Environmental Biology of Fishes 26: 177-199. 1989. @ 1989 Klu~'er Academic Publishers. Printl'd in thl' Netherlands
Ontogenetic diet shifts and resource partitioning among piscivorous fishes in the Venezuelanllanos
Kirk O. Winemiller Departmentof Zoology, The Universityof Texas,Austin, TX 78712,U.S.A
Received18.8.1988 Accepted12.12.1988
Key words: Caquetia, Charax. Diet breadth, Diffuse competition, Gymnotus, Hoplias. Resource overlap, Piranha,Pygocentrus, Rhamdia. Seasonality, Se"asalmus
Synopsis
Resource utilization by nine abundant piscivores from a diverse tropical fish assemblagewas examined over the course of a year. All nine speciesexhibited peak reproduction during the early wet seasonand a similar sequenceof size-dependent shifts from a diet composed primarily of microcrustacea, to aquatic insects, and finally fishes. Three piranha species specialized on fish firis, particularly at subadult size classes (SL 30-80mm). Gradual dessication of the floodplain during the transition season was associated with fish growth, increased fish density, and decreasedaquatic primary productivity and availability of invertebrate prey. Based on 118resource categories, averagepairwise diet overlap was low during all three seasons:wet, transition, and dry. Of 72 species pairings, only one pair of fin-nipping piranhas exhibited high overlap simultaneously on three niche dimensions: food type, food size, and habitat. Adults of two species, a gymnotid knifefish and pimelodid catfish, were largely nocturnal. Patterns of habitat utilization indicate that piranhas may restrict diurnal use of the open-water region by other piscivores. Collective diet overlap of individual piscivore specieswith the other eight feeding guild members and collective overlaps with the entire fish community each revealed two basic seasonal trends. Four species that showed an early switch to piscivory also showed a high degree of diet separation with both the guild and community at large on a year-round basis. The five remaining speciesshowed lowest collective diet overlaps during the transition seasonwhen availability of invertebrates was reduced and fish densities were maximal. Whereas predation may playa role in habitat separation, diffuse competition for food resources during the approximately four-month transition seasonprobably is the principal factor yielding patterns of diet specialization.
Introduction bers provides the most straightforward means of attaining a preliminary assessmentof the relative Most diverse taxonomic groups of organisms ex- importance of interactive processeswithin natural hibit divergence in trophic biology (e .g., cichlid communities. Whereas descriptive studies that re- fishes, Greenwood 1981; ostariophysan fishes, veal patterns of resource subdivision are limited in Fink & Fink 1981).Whether or not ecomorpholog- their abilities to demonstrate the mechanismscre- ical divergence is random or adaptive in the context ating observed relationships (Toft 1985), thought- of community interactions can be debated (Strong ful comparisons and the weight of evidence derived et al. 1979, Grant 1981). The study of resource from numerous studies already have yielded much exploitation among syntopic feeding guild mem- insight (Schoener 1974, Helfman 1978, Sale 1979, 178
Werner 1979.Toft 1985).In a recentreview of the focuseson the piscivore guild as a potential unit for literature on fishes.Ross (1986) observed greatest competitive biological interactions. Given the cli- partitioning on food dimensions.followed by hab- mate and natural history of the Cano Maraca eco- itat and time. Many studiestreating two or more system, several basic questions can be asked: (1) nichedimensions have shown complementarity. in- Do piscivorous fishes partition food and habitat, dicatinga potential effect of interspecificcompeti- and if so, what is the influence of seasonalchanges tion. An interactive effect betweencompetition. in their environment? (2) What are the relation- predation, and temporal shifts in abiotic environ- ships between ontogeny, resource partitioning, mental parametershas been suggestedfor some and seasonal changes in resource availability? (3) systems(Werner & Hall 1977,Werner et al. 1983, Given seasonal patterns of resource availability Mittlebach 1984. and see Schoener1974. Toft and exploitation by consumers,is there evidence of 1985).Ross (1986)observed a generalpattern of interactive effects on piscivore feeding behavior? studiesrevealing less evidence of interactiveeffects (4) Do seasonal trends of pairwise species diet on communitystructure in harsherenvironments. overlaps follow patterns from collective diet over- Seasonalchanges in resourceavailability and on- laps within the piscivore guild or total fish commu- togeneticdiet shifts can affect both predator-prey nity, and if not, which index offers a better assess- and competitive interactionsamong size-structur- ment of the likelihood of interactive effects? ed fish populations. The present study explores seasonaland size-relatedpatterns of resourceex- ploitation within the piscivore guild of a diverse Methods neotropical fish community. Although varying greatlyin both duration and breadthof taxonomic Studysite comparison, several recent studies have shown markedresource partitioning amongtropical fishes During every month of 1984,fishes were collected (Angermeier& Karr 1983,Power 1983,Moyle & from Calio Maraca, a swamp-creek of the Rio Senanayake1984, Schut et al. 1984,Watson & Apure-Orinoco drainage in the western llanos of Balon 1984,Araujo-Lima et al. 1986,Hyslop 1986, Venezuela (8° 52' 30" Lat. N; 69°27' 40" Long. W). Prejs & Prejs 1987,Nico & Taphorn 1988).Fish The region studied (termed an 'estero') lies within populationsfrequently are size structureddue to low, flat terrain that experiences extensive sheet small hatching or neonate size and continuous flooding during the wettest months (late May-Au- growth. Body size significantly affects the size gust). During this time, a large region bordering rangeof food particlespotentially available to con- the creek channel is converted from exposed, sun- sumers(Brooks & Dodson 1965. Werner 1974, baked soil and thorn-scrub habitat into a produc- Magnan& Fitzgerald1984). Furthermore, size dis- tive marsh with diverse aquatic vegetation dom- tributions of fish populations directly influence inated by Cyperus sp., Eichhornia diversifolia, and predation rates on different ageclasses and often Heteranthera reniformis. During the driest months the spatial distribution of fishes (Fraser& Cern (Dec.-May), the aquatic habitat is reduced to a 1982,Mittlebach 1984,Power 1984,Werner & Gil- network of mud-bottom pools blanketed by Pistia liam 1984). stratiotes, Salvinia spp., and Lemna spp.. Dis- The studysite, Calio Maraca,experiences highly solved oxygen concentrations are reduced at this seasonalrainfall that is characteristicof manyparts time and many fishes rely on special respiratory of central and southern Venezuela.Most fishes adaptations for survival. Annual rainfall in the re- reproduceduring the early and peakwet season,a gion was 1300mm during 1984,with 860mm falling situationthat leadsto temporalchanges in average between June and September (Fig. 1). Based on body sizefor mostspecies (Winemiller, unpublish- changes exhibited by stream physico-chemical ed). Although similar data were collectedfor all parameters associated with rainfall and their ef- fishesthat occurred at the site, this investigation fects on aquatic organisms (Winemiller 1987a, 179
