193

Ciguatera in the Introduced Fish Cephalopholis argus (Serranidae)in Hawai‘i and Implications for Fishery Management1

Jan Dierking2,3 and Cara E. Campora4

Abstract: The Peacock gouper (Cephalopholis argus) was introduced to Hawai‘iin 1956 to establish a new fishery. It has become abundant, but the fishery faileddue to concerns about ciguatera fish poisoning, a neurological disease in humanscaused by ingestion of fish containing ciguatoxin. The aim of this study was toprovide better understanding of geographic patterns of ciguatoxicity in C. argusand of the correlation of toxicity with morphometric characters of this species,with the goal to assess the possibility of a safe fishery. Overall, 18.2% of C. argusspecimens from sites around O‘ahu and Hawai‘i Island contained ciguatoxin inconcentrations potentially harmful to humans. This was higher than the rate ofoccurrence in Hawaiian reef fishes in general, and on the scale of ciguatoxicityin species banned from sale in fish markets. Toxicity was high around both ana-lyzed islands. However, toxic individuals were significantly less common aroundO‘ahu than around Hawai‘i Island (8% versus 24%). Regular geographic pat-terns in toxicity within islands (e.g., gradients along coastlines) were not present,and variability in toxicity within each sample site was high. Toxicity was signifi-cantly but weakly positively correlated with C. argus length but not with fishcondition (measured by length at weight). In conclusion, high prevalence of tox-ic individuals, variability in toxicity on all analyzed spatial scales, and low explan-atory power of morphometric characters make the avoidance of ciguatoxic C.argus individuals difficult. A safe fishery for this species in Hawai‘i thereforedoes not appear feasible at present.

Due to their geographic isolation, themain Hawaiian Islands have reef fish com-munities characterized by the lack or limitedpresence of fish families with short to me-dium larval durations and late maturity,which include the commercially importantsnappers (Lutjanidae) and groupers (Serrani-dae), and by high rates of endemism in theextant fish families (Randall 1987). In the1950s, this situation led to the instigationof an introduction program of snappers,groupers, and emperors (Lethrinidae) thathad the objective to create new fisheries. Of12 species in the program, only two snapperspecies and the Peacock grouper (Cephalopho-lis argus Bloch & Schneider) (common namein Hawai‘i: roi), of which 2,385 individualsfrom Moorea (French Polynesia) were ini-tially released, became established (Oda andParrish 1981). After an initial time lag, C.argus populations have grown strongly sincethe 1980s, and the species now surpasses thebiomass of all other nearshore reef fish pred-

Pacific Science (2009), vol. 63, no. 2:193–204: 2009 by University of Hawai‘i PressAll rights reserved

1 This study was funded by the National Oceanicand Atmospheric Administration under award no.NA05NOS4261157 to the Hawai‘i Coral Reef Initiativeand received additional support from the Hawai‘i Coop-erative Fishery Research Unit. Personal financial supportfor J.D. came in part from the Fulbright Association.Manuscript accepted 5 June 2008.

2 Corresponding author.3 Department of Zoology, University of Hawai‘i at

Mānoa, 2538 McCarthy, The Mall, Honolulu, Hawai‘i96822. Current address: Station Marine d’Endoume,Centre d’Océanologie de Marseille, UMR CNRS 6540,Université de la Méditerranée, Rue de la Batterie desLions, 13007 Marseille, France (phone: þ33 (0)491041637; fax: þ33 491 041635; e-mail: jan.dierking@univmed.fr).

4 Department of Cell and Molecular Biology, Univer-sity of Hawai‘i at Mānoa, 1960 East-West Road, Hono-lulu, Hawai‘i 96822.

ators in the main Hawaiian Islands combined.A fishery nevertheless did not develop after C.argus caused a number of ciguatera fish poi-soning incidents in Hawai‘i and was rejectedby consumers thereafter (Dierking 2007).

Ciguatera fish poisoning is a neurologicaldisease that causes severe clinical symptomsin humans (Palafox and Buenconsejo-Lum2001). It is caused by ingestion of ciguatoxin,a natural polyether toxin that is produced bythe benthic dinoflagellate Gambierdiscus toxi-cus (Yasumoto et al. 1980), as well as otherspecies of the genus Gambierdiscus. Cigua-toxin enters the marine food chain whenherbivores consume benthic macroalgae withattached G. toxicus. After ingestion, ciguatoxinis stored in body tissues and can subsequentlybe transferred to higher trophic levels of thefood web when predators prey on toxic herbi-vores (Banner 1974). Ciguatera is confined tothe subtropics and tropics, where it has im-portant public health and economic implica-tions (Lewis 2001). The annual number ofreported incidents in humans ranges from25,000 to 50,000 worldwide (Lehane andLewis 2000) and from 35 to 70 in Hawai‘i(Bruno and Effler 2001), and important fish-eries have been closed due to ciguatera out-breaks (De Sylva 1994).

