Fin-Nipping Piranhas

Home , Caquetaia, Characiformes, Piranha

rf Fin-nipping Piranhas :

KIRK 0 WINEMILLER AND LESLIE C KELSO-WINEMILLER ...... PredatoryResponse Durn AMERIcAsPIRANHAS (Characiformes: Characidae) exhibit of Piranhas to a variety of feeding preferences, including whole fISh, chunks . of fISh flesh, fish fins, scales,and fruit and seeds.3,6,7.13A num- AlternatIve Prey ber of piranhas belonging to the genus Serrasalmusare vora- cious predators that specialize in the fins of other fishes (Figures 1, 2, 7-9).5,7,8,10,13These fin-nipping piranhas are Prey selection by South American common in lowland aquatic habitats throughout tropical South America. fin-nipping piranhas was investi- For example, L G Nico and D C Taphorn6,7reported diets of 6 fin-nipping prey-choicegated in the fieldexperiments. and in laboratory Our com- basin,piranhas and sympatric Winemiller13 in the documented low llanos the(flatlands) ecology of Venezuela's3 species (SerrasalmusRio Apure .

parative data from laboratory irritans, S rhombeus,S medinai; Figures 3&4) from a floodplain system in assemblagethat differentexperiments piranha speciesindicate select theMostwestern Serrasalmusllanos of speciesVenezuela. in the llanos exhibit their highest preference . prey speciesin differentproportions, for fins during the juvenileand subadultstages and switchgradually to and that the differences are associ- whole fish as they grow larger.5,7,13However, fins can comprise either a ated with the bodyform and swim- minor or major fraction of the diet of adult fishes depending on the ming behavior of predator and prey. species and season. High rates of fin cropping and extensive fin damage In the field, cichlid fishes were the have been documented in the field,8,10,13,15but which species are most vul- most preferred prey for fin-nipping nerable to fin-nipping piranhas is unknown. Preliminary evidence sug- piranhas. We tested the hypothesis gests that perciform fishes of the family Cichlidae (Perciformes: Cichlidae) that some South American cichlid may be particularly vulnerable to fin predation.5,15 fishes suffer lower levels of fin preda~on~ec~use of theircauda~ Pre y Defenses (tatl) mImIcry of the head regIon. Mostofthelargecichlidsthatin- habit lowland aquatic habitats with Fish fins are nutritious6 and represent a renewable food resource that piranhas have densely scaled medi- piranhas can crop. But extensive and frequent removal of fin tissue should .- an fins and bright caudal ocelli have a significant negative effect on the survivorship, growth, and repro- (eyespots) at the base of the tail ductive potential of the victims. Fishes can avoid fin predators by a num- that mimic the size and vertical ber of mechanisms, including hiding or crypsis. Schooling by small fIShes " position of the true eye. may confuse or frustrate pursuit predators.9 The threat of injury from spines (eg, Pimelodus blochii, Pimelodidae), electric shock (Electrophorus electricus, Electrophoridae), or predation from large piscivores (eg, Cichla Figure1. ocellaris, Cichlidae; Figure 6C) may also inhibit fin predators. Pygocentrosnattereri (red belly piranha) Winemiller15 presented field evidence that the large cichlid Astronotus is thevoracious schooling predator that most ocellatus suffers significantly lower levels of fin predation (in terms of both peopIeassodatewiththe~piranha.Con- frequency and size of nips) than a coexisting cichlid (Caquetaia kraussii) belltraryand to popular.L- . myth, the diets of the red f . .1 . h b. fr.. . d f, d. h b. H d h !y OU"" pi" anhasconSIS. t no t 0).f peop Ie, 0 stmt ar sIZe, a Itat a IllitIeS, an ee mg a Its. e .propose t at butfish flesh,fins, and evenplant material. Astronotus gains some measure of defense against fin predation from its AARONNORMAN special morphology and coloration pattern that causes the caudal region

