UC Davis UC Davis Previously Published Works Title Resource use by five sympatric parrotfishes in the San Blas Archipelago, Panama Permalink https://escholarship.org/uc/item/0bt697fg Journal Marine Biology, 125(3) ISSN 0025-3162 Authors McAfee, ST Morgan, SG Publication Date 1996-05-01 Peer reviewed eScholarship.org Powered by the California Digital Library University of California Marine Biology (1996) 125:427-437 (cid:14)9 Springer-Verlag 1996 S. T. McAfee (cid:12)9 S. G. Morgan Resource esu yb five sympatric parrotfishes ni the naS Bias Archipelago, Panama Received: 25 September 1995/Accepted: 6 November 1995 Abstract Resource use by five sympatric species of par- changing access to different types of food. All of these rotfish was quantified in the San Blas Archipelago of parrotfishes fed throughout the daytime, and resource the Republic of Panama from March to August 1987. use did not differ between morning and afternoon. Detailed observations of parrotfishes on patch reefs and surrounding seagrass beds showed that they parti- tion resources with respect to habitat, food and size, noitcudortnI but not time. Although parrotfishes shared resources, the proportions of each resource used differed signifi- How so many species of fishes coexist on coral reefs cantly among species. Scarus iserti (Bloch) scraped remains a major ecological question. Niche diversifica- filamentous microalgae that grew from eroded coral tion through competition initially was favored as an pavement on lower slopes of patch reefs and in "halos," explanation, but strong evidence that competition lim- the area of sparse vegetation surrounding reefs. amosirapS ited distributions and abundances of reef fishes was ediriv (Bonnaterre) foraged on upper slopes of patch reefs difficult to obtain. Meanwhile, facilitation, disturbance, where they mostly took bites from dead coral and asso- predation and recruitment also have been proposed as ciated algae. .S aurofrenatum (Cuvier and Valenciennes) processes that structure reef fish communities (see Ebel- had the broadest diet, which consisted mostly of sea- ing and Hixon 1991; Sale 1991 for reviews). Although grasses and macro- and microalgae that were attached to multiple forces likely structure these communities, dead coral on lower reef slopes and in halos. Although .S compelling evidence for any model remains elusive. chrysopterum (Bloch and Schneider) commonly occurred A variety of predictions must be satisfied to demon- on patch reefs, it primarily foraged in seagrass beds that strate that any of these processes regulate community surround them. .S rubripinne (Cuvier and Valenciennes) structure (see Ebeling and Hixon 1991 for review), and was distributed most widely, ranging from seagrass such demonstrations likely would derive from a sub- beds to reef crests, where it took bites from seagrasses, stantial knowledge of complex communities. Detailed dead coral and macroalgae. Juveniles of all species observational studies provide the basis for much of this occurred on lower slopes or in halos where they understanding. scraped filamentous microalgae from coral pavement. Careful observations of resource use formed the As they matured, parrotfishes moved into other habitats foundation for competition theory and continue to be important to its evaluation even though the merits of simplifying experimental manipulations have become Communicated by J.P. Grassle, New Brunswick recognized widely (Wiens 1977; Connell 1980, 1983; S.T. McAfee t Alley 1982; Schoener 1982; Sale 1991). A major tenet of Moss Landing Marine Laboratories, P.O. Box 450, competition theory is that animals coexist by partition- Moss Landing, California 95039, USA ing space, food or time. Fine partitioning of resources is S.G. Morgan () apparent for some coral reef fishes but many other reef Marine Sciences Research Center, fishes do not appear to be particularly specialized (Sale State University of New York, and Dybdahl 1975; Clarke 1977; Sale 1977; Talbot et al. Stony Brook, New York 11795-5000, USA 1978; Ross 1986). One such group is parrotfishes. Present address: Parrotfishes are large, abundant, conspicuous