Due to consumer concerns about ciguaterafish poisoning, C. argus is of no commercialvalue in Hawai‘i. Moreover, in contrast tooriginally envisioned benefits of the intro-duction, the large quantity of prey fishesconsumed by C. argus populations is todayconsidered a potential threat to native fishesand fisheries (Earle 2005). In contrast, mostgroupers worldwide are in decline due totheir value as food fish and their vulnerabilityto fishing pressure, and are considered impor-tant in maintaining ecosystem functioning(Morris et al. 2000). Cephalopholis argus fallsunder the latter pattern in several of its nativehabitats (Heemstra and Randall 1993). It isoverfished in Guam, where it is not cigua-toxic, and its wholesale value is $20/kg in theAsian live reef fish trade (Lee and Sadovy1998). Considering this situation, better un-derstanding of ciguatera in C. argus in Ha-wai‘i may have the potential to lead to a safefishery in the future. Questions with direct

potential applications include How ciguatoxicare individuals of C. argus in Hawai‘i, and,even if overall toxicity is high, are certain in-dividuals (such as those below a certain sizethreshold or from certain areas) predictablysafe for consumption?

However, few studies (and none on agrouper) have explicitly looked at the roleof the geographic origin or morphometriccharacters in ciguatoxicity, and despite recentadvances in the understanding of the biologyof ciguatera, many questions remain (Lehaneand Lewis 2000). In this situation, the aim ofour study was to assess the feasibility of a C.argus fishery in the main Hawaiian Islandsand to enhance general understanding of ci-guatera by (1) screening C. argus populationsfor ciguatoxin concentrations, (2) analyzinggeographic patterns in C. argus toxicity inthe main Hawaiian Islands, and (3) analyzingthe association of ciguatoxin concentrationwith morphometric characters.

materials and methods

Site Selection and Sampling

A total of 291 specimens of C. argus from sixsites around the island of O‘ahu and 11 sitesalong the west coast of the island of Hawai‘i(Figure 1) (estimated site extent 10,000 m2)was sampled by spearfishing on scuba in July2003 (O‘ahu, n ¼ 106; Hawai‘i, n ¼ 185) (nospecimens were caught at site KA-W07).Specimens were placed on ice after the divesand transported to the laboratory within 2hr. There, standard length (SL) and totallength (TL) of each specimen were recordedto the nearest millimeter, and wet weight(W ) (O‘ahu samples only, due to scale mal-functioning during Hawai‘i Island fieldwork)to the nearest 5 g. Sex was determined basedon visual inspection of gonads. Mean SLranged from 13.2 to 44.0 cm and W from0.185 to 2.100 kg. Muscle tissue samples(@0.5 cm3) from the head, body, and tail re-gion of each specimen were frozen at �20�Cfor ciguatoxin analysis.

O‘ahu is the human population center ofthe main Hawaiian Islands and has the high-est number of recreational fishermen of any

194 PACIFIC SCIENCE . April 2009

of the main Hawaiian Islands. Hawai‘i Islandhas a smaller population, but fishing for homeconsumption is still common (Schmitt 2002).Ciguatera fish poisoning incidents are partic-ularly common along the west coasts ofO‘ahu and Hawai‘i Island (Bruno and Effler2001). Cephalopholis argus densities along thelatter coastline are among the highest in themain Hawaiian Islands, with on average 7.2individuals/1,000 m2 in coral reef habitats(Dierking 2007). This coast has also beenthe recent focus of resource management ef-forts to sustain reef fish populations (Tissotet al. 2004).

Fish Condition

Fish condition factors are based on the as-sumption that heavier fish of a given length

are in better condition (Bolger and Connolly1989). The measure of fish condition used inthis study was the relative condition factorKn ¼ W=aTLn, where a and n are the inter-cept and slope of the exponential form ofthe length-weight equation of the populationof interest. The length-weight equation wasdetermined by least-squares regression oflog10 W versus log10 TL. Due to the lack ofW data for Hawai‘i Island samples, the calcu-lation of Kn was limited to O‘ahu samples.

Ciguatoxin Analysis

Ciguatoxin concentrations in C. argus tissuewere determined with the Membrane Immu-nobead Assay, which is based on the specificbinding of a monoclonal antibody (MAb-ciguatoxin) to the ciguatoxin molecule (Ho-

Figure 1. Map of the main Hawaiian Islands (inset) and the islands of O‘ahu (A) and Hawai‘i (B). Black dots marksample sites (OA, O‘ahu; KA, Hawai‘i Island; W, West; S, South; E, East).

