First data on community structure and trophic networks oF uvea coral reeF Fish assemblages (wallis and Futuna, south paciFic ocean) by Laurent Wantiez & Claude Chauvet (1) abstract. - Coral reef fish communities of uvea (Wallis and Futuna) fringing reefs, mid-lagoon reefs, inner barrier reefs and outer barrier reef slopes were studied along four transects using visual censuses during September 1999. a total of 194 fish species in 33 families were recorded. Labridae (34 species), Pomacentridae (33 species) and Chaetodontidae (23 species) were the main families. the composition of the species suggests that uvea is located at a faunal junction between Pacific biogeographic regions. Mean density (2.5 fish/m2) was similar to other Pacific reef fish communities whereas mean biomass (43 g/m2) was within the lowest known values. Four fish assemblages were identified along an inshore-offshore gradient and related to substrate characteristics (live corals, dead corals, seagrass, algae and sand): One seagrass bed assemblage, two lagoon assemblages and one outer barrier reef slope assemblage. the major energy intake comes from lower levels of the trophic network, grazers (45.1% biomass) being the main trophic group. Other important trophic categories came from upper trophic levels: macro-carnivores (24.4%) and piscivores (14.5%). Micro-algae → grazers → piscivores was the most important energy path, before macro-invertebrates → macro-carnivores → piscivores. Plankton represented a minor resource (only 5% planktivores) during the study, which is unusual. Spatial variations in trophic networks occurred among the four fish assemblages with significant variations for grazers and macro-carnivores, and high variations for piscivores, planktivores and corallivores. rÉsumÉ. - Premières données sur la structure des communautés ichtyques et les réseaux trophiques des récifs coralliens d’uvéa (Wallis et Futuna, Sud Pacifique). Les communautés de poissons des récifs frangeants et intermédiaires du récif barrière interne et de la pente externe d’uvéa (Wallis et Futuna) ont été étudiées en septembre 1999 le long de 4 radiales par comptages en plongée. un total de 194 espèces pour 33 familles a été recensé. Les Labridae (34 espèces), les Pomacentridae (33 espèces) et les Chaetodontidae (23 espèces) sont les familles les plus diversifiées. La composition spécifique semble indiquer qu’uvéa se situe à la fron- tière de plusieurs régions biogéographiques du Pacifique. La densité moyenne (2,5 ind./m2) est globalement comparable aux autres communautés de poissons récifaux dans le Pacifique. en revanche, la biomasse moyenne (43 g/m2) figure parmi les plus faibles valeurs signalées dans la région. Quatre assemblages ont été identifiés le long d’un gradient côte- large. ils ont pu être reliés aux caractéristiques du substrat (corail vivant, corail mort, phanérogames, algues et sable) : un assemblage caractéristique des herbiers à phanérogames, deux assemblages lagonaires et un assemblage typique de la pente externe. La principale source d’énergie utilisée par ces communautés provient des plus bas niveaux trophiques, les brouteurs (45,1% de la biomasse) constituant le principal groupe trophique. Les autres principales catégories trophiques proviennent de niveaux trophiques supérieurs : les macrocarnivores (24,4%) et les piscivores (14,5%). La principale voie de transfert d’énergie est micro-algues → brouteurs → piscivores, devant la voie macro-invertébrés → macrocarnivores → piscivores. Le plancton était une source d’énergie peu utilisée (seulement 5% de planctophages) pendant cette étude, ce qui est inhabituel. Les réseaux trophiques des quatre assemblages présentaient une variabilité spatiale. Ces variations ont été significatives pour les brouteurs et les macrocarnivores, et fortes pour les piscivores, les planctophages et les corallivores. Key words. - Fish assemblages - new Caledonia - trophic networks - Coral reefs - habitat. high diversity, complex community structures and distinguish fringing reef, island and platform reef and bar- trophic networks characterize coral reef fish communities. rier reef assemblages (Williams, 1982; Williams and Spatial patterns of diversity and community structures have hatcher, 1983; McGehee, 1994; Wantiez et al., 1997; been well studied (Williams, 1991). Within reef, habitat Grimaud and Kulbicki, 1998; Letourneur et al., 1998a). the characteristics are known to be the main factors in structur- status of the reef may also have a significant impact on ing communities from the inner reef flat to the reef slope exploited fish communities, protected vs. unprotected reefs (Russ, 1984; Galzin, 1987; Letourneur and Chabanet, 1994; (Roberts and Polunin, 1991, 1992; Wantiez et al., 1997; McGehee, 1994; Meekan et al., 1995; newman et al., 1997; Sarramégna, 2000), disturbed vs. undisturbed reefs Sarramégna, 2000). Within a region, distance offshore and (Chabanet et al., 1995; Ohman et al., 1997; Jones and Syms, terrestrial runoff gradient are also significant factors that 1998). (1) LeRveM, université de la nouvelle-Calédonie, BP 4477, 98847 nouméa cedex, nOuveLLe-CaLédOnie. [[email