An Analysis of the Fish and Macrobenthos Along the Sand Island Ocean Outfall Using Remote Video: Vii

HAwAU-S-97-002 C2

AN ANALYSIS OF THE FISH AND MACROBENTHOS ALONG THE SAND ISLAND OCEAN OUTFALL USING REMOTE VIDEO: VII. 1996 DATA

Richard E. Brock

PROJECT REPORT PR-97-05

March 1997

WATER RESOURCES RESEARCH CENTER UNIVERSITY OF HAWAI'I AT MANOA Honolulu, Hawai'i 96822 AUTHOR:

Dr. Richard E. Brock Associate Researcherand Fisheries Specialist Sea Grant Extension Service Marine Science Building 204 University of Hawai'i at Manoa Honolulu, Hawai'i 96822 Tel.: 808/956-2859 FAX: 808/956-2858

$5.00/copy

Please make remittance in U.S. dollars from a U.S. bank or international money order to: ResearchCorporation of theUttiversity of Hawaii

Mail to: Water Resources Research Center University of Hawai'i at M5aoa 2540 Dole St., Holmes Hall 283 Honolulu, Hawai'i 96822 ~ U.S,A. Attn: Publications Office

NOTE: Pleaseindicate PR-97-05 on checkor moneyorder for our reference,

Richard E. Brock

Project ReportPR-97-05

March 1997

PREPARED FOR Departmentof WastewaterManagement City and County of Honolulu Project Report for "The Assessmentof the Impactof OceanSewer Outfalls on the Marine Environment off Oahu, Hawaii" ProjectNo.: C39805 Project Period: 1 January1995-30 September 1997 Principal Investigator: RogerS. Fujioka

WATER RESOURCES RESEARCH CENTER University of Hawai'i at Manoa Honolulu, Hawai'i 96822 Any opinions, findings, and conclusionsor recommendationsexpressed in this publication are those of the author and do not necessarilyreflect the view of the Water ResourcesResearch Center. ABSTRACT

Becausethe diffuser of the SandIsland OceanOutfall lies below safe diving depths,a remotely controlled video camerasystem was usedto determinethe statusof the fish and diurnally exposedmacrobenthos resident to the diffuser,The useof a remotelyoperated vehicle is stipulatedin the NationalPollutant Discharge Elimination System301 h! waiver permit for the Sand Island Wastewater Treatment Plant. Video reconnaissancewas completed over the entire 1,036 m length of the outfall diffuser. Five visual "transects," which "sampled" approximately41% of the total diffuser length, were establishedon the diffuser pipe. Video sampling of the diffuser marine communitieswas carried out annually from 1990 through 1996, Only a few species of diurnally exposed macroinvertebratesare evident on the videotapesof the diffuser; the numbersare insufficient for any meaningfulanalysis. In 1996, 29 fish species,931 individuals!having an estimated biomass ranging from 12to 293g/m2 incan 92 g/m>! were censused,In the 1991through 1995period the numberof fish species encounteredduring a surveyranged from 22 species in 1993!to 31 species in 1992!,the total number of individuals from 279 in 1993! to 2,936 in 1992!, and the mean biomass from 21 g/m2 in 1993!to 53 g/m2 in 1992!.Because the 1990video census covered only theterminal 183 m of the diffuser, whereas the later surveys were spread out along the entire diffuser length,a directcomparison cannot be madebetween the 1990data and the datafor subsequent years.The 1996census noted one new fish speciesfor every 15.3m of substratumsampled and one fish for every 0.7 m2. In the 1991 through 1996 period, measuresof the fish community numberof species,number of individuals,and biomass! have oscillated annually after an initial increase from 1991 to 1992 and a decrease in 1993. From a statistical perspective,changes in the meannumber of speciesper transectand the mean number of individual fishesper transectare significant Kruskal Wallis ANOVA!; changesin the biomass of fishes over the sameperiod are not significant. Thesechanges in the fish community are attributed to changesin the generalviewplane of the videotapesrecorded in 1994 and later years from that recordedin earlier years, as well as to a changein the resolution of the videotapefrom which the dataare derived.Poorer camera resolution results in lower counts; cameraresolution is affectedby local wind andcurrents interacting with the camera,tether, and support vessel,as well as by water visibility. Controlling thesesources of variation inherent with the use of the remotely operatedvideo systemis difficult if not impossible. Until an alternative can be found, the remotely controlled video system is the only low-cost means available to view the marine communities on the diffuser. Until a more accurate means of visual assessmentis available,the biological data generatedby the remotely operatedvideo camera shouldbe viewedas qualitative, with little statisticalrigor.

CONTENTS

INTRODUCTION .

MATERIALS AND METHODS

RESULTS ..

DISCUSSION . 17

REFERENCES CITED .. 20

Figure

l. Schematic of the 1,036-m-long Sand Island Ocean Outfall diffuser pipe showing the approximatelocations of the five transectsestablished in August 1991 and monitored annually using a remotely controlled video recording system.

Tables

1. Family and species of fishes censused on five transects along the 1,036-m-longdiffuser pipe of the Sand!sland OceanOutfall as delineated using a remotely controlled video camera systemon 26 September1996.... 2. Summary of the diurnally exposedmacroinvertebrates censused on five video transectscarried out along the 1,036-m-longdiffuser pipe of the Sand Island Ocean Outfall on 26 September 1996 . 3. Summary of diurnally exposed macroinvertebrates censused on five transectsalong the 1,036-m-longdiffuser pipe of the Sand Island Ocean Outfall as delineated using a remotely controlled video camera system in the annual surveys from 1991 through 1995. 4. Summaryof the characteristicsof five transectscarried out at various points along the 1,036-m-longSand Island OceanOutfall diffuser, with data from the fish censuses camed out at each transect in September 1996... 10 5, Family and speciesof fishes censusedon five transectsalong the 1,036-m-longdiffuser pipe of the Sand Island OceanOutfall as delineated using a remotely controlled video camerasystem on 22 August 1991 6. Family and speciesof fishes censusedon five transectsalong the 1,036-m-longdiffuser pipe of the Sand Island OceanOutfall as delineated using a remotely controlled video camera system on 28 August 1992 7. Family and speciesof fishes censusedon five transectsalong the 1,036-m-longdiffuser pipe of the SandIsland OceanOutfall as delineated using a remotely controlled video camerasystem on 5 August 1993. 13 8. Family and speciesof fishes censusedon five transectsalong the 1,036-m-longdiffuser pipe of the Sand Island OceanOutfall as delineated using a remotely controlled video camerasystem on 10 August 1994 14 9. Family and speciesof fishes censusedon five transectsalong the 1,036-m-long diffuser pipe of the Sand Island Ocean Outfall as delineated using a remotely controlled video camerasystem on 16 November 1995.... 15 10. Comparativesummary of fish community developmentmeasured over six yearsat five locations along the 1,036-m-longSand Island Ocean Outfall diffuser, . 11. Summary of the nonparametricStudent Newman Keuls multiple rangetest on rankedvalues of parametersmeasured in the fish community at five permanenttransects along the 1,036-m-long Sand Island diffuser censusedannually from 1991 through 1996 . INTRODUCTION

