A Karenia brevis Harmful Algal Bloom affected coastal waters shallower than 50 m off west-central Florida from January 2005 through January 2006, showing a sustained anomaly of ∼1 mg chlorophyll m−3 over an area of up to 67,500 km2. Red tides occur in the same area (approximately 26–29°N, 82–83°W) almost every year, but the intense 2005 bloom led to a widespread hypoxic zone (dissolved oxygen <2 mg L−1) that caused mortalities of benthic communities, fish, turtles, birds, and marine mammals. Runoff alone provided insufficient nitrogen to support this bloom. We pose the hypothesis that submarine groundwater discharge (SGD) provides the missing nutrients, and indeed can trigger and support the recurrent red tides off west-central Florida. SGD inputs of dissolved inorganic nitrogen (DIN) in Tampa Bay alone are ∼35% of that discharged by all central Florida rivers draining west combined. We propose that the unusual number of hurricanes in 2004 resulted in high runoff, and in higher than normal SGD emerging along the west Florida coast throughout 2005, initiating and fueling the persistent HAB. This mechanism may also explain recurrent red tides in other coastal regions of the Gulf of Mexico.
 Harmful algal blooms (HABs, or red tides) on the west Florida shelf (WFS) are primarily caused by the toxic species Karenia brevis. This species can produce brevetoxins that accumulate in bivalves, cause mortalities of marine organisms, and lead to irritation of the eye and respiratory systems of animals including humans. The earliest documented red tide on the WFS dates to 1854 [Ingersoll, 1882], but over the past 120 years there are numerous reports of such events off the same west-central Florida area (approximately 26–29°N, 82–83°W). They occur almost every year between August and March, but also at other times [Feinstein et al., 1955; Tester and Steidinger, 1997].
 For over a hundred years, scientists have puzzled about the source of the nutrients that initiate and maintain such extensive, toxic blooms in this specific confined area. Much informal discussion has centered on the role of fresh water discharge from either local rivers or eastward dispersal of Mississippi River water [Feinstein et al., 1955; U.S. Bureau of Commercial Fisheries, 1958, available at http://www.floridamarine.org/features/view_article.asp?id=2853]. The more recent Ecology of HABs (EcoHAB) program (1998–2001) provided insights into the nutrient dynamics of red tides [Walsh and Steidinger, 2004], suggesting that there is phosphorus and nitrogen deficiency from estuarine fluxes [Vargo et al., 2004] (G. A. Vargo et al., Nutrient availability in support of Karenia brevis blooms on the central West Florida Shelf: What keeps Karenia blooming?, submitted to Cont. Shelf. Res., 2006).
 The following complicated hypothesis was proposed by EcoHAB researchers to explain the initiation and maintenance of K. brevis blooms under nitrogen limitation [Lenes et al., 2001; Walsh and Steidinger, 2001; Walsh et al., 2003]: Phosphorus-rich river water is delivered to the coast, where Trichodesmium blooms are stimulated by deposition of iron-rich Saharan dust, fixing nitrogen. These blooms decompose upon sinking, releasing dissolved organic nitrogen and stimulating a toxic dinoflagellate seed population located near the bottom. Coastal upwelling moves this population onshore in the bottom Ekman layer, toward coastal regions rich in colored dissolved organic matter (CDOM) where light-inhibition is alleviated. Small K. brevis blooms would use dead fish as a supplementary nutrient source and swim toward the surface using positive phototaxis, and further protect themselves from light-inhibition by self-shading.
 This sequence of events, which was proposed to repeatedly lead to the unfortunate outcome of HABs only in a restricted area off west-central Florida (or immediately off Texas, according to the authors), has not been independently validated. It seems unlikely that the extensive Saharan dust deposition throughout the tropical and subtropical Atlantic, the Caribbean, and the Gulf of Mexico would have such consequences only in this small area.
 In January 2005, a red tide started off west central Florida that lasted through early January 2006. In July and August 2005, this event caused hypoxic (“dead”) zones off west-central Florida and within Tampa Bay, Sarasota Bay, and Charlotte Harbor. Benthic communities, fish, turtles, birds, manatees and other marine mammals suffered extensive mortality. The event is hard to explain with the “dust” hypothesis, and it has sparked citizen activism and speculation as to the role of nutrient inputs from agriculture and phosphate mines, and possible links to global warming.
 Were environmental factors different in 2004–2005 compared to other years? Here we examine the origin, intensification, and longevity of this bloom and focus on submarine groundwater discharge (SGD) as a potential source of nutrients that has been previously ignored.
