Swift laboratory thermal evolution of wing shape (but not size) in Drosophila subobscura and its relationship with chromosomal inversion polymorphism.

Abstract

Latitudinal clinal variation in wing size and shape has evolved in North American populations of Drosophila subobscura within about 20 years since colonization. While the size cline is consistent to that found in original European populations (and globally in other Drosophila species), different parts of the wing have evolved on the two continents. This clearly suggests that 'chance and necessity' are simultaneously playing their roles in the process of adaptation. We report here rapid and consistent thermal evolution of wing shape (but not size) that apparently is at odds with that suggestion. Three replicated populations of D. subobscura derived from an outbred stock at Puerto Montt (Chile) were kept at each of three temperatures (13, 18 and 22 degrees C) for 1 year and have diverged for 27 generations at most. We used the methods of geometric morphometrics to study wing shape variation in both females and males from the thermal stocks, and rates of genetic divergence for wing shape were found to be as fast or even faster than those previously estimated for wing size on a continental scale. These shape changes did not follow a neat linear trend with temperature, and are associated with localized shifts of particular landmarks with some differences between sexes. Wing shape variables were found to differ in response to male genetic constitution for polymorphic chromosomal inversions, which strongly suggests that changes in gene arrangement frequencies as a response to temperature underlie the correlated changes in wing shape because of gene-inversion linkage disequilibria. In fact, we also suggest that the shape cline in North America likely predated the size cline and is consistent with the quite different evolutionary rates between inversion and size clines. These findings cast strong doubts on the supposed 'unpredictability' of the geographical cline for wing traits in D. subobscura North American colonizing populations.

Full text

Page 1

Swift laboratory thermal evolution of wing shape (but not size)
in Drosophila subobscura and its relationship with chromosomal
inversion polymorphism
M. SANTOS,* P.F. IRIARTE,* W. CE´SPEDES,* J. BALANYA`,? A. FONTDEVILA* & L. SERRA?
*Grup de Biologia Evolutiva (GBE), Departament de Gene `tica i de Microbiologia, Universitat Auto `noma de Barcelona, Barcelona, Spain
?Grup de Biologia Evolutiva (GBE), Departament de Gene `tica, Facultat de Biologia, Universitat de Barcelona, Barcelona, Spain
Introduction
Patterns of morphological variation mostly involving
size-related dimensions across latitudinal/altitudinal gra-
dients are often interpreted in relation to climatic
conditions, mainly temperature. In endotherms this is
exemplified by Bergmann’s rule: ‘the smaller-sized geo-
graphical races of a species are found in the warmer parts
of the range, the larger-sized races in the cooler districts’
(Mayr, 1942). Geographical clines in body size, with
genetically larger individuals derived from higher lati-
tudes, have also been documented in a number of
ectothermic animals, particularly insects from the genus
Drosophila (e.g. Stalker & Carson, 1947; Prevosti, 1955;
David et al., 1977; Coyne & Beecham, 1987; Pegueroles
et al., 1995; James et al., 1997; van’t Land et al., 1999;
Huey et al., 2000). Is there a Bergmann’s rule in
Correspondence: Mauro Santos, Departament de Gene `tica i de Microbio-
logia, Facultat de Cie `ncies, Edifici Cn, Universitat Auto `noma de Barce-
lona, 08193 Bellaterra, Barcelona, Spain.
Tel.: +34 93 581 2725; fax: +34 93 581 2387;
e-mail: mauro.santos@uab.es
J. EVOL. BIOL. 17 (2004) 841–855 ª 2004 BLACKWELL PUBLISHING LTD
841
Keywords:
clines;
Drosophila subobscura;
evolutionary rates;
geometric morphometrics;
inversion polymorphism;
Procrustes analysis;
thermal evolution;
wing shape;
wing size.
Abstract
Latitudinal clinal variation in wing size and shape has evolved in North
American populations of Drosophila subobscura within about 20 years since
colonization. While the size cline is consistent to that found in original
European populations (and globally in other Drosophila species), different parts
of the wing have evolved on the two continents. This clearly suggests that
‘chance and necessity’ are simultaneously playing their roles in the process of
adaptation. We report here rapid and consistent thermal evolution of wing
shape (but not size) that apparently is at odds with that suggestion. Three
replicated populations of D. subobscura derived from an outbred stock at Puerto
Montt (Chile) were kept at each of three temperatures (13, 18 and 22 ?C) for
1 year and have diverged for 27 generations at most. We used the methods of
geometric morphometrics to study wing shape variation in both females and
males from the thermal stocks, and rates of genetic divergence for wing shape
were found to be as fast or even faster than those previously estimated for
wing size on a continental scale. These shape changes did not follow a neat
linear trend with temperature, and are associated with localized shifts of
particular landmarks with some differences between sexes. Wing shape
variables were found to differ in response to male genetic constitution for
polymorphic chromosomal inversions, which strongly suggests that changes in
gene arrangement frequencies as a response to temperature underlie the
correlated changes in wing shape because of gene-inversion linkage disequil-
ibria. In fact, we also suggest that the shape cline in North America likely
predated the size cline and is consistent with the quite different evolutionary
rates between inversion and size clines. These findings cast strong doubts on
the supposed ‘unpredictability’ of the geographical cline for wing traits in
D. subobscura North American colonizing populations.
doi:10.1111/j.1420-9101.2004.00721.x

Page 2

ectotherms (cf. Mousseau, 1997; Partridge & Coyne,
1997; Van Voorhies, 1997)? Fitness costs related to the
effects of surface/volume ratio on heat loss are assumed
to underlie the rule, but it is obvious that size clines in
ectotherms warrant a different explanation because small
insects adopt ambient temperature almost instantane-
ously (Stevenson, 1985).
Although temperature is not the only factor that varies
with latitude, laboratory studies carried out with a
number of Drosophila species have repeatedly observed
thermal selection on body dimensions that goes in the
predicted direction according to the size clines (Ander-
son, 1966; Powell, 1974; Cavicchi et al., 1985, 1989;
Partridge et al., 1994a). There are, however, some
potentially important shortcomings with these experi-
ments. Thus, no replicated populations were kept in
Powell’s or Cavicchi’s et al. studies and, hence, the
‘random-walk’ hypothesis (which claims that evolution-
ary rates generally do not exist) cannot be discarded (see
Bookstein, 1991, pp. 393–398). However, Anderson’s
and Partridge’s et al. thermal stocks were replicated but
flies were maintained in population cages by regularly
introducing bottles with fresh food and removing them
after several weeks, a routine that does not allow control
of larval densities. Crowding conditions differ between
population cages maintained at different temperatures,
with higher larval densities in warmer environments
(L. Partridge, pers. comm., 1998). There is conflicting
evidence about the importance of larval density on the
evolutionary responsesof
(cf. Roper et al., 1996; Santos et al., 1997); however, it
is well established that harmful waste products accumu-
late in crowded cultures (Borash et al., 1998, 2000) and
many of the adaptations to different levels of larval
crowding (Joshi & Mueller, 1988, 1993; Guo et al., 1991;
Mueller et al., 1993; Borash et al., 1998) involve changes
in larval behaviour and physiology that may impinge on
other phases of the life cycle (see e.g. Joshi & Mueller,
1996; Santos, 1996; Santos et al., 1997; Houle & Rowe,
2003). Therefore, although the previous studies – as well
as those that have also reported evolutionary responses
in other fitness-related traits in the thermal stocks (Huey
et al., 1991; Partridge et al., 1994b; James & Partridge,
1995; Azevedo et al., 1996) – do demonstrate rapid
adaptations, it is by no means obvious that temperature
is the main factor or that body size is the target of
selection (see Bochdanovits & De Jong, 2003).
The historically Paleartic species D. subobscura provides
a suitable model system to study the dynamics of clinal
variation and the biological effects of the current
greenhouse-induced increase in world temperatures
(Houghton et al., 2001) for a number of reasons. First,
there is an extensive amount of information on the
geographical distribution of chromosomal arrangements
in this particularly inversion-rich species, with the
so-called standard arrangements in the five (out of
sixconsidering thedot
body sizein
Drosophila
chromosome)acrocentric
chromosomes increasing in frequency with latitude in
European populations (Krimbas & Loukas, 1980; Meno-
zzi & Krimbas, 1992). Most important, similar clines
quickly developed (within about 7 years or ?35 gener-
ations) after the double colonization of South and North
America by the species (Prevosti et al., 1985, 1988),
strongly supporting the hypothesis that environmental
latitudinal gradients are responsible for the clines. Fur-
ther support comes from several independent observa-
tions showing within-population long-term directional
trends in the inversion polymorphism that correlated
with expectations from the latitudinal clines (Orengo &
Prevosti, 1996; Sole ´ et al., 2002), although a putative
critical role played by a shift in temperatures could only
be ascribed to those trends tracked for the gene arrange-
ments of chromosome O for which relatively long time-
series matching both the inversion frequencies and the
variation in temperatures exist (Rodrı ´guez-Trelles et al.,
1996; Rodrı ´guez-Trelles & Rodrı ´guez, 1998).
Secondly, parallel body size clines as the long-
standing ones in native European populations (Prevosti,
1955; Misra & Reeve, 1964; Pfriem, 1983; Pegueroles
et al., 1995) appeared in New World populations too,
although it took about 20 years after the colonization
for the clines to build up (Huey et al., 2000; Gilchrist
et al., 2001). Interestingly, the size clines are not
isometric in the sense that the relative contribution of
two portions of longitudinal vein IV used to measure
wing length (WL) also changed with latitude but in an
opposite way according to the northern hemisphere
continent (i.e. Europe vs. North America). On the basis
of previous studies in D. melanogaster (Gilchrist et al.,
2000) it was hypothesized that the wing shape variation
in D. subobscura may simply represent drift around an
optimum. Consistent with this idea, recent biometric
and quantitative trait loci (QTL) analyses suggest that
wing size and shape have a contrasting genetic archi-
tecture; the former likely being subjected to directional
selection and the latter to optimizing selection and
regulated largely independently of wing size, with up to
50 loci throughout the D. melanogaster genome having a
significant and generally additive effect on wing shape
as well as minor pleiotropic effects on fitness (Weber
et al., 1999, 2001; Zimmerman et al., 2000; Gilchrist &
Partridge, 2001). Birdsall et al. (2000) used the method
of relative warps (Bookstein, 1991) to study wing shape
variation in 12 inbred lines from D. melanogaster at 18
and 25 ?C. They found that the two rearing tempera-
tures caused differences in wing area of up to 20%, but
wing shape seemed to be independent of sex and
temperature effects on cell growth and density. This is
important because the cellular basis of the body size
cline for D. subobscura in North America (latitudinal
variation in cell area) is different from that in Europe
and South America clines (latitudinal variation in cell
number; Calboli et al., 2003). Therefore, the opposite
latitudinal gradients for wing shape (Huey et al., 2000)
842
M. SANTOS ET AL.
J. EVOL. BIOL. 17 (2004) 841–855 ª 2004 BLACKWELL PUBLISHING LTD

