COMMUNITY STRUCTURE OF FISH AND MACROBENTHOS
AT SELECTED SITES FRONTING SAND ISLAND, OAHU,
IN RELATION TO THE SAND ISLAND DEEP OCEAN OUTFALL,
YEAR 31992





Richard E. Brock






Special Report 01.29:93








January 1993







PREPARED FOR
City and County of Honolulu
Department of Public Works
Project Completion Report
for
The Assessment of the Impact of Ocean Outfalls
on the Marine Environment off Oahu, Hawaii
Project No.:  C-59390
Project Period:  1 November 199231 January 1993
Principal Investigator:  Roger S. Fujioka


WATER RESOURCES RESEARCH CENTER
University of Hawaii at Manoa
Honolulu, Hawaii  96822

ABSTRACT
This report constitutes the third year of an annual monitoring (carried out on 2122 
December 1992 and 25 January 1993) of shallow marine communities inshore of the 
Sand Island deep ocean outfall. This quantitative monitoring effort focuses on benthic 
and fish community structure and is designed to detect changes in these communities. 
Marine communities offshore of Honolulu have received considerable perturbation over 
the last 100 years. Raw sewage was dumped in shallow water until 1978; point and non-
point sources of pollution from both urban sources and industry continue. All of these 
disturbances may serve to obscure any possible impacts from the deep ocean outfall 
discharge. The marine communities show a considerable range in development that is 
probably related to past (historical) impacts. Stations have been sited to take advantage of 
these gradients. Analysis of the first years data showed that there had been no 
statistically significant change in the biological measures (i.e., percent coral cover, 
number of coral species, number of invertebrate species, total number of invertebrates 
counted, number of fish species, total number of fishes counted, and the biomass of fishes 
present at each station) quantified in the study during that period. Hurricane Iniki which 
occurred in September 1992 impacted marine communities along the south shore of 
Oahu. Considerable damage was incurred by the coral communities especially at the 
westernmost study site. Despite the considerable impact of this storm, statistical analysis 
of the most recent data show that the changes which have occurred are not statistically 
significant. 

INTRODUCTION
Purpose
In recent years controversy has arisen regarding the impact that sewage effluent from the Sand Island 
Wastewater Treatment Plant may have on inshore coral reef species. Much of the geographical area of 
concern in this study was impacted by the release of 3 m3/sec (62 mgd) of raw sewage in 10 m of water off 
Sand Island from 1955 to 1977. Starting in 1978 sewage received advanced primary treatment and was 
released further offshore of Sand Island from a deep ocean outfall (6773 m depth). Despite studies that 
demonstrated the recovery of inshore benthic communities once the shallow sewage stress was removed 
(e.g., Dollar 1979), concern continues over the possible impact that the release of sewage from the deep 
ocean outfall may be having in the shallow (< 20 m) marine communities fronting Honolulu and Sand 
Island. Accordingly, this study was undertaken commencing in 1990 in an attempt to quantitatively 
ascertain the impacts that may be occurring. This document presents the results of the third annual survey 
carried out in 1992.

Strategy
Marine environmental surveys are usually performed to evaluate the feasibility of an ecosystem response to 
specific proposed activities. Appropriate survey methodologies reflect the nature of the proposed action(s). 
An acute potential impact (as channel dredging) demands a survey designed to determine the route of least 
harm and the projected rate and degree of ecosystem recovery. Impacts that are more chronic or progressive 
require different strategies for measurement. Management of chronic stress to a marine ecosystem demands 
identification of system perturbations which exceed boundaries of natural fluctuations. Thus a thorough 
understanding of normal ecosystem variability is required in order to separate the impact signal from 
background noise. Infrequent natural events may add considerably to the variability or background 
noise measured in a marine community. In September 1992 Hurricane Iniki struck the Hawaiian Islands 
and impacted some marine communities along Oahus south shore. This rare event has provided this study 
with information on the possible extreme of such natural impacts.
Rare storm events not withstanding, the potential impacts confronting the marine 
ecosystem offshore of Sand Island and Honolulu Harbor are most probably those 
associated with chronic or progressive stresses. Because of the proximity of the 
population center and industry, marine communities fronting Honolulu Harbor are 
subjected to a wide array of impacts not usually occurring in other Hawaiian coral 
communities. Thus a sampling strategy must attempt to separate impacts due to 
wastewater treatment plant effluent on coral reef communities located at some distance 
shoreward from a host of other perturbations occurring in the waters fronting Honolulu.
Honolulu Harbor is the primary commercial port for the State of Hawaii and has been 
so since before the turn of the century (Scott 1968). The harbor is the result of dredging 
what was originally the drainage basin of Nuuanu Stream. This dredging commenced 
before 1900 and periodic maintenance dredging occurs up to the present time; up until 
about 1960 spoils were dropped just outside of the harbor generally to the east of the 
Honolulu Sewer Outfall. Besides shipping, the harbor is ringed with industry; pineapple 
canneries, gas and oil storage, and numerous other businesses are operating or have 
operated in the past. Storm drainage into the harbor and nearby Keehi Lagoon carries 
runoff from Honolulus streets and suburbs into the ocean. Pollution is well-known in the 
harbor and Cox and Gordon (1970) cite references describing these conditions as early as 
1920. Sewage has been pumped into the ocean offshore of Kewalo and Sand Island since 
the 1930s. These early inputs were all raw sewage released in water not exceeding 20 m 
deep. The actual point of release varied through time as different pipes were constructed 
and used. The multitude of perturbations that have occurred or are still occurring in 
shallow water (< 20 m) up until the construction of the present deep water outfall in 1978 
may serve to obscure the impacts of the present discharge.
The waters fronting Sand Island into which the deep ocean outfall discharges may be 
considered in terms of gradients. There are numerous gradients due to point (storm 
drains, streams, etc.) and non-point inputs into Honolulu Harbor and environs from the 
above-mentioned activities. Because many of these inputs have been occurring for a 
considerable period of time, the species composition and functional relationships of the 
benthic and fish communities at any given location in the waters offshore of Honolulu are 
those that have evolved under the influence of these ongoing perturbations.
