Natural selection for earlier male arrival to breeding grounds through direct and indirect effects in a migratory songbird

Na u al selec ion o ea lie male a i al o b eeding g ounds h ough di ec and indi ec e ec s in a mig a o y songbi d William Velmala 1,2,† , Samuli Helle 1,† , Ma kus P. Ahola 1,3 , Ma cel Klaassen 4,5 , Esa Lehikoinen 1 , Kalle Rainio 1 ,P € ai i M. Si ki€ a 1,2 & Toni Laaksonen 1 1 Sec ion o Ecology, Depa men o Biology, Uni e si y o Tu ku, Tu ku FI-20014, Finland 2 Finnish Museum o Na u al His o y, Uni e si y o Helsinki, P.O. Box 17, FI-00014 Helsinki, Finland 3 Na u al Resou ces Ins i u e Finland, I € ainen Pi k€ aka u 3, FI-20520 Tu ku, Finland 4 Depa men o Animal Ecology, Ne he lands Ins i u e o Ecology, D oe endaalses eeg 10, 6708 PB Wageningen, The Ne he lands 5 Cen e o In eg a i e Ecology, School o Li e and En i onmen al Sciences, Deakin Uni e si y, Wau n Ponds, Vic. 3216, Aus alia Keywo ds Fi ness, li e his o y, mic oe olu ion, seasonal in e ac ions, sexual selec ion, iming o mig a ion. Co espondence William Velmala, Sec ion o Ecology, Depa men o Biology, Uni e si y o Tu ku, Tu ku FI-20014, Finland. Tel: +358 50 4085039; Fax: +358 2 3336598; E-mail: [email p o ec ed] Funding In o ma ion The s udy was inancially suppo ed by he Academy o Finland (p ojec no. 130436 o TL). SH was unded by The Tu ku Collegium o Science and Medicine. Recei ed: 17 Decembe 2014; Re ised: 19 Janua y 2015; Accep ed: 21 Janua y 2015 Ecology and E olu ion 2015; 5(6): 1205–1213 doi: 10.1002/ece3.1423 † These au ho s con ibu ed equally o his wo k. Abs ac Fo mig a o y bi ds, he ea lie a i al o males o b eeding g ounds is o en expec ed o ha e i ness bene i s. Howe e , he selec ion di e en ial on male a i al ime has a ely been decomposed in o he di ec e ec o male a i al and po en ial indi ec e ec s h ough emale ai s. We measu ed he di ec ional selec ion di e en ial on male a i al ime in he pied lyca che (Ficedula hypol- euca) using da a om 6 yea s and annual numbe o ledglings as he i ness p oxy. Using s uc u al equa ion modeling, we we e able o ake in o accoun he empo al s uc u e o he b eeding cycle and he hie a chy be ween he examined ai s. We ound di ec ional selec ion di e en ials o ea lie male a i al da e and ea lie emale laying da e, as well as s ong selec ion di e en ial o la ge clu ch size. These selec ion di e en ials we e due o di ec selec ion only as indi ec selec ion o hese ai s was nonsigni ican . When decomposing he di ec selec ion o ea lie male a i al in o di ec and indi ec e ec s, we disco e ed ha i was almos exclusi ely due o he di ec e ec o male a i al da e on i ness and no due o i s indi ec e ec s ia emale ai s. In o he wo ds, we showed o he i s ime ha he e is a di ec e ec o male a i al da e on i ness while accoun ing o hose e ec s ha a e media ed by e ec s o he social pa ne . Ou s udy hus indica es ha na u al selec ion di ec ly a o ed ea lie male a i al in his lyca che popula ion. In oduc ion The annual cycle o mig a o y animals in empe a e and a c ic zones consis s o a leas ou cha ac e is ic pe iods: b eeding, o e win e ing, and wo mig a o y pe iods be ween he espec i e b eeding and o e win e ing g ounds. Mig an s a e aced wi h he inc easing challenge o iming each o hese pe iods op imally in he ace o ongoing changes in en i onmen al condi ions (Mølle e al. 2008b; Ca ey 2009; Knudsen e al. 2011; McNama a e al. 2011). To unde s and how iming may e ol e in esponse o hese changes, knowledge abou he way how selec ion wo ks on he iming o hese phases in he annual cycle is needed (e.g., Gunna sson e al. 2006; Go do e al. 2013). In e i o ial mig a o y bi ds, males usually a i e o he b eeding g ounds ea lie han emales (Mo bey and Ydenbe g 2001), which helps hem o claim and es ablish high-quali y e i o ies (e.g., Aebische e al. 1996; Hassel- quis 1998; Smi h and Moo e 2005). Ea ly-a i ing males ª2015 The Au ho s. Ecology and E olu ion published by John Wiley & Sons L d. This is an open access a icle unde he e ms o he C ea i e Commons A ibu ion License, which pe mi s use, dis ibu ion and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly ci ed. 1205 ha e also been ound o a ac highe -quali y ma es (e.g., Ala alo e al. 1984; Rubolini e al. 2004), es ablish la ge ha ems (Hasselquis 1998), and ha e mo e ex a-pai ma ing oppo uni ies (Reudink e al. 2009; Coope e al. 2011), he eby inc easing hei b eeding success. In addi- ion, ea ly-a i ing males and emales ha e been shown o pe o m be e in b eeding in e ms o la ge clu ch sizes, mo e ledglings (Po i 1998; H€ o ke 2002; T yjanowski e al. 2004; Se gio e al. 2007), and mo e ec ui ing o sp ing (Mølle 1994; Hasselquis 