:>ecologia (989) 8 :5- (k(:i)logia D Springer-Verlag 989 Paerns of variaion in life hisory among Souh American fishes in seasonal environmens Kirk O. WillelDiUer Deparmen of Zoology, The Universiy of Texas, Ausin, TX 787. USA smary. Ten rais relaed o life hisory heory were measured or esimaed for 7 freshwaer fish species from wo locaions in he Venezuelan llanos. Mulivariae saisics and cluser analysis revealed hree basic endpoin paerns bounding a wo-dimensional coninuum. A suie of aribues associaed wih parenal care and aseasonal reprodo<:- ion appeared o correspond o an equilibrium sraegy. A second group of small fshes was disinguished by rais associaed wih rapid colonizing abiliy: early mauraion. coninuous reproducion, and small cluches. The hird basic paern was associaed wih synchronized reprodo<:- ion during he early we season. high fundiy, absence of parenal care. and breeding migraions. A subse of mosly small fishes exhibiing lile or no parenal care. small cluches. and wo o four monh reproducive seasons was inermediae bew he opporunisic (rapidly colonizing) and seasonal sraegies. All en life hisory variables showed significan effecs of phylogeny. The cluser of species corresponding o he equilibrium group was dominaed by siluriform fishes and perciforms of he Cichlidae. The opporunisic cluser was dominaed by cyprinodoniform and characiform fishes. whereas he seasonal cluser conained primarily characiform and siluriform fishes. Seven of nine rais were significanly correlaed wih body lengh. The hree reproducive paerns are inerpreed as being adapaive wih respec o relaive inensiy and predicabiliy of emporal and spaial variaion in abioic environmenal parameers. food availabiliy. and predaion pressure. Key wonls: Llanos - Reproducion - Sraegies - Tropical fishes-venezuela The diversiy of mehods by which organisms reprodx:e has long inrigued nauraliss. One can argue ha rais associaed wih reproducion should be subjec o inense naural selecion, as hese direcly affec he individual's geneic represenaion wihin subsequen generaions. Alernaive modes of reproducion have obvious demographic implicaions, since all populaions experience eiher gradual or periodic urn-overs. Alernaive paerns of reproducion and he degree o which hese are sful in differen environmenal seings have formed he basis for boh heoreical and empirical research under he heading life hisory (Seams 97; Souhwood 977; Horn 978; Sibley and Calow 98a). n he mos basic erms, life hisory heory involves consrains, or rade-offs, among reproducive and demographic parameers (Seams 98a; Reznick 985; Pease and Bull 988). For example, models have incorporaed negaive correlaions beween curren invesmen and fuure invesmens in reproducion, juvenile survivorship and cluch size. reproducive effor and adul survivorship, and mean generaion ime and he inrinsic rae of increase. Alernaive life hisory predicions can be esed by evaluaing paerns of covariaion among reproducive and demographic parameers among heerospeciflc populaions or conspecific demes ha have experienced differen environmenal condiions for long ime periods. The presen sudy consiues one such es. characeridng paerns of reproducion in ropical fishes wih a series of dependen variables. The degree of species clusering in mulivariae space indicaes he likelihood ha cerain combinaions of rais will occur ogeher in naure. Finally, evaluae species clusers for associaion wih phylogeny and ogical facors. Tropical freshwaer fishes exhibi large diversiy in morphological, physiological. and logical aribues (Lowe- McConnell 987). Despie few comparaive sudies of ropical fish reproducive sraegies o dae (Africa-Welcomme 99; Cenral America-Kramer 978; Asia-DeSilva e al. 985), hese diverse fish assemblages provide excellen sysems for evaluaion of life hisory paerns. The grea drainage basins of Souh America harbor he mos diverse freshwaer fish assemblages on earh (ca. described species). Raher paradoxically, his ichhyofauna is derived from only a few basic socks (species belonging o he Characiformes. Siluriformes. and Cichlidae (Perciformes) comprise abou 9% of all species, Lowe McConnell 987). invesigaed fish reproducive biology in he llanos of Venezuela during welve coive monhs in 98. The region has an average annual emperaure of. C and a highly seasonal paern of rainfall, wih 89 mm of precipiaion occurring from June-November and 78 mm from December- May in 98. Despie exreme changes in environmenal condiions caused by seasonal rainfall. fishes exhibied a wide range of reproducive characerisics in he llanos. This repor describes hese characerisics for 7 species and inerpres he basic paerns wihin an logical conex. Mehods Collecions During every monh of 98, fishes were colleced from wo locaions in he sae of Ponuguesa, Venezuela. The
more faunisically diverse sie, Cano Maraca, is a swampcreek locaed in he wesern llanos region (8 5'" La. N; 9 5'" Long. W). The area sudied (ermed an "esero") lies wihin low, fla errain ha experiences exensive shee flooding during he wees monhs (June-Augus). During his ime, he creek's broad floodplain is convered from exposed, sun-baked soil and horn-scrub habia ino a producive marsh. The aquaic habia is reduced o a nework of pools (average deph ca.. m) during he dries monhs (December-May). Dissolved oxygen concenraions are reduced during his ime, and many fishes rely on special respiraory adapaions for survival (Winemiller 988). The oher sie, Cano Volcan (8 59'5" La. N; 9 5'" Long. W), is a hird order sream in he lowes ier of he Andean Piedmon. This narrow sream flows hrough deciduous fores and conains boh pool and rime habias. Cano Volcan differs from C. Maraca in having a slighly seeper gradien, more sable dry season discharge, and a more foresed waershed. Cano Volcan experiences frequen bu brief flash floods during we season downpours. Physico-chemical daa for each sie during he year sudied are given in Winemiller (987). Fishes were colleced by dipne, seine (. mm mesh/.5 m lengh;.7 mm/ m), experimenal gillne, and hook and line. A each sie, an aemp was made o sample he enire fish communiy such ha he sample for each species refleced is relaive abundance and populaion size srucure during each monh. The collecing effor expended each monh was approximaely equal for each sie, consising of hree or four hauls of he m seine in open waer habias, and alernae use of he.5 m seine, dipnes, and gillnes for a minimum of h wihin open waer, shoreline, vegeaion, and peripheral pool habias. Sampling was erminaed when wo hours of collecing yielded no addiional rare species. Collecions used for comparisons of relaive populaion densiies were made during eiher one or wo days beween he h and 8h of each monh. Addiional collecions were made on oher daes in order o supplemenhe numbers of uncommon species aken in he sandard samples. Excep for very abundan species (for which a subsample represenaive of he abundance rank in he collecion was reained), all colleced individuals were preserved in 5% formalin o assure preservaion of viscera and somach conens. Following examinaion, preserved specimens were deposied in he Museo de Hisoria Naural de la Universidad de los Llanos Occidenales Esquiel Zamora, Guanare, Poruguesa and he Naural Hisory Collecion of he Texas Memorial Museum, Universiy of Texas, Ausin. During March, June, July, Sepember, and November of 98, several collecions were made a Represa Les Majaguas, a 5 ha reservoir in wesern Poruguesa (9 '" La. N; 9'" Long. W). Fishes were colleced from he reservoir by seine (. mm/.5 m), dipne, and hook and line. Snorkeling observaions provided an addiional means for deermining he presence of fishes in clear regions of he reservoir. Two regions were sampled: a narrow arm of he lake direcly below an inflow canal (from he Rio Cojedes), and he main body of he reservoir, boh near and offshore. Physico-chemical condiions wihin he main body of he reservoir are relaively consan, while he former sie experiences radial changes associaed wih seasonal rainfall (e.g. urbidiy, deph, and flow rae all increase during he we season). Voucher specimens were pre- served in % fonnalin and deposied in he MHN (UNEL- LEZ) and he TNHC (TMM). The daa Several aribues relaed o reproducion and populaion srucure were measured direcly and oher life hisory rais were calculaed or esimaed from hese. All preserved specimens were idenified and measured for sandard lengh (SL). When available, specimens