Phylogeny, historical biogeography and the evolution of migration in accipitrid birds of prey (Aves: Accipitriformes)

O nis Hunga ica 2014. 22(1): 15–35. doi: 10.2478/o hu-2014-0008 Phylogeny, his o ical biogeog aphy and he e olu ion o mig a ion in accipi id bi ds o p ey (A es: Accipi i o mes) Jenő nagy1 & Jácin ökölyi2,* Jenő Nagy & Jácin Tökölyi 2014. Phylogeny, his o ical biogeog aphy and he e olu ion o mig a ion in accipi id bi ds o p ey (A es: Accipi i o mes). – O nis Hunga ica 22(1): 15–35. Abs ac Mig a ion plays a undamen al pa in he li e o mos empe a e bi d species. The e gu la , la gescale seasonal mo emen s ha cha ac e ize empe a e mig a ion sys ems appea o ha e o igina ed in pa allel wi h he pos glacial no he n expansion o opical species. Mig a o iness is also in- luenced by a numbe o ecological ac o s, such as he abili y o su i e ha sh win e s. Hence, unde s anding he o igins and e olu ion o mig a ion equi es in eg a ion o he biogeog aphic his o y and ecology o bi ds in a phylogene ic con ex . We used molecula da ing and ances al s a e econs uc ion o in e he o igins and e olu- iona y changes in mig a o y beha io and ances al a ea econs uc ion o in es iga e his o ical pa e ns o ange e olu ion in accipi id bi ds o p ey (Accipi i o mes). Mig a ion e ol ed mul iple imes in bi ds o p ey, he ea - lies o which occu ed in ue hawks (Accipi inae), du ing he middle Miocene pe iod, acco ding o ou analy- ses. In mos cases, a opical ances al dis ibu ion was in e ed o he nonmig a o y ances o s o mig a o y line- ages. Resul s om di ec ional e olu iona y es s indica e ha mig a ion e ol ed in he opics and hen inc eased he a e o coloniza ion o empe a e habi a s, sugges ing ha empe a e species migh be descendan s o opi- cal ones ha dispe sed in o hese seasonal habi a s. Finally, we ound ha die gene aliza ion p edic s mig a o i- ness in his g oup. Keywo ds: ances al a ea econs uc ion, annual cycle, compa a i e, die specializa ion, diu nal bi ds o p ey, mo- lecula da ing, seasonali y Össze oglalás A leg öbb mé sékel ö i madá aj éle ciklusában alap e ő sze epe öl be a onulás. A endsze es, nagy ki e jedésű mozgások, melyek a mé sékel ö i onulási endsze eke jellemzik, egyes el é elezések sze in a ópusi ajok posz glaciális, északi i ányú e jeszkedésé el pá huzamosan jelen ek meg. Ezen elül a onulás elő o dulásá számos ökológiai ényező is be olyásolha ja, min például a kö nyeze szezonali ásának mé éke agy a éli úlélés be olyásoló ényezők. A onulás e ede e és e olúciója ezé csak úgy é he ő meg, ha a mada ak biogeog á iai ö é- ne iségé és ökológiájá ilogene ikai kon ex usban anulmányozzuk. Jelen izsgála ban a ágómadá alakúak (Acci- pi i o mes) onulásának e olúciójá elemez ük kompa a í módsze ekkel. Első lépésben lé ehoz unk egy osszilis ada ok alapján da ál molekulá is ö zs á , amelyen jelleg ekons ukció égez ünk és ekons uál uk a ajok ősi el e - jedési e üle é . Az elemzéseink alapján a onulás öbbszö alakul ki a agadozók ese ében, legko ábban a héja o - mákon (Accipi inae) belül, élhe ően a Miocén közepén. A leg öbb ese ben a onuló leszá mazási onalak nem o- nuló őseinél ópusi el e jedés e kö e kez e he ünk. A di ekcionális e olúciós esz alapján a onulás a ópusokon jelen meg és megnö el e a mé sékel égö kolonizációjának á ájá . Esze in ehá a mé sékel ö i agadozómadá a- jok onuló ópusi ajok leszá mazo ainak ekin he ők, melyek az e ősen szezonális, északi élőhelyek i ányába e - jeszked ek. Végeze ül nega í kapcsola o alál unk a onulás megjelenése és a áplálékspecializáció mé éke közö . Kulcssza ak: é es ciklus, jelleg ekons ukció, kompa a í , molekulá is da álás, nappali agadozómada ak, sze- zonali ás, áplálékspecializáció 1 MTA-ELTE-MTM Ecology Resea ch G oup, 1117 Budapes , Pázmány Pé e sé ány 1/C, Hunga y, e-mail: jenon- agy[email p o ec ed] 2 MTA-DE “Lendüle ” Beha iou al Ecology Resea ch G oup, Depa men o E olu iona y Zoology and Human Biology, Uni e si y o Deb ecen, 4032 Deb ecen, Egye em é 1., Hunga y * co esponding au ho : [email p o ec ed] ORNIS HUNGARICA 2014. 22(1) 16 In oduc ion Bi ds o igina ed, acco ding o ou cu en knowledge o he ossil eco d, abou 150 200 million yea s ago du ing he geologi- cal e a o Ju assic (Padian & Chiappe 1998). The appea ance o powe ed ligh , p o bab ly in combina ion wi h se e al o he a i- an ea u es such as wa mbloodedness and he p esence o ex ensi e pa en al ca e has uelled he di e si ica ion o his g oup o e eb a es, which seems o ha e accele a ed a ound, o sho ly a e he C e aceous Pa leo gene bounda y (E icson e al. 2006, B own e al. 2008, Je z e al. 2012). The widesp ead occu ence o bi ds is g ea ly acili a ed by hei excellen dis- pe sal capabili ies. This is pe haps mos clea ly seen in mig a o y bi ds, which can a el housands o kilome es on con inen- al scale wi hin a single yea . Mig a ion is a cha ac e is ic ea u e o bi ds ha is ex- emely common especially in species in- habi ing he No he n Hemisphe e empe a e zone and he A c ic, bu i also occu s, al hough in less ex eme o ms, in o he e- gions o he globe in he o m o in a opi- cal mig a ion sys ems (Ale s am 1993, Be  hold 2001, New on 2008). Mig a ion i sel p o oundly in luences he dis ibu ion, ecology and di e si ica ion o bi ds (New on 2008), hence i is no su p i sing ha a g ea deal o in o ma ion has ac- cumula ed on i s in e nal, p oxima e de e - minan s and i s phenology e e since Hans Ch is ian Mo ensen s a ed inging bi ds a he e y end o he 19 h cen u y. These s u dies e ealed ha he mig a o y pheno ype is de- e mined by a se o complex and igh ly e gu la ed mechanisms (Gwinne 1990), which includes, among o he s (1) senso y elemen s unde lying o ien a ion and na iga ion, (2) mechanisms esponsible o he egula ion o mig a o y es lessness (‘zugun uhe’) du ing he annual cycle and (3) a ange o physio- logical adap a ions ha co e he me abolic equi emen s o longdis ance ligh s du ing mig a ion. Unde s anding how such a comp lex pheno ype could ha e e ol ed is a majo challenge in o ni hology. While a weal h o in o ma ion has accu- mula ed on he de ails o he p ocess o mig a ion, compa a i ely li le is known abou how mig a ion o igina ed and e ol ed