AMER. ZOOL., 28:899-912 (1988)
Thermoregulation, Flight, and the Evolution of Wing Pattern i n
Pierid Butterflies: The Topography of Adaptive Landscapes
JOEL G. KINGSOLVER
Department of Zoology, NJ-15, University of Washington,
Seattle, Washington 98195
SYNOPSIS. This paper describes a case study of adaptation, constraint, and evolutionary
innovation in pierid butterflies. I develop a framework for discussing these issues that
focuses on the questions: What is the form of the adaptive landscape relating fitness to
phenotypic characters? How do such landscapes differ for evolutionarily related groups?
I examine the evolution of wing pigment patterns and thermoregulatory behavior for
butterflies in two subfamilies in the family Pieridae, with three principal results. First, I
show that thermoregulation can be an important component of fitness in pierids, and that
wing color and thermoregulatory behavior are important phenotypic characters determining thermoregulatory performance and the adaptive landscape. Second, I show how
limits on possible variation in wing color and behavior constrain evolution within one
subfamily of pierids, and how these constraints are set by the physical and biochemical
mechanisms of adaptation. Third, I show how evolutionary innovation may have resulted
from the addition of a new, behavioral dimension to the landscape, and how this addition
has altered the functional interrelations among various elements of the wing color pattern.
I suggest that comparative analyses of the form and determinants of the adaptive landscape
may be useful in identifying evolutionary innovations, and complement theoretical analyses
of evolutionary dynamics on such fitness surfaces.
INTRODUCTION
Adaptation and constraint
This paper presents a case study of adaptation, constraint, and evolutionary innovation in pierid butterflies. I begin by
developing a framework for examining this
case study that focuses attention on a more
general issue: The topography of adaptive
landscapes.
The adaptive landscape has been useful
in studying adaptation and constraint both
as a metaphor and as an analytic tool. In
Wright's (1931, 1969, 1977) original formulation, as part of his shifting balance
theory of evolution, the adaptive landscape
is the relation of mean population fitness
to population gene frequencies; evolution
is then represented as a trajectory on this
fitness surface in gene frequency space. In
recent quantitative genetic models of evolution {e.g., Lande, 1979; Lande and
Arnold, 1983) the landscape is afitnesssurface as functions of phenotypic character
values.
1
From the Symposium on Energetics and Animal
Behavior presented at the Annual Meeting of the
American Society of Zoologists, 27-30 December
1986, at Nashville, Tennessee.
The utility of the landscape concept lies
in its representation of adaptation and evolutionary change. For natural selection
without frequency dependence, evolution
proceeds "uphill" (towards increasing mean
population fitness) on the landscape surface for a variety of genetic systems (Fisher,
1930; Wright, 1931; Kingman, 1961;
Lande, 1979); and evolutionary equilibria
lie at or near peaks in the landscape. In
this view, adaptation is the state of being
on such adaptive peaks, or higher on the
slope than some reference {e.g., ancestral,
sister) group. In practice, most studies of
adaptation relate genotypes or phenotypic
characters to some measure of performance which is correlated to individual
fitness, and adaptations are phenotypic
characters that improve (or maximize) performance (Arnold, 1986).
Constraints on adaptive evolution are
manifested as limits on genetic and phenotypic variation and covariation, and
affect the population's location and movement on the fitness surface. First, at any
point in time the population occupies only
a restricted region in phenotypic space, and
thus experiences only a restricted region
of the adaptive surface. (I assume here that
the phenotypic space and its associated fit-
899
900
JOEL G. KINGSOLVER
ness surface exist independent of the existence of populations currently occupying
that space and surface: see Burger [1986]
for discussion.) As a result, the ranges of
genotypes and phenotypes in the population limit the available variation on which
selection can act. Second, in quantitative
genetic models phenotypic and genetic
covariances strongly influence the rates and
directions of evolution on the fitness surface (Lande, 1979, 1980; Lande and
Arnold, 1983; Burger, 1986). Along these
lines, several recent studies have modeled
constraints by partitioning genetic variance-covariance into components attributed to developmental (Cheverud, 1984)
and phylogenetic (Cheverud et al., 1985;
see also Felsenstein, 1985) factors. At the
same time, empirical studies have tried to
demonstrate how developmental mechanisms determine the relation of genetic to
phenotypic change, and to evaluate how
such mechanisms may constrain phyletic
evolution (Alberch et al., 1979; Alberch
and Alberch, 1981; Alberch and Gale,
1985; Hanken, 1983). Thus, the adaptive
landscape can provide a framework that
represents both adaptation and constraint.
Topography of the landscape
Metaphorically, the discussion thus far
might be considered the "statics and
dynamics" of the adaptive landscape: Given
a particular landscape, what factors determine evolutionary change along its surface? An alternative perspective, and the
subject of this paper, might be called the
"topography" of the landscape: What is
the form of the landscape? What features
of organisms and environments determine
the form of this surface? To address these
questions, we need first to decide what
topographic features are of evolutionary
importance, and second to locate such features in phenotypic (or gene frequency)
space.
The analyses of evolutionary dynamics
summarized above identify the important
topographies: There are at least four general features that are relevant to the origin
of evolutionary innovations. First, the rate
of evolution is directly related to the selection gradient, or equivalently the steepness
of the fitness surface (Fisher, 1930; Lande,
1979). Second, saddle points represent
potential areas for disruptive selection and
possible "branch" points between alternative adaptive peaks (Felsenstein, 1979;
Kirkpatrick, 1982; Slatkin, 1984; Lande,
1985; Newmans/al., 1985; Burger, 1986).
Third, the addition of new dimensions to
the landscape implies a range of phenotypic values for a character that was previously monomorphic (or absent). Fourth,
ridges and inversions of curvature in the
landscape represent regions where the
nature of functional interrelationship
between two or more characters is changing (Burger, 1986).
