The Community Structure of Coral Reef Fishes

The University of Chicago

The Community Structure of Coral Reef FishesAuthor(s): G. R. V. Anderson, A. H. Ehrlich, P. R. Ehrlich, J. D. Roughgarden, B. C. Russelland F. H. TalbotSource: The American Naturalist, Vol. 117, No. 4 (Apr., 1981), pp. 476-495Published by: The University of Chicago Press for The American Society of NaturalistsStable URL: http://www.jstor.org/stable/2460456 .

Accessed: 23/08/2013 09:19

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

The University of Chicago Press, The American Society of Naturalists, The University of Chicago arecollaborating with JSTOR to digitize, preserve and extend access to The American Naturalist.

http://www.jstor.org

This content downloaded from 192.236.36.29 on Fri, 23 Aug 2013 09:19:07 AMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=ucpress

http://www.jstor.org/action/showPublisher?publisherCode=amsocnat

http://www.jstor.org/stable/2460456?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp


Vol. 117, No. 4 The American Naturalist April 1981

THE COMMUNITY STRUCTURE OF CORAL REEF FISHES

*G. R. V. ANDERSON, tA. H. EHRLICH, tP. R. EHRLICH, TJ. D. ROUGHGARDEN, ?B. C. RUSSELL, AND ?F. H. TALBOT

Australian National Parks and Wildlife Service, Canberra City. Australian Capital Territory 2601. Australia; tDepartment of Biological Sciences, Stanford University, Stanford. California 94305; tHopkins Marine Station. Pacific Grove, California 93950; ?Center for Environmental Studies,

Macquarie University, North Ryde, New South Wales 2113, Australia

Submitted June 1, 1979; Accepted June 26, 1980

In this paper we discuss the structure of coral reef fish communities as exem- plified by the butterfly fishes, Chaetodontidae, of the northern Great Barrier Reefs of Australia. The study is focused on the question of whether (a) reef fish communities are structured in ways similar to terrestrial vertebrate communities (Brown and Lieberman 1973; Cody 1968; Diamond 1975; Pianka 1976; Pulliam 1975; Schoener 1974) and are potentially explainable by a suitable extension of the theory which is being developed for terrestrial vertebrate communities, or (b) these communities are structured in fundamentally different ways and are to be explained by a theory which places emphasis on the dynamics of fish larval dispersal, larval habitat preferences, and the stochastic recruitment of larvae to local regions of a reef. This latter position is espoused in an important paper by Sale (1977). We present data on chaetodontid fishes and review the issues involved. We conclude that the available data falsify neither hypothesis and that they will be extremely difficult to distinguish in practice.

As presented in detail below, our chaetodontid data show (1) there are conspicuous differences in the niches of many of the species which locally coexist; (2) geographical replacement does occur between species from the same niche; and (3) the species from replicates of small sites within a habitat type are distributed in ways consistent with both hypotheses.

MATERIALS AND METHODS

Field site. -The work on chaetodontids was based at the Lizard Island Re- search Station on the northern part of the Australian Great Barrier Reefs. The station lies some 30 km from Cape Flattery on the mainland shore of Queensland, slightly more than halfway to the outer barrier reef (fig. 1). Studies of the feeding behavior of a large number of chaetodontids were possible on reefs a few hundred meters off the station; power boats provided easy access to a large variety of

Am. Nat. 1981. Vol. 117, pp. 476-495. ? 1981 by The University of Chicago. 0003-0147/81/1704-0005$02.00

476



COMMUNITY STRUCTURE OF CORAL REEF FISHES 477

1450 15' 1450 30'

14030' -_,,____ _

BOMMIES OUTER BARRIER - | { . ) REEF (2 sites)

10 km

E~REEF

INNER BARRIER REEF (11 sites) BOMMIES'> ,-'

"1

FAR CENTRAL

REEFS (4 sites) MAC'S REEF

] / ( v ~~~~~~LIZARD I

NEAR CENTRAL / EAGLE I REEFS (5 sites)

14045 / --;__-.____

MARTIN REEF

NEAR SHORE LINNET REEF REEF (1 site)

MAXWELL REEF

N

CAPE FLATTER (mainland) DECAPOLIS REEF

FIG. 1.-Map of transect on Australian Great Barrier Reefs.

habitats from the shore to the outer barrier. The work at Lizard was done in a 2-wk period, December 3-16, 1976.

Microdistribution. -The abundance of various chaetodontids was assayed at 23 different stations lying on a transect (fig. 1) from the outer barrier (50 km from the Queensland shore) to Decapolis Reef (6 km from the shore). Records were kept of 23 species of Chaetodon (sensu lato), four species of Heniochus, two of For- cipiger, as well as Chelmon rostratus and Coradion chrysozonus. All size classes that had assumed the adult pattern were counted. Censuses were done by a team of three divers swimming parallel transects 6-8 m apart. Swimming rates were calibrated so that a standard distance of 100 m could be covered and all chaetodontids within 2.5 m on either side of the swimmer were recorded (giving a total coverage of 500 m2 per census unit). No attempt was made to keep track of individuals that left the field of view, so some multiple counting, especially of more mobile species, undoubtedly occurred. When the reef surface sloped, the three divers swam at different depths, most often at about 4, 8, and 12 m. In most localities more than 1,200 m of transect were censused, covering more than 6,000 mi2. The results of different censuses by the three individuals and repeat censuses by the same individual were usually quite similar, with the exception of the results for Heniochus. The Heniochus species presented problems of sight identification as well as of counting because of their tendency to conceal themselves in crevices



