BULLETIN OF MARINE SCIENCE, 63(3): 605–610, 1998 NOTES THE WRASSE HALICHOERES CYANOCEPHALUS (LABRIDAE) AS A SPECIALIZED CLEANER FISH Ivan Sazima, Rodrigo L. Moura and João Luiz Gasparini Cleaning symbiosis in fishes involves feeding by microcarnivores, the cleaners, on ectoparasites and other material from the body surface of cooperating hosts, the clients (Losey, 1987). Cleaning wrasses (Labridae) are well known components of reef commu- nities and range from highly specialized cleaners of the genus Labroides (Youngbluth, 1968; Potts, 1973) to less specialized cleaners in the genera Bodianus, Semicossyphus, and Thalassoma, which clean mostly as juveniles (Hobson, 1968, 1974; Itzkowitz, 1979; Johnson and Ruben, 1988; Van Tassell et al. 1994). The genus Halichoeres is the most species-rich within the family Labridae (Nelson, 1994). These wrasses are highly mobile, diurnal feeders and their foraging methods in- clude scanning and picking small prey, following other fishes, and even use of anvil rocks (Hobson, 1968; Wainwright, 1988; Coyer, 1995). Although the feeding habits of wrasses in general are regarded as well suited for cleaning (Hobson, 1976; Losey, 1987), we found reference to cleaning activities for only three species of Halichoeres, the Pacific H. semicincta (Hobson, 1976) and H. nicholsi (McCourt and Thomson, 1984), and the West- ern Atlantic H. poeyi (Szpilman, 1991). Here we report on cleaning activities of the yellowcheek wrasse, Halichoeres cyanocephalus in the western South Atlantic. This distinctive, bright yellow and blue wrasse occurs in deeper water than do most other tropical West Atlantic species of Halichoeres, and its habits are little known (Randall and Böhlke, 1965; Humann, 1994). We observed that feeding activities of juvenile H. cyanocephalus are centered on defined cleaning stations, as in specialized cleaners of the genera Labroides and Symphodus (Potts, 1973; Moosleitner, 1980), but different from other species of Halichoeres for which clean- ing is recorded. MATERIAL AND METHODS Field work was conducted on rocky reefs at Escalvada Island (20°42'S, 40°24'W), off the coast of Espírito Santo, southeastern Brazil, in January and May 1997, and on coral reefs at Tamandaré (08°22'S, 35°05'W), off the coast of Pernambuco, northeastern Brazil in March 1997. At the Escalvada site, the bottom is sandy with scattered patches of coarse gravel and occasional isolated rocks or slabs. The rocky substrate is sparsely to thickly colonized by red and brown algae, and anthozoans including the gorgonians Phyllogorgia dilatata and Plexaurella sp. Depth ranged from 6 to15 m and visibility averaged 10 m. At the Tamandaré site, the bottom is sandy with calcareous patch reefs with rubble at the margins. The reefs are colonized by brown and red algae, and antho- zoans including colonies of the large star coral Montastraea cavernosa and the porous coral Porites spp. Depth ranged from 27 to 35 m and visibility averaged 18 m. Four days of field work were spent at the Escalvada site and three days at the Tamandaré site, totaling 21 h of scuba diving. Cleaning activities were observed directly, video-recorded and photo- graphed during daylight hours. Besides general observations on the behavior of the cleaners and their clients (Hobson, 1971; Moosleitner, 1980), we focused on the following questions: Do juve- 605 606 BULLETIN OF MARINE SCIENCE, VOL. 63, NO. 3, 1998 nile H. cyanocephalus clean their clients at defined cleaning stations? What client species do seek these stations? What are the feeding habits and size of these clients? We centered the behavioral observations on cleaning events, i.e., the period of association between a single cleaner and a single host beginning with physical contact and ending when the cleaner or the client retreats (see Johnson and Ruben, 1988 for definitions and procedures). During observational sessions of 15 to 50 min we used “focal animal” and “all occurrences” samplings (Altmann, 1974), totaling 410 min of direct observation (375 min at Escalvada and 35 min at Tamandaré). Total length of each individual cleaner or client was estimated against a rule placed on or nearby the cleaning station, the estimates being initially checked for accuracy against collected specimens. One cleaner individual was collected at each site and deposited in the fish collection of the Museu de História Natural, Universidade Estadual de Campinas (ZUEC 3151 and 3184). Nine recorded cleaning events are on file in the ZUEC video-tape collection (tape # 3). Classification of feeding modes of fish clients on Table 1 are based on field observations and literature data (Randall, 1967). RESULTS AND DISCUSSION Cleaning was recorded for 10 juveniles of Halichoeres cyanocephalus (total length ranging from 5 to 8 