Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Biological Dynamics of Forest Fragments Project wikipedia , lookup
Occupancy–abundance relationship wikipedia , lookup
Theoretical ecology wikipedia , lookup
Behavioral ecology wikipedia , lookup
Maximum sustainable yield wikipedia , lookup
Molecular ecology wikipedia , lookup
(k(:i)logia D Springer-Verlag 1989 :>ecologia (1989) 81 :225-241 Patterns of variation in life history among South American fishes in seasonalenvironments Kirk O. WillelDiUer Departmentof Zoology, The University of Texas,Austin, TX 78712.USA s..mary. Ten traits relatedto life history theory weremeasured or estimatedfor 71 freshwaterfish speciesfrom two locations in the Venezuelanllanos. Multivariate statistics and cluster analysis revealedthree basic endpoint patterns bounding a two-dimensionalcontinuum. A suite of attributesassociatedwith parentalcareand aseasonalreprodo<:tion appearedto correspond to an equilibrium strategy. A secondgroup of small fISheswas distinguishedby traits associatedwith rapid colonizing ability: early maturation. continuous reproduction, and small clutches. The third basic pattern was associatedwith synchronizedreprodo<:tion during the early wet season.high f~undity, absence of parental care. and breeding migrations. A subset of mostly small fishes exhibiting little or no parental care. small clutches.and two to four month reproductiveseasons wasintermediatebetw~ the opportunistic (rapidly colonizing) and seasonalstrategies.All ten life history variables showedsignificant effectsof phylogeny.The cluster of speciescorrespondingto the equilibrium group wasdominated by siluriform fishes and perciforms of the Cichlidae. The opportunistic cluster was dominated by cyprinodontiform and characiform fishes.whereasthe seasonalcluster contained primarily characiform and siluriform fishes. Seven of nine traits weresignificantly correlatedwith body length. The three reproductive patterns are interpreted as being adaptativewith respectto relative intensity and predictability of temporaland spatialvariation in abiotic environmental parameters.food availability. and predation pressure. Keywonls:Llanos- Reproduction- Strategies- Tropical fishes-Venezuela The diversity of methods by which organisms reprodtx:e has long intrigued naturalists. One can argue that traits associatedwith reproduction should be subject to intense natural selection, as these directly affect the individual's geneticrepresentationwithin subsequentgenerations.Alternative modes of reproduction have obvious demographic implications,sinceall populationsexperienceeither gradual or periodic turn-overs.Alternative patternsof reproduction and the degreeto which these are s~ful in different environmentalsettingshaveformed the basisfor both theoretical and empirical researchunder the headinglife history (Steams 1976; Southwood 1977; Horn 1978; Sibley and Calow 1986a).In the most basic terms, life history theory involvesconstraints,or trade-offs,among reproductiveand demographic parameters (Steams 1983a; Reznick 1985; Peaseand Bull 1988).For example, models have incorporated negativecorrelations betweencurrent investmentand future investments in reproduction, juvenile survivorship and clutch size. reproductive effort and adult survivorship, and meangenerationtime and the intrinsic rate of increase. Alternative life history predictions can be testedby evaluating patterns of covariation among reproductive and demographic parameters among heterospeciflCpopulations or conspecificdemesthat have experienceddifferent environmental conditions for long time periods. The presentstudy constitutes one such test. characteriDng patterns of reproduction in tropical fishes with a seriesof dependentvariables.The degreeof speciesclustering in multivariate space indicates the likelihood that certain combinations of traits will occur together in nature. Finally, I evaluate species clustersfor associationwith phylogeny and ~Iogical factors. Tropical freshwaterfishesexhibit large diversity in morphological, physiological. and ~logical attributes (LoweMcConnell 1987).Despitefew comparativestudiesof tropical fish reproductive strategiesto date (Africa-Welcomme 1969; Central America-Kramer 1978; Asia-DeSilva et al. 1985),thesediverse fish assemblagesprovide excellentsystemsfor evaluation of life history patterns. The great drainagebasinsof South America harbor the most diversefreshwater fish assemblages on earth (ca. 2400describedspecies). Rather paradoxically, this ichthyofauna is derived from only a few basic stocks (speciesbelonging to the Characiformes.Siluriformes. and Cichlidae (Perciformes)comprise about 93% of all species,Lowe McConnell 1987).I investigated fish reproductive biology in the llanos of Venezuela during twelve co~tive months in 1984.The region has an averageannual temperatureof 26.2°C and a highly seasonal pattern of rainfall, with 1189mm of precipitation occurring from June-Novemberand 178mm from DecemberMay in 1984. Despite extreme changesin environmental conditions caused by seasonalrainfall. fishes exhibited a wide rangeof reproductivecharacteristicsin the llanos. This report describesthesecharacteristicsfor 71 speciesand interprets the basicpatterns within an ~logical context.~ Methods Collections During every month of 1984, fishes were collected from two locations in the state of Ponuguesa, Venezuela. The 226 servedin 10% fonnalin and depositedin the MHN (UNELmore faunistically diversesite, Cano Maraca, is a swampLEZ) and the TNHC (TMM). creek located in the western llanos region (8°52'30" Lat. N; 69°53'30" Long. W). The area studied (termed an "estero") lies within low, flat terrain that experiencesextensive Thedata sheet flooding during the wettest months (June-August). Severalattributes related to reproduction and population During this time, the creek'sbroad floodplain is converted structureweremeasureddirectly and other life history traits from exposed,sun-bakedsoil and thorn-scrub habitat into werecalculatedor estimatedfrom these.All preservedspeca productive marsh. The aquatic habitat is reduced to a imens were identified and measured for standard length network of pools (averagedepth ca. 1.0m) during the driest months(December-May).Dissolvedoxygenconcentrations (SL). When available, 30 specimensof each speciesfrom arereducedduring this time, and many fishesrely on special eachmonthly collection at both Calio Maraca and C. V 01can were dissected.The condition of the gonad was coded respiratory adaptationsfor survival (Winemiller 1988). basedin its relative sizeand color using the following cateThe other site, Cano Volcan (8°59'15" Lat. N; gories:clear (1), translucent(2), opaque(3), small (1), medi69°53'30" Long. W), is a third order stream in the lowest um/small (2), medium (3), medium/large (4), and large (5). tier of the Andean Piedmont. This narrow stream flows The overall gonad code for each individual was equal to through deciduousforest and contains both pool and rime (color code+size code) divided by two, and as a result, habitats. Cano Volcan differs from C. Maraca in having a slightly steepergradient,more stabledry seasondischarge, values ranged from 1.0 to 4.0. Although the same basic and a more forested watershed.Cano Volcan experiences criteria were usedin coding male and femalegonads,intersexual and interspecific differencesin the size, color, and frequent but brief flash floods during wet seasondowntexture of fully mature gonads were taken into account. pours. Physico-chemicaldata for each site during the year Undevelopedovarieswere tiny transparent, globular or lastudiedare given in Winemiller (1987). mellar structures.Following a gradual transition, fully-deFishes were collected by dipnet, seine (3.2 mm mesh/ veloped fish ovaries were large opaque-yellow or orange 2.5 m length; 12.7mm/20 m), experimental gillnet, and structureshaving a grainy appearancedue to the presence hook and line. At eachsite, an attempt was madeto sample of mature oocytes. Mature ovaries are cylindrical or lie the entire fish community such that the sample for each as sheetsalong the lateral walls of the abdominal cavity. speciesreflectedits relative abundanceand population size structureduring eachmonth. The collectingeffort expended Immature testesappearedas transparentthreadlike or minute lamellar structures in contrast to the smooth, opaque, each month was approximately equal for each site, conmilky-white appearance of mature testes. Fully-mature sisting of three or four hauls of the 20 m seine in open testeswere either cylindrical. lamellar. or multilobed (the water habitats,and alternateuseof the 2.5 m seine,dipnets, latter condition occursin catfishesof the Pimelodidae).For and gillnets for a minimum of 6 h within open water, shoreseveralspecies,gonad codeswere regressedagainstgonadoline, vegetation,and peripheralpool habitats. Samplingwas somatic indices to test their reliability as overall indices terminated when two hours of collecting yielded no addiof gonadal development(Fig. 1). tional rare species.Collectionsusedfor comparisonsof relaDiameters of ten of the largest oocytesin each mature tive population densitieswere made during either one or ovary were determined using a dissectingmicroscopeand two days betweenthe 11th and 28th of each month. Addian ocular micrometer. Oocyte sizeswithin each examined tional collections were made on other dates in order to ovary wereclassifiedasextremelyuniform, moderatelyunisupplementthe numbersof