Harmful Algae HA 39 -25 merged.pdfMahkota, g 25200 Kuantan, Pahang, Malaysia Department of Botany,...

Sampling harmful benthic dinoflagellates: Comparison of artificialand natural substrate methods

Patricia A. Tester a,*, Steven R. Kibler a, William C. Holland a, Gires Usup b,Mark W. Vandersea a, Chui Pin Leaw c, Lim Po Teen d, Jacob Larsen e,Normawaty Mohammad-Noor f, Maria A. Faust g, R. Wayne Litaker a

a National Oceanic and Atmospheric Administration, National Ocean Service, National Centers for Coastal Ocean Science,

Center for Coastal Fisheries and Habitat Research, 101 Pivers Island Road, Beaufort, NC 28516, USAb Program Sains Laut, Pusat Pengajian Sains Sekitaran dan Sumber Alam, Fakulti Sains dan Teknologi, Universiti Kebangsaan Malaysia,

43600 Bangi Selangor, Malaysiac Institute of Biodiversity and Environmental Conservation, Universiti Malyasia Sarawak, 94300 Kota Samarahan, Sarawak, Malaysiad Aquatic Sciences Program, Faculty of Resource Science and Technology, Universiti Malaysia Sarawak, 94300 Kota Samarahan, Sarawak, Malaysiae IOC Science and Communication Centre on Harmful Algae, Department of Phycology and Mycology, Øster Farimagsgade 2D, DK-1353 Copenhagen K, Denmarkf Institute of Oceanography and Maritime Studies, Kulliyyah of Science, International Islamic University Malaysia, Jalan Sultan Ahmad Shah, Bandar Indera

Mahkota, 25200 Kuantan, Pahang, Malaysiag Department of Botany, National Museum of Natural History, Smithsonian Institution, Washington, DC 20560, USA

1. Introduction

The resurgence of interest in the biodiversity of harmful benthicdinoflagellates, most notably the ciguatera-associated genus

Gambierdiscus (Fraga et al., 2011; Nishimura et al., 2013; Testeret al., 2013) has been facilitated by recent advances in taxonomy(Litaker et al., 2009) and molecular detection and quantificationmethods (Murray et al., 2009; Penna et al., 2010; Nagahama et al.,2011; Perini et al., 2011; Accoroni et al., 2012; Pfannkuchen et al.,2012; Vandersea et al., 2012). However, before the full potential ofmolecular assays can be utilized, especially species-specificquantitative polymerase chain reaction assays (qPCR), problemsinherent to sampling benthic harmful algal bloom (BHAB)dinoflagellates need to be addressed (GEOHAB, 2012). As a group,BHAB dinoflagellate species co-occur globally in shallow, tropicaland subtropical environments where they are typically associatedwith benthic substrates. The most common substrates colonizedby BHAB dinoflagellates include macroalgae, algal turf, seagrasses,

Harmful Algae 39 (2014) 8–25

A R T I C L E I N F O

Article history:

Received 14 March 2014

Received in revised form 17 June 2014

Accepted 17 June 2014

Available online

Keywords:

Gambierdiscus

Ciguatera fish poisoning

Ostreopsis

Prorocentrum

Cell-based monitoring

A B S T R A C T

This study compared two collection methods for Gambierdiscus and other benthic harmful algal bloom

(BHAB) dinoflagellates, an artificial substrate method and the traditional macrophyte substrate method.

Specifically, we report the results of a series of field experiments in tropical environments designed to

address the correlation of benthic dinoflagellate abundance on artificial substrate and those on adjacent

macrophytes. The data indicated abundance of BHAB dinoflagellates associated with new, artificial

substrate was directly related to the overall abundance of BHAB cells on macrophytes in the surrounding

environment. There was no difference in sample variability among the natural and artificial substrates.

BHAB dinoflagellate abundance on artificial substrates reached equilibrium with the surrounding

population within 24 h. Calculating cell abundance normalized to surface area of artificial substrate,

rather than to the wet weight of macrophytes, eliminates complications related to the mass of different

macrophyte species, problems of macrophyte preference by BHAB dinoflagellates and allows data to be

compared across studies. The protocols outlined in this study are the first steps to a standardized

sampling method for BHAB dinoflagellates that can support a cell-based monitoring program for

ciguatera fish poisoning. While this study is primarily concerned with the ciguatera-associated genus

Gambierdiscus, we also include data on the abundance of benthic Prorocentrum and Ostreopsis cells.

� 2014 Published by Elsevier B.V.

* Corresponding author at: Center for Coastal Fisheries and Habitat Research,

National Centers for Coastal Ocean Science, National Ocean Service, NOAA, 101

Pivers Island Road, Beaufort, NC 28516, USA. Tel.: +1 252 728 8792;

fax: +1 252 728 4537.

E-mail addresses: [email protected], [email protected] (P.A. Tester),

[email protected] (S.R. Kibler), [email protected] (W.C. Holland),

[email protected] (G. Usup), [email protected] (M.W. Vandersea),

[email protected] (C.P. Leaw), [email protected] (L.P. Teen),

[email protected] (J. Larsen), [email protected] (N. Mohammad-Noor),

[email protected] (M.A. Faust), [email protected] (R.W. Litaker).

Contents lists available at ScienceDirect

Harmful Algae

jo u rn al h om epag e: ww w.els evier .c o m/lo cat e/ha l

http://dx.doi.org/10.1016/j.hal.2014.06.009

1568-9883/� 2014 Published by Elsevier B.V.

http://crossmark.crossref.org/dialog/?doi=10.1016/j.hal.2014.06.009&domain=pdf



mailto:[email protected]












http://www.sciencedirect.com/science/journal/15689883

www.elsevier.com/locate/hal


coral rubble, rocks and sediments (Bomber and Aikman, 1989;Aligizaki et al., 2009; Cohu et al., 2011). The abundance of BHABdinoflagellates is most commonly quantified by collection ofmacrophytes, which are shaken in ambient seawater to suspendthe attached cells. The BHAB dinoflagellate cells are thenconcentrated and enumerated using standard microscopy meth-ods (Litaker et al., 2010 and references therein). BHAB dinoflagel-late abundances are generally expressed as cells g�1 wet weight(=fresh weight) of macrophyte (e.g., Yasumoto et al., 1979;Chinain et al., 1999; Mangialajo et al., 2011). However, colonizedsubstrates often possess complex morphologies with a wide rangeof surface area to mass ratios, making comparison of BHABdinoflagellate cell abundances among different substrates prob-lematic. The advantages of normalizing cell abundance to algalsurface area (cells cm�2) rather than algal mass (cells g�1 wetweight) was identified by Bomber et al. (1985) and Lobel et al.(1988), although methods for measuring algal surface areas aredifficult and often impractical. Other problems inherent to themacrophyte method include inconsistent distribution of macro-algae in time and space, scarcity or lack of the targeted macroalgalspecies among different environments, unequal dinoflagellateabundances among different macrophytes and discontinuous orpatchy distribution of BHAB dinoflagellate cells. This variabilitymeans that a relatively large number of replicate samples may berequired for a statistically robust measure of cell abundance(Lobel et al., 1988).

As an alternative to macrophytes, some researchers have usedartificial substrates to assess BHAB dinoflagellate abundance.Caire et al. (1985) employed fabric strips suspended in the watercolumn to monitor the Gambierdiscus population at an atoll inFrench Polynesia. Similarly, artificial materials (test tube brushes,plastic plates) have been used to compare the abundance ofProrocentrum lima on substrates with different surface areas in theFlorida Keys, USA (see Bomber and Aikman, 1989). Kibler et al.(2010), Tester et al. (2010) and Tan et al. (2013) used measuredpieces of fiberglass screen and Ishikawa et al. (2011) deployedfabric tubes (cotton 65%, synthetic 35%) as substrates to collectBHAB dinoflagellate cells in other tropical and subtropicalecosystems.

Artificial substrates offer numerous advantages over macro-phytes. The most important advantage is that dinoflagellate cellabundances can be more easily normalized to a known surface area(cells cm�2, cells 100 cm�2) for comparison among studies.Artificial substrates can be readily deployed across multiplespatial and temporal scales in any environment independent ofthe availability of macroalgae or other natural substrates. They canbe easily randomized, allowing the design of statistically rigorousfield studies. Significantly, artificial substrates also eliminatedinoflagellate-macroalgae preference effects, grazing by fish orother fauna and algal palatability considerations (see Cruz-Riveraand Villareal, 2006). Another advantage of samples collected fromartificial substrate is that the samples tend to be cleaner than thosefrom natural substrates with fewer contaminating biota orparticulates. This is likely a consequence of the short incubationtime. A disadvantage of the artificial substrate method is that eachsampling site must be visited twice, once to deploy the substratesand again to retrieve them.

In this study we compare two collection methods for measuringBHAB dinoflagellate abundances, an artificial substrate methodand the traditional macrophyte method. The objective was todevelop a widely applicable, statistically robust sampling methodwhereby cell abundances can be normalized across differentstudies. While this effort was primarily concerned with quantify-ing Gambierdiscus abundance as a cell-based monitoringprotocol for ciguatera fish poisoning (CFP), we also report dataon the abundance of benthic Ostreopsis and Prorocentrum cells.

Development of a universally adopted, fully validated samplingprotocol will help resolve long-standing questions such as thepotential environmental triggers for species-specific bloomformation, species toxicity, seasonality of abundance and environ-mental risks of BHAB events.

2. Methods

2.1. Sample sites

The feasibility of using artificial substrate (fiberglass screen)for quantifying the abundance of benthic dinoflagellates wastested in a range of tropical marine coastal environments in theCaribbean (Belize, Central America) as well as in the Indo-Pacific(Malaysia). Screen and comparative macrophyte (algae andseagrass) samples were collected in an array of habitats fromthe central lagoon system of Belize, Central America near CarrieBow Cay (16.80258 N, 88.08208 W) during May of 2009 andJanuary of 2012 (Fig. 1A). This portion of the Belizean centrallagoon is the type locale for a number of BHAB Gambierdiscus,Ostreopsis and Prorocentrum species (Faust, 1993, 1994, 1999;Faust and Morton, 1995; Faust et al., 2008; Litaker et al., 2009) andwas an ideal location to test the new sampling method. Malaysianscreen and macrophyte samples were collected in May 2012 alongthe eastern and western coasts of Pulau Sibu (2.21338 N,104.06768 E) and on the west coast of Pulau Tinggi (2.29438 N,104.11778 E) as part of the International Training Workshop onthe Ecology and Taxonomy of Benthic Marine Dinoflagellates held21–31 May 2012 in Pulau Sibu and the Universiti Kebangsaan,Malaysia. The field sites were protected islands located on thesoutheast coast of the Malay Peninsula, 30–33 km SSE of the city ofMersing (2.43578 N, 103.83088 E, Fig. 1B). Habitats sampledincluded protected mangrove embayments, island fringe envir-onments, lagoonal patch reefs, seagrass beds, coral fore and backreef sites, as well as rocky hard bottom areas (Tables 1 and 2).Screen and macrophyte samples were collected from 0.2 to 20 min relatively pristine environments as well as those heavilyimpacted by humans.

2.2. Screen sampling method

In order to test the screen method for characterizing BHABdinoflagellate abundance, this study was designed to address fourmain topics: (1) How long does it take for BHAB cells on theartificial substrate to achieve equilibrium with the surroundingcell abundances? (i.e., incubation or soak time); (2) How does thesize of the sampling screen (artificial substrate) affect cellabundance estimates?; (3) How well does the abundance of BHABcells associated with screens correlate with cell abundances frommacrophytes (natural substrate)?; and (4) How many replicatescreens are needed to assess BHAB dinoflagellate abundances formonitoring purposes?

The artificial substrate used in this study consisted of pieces ofblack fiberglass screen (window screen) cut into rectanglesmeasuring 10.2 cm � 15.2 cm (Fig. 2A). Each screen was attachedto monofilament fishing line and suspended in the water columnwithin �20 cm of the seabed using a weight and small subsurfacefloat (Fig. 2B). The subsurface floats were used to limit the length ofmonofilament line and avoid disturbance to the screen. Afterplacement, the screens were allowed to incubate for a definedperiod of time before being retrieved. For retrieval, a 775 ml plasticwide-mouth jar filled with ambient seawater was positionedbeside each screen before the screen was gently removed from themonofilament line and transferred to the jar (underwater) withoutbeing folded. The jar was then capped and returned to thelaboratory for processing.