Page 3

do not seem to be related to the cellular bases of the
clines, as suggested by the work of Birdsall et al. (2000).
Third and finally, Orengo & Prevosti (2002) recently
presented evidence for a positive relationship between
wing size and standard gene arrangement dose, an
expected trend according to the latitudinal clines.
However,theyused two
D. subobscura males and the likely presence of nongenetic
effects on body size precludes any firm conclusion. In
addition, the different evolutionary rates observed in
colonizing populations when comparing chromosomal
polymorphism and wing size (see above) is somewhat
puzzling from that putative genetic connection. Further
work is clearly needed.
We havedevelopeda
D. subobscura populations kept at three temperatures
(13, 18 and 22 ?C) to study the short- and long-term
effects of thermal selection on chromosomal inversion
polymorphism and wing size and shape. A large stock
from Puerto Montt (Chile) was chosen as the base
population because (1) the species was detected for the
first time in America at this locality in February 1978
(Brncic et al., 1981); (2) the stock harboured all poly-
morphic chromosome arrangements involved in the New
World latitudinal clines (Prevosti et al., 1988); and (3) the
introduction of the species into South and North America
(first detected in Port Townsend, Washington, in 1982;
Beckenbach & Prevosti, 1986) was the result of a single
colonizing event from a Paleartic population (Mestres
et al., 1992), which provides a unique opportunity to
empirically test how replicated clinal patterns in nature
relate to temperature. Our rearing protocol allows con-
trolling for larval densities, thus minimizing those
potentially spurious correlated responses that may arise
due to other factors not related to thermal selection. Here
we report the initial results showing no size differences
according to thermal regimes, a probably unsurprising
finding for populations that have diverged for 27 gener-
ations at most. However, the analyses of wing shape by
using the shape index in Huey et al. (2000) and the
framework of geometric morphometrics (Bookstein,
1991; Dryden & Mardia, 1998) revealed consistent and
significant differences among temperatures, a somewhat
unexpected finding if shape clines were historically
contingent as formerly sustained (Huey et al., 2000).
We discuss the results in the light of available informa-
tion on Drosophila wing shape, and provide some empir-
ical evidence as to suggest that the highly congruent
results across replicated populations are related to gene-
inversion linkage disequilibria.
samplesof wild-caught
setof threereplicated
Materials and methods
Thermal selection stocks
The D. subobscura populations originated from 93 isofe-
male strains derived from a large outbred stock collected
by Dr J. Balanya `, Dr G. W. Gilchrist, Dr R. B. Huey and Dr
M. Pascual in Puerto Montt (Chile; 41?28¢S) in November
1999. The isofemale lines were kept in 90-mL bottles
(?30–40 breeding adults/bottle) at 18 ?C for more than
1 year (?16 generations) prior to the establishment of the
thermal stocks and, hence, the experimental material had
likely ceased to undergo rapid adaptation to laboratory
conditions (Matos et al., 2000). A large outbreeding
population was founded in March 2001 by randomly
dumping ?25 pairs of virgin flies from each isofemale line
into three Plexiglas cages (27 · 21 · 16 cm3) and main-
tained at 18 ?C (12 : 12 light/dark cycle). A large number
of eggs were collected from these cages, and emerging
adults were randomly dumped into three new cages
(18 ?C; 12 : 12 light/dark cycle). Eggs were sampled from
these cages over 10 consecutive days and placed in
130-mL bottles (?200–250 eggs per bottle) containing
50 mL of David’s killed-yeast Drosophila medium (David,
1962). A total of 225 bottles were set up and randomly
distributed into nine groups with 24 bottles each. Three
groups were allocated at 15 ?C (12 : 12 light/dark cycle),
three at 18 ?C (12 : 12light/dark cycle)andthree at 21 ?C
(12 : 12 light/dark cycle). Therefore, the three replicated
thermal selection stocks were established in May 2001.
The extra nine bottles were used to individually cross a
random sample of emerging males to three to four virgin
females from the ch-cu marker strain in order to estimate
chromosome arrangement frequencies in the initial pop-
ulations. This strain is homozygous for the morphological
recessive markers on the O chromosome cherry eyes (ch)
and curled wings (cu) (Koske & Maynard Smith, 1954),
and its genetic background is highly homogeneous and
fixed for the standard gene arrangements in all major
acrocentric chromosomes but chromosome O, where it is
fixed for gene arrangement O3+4(Lankinen & Pinsker,
1977). Whenever feasible, one F1 female third-instar
larva derived from each cross with the homozygous ch-cu
stock was examined for its inversion loops in polythene
chromosomes to determine the gene arrangements of one
set of the chromosomes from the wild-type male.
After two generations of acclimatization the 15 ?C
stocks were transferred to the final temperature of 13 ?C,
and the 21 ?C stocks to 22 ?C. Previous laboratory
observations (reviewed in Krimbas, 1993) indicate that
optimal temperature for D. subobscura is approximately
18 ?C, and that males can become sterile at 25 ?C or even
lower. Latitudinal clines for optimal temperatures are
quite possible to occur; however, the three temperatures
explored in our thermal selection stocks likely cover
much of the physiologically tolerable range in this species.
All populations are maintained on a discrete genera-
tion, controlled larval crowding regime as follows. Prior
to initiating a new generation, eclosed adults from the
bottles are dumped into a Plexiglas cage and supplied
with liberal amounts of food (two 90 mm Ø Petri dishes
with Drosophila medium supplemented with active dried
yeast) before egg collections. The number of breeding
Thermal evolution in Drosophila subobscura
843
J. EVOL. BIOL. 17 (2004) 841–855 ª 2004 BLACKWELL PUBLISHING LTD

Page 4

adults per population is typically well over 1500 flies.
Once females reach their peak of fecundity (?12–13 days
after emergence at 13 ?C, ?7–8 days at 18 ?C and
?5–6 days at 22 ?C) eggs are collected over a 7-day
period at 13 ?C, 5-day at 18 ?C and 3-day at 22 ?C, and
placed in 130-mL bottles (?200–250 eggs per bottle) as
previously described, with a total of 24 bottles per
population. Generation times (eggs tnfi
?46 days at 13 ?C, ?33 days at 18 ?C and ?25 days at
22 ?C. Consequently, the three sets of populations differ
only in the temperature they experience (humidity was
not strictly controlled but adult flies in the cages had
continuing access to fresh and moist food).
In May 2002 (after nine generations at 13 ?C,
12 generations at 18 ?C, and 15 generations at 22 ?C),
samples from all populations were obtained by placing
eggs into eight additional 130-mL bottles per population.
These bottles were cultured at 18 ?C and emerging adults
were dumped into Plexiglas cages for egg collections.
Eggs for the experiment were collected over a 48-h
period by placing Petri dishes containing nonnutritive
agar with a generous smear of live yeast in the cages.
Larval density was controlled during culturing (100–110
eggs per bottle), and a total of 12 bottles per population
were placed at 18 ?C on the same incubator shelf.
Therefore, the experiment was designed so that the
parents of sampled flies had also been reared at the same
temperature, to control for the possibility of nongenetic
parental effects on offspring size (Crill et al., 1996).
Emerging flies were separated by sex; females stored in
Eppendorf tubes with a 3 : 1 mixture of alcohol and
glycerol at 4 ?C, and a sample of males (125–150 males
per population randomly chosen from the 12 replicated
bottles) were individually crossed in vials (2 · 8 cm
containing 6 mL of food) to three to four virgin females
from the ch-cu marker strain in order to estimate
chromosome arrangement frequencies as previously
tn+1) are
described. After approximately 9 days the males were
removed from the vials and individually fixed in a 3 : 1
mixture of alcohol and glycerol at 4 ?C.
All fly handling was done at room temperature using
CO2anaesthesia, on flies not less than 6 h after eclosion.
Wing size and shape
Definitions
Morphometrics involves the quantitative study of form,
and it is naturally understood that form consists of size
and shape (Needham, 1950). An important contribution
of geometric morphometrics is the clear definition of size
and shape (Dryden & Mardia, 1998). Size is defined as
any positive real-valued function from a landmark
configuration (i.e. a set of points that can be precisely
located)matrix
X
that
g[aX] ¼ ag[X] for any positive scalar a. The shape of a
set of p landmark points is the geometrical information of
the configuration of points that is invariant to transla-
tion, rotation and rescaling.
satisfies the condition
Wing measurements
We analysed here the wing size and shape of flies from
each experimental population. Both wings were removed
and mounted on microscope slides in DPX under cover-
slips from 100 females per population selected haphaz-
ardly from the 12 replicated bottles (see above), as well as
from all males crossed to the ch-cu marker strain when-
ever information of their chromosomal arrangements
was available (>100 males per population; see Table 1).
All the data used here are from the left wings. Wing
images were captured using a compound microscope
(Zeiss Axioskop, Jena, Germany), with low power
objective (2.5·) and attached video camera (Sony CCD-
Iris, Tokyo, Japan), connected to a PC computer with
MGI VideoWave software. Female wings were digitized
Table 1 Mean (±SD) of the basal (L1) and distal (L2) segments of wing longitudinal vein IV, and centroid size (in a normalized form; see
Dryden & Mardia, 1998, p. 24) in Drosophila subobscura for each thermal regime and replicated population (values are given in millimetres;
1 mm ¼ 144 pixels).
Temperature (?C)Replicate
FemalesMales
n
L1L2 Centroid size
n
L1L2Centroid size
13R1
R2
R3
Total
100
100
100
300
1.521 ± 0.040
1.480 ± 0.046
1.479 ± 0.044
1.493 ± 0.048
1.283 ± 0.035
1.250 ± 0.037
1.256 ± 0.036
1.263 ± 0.038
1.014 ± 0.021
0.986 ± 0.023
0.989 ± 0.022
0.996 ± 0.025
115
115
107
337
1.356 ± 0.040
1.322 ± 0.040
1.322 ± 0.046
1.334 ± 0.045
1.154 ± 0.034
1.125 ± 0.034
1.136 ± 0.036
1.138 ± 0.037
0.907 ± 0.019
0.884 ± 0.020
0.887 ± 0.022
0.893 ± 0.023
18R1
R2
R3
Total
100
100
100
300
1.455 ± 0.052
1.452 ± 0.050
1.472 ± 0.055
1.460 ± 0.053
1.247 ± 0.040
1.248 ± 0.037
1.262 ± 0.038
1.252 ± 0.039
0.975 ± 0.027
0.972 ± 0.027
0.986 ± 0.030
0.978 ± 0.028
134
128
132
394
1.297 ± 0.045
1.295 ± 0.049
1.326 ± 0.046
1.306 ± 0.048
1.122 ± 0.033
1.130 ± 0.033
1.142 ± 0.035
1.131 ± 0.035
0.873 ± 0.023
0.873 ± 0.024
0.889 ± 0.024
0.878 ± 0.025
22 R1
R2
R3
Total
100
100
100
300
1.472 ± 0.052
1.478 ± 0.050
1.462 ± 0.050
1.471 ± 0.051
1.246 ± 0.039
1.257 ± 0.037
1.244 ± 0.038
1.249 ± 0.038
0.985 ± 0.026
0.990 ± 0.024
0.979 ± 0.025
0.984 ± 0.025
117
109
111
337
1.323 ± 0.041
1.313 ± 0.049
1.313 ± 0.042
1.317 ± 0.044
1.139 ± 0.034
1.133 ± 0.035
1.136 ± 0.037
1.136 ± 0.035
0.889 ± 0.021
0.883 ± 0.025
0.884 ± 0.019
0.885 ± 0.022
844
M. SANTOS ET AL.
J. EVOL. BIOL. 17 (2004) 841–855 ª 2004 BLACKWELL PUBLISHING LTD