Dobzhansky (1937) and others have
mentioned the tendency of adaptive peaks
to cluster close together, but the problem
of locating important landscape features
has not been systematically addressed. At
least four general approaches come to
mind: Survey, comparative, experimental,
and a priori approaches. First, as suggested
by Lande and Arnold (1983), one can use
techniques from selection analysis to estimate the slope and curvature of the landscape {i.e., the selection gradient) for a series
of populations and species. Such a survey
would be extremely valuable in evaluating
patterns of phenotypic selection in nature,
but would yield primarily local estimates
of the landscape at the current location of
particular populations. Mitchell-Olds and
Shaw (submitted) discuss some of the difficulties in statistical inference and biological interpretation with this approach. Second, a comparative approach among
related taxa might be used to form a picture of the landscape from the fragments
currently occupied by those taxa (see Felsenstein [1985] for a recent discussion of
the comparative method and its methodological difficulties). Third, relevant experimental manipulation of phenotype can
be used in some systems to systematically
examine the relation of phenotype to fitness components. This approach may be
particularly useful ; n evaluating the landscape surface in regions of phenotypic space
not currently occupied by any populations.
Fourth, one can try to identify a priori characteristics of organisms and/or environments that give rise to important topographic features: That is, to identify
BEHAVIORAL INNOVATION IN PIERIDS
evolutionary innovations. Clearly, these
approaches are complementary, not
mutually exclusive, but the a priori
approach has not been systematically discussed, much less applied. The experimental approach has been frequently used to
relate phenotype to fitness, but has not been
systematically applied towards estimating
the fitness surface in currently unoccupied
phenotypic space.
A case study
In this paper I present a case study of
adaptation, constraint, and evolutionary
innovation from this perspective of the
topography of the adaptive landscape, using
primarily the comparative and experimental approaches. The study concerns the
evolution of thermoregulation and wing
color in pierid butterflies, and addresses
the question: How do the adaptive landscapes relating wing color to thermoregulatory performance differ for evolutionarily related groups of pierids? There are
three parts. First I show that thermoregulation can be an important component of
fitness in pierids, and that wing color and
thermoregulatory behavior are important
phenotypic characters determining thermoregulatory performance and the adaptive landscape. Second, I show how limits
on possible variation in wing color and
behavior constrain evolution within one
subfamily of pierids, and how these constraints are set by the physical and biochemical mechanisms of adaptation. Third,
I show how evolutionary innovation in
another pierid subfamily has resulted from
the addition of a new, behavioral dimension to the adaptive landscape, and how
this addition has altered the functional
interrelations among other phenotypic
characters. I suggest that such comparative
analyses of adaptive topographies may be
useful in identifying constraints and innovations in evolution.
THERMOREGULATION AS A
COMPONENT OF FITNESS
Most pierid butterflies require body temperatures of 28 to 40°C to take off and fly
(Watt, 1968; Kingsolver, 1985a). They
achieve these elevated body temperatures
by behavioral thermoregulation, princi-
901
pally by orientation to solar radiation
(review in Kingsolver, 19856). The relationships between weather, body temperature, and flight activity have been quantified for Colias (Leigh and Smith, 1959;
Watt, 1968; Roland, 1982; Kingsolver,
1983a, b), Pieris (Cappuccino and Kareiva,
1985; Kingsolver, 1985a; Ohsaki, 1986),
and Anthocharis (Courtney and Duggan,
1983). One result that has emerged from
these studies is that in many temperate
populations of these species, weather
strongly limits the amount of time available
for flight activity. For example, in montane
Colias species, the average available flight
activity time ranged from 3 to 10 hr/day,
depending on elevation (Kingsolver,
1983a); similar values have been reported
for Pieris in montane Colorado (Kingsolver, unpublished results) and in Connecticut
(Cappuccino and Kareiva, 1985), and
Anthocharis in Great Britain (Courtney and
Duggan, 1983). Many mark-recapture
studies (reviewed in Courtney, 1986) have
shown that the mean adult lifespan in most
pierids is on the order of 3-7 days. These
results suggest that available flight time may
be a limited resource in many temperate
pierid populations.
Because females must fly to find oviposition sites (and most pierids lay eggs singly
on plants), one possible consequence of
limitations on flight activity time is a reduction in realized fecundity of females. Three
lines of evidence now support this hypothesis: 1) behavior, 2) demography, 3) and
resource allocation.
1) Experiments in both field cages and
environmental chambers with C. eurytheme,
Anthocharis cardamines, and Pieris rapae show
that daily egg production is correlated with
air temperature and/or solar radiation
(Stern and Smith, 1960; Gossard and Jones,
1977). For Colias in Colorado, Kingsolver
(19836) estimated realized fecundity by
combining field data on available and realized flight times, activity budgets, longevity, and maximum oviposition rates; these
estimates suggested that Colias in high elevation populations can lay only 20-50% of
their full complement of eggs as a result
of limited flight time. 2) Comprehensive
population demographical studies have
been done in three species of temperate
902
JOEL G. KINGSOLVER
ulatory posture, lateral basking, to elevate
their body temperatures into the range
required for active flight (Watt, 1968;
Douglas and Grula, 1978). In this posture
(Fig. 1), the wings are closed over the dorsum, and the ventral hind wing (VHW) surface is oriented perpendicular to the solar
beam; radiation is absorbed by the basal
regions of the VHW, and the resulting heat
is transferred to the body, elevating the
body temperature. Because of the thinness
and low thermal conductivity of the wings,
only radiation absorbed in the basal portions of the VHW is effectively transferred
as heat to the body. When the body temperature exceeds 40-42°C, Colias exhibits
a heat-avoidance posture in which the wings
and body are oriented parallel to the solar
beam, reducing the radiative heat load.
