Patterns of Cleaner Wrasse Density Among Three Regions of the Pacific

MARINE ECOLOGY PROGRESS SERIES Published February 11 Mar Ecol Prog Ser

Patterns of cleaner wrasse density among three regions of the Pacific

Celine Arnalll*, Serge Morandl, Michel ~ulbicki~

'Centre de Biologie et dtEcologie Tropicale et Mediterraneenne, Laboratoire de Biologie Animale (UMR 5555 CNRS), Universite de Perpignan, F-66860 Perpignan Cedex, France 'ORSTOM B.P. A5. Noumea, New Caledonia

ABSTRACT: Despite years of studies, cleaning interactions between cleaner and host fishes are still not well understood. Relating density of cleaner wrasse with fish species richness, density and biomass across different geographic localities may help to better understand cleaner/host fish relationships. In this study, we explored the patterns of variation in cleaner wrasse densities among 3 reglons of the Pacific. An increasing gradient of cleaner wrasse density was observed from Ouvea (West Pacific) to Tuamotu (Central Pacific). This increasing density followed a decreasing gradient in the density of other reef fishes. Within each locality, densities of cleaner wrasse were positively correlated with densities of reef fishes. In order to better understand the underlying influences on the density of cleaner wrasse, the densities of reef fishes were examined in relationship to fish body size, fish sedentary, territorial and mobile behavior, and gregariousness. We showed that there is no relationship between the density of cleaner wrasse and host fish sedentary, territorial or mobile behavior, but that there are positive relationships with both host fish body size and host gregariousness. We hypothesize that the positive relationships are closely linked with fish ectoparasite loads.

KEY WORDS: Cleaner fish . Labrojdes spp. . Biogeography. Indo-Pacific . Fish density . Fish body size . Fish gregariousness

INTRODUCTION cleaner wrasses belonging to the Labridae family are quite common (van Tassel et al. 1994). Cleaning symbioses have been widely described in Surprisingly, although many studies have investi- the literature (Feder 1966, Youngbluth 1968, Hobson gated the nature of cleaning symbioses, few efforts 1971, Losey 1972, 1974, 1979, 1993, Gorlick et al. 1978, have been made to understand interactions between Itzkowitz 1979, Poulin 1993, Grutter 1995, 1996a,b, cleaner fishes and other members of the fish commu- Losey et al. 1995, Arnal 1996, Poulin & Grutter 1996, nity (Arnal & C6te 1998). Moreover, factors that may Grutter & Poulin 1998a). This relationship involves a influence cleaner fish densities are poorly known. cleaner fish, which removes ectoparasites, diseased Cleaning stations were, at first, described as determin- tissues and mucus from the surface of a client fish. ing the distribution and density of fish populations Although this relationship has been described as a (Slobodkin & Fishelson 1974). Johnson & Ruben (1988) textbook example of mutualism, the cost and benefit emphasized that the density and diversity of reef fishes for cleaners and hosts are not well understood (Losey plus the availability of suitable substrates promote the et al. 1995, Poulln & Grutter 1996, Arnal & CBte 1998, density and diversity of cleaners, rather than the Grutter & Poulin 1998a). Cleaning interactions occur reverse. Therefore, increased knowledge of the geo- usually at cleaning stations where host fishes visit a graphical variation in cleaner fish densities may open settled cleaner fish. Many fish species have been new avenues toward better understanding of cleaner/ described as cleaners; among these species, the host fish relationships. In the Pacific, fish species richness, density and biomass show variations between biotopes (Kulbicki 1997).

O Inter-Research 1999 Resale of fullarticle not permitted 214 Mar Ecol ProgSer l??:213-220, 1999

