lable at ScienceDirect

Estuarine, Coastal and Shelf Science 89 (2010) 62e72

Contents lists avai

Estuarine, Coastal and Shelf Science

journal homepage: www.elsevier .com/locate/ecss

Short-term variability of suspended sediment and phytoplankton inTampa Bay, Florida: Observations from a coastal oceanographic towerand ocean color satellites

Zhiqiang Chen 1,*, Chuanmin Hu, Frank E. Muller-Karger, Mark E. LutherCollege of Marine Science, 140 Seventh Avenue South, Saint Petersburg, Florida, 33701, USA

a r t i c l e i n f o

Article history:Received 16 February 2009Accepted 25 May 2010Available online 2 June 2010

Keywords:short-term variabilityphytoplanktonsedimentremote sensingbio-optic sensorsU.S.A.FloridaTampa Bay

* Corresponding author.E-mail address: zchen@marine.usf.edu (Z. Chen).

1 Present address: South Florida Water ManagemRoad, West Palm Beach, FL 33406, U.S.A.

0272-7714/$ e see front matter � 2010 Elsevier Ltd.doi:10.1016/j.ecss.2010.05.014

a b s t r a c t

We examined short-term phytoplankton and sediment dynamics in Tampa Bay with data collectedbetween 8 December 2004 and 17 January 2005 from optical, oceanographic, and meteorological sensorsmounted on a coastal oceanographic tower and from satellite remote sensing. Baseline phytoplankton(chlorophyll-a, Chl) and sediment concentrations (particle backscattering coefficient at 532 nm, bbp(532)) were of the order of 3.7 mg m�3 and 0.07 m�1, respectively, during the study period. Both showedlarge fluctuations dominated by semidiurnal and diurnal frequencies associated with tidal forcing. Threestrong wind events (hourly averaged wind speed >8.0 m s�1) generated critical bottom shear stress of>0.2 Pa and suspended bottom sediments that were clearly observed in concurrent MODIS satelliteimagery. In addition, strong tidal current or swells could also suspend sediments in the lower Bay.Sediments remained suspended in the water column for 2e3 days after the wind events. Moderate Chlincreases were observed after sediment resuspension with a lag time of e1e2 days, probably due torelease of bottom nutrients and optimal light conditions associated with sediment resuspension andsettling. Two large increases in Chl with one Chl > 12.0 mg m�3 overe2 days, were observed at neap tides.For the study site and period, because of the high temporal variability in phytoplankton and sedimentconcentrations, a monthly snapshot can be different by �50% to 200% from the monthly “mean” chlo-rophyll and sediment conditions. The combination of high-frequency observations from automatedsensors and synoptic satellite imagery, when available, is an excellent complement to limited fieldsurveys to study and monitor water quality parameters in estuarine environments.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Estuaries are highly dynamic environments where rivers, winds,and tides interact to determine physical, chemical, and biologicalvariability. These driving factors lead to a wide range of temporalscales of phytoplankton and sediment variability in estuaries(Cloern et al., 1989; Cloern, 1991; Harding, 1994; Li and Smayda,2001; Roegner et al., 2002). Superimposed on “periodic” varia-tions due to diel variability, tides, and seasonal cycles are aperiodicor episodic (short-term) meteorological events. Wind pulsesmodify estuarine circulations and water levels and generate wavesand currents that suspend sediments (e.g., Schoellhamer, 1995) andmix nutrients and benthic algae into overlying waters (Lawrenceet al., 2004; Yeager et al., 2005).

ent District, 3301 Gun Club

All rights reserved.

Short-term variability in sediment has been well documentedusing optical and/or acoustic sensors (Schoellhamer, 1995; Jing andRidd, 1996; Li and Amos, 2001). However, one of the main obstaclesto studying short-term variability of phytoplankton in estuaries hasbeen lack of reliable and cost-effective means for high-frequencysampling (e.g., Roegner et al., 2002). As a result, there are very fewattempts to examine the relationship between sediment andphytoplankton dynamics, while this interaction is known to becritical in some estuaries (May et al., 2003; Desmit et al., 2005).

Tampa Bay is the largest estuary in Florida with a surface area ofe1000 km2. The average depth of the Bay ise4.0 m, with a dredgedchannel (>10 m) extending from the mouth of the bay to the upperbay. Tampa Bay is often divided into 4 segments based on geo-morphologic differences and salinity regimes (Lewis andWhitman,1985), namely Old Tampa Bay (OTB), Hillsborough Bay (HB), MiddleTampa Bay (MTB) and Lower Tampa Bay (LTB) (Fig. 1). An effort tocontrol severe eutrophication problems in the 20th century led toa systematic water quality monitoring program across the Bay(Boler et al., 1991). The monitoring program conducts field surveys

Fig. 1. Satellite image of Tampa Bay collected on 20 December 2004 with ModerateResolution Imaging Spectroradiometer (MODIS/Aqua). Four bay segments are outlinedwith dashed lines: Old Tampa Bay (OTB), Hillsborough Bay (HB), Middle Tampa Bay(MTB), and Lower Tampa Bay (LTB). Data collection locations are annotated: Towerstation (star) at Manatee Channel collected meteorological, oceanographic, and opticaldata; SunshineSkyway Bridge station (triangle) collected Acoustic Doppler CurrentProfile (ADCP) current data. The EPCHCTampa Bay water quality monitoring station(square) is the site of a 7-year (1997e2003) monthly time series of chlorophyll(mg m�3) and turbidity (NTU) data. The inset showing the location of Tampa Bay in theState of Florida.

Z. Chen et al. / Estuarine, Coastal and Shelf Science 89 (2010) 62e72 63

only once per month, which is customary for water quality surveysin other estuaries. These monthly data have been widely used tocharacterize long-term changes in water quality (Janicki et al.,2001; Schmidt and Luther, 2002). However, there has been littleinformation available on short-term variability of phytoplanktonand sediment concentrations in the Bay.

In this paper we examine the short-term variability of phyto-plankton and sediment in Tampa Bay using data collected fromautomated sensors mounted on a coastal oceanographic tower andfrom satellites. Our objectives were: (1) to characterize short-termpatterns of variations in phytoplankton and sediment in TampaBay; (2) to understand underlying mechanisms responsible for theobserved variability; and (3) to discuss implications of short-termvariability on water quality assessments for Tampa Bay and otherestuaries. We found that high-frequency in situ data and synopticsatellite remotely sensed data are important complements, ifavailable, to field surveys that are limited in scope due to personneland other costs.