344 NATIONAL GEOGRAPHIC RESEARCH &: EXPLORATION 9(3):344-357; 1993

As an experimental descrip- to mimic the fish'shead region (Figure 6A&:B).The medianfins (top, bot- tion of piranha diet, it is a tom, and tail) of Astronotusare almost completely coveredwith small, densedark-colored scales. When held in their normal positions,the medi- contribution to thefield. I an fins overlapposteriorly such that the body silhouetteis almostperfectly think the readership of your symmetricalwhen viewed from the side. Mimicry of the head region by journal will benefitfrom the tail is enhancedby the distinct, orangecaudal ocellus (eye spot) which this paper; it is a contribu- closely matches the size, color, and vertical position of the true eye. tion to our knowledgeof Except for the caudal eyespot, the coloration pattern of wild-type Astronotusis drab olive or gray and cryptic againstthe normal surround- the diet of piranhas, and the ings of the fish (ie, decayingleaves and roots of floating aquatic plants). strategiesthat their prey use Caquetaiais believedto be a fairly recentinvader of the Orinoco lowlands, to defend themselves The having originated from Lake Maracaiboand coastaldrainages.4 Lacking manuscript contributes to the orange caudal ocellus, densely-scaledmedian fins, and overlapping marginsof the median fins, Caquetaiasuffers very high levelsof fin preda- the knowledge of the com- tion when piranhasare most densein the westernllanos.15 plex ecologyof largeneo- The caudal ocellus and densely-scaledmedian fins of Astronotusand tropical floodplain rivers. other large cichlids (Cichla spp, Crenicichlaspp; Figure 6C&:D) of the South American lowlands are deterrentsto fin predationwith presenceof REVIEWERS the caudalocellus highly correspondentwith extensivecaudal fin squama- tion.15Winemiller hypothesizesthat by virtue of their greatersize: Larger cichlids are more exposedto fin predatorsthan smaller cichlids; and the headregion of a largecichlid posesa greaterthreat of injury to fin nippers than that of small cichlids. Observationsof captive cichlids indicate that they are awareof the predationthreat posed by piranhas,and that they frequently repel attacks with frontal agonostic displays and assaults (Winemiller,unpublished observations).IO.15 T M Zaretll,17hypothesized that the caudaleyespot of the peacockcich- lid Cichla ocellarisserves primarily as a signal of speciesidentity to larger conspecificsso that cannibalism might be averted.WinemillerI5 argues that the fin predation-caudal head mimicry hypothesisposes a simpler explanation for the presenceof bright caudal ocelli in South American cichlids than the earlier hypothesisof inhibition of cannibalism.The inhibition of cannibalismhypothesis seems implausible given the extremely high diversity and abundanceof heterospecificpredators within the native rangeof that species(Zaret's hypothesis was initially formulatedbased on evidencefrom an introduced population in Lake Gatlin, Panama).For a small Cichlawith a bright eyespot,the advantageof signalingspecies iden- tity to larger piscivorousconspecifics would be outweighedby the disad- Figure 2. vantageof being more conspicuousto other predators. Piranhajaws in vivo. When thejaws The fin predation-caudal head mimicry hypothesis and alternative close,the razor-sharp triangular teeth hypothesescan be testedwith further comparisonsof field data,plus con- interlock tightly and can cleanly shear piecesof fish flesh, bones,or fins. trolled experimentsinvolving manipulation of the false eyespot.We first PAUL A ZAHL presentan analysisof fin damagein a natural assemblageof fishesin the western llanos of Venezuela.Second, we present results of laboratory experimentswith fin-nipping piranhasand alternativeprey: Methods

KIRK 0 WINEMIliER, assistant professor, FIELD DATA and LESUE C KELSO-WINEMILLER, research associate,Depanment of Wildlife and We measured standard length (SL) and fin damage of all (ish specimens FisheriesSciences, Texas A&M Uni- taken during monthly sampling of Cafio Maraca in Portuguesa state, versity, CollegeStation, TX 77843-2258. Venezuela, throughout 1984.16Cafio Maraca is a stream-swamp within the

346 RESEARCH &: EXPLORATION 9(3): 1991 A B

c D

E F

Rio Apure drainagethat experienceslarge changesin habitat characteris- Figure 3. tics due to highly seasonalrainfall.16 On both field-preservedspecimens Examplesof fin-nipping piranhasfrom and experimentalfishes, the depth of eachfin nip was measuredfollowing Venezeula:A, Pygocentruscaribe, themethods reported in Winemiller.is juvenilesare facultative fin predators and adultsshear off chunksof fish flesh. LABORATORY EXPERIMENTS B, Serrasalmusmanueli,juveniles are fin specialistsand adultsshear off fine fish Six juvenile fin-nipping Serrasalmusirritans, 1 juvenile 5 medinai,and 1 flesh and sometimeseat fruit. juvenile 5 elongatus were collected from ponds (prestamos) of the C, 5 rhombeus,juvenilesare fin specialists UNELLEZ experimental ranch in Apure state. Fishes were captured by and adultsshear off fish flesh and consume seine and cast net, and later were transported to Knoxville, Tennessee, wholefishes. D, 5 medinai,juvenilesand small adultsare fin spedalists. where behavioralexperiments were conducted.The piranhaswere housed E, 5 irritans, juvenilesand adultsare fin either separatelyor as pairs in separatelS0-L glassaquaria and fed gold- spedalists.F, 5 elongatus,juvenilesand fish (Carassiusauratus) until their usein the feedingtests. small adultsare fin spedalists. An experimental assemblageof South American fishes (Table 1) was KIRK 0 WINEMILLER