three wet months, collecting effort expended was approximately equal during each sampling period. More total time was spent collecting during the first ! three months of the wet seasondue to the compara- ~ tive inefficiency of adult fish captures in the newly- i.. 0 expanded aquatic environment. All aquatic hab- itats in the estero region were sampled for a mini- ...... 10. ,.. -0' A~, _0' .lu. .lul ... 00' Mo. 0.. mum of six hours (habitats combined), and a col- lecting outing was terminated when two hours of sampling yielded no additional species.Collections Fig. 1. Average monthly rainfall (closed symbols) and maximum used for comparisons of relative densities were midpool depth measured each month (open symbols) during the made during either one or two days between the study period at Cano Maraca. 11th and 28th of each month. Additional collec- tions were made on other dates to supplement stan- 1987b), the annual cycle was divided into three dard samples of uncommon species. A complete terms of equal duration: dry season (Jan.-Apr.), listing of collecting dates and sample sizes for pre- wet season (May-Aug.), and transition season served specimens appears in Winemiller (1987a). (Sept.-Dec.). Dry season conditions gave way to Collecting activities should not have affected nat- the wet season very abruptly with the commence- ural fish densities between sampling periods, be- ment of heavy rains, whereas the wet seasonslowly cause(1) only a fraction of the total aquatic habitat graded into the next dry season over a prolonged was sampled on a given date and (2) fishes could period (i.e., transition season). During the transi- move freely between the estero region and the tion season, fish densities increased and biotic in- pennanent channel downstream (except during teractions intensified within gradually diminishing March and April). All collected individuals were aquatic habitats in the floodplain. Estimates of fish preserved in 15% formalin rather than the usual biomassat Cano Maraca at the end of the transition 10% to safeguard against decomposition of stom- season are among the highest densities recorded ach contents. The largest fishes were cut on the for natural freshwater ecosystems(141-165g m-2, lateral wall of the visceral cavity to hasten pene- Winemiller 1987a). All available climatic and ec- tration of formalin. Following examination, a por- ological information indicate that 1984was not un- tion of the fishes collected was deposited with the usual in the sequence or magnitude of events. For Museo de Ciencias Naturales de UNELLEZ, Gua- example, the rainy seasonsometimes beginsearlier nare, Portuguesa, Venezuela. The remainder was or later than the last week of May, yet the first rains deposited in the Natural History Collection of the are typically heavy, persist for several months, Texas Memorial Museum, University of Texas, slowly subside, and nearly cease all together for Austin, Texas. several months during the peak dry season. Separate samples were taken on 27 and 28 Sep- tember, 25 and 26 October, 22 November, and II December from three habitats: open water mid- Collections pool (day and night), shallow pool edge, and dense vegetation. The open water habitat was sampled in Fishes were collected by seine (3.17mm un- a large pool at the lower end of the estero with the stretched mesh. 2.5m length; 12.7mm mesh. 20m 20m, 12.7mm mesh seine. These large seine sam- length). experimental gillnet. and dipnet. On each ples were preserved in 15% formalin and kept sep- outing. an attempt was made to sample the entire arate from small seine (2.5 m, 3.17mm mesh) sam- fish community such that the sample for each spe- ples taken from the same pool. Becausethe large cies reflected its relative abundance and size struc- mesh seine allowed small fishes to pass through, ture during each month. With the exception of and large fishes easily avoided the small seine, 180 values obtained for the number of individuals of for stomach content analysis. For certain piscivo- each species in the two open water samples were rous species, such as Hop/ias ma/abaricus and Py- later combined as one sample. Night sampleswere gocenlrus nolalus, all available specimenswere dis- taken from the same open water location with both sected. When more than 30 specimenswere avail- seines on 27 September and 25 October (1800- able, a subsample was chosen for dissections that 2400h). The 2.5 m seine was used to sample open included representatives spanning the total spec- and sparsely vegetated, shallow shoreline habitat trum of size classes in the collection. During the during the morning of each outing. For the vegeta- course of one year, 83 fish specieswere collected tion habitat samples, the 2.5 m seine was rolled on from the estero. Each monthly sample produced poles attached to each end until the total area of between 40 and 60 species (Winemiller 1987a). exposed net was 4.2 m2. The weighted end of the This paper reports findings for the nine most abun- seine was passed below dense massesof floating dant piscivorous speciesonly: Hop/ias malabaricus vegetation and quickly brought to the surface on (Erythrinidae), Charax gibbosus (Characidae), the other side, enveloping a quantity of aquatic Pygocenlrus nolalus (Characidae), Serrasa/muscf. vegetation approximately equal to the area of the irritans (Characidae), Se"asalmus cf. medini, Ser- seine and entrapping all fishes contained therein. rasalmus rhombeus, Gymnolus carapo (Gymnoti- The vegetation was then hauled to the adjacent dae), Rhamdia sp. (Pimelodidae), and Caquelia shoreline where fIShes were extracted and pre- kraussii (Cichlidae). For the purposesof this study, served. piscivores were defined as fishes averaging at least 50% fish prey in the diet of the largest adult size class. Six piscivorous specieswere rare in the local Diet analysis ecosystemand omitted from the analysis [i.e., Ac- eslrorhynchus microlepis (Characidae), Ageneio- All preserved specimens were identified, counted, sus brevifilis (Ageneiosidae), Crenicich/a saxalilis and measured for standard length to the nearest (Cichlidae), Hoplerylhrinus unilaeniaIUS(Erythri- 0.1 mm (alliengtbs are reported hereafter as SL). nidae), Pseudoplalysloma fascialum (Pimelodi- Fisb species names were provided by Donald Ta- dae), and Slernopygusmacrurus (Sternopygidae»). phorn of UNELLEZ and Francisco Mago-Leccia Additionally, four common speciesthat consumed of UCV (Caracas). and voucher specimens are fishes in minor proportions were excluded from the available in the Museo de Historia Natural (UN- comparisons [Aequidens pulcher (Cichlidae), As- ELLEZ) and The Texas Natural History Collec- lronOIUS ocellalus (Cichlidae). Parauchenip'erus tion (TMM. UT, Austin). Many Venezuelan spe- galealus(Auchenipteridae), and Synbranchusmar- cies could not be given tentative speciesnames due moralus (Synbranchidae)J. to the current state of taxonomic confusion. Conse- Food items in the anterior half of the gut were quently, a few species were designated as distinct removed, examined under a dissectingmicroscope, forms by 'sp.'. Two piranhas were initially identi- and identified at the species(fishes) or ordinal level fied as Se"asalmus eigenmanni and Serrasalmus (invertebrates). Broader food categories were not sp. (Taphorn & Lilyestrom 1985).The piranha sub- used for numerical analyses,because (I) lower tax- family (Serrasalminae) is currently under revision. onomic categories generally represented one or According to William L. Fink (personal communi- more renewable resources, and (2) lumping re- cation). these species correspond to Se"asalmus source states often inflates niche overlap values cf. irritans and Se"asalmus d. medini, respective- (Greene & Jaksic 1983). Prey items then were sort- Iy. Caquetia kraussii was formerly listed as Petenia ed, carefully blotted with a dry paper towel, and kraussii (Mago-Leccia 1970. and see Kullander while still moist, measured in an appropriately 1983). sized graduated cylinder by water displacement. Whenever available, 30 specimens of each spe- The degree of resolution for very small items was cies from each monthly collection were dissected 0.005 mi. For volumes less than 0.005ml, a value 181 was estimated by spreading the item on a glassslide mirrors the feeding habits of the population at and comparing it to a similarly spread substance large. This study employed niche overlap (fp) as a known to equal 0.005 mi. Except for highly frag- measureof similarity only, and not as a direct esti- mented material. the size interval associated with mate of interspecific competition (a) in the manner each identified prey item was recorded (10mm in- previously used by some investigators. tervals measured along the longest particle axis). Following the method of Lawlor (1980), electiv- An ocular micrometer aided linear measurements ities were used in place of Pi~ for an additional set of small prey items. Diet breadths were calculated of diet overlap measures adjusted for prey relative by Simpson's measure (Simpson 1949). Asymmet- abundances (hereafter referred to as diet similar. rical diet overlaps were calculated by MacArthur & ities). Consumer electivities (ei,) were calculated as Levins' (1967) formulas:
