Transactions of the Royal Society of S. Aust. (2005), 129(2), 193-201.

1 Senior Research Fellow, South Australian Research andDevelopment Institute, PO Box 120 Henley Beach, SouthAustralia 5022. Email: shepherd.scoresby@saugov.sa.gov.au

2 Plant Biodiversity Centre, Hackney Rd, Hackney, South Australia5069.

3 9 Giles St., Glenelg, South Australia 5045.

CLEANING SYMBIOSIS AMONG INSHORE FISHES AT ALTHORPE ISLAND,SOUTH AUSTRALIA AND ELSEWHERE

by S. A. SHEPHERD1, J. TEALE2 & D. MUIRHEAD3

Introduction

Cleaning symbiosis among fishes is a commonmutualism among tropical fishes in which cleanerfish benefit by feeding on ectoparasites on clientspecies, which in turn benefit by their removal(Losey et al. 1999). Information on the similartemperate symbiosis is scarce, and in southernAustralia is available only from the accounts ofHutchins (1979, 1991a) in southwestern Australia,and brief references elsewhere (e.g. Hutchins 1991b;Edgar 1997; Hutchins & Swainston 1999; Morrison& Storrie 1999; Kuiter 2000). However, studies ofcleaning symbiosis in other temperate waters of theworld are more numerous e.g. the northern andsouthern Atlantic (Henriques & Almada 1997;Zander et al. 1999; Sazima et al. 2000), theMediterranean (reviews of Moosleitner 1980 andZander et al. 1999; Zander & Nieder 1997; Zander &Sötje 2002), the Black Sea (Darkov & Mochek1980), the Gulf of California (McCourt & Thomson1984), California (Hobson 1971), and New Zealand(Ayling & Grace 1971).

The prevalence of such cleaning behaviour amonglabrids in particular has recently generated muchinterest in their use as a benign method of reducingparasite loads on fish farms (Young 1996).Ectoparasites are a problem in southern Australianfish farms, and observations on the symbiosis, andthe cleaner and host species involved, will bevaluable in assessing the merit of such biologicalcontrol.

In this paper we describe aspects of the ecology ofcleaning behaviour from accumulated observations(a) at West I., South Australia (35° 37� S, 138° 35� E)over 15 years, (b) during the 2004 expedition and alater visit to Althorpe I. (34° 22� S; 136° 52� E), (c)at Cape Jervis (35° 35� S; 138° 6� E), and (d) underRapid Bay jetty (35° 32.8� S; 138° 12.4� E) withmany casual observations elsewhere.

Methods

During long-term underwater studies of abalone atWest I. (Shepherd 1998), visits were made there bySAS at 1-2 month intervals for 1-2 days from 1983-2000, and less often after that. Fish cleaningbehaviour was first observed in Dec. 1985 in AbaloneCove at 4 m depth, near an abalone monitoring site, atthe junction of a vertical granite wall and a seagrassbed of Amphibolis antarctica. Thereafter, the cleaningstation was monitored for evidence of the behaviour

Summary

SHEPHERD, S. A., TEALE, J. & MUIRHEAD, D. (2005). Cleaning symbiosis among inshore fishes at Althorpe Island,South Australia and elsewhere. Trans. R. Soc. S. Aust. 129(2), 193-201, 30 November, 2005.

Host-cleaner symbiosis among fish, not previously reported for South Australia, was investigated at AlthorpeI., West I., Cape Jervis and other sites over short to long periods. Five cleaner fish species, one crustaceancleaner species, and 23 client fish species are recorded. The cleaning symbiosis is described in some detail fortwo cleaners, the moonlighter, Tilodon sexfasciatus, and the pencil weed whiting, Siphonognathus beddomei, andin less detail for three other cleaners, the western cleaner clingfish, Cochleoceps bicolor, the black-spottedwrasse, Austrolabrus maculatus, and the old wife, Enoplosus armatus. Cleaning behaviour appears to befacultative, and its timing dependent on cleaner behaviour. At West I. it was restricted to a few occasions per dayfrom late morning to late afternoon during summer months at a fixed cleaner station, whereas at Althorpe I. andCape Jervis cleaner stations were often not fixed, and cleaning took place opportunistically throughout the day.A third cleaner, the western cleaner clingfish, Cochleoceps bicolor, appears to clean mainly deeper waterspecies. Communicative and contextual signals between client and cleaner are described. They include: guildsigns by the cleaner; posing by the client, in which the body is inclined at various angles; and fixed cleanerstations defined variously by topography, or presence of sponges. At Althorpe I. zebra fishes that posed at anangle head up received more attention from cleaners than those that posed horizontally, and those that posed insmall host groups received more attention than those in larger groups.

KEY WORDS: Fish cleaning; cleaning symbiosis; posing behaviour; fish behaviour; ectoparasites; southernAustralia; cleaner; Tilodon sexfasciatus, Siphonognathus beddomei, Cochleoceps bicolor, Austrolabrusmaculatus, Enoplosus armatus.

S. A. SHEPHERD, J. TEALE & D. MUIRHEAD194

up to eight times per visit at different times of dayuntil June 2001 when the study terminated.