1998). The o e all ad an ages o a i ing ea ly, he e o e, seem o be well es ablished o bo h sexes, al hough i has been shown ha a i al da e in he wo sexes may also be unde di e gen selec ion (Mølle 2007). Howe e , we ha e gene ally a e y limi ed knowledge on wha he spe- ci ic selec ion pa hways a e o ea ly male a i al. Few o - mal analyses using di ec ional selec ion di e en ials ha e been conduc ed on how na u al selec ion ac s on a i al ime (Mølle 2007; Mølle e al. 2008a; Tepli sky e al., 2011; Gienapp and B egnballe 2012; A naud e al. 2013). Only one s udy has p e iously conside ed he po en ial di e en pa hways o selec ion on a i al da e, ha is, dis- inguishing be ween i s di ec e ec s on i ness and i s indi ec e ec s h ough ea lie laying da e and la ge clu ch size ha a e likely o depend on he quali y o he emale (No is e al. 2004). Analyzing da a om Ame i- can eds a s (Se ophaga u icilla), No is e al. (2004) ound ha he a i al da e did no ha e a signi ican di ec e ec on he numbe o ledglings in ei he sex. Ins ead, he indi ec i ness e ec s o a i al da e, ia emale laying da e and ledging da e, we e ound o be signi ican . Howe e , he pa h analysis o No is e al. (2004) was no used o es ima e he di ec ional selec ion di e en ial and i s componen s ha a e ele an measu es when compa ing he s eng h o selec ion be ween species and ai s (Scheine e al. 2000). To add ess hese issues, we s udied whe he he e is selec ion on he iming o male a i al in he pied ly- ca che (Ficedula hypoleuca), a small mig an songbi d b eeding in Eu asia and win e ing in sub-Saha an A ica. We quan i y he selec ion on male a i al da e by calcula - ing di ec ional selec ion di e en ial using s uc u al equa- ion modeling (SEM), which enables us o examine o wha ex en selec ion a ises h ough di e en pa hways (Scheine e al. 2000). The majo bene i s o applying SEM in selec ion s udies a e ha i enables (1) he modeling o a mo e biologically ealis ic scena io o mul i a ia e na u- al selec ion compa ed o uni a ia e app oaches and mul i- ple eg ession models (Lande and A nold 1983; Mo issey 2014) and (2) he sepa a ion o selec ion di e en ials in o di ec and indi ec selec ion and hei componen s (Schei- ne e al. 2000). In o he wo ds, SEM akes in o accoun he hie a chical sequence o he di e en a iables, such as he causally in e ela ed li e his o y e en s du ing a b eed- ing cycle. This me hod is s ill su p isingly a ely used in s udies quan i ying na u al selec ion (Mo issey 2014), despi e i s ob ious bene i s when analyzing mul i a ia e na u al selec ion om phenomena ha a e empo ally s uc u ed. To ou knowledge, he e a e no p e ious s ud- ies ha ha e used his app oach o model he selec ion on iming o mig a ion in a mig an bi d. Ma e ials and Me hods S udy species The pied lyca che is a long-dis ance mig an , b eeding in empe a e and bo eal o es s om Wes Eu ope o wes- e n Sibe ia, and win e ing in wes e n sub-Saha an A ica. I is a small (weigh ing 16 g on a e age) insec i o ous, ca i y-b eeding passe ine bi d, which spends eigh mon hs a yea away om he b eeding a eas, ei he on mig a ion o in win e ing g ounds in A ica. In Finland, pied ly- ca che s a i e be ween la e Ap il and ea ly June and he las bi ds depa by la e Augus . The pied lyca che is an ex ensi ely s udied model species, as i is abundan h oughou i s ange and eadily accep s man-made nes boxes, e en p e e ing hem o na u al ca i ies. (e.g., Lundbe g and Ala alo 1992) S udy si e Ou s udy si e is si ua ed on Ruissalo, an island o 9 km 2 in he icini y o he ci y o Tu ku (60°260N, 22°100E) in sou hwes e n Finland. The s udy si e consis s o coni e - ous o es , mainly sco s pine (Pinus syl es is), and decid- uous o es , wi h oak (Que cus obu ), sil e bi ch (Be ula pendula), and small-lea ed lime (Tilia co da a) as he main ee species. The s udy was conduc ed in 2005–2006 and 2008–2011. Each yea , 230 imbe nes boxes (inside dimensions 12.5 912.5 cm, inside heigh 23.6 cm, en ance hole 32 mm) we e a ailable and moni o ed, excep o 2010–2011 when only 195 nes boxes we e moni o ed. The nes boxes we e se as lines along pa hs and oads, he nea es -neighbo nes -box dis ance being 20–40 me e s. Each yea , by he ime he pied lyca che s a i ed, a p opo ion o he nes boxes was al eady occu- pied by ei he g ea i s (Pa us majo ) o blue i s (Cyan- is es cae uleus). The numbe o b eeding pai s o pied lyca che s in he whole s udy a ea a ied annually, om 79 o 124, wi h a mean o 101 b eeding pai s, al hough much ewe we e a ailable o he analysis (see below). The es o he nes boxes we e inhabi ed by i s o emained uninhabi ed. Pied lyca che nes -box occupancy o all nes boxes hus a ied om 40.5 o 53.9%, wi h a mean o 45.9%. 