Download Morphological divergence patterns among populations of

Biological Journal of the Linnean Society, 2008, 93, 799–812. With 7 figures
Morphological divergence patterns among populations
of Poecilia vivipara (Teleostei Poeciliidae): test of an
ecomorphological paradigm
JOSÉ L. GOMES JR and LEANDRO R. MONTEIRO*
Laboratório de Ciências Ambientais, CBB, Universidade Estadual do Norte Fluminense. Avenue.
Alberto Lamego, 2000. Campos dos Goytacazes, RJ, cep 28013-600, Brazil
Received 17 December 2006; accepted for publication 30 June 2007
The structure of body size and shape divergence among populations of Poecilia vivipara inhabiting quaternary
lagoons in South-eastern Brazil was studied. This species is abundant throughout an environmental gradient
formed by water salinity differences. The salinity gradient influences the habitat structure (presence of macrophytes) and the fish community (presence of large predators). Size and shape variation within and among
populations was quantified by geometric morphometrics and analysed by indirect and direct gradient ordinations,
using salinity and geography as a framework. Morphological divergence was associated with the salinity gradient.
The evolutionary allometries observed were independent of within-group static allometries. Sexually dimorphic
patterns were observed in size variation and within-population allometries. Specimens from freshwater (higher
predation) sites presented smaller sizes, relatively longer caudal regions, lower anterior regions and a ventrally
displaced eye. These features are consistent with an ecomorphological paradigm for aquatic organisms from
populations subject to intense predation. A process of directional selection is postulated as the most likely force
driving diversification among P. vivipara populations. © 2008 The Linnean Society of London, Biological Journal
of the Linnean Society, 2008, 93, 799–812.
ADDITIONAL KEYWORDS: body shape – body size – evolutionary allometry – geographical variation –
geometric morphometrics – natural selection – predation.
INTRODUCTION
The study of body shape variation in fish populations
subject to environmental gradients has provided interesting examples of natural selection. The most commonly studied agents of selection for body shape
divergence among populations of the same species are
water flow (streams/lakes) (Brinsmead & Fox, 2002;
Langerhans et al., 2003), habitat use/feeding efficiency
(Bourke, Magnan & Rodríguez, 1997; Walker, 1997;
Svanbäck & Eklöv, 2003, 2004), and predation/escape
performance (Walker, 1997; Ghalambor, Walker &
Reznick, 2003, Ghalambor, Reznick & Walker, 2004;
Langerhans & DeWitt, 2004; Langerhans et al., 2004).
Experimental evidence has been reported of feeding
*Corresponding author. Current address: Department of
Biological Sciences, The University of Hull, Cottingham
Road, Hull HU6 7RX, UK. E-mail: [email protected]
habits generating phenotypic plasticity in body shape,
usually associated with the anterior region and the
trophic apparatus (Robinson & Wilson, 1995; Ruehl &
DeWitt, 2005). Body size variation is also correlated
with environmental differences experimented by populations, particularly with predation, as a consequence
of life-history evolution (Endler, 1995; Reznick et al.,
1997). On the other hand, environmental variables,
such as water salinity, can cause nonselective size
variation by influences on the metabolism (Trexler,
1989; Bouef & Payan, 2001).
The findings of shared convergent patterns of
interspecific shape divergence in different aquatic
vertebrate species subject to similar environmental
gradients has allowed the elaboration of an ecomorphological paradigm for the correlation of prey species
body shape with predation pressure (Langerhans &
DeWitt, 2004; Langerhans et al., 2004). This paradigm provides a framework for the interpretation of
© 2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 799–812
799
800
J. L. GOMES JR and L. R. MONTEIRO
body shape divergence patterns and a functional
hypothesis for predation as an agent of selection. This
interpretation, along with quantitative genetics
theory of genetic drift as a null model (Lande, 1979;
Ackermann & Cheverud, 2002; Monteiro & Gomes,
2005), allows for robust assessments of evolutionary
processes acting on interpopulational phenotypic
divergence patterns, even for species not yet exhaustively studied.
The livebearing fish Poecilia vivipara Bloch &
Schneider 1801 is abundant and widespread in the
recently formed (approximately 5000 years ago) quaternary lagoons of Northern Rio de Janeiro State. Its
high tolerance to environmental extremes, particularly salinity and temperature (Trexler, 1989), makes
it one of the few species present in all lentic environments of the region (Bizerril & Primo, 2001). The
lagoons present an environmental gradient as a result
of water salinity, where the water bodies closer to the
ocean present brackish to salt water (sometimes more
saline than the sea), and the water bodies farther from
the ocean present freshwater. The salinity gradient
influences habitat structure (absence of macrophytes
in high salinity; Suzuki, Ovalle & Pereira, 1998, 2002)
and the fish community (freshwater predators absent
from higher salinity sites; Bizerril & Primo, 2001).
Being omnipresent throughout the environmental
gradient, P. vivipara populations are exposed to varied
conditions, such as different predation pressures,
habitat structures, food availability, and physicochemical water characteristics.
A previous study has demonstrated that the magnitude of body shape variation among P. vivipara
populations is high, even for sites within the same
lagoon, but with different environmental characteristics (Neves & Monteiro, 2003). Divergence rate tests
applied to morphological differences among population means with known maximum times of separation
indicated directional selection as the most probable
process influencing shape diversification among
P. vivipara populations (Monteiro & Gomes, 2005).
Although suggestive of natural selection, these previous results lack the determination of possible agents
of selection to explain the body shape divergence
observed.
The present study aimed to evaluate possible evolutionary mechanisms and causal factors influencing
the morphological variation among populations of
P. vivipara exposed to an environmental gradient
related to water salinity and predation. Accordingly,
we compared observed divergence patterns with evolutionary neutral expectations from spatial analysis
and quantitative genetics. We also compared our
results with proposed ecomorphological paradigms
associated with different agents of selection
(predation, habitat structure, food availability) aiming
to identify convergent patterns of shape changes that
can be used as evidence of natural selection.
MATERIAL AND METHODS
STUDY AREA
The quaternary plains at Northern Rio de Janeiro
State (Fig. 1) were formed by sediment deposition
during formation of the Paraíba do Sul River delta,
and sea level fluctuations associated with palaeoclimatic modifications during the Holocene (starting
approximately 5100 years ago; Martin et al., 1997).
