WING BEAT FREQUENCIES IN DROSOPHILA
MELANOGASTER SELECTED FOR DIFFERENT WING
ESTELLE J.K. NOACH*, HERMAN VAN DER KLIS, GERDIEN DE JONG
and WILLEM SCHARLOO
Department of Botanical Ecology
and Evolutionary Biology,
Padualaan 8, 3584 CH Utrecht, The Netherlands)
Wing beat frequencies
of w.b.f. were
gaster lines selected for Long and Short wings at two different temperatures. Flies were
glued with their thorax to a thin thread and wing beat frequency
by use of a
stroboscope. Flights were performed at two different ambient or Flight temperatures. Wing
length and wing area were measured afterwards.
Wing beat frequency increased with increasing temperature except in the Long line selected
at 20°C: most lines showed phenotypic plasticity in w.b.f. In the lines selected at 25°C the
Short line had a significantly
higher w.b.f. than the Control and Long line at both Flight
temperatures. The genetic differences in w.b.f. between the lines remained constant. Lines
selected at 20°C showed no differences in w.b.f. at both Flight temperatures. Significant
interactions were observed.
Various hypotheses are posed for the adaptive significance
of changes in w.b.f. due to tem-
perature, in terms of wing load, flight muscles and optimal flight temperatures.
wing beat frequency, wing length, Drosophila, temperature.
Phenotypic plasticity (i.e., the ability of an individual organism to alter its
physiology/morphology in response to changes in environmental conditions
(SCHLICHTING, 1986)) has recently been extensively studied in Drosophila
(e.g., SCHEINER & LYMAN, 1989, 1991; DAVID et al., 1994; BARKER &
KREBS, 1995; NOACH et al., 1996). Most studies have explored morpho-
logical characters. To get insight into the plasticity of wing length, NOACH
et al. (1997) performed artificial selection on wing length in Drosophila
melanogaster at different temperatures, and examined correlated responses
in other morphological characters (thorax length, and cell size and cell
number in the wing). In addition, they studied correlated responses in
physiological traits, e.g., locomotor activity (NOACH et al., submitted). In
the present paper, results of experiments will be presented concerning wing
beat frequency (also a physiological trait) and its plasticity at different am-
bient temperatures in the same selection lines as used in the locomotor
The relationship between wing length and wing beat frequency (w.b.f.) in
Drosophila has not yet fully been studied. REED et al. ( 1942) observed that
different species had a characteristic range of w.b.f.'s, and that w.b.f. was a
heritable physiological trait. Other studies explored flight duration and w.b.f.
(GRAVES et al., 1988), course-control, metabolism and wing interference
during tethered flight (goth, 1987), and causes of variation in wing loading
among different Drosophila species (STARMER & WOLF, 1989).
The adaptive significance of different w.b.f.'s in relation to body size re-
mains open to investigation. REED et al. (1942) observed a positive relation
between w.b.f. and temperature of origin of different Drosophila species
(habitat temperature), and also a positive relation between w.b.f. and ambi-
ent temperature within Drosophila species, as did UNWIN & CORBET (1984).
Furthermore, STALKER & CARSON (1949) and STARMER & WOLF (1989)
suggested that the adaptive significance of a lower thorax/wing-ratio in
cooler environments would allow a lower w.b.f. and would improve fly-
ing at cooler temperatures. However, these findings were all done either
when different Drosophila species were compared or when one species was
reared at different temperatures. In this paper, the relationship between an
adaptive mechanism for wing beat frequency, body size, and temperature is
discussed by using lines selected for different wing lengths.
MATERIAL AND METHODS
Stocks and rearing
In a French suburban population of Drosophila melanogaster (collected in
October 1990) directional selection for Long and Short wings has been per-
formed at 20°C and 25°C separately, as described in NOACH et al. (1997). At
both Selection Temperatures an unselected Control line was reared. All flies
were reared on a Mittler-Bennett medium (MITTLER & BENNETT, 1962).
