triglochin concinna, a species that had low biomass but was capable of reducing assemblage performance, presumably by concentrating N. Evidence for strong species selection effects led us to predict that three species would eventually dominate our parallel field experiment that tested the same assemblages. Exactly that happened in nine years, but (we predict) without losing function, because the site retained the three highest-performing species. Biodiversity loss was nonrandom in the field, and because trios with two top performers sustained critical functions in the greenhouse, we predict that many functions will not decline, even if the salt marsh becomes dominated by a single species, e.g., S. virginica. Unmeasured functions (e.g., resilience) might not persist, however. Knowing how assemblages perform made biodiversity-ecosystem function theory both explanatory and predictive.