P.A. Tester et al. / Harmful Algae 39 (2014) 8–25 9

2.3. Screen surface area

Screen surface area (AScr) was estimated using the surface areaand number of fiberglass filaments in the X and Y directions

(i.e., length and width) using the expression AScr = AXNX +AY NY � NXNY16r2, where AX is the surface area of each screenfilament in the X-direction, AY represents the surface area of eachscreen filament in the Y-direction and NX, NY are the number of

Fig. 1. Maps showing sampling locations: (A) Barrier reef and coral cays in the central lagoon of Belize, Central America and (B) Islands of Pulau Sibu and Pulau Tinggi,

Malaysia.

Source: Maps adapted from Google Earth Pro (v 7.0.3.8542, Google, Inc. 2013).

P.A. Tester et al. / Harmful Algae 39 (2014) 8–2510

filaments in the X (NX = 60) and Y (NY = 107) dimensions (Fig. 3).The expression NXNY16r2 represents surface area of intersectingfilaments as estimated by Weisstein (2013). The surface area ofeach filament in the X and Y directions were calculated as the areaof a cylinder (AX = 2prLX, AY = 2prLY), where r represents filamentradius (mean = 13 mm) and LX, LY correspond to filament lengths inrespective directions. Using this method, the surface area of each10.2 cm � 15.2 cm screen equaled 166 cm2.

2.4. Macrophyte sampling method

BHAB dinoflagellates were collected from macroalgae and asmall number of seagrass samples. Although the macroalgae variedfrom location to location, the most common genera collectedincluded various phaeophytes (Dictyota spp. J. V. Lamouroux andPadina spp. Adanson), rhodophytes (Amphiroa sp. J. V. Lamouroux,Acanthophora sp. J. V. Lamouroux, Laurencia sp. J. V. Lamouroux)and chlorophytes (Caulerpa sp. J. V. Lamouroux) (Guiry and Guiry,2012). Macroalgae were not identified to species. Seagrass sampleswere collected from beds of Thalassia testudinum Banks et Konig.Individual macrophyte samples were placed in 775 ml plastic jarswith ambient seawater. For small macroalgal specimens, the entire

thallus was collected without the holdfast. For larger specimens, asection of the thallus sufficient to fill approximately one third ofthe jar was pinched off and gently placed into the sample jar.Seagrass samples were similarly collected using scissors to gentlyremove enough material to fill approximately one quarter of thesampling container.

2.5. Sample processing

Approximately 20% of the seawater in each sample jar waspoured through a 300–500 mm sieve into a one liter graduatedcylinder. The jar was closed and shaken vigorously for 5–10 s todislodge dinoflagellates attached to the substrate (artificial ormacrophyte). The remaining homogenate was then pouredthrough the sieve into the cylinder and the total volume wasrecorded. Sieving was used to remove coarse sediment, detritusand other large materials from the sample. Macrophyte specimenswere set aside for weight determination; the sampling screenswere discarded. New screens were used for each experiment.Preliminary experiments using microscopy showed that a negligi-ble number of dinoflagellate cells remained attached to the screensafter shaking.

Table 1Sampling locations for screen vs. macrophyte (SM) experiments completed at sites in: (A) in Belize during May 2009, (B) in Belize during January 2012 and (C) in Malaysia

during May 2012.

Site Location Date Latitude Longitude Description

A. Belize screens vs. macrophyte sites – May 2009

SM1 Blue Ground Range 03 May 2009 16.7964 �88.1352 Mixed bed of Dictyota, Laurencia and

Acanthophora at a lagoonal patch reef

SM2 Carrie Bow Cay 04 May 2009 16.7964 �88.1352 Combined beds of Dictyota and

Acanthophora on E and W sides of island

SM3 Peter Douglas Cay 05 May 2009 16.7073 �88.1709 Caulerpa bed at the mouth of a eutrophic

mangrove embayment

SM4 Southwater Cay 06 May 2009 16.8151 �88.0824 Padina bed in a sheltered harbor

SM5 The Sand Bores 07 May 2009 16.7688 �88.1136 Amphiroa bed at a lagoonal patch reef

SM6 Carrie Bow Cay 12 May 2009 16.8025 �88.0823 Thalassia bed in a shallow back reef site

SM7 Carrie Bow Cay 12 May 2009 16.8029 �88.0821 Padina bed in a shallow back reef site

B. Belize screens vs. macrophyte sites – January 2012

SM8 Carrie Bow Cay 20 Jan 2012 16.8031 �88.0821 Acanthophora bed at N end of island

SM9 Southwater Cay 21 Jan 2012 16.8151 �88.0823 Padina bed on W side of island

SM10 Carrie Bow Cay 25 Jan 2012 16.8024 �88.0817 Dictyota bed on fore reef to E of island

SM11 Twin Cays 26 Jan 2012 16.8346 �88.1039 Acanthophora bed in mangrove channel

on N end of island

SM12 Carrie Bow Cay 30 Jan 2012 16.8024 �88.0817 Padina bed on E side of island

C. Malaysia screens vs. macrophyte sites – May 2012

SM13 Pulau Sibu 23 May 2012 2.1945 104.0710 Padina bed SW end of island

SM14 Pulau Sibu 23 May 2012 2.2144 104.0766 Padina bed E side of island

SM15 Pulau Sibu 24 May 2012 2.2185 104.0742 Padina bed E side of island

SM16 Pulau Tinggi 24 May 2012 2.2923 104.1040 Padina bed W side of island

SM17 Pulau Tinggi 25 May 2012 2.2925 104.1036 Padina bed W side of island

SM18 Pulau Tinggi 25 May 2012 2.2934 104.1016 Dictyota bed W side of island

Table 2Locations where sample size (SS) experiments were completed in Belize during May 2009.

Site Location Date Latitude Longitude Description

Belize sample size sites – May 2009

SS1 Peter Douglas Cay 05 May 2009 16.7073 �88.1709 Caulerpa bed at the mouth of a eutrophic mangrove embayment

SS2 Outer Twin Bays 02 May 2009 16.8284 �88.1063 Outer portion of Twin Bays, at Twin Cays

SS3 Northeast Peter Douglas Cay 02 May 2009 16.7135 �88.1730 Steep coral ridge along eastern face of Peter Douglas Cay

SS4 Blue Ground Range 03 May 2009 16.7964 �88.1352 Lagoonal patch reef site

SS5 Southwater Cay 06 May 2009 16.8151 �88.0823 Sheltered harbor with pilings and docks

SS6 The Sand Bores 06 May 2009 16.7688 �88.1136 Lagoonal patch reef site

SS7 Deep Peter Douglas Cay 08 May 2009 16.7065 �88.1703 Deep (20 m) slope of steep coral ridge along SE side of island

SS8 Shallow lagoon 10 May 2009 16.8040 �88.0864 Shallow (�2 m) Thalassia bed located 0.5 km west of Carrie Bow Cay

SS9 Deep lagoon 10 May 2009 16.7583 �88.1142 Deep (20 m) Thalassia bed near SS6

SS10 Carrie Bow Cay fore reef 10 May 2009 16.8025 �88.0784 Deep (20 m) fore reef site located 0.4 km east of Carrie Bow Cay

SS11 Curlew Bank fore reef 12 May 2009 16.7881 �88.0760 Deep (20 m) fore reef site located 0.3 km east of Curlew Bank

SS12 Twin Cays grass bed 12 May 2009 16.8246 �88.1025 Thalassia bed in a shallow (�1.5 m) trough along SE side of the big island


The sieved seawater sample was re-homogenized and 50–775 ml were gravity filtered through a 25 mm diameter piece of20 mm pore size nylon mesh to collect the BHAB dinoflagellatecells. If gravity was insufficient to filter the entire aliquot throughthe mesh, a hand vacuum pump was used to gently filter theremainder of the sample (<5 cm Hg). Any particulate materialsremaining in the graduated cylinder were rinsed onto the meshwith filtered seawater. The piece of nylon mesh was thentransferred to a 15 ml screw-cap tube containing 5–10 ml of GF/F (Whatman1, Piscataway, NJ, USA) filtered seawater and thesample was preserved with 1–2 drops of neutral Lugol’s iodinesolution or 1% glutaraldehyde (Throndsen, 1995).

The cell abundances of Gambierdiscus, Ostreopsis and Prorocen-

trum in each screen or macrophyte sample were determined usingmicroscopy. Specifically, each preserved sample was shaken tosuspend BHAB dinoflagellate cells and an aliquot was transferredto one well of a 6-well plate containing 2–3 ml of GF/F filteredseawater. The nylon mesh was discarded. The sample was allowedto settle for 12–24 h and cell abundance was determined byinverted light microscopy (Utermohl, 1958). Between two and fouraliquots were counted from each sample. For screen samples,BHAB dinoflagellate concentrations were expressed as cells100 cm�2, calculated using the average abundance of cells in eachsample, the surface area of each screen (166 cm2) and anappropriate volumetric conversion factor. For macrophyte sam-ples, BHAB dinoflagellate concentrations were determined using asimilar method, except concentrations were normalized to the wetweight of the macrophyte specimen and expressed as cells g�1 ofalgae or seagrass.

2.6. Incubation time experiment

To assess the recruitment of benthic dinoflagellate cells to theartificial substrate, an experiment testing different incubationtimes was conducted at Carrie Bow Cay, Belize (CBC) during 19–21January 2012. A group of six screens was moored as describedabove at a shallow back reef site characterized by a mixed coral,seagrass and macroalgae assemblage. One screen was collected at6, 12, 18, 24, 36 and 48 h and abundances of Gambierdiscus,Ostreopsis and Prorocentrum cells were determined as describedabove (cells 100 cm�2).

2.7. Screen size experiment

In order to examine the effect of screen size on recruitment ofBHAB dinoflagellate cells, an experiment was conducted in theshallow back reef environment at CBC during 21–22 January 2012.Two different sized screens, 166 cm2 (n = 3) and 257 cm2 (n = 3),were incubated for 24 h and Gambierdiscus, Ostreopsis andProrocentrum abundances were quantified as detailed above (cells100 cm�2). A student’s t-test was used to assess normallydistributed data for significant differences in recruitment of BHABdinoflagellates between the two sized screens (Sokal and Rohlf,1995). A Mann–Whitney rank sum test was used to assess non-normally distributed data (Zar, 1996).

2.8. Screen vs. macrophyte experiments

To directly compare dinoflagellate cell abundances on artificialand natural substrates over a range of different environments(designated as screen vs. macrophyte sites, SM), replicate screenand macrophyte samples were collected in different habitats. Inthe first set of experiments, samples were collected from sevensites in Belize in May 2009 (Table 1A). At each site, 6–9 individualscreens were placed haphazardly at locations within a 2–3 mcircular area of a macrophyte bed. After the screens were deployed

for 24 h, each was collected and a similar number of macrophytesamples were taken within the same area. The same method wasused for SM experiments in the shallow back reef area surroundingCarrie Bow Cay, Belize in January 2012. The only difference wasthat sample size was limited to three screens and threemacrophyte samples (Table 1B).

To improve randomization of individual screen and macrophytesamples, a more rigorous sampling protocol was tested at six sitesat Pulau Sibu and Pulau Tinggi, Malaysia in May of 2012 (Table 1C).For this method, the centers of pre-selected, monospecific algalbeds (Padina sp., Dictyota sp.) were used as reference points forplacement of six replicate screens. A circular grid with twoconcentric rings at approximately 0.5 and 1 m from the center ofeach bed was utilized to determine the placement of each screenusing a random number generator (Fig. 2C). After 24 h, screenswere collected. Replicate algae samples were then selectedrandomly from six other locations on the same sampling grid.Processing of screen and algae samples and counting of BHABdinoflagellates were completed as described above. The relation-ship between mean screen and macrophyte abundances was thenanalyzed using linear regression analysis of log-transformed data

Fig. 2. Artificial substrate and sampling methods. (A) Photo of fiberglass screen

section (166 cm2) used as an artificial substrate for BHAB dinoflagellate collection.