1206 ª2015 The Au ho s. Ecology and E olu ion published by John Wiley & Sons L d. Selec ion o Ea lie A i al in a Songbi d W. Velmala e al. Field obse a ions The a i al da es o indi idual pied lyca che males o he s udy a ea we e assessed by daily moni o ing o he whole s udy a ea. The moni o ing co e ed he en i e pe - iod o he males’ a i al, om la e Ap il un il la e May. Nes -si e moni o ing was conduc ed be ween 7 AM and 1 PM (UTC +2) by slowly walking h ough he s udy a ea and s opping a nes boxes, spending 3–4 min in he icini y o each nes box. The e we e 2–4 obse e s in ol ed in he moni o ing each day, and each obse e was andomly selec ed o moni o ing a speci ic subse o nes boxes on a daily basis. Flyca che s we e de ec ed using isual and audi o y cues. The males can be indi idually and unambiguously ecognized based on he p esence o colo ed and alumi- num ings, and plumage de ails, including colo a ion o he back, ump pa e n, shape and size o he whi e o e- head pa ch, and he amoun o whi e on he ail and wings. The shape o he o ehead pa ch was cha ac e ized as a uni o m block, sepa a e do s, o linked do s. The size o he o ehead pa ch was assigned as la ge, medium, small, o e y small/absen . Do sal colo a ion was assessed using a scale om 1 o 7 designed by D os (1936): 1 being black and 7 being emale-like b own. In addi ion o cha ac e de e mina ion using binocula s, digi al pho og- aphy wi h a elepho o lens was used o eco d indi idual plumage de ails in 2010 and 2011. The a i al da e es ima e is he i s da e a male was obse ed in he s udy a ea du ing he daily moni o ing. We included only hose indi iduals ha s ayed b eeding in he s udy a ea and whose iden i y could la e be con- i med on plumage cha ac e s while caugh a he nes box hey we e b eeding in (see easoning below). To a oid in luencing he males’ e i o y selec ion, hey we e no apped and inged o o he wise in e e ed wi h du - ing his pe iod (excep in 2009 when males we e cap u ed some days a e a i al). Ins ead, each male was s udied wi h binocula s o indi idual iden i ica ion. Females we e no included in he s udy because, due o he ex emely ine di e ences in plumage cha ac e s, hey canno be eliably iden i ied indi idually, making he de e mina ion o exac a i al day impossible wi hou in e e ence (i.e., ca ching and inging). The b eeding pe o mance o each male and his ma e was moni o ed h oughou he b eeding season using he ollowing indica o s: laying da e, clu ch size, and numbe o ledged young (Table 1). Bo h adul s we e cap u ed when he chicks we e 6–7 days old. While in he hand, he iden i y o a male was con i med o ela e o he same e i o ial male ha was obse ed in he icini y o he nes box du ing he a i al pe iod. F om he whole b eed- ing popula ion, we excluded he ones who abandoned hei nes o whose nes s we e p eda ed a an ea ly s age (and hus, hei iden i y could no be e i ied), o who we e andomly chosen o o he s udy expe imen s ha migh comp omise ma e choice, clu ch size, o numbe o ledglings. S a is ical analyses We used s uc u al equa ion modeling (SEM), o in his case pa h analysis, o s udy he s eng h o na u al selec- ion on male a i al da e, emale laying da e, and clu ch size by es ima ing di ec ional selec ion di e en ials o hese ai s (Scheine e al. 2000). The logic o using SEM in s udies o na u al selec ion is shown in Figu e 1. The ela i e numbe o o sp ing p oduced ( ledglings) was used as annual i ness measu e he e. The di ec ional selec- ion di e en ial is es ima ed by he co a iance be ween a ai and i ness (Fig. 1). I can be decomposed in o di ec and indi ec selec ion, which, in u n, may in ol e se e al di e en pa hways, depending on model complexi y (Scheine e al. 2000). In SEM amewo k, di ec selec ion on a ai is es ima ed by summing i s di ec e ec on i - ness and indi ec e ec s h ough ai s ha go o wa d owa d i ness in a SEM diag am (Scheine e al. 2000; Fig. 1). I se e al such media ing ai s a e included in he model, he indi ec e ec o di ec selec ion can be u he decomposed in o speci ic indi ec e ec s (Fig. 1). The di ec e ec o a ai on i ness is he selec ion g adien , o he slope o he (pa ial) eg ession coe icien (Lande and A nold 1983). Indi ec selec ion (o noncausal selec- ion due o sha ed causes) on a ai is es ima ed by sum- ming i s e ec s h ough backwa d-going ai s ha a e connec ed o i ness in a pa h diag am (Scheine e al. 2000). Again, depending on model complexi y and on he posi ion o a a iable o in e es in a s uc u al model, indi ec selec ion (like di ec selec ion) can be decom- posed in o speci ic indi ec e ec s (Fig. 1). Based on he ecology o mig an bi