mem- 1Marine Sciences Research Center, State University of New York, bers of coral reef fish communities that forage over reefs Stony Brook, New York 11795-5000, USA during the daytime (Robblee and Zieman 1984; Lewis 428 and Wainwright 1985). Scarids 97( species) comprise parrotfishes that occur on patch reefs along the Carib- one of the two most speciose families of primarily bean coast of Panama. herbivorous reef fishes (Choat 1991). With beak-like terminal jaws, parrotfishes feed nonselectively by tak- ing bites from multispecific algal turfs or by scraping slairetaM dna sdohtem dead coral for endolithic microalgae. They also feed selectively by taking deliberate bites of seagrasses, mac- roalgae and live coral (Hiatt and Strasburg 1960; Lit- The present study was conducted in the San Blas Archipelago, off the Caribbean coast of the Republic of Panama, on 17 patch reefs tler et al. 1989; Bellwood and Choat 1990; Choat 1991; within a 3 km 2 area near the Smithsonian Tropical Research Insti- Bruggemann et al. 1994). Strong, well-developed phar- tute's field station from 30 March to 9 August 1987 (see Robertson yngeal mills grind the ingested matter (Frydl and 1987 for location of study site). These patch reefs are sheltered from Steam 1978; Lobel and Ogden 1981; Bruggemann et al. waves by a barrier reef and experience high sedimentation and 1994). Laterally compressed fusiform bodies, and well- occasional low salinities from discharge by several large rivers (Og- den and Buckman 1973). Patch reefs generally occur in less than developed pectoral and pelvic fins permit slow swim- 10 m of water shoreward of a barrier reef and are comprised pri- ming and precise control of body orientation during marily of finger coral (Porites porites), staghorn coral (Acropora episodes of continuous, rapid grazing over small areas cervicornis), elkhorn coral (Acropora stagnalis), fluted coral ( Agaricia of the reef (Choat 1991). Parrotfishes are protogynous agaricites) and brain coral (Monastrea spp.). Mixed seagrasses (Thalassia testudinum and Syringodium filiforme), silty sand, gor- hermaphrodites and often have dichromatic life history gonians and scattered coral heads surround reefs. phases, termed initial (generally female) and terminal Feeding habits of parrotfishes were studied while snorkeling or (always male) phases, which are readily distinguished in SCUBA diving on the 17 reefs from dawn to dusk when these the field (Barlow 1975; Choat and Robertson 1975). diurnal parrotfishes were active. A total of 198 individuals 36 Caribbean scarids spawn every afternoon throughout Scarus iserti (Bloch), 47 Sparisoma viride (Bonnaterre), 41 S. auro- frenatum (Cuvier and Valenciennes), 38 S. chrysopterum (Bloch and the year, although spawning activity increases during Schneider), 36 S. rubripinne (Cuvier and Valenciennes) that were at summer (Colin 1978; Robertson and Warner 1978). least 2-cm long were followed for 5 min each. The number of bites Many parrotfishes retire to holes in the reef to sleep at taken from algae and other items, estimated total length, phase night (Winn and Bardach 1959). (initial or terminal) and time of day were recorded for each fish Although 31 species of scarids occur in the Carib- followed. Foods consisted of four macroalgae (Udotea spp., Penicil- lus spp., Halimeda spp., Dictyota spp.), two seagrasses (Thalassia bean (Randall 1968), only five species of parrotfishes testudinum, Syringodium filiforme), a thin layer of flocculant matter are abundant on patch reefs along the Caribbean coast covering smooth eroded coral, which was termed coral "pavement", of Panama. Preliminary observations indicated that "algal turf', which grew on dead coral, and sand. Multispecific algal Scarus iserti was the most abundant parrotfish fol- turf consists of macroalgae as well as filamentous, crustose and unicellular algae with representatives from five divisions (Cy- lowed by four species in the genus Sparisoma: S. viride, anophyta, Chlorophyta, Chrysophyta, Phaeophyta and Rhodo- S. aurofrenatum, S. chrysopterum, and S. rubripinne. phyta; Carpenter 1986; Bruggemann et al. 