(B) Schematic showing anchoring assembly for screens with subsurface float and

bottom weight. Sampling device was constructed of heavy gauge monofilament

fishing line, swivels, weights and floats. (C) Schematic of circular grid used to

randomize screen and algae samples (SM13–SM18, Table 1C) for comparative

experiments conducted at Pulau Sibu and Pulau Tinggi, Malaysia in May 2012.

Concentric circles were located 0.5 and 1 m from the center of the algal bed and

sample locations on the grid were selected using a random number generator.


(base 10). A value of one was added to all abundances prior toanalysis to allow logarithmic transformations.

The variability among replicate samples of BHAB dinoflagellatesassociated with artificial and natural substrates was comparedusing the coefficient of variation (CV). This comparison wasaccomplished using the D’AD test developed by Feltz and Miller(1996) and Miller (1991), which compares the CV of each group(screens, macrophytes) to the CV of the pooled population using ax2 distribution. Because the D’AD test assumes data are normallydistributed, and both Gambierdiscus and Ostreopsis abundancesdeviated from normality (Shapiro–Wilk test, p < 0.05), SMabundance data for these two genera were first square roottransformed prior to CV analysis. Despite persistent deviationsfrom normality in one group (Gambierdiscus abundance onmacrophytes), the D’AD test was applied because it is robust fornon-normal data (Feltz and Miller, 1996).

2.9. Sample size experiments

To estimate the number of replicate screens necessary to assessBHAB abundances in a range of environments, a series of samplesize (SS) experiments was conducted using replicate screens at 12sites in Belize during May 2009 (Table 2). It is a premise common tomost ecological sampling that the mean of all replicate samples ateach site yields the best estimate of abundance. However, sincelogistical factors and manpower limitations often constrain thenumber of samples that can be collected in monitoring programs, itis desirable to utilize the fewest number of screens possible. Basedupon results of previous sampling efforts conducted in theCaribbean region as well as a review of the BHAB literature, thecoefficient of variation (CV) was used as a criterion to define anacceptable degree of variability among replicate samples. Specifi-cally, a CV threshold of 100% (N100) was selected as a referencevalue to delineate variability around the sample mean. Compara-tive data from the literature are rare, either because samples werenot replicated or the replicate values were not reported. However,there are a few BHAB studies where it is possible to addressvariability among replicates, albeit on natural substrates. Forinstance, Carlson (1984) reported CVs of 86–170% for Gambierdis-

cus, 89–200% for Ostreopsis and 64–253% for Prorocentrum

abundances on macroalgae from sites in the US and British VirginIslands. Similarly, Ballantine et al. (1988) and Yasumoto et al.(1979, 1980) reported CVs of 29% to >200% for BHAB cellabundances among macrophytes collected at sites in Puerto Ricoand the Pacific, respectively. More recent data from the Mediter-ranean Sea and the Indo-Pacific demonstrated BHAB abundances

with similar levels of variability (Marasigan et al., 2001; Aligizakiet al., 2009; Shears and Ross, 2009; Cohu et al., 2011). Given therelatively large degree of sample variability encountered in thisand other studies, our decision to use a CV of 100% represents arelatively rigorous level of precision.

For the SS study, nine replicate screens were attached to a0.25 m2 quadrat made of plastic plumbing pipe and incubated ateach sampling site for 24 h before screens were collected andprocessed as above. For 10 of the 12 sites all nine screens werecollected successfully, but at the other two sites screens were lostduring incubation (n = 8 at SS4, n = 7 at SS6, Table 6). Because theaverage abundance of BHAB dinoflagellates calculated at each sitedepended not only on the number of replicate screens, but also onwhich replicates were used, all possible unique combinations ofreplicate abundances of Gambierdiscus, Ostreopsis and Prorocen-

trum cells were compiled using MATLAB (R2013a, The Mathworks,Inc. Natick, MA, USA) for sample sizes between n = 2 and n = 9.Once all combinatorial data sets were generated, the means (x,cells 100 cm�2) and CVs (%) were calculated for each sample size.The result was an array of average cell abundances at each site,most of which were characterized by a gradual reduction in the CVas the sample size increased. The sample size necessary to achievea CV of 100%, N100, was then determined graphically forGambierdiscus, Ostreopsis and Prorocentrum by plotting samplesize vs. CV for n = 2, 3, . . ., n. Sites where the CV did not declinebelow the reference level of 100%, would require sample sizes >9to accurately estimate cell abundances. This situation primarilyarose when cell abundances were very low (<20 cells 100 cm�2),when BHAB cells were lacking on some replicate screens (i.e., 0cells 100 cm�2) or when both conditions occurred.

In order to determine the number of macrophyte samplesneeded to characterize BHAB abundance, the sample size analysisprocedure was repeated using replicate macrophyte samples fromthe screens vs. macrophyte experiments (see above). The smallsample size (n = 3) at the SM sites in Belize during 2012 precludedcalculation of N100.

2.10. Data analysis

Most of the data analyses and all of the graphical visualizationswere completed using functions incorporated in SigmaPlotsoftware (v. 11.0, Systat Software, Inc., San Jose, CA, USA).Significant differences among groups (screen size experiment,Section 2.7) were assessed using Student’s t tests (normal data) orMann–Whitney rank sum tests (non-normal data) after normalityand equality of variances were verified using the assumptionchecking functions of SigmaPlot software. Normality of data wasaddressed using a Shapiro–Wilk test and equality of variances wasaddressed using a constant variance test based on a Spearman rankcorrelation. Linear regression analyses (screen vs. macrophyteexperiments, Section 2.8) were completed following similarprocedures. Statistical testing for CV comparisons among screenand macrophyte data and sample size experiments are described inSections 2.8 and 2.9.

3. Results

3.1. Incubation time and screen size

The incubation time experiment conducted in Belize in January2012 demonstrated the abundance of benthic dinoflagellatesreached a plateau when screens were incubated for one full day.Gambierdiscus, Ostreopsis and Prorocentrum cells began associatingwith the screens within 6 h of being deployed. After the first 12 h,cell concentrations had increased 2–10-fold relative to the 6 hsample (Fig. 4). Screen-associated dinoflagellate populations

Fig. 3. Schematic of screen material for determination of surface area. Surface area is

determined by considering screen filaments as a series of long, thin cylinders with

radius r, lengths in the X (LX) and Y (LY) directions, and filament areas AX = 2prLX and

AY = 2prLY, respectively. Surface area of the entire screen (AScr) is determined using

the expression AScr = AXNX + AYNY � NXNY16r2 after Weisstein (2013). See text for

details.


continued to increase until 24 h, when Gambierdiscus, Ostreopsis

and Prorocentrum cell abundances reached 211, 238 and 1597 cells100 cm�2, respectively. The much higher Prorocentrum abun-dances, compared to the other two genera, reflected the greaterpopulation of this genus at the sampling site during theexperiment. Between 24 and 48 h, there was little change in

Gambierdiscus or Prorocentrum abundances, whereas Ostreopsis

abundance declined slightly (Fig. 4).The screen size experiment showed differently sized screens,

166 and 257 cm2, were equally effective at characterizing theabundances of BHAB dinoflagellates. After the 24 h incubation,average Gambierdiscus abundances reached virtually identicalconcentrations of 41 � 18 and 40 � 13 cells 100 cm�2 on the 166 and257 cm2 screens, respectively (Fig. 5, Student’s t-test = 0.07, p = 0.95).Similarly, mean Ostreopsis abundances on the 166 and 257 cm2

screens reached 67 � 14 and 64 � 10 cells 100 cm�2, respectively,while Prorocentrum abundances reached 1163 � 109 cells 100 cm�2

and 868 � 264 cells 100 cm�2 (Fig. 5). The lower mean abundance ofProrocentrum associated with the larger screen (868 cells 100 cm�2)was attributed to a single replicate with substantially fewer cells (607cells 100 cm�2). Because the abundance data for Ostreopsis andProrocentrum were not normally distributed, Mann–Whitney ranksum tests were used to test for differences among the two screensizes, where T represents the Mann–Whitney test statistic. Testresults indicated the number of cells recruited to the different screenswere not significantly different for Ostreopsis (T = 11.0, p = 1.0) orProrocentrum (T = 14.0, p = 0.20).

3.2. BHAB abundance on screens vs. macrophytes

The results of the screen vs. macrophyte (SM) experimentsshowed BHAB dinoflagellate abundances varied among thedifferent sampling sites and time periods by 4–5 orders ofmagnitude. Overall, the widest range in dinoflagellate abundancesoccurred among sites in Belize during 2009 when Gambierdiscus

concentrations ranged from a few cells per sample (screens ormacrophyte) to 35,086 cells 100 cm�2 on screens (GScr) and7460 cells g�1 of macrophyte (GMac, Fig. 6A and Table 3). During thesame study period, Ostreopsis reached maximum abundances of8149 cells 100 cm�2 (OScr) and 1464 cells g�1 (OMac, Fig. 6B andTable 4) while Prorocentrum abundances (PScr, PMac) reached 9817cells 100 cm�2 and 3063 cells g�1, respectively (Fig. 6C andTable 5).

In contrast to the 2009 sampling period, samples from Belize in2012 exhibited lower abundances consistent with the lower watertemperatures during January. During this period, GScr and GMac

reached only 822 cells 100 cm�2 and 134 cells g�1 (Fig. 6D

Fig. 4. Time series of Gambierdiscus, Ostreopsis and Prorocentrum cell abundances

(cells 100 cm�2) associated with screens incubated at Carrie Bow Cay, Belize for 6,

12, 18, 24, 36 and 48 h on 18–20 January 2012 (n = 1).

Fig. 5. Abundances (cells 100 cm�2) of (A) Gambierdiscus, (B) Ostreopsis and (C) Prorocentrum cells associated with screens of two different sizes (166 cm2, 257 cm2, n = 4)

incubated for 24 h at Carrie Bow Cay, Belize, 21–23 January 2012.


and Table 3), OScr and OMac reached only 592 cells 100 cm�2 and294 cells g�1 (Fig. 6E and Table 4), and Prorocentrum reachedmaximal abundances of 4093 100 cm�2 and 787 cells g�1 (Fig. 6Fand Table 5). Some of the lowest overall BHAB dinoflagellateabundances occurred at the Malaysian sites during May 2012where the GScr and GMac maxima reached only 114 cells 100 cm�2

and 35 cells g�1 (Fig. 6G and Table 3), respectively, OScr and OMac

reached 1898 cells 100 cm�2 and 346 cells g�1 (Fig. 6H and

Table 4), and PScr and PMac reached 2329 cells 100 cm�2 and432 cells g�1 (Fig. 6I and Table 5).

Comparison of average dinoflagellate BHAB abundancesshowed cells associated with screens paralleled those on macro-phytes at most sites. This was most apparent for the 2009 datafrom Belize where the sites with the highest abundances onscreens were generally the same as those with the highestabundances on macrophytes (Fig. 6). For example, sites SM3, SM6

Fig. 6. Results of Screen vs. Macrophyte (SM) experiments. Abundances of Gambierdiscus, Ostreopsis, and Prorocentrum cells associated with replicate screen (black bars, cells

100 cm�2) and algae samples (white bars, cells g�1) collected at sites in Belize during May 2009 (A–C), in Belize during January 2012 (D–F) and in Malaysia during May 2012

(G–I). Abundances are given as mean and standard deviation of the mean (error bars) to show variability among replicates and association among screen and macrophyte

data. Refer to Table 1 for site descriptions. The lines connecting the bars are included to visualize trends and do not reflect direct relationships.


and SM7 exhibited the three highest mean abundances ofProrocentrum (6529, 6655, 2806 cells 100 cm�2, respectively;Fig. 6C and Table 5) associated with screens. These three sites alsohad the three highest abundances of Prorocentrum cells onmacrophytes (2299, 3063 and 616 cells g�1, respectively). Al-though a similar relationship was apparent in the Belize 2012 andMalaysia 2012 SM experiments, there were a few samples where

the abundances varied to a greater degree than observed in Belizein 2009 (Fig. 6F and I, Table 5).