in bi ds. This is no su p ising, since beha io- al ai s, such as he mig a o y ha bi s o a species do no ossilize and hence ou cu - en ideas o i a e s ic ly in e ed om phylogene ic o biogeog aphic s udies. Cu - en heo ies o he e olu ion o mig a- ion can be di ided in o wo g oups (Rap- pole & Jones 2002, B ude e & Salewski 2008). The ‘ opical o igin’ hypo hesis p o- poses ha mig a o y bi ds de i e om spe- cies inhabi ing egions whe e en i onmen- al ac o s we e cons an du ing he yea , so he e was no need o mig a ion. These species could ha e colonized mo e sea- sonal, no he n habi a s, which, du ing he summe mon hs p o ided app op ia e con- di ions o ep oduc ion. Howe e , du ing win e ood a ailabili y dec eased, hence hese bi ds we e o ced o e u n o sou he n la i udes (Rappole & Jones 2002, B ude e & Salewski 2008). Cox (1985) de eloped a s eppings one model o his hypo hesis. Acco ding o his model, esou celimi a- ion due o compe i ion o ood o ced ce - ain opical esiden species o expand hei ange o he sub opics. These bi ds con i nu ed o e u n o he opics du ing he win e , esul ing in he o ma ion o pa ial mig an species. These pa ial mig an s hen con inu ed o sp ead o highe la i udes whe e hey we e able o b eed success ully while s ill e u ning o he o iginal a ea in he win e 17 J. Nagy & J. Tökölyi (S iles 1980, Cox 1985). Thus, his hypo he sis p edic s ha mig a o y species e ol ed om opical ances o s. Se e al lines o e idence suppo he o pi cal o igin o longdis ance mig an s. Jo- seph e al. (1999), in a s udy o wade s, analyzed he e olu ion o b eeding and win- e ing dis ibu ion o 16 species om he genus Cha ad ius using phylogene ic me hods. By econs uc ing he hypo hesized dis ibu ion o he ances o s o hese bi ds, hey showed ha species ha a e mig a o y oday de i e om ances o s whose b ee ding and nonb eeding anges we e loca ed in he opical zone. Ano he simila s udy in es- iga ed he e olu ion o mig a ion in Ca- ha us h ushes (Ou law e al. 2003). This s udy showed ha No h Ame ican (mig a- o y) h ushes a e sis e o opical species, and he ances al a ea o he whole lineage was in e ed o be in he Neo opics, p o- iding u he suppo o he “ opical o i- gin” hypo hesis. This pa e n is no es ic ed o in e speci ic compa isons bu is also seen among popula ions di e ing in mig a- o y s a us. Fo ins ance, in a s udy o No h Ame ican Chipping Spa ows (Spizaella passe ina), Milá e al. (2006) ha e shown ha he no he n, longdis ance mig an popula ions descend om nonmig a o- y Mexican popula ions, which colonized No h Ame ica a e he las glacial maxi- mum 18,000 yea s ago. Thus, longdis ance mig a ion and coloniza ion o empe a e e- gions in his species de eloped in andem. O he hypo heses on he o igin o mig a- ion emphasize he impo ance o chan ges on he b eeding e i o ies o bi ds wi h a no he n dis ibu ion (‘no he n o igin’ hy- po hesis) (Bell 2000, 2005, B ude e & Salewski 2008). Acco ding o hese suppo- si ions, clima ic o o he ecological chan ges (e.g. global cooling) could ha e led o he e olu ion o mig a ion by o cing nonmig a o y empe a e and a c ic species o lea e he e home anges du ing he win e (Bell 2000, 2005), esul ing in mig a o y s a e- gies which allowed he su i al o popu- la ions in a s ongly seasonal milieu (Bell 2000, 2005). Thus, his hypo hesis p edic s ha mig a ion e ol ed om empe a e esi den species, a p edic ion ha has ecei ed ela i ely low suppo o da e. I is clea , howe e , ha mig a ion can e ol e wi h- ou he expansion o he b eeding anges, as exampli ied by he la ge numbe o in- a opical mig an s (e.g. Boyle & Conway 2007, Boyle e al. 2011). Compa a i e s u dies o he occu ence o mig a ion among some o hese opical axa, such as he pas- se ine g oup Ty anni e ealed ha a num- be o ecological ai s, speci ically die and habi a , p edic whe he a species is mig a- o y o no in he opics (Boyle & Conway 2007, Boyle e al. 2011). Thus, yea  ound a ia ion and p edic abili y o ood sou ces (Boyle & Conway 2007, Boyle e al. 2011), as well as he abili y o exploi hese sou  ces (Bell 2011, Boyle e al. 2011) appea s o p edispose some a ian axa o mig a ion. As he examples abo e sugges , he e olu- ion o mig a ion in bi ds is a complex p ob- lem ha equi es an in eg a i e app oach combining aspec s o he his o ical biogeo g aphy ( ange expansions), ecology (habi- a , ood a ailabili y) and beha io al ecology (die specializa ion) o bi ds. Ye , phyloge- ne ic s udies o en a ge only one o hese aspec s, while igno ing o he s. He e, we de- sc ibe an a emp o such an in eg a i e app oach using accipi id bi ds o p ey (Accipi i o mes) as a model g oup. This axon is ideal o ou pu pose because i includes bo h opical and empe a e species and he e is wide a ia ion in mig a o y beha io , habi- a and die wi hin he g oup (Fe gusonLees ORNIS HUNGARICA 2014. 22(1) 18 & Ch is ie 2001). Speci ically, we in es iga e he ollowing: (1) he phylogeny and his o i cal biogeog aphy o diu nal bi ds o p ey; (2) he e olu iona y o igins o mig a ion in ap o s in a phylogene ic con ex and (3) he ecological and beha io al ai s ha a e associa ed wi h mig a o iness in his g oup. The ai s we in es iga e a e body mass, die and habi a . Body mass in luences nea ly all aspec s o bi ds’ li e and could be a c ucial ac o de e mining which species can su - i e pe iods o ood sho age and/o cold whea he (e.g. New on 2008). The e o e, we hypo hesize ha la ge bi ds a e mo e like- ly o become esiden s (o ice e sa: esi- den s migh be selec ed o become la ge ) (Tökölyi & Ba a 2011). Die can be impo - an o wo easons: i s , ap o s eeding on wa mblooded p ey o ca cass a e mo e like- ly o su i e he win e a empe a e la i udes and hence be nonmig a o y. Second, species wi h a gene alis die should be mo e likely o subsis du ing pe iods o esou ce sho age by inding al e na i e ood sou ces, hence we p edic ha ood gene alis s a e mo e e- quen in mig an s. Las ly, habi a ype was hypo hesized o be impo an in he e olu- ion o opical empe a e mig a ion sys ems because i could ha e de e mined he a aila- bili y o sui able co ido s o opical em- pe a e dispe sal ou es (Rappole & Jones 2002). Rappole and Jones (2002) no ed ha he majo i y