478 THE AMERICAN NATURALIST

TABLE 1

RANK ABUNDANCES OF FIRST EIGHT OF FIFTEEN SPECIES

TRANSECTS DIVERS

1 2 3 4 5 All 100 m 100 m 140 m 140 m 80 m 1 2 3* 1974

Chaetodon plebeius ... 1 1 3.5 1 1 2.5 1 2 1 1 (59)

C. trifasciatus ........ 2 2 5 3 2 1 2 1 3 1 (45)

C. melannotus ........ 3 4 7 8 3 2.5 5 7.5 2 1 (26)

C. citrinellus ......... 4 4 2 ... 6 5 3 3 6 2 (24)

C. auriga .5 4 7 6.5 4 5 5 4.5 4 2 (23)

C. vagabundus 6 10 1 5 6 ... 5 6.5 6 2 (21)

C. kleinii .7 6 10.5 3 8 5 7 4.5 8.5 3 (17)

C. aureofasciatus 8 7 3.5 3 9.5 .. 8.5 6.5 6 2 (15)

NOTE.-( ) = numbers observed. * One less transect (see text for explanation).

and caves, and therefore we have not used these data. Problems of distinguishing the two Forcipiger species led us to pool the data from them.

Table 1 shows the rank order abundances of the eight commonest species (out of 14 species present + one Chaetodon aureofasciatus-Chaetodon rainfordi hybrid individual) for five transects across the main study reef near the Lizard Island station on December 6, 1975. The first four transects were swum by three different observers and the last one by two. Note that, although there is the expected variance from transect to transect and diver to diver, the basic pattern of abundance remains the same. That this general pattern is stable over time is evidenced by the column labeled "1974" which presents data from Ehrlich et al. (1977) gathered on the same reef in October-November 1974. In making observations on chaetodontid behavior on the same study reef, they classified species into rough abundance categories: (1) common to abundant, (2) common, and (3) rare.

It is interesting to note that of the 15 taxa found in the one day set of transects, two were found that were not present in 1974: A pair of Chaetodon bennettii was observed, a species that is rare everywhere, and the hybrid individual Chaeton aureofasciatus-rainfordi. Of the 12 species ranked common or common to abundant in 1974, seven were in the top eight in 1976. The eighth was Chaetodon kieinii which was scored as rare in 1974. Chelmon rostratus, which was common in 1974, ranked ninth of 15 in 1976, and Chaetodon rainfordi, also common in 1974, was tenth in 1976. Of the common species in 1974 only Chaetodon trifascialis was rare in 1976; nine of the top 10 were the same.




Ten isolated large coral heads ("bommies") behind the outer barrier were censused completely. The bommies rise to within 2 m of the surface from a sand bottom at approximately 20-30 m depth and are isolated from one another by expanses of open water where adult chaetodontids are not observed swimming. The areas of the bommies were measured so that they could be compared to the linear censuses. (Whether the vertical faces of the bommies were in some sense equivalent to equal areas of horizontal substrate could not be determined, but several species of chaetodontids were found on them).

Microhabitat and trophic analyses.-Approximately 100 person hours were spent over the 2-wk period by three divers making detailed notes on habitat associations and feeding behavior of chaetodontids. Particular attention was paid to association with various kinds of corals and to recording sequences of biting of different substrates. One to three specimens of each species were collected with hand spears and their gut contents examined microscopically. A fish species was placed in a trophic category only if we had obtained data on both stomach contents and observed feeding behavior in the field. Our data do not speak to possible seasonal differences in feeding behavior, but more casual observations by us at Lizard and other Pacific localities make major seasonal differences seem relatively unlikely. In general our results confirm observations found in the literature (e.g., Reese 1977).

Geographic distribution.-A search of the literature has been carried out to determine the geographic occurrence of the different species. This was sup- plemented by field notes and censuses carried out by the Ehrlichs in the Society Islands and Fiji.

Morphometrics. -In order to determine if there were gross differences in size and shape that might correlate with observed differences in distribution or behavior, six measurements (standard length, snout length, head length, body length, jaw gape, and snout angle) were made of 153 specimens of 28 species of Chaetodontidae in the collection of the Australian Museum. The measurements were the same as those taken by Jones (1968, fig. 4) in his study of guilds in acanthurids (surgeon-fishes). Standard vernier calipers were used except for measuring the angle between the eye and mouth which was done with a protrac- tor. Standard length measurements correlate well with maximum size estimates given by Steene (1978) and with our subjective impressions of the dominant size classes at our study sites. Therefore problems of size bias in the museum sample would appear to be minimal.

RESULTS

The results are summarized in six tables. Table 2 shows assessments of food utilization based on observations of foraging behavior in the field and analysis of stomach contents from fish that were speared in the study sites. Table 3 presents the morphometric data obtained from preserved specimens from the collections of the Australian Museum. These two tables allow us to classify every species into trophic and morphometric categories. This classification then forms the basis for




TABLE 2

TROPHIC CATEGORIES

Soft Coral and Noncoralline Hard Coral Some Hard Coral Invertebrates Generalistst

Chaetodon aureofasciatus Chaetodon kleinii* Chaetodon auriga Chaetodon citrinellus C. baronessa C. lineolatus C. auriga C. ephippium C. ornatissimus C. melannotus Chelmon rostratus C. ulietensis C. plebeius C. unirnaculatus Forcipiger spp. C. vagabundus C. rainfordi C. speculum C. trifascialis C. trifasciatus

* Especially polychaetes and crustaceans. t Plus plankton.

