Female preference for apparently symmetrical male sexual ornaments in the barn swallow Hirundo rustica

Summary The outermost tail feathers of barn swallows Hirundo rustica apparently reliably signal the quality of males, because individuals with the longest tails have the lowest degree of fluctuating asymmetry (random deviations from symmetry in the otherwise symmetrical tail trait) despite the size of their secondary sexual character. I experimentally tested whether females preferred males with symmetrical tails without altering the aerodynamic properties of birds by painting the tips of the outermost tail feathers with white or black correction fluid. Unmated males were randomly assigned to one of four treatments: (i) asymmetrical tails, where the outermost 20 mm of one tail feather was painted white and the other black; (ii) symmetrical tails where the outermost 10 mm of both tail feathers was painted white (symmetric I); (iii) symmetrical tails where the outermost 20 mm of both tail feathers was painted white (symmetric II); or (iv) controls where the outermost 20 mm of both tail feathers was painted black. The experimental treatment affected the duration of the premating period since it took longer for asymmetrical males to acquire a mate than for either group of symmetrical males or control males. This gave rise to a delayed start of laying among males with apparently asymmetrical tails. The seasonal production of fledglings therefore decreased from control males through males with either symmetrical treatment to males with the asymmetrical treatment. Females therefore pay direct attention to the level of fluctuating asymmetry in secondary sexual characters even when the asymmetry does not affect the aerodynamic properties of males.     

White tail markings are an indicator of quality and affect mate preference in rock sparrows

In birds, colourful and elaborate feathers are important traits in mate choice. Distinct tail white patches are present in many species of birds, but they remain little studied. Tail markings may indeed have a signal function because in many species males spread the tail offering a good view of these markings to females during courtship behaviour. Here, we investigated whether white tail spots in male rock sparrow, Petronia petronia, play a role in mate choice. In a free-living population of rock sparrows, we found a reduction in white tail spots size as the breeding season progressed due to abrasion, which was expected if tail spots act as a reliable quality indicator (i.e. a handicap). The same reduction was found under captive conditions, and males in worse condition (individuals that lost more weight) abraded a bigger part of white. This suggests that white tail markings are an indicator of male quality. In captivity, we measured female preference for males differing in white patch size in a mate choice experiment. The experimental reduction of the size of the males’ white spots resulted in a lower sexual interest by females. During courtship display, male rock sparrow shows a yellow breast patch (a carotenoid-based, sexually selected ornament) together with the white spots in the tail. The sizes of these two traits are positively correlated, but only the abraded white area in the tail correlates with a surrogate of individual quality (lost of weight). In conclusion, we can assert that the size of the white spots is preferred by female rock sparrows and it is a part of a multiple signal system.     

Diet quality affects an attractive white plumage pattern in dark-eyed juncos (Junco hyemalis)

Sexually selected traits that act as signals of quality often display some degree of condition dependence. In birds, condition dependence of ornamental plumage is often mediated by production costs related to acquisition or allocation of dietary resources. White plumage ornaments, however, have often been assumed to be inexpensive because their production requires neither pigment nor specialized feather structure. In male dark-eyed juncos (Junco hyemalis), the size of a white patch on the tail contributes to attractiveness and mating success. Using captive males, we examined the effects of diet quality on the size and brightness of the tail-white patch. After removing four tail feathers to induce replacement, we maintained subjects on a subsistence (low-protein) or enriched (high-protein) diet while induced feathers grew. Birds that received an enriched diet grew their feathers more quickly and grew larger, brighter white patches. Feather growth rate was positively correlated with the increase in the size of the tail-white patch, a relationship that was stronger in the subsistence diet group. However, within diet treatments, faster-grown feathers were slightly duller. Taken together, these results suggest that variation in diet quality may lead to condition-dependent expression of tail white and that condition dependence may be stronger in more stressful environments. We suggest a mechanism by which increased feather growth rate may lead to an increase in the size of the tail-white patch and discuss potential trade-offs between signal size and brightness.     

