male guppy colour through time and

Document technical information

Format pdf
Size 509,6 kB
First found фев 4, 2016

Document content analysis

Language
English
Type
not defined
Concepts
no text concepts found

Transcript

bs_bs_banner
Biological Journal of the Linnean Society, 2014, 112, 108–122. With 3 figures
Using adaptive traits to consider potential
consequences of temporal variation in selection:
male guppy colour through time and space
KIYOKO M. GOTANDA1,2* and ANDREW P. HENDRY1,2
1
Department of Biology, McGill University, 1205 Docteur Penfield Avenue, Montreal, Quebec, H3A
1B1 Canada
2
Redpath Museum, McGill University, 859 Sherbrooke West, Montreal, Quebec, H3A 0C4 Canada
Received 6 November 2013; revised 4 January 2014; accepted for publication 5 January 2014
Temporal variation in selection is typically evaluated by estimating and comparing selection coefficients in natural
populations. Meta-analyses of these coefficients have yielded important insights, but selection coefficients are
limited in several respects, including low statistical power, imperfect fitness surrogates, and uncertainty regarding
consequences for trait change. A complementary approach without these limitations is to examine temporal
variation in adaptive traits themselves, which is mechanistically easier and more directly relevant to evolutionary
consequences. We illustrate this approach by analyzing the colour patterns of male guppies, Poecilia reticulata,
from each of six sites in Trinidad in each of 6 years. This system is particularly appropriate for our study because
key aspects of colour variation are genetically-based and responsive to selection. However, although spatial
patterns of colour variation have been extensively considered in this system, no study has yet formally assessed
annual temporal variation in non-manipulated populations. Matching previous conclusions for the guppy system,
we find that guppies from different sites manifest different colour patterns in association with different predation
regimes. We here add the new finding that, although some temporal variation is present, spatial patterns of colour
variation are generally consistent across years. These results suggest that, when considering adaptive traits,
spatial variation is more important than temporal variation, although our study system might be exceptional in
this regard. Additional studies examining spatiotemporal variation in adaptive traits could help to improve our
understanding of the role that spatiotemporal variation in selection plays in the evolutionary process. © 2014 The
Linnean Society of London, Biological Journal of the Linnean Society, 2014, 112, 108–122.
ADDITIONAL KEYWORDS: adaptation – Poecilia reticulata – selection – spatial variation.
INTRODUCTION
The extent to which selection acting on natural populations is temporally consistent or variable has been
the subject of several recent meta-analyses of published selection coefficients, but the answer remains
unresolved (Siepielski, DiBattista & Carlson, 2009;
Bell, 2010; Kingsolver et al., 2012; Morrissey &
Hadfield, 2012). Some meta-analyses suggest that
substantial temporal variation is present (Siepielski
et al., 2009; Kingsolver et al., 2012), whereas others
*Corresponding author.
E-mail: [email protected]
108
suggest that apparent temporal variation is an artefact of sampling error (Morrissey & Hadfield, 2012). It
will be hard to conclusively resolve this question on
the basis of selection coefficients alone because it is
very difficult to accurately and precisely measure
selection in natural populations (Kingsolver et al.,
2001; Hereford, Hansen & Houle, 2004; Hersch &
Phillips, 2004; Morrissey, Kruuk & Wilson, 2010).
Furthermore, selection coefficients cannot, by themselves, resolve the critical question of whether or not
temporal variation in selection has important evolutionary consequences for adaptive traits.
A complementary approach that avoids the above
limitations of selection coefficients is to directly
examine temporal variation in the adaptive traits
© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 112, 108–122
SPATIOTEMPORAL VARIATION IN GUPPY COLOUR
themselves (Grant & Grant, 2002; Aguirre & Bell,
2012), although this approach has caveats of its own
(see Discussion). One advantage is that measurement
of phenotypic trait values is easier and subject to less
bias than is the estimation of selection coefficients.
The reason is that estimating selection also requires
accurately estimating fitness, which can be a difficult
undertaking for both practical and theoretical
reasons, especially in natural populations (Kingsolver
et al., 2001; Hereford et al., 2004; Hersch & Phillips,
2004; Morrissey et al., 2010). Indeed, a number of
studies have found that selection coefficients are often
not predictive of evolutionary change (Merilä,
Sheldon & Kruuk, 2001) and that estimates of selection in well-adapted populations will not reflect the
selection that drove that adaptation in the first place
(i.e. selection erases its traces; Haller & Hendry,
2014). Thus, another advantage of monitoring adaptive traits themselves is that they point more directly
to any realized consequences of temporal variation in
selection. Although temporal variation in traits will
not necessarily match temporal variation in selection,
it should provide a valuable and complementary
contribution to discussions about the importance of
temporal variation in the evolutionary process. We
emphasize that we are not advocating a replacement
for estimating selection on adaptive traits, nor are
we dismissing the valuable contributions selection
estimates can provide to understanding adaptation.
Rather, we are suggesting a complementary approach
for gaining further insight into the importance of
temporal variation in selection.
We further suggest that a comparative approach
would be especially useful for evaluating the importance of temporal variation (in selection or adaptive
traits). That is, we should be asking how important
temporal variation is in relation to other aspects of
variation, with the most obvious candidate being
spatial variation among conspecific populations
(Blanckenhorn et al., 1999; Gosden & Svensson, 2008;
Weese et al., 2010; Sternalski, Mougeot & Bretagnolle, 2012; Cox & Rabosky, 2013). It is generally
accepted that spatial variation in selection, and hence
adaptation, is very strong (Endler, 1986; Schluter,
2000), which provides a useful basis for assessing the
importance of temporal variation. If, for example,
temporal variation is as strong as spatial variation,
we can safely assume that both are important in the
evolutionary process. In some situations, temporal
variation might be synchronized across a number of
populations (e.g. through regional climate forcing), in
which case the main effect of time would be of interest. In most situations, however, temporal variation
will be asynchronous across space (e.g. through local
fluctuations in biotic and/or abiotic conditions), in
which case the interaction between space and time
109
would be of greatest interest. That is, the key question becomes: how consistent through time are the
spatial patterns of variation?
STUDY
SYSTEM
We implement the above ideas through an analysis of
spatiotemporal variation in the colour patterns of
male Trinidadian guppies, Poecilia reticulata (Peters).
This system is appropriate for several reasons. First,
guppy colour is known to differ among populations in
response to spatial variation in selective pressures,
such as predation (Endler, 1980; Magurran, 2005),
resource levels (Kodric-Brown, 1989; Grether, Hudon
& Endler, 2001a; Millar et al., 2006; Schwartz &
Hendry, 2010), parasites (Houde & Torio, 1992), and
sexual selection (Endler & Houde, 1995; Houde, 1997;
Schwartz & Hendry, 2007). In general, sexual selection by females results in positive selection for conspicuous colour such as orange (Houde, 1997),
whereas predation by visual predators results in
negative selection conspicuous colours (Endler, 1980;
Magurran, 2005). This spatial variation in colour
provides a useful benchmark for considering the
effects of temporal variation. Second, although some
aspects of guppy colour, such as carotenoid deposition,
brightness, and chroma, are influenced by plasticity
(Grether et al., 2001a; Grether, Cummings & Hudon,
2005; Miller & Brooks, 2005; Ruell et al., 2013), the
size, number, and location of colour spots is highly
heritable and differences within and among populations are mostly genetically based (Houde, 1992;
Brooks & Endler, 2001; Karino & Haijima, 2001;
Hughes, Rodd & Reznick, 2005; Tripathi et al., 2009;
Gordon, López-Sepulcre & Reznick, 2012). These
properties increase the likelihood that spatiotemporal
variation in the colour properties documented in the
present study reflects spatiotemporal variation in
underlying genotypes; however, with a field study
such as ours, we cannoy be certain to have completely
eliminated plastic effects. Third, the generation
length of guppies is short enough (approximately 1.8
generations per year; Reznick et al., 1997) that temporal variation in selection could drive adaptive
genetic change, and thus phenotypic change, on a
yearly time scale (Endler, 1980).
