January1986] ClutchSize in Motmots 13

B., J. J. ReyesR., Wilfrido ContrerasD.) of Direcci(•n KOEt•IC,W. D. 1982. Ecologicaland social factors Generalde la FaunaSilvestre, Secretarla de Agricul- affectinghatchability of eggs.Auk 99: 526-536. tura y RecursosHidr•ulicos (M•xico, D.F.). Officials LACK,D. 1947. The significanceof clutchsize. Ibis of the M•rida office of Divisi6n Hidrom•trica, SARH, 89: 302-352. kindly gave us accessto weather data. George and MARTIN, M. W., & R. F. MARTIN. 1985. Nestling JanetCobb, Joann Andrews, and EdwardKurjack were feeding schedulesof Turquoise-browedMot- generous hosts in Yucatfm. We thank Anne Clark, mots in Yucatan, Mexico. Wilson Bull. 97: 372- Paul Mason, and two anonymousreviewers for use- 374. ful criticismsof a draft of this paper. Partial support OREJUELA,J. E. 1977. Comparativebiology of Tur- for this study was provided by The Explorers Club quoise-browedand Blue-crowned motmots in the and by the TexasMemorial Museum-The University YucatanPeninsula, Mexico. Living Bird 16: 193- of Texasat Austin.The Museumof Zoologyof Lou- 208. isianaState University provided computertime. SCnP•EmER,R. W., & E. A. SC•P•EmER.1983. The world accordingto E1Nifio. Abstr., 101ststated meet- LITERATURE CITED ing of Amer. Ornithol. Union, New York. Scot'r, P.E. 1984. Reproductionof the Turquoise- CLARIC,A. B., & D. S. Wig,SOtS. 1981. Avian breeding browed Motmot in Yucat•n, Mexico. Unpub- adaptations:hatching asynchrony,brood reduc- lished M.A. thesis, Austin, Univ. Texas. tion, and nest failure. Quart. Rev. Biol. 56: 253- --, ,•' R. F. MARTIN. 1983. Reproductionof the 277. Turquoise-browedMotmot at archaeological CONOVER,W.J. 1980. Practical non-parametric sta- ruins in Yucat•n. Biotropica 15: 8-14. tistics.New York, John Wiley & Sons.

The Frank M. ChapmanMemorial Fund givesgrants in aid of ornithologicalresearch and alsopostdoctoral fellowships. While there is no restriction on who may apply, the Committee particularly welcomes and favorsapplications from graduatestudents; projects in gamemanagement and the medicalsciences are seldom funded.Applications are reviewed oncea year and should be submittedno later than 15 January-with all supporting material. Applicationforms may be obtainedfrom the Frank M. Chapman Memorial Fund Committee,% JaneConnelly, The AmericanMuseum of Natural History, CentralPark Westat 79th Street, New York, NY 10024-5192. Dr. Mary McKitrick was appointedChapman Fellow for the period May 1985 through May 1986. She is studyingthe significanceof individual variation in M. flexor crurislateralis in the Tyrannidae. Dr. Nina Pierpontwas appointedChapman Fellow for the period October1985 through October1986. She is studyingthe evolution of diversity in woodcreepers(Aves: Dendrocolaptidae). Chapmangrants during 1985,totalling $25,482with a mean of $499.65,were awardedto: David J. Ander- son,booby breeding and feedingand marineresources in the Galapagos;John Anderson, tests of cooperative foragingin the White Pelican;Dr. BonnieBowen, genetic structure and paternityin communallybreeding jays;Angelo Capparella,effects of riverine barrierson gene flow in Amazonianforest undergrowth birds; William J. Carmen,social behavior and ecologyof the CaliforniaScrub Jay; Richard Casey, the effectof social interactionduring the criticalperiod of songlearning in White-crownedSparrows; Peng Chai, avian feeding behaviorand its implicationsfor the evolution of neotropicalbutterfly mimicry; JackClinton-Eitniear, Ca- thartesburrovianus: movement patterns and systematics;Nonie Coulthard,the White-throatedBee-eater (Mer- opsalbicollis): a study of cooperativebreeding in relation to ecologicalfactors; Marilyn England,breeding biologyof the Northern Harrier on Long Island,New York; ChristianErard, systematics of African Musci- capidae;Mark A. Finke, male parental care and the evolution of reproductiveeffort: an experimentaltest; Dr. JonFjeldsa, museum work for a field guideand biogeographicanalysis of the Andeanavifauna; Ronald C. Gaines,nest site tenacity and philopatry of the FerruginousHawk; Steven M. Goodman,museum tour for the birds of Egypt and project;Frederick P. Greene, determinantsof guild structureamong insectivorous birds;Jeffrey G. Groth,systematics of the Red Crossbillin westernNorth America,using vocal, morphological and electrophoreticdata; John M. Hagan, colonialnesting in Ospreys;Dionyssia Hatzilacos, breeding and feeding biology of the White Pelican (Pelecanusonocrotalus) nesting at Lake Mikra Prespa--measuresfor protection;P. Hibbard,John P. Dunning, LaurieM. McKean,and RichardBowers, the efficiencyof foraging behavior:an experimentalapproach; Geoffrey E. Hill, the reproductiveconsequences of sub-adultplumage

(continuedon p. 22) MOONLIGHT AVOIDANCE BEHAVIOR IN LEACH'S STORM-PETRELS AS A DEFENSE AGAINST SLATY-BACKED GULLS

YUTAKA WATANUKI Instituteof AppliedZoology, Faculty of Agriculture,Hokkaido University, Kitaku kita 9 nishi9, Sapporo060, Japan

ABSTRACT.--Diurnalactivity patternsof Leach'sStorm-Petrels (Oceanodroma leucorhoa) and Slaty-backedGulls (Larusschistisagus) were investigated.The petrels reducedactivity in moonlightin May and Junewhen the predationrate by gulls was relativelyhigh. Petrel activitylevels were inverselycorrelated with light intensitiesand the correspondingrisk of predationby the gull. Thissuggests that nocturnalactivity and moonlightavoidance by the petrel in its colonyare an effectivedefense against diurnal predators.Activity synchroni- zationof the petrelwas most marked during the full moon,further supportingthe predator- avoidancehypothesis. Received 15 October1984, accepted 27 April 1985.

MANX Shearwater (Puffinuspuffinus; Harris which are coveredwith Calamagrostislangsdorffii, iso- 1966), Leach's Storm-Petrel (Oceanodromaleu- lated rockstacks, and cliff ledges.A few pairsof Black- corhoa;Harris 1974), Fork-tailed Storm-Petrel (O. tailed Gulls (L. crassirostris)nested on these sites. furcata;Harris 1974,Boersma et al. 1980),and The Slaty-backedGull is an important predator of Cassin'sAuklet (Ptychoramphusaleuticus; Tho- adult petrels on the island. Although Jungle Crows (Corvusmacrorhynchos) excavated petrel burrows and resen 1964, Manuwal 1974) are strictly noctur- ate adults, eggs, and chicks,predation by the small nal in their colonies and are less active on crow population (13 pairs) was not significant. moonlit nightsthan on dark nights.Cody (1973) Activitypatterns.--Observations of flying birdswere discussedthe nocturnal activity of alcids as a made from a blind set on top of a headland about 25 defenseagainst diurnal predators.Gross (1935), m abovesea level. For 5 rain every 30 rain, I counted M.P. Harris (1966), and S. W. Harris (1974) also all birds passingan imaginary 20 x 30-m plane ori- suggestedthat nocturnal procellariiformsare ented vertically with reference to the cliff face op- vulnerableto diurnalpredatory gulls on moon- posite the headland. Two 6-volt electric lights, one lit nights. However, the relationship between set horizontally and the other about 45ø upward at daily activities of procellariiformsand preda- the blind, lit the plane facing toward the sea from tion risk has not been studied. the lower corner to the oppositeside. This reduced the effectof the light on landing birds approaching I studied Leach's Storm-Petrels and Slaty- from the sea.If all the landing petrels were attracted backedGulls (Larusschistisagus) on Daikoku Is- instantaneouslyto the lights, the number of petrels land. Their activity patternsare describedand flying through the lights would increaseduring the the nocturnalbehavior of the petrels in the col- observations. However, the number was rather con- ony is discussedas predator avoidance. stant during 5-rain observations(Fig. 1). A few pe- trelsand gullsflew circularlyin the lights;these were excluded from the data. Observations on moonlit

STUDY AREA AND METHODS nights showedthat Slaty-backedGulls did not avoid the lights. No differential responseto the lights by Thecolony.--The study was conducted on Daikoku gulls and petrelswas assumed.Observations started Island (42ø52'N, 144ø52'E),Akkeshi, Hokkaido, be- in daylight and lasted24 h (n = 19 days).Data taken tween late April and early October1982. The island during a day with dense fog (13-14 May) were ex- is 6.1 km in circumference and treeless, with the ex- cluded from the analysis becauseof low visibility. ception of birch (Betulaermanii) groves in the ravines. Data of 27-28 April were included in those of May. Leach'sStorm-Petrel (the only petrel breedingon the Light intensitywas measured by a lux-meter(+ 10%) island) nestsin the interior partsof the island, which set horizontally and was divided into four classes: is covered with Artemisamontana and Urtica platy- dark (0 lux), moonlight with no cloud cover(<1 lux), philla. Abe et al. (1972) estimatedthat there were twilight (1-5 x 10ß lux), and daylight (>5 x 10ß lux). 1,070,000breeding pairs of petrels,but a more recent Time of sunrise, sunset, moonrise, and moonset were estimate is 415,000 (Watanuki 1985b). About 3,500 from the astronomical tables for Kushiro, about 40 pairsof Slaty-backedGulls nested on maritimeslopes, km west of Daikoku Island. 14 The Auk 103: •4-22. January 1986 January1986] MoonlightAvoidance in Leach's Petrel 15

TABLEI. Breedingschedule of the petrel and its nest duties shown by egg-daysand chick-daysin the study plots.

May June July Aug Sept No. of eggslaid 0 335 16 0 0 No. of chicks hatched 0 0 199 64 0 Egg-days 0 5,785 6,305 390 0 Chick-days 0 0 1,965 7,220 6,935

marked by individually numbered rings. Their nests were checkedevery day between4 and 9 July to de- termine the length of incubation stints. I estimated I I I I i I I I I chick feeding frequency by weighing 15 chicks at 1 2 3 4 5 0900 and 2100 on 9 August, and at 0300 and 1500 on 10 August. A feeding was assumedto have occurred Time after lighting (min.) when a chickgained weight during this interval.This is a minimum estimatesince more than one feeding Fig. I. Number of flying Leach'sStorm-Petrels per may occurin the interval. Night nest attendancein 30 s in relation to time after lighting by two 6-volt the prelaying period was estimatedby checking if lights, one set horizontally and the other about 45ø toothpicksplaced at the entranceof the active bur- upward. The 5-rain experimentwas repeatedthree rows the previousday were moved. times. The significanceof the correlation coefficient(r)' and the differencebetween values was examinedby ANOVA (Sokal and Rohlf 1969). Overlap of petrel and gull activity was assessedby Pianka'soverlap index (Pianka 1973): RESULTS O• = ß xiyi/(g x? ß y?)% Breedingof petrels.--Petrels arrived in late where x, = the number of petrels flying per 5 rain and y• = the number of gulls flying per 5 min. The April and laid eggs between early June and time dispersionof petrel activity between 1930 and early July. Chicks hatched between mid-July 0230 was determined by Morishita's If index (Mor- and mid-Augustand fledged between late Sep- ishita 1959): tember and mid-October.The incubation duty 15 = K[• x,(xi - I)]/2 x,(• x, - 1), as shown by egg-dayswas high in June and July, and the chick feeding duty shown by where xi = the number of petrels flying per 5 rain chick-dayswas high in August and September and K = the number of samplings(K = 15 in this (Table 1). case). Breeding Leach'sStorm-Petrels visited their Predationby gulls.--I estimated the number of nests0.6 times/night (a total of 63 nestsvisited Leach's Storm-Petrelseaten by Slaty-backedGulls from the energyrequirements of gulls and their food out of 105 nestschecked) in the prelaying pe- composition.The food habits of the gulls were de- riod, fed a chick at least 1.2 times/night (n = termined by collecting pellets and food remains at 15),and exchangedincubation duties at 2.9 days their nestingsites. During the study,collections were (n = 17) on the average.These figures agree made every 5 days in 4 study quadratswith a total with those of other observers (Gross 1935, Wil- of 114 nests,yielding 1,143samples. The energyre- bur 1969). quirementsof Slaty-backedGulls were estimatedby Light intensityand activities.--Leach'sStorm- usinga simple modificationof the model of Furhess Petrels arrived at Daikoku Island about 1 h af- (1978).Calculations of energyrequirements and food ter sunsetand left 1-2 h before sunrise(Fig. 2). compositionof the gulls are shown in Appendices1 and 2. The number of petrels killed was calculated The mean number of flying petrels and gulls for 5-day periodsand was combinedmonthly. was related to light intensity (Fig. 3). Petrels Petrelbreeding.--Ten 4 x 10-m study quadratscon- were nocturnal and rarely flew in twilight or taining a total of 351active petrel burrowswere used. daylight on the island. More petrels flew in I checkedthe nestsevery 5 days and recordedthe darknessthan in moonlightin May (F•,• = 6.50, contents.Twenty-five pairs incubatingeggs were P < 0.05) and June (F•,42= 7.79, P < 0.01); re- 16 YUTAKAWATANUKI [Auk,Vol. 103

20

10

20

-10

Sep i!i -2O -10

-20 •

-10

-lO

18 O0 06 18 oo 06 18 oo 06

Time of day (hrs) Fig.2. Daily changesin the numberof flyingpetrels (dark lines) and gulls (light lines)per 5 min. Heavily shadedareas represent dark times,and lightly shadedareas represent moonlight times.Vertical lines indicate sunrise and sunset. duced activity in moonlight also occurredin killed immediately by Slaty-backedGulls and July, although the difference was not signifi- 1 by Jungle Crows.This suggeststhat there is cant (F,,56= 3.26, NS). Petrels did not avoid considerablepotential predation risk in twi- moonlight in August (F,,46= 0.30, NS) and Sep- light and daylight hours. tember (F•,?2= 3.51, NS). The mean number of The time overlap of petrel and gull activity, petrels flying at each light intensity was neg- shownby lnOxy,was positivelycorrelated with atively correlatedwith that of the gulls (r = the length of the period of moonlight (h; r = -0.662, n = 20, F•,•8= 14.4, P < 0.01; Fig. 3). 0.812, n = 15, F•,• = 21.29, P < 0.01; Fig. 4) Slaty-backedGulls rarely flew in the dark when the data of the dayswith no overlap (i.e. (Figs.2, 3). More gulls flew in moonlight than Oxy = 0) were excluded.Encounter-rate indices in darknessin all the months(May: F•,, = 56.13, (number of petrels x number of gulls) were P < 0.01; June: F•,• = 12.33, P < 0.01; July: calculateddirectly from Fig. 3 and were in the F•,s6= 25.3, P < 0.01; August: F1,46= 5.13, P < order moonlight > dark > twilight > daylight. 0.05; September:F•,?2 = 4.09, P < 0.05). Gull I do not have direct evidence of predation un- activity levels were higher in twilight than in der moonlight, although the indirect evidence moonlight in all months (May: F•,7s= 26.96, above corroboratesGross's (1935) finding that P < 0.01; June: Fi,s• = 36.15, P < 0.01; July: mostencounters between gulls and petrels oc- F•,• = 81.96, P < 0.01; August: Fl,•5 = 79.12, cur during moonlight. P < 0.01; September:Fl,7O = 168.71, P < 0.01) Activitypatterns of petrels.•Most petrel activ- and higher in twilight than in daylight in May ity occurredat about2400 or between2200 and (F,,m --- 34.62, P < 0.01) and June (F•,•s= 11.96, 0100on nightswith a new moon,0130 on nights P < 0.01). with a full moon, 2200 on nights with a half- I observedLeach's Storm-Petrels flying in moon rising at midnight, and 0100 on nights daylight 11 times. Six of these petrels were with a half-moon setting at midnight (Fig. 2), January1986] MoonlightAvoidance inLeach's Petrel 17

o o

o o o o o

Jun.

lOO

5o

<• Jul. 43 lOO

5o 1 2 3 4 5 6 7 8 9 10 Moonlight time (hrs)

Aug. 33 Fig. 4. The correlationbetween length of moon- light and activity overlap of petrels and gulls (log•0Oxy).

ratherthan reactingdirectly to the presentlight 1001Sep. 26 intensity. On 13-14 May, a half-moon night 28 46 (moonriseat 2300) with densefog, the peak of ,.r'--• I '•"•:'• I J petrel activity occurredat 0030 rather than 2200, D M T as would be expectedif there had been no fog Light intensity (the data of this day were excluded in the nu- merical analysis).It seems,therefore, that pe- Fig. 3. Mean number of flying petrels (shaded bars)and gulls (open bars) per 5 rain under dark (D), trel activity is not controlledonly by endoge- moonlight (M), twilight (T), and daylight (DL) con- nousrhythms keyed to lunar cycles,as reported ditions. Samplesizes of gulls and petrelsare the same in fruit bats (Morrison 1978). Because of the and are shown above the bars. low visibility on 13-14 May, petrelsmight have found it more difficult to locate the island and thus delayed their arrival. This suggeststhat as has been reported in Cassin'sAuklet (Ma- petrel activity at the colony was influenced by nuwal 1974) and in Leach's Storm-Petrel in light intensity, anticipation of the lunar cycle, California (Harris 1974). The petrels on Dai- and weather conditions. koku Island arrived later on full-moon nights On nights with a full moon, petrel activity (5-6 May, 5-6 June, and 5-6 August) and ear- was highly synchronized in May, June, and lier on nights when the half-moon roseat mid- August but was dispersedin September(Fig. night (10-11 June,8-9 July, and 11-12 August) 2). The synchronyof petrel activity at night than they did on nights without a moon (27- was positively correlated to the length of 28 April, 23-24 May, 19-20 July, and 15-16 Au- moonlighttime in May, June,July, and August gust; Fig. 2). (r = 0.843, n = 14, F•,•2= 29.47, P < 0.01) but On nights in May, June, and July with quar- not in September(Fig. 5). ter or half-moons,the mean number of petrels Predationby gulls.--The number of flying gulls flying per 5 min during darkness(œ = 155.6, remained high in May, June, and August, but n = 42) was significantlyhigher than that (œ= decreasedin September(F4,•3 = 5.45, P < 0.01; 100.0,n = 30) on moonlessnights in the same Table 2) because nonbreeders and unsuccessful months (F•,3o= 37.47, P < 0.01). The petrels, breedersleft the colonyin Augustand Septem- therefore, seemedto anticipate the lunar cycle ber. The total number of petrels was large in and synchronizetheir activity to the hours of May, June,and July, decreasedin August, and darknesson clear nights with a half-moon, was small in September(F4,•3 = 8.59, P < 0.01; 18 YUTAKA¾VATANUKI [Auk, Vol. 103

around the colony is crucial in avoiding gull 2.0 predation. Petrel adults,eggs, and chicksin the nest burrows are protected from gull preda- tion, and petrel foraging areasdo not overlap those of gulls (Pearson 1968, Wiens and Scott 1975). Therefore, only adult petrels flying on or near coloniesare subjectto gull predation. The negative correlationbetween petrel and 1.5 gull activity levels (Fig. 3) indicatesthat pred- ator avoidancemay be the leading causeof pe- trel nocturnal behavior at Daikoku. Petrels avoided moonlight in May and June when the predationrate was relatively high, but they did not avoid it in Septemberwhen the predation 1.0 I • I I I I I rate was minimal (Fig. 3, Table 3). The partial 1 2 3 4 5 6 7 8 9 10 correspondencebetween seasonal change in predation rates and seasonal variation in Moonlight time (hr•) moonlight effects on petrel activity further Fig. 5. The correlationbetween length of moon- supports this hypothesis. light and synchronyof petrel activity shown by the The medium synchronyof petrel activity on If index. ß = May, ß = June,ß = July, ß = August, nights with half-moons (Fig. 5) probably was and O = September.Regression line a is for the data the result of short periods of darknessin which of May, June,July, and August;b is for thoseof Sep- petrels were free from predation. The greatest tember. activity synchrony occurredon nights with the highest predation risk (full-moon nights in May, June, and August) and the least syn- Table 2). The number of petrels flying at night chrony on the night without predation risk was positively correlated with the dry-weight (September;Fig. 5). Numbers of flying gulls percentageof petrel remains (Appendix 2) in were relatively constant on the full-moon the gull diets sampledin each5-day period (r = nights comparedwith those of petrels (Fig. 2). 0.886, n = 18, Fz,z6= 58.42, P < 0.01). Therefore, highly synchronouspetrel activity Monthly changesin petrel numbers, num- on these nights may dilute the predator's ef- bers killed by gulls, and predation rates are fects by minimizing the risk that a particular shown in Table 3. Gulls ate many petrels in petrel encountersa gull (Wilson 1975: 41-42). May and June, more in July, fewer in August, Nest duties.--Although I could not distin- and few in September,a pattern matching the guish nonbreeders from breeders during ob- seasonalcycle of estimatedgull energyrequire- servations, nonbreeders were inferred to be ments (Appendix 1) and food composition(Ap- abundant in June and July (Ainley et al. 1974), pendix 2). Thus, the monthly petrel kill ap- and probablyin May, becausethe total number pears to vary with gull energy requirements of flying petrels was high in thesemonths (Ta- and availability of the prey. ble 2). Moonlight avoidance behavior disap- peared in August and September; therefore, DISCUSSION only nonbreeders may avoid moonlight. Be- causenonbreeders do not have as many nest Predatoravoidance.--Slaty-backed Gulls are dutiesas breeders, the threat of predationwould the most important predators of adult petrels be a crucial factor determining their activity on Daikoku Island. I estimatedthat annual pre- pattern. Eight of 12 pairs exchangedtheir in- dation by the gulls approximates13.2% of all cubation duties on a clear, moonlit night (8-9 adult petrels. This appearsto be a potentially July). This implies that moonlight does not re- strong selectionpressure on Leach'sStorm-Pe- strict breeders' activities during the incubation trel, whose adult survival rate is high (Hun- period, although more data are needed to sub- tington and Burtt 1970). stantiatethis. Harris (1966) reported moonlight The daily activity pattern of adult petrels avoidance behavior in nonbreeding Manx January1986] MoonlightAvoidance in Leach'sPetrel 19

TABLE2. Monthly changein the meanof the total numberof flying birdsin 24 h.

May June July Aug Sept No. of daysobserved 4 3 4 3 4 No. of petrels(•) 1,330.3 1,516.7 1,493.8 861.3 187.0 No. of gulls(•) 316.5 340.7 350.5 305.7 152.3

Shearwaters, which were killed by Herring avoidance behavior (Harris 1974). On the other Gulls (L. argentatus)and Great Black-backed hand, Audubon'sShearwaters (Puffinus lhermi- Gulls (L. marinus)(Harris 1965). nieri)and GalapagosStorm-Petrels (Oceanodro- On the other hand, seasonalchanges in the raatethys) are diurnal on their breedingisland, nest duties of breeders may contribute to the where the most important predator is a noctur- seasonal changes in the petrel response to nal or crepuscularowl (Asiogalapagoensis; Har- moonlight. Breeding petrels returned to the ris 1969a, b). Thus, although the predator- colony more frequently in the nestling period avoidancehypothesis generally is supported, than in the incubationperiod. The relative im- further investigationson intercolonial and in- portanceof nest dutiesin the prelaying period terspecificvariation of predationrisk and diur- may be small becauseprelaying exodusis sug- nal activity patterns are needed. gestedin Leach'sStorm-Petrel (Gross 1935, Wa- Foraging.--Grubb (1974) suggested that tanuki 1985b). I assumed,therefore, that nest- abundant food in the ocean'supper waters at duty activity was seasonaland correlatedwith night has been a strong selectiveforce toward the growth of chicks(Table 1). High predation nocturnal activity of Leach's Storm-Petrelsat risk and low nest duties seemed to cause breed- sea.Food of Leach'sStorm-Petrels breeding on ers to avoid moonlight in May and June, but Daikoku Island may be most abundant at night increasingnest duties in July might have taken (Watanuki 1985a). The petrels stay on the is- priority over the high predation risk and con- land for part of the night and thus lose some sequently changed breeders' response to opportunities for feeding at night when prey moonlight. is abundant.A pair exchangesincubation du- B•dard (1976) criticized the predator-avoid- ties at night every 2.9 days, and hence it loses ance hypothesis(Cody 1973) becausethere is one night of feeding opportunitiesabout every intensepredation on somealcid coloniesin the 3 days. Missed feeding opportunities seem to Arctic, where daylight is continuousduring the be potentially crucial becausea petrel must re- summer. On South Orkney Island (60ø43'S, cover 11% of body weight lost during an in- 45ø38'W),where the period of darknessis short cubation stint (Watanuki 1985b). In addition, a in summer, the Black-bellied Storm-Petrel (Fre- pair feeds a chick at least 1.2 times per night gettatropica) is preyed on by the diurnal Great and thus losesmore opportunities. Skua (Catharactaskua) and shows nocturnal ac- Sensoryconstraints probably affectforaging tivity (Beck and Brown 1971). Leach's Storm- time. If Leach'sStorm-Petrels rely heavily on Petrels on Kent Island (44ø02'N, 124ø09'W) and visual cuesto locatefood, they would have the on Little River Rock (41ø02'N, 124ø09'W) also advantage of feeding during the day or on a are killed by diurnal gulls and show nocturnal moonlit night. Leach'sStorm-Petrels can locate activity (Gross 1935) as well as moonlight food by smell at night as well as during the

TABLE3. Monthly changein the numberof petrelspotentially available, number killed by gulls, and pre- dation rate.

May June July Aug Sept No. of petrels' 987,700 908,700 760,500 589,300 547,800 No. killed 36,676 29,376 49,065 15,176 321 Predation rate (%) 3.7 3.2 6.5 2.6 0.1 Includesnonbreeders in May, June,and July.Nonbreeders were assumedto be 16%of the total population (Wiens and Scott 1975). Unsuccessfulbreeders were assumedto have left the colony as soon as they failed. 20 YUTAKAWATANUKI [Auk,Vol. 103 day (Grubb 1972).Gordon (1955) suggestedthat GRUBB,T. C. 1972. Smell and foraging in shear- they are nocturnalfeeders, although they also waters and petrels. Nature 237: 404-405. feed during the day (H. Ogi pers. comm.). Di- 1974. Olfactory navigation to the nesting rect testsof Grubb'shypothesis are lacking. In- burrow in Leach's Petrel (Oceanodromaleucor- hoa). Anita. Behav. 22: 192-202. direct evidencesuggests that the petrelsare not HARRIS,M.P. 1965. The food of some Larusgulls. obligatorynocturnal feeders and that abundant Ibis 107: 43-53. prey at night is not always an important factor --. 1966. Breedingbiology of Manx Shearwater affecting the petrels' nocturnal activity at the Puffinuspuffinus. Ibis 108: 17-33. breedingcolony. In conclusion,predation risk 1969a. Food as a factor controlling the and nest duties seem to be more important fac- breeding of Puffinuslherminieri. Ibis 11I: 432-456. tors determining petrel activity at the colony ß 1969b. The biology of storm petrels in the than food availability. GalfipagosIslands. Proc. California Acad. Sci. 37: 95-166.

ACKNOWLEDGMENTS HARRIS,S.W. 1974. Status,chronology and ecology of nesting storm-petrelsin northern California. I thank the Akkeshi Marine BiologicalStation for Condor 76: 249-261. the use of its facilities and H. Uno for his excellent HUNT, G. L., JR. 1972. Influence of food distribution logisticalsupport in the field. H. Abe and A. C. Tho- and human disturbanceon the reproductivesuc- resen made valuable commentson early drafts of the cessof Herring Gulls. Ecology53: 1051-1061. manuscript,and the Boardof Educationof Akkeshi- HUNTINGTON,C. E., •t E. H. BURTT,JR. 1970. Breed- cho gave permissionto work on the island. ing age and longevity in Leach'sPetrel Ocean- odroma leucorhoa. Abstr., 15th Intern. Ornithol. LITERATURE CITED Congr.: 663. MANUWAL,D. A. 1974. The natural history of Cas- ABE, M., N. ICHIDA, H. SHIMIZU, M. HASHIMOTO, O. sin'sAuklet (Ptychoramphusaleuticus.). Condor 76: YUNOKI,N. OZAWA,• S. OZAWA. 1972. [A new 421-431. attempt to estimatethe number of nestingpetrel, MORISHITA,M. 1959. Measuring of dispersionof Oceanodromaleucorhoa leucorhoa,and wildlife of individual and analysis of distribution pat- the Daikoku Island, Hokkaido.] Tori 21: 346-365. terns. Mere. Fac.Sci., Kyushu Univ. Set. E, Biol. AINLEY, D. G., S. MORRELL,& T. J. LEWIS. 1974. Pat- 2: 215-235. tern in the life historiesof storm petrelson the MORRISON,D.W. 1978. Lunar phobia in a neotrop- Farallon Islands. Living Bird 13: 295-312. ical fruit bat Artibeusjamaicensis (Chiroptera: BECK,J. R., & D. W. BROWN. 1971. The breeding Phyllostomidae).Anita. Behav.26: 852-855. biology of the Black-bellied Storm-Petrel Freget- PEARSON,T.H. 1968. The feeding biology of seabird ta tropica.Ibis 113: 73-90. speciesbreeding on the Fame Islands, North- B•DARD,J. 1976. Coexistence, coevolution and con- umberland. J. Anita. Ecol. 37: 521-552. vergent evolution in seabird communities: a PIANKA, E. R. 1973. The structure of lizard com- comment.Ecology 57: 177-184. munities. Ann. Rev. Ecol. Syst. 4: 53-74. BOERSMA,P. D., N. T. WHEELWRIGHT,M. K. NERINI, •t SIBLY, R. M., & R. H. McCLEERY. 1983. Increase in E. S. WHEELWRIGHT.1980. The breeding biology weight of Herring Gulls while feeding. J. Anita. of the Fork-tailedStorm-Petrel (Oceanodroma fur- Ecol. 52: 35-50. cata). Auk 97: 268-282. SOKAL,R. R., & F. J. ROHLF. 1969. Biometry. San BRISBIN,I. L., JR. 1968. A determination of the ca- Francisco, W. H. Freeman and Co. loric densityand majorbody componentsof large THORESEN,A.C. 1964. The breeding behavior of the birds. Ecology49: 792-794. Cassin Auklet. Condor 66: 456-476. COPY, M. L. 1973. Coexistence, coevolution and convergent evolution in seabird communities. WATANUKI,Y. 1984. Food intake and pellet of Black- Ecology 54: 31-44. tailed Gull, Larus crassirostris.J. Yamashina Inst. FURNESS,R. W. 1978. Energy requirementsof sea- Ornithol. 16: 168-169. bird communities:a bioenergeticmodel. J. Anita. ß 1985a. Food of breeding Leach'sStorm-Pe- Ecol. 47: 39-53. trels (Oceanodromaleucorhoa). Auk 102: 884-886. GORDON,M. S. 1955. Summer ecology of oceanic ß 1985b. Breedingbiology of Leach'sStorm- birds off southern New England. Auk 72: 138- Petrels Oceanodroma leucorhoaon Daikoku Island, 147. Hokkaido,Japan. J. YamashinaInst. Ornithol. 17: GROSS,W. A. O. 1935. The life history cycle of 9-22. Leach's Petrel (Oceanodromaleucorhoa leucorhoa) WIENS,J. A., •t J. M. SCOTT. 1975. Model estimation on outer sea islands of the Bay of Fundy. Auk of energy flow in Oregon coastalseabird popu- 52: 328-399. lations. Condor 77: 439-452. January1986] MoonlightAvoidance inLeach's Petrel 21

WILBUR,H.M. 1969. The breedingbiology of Leach's ments were estimatedfrom foraging time, percent- Storm-Petrel Oceanodroma leucorhoa. Auk 86: 433- age of flapping flight, and energy costsof flapping 442. and gliding flight. Foraging times of nonbreeders WILSON,E. O. 1975. Sociobiology,a new synthesis. were not measured, but were assumed to be similar Cambridge,Massachussetts, Belknap Press. to that of breeders.Existence energy requirementsof adults and chicks were estimated from their body weight and the mean daily ambient temperature. APPENDIX 1 I counted the number of adults on breeding sites and surrounding areas about every 3 weeks. About Calculationof energyrequirements of the Slaty-backed 5,000 birds were on the breeding sites in June and Gull.--A model similar to Furness's (1978) was used July. Nest-sitetenacity of parentsduring daytime in to estimate the energy requirement of Slaty-backed these months was about 70%;accordingly, the num- Gulls.Egg-laying, hatching, chick survival, and chick ber of breeders was estimated to be about 7,000. About growth were monitored at 5-day intervals in the 4 2,000 birds were in surrounding areasand were as- study quadrats(Fig. 6). Mean daily energy require- sumed to be nonbreeders at their maximum number meritsfor egg production,chick existence,and chick in late May. growth at 5-day intervals were then estimated.The Other species-specificparameters are: egg weight averagepercentage of time spent foraging away from at laying, 107 g (n = 25); adult weight, 1,100 g (n = the nest site by adultswas determinedby monitoring 1); arrival of last breeder, 1 May; departure of first 10-20 pairs at nest sitesevery 30 rain at night using breeder, 30 July; departure of last breeder, 25 Sep- a starlight scope.Birds away from the nestswere as- tember;date of peak number of nonbreeders,26 May. sumedto be foraging in flight. Seasonalchanges in The following dates were assumed:arrival of first the foraging time of breedersare shown in Fig. 6b. breeder, 1 March; arrival of first nonbreeder, 1 April; Percentageof flapping flight, measuredby scanning departureof last nonbreeder,30 July. Model-specific flying birds for 1 rain every 5 rain in various wind parametersare the sameas in Furhess(1978). conditions,averaged 50%. Breeders'activity require- APPENDIX 2

Foodof Slaty-backedGulls.--Pellets and food re- Laying mains were dried to constant weight at about 20øC • 100t JkA Hatching, • Fledging g50[ /" 90 80

70

80

50

4O

30 /

70

• •o 6O

Apr, May Jun, Jul, Aug. Sep. Fig. 6. Breeding phenology and energy require- mentsof Slaty-backedGulls. (a) Seasonalchanges in egg-laying, hatching, and fledging; (b) number of May Jun Jul Aug Sep hoursof foraging activity per day;(c) energy require- Fig. 7. Seasonalchanges in diet of Slaty-backed ments.The curvesin (c) representthe increasingen- Gulls shown by dry-weight percentages.(a) Adult ergy requirementsof nonbreeders,chicks, breeders, Leach's Storm-Petrel; (b) fish (dark line) and other and the total population. items (broken line). 22 YUTAKAWATANUKI [Auk, VoL 103 for at least30 days.Adult Leach'sStorm-Petrels, fish, fish, 1,500; crab with shell, 700; shellfish with shell, shellfish, crabs, and chickens were the main food 500; chicken, 3,000 (Brisbin 1968, Hunt 1972, Sibly items.The following equationwas used to determine and McCleery 1983).R•s were assumedto be the same the total wet weight of the ith food: W• = (D• x R• x as those of Black-tailed Gulls (Larus crassirostris)and TER)/(• D•R.C,),where D, = dry weight of the ith food, the following: adult Leach'sStorm-Petrel, 5.9; fish, Ri = regurgitation coefficient of the ith food (show- 317.3; crab, 6.7; shellfish, 2.2; chicken, 30.0 (Watanuki ing the ratio betweenwet weight of intake and dry 1984). The number of petrels killed by gulls was weight of pellets), TER = total energy requirement calculatedby dividing the total petrelwet weight by of the gulls for 5 days,and C• = caloricvalue of the the averageindividual adult weight (48 g, Watanuki ith food. Cis (kcal/kg wet weight) were assumedto 1985b).Seasonal changes in the gulls' diet composi- be the following: adult Leach'sStorm-Petrel, 2,600; tion are shown in Fig. 7.

(continuedfrom p. 13) in male Black-headedGrosbeaks; Adan (Hussein)Isack, biology of the GreaterHoneyguide (Indicator indica- tor); Dr. Pedro Jordano,pattern of fruit use by wintering frugivorous birds and their implications for bird- dispersedplants in Mediterranean habitats;Frank J. Joyce,nest site selectionby three passerineassociates of Hymenoptera; Donald M. Kent, foraging strategiesof Snowy Egrets (Egrettathula) in the salt marshesof Massachusetts;Roni King, winter territoriality in migratory European Robins (Erithacusrubecula)--habitat selectionand winter survivorship;Francis R. Lambert,co-adaptation between frugivorous birds and fig trees in Malaysianlowland forest;Jeanette L. Lebell, microgeographicvariation in the flight whistle of the Brown- headed Cowbird; David E. Manry, a preliminary study of blood and feather pulp proteins in two South Americanibis (Threskiornithidae)species; Jean-Louis Martin, the population structureof Paruscaeruleus L. (Aves):geographical variation and speciationin the Mediterranean region; Jon Miller, breeding distribution and origin of Water Pipits in the Sierra Nevada;David Morimoto, effectsof forest fragmentationon avian communitystructure and species-habitatrelationships in the southeasternMassachusetts Pine Barrens;Jay Pitocchelli,speciation in the genus Oporornis;Richard O. Prum, courtshipbehavior and ecologyof Masius chrysopterusin Ecuador;Dr. Michael R. W. Rands,the breedingbehavior and habitatof the Arabian Bustard, Ardeotisarabs; Pamela C. Rasmussen,relationships of Fuego-PatagonianBlue-eyed Shags; Mark D. Reynolds, socialbehavior of Yellow-billed Magpies;Dr. Gary Ritchison,the significanceof song repertoiresin the Northern Cardinal; Jeffrey A. Schwartz, vocal similarity in mated Ring-billed Gulls as an aid in parental recognitionby young; Dr. Ron Scogin, floral color and hummingbird vision; Peter E. Scott, the nesting ecologyof desert hummingbirdsin relation to the pollination ecologyof certain nectar plants; Patricia Serrentino,the breedingecology and behaviorof the Northern Harrier (Circuscyaneus) in CoosCounty, New Hampshire;Laurie J. Stuart-Simons,food limitation in birds;Richard R. Snell, hybridization,isolation and speciesrecognition in arcticgulls; Carol Spaw, thick-shelled eggs and their evolutionaryimplications; Charles Sullivan, nest behaviorand developmentof young in the JabiruStork; Dr. Kimberly A. Sullivan, energetics and the development of time-budgeting;Richard John Watling, investigation of statusof Ogea Flycatcher and Blue-crownedLory; Dr. G. CauseyWhittow, water lossfrom eggsof Great Frigatebird(Fregata minor); Dr. David Winkler, a general model of parental care with experimentaltests on the Tree Swallow; John L. Zimmerman, Ortstreuein Henslow's Sparrows (Ammodramushenslowii) and movementsin responseto spring burning of tallgrassprairie. RELATIONSHIPS AMONG TERRITORY SIZE, HABITAT, SONG, AND NESTING SUCCESS OF NORTHERN CARDINALS

RICHARD N. CONNER, 1 MARY E. ANDERSON,2'3 AND JAMESG. DICKSON1 •WildlifeHabitat and SilvicultureLaboratory, Southern Forest Experiment Station, USDA ForestService, Nacogdoches, Texas 75962 USA, and 2Departmentof Biology,Stephen F. AustinState University, Nacogdoches,Texas 75962 USA

A13STRACT.--Wecollected data from 30 territories of Northern Cardinals (Cardinaliscardi- nalis)during 3 breedingseasons (1979-1981) in easternTexas. Territory size waspositively correlatedwith tree height and negativelycorrelated with the densityof foliageat ground level and at 3 m abovethe ground,and with the densityof shrubs.Nesting success was positivelycorrelated with presenceof patchyunderstory foliage and arthropodbiomass in territories.Cardinals with low songcomplexity and shortersongs held better-qualityterri- toriesand had better nestingsuccess than cardinalswith more complexsongs. We suspect that youngmales may use long, complexsongs to establisha territoryinitially, and in subsequentyears put lesstime and effortinto songand more into careand defenseof young. Received15 January1985, accepted7 June1985.