TABLE 3

MORPHOMETRIC CATEGORIES

Small Large SL < 90 mm SL > 90 mm

Typical .............. Chacetodon plebehus (HC) Chaetodoti speculum (HC) C. trifascialis (HC) C. lineolatus (SC)

SL-Soth > 1.20 C. trifasciatus (HC) C. melannotus (SC) Depth ' C. kheinii (SC) C. unimaculatus (SC) C. citrinellus (G) C. auriga (NC)

Forcipiger spp (NC) C. ephippium (G) C. ulietensis (G) C. ivagabundus (G)

Extra deep bodied .... . Ciactodon auireofasciatus (HIC) Chaetodon ornatissimtus (HC) SL-Snout C. baronessa (HC) Chelmon rostratus (NC) Depth < 1.20) C. rainfordi (HC)

NOTE.-HC = hard coral; SC = soft coral; NC = noncoralline invertebrate; G = generalist.

interpreting the fish census data in Table 4. The census data refer to the long 50 km transect from the near shore to the outer barrier. Table 5 presents census results for ten replicate sites within one habitat type and is logically independent of the first three tables. Tables 6 and 7 compare some general features of the community structure of butterfly fish on the Great Barrier Reef with the structure at other places throughout the South Pacific.

Butterfly fishes can be classified into four primary food categories (table 2). Eight species can be seen nipping almost exclusively on the surface of hard corals. These species contain essentially nothing in their stomachs other than material which appears to be coral tentacles. We term these species hard coral specialists (HC). There appears to be further specialization within this group with respect to



COMMUNITY STRUCTURE OF CORAL REEF FISHES

C-5~~~~~~~~~-

Co)~~~~~~~-

6~~

u 0 ~ C) ~ ~ ~ ~ ~ * E~~* I




CO m-ON0OON_

C) ~

0 -e

O ~

LL1 -

00 e00e

~~ L.....

< ~ ~ ....

z C)~~~~~~C LL ~ t C):

Q A O

- _00S00&




TABLE 6

FAUNAL SIZE AND DEGREE OF PLANKTIVORY

Great Barrier Society Reefs" Fijit Islandst Hawaii?

Chaetodontids'" (excluding HITnfiochus)

Total species ........ ....... 29 23 24 19 Total planktivores 1........... 2 4 4 % Planktivores ....... ...... 3 7 17 21

Pomacentrids# Total species ........ ....... 91 60 30 15 Total planktivores ........... 47 35 16 9 % Planktivores ....... ...... 52 58 53 60

* Our observations at Lizard Island and Goldman 1973. t Observations by PRE and AHE, 1977 and Carlson 1975. t Observations by PRE and AHE, 1971 and 1977 and Bagnis et al. 1972. ? Observations by PRE and AHE 1969 to 1976; J. Randall personal communication; various

publications, especially Hobson 1974. 1I The following were classified doing a major portion of their feeding as planktivores on the basis of

our observations and reports in the literature: Chaetodon kleinii, C. trichlrous, C. corallicola (= kleinii), C. miliaris, Hemitaurichthys polylepis, and H. thoinsoni. Here and in table 6, C. peleit'ensis was considered a synonym of C. punctatofasciatus. We made several observations of paired individuals, one with "pelewensis" pattern and one with "punctatofasciatus" pattern. One individual in the Australian Museum collections has one pattern on one side and one on the other. C. triangullulrn was considered a synonym of C. baronessa.

# Data abstracted from Allen (1975). The following were classified as planktivores: Amphiprion, Preinnas, Acatithochromis, Chrornis, Dascvllus, Lepidozygus, Amnblyglyphidodoni, Glyphidodotitops (8 spp.), Neopomiacentrus, Pomiacentrus (10 spp), Pomnachronfis. Pristotes. Decisions on which genera and species were classed as planktivores were made before distributions were examined.

TABLE 7

FAUNAL SIMILARITY OF CHAETODONTIDS

GBR Fiji Samoa* Soc. Is. Hawaii

Great Barrier Reefs ...... ... .76 .62 .66 .41 Fiji .................... 97 ... .74 .83 .52 Samoa .................. ..90 .85 ... .90 .50 Society Islands .79 .79 .79 .75 ... .63 Hawaii .................. ..63 .63 .53 .79

NOTE.-Proportion of species in row locality also present in column locality. * Data from Schultz 1943. Other data as indicated in footnotes to table 6.

the species of hard coral that are used, but we could not document this because of difficulties with coral identification. Four species can be seen nipping often on soft corals. The stomach contents in this group reveal pieces of soft coral and often some evidence of hard coral as well. We term this group the soft coral eaters (SC). However this group does not use soft corals exclusively. Three species are clearly involved mostly with noncoralline invertebrates, especially polychaetes and crustaceans. These fish usually feed over dead coralline substrates which are




encrusted with epizoans. We term this group the noncoralline invertebrate feeders (NC). Four species were generalists (G) in the sense that their stomach contents and feeding behavior reveal the use of hard corals, soft corals, polychaetes, and crustaceans.

Table 3 presents the summary of some of our morphometric data. The 19 species shown in the table range from very small to quite large and vary in shape from "typical" chaetodontid form, as seen in C. trifasciatus, to extra deep bodied, typified by C. baronessa. We divided the size into two categories based on the standard length of the fish and the shape into two categories "typical" and "extra deep bodied" based on an index obtained by dividing the body depth into the standard length minus the snout length. The criteria are shown in the table.

