Female Site Fidelity and Polygyny in the Blackpoll Warbler (Dendroica Striata)

FEMALE SITE FIDELITY AND POLYGYNY IN THE BLACKPOLL WARBLER (DENDROICA STRIATA)

BONITA C. ELIASON JamesFord Bell Museum of NaturalHistory, Department of Ecologyand Behavioral Biology, Universityof Minnesota,Minneapolis, Minnesota 55455 USA

ABSTR^CT.--From1980 to 1982,8-30% of male BlackpollWarblers (Dendroica striata) studied on Kent Island, New Brunswick,were bigamouseach year (2 = 16.8%).I testedwhether differencesin the qualityof the breedingsituation can give rise to polygynousmatings. Five territory parametersand two male parameterswere usedas measuresof quality. Male arrival timeswere correlatedwith malemating status and the reproductivesuccess of monogamous females.Territories of bigamousmales had morelarge conifersthan did thoseof monogamous males.Territory parametersdid not appear to influence female choices. Strongsite attachment by femalesmay lead them to matebigamously when few malesare availableearly in the seasonnear the females'former nest sites. The reproductivesuccess of secondaryfemales was not differentfrom that of monogamousfemales. Any costsof mating bigamouslymay be offsetby increasedreproductive success associated with early nest initiation and competitiveadvantages conferred by site dominance. Site attachmentand return patternscan accountfor the incidenceof polygyny in this population,but probablycould not give rise to high levelsof polygynyor large harems. More informationis neededto evaluatethe influenceof site fidelity on mating decisionsof femalesin otherpolygynous passerine species. Received 5 November 1985, accepted I May 1986.

RESEARCHon the evolution of polygyny in In the original formulationof the polygyny passerinebirds has focusedon speciesin which threshold model, male quality and territory polygynyis common(see Wittenberger 1979, quality were assumedto be highly correlated Vehrenkamp and Bradbury 1984 for reviews). (Orians 1969). Weatherhead and Robertson Relatively little attention has been directed at (1977, 1979) suggestedthat when male quality understandingthe occurrenceof polygynyat and territory quality are not highly correlated, low levels in primarily monogamousspecies females might choose mates based on male (Ford 1983).I attemptedto identify factorsthat quality alone.Although the detailsof their the- causelow levels of polygyny in a population ory have been disputed(Heisler 1981, Searcy of the BlackpollWarbler (Dendroicastriata). and Yasukawa1981, Wittenberger 1981), their Early researchon mating systemsin territo- ideas stimulated studies of the role of male rial passerine birds demonstrated that the characteristicsas well asterritory parametersin number of matesacquired by a male was cor- influencing female mate choice. related with the resources he defended. This I evaluatedthe role of five territory-quality led to the formulationof the polygynythresh- parametersand two male-qualityparameters in old model, which explainshow differencesin determining mating statusin a population of territory quality can give rise to polygyny in the Blackpoll Warbler. The mating status of territorial passerinebirds (Verner 1964, Verner malescould be explainedby male and territory and Willson 1966, Orians 1969). The model was parameters,but femalesettling datescould not. incorporatedinto a moregeneral theory on the The reproductive successof monogamousfe- evolutionof matingsystems (Emlen and Oring males was correlated with male arrival dates. 1977).In the model,female fitness is presumed The polygyny thresholdmodel assumesthat to be a functionof accessto resources.Bigamy femalesare free to mate with any male. This occurswhen differencesin territoryquality are implies that they survey all opportunities large enough that a female can have higher availableat the time of settling before choos- fitnessif sheshares resources on a high-quality ing. In reality, the tendency to return to a fa- territory of a mated male than if she matesmo- miliar site often constrains the choices of both nogamouslyon a low-qualityterritory. males and females that have bred previously