protected]] Cybium 2003, 27(2): 83-100. Structure and trophic network of Uvea reef fish Wantiez & chauvet Modern trophic studies take into account all food types materials and methods consumed by a species to set up trophic networks. Such methods minimize biases introduced by traditional classifi- study site cation of a species into a single trophic category corre- uvea (Fig. 1) is the main island (75.4 km2) of Wallis and sponding to the main type of food consumed (Parrish et al., Futuna French territory. Futuna and alofi are located 235 1986). these methods also avoid the creation of a heteroge- km in the Southwest, niua Fo’ 249 km in the Southeast, and neous omnivorous category (thollot et al., 1999). Savaï 352 km in the West. uvea is located at the border of Consequently, trophic networks and main energy paths are the Pacific Plate, close to the West-central part of the more easily identifiable with these methods (Wantiez, australian Plate. in this region, oceanic surface currents 1994a; Letourneur, 1998). they have only been used to headed West and South-West (Reverdin et al., 1994) even in describe coral reef fish trophic networks in Reunion island an el niño situation (delcroix and henin, 1989). in the light (Letourneur, 1998), Moorea (French Polynesia) (Letourneur of these characteristics and considering reef fish larval et al., 1997a) and new Caledonia (Kulbicki, 1992; Kulbicki phase duration and swimming capabilities, uvea is a remote et al., 1996; Letourneur et al., 1997a). Consequently, Pacific island where coral reef fish community is probably trophic networks and energy paths of Pacific coral reef fish characterized by self-recruitment. a lagoon of 200 km2 sur- communities are still poorly described. Such information is rounds this volcanic island. the island and its lagoon are fundamental to the understanding of the functioning of located between latitudes 13°10’ and 13°23’S, and between these communities (Sale, 1991) and essential to the plan- longitudes 176°06’ and 176°17’W. More than 20 coralline ning of adequate management and protection policies. or basaltic islets are present in the lagoon. the lagoon geo- in the Pacific, most surveys of coral reef fish communi- morphology is complex with the presence of basaltic ridges, ties have been done in the Great Barrier Reef and hawaii which separate shallow areas (< 10 m) from deep hollows (Sale, 1991). Several French Polynesian, Micronesian and where depth can exceed 50 m. the lagoon is limited by a new Caledonian reefs have also been well described barrier reef interrupted by three small passes on the western (Kulbicki 1992, 1997). On the other hand, the study of border and one larger pass in the South. an important tidal majority of West Pacific high islands coral reefs have been range (> 2 m) characterizes this lagoon and a first estima- incomplete, even though these habitats are the most com- tion indicates that one third of the lagoon waters is probably mon in this region. Some of these communities have never renewed at each tide. been studied. it is known that biogeographic regions and a stratified sampling design was selected to identify differences between coral reef geomorphology are signifi- coral reef fish community structure and trophic networks in cant in structuring reef fish communities (harmelin-vivien, uvea. this design is the most appropriate to identify com- 1989; thresher, 1991; Kulbicki, 1992, 1997; Lieske and munity structure at a local scale using multivariate methods. Myers, 1994; Letourneur et al., 1997a). Consequently, at two levels of stratification were selected and based on the present there is a bias for certain biogeographic regions, reef knowledge of the general organization of reef fish commu- types and habitats in studies on reef-fish ecology (Ohman et nities among reefs within this region (Williams, 1991). al., 1997). For the Pacific region, this raises the importance Spatial cover of sampling was privileged in order to get data of identifying the structure and functioning of high island on a maximum of reefs with the allocated sampling effort. communities, and analyzing the differences of these with the first level of stratification was a coast to ocean transect, the Great Barrier Reef and atoll reefs. four stations being sampled: one fringing reef, one interme- no data has ever been published on Wallis and Futuna diate reef, one inner barrier reef and one outer barrier reef. (French territories) coral reef fish communities, with the the second level of stratification allocated each set of 4 sta- exception of a limited edition report regrouping only quali- tions around the island in a systematic way, uvea lagoon tative results (Richard et al., 1992) and a study of tradi- being divided into four sectors (Fig. 1). Because of bad tional fishery in Futuna (Galzin and Maugé, 1981). this weather conditions, no intermediate reef was sampled in article presents the first data on species richness, density, sector 3, no outer barrier reef was sampled in sector 1 and biomass, community structure and trophic functioning of the outer barrier reef station