WINEMILLER &: KELSO-WINEMILLER: FIN-NIPPING PIRANHAS 347 Table 1. Frequencies of Nip Damage from 3 Piranha Species in an Experimental Fish Assemblage

PREY SPECIES n S ELONGAWS S IRRITANS SMEDINAI

AnostomusSpp (ff) 2 0 0 0 Hemigrammussp (ff) 2 lc lc 0 Metynnisspp (db) 4 1 2c 7c Moenkhausiasanctaefilomenae (ff) 2 lc 0 0 Piaractusbrachypopmus (db) 1 0 lc 7c,4d Thayeriaboehlkei (ff) 3 2c 0 0 CHARACIFORMES 14 5c 4c 14c,4d Aptemotusalbifrons 1 0 0 0 GYMNOTIFORMES

Bunocephalussp (ff) 1 0 0 0 Corydorasspp (db) 6 lc 2c 2c Chaetostomasp (ff) 1 0 0 0 Farlowellasp (ff) 1 0 0 0 Pekoltiavittatus (db) 1 0 0 0 Platydorasamaurus (db) 1 0 0 0 Pimelodellasp (£t) 1 0 0 0 Pterygoplichthyssp (db) 2 lc 4c,3d, la, lp 0 Rineloricariasp (ff) 1 0 lp 0 Trachelyichthyssp (£t) 1 0 0 0 Figure 4. SILURIFORMES 16 2c 6c,3d, la, 2p 2c Serrasalmus rhombeus feeds heavily on fish fins when young and later shifts to a Poedliareticulata 3 lc lc 0 diet of whole fish and fish flesh. Unlike the CYPRINODONllFORMES red belly piranhas,S rhombeus tends to Aequidenscurviceps (db) 1 lc 0 0 hunt alone or in small loose-knit groups. Apistogramma sp (ff) 2 0 2c 0 AARON NORMAN Satanopercapellegrini (db) 26 lac, ld 26c, Id 12c, Id, Ip Mesonautafestivum (db) 2 0 Id Id Nannocara anomalum (ff) 2 0 0 0 PERCIFORMES 33 llc, ld 29c, Id 12c,2d, Ip

~TAL FISHES 67 19c, Id 4Oc, 4d, la, 2p 28c,6d, Ip

a=analfin, c=caudalfin, d=dorsal fin, db=deep-bodied(deep-bodied fIShes were thosein which maximum body depth, including the extendeddorsal fin, was ->0.5 TL, where TL=SL + length of caudal fin.), ff=fusiform, p=pectoral fin.

acclimated in a 475-L aquarium landscapedto simulate a stream-edge habitat. Someof the fISheswere capturedin Venezuelaand transportedto the lab, but the majority were purchasedfrom a local fish supplier. The aquariumwater was aerated,filtered, and maintainedat 25°C.Three small submersible pumps maintained moderate water current in half of the aquarium, and rocks, woody debris, and plastic plants provided cover. Artificial light was maintainedon a 14-h-light : 10-h-darkcycle. Resident fisheswere fed frozen brine shrimp. A singlejuvenile 5 elongatus(SL=65 mm) was introduced to the assemblageand allowed to hunt for 1 week. Figure 5. After the removalof the piranha, the frequenciesand locationsof fin nips Relationshipbetween the maximumstan- on residentswere recorded.Residents were monitored each day until all dard lengthrecorded (MSL) and the total fishes had completely regeneratedtheir fins, a period of 31 to 39 days. numberof fin nips countedin the sampleof eachspecies at Calio Maraca, Venezuela, Once regenerationwas complete,the test was repeatedwith a singlejuve- in 1984 (numberof nips=O.337[MSL]-1.60; nile 5 medinai (SL=63mm) for 2 days (yielding a total number of nips r2=O.23; F=16.81; DF=1,58; p

