Behavioral Ecology Vol. 12 No. 4: 513-515
© 2001 International Society for Behavioral Ecology
Forum |
Flight, fitness, and sexual selection: a response
Department of Biological and Molecular Sciences, University of Stirling, Stirling FK9 4LA, UK
Address correspondence to K.L. Buchanan. E-mail: buchanankll{at}cf.ac.uk .
Received 6 October 2000; revised 6 November 2000; accepted 11 November 2000.
We are disappointed to find that Møller and Barbosa
(2001
) view our recent
reporting of a naturally selected portion of the tail streamer of the barn
swallow as unconvincing. They raise four points, which we will deal with in
turn.
The lack of standardization of environmental conditions is an unfortunate consequence of conducting this kind of work in the field. The general approach to this problem in the ecological literature is to use random allocation of birds to experimental groups. This makes it unlikely that different groups vary systematically in environmental conditions and would mean that significant results were harder to obtain with differing environmental conditions inflating the error term in any analysis. Møller and Barbosa raise two particular issues—nest site geography and breeding effort.
We agree that the geography of the nest site may have some influence on the
flight trajectory and this is why the farm was entered (as a categorical
variable) into the analyses, to control for variation in the building
structure at the nest site. The farm factor explained significant variance in
three variables (Buchanan and Evans,
2000: Table 4). We therefore feel that we have controlled for any
variance due to this factor. We would also point out that all the flights we
recorded were outside buildings and were therefore not constrained by internal
structures within the buildings.
Møller and Barbosa also argue that the manipulations should have been standardized both to the breeding attempt and to chick age. Unfortunately we did not record breeding variables during our study as we focussed entirely on the flight of the birds. We reiterate that birds were randomly allocated into experimental groups, therefore any differences in breeding activity between birds would increase the variance between them making significant results harder, rather than easier to obtain. Furthermore, it should be noted (as stated in the Methods) that date of the assessment of flight performance was added into all models during the analysis and was eliminated each time, as it was not found to explain significant variation in individual flight performance. We would also point out that we calculated repeatability estimates for all our flight variables, and these suggested that the between individual variation in flight performance was significantly greater than the within individual variation. This result is unlikely if stage of breeding influenced flight performance.
Møller and Barbosa note a "lack of consistency" between
the results reported in Evans
(1998) and those reported in
Buchanan and Evans (2000), when
referring to the effect a 20 mm reduction has on flight performance. There are
two important points to note here. First, the nature of the manipulations
differed between the two studies; in the earlier study two types of
manipulation were used and for both types the original feather was cut and
replaced with an appropriate length of feather, following the example set by
Møller's work. However, in Buchanan and Evans
(2000) the original streamers
were reduced in length by trimming the tip without cutting the feather. The
difference in the nature of these manipulations will have produced
fundamentally different effects on flight performance, but does not negate the
conclusions of either study. One of the reasons Buchanan and Evans
(2000) used their manipulation
method was because Evans (1998)
had clearly demonstrated that the type of manipulation carried out had a
strong effect on the results obtained and that cutting a tail feather at the
base is likely to have fundamental effects on its aeordynamic properties. We
would strongly argue that the manipulations carried out by Buchanan and Evans
(2000) should be interpreted as
having greater biological relevance to the effects of tail length on flight
performance in the barn swallow, as in this study the integrity of the feather
remained complete. The effect of cutting the feather has been seen in a more
recent experiment (Evans MR, Cherry MI and Dowse A, unpublished data). For
most flight variables cutting the feather results in a stepwise change in
performance. For example, the mean flight velocity of sunbirds (Nectarinia
violacea) with cut tail feathers was just over 3 m/s faster than that of
sunbirds with uncut feathers of the same length. We would suggest that, where
possible, cutting tail feathers should be avoided as it disrupts the
aeroelastic properties of the feather. However, sometimes it is unavoidable,
for example when conducting tail elongation manipulations.
Second, Buchanan and Evans
(2000) pointed out in some
detail that original streamer length greatly affects the way in which tail
reductions affect flight performance. Figure 2 in Buchanan and Evans
(2000) demonstrates that for
both mean velocity and maximum agility the direction of the change between the
control treatment and the 20 mm reduction is dependent on the original tail
length of the birds within the treatment groups.
