Male and female birds typically form strong

Plygyny and extrapair fertilizatins in eastern red-winged blackbirds (Agelaius pheniceus) Parentage f nestling red-winged blackbirds (Agelaius pheniceus) frm an eastern ppulatin was determined using DNA fingerprinting techniques. Of 235 nestlings surveyed, 58 had fingerprints excluding the male, but nne excluded the female tending the nest. Data n pairing status during the female's fertilizable perid was available fr 232 ffspring; 55 (25% f 1988 nestlings, 23% f 1989 nestlings) f thse were sired thrugh extrapair cpulatins. Of these 55 ffspring, 33 culd be assigned t nearby territry hlders; 16 f the remaining nestlings may have been sired by nearby males that were nt captured. During bth years, 44% f territrial males had mre than ne female nesting simultaneusly n their territry. The number f extrapair fertilizatins gained by males increased significandy with harem size in 1 year. Paternity (die prprtin f nesdings n the territry sired by die territry hlder) shwed a psitive but nnsignificant increase widi harem size in bdi years. There was n apparent cst in paternity fr males guarding tw r mre fertilizable females at the same time. The brds f females that were fertilizable at die same time andier female was setding n die same territry tended t have a greater prprtin f extrapair fertilizatins (0.37) than did die brds f dier females within harems (0.15). Established fertilizable females were chased significantly mre by die territry wner and by extrapair males when a new female was setding. There were n assciatins between a male's paternity r success at gaining extrapair fertilizatins and his age r clr-band cmbinatin. Overall, extrapair fertilizatins had litde effect n die relatinship between fledgling success and harem size and appeared t have a minimal impact n die verall intensity f sexual selectin n males. Key wrds: paternity, sexual selectin, DNA fingerprinting, mating systems, reprductive success, extrapair cpulatins, mixed reprductive strategy. [Behav Ecl 4:49-60 (1993)] Male and female birds typically frm strng scial assciatins widi each ther during breeding (Lack, 1968). Cpulatins utside diese scial assciatins (extrapair cpulatins, r EPCs) are nw recgnized as an imprtant mixed mating strategy (sensu Trivers, 1972) in scially mngamus and plygynus birds (reviewed by Birkhead, 1987; Birkhead and Mailer, 1992; Birkhead et al., 1987; McKinneyetal., 1984; Westneatetal., 1990). Recent studies using genetic techniques t analyze die cnsequences f EPCs (extrapair fertilizatins, r EPFs) have revealed cnsiderable variatin in die frequency f EPFs. Burke et al. (1989) and Gytlensten et al. (1990) fund few if any EPFs in dunncks (Prunella mdularis) and wd and willw warblers (Phyllscpus sibilalrix and P. Irchilus), respectively. In cntrast, 35% 40% f chicks came frm EPCs in indig buntings (Passerina cyanea; Payne and Payne, 1989; Westneat, 1987,1990) and white-crwned sparrws (Zntrichia leucphrys; Sherman and Mrtn, 1988) and 65%-l 00% f die chicks were sired by males utside die scial grup in splendid fairy wrens (Malurus splendens; Brker et al., 1990). Studies f many dier species have revealed a similar diversity in die frequency f EPFs (reviewed by Birkhead and Meller, 1992). Explanatins fr this diversity are incmplete. In part diis is because few details abut die csts and benefits t die participants are knwn, despite die increasing attentin paid t EPCs. The benefits t males f pursuing EPCs seem clear; males can increase dieir reprductive success widiut prviding care t resultant yung. Yet pursuit f EPCs can have csts, such as reducing die time and energy available fr parental care t yung in dier nests, expsing males t predatrs r sexually transmitted diseases, r cmpeting widi dier mating tactics such as mate guarding r attractin f additinal mates (Westneat et al., 1990). This last pssibility is receiving increasing attentin. EPCs are knwn t ccur in sme plygynus birds (see Alatal etal., 1984; Bllinger and Gavin, 1991;Gavin and Bllinger, 1985;Gibbs etal., 1990). Plygyny and EPCs culd interact in tw ways. First, plygynus males might be faced widi a cmplicated trade-ff between mate guarding and/r pursuing EPCs and being plygynus (Alatal et al., 1987; Westneat etal., 1990). Plygynus males may be incapable f guarding several females at nce (e.g., Alatal et al., 1984) r incapable f guarding ne female while attempting t attract new females. Behaviral data n mate guarding in plygynus great reed warblers (Acrcephalus arundinaceus) suggest diat males decrease guarding when attempting t attract additinal females (Hasselquist and Bensch, 1991). Guarding several females, attempting t attract new females, and prviding parental care t several brds might all reduce the time and effrt available t a male fr pursuing EPCs. Gibbs et al. (1990) reprted diat die reprductive success f male red-winged blackbirds (Agelaius pheniceus) was nt assciated widi harem size after David F. Westneat Sectin f Genetics and Develpment, Bitechnlgy Building, Crnell University, Ithaca, NY 14853, USA D. F. Westneat is nw at the Center fr Evlutinary Eclgy, T. H. Mrgan Schl f Bilgical Sciences, 101 Mrgan Building, University f Kentucky, Lexingtn, KY 40506-0225, USA. Received 16 Octber 1991 Revised 25 March 1992 Accepted 6 April 1992 1045-2249/93/$4.00 1993 Internatinal Sciety fr Behaviral Eclgy Dwnladed frm http://behec.xfrdjurnals.rg/ at Penn State University (Patern Lib) n September 18, 2016 Westneat Plygyny and EPFs in red-winged blackbirds 49

the effects f EPFs were included, suggesting that plygyny interfered with bth the pursuit f EPCs and the guarding f females against EPCs. Similarly, Bllinger and Gavin (1991) reprted that primary female bblinks (Dlichnyx ryzivrus) were mre likely t have their ffspring sired by EPCs than secndary females. The secnd way in which plygyny and EPCs culd interact is if plygynus pairings and EPCs are achieved in similar ways. Fr example, plygynus males might simply be better at attracting and guarding females than mngamus males. Differences in male fighting ability culd lead t sme males defending larger territries and thus attracting mre females. Such fighting ability might als give males an advantage in prtecting their wn paternity while als utcmpeting ther males fr EPCs (cmpetitive male hypthesis). Alternatively, sme males might be preferred by females (female preference hypthesis). These males might attract mre females t their territries. Thse females might avid EPCs, whereas females settling n territries f less preferred males might engage in EPCs with the preferred males. Bth the cmpetitive male and female preference hypdieses culd lead t plygynus males gaining mre EPCs and/ r having higher paternity than mngamus males. I tested these hypthesized relatinships between EPCs and plygyny in an eastern ppulatin f red-winged blackbirds. The red-winged blackbird is a well-studied plygynus icterid abundant in Nrth America, and EPCs have been reprted in this species by several researchers. Bray et al. (1975) and Rberts and Kennelly (1980) vasectmized male red-winged blackbirds and fund substantial numbers f fertile eggs in nests f thse males' mates. Mnnett et al. (1984) and Orians and Davies (unpublished data) reprted bservatins f EPCs in western red-winged blackbirds. Recently, Gibbs et al. (1990) used DNA fingerprinting t shw that 28% f chicks in an Ontari ppulatin were sired by nearby males. Males widi larger harems appeared t fare prly in terms f