What causes the sporadic phytoplankton bloom southeast of Madagascar?

A large, dendritic phytoplankton bloom develops in the austral summer of roughly every other year in the Indian Ocean southeast of Madagascar. It starts in February or March and dissipates by the end of April. It was observed in 1997, 1999, 2000, 2002, and 2004 and was absent in 1998, 2001, 2003, and 2005. A. Longhurst, who described it in 2001, suggested that the bloom is caused by the entrainment of nutrient-rich waters into a deepening mixed layer. At the time the bloom had been seen to fail only once, in 1998, and Longhurst attributed that to reduced entrainment due to low winds during the strong La Nina. Since reanalysis winds are not correlated with the occurrence/failure of the bloom, and since the bloom starts before sea surface temperature (SST) peaks and increased chlorophyll is found with high rather than low SST, entrainment is an unlikely cause. Argo float profiles from 2004 show the bloom within a shallow mixed layer over a very strong pycnocline. This indicates that the bloom is caused not by entrainment of nutrient-rich waters from below, but by another mechanism such as nitrogen fixation by diazotrophs that are retained within a shallow, well-lit layer, or nutrient fluxes due to vertically mobile plankton. This bloom is among the strongest features of interannual variability in the ocean color time series, second only to El Nino-Southern Oscillation (ENSO)- related variability in terms of both the area effected and magnitude of the anomaly. While the mechanisms with which biology responds to the physical changes associated with ENSO are well-known and rather obvious in the strong correlations with physical indicators such as sea surface height or temperature, the occurrence/failure of a strong Madagascar bloom does not have an obvious correlation with any physical parameter. Of the many variables tested, only the landfall of tropical cyclones on the island of Madagascar correlated with the occurrence of the bloom. A hypothesis is presented, without any proof other than that it explains the location, seasonal timing, interannual variability, and observations of temperature and mixed layer depth, on the basis of the premise that runoff from the island alleviates iron limitation of diazotrophs. Definitive determination of the nature of the bloom will probably require in situ sampling.