Concurrently with observations of the behaviour ofthe blue groper, Achoerodus gouldii (Shepherd &Brook 2005), and the moonlighter, Tilodonsexfasciatus, fish cleaning by the latter species wasmonitored in Mooring Bay, Althorpe I. near shore ata depth of 3 m among granitic blocks and sheetgranite. The site was monitored several times perhour between 0900 and 1700 h from 2-11 Februaryand 9-13 April, 2004 and from 7-11 February 2005.In 2005 fish cleaning was recorded mainly duringconcurrent time budget studies on the moonlighter.

At Cape Jervis cleaning was observed at asheltered site 250 m north of Cape Jervis lighthouseat about 3 m depth among reef of high relief withnumerous caves, in which five moonlighters resided.At the Rapid Bay jetty site, throughout its >300 mlength, the array of vertical piles and the mass of oldpiles, concrete cladding and debris strewn on thebottom, together formed a complex reef habitatwhere three cleaner species, the western cleanerclingfish, Cochleoceps bicolor, the black-spottedwrasse, Austrolabrus maculatus, and themoonlighter, T. sexfasciatus, were common.Cleaning occurred at depths of 3 – 10 m at stationsadjacent to piles, among debris or near sponge.

For each cleaning event observed, data recordedwere: (a) the specific identity, number and size of thehost or client fish and cleaner fish present; (b) theduration of the behaviour, including anticipatoryhovering or slow swimming in the vicinity; (c) poseof the client species; and (d) the number of bitesobserved for each fish. In 2004 and 2005 at AlthorpeI. we noted more carefully the angle of pose of theclient, and categorised poses as vertical (head up),inclined (with head up at ~ 40 – 60° from horizontal),horizontal, or nose down (at an angle of ~ 20 – 40°from horizontal). In the horizontal category it wasnot always clear whether a fish was a posing client,or a hovering onlooker, and in cases of doubt it wasincluded in the latter category.

From observations on the frequency of sightingsof cleaning events at West I., an estimate of theexpected number of cleaning events there on anyday was derived as follows (see Dixon et al. 2005).Let x be the expected mean number of dailycleaning events during time period tD (in minutes).If s is the total number of sightings of fish cleaningon n inspections, and if the effective observationtime by the observer watching the cleaning stationis t0 min. each during the period tD, then it can beshown that:

s/n.t0 = x/tD

i.e. x = s.tD / n.t0

The standard error of s, a binomial estimatedquantity, is given by:

�s.(n-s)/n

Hence, from this relation, and knowing thevariance of t0, the standard error of x can becalculated.

Note that for each observed cleaning event lasting eminutes, if the average diver observation time is 3min. (see below) then t0 = e+ 6, because a positivesighting will result if the diver observationscommence up to 3 min. before and after time period e.

Results

Five cleaner fish species, one cleaner shrimp and23 client species, are recorded from South Australia(Table 1). The posing position of the client fish variedwithin and between species, and also according to thecleaner species. Typically the client fish laymotionless, dorsal side uppermost, and with finserect or extended. Some client species posed inclinedwith head upward at various angles (e.g. Fig. 5),while other species posed mainly horizontally (Figs1-4). In the presence of a western cleaner clingfish ora pencil weed whiting, the client fish often opened itsmouth and gills (Fig. 2). A summary of fish cleaningobservations at various sites, together with data onthe species involved, the duration of cleaning events,size of client groups and number of bites per fish, isgiven in Table 1, and considered in more detail belowfor the main sites investigated.

West IslandAt West I. fish cleaning occurred at the same station

at the junction of Amphibolis antarctica seagrass bedsand a rock wall during the 15-year period, but was notobserved elsewhere despite frequent searches.Normally a small school of zebra fish arrived at thecleaning station, between seagrass and a vertical wall,and hovered in the vicinity. Client fish variouslyposed horizontally, inclined or vertically, while othersswam slowly or hovered nearby. Cleaners thenemerged from the seagrass and swam about a host,searching for and apparently picking ectoparasitesfrom its flanks. On four occasions other client specieswere present, in addition to the zebra fish (Table 1). Ablue-throated wrasse occasionally paused nearby orwas pursued by a cleaner, but cleaning was onlyobserved once. On the five occasions when theinception of cleaning was observed, the client(s)initiated cleaning four times, by posing first andwaiting for cleaners to emerge, and on one occasionthe cleaner approached a hovering fish. All cleaningevents were recorded between December and Marchbetween 1100 and 1500 h between 1-5 h (mean 2.4 h;

CLEANING SYMBIOSES 195

Figs 1-6. 1. Blue devil, Paraplesiops meleagris, being cleaned on the left flank by a western cleaner clingfish, Cochleocepsbicolor. Seacliff Reef, 12 m depth. 2. Western blue groper, Achoerodus gouldii, posing for cleaning by a group of small,almost transparent, western cleaner clingfish, C. bicolor, on the client’s flank. Western River, Kangaroo I., 16 m depth.3. Spotted boarfish, Paristiopterus gallipavo, posing for cleaning by western cleaner clingfish, C. bicolor, which isapproaching the client near the mouth. Wirrina Reef, 22 m depth. 4. Harlequin fish, Othos dentex, being cleaned on theflank by a western cleaner clingfish, C. bicolor. Snug Cove, Kangaroo I., 14 m depth. 5. Magpie perch, Cheilodactylusnigripes, being cleaned on the flank by a western cleaner clingfish, C. bicolor. Note the angled pose of the client fish.Seacliff Reef, 12 m depth. 6. Rainbow cale, Odax acroptilus, posing for a pencil weed whiting, Siphonognathusbeddomei, approaching from the rear. Note the bright eye-spot of the cleaner. Moana Reef, 10 m depth. All photos D. M.