e='/ RbPi/ ' . I L P;/'IA (fJik= --'-I~I- . where Rj is the relative proportion of food item i in L plj2 i-I the environment, and pjj is the proportion of item; in the diet of consumer speciesj. The relative abun- and dance of each prey item in the environment was estimated from the column sums in the community . resource matrix (items summed across all fish spe- I: PtIJ~ cies examined, both piscivorous and nonpiscivo- fit! a I-I . rous). Although it is labor intensive, this method of I-Ir P1.2 calculating Rj has the advantage of basing total prey availability on the overall performance of the en- tire fish community. Assessmentof prey R;s based where Pij is the proportion of prey item i consumed on independent sampling efforts risk (1) under- byspeciesj. andpik is the proportion of the ith prey estimating availability when microhabitats avail- item consumed by species k. These measures re- able to fish are undersampled, and (2) overestimat- flect differences in the magnitude of overlap be- ing availability when large numbers of prey that are tween two populations due to unequal niche actually invulnerable to predation are sampled breadths. For example, one would expect diet (Wallace 1981). In effect, the method estimates overlap of the more specialized bluegill sunfish. food availability using fishes rather than humans as Lepomis macrochirus. on the green sunfish. L. samplers of their environment. The method is cyanellus. to be less than that of the latter on the more reliable when based on diverse assemblages former (Werner & Hall 1977). The asymmetrical containing many ecomorphotypes, because any measure has been viewed as desirable in the sense single specieswould have a relatively small influen- that potential for exploitation competition for re- ce on estimates of availability. sources is more realistically portrayed as unequal Ecological measuresand statistics were comput- for specieshaving different diet breadths (Slobod- ed from the three seasonal resource matrices (9 chikoff & Schulz 1980). Whereas this and other predators x 118 potential prey categories). For measures of niche overlap present difficulties for analysis of ontogenetic diet shifts, dissected speci- statistical inference when three or more speciesare mens were grouped into either 10 or 20 mm in- compared. they at least provide a simple. reliable tervals, depending on the total range of SLs record- index of diet similarity. All PijSwere based on the ed for each species. The average proportion of total volumes of the food items consumed during a microcrustacea,aquatic insects, and fish prey items specified period. This method assumesthat stom- was determined for each fish size interval. Fish ach contents from a large sample of individuals scales were omitted from this analysis, because 182 scalescomprised a minor portion of diets and none ta argentea. numerically dominated the estero of the nine speciesare believed to rely on scalesas a (Winemiller 1987a). significant food resource (sensu Roberts 1970, Sa- Average size of all nine piscivores was greater rima 1983). Most scales not associated with flesh during the transition season than the wet season were probably taken during aggressiveinteractions due primarily to growth of juveniles produced dur- by these species.All statistical tests were computed ing June and July (Fig. 2). Hoplias. Charax, Pygo- by SAS (SAS Inst. 1985). A generalized linear centrus, Gymnotus, and Caquetia were extremely model (GLM) procedure was used for calculation abundant within shrinking aquatic habitats during of ANOVA statistics for comparisons of data sets the transition season. Populations of Hoplias, with large unequal sample sizes. Charax, and Caquetia consisted of a mixture of adults, subadults, and juvenile size classesduring the transition season. Both adult and subadult size Results classesof Rhamdia and Gymnotus were taken dur- ing the transition season, whereas the four piranha The piscivore guild of the dry season was dom- species primarily consisted of subadults. During inated by adult Hop/ias ma/abaricus, followed by late October, dissolved oxygen concentration ex- Caquetia kraussii (17-185 mm), adult Gymnotus hibited a temporary reduction (mean dissolved carapo. and adult Rhamdia sp. (Fig. 2). During the 0] = 0.73 ppm), which probably resulted from a harshest months of the dry period (Feb.-early pulsed increase in total microbial metabolism asso- May), Charax gibbosus and the four piranha spe- ciated with the death of aquatic macrophytes in cies (Pygocentrusand Se"asalmus spp.) emigrated shallow standing waters of the drying floodplain from the estero region to deeper aquatic habitats (Winemiller 1989). At this time, all four piranha downstream. Collections made on Caito Maraca at specieswere rare or absent from the estero region a location 14km east (downstream) of the estero of Cano Maraca. The four species reappeared on during the height of the dry seasonyielded several the site after dissolved oxygen had increasedwithin individuals of Charax. Se"asa/mus irritans, and S. the stream channel (0] = 2.s-6.0ppm) within six medini. Small juveniles of Hop/ias and especially weeks. Young-of-the-year Pygocentrus (76- Caquetia were taken at various intervals during the 127mm) were taken in large numbers during No- four month dry season,indicating that at least por- vember and December (Fig. 2). tions of the local populations were reproductively active. Charax, Pygocentrus, and S. irritans returned to Ontogeneticdiet shifts the estero within one week of the abrupt onset of the rainy season. All individuals of these species Of m9 specimens from Cafio Maraca that were taken during June were adults with gonads in ad- dissected for diet analysis, the nine major pisci- vanced states of maturation. Eleven large male vores comprised 1980(20%). Small juveniles of all Pygocentrusin black nuptial coloration were taken nine speciesexploited aquatic microcrustacea and in shallow water among emergent stands of aquatic aquatic insects (Table 1, Fig. 3). Pygocentrus macrophytes with one haul of the 2.5 m seine on showed two distinct ontogenetic shifts in basic diet June 6. Each male had milt-laden testes, and their components: from primarily microcrustacea to aq- stomachs contained either large cycloid scales or uatic insects (20-40 mm), and from aquatic insects plant debris. Juveniles of all nine specieswere col- to fishes(4()...()() mm). Rhamdia experienced similar lected during July and August. During the wet diet shifts between 2a-40 mm and ro-so mID re- season, small omnivorous and herbivorous char- spectively. A mixture of microcrustacea and aquat- acids (especially Astyanax bimacuJarus.Markiana ic insects dominated the diets of juvenile Hoplias geayi. Ctenobrycon spi/urus, and Odontosti/be «50mm), S. irritans «30mm), S. medini Du/cher), and the detritivorous curimatid, Curima- «40mm), S. rhombeus «40mm). GymnotILS 183
, .. A . 0 N 0 Monlll
, M AM" .. Mon."
Fig. 2. Mean monthly standard lengths {or each piscivore speciesat Cafio Maraca. Error bars represent one standard deviation and coUection sample sizesappear within each open bar. 184
«I20mm). and Caquetia «40mm. Fig. 3). equal portions of unidentified fish fins and whole Ephemeroptera nymphs alone comprised 65% of Astyanax bimaculatus. The largest S. medini speci- the diet of juvenile Charax «30mm. Table I). mens examined (71.6-80.6mm) corresponded to Gymnotus smaller than 40 mm consumed62Ofo chi- subadult size classes. and all contained fins (the ronomid larvae (Diptera). whereas between 120 largest specimen contained fins and associated and 240 mm. Ephemeroptera. Odonata nymphs. flesh of Hoplias). Between 40 and 50mm. Pygo- and chironomid larvae were predominant in Gym- centrus shifted to a diet comprised primarily of notus stomachs. Juvenile Caquetia between 40 and whole fishes and sheared pieces of fish flesh. A 70 mm preyed heavily upon a wide variety of aquat- wide variety of fish specieswere taken by Pygocen- ic insects (Table 2). Irus. but primarily Curimata argentea. small omni- The shift to a diet comprised primarily of fish vorous characids. and the large detritivorous char- components occurred between 20 and 30 mm for S. aciform. Prochilodus mariae (Prochilodontidae). irritollS, and between 30 and 40 mm for S. medini Adult Pygocentrusconsumed portions of other ver- and S. rhombeus (Fig. 3). These narrow-bodied tebrates (turtles. birds. mammals) when fishes piranhas were fin specialists at subadult and adult were dispersed at low densities during the early wet size classes(Tables 2.3). Fins comprised the major season. Hoplias also shifted to a diet comprised of fraction of the diet of S. rhombeus larger than diverse fISh species between 40 and 50 mm. but SOmm, although this sample consisted of only 2 unlike Pygocentrus. Hoplias always swallowed its specimens. A 195mm specimen contained nearly prey whole. Hoplias is capable of consuming rela-
T~ 1. Resource matrix of dominant rood items consumed by juvenile size classesor piscivorous fishes at Caito Maraca. Utilization is expressedas a percentage of the total volume of the prey item basedon all specimens. Nine fIShand 20 non-fish itenlS with a combined volume of 45.2 liters are not listed (HMA = Hopl;asmolabaricus. CGI - Charaxg;bbosus. PNO - Pygocenrrusnolalus. SIR = Serrasalmus;rr;lans. SME = Serrasalmus medini. SRH - Serrasalmusrhombeus. GCA ~ Gymnotus carapo. RHA = Rhamd;a sp.. CKR - Caqwlio trows;;).