cm), each of them tending a defined cleaning station. At the Escalvada site, four stations were recorded in depths ranging from 10 to 16 m and one at 6 m, all of them centered on small isolated rocks on a bottom of sand and gravel (Fig. 1). At the Tamandaré site, five stations were recorded in depths ranging from 29 to 33 m at the margins of the patch reefs; all but one of these included a portion of hard substrate and marginal rubble (one station was centered on a reef groove). At both sites, the cleaners moved in a radius of about 50 to100 cm around their cleaning stations. Although defined cleaning stations are known for several cleaning labrids in the genera Labroides, Symphodus (Crenilabrus), and Thalassoma (Potts, 1973; Hobson, 1974; Moosleitner, 1980; Snelson et al., 1990), no such stations were previously recorded for species of Halichoeres. Hobson (1976) reports that an individual of H. semicincta meandered over a relatively large area and approached unsoliciting client fishes as far as 5 m from the cleaner. We have recorded similar cleaning behavior in two other Halichoeres species, H. poeyi and H. brasiliensis, which also seem to wander over a relatively large area, and approach and clean posing as well as non-posing clients (unpubl. observ.). At a cleaning station, H. cyanocephalus swam in a characteristic teetering fashion most of the time. The cleaner would swim forward slowly, or would remain almost stationary, repeatedly raising and dropping its body in the water column. The extent of these up and down movements ranged from 2.5 to 3.5 cm for a 5 cm long individual. The seesaw swimming rendered the yellowcheek wrasse very conspicuous at its station and, along with its vivid yellow and blue pattern, could act as an additional visual guide for the clients. Potts (1973) comments on so called “cleaner dance” performed by Labroides dimidiatus in the presence of some preferred clients or a large fish, and speculates that this swimming pattern has an appeasing function. Besides the presumed function as a visual cue for prospective fish clients, the seesaw swimming of H. cyanocephalus may have a similar appeasing function even if the recorded clients are not especially large or piscivorous. When picking planktonic organisms in the water column or when chased by other fishes, the yellowcheek wrasse moved quickly in the usual labrid swimming style. Six species of client fishes were recorded on two selected cleaning stations at the Escalvada site in about five hours of observation (Table 1). We have no comparable quan- titative data for the Tamandaré site but we recorded six species of clients on three stations: NOTES 607 Figure 1. A juvenile Halichoeres cyanocephalus cleaning a Chromis multilineata client at a cleaning station centered on three small rocks at Escalvada Island, off southeastern Brazil (based on a color photograph). Scale bar equals 10 cm. the haemulids Anisotremus virginicus and Haemulon sp., the pomacanthid Holacanthus tricolor, the labrid Bodianus rufus and the scarid Scarus aff. taeniopterus, as well as Pseudupeneus maculatus. All of the observed client species of H. cyanocephalus are diurnal, medium-sized reef fishes and include grazing herbivores, digging or picking predators on invertebrates (invertivores), omnivores, and plankton-pickers. This restricted assemblage of clients is known for two other Western Atlantic labrid cleaners, Bodianus rufus and Thalassoma bifasciatum (Darcy et al., 1974; Johnson and Ruben, 1988; Wicksten, 1995). A much more diverse assemblage of clients is recorded for the highly specialized Pacific cleaning wrasse Labroides phthirophagus and the Western Atlantic cleaning goby Elacatinus evelynae, which also clean large reef piscivores such as groupers and morays (Youngbluth, 1968; Johnson and Ruben, 1988). At the Escalvada site, the brown chromis, Chromis multilineata, was the most common client of the yellowcheek wrasse (Table 1). In order to be cleaned, a brown chromis would abruptly descend 4 to 6 m from the water column and hover close to the bottom, posing in a horizontal (Fig. 1) or slightly oblique, head-up, posture. Other fish clients also solicited cleaning service while hovering close to the wrasse and assumed characteristic poses (see Losey, 1971 and Moosleitner, 1980 for classification of client poses). The single indi- vidual of B. rufus was cleaned while swimming very slowly. Posing is not necessary to initiate cleaning but seems to increase the efficiency of the cleaner (Losey, 1971). Seven timed cleaning events involving brown chromis ranged from 12 to 30 s. Cleaning events recorded by other investigators for several cleaner species lasted from one second to sev- eral minutes and may be related to the cleaner’s preference for certain client species,