From Animal Behavior, 1983, 31, 689-700

Abstract

Suggested roles of avian wing sounds in territorial defence have not been tested experimentally. The non-facultative, shrill wing whistle of male broad-tailed hummingbirds occurs during aerial displays used in defence of courting territories. To investigate the roles of the wing whistle in territorial behaviour, we observed males before and after being experimentally silenced, and after having wing whistles restored. Silenced birds intruded in territories more easily than normal males, suggesting that the wing whistle is a conspicuous signal advertising the presence of adult, male birds. Birds were less active and aggressive in territory defence after silencing than they were either before silencing or after restoration of the wing whistle. Silenced birds tended to lose their territories to rival hummingbirds more readily than non-silenced birds. We propose that silenced birds defend their territories less effectively because they cannot communicate threat and because they lack sensory feedback that normally enhances a bird's aggressiveness. Results of this study suggest that the wing whistle of male broad-tailed hummingbirds is important in maintenance of courting territories and hence in reproductive success.

Sound production in animals has evolved in many cases to enhance communication among individuals. It has long been thought that some bird songs function in territory advertisement (Thorpe 1961; Armstrong 1963), and recent experimental evidence supports this theory (Peck 1972; Smith 1976; Krebs 1977; Smith 1979). There has been less investigation of the potential territorial roles of mechanically produced avian sounds. It has been suggested that wing sounds in spruce grouse (MacDonald 1968), flappet larks (Wickler 1967; Payne 1973; Bertram 1977), and woodcocks (Sheldon 1967) probably function in territorial advertisement, as do tail feathers of snipe (Jellis 1977). Wing rattling in todies occurs during territory defence (Kepler 1977), and drumming in woodpeckers may facilitate territory defence (Winkler & Short 1978). Direct evidence has not been provided to test the behavioural roles of any of these sounds.

Male hummingbirds (Trochilidae) are typically territorial, defending mating and feeding territories during the breeding season (Wolf & Stiles 1970), and feeding territories at other times (Pitelka 1942). Males of a number of hummingbird species reportedly produce wing sounds in flight. These species include black-chinned hummingbirds (Archilocus alexandri), Costa's hummingbirds (Calypte costa), broad-billed hummingbirds (Cynanthus latirostris) (Bent 1964), Lucifer hummingbirds (Calothorax lucifer), and violet sabrewings (Campylopterus hemileucurus) (Wagner 1966).

Wing sounds are also produced by members of the genus Selasphorus, including three temperate zone species, the broad-tailed hummingbird (S. platycercus), the rufous hummingbird (S. rufus), and the Allen's hummingbird (S. sasin), and by at least one tropical species, the scintillant hummingbird (S. scintilla). The wing sound of S. rufus is thought to aid in courtship of female conspecifics and intimidation of male conspecifics (Sprot 1927). Pitelka (1951) suggested that the wing sound of S. sasin announces the presence of males during circuit flights over their territories.

During rapid flight male broad-tailed hummingbirds (S. platycercus) produce a particularly shrill, buzzing whistle as air rushes through slots created by the tapered tips of the ninth and tenth primary wing feathers (Fig. 1) (Ridgeway 1877; Woodbury & Sugden 1938; Wagner 1948; Armstrong 1963). It is uncertain whether the wing whistle is produced all year round or primarily during the breeding season (see Henshaw in Ridgeway 1892, and Wagner 1948 for opinions). The sound resembles the buzz of a cicada (Cicadidae) (Bailey 1928). The sound is probably perceptible at 50 m by males, and at more than 75 m by female conspecifics (Wagner 1948). Ridgeway (1877) wrote that dives and chases by males, "accompanied as they were by the shrill piercing noise, invariably put to flight all birds assaulted". Linsdale (1938) documented an especially loud buzz occurring in flight after a territorial chase. According to Wagner (1948, 1966), the wing whistle, which is obvious during territorial display flights, may be used in mate attraction and courtship.