in sector 2 was located out of these populations, and their relationships with substrate the transect line. characteristics. these data, which are very important for Wallis and Futuna environmental authorities, also constitute sampling methods an important contribution to the understanding of high Reef fish communities were studied by visual census island reef fish communities, and provide interesting results using the point transect method (Buckland et al., 1993), on the biogeographic distribution of coral reef species between 11 and 20 September 1999. each station was sam- across the Pacific. pled by three point counts, the first on the reef flat, the 84 Cybium 2003, 27(2) Wantiez & chauvet Structure and trophic network of Uvea reef fish second on the reef crest and the third on the reef slope, in Serranidae, large Labridae, Scaridae, acanthuridae, order to sample each part of the reef. in each habitat, a diver Siganidae) were sampled within a visibility radius. the vis- randomly chose a point. to limit the disturbance induced by ibility radius is the maximum distance at which the diver his arrival, the diver waited 5 minutes at this point before can identify large species without confusion. Fish beyond beginning the sampling. then, he counted all fish by spe- 10 m were not sampled. this method minimizes the biases cies, estimated their size and the distance from the point. induced by the low detectability of small species (Kulbicki Fish species were divided into two categories (appendix 1). and Sarramegna, 1999), and minimizes the biases induced Small or cryptic species (such as Chaetodontidae, by the avoidance of the diver by large species easily detect- Pomacentridae, small Pomacanthidae, small Labridae) were able in a 10-m radius (Kulbicki, 1994). sampled in a 3-m radius around the point, and large species the point transect method was choosen to take into (all commercial species such as Carangidae, Lethrinidae, account the general organization of coral reef fish commu- nities within the reef (Williams, 1991) and to privilege spatial cover of sampling. the method is simple, fast, objec- 170°e 180° 170°W tive, repeatable and has been found to be extremely effec- uvéa tive for counting reef fishes (Bohnsack and Bannerot, Samoa 1986). this method is also well adapted to heterogeneous vanuatu Futuna environments (harmelin-viven et al., 1985). Relatively Fidji 15°S shorter count durations than for linear transects increase the number of stations and habitats within a station that can be sampled in a given survey period (Jennings and Polunin, niue 1996). Consequently, this method is adapted to identifica- 20°S tion and analysis of community structure. Moreover, push- tonga pull effects due to the diver presence are reduced due to the nouvelle-Calédonie short period of time spent in a single census area (Samoilys and Carlos, 1992; Jennings and Polunin, 1996). 176°15’W ∞ 176°10’W the substrate was sampled by the line intercept transect 25 method (english et al., 1997) in the same area as the point transects. thirty possible substrate categories were recorded Sector 3 along a 50 m transect which was laid on the bottom from the Sector 2 reef flat to the reef slope. this method is reliable and effi- ∞ cient for obtaining quantitative percent cover data (english 34 ∞∞ 13°15’S ∞ 32 et al., 1997). 33 35 ∞ data analysis ∞ ∞ Fish species richness per station was estimated from the 22 23 24 three points sampled at each station. at each station, density was calculated by: ∞ 3 ∞ uvea 12 13 ∞ ∑ Dk k = 1 D = 13°20’S ∞ 14 3 42 Sector 1 where D = density at the station (fish/m2), Dk = density of point transect k at the station (fish/m2). density of point ∞ transect k was calculated as follows: ∞ Sector 4 43 ∞ 44 Ni Nj 45 ∑ n ∑ n i j i = 1 j = 1 D = + k 13°25’S πR21 πR22 where n = abundance of species i at point transect k, Figure 1. - Location of the study area and the stations sampled in i uvea. Nl = total number of large species counted in a R1 at point Cybium 2003, 27(2) 85 Structure and trophic network of Uvea reef fish Wantiez & chauvet transect k, R = visibility radius, n = abundance of species j ous multivariate analyses. the individual was considered as 1 j at point transect k, N = number of small species counted in the basic functional component in the determination of spe- j a 3-m radius at point transect k, R = 3 m. cies structure (Wantiez et al., 1996). Consequently, the 2 the weight of fish was calculated from length-weight analysis was performed on densities. as the data matrix was relationships previously defined (Kulbicki et al., 1993; characterized by numerous zeros because of the dominance Kulbicki et al., 1994; Letourneur et al., 1998b). Biomass of a small number of species, a correspondence analysis per station was calculated in a similar way to density: using a χ2 similarity coefficient was selected (Legendre and Legendre, 1984). all species were used in the analysis, 3 Ni Nj since rare species can be indicators of brief environmental ∑ B ∑ w ∑ w perturbations (Wantiez et al., 1996). no transformation was k i j B = k = 1 and B = i = 1 + j = 1 applied to