Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Harmful Algae. Author manuscript; available in PMC 2008 April 24.
Published in final edited form as:
Harmful Algae. 2007; 6(2): 232–252.
doi:  10.1016/j.hal.2006.08.005
PMCID: PMC2330169

Long-term increase in Karenia brevis abundance along the Southwest Florida Coast


Data collected along the southwest coast of Florida between Tampa Bay and Sanibel Island on the abundance of the toxic dinoflagellate Karenia brevis from 1954 to 2002 were examined for spatial and temporal patterns. K. brevis was found to be approximately 20-fold more abundant within 5 km of the shoreline than 20-30 km offshore. Overall, K. brevis was approximately 13-18-fold more abundant in 1994-2002 than in 1954-1963. In 1954-1963, K. brevis occurred primarily in the fall months. In 1994-2002, it was more abundant not only in the fall, but also in the winter and spring months. It is hypothesized that greater nutrient availability in the ecosystem is the most likely cause of this increase in K. brevis biomass, and the large increase in the human population and its activities in South Florida over the past half century is a major factor.

Keywords: Karenia brevis, Red tide, Harmful algal blooms, Florida, Nutrients, Eutrophication

1. Introduction

1.1. History of Karenia brevis in southwest Florida

Blooms of Karenia brevis, a photosynthetic dinoflagellate that produces a suite of similar compounds called brevetoxins (Baden, 1989), often occur along the southwest coast of Florida (Steidinger et al., 1998; Kusek et al., 1999). These toxins affect vertebrate nervous systems by altering sodium channels (Kirkpatrick et al., 2004). K. brevis occurs naturally in the Gulf of Mexico at concentrations up to around 103 cells L-1 (Tester and Steidinger, 1997). It also occasionally generates large dense blooms (>105 cells L-1) that can kill fish, marine mammals, and other marine life (Landsberg, 2002; Flewelling et al., 2005). The organism is also filtered out of the seawater as food by shellfish, resulting in high concentrations of brevetoxin in shellfish that can harm human consumers. Brevetoxin can also become aerosolized, causing respiratory problems in humans and marine mammals (Backer et al., 2003).

Early Spanish explorers in the 15th and 16th centuries such as Cabeza de Vaca described fish kills in the Gulf of Mexico that appear similar to the fish kills generated by K. brevis in present times (Tester and Steidinger, 1997; Kusek et al., 1999). Major fish kills off the Florida coast were definitely documented by 1844. Davis (1948) documented that the fundamental cause of the major fish kills was the toxic dinoflagellate Gymnodinium breve, subsequently renamed Ptychodiscus brevis and currently known as Karenia brevis in honor of the many decades of dedicated research on this species by Dr. Karen Steidinger. These early harmful algal blooms (HABs) indicate that they can occur naturally when there has been little disturbance of the ecosystem by humans. While HABs do occur naturally, it is plausible to hypothesize that they are increasing as a result of human activities. In many parts of the world, HABs have increased in frequency and areal extent, partly attributed to human activities releasing more nutrients into coastal waters (Smayda, 1990; Hallegraeff, 1993; Anderson et al., 2002; Glibert et al., 2005). Parsons et al. (2002) have shown an increase in the toxic diatom Pseudo-nitzschia off the Mississippi delta that corresponds to increased nutrients from that river. Lam and Ho (1989) found that an increase in dinoflagellate blooms in Tolo Harbor, Hong Kong corresponded with an increase in the number of people in its watershed. This investigation was conducted to determine if K. brevis blooms have increased over the past half-century along the southwest coast of Florida.

1.2. Site description

K. brevis blooms occur primarily over the inner West Florida Shelf (Fig. 1), a shallow carbonate platform that extends approximately 200 km west of the Florida coastline (Weisberg et al., 1996; Meyers et al., 2001). The limestone is riddled with solution holes, sinkholes, and a few submarine springs. This hard bottom is overlain by a patchy, thin layer of sediments, primarily quartz sand inshore and biogenic carbonate sand and shell offshore (Brooks, 1973). The outermost shelf waters are most influenced by variations in the Loop Current, while inshore waters are more influenced by wind and land runoff (He and Weisberg, 2003). The inshore coastal currents are strongly wind driven, generally flowing to the north in the summer and to the south in the winter (Weisberg et al., 1996; Yang and Weisberg, 1999). There is evidence that advection of the inshore shelf waters is much less than that of the offshore shelf waters due to a persistent transport barrier in the middle of the shelf area (Yang et al., 1999; Olascoaga et al., in press).

Fig. 1
Overall average concentration of K. brevis from 1954 to 2002. These data were interpolated using the Inverse Distance Weighted function withing ArcGIS Spatial Analyst (Search radius distance = 11819.628 m). Data from Florida Fish and Wildlife Research ...

Rainfall in South Florida is highly seasonal, leading to highly seasonal river flow as well (Fig. 2). Most of the rivers flow into large bays or lagoons behind barrier islands before the water enters the shelf waters.

Fig. 2
Average monthly rainfall and river flow over the period 1968-2005. Rainfall data at Ft. Myers from South Florida Water Management District (SFWMD). Caloosahatchee River flow at station S79 from SFWMD. Peace River flow at station 02296750 from United States ...

The actual size of the watershed for southwest Florida coastal waters has increased considerably as a result of large scale water diversions carried out in South Florida for wetland drainage, agricultural expansion, urban development, flood control, and water supply purposes. There was approximately a 4-fold enlargement of the Caloosahatchee River watershed as a result of the construction of a canal between Lake Okeechobee and Lake Hicpochee (Fig. 3). Initially, the Caloosahatchee River headwaters started at Lake Hicpochee and there was no direct connection to Lake Okeechobee. In the 1880s, a small canal was constructed between Lake Hicpochee and Lake Okeechobee. It was expanded in the 1920s and 1930s (Steinman et al., 2002) and even more in the early 1960s (Gunter and Hall, 1962). As a result, the entire watershed of the Kissimmee River, along with Taylor Creek and Fisheating Creek became part of the Caloosahatchee watershed. Much of the Kissimmee River basin was drained and converted to agriculture and ranching operations (McPherson and Halley, 1996; Steinman et al., 2002). In the 1960s, the northern third of the Everglades were drained and turned into agricultural fields, primarily sugar cane (Snyder and Davidson, 1994). As a result of backpumping into the lake, these agricultural lands to the south and east of Lake Okeechobee also became part of the watershed.