The purpose of this study was to investigate the suggested roles of the wing whistle of male broad-tailed hummingbirds in territory defence. If the wing whistle functions as a social signal, its removal should result in behavioural changes in manipulated birds as well as the birds around them. Specifically, if the wing whistle is an aggressive signal, wing-silenced birds might be less effective in territory defence, substitute other aggressive behaviour for the missing whistle, or switch to less expensive defences if energy became limited. Flexibility in territorial behaviour has been documented as a response to variables such as resource quality (e.g. Ewald & Carpenter 1978). Since changes in an aggressive signal should affect the behaviour of the signal's receivers, we also predicted that conspecific males should be more aggressive and effective than normal at intruding into territories of silenced males. To test these predictions, territorial, male broad-tailed hummingbirds were observed before and after experimental silencing of the wing whistle, and interactions of these birds with conspecific males were analysed.

Methods

Study Site

We observed breeding broad-tailed hummingbirds between 29 May and 7 July 1981 at the Rocky Mountain Biological Laboratory, located at an elevation of 2900 m in the Elk Mountains of Central Colorado. The laboratory is situated in a steep valley dominated by subalpine meadow that is interrupted by patches of willows (Salix sp.). Adult males defended courting territories for approximately 6 weeks, from early June to mid-July. Post-breeding rufous hummingbirds began to migrate south through the study site in early July. An estimated 13 of 62 (21%) male broad-tailed hummingbirds at the study site were territorial, while the remainder were non-territorial and presumably competed for territory sites. We found 13 courting territories of male broad-tailed hummingbirds in the study area; they surrounded and enclosed willow patches and varied in size from 1800 square metres to 8000 square metres (X = 4343.9 square metres, N = 10). We estimated territory sizes from chase distances, location of intruders that elicited chase responses, and location of circuit flights. Willows, typically 1 m in height, covered 60-70% of the area of any given territory and served as perches.

Courtship and probably copulation occurred within territories, but nesting did not. Nectar sources within territories were rare except for willow blossoms, which were used by hummingbirds for 2 weeks in early June (personal observation). We placed a single feeder containing 30% sucrose (by volume) solution at the approximate centre of each territory to control for variation in food availability to territory owners. Territories were established before we distributed feeders, hence territories would have been defended without the supplemental feeding. All observations were made at least 1-1.5 days after birds had established territories, to control for the amount of time required for new owners to equilibrate in terms of defence costs (Copenhaver & Ewald 1980).

Capture and Mark Techniques

Most hummingbirds were captured using traps; butterfly nets and hummingbird mist nets were also used (capture permit 20753-A, first author, and 20753, second author). Traps were constructed by sewing fibreglass insect screening to a wire frame 0.6 m long and 0.4 m in diameter. A feeder was hung inside and a trap door could be closed using guy lines operated by a person hidden in the willows several metres behind the trap.

Two methods were used to mark captured birds. Coloured flagging was glued to the hummingbird's back using Duco cement (see methods in Baltosser 1978). Alternatively, Hyplar acrylic paints of similar colours were used to place a dot of colour on the bird's back or to paint the three central rectrices. Markings were easily visible with binoculars and frequently with the naked eye. Neither capturing nor marking had apparent effects on the territorial behaviour of most birds.

Observations

We chose to observe seven territory sites that were centrally located in the study area, with two of the sites separated by about 50 m and 250 m, respectively, from the others, and five of the sites clumped with shared borders. These seven sites offered high fequencies of behavioural interactions. Five of the seven sites were occupied by one individual throughout the entire breeding season, whereas a turnover of owners occurred at the others. We observed eight territorial males, one of which served first as a control and later as an experimental subject.