EPFs, but nne f the hypthesized relatinships between EPCs and plygyny were examined directly. In this paper I present results suggesting a different relatinship between EPCs and plygyny than diat reprted by Gibbs et al. (1990). METHODS Study ppulatin I studied the parentage f nesding red-winged blackbirds in a ppulatin inhabiting small pnds, cattail edges arund a shallw lake, and in several ld fields within Crnell University's Experimental Eclgy Pnd Unit 1, 5 km nrtheast f Ithaca, New Yrk, USA (42 N, 76 E). Data presented here cver the breeding seasns f 1988 and 1989. A detailed descriptin f die study site can be fund in Westneat (1992a). Field methds In mid-april in bdi years, I began trapping males using treadle traps baited with cracked crn. Sme females were als trapped during this time. By die end f April, all f the territrial males n die main study area had been captured. In 1988,1 cntinued trapping, fcusing n females until they began building nests. In 1989,1 reduced trapping effrts by 1 May and minimized capture f females befre and during nest building. In bdi years, abut ne- Uiird f the females (14 and 12, respectively) were trapped after incubatin had started. In 1988, seven females were never banded, and in 1989 tw eluded capture. I captured sme females in bth years by placing a treadle trap n the nest when the nesdings were 4-6 days ld. I banded captured birds widi diree clred, plastic bands (Hughes Ltd.; pssible clrs were yellw, red, green, dark blue, and light blue) and a numbered aluminum band, with nly tw bands per leg. The clr cmbinatin diat each bird received was determined arbitrarily using the fllwing rules; I attempted t band neighbrs using different clrs and avided using tw bands f the same clr n ne leg. Sme clr bands, particularly red, faded frm 1 year t the next and s were replaced if the bird was recaptured. I tk bld samples frm the brachial vein in the right wing f all birds n initial capture. Between 0.1 and 0.4 ml f bld was taken and placed in tw vials cntaining 0.1 ml f TNE (10 mmtris- 10 mm NaCl-2 mm EDTA, ph 8.0) buffer (Quinn and White, 1987). I dien flded die wing back alng die side f die bird and held die wing fr several minutes befre releasing die bird. I tk bld samples frm nesdings at 4 6 days after hatching. As widi die adults, I flded die wing ver and pressed it against die bdy t help stp the bleeding and then placed die nestling back int die nest. Behaviral bservatins My assistants and I began intensive bservatins f male and female reprductive behavir in late April in bth 1988 and 1989. We fcused n females widi bands, but sme unhanded females were bserved when diere were few dier unhanded females nearby that culd create cnfusin abut identity. Once a female started building a nest, I began 60-min bservatin perids f fcal pairs (die territrial male and ne f die females n his territry). Each fcal pair was bserved n at least ne and as many as 10 different days during nest building and egg laying. Observatins spanned all daylight hurs (0500 h-2000 h Eastern Standard Time). I determined territry bundaries by bserving where particular males sang and where diey respnded t natural intrusins by dier males. When diey were n die territry, bdi members f die fcal pair culd usually be bserved simultaneusly. Hwever, when a chice had t be made, I fcused n die female. Vice descriptins f male and female behavirs were recrded n tape and later transcribed. During bservatin sessins I nted die ccurrence f any physical interactin between die female and her mate r extrapair males. These included curtship displays by male r female, chases, and attempted r successful muntings. Observatins were made frm blinds n statinary twers, cars, r prtable canvas blinds. The latter were particularly effective as die bserver culd fllw die birds fairly clsely widiut disturbing diem. Hwever, windy cnditins, a fairly cmmn ccurrence n die study site, made ex- Dwnladed frm http://behec.xfrdjurnals.rg/ at Penn State University (Patern Lib) n September 18, 2016 50 Behaviral Eclgy Vl. 4 N. 1

tensive use f these blinds difficult, and s either a car r twer blind was used in adverse cnditins. DNA fingerprinting I analyzed parentage using standard DNA fingerprinting techniques (Burke and Brufrd, 1987; Burke et al., 1989; Gibbs et al., 1990; Jeffreys et al., 1985a,b; Westneat, 1990). Bld samples cllected frm adults and nestlings were stred at -20 C t -80 C until analysis. DNA was extracted frm the bld using a mdificatin f a prcedure described by Quinn and White (1987). Fr 20 f the 235 nestlings analyzed, DNA was btained frm tissue ther than bld using the extractin technique described in Westneat (1990). I islated DNA frm the carcasses f seven nestlings fund dead in the nest, eight embrys in unhatched eggs, and feather pulp and tissue salvaged frm remains f seven nestlings (tw f which had bld samples taken previusly) after a predatr had visited the nest (representing tw brds). In these latter tw cases, multiple samples were cllected and analyzed, and differences amng fingerprint patterns were taken as evidence that the samples were frm different nestlings. The surce f the DNA did nt affect the ability t examine parentage, except that in several instances t little DNA was available t run mre than ne r tw gels. Abut 12 Mg f DNA was digested with the restrictin enzyme Haelll. Digested DNA was precipitated, pelleted, and washed twice with 70% ethanl, then resuspended in 23 n\ f TE (10 mm Tris- 1 mm EDTA, ph 7.4). The ttal amunt f digested DNA in 20 /xl was adjusted t within 1 jig fr all lanes (usually 6-8 ng f DNA). Iplaced the samples in 0.8% agarse gels (mlds and buffer tanks manufactured by Bethesda Research Labs) and subjected them t a vltage differential fr 1700 vlt-h (apprximately 48 h) in 1 x TBE buffer (0.089 M Tris, 0.089 M brate, 0.002 M EDTA). A pump circulated the buffer slutin during electrphresis. Gels were treated as in Westneat et al. (1988), and DNA was transferred vernight t nyln membranes (Zetabind, AMF Cun) with 1 M ammnium acetate, 0.04 M NaOH. These membranes were then baked at 80 C fr 2 h. I placed between fur and eight membranes in plexiglass tubes with 30 ml f prehybridizatin slutin (NaPi; Westneat et al., 1988). Between 50 and 200 ng f prbe DNA was labeled by randm priming (Behringer Mannheim). Unincrprated nucletides were separated frm labeled prbes using clumns f Sephadex G-50 (medium) in TE. 1 then placed ne reactin per tw t three membranes in the plexiglass tubes with fresh NaPi. Hybridizatin ccurred at 60 C fr 48 h. Washes depended n which f the three prbes (Ml3, per, and 18.15) were used. All hybridized membranes, regardless f prbe, were initially washed twice at rm temperature and nce at 60 C with 2 x SSC and 0.5% sdium ddecyi sulfate (SDS). I then wrapped membranes prbed with M13 and placed them n film (Kdak XAR). Membranes prbed with per and 18.15 were washed an additinal ne r tw times with 1 x SSC at 65 C. Films were expsed withut intensifying screens fr 2 days t 3 weeks, depending n the strength f the signal. I then stripped membranes using 0.4 N NaOH at 42 C, neutralized in 0.2 M Tris, 0.5% SDS, 0.1 x SSC, and washed them in 0.5% SDS, 0.1 x SSC at 65 C. If nt reprbed immediately, the membranes were stred in 0.5 x TBE at 4 C. I islated the three prbes by cutting fragments ut f the apprpriate vectr DNA. I islated a 2.1-kb fragment f Ml 3 mpl 8 n lw-melting-pint agarse after cutting the fragment ut f whle M13 with the enzymes Haelll and Clal. A 2.5-kb fragment