744 ShortCommunications andCommentaries [Auk,Vol. 111

ervation of avian blood and tissue samplesfor predominantlymonogamous : Genetic evi- DNA analysis.Can. J. Zool. 69:82-90. dence. Anim. Behav. 35:877-886. SHERMAN,P. W. ANDM. L. MORTON.1988. Extra-pair WESTNEAT,D.F. 1990. Geneticparentage in the In- fertilizationsin mountain White-crownedSpar- digo Bunting: A study using DNA fingerprint- rows. Behav. Ecol. Sociobiol. 22:413-420 ing. Behav. Ecol. Sociobiol. 27:67-76. SH•N, H.-S., T. A. BARGIELLO,B. T. C•a•ac, F. R. JACKSON, WESTNEAT,D. F., P. W. SH•RMAN, AND M. L. MORTON. AND M. W. YOUNG. 1985. An unusual coding 1990. The ecology and evolution of extra-pair sequencefrom a Drosophilaclock gene is con- copulations in . Curt. Ornithol. 7:331-369. served in vertebrates. Nature 317:445-448. WETTON, J. H., R. E. CARTER,D. T. PARICIN,AND D. SMITH, H. G., R. MONTGOMERIE,T. POLDMAA, B. N. WALTERS.1987. Demographicstudy of a wild WHITE,AND P. T. BOAG. 1991. DNA fingerprint- HouseSparrow population by DNA fingerprint- ing reveals relation between tail ornaments and ing. Nature 327:147-149. cuckoldry in Barn Swallows, Hirundorustica. Be- hav. Ecol. 2:90-98. Received30 June1993, accepted 17 November1993. WESTNEAT,D. F. 1987. Extra-pair fertilizations in a

The Auk 111(3):744-748, 1994

Impact of InterspecificAggression and Herbivory by Mute Swanson Native Waterfowl and Aquatic Vegetation in New England

MICHAEL R. CONOVER • AND GARY S. KANIA 2 •BerrymanInstitute and Department of Fisheriesand Wildlife, Utah State University, Logan,Utah 84322, USA; and 2Depart:aentof WildlifeManagement, National Rifle Association, 11250 Waples Mills Rd., Fairfax,Virginia 22030, USA

One of the greatestthreats to American avifauna is whether theseconcerns are real or whether swanpop- the establishmentof free-rangingexotic avian pop- ulationsshould be controlled.At present,Mute ulations (Temple 1992). Despite evidenceof substan- are protectedin some statesand unprotectedin oth- tial harm causedby exotic birds, most exotic ers;in still others,government employees attempt to are so poorly studiedthat many of their alleged en- control populationsby shaking eggs and re- vironmental impacts remain largely undocumented moving adults (Allin et al. 1987). In this study, we (Temple 1992). One such exotic speciesis the Mute examined interspecificaggression by free-ranging Swan (Cygnusolor), a Eurasianspecies that has been adult Mute Swanswith breeding territories in fresh- introduced severaltimes into North America, begin- water ponds and the impact of their herbivory on ning in the late 1800s(Allin et al. 1987).By the 1970s, aquatic vegetation. free-ranging populations existed in Michigan, Min- Methods.--Territorial pairs of free-ranging Mute nesota,Wisconsin, Wyoming, British Columbia, On- Swanswere observedat 15 freshwaterponds (2 to 30 tario, and in Atlantic coastalstates from Maryland to ha) in New Haven County,Connecticut from 1982to Massachusetts(Allin 1981). In the Atlantic Flyway, 1987.These ponds represented all freshwatersites in free-rangingpopulations increased from 200 birds in the study area known to have nestingpairs of free- 1955 to 5,300 in 1987 (Allin et al. 1987). ranging Mute Swans in 1982. Observationswere lim- Some biologistsare concernedthat the increasing ited to freshwater sites because these were the main populationof free-rangingMute Swansmay havean areas where Mute Swans came into contact with na- adverse impact on native waterfowl, owing to the tive waterfowl. swan'saggressive nature (Reese1975, Williams 1989). Data on the impactof interspecificaggression were Swanssometimes attack other waterfowl, causingin- collectedon both membersof a swan pair simulta- jury or death (Stone and Marsters1970, Willey and neously;data for males and femaleswere analyzed Halla 1972,Allin et al. 1987).Furthermore, aggressive separately.Sex of pair memberswas determinedby swansmay displaceother waterfowl (Willey and Hal- the larger body size and larger fleshy knob on the la 1972). An additional concern is that the foraging forehead of males (Bellrose 1980). Pairs were observed behavior of swansmay adverselyaffect aquaticplant year-round for 30-rain periods, randomly selected biomass,reducing the food available for other wa- amongdaylight hours.Observations were madefrom terfowl. Currently, data are insufficient to judge shore,usually from a car to minimizedisturbance of July 1994] ShortCommunications and Commentaries 745