In recentyears controversy has arisen regarding the impactthat sewageeffluent from the Sand Island Wastewater Treatment Plant may have on marine communities resident to the receiving waters. The outfall was constructed in 1975, and screenedsewage has been dischargedsince 1976.The oceanportion of the outfall is comprisedof 2,780 m of 2,1-m- diameter reinforcedconcrete pipe that terminatesin a 1,036-m-longdiffuser. The diffuser is madeup of reinforcedconcrete pipe of three diameters:490 m of 2.1-m-diameterpipe, 271 m of 1,7-m-diameterpipe, and 275 m of 1.2-m-diameterpipe at the terminus.Along its length, the diffuser, which lies in waterfrom 68 to 73 m in depth,has 282 portsthat rangefrom 7.6 to 9 cm in diameter.The diffuser restson a gravelpad and has someballast rock placed at the junctures betweensections. Fishes and macroinvertebrateshave taken up residencealong most of the length of the deep oceanoutfall. This study has been undertakenin an attempt to semiquantitativelyascertain the impactsthat may be occurring to the communitiesresident to the dischargeport areasof the outfall. This report presentsa synopsisof the data from the seventhannual sampling effort carriedout on 26 September1996 and comparatively analyzes these data with information coUectedannually since 1991.

MATERIALS AND METHODS

A remotely controlledvideo camerawas usedto conductthe censusbecause the fish and diurnally exposedmacroinvertebrate communities of interest to this study reside in waters below safe diving depths.In addition,the systemwas usedbecause the waiver permit issued by the U.S. EnvironmentalProtection AgencyfHawai'i Departmentof Healthrequires the use of a remotelyoperated vehicle. There are a numberof drawbacksas well aspositive aspects to using a video camera system to visually census fishes and diurnally exposed macroinvertebrates.The drawbacksinclude problemswith cameraresolution, making species and size identificationsdifficult, andthe problemof adequatelycontrolling the camera to focus- in on rapidly fleeingfishes, adding further difficulty to identificationproblems. On the positive side, a permanentrecord of' the organismsin the pathof the camerais obtained.An additional benefit to using a videosystem is that it eliminatesthe needfor diving to greatdepths. There are somewelf-known problems with using visual censusmethods to assesscoral reef fish populations,regardless of whether a cameraor diver is in the water conductingthe census. One of these is the simple frightening of wary fishes on the approach of the diver or camera. Another is the underestimationof cryptic species such as moray eels family Muraenidae!and nocturnalspecies such as squirrelfishes family Holocentridae!and bigeyes or 'aweoweo family Priacanthidae!.This problemis compoundedin areasof high relief andcoral coveragethat affords numerous shelter sites. Species lists andabundance estimates are more accurate for areas of low relief, although some fishes with cryptic habits or protective coloration, such as scorpionfishesor nohu family Scorpaenidae!and catfishes family Bothidae!, might still be missed.Another problem is the reducedeffectiveness of the visual censustechnique in turbid water.This is compoundedby the difficulty of countingfishes that move quickly or are very numerous.Additionally, bias relatedto the experienceof the census taker should be considered in making comparisons between surveys. Despite these problems, the visual censustechruque carried out by divers is probablythe most accurate,nondestructive assessmentmethod currently availablefor counting diurnally active fishes Brock 1982!,Use of a remotely controlled video systeinto obtain censusdata compoundsmany of the above problems,but it is probablyone of the mostcost-effective methods available for assessingfish communitiesat depthsbelow safediving limits. Other than exposedsessile species corals in shallowwater and somesponges in deeper water!, most tropical marine invertebratesare cryptic, remainingunder shelteruntil darkness when they emergeto feed.Only a fewmotile macroinvertebratesremain fully exposedduring the day; amongthese are someholothtuian seacucumber! and echinoid seaurchin! species. Problemswith speciesidentification preclude the enumerationof mostof the diurnally exposed invertebrates. Identification of holothurians is based on a microscopic examination of skin spiculeconfiguration, and spiculesare also used for the identificationof sponges.Thus, in this study,the identification and enumerationof exposedmacroinvertebrates are confined to large arthropods spiny lobsters! and seaurchins; as for speciesof holothurianspresent along the SandIsland diffuser pipe, educated"guesses" are made. This study utilized a remotelycontrolled video systemto visually assessthe fish and macroinvertebrate populations resident to the diffuser pipe. The video "transect" was undertakenby the OceanographicTeam of the Departmentof WastewaterManagement, City and County of Honolulu. In the first tlat annualsurveys, the video cameragenerally traveled from 0.5 to 1.5 m above the diffuser pipe, occasionally moving to the right or left side and down! to survey the substratumalongside the pipe. The cameraviewed a pathfrom about 1.5 to 3 m in width. For purposesof data analysis, we assumedthat the camera path was approximately2 m in width, andwe attemptedto countonly fishesseen in this path.At times, the camera would tilt up toward the horizon!, allowing a viewing aheaddown the pipe. Visibility underthese circumstances ranged from about0.5 m in a dischargeplume! to about8 m, which is approximately the length of one pipe section.In the fourth 993! survey, the cameraroughly followed the samepath, but on occasionit roseto about4 m abovethe pipe. In doingso, a widerfield of viewwas attained, but with the small average size of fishpresent this meant a decreasein resolutionwith greaterheight above the pipe, resulting in a decreasein apparentabundance of fishes. Thepath of thecamera in the 1994and 1995 surveys was quite different. The camera remainedalong one inshore!side of thepipe while travelingtoward the outfall terminus, and on reachingthe terminus, the camera was moved to theopposite offshore! side, for its travel backto thebeginning of thediffuser. The censuses for the 1994and 1995 surveys were carried out at each of the five transect sites on the offshore side of the diffuser only, to make data betweensurveys comparable. For the 1996survey, the camera traveled down the offshore side of the pipe. Thecamera grossly underestimates the number of fish andinvertebrates, thus everything seenin an arbitrary2-m-wide path was counted, regardless of whetherit wasencountered directlybelow the camera as when viewing from above!or severalmeters ahead as when the camera is in a horizontal position!. The fish census involved not only the counting of populationsbut also the estimating of lengthsof all fishesfor lateruse in calculatingstanding crop. The standingcrop of all fisheswas estimated by useof linearregression techniques Ricker 1975;Brock andNorris 1989!.Species-specific regression coefficients have been developedover the last 30 yearsby the authorand others at the Universityof Hawai'i, the Naval UnderseaCenter seeEvans 1974!,and the Hawai'i Division of Aquatic Resourcesfrom weightand body length measurements of captured fishes; for manyspecies, sample sizes were in excess of a hundred individuals.