the data because a higher inertia was extracted by 3 k πR2 πR2 the analysis when raw data was used. the results of the 1 2 factorial analysis were completed with a cluster analysis. this method is recommended by Lebart et al. (1997) to where B = biomass at the station (g/m2), B = biomass of k facilitate the identification of samples and species assem- point transect k at the station (g/m2), w = weight of species i blages from the projections on the factorial charts. two i at point transect k (g), w = weight of species j at point j clusters were calculated using the Ward aggregative method transect k (g/m2). with (1-r) Pearson distance (Lebart et al., 1997), one on the the percentage of each substrate category at each station samples scores and one on the species scores on the corre- was calculated by: spondence analysis axes. differences between the identified L communities characteristics (species richness per station, % = i 100 i L density, biomass) were tested using Kruskal-Wallis test and t multiple comparison between treatment (Siegel and where% = percentage of substrate category i, L = total i i Castellan, 1988). length of category i, L = total length of transect (50 m). t Relationships between ichthyofauna and substrate char- each species was attributed a diet having nine possible acteristics were quantified using multiple regression analy- components: fish, macro-invertebrates (> 2mm), micro-in- sis with a stepwise procedure because of possible intercor- vertebrates (≤ 2 mm), zooplankton, other plankton, macro- relations among the predictor variables (Sokal and Rohlf, algae, micro-algae assemblage, coral and detritus. the 1995). using this method it was possible to rank the contri- micro-algae category was extended to micro-algae assem- butions of the different substrate categories. this procedure blage because grazers consume micro-algae and also organ- was applied to the total species richness and log-transformed ic detritus that accumulate within micro-algae communities. densities and biomasses to fit a normal distribution. it was Consequently, grazers viewed traditionally as herbivores also applied to log-transformed densities of the characteris- are now clearly partly detritivorous (Choat, 1991; arias tics species of each fish assemblage identified using the Gonzales et al., 1997). it was not possible to identify the multivariate analyses. origin of the organic matter (micro-algae and detritus) con- sumed by grazers. however this confusion was not a major problem in the present study because both categories come results from the lowest trophic levels. diets were determined using stomach contents data (iRd database, noumea) and com- pleted by a bibliographical search (Kulbicki et al., 1994; in total, 193 fish species belonging to 33 families were Wantiez, 1994b; thollot, 1996; Letourneur et al., 1997b). a recorded during this study (appendix 1). Labridae (34 spe- given species may have several components in its diet and cies), Pomacentridae (33 species) and Chaetodontidae (23 so is included in several trophic categories. the weight con- species) were the most diversified families. Mean species tribution of a species to a trophic category was considered richness per station was 46.9 ± 4.3, mean density 2.53 ± according to the importance of the corresponding food item 0.72 fish/m2 and mean biomass was 43.0 ± 15.3 g/m2. in the diet of the species. this procedure is more precise Multiple regression analysis showed that these parame- than the usual attribution of a species to a single trophic ters were seldom correlated with substrate categories. category (Wantiez, 1994a). trophic networks were studied Species richness per station was negatively correlated only using biomass index, which is the best representation of with sand cover (r = -0.57, p < 0.05) and log-transformed energy flow when no productivity estimates are known density was positively correlated only with algae cover (all (Wantiez, 1994a). types) (r = 0.55, p < 0.05). no significant correlation Reef fish community structure was studied using vari- appeared between biomass and substrate categories. 86 Cybium 2003, 27(2) Wantiez & chauvet Structure and trophic network of Uvea reef fish community structure (see appendix 1). Species groups 1 and 2 were characteris- Four assemblages were identified using a correspond- tic of SB and Lt2 assemblages respectively. Both species ence analysis (Fig. 2): one outer slope community (OS), one groups 3 and 4 were characteristic of OS assemblage. lagoonal type 1 community (Lt1), one lagoonal type 2 com- Characteristic species of Lt1 assemblage were mainly munity (Lt2) and one seagrass bed community (SB). these found in species group 5 along with a few Lt2 species, but assemblages were validated by a cluster analysis of samples also in species group 1 and 4. the cluster analysis confirms scores, only Lt2 assemblage was constituted by two station the general trends of the correspondence analysis and groups (Fig. 3). the cluster was calculated on the first six underlines that Lt1 assemblage has intermediate character- axes extracted by the correspondence