s.e. 0.4 h) after low water.In all there were 508 monitoring occasions

distributed across all months of the year (meannumber of occasions 42.3 (s.e. 4.8) per month), andfish cleaning was observed only on 16 occasions,between the months described, with sea

temperatures >16° C. We tested for independencebetween the number of cleaning events and thecumulative number of monthly monitoringoccasions for the 15-year period, and rejected thenull hypothesis of non-independence (χ2 = 26.6;P<0.001) i.e. there was a highly significant

111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111 2

3 4

5 6

S. A. SHEPHERD, J. TEALE & D. MUIRHEAD196

TABLE 1. Cleaner and host species with their respective length range (L), mean client abundance (A) per cleaning event,location of cleaning, and duration, D. N = number of cleaning episodes observed. Standard errors in brackets. Nd = nodata. All sites are in South Australia, except where indicated otherwise.

Cleaner species Client species N Location and notes

Pencil weed whitingSiphonognathus beddomei

resident in seagrass Amphibolis antarctica.

L= 5-8 cmA=1.6 (s.e. 0.1)

Pencil weed whitingSiphonognathus beddomei

resident in seagrassL=5-8 cm; A=1-2.

Moonlighter Tilodon sexfasciatus

resident in cavesL= 10-20 cm

A=1.2 (s.e. 0.1)

Moonlighter T. sexfasciatus

resident in cavesL= 8-17 cm; A=1.

Western cleaner clingfishCochleoceps bicolor

L= 3 cm; A=3.

Western cleaner clingfishC. bicolor L= 3 cm; A=1.Resident among sponge,

ascidians and rarely bryozoans

Zebra fish Girella zebraL=15-30 cm; A=4.1 (s.e. 0.4)

Magpie perch Cheilodactylus nigripesL=25 cm; A=1

Bluethroated wrasse Notolabrus tetricus L=20-25 cm; A=1

(1) Spiny-tailed leatherjacketAcanthaluteres brownii L=20 cm; A=1

Bluethroated wrasse N. tetricusL=15 cm; A=1

(2) Rainbow cale Odax acroptilusL=30 cm; A=1 (Fig. 6)

(3) Senator wrasse, Pictilabruslaticlavius A=1

Maori wrasse Ophthalmolepis lineolata A=1

Rainbow cale Odax acroptilus A=1Red mullet Upeneichthys vlamingii

A=1Zebra fish G. zebra

L=20-35 cm; A=2.5 (s.e. 0.5)Magpie perch C. nigripes

L=30-35 cm; A=1Bluethroated wrasse N. tetricus

L=35 cm; A=1Silver drummer Kyphosus sydneyanus

L=25 cmBlue groper Achoerodus gouldii

L=80 cmOld wife Enoplosus armatus L=20 cmMoonlighter T. sexfasciatum L=15 cm

Zebra fish G. zebraL=20-25 cm; A=3.6 (s.e.0.7)

Magpie perch C. nigripesL=22 cm; A=1

Blue groper A. gouldiiL=27-35 cm; A=1Dusky morwong

Dactylophora nigricansL=45 cm; A=1

Blue groper A. gouldiiL=30 cm; A=1 (e.g. Fig. 2)Harlequin fish Othos dentex

Magpie perch C. nigripesL=25-30 cm (Fig. 5)

Zebra fish G. zebra L=30 cm; A=1Harlequin fish O. dentexL=70 cm; A=1 (Fig. 4)Blue groper A. gouldii

L=30-80 cm; A=1 (Fig. 2)

16

2

2

1

1

211

1

11

51

4

8

1

1

115

1

3

1

1

16

32

3

Abalone Cove, West I. nearvertical wall and seagrass

D=31sec. (s.e. 3)Bites/fish 3.2 (s.e. 0.5)

(1)Rapid Head, (2) Moana reef,(3) Second Valley.

((3) A.Hirst pers. comm.)Poses generally horizontal, but

maori wrasse with inclined pose.

Mooring Bay, Althorpe I.D=50 sec. (s.e. 9).

Bites/fish – see Table 2.Poses of the zebra fish and

magpie perch varied (see text),but poses of remaining

species horizontal.

Cape Jervis, 300 m north oflighthouse, 4 m deep, near rock

wall and cave.D=53 sec. (s.e. 15)

Bites/fish 1.5 (s.e. 0.2)

Recherche Archipelago, W.A. Hutchins (1991a,b)

Morrison & Storrie (1999)Rapid Bay jetty, Western River,K.I., Snug Cove, K.I. Second

Valley, Moana reef, Aldinga drop-off, Hallett Cove reef, Seacliff

reef, Glenelg dredge site. Magpie perch and zebra fishoccasionally posed inclined

CLEANING SYMBIOSES 197

TABLE 1. Cleaner and host species with their respective length range (L), mean client abundance (A) per cleaning event,location of cleaning, and duration, D. N = number of cleaning episodes observed. Standard errors in brackets. Nd = nodata. All sites are in South Australia, except where indicated otherwise (cont.).