Ciguatera in Cephalopholis argus in Hawai‘i . Dierking and Campora 195

kama et al. 1998). The Membrane Immuno-bead Assay is rapid and affordable, thus allow-ing large-scale screenings of fish populationsthat are difficult to accomplish with morecost- and time-intensive mass spectrometry(Lewis et al. 1999) or tissue culture (BotteinDechraoui et al. 2005) approaches. TheMembrane Immunobead Assay is the methodused by the Hawai‘i Department of Healthfor the validation of ciguatoxicity in fish im-plicated in ciguatera fish poisoning incidents(Hokama and Ebesu 2001).

Membrane Immunobead Assay proce-dures followed Hokama et al. (1998), withsmall modifications as described by Dierking(2007). Briefly, Membrane Immunobead As-say scores were obtained by visual compari-son of test stick colorations with a standardscale (Table 1). Because muscle tissue fromdifferent body regions can vary in concentra-tions (Kimura et al. 1982), test scores (eachwith a possible range of 0–2 [Table 1]) ofwhite muscle tissue from the head, body, andtail of each specimen were added to an overallfish score (possible range 0–6), which was in-terpreted regarding the toxicity and safety ofthe fish for human consumption (Table 2).

Validation of Membrane Immunobead AssayResults with the Neuroblastoma Cell Bioassay

Because the Membrane Immunobead Assayhas not routinely been used in prior pub-lished studies, a subset ðn ¼ 30Þ of C. argusmuscle tissue samples (15 samples that tested

negative using the Membrane ImmunobeadAssay and 15 samples that tested borderline½n ¼ 8� or positive ½n ¼ 7�) was retested withthe accepted Neuroblastoma cell bioassay forthe detection of ciguatoxin (Manger et al.1995) (for prior applications see, e.g., BotteinDechraoui et al. 2005).

The specific Neuroblastoma cell assaymethodology followed Campora et al.(2008). In short, the presence of ciguatoxinin muscle tissue was determined by compar-ing the survival rate of neuroblastoma cellsin treatment wells containing fish extractwith that in control wells without extract,with lowered survival in treatment wells indi-cating cytotoxicity. Differences between con-trol and treatment wells were assessed byone-way analysis of variance (ANOVA) fol-lowed by Tukey’s test for multiple compari-sons. Samples with significantly lower cellsurvival in treatment wells were scored as‘‘positive,’’ samples with posttest criticalvalues above 3.50 but lack of significanceðP > :05Þ were scored as ‘‘borderline,’’ andall other samples were scored as ‘‘negative.’’

The agreement of the proportion of nega-tive, borderline, and positive results obtainedwith the Membrane Immunobead Assay andthe Neuroblastoma cell assay was then as-sessed with a Chi-square test. In addition,the correlation of results (with test scorestranslated to numerical values: negative, 0;borderline, 0.5; positive, 1) was assessed by

TABLE 1

Conversion Table for Determination of MonoclonalImmunobead Assay Scores and Ciguatoxin

Concentrations from Coloration of the MonoclonalImmunobead Assay Membrane

MembraneColorationa Score

Ciguatoxin(ng/g)

No color 0 0Faint blue 0.5 0.08Distinct blue 1 0.16Intense blue 2 0.32

a For a graphic display of membrane colorations for a widerange of ciguatoxin concentrations obtained by running Mono-clonal Immunobead Assays on standards of known concentration,see Hokama et al. (1998).

TABLE 2

Conversion Table for Interpretation of Overall FishScores

Fish ScoreRangea

ScoreClassb Interpretation

2.0–4.0 Positive Unsafe (incident possible)>4.0 Strongly

positiveUnsafe (incident likely)

a Overall fish scores are the sum of the three individualscores obtained from the head, body, and tail of each specimen(see Table 1).

b Boundaries and interpretations were previously establishedby Hokama et al. (1998) based on Monoclonal Immunobead As-says of fish that had caused ciguatera fish poisoning and of fishthat were consumed without causing incidents.

196 PACIFIC SCIENCE . April 2009

calculating the Pearson correlation coeffi-cient.