Wing beat frequency
Wing beat frequency was tested at two different temperatures, 20°C and
25°C. These temperatures will be referred to as 'Flight Temperatures'. All
flies were reared at the temperature they were selected in. Wing beat fre-
quencies at Flight Temperature 20°C were tested in generation 42 of the
lines selected and reared at 25°C and generation 29 of the lines selected
and reared at 20°C. Wing beat frequencies at Flight Temperature 25°C were
tested in generation 44 of the 25°C selection lines and generation 30 of the
20°C selection lines. Flies were allowed to acclimatize to the temperature
at which the experiment was conducted at least 24 h prior to the start of
each flight. Cold-anaesthetized flies were attached individually with their
thorax to a very thin iron thread (0 0.18 mm) with a cyano-acrylate glue
(tethered flight). Only male flies of 4 to 5 days old were used. All three
lines (Long, Short and Control) of one Selection Temperature were tested
simultaneously, with two or three flies of each line in the same session
(max. 12 flies per flight-session). At Flight Temperature 20°C, 30 flies of
each line were tested, whereas 20 flies of each line at Flight Temperature
25°C were observed. Wing beat frequency of each fly was recorded every
five minutes in a period of an hour with a Digital Stroboscope (GcnRad
1546, Strobotac) in strokes/minute. At the most, two minutes were needed
to measure all the flies. After an hour, each fly was released from its thread
and frozen. Wing length (from the middle of the anterior cross-vein until
the end of the 3`d longitudinal vein) and total wing area of each fly were
measured, to show that (wing length)2 is a good indicator for wing area.
For a description of the measurement-methods of wing length see Noach et
al. (1997). To measure total wing area, wings were microscopically exam-
ined (objective 1.Ox, projective 2.0x) and scanned with a Sony b/w CCD
camera, type XC-77CE. The VIDAS image analysis system (Kontron/Zeiss,
Eching, Germany) was used to quantify the wing area.
Computations and statistics
The first six wing beat frequency-measurements (30 minutes) of each in-
dividual were used, since w.b.f. remained very constant during this time.
One-way ANOVA's on w.b.f.'s for each Selection Temperature-Flight Tem-
perature combination revealed the influence of Flight Temperature on w.b.f.
of the lines. Two-way nested ANOVA's with a posteriori Bonferroni tests
on w.b.f.'s (with factors Line and Flight Temperature) per Selection Tem-
perature revealed whether w.b.f.'s of the selected lines differed significantly
from each other and from the unselected Control line. Plasticity of the
w.b.f. (i.e., the difference in w.b.f.'s between Flight Temperatures) in the
wing length selection lines was investigated also with two-way ANOVA's.
Wing lengths and wing areas of the Long, Short and Control line differed
significantly from each other at each Selection Temperature in the gener-
ations in which wing beat frequency (w.b.f.) was tested (fig. 1). This was
confirmed by one-way ANOVA's per Selection Temperature-Flight Tem-
perature combination (tables not shown). W.b.f.'s of flies remained very
constant during the first 30 minutes of a flight (fig. 2).
Since the two subsets of flies (selected and raised at 20°C and selected
and raised at 25°C) are two independent groups, no direct comparisons
between Selection Temperatures can be made. Therefore, ANOVA's have
been performed for each Selection Temperature separately. All lines show
a significantly higher w.b.f. at Flight Temperature 25°C compared to Flight
Temperature 20°C, except for the Long line selected at 20°C (fig. 3; one-way
ANOVA's not shown). First, the lines selected at 25°C will be considered:
no significant difference in w.b.f. between the Long and Control line at
each Flight Temperature is seen in spite of the significant difference in wing
length and (thus) in wing area between those lines. In contrast, the Short line
Fig. 1. Mean wing length (f sem) of male flies of lines selected
on wing length at Selection
20°C and of lines selected at Selection Temperature
25°C in the generations
which wing beat frequency was tested at Flight Temperatures
20°C and 25°C (see Material
and Methods for the test-generations).