Page 5

by M.S. and male wings by P.F.I., but all wings were
measured by one of us (M.S.). Calibration of the optical
system was checked at each session. x and y coordinates
of 13 morphological landmarks (Fig. 1) were recorded for
each wing using the image processing and analysis
program for the IBM PC Scion Image (based on the
NIH-Image for Macintosh and available at http://
www.scioncorp.com). All landmarks are at the intersec-
tions of wing veins or at points where veins reach the
wing margin and are easy to locate precisely, and can
therefore be considered type 1 landmarks according to
Bookstein (1991), pp. 63–67) or anatomical landmarks
according to Dryden & Mardia (1998), p. 3). Using the
original landmark coordinates we followed Robertson &
Reeve (1952) and Prevosti (1955) by calculating WL as
the combined lengths of the basal (i.e. the Euclidean
distance between landmarks 9 and 13 and labelled as L1)
and distal (Euclidean distance between landmarks 13 and
5 and labelled as L2) segments of longitudinal vein IV.
These linear measurements have been used to study size
and shape clines in D. subobscura (e.g. Pegueroles et al.,
1995; Huey et al., 2000; Gilchrist et al., 2001).
Procrustes analysis
Procrustes methods allow comparing configurations of
landmarks by optimally superimposing (according to a
least-squares criterion) homologous landmarks in two
or more specimens to achieve an overall best fit (Rohlf,
1990, 1999; Rohlf & Slice, 1990; Klingenberg &
McIntyre, 1998). When several objects (e.g. wings) are
fitted using Procrustes superimposition (as was done in
the present work) the method has been called ‘gener-
alized Procrustes analysis’ (GPA) (see Dryden & Mardia,
1998, pp. 44–47). There has been some controversy on
the relative merits of GPA over alternative approaches,
but Rohlf (2000, 2003) has shown that the mean shape
in a population can be reliably estimated using GPA. In
brief, the procedure can be described as follows (Rohlf &
Slice, 1990; Bookstein, 1996): (1) shift the x and y
coordinates to the origin (0, 0) and scale the configu-
rations to unit centroid size (defined as the square root
of the sum of squared distances of a set of landmarks
from their centroid or, equivalently, the square root of
the sum of the variances of the landmarks about that
centroid in x and y directions; Slice et al., 1996); (2)
rotate the configurations against a single reference
specimen to achieve an optimal fit of corresponding
landmarks; (3) a single overall consensus configuration
is computed as the average coordinates of corresponding
landmarks in the aligned configurations; (4) repeat steps
2 and 3 to minimize the sum of the squared distances
between the landmarks of all objects in the sample and
the corresponding landmarks of the consensus config-
urations. The final step was done without additional
scaling and, consequently, we performed a partial
Procrustes fit according to Dryden & Mardia (1998);
see also Rohlf, 1999). [Rescaling the coordinates of each
configuration by the scaling option, 1/cos(q) (see Rohlf,
1999) would make very little difference, in the order of
?0.0015 and ?0.0004% of the shape variation in the
female and male data sets, respectively]. The variation
in the landmark coordinates that remains after Procrus-
tes superimposition is a complete and nonredundant
description of the variation in shape, and the usual
linear multivariate methods focus on these coordinates
(see below).
In this work we used MATLABMATLAB (V.5 and V.6; The
MathWorks, Inc. 1998, 2002) for morphometric analyses,
and results were checked with the ‘tps’ series of programs
by J. F. Rohlf (available at http://life.bio.sunysb.edu/
morph). Some helpful functions in morphometrics from
the MATLABMATLAB toolboxes Res5 and Res6 developed by R. E.
Strauss were also used (available at http://www.biol.
ttu.edu/Strauss/Matlab/matlab.htm).
Statistical analyses
The unit of analysis here is the population, and the
three replicated populations (R1, R2 and R3) of each
thermal stock were treated as a random factor nested
within experimental temperature (13, 18 and 22 ?C),
which was a fixed effect in the ANOVA
Rohlf, 1981).
ANOVAs (see Sokal &
Allometry
To test for size effects on shape variation we carried out
multivariate regressions of Procrustes coordinates on
centroid size (Dryden & Mardia, 1998). These regressions
generally accounted for around 4% of total Procrustes
sums of squares; however, multivariate analyses using
Fig. 1 Image of Drosophila subobscura left wing indicating the 13
landmarks used in this work. The lengths of the proximal (L1) and
distal (L2) segments of longitudinal vein IV were calculated as the
linear distance between landmarks 9 and13, and landmarks 13 and
5, respectively.
Thermal evolution in Drosophila subobscura
845
J. EVOL. BIOL. 17 (2004) 841–855 ª 2004 BLACKWELL PUBLISHING LTD

Page 6

the residuals of a regression on centroid size produced
results that were qualitatively identical to those of the
complete variation. Therefore, we only report the results
of analyses of the total shape variation.
Procrustes ANOVA
As pointed out by Klingenberg & McIntyre (1998),
calculation of Procrustes coordinates is based on the
algebra of sums of squares, and the variance in the set of
optimally aligned landmark configurations can be parti-
tioned in a way analogous to the deviations from a grand
mean in conventional
ANOVAANOVA (Goodall, 1991). The
coordinates of the Procrustes-aligned configurations are,
therefore, amenable to the preceding two-level nested
ANOVA model.
For each of the x and y coordinates of the aligned
configurations a separated two-level nested ANOVA
run, and the resulting sums of squares for temperature,
replicate and error in the Procrustes
obtained after summing the corresponding sums of
squares across x and y coordinates of all landmarks.
There are more degrees of freedom in Procrustes ANOVA
than in conventional ANOVAANOVA (Goodall, 1991) because the
squared deviations are summed over all the landmark
coordinates. Therefore, the number of degrees of freedom
is that for ordinary ANOVAANOVA times the shape dimension;
namely, 2p ) 4 for two-dimensional coordinate data,
where p is the number of landmarks.
To avoid making assumptions about the specific distri-
bution of wing shapes around the mean landmark
configuration we used permutation tests (Manly, 1997)
for testing the statistical significance of ANOVA
Permutation tests also avoid the rather stringent statis-
tical constraints of the covariance structure described by
Goodall (1991) (see also Rohlf, 2000). For the two-level
nested
ANOVA ANOVA model randomization is a two-stage
process: (1) random permutations among subgroup
(replicated populations) within group (experimental
temperature) for the among-subgroup F-statistics; and
(2) random permutations among subgroup and group for
the among-group F-statistics. Each test used 10 000
random permutations of the observations.
ANOVA
ANOVA
ANOVA was
ANOVAANOVAs were
ANOVA
ANOVA effects.
Localized variation
In order to assess how much of shape variation was due
to each landmark, we followed Klingenberg & McIntyre
(1998) and decomposed the Procrustes mean squares for
each effect in the two-level nested
according to the landmarks. Thus, we summed x and y
mean squares of each landmark separately and com-
putedthe variancecomponents
expected mean squares (Sokal & Rohlf, 1981). Because
the least-squares algorithm tends to spread variation
from variable landmarks to the others, this approach
should be taken cautiously if one or a few landmarks
are much more variable than the rest (Chapman, 1990;
Walker, 2000).
ANOVAANOVA model
according tothe
Shape variability
We used principal component analysis (PCA) (see e.g.
Jolliffe, 1986) to investigate patterns of covariation in the
positions of landmarks, which is a usual method in the
context of shape analysis (Dryden & Mardia, 1998;
Klingenberg & McIntyre, 1998; Klingenberg & Zaklan,
2000). The analyses must use covariance matrices of the
coordinates of superimposed landmarks to avoid pro-
blems related to rotations of the coordinate system, and
principal components coefficients can be presented
graphically by drawing lines centred at the mean location
of each landmark and ending at an arbitrary number of
standard deviations away from that mean in the direction
to which the landmark would shift.
The computer programs used for statistical data
analyses were MATLABMATLAB (V.5 and V.6, The MathWorks,
Inc. 1998, 2002), and the statistical software packages
STATISTICA V.6 (StatSoft, Inc., 2003) and SPSS V.11
(SPSS Inc., 2001). They were run on a Pentium?4
(1.60 GHz) PC-compatible.
STATISTICA
Results
Wing size and shape index
The mean values for the basal (L1) and distal (L2) WLs
and centroid sizes were calculated for females and males
from each population (Table 1). Flies were considerably
larger than the F2offspring from a wild sample collected
in 1986 at the same locality of Puerto Montt and raised
under similar conditions of food and temperature (see
Pegueroles et al., 1995). Females were approximately
11% bigger than males, and average sizes for the 13 ?C
thermal selection regime were slightly bigger than those
at warmer temperatures. There was, however, substan-
tial variation among replicated populations and the two-
level nested ANOVA ANOVAs did not detect any significant effect
of thermal selection regimes on WL or centroid size. In
addition, there was no indication of a linear trend
between size and temperature (Table 2).
However, significant effects of thermal selection
regimes (but not replicates) were observed for the length
of L1 relative to total WL, and the patterns were similar
in both sexes (Fig. 2). Nonlinear (deviation) effects were
significant and, therefore, this shape index did not show
a neat linear trend with temperature (Table 2). It is
worth mentioning that post hoc comparisons for females
only detected statistically significant differences when
contrasting 18 ?C vs. 13 ?C or 22 ?C; namely, the length
of L1 relative to that observed at 18 ?C significantly
increased at the lowest and highest temperatures in
females (but not in males where the shape index was
significantly higher only at the lowest temperature;
analyses not shown). When globally considered, these
findings could be compatible with the contrasting
patterns found in the native Paleartic populations (i.e.
a positive correlation between the shape index and
846
M. SANTOS ET AL.
J. EVOL. BIOL. 17 (2004) 841–855 ª 2004 BLACKWELL PUBLISHING LTD