Four points can be made concerning the relationship between the trophic categories and the morphometric categories. (1) Of the eight hard coral feeders, six are small and only two are large. However they are evenly divided between the two shape classes. (2) Of the four soft coral feeders, all are typical; three are large and one is small. (3) The three noncoralline invertebrate feeders are all large; two are typical and one is extra deep bodied. (4) The four generalists are typical but vary in size. These morphometric data represent only some of the very obvious features of the morphological differentiation among the species. The fishes also differ markedly in snout length and in the position where the snout inserts into the head. Dentition also varies markedly among species. Hobson (1974) has suggested for the two species of Forcipiger that these differences are of functional sig- nificance. For the remaining chaetodontids, however, the relationship between form and function remains to be investigated.

Table 4 and figure 2 present the census data from the 50-km transect from a near shore reef to the outer barrier reef. For each group of sites the fish species are classed into log2 abundance categories. This table reveals nine major points about the community structure of butterfly fish from the perspective of geographical ecology.

1. Of the more than 20 species of Chaetodon (sensu lato), Chelmon, and Forcipiger which are regionally available, the number of species which co-occur locally (defined as being present in more than one fish per 100 m of transect) lies between four and nine. Typically six to seven more species co-occur locally in "appreciable" abundance while seven more species are present as one or two isolated individuals in the study site. (These are classed as rare in table 4.)

2. There is niche diversification within every group of sites. All groups of sites contain members from three of the four trophic categories at abundances of more than one fish per 100-m transect. If rare fish are included, then all groups of sites contain members from all four trophic categories.

3. In every habitat the numerically dominant species are small hard coral specialists.

4. The butterfly fish on the outer barrier are all typical in form. The typical species are generally replaced by extra-deep-bodied species as one proceeds from the outer barrier to the near shore reefs. The extra-deep-bodied species possibly can hold a stationary position in quiet water and may be able to forage in complex crevices more readily than typical forms. The typical forms tend to dart and to




swim quickly in their foraging. On the outer barrier there is more surge and reef is more compacted than near shore, so the extra-deep-bodied form is presumably at a disadvantage there.

5. Every community includes generalist species but they are never the most abundant species.

The remaining four points concern the details of each feeding complex. 6. In the hard coral complex both of the large species, C. speculum and C.

ornatissirnus, are rare or absent from all sites; it is only the small hard coral feeders who attain appreciable abundance. Chaetodon trifasciattus is common from the outer reef through the far central reefs. Thereafter it becomes uncommon at the near central reefs and rare at the near shore reef. It is present at all sites. Chaetodon plebeius is uncommon but decidedly present from the outer barrier through the far central reefs. It becomes common in the near central reefs and drops to rare at the near shore reef. This species is also present at all sites. Chaetodon baronessa is rare in the outer barrier, uncommon from the inner barrier through the near central reefs, and rare again at the near shore reef. This species is present at all sites too but is principally found between the extremes of the outer barrier and the near shore reef. Chaetodon aureofasciatus is absent at the outer and inner barrier reef. It is rare at the far central reefs, common at the near central reefs, and extremely common at the near shore reef. Similarly, C. rainfordi is absent at the outer and inner barrier reef, is rare at the far central reefs, rises to uncommon at the near central reefs, and becomes common at the near shore reef. There is thus clear geographical replacement of the C. trifasciatls, C. plebeius, and C. baronessa group by the C. aureofasciatus and C. rainfordi group along the overall transect from the outer barrier reef to the near shore reefs.

7. The soft coral feeding complex principally involves two species, the small C. kleinii and the large C. melannotus. The other two species are rare at or absent from all the sites. Chaetodon kleinii is uncommon on the outer barrier, rare on the inner barrier, uncommon on the far central reefs, rare on the near central reefs, and absent from the near shore reef. Chaetodon inelannotus is rare on the outer barrier, uncommon at the next three habitats, and rare at the near shore reef. The data suggest that C. klieini and C. melannotus coexist through much of their species ranges but that C. kleinii's range tends toward the outer barrier while C. melainnotius tends toward the near central reefs.

8. The noncoralline invertebrate feeding complex also principally involves only two forms, C. auriga and Forcipiger spp., both of which are in the same size and body depth class. (Both Forcipiger longirostris and F. flavissimus are prob- ably present in our censuses, but the two species could not be distinguished with certainty in the field.) Although these two species have structurally different mouths and, at least in Hawaii, different feeding preferences [Hobson 1974], both are noncoralline invertebrate feeders. However Forcipiger spp. has a noticeably longer snout than C. auriga. The third species, Chelmon rostratus, is rare or absent from all sites. Forcipiger spp. is uncommon on the outer barrier and rare on the inner barrier and elsewhere. Chaetodon auriga is absent on the outer barrier reef, rises to uncommon on the inner barrier reef, and is generally uncommon at the other sites. (Although it is recorded as rare in the far central reefs, its




_ I I I I1I

EXTREMELY COMMON

aureofasciatus

VERY COMMON

trifasciatus COMMON _- d S

rainfordi

UNCOMMON A

RARE bronessa

OUTER INNER FAR NEAR INSHORE BARRIER BARRIER CENTRAL CENTRAL REEFS

REEFS REEFS REEFS REEFS

FIG. 2a.-Hard coral feeders.

EXTREMELY COMMON

VERY COMMON

COMMON

melanotus UNCOMMON

RARE V kleinii

_ I I I II



FIG. 2b.-Soft coral feeders.