2. Materials and methods

2.1. In situ automated sensors

An array of automated bio-optical, hydrographic and meteoro-logical sensors was deployed at a coastal oceanographic tower

station in Tampa Bay, Florida, to measure physical conditions andphytoplankton and sediment concentrations from 8 December2004 to 17 January 2005. The stationwas located near themiddle ofthe Bay (27. 661�N, 82.583�W, Fig. 1) at a bottom depth of 4.6 m.

Two types of optical sensors were deployed. One was theWETLabs� ECO-BBSB sensor to measure particle backscatteringcoefficient at 532 nm (bbp(532), which is a proxy for suspendedsediment concentration (e.g., D’Sa et al., 2007). The other was theWETLabs� ECO-FLNTUSB sensor to measure in vivo chlorophyll-a fluorescence near 683 nm, used as a proxy for chlorophyll-a concentration (Chl). Both sensors had an internal battery forcontinuous data logging. To determine possible vertical differencesin phytoplankton and sediment between the surface and bottomlayers, one set of sensors (i.e., backscattering sensor and fluorom-eter) was installed at 1 m depth below the surface with sensorsfacing down to minimize possible interference from sunlight. Theotherwas installed at 1.5m above the bottomwith sensors facing upto minimize possible interference from bottom reflectance.Sampling frequencywas set to1h�1. The recorded rawdata (voltage)were processed with the WETLabs� ECOView software and con-verted to bbp(532) (m�1) and chlorophyll-a concentration (Chl,mg m�3) using calibration coefficients provided by the manufac-turer. Those “default” Chl data were further calibrated using chlo-rophyll-a concentrations determined from discrete bottle samplescollected concurrently near the tower station and analyzedwith thestandard fluorometric method (Strickland and Parsons, 1972).

Although Chl derived from in vivo fluorescence may not be anaccurate proxy of biomass for every case due to environmentalinfluences on the physiology and taxonomic status of the phyto-plankton assemblages, in practice numerous publications showexcellent correlation between the two for various environmentsincluding estuaries (Li and Smayda, 2001; Roegner et al., 2002). Wehave shown that in Tampa Bay there is a good correlation betweenin vivo fluorescence and Chl for Chl between 10 and 30 mgm�3 (Huet al., 2004). Newer results from a cruise survey in April 2008further confirmed this finding for low Chl (1.3e2.5 mg m�3, seesupplemental materials). However, a close examination of the invivo fluorescence found that under high solar irradiation aroundnoon, fluorescence was affected by non-photochemical quenchingeffects (see results below). We therefore used the method ofMorrison (2003, equation 18 and derived parameters) to model thefluorescence efficiency (or quantum yield of fluorescence) asa function of photosynthetically active radiation (PAR). A PARthreshold was determined by visual comparison of Chl and PARtime series, with the latter corrected for light attenuation betweensurface and 1 m using a PAR attenuation coefficient of 0.8 m�1 (ourunpublished data). The estimated PAR threshold was e100 mmolphotons m�2 s�l, close to those reported elsewhere (Marra, 1997;Morrison, 2003). Then, we used three non-photochemicalquenching values, namely 0.1, 0.5 and 0.9, to represent minimum,medium, and maximum non-photochemical effects, respectively,to obtain the “true” Chl values.

Salinity, temperature, and water pressure were measured witha Sea Gauge (Sea Bird Electronics) sensor deployed about 2.2 mabove the sea floor. Wave height and period were computed usingthe standard Sea bird Electronics wave and tide recorder software,which uses linear wave theory and the standard surface gravitywave dispersion relation. Several above-water sensors measuredwind speed, barometric pressure, and solar PAR. Water level wasestimated from water pressure and barometric pressure Detaileddescriptions of sampling of these data were given by Sopkin et al.(2007). The meteorological and oceanographic data were nor-mally sampled at higher frequency (once per 6 or 15 min) andbinned into hourly averages to facilitate comparisons with theoptical data.

Z. Chen et al. / Estuarine, Coastal and Shelf Science 89 (2010) 62e7264

2.2. Ancillary data

Near-surface current data from the Sunshine Skyway Bridgestation (27.620�N, 82.6558�W, e8.5 km away from our samplingstation, Fig. 1) were used to represent general current conditions atthe tower station, where no current data was available during thestudy period. These data were combined with wind data to esti-mate vertical diffusivity (units of m2 s�1) to represent water columnstability (James, 1977). While currents in Tampa Bay vary withlocation and bathymetry (Weisberg and Zheng, 2006), we assumedthat currents at the Sunshine Skyway Bridge were qualitativelyrepresentative of those at the optical instrumentation towerbecause the two stations were relatively close and shared similargeomorphology. Indeed, water level at the tower stationwas highlycorrelated with currents observed at the bridge station, suggestingcoherence of the current data at these two stations.

Monthly chlorophyll and turbidity (Nephelometric TurbidityUnit, NTU, another common proxy for sediment concentration)observations between 1997 and 2003 were collected by the Envi-ronmental Protection Commission of Hillsborough County (EPCHC)at a water quality station located at 27.666�N, 82.5992�W, ore1.7 kmaway from the tower station (Fig.1). These datawere comparedwiththe sensor data and helped evaluate the sensitivity of a monthlyobservationprogram in long-termanalyses ofwater quality changes.

2.3. Satellite data

We previously described assessments based on satellite-derivedturbidity and water clarity for Tampa Bay (see Chen et al., 2007a,b).Here, we used remote sensing reflectance at 645 nm (Rrs(645), sr�1)as a proxy for suspended sediment concentration (Hu et al., 2004).Imagery from the Moderate Resolution Imaging Spectroradiometer(MODIS) 250-m channels from both Terra (overpass at local time ofe10:30 am) and Aqua satellites (overpass at local time ofe1:30 pm)were used to increase the spatial and temporal coverage (Hu andMuller-Karger, 2003). This combination also made it possible tostudy sediment variability within a tidal cycle (Doxaran et al., 2009).Similarly, MODIS/Aqua and the Sea-viewing Wide Field-of-viewSensor (SeaWiFS) 1-km data were processed using a semi-analyticalalgorithm (Lee et al., 2005; Chen et al., 2007b) to obtain light atten-uation coefficients, (Kd, m�1), at 488 nm and 490 nm, respectively.Reliable estimates of satellite-derived Chl in the Bay were, however,not derived because of interference by colored dissolved organicmatter and/or bottom reflectance (Chen et al., 2007c). MODIS/Terradata were not used to estimate Kd due to several problems with theinstrument, particularly in the blue bands (Franz et al., 2006).