Thus, the ventral wing surfaces are used
as radiation-absorbing devices.
One consequence of this behavior is that
wing coloration affects radiation absorption and thus the body temperature of the
butterfly. Wing coloration in Colias is
determined by a two-pigment system: a
black, melanic pigment and several yelloworange pteridine pigments (Watt, 1967).
The relative proportions of melanic- and
pteridine-pigmented wing scales determine the wing color in a particular wing
region. Because of the lateral basking posture and poorly conducting wings of Colias,
only the wing color on the basal VHW is
relevant to body temperature and thermoregulation; color in the other wing
regions is thermally irrelevant (Kingsolver
and Moffat, 1982). The aspect of wing color
that is directly relevant to radiation absorption is the solar absorptivity, defined as the
fraction of solar radiation striking a surface
which is absorbed by it.
The degree of melanization on the basal
VHW varies considerably biogeographically and seasonally in and among Colias
species. Watt's (1968, 1969) studies show
that much of this variation is adaptive:
THERMOREGULATION AND
greater melanization leads to higher body
WING COLOR IN SULFURS
temperatures for a given set of environ(SUBFAMILY COLIADINAE)
mental conditions, and heavy melanization
Thermoregulatory behavior and
is associated with populations in relatively
physical mechanisms
colder conditions. Furthermore, VHW
Coliadine butterflies studied to date {Co- melanization is influenced by photoperiod
lias and Nathalis) use a specific thermoreg- (and in some cases, temperature) during
pierids: C. alexandra (Hayes, 1981, 1984),
Anthocharis cardamines (Courtney and Duggan, 1983), and Leptidea sinapis (Warren,
1981). In all three cases, the major determinant of population size is the realized
fecundity of females in the previous generation. Courtney (1984) has recently summarized field data on realized fecundity in
Lepidoptera, showing that the mean realized fecundity was less than !/s of the maximum for all species that lay eggs singly.
In a general review of demographic studies
of temperate Lepidoptera, Dempster
(1983) showed that reductions in realized
fecundity was the single most important
factor determining population variation in
8 of 16 species studied. 3) Springer and
Boggs (1986) examined oocyte production
in two Colorado populations of Colias philodice eriphyle that differ significantly in their
available flight activity times (Kingsolver,
1983a), to test the hypothesis that reduced
flight time will select for reduced oocyte
production. By rearing females from the
two populations under identical conditions, they showed that the direction and
magnitude of the difference between populations in number of oocytes are closely
correlated with available flight activity time.
All of these results support the claim that
"reduced fecundity as a consequence of
poor weather seems to be a recurrent theme
in pierid biology" (Courtney, 1986, p. 76).
Since the principal function of thermoregulation in pierids is to maintain the body
temperatures needed for flight, this argues
strongly that thermoregulation is an
important component of fitness in temperate pierid butterflies, primarily via its effects
on flight activity time. Thus we shall use
flight activity time, subject to certain overheating constraints, as an indirect measure
of fitness, and consider the relation of relevant phenotypic characters to the form of
the flight time surface—i.e., the adaptive
landscape.
903
BEHAVIORAL INNOVATION IN PIERIDS
Wing Angle
Orientation Angle
' I'
B = 0°
7 = 90° or 270°
Lateral
8 = 90°
7 =0°
Dorsal
' i '
5°<6 < 90
r =o°
Reflectance
FIG. 1. Common basking postures of butterflies. A) Definitions of the wing angle (9) and body orientation
angle (7) for a basking butterfly. B) Lateral, dorsal, and reflectance basking postures. Modified from Kingsolver
(1985a).
904
JOEL G. KINGSOLVER
larval development in multivoltine species,
such that seasonal variation in melanization in a population is associated with seasonal climatic changes (Watt, 1969; Hoffman, 1974, 1978; Douglas and Grula,
1978). Clearly, VHW melanization in Colias represents an adaptive feature for thermoregulatory performance.
If thermoregulatory performance, via its
effects on available flight activity time, is a
component of fitness, we may ask: What is
the functional relationship of thermoregulatory characteristics like VHW color to
fitness—that is, what is the form of the
adaptive landscape? I have addressed this
question by developing and testing a series
of mechanistic models that relate weather
conditions at a site and thermoregulatory
characteristics of the butterfly to flight
activity time (Kingsolver, 1983a; Kingsolver and Watt, 1983, 1984). Four characteristics of the butterfly are needed: body size
(thoracic diameter); thermoregulatory
posture; solar absorptivity of the basal
VHW; and the thickness of "fur" (setae)
on the ventral pterothorax. One important
constraint in the model is that resulting
from overheating: when butterflies experience body temperatures of 40-42°C, even
when using the heat-avoidance posture,
survivorship and fecundity are significantly
reduced (Kingsolver and Watt, 1983). For
present purposes, then, we can use such
models to answer the following two questions: Given a set of meteorological conditions during the flight season at a particular site, and the thermoregulatory
characteristics of a coliadine butterfly, what
is the predicted mean daily available flight
activity time? What is the probability that
the butterfly will experience body temperatures in excess of 42°C, even if it uses the
heat-avoidance posture?