uncommonspeciestaken in the form, moderately variable, or extremely variable. The standard samples.Except for very abundant species(for number of separateoocyte sizeclassespresentin eachmawhich a subsamplerepresentativeof the abundancerank ture ovary was recorded whenever these were essentially in the collection wasretained),all collectedindividuals were non-overlapping. The fat content of the body cavity was preservedin 15% formalin to assurepreservationof viscera and stomach contents. Following examination, preserved coded using the following criteria: 1= none, 1.5= tracesof fat in connective tissue of the coelomic cavity lining, 2 = specimensweredepositedin the Museo de Historia Natural small amounts around visceraand traces in the connective de la Universidad de los Llanos OccidentalesEsquiel Zatissueof the coelomiccavity lining, 2.5= moderatedeposits mora, Guanare, Portuguesaand the Natural History Colaround the visceraand a thin layer on the coelomic cavity lection of the TexasMemorial Museum, University of Texlining, 3 = large deposits around the viscera and coelomic as, Austin. cavity lining, but not filling the coelom, 3.5= very large During March, June, July, September,and November deposits filling the coelom. but not producing a visible of 1984,severalcollectionsweremadeat RepresaLes Majabulge, and 4 = coelom packed with fat depositsproducing guas, a 4250ha reservoir in western Portuguesa(9°40'00" a bulge of the belly region. Mean standard lengths, ratio Lat. N; 69000'00" Long. W). Fishes were collected from of immature to adult SL's, mean gonad codes,and mean the reservoir by seine (3.2 mm/2.5 m), dipnet, and hook fat codes were plotted for common speciesby month for and line. Snorkeling observationsprovided an additional estimation of the duration of breeding seasons.Immatures meansfor determiningthe presenceof fishesin clear regions weredefinedasstandardlengthsfalling below the minimum of the reservoir.Two regionswere sampled: a narrow arm length for which a fully gravid individual was observed of the lake directly below an inflow canal (from the Rio among all collectionsof a given species. Cojedes),and the main body of the reservoir, both near Ten variables related to life history theory were either and offshore. Physico-chemicalconditions within the main measured,coded, or estimated for 59 speciesfrom Caiio body of the reservoir are relatively constant, while the Maraca and 12 speciesfrom Catio Volcan. former site experiencesradial changesassociatedwith sea1) The estimate of annual population density fluctuation sonal rainfall (e.g. turbidity, depth, and flow rate all inwasequal to the coefficient of variation (CV) of the populacreaseduring the wet season).Voucher specimenswerepre- 227 >< W Q ~ )( W Q U ~ 0( ~ 0 II) 0 Q 0( Z 0 ~ U ~ C ~ 0 In ~ g C Z 0 ~ Astyanaxb/msculatus o.a 0'» >< w 0 ~ u ~ C 2 0 II) 0 0 C Z 0 " 0.8 )( 0.- ~ UI 0 u i= c ~ 0 CI) 0 0 C Z 0 " 0.15 0.10 0.- Rhamdis.,,1 0.15 0.10 ~ a.0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 0.1 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 GONAD CODE GONAD CODE F"tg.1. Linear regressions of gonadosomatic index with gonad code for two tetras and two catfishes from the Venezuelan llanos. Statistics for each regression are as follows: Bryconamericus beta (females) r=0.76. F23.1= 31.1, P<O.(xx)I; B. beta (males) r=O.60, F23.1= 13.2, P<0.OO1; Astyanax bimacuJatus (females) r=0.86, F23.1=66.3, P<0.(xx)1; A. bimacuialus (males) r=0.80. F23.1=41.0, P<0.(xx)1; Ancistrus triradiatus (females) r=0.86, F1s.1=52.0, P<O.(xx)I; A. triradialUS (males) r=0.97, F10.1=145.7. P<0.OOO1; Rhamdia sp.l (females) r=0.95. F 19.1= 139.9, P<0.OOO1; Rhamdia sp.l (males) r=0.95. F 10.1= 101.7, P<0.(XX)1 tion sampleN's from the 12 standard monthly coUections at eachsite. 2) Mean generation time was estimated as the average number of months from fertilization to the first bout of reproduction by new adults. Estimateswere basedon the breedingseason,monthly sizedistributions, and minimum size of maturation for each species.For example, mean generation time for Curimata argentea was estimated as approximately 12 months, since most reproduction occurred during the first month of the wet seasonand maturation required a period wen in excessof the three months during which offspring werecoUected(Fig. 2). Averagegeneration times of more continuously breeding speciessuch as Hoplias malabaricus, Roeboidesdayi, and Aequidens pulcher(Figs. 2 and 3) weremore difficult to estimatebased solely on SL frequencyhistograms,and thus required careful considerationof growth rates. 3) Duration of the breeding season(in months) was estimated from plots of mean gonad codes,visceral fat, SL's, and relative proportions of juveniles to adults by month at eachsite (Fig. 4). 4) In order to retain as much information as possible for statisticaltests,I choseto make a very rough approximation of the number of reproductive bouts per year, rather than codeeachspeciesin a conservativebinary fashion (i.e. single versusmultiple brooded). Average reproductive bouts per year per female was estimatedfrom plots of mean gonad codesby month and recordsof sizedistributions of oocytes within mature ovaries.Theseestimateswere very conservative, particularly for speciesthat showednearly continuous breeding seasons(some of these fishes may even spawn on a daily basis during peak reproductive periods). Bias againstthe highestvaluesfor number of reproductivebouts per year due to conservativeestimateswould not greatly affect overall ordination of specieson the variable (i.e. estimatesof bouts per year can be more appropriately viewed as ranks rather than parametric values). 5) Fecundity was recorded as the averagenumber of the largestoocyte sizeclassbasedon three fully-gravid females. For small species,the total number of oocyteswas counted directly. The number of oocytes from large ovaries was estimated by counting and weighing 500 mature oocytes (blotted dry, wet weight to nearest0.01 g), then weighing the entire ovary, and solving for the total number of oocytes. 6) Maximum oocyte size was equal to the diameter of the largestoocyte in fully developedovaries(nearest0.05mm). Intraspecific variation in diameter of the largest oocyte in mature ovaries was small, ranging from lessthan 0.05 mm in specieswith small oocytes to not more than 0.55 mm in specieswith vary large oocytes(e.g.Ancistrustriradiatus). 7) Becausetropical fishesexhibit parental care in a variety of forms, it was quantified as the sum of A + B + C + D from the following: Cur/mata .rgentea HopI/as ma/abarlcus 10 20 30 40 50 10 70 10 20 30 40 50 60 70 10 20 30 40 50 80 70 10 20 30 .0 10 80 :J .. 10 50 100 '10 200 250 .. "i ~ '0 '> :a .5 .. i; ~ ~ '> :c ..s - 0 .. $ .0 E ~ z 0 ~ OJ .c E ~ z 50 10 50 100 100 150 150 200 200 250 250 SL Interval SL Interval Fig. 2. Standard length frequency histograms during each month for Hop/ios malabaricus(extendedbreeder) and Curimata argentea (cyclic breeder)at Cano Maraca ~ 229 Roebolde. dayl Aequldenspulcher 10 20 30 40 10 10 20 30 40 50 ,. 20 30 40 50 20 30 40 so 12.. May I0'- 10 ~ i; ~ 'C "> :cc " i;~ 'C "> :c .5 0 -0 .. . .0 E ~ z ~ e .c E ~ z .., HI... e. o. 1. 10 20 10 .0 10 SL Interval 10 a, 30 40 50 ., SL Interval Fig. 3. Standard length frequency histogramsduring each month for two fishes with extendedbreeding periods, Roeboidesday; and AeQuidens Dutcher.at Caiio Maraca 230 ! iu 1 1 & ~ c. . I .I -( f, ,,~-\,\ " .~ ~ ,f'f,f' J 0---0 I -- , , ,.0---0-- '. "V .'0" .0 ! ""/ ,°" "01 .» eE I ; . 8I ... 100 uj. 4.M.M"".80MD . . . . . . . . . . . J. A . .. -- J J . 0 0 Month ~ ~ i .. ~ 0 <1 " ~ i P"', ~ "S 0.1 ~ > 0.. ... 0.. .2 i ~ 0.2' 2 ... .,..4. , , , , ! . , "'-0-_.0 , b," '. I '. r. V r .m :. .. , . . . . . . . . i' : i s '; ~ c ; ... I O.t ~ ... . ~, 0.4 .. ~ J'M.MJJ.80NO . 0.2 - r"... , , , °'. -" I . .--:$ 6 ,'b, '" .,.:::'~::X""-' I ~: ~ 0 , '. . . . . . . . . J'M.MJJ.80ND ~. J ~."20 Comparison of reproductive Month measuresof a multiple--spawningcharacid (Bryconomericw beta) and cicblid (Caquetia kraw.ril) with a seasonally-spawning characid (Astyonax bimacrdatus)and cichlid (CichlD.romaor~) from the Venezuelanllanos (ciORd symbobcorrespondto meanfat code-topponeb. and mean SL-bottom~b; opensymbolscorrespondto mean gonad code-toppane-b. and ratio of juvenilesto adJdts-bottom panels) A = specialp1ac:ement of zygotes(1) or zygotesand larvae (2) B = brief period of nutritional contribution to larvae (2), or long period of contribution to larvae or embryos (4) C = brief period of parental protection by one sex (1), or both (2), or long period of parental protection by one sex(2), or both (4) D-extremely long period of gestation (4) (applied only to Potamotrygon) Parentalcarecodesrangedfrom 1 to 8 and, with few exceptions, most details of reproductive behavior of a species were documentedby field observationsof fishes collected at the sites,or the scientificand aquarium literature. 8) Dry season age distribution of each population was coded as either O-No adults (applied only to two species of annual k.illifish), I-Skewed in favor of juvenile and subadult sizeclasses.2-Approximately even distribution of immature and adult size classes,and 3-Skewed in favor of adult sizeclasses. 9) Wet season age distribution of each popUlation was coded as either 1, 2, or 3 as before. 