I i X X I EXTREMELY

COMMON

VERY COMMON

COMMON

Forcipiger C. auriga UNCOMMON | \ ~ ~

Chelmon RARE _A

. I I I I I .



FIG. 2c -Noncoralline invertebrate feeders.

EXTREMELY COMMON

VERY COMMON

COMMON

/ephippium citrinellus vagabundus UNCOMMON / 6

RARE /



FIG. 2d.-Generalist feeders.




abundance was .9 fish/l00 m and thus is on the borderline between the rare and uncommon categories in this area). In this complex we see an abrupt geographical replacement of the Forcipiger by C. auriga as one proceeds shoreward from the outer barrier. The switchover occurs very near the outer barrier reef.

9. The generalist feeding complex consists of three large species, C. ephippium, C. ulietensis, and C. vagabundus and one small species, C. citrinellus. Both C. ephippium and C. ulietensis are uncommon on the outer barrier and drop to rare or absent from the sites shoreward from there. Chaetodon citrinellus is rare on the outer barrier, rises to uncommon from the inner barrier through the far central reefs, drops to rare at the near central reefs, and is absent from the near shore reef. Finally, C. vagabundus is absent from the outer barrier and rare in the inner barrier. It rises to uncommon at the far central reefs and holds that status all the way into the near shore reef. The overall picture in this feeding complex is that the pair, C. ephippium and C. ulietetisis, is replaced by C. citrinellus, which is itself then replaced by C. vagabundus on the transect from the outer barrier reef to the near shore reef.

In general terms we see that geographical replacement of species along the reef to near shore transect is a conspicuous phenomenon which occurs among the fishes from each of the feeding categories. This basic fact must be accounted for by any theory of the community structure of coral reef fishes.

Table 5 presents counts of individuals for 10 replicate sites within one habitat type. The habitat consists of the region slightly shoreward from the outer barrier reef. This region contains many separate and unconnected chunks of coral reef, of roughly cylindrical shape, which rise from the sand and rubble floor. The depth of the floor is typically, 20 m (it was measured in each case). The bommies rise to within 2 m of the water surface. The bommies we studied varied in diameter from 12 m to 26 m. It was possible for the census team to swim around them and over them completely and to census the entire bommie. One set of five bommies was shoreward of the Day Reef and the other set of five was shoreward of the Carter Reef. Although one might expect the fish fauna of bommies behind a particular reef to be more similar to one another than to the fish from the bommies behind another reef, we did not find this. The data in Table 5 show that the fish fauna on all the bommies appear very similar to one another.

It would be useful if the distribution of each species over the replicate sites could be subjected to a statistical test of randomness. This is not possible because the sites are not all of equal area and could not reasonably be assumed to be equal in other respects (shape, surface structure, invertebrate fauna).

In addition it has become clear that discovery of a significant deviation from randomness toward clumping would not falsify the "random recruitment" hypothesis. That hypothesis would be consistent with a clumped distribution on the basis of selective settling based on characteristics of the substrate but not on characteristics of the fish community already resident. Similarly, overdispersion could result if physically suitable settling sites were themselves overdispersed.

This raises the question of whether any investigation of the arrangement of individuals within or among small study sites that are close to one another can ever contribute to the testing of the random recruitment hypothesis (e.g., Talbot et




al. 1978). It is clear that the scale problem is central to the resolution of the issues raised by Sale (1977).

Tables 6 and 7 present data on the species composition of butterfly fishes at several geographically different sites from the Great Barrier reefs east and north- east across the South Pacific. Some of these data were gleaned from the literature and others were established by limited (a few hundred m of transect) censuses carried out by P. R. E. and A. H. E. Data for pomacentrid species (Allen 1975) are also presented for comparison. The data show major changes in the structure of the chaetodontid-pomacentrid community, including significant decline in chaetodontid species diversity in the oceanic islands and niche shifts into qualitatively new feeding categories. The new feeding category is that of planktivory, a behavior seen facultatively in C. k/einii on the Barrier Reefs and not in any other chaetodontid there. The major shift into this trophic category by butterfly fishes appears to be correlated with a dropout of the damsel fishes which, along with apogonids, dominate in this role on the Barrier Reefs. The decline in their species abundance is shown for both total species and planktivorous species at the bottom of table 6. We interpret these major differences in community structures as signs of great isolation between communities that point to a lack of an ecologically important dispersal by reef fish larvae across the Pacific Ocean.

DISCUSSION

To present a clear picture of the specific issues involved, we begin with a summary of and comment on Sale's (1977) important paper. He makes five principal points.

1. Reef fishes are not specialized and overlap extensively in food and microhabitat. This observation is important in that Kohn's (1959, 1968, 1971) work on reef Conus snails did establish a remarkably high degree of specialization and relatively low food overlap in these predatory snails. Sale emphasizes that Conus snails do not provide a good empirical model for reef fish. We offer three comments on this point. First, the claim that fish are not specialized and overlap extensively is a value judgment on the data. It is generally observed, as acknowl- edged by Sale, that reef fish do exhibit habitat and resource partitioning. For example, Sale's (1975) data reveal characteristic habitat associations of damsel fishes, and our data below demonstrate characteristic habitats for butterfly fishes. Furthermore, as Sale notes, Roughgarden's (1974) analysis of Randall's (1967) data demonstrates substantial resource partitioning with respect to prey size in serranids (groupers and relatives), which are predatory, although no such partitioning occurs in scarids (parrot fishes), which are herbivorous. Second, as Sale has noted, Current community theory is not restricted to coexisting species which exhibit wide niche separation and narrow niche widths. Indeed, according to niche theory, overlap may be nearly total and specialization nearly absent, provided the predation pressure is sufficiently strong (Roughgarden and Feldman 1975) and/or provided the utilization curves and the resource distributions satisfy certain conditions (Roughgarden 1974). Thus the occurrence of broad overlap and a near absence of specialization in reef fish ecology does not ipso facto entail that current




community theory is inappropriate for reef fish. Third, the remarkable degree of food specialization revealed by Conus snails is not generally found in terrestrial vertebrate communities. Indeed, in birds, lizards, and mammals (see references in introduction) resource utilization curves characteristically show broad overlaps, with significant, but rarely spectacular, mean differences. Thus the differences between reef fish and Conus obviously do not imply that reef fish communities also differ from terrestrial vertebrate communities in this regard.