ds and on he ques- ions asked in his s udy, we o mula ed an a p io i s uc- u al equa ion model assuming ha he male a i al da e Table 1. Desc ip i e s a is ics o he a iables used in he s uc u al equa ion modeling, pooled o e he whole s udy pe iod. In he selec- ion analysis, howe e , annual ai means we e used in s anda diza- ion. In male a i al da e and emale laying da e, he i s e en was gi en alue he 0, and subsequen e en s a e days a e he i s e en . Mean Min–max SD N % missing alues A i al da e 13.40 0–31 5.74 363 – Laying da e 14.70 0–30 4.42 363 – Clu ch size 6.56 2–9 0.80 363 – Numbe o ledglings 5.70 0–9 1.50 261 28.1 ª2015 The Au ho s. Ecology and E olu ion published by John Wiley & Sons L d. 1207 W. Velmala e al. Selec ion o Ea lie A i al in a Songbi d has a di ec e ec on i ness as well as an indi ec e ec on i ness h ough bo h emale laying da e and clu ch size (Fig. 2). The model also assumes ha he emale laying da e has a di ec e ec on i ness bu also an indi ec e ec ia clu ch size and ha clu ch size has a di ec e ec on i ness only (Fig. 2). Because he ai s included in ou analysis di e ed in hei measu emen scale and sample a iance (Table 1), we ollowed He e o d e al. (2004) o ob ain mean-s anda dized selec ion es ima es and di ided all ai s (including ledgling numbe o mea- su e ela i e i ness) by hei annual ai means p io o analysis. The esul ing mean-s anda dized selec ion coe i- cien s ep esen he p opo ional change in i ness o a p opo ional change in he mean o he ai in ques ion. This makes i possible o compa e ou esul s wi h o he ai s, popula ions, o species mo e eliably had we used s anda diza ion based on pheno ypic s anda d de ia ions (He e o d e al. 2004; Ma sumu a e al. 2012). Howe e , mean s anda diza ion equi es a iables wi h na u al o i- gin ha can be es ablished as equal wi hin di e en s ud- ies (He e o d e al. 2004). The e o e, o bo h male a i al da e and emale laying da e, he da e o he i s e en (i.e., he a i al da e o he i s male o he i s laying da e o he season) was gi en he alue o ze o and he subsequen e en s we e sco ed by days since ha e en on a yea ly basis (Houle e al. 2011). Howe e , we use e ms “a i al da e” and “laying da e” (ins ead o “day”) h oughou he a icle in spi e o he abo e-men- ioned ans o ma ion. The i o he a p io i model o he obse ed da a was examined using he chi-squa e es ( 2 ) and he ollowing i indices: he oo mean squa e e o o app oxima ion (RMSEA), s anda dized oo mean squa e esidual (SRMR), he compa a i e i index (CFI), and he Tucke –Lewis index (TLI) (Wes e al. 2012). Bo h RMSEA and SRMR a e badness-o - i measu es, whe e 0 indica es a pe ec i o he model. In con as , in bo h CFI and TLI, a alue app oaching 1 indica es good model i (Wes e al. 2012). RMSEA has he added bene i o p o iding 90% con idence in e als o he es ima e, and i can be used o es he null hypo hesis ha he es ima e is <0.5, indica ing a good i (Wes e al. 2012). The ough cu o alues used o indica e a well- i ing model o SRMR, CFI, and TLI we e <0.08, >0.95, and >0.95, espec i ely (Wes e al. 2012). As we had epea ed obse - a ions o some males (n=42) om di e en yea s, we used a design-based clus e ing me hod ha co ec s o nonindependence o da a poin s by adjus ing pa ame e SEs wi hou explici ly es ima ing his dependency, as ca - ied ou in mixed modeling (Kal on 1977). Model pa am- e e s we e es ima ed using obus maximum-likelihood es ima ion (MLR) ha allows o non-no mal esponse dis ibu ions (he e, he ela i e ledgling numbe had a skewness o 1.41 and a ku osis o 2.47), and missing da a we e handled using ull-in o ma ion maximum like- lihood (FIML). Analyses we e conduc ed wi h Mplus ( e sion 7.3; Mu h en and Mu h en 1998–2012). Resul s The a p io i s uc u al equa ion model showed good i o he da a (n=363, 2ml =0.24, d =1, P=0.63; RMSEA (90% CIs) =0.00 (0.00, 0.11), P=0.75; CFI =1.00; TLI =1.08; SRMR =0.007). The model indica ed s a is i- cally signi ican di ec ional selec ion di e en ial o ea lie male a i al da e and emale laying da e and o la ge emale clu ch size (Table 2). A 10% shi owa d an ea - lie male a i al da e and emale laying da e gi es an expec ed p opo ional inc ease in i ness by 0.56% and 0.81%, espec i ely (Table 2). A 10% la ge clu ch size inc eases i ness by 9.65% (Table 2). The di ec ional selec- ion di e en ials o male a i al da e and emale laying da e did no s a is ically di e om each o he (z=0.68, P=0.50), bu selec ion di e en ial o clu ch size was sig- ni ican ly s onge han ha o male a i al da e (z=10.7, P<0.0001) and emale laying da e (z=10.6, P<0.0001). Indi ec selec ion o male a i al da e was no de ined in ou model, and hus, i s di ec