Association of Ciguatoxicity with GeographicFactors and C. argus Morphometry

Differences in the mean toxin concentrationof C. argus from O‘ahu and Hawai‘i Island,and differences in the prevalence of toxicindividuals around the two islands, were as-sessed with a two sample t-test and a Chi-square test, respectively. An analysis of cova-riance (ANCOVA) type general linear modelwith site as explanatory and ciguatoxin scoreas response variable, and C. argus SL as cova-riate, was used for the analysis of the relation-ship of geographic location, C. argus size, andciguatoxicity. Patterns in toxicity along coast-lines were analyzed by comparing side-by-side box plots of toxicity at each site. Finally,the Pearson correlation factor was calculatedto test the association of fish condition andciguatoxicity. Minitab (Version 14, Minitab,Inc., State College, Pennsylvania) was the sta-tistical software for all analyses, with resultsconsidered significant at P < :05.

results

Membrane Immunobead Assay Validation

Using the Membrane Immunobead Assay, of30 C. argus muscle tissue samples, 15 testednegative, 8 borderline, and 7 positive forthe presence of ciguatoxin. Using the Neuro-

blastoma cell assay on the same samples, 20samples tested negative, 5 borderline, and 5positive. Overall, the proportion of negative,borderline, and positive results obtained withthe two tests did not differ significantly (Chi-square analysis, P ¼ :419). Moreover, resultswere strongly and significantly correlated(r ¼ 0:64, P < :001).

Overall Sample Description

The screening of 291 specimens for toxicityprovides a general overview of ciguatera inC. argus in the main Hawaiian Islands.Forty-six percent of specimens tested nega-tive, 36% marginal, 14% positive, and 4%strongly positive for ciguatoxin (Figure 2),which means that 82% of fish were generallysafe and 18% potentially unsafe for humanconsumption (Table 2). The mean fish toxic-ity score of 1.25 fell into the ‘‘marginal’’ scorecategory, which would generally be consid-ered safe for consumption (Table 2). How-ever, variability was large, which is reflectedin the large standard deviation and wide 95%confidence interval for ciguatoxin scores inthe overall sample (Table 3).

Large-Scale Geographic Patterns in CiguatoxinConcentrations

Mean ciguatoxin concentration in C. arguswas significantly lower for O‘ahu than forHawai‘i Island (Table 3) (t-test with samplesite mean as sampling unit [t ¼ 3:82, df ¼

Figure 2. Ciguatoxicity of C. argus in the O‘ahu, Hawai‘i Island, and overall sample (sample sizes in parentheses). Fishin the negative and marginal score classes are considered safe, and fish in the positive and strongly positive classes areconsidered unsafe for human consumption.

Ciguatera in Cephalopholis argus in Hawai‘i . Dierking and Campora 197

10, P < :01]). The prevalence of individualsunsafe for human consumption was alsosignificantly lower (7.5% for O‘ahu and24.3% for Hawai‘i Island) (Chi-square test[n ¼ 291, df ¼ 1, P < :001]) (Figure 2). Boxplots of toxicity at the sample sites groupedin order of their relative geographic position

along coastlines did not reveal clear geo-graphic patterns (Figure 3). In particular,individuals containing ciguatoxin in high con-centrations occurred along all coastlines, andno spatial gradients in toxicity were apparent.

Between-Site Differences and the Role of Lengthand Fish Condition in Ciguatoxicity

Mean ciguatoxin concentration differed sig-nificantly between sites, and C. argus SL andciguatoxin concentration were significantlypositively associated (ANCOVA: F ¼ 2:17;df ¼ 1, 16; site: P ¼ :006; SL: P ¼ :03) (Ta-ble 4). The increase in toxicity with size oc-curred consistently across sites, as indicatedby the nonsignificant interaction of site andSL when the interaction term ‘‘site*C. argusSL’’ was added to the general linear model(F ¼ 0:72, df ¼ 16, P ¼ :77). The proportion

Figure 3. Box plots of the ciguatoxicity (measured by Membrane Immunobead Assay score) in C. argus populations ateach site, in order of the relative position of sites along coastlines (also see Figure 1). (OA, O‘ahu; KA, Hawai‘i Island;sample sizes at each site in parentheses). Boxes represent the interquartile range, horizontal lines through the boxes themedian, and black dots the mean. Outliers are marked by asterisks (*). The dashed horizontal lines mark ciguatoxinscore class boundaries, with fish below the lower boundary (a2) generally safe and those above generally unsafe forconsumption. Fish above the upper boundary (>4) would likely have caused ciguatera incidents.

TABLE 3

Ciguatera in the O‘ahu, Hawai‘i Island, and OverallSample

Region n xG 1 SD 95% CI Range

O‘ahu 106 0.73G 1.07 0–2.8 0–6Hawai‘i I. 185 1.55G 1.48 0–4.4 0–6Overall 291 1.25G 1.40 0–4.0 0–6

Note: Mean toxicity differed significantly between islandsðP < :01Þ. SD, standard deviation; CI, confidence interval.