o longdis ance mig an s in he Nea c ic spend he win e in o es s, whe eas almos none o he Palea c ic/Asian mig an s do so. They p oposed ha he lack o o es ed habi a s in No h A ica could ha e ac ed as a dispe sal ba ie , e ec i ely il e ing ange expansions om sou h. The e o e, we es ed whe he habi a ype is associa ed wi h mig a o iness in bi ds o p ey. Me hods Phylogene ic econs uc ion and molecula da ing The lis o genes used o econs uc he molecula phylogeny o bi ds o p ey o Table 1. Gene sequences used o econs uc he phylogeny o Accipi i o mes, hei a ailabili y, leng h and he mos app op ia e e olu iona y model o sequence e olu ion applying o hem, as e alua ed by jModelTes 1. ábláza A ágómadá -alakúak ö zs a- ekons ukciójához használ génszek enciák ne e, elé he ő- sége ( ajok száma), hossza és a jModelTes ál al megha á ozo legmeg elelőbb e olúciós modell Gene (Abb e ia ion) No species No. bp Model 12S ibosomal RNA (12S) 74 900 TIM2+I+G 16S ibosomal RNA (16S) 53 1527 GTR+I+G ATP syn hase F0 subuni 6 (ATP6) 56 684 T N+I+G ATP syn hase F0 subuni 8 (ATP8) 56 168 T N+I+G β- ib onigen in on 7 (BFI7) 69 922 TVM+G Cy och ome c oxidase subuni 1 (COX1) 86 1551 TIM2+I+G Cy och ome b (CYTB) 164 1146 TIM3+I+G NADH-ubiquinone oxido educ ase subuni 2 (ND2) 151 1047 GTR+I+G NADH-ubiquinone oxido educ ase subuni 2 (ND6) 66 525 GTR+I+G Recombinase ac i a ing gene 1 (RAG1) 87 2872 GTR+I+G 19 J. Nagy & J. Tökölyi which gene ic da a is a ailable is shown in Table 1. All sequences we e e ie ed om GenBank (h p://ncbi.nlm.nih.go /). Se- quences we e aligned using MAFFTLINSI (Ka oh e al. 2005) wi h de aul pa ame e s and alignmen s we e isually checked. Two alignmen s (12S and 16S) con ained mul- iple indels and we e un h ough Gblocks (Cas esana 2000) o emo e poo ly aligned posi ions. Sequence managemen was done in he R s a is ical en i onmen (R De elop- men Co e Team 2012) using unc ions om lib a ies ape (Pa adis e al. 2004) and seqin (Cha i & Lob y 2007). Alignmen s we e conca ena ed and spe- cies wi h ew da a (<500 nucleo ides) we e emo ed. The median sequence leng h o he emaining 180 species was 1,038 base pai s ( ange 5192,872) (Table 1). This axo nomic sample ep esen s app oxima ely 70% o ex an species. The esul ing spa se supe ma ix was used o econs uc he phylogene ic ela- ionships o he 180 species. Fi s , RAxML (S ama akis 2006) was used o ob ain a s a ing ee o phylogeny es ima ion. We used a apid boo s ap analysis wi h 100 boo s ap eplica es ollowed by a sea ch o he bes sco ing maximum likelihood ee, using he GTR+I+G model o e olu ion. The Sec e a y Bi d Sagi a ius se pen inus (Sagi a iidae) was used as an ou g oup in his p ocess; he sis e ela ionship be ween Sagi a iidae and he es o he Accipi i- o mes (Pandionidae and Accipi idae) is well suppo ed om molecula phylogene ic s udies (e.g. E icson e al. 2006, B own e al. 2008, Hacke e al. 2008). Nex , he bes sco ing ee ob ained om his analysis was used as a s a ing ee in a Bayesian MCMC analysis (implemen ed in BEAST; D ummond & Rambau 2007) o simul aneously econs uc he phylo ge ny and di e gence imes o bi ds o p ey. The en gene segmen s we e pa i ioned sepa a ely and each gene segmen was as- signed i s own bes  i e olu iona y mo del, as e alua ed by Akaike In o ma ion C i- e ion (AIC) in he so wa e jModelTes 2 (Guindon & Gascuel 2003, Da iba e al. 2012) (Table 1). Molecula da ing was done using an un- co ela ed elaxed molecula clocks app oach, which akes in o accoun a ia ion in he a e o molecula e olu ion among line ages (D ummond e al. 2006). Th ee os- sil cons ain s we e used o da e he phy- logene ic ee ( ollowing do Ama al e al. 2009 and e e ences he ein): (1) he mini mum age o di e gence be ween Pandio- nidae and Accipi idae was se o 37 Mya, based on he oldes known ossil belon ging o Pandionidae (Ha ison & Walke 1976); (2) he maximum age o di e gence o Bu- eo galapagoensis was se o 4 Mya and (3) he maximum age o di e gence o B. soli a ius was se o 5.1 Mya. B. galapa go- ensis and B. soli a ius a e bo h island spe- cies ( es ic ed o he Galapagos Islands and Hawaii, espec i ely) and he la e wo age cons ain s a e based on he assump- ion ha hese species canno be olde hen he islands which hey inhabi (do Ama al e al. 2009). Two independen BEAST uns we e pe o med, each allowed o un o 50,000,000 gene a ions wi h a hinning in e al o 5,000 gene a ions. Con e gence was e alua ed by checking e ec i e sample size (ESS) o pa ame e s in T ace (Ram- bau & D ummond 2012). All pa ame e s had ESS alues >100 and mos we e >>200. The wo uns we e combined (a e emo ing 10% bu nin) and esampled a in- e als o 10,000 gene a ions o yield 9,000 ees ha ep esen s a sample o he pos- e io dis ibu ion o phylogene ic ees. A ORNIS HUNGARICA 2014. 22(1) 20 maximum clade c edibili y ee was gene a ed om his sample in T eeAnno a o (Rambau & D ummond 2012). To isualize di e si ica ion a e h ough ime, we c ea ed a lineages h ough ime plo o 100 ees selec ed andomly om he pos e io sample and he maximum clade c edibili y ee. Ances al a ea econs uc ion We collec ed b eeding season dis ibu ion da a on 180 species om Fe gusonLees and Ch is ie (2001). Species we e sco ed as p esen /absen in he ollowing biogeo- g aphical ealms: Nea c ic, Palea c ic, Neo opical, A o opical, Malagasy, Indoma- layan, Aus alasian. The delimi a ion o hese ealms is based on Fe gusonLees and Ch is ie (2001). We in e ed ances al anges based on his dis ibu ion da a by employing p o ba bilis ic his o ical biogeog aphy me hods using he BioGeoBEARS R package (Ma z- ke 2013). These me hods model geog aphic ange e olu ion by assuming di e en o ms o anagene ic and cladogene ic changes in geog aphic dis ibu ion du ing specia ion e en s: dispe sal, ex inc ion, ica iance, sympa ic specia ion and ounde e en spe- cia ion. We e alua ed which o he adi io nally used his o ical biogeog aphic models bes i s ange e olu ion in bi ds o p ey by calcula ing and compa ing six models using Akaike In o ma ion C i e ion. These mo dels a e he Dispe salVica iance (DIVA) model (Ronquis 1997), he Dispe salEx- inc ionCladogenesis (DEC) model (Ree e al. 2005, Ree & Smi h 2008) and he Bay A ea