THEfunctional significance of songin birds repertoiresalso had the highest successin ex- is a widely studiedphenomenon. In particular, cluding other malesand attractingfemales. If the significanceof songrepertoires has recent- maleswith largerrepertoires or someother song ly receivedextensive examination, resulting in characteristic are able to establish and maintain many hypotheses.Krebs (1977a) summarized larger or better-qualityterritories, such males thesehypotheses as follows:(1) repertoiresen- shouldhave a highernesting success than males hance individual recognition (Emlen 1971, with territoriesof lesserquality. High-quality Brooks and Falls 1975); (2) repertoires are a territoriesshould have a greateravailability of product of sexual selection (Catchpole 1973, food and nest concealmentthan low-quality Kroodsma 1977); and (3) repertoires increase territories. Females from an inbred strain of successin territorial competitionby allowing canaries(Serinus canaria) exposed to playbacks acquisitionof larger (Krebs 1977b) or better- of maleswith large repertoiresbuilt nestsfast- quality (Howard 1974) territories. er and laid larger clutchesthan did femalesex- Large repertoires could permit matched posedto small repertoires(Kroodsma 1976). countersingingamong males establishingor Female birds may select males on the basis defendingterritories (Lemon 1968, Verner 1975) of songquality or selectthe territory of a male or reduce habituation of listeners (Hartshorne becauseof the habitat'sappearance through 1956). Krebs (1977a) suggesteda new mecha- mechanismssimilar to those by which males nism(the BeauGeste hypothesis) whereby large may identify habitatquality. If either method song repertoires could be used to create the wasused, pairs in higher-qualityhabitat (more illusion that a particular habitat area was al- food and betterconcealment available) might readysaturated with numerousterritorial males. be expectedto producemore eggs or youngon Holding territory size constant, repertoire the averagethan pairs in lower-qualityterri- size of Northern Mockingbirds(Mimus poly- tories.Although predation would have a sub- glottos)in central and western Texas was cor- stantialimpact on nestingsuccess (Best 1978), relatedwith measurementsof territory quality vegetationstructure or quality may also have (Howard 1974).Male mockingbirdswith larger an influenceon predationand nestingsuccess of birds (Bestand Stauffer 1980). Vegetation characteristics around nests were correlated • Present address:Math and Science Division, Bee with nestingsuccess of VesperSparrows (Pooe- County College, Beeville, Texas78102 USA. cetesgramineus) in a 1-yr studyin WestVirginia 23 The Auk 103:23-31. January1986 24 CONNER,ANDERSON, AND DICKSON [Auk, Vol. 103

T^I3LEI. Characteristicsof cardinal songsand I00 randomly selectedsyllables from 30 territories over a 3-yr period in easternTexas.

Variable code Description Mean SD Songs ASYB No. of syllablesper song 10.45 2.49 DSYB No. of different syllablesper song 1.61 0.23 TRAN No. of syllabletransitions per song 0.62 0.23 SBVS Total versatility of syllablesper song 0.08 0.08 SLGT Songlength (s) 2.78 0.59 ISI Intersonginterval (s) 6.33 1.33 NSGS No. of songsrecorded per bird 311.27 267.87 SBRP Syllablerepertoire per male 15.50 4.52 STRP Song-typerepertoire per male 18.67 8.16 SGVS Total versatility of songtype per male 7.62 2.90 RESB Residualof syllablerepertoire -0.22 2.76 REST Residualof song-typerepertoire -0.16 4.26 Syllables FMOD Frequencymodulation (degree) 46.20 5.82 FEMP Frequencyemphasized (kHz) 2.2O O.42 HIFR High frequency(kHz) 4.19 0.58 LOFR Low frequency (kHz) 1.33 0.24 FRGE Frequencyrange (kHz) 2.87 0.56 SYBL Syllable length (s) 0.24 0.04 ISBI Intersyllableinterval (s) 0.13 0.02 NELM No. of elementsper syllable 2.84 1.26

(Wray and Whitmore 1979).Concealment, nest shortleaf pine (P. echinata),loblolly pine, post oak height, and adult body weight were also sig- (Quercusstellata), and sweetgum (Liquidambarstyraci- nificantly related to nesting successin a ripar- fiua).Dominant tree species in the matureforest along ian bird communityin Iowa (Bestand Stauffer the streamswere sweetgum,American beech (Fagus 1980). grandifolia),red maple (Acerrubrum), sweetbay (Mag- We examined relationships of territory size noliavirginiana), and hickory(Carya spp.). Winged su- mac (Rhuscopallina), smooth sumac (R. glabra),and and habitatquality with songcharacteristics and sweetgumwere dominant in the pine plantation. nestingsuccess of Northern Cardinals(Cardi- Songrecording.--We recorded songs of 30 territorial naliscardinalis) in eastern Texas to evaluate as- male cardinals from the end of March through the sociationsof songwith habitatand nestingsuc- middle of June with a Uher 4000 report IC tape re- cess. corder and a 40-cm Dan Gibson Parabolic EPM P-650 electronicmicrophone. The time spanover which we METHODS recordedsongs enabled us to samplesongs of all males throughoutall phasesof the nesting season.Males Studyarea.--We selecteda diverse"edge" stand (15 were color marked (colored bands and tarsi stream- ha) for the study in a pine (Pinusspp.)-hardwood ers) for individual identification. To minimize sound forest 15 km southof Nacogdoches,Texas. Northern distortion, we made most recordingswithin 8-15 m Cardinalswere studied in this area during three con- of males.We did not use playbacksof song to elicit secutivebreeding seasons(I 979-198I). Approximate- singingbehavior. Only maleswith 89 or more songs ly half of the standwas mature forest (18-22 m tall); recordedwere used for analyses(n = 30 males). most of the remainder was a 4-yr-old loblolly pine All field recordingswere reviewed at one-quarter (Pinustaeda) plantation with densepine and decid- tapespeed (2.4 cm/s);syllable types (see Lemon 1965) uousfoliage that was2-3 m high in 1979.The mature were initially determinedand identified at this re- forestand the pine plantationhad patchyfoliage and duced speed. The different syllable patterns were rangedfrom xeric,sandy hilltops to roesicsites along verifiedwith spectrogramsmade on a Kay Elemetrics two streams.The streamsran through forestedstrips model 7029A Sona-Graph,using the wide band filter (30-50 m wide) in the pine plantation.Several small setting with a frequency range of 80-8,000 Hz. In- openings(0.45 ha total) in the matureforest had a formation measured included the kinds and numbers ground cover of grassesand forbs. Dominant tree of syllablesin eachsong, song length, and intersong speciesin uplandportions of the matureforest were interval (Table I). Song lengths and intervals were January1986] CardinalHabitat, Song, and Nesting 25 measuredto the nearest0.1 s with an electronicstop- songvariation and the total number of syllablesin a watch. song. The formula for syllable-type versatility is One hundred syllablessung by eachterritorial male written as: cardinal were sampled randomly and spectrograms syllable-typeversatility = (a x b)/c, were made of these 3,000 syllables.We made 8 mea- surements on each syllable: frequency range, fre- where a = number of different syllables/song,b = quency modulation, frequency emphasized,highest number of syllable transitions/song,and c = total frequency, lowest frequency, syllable length, num- numberof syllables/song. ber of elementsin each syllable, and the following In addition to syllable-type versatility, we calcu- intersyllable interval (after Morton 1975; Table 1). lated total versatility of song types per male (SGVS) We measuredthe frequencyemphasized at that point and transitionsbetween song types as describedby in the syllablewhere mostof the energy was focused, Kroodsmaand Verner (1978) (Table 1). Song types as indicatedby the darkestportion of each syllable were listed in sequence,and 10-songsequences were on the spectrograms. usedto calculatetotal songversatility. If fewer than Frequencymodulation of syllableswas determined 11 songswere sung in a bout, total song versatility geometrically as an angular measurementof the rate was not calculated. of frequencychange. Each syllable was divided into Territorymapping.--We delineated the territoriesof four equal segments,with five vertical lines; this pro- Northern Cardinalsby intensivespot-mapping, not- vided five possible points of intersection with the ing the outmostlimits where each male sang(Inter- syllable. At each point, the angle of inclination was national Bird Census Comm. 1970), by observing measuredin degreeswith respectto the horizontal movementsof males in great detail, and sometimes axis. Eachangle was recordedas an acute angle, dis- by flushing males.Male and female cardinalswere regarding the quadrant in which it fell. This simpli- captured at 11 different mist-net stations(each sta- fied the data by ignoring the various modes of in- tion had a 2.1 x 12.8-m net with 3.8-cm mesh) set up flection inherent in cardinal syllables.The five angle throughout the study area, color marked, and re- measurementswere averaged and used as an index leased.Poor visibility and impenetrability of vege- of frequency modulation, a measureof frequency tation precluded extensive use of Wiens's (1969) changeper unit of time. flushing technique. A 40-m grid with numbered The intersyllableinterval wasthe period of silence stakesand flagged poles positioned at all intersec- separating the syllable measured from the syllable tionsof grid linesprovided reference points for map- that followed it. This measurementwas taken only ping territories on the entire 15-ha study area. About between like syllables. 230-250 h were spent from the end of March to 15 As measuresof repertoire size, we determined the June each year (1979-1981) to determine territory number of different syllablesused by each male in boundaries. Territories were drawn on maps of the all songs,as well as the numberof songtypes used study area, cut out, and their areasmeasured with a by each male (after Lemon 1965). Both the number LiCor conveyorbelt (LI-3000) area meter. of different syllablesand the number of song types Samplingterritory vegetation.--The geometric center sung by each male correlatedsignificantly with the of eachcardinal territory was locatedin the field by number of songsrecorded (r = 0.69 and 0.73, respec- using grid maps.Three plots, 8 m in radius (0.02 ha), tively; P < 0.001). Thus, to correctfor differencesin were centered 20 m from each territory center at numbersof songsrecorded, we calculatedregression compassbearings of 90ø , 210ø , and 330ø . The height equationsfor each male to obtain a "predicted" num- and speciesof trees greater than 5 cm in diameter at ber of syllablesor songtypes for a given number of breastheight (DBH), basalarea, maximum height of songsrecorded. A log transformationof the number vegetation,percentage of canopyclosure, and foliage of songscorrected for a slightly parabolicrelation- density were measuredwithin these plots (Table 2). ship between songsrecorded and both variables.We The directionsfor foliage density measurements, used, as new variables, the residuals for each obser- either north and east or south and west, were deter- vation (the difference between the observed number mined randomly for each plot. Foliage density was of syllables,or song types,and the predicted)that estimatedfor ground level and for 1-, 2-, and 3-m were calculatedby the regressionprogram [residual height intervals with a 0.5-m-square(0.25 m2) grid- for the number of syllables(RESB): r 2 = 0.56, n = 30, ded board (MacArthur and MacArthur 1961). Mid- F•,28= 36.2, P < 0.001; residual for the number of story and canopy foliage density (higher than 7 m) syllabletypes (REST):r 2 = 0.72, n = 30, F•.•8= 70.9, and percentageof canopyclosure above 20 m were P < 0.001; Table 1]. estimated with a reflex camera (MacArthur and Horn To evaluatesong complexity,we calculatedan in- 1969) positionedat the center of each 0.02-ha plot. dex of syllableversatility within songsfor each car- Foliageheight diversity (MacArthur and MacArthur dinal. Modifying a techniquedeveloped by Kroods- 1961)and estimatesof verticaland horizontalpatch- ma and Verner (1978), total syllable-typeversatility inesswere calculatedfor eachterritory (Table2). Ver- was defined as a ratio between the sources of within- tical patchinesswas calculatedby subtractingfoliage 26 CONNER,ANDERSON, AND DICKSON [Auk, Vol. 103

TABLE 2. Habitat characteristics measured within 27 cardinal territories in eastern Texas.

Variable code Description Mean SD TF0-3 Total foliage density,ground to 3 m 0.60 0.25 (m2/m 3) FDG Foliage density, ground level (m2/m3) 0.27 0.17 FDI Foliagedensity, I m (m2/m3) 0.13 0.05 FD2 Foliage density, 2 m (m2/m3) 0.12 0.06 FD3 Foliagedensity, 3 m (m•/m •) 0.08 0.04 FD7 Foliage density, 7-13 m (m2/m3) 0.04 0.06 FDI3 Foliage density, 13-20 m (m2/m3) 0.04 0.06 FD20 Foliage density, > 20 m (m2 ! m•) 0.20 0.37 %CC Canopyclosure (%) 26.85 25.36 VHGT Vegetation height (m) 13.41 6.70 #TRSP No. of tree species 3.12 1.97 #TRES No. of tree stems >6 cm 7.47 3.95 TSPD Tree speciesdiversity (H') 0.79 0.55 BA Basalarea (m2/ha) 8.21 5.18 #SHSP No. of shrubspecies 8.37 4.92 #SHRB No. of shrub stems 29.96 13.93 SSPD Shrub speciesdiversity (H') 4.24 2.16 %GC Ground cover (%) 42.91 26.73 TSIZ Territory size (ha) 0.64 0.14 ARBM Arthropodbiomass (g) 3.78 1.52 HPTCH Horizontalpatchiness at 2 m (I - J') 0.09 0.13 VPTCH Vertical patchiness(1 - J') 0.11 0.07 FHD Foliageheight diversity (H') 1.24 0.09

height equitability(J') from 1.0. Horizontal patchi- fledglingscould be heard beggingfor food in shrubs nessof shrub-level vegetation at I, 2, and 3 m was near their nest site. Our presencemay have influ- estimatedfor each territory by calculatingthe equi- enced nest success(Bart and Robson 1982). The dis- tability of the sumsof foliage densitiesat each hor- turbancefactor should have been fairly equal in all izontalfoliage sampling point and subtractingit from territories becauseall nestswere visited at an equal 1.0. rate, and territories in general were traversed at A smallersubplot (3 per territory), 1.8 m in radius equivalentfrequencies while mappingbird positions (0.001 ha), was establishedwithin each 0.02-ha plot. and recording male song. Unsuccessfulnests were It was randomly located3 m from the center of the classifiedaccording to the mostlikely causefor fail- plot. The speciesand numberof stemsof all woody ure: fire ant (Solenopsisinvicta) predation or Brown- vegetationless than 6 cm DBH, percentageof ground headed Cowbird (Molothrusater) parasitism;undis- cover, and average height of shrubby vegetation turbed empty nest--snake or bird predator; and dis- (vegetation<3 m high) were measuredin this sub- furbed empty nest--mammalianpredator (Best 1978). plot (Table I). Shrub speciesdiversity (H') and eq- Samplingnest-site vegetation.--An 8-m-radius plot uitability (J') were calculated.Percentage of ground (0.02 ha) was centered on each nest site to sample cover was determined by looking straight down vegetation(Table 3). Foliagedensities, basal area, and through a tube, 11 cm long and 4 cm in diameter,at vegetationheight were measuredin the sameman- the centerof the subplotfrom 1.3 m abovethe ground ner as for the territory plots. At each nest,the height and estimatingthe proportionof herbaceouscover of the nest tree and height of the nest above ground to bare ground. were measuredwith a metric tape when possible, Nest observations.--Whileterritories were being and otherwisewith an Abneylevel. Estimates of nest mappedand songsrecorded, we searchedthe study concealment were obtained with the aid of a 0.5-m- area for cardinalnests. Only what we believedto be squaredensity board similar to the board used by first nestingattempts were evaluated.After a nest MacArthurand MacArthur (1961).The boardwas po- was located,we visited it every day or two to check sitionedvertically 0.5 m fromthe neston botha north- how many eggs or young were present. We used a southand east-westbearing to estimateconcealment mirror on the end of a pole to examine nestsabove from horizontal directions.The observer,facing the eye level. Young cardinalswere consideredto have board 0.5 m from the nest on the oppositeside of the fledged successfullyif they were seen in the nest nest from the board (1 m from the board), looked until at least 9 days old and subsequentlywere not through the vegetationand estimatedthe percentage found in an undisturbed nest (Table 3). Typically, of the board that was obscuredby foliage. This pro- January1986] CardinalHabitat, Song, and Nesting 27

TABLE 3. Characteristics measured at 30 cardinal nest sites in eastern Texas.

Variable code Description Mean SD NFDG Foliage density, ground level at nest site 0.28 0.18 (m2/m 3) NFD1 Foliage density, 1 m at nest site (m2/m•) 0.23 0.16 NFD2 Foliage density, 2 m at nest site (m2/m•) 0.32 0.56 NFD3 Foliage density, 3 m at nest site (m2/m•) 0.15 0.17 NFD7 Foliage density, 7-13 m at nest site (m2/m•) 0.02 0.07 NFD13 Foliage density, 13-20 m at nest site (m2/m3) 0.02 0.04 NFD20 Foliage density, >20 m at nest site (m2/m•) 0.29 1.40 NVHGT Vegetationheight at nestsite (m) 11.98 8.10 NBA Basal area at nest site (m2/ha) 7.43 5.68 NHGT Nest height (m) 1.58 0.73 HTNT Height of nest plant (m) 2.88 0.99 FDHZ Horizontal foliage concealmentof nest (%) 45.09 26.51 FDVT Vertical foliage concealmentof nest (%) 64.57 26.37 FLDG No. of young fledged/nest 0.70 0.97 YONG Maximum number of young in nest 1.04 1.15 EGGS Maximum number of eggsin nest 2.04 0.82

cedurewas repeatedwith the board 0.5 m above the 21), was usedto evaluatevariables that might be im- nest and the observer 0.5 m below the nest to esti- portant to nesting success.Rao's V was used as the mate vertical concealment. We did not measure con- stepwiseselection method, with an F value of 3.2 set cealmentabove the nestseparately from concealment for entry of variables into the analysis. Becauseof below the nest. unequal group sample sizes, we checked the influ- Arthropodsampling.--A crude estimate of food ence of heteroscedasticityby making a log transfor- availability was obtainedby samplingeach territory mationon all variablesand recalculatingthe discrim- with 38-cm-diameterinsect sweep nets. Although not inant function (DFA). The results of the DFA using a directsample of cardinalfood, we assumedthat our log-transformedvariables were consistentwith the measureof arthropodbiomass would give a relative DFA using raw data, indicating the acceptabilityof indexof the overallfood availability. Five 200-sweep the initial DFA (Pimentel 1979). samples(1,000 sweeps/territory)were taken from random placeson 5 different days immediatelyfol- RESULTS lowing nestingin eachterritory. Thesesamples were oven dried for 48 h and weighed on an analytical Cardinal territory size was related to vege- balance. tation structure. Territory size was positively Data analyses.--Of43 territories located and stud- correlatedwith vegetation height and nega- ied, only 21 had measurementsfor all variables(-> 89 tively correlatedwith total foliage density from songs recorded and all nest data). Correlation anal- the ground up to 3 m, foliage densityat ground yseswere usedto examinedegrees of associationbe- level, and number of shrubs (Table 4). Thus, tween variables.We realize that there are potential problemsassociated with interpretationsof multiple territories were larger in areas having mature bivariate correlations if conclusions are focused on trees and smaller where only shrubby vegeta- individual correlations.We have evaluatedpatterns tion was present. Previous studies have indi- of numeroussignificant correlations in this paperto cated that cardinalshave a strong affinity for gain insight into ecologicalrelationships. Although understory foliage (Dow 1970, Conner et al. many significantcorrelations were detected,we pre- 1983) and often nest in shrubs (Stauffer and sent only thoserelevant to relationshipsamong ter- Best1980). Because understory foliage was neg- ritory size, habitat, song, and nesting successof atively correlatedwith overstory foliage (per- Northern Cardinals.Sample sizes for evaluatingde- centageof canopyclosure was negatively cor- greesof relationshipvaried: nesting success vs. song related, P < 0.001, with understory foliage variables,n • 21; nesting successvs. territory size, arthropodbiomass, and vegetationvariables, n = 23; variables: FDG, r -- -0.69; FD1, r -- -0.58; FD2, and songvs. territory size, arthropodbiomass, and r --- -0.58; and TF0-3, r = -0.76), the increase vegetation variables, n = 30. Discriminant function in cardinal territory size in areaswhere mature analysis(Hull and Nie 1981) comparingterritories trees were present was not surprising. Cardi- that fledgedyoung vs. thosenot fledgingyoung (n = nals may require a thresholdamount of under- 28 CONNER,ANDERSON, AND DICKSON [Auk,Vol. 103

TABLE4. Correlationsof habitat,song, territory size and quality, and nestingsuccess of Northern Cardinals (seeTables 1-3 for variable codes).Sample size varies for the entriesin this table;see Methods. Correlations with P > 0.10 and variableswithout any significantcorrelations are omitted from the table.a

TSIZ ARBM SBVS ASYB SGVS RESB YONG b FLDG b

TSIZ 0.25 c ARBM -0.34* -0.40* 0.45* 0.40* SBVS - 0.34* 0.32* - 0.36 c SGVS 0.25 c 0.32* SLGT -0.45** 0.83*** %GC 0.72*** -0.42* FDG -0.43** 0.57*** 0.43* 0.40* FD1 0.64*** -0.34* 0.37* 0.37* FD2 0.41' 0.51'* FD3 - 0.46 * * -0.30 c -0.36* TF0-3 -0.45** 0.51'* -0.29 c 0.33* 0.38* 0.45* FD20 SSPD 0.54*** -0.53*** 0.32* #SHRB -0.37* HPTCH 0.41' VHGT 0.40** -0.33* 0.33* -0.30 • -0.28 c #TRES -0.44** -0.38* -0.46* BA 0 53*** -0.39* -0.44*

* = P < 0.05, ** = P < 0.01, *** = P < 0.001. Spearmanrank correlations. P < 0.10. storyfoliage in their territories.If shrubfoliage (Table 4). This suggestsan associationbetween is sparse, territory size may be increased to large syllable repertoiresand vegetation char- compensate. acteristicof good-qualityterritories. Although Arthropod biomasswas positivelycorrelated understoryfoliage variablesand shrub species with foliage density at ground level and at 1 diversity were correlatedwith arthropod bio- m, total foliage densityat 0-3 m, percentageof mass,the residual of the number of syllables ground cover,and shrub speciesdiversity (Ta- was not. The residual of the number of sylla- ble 4). Arthropod biomasswas negatively cor- bles also was not significantlycorrelated with related with number of trees, basal area, and any measureof nesting success. vegetationheight. Nesting successin the long run should be Total syllableversatility (SBVS),our measure the bestindex to what mightbe quality habitat. of within-song complexity,was negatively cor- Although our study was only 3 yearsin length, related with arthropodbiomass, foliage density some relationshipswere observed. Fledging at 3 m, and shrub speciesdiversity (Table 4). successand number of young in nests were The averagenumber of syllablesused in each positively correlatedwith arthropod biomass songand songlength alsocorrelated negative- (Table 4). The number of eggsin nestswas not ly with arthropodbiomass. These resultssug- correlatedwith arthropodbiomass or any other gestthat cardinalshaving the shortestand least variablemeasured. We had originally expected versatile songswere on the best-qualityterri- clutch size to be related to food availability. tories. Total song versatility correlated posi- Fledging successand the number of young in tively with total syllableversatility but was not nestswere also positively correlatedwith fo- correlated with arthropod biomass.No other liage density at the ground, at 1 m above the songvariable was significantlycorrelated with ground, and with total foliage density at 0-3 arthropod biomass(P > 0.05). m. In addition, fledging successwas positively Of our measuresof repertoire size, only the correlatedwith foliage density at 2 m and hor- residual of the number of syllables sung by izontal patchiness at 2 m above the ground. males(RESB) was positivelycorrelated with fo- Bothfledging success and the numberof young liage density at 2 m, total foliage density from in nests were negatively correlatedwith the the ground to 3 m, and shrub speciesdiversity number of trees and tree basal area (Table 4). January1986] CardinalHabitat, Song, and Nesting 29

Collectively, these resultssuggest that the bet- TAI•Lil5. Correlationsof original territory and song ter-quality territories were areas with fewer variables with the discriminant function (canoni- treesand more understoryfoliage. Originally, cal variate) resulting from a 2-group DFA used to discriminatebetween territories successfully fledg- we had expectedsome aspect of nestingsuccess ing young(n = 7) and territorieswhere nests failed to be positively correlated with territory size, (n = 14). (See Tables 1-3 for variable codes and but none was. units.) No significantcorrelation between fledging successand nest height existed (rs = 0.01, P > Variable Pearson correlation a 0.05). However, territories with successfulnests FD2 0.75*** had higher foliage density and patchinessat 2 HPTCH 0.64'** SLGT -0.60** m throughout the territory than did territories VPTCH 0.57** with unsuccessful nests (ANOVA, P < 0.05). TF0-3 0.54** This suggeststhat abundanceand distribution FHD - 0.53 * * of foliage at 2 m above the ground throughout FD! 0.51'* the territory may be important for cardinals. ARBM 0.47'* ASYB -0.46* This is reinforcedby the fact that cardinal nests #TRES -0.43* in the studyaveraged !.6 m abovethe ground. VHGT - 0.41 * Nesting successvariables were not signifi- '* = P < 0.05, ** = P < 0.01, *** = P < 0.001. cantly correlated with cardinal song character- istics (P > 0.05). However, total syllable ver- satility had a nearly significant negative territorieshaving unsuccessfulnests (P < 0.006, correlationwith number of young in nests(rs = 90.5% of casescorrectly classified,eigenvalue -0.36, P < 0.053). Studies conducted over a of the discriminant function = 1.048). Three longer period of time with a larger sample are variables entered the stepwise DFA: foliage probably needed to demonstratestrong rela- density at 2 m, song length, and versatility of tionships between song characteristics and syllablesper song.Examination of the correla- nesting success. tions of the original territory and song vari- Cowbird parasitism and predation affected ables with the discriminant axis (canonical nesting success.Causes of nesting failures in variate) indicated that successful nests were our study area over 3 years were: 7 to Brown- positively associatedwith increasingvalues of headed Cowbird parasitism,5 to small mam- foliage density at 2 m, horizontal patchinessat mals and snakes,! to a large mammal, and ! to 2 m, vertical patchiness,total foliage density fire ants. We observed partial clutch losseson from the ground to 3 m, foliage density at 1 m 9 occasions,which may have causedan absence (i.e. dense, patchy understory foliage), and ar- of correlationbetween the numberof eggsand thropod biomass(Table 5). Successfulnests were the number fledged (rs = 0.08, P > 0.05). The negatively associatedwith increasing song numberof young and the numberfledged were length, foliage height diversity, number of syl- significantlycorrelated (Spearman rank corre- lables per song, number of tree stems, and lation, rs = 0.70, P < 0.001). Also, nests found vegetation height. Thus, unsuccessfulnests in early stagesof the nestingcycle (nestbuild- were associatedwith increasing vegetation ing and egg-laying) may have had a higher height, density of trees,and foliage height di- probability of parasitismthan nestsfound dur- versity,and male territorial cardinalsthat sang ing the nestling phase (Mayfield 1961). Al- long songswith higher within-song versatility. though we expected nest concealment to have Resultsof the DFA agree closelywith our in- an effecton nest parasitismand predation, no terpretationof the correlationpatterns among significantcorrelations existed between fledg- the univariate variables (Table 4). ing successand foliage density in the nest vi- cinity (NFD! through NFD3, Table 3) or nest DISCUSSION concealment (FDHZ and FDVT). A stepwise2-group DFA using song and ter- Our results suggest some interesting rela- ritory variables(Tables 1, 2) successfullydis- tionships. Cardinals with lessversatile or com- criminated between territories having success- plex songs may hold better-quality territories ful nests(nests fledging at leastone young) and (territorieswith more understoryfoliage and 30 CONNER,ANDERSON, AND DICKSON [Auk, VoL 103

TABLE6. Songcharacteristics of a male cardinalthat held the sameterritory for threesuccessive breeding seasons(see Table 1 for variable codesand units).

Year NSGS SBVS ASYB DSYB TRAN SLGT SBRP

1979 316 0.094 7.1 1.4 0.36 1.96 18 1980 398 0.140 7.5 1.5 0.50 2.41 17 1981 187 0.177 11.6 1.8 0.78 2.49 18

greater arthropod biomass)and fledge more negatively related to fledging success.How- young than cardinalswith more complexsongs. ever, this hypothesis also suggeststhat song This suggeststhat natural selectionwould fa- complexity should decrease over successive vor males with the least song complexity, a years. Such was not the case with one male concept that at first appears counterintuitive. cardinal that nestedin the sameterritory for 3 Severalalternative explanationsmay clarify consecutive years. Total syllable versatility, this apparent contradiction.Fledging success number of syllablesper song, number of dif- may be confoundedby factorsother than over- ferentsyllables used per song,and songlength all territory quality. Predation and nest para- increasedsteadily for this individual during the sitism can both greatly affect nesting success study (Table 6). Syllable repertoire (SBRP)for and yet not be related to territory quality from this male remained relatively constant,and no a "food" perspective.Our visits to nests may trend or pattern was observedwith other song have increasedthe probability of cowbird par- variables. Similarly, repertoires of Great Tits asitismand predation.Although nest conceal- also do not increaseyearly (Krebs et al. 1978). ment should be related to cowbird parasitism The changewe observedin song complexity and predation, our measurementsof conceal- over time for one male cardinal is obviously ment may have been insufficientto document insufficientto establishany trends.Studies on this relationship. Wray and Whitmore (1979) additional males are needed to provide suffi- made more extensive measurements of nest cient data for a more accurate determination. concealmentthan we did and detecteda sig- nificant relationship between foliage proximal ACKNOWLEDGMENTS to the nest and nesting success. We thank L. B. Best, D. E. Kroodsma, R. E. Lemon, Our measuresof song complexity and rep- C. J. Ralph,and J.Verner for numeroushelpful com- ertoire size may also be confounded.Howard mentsand suggestionson the manuscript.This study (1974) speculatedthat both repertoire size and was endorsedand funded in part by the U.S. Man reproductivesuccess might be correlatedwith and the BiosphereProgram (MAB-3), which contrib- age.A studyof the Great Tit (Parusmajor) dem- utes to grazing land managementobjectives. We also onstrated that males with intermediate-size thank the John Casonfamily and the SouthlandDi- repertoireshave the highestlifetime reproduc- vision of St. Regis Paper Co. for use of their lands during the study.Mention of commercialproducts tive success(McGregor et al. 1981). Older male in this paper doesnot constituteendorsement by the White-crowned Sparrows (Zonotrichia leuco- USDA to the exclusionof other productsthat might phrys)hold largerterritories and are more often be suitable. successfulin breedingthan young males(Ralph and Pearson 1971). Male Field Sparrows (Spi- LITERATURE CITED zella pusilla)with larger territories mate more frequently than males with smaller territories BART,J., & D. S. ROBSON.1982. Estimating survivor- (Best 1977). ship when the subjectsare visited periodically. Young male cardinalsmay use longer, high- Ecology63: 1078-1090. ly complexsongs when first establishinga ter- BEST,L. B. 1977. Territory quality and mating suc- cessin the Field Sparrow (Spizellapusilia). Con- ritory. In subsequentbreeding seasons,age or dor 79: 192-204. prior ownership of a particular territory may 1978. Field Sparrow reproductivesuccess permit the male to put "lesseffort" into song and nesting ecology.Auk 95: 9-22. (complexity)and allow more effort for nestde- --, & D. F. STAUFFER.1980. Factorsaffecting fenseand careof young. This hypothesismesh- nesting successin riparian bird communities. es with our results of song complexity being Condor 82: 149-158. January1986] CardinalHabitat, Song, and Nesting 3l

BROOKS,R. J., & J. B. FALLS. 1975. Individual rec- ß 1977. Correlatesof songorganization among ognitionby songin White-throatedSparrows. II. North American wrens. Amer. Natur. 111: 995- Featuresof songused in individual recognition. 1008. Can. J. Zool. 53: 1749-1761. ß& J. WERNER.1978. Complexsinging behav- CATCHPOLE,C.K. 1973. The functionsof advertising iors amongCistothorus wrens. Auk 95: 703-716. song in the SedgeWarbler (Acrocephalusschoe- LEMON,R.E. 1965. The songrepertoires of cardinals nobaenus)and the Reed Warbler (A. scirpaceus). (Richraondenacardinalis) at London, Ontario. Can. Behaviour 46: 300-320. J. Zool. 43: 559-569. CONNER,R. N., J. G. DICKSON,B. A. LOCKE,& C. A. 1968. The displaysand call notesof cardi- SEGELQUIST.1983. Vegetation characteristics nals. Can. J. Zool. 46: 141-151. important to common songbirdsin east Texas. MACARTHUR,R. H., & H. S. HORN. 1969. Foliage Wilson Bull. 95: 349-361. profile by vertical measurements.Ecology 50: DOT, D.D. 1970. Distribution and dispersalof the 802-804. cardinal, Richmondena cardinalis, in relation to ß& J. W. MACARTHUR.1961. On bird species vegetationalcover and fiver systems.Amer. Midi. diversity. Ecology42: 594-598. Natur. 84: 198-207. MAYFIELD,H. 1961. Nestingsuccess calculated from EMLEN,S. T. 1971. The role of songin individual exposure.Wilson Bull. 73: 255-261. recognitionin the Indigo Bunting. Z. Tierpsy- MCGREGORßP. K., J. R. KREBS,& C. M. PERRINS.1981. chol. 28: 241-246. Songrepertoires and lifetime reproductivesuc- HARTSHORNEßC. 1956. The monotony-thresholdin cessin the Great Tit (Parusmajor). Amer. Natur. singing birds. Auk 72: 176-192. 118: 149-159. HOWARD, R. D. 1974. The influence of sexual selec- MORTON,E. $. 1975. Ecologicalsources of selection tion and interspecificcompetition in mocking- on avian sounds. Amer. Natur. 109: 17-34. bird song(Mimus polyglottos). Evolution 28: 428- PIMENTEL,R. A. 1979. Morphometrics:the multi- 438. vafiateanalysis of biologicaldata. Dubuque, Iowa, HULL, C. H., & N.H. NIE. 1981. SPSSupdate 7-9. Kendall/Hunt. New Yorkß McGraw-Hill. RALPH,C. J., & C. A. PEARSON.1971. Correlation of INTERNATIONAL BIRD CENSUS COMMITTEEß 1970. An age, size of territory, plumages,and breeding international standardfor a mapping method in successin White-crownedSparrows. Condor 73: bird census work. Audubon Field Notes 24: 722- 77-80. 726. STAUFFER,D. F., & L. B. BEST. 1980. Habitat selection KREBS,J. R. 1977a. The significanceof songreper- by birds of riparian communities:evaluating ef- toires: the Beau Geste hypothesis.Anim. Behav. fectsof habitat alterations.J. Wildl. Mgmt. 44: !- 25: 475-478. 15. 1977b. Songand territory in the GreatTit. WERNER,J. 1975. Complex song repertoire of male Pp. 47-62 in Evolutionary ecology (B. Stone- Long-billed Marsh Wrens in easternWashing- house and C. M. Perrins, Eds.). London, Mac- ton. Living Bird !4: 263-300. millan. WIENS,J.A. 1969. An approachto the studyof eco- ß R. ASHCROFT,& M. I. WEBBER.1978. Song logicalrelationships among grassland birds. Or- repertoiresand territorydefence in the GreatTit. nithol. Monogr. No. 8. Nature 271: 539-542. WRAY,T., & R. C. WmTMORE.1979. Effectsof vege- KROODSMA,D.E. 1976. Reproductivedevelopment tation on nesting successof Vesper Sparrows. in a female songbird:differential stimulation by Auk 96: 802-805. quality of male song.Science 192: 574-575. FAMILY ASSOCIATIONS IN COMMUNALLY ROOSTING BLACK VULTURES

PATRICIA PARKER RABENOLD 1 Departmentof Biology,University of North Carolina,Chapel Hill, North Carolina27514 USA

ABSTI•ACT.--Observationsof marked individuals in a partially marked population of Black Vultures (Coragypsatratus) show that immediate family membersmaintain close contact throughoutthe year. Associationsbetween mates and between parentsand offspringare primarily responsiblefor this. Mutual allopreening, feedings, and intercessionsin fights occur almost exclusively within families. Adults known to breed within the study area (neighbors)associate more stronglythan nonneighbors,although not as stronglyas mates. Certain familiesassociate preferentially with certain other families.Strength of interfamily associationsis not related to distance between nest sites within the study area. I argue, therefore,that birds activelyseek particular individuals as associates.Further knowledgeof breedingdispersal of young of known parentagewill provide cluesto relatednessof highly associatedneighbors. Strong associations among related individuals can help explain the use of avian communalroosts as information centers.Received 27 April 1984,accepted 10 June1985.