We would also like to correct three factual errors made by Møller
and Barbosa. It should be noted that while Evans
(1998) reports an effect on
mean agility, Buchanan and Evans
(2000) refer to maximum
agility—as such, the results of the two studies cannot be compared. The
mean agility of birds in Buchanan and Evans
(2000) is 369 ± 12.6
degrees/s, which is very similar to that in Evans
(1998). Møller and
Barbosa also suggest that in Evans
(1998) birds with shortened
tails had a mean velocity 10 m/s greater than controls; the real difference is
8.3 m/s (Evans 1998) which
compares with a difference of +1.3 m/s for a male with a 120 mm streamer and
-1.0 m/s for a male with a 100 mm streamer. Finally, Evans
(1998) did not conduct tip
manipulations, as suggested by Møller and Barbosa. Two manipulations
types were used by Evans
(1998); feather was added or
removed from the basal portion of the tail feather or from the streamer, but
in both cases the feather was cut and rejoined. If this technique had not been
used the streamer and basal manipulation types could not have been
compared.
We would also like to stress the difference in between the manipulation
technique that we have used and that adopted by Møller and his
co-workers. We have trimmed the feather while leaving the original feather
intact—allowing it to move appropriately. Møller has adopted a
technique where the entire feather is cut and rejoined, but this is likely to
have negative effects on the aeroelastic properties of the feather. The other
unfortunate consequence of Møller's manipulation technique is that it
changes the size of the non-sexually dimorphic basal part of the feather while
leaving the sexually dimorphic streamer unaffected. We have adopted a
technique that allows us to manipulate the streamer itself, rather than the
basal part of the feather which is not of interest. Evans
(1998) found that the type of
manipulation used had large effects on the size and shape of the relation
between manipulation and flight. This result has recently been confirmed using
doubly labelled water to measure daily energy expenditure. There is little
effect of manipulation when basal manipulations are conducted
(Cuervo et al., 1996; Hall A,
personal communication) while streamer manipulations show pronounced U-shaped
relationships (Hall A, personal communication).
Møller and Barbosa suggest that as Buchanan and Evans
(2000) did not report any
fitness benefits from the aerodynamic parameters examined, that none exist. We
would suggest that this is a rather hasty interpretation of our results. The
concept of fitness is defined as "the average per capita lifetime
contribution of individuals of that genotype to the population after one or
more generations" (Futuyma,
1998). As such, it is probably impossible to measure fitness
directly. The majority of field studies tend to assess surrogate measures of
fitness for example, the length of the pre-mating period, percent of second
clutches, and annual reproductive output
(Møller, 1988). None of
these variables measure fitness directly, although it is assumed that they
vary with fitness. Just as Møller and co-workers have used surrogate
fitness measures for the benefits of streamer possession, we were faced with
identifying surrogate fitness measures for the costs of streamer possession.
Certainly it is true that our study did not attempt to demonstrate that
certain flight performance characteristics result in for example, higher
reproductive success for individuals. But we would suggest that flight
performance is indisputably linked to foraging success, which in turn is
likely to have fitness consequences. On a conceptual note it is difficult to
see how any experiment to examine the fitness consequences of tail
manipulation could be conducted. We assume (under the adaptationist paradigm)
that any trait is at its optimum at its current level. So we would assume that
swallow streamer length was currently at its equilibrium optimal size for each
individual. Therefore, any manipulations are ultimately and inevitably going
to have negative effects on fitness. We would like to point out that the
theoretical framework for interpreting changes in flight performance with
changes in tail length is covered in detail in Evans and Thomas
(1997), and this framework
assumes that deviations from the optimal tail length would be deleterious in
terms of fitness. Møller
(1988) demonstrated that
streamer elongations improved annual reproductive success. This did not mean
that swallows were somehow sub-optimal and should be producing longer
streamers. Møller (1989)
showed that males given long streamers incurred costs in following
seasons—this was taken as evidence that streamers were costly. The
negative effect of elongated streamers on the following season could be
conceptualized as balancing their positive effect on the current season
producing an optimal overall streamer length. Therefore, any experiment that
attempted to examine the effect of tail manipulations on fitness would be
doomed to failure—negative fitness effects would inevitably be detected
with any deviation from the observed value. It is necessary to break down the
selective pressures on trait size, to quantify the selection leading to
optimal size for each individual. While Møller and his co-workers have
focussed on the surrogate measures of the benefits of streamer possession, we
have tried to examine surrogate measures of the costs—and we would argue
that both are needed for a full understanding of the evolution of the trait.