analysis, which istics between SB and OS assemblages. Comparison of extracted more than 70% of the total inertia. Characteristic substrate characteristics between OS, Lt1 and Lt2 showed species of these four assemblages were identified using the significant differences for algae and rocky substrate projection of the species (Fig. 2), which have absolute or (Kruskal-Wallis test, p < 0.05) (Fig. 5). relative contributions to the first two axes of the corre- the outer slope (OS) community was characterized by spondence analysis higher than 10%. these species catego- fish species associated with live corals (appendix 1): ries were compared with the results of a cluster of species Chaetodontidae, Pomacanthidae and Pomacentridae. Other scores on the first six axes extracted by the correspondence species common to outer slopes were also characteristic of analysis which classified the species in 5 groups (Fig. 4) this assemblage (appendix 1): Elagatis bipinnulata, 3 axis 2 (12.2%) Outer slope 2 45 1 35 25 14 34 24 23 axis 1 (17.6%) 43 0 44 13 -3 -2 -112 0 42 32 1 2 Lagoon type 1 -1 Lagoon type 2 Figure 2. - Projection on the stations (numbers) and the species (symbols) 22 -2 on the first two axes determined by the correspondence analysis. Only species -3 with absolute or relative contributions higher than 10% on one of the two first Seagrass bed axes are displayed. the names of the -4 species are given in appendix 1. St12 St44 Lt1 St24 St43 SB St22 St25 OS St35 St45 St13 St23 Figure 3. - Cluster analysis of the St42 samples scores (stations, St) on the Lt2 St32 first six axes extracted by correspon- dence analysis (Fig. 2) using Ward St14 aggregation method and (1-r) Pearson distance. OS: outer slope; Lt1: lagoon St34 type 1; Lt2: lagoon type 2; SB: sea- grass bed. Lt1 Cybium 2003, 27(2) 87 Structure and trophic network of Uvea reef fish Wantiez & chauvet Aphareus furca and Chaetodon ornatissimus. Species rich- cluster group 1) and OS assemblage (Chrysiptera taupou, ness per station, density and biomass of this assemblage Dascyllus reticulatus, Coris gaimard and Naso lituratus, were not significantly different from Lt1 and Lt2 (tab. i). included in cluster group 4) (appendix 1). this assemblage this assemblage occurred on stations characterized by a had a significantly lower density than Lt2 assemblage substrate colonized by live corals and coralline algae (tab. i). this assemblage was located on stations character- (Fig. 5). Log-transformed densities of the characteristic spe- ized by rocky and sandy substrates with dead corals (Fig. 5) cies of OS assemblage were negatively related to sand cover but no significant relationships were identified between log- and positively related to dead corals with algae and live cor- transformed densities of the characteristic species of Lt1 als cover (tab. ii). fish assemblage and substrate characteristics. the first lagoon assemblage (Lt1) was characterized by the second lagoon assemblage (Lt2) was characterized common lagoon species and herbivorous species by planktivorous species (appendix1): Spratelloides sp,. (appendix 1): Pomacentridae, Labridae, Scaridae and Pseudanthias spp. and Pomacentridae related to live corals acanthuridae. Some of the characteristic species are com- (Chromis viridis, Pomacentrus coelestis, Pomacentrus mon to SB assemblage (Halichoeres trimaculatus, Scarus pavo). Benthic carnivores are also characteristic of this sp., Acanthurus blochii and Zanclus cornutus, included in assemblage (Mulloidichthys flavolineatus, Monotaxis gran- 1 2 3 Figure 4. - Cluster analysis of the spe- cies scores on the first six axes extract- ed by correspondence analysis (Fig. 2) using Ward aggregation method and (1-r) Pearson distance. the species 4 included in the analysis have absolute or relative contributions to one of the first two axes higher than 10%. the 5 numbers are the species groups. the names of the species are given in appendix 1. Figure 5. - Substrate characteristics of the 4 assemblages identified using a correspondence analysis (Fig. 2). OS: outer slope; Lt1: lagoon type 1; Lt2: lagoon type 2; SB: seagrass bed. 88 Cybium 2003, 27(2) Wantiez & chauvet Structure and trophic network of Uvea reef fish table i. - Mean species richness (Sr per station), densities (fish/m2) and biomasses (g/m2) in the different fish assemblages identified in uvea, using a correspondence analysis (Fig. 2). values in parentheses are standard deviation; OS: outer slope; Lt1: lagoon type 1; Lt2: lagoon type 2; SG: seagrass bed (not included in the statistical test); difference: Kruskal-Wallis test and multiple comparisons between treatment (Siegel and Castellan, 1988) between OS, LT1 and LT2; NS: not significant (p > 0.05); *: significant (0.01 < p ≤ 0.05). OS LT1 LT2 Difference SB Sr per station 56.3 (2.52) 41.3 (9.00) 47.5 (5.47) NS 38 Density 1.82 (0.89) 1.43 (0.23) 3.69 (1.27) *(LT1 < LT2) 2.14 Biomass 53.6 (23.0) 44.4 (44.9) 42.5 (22.4) NS 8.5 table ii. - Stepwise multiple regression analyses between log (den- species related to soft bottom substrate colonized by sea- sity +1) of the characteristic species of the fish assemblages