of each species from each monhly collecion a boh Calio Maraca and C. V - can were disseced. The condiion of he gonad was coded based in is relaive size and color using he following caegories: clear (), ranslucen (), opaque (), small (), medium/small (), medium (), medium/large (), and large (5). The overall gonad code for each individual was equal o (color code+size code) divided by wo, and as a resul, values ranged from. o Alhough he same basic crieria were used in coding male and female gonads, inersexual and inerspecific differences in he size, color, and exure of fully maure gonads were aken ino accoun. Undeveloped ovaries were iny ransparen, globular or lamellar srucures. Following a gradual ransiion, fully-developed fish ovaries were large opaque-yellow or orange srucures having a grainy appearance due o he presence of maure oocyes. Maure ovaries are cylindrical or lie as shees along he laeral walls of he abdominal caviy. mmaure eses appeared as ransparen hreadlike or minue lamellar srucures in conras o he smooh, opaque, milky-whie appearance of maure eses. Fully-maure eses were eiher cylindrical. lamellar. or mulilobed (he laer condiion occurs in cafishes of he Pimelodidae). For several species, gonad codes were regressed agains gonadosomaic indices o es heir reliabiliy as overall indices of gonadal developmen (Fig. ). Diameers of en of he larges oocyes in each maure ovary were deermined using a dissecing microscope and an ocular micromeer. Oocye sizes wihin each examined ovary were classified as exremely uniform, moderaely uniform, moderaely variable, or exremely variable. The number of separae oocye size classes presen in each maure ovary was recorded whenever hese were essenially non-overlapping. The fa conen of he body caviy was coded using he following crieria: = none,.5 = races of fa in connecive issue of he coelomic caviy lining, = small amouns around viscera and races in he connecive issue of he coelomic caviy lining,.5 = moderae deposis around he viscera and a hin layer on he coelomic caviy lining, = large deposis around he viscera and coelomic caviy lining, bu no filling he coelom,.5 = very large deposis filling he coelom. bu no producing a visible bulge, and = coelom packed wih fa deposis producing a bulge of he belly region. Mean sandard lenghs, raio of immaure o adul SL's, mean gonad codes, and mean fa codes were ploed for common species by monh for esimaion of he duraion of breeding seasons. mmaures were defined as sandard lenghs falling below he minimum lengh for which a fully gravid individual was observed among all collecions of a given species. Ten variables relaed o life hisory heory were eiher measured, coded, or esimaed for 59 species from Caiio Maraca and species from Caio Volcan. ) The esimae of annual populaion densiy flucuaion was equal o he coefficien of variaion (CV) of he popula-
7 >< W Q U ( ) Q ( Z )( W Q U C n g C Z '» Asyanax b/msculaus o.a Rhamdis.,, >< w u C ) C Z ".8.-.5. u i= c C). C Z.- " )( U.5 a.-.5..5..5..5..5...5..5..5. GONAD CODE GONAD CODE F"g.. Linear regressions of gonadosomaic index wih gonad code for wo eras and wo cafishes from he Venezuelan llanos. Saisics for each regression are as follows: Bryconamericus bea (females) r=.7. F. =., P<O.(xx); B. bea (males) r=o., F. =., P<.OO; Asyanax bimacujaus (females) r=.8, F. =., P<.(xx); A. bimacuialus (males) r=.8. F. =., P<.(xx); Ancisrus riradiaus (females) r=.8, Fs.=5., P<O.(xx); A. riradialus (males) r=.97, F.=5.7. P<.OOO; Rhamdia sp.l (females) r=.95. F 9. = 9.9, P<.OOO; Rhamdia sp.l (males) r=.95. F. =.7, P<.(XX).5 ion sample N's from he sandard monhly couecions a each sie. ) Mean generaion ime was esimaed as he average number of monhs from ferilizaion o he firs bou of reproducion by new aduls. Esimaes were based on he breeding season, monhly size disribuions, and minimum size of mauraion for each species. For example, mean generaion ime for Curimaa argenea was esimaed as approximaely monhs, since mos reproducion occurred during he firs monh of he we season and mauraion required a period wen in excess of he hree monhs during which offspring were coueced (Fig. ). Average generaion imes of more coninuously breeding species such as Hoplias malabaricus, Roeboides dayi, and Aequidens pulcher (Figs. and ) were more difficul o esimae based solely on SL frequency hisograms, and hus required careful consideraion of growh raes. ) Duraion of he breeding season (in monhs) was esimaed from plos of mean gonad codes, visceral fa, SL's, and relaive proporions of juveniles o aduls by monh a each sie (Fig. ). ) n order o reain as much informaion as possible for saisical ess, chose o make a very rough approximaion of he number of reproducive bous per year, raher han code each species in a conservaive binary fashion (i.e. single versus muliple brooded). Average reproducive bous per year per female was esimaed from plos of mean gonad codes by monh and records of size disribuions of oocyes wihin maure ovaries. These esimaes were very conservaive, paricularly for species ha showed nearly coninuous breeding seasons (some of hese fishes may even spawn on a daily basis during peak reproducive periods). Bias agains he highes values for number of reproducive bous per year due o conservaive esimaes would no grealy affec overall ordinaion of species on he variable (i.e. esimaes of bous per year can be more appropriaely viewed as ranks raher han parameric values). 5) Fecundiy was recorded as he average number of he larges oocye size class based on hree fully-gravid females. For small species, he oal number of oocyes was couned direcly. The number of oocyes from large ovaries was esimaed by couning and weighing 5 maure oocyes (bloed dry, we weigh o neares. g), hen weighing he enire ovary, and solving for he oal number of oocyes. ) Maximum oocye size was equal o he diameer of he larges oocye in fully developed ovaries (neares.5 mm). nraspecific variaion in diameer of he larges oocye in maure ovaries was small, ranging from less han.5 mm in species wih small oocyes o no more han.55 mm in species wih vary large oocyes (e.g. Ancisrus riradiaus). 7) Because ropical fishes exhibi parenal care in a variey of forms, i was quanified as he sum of A + B + C + D from he following:
Hop/as ma/abarlcus Cur/maa.rgenea 5 7 :J 5 7 5 ' 5 "i ' '> :a.5 - $. E z i; '> :c s - OJ.c E z 5 8 7 5 5 5. 8 5 5 5 SL nerval SL nerval Fig.. Sandard lengh frequency hisograms during each monh for Hop/ios malabaricus (exended breeder) and Curimaa argenea (cyclic breeder) a Cano Maraca
9 Roebolde. dayl Aequldens pulcher 5 May -,. 5 i; 'C "> :cc -.. E z '- so " i; 'C "> :c.5 - e.c E z, H. e. o... SL nerval a, 5., SL nerval Fig.. Sandard lengh frequency hisograms during each monh for wo fishes wih exended breeding periods, Roeboides day; and AeQuidens Ducher. a Caiio Maraca
! i u & c..,.. f,,,-\,\ f,f' -( ",f' ---,.----- --, '. J.. "V '" ""/, "! " e-.» E ; uj............ J..M.M"".8MD -- A. J J Monh... 8 i <. "S >.. i.' P"',,,,,! "'--_. " i. '. '. i' '; c ; O. - ". b," '. r. :. V r i,.,. : -. '.... -.m s,'b,.,.........--:$ '"........ J'M.MJJ.8NO,,,r". J'M.MJJ.8ND - :. J ".,. Comparison of reproducive Monh measures of a muliple--spawning characid (Bryconomericw bea) and cicblid (Caqueia kraw.ril) wih a seasonally-spawning characid (Asyonax bimacrdaus) and cichlid (CichlD.roma or) from he Venezuelan llanos (ciord symbob correspond o mean fa code-op poneb. and mean SL-boom b; open symbols correspond o mean gonad code-op pane-b. and raio of juveniles o adjds-boom panels) A = special pac:emen of zygoes () or zygoes and larvae () B = brief period of nuriional conribuion o larvae (), or long period of conribuion o larvae or embryos () C = brief period of parenal proecion by one sex (), or boh (), or long period of parenal proecion by one sex (), or boh () D-exremely long period of gesaion () (applied only o Poamorygon) Parenal care codes ranged from o 8 and, wih few excepions, mos deails of reproducive behavior of a species were documened by field observaions of fishes colleced a he sies, or he scienific and aquarium lieraure. 8) Dry season age disribuion of each populaion was coded as eiher O-No aduls (applied only o wo species of annual k.illifish), -Skewed in favor of juvenile and subadul size classes. -Approximaely even disribuion of immaure and adul size classes, and -Skewed in favor of adul size classes. 9) We season age disribuion of each populaion was coded as eiher,, or as before. ) The maximum sandard lengh among all specimens of each species was measured o he neares. Mm. Maximum lengh was used in analyses raher han average lengh of maure aduls, since i beer reflecs geneic poenial among fishes exhibiing indeerminae growh. For en species ha occurred a boh sies, daa from he sie wih he highes populaion densiy were used in