As noted above, if impacts are occurring in the shallow marine communities fronting 
Honolulu due to the deep ocean outfall, these are probably chronic in nature thus causing 
a slow decline in the communities so impacted. Gradients of stress or impact should 
be evident with distance from impact source(s). Thus to quantitatively define these 
impacts, one should monitor these communities through time in areas suspected of being 
impacted as well as in similar communities at varying distances away from the suspected 
source(s). This rationale has been used in developing the sampling strategy for this study.

MATERIALS AND METHODS
The quantitative sampling of macrofauna of marine communities presents a number of problems; many of 
these are related to the scale on which one wishes to quantitatively enumerate organism abundance. Marine 
communities in the waters fronting Sand Island may be spatially defined in a range on the order of a few 
hundred square centimeters (such as the community residing in a Pocillopora meandrina coral head) to 
major biotopes covering many hectares. Because considerable interest focuses on visually dominant corals, 
diurnally exposed macroinvertebrates, and fishes, we designed a sampling program to attempt to delineate 
changes that may be occurring in communities at this scale.
Three sites were selected for the monitoring of benthic and fish community response 
to possible sewage impacts. The approximate locations of these sites (shown in Fig. 1) 
are close to some stations used by Dollar (1979). The sites and the rationale for their 
selection are given below:
Site A (Kewalo Landfill) was utilized as a control area lying east of the present deep 
ocean outfall in about 16 m of water (Fig. 1). Prevailing currents create a westerly 
movement of sewage effluent (Dollar 1979) thus the shallow Kewalo Landfill area is 
probably not directly impacted. At this location, corals occur in areas of emergent 
limestone. Local coverage over short linear distances may exceed 30 percent. This station 
is in the vicinity of Dollars (1979) Station 2.
Site B (Kalihi Channel) is located about 120 m east of the Kalihi Entrance channel in 
approximately 15 m of water. This station is about 900 m west of the bypass (old) outfall 
in an area heavily impacted by the old (19551977) shallow water discharge, and is very 
close to Dollars (1979) Station 14. Again there is emergent limestone at this station but 
coral coverage is low (< 1%).
Site C (Reef Runway) is located in an area of complex limestone substratum, in water 
ranging from 7.5 to 12 m deep fronting Honolulu International Airports Reef Runway. 
This station is close to Brocks (1986) station that was monitored quarterly in 197778 
(AECOS, Inc. 1979) and again in 1986. It is close to Dollars (1979) Station 19. This site 
was moderately impacted by the old shallow water sewage outfall (Dollar 1979).
At each site two transect lines have been permanently established using metal stakes 
and plastic coated no. 14 copper wire. Transects are 20 m in length and have an 
orientation perpendicular to shore. Two transects have been established at each location 
to provide some replication. Both sample approximately the same benthic community. 
On each transect there are five permanently marked locations (0 m, 5 m, 10 m, 15 m, and 
20 m) for taking photographs of the benthic communities. Cover estimates were also 
made in the field with a 1 ? 1 m quadrat placed at the -1 to 0 m, 4 to 5 m, 9 to 10 m, 14 to 
15 m, and 19 to 20 m marks on the transect line in each survey.
Fish abundance and diversity is often related to small-scale topographical relief over 
short linear distances. A long transect may bisect a number of topographical features 
(e.g., coral mounds, sand flats, and algal beds) thus sampling more than one community 
and obscuring distinctive features of individual communities. To alleviate this problem, a 
short transect (20 m in length) has proven adequate for sampling many Hawaiian benthic 
communities (see Brock 1982, Brock and Norris 1989).
Information collected at each transect location includes a visual assessment of fishes, 
benthic quadrats for cover estimates of sessile forms (algae, corals, and colonial 
invertebrates) and counts along the transect line for diurnally exposed motile 
macroinvertebrates. Fish censuses are conducted over a 20 ? 4 m corridor (the permanent 
transect line) and all fishes within this area to the waters surface are counted. A single 
diver equipped with SCUBA, and a slate and pencil enters the water, counts and notes all 
fishes in the prescribed area (method modified from Brock 1954). Besides counting the 
numbers of individuals of all fishes seen, the lengths of each is estimated; these length 
data are later used in the estimation of fish standing crop by linear regression techniques 
(Ricker 1975). Species specific regression coefficients have been developed over the last 
thirty years by the author and others at the University of Hawaii, Naval Undersea Center 
(see Evans 1974) and the Hawaii State Division of Aquatic Resources through capturing, 
weighing and measuring fishes; for many species the coefficients have been developed 
using sample sizes in excess of a hundred individuals. Two weeks were allowed to elapse 
from the time of station selection and marking to the time of the first fish census to 
reduce the bias caused by wary fishes. The same individual (R. Brock) performed all fish 
censuses to reduce bias. 
Besides frightening wary fishes, other problems with the visual census technique 
include the underestimation of cryptic species such as moray eels (Family Muraenidae) 
and nocturnal species, e.g., squirrelfishes (Family Holocentridae), bigeyes or aweoweos 
(Family Priacanthidae), etc. This problem is compounded in areas of high relief and coral 
coverage that affords numerous shelter sites. Species lists and abundance estimates are 
more accurate for areas of low relief, although some fishes with cryptic habits or 
protective coloration (e.g., the nohus, Family Scorpaenidae; the flatfishes, Family 
Bothidae) might still be missed. Obviously the effectiveness of the visual census 
technique is reduced in turbid water and species of fishes which move quickly and/or are 
very numerous may be difficult to count. Additionally, bias related to the experience of 
the diver conducting counts should be considered in making comparisons between 
surveys. In spite of these drawbacks, the visual census technique probably provides the 
most accurate nondestructive assessment of diurnally active fishes presently available 
(Brock 1982).
A number of methods were utilized to quantitatively assess benthic communities at 
each station; these methods included the use of photographs taken at locations marked for 
repeated sampling through time (each covering 0.67 m2) and 1 ? 1 m quadrats also placed 
at marked locations for repeated measurements. The photographs and quadrats were both 
used to estimate coverage of corals and other sessile forms. Photographs provide a 
permanent record from which to estimate coverage and were used in 1991 and 1992; the 
1 ? 1 m quadrats were used for an in the field appraisal of coverage in the three 
surveys. Cover estimates from photographs and quadrats were all recorded as percent 
cover. Diurnally exposed motile macroinvertebrates greater than 2 cm in some dimension 
were censused in the same 4 ? 20 m corridor used in the fish counts. 