the birds.During an observationperiod, we notedall T^BLE1. Number of aggressiveinteractions per hour aggressivebehaviors by swans,the other speciesin- (œñ St) initiatedby free-rangingmale (n = 15)and volved, and the outcomes.Agonistic behaviors by female (n = 15) Mute Swans against swans and swansincluded threats (raising wings above back), other waterfowl at 15 freshwater sites in Connect- icut. chases,and physical attacks. All sites were searched weekly from a boat or shorefor nestingwaterfowl. Samplesizes for all statisticaltests were the number Speciesbeing attacked Male Female pa of subjectsunder observation.Paired t-testswere used to ascertainif there were sexualdifferences in ag- 0.62 ñ 0.21 0.31 ñ 0.11 0.03 gressivebehavior. To test whether seasonalvariation 0.10 ñ 0.03 0.07 ñ 0.01 0.07 occurredin swanaggressiveness, data on eachswan's American Black 0.03 ñ 0.01 0.01 ñ 0.001 0.10 behaviorduring March to May, Juneto August,and 0.38 ñ 0.29 0.01 ñ 0.01 0.11 Septemberto Februarywere analyzedusing a two- Human 0.25 ñ 0.07 0.16 ñ 0.06 0.08 wayANOVA (seasonsvs. subjects). Linear regressions Total interspecific were usedto compareinterspecific aggression rates aggression 0.80 ñ 0.30 0.31 ñ 0.09 0.04 of individual swansto their intraspecificaggression Paired one-tailed t-test comparingmales and female swans. rates. To determinewhether waterfowlwere avoiding areas near swans, we prepared a map of each site site was consideredas being composedof five sec- noting water depths.At the beginningof eachob- tions. Eachtime a site was sampled,all of the above- servation period, we noted a swan's location and ran- groundvegetation was harvestedwithin a 0.33 x 0.33 domlyselected from the mapanother point of similar m sampling frame placed within each section. Sub- depth.At 5-rainintervals, we thenvisually estimated sequent sampling was taken elsewhere in each sec- the distancefrom the nearestindividual of everywa- tion so that no area was harvested more than once. terfowl speciespresent to both the swan and the ran- Samplingwas conductedin late May or early June domlyselected point. For eachindividual swan,these when the exclosureswere established(year 1, spring), datafrom all observationperiods were then combined three months later (year 1, fall), and 15 months later to yield a single mean distancevalue between it and (year 2, fall). Vegetationwas segregatedby species, each waterfowl species.A similar mean value was driedfor morethan 24 h at 110øC,and weighed. Paired obtainedfor the pairedrandom points. Hence, sample t-testswere usedto comparethe biomassof grazed sizes consisted of the number of individual swans and ungrazedsites. One grazedand one ungrazed under study. A paired t-test was then used to test samplewere obtainedfrom eachpond with eachsam- whethermean distances between a waterfowlspecies ple consistingof 10 sampleframes collected from the and a swan differed from the distance between that samepond (five sampleframes per exclosurex two speciesand a random point. exclosuresper pond). Hence, for thesestatistical tests, The experimenton the impactof swanherbivory the samplesize equalled the number of ponds under was conductedat 12 of the freshwaterponds used in study. the other part of this study.Exclosures were erected Results.--During405.0 h of observationon 15 males in late May or early June at two ponds in 1983, six and398.5 h on 15females, we observed870 aggressive ponds in 1984, two in 1985, one in 1986, and one in interactionsby free-ranging Mute Swans;410 were 1987.At each pond, we locatedtwo sites (3 x 3 m directedagainst other swansand 460 againstother each)between 20 and 50 m apart that were similar in species.Swans were aggressivetowards humans 174 water depth (between0.5 and 1.0 m) and appeared times (usuallyinvolving a threat displaydirected at to have similar vegetation. One of these sites was peopletrying to feed them). They were alsoaggres- randomly selectedto serve as an open site where sive towardsGreat Blue Herons (Ardeaherodias) three swanscould graze (hereafterreferred to as grazed times, (Larusspp.) three times, scaups(Aythya site)and markedwith a woodenpost. The other site spp.) twice, Gadwalls (Anasstrepera) once, domestic served as the ungrazed site and a 3 x 3 m swan ducksseven times, and domesticgeese 21 times.On exclosure was centered on it. The exclosure was built all other occasions,interspecific aggressive behavior by attaching a 0.3 to 0.5 m wide -wire fence was directedat (Anasplatyrhynchos), Amer- to eight posts.The fence was built so that its bottom icanBlack (A. rubripes),and CanadaGeese (Branta edge was level with the water surfaceso that it would canadensis;Table 1). excludeswans but not other aquaticherbivores such There were no significantdifferences among years as fish, turtles,diving ducks,and muskrats.Another in the aggressivebehavior of swans.Hence, data were pair of grazedand exclosuresites was set up on the summarizedfor eachindividual subjectacross years oppositeside of the pond. Data from thesetwo pairs and statistical tests conducted on the combined data. werealways combined for analysisto providea single Males were more aggressive than females towards grazedand ungrazedvalue for eachpond. both conspecificsand other species(Table 1). Rates Forvegetation sampling, each grazed and ungrazed of interspecificand intraspecificaggression were not 746 ShortCommunications and Commentaries [Auk, Vol. 111