Left bar: Long lines; middle bar: Control lines; right
bar: Short lines.
Fig. 2. Example of the wing beat frequencies
of three simultaneously flying flies in the first
30 minutes of a flight-session.
Observations at 5 min intervals until t = 30 min are used in
the nested anova (table I). Lines selected and raised at 25°C. Squares: Long lines; circles:
Control lines; triangles: Short lines.
has a significantly higher w.b.f. than the Control and Long line at both Flight
Temperatures (fig. 3a; table I). No significant Genotype-by-Environment t
interactions in w.b.f.'s are observed when w.b.f.'s are measured at different
ambient temperatures, indicating that plasticity of w.b.f. is similar in all
three lines (table I, Line * Flight), i.e., the slopes of the reaction norms
are similar. A different outcome is obtained in the lines selected at 20°C
(fig. 3b; table I): except for a marginal significant difference in w.b.f.'s
of the Long and Short line at Flight Temperature 20°C (p < 0.05), no
significant differences in w.b.f.'s between the lines are observed at each
Flight Temperature. In contrast to all other lines, w.b.f. of the Long line
(selected at 20°C) is the same at both Flight Temperatures. This is seen in
the significant Genotype-by-Environment interaction of Line * Flight of the
Long line, compared to the Short and Control line (table I). Again, wing
length and wing area of the lines selected at 20°C differ significantly.
It was not possible either to weigh individual flies or to measure thorax
length after a successful flight, because almost always remainders of glue
stayed on the thorax of a fly after it had been released. Therefore, wing load
could not be calculated directly. However, both fresh body weight and wing
Fig. 3. Mean Wing beat frequency
(-b sem) of male flies tested at Flight Temperatures
and 25°C. a. Lines selected on wing length at 25°C; b. Lines selected on wing length at
20°C. Squares: Long lines; circles: Control lines; triangles: Short lines.
length, but not wing area, were registered during the selection procedure
in roughly the same generations as in which the wing beat experiment was
performed (generation 30 in the lines selected at 20°C; generation 44 in the
lines selected at 25°C). With these data an estimate of wing load for all lines
is made, using: wing load = fresh body weight/wing area or wing load =
fresh body weight/(wing length)2 (table II). To check whether (wing length)2
Two-way nested ANOVA's
per Selection Temperature,
with factors Line and Flight Tem-
perature on wing beat frequencies (w.b.f.). The first 5 observations
of each individual (30
minutes of flight) did not differ significantly. Indented characters refer to the a posterior
Bonferroni tests. Wing length selection lines were tested at Flight Temperatures
25°C. Lines were selected at 20°C and at 25°C.
n.s.: not significant (p > 0.05); *: p < 0.05; **: p < 0.01; ***: p < 0.001.
Estimates of wing load of male flies in all lines. Wing load is calculated as (body weight)/
(wing length)2. Body weight and wing length are measured in flies of both selection tem-
peratures in the same generation as the flies used in the w.b.f.
but in individuals
other than those used in the flight-performance experiments.
indeed was a good indicator for wing area, both wing area and wing length
of the test flies were measured. The overall correlation coefficient between
wing area and (wing length)2 was 0.95, showing that (wing length)2 is a
good indicator for wing area. At both Selection Temperatures, wing load
is highest in the Short lines, followed by the Control lines and lowest in
the Long lines: a fly selected for small wings has a higher body weight
relative to wing size than a control fly or fly selected for longer wings.
In the lines selected at 25°C, a linear relationship between wing load and
Fig. 4. Relationship
wing beat frequency
and wing load (= fresh body weight/(wing
length)2) of male flies at Flight Temperatures
20°C and 25°C. a. Lines selected on wing
length at 25°C; b. Lines selected on wing length at 20°C. Open symbols: Flight Temperature
20°C; Closed symbols: Flight Temperature
wing-beat frequency is found when plotting these two characters for each
Selection Temperature and for each Flight Temperature (fig. 4): a short-
winged fly needs to beat its wings faster to remain in tethered flight than a
fly with non-selected or long-selected wings. In the lines selected at 20°C
a similar relationship exists at Flight Temperature 25°C. However, at Flight
Temperature 20°C an unexpectedly lower w.b.f. is observed in the Short and
Control line as compared to the Long line.
