tion of phytoplankton, as has been demonstrated in the Baltic Sea (Wulff et al. 1990). Where long-term nutrient data are available, the increased occurrence of nuisance algal blooms has always been found to be correlated with a decrease in Si:N and Si:P ratios (Smayda 1989 and references therein). Net primary production probably remains controlled by nitrogen or phosphorus availability throughout the range of silicon availabilities (Howarth 1988), but the relative availability of silicon may well control the abundance of diatoms versus other phytoplankton species, thereby setting the stage for nuisance blooms (Smayda 1989).
Trace-metal availability may be another factor in the initiation of many nuisance algal blooms, with high iron availability favoring the bloom-forming species. A variety of physiological studies with pure cultures and short-term enrichment studies with natural populations have shown that the requirement for iron is high for red-tide dinoflagellates (Wilson 1966, Martin and Martin 1973, Graneli et al. 1986, Doucette and Harrison 1990), for the brown-tide algae (Cosper et al. 1990), and for cyanobacteria (Wurtsbaugh and Horne 1983, Howarth et al. 1988b). Red-tide outbreaks over a 25-year period in Florida have been correlated with iron inputs from rivers (Kim and Martin 1974) although inputs of a number of other substances are undoubtedly correlated with the iron inputs (Wells et al. 1991). However, most iron is not directly available to phytoplankton, and the available fraction of iron is not correlated with the outbreak of dinoflagellate blooms in Maine (Wells et al. 1991). Proof of a critical role for iron in initiating blooms must await further study.
Many estuaries and coastal marine ecosystems receive excessive inputs of nutrients, and sewage effluent is often a major component of this. These excessive inputs cause eutrophication, leading to anoxic and hypoxic conditions, loss of seagrass and algal beds, and damage to coral reefs. In many temperate estuaries and coastal seas, nitrogen is the primary nutrient of concern leading to increased eutrophication. Phosphorus controls eutrophication in some coastal marine ecosystems, at least during some seasons, and may be the primarily controlling element in tropical seas. To control coastal eutrophication, both nitrogen and phosphorus need to be controlled (D'Elia et al. 1986, Howarth 1988, Graneli et al. 1990). For phosphorus, the approach used in lakes—correlational models relating phosphorus to phytoplankton in lakes—should provide an adequate strategy for determining acceptable inputs to most temperate estuaries and coastal marine ecosystems as well. Acceptable inputs for nitrogen to marine ecosystems can be developed from similar models relating nitrogen inputs to phytoplankton production or biomass. However, such approaches may not adequately protect tropical and