Times of observation were typically 0700 to 1200 hours because territorial activity was high and relatively constant during morning hours. Observers sat approximately 5 m from territory boundaries and recorded frequencies and durations of behavioural events important in territorial defence. Events observed included chases, vocal chatter, circuit flights (flights around a territory within its boundaries, sensu Pitelka (1951), occurring either in response to males heard outside the territory or when no environmental stimulus was evident), pendulum dives (vertical, U-shaped dives approximately 30 m in height (Bene 1947; Wagner 1948)), shallow dives directed at conspecifics, and the presence of intruding or passing hummingbirds. These flights and chattering are frequent and important in territory defence in many temperate zone hummingbirds (Pitelka 1951; Lyon 1973; Ewald & Carpenter 1978).

Experiments

The male birds used in our experiments were each observed for 2 days before treatment to silence the wing whistle. To fully silence the sound, we applied a thin film of Bond 527 quick-drying glue (Bond Adhesives Co.) with a toothpick to the narrowed portions of the ninth and tenth primary wing feathers (i.e. the distal 7 mm for the thenth, and distal 3 mm for the ninth). Glue weighed less than 0.024% of a bird's weight (mean bird weight = 3.58 g, N = 4; glue dry weight = 0.00085 g measured by a Mettler H20 balance). According to formulas presented by Feinsinger et al. (1979), we estimated that the wing disc loading (mass of the bird/area swept out by the wing) was thereby increased by 0.02%, and the cost of hovering flight was increased by 0.01%. The cost of forward flight is probably less than that of hovering flight (Greenewalt 1975). Experimental birds probably could compensate for the slight increase in flight costs by ingesting more food from feeders available at territory sites. Silenced birds did not preen more often than usual or exhibit any behavioural changes other than those recorded in interactions with conspecifics.

We observed the experimental birds again after silencing, and thus each bird served as its own control. Silenced birds on their territories were just as easily observed as normal, non-silenced birds, and hence observations were not biased against silenced birds. Five birds were each observed for the 2 or 3 days following silencing. Two others were observed until they abandoned their territories on the tenth and fourteenth days, respectively, after silencing. Sample sizes of experimental silencing trials were small because of (1) the limited number of males with territories having sufficient behavioural interactions, and (2) the difficulties with recapturing techniques at the start of the season.

The wing whistles of three birds were restored by removing the glue with acetone; glued portions were dipped in acetone and swabbed with acetone-soaked cotton swabs, during which proper ventilation was used to prevent fumes from affecting the birds. Feathers appeared completely normal after treatment with acetone.

Two sham-glued birds, with glue placed on equivalent surface areas of non-narrowed portions of primary wing feathers, served as controls for the presence of glue on the wings. Additionally, the acetone treatment, a procedure requiring manipulations similar to those of the gluing procedure, caused none of the behavioural changes observed in glued birds.

Observations of control birds were paired with appropriate experimental birds to control for daily and seasonal variations in territorial behaviour. Territorial activity varied among days because of weather (e.g. on windy days birds were less active), and varied seasonally with amount of courtship activity and with number of conspecific and heterospecific hummingbirds present at the study site. One control bird, which had been captured and marked, served as the control for one of the seven silencing trials. This bird's territory was located adjacent to that of the experimental bird. A second control bird was captured and marked, first silenced for 6 days, then treated with acetone to restore its wing whistle. This bird served as the control for the other six trials, and defended a territory approximately 250 m from those of the experimental birds.

Analysis

Tapes of wing whistles and vocalizations of several territorial males were made using a Uher 4000 Report IC Tape Recorder with a Uher M517 microphone mounted in a parabolic reflector. Durations and frequencies of territorial events were analysed using a randomization test to compare pre- and post-treatment behaviour in experimental birds. Control data were not subjected to statistical analysis since the same bird was observed as a control for six of the seven trials. However, mean control data for the seven trials are portrayed graphically with mean experimental data. Percentage data were arc-sine transformed. Behavioural events with frequencies less than or equal to 2/h were not included in analysis of data with percentages; this involved observations during less than 10% of the total hours.