1994). Algal turfs form Several other species were uncommon on these reefs, a single category, because the plant species are interspersed and including Scarus taeniopterus, Sc. guacamaia, Sc. co- parrotfishes are nonselective at this level (Bruggemann et al. 1994). eruleus, Sc. coelestinus and Sc. vetula (also see Wein- Bites frequently were taken from seagrasses, coral pavement, dead coral, Halimeda spp. and Dictyota spp. Foods that were eaten stein and Heck 1979). In addition, the small parrot- infrequently (Udotea spp., Penicillus spp., live coral, sponges) and fishes, S. radians and S. atomarium, occurred in seagrass bites of sand were combined into a single category called "other." beds adjacent to reefs, while Cryptotomus roseus were The availability of these six food types was quantified by record- found in seagrass beds farther from patch reefs. ing the occurrence of plants, macroinvertebrates or substrates on Preliminary observations on patch reefs in the San two of the reefs, "Smithsoniantupo-l" and "Porvenir-20," which were approximately 80 and 25 m in diameter, respectively. Potential Blas Archipelago revealed little evidence of resource foods were recorded at 20-cm intervals along 20-m long haphazardly partitioning in this assemblage of parrotfishes, and placed transects. A total of 51 transects were followed in five zones: evidence from other functional groups of parrotfishes crest, upper slope, lower slope, halo and adjacent seagrass bed. The has been mixed (Choat 1969; Roughgarden 1974; Russ crest was defined as the reef flat that was awash at low tide. The ;4891 Lewis and Wainwright 1985). Therefore, a detailed upper slope extended 3 m below the reef crest, and the lower slope extended 3 m above the bottom. The halo was the 3-m zone of sparse evaluation of resource use by parrotfishes would augment vegetation that surrounded the reef, and the seagrass zone extended the body of knowledge needed to evaluate the importance 3 m into seagrass beds surrounding the halo. of competition in reef fish communities and provide new Parrotfishes on Smithsoniantupo-1 and Porvenir-20 reefs were information on the ecology of this important group. censused visually within each of the five zones between dawn and dusk (06:30 to 19:00 hrs) using a technique developed by Brock Despite considerable interest in resource partitioning (1954), reviewed by Sale (1980) and modified by Lewis and Wain- by reef fishes, few investigators simultaneously have wright (1985). Parrotfishes within 5.1 m of a 20-m transect line were examined resource use along all three major niche counted; each fish was categorized by species, phase, estimated total dimensions: space, food and time (see Ross 1986 for length (see Table 1 for size classes of each species) and time of day. review). In addition, ontogenetic shifts in resource use Brock (1982) tested the reliability of this technique and found excel- lent agreement between the number of parrotfishes that were cen- may occur. The purpose of the present study was to sused visually and the number that were removed after rotenone was determine spatial, temporal and developmental pat- applied to a patch reef. Lengths of fishes were validated periodically terns of habitat and food use by the five most common by comparing the estimated length of fish to background landmarks 924 Table I suracS iserti and amosirapS spp. Size classes based on total among species (Fig. 1; Table 2). Scarus iserti ate micro- lengths (cm) of five species of parrotfishes observed on 71 patch sfeer algae associated with coral pavement, Sparisoma viride in the San Blas Archipelago, Panama primarily took bites from dead coral and S. chryso- seicepS Small Medium Large pterum mostly ate seagrasses. Both S. aurofrenatum and S. rubripinne mostly consumed seagrasses and dead .cS itresi 3-9 10-14 > 41 .S ediriv 3-17 81 23 >32 .S mutanerforua 5-13 41 22 >22 .S muretposyrhc 4-15 16-27 >27 .S ennipirbur 5 71 18-30 >30 that were later measured. Estimated lengths and measured distances of landmarks were generally within 2 cm for small fish and 3 cm for -..._ large fish. Fishes were censused along transects in the five zones of 02 the two reefs a total of 501 times. Transects were removed and -- replaced haphazardly. .si k Behavioral interactions of parrotfishes were observed for indica- tions of competitive behaviors. The participants were identified, their lengths were estimated, displays, postures and color changes were described and durations of encounters were noted. Niche overlap indices, each with their particular biases, were not calculated for the many variables that were included in this multi- dimensional study of resource use. Instead, the Scheirer-Ray-Hare -->~ ~ " ~ , . I j~./..