COMMUNALroosts and breeding colonies shareor cooperativelydefend its find. If these could serve as "information centers" to which associations of individuals are stable over time, unsuccessfulforagers return and follow suc- it should be possibleto detect relative differ- cessful individuals to a known food source encesin the strengthsof associationsbetween (Ward and Zahavi 1973). Explaining the appar- different individuals in communalroosting and ent aid-giving of successfulforagers remains a feeding situations.There is evidence suggest- difficultyof this hypothesis.Some workers em- ing that parent-offspringties persistfor several phasize the probability of high levels of relat- months past fledging in Black Vultures (Jack- ednesswithin colonies(e.g. Erwin 1978,Waltz son 1975, McHargue 1977, this study). This be- 1982) to help explain why successfulforagers havior would help explain the use of roostsas might tolerate any costsof being followed. By information centers. Preferential association sharing food with closely related individuals, and preferential behavioralinteractions among successfulforagers could increasetheir inclu- individualsin a populationcould structurethe sive fitness (Hamilton 1964). If mechanisms ex- distribution of benefits accruing from shared ist that allow preferential sharing with close information at communal roosts. The mainte- relatives, inclusive fitness gains could out- nance of family associationsoutside the breed- weigh costsof sharing. ing seasonis the subjectof this paper. Black Vultures (Coragypsatratus) form large To examine associations of individuals with- communal roostseach night throughout the in BlackVulture communalgroups, I ask spe- year, although they nest in pairs in isolated, cifically (1) whether mated pairs and individ- well-protected locations. Indirect evidence uals of known relatedness (parents and suggeststhat BlackVulture roostsmay serveas offspring or siblings) are seen together more food-finding information centers.For instance, often than predicted by chance;(2) whether morning roost departures are significantly these intrafamilial associations differ from as- clumped both in time and in direction, and sociations of the same individuals with birds these clumped departurescannot be explained not known to be related (extrafamilial associa- by weatherconditions (Rabenold 1983). tions); (3) whether pairs nesting in an area as- A successfulforager might return to the roost sociatemore than expectedby chancewith oth- to recruit particular individuals with whom to er pairs nestingin the samearea, and whether these neighborhood associationsdiffer in de- gree from associationswithin families or be- • Present address:Department of Biological Sci- tween nonneighbors;and (4) whether strength ences, Purdue University, West Lafayette, Indiana of associationsbetween neighbors is depen- 47907 USA. dent on the distance between their nest sites.

32 The Auk 103:32-41. January1986 January1986] BlackVulture Family Associations 33

METHODS bird, includingmutual allopreening, fights (initiator, if seen,type of fight, and outcome),and beggingby Studyarea and population.--The study area, situated young birds (adults to whom begging was directed in farmlandsof ChathamCounty in central North and whether the youngbird was ignored,rebuffed, Carolina and covering approximately250 km2, is or fed by regurgitation).I performed318 roostcen- composedof 60% agricultural land and 40% wood- susesduring the study. land; it containsno towns (Reeveset al. 1970).The Feeding groupswere located by following birds main farm productsof this area are poultry, swine, out of the roostto their morning feeding sites,or by dairyand beefcattle, and tobaccoproduced by many driving a prescribedcircuit through the study area small farms.I have studiedthe populationof Black during midmorning while watching for groups of Vulturesthere since autumn 1976, and I begana trap- vultures circling or descending.At all feedings, I ping and markingprogram in 1977.Birds were cap- countedthe number of BlackVultures present,iden- turedprimarily in a walk-in funneltrap baitedwith tified tagged birds, and recorded social interactions carrion.By summer 1981, 344 BlackVultures had been involving marked birds. I performed 86 feeding cen- markedwith individuallynumbered vinyl wing tags susesduring this study. (Wallaceet al. 1980)legible at closerange with 10x Ascertainingfamily membership.--InNorth Carolina, binocularsand at greaterdistance with a 15-60x zoom breeding Black Vulture pairs nest in isolated loca- telescope. tions usually deep in woods.Between 1977 and 1981 Basedon the proportion of birds marked at roosts I located14 suchnest siteswithin the study area.At and feedingareas and an estimatedannual mortality 8 of thesethe sameadults returned eachyear to breed. of 12.5%(the annual mortality of breedingadults All adultsat these8 nestswere markedwith patagial during this study), approximately1,200 Black Vul- tags.Each pair produced1 clutchof 2 eggseach year tures regularly use the area year-round.Not all of (although2 early failuresresulted in later relaying), thesebirds roost in the studyarea on anygiven night, and all fledgedyoung have been tagged at eachnest as there is considerable movement in and out. Sea- in every year since its discovery. In all, 46 young sonalestimates of the numberof birdspresent range birds of known parentagehave been taggedin the 8 from 358/night during winter (mean size of winter families,and different-agesibs (produced by the same roosting groups multiplied by mean number of ac- parentsin different years)are known for all sites. tive roosts/night)to 242/night in summer.The range Associationmeasures.--I used only observationsof overwhich these birds forage is unknown,although feeding or roosting groupsof birds to test the sig- somemovement of marked birds outsidethe study nificance of associations between individuals in the area suggestsvery large rangesfor someindividuals. groups.Sightings of birds at nest siteswere not in- Of the 344 marked Black Vultures, over 100 are seen cluded.I alsoexcluded observations made during the regularlywithin the studyarea, 2 have beenseen at early breeding months of March, April, and May, a distanceof 95 km (and have since returned), and 1 when movement of breeding adults is restrictedow- wasseen repeatedly at a communalroost 170 km away. ing to incubationand brooding duties. Only the first Daily rangesare not likely to includesuch long sighting of an individual at a particularsite on any movements. day was used. To test associationswithin the group The studyarea contains seven roost sites used by of birds regularly using the area, I attemptedto omit the local population. Roostsare all in mixed conifer- transientsand irregular visitorsby consideringonly hardwood forest near small creeks. Roost sites are thoseanimals sighted on at least 10 different days.If traditionalin the sensethat eachis regularlyattend- observationsof two birds did not overlap completely ed when birds are feeding nearby, and local resi- as a result of different marking dates or disappear- dentsreport that the samesites have been occupied ance of either one, the measurement of association sporadicallyfor many years,although no roostsite was limited to the period of potential overlap. is occupiedevery night. My observationsof marked Of the 344 marked Black Vultures, 102 (50 adults, birds indicate that within a certain area individuals 52 young) were resightedat least 10 times (maximum usea numberof roostsover the courseof a year,but number of sightingsfor any individual in groups= often usea singleroost for severaldays in a row. 61), and thesebirds are consideredthe coregroup of Censusesof feeding and roosting groups.--Each roost vulturesregularly using the study area. These 102 was censusedregularly through the courseof the individualswere consideredby pairs in three mea- study.If a roostwas occupied, I countedthe number sures:the Chi-squaretest of independencewithin of BlackVultures present and identified tagged birds. pairs;the point correlationcoefficient V (Pielou 1976); Censuseswere usually complete,except in summer and Dice's(1945) "associationindex." The point cor- when foliagesometimes partially obscured the roost. relationcoefficient, an indexranging from - 1 to + 1, Measuresof association,therefore, represent mini- wasused as a comparativemeasure between subclass- mum estimates of actual associations between indi- es of the population(e.g. between related pairs and viduals.When scanninggroups for tags,I recorded unrelated pairs) and as a meansof assigningdirec- all socialinteractions involving at least one marked tion to resultsof the Chi-squaretest (the Chi-square 34 PATRICIAPARKER RABENOLD [Auk,Vol. 103

TABLEI. Associationof 8 matedpairs of BlackVul- ciations at food, significantly more than the turesoutside of incubationand broodingperiods. random expectation of 5% (XL P < 0.01). No

Number of censuses age-paired category (e.g. adult-adult, adult- including: Measuresof young, young-young) composed of known association a family memberspaired with extrafamilialbirds, A A B Neither and only only A nor Dice's differed significantly from 0.05 in the propor- Pair B (a) (b) (c) B X2 index tion of positiveassociations at feeding sites.Be- I 15 13 8 158 49.93 0.588 causeassociation rules among family members 2 20 14 13 132 42.28 0.597 apparentlydo not differ between roostingand 3 9 3 3 57 30.42 0.750 feeding, I pooled roosting and feeding obser- 4 10 2 4 52 30.58 0.769 vations for the remainder of the analysis. 5 9 1 15 140 42.51 0.529 Associations between mates.--Mated adults al- 6 I0 6 3 151 66.92 0.690 7 8 6 16 104 13.52 0.421 ternate nest shifts of approximately 24 h dur- 8 15 I0 5 109 47.07 0.667 ing the 55 days of incubationand brooding. I have 62 observationsof an off-duty bird at- ap < 0.001for all Chi-squarevalues. Dice'sindex = 2a/(2a + b + c). tending a communal roost, and in 27 visits to nests at dawn I never found more than 1 adult present.Except during March, April, and May, test is two-tailed and will not differentiate between when they were incubating and brooding, significantlypositive and significantlynegative as- sociations). Dice's index was also used to make com- membersof the 8 mated pairs with known nest parisons and to validate results based on the other siteswere seen together more often than ex- two measures. pectedby chance(Table 1, X2, P < 0.001 for all). However, none showed complete overlap, as all 16 marked adults of known nest site occa- RESULTS sionally were seen without their mates.Of the Associationsin feedingand roostingcontexts.- associations of these 16 individuals with all Individual vultures usually were resighted other adultsthat were not family members,only more often at roosts than at food for two rea- 18%were positive, significantly fewer than be- sons:(1) I made many more censusesat roosts tween mates (Fig. 1). Adults clearly associate than at food (318 vs. 86), and (2) roostinggroups with their mates more than with other adults were muchlarger than feedinggroups (means = outsidethe breeding season,but they also reg- 82 and 22). While 102 birds were resighted10 ularly associatewith some adults that are not or more times, only 53 of these had at least 10 their mates. resightingsat roostsand only 11 had at least Associationsbetween parents and offspring.- 10 resightingsat feeding sites.The small feed- Among parent-offspringpairs, 23 of 36 (64%) ing sample makesfull comparisonsof associa- showedpositive association through the course tion rules between the two contexts impossi- of this study (Fig. 1). I have insufficientyearly ble. However, 7 of the 11 birds resighted 10 or resightings of almost all pairs to test the de- more times at food were local breeders com- cline of parent-offspringassociations as the prising 4 mated pairs. These 4 pairs did not young birds mature,so for this analysisI pooled differ between roosting and feeding contexts; all parent-offspring pairs regardless of off- all were positively associatedin both situations springage. In general,resighting rates of young (P < 0.05, X2 test). If a larger subsampleof the birds decreasedas the birds aged so that 3- or 102 is examined, using birds with as few as 2, 4-yr-old offspringare resightedless frequently but on average 6, feeding sightings, sufficient than either their parentsor their youngersib- pairs exist to compare within-family associa- lings. However, when in the study area, sub- tions at roosts(68 pairs) against those at food adults occurred with their parents with the (37 pairs).Although feeding groupsare smaller sameprobability as did younger sibs. Associa- and random mixture would produce many tions of the 16 breeding adults of known nest fewer positiveassociations at food than at roosts, site with all young known not to be their off- the proportion of positive associationsdid not spring yielded only 14% positive associations. differ between roostsand food for mates, par- Parent-offspringassociations included signifi- ent-offspringpairs, or siblings.Pooled intrafa- cantly more positive associationsthan did par- milial associationsincluded 32% positive asso- ent-nonoffspringassociations. The conversere- January1986] BlackVulture Family Associations 35

•••'///////J 8pairs Mates : 'X'L[•] 477pairs Breeders- Nonmates

-X- 433pairs Breeders-Nonoffspring •----"JJ'•J-X- 'X'bi••• 774.•6pairs Parents-OffspringYoung-Nonparents ', Iv/////•'•..X.• • -J

•jj 69228 pairs SiblingsNonsiblings ', •

-X-•-X- 237672pairs Pooled TntrafamilyExtrafamily : •:.•::•/•.•J•-'•JJ•

0.0 0.25 0.50 0.75 1.00 0.75 0.50 0.25 0.0 PROPORTION OF POSITIVE ASSOCIATION iZNDEX ASSOCIATIONS (3(2) (PROPORTION OF SIGHTINGS TOGETHER)

Fig. 1. Left histogram:proportion of significantpositive associations (X 2, P < 0.05) within families (shaded bars) and between presumedunrelated (extrafamilial) birds (open bars), by age/statusclass. *.* indicates significantdifference in proportionsat P < 0.001, using test of equality of 2 percentages(Sokal and Rohlf 1969). Right histogram:mean association-indexvalues (+SD) within families (shaded bars) and between presumedunrelated (extrafamilial) birds (open bars),by age/statusclass. ** indicatessignificant differences in distributionsat P < 0.001, * indicatesP < 0.01 using Kolmogorov-Smirnov2-sample test (Siegel 1956). ns = not significant. lationship between offspring and all adults groups (Table 2). They also had significantly known not to be their parents was similar al- greaterposttledging survival (26 of 28, or 93%, though not as pronounced (Fig. 1). Associa- were known to have survived more than 2 tions of young of known nest site with all monthspast tledging) when comparedwith the nonparent adults included 37% positive asso- two families with low parent-offspringassoci- ciations,significantly lower than the compa- ation (X2, P < 0.01). I know little about the as- rable parent-offspringproportion. Young birds sociationsof parents with young that disap- were found with their parentsmore often than peared within 2 months of tledging becauseI with other adults. do not have the requisite 10 resightingsfor Parent-offspringassociations varied widely these offspring. These calculationsare based amongfamilies but were fairly consistentwith- only on associationsof parentswith surviving in families (Table 2). Within two families iden- offspring.For the two families with low asso- tified as TB and AH, no significantly positive ciations between parents and surviving off- associationswere found in 9 pairings of par- spring, I assume these parents treated their entswith survivingoffspring. To the extentthat nonsurviving offspring similarly. associationsbetween parentsand volant young Associationsbetween siblings.--The proportion representthe degree of posttledgingparental of positive associationsamong sib pairs (same- care,these two breeding pairs investedlittle if and different-agepairs pooled) was 55%, not anything in their young after tledging. Their significantlydifferent from the 46%positive as- overallreproductive success perhaps reflects the sociationsfound when known-nestyoung were costof not maintaining contactwith offspring; pairedwith all other nonsiblingyoung (Fig. 1). 13 young fledged at thesetwo nestsover 4 yr, This seemingly low proportion of positive as- but only 6 (46%) were known to have survived sociationsamong sibs (when compared with more than 2 months after fledging. The re- other intrafamilial categories)was not affected maining families showedpositive associations by separatingsibs of different ages.Positive as- in 23 of 27 (85%) parent-surviving offspring sociationsoccurred in 7 of 12 (58%) same-age pairingsduring all posttledgingappearances in sib setsand 8 of 16 (50%) different-agesib sets. 36 PATRICIAPARKER RABENOLD [Auk, Vol. 103

TABLE2. Fledgling survivorshipand strengthof parent-offspringassociation.

Association measures a

No. of X2 values that were: Proportion of Dice's index fledglings Significantly Family code surviving positive Positive Negative Mean $D LA 6/6 6 2 0 0.329 0.182 DC 4/6 1 1 0 0.331 0.063 ML 6/6 6 0 0 0.550 0.275 LE 7/7 6 0 0 0.664 0.149 $A 3/3 4 1 0 0.422 0.274 TB 2/5 0 2 1 0.116 0.102 AH 4/8 0 0 6 0.000 0.000 Parentswith surviving offspring.

In fact,young birds showedstrong positive as- than the 18% positive associationsbetween sociationsregardless of relatedness.Young ap- breedingadults and all extrafamilialadults (Ta- parently make no active attempt to follow or ble 3). However, the proportion of positive as- remain with their siblings as long as they are sociations among neighbors was less than with a group of young vultures. among mates (Table 3). Closer inspection re- Pooled intrafamilial associations contained a veals that some breeding pairs associated higher proportionof positiveassociations than stronglywith certainother pairs.Furthermore, did pooled extrafamilial associations(Fig. 1). becauseparents and offspring showed signifi- Overall, immediate family members occurred cant levels of association,entire family groups with each other in communal roosts and feed- from different nest sites were positively asso- ing groupsoutside the breeding seasonmore ciated. Levels of associationsbetween neigh- often than they occurredwith other presum- boring breeding pairs could be a function of ably unrelated individuals. High intrafamilial distancebetween nest sites.This would imply associations,however, resulted primarily from passiveco-occurrence due to degree of home- parent-offspring and mated pair associations, range overlap in animals restricting their rather than sibling associations. movements to the vicinity of their nest sites. Associationsbetween neighbors.--All marked There was no correlation between degree of breedingadults of known nestsite were tested associations and distance between nest sites in pairswith all other nonmatebreeding adults within the studyarea (Fig. 2). A linear regres- of known nest site for association,excluding sion of mean V valuesfor each distanceagainst the months of March, April, and May. Of 82 distancebetween nest was not significant(r = pairs, 28 (34%)had positive associations,more -0.105, P >> 0.05).

TABLE3. Comparisonof associationsbetween breeding adult Black Vultures and other adults.

Measures of association

X2 testsa Dice's index Type of association n p n • SD Breedingadults with all nonmateadults 477 0.182 478 0.135 0.115 (P < 0.01)b (D = 0.275, P < 0.001)c Breedingadults with neighboringbreeders 82 0.341 86 0.273 0.210 (P < 0.001)b (D = 0.520, P < 0.05)c Within matedpairs 8 1.000 8 0.628 0.115 an = numberof dyads,p = proportionof associationsthat were significantlypositive (P < 0.05). bTest of equality of two percentages(Sokal and Rohlf 1969). • Kolmogorov-$mirnovtwo-sample test (Siegel 1956); D = greatestinterval differencebetween two cumu- lative frequencydistributions. January1986] BlackVulture Family Associations 37

TABLE 4. Occurrences of behavioral interactions be- tween different classes of Black Vultures.

No. of casesinvolving:

ents/ off- Sib- Extra- Type of behavior Mates spring lings family Mutual allopreening 27 26 8 9 Postfledgingfeedings 0 41 0 0 Interference at roost 2 2 2 0 Fig. 2. Point correlation coefficient (V) for all Interference at food 3 7 0 0 breedingadults paired with nonmate locally breed- ing adults by distance between nests. V values for eachnest-nest comparison (nest A adults paired with nest B adults) are connectedby vertical lines. Dia- than expected,based simply on the represen- monds indicate nest-nestcomparisons in which all tation of sibsin the populationof young birds pairs are significantlypositively associated (X 2, P < (X2 = 66.87, P < 0.001). 0.05). Forty-one postfledging feedings involving known parent-offspring pairings were ob- Comparisonsof associationmeasures.--All re- served. These feedings occurred as late as 8 sultsfrom Dice'sassociation index agreedwith months after fledging. Adults were never ob- resultsof Chi-squaretests (Fig. 1). Association- servedto feed young other than their own in index valueswere greater among matesthan spite of beggingattempts by other young. Un- unmatedadults, and among parent-offspring relatedyoung attractedby conspicuousregur- pairs than among both parent-nonoffspring gitative feeding were always driven away by young and offspring-nonparentadults. Index pecksfrom the feeding adult. valuesdid notdiffer significantly between pairs Central to the association of kin in Black Vul- of sibsand pairsof nonsiblingyoung. Pooled tures is the fact that individuals will come to intrafamilial association-index values exceeded the aid of kin that are being activelydisplaced extrafamilial values. Association indices be- or physicallyattacked by unrelated birds. Al- tween neighboring breeding adults (extrafam- though only 16 instanceswere observed,all in- ilial neighbors) were less than between mates volved intercessionon behalf of known family (Table 3), but greater than between breeding members (Table 4). adults and all nonmate adults (extrafamilial Familygatherings at nestsites.--In addition to adult population)(Table 3). preferringthe companyof kin and behaving Behavioralcorrelates offamily associations.--Be- preferentially toward them in communal situ- havioral correlatesof family associationsin- ations,Black Vultures rendezvous with family clude allopreening,feeding, and defending membersat their uniquenest sites. After fledg- family membersagainst attacks by unrelated ing, young birds that become separatedfrom birds. Of 70 mutual allopreeninginteractions their parents meet them back at their nest sites. involving at least one known family member Preening and feedingstake place there before in communallyroosting or feedinggroups, 61 birds depart for the evening communalroost. were between family members(Table 4). Of the In 188 midafternoon nest-site observations af- 9 instancesof extrafamilialallopreening, 7 in- ter August fledging through the end of Decem- volvedyoung birds with otherunrelated young ber, at leastone family memberwas present in (with whom they were positivelyassociated), 101 cases.Both adults and young (at least one and 2 involvedbreeding adults mutually allo- of each) were present in 62 cases,and all four preening with adults from neighboring nest nuclear family members were present in 34 sites(with their matespresent in both cases). cases.In someof the meetingsof an entire fam- Although mostextrafamilial observationswere ily, the four membersarrived together from the of youngallopreening unrelated young, young samedirection, whereas in othermeetings some apparentlydo not allopreenother young in- or all membersarrived singly. Nest-site ren- discriminatelywith regard to relatedness.A1- dezvousof families probably occur daily for lopreening between sibs was more common about 3 months after fledging and decline 38 PATRICIAPARKER RABENOLD [Auk,Vol. 103 slowly in frequency thereafter. By 5 months regularly at a single site leads to high associa- after fledging, young were no longer observed tion values among regular visitors to the roost, at the nest sites, and meetings of family mem- primarily young birds, and accounts for the bers other than mates occurred at roosts and high proportionof positiveassociations among feeding sites. unrelated young (Fig. 1) and the higher overall Courtshipand copulation.--Courtshipand cop- proportion of positive extrafamilial associa- ulation did not occur in communal situations. tions involving young from known nest sites Male Black Vultures have a characteristic court- (41%)compared with that of their parents(16%) ship posture that immediately precedes copu- (Fig. 1, P < 0.05). lation and is presentedto the female at the nest Factorsfavoring maintenance of contactamong site. I have witnessed4 copulations;each was family members.--High association between preceded by the courtship posture and oc- mates would be expected in animals that are curredat the nest site. On only 7 occasionsdur- highly gregariousimmediately preceding and ing hundredsof hours of roostand feeding ob- during courtshipand egg-laying, when mate servationsduring Februaryand March (eggsare guardingmight be necessary.However, in Black laid in mid-March) have I seen birds assume Vulturesmates are highly associatedyear-round the male courtship posture. In these 7 in- (except during incubation and brooding, when stances,the posturing bird was approached they take alternate nest shifts), with no marked quickly by severalnearby birds of both sexes increaseimmediately preceding reproduction. that began pecking and biting him violently Although mate guarding could contribute to until he fled the area. high associationduring the reproductiveperi- od, courtshipand copulationdo not occur in DISCUSSION communal situations, but are confined to the nest site and isolated from other birds. Because In their lifetimes, Black Vultures may use birds adopting the courtshipposture in groups many communalroosts. At any one roost the are attackedby nearbybirds, strictmate guard- compositionof individuals changes nightly, ing by individual males is unnecessary. resulting in large-scalemixing of individuals. Parent-offspring bonds persist in part be- Nonetheless, family members maintain close causeof extended postfledgingdependence of contact outside the breeding season in com- young on direct parental feeding. Familieswith munal feeding and roostinggroups. A signifi- low parent-offspring association had lower cantly greater proportion of positive associa- fledgling survivorship than those with high tions involved mated pairsand parent-offspring parent-offspring association. Regurgitative pairs than their extrafamilial pairings, and feedings of offspring fledged in August oc- pooled intrafamilial associationscontained a curred as late as the following April. During greater proportion of postive associationsthan this period young birds will not compete for pooled extrafamilial associations. food at a carcassattended by many adults.They Young birds, in general, have more positive wait at the side for their parents to emerge, extrafamilial associationsthan do their parents. then beg loudly for regurgitativefeedings. The This age difference in associationpatterns may postfledging nest-site rendezvous underscore be partially explained by differential move- the dependenceof young on parents for feed- ment ratesand the relative site fidelity of young ings and feeding assistancein groups. Al- birds during the winter (RabenoldMS). Black though family members,and especiallyyoung Vultures form larger roosting groups during birds, often become separated in hectic morn- winter than summer. The winter roosting sites ing roost departures,daily reassemblyat nest are stable within a winter, although smaller sites assuresthat family members maintain satelliteroosts occasionally may be used.Turn- contactduring the period when young are de- over rates for adults are greater than turnover pendent on parents for feeding and still unfa- rates for young birds at the winter roost, and miliar with many partsof their ranges.During young birds are more likely to return after an their first spring (almost1 yr old) young birds absence.Young birds evidently tend to wait at begin to join large feeding groups, and their the winter roost, whereas adults (including parents aid them by threatening other adults their parents) circulate among several roosts. that attemptto drive theseoffspring away from This tendency for young birds to congregate food. This form of assistance has been docu- January1986] BlackVulture Family Associations 39 mented in other specieswith long-lived asso- sites;those nesting closer together might have ciations between parents and young (Scott more extensiverange overlap than those nest- 1980). As juveniles mature and become more ing farther apart. No suchrelationship was ev- competitive feeders,the relationship with their ident when the associationvalues of breeding parents may evolve into mutual aid-giving adultsnesting within this study area were plot- during competition within a large feeding ted againstdistance between nestsites (Fig. 2). group. That parent-offspring associationsper- It appears instead that some families have sist well beyond the age of strict dependence strongalliances with othersand that this pat- suggeststhat long-term associationscontinue tern of alliances between families is not a func- to be mutually beneficial. tion of distance between nest sites. Aid in feeding may be the largestsingle fac- It is possible,however, that the greatestin- tor favoring retention of strong ties among ternest distance considered (< 11 kin) is trivial members of Black Vulture families. Associa- to a vulture and that it is not surprising, there- tions of family membersremained high when fore, that associationstend to be high with no feeding observationsalone were considered, significantdecline over this distance.My ob- while extrafamilial associations fell to the level servationsof breeding adults suggestthat their expected from random assortment. Because movementsare heavily concentratedwithin 20 their food supply is extremely patchy in space km of their nest sites throughout the year. and time, an individual's chancesof finding Within this restricted range, internest dis- food while alone may be small, and competi- tancesapproaching 11 km include a significant tion at feeding sites can be fierce. Becausea portion of the estimated range of breeding single sourceusually containsenough food for adults. Furthermore, the daily afternoon meet- severalbirds, a finder should recruit his family ings at nest sitesmake it unlikely that families members preferentially to share food with would travel great distancesout of the study them, thereby increasing his inclusive fitness area. If greater internest distanceswere consid- by fostering survival of genetically correlated ered, a decline in associationsmight be found individuals (Hamilton 1964). In addition, a co- that would support the passiverange-overlap alition of individualsmay betterdefend a small hypothesis.Associations must decline with in- carcassor part of a large one.As a consequence, creasinginternest distance.Despite its poten- eachmay feed better than if competingalone. tially large range, an individual bird coversa Are associationsbased merely on range over- finite area,and associationsmust eventually fall lap?--If young birds cover large ranges cen- to zero. It is therefore of interest to look not tered on their natal nest sitesuntil they become only for a decline in associationswith distance, reproductiveand adults centertheir rangeson but at the range of associationvalues within a their breeding sites, immediate family mem- reasonabledistance from eachfamily's nest site. bers would have extensiveor completerange The passiverange-overlap hypothesis predicts overlap. Although their ranges would overlap a smoothdecline from center of range to edge. extensively with many other individuals' Not only was no decline found, but association ranges,family memberswould have stronger values at each internest distancevaried widely associationswith closelyrelated birds than with (Fig. 2); negativeassociation values were found others simply as a result of passiveassociation at even very small internest distances.In fact, from sharing an identical range, even if the for only one family was the most closely asso- birds moved independently of each other ciated family found at the next-nearestnest site. within this range. Extrafamilial associationval- The lack of correlation between associations ues should vary widely depending on the ex- and internestdistances suggests that particular tent of range overlap,which may be correlated associations,both at family and neighborhood with the distance between nest sites. Mates levels, are based on some attribute of the birds would have the highest associations,followed other than wide overlap of ranges. The active by neighboring breeding adults, followed by nature of intrafamilial associationsis also sup- nonmateadults, including all adultsregardless ported by conspicuousallopreening and in- of nestsite. These predictions match my results tercessionsin fights. Becauseintrafamilial as- (Table 3). However, within the neighbor- sociationis so marked, it may be that breeding neighbor pairs, associationvalues should de- adults of highly associatedfamilies are also cline with increasing distance between nest closely related. Although no birds marked as 40 PATRICIAPARKER RABENOLD [Auk, Vol. 103 nestlingshave bred in the study area, 3- and 4- is sufficiently patchy, entire families could yr-old birds are still in the area and use local searchwithout success,and they couldthen rely roosts.Age of first breeding is not known for on the larger group for foraging information. this species.If many birds breed in their natal Even if they are followers,cooperation of en- area and intrafamilial associationspersist, it is tire families in competitive groups may in- likely that neighborsmay include parent-off- creasetheir chancesof feeding. If families or spring or sibling relationships. larger alliances of families function as units Roostsas informationcenters.--Perhaps the competing for food, no one nest site would chief theoretical criticisms of Ward and Zaha- providea suitablemeeting place. Families could vi's (1973)information-center hypothesis is that instead meet allies and relatives at neutral roost it relies on reciprocationbetween finders and sites.Beyond the observationsreported here of followers.Unless reciprocationis assured,the family membersaiding each other in roosting participation of successfulforagers appears and feeding interactions,detailed observations paradoxical.Why should they return, only to of feeding interactionsare neededto verify the be followed by competitors,unless there is a competitive advantage of individuals feeding yet stronger advantage in joining an aggrega- in the presenceof family membersor allies. tion, such as reduced risk of predation (Alex- The evidence reported here suggeststhat ander 1974, Hoogland and Sherman 1976)?A Black Vulture communal roosts serve as meet- commonassumption is that food is sufficiently ing placesfor family groups and allied fami- abundant within a patch that competition lies. Family members associatewith each other would be weak and attemptsto dissuadefol- preferentiallyand assisteach other in compet- lowers would be pointless(Krebs 1978). How- itive interactionsin large roostingand feeding ever, there must be a maximum number of fol- groups.The extent to which kin selectionhas lowers beyond which this would not be true. shaped the evolution of communal roosts in Furthermore, the field evidence in support of Black Vultures cannot be estimated without this hypothesishas to date been gathered on greater knowledge of the relatednessof allied colonially breeding birds (Krebs 1974, Erwin family groups.If young birds eventually breed 1978,Waltz 1981),in which successfulforagers near their natal areasand family bondspersist, must eventually return to feed young. In these allied neighbors may also be closely related. cases,predator defense or localizednesting sites The possibility that large roosting groups con- becomeparticularly plausible explanationsfor sistof a few large extendedfamily units or clans aggregation. Black Vultures have no known is intriguing in light of the information-center predatorsas adults and thus representa fairly hypothesis. This study demonstratespositive simple systemfor examiningcertain aspectsof associationsand cooperative behavior among the information-centerhypothesis. Because they family members,and I show elsewhere(Raben- are not colonial breeders,members of the pop- old MS) that following from the roostand con- ulation that are vulnerable to predation (nest- tinued participation in a roost might be dis- lings) are not presentin the aggregations.Also, couraged in unrelated birds by aggressive successfulforagers need not return to a specific interactions in the roost. While these results do aggregationsite to feed nestlingsand they could not present direct evidence that foraging in- conceivably avoid competitors by sleeping formation is transferredat roosts,they suggest alone, although I have no evidence that such that potential recipients of information are "cheating" occurs.While a single bird roosting often family members,a conditionunder which solitarily would be difficult to detect, I have the evolution of information sharing is more many examplesof birds seenat carcassesjoin- likely than amongbirds assorted randomly with ing a roostinggroup the sameevening (Raben- respectto relatedness. old MS). The paradoxof the returning successfulfor- ACKNOWLEDGMENTS ager is partially resolvedif it returnsto lead its This work was supportedin part by a grant from family membersto food. But why should fam- the Frank M. Chapman Fund of the American Mu- ilies not maintain separate fixed aggregation seumof Natural History and in part by a UNC Re- centers(such as their nest sites)to which they search Fellowship. The cooperation of local land- return with foraging information to avoid owners,especially the A. J.Perry family and the entire sharing with nonrelatives?If the food supply Lindley clan, is greatly appreciated.Celia Lewis was January1986] BlackVulture Family Associations 41 an untiring assistantin all phases. B. Brown, S. McHARGUE,L.H. 1977. Nesting of Turkey and Black Campbell, C. Christensen,K. Long, K. Meyer, H. vultures in Panama. Wilson Bull. 89: 328-329. Mueller, B. Simpson,M. Wallace,and especiallyP. PIELOV,E.C. 1976. Mathematicalecology. New York, Frederickand K. Rabenoldhelped in variousphases Wiley. of the fieldwork. E. Patterson and B. Bromer wrote RABENOLD,P.P. 1983. The communal roost in east- and rewrote many computerprograms. The manu- ern cathartid vultures--an information center? scripthas profited considerably from commentsby J. Pp. 303-321 in Vulture biology and management A. Jackson, H. C. Mueller, K. N. Rabenold, R. H. Wi- (S. R. Wilbur and J. A. Jackson, Eds.). Los An- ley, E. C. Waltz, and a very helpful anonymousre- geles,Univ. California Press. viewer. I especiallythank my majorprofessor, H. C. REEVES,T. L., g. BLAIR,J. O. FEARRINGTON,B. HORNY, Mueller, for guidanceand encouragement. & T. TALLEYß1970. Land developmentpotential study. Prepared for the County of Chatham, LITERATURE CITED North Carolina. HUD Project#NCP77. SCOTT,D. K. 1980. Functional aspectsof prolonged ALEXANDER,R.D. 1974. The evolution of social be- parental care in Bewick's Swans. Anim. Behav. havior. Ann. Rev. Ecol. Syst.4: 325-383. 28: 938-952. DICE,L.R. 1945. Measuresof the amountof ecologic SIEGEL,S. 1956. Nonparametricstatistics for the be- associationbetween species.Ecology 26: 297-302. havioral sciences. New York, McGraw-Hill. ERWIN,R. M. 1978. Coloniality in terns: the role of SOKAL,R. R., & F. J. ROHLF. 1969. Biometry. San socialfeeding. Condor 80: 211-215. Francisco, W. H. Freeman. HAMILTON,W.D. 1964. The genetical evolution of WALLACE, M.P., P. G. PARKER,& S. A. TEMPLE. 1980. social behavior. J. Theoret. Biol. 7: 1-52. An evaluation of patagial markers for cathartid HOOCLANV, J. L., & P. W. SIaERMAN. 1976. Advan- vultures. J. Field Ornithol. 51: 309-314. tagesand disadvantagesof Bank Swallow (Ripar- WALTZ,E.C. 1981. The information center hypoth- ia riparia)coloniality. Ecol. Monogr. 46: 33-58. esisand colonialnesting behavior. Unpublished JACltSON,J.A. 1975. Regurgitativefeeding of young Black Vultures in December. Auk 92: 802-803. Ph.D. dissertation,Syracuse, State Univ. New York Coil. Env. Sci. and Forestry. KREBS,J. R. 1974. Colonial nesting and socialfeed- ß 1982. Resource characteristics and the evo- ing as strategiesfor exploitingfood resourcesin lution of information centers. Amer. Natur. 119: the Great Blue Heron (Ardea herodias).Behaviour 73-90. 51: 99-134. WARD,P., & A. ZAHAVl. 1973. The importanceof ---. 1978. Colonialnesting in birds,with special certain assemblagesof birds as "information- referenceto the Ciconiiformes.Pp. 299-314 in centres"for food-finding.Ibis 115:517-534. Wading birds (A. Sprunt, J. C. Ogden, and S. Winckler, Eds.). New York, Natl. Audubon Soc. Res. Rept. No. 7. COMPARATIVE REPRODUCTIVE SUCCESS OF YELLOW-SHAFTED, RED-SHAFTED, AND HYBRID FLICKERS ACROSS A HYBRID ZONE

WILLIAM S. MOORE 1 AND WALTER D. KOENIG 2 •Departmentof BiologicalSciences, Wayne State University, Detroit, Michigan 48202 USA, and 2HastingsReservation and Museum of VertebrateZoology, University of California, StarRoute Box 80, CarmelValley, California 93924 USA

ABSTR^CT.--Alternativehypotheses of hybrid zones make specificpredictions about re- productivecomponents of fitnessin the hybrids. The dynamic-equilibriumand reinforce- ment hypothesesare premised on reduced hybrid fitness, which should be apparent as reducedclutch or brood size or as increasedembryonic mortality. The hybrid-superiority and introgressionhypotheses predict normal clutchand broodsize and embryonicmortality. Reproductivesuccess was measuredat four study siteson a transectacross the hybrid zone betweenthe Yellow- (Colaptesauratus auratus) and Red-shafted(C. a. caret)subspecies of the Northern Flicker. Two additional clutch size samplesrepresenting pure Yellow- and Red- shafted flickers were obtained from museum egg collections.Mean clutch size did not differ significantly among the six samples.Factorial ANOVAs showed that early clutches and broodsare larger than late clutchesand broods,but no significant difference was detected betweenhybrid and parentalstudy sites. Analyses of the effectof phenotype(yellow-shafted, red-shafted,hybrid) also suggestthat neither clutch size nor brood size is affected,with the exceptionthat hybrid males sired significantlysmaller broods. Finally, there were no sig- nificant effectsof type of cross(red-shafted male x hybrid female, etc.) on the ratio brood- size / clutch-size. The only evidencefor reducedhybrid fitnesswas in the test where males with hybrid phenotypesappear to have sired smallbroods. This may indicatethat abnormalbehavior of hybrid malesaffects female fecundity,but it is alsoplausible that this marginallysignificant result is a type I statisticalerror. The overall lack of evidencefor reducedhybrid fitnessis inconsistentwith either the dynamic-equilibriumor reinforcementmodels. Of the two re- maining alternatives,the boundedhybrid-superiority model appearsthe more likely expla- nation of the Northern Flicker hybrid zone becauseearlier work (Moore and Buchanan1985) showed that the hybrid zone is not becoming broader, as predicted by the introgression model. Received20 February1985, accepted 2 July 1985.