Our results should be regarded as drawing attention to the fact that earlier
work has treated aerodynamics as a black box. We feel it is reasonable to
assume that differences in for example, the prey caught by swallows with
different tail manipulations occur because of changes in flight rather than
different food preferences of birds with different tail manipulations.
Additionally, these experiments allowed us to examine the extent to which
sexual selection was responsible for streamer elongation, a question that came
to prominence with the publication of a plausible mechanism for streamer
elongation under natural selection
(Norberg, 1994). Examination
of variables like annual reproductive output would be unable to distinguish
natural and sexual selection, as both hypotheses would predict increased
success with increased streamer length. In order to determine the extent of
sexual selection and natural selection the costs have to be examined. Our data
strongly suggest that the bulk of the streamer has been produced by natural
selection but that sexual selection has extended the streamer beyond the
aerodynamic optimum by c15 mm.
Møller and Barbosa suggest that we have been attempting to demonstrate that males benefit from a long tail streamer, when in fact this is an over-simplification of the case. Instead we would argue that our results show that males can benefit from a long tail streamer, depending on their original morphology, but that from an aerodynamic point of view streamers should all be shorter than current lengths. The central aim of our study was, therefore, to demonstrate that both natural and sexual selection have been important in shaping the tail streamer of the barn swallow—not to investigate fitness benefits.
Møller and Barbosa suggest that variation in the time between tail
manipulation and the assessment of flight performance could cause major error
in the results due to differences in the degree of habituation. We would
suggest that if this were the case we would not have found significant
repeatabilities (e.g., mean velocity 73.6%, p <.05) for the flight
parameters measured on different days post-manipulation. However, in an effort
to address this issue directly we have reanalyzed the results reported in
Buchanan and Evans (2000)
including the time between manipulation and filming in each of the glm models
determining the effects of manipulation on flight performance. We can confirm
that the time between manipulation and filming was not a significant predictor
of flight performance for any of the flight variables (mean velocity,
F1,34 = 0.76, p =.39; maximum curvature,
F1,26 = 0.14, p =.714; maximum agility,
F1,29 = 3.82, p =.06; mean xy rate change,
F1,26 = 0.02, p =.887; mean acceleration,
F1,33 = 3.31, p =.078; PC1,
F1,32 = 0.07, p =.791; PC2,
F1,44 = 2.21, p =.144). Furthermore, including
time between manipulation and filming did not significantly affect any of the
model variables for the results originally reported in Buchanan and Evans
(2000).
Møller and Barbosa go on to address the optimal flight strategies of
birds with different tail lengths, but unfortunately have misinterpreted
Figure 3 of Buchanan and Evans
(2000). The issue of how
original tail length contributes to the effect of tail manipulation on flight
is a complex one, but is dealt with in detail in Buchanan and Evans
(2000). To take velocity as an
example, it is clear that not all birds have the same optimal flight speed. As
the sexually selected portion of the tail is removed and the tail is reduced
towards its aerodynamic optimum, birds with long tail streamers increase their
velocity, whilst birds with short tail streamers decrease their velocity
(Buchanan and Evans, 2000:
Figure 3). Møller and Barbosa suggest flying faster is beneficial while
flying slower is costly—or vice versa. However, the situation is
unlikely to be this simple: covariation of morphological characteristics means
that individual birds are adapted to fly at a variety of optimal speeds,
depending on characteristics such as weight, wing length, and tail fork
depth.
Møller and Barbosa are also concerned as to the biological relevance
of the flights filmed. Certainly it is true that the birds were not filmed in
foraging flight, but observation confirms that swallows have similar flight
characteristics during foraging flights and when visiting the nest site
(Buchanan et al., unpublished data). We are currently quantifying individual
flight characteristics of swallows filmed in foraging flight. However, in
defense of our measurements of flight, we find it unlikely that
"provisioning nestlings is biologically and aerodynamically
irrelevant" as it has been demonstrated that the efficiency of parental
provisioning is important for nestling success
(de Lope and Møller,
1993) and therefore probably also for individual fitness. We also
suggest that if our results were indeed irrelevant then it would be unlikely
that we would have found similar results by using stereo-video of mobbing
flight (Allombert S, unpublished data) and in flight mazes
(Rowe et al., 2001).