model (Landis e al. 2013), oge he wi h he combina ion o hese h ee wi h ounde e en specia ion. The h ee base- line models all assume dispe sal, ex inc ion, sympa ic specia ion and ica iance as pos- sible ange e olu ion mechanisms bu di e in he way hey ea sympa ic and ica ian specia ion e en s: he DIVA model allows na owscale sympa y bu bo h na ow and widesp ead ica iance. The DEC model as- sumes na owscale and subse sympa y, bu only na owscale ica iance whe eas BayA ea assumes na owscale and wide scale sympa y o occu (Ma zke 2013). We used he bes i o hese models o es i- ma e he mos likely ancien dis ibu ions a each node (ances o s a e) o he phylogeny. The me hod also gi es a ela i e p obabili- y, anging om 0 o 1, which gi es he p o babili y ha he node was in he gi en s a e. The highe his alue, he highe is ou con- idence in he ac ual econs uc ion is co - ec and unce aini y in he ances al ange econs uc ion is low. These analyzes we e done using he maximum clade c edibili y ee. Li e his o y da a All da a, wi h he excep ion o body mass in o ma ion, we e collec ed om Fe gu- sonLees and Ch is ie (2001), complemen ed om he Global Rap o In o ma ion Ne wo k (2013), i necessa y. Mig a o- y beha io was ca ego ized based on p e- ious phylogene ic s udies (e.g. Kondo & Omland 2007, do Ama al e al. 2009) as: (1) nonmig a o y (no seasonal mo emen s p e sen ), (2) pa ially mig a o y (pa o he popula ions, o pa o he indi iduals wi h- in he species pe o m egula seasonal mo emen s) and (3) comple ely mig a o y (all popula ions and indi iduals mig a o y). We used his a iable o in e a es o e o- lu ion o and om comple e mig a ion (see below). Howe e , he numbe o comple ely mig a o y species was ela i ely low in ou 21 J. Nagy & J. Tökölyi sample (N=13), he e o e, mig a ion was bi- na ized in all o he analyzes as ei he mig a- o y o nonmig a o y. In o ma ion on body mass was ob ained om Dunning (2008), and in a ew cases om Fe gusonLees and Ch is ie (2001). We used he a e age o male and emale body masses (log ans o med) when hey we e a ailable; howe e , in 8 cases da a on male o emale body mass was a ailable on- ly. Fo 29 species no eliable body mass da- a could be ound. Die (win e die ) was ca ego ized ollow- ing Roulin and Wink (2004). These au ho s assigned a ela i e impo ance alue an ging om 1 o 9 o each o nine ood ca e go ies (li e bi ds, mammals, ep iles, ish, amphibians, c us aceans, insec s, wo ms and ca ion) based on desc ip ions o indi i dual species’ die in Fe gusonLees and Ch is ie (2001). Food ypes ha do no ap- pea in he die o a species ecei ed a sco e o 9, whe eas he mos impo an ood ype ecei ed a sco e o 1. F om hese alues, we calcula ed eliance on wa mblooded p ey and ca ion as he minimum o he impo - ance sco es ecei ed o bi d o mammal p ey o ca ion. Die speci ici y was es i- ma ed by coun ing he numbe o ood ypes in he die o a gi en species ha ecei ed a sco e <9. Finally we classi ied habi a ype as open o closed based on desc ip ions in Fe gu- sonLees and Ch is ie (2001). Compa a i e analyzes We pe o med Bayesian ances al s a e e- cons uc ion in BEAST (D ummond & Rambau 2007) o in e he mig a o y be- ha io a ances al nodes in he phylogeny o bi ds o p ey. Mig a o y beha io was e- coded as a bina y a iable o his analysis (as ei he mig a o y o nonmig a o y, hus comple e and pa ial mig an s we e colla ed). Bayesian ances al s a e econs uc ion akes in o accoun phylogene ic unce ain- y and calcula es he p obabili y ha a gi en node was mig a o y o nonmig a o y, based on he ai alues o i s ances o and descendan s. By aking in o accoun unce - aini y in phylogene ic econs uc ion, his me hod is subs an ially be e han pa simo- nybased econs uc ions, whose ou come is condi ional on a single (possibly e one- ous) opology. In addi ion, Bayesian ances- al s a e econs uc ion also akes in o ac- coun di e ences in b anch leng hs, which makes hem mo e ealis ic han pa simo- nybased me hods. Nex , we in es iga ed how changes in mi- g a o y beha io occu ed on he phyloge- ny by es ima ing ansi ion a es be ween he h ee le els o mig a o y beha io (i.e. he a e o ansi ion om mig a o y o pa - ial o comple e mig an and ice e sa, and he a e o ansi ion om pa ial o comp le e mig an and ice e sa). This analysis was done using he Mul iS a e module o BayesT ai s 1.0 (Pagel e al. 2004). We de e mined whe he body size (log ans o med), habi a ype, die b ead h, eliance on wa mblooded p ey and geo- g aphical loca ion (Old s. New Wo ld) a - ec s mig a o y beha io by cons uc ing a mul i a ia e phylogene ic gene alized line- a mixed models as implemen ed in he MC- MCglmm package in R (Had ield & Naka gawa 2010), wi h hese ai s as dependen a iables. We also included he in e ac ion be ween habi a and geog aphical occu - ence o model he di e ences in habi a use among Old Wo ld and New Wo ld mig an s. Las ly, we es ed he co ela ed e olu- ion among mig a ion and explana o y a i ables ound o be signi ican in he mul i a ORNIS HUNGARICA 2014. 22(1) 22 ia e analyses using he Disc e e module o BayesT ai s 1.0 (Pagel & Meade 2006). This me hod e alua es ansi ion a es among pai s o bina y ai s on a phylogeny e ea ling de ails o co ela ed e olu ion among ai s. Fo example, when analy zing he co - ela ed e olu ion be ween mig a ion and di- e speci ici y one can ask whe he mig a ion is mo e likely e ol e in gene alis (o spe- cialis ) lineages o ice e sa: does a gene alis (o specialis ) die e ol e mo e likely in mig an s? Since his analysis can hand le only bina y ai s we dicho omized die speci ici y as specialis (<5 ood ypes con- sumed) o gene alis (a leas 5 ood ypes consumed). Resul s Phylogeny and di e si ica ion o Accipi i o mes Figu e 1 shows he ela ionship be ween majo lineages o ap o s. Ou analysis e- co e ed he ela ionships among majo line ages o bi ds o p ey desc ibed in p e i- ous molecula phylogene ic s udies (Wink & Saue Gü h 2004, Le ne & Mindell 2005, G i i hs e al. 2007). C own g oup Accipi i o mes (i.e. he spli be ween Sa- gi a ius and he es o he species) is in- e ed o ha e o igina ed ~44 million yea s ago (95% highes pos e io densi y in e al: 56.4 – 37.4), du ing he Eocene pe