Prey category HMA 001 PNO SIR SME SRH OCA RHA CKR
Vegetative detritus o.J2 - 0.17 0.61 6.01 0.12 o.os Copepoda 0.07 o.20 I.B2 0.24 0.51 0.61 o.45 Oadocera 2.29 I. 27 17,10 11.96 16.30 5.18 9.61 9.12 O.8S Amphipocia 0.40 I. 40 O.,59 O.M 4.07 9.,2S Eubranchipoda 0.02 2.(1J 4.50 O.29 Ostracoda 0.03 o.26 I.96 0.09 1.~ 4.84 o.48 Ephemeroptera 17.~ 65.40 71.79 16.73 63.89 20.78 1.~ 10.66 37.18 Odonata nymphs 20.12 4.03 2.49 6.92 6.38 Misc. aq. Hemiptera 0.02 9.78 4.87 Corixidae (Hemiptera) 0.42 7.77 - 4.39 Trichoptera O.20 O.72 1.04 5.66 O.21 O.04 Aq. Coleoptera (Iarv) 0.47 O.a I.42 4.. 1.8 0.49 1.53 Aq. Coleoptera (adlt) O.~ O.87 14.25 1.68 aaironomid larvae 1.86 21.82 I.~ 1.72 4.35 1.04 61.- 25.84 2.41 Mosquito larv.(Diptera) 0.03 O.(M I.45 0.05 0.41 12.74 Misc. aq. insects 2.15 O.88 O.31 1.20 21.37 4.12 Fish fins 68.23 3.55 Misc. fish 8.63 4.19 2.4S 1.53 At'qllidt'tU 5.82 PMCilia 3.61 Hoplias 3.88 Rachovio 7.49 Coq~lia 7.21 185
s.lr888tnus thombeua '.0- Uo; ..- u- ..-..~- OA-
u.
M.L:: ~ T - . .'-~~!' 0 00 10 '. ,. - ~~(-)
GymnotJ8~
. ~ '20 '80 200 200 280 320 LengtllrMrval (mm)
fI*IIct8 SII.
fO 80 120 L8ngtI kI-.V8J (mm) s.r..tInu8lni18ns
u, .0 80 120 .j u- .-."'Aq.- ~ k8V8I (mm) 0.0- .- ! ~ kt8usU 0.2- ..
0.0- .. 0 20 .. 80 10 '00 120 '.. '.0 L8ngtI ~ (mm) u r ;.~+A4- s ~ ~ ..- u u
0.0 ~0 20~:2::::::::::::::;?:!':. 40 " 10 100120 140 ," ," 200220 2-0- :- LqfI J..-,y8I(mm)
Fig. 3. Volumetric proportions of microcrustacea. aquatic insects. and fish prey categories taken by piscivores grouped by length interval. 186 tively large prey items. For example, a 305mm upon small characids, especially Astyanax bimac- individual had consumed a 95 mm specimenof the ulatus, Odontostilbe pulcher, and Ctenobrycon spi- heavily armored, spined catfish, Hoplosternum lit- lurus. Ctenobrycon spilurus and Curimata argentea torale (Callichthyidae). A 288rnrn Hoplias con- were the primary prey of large adult Gymnotus tained a subadult Prochilodus measuring over a (>200mm). Rhamdia larger than 6Omm con- third its own body length. Cannibalism was observ- sumed mostly whole Curimata, but also fresh re- ed among all size classesof Hoplias. mains of Astyanax and Prochilodus that apparently Between 50 and 60 mm, Charax shifted from resulted from frenzied attacks by schooling Pygo- feeding primarily on aquatic insects to small char- centrus. The proportion of fish in stomachs was acid fishes, particularly Odontostilbe pulcher. Like highly correlated with standard length (Table 4). Charax, Caquetia over 70mm preyed most heavily
Table 2. Partial resource matrix of dominant food items consumedby subadult size classesof piscivorous fishes at Cano Maraca. Seven fish and 28 non-fISh items totalling 43.4 liters are not listed.
Prey category HMA CGI PNO SIR SME SRH GCA RHA CKR
- - Fine detritus 0.01 o.~ 0.11 ., .~85 0 .40 Aquatic vegetation 0.21 4.37 - .~ - O. 10 O. 17 0.01 ,ii,. 8. 3S 1.24 Clams (Mollusca) 0.02 - ~ Oadocera 0.05 - ~ ':;; 3. 44 1. 85 0.79
... ! .,. S. 6. 43 O.~ Eubranchipoda 0.04 2". OS 0.46 'J' ,.. 3. 89 7.72 Shrimp (Decapoda) ~: ;~~ 7.48 6.25 0.04 ,.. j,..;!!'" 37. 46 7. 54 2.02 Ephemeroptera ;, 20. 20. Odonata nymphs 4.76 10.35 0.57 :;rc~ .- 82 12 2.41 Misc. aq. Hemiptera 0.57 2.52 0.03 ~: -~~ .~ S.07 Corixidae (Hemiptera) 0.04 14.89 ",~ ~ ~ 2.38 Aq. Coleoptera (Iarv) 0.10 0.36 ...;'~ ~;~ . 2.,84 4. 03 2.77 Aq. Coleoptera (adlt) 0.45 0.11 J O. 02 O.40 12.63 Chironomid larvae 0.07 0.26 ~.' 93 19. 89 0.37 Misc. aq. insects 0.26 0.19 0.01 1.,24 O.79 6.54 Onhoptera 1.05 O.99 15.21 Fish fins 0.72 0.07 M.17 100 M). ,'T7 Misc. fish flesh 0.18 32.51 3. 58 1.16 Misc. fish (whole) 10.52 9.59 6.89 2.41 1. 99 4.54 Corydoras c~; 7. 9S Odonloslilbe 3.58 40.16 14.22 Aequidens 4.37 ,.. 1.00 ~ Gephyrocharax 7.56 0.50 - Hemigrammus 3.73 ~~ 1.00 Poecilia 2.01 :: . ~ c" 1.33 Curimala 25.79 10.37 - ~;~ ... Astyanax 5.87 17.19 .)"'"- Hoplios 8.31 ,- Rachovia 3.58 Triportheus 4.37 Markiano 5.S9 Caquelia 9.67 1.26 Cheirodontops 3.SJ - - Gymnotus 9.93 Hypostomus 15.68 Apislogramma 5.16 187
Tobit' 3. Panial resource matrix of dominant food items consumed by adult size classesof pisciyorous fishes at Cailo Maraca. Twenty four fish and 32 non.fish items totalling 92.9 liters are not listed.
Prey category HMA CGI PNO SIR SME SRH GCA RHA CKR
Vegetative detritus 0.01 0.86 3.56 3.13 o.~ - 0.04 5.45 Aquatic vegetation 0.15 1.01 S.SS 2.53 2.17 1.91 7.27 Hard fruits, seeds 0.29 0.31 ff~% O.~ 0.84 7.27 Fish ew 0.17 12.01 1.19 Snails (Mollusca) ~"t~ - 4.86 1.88 .. Shrimp (Decapoda) 0.16 12.13 0.92 - ~ 16.30 8.96 4.35 Odonata nymphs 0.26 0.04 ~ IS.tIJ 1.24 3.46 3.46 Orthoptera 0.06 0.86 2.33 0.44 Arachnida 0.13 S.20 4.15 Ctenoid scales 3.15 0.05 2.00 Fish fins 0.31 5i.S3 76.19 .!6..54 Misc. fish flesh O.OS 0.47 14.46 3.64 0.89 Misc. fish (whole) 3.07 23.18 2.22 50 6.70 1.20 22.19 OdontostiJbe O.~ 37.13 0.07 - - 2.81 7.00 Ctenobrycon 4.16 2.30 - 11.38 6.92 Pimelodella 2.34 J7,jS : Curimata 20.19 I.~ 22.64 .. 10.54 26.00 30.63 Astyanax 4.43 3.74 ;. e .44 0.84 9.36 ProchiJidus 25.23 7.91 . 9.88 Rineloricaria 6.17 0.14 ... Hoplias 2.24 .. n... .,,; - S.:M> . :..;:: Triportheus ,'" Roeboides 17.17 "" .. 3.10 Hoplostemum ~.67 ~ Non-fish vertebrates S.SS ~
Table 4. Product moment correlations between the proportion P
u 1_1n 1 J 1_- . and Gymnotus in all three size classesappeared to t8.PI.Mf"'_U L_. m~. be frequent foragers. To some degree. frequency food ~ of piscivorous feeding may be overestimated by stomach contents. becausecompact bones and stiff Fig. 4. Relativeseasonal availability of fish food resourcesas indicatedby total volumesconsumed by the completeCano fin rays are digested more slowly than soft tissues. Maracafish fauna (total volumesrecorded in liters for each Additionally. different digestion rates for predom- seasonappear in the key; AQ = aquatic.TR = terrestrial). inantly juvenile versus adult prey probably affects the frequency of empty stomachs.Chitinous matter To some degree. seasonaldifferences in the fre- associatedwith invertebrate feeding by smaller pis- quency of empty stomachswere due to ontogenetic civores passesthrough guts more slowly than ver- diet shifts. Although no significant overall effect of tebrate flesh. fish size on frequency of empty stomachs was Diet breadth among piscivoresshowed no signif- noted. the largest fishes had significantly more icant overall effect of season(ANOV A). Mean diet empty stomachs than the smallest size class(Table breadth among piscivores was 4.97 during the wet 5). The feeding rate of large piscivores. such as season.4.72 during the transition season.and 3.89
Tllblt 5. Percent empty stomachs encountered during three seasonsand within three piscivore size classes.