The construction of the river delta continuously
formed river arms, which were later abandoned and
became long, narrow lagoons scattered all over the
region (Martin et al., 1997; Soffiati, 1998; Primo, Bizerril & Soffiati, 2002). Historically, there was considerable contact among the lagoons and the main river
body (particularly during floods). However, the construction of a large network of drainage and irrigation
channels by the National Department of Work
Against Drought (DNOS) after 1950, considerably
decreased the water level in the region and isolated
most lagoons from all others (Primo et al., 2002).
The distance from the sea determined a considerable
amount of variation in environmental conditions
(related to water salinity differences) of different
lagoons, as well as environmental gradients within
lagoons (Suzuki et al., 1998, 2002).
The Feia and Campelo lagoons are the oldest in the
region. Their origin is related with river sediments
deposited during the first development of the Paraíba
do Sul River delta southwards, starting 5000 years
ago (Martin et al., 1997; Soffiati, 1998). Subsequently,
the main river mouth was displaced towards the east
and grew in that direction, mostly by marine sediment
deposition. The eastern lagoons (such as Açu, Grussaí,
and Iquipari), closer to the ocean were mostly formed
by abandonment of ancient river arms (Primo et al.,
2002). These are younger lagoons (up to 2000 years)
and present a salinity gradient, decreasing from the
sites close to the ocean (the sand bars) towards the
interior of the continent. The salinity gradients influence the composition of the fish community in the
region (Bizerril & Primo, 2001). Hoplias malabaricus,
the main predator of P. vivipara (Mazzoni & IglesiasRios, 2002), is a freshwater species absent from higher
salinity sites (Garcia et al., 2003). The marginal vegetation (macrophytes) is usually absent from higher
salinity sites and abundant in freshwater sites
(Suzuki et al., 1998, 2002). The mean salinities of
collection sites are listed in Table 1.
FIELD
AND LABORATORY METHODS
The specimen collections were carried from May 2002
to September 2004. Specimens were collected with a
© 2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 799–812
MORPHOLOGICAL DIVERGENCE AND ECOMORPHOLOGY
801
Brazil
o
40 W
41oW
22oS
Rio de Janeiro
4
7
18
22
15
23
9
26
8
6
10
12
16
17
11
24
19
20
25
3
13
2
1
21
5
14
0
5
10
N
20
km
Figure 1. Map showing the geographical distribution of sampling sites in Northern Rio de Janeiro State. Numbering
corresponds to Table 1.
pole seine (1-mm2 mesh), humanely killed in iced
water, fixed in 10% formalin an stored in 70% alcohol.
From each site, a number of specimens (N = 34–60) of
each sex were used (Table 1), summing up to 1376
males and 1538 females. Only adult animals (sexually
mature) were included in the sample. All females
were carrying developing embryos and the males had
a fully developed gonopodium and visible secondary
sexual characters. Where the lagoons presented a
salinity gradient, two or more sites were sampled
at different positions along the gradient. Salinity in
each site was measured with a densimeter (Seatest,
© 2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 799–812
802
J. L. GOMES JR and L. R. MONTEIRO
Table 1. Samples of Poecilia vivipara collected in the lagoons
Sample sizes
Number
Site name
Females
Males
Salinity (ppt)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Açu Lagoon - Interior
Açu Lagoon - Bridge
Açu Lagoon - Sandbar
Barra do Itabapoana
Barra do Furado
Bom Jardim
Buena
Cacimba
Campelo Lagoon
Cataia Lagoon
Cima Lagoon
Fazenda Frecheira
Feia Lagoon - Ponta Grossa
Feia Lagoon - Canto do Sobrado
Três Vendas
Grussai Lagoon - Sandbar
Grussai Lagoon - Interior
Imburi
Iquipari Lagoon - Sandbar
Iquipari Lagoon - Interior
Lagamar
Praia Lagoon - Site 2
Praia Lagoon - Site 1
Taí Grande Lagoon
Taí Pequeno Lagoon
Limpa Lagoon
60
60
60
60
60
60
60
60
60
59
60
60
60
60
60
60
59
60
60
51
59
60
60
52
59
59
46
60
50
60
60
60
35
60
60
60
59
60
59
60
60
34
60
60
50
54
60
57
60
40
52
–
13*
26.9*
30.5*
12
16
0
0
0
0*
0
0
0
0
0
0
7*
2.3*
0
11.1*
1.7*
5
0
0
0
0
0
Numbers shown correspond to numbering in figures. Salinities marked with an asterisk represent the means of a 2-year
time series.
Aquarium Systems). For some lagoons (Table 1), a
2-year monthly time series was available from an
independent study (M. S. Suzuki, unpubl. data), and
a mean salinity was calculated for each site. The
specimens were photographed by a digital camera
(Pixera Professional) and the coordinates of 13 landmarks (Fig. 2A, B) were registered for each individual, using the TpsDig software (Rohlf, 2006).
Because of the conspicuous sexual dimorphism in
body size, coloration and the shape difference caused
by the anal fin modification in males (to form the
gonopodium), sexes were analysed independently.
GEOMETRIC
AND MULTIVARIATE METHODS
Body size was estimated as centroid size: the square
root of the summed squared distances from each
landmark to the configuration centroid, which is the
average of all landmarks (Bookstein, 1991; Monteiro
& Reis, 1999). Body size variation among sites
was assessed by analysis of variance (ANOVA) and a
Figure 2. Landmarks digitized on specimens for males
(A) and females (B).
© 2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 799–812
MORPHOLOGICAL DIVERGENCE AND ECOMORPHOLOGY
regression of mean centroid sizes on the mean salinity
values for each site.
The landmark configurations were superimposed
separately for males and females by Procrustes superimposition (Rohlf & Slice, 1990). This least squares fit
effects linear transformations of similarity (scaling,
translation and rotation), to minimize the summed
squared distances among corresponding landmarks in
a sample, allowing for the estimation of a mean shape
(Monteiro & Reis, 1999). The aligned co-ordinates
were projected to the space spanned by partial warps
and uniform components (Bookstein, 1991), calculated using the grand mean shape (over all samples)
as reference.
The among-site variation was assessed by submitting the matrix of shape variables (partial warps and
uniform components) to a canonical variates analysis
(Monteiro & Reis, 1999). Minimum spanning trees
were calculated from Mahalanobis distances and combined to the scatterplot of canonical scores to provide a
more accurate view of the shape similarities and to
check the efficiency of dimensionality reduction. The
importance of allometric patterns of shape differentiation to the observed among-site divergence patterns
was assessed by a multivariate analysis of covariance
(MANCOVA), using partial warps and uniform component scores as dependent variables and centroid size as
covariate. The allometric patterns influencing major
axes of variation within-samples were assessed by
regressions of relative warps (principal components of
the shape space spanned by partial warps and uniform
components) on centroid size.