iod. The Figu e 1. Simpli ied phylogeny showing majo sublineages o Accipi i o mes 1. áb a Egysze űsí e ö zs a a ágómadá -alakúak őbb csopo jainak ilogene ikai iszonyai ól 23 J. Nagy & J. Tökölyi spli be ween Pandion and he es o Ac- cipi i o mes occu ed ~39.5 million yea s ago (95% highes pos e io densi y in e al: 49.5–37). The wo ea lies b anchings wi h- in Accipi idae esul ed in he appea ance o elanid ki es (Elaninae) ~34.7 million yea s ago (95% highes pos e io densi y in e al: 44.3–29.4) and he g oup con aining Gy- pae inae and Pe ninae (~27.7 million yea s ago; 95% highes pos e io densi y in e al: 35.3–23.2). Then, a ound he s a o he Miocene pe iod he di e si ica ion o ap- o s accele a ed and con inued a a high a e un il ecen imes (Figu e 2). Ances al a ea econs uc ion The bes model desc ibing ange e olu ion in bi ds o p ey was he DEC model con- aining ounde e en specia ion. Based on his model, a sou he n o igin was in e ed o all ap o sub amilies. These analyses sugges ha Accipi inae, Aegypiinae and Gypae inae ha e an A o opical o igin, whe eas Elaninae and Pe ninae de i e om he Neo opics. Bu eoninae and Ha piinae had a join A o opical/Neo opical dis- ibu ion acco ding o ou econs uc ion, whe eas Aquilinae ha e an A o opical/ Neo opical and Indomalayan o igin, al- hough we no e ha he accu acy o hese econs uc ions is qui e low (<0.3) (Table 2). Ci cea inae we e assigned an Indoma- layan o igin wi h ela i ely high p obabili y (0.87) (Table 2). Las ly, he mos likely ances al dis ibu ion o Haliaee inae was Aus alasia. Figu e 2. Lineages- h ough- ime plo showing he pa e n o di e si ica ion and accumula ion o ap- o species h ough ime as econs uc ed by ou mul i-gene elaxed molecula da ing ana- lysis. G ey lines show andom ees (N=100) om he pos e io sample o da ed ul ame ic ees ob ained om BEAST; he black line deno es he maximum clade c edibili y ee 2. áb a A ágómadá -alakúak di e zi ikációja a molekulá is da álás alapján. A szü ke onalak a ö zs a- ekons ukció so án lé ehozo posz e io min ából 100 éle lensze űen ki álasz- o ul ame ikus a, a eke e onal pedig az összegze min ából számol ul ame ikus a ágainak számá mu a ja az idő ügg ényében ORNIS HUNGARICA 2014. 22(1) 30 mig a ion occu ed wi h a highe a e in lin- eages wi h a opical dis ibu ion (9.08 s. 0.73 in lineages wi h a non opical dis i- bu ion) (Figu e 5A). Fu he mo e, ansi- ions o a non opical dis ibu ion a e much mo e likely in mig a o y (30.01) han in nonmig a o y lineages (0.15). The an- si ion a es also e ealed ha swi ches in he geog a phic dis ibu ion om a opical o a non opical dis ibu ion o ice e sa a e i ually lacking in nonmig a o y line- ages, whe eas hey occu a a ela i ely high a e in mig an s (Figu e 5A). In he second case, ansi ion a es indica e ha mig a ion Pa ame e es ima e lowe 95% con idence in e al uppe 95% con idence in e al P- alue Habi a (Fo es /Open) 1.30 0.13 2.27 0.02 Old/New Wo ld 0.78 -0.75 1.75 0.16 log(body mass) -0.06 -0.51 0.33 0.78 Die gene alism 0.44 0.21 0.75 <0.01 Reliance on wa m-blooded p ey o ca cass 0.00 -0.21 0.23 0.99 Habi a : Old/New Wo ld in e ac ion -1.39 -3.35 0.40 0.18 Table 4. Fac o s a ec ing he occu ence o mig a ion in diu nal bi ds o p ey (N=151 species); pa- ame e es ima es, hei 95% con idence in e als and P- alues om a mul i a ia e mixed e ec models con olling o phylogeny 4. ábláza A onulás elő o dulásá be olyásoló ényezők ágómadá -alakúaknál (N=151 aj); iloge- ne ikai iszonyok a kon ollál öbb ál ozós ke e lineá is modellből szá mazó becsül é ékek, azok 95%-os kon idenciain e alluma és a P-é ékek Figu e 5. Resul s om pai wise di ec ional es be ween mig a o y beha io (mig a o y o non-mi- g a o y) and (A) geog aphic dis ibu ion: opical (species no p esen in he Nea c ic o he Palea c ic) o non- opical (species p esen in he Nea c ic o he Palea c ic); (B) die speci- ici y: specialis (consumes <5 ypes o ood) o gene alis (a leas 5 ood ypes consumed). The g aph show ansi ion a es among pai s o ai s indica ing he a e wi h which hese e olu iona y changes a e in e ed o ha e occu ed on he phylogeny 5. áb a A onulási iselkedés és a öld ajzi el e jedés (A) ille e a áplálékspecializáció (B) közö i di- ekcionális esz ek e edményei. A onulási iselkedés kódolása: onuló ( észlegesen agy eljesen) agy nem onuló. A öld ajzi el e jedés kódolása: ópusi (nem o dul elő sem a Pa- lea k iszban sem a Nea k iszban) agy nem ópusi (az előző ellen e je). A áplálékspeciali- záció kódolása: specialis a (<5 áplálék ípus ogyasz ) agy gene alis a (az előző ellen e - je). Az áb a a különböző jellegpá ok közö i e olúciós anzíciós á áka mu a ja 31 J. Nagy & J. Tökölyi is mo e likely o a ise in specialis lineages and ha a gene alis die is mo e likely o e ol e in mig an s han in nonmig an s. Hence, i appea s ha die b ead h e ol es as in esponse o he selec i e en i onmen s imposed by mig a o iness, a he han p e- disposing species o mig a ion. Discussion Phylogeny o Accipi i o mes The phylogene ic ela ionships among and wi hin majo lineages o Accipi i o mes ha e been ex ensi ely s udied be o e (e.g. Wink & Saue Gü h 2004, Helbig e al. 2005, Le ne & Mindell 2005, G i hs e al. 2007, do Ama al e al. 2009). He e, we combined all a ailable gene ic in o ma ion o p oduce a mul igene phylogeny o Ac- cipi i o mes wi h a b oad axonomic sam- pling, including app oxima ely wo hi ds o ex an species o accipi id bi ds o p ey. The phylogene ic hypo hesis ob ained om his analysis is b oadly cong uen wi h p e ious epo s showing ha se e al adi ionally es- ablished clades a e in ac polyphyle ic o pa aphyle ic. Fo ins ance, Old Wo ld ul- u es o m a polyphyle ic clade comp ised o : (1) Gypae inae which is monophyle - ic wi h Pe ninae and includes he Bea ded Vul u e Gypae us ba ba us and he Egyp ian Vul u e Neoph on pe cnop e us and (2) Ae- gypiinae which con ains all emaining Old Wo ld ul u es and is he sis e clade o Ci - cae inae. Accipi e hawks a e pa aphyle ic and should include ha ie s (Ci cus spp.), which a e closely ela ed o he clade con- aining goshawks (see also B eman e al. 2013 o a mo e de ailed analysis). In addi- ion, we also obse ed widesp ead pa aphy- ly in aquiline eagles