Results

Territory Defence

When the wing whistles of territorial males were experimentally silenced, their territorial behaviour altered. This was not the case with the control birds with which experimental birds were paired in time (but not in a statistical sense). Experimental birds chased fewer intruding males after being silenced than before (P = 0.03, N = 6 birds, Fig. 2a). They perch-chattered (i.e. remained perched while chattering, instead of initiating a chase within 1 s) at more intruders after being silenced (P = 0.03, N = 6 birds, Fig. 2b, often allowing intruders to fly to them, hover over them, or direct shallow dives at them. Twice, we observed adult, male conspecifics directing pendulum dives at silenced territury owners, while we observed none directed at normal, non-silenced owners. A territory owner was considered "passive" if it remained perched while allowing any of these types of intruder behaviour, or if it was chased by an intruder. More intruders were received passively by territory owners (i.e. owners were passive during more territory intrusions) after the owners were silenced than before (P = 0.016, N = 6 birds, Fig. 2c).

Territorial chases by silenced birds appeared to be slower than those of non-silenced and sham-glued birds, although flight speeds were not measured. Fewer territorial chases by silenced birds were successful relative to those of non-silenced birds (P = 0.03, N = 6 birds, Fig. 3a); an unsuccessful chase was one in which the intruder hovered, chased, or dived at the silenced owner rather than leaving the territory readily. Means for the control birds in Fig. 3a are low because the bird that served as a control for six of the silencing trials directed most of its chases at a neighbouring male during boundary disputes. The neighbouring male apparently defended portions of the control bird's territory, and was therefore unusual in that it did not leave the control bird's territory readily.

Chases were shorter in duration (P = 0.03, N = 6 birds, Fig. 3b) and in distance after birds were silenced; chases frequently terminated within territorial boundaries rather than extending beyond them as chases did before birds were silenced. Territory owners performed fewer circuit flights after silencing than before (P = 0.016, N = 7 birds, Fig. 3c). The number of male intruders per hour did not alter significantly after territory owners were silenced (X before = 9.3, X after = 9.2, P > 0.05, N = 7 birds)>

To investigate compensation for lack of the wing whistle by use of chatter during flight, frequencies of chatter vocalizations were measured. More chases involved chatter by territory owners after birds were silenced than before (P = 0.03, N = 6 birds, Fig. 3d) More circuit flights involved chatter in four of six birds after silencing than before (X before = 15.5, X after = 28.1, P > 0.05, N = 6 birds, Table I). Durations of chatter during chases and circuit flights were not recorded.

Territory sizes decreased by an average of 36.4% (N = 8 territories, measured before and 2 days after owners were silenced), presumably as a result of delayed chase responses and shorter chase distances of territory owners following silencing. Two territory owners were observed for a week after being silenced to test the possibility that birds might eventually return to normal behaviour with normal territory sizes. Data for neither bird showed any significant return to normal behaviour in the 10 variables measured.

Data on territory losses to rival males, mostly migrant male rufous hummingbirds, suggest that silenced broad-tailed hummingbirds tended to lose their territories more readily than non-silenced conspecifics. Four silenced birds left or lost their territories to rival males by early July; two of these were dominated by single, migrant rufous hummingbirds and one by a single conspecific (Table II). Seven of eight non-silenced birds, and both birds with wing whistles restored, did not abandon their territories until 14 July or later, and then mostly because they were dominated aggressively by large numbers of rufous hummingbirds (Miller, in press a). Dates of territory loss (or abandonment) differ by a randomization test for silenced and non-silenced birds, both for 1981 alone (P = 0.008) and for 1980 and 1981 together (P = 0.002). Whistle-restored birds and a sham-glued bird fall within the range of dates for non-sioenced birds. Birds not represented in Table II had died or left the study area because of experimental manipulation. Non-silenced birds apparently maintained territories later in the season than silenced birds and resisted domination by single rival males. More complete data on the relative length of territory ownership for silenced and non-silenced birds are not available.