2. Many features of the local environment of reef fish are unpredictable. This point is especially directed to the occurrence of suitable space for fish to live in. According to Sale (1977), space becomes available by mechanisms whose effect and action are unpredictable to the fishes both in space and time.

3. Fish typically produce large numbers of pelagic larvae which may disperse over great distances by remaining in the plankton for a long time. Sale (1977) suggests this characteristic trait of fish may be viewed as an adaptation to the unpredictable occurrence of suitable living space mentioned in the preceding point.

4. The existence of a large and well-mixed pool of larvae from many species has led Sale to suggest that recruitment to a site within a habitat zone may be independent of the structure of the resident fish community at that site. We offer three comments on this point.

First, this point is absolutely critical to the controversy. Community theory, and niche theory in particular, proceeds by writing down models wherein the dynamics of each population depend on the population sizes of all the species residing in the local community. The model expresses the interactions among the resident populations. To date competitive interactions have received most attention in models, but the inclusion of predators and other extensions are tractable; they simply have not yet been done. If the local community structure is controlled by variables outside the local system, however, then the approach of community theory is itself inappropriate.

Second, if the recruitment to a site is independent of the resident fish community at that site, then the situation may simply be one of scale. Perhaps the original study site is too small, and if a larger study site were considered then the larval dispersal would be contained within it. As a result recruitment to the site would then be determined by the resident fish community within the site. This possibility deserves serious consideration because the practical difficulties of field work under water have often led marine ecologists to lay out and work very small study sites relative to terrestrial vertebrate standards. One would not think of applying and testing community theory with the birds on a single tree or the lizards in a single bush. Similarly, many underwater study sites are obviously too small to represent a population's dynamics, and, lacking knowledge of the spatial extent of the demographic units, little can be said about dynamics (Ehrlich et al. 1975).

Third, the real issue is whether recruitment to a local site is only a matter of scale. We may envision two possible situations. In the first, the exchange of larvae between sites decreases with the distance between sites. This in fact does often apply to terrestrial dispersal (see Harper [1977] in plants; Richardson [1970] in Drosophila; Blair [1960] in lizards). If this situation applies to fish larvae too, then




the proportion of larvae recruited to a site from the outside decreases as the area of the site increases, and community theory should apply to a site that is sufficiently large. The appropriate radius for a roughly circular site would be, say, five times the average dispersal distance of the larvae. In the second situation we may imagine that all sites, even those very distant from one another, are con- nected to each other as a result of sharing a common pool of larvae. The repro- ductive output from each site leaves that site and enters a long phase of larval existence in the plankton. The currents carry the larvae and mix the output from all the sites. Then the larvae settle out from a homogeneous planktonic pool of larvae (at a low net rate) as suitable space occasionally becomes available. Settling is statistically independent of the resident fishes in the vicinity, and the new arrival holds on to the space it settles into. It then follows that the adult fish communities at a site represent a random sample drawn from the common larval pool under the constraints presented by the distribution pattern of physical substrates. This sample happened to settle in the site and then grew up there. If this applies to reef fish then typical terrestrial vertebrate community theory is certainly inappropriate to explain the fish community structure even for large study sites. There are formal similarities between this second situation and the so-called Levine multiple niche polymorphism model in population genetics.

5. Sale's final point is that he envisions that the explanation for the coexistence of reef fishes somehow lies in the stochastic process of larval recruitment to local areas of a reef. Sale does not actually offer a mathematical model or a complete verbal account of the larval recruitment process and how that process explains the coexistence of many species of reef fish. However, he does suggest that a model for regional coexistence (as distinct from local coexistence) along the lines of the Levins and Culver (1971) model may be appropriate. There is a crucially important point to consider here. There are two sides to the empirical phenomenon which must be explained. As is well known, a great many species of reef fishes do coexist locally, and of course this fact must be accounted for. But also, there are many species that are not observed coexisting. There are, as we have demonstrated, species of butterfly fishes found inshore in comparatively turbid water and different species which are found toward the outer barrier in clearer water. Still other species can be assigned to intermediate habitats. All these species presumably can reach one another's habitats and are thus potentially able to co-occur locally. Yet they do not all co-occur and any theory of reef fish communities must account for this as well as for the coexistence. Sale (1977) explains the pattern of distribution across broad habitat zones on the basis of larval settling preferences. He writes, "reef fishes are habitat specialists on a broad scale, and we can expect that newly settling larvae of each species will show appropriate habitat preferences" (p. 351). Thus a theory for reef fish communities, like any other community theory, must satisfy two criteria: (1) It must explain the mechanism leading to coexistence of a great many species; and (2) it must explain why those that do coexist are a proper subset, often a small subset, of all the species that are regionally available.