The spatial pattern of morphological differentiation
was assessed by a comparison of shape distances
(Procrustes distances) among sites with geographical
distances. The Procrustes distances represent the
metric of the shape space used for statistical comparisons using geometric morphometric methods (Dryden
& Mardia, 1998). The Procrustes and geographical
distances were compared directly by Mantel (matrix
correlation) tests and by multivariate Mantel correlograms (Legendre & Fortin, 1989; Legendre, 1993).
The correlograms divide the distance matrices by
classes belonging to different intervals of geographical distances (lags) and calculate a matrix correlation
for each interval. Bootstrap confidence intervals for
the matrix correlation estimates were calculated with
10 000 replications. Partial Mantel tests and correlograms were calculated by removing the influence of
salinity from shape distances. A matrix of salinity
differences was calculated using the module of the
salinity difference between each pair of sites. The
matrices of Procrustes distances obtained for different
sexes were also compared by matrix correlation.
The association of shape variables (mean site scores
of partial warps and uniform components) with mean
803
water salinity and mean centroid size was assessed by
a two-block partial least squares analysis (Rohlf &
Corti, 2000). This analysis searches for pairs of vectors
(shape space-variable space) explaining maximum
covariance between the two blocks of variables
(Monteiro, Duarte & Reis, 2003). To assess the existence of a linear spatial gradient, the latitude and
longitude for each sampled site were included in the
variable matrix. The significance of covariance
explained was tested by a permutation test (10 000
replications). The proportion of shape variation
explained by the partial least squares vectors was
assessed by the coefficient of determination for the
multivariate regression (Monteiro, 1999) of shape variables (partial warps and uniform components) on
partial least squares vector scores.
A test for the hypothesis that a neutral process
(genetic drift) would be sufficient to explain the amongsite shape variation patterns was performed (Ackerman & Cheverud, 2002; Marroig & Cheverud, 2004).
This test is based on the prediction from quantitative
genetic theory that, under genetic drift, the amount of
divergence among populations should be proportional
to the variation in the ancestral population (Lande,
1979), which can be estimated by the pooled withingroups additive genetic covariance matrix or its surrogate, the pooled within-group phenotypic covariance
matrix. A large amount of evidence is available in the
literature supporting the use of phenotypic covariance
matrices as surrogates for genetic covariance matrices
(Cheverud, 1988; Roff, 1996). To compare the proportionality between magnitudes and patterns of variance
distribution in among and within-group matrices, we
calculated eigenvalues of within-group and amonggroup covariance matrices of shape variables. A regression of log-transformed among-group eigenvalues on
log-transformed within-group eigenvalues was performed and 95% confidence intervals (CIs) for the
regression coefficients were calculated (Ackerman &
Cheverud, 2002). Regression coefficients equal (or not
statistically different than) to 1 indicate an adjustment
to the expectation of genetic drift. Regression coefficients significantly larger than 1 indicate a process of
directional selection influencing morphological diversification because there is more variation amonggroups than expected by the genetic drift model
(Marroig & Cheverud, 2004).
The software used for the geometric analyses
were the TPS series (TPSRelw, TPSPLS, TPSRegr;
Rohlf, 2006). The canonical variates analyses and the
calculations of among-groups and within-group covariance matrices were performed on NTSYS-pc (Rohlf,
2000). ANOVAS and MANCOVAS were performed
with the package ‘stats’, which is part of the R-system
(R Development Core Team, 2006). The matrix correlations and correlograms were calculated with the
© 2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 799–812
804
J. L. GOMES JR and L. R. MONTEIRO
add-on package ‘ecodist’ (Goslee & Urban, 2006) in
the R-system.
RESULTS
BODY
SIZE VARIATION
10
The ANOVAs showed that mean differences in body
size, were highly significant for males (F = 168.22,
P < 0.0001) and females (F = 144.67, P < 0.0001) from
different sites. Body sizes were larger in lagoons
with salinity gradients (Açu, Iquipari, and Grussai
Lagoons). For both sexes, samples from sand bar sites
showed larger mean body sizes than samples from
interior sites (Fig. 3). Within- and among-group size
variation was larger for female than for male samples
(Fig. 3). A positive association of mean body size and
mean salinity was observed in both sexes. The estimated regression equations (using only site means
8
6
4
2
Centroid Size
A
0
5
10
15
20
25
30
25
30
10
Salinity (ppt)
8
6
4
2
Centroid Size
B
0
5
and depicted in Fig. 3) were: female mean size = 3.73 +
0.09 ¥ Salinity (F = 63.52, R2 = 0.7342, P < 0.0001);
and male mean size = 2.96 + 0.07 ¥ Salinity (F = 60.4;
R2 = 0.7242, P < 0.0001).
10
15
20
Salinity (ppt)
Figure 3. Scatterplots showing the association of centroid
size with salinity for different sites. A, female data set. B,
male data set. Dashed line indicates the least squares
regression.
BODY
SHAPE VARIATION
The ordination of samples by canonical variates was
significant for both sexes. In the female data set
(Wilks’ lambda < 0.0001; F = 26.572; P < 0.0001), the
first two canonical axes depict 44.87% of variation among sites (Fig. 4). The first canonical axis
(representing 32.97% of variation) ordinate samples
from freshwater sites with positive scores and saltwater sites with negative scores. Shape differences
associated with freshwater sites in the first axis are a
more ventrally positioned eye, a more dorsally positioned pectoral fin, an anteriorly positioned (and relatively larger) mouth, a lower body height, and a
longer caudal peduncle (the quadrilateral formed by
landmarks (#10, #11, #12, #13; (Fig. 4). The second
canonical axis (representing 11.90% of variation) does
not ordinate sites in a clear spatial or environmental
pattern (Fig. 4). The minimum-spanning tree connects samples which are closest to each other in the
space defined by Mahalanobis distances (i.e. the
whole set of canonical variates). The connection of
sites that appear distant in the two-dimensional scatterplot, such as sites #13 (Ponta Grossa) and #14
(Canto do Sobrado), reflects the loss of information
during the projection from a multidimensional space
to a two-dimensional scatterplot.