cntaining a clne f the muse per gene (Gibbs et al., 1990; Shin et al., 1985) was excised with EcRl and Hindlll. The 33.15 prbe (Jeffreys et al., 1985a,b) was similarly excised frm plasmid Psptl 8 (Carter et al., 1989). I islated all fragments n lw-melting-pint agarse and purified them using a GeneClean kit (Bilabs). A ttal f 82 adults (59 in 1988 plus 23 new adults in 1989) and 235 ffspring (112 in 1988 and 123 in 1989) were analyzed with at least tw prbes. Nine membranes (with 21 families and 74 nestlings) had weak Ml3 prbings, perhaps due t less DNA and a new batch f prbe that was less reliable. I used the 18.15 prbe as the secnd prbing n these membranes. Fur nestlings culd be scred fr nly ne prbe, perhaps due t abnrmally less DNA in their lanes. The data frm this ne prbing were initially treated in the same way as the cmbined data n tw prbes fr the remaining nestlings. Scring and interpretatin Scring fllwed the general methds utlined in Westneat (1990). Scring was nt dne blindly; nestlings were always run within a few lanes f each scial parent. I defined scial parents as the female attending the nest and the male hlding the territry at the time the nest was built and eggs were laid. Scring cnsisted f placing acetate verlays n the develped autradigraph, marking bands with clred pens, and judging whether r nt bands had migrated sufficiently similar distances t be called the same. I used several rules f thumb fr judging similarity. First, if the center f a band was within 0.5 mm f anther band, they were judged the same (Smith et al., 1991; Westneat, 1990). Bands that had migrated the same distance, but were f markedly different intensities in lanes that had similar amunts f DNA, were cnsidered assymetrically shared. That is, I cnsidered the darker band t be cmpsed f tw fragments, ne that was shared with the lighter band in the ther lane and ne that was nt. I did nt scre light bands in lanes cntaining mre DNA than in ther lanes, whereas I did scre such bands when a lane had less DNA than surrunding lanes. In all cases, I chse t scre cnservatively: If pssible, I matched bands in the nestlings with bands in the scial parents. Thus, bands in nestlings that were nt present in either scial parent had t be clearly different t be cnsidered nvel fragments. I used data n nvel fragments and prprtins f bands shared between ffspring and each scial parent t determine exclusins (e.g., Westneat, 1990). Thse ffspring that did nt match their scial father were reanalyzed n new gels cntaining bth putative parents and all males defending territries within 200 m. If n nearby male culd Dwnladed frm http://behec.xfrdjurnals.rg/ at Penn State University (Patern Lib) n September 18, 2016 Westneat Plygyny and EPFs in red-winged blackbirds 51

Table 1 Number f bands scred (±SD) and band-sharing cefficients (±SD) amng apparently unrelated adults fr each f the three fingerprinting prbes Prbe M13 M2.5 18.15 N. f bands Prprtin f bands shared 26.9 ± 7.1 (158) 0.32 ± 0.08 (145) 23.1 ± 5.2 (235) 0.31 ± 0.08 (175) 21.6 ± 4.9 (74) 0.32 ±0.06 (41) Sample sizes are in parentheses; fr the number f bands, sample sizes are numbers f nestlings scred fr each prbe, fr band-sharing cefficients, sample sizes are numbers f dyads. Dyads represent cmparisns with individuals run n the same gel within tw lanes. be assigned as the true father, I ran a new gel with additinal, mre distant males. Statistical analyses Fr analyses f band-sharing cefficients, I cnsidered dyads cntaining ne individual in cmmn as independent events. The band-sharing cefficient is a functin f bth individuals' fingerprints and their genetic relatedness. Thus, the cefficients frm tw dyads cntaining a cmmn individual are nt instances f pseudreplicatin. Because I nly scred individuals within several lanes f each ther, the prprtin f dyads that have duplicated individuals is relatively small. T analyze patterns f parentage I generally assumed that ffspring within a brd did nt represent independent inseminatins. I thus used prprtins f ffspring within a brd (r fr an adult within a seasn) as a dependent variable in many analyses. This variable is nt cmpletely satisfactry, as many individual brds r birds had paternity values f 1 r 0, which I treated the same regardless f the number f ffspring invlved (i.e., a brd size f 1 was weighted the same as a brd size f 5). In general, the alternative methd, using ffspring as independent events, gave qualitatively the same results, s I have reprted nly the frmer. Fr mst f the analyses n males, I pled all brds n a male's territry t arrive at verall paternity (number f ffspring n the territry sired by the territry hlder). Because sme males were the same frm ne year t the next, I analyzed the relatinship between plygyny and EPCs separately fr each year. RESULTS DNA fingerprints Numbers f bands and average prprtins f bands shared between adults in the ppulatin are shwn in Table 1. I used a sum f bands shared fr bth prbes t cmpare individuals. Unless therwise specified, this sum is f the tw prbes detecting the mst bands. The band-sharing cefficient between adults averaged 0.318, with a range f 0.190-0.530. The prprtin f bands shared was n different amng male-male dyads (0.316, n = 99), male-female dyads (0.322, n = 70) r female-female dyads (0.304, n = 5; ANOVA, F ia11 = 0.29, p >.70). Analyses f parentage Cmparisn f nestling fingerprints t thse f the scial parents revealed cmplete cmpatability fr 104 ffspring. In the remaining 121, ne r mre fragments in the nestlings' fingerprints was nt present in either parent's fingerprints. The distributin f nestlings with different numbers f nvel fragments was bimdal, but the mdes were nt entirely distinct (Figure 1). Nvel fragments arise frm either mutatin r thrugh EPF r intraspecific brd parasitism. Mutatin rates, which have ranged between 0.004 and 0.011 per fragment in ther studies (e.g., Burke and Brufrd, 1987; Jeffreys et al., 1985b; Westneat, 1990), thus were expected t result in a small number f nvel fragments, whereas mismatch f at least ne parent shuld have resulted in a larger number. The exact number due t either depends in part n the number f fragments scred; in the red-winged blackbirds that number averaged arund 45 (sum f tw prbes). The number f nvel bands expected if ne parent was mismatched als depends n the backgrund prprtin f bands shared (x) in the ppulatin. In the Ithaca redwinged blackbirds, x = 0.318. The average allele frequency (q) was 0.174 (x = 2^ q*; Jeffreys et al., 1985b). A parent wuld thus be expected t share (1 + q q*)/(2 q) bands with its ffspring (Jeffreys et al., 1985a), r 0.626, leaving 0.374 (r abut 17 bands) that shuld have been uniquely shared with the ther parent. If that parent was mismatched, then 0.318 f thse bands were expected t be shared with the actual parent, leaving an expected 11 nvel bands. This value is a mean with a variance that depends n the number f bands scred and the band sharing between all the individuals invlved. Thus, nestlings might have few nvel bands and yet nt be the ffspring f ne f the tw scial parents. Calculating the number f nvel bands needed t exclude parentage can be difficult, especially in this case, because the distributin was cntinuus (Figure 1). Hwever, I cnservatively assumed that nestlings with nly ne r tw nvel fragments were unlikely t have misassigned parents (adding in cases f three nvel fragments des nt change the general result). I used thse nestlings t estimate mutatin rates and then calculated the prbabilities f nestlings having three r mre nvel fragments frm mutatin alne. The prbability that a nestling (with between zer and five nvel bands) had at least ne nvel band was 0.40. The prprtin Dwnladed frm http://behec.xfrdjurnals.rg/ at Penn State University (Patern Lib) n September 18, 2016 52 Behaviral Eclgy Vl. 4 N. 1