Assessing the aerodynamic effects of tail elongations in the house martin (Delichon urbica): implications for the initial selection pressures in hirundines

Of the three species of hirundine that breed sympatrically across the U.K., one, the barn swallow, has outer tail feathers elongated into streamers, whereas the other two species, the house martin and the sand martin, do not. The tail streamer of the barn swallow is regarded as a classic example of a sexually selected trait. Recent evidence, however, has suggested that streamers may have evolved largely through natural selection for enhanced flight performance and increased maneuverability. We tested the hypotheses that small streamers (1) increase performance in turning flight, but (2) decrease performance in flight variables related to velocity. We manipulated the lengths of house martin outer tail feathers and measured changes in their free-flight performance, using stereo-video to reconstruct the birds" three-dimensional flight paths. Five flight variables were found to best describe individual variation in flight performance. Of these five, the three variables determining maneuverability predicted that flight performance would be optimized by a 6- to 10-mm increase in the length of the outer tail feathers. In contrast, for mean velocity and mean acceleration, extension of the outer tail feathers appears to have a detrimental effect on flight performance. We suggest that the initial selection pressure for streamers in ancestral short-tailed "barn swallows" was via natural selection for increased maneuverability. In addition, we propose that the benefits of increased maneuverability have differed between hirundines in the past, such that the cost of increasing the length of the outer tail feather has, to date, outweighed the benefits of doing so in streamerless hirundines. Received: 14 February 2000 / Revised: 2 July 2000 / Accepted 18 July 2000     

Costs of overlapping reproduction and moult in passerine birds: an experiment with the pied flycatcher

The process of moult in birds requires the investment of substantial amounts of energy and nutrients in feather production and attendant processes. Flight performance may be reduced during the moult of the wing feathers, and moulting birds may suffer from an increased predation risk. These factors may explain why the moult is usually separated in time from other energetically demanding processes, such as reproduction. In this study, we investigated the importance of the temporal separation of moult from breeding activities in the pied flycatcher Ficedula hypoleuca. We induced a moult-breeding overlap by removing the two innermost primaries on both wings, thus imitating the natural loss of these feathers during the initial stage of the moult. The experiment probably did not stimulate feather regrowth, but may have reduced flight performance. Just before fledging of the young, manipulated males and females, as well as the fledglings, had a lower body condition than control birds. The return rate of adult males in the year following the manipulation was significantly lower for males in the manipulated group than for control males. The return rate of females was lower than that of males, but there was no difference between the two female groups. The number of nestlings recruited in the year after the manipulation was significantly lower for the group with an induced moult-breeding overlap than for the control group. The reproductive performance in recruited and returning old females was independent of the manipulation, and in returning adults and recruits the feather quality in terms of number of barbs/mm was independent of the manipulation. These results suggest that a moult-breeding overlap induces fitness costs in terms of reduced recruitment rate and adult survival in a long-distance migrant. Received: 17 September 1997 / Accepted after revision: 23 February 1998     

Moulting patterns drive within-individual variations of stable isotopes and mercury in seabird body feathers: implications for monitoring of the marine environment

One major limitation in the use of body feathers of seabirds as a monitoring tool of the trophic structure and contamination levels of marine ecosystems is the degree of heterogeneity in feather chemical composition within individuals. Here, we tested the hypothesis that moulting patterns drive body feather heterogeneity, with synchronous moult minimizing within-individual variations, in contrast to asynchronous feather growth. Chicks of white-chinned petrels Procellaria aequinoctialis (representative of bird chicks) and adults of king penguins Aptenodytes patagonicus (representative of adult penguins) that moult their body feathers synchronously showed very low within-individual variations in their feather δ¹³C and δ¹⁵N values and mercury (Hg) concentrations. By contrast, body feathers of adults of Antarctic prions Pachyptila desolata (representative of adult seabirds with asynchronous feather growth during a protracted moult) presented much higher within-individual variances for the three parameters. These findings have three important implications for birds presenting a synchronous body moult. (1) They suggest that all body feathers from the same individual have identical δ¹³C and δ¹⁵N values and Hg content. (2) They predict negligible within-individual variations in the body feather values of other useful stable isotopes, such as δ²H and δ³⁴S, as well as in the concentrations of other compounds that are deposited in the keratin structure. (3) Analysis of one or any number of pooled body feathers is equally representative of the individual. In conclusion, we recommend that long-term routine monitoring investigations focus on birds presenting synchronous rather than asynchronous moult of body feathers both in marine and terrestrial environments. This means targeting chicks rather than adults and, for seabirds, penguins rather than adults of flying species.     