HYBRID zones often occur between avian diverged in isolation but not to the extent that populations that have diverged to near the hybridization would disrupt distinctly co- species level but retain contiguous distribu- adapted gene complexes resulting in hybrid tions as a result of either secondarycontact or breakdown or hybrid unfitness. He further parapatric divergence (Short 1969, Moore 1977, suggestedthat the secondarycontact was estab- Rising 1983a). Typically, hybridization is lished as distinct ecological communities ex- rampant in the zone but the zone itself is very panded from refugia to form a "suture zone" narrow and seemsto act as a barrier separating in the sense of Remington (1968) and that the markedly divergent plumage patterns or song genes of the divergent charactersare either types, or both. Furthermore, the biogeography adapted in their respective communities or of most avian hybrid zones suggeststhat they closely linked to genes that are. The hybrid arosein antiquity, probably as a result of Pleis- zone persists,then, becauseit occursin an eco- tocene glaciation, and therefore seem to be tone in which neither parental taxon is partic- evolutionarily stableconfigurations (Short 1970, ularly well adapted. This was termed the hy- Moore 1977; but see Barrowclough 1980). brid-superiority model, although bounded Three theories have been proposed to ex- hybrid superiority is more apt becauseit im- plain the existence of stable hybrid zones. plies that hybrid superiority is restricted geo- Moore (1977)suggested that hybrid zonesrep- graphically. resentsecondary contacts between taxa that had Barton (1979) and Barton and Hewitt (1981) 42 The Auk 103: 42-51. January1986 January1986] ReproductioninHybrid Flickers 43 developed an alternative hypothesisthat ex- populationsof the Northern Flicker (Colaptes plainsstable hybrid zonesby balancingoppos- auratus)along a transectacross the hybrid zone ing forces.The essentialfeature of the Barton- between the Yellow- and Red-shafted flickers Hewitt dynamic-equilibriummodel is hybrid un- (C. a. auratusand C. a. cafer) in western Ne- fitness,which prevents the hybrid zone from braskaand easternWyoming. The two central becoming broader through introgressive hy- localessupport hybrid populations, whereas the bridization; less-fit hybrids form a narrower distal locales represent parental populations hybrid zone. Barton(1979) showedthat hybrid (see Fig. 1). Our purpose is to determine unfitnessitself can fix the width of the hybrid whether hybridization between these well- zone and further showed that variations in marked subspecieshas disrupted coadapted population density could prevent the hybrid gene complexesand whether hybrid break- zone from "flowing" geographically. down is essentialto understandingthis hybrid The third alternativeis that the hybrid zone zone.Data pertainingto historicalstability were appearsstable only to a short-lived observer reportedearlier (Moore and Buchanan1985). but actually is growing broader through intro- The structureof the hybrid zone has been gressive hybridization or becoming more re- reportedby Short (1965) and Moore and Bu- strictedas premating reproductiveisolation is chanan (1985). To summarize briefly, the Yel- reinforcedby selectionagainst hybrids (Wilson low-shaftedFlicker is broadlydistributed in di- 1965, Remington 1968). Whether introgression verse woodland types across eastern North or reinforcementis expectedwould dependon America, whereas the Red-shafted Flicker is the whether hybrid phenotypesare at leastas fit as western North American counterpart.The hy- the parental phenotypesor lessfit, respective- brid zone occurs on the western Great Plains, ly. primarily in riparian woodlands(cottonwood, The three alternativesprovide radically dif- peach-leavedwillow, and green ash) but also ferent explanationsof hybrid zones,each with in the coniferous forests of the Black Hills, the its own implicationsfor speciationtheory. Fur- Pine Ridge region of Nebraskaand South Da- thermore,the alternativesappear to be testable, kota,and in the CypressHills of Saskatchewan. as the dynamic-equilibriumand reinforcement The hybrid zone may be continuousin the sol- models predict somemeasure of hybrid break- idly forestedmontane regions of Alberta, Brit- down, whereasthe hybrid-superiorityand in- ish Columbia, and southern Alaska (Short 1965, trogressionmodels do not. More precisely,ac- Moore pets. obs.).The two subspeciesexhibit cording to the dynamic-equilibrium and contrasting conditions for 6 (5 in females) reinforcement models, one would expect in- plumagecharacters; e.g. red-shaftedmales have creasesin infertility and developmentaberra- a red malar stripe,whereas that of the yellow- tions in hybrid phenotypes.The pairs of alter- shafted is black. Hybrids exhibit intermediate natives can be tested further by studying the variations. These charactersare easily scored, historical stability of the hybrid zone. even with binoculars.Mating is random with The idea that hybridization between dis- regard to the plumage characters(Short 1965, tinctly coadaptedgene poolsresults in hybrid Bock1971, Moore and Buchanan1985). The hy- breakdown is an old idea in evolutionary bi- brid zone varies in width and is asymmetrical, ologythat is of particularimportance to specia- with introgressionevident farther west of cen- tion theory. Collectively,the assortedmaladies ter than east (see Fig. 1). Population densities that are expectedto result are termed postmat- are uniformly high along the study transect ing reproductive isolating mechanisms.Al- (Moore and Buchanan 1985). though hybrid breakdown could manifest at any stagein the life cycle,the maladiesmost MATERIALS AND METHODS often mentioned affectthe early development or fertility of hybrid phenotypes.Thus, where Active nestswere locatedthroughout May and June natural hybridization has disrupted coadapted in 1981-1984 at four study sites on a transectacross the hybrid zone (Fig. 1). The study sites are: Suth- gene complexes,one would expecthigher fre- erland (JamesFIaugland farm on the south bank of quenciesof developmentalanomalies and re- the South Platte River between the river and inter- duced fertility. In birds this should appear as statehighway 1-80, 1.75 km south-southeastof South- reduced brood and clutch sizes,respectively. erland, Lincoln Co., Nebraska, R33W, T14N, secs. 32, We reportclutch- and brood-sizedata for four 33); Bridgeport(Blanchard-Lindgren ranch on the 44 MOOREAND KOENIG [Auk,Vol. 103

CON?OUR MBP OF ?LIOKER HYBRID ZONE

o '•11'•"H 111'2'• 'N 108'•,8'H 108"12'H 10'3'•6 'H lnl* H 98'2•'H 95"•'H LONGITUDE Fig. 1. Provisional contour map of the Northern Flicker hybrid zone on the western Great Plains. The solid squaresmark the four live-studysites (from eastto west:Sutherland, Bridgeport, Morrill, and Wheat- land). The contourlevels indicatethe transitionin Short's(1965) hybrid index (HI) from yellow-shaftedin the east (HI = 0) to red-shaftedin the west (HI = 23). The dashedcontours and questionmarks indicate poorlycollected areas that cannotbe accuratelymapped. The map is accuratein the regionof the live-study sites.

south bank of the North Platte River, 7.5 km west- pleteness,the eggsin a given nestwere countedover northwestof Bridgeport,Morrill Co.,Nebraska, R51W, severaldays as laying progressedand incubationbe- T20N, secs. 21, 22, 23); Morrill (north bank of the gan. When clutch size remained the samefor at least North Platte River, 4.8 km west-northwest of Morrill, two successivedays, the clutch was consideredcom- Scottsbluff Co., Nebraska, R58W, T23N, secs. 13, 14); plete. In some casesit was apparent that the clutch and Wheatland (Laramie River at the confluence of was complete at the time the nest was discovered Sybille Creek, 13 km west-northwestof Wheatland, becauseclutch size did not increase or the eggs Platte Co., Wyoming). Bridgeportand Morrill were hatchedin lessthan 11 days,the normal incubation establishedas study sites in 1981; Sutherland and period for flickers(Sherman 1910,Moore pets. obs.). Wheatland were not established until 1982. Broodsize.--An accessport was cut into the nest Nest cavities were reached with the aid of a 9.6-m cavity during the broodingperiod. At that time, the extensionladder and climbinggaffs where necessary. hatchlingswere banded and counted,and the adults Cavity interiors were viewed with a 6-cm circular were trapped, banded, and scoredphenotypically. mirror on a handle and a penlight. Usually, the accessport was cut and the hatchlings The following data were collectedfor each nest: banded in the afternoon of one day, and the adults score for each plumage characterfor both parents were trapped the following morning. Adults were (seebelow), clutch size, brood size, and, when pos- trapped in the nest cavity by a remote-activatedtrap sible, date when the first egg was laid. door modified from a common rat trap. The trapped Clutchsize.--It was necessaryto establishthat the adult was then removedthrough the accessport. Fol- clutcheswere completeat the timesof the final counts. low-up observationson numerousnests suggest that With rare exceptions,flickers lay one egg per day the trapping and banding activities did not perma- (Sherman 1910, Moore pets. obs.). To establishcorn- nently disrupt the nesting process;nevertheless, January1986] ReproductioninHybrid Flickers 45 brood size was basedon the number of hatchlingsin TABLE1. Average hybrid index scores,percentage the nest at the time it was first disturbed rather than transition from pure yellow-shafted(0%) to pure the number that subsequentlyfledged. The ages of red-shafted(100%) scores, and samplesizes (n) on a transectacross the Northern Flicker hybrid zone. the broods at the time hatchlings were counted var- ied between 10 and 17 days after hatching. Percentage In mostcases, Short's (1965) hybrid index (HI) was of pure determined for both adults at the time the nest was red-shafted discovered.This initial phenotypicdetermination was score done by observingthe adults around the nest with Average hybrid index Fe- binoculars. In those cases where the adults were sub- Studysite Female(n) Male (n) male Male sequentlytrapped, the HI was redetermined on the bird in hand. Priority was given to His determined Sutherland 0.73 (11) 0.69 (13) 3.8 3.0 on hand-held birds, but in some cases when it was Bridgeport 4.85 (26) 7.09 (32) 25.5 30.8 not possibleto trap one of the adults (usuallythe Morrill 13.64 (25) 16.22 (27) 71.8 70.5 Wheatland 15.71 (14) 18.19 (16) 82.7 79.1 female),the HI determinedusing binoculars was used. The concordancebetween scoresdetermined using binocularsand in the hand was strong. Short's(1965) hybrid index is basedon 6 plumage within longitudes105-113øW and latitudes38-42øN. characters in males. The 6 characters are listed ac- These two blocks are roughly contiguouswith the cording to the following format: character(abbrevi- easternand western ends of the study transect;they ation, auratus/cafer),where auratusand cafer repre- are narrow so as to avoid varianceresulting from the sent the relative states for the parental subspecies. strong latitudinal gradient in clutch size in Colaptes The plumage charactersare: crown color (CROWN, (Koenig 1984).Koenig's data were extractedfrom the gray/brown); ear covert (EAR, brown/gray); throat egg collectionsof museums,and we followed his color (THROAT, brown/gray); red nuchal patch protocolfor determining the completenessof clutch- (NUCHAL, present/absent); shaft color (SHAFT, es. Only clutcheswhere incubationwas evidenced lemon-yellow/salmon-red);and malar stripe (MA- by embryonic development were included in our LAR, black/red). All of thesecharacters except MA- samples. LAR can be scored in females; females lack the malar stripe.Each character is given an integer scoreon the RESULTS scale 0-4, where 0 and 4 represent the auratusand caferparental states, respectively, and the integersI- 3 representproportionally intermediate conditions. Phenotypesof adultsat the four studysites.- In theory,then, pure auratusmales and femaleswould The averagehybrid index (HI) scoresfor nest- have His of 0, and cafermales and femaleswould ing adults at the four study sitesare presented score24 and 20, respectively.In practice,however, in Table 1. The two end sites, Sutherland and CROWN scoresare never higher than 3 in this por- Wheatland, ideally representpure yellow- and tion of the speciesrange. We reserved the scoreof 4 red-shaftedlocales for comparisonwith the hy- for the deeper brown crown color of the northwest- brid locales, Bridgeport and Morrill. Suther- ern race (Short 1965), but this variation appearsun- land is close to pure yellow-shafted, but sig- related to hybridization. In any case,pure red-shaft- nificant numbers of hybrid phenotypes are ed malesand femaleswould have hybrid indicesof included in the Wheatland samples. For this 23 and 19, respectively,at the latitudeof our transect. Becausesome nests were depredated after clutch reason and becausesample sizes from Wheat- size was determined and others were found after the land are small, greater considerationshould be eggs hatched, there is variation in sample size for given to Sutherlandas a parental population clutchesand broods.Sample sizesare given in the for comparison. results section in the context of specific tests. SPSS Descriptivestatistics.--The distributions of (release9.2) was usedto perform statisticaltests (Nie clutch and brood sizes for the six samplesare et al. 1975). illustrated in Fig. 2, along with the means, Sutherland and Wheatland were intended to serve standard deviations, and sample sizes. Brood- as yellow-shaftedand red-shaftedreference samples, size data were not available for pure red- and respectively.However, there is someevidence of in- yellow-shafted samplesbased on museum egg trogressionat both localities. Two additional refer- encesamples were extractedfrom the clutch-sizedata collectionsand so only clutch-sizedata are giv- compiledby Koenig (1984). One sample represents en for these samples. pure yellow-shafted and includes clutcheswithin the The distributions of clutch and brood sizes block delineated by longitudes 89-96øW and lati- at all localesappeared sufficiently close to nor- tudes 38-42øN; the second sample includes clutches mal distributionsthat parametricstatistical tests 46 MOOREAND KOENIG [Auk,Vol. 103

Clutch Size Brood Size study sites,it seemedas though clutch sizesof early-nestingpairs were larger than those of late nesters and that there was also variation between years. Uncontrolled sourcesof vari- ance could obscureany underlying difference between locales or alternatively could bias the test. We tested for the effects of seasonal vari- ation (early vs. late clutches)and between-year variation by a factorial ANOVA with site (1- 4), season(early, late), and year (1982, 1983, 1984) as factors. Three problemsin the design of this three- factor ANOVA should be mentioned. First was how to separateearly and late clutches.We plotted a histogramof first-egg dates.A con- spicuousbimodality was manifest, with a val- ley around 20-21 May. This is consistentwith the field observationof a period in late May during which new nests are more difficult to find than earlier or later. Thus, two levels were defined for the factor season;early comprises nests in which laying began 20 May and ear- lier, and late comprisesnests in which laying began 21 May and later. The secondproblem is that Sutherland and Wheatland were not es- tablished as study sites until 1982; thus, data are availableat Bridgeportand Morrill for 4 yr, 1981-1984,but for only 3 yr at Wheatland and Fig. 2. The distributionsof clutchand brood sizes Sutherland. For this analysis,then, only the acrossthe Northern Flicker hybrid zone. The ex- years 1982-1984 were considered.The third treme upper and lower clutch panels are based on problem is that the numbers of observations museum egg collections;the other distributionsare (clutches or broods) in the 24 cells resulting basedon the four live-study sites.The horizontal line from the cross classification of the three factors above each histogram is a 95% confidenceinterval centered at the mean. are unequal, which means that the effectsare not orthogonal.The regressionapproach of ad- justing sumsof squareswas used to correctfor were valid in mostinstances and provided more nonorthogonality(Nie et al. 1975). sensitivestatistical tests than did correspond- The resultsof this analysisare that the effect ing nonparametric tests. of seasonwas very significant (P < 0.002), but Clutch-size comparisonsacross study sites.- neither site (P < 0.382) nor year (P < 0.240) Variances in clutch size were homogeneous was significant,nor were any of the interaction acrossthe four study sites and two museum terms. samples(Bartlett-Box F-test, P < 0.87). This is Becausethe factor year was not significant of interestbecause disruption of coadaptedgene but seasonwas, the most parsimoniousANO- complexesvia hybridization might be expected VA model for the final comparisonof mean to increase variance in clutch size. Although clutch size between sites consideredjust two not significant,clutch-size variances were low- factors, site and season.In this ANOVA (Table est at the two hybrid locales.Because the vari- 2), again using the regressionapproach to cor- anceswere homogeneous,we comparedmean rect for nonorthogonality, the effect of season clutch size acrosslocales by completely ran- remained significant (P < 0.001), with early domized-designANOVA. In this analysis,the clutchesaveraging 7.82 eggs vs. 6.80 for late mean clutch size did not differ significantly in clutches, whereas the effect of site was even any of the six samples(P < 0.363). less significant (P < 0.561) than in previous As the data were collected from the four live- tests. January1986] Reproductionin Hybrid Flickers 47

Brood-sizecomparisons across study sites.--Brood T^•3t•E2. The effect of site and season on mean clutch size was analyzed in exactly the same manner size along a transectacross the Northern Flicker as clutch size with similar results.The analysis hybrid zone. of clutch size provided no statistical evidence for reduced hybrid fitness,and the brood-size Sum of Mean statis- Attained statistics were even less compelling. Mean Source squares df square tic significance brood size was actually greater at Bridgeport Total 180.39 95 ------than Sutherland(Fig. 2), although,again, there Site 3.63 3 1.21 0.69 P < 0.561 were no significantly different means. Brood Season 22.00 1 22.00 12.52 P < 0.001 size was correlated with clutch size (r = +0.62, Site by season 0.12 3 0.04 0.02 P < 0.995 P < 0.0001), and so the more refined factorial Error 154.64 88 1.76 -- -- designsyielded much the same results as the clutch-size analyses,viz., year did not affect brood size but season did (P < 0.014). In the two-factor ANOVA, site did not affect brood previousanalysis. Females with HI scoresof 3- size (P < 0.651). 17 inclusive were classified as hybrids, those The analysesof clutch and brood size were with scores<3 asyellow-shafted, and those> 17 similar, but it is important to note that clutch as red-shafted. For males, individuals scoring and brood size probably would be affectedby between 3 and 21 inclusive were classified as different componentsof fitness.Although ab- hybrids, those <3 asyellow-shafted, and those normal male behavior conceivablycould affect >21 as red-shafted. the female reproductivecycle (Lehrman 1965), The resultsof theseanalyses are summarized clutch size should be affectedprimarily by fe- in Table 3. Adults from the four study sites male fertility, whereas brood size could be af- were pooled and reclassifiedby phenotype. fected by male fertility as well as zygotic mor- Again, to control the effect of early vs. late tality. It is possible to assessthe effect of nesting, a two-factor ANOVA, season (early, hybridization on development alone by ana- late) x phenotype(yellow-shafted, hybrid, red- lyzing brood size while statisticallycontrolling shafted),was employed for the comparisonsof for clutch size. This might be accomplishedby clutch and brood sizes. However, because there comparing the slopes of lines of brood sizes were no data for late broodssired by red-shaft- regressedon clutch sizes for the four sites by ed males, the ANOVA of male brood was re- analysisof covariance.However, such an anal- stricted to two phenotypic classes,i.e. season ysis may not be valid; brood sizesare not nor- (early, late) x phenotype (yellow-shafted,hy- mally distributed for a given clutch size be- brid). In the four testsof the effect of pheno- causein too many casesbrood size is equal to type on clutch and brood size, hybrid males clutch size. A cruder but safer test can be ac- sired significantly (P < 0.04) smaller broods complishedby comparingthe ratios of brood- than nonhybrids; the other three testswere not size/clutch-size across sites by the nonpara- significant. metric Kruskal-Wallis test. The test of the null This result provided some evidence for re- hypothesisthat the central tendencies(means) ducedhybrid fitness.However, a more detailed of this ratio are identical across sites cannot be examinationof this apparentreduction in brood rejected (P < 0.485; sample sizes: Sutherland = size was enigmatic. Either reduced fertility in 20, Bridgeport = 26, Morrill = 18, Wheatland = hybrid malesor increasedfrequency of devel- 8). Although the validity of an analysisof co- opmental abnormalitiesin zygotessired by hy- variance is suspect,such an analysisgave a very brid males may be the cause. In either case, similar result. Thus, there is not evidence of clutch size should not be reduced, but the ef- hybrid unfitness in the form of reduced brood fects should be apparent in a reduced ratio of size at hybrid locales. brood-size/clutch-size. A paradox is then ap- Clutch- and brood-sizecomparisons by pheno- parent in Table 3 becausethe brood-size/clutch- type.--Males and females can be classified as size ratio was not significantin itself (Kruskal- pure yellow-shafted, pure red-shafted, or hy- Wallis test), and both a reduced clutch size and brid based on HI scores. Clutch and brood sizes an increasedzygotic mortality seemedto con- can then be compared between phenotypic tribute about equally to the statisticallysignif- classesas opposedto between locales,as in the icant reduction in brood size. It is possiblethat 48 MOOREAND KOENIG [Auk,Vol. 103

TABLE3. Averageclutch and broodsizes and their ratiosfor pure Yellow-shafted,pure Red-shafted,and hybrid flickers(sample sizes in parentheses).

Clutch size Brood size Early Late Early Late Brood/ clutch Femalephenotype Yellow-shafted 7.42(12) 6.63(8) 6.11(9) 5.00(6) 0.835(15) Hybrid 7.84(25) 6.71(17) 6.53(28) 5.83(12) 0.833(34) Red-shafted 7.40(5) 8.00(2) 6.80(5) 6.00(1) 0.903(5) Test Phenotype P < 0.61a P < 0.39a p < 0.68c Season P < 0.37 P < 0.18 Male phenotype Yellow-shafted 8.06(16) 7.50(2) 7.00(15) 7.00(2) 0.898(16) Hybrid 7.27 (30) 6.61 (26) 6.14 (28) 5.40 (20) 0.826(42) Red-shafted 8.60 (5) 7.00 (2) 8.00 (4) -- 0.866(4) Test Phenotype P < 0.15' P < 0.04b P < 0.17c Season P < 0.09 P < 0.53 ANOVA: phenotype(yellow, hybrid, red) x season(early, late). ANOVA: phenotype(yellow, hybrid) x season(early, late). Kruskal-Wallis test.

two samplingerrors compounded to producea DISCUSSION false rejectionof the null hypothesisor, alter- natively, that hybrid malessomehow affected Severalpoints concerningecologically and the egg-laying capacityof their mates--per- geographicallyadaptive variation in clutchsize hapsthrough behavior (Lehrman 1965). need to be considered,as they may be con- Theeffects of particularcrosses on hatchlingsuc- founding our attempt to identify the effectsof cess.--The data set allowed one additional set hybridization (Koenig 1982, 1984). Latitudinal of testsfor reducedhybrid fitness:the effects variation acrossthe four study sites, which of specific crosses(e.g. yellow-shafted fe- spanned about 1ø, is minimal. According to male x hybrid male, etc.) on clutch and brood Koenig's(1984) regression of clutchsize on lat- size (Table 4). Of the nine possibletypes of itude for the Northern Flicker,the expecteddif- crosses,none involved red-shafted x yellow- ference in clutch size between Sutherland and shafted and only one involved red-shaftedx Wheatlandis only 0.1 egg.However, ecological red-shafted.The average ratio (brood-size/ differencesbetween study sites may have influ- clutch-size)for all nestswas pooled from the enced our results. In particular, there is both four sites(Table 4). The ratio for hybrid males an elevational increase from Sutherland (902 weighted and averagedacross all types of fe- m) to Wheatland (1,448 m) and an ecological maleswas 0.826.The grand mean for all crosses gradientof increasingaridity and harsherwin- was 0.838. The means (central tendencies) were ter conditions.These differences may produce comparedby the Kruskal-Wallis test. The null greater seasonalityin productivity, found by hypothesisthat the sum of ranks for the six Koenig to significantly influence clutch-size classes(excluding the crosswhere n = 1) are variationin the Northern Flicker,and thus may identical could not be rejected(P < 0.336). Al- be partially responsiblefor the (nonsignifi- though not significant,the hatchlingsurvival cantly) larger clutch sizesfound in Wheatland rate (brood-size/clutch-size) was lowest for the and in the pure red-shaftedsamples. crosshybrid x hybrid. A similar analysiswas Suchan effectis likely to be minimal for sev- made for clutch size, but there were no signif- eral reasons.Winter actualevapotranspirations icantdifferences between classes as defined by (AE), defined as the sum for the three consec- the type of cross.This is not surprisingbecause utive months yielding the lowest AE values, the female alone probably determines clutch are 0 for all sites (Thornthwaite Assoc. 1964). size. AE was used by Ricklefs (1980) and Koenig January1986] ReproductioninHybrid Flickers 49

TABLE4. Hatchling survivorshipas measuredby the ratio of brood size to clutchsize for variousnatural crossesfrom the hybrid zone (samplesizes in parentheses).

Phenotype of male parent Yellow-shafted Hybrid Red-shafted Weighted meana Phenotypeof female parent Yellow-shafted 0.838 (21) 0.823 (12) -- 0.833 (33) Hybrid 0.911(5) 0.811(26) 0.888(3) 0.833(34) Red-shafted -- 0.929 (4) 0.800 (1) 0.903 (5) Weightedmean a 0.852(26) 0.826(42) 0.866(4) 0.838(72) Weighted meansfor each phenotype averagedover all crosses.

(1984) as an index of winter productivity. Oth- rather a failure to find a reduction in these re- er factorspotentially influencing seasonalityof productive parametersin hybrid phenotypes. resourceseither did not vary among sites(pop- If reducedhybrid fitnesswere a factormain- ulation density) or do not influence geographic taining the Northern Flicker hybrid zone, we patternsof clutch size in this species(summer would expecta reductionin one or more of the AE; Koenig 1984).Thus, we feel that sourcesof three reproductive parameters (clutch size, ecological variation in clutch size are as com- brood size, or the ratio of the two) at the two parable among these sites as possiblein any central hybrid locales(Bridgeport and Morrill) study encompassinga geographicalgradient. relative to the two distal locales (Sutherland One ecologicalfactor, date, had a significant and Wheatland)and the two museumegg col- influence on clutch size, with clutch size being lection samples.This was not the case.Reshuf- significantly lower in late clutches. Koenig fling the data such that the reproductivepa- (1984) found that clutch size increaseswith date, rameters were compared across adult whereasthere was a highly significantdecrease phenotypes(Table 3) resulted in one signifi- in late clutchessizes at the four live-study sites cant test. Specifically,males with hybrid phe- reportedhere. The obvious,although not nec- notypesseemed to be associatedwith reduced essarilycorrect, explanation for this seemingly brood size, but, as discussed in the results, this contradictoryresult is that Koenig's samplein- was an enigmatic result when explained in cluded clutchesfrom a broad range of latitudes; terms of reducedhybrid fitness.When the data clutches from low latitudes are smaller, and set was reshuffled again, such that develop- they are also initiated earlier, on average,be- mental success(brood-size/clutch-size) could causebreeding commences earlier in the south. be comparedbetween different types of cross Our goal was to determine whether there is (Table4), there were no significantdifferences. reduced fertility or increased developmental Consideringcomparisons for clutch size, brood failure in flickers from the Northern Flicker size, and the ratio between the two, 10 statis- hybrid zone. Evidence of reduced hybrid fit- tical testswere performed where reduced hy- nesswould suggestthat this apparently stable brid fitnessmight have manifested.Only one hybrid zone is best explained by a dynamic- test was significant (P < 0.04), and this could equilibrium model. Failure to find this kind of be attributed reasonablyto chancealone. dysgenesiswould favor a boundedhybrid-su- Becauseof limited sample sizes we cannot periority model or some other explanation stateconclusively that there is no reduction in (Moore 1977). In testing the alternatives, it is hybrid fitnessin the Northern Flicker hybrid important to point out that the hybrid-superi- zone; however, we can state that if it is there, ority model refers to ecologicalor exogenous it is very slight. This overall lack of evidence fitnessparameters, whereas the dynamic-equi- for hybrid unfitnessis inconsistentwith either librium model requires depressionof endoge- the dynamic-equilibrium or the reinforcement nous fitnessparameters in hybrids, i.e. infertil- model. Of the two remaining alternatives ity, developmental aberrations, failure to (Moore 1977), the bounded hybrid-superiority maintain physiologicalhomeostasis, etc. Thus, model appearsthe more likely explanationof the bounded hybrid-superiority hypothesis the Northern Flicker hybrid zone becausethe doesnot predict increasedclutch size, etc., but hybrid zone is not growing broader, as pre- •0 Moo• ^Nt•KO•N•G [Auk,Vol. 103 dicted by the introgressionmodel (Moore and tive isolatingmechanisms do not evolve in hy- Buchanan 1985). brid zoneseven when the hybrids are selected The Northern Flicker hybrid zone is one of against.When somemeasure of premating re- a substantialnumber of avian hybrid zones(see productiveisolation (assortative mating) is seen Short 1969, Moore 1977, and Rising 1983a for in a hybrid zone, it is possiblethat it evolved reviews). Although few of these have been in situ as an adaptive response to selection studied in detail, it is apparent that the flicker againsthybrids. Alternatively, it is possiblethat hybrid zone is not paradigmatic for all avian its evolutionwas incidental to the hybridzone. hybrid zones.For example,the hybrid zone be- In the caseof the sapsuckers,for example,the tween the Blue-winged (Vermivorapinus) and red-breasted are known to return to seasonal Golden-winged (V. chrysoptera)warblers has nestingareas earlier than the red-naped,and moved substantially in historical times (Gill this no doubt contributes to the low level of 1980). Reproductive isolation appears to have interspecificmatings. It is possiblethat selec- evolved in a portion of the Northern Oriole tion hasfavored early and late migration in the (Icterusgalbula) hybrid zone (Corbin and Sibley respectivespecies and this, incidentally,has re- 1977), although major portions of the hybrid suited in a reducedlevel of hybridization in the zone remain unchanged (Rising 1983b). John- limited areasof sympatry.Although it is im- son and Johnson(1985) describednatural hy- portant to recognize these alternatives as a bridization between the Red-breasted(Sphyra- guide to future research,the present data are picusruber daggetti) and Red-naped(S. nuchalis) too few to develop a reasonedinference as to sapsuckers.This is presumablyan old second- the relationshipsamong different kinds of hy- ary contactand hybridization persists,but this brid zones. situation ':sbest describedas a zone of overlap and hybridization (Short 1969) becauseparen- ACKNOWLEDGMENTS tal phenotypespredominate numerically. The We thank JamesHaugland of Sutherland, Nebras- sapsuckers clearly mate assortatively, and ka; Donald Lindgren of Bridgeport, Nebraska; Sam Johnsonand Johnson(1985) argued from in- Masid of Morrill, Nebraska;and Buddy Wilson of direct evidence that hybrids assortedfrom the Wheatland,Wyoming for allowing us to conductthis F•swere selectedagainst. The Northern Flicker researchon their property.We thank Thad Grudzien hybrid zone, in contrast,appears stable, and for critically reading the manuscriptand Lori Mer- there is no evidence of preferential mating lotti for helping with the preparation of the manu- (Moore and Buchanan 1985) or of hybrid break- script.We thank Ned K. Johnson,Peter G. Connors, down. Furthermore, hybridization is rampant, and an anonymousreviewer for many thoughtfuland useful commentsthat were incorporated into the re- one infrequently seesa pure Red- or Yellow- vised manuscript.This researchwas sponsoredby shaftedFlicker in the center of the hybrid zone, NSF grants BSR 81-05195 and BSR 83-20605 to Wil- and the zone coversa large geographicalarea liam S. Moore. Compilation of the museum data was (Short 1965). made possibleby a grant from the Frank M. Chap- A monumental questionis whether one type man Memorial Fund to Waiter D. Koenig. of hybrid zone evolves into another, particu- larly as a result of selectionforces operating LITERATURE CITED within the hybrid zone. This appearsto be the BARROWCLOUGH,G.F. 1980. Geneticand phenotypic case in the Northern Oriole. On the other hand, differentiation in a Wood Warbler (genus Den- the flicker hybrid zone must be at least as old, droica)hybrid zone. Auk 97: 655-668. but evolution of reproductiveisolation has not BARTON,N.H. 1979. The dynamicsof hybrid zones. progressedthere (Moore and Buchanan 1985). Heredity 43: 341-359. Dowling and Moore (1984, 1985) described a , & G. H. HEWITT. 1981. Hybrid zones and zoneof overlapand hybridizationbetween the speciation. Pp. 109-145 in Essayson evolution commonand the striped shiner (Notropis)that and speciationin honor of M. J. D. White (W. R. Atchley and D. S. Woodruff, Eds.).Cambridge, is comparableto the sapsuckerhybrid zone in England, Cambridge Univ. Press. a number of respects.Despite the antiquity of Boche,C.E. 1971. Pairing in hybrid flicker popula- that zone and clear evidence for selection tions in eastern Colorado. Auk 88: 921-924. against hybrids, reproductive isolation be- CORBIN,K. W., & C. G. $IBLEY.1977. Rapid evolution tween the shiners is far from complete.This in orioles in the genus Icterus.Condor 79: 335- raisesthe possibilitythat premating reproduc- 342. January1986] Reproductionin Hybrid Flickers 51

DOWLING, T. E., & W. S. MOOREß 1984. Level of re- BRENNER,& D. H. BENTß 1975. SPSS, statistical productiveisolation between two cyprinid fish- packagefor the socialsciences, 2nd ed. New York, es,Notropis cornutus and N. chrysocephalus.Copeia McGraw-Hillß 1984: 617-628. REMINGTON,C.L. 1968. Suture-zonesof hybrid in- ß & -- 1985. Evidence for selection teractionbetween recently joined biotas.Pp. 321- againsthybrids in the family Cypfinidae (genus 428 in Evolutionarybiology, vol. 2 (TßDobzhan- Notropis).Evolution 39: 152-158. sky, M. K. Hecht, and W. C. Steere, Eds.). New GILL,F. Bß 1980. Historical aspectsof hybridization York, Appleton-Century-Crofts. betweenBlue-winged and Golden-wingedwar- RICKLEESßR. E. 1980. Geographicvariation in clutch biersß Auk 97: 1-18. size among passefine birds: Ashmole'shypoth- JOHNSON,N. K., & C. B. JOHNSONß1985. Speciation esisß Auk 97: 38-49. in sapsuckers(Sphyrapicus): II. Sympatry, hybrid- RISING,J.D. 1983a. The Great Plainshybrid zonesß ization, and mate preferencein S. ruberdaggetti Pp. 131-157 in Current ornithology, vol. 1 (R. F. and S. nuchalis. Auk 102: 1-15. Johnston,Ed.). New York, Plenum Publ. Corpß KOENIG,W. D. 1982. Ecologicaland social factors --. 1983b. The progressof oriole hybridization affectinghatchability of eggs.Auk 99: 526-536ß in Kansas, USA. Auk 100: 885-897. ß 1984. Geographicvariation in clutch size in SHERMANßA. g. 1910. At the sign of the Northern the Northern Flicker (Colaptesauratus): support Flicker. Wilson Bull. 22: 133-171. for Ashmole'shypothesisß Auk 101:698-706ß SHORT,L.L. 1965. Hybridization in the flickers (Co- LEHRMAN, D. S. 1965. Interaction between internal laptes)of North AmericaßBullß Amer. Mus. Nat. and external environmentsin the regulation of Histß 129: 307-428. the reproductivecycle of the Ring Dove.Pp. 355- 1969. Taxonomicaspects of avian hybrid- 380 in Sex and behavior (F. A. BeachßEd.). New izationß Auk 86: 84-105. York, John Wiley & Sons. 1970. A reply to Uzzell and Ashmole.Syst. MOORE,W.S. 1977. An evaluationof narrowhybrid Zool. 19: 199-201. zones in vertebratesß Quartß Rev. Biol. 52: 263- THORNTHWAITEASSOCIATESß 1964. Average climatic 277. water balancedata of the continents,part VII. ß & D. B. BUCHANANß1985. Stability of the Lab. Climatol. Publ. Climatol. 17: 419-615. Northern Flicker hybrid zone in historicaltimes: WILSON,E. O. 1965. The challenge from related implicationsfor adaptivespeciation theory. Evo- speciesßPp. 7-25 in The geneticsof colonizing lution 39: 135-151. species(H. G. Baker and G. L. Stebbins,Eds.). NIE, N.H., C. H. HULL, J. G. JENKINSßK. STEIN- New Yorkß Academic Pressß

The American Ornithologists'Union solicitsnominations for its Brewsterand CouesAwards. Nominations and supportingmaterials should be sentto Dr. Elsie C. Collias,Department of Biology,University of California, Los Angeles,California 90024.Materials must be receivedbefore ! March 1986.

WisconsinProject Loon Watch is acceptingapplications for its firstannual Sigurd T. OlsonCommon Loon ResearchAward for researchon CommonLoons in the LakeSuperior-Lake Michigan region of the United Statesand Canada.To apply, submita brief (maximum10 pages)description of the proposedresearch programand curriculumvitae to Dr. Paul I. V. Strong,Director, Wisconsin Project Loon Watch,Sigurd OlsonEnvironmental Institute, Northland College,Ashland, Wisconsin 54806 USA no later than ! March 1986. Proposalsby studentsshould be accompaniedby two letters of recommendation.The $1,000 award will be grantedon the basisof the project'spotential to betterunderstand and manageUpper Great Lakes populationsof Common Loons. BROOD PARASITISM IN A HOST GENERALIST, THE SHINY COWBIRD: I. THE QUALITY OF DIFFERENT SPECIES AS HOSTS

PAUL MASON 1 Departmentof Zoology,University of Texas,Austin, Texas 78712 USA

ASSTRACT.--TheShiny Cowbird (Molothrusbonariensis) of South America, Panama, and the West Indies is an obligate brood parasiteknown to have used 176 speciesof birds as hosts. This study documentswide variability in the quality of real and potential hostsin terms of responseto eggs, nestling diet, and nest survivorship. The eggs of the parasiteare either spotted or immaculate in eastern Argentina and neighboring parts of Uruguay and Brazil. Most speciesaccept both morphs of cowbird eggs,two reject both morphs, and one (Chalk- browed Mockingbird, Mimus saturninus)rejects immaculate eggs but acceptsspotted ones. No species,via its rejection behavior, protectsthe Shiny Cowbird from competition with a potentialcompetitor, the sympatricScreaming Cowbird (M. rufoaxillaris).Cross-fostering ex- periments and natural-history observationsindicate that nestling cowbirds require a diet composedof animal protein. Becausemost passerinesprovide their nestlingswith suchfood, host selectionis little restricted by diet. Species-specificnest survivorship, adjustedto ap- propriatevalues of Shiny Cowbird life-history variables,varied by over an order of mag- nitude. Shiny Cowbirds peck host eggs.This density-dependentsource of mortality lowers the survivorshipof nestsof preferred hostsand createsnatural selectionfor greater gener- alization. Host quality is sensitive to the natural-history attributes of each host speciesand to the behavior of cowbirds at nests.Received 4 June1984, accepted26 June1985.