Wing beat frequency and temperature
Wing beat frequency and its plasticity have been investigated at two dif-
ferent ambient or Flight Temperatures (20°C and 25°C) in Drosophila
melanogaster lines selected for different wing lengths. Except for the Long
line selected at 20°C, wing beat frequency (w.b.f.) became significantly
higher at the higher temperature: w.b.f. proved to be a phenotypically plas-
tic character. A similar result was observed in other flies (Calliphora and
Musca, see UNWIN & CORBET, 1984). In an early study of REED et al.
(1942) it was already observed that, in general, higher wing beat frequen-
cies were found for different Drosophila species living in warmer habitats,
and, within species, at higher ambient temperatures.
Wing beat frequency and wing load
In our lines selected at 25°C, the short-winged flies had a significantly
higher w.b.f. than those selected for long wings. By estimating the wing
load of the various lines, it became clear that the short-winged flies had
a higher wing load than the long-winged ones. This held true for flies of
each Selection Temperature. Flies selected for short wings had a higher
body weight relative to wing size than control or long-winged individuals
BYRNE et al. (1988) stated that a positive relation between wing load and
w.b.f. was observed in insects except for those weighing less than 0.03 g.
They suggested that very small insects use different strategies for flying
than large insects, namely the 'clap-and-fling' method (WEIS-FOGH, 1973)
or 'squeeze-and-peel' method (ZANKER, 1990) with which an extra lift is
generated. However, in our experiment with male Drosophila melanogaster
varying in wing length and weighing less than 0.03 g, the positive rela-
tion between wing load and w.b.f. was still present in three out of four
comparisons (fig. 4). This is supportcd by data of DAVID et al. (1994),
who raised male Drosophila melanogaster at increasing temperatures and
showed a correspondingly higher wing load and w.b.f.
Wing beat frequency and genetic variation
In five out of our six plasticity measurements, w.b.f. was significantly higher
at the higher Flight Temperature (25°C). In these five measurements, the
reaction norms were parallel and no significant genotype-by-environment
interactions were observed. When testing two different Drosophila species
over a temperature range, REED et al. (1942) found that the shapes and
slopes of the two w.b.f.-curves were highly similar, but differed in height.
They concluded that the difference in w.b.f.'s between the two species,
measured at one test temperature, was not due to a 'physiological differ-
ence'. However, the fact that w.b.f. within species increases with increasing
ambient temperature might be due to a better muscle efficiency at higher
temperatures, and thus to a physiological difference. Wing beat frequency
measuremcnts of our flies, all of the same species and originating from the
same base population, agree with these findings, except for the Long line
selected at 20°C. Artificial selection for wing length caused genetic dif-
ferences in w.b.f.'s between the lines (different heights), but this variation
remained constant between the lines over a range of temperatures. Within
each line a similar energetic constraint might be present, resulting in a re-
action norm of w.b.f. and ambient flight temperature with an optimal slope
(see next paragraph).
Adaptive significance of wing beat frequency
REED et al. (1942) assumed that the energy output per wing stroke should be
proportional to the volume of the thoracic flight muscles. By plotting certain
thorax-wing dimensions of several Drosophila species, which varied in body
size, against the corresponding w.b.f.'s, REED et al. sought an optimal, i.e.,
adaptive reaction norm of these two characters in different species. In
our data, the two Control lines, each flying at the temperature at which
they were raised, are assumed to be best adapted to their own laboratory
environment. They have diverged in size and wing load: the thorax length of
both Control lines are approximately the same, whereas the wing length of
the 20°C Control line is larger compared to the 25°C Control line. A graph
of these Control data, similar to that of REED et al. (1942) (i.e., plotting in
one graph the w.b.f. of the 20°C Control line at Flight temperature 20°C,
and the w.b.f. of the 25°C Control linc at Flight temperature 25°C), is
assumed to have an optimal slope. In view of our selection lines, two trains
of thought could be pursued: one in which a relation between an, for the
laboratory, optimal thorax/wing ratio and optimal wing beat frequency is
assumed; the other considers the relation between flight muscle volume and
wing beat frequency.
