Long forked tail ornament in male barn swallows (Hirundo rustica) were suggested to impose a condition-dependent viability cost ( Møller 1989; reviewed in Møller 1994; Møller & de Lope 1994; Møller et al. 1995): long tails impair male flight and foraging ability in terms of mean size of prey captured. In a recent study, we ( Matyjasiak et al. 1999) showed such a foraging cost of an experimentally imitated tail ornament in the sand martin (Riparia riparia), which have no tail ornament. We lengthened the tail in females, instead of in males, and thus controlled experimentally for the possible effect of differential allocation of female parental expenditure (differential-allocation hypothesis, Burley 1986).

Cuervo (2000) raises important issues in his comments about our paper. He questions the method used in our study ( Matyjasiak et al. 1999). He points out that ‘in order to test the cost of flight of an ornament, the trait to be experimentally manipulated has to be an ornament’, and that we should have not only elongated but also shortened tail feathers. Secondly, he suggests that we did not provide, in our article, the exclusive evidence for the handicap model of sexual selection. Thirdly, he disagrees with our suggestion that the results from the barn swallow experiments could be confounded by the differential allocation of female parental expenditure. Here, we outline the areas of agreement and disagreement between Cuervo's critique and our paper.

Firstly, we agree with Cuervo regarding the importance of tail shortening in studies of fully developed tail ornaments, which has already been pointed out by other authors ( Thomas & Rowe 1997; Evans & Thomas 1997). However, shortening of an ornament that is at equilibrium would always produce a decrease in costs. Therefore it will only confirm that an ornament is costly. However, if shortening is done to individuals of various, known condition, it may result in crucial information for distinguishing between hypotheses of sexual selection. We agree that for a character that is not an ornament, as in our experiments, shortening would not give new insights apart from another way of confirming the measures used in the barn swallow studies.

However, we disagree with Cuervo's claims of irrelevance of our experiment to the issues of the evolution of signals. We have chosen sand martins as a model species because the shape of its tail resembles that in ancestors of modern tail-ornamented swallows, from which tail feather elongation under sexual selection may have started ( Matyjasiak et al. 2000). Possible scenarios of the early evolution of a tail ornament may be examined by experimentally adding such an ornament, which requires manipulating original, non-ornamental traits (e.g. Goötmark 1994, 1996). We disagree with Cuervo that the existence of traits that may reduce the costs of ornaments (e.g. Møller 1996) eliminates the validity of tail manipulation studies in species without ornaments. We believe that such cost-reducing traits may not have existed during the early stages of evolution of tail ornaments (as well as other sexual traits) but may have developed later, and our experiment may represent such a situation very well. Hence, by imitating the initial development of a forked tail ornament in sand martins, we performed a biologically relevant manipulation.

Secondly, we do not state in our paper that we have tested or proved the handicap principle. We only mention in the last paragraph of the Discussion, that our results are consistent with this principle. However, we do admit that mentioning only the handicap hypothesis in the Introduction and Discussion only, and omitting other processes by which costly tail ornaments might arise, was unwarranted, as Cuervo properly argued; this might have left a false impression that our goal was to test the handicap hypothesis. We do believe that we have properly measured the costs of tail elongation, and this was our goal, as stated in the Abstract and in the last paragraph of the Introduction. Nowhere in the paper did we state that we actually aimed to test the differential costs of ornaments that are crucial to the handicap hypothesis. This issue was the objective of a different paper ( Matyjasiak et al. 2000) in which we consider conditions important for the early evolution of tail ornaments, with special emphasis on the handicap principle.

Thirdly, we are more cautious at interpreting the barn swallow studies and we do not exclude a possibility that differential allocation of parental expenditure can affect the size of prey brought by males. Differential allocation does exist in this species ( Møller 1992; de Lope & Møller 1993), as shown by the higher feeding rate by females responding to male higher attractiveness (longer tails). We believe that even though simple logical, intuitive reasoning may suggest that prey size should be unaffected by a differential allocation mechanism, it is not just reasoning but empirical proof that is needed here. Because there was no proof for a lack of the effect of differential allocation on the size of prey, we thought of an independent way of illustrating the costs of tail elongation in swallows, using a species without possible differential allocation effects. We thought that if we obtained results confirming the barn swallow studies by Møller and collaborators, we could help in validating the measures of tail elongation costs used in those studies. This was possible using the sand martin females for reasons discussed in the paper, all of which point to the lack of any differential allocation effects in this species. From this point of view, experimentally coupling tail elongation with tail shortening in our study appears unnecessary.

Even though, as argued by Cuervo (2000), it seems likely that the prey size in male swallows is unaffected by a change in feeding rate of females in response to male attractiveness (an assumption inherent in the barn swallow studies), another scenario is also possible. Imagine a female that has increased the amount of food for nestlings, in response to increased sexual attractiveness of her male partner due to experimental elongation of his tail. In such a situation the male has been relieved from a considerable duty of providing the young with a large amount of food. Hence, it is possible that such a male shifts from maximising the energy brought to nestlings per unit time (i.e. maximising foraging rate) to the criterion of obtaining the daily energetic needs at the least expense (i.e. minimising foraging costs). Such shifts may lead to including some non-preferred, small insects that can be captured quickly and inexpensively in terms of energy, because it does not require the use of expensive flapping flight, which is required to capture larger, preferred insects ( Waugh 1978; Bryant & Turner 1982; Turner 1980, 1982). By including more small insects in their catch at the expense of a reduced foraging rate, attractive males could save energy for future use. Barn swallows appear to compromise between maximising their foraging efficiency (maximising foraging gains per costs) and maximising energy intake per unit time (see fig. 5.7 in Turner 1980 and table IV in Turner 1982). Hence, if sexually attractive males aim at minimising foraging costs rather than maximising foraging rate for nestlings, then we cannot exclude the possibility that this may result in smaller insects being caught, on average, by males with experimentally longer tails. By a similar reasoning, it is theoretically possible that differential allocation effects could lead to larger mean prey size in males with experimentally shortened tails - an effect actually shown in the barn swallow studies ( de Lope & Møller 1993; Møller & de Lope 1994; Møller et al. 1995). Hence, whether shortening, elongating or performing both experimental manipulations, in our view we cannot be entirely sure whether the results are affected by differential allocation. This was sufficient motivation for us to investigate, independently, the usefulness of the change in prey size as an indicator of foraging costs due to elongated tails.

In contrast to Cuervo's opinion, we believe that our results are relevant to the issues important for the evolution of forked tail ornaments. We have measured the costs of a character that imitates the hypothetical early stages of the evolution of sexual ornaments, and we may use these results to discuss the early evolution of sexual ornaments (see Matyjasiak et al. 2000). We also confirmed the validity of measures of tail elongation costs used in the barn swallow studies.