782 The Auk 103: 782-790. October 1986 October1986] FemaleSite Fidelity and Polygyny 783

(Hinde 1956, Greenwood 1980, Oring 1982). 35 male-yearsof data was obtained.These data rep- Female Blackpoll Warblers tend to breed near resent 25 different males and 22 different females. To their former nest sites. Local sex ratios are distinguishthe matesof bigamousmales, I term the sometimesskewed by differential returns of first female to initiate her nest the "primary" female males and females. Given these two factors, the and the secondthe "secondary"female. If differencesin the quality of the breeding situa- benefitsof early nest initiation may causefe- tion give rise to polygynous matings, variation in males to mate with mated males. I assess the measuresof quality should explain (1) the division costsand benefitsto femalesof mating biga- of malesinto mating-statuscategories, (2) the settling mouslyunder these circumstances.Finally, I patterns of the first females on all territories, and (3) suggestthat high nest successmay make the the reproductive successof females within mating- island a favorableplace to breed, enhancing statuscategories (Vehrenkamp and Bradbury 1984). the importanceof site tenacity in maintaining To get sample sizes large enough to test these pre- a breeding spot there. dictions,it was necessaryto pool the information for 1980-1982.

STUDY AREA AND METHODS Beginning in 1980, I used the following parameters as measuresof territory quality: (1) territory size, The study was conductedin a 35-ha spruce-firfor- (2) coniferdensity, (3) total conifers,(4) large conifer est on Kent Island, New Brunswick,during May- density, and (5) total large conifers.The birds forage August 1979-1982. Kent is an 80-ha island located 20 at midcanopy in conifers up to 10 m tall. Microlepi- km southeast of the Maine-New Brunswick border dopteralarvae that are inside needlesand closedbuds in the Bayof Fundy.The vegetationof the islandhas are an important part of the diet (Eliason unpubl. been describedby McCain (1975).Blackpoll Warblers data). Becauseof these factors, direct assessmentof used only the forestedparts of the island. food on eachterritory proved to be beyond the scope Birdswere netted and marked with unique com- of this project.Male age and arrival dateswere used binations of three plastic color bands and one num- as measuresof male quality. bered aluminum band. Blackpoll Warblers are sex- I determinedterritory size by tracing the outlines ually dichromatic,and therefore males and females of the territories with a plane planimeter. For this were distinguishedreadily. During the 4 yearsof the analysisI used territory maps for the 2-week period study41 adultsand 173nestlings were color-banded. when mostfemales settled on territories.In this pop- For the purpose of analysis, "old" birds are those ulation, more than 90%of foraging by BlackpollWar- known to be 2 years old or older. Four males and 1 biers was done in conifers(Eliason unpubl. data), so female that were banded before 1979 provided some conifer density on the territory was taken as a mea- known-age birds at the outset.Fourteen birds band- sure of availableforaging space.Tree densitieswere ed as nestlingsreturned in subsequentyears, provid- determined by counting trees (by speciesand size ing a group of known-ageyearlings. class)in 10 x 10-m plots. The number of plots used Arrival dates of both sexeswere determined by to sample each territory was proportional to the size patrolling the studyarea daily, looking and listening of the territory, so that sampling intensities were for birds. I use the term "settling date" for the first equal(about 6% of territory area).The third param- day a bird was seen on a territory on which it sub- eter, total number of conifers per territory, is the sequently bred. Territory boundaries were deter- productof territory size and coniferdensity. Forag- mined by noting male singing postsand sitesof in- ing observationsalso indicated that coniferssmaller teractionbetween males. Points were plottedon maps than 5 m tall and 8 cm in diameterat breastheight of the studyarea made from aerialphotographs. Each were usedinfrequently by the birds. The large co- week maps of all territories were drawn by connect- nifer parameterswere calculatedby including only ing observationpoints accumulated during the week. trees above that minimum size. After the territoryestablishment period, I visitedeach Male mating-statuscategories were: bigamous, mo- territory at least once every 2 days for a minimum of nogamous,or unmated. Logisticregression was used 1-2 h to determinemating relationships,monitor in- to constructthe best possiblemodel to explain vari- teractionsbetween mates,and follow reproductive ancein the mating statusof males.The five territory- chronology. quality parametersand two male-qualityparameters A maximum of 15 males held territories on the were used as potential explanatoryvariables. island in any year. The first field seasonwas devoted An explanatoryregression model was first sought to a feasibilitystudy during which the breedingbi- for the responsecategories of mated vs. unmated. ology of 5 males and their mates was studied. The The analysiswas then repeatedwithin the matedcat- breeding habitatis dense,and nesting femalesare egoryfor bigamousvs. monogamous.The procedure secretive. To be reliable, I concentrated on the activ- was performedwith the computerprogram LOGIT ities of 8-12 malesper year in 1980-1982.A total of (K. Larntz and S. Weisberg,Dept. Applied Statistics, 784 BONITAC. ELIASON [Auk, VoL 103