At this point we can summarize the two principal views of how coral reef fish communities are structured. We stress that neither of these views has been




developed into a logically complete theory at this time and further theoretical work is required. Nonetheless, the views are quite different from one another and we can, to some extent, state predictions or propositions which, if confirmed, would tend to support one over the other.

An important beginning to a theoretical formulation of the view that traces to Sale has very recently been made by Chesson and Warner (1981).

1. The coral reef fish community is structured in ways that parallel the community structure of vertebrates in terrestrial situations, with the one possible exception that the scale of the phenomenon may be larger. The scale may be larger because of the long dispersal distances of fish larvae and often fish eggs as well. Local coexistence of species is facilitated by the resource partitioning among them with respect to food and to microhabitat variables like water depth. That the fishes that do coexist locally are a subset of those regionally available is explained as a result of competition by either or both of two distinct causal pathways. One pathway is that there is strong present-day competition resulting in ecological competitive exclusion. The other pathway is that the present-day competition is weak, perhaps has always been weak, but has led to the evolution of differing habitat preferences. In either case we would expect on long transects across habitats that fish with similar food habits and microhabitat preferences (i.e., similar niches) would replace one another more or less abruptly along the habitat gradient. In short, (1) the local community structure involves coexistence of fishes whose niches differ with respect to food and microhabitat; and (2) geographical replacement occurs along habitat gradients for fishes whose niches are the same with respect to food and microhabitat. Such patterns are very frequently observed in terrestrial vertebrate communities and are the basic observation that mathematical niche theory is addressed to explaining. Niche theory in fact provides several alternative formulations all based on competition and predation models which predict both local coexistence and geographical replacement. At the start of our research none of us expected this view to be sustained.

2. The coral reef fish community is structured principally by the dynamics of larval dispersal and settling. Population interactions, especially competition, are presumed to be insignificant. A local community of resident fishes within a given habitat type is a random sample of the larvae from a common larval pool. It is further postulated, however, that larvae have habitat preferences and settle only under specific environmental conditions. That the fishes that do coexist locally are a subset of the fishes regionally available is explained by reference to the physiol- ogy of larval settling. According to this conception, (1) the high local diversity in reef fish communities does not actually pose a problem because in the absence of competition there would appear to be no difficulty to obtaining the coexistence of many species; and (2) the geographical replacement of species along a long habitat transect reflects the habitat preferences of larvae in their settling. All of us were sympathetic to this view of the maintenance of coral reef diversity.

To test the appropriateness of these two views we examined the following three points in the field: (1) Are there conspicuous differences in niches of the butterfly fish species that do coexist in each habitat type? (2) In a 50-km transect from the




shore to the outer barrier reef, does geographical replacement among butterfly fish occur among species of the same niche or does the geographical replacement seem to be independent of the niches of the species involved? (3) In a set of 10 replicate local communities of butterfly fish within the same habitat type, are the species- abundance distributions those that would be expected by chance recruitment to those communities?

Our data have shown: (1) There are conspicuous differences in the niches of most of the species that locally coexist; (2) geographical replacement does occur between species from the same niche; and (3) at many sites within a habitat type the results of censuses are similar to one another in space and time.

Our data on the community structure of coral reef fish thus are not particularly surprising from the point of view of a terrestrial vertebrate ecologist. Yet it is precisely this fact that is significant. There are niche differences between closely related fishes that coexist in a local region, a result confirming the observations of workers such as Jones (1968) working with acanthurids and Hobson (1974) on chaetodontids and other groups. More importantly the fish that do coexist locally are a subset of all those regionally available, and there is geographical replacement of species with the same feeding habits along a 50-km habitat transect running from the outer barrier reef to a near shore reef. These facts accord nicely with the patterns in terrestrial vertebrate communities and do not license the belief that the approach of mathematical community theory is a priori inappropriate for coral reef fish communities. We do not imply that community theory has been well established in terrestrial vertebrate communities. What we do assert is that community theory is not automatically destined to be less successful with reef fish than with terrestrial vertebrate communities simply because reef fish and their larvae are under water.

We close with a remark on the problem of scale. We had expected the scale of species replacement of reef fish to be very coarse compared with that of terrestrial vertebrates because of the long dispersal of fish larvae and eggs. We had initially not expected much geographical replacement along the 50-km transect and had expected instead to observe species replacement principally between the Barrier Reefs and islands of Oceania hundreds and thousands of kilometers away. How- ever, substantial geographical replacement does occur in the 50-km transect, and the spatial scale of this phenomenon is quite the same as that in the geographical replacement of bird and lizard species along altitudinal transects on mountains (e.g., Diamond 1975). In a similar vein the huge degree of community differentiation between the Barrier Reef and islands of Oceania is not simply a matter of species replacement but bespeaks a rather major community reorganization. This major differentiation of community structure over very long distances seems quite similar to the differentiation between bird and lizard community structure on continents and on oceanic islands (e.g., Darlington 1957). If so, we are again observing terrestrial and marine biogeographic phenomena on the same spatial scale. The distance in ocean waters seems no shorter to fishes than does that of the ocean air to birds. Reef fish larvae and eggs may be collected in plankton, often far out to sea, but various mechanisms have also been described that tend to limit




their dispersal (for review, see Ehrlich 1975). There seems no reason at present to believe that the ecological and evolutionary consequences of this dispersal are more significant than those of ordinary waif dispersal of terrestrial animals.