If macrothalloid algae were encountered in the 1 ? 1 m quadrats or photographs, they 
were quantitatively recorded as percent cover. Emphasis was placed on those species that 
were visually dominant and no attempt was made to quantitatively assess the multitude of 
microalgal species that constitute the algal turf so characteristic of many coral reef 
habitats. 
As requested by permit agencies, simple physical measurements were made at the 
three sites while in the field. Measurements were made of percent oxygen concentration 
and temperature with a YSI Model 57 Oxygen meter, salinity was taken with a hand held 
refractometer and a 12-inch secchi disk was used to determine water clarity.
Data were subjected to simple non-parametric statistical procedures provided in the 
SAS Institute statistical package (SAS Institute 1985). Non-parametric methods were used 
to avoid meeting requirements of normal distribution and homogeneity of variance in the 
data. Data analysis utilized the Kruskal-Wallis one-way analysis of variance which was 
used to discern statistically significant differences among ranked means for each transect 
site and sample period; this procedure is outlined by Siegel (1956) and Sokal and Rohlf 
(1981). The a posteriori Student-Neuman-Keuls multiple range test (SAS Institute, Inc. 
1985) was also used to elucidate differences between locations.
During the course of the fieldwork, an effort was made to note any green sea turtles (a 
threatened species) within or near the study sites.
RESULTS
Field sampling was first undertaken on 2729 December 1990. Station locations were selected and marked 
in November 1990 The permanent pins were deployed about a week later. Figure 1 presents the 
approximate locations of the three stations, each with a pair of transects; Figures 2, 3, and 4 are sketches 
showing the orientation of the permanent photographic quadrats on each transect line. The 1991 data were 
collected on 56 December 1991, and the 1992 information was taken on 2122 December 1992 as well as 
on 25 January 1993 from the same locations.
Malfunction of a new Nikonos V camera caused the loss of all photographic quadrat 
data for all stations in the first (1990) field effort. Subsequently, the annual photography 
effort has been carried out by Mr. A. Muranaka (City and County of Honolulu). 
However, the 1990 visually assessed square meter quadrat data provided information on 
benthic coverage in this first annual effort. Subsequent surveys have used both 
photographic and quadrat methods to assess the benthic communities. It should be noted 
that the numbering of photoquadrats has changed from the 1991 and 1992 surveys but the 
locations are the same.
The results are presented below by station. All transects have an orientation that is 
perpendicular to shore.
Site A  Kewalo Landfill Station
This station is located 600 m offshore of the old Kewalo Landfill in water ranging from 17 to 18 m deep on 
a substratum dominated by limestone with moderate coral community development. The two transects are 
35 m apart out of visual range of one another (see Fig. 2). Water clarity at this station is usually in the range 
from 15 to 20 m.
A summary of the data collected at Transect 1 in December 1992 is presented in 
Table 1. In the quadrat survey, six coral species were encountered having a mean 
estimated coverage of 18 percent; the dominant species are Porites lobata and 
Pocillopora meandrina. One algal species (Amansia glomerata) was noted in the 
quadrats. The macroinvertebrate census noted one cone shell (Conus lividus), two 
polychaetes (the Christmas tree worm - Spirobrachus giganteus corniculatus and the 
featherduster worm - Sabellastarte sanctijosephi), and three echinoderms (the long-
spined sea urchin or wana - Echinothrix diadema, the sea star - Linckia diplax, and the 
boring sea urchin - Echinostrephus aciculatum). The results of the fish census carried out 
at Transect 1 are given in Appendix Table A. Table 2 presents the results of the 
photographic survey carried out on 3 August 1992. The mean coral coverage in the 
photographic survey was estimated to be 12.9 percent with Porites lobata being the 
dominant coral. Interestingly, in this survey an adult helmet shell (Cassius cornuta) was 
present in the 20 m quadrat occupying about 3 percent of the substratum.
In total 36 species of fishes and 312 individuals were encountered on Transect 1. 
The most common species include the yellowstripe goatfish or weke (Mulloides 
flavolineatus), the damselfishes (Chromis ovalis and C. vanderbilti) as well as the sleek 
unicornfish or kala holo (Naso hexacanthus). The standing crop of fishes on this transect 
was estimated to be 736 g/m2 and the species contributing most heavily to this biomass 
were the weke (Mulloides flavolineatus - 78% of the total biomass) and the kala holo 
(Naso hexacanthus - 7% of the total biomass).
Transect 2 was also established offshore of the Kewalo Landfill approximately 35 
m west of Transect 1 in water ranging from 17 to 18.2 m deep. Table 3 presents a 
summary of the biological information collected at this transect site. The quadrat survey 
noted two macroalgal species (Amansia glomerata and Botryocladia scottsbergii) and six 
coral species (Porites lobata, P. compressa, Pocillopora meandrina, P. eydouxi, 
Leptastrea purpurea, and Montipora verrucosa) having an average coverage of 20.7 
percent. The largest contributor to this coverage was Porites lobata. The estimated coral 
coverage has decreased approximately 8 percent since the 1991 survey, probably due to 
Hurricane Iniki which impacted the Hawaiian Islands on 11 September 1992. There were 
numerous broken coral fragments in the quadrats and elsewhere. The invertebrate census 
counted two polychaete species (the Christmas tree worm - Spirobranchus giganteus 
corniculatus and the featherduster worm - Sabellastarte sanctijopsephi), the pearl oyster 
(Pinctado marginifera), hermit crab (Aniculus strigatus), and one sea urchin species 
(Echinothrix diadema). The photographic quadrat survey carried out on 3 August 1992 
(before the hurricane) noted three algal species (mean coverage 2%), an unidentified red 
sponge, and four coral species with a mean coverage of 27.8 percent (Table 2).
The results of the fish census are presented in Appendix Table A. Twenty-six fish 
species were identified (240 individuals) on this transect; the most abundant species 
included the yellowstripe goatfish or weke (Mulloides flavolineatus) and the damselfish 
(Chromis vanderbilti). The standing crop of fishes was estimated to be 247 g/m2. The 
species that contributed the most to this estimated weight was the yellowstripe goatfish or 
weke (Mulloides flavolineatus - 93% of the total biomass).
Site B  Kalihi Entrance Channel
Two transects (numbers 3 and 4) were established on a limestone substratum about 120 m east of the Kalihi 
Entrance Channel in 13.7 to 15 m of water. This station is located about 2.2 km seaward of Mokauea Island 
situated in Keehi Lagoon, and about 900 m west of the old outfall which is now used as an emergency 
bypass. Much of the substratum in the vicinity of this station is comprised of sand and rubble. An area of 
low emergent limestone approximately 60 m wide and 110 m in length with the long axis oriented 
perpendicular to shore is present. Transect 3 is located on the deeper end of this hard substratum area. 