analyses. No surprisingly, a few life hisory variables revealed inerregional differences wihin a species (hese provide a basis for comparisons in a fuure repor). The freshwaer singray, Poamorygon orbignyi, was colleced from oher sies in he llanos (esado Apure) and included in he cluser analysis, since i has very disincive life hisory characerisics (e.g. an invarian cluch size of wo, vivipariy, and a long gesaion period). Wih he excepion of Prochi/odus mariae (boh sies) and Brycon whiei (C. Volcan), and he freshwaer singray, only species known o be reproducing a he wo sies were included in he analysis. mmaure yearlings, non-gravid, and gravid aduls of Prochi/odus and Brycon were colleced, bu spawning ook place a remoe downsream locaions. A he beginning of he rainy season, long disance migraions (aduls downsream, immaures upsream) are a conspicuous feaure of he ecology of hese species (Lilyesrom 98; Lilyesrom and Taphom 98). f one or wo variables could no be esimaed for a species wih confidence, he value assigned o is closes phylogeneic relaive was used as an esimae of he unknown characer (This was performed for only of 7, or abou % of he observaions in he daa marix). Because he reproducive biology of he genus is disinc and well documened (Wourms 97; L.G. Nico and D.C. Taphom pers. comm.), he annual killifish, Pero- /ebias hoignei, was included in he daa se, even hough only wo individuals were colleced a Cano Maraca during he we season. Analyses A principal componens analysis was compued from he correlaion marix derived from 7 fish species and all life hisory variables in he original daa se (Appendix ). Principal componens analysis is an ordinaion mehod ha allows a mulidimensional swarm of correlaed daa poins o be viewed wihin wo or hree new orhogonal dimensions. When presen, inrinsic paerns wihin he mulidimensional swarm will emerge, generally wihin a plo of he firs wo or hree componens (i.e. independen axes derived from he original dependen variables; Pielou 98). The srengh of each componen is refleced by is associaed eigenvalue, wih values greaer han. being generally acceped as explaining a significan proporion of he variance in he mulidimensional daa se. Correlaions beween he firs wo principal componen scores for each species and he original life hisory variables for each species were compued o provide an index of he relaive imporance of each variable in he ordinaion of species ino reproducive sraegies. n accordance wih recen consideraions of he effec of allomery on life hisory rais (Calder 98; Dunham and Miles 985; Gileman 98), addiional PCA's were compued on modified daa marices wih ) log-ransformed SL and fecundiy and ) wih nine dependen variables correced for SL (residuals from log SL linear regression). To es he hypohesis ha paerns of inerspecific variaion among life hisory rais have a phylogeneic basis (Dunham and Miles 985), a nesed analysis of variance was performed on each of nine life hisory variables, using family nesed wihin order and log lengh as a covariae. A cluser analysis was performed using Euclidean disances based on all aribues among 7 species (singray included). Each of he en life hisory variables was san- dardized (mean =, sandard deviaion = ) prior o compuing Euclidean disances. Clusering was performed by he average linkage mehod (Sokal and Michener 958), where he disance beween wo clusers is he average disance beween observaions and/or clusers. Canonical discriminan funcion analyses were performed o explore paerns of wihin class variaion relaive o beween class variaion among life hisory variables. The ype of we season age disribuions ( saes), dry season age disribuions ( saes) and parenal care ( saes) were used as class variables wih he oher nine rais as dependen variables. The remaining seven life hisory variables conained en or more saes, and were hus reaed as dependen variables for he purposes of canonical discriminan funcion analysis. Resuls Univariae saisics Compared wih emperae zone assemblages (Mabon 98; Wooon 98), he fishes of boh Caiio Maraca and C. Volcan exhibied an exremely wide range of life hisory aribues (Appendix ). Large inerspecific variaion in fecundiy, body size, and he iming of reproducion was apparen wihin mos orders and families. For example, boh aseasonal and seasonally-spawning species were observed in he speciose Characidae as well as he comparaively species-poor Cichlidae (Fig. ). Diameer of maure oocyes ranged from a low of.5 mm in Curimaa argenea o a high of. rom in Ancisrus riradiaus (i.e., excluding 5. mm repored for Poamorygon by Thorson e a. 98). Fory seven species, including Curimaa argenea, Asyanax bimacujaus. Brycon whiei, and Eigenmannia virescens, exhibied virually no parenal care following spawning. Cichlids and loricariid cafishes exhibied highes levels of parenal care. lnercorrelaions among he en life hisory variables appear in Table. Significan large posiive correlaions (r -.5, P<.) were obained for generaion ime wih log SL, log fecundiy wih log SL, and egg size wih parenal care. Significan large negaive correlaions (r.5, P <.) were obained for generaion ime wih lengh of breeding season, reproducive bous per year and we season age disribuion; log fecundiy wih we season age disribuion; and log SL wih we season age disribuion. Tweny six life hisory variable correlaions were negaive and 9 were posiive or zero. When size was facored ou of he analysis using residuals from he log lengh regression, signs and relaive magniudes of correlaions remained essenially unchanged (Table ). Mulivariae saisics When all en variables were included in he daa marix (raw scores for 8 variables plus log fecundiy and log SL), he fls hree principal componens reduced variaion by 7 percen (Table ). The resul was very similar when SL and fecundiy were unransformed (firs hree PC's modelled 7% of oal variaion). The firs axis was influenced more or less equally by seven variables (populaion flucuaion, egg size, and parenal care excluded). The second axis was derived primarily from egg size, parenal care, and log lengh. Populaion flucuaion, lengh of breeding season, dry and we season size disribuions were also significanly correlaed wih PC (Table ). The hird principal compo-
Table. Correlaion marix of absolue life hisory variables for fshes from wo sies in he Venezuelan llanos. Values in paralhee8 are for residuals of regression wih log SL Life hisory rai Flucuaion Generaion ime Lengh breeding season Bous per year Parenal care Dry size disr. We size disr. Flucuaion Generaion ime Lenlh breeding season Bous per Year Fecundiy Egg size Parenal care Dry size disr. We size disr. Maximum lengh x -. -.--.. -.8- -.-- -.7 -. -. ( -.9 x.s** -.**.** O.S* -..9** -.S7** -.S7 :-.9)** (.) (-.) :-.)** (-.S)** (.)** )( (.9)** (-.5)* O." )( (-.9)** -.S** -.59** )(..8 -O.OS..*. -.9 -..* -.7.*...** -.7 -.S. -.7. (-.) (-.).* (-.9) (-.9) (-.) (-.8) (.5)** (-.).* (.7). (.5) (-.5)** (.) (.) (.) (-.75) (.)*. (-.9)** x (-.)** (.5)*. (-.) (.) (-.) (.7)..5** x (-.8)** (:». -.. -.** x (-.) -. -.5 -.5 x.7**.5.. -.7 Mean val. P<O.OS. P<O.Ol..9.78.8 5.9,d.9.9.7. T. Correlaions of ran wo principal componen axes wih he original life hisory variables - PC! PC PC Variable Absolue daa..9. Flucuaion Generaion ime Lengh breeding season Bous per year Fdiy Egg- Parenaicare Dry Si7 disr. We - disr. CumuJai v8riaa...7 PC.. -.5 -.9.. -..5 -.85 Maximum lenlh ----. - PC! PC PC Variable Lenlh-adjused daa PC -....57 -..5.57 -.8 -.8.5 Eigenvalue Cumulaive - - vari8dce PC.9.9 -.5.57..89 -. -. -.75..7...59. -.77 Flucuaion Generaion ime lengh breeding season Bous per year Fecundiy Egg size Parenal care Dry si: disr. We sil:disr. PCl PC "- PC -.7 -. -.8 -.WJ -.8.5. -.9.7 -..8.7 -..7.5.7....7.9 -.5.9.9.97.5 -O.S -.7 ire hisory variable Flucuaion Generaion ime Lengh breeding season Bous per year Log fecundiy Egg size Parenal care Dry disr. We disr. Log lengh. p<o.os, p<o.oos PC.7.8 -.9 -.79.8.5 -.. -.77.9 PC -.7.9..9.5.85.8 -.. -.5..5 nen axis was influenced primarily by populaion flucuaion, dry season age disribuion, and reproducive bous per year. The fourh axis modelled only 7% of he mulidimensional variance and was derived primarily from hree variables: lengh of breeding season (eigenvecor=-.5), log fecundiy (-.57), and egg size (.5). A plo of species by heir scores on he firs wo principal componens reveals a more or less riangular scaer wih apices corresponding o hree fairly disincive suies of characerisics (Fig. 5). Species wih inermediae scores on PC and high scores on PC exhibied he following aribues: well-developed parenal care, prolonged breeding seasons, repeaed reproducion, evenly disribued size classes hroughou he year, long generaion imes large eggs, and large body size. Acxording o sample CV values. hese species ended o exhibi comparaively small-scale flucuaions in local populaion densiy over he course of he year. A foureen species cluser around he firs apex was dominaed by cafishes (Siluriformes) and cichlids (Fig. ). The second apex resuled from low scores on PC and inermediae or low scores on PC (Fig- 5). This region of wo-dimensional componen space is characcrlud by lile or no parenal care, prolonged breeding seasons, re-