Species Wet Transition Dry Tolal sampl,
Hoplias 10.9 51.3 54.2 389 Charax 2.8 31.8 81.8 29S Pygoct'nlrus 6.4 0 238 S. irrilans 0 4.2 76 S. mt'dini 0 0 0 68 S. rhombt'us 4.3 3.3 SO Gymnolus 1.1 7.4 4.8 24M Rhamdia 0 18.6 29.3 124 Caqut'tia 5.5 .$.1 13.8 370
Mean 3.4 13.4 30.6 F= 3.67.2 DF. P <0.05
Juveniles SubadullS Adults --Species Hop/iDS 6.5 31.2 51.1 Charax 3.6 26.9 35.3 Pygocentrus 0 2.0 2.4 S. i"itans 0 3.8 0 S. medini 0 0 0 S. rhombeus 6.4 0 0 Gymnolus 0 3.0 6.8 Rhamdia 0 12.0 21.2 Caqwlia 0.6 4.8 36.4
Mean 1.9 9.3 17.1 F= 2.87.2 OF. P- 0.07 189
50
40
30
20
10 HMA CGI ~ SIR SME IRK ~ MIA CKR SpecIH 0 0.2 0.4 0.8 0.8 1.0 1.2 50 Fig. 5. Averagediet breadthsof piscivoresduring the wet, transition, and dry seasonsat Cano Maraca (HMA = Hoplias 40 malabaricus. CGI = Clulrax gibbosus. PNO = Pygocmlrus no- tatus, SIR = Se"QSalmusirritans, SME = Se"asalmus medini, SRH = Serrasalmus rlJombeus. GCA = Gymnotus carapo. 30 RHA = RhamdiQ sp., CKR = Caquffla kraussii). -0 20
10 during the peak dry season.The three Se"asa/mus speciesexhibited relatively narrow diet breadths as a result of their fin-nipping specialization. Diet breadths showed a trend of lower values during drier periods for Serrasa/musspp., Hop/ias, Char- ax. and Rhamdia (Fig. 5). Pygocentrusand Caque- ria exhibited trends of somewhat more generalized feeding during drier periods. Gymnotus wasa more generalized feeder during the transition season than either the wet or dry seasons. During the transition season,Gymnotus (31-3
-.. ... E ~2 ; g V s. U N '0 .. c >- 0 U ~ D- i ~ L C . U ~
Fig. 8. Relative proponions of individuals of each speciestaken Spec- from four microhabitat samples during four months of tbe tran- sition season at Cado Maraca (species codes are the sameas in Fig. 7. Mean prey sizes encountered in diets or CaJ\o Maraca Fig. 5). piscivor~s during the transition season. Error ban represent on~ standard deviation. th~ number or pr~y items measuredappean within eachopen bar. and meansror speciesjoined by lines utilization during the transition season (Fig. 8). below the x axis were not significantlydifferent at a = 0.05 Hop/ias and Charax were habitat generalists, oc- according to Duncan"s Multiple Range Test (speciescodes are curring with similar frequencies in all three habitats the same as in Fig. 5). and nocturnally in midpool regions to somedegree. The four piranha species were essentially diurnal, (Pygocenlrus x S. rhombeus). Fin specialists, S. midpool, open-water dwellers. Serrasa/musrhom- irritans x S. rhombeus provided the only transition beusappeared to utilize vegetation/edge habitats to seasonsimilarities exceeding0.31 (tp, = 0.83, 1.19), a small extent. Gymnotus. Rhamdia, and Caquelia whereas Hoplias on Charax (tp, = 0.96) represent- were primarily vegetation dwellers, particularly by ed the only pairing above 0.24 during the dry sea- day. Both Gymnotus and Rhamdia foraged in son. deeper, open-water nocturnally, the former aided by electroreception (Bullock 1982) and the latter by keen olfactory capabilities associated with its Preysize long barbels. Caquelia. large adults in particular, inhabited nearshore/edge habitats in addition to A significant relationship exists between prey size thick stands of aquatic macrophytes (29%). and piscivore species (F = 65.51. 8 OF. In contrast with diet, habitat overlap (p,/-based) P
Although some opportunity for microhabitat spe- cialization probably does exist within the turbid waters of the creek during these periods (e.g., ben- thic versus midlevel or surface), each habitat yield- ed rather consistent mixtures of species in seine hauls. If one considers three niche axes simultaneously (prey type, prey size, habitat), and assumesconser- HMA CGI PNO SIR SME SRH
HopI/..nl8/.urlcu. 15 - - 0- - ~Ut«TY Co aULD ~ . > 10 0
.. 'ij
0 0
.. 8 -0 I-
We' T'8n8111on 15
10 .Q. ~. 0> . ~ 0 Wet r'8n8ltlon .. "0 GynwlOlu8 Qrapo I-
S."...,- /nt,.,. 15 A. ~ ~ > 0
.. T'8naltlon :; 5
'i ~~~~~~04 "0 8~~~~~~~~-~~~~~~-~~.. I- 0 Wet Trlnlillen
s.n.sa/mus ,ho~s
Do ~ . > 0
i 'i .. 0 I-
We' Transition
Fig. 10. Total diet overlap exhibited by each piscivore species during different seasons at Caiio Maraca. Total community diet overlaps are represented by open symbols. and total piscivore guild overlaps are plotted as closed symbols. 193 during the transition season, both within the pisci- Werner & Hall 1979. Machado-Allison & Garcia vore guild and the community at large (Fig. 10). In 1986.Keast 1985b). Morphological changesassoci- the casesof Gymnotus, Caquetia. and Serrasa/mus ated with greater trophic specialization in the form medini. collective guild overlaps were low relative of piscivory should reduce the efficiency of mi- to overall community overlaps during both the wet croinvertebrate feeding. Following this reasoning. and transition seasons.Rhamdia exhibited a slight Griffiths (1975) recognized a dynamic trade-off be- increase in total guild overlap between the wet and tween maximizing prey number versusprey energy transition seasons, although overall community in the feeding strategies exhibited by piscivorous overlap declined. yellow perch. Perca flavescens, during ontogeny.