Transect 4 parallels Transect 3 but is shoreward of this and approximately 8 m to the west (see Fig. 3). 
During the 1992 survey water clarity at this station ranged from 10 to 27 m during our visits. The lack of 
appropriate hard substratum necessitated establishing the two transects at this station on an end to end 
fashion relatively close to one another (8 m apart). Because of the close proximity, the fish censuses at 
these stations were carried out on both transects prior to any other data collection.
Transect 3 has an orientation perpendicular to shore on the limestone substratum in 
water 14.6 to 15 m deep. Table 4 presents a summary of the biological observations made 
at Transect 3. The quadrat survey noted two algal species (Desmia hornemannii and limu 
kohu or Asparagopsis taxiformis together having a mean coverage of 0.1%), the red 
encrusting sponge (Spirastrella coccinea), one soft coral (Palythoa tuberculosa), and four 
coral species (Porites lobata, Pocillopora meandrina, Montipora verrucosa, and M. 
patula) having a mean estimated coverage of 2 percent (a decrease of 2% over previous 
surveys). The invertebrate census noted one rock oyster (Spondylus tenebrosus), octopus 
or hee (Octopus cyanea), and three sea urchin species (Echinostrephus aciculatum, 
Echinometra mathaei and Echinothrix diadema) as well as the starfish (Linckia diplax). 
The photographic quadrat survey found an unidentified red sponge species and four coral 
species (Porites lobata, Pocillopora meandrina, Montipora verrucosa?, and Fungia 
scutaria) having a mean coverage of 4 percent.
The fish census (App. Table A) found 15 species, 33 individuals and an estimated 
standing crop of 30 g/m2. The most abundant fishes at Transect 3 included the manybar 
goatfish or moano (Parupeneus multifasciatus) and the lei triggerfish or humuhumu lei 
(Sufflamen bursa). The fish species contributing heavily to the biomass on Transect 3 
included a single tableboss or aawa (Bodianus bilunulatus making up 38% of the 
biomass), an orangebar surgeonfish or naenae (Acanthurus olivaceus - 19% of the 
total), a rockmover (Novaculichthys taeniourus - 14%), and five humuhumu lei 
(Sufflamen bursa - 14% of the total weight). 
Transect 4 also sampled the benthic and fish community present in the vicinity of the 
Kalihi Entrance Channel. As with the previous transect, Transect 4 sampled the limestone 
substratum at a depth ranging from 13.7 to 14 m. Table 5 presents a summary of the 
biological data collected on Transect 4. The quadrat survey noted two algal species 
(Desmia hornemannii and limu kohu or Asparagopsis taxiformis), the red sponge 
(Spirastrella coccinea), and four coral species (Porites lobata, Pocillopora meandrina, 
Montipora verrucosa, and M. patula). Coral coverage was estimated to be 3.2 percent 
and both Porites lobata as well as Pocillopora meandrina were the major contributors to 
this coverage. The invertebrate census noted one small cone shell (Conus lividus), four 
sea urchin species (the boring urchin - Echinostrephus aciculatum, the green urchin 
Echinometra mathaei, and long spined urchin or wana - Echinothrix diadema) as well as 
the starfish Linckia diplax. In the photographic quadrat survey an unidentified red sponge 
species was seen and two corals (Porites lobata and Pocillopora meandrina) having a 
mean coverage of 6.3 percent. 
The fish census noted 27 individual fishes among 9 species (App. Table A). The most 
common fishes present on this transect include the twospot wrasse (Cheilinus 
bimaculatus) and the lei triggerfish or humuhumu lei (Sufflamen bursa). The standing 
crop of fishes on Transect 4 is estimated to be 14 g/m2 and the important contributors to 
this biomass include a stripebelly puffer or keke (Arthron hispidus - 46% of the total) and 
the lei triggerfish or humuhumu lei (Sufflamen bursa) which contributed 30 percent of the 
total weight present at this transect site.
Site C  Reef Runway 
Two transects (numbers 5 and 6) were established on limestone substratum offshore of 
the Honolulu International Airport Reef Runway. This station lies between 760 and 840 
m seaward of the runway in water ranging from 9.1 to 11.6 m deep. The substratum of 
this area is a mosaic of emergent limestone spur and groove formations grading seaward 
into a series of low limestone mounds. The general orientation of the spur and groove 
formations is perpendicular to the shoreline and direction of usual wave impact. The 
spurs are from 5 to 40 m in width, 30 to 80 m in length and are spaced from 10 to 100 m 
apart. Sand is the dominant substratum in the intervening areas. The maximum 
topographical relief formed by these spurs is about 3.5 m. Just seaward of the spurs and 
grooves is a zone of low emergent limestone; these patches of hard bottom are from 5 
? 10 m to several hundred square meters in size; spacing of these limestone areas is 
between 10 to 50 m and sand is again found in the interven-ing areas. Corals are 
restricted to the areas of hard substratum. Water clarity at this station ranged from 7 to 20 
m during our 1992 visits; usually the clarity did not exceed 12 m. Hurricane Iniki caused 
considerable damage to the benthic communities at Station C. A large (approximately 60 
m in diameter) sand patch located between Transects 5 and 6 has disappeared and has 
been replaced by coral rubble. Much of the hard substratum on both transects was broken 
and the underlying limestone rock exposed and crevices and holes were filled with coral 
rubble. The result of this has been a change in the abundance of both invertebrates and 
fishes at this location.
Both transects were established on spurs or ridges of limestone (see Fig. 4). Transect 
5 was established on a limestone ridge at a depth of 9.1 to 11 m. Table 6 presents the 
results of the biological survey carried out at Station 5. The quadrat survey noted the 
coralline algal species (Porolithon onkodes), two soft corals (Anthelia edmondsoni and 
Palythoa tuberculosa), and four coral species (Porites lobata, P. compressa, Pocillopora 
meandrina, and Pavona duerdeni) having a mean coverage of 0.4 percent. This coverage 
is down from the previous survey (2.1%). The invertebrate census found one hermit crab 
(Calcinus herbstii) and two sea urchin species, the green sea urchin (Echinometra 
mathaei) and the black sea urchin (Tripneustes gratilla). The photographic quadrat 
survey completed after Hurricane Iniki noted the encrusting coralline algae (Porolithon 
onkodes), soft coral (Palythoa tuberculosa), and two coral species (Porites lobata and 
Pocillopora meandrina) having a mean estimated coverage of 0.2 percent (down 2% 
from the previous year).