Studies analyzing spatial variation in traits often
replicate their samples in at least 2 years (examples
in supplementary data from Siepielski et al., 2013) to
confirm temporal consistency of the observed patterns. Remarkably, none of the many studies of male
guppy colour in non-manipulated populations has yet
replicated their samples across multiple years.
Instead, the few studies with replicates across years
have focused on the effects of major human-caused
disturbances, such as introduction to a new predation
© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 112, 108–122
110
K. M. GOTANDA and A. P. HENDRY
regime (Endler, 1980; Karim et al., 2007; Kemp et al.,
2009), introduction to a new locality (Kudo & Karino,
2013) or the clear-cutting of riparian vegetation
(Schwartz & Hendry, 2010). As a result, the extent of
temporal variation in male guppy colour remains
unknown, although such variation appears to be possible. In particular, each of the above-mentioned
selective agents (predation, resource levels, parasitism, and sexual selection) likely varies through time
in response to temporal variation in flooding (van
Oosterhout et al., 2007; Weese et al., 2011), population
density (Reznick, 1989; Reznick, Bryant & Bashey,
2002), sex ratio (Rodd & Reznick, 1997; Pettersson
et al., 2004; McKellar, Turcotte & Hendry, 2009), and
parasitism (van Oosterhout et al., 2007; Gotanda
et al., 2013). Moreover, a mark–recapture study found
some evidence of spatiotemporal variation in viability
selection on male guppy colour, although that study
was attended by the above mentioned concerns surrounding selection estimates (Weese et al., 2010).
In the present study, we assess spatiotemporal variation in male guppy colour through the analysis of
photographs from field collections in each of six consecutive years (approximately 10 guppy generations)
at each of six different sites in Trinidad. Using
genetically-based and known-adaptive colour metrics
(sizes and numbers of spots of different colours) that
have been reported in many previous studies, we
partition the variation among effects of site, year, and
their interaction. We then focus our interpretation on
the extent to which spatial patterns of variation in
guppy colour are consistent through time.
MATERIAL AND METHODS
Each year from 2005 to 2010, live guppies were
sampled from each of six different sites (see Supporting
information, Fig. S1) along the Marianne and Paria
rivers of northern Trinidad. For each year, these
samples were always taken during the dry season in
the months of February and March. Based on our
previous work in these watersheds (Crispo et al., 2006;
Hendry et al., 2006; Millar et al., 2006; Schwartz &
Hendry, 2010; Weese et al., 2010), we selected sites
that (1) were far enough apart so that dispersal
between them was minimal making them effectively
independent sites for this study; (2) represented a
range of different environments so that adaptive
spatial variation would be expected; and (3) had large
population sizes so that genetic drift would be
minimal. In accordance with standard practice, we
classified sites as high-predation (HP) or low-predation
(LP) according to whether highly piscivorous fishes
were present or absent, respectively (Endler, 1980;
Rodd & Reznick, 1997; Martin & Johnsen, 2007). On
the Marianne River, two sites (M7 and M14) were HP
and two sites (M4 and M16) were LP. On the Paria
River, both sites (P4 and P7) were LP, as is the case for
all sites in the Paria (Millar et al., 2006, fig. 1; see also
Supporting information, Fig S1).
At each site in each year, we used butterfly nets to
capture 16–21 male guppies (N = 20 for most), except
one sample that, as a result of logistical constraints,
had only six males (see Supporting information,
Tables S1, S2). The fish were transported to our field
station in Trinidad, where they were euthanized with
an overdose of tricaine methanesulfonate (Finquel
MS-222; Argent Laboratories Group) buffered with
NaHCO3 to obtain a neutral pH. This process helps
standardize some of the plastic components of male
guppy colour (Endler, 1991). Immediately after euthanasia, each guppy was placed left-side-up on a light
background and blotted with a paper towel to reduce
reflective glare. A digital camera was situated directly
overhead and illumination was provided by two full
spectrum fluorescent bulbs (Coralife 18′′ 10 000 K
Daylight). Two photographs were taken of each guppy,
one with and one without a flash, to be compared
side-by-side when collecting colour data (Millar et al.,
2006; Karim et al., 2007; Millar & Hendry, 2012). An
acrylic colour standard and a ruler were visible in
each photograph.
Our analyses of colour variation were based on the
size and number of colour spots as inferred from the
photographs. This method of analysis represents
the classic and most common approach, which means
that our results are directly relevant to previous work
(Endler, 1980; Kodric-Brown, 1985; Houde & Endler,
1990; Pilastro et al., 2004; Millar et al., 2006; Weese
et al., 2010). Moreover, analyses based on these
metrics have yielded many robust conclusions regarding the role of natural and sexual selection in shaping
guppy colour (Endler, 1980; Houde, 1997; Magurran,
2005). We would ideally also have included analyses
based on spectrophotometry and visual modelling
(Grether et al., 2001a, 2005; Endler & Mielke, 2005;
Kemp, Reznick & Grether, 2008; Kemp et al., 2009),
which provides information on how colours are perceived by guppies and their predators (Kemp et al.,
2008, 2009). Although we collected spectrophotometric
data in 2009 and 2010, these data were not available
for prior years and were not analyzed in the present
study because our goal was to consider variation in an
adaptive trait over a longer time frame.
Since 2008, the photographs were captured in both
RAW and JPEG formats, with the former being preferable for colour analysis (Stevens et al., 2007).
However, only JPEGs had been captured prior to 2008,
and so only JPEGs were used for the present analysis,
as has also been the case in other guppy studies (Miller
& Brooks, 2005; Pitcher, Rodd & Rowe, 2007; Kudo &
Karino, 2012; Kudo & Karino, 2013). All photographs
© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 112, 108–122
SPATIOTEMPORAL VARIATION IN GUPPY COLOUR
were taken using the highest possible resolution available for the camera. One individual (KMG), who was
blind to the year and site of origin of each guppy,
analyzed all 695 photographs in random order. Each
photograph was imported as a raster image into
MAPINFO PROFESSIONAL, version 6 (Pitney Bowes
Software). Individual spot colours were then classified
as black, fuzzy black, orange (includes red), yellow,
violet, silver, blue, and green (sensu Endler, 1978;
Endler, 1991; Millar et al., 2006; Karim et al., 2007).
Flash and nonflash photographs were compared sideby-side to prevent ambiguity in colour perception
(Millar et al., 2006; Karim et al., 2007; Millar &
Hendry, 2012), and a colour standard was used as a
visual reference in all photographs. Data were collected on the flash photographs. Each spot on the body
of the fish (i.e. fins were excluded) was individually
outlined in MAPINFO and its area (mm2) was
recorded. These areas were summed across all spots of
a given colour on a given fish. The resulting ‘total area’
of each colour on each fish was then divided by the total
body area of the fish (again excluding fins) to yield the
‘relative area’ of each colour. In addition, the number of
spots in each colour category was recorded to yield
‘spot number’ for each guppy. Previous work has shown
that these measurements are highly repeatable
(Gotanda et al., 2013).