For the first supposition, it should be checked whether selection on wing
length has caused a proportional correlated response in thorax length. A
low correlated response in thorax length could result in a suboptimal w.b.f.,
because an optimal thorax to wing length ratio, as assumed to exist in the
Control lines, is no longer present. The w.b.f. of a line with a thorax/wing
ratio closest to the Control line should approach the w.b.f. of the Control
line. A line with a highly deviating thorax/wing ratio should result in a
w.b.f. deviating considerably from the Control line. This reasoning will
now be applied to our data: the thorax length to wing length ratio in the
selection lines has been calculated in the same generation in which the
w.b.f. experiment has been performed. At Selection Temperature 25°C,
both the deviations of this ratio in the selected lines (Long and Short) and
the Control line were equal (thorax length/wing length ratio for Long: 0.63;
Control: 0.68; Short: 0.74). In the lines selected at 20°C, the deviation
of the ratio in the Long line from the Control was larger than that in the
Short line from the Control (thorax length/wing length ratio Long: 0.57;
Control: 0.64; Short: 0.65; Deviation Control-Long = 0.063; Deviation
Short-Control = 0.016). Now, the thorax/wing ratios and the corresponding
wing beat frequencies can be compared. In the lines selected and flown at
25°C the w.b.f. of the Short line deviates more from the Control than the
w.b.f. of the Long line. This is not as was expected on the basis of the
thorax/wing ratio. At Selection and Flight Temperature 20°C the w.b.f. of
the Long line deviates somewhat more from the Control than the w.b.f. of the
Short line, as was expected (fig. 3). It has to be concluded that the w.b.f.'s
are not unequivocally dependent on the ratio of thorax to wing length: a
change in the thorax/wing ratio often, but not always automatically, leads
to a change in w.b.f.
The second train of thought involves the flight muscles. A positive relation
between thorax volume and flight muscle volume can be expected (REED et
al., 1942; DAVID et al., 1994). As the flight muscles are situated within the
thorax, the flight-muscle volume could be related to a correlated response of
thorax length to selection on wing length. It is not known what minimal or
maximal muscle size in relation to body size is necessary in order to make
flying possible. Preliminary data of DAVID et al. (1994) perhaps suggest that
morphological changes in wing length and thorax length bring about a more
efficient use of the flight muscles as well as an increased wing load due to
an increase in rearing temperature. If muscle volume shows a correlated
adaptive response to selection on wing length, a similar response in w.b.f.
will be expected, and the relation between wing load and w.b.f. might well
remain the same. Howcver, if wing length changes independently from
muscle volume, a different w.b.f. response may be expected; REED et al.
(1942) suggested that the line with the smallest wings has the highest w.b.f.,
indicating perhaps more muscle power than necessary. Our data show that
the w.b.f. in the Short line selected at 25"C, has increased relatively to
the Control line. According to the reasoning by REED et al., a change in
wing length independently from thorax or muscle volume should then have
occurred. In our data, the Short line is the only one in which an independent,
non-adaptive change in wing length from muscle volume has occurred. It
might be suggested that in the others an adaptive muscle change might have
occurred. This second train of thought leads to the conclusion that in some
cases an independent change is found, whereas in others a change in wing
length is accompanied by a change in flight muscle, or thorax volume.
To conclude our ideas on the adaptive significance of wing beat frcqucncy,
the implications made by STARMER & WOLF (1989) will be considered:
they suggested from their results in tcn different Drosophila species, that
smaller wing size and higher wing load could be selectively advantageous
if manoeuvrability, shorter flights, and perhaps morc hovering are impor-
tant at warmer temperatures. However, this interesting assumption does not
contribute to our understanding of how the mechanisms by which changes
in w.b.f. occur, arc brought about. Since no differences in genotypc-by-
environment interaction are observed, one might think that w.b.f. is con-
strained by muscle volumc and wing length or wing load.