120 100 en 80 t? 60 " I 40 20 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 17 N. f nvel bands Figure 1 Frequency histgram f the number f red-winged blackbird nesdings widi different numbers f nvel fragments (fragments nt shared with either scial parent). Dwnladed frm http://behec.xfrdjurnals.rg/ at Penn State University (Patern Lib) n September 18, 2016 f nestlings with at least tw nvel bands was 0.14, which crrespnds t a mutatin frequency f 0.37 per individual. The average f thse tw (0.38) gives the best guess at the actual mutatin rate per individual, with the rate/fragment = 0.008 (0.38/ 45). With that estimate, the expected prbability f bserving three nvel bands frm mutatin alne is 0.38s = 0.05, fur nvel bands is 0.384 = 0.02, and five nvel bands is 0.385 = 0.008. Given that I analyzed 235 nestlings, I thus expected abut 12, 5, and 2 nestlings t have 3, 4, and 5 nvel fragments, respectively. The bserved values fr three and fur nvel bands were clse t the expectatins, whereas that f five was mre than expected (Figure 1). I cncluded then that ffspring with three r fewer nvel fragments were highly likely t be descendent frm bth scial parents, whereas thse with five r mre nvel fragments were unlikely t be descendent frm at least ne scial parent. Fr nesdings with fur nvel fragments, I needed additinal infrmatin t either assign r exclude ne r bth parents. Band-sharing prprtins prvide such additinal infrmatin abut parentage. Nestlings with zer r ne nvel fragment shared 0.63 ± 0.07 f their fragments with each parent (Figure 2a,b). The lwer, ne-tailed, 99% limit f this distributin was 0.47, indicating that the prbability that direct descendents wuld have a cefficient less than 0.47 wuld be less than 1%. I therefre used tw cnservative criteria t exclude parentage: (1) mre than three nvel fragments and (2) a band-sharing cefficient <0.45 with at least ne parent. The cmbinatin f bth criteria resulted in 54 nestlings excluded as descendents f at least ne f their scial parents (Figure 2a,b). Nte, hwever, that in Figure 2, nine nestlings fit ne criterin but nt the ther; that is, either they had mre than three nvel fragments but a band-sharing cefficient with bth parents >0.45, r they had fewer than fur nvel fragments and a band-sharing cefficient with at least ne parent <0.45. Of these nine ambiguus cases, three were prbed with a third prbe (Ml3), revealing at least eight additinal fragments. That infrmatin remved the ambiguity in tw cases; withut Ml3 the nestlings had 5 and 10 nvel fragments, with Ml3 they had 9 and 12 nvel bands and a band-sharing cefficient with the male f 0.42 and 0.43. In the ther case Ml3 added n mre nvel bands but cnfirmed the high prprtin f bands shared with bth male and female (0.50 and 0.64); therefre, I cncluded the nestling was descendent frm bth putative parents. Three ther nestlings had a cefficient with the female fjust under 0.45 and zer nvel fragments (Figure 2b). One f these was prbed nly nce (see Methds) and anther had a slightly lighter lane. Given diat diese three had n nvel fragments, I cncluded they were ffspring f dieir putative mther. Of the remaining three nestlings, ne had 11 nvel fragments and a cefficient f 0.49 with the male and 0.60 with die female. Because it was unlikely diat 11 nvel fragments ccurred by mutatin alne, I cncluded this nestling excluded the scial father. Anther nesding had nly diree nvel bands but a band-sharing cefficient f 0.32 with the male. Because this cefficient is als very unlikely if the nesding was descendent frm the male, I cncluded diat it excluded die scial fadier. The ne remaining nesding had a truly ambiguus fingerprint, as it had a cefficient widi die male f 0.47 and fur nvel fragments. T be cnservative, I assigned diis nesding t its scial parents. I reanalyzed all excluded nesdings, including die fur ambiguus nes abve, t assign parentage (see belw). These reanalyses cnfirmed die decisins made abut dieir parentage. I cncluded diat 58 nestlings had fingerprint patterns excluding at least ne f the putative parents. As can be seen frm Figure 2a, diese nesdings had lw band-sharing cefficients widi die resident male and, as shwn in Figure 2b, high cefficients widi die resident female. This indicated diat all 58 nestlings came frm matings between die female and andier male and nne frm intraspecific brd parasitism. \ "5 04 00 4 S II! t 2 4 8 10 12 14 18 18 20 1.0 OS!! i 0.0 0 6 8 10 12 14 16 18 20 Number f nvel bands Westneat Plygyny and EPFs in red-winged blackbirds Figure 2 Relatinship between bandsharing prprtins and number f nvel bands fr each nesding red-winged blackbird. Band-sharing prprtins are with (a) the scial fadier and (b) the scial mther. Dashed lines indicate die criteria fr excluding parentage (see text). 53

nity S Q. 1.0, 0.8-0.6-0.4-8 0 The frequency f EPFs was thus 0.25 in 1988 and 0.23 in 1989. Over bth years, 28 f 68 brds (41%) had at least ne ffspring frm EPCs, 27 f 62 females (44%) were invlved in at least ne EPC, and 23 f 43 males (53%) were cucklded at least nce. Only ne nestling was knwn t be sired thrugh a female switching scial mates during the fertilizable perid. 0.2- Figure 3 The relatinship between paternity (prprtin f nestlings n a territry sired by the territry wner) and harem size (the maximum number f females nesting simultaneusly n a male's territry) fr (a) 1988 males and (b) 1989 males. Values within circles represent the number f cincident pints. A Kendall's rank crrelatin test was used t test fr a significant assciatin; in neither year was it significant (see text). ity c 0. 0.0 1 0-08 - 0.6-04 - 02-0.0 1 2 8 Harem size GGOO Harem size I was nt certain that all 58 excluded nestlings came frm EPCs. By definitin, EPCs ccur between individuals wh are nt paired t each ther. In this study, I cnsidered females as paired with males if the male was defending the territry at the time the female built her nest and laid her eggs. In three instances, males disappeared frm their territries after their female started incubating and were replaced befre bld samples were taken frm nestlings. In these cases the replaced male was cnsidered the scial father. In tw f thse instances these scial mates did nt sire sme yung in their nests (ttal f three nestlings). In anther instance, a male was replaced 2 days befre the first egg was laid. Only the replacement male was seen cpulating, but befre ding the genetic analyses I cnsidered the scial father as uncertain. Indeed, tw f three nestlings were nt sired by the replacement male. Hwever, they als were nt sired by the riginal male that was replaced. Because the female must have cpulated with a male t whm she was nt paired at the time, I cnsidered these tw yung as arising frm EPCs; the third was an ffspring f the replacement male, wh in subsequent analyses was cnsidered the scial father. Finally, ne nest was fund underging Incubatin in a rarely visited territry. A banded male defended the nest and was seen carrying fd during the nestling stage but culd nt be cnfidently cnsidered the scial parent. I analyzed the fingerprints f the nestlings as if the banded male was the scial father. All three nestlings in this nest were