Fig. 3
Map of original Caloosahatchee River watershed and the current watershed after the connection to Lake Okeechobee was built. Data from SFWMD.

Along with these large-scale changes in the watershed, the human population along the southwest coast of Florida has increased 10-40-fold during the past half century (Table 1). There have also been dramatic increases in the past few decades in nutrient and chlorophyll concentrations (Turner et al., 2006) and in macroalgal abundance (Lapointe and Bedford, in press) in these coastal waters. Because of natural deposits of phosphorite along the west coast of Florida, nitrogen:phosphorus ratios are usually well below the Redfield ratio of 16:1 and phytoplankton along the coastline are usually nitrogen limited (Walsh and Steidinger, 2001; Brand, 2002; Vargo et al., 2004).

Table 1
Human population in Southwest Florida over time by county (from Florida Statistical Abstracts, Floyd et al., 2004)

2. Methods

2.1. The main data set

To test the hypothesis that K. brevis populations have increased, data on its abundance collected from January 1954 to May 2002 were used (Florida Fish and Wildlife Research Institute, 2002). Data after May 2002 were not available at the time of this analysis. This data set has few nutrient and other ancillary data associated with it. Because the data were collected by various people and organizations using diverse sampling approaches, the distribution of the data over time and space is not systematic, making data analysis more difficult. Furthermore, some of the data were collected not as a part of an unbiased sampling program, but rather to sample K. brevis blooms. As a result, the raw data most likely overestimate the average abundance of K. brevis in time and space. An examination of the entire data set revealed widespread sampling: both the east and west coasts of Florida, throughout the Florida Keys in the south and along the entire northern coast of the Florida Panhandle. Many of these areas were only sporadically sampled, probably in response to apparent K. brevis blooms. The highest density of data was along the shoreline from Tampa Bay to Sanibel Island, where routine sampling was undertaken to protect human health. This area (Fig. 4) was separated into six 5-km strips for a more detailed examination. The data were examined considering the possibility of spatial and temporal biases as a result of nonrandom sampling. The potential for bias was tested several ways.

Fig. 4
Areas used to analyze inshore-offshore gradients in K. brevis abundance and salinity.

2.2. Reduction of bias

Because K. brevis occurrence is sporadic and somewhat seasonal, the data were examined for seasonality in the number of samples taken. An unbiased sampling program would take equal numbers of samples each month, while a biased sampling program aimed at collecting K. brevis would show considerable seasonality. This was examined by calculating the coefficient of variation for number of samples collected each month. The sporadic nature of K. brevis abundance was also handled by using average data over large areas and long time spans (approximately decadal) that integrate the spatial heterogeneity and temporal variability.

Satellite imagery has been used to locate and track K. brevis blooms, but reliance upon such data could have led to a bias of more blooms sampled in recent years than 50 years ago when satellite imagery was not available. As satellite imagery can only detect K. brevis blooms above a concentrations of around 105 cells L-1 (Tester et al., 1998; Tomlinson et al., 2004), the data were conservatively reanalyzed with the assumption that all samples over 105 cells L-1 were the result of sampling bias and should be eliminated. Similarly, the death of fish also occurs only above around 105 cells L-1 (Tester et al., 1998), while visual detection of K. brevis blooms only occurs above 106 cells L-1 (Steidinger et al., 1998). Therefore, eliminating all data over 105 cells L-1 avoids the bias of deliberately sampling known blooms.

Similarly, it is possible that more sampling was conducted when noticeable blooms occurred than when no bloom were noticed. To reduce this potential problem, the data were averaged for each month first and then all the monthly averages over the time periods were averaged and compared. This prevented the overall averages from being possibly overly influenced by a large number of samples being collected in one month due to a large bloom.

Because one or a few large blooms could greatly influence averages, the concentration frequency spectra were examined in each of the data sets to make sure they were not highly skewed. Examination of the concentration frequency spectra above and below 105 cells L-1 (optically detectable) was also used to evaluate the extent to which biased sampling could influence the results.

2.3. Statistics

Because of the wide range of concentrations of K. brevis observed, a variety of statistical tests were used, with an emphasis on heteroscedastic and nonparametric tests, and logarithmic transformations. In most data sets, mean concentrations of K. brevis were compared using a two-tailed heteroscedastic t-test on both raw data and logarithmically transformed data. For logarithmic transformations, samples with 0 cells L-1 were given a value of 1 cell L-1. The Mann-Whitney nonparametric test was also used for comparing mean concentrations. The concentration frequency distributions were compared by a Chi-square test on data that were binned by samples having concentrations within a certain range (0, 1-1000, 1000-3000, 3000-10,000 cells L-1, etc.). Seasonal comparisons were made by Chi-square tests on data binned by month.

3. Results

3.1. Overall distribution

For the overall data set from 1954-2002, the data indicate the most intense blooms occurred along the shoreline from Tampa to Ft. Myers, with more sporadic blooms offshore, to the south towards Naples and the Florida Keys, and to the north of Tampa Bay (Fig. 1). Because K. brevis blooms are sporadic both spatially and temporally, however, the spatial distribution at any one time would rarely appear as the long-term average distribution.

3.2. Sampling intensity and bias

An examination of number of samples taken each month indicated a dramatic decline in 1963 and a dramatic increase again in 1994. For the area shown in Fig. 4, 2158 samples were taken in the 1954-1963 period (average of 18 samples per month), 1610 samples were taken in the 1964-1993 period (average of 4.5 samples per month), and 3312 samples were taken in the 1994-2002 period (average of 32.8 samples per month). The fewer samples in the longer three-decade period of 1964-1993 indicated a much lower sampling effort, suggesting less systematic sampling and more potential for bias, compared to the 1954-1963 and 1994-2002 periods, when more samples were taken in a much shorter time. As K. brevis is highly seasonal, a biased sampling program would also be expected to be seasonal. Both the 1954-1963 and 1994-2002 periods had a coefficient of variation of 28% for monthly sampling, while the 1964-1993 period had a 64% coefficient of variation. This indicates much more seasonal sampling bias in the 1964-1993 period than the 1954-1963 and 1994-2002 periods. Because of this potential bias and the fewer number of monthly samples taken, the 1964-1993 data were excluded from the temporal analyses. Therefore, the analysis focuses on the 1954-1963 and 1994-2002 periods because of the larger number of samples per year and the reduced sampling bias.