VARIATIONin resourcequality can have great parasitized176 species(Friedmann et al. 1977). ecologicaland evolutionary consequences.Ob- The Shiny Cowbird is sympatric with a poten- ligate brood parasites never build nests but tial competitor, the ScreamingCowbird (M. ru- leave the care of their eggsand young to other foaxillaris),in Argentina, Uruguay, and neigh- species,their hosts. The parental behavior of boring parts of South America. This latter hosts is a critical and quantifiable resource to species is extremely specialized on the coop- brood parasites.The first task in understanding eratively nesting Bay-wingedCowbird (M. bad- the use of resourcesis to ascertainthe quality ius), although anecdotal reports of its use of of each alternative. I surveyedthe quality of other hosts exist (Hudson 1874, 1920; Grant various passerinespecies as hosts of the Shiny 1911,1912; Pereyra 1938;Hoy and Ottow 1964). Cowbird (Molothrusbonariensis) in BuenosAires Experiments that simulate natural parasit- Province, Argentina. Host quality is tractable ism provide two kinds of important informa- to analysis becauseselection is spatially and tion. First, the technique identifies individuals temporallyfocused at nests.Dimensions of host that reject parasitic eggs. This response is ap- quality examined include responseto parasitic parently speciestypical for North American eggs,nestling diet, and characteristicsurvivor- birds (Rothstein 1975a, b). Rejecter speciescan ship of eachspecies' nests. This is the only sys- easily be regardedas unsuitable to cowbirdson tematic attempt to characterizethe quality of the basis of this single criterion (Rothstein an array of speciesfor any brood parasite. 1982). Second, if a speciesis an accepter, then The Shiny Cowbird is widely distributed observed parasitism accurately reflects use of throughout South America (Friedmann 1929). that species. If both Shiny Cowbird and It is an extreme host generalist, known to have Screaming Cowbird eggs appear in nests, the two parasitic speciesare potentially in compe- tition. Experiments in artificial parasitism can • Present address: Department of Biological Sci- reveal if certain species reject eggs of one ences,University of California, Santa Barbara,Cali- speciesbut acceptthose of the other. fornia 93106 USA. Categorizationof speciesas acceptersor re-

52 The Auk 103:52-60. January1986 January1986] Qualityof Shiny Cowbird Hosts 53 jecters is complicatedby extreme variation in as acceptedif it remained in the nest 5 days or more Shiny Cowbird eggs. Some speciesmay reject (seeRothstein 1975a). Large series of experimentswith one egg type but accept many others. Despite North American bird speciesindicate that almost all the unusual level of variation, two broad classes rejectionsoccur within this period (Rothstein 1975a, Mason unpubl. data). of eggsare recognized:spotted and immaculate Specieswere categorized as acceptersor rejecters (Friedmann 1929, Fraga 1978, Gochfeld 1979). for specificegg types if two-thirds or more of the More importantly, I found no evidence that responseswere consistent.Usually, responsesare birds discriminated against eggs within each more consistent, but Rothstein (1975b) used this category,and this is the primary justification truncation criterion to show that species'responses for the dichotomous categorization (see Re- were distributed bimodally. Specieswere not cate- suits). Screaming Cowbird eggs (described in gorizedif the frequencyof rejectionfell betweenone- detail by Fraga 1983a)are spotted,but the mac- third and two-thirds. Species were tentatively re- ulations are less distinct. Most also possess garded as acceptersof any egg morph with which miniscule scrawls. they were not testedif (1) the speciesclearly accepted Nestling diet is anothercritical aspect of host cowbirdeggs widely divergentfrom its own eggs;or (2) the untested morph closely resembled its own. quality. Cross-fostering,the introduction of For example,a specieslaying an immaculatewhite parasiticyoung into nests,is a difficult but in- egg and shown experimentallyto acceptspotted eggs formative technique capable of assessinghost would be tentatively categorized as an accepter of dietaryquality. Failing this,descriptions of food immaculatecowbird eggs. items brought to nestsmay be adequate. To assessdietary quality and responseto nestlings, Nests are subject to total failure for reasons I augmentedobservations of nestling development such as predation and severe weather. Survi- with cross-fosteringexperiments. If a nestling cow- vorship estimatesare appropriate measuresof bird remained healthy on subsequentvisual inspec- this aspectof host quality. tions, dietary quality was scoredas acceptable.Un- healthybirds were recognizedby any of the following symptoms:pallor, emaciation,loss of thermoregula- METHODS tory ability, or moribund passivity.Ideally, cross- fosteredcowbirds should be larger than host young Observationswere carriedout almostdaily for two to eliminate the possibilitythat subsequentstarva- seasons(1977-1978 and 1978-1979) at two sites 17.6 tion of the cowbird is the result of a competitive dis- km apart near Magdalena, Buenos Aires Province, advantageand not of dietary quality. Argentina. Descriptionsof the study sites, the avi- Host specieswere scoredas acceptablewith respect fauna,and my generalfield techniquesappear else- to diet if cross-fosterednestlings fledged. If nestlings where (Mason 1985). The habitat consistedof pas- appearedhealthy before nest failure (occurring for ture, with clumps of trees and marsh. I searched reasonsunrelated to parasitism),the host was scored continuallyfor neststo carry out manipulationsand as tentatively acceptable. make observations. Of course, nests that were diffi- Nest survivorship can estimate the parasite's ex- cult to find or examine are underrepresented in the pectation of success,provided the host fulfills the sample. other requirementsfor acceptability.I measuredthe The technique of artificial parasitism followed probabilityof survivalfor the egg and nestlingphas- Rothstein (1971), except that I occasionallyplaced 2 es, consideringonly those sourcesof mortality that eggsin a single nest. The experimentsused real and affected all nest occupantssimilarly and simulta- artificial eggsof 3 experimentalmorphs: spotted M. neously.Survivorship of eachinterval wascalculated bonariensis,white M. bonariensis,and M. rufoaxillaris. by raisingdaily survivorshipto a power equal to the Dimensionsand weights of real and artificial eggs length of that interval in days(Mayfield 1975,Hens- can be found in Mason (1980). Experimental parasit- let and Nichols 1981). Because cowbirds differ from ism was performed as early as possible,and always most of their hosts in length of the incubation and before 1200. This correspondsto the natural laying nestlingperiods, I usedthe appropriatevalues of these behavior of Shiny Cowbird females(Hoy and Ottow Shiny Cowbird life-history variables (11.9 days and 1964). 13.9 days, respectively;Mason unpubl. data) to find There are two mutually exclusive responsesto the specificvalue of the host to cowbirds.The length parasiticeggs: acceptance for incubation,or rejection. of the egg phase includesthe laying period of the Rejection includes any behavior of the host species host as well as the incubationperiod of the cowbird that doesnot result in successfulincubation: ejection egg. Survivorshipto fledging is the product of sur- of the parasiticegg, nest desertion,or burial of the vivorship during the egg and the nestling phases. parasiticegg. I scoredan experimentallyplaced egg The survivorship estimatesreflect freedom of the nest 54 PAULMASON [Auk, Vol. 103

TABLE1. Passefineresponse to cowbird eggs. Experimental eggs were of 3 morphs: immaculate Shiny Cowbird (Shiny, I), spottedShiny Cowbird (Shiny, S), and ScreamingCowbird (Screaming).

Responseto experimental egg morphsa Shiny, I Shiny, S Screaming Host speciesb R/T (Status) R/T (Status) R/T (Status) RufousHornero, Furnariusrufus (I) 4/5 (Rej) 5/6 (Rej) 4/4 (Rej) Wren-like Rushbird,Phleocryptes melanops (I) 0/5 (Acc) 0/7 (Acc) 0/4 (Acc) Tufted Tit-Spinetail,Leptasthenura platensis (I) 0/1 (Acc) 0/4 (Acc) 0/2 (Acc) Little Thornbird, Phacellodomussibilatrix (I) -- (Acc?)c 0/1 (Acc) -- (Acc?) Freckle-breasted Thornbird, Phacellodomusstriaticollis (I) -- (Acc?) 0/1 (Acc) 0/2 (Acc) Firewood-gatherer, Anumbiusannumbi (I) -- (Acc?) 0/6 (Acc) 0/4 (Acc) Vermillion Flycatcher,Pryocephalus rubinus (S) 0/6 (Acc) 0/6 (Acc) 0/3 (Acc) Yellow-browed Tyrant, Satrapaicterophrys (S) l*/3a (Acc) 0/2 (Acc) 0/1 (Acc) Cattle Tyrant, Machetornisrixosus (S) 0/2(Acc) -- (Acc?) -- (Acc?) Fork-tailed Flycatcher,Tyrannus savana (S) 3/3 (Rej) 2/2 (Rej) -- (Rej?) Great Kiskadee,Pitangus sulphuratus (S) 0/4(Acc) 1/2 ? 1/2 ? White-crestedTyrannulet, Serpophagasubcristata (I) -- (Acc?) 0/1 (Acc) 0/1 (Acc) House Wren, Troglodytesaedon (S) 0/3 (Acc) -- (Acc?) -- (Acc?) White-rumped Swallow, Tachycinetaleucorrhoa (I) -- (Acc?) 0/3 (Acc) 0/1 (Acc) Brown-chestedMartin, Phaeoprognetapera (I) -- (Acc?) 0/1 (Acc) -- (Acc?) Chalk-browed Mockingbird, Mimus saturninus(S) 8'/11 (Rej) 2'/10 (Acc) 6/12 ? Creamy-belliedThrush, Turdusamaurochalinus (S) 0/1 (Acc) -- (Acc?) -- (Acc?) Masked Gnatcatcher,Polioptila dumicola (S) 0/1 (Acc) -- (Acc?) -- (Acc?) Bay-winged Cowbird, Molothrusbadius (S) 0/6 (Acc) -- (Acc?) 0/1 (Acc) Blue-and-yellowTanager, Thraupis bonariensis (S) 0/1 (Acc) -- (Acc?) 0/1 (Acc) Saffron Finch, Sicalisfiaveola (S) 0/4 (Acc) 0/8 (^cc) 2**/8 (Acc) Grassland Yellow-Finch, Sicalis luteola (S) 0/2 (Acc) -- (Acc?) -- (Acc?) GrasslandSparrow, Ammodramushumeralis (S) 0/1 (Acc) -- (Acc?) -- (Acc?) Rufous-collaredSparrow, Zonotrichiacapensis (S) 2**/6 (Acc) 1'/5 (Acc) 0/7 (Acc) Hooded Siskin, Carduelismagellanica (S or I) -- ? 0/1 (Acc) 0/1 (Acc) House Sparrow, Passerdomesticus (S) 0/1 (Acc) 0/1 (Acc) -- (Acc?) ' R/T representsthe numberof rejectionsout of the numberof trials.Acc = accepts,Rej = rejects. bCommon names according to Meyer de Schauensee(1970). The letter in parenthesesfollowing each speciesname roughly categorizesthe eggsof that speciesas spotted(S) or immaculate(I). More detailed descriptionsof eggsare in Mason (1985). cSpecies that acceptedeggs widely divergentfrom their own, but untestedfor morphsmore similarto their own, are tentatively scoredas accepters,as indicatedby a questionmark. a Each asterisk indicates a case of desertion. from completeloss, but do not include events that Two species,the Chalk-browed Mockingbird causedifferential mortality within the nest. and the Great Kiskadee, showed intermediate Scientificnames of speciesare given in Table 1. levelsof ejectionfor certainmorphs. In the case of the kiskadee,the problem may be one of RESULTS smallsample sizes (4 experimentswith spotted eggs),but this is not true for the mockingbird, Responsesto eggs.--I performed 187 experi- which rejectedScreaming Cowbird eggs from mental egg manipulationsand scored 42 re- 6 of 12 nests. I have not assignedrejecter or sponsesas rejections.Species could almost al- accepterstatus to either species. ways be categorizedas acceptersor rejecters, Eight rejectionsoccurred by desertionand althoughonly for a specificegg morph in the none by egg burial. Unlike ejection,desertion caseof the Chalk-browed Mockingbird (Table was never expressedconsistently. No species 1). Speciesclassified as rejecters of all eggs(Ru- could be classifiedas a rejecteron the basisof fous Hornero, Fork-tailed Flycatcher)may be desertion. referredto asdual rejecters,and thoseclassified No evidence suggeststhat response to arti- asaccepters of all asdual accepters.The Chalk- ficialeggs differs from responseto naturaleggs. browed Mockingbird is a differential accepter, To use the Chalk-browed Mockingbird as an favoringthe spottedShiny Cowbird egg morph. example,responses to real and artificial eggs January1986] Qualityof ShinyCowbird Hosts 55

TABLE2. Resultsof cross-fosteringexperiments. Number of experimentsfor each host speciesis indicated in parentheses.See text for discussion.

Host species Results Rufous Hornero (1) 4-day-old nestling lived 4 more days before starvation, losing in compe- tition with much bigger host nestlings Firewood-gatherer(1) Lived 3-4 days before death of all nestlingsdue to botfly parasitism Vetmillion Flycatcher(1) Lived 2-3 days until predation Yellow-browed Tyrant (1) Lived 5 days until predation Great Kiskadee (1) Nestling of several days fledged after 9 more days in nest Chalk-browed Mockingbird (2) Nestling at first nest lived 4-5 days until predation at nest; nestling at secondnest lived 6-7 days until predation at nest Creamy-belliedThrush (1) Fledgedafter 14-15 daysin nest,but at very low weight, having hatched 4 days after host nestling Saffron Finch (2) Nestling Shiny Cowbird lived 2 days until death of all nestlings due to botfly parasitism;nestling ScreamingCowbird failed within 2 days House Sparrow (1) Failed within 2 days

were statisticallyindistinguishable (P > 0.5 for becausehost young had a substantial size ad- all comparisonsamong immaculate and spotted vantage (Table 2). The hornero has been known Shiny Cowbird eggs and ScreamingCowbird to rear cowbirds in other areas of South Amer- eggs;Fisher's exact test, Siegel 1956). Further- ica (Friedmann 1929, Friedmann et al. 1977, more, the assignmentof speciesas acceptersor Salvador 1983). A secondthrush nest was found rejecters(Table 1) agreeswith Fraga(1978, 1980, with 2 cowbirds about to fledge. The experi- 1983b,1985) for 11 commonspecies, including mental nestling (in the thrush nest) weighed all rejecters.Observations of natural parasitism 22.0 g at fledging, while the 2 nestlingsat the agree (Mason 1986). other nest weighed 43.0 and 45.5 g. Lack of competitionfrom the ScreamingCow- Dietary suitability (Table 3) was judged by 3 bird.--In all but one case (Chalk-browed Mock- criteria: speciesreported to successfullyrear ingbird), birds responded to ScreamingCow- naturally placed eggs (Friedmann 1929, 1963; bird eggsas they did to spottedShiny Cowbird Friedmannet al. 1977;Fraga 1978;this study), eggs.No host speciestotally protectsthe Shiny speciesthat feed young animal protein (Mason Cowbird from competitionwith the Screaming 1985),and the resultsof the cross-fosteringex- Cowbird via its rejection behavior. Since periments. ScreamingCowbird eggs were found only in Nest survivorship.--Completedata on daily nestsof the Bay-wingedCowbird (Mason 1980), survivorshipare in Mason (1985). Daily survi- host selection in the Shiny Cowbird is unaf- vorship for the egg phase ranged from 0.988 fectedby nestling competitionwith the former (RufousHornero, n = 15 nests,244 days of ob- speciesin all other nests. servation) to 0.852 (White-crested Tyrannulet, Nestlingdiet.--I performed 11 cross-fostering n = 4 nests,27 days of observation).Daily sur- experiments (Table 2). Only 2 species were vivorship for the nestling phase ranged from scored as having unacceptablediets: the Saf- 0.996 (RufousHornero, n = 13 nests,285 days fron Finch and the House Sparrow. Observa- of observation)to 0.857(Fork-tailed Flycatcher, tions of natural parasitismsupport the experi- n = 4 nests, 28 days of observation). mentsin the caseof the finch: 2 Shiny Cowbirds The hornero'ssuperior survivorship for both in separatenests both failed to survive 2 days. phasesof the nestingcycle is largely attribut- No fledgingrecords exist for either species.The able to its remarkable domed mud nest, which remaining species were judged as acceptable, contributes to superior survivorship in other and observationssupport this interpretation. birds that also use it (Mason 1985). The differ- Differential mortality of the cowbird nestling ence between survivorship in hornero and occurred at 1 nest of the Rufous Hornero, and mockingbird nests was significant for both 1 nest of the Creamy-bellied Thrush. Both phasesof the nestingcycle (Table 4). Mocking- deathswere attributed to intranestcompetition bird survivorship exceededthat of the Rufous- 56 PAULMASON [Auk, Vol. 103

TABLE3. Dietary suitability of different host species.Descriptions of food items can be found in Mason (1985). Cross-fosteringexperiments are discussedin Table 2 and text.

Natural Successful successful Animal protein cross- Host species rearing? provided? fostering? Rufous Hornero Yes a Yes Yes? Wren-like Rushbird Yes Freckle-breasted Thornbird Yes Firewood-gatherer Yesa Yes Vermillion Flycatcher Yes Yes Yellow-browed Tyrant Yesb Yes Yes Cattle Tyrant Yesa Fork-tailed Flycatcher Yesa Tropical Kingbird Yesa (Tyrannusmelancholicus) Great Kiskadee Yes Yes White-rumped Swallow Yesb House Wren Yes •,b Chalk-browed Mockingbird Yesa.b Yes Rufous-bellied Thrush Yes •.b (Turdusrufiventris) Creamy-bellied Thrush Yesb Yes? Masked Gnatcatcher Yes a Bay-wingedCowbird Yes• Yes Yellow-wingedBlackbird Yesb (Agelaiusthilius) Blue-and-yellowTanager Yes• Saffron Finch No b No No? Grassland Yellow-Finch No • No Rufous-collaredSparrow Yes•,b Yes Hooded Siskin Yes ' House Sparrow No• Variable No Recordsfrom literature (Friedmann 1929, 1963;Friedmann et al. 1977; Salvador 1983;Fraga 1985). Observationsof this study. collared Sparrow only for the nestling phase rumped Swallow) tended Shiny Cowbird when all sourcesof mortality were considered. fledgings more frequently than any other However, 13 mockingbird nest failures were species(Table 6; Mason 1986). attributedto peckingby Shiny Cowbirds,while only 1 suchloss occurred at sparrownests, and DISCUSSION none occurred at nests of the hornero. When this sourceof mortality was removed,the daily Responseto eggs.--The 26 speciessurveyed survivorship of mockingbird nests improved responded to 3 particular morphs of cowbird significantly,surpassing that of the sparrowbut eggs in a species-typicalfashion, with only 2 remaining inferior to that of the hornero. minor exceptions.This agreeswith data from The productof the survivorshipprobabilities North America (Rothstein 1975a, b). In further for the egg and nestling phasesrepresents the agreement,rejection is almost always accom- probabilityof a nestfledging at leastone young plished by ejection of the parasitic egg. Two Shiny Cowbirdunder ideal conditions.The ac- specieswere classifiedas dual rejecters(Rufous cepter speciesdiffered by almost an order of Hornero, Fork-tailed Flycatcher) and one as a magnitude;when rejecterswere considered,the differentialaccepter favoring the spottedmorph differencewas greater still (Table 5). Rejecters (Chalk-browed Mockingbird). Speciesthat re- are included to show the success a female could ject cowbird eggsare obviouslyunsuitable as experiencewere she to lay a mimetic egg. The hosts. RufousHornero, in the absenceof rejectionbe- Desertionwas infrequent, and egg burial was havior, would be the best host choice. The best never observed. This also resembles most North accepterspecies judged by survivorship(White- American experiments in artificial parasitism January1986] Qualityof Shiny Cowbird Hosts 57

T^BLE4. Comparisonsof daily mean survivorshiprates. Numbers are presentedas daily survivorship(SD, no. of nestssurveyed, no. of observationdays).

Rufous Hornero vs. Chalk-browed Mockingbird Eggs 0.988 (0.007, 15, 244) 0.922 (0.012, 59, 477) 4.75 0.999+ <0.001, 1-tailed Eggs' 0.988 (0.007, 15, 244) 0.950 (0.010, 59, 477) 3.11 0.999 =0.001, 1-tailed Nestlings 0.996 (0.004, 13, 285) 0.953 (0.014, 28, 235) 2.95 0.998 =0.002, 1-tailed Chalk-browed Mockingbird vs. Chalk-browed Mockingbird' Eggs 0.922 (0.012, 59, 477) 0.950 (0.010, 59, 477) 1.79 0.963 =0.037, 1-tailed Chalk-browed Mockingbird vs. Rufous-collaredSparrow Eggs 0.922 (0.012, 59, 477) 0.899 (0.020, 45, 227) 0.99 0.839 =0.161, NS Eggs' 0.950 (0.010, 59, 477) 0.903 (0.020, 45, 227) 2.10 0.982 =0.018, 1-tailed Nestlings 0.953 (0.014, 28, 235) 0.901 (0.024, 26, 161) 1.87 0.969 =0.031, 1-tailed Nest failuresdue to peckingof eggsby Shiny Cowbirdseliminated from calculationof survivorship.

(Rothstein 1975a, b; but see Clark and Robert- birds (Postand Wiley 1976, 1977;Friedmann et son 1981).Desertion may be a responseto hu- al. 1977). man presence or some disturbance at the nest Only 3 species(Saffron Finch, GrasslandYel- other than the parasiticegg (Rothstein 1975a, low-Finch, Hooded Siskin) at my study sites 1976),but my techniquescould not identifythis are seed specialists.Neither finch is known to possibility. have reared cowbirds to fledging (Friedmann The speciesmost likely to desertwas the Ru- et al. 1977). Shiny Cowbird deaths in nestsof fous-collaredSparrow, which rejectedonly 3 of the Grassland Yellow-Finch have been attrib- 18 artificially placed cowbird eggs,all by de- uted to diet (Salvador1983, Fraga 1985). Other sertion. Fraga (1978) also found desertionto be seed specialists(Carduelis spp.) may augment infrequent. Hudson (1920) claimed that deser- the nestling diet with aphids, although reports tion was the typical responseto parasitism in are variable (Bent and collaborators 1968, the Vetmillion Flycatcher, but I observed ac- Friedmann et al. 1977). Friedmann et al. (1977: ceptancesin all 19 experiments. 43) and Middleton (1977) observed Brown- Cowbird eggs incubated in nests of differ- headed Cowbirds to die in goldfinch nests. ential acceptersfulfill the biological criterion The failure of a cross-fostered cowbird to sur- for mimicry. Their resemblanceto host eggsis vive in a nest of the House Sparrow is consis- adaptive and subjectto selection.Human stan- tent with observationsby Salvador(1983) and dardsof similarity are misleadingand inappro- the lack of fledging records(Friedmann et al. priate. For example, Chalk-browed Mocking- 1977). Because seeds often form a substantial bird eggsare easilydistinguished (by humans) portion of the nestling diet, failures are prob- from spotted Shiny Cowbird eggs that are ably due to dietary restriction rather than to readily accepted.Three additional speciesare discrimination against young cowbirds (Bar- suspectedof being differential accepters(Fraga rows 1889, Seel 1969, Anderson 1978, Eastzer 1985). et al. 1980). Nestlingdiet,--Cowbirds appear to require Nest survivorship.--Onesource of nest failure animal protein. Becausemost passetinesfeed waspecking by Shiny Cowbirds.Nests of high- their young arthropods(Hamilton and Orians ly preferred hostsare often multiply parasit- 1965,Skutch 1976) and measuresof overlap in ized by several females (Fraga 1985; Mason nestlingdiets of sympatricspecies are typically 1980,1986) and consequentlysubject to higher high (Orians and Horn 1969, Anderson 1978, levels of pecking. This behavior depressedthe Maher 1979),cowbirds are probablylittle re- successof Chalk-browed Mockingbird neststo strictedin hostchoice by factorsassociated with a level not significantlydifferent from that of nestlingnutrition. This claim is supportedby the Rufous-collaredSparrow (Table 4). Pecking the large and taxonomically diverse list of is sometimesless drastic, and cowbird eggs are species known to have reared Brown-headed occasionallylost during otherwise successful (M. ater),Bronzed (M. aeneus),and Shiny cow- nesting attempts. Nonetheless, lossesto peck- 58 PAULMASON [Auk, Vol. 103

TABLE5. Survivorshipestimates of Shiny Cowbirdsin nestsof different hostspecies. Species with unsuitable nestlingdiets are not included.

Survivorship Eggs Nestlings Overalla Rufous Hornero b 0.823 (15, 244)c 0.952 (13, 285) 0.783 Wren-like Rushbird 0.596 (20, 271) 0.715 (11,126) 0.426 Tufted Tit-Spinetail 0.306(4, 25) 1.000(2, 31) 0.306 Freckle-breastedThornbird 0.211 (7, 60) 0.746 (4, 48) 0.157 Firewood-gatherer 0.607(11,120) 0.671(8, 106) 0.407 Vermillion Flycatcher 0.403(22, 151) 0.405(14, 143) 0.163 Yellow-browedTyrant 0.713(10, 95) 0.486(8, 79) 0.347 Fork-tailedFlycatcher b 0.549(7, 72) 0.117(4, 28) 0.064 Great Kiskadeea 0.489(7, 106) 0.481(3, 39) 0.235 White-rumpedSwallow 0.584(9, 89) 0.819(5, 70) 0.478 House Wren 0.273 (9, 44) 0.624 (3, 41) 0.170 Chalk-browedMockingbird' 0.326(59, 477) 0.514(28, 235) 0.168 Rufous-bellied Thrush 0.370 (3, 30) 0.408 (2, 16) 0.151 Bay-wingedCowbird 0.190(10, 82) 1.000(2, 20) 0.190 Rufous-collaredSparrow 0.227(45, 227) 0.233(26, 161) 0.053 Overall survivorship= (egg survivorship)x (nestling survivorship). Rejectscowbird eggs. Mean survivorship(no. of nestssurveyed, no. of observationdays). Responseto spottedeggs uncertain (Mason 1986). Rejectsimmaculate cowbird eggs.

ing are density-dependentforms of mortality Chalk-browed Mockingbird's larger size at becausethey are inflicted by cowbirds hatching and faster growth rate (Fraga 1985, The survivorshipprobabilities (Table 5) rep- Mason 1985). I observedonly 1 caseof starva- resentthe freedomof nestsfrom completeloss, tion among16 cowbirdnestlings in 10 broods, and estimate the success of cowbirds under ide- but Fraga (1985) reported 12 such casesamong al circumstances.Differential mortality of cow- 20 nestlingsfrom 15 nests.The reasonsfor in- birds within a nest will lower success below creasedstarvation at Fraga'ssite are unknown. rated survivorship. Competition among nest- Competition between host and parasite is lings for the limited food available can intro- unlikely to adversely affect cowbirds in nests duce differential mortality asin adaptivebrood of the Rufous-collaredSparrow. Nestling spar- reduction(Lack 1954):competition is size and rows are smaller at hatching and grow more age mediated, and inequalities in competitive slowly (King 1973, Fraga 1978, Mason 1985). ability are introduced by the amount of food Fraga(1985) found that the overall successof supplied,size at hatching,magnitude of hatch- naturally laid cowbird eggsin Rufous-collared ing asynchrony,and normal nestling growth Sparrownests was 0.056(vs. 0.053in Table 4). rate. On the other hand, competitionis likely if Successfulparasitism requires that female sparrownests are multiply parasitized.The nu- cowbirds properly synchronizetheir laying tritional demandsof two cowbird young prob- with that of the host.Eggs laid late may fail to ably exceedthe capabilitiesof that hostto rear hatch,or hatchso late that the nestlingwill be two healthy cowbirds.Fraga (1978, 1983b,1985) competitively inferior. The extremely low reported no casesof sparrows fledging two fledgingweight of the cowbirdnestling in the cowbirds, while King (1973) showed that nest of the Creamy-bellied Thrush and the growth rate and fledgingweight were both re- death of the cowbird nestling cross-fosteredin ducedwhen two cowbirdnestlings were in the a nest of the Rufous Hornero are illustrative nest. In the region of South America studied, (Table 2). multiple parasitismis almostentirely restricted The life-historycharacteristics of somespecies to large, highly preferredhosts; smaller hosts suggestthat severe nestling competitionmay (like the sparrow)are slightly used (Mason be difficult to avoid. The advantageof the cow- 1986). bird's shorterincubation period is offsetby the Friedmann (1963) pointed out that the suc- January1986] Qualityof ShinyCowbird Hosts 59

TABLE6. Observations of host species attending A. Landa (Estancia E1 Talar). The staff of the Museo Shiny Cowbird fledglings. Argentino de CienciasNaturales gave me consider- able help and encouragement.In particular,I thank Host species Date of observation J. M. Gallardo (Director), J. R. Navas (Jefe, Divisi6n White-rumpedSwallow (7 nests) de Ornitologia),and M. Canevari.R. M. Fragaand S. 1 14 Nov 1978, 18 Nov 1978 I. Rothstein have been sources of intellectual stimu- 2 24 Nov 1978, 30 Nov 1978 lation and encouragement.They, L. F. Kiff, V. Nolan, 3 27 Nov 1978, 28 Nov 1978 Jr., R. Robertson,S.S. Sweet,and an anonymousre- 4 8 Dec 1978, 12 Dec 1978 viewer kindly read and commentedon earlier ver- 5 9 Dec 1978, 14 Dec 1978 sionsof the manuscript.Financial support was in part 6 19 Dec 1978 provided by the Frank M. Chapman Fund of the 7 23 Dec 1978 American Museum of Natural History, New York. I House Wren 22 Nov 1978 was supported by NSF grant BSR 8214999 to S. I. Yellow-winged 12 Jan 1979 Rothstein during the preparation of the final draft of Blackbird Masked Gnatcatcher 2 Feb 1979 this paper. LITERATURE CITED

ANDERSON,T. R. 1978. Populationstudies of Euro- cessof cowbirds is characteristicallyless than pean sparrowsin North America. Occ. Pap. Mus. that of their hosts. Three important sourcesof Nat. Hist. Kansas 70: 1-58. mortality (diet, density-dependent egg peck- BARROWS,W. B. 1889. The English Sparrow (Passer ing, and the adverse affect of intranest com- domesticus)in North America, especiallyin re- lationsto agriculture.U.S. Dept. Agr. Bull. 1: 1- petition) are probablythe main reasonsfor this 405. observation. BENT, A. C., & COLLABORATORS.1968. Life histories Host quality is sensitive to the natural-his- of North American cardinals, grosbeaks,bunt- tory attributes of nesting speciesand to cow- ings,towhees, finches, sparrows, and allies, part bird behavior. Some adaptations of cowbirds 1 (O. L. Austin, Ed.). New York, Dover Publ. can circumvent apparent barriers to successful CLARK,K. L., & R. J. ROBERTSON.1981. Cowbird par- parasitism. Egg rejection can be overcome by asitismand the evolution of anti-parasitestrat- mimicry, ashas been documentedfor the Chalk- egiesin the Yellow Warbler. Wilson Bull. 93: 249- browed Mockingbird (Fraga 1985, this study). 258. The Rufous Hornero is apparently a frequent EASTZER,D., P. R. CHU, • A. P. KING. 1980. The host in other parts of Argentina (Friedmann young cowbird: average or optimal nestling? Condor 82: 417-425. 1929, Friedmann et al. 1977, Salvador 1983), de- FRAGA,R. M. 1978. The Rufous-collaredSparrow as spite its status as a rejecter in Buenos Aires a host of the Shiny Cowbird. Wilson Bull. 90: Province. In the caseof large hosts,the shorter 271-284. the host incubation period, the more likely 1980. The breeding of Rufous Horneros cowbird nestlings will experience severe com- (Furnariusrufus). Condor 82: 58-68. petition. Egg pecking may reduce nestling 1983a. The eggsof the parasiticScreaming competitionby selectivelydestroying host eggs, Cowbird(Molothrus rufoaxillaris) and its host,the whose numbers vary inversely with those of Bay-winged Cowbird (M. badius):is there evi- the cowbird in Chalk-browed Mockingbird dencefor mimicry?J. Ornithol. 124: 187-193. nests(Mason 1980, Fraga 1985). However, en- 1983b. Parasitismode cria del renegrido (Molothrusbonariensis) sobre el chingolo (Zono- tire clutchesare often destroyed, thus increas- trichiacapensis): nuevas observaciones y conclu- ing the relative value of other species.A gen- siones. Hornero 12, No. Extraord.: 245-255. eralized pattern of resourceuse results from 1985. Host-parasite interactions between the Shiny Cowbird's ability to exploit several Chalk-browed Mockingbirds and Shiny Cow- avenues of adaptive niche expansion. birds.Pp. 829-844in Neotropicalornithology (P. A. Buckley, M. S. Foster, E. S. Morton, R. S. ACKNOWLEDGMENTS Ridgely, and F. G. Buckley, Eds.). Ornithol. Monogr. No. 36. My fieldwork in Argentina was made possibleby FRIEDMANN,H. 1929. The cowbirds:a study in the the kind help and cooperationof a numberof people biology of social parasitism. Baltimore, C. C. and agencies.M. Christie first took me into the field Thomas. in Argentina. I spent my first seasonwith the Earn- ß 1963. Host relations of the parasiticcow- shawfamily (EstanciaSan Isidro), and my secondwith birds. U.S. Natl. Mus. Bull. 233: 1-276. 60 PAUUMASON [Auk,Vol. 103

, L. F. KIFF, & S. I. ROTHSTEIN. 1977. A further MAYFIELD,H. F. 1975. Suggestionsfor calculating contribution to knowledge of the host relations nesting success.Wilson Bull. 78: 456-466. of the parasitic cowbirds. Smithsonian Contrib. MEYERDE SCHAUENSEE, g. 1970. A guide to the birds Zool. 235: 1-75. of South America. Wynnewood, Pennsylvania, GOCHFELD,M. 1979. Broodparasite and host coevo- Livingston Publ. Co. lution:interactions between Shiny Cowbirdsand MIDDLETON,A. L.A. 1977. Effect of cowbird para- two speciesof meadowlarks. Amer. Natur. 113: sitism on American Goldfinch nesting. Auk 94: 865-870. 304-307. GRANT, C. H. [5. 1911. List of birds collected in Ar- ORIANS,G. H., •t H. S. HORNß1969. Overlap in foods gentina, Paraguay,Bolivia, and southern Brazil, and foragingof four speciesof blackbirdsin the with field notes, Part I, Passeres.Ibis 5, 9th Set.: potholesof centralWashington. Ecology 50: 930- 104-131. 938. ß 1912. Notes on some South American birds. PEREYRA,J.A. 1938. Aves de la zona tiberefta nor- Ibis 1, 10th Set.: 273-280. deste de la Provincia de Buenos Aires. Mem. Jar- HAMILTON, W. J., III,& G. H. ORIANS. 1965. Evolu- din Zool. La Plata: 1-304. tion of brood parasitism in altricial birds. Con- POST,W., & J. W. WILEYß 1976. The Yellow-shoul- dor 67: 361-382. deredBlackbird--present and future. Amer. Birds HENSLER,G. L., •r J. D. NICHOLSß1981. The Mayfield 30: 13-20. method of estimating nesting success:a model, ß 1977. Reproductiveinteractions of the Shiny estimators, and simulation results. Wilson Bull. Cowbird and the Yellow-shouldered Blackbird. 93: 42-53. Condor 79: 176-184. HOY, G., •t J. OTTOW. 1964. Biologicaland oological ROTHSTEIN,S. I. 1971. Observationand experiment studies of the molothrine cowbirds (Icteridae) of in the analysisof interactionsbetween brood Argentina. Auk 81: 186-203. parasitesand their hosts.Amer. Natur. 105: 71- HUDSON,W. H. 1874. Notes on the procreant in- 74. stinctsof the three speciesof Molothrusfound in ß 1975a. An experimental and teleonomic in- BuenosAyres. Proc. Zool. Soc.London: 153-174. vestigationof avian brood parasitism.Condor 77: ß 1920. Birds of La Plata, vol. 1. New York, E. 250-271. P. Dutton. --. 1975b. Evolutionaryrates and host defenses KING,J. R. 1973. Reproductiverelationships of the againstavian brood parasitism.Amer. Natur. 109: Rufous-collaredSparrow and the Shiny Cow- 161-176. bird. Auk 90: 19-34. 1976. Experiments on defenses Cedar LACK,D. 1954. The natural regulation of animal Waxwings use against cowbird parasitism.Auk numbers. Oxford, Clarendon Press. 93: 675-691. MAHER,W.J. 1979. Nestling diets of prairie passer- 1982. Successes and failures in avian and ine birds at Matador, Saskatchewan, Canada. Ibis nestlingrecognition with commentson theutility 121: 437-452. of optimality reasoning.Amer. Zool. 21: 547-560. MASON,P. 1980. Ecologicaland evolutionary as- SALVADOR,S. A. 1983. Parasitismo de cria del re- pectsof host selectionin cowbirds.Unpublished negrido (Molothrusbonariensis) en Villa Maria, Ph.D. dissertation, Austin, Texas, Univ. Texas. C6rdoba, Argentina (Aves: Icteridae). Hist. Nat. ß 1985. The nesting biology of somepasser- 3: 149-158. ines of BuenosAires, Argentina. Pp. 954-972 in SEEL,D.C. 1969. Food, feeding ratesand body tem- Neotropical ornithology (P. A. Buckley, M. S. perature in the nestling House Sparrow Passer Foster, E. S. Morton, R. S. Ridgely, and F. G. domesticus at Oxford. Ibis 111: 36-47. Buckley, Eds.). Ornithol. Monogr. No. 36. SIEGEL,S. 1956. Nonparametricstatistics for the be- ß 1986. Broodparasitism in a host generalist, havioral sciences. New York, McGraw-Hill. the Shiny Cowbird: II. Host selection.Auk 103: SKUTCH,A. F. 1976. Parent birds and their young. 61-69. Austin, Texas, Univ. Texas Press. BROOD PARASITISM IN A HOST GENERALIST, THE SHINY COWBIRD: II. HOST SELECTION

]•AUL MASON • Departmentof Zoology,University of Texas,Austin, Texas 78712 USA

AI•STI•ACT.--Hostselection by brood parasiticShiny Cowbirds (Molothrusbonariensis) was studiedat two sitesin BuenosAires Province,Argentina. The eggsof the parasiteare either spottedor immaculate,so host selectionwas studiedwith respectto egg type as well as to site. Immaculate eggs were rare at both sites. Cowbirds in this region prefer to parasitize nestsof birds larger than themselves.This preferencecontrasts with that of almostall other brood parasites,and even that of Shiny Cowbirds in other parts of the species'range. One large species,the Chalk-browedMockingbird (Mimussaturninus) was used frequently and consistentlyat both sites.At least two other large species(thrushes) were used at Site I, but neither was present at Site II. Cowbirds were more specializedon large hostsat Site I: only 9% of the spottedeggs were laid in nestsof small birds at Site I, whereas 35% were laid in suchnests at Site II. Small species,such as the Rufous-collaredSparrow (Zonotrichiacapensis), were usedto a greaterextent at Site II than Site I (where someof the samespecies were not usedat all). The shift to smallerhosts is probablya responseto a changein the structureof the community;large host speciesare relatively lessabundant at Site II. Rejecterspecies are large, and all were parasitizedwhile many smaller accepterspecies were unmolested.Surprisingly, more immaculateeggs were laid in nestswhere they had little chanceof successfulincubation, rather than in nestsof accepters.No evidencesuggests that host races("gentes") are formed. To the contrary, female cowbirdslaying different egg types apparently selecthosts in the same manner. Received8 August1984, accepted 25 July 1985.