To reduce the number of variables for analysis, the
above colours were grouped into biologically relevant
categories. These groups have different physiological
bases, structural bases, functional interpretations,
and selective relevance (Endler, 1978; Kodric-Brown,
1989; Hearing, 1993; Brooks & Endler, 2001; Grether
et al., 2001a, 2001b, 2005; McGraw et al., 2002;
Blows, Brooks & Kraft, 2003; Griffith, Parker &
Olson, 2006). ‘Carotenoid colours’ consisted of the sum
of orange and yellow spots, although note that pigments other than carotenoids also contribute to these
spots (Grether et al., 2001a, 2005). ‘Structural colours’
are colours that are iridescent and have higher levels
of reflection, and consisted of the sum of blue, violet,
and silver spots (Endler, 1978; Brooks & Endler,
2001). ‘Melanic colours’ consisted of the sum of black
and fuzzy black spots. In addition, green was retained
in the analysis as a separate colour category (Weese
et al., 2010). To meet statistical assumptions of normality, data were transformed as appropriate.
All data analyses were conducted in R, version
2.15.2, R Core Team, 2012) and involved multivariate
and univariate general linear models with year
(2005–2010), site (M4, M7, M14, M16, P4, and P7),
the year-by-site interaction, and body size as a
covariate. In these models, site was clearly a fixed
effect because the different sites were specifically
chosen for their properties as described above. By
contrast, year can be considered to be a random effect
111
because the specific years should be considered a
random selection of six consecutive years. We therefore assessed statistical significance on the basis of
mixed models. However, our primary goal was to
directly compare the effects and effect size of site,
year, and their interaction in a single model, and so
we additionally analyzed models with year specified
as a fixed effect. Statistical results were similar
between the two approaches (i.e. year as fixed or
random), and so we only report the fixed-effects
analysis (mixed model results appears in the Supporting information, Table S8). In addition, order of entry
of terms into the model did not alter conclusions, and
so we report models where the site term was fit first.
In each case, we first fit multivariate analyses of
covariance with all colour categories (carotenoid
colour, structural colour, melanic colour, and green)
considered simultaneously as response variables. We
then fit separate analyses of covariance for each
colour category. All of these models were fit separately
for two different classes of colour metrics: relative
areas (arcsine square root transformed) and spot
numbers (square root transformed), as described
above. The size of individual spots was not analyzed
because it was statistically redundant to relative
areas and spot numbers. Post-hoc Tukey’s honestly
significant difference tests were run following the
analyses of covariance to determine which of the sites
differed from each other for each colour category.
Effect sizes for the fixed effect models were calculated
as partial η2 based on Wilk’s partial η2 for multivariate tests and sums of squares for the univariate tests
(Langerhans & DeWitt, 2004) using the heplot
package in R. Ranges for partial η2 were generated
when the data were jack-knifed by omitting one row
of data at a time and repeating this with all rows
(Langerhans & DeWitt, 2004).
For the interpretation of effects from the above
models, first, the site term reflects consistent (across
years) spatial variation in colour patterns, which we
expect from previous work to be a consequence of
spatial variation in selection. Second, the year term
reflects consistent (across sites) temporal variation
in colour pattern. We assume this variation is
a consequence of improvements of both our photographic methods and camera/flash equipment
through time rather than any biologically significant
regional trends in colour (see Results). Third, the
site-by-year interaction term reflects differences
between years in the patterns of spatial variation
(e.g. guppies from one site are more colourful than
those from another site in some years but not other
years). This variation would be consistent with temporal variation in selection that differed between
sites, although other factors could also contribute
(see Discussion.)
© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 112, 108–122
112
K. M. GOTANDA and A. P. HENDRY
Table 1. Results from multivariate analyses of covariance using Wilk’s lambda test statistic with site fitted before year
as fixed effects and body size (mm2; log transformed) as the covariate
Relative area colour
Site
Year
Year-by-site
Body size
Spot number
Site
Year
Year-by-site
Body size
F
d.f.
P
Wilk’s partial η2
36.106
30.040
2.892
20.224
202 173
202 173
1 002 600
4 655
< 0.001
< 0.001
< 0.001
< 0.001
0.176
0.173
0.099
0.084
(0.175–0.178)
(0.171–0.176)
(0.097–0.102)
(0.079–0.087)
25.451
15.339
2.362
12.594
202 173
202 173
1 002 600
4 655
< 0.001
< 0.001
< 0.001
< 0.001
0.135
0.096
0.083
0.081
(0.133–0.138)
(0.095–0.097)
(0.081–0.084)
(0.075–0.085)
Estimates of variance were calculated as multivariate partial η2 also utilizing Wilk’s lambda. Numbers in parentheses are
the ranges generated when the data were jack-knifed. The dependent variables were relative area (arcsine square root
transformed) and spot number (square root transformed) for all colour categories (carotenoid colour, structural colour,
melanic colour, and green). Statistical significance (P < 0.05) is indicated in bold.
Finally, we placed our results in the context of an
earlier, single-year survey that included our six sites
and 23 additional sites (Millar et al., 2006). That
earlier study found a strong negative association
across populations between mean relative areas of
orange and mean relative areas of blue. We therefore
used Pearson product-moment correlations to test, in
the present data set, for the same relationship among
all years and within each year.
RESULTS
Multivariate analyses of relative colour area and of
spot number revealed significant effects of all the
independent variables, body size, site, year, and the
year-by-site interaction, in all cases (Table 1). Of
these effects, site and year explained the most variance in relative colour area and site explained the
most variance in spot number (Table 1). Relative to
site, the year-by-site interaction explained 44% as
much of the variation (difference between the larger
and smaller value divided by the larger value) in
relative colour area and 39% as much of the variation
in spot number (Table 1).
Univariate analyses revealed highly significant
effects of all the independent variables on most colour
categories and colour metrics (Table 2). Site explained
the most variance in carotenoid colour relative area
and spot number, structural colour relative area and
spot number, and melanic colour spot number; for
carotenoid colour spot number, year explained almost
the same amount. Year explained the most variance
for the other colour variables, although site explained
almost as much for green relative area. Relative to
site, the year-by-site interaction explained less of the
variation in all cases (6% as much for green spot
number to 72% as much for carotenoid colour relative
area) except for melanic colour relative area, where
neither effect was strong. The strength of the year
effect and the overall trend through time toward
increasing total colour is almost certainly the result of
improvements in photographic methods and equipment. In particular, we gradually transitioned from
point and shoot cameras with onboard flashes to
digital SLRs with commander ring flash set-ups. We
also transitioned from rudimentary backgrounds and
automatic camera settings (which are sensitive to
changing conditions) to standardized backgrounds
and full-manual control of settings (e.g. shutter speed,
aperture, white balance, and ISO).
We now consider colour variation among specific
sites, as well as temporal consistency in that variation. In this assessment, we will often refer to predation regime. Our intent in making this reference is
not to imply a test for how predation regime influences colour because the present study was not
designed as another test of this well-established association (Endler, 1980; Rodd & Reznick, 1997; Millar
et al., 2006; Karim et al., 2007; Martin & Johnsen,
2007; Kemp et al., 2009; Weese et al., 2010). Instead,
our references to predation regime draw on this prior
knowledge to aid post-hoc interpretations of the
causes of spatial variation in guppy colour.
Sites where guppies had the highest relative area of
total colour were LP sites (M4, M16, P4, and P7), and
this pattern was relatively consistent across all years
(Fig. 1; see also Supporting information, Table S1).