ional selec ion di e en- ial equaled di ec selec ion (Table 2). The es ima es o indi ec selec ion o emale laying da e and clu ch size we e e y small and s a is ically nonsigni ican , showing Selec ion di e en ial Di ec selec ion Indi ec selec ion Di ec e ec s Indi ec e ec s Speci ic (non-causal) indi ec e ec s Speci ic (causal) indi ec e ec s Figu e 1. The pa i ioning o selec ion di e en ial in o di ec and (noncausal) indi ec selec ion. Di ec selec ion can be u he decomposed o di ec e ec s (i.e., selec ion g adien s) and o indi ec e ec s. Depending on a model s uc u e and complexi y, indi ec e ec s o di ec selec ion as well as indi ec selec ion can be u he decomposed o hei espec i e speci ic indi ec e ec s. 1208 ª2015 The Au ho s. Ecology and E olu ion published by John Wiley & Sons L d. Selec ion o Ea lie A i al in a Songbi d W. Velmala e al. ha di ec ional selec ion di e en ial o hese ai s was mainly due o di ec selec ion (Table 2). Selec ion g adi- en s (i.e., di ec e ec s) o all ai s we e s a is ically sig- ni ican (Fig. 2; Table 2). The o al indi ec e ec s o di ec selec ion h ough o wa d media ing ai s included in he model we e no s a is ically signi ican o any o he ai s s udied (Table 2), indica ing ha , o example, selec ion o male a i al da e was owing o i s di ec e ec on i ness. The pai wise associa ions in he aw da a be ween all ai s used in he analysis (male a i al da e, emale laying da e, clu ch size, and numbe o ledglings) a e shown in Figu e 3. Discussion We in es iga ed he s eng h o na u al selec ion on im- ing o male sp ing a i al in he pied lyca che by aking ad an age o s uc u al equa ion modeling (SEM). SEM enabled he ealis ic modeling o selec ion episodes on empo ally o de ed b eeding ai s. We ound di ec ional selec ion di e en ial o an ea lie male a i al da e. Impo an ly, his selec ion coe icien esul ed om he di ec e ec o male a i al da e on i ness and no om he indi ec e ec s on i ness h ough emale ai s, ha is, laying da e and clu ch size. The magni ude o selec ion on he male a i al da e was simila o selec ion o ea - lie emale laying da e. Bo h o hese we e, howe e , se - e al o de s o magni ude lowe han selec ion o la ge clu ch sizes, which logically a ises om he s ong co ela- ion be ween clu ch size and numbe o ledglings in he cu en sample. A weal h o p e ious s udies has shown co ela ions be ween he iming o b eeding (i.e., laying da e) and b eeding success ( o he pied lyca che , see, e.g., Lund- be g and Ala alo 1992; Canal e al. 2012), including he associa ions be ween male a i al da e and ep oduc i e success (e.g., Ala alo e al. 1984; Aebische e al. 1996; Hasselquis 1998; No is e al. 2004; Rockwell e al. 2012). Howe e , only ew s udies ha e in es iga ed how na u al selec ion ac s on sp ing a i al da es by quan i y- ing selec ion coe icien s (Mølle e al. 2008a; Gienapp and B egnballe 2012; A naud e al. 2013), and e en ewe ha e examined he pa hways linking a i al ime wi h i - ness. No is e al. (2004) s udied he Ame ican eds a and es ima ed he di e en pa hways o selec ion on a i- al da e o i ness due o di ec and indi ec e ec s, bu hey did no examine he s eng h o na u al selec ion (i.e., he di ec ional selec ion di e en ial) o male a i al da e o o emale ai s. The cu en s udy he e o e seems o be he i s one ully u ilizing he insigh ob ained om he SEM o s udying na u al selec ion on iming o sp ing a i al in a mig an bi d. Ou esul s show, o he i s ime, a di ec e ec o male a i al da e on i ness while accoun ing o hose indi ec e ec s ha a e media ed by po en ial e ec s o he social pa ne . In o he wo ds, di ec ional selec ion di e en ial and di ec selec ion o ea lie male a i al we e due o i s di ec e ec on i ness and no due o i s Male a i al da e Female laying da e Clu ch size Numbe o ledglings –0.04 (–0.11, 0.03) –0.07 (–0.11, –0.03) –0.06 (–0.11, –0.01) 0.96 (0.78, 1.15) –0.01 (–0.04, 0.01) 0.26 (0.21, 0.30) 0.01 (0.01, 0.02) 0.05 (0.03, 0.07) Figu e 2. The s uc u al equa ion model used o es ima e di ec ional selec ion di e en ial on male a i al da e, emale laying da e, and clu ch size. Rela i e numbe o ledglings was used as a p oxy o i ness. A ows be ween boxes ep esen assumed causal associa ions be ween he mean-s anda dized ai s, and open sho a ows deno e esidual a iances o dependen a iables. Di ec e ec s (pa hs) om a ai o i ness can be conside ed selec ion g adien s. Numbe s in pa en heses a e 95% con idence in e als. Table 2. The p opo ional model-p edic ed di ec ional selec ion di e en ial o male a i al da e, emale laying da e, and clu ch size on numbe o ledglings (i.e., i ness), decomposed in o di ec and indi ec selec ion, whe e di ec