Four birds were observed immediately after silencing to determine whether the passiveness of silenced birds (1) was a delayed response to intruder behaviour, or (2) was an immediate response to loss of the wing whistle. The first alternative is correct if freshly-silenced birds used normal "beeline" chases (direct, rapid chases begun as soon as intruders entered territories) during initial encounters, but later used passive, delayed responses. Two birds used beeline chases when first silenced to exclude the first four intruders and the first intruder, respectively. All subsequent chases were more passive. Two birds did not use beeline chases at any time following silencing; their behaviour supports the second alternative. No definite conclusion may be drawn.

Territory Intrusion

Territory owners responded to experimentally silenced, intruding male conspecifics differently than to normal, non-silenced intruders. We observed two different owners directing pendulum dives at a silenced intruder, whereas we never observed owners directing pendulum dives at non-silenced, adult intruders. Silenced intruders, more than normal intruders, went unnoticed altogether or were unnoticed initially upon entry before being chased out by territory owners (Fig. 4). Flight patterns of silenced intruders that might have caused them to go unnoticed, e.g. flying low in the vegetation, were not observed. Although both silenced and normal intruders were chased off territories eventually, territory owners initially behaved passively toward normal intruders more often than toward silenced intruders (Fig. 4).

Restoration of the Wing Whistle

Male hummingbirds exhibited a return to normal territorial behaviour accompanying the return of the wing whistle. For bird OT, as shown in Fig. 5, most variables that altered significantly in all birds as a result of silencing returned after whistle restoration to almost exactly the same levels as before silencing. The only variable that did not alter as a result of silencing for bird OT was chase duration (Fig. 5e); nevertheless, it was the same after whistle restoration as before silencing. For two other birds with restored wing whistles, no observations were made before silencing and hence we cannot assert that they returned to presilenced levels of behaviour.

The three whistle-restored birds did not differ from the six (seven for Fig. 5f) non-silenced birds for variables in Figs 5a, 5e, and 5f (P > 0.05; randomization test). They did differ for variables in Figs 5b and 5c (P = 0.02; randomization test) but not in the predicted direction; that is, whistle-restored birds were even less passive toward intruders (less like silenced birds) than were non-silenced birds.

Sham Controls

Two sham-glued birds functioned as controls for the presence of glue on the primary wing feathers. Sham bird 2 (S2), with glue placed on the tips of the seventh and eighth primaries, produced an essentially normal wing whistle. Sham bird 1 (S1), with glue placed over equivalent surface areas on proximal portions of the ninth and tenth primaries, produced no wing whistle for two days following treatment. Within a week the whistle was produced again, but it varied in volume (from about 75% to 100% of that of normal whistles) during the remainder of the study. During several types of flights in which other birds produced whistles of low folume, bird S1 was silent.

It is difficult to interpret data for the sham-glued birds because of the small sample size. For several behavioural variables, the sham-glued birds were normal relative to untreated birds (Figs 2a, 2b, 2c, 3b), whereas for other variables, at least one sham-glued bird behaved more like silenced birds (*Figs 3a, 3c, 3d). The frequency of circuit flights for sham bird S2 actually falls above the highest values for silenced birds, and below values for six of the eight non-silenced birds (Fig. 3c). Without a larger sample size, we consider sham-glued birds to have behaved normally in terms of frequency of circuit flights. However, sham bird S1 performed fewer chases successfully (Fig. 3a) and more chases with chatter (Fig. 3d) than untreated birds. The values of these two variables may be interpreted in light of the fact that bird S1 produced a wing whistle of subnormal volume at times; this might cause some compensation with chatter during flight, and might cause some intruders to turn back while being chased and investigate the territory owner (bird S1) (see Discussion).

Hence, variations from predicted values in data for sham-glued birds might be explained by slight auditory consequences of the sham-gluing treatment, rather than the simple presence of glue on the primaries. In the absence of stronger, more complete controls for the presence of glue, we cautiously interpret the present data to indicate that gluing of the primaries did not itself cause behavioural changes in experimental birds.