A number of screen and macrophyte samples contained zeroGambierdiscus cells. The frequency of zeroes was highest at siteswith low overall dinoflagellate abundances, such as site SM14,where four of six macrophyte samples had zero Gambierdiscus cells(Fig. 6G). Similarly, zero Ostreopsis cells were present in three of

Table 3Results of screens vs. macrophyte (SM) experiments for Gambierdiscus cells at sites in Belize during May 2009 and January 2012 and in Malaysia during May 2012. Data are

given as the mean abundances, x (cells 100 cm�2), standard deviation (Std, cells 100 cm�2), minimum (Min), maximum (Max), coefficient of variation (CV, %) and sample size,

n. ‘‘n/a’’ Indicates sample size insufficient to reach N100 threshold.

Site Gambierdiscus

Screen samples Macrophyte samples

x cells

100 cm�2

Std cells

100 cm�2

Min Max CV (%) N100 n x cells g�1 Std cells

100 cm�2

Min Max CV (%) N100 n

Belize 2009

SM1 96 69 23 233 71 5 8 15 8 9 29 54 2 5

SM2 37 35 0 94 97 9 9 6 12 0 36 205 n/a 9

SM3 20,798 14,086 5308 35,086 68 4 5 4524 1876 2538 7460 41 4 5

SM4 319 127 164 470 40 2 5 41 21 15 66 52 2 5

SM5 33 27 0 70 81 5 5 3 2 2 7 63 2 5

SM6 363 127 172 517 35 2 6 75 33 41 129 44 2 6

SM7 431 94 329 587 22 2 6 59 14 41 76 24 2 6

Belize 2012

SM8 359 102 254 458 28 n/a 3 42 20 20 58 47 n/a 3

SM9 146 31 127 182 21 n/a 3 34 20 22 57 59 n/a 3

SM10 200 31 176 235 16 n/a 3 20 3 17 23 17 n/a 3

SM11 36 40 12 82 113 n/a 3 8 3 5 10 36 n/a 3

SM12 752 62 705 822 8 n/a 3 104 36 64 134 35 n/a 3

Malaysia 2012

SM13 8 8 0 20 102 n/a 6 2 1 1 3 49 2 6

SM14 24 14 0 39 57 4 6 1 3 0 8 208 n/a 5

SM15 73 30 42 114 41 2 6 21 12 5 35 56 3 6

SM16 26 16 11 53 61 2 6 10 12 0 33 120 n/a 6

SM17 47 25 11 76 54 3 6 7 5 0 12 68 4 6

SM18 22 11 11 33 50 2 5 4 5 0 14 122 n/a 6

Table 4Results of screens vs. macrophyte (SM) experiments for Ostreopsis cells at sites in Belize during May 2009 and January 2012 and in Malaysia during May 2012. Data are given

as the mean abundances, x (cells 100 cm�2), standard deviation (Std, cells 100 cm�2), minimum (Min), maximum (Max), coefficient of variation (CV, %) and sample size, n. ‘‘n/

a’’ Indicates sample size insufficient to reach N100 threshold.

Site Ostreopsis


x cells

100 cm�2

Std cells

100 cm�2


100 cm�2


Belize 2009

SM1 9 12 0 23 138 n/a 8 21 31 0 74 149 n/a 5

SM2 600 451 164 1409 75 7 9 188 228 46 747 121 n/a 9

SM3 160 224 0 470 140 n/a 5 36 49 0 90 137 n/a 5

SM4 554 159 329 728 29 3 5 937 335 558 1282 36 3 5

SM5 0 0 0 5 0 0 0 5

SM6 6346 2366 1691 8149 37 2 6 1338 81 1258 1464 6 2 6

SM7 3685 384 3264 4180 10 2 6 294 44 242 343 15 2 6

Belize 2012

SM8 123 54 63 129 44 n/a 3 12 10 5 23 86 n/a 3

SM9 31 12 18 42 39 n/a 3 7 8 0 16 117 n/a 3

SM10 168 90 63 222 54 n/a 3 11 3 8 15 31 n/a 3

SM11 254 143 95 370 56 n/a 3 28 6 23 34 22 n/a 3

SM12 451 133 328 592 29 n/a 3 221 64 172 294 29 n/a 3

Malaysia 2012

SM13 80 63 36 203 78 4 6 30 11 16 42 36 2 5

SM14 1268 407 770 1898 32 2 6 45 20 23 79 44 2 6

SM15 768 460 349 1560 60 2 6 232 88 152 346 38 3 6

SM16 157 70 59 261 44 2 6 56 37 20 116 66 4 6

SM17 137 77 22 235 56 3 6 39 12 27 59 31 2 6

SM18 359 194 141 603 54 2 5 33 7 21 42 23 2 6


five screen samples at site SM3. Ostreopsis cells were absent in allscreen and macrophyte samples collected at site SM5 (Fig. 6B). Incontrast, Prorocentrum cells were present in every sample taken inthis study.

The results of the CV analyses indicated the variability amongBHAB abundances on screens and macrophytes did not differ.Neither the means nor variances in BHAB abundance among screenand macrophyte samples could be equated because they werenormalized to different quantities (cells 100 cm�2 vs. cells g�1).However, the CVs of the Gambierdiscus cell abundances from thetwo substrates were unit-less (CV ¼ std=x), and therefore directlycomparable. The results of the D’AD test showed CVs for pooledGambierdiscus (D’AD = 0.64, p > 0.05, n = 18), Ostreopsis

(D’AD = 0.01, p > 0.05, n = 18) and Prorocentrum (D’AD = 1.43,p > 0.05, n = 18) abundances on screens were not significantlydifferent than those on macrophytes.

Despite the wide range of BHAB cell abundances and highsample variability at some sites, regression analysis showed therewas a positive, linear relationship between abundance of BHABdinoflagellate cells on artificial and natural substrates. Thisrelationship was most apparent during the Belize 2009 samplingperiod, when average Gambierdiscus abundance on screens wasstrongly related to cell abundance on macrophytes (GMac, Fig. 7A,r2 = 0.99, p < 0.001). Similar linear relationships were apparent

for both Ostreopsis (Fig. 8A, r2 = 0.84, p = 0.02) and Prorocentrum

(Fig. 9A, r2 = 0.81, p = 0.004) during this sampling period.Although the sample size and number of sites from Belize in2012 (n = 3, 5 sites) was smaller than in 2009 (n = 5–9, 7 sites),there were significant linear relationships among mean GScr andGMac (Fig. 7B, r2 = 0.83, p = 0.021) and mean OScr and OMac (Fig. 8B,r2 = 0.80, p = 0.027). However, the relationship between meanProrocentrum abundances on screens and macrophytes was notsignificant, largely due to relatively high cell abundances onscreens compared to macrophytes at site SM9 (Figs. 6F and 9B,Table 5).

The weakest relationship among screen and macrophytesamples was evident during the 2012 sampling period in Malaysiawhere all BHAB cell abundances were relatively low (Fig. 6G–I).Regression analysis showed neither Gambierdiscus, Ostreopsis norProrocentrum average abundances from screens were linearlyrelated to those from macrophytes (Figs. 7C, 8C and 9C, p > 0.05).However, the screens vs. macrophyte relationships for bothGambierdiscus (p = 0.055) and Prorocentrum (p = 0.054) were justabove the p = 0.05 threshold. A larger sample size would likelyhave resulted in a significant linear fit. It is possible that therelatively large regional tidal range at Malaysian sampling sites (1–3 m, Taira et al., 1996; Wan-Jean, 2005) played a role distributingBHAB cells among benthic substrates, thereby weakening the

Fig. 7. Results of Screen vs. Macrophyte (SM) experiments. Mean abundances of Gambierdiscus cells associated with replicate screen (cells 100 cm�2) and macrophyte samples

(cells g�1 wet weight) collected at sites in (A) Belize during May 2009, (B) Belize during January 2012, (C) Malaysia during May 2012 and (D) pooled data from all three

locations. Linear regression (solid lines) and 95% confidence intervals (dotted lines) are shown, together with regression equations, coefficients and p-values. The regression

line in panel C is not shown because the relationship was not significant (p � 0.05). Data were log-transformed prior to regression analysis and have been plotted on a linear

scale.


linear relationship among BHAB abundances on screens andmacrophytes.

Although there were weak linear relationships in BHABdinoflagellate abundance in some of the screen vs. macrophyteexperiments, the pooled data from all sampling periodsexhibited significant linear relationships among abundances onscreens and macrophytes (p < 0.001) for Gambierdiscus (Log GScr =0.745 + 0.995 � Log GMac, r2 = 0.95), Ostreopsis (Log OScr =0.615 + 0.999 � Log OMac, r2 = 0.71) and Prorocentrum (Log PScr =1.39 + 0.712 � Log PMac, r2 = 0.66) (Figs. 7D, 8D and 9D, respectively).The robust linearity among the SM cell abundance data demon-strated that screens were an excellent substitute for macroalgae andseagrasses. This was true despite the samples having been collectedfrom a variety of benthic habitats with different tidal regimes andover multiple years.

3.3. Required sample size

The sample size experiments demonstrated dinoflagellateBHAB abundance could be assessed adequately using fewer thannine replicate screens, provided cell abundances were sufficientand sample variability was not extreme. An example of the samplesize analysis is given in Fig. 10, which shows eight replicatemeasurements of Gambierdiscus abundance (96 � 69 cells

100 cm�2) from screens placed at site SS4 in Belize during 5–6May 2009. Mean Gambierdiscus abundance depended greatly uponthe number of screens that were used. For instance, when only twoscreens were utilized, mean Gambierdiscus abundances calculated atthis site ranged from 35 to 198 cells 100 cm�2, contingent upon whichtwo replicates were used (Fig. 10D). Depending on the number ofreplicates used, either 3, 4, 5, 6 or 7, the range of means varied from 47to 163, 53 to 140, 56 to 126, 62 to 117, and 77 to 107 cells 100 cm�2,respectively. The possible CVs followed a similar pattern, rangingfrom 0 to 116% for two replicates, decreasing to 57–79% as sample sizeincreased to seven and reaching 71% when all eight screens wereincluded (Fig. 10G). The range of CVs declined below the 100%reference level when at least five replicate screens were used (dashedline, Fig. 10G). In other words, the analysis showed only five replicatescreens were necessary at site SS4 to achieve an acceptable estimateof Gambierdiscus abundance (i.e., N100 = 5). Because of the highabundance of Prorocentrum cells and relatively small samplevariability among replicate screens, the sample size analysis showedas few as two screens were adequate to assess Prorocentrum

abundance at SS4 (Fig. 10I).In contrast, the mean Ostreopsis abundance at SS4 was low

(9 � 12 cells 100 cm�2), representing a CV of 138% among the eightscreens (Table 6). Furthermore, Ostreopsis cells were absent from fiveof the eight screens incubated at this site (Fig. 10B). Although the

Fig. 8. Results of screen vs. macrophyte (SM) experiments. Mean abundances of Ostreopsis cells associated with replicate screen (cells 100 cm�2) and macrophyte samples



line in panel C is not shown because the relationship was not significant (p � 0.05). Data were log-transformed prior to regression analysis and have been plotted on a linear

scale.


range of possible mean abundances narrowed from 0 to 16 cells100 cm�2 when n = 2 to 11–12 cells 100 cm�2 when n = 7 (Fig. 10E),the range of possible CVs did not follow this pattern. Instead, CVsincreased with sample size due to the extremely high variabilityamong replicates, ranging from 0 to 141% at n = 2 to nearly 250%when n = 6 (Fig. 10H). While CVs began to decline at n > 6, they didnot decrease below 100%, precluding determination of N100. A similarpattern of increasing CVs with sample size occurred at other siteswith multiple zero abundance values.