348 RESEARCH &: EXPLORATION 9(3); 1993 A

c D eyespot in Astronotus,we set up the following conditions in 4 150-L Figure 6. aquaria:4 wild-type Astronotuswith 4 red Astronotusand 1 5 irritans (3 A, Astronotus ocellatus,a South aquaria),and 8 wild-type Astronotuswith 1 5 irritans (1 aquarium).Wild- Americancichlid with a caudalocellus and type and "red" variety Astronotuswere purchasedfrom tropical fish whole- scaledmedian fins that result in mimicry of the headregion by the caudalregion, salersin the United Statesand werehoused in a 2-m-diameterplastic wad- B, close-upof caudalocellus of Astronotus, ing pool and fed goldfish, frozen brine shrimp, and commercialpellets C, Cichla ocellaris,a largepiscivorous beforeuse in the experiments.Except for the presenceof a bright orange cichlidfrom SouthAmerica with a caudal caudalocellus, wild-type Astronotusare mottled tan and gray and can fade ocellusand scaledmedian fins, and to a dull gray dependingon environment and mood. Red Astronotusare D, Crenicichla geayi,a SouthAmerican the product of selectivebreeding and differ from wild-type fIShin lacking pike cichlidwith a caudalocellus. KIRK 0 WINEMILLER the caudalocellus and having uniform orange-brownflanks. Each experimental aquarium contained gravel, 3 large r.pcks, 2 pieces of woody debris, and 1 plastic plant anchored to the bottom. In addition, each aquarium contained a box filter for aeration and a submergedaquarium heaterthat maintainedwater at 25°C.Astronotus were fed pelletsand brine &hrimp during the experiments.After 5 days, the experimentwas termi- natedand fish lengthsand nip depthswere measured. To test for the effect of eyespots,we maskedthe natural eyespotsof 1 group of Astronotus with a 10% AgNO3 (silver nitrate) solution. We swabbedAgNO3 over the region of the caudal eyespot of 8 wild-type Astronotusand returned them to the stock pool. After 14 days,the scales in the areatreated with AgNO3had turned opaquegray and all of the eye- spotswere completelymasked. To test for the effectof fin squamationand

WINEMILLER &: KELSO-WINEMILLER: FIN-NIPPING PIRANHAS 349 AsTRONOTUS VS UPOMIS Table 2. Nipped Individuals, Cafto Maraca, 1984

AQUARIUM 1 NIPPED INDIVIDUALS TOTAL INDIVIDUALS PERCENTAGE Serrasalmus initans-iOO mm SL IN SAMPLE IN SAMPLE unmasked Astronotus ocellatus-II5, 133 mm masked Astronotus ocellatus-II3 mm Characiformes 784 14822 5.3 Lepomis macrochiras-I24, 125, 126 mm Gyrnnotiformes 54 667 8.1 Silumormes 234 3434 6.8

AQUARIUM 2 Cyprinodontiformes 3 341 0.1 Serrasalmus-78 mm Perciformes 306 1684 18.2 unmasked Astronotus-130 rom Perciformes(minus Apistogramma) 306 1505 20.3 masked Astronotus-124, 125 rom Lepomis-125, 127, 150 rom Perctfonn numbers are presented with and without the dwarf cichlid Apistogramma. Order was significantly ass0- ciated with the relative frequencies o~ nipped individu~ls in relation to frequencies o.f sampled individuals. (X2=345.1; DF=4, p

FIELD DATA Except for a single immature specimenof 5 medinaicollected at the start of the peak dry season Oanuary), piranhas were present at the Cafio Maraca field site only during the wet seasonOune to August) and the transition to dry season(September to December).I3During the peak dry season, resident fishes at Cafio Maraca must contend with a much reduced aquatic environment, high temperatures(>30°C), and frequent hypoxia.I4Near the end of the transition season,the 4 piranha speciesat Cafio Maraca (Pygocentruscaribe, 5 irritans, 5 medinai, 5 rhombeus) migrate downstreamto deeper,more stable aquatic environments.They return to the site with the beginning of each new wet season.I3 Characteristicpiranha-nip scarson the fins of fIShesat Cafio Maracacor- relatedwith the presenceof piranhasat the site. Elevennips were recorded among the 6825 specimenscollected during the dry season,and all of

350 RESEARCH & EXPLORA TTON qm. Tqq, WILD-TYPE VS "RED" AsTRONOTUS

AQUARIUM 1 SL Serrasalmus irritans-77 rnm SL Wild-type Astronotus ocellatus-96, 101, 105, CHARACIFORMES III rnm nr species= 29 "Red"Astronotus ocellatus-89, 90, 91. 94 rnm nr individuals = 784 sum 196 797 239 98 22 1350 AQUARIUM 2 mean 81.8 0.25 1.02 0.30 0.12 0.03 1.72 Serrasalmus-69 mm SD 40.5 0.55 0.67 0.61 0.42 0.28 1.68 Wild-type Astronotus--8S, 86, 88, 90 mm GYMNOTIFORMES* "Red" Astronotus--83, 84, 87, 91 mm nr species= 4

nr individuals = 54 AQUARIUM 3 sum 32 61 0 93 Serrasalmus-86 mm mean 197.6 0.59 1.13 0 1.72 Wild-type Astronotus-97, 100, 100, 103 rom SD 78.0 0.50 1.35 0 1.16 "Red" Astronotus-87, 89, 89, 96 rom SILURIFORMES nr species 17 = AQUARIUM 4 nr individuals 234 = Serrasalmus-64 mm sum 84 247 17 18 6 372 Wild-typeAstronotus-99, 109, 110, 115 mm mean 90.3 0.36 1.05 0.07 0.08 0.03 1.59 "Red" Astronotus--84, 85, 90, 90 rnrn SD 54.8 0.72 0.56 0.26 0.28 0.16 1.18 CYPRINODONTIFORMES nr species= 2 AQUARIUM 5 (CONTROL) nr individuals = 3 Serrasalmus-98 mm sum 0 3 0 0 0 3 Wild-type Astronotus-lO2. Ill. 115, 117.117.l21.l22.123mm mean 20.7 0 1 0 0 0 1 SD 0.61 0 0 0 0 0 {\ PERCIFORMES nr species= 5 nr individuals = 306 sum 70 353 108 26 31 588 mean 62.7 0.23 1.15 0.35 0.09 0.10 1.92 SD 40.4 0.56 0.85 0.81 0.32 0.45 2.25