RESULTS

Video "transects"of the fish communitiesresident to the SandIsland diffuser pipe have been carried out on sevenoccasions: 7 November 1990,22 August 1991,28 August 1992, 5 August1993, 10 August 1994, 16 November 1995, and 26 September 1996. The November 1990survey only coveredthe final 183m of the 1,036-m-longdiffuser pipe. In the more recentsurveys, the camera commenced just shorewardof the first dischargeport on the diffuserand "sampled" the fish andmacrobenthos for theentire diffuser length. The 1991 surveytape was viewed several times to determinewhere representative transects could best be established.Five transectsites selectedas being representativesections of the diffuser pipe weresampled using the visual census technique. These transects were located using known pointson the pipe and by countingsections of pipefrom those points. The five transectsites wereagain sampled in subsequentsurveys, thus aHowing for datacomparison between years. The locationof eachtransect is shownin Figure 1. The transectsrange from 44 to 110m in length; thus,in total, approximately848 m of substratumwere sampledin this survey.The results of all fish censusesare presentedin Table 1, and the data for each transect are discussed below. In tallying the numberof speciesseen on a giventransect, all fishes that could not be positively assignedto a givenspecies were lumped into groups such as "labrid unidentified"!; in the tally of species,each of thesegroups was countedas a singlespecies, even though more than one species may have been present. The diurnaHy exposed macroinvertebrates that were present in each transect area are tallied in Table 2, and the macroinvertebrate censusdata from previousannual surveys are presented in Table 3 for comparativepurposes. Transect1 commenced58 m downfrom the beginningof the diffuser pipe andcontinued for approximately95 m alongthe pipe toward the terminus Figure 1!. This transectsampled 13.5sections of the 2.1-m-diameterdiffuser pipe. The depthat the top of the diffuser pipe was approximately 68.5 m at the start of the transect and about 70.1 m at the end. Ten species of fish 45 individuals! were noted on this transect, and the biomass was estimatedat 21 g/rn~.This amountsto one new speciesencountered for every 19.0 m~ of substratumsampled and one individual fish seenfor every 0.8 m~.Of the speciesthat could be identified, the most abundanton this transect was the damselfish Chromis sp., probably C. hanui and/or C. agilis!. In termsof standingcrop, two bluespineunicornfish or kala Naso unicornis! accountedfor 69% of the total. One macroinvertebratespecies, the brown sea cucumber or Bohadschia vitiensis, was censusedon this transect Table 2!. Transect2 commenced355 m down from the beginning of the diffuser pipe in about 71 m of waterand ended 91 m downthe pipe from that point Figure 1! in approximately 71.3 m of water depth to the top of the pipe!. This transectsampled 12.5 sectionsof the 2.1-m-diameter portion of the diffuser pipe. Seventeenspecies of fishes were censused Table 1!; this translatesto one new speciesseen for every 10.7m> of substratumsampled. In total, 310 individual fishes were counted, and one new fish was seenfor every 0.6 m~ of substratumsampled. The mostcommon fish speciesseen were the damselfish Chromis sp. 20% of the total! andthe juvenile snapper Lujtanus sp. 52% of the total!. Importantspecies by weight were two morayeels or puhi Gymnothorar sp. and G. flavimarginatus 44% of the total! and thejuvenile snappers Luj tanus sp. 15%of the total!. The biomassof fisheson Transect 2 was estimatedat 79 g/m>. Three macroinvertebratespecies were seenon this transect Table 2!. Of interest was a single spiny lobster or 'ula Panulirus marginatus! estimated to ~eigh 0.6 kg. Seventy-threemeters from the end of Transect 2, Transect 3 was established see Figure 1!. The waterdepth at the beginningof the transectwas 71.3 m to the top of the pipe; at the end of the transectat 1 10m away,it was 71.9 m. Transect3 sampled2.5 sectionsof the C Q. C III ID II! III 0 Oo C IO C 4 O O

O « C V 0 C g g 2 4! ~ cn 8 X 0 0 III C Z0 tO 2' Io 4 50 C Q O IO IO C G « Kg O. O E 4 g 0 III 9 III l3 III Ill III ~ 0 O 0 c C 0 0« Cl8: E III IO m 0l C C CI IN IO g E n~o III C= III III 0 'll O c m C CL C «C III p~c 0'C 9 6 ~P 0 r O 0 IllCOe C 1~ 0l C

U-%8 C C TABLE 1. Family and Speciesof FishesCensused on Five TransectsAlong the 1,036-m-Long DiffuserPipe of the SandIsland Ocean Outfall asDelineated Using a RemotelyControlletl Video CameraSystem on 26 September1996 Tratlsec't FAMILY and Species

MURAEMDAE Gymnorhorax sp. Gymnorhorax flavimargi natus SERR AMDAE Pseudanrhias sp. 26 216 22 71 APOGOMDAE Apogon kallopterus CARANGIDAE Caranx melampygus LUTJAMDAE Lujranus sp. 162 380 48 MULLIDAE Parupeneus mulrifasciarus 20 14 CHAETODONTIDAE Chaetodon kleinii Chaerodon mi liaris Chaetodon sp, POMACENTRIDAE Chromis hanui 27 20 Chromis agilis 1 Chromis sp. 192 61 416 52 13 Plectroglyphidodon j ohnstoni anus 1 Pomacentrid unidentified LAB RIDAE Bodianus bi lunulatus Pseudojuloides cerasinus 1 Thalassoma sp. 15 30 Labrid unidentified 21 GOBHDAE Pterelotris heteropterus ACANTHUIUDAE Acanthurus xanthopterus Acanthurus sp. Naso unicornis Naso hexacanthus Zanclus cornutus 12 BALISTIDAE Su flamen fraenatus TETR 4.ODONTIDAE A rothron hispidus Arothron sp. Canthigasrer coronata Total No, of Species 10 17 14 7 12 Total No. of Individuals 245 310 1,111 134 131 Biomass g/m2! 21 79 55 12 293 No>E:Areas sampled on the five transectsvaried: 190 m for Transect1, 182m for Transect2, 220m for Transect3, l68 m for Transect4, and 88 m for Transect 5. The numbersof individuals of each speciescensused are given in the body of the table. Totals for numbersof speciesand individuals and an estimateof biomassfor each transect are given at the foot of the table. TABLE2. Summaryof the Diurnally ExposedMacroinvertebrates Censused on Five Video TransectsCarried out Along the 1,036-m-LongDiffuser Pipeof the SandIsland OceanOutfall on 26 September1996

Transect PHYLUM and Species

ARTHROPODA Panulirus marginatus

ECHINODERMATA Bohadschia vi tiensis ?! Chondrocidarisgi gantea Culcita novaeguineae Holothuria atra 12

Total No. of Species 1 3 1 2 1 Total No, of Individuals 1 4 12 13 5

NorE: The numbers of individuals of each speciescensused are given in the body of the table. Totals for numbers of speciesand individuals for eachtransect are given at the foot of the table.