2.4. Data analysis

A spectral power analysis was applied to the hourly data of Chland backscattering coefficient to determine dominant frequencies.All hourly time series datawere then low-pass filtered with a cutofftime of 36 h to extract subtidal variability. To objectively analyzethe subtidal variations and identify event-related anomalies, therobust data analysis method proposed by Hoaglin et al. (2000) wasused to estimatemeans and standard deviations of the filtered data.The Hoaglin et al. (2000) statistical analysis is less sensitive tooutliers caused by anomalous events and thus provides betterestimates of means and variations under normal conditions. Thestatistical analysis showed that the mean or baseline Chl was about3.7 mg m�3. This was very close to the histogram mode value(e3.5 mg m�3) of the Chl time series. Thresholds were then calcu-lated as three times the standard deviations above the “mean”conditions (i.e., 99% confidence level) of the filtered data. Anyvalues greater than the thresholds were considered to be

anomalies. Consecutive anomalies constituted an event. Theanomalies and the duration of an event depend on the chosenthresholds, and we used a 99% confidence level to identify them.

Maximum wave bottom shear stress was estimated using themodel proposed by Li and Amos (2001). In principle, bottom shearstress (sb, Pa) includes effects from waves, currents, and interac-tions between waves and currents near the bottom boundary layer.Due to a lack of current data near the bottom boundary layer, onlywave-induced bottom shear stress (sw) was estimated. We usedlinear wave theory as follows:

L ¼ g � T2

2� ptanðk� hÞ (1)

Ub ¼ p� HðT � sinðk� hÞÞ (2)

Ab ¼ Ub � 2� pT

(3)

sw ¼ 0:5� r� fw � U2b (4)

fw ¼ exp

5:213�

�KbAb

�0:194

�5:997

!;

AbKb

> 1:7 (5)

fw ¼ 0:28;AbKb

� 1:7 (6)

where L, T, H, h, k are wave length (m), wave period (s), wave height(w0.707 times of wave significant height, m), water depth (m), andwave number (2 � p/L, m�1), respectively; Ub, Ab, fw, r are waveorbital velocity (m s�1), near-bed wave orbital amplitude (m), wavefriction factor (dimensionless), and seawater density (kg m�3),respectively. The bottom roughness height (Kb) can be assumeda constant within a limited geographic region (Peter Howd/USGS,pers. Comm.), and was taken as 0.003 m as obtained in Old TampaBay by Schoellhamer (1995).

3. Results

3.1. Physical conditions

We observed five distinct wind events: 11, 14, 19, 26 December2004, and 13 January 2005 (Figs. 2A and 3C). During these events,wind speeds exceeded 8.0 m s�1 fore2e3 days and wind directiontypically changed from southerly/southeasterly (from the south orsoutheast toward the north or northwest) to northeasterly/north-westerly. This transition is common in the region as cold frontspass, coming from the north or northwest (Wang, 1998).

Water levels showedmixed semidiurnal and diurnal oscillationswith maxima of about 1.0 m (Fig. 2B). In addition, the neap (e.g.,about 17 December 2004) and spring (e.g., about 10 January 2005)tides were detected. Current speeds ranged from 0.1 to 1.2 m s�1.Under similar wind conditions (<3.0 m s�1), a daily average ofhourly sampled current speed on 10 January 2005 (a spring tide)was about 0.60 m s�1, about twice of that on 17 December 2004 (aneap tide). Water motions within a tidal cycle followed the direc-tion of the navigation channel located nearby the Sunshine SkywayBridge station, which is oriented primarily along 62� or 242� true(Fig. 2C). Salinity ranged from 26 to 31 and showed similar fluc-tuations as water level (r ¼ 0.72, n ¼ 961), with higher salinityoccurring during flood tides (Fig. 2D). Water temperature rangedfrom 14 to 22 �C and mirrored the passages of cold atmosphericfronts (Fig. 2E) with cold fronts depressing water temperature.

08 12 16 20 24 28 01 05 09 13 1712.0

14.0

16.0

18.0

20.0

22.0

Tem

pera

ture

(ο C)

temperature

filter temperature

24.0

26.0

28.0

30.0

32.0

Salin

ity

salinity

filter salinity

−0.6

−0.3

0.0

0.3

0.6

0.9

Cur

rent

(m

s−1

)

−0.05

0.00

0.50

1.00

1.50

Wat

er le

vel (

m) water level

filter water level

−10

0

10

Win

d (m

s−1

)

December 2004 January 2005

A

B

C

D

E

spring tide neap tide

Deep channel orientation 62°

Fig. 2. Hourly time series of (A) winds (m s�1); (B) water levels (m); (C) currents (m s�1); (D) salinity; and (E) temperature (�C). Thick red lines show subtidal variations, representedas low-pass filtered data with a cutoff time of 36 h, and the results are shown as thick lines. The arrow in panel C shows the deep channel orientation, which mainly controls thecurrent direction. The time shown here and in the following figures is local time (the U.S. Eastern Standard Time), starting from mid-night.

Z. Chen et al. / Estuarine, Coastal and Shelf Science 89 (2010) 62e72 65

Solar irradiance showed typical diurnal cycles modulated bycloud cover (Fig. 3A). Wave periods varied from 1 to 9 s with mostperiods <2 s (Fig. 3B). Such short waves are consistent with thelimited size of the Bay and the short wind fetch. However, therewere occasions when waves of longer periods (>4 s) wereobserved. These appeared to be independent of wind speed (e.g.,5e10 January 2005). A likely cause for the longer waves is swellspropagating into the Bay from the West Florida Shelf (WFS).Significant wave height was generally <1 m, with 5 peaks in therecord (Fig. 3C). The wave heights were closely correlated withwinds (r ¼ 0.85, n ¼ 961), suggesting that wave height was deter-mined primarily by local winds. Wave bottom orbital velocity andbottom shear stress exhibited similar patterns as wave height andwind speed, except for some moderate peaks between 5 and 10January 2005 when wind speed was low. Peaks on these datescoincided with the longer wave periods (Fig. 3B), suggesting thatbottom shear stress can be influenced by other factors than localwind, such as swells propagating from the adjacent shelf.