We have now investigated the selection pressures acting on tail morphology
in a variety of hirundine species, and obtained highly consistent results
(Buchanan and Evans, 2000;
Park et al., 2000;
Rowe et al., 2001). In
particular, a recent study examining the effects of streamer manipulation in
barn swallows on flight time through large flight maze has confirmed the
U-shaped relationship between tail manipulation and flight performance. This
provides additional evidence for the involvement of both natural and sexual
selection, in streamer evolution (Rowe et
al., 2001). Park et al.
(2000) also found that the
addition of small streamers to house martins (Delichon urbica)
resulted in an increase in manoeuvrability during turning flight. Overall,
these results suggest that tail streamers are predominantly naturally selected
structures improving maneuverability but have been exaggerated by sexual
selection in barn swallows resulting in a c15 mm extension beyond the
aerodynamic optimum.
In conclusion we would argue that while our data may contain error, it is
not systematically biased. Møller and Barbosa have some interesting
ideas as to the minimization of any errors included, but this would only
increase the strength of the significant relationships detected. We agree that
fitness has not been measured by this study—but we would argue that the
aim of the study was to demonstrate that both natural and sexual selection
have played a role in promoting the evolution of the tail streamer in the barn
swallow and to document their relative importance. We would also point out
that fitness has not been assessed in any similar study including those of
Møller. The exhaustive body of work by Møller and his colleagues
confirms the importance of sexual selection for tail exaggeration, however our
study confirms that the tail streamer has also evolved under influence from
natural selection. Our work has enabled us to quantify the extent of the
sexually selected component of streamer length which until recently seemed to
be a matter of confusion, for example "tail ornaments in the... swallow
have evolved as a result of female choice"
(Møller, 1988); in
contrast to "it is widely accepted that sexual selection cannot on its
own be responsible for the elongated tail"
(Hedenström
and Møller, 1999). We suggest that the way forward would be
to conduct experimental manipulations of the sexually dimorphic trait (i.e.,
the streamer rather than the basal part of the feather) using manipulations
that are smaller than the proposed sexually selected part of the streamer
(i.e., <15 mm). This would put studies of the benefits of streamer
possession on the same footing as the existing studies of the costs and allow
informed discussion of the selective pressures on tail length in this
species.
REFERENCES
Buchanan KL, Evans MR, 2000. The effect of tail
streamer length on aerodynamic performance in the barn swallow. Behav
Ecol 11:
228-238.
Cuervo JJ, de Lope F, Møller AP, Moreno J, 1996. The energetic cost of tail streamers in the barn swallow (Hirundo rustica). Oecologia 108: 252-258.[Web of Science]
de Lope F, Møller AP, 1993. Female reproductive effort depends on the degree of ornamentation of their mates. Evolution 47: 1152-1160.[Web of Science]
Evans MR, 1998. Selection on swallow tail streamers. Nature 394: 233-234.
Evans MR, Thomas ALR, 1997. Testing the functional
significance of tail streamers. Proc R Soc Lond B
264: 211-217.
Futuyma DJ, 1998. Evolutionary biology, 3rd ed. Sunderland, Massachusetts: Sinauer.
Hedenström A, Møller AP, 1999. Length of tail streamers in barn swallows. Nat Lond 397: 115.
Møller AP, 1988. Female choice selects for male sexual tail ornaments in the monogamous swallow. Nature 332: 640-642.
Møller AP, 1989. Viability costs of male tail ornaments in a swallow. Nature 339: 132-135.
Møller AP, Barbosa A, 2001. Flight, fitness and
sexual selection. Behav Ecol 12:
511-512.
Norberg RA, 1994. Swallow tail streamer is a
mechanical device for self-deflection of tail leading edge, enhancing
aerodynamic efficiency and flight manoeuvrability. Proc R Soc Lond
B 257:
227-233.
Park KJ, Evans MR, Buchanan KL, 2000. Evolution of tail streamers in hirundines: experimental evidence from the house marten Delichon urbica. Behav Ecol Sociobiol 48: 364-372.[Web of Science]
Rowe LV, Evans MR, Buchanan KL, 2001. The function and
evolution of the tail streamer in hirundines. Behav Ecol
12: 157-163.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||