(and climatic) gradient in Colorado: Montrose, elevation h = 1.5 km, with a resident
population of C. philodice eriphyle; Skyland,
h = 2.8 km, with a separate population of
C. p. eriphyle; and Mesa Seco, h = 3.3-3.6
km, with C. meadii. For each site, predicted
flight time is given as a function of VHW
absorptivity and fur thickness; those characteristics that violate an overheating constraint (where body temperatures exceeding 42°C occur more than 1% of the time)
are also identified. Three points are worth
mentioning. 1) For all sites, predicted flight
time is a rapidly increasing function of
VHW absorptivity, and this slope is particularly high for Mesa Seco, the highest elevation. If we consider flight time as a fitness
component, this result implies that the
selection gradient (sensu Lande and Arnold,
1983) is rather steep. 2) For each site, there
is a combination of absorptivity and fur
that yields the maximum flight time without violating the overheating constraint
(defined as the optimum). The measured
characteristics for the resident Colias population at each (given on the figures) yields
a flight time quite similar to that for the
optimum characteristics, especially for the
higher elevation sites where flight time is
limited. 3) Even when we consider the optimal phenotype for each site, there are large
differences among the sites in available
flight time (which are not due to differences in cloudiness among sites, since the
meteorological data are for sunny days). In
fact, the principal determinant of these differences is significant differences in shortterm meteorological variation that affect
the probability of overheating (see Kingsolver and Watt [1983, 1984] for detailed
discussions). If the location of the 'optimal'
phenotype in each environment is considered an adaptive peak, it is clear that the
heights of these peaks for Colias are different in different environments.
Model simulations yield the predicted
adaptive landscape, where wing absorptivity and thoracic fur thickness are the phenotypic characters and flight time (subject
to the overheating constraint) is the component of fitness. Figure 2 gives landscapes
for three sites where Colias butterflies have
resident populations, along an elevational
Because changes in ventral wing melanism are the primary determinant of solar
absorptivity in coliadines, it is useful to consider the landscape in terms of basal and
medial/distal melanin (Fig. 6, top). This
emphasizes two important results: That for
any particular thermal environment there
is an intermediate degree of basal mela-
Microevolution of wing color
BEHAVIORAL INNOVATION IN PIERIDS
55
65
70
Solar Absorptivity (%)
10 - 1.5,
S. Montrose
0.
£•
6-
Skyland
M.5
/
4 -
>
1.0/
/
/
905
nism yielding the maximum thermoregulatory performance; and that medial/distal
melanism does not affect performance.
Constraints enter the landscape in two
ways that underscore the utility of a model
reflecting the mechanisms of adaptation.
The two features that characterize thermoregulation in coliadines are the pteridine/melanin pigment systems for wing
color determination, and the use of lateral
basking as a thermoregulatory mechanism.
The pigment systems define the range of
possible wing absorptivities to between yellow (solar absorptivity a = 0.4) and black
(a = 0.7); similarly the range of possible fur
thicknesses that have useful thermal effects
is from zero to about the size of the thoracic radius (Kingsolver and Moffat 1982;
Kingsolver, 1983a). These ranges, which
reflect the levels of variation in these characters for the genus Colias, explicitly define
the parameter sets considered in the model.
In fact, one can use these limits to identify
the range of meteorological conditions to
which "Co/tas-type" butterflies can adapt
(Fig. 3). The second constraint, implicit in
our model but equally important, is thermoregulatory posture. All coliadine butterflies studied to date use lateral basking,
and this feature is one of the principal
determinants of the functional relationship between VHW color and flight time.
One can imagine other thermoregulatory
mechanisms, not found in coliadines, that
would change the form of the adaptive
landscape.
2-
I draw two conclusions from this analysis. First, Colias butterflies are indeed
adapted for effective thermoregulation to
the particular environmental conditions
70
55
65
45
they experience, but because of constraints
Solar Absorptivity (%)
due to overheating the adaptive peaks differ in height in different thermal environFIG. 2. Predicted flight time (hr/day) as a function
of wing solar absorptivity (%) and thoracic fur thick- ments. Thus, the degree of local adaptaness (mm) for Colias butterflies at three sites in central tion that can be attained by "Coftas-type"
Colorado. Light lines plot values for different fur butterflies may be limited in some selective
0.0/
thicknesses from 0 to 1.5 mm as indicated. Heavy lines
mark boundaries beyond which the overheating constraint, i.e., body temperature exceeds 42°C more than
1 % of the time for a butterfly in heat-avoidance posture, is violated. Solid symbols with error bars are the
predicted means, and standard deviations, of cumulative flight time based on observed means and standard deviations of absorptivity and fur thickness for
resident male Colias at each site. (A) C. meadii at Mesa
Seco p , 3.3-3.6 km elevation. (B) C. philodice eriphyle
at Montrose (•, 1.5 km elevation), and C. p. eriphyle
at Skyland (A, 2.6-2.7 km elevation). See Kingsolver
(1983a). The optimal solution (O) at each site is indicated. Modified from Kingsolver (1983a).
906
JOEL G. KINGSOLVER
!50r
125
< 100
o
"S 75
Qcn
T3
E
50
MINIMUS
25
13
18
23
28
33
38
Air temperature (°C)
FIG. 3. Flight space diagrams, based on model simulations, indicating the ranges of wind speeds (cm/sec) and
air temperatures (°C) in which Colias butterflies are able to achieve the body temperatures necessary for active
flight. The area enclosed by the line is the flight space. For this plot, solar radiation load perpendicular to
the solar beam is 110 mW/cm 2 . Given are the flight spaces for two hypothetical Colias species, representing
the extremes possible for "Colias-lype" butterflies: C. maximus (solid line), with all-black wing bases and a thick
fur layer; and C. minimus (dashed line), with all-yellow wing bases and no thoracic fur. See text. Modified
from Kingsolver and Watt (1984).
environments. Second, the use of lateral
basking posture as a thermoregulatory
mechanism in coliadines determines the
relationship of wing color to fitness; but
the lack of variation in basking posture is
a potential constraint on adaptive evolution in coliadines.
influenced by photoperiod during the larval period (reviews in Shapiro, 1976, 1984).