10) The maximum standard length among all specimens of eachspecieswas measuredto the nearest0.1 Mm. Maximum length was usedin analysesrather than averagelength of mature adults, since it better reflects genetic potential among fishesexhibiting indeterminategrowth. For ten speciesthat occurred at both sites, data from the site with the highest popUlation density were used in 231 analyses.Not surprisingly, a few life history variables revealedinterregional differenceswithin a species(theseprovide a basisfor comparisonsin a future report). The freshwater stingray, Potamotrygonorbignyi, was collected from other sites in the llanos (estado Apure) and included in the cluster analysis,sinceit has very distinctive life history characteristics(e.g. an invariant clutch sizeof two, viviparity, and a long gestation period). With the exception of Prochi/odusmariae (both sites) and Brycon whitei (C. Volcan), and the freshwater stingray, only speciesknown to be reproducingat the two siteswereincludedin the analysis. Immature yearlings,non-gravid, and gravid adults of Prochi/odus and Brycon were collected, but spawning took place at remote downstream locations. At the beginning of the rainy season,long distancemigrations (adults downstream,immatures upstream)are a conspicuousfeature of the ecology of thesespecies(Lilyestrom 1983; Lilyestrom and Taphom 1983).If one or two variables could not be estimatedfor a specieswith confidence,the value assigned to its closestphylogeneticrelative was used as an estimate of the unknown character (This was performed for only 21 of 720, or about 3% of the observations in the data matrix). Becausethe reproductive biology of the genus is distinct and well documented(Wourms 1972; L.G. Nico and D.C. Taphom pers.comm.), the annual killifish, Ptero/ebias hoignei, was included in the data set, even though only two individuals werecollectedat Cano Maraca during the wet season. Analyses A principal components analysis was computed from the correlation matrix derived from 71 fish species and all 10 life history variables in the original data set (Appendix 1). Principal components analysis is an ordination method that allows a multidimensional swarm of correlated data points to be viewed within two or three new orthogonal dimensions. When present, intrinsic patterns within the multidimensional swarm will emerge, generally within a plot of the first two or three components (i.e. independent axes derived from the original dependent variables; Pielou 1984). The strength of each component is reflected by its associated eigenvalue, with values greater than 1.0 being generally accepted as explaining a significant proportion of the variance in the multidimensional data set. Correlations between the first two principal component scores for each speciesand the original life history variables for each species were computed to provide an index of the relative importance of each variable in the ordination of species into reproductive strategies. In accordance with recent considerations of the effect of allometry on life history traits (Calder 1984; Dunham and Miles 1985; Gittleman 1986), additional PCA's were computed on modified data matrices with 1) log-transformed SL and fecundity and 2) with nine dependent variables corrected for SL (residuals from log SL linear regression). To test the hypothesis that patterns of interspecific variation among life history traits have a phylogenetic basis (Dunham and Miles 1985), a nested analysis of variance was performed on each of nine life history variables, using family nested within order and log length as a covariate. A cluster analysis was performed using Euclidean distances based on all 10 attributes among 72 species (stingray included). Each of the ten life history variables was stan- dardized (mean= 0, standard deviation= 1) prior to computing Euclidean distances.Clustering was performed by the average linkage method (Sokal and Michener 1958), where the distancebetweentwo clustersis the averagedistancebetweenobservationsand/or clusters.Canonical discriminant function analyseswere performed to explore patterns of within classvariation relative to betweenclassvariation among life history variables. The type of wet season age distributions (3 states),dry seasonage distributions (4 states)and parental care (6 states)were usedas classvariableswith the other nine traits as dependentvariables.The remainingsevenlife history variablescontainedten or more states, and were thus treated as dependent variables for the purposesof canonical discriminant function analysis. Results Univariatestatistics Comparedwith temperatezone assemblages (Mabon 1984; Wooton 1984), the fishes of both Caiio Maraca and C. Volcan exhibited an extremely wide range of life history attributes (Appendix 1). Large interspecificvariation in fecundity, body size,and the timing of reproduction was apparent within most orders and families. For example,both aseasonaland seasonally-spawningspecieswere observed in the specioseCharacidae as well as the comparatively species-poorCichlidae (Fig. 4). Diameter of mature oocytes ranged from a low of 0.45 mm in Curimata argentea to a high of 4.00 rom in Ancistrus triradiatus (i.e., excluding 35.00mm reported for Potamotrygon by Thorson et aI. 1983). Forty seven species,including Curimata argentea, Astyanax bimacuJatus.Brycon whitei, and Eigenmanniavirescens,exhibitedvirtually no parental carefollowing spawning. Cichlids and loricariid catfishesexhibited highestlevels of parental care. lntercorrelations among the ten life history variables appear in Table 1. Significant large positive correlations (r~ -0.5, P<0.01) were obtained for generationtime with log SL, log fecundity with log SL, and eggsizewith parental care. Significant large negative correlations (r ~ 0.5, P < 0.01) were obtained for generation time with length of breedingseason,reproductive bouts per year and wet season agedistribution; log fecundity with wet seasonagedistribution; and log SL with wet seasonage distribution. Twenty six life history variable correlations were negative and 19 were positive or zero. When size was factored out of the analysisusing residuals from the log length regression, signsand relative magnitudesof correlations remained essentiallyunchanged(Table 1). Multivariate statistics When all ten variables were included in the data matrix (raw scoresfor 8 variables plus log fecundity and log SL), the flISt three principal components reduced variation by 76 percent (Table 2). The result was very similar when SL and fecundity were untransformed (first three PC's modelled 70% of total variation). The first axis was influenced more or lessequally by sevenvariables(population fluctuation, egg size,and parental care excluded).The secondaxis was derived primarily from egg size,parental care, and log length. Population fluctuation, length of breeding season, dry and wet seasonsizedistributions werealso significantly correlated with PC2 (Table 3). The third principal compo- ~ 232 Table I. Correlation matrix of absolute life history variablesfor fIShesfrom two sites in the Venezuelanllanos. Values in paralthe1e8 are for residualsof regressionwith log SL Life history trait Fluctuation Generationtime Lenlth breedingseason Bouts per Year Fecundity Eggsize Parentalcare Dry sizedistr. Wet sizedistr. Maximum length Fluctuation Generation time Length breeding season x ( -0.09 x 0.6S** -0.62** 0.62** O.IS* -0.01 0.49** -0.S7** :-0.29)** :-0.63)** )( O.~" -0.04 -0.30-- 0.10 0.01 -0.18-0.34-- -0.07 -0.10 -0.04 Mean val~ 1.06 . -0.S7 10.69 Bouts per year Parental care (0.13) (-0.S3)** (0.39)** )( -0.3S** -0.59** 0.12 0.41.* -0.46.* 0.46.. -0.2S. 0.08 0.00 -0.73.* 0.46** -0.37.. 3.78 (-0.01) (0.41)** (-0.25)* (-0.49)** )( 2.80 -O.OS -0.09 0.43.. -0.67 0.61.. (-0.22).. (-0.36).* (-0.16) (-0.18) (0.27). (0.50).. (0.31).. (0.11) (-0.49)** (-0.33)** x (0.62).. 0.65** x -0.26. -0.34** -0.16 -0.05 0.47** 0.25. 5.269~,d 1.39 1.94 Wet size distr. Dry size distr. (-0.09) (0.53)** (-0.45)** (-0.75).. (0.45)*. (-0.34).. (-0.38)** x -0.15 0.11 (-0.09) (-0.30).* (0.41) (0.31)*. (-0.44).. (0.27). (0:11». (-0.10) x -0.70.. 2.67 1.46 P<O.OS...P<O.Ol T~ 3. Correlations of rant two principal component axes with the original life history variables tire history variable PC1 PC2 Fluctuation Generationtime Length breedingseason Bouts per year Log fecundity Egg size Parentalcare Dry ~ distr. Wet ~distr. Log length 0.07 0.86.. -0.37.. 0.09 0.36.. 0.09 0.05 Absolute data - CumuJati~ v8riaIa PC! PC2 PC3 4.01 2.29 1.30 Variable PCI Fluctuation Generationtime Lengthbreeding season Bouts per year F~dity EggParentaicare Dry Si7£ distr. Wet- 0.401 0.630 0.760 distr. Maximum lenlth 0.034 0.430 -0.345 -0.396 0.410 0.023 -0.060 0.315 -0.385 0143 --- -. - PC2 -0.244 0.061 0.243 0.057 -0.033 0.560 0.570 -0.286 -0.168 0.354 PC3 0.692 0.692 -0.051 0.357 0.003 0.089 -0.166 -0.424 -0.275 0.230 Lenlth-adjusteddata ~ Eigenvalue PC! PC2 PC3 Variable 3.71 1.62 1.03 - Cumulative - -~ vari8DCe . 0.412 0.592 0.707 PCl Fluctuation -0.073 Generationtime -0.364 length breedingseason -0.381 Boutsper year -0.3WJ -0.368 Fecundity Eggsize 0.325 Parentalcare 0.314 Dry si2I:distr. -0.394 Wet si2l:distr. 0.276 I PC1 -0.611 0.281 0.127 -0.366 0.071 0.325 0.472 0.243 0.061 PC3 "-1 0.270 0.390 -0.546 0.090 0.090 0.497 0.151 -O.1S1 -0.227 . p<o.os, ..p<o.oos -0.69.. -0.79.. 0.82.. 0.05 -0.12 0.63.. -0.77.. 0.69.. 0.85.. 0.86.. -0.43. -0.25. 