University of Minnesota). With this method, the sta- quality parameterscould explain the variance in fe- tistic being minimized is the deviance (see equation malereproductive success (prediction 3). The polyg- below). The significanceof various modelsis tested yny thresholdmodel allowsone to make predictions by calculatingthe differencebetween the deviance aboutthe relationshipof territory quality and female values associated with each model. This difference is fitnessonly within mating-statuscategories (Vehren- distributedapproximately as X2(kl k2df • where k, = the kamp and Bradbury 1984). I did this analysisfor mo- numberof explanatoryvariables in the modelsbeing nogamousfemales only becausesample sizes of non- compared(McCullagh and Nelder 1983). monogamousfemales were too small. A step-up method of analysiswas used in which Multiple regressionanalyses were performed us- models with one explanatory variable were com- ing the program MULTREG (Weisberg1982). All oth- pared with the baselinedeviance of a model with no er statisticalanalyses were performed using SPSS variables{baseline deviance = 2nxlog,(n•+ no)/n•+ 6000, version 8.3 on the University of Minnesota 2nolog,[(n•+ no)/no],where nl = the number of in- computer system. dividuals in a response category}. Additional vari- As part of the assessmentof possible coststo fe- ables then were added to the best single-variable malesof their mating choices,the numberof feeding model to see if a better fit could be obtained. Multi- trips of both sexesto the nestwere countedduring ple-variable models were deemed better than single- three 30-min periodseach day throughoutthe nest- variable models only if they produced significantly ling period. Means of feeding ratesfor days5-8 post- smaller deviance values than the best model with hatchingwere used to comparethe attentivenessof one fewer variable. birds of different mating status. Nests were located by following femalesduring building and incubation.Nests were checkeddaily RESULTS to determine the date of first laying, number of eggs laid, hatchingdate, and number hatched.The day MATING SYSTEM the first egg of a clutchhatched was designated day 1. Nestlingswere weighed and measureddaily from Bigamy occurredin all years, ranging in fre- hatchingthrough day 8. Youngwere bandedat day quency from 8 to 40% (t = 22.6%) of males of 6. After day 8 nestswere checkedthree timesdaily known mating status (Table 1). Territory pa- to determine the number of young fledged, which rameters were not measured in 1979. The mean was used as a measureof female reproductivesuccess.I did not anticipate second broods and did not level of bigamy for the 3 yearsfor which I have seek them in 1979 and 1980. In 1981 and 1982 terri- territory-quality information was 16.7%.Biga- tories were monitored closely throughout the re- my occurredin first broodsonly; no bigamous nesting period. Season-long reproductive success pair bonds lasted beyond the first brood. No comparisonsrefer to 1981-1982only. monogamousfemales switched mates between Becausefemales do not sing, they are lessconspic- broods.Two femalesswitched after matingwith uousthan males.Female settling dates are thus more bigamousmales for their first broodsto mate difficultto determine.I wasconfident of the accuracy with previously unmated males for a second of settlingdates for femalesthat were engagedin the brood.The frequencyof known unmated males early stagesof courtship when first seen (n = 11). at the time females were involved in their first Eight of these 11 females were the first females to settle on their territories. I performed a correlation broods ranged from 0 to 40% (œ= 17%) over analysis on this sample to test the prediction that the 4 years. I saw no unmated femalesat the female settling dates are correlatedwith territory time females were involved in their first broods. quality. I had a larger sampleof femaleswith known This population exhibits "facultative polyg- first-eggdates than of femaleswith known settling yny" (Ford 1983)because polygyny occursreg- dates,and settlingdates and first-eggdates were pos- ularly, but less than 20% of males have more itively correlatedfor the 11 femaleswhose settling than one mate. dates were known with confidence (r = 0.58, t9 = 2.15, P = 0.06). I performed a multiple regression analysisusing first-eggdates as the dependentvari- TERRITORYAND MALE QUALITY able and the five territory-qualityparameters plus male arrival datesas potential explanatoryvariables. Male matingstatus.--The best explanation of Male age was not included in this or the fitnessanal- the fit of males into mated vs. unmated cate- ysisdescribed below becausetoo few yearling males gorieswas provided by a regressionmodel with were mated. arrival time as the only explanatoryvariable. I also used multiple regressionanalysis to explore The addition of territory-quality measuresdid whethersome combination of the territory-and male- not improve the fit (Table 2). A regression October1986] FemaleSite Fidelity and Polygyny 785

