Академический Документы
Профессиональный Документы
Культура Документы
MINI REVIEW
The evolution of trade-offs: where are we?
Keywords: Abstract
acquisition; Trade-offs are a core component of many evolutionary models, particularly
allocation; those dealing with the evolution of life histories. In the present paper, we
quantitative genetics; identify four topics of key importance for studies of the evolutionary biology of
trade-offs; trade-offs. First, we consider the underlying concept of constraint. We
Y-model. conclude that this term is typically used too vaguely and suggest that
constraint in the sense of a bias should be clearly distinguished from
constraint in the sense of proscribed combinations of traits or evolutionary
trajectories. Secondly, we address the utility of the acquisitionallocation
model (the Y-model). We find that, whereas this model and its derivatives
have provided new insights, a misunderstanding of the pivotal equation has
led to incorrect predictions and faulty tests. Thirdly, we ask how trade-offs are
expected to evolve under directional selection. A quantitative genetic model
predicts that, under weak or short-term selection, the intercept will change
but the slope will remain constant. Two empirical tests support this prediction
but these are based on comparisons of geographic populations: more direct
tests will come from artificial selection experiments. Finally, we discuss what
maintains variation in trade-offs noting that at present little attention has been
given to this question. We distinguish between phenotypic and genetic
variation and suggest that the latter is most in need of explanation. We suggest
that four factors deserving investigation are mutation-selection balance,
antagonistic pleiotropy, correlational selection and spatio-temporal variation,
but as in the other areas of research on trade-offs, empirical generalizations are
impeded by lack of data. Although this lack is discouraging, we suggest that it
provides a rich ground for further study and the integration of many
disciplines, including the emerging field of genomics.
primarily to a single aspect. To help distinguish the two use. The ambiguity in the literature has certainly not
usages, we use the subscripted symbols X1 and X2, when lessened since this call in 1994 and so, because of the
a functional trade-off is assumed, whereas when the potential confusion in the meaning of constraint, we
trade-off is measured by a regression analysis we use Y advocate not using the term unless it is precisely defined
and X, although the distinction clearly becomes blurred in the given context.
at times.
Trade-offs between life history traits such as between
Quantitative genetics and absolute constraints
fecundity and survival, measured as either functional or
statistical relationships (more frequently the latter) have Trade-offs are specified in quantitative genetics by a
been demonstrated in a large number of studies and negative genetic covariance between traits, a covariance
numerous taxa in laboratory, semi-natural and natural that could be caused by antagonistic pleiotropy or linkage
populations (e.g. Reznick, 1985; Partridge & Sibley, 1991; disequilibrium. In both cases a causal connection is
Roff, 1992, 2002; Stearns, 1992; Gustafsson et al., 1994; inferred. This specification is in terms of the bivariate
Ots & Horak, 1996; Sinervo & DeNardo, 1996; Zuk, relationship and hence does not take into account the
1996). While neither the existence of fitness trade-offs effect of interactions with other variables. However, the
nor their central place in shaping evolutionary trajector- sum total of all interactions with other traits is taken into
ies is in doubt, there is still little understanding, from account by use of the entire genetic and phenotypic
either a theoretical or empirical perspective, of how such variancecovariance matrix to predict response to selec-
trade-offs evolve (Houle, 1991; Chippendale et al., 1996; tion. Given these matrices it is possible to define precisely
Fry, 1996; Reznick et al., 2000; Roff & DeRose, 2001; the circumstances under which some evolutionary tra-
Roff, 2002). In this brief review we discuss four main jectories are not permitted. An important conclusion of
topics that we believe are important for an understanding such analyses, which we describe below, is that when
of how trade-offs evolve. These topics encompass major trade-offs involve more than two traits, some evolution-
issues that remain unresolved or misunderstood in the ary trajectories may be proscribed even if all of the
current literature and are discussed under the following bivariate correlations are greater than )1. Thus, absolute
headings: genetic constraints may exist, at least in the short term, in
1 Problems with the concept of constraint. spite of imperfect genetic correlations.
2 Insights from models of resource acquisitionalloca- Variancecovariance matrices are symmetric and
tion. hence can be reduced, using principal components
3 The effect of directional selection on the statistical analysis, to a set of orthogonal axes designated by the
description of trade-offs. eigenvectors. Each axis is made up of a linear combina-
4 What maintains variation in the trade-off? tion of the individual traits (the principal component
scores) and there are as many axes as there are traits. The
variance in each principal component is given by the
Problems with the concept of constraint
eigenvalue. If an eigenvalue is zero there is no genetic
variance in the respective direction and hence evolution
Problems of definition
cannot proceed in that direction. To illustrate this we
In referring to the effect of negative genetic correlations consider two situations, one in which there are two traits
on evolutionary change, the term constraint is used in and one in which there are three traits.
two senses: first, it is used in the sense of impeding, but Consider the trade-off illustrated in Fig. 1: from a
not stopping, evolution in particular directions, and quantitative genetic perspective the trade-off is described
second, it is used to mean that there are evolutionary by a bivariate normal distribution (as noted in the
trajectories that are unavailable to selection, termed Introduction, the symbols X and Y are here used to
evolutionarily forbidden trajectories by Kirkpatrick & designate that the traits are being considered within a
Lofsvold (1992) and absolute evolutionary constraints statistical framework). Under this model, provided there
by Mezey & Houle (2005). Unfortunately, it is frequently is variation orthogonal to the axis describing the trade-
not clear in which sense constraint is being used. This is off, which means that the genetic correlation between
not a trivial source of confusion, because under the the two traits is greater than )1, all combinations of
former meaning all character states are possible, whereas values are possible, although the frequency of combina-
under the latter meaning some states are proscribed. We tions obviously varies greatly. Thus, under this model
agree with Perrin & Travis (1992) that it is perfectly selection can, in principle, push a population in any
reasonable to use the term constraint in a mathematical direction and the trait combinations are only constrained
sense, as for example, in the verbal statement of in the sense that for a given selection differential the
X1 + X2 < Z, meaning X1 plus X2 is constrained to be response will be greater in some directions than others.
less than Z. However, because of frequent ambiguity in Long-term evolution to any combination is possible. The
its meaning, we are also sympathetic to the call by Van only case in which evolution is constrained in the sense
Tienderen & Antonovics (1994) for a moratorium on its that some combinations cannot evolve, is that in which
P(X,Y)
Trait Y
Fig. 1 A quantitative genetic view of a
trade-off between traits X and Y shown from
two orientations. Each plot shows the
bivariate normal probability distribution of
Trait Y
breeding values for each trait (where P(X,Y)
denotes the probability of the XY combina-
tion), with the left plot showing it in con-
tour perspective and the right plot showing
Trait X Trait X
Y)
it in 3D mode. The solid lines in the left plot
X,
P(
show the major and minor axes.
the genetic correlation between traits is exactly )1. This negative correlation between traits X and Y that is greater
condition is also specified, as noted above, by one of the than )1. From this we could, incorrectly conclude that
eigenvalues of the matrix being zero. To obtain a visual evolution in any direction is possible. In fact, depending
understanding of this condition consider what happens if on the distribution of points, the projection of points onto
we rotate the axes such that they now fall along the the X-Y plane could produce a zero correlation, and thus
major and minor axes of the bivariate normal distribu- it could appear that there was no trade-off between traits
tion (Fig. 1): the equations specifying this rotation are X and Y. This is what Pease & Bull (1988) referred to as
given by the eigenvectors of the matrix. We now have the problem of dimensionality. A bivariate genetic
two uncorrelated traits made up from a linear combina- correlation less than )1 is clearly insufficient evidence
tion of the original two correlated traits, with the for the conclusion that all evolutionary trajectories are
eigenvalues being the genetic variances of these two possible. In general, statistical representations of bivariate
synthesized traits (Kirkpatrick & Lofsvold, 1992). When trade-offs permit, at best, only weak inferences about
r )1 there is no variation in the direction of the minor how constrained, in the sense of being biased, evolu-
axis (the second eigenvalue) and thus selection will be tionary trajectories are likely to be. While it is probably
totally ineffective in producing a change in this direction. true that, in most cases, selection in the direction of the
Note, however, that there can still be genetic variation largest eigenvalue will be the fastest, failure to include
for both of the original traits. other traits could still lead to misleading predictions.