selec ion is u he decomposed in o di ec and indi ec e ec s. Numbe s in pa en heses a e 95% con idence in e als. A i al da e Laying da e Clu ch size Di ec ional selec ion di e en ial 0.056 (0.108, 0.005) 0.081 (0.129, 0.034) 0.965 (0.783, 1.147) Di ec selec ion 0.056 (0.108, 0.005) 0.084 (0.132, 0.036) 0.964 (0.782, 1.146) Di ec e ec (o selec ion g adien ) 0.059 (0.111, 0.008) 0.070 (0.114, 0.027) 0.964 (0.782, 1.146) Indi ec e ec s 0.003 (0.003, 0.009) 0.013 (0.034, 0.007) – Indi ec selec ion –0.002 (0.003, 0.007) 0.001 (0.001, 0.002) ª2015 The Au ho s. Ecology and E olu ion published by John Wiley & Sons L d. 1209 W. Velmala e al. Selec ion o Ea lie A i al in a Songbi d indi ec e ec s ia emale ai s. While emale ai s mos likely in luence b eeding pe o mance, o example, h ough iming o laying, clu ch size, and le el o pa en al ca e, he male a i al da e seems o ha e a di ec e ec on i ness. Ou esul s hus sugges ha ea ly-a i ing males a e ei he o a highe pheno ypic quali y, ha e be e genes, o a e able o gain be e esou ces, such as high- quali y b eeding e i o y, compa ed o la e -a i ing males. A e iew by He e o d e al. (2004) used 38 s udies ha had epo ed he necessa y in o ma ion o he pos calcu- la ion o mean-s anda dized selec ion g adien s (o which nine we e on bi ds). In hose 38 s udies, he median absolu e alue o mul i a ia e mean-s anda dized selec ion g adien s was 0.86 o li e his o y ai s, 0.38 o ecun- di y, and 0.54 o all es ima es (He e o d e al. 2004). The selec ion g adien s ( he di ec e ec s) o he male a i al da e and emale laying da e (0.06 and 0.07, espec- i ely) in ou s udy can hus be conside ed mino . In con as , he selec ion g adien o clu ch size (0.96) indi- ca es s ong di ec selec ion. Howe e , compa ing selec- ion g adien s, o selec ion di e en ials in pa icula (Scheine e al. 2000), in ou s udy o hose p esen ed in He e o d e al. (2004) is no s aigh o wa d due o Figu e 3. Associa ions be ween all ai s used in he s uc u al equa ion modeling (SEM). All alues a e s anda dized by hei annual mean. 1210 ª2015 The Au ho s. Ecology and E olu ion published by John Wiley & Sons L d. Selec ion o Ea lie A i al in a Songbi d W. Velmala e al. me hodological di e ences. In he cu en s udy, we ha e aken in o accoun he empo al s uc u e o he pied ly- ca che b eeding cycle, which gi es a biologically mo e ealis ic and causally de ined model o na u al selec ion, whe eas mul iple eg ession models, assuming no such hie a chy be ween ai s, a e adi ionally used in selec ion analyses (He e o d e al. 2004; Mo issey 2014). Al hough no ully excluding he possibili y o equi alen models (i.e., models assuming a di e en causal s uc u e ha also show simila i o he da a), ou model i ed he da a well, sugges ing ha he selec ion es ima es ob ained can be conside ed o ha e low bias. We he e o e encou - age u u e s udies o na u al selec ion o use SEM in o de o base he selec ion es ima es on mo e causally plausible models o na u al selec ion in he wild. I should be no ed ha ou s udy only conside s selec- ion on pa en al ecundi y bu no on su i al. The e may also be cos s o , and selec ion agains , ea ly mig a ion and a i al due o ha sh wea he and ood sho age (Mølle 1994; Kokko 1999; B own and B own 2000; Mølle e al. 2008b). Un o una ely, hese a e ex emely di icul ques- ions o s udy in a small songbi d, such as he pied ly- ca che , un il i is possible o ack hem indi idually h oughou hei annual cycle. The impo ance o he ade-o s be ween he po en ial su i al isks associa ed wi h ea ly mig a ion and he bene i s o ea ly a i al o b eeding g ounds hus emain o be s udied. I also emains o be examined how indi idual condi ion (s a e) a ec s hese ade-o s: The isk may be mo e modes i ea ly-a i ing males a e also highe -quali y indi iduals in he sense ha hey a e mo e likely o ou li e disad an a- geous condi ions (e.g., Mølle 1994). In ac , he ad an- age o ea ly b eeding is emphasized when empe a u e p io o b eeding is low, possibly indica ing highe -quali y indi iduals o be less in luenced by ha sh condi ions (Ah- ola e al. 2012). I can be a gued ha he numbe o ec ui s could ha e been used as a i ness p oxy ins ead o he numbe o ledglings, bu se e al au ho s ha e concluded ha ledg- ling numbe is a ele an i ness measu e (e.g., B omme e al. 2004) o e en a gued agains assigning i ness ac oss gene a ions (e.g., Lande and A nold 1983; Che e ud and Moo e 1994; Wol and Wade 2001). In addi ion, in his species, he dispe sal dis ibu ion is pa icula ly wide (Lundbe g and Ala alo 1992; Lehikoinen 2014) and, con- sequen ly, he numbe s o ec ui s eco ded a pa en al b eeding g ounds a e low. The e o e, he numbe o ec ui