198 PACIFIC SCIENCE . April 2009

of variation explained by the associationof ciguatoxicity with site and SL was13.7%. Partitioning of variation showed thatbetween-site differences in toxicity explained12.2% and the association of toxicity and SL1.5% of total variation in ciguatoxicity (Table4). Within-site variation therefore repre-sented 87.8% of total variation, of which allbut the proportion explained by fish SL re-mained unexplained by the factors includedin the analysis. The pattern of high variabilityin toxicity despite the significant increase intoxicity with fish length is illustrated byFigure 4. A second morphometric character,fish condition, was not significantly correlatedwith ciguatoxin concentration ðP ¼ :480Þ(Figure 5).

discussion

The good overall agreement of results ob-tained by the Membrane Immunobead Assay

and the Neuroblastoma cell assay (i.e., insig-nificant differences in the proportion ofnegative, borderline, and positive samples½P ¼ :419� and significant correlation of testscores ½P < :001�) for the same subset of 30

Figure 4. The relationship of toxicity (Membrane Immunobead Assay score) and SL in C. argus from O‘ahu (graytriangles) and Hawai‘i Island (black circles), with least squares regression line for the combined datasets. Dashed hor-izontal lines mark score class boundaries, with fish below the lower line generally safe and fish above this line poten-tially unsafe, and those above the upper line highly unsafe for consumption.

TABLE 4

ANCOVA Table of the Main Effects General LinearModel of Site, C. argus SL (Covariate), and Ciguatoxicity

(Dependent Variable)

Source ofVariation df

Sums ofSquares

Adj.Mean

Squaresa F P

VariationExplained

(%)

C. argusSL

1 15.7 8.52 4.74 .030 1.5

Site 16 62.4 3.89 2.17 .006 12.2Error 273 490.7 1.79Total 290 568.7r 2 13.7

a Adjusted mean squares are based on adjusted sums ofsquares calculated with all model terms included in the model.

Ciguatera in Cephalopholis argus in Hawai‘i . Dierking and Campora 199

C. argus specimens indicates that the conclu-sions derived from the Membrane Immuno-bead Assay in this study are valid. Thehigher proportion of samples containingciguatoxin detected by the Membrane Immu-nobead Assay compared with the Neuroblas-toma cell assay suggests that toxicity estimatesin this paper are conservatively high. How-ever, variation introduced by false positivetests would arguably be random and thereforedoes not appear likely to skew results.

Ciguatera in C. argus in the Main HawaiianIslands

The first objective of this study was the gen-eral screening of C. argus populations in themain Hawaiian Islands for ciguatoxicity to as-sess the human health risk of an unlimitedfishery (i.e., no use of spatial closures, slotsizes, or other measures to avoid ciguatoxic

fish). The mean ciguatoxin score of the over-all population was 1.25, which would be gen-erally safe for human consumption (Table 2).However, the variability in ciguatoxicity washigh, and despite the low mean toxicity,18.2% of individuals were too toxic for safehuman consumption (Figure 2). It is im-portant to note that it is unlikely that all un-safe individuals would have caused ciguaterafish poisoning incidents upon consumption,because human susceptibility to ciguatoxinvaries and because the Membrane Immuno-bead Assay scoring scale was designed with abuilt-in safety margin to avoid false negativetests (i.e., classifying toxic fish as edible) (Ho-kama and Ebesu 2001). At the same time, al-though fish that test marginally for ciguatoxin(35.6% of total) are considered safe, frequentconsumption of such individuals could stilllead to ciguatera fish poisoning incidents, be-cause ciguatoxin can bioaccumulate in humantissue (Lehane and Lewis 2000).

Figure 5. Relative condition Kn and ciguatoxicity (Membrane Immunobead Assay score) in C. argus from O‘ahu.Kn ¼ 1 (marked by solid horizontal line) represents the mean body condition of the population.

200 PACIFIC SCIENCE . April 2009

The observed prevalence of toxic individu-als was much higher than in Hawaiian reeffishes overall (Hokama et al. 1998) and onthe scale of that of species considered highrisk or banned from commercial sale dueto ciguatera (e.g., Bottein Dechraoui et al.2005) (Table 5). The latter include thegreater amberjack (Seriola dumerili ) in Ha-wai‘i and great barracuda (Sphyraena barra-cuda) in Florida (De Sylva 1994). Thefinding of higher-than-average toxicity in C.argus agrees well with official Hawai‘i De-partment of Health statistics, which showthat despite the minor importance of thisspecies compared with popular market reeffishes, it caused more ciguatera fish poisoningincidents than any other species in Hawai‘i inrecent years (21 incidents between 1996 and2000, equivalent to 17% of all reportedcases) (Bruno and Effler 2001). Consideringthe severity of ciguatera fish poisoning, alarge-scale fishery for C. argus does not ap-pear feasible.