3.3. Inshore-offshore concentration gradients

Examination of the inshore-offshore gradient in K. brevis abundance in the area between Tampa and Ft. Myers sectioned into 5 km wide strips (Fig. 4) indicated that, on average, K. brevis was more abundant along the shoreline than farther offshore (Fig. 5). The inshore area also had lower salinity as a result of land runoff. There was a bias toward more samples close to the shore, probably because inshore samples are more accessible and there is more public concern about K. brevis impacts in inshore waters. The data were normalized, thus the average concentrations calculated were considered to accurately reflect the abundance of K. brevis, not the sampling effort. K. brevis was significantly more abundant in the 0-5 km strip than in either the 5-10 km strip or the offshore 25-30 km strip (Table 2). The data indicate that, on average, K. brevis was approximately 20-fold more abundant in the 0-5 km inshore waters than the 20-30 km offshore waters.

Fig. 5
Average abundance of K. brevis and average salinity in each of the 5 km strips shown in Fig. 4 over the 1954-2002 period. To examine the possible relationship with salinity and land runoff, only samples with associated salinity data were used in this ...
Table 2
Statistics of K. brevis abundance comparing the 0-5 km area and the 5-10 km area; comparing the 0-5 km area and the 25-30 km area shown in Fig. 4. Only samples that had associated salinity data were included

3.4. Changes over time

In both the 1954-1963 and 1994-2002 periods, average K. brevis concentrations were higher inshore than offshore, but throughout the transect, cell concentrations were much higher in the 1994-2002 period than the 1954-1963 period (Fig. 6). K. brevis abundance was significantly higher in the 0-5 km strip than either the 5-10 km strip or the offshore 25-30 km strip in both periods (Table 3). The increase in K. brevis abundance in the entire area was approximately 13-fold and statistically significant (Table 4). Even if only samples less than 105 cells L-1 were used to exclude any possible remote sensing bias (Fig. 7), the difference was highly significant (Table 4). When the data were binned by month first, and average concentrations calculated from monthly averages, the increase from 1954-1963 to 1994-2002 was 18-fold and highly significant (Table 4). The increase over time was highly statistically significant for all six of the 5 km strips along the shoreline, even if only samples with less than 105 cells L-1 were used (Table 5). The proportional increase over time was larger offshore than inshore (Fig. 6, Table 5), indicating that the blooms have extended farther offshore over time.

Fig. 6
Comparison of the average abundance of K. brevis in the areas shown in Fig. 4 between the 1954-1963 and 1994-2002 periods. Data from Florida Fish and Wildlife Research Institute (2002).
Fig. 7
Comparison of the average abundance of K. brevis for samples less than 105 cells L-1 in the areas shown in Fig. 4 between the 1954-1963 and 1994-2002 periods. Data from Florida Fish and Wildlife Research Institute (2002).
Table 3
Statistics of K. brevis abundance comparing the 0-5 km area and the 5-10 km area; comparing the 0-5 km area and the 25-30 km area shown in Fig. 4 in both the 1954-1963 and 1994-2002 periods
Table 4
Statistics of K. brevis abundance in the collective area (0-30 km) shown in Fig. 4, comparing 1954-1963 and 1994-2002
Table 5
Statistics of K. brevis abundance in the 5 km strips shown in Fig. 4, comparing 1954-1963 and 1994-2002

As the 0-5 km strip along the shoreline (Fig. 4) had been extensively sampled, there were enough data to examine the concentration frequency spectra and compare the 1954-1963 and 1994-2002 periods. The percentage of samples having “background” levels less than 1000 cells L-1 of K. brevis dropped from around 85% to 47% over that time span (Fig. 8). By contrast, all the concentration categories above “background” levels of 103 cells L-1 were considerably more frequent in the 1994-2002 period than the 1954-1963 period. Samples with blooms over 106 cells L-1 were about eight times more frequent in the 1994-2002 period than the 1954-1963 period. The slightly higher percentages in the 100,000-300,000 cells L-1 category over the 30,000-100,000 cells L-1 category may reflect a sampling bias toward detectable concentrations over 105 cells L-1. The fact that the higher percentages are seen in both the 1954-1963 and 1994-2002 data sets indicates that this possible bias cannot account for the large increase from 1954-1963 to 1994-2002 in K. brevis abundance.

Fig. 8
Comparison of the concentration frequency spectra of K. brevis in the 0-5 km strip shown in Fig. 4 between the 1954-1963 and 1994-2002 time periods. Data from Florida Fish and Wildlife Research Institute (2002).

The highest density that K. brevis blooms achieved were examined by averaging the highest 1% of the data points in the two time periods. For 1954-1963, the top 22 samples (out of 2158 samples) averaged 2.5 million cells L-1, while for 1994-2002, the top 33 samples (out of 3312 samples) averaged 34.7 million cells L-1. T-tests and a Mann-Whitney nonparametric test indicate that the difference is highly significant (P < 0.0001). These data indicate that the blooms have achieved much higher densities in more recent years than a half century earlier.

The examination of mean concentrations, highest achieved concentrations, concentration frequency spectra, and inshore-offshore transect concentrations all indicate that K. brevis has increased substantially in abundance over the past half century in the area examined.

3.5. Changes in seasonality

An examination of the percentage of samples each month showing greater than 3 × 104 cells L-1 and no cells both show that K. brevis is least prevalent in June and July and is most abundant in the fall months (Fig. 9). K. brevis abundance then declined in late winter and spring. A comparison of the seasonal occurrence of K. brevis between the 1954-1963 and 1994-2002 periods (Fig. 10A) indicated a shift in the seasonal pattern. A Chi-square test indicated a probability of less than 0.0001 that the two seasonal patterns are the same. A comparison of the seasonality in cell concentrations for the two periods, also showed a significant change in the seasonal pattern (Fig. 10B). The differences were highly significant (T-test, P < 0.05) for all months except March and April.

Fig. 9
Average monthly frequency of absence of K. brevis and concentrations over 3 × 104 cells L-1 over the 1954-2002 time period in the area shown in Fig. 4. Note that the absence frequency scale is reversed. Data from Florida Fish and Wildlife Research ...
Fig. 10
Comparison of seasonal occurrence and abundance of K. brevis in 1954-1963 and 1994-2002 in the area shown in Fig. 4. Data from Florida Fish and Wildlife Research Institute (2002). (A) Average monthly frequency of K. brevis concentrations over 3 × ...