5 C\ U c. - - -7-5 - - 5 PC Fig. 5. Plo of Venezuelan fish specie scores on he firs wo principal componens (analysis derived from raw life hisory daa wih log fecundiy and log SL). Symbols group species ino hree broad caegories based on a cluser analysis. Top ellipse bounds an equilibrium sraegy, lef eclipse bounds an opporunisic sraegy, and righ ellipse bounds a seasonal reproducive sraegy. The small ellipse idenifies fishes of small body size ha exhibi a sraegy inermediae beween exreme seasonal reproducion and opporunism Opporunisic; 8 ; d 5 N U D Fig. 7. Cluser diagram of Venezuelan fishes based on Euclidean disances compued from en sandardized life hisory aribues. Roo mean square disances beween species is.7. Species number codes are given in Appendix. Explanaion of hree paerns appears in ex - dominaed dry season populaions, juvenile-dominaed we season populaions, long generaion imes, inermediae o - high fecundiies, small oocyes, and inermediae or large -7-5 - - 5 body size. For he mos par, species locaed near he hird PC apex exhibied large local populaion flucuaions. The 8 Fig.. Plo of ropical fish PCA scores as in Fig. 5, bu wih symbols idenifying species by order. Again, he op apex corresponds o orders Characiformes and Siluriformes exclusively species locaed in he region of he hird apex belonged o he equilibrium sraegy, lef apex o opporunism, and righ (Fig. ). Seveneen characiform and siluriform fishes were apex o a seasonal reproducive sraegy Characiforms;. Siluriforms;. Perciforms; A Cyprinodoniforms; Synbranchiforms diae and low PC and low PC scores; Fig. 5). These iner- inermediae beween he second and hird apices (inermemediae fishes were all comparaively small, wih prolonged, ye disincly seasonal reproducion. Egg sizes were peaed bous of reproducion, an even size disribuion during he we season (for some species during he dry season small, fecundiy inermediae, and parenal care absen or weakly developed in his group ha included Corydoras species, Ochmacanhus alern us, and Odonosilbe Dulcher. as well), shor generaion ime, relaively small cluches, small oocyes, small body size, and inermediae populaion flucuaions. Characiformes and Cyprinodoniformes were predominan among he nine species grouped near he second apex (Fig. ). The hird apex was composed of species wih high scores on PC and inermediae or low scores on PC (Fig. 5). Species in his region of principal componen wo-space were characerized by very lile or no parenal care, shor breeding season, few bous of reproducion per year, adul- Cluser analysis Clusering of 7 species based on Euclidean disances resuled in hree major groupings plus wo single species branches (Fig. 7). The freshwaer singray, Poamorygon orbignyi, was included in his analysis (esimaes for several Poamorygon life hisory parameers were based on he closely relaed P. morro using daa from Thorson e al.
98). The same hree basic life hisory sraegies were fonned by hree large clusers, wih Poamorygon and he characid, Brycon whiei, represening exreme examples of wo of hese basic suies of characerisics. Poamorygon could be considered an exreme case of he life hisory sraegy ha corresponds o long generaion ime, well-developed parenal care, large oocyes, and low fecundiy. The same species (5 families, orders) ha comprised he frs apex from PCA produced his cluser (Figs. 5 and 7). Likewise, Brycon represens an exreme case of he sraegy exhibied by he 8 species cluser (9 families, orders) characerized by seasonal reproducion, high or inermediae fecundiy, no parenal care. and generaion imes from o 8 monhs. This large cluser of seasonally reproducing fishes conained wo subunis ha corresponded o 9 large species wih high fecundiies (e.g. Prochi/odus mariae. Rhamdia spp., and Rhamphichhys marmoraus) and 9 small species wih much lower fecundiies (e.g. Characidium spp., Microglanis iheringi, and Ochmacanhus alernus). An eigh species cluser ( families, orders) corresponded o he shor generaion ime, low fecundiy, muliple-brood sraegy of he PCA (Figs. 5 and 7). Creagruus sp. and Hemigrammus sp. we grouped wih aseasonally-reproducing, fas-mauring species by PCA, ye placed wihin he small body size subuni of he seasonallyreproducing group by average clusering (Figs. 5 and 7). Gephyrocharax valenciae was placed wihin he muliplebrood, fas-mauring group by cluser analysis, ye only slighly nearer o small-sized, seasonally-spawning species by he frs wo principal componens. f one examines he original life hisory variables (Appendix ), i is clear ha all hree groups grade ino one anoher as a coninuum of reproducive sraegies. Analysis of variance The composiion of he hree groups clusered near he apices of Figure 5 is associaed o some degree wih phylogeny. For example, five of he species in he firs apex group (high PC scores) are cichlids (Perciformes), seven are cafishes (Siluriformes), one is a characiform (Hoplias malabaricus), and one is a synbranchiform eel (Synbranchus marmoraus). Likewise, he sa:ond group (low PCl scores) is dominaed by characiforms and cyprinodoniforms (wo killifishes and one livebearer). The hird group (high PCl scores) is composed enirely of eras, cafishes, and knifefishes (Characiformes and Siluriformes). On he oher hand, members of he mos diverse order, Characiformes, were found in all hree groups. Resuls of nesed ANOV A show ha order had a significan effec on all en life hisory variables when lengh was no included as a covariae (Type SS, Table ). When log lengh was included as a covariae (Type SS). he main effec of order on we season size disribuion became marginally insignifican (Table ). Family significanly affeced all life hisory variables excep generaion ime. lengh of reproducive season, and log lengh (Table ). Lengh as a covariae did no aler he fracion of family level ess (family nesed wihin order) ha aained saisical significance (P<O.O5). Because considerable variaion in adul size exised wihin many families (SL range 8- mm for Characidae. - mm for Cicblidae. 5- mm for Callichhyidae. - mm for Loricariidae. 8- mm for Pimelodidae). T.. Resuls of nesed ANDY A for family wihin order wih log lengh as a covariae Trai Class variable Type SS Populaion Flucuaion Order.8. Family (order)..9 Lollengh..97 Generaion Order.98. ime LCDh seuod Bous per year Log fecundiy F Family (order').5. Log lengh.8. Order 5.SO.5 p...s7.8.s..7 5.9.(XX).5 Family (order).98.9.9.s7 Log lengh.9.7 Order.8.(xx) 59.7.(xx).78 Family (order').5.cmm».5.cmm» Log lengh.. Order..CMM»7...88 Family (order).8... Log lengh.. Egg size Order.8. Family (order).7. Log lengh 8.99. 7.98 O.lml.77. O.lml Parenal care Order 7.97. 7.9..9 Family (order).59..58. Log lengh 7.. Dry season Order 8.7..7..7 size disribuion Family (order)..5.7.5 Log lengh.. We season Order 5.8..9.OS.7 size disribuion Family (order).9... Log lengh 55.78. Log lengh Order..5 Family (order)..5 paerns of covariaion among life hisory variables could have been affeced by allomery (Seams 98a; Dunham and Miles 985). Log lengh was significanly correlaed wih seven of nine life hisory variables (Table ). Again, resuls from nesed ANOV A wih log SL as a covariae indicae ha lengh influenced he paern of variaion in only one of 8 ess for phylogeneic affecs on life hisory variables (Table ). Whereas he poenial for a high fecundiy migh be expeced o exhibi an allomeric paern, larger fishes did no always exhibi higher fecundiies han smaller fishes (e.g. Paraucheniperus go/eaus. Hyposomus argus. Loricarichhys ypus, Perygoplichhys muliradiaus, Crenicichia geay,'). Canonical discriminan funcion Resuls of he canonical discriminan funcion analyses indicae ha all hree of he designaed class variables bad significan (P<.<XX>) raios of wibin-class o beclass variance among he remaining nine life hisory variables. Canonical correlaions were.88 (we season size dis-