identi- grasses (Fig. 5) (appendix 1): Lethrinus spp., Scolopsis fied by correspondence analysis (Fig. 2) and substrate characteris- trilineatus, Parupeneus multifasciatus, Novaculichthys tics. r: correlation coefficient; *: significant (0.01 < p ≤ 0.05); **: highly significant (p ≤ 0.01) taeniorus, Rhinecanthus aculeatus and Arothron hispidus. this community was characterized by the lowest species Outer slope Lagoon type 2 richness per station and biomass but a medium-high density Substrate class Rank Partial r Rank Partial r (tab. i) because of the presence of numerous juveniles. it Live corals 3 0.58* was not possible to compare this assemblage with OS, Lt1 Algae 1 -0.79** and Lt2 using statistical tests because SB assemblage was Dead corals 2 0.70* 2 -0.62* present at only one station. no significant relationships Other living organisms were identified between log-transformed densities of char- Sand 1 -0.67 * acteristic species of OS fish assemblage and substrate char- Rubble acteristics. Rock Total r2 0.76** 0.64** trophic networks in uvea, the major energy intake came from lower levels doculis, Halichoeres sp., Thalassoma spp.). Some of the of the trophic network (micro-algae assemblage), grazers characteristic species were common in Lt1 assemblage and (45.1%) being the main trophic group (Fig. 6) with Scaridae included in cluster group 5 (Monotaxis grandoculis, (Scarus frenatus, Scarus ghobban and Chlorurus sordidus) Mulloidichthys vanicolensis, Pomacentrus pavo). this and acanthuridae (Acanthurus blochii, Acanthurus lineatus, assemblage had the highest density (tab. i). this commu- Acanthurus triostegus and Ctenochaetus striatus). the other nity was located at stations characterized by a sandy sub- important trophic categories, macro-carnivores (24.4%, strate with rubbles and a lower percentage of algae than Lt1 Caranx ignobilis, Lutjanus fulvus, Lutjanus kasmira and (Fig. 5). Log-transformed densities of the characteristic spe- Monotaxis grandoculis) and piscivores (14.5%, Caranx cies of Lt1 fish assemblage were negatively related to algae ignobilis, Lutjanus kasmira), consume energy from upper and dead corals with algae covers (tab. ii). trophic levels (Fig. 6). using these results, it was possible to the seagrass bed community (SB) was characterized by build a synthetic and schematic model of the trophic net- 1.2% 4.2% 14.5% 45.1% Piscivores Macro-carnivores Micro-carnivores zooplanktivores 24.4% Macro-herbivores Grazers Corallivores detritivorous 4.7% 5.0% Figure 6. - trophic categories of the coral reef fish community of uvea, 0.9% expressed in biomass. Cybium 2003, 27(2) 89 Structure and trophic network of Uvea reef fish Wantiez & chauvet Piscivores (14.5%) Macro-carnivores (24.4%) detritivores (1.2%) Micro-carnivores (4.7%) Corallivores (4.2%) zooplanktivores (5.0%) Micro-invertebrates Macro-invertebrates Coral Grazers Macro-herbivores (0.9%) (45.1%) Detritus Micro-algae Macro-algae Figure 7. - Schematic representation of trophic networks of the fish community in uvea, expressed in biomass. Boxes and arrows involv- ing fishes are in continuous lines and their size are relative to the percentage of overall biomass involved. dotted boxes and arrows involve non-quantified trophic links of organisms other than fish. works of coral reef fish communities in uvea (Fig. 7). this munities in uvea coral reefs (Fig. 7). One other important model presents the different energy transfer paths used by energy path was macro-invertebrates → macro-carnivores the ichthyofauna. the micro-algae → grazers → piscivores → piscivores. Micro–invertebrates, coral and zooplankton path was the most important energy path for reef fish com- were involved in three minor energy paths: coral → coral- Figure 8. - trophic network of the different coral reef fish assemblages identified by the correspondence analysis (Fig. 2), expressed in biomass. OS: outer slope; Lt1: lagoon type 1; Lt2: lagoon type 2; SB: seagrass bed. 90 Cybium 2003, 27(2) Wantiez & chauvet Structure and trophic network of Uvea reef fish livores → piscivores, micro-invertebrates → micro-carni- man landings showed that fish size of these and other com- vores → piscivores and zooplankton → zooplanktivores → mercial species was coherent with a sustainable fishing piscivores. activity (Chauvet, unpubl. data). Consequently, even if these no significant correlation occurred between trophic species are present in uvea, they must be rare. the most groups and substrate categories (p > 0.05) but variations in reasonable hypothesis would be uvea’s location in the mean characteristics of trophic networks occurred between Pacific Ocean. uvea is characterized by low diversity com- the different fish assemblages identified in uvea (Fig. 8). pared with the austro-Malayan maximum diversity area and the differences between OS, Lt1 and Lt2 are not statisti- compared with the closest islands (905 species recorded in cally significant (Kruskal-Wallis test, p > 0.05) with the Samoa, 822 species in Fidji; Kulbicki and Rivaton, 1997). exception of piscivores and grazers (p < 0.05). differences Consequently, uvea could be relatively isolated