Comparing cleaner wrasse density with these 3 factors are given respectively by Harmelin-Vivien et al. in different geographic localities can help to answer (1997), Matoto et al. (1997) and Kulbicki (1997).Within several questions. Firstly, is the density of cleaner each region sampling areas were designated accord- wrasse subject to geographical variation? Secondly, ing to the prominent type of reef present. Within a cer- within a region, considering that reef fishes are poten- tain reef type the sites were separated- by a minimum tial hosts for the cleaners, does the density of cleaner of 1 nautical mile and a maximum of 3; their precise tvrasse vary with reef fish density, number of species or locations were chosen at random. At each site, under- biomass, which should represent the food available for water visual surveys were performed in transects. The cleaners? Finally, does cleaner density vary with fish number of transects per site was 4 in Tuamotu, 1 in densities according to the body size, the sedentary, ter- Tonga Tapu and 2 in Ouvea. In each transect all fish ritorial or mobile behaviors, or the gregariousness of were recorded visually (see Letourneur et al. 1997 for fishes? details of the method). Fish body sizes may affect cleaner wrasse density. Fish density (number of fish ha-'), species richness Several studies have shown that ectoparasite load is (number of species transect-') and biomass (g ha-') often positively correlated with fish body size (Poulin were estimated for each station at each locality (Burn- 1993, Grutter 1995, Poulin & Rohde 1997, Sasal et al. ham et al. 1980, Buckland et al. 1993). In brief, a 50 m 1997, Sasal & Morand 1998). Therefore, we hypothe- transect line was first laid on the bottom. Then a diver size that cleaner wrasses should prefer sites or locali- swam along this transect and recorded all the fish he ties where large fishes occur. observed. For each observation, the diver noted the Reef fish sedentary, territorial and mobile behaviors species, the size of the fish and its perpendicular dis- may have an effect on cleaner fish density. Studies on tance from the transect line. Fish size was estimated in Caribbean cleaner fishes (Thalassoma bifasciatum and 1 cm class intervals for fish less than 10 cm in length, Elacathinus sp.) revealed that territorial fishes, by 2 cm class intervals for fish between 10 and 30 cm, and chasing reef fishes, appear to have a negative influ- 5 cm class intervals for fish longer than 30 cm. Dis- ence on cleaner fish activity when they are settled tances from the transect line were noted in 1 m inter- close to each other (Itzkowitz 1979, Arnal 1996, Arnal vals for up to 5 m, then in 2 m intervals for distances & C6te 1998). Mobile fishes could also be less attrac- between 5 and 10 m; fish beyond 10 m were recorded tive than sedentary fishes for cleaner fish, as they in 5 m intervals. Densities were estimated using dis- would not represent a faithful clientele. Moreover, as tance-sampling algorithms (Buckland et al. 1993) pro- parasite load, generally, is high in sedentary hosts vided in the program DISTANCE (Laake et al. 1994). (C6te & Poulin 1995), cleaner fishes should prefer Biomasses were calculated by using the same algo- sedentary client fishes. rithms applied to weights instead of numbers. To Finally, fish gregariousness is known to be positively obtain weight, length-weight relationships were used correlated with parasite species richness (C6te & from ORSTOM's data base FISHEYE (Labrosse et al. Poulin 1995, Sasal & Morand 1998); solitary fishes are 1997). less parasitized than fishes living in large groups. Fish densities were also estimated in relationship to Therefore, solitary fishes could be less attractive for fish body size, host sedentary, territorial and mobile cleaner wrasses than are gregarious fishes. behavior, and gregariousness. Fishes were divided Our aim in the present study was to test for relation- into 3 size classes: small fishes (c7cm), medium fishes ships between cleaner wrasse density and reef fish (7 to 30 cm), and large fishes (>30 cm). Territorial fish communities within and among regions. We were also are those which defend a territory (i.e. many Pomacen- looking for possible interactions with fish body size, tridae) or which roam over a fixed territory (i.e. most fish sedentary, territorial or mobile behavior, and gre- Chaetodontidae). The term sedentary was used to des- gariousness of fishes. ignate fish that stay within a limited range, usually not leaving their reef for long periods (e.g. most Ser- ranidae). Species were considered mobile if they MATERIALS AND METHODS moved over large expanses of reef; this category in- cluded fishes which do not roam from one reef to The cleaner wrasse genus Labroides (Labridae) was another, if reefs are separated by large expanses of soft investigated. Four speci.es of cleaners were chosen, L. sediment or deep water (e.g. Scandae, many Lut- dimidiatus, L. bicolor, L. pectoralis and L, rubrolabia- janidae or Lethrinidae), but also fishes which cover tus (van Tassel et al. 1994). The study was carried out great distances and are not attached to a particular in 3 different localities: Tuamotu (n = 130),Tonga Tapu reef (e.g.most Carangidae, some Lethrinidae). Finally, (n = 45) and Ouvea (n = 105), where n is the number of 3 different fish group sizes were used to distinguish sites at each locality. The descriptions of these regions gregariousness: solitary or paired fishes (1 or 2 fishes), Arnal et al.: Cleaner wrasse density in the Pacific 215