There are, however, two important differences between pierines and coliadines
relevant to thermoregulation. First, the
background color for pierines is white on
the dorsal wing surfaces, and white or yellow (occasionally orange) on the ventral
THERMOREGULATION AND
surfaces. The yellow pigment in pierines is
WING COLOR IN WHITES
sepiapterin, one of the pterines found in
(SUBFAMILY PIERINAE)
coliadines, whereas the white pigment is
Thermoregulatory behavior and
primarily leucopterin (Watt, 1967). The
physical mechanisms
concentration of sepiapterin in Pieris is
Pierines (subfamily Pierinae), like colia- controlled by a single locus with multiple
dines, use behavioral orientation to the sun alleles (Watt and Bowden, 1966), and indifor behavioral thermoregulation, and wing viduals with yellow on the dorsal surfaces
color plays an important thermoregulatory occur in many populations in pierines. Secrole. As in the coliadines Colias and Nathalis, ond, nearly all pierines studied to date
the degree and pattern of wing melaniza- {Infraphulia is a possible exception; Shation in the pierines Pieris, Tatochila, Phulia, piro, 1985) use a thermoregulatory posInfraphulia, and Piercolias varies biogeo- ture, apparently unique to the group, called
graphically both within and between reflectance basking (Fig. 1). As the name
species. In addition, melanization varies implies, in reflectance basking the wings
seasonally in some pierine species and is are used as solar reflectors (Kingsolver,
907
BEHAVIORAL INNOVATION IN PIERIDS
1985a, b). The wings are held open at an
angle, with the dorsal surface of the body
oriented towards the solar beam; radiation
is reflected off the white dorsal wing surfaces onto the body, increasing body temperature. On the other hand, at high body
temperatures pierines use a heat-avoidance
posture identical to that found in coliadines.
Microevolution of wing color in pierines
Mathematical and physical models, comparative studies, and experimental manipulations of wing pattern (Kingsolver,
1985a, b, 1987) reveal several interesting
interactions between details of the basking
posture (particularly the wing angle, or how
far open the wings are) and the dorsal pattern of melanization (Fig. 4). First, the relationship of basking angle to body temperature depends on wing melanization
pattern: the greater the amount of melanization on the wing margins, the larger
the wing angle at which body temperature
is maximized. Thus, for example, butterflies in the subgenera Artogeia and Pontia,
which differ in the degree of marginal melanization, use significantly different basking wing angles. Furthermore, experimentally increasing the amount of black
pigmentation on the dorsal margins in
Artogeia produces a significant increase in
the wing angle used during basking.
Second, the effect of melanization on
body temperature depends on the wing
region in question (Fig. 4). At the bases of
the dorsal fore wings (DFW) and the anal
regions of the dorsal hind wings (DHW)
radiation absorbed by the wings can be
effectively transferred as heat to the body;
as a result, increased basal melanization
increases body temperature. In contrast,
increased melanization on the medial and
distal wing surfaces decreases the high solar
reflectivity (about 0.8 for white wing surfaces) that is needed for effective reflection
of radiation to the body; as a result, increased melanization in these wing regions
decreases body temperature. In fact, we
can show that these different effects of melanization translate into effects on flight
activity. For example, experimental
manipulations with Artogeia napi macdun-
35
NEUTRAL
I
I REFLECTION
•
ABSORPTION
FIG. 4. Diagram showing the contribution of different dorsal wing regions to increasing body temperature during reflectance basking as a function of wing
angle. The basal regions absorb (absorption, black
areas) radiation, and heat is transferred to the body.
The distal regions do not affect body temperature
(neutral, hatched areas); the medial regions reflect
(reflection, white areas) solar radiation from the wings
to the body. As wing angle decreases, the relative
proportion of the regions contributing to radiation
reflection increases, and the neutral distal regions
decrease. From Kingsolver (1987).
noughii, which has very little dorsal melanization, show that increasing basal black
pigment leads to significantly earlier initiation of flight in the morning, whereas
increasing marginal black pigment has the
opposite effect on flight initiation (Kingsolver, 1987).
The thermal consequences of wing melanization can be summarized using a
"functional map," which describes how
melanization in specific wing regions affects
body temperature (Fig. 5) (Kingsolver,
1987). Three general points emerge from
a comparison of such maps for coliadines
and pierines. First, for coliadines increased
melanization either increases or has no
effect on body temperature; for pierines,
increased melanization can increase,
decrease, or not affect body temperature
depending on the wing region. Thus the
function of wing melanization is partially
reversed in the two groups. Second, the
thermal effects and thermoregulatory
function of melanization on the medial and
distal dorsal wing surfaces in pierines
depend critically on details of the ther-
908
JOEL G. KINGSOLVER
EFFECTS OF MELANIN
DORSAL
VENTRAL
— or 0
MEDIAL/DISTAL MELANIN
Pierine
FIG. 5. Functional map relating wing melanization
in different wing regions to body temperature in Pieris
butterflies. The map is given in terms of the effect of
increased melanin in a wing region on body temperature: + , —, and 0 represent increases, decreases and
no effects in body temperature, respectively. From
Kingsolver (1987).
moregulatory posture, in particular wing
angle, and one cannot interpret the functional significance of wing pattern without
knowledge of such behaviors. Third, the
difference in thermoregulatory mechanism between coliadines and pierines means
that wing color in only a small portion of
the wings affects thermoregulation in coliadines, while the entire dorsal wing surfaces may have thermoregulatory effects in
pierines.
We do not yet have a model like that for
Colias that relates quantitatively thermoregulatory behavior and wing color to body
temperature and flight time, but we can
construct some important qualitative features of the adaptive landscape surface for
pierines (Fig. 6, bottom). First, reflectance
basking represents a new dimension: It is
a qualitatively different posture than lateral basking, and, unlike in laterally basking coliadines, there is a range of reflectance postures (in terms of wing angle)
found in pierines. Second, one can no
longer usefully speak of "wing melanization": Both basal and medial/distal wing
regions can affect thermoregulatory per-
MEDIAL/DISTAL MELANIN
FIG. 6. Proposed general form of the adaptive landscape surface for pierid butterflies in cool environments. The vertical axis is a measure of thermoregulatory performance (flight activity time and
overheating); the horizontal axes are degree of basal
melanin and medial/distal melanin. Above: Laterallybasking Coliadine butterflies, where ventral wing melanin is considered. Note the intermediate maximum
for basal melanin on the surface (see Fig. 2), and that
medial/distal melanin does not affect performance.