When the genetic correlation is greater than )1, the Even if there are no eigenvalues that are exactly zero,
above scenario presents a logical conundrum, because if movement along a particular evolutionary trajectory may
both traits covary positively with fitness there is appar- be very slow if the eigenvalue in that direction is very
ently nothing stopping the population moving off to ever small relative to the other eigenvalues (Blows &
increasing values of each trait. We postulate that this Hoffmann, 2005). If we wish to make statements about
does not happen because the variation about the line the importance of particular trade-offs in modulating and
does not truly reflect a bivariate normal distribution but directing evolutionary change, it is necessary to know
is an approximation built up from the interaction of how this trade-off is integrated with other traits, and thus
multiple traits, which together do prevent evolution in the extent to which variation observed on the X-Y plane
particular directions, or at least so biases it that particular actually represents variation that is in actuality more
directions are highly unlikely. In other words, the restrictive than implied by the simple bivariate statistical
statistical description of the trade-off is a consequence relationship. While this is possible, in principle, by
of a functional trade-off in multivariate space being measuring the variances and covariances of a wide suite
projected onto a two-dimensional surface. To illustrate of traits, such an approach is time consuming, potentially
this, consider the situation in which three traits are costly and not guaranteed to include the requisite suite of
functionally constrained to lie upon a plane as shown in traits. We suggest that a better approach is to combine a
Fig. 2. It is immediately obvious that selection cannot quantitative genetic analysis with a phenotypic analysis
drive the population to any combination of trait values that focuses on the underlying functional relationships,
that lie off the plane, although there can be genetic paying particular attention to the possible influence of
variation for all three traits. However, if we project the unmeasured variables (path analysis may be of consider-
observed trait values onto the X-Y plane (i.e. rotate the able use in this). Charlesworths (1990) analysis of a
axes such that the Z-axis is perpendicular to the surface hypothetical life history with functional constraints
of the page; Fig. 2), we observe a scatter of points with an illustrates this approach, as does the theoretical and
overall trade-off indicated by a statistically significant empirical analyses of the evolution of growth trajectories
1.0 1.0
0.5
0.5
Trait Y 0.0
Trait Y
0.0
0.5
1.0
0.0 0.5
0.2
0.2
0.4
Trait Z
0.4
0.6
Tra 0.6
0.8
it X 0.8 Z 1.0
1.0 it
1.0 Tra 0.2 0.4 0.6 0.8 1.0
Trait X
Fig. 2 An illustration of how a trivariate trade-off may appear as a bivariate trade-off with variation about the trade-off line when one trait is
omitted from the analysis. The actual three trait functional trade-off is shown on the left (Y 1 ) X ) Z, or, because this is the functional form,
X2 1 ) X1 ) X3). The figure on the right shows the plot of X and Y when Z is omitted: now the distribution appears as a bivariate distribution
with a correlation greater than )1 and hence in this projection there appears to be no absolute constraint on the direction of evolution.
Combinations of X and Y that can be achieved are contingent on the fitness of the three-fold combination of X, Y and Z which is limited by the
functional trade-off.
by Kirkpatrick et al. (1990) and Kirkpatrick & Lofsvold most influential models explaining how such positive
(1992). The latter analyses and that of wing shape in correlations can arise in the presence of trade-offs is the
Drosophila melanogaster (Mezey & Houle, 2005) further acquisitionallocation model of Van Noordwijk & de Jong
illustrate the statistical problem of demonstrating that (1986), which was first formulated by James (1974) (his
any given eigenvalue is exactly zero: the best that we can analysis contains an arithmetic error, which, unfortu-
achieve is the statement that a particular eigenvalue is no nately, is reproduced in Roff, 1992 and Roff, 2002) and
greater than some positive value. On the other hand, also derived by Riska (1986). James called the model the
provided the confidence limits are reasonably small, it partitioning of resources model, whereas Riska called it
should be possible to delineate likely from unlikely a variable parts model and De Jong & van Noordwijk
evolutionary trajectories. An alternative approach for (1992) referred to it as the Y-model, which, for
delineating the limitations imposed by functional trade- simplicity, is the name we use in this paper. This model
offs is experimental manipulation of the component posits that within an individual two traits (X1,X2) are
traits (reviewed in Roff, 2002, pp. 132142, and for an determined by the allocation of resources from a com-
excellent discussion on allometric engineering to dem- mon pool, T: T X1 + X2 (note that we here use the
onstrate trade-offs see Sinervo et al., 1992) and this can symbols X1 and X2 rather than X and Y, because we are
profitably be combined with predictions made from more referring to the functional form, rather than the statistical
classic quantitative genetic analyses. manifestation, of the trade-off). Given this mathematical
relationship we can state that the value of trait X2 is
restricted in its possible values by the allocation of some
Insights from models of resource
proportion, P, of the acquired resource T to X1, leaving
acquisitionallocation
(1 ) P)T to be allocated to X2. Using this model, and
assuming that variation exists among individuals, we can
The Y-model
scale up to the level of the population and calculate the
The concept that trade-offs between fitness-related traits covariance between the two traits X1 and X2, denoted as
are fundamental in shaping both evolutionary trajector- rX1X2, as:
ies and equilibrium trait values has proved to be a
rX1 X2 r2T lP 1 lP r2P l2T r2P ; 1
powerful heuristic tool, with strong theoretical and
empirical support (e.g. Roff, 1992, 2002; Stearns, 1992; where l designates mean values and r2 designates
Reznick et al., 2000). Nevertheless, predicted trade-offs, variances. For a fixed acquisition (r2T 0) the covari-
generally measured by their statistical correlation, are not ance between X1 and X2 is negative and the correlation is
always found. In fact, positive phenotypic correlations )1, indicating the functional trade-off. However, when
between traits predicted to be involved in fitness trade- there is variability in acquisition it is possible for the
offs (and hence expected to be negatively correlated) are covariance between X1and X2 to be positive, giving the
uncovered not infrequently in laboratory and natural false impression, as measured by the statistical relation-
populations (Reznick et al., 2000; Roff, 2002). One of the ship between X1 and X2, that there is no trade-off. This
Table 1 Illustrative analysis of the trade-off between allocation to ovaries (O) vs. soma (S) in the wing dimorphic sand cricket, Gryllus firmus.
Acquisition is approximated by the total body mass, T, and P is the proportion allocated to the ovaries. For further details see Crnokrak & Roff
(2002).