Specifically, adjusted means for these four sites were
the highest among all sites within each of the six
years, with only two minor exceptions (Fig. 1; see also
© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 112, 108–122
SPATIOTEMPORAL VARIATION IN GUPPY COLOUR
113
Table 2. Results from the univariate analyses of covariance with year, site, and their interaction as fixed effects; relative
area (arcsine square root transformed) or spot number (square root transformed) for individual colour categories
(carotenoid, structural, melanic, and green) as dependent variables; and body size (mm2; log transformed) as the covariate
Carotenoid
Relative area
Site
Year
Year-by-site
Body area
Spot number
Site
Year
Year-by-site
Body area
Structural
F
P
Partial η2
d.f
13.923
5.814
2.527
2.449
< 0.001
< 0.001
< 0.001
0.118
0.311
0.091
0.088
0.000
5658
5658
25 658
1658
12.982
0.696
3.2717
51.835
< 0.001
0.627
< 0.001
< 0.001
0.100
0.099
0.087
0.003
5658
5658
25 658
1658
Melanic
Relative area
Site
Year
Year-by-site
Body area
Spot number
Site
Year
Year-by-site
Body area
P
Partial η2
d.f
14.085
7.716
3.703
46.801
< 0.001
< 0.001
< 0.001
< 0.001
0.294
0.252
0.123
0.087
5658
5658
25 658
1658
63.313
15.595
3.272
134.399
< 0.001
< 0.001
< 0.001
< 0.001
0.375
0.119
0.111
0.062
5658
5658
25 658
1658
F
Green
F
P
Partial η2
d.f.
F
P
Partial η2
d.f.
1.869
8.750
2.418
0.119
0.098
< 0.001
< 0.001
0.730
0.059
0.276
0.084
0.001
5658
5658
25 658
1658
10.511
6.254
3.482
3.849
< 0.001
< 0.001
< 0.001
0.0502
0.223
0.225
0.117
0.015
5658
5658
25 658
1658
2.695
5.853
1.267
0.178
0.020
< 0.001
0.174
0.674
0.093
0.076
0.046
0.000
5658
5658
25 658
1658
3.523
4.359
2.172
3.825
0.004
< 0.001
< 0.001
0.051
0.081
0.117
0.076
0.009
5658
5658
25 658
1658
Statistical significance (P < 0.05) is indicated in bold.
Supporting information, Table S1). Interestingly, high
total colour was achieved very differently by guppies
from one of these LP sites (M16) than by guppies from
the other LP sites, a pattern again consistent across
years (Fig. 1; see also Supporting information,
Tables S1, S2). In particular, and with only minor
exceptions, guppies from the M4, P4, and P7 sites
exhibited higher carotenoid colour and lower structural colour than did guppies from the M16 site
(Figs 1, 2, 3; see also Supporting information,
Tables S1, S3, S4). Conversely, guppies from the M16
site exhibited lower levels of carotenoid colour, higher
levels of structural colour, and higher levels of green
(Figs 1, 2, 3; see also Supporting information,
Tables S1, S3, S4). Overall, spatial variation in colour
was relatively consistent through time, both in its
generality (guppies from the four LP sites always had
the highest total colour) and in its anomalies (guppies
from one LP site attained high colour in a very
different way than guppies from the other three LP
sites). Consideration of river drainage (Marianne or
Paria) as a potential effect does not significantly alter
our results or interpretations of the data (see Supporting information, Tables S5, S6)
Results for spot number were similar to those
described above for relative colour area, and so they
are only briefly summarized here (for details, see
Supporting information, Fig. S2, Table S2, S3, S4). In
particular, strong spatial patterns were again evident
and were generally consistent through time. As one
example, the site where guppies had the most colour
spots was M16 in all six years (see Supporting information, Fig. S2, Table S2). As another example,
guppies from M7 and M14, both HP sites, generally
had more structural colour spots than did guppies
from three LP sites: M4, P4, and P7 (see Supporting
information, Fig. S2, Tables S2, S3, S4).
Millar et al. (2006) reported a strong negative association between mean relative areas of orange and of
blue among guppy populations from the Marianne
and Paria rivers. We found the same association
(r = −0.881; P = 0.020) in our dataset based on six of
© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 112, 108–122
114
K. M. GOTANDA and A. P. HENDRY
Figure 1. Adjusted means for the relative area of colour categories (all colour, carotenoid colour, structural colour, and
melanic colour) by year and site. Data were arcsine square root transformed. Error bars are not presented for clarity when
interpreting the graph but are reported in the Supporting information (Table S1). High-predation (HP) sites are indicated
with open symbols and dashed lines; low-predation (LP) sites are indicated with filled symbols and solid lines.
these sites when all years were combined (Fig. 2).
This negative relationship was also present across
all six years, although it was statistically significant
in only three years (see Supporting information,
Table S7). Perhaps the most interesting part of this
analysis is the position of M16 in relation to the other
sites. Although an LP site, M16 has colour patterns
more similar to guppies from an HP site than to those
for guppies from other LP sites (Fig. 2). Again, this
distinction for M16 guppies was consistent across all
years.
DISCUSSION
Some temporal variation was evident in the present
study. The main effect of year is likely attributable to
improved photographic methods and equipment (see
Results) and so is unlikely to be biologically relevant.
The interaction between site and year was also often
significant in both multivariate and univariate
models, and might well reflect temporal variation in
selection. Currently, we do not have a specific causal
explanation for this variation because we did not
measure environmental variables at each site over all
6 years. Thus, it might be worthwhile to search for
potential causes of this temporal variation in future
work. However, we do not further discuss this temporal variation for the above reasons and because (as
explained below) its effect was minor in relation to
spatial variation.
Three main observations highlight the predominant
effect of spatial variation over temporal variation in
guppy colour. First, guppies from three populations
(M4, P4, and P7) in LP environments consistently had
the highest total colour, the highest carotenoid colour,
and the lowest structural colour (Fig. 1; see also
Supporting information, Fig. S2). Second, guppies
from the fourth LP population (M16) also had high
© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 112, 108–122
SPATIOTEMPORAL VARIATION IN GUPPY COLOUR
115
Figure 2. The mean relative area of orange and blue spots is inversely related among sites in the Marianne and Paria
Rivers. Error bars denote 95% confidence intervals for all years combined. Grey triangles are data from a previous study
of 29 sites in the Marianne and Paria rivers (Millar et al., 2006).
total colour but differed from the other three LP
populations in consistently having low carotenoid
colour, high structural colour, and high green (Fig. 1;
see also Supporting information, Fig. S2), which is
also consistent with previous work (Schwartz &
Hendry, 2010; Weese et al., 2010). Third, the
previously-described negative association among
populations for relative orange versus blue colour on
male guppies (Millar et al., 2006) was essentially
identical and consistent across years in our new
samples (Fig. 2). These consistent spatial patterns
likely reflect genetically-based adaptive responses to
spatial variation in selection (details below). The
alternatives of drift and plasticity appear less likely
because (1) the populations we sampled are large
enough that genetic drift is likely minor and (2)
variation in the size and number of colour spots is
strongly genetically based (Houde, 1992; Brooks &
Endler, 2001; Karino & Haijima, 2001; Hughes et al.,
2005; Tripathi et al., 2009; Gordon et al., 2012). Of
course, it is important to recognize that a field study
such as ours cannot disentangle genetic effects from
any plastic effects on male guppy colour patterning.
Furthermore, more plastic components of guppy
colour, such as carotenoid deposition, brightness, and
chroma, might vary more temporally and would be
useful to study in future work.
Biological explanations for the first observation
(three LP sites had high colour, high carotenoid, and
low structural colour) and the third observation (a
negative relationship between carotenoid and structural colours) follow naturally from previous work on
guppy colour. In particular, high orange in guppies
from LP sites is considered to reflect strong female
preferences for orange (Kodric-Brown, 1985; Long &
Houde, 1989; Houde & Endler, 1990; Endler & Houde,
1995; Grether, 2000; Brooks & Endler, 2001; Pilastro
et al., 2004) coupled with an absence of major predators that select against orange (Endler, 1978; Endler,
1991; Godin & McDonough, 2003; Millar & Hendry,
2012). In addition, high blue but low orange in
guppies from high-predation (HP) sites (here M7 and
M14) is considered to reflect female preference for
colourful (including blue) males (Kodric-Brown, 1985;
Houde, 1997) and the presence of major predators
that select against orange but are less sensitive to
blue (Endler, 1978; Archer & Lythgoe, 1990; Kröger,
Bowmaker & Wagner, 1999; Kemp et al., 2008).