ribuion),.8 (dry season size disribuion), and.8 (parenal care). For we season size disribuion, he firs canonical eigenvecor had high loadings for log fdiy (.77), egg size (-.58), lengh of breeding season ( -.), and log SL (.5). Taken ogeher, hese daa reveal a paern in which species exhibiingjuvenile-dominaed we season populaions end o have high fecundiies, large body size, comparaively small oocyes, and a brief reproducive season. Conversely, populaions dominaed by aduls during he we season were smaller, less fecund, and had larger oocyes and longer reproducive seasons. Reproducive bous per year (.9) and log fecundiy (-.) had he highes loadings for he firs canonical eigenvecor of dry season size disribuion. n one-dimensional space, species ranged bew adul-dominaed dry season populaions exhibiing repeaed bous of reproducion and low fecundiies, o juvenile-dominaed and unifordl populaions wih relaively fewer bous of reproducion bu larger cluches. The second dry season size disribuion canonical vecor had a correlaion of.77 and highes variable loadings corresponding wih lengh of reproducive season (-.78), parenal care ( -.7), and populaion flucuaion (.). The second canonical vecor discriminaed among dry season disribuion classes along a gradien running from shor breeding season, lile parenal care, variable populaion size, and adul-dominaed dry season populaions a one end, o prolonged breeding. parenal care, more sable populaion densiy, and unifordl dry season size disribuions a he oher. Egg size (.87), lengh of breeding season (.), populaion flucuaion (-.9), and log SL (-.7) had he highes loadings on he firs parenal care canonical eigenvecor. n oher words, species exhibiing greaer levels of parenal care ended o have larger eggs, longer breeding seasons, lower populaion flucuaions, and smaller body size han fishes wih poorly-developed parenal care. Overall, resuls of canonical discriminan funcion reinforced findings ofpca and average clusering by Euclidean disances. Paerns of reproducion appear o be paricularly consisen when a leas wo dimensions are incorporaed ino he model simulaneously. A wo-dimensional gradien having hree disinc endpoin sraegies seems o describe ropical fish life hisory paerns in he simples and mos parsimonious fashion. Small, muliple-cluching fishes exhibiing small invesmens in individual offspring correspond o a life hisory sraegy associaed wih shor generaion imes and rapid populaion urnover. Larger more fecund, seasonally-spawning fishes also inves lile in individual offspring (i.e., small oocyes, no parenal care), bu much more in oal we season reproducive effor. n conras o boh of hese paerns, a group of aseasonally-reproducing fshes had large oocyes, inerdlediae or small cluches, and parenal care of eggs and/or larvae. The las group seems o represen a sraegy associaed wih invesing relaively more in individual offspring. leading o higher survivorship and moderaion of local populaion flucuaions. nally proposed by Pianka (97). Presumably, he suie of characerisics forming his "equilibrium sraegy" is assom.cussion Covar;aion among life hisory variables Even hough mos of he primary measuremens of reproducive and demographic parameers involved housands of examined specimens (Appendix ). populaion flucua- ions, average generaion imes. lengh of breeding season, and number of reproducive bous per year should be considered bes approximaions unil new daa become available. While limiaions of he daa should be aken ino accoun, poenial life hisory rade-offs are implicaed by significan negaive correlaions obained among nine variables ( using lengh adjused values). Egg size and parenal care changed from having zero correlaions wih log fundiy o having significan negaive correiaions when he daa were adjused for lengh (Table ). Because ropical fishes exhibi wide variaion in body form (e.g. ranging from he robus Cich/asoma orinocense o he snake-like Synbrmrchw marmoraw), lengh provided only a rough esimae of body size differences. Even hough gravimeric or volumeric daa could poenially increase he sensiiviy of size-adjusmens, he seven significan lengh correlaions were consisen wih oher comparaive fish sudies (a 978, Mahon 98; Wooon 98). For example generaion ime, fecundiy, and egg size were all correlaed posiively wih body lengh (Table ). Lengh of breeding season. bous of reproducion per year, and he raio of aduls o juveniles during he we season correlaed negaively wih lengh. Several of he observed negaive correiaions should have resuled from fundamenal physiological consrains. Cluch sizes are bound o be lower in species ha exhibi many spawning bous per year, compared o seasonal spawners ha exen oal annual reproducive effon during a consriced ime period. Likewise, higher oal fecundiy is achieved by packing less maer and energy ino individual oocyes. Many of he posiive correlaions among variables would be logically anicipaed as a consequence of secondary consrains imposed by primary negaive consrains (Pease and Bull 988). The difficul ask of assessing which suies of characers migh form adapive life hisory sraegies largely reduces o a problem of criically assessing which aribues consiue key variables responsive o naural selion. ndirec evidence frequenly may be required for idenificaion of crucial rade-offs (Pease and Bull 988). Some variables can be evaluaed singularly wih regard o a variey of environmenal parameers. For example, he iming of reproducion of fishes has been viewed as adapive wih respec o a number of differen environmenal facors (Kranler 978; Johannes 978; Balz 98; Keas 985; and see below). Basic paerns and ecological correlaes Resuls of mulivariae analyses of ropical fish life hisory aribues were robus. yielding similar paerns for daa ses conaining SL as a variable, lengh log-ransformed, daa adjused for lengh effecs, and reaing hree of he en aribues as class variables. Moreover, average clusering and PCA produced nearly idenical groupings among fishes. A plo of he species on he firs wo orhogonal axes resuled in a disribuion wih hree apical cluserings, each having fairly homogeneous life hisory feaures. One suie of life hisory rais always conained inermediae or long generaion imes. large invesmen in individual offspring, delayed mauraion, aseasonal reproducion, and prolonged breeding. To some exen. his associaion of life hisory rais agrees wih he relaive K-sraegy" as origi-
ciaed wih higher juvenile survivorship as resul of greaer parenal invesmen in individual progeny (Sraegies are used here as hypoheses for inerpreing observed paerns as adapive suies of characerisics. Neiher eleology nor cogniion are implied from he presen usage of he erm. cr. Chapleau e al. 988). Wih he larges oocye (diameer esimaed near 5. mm), he smalles cluch (wo offspring), long gesaion period (9- monhs), and lae mauraion (ca. monhs, Thorson e al. 98), he freshwaer singray, PoamorygOll, provides an exreme example of he so called equilibrium sraegy. As a group, cichlids ended o exhibi equilibrium sraegy characerisics (i.e., comparaively large oocyes, brood proecion, and acyclic spawning). Enhanced early survivorship of offspring was indicaed by he presence of numerous juvenile size classes of some cichlids hroughou he dry season (e.g. Caqueia kraussii, Fig. ). -sraegiss apparenly reproduced wih some degree of success, even hough predaion pressure was inense during he early dry season a Calio Maraca. when fishes were encounered a high densiies (Winemiller 987). The wo remaining associaions of life hisory characerisics appear o divide many of he rais comprising Pianka's (97) earlier "r-seleced" suie of aribues. n an aemp no o inroduce new jargon, will refer o he suie of characerisics composed of shor generaion ime, low fecundiy, and minimal invesmen per offspring as he "opporunisic sraegy". The opporunisic sraegy was associaed wih small species (e.g. Poecilia reiculaa. OdonosiJbe