Cleaner species Client species N Location and notes

Black-spotted wrasseAustrolabrus maculatus

Old wife E. armatusL=20 cm; A=3

Cleaner shrimpPericlimenes aesopius

Dusky morwong D. nigricansL=50-80 cm; A=1

Boarfish Pentaceropsis recurvirostrisL=30 cm; A=1

Spotted boarfish Paristiopterus gallipavoL=35 cm; A=1 (Fig. 3)

Red mullet U. vlamingii L=25 cm; A=1Spiny tailed leatherjacket

Acanthaluteres brownii L=30 cm; A=1Scaly fin Parma victoriae

L=15 cm; A=1Blue devil Paraplesiops meleagris

L=25-30cm; A=1 (Fig. 1)Bullseye Pempheris multiradiata

L=15 cm; A=1 Bue groper A. gouldii

L=25- 35 cm; A=1Red mullet U. vlamingii L=25cm; A=5

Dusky morwong D. nigricansL=33 cm; A=1

Zebra fish G. zebra L=22 cm; A=4Blue groper A. gouldii L= 1.2 m; A=1

Leafy seadragon Phycodurus equesFam. Tetraodontidae

2

1

1

11

1

3

1

3

11

13

11

upward. All other species posedhorizontally.

Overall D=30-60 sec.

Rapid Bay jetty. All horizontalposes, except zebra fish (this

study and D. Cowan (pers.comm.))

Anchorage Cove, Pearson I.(Shepherd 2005) Horizontal poseVictor HarborPort Victoria

R.H. Kuiter (pers.comm.)

association between fish cleaning and the summerseason. Next we tested for independence betweenthe number of cleaning events and the cumulativenumber of monthly monitoring occasions during theabove four summer months, and accepted the nullhypothesis of non-independence (χ2 = 4.5; ns) i.e.the probability of cleaning occurring wasapproximately the same during each month fromDecember to March. Next we examined the data fortime of day at which cleaning took place. There were230 monitoring occasions at the cleaning stationbetween December and March, with a cumulativemean of 23.0 (s.e. 4.1) monitoring occasions foreach hour from 0800 to 1800 h, with cleaning eventsobserved only between 1100 and 1500 h. The nullhypothesis of non-independence of cleaning eventsand the number of observations per hour throughoutthe day was rejected (χ2 = 29.4; P<0.001), and weconcluded that fish cleaning was restricted to theperiod 1100 - 1500 h.

Lastly we calculated from the above formula thatthe expected number of cleaning events per day from

December to March inclusive was 3.2 (s.e. 0.2), onthe basis that: (1) n, the number of monitoringinspections, = 132, averaging 3 min. duration each(monitoring inspections were ~2 min. duration,except one in four of them, which was ~6 min. duringquadrat monitoring nearby, thus averaging 3 min.);(2) each cleaning event, including pre-cleaninghovering, lasted 3.1 (s.e. 0.3) min., derived from fivemeasured events, i.e. t0 = 9.1 (see above); and (3)both inspections and cleaning events occurredrandomly during the period 1100-1500 h.

Althorpe IslandThe behaviour of the cleaner moonlighter, Tilodon

sexfasciatus, differed dramatically between 2004 and2005 in Mooring Bay. In February 2004 threemoonlighters sheltered in a cave at the cleaningstation, and emerged from time to time to cleangroups of client species, which gathered at the caveentrance at various times of day. Eight of the ninecleaning events observed took place during theperiod 1500-1800 h, 4-6 h after low tide (mean time

S. A. SHEPHERD, J. TEALE & D. MUIRHEAD198

of cleaning 1558 h) and one at 1045 h at about lowtide. As previously, we tested for independencebetween monitoring observations and cleaningevents, and rejected the null hypothesis (χ2 = 12.6; P<0.005) of non-independence. We concluded thatcleaning behaviour strongly favoured mid- to lateafternoon. In April 2004 no cleaning took place,despite frequent observations over a week, and someposing by zebra fish at the cleaning station.

In 2005 several cleaner moonlighters were alsopresent in the same area, but sheltered in othercrevices nearby. These cleaners rarely cleaned at thecleaning station used in the previous year, butforaged individually or in pairs over a wide area, andwere approached or followed by client fishes insmall to large groups, seeking their attention.Cleaning took place wherever and whenever amoonlighter responded to a client, and was recordedthroughout the day when observations took place(i.e. 0900 - 1700 h), and up to 5 h before and afterlow tide.

Over the two years, 84% of the fishes cleaned, orposing for cleaning, were zebra fish, while six otherspecies were observed posing on rare occasions(Table 1). The zebra fish variously adopted all poses,although the head down pose was rare, occurring ononly 1.2% of occasions. The magpie perch mainly,and the silver drummer and blue groper always posedhorizontally, while the moonlighter posed nose down,and the blue-throated wrasse mostly horizontally, butsometimes inclined head up. Sea sweep, Scorpisaequipinnis, occasionally posed, but were neverobserved being cleaned, while the horseshoeleatherjacket, Meuschenia hippocrepis, on occasionsjoined a group of client species, but never posed, andsometimes seemingly disrupted the event. Cleanersmainly searched the flanks of a client, andoccasionally the pectoral fins, tail or head and mouth.On only two occasions (i.e. 2.8%) did we observecheating, when the cleaner fish aggressively bit afish, causing the client to jolt noticeably and flee.