Data are grouped by prey order. *Gyrnnotiforrns lack dorsal and pelvic fins. Most gyrnnotiforrns also lack caudal fins, however piranha nips suffered by the caudal tip were scored as caudal nips.

thesewere from the earliestdry seasonsamples (18&:28 January 1984).In contrast,>2400 nips were recordedamong the 22 499 fishessampled during the wet period of June to December. In terms of nip-scar frequenciesrecorded in our sampleof fishesfrom Cafio Maraca, characiform fishes (tetras and their relatives) were the group most often exploited by piranhas, followed by perciform cichlid fishes (Table 2). However, characiformswere by far the most abundant fIShesat the site, so that, per capita,perciform fisheswere the group most preferredby fin-nipping piranhas (Table 3). Cichlids comprised <2% of the total individuals sampledat the site, yet they represented>18% of the individuals with piranha fin-nips (Table3). Becauseit inhabits denseveg- etation in shallow water at the edge of the swamp, the dwarf cichlid Apistogrammahoignei is essentiallyinaccessible to piranhas.If we exclude the dwarf cichlid from the calculations,the remaining cichlids comprised >20%of all nipped individuals (Table2). Cyprinodontiforms (annual killifishes) were the leastnumerous and the leastexploited fIShes,both on an absoluteand per capita basis (Tables2&:3). Like the dwarf cichlid, small killifishes inhabit shallow, vegetatedfringe habitats of the swamp and probably come into contact with fin-nipping piranhas only infrequently: Per capita, siluriforms (catfishes) and gymnotiforms (weakly electric knifefIShes)suffered intermediate fin damagewhen basedon either the sampleof nipped fIShes(Table 2) or the total sampleof fIShes(Table 3).

WINEMILLER &: KELSO-WINEMILLER: FIN-NIPPING PIRANHAS 1,1

~ Thefin-nipping habit of Siluriforms were the 2nd most abundant group and gymnotiforms were thesepiranhas has obvious the 2nd rarest (Table 3); both groups are dominated by nocturnal fIShes that hide in aquaticvegetation during the day: advantages.Fins are an The caudalfin is most frequentlyattacked by piranhas,generally suffer- abundant and renewable ing at least3-fold greaterincidence of damagethan the 2nd most attacked resourcethat can be cropped fin (Table3). Except for the catfIShes(siluriforms), the anal fin (unpaired and replenished,much like ventral fin) is the 2nd most attackedfin. Most catfIShesinhabit the stream vegetation that is browsed bottom, hence their anal fins (ventrally positioned fins) are less exposed to piranhas than their dorsal fins. When fin nips were summed for each by herbivores.Also, by fish species,larger species were shown to suffer more fin nips than smaller attacking thefins of larger species (Figure 5). When the dorsal and anal fins of perciforms are fishes, thesepiranhas avoid attacked, these attacks are nearly always directed only at the soft-rayed resourcecompetition with portions rather than the spiny portions. piscivoresof similar size that LABORATORY EXPERIMENTS WITH FISH ASSEMBLAGES swallowtheir prey whole. In the assemblageexperiment, 3 individual fin predatorsrepresenting 3 specieswere introduced 1 at a time into the sameartificial environment (Table 1). The averagenumbers of nips recordedper day for eachof the predatorswere 3 (5 elongatus),17.5 (5 medinai),and 23.5 (5 irritans). 5 elongatus(Figure 3F) has the most fusiform-elongatebody form of the 3 piranhas tested and attackedmostly the cichlid (5atanopercapellegrini) and fusiform characids(Table 4). If fusiform characidsare the preferred prey of 5 elongatus,then its comparativelylow rate of successcould have been associatedwith the lower abundanceof fusiform characidsin our experimentalassemblage. 5 elongatususually hoverednear cover (eg,vegetation, woody debris) and attackedprey by stalking or ambush.Here we define stalking as slow movement (usually with rapid fin fluttering) toward a targetthat appearsunaware of the predation threat,followed by a rapid attack. Ambushing is defined as gliding behind or near cover followed by a rapid dashtoward passingprey: 5 irritans (Figure3E) preyedheavily upon cichlids and to a lesser extent on slow-moving stout-bodied catfishes,such as Pterygoplichthys and Corydorasspp (Tables1&4). The predatorybehavior of 5 irritans was similar to that observedfor 5 elongatus,except that stalking was observed much more frequently than ambushing. 5 medinai (Figure 3D) preyedmost heavily upon deep-bodiedcharaci- forms (Piaractus, Metynnis) followed by cichlids (Tables 1&4). The predatory behavior of 5 medinai was very different from that of the other 2 Figure 7. piranhas.5 medinai is an activeswimmer and appearsto prefer roving the Catoprion mento (wimplepiranha) is a open midwater column to hovering at stations near cover. We observed specialistthat feeds on the scalesof other fishes. The lowerjaw protrudeswhen only a few ambush attacks and none of 5 medinai's open-waterattacks open.When the teethon the lowerjaw are appeared to involve the stealth (fin-fluttering) exhibited by the other raked acrossthe flanks of the victim, scales piranhas.On severaloccasions, 5 medinai removedlong narrow strips of are removedlike a row of shingles. fin tissue from large prey by initiating the bite near the center of the fin AARON NORMAN rather than at the fin margin, the latter being the usual mode of attack. The first contact with the fin appeared to be with the lower jaw in these attacks. Attacks initiated near the center of a fin left long narrow nip wounds. LABORATORY EXPERIMENTSWITH 2 PREY SPECIES In prey-choiceexperiments involving bluegill sunfISh(Lepomis) and oscar cichlids (Astronotus),sunfish received significantly greater fin damage than the cichlids (Table5). If we excludethe caudalfin. the depth of nips