T VT
P VP CovOS
the scenario just described, the strength of the correlation Loman, 2003) argues for a greater need for an under-
should increase as ration is decreased. A critical assump- standing of the functional basis of trade-offs, particularly
tion of these predictions is that the pattern of allocation with respect to the adaptive significance of patterns of
of resources among components does not change with allocation under different acquisition regimes. The
resource availability. Unfortunately, this assumption may Y-model is a powerful conceptual and analytical tool but
not be valid (Sgro & Hoffmann, 2004; and see the a misinterpretation of its predictions has led to an unfor-
example below). tunate number of incomplete empirical investigations.
To illustrate the effect of truncation of the resource
distribution on patterns of allocation, we present data on
The effect of directional selection on
the allocation to ovary mass and the main flight muscles
trade-offs
(the dorso-longitudinal muscles, hereafter DLM) in the
macropterous (long-winged) morph of the sand cricket,
A quantitative genetic perspective
G. firmus. This experiment is described in detail elsewhere
(Roff & Gelinas, 2003) and we note here only the In general, trade-off functions are described empirically
pertinent features: nymphs were raised on ad libitum food by the simple linear regression between the two traits,
but adults were fed either ad libitum or at a rate that had which under the quantitative genetic framework can be
been shown to reduce seven day fecundity (measured by written as:
ovary mass) by approximately one half. On the ad libitum
rPXY rPXY
diet the combined mass of ovaries plus DLM was 0.162 g Y lY 2 lX 2 X e; 9
rPX rPX
(SE 0.004, n 442, CV 48.31%), whereas on the
low ration it was 0.089 g (SE 0.002, n 414, CV where lX and lY are the mean values of traits X and Y
13.3%). (These values differ slightly from those calcula- respectively; rPXY is the phenotypic covariance between
ted from Table 1 of Roff & Gelinas, because in the present traits X and Y, r2PX is the phenotypic variance of trait X
analysis we included only females for which we had both and e is a normally distributed error term (Roff et al.,
variables.) For the present analysis we are concerned 2002). The first terms in parentheses define the intercept
with how the food allocated to the combination of of the regression line and the terms in the second set of
ovaries plus DLM is distributed between these two parentheses define the slope. As the designation of which
components and therefore we can consider the combined trait as the dependent or independent variable is largely
mass as the total acquisition. The low ration reduced total arbitrary, a better statistical model may be to define the
acquisition by 45% and the CV of acquisition by 72%. trade-off as the principal axis of the bivariate normal
The percentage of this total acquisition allocated to DLM distribution that relates the two traits as in Fig. 1, but this
averaged 15.5% for females on the ad libitum diet, and does not change the qualitative predictions.
23.7% for females on the low ration, a difference that is In principle, selection will eventually change the
highly significant (t854 8.1335, P < 0.0001, Kruskal variances and covariances (Bohren et al., 1966; Falconer,
Wallis test, v21 57.80, P < 0.0001; proportions trans- 1989; Roff, 1997) which would thereby change both the
formed using arcsine square-root). In contrast, the intercept and the slope of the trade-off function. How-
correlation between ovary mass and DLM mass was ever, for the infinitesimal model, short-term directional
)0.591 and )0.565 on ad libitum and low rations selection does not change the shape of the distribution of
respectively, a difference that is not significant (t breeding values, except under extreme conditions, and
0.5697, P > 0.5). Thus a severe reduction in acquisition response is dominated by changes in trait means (Barton
did not result in a change in the correlation (i.e. the & Turelli, 1987; Turelli & Barton, 1994). This prediction is
trade-off), but did change the pattern of allocation. supported by the empirical observation that artificial
Phenotypic plasticity in resource allocation is a com- selection experiments over 1015 generations generally
mon phenomenon (e.g. Reznick, 1983; Smith & Davies, have little effect on heritabilities and genetic correlations
1997; Billerbeck et al., 2000; Li et al., 2001; Jordan & (Roff, 1997). With respect to trade-offs, an important
Snell, 2002; Bochdanovits & de Jong, 2003; Sgro & distinction made by De Jong (1990) is between what the
Hoffmann, 2004) and invalidates predictions of the sign author has termed structured pleiotropy and unstruc-
of the correlation between two traits based solely upon tured pleiotropy. Gene substitutions in the former
the observed change in the variance or coefficient of category produce correlated effects on both traits
variation in resource acquisition. The assumption that a whereas those in the second category do not, although
restriction in acquisition will either convert a positive they might still affect both traits. Structural pleiotropy is
covariance to a negative covariance or increase the expected when there is a developmental constraint or
magnitude of the negative correlation is incorrect both functional constraint underlying genetic covariances (De
on theoretical and empirical grounds. That such observa- Jong, 1990, p. 459), and should therefore, be commonly
tions have been made in some cases (e.g. Biere, 1995; found in functionally based trade-offs. Theoretical ana-
Glazier, 1999; Messina & Slade, 1999; Donohue et al., lysis shows that the covariance between two traits is
2000) but not others (e.g. Glazier, 1999; Lardner & more resistant to change when determined by structured
pleiotropy than when determined by unstructured plei- among more than two traits (i.e. the hierarchical
otropy (De Jong, 1990), and provides another reason Y-model may be more appropriate). As formulated, the
why the slope of a trade-off will not vary substantially Y-model is completely deterministic within an individual
under directional selection. but in many, if not most, cases the variation in the two
Thus, for the reasons described above, the slope of the component traits will also be subject to other influences
trade-off function under short-term selection is expected and hence may be treated as separate, although corre-
to remain constant, except under very restrictive condi- lated, traits.
tions (Roff et al., 2002), whereas the intercept will
change because of changes in lY and lX. The change in
Predictions and tests
the intercept for truncation selection, which is that
typically applied in artificial selection experiments, can The above equations predict that, under short term or
be predicted using the standard response equation. For weak selection, the trade-off function, as measured in its
selection on a single trait, Y, the intercept will change statistical context, will evolve by a shift in the intercept
according to: alone, defined either as the regression line or principal
at1 at RY bCRX at ihY hY rPY rA hX brPX ; 10 axis (Roff et al., 2002). As an initial test of this prediction
Roff et al. (2002, 2003) compared two trade-off functions
where at is the intercept at generation t, RY is the direct among geographically widely separated populations of
response of trait Y, CRX is the correlated response of trait G. firmus characterized by different degrees of wing
X, b is the slope of the trade-off function, hX and hY are dimorphism. For females, Roff et al. (2002) used the
the square-roots of the heritabilities, i is the intensity of linear regression between ovary mass (fecundity) and
selection and rA is the genetic correlation between X and DLM mass and for males, Roff et al. (2003) used the
Y. If truncation selection acts simultaneously on traits X linear regression between call duration (probability of
and Y, the intercept will change to: attracting a mate: Crnokrak & Roff, 1995) and DLM mass.
at1 at RY bRX We compared three newly collected populations from
Florida, South Carolina and Bermuda, and a population
ir
at hY rPY hY rA hX that we had maintained in the laboratory for 19 years
1 rP (approximately 80 generations). Assuming that the
brPX hX hX rA hY ; 11 differences in proportion macropterous (assayed both in
where ir is approximately equal to i0[(1 + rP/4)(1 + rP)], the field and in common laboratory conditions) reflected
and i0 is the expected selection when rP 0 (Sheridan & differences in the local selection regimes, we predicted
Barker, 1974). More generally, for any type of directional among-population variation in the intercept of the linear
selection (G. de Jong, personal communication) the regression. In addition, because of the likelihood that
change in trait means is given by: evolution in the laboratory environment had caused
" # changes in the variances of the component traits and the
dX 2 @ ln w
dt rG;X rG;XY @X covariances between them, we predicted that the labor-
; 12
dY rG;XY r2G;Y @ ln w
@ Y
atory population might also differ with respect to the
dt slope of the linear regression.