TABLE1. Male mating status.Percentages of males TABLE3. Mean values of territory parameters for in eachmating-status category in eachyear and the malesin three mating-statuscategories: unmated mean percentagesfor the 4 years are shown. The (U), monogamous(M), and bigamous(B). Parame- 35 male-yearsshown representthe historiesof 25 ter abbreviations are defined in Table 2. different males. Mating status Monog- Parameter U M B Mean Year n amous Bigamous Unmated n 6 16 5 1979 5 60 40 0 1980 8 75 12.5 12.5 T$ (ha) 0.8 1.1 1.2 1.1 1981 12 75 8 17 CD (trees/100 m2) 23.9 37.3 48.2 36.3 1982 10 30 30 40 LCD (trees/100 m2)a 14.5 22.8 37.9 23.7 Mean 60 22.6 17.4 TC (trees/ territory) 2,200 3,480 5,150 3,510 TLC (trees/ territory)b 1,310 2,150 4,040 2,320 model with total large conifersplus arrival time One-way ANOVA, F•,24= 3.83, P = 0.04. as explanatoryvariables provided the best ex- One-way ANOVA, F•,24= 9.46, P < 0.001. planation for the monogamous/bigamousdi- chotomy (Table 2). Male arrival dateswere correlated inversely ritories of bigamous males had significantly with total large conifers(r = -0.47, P = 0.01), more total large conifers (Student-Newman- consistent with the idea that males that arrive Keuls test, P < 0.01) than those of other males, but there was no difference between the terri- first settle on the best territories. The mean values of all territory parameterswere smallestfor tories of unmated and monogamous males. unmated males and largest for bigamousmales, Thus, there appearedto be differencesbetween with the values for territories of monogamous the territories of bigamous males and other males. males intermediate (Table 3). The mean values for large conifer density and total large coni- Femalesettling dates.--There was no signifi- fers were significantlydifferent among the ter- cant correlationbetween female settling dates ritories of males of different mating status.Ter- and any of the six breeding-qualityparameters. No significant regressionmodel could be con- structedto explain the variancein first-eggdates using the six breeding-quality parametersas TABLE2. Derlance valuesfrom logisticregression potential explanatoryvariables. analysis.Values for modelscontaining each of the sevenexplanatory variables alone, plus the two best Femalereproductive success.--The only signif- multiple-variablemodels, are shown, along with icant regressionmodel to explain the variance the baseline deviance values for the responsecat- in the season-longreproductive successof mo- egories of mated vs. unmated and bigamous vs. nogamous females was one with male arrival monogamous.Asterisks indicate best models. dates alone (r = 0.68, t•0 = 2.91, P = 0.02). The Responsevariables addition of the territory parametersdid not improve the fit. Mated vs. un- Bigamousvs. Explanatoryvariables mated monogamous SITE FIDELITY AND ARRIVAL SCHEDULES Baseline 26.4 15.8 In all 12 cases where males with known Territory size (TS) 24.2 15.1 Conifer density (CD) 24.0 15.5 breedinghistories returned, the territoriesthey Total conifers (TC) 23.0 10.8 defendedin their secondyear included part of Large conifer density the territories they had defended in the pre- (LCD) 22.7 13.8 vious year. In 11 of 12 caseswhere femaleswith Total large conifers (TLC) 21.8 5.1 known breeding historiesreturned, they set- Male arrival dates (ARR) 5.4* 13.8 tled on or immediatelyadjacent to the territory Male age (AGE) 11.5 14.6 on which they had nested the year before. In ARR + AGE 5.4 13.6 4 cases the mate returned, but in 7 he did not. ARR + TLC 5.3 0.50 x 10 -•* Members of different age and sex classesre- 786 BONITAC. ELIASON [Auk, Vol. 103