2.1-m-diameterportion of the diffuser pipe and 12.5 sectionsof the 1.7-m-diameterportion. Fourteenspecies of fisheswere counted and one new fish specieswas seen for every 15.7m2 of substratumsampled. The number of individual fishes encounteredon this transect was 1,111, and one new fish was seen for every 0.2 m~ of substratum sampled. The most abundantfish specieson this transectappeared to be Chrornis sp. 7% of the total! and the juvenile snapperLurj'anus sp. 4% of the total!, The standingcrop of fisheswas estimatedat 55 g/m2, 41% of which was comprisedof Lujtanus sp. and 22% of damselfishes Chrornis sp.!. Table 2 presentsa summaryof the macroinvertebratesencountered on Transect 3; one species,the black seacucumber Holothuria arra 2 individuals!,was censused. Transect 4 was established 161 m from the end of Transect 3 toward the diffuser terminus Figure 1!. It commencedat a depthof about70.7 m depthto the top of the pipe! and endedat a depthof 70.1 m. Transect4 sampledabout 84 m of the diffuser pipe, covering 2.5 sections of the 1.7-m-diameterportion of the pipe and 9 sections of the 1.2-m-diameter portion. In total, sevenspecies of fishes 34 individuals! were seenat this transect.Equating thesefigures to the areasampled results in onenew fish specieswas seen for every24.0 m2 andone fish encounteredfor every1.3 m2. The most abundant fish specieson Transect4 were Chromissp., which comprised 39% of thetotal numberof fishesseen, and juvenile Lujtanus sp.,which madeup 36%of thetotal. The standing crop of fishesat Transect4 wasestimated at 12 g/m2, and the speciescontributing most heavily were the bridled triggerfish or humuhumu mimi Sufgamenfraenatus 3 individuals making up 53% of the total! and TABLE3, Summaryof Diurnally ExposedMacroinvertebrates Censused on Five Transects Along the l,036-m-LongDiffuser Pipeof the SandIsland Ocean Outfall as DelineatedUsing a RemotelyControlled Video CameraSystem in the AnnualSurveys from 1991Through 1995

1991 1992 PHYLUM and Species Tfarlscct Transect 1 2 3 4 5 2 3 4 5

ARTHROPOD A Panuli rus marginatus

ECHINODERMATA Bohadschia vitiensis Culcita novaeguineae Diadema setosum 7 1 Holothuria atra 2 2 32 12 1 14 11 4 Tripneustesgratilla 3

Total No. of Species 2 2 2 1 2 1 2 2 1 1 Total No, of Individuals 9 3 33 12 4 2 3 21 11 4

1993 1994 Transect Transcc't I 2 3 4 5 1 2 3 4

ARTHROPOD A Panuli rus marginatus

ECHINODERMATA Bohadschia vitiensis ?! 1 1 2 3 Chondrocidarisgigantea 1 1 1 Culcita novaeguineae 1 1 Holothuria atra 2 2 21 23 7 4 16 10 Tripneustes gratilla 1

Total No. of Species 4 3 4 3 1 2 1 3 1 1 Total No, of Individuals 5 4 25 27 7 2 1 6 16 10

1995 Transect 1 2 3 4 5

ARTHROPOD A Panuli rus marginatus

ECHINODERMATA Bohadschia vitiensis ?! 3 4 Chondrocidaris gigantea 2 1 Holothuria atra 6 16 2 Tripneustes gratilla 1

Total No. of Species 1 1 3 3 Total No, of Individuals 2 1 10 21 2

NvrE: Areas sampledon the five transectsvaried: 190 m for Transectl, 182 m for Transect2, 220 rn for Transect3, 168m2 for Transect4, and88 m~for Transect5. The numbersof individualsof eachspecies are given in the bodyof the table. Totals for numbersof speciesand individuals for each transectare given at the foot of each section of the table. juvenile Lujtanus sp. 6% of the total!. Two macroinvertebratespecies Holothuria atr a and Bohadschiavitiensis! were seen on this transect Table 2!. Transect 5 covered the final 44 m of the diffuser pipe and the terminus. This transect sampled5.5 sectionsof pipe and the diffuser terminus.It commencedat a depthof about71 m depth to the top of the pipe! and endedat a depthof approximately69,5 m at the top of the diffuser terminus. Twelve speciesof fishes 31 individuals! were encounteredduring the census.As for areasampled, one new speciesof fish was seenfor every 7.3 m2 and one new fish for every 0.7 m, The most abundantspecies seen included the brightly colored seabass Pseudanthias sp. 4% of the total! and a small school of blue trevally or 'omilu Caranx rnelampygus 16% of the total!. The standingcrop of fisheson Transect5 wasestimated at 293 g/m>, and the species comprising the greatest proportion were C, rnelampygus 1 individuals inaking up 62% of the total! and Acanthurus xanthopterus one individual making up 6% of the total!. Only one macroinvertebratespecies the brown seacucumber Bohadschia vitiensis! was seenon this transect Table 2!. The results of the five transects carried out in 1996 using the video camera are summarizedin Table4, The visual fish censustransects from the videosshot in 1991through 1995covered the samefive sectionsof the diffuser pipe as thosein the presentsurvey; thus datafor the six yearscan be comparativelyanalyzed. The 1991through 1995 data are presented in Tables 5 through9, respectively,and data for all six yearsare summarizedin Table 10. In general,the 1992and 1994census data showed higher numbersof speciesthan in other years. Thesedifferences may be due to ! bettercontrol of the remotelyoperated camera in viewing

TABLE4. Summaryof the Characteristicsof Five TransectsCarried Out at Various Points Along the 1,036-m-Long Sand Island Ocean Outfall Diffuser, with Data from the Fish CensusesCarried Out at EachTransect in September1996

Transect Parameter

Transect Length m! 95 91 110 84 44 85

AreaSampled m2! 190 182 220 168 88 170

No, of Species 10 17 14 7 12 12

No. of Individuals 245 310 1,111 134 131 386

No. of m2 SampledPer New Species 19.0 10.7 15.7 24,0 7,3 15.3

No, of m~ SampledPer Individual 0,8 0.6 0.2 1.3 0.7 0.7

Biomass 8/m~! 21 79 55 12 293 92

NOTE:Means are presentedin the right column.