where w is mean fitness and the subscript G denotes For both males and females, the three recently collec-
genetic components of variance and covariance. ted populations did differ in intercept but not slope of the
An important caveat must be added to the predicted linear regression, as predicted. Also as predicted, the
response when selection is applied to both traits: if the slope of the regression between ovary and DLM mass of
joint selection acts to increase acquisition (l2T ), say by the lab females differed significantly from the field
selecting on the sum Y + X (assuming this to be equiv- populations (Roff et al., 2002). However, the slope of
alent to selection on X1 + X2), then a change in covari- the regression between call duration and DLM mass in
ance may ensue (see eqn 1). Joint selection on both traits lab males did not (Roff et al., 2003). This difference
may affect not only mean acquisition but also mean between male and female traits was reflected in the
allocation and conceivably the two variances, which proportion macropterous: females from the lab popula-
could also result in a change in the covariance (James, tion had significantly reduced proportion macroptery,
1974, in fact, described changes in covariance caused by while males did not. This suggests that evolution within
selection on both traits but, as noted earlier, his equation the lab environment had altered the characteristics of the
for the covariance is flawed). The important point is that females but not the males.
in the Y-model the two component traits within an Tucic et al. (2005) compared two populations of the
individual are not mathematically independent traits and iris, Iris pumila, with respect to the linear regression
thus it is actually incorrect to regard them as separate between two measures of vegetative reproduction and
traits. The separate traits are in fact acquisition and somatic growth. One population was drawn from a
allocation, which may be much more difficult to meas- population growing on a dune in full sunlight and the
ure, particularly as resources will generally be allocated second from the understorey of a Pinus nigra stand where
light was considerably diminished. For both measures of models. The examples that we have given in this section
the trade-off there was a marginally significant difference illustrate the potential complexity of responses to selec-
(P 0.04) in the regression slope but no difference in tion on trade-offs and the need to understand the
intercept (P 0.1). Although the rhizomes were grown functional basis of the trade-off to understand the
under controlled conditions they were sampled directly response to selection. The evolutionary trajectory of the
from their natal populations, introducing the possibility trade-off can be particularly difficult to predict when
of previous environmental conditions affecting their both acquisition and allocation are allowed to vary in
allocation patterns. Tucic et al. (2005, p. 21) cite a response to selection, as illustrated by shifts in the
number of other studies which counter the specific correlation between longevity and stress resistance
prediction of Roff et al. (2002) that a shift in the slope of (Archer et al., 2003; Phelan et al., 2003; Prasad & Joshi,
the phenotypic linear regression should be less likely 2003; Prasad & Shakarad, 2004) or between larval
than in the intercept. However, in all the cited cases the survival and growth rate (Chippindale et al., 2003)
comparison was between regressions observed under observed in selection experiments in D. melanogaster.
different environmental conditions not different popula-
tions grown under the same environment. Such experi-
What maintains variation in the trade-off?
ments measure the phenotypic plasticity of the trade-off,
not the response to selection and the above theoretical Thus far, we have assumed that, if an underlying
development makes no statement about this circum- functional trade-off exists between traits, it will be
stance. Indeed, as discussed earlier and shown by the expressed in the pattern of statistical variation and
simulation analysis of Malausa et al. (2005), we would covariation of the two traits within populations. How-
expect phenotypic plasticity in allocation and hence that ever, many trade-offs may not be visible because no
the trade-off function, as typically measured by linear variation exists in either trait, i.e. the trade-off is fixed.
regression, could change with respect to both slope and Where trade-offs are visible, an important question is
intercept. why does such variation persist? To answer this ques-
The prediction derived from eqn 11 is predicated on tion we must first determine to what extent the variation
the (co)variances not changing. There is abundant that reveals the trade-off reflects phenotypic plasticity
evidence that genetic and phenotypic (co)variances do versus genetic variation among individuals. Even under a
change in some cases but not others (Roff, 2000; Jones laboratory setting, individuals do not experience exactly
et al., 2003; Cano et al., 2004) although whether this is the same environment and thus some, if not all, of the
due primarily to drift or selection is generally uncertain variation could be a result of phenotypic plasticity. It is
(Roff, 2000, 2004; Steppan et al., 2002; Roff & Mousseau, relatively easy to produce a model in which a trade-off is
2005). If changes in allele frequencies of pleiotropic expressed strictly as a phenotypically plastic response: for
genes affect both traits, then either selection or drift will example, in a beetle such as Stator limbatus in which all
produce proportional changes in variances and covar- larval resources come from a single seed (Fox et al.,
iances (Reeve, 2000; Roff, 2004) and hence the slope of 1997), fitness may well be maximized if females facul-
the trade-off will be preserved but the intercept will tatively increase the size of their eggs on small seeds,
change as selection acts on mean trait values. Similarly, if even though the increase in larval survival may be
as expected for trade-offs, there is strong structured somewhat offset by a trade-off between egg size and
pleiotropy the slope will be resistant to change (De Jong, number (Roff, 2002, pp. 433438). However, it is
1990). The same arguments apply not only to trade-offs certainly evident from the observation of trade-offs
but also to any bivariate relationships, such as the under highly controlled (common garden) conditions
positive covariation between body parts. that there is genetic variation in most trade-offs (e.g.
Ultimately, the stability of the trade-off relationships is Billerbeck et al., 2000; Donohue et al., 2000; Roff et al.,
an empirical question and we require more studies of 2002, 2003; Lee et al., 2003; Roff & Gelinas, 2003).
interpopulation variation in trade-offs, as well as con- Pedigree experiments that examine both the genetic basis
trolled experiments investigating responses to selection of the trade-off and possible genetic variation in plasticity
on the trade-off itself. Common garden comparisons of would be useful to assess the relative importance of these
different populations can provide valuable insights into two sources of variation in generating observed trade-off
the extent and pattern of natural variation in trade-off functions.
functions, but the results of such studies may be difficult A more difficult question to answer is why is there
to interpret because little is known about the patterns of genetic variation underlying the trade-off, whether or
selection in the natural environment. Artificial selection not plasticity is present? Why does not the population
on the trade-off function (or, more accurately, on the collapse to a single combination of traits? In some ways
traits comprising the trade-off) has the advantage that this is the same problem of accounting for variation in
the forces of selection are under strict control and often general, although there are particular features of trade-
responses can be predicted a priori, allowing tests of more offs that make this a separate problem. Four mechanisms
complex or counter-intuitive aspects to the conceptual that could preserve or at least reduce the rate of erosion
of variation resulting in an observed trade-off are (1) but a ridge aligned in the direction of the trade-off: in this
mutation-selection balance, (2) antagonistic pleiotropy, case, multiple combinations are equally fit. Variation
(3) correlational selection, and (4) spatio-temporal het- along the ridge is thus effectively neutral and the erosion
erogeneity. Of these four, correlational selection is of variation will be retarded. Selection favouring trait
particularly noteworthy because it may play a much combinations is called correlational selection and will
more central role in preserving variation in trade-offs generate a genetic correlation between the two traits
than variation in single traits or multiple traits that are under selection as a result of linkage disequilibrium
not functionally interconnected. (Sinervo & Svensson, 2002). Correlations generated by
linkage disequilibrium are relatively unstable, partic-
ularly in small populations where drift can produce wide
Mutation-selection balance
fluctuations in allele frequencies. Pleiotropic mutations
Empirical analyses using Daphnia pulex (Lynch et al., that produce combinations of the type favored by
1998) and D. melanogaster (Houle, 1998) lend support to selection should spread in the population replacing the
the hypothesis that much of the standing genetic vari- linkage disequilibrium genetic correlation with one based
ance in life history traits is because of mutational input on pleiotropy.