In the ordination of the male data set (Wilks’
lambda < 0,0001; F = 21 055; P < 0,0001), the two first
canonical axes explain 45.15% of variation among
sites. The first canonical axis (explaining 33.39% of
variation) ordinates samples from saltwater sites with
positives scores and freshwater sites with negative
scores (Fig. 5). Samples from freshwater sites (negative scores present a more anteriorly positioned mouth,
ventrally displaced eye, dorsally displaced pectoral fin,
lower body height, and longer caudal peduncle. The
second axis (explaining 11.76% of variation) does not
ordinate sites in a clear spatial or environmental
pattern (Fig. 6). The minimum spanning tree suggests
that the reduction of dimensionality caused a smaller
loss of information than in the female data set.
The importance of allometric patterns to the
observed differences among sites was assessed by a
MANCOVA, using centroid size as a covariate and
partial warps and uniform components as variables.
The significant results indicate that the differences in
shape for males and females from different sites were
not fully allometric (Table 2). However, the interaction term of Size ¥ Site (test for parallelism) is significant, indicating that the allometric trajectories in
© 2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 799–812
MORPHOLOGICAL DIVERGENCE AND ECOMORPHOLOGY
805
2.58
8
Canonical Variate 2 (11.90%)
22
15
5
1.19
11
26
4
23
12
19
1
17
-0.21
6
7
13
21
18
3
16
2
9
14
20
24
-1.60
10
25
-3.00
-4.00
-2.03
-0.05
1.92
3.89
Canonical Variate 1 (32.97%)
Figure 4. Canonical variates analysis for the female data set. Numbers follow the sequence in Table 1 and Fig. 1. Grey
lines correspond to minimum spanning tree from Mahalanobis distances. Grey dots correspond to brackish and saltwater
sites, whereas black dots correspond to freshwater sites. Larger grids below the scatterplot correspond to deviations in
shape associated with positive (right) and negative (left) scores of the first canonical axis. Smaller grids on the side
correspond to deviations in shape associated with positive (upper) and negative (lower) scores of the second canonical axis.
Table 2. Multivariate analysis of covariance models for among-site shape differences (partial warps and uniform
components) using centroid size as covariate
Females
Size
Site
Size ¥ Site
Males
Size
Site
Size ¥ Site
d.f.
Wilks’ lambda
F
Numerator d.f.
Denominator d.f.
P (> F)
1
25
25
0.13
0.001520
0.35
437.45
21.24
2.93
22
550
550
1 465
24 340
24 340
< 2.2 ¥ 10-16
< 2.2 ¥ 10-16
< 2.2 ¥ 10-16
1
24
24
0.161
0.004
0.452
309.127
16.672
2.020
22
528
528
1 305
21 300
21 300
< 2.2 ¥ 10-16
< 2.2 ¥ 10-16
< 2.2 ¥ 10-16
d.f., degrees of freedom.
shape space are not parallel among sites (Table 2).
The significant interaction impedes an interpretation
of the Sites term independently from Size. Separate
regressions were adjusted for each relative warp with
centroid size. For the female data set, the single axis
with a significant relationship with size was the first
relative warp, which explained 29.87% of total shape
variation (i.e. considering the entire data set without
regard of sample site). No relationship between the
regression coefficients and salinity was observed
(Fig. 6). The male data set presented association only
between the second relative warp (explaining 13.21%
of total shape variation) and centroid size. Again,
although there is considerable variation on steepness
of trajectories among sites, no relationship between
allometric coefficients and water salinity was
© 2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 799–812
806
J. L. GOMES JR and L. R. MONTEIRO
2.83
25
Canonical Variate 2 (11.76%)
24
1
1.55
3
4
13
7
15
14
0.27
9
18
17
10
20
21
12
23
-1.02
11
6
5
8
2
16
19
22
-2.30
-3.00
-1.25
0.50
2.25
4.00
Canonical Variate 1 (33.39%)
Figure 5. Canonical variates analysis for the male data set. Numbers follow the sequence in Table 1 and Fig. 1. Grey
lines correspond to minimum spanning tree from Mahalanobis distances. Grey dots correspond to brackish and saltwater
sites, whereas black dots correspond to freshwater sites. Larger grids below the scatterplot correspond to deviations in
shape associated with positive (right) and negative (left) scores of the first canonical axis. Smaller grids on the side
correspond to deviations in shape associated with positive (upper) and negative (lower) scores of the second canonical axis.
observed (Fig. 6). Because the samples were composed
by adults, the allometric patterns observed should be
considered static allometries (non-ontogenetic).
The comparison of shape distances (Procrustes)
with geographical distances by matrix correlation was
marginally significant for both sexes. The full matrix
correlation for the female data set was r = 0.187,
P = 0.0601 whereas, for the male data set, the correlation was r = 0.171, P = 0.0584 (tests based on 10 000
permutations). The Mantel correlograms (not shown)
presented significant spatial autocorrelation only at
small geographical scale (less than 16 km for females
and less than 6 km for males). This geographical
range limits the significance of shape-spatial correlation to sites within the same lagoon. The observed
95% CI for matrix correlations in the significant distance class was in the approximate range 0.05–0.20 in
both sexes. The partial matrix correlations and correlograms were all nonsignificant, after removing the
influence of salinity differences among sites from
shape distances. The matrix correlation between
shape distances and salinity differences was r = 0.418,
P = 0.0019 for the female data set and r = 0.379,
P = 0.0001 for the male data set. The comparison of
Procrustes distances among sites obtained for males
and females was significant (r = 0.324, P = 0.0013),
suggesting a moderate correlation among divergence
patterns based on different sexes.
The test for genetic drift performed, using amonggroup and within-group covariance matrices for shape
variables showed regression coefficients significantly
higher than 1 both for females (b = 1.867, 95%
CI = 1.692–2.041) and males (b = 1.862, 95% CI =
1.563–2.161). This result is inconsistent with the
expectation of the genetic drift model, suggesting a
process of natural selection behind the morphological
diversification among sites.