A similar shift in the seasonal pattern was also observed in just the 0-5 km strip along the shoreline, indicating a significant increase from 1954-1963 to 1994-2002 in K. brevis abundance throughout most of the year (Fig. 11). A Chi-square test indicated that the increase was significant (P < 0.0001).

Fig. 11
Comparison of seasonal occurrence and abundance of K. brevis in 1954-1963 and 1994-2002 in the 0-5 km strip of the area shown in Fig. 4. Data from Florida Fish and Wildlife Research Institute (2002). (A) Average monthly frequency of K. brevis concentrations ...

In the 1954-1963 period, K. brevis was sparse in the spring and summer months. On average, there was a dramatic increase in October and high concentrations continued throughout the fall until January and then there was a decline in late winter and spring. By contrast, in the 1994-2002 period, the large increase began in August and K. brevis remained quite prevalent throughout the fall and winter, and into the spring. The decline into the spring was much slower in more recent years than a half century earlier.

4. Discussion

4.1. Overall changes and potential effects

In general, K. brevis was more abundant nearshore than offshore and became more abundant from 1954-1963 to 1994-2002. Overall, concentrations increased approximately 13-18-fold and the blooms extended farther offshore. K. brevis occurred at relatively high concentrations for more months of the year in 1994-2002 relative to 1954-1963, occurring earlier in the fall and extending on through the winter and into the spring.

This extension of the K. brevis blooms into the spring months causes two problems today that were much less of a problem in the past. Late winter and early spring is the prime time for tourists from more northern latitudes to visit Florida. The higher incidence and abundance of K. brevis, with the associated irritating aerosols and fish kills, during this time period can cause problems to the local tourist industry and economy. Springtime is also when West Indian manatees (Trichechus manatus latirostris) migrate from the estuaries back into the coastal waters (Landsberg and Steidinger, 1998; Landsberg, 2002). Half a century ago, K. brevis was usually not very abundant at this time (Figs. (Figs.1010 and and11).11). As a result of K. brevis occurrence now extending more into the spring in recent years, many manatees are now being exposed to brevetoxin and dying (Landsberg and Steidinger, 1998; Florida Fish and Wildlife Research Institute, 2006).

The concentration frequency spectra (Fig. 8) indicated that all concentration categories above the 103 cells L-1 background level, not just the lethal bloom concentrations, have increased over time. The significant increase in these sublethal concentrations of K. brevis could have serious implications. These concentrations are not high enough to kill organisms directly, but perhaps allow for the accumulation of the brevetoxin in their tissues. Biomagnification of this lipid-soluble toxin through the food web could lead to rather high concentrations at higher trophic levels. Flewelling et al. (2005) have argued that this scenario may explain the mortality of many bottlenose dolphins (Tursiops truncatus) and West Indian manatees (Trichechus manatus latirostris) along the west Florida coast in recent years in areas where no large blooms of K. brevis could be found at the time. The high concentrations of brevetoxins observed in the tissues and stomach contents of the marine mammals suggested trophic transfer and biomagnification of brevetoxin (Flewelling et al., 2005). This evidence, along with the observed increase in frequency of sublethal concentrations of K. brevis from 1954-1963 to 1994-2002 (Fig. 8) suggests a possible increasing threat to seafood safety in west Florida coastal waters.

4.2. Possible causes of the apparent increase over time

The data show a significant increase in the average concentrations, peak concentrations, areal extent, and seasonal duration of K. brevis off the southwest coast of Florida over the past half-century. Possible causes for this to be considered here are sampling bias, long term changes or oscillations in the ecosystem, increases in iron-stimulated nitrogen fixation, and increases in nutrients from land.

4.3. Sampling bias

While the data available (Florida Fish and Wildlife Research Institute, 2002) do not indicate the sampling strategies employed, it is almost certain that at least some of the sampling was not unbiased, but rather the result of searches for K. brevis blooms to sample and study, leading to an overestimate of its true average abundance. There was no evidence that sampling bias was so much more severe in 1994-2002 than 1954-1963 that it could result in a 13-18-fold increase in estimated biomass. Indeed, it is more likely that there was more bias intheearly days of K. brevis research to collect samples for examination of the newly discovered toxic dinoflagellate as a source of fish kills, and less bias in more recent times with more emphasis on objective oceanographic sampling. The data on seasonal sampling bias indicated relatively little bias during the 1954-1963 and 1994-2002 periods and no significant difference between the two in monthly sampling variability (both had a coefficient of variation of 28%).

Another potential bias that must be considered is often called “the observer effect”. As there were many more people along the coast and out on the water in 1994-2002 than 1954-1963, more K. brevis blooms could have been reported in the later years that might have gone unnoticed a half-century earlier. A comparison of the number of samples taken indicated no major differences between the 1954-1963 and 1994-2002 periods (2158 and 3312 samples, respectively). Furthermore, all data were normalized to number of samples, so it does not affect the estimated concentration of K. brevis.

To avoid the potential problem of more samples being taken during obvious blooms than when no blooms occur, the data were also binned by month and monthly averages used to compare 1954-1963 to 1994-2002 average concentrations. This analysis showed an 18-fold increase that was highly significant.

One possible bias is the use of remote sensing for the detection of K. brevis blooms, a technology not available in 1954-1963. This is definitely a possibility farther offshore where K. brevis is sparse and not noticed by many people. This is less probable along the shoreline where there is routine sampling whether remote sensing shows the presence of K. brevis or not. Furthermore, people in the 1950s did not need remote sensing to detect red tide along the shoreline because of the observations of dead fish and noxious aerosols. As a result, one would expect remote sensing capabilities to lead to a bias offshore but not along the shoreline. Examination of only data below the remote sensing detection limit of 105 cells L-1 (Fig. 7, Table 5) indicated that any potential remote sensing bias alone cannot explain the large increase in average K. brevis concentration.

The increase in the frequency of K. brevis occurrence at concentrations that are elevated but not noticeable to remote sensing or the human eye (103 to 105 cells L-1), as seen in the concentration frequency spectra (Fig. 8) also indicated that sampling bias, the observer effect, and remote sensing bias alone cannot explain the large increase in K. brevis concentrations. It also cannot explain the change in the seasonal pattern in K. brevis occurrence. It is concluded that the apparent increase in K. brevis abundance is real.