TABLE2. Mean number of aggressiveencounters per chasingthem (Table 3). Swans got close enough on hour initiatedby free-rangingmale and femaleMute 50 occasionsto bite individuals of other species.Mal- Swans during different seasonsat five freshwater lards, American Black Ducks, Canada Geese, and do- sites in Connecticut. mestic geesewere bitten, but injuries did not appear serious.Once a swan used its wing to strike a goose. Season CanadaGeese were observednesting or with young Sep- at 7 of the 15 study sites, Mallards at 12 sites, and Speciesbeing March- June- tember- American Black Ducks at 1 site. We did not observe attacked May August February any instanceswhere swans foiled a nesting attempt Male swans by anotherwaterfowl species.Despite the swans'ag- Mute Swan 0.72 0.42 1.70 • gressive nature, there was no evidence that any wa- Mallard 0.24 0.04 0.11 terfowl speciesavoided areas near swans.Distances American Black Duck 0.05 0.19 0.03 between the closestwaterfowl to a random point and Canada Goose 0.25 0.06 0.14 to a swan did not vary for most waterfowl species Human 0.24 0.51 0.05 (Table 4). Total interspecific When the exclosureswere erectedin May to assess interaction 0.79 0.91 •0.40 the impact of swan herbivory, above-ground plant Female swans biomassfor all plant speciesand for the three most Mute Swan 0.47 0.01 0.69 ubiquitousspecies did not differ betweengrazed and Mallard 0.07 0.05 0.09 ungrazedsites (Table 5). In the fall, there was slightly American Black Duck 0.05 0.00 0.14 moreplant biomassin the exclosuresthan in the grazed Human 0.23 0.09 0.14 sites,but the differenceswere not statisticallysignif- Total interspecific icant.During the winter, exclosuresat five pondswere interaction 0.35 0.21 0.26 destroyedby shifting ice or vandalism.At the re- .' None of differencesamong seasons were significant(P < 0.05)based maining sevenponds, plant biomassin the fall of the on two-wayANOVA (2,8 df). secondyear wasslightly higher in the exclosuresthan at the grazed sites,but the differenceswere not sta- tistically significant. correlated in either males (r 2 = 0.01, P = 0.80) or Discussion.--Insouthern England, some Mute Swans females (r z = 0.01, P = 0.76). occupyand defend their territories year-round, de- Swansoccupied their territoriesthroughout the year pending upon winter weather and availability of food exceptwhen pondsfroze. Five malesand five females resourceson the territory (Scott 1984). Free-ranging were observed often enough in each season(more Mute Swans in Connecticut also occupied their ter- than 10 observationperiods/season) to be included ritories almostyear-round (exceptduring midwinter in the seasonalanalysis. Aggression rates for both when freshwater sites became frozen). Whenever male and female swansdid not vary by season(Table present on their territories, Mute Swans engaged in 2). Males varied in their aggressivenesstoward Amer- both interspecificand intraspecificaggression. Con- ican BlackDucks, CanadaGeese, and all interspecific sequently,they threatenedor attackedboth breeding interactions combined; females did not (Table 2). waterfowl and thosemigrating through or wintering Aggressiveinteractions usually ended when ducks in Connecticut. moved a few metersaway from a threatening swan. Mute Swans spent as much time in interspecific However, Canada Geese often had to swim more than aggressionas in intraspecificaggression. Intensive 50 m, had to fly away, or were chasedinto water less interspecificaggression is rare among mostbirds, but than 5 cm deepor onto dry land beforea swanstopped is common in one group: steamer-ducks(Tachyeres