fishes living in small groups (from 3 to 20 fishes), and Between the 3 localities, only reef fish densities var- fishes living in large groups (more than 20 fishes). ied significantly (Kruskal-Wallis test: H[2,282] = 13.52, All analyses were performed using non-parametric p = 0.001). There were no differences in species rich- statistics since our data did not meet the assumption of ness or fish biomass between Tuamotu, Tonga Tapu parametric testing. Kruskal-Wallis l-way ANOVA by and Ouvea reefs (Kruskal-Wallis tests). Fish densities ranks was used for comparisons between the 3 locali- differed between Tuamotu and Ouvea (Tukey HSD ties. A Bonferroni correction adjusted the level of sig- tests: p = 0.026), with a greater mean (-+ SE) number of nificance from a = 0.05 to a = 0.01. Tukey unequal sam- fish at Ouvea (30922 + 482 fishes ha-') than at ple size HSD tests were used to detect differences Tuamotu (19331 + 113 fishes ha-') (Fig. 1). No differ- between localities (a = 0.05). Associations within each ence in reef fish density between Ouvea and Tonga locality were determined using Spearman's rank cor- Tapu (31082 2 293 fishes ha-', Tukey HSD tests) was relation coefficient (a = 0.05). Because of the relatively recorded. small number of stations visited in Tonga Tapu (n = 45), all results involving this locality should be considered with caution. Variations in reef fish densities in relationship to their body size, sedentary, territorial and mobile behavior, and gregariousness among the three localities RESULTS Densities of small fish differed between Tuamotu, Variations in cleaner wrasse densities, and reef fish Tonga Tapu and Ouvea (Kruskal-Wallis test: H [2,282] densities, species richness and biomass among the = 8.76, p = 0.012). Medium and large fish densities three localities were also different between the 3 localities (Kruskal- Wallis test: all p < 0.001). When using Tukey HSD tests, There was a significant difference in densities of only the density of medium sized fish showed signifi- cleaner wrasse among Tuamotu, Tonga Tapu and cant differences between Tuamotu, Tonga Tapu and Ouvea reefs (Kruskal-Wallis test: H[2,282] = 35.47, p < Ouvea. There were more medium sized fishes in 0.0001). Cleaner wrasse densities (mean + SE) differed Tonga Tapu than in Tuamotu (Tukey HSD tests: p = significantly between Tuamotu (125 _t 15 cleaners ha-') 0.009) (Table 1). Moreover, there was no difference and Ouvea (35 5 8 cleaners ha-') (Tukey HSD tests: p < between medium fish density in Ouvea and Tonga 0.001), and between Tonga Tapu (123 * 36 cleaners Tapu (Tukey HSD tests). ha-') and Ouvea (Tukey HSD tests: p = 0.005) (Fig. 1). The densities of mobile fish in Ouvea, Tonga Tapu The highest densities occurred at Tuamotu and Tonga and Tuamotu were similar (Kruskal-Wallis test: H[2, Tapu; at these 2 locations the mean number of cleaners 2821 = 1.22, p = 0.54). Densities of territorial and did not differ significantly (Tukey HSD test). The 10~7- sedentary fish showed a significant difference among est mean density was observed at Ouvea. the 3 localities (Kruskal-Wallis test: all p < 0.001). However, these differences in densities were con- firmed only for sedentary fish by Tukey HSD tests. 40000 200 Q m CD There were more sedentary fishes at Ouv6a than at L Reef fishes a,3 2 Tuamotu. No difference was found between Ouvea and Tonga Tapu for sedentary fish densities (Tukey -150 11,5 V) HSD tests) (Table 1). V) m The densities of fish from large size groups were the 3 (D same in Ouvea, Tonga Tapu and Tuamotu (Kruskal- -100 5 Wallis test: p = 0.64). However, Tukey HSD tests a showed a significant difference between Tuamotu and CD3 I?. Ouvea (p = 0,005). Densities of solitary or paired fish U" -50 and fish from small groups differed at the 3 localities -=!! (Kruskal-Wallis test: all p < 0.0001) (Table 1). Tukey 5V) U HSD tests showed that more solitary or paired fishes e were found in Tonga Tapu than in Ouvea and Tuamotu -0 S UY (Tukey HSD tests: all p 0.0001); no differences were OUVEA TONGA TUAMOTU found between Ouvea and Tuamotu (Tukey HSD tests: Fig. 1. Mean densities of reef fishes and cleaner wrasse (t p = 0.99). Finally, fish from small sized groups showed standard error) at Ouvea (105 sites), Tonga Tapu (45 sites) significant differences in their densities at the 3 locali- and Tuamotu (130 sites) ties (Tukey HSD tests: all p < 0.006). 216 Mar Ecol Prog Ser 177: 213-220, 1999