It should be noted that a commercial kitto test fish for ciguatoxin (Cigua-Checkby Oceanit) exists. In principle, wide avail-ability of this (or of alternative) test(s) tofishermen could contribute to a safe commer-cialization of C. argus. However, the currentcost to test a single fish is US $10.99 (www.oceanit.com), which appears prohibitive, inparticular when considering that this amountcould be used to purchase nonciguatoxic fishspecies directly.

Geographic Patterns in Ciguatoxicity

Understanding of geographic patterns in C.argus toxicity in Hawai‘i could still lead tosafe fisheries of limited geographic extent.This study concentrated on patterns on threespatial scales within the main HawaiianIslands: (1) between islands, (2) along coast-lines, and (3) between sample sites. Regarding(1), mean toxicity differed significantly be-tween islands (Table 3). The mean toxicityof C. argus from O‘ahu ðx ¼ 0:73Þ fell intothe negative Membrane Immunobead Assayscore class (Table 2), which is consideredsafe for human consumption. Even the muchhigher mean toxicity for Hawai‘i Islandðx ¼ 1:55Þ still fell into the marginal scoreclass, generally considered as safe. However,due to high variability in toxicity, despite thelow mean toxicity for O‘ahu, toxic (outlier)scores occurred frequently, and 7.5% of indi-viduals were unsafe for human consumption(Figure 2). For Hawai‘i Island, where unsafeindividuals were significantly more common(24.3% of total), the case was even less am-biguous. Overall, the risk of a fishery thus ap-pears unacceptably high in both cases.

Second, considering the lack of recogniz-able patterns in mean toxicity along coastlines(Figure 3), as well as the fact that none ofthe coastlines and only 2 of 17 sample sites(KA-W06 and OA-W01) were completelyfree of toxic invididuals (Figure 3), safe re-gional fisheries for C. argus do not appear

TABLE 5

Prevalence of Individuals Considered Unsafe for Human Consumption Due to Risk of Ciguatera in Different Speciesof Reef Fishes

Species/Group Locationa Prevalenceb Test c Source

C. argus MHI 18.2 MIA This studySeriola dumerilid MHI 14.7 Radioimmunoassay Kimura et al. (1982)Caranx spp.e MHI 15.4 MIA Hokama et al. (1998)Sphyraena barracudad Florida 30 Cytotoxicity assay Bottein Dechraoui et al. (2005)Hawaiian reef fishes overall MHI 4 MIA Hokama et al. (1998)

a MHI, main Hawaiian Islands.b Proportion of fish considered unsafe for human consumption in each study.c MIA, Monoclonal Immunobead Assay.d Banned from fish markets due to ciguatera.e Considered as high-risk species.

Ciguatera in Cephalopholis argus in Hawai‘i . Dierking and Campora 201

feasible either. Both safe sites were not repre-sentative of larger geographic trends, becausetoxic specimens occurred at the neighboringsites OA-W02, KA-W05, and KA-W08.These results stand in contrast to the patterndescribed by Lewis (2001) in the Republic ofKiribati, where the southern reef of Tarawaand the western reef Maraki have a high riskof ciguatera, whereas remaining reefs arelow-risk areas. The Kiribati pattern couldoffer the resource management solution ofclosing only the toxic reef to fishing, butsuch an approach does not appear feasiblefor C. argus in Hawai‘i.

Finally, mean toxicity differed significantlybetween sites (Table 4). At the same time, in-dependent of mean toxicity, all sites werecharacterized by large variability in toxicity.This is illustrated by the broad score rangeexpanding into the unsafe score range at allbut three sites and by the occurrence ofhighly toxic outliers at six of the 17 sites (Fig-ure 3). The high variability in toxicity withinsites is reflected in the large proportion ofoverall variation in toxicity that is due towithin-site variation (87.8% of total [Table4]). The practical implication of the describedpatterns is that even individuals caught at siteswith low mean toxicity cannot be consideredsafe for a fishery.