Below: Reflectance-basking Pierine butterflies, where
dorsal melanin is considered. Note the reversal of
curvature of the surface compared to lateral basking.
See text.
formance in pierines (Fig. 6, bottom), in
contrast to coliadines (Fig. 6, top). Furthermore, the functional interrelationships for melanization in different wing
regions changes qualitatively as one moves
from lateral to reflectance basking, and as
reflectance wing angle changes, generating an inversion of curvature in the adap-
BEHAVIORAL INNOVATION IN PIERIDS
tive landscape surface for pierines (Fig. 6,
bottom) relative to coliadines (Fig. 6, top).
909
two, less common, pierid subfamilies are
still unclear. Thus far, the reflectance basking posture is only known among pierines
BEHAVIORAL MECHANISM AND
in association with their characteristic white
EVOLUTIONARY INNOVATION
background, and likely is a derived charThere are two differences relevant to acter; lateral basking is apparently the rule
thermoregulation between coliadine and in the other pierid subfamilies.
pierine butterflies: Background pigment
To understand the evolution of therand basking posture. The genetic basis for moregulation in pierines, then, one must
the difference in pigmentation is rather account for both the novel posture and the
simple. The usual white background in characteristic background pigment of the
pierines results from the pteridine pigment group. But which came first, the pigment
leucopterin, but a yellow background due or the posture? I would argue that the pigto the pteridine sepiapterin occurs in many ment is the more likely first step, on three
pierines, and is controlled by a single locus, grounds. First, both white and yellow backmultiple allele system. The yellow and grounds are found in the pierines and in
orange background in coliadines results the coliadines, with a simple genetic basis,
from several pteridines, including which suggests a common ancestry. Secsepiapterin; but a white background, again ond, there is no known variation in basking
due to leucopterin, is found in the "alba" posture in coliadines: all species studied to
form of females of many coliadine species, date use lateral basking. In contrast, there
and is controlled by a single gene locus is considerable variation in posture in pier(Watt, 1973). Thus, the transition from ines, both within and among species. Not
yellow to white and its converse has a sim- only are there differences in wing angle
ple genetic basis in both groups. The sec- used during reflectance basking, but sevond difference is in basking posture—the eral species apparently use dorsal basking
position of the wings and body relative to (Shapiro, 1985; Kingsolver, unpublished).
the sun. The genetic basis for this behav- Most importantly, female Pieris sometimes
ioral difference is unknown, but in pierines also use lateral basking, generally when they
our experiments show that it is actually are not flying frequently (Kingsolver,
quite plastic for an individual butterfly.
unpublished; Wiernasz, unpublished); and
The result of these two differences is that male and female Tatochila and Infraphulia
the functional significance and nature of sometimes use lateral basking during initial
selection on wing melanization pattern is warm-up after emerging from roosting sites
qualitatively different in these two groups: (D. Wiernasz, personal communication).
For some wing regions the direction of Thus, reflectance basking may not be the
potential selection is actually reversed. only posture used among pierines. Third,
Clearly behavioral posture is an important the posture is functionally useless for therelement in understanding wing color evo- moregulation without white wings, but the
lution in this butterfly family. In terms of converse is not true. "Alba" (white) Colias
the adaptive landscape for pierids, the females are just as effective in thermoreguchanges in background color and basking lating using lateral basking as wild-type
posture have two effects: the addition of a females (Watt, 1973). The reason is that
new behavioral dimension, and an inver- in cold environments it is the amount of
sion of curvature of the fitness surface with melanized wing scales on the VHW that is
respect to melanization—two key topo- important; here the difference between
graphic features for evolutionary innova- white vs. yellow scales is trivial. In very
warm environments where overheating is
tion (see Introduction).
a potential problem, white could in fact be
Cladistic analyses based on other mor- superior to yellow for thermoregulation.
phological characters (Ackery, 1984) indi- A glance at the flight spaces for Colias clarcate that the Pierinae and the Coliadinae ifies these points (Fig. 3). In contrast, mathare each a monophyletic group. The sister ematical and physical models show that
group relations among these and the other
910
JOEL G. KINGSOLVER
reflectance basking with yellow wings would
only be 25% as effective in increasing body
temperature as with white wings (Kingsolver, 1985a). Thus, it is unlikely that this
thermoregulatory posture would have been
established in pierines (or their ancestors)
before the white pigmentation.
Note that white pigmentation need not
have evolved as an adaptation for thermoregulation, because white pigmentation
may have served other functions in pierines and their immediate ancestors. For
example, several studies show that some
Pieris species are relatively unpalatable to
bird predators, resulting from sequestering of compounds from their cruciferous
host plants (Jones, 1932; Marsh and Rothschild, 1974; Kingsolver, in preparation),
and it has been suggested that white represents a type of aposematic coloration in
pierines (Rothschild, 1981). In Colias, and
presumably in Pieris, the white-yellow polymorphism in females affects the nitrogen
balance such that white females mature
more oocytes than yellow females (Watt,
1973). Which if any of these was the original function of white pigmentation is
unknown.
If this scenario is correct, the change in
posture from lateral to reflectance basking
represents a major event in the evolution
of wing patterns in pierids. We are unaccustomed to thinking that such "small"
behavioral changes could represent important evolutionary change, particularly at
the subfamily level; behavioral characters
are often considered relatively (compared
to morphological characters) plastic phenotypic characters. This points to one of
the general lessons of this study: There is
no simple relationship between the degree
of change in a phenotypic character and
its effects on functional performance, on
which selection can act. In terms of the
adaptive landscape, this simply means that
the fitness surface may be steep and/or
bumpy in particular directions and particular locations.