Inter- and intra-sexual patterns of fluctuating asymmetry in the red-billed streamertail: should symmetry always increase with ornament size?

Recent work on fluctuating asymmetry has suggested that ornaments should have higher levels of fluctuating asymmetry than (1) non-ornaments and (2) homologous structures in the non-ornamented sex. In addition, as both ornament size and symmetry should increase with individual quality there should be a tendency for ornament symmetry to increase with ornament size. In non-ornaments, a U-shaped relationship between symmetry and size is expected, with the individuals at the extremes being more asymmetrical than individuals around the optimum. We tested these predictions in the red-billed streamertail (Trochilus polytmus), a sexually dimorphic endemic Jamaican hummingbird. The lengths of four bilaterally symmetrical traits (first and second outermost tail feathers, tarsi and wings) in 43 adult males and 42 females were measured. The second outermost tail feathers of adult males (which are elongated into streamers) were absolutely but not relatively more asymmetrical than non-ornaments (including the homologous feathers in females). When character size was controlled for, wings were shown to be relatively more symmetrical than other traits. Symmetry did not increase with increasing trait size in any of the morphological traits measured. There was a U-shaped relationship between asymmetry and trait size for four traits (adult male streamers, adult male wings and female outer tail feathers). These results do not support any of the predictions made by fluctuating asymmetry hypotheses and suggest that stabilising selection may act on ornaments as well as non-ornaments. These predictions have been supported in swallows and peafowl but not in sunbirds; this may be due to differences in female perception of tail ornaments. Perhaps male tails do not convey information about quality in some species, or there may be inter-specific differences in the relative costs of tail ornaments and the benefit of marginal increases in tail length and symmetry.     

The effects of parasites on male ornaments and female choice in the lek-breeding black grouse (Tetrao tetrix)

Summary We describe the results of two studies of parasitic infection in the black grouse (Tetrao tetrix). The first deals with our own observations of lekking black grouse in which the parasite levels of two blood parasites, the protozoan Leucocytozoon lovati and microfilaria, probably produced by a nematode worm Splendidofilaria tuvensis, were scored. We also obtained measures of age, survival, number of copulations, body mass and length of the ornamental tail feathers (the lyre) of the lekking males. The second study analysed the data from Lund (1954) which involved eight gut parasites obtained from birds which were killed. In the first study we found higher levels of infection of Leucocytozoon in adults relative to young birds. Parasites had no effect on male survival and there was no correlation in infection between the two types of parasites. Birds infected with microfilaria had shorter tail ornaments. There was no relationship between parasitic infection and mating success. However, the data indicated that such a trend indeed may exist for Leucocytozoon and the most successful males on the leks were less often infected by Leucocytozoon than other males. Results of the second study showed a negative relationship between parasite load (a combined measure of all parasites) and both ornamental tail feather length and body mass. These observations are compatible with, but not conclusive evidence for, the hypothesis of Hamilton and Zuk (1982) on the evolution of secondary sexual characters, where females choose to mate with genetically resistant males which show their resistance by expressing larger and more showy secondary sexual characters. Alternative explanations for the observed patterns are: females avoid infected males for some immediate benefit; and/or parasite loads are indicators of general stress rather than genetical resistance. Under the latter hypothesis females could mate with more vigorous males for reasons unrelated to parasite resistance.     

Testosterone and sexual signalling in male house sparrows (Passer domesticus)

Female preference for mates with elaborated ornaments has often been explained on the basis that exaggerated secondary sexual traits might reflect individual quality and females might gather direct and indirect benefits in mating with such males. Sexual signals must however also entail costs to be reliable indicators of male quality. Androgens have been suggested as a physiological link between sexual signals and individual quality for several reasons, including their immunosuppressive effect. In this study, we tested two hypotheses linked to the hormonal basis of sexual signal expression. First, we investigated whether testosterone is correlated with the size of the black feather bib on the throat of male house sparrows (Passer domesticus) - a trait involved in intra- and inter-sexual selection. Second, we tested whether testosterone affects the seasonal exposure of the trait. Observational work conducted in 1998 showed that the testosterone level was positively correlated with badge size both in spring and in the subsequent fall, after molt. In 1999, we experimentally reduced spring testosterone levels using silastic implants filled with cyproterone acetate, an antiandrogen. Male house sparrows implanted with cyproterone acetate showed reduced exposure of the badge, because the white tips of the black feathers of the badge wore off later than in control males implanted with empty silastic tubes. This result suggests that testosterone can be causally involved in the expression of a secondary sexual trait in house sparrows, at least in terms of its seasonal exposure.     