Overall, the sample size experiments indicated N100 forGambierdiscus and Ostreopsis varied substantially among samplinglocations. For Gambierdiscus, the greatest sample variabilityoccurred at sites SS2, SS3, SS6 and SS9 with CVs of 131–150%(Table 6). The CV vs. sample size plots indicated the sample sizewas too small to provide an estimate of N100 at these sites (‘‘n/a’’ inTable 6). Aside from these four locations with high CVs, the N100 forGambierdiscus ranged between two (SS1, SS5, SS12) and nine(SS10) screens. For Ostreopsis, sites SS1, SS2, SS4 and SS6 exhibitedsuch high CVs (125–212%) that N100 could not be estimated. Ateach of these sites, there were no Ostreopsis cells present on themajority of the replicate screens. For the remainder of the samplesites N100 ranged between two (SS3, SS5, SS12) and eight screens(SS7, Table 6).

With the exception of site SS1, the sample size experimentshowed that Prorocentrum abundances were sufficiently high(x ¼ 289�1751 cells 100 cm�2) and variability was sufficiently low(CV 17–64%) to allow estimation of N100 at all sites. At sites SS2–SS12, N100 ranged between two and three screens. Prorocentrum

abundance on the replicate screens at site SS1 ranged between1033 and 107,067 cells 100 cm�2, yielding a CV of 128% (Table 6).This was the most extreme range of cell abundances among thethree genera at a single site observed during this entire study.

In order to compare sample sizes among artificial and naturalsubstrates, data from the SM experiments in Belize 2009 andMalaysia 2012 were re-analyzed using the same method of samplesize analysis. During the 2009 sampling period in Belize,Gambierdiscus abundance was relatively high (sites SM3, SM4,SM6, SM7), CVs were 22–68% and N100 was estimated to be 2–4samples (Table 3). A similar number of samples was necessary formacrophytes at these same sites (N100 = 2–4, CVs 24–52%, Table 3).At sites with lower cell abundances and higher CVs, N100 was equalto five screens (SM1, SM5). A notable outlier among the screens vs.macrophyte experiments was site SM2, where two separate bedsof macroalgae were combined (Dictyota, Acanthophora, Table 1).Here, CVs were elevated in both screen (97%) and algae samples(205%). As a result, N100 was estimated to be seven screens but the

Fig. 9. Results of screen vs. macrophyte (SM) experiments. Mean abundances of Prorocentrum cells associated with replicate screen (cells 100 cm�2) and macrophyte samples



lines in panels B and C are not shown because the relationships were not significant (p � 0.05). Data were log-transformed prior to regression analysis and have been plotted

on a linear scale.


sample size was too small to estimate N100 for the algae samples atthis site (Table 3).

With relatively low cell abundances at all sites in Malaysia2012, screens proved a more consistent substrate than macro-phytes to assess Gambierdiscus populations. Although CVs amongscreen samples were low enough to estimate N100 at sites SM14–SM18 (CVs 41–61%, N100 = 2–4), the sample size for correspondingmacrophyte samples could only be determined at SM15 and 17 dueto higher variability at the remaining sites (CVs 120–208%,Table 3). Only at SM13 was the Gambierdiscus sample variabilityamong screens too high (CV 102%) to allow N100 to be estimated,though the CV was relatively low (49%) in the macrophyte samplesfrom this site (N100 = 2).

The SM results for Ostreopsis cells showed relatively lowvariability on both screens and macrophytes at sites SM4, SM6 andSM7 where only two to three replicate screen or macrophytesamples were needed to reduce CVs below the reference level of100% (Table 4). Variability was much higher at sites SM1 and SM3(CVs 138–149%), preventing N100 determination for either screenor macrophyte samples. The mixed algal bed at SM2 yielded a N100

of seven screens, but variability was too high to allow sample size

determination for the algal samples (CV 121%, Table 4). Thesampling sites in Malaysia during 2012 showed CVs were <100%among both screens and macrophytes (23–78%), despite that only5–6 replicate samples were collected. Sample size analysisdemonstrated that between two and four samples of eithersubstrate were sufficient to yield CVs < 100% (Table 4).

Unlike either Gambierdiscus or Ostreopsis, the screens vs.macrophyte experimental results for Prorocentrum showed lowvariation among sites. Despite a wide range of abundances amongscreens (PScr 211–9817 cells 100 cm�2) and macrophytes (PMac 26–3063 cells g�1), CVs for both substrates were relatively low insamples from Belize 2009 (12–53%) and Malaysia 2012 (CVs 22–77%) (Table 5). As a result, N100 was only two to three samples forscreens and macrophytes from each location.

4. Discussion

This study demonstrated artificial substrate (fiberglass screen)performed as well as macrophytes for quantifying the abundanceof BHAB dinoflagellates. Specifically, a series of parallel experi-ments using artificial substrate and macrophytes showed the

Fig. 10. Example of sample size results. Abundance of (A) Gambierdiscus, (B) Ostreopsis and (C) Prorocentrum cells (cells 100 cm�2) associated with replicate screens moored at

the mangrove bay at Site SS1, Peter Douglas Cay, Belize 5–6 May 2009. (D–F) Average cell abundances vs. sample size (n) for each BHAB genus using all unique random

combinations of the replicate screens (see Section 2). (G–I) CV (%) combinatorial abundance data vs. sample size for each BHAB genus. The broken line denotes �100% CV from

the mean (N100).


abundance of BHAB dinoflagellates recruited to fiberglass screensover a 24 h period was directly related to the abundance of thesame dinoflagellate taxa on macrophytes at the same locations.The strongest linearity between mean screen and macrophyte cellabundances for Gambierdiscus (r2 = 0.99, Fig. 7A), Ostreopsis

(r2 = 0.84, Fig. 8A) and Prorocentrum (r2 = 0.81, Fig. 9A) wereobserved in the 2009 samples from Belize. Despite weak linearityin some of the screen vs. macrophyte experiments, wherevariability was high and cell abundances were low, pooled datafrom all sampling periods showed significant linear relationshipsfor Gambierdiscus (r2 = 0.95, p < 0.001, Fig. 7D), Ostreopsis

(r2 = 0.71, p < 0.001, Fig. 8D) and Prorocentrum (r2 = 0.66,p < 0.001, Fig. 9D). The linearity among screen and macrophyteabundance data was quite remarkable considering the BHABsamples were collected from a diverse array of habitats and avariety of macrophytes (Tables 1 and 2).

Interestingly, both the screen and macrophyte samplesexhibited similar degrees of variability among replicate samples.When the CVs were compared among screens and macrophytes,there were no significant differences in variability among samplesfrom the two substrates for Gambierdiscus, Ostreopsis, or Prorocen-

trum (D’AD test, p > 0.05). The linearity in cell abundancesbetween screen and macrophyte samples indicate the factorsgoverning dinoflagellate recruitment of BHAB cells to artificial andnatural substrates are comparable.

The artificial substrate sampling method employed in this studywas based on the recruitment of BHAB cells to clean, artificialsurfaces. We hypothesized the number of cells recruited to thescreens would be proportional to the overall number of BHAB cellsin the environment. A survey of the literature revealed noconsistent preference of Gambierdiscus species for particularmacrophytes (Litaker et al., 2010). This conclusion was furthersupported in a recent laboratory study (Parsons et al., 2011), whichprovided no evidence that Gambierdiscus cells exhibited apreference for a particular macroalgal species. Other laboratoryexperiments with Gambierdiscus have shown the association ofcells with substrate surfaces is highly ephemeral and that

emigration and immigration behaviors may be prompted by bothphysical and chemical cues (Nakahara et al., 1996). Consistent withthese observations, substantial numbers of BHAB dinoflagellateshave been reported in the water column in addition to beingassociated with benthic substrates (Tindall and Morton, 1998; Vilaet al., 2001; Gayoso et al., 2002; Levasseur et al., 2003; Faust et al.,2005; Faust, 2009; Totti et al., 2010; Cohu et al., 2011). Much likethe colonization of screens, the colonization of newly availablecoral substrate by opportunistic Gambierdiscus cells has beenposited to foster blooms leading to CFP outbreaks (Bagnis, 1987;Bagnis et al., 1988; Kohler and Kohler, 1992). In light of this, itseems likely that BHAB dinoflagellates associate with surfacesfacultatively and their distribution among benthic substrates ishighly transitory (Bomber, 1985; Bomber et al., 1988; Nakaharaet al., 1996; Cruz-Rivera and Villareal, 2006; Totti et al., 2010).

While this study champions the use of fiberglass screens as theartificial substrate of choice, it is likely that any new, clean surfacemay work as well. However, it is important that surface area can bemeasured accurately. Although surfaces like cotton strips, wovenfabric or other complex three dimensional materials can collectBHAB dinoflagellates, their surface areas are not easily quantified.The length of time materials are incubated is another importantconsideration. In this study, an incubation time of 24 h was foundto be ideal for BHAB abundance assessment because Gambierdiscus,Ostreopsis and Prorocentrum abundances reached equilibrium withBHAB cells in the surrounding environment within that timeperiod. Data substantiating this equilibrium were obtained from anexperiment conducted in the back reef environment at Carrie BowCay, Belize. Here, Gambierdiscus, Ostreopsis and Prorocentrum cellsare typically present at a wide range of densities on macrophytes(0–1500, 0–1500, 100–3100 cells g�1, respectively) and in thesediment (100–200, 100–400, 100–300 cells g�1, respectively;Faust, 2009). This study further showed immigration of Gambier-

discus, Ostreopsis and Prorocentrum cells to newly introducedscreens occurred within 6–12 h. After 24 h, cell densities reached aplateau of approximately 200, 205 and 1600 cells 100 cm�2,respectively, and cell abundances remained relatively stable for

Table 5Results of screens vs. macrophyte (SM) experiments for Prorocentrum cells at sites in Belize during May 2009 and January 2012 and in Malaysia during May 2012. Data are

given as the mean abundances, x (cells 100 cm�2), standard deviation (Std, cells 100 cm�2), minimum (Min), maximum (Max), coefficient of variation (CV, %) and sample size,

n. ‘‘n/a’’ Indicates sample size insufficient to reach N100 threshold.

Site Prorocentrum


x cells

100 cm�2

Std cells

100 cm�2


100 cm�2


Belize 2009

SM1 487 189 211 728 39 2 8 131 48 67 203 37 2 5

SM2 684 133 399 845 19 2 9 298 89 127 415 30 2 9

SM3 6529 1176 5120 7609 18 2 5 1392 660 669 2299 47 2 5

SM4 643 184 470 916 29 2 5 274 145 147 517 53 2 5

SM5 535 138 376 728 26 2 5 59 31 26 108 53 2 5

SM6 6655 2697 1613 9817 41 3 6 2580 431 1972 3063 17 2 6

SM7 2806 343 2348 3311 12 2 6 488 113 302 616 23 2 6

Belize 2012

SM8 2932 753 2358 3784 26 n/a 3 387 136 262 531 35 n/a 3

SM9 635 106 524 735 17 n/a 3 383 133 301 536 35 n/a 3

SM10 2152 241 1925 2405 11 n/a 3 152 18 132 165 12 n/a 3

SM11 1015 358 672 1385 35 n/a 3 144 48 114 200 33 n/a 3

SM12 3557 510 3078 4093 14 n/a 3 728 84 632 787 12 n/a 3

Malaysia 2012

SM13 566 230 339 974 41 2 6 73 35 35 113 49 2 5

SM14 740 277 365 1073 37 2 6 145 50 83 217 35 2 6

SM15 1349 487 576 1912 36 2 6 269 111 131 432 41 2 6

SM16 732 111 595 870 15 2 6 159 84 109 328 53 2 6

SM17 1050 681 437 2329 65 2 6 170 80 87 284 47 2 6

SM18 811 621 312 1897 77 3 5 79 17 66 102 22 2 6


the next 24 h (Fig. 4). The consistent cell densities after 24 h werean indication that immigration and emigration rates had reachedequilibrium. Longer incubation times increased the amount ofcontaminating material on each screen, making subsequent cellsorting and enumeration more difficult. Prior testing of the screensubstrate demonstrated BHAB cells were almost completelysupplanted by epiphytic diatoms and other debris after 48 h.Accordingly, the >4 month incubation times used by Caire et al.(1985, cloth strips) or the �1 month period used by Ellsworth et al.(2013, ceramic tiles) would not be appropriate for the screensampling method.