~ ~..~....~ extension of the Kruskal-Wallis test was used to determine similar- ities in resource use by the five species with respect to habitat, diet, time, fish length and phase. This nonparametric test was used in lieu -~.-~ ~<~ ~-~ of a multiway analysis of variance (Sokal and Rohlf .)5991 Nonsig- nificant interaction terms were removed to increase the degrees of freedom, and data were reanalyzed. All data were graphed in three B dimensions to best portray multidimensional patterns of resource use, but this precluded the inclusion of error bars. A two-way test was used to determine the similarity of feeding by the five species of parrotfishes among the five zones. A four-way test 08 was used to determine whether or not the diet of each species shifted ontogenetically, temporally and with respect to phase. Hourly feed- ing rates of parrotfishes were determined to ensure that peak feeding rates did not occur near noon, before data were divided into morn- o4 ings and afternoons. Distributions of potential foods among the five zones were tested using a two-way test. os A three-way test was used to identify differences in abundance of parrotfishes among the five zones of reefs during mornings and afternoons. Distributions of phases and sizes of parrotfishes were analyzed separately in two-way tests. Individuals could be scored as occurring among zones of the reef, but could not be counted again as belonging to particular size classes and phases within zones. There- fore, statistical interactions in multiway tests that would indicate whether or not adults migrated to different areas of reefs to spawn during afternoons were precluded. stluseR O2: 1 Dietary patterns of resource use IS. 51 ! All five parrotfishes ate each of the six types of foods, but the proportions of foods eaten differed significantly o Fig. 1 Scarus iserti and Sparisoma spp. A Mean number of bites of six food items taken by five species of parrotfishes during 5-rain observa- tion periods on 17 patch reefs. 198 individuals (36 Sc. iserti, 47 S. viride, 41 S. aurofrenatum, 38 S. chrysopterum, 36 S. rubripinne) at least 2-cm long censused. B Distribution of foods (mean percentage) among five zones of two of the patch reefs. Foods and substrates identified under 15 transect lines every 20 cm. A total of 15 transects censused. C Mean number of five parrotfishes per 20 x 3 m transects in five zones of two patch reefs. A total of 105 transects censused 430 Table 2 Scarus iserti and Sparisoma spp. Scheirer-Ray Hare tests. Small . Number of bites of six food types (Food) taken by five species of parrotfishes (Species) during 5-rain observation periods conducted on 17 patch reefs. 198 individuals (36 Sc. iserfi, 47 S. viride, 41 S. ,mutane'iforua 38 S. chrysopterum, and 36 S. rubripinne) at least 2-cm long censused. Abundances of food types (Food) available to parrot- fishes that occur in five zones (Zone) of two patch reefs. A total of 15 transects censused. Abundances of five parrotfishes (Species) during mornings and afternoons (Time) in five zones (Zone). A total of 105 transects censused, ns indicates p > 0.05 =, i :I05 Test df SS H p Source :0 Number of bites Food 5 16725235 144.92 <0.001 Species 4 17679898 153.20 <0.001 Food x Species 20 18721920 12.97 < 0.001 .'~ ~,x..~. o~ ~, Error 1152 83168206 Food availability Food 5 2172.10 3.36 ns Medium . Zone 4 10866.16 16.82 <0.010 ..... i! Food x Zone 20 37811.62 58.54 <0.001 Error 60 6641.08 4o ::i 04 Number of fishes Zone 4 1642651 72.78 <0.001 Species 4 391313 17.34 <0.010 Time I 33121 1.47 ns Zone x Species 16 3110748 137.82 <0.001 Error 499 6649516 " ;0 coral, but S. aurofrenatum did not forage in seagrass 2 beds as did S. rubripinne. Furthermore, S. aurofrenatum had a more diverse diet than S. rubripinne; S. auro- ,~..