3.2. Suspended sediments

The two bbp(532) sensors showed nearly identical patterns ofvariability (r ¼ 0.88, n ¼ 961), with values near the surface being onaverage e20% lower than those at the bottom sensor(0.072 � 0.040 m�1 versus 0.092 � 0.044 m�1). Because of the simi-larity in variability between the two records, and since the surface

values are the most relevant for validation of the satellite imagery, inthis study we focused only on the surface values.

The near-surface bbp(532) record showed high variability andranged from 0.03 to 0.20 m�1 with a baseline value of about0.07m�1 (Fig. 4A). Spectral analysis showed that the highest energyof variation was associated with the diurnal frequency, similar tothe water level spectrum (Fig. 5A and B), showing that high-frequency sediment variation was associated with tides. High bbp(532) typically occurred at low water level within a tidal cycle(therefore low salinity waters), particularly during ebb or earlyflood tides, and sometimes displayed a double-peak feature(Fig. 4C). In addition, resuspended sediments on the WFS may alsobe transported into the Bay by tidal advection, as revealed byfollowing remote sensing imagery (Figs. 6 and 7).

The results of the robust statistical analysis showed that therewere three prominent sediment resuspension events, specificallyon 8e16, 25e30 December 2004, and 13e17 January 2005 (Fig. 4A).There was also a moderate and transient sediment resuspensionevent around 9 January 2005. The three prominent events werecoincident with stronger winds and larger wave heights. Sedimentresuspension typically occurred at wind speeds >8.0 m s�1 andbottom shear stress >0.2 Pa (Fig. 3E). The threshold of 0.2 Pa islikely a minimum estimate because current-induced bottom shearstress was not included. The suspended sediments typicallyremained in the water column fore2e3 days after a wind event.Wind events alone, however, did not account for all sediment

December, 2004 January, 2005 08 12 16 20 24 28 01 05 09 13 17

0.2

1.0

3.0

Wav

e st

ress

(N

m−2

)

0.0

0.2

0.4

0.6

Wav

e ve

loci

ty (

m s−1

) 0.0

1.0

2.0

3.0

Wav

e he

ight

(m

)

wave height

filter wave height

0.0

2.0

4.0

6.0

8.0

10.0

Wav

e pe

riod

(s) wave period

filter wave period

0

200

400

600

800

PAR

(W

m−2

)

A

B

C

D

E

Fig. 3. Hourly time series of (A) solar irradiation, PAR (W m�2); (B) significant wave period (s); (C) significant wave height (m); (D) near-bottom wave orbital velocity (m s�1); and(E) near-bottomwave shear stress (Pa). To show the subtidal variations, significant wave height and wave period were low-pass filtered with a cutoff time of 36 h, and the results areshown as thick lines in B and C. The black dashed line in panel E shows the estimated critical shear stress of w0.2 Pa, above which sediment resuspension may occur.

Z. Chen et al. / Estuarine, Coastal and Shelf Science 89 (2010) 62e7266

resuspension, and sediment resuspension was not necessarilyassociated with a wind event. During some wind events, such as on19 December 2004, we did not observe any sediment resuspension.In contrast, the moderate resuspension on 9 January 2005 coin-cided with low wind but longer wave periods (Fig. 3B) and largerbottom shear stress (Fig. 3E). Thus, this resuspension event waslikely not caused by local wind but by swells and/or strong currents(>1.0 m s�1 coincident with a spring tide).

TheMODIS images (Figs. 6 and 7) clearly revealed the spatial andtemporal patterns of sediment variation and demonstrated howsynoptic satellite images complement high-frequency opticalobservations at a fixed station to enable the study of sediment

dynamics. MODIS Rrs(645) values were strongly correlated with insitu bbp(532) (Fig. 5A, r ¼ 0.72, n ¼ 7 for Aqua; r ¼ 0.82, n ¼ 12 forTerra). Fig. 6 shows the sediment variationwithin a tidal cycle, whensediment concentration increased from a low tide to a flood tide,primarily due to tidal advection. The water level was low in themorning (e10:46 am) when MODIS/Terra showed relatively lowsediment concentration in the middle bay. After about 3 h(13:55 pm) when the water level was higher, MODIS/Aqua showedincreased sediment concentration in the middle bay, followinga flood tide.

Fig. 7 shows overall sediment concentrations under normalconditions and wind events. The images show that high

0.000.050.100.150.200.25

bbp(

532)

(m

−1 )

0.000.200.400.600.801.00

Rrs

(645

)(sr

−1 )Aqua Rrs(645)

Terra Rrs(645)

08 12 16 20 24 28 01 05 09 13 170.0

5.0

10.0

15.0

Chl

(m

g m

−3 )

0.00.050.100.150.200.25

bbp(

532)

(m

−1 )

−0.50

0.00

0.50

1.00

wat

er le

vel(

m)

bbp(532)

water level

13 14 15 16 17 18 190.0

5.0

10.0

15.0

Chl

(mg

m−

3 )

0.0

200

400

600

800

PAR

(W

m−

2 )

Chl

Corrected Chl

PAR

B

D

A

C

Fig. 4. Time series of hourly averages of near-surface (A) surface backscattering coefficient (bbp(532), m�1); and (B) surface Chl (mgm�3). Overlaid in Panel A areRrs(645) (�10�2, sr�1)estimated fromMODIS/Aqua (circles) andMODIS/Terra (diamonds) using themethod of Chen et al. (2007a). The horizontal dashed lines in (A) and (B), respectively, represent estimatedbaseline values of 0.07m�1 for backscattering coefficient and 3.7mgm�3 for Chl using a robust statistical analysis. Low-pass filtered bbp(532)and Chl (cutoff time of 36 h) are shown asthe solid thick lines to illustrate their subtidal variations. The time series in the squaresbetweenDec.13e19areenlarged in (C) and (D),withwater levels (also indicating salinity changes)and PAR added. The circle in (C) highlights the double-peak feature of sediments associatedwith tides, while the circle in (D) highlights the Chl decrease for high PAR (solid circles) andcorrected Chl (open circles) using non-photochemical quenching value of 0.5. The horizontal bars in Panel A and B show the general periods of events. Please note that themarked datesstart from mid-night, thus the higher Chl occurred at day time.