(see also Charlesworth & Hughes, 2000; Roff, 2005). The concepts of correlational selection and antagonistic
Whether this is sufficient to maintain the phenotypic and pleiotropy are related in the sense that both predict that
genetic covariation found in trade-offs has not been fitness will be maximized only at certain combinations of
determined, although theory shows that it could be traits. However, the two concepts differ in that pleiotropy
important (Houle, 1991). Evidence suggests that most refers to gene action, specifically, genes that affect more
mutations have deleterious effects on all components of than one phenotypic trait, while correlational selection
fitness and that these tend to be purged from natural refers to the fitness surface. Confusion arises because
populations (Houle et al., 1994, 1997; Estes et al., 2005). antagonistic pleiotropy is defined in terms of changes in
Mutations with antagonistically pleiotropic effects would fitness generated by the pleiotropic phenotypic effects: an
remain segregating in the populations for longer, thereby increase in fitness associated with the value of one trait is
generating variation in the trade-off. correlated with a decrease in fitness associated with the
value of the other trait. This is indeed a form of
correlational selection and it would be expected to
Antagonistic pleiotropy
generate an optimal trait combination (i.e. a single
The conditions for the maintenance of genetic variation fitness peak). However, correlational selection is broader
by antagonistic pleiotropy appear to be quite restrictive, than this, encompassing fitness ridges or even saddles,
requiring the presence of nonadditive genetic effects with regions in which parallel changes in both traits (i.e.
(reviewed in Roff, 1997), although the precise require- both increase or both decrease) have parallel rather than
ments have not yet been ascertained. Surveys of the antagonistic effects on fitness (Phillips & Arnold, 1989).
amount of dominance variance suggest that antagonistic Most relevant for this discussion, correlational selection
pleiotropy is unlikely to be an explanation for the can generate a suite of combinations that have equal
maintenance of genetic variance in morphological traits fitnesses rather than a single fitness peak, and it is this
but could account for that in life history traits (Roff, that may help to maintain variation in fitness trade-offs.
1997). However, Estes et al. (2005) argue that the overall Correlational selection is difficult to detect statistically
negative effects of pleiotropic mutations suggest that but is expected to be common (Schluter & Nychka, 1994;
antagonistic pleiotropy is unlikely. What is needed are Blows & Brooks, 2003). Examples of correlational
quantitative genetic studies that determine not only the selection on trade-offs include the interaction between
additive genetic components of trade-offs but also the color pattern and antipredator behavior in the garter
dominance components. Additionally, we need theoret- snake, Thamnophis ordinoides (Brodie, 1992) and the
ical studies to determine what levels of dominance pygmy grasshopper, Tetrix subulata (Forsman & Appelq-
variance would be sufficient to account for the observed vist, 1998), the trade-off between water-use efficiency
covariances. and leaf size in Cakile edentula (Dudley, 1996), the trade-
off between colour morph and immunocompetence in
the side-blotched lizard, Uta stansburiana (Svensson et al.,
Correlational selection
2002), and the trade-off between size and timing of
If the fitness surface for the two traits of a trade-off has a sexual maturation in the grasshopper Sphenarium purpu-
single peak then, ignoring drift and mutation, the rascens (Castillo & Nunez-Farfan, 1999). Correlational
evolutionary trajectory will take the population eventu- selection may slow the erosion of variation but will not,
ally to the peak provided that the genetic correlation is in the absence of other factors, maintain variation.
not )1. Stabilizing selection will maintain the population Variation will be preserved over the longest period if
at the fitness peak and will eventually erode variation. correlational selection is in the same direction as the
But suppose that the fitness surface contains not a peak major axis of the trade-off, and over time the genetic
(co)variance structure is expected to evolve such that the limit of applicability, and progress in resolving absolute
major axis is aligned with the direction of selection constraints will likely require studies of the mechanistic
(Jones et al., 2003, 2004; Blows et al., 2004; Estes et al., or functional basis of the suite of trade-offs hypothesized
2005) although we certainly do not have a sufficient to restrict evolutionary change.
number of studies yet, to say if this is generally true. In discussing the second topic, acquisitionallocation
models of resource-based trade-offs, we describe
important insights gained from this approach, but also
Spatio-temporal heterogeneity
suggest that a misunderstanding of the equation has led
The optimum combination of trait values will typically to improper predictions and tests of whether such
vary with environmental conditions: for example, in the trade-offs exist. We show how the Y-model can be
case of the seed beetle discussed above, seed size will vary expanded to include other types of trade-offs such as
among host plant species and if these show spatial and that between egg size and number, or that between
temporal variation then so also will the appropriate adult size and development time. The Y-model itself has
combination of egg size and egg number. If such been expanded to include more than two traits (the
variation can be accommodated by phenotypic plasticity hierarchical Y-model: De Laguerie et al., 1991; De Jong,
then genetic variation will not be preserved. The likeli- 1993; Worley et al., 2003; Bjorklund, 2004) and phen-
hood of preservation of genetic variation increases as the otypic plasticity (Malausa et al., 2005) but more theor-
predictability of the environment decreases, but it is etical and empirical research is needed on these
important to note that this requires either the joint expanded models.
effects of spatial and temporal variation or the interaction If variances and covariances are not changed under
of temporal variation with an overlapping age structure directional selection, the answer to the question posed as
(Roff, 2002). The importance of overlapping age struc- our third topic, how does directional selection affect
ture in preserving either phenotypic or genetic variation trade-offs, as expressed by the linear regression between
in trade-offs has been little explored, but would be a the two traits? is that the intercept but not the slope of
fruitful avenue for further research. the linear regression will change. Tests of this prediction
The observation that the genetic correlation underly- using stocks from geographic populations grown under
ing the trade-off can itself be environmentally sensitive common garden conditions were affirmative, but artifi-
(Sgro & Hoffmann, 2004), also begs more study and cial selection experiments would provide a better test. It
theoretical insight. It is not clear if this sensitivity could is important to distinguish the above evolutionary
play any role in preserving variation. This plasticity in the predictions from predictions concerning the purely
correlation no doubt reflects the functional behavior of phenotypic change in the regression expected when the
the underlying genes, some being switched on or off, or same genotypes are reared under different environmen-
up- or down-regulated. Whereas quantitative genetics tal conditions. In this case, there is no reason to expect
can describe these changes in a statistical framework, we that the linear regression will necessarily remain
lack information about the morphological, physiological unchanged in either its slope or intercept (Malausa et al.,
and behavioural changes that occur and the suite of 2005).
genes that are activated. In this regard, further develop- The final topic that we have considered is the
ment of genomics and its application to the present perplexing question of what maintains variation in the
question are crucial (Bochdanovits et al., 2003; Stearns & trade-off. We suggest that this question is best answered
Magwene, 2003; Bochdanovits & de Jong, 2004; Tonsor by distinguishing between phenotypic variation and
et al., 2005). genetic variation. The more difficult question to resolve
is why genetic variation is observed and this variation is,
of course, fundamental for the evolution of trade-offs.