sired by anther male. It is pssible that the banded male bserved at the nest later in the nesting cycle had replaced the resident male after incubatin had started. Because it was nt clear if these three nestlings had been sired thrugh EPCs, I mitted this brd frm all subsequent analyses. Wh lses frm EPCs? Plygyny and apparent success 1 defined harem size as the maximum number f females wh had nests, eggs, nestlings, r recent fledglings (within 1 week) that verlapped. In the 2 years f study, 44% f territrial males had harems f tw r mre females. In each year, ne male (a different male and territry in each) had a harem f fur females whse nests verlapped in time. I analyzed the apparent success f males with different harem sizes in tw ways. First, I tallied the number f nestlings n a territry frm which bld samples were taken. This measure prvided the best sample t test the ptential interactin between EPCs and plygyny. Secnd, I used the number f fledglings as the best estimate f reprductive success available. The number f yung sampled n a territry (1988, Kendall's tau = 0.49, n = 26, p <.001; 1989, Kendall's tau = 0.62, n = 25, p <.001) was highly crrelated with harem size in bth years, whereas the number f fledglings prduced was crrelated with harem size nly in 1989 (1988, Kendall's tau = 0.19, n = 26,/> >.17; 1989, Kendall's tau = 0.57, n = 25, p <.001). Plygyny and paternity Paternity (the prprtin f all yung sampled n a territry in ne year that were sired by the territry wner) was nt significantly assciated with harem size in either year (Figure 3; 1988, Kendall's tau = 0.16, n = 22, p >.25; 1989, Kendall's tau = 0.0, n = 21, p >.95). This result initially suggested that there was n trade-ff between plygyny and prtecting paternity. Hwever, I tested whether males culd (1) divide their attentin amng several females and still guard them all r (2) encurage new females t settle and guard established females. There was n evidence that males suffered a lss in paternity when tw females had verlapping fertilizable perids (estimated as 5 days befre t 2 days after the first egg because mre than 95% f the cpulatins ccurred during that time; Westneat DF, manuscript in preparatin). I analyzed this hypthesis in tw ways. First, I used all brds as independent events, which gave the largest sample size. Females that verlapped had a similar number f EPFs in their brds (X = 0.22, n = 25, SD = 0.34) as females that were fertilizable alne n territries f plygynus males (X =» 0.23, n = 16, SD = 0.35; Mann-Whitney U = 196, nl = 25, n2 = 16, p >.9). Sme f the brds in this analysis were started at the same time n the same territry. T crrect fr pssible prblems f nnindependence, I repeated the analysis by pling brds f the apprpriate type within territries in each year. In this sample, females that were fertilizable simultaneusly had 23% EPFs (n = 10) in their brds, and females fertilizable alne n plygy- Dwnladed frm http://behec.xfrdjurnals.rg/ at Penn State University (Patern Lib) n September 18, 2016 54 Behaviral Eclgy Vl. 4 N. 1

U.90. p>050 Mnganxxjs Ply 1 lemale Ply 2+ temale U. 8 6 5. p > 0 55 Ply settling Mngamus Ply 1 temale Ply 2 * temale Ply settling U = 80.5. p»0.25 U = 65, p < 0.05 0) as 2 0 - > ^E^^B, Mngamus Ply l lemale Ply 2 * lemale Ply settling Mngamus Ply 1 temale Ply 1 lemale Ply 2+ temale nus territries had 14% EPFs (n = 9; Mann-Whitney {/= 39, p >.6). The appearance f a new female cncurrently with the fertilizable perid f an established female may have had sme effect n paternity. Brds f such females cntained 37% EPFs (n = 9), whereas brds started when n female was settling averaged nly 15% EPFs (n = 31), a difference that is nt quite significant (Mann-Whitney U = 94.5, p >.08). This analysis als cntains sme brds n the same territry. If brds f similar type are pled n territries within years and nly ne type f brd is used fr each territry in each year, then brds started when n female was settling averaged 20% EPFs (n = 10), nt significantly different frm the 37% in brds started when a new female was settling (Mann Whitney U D 35, p >.40). Fr six males, I had infrmatin n paternity bth when a new female was settling and when ne was nt (this includes tw cmparisns acrss seasns). Significantly mre EPFs ccurred in the brds started when a new female was settling (netailed Wilcxn rank test, T = 2, n - 6, p <.05). Behavir Details f the mating behavir f males and females during the fertilizable perid will be prvided elsewhere (Westneat DF, in preparatin). Here I fcus n cmparing the mating behavir f males and females n mngamus and plygynus territries. These analyses are based n bservatins cnducted during the perid frm 5 days befre t 2 days after egg laying. Only ne set f bservatins per pair per year was used, and these were averaged Ply settling U = 97, p > 0 70 u = 50, p < 0.01 Mngamus Ply 2* lemale Ply: settling MnganxHJS Ply: 1 temale acrss several bservatin sessins within pairs. In the case f plygynus territries, I used bservatins n tw different females n a particular territry in different categries f the independent variable. Restricting the sample by randmly chsing nly ne pair per territry changed neither the directin nr the statistical significance f the results. There were n significant differences in male r female behavir (number f within-pair curtships, within-pair chases, extrapair chases, female frays ff the study area, male frays, r intruders) n plygynus males' territries where nly ne female was fertilizable versus territries where tw r mre females were fertilizable simultaneusly (Figure 4a-f). Newly arrived females, hwever, had an effect n male behavir directed at the established female. The territrial resident and extrapair males bth chased fertilizable females significantly mre when a new female was present than when ne was nt (Figure 4b,c). Hwever, there were n significant differences in the frequency f withinpair curtships, male and female frays, r the number f intruders (Figure 4a,d f)age and cuckldry Territrial males in the Ithaca ppulatin in 1988 and 1989 were all in at least their secnd breeding seasn. I tested the effect f age n paternity by cmparing the paternity f males that returned in 1989 with their paternity in 1988. The males that returned in 1989 had an average paternity f 0.79 in 1988 and 0.77 in 1989, a difference that was nt statistically significant (Wilcxn signed-ranks Westneat Plygyny and EPFs in red-winged blackbirds Ply: 2* lemale Ply: settling Figure 4 Cmparisns f the rate (per hur) f within-pair and extrapair events by territries that were mngamus, plygynus with ne fertilizable female (Ply: 1 female), plygynus with tw r mre simultaneusly fertilizable females (Ply: 2 + female), and plygynus with a newly settling female at the same time as a fertilizable female (Ply: settling). Behavirs measured were (a) within-pair curtship, (b) within-pair chases, (c) extrapair chases, (d) female frays ff the study area, (e) male frays ff study area, and (f) intrusins n territries. A Mann-Whitney U test was used in all six panels (a-f) t cmpare the behavir f individuals n plygynus territries with ne versus tw fertilizable females r plygynus territries with bth a settling female and a fertilizable female versus thse with just fertilizable females. 55 Dwnladed frm http://behec.xfrdjurnals.rg/ at Penn State University (Patern Lib) n September 18, 2016 U = 23. p > 0 80