10 TABLE5, Familyand Species of FishesCensused on FiveTransects Along the 1,036-m-Long Diffuser Pipe of the Sand Island Ocean Outfall as DelineatedUsing a Remotely Controlled Video CameraSystem on 22 August 1991

Transect FAMILY and Species

MURAXNIDAE Gymnothorax flavimarginatus SEMbQ4IDAE Pseudanthias thompsoni 87 APt3GONIDAE Apogon kallopterus ?! LUTJANIDAE Lutjanus kasmi ra h4ULlJDAE Parupeneus multifasciatus 10 CHAETODONTIDAE Chaetodon multicinctus Chaerodon kleinii ?! Chaetodonsp, Heniockus diphreutes ?! POMACANTHIDAE Hoiocanthus arcuatus

POMACENTRIDAE Chromis hanui 23 Chrornis verator ?! 1 Chromis sp. 31 118 91 30 LAB RIDAE Bodianus bi lunuiatus Thaiassomaduperrey ?! Thalassoma sp, 1 Pseudojuloides cerasinus 4 11 19 Labrid unidentified 69 50 175 62 41 ACANTElURIDAE Acanthurus dussumieri A canthurusxanthopterus Acanthurussp. Zanclus cornutus

8AUSTIDAE Su~n fraenatus MONACANTHIDAE Cantherhines dumerilii

TETKKODONTIDAE Arothron hispidus Arothron meleagris ?! Canthigaster j actator ?! Candugaster coronata 1 1

Total No. of Species 12 13 14 11 11 Total No. of Individuals 169 217 1,045 147 207 Biomass g/m2! 32 8 55 10 106 NOTE:Areas sampled on thefive transectsvaried 190m2 for Transect1, 182m~ for Transect2, 220m~ for Transect3, 168 m for Transect4, and 88 m for Transect 5. The numbersof individuals of each speciesare given in the body of the table. Totals for numbersof speciesand individuals and an estimateof biomass for each transect are given at the foot of the table, Data from Broclt 1992b!

11 TABLE6. Familyand Species of FishesCensused on FiveTransects Along the 1,036-m-Long DiffuserPipe of the SandIsland Ocean Outfall as DelineatedUsing a RemotelyControlled VideoCamera System on 2S August1992 Trallsec t FAMLY and Species

MURAENIDAE Gymnothoraxfkrvi marginatus Gi rnnothoraxsp, A ULOSTOMIDAE A ulostomus chinensis PRIACANTHIDAE Priacanthussp. ?! S ERR/ttNIDAE Pseudanthias thompsoni 24 158 41 148 LUTJANIDAE Lujtanus kasmira 261 Lutjanusfulvus 2 2 Lujtanussp. ?! 722 153 375 MULLIDAE Parupeneusmultifasciatus 24 32 CHAETODONTIDAE Chaetodonmulticinctus ?! Chaetodonkleinii ?! Chaetodonsp. POMACANTHIDAE Holocanthus arcuatus 1 Centropygesp. ?! 16 POMACENTRIDAE Chromis hanui 52 31 28 19 8 Chromissp, 76 126 91 55 11 Pomacentrtd unidentified 11 3 LAB RIDAE Bodianus bilunulatus 1 Thalassomaduperrey ?! 23 17 8 Pseudojuloides cerasinus 3 15 Gomphosusvarius 3 Macropharyvtgodongeo+roy 1 3 Halichoeresornatissimus ?! 1 1 1 Labrid unidentified 111 8356 30 ACANTHURIDAE Acanthurus dussumien Zanclus cornutus BALISTIDAE Su enjraenatus Su@amenbursa ?! MONACANTHIDAE Cantherhi nes dumenlii TETRAODONTIDAE A rothronhispidus Canthigaster j aerator ?!

Total No. of Species 15 20 23 13 12 Total No. of Individuals 308 593 1,144 325 566 Biomass g/m ! 39 45 58 44 77 NOTE:Areas sampled on the five transectsvaried: 190 m for Transect1, 182m for Transect2, 220m for Transect3, 168 m2 for Transect4, and88 m~ for Transect5. The numbersof individualsof eachspecies are given in the body of the table.Totals for numbersof speciesand individualsand an estimateof biomassfor eachtransect are given at the foot of the table. Data from Broctt 1993a! TABLE7. Family and Speciesof FishesCensused on Five TransectsAlong the 1,036-m-Long Diffuser Pipe of the Sand Island OceanOutfall as DelineatedUsing a Remotely Controlled Video CameraSystem on 5 August 1993

Tfallsec t FAMILY and Species

SERRANIDAE Pseudanthias thompsoni

LUTJANIDAE Lujtanus kasmi ra

MULUDAE Parupeneus multifasciatus

CHAETODONTIDAE Chaetodonsp. Heniochus di phreutes

POMACENTRIDAE Chromis hanui 14 Chromis sp, 24 44 Pomacentrid unidentified 5

LAB RIDAE Pseudojuloides cerasinus ?! Thalassonta sp. 2 2 Labrid unidentified 18 15 34 10

ACANTHlURIDAE Acanthurus xanthopterus Acanthurus sp, Ctenochaetus stri gosus ?! Nasa unicornts ?! Zanclus cornutus

BALISTIDAE Sufflamenfraenatus Rhinecanthus aculeatus ?!

MONACANTHIDAE Cantherhi nes dumerilii

TETRAODONTIDAE Arothron hispidus Canthigaster j aerator Canthigaster coronata

Total No. of Species ll 10 11 8 9 Total No. of Individuals 42 52 126 23 36 Biomass g/m2! 6 12 11 39 39

NoTE: Areas sampledon the five transectsvaried: 190 m for Transect1, 182 rn for Transect 2, 220 m for Transect 3, 168 rn for Transect4, and 88 m for Transect 5. The numbersof individuals of each speciesare given in the body of the table. Totals for numbersof speciesand individuals and an estimateof biomass for each transect are given at the foot of the table, Data from Brock 1993b!