352 RESEARCH &: EXPLORATION 9(3); 1993 Table 4. Fin Nip Frequencies by Prey Body Form The observeddifferences and by Taxonomic Order in depth of nip could be

S ELONGATUS 5 IRRITANS SMEDlNAI attributed to differencesin fin structure (Astronotus BODYFORM fusiform 7 (33%) 7 (15%) 2 (6%) vs Lepomis) or body color- deep-bodied 14 (67%) 40 (85%) 33 (94%) ation and selectedbehavior X2= 7 .66, DF=2, p

X2=18.95,DF=4,p

Results from X2 tests of association show that the relative frequency of fin nips was significantly associated with piranha species and both prey body form and taxonomic group.

Table 5. Astronotus-Lepomis Experiment

NR OF NIPS F=65.85;DF=l, 26; p

Lepomis 9.12 mm (5.85) Astronotus 40.9 mm (1.76)

INDEPENDENT V ARIABLE=PRESENCE OR ABSENCE OF EYESPOT (UNMASKED ASTRONOTUS VS OCELLUS-MASKED ASTRONOTUS)** CAUDALFIN

NR OF NIPS F=O.O8;DF=l, 12; p=O.778 DF=I, 1; no comparisonmade DEPTH OF NIPS F=O.Ol;DF=l, 9; p=O.936 DF=I, 1; no comparisonmade

'wild-type Astronotus having dense opaque scales covering the median fins and caudal ocelli vs Lepomis lacking both features "wild-type Astronotus with ocelli vs wild-type Astronotus with masked ocelli Correlations between fish length and nips means are presented only for comparisons with p

LABORATORY EXPERIMENTS WITH 2 PREY MORPHS In tenns of the depth of fin nips, red variety Astronotus lacking the caudal eyespot suffered significantly more fin predation than wild-type con- specifics with eyespots (Table 6). The frequency of fin damage was not significantly different between the wild-type and red varieties (Table 6). When wild-type Astronotus were housed with an equal number of red con-

WINEMIUER &: KELSO-WINEMILLER: FIN - NIPPING PIRANHAS 353 body shapesand habitat preferencesmost similar to its own. Figure 9. Evidence from both field data on nip damageand laboratory prey- As conspicuouspredators of neotropical fish choice experimentsindicates that cichlids are either particularly vulnera- communities. piranhas, egoPygocentrus sp. ble to piranhas, or that they are gready preferredby them. Why should affect not only the population dynamics of their prey. but they also influence many either be the case?It seemsunlikely that the fin tissueof cichlids is more aspects of prey behavior by posing a threat. nutritious than that of most other South American fishes.Without bio- ROSALIA DEMONTE chemical data this possibility remains a possible factor in determining piranha feedingresponses, albeit an unlikely one.WinemillerlS argues that larger cichlids provide easytargets for small fin-nipping piranhasbecause: they are relatively slow; most have large, broad fins; they are diurnally active,large, and hencerelatively conspicuous;and they occupy the same habitats as small juvenile piranhas (eg, near stream margins, often near aquatic vegetation or other physical structures). Prey availability also influencesprofitability, and cichlids are generallyabundant in backwater habitats, like Cafio Maraca, that serve as nursery areas for piranhas. Cichlids (perciforms) were not the most abundant taxonomic groups at Cafio Maraca,but they were nipped with disproportionatelygreater fre- quencythan other fish orders(Tables 2&:3).