4.4. Long term changes or oscillations in the ecosystem

While using the average of a decade of data helps average out the year to year variability in K. brevis occurrence and concentration, one cannot rule out some long term natural change or oscillation in factors that influence the ecology of K. brevis. For example, one possibility is that the North Atlantic Oscillation, Atlantic Multidecadal Oscillation, or periodicity in hurricane activity could ultimately influence the population dynamics of K. brevis. An examination of the timing of those oscillations (Enfield et al., 2001; Stenseth et al., 2002; Chavez, 2004; Trenberth, 2005) however indicates that the 1954-1963 and 1994-2002 periods are approximately in phase, not out of phase with each other. This does not preclude some other oscillatory phenomenon from causing K. brevis to be less prevalent in 1954-1963 than 1994-2002, but we know of none at the present time.

Another possible hypothesis is that the increase in water temperature and change in water mass structure over the past half-century (Houghton et al., 2001; Rayner et al., 2003; Sheppard and Rioja-Nieto, 2005) caused a shift in community structure, but our understanding of it is too poor to document such a change and demonstrate a mechanism by which it could lead to an increase in K. brevis. Any change in the ecosystem is just as likely to cause a decrease in K. brevis as an increase.

The decline of nekton on the west Florida shelf as a result of fishing pressure has probably affected community structure as a result of “top down controls”, but again, no specific mechanism can be identified that would predict an increase in K. brevis as a result. One can envision ways in which changes in temperature and community structure could lead to reduced grazing or competition from other phytoplankton species, but one can just as easily envision exactly the opposite. Furthermore, peak blooms of K. brevis tend to become almost monospecific (Steidinger and Vargo, 1988), so the observed increase over a half-century is not just a shift toward K. brevis being a larger proportion of the phytoplankton community. Overall K. brevis biomass is higher and that implies higher nutrient availability.

4.5. Nutrients

Because most of the nutrients available end up in the biomass of K. brevis as the bloom matures and becomes almost monospecific (Steidinger and Vargo, 1988), the amount of available nutrients determines the maximum biomass of K. brevis that can develop. The 13-18-fold increase in K. brevis abundance from 1954-1963 to 1994-2002 (Table 4) and the increase in the highest achieved concentrations imply an increase in nutrient availability.

Lenes et al. (2001) and Walsh and Steidinger (2001) have hypothesized that iron-rich dust from North Africa stimulates the growth of Trichodesmium, which in turn enriches the ecosystem with nitrogen by way of nitrogen fixation. This hypothesis appears quite plausible. It is less clear that it can explain a 13-18-fold increase in K. brevis over a half-century or the much higher concentrations within 5 km of the coastline. Prospero and Lamb (2003) estimate that there may have been as much as a 4-fold increase in dust from Africa from 1950s to 1980s and no significant changes throughout the 1980s and 1990s. This increase could be a contributor to the increase in K. brevis abundance. As African dust is spread throughout the West Florida Shelf, however, this hypothesis does not explain why concentrations of K. brevis are much higher a long the shoreline than farther offshore (Fig. 5). While iron could be a limiting nutrient offshore, it is unlikely to be a limiting nutrient in the shallow waters along the coastline. Because of large amounts of iron supplied by land runoff and sediments, additional iron from African dust is probably insignificant inshore.

The higher concentrations of K. brevis along the shoreline where salinity is lower (Fig. 5) suggests nutrient-rich water from land based sources is important. Many scientific papers have suggested that there is a connection between land runoff and red tides (Webb, 1886; Agassiz, 1890; Gunter et al., 1947, 1948; Slobodkin, 1953; Gunter and Hall, 1962; Steidinger and Joyce, 1973; Dixon and Steidinger, 2004).

We know of no natural sources of nutrients that have increased 13-18-fold. The most plausible hypothesis for a large increase in nutrient availability is that the large increase in human activities in South Florida is involved. The large increase in the human population along the southwest coast of Florida (Table 1) would be expected to produce more sewage, more disturbance of terrestrial and wetlands ecosystems and their ability to sequester nutrients, and more land surface runoff. There has also been a large increase in agriculture, fertilizer use, mining of phosphate deposits, and oxidation of nitrogen-rich organic peat (McPherson and Halley, 1996). While we do not have historical data on fertilizer use in the watershed of the southwest Florida coast, nationwide trends probably reflect the changes in Florida reasonably well. Heimlich (2003) found that 2.7 million tonnes of nitrogen fertilizer was used in 1960 and 11.4 million tonnes were used in 1980 in the United States. The amount of fertilizer used after 1980 remained about the same through to 1998.

The 4-fold enlargement of the Caloosahatchee River watershed (Fig. 3) alone would have also greatly increased the nutrient load to the coastal waters. The release of buried nutrients as a result of draining the northern Everglades, which led to oxidation of the organic peat and release of the associated nutrients, has been a major factor in the eutrophication of many areas of South Florida, including Lake Okeechobee (Brand, 2002; Steinman et al., 2002), and thus the Caloosahatchee River and coastal waters downstream.

The similar seasonal pattern of flow down the rivers (Fig. 2) and K. brevis (Fig. 9), with the peak in K. brevis prevalence a couple of months after the peak river flow, is suggestive. It is important to recognize that these are just averages over many years, and river runoff does not necessarily result in a K. brevis bloom each year. The timing of individual blooms is actually quite sporadic from year to year. The sporadic timing suggests local physical circulation, species competition, and ecosystem processes determine whether or not a bloom of K. brevis develops. River flow and nutrient availability may determine how large a K. brevis bloom can become once other environmental circumstances allow it to outcompete other phytoplankton species and develop.

Interestingly, a comparison of seasonal flow in the 1994-2002 period with the 1964-1973 period (1954-1963 data are not available) for the Caloosahatchee River (Fig. 12) indicated the same increase and extension into late winter and spring as seen in K. brevis seasonality (Figs. (Figs.1010 and and11).11). The data suggest that increased flow down the Caloosahatchee River in the winter and spring may be related to increased K. brevis downstream in the winter and spring. Interestingly, Gunter and Hall (1962) recognized the potential problem, but argued that increasing the flow of water from Lake Okeechobee down the Caloosahatchee River in spring would not increase K. brevis abundance because it occurs in the fall. The long-term data suggest that increasing flow in the spring has changed the timing of K. brevis occurrence, so that it now often occurs in both the fall and spring.

Fig. 12
Comparison of the average monthly flow through S79 on the Caloosahatchee River between the 1964-1973 and 1994-2002 time periods. Data before 1964 are not available. Data from South Florida Water Management District.