13 TABLE8. Family and Speciesof FishesCensused on Five TransectsAlong the 1,036-m-Long Diffuser Pipe of the SandIsland OceanOutfall as DelineatedUsing a RemotelyControlled Video CameraSystem on 10August 1994 Transect FAMILY and Species

IvIUIVZNIDAE Gymnothorax sp, AULOSTOMIDAE Aulostomus chinensis SERRAMDAE Pseudanthiassp. 18 35 103 APOGONIDAE Apogon kallopterus 17 CARANGIDAE Caranx nrelampygus LUTJANIDAE Lutjanus kasmira 72 20 IvIULLIDAE Parupeneusmulti jasciatus ltfulloidichthys vanicolensis AE Chaetodon kleinii Chaetodon mi liaris POMACANTMDAE Centropyge fi'sheri POMACH'~AE Chromis hanui 18 16 64 !4 4 Chromis agilis 9 7 Chromis sp. !41 90 339 77 22 Pomacentrid unidentified 6 19 4 4 I Plectroglyphidodon j ohnstonianus ?! LAB RIDAE Pseudojuloides cerasinus ?! 18 15 12 Bodianus bilunulatus Anampses chrysocephalus I Halichoe res ornatissimus 3 Labrid unidentified 15 22 50 18 ACAYPlURIDAE Acanthurus sp. ,Vaso unicornis ?! paso brevirostris Zanclus cornutus BALIFIlDAE Sup!amen fraenatus Su~n juscus ?! MON ACANIMDAE Cantherhines dumerilii Aluterascripta lETRAODONIlDAE Arothron hispidus Canthi gaster j actator Canthigastercoronata

Total No. of Species 14 16 24 16 14 Total No. of Individuals 225 194 691 186 177 Biomass g/mr! 16 46 42 43 28 NoTE:Areassampled on the five transectsvaried: 190 m for TransectI, 182 m for Transect2. 220 m for Transect 3. 168m2 for Transect4, and88 m2for Transect5. The numbersof individualsof eachspecies are given in the bodyof the table. Totals for numbersof speciesand individuals and an estimateof biomassfor each transectare given at the foot of the table, Data from Brock 1993b! TABLE9. Familyand Species of FishesCensused on FiveTransects Along the 1,036-m-Long Diffuser Pipeof the SandIsland Ocean Outfall as DelineatedUsing a RemotelyControlled Video CameraSystem on 16November 1995 Transect FAVIILY and Species

ML% ~AE Gymnothorax sp. SERIVAIDAE Pseudanthias sp, 35 l48 45 62 CARANGIDAE Caranx ignobi lis LUTJANIDAE Lutj anus kasmira 64 25 MULLIDAE Parupeneusntultifasciatus 10 Parupeneuspleurosti grna 1 PRIACANIMDAE Priacanthus sp. CH/iiETODONTIDAE Chaetodon kleinii Chaetodon mi liaris Chaetodon sp. POMACAXI'HID AE Centropygefisheri ?! POMACENIRIDAE Chromis sp, 198 268 601 54 26 Pomacentrid unidentified 10 23 3 LABRIDAE Bodianus bilunulatus /Vfacropharyngodongeo+roy ?! Anampses chrysocephalus Labrid unidentified 27 43 65 13 ACANTHIURIDAE A canthurusxanthopterus Aconthurus sp. BALISTIDAE Sujflatnen fraenatus MONACANI"HIDAE Cantherhines dutnerilii DIODONIIDAE Diodon holocanthus TE1RAODON!IDAE Canthigaster cinctus Canthigaster sp. Unidentified fish

Total No. of Species 11 14 14 9 10 Total No. of Individuals 253 433 882 128 100 Biomass g/m2! 19 37 21 24 51 NOM:Areas sampled on the five transects varied: 190 m for Transect1, 182m for Transect2, 220m for Transect3. 168 m for Transect4, and 88 m for Transect5. The numbersof individuals of each speciesare censusedgiven in the bodyof thetable. Totals for nuinbersof speciesand individuals and an estimate of biomassfor eachtransect are given at the foot of the table. TABLE10. Comparative Summary of Fish CommunityDevelopment Measured Over Six Years at Five LocationsAlong the 1,036-m-LongSand Island Ocean Outfall Diffuser

No. of S ies No. of Individuals 1991 1992 1993 1994 1995 1996 1991 1992 1993 1994 1995 1996

1 12 15 11 14 11 10 169 308 42 225 253 245

2 13 20 10 16 14 17 217 593 52 194 433 310 3 14 23 11 24 14 14 1,045 1,144 126 691 882 l,ill

4 11 13 8 16 9 7 147 325 23 186 128 134

5 11 12 9 14 10 12 207 566 36 177 100 131

Mean 12 17 10 17 12 12 357 587 56 295 359 386 Biomass g/m~! 1991 1992 1993 1994 1995 1996

1 32 39 6 16 19 21

2 8 45 12 46 37 79

3 55 58 11 42 21 55

4 10 44 39 43 24 12

5 106 77 39 28 51 293

Mean 42 53 21 35 30 92 NQTE: Data are drawn from Tables 1 and 5 through 9, and meansfor the five transectsare given at the foot of each section of the table. the substratumor ! improvedability to identify fishesrecorded with a videocamera with the passageof time. From a review of all tapes for 1991 through 1996, it is suggestedthat maximumvisibility was about the same 8 m! for all yearsbut coverageby the camerawas better in 1992 and 1994. The control of the camera steadiness and area coverage! in 1993 appearedto be less than in other years, and the resolution in the 1993 videotape was considerablyless. Averagevisibility in 1993 was about 3 m, less than in previous surveys, The net resultis an apparentdecrease in fish speciesand individuals in the 1993data. As noted above,Table 10 presentsa summaryof the parametersmeasured in the fish communityalong the SandIsland diffuser for the six samplingdates. Statistically significant differenceswere found for the meannumber of fish speciesencountered on a transectamong thesix years Kruskal WallisANOVA, df = 5,p ! 0.01;see Siegel 1956!. The nonparametric Student-Newman-Keulsmultiple rangetest on rankedvalues of eachvariable SAS Institute, Inc. 1985!was usedto delineatestatistically significant differences for the meannumber of fish speciesper transect among the six years.These results are presented in Table11; only themean numberof fish speciesfor 1993differed significantly from the otherfive years.The same

16 TABLE 11,Summary of the NonparametricStudent-Newman-Keuls Multiple RangeTest on RankedValues of ParametersMeasured in the Fish Community at Five PermanentTransects Along the 1,036-m-LongSand Island Diffuser CensusedAnnually from 1991Through 1996