4 -, 1CO O) 2 (0 U O 1 2 Harem size In all cases either nne f the males fit r nly ne f them did. I assigned paternity t 33 (57%) f the 58 excluded ffspring t sampled males. Of the 25 unassigned ffspring, 19 (76%) were frm 8 territries n the margin f the study area (including the brd f 3 mitted earlier). All f these territries had at least ne adjacent neighbr that was nt sampled. Tw unassigned ffspring had been sampled as embrys, and there was insufficient DNA t screen mre than a cuple f males. Of the 33 assigned ffspring, 29 (88%) were sired by adjacent territry wners. The remaining fur were sired by males hlding a territry within 150 m f the nest. Six unassigned nestlings must have been sired by males mre than 150 m frm the nest, as all males within that distance were sampled and nne fit. Figure 5 The relatinship between the number f extrapair fertilizatins gained by males and their harem size in (a) 1988 and (b) 1989. Values within circles represent the number f cincident pints. The assciatin was significant in 1988 but nt in 1989 using Kendall's rank crrelatin test (see text). 3 - c CO CD 2 Q. UJ. O Harem size test, T = 44.5, p».05). Five males in 1989 had higher paternity, eight had lwer, and tw had the same (bth had full paternity in bth years). Clr bands Because male red-winged blackbirds have cnspicuus red and yellw epaulets, it was pssible that the clred leg bands influenced hw males were perceived by cnspecifics (e.g., Metz and Weatherhead, 1991), thereby ptentially affecting their paternity. I analyzed the effects f the prprtin f bands that were red, r red and yellw, n paternity. Fr males that were present in bth seasns, I pled infrmatin n paternity in each seasn. There was n assciatin between the prprtin f bands that were red (Kendall's tau = -0.02, n = 28, p >.90), r the prprtin that were red r yellw (Kendall's tau = 0.01, n = 28, p >.90), with paternity. Wh is successful at gaining EPCs? Assignment f paternity T assign paternity, I searched adjacent males, then nearby males (any part f their territry within 150 m f the fcal nest), and then males mre distant fr individuals wh had all r nearly all f an excluded nestling's paternal fragments in their fingerprint. Males having all but tw paternal fragments were cnsidered the actual father. I used this cnservative criterin because the prcess f searching thrugh a grup f males increased the prbability f finding a male thatfit just by chance. The prbability that any particular male shared all but tw paternal fragments (average number f paternal fragments was 17) by chance was 3 x 10" 6. The prbability that 1 f 10 males wuld fit by chance was 3 x 10" 5. Even thugh this is a mean with sme variance, I was cnfident f assigning the crrect male as the sire. Plygyny and pursuit f EPCs Males with large harems gained significantly mre EPFs than males with small harems in 1988 but nt in 1989 (Figure 5; 1988, Kendall's tau = 0.34, n = 27, p <.05; 1989, Kendall's tau = 0.06, n = 25, p >.65). This was true even if I excluded all males hlding territries n the margin f the study area wh might have sired sme yung in nests in unstudied territries (1988, Kendall's tau = 0.67, n = 14, p <.001; 1989, Kendall's tau = 0.04, n = 15, p >.84). The relatinship between actual male reprductive success (numbers f fledglings sired) and harem size varied amng years (Figure 6). In 1988, the relatinship was nt significant (Kendall's tau = 0.13, n = 25, p >.35), whereas in 1989 it was highly significant (Kendall's tau = 0.51, n = 24, p <.001). These relatinships held even when males n the edge f the study area were mitted (1988, Kendall's tau = 0.23, n = 14, p >.20; 1989, Kendall's tau = 0.53, n = 15, p <.01). This difference in actual success between years appears t be due t differences in nestling survival rather than t patterns f EPFs. The number f sampled nestlings sired by the male was significantly assciated with harem size in 1988 (Kendall's tau = 0.46, n = 25, p <.01), suggesting that the lack f a significant assciatin between actual fledgling success and harem size in that year was a cnsequence f the nnsignificant relatinship between number ffledglings prduced n the territry and harem size (see abve). Age I cmpared the number f EPFs gained by males in 1989 with the number they gained in 1988. Males gained 0.43 mre EPFs in 1989 than in 1988, but this is nt significantly different frm 0 (t = 1.2, p >.25). Of the 21 males that returned frm 1988, 13 (62%) gained the same number f EPFs as they did in 1988; 10 (77%) f these gained 0 EPFs in 1988 and 0 in 1989. Five males gained mre EPFs in 1989 and three gained fewer than in 1988. Band clr Neither the prprtin f bands that were red (Kendall's tau = -0.08, n = 34, p >.50) nr thse that were red and yellw (Kendall's tau = 0.09, n = 34, p >.45) was assciated with success at gaining EPFs. Dwnladed frm http://behec.xfrdjurnals.rg/ at Penn State University (Patern Lib) n September 18, 2016 56 Behaviral Eclgy Vl. 4 N. 1

Variance in male mating success Sme males lst and thers gained reprductive success thrugh EPCs. Males that had high paternity gained n mre EPFs than males with lw paternity in bth years (1988, Kendall's tau = 0.05, n = 22, p >.75; 1989, Kendall's tau = -0.09, n = 21, p >.55). Nevertheless, paternity r EPFs gained culd affect variance in reprductive success and hence the intensity f sexual selectin n males. I therefre calculated the mean and variance fr tw measures f reprductive success: number f chicks n a territry that fledged and number f genetic descendents that fledged. All males that defended territries n the study area fr mre than 10 days during the time sme females were fertilizable were included in this analysis. EPFs tended t increase variance in reprductive success regardless f whether territries n the edge were included (Table 2). Hwever, because the effect f variance in reprductive success n the prcess f selectin depends n mean success (Crw, 1958), I als calculated indexes f selectin (/,; the variance in success divided by the square f mean success) fr the full set f males and fr the males n the interir f the study area. Inclusin f EPFs increased /, in three f the fur cmparisns f the number f yung fledged (apparent success; Table 2). Hwever, the effects were generally small (less than 27% in all cases), especially fr the mre restricted sample f males. DISCUSSION This study cnfirms that extrapair cpulatins lead t extrapair fertilizatins in red-winged blackbirds. Apprximately 24% f the yung were sired thrugh these matings, and die majrity f these were sired by neighbring territry wners. These results add t the grwing evidence that males f many territrial passerines fllw a mixed mating strategy. The pursuit f EPCs as a mixed strategy culd affect the peratin f sexual selectin (e.g., Birkhead and Meller, 1992; Gibbset al., 1990;Gwaty, 1985; Westneat, 1987). The pprtunity fr selectin (/,) prvides sme infrmatin abut the verall ptential fr selectin (e.g., Wade, 1987). In the present study, EPFs had a minimal impact n the pprtunity fr selectin f males. Hwever, /, is a rugh index f the pprtunity fr selectin. It is sensitive t the types f data used in the calculatin (e.g., Cabana and Kramer, 1991; Cluttn- Brck, 1988; Dwnhwer et al., 1987) and des nt detect relatinships between specific cmpnents f mating success and male traits. This study has shwn that sme males gain frm EPCs and thers lse. Success and failure in these matings might be influenced by male mrphlgical r behaviral traits. Hw thse relatinships interact with traits imprtant fr success in ther areas remains an interesting and unanswered