An experimental study of male adornment in the scarlet-tufted malachite sunbird: II. The role of the elongated tail in mate choice and experimental evidence for a handicap

Summary Scarlet-tufted malachite sunbirds (Neetarinia johnstoni) are endemic to the high altitude zones of East African mountains. On Mount Kenya the males are resident and territorial all year. The females arrive over an extended period at the beginning of the breeding season, after spending the non-breeding season in nomadic flocks at slightly lower elevations. Males are bright iridescent green with elongated central tail feathers, which are displayed during courtship. We examined the role of the tail in mate choice using natural variation and investigated the influence of tail length on male time budgets by experimental manipulation. Territorial males that paired had longer tails than territorial males which remained unpaired. Males with naturally long tails started breeding earlier, resulting in fledglings being produced earlier in the season when the weather was more clement. Birds which bred together in one year were paired together in the next year if they both survived. Thus the male's tail may only be involved as a cue in the formation of new pairs. Males which were already paired were subjected to one of three experimental treatments — having their tail elongated, shortened, or manipulated but kept at the same length. Males with experimentally shortened tails spent more time in flight and hawked for flying insects with a higher efficiency than control males. Both control and elongated tail males reduced the amount of time in flight and had a lower hawking efficiency after manipulation. These results suggest that the long tail of male scarlet-tufted malachite sunbirds is a handicap. The factors influencing the reproductive success of a pair are discussed. Offprint requests to: M.R. Evans     

Sexual selection in a monomorphic lek-breeding bird: correlates of male mating success in the great snipe Gallinago media

Summary In the lek-breeding great snipe, male morphology, behaviour, and territory features were recorded for individually marked birds on two adjacent leks. Partial correlation sshowed that male mating success, expressed as the number of female solicitations and copulations, was negatively correlated with the distance of a display territory to the lek center and positively correlated with the number of displays per unit time given by a male. No other variables were directly correlated with male matin success. Thus, central males obtain more matings than peripheral males and successful males display more per unit time than do less successful males, independently of position on the lek. Central males were found to be older than peripheral ones and were present more often on the lek. Furthermore, central males had a larger number of white tail feathers, which are usea as visual signals in the displays, but this may be explained by the fact that these males were older. It is suggested that male great snipe are subject to sexual selection mainly in behavioural and vocal cues and that this may explain the absence of size and plumage dimorphism in this species.     

Racketed tail of the male and female turquoise-browed motmot: male but not female tail length correlates with pairing success,performance, and reproductive success

Both males and females of many avian species maintain elaborate plumage traits, and elaborate monomorphic plumage may convey adaptive benefits to one or both sexes as inter- or intraspecific signals. Both sexes of the turquoise-browed motmot (Eumomota superciliosa) are elaborately plumed with long racket-tipped tail. I investigated whether the racketed tail functions as a sexually selected signal in one or both sexes by testing the predictions that males and/or females with the largest tails have: (1) greater pairing success, (2) greater reproductive performance (clutch-initiation date, clutch size, and hatching success), and (3) greater reproductive success. Yearling males with longer denuded rachises (wires) on the central tail feathers had greater pairing success. In addition, adult males with longer wires paired with females who laid larger clutches, had greater hatching success independent of clutch size, and fledged more young. There was no relationship between female tail plumage and pairing success, reproductive performance, or fledgling success. These results are consistent with the hypothesis that male tail plumage functions as a mate choice or status signal, but that the tail of the female does not function in a sexually selected context. I discuss alternative hypotheses for the evolutionary maintenance of the elaborate female tail plumage.     