Old males

Old females

Yearlingmales 4 * [] 1981 iii!i!•1982 6 40 45 20 X nest site

DAYS AFTER FIRST INDIVIDUAL SIGHTED

Fig. 1. First sighting dates of birds in three age 100 m and sex classes for 1981 and 1982. "Old" refers to birds 2 years old and older. Dates were ordered by using the date the first individual was sighted as day 0 and then numbering days consecutively.Asterisks indicate medians. Fig. 2. Territories (stippled) of the first 3 malesto settle in 1982 superimposedon 12 territories (solid lines) and 6 nest sites (X's) for 1981. turned at different times (Fig. 1). Old males arrived first and yearling males latest, with no overlap in the datesthey began advertising on choices in 1981 as well. There was one case of territories. Femalesarrived over a longer peri- bigamythat involved a female(O/B) that had od. Some old femalesarrived before any year- nested on the same territory in 1980, and a ling males.The median date of old female ar- yearling female (M/Y). Although O/B arrived rival was earlier than that of yearling males after M/Y, she initiated her nest first. The male (Fig. 1). made more feeding trips to her nest than to More old females than old males returned in that of M/Y (3.2 vs. 2.7 trips.nestling-•-h-•), someyears, although the return rates of old and the male fed her fledglingsbut not those birds were similar for males (48%) and females of M/Y. In the bigamous mating described (49%) when averagedover the 4 years of the study. Depending on the settling patternsof early males,this couldmean that no unmated TABLE4. Reproductiveoutput of femalesin relation males were available near the areas where re- to their mating status.'Values shown are means + SD, with sample sizesin parentheses. turning femalesformerly nested. This is evidentlywhat happenedin 1982(Fig. Number Number 2), when 7 old femalesand only 3 old males fledged fledged returned. The territories of the 3 returning Mating (firstbrood)** (season-long)* status 1979-1982 1981-1982 males contained portions of what had been multiple territories in 1981, but there were Primary 4.8 + 0.39 8.3 + 0.96 many unoccupiedareas. All 6 femaleswhose (6) (4) nest sitesare shown in Fig. 2 returned in 1982. Secondary 2.9 + 1.30 4.8 + 1.70 Each mated with the male that held the terri- (5) (4) tory closestto her old nest site. This resulted Monogamous 3.5 + 1.05 4.8 + 1.06 in 3 bigamousmatings. (15) (10) Site fidelity appeared to influence mating a ß = p < 0.05, ** = P < 0.01; Kruskal-Wallis test. October1986] FemaleSite Fidelity and Polygyny 787

TABLE5. Male feeding rateson days5-8 post-hatch. TABLE6. Laying datesand reproductivesuccess of femalesin relation to the number of broodsthey Percent- produced.Egg 1 dateswere recordedby making Trips. age of the date the first egg of the seasonwas laid equal nestling-•.h • total trips to 1 and numbering days sequentially.a Female by malea made by status n (:• _+ SD) male Date egg 1 Fledglings No. of laid* produced** Secondary 5 0.53 + 0.53 16.1 broods n (t + SD) (œ+ SD) Monogamous 14 1.89 + 0.62 52.5 1 6 9.8 + 5.3 3.7 + 0.82 a U = 2, P < 0.05;two-tailed Mann-Whitney U-test. 2 13 4.7 + 2.46 6.6 + 1.76