Table 1 Mean densities (+ standard error) of reef fishes accord~ngto their body size (small, medium and large sized fishes), sedentary, territorial or mobile behaviors, and gregariousness (solitary or paired fishes, fishes from small and large sized groups) at the 3 locations: Ouvea (105 sites), Tonga Tapu (45 sites) and Tuamotu (130 sites). (NS: no significant difference between the 3 localities. S: significant difference between the 3 locations shown by Kruskal-Wallis test or Tukey HSD tests only; SS: significant difference between the 3 locations shown by Kruskal-Wallis and Tukey HSD tests)

Ouvea Tonga Tapu Tuamotu Significance I l Body size Small 16688 * 3683 13922 r 2046 9659 * 776 Medium 12830 * 1527 15619 * 1372 8858 * 583 Large 549 * 66 255 t 52 706 * 80 Behavior Territorial Sedentary Mobile Gregariousness 1 or 2 fishes 2379 * 196 6224 + 489 2341 + 152 Small group 4 107 + 353 10042 & 915 6489 * 588 Large group 23161 * 4570 12287 * 2087 10330 * 703

Variations in cleaner wrasse densities and fish DISCUSSION densities, species richness and biomass within reefs Similar biotopes were found in the 3 regions studied. Densities of cleaner wrasse were significantly corre- However, one should be aware of some differences in lated with reef fish densities in Tuarnotu (r, = 0.473, p < habitat and in sampling design between these 3 re- 0.0001), Tonga Tapu (r, = 0.412, p = 0.003) and Ouvea gions. In Ouvea, the atoll has a larger percentage of (r, = 0.412, p i 0.0001). Densities of cleaner wrasse were hard substrate (Kulbicki 1997) than Tonga Tapu or the correlated with densities of reef fish according to their atolls of Tuamotu. Ouvea, like Tonga Tapu, has many size (small, medium and large fish) and their gre- passes, and the oceanic influence is probably much gariousness (solitary, small group size and large group greater than in the atolls of Tuamotu. The size of the size) in Tuamotu and Ouvea (Spearman's rank corre- atolls also plays a role. In Tuamotu the atolls studied lation coefficient: all significant) (Table 2). Cleaner ranged from 1 to 315 km2,whereas Tonga Tapu covers cvrasse density in Tonga Tapu was correlated with the density of large sized fish. When we looked at gregariousness, only solitary or paired Table 2. Correlations (Spearman's rank correlation coefficient, r,) fish were significantly correlated with densities between cleaner wrasse densities and reef fish densities according to of cleaner wrasse at the 3 locations, ~i~hfrom their body size (small, medium and large sized fishes), sedentary, territorial or mobile behaviors, and gregariousness (solitary or paired and large sized were not flshes, fishes from small and large groups) at the 3 locations: Ouvea (n with densities of =leaner at Tonga = 105), Tonga Tapu (n = 45) and Tuarnotu (n = 130). Probability values (Table 2). The densities of cleaner wrasse were are given in parentheses significantly correlated with the density of sedentary fishes at the 3 locations. Cleaner Ouvka Tonga Tapu Tuamotu wrasse density was not correlated with densities of mobile fishes or territorial fishes at Tonga Body size Tapu, but these correlations were significant Small 0.24 (<0.0001) 0.35 (0.014) 0.40 (