PREY DEFENSES AGAINST PIRANHAS Approximately half the 83 fISh species documented at Cafio Maraca were nipped infrequently or not at all. Several traits may account for this appar- ent lack of susceptibility to, or avoidance by; fin-nipping piranhas. First,

WINEMIu..ER &: KELSO- WINEMIIJ..ER : FIN - NIPPING PIRANHAS 355 The South American fish small fishes may have fins so small that they do not provide adequate tar- fauna is the most species- gets for piranhas, or they are not profitable in tenns of net energy gained (eg, energy consumed-energy expended to capture and process the rich on Earth. Associated resource). We found a positive correlation between maximum length of with this rich taxonomic species and the frequency of fin damage in the field. The Cafio Maraca site variety is a diversity of eco- is dominated by small fishes, most of which were rarely exploited by small logical roles and species fin-nipping piranhas. In addition, small fishes can find refuge in the narrow spaceswithin dense stands of aquatic plants or in the shallow water at associations,many of which the very margins of streams and swamps. Many small characifonns, cat- seemrather bizarre in ref- fishes (Silurifonnes), guppies, and killifIShes (Cyprinodontifonnes) proba- erenceto ecologicalfindings bly avoid encounters with fin predators by means of this fine-scale habitat in temperate zones. Feed- separation. Even some of the large nocturnal fishes probably avoid encounters with piranhas by hiding in dense vegetation during daylight ing specialization, such as (eg, knifefishes and catfIShes).At the Cafio Maraca site, the dwarf cichlid fin nipping, and defensive Apistogramma hoignci suffered no fin predation, and this likely resulted adaptations,like eyespotsand from the combined effects of size-dependent profitability (smaller fins pro- head mimicry by the tail vide a smaller target) and habitat selection. region, becomeless surpris- Some of the larger fishes may defend themselves against piranha attacks using spines or frontal displays and biting. Many catfISheshave venomous ing as scientists obtain secretions associated with their sharp pectoral and dorsal spines (eg, more information from the Pimelodella spp [Pimelodidae] and Corydoras spp [Callichthyidae]). The tropics. predatory characifonn Hoplias malabaricus (Erythrinidae) has jaws anned with teeth nearly as fonnidable as those of the piranhas (Figure 2), yet it suffered moderately high incidence of fin predation at Cafio Maraca (143 total nips; 24% of all individuals were nipped). Nearly all South American cichlids have only small conifonn teeth used for grasping, yet some suc- cessfully ward off fin-nipping piranhas with frontal attacks and aggressive displays.IS I Sazima and F A Machadolo describe the group defense tactic used by cichlids (Cichlasoma dimerus and Mesonautafestivum) against fin- nipping Serrasalmus in western Brazil. These cichlids organized themselves into a defensive ring near the substrate with their tails oriented toward the center of the ring and their heads oriented outward to fend off piranhas. This group defense would be particularly adaptive given our finding that piranhas attack the caudal fin much more frequently than other fins (Table 3).2.7,8,10 If larger cichlids are among the most conspicuous and profitable of prey for fin nippers, and if cichlids can repel piranha attacks with frontal displays and bites, then features that confound the discernment of a cichlid's head and tail regions would be advantageous. We tested the hypothesis that the dense, opaque fin scales and caudal eyespots of the largest South American cichlids are adaptions that reduce fin-predation by piranhas.IS Astronotus received fewer nips than a perciform fish of similar size (Lepomis) under the controlled conditions of the laboratory; Even though an entire suite of species-specific traits is associated with the treatment effect in these experiments, we believe that the presence of dense, small opaque scales on the median fins of Astronotus was a key trait leading to the observed differences. These dense scales obscure many of the visual cues that might identify fins for piranhas. The fins of Lepomis, as well as the vast majority of bony fishes (including cichlids from other parts of the world), are transparent, and have soft fin-rays running along their length. The fins of Astronotus and other large South American cichlids, such as Cichla and Crenicichla spp, that coexist with piranhas (Figure 5), are more rigid and opaque, and often closely match the colors of the head region.