Next we examined whether attention from acleaner depended on the size of the client group. Themean number of feeding bites per fish wassignificantly higher for group sizes of one and twocombined, compared with larger groups (t = 3.13; P< 0.002) (Table 2). The difference in mean number ofbites per fish between client groups of 1 - 2 fish and3 or more fish was significant (t = 3.13; P < 0.002).Moreover the proportion of fish receiving cleanerattention declined significantly (χ2 = 30.6; P<0.001)with increasing size of client group, i.e. from 87.5%for single fish down to 25 – 45.6% for groups of �4fish (Table 2). We then examined whether non-horizontal posing by a client resulted in morecleaning bites from a cleaner. Fish posing verticallyreceived a mean of 2.5 (s.e. 0.3) bites per fish, thoseposing inclined upward received 2.0 (s.e. 0.2), thosehead down received 1.8 (s.e. 0.8) while the remainderposing or hovering horizontally received 0.5 (s.e.0.3) bites per fish. The difference in mean bites perfish between those posing vertically and thoseinclined upward was not significant (t = 1.2; ns), butthe difference between those posing horizontally andthose non-horizontally was highly significant (t =5.9; P<0.001). Those adopting a head-down posewere too few to test statistically. From Table 2 it isevident that in general 50 – 80% of fish posedvertically or at an angle for host group sizes 1 – 4,but in larger groups the incidence of posing non-horizontally was less.

Cape JervisPeriodic visits were made to the cleaning station

over 15 months (Table 1), but cleaning was observedonly from December to February between 1300 and1730 h and 2-5 h after low tide. Several moonlightersapparently resided in a cave, near the entrance towhich fishes sometimes hovered and posed forcleaning. However, the cleaning station was veryloosely defined, as at Althorpe I. in 2005, andcleaners often made feeding excursions away from

TABLE 2. Althorpe I. 2004 and 2005. Mean number of cleaning bites per fish for different sized groups of host fishes (allspecies combined), mean percentage of fish in host group cleaned, and proportion of different posing types in each groupsize category. Group Size = number of host fishes present during a cleaning event, E = number of cleaning events, andN = total number of fish. Standard errors in brackets. For the four categories of pose,�= hovering vertically head up,� = inclined head up, � = hovering horizontally, and � = inclined head downward.

Group E N Bites/ % PoseSize fish cleaned � � � �

1 32 32 1.9 (0.3) 87.5 0.16 0.37 0.41 0.062 17 34 1.9 (0.3) 79.4 0.44 0.41 0.12 0.033 4 12 1.3 (0.5) 58.3 0.17 0.25 0.58 04 5 20 0.7 (0.4) 25 0.35 0.35 0.30 0

≥5 6 57 1.3 (0.2) 45.6 0.23 0.19 0.54 0.04

CLEANING SYMBIOSES 199

the cave, during which they cleaned zebra fish andother species that sought attention.

Rapid Bay jetty and elsewhereDuring the many visits to Rapid Bay jetty, cleaning

by the black-spotted wrasse and western cleanerclingfish has been observed mainly from latemorning to mid-afternoon. The western cleanerclingfish has been observed cleaning numerousspecies at many deeper reefs (10 – 20 m) in Gulf StVincent and Investigator Strait where erect, stalked,caliculate (goblet-shaped) or bilamellate sponges(e.g. Callyspongia bilaminata), and large solitaryascidians (e.g. Ascidia sydneiensis and Herdmaniamomus) occur. The clingfish appeared to residewithin the folds, lamellae or cup, as the case was, ofthe sponge, or near the ascidian, and establish acleaning station close by. Cleaning by the clingfishhas been observed in all months of the year, exceptJuly and August, but this could be due to fewer diverobservations in winter months. Client species andsites are listed in Table 1. As yet we have no data onbite rates, due, in the case of the clingfish, to thedifficulty of obtaining data in poor light conditions,and, in the case of the black-spotted wrasse, to thetendency of the wrasse to retreat in the presence of adiver.

Discussion

Most studies of cleaning in temperate regions notethat in these waters cleaning behaviour is incidental,transient or occasional. Except for Hobson’s (1971)study which noted a critical temperature of 12 – 13°C, below which cleaning was rare, this is the firststudy in temperate waters with evidence ofseasonality in the behaviour, presumably due to thelower incidence of ectoparasitism in colder water(Ayling & Grace 1971; Côté & Molloy 2003; but cfHenriques & Almada 1997). At West I. the mean seatemperature range is 13.5 °C (winter) to 21.5 °C(summer) (Shepherd & Womersley 1970), butcleaning was observed only from December withtemperatures >16 °C. However, if 16 °C were atemperature threshold sensu Hobson (1971),cleaning should occur until late May, when in fact itwas not observed after March.