2. Data and Methods
K. brevis cell counts from routine surveys were conducted by the Florida Fish and Wildlife Research Institute (FWRI). Meteorological, river flow, and limited nutrient concentration observations were obtained from the National Hurricane Center and the National Weather Service, the Florida Weather Service, the U.S. Geological Survey (USGS), and the Southwest Florida Water Management District (SFWMD). Limited ground-water nutrient data were collected by the USGS. Daily sea surface temperature (SST), reflectance (SSR), and fluorescence line height (FLH) observations were obtained from AVHRR, SeaWiFS, and MODIS satellite imagery [Hu et al., 2005] to help assess the spatial extent of K. brevis blooms.
3. Results, Discussion, and Conclusion
3.1. 2005 Red Tide Episode
 Blooms detected on 19 December 2004 off Tampa Bay in SeaWiFS and MODIS images contained Heterosigma, which can cause fish kills but is not harmful to humans. On 28 and 29 December, MODIS FLH images revealed a bloom covering ∼1100 km2 immediately off Tampa Bay. By 21 January 2005, the patch had grown to ∼7140 km2 (Figure 1a). A field survey on 19 January revealed high (105–106 cells L−1) to very high (>106 cells L−1) K. brevis concentrations near the 30–40 m isobath. Nil to low K. brevis counts were observed near the coast (Figure 1a).
 Over subsequent months, K. brevis patches were observed along the coast and within the estuaries from Tampa Bay to Charlotte Harbor. The red tide intensified between August and October 2005, occupying the entire inner shelf (<50 m; Figure 1b) in late September. The chlorophyll anomaly of 2005 was sustained (Figure 2) and persisted through January 2006.
 The “dust” hypothesis does not explain the timing, extent and longevity of the 2005 red tide. The 2005 SeaWiFS and MODIS images reveal small Trichodesmium surface slicks near the red tide of 2005. However, we have not identified any out-of-season, significant, or long-lasting dust events in 2005. The SeaWiFS aerosol optical thickness was low throughout the year.
 Satellite-derived SST anomalies showed cooler inner shelf waters by 0.5–1.0°C between March and May 2005 relative to the previous ten-year monthly climatology (Figure 2). This suggests that upwelling of colder waters occurred immediately at the coast in early 2005. The entire shelf showed normal or above average SSTs approximately between June and October. Surface drifters released by NOAA/AOML near the southwest Florida coast during the first half of 2005 showed very slow (<5 cm s−1) southward drift along the coast. After June 2005, inner shelf drifters showed slow motion to the north, parallel to the coast, suggesting that downwelling took place during the second half of 2005. During this period, the red tide intensified and expanded over the entire inner shelf, even as direct and strong mixing was effected by hurricanes Dennis (9–10 July 2005) and Katrina (24–29 August 2005).
3.2. River Nutrients and HABs in Coastal Florida Waters
 Rainfall over Florida between October 2004 and February 2005 was comparable to or below typical climatological values. However, in August and September 2004, Florida experienced four hurricanes (Charley, Frances, Ivan, and Jeanne) and some of the highest precipitation since 1970 (>38 cm in a month; Figure 3). This led to high discharge by Florida rivers into early 2005. The Caloosahatchee River, altered since the 1880's and with flow progressively more regulated since the 1930's, showed increasing flow since 1999 (Figure 3b). The Suwannee, Peace, and Alafia Rivers also showed high flow in 2005 (data not shown). This in part explains the lower coastal and inner shelf salinities detected in winter 2005 surveys relative to recent years (Figure 1a).