pulcher) ha remained reproducively acive despie apparen high juvenile and adul moraliy during he harsh condiions and inense predaion of he dry season. A Calio Maraca, fishes exhibiing he opporunisic suie of aribues buil-up dense local populaions from relaively small pools of adul founders over a six monh period following he iniiaion of rains. This populaion growh was achieved hrough a combinaion of muliple bous of reproducion by older adul survivors wih rapid recruimen of new aduls via rapid mauraion raes. The hird suie of life hisory aribues clearly consiues a "seasonal sraegy", which is characerized by cyclic (ofen annual) reproducion. relaively long generaion imes (usually coinciding wih he reproducive cycle), large cluches. and small invesmen per offspring. Fishes idenified as seasonal-sraegies exhibied a characerisic burs of reproducion wih he early rains, followed by gradual reducions in populaion size due largely o predaion on immaures during he early dry season. Asyanax bimacua- us and Curimaa argenea a Calio Maraca provide good examples of he seasonal sraegy, increasing rapidly from relic, adul dry season populaions o large juvenile populaions in he newly-inundaed swamp floodplain. A major fracion of juveniles produced during his "spring bloom " of producion never survives o mauriy, ulimaely falling prey o birds and predaceous fishes during he early dry season. The seasonal sraegy seems o exploi boh emporal and spaial variaion in qualiy of habias for enhanced juvenile survival and growh. Eigheen of he species grouped wihin he seasonal sraegy exhibied shor-range migraions from he downsream channel ino he upper esero for seasonal breeding, whereas only Hoplias malabaricus, exhibied comparaively less-pronounced local migraions among he equilibrium-sraegiss. The large characiforms, ProchiJodus mariae and BrycOll whiei, exhibied long-disance seasonal migraions. wih aduls moving down from he Andean piedmon o spawn in producive llanos floodplains during he we season. These aduls laer reurn wih yearlings o he piedmon. presumably o escape predaion and he harsh condiions of he floodplains during he dry season. Quaniaive feaures of fish habias were highly variable on a seasonal basis a he sies sudied in Venezuela. and ineresingly, 8 ( families, 5 orders) of 7 species were idenified as relaive seasonal-sraegiss. Th species exhibied gonadal recrudescence during he dry season and a burs of spawning aciviy following he firs heavy rains of he we season. Several seasonally-spawning species appeared o exhibi no more han one or wo bous of reproducion during he firs several weeks following iniiaion of rains (e.g. Asyanax bimaculaus. Triporheus sp.), whereas oher local populaions exhibied susained, ye grealyreduced, levels of spawning for several monhs afer an iniial synchronous burs (e.g. Cenobrycon spijurus. Serrasa/mus irrians). Larval growh and developmen were paricularly rapid and survivorship probably quie high wihin he newly-expanded aquaic environmen. Larval food resources were abundan and adul predaory fishes were a heir lowes densiies during he early we season (Winemiller 987). During he firs hree monhs of he we season, species classified as seasonal-sraegiss dominaed he fish communiy a Calio Maraca, boh in erms of biomass and densiy. As he we season proceeded oward a new dry period, formerly dominan seasonal-sraegiss gradually gave way o species exhibiing opporunisic and equilibrium sraegies. As previously noed, several opporunisicsraegies achieved heir highes densiies (e.g. Roeboides day;. Hemigrammus sp.) during he lae-we/early-dry period of ransiion (Sepember-December) via coninual - ruimen of new individuals, his in spie of increasing predaion pressure and diminishing food resources. As he dry season gave rise o increasingly harsh condiions (January- May), mos local populaions were grealy reduced, paricularly seasonal-sraegiss ha persised as adul remnans of he formerly-dominan local populaions. Several species of opporunisic- and equilibrium-sraegiss coninued reproducion during periods of exremely harsh dry season condiions, alhough juvenile moraliy was probably quie high. Bur e al. (988) hypohesized ha muliple spawning in he neoropical characid, Hyphessobrycon pulchripinnis, increases oal reproducive oupu, which could be adapive a less seasonal, ropical laiudes. Ye many, if no mos. ropical ecosysems experience highly cyclic rainfall. Given he heoreical finding ha rapid mauraion raes maximize he inrinsic rae of populaion increase more efficienly han increasing eiher survivorship or fecundiy (Lewonin 95; Sibly and Calow 98b; and ohers). hypohesize ha colonizing abiliy in he face of inense predaion or unpredicable variaion in qualiy of aquaic habias is he major adapive feaure of he opporunisic sraegy. Mauring rapidly, producing muliple small cluches, and recolonizing ephemeral habias each year, he annual killifishes (Cyprinodonidae) provide compelling illusraions of advanages of he opporunisic sraegy. Resuls from Reznick and Endler's (98) invesigaion of Trinidadian guppies, PoeciJia reiculala, furher suppor his inerpreaion of he opporunisic sraegy. Females from environmens wih greaher hreas of predaion for adul
7 guppies maured a smaller sizes, had shorer inerbrood inervals and allocaed larger fracions of issue o reproducion. The basic life hisory sraegies displayed by neoropical fishes agree well wih he hree paerns ha emerged from Balz's (98) inerspecific comparison of emperae surfperches (Embioocidae). One group of surf perches conained six species having large bodies, moderae o high fecundiies. delayed mauraion, and long life spans. All of hese rais correspond o he relaive seasonal sraegy derived from he presen invesigaion. Likewise, Balz's second group of medium-sized surfperches wih low fecundiy and delayed mauraion mirror many characerisics of he equilibrium sraegy among ncoropical fishes. Moreover, he rais associaed wih his hird group of en small surfperches (i.e., shor-lived, rapidly-mauring, variable cluches) agree well wih he suie of rais describing he opporunisic sraegy. Balz (98) also discussed a rend of higher fecundiy in associaion wih more seasonal environmens. A general hypohesis of life hisory evoluion in response o environmenally-induced variaion in resource availabiliy, predaion pressure, and physiological sress emerges from he wo independen analyses. f ecion derived from environmenal variaion is ulimaely responsible for he evoluion of life hisory rais (Seams 97, 977; Souhwood 977, 988), why do all species a a given sie no respond in a similar manner? For example, Calio Maraca and C. Volcan are highly seasonal environmens, bu only 9% if heir residen fishes were idenified as relaive seasonal-sraegiss. he remainder being eiher opporunisic- or equilibrium-sraegiss. Similarly, Kramer (977) and DeSilva e al. (985) found boh seasonally and coninuously-reproducing osariophysan fishes synopic in small rainfores sreams. The soluion probably lies in he relaionship beween rophic ecology and variaion in resource abundance and predaion pressure. Mos seasonal-sraegiss a he wo sies were omnivorous and insecivorous fishes. The availabiliy of food resources, boh aquaic and erresrially-based, is srongly influenced by asonal rainfall (Winemiller 987). Opporunisic-sraegiss were primarily small fishes wih comparaively broad dies and subjec o inen predaion. Roeboides day;, a faculaive scale predaor, was a parial excepion. Roeboideswiched from preying primarily upon aquaic insecs during he producive rainy season, o a die comprised mosly of scales during he ransiion period preceding peak dry condiions. Many of he equilibriumsraegiss a boh sies were benhic omnivores and piscivores. Relaive o oher rophic guilds, he densiy of adul food resources is less variable wih season for hese species. The period of peak food availabiliy for many adul piscivores (ransiion season) did no correspond wih he period of highes food availabiliy for juveniles (we season). Wih adul food resources available for gamee producion, equilibrium-sraegiss probably produced limied numbers of offspring during he dry season via brood proecion. Reducion in he overall number of breeding fishes during he dry season migh also reduce poenial iner-brood compeiion for limied larval food resources. nformaion on fish species disribuions wihin Represa Las Majaguas was used for an addiional es of he hypohesis ha he seasonal and opporunisic sraegies are more adapive han he equilibrium sraegy in periodicallyvariable environmens. The main body of he lake provides fishes wih a sable habia compared o he shallow coves below he inflow canals ha raise he lake's waer level during he rainy season. For hose species for which did no have daa from Calio Maraca or C. Volcan (N= 5), a relaive life hisory sraegy was assigned based on published informaion on reproducive behavior of he species and sraegies exhibied by closely relaed species from Appendix. For example, all cichlids were assumed o adop a relaive equilibrium sraegy, since all are known o exhibi brood proecion. relaively large oocyes, e ceera. The life hisory sraegies assigned o each species are presened in Appendix. documened 5 relaive equilibrium-sraegiss, four opporunisic-sraegiss, and nine seasonal-sraegiss from he sable lenic region of Las Majaguas. Four equilibrium-. four opporunisic-. and 7 seasonal-sraegiss were colleced from he seasonally-variable canal region of he lake. Cao Maraca, anoher highly seasonal environmen. had equilibrium-. 