I have presented this study of pierid evolution to show how, using a combination
of experiments, observations, and models
with a comparative analysis, one might try
to analyse the topography of an adaptive
landscape, and to use those analyses to
locate topographic features associated with
evolutionary innovations. What cues might
one use to identify such features a priori?
The present study might provide one possible cue: It is the change in physical mechanism, from radiation absorption to radiation reflectance, that drives the important
features of the pierid landscape. In his
studies of Actinopterygian fish, Lauder
(1982) suggested that the existence of two
different biomechanical pathways for jaw
opening enabled the structural and functional diversification of this group.
Whether these or other cues can be used
to locate important topographic features
remains to be seen.
ACKNOWLEDGMENTS
I thank John Gittleman and Steve
Thompson for the opportunity to participate in this symposium. Thanks to Steve
Arnold, Tom Daniel, Sharon Emerson,
John Gittleman, Ray Huey, Ruth Shaw,
Steve Thompson, and Diane Wiernasz for
improving earlier drafts of this manuscript. Research was supported by NSF
grants DEB 8320373 and 8600485 to the
author.
REFERENCES
Ackery, P. R. 1984. Systematic and faunistic studies
on butterflies. In D. Vane-Wright and P. Ackery
(eds.), The biology of butterflies, pp. 9 - 2 1 . Academic
Press, London.
Alberch, P. and J. Alberch. 1981. Heterochronic
mechanisms of morphological diversification and
evolutionary change in the neotropical salamander, Bolitoglossa occidentalis.J. Morphol. 167:249—
264.
Alberch, P. and E. Gale. 1985. A developmental
analysis of an evolutionary trend: Digital reduction in amphibians. Evolution 39:8—23.
Alberch, P., S. J. Gould, G. F. Oster, and D. B. Wake.
1979. Size and shape in ontogeny and phylogeny.
Paleobiology 5:296-317.
Arnold, S.J. 1986. Laboratory and field approaches
to the study of adaptation. In M. E. Feder and
G. V. Lauder (eds.), Predator-prey relationships, pp.
157-179. University of Chicago Press, Chicago.
Burger, R. 1986. Constraints for the evolution of
functionally coupled characters: A nonlinear
analysis of a phenotypic model. Evolution 40:182193.
Cappuccino, N. and P. Kareiva. 1985. Coping with
a capricious environment: A population study of
the rare woodland butterfly Pieris virginiensis.
Ecology 66:152-161.
BEHAVIORAL INNOVATION IN PIERIDS
Cheverud, J. M. 1984. Quantitative genetics and
developmental constraints on evolution by selection. J. Theoret. Biol. 110:155-171.
Cheverud, J. M., M. M. Dow, and W. Leutenegger.
1985. The quantitative assessment of phylogenetic constraints in comparative analyses: Sexual
dimorphism in body weight among primates.
Evolution 39:1335-1351.
Courtney, S. 1984. The evolution of egg clustering
by butterflies and other insects. Am. Nat. 123:
276-281.
Courtney, S. P. 1986. The ecology of pierid butterflies: Dynamics and interactions. Adv. Ecol. Res.
15:51-131.
Courtney, S. P. and A. E. Duggans. 1983. The population biology of the Orange Tip butterfly
Anlhocharis cardamines in Britain. Ecol. Entomol.
8:271-281.
Dempster, J. P. 1983. The natural control of populations of butterflies and moths. Biol. Rev. 58:
461-481.
911
Kingsolver, J. G. 1983i. Ecological significance of
flight activity in Colias butterflies: Implications
for reproductive strategy and population structure. Ecology 64:546-551.
Kingsolver, J. G. 1985a. Thermal ecology of Pieris
butterflies: A new mechanism of behavioral thermoregulation. Oecologia 66:540-545.
Kingsolver, J. G. 19856. Thermoregulatory significance of wing melanization in Pieris butterflies:
Physics, posture, pattern. Oecologia 66:546-553.
Kingsolver, J. G. 1985c. Butterfly thermoregulation:
Organismic mechanisms and population consequences. J. Res. Lepid. 24:1-20.
Kingsolver, J. G. 1987. Evolution and coadaptation
of thermoregulatory behavior and pigmentation
pattern in pierid butterflies. Evolution 41:472490.
Kingsolver, J. G. and R. K. Moffat. 1982. Thermoregulation and the determinants of heat transfer
in Colias butterflies. Oecologia 53:27-33.
Kingsolver, J. G. and W. B. Watt. 1983. ThermoDobzhansky, T. 1937. Genetics and the origin of species.
regulatory strategies in Colias butterflies: TherColumbia Univ. Press, New York.
mal stress and the limits to adaptation in temporally varying environments. Am. Nat. 121:32Douglas, M. M. and J. Grula. 1978. Thermoregu55.
latory adaptations allowing ecological range
expansion by the pierid butterfly, Nalhalis Me.
Kingsolver, J. G. and W. B. Watt. 1984. Mechanistic
Evolution 32:776-783.
constraints and optimality models: Thermoregulatory strategies in Colias butterflies. Ecology
Felsenstein, J. 1979. Excursions along the interface
65:1835-1839.
between disruptive and stabilizing selection.
Genetics 93:773-795.
Kirkpatrick, M. 1982. Quantum evolution and puncFelsenstein, J. 1985. Phylogenies and the comparatuated equilibria in continuous genetic charactive method. Amer. Nat. 125:1-15.
ters. Am. Nat. 119:833-848.