Wing Whistle Audibility

The wing whistle was often audible to the human ear from distances of 75-100 m. Evidence that the wing whistle was audible to conspecific birds comes from observations of territorial males; we noticed that other males that we heard faintly in the distance elicited vocal chatter in the territorial males. Apparently the whistle was at least as perceptible to these males as to us.

We also observed conspecific responses to the wing whistle at closer ranges. Once, while a female was perched under a porch roof, a male performed a pendulum dive that was directly above her, but was visually obstructed by the roof. The female flinched as the male reached the bottom, loudest portion of the dive. Numerous times, when we heard a male aproaching a hummingbird feeder from around the corner of a cabin, we watched feeding females fly rapidly in the opposite direction, and we observed feeding males quickly approach the oncoming male.

Discussion

The wing whistle is a loud, shrill sound that is conspicuous during nearly all flights of male broad-tailed hummingbirds. As such, the wing whistle announces the sex, adult status, and presence of a male. It is probably an important means of communication among males who presumably compete for courting sites and for mates.

The whistle renders males conspicuous to rival males. Because silenced intruders fail to advertise their presence, they go unnoticed by territory owners more often than do non-silenced intruders (Fig. 4). Intruding silenced birds are therefore probably more successful at acquiring resources (although silenced and normal intruders alike are eventually chased from territories; Fig. 4). Kodric-Brown & Brown (1978) proposed that the wing sound of adult male rufous hummingbirds (S. rufus) reduces their success at intruding relative to conspecific females and juveniles. Data from Ewald & Rohwer (1980) for Anna's hummingbirds (Calypte anna) support an alternative: the brighter coloration of adult hummingbirds reduces their intrusion success. Our data from wing-silenced S. platycercus indicate that wing sounds of hummingbirds affect the success of territory intrusions.

The wing whistle probably plays an important role in advertising the presence of territorial males at particular sites. Circuit flights of hummingbirds may serve this function (Wagner 1948; Pitelka 1951), and in male broad-tailed hummingbirds other audible, territorial flights such as dives and chases probably do also. If the whistle during flights of territorial hummingbirds advertises presence and deters potential intruders, then silenced territory owners performing relatively few such (silent) flights might be expected to have relatively high territory intrusion rates. The number of intrusions per hour per territory did not change after we silenced territory owners; however, a better measure would have been intrusion rate/unit area since territory sizes decreased after owners were silenced. It is likely, in light of evidence that territory size often reflects changes in intrusion pressure (Myers et al. 1979; Ewald et al. 1980), that higher intrusion rates were hidden by obervation of intruders over smaller areas.

Role in Aggression

Results of the present study indicate that the wing whistle of male broad-tailed hummingbirds is important in the aggressive defence of territories. We suggest that experimentally-silenced birds may be less effective in territory defence than normal birds because (1) silenced birds cannot communicate threat during territorial displays, and (2) they act less aggressively because their normal behaviour patterns are disrupted by the sudden loss of the shrill sound that typically accompanies flight.

Threat. Territorial hummingbirds are notoriously aggressive, and in species such as broad-tailed hummingbirds which lack song, display flights (such as chases and dives) may intimidate territory intruders (Pitelka 1942). The wing whistle is most noticeable during these territorial displays and apparently enhances the communication of threat.

Specifically, our data show that chases by silenced territorial males are less successful than those of normal males in deterring intruders (Fig. 3a), and conspecific responses to a male's wing whistle indicate it is a threat. Also, silenced territorial birds appear to substitute chatter for wing whistle threat signals. In physical structure, the chatter vocalization is characteristic of aggressive sounds (i.e. it is harsh, covering a wide-frequency band, and is of relatively low frequency; Morton 1977).

Territorial encounters with silenced birds indicate that they are not perceived as threatening. Pendulum dives, which are normally directed toward any bird except another conspecific, adult male, are directed toward silenced, adult males, and shallow dives, normally directed only toward rival males, are not directed toward silenced males. Also, the silent flight of silenced intruders is incongruous with their flight behaviour (which is not the unobtrusive flight behaviour characteristic of silent females). Territory owners consistently investigate and chase thses non-threatening, abnormal intruders more than non-silenced intruders (Fig. 4).

