Macrocosms can be used to study complex ecological processes in a small physical space, but the validity of the studies depends on how well the macrocosm simulates natural ecosystem functions. We measured the standing crop of macroalgae and nutrient levels over 4 years in the Biosphere II macrocosm tropical reef biome at Oracle, Arizona. Ten years after this system was closed to outside introductions, it still contained 35 species of macroalgae, within the range found on natural reefs. The macroalgae community was recognizable as a late-successional, coralline algal turf community similar to those found in the low-energy portions of inner tropical reefs. However, biomass values for the most abundant species were in a continual state of flux over the study, and no single species was dominant. Nutrient levels were also unexpectedly dynamic, with dissolved inorganic phosphorous, nitrate, and ammonium each varying by a factor of 10 over the study. The dynamic nature of biomass and nutrient cycles would make it difficult to use this macrocosm for controlled studies. On the other hand, the community dynamics in the Biosphere II ocean may shed light on processes controlling biodiversity and succession on natural reefs. The apparently chaotic swings in biomass and nutrient levels suggested that the paradox of the plankton, which explains how seemingly uniform aquatic environments can support a wide diversity of planktonic forms, may apply to reef macroalgae as well. The Biosphere II ocean biome demonstrates that a diverse macroalgal reef community can be restored relatively easily, supporting the feasibility of actively restoring damaged natural reefs. Macrocosms such as this could be used in manipulative experiments to study the effects of nutrient enrichment and herbivory on coral–algal phase shifts.