The fish census (App. Table A) counted 136 individuals amongst 23 species. The 
most common species included the manybar goatfish or moano (Parupeneus 
multifasciatus), the brown surgeonfish or maiii (Acanthurus nigrofuscus), and the 
goldring surgeonfish or kole (Ctenochaetus strigosus). The standing crop of fishes on 
Transect 5 was estimated to be 
69 g/m2; the most important contributors to the estimated standing crop include the 
moano (Parupeneus multifasciatus - 19% of the total), the kole (Ctenochaetus strigosus - 
36% of the biomass), the maiii (Acanthurus nigrofuscus - 9%), and the black 
triggerfish or humuhumu eleele (Melichthys niger - 9%). 
Transect 6 was established approximately 80 m seaward of Transect 5. The 
substratum at Transect 6 was similar to Transect 5 and is situated on a limestone spur that 
is about 40 m wide and 80 m long. Water depth at this site varies between 10.7 to 11.6 m. 
A summary of the biological observations made on Transect 6 is given in Table 7. The 
quadrat survey found two algal species (Porolithon onkodes and Cladymenia pacifica 
having a mean coverage of 6.5%), one soft coral (Anthelia edmondsoni), and six coral 
species (Porites lobata, P. compressa, Pocillopora meandrina, Montipora patula, M. 
verrucosa, and Pavona duerdeni) having a mean coverage of 3.1 percent. This coral 
coverage estimate is down 3.6 percent from last years survey. The census of 
macroinvertebrates noted three species:  the terebellid polychaete worm (Loimia 
medusa), the green sea urchin (Echinometra mathaei), and the black sea urchin 
(Tripneustes gratilla). The photo quadrat survey noted coralline algae (Porolithon 
onkodes) with a mean coverage of 15 percent as well as three coral species (Porites 
compressa, P. lobata, and Pocillopora meandrina) with a mean coverage of 5 percent. 
The fish census found 247 individuals belonging to 36 species in the 4 ? 20 m census 
area. The most abundant fishes on Transect 6 included the damselfish (Chromis 
vanderbilti), the brown surgeonfish or maiii (Acanthurus nigrofuscus), the goldring 
surgeonfish or kole (Ctenochaetus strigosus), and the plankton feeding surgeonfish 
(Acanthurus thompsoni). The standing crop of fishes on this transect was estimated to be 
108 g/m2; the largest contributors to this biomass included the bullethead parrotfish or 
uhu (Scarus sordidus - 22% of the total), the goldring surgeonfish or kole (Ctenochaetus 
strigosus - 17%), the black triggerfish or humuhumu eleele (Melichthys niger - 11%), 
and the brown surgeonfish or maiii (Acanthurus nigrofuscus - 8%).
There were from one to three green sea turtles (Chelonia mydas) usually present at 
Transect 6 in the past. No green turtles were seen during the 1992 survey work in the 
vicinity of Transects 5 and 6. A short underwater reconnaissance of the usual resting area 
(a small ledge under a large Porites lobata colony) revealed that the resting site had been 
completely covered with coral rubble from the hurricane (as had most of the depressions 
in the surrounding area).
Physical measurements were made on the morning of 22 December 1992. These data 
are presented in Table 8. Little variation was noted in temperature (22.6 to 22.8C), 
percent oxygen saturation (102 to 105%) or salinity (all 34) despite the fact that 
measurements for oxygen and temperature were made both at the surface and about 1 m 
above the bottom. In all cases the secchi disk measurements did not yield an extinction 
value; water clarity was such that the disk was still plainly visible on the bottom from the 
surface. As has been suggested previously, a better method of determining water clarity 
might be to collect water samples and measure turbidity with a nephalometer in the 
laboratory.
The biological data for all three surveys (1990, 1991, and 1992) are summarized as 
means for each transect in Table 9. The 1990 and 1991 data are from Brock (1992a, 
1992b). Differences are apparent for some of the parameters between the three years. 
Some change is evident in the benthic measures (such as coral cover) between the 1991 
and 1992 (pre- and post-hurricane) surveys and this is to be expected. Despite these 
changes the Kruskal-Wallis ANOVA shows that there are no statistically significant 
changes from the 1990 survey to the most recent 1992 field effort for the mean coral 
cover at a sample site (P > 0.69, df = 2, N.S.), mean number of coral species at each site 
(P > 0.28, df = 2, N.S.), mean number of invertebrate species on a transect (P > 0.52, df = 
2, N.S.), mean number of individual invertebrates at each location (P > 0.74, df = 2, 
N.S.), mean number of fish species on a transect (P > 0.36, df = 2, N.S.), mean number of 
individual fish at a sample site (P > 0.16, df = 2, N.S.), and mean standing crop at a 
station (P > 0.44, df = 2, N.S.). The Student-Neuman-Keuls multiple range test likewise 
demonstrated that there are no statistically significant differences among these parameters 
between the six stations and three sample periods.
The biological parameters measured in the three surveys (i.e., number of coral 
species, percent cover, number of macroinvertebrate species, number of fish species, 
number of individual fish and biomass of fishes), point to the fact that the Kewalo 
Landfill site has the most diverse communities, followed by the Reef Runway. The least 
diverse appears to be the Kalihi Entrance Channel site; this ranking has not changed over 
the three surveys. The low biological diversity at the Kalihi Entrance Channel site is not 
surprising in view of the fact that this station was heavily impacted by the old shallow 
water outfall until 1978. 
From a commercial fisheries standpoint, a number of important species have been 
encountered in the vicinity of the Kewalo Landfill transects and Reef Runway sites 
including goatfishes (weke - Mulloides flavolineatus and wekeula - M. vanicolensis), 
amberjack or kahala (Seriola dumerili), emperor or mu (Monotaxis grandoculis), uku 
(Aprion virescens), and the squirrelfish or menpachi (Myripristes amaenus).