from the occurred in two major categories: grazers were more impor- possible colonization paths across the Pacific (Kulbicki and tant in OS and Lt1 assemblages, and piscivores were more Rivaton, 1997). thresher (1991) indicated that continental important in OS and Lt2 assemblages. SB trophic network margin reefs in the Pacific (including Great Barrier Reef, (one station) could not be statistically compared with the new Caledonia, Fiji), are rich in species of Serranidae, others. however, these networks were relatively close to whereas acanthuridae dominate consistently in Pacific Lt2 networks, with fewer piscivores and more macro-her- Central oceanic reefs (including Guam, Samoa islands, bivores and micro-carnivores. the two main energy paths tahiti). the latest observations gathered in uvea in november remain the same for all four assemblages but two organiza- 2000 indicated that 590 fish species were identified in the tions can be identified. Micro-algae → grazers → piscivores lagoon and the outer reef slope in the 0-25 m depth zone is the more important path for OS and Lt1 assemblages, (Wantiez and Williams, unpubl. data). the species composi- macro-invertebrates → macro-carnivores → piscivores is tion reflected a mixture of indo-Pacific, western Pacific and the more important path for SB assemblage, with both Pacific Plate fauna. this suggests that uvea is situated at a energy paths having a similar importance for Lt2 assem- faunal junction between the western Pacific and the Pacific blage (Fig. 8). variations exist for the other trophic catego- Plate biogeographic regions, as described by Springer ries but they are not statistically significant. (1982), which is consistent with its geographic location. Species richness per station, density and biomass of uvea coral reef fish communities are within the range of the discussion values reported in new Caledonia using visual censuses (Kulbicki et al., 1996; Kulbicki, 1997; Grimaud and species richness, density and biomass Kulbicki, 1998), the Chesterfield islands (Kulbicki et al., the fish species counted during this study are common 1990), Great Barrier Reef (Williams and hatcher, 1983), to healthy coral reefs (Randall et al., 1990; Lieske and and French Polynesia (Galzin, 1987). however, density and Myers, 1994). in 1980, 330 species and 55 families were biomass in uvea are among the lowest values. therefore, recorded in a qualitative study (Richard et al., 1992), which uvea coral reef fish communities have global characteris- included coral reefs and soft bottoms of uvea, Futuna and tics consistent with other populations in the Pacific region alofi islands. unfortunately, it was not indicated where the but they are among the poorest ones. different visual census fish were counted. a total of 69% of the species counted dur- techniques may give different estimations but the order of ing our study was recorded by Richard et al. (1992) and only magnitude of thee estimations should remain the same one of the families (Pseudochromidae) was not recorded by (Kulbicki and Sarramégna, 1999) Richard et al. (1992). Several common species, which are widely distributed on Pacific coral reefs, were not recorded community structure and relationships with substrate in uvea during the present study, in particular several com- characteristics mercial species: Serranidae (Epinephelus spp., Plectropomus uvea fish communities were structured along a coast to leopardus), Lehtinidae (Gymnocranius spp., Lethrinus nebu- outer slope gradient, with one coastal seagrass bed commu- losus), Labridae (Bodianus perditio), acanthuridae (Naso nity, two lagoons communities and one outer slope com- spp.), Siganidae (Siganus spp.), among others. this particu- munity. Such distance offshore organization is classical for larity could be explained by the limited sampling effort to coral reef fish communities of the Great Barrier Reef, complete this inventory but not by the sampling method. Polynesian atolls and several Pacific high islands (Williams however, the habitats where these species are commonly and hatcher, 1983; Galzin, 1987; Wantiez et al., 1997). in found were well sampled. another hypothesis could be an australia, mid-shelf reefs supported higher diversity, den- overfishing of these commercial species. these species are sity and biomass than inshore and outer reefs, along a 200 rare in the commercial catches, but a quick analysis of fisher- km coast to ocean transect across the Great Barrier Reef Cybium 2003, 27(2) 91 Structure and trophic network of Uvea reef fish Wantiez & chauvet (Willams, 1982; William and hatcher, 1983). Such a pattern trophic networks was not observed on a smaller spatial scale in uvea lagoon, the trophic networks of uvea coral reef fish populations with the exception of a higher density for one lagoonal display common characteristics and one particularity com- assemblage (Lt2). pared to similar communities sampled using visual census. terrestrial runoff, which enriches coastal waters, and the major energy paths involve the lower levels of trophic oceanic influences (high energy and oligotrophic