and bu eonine hawks, as epo ed p e iously (Helbig e al. 2005, do Ama al e al. 2009). On he o he hand, we also obse ed se e al disc epancies in he highe le el ela ion- ships o Accipi idae be ween ou econs uc- ions and hose o ob ained om p e ious s udies (e.g. Le ne & Mindell 2005, G i - i hs e al. 2007). Fo ins ance, we eco e ed Aqui linae and Ha piinae as sis e clades, al- bei wi h ela i ely low suppo (pos e io p obabili y: 0.45). Ha pagus ki es we e in- e ed as he sis e g oup o he clade con- aining Bu eoninae and Haliaee inae wi h ela i ely high suppo (pos e io p obabili y: 0.83). Las ly, he sis e ela ionship be ween he clade con aining Aquilinae and Ha pii- nae on one hand and Bu eoninae, Haliaee i- nae and Accipi inae on he o he was s ong- ly suppo ed (pos e io p obabili y: 1). His o ical biogeog aphy and e olu ion o mig a ion in Accipi i o mes Ou ances al s a e econs uc ion sugges s ha mig a o y beha io in bi ds o p ey e ol ed mul iple imes. Mos o hese ap- pea o be ela i ely ecen e en s (occu - ing du ing he Pliocene o Pleis ocene, i.e. <5 million yea s ago). In one case howe e , mig a o y beha io appea s o be much mo e ancien . In ue hawks (Accipi inae) mig a- ion appea s o ha e e ol ed app oxi ma ely 1412 million yea s ago, du ing he middle o he Miocene pe iod. By compa i son, do Ama al e al. (2009) econs uc ed he o i- gin o mig a ion in one Bu eo clade a app oxima ely 5 million yea s ago, a esul ha is suppo edby ou analyses (Figu e 3B). Mo e di ec es ima es based on he age dis- ibu ion o ossilized indi iduals (speci i- cally, he lack o ju eniles) ound a Oldu ai Go ge, in Tanzania, sugges ha his si e was a win e ing loca ion o sho ebi ds belon ORNIS HUNGARICA 2014. 22(1) 32 ging o Cha ad iidae 1.91.74 million yea s ago, implying ha mig a ion was p esen a his ime (Loucha 2008). Hence, ou es i- ma e o he o igin o mig a ion in accipi id hawks appea s o be one o he oldes da es published so a . Such es ima es a e impo - an (ye ema kably lacking) i we a e o un- de s and he e olu ion o mig a ion in a con- s an ly changing spa io empo al con ex a a global scale (Loucha 2008). The middle o he Miocene pe iod saw a se ies o global cooling e en s (Zachos e al. 2001), which esul ed in he expansion o g asslands and con ac ion o o es habi a s, possibly opening new niches o bi ds o p ey. Ou analyses sugges ha accipi id hawks appea ed sho ly be o e his pe iod, p obably in he A o opical ealm and colo- nized o he pa s o he wo ld sho ly he e- a e . Since mos ex an species belonging o his lineage a e a leas pa ly mig a o- y oday, i is likely ha hei ances o al- so pe o med seasonal mig a o y mo e- men s. Al e na i ely, mig a ion could ha e e ol ed sepa a ely in hese lineages due o simila selec i e en i onmen s (i.e. as a consequence o con e gen e olu ion a he han sha ed phylogene ic backg ound). Since mig a ion is a phylogene ically labile ai ha can e ol e e y quickly (see e.g. Zink 2011), independen e olu ion in mul- iple lineages expe iencing simila selec i e en i onmen s is a plausible scena io o he occu ence o mig a o y beha io in clus e s o closely ela ed species. Howe e , his ex- plana ion is clea ly less pa simonious in ex- plaining he e olu ion o mig a ion in ue hawks, since his g oup con ains bo h em- pe a e opical and in a opical mig an s on di e en con inen s, which would im- ply simul aneous, independen selec ion o mig a ion in a wide a ie y o di e en en- i onmen s on di e en pa s o he wo ld. Join econs uc ion o ances al dis ibu- ion and mig a o y beha io sugges s ha in ap o s, mig a ion appea ed mos ly in spe- cies wi h a sou he n o igin. This is u he s eng hened by ou di ec ional analyses, which sugges ha mig a ion is mo e like- ly o e ol e in opical species and ha mig a o y ap o s a e mo e likely o swi ch o a non opical b eeding ange, hence sugges ing ha mig a o y beha io and ange ex- pansions a e e ol ing in pa allel. In e es ingly, we also ound ha he ansi ion a e om a opical o non opical dis ibu ion (and ice e sa) is e y low in nonmig a- o y ap o s, bu no in mig an s, sugges ing ha mig a ion g ea ly enhanced ange ex- pansions in his g oup o bi ds. The sou he n o igin o mig a o y ap o s is in line wi h p e ious s udies ob aining simi- la esul s in a a ie y o axonomic g oups (Joseph e al. 1999, Ou law e al. 2003, Milá e al. 2006). We ha e o emphasize, howe e , ha his esul helps li le in unde s anding he e olu ion o mig a ion in bi ds o p ey. As we ha e shown, all majo lineages wi h- in Accipi idae ace back hei o igin o one o he sou he n biogeog aphic ealms. Hence, bo h mig a o y and nonmig a o y species cu en ly inhabi ing he empe a e zone des cend om he opics. Ou analyses do sug- ges , howe e , ha mig a ion is mo e likely o eme ge in he opics han in he empe a e zone, and ha mig a o y bi ds a e mo e likely o dispe se and swi ch om a opi- cal dis ibu ion o a non opical one. Hence, he ela ionship be ween coloniza ion o he empe a e zone and he e olu ion o mig a- ion could be he e e se o wha adi ional- ly is assumed, i.e. mig a o y bi ds (in a opi cal mig an s) mo e likely o colonize no el habi a s and expand o No h. This hypo he- sis could be es ed in he u u e by in es iga ing he ecological and beha io al ai s p o- 33 J. Nagy & J. Tökölyi mo ing he coloniza ion o empe a e habi a s in a b oade sample o bi ds. Since all bi ds o p ey a e o opical an- ces y, hei ances o s mus ha e unde gone ange expansions o he empe a e egion. Ye , no all o hese species became mig a- o y. We ound ha win e die specializa ion p edic s he occu ence o mig a ion in acci pi id bi ds o p ey, wi h mig a o y species elying on mo e a iable die s. The abili y o eed on a wide sou ce o ood ypes could g ea ly enhance he p obabili y ha a species su i es he win e in he empe a e zone whe e ood a ailabili y is much lowe du ing he win e (New on 2008). Hence, his could a leas pa ly explain in e speci ic di e en ces in mig a o iness. In e es ingly, we ound no associa ion be ween mig a ion and eli- ance on wa mblooded p ey o ca cass, sug- ges ing ha hese ood sou ces alone migh no be enough o sus ain mos species in he empe a e zone du ing win e . Fo ins ance, wo o he