Download (k(:i)logia Patterns of variation in life history among

(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."