9 opporunisic-. and seasonal-sraegiss (Appendix ). Chi Square analysis indicaed a significan associaion bewn reproducive sraegy and habia for a x marix <x Z = df, P <.) and a x marix when opporunisic and seasonal sraegies were combined <x Z = 5., df, P <.). All four of he opporunisic-sraegiss colleced from he lenic body of Las Majaguas were aken from shallow, genly-sloping lioral zones where small. less predicable flucuaions in lake level migh be expeced o yield relaively large-scale habia aleraions. Phylogeneic and allomeric consrains Paerns of covariaion wihin he PCA and he hypohesized reproducive sraegy affiliaions of fishes were influenced by boh body lengh (his being inerpreed as an index of size) and phylogeny, paricularly a he ordinal level (Table ). f life hisory rais represen adapaions, and hence are derived from naural selecion, he isolaion of a phylogeneic componen should no be surprising. Paerns of covariaion among life hisory characers are generally clearer when comparisons are made a higher axonomic levels (e.g. o 978; Seams 98a; Dunham and Miles 985; Gileman 98; Dunham e al. 987) han a lower levels (e.g. Searns 98 b). This observaion is consisen wih he view ha life hisory paerns evolve in response o naural selecion, and are no merely phenoypic arifacs correlaed wih oher rais of he organism. Given he appropriae selecive regime, members of widely differen axa someimes exhibi subsequen convergen in life hisory characers (Tinkle e a. 97; Searns 98a; Dunahm e a. 987). The diverse characiform fishes provide excellen examples of life hisory characer divergence. The order Characiformes exhibis morphological and ecological divergen o a degree ha is perhaps unrivalled by any oher animal order (aery 977). Characifonns spanned he enire specrum of observed reproducive paerns. Obviously, his phenoypic variaion represens evoluionary divergen from an ancesral proocharaciform ha exhibied a more limied suie of life hisory rais. Early proocharaciforms could have been relaive equilibrium-sraegiss, akin o Hoplias malabaricus which possesses morphological characers considered primiive for he order. Or perhaps hey were seasonal-sraegiss, as mos species of he superorder Osariophysi appear o be. Eiher way, given sufficien evoluionary ime (esimaed o be -75 mil-
lion yrs. for characiforms, Fink and Fink 98), a sriking diversiy of rais has evolved from wha was cerainly a more resriced suie of characerisics. f congeners among early proocharacifonns had been compared wih one anoher relaive o more disan axa, phylogeneic design consrains migh have been inferred. On a proximae scale, here is every reason o expec ha phylogeneic design consrains are real (Harvey and Mace 98; Dunham and Miles 985; Gileman 98). For precisely his reason, comparisons among disan axa should serve as more appropriae guides for general heories ofljfe hisory evoluion han inraspecific comparisons. According o Searns (98a), "alhough he comparaive mehod may no be a sharp ool, i can be a srong one wih broader scope of applicaion han oher mehods. cerainly leads o a higher level of generalizaion, and o a broader perspecive on wha causes paerns han does he sricly experimenal mehod." Body size is boh consrained and influenced by evoluion of oher life hisory variables. For example, RofT (98, 98) explained over 8fo of he variaion in age of mauraion for eleos fishes by incorporaing growh funcions and a sandard assumpion for a posiive size-fecundiy relaionship ino he model. n he p sudy, body lengh was posiively correlaed wih fecundiy, boh among and wihin higher fish axa. Logically, a fish mus live long enough and assimilae enough resources in order o achieve a large cluch size. Consequenly, a small cluch can be viewed as a logical consequence of small adul body size in many insances (e.g. Poecilia reiculaa. Rachovia maculipinnis, Creagnllus sp.). Obviously, larger organisms have greaer poenial o produce larger cluches (e.g. Prochilodus mariae. Brycon whiei. Rhamdia spp.). Ye based on inerspecifc comparisons of diverse fishes, high fecundiy does no o be a necessary consequence of large adul body size (e.g. Hyposomus argus and Poamorygon orbignyi, wo of he larges fishes in he analysis had average cluches of only 89 and respecively). CUioS Neoropical fishes inhabiing he same highly-seasonal environmens exhibi a variey of differen life hisory paerns. Mulivariae life hisory paerns of individual species spanned a wo-dimensional coninuum bounded by hree basic suies of characerisics. A so-called equilibrium sraegy was associaed wih sedenary local populaions. relaively sable adul food resources, prolonged breeding seasons, and parenal invesmen in individual offspring, which probably resuls in enhanced juvenile survivorship and reduced flucuaions in local populaion densiy. An opporunisic sraegy was characerized by rapid recolonizaion of disurbed habias by small, rapidly mauring, muliple spawning fishes. Mos fishes in he llanos appeared o be associaed wih a seasonalife hisory sraegy ha explois an annual expansion of aquaic and communiy producion. n he exreme case, he seasonal sraegy was characerized by large adul size, high fecundiy, absence of parenal care, and long-disance spawning migraions o producive, we season floodplains. Acknowkdgs. am graeful o E.R. Pianka and D. Reznick for commening on earlier drafs of he manuscrip. Valuable suggesions ha improved he sudy were provided by JJ. Bull. C.M. Pease. M. Kirkparick C. Thomas. J. Hayes, R. Buskirk and L.E. Gilber. especially hank D.C. Taphorn for his hospialiy in Venezuela. The field assisance of D.C. Taphorn. L.G. Nico. A. Barbarino. P. Rodriguez. and N. Greig is glqly appreciaed. D.C. Taphom provided iniial idcnifk:aions of fishes. hank. P. Urriola. R. Schargel. F. Mago-Lcccia. and C. Hubbs for assisance in obaining pcnnis and visas. The Dircccion Adminisraa:ion y Desarrollo Pcsqucro of he Republic of Venezuela provided a collecing permi. am graeful o D. Urriola for kindly allowing me o work. on his propery. Funds were provided by a research gran from he Naional Geographic Sociey and he Texas Memorial Museum. Refa Balz DM (98) Life hisory variaion among female surfpen:hes (Perciformes: Embioocidoe) Environ Bio Fish :59-7 Bur A, Kramer D, Nakasuru K, Spry C (988) The empo of reproducion in Hyphessobrycon pujchripinnis (Cbaracidae), wih a discussion on he biology of 'muliple spawning' in fishes. Environ. Bioi Fish : 5-7 Calder WA, (98) Size, funcion, and life hisory. Harvard Univ. Press, Cambridge, Mass Cbaplaeu F, Johansen PH, Williamson M (988) The disincion bewec;n paern and pfocql in evoluionary biology: he use and abuse of he erm 'sraegy'. Oikos 5: -8 DeSilva SS, Schu J, Kormulder K (985) Reproducive biology of six Barbus species indigenous o Sri Lanka. Environ. Bio Fish :-8 Dunham AE, Miles DB (98S) Paerns of covariaion in life hisory rais of squamae repiles: he effecs of size and phylogeny. Am Na :-57 Dunham AE, Mi DB, Reznic: DN (987) Life hisory paerns in squamae repiles. n: Gans C, Huey RB (eds) Biology of he repilia, V. Ecology B: defense and life hisory. Academic Press, New York Fink SV, Fink WL (98) nerrelaionships of he osariopiysan fishes (Teleosei). Zool J Linn Soc 7: 97-5 Gery J (977) CharCOids of he World. T.F.H. Publ., Nepune, NJ Gileman JL (98) Carnivore life hisory paerns: allomeric, phylogeneic, and ecological