0 0.1 0.2 0.3 0.4 0.50

50

100

Frequency (h−1)

Quenching parameter 0.1Quenching parameter 0.5Quenching parameter 0.9

0

10

20Chl

0

0.05

0.1

Spec

tral d

ensi

ty bbp(532)

0

10

20

30Water levels

D

C

A

B

Fig. 5. Spectral density of time series of (A) water levels, (B) bbp(532), (C) Chl, and(D) Chl corrected for different non-photochemical quenching effects (see text for moredetails). Both diurnal (w0.04 h�1) and semidiurnal (w0.08 h�1) frequencies are foundin all spectra, while sediment shows fewer coherencies with tides as compared tochlorophyll.

Z. Chen et al. / Estuarine, Coastal and Shelf Science 89 (2010) 62e72 67

concentrations were more prevalent in the shelf and the lowerTampa Bay than in the upper bay. This can be explained by thestrongerwind-inducedwave action in the openwaters (the shallowshelf) than in the semi-closed waters (the upper bay). The imagesalso help interpret the observations that could not be accounted forby sensor data only. For example, no marked increases in sedimentconcentrationswere observed from the sensor data on 20Decemberwhen there was a strong wind event. However, satellite imagesclearly showed that during the same time sediment resuspensionindeed occurred, but it was mostly observed only outside the Bay(Fig. 7C). Likewise, Fig. 7G shows that on 9 January, most of the Bayshowed lower sediment concentration, consistent with the lowwind speed, except in a small patch of moderately high sedimentconcentration near the optical sensors tower.

3.3. Phytoplankton

The chlorophyll values near the bottom at the tower stationgenerally showed similar variability as those near the surface aswell. Because of the similarity and the lack of bottom chlorophylldata after 7 days (the bottom fluorometer failed after about 7 daysof operations), we used only the surface Chl observations in thisstudy.

Chlorophyll ranged from 2.0 to 12 mg m�3 and followed similarhigh-frequency variations as observed inwater level (Figs. 4B,D and5C). Higher Chl usually occurred at lower salinity water or duringearly flood tides. Chl was also generally higher during daytime,particularly during late afternoon. For example, average Chl at 5 pmlocal time (5.1 mg m�3) was statistically higher than at mid-night(4.0 mg m�3) (p < 0.05,n ¼ 40). The increase in Chl during the daywas especially evident in the series corrected for fluorescencequantum yield (Fig. 4D).

Fig. 6. Remote sensing reflectance at 645 nm (Rrs(645), �10�2) (sr�1) from (A) MODIS/Terra (10:46, low tide) and (B) MODIS/Aqua (13:55, flood tide) on 27 December 2004. (C) bbp(532) (m�1) and water level (m) within one tidal cycle on 27 December 2004. White color represents shallow water (<2.8 m, Chen et al., 2007a), black color represents Rrs(645)>0.02 sr�1 due to the bright bottom, grey color represents the land. The location of the in situ station is marked as a white cross. The circles overlaid on the imagery highlight areasof contrasting of sediment concentration between the MODIS images.

Z. Chen et al. / Estuarine, Coastal and Shelf Science 89 (2010) 62e7268

The low-pass filtered Chl data and robust statistical analysisshowed three distinct episodes of Chl that were significantly higherthan the mean baseline condition (p < 0.01) for 16e24 December2004, 28 December 2004 to 1 January 2005, and 13e17 January2005, respectively (Fig. 4B). The first two episodes followed thesediment resuspension events earlier, with a time lag of about 1e2days (Fig. 4A and B). The third episode followed a moderateresuspension event and coincided with the last resuspension event.The first event contained sharp increases in Chl (e.g., >12 mg m�3,nearly 3e4 times of the baseline Chl). These Chl increases are, inpart, attributed to tidal advection of a Chl gradient as describedearlier. However, historical monitoring data showed that Chl in themiddle and lower portions of Tampa Bay were rarely > 5.0 mg m�3

during a normal winter season. Fig. 8A shows that Chl increasesoccurred after sediment resuspension events with a lag time of 1e2days, when water clarity increased. Further, the Chl spike wascoincided with a neap tide when estimated diffusivity was ata smaller level with a lower vertical mixing rate (Fig. 8B). A similar

increase in surface Chl following a sediment resuspension eventand coincident with a neap tide (Fig. 9) was observed in anotherdeployment between 1 November and 13 December 2005 at thesame station.

The above results assumed that chlorophyll-a concentrations(Chl) could be accurately represented by calibrated in vivo fluo-rescence. Whether or not this is true depends on the chlorophyllfluorescence yield (e.g., the ratio of in vivo fluorescence to Chl),which further depends on environmental and physiological factors.Fig. 4D shows that around some solar noons (highest PAR), thecorresponding in vivo Chl fluorescence values were consistentlysuppressed, confirming the occurrence of non-photochemicalquenching in Tampa Bay during the study period. We correctedthese effects using the method described by Morrison (2003).Spectral analysis of the corrected data yielded nearly identicalfrequency patterns to the original results except slightly higherdiurnal and semidiurnal density (Fig. 5C and D). Further, daytimeChl near solar noon generally increased after this correction,

Fig. 7. Remote sensing reflectance at 645 nm (Rrs(645), �10�2) (sr�1) from MODIS/Terra and MODIS/Aqua between 8 December 2004 and 17 January 2005, showing the devel-opment and progression of suspended sediments inside and outside Tampa Bay. The location of the in situ station is marked as a cross.

Z. Chen et al. / Estuarine, Coastal and Shelf Science 89 (2010) 62e72 69

whereas Chl values changed little at night and other times of lowirradiance. Thus, the correction for fluorescence quench yield didnot change the association between sediment suspension/settlingand increases in phytoplankton and marked increases in Chlobserved on 17e18 December 2004.

4. Discussion

4.1. Variability of sediment and chlorophyll-a

Spectral analyses of time series of both sediment and Chlrevealed that tides were the primary forcing to control theobserved short-term variability in Tampa Bay. Similar tidal-driven Chl and sediment changes have been observed in otherestuaries (Cloern et al., 1989; Cloern, 1996; Li and Smayda, 2001;Roegner et al., 2002). These tides-associated sediment and Chlvariations (relatively high Chl and sediment during low tides andlower Chl and sediment during higher tides) are consistent withdistinctive spatial gradients of Chl and sediment from the upperto the lower Tampa Bay (Bendis, 1999). For sediment concen-trations, a double-peak feature is also attributable to the effectsof ebbing and flooding tides on the deposition and resuspension

of suspended sediments (e.g., Jing and Ridd, 1996). However, thisgeneral tidal variation pattern may change during events whenhigh sediment and Chl are observed outside the Bay (e.g., Fig. 6;Roegner et al., 2002).