Discussion Resource partitioning
Ontogeny and diet Food resource partitioning was widespread within the Cafio Maraca piscivore guild, particularly dur- All nine sympatric, tropical piscivores showed dis- ing the transition season. Given the high concord- tinct shifts from invertebrate feeding by small juve- ance between ontogenetic diet shifts, prey avail- niles to primarily fish feeding by subadult and adult ability, and seasonal rainfall, can high levels of size classes.Predominance of fin nipping by sub- resource partitioning be interpreted in relation to adult and adult size classesof serrasalmine pira- biological interactions, especially interspecific nhas is consistent with earlier findings (Goulding competition? Three general, alternative explana- 1980, Machado & Garcia 1986, Northcote et al. tions could account for observed patterns of re- 1986, 1987, Sazima & Zamprogno 1986, Nico & source exploitation. First, resource exploitation Taphorn 1988).Three primary factors probably are could occur at random. Yet sufficient subdivision involved in producing size-related patterns of feed- of available resources was observed among the ing among piscivores. First, juvenile fishes are con- nine piscivores as to render this null hypothesis strained by their small size to exploit relatively unlikely. For example, average seasonaldiet over- small food particles (Werner & Gilliam 1984,Keast laps were considerably lower (q>= 0.18- 0.25) 1985a, 1985b). Only following a period of initial than those obtained for series of randomly con- growth can whole-fish swallowers switch to larger structed 100speciescommunities (q>= ca. 0.75, fish prey. Second, availability of small zooplankton Pianka et al. 1979). Second, contemporary patterns and benthic invertebrates (e.g., aquatic insect lar- of niche partitioning could be founded in genetical- vae) was much greater during the wet seasoncom- ly-based morphological and behavioral traits that pared to either the transition or dry seasons at evolved in response to past environmental condi- Caiio Maraca (Fig. 4). Observed switching from tions that differed from current conditions (sensu invertebrate to fish prey largely coincided with Connell 1980). This second hypothesis contrasts changing food availabilities as indicated by feeding sharply with a third possible explanation. that con- performance of the entire fish assemblage.Where- temporary patterns of resource subdivision are as total aquatic primary production and aquatic adaptive in the context of present day biotic inter- invertebrate availability declined during the gradu- actions. Interspecific competition is foremost al desiccation of the floodplain, fish densities in- among biotic mechanisms cited in studies of com- creased markedly. Jackson (1961) made a similar munity resource partitioning, although the role of observation for African fishes in seasonalaquatic predation has recently received greater attention, habitats. Third, many predatory fishes exhibit particularly with respect to habitat selection (Fras- changesin relative body proportions and other an- er & Cerri 1982, Sib 1982, Werner et al. 1983, atomical traits during growth that are associated Mittlebach 1984, Power 1984. Power et al. 1985. with greater feeding specialization (Werner 1974, Hobson & Chess 1986). 194
In recent reviews of resource partitioning in low- the late wet and dry seasonsin floodplain ecosys- er vertebrates, Toft (1985) and Ross (1986) in- tems. During this 'crunch period', resources be- dicated that causal factors responsible for differ- come increasingly limited (especially resources re- ences in resource use are rarely known. MacNally lated to space) as local stocks of aquatic organisms (1983) and others have argued that descriptive field become denser within shrinking bodies of water. studies alone cannot demonstrate competition. Biotic interactions, especially fish predation, in- Whereas field experiments potentially can demon- tensify at this time, but are piscivores necessarily strate interspecific competition (Schoener 1983), resource-limited? Whereas apparent availability of they are beset with problems of interpretation as fish prey may be high during the transition season, well. Positive, negative, or even a complete lack of niche segregationand trophic specialization among population response to experimental manipula- piscivores should OCCl'rfor two reasons. First, fol- tions potentially are confounded by effects of in- lowing optimal foraging theory (Krebs 1978, Pyke direct interactions (Bender et al. 1984, Tilman 1984), scarcity of invertebrate prey and temporary 1987) or measurement of inappropriate variables abundance of prey fishes would favor exploitation (Diamond 1983, MacNally 1983).Even though de- of the most profitable forms by predators, basedon scriptive studies cannot demonstrate competition their morphology and innate components of feed- directly, they often provide strong circumstantial ing behavior. Profitability would be expressedhere evidence with the potential to support or reject the in terms of the ratio of energy gained by maximiz- hypothesis of interspecific competition. The infer- ing the rate of prey capture, divided by the total ential value of this circumstantial evidence depends energy cost associated with searching, pursuit, upon the strength of comparisons based on inter- handling, digestion, and assimilation. Evolution- population or temporal and spatial environmental ary divergence of prey escape tactics increases re- variation. quirements for specialization on a restricted num- In this study, trophic specializationswere associ- ber of prey species(Rand 1967). Second, the brev- ated with lower diffuse diet overlap during the ity of the seasonof high fish density, together with transition season when fish population densities the high risk of mortality due to predation and dry were highest and availability of invertebrate prey season dessication/anoxia, conveys a tremendous was reduced. Due to the general pattern of in- fitness advantageto individuals that maximize their terspecific synchrony in reproduction with the on- intake of temporarily abundant prey, then survive set of the wet season, ontogenetic shifts toward the ensuing dry season. A disproportionate frac- more specialized piscivore diets largely corre- tion of the next generation would be founded by sponded with the progression of the transition sea- efficient piscivores that are larger and more fecund son and fish growth. These observations are consis- (or more effective in providing parental care) than tent with McKaye & Marsh's (1983)hypothesis that less efficient conspecifics. In effect, a major frac- African cichlid fishes switch from more opportunis- tion of the annual caloric intake of tIanos piscivores tic, generalized diets to more specialized feeding is concentrated into the approximately four-month during periods of relative resource scarcity (i.e., transition season. Relative to other seasons, following the 'niche overlap hypothesis' of Pianka growth and visceral fat deposition is maximal for 1974). Morphological adaptations for specialized most speciesduring the transition season(personal feeding may be selectively advantageousduring a observation). limited period during the annual cycle of ecological Given the selective premium that is placed on events. At Caito Maraca, four speciesthat switched highly efficient foraging by subadult and adult pis- to piscivory at small sizesalso exhibited low collec- civores during the transition season,it appears like- tive diet overlaps throughout the year (Fig. 10). ly that observed trophic specializations are adap- In summarizing work involving tropical fish com- tive in reducing the negative impact of diffuse com- munities, Lowe-McConnell (1987) suggested that petition within this rich fish assemblage.For com- competition is potentially most important during petition to occur, resources must be in limited~ 195 supply and contested by two or more populations. mass in stomach contents was estimated from The presence of numerous heterospecific pisci- length/massrelationships. Both of these factors in- vores during the transition seasonof high fish den- crease risks of sampling error. Third, and most sity could restrict foraging to the most profitable seriously, the taxonomic identities of the eight fish prey types. Reductions in pairwise diet overlap species are confused (see Taphorn & Lilyestrom during periods of reduced food availability have 1985for