waters) networks, grazers (mainly Scaridae and acanthuridae) rep- have been identified as major structuring factors (Letourneur resenting 45.1% of the biomass. the two other important energy sources come from higher trophic levels, macro-in- et al., 1998a). these opposed influences are closely related vertebrates (macro-carnivores = 24.4%, mainly Caran- to substrate characteristics (Letourneur and Chabanet, 1994; gidae, Lutjanidae and Lethrinidae) and fish (piscivores = McGehee, 1994). in the present study, live corals, algae, 14.5%, mainly Carangidae and Lutjanidae). Such organiza- dead corals with algae, rocks or sand were significant struc- tion was observed in the SW lagoon of new Caledonia and turing factors. Fish communities are strongly influenced by Ouvea atoll (new Caledonia), where grazers represented habitat structure and complexity (Friedlander and Parrish, 30-37% of the biomass, macro-carnivores 22-31% and pis- 1998; Ohman and Rajasuriya, 1998), with density and bio- civores 10-18% (Kulbicki, 1997). Piscivores (11%) and mass being higher in complex habitats (Friedlander and grazers (23%) were less important in the Chesterfield Parrish, 1998). this pattern cannot be totally applied to the islands where they were substituted by macro-herbivores uvea lagoon. Lt2 assemblage characterized by a sandy (25%) (Kulbicki, 1997). the importance of grazers in uvea substrate with rubbles had the highest density because of can be explained by the importance of hard substrate (rocks, planktivore schools. Only biomass was higher, but not sta- dead corals and consolidated coralline substrate) represent- tistically different from the other groups, in the most com- ing more than 50% of substrate cover. this substrate is eas- plex habitat (OS) where live and dead corals with algae and ily colonized by turf and filamentous algae, which are the coralline algae represented more than 90% of substrate main primary producers on reefs (Polunin, 1996). Macro- cover. Relationships between fish communities and live carnivores and piscivores classes are generally linked coral cover have been widely studied, but fewer studies because numerous species (i.e., Caranx ignobilis, Lutjanus have examined the relationships between fish communities fulvus and L. kasmira) contribute to the two categories. and other habitat variables (Johns and Syms, 1998). Macro-carnivores are opportunistic species consuming Contrasting patterns were found, ranging from numerous macro-invertebrates and fish and strictly piscivorous spe- significant correlations to no relationship (Johns and Syms, cies are rare (Wantiez, 1994a), none of them were observed 1998). the intensity of the relationships between fish com- during this study. munities and their habitats depends on the diversity of the the particularity of uvea trophic networks comes from habitats studied (Jenning et al., 1996), which could partly the limited presence of zooplanktivores (5% of the biomass, range 1-13%), mainly Caesionidae (Caesio caeruleata) and explain these differences. Live coral cover is usually posi- Pomacentridae (Chromis viridis, Pomacentrus coelestis, tively correlated with diversity and abundance of fish (Johns P. pavo). this pattern is unusual for coral reef fish commu- and Syms, 1998), such a pattern was observed for the char- nities. Similar results have only been reported in tuamotu acteristic species of the OS assemblage. algae cover was atolls (French Polynesia) using visual census, where plank- among the significant factors, being positively correlated tivores represented 7 to 15% of the biomass (Kulbicki, with herbivores (hart et al., 1996). Similar positive correla- unpubl. data). Planktivores are more important in the SW tion occurred for the overall fish density in the present lagoon of new Caledonia (9-33%; Kulbicki, 1992), the study. Chesterfield islands (12-31%; Kulbicki, 1992) and Reunion Consequently, uvea coral reef fish communities appear island (22-55%; Letourneur, 1998) where visual census typical of these communities elsewhere in the world. the methods were used. they are also more important in the specificity of this island is probably its location at the bor- Great Barrier Reef (43-71%; Williams and hatcher, 1983) der of different biogeographic regions. these characteristics where sampling was done using explosives. it appears that are the result of several biological processes such as larvae the limited presence of zooplanktivores in uvea cannot not availability, habitat selection at settlement, differential sur- be linked to a lack of plankton, the biomass of plankton vivorship after settlement, food and shelter availability, being 0.74 g aFdW/100 m3 in the lagoon during this survey regional current patterns, reef size, lagoon morphology and (Wantiez, unpubl. data). the tide can have a significant location (Williams and hatcher, 1984; Grimaud and effect on plankton availability for lagoon fish. the renewal Kulbicki, 1998; Letourneur et al., 1998a; Ohman and of approximately one third of lagoon waters at each tide Rajasuriya, 1998). should export a significant part of plankton production to 92 Cybium 2003, 27(2)