TABLE3. Different responsesby waterfowlto aggressiveencounters with free-rangingmale or femaleMute Swans at 15 freshwater sites in Connecticut.

Percent of all waterfowl responses Speciesbeing Moved Swam Flewto Flew attacked n < 10 m 10-50 m >50 m ashore far side away Male swans Ducks 75 55 29 3 7 3 4 Canada Goose 115 30 24 21 11 10 4

Female swans Ducks 34 74 24 0 3 0 0 Canada Goose 29 69 0 14 7 0 10 July 1994] ShortCommunications andCommentaries 747

TABLE4. Distance(œ + SE) between a free-ranging TABLE5. Above-groundplant biomass(g/m 2) of all Mute Swan (or a random point) and closestwater- plantsand for threeubiquitous species at siteswhere fowl to them at 15 freshwater sites in Connecticut. Mute Swanscould graze and at ungrazedsites that were surroundedby exclasureswhen first erected Distance (m) to (Year 1, Spring; n = 12 ponds),three monthslater Speciesbeing n (Year 1, Fall; n = 12 ponds), and 15 months later attacked (sites) Swan Random point (Year 2, Fall; n = 7 ponds). Male swans Biomass(g/m 2) Mallard a 13 b 49.3 + 9.1 60.0 + 7.5 Sampling American period Grazed Ungrazed t' Black Duck 7 34.7 + 9.4 57.0 + 19.5 Wood Duck 5 46.2 + 15.6 58.2 + 18.3 All plant species Other ducks 4 57.8 + 13.9 53.5 + 22.4 Year I, Spring 36.0 37.9 0.80 Canada Goose 4 82.8 + 27.4 81.2 + 34.1 Year 1, Fall 59.1 40.2 1.38 Female swans Year 2, Fall 75.4 60.9 0.62 Mallard 14 51.6 + 10.4 61.9 + 10.8 Potomogeton spp. American Year 1, Spring 1.8 7.3 1.00 Black Duck 9 33.6 + 11.5 43.6 + 10.5 Year 1, Fall 1.5 0.1 1.19 Wood Duck 6 27.8 + 5.6 39.9 + 16.3 Year 2, Fall 0.1 0.0 1.00 Other ducks 6 69.5 + 22.6 42.3 + 6.7 Canada Goose 4 82.5 + 19.9 109.8 + 30.3 Nymphaea odorata Year 1, Spring 2.3 1.2 1.47 "The only significantdifference (P < 0.05) basedon paired t-tests Year 1, Fall 12.3 16.2 1.45 (two-tailed) was distance between American Black Ducks and male Year 2, Fall 12.6 18.3 1.23 swans. • Samplesizes are lessthan 15 becausesome waterfowl specieswere EIodea canadensis not observed at all sites. Year I, Spring 1.0 1.0 0.00 Year I, Fall 4.6 3.4 1.01 Year 2, Fall 1.6 1.6 0.00 spp.; Livezey and Humphrey 1985, Murray 1985, Nuechterlein and Storer 1985). This raisesthe ques- All comparisonsnonsignificant (P > 0.05). tion of whether steamer-ducks and swans share char- acteristicsthat can explain their intraspecificaggres- that shared a pond with a Mute Swan was the ener- sion. Steamer-ducks have massive heads, thick skin, getic costof evasion.There was, however, consider- and a large size advantage over their interspecific able variation among individual swans in the toler- opponents;these traits reduce the risk of defeat