It is important to note that stabilization of BHAB cell densities at24 h (Fig. 4) was not due the complete utilization of availablesurface area on the screen for colonization by dinoflagellate cells.Rather, abundances on screens incubated for 24 h were propor-tional to ambient cell densities. Where ambient cell abundanceswere higher, the cell abundances on screens reflected this (Tables3–5). For example, in the screens vs. macrophyte and sample sizestudies, the mean abundances of Gambierdiscus reached 35,086cells 100 cm�2 (site SM3), Ostreopsis reached 8149 cells 100 cm�2

(site SM6), and Prorocentrum reached 107,067 cells 100 cm�2 (siteSS1). These abundances were 1–2 orders of magnitude greater thanthose observed in the incubation time study conducted at CarrieBow Cay (Fig. 4). This indicates the available surface area did notlimit cell recruitment.

The data from the screen size experiment has shown area-normalized BHAB abundances were not affected appreciably bythe size of the substrate (within the screen sizes tested). Larger(257 cm2) and smaller screens (166 cm2) yielded the same numberof cells per 100 cm2. Although larger screens may be useful inenvironments with low BHAB cell abundances, overly large screensare not recommended because of the handling difficulties duringsample collection. The screen size used in this study, 166 cm2, wasoperationally defined because it best fit into the wide mouth jarsused to collect samples. Screens larger than the sample jar shouldnot be used because this will necessitate folding or rolling thescreen material, greatly increasing the likelihood that cells will belost during collection. The use of rigid, wide-mouth sample jars isrecommended instead of plastic bags, as the jars do not collapse orchange shape during screen collection and leakage of samples aftercollection is prevented.

The variability of BHAB dinoflagellates observed in this studywas consistent with previous work showing that benthic organ-isms exhibit inherently patchy distributions. Such small scalevariability is the rule in benthic ecology. As a result, large numbersof samples may be necessary to estimate mean abundance withreasonable confidence intervals (Chutter, 1972; Yasumoto et al.,1979; Taylor and Gustavson, 1985; Barbiero et al., 2011; Mavricet al., 2013). Among Ostreopsis and Prorocentrum cells, samplevariability may also be driven by the co-occurrence of species that

Table 6Results of sample size (SS) experiments completed in Belize in May 2009. Data include the abundance of Gambierdiscus cells from screen samples expressed as the mean

abundance, x (cells 100 cm�2), standard deviation (Std, cells 100 cm�2), minimum (Min), maximum (Max), coefficient of variation (CV, %) and sample size, n. ‘‘n/a’’ Indicates

sample size insufficient to reach N100 threshold.

Site x cells 100 cm�2 Std cells 100 cm�2 Min Max CV (%) N100 n

Gambierdiscus

SS1 5027 2847 2264 9617 57 2 9

SS2 13 17 0 47 131 n/a 9

SS3 4 6 0 12 150 n/a 9

SS4 96 69 23 233 71 5 8

SS5 519 156 280 747 30 2 9

SS6 13 18 0 47 138 n/a 7

SS7 39 29 0 93 73 7 9

SS8 104 47 0 163 45 3 9

SS9 127 190 0 584 150 n/a 9

SS10 136 131 0 350 96 9 9

SS11 114 61 23 187 53 4 9

SS12 130 50 93 233 38 2 9

Ostreopsis

SS1 79 106 0 282 133 n/a 9

SS2 8 17 0 47 212 n/a 9

SS3 158 66 47 258 42 2 9

SS4 9 12 0 23 138 n/a 8

SS5 736 190 517 1080 26 2 9

SS6 10 13 0 23 125 n/a 7

SS7 37 31 0 94 86 8 9

SS8 170 88 0 282 52 4 9

SS9 46 24 0 70 52 4 9

SS10 156 69 23 251 44 3 9

SS11 130 102 23 329 79 7 9

SS12 180 66 94 282 37 2 9

Prorocentrum

SS1 25,145 32,162 1033 107,067 128 n/a 9

SS2 626 232 376 1151 37 2 9

SS3 754 222 481 1139 29 2 9

SS4 487 189 211 728 39 2 8

SS5 1751 299 1386 2372 17 2 9

SS6 289 98 164 470 34 2 7

SS7 378 117 164 517 31 2 9

SS8 793 217 376 1080 27 2 9

SS9 410 263 117 939 64 3 9

SS10 527 170 235 798 32 2 9

SS11 540 104 399 705 19 2 9

SS12 467 207 235 775 44 2 9


are primarily planktonic (e.g., Prorocentrum mexicanum, Prorocen-

trum micans, Prorocentrum panamensis; Grzebyk et al., 1998), thosethat are predominantly benthic (e.g., Ostreopsis siamensis, Pro-

rocentrum levis, Prorocentrum lima; Faust, 1993; Faust et al., 2008;Shears and Ross, 2009; Hoppenrath et al., 2013), and those withoverlapping distributions (e.g., Ostreopsis labens, Ostreopsis ovata,Prorocentrum rhathymum; Faust and Morton, 1995; Mangialajoet al., 2011; Pfannkuchen et al., 2012). Sample variability may befurther exacerbated by species that commonly occur in aggregatesor clumps, such as O. ovata (Mangialajo et al., 2011). Overall, ourresults showed 2–8 replicate screens were sufficient to assessBHAB cell abundance at most sites, even when CVs among screensapproached 100% (Table 6). However, when sample variability washigh relative to the mean (CV > 100%), even nine screens wereinsufficient to assess mean abundance. This was most commonwhen data were highly skewed due to a number of replicatescreens with no cells (e.g., GScr site SS3; OMac site SS4; Table 6). Insuch cases more replicates may not improve the estimates of cellabundance. From a management perspective, however, it is notnecessary to add more replicates when there are few target cells,simply to achieve precise estimates of cell abundance.

The artificial substrate method described here has proveneffective in a wide range of habitats, but the method may beoptimized for the BHAB species of interest and the range ofabundances encountered. For species present at moderate to highabundances, the method may be used as it is described here. Inenvironments with very low cell abundance (<100 cells 100 cm�2,<25 cells g�1) the number of cells that are counted in the resultingsample, even after concentration, may be too small to allowsufficient accuracy. However, modifications to the methoddescribed in this study to deal with such low abundances shouldbe considered carefully. For instance, filtering a larger portion ofthe sample volume onto the filter mesh will increase the number ofcells that are preserved for counting, especially if a larger diameterpiece of filter mesh is utilized (e.g., 47 mm). Unfortunately, thelarger volume also increases the amount of contaminating material(other cells, sediment, organic matter, etc.) in the resulting cellsample, making microscopy more difficult and potentiallyrendering the sample virtually uncountable. Another option is toincubate a larger piece of screen (e.g., 250–300 cm2), therebyincreasing the overall number of cells that are collected. However,the larger screen requires a larger sample jar in order to limit loss ofcells from the screen and is more difficult to handle underwater(see above). A third option is to count a larger portion of thepreserved cell sample, although this approach, too, increases theamount of contaminating material in the settled sample. Perhapsthe best solution is to use a finer mesh sieve in preparation forfiltering the sample onto the 20 mm mesh. The addition of this stepmay reduce the amount of contaminating material such that alarger aliquot may be settled for counting. As a result of this study,a pre-filtration step using a 150 mm sieve has been incorporatedinto our standardized sampling protocol.

Considering the current limitations of toxin-based monitoringmethods for BHABs (Lehane and Lewis, 2000; Litaker et al., 2010),cell-based monitoring offers a relatively low cost alternative forassessing the risks of BHABs and mitigating their effects onhumans and coastal environments (Tester et al., 2013). The resultsof this study show the artificial substrate method is directlyapplicable for monitoring Gambierdiscus populations in ciguateraendemic areas. An extensive literature review by Litaker et al.(2010, Fig. 3) concluded that �1000–10,000 cells g�1 represents alevel of concern for potential CFP events. Using the screens vs.macrophyte relationship from Fig. 7D, this abundance equates to5000–53,000 cells 100 cm�2 using the screen method. DenseOstreopsis blooms occur at concentrations >50,000 cells g�1

(Tichadou et al., 2010; Totti et al., 2010; Pfannkuchen et al.,

2012) so a reasonable threshold level of concern for Ostreopsis maybe 10-fold fewer cells, or �5000 cells g�1. Similarly, for Prorocen-

trum lima a threshold of �1000 cells g�1 may represent a level ofconcern for diarrheic shellfish poisoning (DSP) events (Foden et al.,2005; CEFAS, 2012). Using our regression results, these abun-dances are equivalent to 20,000 and 3000 cells 100 cm�2 forOstreopsis and Prorocentrum, respectively (Figs. 8D and 9D).

The data presented here demonstrate the efficacy of the screenmethod for collecting BHAB dinoflagellates as a first step for cell-based monitoring. Because species may vary in toxicity, specieslevel identification using sensitive qPCR analysis is the next step ina monitoring protocol. Subsequently species-specific toxicityinformation can be combined with the taxa and abundance datafrom molecular assays to provide a more refined risk assessment ofthe sampling location. Other advantages of artificial substratesover traditional macrophyte-based collection include: (1) astandardized unit of measurement (cells cm�2 or 100 cm�2) tonormalize BHAB densities from studies conducted globally across avariety of habitats; (2) elimination of destructive macrophytesampling; (3) cleaner samples to facilitate counting and molecularanalysis; (4) elimination of difficulties associated with discontin-uous distribution and/or availability of specific macrophytes; and(5) elimination of issues involving dinoflagellate-macroalgaepreference or macrophyte grazing by fish and other fauna. Themajor disadvantage of the artificial substrate method is that itrequires two trips to each sampling site, one to deploy the screensand the second to retrieve them. Because of the numerousadvantages of the artificial substrate method over naturalsubstrate collection, we advocate the screen sampling methodas a tool for cell-based monitoring efforts to better predict andmitigate outbreaks of CFP, diarrheic shellfish poisoning andOstreopsis-associated maladies in endemic areas.

Acknowledgments

Funding was provided by the UNESCO IOC-Yeosu Project ofKorea. The International Training Workshop on the Ecology andTaxonomy of Benthic Marine Dinoflagellates held 21–31 May 2012in Pulau Sibu and the Universiti Kebangsaan, Malaysia included afield and laboratory training component and was inspired by theoutcome of SCOR and IOC/UNESCO’s GEOHAB: HABs in BenthicSystems Open Science Meeting in Honolulu, HI in June 2010. Partialfunding was provided as an award from the National Oceanic andAtmospheric Administration’s Ecology and Oceanography ofHarmful Algal Blooms program (ECOHAB contribution number#764). Dr. M. Toscano, Smithsonian Institution, kindly providedpartial support for the February 2012 sampling at Carrie Bow Cay,Belize. We thank the dedicated and hardworking Malaysianworkshop participants including (in alphabetical order). GraceAbdala, Kieng Soon Hii, Lu Song Hui, Grace Joy Chin Wei Lie, JuneMoh, Chi-Thoi Nguyen, Giang Tuong Ngoc Nguyen, Ngoc LamNguyen, Arief Rachman, Gan Hui Shan, Mark Skinner, Toh Hii-Tan,Thamrin Thamrin, Ho Van The, Hikmah Thoha, Giang Tuong,Pradem Uttayarnmanee, Noime Walican and Hua Zhang. Also, wehad the excellent assistance of Gan Jia Cheng, Gan Hui Shan, MimiNora Mansor and Zuhaimi Bin Samat from the UniversitiKebangsaan, Malaysia who facilitated the logistics and field workto make this workshop possible. We gratefully acknowledge thehelpful comments from anonymous reviewers that improved tothis manuscript.[SS]

References

Accoroni, S., Romagnoli, T., Pichierri, S., Colombo, F., Totti, C., 2012. Morphometricanalysis of Ostreopsis cf. ovata cells in relation to environmental conditions andbloom phases. Harmful Algae 19, 15–22.


http://refhub.elsevier.com/S1568-9883(14)00111-5/sbref0005



Aligizaki, K., Nikolaidis, G., Katikou, P., Baxevanis, A.D., Abatzopoulos, T.J., 2009.Potentially toxic epiphytic Prorocentrum (Dinophyceae) species in Greek coastalwaters. Harmful Algae 8, 299–311.

