Ethology 103, 283-291 (1997) 0 1997 Blackwell ~issenschafts-Verlag,Berlin ISSN 0179-1613 Centerfor Ecology, Evokution and Behavior, 7:H.Morgan .School of Biological Sciencer, riniversio of Kentucb, 13exington Antipredator Behaviour and Suppressed Aggression by Convict Cichlids in response to Injury-released Chemical Cues of Conspecifics but not to those of an Mopatric Heterospecific BM:IN D. WISENUfi;.ni tk R. CRAIG; SARGENT WISH~D~.:N,B. D. tk SARIXNT, R. C. 1997: Antipredator hchaviour and suppressed aggresyion by convict cichlids in response to injuryreleased chemical cues of conspecifics but not to those of an allopatric heterospecific. Ethology 103, 283-291. Abstract In aquatic environments, chemical cues sene as an important source of information for the dctcctron of predation risk. Here, we investigate the response of convict cichlids, (;lcbIusoma nigr+.rciatum, to injuryrelcased chemical cues. We exposed pairs of juvenile convict cichlids first t~ dechlorinated tap water (control), then hter to one of two test stimuli: 1. chemical cues from injured convict cichlids; or 2. chemical cues from injured mosquito fish, Gambzisiu uznnir. Gambusiu are allopatric and phylogenetically unrelated to convict cichlids. C~umhusiu?kin was used to control for a general response to injured fish. In response to conspecific cues, convict cichlids significantly increased time spent near the bottom of tcst aquaria and time under a shelter object. In rcsponse to (,bm/7usiu skin, convict cichlids tended to increasc ume spent near the tank bottom hut did not increase use of the shelter object. There was a trade-off between antipredator and agonistic behaviours. In response to con\wt cichlid cues, there was a significant reduction in the frequency of approaches and bites. Gumhush skin extract had no significant effect on aggressive behaviour. These data sugest a species-specific antipredator response to conspecific alarm pheromones in a New World cichlid fish and demonstrate a trade-off between predator avoidancc and intraspecific aggression. Further, the presence of an alarm response in this model species sets the stage for thc use of chemical cues as a research tool to manipulate predation risk in studies of the interaction betwccn predation risk and reproductive bchaviour. Corresponding author: Brian WISENDJ~N,Center for Hcolo~y,Evolution and Beha\ ior, T H. Morgan Sch(101 of Bi(11ogical Sciences, 101 Morgan Building, Uni\ ersity of Kentucky, Lexington, liY 4050ti-0225, USA. e-mail: [email protected] Introduction Chemical cues are a rich source of information about the environment. This is particularly true in aquatic habitats where water serves as a good medium for dissolving and dispersing chemical signals (1,Ir.w 2982; HAM 1992). A wide range of aquatic taxa use chemical information to assess predation risk (CHIVERSunpubl. data) including many species of fish (SMITH 1992). Among fishes, attention has focused on the Ostariophysi, LJ. S. Copyright Clcarancc Ccnter Code Statement: 0179-1~13/97/1034-0283$14.00/0 284 €3. D. Vi’lSt.hDF,N & R. C. S,1RGI‘N’l representing 72% of freshwater fish (NEIXIK 1984), and the role of their specialized pheromone system in mediating predator-prey interactions (MAGL’KR:\S 1989; MA’I’I[IS & ShiI’rti 1992, 1993a,b,c; SMI‘I’H1992; CHI~ERS81 SMITH 1994a,b, 199s; hh‘1’~[1Set al. 1995; W1SHNDF.N et al. 1995a; CHIVtXS et al. 1996a). Injuryreleased alarm cues elicit antipredator behaviour from conspecifics and sympatric heterospecifics. Injuryreleased alarm pheromones are not restricted to spccics within the (ha- riophysi. Recently uncovered examples of alarm reactions to conspecific chemical cues for fish taxa outside of the Ostariophysi include the Gasterosteidae (h~ITl~llS8c S~IITH 1993c), Gobiidae (SMITH 1989; SMITH & I,AWREI\;CE 1991; SMITH et al. 19911, (htidac (HI:<;IE et al. 1991), Cyprinodontidae (REXD 1969), Poeciliidae (GARCLAet al. 1992) and Percidae (SMlTtl 1979, 1982; CHIVERS et al. 1995a). The Cichlidae is a speciose family of fish that occur throughout thr. tropics in the Old and New World (BF,RRA 1981). Only a few cichlid species have been tested for a response to injuryreleased chemical cues. Two African species, Oreochrumis mossan//iczis U.-\IS\X’AL 8c Wt\GtIk%Y 1990) and Elupiu mmwp!du (NOBLE1939) exbillit antipredator behaviour in response to injured conspecifics. One South American species, Ptero/dyllum sculare, showed no response (SCHUTZ1956). In this study, we test the convict cichlid, C2da.roonm n&rqfusczatzm, a cichlid species native to Central America (Bussi~;1987), for the presence of an alarm reaction to chemical cues from injured conspecifics. Convict cichlids are commonl>-used in laboratory experiments for studies of parental care (iIRMITA<;E 1960; W1IHI:R & %’F,BER 1971; MYKBEKG1975; BEKNSTElh 1980; NOON,11\: 1983; T(l\&’NstIF.ND & WiOOTTOK 1984; KF.”