356 RESEARCH &: EXPLORATION 9(3): 1993 The caudal eyespot in South American cichlids probably enhances In addition to the insights mimicry of the head region by the tail region by confounding visual cues gainedfrom thefield data, received by fin-nipping piranhas. Our experimental treatment for the presence or absenceof eyespots was inconclusive. On the one hand, wild-type the other aquarium tests Astronotus with eyespots showed significandy less fin damage than red showed taxonomic prey Astronotus that lack eyespots (nip depth, Table 6). On the other hand, our preferences,positive results AgNO3-treated wild-type Astronotus without eyespots did not experience in the Lepomis-Astronotus greater fin damage than untreated wild-types. Based on the results of the various experiments, we conclude that dense scales on the median fins are experiments(obvious species the primary masking characteristic, and that the presence of caudal eye- differenceswere the pres- spots probably transmits additional confounding visual cues to the receiv- enceof both densefin er. The piranhas in our eyespot-manipulation experiments may have been scalesand eyespots),and too large to be threatened by Astronotus. Tail mimicry of the head region would only be an effective deterrent to fin predation if the size ratio somesupport for an effect between predator and prey were large enough to pose a real threat to small of eyespots (the red vs wild- serrasalmine piranhas. type experiment). To reduce levels of fin damage, eyespots need not be 100% effective in CORRESPONDENCE frightening away potential fin predators. For a small piranha, the decision 26 AUGUST1993 whether to approach or flee an object transmitting cues asso(:iatedwith the head region of a large predatory cichlid must be made quickly: These small piranhas often rely on stealth, so that when they receive confusing visual signals, they hesitate and miss opportunities for attack. This confu- sion, or startle effect, on predators has also been proposed as an adaptive function of the eyespot patterns on the tail region of butterfly fishes (Chaetodontidae) of coral reefsl2and on the wings of butterflies and moths.l

REFERENCES 1. Blest, A D: The function of eyespot water fIShesin 2 tropical communities. ACKNOWLEDGMENTS patternsin the Lepidoptera.Behaviour Journalof FishBiology 29:S3-{jS; 1986. Numerouspersons assisted in the field 11:209-255;1957. 10. Sazirna,I; &: Machado,F A: collections, and special thanks go to 2. Foxx, R M: Attack preferencesof the Underwaterobservations of piranhasin A Barbarino,D Taphom, L Nico, and red-belliedpiranha (Rooseveltiella nat- westernBrazil. EnvironmentalBiology of R Beach.Institutional supportwas pro- term). Anilnal Behavior20:280-283; 1972. Fishes28:17-31; 1990. vided by D T aphom of La Universidadde 3. Goulding, M: TheFishes and the 11. Schroder,S L; &: Zaret, T M: The los LlanosOccidentales in Venezuela,and Forest. University of California Press, adaptivesignificance of color patternsin collectingpermits were obtainedfrom Berkeley;1980. Cichlaocellaris. Copeia 1979:43-47; 1979. la Direcci6nAdministaci6n y Desarollo 4. Mago, F: Lista de losPesces de 12. Wickler, W: Mimicry in Plantsand Pesquerode la Republicade Venezuela. Venezuela.Ministerio de Agricultura y Animals. McGrawHill, New York; 1968. We thank L Nico for providing valuable Cria, Caracas,Venezuela; 1970. 13. Winemiller, K 0: Ontogeneticdiet commentson an earlierdraft. This 5. Machado-Allison,A; &: Garcia, C: shifts and resourcepartitioning among researchwas funded by the National Food habitsand morphologicalchanges piscivorousfIShes in the Venezuelan GeographicSociety. during ontogenyin 3 serrasalminefish llanos. EnvironmentalBiology of Fishes speciesof the Venezuelanfloodplains. 26:177-199;1989. Copcia1986:193-195; 1986. 14.Winemiller, K 0: Developmentof 6. Nico, L G: Trophic ecologyof dermal lip protuberancesfor aquaticsur- piranhas(Characidae: Serrasalminae) facerespiration in SouthAmerican from savannaand forestregions in the characidfIShes. Copeia 1989:382-390; Orinoco River basin of Venezuela.PhD 1989. dissertation,University of Florida, IS. Winemiller, K 0: Caudaleyespots Gainesville;1991. asdeterrents against fin predationin the 7. Nico, L G; &: Taphorn, D C: Food neotropicalcichlid Astronotusocellatus. habits of piranhasin the low llanos of Copeia1990:665-673; 1990. Venezuela.Biotropica 20:311-321; 1988. 16.Winemiller, K 0: Spatialand tem- 8. Northcote, T G; Northcote, R G; &: poral variation in tropical fish trophic net- Arcifa, M S: Differential cropping of the works. EcologicalMonographs caudallobes of prey fishesby the piranha, 60:331-367;1990. Serrasalmusspilopleura Kner. 17. Zaret, T M: Inhibition of cannibal- Hydrobilogia141:199-205; 1987. ism in Cichlaocellaris and hypothesisof 9. Sazima,I: Similaritiesin feeding predatormimicry amongSouth American behaviorbetween some marine and fresh- fishes. Evolution31:421-437: 1977.

WINEMILLER &: KELSO-WINEMILLER: FIN-NIPPING PIRANHAS "',,7