By performing integrated physiological and morphological studies, more
information of size of flight muscles, wing beat frequency and their rela-
tionship with temperature should become available to understand thc rela-
tionship between body size and flight mechanism in small insects. This will
be necessary to give a unequivocal opinion upon the adaptive significance
of the wing beat frequency-reaction norms.
The authors would like to thank Dr. Bcrd Bruins for teaching us the practical
procedure of the experiment, Dick Smit for drawing the figurcs, Harry van
der Wildt for writing the computer program for automatic data collection
of the stroboscope measurements, and Dr. Maarten Terlou for his technical
assistance with the image analyzer.
J.S.F. & R.A. KREBS,
1995. Genetic variation and plasticity of thorax length and
wing length in Drosophila aldrichi and D. buzzatii.
J. Evol. Biol. 8: 689-709.
BYRNE, D.N., S.L. BUCHMANN
& H.G. SPANGLER,
wing beat frequency
and body mass in homopterous
insects. J. Exp. Biol. 135: 9-23.
DAVID, J.R., B. MORETEAU,
A. STOCKEL & A.G. IMASHEVA,
1994. Reaction norms of size characters in relation to growth temperature
melanogaster: an isofemale
line analysis. Genet. Sel. Evol. 26: 229-251.
GÖTZ, K.G., 1987. Course-control,
metabolism and wing interference
during ultra long teth-
ered flight in Drosophila melanogaster. J. Exp. Biol. 128: 35-46.
GRAVES, J.L., L.S. LUCKINBILL & A. NICHOLS,
Flight duration and wing beat frequency
in long- and short lived Drosophila melanogaster. J. Insect Phys. 34: 1021-1026.
S. & J. BENNETT,
1962. A simple food medium that requires no live yeast with a
minimum of variables. Dros. Inf. Serv. 36: 131-132.
E.J.K., G. DE JONG & W. SCHARLOO,
1996. Phenotypic plasticity in morphological
traits in two populations
of Drosophila melanogaster.
J. Evol. Biol. 9: 831-844.
NOACH, E.J.K., G. DE JONG
& W. SCHARLOO,
of wings in selection
lines of Drosophila melanogaster. Heredity 78 (in press).
NOACH, E.J.K., G. DE
& W. SCHARLOO,
in Drosophila melanogaster
selected for different wing lengths. (submitted).
REED, S.C., C.M. WILLIAMS & L.E. CHADWICK,
as a character
for separating species, races and geographic varieties of Drosophila. Genetics 27: 349-
S.M. & R.F. LYMAN,
1989. The genetics of phenotypic plasticity.
J. Evol. Biol. 2: 95-107.
S.M. & R.F. LYMAN,
1991. The genetics of phenotypic
II. Response to
selection. J. Evol. Biol. 4: 23-50.
SCHLICHTING, C.D., 1986. The evolution of phenotypic plasticity in plants. Ann. Rev. Ecol.
Syst. 17: 667-693.
H.D. & H.L. CARSON,
1949. Seasonal variation in the morphology
robusta Sturtevant. Evolution 3: 330-343.
W.T. & L.L. WOLF,
1989. Causes of variation
in wing loading among Drosophila
species. Biol. J. Linn. Soc. 37: 247-261.
D.M. & S.A. CORBET,
Wingbeat frequency, temperature
and body size in bees
and flies. Phys. Entomol. 9: 115-121.
WEIS-FOGH, T., 1973. Quick estimates of flight fitness in hummingbirds
J. Exp. Biol. 56: 79-104.
ZANKER, J.M., 1990. The wing beat of Drosophila melanogaster. I. Kinematics. Phil. Trans.
R. Soc. Lond. B. 327: 1-18.