Page 7

latitude) and in recently colonizing populations of North
America (a negative correlation with latitude; Huey et al.,
2000).
Wing shape variability
The Procrustes two-level nested ANOVA
shape variation led to the same previous conclusion
regarding shape index L1/WL; namely, a significant and
consistent (across replicated populations) shape variation
was detected among thermal selection regimes but with a
nonlinear trend for temperature (Table 3a. Notice that
probabilities from the two-stage permutation test are
ANOVA for female wing
higher than those from a ‘true’ F-statistics with the same
degrees of freedom, which clearly indicates that the
degrees of freedom in our Procrustes ANOVA
independent and also suggests correlated landmark
shifts.) However, no statistically significant differences
were detected for males (Table 3b), which might suggest
that they are lagging behind females. [Actually, when
Procrustes distances were used as the dependent variable
(which allows for global tests of shape differences),
significant effects of thermal selection regimes were
detected for both sexes (see below).] A differentiated
pattern between sexes also emerged when variance
components from the Procrustes ANOVA
tioned by landmarks. Thus, landmarks 1, 2 and 13
dominated for temperature effects in females but had
relatively low amounts of variability in males. However,
landmark 5 had the largest temperature effect in males
(Fig. 3). To summarize, although the relative amounts of
variation at each landmark vary markedly among the
factors included in the two-way nested ANOVA
perature effects seem to be significant in both sexes for at
least one landmark involved in the shape index L1/WL
(as could be expected from the results in Table 2).
Principal component analyses were carried out for
overall shape variation among individuals (within sexes)
across the entire wing (i.e. from the covariance matrices
of the coordinates of superimposed landmarks obtained
from the 900 females and the 1068 males), as well as
among thermal regimes (i.e. from the covariance matri-
ces of the mean coordinates at each temperature). For
individual variation the first two principal components
explained at least 63 and 7% of the variability, respect-
ively, and results were fairly consistent between sexes
(Fig. 4). This high level of variability explained by a few
PCs clearly suggests strong dependencies among land-
marks and, hence, the isotropic model (which presumes
that there is an equal amount of nondirectional variation
ANOVA are not
ANOVAs were appor-
ANOVAs, tem-
Table 2 Two-level nested analyses of variance for wing length [WL: as loge(L1 + L2) in pixels; 1 mm ¼ 144 pixels], centroid size (in pixels),
and shape index (basal length/wing length) [as loge(L1/WL)]. The sums of squares for the fixed factor temperature were further decomposed
to test for linear (regression) and nonlinear (deviation) effects.
Source d.f.
WLCentroid size L1/WL
SSMS
FP
SS MS
FP
SSMS
FP
(a) Females
Temperature
Regression
Deviation
Replicate
Error
2
1
1
6
4.49 · 10)2
2.89 · 10)2
1.60 · 10)2
6.11 · 10)2
0.59
2.24 · 10)2
2.89 · 10)2
1.60 · 10)2
1.02 · 10)2
6.58 · 10)4
2.20
2.84
1.57
15.47
0.192
0.143
0.257
<0.001
1063.5
498.8
564.7
1370.4
11592.9
531.7
498.8
564.7
228.4
13.0
2.33
2.18
2.47
17.56
0.179
0.190
0.167
<0.001
7.09 · 10)3
0.94 · 10)3
6.15 · 10)3
1.17 · 10)3
0.27
3.54 · 10)3
0.94 · 10)3
6.15 · 10)3
0.19 · 10)3
0.30 · 10)3
18.23
4.83
31.62
0.64
0.003
0.070
0.001
0.696
891
(b) Males
Temperature
Regression
Deviation
Replicate
Error
2
1
1
6
3.73 · 10)2
1.29 · 10)2
2.44 · 10)2
7.67 · 10)2
0.68
1.86 · 10)2
1.29 · 10)2
2.44 · 10)2
1.28 · 10)2
6.46 · 10)4
1.46
1.01
1.91
19.78
0.304
0.353
0.216
<0.001
789.7
238.6
551.1
1220.7
10709.3
394.9
238.6
551.1
203.4
10.1
1.94
1.17
2.71
20.12
0.223
0.320
0.151
<0.001
8.77 · 10)3
4.62 · 10)3
4.15 · 10)3
3.88 · 10)3
0.36
4.39 · 10)3
4.62 · 10)3
4.15 · 10)3
0.65 · 10)3
0.34 · 10)3
6.79
7.14
6.42
1.88
0.029
0.037
0.044
0.082
1059
Fig. 2 Averages of the relative length (with 95% confidence
intervals) of the basal portion of longitudinal vein IV (L1) to the total
wing length (WL ¼ L1 + L2) vs. experimental temperature for
females (d) and males (j). All three replicated populations within
each thermal regime were pooled because no statistically significant
differences among replicates were detected in the two-way nested
ANOVAANOVAs (see Table 2).
Thermal evolution in Drosophila subobscura
847
J. EVOL. BIOL. 17 (2004) 841–855 ª 2004 BLACKWELL PUBLISHING LTD

Page 8

at each landmark) does not seem to hold (see Dryden &
Mardia, 1998, p. 97). The effects of between-individual
variation were distributed in a relatively even way
among all landmarks. Outer landmarks tended to move
in the direction of an enlarged wing aspect; namely, to
decrease wing width relative to WL. However, the
direction of PCs is arbitrary and all the movements can
be simultaneously reversed by 180?. Therefore, the best
interpretation is to describe overall shape variation along
a wide–narrow direction. The PC2 mainly consisted of a
widening (narrowing) of the posterior compartment of
the wing (which includes those landmarks located below
an imaginary line situated approximately along the
fourth longitudinal vein; see Garcı ´a-Bellido & de Celis,
1992). The PC3 simultaneously involved two landmarks
on the fourth longitudinal vein (5 and 13) shifting to
reverse directions, resulting in an increase (decrease) of
the length of the distal segment L2. Conventional
morphometric methods performed on wing size mea-
surements in samples from Europe and North America
also detected an inverse relationship (accounting for 10%
of the variance) between the proximal (L1) and distal
(L2) portions of longitudinal vein IV, as well as between
WL and width (Gilchrist et al., 2001).
For temperature effects there are only two PCs, and the
features of shape variation are graphically shown in
Fig. 5. Some slight differences between sexes are now
apparent (but recall that Procrustes
detect statistically significant thermal effects on males;
Table 3). The PC1 in females was connected to the large
variability previously detected for landmark 1 (see
Fig. 3), which moves towards landmark 9. However,
several landmarks had relatively large PC1 coefficients in
males. Landmarks 13 and 7, which define the position of
the posterior cross vein, shift to the same direction in
both sexes. Similar to PC3s for individual variation,
landmarks 5 and 13 seem to move in opposite directions
but these shifts were quite small in females. The PC2 was
mainly involved with the shift of landmark 5 in males
and females.
ANOVA ANOVAs did not
Rates of genetic divergence for wing shape
Rates of wing size evolution and divergence on a
continental scale have been estimated to be very fast in
D. subobscura (Huey et al., 2000; Gilchrist et al., 2001). We
did not detect here any significant effect of thermal
selection regimes on wing size, but it could be interesting
to estimate the rates of divergence for wing shape in the
experimental populations (Hendry & Kinnison, 1999
describe this as the synchronic method). To handle wing
shape as a single metric we used Procrustes distances and
calculated them as q ¼ 2 sin)1(dp/2), where dpis the
square root of the sum of square differences between
corresponding points (see Rohlf, 1999). This procedure
obviouslyreducesthe inherently
shape data to a single magnitude of shape differences,
multidimensional
Table 3 Two-level nested analyses of variance for shape using
Procrustes sums of squares as a measure of overall variation in shape.
Statistical significance was computed after 10 000 random permu-
tations (see text for details).
Sourced.f. SSMS
FP
(a) Females
Temperature
Regression
Deviation
Replicate
Error
44
22
22
11.58 · 10)3
2.36 · 10)3
9.22 · 10)3
8.63 · 10)3
696.81 · 10)3
2.63 · 10)4
1.07 · 10)4
4.19 · 10)4
0.65 · 10)4
0.36 · 10)4
4.02
1.64
6.40
1.81
0.019
0.196
0.007
0.071
132
19 602
(b) Males
Temperature
Regression
Deviation
Replicate
Error
44
22
22
5.53 · 10)3
1.98 · 10)3
3.55 · 10)3
10.29 · 10)3
899.41 · 10)3
1.26 · 10)4
0.90 · 10)4
1.61 · 10)4
0.78 · 10)4
0.39 · 10)4
1.61
0.15
2.07
2.02
0.187
0.318
0.113
0.016
132
23 298
Fig. 3 Pie charts indicating the percentage of the variance for the
effects of temperature (black), replicate (white) and error (grey) in
the Procrustes ANOVA ANOVAs on individual landmarks. Those landmarks
used to estimate shape index (basal length/wing length) are
underlined. Significance of temperature effect are indicated by
asterisks (*P < 0.05; ***P < 0.001).
848
M. SANTOS ET AL.
J. EVOL. BIOL. 17 (2004) 841–855 ª 2004 BLACKWELL PUBLISHING LTD