0.54.. nent axis was influenced primarily by population fluctuation, dry seasonage distribution, and reproductive bouts per year. The fourth axis modelled only 7% of the multidimensional variance and was derived primarily from three variables: length of breeding season (eigenvector=0.512),log fecundity (-0.527), and egg size(0.450). A plot of speciesby their scoreson the first two principal componentsrevealsa more or less triangular scatter with apices corresponding to three fairly distinctive suites of characteristics(Fig. 5). Specieswith intermediate scoreson PC1 and high scoreson PC2 exhibited the following attributes: well-developed parental care, prolonged breeding seasons, repeated reproduction, evenly distributed size classesthroughout the year, long generation times large eggs,and large body size.Acxording to sampleCV values. these speciestended to exhibit comparatively small-scale fluctuations in local population density over the course of the year. A fourteen speciescluster around the first apex was dominated by catfishes (Siluriformes) and cichlids (Fig. 6). The secondapex resulted from low scoreson PC1 and intermediate or low scores on PC2 (Fig- 5). This region of two-dimensional component space is charactcrlud by little or no parental care, prolonged breeding seasons,re- 233 5 4 3 2 C\I U c. 0 -2 -3 -7 -5 -3 -1 1 3 5 PC 1 Fig. 5. Plot of Venezuelanfish speciesscoreson the first two principal components(analysisderived from raw life history data with log fecundity and log SL). Symbolsgroup speciesinto three broad categoriesbasedon a clusteranalysis.Top ellipseboundsan equilibrium strategy,left eclipseboundsan opportunistic strategy,and right ellipse bounds a seasonalreproductive strategy. The small ellipse identifies fishes of small body size that exhibit a strategy intermediate betweenextremeseasonalreproduction and opportunism 0 Opportunistic; 8 Seasonal; d Equilibrium 5 4 3 N 2 Fig. 7. Cluster diagram of Venezuelan fishes based on Euclidean distances computed from ten standardized life history attributes. Root mean square distances between species is 4.47. Species number codes are given in Appendix 1. Explanation of three patterns appears in text U D.. 0 -2 -3 -7 -5 -3 -1 1 3 5 PC1 Fig. 6. Plot of tropical fish PCA scoresas in Fig. 5, but with symbols identifying speciesby order. Again, the top apexcorresponds to the equilibrium strategy, left apex to opportunism, and right apexto a seasonalreproductivestrategy0 Characiforms;. Siluriforms; t. Perciforms;A Cyprinodontiforms; 0 Synbranchiforms peatedbouts of reproduction, an evensizedistribution during the wet season(for somespeciesduring the dry season as well), short generation time, relatively small clutches, small oocytes,small body size,and intermediatepopulation fluctuations. Characiformesand Cyprinodontiformes were predominant among the nine speciesgrouped near the second apex (Fig. 6). The third apexwascomposedof specieswith high scores on PC1 and intermediate or low scoreson PC2 (Fig. 5). Speciesin this region of principal component two-space were characterizedby very little or no parental care, short breedingseason,few bouts of reproduction per year, adult- dominated dry seasonpopulations,juvenile-dominatedwet seasonpopulations, long generationtimes, intermediateto high fecundities, small oocytes, and intermediate or large body size.For the most part, specieslocated near the third apex exhibited large local population fluctuations. The 48 specieslocated in the region of the third apex belonged to orders Characiformes and Siluriformes exclusively (Fig. 6). Seventeencharaciform and siluriform fisheswere intermediatebetweenthe secondand third apices(intermediate and low PC1 and low PC2 scores;Fig. 5). Theseintermediate fishes were all comparatively small, with prolonged, yet distinctly seasonalreproduction. Egg sizeswere small, fecundity intermediate, and parental care absentor weakly developed in this group that included Corydoras species,Ochmacanthus alternus, and OdontostilbeDulcher. Clusteranalysis Clustering of 72 speciesbased on Euclidean distancesresulted in three major groupings plus two single species branches (Fig. 7). The freshwater stingray, Potamotrygon orbignyi, was included in this analysis(estimatesfor several Potamotrygon life history parameters were based on the closely related P. morro using data from Thorson et al. 234 T.~ 4. Resultsof nestedANDY A for family within order with 1983). The same three basic life history strategies were log length as a covariate fonned by three large clusters,with Potamotrygonand the characid, Brycon whitei, representingextreme examplesof Trait Classvariable Type I SS two of thesebasicsuitesof characteristics. Potamotrygoncould be consideredan extreme caseof p F the life history strategythat correspondsto long generation time, well-developedparental care, large oocytes,and low 4.38 0.003 Order Population fecundity. The same 14 species(5 families, 3 orders) that Fluctuation comprised the ftrst apex from PCA produced this cluster Family (order) 2.42 0.009 2.36 0.01 (Figs. 5 and 7). Likewise,Bryconrepresentsan extremecase Lollength 0.01 0.974 of the strategyexhibited by the 48 speciescluster (19 famiGeneration Order 6.98 0.0001 4.S7 0.0018 0.6S lies, 2 orders)characterizedby seasonalreproduction, high time or intermediatefecundity, no parental care. and generation Family (order') 1.53 0.12 1.24 0.27 times from 6 to 48 months. This large cluster of seasonally Log length 24.83 0.0001 reproducing fishes contained two subunits that correOrder 5.SO 0.0005 5.96 0.(XX)2 0.50 LCDIth spondedto 29 large specieswith high fecundities(e.g. ProseuoD chi/odusmariae.Rhamdiaspp.,and Rhamphichthys marmorFamily (order) 0.98 0.49 0.91 0.S7 Log length 3.49 0.07 atus)and 19 small specieswith much lower fecundities(e.g. Characidiumspp., Microglanis iheringi, and Ochmacanthus Bouts per Order 20.38 0.(xx)1 59.07 0.(xx)1 0.78 alternus).An eight speciescluster (4 families, 3 orders)coryear Family (order') 3.54 0.CMM»3 3.51 0.CMM»3 respondedto the short generationtime, low fecundity, mulLog length 6.13 0.02 tiple-brood strategyof the PCA (Figs. 5 and 7). Creagrutussp. and Hemigrammussp. we~ grouped with aseasonally-reproducing, fast-maturing speciesby PCA, yet placedwithin the small body sizesubunit of the seasonallyreproducing group by averageclustering (Figs. 5 and 7). Gephyrocharaxvalenciaewas placed within the multiplebrood, fast-maturing group by cluster analysis, yet only slightly nearer to small-sized,seasonally-spawningspecies by the fIrSt two principal components.If one examinesthe original life history variables (Appendix I), it is clear that all three groups grade into one another as a continuum of reproductivestrategies. Analysisof variance The composition of the three groups clustered near the apicesof Figure 5 is associatedto somedegreewith phylogeny.For example,five of the speciesin the first apexgroup (high PC2 scores)are cichlids (Perciformes),sevenare catfishes(Siluriformes),one is a characiform (Hopliasmalabaricus), and one is a synbranchiform eel (Synbranchusmarmoratus). Likewise, the sa:ond group (low PCl scores)is dominated by characiforms and cyprinodontiforms (two killifishes and one livebearer).The third group (high PCl scores)is composedentirely of tetras, catfishes,and knifefishes(Characiformesand Siluriformes).On the other hand, membersof the most diverse order, Characiformes, were found in all three groups. Resultsof nestedANOV A show that order had a significant effect on all ten life history variableswhenlength wasnot included asa covariate(Type I SS,Table 4). When log length was included as a covariate (Type III SS). the main effect of order on wet seasonsize distribution became marginally insignificant (Table 4). Family significantly affectedall life history variablesexcept generation time. length of reproductive season,and log length (Table 4). Length as a covariate did not alter the fraction of family level tests (family nested within order) that attained statistical significance(P<O.O5). Becauseconsiderablevariation in adult sizeexistedwithin many families (SL range 28-313 mm for Characidae. 30-230mm for Cicblidae. 25-216 mm for Callichthyidae. 26-233 mm for Loricariidae. 38-200mm for Pimelodidae). Log fecundity Eggsize Order 24.41 0.CMM»1 17.32 0.0001 0.88 Family (order) 6.38 0.0001 4.01 0.0001 Log length 104.31 0.0001 Order 6.68 0.0001 7.98 O.lml 0.77 Family (order) 4.72 0.0001 4.12 O.lml Log length 48.99 0.0001 Parental care Order 72.97 0.0001 67.96 0.0001 0.91 Family (order) 4.59 0.0001 4.58 0.0001 Log length Dry season Order size distribution Family (order) Log length Wet season Order size distribution Family (order) Log length Log length Order Family (order) 27.64 0.0001 18.37 0.0001 11.37 0.0001 0.76 3.36 0.0005 3.37 0.0005 0.24 0.62 5.86 0.0003 2.39 0.OS2 0.74 2.92 55.78 4.10 1.31 0.002 2.13 0.021 0.0001 0.0035 0.225 patterns of covariation among life history variables could have been affected by allometry (Steams 1983a; Dunham and Miles 1985). Log length was significantly correlated with sevenof nine life history variables (Table 1). Again, results from nested ANOV A with log SL as a covariate indicate that length influenced the pattern of variation in only one of 18 testsfor phylogeneticaffects on life history variables(Table 4). Whereasthe potential for a high fecundity might be expected to exhibit an allometric pattern, larger fishesdid not always exhibit higher fecunditiesthan smaller fishes (e.g. Parauchenipterusgo/eatus.Hypostomus argus. Loricarichthys typus, Pterygoplichthysmultiradiatus, Crenicichiageay,'). Canonicaldiscriminantfunction Resultsof the canonical discriminant function analysesindicate that all three of the designatedclass variables bad significant (P<0.