questin. My results prvide a different picture f the interactin between EPCs and plygyny dian that suggested by Gibbs et al. (1990). In the present study, inclusin f EPFs had litde effect n the relatinship between harem size and reprductive success in either year. In 1988, neither the number f fledglings prduced n die territry nr the number f genetic descendents fledged was significantly assciated with harem size. Hwever, bdi a I 6 ffl 3> O a> Harem size Harem size -- the number f sampled yung n the territry and the number descendent frm the male were highly assciated widi harem size. This difference is due t predatin n nests between die time f sampling and fledging. In 1989, bth apparent and actual fledgling success shwed strng assciatins widi harem size. In bdi years a male's paternity was unrelated t harem size, and die number f EPFs gained was psitively assciated with harem size in 1 f die 2 years. Males widi large harems thus appear t at least maintain paternity and gain mre EPFs dian males widi small harems in sme years. Hw this ccurs is nt clear. It is pssible diat plygynus males are lder and dius mre cmpetitive, such diat diey can defend a mre attractive territry and als get past die mate defense f dier males t gain EPCs. In diis study, diere was n evidence diat age affected success at EPCs, yet sample sizes were lw, and die study did nt cver many years. Plygynus males might als be mre cmpetitive, regardless f age. Data frm ther studies have prvided evidence suggesting male cmpetitiveness and harem size are nt strngly related (Eckert and Weadierhead, 1987). Similar infrmatin n male cmpetitiveness is nt available fr this ppulatin. A secnd pssible explanatin fr die success f plygynus males is diat females might be attracted t sme males regardless f territry quality (Weadierhead and Rbertsn, 1979). Sme females might settle widi dieir chice f mate and dien resist the extrapair attempts f dier males. Odier females might be excluded frm setding n die territries f dieir preferred mate but subse- Figure 6 The assciatin between actual fledgling success f males (number f genetic descendents in pl f fledglings) and harem size. Values within circles indicate the number f cincident pints, (a) Data frm 1988; a slight psitive trend was nt statistically significant, (b) Data frm 1989; a psitive relatinship was statistically significant (see text). Dwnladed frm http://behec.xfrdjurnals.rg/ at Penn State University (Patern Lib) n September 18, 2016 Westneat Plygyny and EPFs in red-winged blackbirds 57

Table 2 Means, variances, and estimates f the pprtunity fr selectin (/,; standardized variance) fr each f tw measures f reprductive success: number f yung fledged frm a male's territry (apparent success) and the number f descendents fledged (actual success) Year 1988 All males Apparent Actual Center males Apparent Actual 1989 All males Apparent Actual Center males Apparent Actual 26 25 14 14 25 24 15 15 Mean 3.2 2.7 3.6 3.4 3.2 3.0 2.9 2.7 Variance 10.7 9.6 13.3 11.8 10.2 9.8 10.4 10.8 1.04 1.32 1.03 1.02 1.00 1.09 1.24 1.48 Analyses were perfrmed n all males and just thse males with all neighbrs fingerprinted (center males) separately in each year. quently engage in EPCs with that male. If s, ne might expect later-settling females t engage in EPCs mre, but there was n assciatin between settlement date and EPFs (Westneat, 1992b). In additin, there was n evidence that females pursued EPCs in this ppulatin f red-winged blackbirds (Westneat, 1992b). Yet females may have mre subtle ways f exerting chice than by pursuing EPCs, such as accepting matings frm preferred mates while resisting attempts frm thers. The verall success f plygynus males ccurred in the face f a pssible trade-ff between EPCs and plygyny. There was weak evidence that paternity suffered when males attempted t becme plygynus. The frequency f EPFs in brds f females in the prcess f nest building r egg laying when a new female was settling was mre than duble that f ther females within harems. Because this trend was nt significant, it wuld be premature t analyze the pssibility f a trade-ff based n that result alne. Hwever, behaviral bservatins shwed a marked effect f newly setding females n the interactins between males and established yet fertilizable females. The frequencies f within-pair and extrapair chases were substantially higher than seen when a new female was nt setding (Figure 4). These within-pair chases usually resulted in reduced interactin between the established female and the new female. In sme cases males cnfined the established female t ne crner f the territry dirughut the bservatin sessin. Female-female aggressin has been interpreted as territrial behavir that may functin t prevent settlement f additinal females (e.g., Beletsky, 1983a,b; Hurly and Rbertsn, 1984; LaPrade and Graves, 1982; Leningtn, 1980; Ner, 1956; Orians, 1961,1980; Searcy, 1986; Yasukawa and Searcy, 1982). By attempting t prevent interactin, males might imprve die chances diat the new female will settle, but at the cst f reducing the effectiveness f mate guarding and lwering paternity in the established female's brd. This cst, hwever, averages abut 0.9 ffspring per brd [difference in paternity 0.37-0.15 = 0.22 x 4 (average brd size)]. This cst is ffset by die chance f siring abut 75% f andier brd f fur, even cnsidering that males were nt always successful at getting new females t stay and nest (2 f 14 such females did nt stay). There was n evidence diat multiple fertilizable females were difficult fr males t guard. This result is cnsistent with data indicating that males can usually guard mates by being present n the territry (Westneat DF, in preparatin). Thus, males can be plygynus withut much cst in paternity if they attract new females early, befre their dier females have started nesting, r late, after dieir dier females are incubating eggs. Whedier males can benefit by influencing die timing f female setdement is an interesting but currendy untested pssibility. The differences between my results and dise f Gibbs et al. (1990) present sme difficulties in frmulating a general explanatin fr the rle f EPCs in this plygynus bird. In dieir study, a significant, psitive assciatin between harem size and apparent success became a nnsignificant, psitive assciatin when die effect f EPFs was factred in. This suggests diat males with larger harems were mre likely t lse ffspring thrugh EPCs and/r less likely t gain diem. Whether such males achieve a net gain r a net lss frm EPCs is an extremely imprtant questin in light f debate ver die csts and benefits f pursuing plygyny (e.g., Searcy and Yasukawa, 1989). Three pssible explanatins fr die differences between die tw studies are (1) sample size, (2) missing infrmatin, and (3) unknwn eclgical factrs. First, Gibbs et al.'s (1990) cnclusin diat realized success is nt assciated widi harem size depends n nt rejecting a statistical null hypthesis. When sample sizes are small r variances are high, die pwer f any statistical test is lw; thus cnfidence in nt rejecting a null hypuiesis is als lw. The Ontari ppulatin studied by Gibbs et al. (1990) may in fact be similar t die Idiaca ppulatin, but because nly 13 males were analyzed in Ontari, additinal variance caused by including EPFs may have reduced die pwer f their statistical analysis, thus giving die nnsignificant result. Secnd, in bdi studies, sme infrmatin n sme males was missing. Fur males in Gibbs et al.'s study were mitted frm die analyses. Three males had nests diat were unsampled (tw f diese sired yung in ther males' nests), and the furth apparently sired n ffspring n any territry. It is hard t knw what effect such missing infrmadn might have n die results. In my study, sme EPFs culd nt be assigned t males. I suspect diis had litde effect n the results f my study; eliminatin f die males hlding territries n the margin f die study area had a minimal impact n mst f die analyses. The tw pints abve might be sufficient t explain die difference between die tw studies, but eclgical differences between ur study ppulatins might have cntributed as well. Males n the Ontari marsh averaged tw t diree mre ffspring dian males in Idiaca, and die indexes f Dwnladed frm http://behec.xfrdjurnals.rg/ at Penn State University (Patern Lib) n September 18, 2016 58 Behaviral Eclgy Vl. 4 N. 1