Page 9

and ignores the direction of landmarks’ shifts against
each other (Goodall, 1991; Klingenberg, 2003; but see
Monteiro et al., 2003). However, the Procrustes distances
are informative as summary statistics and can be used to
investigate evolutionary rates of shape change.
Two-level nested ANOVAANOVAs applied to the Procrustes
distances also rendered significant and consistent (across
replicated populations) shape variation among thermal
selection regimes for both sexes (females: F44,132¼ 6.97,
P < 0.05; males: F44,132¼ 7.51, P < 0.05. Probabilities
were computed after 10 000 random permutations as
described in Materials and methods); and post hoc
comparisons detected statistically significant differences
when contrasting 18 ?C vs. 13 ?C or 22 ?C in females,
and 13 ?C vs. 22 ?C in males (analyses not shown).
Incidentally, although qualitative conclusions obtained
from shape index L1/WL (see Table 2 and above) and
from Procrustes distances are the same, both variables are
loosely correlated (females: Spearman rS¼ )0.020, n.s.;
males: rS¼ )0.068, P < 0.05).
Divergence rates for wing shape were estimated to be
0.0105 haldanes (2.1 · 104darwins; 13 ?C vs. 18 ?C) and
0.0106 haldanes (2.7 · 104darwins; 18 ?C vs. 22 ?C) in
females, and 0.0068 haldanes (1.7 · 104darwins; 13 ?C
vs. 22 ?C) in males. Our approach of comparing only
those means that showed statistically significant differ-
ences in post hoc contrasts can be obviously criticized (see
Hendry & Kinnison, 1999). However, the previous
figures do suggest that rates of genetic divergence for
wing shape can be as fast or even faster than those
estimated for wing size (cf. with the values reported by
Gilchrist et al., 2001).
Relationship between chromosomal polymorphism
and wing shape index
The frequencies of chromosomal gene arrangements in
the original natural population at Puerto Montt, in the
initial founding population, and in the thermal selection
stocks after 1 year of divergence are shown in Table 4.
Three-way log-linear analyses (including experimental
temperature, replicate, and gene arrangement as the
main effects) performed for each chromosome showed
that the frequencies of some gene arrangements have
already changed according to experimental temperature,
but the three replicate populations within each tempera-
ture were homogeneous and could be lumped together
(results not shown). Our aim here, however, is not to
discuss those changes in gene arrangement frequencies
but to relate the simultaneously analysed chromosomal
Fig. 4 Principal components of the covariance patterns in landmarks shifts due to among-individual variation for each sex. The PC coefficients
are shown as a solid line originating at the mean location of the landmark (open circles) and ending at the location to which the landmark
would move to +6 (PC1), +10 (PC2) and +12 (PC3) standard deviations (obviously an exaggeration of the variation in the dataset). The
proportion of total variation accounted by each PC is given in brackets.
Thermal evolution in Drosophila subobscura
849
J. EVOL. BIOL. 17 (2004) 841–855 ª 2004 BLACKWELL PUBLISHING LTD

Page 10

polymorphism and wing shape index in the males
sampled from the thermal stocks. For this purpose we
have used the ‘standard dose’ (i.e. the number of
standard gene arrangements carried out by a male,
which ranges from 0 to 5) as the relevant variable. The
reason is that the various gene arrangements in Table 4
can be divided in two groups based on the correlation of
gene arrangement frequencies and latitude: the ‘cold-
adapted’ group comprised by Ast, Jst, Ust, Estand Ost; and
the ‘warm-adapted’ group involving A2, J1, U1+2, U1+2+8,
E1+2+9+3, E1+2+9+12, O3+4 and O3+4+8 (arrangements
E1+2+9, O3+4+2and O3+4+7do not show a clear latitudinal
pattern; see Menozzi & Krimbas, 1992). Because the
frequencies of some standard gene arrangements were
relatively low, we have grouped the low frequencies at
the upper tail of the distribution and the relationship
between shape index L1/WL and the standard dose is
plotted in Fig. 6 for each experimental temperature. In
all cases, the length of L1 relative to total WL sharply
decreased with the standard dose [dependent variable
loge(L1/WL); 13 ?C: b ¼ )5.16 · 10)3, F1,328¼ 20.11,
P < 0.001;18 ?C:
b ¼ )3.85 · 10)3,
P < 0.001; 22 ?C:
b ¼ )6.46 · 10)3,
F1,358¼ 12.37,
F1,324¼ 26.23,
P < 0.001], and the regression coefficients were not
statistically heterogeneous [F2,1010¼ 1.18, n.s. Overall
b ¼ )5.10 · 10)3(95% confidence limits: )6.42 · 10)3,
)3.78 · 10)3)]. It thus seems quite clear that polymor-
phic gene arrangements in D. subobscura have a consistent
(across temperatures) biometrical effect on wing shape.
The standard dose in the thermal stocks increased with
increasing temperature, which could somewhat explain
the pattern observed in Fig. 2 for males (in fact, the
statistically significant effect for temperature in Table 2b
disappears when the standard dose is introduced in the
analysis as a covariate; results not shown). Interestingly,
the decrease of wing shape index with latitude observed
in North American colonizing populations (Huey et al.,
2000) fully agrees with the present data because the
standard gene arrangements generally increase with
latitude in those populations (Balanya ` et al., 2003).
Actually, the increase of standard dose with latitude for
eight North American populations covering a latitudinal
range of about13?
and
b ¼ 0.072 ± 0.010. As formerly pointed out we did not
aim here to discuss the changes in gene arrangement
frequencies in the thermal selection stocks, but just
sampledin1994was
Fig. 5 Principal components of the covariance patterns in landmarks shifts due to among-temperature variation for each sex. The end points of
the solid lines are at locations displaced +15 (PC1) and +30 (PC2) standard deviations from the mean configuration.
850
M. SANTOS ET AL.
J. EVOL. BIOL. 17 (2004) 841–855 ª 2004 BLACKWELL PUBLISHING LTD