A biological explanation for the second observation
(the M16 LP site had high colour but in a different
© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 112, 108–122
116
K. M. GOTANDA and A. P. HENDRY
Figure 3. Representative photographs of male guppy colour patterning. All photographs are from 2009, although the
colour on each fish shown in the photograph is representative of the overall mean for a given site. Photographs on the
left represent sites where males have less carotenoid colour and photographs on the right represent sites where males
have more carotenoid coloration. Black scale bars = 10 mm. Photographs are not scaled with respect to each other. HP,
high-predation; LP, low-predation.
way from other LP sites: more structural colour than
carotenoid colour) does not emerge so easily from
previous work and so deserves further consideration.
The fact that M16 guppies were anomalous across all
6 years suggests a consistently atypical local selective
pressure. One possibility is a unique predator fauna,
although we have not noted any predator differences
between M16 and the other LP sites during 10 years
of visits. Another possibility is the more open canopy
at M16 in comparison with the other LP sites
(McKellar et al., 2009; Schwartz & Hendry, 2010);
however, we would expect the resulting higher productivity to favour increased (not decreased) carotenoid colour (Grether et al., 2001b). A third possibility
is limited genetic variation in comparison with other
LP sites. This is unlikely because M16 has comparable genetic diversity to other Marianne LP sites
(Crispo et al., 2006), and colour variation is not lower
among individuals within M16 compared to other LP
sites (see Supporting information, Tables S1, S2). A
final possibility is that anomalous patterns of sexual
selection at M16 have favoured an atypical male
colour pattern. This last explanation is attractive
given (1) a lack of support for the above alternatives;
(2) evidence that females from different populations
show different preferences for male colour (Endler &
Houde, 1995; Brooks & Endler, 2001); and (3) evidence for among-population correlations between
female preference and male colour (Houde & Endler,
1990; Schwartz & Hendry, 2007). However, the
reasons why females from this site might prefer an
anomalous colour pattern are not known, although
the initial impetus could be female preferences for
novel males (Hughes et al., 2013). Overall, temporal
consistency of the anomalous colour pattern at this
site suggests the value of more formally exploring
these, as well as other, potential explanations. Of
particular value would be formal selection estimates
© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 112, 108–122
SPATIOTEMPORAL VARIATION IN GUPPY COLOUR
across years at M16 and other sites of particular
interest.
SPACE
VERSUS TIME
We found that, although some temporal variation was
present, it did not strongly modify spatial variation in
male colour, and this result might be considered surprising in light of evidence for temporal variation in
biotic and abiotic factors that influence selection (see
Introduction). We suggest that temporal variation in
selection is indeed present but is overpowered by
spatial variation, at least on the spatiotemporal
scales investigated in the present study. In particular,
(1) LP populations never have dangerous predatory
fishes, whereas HP populations often do (Endler,
1980); (2) canopy cover differs among sites but is
reasonably consistent through time at a given site in
the absence of human disturbances (Schwartz &
Hendry, 2010); (3) parasitism varies much more spatially than temporally (Fraser, Ramnarine & Neff,
2010; Gotanda et al., 2013); and (4) factors (e.g. sex
ratio) that likely influence sexual selection also
appear to vary more spatially than temporally
(McKellar et al., 2009). All of these observations
support our present results in suggesting that the
adaptive consequences of temporal variation in selection are weaker than the adaptive consequences of
spatial variation in selection.
An alternative is that temporal variation in selection is strong but, for various reasons, the response or
consequence of this selective variation does not cause
notable phenotypic changes on the time scale we
considered. First, colour might evolve so slowly that
adaptive phenotypes are reasonably stable despite
temporal fluctuations in selection. This explanation is
unlikely to be sufficient given the high heritability of
colour (Houde, 1992; Brooks & Endler, 2001; Karino
& Haijima, 2001; Hughes et al., 2005; Tripathi et al.,
2009; Gordon et al., 2012) and a number of studies
showing rapid colour changes in guppies introduced
to new environments (Endler, 1980; Kemp et al.,
2008), although this is not always the case (Karim
et al., 2007; Kemp et al., 2009). Second, temporal variation in selection might be swamped, or at least
dampened, by gene flow among sites. This explanation also appears unlikely given our documentation of
low gene flow among these sites (Crispo et al., 2006;
Weese et al., 2011) and the strong spatial variation
that would be likewise opposed by high gene flow.
Additionally arguing against both of these alternatives, the spatial variation in our study is large and
sufficiently stable to suggest that the populations are
reasonably well adapted to alternative fitness peaks.
For all of these reasons, we favour the explanation
that, when considering the variation in an adaptive
117
trait under selection, the selective factors influencing
male guppy colour, and therefore selection itself, vary
more in space than in time.
Our conclusion that temporal variation in selection
contributes less than spatial variation in selection to
colour pattern variation might be used to support the
perspective that directional selection is relatively consistent through time (Morrissey & Hadfield, 2012).
However, we caution against too strong an interpretation in this direction owing to the properties of the
guppy system. In particular, guppies have long been
known to exhibit very strong spatial variation in
selection and traits, and have been a preferred study
system for this reason (Endler, 1978; Endler, 1980;
Magurran et al., 1992; Carvalho et al., 1996; Kelly,
Godin & Wright, 1999; Millar et al., 2006; Kemp et al.,
2009; Millar & Hendry, 2012). Indeed, the spatial
variation in male guppy colour is so strong and clear
that no study of non-manipulated populations had yet
temporally replicated its samples. Moreover, we specifically selected populations for our long-term study
that were known from previous work (Millar et al.,
2006) to occupy different selective environments and
to differ in their colour patterns. In short, the guppy
system in general, and our study sites in particular,
might represent a situation in which the importance
of spatial variation in selection is exceptionally
(perhaps atypically) strong in relation to temporal
variation in selection.
CONCLUSION
AND APPLICATIONS
We found that temporal variation, although present,
was too low to modify interpretations regarding
spatial variation in an adaptive trait: male guppy
colour. We suggest that this result reflects selection
that varies much more in space than in time, at least
for the populations and time frame examined in the
present study. Although these observations might be
taken to support the idea that selection is reasonably
consistent through time, our study system might be
exceptional in this regard. We suggest that analyzing
spatiotemporal variation in adaptive traits provides a
complementary basis for considering the importance
of temporal variation in selection. In doing so, it is
critical to note that the adaptive evolution of traits is
not necessarily expected to match selection coefficients estimated in natural populations (Merilä et al.,
2001; Haller & Hendry, 2014). We are confident that
our results do reflect the action of selection owing to
the specific properties of our study system as discussed above. Similar inferences for other systems
would require similar background evidence that traits
are genetically based and responsive to natural selection, and that genetic drift and gene flow are not too
strong.
© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 112, 108–122
118
K. M. GOTANDA and A. P. HENDRY
Analyses such as those conducted in the present
study require data on genetically-based, adaptive
phenotypes in multiple populations over multiple
years/generations, and so it is worth considering
whether such data are likely to be commonly available. A first important point is that any system with
data to address spatiotemporal variation in selection
automatically has the data to examine spatiotemporal
variation in the traits themselves. Indeed, such
datasets are optimal because they allow consideration
of the extent to which spatiotemporal variation in
evolution and selection line up with each other, as
they might or might not depending on the situation
(see Introduction). A second important point is that
the examination of trait variation will be possible for
many additional data sets because it does not require
individual-based data (population means are sufficient) or fitness estimates. Examples of such data sets
include body size and shape in salmonids (Carlson &
Quinn, 2007), colour patterns in Cepaea snails
(Silvertown et al., 2011), pelvic structures in
threespine stickleback (Klepaker et al., 2012), gall
size in Eurosta flies (Weis, Abrahamson & Andersen,
1992), and morphological traits in deer mice (Pergams
& Ashley, 1999). We note that, for the same reasons
given above, it would also be extremely useful to
obtain estimates of selection for such systems. In
short, we are confident that many data sets exist for
which spatiotemporal variation in adaptive traits can
be examined, but we again caution that interpretations always should be conditioned by information on
the genetic basis and adaptive significance of the
studied traits.
We can see several useful extensions to the approach
advocated here. First, we studied a trait that was
genetically-based, but it also would be interesting to
consider spatiotemporal variation in very plastic
traits. For example, we might predict that plasticity
would lead to greater temporal variation because
plastic traits are presumably able to respond more
quickly to a local environmental change (Stearns,
1989; Lande, 2009; Pfennig et al., 2010). In our system,
this might be true if we had quantified some of the
more plastic components of colour such as carotenoid
deposition, brightness, and chroma. Second, we examined populations for which genetic drift was likely
unimportant, yet it would also be interesting to test
whether very small populations show greater temporal
variation in the spatial patterns of a phenotypic
trait. Third, we emphasized phenotypic variation in
quantitative traits, but similar analyses could be performed for frequencies of phenotypic or genetic
polymorphisms. Overall, we suggest that analyses of
spatiotemporal variation in adaptive traits complement (without replacing) analyses of spatiotemporal
variation in selection coefficients.
ACKNOWLEDGEMENTS
We thank Stephanie Carlson, Ben Haller, Adam
Siepielski, and three anonymous reviewers for comments that improved the manuscript. We also thank
the numerous field assistants who helped with collecting and photographing guppies (L. Baillie, P.
Bentzen, M. Boisjoly, F. Dargent, L. Delaire, S.
Gordon, D. Hoops, F. Pérez-Jvostev, A. McKeller, N.
Millar, S. Muttalib, I. Paterson, A. Schwartz, and M.
Turcotte), as well as Gregor Rolshausen for assisting
with the statistical analysis. All procedures regarding
handling of animals were in accordance with McGill
Animal Care Protocol number 4067. Funding was
provided by the Natural Science and Engineering
Research Council of Canada (NSERC) in the form of
Discovery and Special Research Opportunity grants
to APH and a Vanier Canada Graduate Scholarship
and an NSERC CGS-M to KMG. Additional funding
was provided by the Le Fonds Québécois de la
Recherche sur la Nature et les Technologies (FQRNT)
in the form of a scholarship to KMG.
REFERENCES
Aguirre WE, Bell MA. 2012. Twenty years of body shape
evolution in a threespine stickleback population adapting to
a lake environment. Biological Journal of the Linnean
Society 105: 817–831.
Archer SN, Lythgoe JN. 1990. The visual pigment basis for
cone polymorphism in the guppy, Poecilia reticulata. Vision
Research 30: 225–233.
Bell G. 2010. Fluctuating selection: the perpetual renewal of
adaptation in variable environments. Philosophical Transactions of the Royal Society Series B, Biological Sciences
365: 87–97.
Blanckenhorn WU, Morf C, Muhlhauser C, Reusch T.
1999. Spatiotemporal variation in selection on body size in
the dung fly Sepsis cynipsea. Journal of Evolutionary
Biology 12: 563–576.
Blows MW, Brooks R, Kraft PG. 2003. Exploring complex
fitness surfaces: multiple ornamentation and polymorphism
in male guppies. Evolution 57: 1622–1630.
Brooks R, Endler JA. 2001. Direct and indirect sexual
selection and quantitative genetics of male traits in guppies
(Poecilia reticulata). Evolution 55: 1002–1015.
Carlson SM, Quinn TP. 2007. Ten years of varying lake
level and selection on size-at-maturity in sockeye salmon.
Ecology 88: 2620–2629.
Carvalho GR, Shaw PW, Hauser L, Seghers BH,
Magurran AE. 1996. Artificial introductions, evolutionary
change and population differentiation in Trinidadian
guppies (Poecilia reticulata:Poeciliidae). Biological Journal
of the Linnean Society 57: 219–234.
Cox CL, Rabosky ARD. 2013. Spatial and temporal drivers
of phenotypic diversity in polymorphic snakes. American
Naturalist 182: E40–E57.
© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 112, 108–122
SPATIOTEMPORAL VARIATION IN GUPPY COLOUR
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.
Endler JA. 1978. A predator’s view of animal color patterns.
Evolutionary Biology 11: 319–364.
Endler JA. 1980. Natural selection on color patterns in
Poecilia reticulata. Evolution 34: 76–91.
Endler JA. 1986. Natural selection in the wild. Princeton,
NJ: Princeton University Press.
Endler JA. 1991. Variation in the appearance of guppy color
patterns to guppies and their predators under different
visual conditions. Vision Research 31: 587–608.
Endler JA, Houde AE. 1995. Geographic variation in female
preferences for male traits in Poecilia reticulata. Evolution
49: 456–468.
Endler JA, Mielke PW. 2005. Comparing entire colour patterns as birds see them. Biological Journal of the Linnean
Society 86: 405–431.
Fraser BA, Ramnarine IW, Neff BD. 2010. Temporal variation at the MHC class IIB in wild populations of the guppy
(Poecilia reticulata). Evolution 64: 286–2096.
Godin JGJ, McDonough HE. 2003. Predator preference for
brightly colored males in the guppy: a viability cost for a
sexually selected trait. Behavioral Ecology 14: 194–200.
Gordon SP, López-Sepulcre A, Reznick DN. 2012.
Predation-associated differences in sex linkage of wild
guppy coloration. Evolution 66: 912–918.
Gosden TP, Svensson EI. 2008. Spatial and temporal
dynamics in a sexual selection mosaic. Evolution 62: 845–
856.
Gotanda KM, Delaire LC, Raeymaekers JAM,
Pérez-Jvostov F, Dargent F, Bentzen P, Scott ME,
Fussmann GF, Hendry AP. 2013. Adding parasites to
the guppy-predation story: insights from field surveys.
Oecologia 172: 155–166.
Gotanda KM, Hendry AP. 2014. Data from: using adaptive
traits to consider potential consequences of temporal variation in selection: male guppy colour through time and
space. Dryad Digital Repository doi:10.5061/dryad.pj02h.
Grant PR, Grant BR. 2002. Unpredictable evolution in a
30-year study of Darwin’s finches. Science 296: 707–711.
Grether GF. 2000. Carotenoid limitation and mate preference evolution: a test of the indicator hypothesis in guppies
(Poecilia reticulata). Evolution 54: 1712–1724.
Grether GF, Cummings ME, Hudon J. 2005.
Countergradient variation in the sexual coloration of
guppies (Poecilia reticulata): drosopterin synthesis balances
carotenoid availability. Evolution 59: 175–188.
Grether GF, Hudon J, Endler JA. 2001a. Carotenoid scarcity, synthetic pteridine pigments and the evolution of
sexual coloration in guppies (Poecilia reticulata). Proceedings of the Royal Society of London Series B, Biological
Sciences 268: 1245–1253.
Grether GF, Millie DF, Bryant MJ, Reznick DN, Mayea
W. 2001b. Rain forest canopy cover, resource availability,
and life history evolution in guppies. Ecology 82: 1546–
1559.