SPECIESthat use a wide variety of resources plete descriptor of parasitism in this case. A are known as generalists.The brood parasitic fourth bias concernsparasitism of speciesthat Shiny Cowbird (Molothrusbonariensis) is an ex- reject cowbird eggs. Becauserejecter species treme generalist with respect to the "host generally remove cowbird eggsin a very short niche": its eggs have been found in nests of time, parasitism of rejecters is unlikely to be 201 species(Friedmann and Kiff 1985). observed(Rothstein 1971, 1975, 1977). Experi- Our knowledge of how female cowbirds se- ments simulating parasitism must have been lect hostsis incompleteand biased.First, data performedpreviously to identify rejecters. on parasitismare often gatheredindirectly by This study describesand interpretsthe con- studying a particular host species.Information trasting patterns of host selection by Shiny about the compositionof the host community Cowbird females at two sites in Buenos Aires and the rates of parasitismof other speciesis Province, Argentina. Geographic variation in frequently unavailable.A secondsource of bias host selection remains unexplained. For ex- concernsthe breeding seasonsof host and par- ample,parasitism of the Rufous-collaredSpar- asite. In Tucum•n Province, Argentina, only row (by the nominate subspeciesof the cow- early nestsof the Rufous-collaredSparrow (Zo- bird) varies from 15 to 77% according to site notrichiacapensis) escaped parasitism: all late (Sick 1958;King 1973;Fraga 1978, 1983;Goch- nestswere parasitized(King 1973). Third, sev- feld 1979; Salvador 1983). eral Shiny Cowbirds often parasitize a single I also presentdata on host selectionwith re- nest (Friedmann 1929, 1963; Friedmann et al. gard to egg color. Almost all eggs can be de- 1977). The proportion of nestsparasitized, the scribedas spotted or immaculate in easternAr- most frequently reported statistic,is an incom- gentina, Uruguay, and portions of Brazil. Intermediate eggswith a few fine spotsare rare (Friedmann 1929; Fraga 1978, 1983, 1985; Goch- •Present address:Department of BiologicalSci- feld 1979;Salvador 1983). Some host species, ences,University of California, Santa Barbara,Cali- dual accepters,incubate both morphs. Other fornia 93106 USA. hosts,dual rejecters,remove all cowbird eggs 61 The Auk 103: 61-69. January 1986 62 PAUl,MASON [Auk, Vol. 103 from the nest. One major host (Chalk-browed parasitized) are reported because many nests Mockingbird, Mimussaturninus) is a differential were multiply parasitized.Eggs laid in peculiar accepterthat ejectsimmaculate eggs but accepts circumstances(such that they could not be spotted ones (Fraga 1980, 1985; Mason 1986). scored for acceptanceor rejection), and eggs Severalother important host speciesprobably found in nestsbut later rejectedare included. are differential accepters of cowbird eggs Only 15 immaculateand 6 intermediateeggs (Friedmann et al. 1977, Fraga 1985). These with a few fine spots were seen. Intermediate speciesare particularly important becausethey eggs appear immaculate from a distance of representstrong sources of natural selectionon about 1 m or more. The Chalk-browed Mock- egg morphology and host selection. The eco- ingbird, the only speciesknown to responddif- logicalsituation could potentially result in the ferentially to different egg morphs(Fraga 1985, evolutionof hostraces or "gentes"(Payne 1977). Mason 1986), apparently perceivesintermedi- ate eggsas immaculate(see below). METHODS Comparisonsbetween sites.--Immaculate and intermediate eggs were rare at both sites. At One seasonwas spent at each of two sites located Site I, I found 4 immaculate eggs, 0 interme- 17.6 km apart near Magdalena, Buenos Aires Prov- diate eggs,and 8 spottedeggs in nestsof dual ince, Argentina: Site I was EstanciaSan Isidro and accepters.At Site II, I found 0 immaculateeggs, Site II was Estancia E1 Talar. Further details, includ- 1 intermediate egg, and 53 spotted eggsin such ing the relative abundanceof passefinespecies, are nests. presentedelsewhere (Mason 1985). The first season lastedfrom 20 October 1977 to 24 February 1978,and At both sites,species larger than the cowbird the second from 21 September 1978 to 10 February were preferred as hosts (Site I: G = 57.6, 7 df, 1979. Very little laying by any passerinespecies oc- P < 0.001; Site II: G = 34.0, 7 df, P < 0.001). curs outside this interval (see below and Mason 1985). The mockingbird was the only large species Nests were observedrepeatedly until young fledged present at both sites and representedby sam- or activity at the nest ceased.Observations were car- pies sufficient for statisticalcomparison; it was ried out daily, although most nests were not ob- parasitizedequally at each site (G = 13.3, 9 df, servedwith suchfrequency. Shiny Cowbird eggswere P > 0.1). All thrush nests were parasitized. In scored for color (spotted, immaculate, or intermedi- addition to the 5 nests found in the egg stage ate), date, and host. A cohort of nests of the same (see Appendix), an additional nest of each speciesat each site was then characterizedby a dis- tribution describing the frequency of nests with a specieseach contained 2 cowbird young. certain number of cowbird eggs of a specific type. At Site I, large, abundant passerines(exclud- Distributions obtained in this manner were com- ing the mockingbird) were: Rufous Hornero (a pared using the heterogeneityG-test (Sokal and Rohlf dual rejecter),Great Kiskadee(response to spot- 1981). Other statisticaltests vary accordingto the par- ted cowbird eggs uncertain; Mason 1986), Ru- ticular comparison(see below). The diversity of host fous-bellied Thrush, White-rumped Swallow use at each site was described by calculating the (body size considerablysmaller than that of the Shannon-Weaver index (Peet 1974, Whittaker 1975). cowbird, but included because of its wing- Scientificnames of speciesare given in the Appen- length), and the Brown-and-yellowMarshbird dix. [Pseudoleistesvirescens; no nestsfound, but a fre- quent hostchoice in BuenosAires (Gibson1918; RESULTS Hudson 1920; Friedmann 1929, 1963; Fried- mann et al. 1977)]. At Site II, the thrush was The data base consists of 360 nests (7 nest absent and the Brown-and-yellow Marshbird boxes)of 28 passerinespecies. I found 233 Shiny rare (Mason 1985). Excluding the small-bodied Cowbird eggs in nests of 13 species.The dis- swallow, only 1 large accepterof spotted eggs tribution of 212 spotted Shiny Cowbird eggsin (the mockingbird) was abundant at Site II, as 333 nestsdiscovered in the egg phase is given opposedto at least2 (mockingbird,thrush) but in the Appendix. Speciesare listed by decreas- probably 3 (including the marshbird) such ing winglength of adult females (Mason 1980) speciesat Site I. becausemore accurate data for body size (i.e. The sites differed markedly in that species weight) are lacking. Both intensity (mean num- smaller than the cowbird were used to a much ber of cowbird.eggs per parasitized nest) and greater extent at Site II. The Rufous-collared frequency of parasitism(percentage of nests Sparrow was used significantly more at Site II January1986] HostSelection by ShinyCowbirds 63 than at Site I (G = 9.7, 3 df, P < 0.01). Two oth- T^I•LE1. Diversity in host selection by site.a er species(Yellow-browed Tyrant, House Wren) were ignored by cowbirds at Site I, but both -pilog Pl were parasitizedat Site II, althoughsample sizes are small. The tyrant, wren (natural nest sites), Host diversity measuredby species and sparrow were used equally at Site II (G = Site I 4.4, 6 df, P > 0.50). In addition, the Bay-winged Rufous Hornero 0.013 0.024 Cowbird was used more at Site II than at Site Cattle Tyrant 0.013 0.024 Chalk-browed Mockingbird 0.759 0.091 I, and a single spotted cowbird egg was laid in Rufous-bellied Thrush 0.127 0.114 a nest of the Freckle-breasted Thornbird at Site Creamy-bellied Thrush 0.013 0.024 II the day following loss of that nest. Only 1 Bay-wingedCowbird 0.025 0.040 small species(Saffron Finch) was parasitized at Saffron Finch 0.013 0.024 Rufous-collaredSparrow 0.038 0.054 Site I but not Site II. When frequencies for all H' = 0.395 b small specieswith data from both sites were Site II pooled, smaller specieswere parasitized signif- Rufous Hornero 0.022 0.037 icantly more often (G = 16.9, 5 df, P < 0.005). Freckle-breasted Thornbird 0.007 0.016 House Wrens at Site II nesting in boxes were Yellow-browed Tyrant 0.030 0.045 more heavily parasitizedthan those nesting in Cattle Tyrant 0.007 0.016 natural sites (G = 17.30, 5 df, P < 0.05). A pair Fork-tailed Flycatcher 0.052 0.067 of wrens used a nest box at Site I, but this nest Wkite-rumpedSwallow 0.037 0.053 House Wren 0.059 0.073 was not parasitized. The only other use of a House Wren (nest boxes) 0.141 0.120 nest box was by a pair of Saffron Finches (Site Chalk-browed Mockingbird 0.533 0.146 II). Bay-wingedCowbird 0.015 0.027 Cowbirds selected a more diverse array of Rufous-collaredSparrow 0.096 0.098 hostsat Site II than at Site I (measuredby the H' = 0.698 Shannon-Weaver index; Peet 1974, Whittaker Host diversity measuredby sizec 1975), regardlessof whether hosts were char- Site I acterized by speciesor by size (Table 1). The Large 0.91 ! 0.037 latter analysis was done because a cowbird's Small 0.089 0.093 decision to parasitize a nest may be based on H' = 0.130 some general aspect,such as size, rather than Site II on speciesidentity (Rothstein 1976). Large 0.644 0.123 Parasitismof rejecters.--AlthoughI could not Small 0.356 0.160 estimatethe true frequencyof parasitismof re- H' = 0.283 jecters, I observed some cowbird eggs in nests ap's represent the proportionof spottedeggs found of the RufousHornero and the Fork-tailed Fly- in nestsof different species. catcher(both dual rejecters;Fraga 1980,Mason bHost diversity measured using the Shannon- 1986). Responsescould not be scoredin 3 cases Weaver index (Whittaker 1975):H' = -• pilog Pl of parasitismof the hornero becauseI acciden- cLarge and small reckonedusing winglength rel- ative to the Shiny Cowbird. tally destroyed one nest, or the cowbird egg was laid extremely late in the incubation or nestling period. Two spottedeggs were reject- hornero and the flycatcher,generally rejects ed and later found, one in the entrance of a immaculateeggs. Thirteen immaculateand in- nest and the other directly below a nest. A bro- termediateeggs were observedin 10 mocking- ken eggshell was found below another nest, bird nests.Three were accepted,7 ejected,and probably from an egg laid and rejected be- the remainder could not be scored(predation tween observations. Four nests of the Fork- or collection). At one nest where acceptance tailed Flycatchercontained cowbird eggs.Two was observed,an immaculateegg was rejected nestshad 3 cowbird eggseach (1 of which was the previousday. intermediate), and all eggs were ejected.Nei- Rejecter speciesof South America resemble ther nest was ever known to contain host eggs. North Americanrejecters in being large.Rejec- Two other nestseach received 1 spotted egg, ters (of at least one morph) are, on average, which was ejectedin both cases. larger than accepterspecies (Table 2; P < 0.05, The Chalk-browedMockingbird, unlike the Mann-Whitney U = 5; Siegel 1956). This test, 64 PAU•MA$O• tAuk, Vol. 105

TABLE2. Passefinesize (estimatedby winglength) and responseto cowbirdeggs.

Response to eggsø Fig. 1. Datesthat cowbirdeggs were laid or found. Wing- Immac- lengtha Species Spotted ulate 138 Brown-chested Martin Acc Acc The breeding seasonsof the Chalk-browed 129 Great Kiskadee c ? Acc Mockingbird and the Rufous-collaredSparrow 126 Chalk-browedMockingbird Acc Rej were completelywithin the breeding seasonof 125 Rufous-belliedThrush c Acc? Rej? the cowbird. I divided the breeding seasonsof 123 Creamy-belliedThrush Acc Acc thesetwo hostsinto quartersand tested for het- 119 White-rumped Swallow Acc Acc 112 Fork-tailedFlycatcher Rej Rej erogeneity among the subsamples. For the 107 RufousHornero Rej Rej mockingbird, I pooled both sites because the 104 Shiny Cowbird distributionswere equivalent. For the sparrow, 99 Cattle Tyrant Acc Acc parasitismonly from Site II was analyzed.The 98 Blue-and-yellowTanager Acc Acc subsamplesprovided no evidence of hetero- 93 Bay-wingedCowbird Acc Acc 90 Yellow-browed Tyrant Acc Acc geneityin either species(mockingbird: G = 27.4 85 Firewood-gatherer Acc Acc for 27 df, NS, 0.25 < P < 0.5; sparrow:G = 9.2 81 Vetmillion Flycatcher Acc Acc for 9 df, NS, 0.25 < P < 0.5). 73 House Sparrow Acc Acc 71 Grassland Yellow-Finch Acc Acc DISCUSSION 70 Freckle-breasted Acc Acc Thornbird 67 Rufous-collaredSparrow Acc Acc The extent of generalizationby Shiny Cow- 66 Saffron Finch Acc Acc birds varied between two sites. Most large 64 Tufted Tit-Spinetail Acc Acc speciesapparently were used consistently,re- 63 Wren-like Rushbird Acc Acc gardlessof responseto cowbird eggs. Use of 63 Little Thornbird Acc Acc 57 GrasslandSparrow Acc Acc smaller specieswas variable, but more preva- 51 House Wren Acc Acc lent at Site II. The Chalk-browed Mockingbird 51 Masked Gnatcatcher Acc Acc was parasitizedconsistently and heavily at both 48 White-crestedTyrannulet Acc Acc sites. Other large hosts (thrushes) were para- aRejecters are significantly larger than accepters sitized at Site I, but were absent from Site II. (Mann-Whitney U-test, P < 0.05, U = 5). Large and The Brown-and-yellow Marshbird, probably a small reckonedrelative to the Shiny Cowbird. common host in Buenos Aires Province (Hud- bAcc = accepts,Rej = rejects(Mason 1986). son 1874, 1920; Sclater and Hudson 1888; Gib- cNot included in the analysisbecause response to cowbird eggsnot verified experimentally. son 1918;see reviews by Friedmann 1929, 1963, and Friedmann et al. 1977), was abundant at Site I but rare at Site II. based on winglength, is conservativebecause The Shannon-Weaver indices (Table 1) de- it includes as large birds two Hirundinidae scribe the width of the "host niche" and illus- (White-rumped Swallow, Brown-chestedMar- trate differencesin site-specifichost selection. tin), birds with extremely long wings relative Although the measurementswere determined to body size. in part by my ability to find and examine nests, Seasonality.--Thefirst cowbird egg was laid the occasionalor rare parasitismof a few species on 7 October and the last on 19 January (Fig. not censused would result in only slight 1). Three species,all furnariids (Rufous Hor- changes.If I missedcommon, heavily parasit- nero, Firewood-gatherer, Freckle-breasted ized species,the indiceswould be substantially Thornbird), completeda portion of their breed- altered.The only specieslikely in this category ing before the cowbird (Mason 1985). The last is the Brown-and-yellowMarshbird at Site I. If two species accept all cowbird eggs (Mason this specieswas heavily parasitized, the index 1986), but only a single nest of the thornbird (reckoned by species)would increaseat Site I, was parasitized (see Appendix). Nests of the decreasing the difference between sites. If Rufous Hornero were ranked by date: parasit- reckoned by size, however, the index would ized nests occurred later in the season (Mann- increase the contrast because the marshbird is Whitney U = 8, P < 0.025; Siegel 1956). a large species(Mason 1980). January1986] HostSelection by ShinyCowbirds 65

Conspicuousnessmight predisposea nest to while the small Rufous-collaredSparrow is parasitism (Brown-headed Cowbird, Molothrus parasitizedless. During the last half of a 10~yr ater:Nice 1937,Rothstein 1975; Shiny Cowbird: study of the sparrow at Lobos, Buenos Aires Friedmann 1929, 1963; Gochfeld 1979), but the (Fraga 1978, 1983),parasitism dropped signifi- contrast between sites in parasitism of the Ru- cantly from 72.5 to 43.5%.This coincidedwith fous-collaredSparrow and other small species changesin local agriculturethat attractedthe arguesagainst the importanceof conspicuous- White-browed Blackbird(Sturnella superciliaris), ness.The elaboratenests of furnariids(e.g. the a bird much larger than the sparrow.Two of 6 hornero, Firewood-gatherer,and thornbirds) nestscontained cowbird eggs, but the species are conspicuousto the human observer, but was seen tending cowbird fledglings (Fraga theseare not equally likely to be parasitized. 1985).Parasitism of the mockingbird remained Birds that nest in old hornero nests (Tufted Tit- high (above that of the sparrow;Fraga 1985) Spinetail, White-rumped Swallow, Saffron and unchanged. In Villa Maria, C6rdoba Prov- Finch, House Sparrow) were likewise not ince, the mockingbirdwas usedmore than any equally parasitizeddespite the control for nest other local host (86.9%of nestsparasitized; Sal- site (Appendix; Mason 1985). The strongest vador 1983). Other large hosts also were used, support for the conspicuousnessargument is but only 8 of 22 Rufous-collaredSparrow nests the significantdifference (at Site II) in parasit- held cowbird eggs.With regard to the parasit- ism of wrens nesting in boxesas opposedto ism of this host, Salvador wrote: "at this site, it natural nesting sites. [the sparrow]was not a host of great impor- A more satisfyingexplanation attributes the tance,if we compareit with Mimussaturninus" local differences to the operation of the same [Salvador1983:155 (trans. by PM)]. host-selection mechanism in environments of- Gochfeld (1979) found heavy use (23/24 fering differing arrays of resources(hosts). nests)of the Long-tailedMeadowlark (Sturnella Preference measures have different formal de- loyca)but no use (0/11) of the highly similar scriptions(Ivlev 1961, Murdoch 1969, Rapport Lesser Red-breasted Meadowlark (Sturnella de- and Turner 1970, Chesson 1978, Jaenike 1980), filippi)in BahiaBianca, southern Buenos Aires but nonecan be calculatedhere becauseprecise Province.The low frequencyof parasitismof numericaldensities of host and parasitepopu- the Rufous-collaredSparrow (2/13 nests) re- lations are unavailable. Nonetheless, three dis- semblesmy resultsat Site I, but 0 of 4 nestsof tinct lines of evidenceprovide qualitativesup- mockingbirdscontained cowbird eggs (Goch- port that large hostsare preferred:(1) among feld pets.comm.). Both types of eggsoccurred accepters,larger hosts were parasitized more at the site. The absenceof parasiticeggs in frequently and to a greater extent than were mockingbird and Lesser Red-breasted Mea- smaller hosts;(2) all known rejecter species dowlark nestsmay possiblyreflect rejection be- (which are large)were parasitizedwhile many havior of the hostsrather than avoidanceby accepterswere left unmolested;and (3) when cowbirds. larger hostswere lessabundant or absent,cow- The preferencefor large hosts,a trait unusu- birds used smaller hosts. al in any parasiticbird (Payne 1977), may be a Niche expansionand inclusionof specieslike phenomenonrestricted to the Rio de La Plata the sparrowis probablyadaptive. Pecking of basinand surroundingarea. In the West Indies, host eggsby cowbirdsis a density-dependent there is no clearrelationship between host size form of mortality that sometimesterminates and preferenceranking by Shiny Cowbirds.In nestingattempts. Pecking depressesthe survi- Trinidad the diminutive House Wren may be vorship of mockingbirdnests to a level indis- the mosthighly preferredhost (Manolis 1982), tinguishable from that of the Rufous-collared while in Puerto Rico both large and small hosts Sparrow (Mason 1986). Presumably, if niche may be heavily used(P•rez Rivera 1983,Wiley width remained constant in the two environ- 1985). ments,nest losses to peckingwould further in- Hostselection by femaleslaying contrasting egg creaseat Site II, and cowbirdswould experi- types.--Only 5 immaculate and intermediate ence still lower success. eggs were laid in nests of accepters,whereas Other studies share a common feature that 16 were laid in nestsof speciesthat normally supportsthe claim of a preferencefor larger rejectthese morphs. Females that lay theseeggs hosts:at least one large host is heavily used, clearlydo not placethem preferentiallyin nests 66 PAULMASON [Auk,Vol. 103 where the eggswill be accepted.No evidence ACKNOWLEDGMENTS suggeststhat host selection varies between fe- males. M. Christie helped get my researchprogram start- ed. J. M. Gallardo (Director) and J. R. Navas (Jefe, Fraga (1985) also describedimmaculate eggs Divisi6n de Ornitologia) give me free accessto the laid in and rejected from mockingbird nests. collectionsof the Museo Argentino de CienciasNat- He estimated that at least one-third of the nests urales, Buenos Aires. I am indebted to the estanci- were parasitizedwith white eggs.Descriptions eros, E. M. and D. Earnshaw (EstanciaSan Isidro) and of immaculateeggs below, but spottedeggs in- A. Landa(Estancia E1 Talar), for allowing me to work side, nestsof the Brown-and-yellow Marshbird on their land. They and their families showed me (Hudson 1874, 1920; Sclater and Hudson 1888) the greatesthospitality. R. M. Fragaoffered helpful suggestthat this host is a differential accepter suggestionsand encouragementat all stagesof the research.M. Gochfeldprovided a list of nestshe found like the mockingbirdand that cowbirdsbehave at Bahia Bianca. Earlier versions of the manuscript the same toward the marshbird and the mock- were read and criticized by A.M. Dufty, Jr., R. M. ingbird. This pattern of host selectionis appar- Fraga,M. Gochfeld, W. W. Murdoch, S. I. Rothstein, ently stable because Hudson's observations and S.S. Sweet. Their commentshave considerably were conductedmore than a century ago. improved the final version. The Frank M. Chapman Parasitismof rejecters.--Parasitismof rejecter Fund of the American Museum of Natural History speciesposes a significantand interestingevo- provided partial financial support. I was supported lutionary problem.Selection penalizes females by NSF grant BSR 82-14999 to S. I. Rothstein during that lay either egg morph when they lay in the preparationof the manuscript.Finally, I would nestsof dual rejecters.However, ovipositionin like to acknowledge H. Friedmann, whose 1929 monographand subsequentpublications served as nests of differential accepterspenalizes only my initial inspirationto undertakethis project. certain females.The situation is complex be- cause selection occurs on a phenotype (the LITERATURE CITED morphologyof the egg) that is a productof the maternal genotype. CHESSON,J. 1978. Measuring preference in selective The frequencyof ovipositionin nestsof re- predation. Ecology59: 211-215. jecterssuggests that host race formation does FRAGA,R.M. 1978. The Rufous-collaredSparrow as a host of the Shiny Cowbird. Wilson Bull. 90: not occur in Shiny Cowbirds. Femalesthat lay 271-284. in nests of rejecters almost certainly did not 1980. The breeding of Rufous Horneros fledge from nestsof thosespecies. Similarly, it (Furnariusrufus). Condor 82: 58-68. seemsunlikely that all femaleslaying immac- 1983. Parasitismo de cria del renegrido ulate eggsin mockingbirdnests hatched from (Molothrusbonariensis) sobre el chingolo (Zono- spotted eggs laid in mockingbird nests, al- trichiacapensis): nuevas observaciones y conclu- though this possibility cannot be ruled out. In siones. Hornero 12, No. Extraord.: 245-255. addition, productionof cowbird offspringmay 1985. Host-parasite interactions between be highestfor hostsof low-preferenceranking. Chalk-browed Mockingbirds and Shiny Cow- Fraga(1985) presenteddata that more cowbirds birds. Pp. 829-844 in Neotropicalornithology (P. A. Buckley, M. S. Foster, E. S. Morton, R. S. fledged from nests of Rufous-collaredSpar- Ridgely, and F. G. Buckley, Eds.). Ornithol. rows than of Chalk-browed Mockingbirds,and Monogr. No. 36. he argued that most females laying in mock- FRIEDMANN,H. 1929. The cowbirds:a study in the ingbird nests probably fledged from nests of biology of social parasitism. Baltimore, C. C. other species. Thomas. The host selectionmechanism is clearly un- 1963. Host relations of the parasiticcow- der natural selection,but responseto selection birds. U.S. Natl. Mus. Bull. 233: 1-276. requires(among other things) that there exist , & L. F. KIFF. 1985. The parasitic cowbirds and their hosts. Proc. West. Found. Vert. Zool. 2: proximatecues for ultimate success.Apparent- 226-302. ly, these cuesare often lacking, at least in this --, & S. I. ROTHSTEIN. 1977. A further environment. Some features of Shiny Cowbird contribution to knowledge of the host relations parasitism,such as inclusion of the sparrow in of the parasitic cowbirds. Smithsonian Contr. the array of host choices,appear to be subtle Zool. 235: 1-75. adaptations(Mason 1986).Other features,such GIBSON,E. 1918. Further ornithological notes from as the parasitismof rejecters,remain perplex- the neighbourhoodof Cape San Antonio, Prov- ing. ince of BuenosAyres. Ibis 4, 4th Ser.: 1-38. January1986] HostSelection byShiny Cowbirds 67

GOCHFELD,M. 1979ß Broodparasite and host coevo- PEET,R. K. 1974. The measurement of speciesdi- lution: interactionsbetween Shiny Cowbirdsand versity.Ann. Rev. Ecol.Syst. 5: 285-307. two speciesof meadowlarks. Amer. Natur. 113: PgREZRIVERA, R. A. 1983. An account of bird para- 855-870. sitismby the Shiny Cowbirdin PuertoRico. E1 HUDSON,W. H. 1874. Notes on the procreant in- C6ndor: revista de la Universidad del Turabo 8: stinctsof the three speciesof Molothrusfound in 57-71. BuenosAyres. Proc.Zool. Soc.London: 153-174. RAPPORT,D. J., & J. E. TURNER. 1970. Determination ß 1920. Birds of La Plata, vol. I. New York, E. of predatorfood preferences.J. Theoret.Biol. 26: P. Dutton. 365-372. IVLEV,V.S. 1961. Experimentalecology of the feed- gOTHSTEIN,S.I. 1971. Observationand experiment ing of fishes(D. Scott,trans.). New Haven, Con- in the analysisof interactionsbetween brood necticut, Yale Univ. Press. parasitesand their hosts.Amer. Natur. 105:71- JAENIKE,J. 1980. A relativistic measure of variation 74. in preference.Ecology 61: 990-997. 1975. An experimental and teleonomic in- KING,J. R. 1973. Reproductiverelationships of the vestigationof avianbrood parasitism. Condor 77: Rufous-collaredSparrow and the Shiny Cow- 250-271. bird. Auk 90: 19-34. 1976. Cowbird parasitism of the Cedar MANOLIS,T.D. 1983. Host relationshipsand repro- Waxwingand its evolutionaryimplications. Auk ductivestrategies of the Shiny Cowbird in Trin- 93: 498-500. idad and Tobago. Unpublished Ph.D. disserta- 1977. Cowbird parasitism and egg recog- tion, Boulder, Colorado, Univ. Colorado. nition of the Northern Oriole. Wilson Bull. 89: MASON,P. 1980. Ecologicaland evolutionary as- 21-32. pectsof hostselection in cowbirds.Unpublished SALVADOR,S. A. 1983. Parasitismo de crla del re- Ph.D. dissertation, Austin, Texas, Univ. Texas. negrido (Molothrusbonariensis) en Villa Maria, 1985. The nesting biology of some passer- C6rdoba,Argentina (Aves:Icteridae). Hist. Nat. ines of BuenosAires, Argentina. Pp. 954-972 in 3: 149-158. Neotropical ornithology (P. A. Buckley, M. S. SCLATER,P. L., & W. H. HUDSON. 1888. Argentine Foster, E. S. Morton, R. S. Ridgely, and F. G. ornithology.Vol. I, Passeres.London, R. H. Por- Buckley,Eds.). Ornithol. Monogr. No. 36. ter. ß 1986. Brood parasitismin a host generalist, SICK,H. 1958. Notas bio16gicass6bre o gaud6rio, the Shiny Cowbird: I. The quality of different "Molothrusbonariensis" (Gmelin) (Icteridae, Aves). speciesas hosts.Auk 103: 52-60. Rev. Brasil. Biol. 18: 417-431. MURDOCH,W. W. 1969. Switchingin generalpred- SIEGEL,S. 1956. Nonparametricstatistics for the be- ators: experiments on predator specificity and havioral sciences. New York, McGraw-Hill. stability of prey populations.Ecol. Monogr. 39: SOKAL,R. R., & F. J. ROHLF.1981. Biometry,2nd ed. 335-354. San Francisco, W. H. Freeman. NICE,M.M. 1937. Studiesin the life history of the WHITTAKER,R. H. 1975. Communities and ecosys- Song Sparrow, part I. Trans. Linnaean Soc. New tems, 2nd ed. New York, Macmillan Publ. Co. York 4: 1-247. WILEY,J.W. 1985. Shiny Cowbird parasitismin two PAYNE,R. B. 1977. The ecologyof brood parasitism avian communities in Puerto Rico. Condor 87: in birds. Ann. Rev. Ecol. Syst.8: 1-28. 165-176. 68 P^ut.M^SON [Auk,Vol. 103

APPENDIX. Parasitizedand unparasitizednests.

No. of cowbird eggs per nestb Wing- length* Species 0 1 2 3 4 5 6 7 8 9 10 Freq.c Int. • A. Distribution, frequency,and intensity of parasitism 126 Chalk-browed Mockingbird (Mimus saturninus) Site I (n = 30) 8 8 6 5 0 0 0 1 1 0 1 0.73 2.73 Site II (n = 38) 10 9 6 6 3 3 1 0.74 2.57 125 Rufous-bellied Thrush (Turdusrufiventris) Site I (n = 4) 0 2 1 0 0 1 1.00 2.25 Site II Speciesabsent 123 Creamy-belliedThrush (T. amaurochalinus) SiteI(n= 1) 0 1 1.00 1.00 Site II Speciesabsent 119 White-rumped Swallow (Tachycinetaleucorrhoa) SiteI(n= 1) 1 0.00 -- Site II (n = 9) 5 3 1 0.44 1.25 112 Fork-tailedFlycatcher (Tyrannussavana)' Site I No nests examined Site II (n = 7) 3 2 1 1 0.57 1.75 107 Rufous Hornero (Furnariusrufus)' SiteI(n= 1) 0 1 1.00 1.00 Site II (n = 16) 13 3 0.19 1.00 104 Shiny Cowbird (Molothrusbonariensis) 99 Cattle Tyrant (Machetornis rixosus) SiteI(n= 1) 0 1 1.00 1.00 Site II (n = 2) 1 0.50 1.00 93 Bay-wingedCowbird (Molothrus badius) Site I (n = 10) 8 2 0.20 1.00 Site II (n = 5) 3 2 0.40 1.00 90 Yellow-browed Tyrant (Satrapaicterophrys) Site I (n = 2) 2 0.00 -- Site II (n = 8) 4 4 0.50 1.00 70 Freckle-breasted Thornbird ( Phacellodomusstriaticollis ) Site I (n = 3) 3 0.00 -- Site II (n = 4) 3 1 0.25 1.00 67 Rufous-collaredSparrow (Zonotrichiacapensis ) Site I (n = 32) 29 3 0.09 1.00 Site II (n = 19) 11 5 1 2 0.42 1.63 66 Saffron Finch (Sicalisfiaveola ) Site I (n = 13) 12 1 0.08 1.00 Site II (n = 22) 22 0.00 -- January 1986] Host Selectionby ShinyCowbirds 69

APPENDIX. Continued.

No. of cowbird eggs per nestb Wing- length' Species 0 1 2 3 4 5 6 7 8 9 10 Freq.c Int. a

51 House Wren (Troglodytesaedon) Natural nest sites Site I (n = 3) 3 0.00 -- Site II (n = 8) 3 3 1 1 0.63 1.60 Nest boxes SiteI(n= 1) 1 0.00 -- Site II (n = 5) 0 0 2 0 2 0 0 1 1.00 3.80 B. Speciesnot observedto be parasitizedf 138 Brown-chestedMartin (Phaeoprognetapera) (0, 1) 129 Great Kiskadee(Pitangus sulphuratus) ø (0, 7) 85 Firewood-gatherer(Anumbius annumbi) (0, 11) 81 Vetmillion Flycatcher(Pyrocephalus rubinus) (0, 22) 73 House Sparrow (Passerdomesticus) (0, 3) 71 Grassland Yellow-Finch (Sicalisluteola) (0, 3) 69 Hooded Siskin (Carduelismagellanica) (0, 7) 64 Tufted Tit-Spinetail (Leptasthenuraplatensis) (1, 3) 63 Wren-like Rushbird (Phleocryptesmelanops) (0, 22) 63 Little Thornbird (Phacellodomussibilatrix) (1, 0) 57 GrasslandSparrow (Ammodramushumerails) (0, 1) 51 MaskedGnatcatcher (Polioptila dumicola) (0, 1) 48 White-crestedTyrannulet (Serpophagasubcristata) (1, 3) ? TropicalKingbird (Tyrannusmelancholicus) (1, 0) ? Pipit (Anthussp.) (0, 1) aSpecies listed by decreasingwinglength (measuredfrom skins of adult femalescollected in BuenosAires Province; Mason 1980). bSpotted eggsonly. cFreq. = frequency(proportion of nestsparasitized). a Int. = intensity (mean number of eggsper parasitizednest). • Tyrannussavana and Furnariusrufus rejected cowbird eggs.Pitangus sulphuratus may show intermediate levelsof rejection.Some birds were not testedfor responseto cowbirdeggs (Mason 1986). eSample sizesfor sitesI and II in parentheses. REVERSED SEXUAL SIZE DIMORPHISM: EFFECT ON RESOURCE DEFENSE AND FORAGING BEHAVIORS OF NONBREEDING NORTHERN HARRIERS

ETHAN J. TEMELES Departmentof Zoology,University of California,Davis, California 95616 USA

ABSTR^CT.--Sexualdifferences in resourcedefense and foragingbehaviors during the non- breeding seasonare detailed for Northern Harriers (Circuscyaneus) in California. Female harriershunted more frequently in high (> 0.5m) vegetationthan males.In addition,females huntedat slowerspeeds and useddifferent hunting behaviors than males.Females in high vegetationshowed a significantlygreater response (i.e. attack)rate to approachingharriers than males,and femaleswon nearly all (28/29) aggressiveinteractions with males.These resultssuggest that sexual differences in harrierforaging behavior during the nonbreeding seasonresult from femalesexcluding males from preferredforaging areas and malesadopt- ing alternativeforaging strategies. Foraging strategies of harrier sexesare comparedwith foragingstrategies of sexesof birds in which malesare larger than femalesto examinethe role of bodysize in determiningsexual foraging strategies. Received 7 March 1985, accepted 2 July 1985.

IN many bird species(e.g. ducks,humming- ordinate females trapline-forage among dis- birds,woodpeckers), sexes differ in geographic persed resources, defend poorer-quality distribution, habitat use, or foraging behavior territories, or "rob" male territories (Stiles 1973, during the nonbreeding season(Ligon 1968, Wolf et al. 1976,Feinsinger and Colwell 1978, Feinsinger and Colwell 1978, Hogstad 1978, Kodric-Brown and Brown 1978, Stiles and Wolf Stiles and Wolf 1979, Kilham 1983, Hepp and 1979, Kuban and Neill 1980, Kuban et al. 1983). Hair 1984). Social dominance is one mecha- I examined the consequencesof sexual size nism that may influence nonbreeding sexual reversal on socialdominance and sex-specific foraging differences (Gauthreaux 1978). Usu- foraging behavior in the Northern Harrier ally, socialdominance confers priority of access (Circuscyaneus). Females are considerablylarg- to resources(e.g. food) and is attainedby those er than malesin this species,as in virtually all individuals that are most successfulin aggres- speciesof predatorybirds (Falconiformes,Stri- sive interactions(e.g. fights, chases)with oth- giformes,Stercorariidae). Sexual differences in ers (Morse 1980). distribution,habitat use, or foraging behavior Social dominance during the nonbreeding during the nonbreeding seasonhave been re- period affectsaccess to preferred resourcesand cordedfor harriers (Schipperet al. 1975, Bild- leads to sexual foraging differencesin many stein 1978, Bildstein et al. 1984) as well as for avian species(see above references).In gener- other raptors (Mills 1976, Stinson et al. 1981, al, larger males supplant smaller femalesand Marquiss and Newton 1982) and for nonrap- foragein areasof highestresource density; these torial speciesin which femalesare larger than sexualdifferences in foraging area often are ac- males(e.g. Puttick 1981).However, female so- companiedby sexualdifferences in foraging cial dominanceas the mechanismresponsible behavior when foraging behaviorsvary with for these differenceshas only been hypothe- habitattype. Alternatively, smaller females may sized or inferred (Mills 1976, Mueller et al. 1977, adopt different foraging behaviors to exploit Evans 1980, Boxall and Lein 1982) or deduced the same resources as dominant males. Both from experimental introduction of a captive patterns of sex-specificforaging during the bird into the territory of a resident bird (e.g. nonbreedingseason are well illustratedin many Cade 1955). hummingbirdspecies: larger (heavier),socially I presentdata on harriers that show that sex- dominant males defend high-quality food ter- ual differencesin foraging area and behavior ritories,whereas smaller (lighter), sociallysub- during the nonbreeding seasonare due to fe- 70 The Auk 103:70-78. January1986 January1986] HarrierResource Defense and Foraging 71 malesaggressively excluding males from pre- automobilespeedometer. To insure reliability and ferred foraging areas.I then examine the influ- consistencyof measurementwith this technique, I ence of body size on sex-specificforaging recordedhunting speedsof harriers only when driv- strategiesby comparingthe foragingbehaviors ing parallel to them for a period of >30 s. These measurementswere possibleowing to the numberof of harrier sexesto foraging behaviorsof sexes roadscrossing the studyarea and the harriers'accep- of birds in which males are larger than females. tance of moving automobiles.Because I found that harriers altered their hunting behavior during pe- STUDY AREA AND METHODS riods of high winds, I excluded from my analyses those observations of harriers made when wind The 25-km2 studyarea was located3 km northeast speedsexceeded 16 km/h. I recognized three cate- of Davis, Yolo Co., California. This area is primarily goriesof hunting speeds:(1) -<8 kin/h, (2) 16-24 kin/ agricultural cropland composedof the following h, and (3) ->32km/h. FollowingSchipper et al. (1975) vegetationtypes: corn stubble,rice stubble,winter and Bildstein (1978), I classifieda harrier's hunting wheat, winter-wheat stubble,plowed field, and fal- behaviorin one of three categories:(1) straightflight low field. The observationsreported here were part [lessthan 5 sharp(>30 ø)turns/mini through a field, of a long-termstudy of harrier foragingbehavior and (2) quartering(more than 5 sharp turns/rain) back cover the period 29 November 1983 to 31 January and forth over a field, and (3) border (edge) follow- 1984. These dateswere not arbitrarily chosen:har- ing (e.g. hunting along ditches, fence rows, road riers began to defend areas in late November and shoulders).I did not record subsequentobservations ceasedto defend areas in early February (observa- of a particular sex at a site on the sameday unless tions in the winter of 1984-1985 corroborate this). I two or more birds of the same sex were observed made 20 daysof observationsinvolving 8 h of field simultaneously. time per day. In addition to the general observationsdescribed When a foraging harrier was sighted, I recorded above, I made detailed observations of a smaller the date and time of each observation, the sex and number of harriers to obtain a better understanding ageof the bird, the vegetationtype and height, hunt- of their foraging behaviors.In this set of observa- ing speed (where possible),and hunting behavior tions, each individual harrier was watched for a 30- (where possible).Adult harriers were easily and un- rain focal period (Altmann 1974). During each 30- ambiguouslysexed: adult malesare gray dorsallyand rain period, I recordedthe number of prey capture adult females are brown. Juvenile C. c. hudsoniuscould attempts,the numberof successfulcaptures, the type be distinguishedfrom adult femalesby their rufous, of prey, the time spentin foraging,perching, eating, unstreakedunderbelly (Cramp and Simmons1980). or in aggressiveinteractions, the total area foraged Extremesexual size dimorphismalso facilitatedsex- in, the number of approachesby male/female con- ing harriers of all ages,especially on occasionswhen specifics,and the number of interactionswith male/ a bird was seenflying with a member of the opposite female conspecificsand outcome. sex. Becausethe number of juveniles (= rufous- I defineda captureattempt as an effortby a harrier breastedbirds) seen on the studyarea was extremely to catcha specificprey item; repeatedstrikes during low (9 juveniles comparedwith 153 adults, or 6%), the pursuit of a singleprey item were consideredto and becausestudies indicate that foragingof juvenile representa single capture attempt. Capture success birds differs from adults (e.g. Marr and McWhirter was definedas the percentageof all captureattempts 1982), I excludedjuveniles from the data analyses. that ended in prey capture (see Temeles 1985). Sev- Vegetation types were classifiedinto two cate- eral featuresof my study area and the behaviors of gories on the basisof height. Vegetation was de- harriers allowed me to accuratelyidentify the types scribedas "high" in fallow fields where all vegeta- of prey (birds, rodents)attacked. First, birds "flush" tion was >0.5 m high and as "low" in corn, rice, and out of vegetationwhen attackedby harriers,whereas winter-wheat stubble, winter-wheat fields, and rodents do not, and becausethe vegetation on my plowedfields where all vegetationwas <0.5 m high study site (even in the high areas)never exceeded (n = 500 measurementsof vegetationheight, mea- 1.2 m and most observations were made at distances suredfrom groundsurface to top of plant). Although of lessthan 100 m, it was possibleto seeavian prey vegetationtype and height were the variablesof pri- as they escapedharriers' attacks.Second, on many mary interest for statisticalanalyses, differences in occasionsharriers picked up vole "nests" while ex- vegetationtype and height alsowere associatedwith ecuting unsuccessfulcapture attempts. Third, har- differencesin vegetationdensity: vegetation in high, riers often utilize different attack strategiesfor birds fallowfields was much denser than vegetationin low, and rodents[e.g. they "pounce"at birdsin more rap- stubblefields, where rows were separatedby 7-10 id motions; Schipper et al. (1975), Bildstein (1978); cm of open space(n = 500 measurementsof vegeta- pers. obs.]. tion density,i.e. distancesbetween adjacent plants). All time observations were recorded on a stop- Hunting speeds of harriers were recorded on an watch or measuredfrom tape-recorderplaybacks. By 72 ETHANJ. TEMELE$ [Auk, Vol. 103

TAI•LI•1. Morphologyof 10 male and 10 femaleadult Northern Harriersin the LouisianaState University Museumof Zoologycollection. Data are means+ SD.