Page 11

notice that the trend between the standard dose and
temperature in these stocks after 1 year of divergence
goes in the opposite direction to that expected from the
latitudinal clines.
Discussion
Flies of both sexes that have diverged for 27 generations
at most were found to be about the same size independ-
ently of thermal selection regime. The lack of divergence
in wing size could probably have been expected from
results in the first surveys conducted on D. subobscura flies
from New World populations, which failed to show any
latitudinal size clines 8 years after colonization (?40
generations; see Pegueroles et al., 1995). Further samples
from the laboratory populations collected after longer
divergence times will allow us to test predictions based
on the size clines. However, features of wing shape that
differ between experimental treatments (and sexes) were
found to be consistent within the three replicated
populations, which clearly excludes any explanation of
wing shape variation in D. subobscura grounded on drift
around an optimum (Gilchrist et al., 2001; see below).
Earlier experiments in Drosophila had also detected
progressive differences in wing form between (unrepli-
cated) stocks maintained at different temperatures
(Cavicchi et al., 1978). The changes, however, were not
significant in any single dimension but were detected
after a multivariate analysis. Further artificial selection
experiments (Cavicchi et al., 1981) led the authors to the
conclusion that the individual wing dimensions are
largely inseparable genetically. However, Weber (1990)
has rightly argued that a shortcoming with the latter
experiment is that the authors only selected on one wing
dimension and in one direction, which is equivalent to
directional selection for size (but not shape) because most
of the genetic variance for size in any single dimension is
simply variance for total size. When artificial selection is
performed on angular offsets extensive localized effects
are revealed, making it clear that the wing does not
Table 4 Chromosomal polymorphism in the original natural population at Puerto Montt, in the founding population, and in the thermal
selection stocks after 1 year.
Chromosome
arrangement
Natural
population
(November
1999)
Founding
population
(May 2001)
Experimental populations
13 ?C (generation 9)
(May 2002)
18 ?C (generation 12)
(May 2002)
22 ?C (generation 15)
(May 2002)
R1R2 R3 TotalR1 R2R3 TotalR1 R2R3Total
A
Ast
A2
N
0.507
0.493
134
0.306
0.694
121
0.248
0.752
117
0.157
0.843
115
0.236
0.764
110
0.213
0.787
342
0.198
0.802
126
0.226
0.774
124
0.258
0.742
120
0.227
0.773
370
0.096
0.904
114
0.164
0.836
110
0.224
0.776
107
0.160
0.840
331
J
Jst
J1
N
0.304
0.696
135
0.281
0.719
121
0.145
0.855
117
0.118
0.882
119
0.089
0.911
112
0.118
0.882
348
0.108
0.892
139
0.114
0.886
132
0.194
0.806
134
0.138
0.862
405
0.190
0.810
116
0.188
0.813
112
0.189
0.811
111
0.189
0.811
339
U
Ust
U1+2
U1+2+8
N
0.422
0.348
0.230
135
0.512
0.281
0.207
121
0.265
0.538
0.197
117
0.269
0.479
0.252
119
0.277
0.482
0.241
112
0.270
0.500
0.230
348
0.312
0.529
0.159
138
0.220
0.591
0.189
132
0.239
0.515
0.246
134
0.257
0.545
0.198
404
0.448
0.353
0.198
116
0.339
0.438
0.223
112
0.351
0.477
0.171
111
0.381
0.422
0.198
339
E
Est
E1+2
E1+2+9
E1+2+9+3
E1+2+9+12
N
0.593
0.037
0.059
0.148
0.163
135
0.612
–
0.074
0.231
0.083
121
0.530
–
0.034
0.214
0.222
117
0.462
–
0.067
0.168
0.303
119
0.491
–
0.027
0.161
0.321
112
0.494
–
0.043
0.181
0.282
348
0.755
–
0.036
0.101
0.108
139
0.659
–
0.098
0.076
0.167
132
0.709
–
0.022
0.179
0.090
134
0.709
–
0.052
0.119
0.121
405
0.828
–
0.112
0.043
0.017
116
0.786
–
0.045
0.116
0.054
112
0.865
–
0.045
0.063
0.027
111
0.826
–
0.068
0.074
0.032
339
O
Ost
O3+4
O3+4+2
O3+4+7
O3+4+8
O5
O7
N
0.289
0.081
0.252
0.126
0.170
0.074
0.007
135
0.190
0.050
0.421
0.140
0.149
0.050
–
121
0.111
0.068
0.581
0.085
0.154
–
–
117
0.109
0.034
0.588
0.143
0.109
0.008
0.008
119
0.045
0.080
0.625
0.116
0.125
0.009
–
112
0.089
0.060
0.598
0.115
0.129
0.006
0.003
348
0.101
0.072
0.619
0.108
0.101
–
–
139
0.098
0.045
0.606
0.114
0.136
–
–
132
0.082
0.075
0.500
0.179
0.157
0.007
–
134
0.094
0.064
0.575
0.133
0.131
0.002
–
405
0.121
0.181
0.534
0.043
0.121
–
–
116
0.018
0.259
0.455
0.125
0.125
0.018
–
112
0.090
0.189
0.550
0.108
0.063
–
–
111
0.077
0.209
0.513
0.091
0.103
0.006
–
339
Thermal evolution in Drosophila subobscura
851
J. EVOL. BIOL. 17 (2004) 841–855 ª 2004 BLACKWELL PUBLISHING LTD

Page 12

evolve as a whole unit (Weber, 1990). Nevertheless,
some contrasting patterns between overall wing variation
(Fig. 4) and temperature effects (Figs 3 and 5) are
apparent in our data set. In the first case the shifts of
landmarks in PC1 did not occur in isolation, but also
included most landmarks. Temperature has more locali-
zed effects in the shifts of landmarks. Similarly to the
finding by Klingenberg & Zaklan (2000) in D. melanoga-
ster, the shifts of the anterior (PC2 in Fig. 4) and posterior
(PC3) cross veins appear to be rather independent of each
other. These authors discuss similar displacements detec-
ted in both intra- and interspecific studies of wild-type
flies, as well as in a number of mutant stocks.
In contrast to what has been uncovered for wing size,
there are no consistent patterns between latitude and
wing shape in Drosophila. Shape index L1/WL in
D. subobscura increases with latitude in Europe and
decreases in North America (Huey et al., 2000), but
shows no linear trend in South America (G. W. Gilchrist,
pers. comm., 2002). In D. melanogaster Gilchrist et al.
(2000) found that the main shape canonical variate also
varied significantly between continents. Hoffmann &
Shirriffs (2002) found a nonlinear trend between latitude
and the first shape canonical variate in D. serrata flies
from Australia, although there was a linear change in
wing aspect (the ratio of WL to wing width). Parenthe-
tically, the shape changes brought about by thermal
selection regimes in our D. subobscura stocks were not
associated with changes in wing-aspect ratio [estimated
as (wing length)2/wing area (see Azevedo et al., 1998).
Females: F2,6¼ 1.31, n.s.; males: F2,6¼ 0.84, n.s.].
Overall the results of these studies, as far as can be
judged from the published reports on traditional and
geometric morphometrics, quantitative genetics and QTL
analyses of Drosophila wing shape (Weber, 1990, 1992;
Bitner-Mathe ´ & Klaczko, 1999a,b; Weber et al., 1999,
2001; Birdsall et al., 2000; Klingenberg & Zaklan, 2000;
Zimmerman et al., 2000; Gilchrist & Partridge, 2001),
suggest that size and shape have different genetic
properties and do not respond to the same environmental
factors. In summary, the steady geographical clines for
size across continents and Drosophila species, with wing
size increasing with latitude largely independently of the
underlying details in the genetic architecture (Gilchrist &
Partridge, 1999), strongly support the hypothesis that
body size (or the correlated trait growth rate) is the target
of selection but there is a localized and richly structured
variation in wing shape.
As wing shape seems to be strongly resistant to
environmental influences (Weber, 1990; Birdsall et al.,
2000) and there is little compelling evidence indicating
that natural wing shape changes are adaptive in
Drosophila, it would be premature to attempt to explain
the high rates of genetic divergence in our thermal
selection stocks in functional terms [notwithstanding
Imasheva et al.’s (1995) conclusions]. Wing shape is
indeed remarkably constant within inbred lines (Birdsall
et al., 2000); however, it does not seem to be strongly
canalized against genetic change and responds to diver-
gent selection in the same way as most quantitative traits,
with some genes causing small and localized effects
(Weber, 1990, 1992). Many genes with small additive
effects on wing shape are dispersed along the Drosophila
genome (Weber et al., 1999, 2001; Zimmerman et al.,
2000). This suggests plentiful chances for gene-inversion
linkage disequilibriainthe
D. subobscura, particularly in samples derived from New
World colonizing populations (actually, linkage disequil-
ibria between microsatellite loci are almost absent in
European populations but are detectable in New World
populations; M. Pascual, pers. comm., 2002). The thermal
selection stocks have already diverged for various gene
arrangements with no differences between replicated
populations (see above), and a strong relationship
between shape variables and polymorphic inversions
(i.e. shape index L1/WL sharply decreased at all experi-
mental temperatures as the dose of standard chromoso-
mal arrangements increased) was uncovered. In addition,
our findings strongly suggest that the latitudinal clines for
chromosomal gene arrangements may account for the
wing shape cline in North America colonizing populations
(Huey et al., 2000). Even more, we could hypothesize that
the shape cline in North America predated the size cline
because of the quite different paces between inversion
and size clines (see above). To test this hypothesis we
have reanalysed the data reported in Pegueroles et al.
(1995) for six North American populations sampled by
A. Prevosti andM. Monclu ´s in July 1986 (4 yearsafter the
inversion-rich species
Fig. 6 Averages of the relative length (with 95% confidence
intervals) of the basal portion of longitudinal vein IV (L1) to the total
wing length (WL ¼ L1 + L2) according to the dose (0, 1, 2 and ‡3) of
standard gene arrangements carried by the sampled males at each
experimental temperature. All three replicated populations within
each thermal selection regime were pooled (see text for details).
852
M. SANTOS ET AL.
J. EVOL. BIOL. 17 (2004) 841–855 ª 2004 BLACKWELL PUBLISHING LTD