Other factors, which inhibit cleaning, as noted byHobson (1971), include turbidity and surge. Theformer is unlikely to have biased our results at WestI., where water transparency was generally higher inwinter than summer (Shepherd & Womersley 1970),but the effect of surge or swell is not known becausemonitoring observations were strongly biased towarddays with low swell and better diving conditions.

From evidence that gentle tactile stimuli arerewarding to fish (Losey 1974; Thresher 1977),

Losey (1979) proposed that cleaning behaviourevolved independently in many parts of the worlddue to the tactile reward to the client. Cleanerprecursors exploited the positive response by clientsto such stimulation to gain ready access to a foodsupply. Communicative signals evolved to increasethe probability of the response. These included:adoption of specific poses by the client, and strikingcolouration (a guild sign – see Eibl-Eiblesfeldt 1955)by the cleaner. Contextual signals also evolved,including cleaning stations, ectoparasite loads on theclient, and prospective food for the cleaner. In thisstudy some signals were present and others absent, asmight be expected from a behaviour that is transientand occasional. Zebra fish, and occasionally magpieperch and blue-throated wrasse, but not otherspecies, adopted a characteristic (non-horizontal)pose, while other species adopted a horizontal posealbeit with other indicia such as erect fins or openedgills. Guild signs, in the form of conspicuoustransverse stripes, are conspicuous in themoonlighter, the western cleaner clingfish, and theold wife, but not the pencil weed whiting, which is auniform green colour. However, another possibleguild sign is the false eyespot, present in smallmoonlighters, old wives, black-spotted wrasse andfemale pencil weed whiting. The last species also hasvery conspicuous eyes (Fig. 6), which may serve thesame function.

In this study cleaner stations, when present, hadwell characterised topographies, close to rockoutcrops or vertical walls, or specific sponge orascidian habitats where cleaners variously resided.However, their existence, or persistence apparentlydepended on the idiosyncratic behaviour of thecleaner species. Some moonlighters had cleanerstations, whereas others did not. The pencil weedwhiting and western cleaner clingfish seemed tohave permanent stations. At Althorpe I. in 2004,cleaner moonlighters adopted a ‘stay-close-to-home’behaviour, and had fixed cleaner stations, whereas in2005 they regularly foraged up to 30 m from theirshelter holes and cleaned opportunistically. Theirreduced foraging range may have been induced byseal predators, which were commonly seen at the sitein 2004 but not in 2005 (cf Connell 2002).Conspicuous topographic features characteristic ofcleaning stations were noted by Darkov & Mochev(1980) in the Black Sea, and by Moosleitner (1980)and Zander & Sötje (2002) in the Mediterranean,whereas Ayling & Grace (1971) and Hobson (1971)found that cleaning stations at fixed sites were rare,or, if present, loosely defined.

Diurnal peaks of cleaning activity have beenpreviously recorded by Johnson & Ruben (1988) andSazima et al. (2000), who suggested that theycorresponded with post-feeding periods of host

variable posing behaviour within a species. In thisstudy the 2005 Althorpe I. data suggested a morecomplex picture for zebra fish. While an inclinedpose overall attracted greater cleaner attention, thenumber of fishes in a client group during a cleaningevent was also an important modifying factor. Forsmall client group sizes of one or two, the probabilityof cleaner attention was higher, and hence thebenefits of inclined posing for attention were less,and the incidence of such poses was relatively low.For client groups of five or more, only a minorityadopted an inclined posed, of which 96% receivedcleaner attention, consistent with the hypothesis thatthe rewards for inclined posing were high. Wehypothesize that inclined posing imposes anincreased risk of predation, especially at sitesfrequented by seals, and hence inclined posingmainly occurs when there is competition for cleanerattention and the rewards of ectoparasite removal aresubstantial. Our data at other sites are consistent withthis view, although of themselves inconclusive.

The low frequency (~3 – 4 events per day) and theseasonality of cleaning, as recorded at West I., andpossibly existing at Althorpe I. and Cape Jervis,indicate that in these temperate waters the cleanerspecies involved are facultative cleaners, obtainingonly a fraction of their food from cleaning activities.Cleaner species in other temperate seas are alsofacultative (Hobson 1971; Ayling & Grace 1971;Zander & Sötje 2002), and, according to the last-named authors, there may be a geographic cline fromtropical to boreal waters from professional tooccasional cleaners. However, the western cleanerclingfish may be a full-time cleaner, as at one site gutcontents were entirely parasitic crustaceans (J. B.Hutchins pers. comm.).

Hobson (1971) suggested that cleaning stationsdeveloped mainly at sites where fish tended toaggregate, usually in places of high habitat diversityi.e. with some vertical relief, an observation thataccords with our experience. If cleaner species orcleaner stations prove to be locally scarce in southernAustralia, and cleaning is a significant benefit to fishhealth as suggested by some authors (Losey et al.1999), then their presence may provide a strongargument for establishing reserves in their vicinity.Many of these and related questions, such as theectoparasitic loads on temperate fish, fish cleaningby deeper water species, the effectiveness of cleaning(Grutter 1996), and the adaptive value of ectoparasiteremoval in temperate waters have yet to be answered.