<XX>1)ratios of witbin-class to bet~classvariance among the remaining nine life history variables.Canonicalcorrelations were0.88 (wet seasonsizedis- tribution), 0.84 (dry seasonsizedistribution), and 0.84 (parental care). For wet seasonsizedistribution, the first canonical eigenvector had high loadings for log f~dity (0.77), egg size ( - 0.58),length of breedingseason( - 0.46), and log SL (0.45).Taken together, thesedata reveal a pattern in which speciesexhibitingjuvenile-dominatedwet season populations tend to have high fecundities,large body size,comparativelysmall oocytes,and a brief reproductive season.Conversely,populations dominated by adults during the wet seasonweresmaller,lessfecund,and had larger oocytes and longer reproductive seasons.Reproductive bouts per year (0.91) and log fecundity (- 0.43) had the highest loadings for the first canonical eigenvectorof dry seasonsize distribution. In one-dimensionalspace,species ranged betw~ adult-dominated dry seasonpopulations exhibiting repeatedbouts of reproduction and low fecundities, to juvenile-dominatedand uniforDl populations with relatively fewer bouts of reproduction but larger clutches. The seconddry seasonsize distribution canonical vector had a correlation of 0.77and highestvariable loadingscorrespondingwith length of reproductiveseason( - 0.78), parental care( - 0.67), and population fluctuation (0.46).The secondcanonical vector discriminated among dry season distribution classesalong a gradient running from short breeding season,little parental care, variable population size, and adult-dominated dry seasonpopulations at one end, to prolongedbreeding.parental care, more stablepopulation density, and uniforDl dry seasonsize distributions at the other. Egg size (0.87), length of breeding season (0.42),population fluctuation (-0.39), and log SL (-0.37) had the highestloadingson the first parental carecanonical eigenvector.In other words, speciesexhibiting greaterlevels of parentalcaretendedto havelarger eggs,longer breeding seasons,lower population fluctuations, and smaller body sizethan fisheswith poorly-developedparental care. Overall, resultsof canonicaldiscriminant function reinforced findings ofPCA and averageclusteringby Euclidean distances.Patternsof reproduction appearto be particularly consistentwhenat leasttwo dimensionsare incorporated into the model simultaneously.A two-dimensionalgradient having three distinct endpoint strategiesseemsto describe tropical fish life history patterns in the simplest and most parsimonious fashion. Small, multiple-clutching fishes exhibiting small investmentsin individual offspring correspondto a life history strategyassociatedwith short generation times and rapid population turnover. Larger more fecund, seasonally-spawning fishesalso investlittle in individual offspring (i.e., small oocytes, no parental care), but much more in total wet seasonreproductive effort. In contrast to both of thesepatterns,a group of aseasonally-reproducing fISheshad large oocytes, interDlediate or small clutches,and parental care of eggsand/or larvae. The last group seemsto representa strategyassociatedwith investing relativelymore in individual offspring. leadingto higher survivorship and moderation of local population fluctuations. m.cussion Covar;ationamonglife history variables Even though most of the primary measurementsof reproductive and demographicparametersinvolved thousands of examinedspecimens(Appendix 1). population fluctua- tions, averagegeneration times. length of breedingseason, and number of reproductive bouts per year should be considered best approximations until new data becomeavailable. While limitations of the data should be taken into account, potential life history trade-offs are implicated by 12 significant negative correlations obtained among nine variables (14 using length adjusted values). Egg size and parental care changed from having zero correlations with log f~undity to having significant negative correiations when the data were adjusted for length (Table 1). Because tropical fishes exhibit wide variation in body form (e.g. ranging from the robust Cich/asoma orinocenseto the snake-likeSynbrmrchwmarmoratw), length provided only a rough estimateof body sizedifferences.Even though gravimetric or volumetric data could potentially increasethe sensitivity of size-adjustments,the sevensignificant length correlations were consistent with other comparative fish studies(Ita 1978,Mahon 1984;Wootton 1984).For example generationtime, fecundity, and egg sizewere all correlated positively with body length (Table 1). Length of breeding season.bouts of reproduction per year, and the ratio of adults to juvenilesduring the wet seasoncorrelatednegatively with length. Several of the observed negative correiations should have resulted from fundamental physiological constraints. Clutch sizesare bound to be lower in speciesthat exhibit many spawning bouts per year, compared to seasonal spawnersthat exen total annual reproductiveeffon during a constricted time period. Likewise, higher total fecundity is achievedby packing lessmatter and energyinto individual oocytes.Many of the positive correlations among variables would be logically anticipated as a consequenceof secondaryconstraints imposed by primary negative constraints (Peaseand Bull 1988).The difficult task of assessing which suitesof charactersmight form adaptive life history strategieslargely reducesto a problem of critically assessing which attributes constitute key variablesresponsiveto natural sel~tion. Indirect evidencefrequently may be required for identification of crucial trade-offs(Peaseand Bull 1988). Somevariablescan be evaluatedsingularly with regard to a variety of environmental parameters. For example, the timing of reproduction of fisheshasbeenviewedas adaptive with respectto a number of different environmentalfactors (Kranler 1978; Johannes 1978; Baltz 1984; Keast 1985; and seebelow). Basicpatterns and ecologicalcorrelates Resultsof multivariate analysesof tropical fish life history attributes were robust. yielding similar patterns for data sets containing SL as a variable, length log-transformed, data adjusted for length effects, and treating three of the ten attributes as classvariables.Moreover, averageclustering and PCA produced nearly identical groupings among fishes. At plot of the specieson the first two orthogonal axesresultedin a distribution with three apical clusterings, each having fairly homogeneouslife history features.One suite of life history traits always contained intermediateor long generation times. large investment in individual offspring, delayed maturation, aseasonalreproduction, and prolonged breeding.To someextent. this associationof life history traits agreeswith the relative .. K-strategy" as originally proposed by Pianka (1970). Presumably,the suite of characteristicsforming this "equilibrium strategy" is asso- ciated with higherjuvenile survivorship as result of greater parental investment in individual progeny (Strategiesare used here as hypothesesfor interpreting observedpatterns as adaptive suitesof characteristics.Neither teleology nor cognition are implied from the present usageof the term. cr. Chapleauet al. 1988).With the largestoocyte (diameter estimated near 35.0mm), the smallest clutch (two offspring),long gestationperiod (19- 11months), and late maturation (ca. 43 months, Thorson et al. 1983),the freshwater stingray, PotamotrygOll,provides an extreme example of the so called equilibrium strategy. As a group, cichlids tended to exhibit equilibrium strategy characteristics(i.e., comparativelylarge oocytes,brood protection, and acyclic spawning). Enhancedearly survivorship of offspring was indicated by the presenceof numerousjuvenile sizeclasses of somecichlids throughout the dry season(e.g. Caquetia kraussii, Fig. 4). Equilibrium-strategists apparently reproduced with somedegreeof success,even though predation pressurewas intenseduring the early dry seasonat Calio Maraca. when fishes were encountered at high densities (Winemiller 1987). The two remainingassociationsof life history characteristics appear to divide many of the traits comprising Pianka's (1970) earlier "r-selected" suite of attributes. In an attempt not to introduce new jargon, I will refer to the suite of characteristicscomposedof short generation time, low fecundity, and minimal investmentper offspring as the "opportunistic strategy". The opportunistic strategy was associatedwith small species(e.g. Poecilia reticulata. OdontostiJbepulcher) that remained reproductively active despite apparent high juvenile and adult mortality during the harsh conditions and intensepredation of the dry season. At Calio Maraca, fishesexhibiting the opportunistic suite of attributes built-up dense local populations from relatively small pools of adult founders over a six month period following the initiation of rains. This population growth was achievedthrough a combination of multiple bouts of reproduction by older adult survivors with rapid recruitmentof new adults via rapid maturation rates. The third suite of life history attributes clearly constitutesa "seasonalstrategy", which is characterizedby cyclic (often annual) reproduction. relatively long generation times (usuallycoinciding with the reproductivecycle),large clutches.and small investmentper offspring. Fishesidentified as seasonal-strategies exhibited a characteristic burst of reproduction with the early