119
Griffith SC, Parker TH, Olson VA. 2006. Melanin- vs.
carotenoid-based sexual signals: is the difference really so
black and red? Animal Behaviour 71: 749–763.
Haller BC, Hendry AP. 2014. Solving the paradox of stasis:
squashed stabilizing selection and the limits of detection.
Evolution 68: 483–500.
Hearing VJ. 1993. Unraveling the melanocyte. American
Journal of Human Genetics 52: 1–7.
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.
Hereford J, Hansen TF, Houle D. 2004. Comparing
strengths of directional selection: how strong is strong?
Evolution 58: 2133–2143.
Hersch EI, Phillips PC. 2004. Power and potential bias in
field studies of natural selection. Evolution 58: 479–485.
Houde AE. 1992. Sex-linked heritability of a sexually selected
character in a natural popualtion of Poecilia reticulata
(Pisces: Poeciliidae) (guppies). Heredity 69: 229–235.
Houde AE. 1997. Sex, color, and mate choice in guppies.
Princeton, NJ: Princeton University Press.
Houde AE, Endler JA. 1990. Correlated evolution of female
mating preferences and male color patterns in the guppy
Poecilia reticulata. Science 248: 1405–1408.
Houde AE, Torio AJ. 1992. Effect of parasitic infection on
male color pattern and female choice in guppies. Behavioral
Ecology 3: 346–351.
Hughes KA, Houde AE, Price AC, Rodd FH. 2013. Mating
advantage for rare males in wild guppy populations. Nature
503: 108–110.
Hughes KA, Rodd FH, Reznick DN. 2005. Genetic and
environmental effects on secondary sex traits in guppies
(Poecilia reticulata). Journal of Evolutionary Biology 18:
35–45.
Karim N, Gordon SP, Schwartz AK, Hendry AP. 2007.
This is not déjà vu all over again: male guppy colour in a
new experimental introduction. Journal of Evolutionary
Biology 20: 1339–1350.
Karino K, Haijima Y. 2001. Heritability of male secondary
sexual traits in feral guppies in Japan. Journal of Ethology
19: 33–37.
Kelly CD, Godin JGJ, Wright JM. 1999. Geographical
variation in multiple paternity within natural populations
of the guppy (Poecilia reticulata). Proceedings of the Royal
Society of London Series B, Biological Sciences 266: 2403–
2408.
Kemp DJ, Reznick DN, Grether GF. 2008. Ornamental
evolution in Trinidadian guppies (Poecilia reticulata):
insights from sensory processing-based analyses of entire
colour patterns. Biological Journal of the Linnean Society
95: 734–747.
Kemp DJ, Reznick DN, Grether GF, Endler JA. 2009.
Predicting the direction of ornament evolution in Trinidadian guppies (Poecilia reticulata). Proceedings of the Royal
Society of London B, Biological Sciences 276: 4335–4343.
Kingsolver J, Diamond S, Siepielski A, Carlson S. 2012.
Synthetic analyses of phenotypic selection in natural
© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 112, 108–122
120
K. M. GOTANDA and A. P. HENDRY
populations: lessons, limitations and future directions. Evolutionary Ecology 26: 1101–1118.
Kingsolver JG, Hoekstra HE, Hoekstra JM, Berrigan D,
Vignieri SN, Hill CE, Hoang A, Gibert P, Beerli P.
2001. The strength of phenotypic selection in natural populations. American Naturalist 157: 245–261.
Klepaker T, Østbye K, Bernatchez L, Vøllestad LA. 2012.
Spatio-temporal patterns in pelvic reduction in threespine
stickleback (Gasterosteus aculeatus L.) in Lake Storvatnet.
Evolutionary Ecology Research 14: 169–191.
Kodric-Brown A. 1985. Female preference and sexual selection for male coloration in the guppy (Poecilia reticulata).
Behavioral Ecology and Sociobiology 17: 199–205.
Kodric-Brown A. 1989. Dietary carotenoids and male
mating success in the guppy – an environmental component
to female choice. Behavioral Ecology and Sociobiology 25:
393–401.
Kröger RHH, Bowmaker JK, Wagner HJ. 1999. Morphological changes in the retina of Aequidens pulcher
(Cichlidae) after rearing in monochromatic light. Vision
Research 39: 2441–2448.
Kudo H, Karino K. 2012. Short-term change in male sexually selected traits and female mate preference in the
guppy Poecilia reticulata. Ichthyological Research 59: 1–
7.
Kudo H, Karino K. 2013. Negative correlation between male
ornament size and female preference intensity in a wild
guppy population. Behavioral Ecology and Sociobiology 67:
1931–1938.
Lande R. 2009. Adaptation to an extraordinary environment
by evolution of phenotypic plasticity and genetic assimilation. Journal of Evolutionary Biology 22: 1435–1446.
Langerhans RB, DeWitt TJ. 2004. Shared and unique features of evolutionary diversification. American Naturalist
164: 335–349.
Long KD, Houde AE. 1989. Orange spots as a visual cue for
female mate choice in the guppy (Poecilia reticulata). Ethology 82: 316–324.
Magurran AE. 2005. Evolutionary ecology: the Trinidadian
guppy. New York, NY: Oxford University Press.
Magurran AE, Seghers BH, Carvalho GR, Shaw PW.
1992. Behavioral consequences of an artificial introduction
of guppies (Poecilia reticulata) in N. Trinidad: evidence for
the evolution of anti-predator behaviour in the wild. Proceedings of the Royal Society of London Series B, Biological
Sciences 248: 117–122.
Martin CH, Johnsen S. 2007. A field test of the Hamilton–
Zuk hypothesis in the Trinidadian guppy (Poecilia
reticulata). Behavioral Ecology and Sociobiology 61: 1897–
1909.
McGraw KJ, Mackillop EA, Dale J, Hauber ME. 2002.
Different colors reveal different information: how nutritional stress affects the expression of melanin- and structurally based ornamental plumage. Journal of Experimental
Biology 205: 3747–3755.
McKellar A, Turcotte M, Hendry A. 2009. Environmental
factors influencing adult sex ratio in Trinidadian guppies.
Oecologia 159: 735–745.
Merilä J, Sheldon BC, Kruuk LEB. 2001. Explaining
stasis: microevolutionary studies in natural populations.
Genetica 112: 199–222.
Millar N, Hendry A. 2012. Population divergence of private
and non-private signals in wild guppies. Environmental
Biology of Fishes 94: 513–525.
Millar NP, Reznick DN, Kinnison MT, Hendry AP. 2006.
Disentangling the selective factors that act on male colour
in wild guppies. Oikos 113: 1–12.
Miller LK, Brooks R. 2005. The effects of genotype,
age, and social environment on male ornamentation,
mating behavior, and attractiveness. Evolution 59: 2414–
2425.
Morrissey MB, Hadfield JD. 2012. Directional selection in
temporally replicated studies is remarkably consistent. Evolution 66: 435–442.
Morrissey MB, Kruuk LEB, Wilson AJ. 2010. The danger
of applying the breeder’s equation in observational studies
of natural populations. Journal of Evolutionary Biology 23:
2277–2288.
van Oosterhout C, Mohammed RS, Hansen H, Archard
GA, McMullan M, Weese DJ, Cable J. 2007. Selection by
parasites in spate conditions in wild Trinidadian guppies
(Poecilia reticulata). International Journal for Parasitology
37: 805–812.
Pergams ORW, Ashley MV. 1999. Rapid morphological
change in Channel Island deer mice. Evolution 53: 1573–
1581.