Few long term records are available, but Turner et al. (2006) found over a 10-fold increase in both nitrate in the Peace River and sediment chlorophyll in Charlotte Harbor at the mouth of the Peace River from 1960 to 1980. In sediment cores dated back 200 years, sharp increases in organic carbon, nitrogen, phosphorus, and biogenic silica were observed after 1950 (Turner et al., 2006). The data indicate that nutrient enrichment has occurred in these coastal waters.

Hu et al. (2006) have argued that nutrients from rivers along the west coast of Florida alone are insufficient to generate the recent large blooms of K. brevis and that groundwater flow is a likely major source of nutrients. While estimating the actual flux of nutrients into coastal waters by groundwater is much more difficult than the flux of river input, it is well established that the flux of nutrients through groundwater is a significant source of nutrients to coastal waters (D’Elia et al., 1981; Simmons, 1992; Moore, 1999; Paytan et al., 2006) and should not be ignored. Miller et al. (1990) demonstrated the importance of groundwater flow in Charlotte Harbor near the mouth of the Peace River. Scott et al. (2006) found approximately a 20-fold increase in nitrate concentrations in 13 of Florida’s largest springs from 1992 to 2001, indicating a large increase in groundwater nutrients.

Hu et al. (2006) argued that river flow could only provide 20-30% of the nitrogen needed to generate the K. brevis bloom observed in 2004-2005, but acknowledged the calculation was based only upon inorganic nitrogen, and that organic nitrogen could provide some unknown additional amount of nitrogen to K. brevis blooms. Using the nutrient bioassay methods of Brand (2002), we have estimated the total amount of nitrogen available to phytoplankton in the Caloosahatchee and Peace Rivers is approximately two to three times larger than just the inorganic nitrogen (Brand and Compton, unpublished data). These are the two large strivers along the south west coast of Florida (Fig. 2). Furthermore, in addition to river flows, non-point source flows along the coastline, which is now highly developed in southwest Florida, must also be considered. This does not diminish the potential importance of groundwater fluxes, but does indicate that surface runoff is not necessarily minor. We suspect a combination of river flow, non-point source inputs, and groundwater provide sufficient nutrients to generate the K. brevis blooms observed inshore.

A factor that may be enhancing the effects of land runoff of nutrients is reduced advection due to a persistent transport barrier (Yang et al., 1999; Olascoaga et al., in press) in the nearshore waters where K. brevis is most prevalent (Fig. 1). The long residence time inshore could allow for the buildup of both nutrients and K. brevis populations.

We hypothesize that the many decades of increased nutrient flux to the coastal ecosystem from land have increased the nutrient pool in this ecosystem. These nutrients can be stored in sediments, detritus, and long-lived macroalgae and seagrasses. The data of Turner et al. (2006) on increasing sediment nutrients, and Lapointe and Bedford (in press) on increasing amounts of macroalgae in these coastal waters support this hypothesis. We hypothesize that this large reservoir of nutrients on the West Florida Shelf ultimately allows for the development of blooms with nutrient requirements that exceed the actual input of nutrients at any given time. This could explain why estimates of nutrients from river flow alone (Vargo et al., 2004; Hu et al., 2006) appear to sometimes be insufficient to support some of the largest blooms. It is hypothesized that processes currently not understood allow for the eventual transfer of nutrients from these benthic pools to K. brevis under certain circumstances, promoting its increase in biomass. Walsh et al. (2001) and Walsh and Steidinger (2001) have also suggested that benthic nutrient sources play an important role in the development of K. brevis blooms. A long-term increase in the benthic nutrient pool could then lead to an overall increase in planktonic K. brevis biomass.

Another factor to consider is the life cycle of K. brevis. The entire life cycle of K. brevis is not known, but a benthic stage is suspected (Steidinger et al., 1998). If this is the case, elevated benthic nutrients could also be enhancing benthic populations of K. brevis, ultimately leading to the larger planktonic populations observed.

For blooms that start offshore (Tester and Steidinger, 1997; Steidinger et al., 1998; Walsh and Steidinger, 2001), the initial bloom may not be supported by nutrients from land runoff, but we hypothesize that land-based nutrients contribute to the much higher K. brevis concentrations found inshore (Figs. (Figs.55 and and6)6) once the blooms have been transported inshore. The association of high K. brevis abundance with high silicate concentrations in all four blooms they examined led Vargo et al. (2004) to conclude that local estuaries were a significant source of nutrients to the blooms.

The complete life cycle of K. brevis, the exact pathways of nutrients leading to K. brevis populations, and how K. brevis can outcompete other faster growing phytoplankton species under certain circumstances are not known. Nevertheless, the large increase in K. brevis abundance over the past 50 years requires a large increase in nutrient availability. We hypothesize that the large increase in human activities in South Florida in the past 50 years is ultimately the major source of these additional nutrients. The much higher concentrations of K. brevis inshore than off shore also support the hypothesis that nutrient inputs from land are a major factor.

5. Conclusions

The data indicate that, in general, K. brevis occurs at much higher concentrations nearshore than offshore. Comparing data from 1954-1963 to 1994-2002 periods, K. brevis has become more abundant. On average, abundance became 13-18 times higher and the blooms extended farther offshore. Comparing data from 1954-1963 and 1994-2002, K. brevis occurred at relatively high concentrations for more months of the year during the later time period. The much higher biomass requires more nutrients. It is hypothesized that increased nutrient-rich land runoff from increased human activities in South Florida is the most plausible source of these additional nutrients.