 A moderate K. brevis population (3 × 105 cell L−1, or ∼3 mg Chl m−3) requires 0.056–0.267 μM day−1 dissolved inorganic nitrogen (DIN) and 0.002–0.012 μM day−1 phosphorus (DIP) to sustain a ∼0.2 division day−1 growth rate [Vargo et al., 2004]. Therefore, 3 mg m−3 (Figure 2) within a 5 m surface layer of ∼67500 km2 require ∼13.5 × 106 moles DIN and 2.25 × 106 moles DIP day−1. Major rivers supply an annual average of about 1.14 × 106 moles DIN and 0.55 × 106 moles DIP day−1 (Table 1). Small rivers and non-point coastal runoff may contribute an additional amount equivalent to about half of these fluxes. These estimates include wastewater contributions. Therefore, a total of ∼1.71 × 106 moles DIN and 0.83 × 106 moles DIP day−1 are normally delivered by surface runoff to west-central Florida's estuaries and coast. Dissolved organic nitrogen (DON) from these sources may be 50–100% higher, yet how much of DON is biologically available is unknown (typically between 10–70%) [Kroeger et al., 2006]. Discharge by the major rivers was 30–100% higher after fall 2004, but riverine nutrient concentrations were only 5–20% above normal values. Clearly, rivers are an important nutrient source to coastal blooms, yet even the higher 2004–2005 surface discharges provided barely sufficient DIP and between three and five times less DIN than required to maintain the 2004–2005 bloom. Similarly, the intense blooms were likely not fueled by nutrient recycling (G. A. Vargo et al., Nutrient availability in support of Karenia brevis blooms on the central West Florida Shelf: What keeps Karenia blooming?, submitted to Cont. Shelf. Res., 2006). Over the course of 2005, diffusion, mixing, advection and sinking of the nutrients and particles would have diluted the nutrients and dissipated the bloom.
Table 1. Flow Rate and Nutrient Concentration of Major Central Florida Rivers That Discharge Into the Gulf of Mexico (1997–2004)a
 Groundwater has long been identified as a nutrient source to estuarine and coastal waters [D'Elia et al., 1981; Dowling et al., 2004; Paytan et al., 2006]. Florida has 27 of the USA's 78 first order springs (64 million gallons per day), with the three largest land-based springs located near the west Florida coast (Figure 4) [Rosenau et al., 1977]. Indeed, many large submarine springs are located off west-central and northern Florida where red tides also occur, but the link between SGD nutrient inputs and red tides has not been addressed.
 Despite a study in the Florida Keys [Lapointe et al., 1990], nutrient fluxes from coastal ground water discharges in relation with red tides have received little attention in Florida. Fanning et al.  observed elevated 222Rn activities in bottom waters immediately off Tampa Bay, and Cable et al.  demonstrated the utility of this isotope as a groundwater tracer. Cunningham et al.  examined the hydrogeologic framework that leads to SGD off Charlotte Harbor. SGD nutrients may be enriched over natural levels by septic tank effluents.
 We pose the following hypothesis: the nitrogen demand of the 2005 red tide was satisfied by the elevated runoff and SGD caused by the high precipitation associated with the 2004 hurricanes. Coastal nutrient budgets need to account for benthic fluxes across leaky coastal margins [Swarzenski and Kindinger, 2003; Dowling et al., 2004]. SGD DIN and DIP fluxes within Tampa Bay far exceed those from local rivers, and DON fluxes from the two sources are also comparable (Table 2) [Swarzenski et al., 2006]. Tampa Bay SGD alone provides nearly 35% as much DIN as all north and central Florida rivers draining toward the Gulf of Mexico combined.
All west-coast rivers for north and central FL, moles d−1c
0.83 × 106
1.71 × 106
2.45 × 106
 We propose that the nitrogen inputs to the inner shelf off west-central Florida from these SGD may exceed those from riverine and atmospheric sources [e.g., Swarzenski et al., 2001; Kim and Swarzenski, 2006], such as observed in other coastal areas [Moore et al., 2002]. Further possible evidence of SGD is the cooling anomaly of inner shelf waters in early 2005 (Figure 2). Coastal SGD may delay and modulate the input derived from precipitation. SGD-derived nutrients may trigger and sustain algal blooms, and likely overwhelm any aeolian-derived nutrient and iron contributions to Florida coastal waters.
 Future studies of this region need to test this alternative to the “dust” hypothesis. Since SGD take place around the Gulf of Mexico where red tides occur, this process should be quantified in a systematic manner, for example, by pinpointing the locations of submarine springs off west-central Florida and quantifying their volume flow and nutrient flux.
 This work was supported by NASA (NNS04AB59G and NAG5-10557), the NOAA Coastal Ocean Program (NA04NOS4780022), and the US Office of Naval Research (N00014-02-1-0972 SEACOOS program). PWS was supported by the USGS Coastal Marine Geology Program. We thank Dr. Gabriel Vargo (USF) for his review of the nutrient budgets, David English and James Ivey for their help in obtaining field observations. We are indebted to several Federal and state agencies, including FWRI, SWFWMD, USGS, EPA, NWS, NHC, and NOAA/AOML for sharing their data. We thank the reviewers for their critical yet constructive comments.