Conclusions
We suggest that four phenomena are likely to be
We have discussed four main topics that we believe are important: mutation-selection balance, antagonistic plei-
important for continuing progress in the study of trade- otropy, correlational selection and spatio-temporal het-
offs. The first, the concept of constraint, is important erogeneity. All four may be important, but unfortunately
because it may lead to misconceptions about the limits of no data exist to determine whether one or several play a
evolutionary trajectories. We suggest that an explicit primary role.
distinction be made between the bias introduced by Given the central role that trade-offs play in evolu-
negative genetic correlations and the limitation in phase tionary theory and the evolution of life histories in
space dictated by an eigenvalue of zero. In general, the particular, it is perhaps surprising that we still know so
former is most likely to be the correct interpretation. little about the genetic architecture underlying trade-offs,
However, the existence of absolute constraints is of very the mechanistic basis of practically all trade-offs, or the
considerable importance, and when suggested by empir- evolution of trade-offs in either the short-term or long-
ical data, deserves more detailed study. Here, statistical term. On the other hand, the lack in these areas provides
analyses of variancecovariance matrices reach their a rich ground for further study and the integration of
many disciplines, including the emerging field of Cano, J.M., Laurila, A., Palo, J. & Merila, J. 2004. Population
genomics. differentiation in G matrix structure due to natural selection
in Rana temporaria. Evolution 58: 20132020.
Castillo, R.C.d. & Nunez-Farfan, J. 1999. Sexual selection on
Acknowledgements maturation time and body size in Sphenarium purpurascens
(Orthoptera: Pyrgomorphidae): correlated response to selec-
This work was supported by grant #DEB-0445140 from tion. Evolution 53: 209215.
the National Science Foundation. We are grateful to Charlesworth, B. 1990. Optimization models, quantitative gen-
Gerdien de Jong for her insightful comments on an etics, and mutation. Evolution 44: 520538.
earlier draft of this manuscript. The manuscript was also Charlesworth, B. & Hughes, K.A. 2000. The maintenance of
improved by the constructive comments of two anony- genetic variation in life-history traits. In: Evolutionary Genetics
mous reviewers. (Singh, R.S. & Krimbas, C.B. eds), pp. 369392. Cambridge
University Press, Cambridge.
Charlesworth, B. & Hughes, K.A. 1996. Age-specific inbreeding
References depression and components of genetic variance in relation to
Archer, M.A., Phelan, J.P., Beckman, K.A. & Rose, M.R. 2003. the evolution of senescence. Proc. Natl. Acad. Sci. U.S.A. 93:
Breakdown in correlations during laboratory evolution. II. 61406145.
Selection on stress resistance in Drosophila populations. Charnov, E.L. 1989. Phenotypic evolution under Fishers funda-
Evolution 57: 536543. mental theorem of natural selection. Heredity 62: 113116.
Barton, N.H. & Turelli, M. 1987. Adaptive landscapes, genetic Chippendale, A.K., Chu, T.J.F. & Rose, M.R. 1996. Complex
distance and the evolution of quantitative characters. Genet. trade-offs and the evolution of starvation resistance in
Res. 49: 157173. Drosophila melanogaster. Evolution 50: 753766.
Berven, K.A. 1987. The heritable basis of variation in larval Chippindale, A.K., Gibbs, A.G., Sheik, M., Yee, K.J., Djawdan,
developmental patterns within populations of the wood frog M., Bradley, T.J. & Rose, M.R. 1998. Resource acquisition and
(Rana sylvatica). Evolution 41: 10881097. the evolution of stress resistance in Drosophila melanogaster.
Biere, A. 1995. Genotypic and plastic variation in plant size: Evolution 52: 13421352.
effects on fecundity and allocation patterns in Lychnis flos- Chippindale, A.K., Ngo, A.L. & Rose, M.R. 2003. The devil in the
cuculi along a gradient of natural soil fertility. J. Ecol. 83: 629 details of life-history evolution: instability and reversal of
642. genetic correlations during selection on Drosophila develop-
Billerbeck, J.M., Schultz, E.T. & Conover, D.O. 2000. Adaptive ment. J. Genet. 82: 133145.
variation in energy acquisition and allocation among latitu- Christians, J.K. 2000. Trade-offs between egg size and number in
dinal populations of the Atlantic silverside. Oecologia 122: 210 waterfowl: an interspecific test of the van Noordwijk and de
219. Jong model. Funct. Ecol. 14: 497501.
Bjorklund, M. 2004. Constancy of the G matrix in ecological Crnokrak, P. & Roff, D.A. 1995. Fitness differences associated
time. Evolution 58: 11571164. with calling behaviour in the two wing morphs of male sand
Blows, M.W. & Brooks, R. 2003. Measuring nonlinear selection. crickets, Gryllus firmus. Anim. Behav. 50: 14751481.
Am. Nat. 162: 815820. Crnokrak, P. & Roff, D.A. 2002. Trade-offs to flight capability in
Blows, M.W. & Hoffmann, A.A. 2005. A reassessment of genetic Gryllus firmus: The influence of whole-organism respiration
limits to evolutionary change. Ecology 86: 13711384. rate on fitness. J. Evol. Biol. 15: 388398.
Blows, M.W., Chenoweth, S.F. & Hine, E. 2004. Orientation of De Jong, G. 1990. Quantitative genetics of reaction norms.
the genetic variance-covariance matrix and the fitness surface J. Evol. Biol. 3: 447468.
for multiple male sexually selected traits. Am. Nat. 163: 329 De Jong, G. 1993. Covariance between traits deriving from
340. successive allocations of a resource. Funct. Ecol. 7: 7583.
Bochdanovits, Z. & de Jong, G. 2003. Experimental evolution in De Jong, G. & van Noordwijk, A.J. 1992. Acquisition and
Drosophila melanogaster: Interaction of temperature and food allocation of resources: genetic (co)variances, selection, and
quality section regimes. Evolution 57: 18291836. life histories. Am. Nat. 139: 749770.
Bochdanovits, Z. & de Jong, G. 2004. Antagonistic pleiotropy for De Laguerie, P., Olivieri, I., Atlan, A. & Gouyon, P.-H. 1991.
life-history traits at the gene expression level. Proc. R. Soc. Lond. Analytic and simulation models predicting positive genetic
Ser. B Biol. Sci. 271: S75S78. correlations between traits linked by trade-offs. Evol. Ecol. 5:
Bochdanovits, Z., der Klis, H. & de Jong, G. 2003. Covariation of 361369.
larval gene expression and adult body size in natural Donohue, K., Pyle, E.H., Messiqua, D., Heschel, M.S. & Schmitt,
populations of Drosophila melanogaster. Mol. Biol. Evol. 20: J. 2000. Density dependence and population differentiation of
17601766. genetic architecture in Impatiens capensis in natural environ-
Bohren, B.B., Hill, W.G. & Robertson, A. 1966. Some observa- ments. Evolution 54: 19691981.
tions on asymmetrical correlated responses to selection. Genet. Dudley, S.A. 1996. Differing selection on plant physiological
Res. 7: 4457. traits in response to environmental water availability: a test of
Brodie, E.D.I. 1992. Correlational selection for color pattern and adaptive hypotheses. Evolution 50: 92102.