Pettersson L, Ramnarine I, Becher SA, Mahabir R,
Magurran A. 2004. Sex ratio dynamics and fluctuating
selection pressures in natural populations of the Trinidadian guppy, Poecilia reticulata. Behavioral Ecology and Sociobiology 55: 461–468.
Pfennig DW, Wund MA, Snell-Rood EC, Cruickshank T,
Schlichting CD, Moczek AP. 2010. Phenotypic plasticity’s
impacts on diversification and speciation. Trends in Ecology
& Evolution 25: 459–467.
Pilastro A, Simonato M, Bisazza A, Evans JP. 2004.
Cryptic female preference for colorful males in guppies.
Evolution 58: 665–669.
Pitcher TE, Rodd FH, Rowe L. 2007. Sexual colouration
and sperm traits in guppies. Journal of Fish Biology 70:
165–177.
R Core Team. 2012. R: A language and environment for
statistical computing. Vienna, Austria: R Foundation for
Statistical Computing.
Reznick D. 1989. Life-history evolution in guppies: 2.
Repeatability of field observations and the effects of season
on life histories. Evolution 43: 1285–1297.
Reznick D, Bryant MJ, Bashey F. 2002. r- and K-selection
revisited: the role of population regulation in life-history
evolution. Ecology 83: 1509–1520.
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.
Rodd FH, Reznick DN. 1997. Variation in the demography
of guppy populations: the importance of predation and life
histories. Ecology 78: 405–418.
© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 112, 108–122
SPATIOTEMPORAL VARIATION IN GUPPY COLOUR
Ruell EW, Handelsman CA, Hawkins CL, Sofaer HR,
Ghalambor CK, Angeloni L. 2013. Fear, food and sexual
ornamentation: plasticity of colour development in Trinidadian guppies. Proceedings of the Royal Society of London
Series B, Biological Sciences 280: 1471–2954.
Schluter D. 2000. The ecology of adaptive radiation. Oxford:
Oxford University Press.
Schwartz AK, Hendry AP. 2007. A test for the parallel
co-evolution of male colour and female preference in Trinidadian guppies (Poecilia reticulata). Evolutionary Ecology
Research 9: 71–90.
Schwartz AK, Hendry AP. 2010. Testing the influence
of local forest canopy clearing on phenotypic variation
in Trinidadian guppies. Functional Ecology 24: 354–
364.
Siepielski AM, DiBattista JD, Carlson SM. 2009. It’s
about time: the temporal dynamics of phenotypic selection
in the wild. Ecology Letters 12: 1261–1276.
Siepielski AM, Gotanda KM, Morrissey MB, Diamond
SE, DiBattista JD, Carlson SM. 2013. The spatial patterns of directional phenotypic selection. Ecology Letters 16:
1382–1392.
Silvertown J, Cook L, Cameron R, Dodd M, McConway
K, Worthington J, Skelton P, Anton C, Bossdorf O,
Baur B, Schilthuizen M, Fontaine B, Sattmann H,
Bertorelle G, Correia M, Oliveira C, Pokryszko B,
Oz˙go M, Stalažs A, Gill E, Rammul Ü, Sólymos P,
Féher Z, Juan X. 2011. Citizen science reveals unexpected
121
continental-scale evolutionary change in a model organism.
PLoS ONE 6: e18927.
Stearns SC. 1989. The evolutionary significance of phenotypic plasticity. Bioscience 39: 436–445.
Sternalski A, Mougeot F, Bretagnolle V. 2012. Phenotypic
variation in nestlings of a bird of prey under contrasting
breeding and diet conditions. Biological Journal of the
Linnean Society 107: 799–812.
Stevens M, Parraga CA, Cuthill IC, Partridge JC,
Troscianko TS. 2007. Using digital photography to study
animal coloration. Biological Journal of the Linnean Society
90: 211–237.
Tripathi N, Hoffmann M, Willing EM, Lanz C, Weigel D,
Dreyer C. 2009. Genetic linkage map of the guppy, Poecilia
reticulata, and quantitative trait loci analysis of male size
and colour variation. Proceedings of the Royal Society of
London Series B, Biological Sciences 276: 2195–2208.
Weese DJ, Gordon SP, Hendry AP, Kinnison MT. 2010.
Spatiotemporal variation in linear natural selection on body
color in wild guppies (Poecilia reticulata). Evolution 64:
1802–1815.
Weese DJ, Schwartz AK, Bentzen P, Hendry AP,
Kinnison MT. 2011. Eco-evolutionary effects on population
recovery following catastrophic disturbance. Evolutionary
Applications 4: 354–366.
Weis AE, Abrahamson WG, Andersen MC. 1992. Variable
selection on Eurostas gall size, I: the extent and nature of
variation in phenotypic selection. Evolution 46: 1674–1697.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the online version of this article at the publisher’s web-site:
Figure S1. Map of the Marianne and Paria rivers on the northern slope of the Trinidadian mountain range.
Figure S2. Spot numbers of colour category by year and site
Table S1. Mean ± SE at each site in each year for relative areas (arcsine square root transformed) of each colour
category (carotenoid, structural, melanic, and green). Also shown is the relative area of all colour categories
combined (total).
Table S2. Mean ± SE at each site in each year for number of coloured spots of each colour category (carotenoid,
structural, melanic, and green). Also shown is the mean number of spots of all colour categories combined (total).
Table S3. Adjusted P-values for post-hoc Tukey’s honestly significant difference test results on pairwise
comparisons between sites for carotenoid coloration. Top right is for relative area and lower left is for spot
number. Statistical significance (P < 0.05) is indicated in bold.
Table S4. Adjusted P-values for post-hoc Tukey’s honestly significant difference test results on pairwise
comparisons between sites for structural coloration. Top right is for relative area and lower left is for spot
number. Statistical significance (P < 0.05) is indicated in bold.
Table S5. Results from multivariate analysis of covariance using Wilk’s lambda test statistic with site fitted
before year as fixed effects and body size (mm2; log transformed) as the covariate for sites within the Marianne
river. Estimates of variance were calculated as multivariate partial η2 also utilizing Wilk’s lambda. The
dependent variables were relative area (arcsine square root transformed) and spot number (square root
transformed) for all colour categories (carotenoid colour, structural colour, melanic colour, and green). Statistical
significance (P < 0.05) is indicated in bold.
Table S6. Results from multivariate analysis of covariance using Wilk’s lambda test statistic with site fitted
before year as fixed effects and body size (mm2, log transformed) as the covariate for sites within the Paria river.
Estimates of variance were calculated as multivariate partial η2 also utilizing Wilk’s lambda. The dependent
variables were relative area (arcsine square root transformed) and spot number (square root transformed) for
© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 112, 108–122
122
K. M. GOTANDA and A. P. HENDRY
all colour categories (carotenoid colour, structural colour, melanic colour, and green). Statistical significance
(P < 0.05) is indicated in bold.
Table S7. Pearson-product moment correlation test results comparing the relative area of blue and orange
colour for individual years. Statistical significance (P < 0.05) is indicated in bold.
Table S8. Results from multivariate analysis of covariance (MANCOVA) using Wilk’s lambda test statistic with
year as a randomized block, site as a fixed effect, the year-by-site interaction, and body size (mm2; log
transformed) as the covariate. All MANCOVAs were carried out in the R working environment utilizing the car
package. The dependent variables were relative area (arcsine square root transformed) and spot number (square
root transformed) for all colour categories (carotenoid, structural, melanic, and green). Statistical significance
(P < 0.05) is indicated in bold.
ARCHIVED DATA
Data deposited at Dryad (Gotanda & Hendry, 2014).
© 2014 The Linnean Society of London, Biological Journal of the Linnean Society, 2014, 112, 108–122

Similar documents

×

Report this document