The partial least squares analysis for among-site
shape variation showed a significant pattern of association of shape with size, salinity, and geographical
coordinates for both sexes. For the female data set,
93.78% of total covariances among the two groups of
variables was explained by the first pair of partial least
squares vectors (PLS vectors). The amount of covariance explained was significant in the permutation test
(P = 0.0012). For the male data set, the first pair of PLS
vectors explain 92.52% of total covariance among
variable blocks (permutation test P = 0.0062). This
pair of vectors suggest an interpopulational divergence
pattern associated with mean body size and water
© 2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 799–812
MORPHOLOGICAL DIVERGENCE AND ECOMORPHOLOGY
807
2 4 6 8 10
A
25
26
19
20
0.05
0.00
-0.05
21
22
23
24
0.05
0.00
Relative Warp 1
-0.05
13
14
15
16
17
18
7
8
9
10
11
12
0.05
0.00
-0.05
0.05
0.00
-0.05
1
2
3
4
5
6
0.05
0.00
-0.05
2 4 6 8 10
2 4 6 8 10
2 4 6 8 10
Centroid Size
2
B
Relative Warp 2
0.04
0.02
0.00
-0.02
-0.04
-0.06
0.04
0.02
0.00
-0.02
-0.04
-0.06
3
4
5
6
2
3
4
5
6
21
22
23
24
25
16
17
18
19
20
11
12
13
14
15
6
7
8
9
10
1
2
3
4
0.04
0.02
0.00
-0.02
-0.04
-0.06
0.04
0.02
0.00
-0.02
-0.04
-0.06
5
0.04
0.02
0.00
-0.02
-0.04
-0.06
2
3
4
5
6
2
3
4
5
6
2
3
4
5
6
Centroid Size
Figure 6. Within-group association of major axes of variation and centroid size. A, female data set comparing the first
relative warp and centroid size. B, male data set comparing the second relative warp and centroid size. Lines correspond
to least squares regressions.
salinity differences. The shapes associated with positive values (higher salinity sites) in the first PLS vector
pair present a relatively smaller mouth (and turned
upwards), dorsally displaced eyes and opercles, ven-
trally displaced pectoral fin, higher bodies, and a
shorter caudal peduncle than the shapes from freshwater sites, with negative scores (Fig. 7). These shape
differences are similar to the first canonical axis. The
© 2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 799–812
808
J. L. GOMES JR and L. R. MONTEIRO
Females
Males
Easting
Size
Salinity
Northing
Figure 7. Partial least squares (PLS) results for the association of site mean shapes and geographical-environmental
variables. The grids to the left represent the shape deviations along the first PLS shape vector for females. The grids to
the right represent the shape deviations along the first PLS shape vector for males. Upper grids represent positive
deviations and lower grids represent negative deviations. Arrows in the middle represent the correlations of variables
with the first PLS variable vector, scaled so that the distance of box frames to the horizontal line in the middle correspond
to a correlation of 1.
PLS vector for variables associated with the shape
vector indicates a linear geographically orientated
(southeast–north-west) salinity gradient, influencing
mean body sizes and shapes (Fig. 7). A multivariate
regression of shape variables (partial warps and
uniform components) on the PLS shape vector scores
was used to calculate the percentage of shape variation
explained by the PLS (as the multivariate coefficient of
determination R2 based on Procrustes distances). For
the female data set, the scores of the first PLS shape
vector explain 41.1% of total shape variation among
mean site shapes whereas, for the male data set, the
scores of the first PLS shape vector explain 39.1% of
total shape variation among mean site shapes.
DISCUSSION
The pattern of morphological diversification of P. vivipara populations during colonization of holocenic
lagoons northern Rio de Janeiro is complex and certainly influenced by the mosaic of environmental differences as observed previously (Neves & Monteiro,
2003; Monteiro & Gomes, 2005). The environmental
differences among lagoons are mainly influenced by
water salinity with noticeable consequences on
habitat structure (absence of macrophytes on sites
with higher salinity) and the structure of fish com-
munities (Bizerril & Primo, 2001). The largest and
more abundant predator sharing the same habitat as
P. vivipara is the trahira or wolf fish Hoplias malabaricus (Erythrinidae). Juvenile trahiras (sometimes
small adults) were regularly collected along in the
same seines we used for P. vivipara, but only in freshwater sites (where trahiras are even exploited commercially by local fishermen). The association of
H. malabaricus and P. vivipara as predator and prey
has been reported before, and P. vivipara is one of the
most frequent items in H. malabaricus diet in Rio de
Janeiro (Mazzoni & Iglesias-Rios, 2002). In a fish
community in the venezuelan llanos, juveniles and
subadult trahiras were the most important predators
of Poecilia species (Winemiller, 1989). Hoplias malabaricus is considered a primary freshwater species
and usually absent from sites with brackish to saltwater (Garcia et al., 2003). The only possible predators found in our brackish water collections were
small cichlids (Geophagus brasiliensis and Cichlasoma facetum), which were observed to eat young, but
not adult P. vivipara in the laboratory (J. L. Gomes Jr
& L. R. Monteiro, pers. observ.). Poecilia vivipara
populations from freshwater sites are certainly
exposed to a high predation environment (that partly
explain the size differences observed). On the other
hand, water salinity can have a direct influence on fish
© 2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 799–812
MORPHOLOGICAL DIVERGENCE AND ECOMORPHOLOGY
growth, by reduction of basal metabolic rates, increase
in feeding and food conversion rates and optimization
of osmoregulation in certain salinity levels (Bouef &
Payan, 2001). Therefore, it is important to distinguish
between direct and indirect salinity effects on size
variation and evolutionary allometry. Because the
observed patterns of morphological differentiation
were generally similar in males and females, we first
discuss the shared patterns between sexes and later
point out the dimorphic patterns
BODY
SIZE VARIATION
Our results show that mean body size is positively
associated with water salinity, which could be a direct
salinity effect on growth (Boeuf & Payan, 2001), an
indirect salinity effect because of predation pressure
in different fish communities, or both. In other livebearing species, individuals from low predation sites
are usually larger than those from high predation
sites (Endler, 1995; Johansson, Turesson & Persson,
2004). Results from a common garden experiment
with P. vivipara populations (J. L. Gomes Jr & L. R.
Monteiro, unpubl. data) indicate that individuals
from a brackish water lagoon presented higher
growth rates (and mean body size) than individuals
from a freshwater lagoon even when raised in freshwater (the observed size differences are not due to
plasticity alone).
BODY
SHAPE VARIATION
The major axes of shape variation among sites were
similar in males and females with respect to the
ordination patterns and shape changes described.
Because the ordination of sites is influenced by water
salinity, which has been shown to influence body size,
it is important to assess the importance of allometric
patterns to the observed among-site shape variation.
The results of the MANCOVA indicate that differences
among sites are maintained even after removing
the statistical influence of size (using a separate slopes
model). However, the within-group allometric patterns
were not homogeneous (nonparallel allometric trajectories), which suggests an uncoupling of within-group
(static) and among-group (evolutionary) allometries.