antipredator behavior in the garter snake, Thamnophis ordino- Dudycha, J.L. & Lynch, M. 2005. Conserved ontogeny and
ides. Evolution 46: 12841298. allometric scaling of resource acquisition and allocation in the
Brown, C.A. 2003. Offspring size-number trade-offs in scorpi- Daphniidae. Evolution 59: 565576.
ons: an empirical test of the van Noordwijk and de Jong Ernande, B., Boudry, P., Clobert, J. & Haure, J. 2004. Plasticity
model. Evolution 57: 21842190. in resource allocation based life history traits in the Pacific
oyster, Crassostrea gigas. I. Spatial variation in food abundance. Jones, A.G., Arnold, S.J. & Borger, R. 2003. Stability of the
J. Evol. Biol. 17: 342356. G-matrix in a population experiencing pleiotropic mutation,
Estes, S., Ajie, B.C., Lynch, M. & Phillips, P.C. 2005. Sponta- stabilizing selection, and genetic drift. Evolution 57: 1747
neous mutational correlations for life-history, morphological 1760.
and behavioral characters in Caenorhabditis elegans. Genetics Jones, A.G., Arnold, S.J. & Burger, R. 2004. Evolution and
170: 645653. stability of the G-matrix on a landscape with a moving
Falconer, D.S. 1989. Introduction to Quantitative Genetics. Long- optimum. Evolution 58: 16391654.
mans, New York. Jordan, M.A. & Snell, H.L. 2002. Life history trade-offs and
Foley, P.A. & Luckinbill, L.S. 2001. The effects of selection for phenotypic plasticity in the reproduction of Galapagos lava
larval behavior on adult life-history features in Drosophila lizards (Microlophus delanonis). Oecologia 130: 4452.
melanogaster. Evolution 55: 24932502. Kirkpatrick, M. & Lofsvold, D. 1992. Measuring selection and
Forsman, A. & Appelqvist, S. 1998. Visual predators impose constraint in the evolution of growth. Evolution 46: 954971.
correlational selection on prey color pattern and behavior. Kirkpatrick, M., Lofsvold, D. & Bulmer, M. 1990. Analysis of the
Behav. Ecol. 9: 409413. inheritance, selection and evolution of growth trajectories.
Fox, C.W., Thakar, M.S. & Mousseau, T.A. 1997. Egg size Genetics 124: 979993.
plasticity in a seed beetle: an adaptive maternal effect. Am. Nat. Lardner, B. & Loman, J. 2003. Growth or reproduction?
149: 149163. Resource allocation by female frogs Rana temporaria. Oecologia
Fry, J.D. 1996. The evolution of host specialization: are trade- 137: 541546.
offs overrated? Am. Nat. 148 (Suppl.): S84S107. Lee, C.E., Remfert, J.L. & Gelembiuk, G.W. 2003. Evolution of
Futuyma, D.J. 1998. Evolutionary Biology. Sinauer Associates, physiological tolerance and performance during freshwater
Inc., Sunderland, MA. invasions. Integr. Comp. Biol. 43: 439449.
Geber, M.A. & Griffen, L.R. 2003. Inheritance and natural Li, B., Shibuya, T., Yogo, Y., Hara, T. & Yokozawa, M. 2001.
selection on functional traits. Int. J. Plant Sci. 164: S21S42. Interclonal differences, plasticity and trade-offs of life history
Glazier, D.S. 1998. Does body storage act as a food availability traits of Cyperus esculentus in relation to water availability. Plant
cue for adaptive adjustment of egg size and number in Daphnia Spec. Biol. 16: 193207.
magna?. Freshw. Biol. 40: 8792. Loh, R. & Bitner-Mathe, B.C. 2005. Variability of wing size and
Glazier, D.S. 1999. Trade-offs between reproductive and somatic shape in three populations of a recent Brazilian invader,
(storage) investments in animals: a comparative test of the Zaprionus indianus (Diptera: Drosophilidae), from different
Van Noordwijk and De Jong model. Evol. Ecol. 13: 539555. habitats. Genetica 125: 271281.
Gustafsson, L., Nordling, D., Andersson, M.S., Sheldon, B.C. & Lynch, M., Latta, L., Hicks, J. & Giorgianni, M. 1998. Mutation,
Qvarnstrom, A. 1994. Infectious diseases, reproductive effort selection, and the maintenance of life-history variation in a
and the cost of reproduction in birds. Phil. Trans. R. Soc. Lond. B natural population. Evolution 52: 727733.
Biol. Sci. 346: 323331. Malausa, T., Guillemaud, T. & Lapchin, L. 2005. Combining
Hermida, M., Fernandez, C., Amaro, R. & San Miguel, E. 2002. genetic variation and phenotypic plasticity in tradeoff model-
Heritability and evolvability of meristic characters in a ling. Oikos 110: 330338.
natural population of Gasterosteus aculeatus. Can. J. Zool. 80: Messina, F.J. 1993. Heritabilty and evolvability of fitness
532541. components in Callosobruchus maculatus. Heredity 71: 623629.
Houle, D. 1991. Genetic covariance of fitness correlates: what Messina, F.J. & Slade, A.F. 1999. Expression of a life-history
genetic correlations are made of and why it matters. Evolution trade-off in a seed beetle depends on environmental context.
45: 630648. Physiol. Ent. 24: 358363.
Houle, D. 1992. Comparing evolvability and variability of Mezey, J.G. & Houle, D. 2005. The dimensionality of genetic
quantitative traits. Genetics 130: 195204. variation for wing shape in Drosophila melanogaster. Evolution
Houle, D. 1998. How should we explain variation in the genetic 59: 10271038.
variance of traits? Genetica 102103: 241253. Mitchell-Olds, T. 1996. Genetic constraints on life-history
Houle, D., Hughes, K.A., Hoffmaster, D.K., Ihara, J., Assimac- evolution: quantitative-trait loci influencing growth and
opoulos, S., Canada, D. & Charlesworth, B. 1994. The effects flowering in Arabidopsis thaliana. Evolution 50: 140145.
of spontaneous mutation on quantitative traits. I. Variances Mousseau, T.A. & Roff, D.A. 1987. Natural selection and the
and covariances of life history traits. Genetics 138: 773785. heritability of fitness components. Heredity 59: 181198.
Houle, D., Hughes, K.A., Assimacopoulos, S. & Charlesworth, B. Mueller, L.D., Folk, D.G., Nguyen, N., Nguyen, P., Lam, P., Rose,
1997. The effects of spontaneous mutation on quantitative M.R. & Bradley, T. 2005. Evolution of larval foraging
traits. II. Dominance of mutations with effects on life-history behaviour in Drosophila and its effects on growth and
traits. Genet. Res. 70: 2734. metabolic rates. Physiol. Ent. 30: 262269.
Houston, A.I. & McNamara, J.M. 1999. Models of Adaptive Ots, I. & Horak, P. 1996. Great tits, Parus major, trade health for
Behaviour. Cambridge University Press, Cambridge. reproduction. Proc. R. Soc. Lond. B Biol. Sci. 263: 14431447.
Imasheva, A.G., Moreteau, B. & David, J.R. 2000. Growth Partridge, L. & Sibley, R. 1991. Constraints on the evolution of
temperature and genetic variability of wing dimensions in life histories. Phil. Trans. R. Soc. Lond. B Biol. Sci. 332: 313.