a *U = 14.5, P = 0.03; **U = 6.5, P < 0.01; two- above,only the primary female (O/B) returned tailed Mann-Whitney U-test. to a site on which she had bred previously,but she settled there even though another female was already present. initiated her first brood more than 10 daysafter the first egg of the seasonwas laid had a second brood. Females that attempted two broods COSTSAND BENEFITSOF SECONDARYSTATUS fledged significantly more young than those Costsof secondaryover monogamousstatus.--I that did not (Table 6). found no evidenceof a significantcost to sec- Early nest initiation allows, but does not ondary females comparedwith monogamous guarantee,an opportunity for a secondbrood. femalesin terms of either number of young Only secondaryfemales that switched mates producedor adult survival. Primary females after their first broodsand mated with previ- producedsignificantly more fledglingsthan did ouslyunmated maleshad secondbroods. These monogamous or secondary females, but there previously unmated males were not present was no significant difference between the suc- when the females initiated their first nests. cess of the monogamousand secondaryfemales(Table 4). This was true for comparisons DISCUSSION of first-broodand season-longsuccess. In termsof the number of young produced, One measureof territory quality, the number secondaryfemales were no worse off than the of large coniferson the territory, was correlat- average monogamous female. However, sec- ed with male mating status.Female settlement ondaryfemales received significantly less help choiceswere better explained by site fidelity from their matesin feeding nestlingsrelative than by variations in territory quality. Biga- to monogamousfemales (Table 5). Secondary mousmatings resulted when no unmated males femalesresponded to reducedmale aid by in- were available near the former nest sites of re- creasingnestling feeding ratescompared with turning females.The costof bigamousmatings monogamousfemales. Secondary females made appearedto be low. Secondaryfemales did not as many trips as were made by both members have lower reproductive successor survival of 14 monogamouspairs (3.6 vs. 3.5 trips.nest- than did monogamousfemales. Mating with al- ling-•. h-l). Theremight be a costto secondary ready-mated males allowed females to initiate females because of this extra effort. For exam- their nestsearly enoughto have secondbroods. ple, secondaryfemales might suffer higher The correlation of male arrival time with male overwinter mortality than monogamousfe- mating statusand female reproductive success males. I found no evidence for this, but the is consistent with the idea that the mating samplesizes were small. The averagereturn choicesof female Blackpoll Warblers on Kent rate for adult females was 49%, whereas 50% of Island are influencedby the benefitsof early secondaryfemales (n = 4) from 1979-1981 re- nest initiation. I found the influence of terri- turned in subsequentyears. tory quality to be equivocal in female settle- Benefits:early nest initiation.--Females that had ment choices.In particular, territories of biga- secondbroods initiated their first nestssignif- mous and monogamousmales differed in the icantly earlier than thosethat did not have sec- numbers of total large conifers (TLC), but TLC ond broods (Table 6). In fact, no female that was not correlated with female settling pat- 788 BONITAC. ELIASON [Auk, Vol. 103 terns, nor did it explain the variance in repro- and matingstatus in other species(Searcy 1979, ductive successamong monogamousfemales. Oring 1982). There have been few studies of The differencein the number of large coni- the effectsof female site fidelity on female mate fers between the territories of bigamousmales choicesin polygynousspecies. Although the and other males may be an artifact of small role of territory quality in mate choiceby fe- sample sizes and differences in male return malesin this populationremains equivocal, site patterns due to chancealone. Four of the 5 in- fidelity was a sufficientexplanation for the oc- stancesof bigamy occurredin the part of the currence of polygyny. study area with the highest tree density. The Kent Island may be a favorableplace to breed low return rates of males in that area resulted for reasonsindependent of territory quality. in reduced competition among the males.As a Ricklefs (1969) found that predation was the result, males claimed territories with large most important mortality factor of eggs and numbersof conifers.If bigamy aroseas ! have young for most birds. Predation on eggs and suggested,however, females mated with those nestlingswas very low on Kent Island because males, not because the males' territories had there are no mammalian and few avian pred- large numbersof conifers,but becausethe fe- ators.The minimum number of fledglingspro- males preferred to nest near where they had ducedby any femalein a seasonwas 2 (n = 34). nested formerly. Ninety-eight percent of clutches(n = 48) pro- Territory quality may limit the number of ducedat leastone fledgling (Eliason1986). This females that can breed on a territory. No male compareswith 52.5% for 35 altricial passerine had more than two mates. In one case 5 females species(Nice 1957). A female Blackpoll War- returned that had bred on adjacent territories bler breeding on Kent Island has a high prob- in the previousyear. Two males returned and ability of producingsome young no matterwhat established territories that included the terri- her mating status. tories on which the 5 females had bred. Two High nestsuccess may explain the tendency bigamousmatings resulted that involved 4 fe- of femalesto breed near where they formerly males. The fifth female mated monogamously bred. Birds that nest successfullyhave been on a territory more than 3 territoriesaway from shown to disperse shorter distancesbetween her former breeding spot. This was the only breeding attemptsthan those that are unsuc- one of 12 returning femalesthat did not settle cessful (Catchpole 1972, Harvey et al. 1979, near her former breeding spot. Blockstein1986). If female Blackpoll Warblers Severalfactors may have obscuredrelation- use the simple rule of returning to a site where ships between the quality of the breeding sit- they were successful,then 98%of them should uation and female settling patterns and repro- return to their former sites (92% of them did ductivesuccess. First, females may have chosen SO), where to settle based on characteristics that ! A female Blackpoll Warbler has several op- did not measure.Second, pooling among years tions if she returns to her former territory and may have obscuredthe actualrelationship be- finds no unmated males nearby. If she pairs tween the number of large coniferson the ter- with a mated male and initiates her nest im- ritory and female settling patterns or repro- mediately, she almost certainly can raise some ductive success.Third, the number of young young,and maybe ableto havea secondbrood. fledgedannually may not be a reliable indica- Also, a female that maintainsher positionon a tor of lifetime reproductivesuccess. There may site by becominga secondaryfemale may have be differential postfledgingor overwinter mor- an advantage in competing for breeding op- tality, or a lackof congruencebetween seasonal portunitieson the site in subsequentyears. In and lifetime reproductivesuccess. Finally, even fact,two secondaryfemales that returnedto the if females base their choices of where to settle same site the next year became primary fe- on territory quality, unpredictable nestling males.Primary females had significantlyhigher mortality factorssuch as weather may obscure annual reproductive successthan either sec- relationshipsin a small sample. ondary or monogamousfemales (Table 4), re- Site fidelity by maleshas been recognizedas ceived as much male aid as monogamousfe- a complicating factor in determining the rela- males,and receivedsignificantly more aid than tionship among male quality, territory quality, secondaryfemales (Eliason 1986).A female that October1986] FemaleSite Fidelity and Polygyny 789