approximately 400 km2 and Ouvea 900 km2. Galzin et This may in part explain the importance of the cleaner al. (1994) have demonstrated that fish diversity is a wrasses in Tuamotu compared to Tonga Tapu and function of atoll size. Therefore, diversity should be Ouvea, which are both much further west. higher in Ouvea than in Tonga Tapu or Tuamotu. This Johnson & Ruben (1988) also noted the dependence was indeed observed, but not at a significant level. of cleaner fish density on the concentration of potential Another important difference is the amount of fishing hosts. The present study clearly indicates that cleaner going on at a certain area. Tonga Tapu supports wrasse densities are correlated with overall reef fish approximately 150 persons km-2 of lagoon, Ouvea, 3 densities, within each locality, but that the magnitude persons km-', and Tuamotu approximately 0.5 persons of this relationship depends on the region. However, km-'. Jennings & Polunin (1995) found that fishing cleaner wrasse density may be inore influenced by the pressure was an important factor determining the quality than by the number of potential hosts. As structure of fish assemblages on small islands in the cleaner wrasses primarily feed on ectoparasites from Fijis. In particular, large piscivores and macrocarni- the host's body (Youngbluth 1968, Grutter 1995, vores (Serranidae, Lethrinidae and Lutjanidae) were 1996a,b, 1997b,c, Grutter & Poulin 1998b), most of the affected. observations should be based on the relationships The depth range in Tonga Tapu was from 0.5 to between cleaners and ectoparasite populations, which 10 m whereas in Ouvea and Tuamotu the range stud- depend directly on the host fish propensity to be para- ied was 0.5 to 20 m. This may be responsible for sitized (Gorlick et al. 1978). The relationship between some of the observed differences in the structures of fish density and ectoparasite loads has never been the fish assemblages. Indeed, the abundance of her- studied. bivores tends to decrease with depth, whereas plank- Ectoparasite community richness is primarily deter- tivores show the opposite trend. Herbivores will also mined by water temperature (Poulin & Rohde 1997) increase with hard substrate cover (Letourneur et al. and follows a latitudinal gradient. The temperatures in 1997). these 3 regions are not well documented. However, in An increasing gradient of cleaner wrasse density Ouvea the range is between 19 and 29°C with strong was observed from Ouvea to Tuamotu, whereas there oceanic input, whereas in the shallower atolls of was a decreasing gradient in the density of other reef Tuamotu the water temperature ranges from 22 to fish. Within each locality, densities of cleaner wi-asse 32°C and there is much less oceanic influence. Tonga were positively correlated with densities of reef fishes. Tapu is probably intermediate. Unfortunately, we have Removal experiments on cleaners showed that cleaner no data on ectoparasite densities, nor on ectoparasite wrasse do not seem to have an influence on the abun- species richness, at the 3 localities. dance of reef fishes (Grutter 1997a), but Johnson & In order to better understand the influences on the Ruben (1988) demonstrated that fish densities may density of cleaner wrasse, the densities of reef fishes influence the occurrence of cleaner fishes. were considered in relationship to fish body size, fish These contrary results, i.e. a positive relationship be- sedentary, territorial and mobile behaviors, and gre- tween cleaner densities and fish densities within local- gariousness. In the following, we will only discuss pat- ities and a negative relationship between localities, terns showing a significant relationship with cleaner may be explained in many ways. For instance, the wrasse density (Table 2). availability of suitable substrates promotes cleaner In animal populations, sedentary hosts seem to be wrasse density on a reef (Johnson & Ruben 1988), more heavily parasitized than more mobile hosts (CBte hence, the observed differences between the 3 locali- & Poulin 1995). Therefore, we hypothesized that the ties may reflect a difference in substratum between densities of cleaner wrasses should be correlated with Ouvea, Tonga Tapu and Tuamotu. One of the striking those of sedentary hosts, as such hosts would provide features of the atolls in Tuamotu is the patchy distribu- more food for cleaners. Moreover, cleaner wrasses tion of reefs, compared with the continuous reefs should prefer reefs where sedentary fishes occur, as observed in Ouvea and to a lesser extent Tonga Tapu. such fish species may represent a more faithful clien- Consequently, the patchiness of the reefs at Tuamotu tele. However, cleaner wrasses occurred in larger may help to separate competing fish and therefore numbers on reefs where there were relatively few induce higher densities of cleaner wrasses. sedentary fishes. Cleaner wrasses live in sheltered Geographical variation may also have an influence habitats (Lenke 1991); therefore, the competition for on cleaner wrasse distribution. Reef fish diversity refuges, especially when more sedentary fishes occur, decreases from the West to the Central Pacific. How- may have a negative influence on cleaner wrasse den- ever, some families show the opposite trend, in partic- sity. Moreover, cleaner wrasses may not use fish ular the Labridae, which have a higher relative diver- sedentary behavior as an indication of the ectoparasitic sity in the Central Pacific (Kulbicki & Rivaton 1997). load of their clients. 218 Mar Ecol Prog Ser 177: 213-220, 1999