Fisher, R. A. 1930. The genelical theory of natural selec- Lande, R. 1979. Quantitative genetic analysis of
tion. Clarendon Press, Oxford.
multivariate evolution, applied to brain : body size
Gossard, T. and R.Jones. 1977. The effects of age
allometry. Evolution 33:402-416.
and weather on egg-laying in Pieris rapae L. J.
Lande, R. 1980. Sexual dimorphism, sexual selecAppl. Ecology 14:65-71.
tion, and adaptation in polygenic characters. EvoHanken.J. 1983. Miniaturization and its effects on
lution 34:292-307.
cranial morphology in plethodontid salamanLande, R. 1985. Expected time for random genetic
ders, genus Thorius. II. The fate of the brain and
drift of a population between stable phenotypic
sense organs and their role in skull morphogenstates. Proc. Natl. Acad. Sci. U.S.A. 82:641-642.
esis and evolution. J. Morphol. 177:255-268.
Lande, R. and S.J. Arnold. 1983. The measurement
of selection on correlated characters. Evolution
Hayes, J. L. 1981. The population ecology of a nat37:1210-1226.
ural population of the pierid butterfly Colias alexandra. Oecologia 49:188-200.
Lauder, G. V. 1982. Patterns of evolution in the
feeding mechanism of Actinopterygian fishes.
Hayes, J. L. 1984. Colias alexandra: A model for the
Amer. Zool. 22:275-285.
study of natural populations of butterflies. J. Res.
Lepid. 23:113-124.
Leigh, T. and R. F. Smith. 1959. Flight activity of
Coliasphilodice eurytheme in response to its physical
Hoffman, R. 1974. Environmental control of seaenvironment. Hilgardia 28:569-624.
sonal variation in the butterfly Colias eurytheme:
Effects of photoperiod and temperature on pterLiem, K. 1973. Evolutionary strategies and morphoidine pigmentation. J. Insect Physiol. 20:1913logical innovations: Cichlid pharyngeal jaws. Syst.
1924.
Zool. 22:425-441.
Hoffman, R.J. 1978. Environmental uncertainty and Marsh, N. and M. Rothschild. 1974. Aposematic and
cryptic Lepidoptera tested on the mouse. J. Zool.
evolution of physiological adaptation in Colias
(London) 174:89-122.
butterflies. Amer. Nat. 112:999-1015.
Jones, F. M. 1932. Insect colouration and relative Newman, C. M.,J. Cohen, and C. Kipnis. 1985. NeoDarwinian evolution implies punctuated equilibacceptability of insects to birds. Trans. R. Ent.
ria. Nature 315:400-402.
Soc. London 80:345-385.
Kingman.J. F. C. 1961. A mathematical problem in Ohsaki, N. 1986. Body temperatures and behavioral
population genetics. Proc. Camb. Phil. Soc. 57:
thermoregulation strategies of three Pieris but574-582.
terflies in relation to solar radiation. J. Ethol. 4:
1-9.
Kingsolver, J. G. 1983a. Thermoregulation and flight
in Colias butterflies: Elevational patterns and
Roland, J. 1982. Melanism and diel activity of alpine
Colias. Oecologia 53:214-221.
mechanistic limitations. Ecology 64:534-545.
912
JOEL G. KINGSOLVER
terfly Colias eurytheme. J. Biol. Chem. 242:565Rothschild, M. 1981. Mimicry, butterflies, and plants.
572.
Symb. Bot. Upsal. 22:82-99.
Shapiro, A. M. 1976. Seasonal polyphenism. Evol. Watt, W. B. 1968. Adaptive significance of pigment
polymorphisms in Colias butterflies. I. Variation
Biol. 9:259-333.
in melanin pigment in relation to thermoreguShapiro, A. M. 1984. Experimental studies on the
lation. Evolution 22:437-458.
evolution of seasonal polyphenism. In R. VaneWright and P. Ackery (eds.), The biology of butter- Watt, W. B. 1969. Adaptive significance of pigment
flies, pp. 297-307. Academic Press, New York.
polymorphisms in Colias butterflies. II. Thermoregulation and photoperiodically controlled
Shapiro, A. M. 1985. Behavioral and ecological
melanin variation in Colias eurytheme. Proc. Natl.
observations of Peruvian high-Andean pierid
Acad. Sci. U.S.A. 63:767-774.
butterflies. J. Lepid. Soc. 39:174-177.
Slatkin, M. 1984. Ecological causes of sexual dimor- Watt, W. B. 1973. Adaptive significance of pigment
polymorphisms in Colias butterflies. III. Progress
phism. Evolution 38:622-630.
in the study of the "alba" variant. Evolution 27:
Springer, P. and C. L. Boggs. 1986. Altitude effects
537-548.
on resource allocation to reproduction in Colias
philodice eriphyle. Amer. Nat. 127:252-256.
Watt, W. B. and S. R. Bowden. 1966 Chemical phenotypes of pteridine colour forms in Pieris butStern, V. and R. F. Smith. 1960. Factors affecting
terflies. Nature 210:304-306.
egg production and oviposition in populations of
Colias philodice eurytheme. Hilgardia 29:411-454. Wright, S. 1931. Evolution in Mendelian populations. Genetics 16:97—159.
Warren, M. S. 1981. The ecology of the Wood White
butterfly, Leplidea sinapis. Ph.D. thesis, Univer- Wright, S. 1969. Evolution and the genetics of populations. Vol. 2. The theory of gene frequencies. Univ.
sity of Cambridge, England.
of Chicago Press, Chicago.
Watt, W. B. 1964. Pteridine components of wing
pigmentation in the butterfly Colias eurytheme. Wright, S. 1977. Evolution and the genetics of populations. Vol. 3. Experimental results and evolutionary
Nature 201:1326-1327.
deductions. Univ. of Chicago Press, Chicago.
Watt, W. B. 1967. Pteridine biosynthesis in the but-