does not mate with a mated male when no un- EasternBird BandingAssociation, Northeastern Bird mated males are present near where she for- BandingAssociation, the GraduateSchool of the Uni- merly nestedfaces the possibilitiesof not mat- versity of Minnesota,and the University of Minne- ing or not having a secondbrood. If she decides sota Computer Center. to leave the island, the same uncertainties exist, plus the likelihood of higher predation rates. LITERATURE CITED Assuming the sample of birds I studied was BLOCKSTEIN,D.E. 1986. Reproductivebehavior and typical of the population asa whole, the overall parentalinvestment of Mourning Doves (Zenai- sexratio onceall birds had arrived was slightly da macroura).Unpublished Ph.D. dissertation, biasedtoward malesin yearswhen it deviated Minneapolis, Univ. Minnesota. from equality (Table 1). Female-biasedsex ra- CATCHPOLE,C. 1972. A comparativestudy of terri- tiosexisted early in the seasonbecause yearling tory in the Reed Warbler (Acrocephalusscirpaceus) males returned later than older males. Year- and Sedge Warbler (A. schoenbaenus).J. Zool. lings return later in a number of species(e.g. (London) 166: 213-231. Martin 1970,Nolan 1978,Oring and Lank 1982). ELIASON,B.C. 1986. Mating system,parental care, Oring and Lank (1982) suggestedthat because and reproductive successin the Blackpoll War- yearlingscould not competesuccessfully with bler (Dendroicastriata). Unpublished Ph.D. dissertation,Minneapolis, Univ. Minnesota. more experiencedbirds for territories, they de- EMLEN,$., & L. ORING. 1977. Ecology,sexual selec- layed their arrival until older males were in- tion and the evolution of mating systems.Sci- volved in breeding activities and less commit- ence 197: 215-223. ted to territory defense. An alternative FORD,N. 1983. Variation in mate fidelity in monog- explanationfor the later arrival of yearlingsis amousbirds. Pp. 329-356in Current ornithology, that yearlingsare lesssuccessful in competing vol. 1 (R. Johnston,Ed.). New York, Plenum Publ. for resourcesin wintering areas,take longer to Corp. build up reservesfor migration, and are not GREENWOOD,P. 1980. Mating systems,philopatty, ready to migrate as soon as older birds are and dispersalin birds and mammals.Anim. Be- hav. 28: 1140-1162. (Rohwer and Butler unpubl. ms.). HARVEY, P., P. GREENWOOD, & C. PERRINS. 1979. Site attachment and return patterns can ac- Breedingarea fidelity of the GreatTit (Parusma- count for the incidence of polygyny in this jor). J. Anim. Ecol. 48: 305-313. population, but data are not available to eval- HEISLER,I. 1981. Offspring quality and the polyg- uate the role of these factorsin determining yny threshold: a new model for the sexy son mating statusin other populationsof Blackpoll hypothesis.Amer. Natur. 117:316-328. Warblers.The combinationof site fidelity and HINDE,R. 1956. The biologicalsignificance of the return patterns that led to polygyny in this territories of birds. Ibis 98: 340-369. populationprobably could not give rise to high MARTIN,$. 1970. Polygynyin the bobolink:habitat levels of polygyny or large harems.Long-term quality and the adaptive complex.Unpublished data on breeding locations and mating rela- Ph.D. dissertation,Corvallis, Oregon State Univ. McCAIN, J. 1975. A vegetationalsurvey of the vas- tionships of marked individuals are needed to cular flora of the Kent Island group, Grand Ma- determinethe influenceof site fidelity in other nan, New Brunswick. Rhodora 77: 196-209. speciesthat exhibit low levelsof polygyny. McCULLAGH, P., & J. NELDER. 1983. Generalized lin- ear models. New York, Chapman & Hall. ACKNOWLEDGMENTS NICE,M.M. 1957. Nesting successof altricial birds. Auk 74: 305-321. I thank Bowdoin College, and in particular Dr. NOLAN,V., JR. 1978. The ecology and behavior of CharlesHuntington, for use of their equipmentand the Prairie Warbler, Dendroica discolor. Ornithol. Kent Island facilities. I am grateful to M. Beverage, Monogr. No. 26. S. Davis, R. Desley, M. Helmberger,and A. Kronke ORIANS,G. 1969. On the evolution of mating sys- for assistancewith the fieldwork. D. Blockstein, J. tems in birds and mammals. Amer. Natur. 103: Howitz, K. Larntz, C. Monnett, F. McKinney, L. Rot- 589-603. terman, F. Singer, and D. Sparling made comments ORING,L. 1982. Avian mating systems.Pp. 1-79 in and criticismsof earlierdrafts of this paper.The work Avian biology, vol. 6 (D. S. Farner, J. R. King, was supportedby grantsfrom the Wilkie Fund of the and K. C. Parkes, Eds.). New York, Academic Bell Museum of Natural History, Chapman Fund of Press. the AmericanMuseum of Natural History, SigmaXi, , & D. LANK. 1982. Sexual selection, arrival 790 BONITAC. ELIASON [Auk, Vol. 103