ou ul u e species ha occu in Eu ope ( he G i on Vul u e Gyps ul us, and he Egyp ian Vul u e) a e mig a o y, despi e he ac ha hei majo ood sou ce – ca cass – is mos likely a ailable yea  ound. How- e e , hese ood sou ces migh also show sea- sonal luc ua ions (e.g. Kendall e al. 2012). Al e na i ely, o he ac o s, such as selec ion o ea ly b eeding o ex ended b eeding sea- son could gene a e di e ences in mig a o- iness be ween popula ions o species (e.g. Tökölyi & Ba a 2011, Camacho 2013). Fu - he wo k is equi ed o cla i y ecological de- e minan s o mig a ion in bi ds o p ey. Acknowledgemen s We would like o exp ess ou g a i ude o he wo anonymous e e ees o hei use ul commen s on ou manusc ip . J. N. was sup- po ed by a schola ship om he Uni e si- y o Deb ecen’s Talen De elopmen P og am (DETEP). J. T. was suppo ed by he TÁMOP 4.2.2.C11/1/KONV20120010 p ojec ; he p ojec is co inanced by he Eu- opean Social Fund and he Eu opean Re- gional De elopmen Fund. Compu a ions we e pa ly un on Hunga ian Na ional In- as uc u e Ins i u e supe compu e s (h p:// nii .hu). We a e g a e ul o Józse Büki, head o And ás Ke e Lib a y o O ni hology and Na u e Conse a ion o his help in acces sing li e a u e in an ea ly pa o his s udy. Ale s am, T. 1993. Bi d mig a ion. – Camb idge Uni- e si y P ess do Ama al, F. R., Sheldon, F. H., Gamau , A., Ha - ing, E., Riesing, M., Sil ei a, L. F. & Wajn al, A. 2009. Pa e ns and p ocesses o di e si ica ion in a widesp ead and ecologically di e se a ian g oup, he bu eonine hawks (A es, Accipi idae). – Mo- lecula Phylogene ics and E olu ion 53: 703–715. DOI:10.1016/j.ympe .2009.07.020 Bell, C. P. 2000. P ocess in he e olu ion o bi d mig a ion and pa e n in a ian ecogeog aphy. – Jou - nal o A ian Biology 31: 258–265. DOI:10.1034/ j.1600048X.2000.310218.x Bell, C. P. 2005. The o igin and de elopmen o bi d mig a ion: commen s on Rappole and Jones, and an al- e na i e e olu iona y model. – A dea 93: 115–123. Bell, C. P. 2011. Resou ce bu e ing and he e olu ion o bi d mig a ion. – E olu iona y Ecology 25: 91– 106. DOI:10.1007/s1068201093834 Be hold, P. 2001. Bi d mig a ion: a gene al su ey. – Ox o d Uni e si y P ess Boyle, W. A. & Conway, C. J. 2007. Why mig a e? A es o he e olu iona y p ecu so hypo he- sis. – The Ame ican Na u alis 169: 344–359. DOI:10.1086/511335 Boyle, W. A., Conway, C. J. & B ons ein, J. L. 2011. Why do some, bu no all, opical bi ds mig a e? A compa a i e s udy o die b ead h and ui p e e ence. – E olu iona y Ecology 25: 219–236. DOI:10.1007/s1068201094034 B eman, F. C., Jo daens, K., Sone , G., Nagy, Z. T., Van Houd , J. & Loue e, M. 2013. DNA ba co Re e ences ORNIS HUNGARICA 2014. 22(1) 34 ding and e olu iona y ela ionships in Accipi e B isson, 1760 (A es, Falconi o mes: Accipi i- dae) wi h a ocus on A ican and Eu asian ep e- sen a i es. – Jou nal o O ni hology 154: 265–287. DOI:10.1007/s1033601208925 B own, J. W., Res , J. S., Ga cía–Mo eno, J., So en- son, M. D. & Mindell, D. P. 2008. S ong mi o- chond ial DNA suppo o a C e aceous o igin o mode n a ian lineages. – BMC Biology 6: 6. DOI:10.1186/1741700766 B ude e , B. & Salewski, V. 2008. E olu ion o bi d mig a ion in a biogeog aphical con ex . – Jou nal o Biogeog aphy 35: 1951–1959. DOI:10.1111/ j.13652699.2008.01992.x Camacho, C. 2013. T opical phenology in empe a e egions: ex ended b eeding season in a longdis- ance mig an . – The Condo 115: 830–837. DOI:10.1525/cond.2013.120192 Cas esana, J. 2000. Selec ion o conse ed blocks om mul iple alignmen s o hei use in phylogene ic analysis. – Molecula Biology and E olu ion 17: 540–552. Cha i , D. & Lob y, J. R. 2007. SeqinR 1.02: A con- ibu ed package o he R p ojec o s a is ical compu ing de o ed o biological sequences e  ie al and analysis. pp. 207–232. – In: Bas olla, U., Po o, M., Roman, H. E. & Vend uscolo, M. (eds.) S uc u al app oaches o sequence e olu ion. – Sp inge , Be lin Cox, G. W. 1985. The e olu ion o a ian mig a ion sys- ems be ween empe a e and opical egions o he New Wo ld. – The Ame ican Na u alis 126: 451– 474. Da iba, D., Taboada, D. L., Doallo, R. & Posada, D. 2012. jModelTes 2: mo e models, new heu is ics and pa allel compu ing. – Na u e Me hods 9: 772. DOI:10.1038/nme h.2109 D ummond, A. J., Ho, S. Y., Phillips, M. J. & Rambau , A. 2006. Relaxed phylogene ics and da ing wi h con idence. – PLoS Biology 4: e88. DOI:10.1371/ jou nal.pbio.0040088 D ummond, A. & Rambau , A. 2007. BEAST: bayesian e olu iona y analysis by sampling ees. – BMC E olu iona y Biology 7: 214. DOI:10.1186/1471 2148-7-214 Dunning, J. B. 2008. CRC handbook o a ian body masses. – CRC P ess, Boca Ra on E icson, P. G. P., Ande son, C. L., B i on, T., Elzanows- ki, A., Johansson, U. S., Källe sjö, M., Ohlson, J. I., Pa sons, T. J., Zuccon, D. & May . G. 2006. Di e - si ica ion o Neoa es: in eg a ion o molecula se- quence da a and ossils. – Biology Le e s 2: 543– 547. DOI:10.1098/ sbl.2006.0523 Fe gusonLees, J. & Ch is ie, D. A. 2001. Rap o s o he Wo ld. – Hough on Mi lin Company, Bos on Global Rap o In o ma ion Ne wo k. 2013. Species ac- coun s. – Downloaded om h p://www.global ap- o s.o g. G i i hs, C. S., Ba owclough, G. F., G o h, J. G. & Me z. L. A. 2007. Phylogeny, di e si y, and clas- si ica ion o he Accipi idae based on DNA se- quences o he RAG1 exon. – Jou nal o A ian Biology 38: 587–602. DOI:10.1111/j.2007.0908 8857.03971.x Guindon, S. & Gascuel, O. 2003. A simple, as , and accu a e algo i hm o es ima e la ge phylogenies by maximum likelihood. – Sys ema ic Biology 52: 696–704. DOI:10.1080/10635150390235520 Gwinne , E. 1990. Bi d mig a ion: physiology and eco- physiology. – Sp inge Ve lag, Be lin Hacke , S. J., Kimball, R. T., Reddy, S., Bowie, R. C. K., B aun, E. L., B aun, M. J., Chojnowski, J. L., Cox, W. A., Han, K., Ha shman, J., Huddles on, C. J., Ma ks, B. D., Miglia, K. J., Moo e, W. S., Shel- don, F. H., S eadman, D. W., Wi , C. C. & Yu i, T. 2008. A phylogenomic s udy o bi ds e eals hei e olu iona y his o y. – Science 320: 1763–1768. DOI:10.1126/science.1157704 Had ield, J. D. & Nakagawa, S. 2010. Gene al quan i a- i e gene ic me hods o compa a i e biology: phy- logenies, axonomies, and mul i ai