Territorial fish density was higher in Tonga Tapu were high. This result may be explained by the posi- than in Ouvea and Tuamotu. On Tonga Tapu, cleaner tive correlation between parasite density and gregari- wrasse density was also high. Although the relation- ousness found, by several authors (CBte & Poulin 1995, ship between the density of cleaner wrasses and the Sasal & Morand 1998). This relationship, however, was density of territorial fishes was not significant at this not found for fish from large groups. There were more location, the 2 factors tend to be correlated. However, fish from large groups in Ouvea than in Tuamotu. Our although cleaning by Caribbean cleaner gobies is results seem to show a preference of cleaner wrasse for influenced by fish territoriality (Arnal & CBte 1998) on gregarious fish limited to small sized groups. Mooring a local scale, it has never been observed that fish terri- & Hart (1992) and Loehle (1995) argued that, in mam- toriality influences cleaner density on a larger scale. mal populations, larger host group size should benefit On the contrary, Itzkowitz (1979) observed the forma- from a lower parasite infestation level. If this argument tion of large groups of Caribbean cleaner wrasses is true for fish populations, then fishes living in small Thalassoma bifasciatum as a strategy against the terri- groups should be more parasitized than those found in toriality of damselfishes. This could provide an expla- larger groups, and cleaner wrasses should prefer reefs nation for the large numbers of both cleaner wrasses supporting large densities of fish from small sized and territorial fishes at Tonga Tapu. Moreover, territo- groups. rial fish usually occur in areas with dense cover (Bell & Moreover, fish forming large schools may not be a Galzin 1984). This cover may also be used by cleaner good target for cleaner wrasses for 2 reasons. First, wrasses and thus explain, in part, the relationship some of the fish forming large schools remain high in between territorial fish and cleaners at Tonga Tapu. the water column and are constantly moving. There- Cleaner wrasse density is significantly higher at fore, they are mostly likely inaccessible to cleaner Tuamotu than at Ouvea although the density of territo- wrasses. Secondly, most fish living in large schools are rial fishes is low at both locations. Thus, we were short-lived. Their densities are therefore quite variable unable to detect a specific influence of fish territoriality over time, and cleaner wrasses, which have relatively on cleaner wrasse density. long lifespans compared to this fluctuating resource, Variations in cleaner wrasse densities between may not benefit reliably from species forming large Ouvea, Tonga Tapu and Tuamotu according to reef schools. fish sedentary behavior can not be explained clearly. Our results showed that cleaner wrasse densities Contrary to our hypothesis, cleaner wrasses did not vary among different localities according to the com- occur in large numbers in areas where sedentary fish petition level for shelters among fishes, but also density was high and territorial fish density low. This according to the quality of fish hosts. Several studies may reflect a strategy the cleaners use to avoid compe- emphasized that cleaner wrasses preferentially clean tition with sedentary fishes for shelter. The relationship client fishes harboring large numbers of ectoparasites between fish territoriality and cleaner wrasse density (Poulin 1993, Grutter 1995). Accordingly, we found remains unclear; in some cases, it may cause an aggre- that cleaner tvrasses were more abundant on reefs gation of cleaner wrasse. inhabited by high densities of larger fish and fish living It is known that fish body size is positively correlated in small groups (which are presumably more heavily with parasite loads (Grutter 1995, Poulin & Rohde parasitized). Moreover, the cleaner wrasse can either 1997, Sasat & Morand 1998). Therefore, we predicted be a facultative cleaner (feeding on ectoparasites, positive correlations between high densities of clean- mucus from the host's body surface and on benthic ers and densities of large fishes. The variation in fish crustaceans) or a very specialized cleaner (feeding densities according to body size followed diverse exclusively on ectoparasites from the host's body sur- trends among the reefs. The number of medium sized face), depending on the food available and, therefore, fish increased from Tuamotu to Ouvea. Small fish den- the ectoparasitic loads of reef fishes (Grutter 1997b,c). sity seemed to vary in the same way, but not signifi- Thus, we suggest that variations in densities of cleaner cantly. The variation in the number of large fish fol- wrasse are closely linked to variations in ectoparasite lowed an inverse gradient. There were more large fish populations. at Tuamotu than at Ouvea. In accordance with one of Recent studies demonstrated that the evolution of our hypotheses, cleaner wrasses were more numerous the association between cleaner wrasses and their where larger fish occurred. Thus, on a regional scale, clients varies among client genera independent of densities of cleaner wrasse follow the increase of fish client size (Grutter & Poulin 199813, C6te et al. in press). body size. A phylogenetic control could prove conclusive. Unfor- Fish from small groups were less numerous in Ouvea tunately, our dataset did not allow us to control for phy- than in Tuamotu. High densities of cleaner wrasses logeny, as only 1 variable was available (there was were observed where densities of fish in small groups only 1 value of density of cleaner wrasse at each site Arnal et al.. Cleaner urrasse density in the Pacific 219