listing of regional ichthyofauna). The only been shown in several earlier studies of fish com- unconfused species name, Roeboides dayi (Stein- munities (see Ross 1986).Though basedon limited dachner), is perhaps the most suspect of the eight. sampling, reduced diet overlap among fishes was Prejs & Prejs (1987) found 99.9 and 100% vegeta- reported by Zaret & Rand (1971) and Greenfield et tion and detritus in the stomachs of R. dayi during al. (1983) during the dry seasonin Central Amer- the dry season, whereas fish scales comprised 76 ican streams. Alternatively, lowe-McConnell (N = 121) and 25% (N = 49) of the diet of R. dayi (1964), Knoppel (1970), and Goulding (1980) at Calio Maraca during the transition and dry sea- stressed high dietary overlap in South American sons respectively. Scalesformed 24 (wet) and 15% fish communities, although no attempts were made (dry) of the diet of R. dayi by volume at Calio to study trophic guilds quantitatively over time. Volcan, a small piedmont stream (N = 313,Wine- The findings of the current study contrast sharply miller 1987a). All other Roeboides species, for with those of Knoppel (1970), who interpreted his which data are available, specialize on scales in data as indicating a perplexing lack of correspond- addition to aquatic insects (Sazima 1983, Wine- ence between morphological characteristics and miller 1987a). Most of the fishes referred to as 'R. diets of fishes from five stream collections in cen- dayi' by Prejs & Prejs (1987) were probably Cre- tral Amazonia. Also in opposition to the current nobrycon spilurus (Valenciennes), a morpholog- findings was Knoppel's (p. 346) contention that ically similar, omnivorous characid that is extreme- 'stomach contents according to age of fishes did not ly common in shallow, productive environments merit close consideration'. Quite the contrary, throughout the llanos. Knoppel probably did not examine stomachs and Some temperate zone studies also have demon- size classesclosely enough to discern the obvious strated greater diet segregation among fishes dur- patterns. ing periods of reduced food availability (Nilsson Recently, Prejs & Prejs (1987) observed higher 1955, Harrington & Harrington 1961, Thorman dietary overlap during the dry seasonamong small, 1982). In addition to diet segregation, interspecific vegetation-dwelling fishes in seasonal pools of the size and time related differences in habitat usehave Venezuelan low llanos. These authors interpreted been demonstrated for both temperate and tropical patterns of diet overlap as ecologically unimpor- fish assemblages(Werner & Hall 1977, Moyle & tant, because their data showed reduced feeding Senanayake 1984, and see Ross 1986 for others). rates or a switch to detritivory during the dry sea- Habitat segregation by Calio Maraca piscivores son (apparently, the dynamic transition seasonof probably was as much a function of predator avoid- the llanos was not sampled). Lower availability of ance as food resource segregation. Fishes used aq- insect prey and increasedpredation pressure were uatic vegetation as the principal form of cover for believed to be the primary factors accounting for avoidance of predation by dense schools of red- reduced dry season feeding; however, results ob- bellied piranhas, Pygocentrus noralus. Additional- tained by Prejs & Prejs (1987) must be interpreted Iy, and perhaps to a lesser extent, smaller fin-nip- with caution for several reasons. First, their com- ping Serrasalmusspecies posed a threat to diurnal, munity subset of eight small, vegetation-dwelling open-water dwellers. Where it is common, Pygo- fishes does not represent a trophic guild, but rather centrus may have a pervasive effect on the spatial a diverse subset of small vegetation-dwelling fish- structuring of fish communities, since only highly es. Secondly, total sample sizes for stomach con- cryptic, heavily armored, or fast schooling fishes tent analysis were small (N = 9-30), and prey bio- were generally taken from deep open-water re- 196 gions of Cano Maraca and other ilanos streams the Mato Grossoregion of westernBrazil (I. Sazi- during daylight. Nocturnal predators (e.g. Rham- ma, personalcommunication). Further field and dia and Gymnotw) leave the dense mats of aquatic laboratorystudies are neededto sort out the niche vegetation to forage in the open water habitat un- relationsof fin-nipping piranhasmore accurately. der cover of darkness. Approximately 80 man- hours of nocturnal hook-and-line fishing at a stream in the low llanos of Apure state (location of Conclusions Cano Maporal given in Nico & Taphorn 1988) yielded nine speciesof large piscivores (unpublish- High fish speciesdiversity in the neotropics is asso- ed data). Diurnal hook-and-line fishing over the ciated with impressive morphological and ecolog- same period yielded 13 large piscivorous species, ical diversification (Roberts 1972, Fink & Fink yet only three species occurred in both samples 1979, Goulding 1980, Sazima 1986, Lowe-McCon- (e.g. Pygocentrus notatus, Pseudoplalystomalas- nell 1987). Coexistence of more than 100 fish spe- ciatus, and Plagioscion squamosissimw). Sixty sev- cies in some llanos creeks (Taphorn & Lilyestrom en percent of the nocturnal sample was comprised 1985) may be due in part to relaxed interspecific of siluriforms (catfishes and knifefishes), whereas competition during horizontal flooding and the 92% of the diurnal sample belonged to the orders bloom of primary production that occurs during the Characiformes (piranhas, tetras, and related wet season. Diets of five of the nine piscivores forms) and Perciformes (cichlids and one fresh- examined overlapped more with the rest of the fish water drum). Furthermore, most diurnal piscivores community during this period than any other. were taken near cover at the stream margin, fur. Large diffuse overlap during the wet seasonlargely ther supporting the hypothesis that Pygocentrus resulted from heavy exploitation of abundant mi- restricts diurnal use of the open-water, midpool crocrustaceaand aquatic insect larvae by juveniles. habitat by other piscivores. Lower collective overlap during the transition sea- The three fin-nipping piranhas exhibited the son probably was derived from complex interac- highest similarities in diet and habitat niche para- tions among several factors, including: (1) reduced meters at Cano Maraca. As might be expected due availability of juvenile food resources, (2) fish to genetic and morphological similarities, closely growth and ecomorphological trade-offs favoring related species frequently exhibit relatively less diet shifts, (3) increased fish densities and diffuse ecological segregation than more distant taxa. competition favoring maximal rates of prey exploi- Lack of diet segregation among fin-nipping pira- tation, (4) differential prey profitability due to nhas, especially Serrasalmw irritans and S. rhom- trade-offs associatedwith prey escape tactics, and beus subadults, was due in part to my inability to (5) the threat of predation by piranhas restricting identify fin fragments at the species level in most diurnal foraging by other piscivores in open-water instances. Records of the incidence of fin nips on microhabitats. In highly diverse communities, dif- preserved Cano Maraca specimensclearly indicate fuse overlap with numerous niche neighbors repre- that not all fishes are equally vulnerable to Serra- sents a greater potential negative effect than isolat- salmw attacks (unpublished data). Preliminary ed pairwise interactions. tests of aquarium-housed juveniles and subadults suggest that Serrasalmw species possessdistinct patterns of foraging behavior associatedwith either Acknowledgements stealth in open water (S. irritans; see Nico & Ta- phorn 1988for a brief description of field observa- I thank D.C. Taphom for his support and hospital- tion), persistent chasing with frequent remaneu- ity in Venezuela. The field assistanceof D.C. Ta- vering (S. medini), or use of cover for ambushing phom, L. Nico, A. Barbarino, and N. Greig was (S. rhombew). Similar behavioral patterns were greatly appreciated. I thank P. Urriola, R. Schar- observed among three syntopic piranha species in gel, F. Mago-Leccia, and C. Hubbs for assistance 197 in obtaining permits and visas. The Direcci6n Ad- African lakes. Cornell University Press. Ithaca. 