Page 13

initial colonization in North America) before the size cline
built up. As predicted, for both females [dependent
variableloge(L1/WL);
b ¼ )0.92 ± 0.27 · 10)3]
males (b ¼ )0.49 ± 0.26 · 10)3) the length of L1 relative
to total WL decreased with latitude, but the colonizing
males were lagging behind females and this tendency is
still apparent in more recent samples (see Gilchrist et al.,
2001). In summary, our results cast strong doubts on the
supposed ‘unpredictability’ of the geographical cline in
D. subobscura North American colonizing populations
(Huey et al., 2000).
The use of geometric morphometrics together with
quantitative genetic studies is an ambitious project and
requires large data sets (see Klingenberg & Leamy, 2001;
Klingenberg, 2003). We are currently analysing male’s
multivariate wing shape tangent space in relation to
inversions, in addition to carrying out quantitative
genetic analyses in a set of isochromosomal lines fixed
for different O chromosome arrangements. In the mean-
time, the most parsimonious explanation for our present
results is that changes in gene arrangement frequencies
as a response to temperature likely underlie the corre-
lated changes in wing shape because of gene-inversion
linkage disequilibria.
and
Acknowledgments
This paper is dedicated to Antonio Prevosti for his
constant scientific and personal inspiration, as well as
for his fundamental contribution to the development of
population and evolutionary genetics in Spain. Many
thanks go to Gerdien de Jong and Linda Partridge for
sharing with us manuscripts before publication. We
thank Raymond Huey, Linda Partridge, Francisco Rodrı ´-
guez-Trelles and two anonymous reviewers for valuable
comments on earlier drafts of the manuscript. P.F.I. is
supported by a postdoctoral fellowship (SB2000-0370)
from the Secretarı ´a de Estado de Educacio ´n y Universid-
ades del Ministerio de Educacio ´n, Cultura y Deporte
(Spain), and W.C. by a postgraduate fellowship (FP2000-
7001) from the Ministerio de Ciencia y Tecnologı ´a
(Spain). This work was supported by grants BOS2000-
0295-C02 from the Ministerio de Ciencia y Tecnologı ´a
(Spain), 2001SGR-00207 from the Direccio ´ General de
Recerca (Generalitat de Catalunya) to the GBE, and by
Fundacio ´n Ramo ´n Areces (Spain).
References
Anderson, W.W. 1966. Genetic divergence in M. Vetukhiv’s
experimental populations of Drosophila pseudoobscura. Genet.
Res. Camb. 7: 255–266.
Azevedo, R.B.R., French, V. & Partridge, L. 1996. Thermal
evolution of egg size in Drosophila melanogaster. Evolution 50:
2338–2345.
Azevedo, R.B.R., James, A.C., McCabe, J. & Partridge, L. 1998.
Latitudinal variation of wing: thorax size ratio and wing-
aspect ratio in Drosophila melanogaster. Evolution 52: 1353–
1362.
Balanya `, J., Serra, L., Gilchrist, G.W., Huey, R.B., Pascual, M.,
Mestres, F. & Sole ´, E. 2003. Evolutionary pace of chromoso-
mal polymorphism in colonizing populations of Drosophila
subobscura: an evolutionary time series. Evolution 57: 1837–
1845.
Beckenbach, A.T. & Prevosti, A. 1986. Colonisation of North
America by the European species Drosophila subobscura and
Drosophila ambigua. Am. Midl. Nat. 115: 10–18.
Birdsall, K., Zimmerman, E., Teeter, K. & Gibson, G. 2000.
Genetic variation for the positioning of wing veins in
Drosophila melanogaster. Evol. Devel. 2: 16–24.
Bitner-Mathe ´, B.C. & Klaczko, L.B. 1999a. Heritability, pheno-
typic and genetic correlations of size and shape of Drosophila
mediopunctata wings. Heredity 83: 688–696.
Bitner-Mathe ´, B.C. & Klaczko, L.B. 1999b. Size and shape
heritability in natural populations of Drosophila mediopunctata:
temporalandmicrogeographicalvariation.Genetica105:35–42.
Bochdanovits, Z. & De Jong, G. 2003. Experimental evolution in
Drosophila melanogaster: interaction of temperature and food
quality selection regimes. Evolution 57: 1829–1836.
Bookstein, F.L. 1991. Morphometric Tools for Landmark Data: Geom-
etry and Biology. Cambridge University Press, Cambridge, UK.
Bookstein, F.L. 1996. Biometrics, biomathematics and the
morphometric synthesis. Bull. Math. Biol. 58: 313–365.
Borash, D.J., Gibbs, A.G., Joshi, A. & Mueller, L.D. 1998. A
genetic polymorphism maintained by natural selection in a
temporally varying environment. Am. Nat. 151: 148–156.
Borash, D.J., Pierce, V.A., Gibbs, A.G. & Mueller, L.D. 2000.
Evolution of ammonia and urea tolerance in Drosophila
melanogaster: resistance and cross-tolerance. J. Insect Physiol.
46: 763–769.
Brncic, D., Prevosti, A., Budnik, M., Monclu ´s, M. & Ocan ˜a, J.
1981. Colonization of Drosophila subobscura in Chile. I. First
population and cytogenetics studies. Genetica 56: 3–9.
Calboli, F.C.F., Gilchrist, G.W. & Partridge, L. 2003. Different cell
size and cell number contribution in two newly established
and one ancient body size cline of Drosophila subobscura.
Evolution 57: 566–573.
Cavicchi, S., Giorgi, D. & Mochi, M. 1978. Investigation of early
divergence between populations of Drosophila melanogaster
kept at different temperatures. Genetica 48: 81–87.
Cavicchi, S., Pezzoli, C. & Giorgi, G. 1981. Correlation between
characters as related to developmental pattern in Drosophila.
Genetica 56: 189–195.
Cavicchi, S., Guerra, D., Giorgi, G. & Pezzoli, C. 1985.
Temperature-related divergence in experimental populations
of Drosophila melanogaster. I. Genetic and developmental basis
of wing size and shape variation. Genetics 109: 665–689.
Cavicchi, S., Guerra, D., Natali, V., Pezzoli, C. & Giorgi, G. 1989.
Temperature-related divergence in experimental populations
of Drosophila melanogaster. II. Correlation between fitness and
body dimensions. J. Evol. Biol. 2: 235–251.
Chapman, R.E. 1990. Conventional Procrustes approaches. In:
Proceedings of the Michigan Morphometrics Workshop (F. J. Rohlf &
F. L. Bookstein, eds), pp. 251–267. Special Publication No. 2.
University of Michigan Museum of Zoology, Ann Arbor, MI.
Coyne, J.A. & Beecham, E. 1987. Heritability of two morpho-
logical characters within and among natural populations of
Drosophila melanogaster. Genetics 117: 727–737.
Thermal evolution in Drosophila subobscura
853
J. EVOL. BIOL. 17 (2004) 841–855 ª 2004 BLACKWELL PUBLISHING LTD

Page 14

Crill, W.D., Huey, R.B. & Gilchrist, G.W. 1996. Within- and
between-generation effects of temperature on the morphology
and physiology of Drosophila melanogaster. Evolution 50: 1205–
1218.
David, J. 1962. A new medium for rearing Drosophila in axenic
conditions. Dros. Inf. Serv. 36: 128.
David, J., Bocquet, C. & De Scheemaeker-Louis, M. 1977.
Genetic latitudinal adaptation of Drosophila melanogaster: new
discriminative biometrical traits between European and
equatorial African populations. Genet. Res. Camb. 30: 247–255.
Dryden, I.L. & Mardia, K.V. 1998. Statistical Shape Analysis. John
Wiley & Sons, Chichester.
Garcı ´a-Bellido, A. & de Celis, J.F. 1992. Developmental genetics
of the venation pattern of Drosophila. Annu. Rev. Genet. 26:
275–304.
Gilchrist, A.S. & Partridge, L. 1999. A comparison of the genetic
basis of wing size divergence in three parallel body size clines
of Drosophila melanogaster. Genetics 153: 1775–1787.
Gilchrist, A.S. & Partridge, L. 2001. The contrasting genetic
architecture of wing size and shape in Drosophila melanogaster.
Heredity 86: 144–152.
Gilchrist, A.S., Azevedo, R.B.R., Partridge, L. & O’Higgins, P.
2000. Adaptation and constrain in the evolution of Drosophila
melanogaster wing shape. Evol. Devel. 2: 114–124.
Gilchrist, G.W., Huey, R.B. & Serra, L. 2001. Rapid evolution of
wing size clines in Drosophila subobscura. Genetica 112–113:
273–286.
Goodall, C. 1991. Procrustes methods in the statistical analysis of
shape. J. R. Stat. Soc. B 53: 285–339.
Guo, P.Z., Mueller, L.D. & Ayala, F.J. 1991. Evolution of
behavior by density-dependent selection. Proc. Natl. Acad. Sci.
USA 88: 10905–10906.
Hendry, A. & Kinnison, M.T. 1999. Perspective: the pace of
modern life: measuring rates of contemporary microevolution.
Evolution 53: 1637–1653.
Hoffmann, A.A. & Shirriffs, J. 2002. Geographic variation for
wing shape in Drosophila serrata. Evolution 56: 1068–1073.
Houghton, J. T., Ding, Y., Griggs, D. J., Noguer, M., van der
Linden, P. J., Dai, X., Maskell, K. & Johnson, C. A. (eds) 2001.
Climate Change 2001: The Scientific Basis: Contribution of Working
Group I to the Third Assessment Report of the Intergovernmental
Panel on Climate Change. Cambridge University Press, New
York.
Houle, D. & Rowe, L. 2003. Natural selection in a bottle. Am. Nat.
161: 50–67.
Huey, R.B., Partridge, L. & Fowler, K. 1991. Thermal sensitivity
of Drosophila melanogaster responds rapidly to laboratory
natural selection. Evolution 45: 751–756.
Huey, R.B., Gilchrist, G.W., Carlson, M.L., Berrigan, D. & Serra,
L. 2000. Rapid evolution of a geographical cline in size in an
introduced fly. Science 287: 308–309.
Imasheva, A.G., Bubli, O.A., Lazebny, O.E. & Zhivotovsky, L.A.
1995. Geographic differentiation in wing shape in Drosophila
melanogaster. Genetica 96: 303–306.
James, A.C. & Partridge, L. 1995. Thermal evolution of rate of
larval development in Drosophila melanogaster in laboratory
and field populations. J. Evol. Biol. 8: 315–330.
James, A.C., Azevedo, R.B.R. & Partridge, L. 1997. Genetic and
environmental responses to temperature of Drosophila mela-
nogaster from a latitudinal cline. Genetics 146: 881–890.
Jolliffe, I.T. 1986. Principal Component Analysis. Springer-Verlag,
New York.
Joshi, A. & Mueller, L.D. 1988. Evolution of higher feeding rate
in Drosophila due to density-dependent natural selection.
Evolution 42: 1090–1093.
Joshi, A. & Mueller, L.D. 1993. Directional and stabilizing
density-dependent natural selection for pupation height in
Drosophila melanogaster. Evolution 47: 176–184.
Joshi, A. & Mueller, L.D. 1996. Density-dependent natural
selection in Drosophila: trade-offs between larval food acquisi-
tion and utilization. Evol. Ecol. 10: 463–474.
Klingenberg, C.P. 2003. Quantitative genetics of geometric
shape: heritability and the pitfalls of the univariate approach.
Evolution 57: 191–195.
Klingenberg, C.P. & Leamy, L.J. 2001. Quantitative genetics of
geometric shape in the mouse mandible. Evolution 55: 2342–
2352.
Klingenberg, C.P. & McIntyre, G.S. 1998. Geometric morpho-
metrics of developmental instability: analyzing patterns of
fluctuating asymmetry with Procrustes methods. Evolution 52:
1363–1375.
Klingenberg, C.P. & Zaklan, S.D. 2000. Morphological integra-
tion between developmental compartments in the Drosophila
wing. Evolution 54: 1273–1285.
Koske, Th. & Maynard Smith, J. 1954. Genetics and cytology of
Drosophila subobscura. X. The fifth linkage group. J. Genet. 52:
521–541.
Krimbas, C.B. 1993. Drosophila subobscura. Biology, Genetics and
Inversion Polymorphism. Verlag Dr Kovac ˇ, Hamburg.
Krimbas, C.B. & Loukas, M. 1980. The inversion polymorphism
of Drosophila subobscura. Evol. Biol. 12: 163–234.
van’t Land, J., van Putten, P., Zwaan, B., Kamping, A. & van
Delden, W. 1999. Latitudinal variation in wild populations of
Drosophila melanogaster: heritabilities and reaction norms.
J. Evol. Biol. 12: 222–232.
Lankinen, P. & Pinsker, W. 1977. Allozyme constitution of
standard strains of Drosophila subobscura. Experientia 33: 1301.
Manly, B.F.J. 1997. Randomization, Bootstrap and Monte Carlo
Methods in Biology, 2nd edn. Chapman & Hall, London.
Matos, M., Rose, M.R., Rocha Pite ´, M.T., Rego, C. & Avelar, T.
2000. Adaptation to the laboratory environment in Drosophila
subobscura. J. Evol. Biol. 13: 9–19.
Mayr, E. 1942. Systematics and the Origin of Species. Columbia
University Press, New York.
Menozzi, P. & Krimbas, C.B. 1992. The inversion polymorphism
of Drosophila subobscura revisited: synthetic maps of gene
arrangement frequencies and their interpretation. J. Evol. Biol.
5: 625–641.
Mestres, F., Balan ˜a ´, J., Segarra, C., Prevosti, A. & Serra, L. 1992.
Colonization of America by Drosophila subobscura: analysis of
the O5 inversions from Europe and America and their
implications for the colonizing process. Evolution 46: 1564–
1568.
Misra, R.K. & Reeve, E.C.R. 1964. Clines in body dimensions in
populationsofDrosophilasubobscura.Genet.Res.Camb.5:240–256.
Monteiro, L.R., Diniz-Filho, J.A.F., dos Reis, S.F. & Araujo, E.D.
2003. Shape distances in general linear models: are they really
at odds with the goals of morphometrics? A reply to
Klingenberg. Evolution 57: 196–199.
Mousseau, T.A. 1997. Ectotherms follow the converse to
Bergmann’s rule. Evolution 51: 630–632.
Mueller, L.D., Graves, J.L. & Rose, M.R. 1993. Interactions
between density-dependent and age-specific selection in
Drosophila melanogaster. Funct. Ecol. 7: 469–479.
854
M. SANTOS ET AL.
J. EVOL. BIOL. 17 (2004) 841–855 ª 2004 BLACKWELL PUBLISHING LTD