The allometric differences observed among sites,
where specimens presented different mean sizes, were
not just simple extrapolations of a common allometric
curve. We used the term static allometry (Cheverud,
1982) for the within-group allometry because the
samples were composed exclusively of reproductive
adults, but it is possible that the within-group allometric trajectories incorporate some ontogenetic allometry because growth continues after sexual maturation
(although at a much smaller rate; Snelson, 1989).
809
The shape changes depicted by the canonical axes
are known to be associated with locomotion and foraging patterns. Differences in the relative size of the
caudal region and body depth were reported to influence escape velocity (Walker, 1997). Langerhans &
DeWitt (2004) and Langerhans et al. (2004) showed
that individuals of Gambusia affinis, Poecilia reticulata, and Brachyrhaphis rhabophora from populations
subject to predation present a relatively longer caudal
peduncle, a shallower anterior body, and a lower
position of the eye. This predation morphotype also
presented higher burst swimming speeds (Langerhans
et al., 2004) but, for a different result regarding body
depth, see also Royle, Metcalfe & Lindström (2006).
The lower position of the eye is presumably involved
with enhanced predator detection (Langerhans &
DeWitt, 2004). In the present study, the mean shapes
corresponding to freshwater populations presented a
phenotype convergent with the high predation morphotypes shared by P. reticulata, B. rabdophora, and
G. affinis (Langerhans & DeWitt, 2004). Considering
that the contrast of freshwater and brackish water
environments probably also represents a difference in
predation regimes (as discussed above), we can consider our results to be consistent with the ecomorphological paradigm for predation pressure in aquatic
vertebrates proposed by Langerhans et al. (2004).
Further support to this conclusion is given by the fact
that the evolutionary allometries are not similar to
static within-group allometries, and the observed patterns of morphological divergence among sites are not
consistent with the expectation of genetic drift, according to quantitative genetic models for the comparison
of shape variation patterns within and among groups
(results from present study). Rate tests for morphological divergence performed in a different study of the
same system (Monteiro & Gomes, 2005) also rejected
the hypothesis of genetic drift. The tests of neutral
evolution and the convergent patterns of shape divergence among P. vivipara and the other three Poeciliids
suggest directional selection by predation as the
process behind the major axes of shape variation.
Other shape differences among sites include changes
in the anterior region, such as the position of the
pectoral fin, and the mouth angle. Such shape differences are likely to be influenced by feeding modes
(scraping ¥ pelagic feeding and benthic ¥ surface
feeding) (Robinson & Wilson, 1995; Ruehl & DeWitt,
2005). The changes in mouth angle associated with the
canonical axes (most apparent for males) might reflect
differences in food availability in different lagoons. The
icon shape associated with positive scores in the first
canonical axis in Figure 6 (the male ordination) shows
a mouth angle (line between landmarks 1 and 2)
directed upwards, whereas the icon associated with
negative scores for the same axis shows a mouth angle
© 2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 799–812
810
J. L. GOMES JR and L. R. MONTEIRO
more anteriorly directed. Considering that the sites
associated with the positive scores present higher
salinity and less macrophytes (open water) for scraping, the observed pattern of shape differences is consistent with the one obtained experimentally by Ruehl
& DeWitt (2005) for G. affinis with attached food
versus free prey diets.
The spatial analysis showed a local pattern of
shape divergence (autocorrelated sites were only
those within the same lagoon) for both sexes. The lack
of larger scale autocorrelations reflect the way the
salinity environmental gradient is organized. As
shown by the PLS analysis, the most important geographical factor influencing the salinity gradient is
distance from the sea (with similar environments
among distant lagoons and differences among closely
spaced lagoons). Even the small scale autocorrelation
is lost when the salinity effect is removed from the
shape distances, suggesting that environmental differences are more important than geographical proximity to determine shape differences among sites. The
divergence among specimens from different sites can
be expected to be large, even within the same lagoons,
because the migration rates for such small-sized livebearing species are quite small (even for pools separated by a few meters) (Chapman & Warburton,
2006), and gene flow is greatly influenced by geographical distance and barriers (Crispo et al., 2006).
The lack of spatial pattern provides further evidence
with respect to the evolutionary processes at stake in
the morphological divergence among populations.
Stochastic processes are expected to generate monotonically decreasing correlograms and higher matrix
correlations (Telles & Diniz-Filho, 2000).
SEXUALLY
DIMORPHIC PATTERNS
The results were generally similar for males and
females, even with the strong sexual dimorphism
characteristic of livebearing fish species. There are,
however, differences regarding the observed allometric patterns, which reflect differences in within-site
body size variation that might arise from different
selective forces acting on the sexes. The males present
consistently smaller means and variances than
females, possibly because of selection for small males
in opportunistic matings (Bisazza & Pilastro, 1997;
Pilastro, Giacomello & Bisazza, 1997). Opportunistic
matings play an important role in P. vivipara reproduction, as commonly observed both in laboratory
and the field (J. L. Gomes Jr & L. R. Monteiro, pers.
observ.). The males lack bright colour spots, with the
exception of a temporary carotenoid-related orange
coloration in the throat region (more common in
saltwater lagoons). Females, on the other hand, can
maximize reproductive fitness by attaining a larger
size (Reznick, 1983; Cheong et al., 1984); hence, the
larger mean sizes and within-group variation among
females. Whenever the sexes are subject to different
selection pressures, it is expected that their responses
differ (Hendry et al., 2006). In the present study, the
patterns of divergence were similar, but the correlation between Procrustes distances among sites using
different sexes was moderate. These results suggest
that the shared patterns of among site divergence are
important and probably caused by the same effects,
but the differential selective pressures still cause
variation in responses of males and females. The
dimorphism of shape variation patterns in P. vivipara
is, nevertheless, smaller than the dimorphism in
P. reticulata (Hendry et al., 2006), where males are
much more conspicuous than females and present
courtship behaviours.
The most obvious dimorphic pattern is the difference
in within-group allometries. The first relative warp
(the major axis of within-group variation) is the principal component associated with body size for females
whereas, for the male data set, the axis associated with
size is the second relative warp. The within-group
allometric patterns are the major axis of shape variation for females, but a secondary axis of shape variation for males. This result is probably a reflection of
the smaller magnitude of within-group size variation
observed for males, which, in turn, is a consequence of
dimorphic growth patterns (Snelson, 1989; Hendry
et al., 2006), and different selection pressures.