Bagnis, R., 1987. Ciguatera fish poisoning: an objective witness of the coral reefstress. In: Human Impacts on Coral Reefs: Facts and Recommendations.Antenne Museum, French Polynesia, pp. 241–253.

Bagnis, R., Bennet, J., Barsinas, M., Drollet, J.H., Jacquet, G., Legrand, A.M., Cruchet,P.H., Pascal, H., 1988. Correlation between ciguateric fish and damage to reefs inthe Gambier Islands (French Polynesia). In: Choat, J.H., Barnes, D., Borowitzka,M.A., Coll, J.C., Davies, P.J., Flood, P., Hatcher, B.G., Hopley, D., Hutchings, P.A.,Kinsey, D., Orme, G.R., Pichon, M., Sale, P.F., Sammarco, P., Wallace, C.C.,Wilkinson, C., Wolanski, E., Bellwood, O. (Eds.), Proceedings of the 6th Interna-tional Coral Reef Symposium, Contributed Papers, 6th International Coral ReefSymposium Executive Committee, vol. 2, Townsville, Australia, pp. 195–200.

Ballantine, D.L., Tosteson, T.R., Bardales, A.T., 1988. Population dynamics andtoxicity of natural populations of benthic dinoflagellates in southwesternPuerto Rico. J. Exp. Mar. Biol. Ecol. 119, 201–212.

Barbiero, D.C., Macedo, I.M., Mais, B., Zalmon, I.R., 2011. Comparative study of theestimated sample size for benthic intertidal species and communities. Lat. Am.J. Aquat. Res. 39, 93–102.

Bomber, J.W., 1985. Ecological studies of benthic dinoflagellates associated withciguatera from the Florida Keys. M.S. ThesisFlorida Institute of Technology,Melbourne, FL.

Bomber, J.W., Aikman, K.E., 1989. The ciguatera dinoflagellates. Biol. Oceanogr. 6,291–311.

Bomber, J.W., Morton, S.L., Babinchak, J.A., Norris, D.R., Morton, J.G., 1988. Epiphyticdinoflagellates of drift algae – another toxigenic community in the ciguaterafood chain. Bull. Mar. Sci. 43, 204–214.

Bomber, J.W., Norris, D.R., Mitchell, L.E., 1985. Benthic dinoflagellates associatedwith ciguatera from the Florida Keys. II. Temporal, spatial and substrateheterogeneity of Prorocentrum lima. In: Anderson, D.M., White, A.W., Baden,D.G. (Eds.), Toxic Dinoflagellates. Elsevier Scientific, Amsterdam, pp. 45–50.

Caire, J.F., Raymond, A., Bagnis, R., 1985. Ciguatera: study of the setting up and theevolution of Gambierdiscus toxicus population on an artificial substrate intro-duced in an atoll lagoon with follow up of associated environmental factors. In:Delsalle, B., Galzin, R., Salvat, B. (Eds.), Proceedings of the Fifth InternationalCoral Reef Congress: French Polynesian Coral Reefs, vol. 1, Tahiti, 27 May–01June, pp. 429–435.

Carlson, R.D., 1984. Distribution, periodicity and culture of benthic/epiphyticdinoflagellates in a ciguatera endemic region of the Caribbean. PhD Disserta-tionDepartment of Botany in the Graduate School, Southern Illinois Universityat Carbondale, Carbondale, IL.

Centre for Environment, Fisheries and Aquaculture Science (CEFAS), 2012.In:Annual report on the results of the Biotoxin and Phytoplankton Official ControlMonitoring Programmes for England and Wales – 2012. Lowestoft, UK, http://www.food.gov.uk/multimedia/pdfs/annual-report-dec12.pdf (accessed03.03.14).

Chinain, M., Germain, M., Deparis, X., Pauillac, S., Legrand, A.M., 1999. Seasonalabundance and toxicity of the dinoflagellate Gambierdiscus spp. (Dinophyceae),the causative agent of ciguatera in Tahiti, French Polynesia. Mar. Biol. 135,259–267.

Chutter, F.M., 1972. Reappraisal of Needhamand Usinger’s data on the variability ofa stream fauna when sampled with a Surber sampler. Limnol. Oceanogr. 17,139–141.

Cohu, S., Thibaut, T., Mangialajo, L., Labat, J.P., Passafiume, O., Blanfune, A., Simon, N.,Cottalorda, J.M., Lemee, R., 2011. Occurrence of the toxic dinoflagellate Ostreop-sis cf. ovata in relation with environmental factors in Monaco (NW Mediterra-nean). Mar. Pollut. Bull. 62, 2681–2691.

Cruz-Rivera, E., Villareal, T.A., 2006. Macroalgal palatability and the flux of ciguateratoxins through marine food webs. Harmful Algae 5, 497–525.

Ellsworth, A., Brandt, A., Parsons, M., 2013. Monitoring strategies for the ciguatera-causing benthic dinoflagellate, Gambierdiscus. In: Seventh Symposium onHarmful Algae in the U.S. Sarasota, FL, USA, 27–31 October (abstract).

Faust, M.A., 1993. Prorocentrum belizeanum, Prorocentrum elegans, and Prorocentrumcaribbaeum, three new benthic species (Dinophyceae) from a mangrove island,Twin Cays, Belize. J. Phycol. 29, 100–107.

Faust, M.A., 1994. Three new benthic species of Prorocentrum (Dinophyceae) fromCarrie Bow Cay, Belize: P. sabulosum sp. nov., P. sculptile sp. nov., and P.arenarium sp. nov. J. Phycol. 30, 755–763.

Faust, M.A., 1999. Three new Ostreopsis species (Dinophyceae): O. marinus sp. nov.,O. belizeanus sp. nov., and O. caribbeanus sp. nov. Phycologia 38, 92–99.

Faust, M.A., 2009. Ciguatera-causing dinoflagellates in a coral-reef-mangrove eco-system, Belize. Atoll Research Bulletin No. 569, , pp. 1–30.

Faust, M.A., Litaker, R.W., Vandersea, M.W., Kibler, S.R., Tester, P.A., 2005. Dinofla-gellate diversity and abundance in two Belizean coral-reef mangrove lagoons: Atest of Margalef’s mandala. Atoll Research Bulletin No. 534, , pp. 103–131.

Faust, M.A., Morton, S.L., 1995. Morphology and ecology of the marine dinoflagellateOstreopsis labens sp. nov. (Dinophyceae). J. Phycol. 31, 456–463.

Faust, M.A., Vandersea, M.W., Kibler, S.R., Tester, P.A., Litaker, R.W., 2008. Prorocen-trum levis, a new benthic species (Dinophyceae) from a mangrove island, TwinCays, Belize. J. Phycol. 44, 232–240.

Feltz, C.J., Miller, G.E., 1996. An asymptotic test for the equality of coefficients ofvariation from k populations. Stat. Med. 15, 647–658.

Foden, J., Purdie, D.A., Morris, S., Nascimento, S., 2005. Epiphytic abundance andtoxicity of Prorocentrum lima populations in the Fleet Lagoon, UK. Harmful Algae4, 1063–1074.

Fraga, S., Rodriguez, F., Caillaud, A., Diogene, J., Raho, N., Zapata, M., 2011. Gam-bierdiscus excentricus sp. nov. (Dinophyceae), a benthic toxic dinoflagellate fromthe Canary Islands (NE Atlantic Ocean). Harmful Algae 11, 10–22.

Gayoso, A.M., Dover, S., Morton, S., Busman, M., Moeller, P., Fulco, V.K., Maranda, L.,2002. Diarrhetic shellfish poisoning associated with Prorocentrum lima (Dino-phyceae) in Patagonia Gulfs (Argentina). J. Shellfish Res. 21, 461–463.

GEOHAB, 2012. Core research project: HABs in benthic systems. In: Berdalet, B.,Tester, P., Zingone, A. (Eds.), Global Ecology and Oceanography of Harmful AlgalBlooms. IOC of UNESCO and SCOR, Paris and Newark, 64 pp.

Grzebyk, D., Sako, Y., Berland, B., 1998. Phylogenetic analysis of nine species ofProrocentrum (Dinophyceae) inferred from 18s ribosomal DNA sequences,morphological comparisons, and description of Prorocentrum panamensis, sp.nov. J. Phycol. 34, 1055–1068.

Guiry, M.D., Guiry, G.M., 2012. AlgaeBase. World-wide Electronic PublicationNational University of Ireland, Galway. , http://www.algaebase.org (accessed03.03.2014).

Hoppenrath, M., Chomerat, N., Horiguchi, T., Schweikert, M., Nagahama, Y., Murray,S., 2013. Taxonomy and phylogeny of the benthic Prorocentrum species (Dino-phyceae)—a proposal and review. Harmful Algae 27, 1–28.

Ishikawa, A., Takeichi, Y., Sakamoto, S., Yamaguchi, M., 2011. Year-round occurrenceof the benthic dinoflagellate Gambierdiscus sp. in temperate coastal waters ofJapan.In: Oral Presentation, North Pacific Marine Science Organization (PICES)Meeting. Khabarovsk, Russia, 14–23 October 2011, http://www.pices.int/publications/presentations/PICES-2011/2011-S5/2011-S5.aspx (accessed 03.03.14).

Kibler, S.R., Holland, W.C., Vandersea, M.W., Faust, M.A., Litaker, R.W., Tester, P.A.,2010. A collection method for ciguatera-associated dinoflagellates. In: Ho, K.-C.,Zhou, M.-J., Qi, Y.Z. (Eds.), Harmful Algae 2008. International Society for theStudy of Harmful Algae and Environmental Publication House, Copenhagen andHong Kong, pp. 205–208.

Kohler, S.T., Kohler, C.C., 1992. Dead bleached coral provides new surfaces fordinoflagellates implicated in ciguatera fish poisonings. Environ. Biol. Fish. 35,413–416.

Lehane, L., Lewis, R.J., 2000. Ciguatera: recent advances but the risk remains. Int. J.Food Microbiol. 61, 91–125.

Levasseur, M., Couture, J.-Y., Weise, A.M., Michaud, S., Elbrachter, M., Suave, G.,Bonneau, E., 2003. Pelagic and epiphytic summer distributions of Prorocentrumlima and P. mexicanum at two mussel farms in the Gulf of St. Lawrence, Canada.Aquat. Microb. Ecol. 30, 282–293.

Litaker, R.W., Vandersea, M.W., Faust, M.A., Kibler, S.R., Chinain, M., Holmes, M.J.,Holland, W.C., Tester, P.A., 2009. Taxonomy of Gambierdiscus including four newspecies, Gambierdiscus caribaeus, Gambierdiscus carolinianus, Gambierdiscus car-penteri and Gambierdiscus ruetzleri (Gonyaulacales, Dinophyceae). Phycologia48, 344–390.

Litaker, R.W., Vandersea, M.W., Faust, M.A., Kibler, S.R., Nau, A.W., Holland, W.C.,Chinain, M., Holmes, M.J., Tester, P.A., 2010. Global distribution of ciguateracausing dinoflagellates in the genus Gambierdiscus. Toxicon 56, 711–730.

Lobel, P.S., Anderson, D.M., Durand-Clement, M., 1988. Assessment of ciguateradinoflagellate populations – sample variability and algal substrate selection.Biol. Bull. 175, 94–101.

Mangialajo, L., Ganzin, N., Accoroni, S., Asnaghi, V., Blanfune, A., Cabrini, M.,Cattaneo-Vietti, R., Chavanon, F., Chiantore, M., Cohu, S., Costa, E., Fornasaro,D., Grossel, H., Marco-Miralles, F., Maso, M., Rene, A., Rossi, A.M., Sala, M.M.,Thibaut, T., Totti, C., Vila, M., Lemee, R., 2011. Trends in Ostreopsis proliferationalong the Northern Mediterranean coasts. Toxicon 57, 408–420.