Results of a sound analysis suggest that the wing whistle of male broad-tailed hummingbirds varies in sound frequency and volume with the level of aggressive motivation of the bird (Miller, in press b). This graded threat may be advantageous to a territorial hummingbird because energetically, a warning costs less than an attack, and also allows males to settle conflicts on the basis of differing motivations (van Rhijn 1980).

Enhancement of aggressive state. The wing whistle is undoubtedly extremely loud to the bird producing it. The sound itself, or vibrations associated with it, may create sensory feedback, enhancing the aggressive motivational state of the bird. If so, the sudden loss of this sound would disrupt the normal behaviour of the bird because the bird may no longer perceive itself as a normal male. Our data suggest the males are less aggressive after silencing, implying that the wing whistle enhances the aggressive state of male.

Experimentally silenced territory owners often perch low in the vegetation, have difficulty expelling intruders (Fig. 3a), and ultimately tend to lose their territories to rival males (Table II). Fewer intruders are chased by silenced owners (Fig. 2a), and individual chases are less intense because of delayed chase responses (Fig. 2c), shorter chase durations (Fig. 3b), and slower, less direct flight. Fast, direct flight is generally a threatening behaviour (Brown 1975). Chases of non-silenced birds but not silenced birds typically extend far beyond territory boundaries, a behaviour that in hummingbirds may deter intruders by indicating aggressive ability and ownership (Copenhaver & Ewald 1980).

DIsruption of normal aggressive behaviour is also implied by the difficulty of silenced birds in establishing new territories. We saw no silenced, territorial males (of 11) usurp territories or establish themselves on new sites, whereas of approximately 50 non-silenced males, at least 15 established new sites during the course of the summer in 1981, and at least five usurped sites. Silenced males may not have attempted territory establishment because of their disrupted aggressive state, or they may have made unsuccessful attempts, but in either case their aggressive abilities apparently were reduced. All the data thus imply that the wing whistle enhances territorial aggression.

An alternative explanation for the changes in territorial behaviour after birds are silenced might be that experimental birds hide their abnormal silence by switching to defences that involve less flight. It is difficult to disprove either explanation with the data at hand, but comparison of the present study with similar studies (see below) provides support for our suggestion that loss of the wing whistle causes loss of aggressive motivation.

Evolution of the Wing Whistle

Sounds occurring in the repertoire of a species presumably have been selectively favoured because they increase the fitness of the producer. Since broad-tailed hummingbirds defend feeding territories during the non-breeding season (DesGranges & Grant 1980), and defend courtship territories used in their promiscuous, competitive mating system (Wolf & Stiles 1970; this study), both the feeding and mating systems could provide selection for an effective aggressive signal. Wagner (1966) has suggested that the wing whistle is under sexual selection for mate attraction and stimulation. Production of sensory feedback that enhances a hummingbird's aggressive state may be a secondarily-derived advantage correlating with the type of social signal that evolved. Selection for a social signal used during the breeding season may be particularly intense because of the interspecific dominance of migrant rufous hummingbirds, whose exclusion of broad-tailed hummingbirds from territories in July (Miller, in press a) may place a constraint on the breeding season of broad-tailed hummingbirds. This constraint seems likely, since a socially subordinate species usually narrows its niche when sympatric with a socially dominant species (Morse 1974). The occurrance of mechanically-produced wing sounds among hummingbirds does not correlate either with migratory habits or with the use of high-elevation habitats. However, the open habitat of broad-tailed hummingbirds may be important in selecting for the use of conspicuous aerial displays enhances by acoustic signals.