DISCUSSION
Since their delineation in December 1990, the six transects have been visited on a number of occasions to 
insure that permanent markers are remaining in place, etc. During these visits reconnaissance surveys have 
been carried out in the areas surrounding the selected stations. At a minimum, these qualitative surveys 
have covered about 4 hectares around each of the three sites. These qualitative observations suggest that the 
marine communities sampled at the three stations are representative of those found in the surrounding 
areas.
The working hypothesis is that all three study sites, being situated in relatively 
shallow water, are outside of the zone of influence of the present deep water outfall. 
However if impacts from the present deep ocean outfall are occurring to the shallow 
water coral reef areas shoreward of the outfall, our monitoring should be able to 
quantitatively discern these impacts. Because of bottom time constraints, potential 
dangers with deep diving and the fact that coral community development is usually 
greatest in water less than 30 m of depth, the placement of biological monitoring stations 
was restricted to waters 20 m or less in depth in this study. Monitoring the shallow water 
stations provides additional information regarding the recovery of these communities to 
the perturbation of raw sewage released from the old shallow water outfall that was 
terminated in 197778. Dollars (1979) study showed that the Kewalo Landfill station 
was not directly impacted by the old outfall but the Kalihi Entrance Channel station was 
acutely perturbed and the station offshore of the Reef Runway received an 
intermediate level of disturbance. Additionally in the mid-1970s the reef runway was 
constructed which must have contributed to the disturbance of benthic communities at 
this station (Chapman 1979). The results of these impacts are still evident in the average 
coral cover estimates made at these stations: the mean coverage offshore of the Kalihi 
Entrance Channel is only 3 percent, at the Reef Runway station it is 4 percent, and 
offshore of the Kewalo Landfill it is 23 percent.
The shallow marine ecosystem fronting Sand Island and Honolulu has received 
considerable perturbation from human activity over the last 100 years. Among the 
disturbing influences was the disposal of raw sewage effluent in shallow water between 
the 1930s and 197778 when the deep ocean outfall became operational. In the period 
from 1955 through 1977 the shallow outfall released 3 m3/sec (62 mgd). Dollar (1979) 
noted two distinct zones of impact to marine communities:  the area of acute 
perturbation was an ellipse 500 m to the east and 1000 m to the west of the outfall. 
Outside of this area the impacts were evident in a decreasing gradient with distance from 
the outfall. The maximal extent of impact attributed to this sewage input was 1.9 km to 
the east and 5.8 km to the west of the outfall. The ellipsoid shape of the zone of influence 
was attributed to the predominant westerly direction of current flow.
The Kewalo Landfill station is 4.75 km east and inshore of the terminus of the deep 
ocean outfall, the Kalihi Entrance Channel station is about 2.1 km east and inshore of the 
terminus and the Reef Runway station is about 3.25 km inshore and west of the deep 
ocean outfall terminus (Fig. 1). Presumably the present outfall releases the sewage below 
the thermocline and little interaction occurs with the inshore biota. If however the 
material was carried into inshore waters, impacts to shallow marine communities would 
occur in those communities situated primarily to the west of the outfall based on Dollars 
(1979) findings.
The Kewalo Landfill station served as a control site in this study. Despite the fact 
that coral coverage and fish community development is greater at this location, the 
Kewalo Landfill station has received perturbations in the past. The two transects (T-1 and 
T-2) that sample the Kewalo Landfill site are situated close to an old, non-operable 
sewage discharge pipe. Operations utilizing this pipe ceased sometime prior to 1955, and 
the pipe was probably used sometime in the 1940s (Mr. A. Muranaka, Oceanographic 
Team, Division of Wastewater Management, City and County of Honolulu). The 
development of Kewalo Basin and the entrance channel in the mid-1930s would have 
created considerable turbidity that probably impacted this site, some 200 m to the west. 
From the historical perspective, human induced perturbations have occurred in probably 
all marine communities situated in shallow waters fronting Honolulu over the last 100 
years. The Kewalo Landfill site was selected as the control site for this study because 
of relatively diverse coral and fish communities present as well as its location well to the 
east of the present deep ocean outfall (presumably out of the zone of influence).
On 11 September 1992 the Hawaiian Islands were struck by Hurricane Iniki. The 
hurricane passed directly over Kauai with sustained winds of 144 mph and gusts to 172 
mph resulting in considerable damage to improvements and forests of that island and the 
west (leeward) coast of Oahu. To a lesser extent, high surf caused damage to marine 
communities along the south, east and west shores of Oahu, Kauai, Maui, Lo(,a)nai 
and Hawaii; this damage was primarily to coral communities. In many areas a large 
amount of sand and other loose material was moved and/or advected out of the shallow 
areas (i.e., depths of less than 27 m) into deeper waters. On Oahu, storm waves 
emanating from the southeast were estimated to exceed 6 to 7 m in height and were 
breaking in water at least 20 m deep (personal observations). 
Storm damage to benthic and fish communities is frequently patchy resulting in a 
mosaic of destruction (personal observations, Walsh 1983) and occasional storm events 
generating high surf are important factors in determining the structure of Hawaiian coral 
communities (Dollar 1982). Numerous studies have shown that storm generated surf may 
keep coral reefs in a non-equilibrium or sub-climax state (Grigg and Maragos 1974; 
Connell 1978; Woodley et al. 1981; Grigg 1983). Indeed, the large expanses of near-
featureless lava or limestone substratum present around much of the Hawaiian Islands at 
depths less than 30 m attest to the force and frequency of these events (Brock and Norris 
1989). These same wave forces also impact fish communities (Walsh 1983).
Hurricane Iniki caused damage to coral communities at all three study sites; the 
greatest impact occurred to the benthic communities at Station C (Reef Runway) where 
areas of the Porolithon covered substratum (up to 1 ? 2 m in area and up to 0.75 m in 
depth) were completely removed; other sites were entirely covered with coral rubble at 
scales from 10 m2 to over 30 m2. In some cases a blanket up to 0.5 m deep of rubble 
buried coral colonies or has killed the lower portions of larger colonies. The hurricane 
broke many coral colonies into pieces; some of these have survived where they have 
lodged into the substratum. These live fragments are responsible for the increase in the 
number of coral species seen in some quadrats between the pre- (1991) and post-
hurricane (present) surveys. This same phenomenon (i.e., live fragments) has also served 
to lessen the decrease in coral cover encountered in some of the quadrats where coverage 
was low prior to the storm. Despite these large changes, many of the benthic components 
survived as shown in the results above. However, since Hawaiian corals are relatively 
slow growing, it will be years before the impact of this large storm will no longer be 
evident in the benthic communities of the study sites.