Morphometrics and Ciguatoxicity

A strong correlation between fish morpho-metrics and ciguatoxicity would enhance ourability to predict the toxicity of wild-caughtC. argus. Individuals of widely varying sizeand condition were present at each samplesite. Could the high variability of toxin con-centrations within subpopulations be linkedto variation in size or condition? Regardingsize, the significant positive correlation ofciguatoxicity with SL shows that a link ex-isted. However, length explained only a mi-nor proportion of the variation in toxicity(r2 ¼ 1:5% [Table 4]). Although large indi-viduals were on average slightly more toxicthan small ones, the many exceptions to thispattern (in particular, the frequent occurrenceof large individuals free of ciguatoxin, and ofsmall but highly toxic individuals [Figure 4])

suggest that bioaccumulation of ciguatoxinover time does not play an important role inC. argus. In contrast, ciguatoxicity and bodysize are strongly positively correlated in sev-eral other reef fishes (Lehane and Lewis2000) (e.g., in the two-spot red snapper [Lut-janus bohar] in the Line Islands [Helfrich et al.1968]). Although the latter pattern couldoffer the fishery management solution ofmarketing only fish below a predetermined‘‘cut-off size,’’ the divergent pattern in C. ar-gus prevents the use of this approach to estab-lish a safe fishery.

Ciguatoxin exerts negative effects that re-sult in reduced condition in numerous spe-cies, including several freshwater and marinefishes (Lewis 2001, Landsberg 2002). In con-trast, because fish condition was unrelated totoxicity in C. argus (Figure 5), ciguatoxic indi-viduals of this species cannot be distinguishedfrom nonciguatoxic individuals based on dif-ferences in weight-at-length. Condition istherefore not a useful factor in fishery man-agement of C. argus in Hawai‘i.

The Patterns behind the Patterns

The foodchain concept of ciguatera (Banner1974) suggests that patterns of fish toxicityare linked to the frequency and extent of G.toxicus blooms, which to date are not well un-derstood (Lehane and Lewis 2000). This sec-tion represents an attempt to elucidate bloompatterns in Hawai‘i by deduction from thepatterns in toxicity observed in C. argus. Inparticular, which pattern could account forthe high within-site variability in toxicity andthe seemingly unpredictable occurrence oftoxic individuals? Cephalopholis argus livesin harems occupying home ranges with dis-tinct spatial boundaries, which are activelydefended against intrusion from neighbor ha-rems. Home ranges are smaller (100 to 2,000m2 [Shpigel and Fishelson 1991]) than samplesites in this study (10,000 m2) (i.e., each sam-ple site encompassed multiple home ranges).Gillespie et al. (1985) observed that ciguaterablooms can occur in small, localized patches.Presence of such a pattern in Hawai‘i wouldmean that only some of the home rangeswithin sample sites overlap with blooms, and

202 PACIFIC SCIENCE . April 2009

result in highly variable uptake of ciguatoxinbetween C. argus from different harems atthe same site. Variability in toxicity withinsample sites observed here would thus beconsistent with small, highly heterogeneousblooms being the norm in Hawai‘i.

In conclusion, this study underlines thecomplexity of patterns of ciguatera in fishand suggests that G. toxicus blooms may behighly localized. The high prevalence of toxicC. argus individuals and the large variability intoxicity on all analyzed spatial scales, as wellas the difficulty of predicting toxicity basedon simple factors, currently stand in the wayof a safe fishery for C. argus in Hawai‘i.

acknowledgments

We thank C. Birkeland, I. Williams, Y. Ho-kama, S. Conant, R. Kinzie, and W. Walshfor suggestions that improved this paper. A.Meyer, T. Clark, S. Fujimoto, R. Robertson,S. Cotton, and B. Carmen assisted with field-work and C. Suma with laboratory work.

Literature Cited

Banner, A. H. 1974. The biological originand transmission of ciguatoxin. Pages 15–36 in H. J. Hum and C. E. Lance, eds.Bioactive products from the sea. MarcelDekker, New York.

Bolger, T., and P. L. Connolly. 1989. The se-lection of suitable indices for the measure-ment and analysis of fish condition. J. FishBiol. 34:171–182.

Bottein Dechraoui, M.-Y., J. A. Tiedeken, R.Persad, Z. Wang, H. R. Granade, R. W.Dickey, and J. S. Ramsdell. 2005. Use oftwo detection methods to discriminateciguatoxins from brevetoxins: Applicationto great barracuda from Florida Keys.Toxicon 46:261–270.

Bruno, P. P., and P. V. Effler. 2001. Cigua-tera: A five year review. Pages 4–6 inCommunicable disease report. Hawai‘iState Department of Health, Honolulu.

Campora, C. E., J. Dierking, C. S. Tamaru,Y. Hokama, and D. Vincent. 2008. Detec-tion of ciguatoxin in fish tissue using a

sandwich ELISA and neuroblastoma cellbioassay. J. Clin. Lab. Anal. 22:246–253.

De Sylva, D. P. 1994. Distribution and ecol-ogy of ciguatera fish poisoning in Florida,with emphasis on the Florida Keys. Bull.Mar. Sci. 54:944–954.