Parameter Year and Mean

Mean No. of Species Per Transect 1994 1992 1991 1996 1995 1993 17 17 12 12 12 10

Mean No. of individuals Per Transect Year 1992 1996 1995 1991 1994 1993 Mean 587 386 359 357 295 56

B iomass g/mz! 1996 1992 1991 1994 1995 1993 92 53 42 35 30 21

NoTE: The meanson a per-transectbasis are given in the body of the table. Horizontal lines connect meansthat do not differ significantly; breaks in the line show significant differences p > 0.05!. analyseswere performedfor the mean numberof individual fishes censusedon a transect. Again, the Kruskal Wallis ANOVA pointedout that a statisticallysignificant difference exists in the meannumber of individual fishesper transectamong the six years df = 5, p > 0.01!. The Student-Newman Keuls multiple rangetest found no statistically significantdifferences amongthe six years.No statisticallysignificant differenceswere found in the meanstanding crop of fishes encounteredon a transect among the six sampling dates Kruskal Wallis ANOVA, df = 5, p > 0.17, not significant; see also Table 11!. The 1990videotape only recordedthe fishes presenton the final 183 m of the diffuser Brock 1992a!;thus the data are not directly comparableto the 1991through 1996transect information. In the 1990study, Brock 992a! found 16species of fishesand one new species for every22.9 m2 of substratumsampled, and 67 individual fishes and one new fish for every 5.5 m2.The standingcrop of fisheswas estimatedat 17 g/m2.Based on the criteria used i.e,, biomassand numberof squaremeters sampled to encountera new speciesor an individual fish!, the fish communitiesalong the diffuser pipe werefound to increasefrom 1990to 1991,

D I SC USS ION

Since 1991,the estimatedstanding crop of fishes rangedfrom 6 Transect1, 1993! to 293 g/m2 Transect5, 1996!.The overall meanstanding crop all years!is 46 g/m2:in 1991,it was42 g/m in 1992,53 g/m2 in 1993,21 g/m2.in 1994,35 g/m ' in 1995,30 g/m2and in 1996, 92 g/m2.The censusof Transect5 in 1996resulted in a high biomassestimate 93 g/m2!; this was due to the chanceencounter with a school of blue trevally or 'omilu

17 Caranx melampygus!that wanderedthrough the path of the video camera.These fishes may have been attracted to the camera due to the noise and light coming from it. If these C. melampygusare removedfrom the biomassestimate, the standingcrop for 1996becomes 56g/m>. The grand mean estimate of thebiomass would then be 40 g/mz all yearscombined!. Goldman and Talbot 975! suggestedthat a reasonablemaximum biomass of coral reef fish is about 200 g/m~. Spaceand cover are important agentsgoverning the distribution of coral reef fishes Sale 1977!.Similarly, the standingcrop of fisheson a reefis correlatedwith the degreeof vertical relief. Thus Brock 954!, using visual techniqueson Hawaiian reefs, estimatedthe standingcrop of fishesto rangefrom 4 g/m~on sandflats to a maximumof 186 g/mz in an areaof considerablevertical relief. Thelarge variation seen in standingcrop of fishes on coral reefs is tied to the structural diversity of the habitat Risk 1972!. Some authors Risk 1972; Gladfelter and Gladfelter 1978; Brock et al. 1979; Ogden and Ebersole 1981; Anderson et al. 1981; Shulman et al. 1983; Shulman 1984; Eckert 1985; Walsh 1985; Alevizon et al. 1985!view reef structureas an importantfactor in determiningthe speciescomposition of coral reef fish communities. Thus some evidence suggeststhat both the biomass and species compositionare influenced by the complexityof the local topography. The substratumin the vicinity of the SandIsland outfall diffuser appearsto be a sandy plain.Sand habitats typically support a lowdiversity of fish speciesand biomass, i.e., biomass rangingfrom 0,5 to 20 g/m> Brock 1954;Brock et al. 1979;Brock andNorris 1989!.The diffuserpipe situatedon a gravelpad with someballast stone placed at theends of mostpipe sectionsprovides additional local topographicalstructure, which has probably influencedthe developmentof thefish community.Because of thesmall graded sizes used, the ballast stone and gravel pad provide only small-scaleshelter, Small-scale shelter favors speciesthat are either small as adultsor juveniles of larger species.The averagesize of the fishescensused in this survey supportsthis contention.Additionally, many of the larger fishes seen especially Acanthurusxanthopterus! were in the vicinity of known areasof topographicalrelief, suchas the set of large tires nearthe start of the diffuser pipe, the discarded55-gallon drum, and the constructiondebris. The accuracyof censusingis lesswith smallerfishes. The data from 1991to presentsuggest that thereis considerablevariation in the fish and invertebratescounts over time. The identificationof somespecies of fishes in this study was not difficult becauseof their size such as adult Acanthurus xanthopterus and Bodianus bilunulatus! or color such as Zanclus cornutus!. In past years some speciessuch as the bluelined snapperor ta'ape Lujtanus kasmira! occurredin suchhigh abundancethat species identification was not difficult. Despitethis, a numberof fisheshave consistently been difficult or impossible to identify becauseof ! poor cameraresolution, ! their rapid movementto cover, ! their small size, or ! their being on the peripheralfield of view. Someof these

18 fishes were small damselfishes probably Chrornishanui and/or C. agilis!, small LutJ'anussp., and small wrasses family Labridae,possibly Cheilinus bimaculatus, Pseudocheilinusspp., Thalassornaspp., or Pseudojuloidescerasinus!. As for abundance,these unidentified fishes areimportant, but they generallycontribute little to the biomassestimates because of their small size. Among the abundantfishes surveyed in mostyears are thejuvenile snappersseen close to the substratum.It hasbeen assumed that thesejuveniles are newly recruitedbluelined snappers or ta'ape Lujtanus kasrnira!, but they may representjuveniles of a numberof other snapper speciesincluding the pink snapperor 'opakapaka Pristipomoides filarnentosus!, which is an important speciesin the Hawaiianbottomfish fishery. If the latter is true, the diffuser may be providing significantlocal habitatto a commerciallyimportant species. The countsfor the parametersmeasured in the 1994fish censusappear to be higher than those for 1993.The changein operationof the cameracommencing in 1994to view the fish community along eachside of the diffuser ratherthan continually crossingover the pipe from side to side as donein previousyears may accountfor someof the increase.The bettercamera resolution in 1994, comparedwith 1993, is probably another reasonfor the increasein the apparentabundance of fisheson the diffuser. In 1995the cameraagain traveledconsistently along one side of the diffuser, but becauseit was often elevated such that identification of fishes was difficult, lower estimates of abundance and biomass resulted. In 1996 the camera again traveledalong one side of the diffuser but generallyviewed the substratumjust directly below, resulting in better coveragefor small speciesthan for larger, more wary species. Overall, the variability in fish counts over the years is influenced by many other factors. Manipulation of the remotelyoperated and tetheredvideo camerais difficult, especiallywhen consideringthat more than 60 m of electricalcable are between the cameraand the operatoron the surfacevessel. Not only doesthe wind createdifficulties with keepingthe surfacevessel on station,but currentsmay interactwith the cableand camera below. Thesefactors serve to move the camera away from the area of interest, making accurate censusing of fishes and invertebrateson the resulting videotapevery difficult at best. Theseproblems, coupled with highly variable turbidity due to the interactionof material dischargedfrom the diffuser ports with local currents, result in highly variable visibility that serves to add variability to the counts. Controlling the sourcesof variationinherent with the useof the remotelycontrolled video systemis difficult if not impossible.The remotelycontrolled video cameraused for the annual inspectionby Departmentof WastewaterManagement personnel probably provides sufficient resolution and information with respectto the physicalstatus of the outfall anddiffuser, but it appearsto be inadequatefor monitoring the statusof fish and macrobenthoson the diffuser.