~ .4;~ "~ .r o..,:~, -'-'~,~>. frenatum took more bites from coral pavement, Hali- meda spp., Udotea spp., Sargassum spp., crabs, urchins Large and sponges. Although S. aurofrenatum is more omniv- orous than the other four species, many parrotfishes (cid:12)9 45, prefer sponges to algae when available (S.T. McAfee unpublished data; J. Wulff personal communication). Nonsignificant ontogenetic shifts in diet were evident for four parrotfishes (Figs. ,2 ;3 Table .)3 Juveniles of all five species took bites from pavement in halos and on lower slopes, and all mature fishes, except Scarus iserti, switched to other foods. Shifts in diet corresponded to shifts in habitat (see subsection "Spatial patterns of 5 ; resource use"); parrotfishes generally ate the com- monest foods wherever they occurred. Food availability These patch reefs were surrounded by expansive beds Fig, 2 Scarus iserti and Sparisoma spp. Mean number of bites of six of Thalassia testudinum and Syringodium filiforme, and food items taken by three size classes of five species of parrotfishes during 5-min observation periods on 17 patch reefs. See Fig. 1 for small amounts of green algae (Udotea spp. and Hali- numbers of parrotfishes censused meda spp.) were interspersed among seagrasses (Fig. ;1 Table .)2 Seagrasses were less dense immediately next meda spp., Penicillus spp. and Udotea spp. occurred in to patch reefs, presumably due to grazing by urchins halos. Lower slopes, upper slopes and crests of patch and herbivorous fishes (Randall 1965; Ogden et al. reefs were comprised mostly of dead coral and asso- 1973). Consequently more sand, coral pavement, Half ciated multispecific algal turf, lesser amounts of 134 gninroM ~ noonretfA "" 2 " " . . . . laitinI Phase ~ lanimreT Phase 04 0~ 0~ Fig. 3 Scarus iserti and Sparisoma spp. Mean number of bites of six curred in halos and on lower and upper slopes of reefs, food types taken by five species of parrotfishes during 5-rain obser- but the extent to which zones were used by each species vation periods mornings and afternoons on 71 patch reefs (upper differed. Seagrass beds and reef crests were used less .)smargaid Mean number of bites of food types taken by initial and often than the other three zones by most parrotfishes. terminal phase parrotfishes during 5-rain observation periods on patch reefs (lower diagrams). eeS Fig. 1 for numbers of parrotfishes Three species of parrotfishes primarily occurred on censused reef slopes and in halos. Scarus iserti was most abun- dant on lower reef slopes but also was common in halos (Fig. .)1 Sparisoma viride was most abundant on upper Halimeda spp. and Dictyota spp. (which also grew on reef slopes, and .S aurofrenatum was most common in dead coral), and live coral. In addition, a small amount halos and on lower slopes. S. chrysopterurn primarily of coral pavement extended from the halo onto lower occurred in seagrass beds and halos but also was com- slopes of reefs. mon on lower slopes. .S rubripinne is the only parrotfish that was common in all five zones, but it was present in lower abundance. Thus despite overlapping distribu- Spatial patterns of resource use tions, highly significant differences in habitat use by parrotfishes were evident. Each of the five species of parrotfish had a unique Highly significant ontogenetic shifts in habitat use pattern of distribution among zones despite overlap in were evident for four species (Fig. 4; Table .)4 Juvenile habitat use (Fig. ;1 Table .)2 All parrotfish species oc- fishes at least 2-cm long primarily occurred around 432 Table 3 Scarus iserti and Sparisoma spp. Four-way Scheirer- llamS J { 0 i Ray-Hare tests of number of bites of six foods (Food) taken by small, medium and large (Size) and initial and terminal phase (Phase) parrotfishes during mornings and afternoons (Time) on 17 patch reefs, ns indicates p > 0.05. See Table 1 for numbers of parrotfishes censused Species df SS H p Source 4 { Sc. iserti Food 5 4974261 90.8413 <0.001 Size 2 2435 0.0445 ns Phase 1 54 0.0010 ns Time I 24 0.0004 ns Error 206 6796132 0 S. viride Food 5 13415173 115.6547 <0.001 Size 2 2048766 14.6990 < 0.050 ..,,5~.