associaions. Am Na 7: 7-77 Harvey PH, Mace GM (98) Comparisons be axa and adapive rends: problems of mehodology. n: Kin,' Con. Sociobiology Group (eds) Curren problems in sociobiology. Cambridge Universiy Press, Cambridge, pp - Horn HS (978) Opimal acics of reproducion and life-hisory. n: Krebs Jr, Davies NB (eds) Behavioural ecology: an evoluionary approach. Sinauer Associaes, Sunderland, Mass., W -9 o Y (978) Comparaive ecology. Cambridge Universiy Press, Cambridge Johannes RE (978) Reproducive sraegies of coasal marine fishes in he ropia. Environ Bio Fish : 5-8 Keas A (985) Developmen of dieary lizaions in a summer communiy of juvenile fishes. Environ Bio Fish :- Kramer DL (978) Reproducive seasonaliy in he fishes of a ropi. callream. Ecology 59: 97-985 Lilyesrom C (98) Aspecos de a biologia del coporo (Prochil>du.r morioe). Rev UNELLEZ Cienc Tecnoll : 5- Lilyesrom C, Taphom D (98) Aspecos de la biologia y conservacion de a palambra (Brycon whil') Myers y Weizman,. Rev UNELLEZ Cienc(fecnol : S-59 Lowe--McConnell RH (987) Ecological sudies in ropical fish communiies. Cambridge Univ. Press, Cambridge Mabon R (98) Divergen srucure in fish axocenea of norh emperae sreams. Can J Fish Aqua sa :-5 Pease CM, Bun JJ (988) A criique of mehods for measuring life hisory rade-offs. J Evol Bioi :9- Pianka ER (97) On r- and K-selccion. Am Na :S9-S9
Pielou EC (98) The inerpreaion of ecological daa. John Wiley & Sons, New York Reznick D, Endler JA (98) The impac of predaion on life hisory evoluion in Trinidadian guppies (PoeciJia reicujaa). Evoluion :-77 RofT DA (98) The evoluion of life hisory parameers in eleoss. Can J Fish Aqua Sci : 989-(MX) RofT DA (98) Predicing body size wih life hisory models. Biosci :- Sibley R. Calow P (98a) Physiological ecology of animals: an evoluionary approach. Blackwell Scienific Publ Oxford Sibley R. Calow P (98b) Why breeding earlier is always worhwhile. J Theor Bioi :-9 Sokal RR. Michener CD (958) A saisical mehod for evaluaing sysemaic relaionships. Univ Kansas sa Bull 8: 9-8 Souhwood TRE (977) Habia. he emple for ecological sraegies? J Anim Eco:7-5 Souhwood TRE (988) Tacics. sraegies and emplaes. Oikos 5:-8 Searns SC (97) Life-hisory acics: a review of he ideas. Q Rev Bioi 5 :-7 Searns SC (977) The evoluion of life-hisory rais: a criique of he heory and a review of he daa. Ann Rev Ecol Sys 8:5-7 Searns SC (98a) The influence of si and phylogeny on paerns 9 Searns SC (98b) A naural experimen in life-hisory evoluion: field daa on he inroducion of mosquiofish (Gambusia af.fmis) o Hawaii. Evoluion 7:-7 Thorson TB. Langhammer JK, Oeinger M( (98) Reproducion sand developmen of he Souh American freshwaer singrays, Poamorygon circujoris and P. morro. Env Bioi Fish 9:- Tinkle DW, Wilbur HM, Tilley SO (97) Evoluionary sraegies in lizard reproducion. Evoluion :55-7 Welcomme RL (99) The biology and ecology of he fishes of a small ropical sream. J Zoo58:85-59 Winemiller KO (987) Tess of ecomorphologicaj and communiy level convergence among neoropical fish assemblages. Ph.D. dissern., Universiy of Texas, Ausin Winemiller KO (989) Developmen of dermal lip prouberances for aquaic surface respiraion in Souh American characid fishes. Copeia 989:8-9 Wooon RJ (98) nroducion: sraegies and acics in flsh reproducion. n: Pos OW, Wooon RJ (cds) Fish reproducion: sraegies and acics. Academic Press, New Yor, London, pp - Wourms P (97) Developmenal biology of annual fishes. pre-embryonic and embryonic diapause of variable duraion in he eggs of annual flshes. Exp Zool 8: 89- of covariaion among life-hisory rais in mammals. Oikos (Copenhagen) : 7-87 Submied February, 989 { Acceped May,989 order Myliobaifonnes family Poamorygonidae. Poamorygon orbignyi. order Characifonnes family Eryhrinidae. Hop/ias ma/abaricus.. Hop/eryhrinus.78 uniaeniaus family Lebiasinidae. Characidium sp.. 5. Lebiasina eryhrinoides.. Pyrrhu/ina /ugubris.7 family Anosomidae 7. Leporinus friderici.5 8. Schizodon isognahus.5 family Prochilodonidae 9. Prochi/odus mariae.9 family Curimaidac. Curimaa argenea.7 family Characidac. Aphyocharax a/bumus.9. Asyanax ineger.. Asyanax bimacu/aus.97. Asyanax meae.97 5. Asyanax superbus.. Brycon whiei. 7. Bryconamericus bea. 8. Bryconamericus.5 deuerodonoides 9. Charax gibbosus.. Cheirodonops geayi.9. Corynopoma riisei.8. Creagruu sp..9. CenobrycOll spihlrus.9. Gephyrocharax.85 valenciae ): ) r U ;.) y.. - :). J Z - - + + ; 5 88 8 895 895 5. 8 5..5.5.... :-c 58.5 - S + + + +-- +** S+**. 7. 5 8 5 5 S5 5 5 7 9 895. W 7 8 87 958 8 755 79 8 8 5 9 755 7.5..9..5.5.95.85..75.85.8.75.7 l 9 7 7 9 9 9 8 8 S 7 SO 5 95 85 89 75 5 89. Opporunisic Opporunisic Oppor./Seas. Oppor./Seas.
Species F\<:. Dry disr 5. Hemigrammus.fp... Markiana geayi.7 7. Odono.fiJbe,uk_r. 8. Pygocenrus noaus.9 9. Roeboide.J dayi.8. Serra.sa/mus irrian.s.. Serra.JOlmlLf medini.9. Serra.salmus rhombeus.. TeragOlloperus. argeneus. Triporheus.fp..5 family Gaseroplecidae 5. Thoracocharax.feJlaus.89 order Silurifonnes suborder Gymnoloidei family Apleronoidae. Adonosernarchus.8 devdlanzi family Gymnolidae 7. Gymnous carapo. family Hypopomidae 8. Hypopomusp..95 family Rhamphichlhyidae 9. Rhamphichhys.5 mannoraus family Slemopygidae. EigenmanniD vire.fcerrs. suborder Siluroidei family Auchenipleridae. Enomocorus gameroi.. Parauc-niperus.8 gajeaus family Aspredinidae. Bnocephohl.f.fp..8 family Pimelodidae. Microg/anis iheringi. 5. PimeJodeJJa sp.l.. PimeJodeJa sp..57 7. PimeJodeJJa sp.. 8. Rhamdia.fp.l.8 9. Rhamdia sp..99 family Ageneiosidae so. AgeneiO.fUS viaus.7 family TrichomYCleridae 5. Oclmacanhus ajerm.f.8 family Callichlhyidae 5. Corydora.s -.7 5. Corydara.s.9,yepenirionaJis 5. Corydara.s habro.fus.8 55. HopJO.Jemum Jiorak.8 family Loricariidae 5. Ancisrus sp..5 57. Hypopopoma.fp..8 58. Hypo.somus argus. 59. Loricarichhy.f ypus.85. OocincJusp..88. PerygopJichhy.f.87 mujiradiaus. RineJoricaria. caraca.serrsis order Cyprinodonifonnes family Cyprinodonidae. Perolebw hoigmi.97. RachoiD macldipillnis.97 family Poeciliidae We Gen- Season Bous disr. era. 7- S 7- - + + + Lengh N Sia. Sracay 75.8 5 755.8 5.5 l5 57. 97 M 8 (XX) S 9 J + 8., TSO.5 in.85.9 S 5 + 78. 58 5.. + 9 SO 587 - S8S..55.75.85..S 8 8 8 8 9 9 O 8 75. 9 + 8 8 8 - -5 - - + + + + + + + 5+ + 5 8 8 5 89 5 7.5.5.7....5.5 O..5. 5 5 8 8 9 8 77 8 95 8.,..5 + 55.7 8 -S S + + 8 8. l.s Opporunisic Opporunisic
5. PoeciliQ reiculaq.97 order Perciformes family Cichlidae. Aequidens pulcher.55 7. Apisogramma hoignei.95 8. AsroMus oce/laus. 9. Caqueaia kraussii.7 7. Cichlasoma orinocense.5 7. Crenicichla geayi.5 order Synbranchiformes family Synbranchidae 7. Synbranchus.8 marmoraus J, + 9.9 89 Opporunisic.. f - 8 - - }+-.., + + + + 87 59 7 87.75.5..85.85.8 7 89 8 55 J!}i: Opponunisic + SO.5.5 j The reproducive sraegy assigned o each species for he Chi Square analysis appears in he las column. (* =Cano Maraca, = C. Volcan, = boh, = colleced from oher sies in he llanos; ** conservaively esimaed from disribuions of oocye size classes and duraion of breeding season - acual value is probably many imes higher - some species may even spawn daily) Appendix. Reproducive sraegies assigned o species inhabiing lenic and loic habias a represa Las Majaguas Species family Poamorygonidae Poamorygon orbignyi family Eryhrinidae Hoplias malabaricus family Lebisasinidae Py"hulina lugubris family Prochilodonidae Prochilodus mariae family Anosomidae Leporinus friderici Schizodon isognahus family Characidae Asyanax bimaculaus Asyanax sp. Bryconamericus bea Bryconamericus deuerodononoides Bryconamericus sp. Colossoma macropomum Creagruus -'p. cf beni Gephyrocharax valenciae Hemigrammus marginaus Hemigrammus sp. Meynnis sp. cf orinocense Paragoniaes alburnus Pygocenrus noaus Roeboides doyi. Lenic Loic Species Lenic Loic SerrasaJmus irrians SerrasaJmus rhombeu.f family Aucheniperidae Paraucheniperus gajeaus family Pimelodidae. PseudopJaysoma fasciaum family Loricariidae Hyposomus argus * Opporunisic Opporunisic * Opporunisic * inhabiin shallow shoreline habias only Opporunisic Opporunisic. Opporunisic Loricaria sp. RineJoricaria caracasensis family Poeciliidae PoeciJia reicujaa family Cichlidae Aequidens diadema Asonous oceljaus Caqueia kraussii CichJa oceljaris CichJa emensis Cichlasoma orinocense Cichlasoma psiacum CichJasoma severum Crenicichla geayi Crenicichla Jugubris Opporunisic. - Opporunisic Geophagus jurupari Geophagus surinamensis family Synbranchidae Synbranchus marmoraus s E.wl. S.e."