We would like to thank the Florida Fish and Wildlife Research Institute and South Florida Water Management District for access to their data files. This research was financially supported by the Cove Point Foundation; Lee County, FL; Bonita Springs, FL; NSF #OCE0432368; NIEHS P50 ES12736.[SS]


  • Agassiz A. Note of Shaler, 1890. Bull. Mus. Comp. Zool. Harv. 1890;16:157–158.
  • Anderson DM, Glibert PM, Burkholder JM. Harmful algal blooms and eutrophication: nutrient sources, composition, and consequences. Estuaries. 2002;25:704–726.
  • Backer L, et al. Recreational exposure to aerosolized brevetoxins during Florida red tide events. Harmful Algae. 2003;2:19–28. [PubMed]
  • Baden DG. Brevetoxins: unique polyether dinoflagellate toxins. FASEB J. 1989;3:1807–1817. [PubMed]
  • Brand LE. The transport of terrestrial nutrients to South Florida coastal waters. In: Porter JW, Porter KG, editors. The Everglades, Florida Bay, and Coral Reefs of the Florida Keys. CRC Press; Boca Raton, Florida: 2002. pp. 353–406.
  • Brooks HK. Geological oceanography. In: Jones JI, Ring RE, Rinkel MO, Smith RE, editors. A Summary of Knowledge of the Eastern Gulf of Mexico 1973. State University System of Florida Institute of Oceanography; St. Petersburg, FL: 1973. pp. IIE1–19.
  • Chavez FP. Biological consequences of interannual to multidecadal variability. In: Robinson AR, Brink KH, editors. The Sea. Vol. 13. Harvard University Press; 2004. pp. 643–679.
  • Davis CC. Gymnodinium brevis sp. nov., a cause of discolored water and animal mortality in the Gulf of Mexico. Bot. Gaz. 1948;109:358–360.
  • D’Elia CF, Webb KL, Porter JW. Nitrate-rich groundwater inputs to Discovery Bay, Jamaica: a significant source of N to local coral reefs? Bull. Mar. Sci. 1981;31:903–910.
  • Dixon LK, Steidinger KA. Correlation of Karenia brevis presence in the eastern Gulf of Mexico and with rainfall and riverine flow. In: Steidinger KA, Landsberg JH, Tomas CR, Vargo GA, editors. Harmful Algae. UNESCO; 2004. 2002. pp. 29–31.
  • Enfield DB, Mestas-Nunez AM, Trimble PJ. The Atlantic multidecadal oscillation and its relation to rainfall and river flows in the continental U.S. Geophys. Res. Lett. 2001;28:2077–2080.
  • Flewelling LJ, et al. Red tides and marine mammal mortalities. Nature. 2005;435:755–756. [PMC free article] [PubMed]
  • Florida Fish and Wildlife Research Institute . Version 2.0 (1954-2002) FWRI Data CD; 2002. Red Tides in Florida: Harmful Algal Bloom Historical Database.
  • Florida Fish and Wildlife Research Institute Red tide contributes to second-highest mortality year on record. 2006. Press release: (
  • Floyd SS, Wilson PA, Middleton P. Florida Statistical Abstract. University of Florida; 2004.
  • Glibert PM, Seitzinger S, Heil CA, Burkholder JM, Parrow MW, Codispoti LA, Kelly V. The role of eutrophication in the global proliferation of harmful algal blooms. Oceanography. 2005;18:198–209.
  • Gunter G, Hall GE. Biological investigations of Caloosahatchee estuary in connection with Lake Okeechobee discharges through Caloosahatchee River. 1962. A Report to the District Engineer, Jacksonville District, Corps of Engineers.
  • Gunter G, Walton Smith FG, Williams RH. Mass mortality of marine animals on the lower west coast of Florida, November 1946-January 1947. Science. 1948;105:256. [PubMed]
  • Gunter G, Williams RH, Davis CC, Walton Smith FG. Catastrophic mass mortality of marine animals and coincident phytoplankton bloom on the west coast of Florida, November 1946-August 1947. Ecol. Monogr. 1947;18:309–324.
  • Hallegraeff GM. A review of harmful algal blooms and their apparent global increase. Phycologia. 1993;32:79–99.
  • He R, Weisberg RH. A loop current intrusion case study on the West Florida Shelf. J. Phys. Oceanogr. 2003;33:465–477.
  • Heimlich R. Agriculture Handbook No. AH722. Economic Research Service, United States Department of Agriculture; 2003. Agricultural resources and environmental indicators, 2003.
  • Houghton JT, Ding Y, Griggs DJ, Noguer M, van der Linden PJ, Dai X, Maskell K, Johnson CA. Climate Change 2001: The Scientific Basis. Cambridge University Press; 2001.
  • Hu C, Muller-Karger FE, Swarzenski PW. Hurricanes, submarine groundwater discharge, and Florida’s red tides. Geophys. Res. Lett. 2006;33:L11601. doi:10.1029/2005GL025449.
  • Kirkpatrick B, et al. Literature review of Florida red tide: Implications for human health effects. Harmful Algae. 2004;3:99–115. [PMC free article] [PubMed]
  • Kusek KM, Vargo G, Steidinger K. Gymnodinium breve in the field, in the lab, and in the newspaper-a scientific and journalistic analysis of Florida red tides. Contri. Mar. Sci. 1999;34:1–229.
  • Lam CWY, Ho KC. Red tides in Tolo Harbour, Hong Kong. In: Okaichi T, Anderson DM, Nemoto T, editors. Red Tides: Biology, Environmental Science, and Toxicology. Elsevier; New York: 1989. pp. 49–52.
  • Landsberg JH. The effects of harmful algal blooms on aquatic organisms. Rev. Fish. Sci. 2002;10:113–390.
  • Landsberg JH, Steidinger KA. A historical review of Gymnodinium breve red tides implicated in mass mortalities of the manatee (Trichechus manatus Latrirostris) in Florida, USA. In: Reguera B, Blanco J, Fernandez ML, Wyatt T, editors. Harmful Algae. Intergovernmental Oceanographic Commission of UNESCO; 1998. pp. 97–100.
  • Lapointe BE, Bedford BJ. Harmful Algae. Red drift seaweed blooms emerge in Lee County, Florida, USA: evidence of escalating coastal eutrophication. in press.
  • Lenes JM, Darrow BP, Cattrall C, Heil CA, Callahan M, Vargo GA, Byrne RH. Iron fertilization and the Trichodesmium response on the West Florida shelf. Limnol. Oceanogr. 2001;46:1261–1277.
  • McPherson BF, Halley R. The South Florida Environment: A Region Under Stress, U.S. Geological Survey Circular; 1996. p. 1334.