Acknowledgements

We thank R. K. Lewis and A. C. Cheshire forpromoting the expedition to Althorpe I., the crew ofRV Ngerin for logistic support for SAS, John and

species, when they had time to seek a cleaningstation. At West I. schools of zebra fish often arrivedat specific points in their ranges at about the sametime of day. This may have been influenced by thetidal cycle as cleaning was always recorded on anincoming tide. We hypothesize that peaks of cleaningactivity are influenced by the activity periods of theclient. It is curious that at both West I. and AlthorpeI. these temporal cleaning peaks occurred only atfixed cleaning stations, suggestive of a combinedtemporal/topographic cue.

Cleaning a client’s pharynx and gills by enteringthe mouth and opercular openings has been observedonly in the case of the western cleaner clingfish withseveral client species (e.g. blue groper, blue devil,dusky morwong, magpie perch, and harlequin fish).This is not surprising as this behaviour is onlypossible with a small cleaner and a relatively largefish (cf. Ayling & Grace 1971).

The 23 client species identified in this study (Table1) are all common in reef habitat, and are likely site-attached but with differing range sizes. If specieswith large ranges had access to more than onecleaner fish or station (termed choosy clients), thenmarket theory predicts that they should have priorityof access to a cleaner fish, compared with residentfish without such choice (Bshary & Noë 2003). AtAlthorpe I., Cape Jervis and West I., zebra fishseemed to include both residents (some wererecognizable by markings) and roving groups, but wecould not distinguish them with certainty. Hence wecould not tell whether choosy clients claimed priorityover residents. However, queuing for cleaner serviceoccurred when larger client groups were present, andwe never observed queue-jumping. The mostcommon cleaner in shallow waters, the moonlighter,had a mean density of 4.3 (s.e. 0.9) per 2000 m2 at 48fish survey sites in southern Gulf St Vincent andInvestigator Strait (SAS unpublished data), sopotential cleaners are often spaced apart at a scale oftens to hundreds of metres on coastal reefs. Hencewe hypothesize that these cleaners may be inrelatively short supply at many shallow coastal reefssites. Darkov & Mochek (1980) also believed thatthere was a general shortage of cleaners in theirwaters. The black-spotted wrasse (see Shepherd &Brook 2003) and the western cleaner clingfish arecommon deeper water species, but their abundanceon offshore reefs is still largely unknown

Casimir (1969), cited in Zander & Sötje (2002),suggested that the degree of inclination in the pose ofa client species indicated its appetitive behaviour,which is greater with steeper slope, and Côté et al.(1998) found that fish with an inclined posedenjoyed twice the chance of cleaner attentioncompared with non-posing fish. Others (McCourt &Thompson 1984; Zander et al. 1999) have reported

S. A. SHEPHERD, J. TEALE & D. MUIRHEAD200

CLEANING SYMBIOSES 201

Erika Lawley for their hospitality on Althorpe I., andMichael Clark, James Brook, and Renate Velzeboerfor assistance in the field. Alastair Hirst, SandraLeigh, Steve Reynolds, Rudie Kuiter and Barry

Hutchins kindly provided data and/or advice, andRick McGarvey gave statistical advice. Commentsby Sue Murray-Jones, Paul Jennings and SharonDrabsch improved the manuscript.

References

AYLING, A. M. & GRACE, R. V. (1971) Cleaning symbiosisamong New Zealand fishes. N. Z. J. Mar. Freshw. Res. 5,205-218.

BSHARY, R. & NOE, R. (2003) Biological markets: theubiquitous influence of partner choice on the dynamicsof cleaner fish – client reef fish interactions. InHammerstein, P. (Ed.) “Genetic and Cultural Evolutionof Cooperation” pp. 168-184. (MIT Press, Cambridge,MA).

CASIMIR, M. J. (1969) Zum Verhalten des PutzerfischesSymphodus melanocercus (Risso). Z. Tierspsych. 26,225-229.

CONNELL, S. D. (2002) Effects of a predator and prey on aforaging reef fish: implications for understandingdensity-dependent growth. J. Fish Biol. 60, 1551-1661.

CÔTÉ, I. M. & MOLLOY, P. P. (2003) Temporal variation incleaner fish and client behaviour: does it reflectectoparasite availability? Ethology 109, 487-499.

_________, ARNAL, C. & REYNOLDS, J. D. (1998) Variationin posing behaviour among fish species visiting cleaningstations. J. Fish Biol. 53 (Supplement A), 256-266.

DARKOV, A. A. & MOCHEK, A. D. (1980) Cleaning symbiosisin Black Sea fishes. J. Ichthyology 20, 152-155.

DIXON, P. M., ELLISON, A. M. & GOTELLI, N. J. (2005)Improving the precision of estimates of the frequency ofrare events. Ecology 86, 1114-1123.

EDGAR, G. J. (1997) “Australian Marine Life: the Plants andAnimals of Temperate Waters”. (Reed Books,Melbourne.)

EIBL-EIBESFELDT, I. (1955) Uber Symbiosen, Parasitismusund andere besondere zwischenartliche Beziehungentropischer Meeres-fische. Z. Tierspsych. 16, 19-25.

GRUTTER, A. (1996) Parasite removal rates by the cleanerwrasse Labroides dimidiatus. Mar. Ecol. Prog. Ser. 130,61-70.