or ance of other waterfowl. The more aggressiveones injury from interspecificaggression (Nuechterlein and chasedother waterfowl, especiallygeese, that were Starer 1985). Likewise, swans have a large size ad- in their vicinity. The more tolerant swansdid not. vantage over their interspecificavian opponents and The swans'interspecific aggression also could ad- facelittle risk of injury during their attackson other verselyimpact other waterfowlby interfering with birds. None of the birds we observed being threat- breedingattempts, but suchwas not observed.Unlike ened or attackedby swansfought back. Kania and Smith (1986), we were unable to document Advantagessteamer-ducks or Mute Swansgain from any incidentsof a swancausing a nestingattempt to interspecificaggression are unclear.Nuechterlein and fail. Starer (1985) proposed that male steamer-ducksex- Nuechterlein and Starer(1985) reported that steam- hibit interspecificaggression to displaytheir fighting er-duckswere able to excludemolting Red Shovelers ability to females and to reduce food competition. (Anas platalea)from their territories. We found no Livezey and Humphrey (1985) argued that the ad- indication that the distances between waterfowl and vantageswhich steamer-ducksgain from interspecific a swan differed from the distance between waterfowl aggressionmay includeprotection of young,defense and a randompoint. If waterfowl were avoiding areas of food from marginalcompetitors, sexual advertise- near Mute Swans, a nonrandom distribution would ment, and practice for intrageneric combat. have been expected,but such was not found. One Biologistsare concernedthat the expandingpop- difference between our observations and those of ulationsof Mute Swansin the United Statesmay have Nuechterlein and Starer (1985) is that we did not limit an adverseimpact on native waterfowl either because our observationsto molting waterfowl. The birds in of their aggressivenature or their herbivoryon aquat- our study could easily escapefrom a swan if threat- ic plants (Reese 1975, Allin et al. 1987, Kania and ened. However this would not be true when geese Smith 1986,Williams 1989).We found that Mute Swans were molting and unable to fly. Hence, aggressive engagedin high rates of aggressionagainst native Mute Swansmay excludemolting geesefrom their ducks and geese, but that these episodes consisted territories. mainly of threatsand chases,and rarely resultedin Swansalso could have a deleterious impact on na- physicalcontact. Hence, one cost for nativewaterfowl tive waterfowl if their herbivory impactedaquatic 748 ShortCommunications andCommentaries [Auk,Vol. 111