19 Until an alternative can be found, the remotely operatedvideo system is the only low-cost means available to view the marine communities on the diffuser. Until a more accurate means of visual assessmentis available,the biologicaldata generated by the remotelycontrolled video camerashould be viewedas qualitative, with little statisticalrigor.

REFERENCES CiTED

Alevizon, W,, R. Richardson,P. Pitts and G. Serviss. 1985,Coral zonation and patternsof community structurein Bahamianreef fishes.Bull Mar. ScL 36.304-318.

Anderson, G.R.V,, A,H. Ehrlich, P.R. Ehrlich, J.D. Roughgarden,B.C, Russell and F,H, Talbot, 1981,The communitystructure of coral reef fishes,Am. Nat. 117:476 495.

Brock, R.E. 1982. A critique on the visual census method for assessingcoral reef fish populations.Bull, Mar. Sci. 32:269-276.

Brock, R.E. 1992a. An analysis of the fish communities along the Sand Island-deep ocean outfall using remote video I. 1990 data. Special Rep. 04.02:92, Water Resources ResearchCenter, University of Hawaii at Manoa,Honolulu, Hawaii. 6 pp. Brock, R.E. 1992b.An analysisof the fish communities along the Sand Island deep ocean outfall using remote video II. 1991 data. Special Rep. 04.08:92, Water Resources ResearchCenter, University of Hawaii at Manoa,Honolulu, Hawaii. 14 pp.

Brock, R.E. 1993a.An analysisof the fish communitiesalong the Sand Island deep ocean outfall using remote video III. 1992 data. Special Rep. 01.15:93, Water Resources ResearchCenter, University of Hawaii at Manoa,Honolulu, Hawaii. 15 pp.

Brock, R,E. 1993b. An analysisof the fish and macrobenthosalong the Sand Island deep ocean outfall, 0'ahu, Hawai'i, using remote video IV 1993. Proj. Rep. PR-94-12, W'ater Resources Research Center, University of Hawaii at Manoa, Honolulu, Hawaii. 17 pp. Brock, R.E. 1995. An analysisof the fish and macrobenthosalong the Sand Island Ocean Outfall using remote video: V. 1994 data. Proj. Rep. PR-95-06, Water Resources ResearchCenter, University of Hawai'i at Manoa, Honolulu, Hawai'i. 19 pp.

Brock, R.E. 1996. An analysisof the fish and macrobenthosalong the Sand Island Ocean Outfall using remote video: VI. 1995 data. Proj. Rep. PR-96-06, Water Resources ResearchCenter, University of Hawai'i at Manoa, Honolulu, Hawai'i. 20 pp.

Brock, R.E., C. Lewis and R.C. Wass. 1979. Stability and structure of a fish community on a coral patch reef in Hawaii. Mar. BioL 54:281 292. Brock, R.E. and J.E. Norris. 1989.An analysisof the efficacy of four artificial reef designsin tropical waters.BulL Mar. Sci. 44:934 941. Brock, V.E. 1954. A preliminary report on a method of estimating reef fish populations. J. WildL Mgmt. 18:297-308.

20 Eckert, G.J. 1985. Settlement of coral reef fishes to different natural substrata and at different depths. In Proceedingsof the 5th International Coral Reef Congress,vol. 5, Tahiti, 385-390.

Evans, E.C., ed. 1974.Pearl Harbor biological survey final report. Report No. NUC-TN- 1128, Naval Undersea Center, Hawaii Laboratory.

Gladfelter, W,B., and E.H, Gladfelter. 1978.Fish community structure as a function of habitatstructure on West Indianpatch reefs, Rev, Biol. Trop. 26I'Supplement1!:65 84,

Goldman,B., and F.H. Talbot. 1975.Aspects of the ecology of coral reef fishes. In Biology and geology of coral reefs,ed. O.A. Jonesand R. Endean,vol. III, Biology 2, 124 154. New York: Academic Press.

Ogden,J.C., and J.P. Ebersole.1981. Scale and community structureof coral reef fishes: A long-term study of a largeartificial reef.Mar. Ecol. Prog. Ser.4:97-104.

Ricker, W.E. 1975.Computation and interpretationof biological statisticsof fish populations. Bull. Fish. Res.Bd. Canada191. 382 pp.

Risk, M.J. 1972.Fish diversity on a coralreef in the Virgin Islands.Atoll Res.Bu'll. 153:1 6

Sale, P.F. 1977, Maintenance of high diversity in coral reef fish communities. Am. Nat.. 111:337-359.

SAS Institute, Inc. 1985.SAS user's guide: Statistics,version 5 edition. SAS Institute, Inc. 956 pp. Shulman, M.J. 1984. Resourcelimitation and recruitment patterns in a coral reef fish assemblage. J. Exp. Mar. Biol. EcoL 74:85 109.

Shulman, M,J., J.C. Ogden, J.P. Ebersole, W,N. McFarland, S.L. Miller, and N.G. Wolf. 1983.Priority effects in the recruitmentof juvenile coral reef fishes.Ecology 64:1508- 1513.

Siegel, S. 1956. 1Vonparametricstatistics for the behavioral sciences. New York: McGraw- Hill Book Co. 312 pp,

Walsh, W.J. 1985. Reef fish community dynamics on small artificial reefs: The influence of isolation, habitat structure, and biogeography. Bull. Mar. Sci. 36:357-376.