HENRIQUES, M. & ALMADA, V. C. (1977) Relativeimportance of cleaning behaviour in Centrolabrusexoletus and other wrasse at Arrábida, Portugal. J. mar.biol. Ass. U.K. 77, 891-898.

HOBSON, E. S. (1971) Cleaning symbiosis among Californiainshore fishes. Fish. Bull. 69, 491-523.

HUTCHINS, J. B. (1979) “The Fishes of Rottnest Island”(Creative Research, Perth).

_________ (1991a) Southern Australia’s enigmaticclingfishes. Aust. Nat. History 23, 627-633.

_________ (1991b) Description of three new species ofgobiesocid fishes from southern Australia, with a key tothe species of Cochleoceps. Rec. West Aust. Mus. 15,655-672.

_________ & SWAINSTON, R. (1999) Sea Fishes of SouthernAustralia. 2nd Ed. (Swainston Publishing, Perth).

JOHNSON, W. S. & RUBEN, P. (1988) Cleaning behavior ofBodianus rufus, Thalassoma bifasciatum, Gobiosomaevelynae and Periclimenes pederseni along a depthgradient at Salt River Submarine Canyon, St Croix.Envir. Biol. Fish. 23, 225-232.

KUITER, R. H. (2000) “Coastal fishes of South-easternAustralia”. (Gary Allen, Sydney).

LOSEY, G. S. (1974) Cleaning symbiosis provides a positivereinforcer to fish. Science 184, 179-180.

_________ (1979) Fish cleaning symbiosis: proximatecauses of host behaviour. Anim. Behav. 27, 669-685.

_________, GRUTTER, A. S., ROSENQUIST, G., MAHON, J. L.& ZAMZOW, J. P. (1999) Cleaning symbiosis: a review. pp.379-395 In Almada, V. C., Oliveira, R. F. & Gonçalves, E. J. (Eds) “Behaviour and conservation of littoralfishes”. (Instituto Superior de Psicologia Aplicada,Lisbon).

MCCOURT, R. M. & THOMSON, D. A. (1984) Cleaningbehavior of the juvenile Panamic Sergeant Major,Abudefduf troschelii (Gill), with a resumé of cleaningassociations in the Gulf of California and adjacentwaters. Calif. Fish and Game 70, 234-239.

MOOSLEITNER, V. H. (1980) Putzerfische und-garnelen imMittelmeer (Cleanerfish and Cleanershrimps in theMediterranean). Zool. Anz., Jena 205, 219-240.

MORRISON, S. & STORRIE, A. (1999) “Wonders of WesternWaters: the Marine Life of South-Western Australia”.(Dept Conservation & Land Management, Perth).

SAZIMA, I., SAZIMA, C., FRANCINI-FILHO, R. B. & MOURA,R. L. (2000) Daily cleaning activity and diversity ofclients of the barber goby, Elacatinus figaro, on rockyreefs in southeastern Brazil. Env. Biol. Fish. 59, 69-77.

SHEPHERD, S. A. (1998) Studies on southern Australianabalone (genus Haliotis) XIX. Long-term juvenilemortality dynamics. J. Shellfish Res. 17, 813-825.

_________ (2005) Ontogenetic changes in diet, feedingbehaviour and activity of the western blue groper,Achoerodus gouldii. Rec. W. Aust. Mus. (in press).

_________ & BROOK, J. B. (2003) Encounter 2002expedition to the Isles of St Francis, South Australia: reeffishes. Trans. R. Soc. S. Aust. 117, 269-279.

_________ & _________ (2005) Foraging ecology of thewestern blue groper, Achoerodus gouldii, at the AlthorpeIslands, South Australia. Trans. R. Soc. S. Aust. 129, 202-208.

_________ & WOMERSLEY, H. B. S. (1970) The sublittoralecology of West Island, South Australia. 1.Environmental features and algal ecology. Trans. R. Soc.S. Aust. 94, 105-138.

THOMPSON, S. K. (1992) “Sampling”. (John Wiley & SonsInc.: New York).

THRESHER, R. (1977) Pseudo-cleaning behavior of Floridareef fishes. Copeia 1977, 768-769.

YOUNG, C. M. (1996) Wrasse as cleaner fish. pp. 204-210.In Sayer, M. D. J., Treasurer, J. W. & Costello, M. W.(Eds) “Wrasse: Biology and Use in Aquaculture”(Fishing News Books Ltd, Oxford).

ZANDER, C. D. & NIEDER, J. (1997) Interspecificassociations in Mediterranean fishes: feedingcommunities, cleaning symbiosis and cleaner mimics.Vie Milieu 47, 203-212.

_________ & SÖTJE, I. (2002) Seasonal and geographicaldifferences in cleaner fish activity in the MediterraneanSea. Helgol. Mar. Res. 55, 232-241.

_________, MEYER, U. & SCHMIDT, A. (1999). Cleaner fishsymbiosis in European and Macronesian waters. In:Almada, V. C., Oliviera, R. F. & Gonçalves, E. J. (Eds)“Behaviour and Conservation of Littoral Fishes”. pp.397-422. (Instituto Superior de Psicologia Aplicada: Lisboa).