Dominance, biomass and extinction resistance determine the consequences of biodiversity loss for multiple coastal ecosystem processes.

Key ecosystem processes such as carbon and nutrient cycling could be deteriorating as a result of biodiversity loss. However, currently we lack the ability to predict the consequences of realistic species loss on ecosystem processes. The aim of this study was to test whether species contributions to community biomass can be used as surrogate measures of their contribution to ecosystem processes. These were gross community productivity in a salt marsh plant assemblage and an intertidal macroalgae assemblage; community clearance of microalgae in sessile suspension feeding invertebrate assemblage; and nutrient uptake in an intertidal macroalgae assemblage. We conducted a series of biodiversity manipulations that represented realistic species extinction sequences in each of the three contrasting assemblages. Species were removed in a subtractive fashion so that biomass was allowed to vary with each species removal, and key ecosystem processes were measured at each stage of community disassembly. The functional contribution of species was directly proportional to their contribution to community biomass in a 1:1 ratio, a relationship that was consistent across three contrasting marine ecosystems and three ecosystem processes. This suggests that the biomass contributed by a species to an assemblage can be used to approximately predict the proportional decline in an ecosystem process when that species is lost. Such predictions represent "worst case scenarios" because, over time, extinction resilient species can offset the loss of biomass associated with the extinction of competitors. We also modelled a "best case scenario" that accounts for compensatory responses by the extant species with the highest per capita contribution to ecosystem processes. These worst and best case scenarios could be used to predict the minimum and maximum species required to sustain threshold values of ecosystem processes in the future