-.~ -~ Phase 1 147441 1.0578 ns Time 1 216947 1.5565 ns muideM J Error 206 20632928 S. aurofrenatum Food 5 2527456 46.7970 <0.001 Size 2 225742 4.1797 ns Phase 1 1701 0.0315 ns Time 1 69600 1.2887 ns ,L i Error 206 10407688 S. chrysopterum Food 5 3359736 61.8385 <0.001 Size 2 541914 6.6096 < 0.050 Phase 1 46681 0.5694 ns Time 1 60 0.0007 ns Size x Time 2 528054 6.4406 < 0.050 Error 206= 18592326 S. rubripinne r .~. ~ :::: ......... . Food 5 6465529 46.1342 < 0.001 Size 2 1471752 9.0753 < 0.050 Phase 1 230 0.0014 ns Time 1 831461 5.1270 <0.050 Error 206 23135740 egraL halos or lower reef slopes and moved into other zones as they grew. Scarus iserti moved from halos and lower slopes to lower and upper slopes as they matured. Sparisoma viride shifted from lower slopes to upper slopes, and S. aurofrenatum moved from halos and lower slopes to upper slopes. S. chrysopterum shifted from halos to seagrass beds as they developed. S. ru- bripinne typically moved from lower slopes to atl zones, but the apparent ontogenetic shift in habitat use was not significant because few fishes occurred in transects. The distributions of initial and terminal phase par- rotfishes often differed (Fig. ;5 Table 5) because ter- minal phase parrotfishes were associated with preferred spawning sites. Terminal phase Scarus iserti spawned on upper and lower slopes, but initial phase fish pri- Fig. 4 Scarus iserti and Sparisoma spp. Mean number of three size marily occurred on lower slopes and in halos. Sparisoma classes of five parrotfishes per 20 x 3 m transect in five zones of two patch reefs. A total of 105 transects censused viride and S. aurofrenatum spawned with females on upper reef slopes. However, most initial phase S. auro- frenatum occurred in halos and on lower slopes, whereas 433 Table 4 Scarus iserti and Sparisoma spp. Two-way Scheirer- tween Scarus iserti and .S coeruleus. In these cases, Ray-Hare tests of abundances of three sizes (Size) of parrotfishes in fishes engaged in erect fin posturing, color changes, and five zones (Zone) on two patch reefs. A total of 501 transects chasing. Parrotfishes also were observed acting aggres- censused, ns indicates p > 0.05 sively toward nonscarids on two occasions, when S. aurofrenatum pursued black-ear wrasses (Halichoeres Species df SS H p Source poeyi), apparently to steal food. Sc. iserti Zone 4 8066631 35.115 < 0.001 Size 2 5068135 22.062 <0.001 noissucsiD Zone x Size 8 5221037 22.728 <0.010 Error 300 53776007 Parrotfishes showed significant differences in resource S. viride Zone 4 14327828 73.973 <0.001 use with respect to habitat, food and size, but not time. Size 2 1372205 7.085 < 0.050 Parrotfishes that used similar spatial resources used Zone (cid:141) Size 8 5975853 30.853 <0.001 different food resources. Adult Scarus iserti, Sparisoma Error 300 39143040 viride and .S aurofrenatum frequented upper reef slopes S. aurofrenatum but ate different foods (Table .)6 Sc. iseni scraped Zone 4 3241888 583.61 <0.001 microalgae from pavement, .S viride took bites from Size 2 1750490 8.847 <0.050 dead coral and .S aurofrenatum ate a wide variety of Zone x Size 8 4025605 20.346 <0.050 Error 300 53107985 foods, including seagrasses, various macroalgae and microalgae growing from dead coral and pavement, S. chrysopterum crabs, urchins and sponges. Adult .S chrysopterum Zone 4 4067150 21.092 <0.001 Size 2 638538 113.3 ns foraged on seagrasses in seagrass beds, and .S ru- Zone (cid:141) Size 8 3207695 536.61 <0.050 bripinne ranged over all five zones taking bites from Error 300 52635627 dead coral, seagrasses and macroalgae. Thus, each spe- S. rubripinne cies had a unique pattern of resource use, although Zone 4 6293012 33.756 <0.001 none of the parrotfishes exclusively used a particular Size 2 583315 3.129 ns habitat or food. This type of resource use may be Zone x Size 8 1837335 9.855 ns common among reef fishes, many of which