Marasigan, A.N., Tamse, A.F., Fukuyo, Y., 2001. Prorocentrum (Prorocentrales: Dino-phyceae) populations on seagrass-blade surface in Taklong Island, GuimarasProvince, Philippines. Plankton Biol. Ecol. 48, 79–84.

Mavric, B., Urbanic, G., Lipej, L., Simboura, N., 2013. Influence of sample size onecological status assessment using marine benthic invertebrate-based indices.Mar. Ecol. 34, 72–79, http://dx.doi.org/10.1111/j.1439-0485.2012.00526.x.

Miller, G.E., 1991. Asymptotic test statistics for coefficients of variation. Commun.Stat. Theory Methods 20, 3351–3363.

Murray, S., Ip, C.L.C., Moore, R., Nagahama, Y., Fukuyo, Y., 2009. Are Prorocentroiddinoflagellates monophyletic? A study of 25 species based on nuclear andmitochondrial genes. Protist 160, 245–264.

Nagahama, Y., Murray, S., Tomaru, A., Fukuyo, Y., 2011. Species boundaries in thetoxic dinoflagellate Prorocentrum lima (Dinophyceae, Prorocentrales), based onmorphological and phylogenetic characters. J. Phycol. 47, 178–189.

Nakahara, H., Sakami, T., Chinain, M., Ishida, Y., 1996. The role of macroalgae inepiphytism of the toxic dinoflagellate Gambierdiscus toxicus (Dinophyceae).Phycol. Res. 44, 113–117.

Nishimura, T., Sato, S., Tawong, W., Sakanari, H., Uehara, K., Shah, M.R., Suda, S.,Yasumoto, T., Taira, Y., Yamaguchi, H., Adachi, M., 2013. Genetic diversity anddistribution of the ciguatera-causing dinoflagellate Gambierdiscus spp. (Dino-phyceae) in coastal areas of Japan. PLoS ONE 8 (4) e60882, http://dx.doi.org/10.1371/journal.pone.0060882.

Parsons, M.L., Settlemier, C.J., Ballauer, J.M., 2011. An examination of the epiphyticnature of Gambierdiscus toxicus, a dinoflagellate involved in ciguatera fishpoisoning. Harmful Algae 10 (6) 598–605, http://dx.doi.org/10.1016/j.hal.2011.04.011.

Penna, A., Fraga, S., Battocchi, C., Casabianca, S., Giacobbe, M.G., Riobo, P., Vernesi, C.,2010. A phylogeographical study of the toxic benthic dinoflagellate genusOstreopsis Schmidt. J. Biogeogr. 37, 830–841.

Perini, F., Casabianca, A., Battocchi, C., Accoroni, S., Totti, C., Penna, A., 2011. Newapproach using the real-time PCR method for estimation of the toxic marine










































http://www.food.gov.uk/multimedia/pdfs/annual-report-dec12.pdf

http://www.food.gov.uk/multimedia/pdfs/annual-report-dec12.pdf



















































http://www.algaebase.org/



http://www.pices.int/publications/presentations/PICES-2011/2011-S5/2011-S5.aspx

http://www.pices.int/publications/presentations/PICES-2011/2011-S5/2011-S5.aspx



























http://dx.doi.org/10.1111/j.1439-0485.2012.00526.x












http://dx.doi.org/10.1371/journal.pone.0060882






dinoflagellate Ostreopsis cf. ovata in marine environment. PLoS ONE 6 (3)e17699, http://dx.doi.org/10.1371/journal.pone.0017699.

Pfannkuchen, M., Godrijan, J., Pfannkuchen, D.M., Ivesa, L., Kruzic, P., Ciminiello, P.,Dell’Aversano, C., Dello Iacovo, E., Fattorusso, E., Forino, M., Tartaglione, L.,Godrijan, M., 2012. Toxin-producing Ostreopsis cf. ovata are likely to bloomundetected along coastal areas. Environ. Sci. Technol. 46, 5574–5582.

Shears, N.T., Ross, P.M., 2009. Blooms of benthic dinoflagellates of the genusOstreopsis; an increasing and ecologically important phenomenon on temperatereefs in New Zealand and worldwide. Harmful Algae 8, 916–925.

Sokal, R.R., Rohlf, F.J., 1995. Biometry. W.H. Freeman and Company, New York, NY.Taira, K., Saadon, M.N.B., Kitagawa, S., Yanagi, T., 1996. Observation of temperature

and velocity in the coastal water off Kuala Terengganu, Malaysia. J. Oceanogr.52, 251–257.

Tan, T.-H., Lim, P.-T., Mujahid, A., Usup, G., Leaw, C.-P., 2013. Benthic harmfuldinoflagellate assemblages in a fringing reef of Sampadi Island, Sarawak,Malaysia. Mar. Res. Indonesia 38, 11–21.

Taylor, F.J.R., Gustavson, M.S., 1985. An underwater survey of the organism chieflyresponsible for ‘‘ciguatera’’ fish poisoning in the eastern Caribbean region: thebenthic dinoflagellate Gambierdiscus toxicus. In: Stefanon, A., Flemming, N.J.(Eds.), Proceedings of the Seventh International Diving Science Symposium.Padova University, Padova, Italy, pp. 95–111.

Tester, P.A., Kibler, S.R., Vandersea, M.W., Holland, W.C., Litaker, R.W., 2010.Sampling methodology for benthic dinoflagellates. In: Oral Presentation. GEO-HAB Open Science Meeting on the Core Research Project: HABs in BenthicSystems, Honolulu, Hawaii, USA, UNESCO-IOC SCOR, 21–23 June.

Tester, P.A., Vandersea, M.W., Buckel, C.A., Kibler, S.R., Holland, W.C., Davenport,E.D., Clark, R.D., Edwards, K.F., Taylor, J.C., Vander Pluym, J.L., Hickerson, E.L.,Litaker, R.W., 2013. Gambierdiscus (Dinophyceae) species diversity in the FlowerGarden Banks National marine Sanctuary, Northern Gulf of Mexico, USA.Harmful Algae 29, 1–9.

Throndsen, J., 1995. Estimating cell numbers. In: Hallegraeff, G.M., Anderson, D.M.,Cembella, A.D., Enevoldsen, H.O. (Eds.), IOC Manuals and Guides No. 33: Manualon Harmful Marine Microalgae. UNESCO, Paris, pp. 63–80.

Tichadou, L., Glaizal, M., Armengaud, A., Grossel, H., Lemee, R., Kantin, R., Lasalle, J.L.,Drouet, G., Rambaud, L., Malfait, P., de Haro, L., 2010. Health impact of unicel-lular algae of the Ostreopsis genus blooms in the Mediterranean Sea: experienceof the French Mediterranean coast surveillance network from 2006 to 2009.Clin. Toxicol. 48, 839–844.

Tindall, D.R., Morton, S.L., 1998. Community dynamics and physiology of epiphytic/benthic dinoflagellates associated with ciguatera. In: Anderson, D.M., Cembella,A.D. (Eds.), Physiological Ecology of Harmful Algal Blooms. Springer-Verlag,Berlin, pp. 293–313.

Totti, C., Accoroni, S., Cerino, F., Cucchiari, E., Romagnoli, T., 2010. Ostreopsis ovatabloom along the Conero Riviera (northern Adriatic Sea): relationships withenvironmental conditions and substrata. Harmful Algae 9, 233–239.

Utermohl, H., 1958. Zur Vervollkommnung der quantitativen Phytoplankton-Meth-odik. Mitteilungen/Int. Vereinigung theor. angew. Limnol. 9, 1–38.

Vandersea, M.W., Kibler, S.R., Holland, W.C., Tester, P.A., Schultz, T.F., Faust, M.A.,Holmes, M.J., Chinain, M., Litaker, R.W., 2012. Development of semi-quantita-tive PCR assays for the detection and enumeration of Gambierdiscus species(Gonyaulacales, Dinophyceae). J. Phycol. 48, 902–915.

Vila, M., Garces, E., Maso, M., 2001. Potentially toxic epiphytic dinoflagellateassemblages on macroalgae in the NW Mediterranean. Aquat. Microb. Ecol.26, 51–60.

Wan-Jean, L., 2005. Aspects of the ecology of intertidal barnacles on the shores ofpeninsular Malaysia and Singapore. MS Thesis Department of BiologicalSciences, National University of Singapore, Singapore.

Weisstein, E.W., 2013. Steinmetz Solid. From MathWorld—A Wolfram Web Resource,http://mathworld.wolfram.com/SteinmetzSolid.html (accessed 03.03.2014).

Yasumoto, T., Inoue, A., Bagnis, R., 1979. Ecological survey of a toxic dinoflagellateassociated with ciguatera. In: Taylor, D.L., Seliger, H.H. (Eds.), Toxic Dinoflagel-late Blooms. Elsevier North Holland, Inc., New York, pp. 221–224.

Yasumoto, T., Inoue, A., Ochi, T., Fujimoto, K., Oshima, Y., Fukuyo, T., Adachi, R.,Bagnis, R., 1980. Environmental studies on a toxic dinoflagellate responsible forciguatera. Bull. Jpn. Soc. Sci. Fish. 46, 1397–1404.

Zar, J.H., 1996. Biostatistical Analysis. Prentice Hall, Upper Saddle River, New Jersey.



















































http://mathworld.wolfram.com/SteinmetzSolid.html







Corrigendum

Corrigendum to ‘‘Sampling harmful benthic dinoflagellates:Comparison of artificial and natural substrate methods’’[Harmful Algae 39 (2014) 8–25]

Patricia A. Tester a,*, Steven R. Kibler a, William C. Holland a, Gires Usup b,Mark W. Vandersea a, Chui Pin Leaw c, Po Teen Lim c, Jacob Larsen d,Normawaty Mohammad-Noor e, Maria A. Faust f, R. Wayne Litaker a

a National Oceanic and Atmospheric Administration, National Ocean Service, National Centers for Coastal Ocean Science, Center for Coastal Fisheries and

Habitat Research, 101 Pivers Island Road, Beaufort, NC 28516, USAb Program Sains Laut, Pusat Pengajian Sains Sekitaran dan Sumber Alam, Fakulti Sains dan Teknologi, Universiti Kebangsaan Malaysia, 43600 Bangi, Selangor,

Malaysiac Institute of Ocean and Earth Sciences, University of Malaya, 16310 Bachok, Kelantan, Malaysiad IOC Science and Communication Centre on Harmful Algae, Department of Phycology and Mycology, Øster Farimagsgade 2D, DK-1353 Copenhagen K,

Denmarke Institute of Oceanography and Maritime Studies, Kulliyyah of Science, International Islamic University Malaysia, Jalan Sultan Ahmad Shah, Bandar Indera

Mahkota, 25200 Kuantan, Pahang, Malaysiaf Department of Botany, National Museum of Natural History, Smithsonian Institution, Washington, DC 20560, USA

The authors regret that in the above paper the authors’ affiliations were not complete. These have been published correctly above.The authors would like to apologise for any inconvenience caused.

Harmful Algae 39 (2014) 374

DOI of original article: http://dx.doi.org/10.1016/j.hal.2014.06.009

* Corresponding author at: Center for Coastal Fisheries and Habitat Research, National Centers for Coastal Ocean Science, National Ocean Service, NOAA, 101 Pivers Island

Road, Beaufort, NC 28516, USA. Tel.: +1 252 728 8792; fax: +1 252 728 4537.

E-mail addresses: [email protected], [email protected] (P.A. Tester).

Contents lists available at ScienceDirect

Harmful Algae

jo u rn al h om epag e: ww w.els evier .c o m/lo cat e/ha l


1568-9883/� 2014 Elsevier B.V. All rights reserved.







http://www.sciencedirect.com/science/journal/15689883

www.elsevier.com/locate/hal


Harmful Algae HA 39 -25 merged.pdfMahkota, g 25200 Kuantan, Pahang, Malaysia Department of Botany,...

Documents

Transcript of Harmful Algae HA 39 -25 merged.pdfMahkota, g 25200 Kuantan, Pahang, Malaysia Department of Botany,...