Dierking, J. 2007. Effects of the introducedpredatory fish Cephalopholis argus on nativereef fish populations in Hawaii. Ph.D.diss., University of Hawai‘i at Mānoa, Ho-nolulu.

Earle, J. L. 2005. Have we created a monster?Hawaii Fish. News 31:14.

Gillespie, N. C., M. J. Holmes, J. B. Burke,and J. Doley. 1985. Distribution and peri-odicity of Gambierdiscus toxicus in Queens-land, Australia. Pages 183–188 in D. M.Anderson, A. White, and D. G. Baden,eds. Toxic dinoflagellates. Elsevier, Lon-don.

Heemstra, P. C., and J. E. Randall. 1993.FAO species catalogue. Vol. 16: Groupersof the world. FAO, Rome.

Helfrich, P., T. Piyakarnchana, and P. S.Miles. 1968. Ciguatera fish poisoning: I.The ecology of ciguateric reef fishes inthe Line Islands. Occas. Pap. BernicePauahi Bishop Mus. 23:305–382.

Hokama, Y., and J. S. M. Ebesu. 2001. Cigua-tera fish poisoning—a foodborne disease.J. Toxicol. Toxin Rev. 20:85–139.

Hokama, Y., W. E. Takenaka, K. L. Nishi-mura, J. S. M. Ebesu, R. Bourke, and P. K.Sullivan. 1998. A simple membrane immu-nobead assay for detecting ciguatoxin andrelated polyethers from human ciguateraintoxication and natural reef fishes. J.Assoc. Off. Anal. Chem. Int. 81:727–735.

Kimura, L. H., M. A. Abad, and Y. Hokama.1982. Evaluation of the radioimmunoassay(RIA) for detection of ciguatoxin (CTX )in fish tissue. J. Fish Biol. 21:671–680.

Landsberg, J. H. 2002. The effects of harmfulalgal blooms on aquatic organisms. Rev.Fish. Sci. 10:113–390.

Lee, C., and Y. Sadovy. 1998. A taste for livefish: Hong Kong’s live reef fish market.Naga 21:38–42.

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

Ciguatera in Cephalopholis argus in Hawai‘i . Dierking and Campora 203

Lewis, R. J. 2001. The changing face of ci-guatera. Toxicon 39:97–106.

Lewis, R. J., A. Jones, and J. P. Vernoux.1999. HPLC/tandem electrospray massspectrometry for the determination ofsub-ppb levels of Pacific and Caribbean ci-guatoxins in crude extracts of fish. Anal.Chem. 71:247–250.

Manger, R. L., L. S. Leja, S. Y. Lee, J. M.Hungerford, Y. Hokama, R. W. Dickey,H. R. Granade, R. Lewis, T. Yasumoto,and M. M. Wekell. 1995. Detection of so-dium channel toxins: Directed cytotoxicityassays of purified ciguatoxins, brevetoxins,saxitoxins, and seafood extracts. J. Assoc.Off. Anal. Chem. Int. 78:521–527.

Morris, A. V., C. M. Roberts, and J. P. Haw-kins. 2000. The threatened status ofgroupers (Epinephelidae). Biodiv. Con-serv. 9:919–942.

Oda, D. K., and J. D. Parrish. 1981. Ecologyof commercial snappers and groupers in-troduced to Hawaiian reefs. Pages 59–67in Proc. 4th Int. Coral Reef Symp., Ma-nila.

Palafox, N. A., and L. E. Buenconsejo-Lum.2001. Ciguatera fish poisoning: Review ofclinical manifestations. J. Toxicol. ToxinRev. 20:141–160.

Randall, J. E. 1987. Introductions of marinefishes to the Hawaiian Islands. Bull. Mar.Sci. 41:490–502.

Schmitt, R. C., ed. 2002. State of Hawai‘idata book. Hawai‘i Department of Busi-ness, Economic Development, and Tour-ism, Honolulu.

Shpigel, M., and L. Fishelson. 1991. Territo-riality and associated behaviour in threespecies of the genus Cephalopholis (Pisces:Serranidae) in the Gulf of Aqaba, RedSea. J. Fish Biol. 38:887–896.

Tissot, B. N., W. J. Walsh, and L. E. Hal-lacher. 2004. Evaluating effectiveness of amarine protected area network in WestHawai‘i to increase productivity of anaquarium fishery. Pac. Sci. 58:175–188.

Yasumoto, T., Y. Oshima, Y. Murakami, I.Nakajima, R. Bagnis, and Y. Fukuyo.1980. Toxicity of benthic dinoflagellates.Bull. Jpn. Soc. Sci. Fish. 46:327–331.

204 PACIFIC SCIENCE . April 2009