do not show Error 300 49824618 obvious differences in resource use (Sale and Dybdahl 1975; Clarke 1977; Sale 1977; Talbot et al. 1978; Ross 1986). However, whether or not reef fish assemblages initial phase .S viride co-occurred with terminal phase showing this type of resource use are maintained by fish. .S chrysopterum typically spawned over grassbeds partitioning resources is equivocal. Although parrot- where adults were most abundant, and initial phase fish fishes that used similar resources in one niche dimen- occurred mostly in halos and secondarily in seagrass sion used different resources in another dimension, as beds and on lower slopes. Initial and terminal phase .S predicted by competition theory, the overlap in re- rubripinne frequented all zones; however, most terminal source use is not consistent with competitive exclusion phase fishes occurred on upper slopes, except during through niche separation. afternoons, when they swam off reefs to spawn over Aggressive, nonsexual encounters among parrot- seagrass beds. fishes were rare, which suggests that competition among these species was minimal. Either competition may not be important in structuring this parrotfish Temporal patterns of resource use assemblage, or the abundance of algae on patch reefs reduced competitive interactions among parrotfishes. Food and habitat use generally did not change signifi- The decimation of the herbivorous sea urchin (Diadema cantly during the day (Figs. ,3 ;5 Tables ,2 .)3 Although )nrurallitna by a waterborne disease throughout the parrotfishes tended to eat more in the afternoon than in Caribbean in 1983 (Lessios 1988a) led to measurable the morning, temporal changes in the composition of increases in the abundance of filamentous and frondose diets were not found for any species (Fig. ;3 Table .)3 algae (Littler and Littler 1984; Hay and Taylor 1985). This increase in algal abundance has been correlated with increases in abundances of parrotfishes and other Behavioral interactions herbivorous reef fishes in the San Blas islands and elsewhere in the Caribbean (Hay and Taylor 1985; Only four competitive, nonsexual interactions between Carpenter 1990; Robertson 1991). Reduced urchin parrotfishes were seen during the hundreds of hours densities also corresponded to invasion of shallow reef that they were observed. Three of the four encounters habitats by herbivorous fishes that previously had been involved Sparisoma viride, and the other occurred be- found deeper on reefs (Carpenter 1988; Morrison 1988), 434 Afternoo~~ ''/ O2 i 6" Fig. 5 Scarus iserti and Sparisoma spp. Mean number of five parrot- ago, aggressive nonsexual interactions among parrot- fishes per 20 x 3 m transect in five zones of two patch reefs mornings fishes may increase, population sizes may diminish, and and afternoons (upper diagrams). Abundances of initial and terminal habitat use and foraging behaviors may be expected to phase parrotfishes in the five zones (lower diagrams). A total of 105 change if competition is important in structuring this transects censused assemblage of parrotfishes. The abundance of algae raises another point con- suggesting that grazing by urchins also may affect habi- cerning the foraging ecology of parrotfishes. ademilaH tat use and foraging behavior of parrotfishes. Despite spp., Dictyota spp. and Udotea spp. produce secondary two to three-fold increases in the numbers of parrot- compounds that deter herbivory by reef fishes (Hay fishes and other herbivorous fishes, algae continued to 1991). Although none of these macroalgae was a pri- increase, indicating that increased grazing by parrot- mary component of the diet of any of the five species of fishes does not compensate for the decline of urchins parrotfishes, they comprised a considerable portion of (Carpenter 1990; Robertson 1991). When this study was the diet of amosirapS ennipirbur and .S ,mutanerforua conducted at San Blas in 1987, .D antillarum popula- and commonly were eaten by the other three species. tions were still < 10% of previous mean densities (3.5 Therefore, secondary compounds of macroalgae may m -2) (Lessios et al. 1988b; Robertson 1991). As urchin not strongly deter herbivory by these parrotfishes be- populations continue to rebound and algae becomes cause they were commonly eaten despite the higher less abundant on patch reefs in the San Blas Archipel- abundances of algal turf and seagrasses. 