The hurricane also impacted the fish communities at the sample sites. Coral rubble 
deposited in depressions serves to lessen the rugosity of the submarine topography (i.e., 
shelter available to fishes). The loss of local shelter causes fishes to move and take up 
residence elsewhere; at Stations 5 and 6 (Reef Runway) where considerable rubble was 
present, many of the resident fishes (such as a school of emperor or mu (Monotaxis 
grandoculis) were no longer on Transect 6 but had moved about 100 m east to an area 
where the coral and benthic community remained relatively intact.
Despite the impact of Hurricane Iniki, the summary data in Table 9 which spans three 
years (December 1990 and December 1992) shows that there has been no statistically 
significant change in the biological parameters measured in this study. The most variable 
parameters through time have been the number of fish censused and the estimated 
standing crop of fish; these changes have been greatest at the Kewalo Landfill site. 
Relative to many other locations in the Hawaiian Islands, the fish community is well 
developed at the Kewalo Landfill station. The high standing crop estimates in 1990 and 
1992 are much greater than found on most coral reefs; the maximum fish standing crop 
encountered on natural coral reefs is about 200 g/m2 (Goldman and Talbot 1975; Brock et 
al. 1979). There are two explanations for the high biomass of fishes at the Kewalo 
Landfill site; these are (1) the shelter created by the old sewer pipe locally enhances the 
fish community and (2) chance encounters with roving predators or planktivorous and/or 
other schooling species during censuses.
Space and cover are important agents governing the distribution of coral reef fishes 
(Risk 1972; Sale 1977; 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). Similarly, the standing crop of fishes on a reef is 
correlated with the degree of vertical relief of the substratum. Thus Brock (1954) using 
visual techniques on Hawaiian reefs estimated the standing crop of fishes to range from 4 
g/m2 on sand flats to a maximum of 186 g/m2 in an area of considerable vertical relief. If 
structural complexity or topographical relief is important to coral reef fish communities, 
then the addition of materials to increase this relief in otherwise barren areas may serve to 
locally increase the biomass of fish. The additional topographical relief is usually in the 
form of artificial reefs but any underwater construction activity (such as the deployment 
of a sewer line) will have a similar effect. The old sewer discharge pipe is set above the 
seafloor creating considerable local topographical relief (about 2 m high) in an area 
where the maximum natural vertical relief does not exceed 25 cm. The shelter and high 
topographical relief must foster greater development of the fish community (see Brock 
and Norris 1989).
Chance encounters with large roving predators (such as uku, Aprion virescens; mu, 
Monotaxis grandoculis; kahala, Seriola dumerili; papio, Caranx melampygus or C. 
orthrogrammus) or schools of planktivorous fishes (opelu, Decapterus macarellus; kala 
holo, Naso hexacanthus; kala lolo, N. brevirostris; lauwiliwili, Chaetodon miliaris; 
mamo, Abudefduf abdominalis) or other schooling species (weke, Mulloides 
flavolineatus) may greatly increase the counts and biomass of a particular transect. The 
presence of the sewer pipe serves to focus numerous predators and schooling fishes in the 
vicinity of the two transects at the Kewalo Landfill site and an encounter with these 
fishes during a census will result in high biomass estimates. Chance encounters with a 
small school of mu or emperor (Monotaxis grandoculis) at Station 6 (Reef Runway) 
accounts for 51 percent of the biomass at that station in 1990; on Transect 2 (Kewalo 
Landfill) the two planktivorous surgeonfishes (kala holo and kala lolo - Naso 
hexacanthus and N. brevirostris) accounted for 40 percent of the biomass and the two 
roving predators the kahala (Seriola dumerili) as well as the papio (Caranx 
orthogrammus) contributed 21 percent to the biomass estimate for that transect in 1990. 
In 1991 these planktivorous surgeonfishes and some predators were present around 
Transects 1 and 2 at the Kewalo Landfill site but did not enter the actual census area 
while the counts were proceeding, thus do not appear in the data. In 1992 the large school 
of yellowstripe goatfish or weke (Mulloides flavolineatus) that are resident to the old 
sewer pipe made up 78 percent of the biomass of Transect 1 and 93 percent of the 
standing crop present on Transect 2.
Making biological measurements underwater can often be a time-consuming process; 
use of the photographic technique lessens bottom time in measuring coral and other 
benthic species coverage. However as noted by Brock (1992b) inspection of the results of 
the coral coverage data from visual appraisal of quadrats in the field relative to the data 
from the photographic method points out several things. First, mean coral coverage 
estimates are in reasonable agreement by either method and the regression of visual 
versus the photographic coverage data show a statistically significant relationship. 
However, the photo quadrat technique does not discern small coral colonies or other 
small colonial benthic species such as the soft coral Anthelia edmondsoni; these are easily 
seen in the field using the visual assessment method. Both methods work but the 
technique selected should be done so keeping the objectives of the study in mind. This 
study will continue to use both methods.
In the present survey the photographs for Transects 14 were collected in August 
1992 and for Transects 5 and 6 in late September. The hurricane struck on 11 September 
thus the photo quadrat data from stations 5 and 6 represent post-hurricane conditions. The 
visually assessed (field) quadrat data were all collected after the hurricane. Thus the 
photo and field quadrat data are not all directly comparable in the 1992 dataset.
The six transects selected for this study show a considerable range in community 
development that is probably related to past (historical) impacts. Separating the impact of 
advanced primary treated effluent released at depth from a multitude of other ongoing 
and historical impacts that have occurred in and to the shallow marine communities 
fronting Sand Island is difficult at best. The added natural disturbance of Hurricane Iniki 
on 11 September 1992 is an additional impact to these communities that varies 
tremendously with location. However, the siting of these permanent stations to capitalize 
on presumed gradient(s) of impact created by the variety of land derived sources as well 
as the repeated sampling of these permanent stations should allow delineation of any 
changes attributable to the Honolulu deep ocean outfall. The sampling of these stations 
over the first two years (19901991) shows that there was little change to the 
communities over that period of time suggesting that there was no quantitatively 
definable impact to shallow water benthic and fish communities due to the operation of 
the Honolulu deep ocean outfall. Many of the changes seen in the 1992 survey appear to 
be related to the natural storm event that occurred in September of that year.