corresponding to other fish densities). Taking this into Grutter AS (1996b) Experimental demonstration of no effect account, host body size may not be a good predictor of by the cleaner wrasse Labroides dimidiatus on the host cleaner wrasse host preference. However, our results fish Pomacentrus moluccensis. J Exp Mar Biol Ecol 196: 285-298 on host fish gregariousness should open new perspec- Grutter AS (1997a) Effect of the removal of cleaner fish on the tives in the study of the cleanedhost fish relationship, abundance and species composition of reef fish. Oecologia as most recent studies considered host body size the 111:137-143 only explanation for host preference by cleaner fishes Grutter AS (1997b) Spatiotemporal variation and feeding selectivity in the diet of the cleaner fish Labroides dimidia- (Poulin 1993, Grutter 1995, Arnal 1996, Poulin & Grut- tus. Copeia 1997 (2):346-355 ter 1996). Grutter AS (1997~)Size-selective predation by the cleaner In this study, we have assumed that cleaner wrasse fish Labroides dimidiatus. J Fish Biol50:1303-1308 densities increase with ectoparasite availability, and, Grutter AS, Poulin R (1998a) Cleaning on coral reef fishes by thus, we assumed that cleaning activity is higher in the wrasse Labroides dimjdiatus: influence of client body size and phylogeny. Copeia 1998 (1):120-127 such localities. This assumption however was not Grutter AS, Poulin R (1998b) Intraspecific and interspecific tested. Future studies investigating the cleaning relationships between host size and the abundance of par- behavior of cleaner wrasses should investigate asitic larval gnathiid isopods on coral reef flshes. Mar Ecol whether the cleaning efficiency of wrasses is a function Prog Ser 164:263-271 Harmelin-Vivien M, Kulbicki M, Galzin R. Mou Tham G of their density. Furthermore, future studies should (1997) Reef fish community structure within atoll lagoons take into account composition and abundance of in French Polynesia (Tuamotu Archipelago). In: Proceed- ectoparasites, as cleaning activity may also be influ- dings 5th Indo-Pacific Fish Conference, Noumea, New enced by these factors (Grutter 1995, 1996a,b, 1997b). Caledonia Hobson ES (1971) Cleaning symbiosis among California inshore fishes. Fish Bull US 69(3):491-523 Acknowdedgements. We thank A. Grutter for helpful com- Itzkowitz M (1979) The feeding strategies of a facultative ments on the n~anuscript. cleaner fish: Tl~alassoma bifasciatum (Plsces: Labridae). J Zoo1 187:403-413 Jennlngs S, Polunin NVC (1995) Comparative size and composition of yield from six Fijian reef fisheries J Fish Biol LITERATURE CITED 46:28-46 Johnson W, Ruben P (1988) Cleaning behaviour of Bobianus Arnal C (1996) Interactions between cleaning gobies and ter- rufus, Thalassoma bifasciatum, Gobiosoma evelynae and ritorial damselfishes on coral reefs. MSc thesis, University Periclimenespedersoni along a reef depth gradient at Salt of East Anglia, Norwich River Submarine Canyon, St Croix. Environ Biol Fish 23: Arnal C, C6te IM (1998) Interactions between cleaning gobies 225-232 and territorial damselfish on coral reefs. Anirn Behav 55: Kulbicki M (1997) Bilan de 10 ans de recherche (1985-1995) 1429-1442 par I'ORSTOM sur la diversite, la densite, la biomasse et la Bell JD, Galzin R (1984) Influence of live coral cover on coral- structure trophique des communautes de poissons lago- reef fish communities. Mar Ecol Prog Ser 15:265-274 naires et recifaux en Nouvelle-Caledonie. Cybium 21: Buckland ST, Anderson DR, Burnham KP, Laake JL (1993) 47-79 Distance sampling-estimating abundance of biological Kulbicki M, Rivaton J (1997) Inventaire des poissons lago- populations. Chapman & Hall, London naires et recifaux de Nouvelle-Caledonie. Cybium 21(1) Burnham KP, Anderson DR, Laake JL (1980) Estimation of (Suppl):81-98 density from Line transect sampling of biological popula- Laake JL, Buckland ST, Anderson DR, Burnham KP (1994) tions. Wildlife Monogr 32:l-202 Distance sampling. Abundance estunation of biological C6te IM, Pouhn R (1995) Parasitism and group size in social populations. DlSTANCE user's guide. Version 2.1 Col- animals: a meta-analysis. Behav Ecol6:159-165 orado Cooperative Fish & Wildlife Research Unit, Col- C6te IM, Arnal C. Reynolds JD (in press) Variation among fish orado State University, Fort Collins species in posing behaviour at cleaning stations. J Fish Labrosse P, Letourneur Y, Kulbicki M, Magron F (1997) FISH- Biol EYE a new database on the biology and ecology of Feder HM (1966) Cleaning symbiosis in the marine environ- lagoonal and reef fishes of the South Pacific. In: Proced- ment. In: Henry SM (ed) Symbioses. Academic Press, New ings 5th Indo-Pacific Fish Conference Noumea, New York, p 335-377 Caledonia Galzin R. Planes S. Dufour V, Salvat B (1994) Variation in Lenke R (1991) The opercular gland of the cleaner-wrasse diversity of coral reef fish between French Polynesian Labroides dimidiatus (Labridae), a light, electron and atolls. Coral Reefs 13.175-180 scanning electron microscopic investivation. J Fish Biol Gorlick DL, Atkins PD, Losey GS (1978) Cleaning stations as 39:383-392 water holes, garbage dumps and sites for the evolution of Letourneur Y,Kulbicki M, Labrosse P (1997) Spatial structure reciprocal altruism. Am Nat 112:341-353 of commercial reef fish communities along a terrestrial Grutter AS (1995) Relationship between cleanlng rates and gradient in the Northern Lagoon of New Caledonia. Envi- ectoparasite loads in coral reef fishes. Mar Ecol Prog Ser ron Biol Fish 51:141-159 118:51-58 Loehle C (1995) Social bariiers to pathogen transnlission in Grutter AS (1996a) Parasite removal rates by the cleaner wild animal population. Ecology 76(2) 326-335 wrasse Labroides dimidiatus. Mar Ecol Prog Ser 130: Losey GS (1972) The ecolog~cal importance of cleaning sym- 61-70 biosis. Copeia 1972:820-833 220 iMar Ecol Prog Ser 177: 213-220, 1999