plant biomass,thereby reducing either the vegetative KRULL,J. N. 1970. Aquatic plant-macroinvertebrate or macroinvertebrate (Krul11970) food resourcesused associationsand waterfowl.J. Wildl. Manage.34: by native waterfowl. In this study, we were unable 707-718. to documentany effect of swan herbivory on aquatic LIVEZEY, B.C., AND P.S. HUMPHREY. 1985. Territo- vegetation. However, swan populations in the area riality and interspecificaggression in steamer- are increasing(Allin et al. 1987) and breeding terri- ducks. Condor 87:154-157. tories are becoming smaller. For instance, in 1983 MURRAY,B. G., JR. 1985. Interspecificaggression in there were two breedingpairs on Lake Whitney, but steamer-ducks. Condor 87:567. by 1990 there were six. Hence, while current swan NUECHTERLEIN,G. L., AND R. W. STORER.1985. Ag- densitiesare not having an impact on aquaticplant gressivebehavior and interspecifickilling by Fly- biomassin New Englandponds, this may changeif ing Steamer-Ducksin Argentina. Condor 87:87- swan densitiesincrease substantially. 91. Acknowledgments.--Wethank P.M. Piconeand B. REESE,J.G. 1975. Productivity and managementof Young for their help in collecting these data. feral Mute Swansin ChesapeakeBay. J. Wildl. Manage. 39:280-286. LITERATURE CITED Scour,D.K. 1984. Winter territorialityof Mute Swans Cygnusolor. Ibis 126:168-176. ALLIN,C.C. 1981. Mute Swans in the Atlantic Fly- STONE,W. B., ANDA.D. MARSTEPS.1970. Aggression way. Proc. Int. Waterfowl Sympos.4:149-152. among captive Mute Swans. N.Y. Fish Game J. ALLIN, C. C., G. G. CHASKO,AND T. P. HUSBAND. 1987. 17:50-52. Mute Swansin the Atlantic Flyway: A review of TEMPLE,S.A. 1992. Exoticbirds: A growingproblem the history,population growth and management with no easysolution. Auk 109:395-397. needs. Trans. Northeast Sect. Wildl. Soc. 44:32- WILLEY, C. H., AND B. F. HALLA. 1972. Mute Swans 47. of RhodeIsland. Rhode Island Dep. Nat. Resour., BELLROSE,F. C. 1980. Ducks, geese, and swans of Div. Fish Wildl. Pam. 8. North America, 3rd ed. StackpoleBooks, Harris- WILLIAMS,W. 1989. Dark side of a classicbeauty. burg, Pennsylvania. Natl. Wildl. 27(2):42-49. KANIA, G. S., AND H. R. SMITH. 1986. Observations of agonistic interactions between a pair of feral Received15 July 1993,accepted 13 December1993. Mute Swansand nesting waterfowl. Conn. War- bler 6:35-37.

The Auk 111(3):748-750, 1994

Siblicide and Cannibalism at Northern Goshawk Nests

CLINT W. BOALAND JOHNE. BACORN Schoolof RenewableNatural Resources, University of Arizona,Tucson, Arizona 85721, USA

In many asynchronouslyhatching birds, broodsize Siblicideand cannibalismis uncommonamong avi- decreasesduring the nestling period in a character- an species(Mock 1984, Mock et al. 1990,Stanback and istic pattern, starting with the last-hatchedchick. This Koenig 1992). Most reports are based on indirect ev- has been widely interpreted as a systemby which idence, such as remainsof nestlingsfound in nests family size is adjusted to match available levels of (Heintzelman 1966, Pilz 1976, Moss 1979, Bechard essentialparentally provided resources(Lack 1954). 1983),and may fail to unequivocallyidentify the cause One causeof this mortality in someraptors and other of death. Observational accounts of siblicidal and can- predatorybirds is fatal sibling aggression(e.g. O'Con- nibalistic events are few (Newton 1978, Pilz and Sei- nor 1978, Stinson 1979, Mock et al. 1990), after which bert 1978, Jonesand Manez 1990, Bortolotti et al. 1991, the victim'stissues are sometimesingested by family Negro et al. 1992). The key events tend to be brief members(Ingram 1959, Mock 1984, Bortolotti et al. (Mock 1984) and may go unwitnessedunless a nest 1991).There is somecontroversy over how often con- is under constant observation. sumptionof the victim occursand, asa separateissue, We report the observationof a nestling Northern how important an evolutionarycomponent the can- Goshawk(Accipiter gentilis) killing and cannibalizing nibalism per se may be (reviews in Elgar and Crespi a siblingafter the adult femaledisappeared from the 1992,Stanback and Koenig 1992). nest area.In addition, we describea separateincident