REFERENCES CITED
AECOS, Inc. 1979. Postconstruction water quality, benthic habitat and epifaunal survey 
for the reef runway, Honolulu International Airport. Final report, Part B: Benthic 
biology. In Honolulu International Airport reef runway postconstruction 
environmental impact report. Vol. 2, Technical report, ed. G.A. Chapman, 69 p., 
Honolulu:  Parsons Hawaii.
Alevizon, W.; Richardson, R.; Pitts, P.; and Serviss, G. 1985. Coral zonation and patterns 
of community structure in Bahamian reef fishes. Bull. Mar. Sci. 36:30418.
Anderson, G.R.V.; Ehrlich, A.H.; Ehrlich, P.R.; Roughgarden, J.D.; Russell, B.C.; and 
Talbot,  F.H. 1981. The community structure of coral reef fishes. Am. Nat. 117:476
95.
Brock, R.E. 1982. A critique on the visual census method for assessing coral reef fish 
populations. Bull. Mar. Sci. 32:26976.
Brock, R.E. 1986. Postconstruction biological survey for the Honolulu International 
Airport reef runway: eight years later. In Survey of the water quality, benthic habitat 
and infaunal populations for Keehi Lagoon, Hickam Harbor and marine pond, 
Honolulu International Airport, ed. OI Consultants, Inc., pp. B1B28, Prepared for 
KFC Airport, Inc., Honolulu.
Brock, R.E. 1992a. Community structure of fish and macrobenthos at selected sites 
fronting Sand Island, Oahu, in relation to the Sand Island deep ocean outfall, 
December 1990 survey. Spec. Rep. 03.24:92, Water Resources Research Center, 
University of Hawaii, Honolulu. 27 p.
Brock, R.E. 1992b. Community structure of fish and macrobenthos at selected sites 
fronting Sand Island, Oahu, in relation to the Sand Island deep ocean outfall, 
December 1991 survey. Spec. Rep. 04.30:92, Water Resources Research Center, 
University of Hawaii, Honolulu. 36 p.
Brock, R.E.; Lewis, C.; and Wass, R.C. 1979. Stability and structure of a fish community 
on a coral patch reef in Hawaii. Mar. Biol. 54:28192.
Brock, R.E., and Norris, J.E. 1989. An analysis of the efficacy of four artificial reef 
designs in tropical waters. Bull. Mar. Sci. 44:93441.
Brock, V.E. 1954. A preliminary report on a method of estimating reef fish populations. 
J. Wildlife Mgmt. 18:29708.
Connell, J. 1978. Diversity in tropical rain forests and coral reefs. Science 199:130210.
Cox, D.C., and Gordon, Jr., L.C. 1970. Estuarine pollution in the State of Hawaii. Vol. 1, 
statewide study. Tech. Rep. 31, Water Resources Research Center, University of 
Hawaii at Manoa, Honolulu. 151 p.
Dollar, S.J. 1979. Ecological response to relaxation of sewage stress off Sand Island, 
Oahu, Hawaii. Tech. Rep. 124, Water Resources Research Center, University of 
Hawaii at Manoa, Honolulu. 78 p.
Dollar, S.J. 1982. Wave stress and coral community structure in Hawaii. Coral Reefs 
1:7181.
Eckert, G.J. 1985. Settlement of coral reef fishes to different natural substrata and at 
different depths. In Proceedings of the 5th International Coral Reef Congr. 5:38590.
Evans, E.C., ed. 1974. Pearl Harbor biological survey - final report. Naval Undersea 
Center, Hawaii Laboratory. Report No. NUC-TN-1128.
Gladfelter, W.B., and Gladfelter, E.H. 1978. Fish community structure as a function of 
habitat structure on West Indian patch reefs. Rev. Biol. Trop. 26 (Supplement 1):65
84.
Goldman, B., and Talbot, F.H. 1975. Aspects of the ecology of coral reef fishes. In 
Biology and Geology of Coral Reefs, Vol. III, Biology 2, eds. O.A. Jones and R. 
Endean, 
pp. 12454, New York:  Academic Press.
Grigg, R. 1983. Community structure, succession and development of coral reefs in 
Hawaii. Mar. Ecol. Prog. Ser. 11:114.
Grigg, R., and Maragos, J. 1974. Recolonization of hermatypic corals on submerged lava 
flows in Hawaii. Ecology 55:38795.
Odgen, J.C., and Ebersole, J.P. 1981. Scale and community structure of coral reef fishes:  
A long-term study of a large artificial reef. Mar. Ecol. Prog. Ser. 4:97104.
Ricker, W.E. 1975. Computation and interpretation of biological statistics of fish 
populations. Bull. Fish. Res. Bd. Canada 191. 382 p.
Risk, M.J. 1972. Fish diversity on a coral reef in the Virgin Islands. Atoll Res. Bull. 
153:16.
Sale, P.J. 1977. Maintenance of high diversity in coral reef fish communities. Am. Nat. 
111:33759.
SAS Institute, Inc. 1985. SAS users guide:  Basics, version 5 edition. Cary, N.C. SAS 
Institute Inc. 1290 p.
Scott, E.B. 1968. The saga of the Sandwich Islands. Crystal Bay, Lake Tahoe, Nevada:  
Sierra-Tahoe Publishing Co. xv + 933 p.
Shulman, M.J. 1984. Resource limitation and recruitment patterns in a coral reef fish 
assemblage. J. Exp. Mar. Biol. Ecol. 74:85109.
Shulman, M.J.; Ogden, J.C.; Ebersole, J.P.; McFarland, W.N.; Miller, S.L.; and Wolf, 
N.G. 1983. Priority effects in the recruitment of juvenile coral reef fishes. Ecology 
64:150813.
Siegel, S. 1956. Nonparametric statistics for the behavioral sciences. New York:  
McGraw-Hill Book Co. xvii + 312 p.
Sokal, R.R., and Rohlf, F.J. 1981. Biometry. 2nd ed. New York:  W.H. Freeman. 855 p.
Walsh, W.J. 1985. Stability of a coral reef fish community following a catastrophic 
storm. Coral Reefs 2:4963.
Walsh, W.J. 1985. Reef fish community dynamics on small artificial reefs:  The influence 
of isolation, habitat structure, and biogeography. Bull. Mar. Sci. 36:35776.
Woodley, J.D., and 19 others. 1981. Hurricane Allens impact on Jamaican coral reefs. 
Science 214:74955.