Sexual selection and the tail ornaments of North American barn swallows

Summary In this study conducted in southeastern Ontario, Canada, we manipulated the length of outer tail feathers (streamers) of male barn swallows (Hirundo rustica) shortly after they returned to four small breeding colonies in the spring. Both streamers on most males in these colonies were experimentally either lengthened or shortened by 20 mm and as a result 10 randomly chosen males of each category had their streamer lengths manipulated before the fertile period of their mate. Males with elongated streamers had a significantly shorter pre-laying period (from arrival until first egg date) than those with shortened streamers and we interpret this as indicating that they obtained their mates more rapidly. Females mated to elongated males had significantly longer tails and laid their first egg significantly earlier than those mated to shortened males; no other indices of reproductive success differed between these two experimental groups. These results provide some support for previous work suggesting that females prefer males with longer streamers and thus that streamer length in this species is under the influence of sexual selection. We also found that males with longer natural streamers were significantly more likely to break them and that males were much more likely to break their streamers than were females. Since males participate in incubation in North America but not in Europe, we suggest that the aerodynamic disadvantages of streamer breakage that result from incubation attentiveness are at least partly responsible for the shorter streamer length of males in North America. Offprint requests to: R. Montgomerie     

Feather coloration in museum specimens is related to feather corticosterone

Colorful ornaments in birds are often sexually selected signals of quality, and variation in ornament expression may be mediated by physiological stress through the secretion of corticosterone. However, testing for links between ornamentation and corticosterone often requires sampling live animals, and such physiological measures may not be matched in the time span in which they were sampled (e.g., very dynamic plasma corticosterone vs. plumage coloration, which is relatively static). Here, we use museum specimens to test for a link between the color of a sexual ornament and feather corticosterone at the time of ornament formation. In red-winged blackbirds, Agelaius phoeniceus, carotenoid-based epaulets appear to be important in male–male social interactions, territory maintenance, and female choice. We measured reflectance spectra of adult male epaulets and plucked adjacent feathers for corticosterone analysis via radioimmunoassay. We controlled for differences in the number of mates, specimen age, and geography by selecting only males with one mate and only birds collected in Florida during a 3-year period. Epaulet hue and red chroma did not vary with feather corticosterone, but males whose epaulets scored high for mean brightness and red brightness had significantly lower corticosterone than males with low brightness scores. This correlation with brightness but not hue or chroma is consistent with an effect of corticosterone (CORT) on feather microstructure, with elevated CORT leading to lower reflectance of white light from the keratin matrix surrounding the carotenoid pigments.     

The effect of rearing environment on blue structural coloration of eastern bluebirds (<Emphasis Type="Italic">Sialia sialis</Emphasis>)

We used a brood-size manipulation to test the effect of rearing environment on structural coloration of feathers grown by eastern bluebird (Sialia sialis) nestlings. Ultraviolet (UV)-blue structural coloration has been shown to be sexually selected in this species. Our experimental design took advantage of the growth of UV-blue wing feathers in nestlings that are retained as part of the first nuptial plumage. We cross-fostered nestlings to create enlarged and reduced broods with the purpose of manipulating parental feeding rates and measured the effect on nestling growth and plumage coloration. Brood size influenced feeding rates to offspring, but the effect varied with season. In general, male nestlings reared in reduced broods were fed more often, weighed more, and displayed brighter structural plumage compared to nestlings reared in enlarged broods. Female nestlings appeared to experience less adverse affects of brood enlargement, and we did not detect an effect of brood-size manipulation on the plumage coloration of female nestlings. Measures of plumage coloration in both males and females, however, were correlated to hatching date and nestling mass during feather development. These data provide empirical evidence that environmental quality can influence the development of the blue structural coloration of feathers and that males may be more sensitive to environmental fluctuations than females.     

Achromatic color variation in black-capped chickadees,Poecile atricapilla: black and white signals of sex and rank

Sexual dichromatism and phenotypic variation in elaborate male traits are common products of sexual selection. The spectral properties of carotenoid and structurally-based plumage colors and the patch sizes of melanin-based plumage colors have received considerable attention as sexual signals in birds. However, the importance of variation in achromatic plumage colors (white, gray and black) remains virtually unexplored, despite their widespread occurrence. We investigated a potential signal function of the achromatic black and white plumage of black-capped chickadees (Poecile atricapilla). We captured and color-banded 178 free-living chickadees and assessed winter flock dominance hierarchies by tabulating pairwise interactions at feeders. We recaptured 73 of these birds and measured plumage coloration for six body regions using a reflectance spectrometer and the area of melanin-based plumage patches from standardized photographs. We found extensive individual variation in chickadee plumage traits and considerable sexual dichromatism. Male black-capped chickadees have significantly brighter white plumage than females, larger black patches, and greater plumage contrast between adjacent white and black plumage regions. We also found rank differences in the plumage reflectance of males; high-ranking males, who are preferred by females as both social and extra-pair partners, exhibit significantly darker black plumage and grow their feathers more rapidly than low-ranking males. This variation among individuals reveals a potential signal function for achromatic plumage coloration in birds.     