rains, followed by gradual reductions in population size due largely to predation on immaturesduring the early dry season.Astyanaxbimacu1aIus and Curimata argenteaat Calio Maraca provide good examplesof the seasonalstrategy, increasingrapidly from relict, adult dry seasonpopulations to largejuvenile populations in the newly-inundatedswamp floodplain. A major fraction of juveniles produced during this "spring bloom " of production never survivesto maturitY, ultimately falling prey to birds and predaceousfishes during the early dry season.The seasonalstrategyseemsto exploit both temporal and spatial variation in quality of habitats for enhanced juvenile survival and growth. Eighteen of the species grouped within the seasonalstrategyexhibited short-range migrations from the downstream channel into the upper esterofor seasonalbreeding,whereasonly Hopliasmalabaricus,exhibited comparatively less-pronouncedlocal migrations amongthe 14equilibrium-strategists.The largecharaciforms, ProchiJodusmariae and BrycOll whitei, exhibited long-distance seasonal migrations. with adults moving down from the Andean piedmont to spawn in productive llanos floodplains during the wet season.Theseadults later return with yearlingsto the piedmont. presumablyto escape predation and the harsh conditions of the floodplains during the dry season. Quantitative featuresof fish habitats were highly variable on a seasonalbasis at the sites studied in Venezuela. and interestingly, 48 (20 families, 5 orders) of 71 species wereidentified as relative seasonal-strategists. Th~ species exhibited gonadal recrudescence during the dry seasonand a burst of spawningactivity following the first heavy rains of the wet season.Severalseasonally-spawningspeciesappearedto exhibit no more than one or two bouts of reproduction during the first severalweeks following initiation of rains (e.g.Astyanaxbimaculatus.Triportheussp.), whereas other local populations exhibited sustained,yet greatlyreduced, levels of spawning for several months after an initial synchronousburst (e.g. CtenobryconspiJurus.Serrasa/musirritans). Larval growth and developmentwere particularly rapid and survivorship probably quite high within the newly-expandedaquatic environment. Larval food resourceswere abundant and adult predatory fisheswere at their lowest densities during the early wet season(Winemiller 1987). During the first three months of the wet season,species classified as seasonal-strategistsdominated the fish community at Calio Maraca, both in terms of biomass and density. As the wet seasonproceededtoward a new dry period, formerly dominant seasonal-strategistsgradually gave way to speciesexhibiting opportunistic and equilibrium strategies.As previously noted, severalopportunisticstrategiesachieved their highest densities (e.g. Roeboides day;. Hemigrammussp.) during the late-wet/early-dry period of transition (September-December)via continual ~ruitment of new individuals, this in spite of increasingpredation pressureand diminishing food resources.As the dry seasongave rise to increasinglyharsh conditions (JanuaryMay), most local populations weregreatly reduced,particularly seasonal-strategiststhat persisted as adult remnants of the formerly-dominant local populations. Severalspecies of opportunistic- and equilibrium-strategistscontinued reproduction during periods of extremely harsh dry season conditions, although juvenile mortality was probably quite high. Burt et al. (1988)hypothesizedthat multiple spawning in the neotropical characid, Hyphessobryconpulchripinnis, increasestotal reproductive output, which could be adaptive at less seasonal,tropical latitudes. Yet many, if not most. tropical ecosystemsexperiencehighly cyclic rainfall. Given the theoretical finding that rapid maturation rates maximizethe intrinsic rate of population increasemore efficiently than increasing either survivorship or fecundity (Lewontin 1965; Sibly and Calow 1986b; and others). I hypothesizethat colonizing ability in the face of intense predation or unpredictable variation in quality of aquatic habitats is the major adaptive feature of the opportunistic strategy. Maturing rapidly, producing multiple small clutches, and recolonizing ephemeral habitats each year, the annual killifishes (Cyprinodontidae) provide compelling illustrations of advantagesof the opportunistic strategy. Results from Reznick and Endler's (1982) investigation of Trinidadian guppies,PoeciJiareticulala, further support this interpretation of the opportunistic strategy. Femalesfrom environmentswith greather threats of predation for adult 237 guppies matured at smaller sizes, had shorter interbrood intervals and allocated larger fractions of tissue to reproduction. The basiclife history strategiesdisplayedby neotropical fishesagreewell with the three patterns that emergedfrom Baltz's (1984) interspecific comparison of temperate surfperches(Embiotocidae). One group of surfperches contained six specieshaving large bodies, moderate to high fecundities. delayed maturation, and long life spans. All of thesetraits correspondto the relative seasonalstrategy derived from the present investigation. Likewise, Baltz's secondgroup of medium-sizedsurfpercheswith low fecundity and delayed maturation mirror many characteristics of the equilibrium strategyamongncotropical fishes.Moreover, the traits associatedwith this third group of ten small surfperches (i.e., short-lived, rapidly-maturing, variable clutches)agree well with the suite of traits describing the opportunistic strategy. Baltz (1984) also discusseda trend of higher fecundity in associationwith more seasonalenvironments. A general hypothesis of life history evolution in responseto environmentally-induced variation in resource availability, predation pressure,and physiological stressemergesfrom the two independentanalyses. If ~Iection derived from environmentalvariation is ultimately responsiblefor the evolution of life history traits (Steams 1976, 1977; Southwood 1977, 1988), why do all speciesat a given site not respond in a similar manner? For example,Calio Maraca and C. Volcan are highly seasonal environments,but only 69% if their resident fishes were identified as relative seasonal-strategists.the remainder being either opportunistic- or equilibrium-strategists. Similarly, Kramer (1977) and DeSilva et al. (1985) found both seasonallyand continuously-reproducingostariophysan fishes syntopic in small rainforest streams.The solution probably lies in the relationship betweentrophic ecologyand variation in resourceabundanceand predation pressure.Most seasonal-strategistsat the two sites were omnivorous and insectivorous fishes. The availability of food resources, both aquatic and terrestrially-based, is strongly influencedby ~asonal rainfall (Winemiller 1987). Opportunistic-strategistswere primarily small fishes with comparativelybroad diets and subjectto inten~ predation. Roeboidesday;, a facultative scalepredator, was a partial exception.Roeboidesswitchedfrom preying primarily upon aquatic insects during the productive rainy season,to a diet comprisedmostly of scalesduring the transition period precedingpeak dry conditions. Many of the equilibriumstrategistsat both sites were benthic omnivores and piscivores. Relative to other trophic guilds, the density of adult food resourcesis lessvariable with seasonfor thesespecies. The period of peak food availability for many adult piscivores(transition season)did not correspondwith the period of highestfood availability for juveniles (wet season).With adult food resourcesavailablefor gameteproduction, equilibrium-strategistsprobably produced limited numbers of offspring during the dry seasonvia brood protection. Reduction in the overall number of breeding fishes during the dry seasonmight also reducepotential inter-brood competition for limited larval food resources. Information on fish speciesdistributions within Represa Las Majaguaswas usedfor an additional test of the hypothesis that the seasonaland opportunistic strategies are more adaptivethan the equilibrium strategyin periodicallyvariableenvironments.The main body of the lake provides fisheswith a stable habitat compared to the shallow coves below the inflow canals that raise the lake's water level during the rainy season.For those speciesfor which I did not have data from Calio Maraca or C. Volcan (N= 15), a relative life history strategy was assignedbasedon published information on reproductive behavior of the species and strategiesexhibited by closely related speciesfrom Appendix 1. For example,all cichlids were assumedto adopt a relative equilibrium strategy,sinceall are known to exhibit brood protection. relatively large oocytes,et cetera.The life history strategiesassignedto eachspeciesare presented in Appendix 2. I documented15 relative equilibrium-strategists,four opportunistic-strategists,and nine seasonal-strategistsfrom the stable lentic region of Las Majaguas. Four equilibrium-. four opportunistic-. and 17 seasonal-strategists were collected from the seasonally-variablecanal region of the lake. CaIto Maraca, another highly seasonal environment. had 13 equilibrium-. 