The subtidal variation of sediment is primarily controlled by localwindswith a critical bottomshear stress ofe0.2 Pa (Figs. 4 and7). Thisobservation is consistent with those from other estuarine andcoastal waters (Lawrence et al., 2004) and earlier results fromTampa Bay (Schoellhamer, 1995), and is an illustration of strongbenthic-pelagic coupling in shallowwaters (Cloern,1996). However,our results showed that wind forcing alone could not account for allsediment resuspension events. The first case is the sedimentresuspension event associated with lowwind. Schoellhamer (1996)found that large vessels in a dredged ship channel could result insolitary wave-induced sediment resuspension. This mechanismmay be responsible for the moderate resuspension event around 9January 2005. Another possible reason of this resuspension event isthe strong current speed (>1.0 m s�1), which is significantly higherthan those observed (e0.2m s�1) by Schoellhamer (1995). The secondcase is the low sediment concentrationwith high wind at out studysite (Fig. 7C). As shown in the satellite images, significant sedimentresuspension did occur outside the bay on the shelf, yet such

08 12 16 20 24 28 01 05 09 13 170.0

2.0

4.0

6.0

Dif

fusi

vity

(m

2 s−

1 , x10

−3 )

0.0

2.0

4.0

6.0

8.0

Kd

(m−

1 )

SeaWiFS Kd(490)Aqua Kd(488)

A

B

Fig. 8. Time series of (A) light attenuation coefficients, Kd (m�1), from SeaWiFS (Kd(490), filled circles) and MODIS/Aqua (Kd(488), open circles) and (B) daily averagedvertical diffusivity (m2 s�1) estimated using the method of James (1977) fromwind andcurrent data. Vertical lines represent the times when increased Chl values wereobserved (Fig. 5B). The graphs illustrate that if a neap tide (low diffusivity) coincidedwith sediment settling (17e20 December), a condition resulting in replete nutrientsand favorable light, significant phytoplankton growth could occur.

Z. Chen et al. / Estuarine, Coastal and Shelf Science 89 (2010) 62e7270

resuspension was not propagated in the bay. Clearly, the resus-pension mechanisms are different even for these two adjacentwater bodies.

The 1e2 days temporal lag of Chl increase, following sedimentresuspension events, suggests a coupling between sedimentresuspension and phytoplankton growth, which were alsoobserved in other studies (Cloern, 1987; Lawrence et al., 2004;Stephen et al., 2004). For Tampa Bay, previous studies have sug-gested that phytoplankton biomass are limited by the nutrientavailability and primarily by nitrogen (Vargo et al., 1991; Johansson,

0.0

3.0

6.0

9.0

12.0

Fluo

resc

ence

(m

g m

−3 )

0.000.020.040.060.080.10

bbp

(532

) (m

−1 )

01 05 09 13 17 2−0.5

0.0

0.5

1.0

Wat

er le

vel (

m)

A

B

C

November 2005

Neap tides

Fig. 9. Time series of (A) uncorrected Chl (mg m�3), (B) backscattering coefficient at 532 nmDecember 2005. The highlighted area shows increased Chl coincident with a neap tide andoccurred around 12 November, but sediment resuspension did not occur. Correspondingly,

2000), and that sediment resuspension could be an importantnutrient source in the lower Bay (Carlson and Yarbro, 2006). This“newly available” nutrient can promote phytoplankton growth 1e2days after the settlement of the resuspended sediments andimprovement of water clarity. Aside from these mechanisms, whatother factors could contribute to the 3e4 Chl increases on 17e18December 2004? A close examination of all times series datasuggests a potential link to tidal stirring. Previous studies haveshown that tidal stirring is a major factor in driving phytoplanktonvariability in estuaries (Cloern, 1991; Monbet, 1992; Howarth et al.,2000; Wong et al., 2007). Small diffusivity during the neap tidesindicates that vertical mixing rates were lower than other periods.Together with increased water clarity, the decreased diffusivity islikely to stimulate surface phytoplankton growth throughincreasing the ratios of photic depth to mixed layer depth (Alpineand Cloern, 1988). A similar increase in surface Chl between 1November and 13 December 2005 at the study site further confirmsthis unique combination of physical forcing (sediment resus-pension and neap tides) in modulating phytoplankton growth. Thisobservation is consistent with the modeling results by May et al.(2003): phytoplankton variability may be related to not onlyphysical forcing, but also their relative timings.

4.2. Implications for water quality monitoring

The Environmental Protection Commission of HillsboroughCounty (EPCHC) water quality monitoring program performsmonthly surveys of Tampa Bay spanning periods ofe3 weeks, withone week dedicated to sampling a segment of Tampa Bay (Boleret al., 1991). From both sensor and satellite data, we observedsignificant short-term variations of sediment and chlorophyll,which is driven by both tides and winds. Below we address animplication of the high-frequency variations for analyses based onmonthly sampling.

Average conditions for a specificmonth are theoretically derivedby averaging all data collected over the month, which can be rep-resented by a single aggregated mean value (mm). If a single

fluorescence

bbp (532)

1 25 29 03 07 11

water level

Decemeber 2005

, and (C) water level (m) observed at the same tower station from 1 November to 13decreased sediment concentrations around 25 November. Note that a neap tide alsothere was no subsequent Chl increase.

Z. Chen et al. / Estuarine, Coastal and Shelf Science 89 (2010) 62e72 71

snapshot measurement (mi) is used to serve as a monthly mean asdone in most estuarine water quality assessments, then the relativeerror can be estimated as:

relative errorðREÞ ¼ mi�mmmm

� 100% (7)

The data between 8 December 2004 and 7 January 2005provide an opportunity to evaluate such errors at a single site. Theresults show that the potential errors introduced by using a singlesnapshot observation to represent the monthly mean for the studysite can range from �50% to 200% for both chlorophyll and sedi-ment concentrations. These errors are larger than, or at least arecomparable to, the seasonal and interannual variations in Chl andturbidity observed from the once-per-month EPCHC surveys(Fig. 10). Because our high-frequency data were collected inDecember and January, thus were particularly prone to cold frontpassages, whether or not the same conclusion is applicable to otherseasons requires further investigation. During the local hurricaneseason (1 Junee30 November), episodic high wind and rainfalloccur, inducing sediment resuspension and land-based runoff.These events create similar short-term variability (Schoellhamer,1995). Thus, for any single site, once-per-month observations maybe inadequate to accurately estimate seasonal or interannual vari-ability. However, the EPCHC data were never used to derive timeseries at a single location, but used to compute averages for eachbay segment. Fig. 10 shows that if multi-year monthly snapshotdata from a single site are averaged, the climatological seasonaltrends become apparent. Similarly, when data from multiple loca-tions from a bay segment are averaged, the data become less noisyand may be used to study long-term changes. However, theuncertainties from these relatively scarce measurements need to beassessed with more high-frequency data as obtained here.