SUMMARY

Observations of the community of chaetodontid fishes at the northern end of the Great Barrier Reefs are consistent with the hypothesis that it is structured in ways similar to terrestrial vertebrate communities. Our data do not support the need for alternative hypotheses centered on larval habitat preferences and stochastic recruitment. Broad geographic patterns of chaetodontids in the Pacific are similar to those seen in terrestrial vertebrates.

ACKNOWLEDGMENTS

We are indebted to Jan Carey for assistance in the field and to John C. Ogden, Robert T. Paine, John E. Randall, Ernst S. Reese, Peter F. Sale, Thomas N. Schoener, Victor G. Springer, and Hugh Sweatman for critical readings of the manuscript. This work was supported by a grant to Paul R. Ehrlich and Jonathan D. Roughgarden from the National Science Foundation's U.S.-Australia Co- operative Science Program.

LITERATURE CITED

Allen, G. R. 1975. Damselfishes of the South Seas. T. F. H. Publications, Neptune City, N.J. Bagnis, R., P. Mazellier, J. Bennett, and E. Christian. 1972. Poissons de Polynesie. Les Editions du

Pacifique, Papeete, Tahiti. Blair, W. F. 1960. The rusty lizard. University of Texas Press, Austin. Brown, J. H., and G. A. Lieberman. 1973. Resource utilization and coexistence of seed-eating desert

rodents in sand dune habitats. Ecology 54:788-797. Carlson, B. A. 1975. A preliminary check-list of the fishes of Fiji (based on specimens in the marine

reference collection of the University of the South Pacific). Suva, Fiji. Mimeographed. Chesson, P., and R. Warner. 1981. Environmental variability promotes existence in lottery competi-

tive systems. Am. Nat. 117. Cody, M. L. 1968. On the methods of resource division in grassland bird communities. Am. Nat.

102:107-148. Darlington, P. J. 1957. Zoogeography. The geographical distribution of animals. Wiley, New York. Diamond, J. M. 1975. Assembly of species communities. Pages 342-444 in M. L. Cody and J. M.

Diamond, eds. Ecology and evolution of communities. Belknap, Cambridge, Mass. Ehrlich, P. R. 1975. The population biology of coral reef fishes. Annu. Rev. Ecol. Syst. 6:211-247.

1977. The behavior of chaetodontid fishes with special reference to Lorenz's "poster coloura- tion" hypothesis. J. Zool., Lond. 183:213-228.

Ehrlich, P. R., R. R. White, M. C. Singer, S. W. McKechnie, and L. E. Gilbert. 1975. Checkerspot butterflies: a historical perspective. Science 188:221-228.

Goldman, B. 1973. Aspects of the ecology of the coral reefs of One Tree Island. Ph.D. diss. Macquarie University, North Ryde, Australia.

Harper, J. L. 1977. Population biology of plants. Academic Press, New York. Hobson, E. S. 1974. Feeding relationships of teleostian fishes in coral reefs in Kona, Hawaii. Fisheries

Bull. 72:915-1031.




Jones, R. W. 1968. Ecological relationships in Hawaiian and Johnston Island Acanthuridae (sur- geonfishes). Micronesica 4:309-361.

Kohn, A. J. 1959. The ecology of Conus in Hawaii. Ecol. Monogr. 29:47-90. . 1968. Microhabitats, abundance, and food of Conus on atoll reefs in the Maldive and Chagos

Islands. Ecology 49:1046-1061. . 1971. Diversity, utilization of resources, and adaptive radiation in shallow water marine

invertebrates of tropical oceanic islands. Limnol. Oceanogr. 16:332-348. Levins, R., and D. Culver. 1971. Regional coexistence of species and competition between rare

species. Proc. Natl. Acad. Sci. USA 68:1246-1248. Pianka, E. R. 1976. Competition and niche theory. Pages 114-141 in R. M. May, ed. Theoretical

ecology, principles and applications. Saunders, Philadelphia. Pulliam, H. R. 1975. Coexistence of sparrows: a test of community theory. Science 189:474-476. Randall, J. E. 1967. Food habits of reef fishes of the West Indies. Stud. Trop. Oceanogr. (Miami)

5:665-847. Reese, E. S. 1977. Coevolution of corals and coral feeding fishes of the family Chaetodontidae. Proc.

3d Int. Coral Reef Symp. 1:267-274. Richardson, R. H. 1970. Models and analyses of dispersal patterns. Pages 79-103 in Ken-ichi Kojima,

ed. Mathematical topics in population genetics. Springer-Verlag, New York. Roughgarden, J. 1974. Species packing and the competition function with illustrations from coral reef

fish. Theor. Popul. Biol. 5:163-186. Roughgarden, J., and M. Feldman. 1975. Species packing and predation pressure. Ecology 56:489-

492. Sale, P. 1975. Patterns of use of space in a guild of territorial reef fishes. Mar. Biol. (Berl.) 29:89-97.

. 1977. Maintenance of high diversity in coral reef fish communities. Am. Nat. 111:337-359. Schoener, T. W. 1974. Resource partitioning in ecological communities. Science 185:27-39. Schultz, L. P. 1943. Fishes of the Phoenix and Samoan Islands collected in 1939 during the expedition

of the U.S.S. "Bushnell." Bull. USNM 180. Steene, R. C. 1978. Butterfly and angelfishes of the world. Australia. Wiley-Interscience, New York. Talbot, F. H., B. C. Russell, and G. R. V. Anderson. 1978. Coral reef fish communities: unstable,

high-diversity systems? Ecol. Monogr. 48:425-440.



Documents

The Community Structure of Coral Reef Fishes