9 opportunistic-. and 44 seasonal-strategists (Appendix 1). Chi Squareanalysisindicated a significant associationbetw~n reproductive strategy and habitat for a 3 x 3 matrix <x Z = 16.3.4 df, P < 0.001) and a 3 x 2 matrix when opportunistic and seasonalstrategies were combined <x Z = 15.6,2 df, P < 0.001).All four of the opportunistic-strategistscollected from the lentic body of Las Majaguas were taken from shallow, gently-sloping littoral zones where small. less predictable fluctuations in lake level might be expectedto yield relatively large-scale habitat alterations. Phylogeneticand allometric constraints Patterns of covariation within the PCA and the hypothesized reproductive strategy affiliations of fishes were influencedby both body length (this being interpreted as an index of size) and phylogeny, particularly at the ordinal level (Table 4). If life history traits representadaptations, and henceare derived from natural selection,the isolation of a phylogeneticcomponentshould not be surprising. Patterns of covariation among life history charactersare generally clearer when comparisons are made at higher taxonomic levels (e.g. Ito 1978; Steams 1983a; Dunham and Miles 1985; Gittleman 1986; Dunham et al. 1987) than at lower levels(e.g.Stearns1983b). This observationis consistent with the view that life history patterns evolve in responseto natural selection,and are not merely phenotypic artifacts correlatedwith other traits of the organism.Given the appropriate selectiveregime,membersof widely different taxa sometimesexhibit subsequentconvergen~ in life history characters(Tinkle et aI. 1970; Stearns1983a; Dunahm et aI. 1987). The diverse characiform fishes provide excellentexamplesof life history character divergence.The order Characiformesexhibits morphological and ecological divergen~ to a degree that is perhaps unrivalled by any other animal order (aery 1977).Characifonns spannedthe entire spectrum of observed reproductive patterns. Obviously, this phenotypic variation representsevolutionary divergen~ from an ancestralprotocharaciform that exhibited a more limited suite of life history traits. Early protocharaciforms could have been relative equilibrium-strategists, akin to Hoplias malabaricuswhich possesses morphological characters considered primitive for the order. Or perhaps they were seasonal-strategists,as most speciesof the superorderOstariophysiappearto be. Either way, given sufficient evolutionary time (estimated to be 100-175 mil- lion yrs. for characiforms, Fink and Fink 1981), a striking diversity of traits has evolved from what was certainly a more restricted suite of characteristics. If congeners among early protocharacifonns had been compared with one another relative to more distant taxa, phylogenetic design constraints might have been inferred. On a proximate scale, there is every reason to expect that phylogenetic design constraints are real (Harvey and Mace 1982; Dunham and Miles 1985; Gittleman 1986). For precisely this reason, comparisons among distant taxa should serve as more appropriate guides for general theories ofljfe history evolution than intraspecific comparisons. According to Stearns (1983a), "although the comparative method may not be a sharp tool, it can be a strong one with broader scope of application than other methods. It certainly leads to a higher level of generalization, and to a broader perspective on what causes patterns than does the strictly experimental method..." Body size is both constrained and influenced by evolution of other life history variables. For example, RofT (1984, 1986) explained over ~8fo of the variation in age of maturation for 30 teleost fishes by incorporating growth functions and a standard assumption for a positive size-fecundity relationship into the model. In the p~t study, body length was positively correlated with fecundity, both among and within higher fish taxa. Logically, a fish must live long enough and assimilate enough resources in order to achieve a large clutch size. Consequently, a small clutch can be viewed as a logical consequence of small adult body size in many instances (e.g. Poecilia reticulata. Rachovia maculipinnis, Creagnllus sp.). Obviously, larger organisms have greater potential to produce larger clutches (e.g. Prochilodus mariae. Brycon whitei. Rhamdia spp.). Yet based on interspecifIC comparisons of diverse fishes, high fecundity does not ~ to be a necessaryconsequence of large adult body size (e.g. Hypostomus argus and Potamotrygon orbignyi, two of the largest fishes in the analysis had average clutches of only 289 and 2 respectively). C~UIioIIS Neotropical fishesinhabiting the samehighly-seasonalenvironmentsexhibit a variety of different life history patterns. Multivariate life history patterns of individual species spanneda two-dimensionalcontinuum bounded by three basicsuitesof characteristics.A so-calledequilibrium strategy was associatedwith sedentarylocal popUlations.relatively stable adult food resources,prolonged breedingseasons,and parentalinvestmentin individual offspring, which probably resultsin enhancedjuvenile survivorship and reduced fluctuations in local population density. An opportunistic strategywas characterizedby rapid recolonization of disturbed habitats by small, rapidly maturing, multiple spawningfishes. Most fishes in the llanos appearedto be associatedwith a seasonallife history strategythat exploits an annualexpansionof aquaticand community production. In the extremecase,the seasonalstrategywascharacterized by large adult size,high fecundity, absenceof parental care, and long-distancespawningmigrations to productive, wet seasonfloodplains. Acknowkdg~ts. I am grateful to E.R. Pianka and D. Reznick for commentingon earlier drafts of the manuscript.Valuablesuggestionsthat improved the study wereprovided by JJ. Bull. C.M. Pease.M. Kirkpatrick C. Thomas. J. Hayes,R. Buskirk..and L.E. Gilbert. I especially thank D.C. Taphorn for his hospitality in Venezuela.The field assistanceof D.C. Taphorn. L.G. Nico. A. Barbarino. P. Rodriguez.and N. Greig is glQtly appreciated.D.C. Taphom provided initial idcntifK:ationsof fishes. I thank. P. Urriola. R. Schargel.F. Mago-Lcccia. and C. Hubbs for assistance in obtaining pcnnits and visas. The Dircccion Administraa:ion y Desarrollo Pcsqucroof the Republic of Venezuelaprovided a collecting permit. I am grateful to D. Urriola for kindly allowing me to work. on his property. Funds were provided by a research grant from the National GeographicSocietyand the TexasMemorial Museum. Refa~ Baltz DM (1984) Life history variation among femalesurfpen:hes (Perciformes:Embiotocidoe)Environ BioI Fish 10:159-171 Burt A, Kramer DI, Nakatsuru K, Spry C (1988) The tempo of reproduction in HyphessobryconpuJchripinnis(Cbaracidae), with a discussion on the biology of 'multiple spawning' in fishes.Environ. Bioi Fish 22: 15-27 Calder WA, III (1984) Size, function, and life history. Harvard Univ. Press,Cambridge,Mass Cbaplaeu F, JohansenPH, Williamson M (1988) The distinction betwec;npattern and pfOCQl in evolutionary biology: the use and abuseof the term 'strategy'. Oikos 53: 136-138 DeSilva SS, Schut J, Kortmulder K (1985) Reproductivebiology of six Barbus speciesindigenous to Sri Lanka. Environ. BioI Fish 12:201-218 Dunham AE, Miles DB (198S)Patternsof covariation in life history traits of squamatereptiles: the effectsof sizeand phylogeny ~. Am Nat 126:231-257 Dunham AE, MiI~ DB, Reznic:t DN (1987) Life history patterns in squamatereptiles. In: Gans C, Huey RB (eds) Biology of the reptilia, V. 16 Ecology B: defenseand life history. Academic Press,New York Fink SV, Fink WL (1981) Interrelationshipsof the ostariopitysan fishes(Teleostei).Zool J Linn Soc 72:297-353 Gery J (1977) CharICOidsof the World. T.F.H. Publ., Neptune, NJ Gittleman JL (1986) Carnivore life history patterns: allometric, phylogenetic,and ecologicalassociations.Am Nat 127:744-771 Harvey PH, Mace GM (1982) Comparisons bet~ taxa and adaptive trends: problems of methodology. In: Kin,'1 Con. Sociobiology Group (eds) Current problems in sociobiology. CambridgeUniversity Press,Cambridge,pp 343-362 Horn HS (1978) Optimal tactics of reproduction and life-history. In: Krebs Jr, Davies NB (eds) Behaviouralecology: an evolutionary approach. Sinauer Associates, Sunderland, Mass., W 411-429 Ito Y (1978) Comparative ecology. Cambridge University Press, Cambridge Johannes RE (1978) Reproductive strategiesof coastal marine fishesin the tropia. Environ BioI Fish 3: 65-84 Keast A (1985)Developmentof dietary ~~lizations in a summer community of juvenile fishes.Environ BioI Fish 13:211-224 Kramer DL (1978) Reproductive seasonality in the fishes of a tropi. calltream. Ecology 59: 976-985 Lilyestrom C (1983)Aspectosde Ia biologia del coporo (Prochilt>du.rmorioe).Rev UNELLEZ Cienc Tecnoll : 5-11 Lilyestrom C, Taphom D (1983)Aspectosde la biologia y conservacion de Ia palambra(Brycon whil~') Myers y Weitzman,1~. Rev UNELLEZ Cienc(fecnol 1: S3-59 Lowe--McConnell RH (1987) Ecological studies in tropical fish communities. Cambridge Univ. Press, Cambridge Mabon R (1984) Divergent structure in fish taxoceneaof north temperatestreams.Can J Fish Aquat sa 41 :330-350 PeaseCM, Bun JJ (1988) A critique of methods for measuring life history trade-offs.J Evol Bioi 1:293-303 Pianka ER (1970)On r- and K-selcction. Am Nat 100:S92-S91 239 StearnsSC (1983b) A natural experimentin life-history evolution: field data on the introduction of mosquitofish(Gambusiaaf.fmits) to Hawaii. Evolution 37:601-617 Thorson TB. LanghammerJK, Oetinger M( (1983) Reproduction sand developmentof the South American freshwaterstingrays, PotamotrygoncircuJorisand P. morro. Env Bioi Fish 9:3-24 Tinkle DW, Wilbur HM, Tilley SO (1970)Evolutionary strategies in lizard reproduction. Evolution 24:55-74 Welcomme RL (1969) The biology and ecology of the fishes of a