The results show that it is useful to combine the monthly fieldsurveys with high-frequency observations from buoys and synopticremote sensing imagery to properly assess water quality variationsin Tampa Bay. The implementation of such an integrated system,however, is not straightforward. Continuous measurements ofwater quality parameters such as turbidity and Chl from mountedsensors are rarely available, and the accuracy of remote sensingproducts are often not satisfactory for shallow estuaries. However,recent advances in algorithm development for Tampa Bay (Chen

Fig. 10. Monthly time series of (A) turbidity (NTU) and (B) Chl (mg m�3) at one of theEPCHC stations (See Fig. 1 for station location) from 1997 to 2003. The thicker linesrepresent the 7-year averages.

et al., 2007a,b) make it possible to establish a remote sensingbased water quality monitoring system. We expect to carry out thistask in the near future.

5. Conclusion

Using high-frequency bio-optical, physical, and meteorologicalobservations collected from a Tampa Bay tower, we showedsignificant short-term variability of phytoplankton and sediments,driven primarily by tides, winds, and associated currents.Combined with synoptic satellite observations, the high-frequencyobservations can complement those from monthly surveys to helpbetter characterize patterns of variability and advance our under-standing of sediment and phytoplankton dynamics in highly vari-able estuaries.

Acknowledgements

This work was funded and logistically supported through theTampa Bay study of the U.S. Geological Survey (USGS), Coastal andMarine Geology Program. Additional financial support wasprovided by the University of South Florida (USF)-USGS Coopera-tive Graduate Assistantship Program, and the National Aeronauticsand Space Administration (NASA grants NNX09AT50G andNNX09AV56G and NNS04ab59G). The field data were collectedonboard the USGS R/V Mako under assistance of Jordan Stanford,Keith Ludwig (from USGS), Laural Wetzell, and Sherryl Gilbert(from USF). The Tampa Bay water quality data were graciouslyprovided by the Environmental Protection Commission of Hills-borough County. We thank David English for his assistance in col-lecting data during the April 2008 cruise. The critical commentsfrom three anonymous reviewers and the journal editor are greatlyappreciated.

Appendix. Supplementary data

Supplementary data associated with this article can be found inthe online version, at doi:10.1016/j.ecss.2010.05.014.

References

Alpine, A.E., Cloern, J.E., 1988. Phytoplankton growth rates in a light-limited envi-ronment, San Francisco Bay. Marine Ecology Progress Series 44, 167e173.

Bendis, B., 1999. Water quality trends in Tampa Bay, Florida. Master thesis.University of SouthFlorida.

Boler, R.N., Molloy, R.C., Lesnett, E.M., 1991. Surface water quality monitoring by theEnvironmental Protection Commission of Hillsborough country. In: Treat, S.F.,Clark, P.A. (Eds.), Proceedings from BASIS II (Tampa, FL).

Carlson, P., Yarbro, L., 2006. Benthic Oxygen and Nutrient Fluxes in Tampa Bay:a Final Report to the Tampa Bay Estuary Program. Florida, Department ofEnvironmental Protection and U.S. Environmental Protection Agency Gulf ofMexico Program. Tampa Bay Estuary Program, Technical Publication # 04e06.

Chen, Z., Hu, C., Muller-Karger, F.E., 2007a. Monitoring turbidity in Tampa Bay usingMODIS/Aqua 250-m imagery. Remote Sensing of Environment 109, 207e220.

Chen, Z., Muller-Karger, F.E., Hu, C., 2007b. Remote sensing of water clarity in TampaBay. Remote Sensing of Environment 109, 249e259.

Chen, Z., Hu, C., Conmy, R.N., Swarzenski, P., Muller-Karger, F.E., 2007c. Coloreddissolved organic matter in Tampa Bay, Florida. Marine Chemistry 104, 98e109.

Cloern, J.E., 1987. Turbidity as a control on phytoplankton biomass and productivityin estuaries. Continental Shelf Research 7, 1367e1381.

Cloern, J.E., 1991. Tidal stirring and phytoplankton bloom dynamics in an estuary.Journal of Marine Research 49, 203e221.

Cloern, J.E., 1996. Phytoplankton bloom dynamics in coastal ecosystems: a reviewwith some general lessons from sustained investigation of San Francisco Bay,California. Review of Geophysics 34, 127e168.

Cloern, J.E., Powell, T.M., Huzzey, L.M., 1989. Spatial and temporal variability inSouth San Francisco Bay (USA). II temporal changes in salinity, suspendedsediments and phytoplankton biomass and productivity over tidal time scales.Estuarine Coastal and Shelf Science 28, 599e613.

Desmit, X., Vanderborght, J.P., Regnier, P., Wollast, R., 2005. Control of phyto-plankton production by physical forcing in a strong tidal, well-mixed estuary.Biogeoscience 2, 205e218.

Z. Chen et al. / Estuarine, Coastal and Shelf Science 89 (2010) 62e7272

Doxaran, D., Froidefon, J., Castaing, P., Babin, M., 2009. Dynamics of the turbiditymaximum zone in a macrotidal estuary (the Gironde, France): observations fromfield and MODIS satellite data. Estuarine Coastal and Shelf Science 81, 321e332.

D’Sa, E.J., Miller, R.L., McKee, B.A., 2007. Suspended particulate matter dynamics incoastal waters from ocean color: application to the northern Gulf of Mexico.Geophysical Research Letter 34, L23611. doi:10.1029/2007GL031192.

Franz, B.A., Werdell, P.J., Meister, G., Kwiatkowska, E.J., Bailey, S.W., Ahmad, Z.,McClain, C.R., 2006. MODIS land bands for Ocean remote sensing Applications.Proc. Ocean Optics XVIII, 9e13. Montreal, Canada October 2006.