CONCLUSIONS
Livebearing fishes are extremely tolerant to environmental variation, easily kept in captivity, and usually
abundant and widespread. They are therefore excellent models for the study of morphological, behavioural, and life-history adaptation. Poecilia vivipara
is abundant in all lentic environments in the holocenic lagoons in Northern Rio de Janeiro (particularly
in salt or brackish water), facing physical and biological differences in their environments (a magnitude of
variation seldom found in other poeciliid models).
This system should allow for the understanding of the
combined and unique effects of different environmental variables on body shape, size, and life history.
The pattern of size and shape divergence observed
among populations of P. vivipara inhabiting different
sites is influenced by the gradient formed by water
salinity. However, the evolutionary allometry observed
seems to be caused by an independent combination of
size and shape divergence in response to exposure to
predators in freshwater sites. The major axis of
among-site shape differences depicts a convergence
with contrasts between predator and predator-free
sites in the literature for three other poeciliid species.
© 2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 799–812
MORPHOLOGICAL DIVERGENCE AND ECOMORPHOLOGY
The observed convergence, the different within-group
allometries, the neutral hypothesis tests, and the lack
of a spatial structure, along with previous published
rate tests, suggest an evolutionary process of directional selection as responsible for the among-site
divergence pattern. The ecomorphological paradigm
proposed by Langerhans et al. (2004) provides both a
functional hypothesis relating a possible adaptation to
its selection agent, and the comparative test by shared
convergences with different lineages (Langerhan &
DeWitt, 2004).
Further research in this system should be directed
to an assessment of the genetical divergence structure
among populations, common garden experiments
(partially accomplished), field work with the postulated major predators, and the life-history evolution
of P. vivipara populations.
ACKNOWLEDGEMENTS
The authors want to thank A. C. Pessanha for assistance in field and laboratory work. Previous versions of
this manuscript have been greatly improved by the
comments of C. R. Ruiz-Miranda, M. S. Suzuki, M. J.
Cavalcanti, M. C. Gaglianone, and M. R. Nogueira.
The collection permits were issued by The Instituto
Brazileiro do Meio Ambiente e dos Recursos Naturais
Renováveis (IBAMA). This work was funded by
research grants and fellowships from the Conselho
Nacional de Desenvolvimento Científico e Tecnológico
(CNPq), Fundação de Amparo à Pesquisa do Estado do
Rio de Janeiro (FAPERJ), and Universidade Estadual
do Norte Fluminense. This is contribution number 103
from the Graduate Program in Ecology and Natural
Resources, Universidade Estadual do Norte Fluminense.
REFERENCES
Ackermann RR, Cheverud JM. 2002. Discerning evolutionary processes in patterns of tamarin (genus Saguinus)
craniofacial variation. American Journal of Physical
Anthropology 117: 260–271.
Bisazza A, Pilastro A. 1997. Small male mating advantage
and reversed size dimorphism in poeciliid fishes. Journal of
Fish Biology 50: 397–406.
Bizerril CRSF, Primo PBS. 2001. Peixes de águas interiores
do Estado do Rio de Janeiro. Rio de Janeiro: FEMARSEMADS.
Bookstein FL. 1991. Morphometric tools for landmark data
geometry and biology. New York, NY: Cambridge University
Press.
Bouef G, Payan P. 2001. How should salinity influence fish
growth? Comparative Biochemistry and Physiology C 130:
411–423.
811
Bourke P, Magnan P, Rodríguez MA. 1997. Individual
variation in habitat use and morphology in brook charr.
Journal of Fish Biology 51: 783–794.
Brinsmead J, Fox MG. 2002. Morphological variation
between lake- and stream-dwelling rock bass and pumpkinseed populations. Journal of Fish Biology 6: 1619–1638.
Chapman P, Warburton K. 2006. Postflood movements and
population connectivity in gambusia (Gambusia holbrooki).
Ecology of Freshwater Fish 19: 357–365.
Cheong RT, Henrich S, Farr JA, Travis J. 1984. Variation
in fecundity and its relationship to body size in a population
of the Least killfish, Heterandria formosa (Pisces: Poeciliidae). Copeia 1984: 720–726.
Cheverud JM. 1982. Relationships among ontogenetic,
static, and evolutionary allometry. American Journal of
Physical Anthropology 59: 139–149.
Cheverud JM. 1988. A comparison of genetic and phenotypic
correlations. Evolution 42: 958–968.
Crispo E, Bentzen P, Reznick DN, Kinnison MT, Hendry
AP. 2006. The relative influence of natural selection and
geography on gene flow in guppies. Molecular Ecology 15:
49–62.
Dryden IL, Mardia KV. 1998. Statistical shape analysis.
New York, NY: John Wiley and Sons.
Endler JA. 1995. Multiple-trait coevolution and environmental gradients in guppies. Trends in Ecology and Evolution
10: 22–29.
Garcia AM, Raseira MB, Vieira JP, Winemiller KO,
Grimm AM. 2003. Spatiotemporal variation in shallowwater freshwater fish distribution and abundance in a large
subtropical coastal lagoon. Environmental Biology of Fishes
68: 215–228.
Ghalambor CK, Reznick DN, Walker JA. 2004. Constraints on adaptive evolution: the functional trade-off
between reproduction and fast start swimming performance
in the Trinidadian guppy (Poecilia reticulata). American
Naturalist 164: 38–50.
Ghalambor CK, Walker JA, Reznick DN. 2003. Multi-trait
selection, adaptation, and constraints on the evolution of
burst swimming performance. Integrative and Comparative
Biology 43: 431–438.
Goslee S, Urban D. 2006. Ecodist: dissimilarity-based functions for ecological analysis. R package, version 1.01. Available at http://www.r-project.org.
Hendry AP, Kelly ML, Kinnison MT, Reznick DN. 2006.
Parallel evolution of the sexes? Effects of predation and
habitat features on the size and shape of wild guppies.
Journal of Evolutionary Biology 19: 741–754.
Johansson J, Turesson H, Persson A. 2004. Active selection for large guppies, Poecilia reticulata, by the pike
cichlid, Crenicichla saxatilis. Oikos 105: 595–605.
Lande R. 1979. Quantitative genetic-analysis of multivariate
evolution, applied to brain–body size allometry. Evolution
33: 402–416.
Langerhans RB, DeWitt TJ. 2004. Shared and unique features of evolutionary diversification. American Naturalist
164: 335–349.
Langerhans RB, Layman CA, Langerhans AK, DeWitt
© 2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 799–812
812
J. L. GOMES JR and L. R. MONTEIRO
TJ. 2003. Habitat-associated morphological divergence in
two Neotropical fish species. Biological Journal of the
Linnean Society 80: 689–698.