s is no an app op ia e o eliable i ness measu e in his species. In conclusion, we ha e shown ha he e is di ec ional selec ion o ea ly male a i al o b eeding g ounds in a mig a o y bi d. By aking in o accoun he sequen ial na - u e o he ecundi y ai s in ques ion, we es ima ed he s eng h o selec ion in a s anda dized manne as well as de ined he speci ic pa hways o di ec ional selec ion o ea lie male a i al da e. We wan ed o s udy whe he an associa ion be ween a i al da e and ledgling numbe migh a ise only h ough emale ai s, such as he laying da e o clu ch size, bu i does no seem o be so. As ou s udy emphasizes, he iming o di e en pe iods wi hin he annual cycle can clea ly be associa ed wi h di ec i - ness consequences. By s udying how na u al selec ion ac s on each o hese pe iods, i will be possible o iden i y wha he ela i e impo ance is o hese pe iods o bi ds, and gain impo an knowledge o e alua ing he po en- ial o mig an s o cope when he en i onmen changes. Acknowledgmen s We hank Sa a Calhim, Joe He e o d, and Ha y Ko hals o hei help wi h he s udy. Suzanne Collins om he Uni e si y o Tu ku G adua e School (UTUGS) made nume ous co ec ions o he manusc ip . The s udy was inancially suppo ed by he Academy o Finland (p ojec no. 130436 o TL). SH was unded by The Tu ku Colle- gium o Science and Medicine. Con lic o In e es The au ho s decla e hey ha e no con lic o in e es s. Re e ences Aebische , A., N. Pe in, M. K ieg, J. S ude , and D. R. Meye . 1996. The ole o e i o y choice, ma e choice and a i al da e on b eeding success in he Sa i’s Wa ble Locus ella luscinioides. J. A ian Biol. 27:134–152. Ahola, M. P., T. Laaksonen, T. Ee a, and E. Lehikoinen. 2012. Selec ion on laying da e is connec ed o b eeding densi y in he pied lyca che . Oecologia 168:703–710. Ala alo, R. V., A. Lundbe g, and K. S  ahlb and . 1984. Female ma e choice in he pied lyca che Ficedula hypoleuca. Beha . Ecol. Sociobiol. 14:253–261. A naud, C. M., P. H. Becke , F. S. Dobson, and A. Cha man ie . 2013. Canaliza ion o phenology in common e ns: gene ic and pheno ypic a ia ions in sp ing a i al da e. Beha . Ecol. 24:683–690. B omme , J. E., L. Gus a sson, H. Pie i€ ainen, and J. Me il€ a. 2004. Single-gene a ion es ima es o indi idual i ness as p oxies o long- e m gene ic con ibu ion. Am. Na . 163:505–517. B own, C. R., and M. B. B own. 2000. Wea he -media ed na u al selec ion on a i al ime in cli swallows (Pe ochelidon py hono a). Beha . Ecol. Sociobiol. 47:339–345. Canal, D., R. Jo ani, and J. Po i. 2012. Mul iple ma ing oppo uni ies boos p o and y in a pied lyca che popula ion. Beha . Ecol. Sociobiol. 66:67–76. ª2015 The Au ho s. Ecology and E olu ion published by John Wiley & Sons L d. 1211 W. Velmala e al. Selec ion o Ea lie A i al in a Songbi d Ca ey, C. 2009. The impac s o clima e change on he annual cycles o bi ds. Phil. T ans. R. Soc. 364:3321–3330. Che e ud, J. M., and A. J. Moo e. 1994. Quan i a i e gene ics and he ole o he en i onmen p o ided by ela i es in he e olu ion o beha io . Pp. 67–100 in C. R. B. Boake, ed. Quan i a i e gene ic s udies o beha io al e olu ion. Uni e si y o Chicago P ess, Chicago. Coope , N. W., M. T. Mu phy, L. J. Redmond, and A. C. Dolan. 2011. Rep oduc i e co ela es o sp ing a i al da e in he Eas e n Kingbi d Ty annus y annus. J. O ni hol. 152:143–152. D os , R. 1936. € Ube das B u kleid m€ annliche T aue liegen € ange, Muscicapa hypoleuca. Vogelzug 6:179– 186. Gienapp, P., and T. B egnballe. 2012. Fi ness consequences o iming o mig a ion and b eeding in co mo an s. PLoS One 7:e46165. Go do, O., P. T yjanowski, J. Z. Kosicki, and M. Fulin. 2013. Complex phenological changes and hei consequences in he b eeding success o a mig a o y bi d, he whi e s o k Ciconia ciconia. J. Anim. Ecol. 82:1072–1085. Gunna sson, T. G., J. A. Gill, P. G. A kinson, G. G elinaud, P. M. Po s, R. E. C oge , e al. 2006. Popula ion-scale d i e s o indi idual a i al imes in mig a o y bi ds. J. Anim. Ecol. 75:1119–1127. Hasselquis , D. 1998. Polygyny in g ea eed wa ble s: a long- e m s udy o ac o s con ibu ing o male i ness. Ecology 79:2376–2390. He e o d, J., T. F. Hansen, and D. Houle. 2004. Compa ing s eng hs o di ec ional selec ion. How s ong is s ong? E olu ion 58:2133–2143. H€ o ke , H. 2002. A i al o Pied A oce s Recu i os a a ose a a he b eeding si e: e ec s o win e qua e s and consequences o ep oduc i e success. Pp. 379–387 in C. Bo h and T. Pie sma, eds. The a ian calenda : explo ing biological hu dles in he annual cycle. P oceedings o he 3 d Con e ence o Eu opean O ni hologis s’ Union, G oningen, Augus 2001. A dea 90 (special issue). Houle, D., C. Pelabon, G. P. Wagne , and T. F. Hansen. 