Measurement Male Female P Wing chord(ram) 346.4 _+9.02 383.9 _+7.68 <0.001 Bill chord (ram) 16.4 +__0.70 19.3 +__0.42 <0.001 Middle toe length (ram) 34.9 +__1.45 41.2 + 1.03 <0.001 Hallux claw arc (ram) 22.5 + 0.67 27.6 + 1.37 <0.001

using different number combinationsand colorsof they generallydid not utilize the samearea for more surveyor's tape tied on many markers (e.g. fence wires) at regular intervals on flat land, I estimated than 14 days. Consequently,I never made a focal observationof a harrier if a previousfocal observa- foraging areas hunted in by harriers. Distancesbe- tion had been made of a harrier of the same sex in tween markers were measured using a 100-m field tape measureor, in caseswhere the layout of roads the samearea within 14 days, unlessthe new indi- permitted, an automobileodometer. Odometer mea- vidual differed noticeablyin plumage from the old individual. Nonparametricstatistical procedures fol- surementswere checkedfor accuracyagainst tape low Daniel (1978) and Siegel (1956); parametric sta- measurementsand revealed no differences.Foraging areaswere measuredafter each focal period. tistical proceduresfollow Snedecor and Cochran (1967).Significance levels were determinedfrom ta- I defined an approachas any harrier (male or fe- bles in Hald (1952). male) coming within 100 m of the focal bird. I rec- ognized four different harrier aggressivebehaviors, which I placed into two categorieson the basisof RESULTS presenceor absenceof physicalcontact. Noncontact behaviors were (1) Chase,in which the focal bird flew Morphology.--Sizedifferences between sexes at the approachingbird and then flew after the ap- of the North American race C. c. hudsonius are proachingbird as it fled, from behind and on level highly significant (Table 1; P < 0.001). Cramp with it; and (2) Escort,in which the focal bird flew at and Simmons (1980) presented morphological the approachingbird and then flew after the ap- measurementsfor the European subspeciesC. proachingbird as it fled, from behind and belowit, c. cyaneusand verified that all sexual size dif- both birds flying far out of the focal bird's foraging area (seeBildstein and Collopy 1985).Chase and es- ferencesare significant.Smaller body weights cort behaviorsdiffered at a quantitative as well as a and winglengths of males correspondedto qualitative level; escortswere considerablylonger smallerwing loading and a large differencein than chases (P < 0.001, t-test; mean + SD escort = the estimatedmetabolic cost of flight (Table 2). 35.3 _ 14.2 s, n = 30; mean + SD chase = 18.3 +_ 8.9 Habitatuse and hunting speeds.--Sexes differed s, n = 21; no sex differences). Contact behaviors were with respect to vegetation height, where the (1) Short contact, in which the focal bird flew at and distributions were 83 females and 21 males in then hit or talon-grappledwith the approachingbird; high (>0.5 m) vegetation,and 25 femalesand and (2) Longcontact, in which the focal bird flew at 24 malesin low (<0.5 m) vegetation(x 2 = 13.3, and then repeatedly hit and grappled with the ap- df = 1, P < 0.0005). Females were observed to proachingbird, both birds falling to the groundsev- hunt more often at slow (-<8 kin/h) speeds, eral times. Long and short contactbehaviors also dif- whereas males were observed to hunt more feredconsiderably in length (P < 0.001,t-test; mean +_ SD long = 65.0 + 17.5 s, n = 8; mean _+SD short = often at higher (->16 kin/h) speeds(X 2= 42.2, 19.2+ 9.2 s, n = 6; femalesonly). I defineda winner df = 2, P < 0.0005; number of observations of of an aggressiveencounter as a bird that succeeded eachsex at given speeds,-<8 kin/h: 34 females, in driving the otherbird (= loser)out of the foraging 1 male; 16-24 kin/h: 6 females, 15 males; >-32 area within 2 min after initiation of the encounter. kin/h: 1 female, 11 males). The sexual differ- In addition, I recordedother aggressiveinteractions encesin hunting speedswere associatedwith observed(outside of focalsamples). vegetationheights. In high vegetation,31/32 For purposesof statisticalindependence of data, I femalesobserved were hunting at speeds-<8 attemptedto minimizerepeat observations of the same kin/h, whereas 16/16 males observed were individual in two ways. First, I tried to identify as many individuals as possibleby noting plumage hunting at speeds>-16 km/h (X2 = 43.8, df = 1, characteristics(e.g. missingor damagedfeathers) and P < 0.0005). In contrast,there were no signifi- favorite perch sites.Second, my observationsof har- cant differencesbetween the sexesfor hunting riers that couldbe positively identified indicatedthat speedsin low vegetation. Moreover, sexesof January1986] HarrierResource Defense and Foraging 73

T^•.z 2. Measurementsof winglength,body weight, T^B•,œ3. Hunting areas(kin 2) determined during 30- and wing loading and calculatedminimum power min focal observationsof Northern Harriers by sex required to fly, P•in (W, 1 W = 1 J/s) for sexesof and vegetation height class.n = samplesize. Northern Harriers. Vegeta- Wing tion Huntingarea Winglength' Body wt • loading* height Sex (m) (kg) (g/cm2) P•i•c Sex class Median Mean + SD n Male 0.344 (20)a 0.350 (90) 0.270 (2) 3.02 Female High 0.149 0.185 + 0.147 16 Female 0.385 (20) 0.531 (97) 0.315 (2) 5.12 Low 0.932 1.018 + 0.245 3 ßData from Snyder and Wiley (1976). Male High 1.126 0.977 + 0.340 4 Low 1.087 0.964 + 0.405 4 bAverage wing loading from data in Poole (1938) and Brown and Amadon (1968). cP•n calculated from Pennycuick's (1972) equa- tions. Wing span, b, was estimated from Greenewalt (1962), where b = 2(winglength)/0.62. Air density at ing, whereas 17/17 females observed in low sea level was used (1.22 kg/m•). vegetation hunted by straight-flying and bor- a Sample sizesare given in parentheses. der-following (X2 = 44.9, df = 1, P < 0.0005). Hunting areas.--Hunting areas (as deter- harriersshowed opposite results in the relation mined during each 30-rain focal observation) of hunting speedto vegetationheight. Hunt- are presentedfor harrier sexeswith respectto ing speedsof female harriers were negatively vegetationheight in Table 3. A Kruskal-Wallis associatedwith vegetationheight: 31/34 obser- ANOVA of the mediansof thesehunting areas, vationsof femaleshunting at speeds-<8 km/h with sex and vegetation height classes(e.g. fe- were in high vegetation, whereas 6/7 obser- males observed in high vegetation) as factor vationsof femaleshunting at speeds-> 16 km/h levels, was highly significant (H = 18.0, P < were in low vegetation (X2= 15.8, df = 1, P < 0.005; n • 27 different individuals). Two-tailed 0.0005).Hunting speedsof male harriers were nonparametricmultiple comparisonsindicated positively associatedwith vegetation height: that the median of the hunting areasof female 10/16 observationsof maleshunting at speeds harriers in high vegetation was significantly -<24 km/h were in low vegetation, whereas smaller than the median of the hunting areas 10/11 observationsof maleshunting at speeds of male harriers in high or low vegetation, or ->32 km/h were in high vegetation (Fisher ex- female harriers in low vegetation (P < 0.05 act test, two-tailed, 0.01 < P < 0.02). overall). Hunting behaviorand habitat.--Females were Captureattempts.--Capture rates varied be- observed to hunt most often by quartering. tween harrier sexeswith respect to vegetation Males hunted by straight-flying or border-fol- height (Table 4). A Kruskal-Wallis ANOVA of lowing (X2 = 42.7, df = 2, P < 0.0005; number median capture rates, with sex and vegetation of observationsof eachsex using given behav- height classesas factor levels, was highly sig- iors,quarter: 40 females,2 males;straight flight: nificant (H = 12.6, 0.005 < P < 0.01; n = 32 dif- 5 females,26 males;border-following: 16 fe- ferent individuals). Female harriers that hunt- males,10 males).The relation amongsex, hunt- ed in high vegetationhad significantly more ing behavior,and vegetationheight paralleled capture attempts per minute than female har- the relation among sex, hunting speed, and riershunting in low vegetation(0.01 < P < 0.05 vegetation height. Sexual differencesin hunt- overall; two-tailed nonparametric multiple ing behaviors also were associatedwith high comparisons)and tended to have more capture vegetation. In high vegetation, 40/44 females attemptsper minute than male harriers hunt- were observedto hunt by quartering, whereas ing in high or low vegetation (Table 4). 20/22 maleswere observedto hunt by straight- Capturesuccess.--Capture successes for differ- flying and border-following (X2 = 42.4, df = 1, ent prey types are presentedfor harrier sexes P < 0.0005). No significant differences be- with respect to vegetation height in Table 5. tween the sexes were observed for distribu- No significantdifferences in harriers' capture tions of hunting behaviorsin low vegetation. successes were observed between sexes within Quartering by female harriers was associated vegetation height classes,between sexes, be- with vegetation height: 40/44 females ob- tween vegetation height classes,or between served in high vegetationhunted by quarter- prey types (i.e. birds, rodents),but capturesuc- 74 ETHANJ. TEMELES [Auk, Vol. 103

TABLE4. Number of prey captureattempts per mi- TABLE5. Capture successesof Northern Harriers nute of hunting by Northern Harriers with respect with respectto sexand vegetationheight classfor to sexand vegetationheight class. n = samplesize. different prey types. Numeratorsare number of successfulattempts, denominators are total num- ber of attempts. Vegeta-tion No.of attempts height perrain hunt Sex class Median Mean + SD n Vegeta-tion Preytype height Pheas- Small Female High 0.50 0.53 + 0.34 16 Sex class Rodents ants birds Low 0.00 0.15 + 0.37 7 Male High 0.04 0.05 + 0.05 4 Female High 7/111 0/11 0/2 Low 0.00 0.05 + 0.06 5 Low 0/3 -- 0/2 Male High I / 14 -- 0 / 1 Low -- -- 0/4 cesseswere extremely low [overall capture suc- cess, rodents: 8/128 (6.25%), birds: 0/20 (0%)]. When hunting attemptsby harrier sexeswith- corrected for ties, Hc = 1.8, P > 0.1; n = 36 dif- in each vegetation height classwere combined, ferent individuals). Responsesto approaching a significantdifference in the distribution of harriers did differ significantly among sex- attempts at prey types was observed between vegetationheight factor levels (Table 8; Krus- vegetation height classes.The distributions kal-Wallis ANOVA, H = 13.8, 0.001 < P < were 125 and 3 attacksat rodents in high and 0.005; n = 27 different individuals). The medi- low vegetation, respectively, and 14 and 6 at- an responserate (number of attacks/30 rain) of tacks at birds in high and low vegetation, re- females in high vegetation (median = 2) was spectively (X2 = 18.6, df = 1, P < 0.0005). significantly higher (P < 0.05 overall; two- Aggressiveinteractions.--Physical contact oc- tailed nonparametric multiple comparisons) curred more frequently in female-female ag- than median responserates of males in high gressiveinteractions than in female-male ag- (median = 0) or in low vegetation (median = gressive interactions; 14/40 female-female 0). To summarize, females won virtually all ag- interactionsinvolved contact,whereas only 2/ gressiveinteractions with males,and although 29 female-male interactions involved contact both sexeswere approachedby conspecificsat (Table 6; X2 = 7.5, df = 1, P < 0.01). The differ- the samerate, only femalesusually responded ence between harrier sexeswas also significant to approachesby conspecifics. in the outcomes of aggressiveinteractions: in female-female interactions, a female that drove DISCUSSION her opponent out of the foraging area within 2 rain from the start of the interaction was Frequencyof aggressionin winteringraptors.- identified in 19/40 interactions, whereas in fe- Harrier sexes differ in degree of resource de- male-male interactions, females won 28/29 in- fense, with females typically defending, and teractions (X2= 18.6, df = 1, P < 0.0005). Too males typically not defending, foraging areas. few male-male interactions (n = 3) were ob- The aggressiveinteractions I observedmay be served for statisticalanalyses (see Table 6). attributed to high harrier densitiesand high, From the above data (especiallythe paucity concentratedprey densities.Harriers were ex- of male-male interactions), I tentatively con- tremely abundant in the study area. Densities cluded that typically only females defended of 8 harriers/kin 2 and 15 raptors/kin 2 were foragingareas. However, the low frequencyof common, especially in high-vegetation areas. male interactions could have been the result of These high raptor densitiesmay have resulted differencesbetween the sexesin approach fre- in more frequent aggressive interactions. A quencies.To clarify this issue,I examined ap- similar situation was reported for Snowy Owls proach and attack frequencies for focal birds (Nycteascandiaca; Evans 1980). Probably,these with respectto sex and vegetation height (i.e. high raptor densities occurred in responseto foraging area).No significantdifferences were either high or nonuniform (or both) prey den- found in approachfrequencies (number of ap- sitieswithin the studyarea (Snyderand Snyder proaches/30rain) with respectto sexand vege- 1970, Newton 1979). However, under most cir- tation height (Table7; Kruskal-WallisANOVA cumstancesprey densities may not be high January1986] HarrierResource Defense and Foraging 75

T^I•I•E 6. Number of observations of inter- and in- T^BLE7. Number of approachesby conspecificsto trasexual aggressiveinteractions of nonbreeding individual Northern Harriers per 30-rain focalob- Northern Harriers. Male-male interactions (2 es- servationby sex and vegetationheight class.n = corts, 1 chase) are not included. Also not included sample size. are 2 avoidance movements by males at the ap- proach of females. Vegeta- tion No.of approaches Short Long height Sexes Chase Escort contact contact Sex class Median Mean + SD n Female-female 13 13 6 8 Female High 1.50 1.72 + 1.32 18 Female-male 8 17 I I Low 1.00 1.22 + 1.09 9 Male High 1.00 1.00 ñ 1.00 5 Low 1.50 1.25 + 0.50 4 enough for raptors to be concentratedin one area, and hence aggressive interactions be- tween individuals may occur so infrequently vegetation areas becausecapture successwas as to be rarely or never recorded.In addition, higher there, since I found no differences in degree of aggressionhas been shown to vary capturesuccess between high- and low-vege- with resource quantity or quality (Ewald and tation areas. Carpenter 1978, Frost and Frost 1980). Conse- Sexualdifferences in foragingbehaviors.--Sexual quently, interactions between raptors may be differencesin flight speeds and foraging be- much lessovert than in this study and may take haviorswere observedonly in high-vegetation the form of avoidance(rather than aggressive) areas,where female harriers hunted by slow behaviors, which are more difficult to discern. quartering whereas male harriers hunted by These are important points to consider in un- fast,straight flight. The highesthunting speeds derstanding sexual differences in wintering recordedwere for maleharriers hunting in high distributions of birds, because failure to ob- vegetation, which contradicts all previous serve overt, aggressive interactions may not studies on the relationship between harrier necessarilyindicate an absenceof social dom- hunting speedsand vegetationtype (Brown and inance. Amadon 1968, Schipper et al. 1975, Bildstein Sexualdifferences in foragingareas.--These re- 1978).I proposethat fast foraging flight was an sults support the view that niche differences alternative male strategy that allowed male between nonbreedingharrier sexesarise from harriersto foragein areasdefended by females female dominance.I suggestthat female dom- while minimizing detection. Ideally, this hy- inance conferredpriority of accessto preferred pothesized shift should be tested by experi- resources(i.e. the high-vegetation areas) and ments in which male foraging behavior and that as a result of female attacks, male harriers hunting speed are examined in the absenceof and subordinate females shifted from pre- females (Morse 1974, Peters and Grubb 1982). ferred to less-preferredforaging areas in low However, my observationssuggest that fast vegetation. Some evidence supports the con- male flight was a behavioralshift from the slow- clusionsthat the high-vegetationareas defend- quartering hunting behavior usually associated ed by femaleswere preferred and that subor- with high vegetation. First, one male that en- dinates were forced to forage elsewhere. For tered a high-vegetation area in the absenceof example, females in high vegetation hunted a defendingfemale hunted at slow speeds(-<8 over much smaller foraging areasthan either km/h) by quarteringflight and made as many femalesin low vegetationor malesin high and capture attempts per minute as females hunt- low vegetation,which is expectedfrom the nu- ing that area. (He eventually was chasedaway merous studies that show inverse correlations by a female.)Second, at the beginning of the betweensize and quality of foraging areas(see 1984 breeding season,male harriers began to Hixon 1980).In addition, femalesin high vege- attack femalesby chasingor hitting them (al- tation showeda much greater attack response though they were unable to drive femalesfrom to approaching harriers than males, and fe- their foraging areas and hence "win" the in- maleswon virtually all aggressiveinteractions teraction).Such attacks by breeding maleshave with male harriers.Finally, it is unlikely that been observedin other avian species(Arm- subordinateharriers preferred to hunt in low- strong 1955, McLaren 1975). At this time, both 76 ETHANJ. TœMEL• [Auk,Vol. 103

TABLE8. Number of responses(e.g. attacks)to con- strike) increaseswith increasing body mass specificsby individual Northern Harriers per 30- (Anderssonand Norberg 1981), weight (and min focalobservation with respectto sexand vege- tation height class(approached individuals only). overall size) may be the decisivefactor in ag- n = sample size. gressive interactions between harriers, espe- cially between males and females,where fe- Vegeta- males have a size advantage(Tables 1, 2). [On tion No.of responses this point, Marquiss and Newton (1982) com- height Sex class Median Mean + SD n mented that female Eurasian Sparrowhawks (Accipiternisus) often kill and eat smaller male Female High 2 2.00 + 1.15 13 Low 1 0.86 + 0.69 7 sparrowhawksduring the nonbreeding sea- son.] Flight costsfor male harriers are much Male High 0 0.00 + 0.00 3 Low 0 0.00 + 0.00 4 less than for female harriers, however, owing to their smaller size and lighter wing loading (Table 2). Hence, males may be better able to survive in and to exploit marginal environ- male and female harriershunted in high vege- ments. tation at slow speedsby quartering.Finally, no Similarities between harriers and humming- differences were observed between sexes of birds are apparent in the types of aggressive harriers that foraged in areasof low vegetation interactionsbetween the sexes.In this study, (seeResults), yet female (but not male) hunting female harriers engagedin prolonged interac- speedsand foragingbehaviors changed in high tions marked by much hitting, talon-clasping, vegetation in the manner (i.e. slow, quarter) and tumbling to the ground. Often an intrud- predicted by previous studies (Brown and ing female returned to a residentfemale's area Areadon 1968, Schipper et al. 1975, Bildstein and repeatedlyharrassed her. Male harriers, on 1978). the other hand, were wary and inconspicuous. Comparisonswith large male-small female They entered a female's area quickly by flying species.--Theforaging strategiesused by sexes low to the ground and fled immediately when of Northern Harriers during the nonbreeding discovered and chased by a resident female. seasonshow striking resemblancesto thoseof Rapid flight is especiallyimportant if a male other birds, especially hummingbirds. Given captures prey in a female's area: one male I that both harriers and hummingbirds are dis- observedfleeing a female'sarea with prey was tinguishedby the fact that they forageprimar- struckby the femaleand had the prey snatched ily on the wing, perhapsthis is not surprising. from him. These observations of harrier strat- Resourcedefense by hummingbirds is associ- egiescorrespond well to the "challenger"male ated with differences in body weight and (large, conspicuous,aggressive) and "robber" winglength (i.e. wing disc loading; Feinsinger female (small, submissive, inconspicuous) and Chaplin 1975, Kodric-Brown and Brown hummingbirdsdiscussed by Kodric-Brownand 1978,Feinsinger et al. 1979). Heavy weight and Brown (1978). proportionally shorter wings result in greater The above comparisonsbetween harriers and maneuverability, which presumably is an ad- hummingbirds illustrate the significance of vantagein territorial defenseby large species body size in determining both the outcomeof and the larger sex(usually males in humming- aggressiveinteractions and the foraging strat- bird speciesweighing more than 4 g; Las- egies of sexes of nonbreeding birds. Domi- iewski and Lasiewski1967, Carpenter 1976).The nance by the larger sex may reduce food com- lighter weight and longer wings of smaller petition and may insure priority of accessto species and female hummingbirds result in preferred resources(Gauthreaux 1978, Morse more energeticallyefficient flight and hencea 1980). However, the lower energy demandsof better ability to exploit resources,which may individuals of the smaller sex may allow them be important in marginal habitats or when a to survive where larger individuals cannot bird is restrictedto foragingaround a nestarea (Morse 1980). (Kodric-Brown and Brown 1978). ACKNOWLEDGMENTS In harriers,heavy weight alsomay be advan- tageousin territorial defense, but in a some- I thank J. V. Remsen, Jr., and the Louisiana State what different way. Becauseterminal speed in University Museum of Zoology for permitting me to a dive (and presumably the force of a raptor's measurespecimens of Northern Harriersin their col- January1986] HarrierResource Defense and Foraging 77 lection. I thank T. Schoenerfor advice and support ing strategyin hummingbirds.Amer. Natur. 109: and D. Amadon, K. Bildstein,G. S. Millsß R. Reyn- 217-224. olds, D. Spiller, and J. Stamps for comments on the ß & R. K. COLWELL.1978. Community orga- manuscript.I especiallyappreciate A. Hedrick's as- nization amongnectar-feeding birds. Amer. Zool. sistanceand encouragement.I thank the Davis Au- 18: 779-795. dubon Society,Heidrick Farms,Inc., and Hunt-Wes- , --, J. TERBORGH,•r S. B. CHAPLIN. 1979. son FoodsßInc., for allowing me accessto harrier Elevationand the morphology,flight energetics, habitat on their lands. This study was supportedby and foragingecology of tropicalhummingbirds. Graduate Student Research Awards, Jastro-Shields Amer. Natur. 113: 481-497. Awardsßand Regents'Fellowships from the Univer- FROST,S. K., & P. G. H. FROST.1980. Territoriality sity of California, Davis. and changesin resourceuse by sunbirdsat Leon- otisleonurus (Labiatae). Oecologia 45: 109-116. LITERATURE CITED GAUTHREAUX,S. A., JR. 1978. The ecologicalsignif- icance of behavioral dominance.Pp. 17-54 in ALTMANN,J. 1974. Observationalstudy of behav- Perspectivesin ethology,vol. 3 (P. P. G. Bateson iour: sampling methods. Behaviour 79: 272-312. and P. H. Klopfer,Eds.). New York, Plenum Press. ANDERSSON,M., ,•' R. A. NORBERG. 1981. Evolution GREENEWALT,C.H. 1962. Dimensionalrelationships of reversedsexual dimorphism and role parti- for flying animals. SmithsonianMisc. Coil. No. tioning among predatory birds, with a size scal- 144(2). ing of flight performance.Biol. J. Linnaean Soc. HALE),A. 1952. Statistical tables and formulas. New 15: 105-130. YorkßJohn Wiley & Sons. ARMSTRONGßE.A. 1955. The wren. London, Collins. HEPP,G. R., & J. D. HAIR. 1984. Dominance in win- BILE)STEIN,K. L. 1978. Behavioralecology of Red- tering waterfowl (Anatini): effects on distribu- tailed Hawks (Buteojamaicensis), Rough-legged tion of the sexes. Condor 86: 251-257. Hawks(B. lagopus), Northern Harriers (Circus cy- HIXON,M.A. 1980. Food productionand competi- aneus),American Kestrels (Falco sparverius) and tor density as the determinantsof feeding ter- other raptorial birds wintering in southcentral ritory size. Amer. Natur. 115: 510-530. Ohio. Unpublished Ph.D. dissertation,Colum- bus, Ohio State Univ. HOGSTAD,O. 1978. Sexual dimorphism in relation to winter foraging and territorial behavior of the , W. S. CLARKßD. L. EVANSßM. FIELE),L. SOUCY, Three-toed Woodpecker Picoidestridactylus and ,•' E. HENCKEL.1984. Sexand age differencesin three Dendrocoposspecies. Ibis 120: 198-203. fall migration of Northern Harriers. J. Field Or- nithol. 55: 143-150. KILHAM,L. 1983. Life history studiesof woodpeck- ers of eastern North America. Publ. Nuttall Or- --, & M. W. COLLOP¾.1985. Escortingflight and nithol. Club No. 20. agonistic interactions in wintering Northern Harriers. Condor 87: 398-401. KOE)RIC-BROWN,A., •r J. H. BROWN. 1978. Influence BOXALL,P. C., & M. R. LEIN. 1982. Territorialityand of economicsßinterspecific competitionß and sex- habitatselection of femaleSnowy Owls (Nyctea ual dimorphismon territoriality of migrant Ru- scandiaca)in winter. Can. J. Zool. 60: 2344-2350. fous Hummingbirds.Ecology 59: 285-296. BROWNßL., ,•' D. AMADON.1968. Eagles,hawks and KUBAN,J. F., J. LAWLEY,•r R. L. NEILL. 1983. The falcons of the world. New Yorkß McGraw-Hill. partitioningof floweringcentury plants by Black- CADE,T. J. 1955. Experimentson winter territori- chinned and Lucifer hummingbirds.Southwest- ern Natur. 28: 143-148. ality of the American Kestrel, Falcosparverius. Wilson Bull. 67: 5-17. , & R. L. NEILL. 1980. Feeding ecology of CARPENTERßF. L. 1976. Ecologyand evolutionof an hummingbirds in the highlands of the Chisos Andean hummingbird ( Oreotrochilusestrella). Mountainsß Texas. Condor 82: 180-185. Univ. California Publ. Zool. 106: 1-74. LASIEWSKI,R. C., & R. J. LASIEWSKI.1967. Physiolog- CRAMP,S., '•' K. E. L. SIMMONS. 1980. Handbook of ical responsesof the Blue-throated and Rivoli's the birds of Europe,the Middle Eastand north hummingbirds. Auk 84: 34-48. Africaß vol. II. Oxfordß Oxford Univ. Press. LIGON,J. D. 1968. Sexual differences in foraging DANIEL,W. W. 1978. Appliednonparametric statis- behavior in two speciesof Dendrocoposwood- tics. Boston,Houghton Mifflin. peckers.Auk 85: 203-215. EVANSßD. L. 1980. Vocalizations and territorial be- MARQUISS,M., ,•' I. NEWTON. 1982. Habitat prefer- havior of wintering Snowy Owls. Amer. Birds encein male and female SparrowhawksAccipiter 34: 748-749. nisus. Ibis 124: 324-328. EWALD, P. W., & F. L. CARPENTER. 1978. Territorial MARR, T. G., ,•' D. W. McWHIRTER. 1982. Differential responsesto energy manipulations in the Anna hunting successin a group of Short-earedOwls. Hummingbird. Oecologia31: 277-292. Wilson Bull. 94: 82-83. FEINSINGER,P., & S. B. CHAPLIN. 1975. On the rela- MCLAREN,M.A. 1975. Breedingbiology of the Bo- tionshipbetween wing discloading and forag- real Chickadee. Wilson Bull. 87: 344-354. 78 ETHANJ. TEMELES [Auk,Vol. 103

MILLS, G. S. 1976. American Kestrel sex ratios and SIEGEL,S. 1956. Nonparametricstatistics. New York, habitat separation.Auk 93: 740-748. McGraw-Hill. MORSE, D.H. 1974. Niche breadth as a function of SNEDECOR,G. W., & W. G. COCHRAN. 1967. Statistical social dominance. Amer. Natur. 108: 818-830. methodsß 6th ed. Ames, Iowa State Univ. Press. 1980. Behavioralmechanisms in ecology. SNYDER,N. F. R., & H. A. SNYDER. 1970. Feeding Cambridge, Massachusetts,Harvard Univ. Press. territoriesin the EvergladeKite. Condor 72: 492- MUELLERßH. C., D. D. BERGER,& G. ALLEZ. 1977. The 493. periodic invasions of Goshawks. Auk 94:652 --, & J. W. WILEY. 1976. Sexual size dimor- 663. phism of hawks and owls of North America. Or- NEWTON,I. 1979. Population ecology of raptors. nithol. Monogr. No. 20. Vermillion, South Dakota, Buteo Books. STILES,F.G. 1973. Food supply and the annual cycle PENNYCUICK,C. 1972. Animal flight. London, Ed- of the Anna Hummingbird. Univ. California ward Arnold. Publ. Zool. 97: 1-109. PETERS,W. D., & T. C. GRUBB,JR. 1982. An experi- ß& L. L. WOLF. 1979. Ecologyand evolution mental analysis of sex-specificforaging in the of lek mating behavior in the Long-tailed Her- Downy Woodpecker,Picoides pubescens. Ecology mi •_Hummingbird. Ornithol. Monogr. No. 27. 64: 1437-1443. STINSON,C. H., D. L. CRAWFORD,• J. LAUTHNER. 1981. POOLE,E. L. 1938. Weights and wing areas of 143 Sex differences in winter habitat of American speciesof North American birds. Auk 55: 511- Kestrelsin Georgia.J. Field Ornithol. 52: 29-35. 517. TEMELES,E.J. 1985. Sexualsize dimorphism of bird- PUTTICK,G. M. 1981. Sex-related differences in for- eating hawks: the effect of prey vulnerability. aging behavior of Curlew Sandpipers.Ornis Amer. Natur. 125: 485-499. Scandinavica 12: 13-17. WOLF, L. L., F. G. STILES,& F. R. HAINSWORTH. 1976. SCHIPPER,W. J. A., L. S. BUURMA,& P. BOSSENBROEK. Ecologicalorganization of a tropical, highland 1975. Comparativestudy of hunting behaviour hummingbirdcommunity. J. Anim. Ecol.45: 349- of wintering Hen Harriers Circuscyaneus and 379. Marsh Harriers Circusaeruginosus. Ardea 63: 1- 29.

"FederalWildlife Permit Procedures"(vol. 1 of "ControlledWildlife"), compiledby CarolEstes and Keith W. Sessions,is now available.It is a comprehensiveguide to the labyrinth of federal statutesrelating to wildlife. Includesdescriptions of the variousprocedures involved in obtainingand transportingwildlife and for dealing with bird specimens(living or dead). Available for $55.00 from: Associationof Systematic Collections,Museum of Natural History, University of Kansas,Lawrence, Kansas 66045. DISPERSAL IN WHITE-CROWNED SPARROWS: A COMPUTER SIMULATION OF THE EFFECT OF STUDY-AREA SIZE ON ESTIMATES OF LOCAL RECRUITMENT

MICHAEL A. CUNNINGHAM • Departmentof Zoologyand Entomology,Colorado State University, Fort Collins, Colorado 80523 USA

A13STRACT.--Considerationof the problem of samplinga populationwithin a finite study areasuggested that recruitmentresults might be relatedto the size of the studyarea. Spe- cifically, the proportion of recruitsobserved within a defined study area that originate in that area (local recruits)should be a function of the proportion of the population that the study area represents.Computer simulationsof dispersal,using Nuttall's White-crowned Sparrow(Zonotrichia leucophrys nuttalli) as a model, were designedto test this hypothesis, and the effectof territory size and populationsize on this relationship.The simulations confirmedthat localrecruitment is a functionof the proportionof the populationsampled in the studyarea. Additionally, these simulations demonstrated that by holding the distri- bution of dispersaldistances constant, local recruitmentfor a given proportionof the pop- ulationsampled increased with territorysize and populationsize. These latter results,how- ever,are artifacts of the simulationsand have no biologicalsignificance. Finally, the proportion of local recruits predicted for a natural population was remarkably close to that obtained from fieldwork.I suggestthat conclusionsabout dispersal and populationstructure should not be drawn from proportionsof local recruitmentwithin finite studyareas, and require more detailedanalysis of dispersal.Received 7 December1984, accepted 11 July1985.

ANALYSESof dispersal(Howard 1960), and the tall's White-crowned Sparrows (Zonotrichialeu- attendantgene flow (Wright 1940, 1943, 1946, cophrysnuttalli) is an example of this latter 1951; Murray 1967; Crumpacker and Williams approach. 1973; Powell et al. 1976; Endler 1977), are im- Estimatingdispersal from local turnover has portant to understandingpopulation structure an inherent problem: the proportion of "local" and evolutionaryprocesses (Barrowclough 1978, recruits (i.e. those produced within the study Greenwood and Harvey 1982, Shields 1982). area) is related to the proportion of the breed- Conclusionsabout dispersalare often not de- ing population sampledin the study area. Con- rived from observedzygote-to-zygote dispersal sider a small hypothetical study area including distances.Instead, a study area is defined with- only a single territory, and a dispersalrule that in a population and an attempt is made to mark individuals cannot settle within their natal ter- all individualsproduced within that study area. ritory. Clearly, the proportion of recruitsto the In subsequentbreeding seasons,the study area study area that originated there (local recruits) is censusedto determine what proportion of will be zero. At the other extreme, a study area the breeding adults originated within the study encompassingthe entire breeding population area. From such "local turnover," inferences are will necessarily contain 100% local recruits. drawn aboutthe magnitudeof dispersalamong Most study areaslie somewhere between these study populations(Petrinovich et al. 1981). Pe- extremes, representing some portion of the trinovich and Patterson's work (1982; see also population of interest. Baptistaand Morton 1982) on the stability, re- To demonstratethe relationship between lo- cruitment, and population structure of Nut- cal recruitment and the proportion of a popu- lation sampled,I designeda computer simula- tion of dispersal and sampled defined study • Presentaddress: Department of Zoology, Ohio areas.To make this simulation representative State University, Lima Regional Campus,4240 Cam- of a natural population, I drew the necessary pus Drive, Lima, Ohio 45804 USA. parameters from the literature on the demog- 79 The Auk 103:79-85. January1986 80 MICHAELA. CUNNINGHAM [Auk,Vol. 103 raphy and dispersalof Nuttall's White-crowned through the six neighboring territories. Dispersal Sparrow. The general conclusionsfrom this distanceswere selectedprobabilistically from a two- analysisshould, however, be applicableto es- parameter gamma distribution fitted by Baker and sentially any population of dispersingorgan- Mewaldt (1978) to dispersaldata for both sexesfrom White-crownedSparrow populations at Point Reyes, isms.Finally, I simulatedPetrinovich and Pat- California.For this distribution,gamma = 0.89352and terson's(1982) study conditionsto see if the the two parameterswere calculatedto be alpha (c0 = proportion of local recruits predicted from my 0.6 and beta (/•) = 366.8. Alpha is a shapeparameter model would be comparableto their empirical and beta is a scale parameter, where the mode = result. and the mean = (c• + I)/• (Johnson and Kotz 1970, Hastingsand Peacock1975). Dispersal distances were COMPUTER SIMULATIONS calculated in meters and converted into territorial units by dividing that distanceby the territorial di- Simulationswere programmedin ApplesoftBASIC ameter.Hexagonal polygons have two diameters,side- on an Apple II+ computer. The first simulation was to-side (minimum) and corner-to-corner (maximum). designed to determine the relationship between lo- All simulationsin this study used minimum diame- cal recruitmentand the proportion of a population ters for territory diameters. sampled,and to determinethe effectof territory size Dispersingindividuals were moved to the target on that relationship, for a given dispersaldistribu- territory, determinedby the dispersaldirection and tion. The population was defined as a squarearray of distance,using the hex addressmanipulation. If the 625 territories, 25 on a side. This approximatesthe territory was available (not occupiedby a surviving size of a songdialect populationat Point ReyesNa- adult or previously dispersedjuvenile), the dispers- tional Seashore,California (300-800 pairs/dialect, ing juvenile occupiedthat territory; if the territory Baker 1981;pers. obs.).The population was storedin was occupied, the dispersing juvenile searched the computer memory as a two-dimensional array and six neighboring territories and occupied the first treatedas a hexagonalarray by manipulationof the availableterritory encountered.Once the dispersing addresses(Page and Didday 1980).The advantageof juvenile settled, it was noted whether it had origi- the hexagonal arrangement was that each territory nated from within or from outsidethe study area. (with the exceptionof boundaryterritories) was sur- Following this, or if the disperserwas not able to rounded by six neighboring territorieswhose centers settle in the target territory or one of the six neigh- wereequidistant from the centerof the territorythey boring territories,a new territory of origin was ran- bounded. In nature, the spatial arrangement of ter- domly selectedand dispersalof a new individual was ritories may be highly variable, but an approxima- initiated in like manner. The populationsdefined in tion to hexagonalpacking is common(Nice 1937,Lack these simulations were finite, and the boundaries 1966, Leuthold 1966, Grant 1968, Southern 1970, Bar- were absorbing;i.e. if a dispersingjuvenile contacted low 1974,Wilson 1975,Ligon and Ligon 1982).Study a boundaryhe waseliminated, and dispersalof a new areaswere defined as squarearrays centered within juvenile was initiated. The number of juvenilesthat the populationand representedsamples of the pop- could dispersefrom a given territory was not limited ulation ranging from approximately5% to 95%at ap- (White-crowned Sparrows may produce more than proximately 5% intervals. one clutch/seasonwith up to 4-5 young/clutch), but Each simulation began by establishing the status on each iteration the territory of origin was again of all territories within the population. A survivor~ chosenrandomly. Dispersaliterations continued un- ship test determined whether each territory was oc- til all territorieswithin the studyarea were occupied. cupiedby a surviving adult or was availableto a re- The study area was then censusedto determine the cruit. Survivorshipwas 50% for eachadult generation proportionof recruitsthat originatedwithin the study (Mewaldt 1964,Baker et al. 1981).For simplicity,each area. territory wasoccupied by a singleindividual rather This simulation used four different territory di- than a pair [simulating pairs would have required ameters (20, 40, 80, and 160 m) and 16 study-area running the program twice (once for each sex, as- proportions(from about 6% to 92% at approximately sumingthe sexesdisperse independently) and aver- 5% intervals). The simulation was repeated 20 times aging the resultsto obtain a single proportion],and for each of the combinationsof territory diameter recruitment was limited to first-yearbirds, i.e. adult and study-areaproportion. This rangeof territory di- movement was not considered. ameterswas chosenbecause it bracketsthose report- Dispersalbegan by randomly selectinga territory ed for Nuttall's White-crowned Sparrow. The small- of origin from the entire population,then determin- est territories were reported in San Francisco and ing a targetterritory by randomlygenerating a dis- averaged1,127 m2 (Pattersonand Petrinovich 1978, persal direction and distance.Dispersal directions Petrinovich and Patterson 1982). The minimum di- were chosenrandomly to be alongone of the six axes ameter of hexagonalterritories of this area would be January1986] Study-areaSize, Local Recruitment, and Dispersal 81

TABLE1. Power-curveregression coefficients for dis- persalsimulations.'