models o con inuous and ca ego ical cha ac e s. – Jou nal o E olu iona y Biology 23: 494–508. DOI:10.1111/ j.14209101.2009.01915.x Ha ison, C. J. O. & Walke , C. A. 1976. Bi ds o he B i ish uppe Eocene. – Zoological Jou - nal o he Linnean Socie y 59: 323–351. DOI:10.1111/j.10963642.1976. b01017.x Helbig, A. J., Kocum, A., Seibold, I. & B aun, M. J. 2005. A mul igene phylogeny o aquiline eagles (A es: Accipi i o mes) e eals ex ensi e pa a- phyly a he genus le el. – Molecula Phylogene ics and E olu ion 35: 147–164. DOI:10.1016/j. ympe .2004.10.003 Je z, W., Thomas, G. H., Joy, J. B., Ha mann, K. & Mooe s, A. O. 2012. The global di e si y o bi ds in space and ime. – Na u e 491: 444–448. DOI:10.1038/na u e11631 Joseph, L., Lessa, E. P. & Ch is idis, L. 1999. Phyloge- ny and biogeog aphy in he e olu ion o mig a ion: sho ebi ds o he Cha ad ius complex. – Jou nal o Biogeog aphy 26: 329–342. DOI:10.1046/j.1365 2699.1999.00269.x Ka oh, K., Kuma, K., Toh, H. & Miya a, T. 2005. MAFFT Ve sion 5: imp o emen in accu acy o mul iple sequence alignmen . – Nucleic Acids Re- sea ch 33: 511–518. DOI:10.1093/na /gki198 Kendall, C., Vi ani, M. Z., Ki ui, P., Thomse , S. & Gi hi u, M. 2012. Mechanisms o coexis ence in ul u es: unde s anding he pa e ns o ul u e 35 J. Nagy & J. Tökölyi abundance a ca casses in Masai Ma a Na ional Re- se e, Kenya. – The Condo 114: 523–531. Kondo, B. & Omland, K. E. 2007. Ances al s a e e- cons uc ion o mig a ion: mul is a e analysis e- eals apid changes in New Wo ld o ioles (Ic e us spp.). – The Auk 124: 410–419. DOI:10.1642/0004 8038(2007)124[410:ASROMM]2.0.CO;2 Landis, M. J., Ma zke, N. J., Moo e, B. R. & Huelsen- beck, J. P. 2013. Bayesian analysis o biogeog a- phy when he numbe o a eas is la ge. – Sys ema ic Biology 62: 789–804. DOI:10.1093/sysbio/sy 040 Le ne , H. R.L. & Mindell, D. P. 2005. Phylogeny o eagles, old wo ld ul u es, and o he Accipi idae based on nuclea and mi ochond ial DNA. – Mo- lecula Phylogene ics and E olu ion 37: 327–346. DOI:10.1016/j.ympe .2005.04.010 Loucha , A. 2008. Eme gence o long dis ance bi d mig a ions: a new model in eg a ing global clima e changes. – Na u wissenscha en 95: 1109–1119. DOI:10.1007/s0011400804353 Ma zke, N. J. 2013. P obabilis ic his o ical biogeog a- phy: new models o ounde e en specia ion, im- pe ec de ec ion, and ossils allow imp o ed accu- acy and model es ing. – Ph.D. hesis, Depa men In eg a i e Biology and Designa ed Emphasis in Compu a ional and Genomic Biology, Uni e si- y o Cali o nia, Be keley. A ailable a h p://phy- lo.wikido .com/local iles/biogeobea s/Ma zke_ PhD_FINAL_ 5_w_ e s.pd . Milá, B., Smi h, T. B. & Wayne, R. K. 2006. Pos gla- cial popula ion expansion d i es he e olu ion o long–dis ance mig a ion in a songbi d. – E olu ion 60: 2403–2409. DOI:10.1111/j.00143820.2006. b01875.x New on, I. 2008. The mig a ion ecology o bi ds. – Aca demic P ess, London Ou law, D. C., Voelke , G., Milá, B. & Gi man, D. J. 2003. E olu ion o longdis ance mig a ion in and his o ical biogeog aphy o Ca ha us h ushes: a molecula phylogene ic app oache. – The Auk 120: 299–310. DOI:10.1642/00048038(2003)120[0299: EOLMIA]2.0.CO;2 Padian, K. & Chiappe, L. M. 1998. The o igin and ea ly e olu ion o bi ds. – Biological Re iews 73: 1–42. DOI:10.1111/j.1469185X.1997. b00024.x Pagel, M., Meade, A. & Ba ke , D. 2004. Bayesian es ima ion o ances al cha ac e s a es on phy- logenies. – Sys ema ic Biology 53: 673–684. DOI:10.1080/10635150490522232 Pagel, M. & Meade, A. 2006. Bayesian analysis o co - ela ed e olu ion o disc e e cha ac e s by e e sib lejump Ma ko Chain Mon e Ca lo. – The Ame i- can Na u alis 167: 808–825. DOI:10.1086/503444 Pa adis, E., Claude, J. & S imme , K. 2004. APE: ana lyses o phylogene ics and e olu ion in R language. – Bioin o ma ics 20: 289–290. DOI:10.1093/bioin- o ma ics/b g412 Rappole, J. H. & Jones, P. 2002. E olu ion o Old and New Wo ld mig a ion sys ems. – A dea 90: 525– 537. R De elopmen Co e Team. 2012. R: a language and en i onmen o s a is ical compu ing. – R Founda- ion o S a is ical Compu ing, Vienna. A ailable a h p://www. R–p ojec .o g. Ree, R. H., Moo e, B. R., Webb, C. O. & Donoghue, M. J. 2005. A likelihood amewo k o in e - ing he e olu ion o geog aphic ange on phy- logene ic ees. – E olu ion 59: 2299–2311. DOI:10.1111/j.00143820.2005. b00940.x Ree, R. H. & Smi h, S. A. 2008. Maximum likelihood in- e ence o geog aphic ange e olu ion by dispe sal, local ex inc ion, and cladogenesis. – Sys ema ic Bio logy 57: 4–14. DOI:10.1080/10635150701883881 Ronquis , F. 1997. Dispe sal ica iance analysis: a new app oach o he quan i ica ion o his o ical bio- geog aphy. – Sys ema ic Biology 46: 195–203. DOI:10.1093/sysbio/46.1.195 Roulin, A. & Wink, M. 2004. P eda o p ey ela ion- ships and he e olu ion o colou polymo phism: a compa a i e analysis in diu nal ap o s. – Biologi cal Jou nal o he Linnean Socie y 81: 565–578. DOI: 10.1111/j.10958312.2004.00308.x S ama akis, A. 2006. RAxMLVIHPC: maximum like- lihoodbased phylogene ic analyses wi h housands o axa and mixed models. – Bioin o ma ics 22: 2688–2690. DOI:10.1093/bioin o ma ics/b l446 S iles, F. G. 1980. E olu iona y implica ions o habi- a ela ions be ween pe manen and win e esiden landbi ds in Cos a Rica. pp. 421–435. – In: Keas , A. & Mo on, E. S. (eds.) Mig an Bi ds in he Neo- opics. – Smi hsonian Ins i u ion P ess, Washing- on D.C. Tökölyi, J. & Ba a, Z. 2011. B eeding phenology de- e mines e olu iona y ansi ions in mig a o y be- ha iou in inches and allies. – Oikos 120: 184– 193. DOI:10.1111/j.16000706.2010.18592.x Wink, M. & Saue –Gü h, H. 2004. Phylogene ic e- la ionships in diu nal ap o s based on nucleo ide sequences o mi ochond ial and nuclea ma ke genes. pp. 483–498. – In: Mo aJunio , J. C., Bue- no, A., Chancello , R. & Meybu g, B. U. (eds.) Rap o s wo ldwide. WWGBP, Be lin Zachos, J., Pagani, M., Sloan, L., Thomas, E. & Billups, K. 2001. T ends, hy hms, and abe a ions in global clima e 65 Ma o p esen . – Science 292: 686–693. DOI:10.1126/science.1059412 Zink, R. M. 2011. The e olu ion o a ian mig a ion. – Biological Jou nal o he Linnean Socie y 104: 237–250. DOI:10.1111/j.10958312.2011.01752.x