selectin (standardized variance in success) in Ithaca (Table 2) were tw t fur times that fund in Ontari (0.25-0.39; Gibbs et al., 1990). Harem size has been shwn t be affected by sme aspects f territries in sme ppulatins but nt in thers (e.g., Hlm, 1973;Leningtn, 1980; Pieman, 1981; Weatherhead and Rbertsn, 1977; Yasukawa, 1981), and these same differences might affect patterns f EPCs, particularly if in Ithaca a wider range f territry types were sampled. Fr example, it is pssible that fd resurces and/r the structure f the territry affect the abilities f males t guard their mates (e.g., Sherman and Mrtn, 1988; Westneat et al., 1990). In sme areas the pursuit f plygyny and the ability t guard mates may be incmpatible, whereas in thers they might nt be, perhaps due t the lcatin f fd resurces r the penness f the habitat. In additin, females may be chsing places t settle separately frm their chice f genetic mate in sme areas but nt in thers, perhaps due t differences in the csts and benefits f engaging in EPCs (e.g., Westneat etal., 1990). Finally, the results f this study suggest that the timing f female settlement may influence paternity. In the Ithaca blackbirds, females wh were fertilizable at the time anther female was settling n a territry tended t have brds with a higher prprtin f extrapair fertilizatins. This suggests that there can be a cst t males fr pursuing plygyny in certain circumstances. If secndary females in the Ontari ppulatin tend t settle mre simultaneusly with the fertilizable perids f primary females than in Ithaca, then lwer paternity f plygynus males might be the result. At present, it is nt pssible t evaluate any f these eclgical explanatins because the apprpriate infrmatin is nt available frm bth ppulatins. Clearly, cmparable data frm this and ther species in which the frequency f EPFs differs amng ppulatins (e.g., pied flycatchers; Alatal et al., 1984; Lifjeld et al., 1991) will help reveal many f the factrs influencing the evlutin f this mixed mating strategy. I am deeply indebted t C. F. Aquadr, B. Nn, H. K. Reeve, J. Clesn, C. Yn, and T. Byce fr their advice and supprt in develping the fingerprinting techniques. I thank Alec Jeffreys and the Lister Institute, Ry Carter, lisle Gibbs, and Mike Yung fr generusly supplying the fingerprinting prbes. B. Jhnsn and N. Hairstn were a tremendus help with the lgistics f using Crnell's pnds facility. L. Breen and G. Crutcher cmpleted much f the lab wrk, and C. Crsn and C. Butler wrked very hard in the field. I als thank A. B. Clark fr her cllabratin n the Ithaca blackbirds. Capture and banding activities were authrized under federal banding permit n. 2213, federal cllecting permit n. PRT-725174, and New Yrk State permit n. SCL 88-010. Finally, I appreciate the helpful cmments f Paul Sherman, Jeff Dickhaut, Greta Crutcher, Bb Mntgmerie, Patrick Weatherhead, and an annymus reviewer n previus drafts f the manuscript. This study was funded by Natinal Science Fundatin grants BSR 8700159 and BSR 890-06562, indirect assistance frm a Natinal Institutes f Health grant t C. F. Aquadr, and supprt frm Crnell University and the University f Kentucky. REFERENCES Alatal RV, Gttlander K, Lundberg A, 1987. Extra-pair cpulatins and mate guarding in the plyterritrial pied flycatcher Ficedula hypleuca. Behaviur 101:139-155. Alatal RV, Gustavssn L, Lundberg A, 1984. High frequency f cuckldry in pied and cllared flycatchers. Oiks 42:41-48. Beletsky LD, 1983a. Aggressive and pair-bnd maintenance sngs f female red-winged blackbirds (Agelaius pheniceus). Z Tierpsychl 62:47 54. Beletsky LD, 1983b. Aggressive respnse t 'self sngs by female red-winged blackbirds, Agelaius pheniceus. CanJ Z! 61:462-465. Birkhead TR, 1987. Sperm cmpetitin in birds. Trends Ecl Evl 2:268-272. Birkhead TR, Atkin L, Mailer AP, 1987. Cpulatin behaviur f birds. Behaviur 101:101-138. Birkhead TR, Meller AP, 1992. Sperm cmpetitin in birds: evlutinary causes and cnsequences. Orland, Flrida: Academic Press. Bllinger EK, Gavin TA, 1991. Patterns f extra-pair fertilizatins in bblinks. Behav Ecl Scibil 29:1-7. Bray O, KennellyJJ, Guarin JL, 1975. Fertility f eggs prduced n territries f vasectmized red-winged blackbirds. Wilsn Bull 87:187-195. Brker MG, Rwley I, Adams M, Baverstck PR, 1990. Prmiscuity: an inbreeding avidance mechanism in a scially mngamus species? Behav Ecl Scibil 26: 191-199. Burke T, Brufrd MW, 1987. DNAfingerprintingin birds. Nature 327:149-152. Burke T, Davies NB, Brufrd MW, Hatchwell BJ, 1989. Parental care and mating behavir f plyandrus dunncks Prunella mdularis related t paternity by DNA fingerprinting. Nature 338:249-251. Cabana G, Kramer DL, 1991. Randm ffspring mrtality and variatin in parental fitness. Evlutin 45:228-234. Carter RE, Wettn JH, Parkin DT, 1989. Imprved genetic fingerprinting using RNA prbes. Nucleic Acids Res 17:5876. Cluttn-Brck TH, 1988. Reprductive success. In: Reprductive success (Cluttn-Brck TH, ed). Chicag: University f Chicag Press; 472-485. Crw JF, 1958. Sme pssibilities fr measuring selectin intensities in man. Hum Bil 30:1-13. DwnhwerJF, Blumer LS, Brwn L, 1987. Opprtunity fr selectin: an apprpriate measure fr evaluating variatin in the ptential fr selectin? Evlutin 41: 1395-1400. Eckert CG, Weatherhead PJ, 1987. Cmpetitin fr territries in red-winged blackbirds: is resurce-hlding ptential realized? Behav Ecl Scibil 20:369-375. Gavin TA, Bllinger EK, 1985. Multiple paternity in a territrial passerine: the bblink. Auk 102:550-555. Gibbs HL, Weatherhead PJ, Bag PT, White BN, Tabak LM, Hysak DJ, 1990. Realized reprductive success f plygynus red-winged blackbirds revealed by DNA markers. Science 250:1394-1397. Gwaty PA, 1985. Multiple parentage and apparent mngamy in birds. In: Avian mngamy (Gwaty PA, Mck DW, eds). Omithl Mngr 37:11-21. Gyllensten UB, Jakbssn S, Temrin H, 1990. N evidence fr illegitimate yung in mngamus and plygynus warblers. Nature 343:168-170. Hasselquist D, Bensch S, 1991. Trade-ff between mate guarding and mate attractin in the plygynus great reed warbler. Behav Ecl Scibil 28:187-193. Hlm CH, 1973. Breeding sex ratis, territriality, and reprductive success in the red-winged blackbird (Agelaius pheniceus). Eclgy 54:356-365. Hurly TA, Rbertsn RJ, 1984. Aggressive and territrial behavir in female red-winged blackbirds. Can J Zl 62:148-153. Jeffreys AJ, BrkfieldJFY, Semenff R, 1985a. Psitive identificatin f an immigratin test-case using human DNA fingerprints. Nature 317:818-819. Jeffreys AJ, Wilsn V, Thein SL, 1985b. Hypervariable Dwnladed frm http://behec.xfrdjurnals.rg/ at Penn State University (Patern Lib) n September 18, 2016 Westneat Plygyny and EPFs in red-winged blackbirds 59