Popula- Median tion Terri- terri- size tory tories (terri- dia- dis- tories) meter persed a b 625 20 18.75 1.0156 0.6942 625 40 9.38 1.0175 0.6090 625 80 4.69 1.0220 0.5082 625 160 2.34 0.9797 0.2932 196 b 40 9.38 1.1349 0.8432 aPower function: y = axb;median dispersaldis- tance = 375 m; dispersalparameters: alpha = 0.60, beta = 366.8, gamma = 0.89352. bSimulation comparable to the field study by Pe- trinovich and Patterson (1982).

the resultsfrom the previous simulation, which used Fig. 1. Proportion of local recruits as a function the same dispersaldistribution and territory diame- of the proportion of the population sampled in the ter, I was also able to determine the effect of popu- studyarea. Eachcurve representsthe plotted means lation size on the proportion of local recruits ob- of 20 runs for eachof 16 study-areaproportions, for servedwithin a study area. territory diametersof 20, 40, 80, and 160 m. The pop- ulation size was 625 territories,and the dispersalpa- rameterswere: alpha = 0.6, beta = 366.8, and gam- RESULTS AND DISCUSSION ma = 0.89352. Two clear relationships emerged from the first simulation. First, the proportion of local approximately36 m. The averageterritories at Point recruits is an increasing function of the pro- Reyesrange in sizefrom 2,360m 2 (Ralph and Pearson 1971) to 4,452 m2 (Chamberlain 1972). These figures portion of the breeding population that the produceminimum diametersof hexagonalterritories study area encompasses(Fig. 1). The shape of in the range of 52-72 m. these curvessuggests a simple power function A second simulation was performed to compare of the form: y = axb, where a and b are con- the proportion of local recruits predicted from my stants.Theoretically, for this particular prob- model with an empirical result (Petrinovich and Pat- lem, y (the proportion of local recruits) should terson 1982).The Presidiostudy area of Petrinovich equal 1.0 when x (the proportion of the popu- and Pattersonrepresented approximately 15% of the lation sampled in the study area) is 1.0. The Presidio dialect area (as determined by measure- simulationconfirms this (Fig. 1). Thus, the con- mentstaken from Fig. 1 of Petrinovichand Patterson stant a should equal 1.0, and the power func- 1982) and contained about 31 territories. Assuming the population density of the entire dialect area was tion reducesto y = xb. Within the range of pa- the same as in the study area, the population of the rameters used in this simulation, the exponent Presidio dialect included about 203 territorial pairs. b should assume some value between 0.0 and For the purposeof simulation, the nearestsquare ar- 1.0. For eachof the four territory diametersused ray of territories, representing the Presidio dialect in this simulation, I fitted power curves to the population,is a 14 x 14 array of 196 territories.Be- mean values of the 20 runs for each study-area causePetrinovich and Patterson did not provide a proportion, and the theoretical point of x = 1.0, dispersaldistribution for their study population, I y = 1.0. The theoretical point of x = 0.0, y = 0.0 assumeddispersal followed that reported by Baker could not be included becauseof logarithmsin and Mewaldt (1978). I used40 m as the territory di- the curve-fittingprogram. In all four casesthe ameter becauseit approximated the diameter of a Presidioterritory (for a hexagonalterritory of 1,127 coefficienta was within 3% of the expectedval- m2, min. diam. = 36.1, max. diam. = 41.7). The sim- ue of 1.0 (Table 1). ulation was repeated 20 times for each of 11 study- The second relationship suggested by this area proportions (from about 5% to 86% at approxi- simulation is that for a given distribution of mately5-10% intervals).Finally, by comparisonwith dispersaldistances and a given proportion of MICHAELA. CUNNINGHAM [Auk,Vol. 103

1.o_ local recruits),but they should not changethe

.9_ general trends in the relationships described here. .8_ The result of decreasingdispersal (higher lo- cal recruitment) with increasing territory size (Fig. 1), in addition to being an artifact of the simulation, probably does not apply to natural populations.Dispersal is a discreterather than a continuousprocess because territorial species dispersefrom territory to territory and not sim- ply absolutedistances (Shields 1982).In natural populations,we would expectthat as territory size increasesso will absolute dispersal dis- tances, because individuals are forced to move longer distanceswhile crossingthe samenum- ber of territories. With regard to population PROPORTION OF POPOLAT•ON SAMPLED 8Y STOO¾ AREA size, Nuttall's White-crowned Sparrow pre- sents some special problems becausethis sub- Fig. 2. Proportionof local recruitsas a function speciesis subdividedby song dialects.In this of the proportionof the populationsampled in the study I assumedthat song dialectsrepresent study area. The curvesrepresent the plotted means of 20 runs for eachof 16 (upper curve)and 11 (lower distinct populations; however, this has been a curve)study-area proportions, for populationsof 625 point of somedebate (for review seeBaker and (upper curve) and 196 (lower curve) territories.The Cunningham 1985). If song dialects are not territory diameterwas 40 m, and the dispersalparam- (relatively) distinct populations, the actual eters were: alpha = 0.6, beta = 366.8, and gamma= population size would be much larger. This 0.89352.The upper curve is from the first simulation would change the position of the recruitment and is also shown in Fig. 1. curve, as demonstratedby the secondsimula- tion (Fig. 2). In natural populations the dis- persal distribution itself is an important deter- the populationsampled in the study area, the minant of the "effective" population size proportion of local recruits increaseswith in- (Wright 1943, 1951; Baker 1981); for dialect creasingterritory diameter (Fig. 1). That is, for species,such as the White-crowned Sparrow, a given distribution of absolutedispersal dis- the effectivepopulation size will depend upon tances,as the territory diameter increases,the both the dispersaldistribution and the effectof number of territories disperseddecreases (Ta- dialect boundaries (Baker and Mewaldt 1978, ble 1). The secondsimulation, with dispersal Baker 1981). distribution and territory diameter held con- The more important objective of the second stant, indicated that, for a given proportion of simulation was to comparethe proportion of the population sampled,the proportion of local local recruits predicted from the model with recruitsincreased as the size of the population that observedpreviously (Petrinovich and Pat- increased (Fig. 2). Both of these results are terson 1982). Petrinovich and Pattersonreport- purely artifactsof the simulations,and for the ed that only 16%of all recruitswere produced same reason.As the area occupiedby a popu- within their study areas. Elsewhere (Petrino- lation increases,either becauseterritory size or vich et al. 1981) they concluded,from this low population size increases,the study area must proportion of local recruits, that there must be likewise increaseto sample the same propor- significant cross-dialectdispersal. They count- tion of the population. If the dispersal distri- ed adults that moved into their study areasas bution remainsconstant and the study-areasize recruits, as well as first-year birds. In my sim- is increased,then fewer of the individuals pro- ulations,I consideredonly juveniles asrecruits. duced within the study area dispersebeyond For comparisonI usedonly the first-yearbirds its boundaries.It is important to keep thesere- (nestlings, fledglings, and first-year brown lationships in mind when considering these crowns) from Petrinovich and Patterson's(1982) simulations,as they influencethe absoluteval- data. Of the 182 first-yearbirds that settled in ues of the dependent variable (proportion of their study areas,51 had been banded there as January1986] Study-areaSize, Local Recruitment, andDispersal 83 nestlings, giving a local recruitment of 28%. prising about the findings of Petrinovich and These recruitment figures are combined data Patterson, and inferences about interdialect ex- from two study sites, the Presidio and Twin change made from these data (Petrinovich et Peaks, but the authors stated that the data were al. 1981)might be invalid. In fact, neither these treated separatelyonly in those instancesin simulations nor the results of Petrinovich and which the results were different for the two Pattersondirectly addressthe issueof interdi- study areas.I assumedthat the data from the alect dispersal.Some qualifications are in order Presidio and Twin Peaks were similar. It is dif- because there are a number of complicating ficult to determine what proportion of the San factorsthat render this simulation only an ap- Franciscodialect population the Twin Peaks proximationof the populationthat Petrinovich study area represented,because the San Fran- and Patterson studied. ciscodialect population is highly fragmented Becauseof the lack of detailed dispersaldata in an urban area, and the full extent of the from the Presidio population, I used the dis- dialect is not shown on the map provided (Pe- persal distribution parametersfrom the Point trinovich and Patterson 1982). However, the Reyespopulations (Baker and Mewaldt 1978). Presidio dialect area is indicated (see also Pe- This distribution is given in absolutedistances trinovich et al. 1981), and the Presidio study rather than number of territories moved. The area representedabout 15%of the dialect area. territories at the Presidio tend to be only 25- The resultsof the secondsimulation are plot- 50% as large as those at Point Reyes. If birds ted in Fig. 2 (lower curve),and the power-curve disperseby territorial units rather than by ab- regressioncoefficients are given in Table 1. This solute distance(Shields 1982), and if dispersal regression was not as close to the theoretical is similar at the Presidio and Point Reyes, then asymptote,where the coefficienta = 1.0, as the the dispersal distancesshould be proportion- regressionsin the previous simulation (Table ately shorter for the Presidio population. A dis- 1). Using the simplified power function, y = xb, tribution of shorter dispersal distancesshould the predicted proportion of local recruitswas increasethe proportion of local recruits for a 20%for a studyarea containing 15% of the pop- given-size study area. ulation, but this curve did not fit the data well The physical habitat at the Presidio is quite and yielded predictedvalues almost 12% below different from that at Point Reyes (Baker and that of the fitted power curve. The power func- Mewaldt 1981;pers. obs.).In the Presidio,suit- tion usingthe calculatedcoefficient (a) of 0.1349 able breeding habitat is patchy, crossedby ma- provided a better fit than the simplified func- jor highways, and fully within an urban area, tion, but the fit still was not good becauseit while Point Reyesprovides larger tractsof con- predictedvalues less than the observedmeans tinuous breeding habitat in a natural "wilder- for x in the range of 8-41% and greater than ness" area. Thus, birds at the Presidio common- the observedmeans for x in the range of 51- ly may have to disperse across areas of 100%. Using this fitted power curve, the pre- unsuitablehabitat before locating an available dicted value for a study area containing 15%of territory. A patchy distribution of breeding the population was 23%. The observed means habitat would tend to increase the absolute dis- (from 20 computer runs) for study areas con- persal distances,thus decreasingthe propor- taining 13% and 18% of the population were tion of local recruits,perhaps counteractingthe 21%and 35%,respectively. Linear interpolation effectof smaller territory size. between these values for a study area contain- The results of my simulations support the ing 15% of the population yields 27% local re- argument that the proportion of local recruits cruits. observedwithin a study area depends on the All three of the predicted values for the pro- proportion of the breeding population encom- portion of local recruits,from a study area con- passedby that study area.This relationship can taining about 15% of a population of 196 ter- be describedby a simplepower function.While ritories, were close to the 28% local recruits my simulation only roughly approximatesthe observedby Petrinovich and Patterson(1982), conditionsat the Presidio, it suggeststhat the the third being remarkably close. This close proportion of local recruits observed there correspondencebetween observedvalues of lo- might not be unexpectedly small; it does not cal recruitment and those predicted by my necessarilyindicate that "a sizableproportion model suggeststhat there may be nothing sur- of banded nestlingsdispersed out of the Pres- 84 MICHAELA. CUNNINGHAM [AukßVol. 103

idio" (Petrinovichet al. 1981).In fact, the pro- CHAMBERLAIN,B. 1972. Observationsof breeding portionof localrecruits observed within a study White-crownedSparrows. Point ReyesBird Obs. area is a function of the size of the study area Newsletter No. 23. in relation to the dispersaldistribution itself, CRUMPACKER,D. W.ß • J. S. WILLIAMS.1973. Densityß which is the important determinant of the "ef- dispersion,and population structurein Drosoph- fective" population size. Reliable conclusions ila pseudoobscura.Ecol. Monogr. 43: 499-538. ENDLER,J. A. 1977. Geographic variation, specia- about dispersaland population structure can- tion, and dines. PrincetonßNew Jersey,Prince- not be made from proportionsof local recruit- ton Univ. Press. ment in finite study areas. To evaluate the GRANTßP.R. 1968. Polyhedral territoriesof animals. structure of natural populations, detailed anal- Amer. Natur. 102: 75-80. yses of dispersal are needed that consider the GREENWOODßP. J., •r P. H. HARVEY. 1982. The natal territorial nature of the population, the effects and breeding dispersalof birds. Ann. Rev. Ecol. of social interactions(e.g. dialect effects),and Syst. 13: 1-21. the distribution of suitable habitat. HASTINGS,N. A. J., & J. B. PEACOCK.1975. Statistical distributions. New York, Halstead PressßJohn ACKNOWLEDGMENTS Wiley & Sons. HOWARD, W. E. 1960. Innate and environmental I acknowledge L. Petrinovich and T. L. Patterson, dispersal of individual vertebrates.Amer. Midi. whosework wasthe stimulusfor this analysis.While Natur. 63: 152-161. I have been critical of their interpretations,I admire JOHNSONßN. L., & S. KOTZ. 1970. Distributions in the important empirical contribution of their exten- statistics: continuous univariate distributions-- sivework. I am gratefulto M. C. BakerßC. R. Tracy, 1. Boston,Houghton Mifflin. R. M. Caseyßand W. M. Shieldsfor readingand com- LACK,D. 1966. Population studiesof birds. Oxfordß menting on this paper.! am especiallygrateful to M. Oxford Univ. Press. C. Baker and C. R. Tracy for their advice and en- LEUTHOLD,W. 1966. Territorial behaviorof Uganda couragement,and to C. R. Tracy for allowing me to kob. Behaviour 27: 255-256. pursue this exercisein partial fulfillment of the Ph.D. LIGON,D. J., & S. H. LIGON. 1982. The cooperative preliminary examination requirement at Colorado breeding behavior of the Green Woodhoopoe. StateUniversity. Finally, I would like to expressmy Sci. Amer. 247: 126-134. appreciation to L. Petrinovich and D. DeSante for MEWALDT, L. R. 1964. Effects of bird removal on a their helpful reviews of this paper. winter populationof sparrows.Bird-Banding 35: 184-195.

LITERATURE CITED MURRAY,B.G. 1967. Dispersal in vertebrates.Ecol- ogy 48: 975-978. BAKER,M. C. 1981. Effective population size in a NICE, M. M. 1937. Studies in the life history of the songbird:some possibleimplications. Heredity Song Sparrow. Trans. Linnaean Soc. New York 46: 209-218. 4: 1-247. -, & M. A. CUNNINGHAM.1985. The biologyof PAGE,R., & R. DIDDAY. 1980. Fortran 77 for humans. bird song dialects.Behav. Brain Sci. 8: 85-133. St. Paul, Minnesota, West Publ. Co. ß •r L. R. MEWALDT.1978. Song dialectsas PATTERSON,T. L., & L. PETRINOVICH.1978. Territory barriers to dispersal in White-crowned Spar- size in the White-crowned Sparrow (Zonotrichia rowsßZonotrichia leucophrys nuttalli. Evolution 32: leucophrys):measurement and stability. Condor 712-722. 80: 97-98. ß& •. 1981. Responseto "Songdialects PETRINOVICH,L., •r T. L. PATTERSON.1982. The White- as barriers to dispersal: a re-evaluation." Evolu- crowned Sparrow: stabilityß recruitment, and tion 35: 189-190. population structure in the Nuttall subspecies , --, & R. M. STEWART.1981. Demogra- (1975-1980). Auk 99: 1-14. phy of White-crownedSparrows (Zonotrichia leu- ß--, & L. F. BAPTISTA.1981. Song dialects cophrysnuttalli). Ecology 62: 636-644. as barriers to dispersal:a re-evaluation. Evolu- BAPTISTA,L. F., •r M. L. MORTON.1982. Songdialects tion 35: 180-188. and mate selection in montane White-crowned POWELL,J. R., T. DOBZHANSKY,J. E. HOOKß & H. E. Sparrows.Auk 99: 537-547. WISTRAND.1976. Geneticsof natural popula- BARLOW,G.W. 1974. Hexagonal territories. Anim. tions. XLIII. Further studieson ratesof dispersal Behav. 22: 876-878. of Drosophilapseudoobscura and its relatives. Ge- BARROWCLOUGH,G. F. 1978. Sampling bias in dis- netics 82: 493-506. persalstudies based on finite area.Bird-Banding RALPH, C. J., •r C. A. PEARSON. 1971. Correlation of 49: 333-341. age,size of territory,plumageß and breeding suc- January1986] Study-areaSize, Local Recruitment, and Dispersal 85

cessin White-crownedSparrows. Condor 73: 77- WRIGHT,$. 1940. Breedingstructure of populations 80. in relation to speciation.Amer. Natur. 74: 232- SHIELVS,W. M. 1982. Philopatry, inbreeding, and 248. the evolution of sex. Albany, State Univ. New ß 1943. Isolationby distance.Genetics 28: 114- York Press. 138. SOUTHERN,I'•. N. 1970. The natural control of a pop- 1946. Isolation by distance under diverse ulation of Tawny Owls (Strixaluco). J. Zool. Lon- systemsof mating. Genetics31: 39-59. don 162: 197-285. ß 1951. The genetical structure of popula- WILsota,E. O. 1975. Sociobiology,the new synthe- tions. Ann. Eugenics15: 323-354. sis. Cambridge, Massachusetts,Harvard Univ. Press.

Repository for Voucher Photographsat VIREO

Photographsto documentcritical aspectsof researchmethods and resultsin paperspublished in TheAuk canbe depositedwith VisualResources for Ornithology(VIREO) for long-termarchival care. VIREO accession numbersfor thesephotographs can then be publishedto enable subsequentaccess to the photographsfor comparison,research, and teaching. Voucherphotographs should documentcritical and unique featuresof the publishedwork, and normally shouldnot be photographspublished with the paper. Photographsof habitatand study-siteslides are of exceptionalimportance for ecologicaland behavioralstudies. Specific plumages, soft-body part colors,or morphologicalcharacters central to methodsand resultsalso would be appropriate,as would photographs illustratingnovel techniques. Preferably,authors will depositoriginal photographswith VIREO; if needed,a reproduction-qualitydu- plicateis then returned.If the photographermust retain the original, then VIREO will placea reproduction- quality duplicatein archival storage.In this caseVIREO requirestemporary access to the original for dupli- cation. Authorsinterested in using this serviceshould contact VIREO immediatelyafter being notified of the acceptanceof their papersfor publication in TheAuk: VIREO, Academy of Natural Sciences,19th and the Parkway, Philadelphia, Pennsylvania19103 USA (215-299-1069). January1986] ClutchSize in Motmots 13

B., J. J. ReyesR., Wilfrido ContrerasD.) of Direcci(•n KOEt•IC,W. D. 1982. Ecologicaland social factors Generalde la FaunaSilvestre, Secretarla de Agricul- affectinghatchability of eggs.Auk 99: 526-536. tura y RecursosHidr•ulicos (M•xico, D.F.). Officials LACK,D. 1947. The significanceof clutchsize. Ibis of the M•rida office of Divisi6n Hidrom•trica, SARH, 89: 302-352. kindly gave us accessto weather data. George and MARTIN, M. W., & R. F. MARTIN. 1985. Nestling JanetCobb, Joann Andrews, and EdwardKurjack were feeding schedulesof Turquoise-browedMot- generous hosts in Yucatfm. We thank Anne Clark, mots in Yucatan, Mexico. Wilson Bull. 97: 372- Paul Mason, and two anonymousreviewers for use- 374. ful criticismsof a draft of this paper. Partial support OREJUELA,J. E. 1977. Comparativebiology of Tur- for this study was provided by The Explorers Club quoise-browedand Blue-crowned motmots in the and by the TexasMemorial Museum-The University YucatanPeninsula, Mexico. Living Bird 16: 193- of Texasat Austin.The Museumof Zoologyof Lou- 208. isianaState University provided computertime. SCnP•EmER,R. W., & E. A. SC•P•EmER.1983. The world accordingto E1Nifio. Abstr., 101ststated meet- LITERATURE CITED ing of Amer. Ornithol. Union, New York. Scot'r, P.E. 1984. Reproductionof the Turquoise- CLARIC,A. B., & D. S. Wig,SOtS. 1981. Avian breeding browed Motmot in Yucat•n, Mexico. Unpub- adaptations:hatching asynchrony,brood reduc- lished M.A. thesis, Austin, Univ. Texas. tion, and nest failure. Quart. Rev. Biol. 56: 253- --, ,•' R. F. MARTIN. 1983. Reproductionof the 277. Turquoise-browedMotmot at archaeological CONOVER,W.J. 1980. Practical non-parametric sta- ruins in Yucat•n. Biotropica 15: 8-14. tistics.New York, John Wiley & Sons.

The Frank M. ChapmanMemorial Fund givesgrants in aid of ornithologicalresearch and alsopostdoctoral fellowships. While there is no restriction on who may apply, the Committee particularly welcomes and favorsapplications from graduatestudents; projects in gamemanagement and the medicalsciences are seldom funded.Applications are reviewed oncea year and should be submittedno later than 15 January-with all supporting material. Applicationforms may be obtainedfrom the Frank M. Chapman Memorial Fund Committee,% JaneConnelly, The AmericanMuseum of Natural History, CentralPark Westat 79th Street, New York, NY 10024-5192. Dr. Mary McKitrick was appointedChapman Fellow for the period May 1985 through May 1986. She is studyingthe significanceof individual variation in M. flexor crurislateralis in the Tyrannidae. Dr. Nina Pierpontwas appointedChapman Fellow for the period October1985 through October1986. She is studyingthe evolution of diversity in woodcreepers(Aves: Dendrocolaptidae). Chapmangrants during 1985,totalling $25,482with a mean of $499.65,were awardedto: David J. Ander- son,booby breeding and feedingand marineresources in the Galapagos;John Anderson, tests of cooperative foragingin the White Pelican;Dr. BonnieBowen, genetic structure and paternityin communallybreeding jays;Angelo Capparella,effects of riverine barrierson gene flow in Amazonianforest undergrowth birds; William J. Carmen,social behavior and ecologyof the CaliforniaScrub Jay; Richard Casey, the effectof social interactionduring the criticalperiod of songlearning in White-crownedSparrows; Peng Chai, avian feeding behaviorand its implicationsfor the evolution of neotropicalbutterfly mimicry; JackClinton-Eitniear, Ca- thartesburrovianus: movement patterns and systematics;Nonie Coulthard,the White-throatedBee-eater (Mer- opsalbicollis): a study of cooperativebreeding in relation to ecologicalfactors; Marilyn England,breeding biologyof the Northern Harrier on Long Island,New York; ChristianErard, systematics of African Musci- capidae;Mark A. Finke, male parental care and the evolution of reproductiveeffort: an experimentaltest; Dr. JonFjeldsa, museum work for a field guideand biogeographicanalysis of the Andeanavifauna; Ronald C. Gaines,nest site tenacity and philopatry of the FerruginousHawk; Steven M. Goodman,museum tour for the birds of Egypt and project;Frederick P. Greene, determinantsof guild structureamong insectivorous birds;Jeffrey G. Groth,systematics of the Red Crossbillin westernNorth America,using vocal, morphological and electrophoreticdata; John M. Hagan, colonialnesting in Ospreys;Dionyssia Hatzilacos, breeding and feeding biology of the White Pelican (Pelecanusonocrotalus) nesting at Lake Mikra Prespa--measuresfor protection;P. Hibbard,John P. Dunning, LaurieM. McKean,and RichardBowers, the efficiencyof foraging behavior:an experimentalapproach; Geoffrey E. Hill, the reproductiveconsequences of sub-adultplumage

(continuedon p. 22) 22 YUTAKAWATANUKI [Auk, VoL 103 for at least30 days.Adult Leach'sStorm-Petrels, fish, fish, 1,500; crab with shell, 700; shellfish with shell, shellfish, crabs, and chickens were the main food 500; chicken, 3,000 (Brisbin 1968, Hunt 1972, Sibly items.The following equationwas used to determine and McCleery 1983).R•s were assumedto be the same the total wet weight of the ith food: W• = (D• x R• x as those of Black-tailed Gulls (Larus crassirostris)and TER)/(• D•R.C,),where D, = dry weight of the ith food, the following: adult Leach'sStorm-Petrel, 5.9; fish, Ri = regurgitation coefficient of the ith food (show- 317.3; crab, 6.7; shellfish, 2.2; chicken, 30.0 (Watanuki ing the ratio betweenwet weight of intake and dry 1984). The number of petrels killed by gulls was weight of pellets), TER = total energy requirement calculatedby dividing the total petrelwet weight by of the gulls for 5 days,and C• = caloricvalue of the the averageindividual adult weight (48 g, Watanuki ith food. Cis (kcal/kg wet weight) were assumedto 1985b).Seasonal changes in the gulls' diet composi- be the following: adult Leach'sStorm-Petrel, 2,600; tion are shown in Fig. 7.

(continuedfrom p. 13) in male Black-headedGrosbeaks; Adan (Hussein)Isack, biology of the GreaterHoneyguide (Indicator indica- tor); Dr. Pedro Jordano,pattern of fruit use by wintering frugivorous birds and their implications for bird- dispersedplants in Mediterranean habitats;Frank J. Joyce,nest site selectionby three passerineassociates of Hymenoptera; Donald M. Kent, foraging strategiesof Snowy Egrets (Egrettathula) in the salt marshesof Massachusetts;Roni King, winter territoriality in migratory European Robins (Erithacusrubecula)--habitat selectionand winter survivorship;Francis R. Lambert,co-adaptation between frugivorous birds and fig trees in Malaysianlowland forest;Jeanette L. Lebell, microgeographicvariation in the flight whistle of the Brown- headed Cowbird; David E. Manry, a preliminary study of blood and feather pulp proteins in two South Americanibis (Threskiornithidae)species; Jean-Louis Martin, the population structureof Paruscaeruleus L. (Aves):geographical variation and speciationin the Mediterranean region; Jon Miller, breeding distribution and origin of Water Pipits in the Sierra Nevada;David Morimoto, effectsof forest fragmentationon avian communitystructure and species-habitatrelationships in the southeasternMassachusetts Pine Barrens;Jay Pitocchelli,speciation in the genus Oporornis;Richard O. Prum, courtshipbehavior and ecologyof Masius chrysopterusin Ecuador;Dr. Michael R. W. Rands,the breedingbehavior and habitatof the Arabian Bustard, Ardeotisarabs; Pamela C. Rasmussen,relationships of Fuego-PatagonianBlue-eyed Shags; Mark D. Reynolds, socialbehavior of Yellow-billed Magpies;Dr. Gary Ritchison,the significanceof song repertoiresin the Northern Cardinal; Jeffrey A. Schwartz, vocal similarity in mated Ring-billed Gulls as an aid in parental recognitionby young; Dr. Ron Scogin, floral color and hummingbird vision; Peter E. Scott, the nesting ecologyof desert hummingbirdsin relation to the pollination ecologyof certain nectar plants; Patricia Serrentino,the breedingecology and behaviorof the Northern Harrier (Circuscyaneus) in CoosCounty, New Hampshire;Laurie J. Stuart-Simons,food limitation in birds;Richard R. Snell, hybridization,isolation and speciesrecognition in arcticgulls; Carol Spaw, thick-shelled eggs and their evolutionaryimplications; Charles Sullivan, nest behaviorand developmentof young in the JabiruStork; Dr. Kimberly A. Sullivan, energetics and the development of time-budgeting;Richard John Watling, investigation of statusof Ogea Flycatcher and Blue-crownedLory; Dr. G. CauseyWhittow, water lossfrom eggsof Great Frigatebird(Fregata minor); Dr. David Winkler, a general model of parental care with experimentaltests on the Tree Swallow; John L. Zimmerman, Ortstreuein Henslow's Sparrows (Ammodramushenslowii) and movementsin responseto spring burning of tallgrassprairie. January1986] Reproductionin Hybrid Flickers 51

DOWLING, T. E., & W. S. MOOREß 1984. Level of re- BRENNER,& D. H. BENTß 1975. SPSS, statistical productiveisolation between two cyprinid fish- packagefor the socialsciences, 2nd ed. New York, es,Notropis cornutus and N. chrysocephalus.Copeia McGraw-Hillß 1984: 617-628. REMINGTON,C.L. 1968. Suture-zonesof hybrid in- ß & -- 1985. Evidence for selection teractionbetween recently joined biotas.Pp. 321- againsthybrids in the family Cypfinidae (genus 428 in Evolutionarybiology, vol. 2 (TßDobzhan- Notropis).Evolution 39: 152-158. sky, M. K. Hecht, and W. C. Steere, Eds.). New GILL,F. Bß 1980. Historical aspectsof hybridization York, Appleton-Century-Crofts. betweenBlue-winged and Golden-wingedwar- RICKLEESßR. E. 1980. Geographicvariation in clutch biersß Auk 97: 1-18. size among passefine birds: Ashmole'shypoth- JOHNSON,N. K., & C. B. JOHNSONß1985. Speciation esisß Auk 97: 38-49. in sapsuckers(Sphyrapicus): II. Sympatry, hybrid- RISING,J.D. 1983a. The Great Plainshybrid zonesß ization, and mate preferencein S. ruberdaggetti Pp. 131-157 in Current ornithology, vol. 1 (R. F. and S. nuchalis. Auk 102: 1-15. Johnston,Ed.). New York, Plenum Publ. Corpß KOENIG,W. D. 1982. Ecologicaland social factors --. 1983b. The progressof oriole hybridization affectinghatchability of eggs.Auk 99: 526-536ß in Kansas, USA. Auk 100: 885-897. ß 1984. Geographicvariation in clutch size in SHERMANßA. g. 1910. At the sign of the Northern the Northern Flicker (Colaptesauratus): support Flicker. Wilson Bull. 22: 133-171. for Ashmole'shypothesisß Auk 101:698-706ß SHORT,L.L. 1965. Hybridization in the flickers (Co- LEHRMAN, D. S. 1965. Interaction between internal laptes)of North AmericaßBullß Amer. Mus. Nat. and external environmentsin the regulation of Histß 129: 307-428. the reproductivecycle of the Ring Dove.Pp. 355- 1969. Taxonomicaspects of avian hybrid- 380 in Sex and behavior (F. A. BeachßEd.). New izationß Auk 86: 84-105. York, John Wiley & Sons. 1970. A reply to Uzzell and Ashmole.Syst. MOORE,W.S. 1977. An evaluationof narrowhybrid Zool. 19: 199-201. zones in vertebratesß Quartß Rev. Biol. 52: 263- THORNTHWAITEASSOCIATESß 1964. Average climatic 277. water balancedata of the continents,part VII. ß & D. B. BUCHANANß1985. Stability of the Lab. Climatol. Publ. Climatol. 17: 419-615. Northern Flicker hybrid zone in historicaltimes: WILSON,E. O. 1965. The challenge from related implicationsfor adaptivespeciation theory. Evo- speciesßPp. 7-25 in The geneticsof colonizing lution 39: 135-151. species(H. G. Baker and G. L. Stebbins,Eds.). NIE, N.H., C. H. HULL, J. G. JENKINSßK. STEIN- New Yorkß Academic Pressß

The American Ornithologists'Union solicitsnominations for its Brewsterand CouesAwards. Nominations and supportingmaterials should be sentto Dr. Elsie C. Collias,Department of Biology,University of California, Los Angeles,California 90024.Materials must be receivedbefore ! March 1986.

WisconsinProject Loon Watch is acceptingapplications for its firstannual Sigurd T. OlsonCommon Loon ResearchAward for researchon CommonLoons in the LakeSuperior-Lake Michigan region of the United Statesand Canada.To apply, submita brief (maximum10 pages)description of the proposedresearch programand curriculumvitae to Dr. Paul I. V. Strong,Director, Wisconsin Project Loon Watch,Sigurd OlsonEnvironmental Institute, Northland College,Ashland, Wisconsin 54806 USA no later than ! March 1986. Proposalsby studentsshould be accompaniedby two letters of recommendation.The $1,000 award will be grantedon the basisof the project'spotential to betterunderstand and manageUpper Great Lakes populationsof Common Loons. 78 ETHANJ. TEMELES [Auk,Vol. 103

MILLS, G. S. 1976. American Kestrel sex ratios and SIEGEL,S. 1956. Nonparametricstatistics. New York, habitat separation.Auk 93: 740-748. McGraw-Hill. MORSE, D.H. 1974. Niche breadth as a function of SNEDECOR,G. W., & W. G. COCHRAN. 1967. Statistical social dominance. Amer. Natur. 108: 818-830. methodsß 6th ed. Ames, Iowa State Univ. Press. 1980. Behavioralmechanisms in ecology. SNYDER,N. F. R., & H. A. SNYDER. 1970. Feeding Cambridge, Massachusetts,Harvard Univ. Press. territoriesin the EvergladeKite. Condor 72: 492- MUELLERßH. C., D. D. BERGER,& G. ALLEZ. 1977. The 493. periodic invasions of Goshawks. Auk 94:652 --, & J. W. WILEY. 1976. Sexual size dimor- 663. phism of hawks and owls of North America. Or- NEWTON,I. 1979. Population ecology of raptors. nithol. Monogr. No. 20. Vermillion, South Dakota, Buteo Books. STILES,F.G. 1973. Food supply and the annual cycle PENNYCUICK,C. 1972. Animal flight. London, Ed- of the Anna Hummingbird. Univ. California ward Arnold. Publ. Zool. 97: 1-109. PETERS,W. D., & T. C. GRUBB,JR. 1982. An experi- ß& L. L. WOLF. 1979. Ecologyand evolution mental analysis of sex-specificforaging in the of lek mating behavior in the Long-tailed Her- Downy Woodpecker,Picoides pubescens. Ecology mi •_Hummingbird. Ornithol. Monogr. No. 27. 64: 1437-1443. STINSON,C. H., D. L. CRAWFORD,• J. LAUTHNER. 1981. POOLE,E. L. 1938. Weights and wing areas of 143 Sex differences in winter habitat of American speciesof North American birds. Auk 55: 511- Kestrelsin Georgia.J. Field Ornithol. 52: 29-35. 517. TEMELES,E.J. 1985. Sexualsize dimorphism of bird- PUTTICK,G. M. 1981. Sex-related differences in for- eating hawks: the effect of prey vulnerability. aging behavior of Curlew Sandpipers.Ornis Amer. Natur. 125: 485-499. Scandinavica 12: 13-17. WOLF, L. L., F. G. STILES,& F. R. HAINSWORTH. 1976. SCHIPPER,W. J. A., L. S. BUURMA,& P. BOSSENBROEK. Ecologicalorganization of a tropical, highland 1975. Comparativestudy of hunting behaviour hummingbirdcommunity. J. Anim. Ecol.45: 349- of wintering Hen Harriers Circuscyaneus and 379. Marsh Harriers Circusaeruginosus. Ardea 63: 1- 29.

"FederalWildlife Permit Procedures"(vol. 1 of "ControlledWildlife"), compiledby CarolEstes and Keith W. Sessions,is now available.It is a comprehensiveguide to the labyrinth of federal statutesrelating to wildlife. Includesdescriptions of the variousprocedures involved in obtainingand transportingwildlife and for dealing with bird specimens(living or dead). Available for $55.00 from: Associationof Systematic Collections,Museum of Natural History, University of Kansas,Lawrence, Kansas 66045. January1986] Study-areaSize, Local Recruitment, and Dispersal 85

cessin White-crownedSparrows. Condor 73: 77- WRIGHT,$. 1940. Breedingstructure of populations 80. in relation to speciation.Amer. Natur. 74: 232- SHIELVS,W. M. 1982. Philopatry, inbreeding, and 248. the evolution of sex. Albany, State Univ. New ß 1943. Isolationby distance.Genetics 28: 114- York Press. 138. SOUTHERN,I'•. N. 1970. The natural control of a pop- 1946. Isolation by distance under diverse ulation of Tawny Owls (Strixaluco). J. Zool. Lon- systemsof mating. Genetics31: 39-59. don 162: 197-285. ß 1951. The genetical structure of popula- WILsota,E. O. 1975. Sociobiology,the new synthe- tions. Ann. Eugenics15: 323-354. sis. Cambridge, Massachusetts,Harvard Univ. Press.

Repository for Voucher Photographsat VIREO

Photographsto documentcritical aspectsof researchmethods and resultsin paperspublished in TheAuk canbe depositedwith VisualResources for Ornithology(VIREO) for long-termarchival care. VIREO accession numbersfor thesephotographs can then be publishedto enable subsequentaccess to the photographsfor comparison,research, and teaching. Voucherphotographs should documentcritical and unique featuresof the publishedwork, and normally shouldnot be photographspublished with the paper. Photographsof habitatand study-siteslides are of exceptionalimportance for ecologicaland behavioralstudies. Specific plumages, soft-body part colors,or morphologicalcharacters central to methodsand resultsalso would be appropriate,as would photographs illustratingnovel techniques. Preferably,authors will depositoriginal photographswith VIREO; if needed,a reproduction-qualitydu- plicateis then returned.If the photographermust retain the original, then VIREO will placea reproduction- quality duplicatein archival storage.In this caseVIREO requirestemporary access to the original for dupli- cation. Authorsinterested in using this serviceshould contact VIREO immediatelyafter being notified of the acceptanceof their papersfor publication in TheAuk: VIREO, Academy of Natural Sciences,19th and the Parkway, Philadelphia, Pennsylvania19103 USA (215-299-1069).