8J9 lIP. ministracion y Desarrollo Pesquero of the Repub- Griffiths. D. 1975. Prey availability and the food of predators. lic of Venezuela provided a collecting permit. I am Ecology 56: 1209-1214. grateful to D. Urriola for kindly allowing me to Goulding. M. 1980. The fishes and the forest: explorations in Amazonian natural history. University of California Press. work on his property. This study was derived from Los Angeles. 21M)pp. a doctoral dissertation submitted to the Depart- Harrington. R.W. & E.S. Harrington. 1961. Food selection ment of Zoology of the University of Texas in among fishes invading a high subtropical salt marsh: from partial fulfillment of the requirements for the onset of ~ng through the progress of a mosquito brood. Ecology 42: 646-«16. Ph.D. I thank my advisors E.R. Pianka and C. Helfman. G.S. 1978.Patterns of community structure in fishes: Hubbs for their support. Funds were provided by a summary and overview. Env. Bioi. Fish. 3: 129-148. research grant from the National Geographic So- Hobson. E.S. & J.R. Chess. 1986. Relationships among fishes ciety and an NSF predoctoral research grant. and their prey in a nearshore sand community off southern California. Env. Bioi. Fish. 3: 201-226. Hyslop. E.J. 1986.The food habits of four small-sizedspecies of Mormyridae from the floodplain pools of the Sokoto-Rima Referencescites river basin. Nigeria. J. Fish Bioi. 28: 147-151. Jackson. P.B.N. 1%1. The impact of predation. especially by Angermeier, P.L. '" J.R. Karr. 1983. FIShcommunities along the tiger-fish (Hydrocyon vinalllS Cast.) on African fresh- environmental gradients in a system of tropical streams. Env. water fishes. PrOto Zool. Soc. Lond. 136: M3-622. Bioi. Fish. 9: 117-135. Keast. A. 1985a.The piscivore feeding guild of fishes in a small Araujo-Ulna. C.A.R.M., L.P.S. Portugal '" E.G. Ferreira. freshwater ecosystem. Env. Bioi. Fish. 12: 119-129. 1986. FISh-macrophyte relationship in the Anavilhanas Ar- Keast. A. 1985b. Development of dietary specializations in a chipelago. a black water system in Central Amazon. J. Fish summer community of juvenile fishes. Env. Bioi. Fish. 13: Bioi. 29: 1-11. 211-224. Bender, E.A.. T.J. Case & M.E. Gilpin. 1984. Perturbation Knoppel. H.-A. 1970.Food of Central Amazonian FIShes:con- experiments in community ecology: theory and practice. tribution to the nutrient-ecology of Amazonian rain-forest- Ecology 65: 1-13. streams. Amazoniana 2: 257-352. Brooks. J.L. &. S.J. Dodson. 1965. Predation. body size and Krebs. C.J. 1978. Ecology: the experimental anaJysisof distri- compostition of plankton. Science 150: 28-35. bution and abundance. 2nd ed. Harper & Row. New York. Bullock. T.H. 1982.Electroreception. Ann. Rev. Neurosci.5: 678 pp. 121-170. Kullander. S.O. 1983.A revision of the South American cichlid Connell, J.H. 191KJ.Diversity and the coevolution of compet- genus Cich/atonw (Teleostei: Cichlidae). Swedish Mus. Nat. itors, or the ghost of competition past. Oikos 35: 131-138. Hist.. Sttx:kholm. 296 IIp. Diamond. J.M. 1983. Laboratory, field and natural experi- Lawlor. L. R. 1~. Overlap. similarity. and competition coeffi. ments. Nature 304: 586-587. cients. Ecology 61: 245-251. Fink. S.V. '" W.L. Fink. 1981. Interrelationships of the diel activity. food habits and spatial distribution of juvenile Sale. P.G. 1979. Habitat panitioning and competition in fISh creek chub. ~moli(us atromacu(alus. Env. Bioi. Fish. 11: communities. pp. 323-331. In: H.E. Oepper (ed.) Predator- 301-307. Prey Systemsin Fisheries Management. Spon Fishing Inst.. Mago-Leccia. F. 1970. Preliminary studies on the ecok>gyof Washington. D.C. fishes in the ilanos. Acta Bioi. Venez. 7: 71-102. (In Spanish). Sazima. I. 1983. Scale-eating in characoids and other fishes. McKaye. K.R. & A. Marsh. 1983.Food switching by two spe- Env. Bioi. Fish. 9: 87-101. cialized algae-scrapingcichlid fIShesin Lake Malawi. Africa. Sazima. I. 1986.Similarities in feeding behavior between some Oecologia 56: 245-248. marine and freshwater fIShesin two tropical communities. J. Mittlebach. G.G. 1984.Predation and resource partitioning on Fish Bioi. 29: 53-65. two sunfishes (Centrarchidae). Ecology 65: 499-513. Sazima. I. & C. Zamprogno. 1985. Use of water hyacinths as Moyle. P.B. &. F.R. Senanayake. 1984. Resource pamtioning shelter. foraging place. and transpon by young piranhas. among the fIShes of rainforest streams in Sri Lanka. J. of ~muGlmus .Jpilop/~ura. Env. Bioi. Fish. 12: 237-240. Zool. London 202: 195-223. Schoener. T.W. 1974. Resource partitioning in ecological com- Nico. L.G. &. D.C. Tapbom. 1988.Food habits of piranhas in muniajes. Science 18S:27-39. the low llanos of Venezuela. Biotropica 20: 311-321. Sdtoener. T.W. 1983.Field experiments on interspecific compe- Nilsson. N.A. 1955. Studies on the feeding habits of trout and tition. Amer. Nat. 122: 240-285. char in Donh Swedish lakes. Rep. Inst. Freshw. Res. Drott- Schut. J.. S.S. deSilva & K. Konmulder. 1984. Habitat, associ- ningholm 36: 163-225. ationsand competiajonof eight Barbus (= Punlius)species Nonhcote. T.G.. R.G. Nonhcote &. M.S. Arcifa. 1986. Differ- (Cyprinidae) indigenous to Sri Lanka. Netherlands J. ?ro. ential cropping of the caudal fin lobes of prey fisbes by the 34: 159-181. piranha. S~"a.salmusspilopltUra Kner. Hydrobiologia 141: Sih. A. 1982.Foraging strategies and the avoidance of predation 199-205. by an aquatic insect. Nolon«ta hoffmanni. Ecology 63: 786- Nonhcote. T.G.. M.S. Arcifa & O. Froehlich. 1987.Fm-feeding 796. by the piranha (S~~aImus spilopl~ura Kner): the cropping Simpson. E.H. 1949. Measurement of diversity. Nature 163: of a novel renewable resource. Proc. 5th Congr. Europ. Ich- 688. thyol. Stockholm 1985: 133-143. Siobodcbikoff. C.N. .It W.C. Schulz. I~. Measures of niche Pianka. E.R. 1974. Nidle overlap and diffuse competition. overlap. Ecology 61: IOSI-IOS5. Proc. Nat. Acad. Sci.. USA 71: 2141-2145. Strong, D.R.. A. Szyska & D. Simberloff. 1979. Tests of com- Pianka. E.R. 19m. Guild structure in desen lizards. Oikos 35: munity-wide character displacement against null hypotheses. 194-201. Evolution 33: 897-913. Pianka. E.R.. R.B. Huey &. L.R. Lawlor. 1979. Niche segre- Taphom. D.C. .It C.G. Ulyestrom. 1985. Fishes of the Fernan- gation in desen lizards. pp. 67-115./n: D.J. Horn. R. Mitch- do Corrales module. Results of the ichthyological investiga- eU. &. G.R. Stairs (ed.) Analysis of Ecological Systems.Ohio tion of project CONICIT.PIMA-18. Rev. Cie. Tccn.. UN- State University Press. Columbus. ELLEZ 3: 55-8S. Power. M.E. 1983. Grazing responsesof tropical freshwater Thorman, S. 1982.Niche dynamics and resource panitioning in fIShesto different scalesof variation in their food. Env. Bioi. a fish guild inhabiting a shallow estuary on the Swedishwest FISh. 9: 25-37. coast. Oikos 39: 32-39. Power. M.E. 1984. Depth distribution of armoured catfish: Tilman, D. 1987.The imponance of mechanismsof interspecific predator-induced resource avoidance? Ecology 65: 523-528. competition. Amer. Nat. 129: 769-774. Power. M.E.. W.J. Matthews &. A.J. Stewart. 1985. Grazing Toft. C.A. 1985.Resource panitioning in amphibians and rep- minnows. piscivorous bass and stream algae: dynamics of a tiles. Copeia 1985: 1-21. strong interaction. Ecology 66: 1448-1456. Wallace. R.K. 1981. An assessmentof diet-overlap indexes. Prejs. A. & K. Prejs. 1987. Feeding of tropical freshwater Trans. Amer. Fish. Soc. 110: 72-76. fishes: seasonality in resource availability and resource use. Watson. D.J. .It E.K. Balon. 1984. Ecomorphological analysis Oecologia 71: 397-404. of fIShtax Werner. E.E.. J.F. Gilliam. D.J. HaU & G.G. Mittlebach. Winemiller. K.O. 198Th. Feeding and reproductive biology of 1983.An experimental test of the effects of predation risk on the currito. Hoplostt'mum littoralt'o in the Venezuelan ilanos habitat use in fish. Ecology 64: 1540-1548. with comments on the possible function of the enlarged male Werner. E.E. & D.J. Hall. 1977.Competition and habitat shift pectoral spines. Env. Bioi. Fish. 20: 219-227. in tWo sunfishes (Centrarchidae). Ecology 58: 869-876. WinemiUcr. K.O. 1989. Development of derrnallip protuber- Werner. E.E. " D.J. Hall. 1979.Foraging efficiency and hab- ancesfor aquatic surface respiration in South American char- itat switching in competing sunfishes. Ecology ti>: 256-264. acid fishes. Copeia (in press). Winemiller, K.O. 1987a.Tests of ecomorphological and com- Zaret, T.M. & A.S. Rand. 1971.Competition in tropical stream munity level convergence among neotropical rlsh assemblag- fishes: SUPf'Onfor the competitive exclusion principle. Eco- es. Ph.D. Dissertation, University of Texas. Austin. 370 pp. ~ 52: 336-342.