Using carbon and nitrogen isotopic values of body feathers to infer inter- and intra-individual variations of seabird feeding ecology during moult

To determine whether stable isotope measurements of body feathers can be used to investigate the isotopic niche of moulting (inter-nesting) adult seabirds, we examined the stable carbon (δ¹³C) and nitrogen (δ¹⁵N) isotopic composition of body feathers of breeding wandering albatrosses (Diomedea exulans) from Crozet Islands, southern Indian Ocean. First we showed that the isotopic composition of body feathers was not significantly different from that of wing feathers, being thus a safe alternative to flight feathers whose collection impairs the birds’ flying ability. Second, we looked at the variances in δ¹³C and δ¹⁵N values resulting from the isotopic measurement of a single feather, four different feathers, and a pool of four feathers per bird, to delineate the best isotopic analytical procedure. A two-step protocol is proposed that allows investigating both the intra- and inter-individual components of the niche width of the species. In a first step, isotopic measurements on a single feather per bird are used to define isotopic specialist from isotopic generalist populations. In a second step and for generalist populations only, measurements on additional (three) feathers per bird are used to delineate type A from type B isotopic generalists (Bearhop et al. in J Anim Ecol 73:1007–1012, 2004). Third, from a biological point of view, our data showed different moulting isotopic niches for adult males and females, and also within female wandering albatrosses. Since the isotopic composition of body feathers in this species reflects that of wing feathers, our results suggest that, after validation, body feathers have the potential for investigating the foraging ecology of other Procellariiforms and seabirds during the poorly known inter-nesting period.     

White tail plumage and brood sex ratio in dark-eyed juncos (<Emphasis Type="BoldItalic">Junco hyemalis thurberi</Emphasis>)

Theory, empirical examples, and recently, proximate mechanisms point to the possibility of adaptive sex ratio adjustment in various organisms. General predictions state that a female should adjust her offspring sex ratio to maximize the benefits or minimize the costs of reproduction given her physical condition or current social and environmental conditions. I tested for an influence of male attractiveness on brood sex ratio in a population of dark-eyed juncos (Junco hyemalis thurberi) by manipulating a male’s white outer tail feathers (“tail white”). Experimentally increasing male tail white did not significantly affect sex ratio, nor was premanipulated male tail white significantly related to brood sex ratio. However, the amount of white on the female’s outer tail feathers, independently of female condition, was positively related to the number of sons in a brood. Determining how a female’s potential genetic contribution to her sons’ attractiveness influences offspring sex ratio should be a priority for future research.     

Schwermetalle in Federn

Results from the literature concerning feathers as a matrix for heavy metal accumulation were presented. Especially problems like the differentiation between endo and exogenous contributions, cleaning procedures or the right choice of feather parts were described. The question of the use of feathers for monitoring air, food or bird contamination is also discussed, as well as the advantages or disadvantages of this matrix in comparison to other methods.     

Sexual ornamentation,condition, and immune defence in the house sparrow Passer domesticus

Female mate preferences for extreme male ornaments may arise and be maintained because males signal their ability to raise an immune defence against parasites by the size of their ornaments. The bursa of Fabricius is an organ involved in antibody synthesis in young birds, but regresses before sexual maturity. The relationship between badge size, condition, and immune defence was investigated in male house sparrows, Passer domesticus, which have a bib of black feathers under their beak, a secondary sexual character currently subject to a directional female mate preference. The size of the bursa of Fabricius decreased during the first year of life in house sparrows. Males with large secondary sexual characters had a smaller bursa than males with small badges. Individuals with a large directional asymmetry in relative testis size had a larger bursa. The abundance of one parasite (a mallophagan) was positively related to bursa size. There was a positive relationship between relative volume of bursa of Fabricius and the number of fault bars on feathers, which is a measure of body condition. These results suggest that males with large badges have lower current levels of immune response than males with small badges, and that males in good body condition, as evidenced from the absence of fault bars, had a relatively small bursa of Fabricius. Received: 6 October 1995/Accepted after revision: 2 June 1996