Drosophila: opposite trends in two sibling species. Genet. Res. Pease, C.M. & Bull, J.J. 1988. A critique of methods for
Camb. 76: 237247. measuring life history trade-offs. J. Evol. Biol. 1: 293303.
James, J.W. 1974. Genetic covariances under the partition of Perrin, N. & Travis, J. 1992. On the use of constraints in
resources model. Appendix 1 in Sheridan and Barker (1974). evolutionary biology and some allergic reactions to them.
Aust. J. Biol. Sci. 27: 99101. Funct. Ecol. 6: 361363.
Phelan, J.P., Archer, M.A., Beckman, K.A., Chippindale, A.K., Sheridan, A.K. & Barker, J.S.F. 1974. Two-trait selection and the
Nusbaum, T.J. & Rose, M.R. 2003. Breakdown in correlations genetic correlation. I. Prediction of responses in single-trait
during laboratory evolution. I. Comparative analyses of and in two-trait selection. Aust. J. Biol. Sci. 27: 7588.
Drosophila populations. Evolution 57: 527535. Sinervo, B. & DeNardo, D.F. 1996. Costs of reproduction in the
Phillips, P.C. & Arnold, S.J. 1989. Visualizing multivariate wild: path analysis of natural selection and experimental tests
selection. Evolution 43: 12091222. of causation. Evolution 50: 12991313.
Prasad, N.G. & Joshi, A. 2003. What have two decades of Sinervo, B. & Svensson, E. 2002. Correlational selection and
laboratory life-history evolution studies on Drosophila melano- the evolution of genomic architecture. Heredity 89: 329
gaster taught us?. J. Genet. 82: 4576. 338.
Prasad, N.G. & Shakarad, M.N. 2004. Genetic correlations: Sinervo, B., Doughty, P., Huey, R.B. & Zamudio, K. 1992.
transient truths of adaptive evolution. J. Genet. 83: 36. Allometric engineering: a causal analysis of natural selection
Reeve, J.P. 2000. Predicting long-term response to selection. on offspring size. Science 258: 19271930.
Genet. Res. 75: 8394. Smith, D.E.C. & Davies, R.W. 1997. Changes in energy
Reznick, D. 1983. The structure of guppy life histories: the allocation by the predator Nephelopsis obscura exposed to
tradeoff between growth and reproduction. Ecology 64: 862 differences in prey availability. Can. J. Zool. 75: 606612.
873. Stearns, S.C. 1992. The Evolution of Life Histories. Oxford Univer-
Reznick, D. 1985. Costs of reproduction: an evaluation of the sity Press, New York.
empirical evidence. Oikos 44: 257267. Stearns, S.C. & Magwene, P. 2003. The naturalist in a world of
Reznick, D., Nunney, L. & Tessier, A. 2000. Big houses, big cars, genomics. Am. Nat. 161: 171180.
superfleas and the costs of reproduction. Trends Ecol. Evol. 15: Stephens, D.W. & Krebs, J.R. 1986. Foraging Theory. Princeton
421425. University Press, Princeton, NJ.
Riska, B. 1986. Some models for development, growth and Steppan, S.J., Phillips, P.C. & Houle, D. 2002. Comparative
morphometric correlation. Evolution 40: 13031311. quantitative genetics: evolution of the G matrix. Trends Ecol.
Roff, D.A. 1992. The Evolution of Life Histories: Theory and Analysis. Evol. 17: 320327.
Chapman and Hall, New York. Svensson, E.I., Sinervo, B. & Comendent, T. 2002. Mechanistic
Roff, D.A. 1997. Evolutionary Quantitative Genetics. Chapman and and experimental analysis of condition and reproduction in a
Hall, New York. polymorphic lizard. J. Evol. Biol. 15: 10341047.
Roff, D.A. 2000. Trade-offs between growth and reproduction: Tanaka, Y. 1996. How is life history variation generated from the
an analysis of the quantitative genetic evidence. J. Evol. Biol. genetic resource allocation? Res. Popul. Ecol. 38: 1117.
13: 434445. Tonsor, S.J., Alonso-Blanco, C. & Koornneef, M. 2005. Gene
Roff, D.A. 2002. Life History Evolution. Sinauer Associates, function beyond the single trait: natural variation, gene
Sunderland, MA. effects, and evolutionary ecology. in Arabidopsis thaliana. Plant
Roff, D.A. 2004. The evolution of genetic architecture. In: Cell Environ. 28: 220.
Phenotypic Integration (M. Pigliucci & K. Preston, eds), pp. 345 Tucic, B., Pemac, D. & Avramov, S. 2005. Testing the
365. Oxford University Press, Oxford. predictions of an evolutionary trade-off model using Iris
Roff, D.A. 2005. Variation and life history evolution. In: pumila plants from an open and a shaded habitat. Plant Sp.
Variation: a Central Concept in Biology (B. Hallgrimsson & B. K. Biol. 20: 1722.
Hall, eds), pp. 333355. Elsevier Academic Press, New York. Tuomi, J., Hakala, T. & Haukioja, E. 1983. Alternative concepts
Roff, D.A. & DeRose, M. 2001. The evolution of trade-offs: of reproductive effort, costs of reproduction, and selection in
effects of inbreeding on fecundity relationships in the cricket, life-history evolution. Am. Zool. 23: 2534.
Gryllus firmus. Evolution 55: 111121. Turelli, M. & Barton, N.H. 1994. Genetic and statistical analyses
Roff, D.A. & Gelinas, M.B. 2003. Phenotypic plasticity and the of strong selection on polygenic traits: what, me normal?
evolution of trade-offs: the quantitative genetics of resource Genetics 138: 913941.
allocation in the wing dimorphic cricket, Gryllus firmus. J. Evol. Van Noordwijk, A.J. & de Jong, G. 1986. Acquisition and
Biol. 16: 5563. allocation of resources: their influence on variation in life
Roff, D.A. & Mousseau, T.A. 2005. The evolution of the history tactics. Am. Nat. 128: 137142.
phenotypic covariance matrix: evidence for selection and drift Van Tienderen, P. & Antonovics, J. 1994. Constraints in
in Melanoplus. J. Evol. Biol. 18: 11041114. evolution: on the baby and the bath water. Funct. Ecol. 8:
Roff, D.A., Mostowy, S. & Fairbairn, D.J. 2002. The evolution of 139140.
trade-offs: testing predictions on response to selection and Worley, A.C., Houle, D. & Barrett, S.C.H. 2003. Consequences of
environmental variation. Evolution 56: 8495. hierarchical allocation for the evolution of life-history traits.
Roff, D.A., Crnokrak, P. & Fairbairn, D.J. 2003. The evolution of Am. Nat. 161: 153167.
trade-offs: geographic variation in call duration and flight ability Zuk, M. 1996. Trade-offs in parasitology, evolution and behav-
in the sand cricket, Gryllus firmus. J. Evol. Biol. 16: 744753. ior. Parasitol. Today 12: 4647.
Schluter, D. & Nychka, D. 1994. Exploring fitness surfaces. Am.
Nat. 143: 597616. Received 10 July 2006; revised 1 September 2006; accepted 1 September
Sgro, C.M. & Hoffmann, A.A. 2004. Genetic correlations, 2006
tradeoffs and environmental variation. Heredity 93: 241248.