Harding Jr., L.W., 1994. Long-term trends in the distribution of phytoplankton inChesapeake Bay: roles of light, nutrients, and stream flow. Marine EcologyProgress Series 104, 267e291.

Hoaglin, D., Mosteller, F., Tukey, J.M., 2000. Understanding Robust and ExploratoryData Analysis. John Wiley & Sons, Inc, pp 470.

Howarth, R.W., Swaney, R., Butler, T.J., Marino, R., 2000. Climate control entrophi-cation in the Hudson River estuary. Ecosystem 3, 210e215.

Hu, C., Muller-Karger, F.E., 2003.MODISmonitors Florida’s ocean dispersal of the PineyPoint phosphate treated wastewater. The Earth Observer (NASA) 15 (6), 21e23.

Hu, C., Chen, Z., Clayton, T., Swarzenski, P., Brock, J., Muller-Karger, F.E., 2004.Assessment of estuarine water quality monitoring with MODIS medium-reso-lution bands: Initial results from Tampa Bay. FL. Remote Sensing of Environ-ment 93, 423e441.

James, I.D., 1977. A model of the annual cycle of temperature in a frontal region ofthe Celtic Sea. Estuarine Coastal Marine Science 5, 339e353.

Janicki, A., Pribble, R., Janicki, S., Winowitch, M., 2001. An Analysis of Long-termTrends in Tampa Bay Water Quality Tampa Bay Estuary Technical Report#04e01.

Jing, L., Ridd, P.V., 1996. Wave-current bottom shear stresses and sediment resus-pension in Cleveland Bay. Australia Coastal Engineering 29, 169e186.

Johansson, J.O.R., 2000. Historical overview of Tampa Bay water quality and seagrassissues and trends. In: Greening, H. (Ed.), Seagrass Management: It’s Not JustNutrients! Symposium Proceedings, pp. 22e24. August 2000, St. Petersburg, FL.

Lawrence, D., Dagg, M.J., Lou, H., Cumming, S.R., Ortner, P.B., Kelble, C., 2004. Windevents and benthic-pelagic coupling in a shallow subtropical bay in Florida.Marine Ecology Progress Series 266, 1e13.

Lee, Z.P., Du, K.P., Arnone, R., 2005. A model for the diffuse attenuation coefficientof downwelling irradiance. Journal Geophysical Research 110, C02016.doi:10.1029/2004JC002275.

Lewis, R.R., Whitman, R.L., 1985. A new geographic description of the boundariesand subdivisions of Tampa Bay. Pages 10e17 in Proceedings of the Tampa BayArea Scientific Information Symposium, Report Number 65, Florida Sea GrantCollege.

Li, M.Z., Amos, C.L., 2001. SEDTRANS96: the upgraded and better calibrated sedi-ment-transport model for continental shelves. Computational Geoscience 27,619e645.

Li, Y., Smayda, T.J., 2001. A chlorophyll time series for Narragansett Bay: assessmentof the potential effect on tidal phase on measurement. Estuaries 24, 328e336.

Marra, J., 1997. Analysis of diel variability in chlorophyll fluorescence. JournalMarine Research 55, 767e784.

May, C.L., Koseff, J.R., Lucas, L.V., Cloern, J.E., Schoellhamer, D.H., 2003. Effects ofspatial and temporal variability of turbidity on phytoplankton blooms. MarineEcology Progress Series 254, 111e128.

Monbet, Y., 1992. Control of phytoplankton biomass in estuaries: a comparativeanalysis of microtidal and macrotidal estuaries. Estuaries 15, 563e571.

Morrison, J.R., 2003. In situ determination of the quantum yield of phytoplanktonchlorophyll a fluorescence: a simple algorithm, observations, and a model.Limnology and Oceanography 48, 618e631.

Roegner, G.C., Hickey, B.M., Newton, J.A., Shank, A.L., Armstrong, D.A., 2002. Wind-induced plume and bloom intrusions into Willapa Bay, Washington. Limnologyand Oceanography 47, 1033e1042.

Schmidt, N., Luther, M.E., 2002. ENSO impacts on salinity in Tampa bay, Florida.Estuaries 25, 976e984.

Schoellhamer, D.H., 1995. Sediment resuspension mechanisms in old Tampa Bay,Florida. Estuarine Coastal and Shelf Science 40, 603e620.

Schoellhamer, D.H., 1996. Anthropogenic sediment resuspension mechanisms ina shallow microtidal estuary. Estuarine Coastal Shelf Science 43, 533e548.

Sopkin, K., Mizak, C., Gilbert, S., Subramanian, V., Luther, M., Poor, N., 2007.Modeling air/sea flux parameters in a coastal area: a comparative study ofresults from the TOGA COARE model and the NOAA Buoy Model. AtmosphericEnvironment 41, 4291e4303.

Stephen, E.D., Cable, J.E., Childers, D.L., Coronado-Malina, C., Day, J.W., Hittle, C.D.,Madden, C.J., Reyes, E., Rudnick, D., Sklar, F., 2004. Importance of storm eventsin controlling ecosystem structure and function in a Florida Gulf Coast Estuary.Journal of Coastal Research 20, 1198e1208.

Strickland, J.D.H., Parsons, T.R., 1972. A Practical Handbook of Sea Water Analysis.Fisheries Research Board of Canada.

Vargo, G.A., Richardson, W.R., Howard, D., Paul, J.H., 1991. Phytoplankton productionin Tampa bay and two tidal rivers: the Alafia and little Manatee. In: Treat, S.,Clark, P. (Eds.), Proceedings of Tampa Bay Area Scientific Information Sympo-sium 2 (Tampa, FL).

Wang, J.D., 1998. Subtidal flow patterns in western Florida Bay. Estuarine Coastaland Shelf Science 46, 901e905.

Weisberg, R.H., Zheng, L.Y., 2006. Circulation of Tampa Bay driven by buoyancy,tides, and winds, as simulated using a finite volume coastal ocean model.Journal of Geophysical Research 111, C01005. doi:10.1029/2005JC003067.

Wong, K.T.M., Lee, J.H.W., Hodgkiss, I.J., 2007. A simple model for forecast of coastalalgal blooms. Estuarine Coastal and Shelf Science 74, 175e196.

Yeager, C.L.J., Harding, L.W., Mallonee, M.E., 2005. Phytoplankton production,biomass and community structure following a summer nutrient pulse inChesapeake Bay. Aquatic Ecology 39, 135e149.