.hl.F,YSlDE 1985; I<EENLRYSIDF et al. 1985; LA\~I’:RY& KI+XI,F.YSIDF1990; R:\N- C;H12EY & GODIN 1992; FR.\SERet al. 1993; NLTT;\I~Id& I<~Eb!I~EYSII~H1993; RREHS 1993, 1994a,b; l,AVI+,RY& REEBS 1994; FR.ISER& I<EENI.EYSIDF 1995; I2.\\’F.RY 1995; SMITH & WCXYlTOX 1995; \X;lSEKDEN et al. 1995b), and foraging and aggrcssion (PEIKF et al. 1971; PEEI(R& PEEICE1982; KOOPS& GRANT1993; Iam,r3’ & GRXST1993a,b; GRAKDa GRANT1994). Convict cichlid parents jointly defend their free-swimming young against predators for 4-bwk ~ISENDFK1994, 1995). Parental care is therefore a form of antipredator behaviour that directly affects fitness. Predation risk, and the pcrccption of risk, should be an important consideration in studies of parental care. Scveral studies hwe incorporated predation risk into studies of reproductive behaviour in fish (M.GURR.\N & SIiGklHRS 1990; MAGNEIAGEN 1991; MA<;NtIAC;EN & VtSTERGiL4RD 1991, 1003; FOKSGKBN 1992; FORSGREN81 ]LIAGhtlAGF.N 1993; CHI\T,RS et al. 1995b). Determining the presence of an alarm reaction in response to injured conspecifics by convict cichlids will provide a valuable research tool to experimentally manipulate perceived of predation risk in studies of the effects of predation risk on reproductive behaviour. The use of chemical cues to induce the perception of predation risk is preferable to the use of predator models because chemical cues are more realistic. Moreover, chemical cues can be presented in a more standardized manner than is the case for live predators behind transparent partitions. Materials and Methods \Ve obtained juvenile convict cichlids from a commercial tropical fish dealer. \Ye placcil two con\ ict cichlids into each of tcn, 9-1 tcst aquaria (30.5 x lS x 20 cm high). Fach tat aquarium contained a >!km laycr of gravel Alarm Reaction by (:onvicr (:ichlids ti? Chemical Cues 285 and a small shelter ohject composed of a hlacl, plcviglass roof (8 x 12 cm) raised 5 cm above the suhst~itui~ihy three legs cif transparent rigid plastic tubing. The tanks were filled to a depth of IOcni above the gravel with dcchlorinated tap water. A horizontal line 5 cni above the gravel was drawn across the fronr IJF wcti t:inL. Thc tank water was aerated hut not filtered. Test stimuli were inlected through a second line of .~qi.a:iuni airline tubing atrached to the airline supplying air to the airstonc in each tanh. Tcsr stiinuh, dischargcd into the upward stream of bubbles from the airstone, dispersed quickly thrwgh~~utthe test aquarium Fish were allo~ved24h to acclimate to test aquaria hchc trials licgan. Each pair of iirh wa., iested twice using the sanie protocol, with the exception of the stimulus used (wucr control <>rsLin eXti-;ict). In tnc morning (0900L1200 h), each pair of fish was tested for a response to 5 ml of dechlorinatcd rap wirer, to coptrd for the effect of stimulus injection. In the afternwn (1400L1700 h), each pair of lich was retested \vith ow of two test stimuli: (1) 5 ml of rinse water from an iniured ciimict cichlicl, or (2) 5 ml of rinsc water frim an inltircd mosquito fish, Gumhusza afinbr. \Ve uscd Gunihusiu skin to tect for a general response to injureii tish. Xlrrsquirr, hvh arc tion- ostariophysan species in the family Poccihidac and arc phylr~geneticallyunrehted ti>cichlidc. Gunhriodo not co- occur with convict cichlids in the rearing ponds from which wc ohtarncd our fish (,I. Rc )IwRFS, pcr:;. comni.) Test stimuli were prepared as follciws. Iminediatel! hefim each trial, a tisli was killed in\tantls lx its vcrtebral column behind the head with a scalpel hlade. Then I0 \ertical cuts were inade nitli ii dpclon cach side of the hod). The fish was then rinsed with Sin1 r~fdechlvrinatedtap \Later. Rinsc writer w:is sollcctecl in :I beaker and uscd immediately. Time between stimulus collection and injcctiiin in a trial was 2 - niin (hi! female Gunh& wcre used to prepare stimuli ro ensure that thc hdy size of stimulus fish \vas approvini it,:l! equal to that of fish uscd in the cichlid treatment. Mean (+ SD) standard Icngth (mm) of ctimulur firh \\'as 3') 6 + 3.8inm and 38.9 + 3.7 mm for convict cichlids and C;iifih&, respcctivcly (t = 0.3l. df = 26, p = 0.623). \Ye ran a total of 28 trials, each lasting for 11 min: 14 trials using water in rhe morning followed by Gum/i//.riu skin in the afternoon, and 14 trials using water in the morning and convict cichlid skin in the afternoon. I3efore each trial, we prepared test stimuli (5 ml water, <;urn/hrio or cichlid cues). We then wirhdrcw 60 nil of water through the injection tube froin the test tank inti, a 6Ohi syringe, discarded it to rcmmc st:ile \v.itcr from the injecrion tube, and withdrew another 60 ml iif tank water and saved it for later use. ;It l(1-s iiitenalq for i inin, we scored the numhcr of fish (0, 1, 2) positioned below the horizontal line on the front panc of ttic aquariuin, and the numhcr of fish under rhe shelter ohicct.