small tropical stream.J ZooI158:485-529 Winemiller KO (1987)Tests of ecomorphologicaJand community level convergenceamong neotropical fish assemblages.Ph.D. dissertn.,University of Texas,Austin Winemiller KO (1989) Developmentof dermal lip protuberances for aquatic surface respiration in South American characid fishes.Copeia 1989:382-390 Wootton RJ (1984) Introduction: strategiesand tactics in fLShreproduction. In: Potts OW, Wootton RJ (cds) Fish reproduction: strategiesand tactics. Academic Press,New Yort, London, pp 1-12 Wourms IP (1972) Developmental biology of annual fishes III. pre-embryonic and embryonic diapause of variable duration in the eggsof annual fLShes. I Exp Zool 182:389-414 Pielou EC (1984)The interpretation of ecologicaldata. John Wiley & Sons,New York ReznickD, Endler JA (1982)The impact of predation on life history evolution in Trinidadian guppies(PoeciJiareticuJata).Evolution 36:160-177 RofTDA (1984)The evolution of life history parametersin teleosts. Can J Fish Aquat Sci 41:989-1(MX) RofT DA (1986) Predicting body size with life history models. Biosci36:316-323 Sibley R. Calow P (1986a) Physiologicalecology of animals: an evolutionaryapproach.Blackwell Scientific Publ.. Oxford Sibley R. Calow P (1986b) Why breedingearlier is alwaysworthwhile. J Theor Bioi 123:311-319 Sokal RR. MichenerCD (1958)A statisticalmethodfor evaluating systematicrelationships.Univ Kansassa Bull 38: 1409-1438 SouthwoodTRE (1977)Habitat. the templet for ecologicalstrategies?J Anim EcoI46:337-365 Southwood TRE (1988) Tactics. strategiesand templates.Oikos 52:3-18 StearnsSC (1976) Life-history tactics: a review of the ideas. Q Rev Bioi 51:3-47 StearnsSC (1977) The evolution of life-history traits: a critique of the theory and a review of the data. Ann Rev Ecol Syst 8:145-171 StearnsSC(1983a)The influenceof si~ and phylogenyon patterns of covariationamonglife-history traits in mammals.Oikos (Copenhagen)41: 173-187 order Myliobatifonnes family Potamotrygonidae 1. Potamotrygonorbignyi 0.46 ): order Characifonnes family Erythrinidae 2. Hop/iasma/abaricus 0.63 2 3. Hop/erythrinus 1.78 3 unitaeniatus family Lebiasinidae 4. Characidiumsp. 1.16 3 5. Lebiasinaerythrinoides 0.46 3 6. Pyrrhu/ina/ugubris 1.27 3 family Anostomidae 7. Leporinusfriderici 1.54 3 8. Schizodonisognathus 1.54 3 family Prochilodontidae 9. Prochi/odusmariae 0.91 3 family Curimatidac 10.Curimataargentea 1.17 .) family Characidac 11.Aphyocharaxa/bumus 0.90 y 12.Astyanaxinteger 1.00 .3 13.Astyanaxbimacu/atus 0.97 .3 14.Astyanaxmetae 0.97 3 15.Astyanaxsuperbus 1.00 3 16.Bryconwhitei 1.00 3 17.Bryconamericus beta 0.61 218. Bryconamericus 1.15 3 deuterodonoides 19. Charaxgibbosus 1.10 3 20. Cheirodontops geayi 1.39 3 21. Corynopomariisei 0.80 2 22. Creagrutussp. 0.91 2 23. CtenobrycOll spihlrus 0.69 :) 24. Gephyrocharax 0.85 3 valenciae Submitted February 2, 1989{ AcceptedMay 30,1989 35.00 8 450 2 4 Equilibrium 2462 6024 2.00 1.50 352 213 521 13 3 1 Equilibrium Seasonal 154 2688 82 0.65 1.30 1.00 28 121 35 234 332 353 .. 2 .. Seasonal Seasonal Seasonal 1-2 1-2 28950 28950 1.00 1.00 252 2S5 71 91 24 ~ 28950 1.00 2W 363 3 Seasonal 1 12 2:-c~ 3528 0.45 92 1671 3 Seasonal 2 t J Z 2 11 12 12 12 12 24 4 6 12 6 617 4+ 2 8400 4287 1+ 9528 1+ 800 2 171545 1 796 6+-243 3 0.65 1.00 0.90 1.00 0.65 1.50 0.95 0.85 37 92 91 106 64 313 44 39 SO1 35 965 185 10 14 1341 289 .. 2 3 2 2 ~ 3 2 Seasonal Seasonal Seasonal Seasonal Seasonal Seasonal Opportunistic Seasonal 2 2 2 2 t 2 6 12 4 4 11 4 6 2 10 6 2-3 S 4 t 4+** 2 3 S+** 280 1108 135 94 755 734 1.00 0.75 0.85 0.80 0.75 0.7~ 101 28 43 28 S2 37 324 112 323 754 1530 819 1 1 2 2 . 3 Seasonal Seasonal Opportunistic Opport./Seas. Seasonal Opport./Seas. ) 2 r 2 .1 43 11 12 12 2; ~ 4+ t+ 11 U 12 4 2 3 2 1 1; 12 12 2 .. .. .. 2 .2 7. l t t 1 t t t 1 Seasonal Seasonal 240 Species F1\1<:t. 25. Hemigrammus .fp. 1.23 26. Markiana geayi 1.07 27. Odonto.ftiJbe ,uk_r 1.06 28. Pygocentrusnotatus 1.19 29. Roeboide.J dayi 0.84 30. Serra.sa/mus irritan.s 1.60 31. Serra.JOlmlLf medini 1.92 32. Serra.salmus rhombeus 1.21 33. TetragOllopterus 1.34 argenteus 34. Triportheus.fp. 1.35 family Gasteroplecidae 35. Thoracocharax.fteJlatus0.89 order Silurifonnes suborderGymnoloidei family Apleronotidae 36. Adontosternarchus 1.82 deVtDlanzi family Gymnolidae 37. Gymnotuscarapo 0.66 family Hypopomidae 38. Hypopomussp. 0.95 family Rhamphichlhyidae 39. Rhamphichthys 1.56 mannoratus family Slemopygidae 40. EigenmanniD vire.fcerrs 1.44 suborderSiluroidei family Auchenipleridae 41. Entomocorus gameroi 42. Parauc-nipterus gaJeatus family Aspredinidae 43. BIInocephohl.f .fp. family Pimelodidae 44. Microg/anisiheringi 45. PimeJodeJJa sp.l 46. PimeJodeJIa sp.2 47. PimeJodeJJa sp.3 48. Rhamdia.fp.l 49. Rhamdiasp.2 family Ageneiosidae so. AgeneiO.fUS vittatus family TrichomYCleridae Dry distr Wet Gen- Season Bouts distr. erat. 2 t 2 t 2 t t I t 11 52.Corydora.s 53.Corydara.s ,yeptentirionaJis 54. Corydara.s habrO.fUS 55. HopJO.Jtemum Jittorak family Loricariidae 56. Ancistrussp. 57. Hypoptopoma.fp. 58. Hypo.stomus argus 59. Loricarichthy.ftypus 60. OtocincJus sp. 61. PterygopJichthy.f muJtiradiatus 62. RineJoricaria 4+ .. 1 3+.. 2 1-2 1 12 2 2 12 12 0.68 2 12 .2 12 12 12 12 12 12 1. .80 120 560 Seasonal 755 0.80 42 351 Seasonal 323 .35 ~ Seasonal 3 567 3 323 12 2 I t(XX) 2 2 3 t 2 2 2 2 2.00 3S 297 362 1M 68 Seasonal 392 2J Seasonal Equilibrium 1.35 in 232 Seasonal 2+ ., 238 TSO 0.85 1.90 42 ItS 56 222 Seasonal Seasonal 4+ 178 .00 58 652 Seasonal 38 68 80 68 134 200 292 9 3Ot 208 122 36 Seasonal Seasonal Seasonal Seasonal Seasonal Seasonal 130 19 Seasonal 38 4~ Seasonal 44 56 281 210 Seasonal Seasonal 2 25 218 477 138 Seasonal Equilibrium 4 68 00 190 224 26 233 114 43 230 233 395 138 3+ t 2 2 t t 649 1.10 31SO 0.55 4313 0.75 1587 0.85 111.10 11S8S 1.0S 750 1.27 tl5 323 2.17 0.93 Stratcay 3 3 3 3175 12 Sita. 6+.. 12 1.60 0.86 0.63 1.33 1.57 0.63 0.82 0.99 7-1 2 S 3 7-12 3 3 12 51. Ocltmacanthus aJterm4.f 0.83 family Callichlhyidae 4 12 4 12 3 12 12 12 12 Length N I.~ 12 3 2+ 238 12 12 2 1 1+ 3+ 152 143 1.45 1.25 1 1 2 0.86 0.83 2 2 12 12 2-4 2 3+ 2+ 12 2684 0.70 2.00 0.65 1.08 0.40 0.85 0.88 0.87 2 2 I 1 1 2 12 12 12 12 8 12 4-5 2 3-4 3-4 4 3 4+ 4+ 3+ 5+ 4 3+ 48 52 289 421 52 763 4.00 1.6 3.25 3.05 O.~ 3.50 12 .5 3+ 255 1.70 113 328 Equilibrium 1 1 4-S S 12+.. 12+.. 8 8 1.30 l.3S 33 32 2 126 Opportunistic Opportunistic 0.62 . 4 4 4 caraca.serrsis order Cyprinodontifonnes family Cyprinodontidae 63. Pterolebw hoigMi 1.97 64. Rachot1iD macldipillnis 1.97 family Poeciliidae 2 2 ., 1 .1 .. I 1 Equilibrium Seasonal Equilibrium Equilibrium Seasonal Equilibrium 241 65. PoeciliQreticulatQ 0.97 order Perciformes family Cichlidae 66. Aequidenspulcher 0.55 67. Apistogrammahoignei 0.95 68. AstroMtus oce/latus 1.04 69. Caquetaiakraussii 0.74 70. Cichlasomaorinocense 0.51 71. Crenicichlageayi 0.54 order Synbranchiformes family Synbranchidae 72. Synbranchus 1.08 J .~ 2. 1 1 f , 1 1I 1 t 1 2 2 11 4+ 19 1.90 4 26 1489 3 Opportunistic I 4 14 8 12 12 ~10 6-12 6 11-12 6 6 }+2. 1.75 0.65 2.40 1.85 1.85 1.80 6 4 7 6 6 6 89 30 181 230 112 100 1012 220 106 550 301 106 J 3+ 1+ 2+ 871 59 2301 3702 1287 230 Equilibrium Opponunistic Equilibrium Equilibrium Equilibrium Equilibrium 12 1 2+ ISO .50 2 330 1.5 ., ~+ !}i: j Equilibrium marmoratus The reproductive strategy assignedto each speciesfor the Chi Square analysis appears in the last column. (*1 =Cano Maraca, 2= C. Volcan, 3 = both, 4 = collected from other sites in the llanos; ** conservativelyestimated from distributions of oocyte size classes and duration of breedingseason- actual value is probably many times higher - somespeciesmay evenspawndaily) Appendix2. Reproductivestrategiesassignedto speciesinhabiting lentic and lotic habitats at represaLas Majaguas - Species family Potamotrygonidae Potamotrygon orbignyi family Erythrinidae Hoplias malabaricus family Lebisasinidae Py"hulina lugubris family Prochilodontidae Prochilodus mariae family Anostomidae Leporinus friderici Schizodon isognathus family Characidae Astyanax bimaculatus Astyanax sp. Bryconamericus beta Bryconamericus deuterodontonoides Bryconamericus sp. Colossoma macropomum Creagrutus -'p. cf beni Gephyrocharax valenciae Hemigrammus marginatus Hemigrammus sp. Metynnis sp. cf orinocense Paragoniates alburnus Pygocentrus notatus Roeboidesdoyi . Lentic Lotic Equilibrium Equilibrium Equilibrium Seasonal. Seasonal Seasonal Seasonal Seasonal Seasonal Seasonal* Seasonal Opportunistic * Opportunistic * Seasonal Seasonal Seasonal Opportunistic Seasonal Seasonal Seasonal Seasonal Opportunistic Seasonal Seasonal Opportunistic. Opportunistic inhabitin~ shallow shoreline habitats only Species SerrasaJmusirritans SerrasaJmusrhombeu.f family Auchenipteridae Parauchenipterus gaJeatus family Pimelodidae PseudopJatystomafasciatum family Loricariidae Hypostomus argus Loricaria sp. RineJoricaria caracasensis family Poeciliidae PoeciJia reticuJata family Cichlidae Aequidens diadema Astonotus ocelJatus Caquetia kraussii CichJa ocelJaris CichJa temensis Cichlasoma orinocense Cichlasoma psittacum CichJasomaseverum Crenicichla geayi Crenicichla Jugubris Geophagusjurupari Geophagussurinamensis family Synbranchidae Synbranchus marmoratus Lentic Lotic Seasonal Seasonal Seasonal Seasonal Seasonal Seasonal Equilibrium Equilibrium Equilibrium Seasonal Opportunistic. Opportunistic Equilibrium Equilibrium Equilibrium Equilibrium Equilibrium Equilibrium Equilibrium Equilibrium Equilibrium Equilibrium Equilibrium Equilibrium s~~ Equilibrium Et.wl. S...e."