2011. Measu emen and meaning in biology. Q. Re . Biol. 86:3– 34. Kal on, G. 1977. P ac ical me hods o es ima ing su ey sampling e o s. Bull. In . S a . Ins . 47:495–514. Knudsen, E., A. Lind en, C. Bo h, N. Jonz en, F. Pulido, N. Saino, e al. 2011. Challenging claims in he s udy o mig a o y bi ds and clima e change. Biol. Re . 86:928– 946. Kokko, H. 1999. Compe i ion o ea ly a i al in mig a o y bi ds. J. Anim. Ecol. 68:940–950. Lande, R., and S. J. A nold. 1983. The measu emen o selec ion on co ela ed cha ac e s. E olu ion 37:1210–1226. Lehikoinen, E. 2014. Eu opean pied lyca che . Pp. 542–551. in J. Valkama, P. Sau ola, A. Lehikoinen, E. Lehikoinen, M. Piha, P. Sola, and W. Velmala, eds. The innish bi d inging a las. Vol II. Finnish Museum o Na u al His o y and Minis y o En i onmen , Helsinki. Lundbe g, A., and R. V. Ala alo. 1992. The pied lyca che . T. and A. D. Poyse , London, U.K.. Ma sumu a, S., R. A linghaus, and U. Dieckmann. 2012. S anda dizing selec ion s eng hs o s udy selec ion in he wild: a c i ical compa ison and sugges ions o he u u e. Bioscience 62:1039–1054. McNama a, J. M., Z. Ba a, M. Klaassen, and S. Baue . 2011. Cues and he op imal iming o ac i i ies unde en i onmen al changes. Ecol. Le . 14:1183–1190. Mølle , A. P. 1994. Pheno ype-dependen a i al ime and i s consequences in a mig a o y bi d. Beha . Ecol. Sociobiol. 35:115–122. Mølle , A. P. 2007. Ta dy emales, impa ien males: p o and y and di e gen selec ion on a i al da e in he wo sexes o he ba n swallow. Beha . Ecol. Sociobiol. 61:1311–1319. Mølle , A. P., J. Balbon  ın, J. J. Cue o, I. G. He mosell, and F. de Lope. 2008a. Indi idual di e ences in p o and y, sexual selec ion, and i ness. Beha . Ecol. 20:433–440. Mølle , A. P., D. Rubolini, and E. Lehikoinen. 2008b. Popula ions o mig a o y bi d species ha did no show a phenological esponse o clima e change a e declining. P oc. Na l Acad. Sci. USA 105:16195–16200. Mo bey, Y. E., and R. C. Ydenbe g. 2001. P o and ous a i al iming o b eeding a eas: a e iew. Ecol. Le . 4:663–673. Mo issey, M. B. 2014. Selec ion and e olu ion o causally co a ying ai s. E olu ion 68:1748–1761. Mu h en, L. K., and B. O. Mu h en. 1998–2012. Mplus use ’s guide, 7 h edn. Mu h en and Mu h en, Los Angeles, CA. No is, D. R., P. P. Ma a, T. K. Kyse , T. W. She y, and L. M. Ra cli e. 2004. T opical win e habi a limi s ep oduc i e success on he empe a e b eeding g ounds in a mig a o y bi d. P oc. R. Soc. Lond. 271:59–64. Po i, J. 1998. A i al ime om sp ing mig a ion in male Pied Flyca che s: indi idual consis ency and amilial esemblance. Condo 100:702–708. Reudink, M. W., P. P. Ma a, T. K. Kyse , P. T. Boag, K. M. Langin, and L. M. Ra cli e. 2009. Non-b eeding season e en s in luence sexual selec ion in a long-dis ance mig a o y bi d. P oc. R. Soc. Lond. 276:1619–1626. Rockwell, S. M., C. I. Boce i, and P. P. Ma a. 2012. Ca y- o e e ec s o win e clima e on sp ing a i al da e and ep oduc i e success in an endange ed mig a o y bi d, Ki land’s Wa ble (Se ophaga ki landii). Auk 129:744–752. Rubolini, D., F. Spina, and N. Saino. 2004. P o and y and sexual dimo phism in ans-Saha an mig a o y bi ds. Beha . Ecol. 15:592–601. Scheine , S. M., R. J. Mi chell, and H. S. Callahan. 2000. Using pa h analysis o measu e na u al selec ion. J. E ol. Biol. 13:423–433. Se gio, F., J. Blas, M. G. Fo e o, J. A. Don aza , and F. Hi aldo. 2007. Sequen ial se lemen and si e dependence in a mig a o y ap o . Beha . Ecol. 18:811–821. 1212 ª2015 The Au ho s. Ecology and E olu ion published by John Wiley & Sons L d. Selec ion o Ea lie A i al in a Songbi d W. Velmala e al. Smi h, R. J., and F. R. Moo e. 2005. A i al iming and seasonal ep oduc i e pe o mance in a long-dis ance mig a o y landbi d. Beha . Ecol. Sociobiol. 57: 231–239. Teplinsky, C., N. G. Mouawad, J. Balbon in, F. de Lope, and A. P. Mølle . 2011. Quan i a i e gene ics o mig a ion synd omes: a s udy o wo ba n swallow popula ions. J. E ol. Biol. 24:2025–2039. T yjanowski, P., T. H. Spa ks, J. P aszyk, and J. Kosicki. 2004. Do Whi e S o ks Ciconia ciconia always p o i om an ea ly e u n o hei b eeding g ounds? Bi d S udy 51:222–227. Wes , S. G., A. B. Taylo , and W. Wu. 2012. Model i and model selec ion in s uc u al equa ion modeling. Pp. 209– 231 in R. H. Hoyle, ed. Handbook o s uc u al equa ion modeling. The Guil o d P ess, New Yo k, NY. Wol , J. B., and M. J. Wade. 2001. On he assignmen o i ness o pa en s and o sp ing: whose i ness is i , and when does i ma e ? J. E ol. Biol. 14: 347–356. ª2015 The Au ho s. Ecology and E olu ion published by John Wiley & Sons L d. 1213 W. Velmala e al. Selec ion o Ea lie A i al in a Songbi d