Page 15

Needham, J. 1950. Biochemistry and Morphogenesis, 2nd edn.
Cambridge University Press, Cambridge.
Orengo, D.J. & Prevosti, A. 1996. Temporal changes in
chromosomal polymorphism of Drosophila subobscura related
to climatic changes. Evolution 50: 1346–1350.
Orengo, D.J. & Prevosti, A. 2002. Relationship between
chromosomal polymorphism and wing size in a natural
population of Drosophila subobscura. Genetica 115: 311–318.
Partridge, L. & Coyne, J.A. 1997. Bergmann’s rule in ectotherms:
is it adaptive? Evolution 51: 632–635.
Partridge, L., Barrie, B., Fowler, K. & French, V. 1994a.
Evolution and development of body size and cell size in
Drosophila melanogaster in response to temperature. Evolution
48: 1269–1276.
Partridge, L., Barrie, B., Fowler, K. & French, V. 1994b. Thermal
evolution of pre-adult life-history traits in Drosophila melano-
gaster. J. Evol. Biol. 7: 645–663.
Pegueroles, G., Papaceit, M., Quintana, A., Guille ´n, A., Prevosti,
A. & Serra, L. 1995. An experimental study of evolution in
progress: clines for quantitative traits in colonizing and
Paleartic populations of Drosophila. Evol. Ecol. 9: 453–465.
Pfriem, P. 1983. Latitudinal variation in wing size in Drosophila
subobscura and its dependence on polygenes of chromosome O.
Genetica 61: 221–232.
Powell, J.R. 1974. Temperature related divergence in Drosophila
body size. J. Hered. 65: 257–258.
Prevosti, A. 1955. Geographical variability in quantitative traits
in populations of Drosophila subobscura. Cold Spring Harbor
Symp. Quant. Biol. 20: 294–298.
Prevosti, A., Serra, L., Ribo ´, G., Aguade ´, M., Sagarra, E.,
Monclu ´s, M. & Garcı ´a, M.P. 1985. The colonization of
Drosophila subobscura in Chile. II. Clines in the chromosomal
arrangements. Evolution 39: 838–844.
Prevosti, A., Ribo ´, G., Serra, L., Aguade ´, M., Balan ˜a ´, J., Monclu ´s,
M. & Mestres, F. 1988. Colonization of America by Drosophila
subobscura: experiment in natural populations that supports
the adaptive role of chromosomal-inversion polymorphism.
Proc. Natl. Acad. Sci. USA 85: 5597–5600.
Robertson, F.W. & Reeve, E.C.R. 1952. Studies in quantitative
inheritance. I. The effect of selection of wing and thorax
length in Drosophila melanogaster. J. Genet. 50: 414–448.
Rodrı ´guez-Trelles, F. & Rodrı ´guez, M.A. 1998. Rapid micro-
evolution and loss of chromosomal diversity in Drosophila in
response to climate warming. Evol. Ecol. 12: 829–838.
Rodrı ´guez-Trelles, F., Alvarez, G. & Zapata, C. 1996. Time-series
analysis of seasonal changes of the O inversion polymorphism
of Drosophila subobscura. Genetics 142: 179–187.
Rohlf, F.J. 1990. Rotational fit (Procrustes) methods. In:
Proceedings of the Michigan Morphometrics Workshop (F. J. Rohlf
& F. L. Bookstein, eds), pp. 227–236. Special Publication No. 2.
University of Michigan Museum of Zoology, Ann Arbor, MI.
Rohlf, F.J. 1999. Shape statistics: Procrustes superimpositions
and tangent spaces. J. Classif. 16: 197–223.
Rohlf, F.J. 2000. Statistical power comparisons among alternat-
ive morphometric methods. Am. J. Phys. Anthropol. 111: 463–
478.
Rohlf, F.J. 2003. Bias and error in estimates of mean shape in
morphometrics. J. Hum. Evol. 44: 665–683.
Rohlf, F.J. & Slice, D. 1990. Extensions of the Procrustes method
for the optimal superimposition of landmarks. Syst. Zool. 39:
40–59.
Roper, C., Pignatelli, P. & Partridge, L. 1996. Evolutionary
responses of Drosophila melanogaster life history to differences
in larval density. J. Evol. Biol. 9: 609–622.
Santos, M. 1996. Apparent directional selection of body size in
Drosophila buzzatii: larval crowding and male mating success.
Evolution 50: 2530–2535.
Santos, M., Borash, D.J., Joshi, A., Bounlutay, N. & Mueller,
L.D. 1997. Density-dependent natural selection in Drosophila:
evolution of growth rate and body size. Evolution 51: 420–432.
Slice, D.E., Bookstein, F.L., Marcus, L.F. & Rohlf, F.J. 1996.
A glossary for geometric morphometrics. In: Advances in
Morphometrics (L. F. Marcus, M. Corti, A. Loy, G. J. P. Naylor
& D. E. Slice, eds), pp. 531–551. Plenum Press, New York
(URL: http://life.bio.sunysb.edu/morph).
Sokal, R.R. & Rohlf, F.J. 1981. Biometry, 2nd edn. Freeman, New
York.
Sole ´, E., Balanya `, J., Sperlich, D. & Serra, L. 2002. Long-term
changes in the chromosomal inversion polymorphism of
Drosophila subobscura. I. Mediterranean populations from
southwestern Europe. Evolution 56: 830–835.
SPSS Inc. 2001. SPSS for Windows (URL: http://www.spss.com).
Stalker, H.D. & Carson, H.L. 1947. Morphological variation in
natural populations of Drosophila robusta Sturtevant. Evolution
1: 237–248.
StatSoft, Inc. 2003. STATISTICA (Data Analysis Software System),
version 6. (URL: http://www.statsoft.com).
Stevenson, R.D. 1985. Body size and limits to daily range of body
temperature in terrestrial ectotherms. Am. Nat. 125: 102–117.
The MathWorks, Inc. 1998, 2002. Matlab, V.5 and V.6. The
Language of Technical Computing (URL: http://www.math-
works.com).
Van Voorhies, W.A. 1997. On the adaptive nature of Bergmann
size clines: a replay to Mousseau, Partridge and Coyne.
Evolution 51: 635–640.
Walker, J.A. 2000. Ability of geometric morphometric methods
to estimate a known covariance matrix. Syst. Biol. 49: 686–696.
Weber, K.E. 1990. Artificial selection on wing allometry in
Drosophila melanogaster. Genetics 126: 975–989.
Weber, K.E. 1992. How small are the smallest selectable domains
of form? Genetics 130: 345–353.
Weber, K., Eisman, R., Morey, L., Patty, A., Sparks, J., Tausek,
M. & Zeng, Z.-B. 1999. An analysis of polygenes affecting wing
shape on chromosome 3 in Drosophila melanogaster. Genetics
153: 773–786.
Weber, K., Eisman, R., Higgins, S., Morey, L., Patty, A., Tausek,
M. & Zeng, Z.-B. 2001. An analysis of polygenes affecting wing
shape on chromosome 2 in Drosophila melanogaster. Genetics
159: 1045–1057.
Zimmerman, E., Palsson, A. & Gibson, G. 2000. Quantitative
trait loci affecting components of wing shape in Drosophila
melanogaster. Genetics 155: 671–683.
Received 21 March 2003; revised 21 January 2004; accepted 23 January
2004
Thermal evolution in Drosophila subobscura
855
J. EVOL. BIOL. 17 (2004) 841–855 ª 2004 BLACKWELL PUBLISHING LTD