  • Meyers SD, Siegel EM, Weisberg RH. Observations of currents on the West Florida Shelf break. Geophys. Res. Lett. 2001;28:2037–2040.
  • Miller RL, Kraemer TF, McPherson BF. Radium and radon in Charlotte Harbor estuary, Florida. Est. Coast. Shelf Sci. 1990;31:439–457.
  • Moore WS. The subterranean estuary: a reaction zone of ground water and seawater. Mar. Chem. 1999;65:111–125.
  • Olascoaga MJ, Rypina II, Brown MG, Beron-Vera FJ, Kocak H, Brand LE, Halliwell GR, Shay LK. Persistent transport barrier on the West Florida Shelf. Geophys. Res. Lett. in press.
  • Parsons ML, Dortch Q, Turner RE. Sedimentological evidence of an increase in Pseudo-nitzschia (Bacillariophyceae) abundance in response to coastal eutrophication. Limnol. Oceanogr. 2002;47:551–558.
  • Paytan A, Shellenbarger GG, Street JH, Davis K, Young MB, Moore WS. Submarine groundwater discharge: An important source of new inorganic nitrogen to coal reef ecosystems. Limnol. Oceanogr. 2006;51:343–348.
  • Prospero JM, Lamb PJ. African droughts and dust transport to the Caribbean: climate change implications. Science. 2003;302:1024–1027. [PubMed]
  • Rayner NA, Parker DE, Horton EB, Polland CK, Alexander LV, Rowell DP, Kent EC, Kaplan A. Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res. 2003;108(D14):4407. doi:10.1029/2002JD002670.
  • Scott TM, Means GH, Meegan RP, Means RC, Upchurch SB, Copeland RE, Jones J, Roberts T, Willet A. Springs of Florida. Florida Geol. Surv. Bull. 2006;66:1–347.
  • Sheppard C, Rioja-Nieto R. Sea surface temperature 1871-2099 in 38 cells in the Caribbean region. Mar. Environ. Res. 2005;60:389–396. [PubMed]
  • Simmons GM. Importance of submarine groundwater discharge (SGWD) and seawater cycling to material flux across sediment/water interfaces in marine environments. Mar. Ecol. Prog. Ser. 1992;84:173–184.
  • Slobodkin LB. A possible initial condition for red tides on the coast of Florida. J. Mar. Res. 1953;12:148–155.
  • Smayda TJ. Novel and nuisance phytoplankton blooms in the sea: evidence for a global epidemic. In: Graneli E, Sundstrom B, Edler L, Anderson DM, editors. Toxic Marine Phytoplankton. Elsevier Science; 1990. pp. 29–40.
  • Snyder GH, Davidson JM. Everglades agriculture: past, present, and future. In: Davis SM, Ogden JC, editors. Everglades: The Ecosystem and Its Restoration. St. Lucie Press; Delray Beach, Florida: 1994. pp. 85–115.
  • Steidinger KA, Joyce EA. Florida Dept. Nat. Res. St. Petersburg: 1973. Florida red tides; p. 26. (Educ. Ser. No. 17).
  • Steidinger KA, Vargo GA. Marine dinoflagellate blooms: dynamics and impacts. In: Lembi C, Waaland JR, editors. Algae and Human Affairs. Cambridge University Press; 1988. pp. 373–401.
  • Steidinger KA, Vargo GA, Tester PA, Tomas CR. Bloom dynamics and physiology of Gymnodinium breve with emphasis on the Gulf of Mexico. In: Anderson DM, Cembella AD, Hallegraeff GM, editors. Physiological Ecology of Harmful Algal Blooms. Springer-Verlag; Berlin: 1998. pp. 133–153.
  • Steinman AD, Havens KE, Carrick HJ, VanZee R. The past, present, and future hydrology and ecology of Lake Okeechobee and its watersheds. In: Porter JW, Porter KG, editors. The Everglades, Florida Bay, and Coral Reefs of the Florida Keys, An Ecosystem Sourcebook. CRC Press; 2002. pp. 19–37.
  • Stenseth NC, Mysterud A, Ottersen G, Hurrell JW, Chan K-S, Lima M. Ecological effects of climate fluctuations. Science. 2002;297:1292–1296. [PubMed]
  • Tester PA, Steidinger KA. Gymnodinium breve red tide blooms: Initiation, transport, and consequences of surface circulation. Limnol. Oceanogr. 1997;42:1039–1051.
  • Tester PA, Stumpf RP, Steidinger KA. Ocean color imagery: what is the minimum detection level for Gymnodinium breve blooms. In: Reguera B, Blanco J, Fernandez ML, Wyatt T, editors. Harmful Algae. Intergovernmental Oceanographic Commission of UNESCO; 1998. pp. 149–151.
  • Tomlinson MC, Stumpf RP, Ransibrahmanakul V, Truby EW, Kirkpatrick GJ, Pederson BA, Vargo GA, Heil CA. Remote Sensing Environ. Vol. 91. 2004. Evaluation of the use of SeaWiFS imagery for detecting Karenia brevis harmful algal blooms in the eastern Gulf of Mexico; pp. 293–303. 2004.
  • Trenberth K. Uncertainty in hurricanes and global warming. Science. 2005;308:1753–1754. [PubMed]
  • Turner RE, Rabalais NN, Fry B, Atilla N, Milan CS, Lee JM, Normandeau C, Oswald TA, Swenson EM, Tomasko DA. Paleo-indicators and water quality change in the Charlotte Harbor estuary (Florida) Limnol. Oceanogr. 2006;51:518–533.
  • Vargo GA, Heil CA, Ault DN, Neely MB, Murasko S, Havens J, Lester KM, Dixon LK, Merkt R, Walsh J, Weisberg R, Steidinger KA. Harmful Algae 2002. Florida Fish and Wildlife Conservation Commission, Florida Institute of Oceanography and Intergovernmental Oceanographic Commission of UNESCO; 2004. Four Karenia brevis blooms: A comparative analysis; pp. 14–16.
  • Walsh JJ, Steidinger KA. Saharan dust and Florida red tides: The cyanophyte connection. J. Geophys. Res. 2001;106(C6):11597–11612.
  • Walsh JJ, Penta B, Dieterle DA, Bissett WP. Predictive ecological modeling of harmful algal blooms. Human Ecol. Risk Assess. 2001;7:1369–1383.
  • Webb JG. The mortality of fish in the Gulf of Mexico. Bull. U.S. Fish. Comm. 1886;6:11–13.
  • Weisberg RH, Black BD, Yang H. Seasonal modulation of the west Florida continental shelf circulation. Geophys. Res. Lett. 1996;23:2247–2250.
  • Yang H, Weisberg RH. Response of the West Florida Shelf circulation to climatological wind stress forcing. J. Geophys. Res. 1999;104:5301–5320.
  • Yang H, Weisberg RH, Niiler PP, Sturges W, Johnson W. Lagrangian circulation and forbidden zone on the West Florida Shelf. Cont. Shelf Res. 1999;19:1221–1245.