Langerhans RB, Layman CA, Shokrollahi AM, DeWitt
TJ. 2004. Predator-driven phenotypic diversification in
Gambusia affinis. Evolution 58: 2305–2318.
Legendre P. 1993. Spatial autocorrelation: trouble or new
paradigm? Ecology 74: 1659–1673.
Legendre P, Fortin M. 1989. Spatial pattern and ecological
analysis. Vegetatio 80: 107–138.
Marroig G, Cheverud JM. 2004. Did natural selection or
genetic drift produce the cranial diversification of neotropical monkeys? American Naturalist 163: 417–428.
Martin L, Suguio K, Dominguez JML, Flexor JM. 1997.
Geologia do quaternário costeiro do litoral Norte do Rio de
Janeiro e do Espírito Santo. Belo Horizonte: CPRM.
Mazzoni R, Iglesias-Rios R. 2002. Distribution patterns of
two fish species in a coastal stream in Southeast Brazil.
Brazilian Journal of Biology 62: 171–178.
Monteiro LR. 1999. Multivariate regression models and geometric morphometrics: the search for causal factors in the
analysis of the shape. Systematic Biology 48: 192–199.
Monteiro LR, Duarte LC, Reis SF. 2003. Environmental
correlates of geographical variation in skull and mandible
shape of the punaré rat Thrichomys apereoides (Rodentia:
Echimyidae). Journal of Zoology 261: 47–57.
Monteiro LM, Gomes JL Jr. 2005. Morphological divergence rates for natural selection: uncertainly of parameter
estimation and robustness of results. Genetics and Molecular Biology 28: 345–355.
Monteiro LR, Reis SF. 1999. Princípios de Morfometria
Geométrica. Ribeirão Preto: Holos Editora.
Neves FM, Monteiro LR. 2003. Body shape and size divergence among populations of Poecilia vivipara in coastal
lagoons of Southeastern Brazil. Journal of Fish Biology 63:
928–941.
Pilastro A, Giacomello E, Bisazza A. 1997. Sexual selection for small size in male mosquitofish Gambusia holbrooki. Proceedings of the Royal Society Series B, Biological
Sciences 264: 1125–1129.
Primo PBS, Bizerril CRSF, Soffiati A. 2002. Lagoas
do Norte Fluminense-Perfil Ambiental. Rio de Janeiro:
SEMADS.
R Development Core Team. 2006. R: a language and
environment for statistical computing. Available at http://
www.r-project.org
Reznick D. 1983. The structure of guppy life histories tradeoff
between growth and reproduction. Ecology 64: 862–973.
Reznick DN, Shaw FH, Rodd FH, Shaw RG. 1997. Evaluation of the rate of evolution in natural populations of
guppies (Poecilia reticulata). Science 275: 1934–1937.
Robinson BW, Wilson DS. 1995. Experimentally induced
morphological diversity in trinidadian guppies. Copeia
1995: 294–305.
Roff DA. 1996. The evolution of genetic correlations: an
analysis of patterns. Evolution 50: 1392–1403.
Rohlf FJ. 2000. NTSYS-Pc: numerical taxonomy and
multivariate analysis system, Version 2.0. Setauket: Applied
Biostatistics Inc.
Rohlf FJ. 2006. Morphometrics at SUNY Stony Brook. Available at http://life.bio.sunysb.edu/morph/
Rohlf FJ, Corti M. 2000. Use of two-block partial least
squares to study covariation in shape. Systematic Biology
49: 740–753.
Rohlf FJ, Slice DE. 1990. Extensions of the Procrustes
method for the optimal superimposition of landmarks.
Systematic Biology 11: 1–16.
Royle NJ, Metcalfe NB, Lindström J. 2006. Sexual selection, growth compensation and fast-start swimming performance in green swordtails, Xiphophorus helleri. Functional
Ecology 20: 662–669.
Ruehl CB, DeWitt TJ. 2005. Trophic plasticity and finegrained resource variation in populations of western mosquitofish, Gambusia affinis. Evolutionary Ecology Research
7: 801–819.
Snelson F. 1989. Social and environmental control of life
history traits in poeciliid. In: Meffe A, Snelson FF, eds.
Ecology and evolution of poeciliid fishes (poeciliidae).
Princeton, NJ: Prentice Hall, 149–161.
Soffiati A. 1998. Aspectos históricos das lagoas do Norte do
Estado do Rio de Janeiro. In: Esteves FA, ed. Ecologia das
Lagoas Costeiras do Parque Nacional da Restinga de Jurubatiba e do Município de Macaé, RJ. Rio de Janeiro:
NUPEM-UFRJ, 1–35.
Suzuki MS, Figueiredo RO, Castro SC, Silva CF, Pereira
EA, Silva JA, Aragon GT. 2002. Sand bar opening in a
coastal lagoon (Iquipari) in the northern region of Rio de
Janeiro State: hydrological and hydrochemical changes.
Brazilian Journal of Biology 62: 51–62.
Suzuki MS, Ovalle ARC, Pereira EA. 1998. Effects of sand
bar openings on some limnological variables in a hypertrophic tropical coastal lagoon of Brazil. Hydrobiologia 368:
111–122.
Svanbäck R, Eklöv P. 2003. Morphology dependent foraging
efficiency in perch: a trade-off for ecological specialization?
Oikos 102: 273–284.
Svanbäck R, Eklöv P. 2004. Morphology in perch affects
habitat specific feeding efficiency. Functional Ecology 18:
503–510.
Telles MPC, Diniz-Filho JAF. 2000. Null expectations of
spatial correlograms under a stochastic process of genetic
divergence with small sample sizes. Genetics and Molecular
Biology 23: 739–743.
Trexler JC. 1989. Phenotypic plasticity in poeciliid life histories. In: Meffe A, Snelson FF, eds. Ecology and evolution of
poeciliid fishes (poeciliidae). Princeton, NJ: Prentice Hall,
201–213.
Walker JA. 1997. Ecological morphology of lacustrine
threespine stickleback Gasterosteus aculeatus L. (Gasterosteidae) body shape. Biological Journal of the Linnean
Society 61: 3–50.
Winemiller KO. 1989. Ontogenetic diet shifts and resource
partitioning among piscivorous fishes in the Venezuelan
llanos. Environmental Biology of Fishes 26: 177–199.
© 2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 799–812