Losey GR (29741 Cleaning symbiosis in Puerto Rico with com- Poulin R, Grutter AS (1996) Cleaning symbioses: proximate parison to the tropical Pacific. Copeia 1974.960-970 and adaptative explanations. BioScience 463512-517 Losey GS (1979) Fish cleaning symbiosis: proximate causes of Poulin R,Rohde K (1997) Comparing the richness of metazoan host behaviour. Anim Behav 27:669-685 ectoparasite communities of marine fish: controlhg for Losey GS (1993) Knowledge of proximate causes aids our host phylogeny. Oecologia 110:278-283 understanding of function and evolutionary history. Mar Sasal P, Morand S 11998) Comparative analysis: a tool for Behav Physiol23:175-176 studying monogenean ecology and evolution. Int J Para- Losey GS, Mahon JL, Danilowcz BS (1995)Innate recognition sit01 28:1637-1644 by host f~shof their cleaning symblont. Ethology 100 Sasal P, Morand S, Guegan JF (1997) Determinants of parasite 227-283 specles richness in Mediterranean marine fish. Mar Ecol Matoto S, Ledua LL, Mou Tham G, Kulbicki M, Dalzell P Prog Ser 149:61-71 (1997) The aquarium fish fishery in Tonga Tapu Tapu- Slobodkin LB, Fishelson L (1974) The effect of the cleaner fish Tonga Tapu. Status and recommendations for manage- Labroides dimidiatus on the point diversity of fishes on the ment. Country Assignment Report, South Pacific Com- reef front at Eilat. Am Nat 108:369-376 mission van Tassel JL, Brito A, Bortone SA (1994) Cleaning behav~our Mooring MS, Hart BL (1992) Animal grouping for protection among marine fishes and invertebrates in the Canary from parasites: selfish and encounter-dilution effect. Islands. Cybium 18:117-127 Behaviour 123:173-193 Youngbluth MJ (1968) Aspect of the ecology and ethology of Poulin R (1993) A cleaner perspective on cleaning symbiosis. the cleaning fish, Labroides phtirophagus, Randall. Rev Fish Biol Fish 3:75-79 Z Tierpsychol 25:915-932

Editorial responsibility; Ofto Kinne (Editor), Submitted: April 24, 1998; Accepted: October 12, 1998 Oldendorf/Luhe, Germany Proofs received from author(s): January 22, 1999