435 Schoener (1974) suggested that habitat dimensions tebrates apparently partition food resources, and space are more important than food dimensions, which in and time are of secondary and tertiary importance, turn are more important than temporal dimensions of respectively (Branch 1984; Ross 1986). In the present resources for niche separation. However, coral reef study, habitat use and diet were interrelated and could fishes, temperate reef fishes and mobile marine inver- not be ranked. Although resources were not partitioned temporally, feeding rates of parrotfishes tended to in- crease in the afternoon. This may have resulted from Table 5 Scarus iserti and Sparisoma spp. Two-way Scheirer-Ray- Hare tests of abundances of initial and terminal phase (Phase) high levels of photosynthate in productive turf and parrotfishes in five zones (Zone) on two patch reefs. A total of 501 seagrass communities at this time (Polunin and transects were censused Klumpp 1989). Although typically overlooked in resource use stud- Species df SS H p ies, ontogenetic shifts in habitat and diet were observed Source for four parrotfish species. Adults of these species often Sc. iserti occurred on the upper slope and in seagrass beds, Zone 4 4830076 51.416 < 0.001 whereas juvenile fishes primarily occurred in halos and Phase 1 895895 9.537 <0.001 on lower reef slopes where they apparently exploited Zone x Phase 4 1707009 171.81 <0.001 diatoms and small protein-rich invertebrates living in Error 200 12200804 algal mats (see Bellwood 1988; Bruggemann et al. 1994) S. viride and epiphytes on macroalgae and seagrasses. On- Zone 4 1641556 66.685 <0.001 Phase 1 5798320 18.685 < 0.001 togenetic changes (nonsignificant) in diet likely resulted Zone x Phase 4 2042358 23.489 <0.001 from a mechanical limitation in jaw structure. Small Error 200 8690398 parrotfishes (2 to 4 cm) could not bite through most S. aurofrenatum macroalgae, seagrasses or coral exoskeletons. On- Zone 4 1831905 21.472 <0.001 togenetic shifts in the diets ofScarus iserti did not occur Phase 1 1668538 19.557 < 0.001 because these small parrotfishes consumed mostly dia- Zone x Phase 4 1785010 20.922 <0.001 toms and filamentous algae throughout their lifetimes. Error 200 12545973 In conclusion, the five sympatric species of parrot- S. chrysopterum fishes shared resources, and few aggressive nonsexual Zone 4 1803130 23.094 <0.001 encounters occurred in the vicinity of patch reefs. How- Phase 1 1218547 15.607 < 0.001 Zone x Phase 4 2171603 27.813 <0.001 ever, significant differences in space and food use were Error 200 11125294 observed. Whether or not similar patterns of resource S. rubripinne use by these species occur on barrier reefs remains to be Zone 4 1877761 21.638 <0.001 examined. The observed interspecific differences in re- Phase 1 483546 5.572 < 0.050 source use may occur if this assemblage of parrotfishes Zone x Phase 4 804761 9.274 0.050 was structured by competition, predation, larval recruit- Error 200 14970831 ment or chance. Additional surveys and complementary Table 6 Scarus iserti and Sparisoma spp. Summary of Species Habitat Food habitat and food use by five sympatric parrotfishes (small Overall Small Large Small Large and large) in five zones (reef crest, upper reef slope, lower reef Sc. iserti Upper Upper Pavement Pavement slope, halo and surrounding Lower* Lower Lower seagrass) of Caribbean patch Halo* Halo reefs. Asterisks indicate zones S. viride Upper* Upper Pavement Dead Coral where fishes predominated Lower Lower Dead Coral S. aurofrenatum Upper Upper Pavement Dead Coral Lower* Lower Dead Coral Seagrasses Halo* Halo Seagrasses IIaIimeda spp. S. chrysopterum Lower Pavement Seagrasses Halo* Halo Seagrass* Seagrass S. rubripinne Crest Crest Pavement Seagrasses Upper Upper Seagrasses Dead Coral Lower Lower Lower Dead Coral Halo Halo Seagrass Seagrass