times,philopatry, and sitefidelity in the polyan- WEATHERHEAD,P., & R. ROBERTSON. 1977. Harem size, drous Spotted Sandpiper. Behav. Ecol. Sociobiol. territory qualityßand reproductivesuccess in the 10(3): 185-192. Red-winged Blackbird(Agelaius phoeniceus). Can. RICKLEFS,R. 1969. An analysisof nesting mortality J. Zool. 5: 1261-1267. in birds. Smithsonian Contr. Zool. 9: 1-48. --, & --. 1979. Offspring quality and the SEARCYßW. 1979. Male characteristicsand pairing polygyny threshold: "the sexy son hypothesis." successin Red-winged Blackbirds.Auk 96: 353- Amer. Natur. 113: 201-208. 363. WEISBERG,S. 1982. Multreg users manual, version , • K. YASUKAWA.1981. Does the "sexy son" 4.1. Technical Rept. #298, Sch. Statistics,Univ. hypothesisapply to mate choice in Red-winged Minnesota. Blackbirds? Amer. Natur. 117: 343-348. WITTENBERGERßJ. 1979. The evolution of mating sys- VEHRENKAMP,S., & J. BRADBURY.1984. Mating systems in birds and mammals. Pp. 271-349 in tems and ecology. Pp. 251-278 in Behavioral Handbookof behavioralneurobiology: social be- ecology: an evolutionary approach, 2nd ed. (J. havior and communication (P. Marler and J. Krebs and N. Davies, Eds.). Sunderland, Massa- Vandenberg, Eds.).New York, Plenum Press. chusetts, Sinauer Assoc. 1981. Male quality and polygyny: the "sexy VERNERßJ. 1964. Evolution of polygamyin the Long- son" hypothesisrevisited. Amer. Natur. 117:329- billed Marsh Wren. Evolution 18: 252-261. 342. ß & M. WILLSON. 1966. The influence of hab- itats on mating system of North American passerine birds. Ecology47: 143-147.