Relevance to Similar Studies

Results of the experimental removal of the wing whistle of male broad-tailed hummingbirds resemble those of other studies involving the removal of socially significant visual or acoustic signals from male animals. Removal of antlers of red deer (Lincoln 1972), of epaulet coloration of red-winged blackbirds (Peek 1972; Smith 1972), of dark throat plumage of Harris sparrows (Rohwer 1977), and of face masks of yellowthroats (Lewis 1972) reduced their success in social aggressive encounters. However, in these cases, unlike in the present study, experimental males spent more time and energy than normal in intrasexual behavioural conflicts in an attempt to maintain territories or social dominance. The increased aggressive behaviour apparently was a response to the bold behaviour of surrounding individuals.

The signals investigated in the above studies appear to have varying importance in competitive encounters. For instance, experimental Harris sparrows compensated for loss of the ability to signal dominance by using other aggressive behaviour; experimental red-winged blackbirds compensated to some degree, but they often experienced territory loss; and experimental red deer stags soon adopted submissive behaviour and avoided conflicts even though in initial conflicts they were successful in maintaining dominance. Experimental hummingbirds in the present study fall even farther along this continuum because they adopted non-aggressive behaviour immediately after they could not signal aggressive intent.

We suggest that the aggressive motivation of red deer and broad-tailed hummingbirds were the most disrupted, red-winged blackbirds less so, and Harris sparrows the least disrupted by releaser removal. The differences in the awareness of signals are reflected in the ranking of disruption of aggressive motivation.

There is evidence from experiments with muted red-winged blackbirds that may be contrary to our theory. If song creates sensory feedback enhancing aggressiveness, then muting should cause reduced aggressiveness. Peek (1972) and Smith (1976) found that males with muted song continued to display visually and apparently defended territories vigorously and effectively. However, in Smith's (1979) study, muted birds (N = 3) experienced higher intrusion rates and lost their territories, suggesting disruption of aggressive motivation. Also, although 3 of 12 muted males in Peek's (1972) study were aggressive enough to regain territories that they had lost after they were muted, these birds had little competition in doing so. If muted red-winged blackbirds do not lose aggressive motivation, it might be because either (1) their muted vocal systems still produce some sound (Smith 1976), or (2) the threat display is still partly functional because birds can still display visually by spreading their wings. In none of the studies of muted red-winged blackbirds were the researchers investigating loss of aggressive motivation, hence more data are needed to test our theory of feedback enhancement of motivation.

The above studies of other animal species involving the removal of releasers used in social aggression generally support the hypothesized functions of the wing whistle in communicating threat and enhancing the aggressive state by physical feedback during territory defence by male broad-tailed hummingbirds. Results of the present study also suggest that the wing whistle advertises the presence of an adult, male hummingbird. This conspicuous acoustic signal, by mediating intrasexual conflicts for mates that visit courting territories during the breeding season, may enhance reproductive success in male broad-tailed hummingbirds. Further investigation is needed to test potential roles of the wing whistle in courtship and in winter territoriality.

Acknowledgments

This paper is part of a Master's thesis submitted to the University of Maryland by the senior author. We gratefully acknowledge the many forms of help provided by Wolfgang M. Schleidt and Eugene S. Morton throughout the study. Allen W. Smeltzer designed the hummingbird trap and offered much encouragement; he and Rebecca Craft provided invaluable field assistance. Statistical advice was patiently given by Svata M. Louda, Estelle Russek, and Stephen G. Vail. William A. Calder III and Nicholas M. Waser helped band many of the birds used in this study, and they offered ideas and equipment for field work. Critical reading of drafts of various manuscripts was provided by Wolfgang M. Schleidt, Eugene S. Morton, Robert S. Fritz, Paul W. Ewald, and many others. This research was supported by grants to S.J.M. from the Frank M. Chapman Memorial Fund, the Harris Foundation, and Sigma Xi, the Scientific Research Society, and NSF Grant DEB 78-07784 to D.W.I.

[Note: the references for this paper were illegible on my photocopy. If you need them, please ask your librarian to obtain a copy of the journal, or write the author for a reprint. -LC]