A Bird’s-Eye View on Evolution of Seasonal Migration

A Bi d’s-Eye View on E olu ion o Seasonal Mig a ion Disse a ion in ul illmen o he equi emen s o he deg ee Doc o e um na u alium o he Facul y o Ma hema ics and Na u al Sciences a he Ch is ian Alb ech s Uni e si y o Kiel Submi ed by Jun Ishigohoka Max Planck Ins i u e o E olu iona y Biology Pl¨on, Ge many May 2024 Fi s e e ee: P o D Mi iam Lied ogel Second e e ee: P o D Hin ich Schulenbu g Examine : P o D Julien Yann Du heil Chai pe son: P o D F ank Kempken Da e o o al examina ion: 2024-06-07 (F i) Summa y Wha makes animals special is ha hey can ac i ely mo e a a geog aphic scale, which allows hem o eloca e hemsel es om an ad e se en i onmen o o he places in mo e a ou able condi ions o su i e and ep oduce. Seasonal mig a ion is a ype o animal mo emen ypically be ween b eeding and win e ing si es wi h egula i y in iming and o ien a ion. Seasonal mig a ion is bo h ecologically and e olu iona ily impo an , because i is a beha iou al adap a ion o seasonal changes in en i onmen , and i de ines he dis ibu ion o animals when hey ma e and ep oduce. Since he dawn o e hology, mechanism, de elopmen , ecological unc ion and e olu ion o seasonal mig a ion ha e been hea ily s udied in bi ds. In his hesis, I add ess how seasonal mig a ion e ol es a a mic oe olu iona y ime scale. To his end, I s udy he Eu asian blackcap Syl ia a icapilla, o “blackcap”. Blackcaps a e a common songbi d species b eeding widely in Eu ope and no h A ica, anging om Ibe ia o Caucasus and Scandina ia, as well as on Maca onesian and Medi e anean islands. The blackcap is a pe ec species o s udy mic oe olu ion o mig a ion because hey ha e he i able a ia ion in popula ion- ypical mig a o y beha iou in p opensi y, dis ance, o ien a ion and iming. Using popula ion genomics and epigenomics, I in es iga e he e olu iona y his o y and molecula mechanisms o di e si ica ion o seasonal mig a ion. The aims o applica ion o popula ion genomics on he blackcap sys em a e o iden i y possible genomic a ge s o selec ion associa ed wi h di e gence o mig a o y beha iou and o unde s and he his o y o popula ions wi h di e en mig a o y pheno ypes. I s a wi h assessing he e ec s o ecombina ion a e, ins ead o selec ion and ue demog aphic e en s, on hese wo ypes o popula ion genomics analyses using simula ion and empi ical analysis on blackcap genomes. I ind ha genomic egions wi h low ecombina ion a es end o ha e dis inc pa e ns o gene ic a ia ion because he a iance in unde lying gene ic ances y educes, ins ead o by selec i e p ocesses. I also ind ha he p esence o high- ecombining genomic egions a ec s demog aphy in e ence using me hods based on he ances al ecombina ion g aph, a s uc u e ep esen ing gene ic ances ies along ecombining ch omosomes, because unde lying genealogies a e no ep esen ed by su icien ly many mu a ions in high- ecombining egions. Th ough empi ical alida ion o genome scans and demog aphy in e ence using blackcap genome da a, I ound one polymo phic in e sion wi h a g adien in equency among popula ions wi h di e en mig a o y pheno ypes and be ween con inen and islands, which is unlikely o happen neu ally unde he es ima ed demog aphy o blackcap popula ions. By compa ing popula ion gene ic simula ions unde he blackcap demog aphy and he obse ed in e sion equencies, I ind ha he in e sion is unde nega i e equency-dependen selec ion wi h di e en op imal in e sion equencies be ween con inen and island. I also de e mined how blackcap popula ions spli in his o y. I ind ha he blackcap popula ions spli in wo phases. An ances al mig an popula ion spli in o Caucasus (eas e n) and wes e n e ugia popula ions du ing one o he Pleis ocene glacial cycles. The popula ion expanded and spli a e he las glacial maximum, and he e we e a leas ou independen ansi ion e en s om mig an s o esiden s. Focusing on one pai o mig an and esiden popula ions, I in es iga e di e en ial 4 epigene ic egula ion be ween seasons and popula ions in he hypo halamus, a s ong candida e b ain egion unde lying he e olu iona y ansi ion. By combining a common ga den expe imen and quan i ica ion o single cell ch oma in accessibili y, I ind ha he amoun o seasonal change in ch oma in accessibili y is educed in some genomic egions in esiden s compa ed o mig an s. Finally, I syn hesise he popula ion genomic and single cell epigenomic insigh s in he ligh o phylogeog aphy and e olu ion o pheno ypic plas ici y. This hesis he e o e p o ides a s ep o wa d o he holis ic unde s anding o how animal beha iou may e ol e by his o ic changes in en i onmen ia modi ica ion o di e en ial epigene ic egula ion. 5 Zusammen assung Eine Besonde hei on Tie en is , dass sie sich ak i au eine g oß lächigen geog a ischen Skala bewegen können. Dies e möglich es ihnen, sich aus eine ungüns igen Umgebung an ande e O e mi güns ige en Bedingungen zu begeben, um zu übe leben und sich o zup lanzen. Dieses saisonale Wande - ode Zug e hal en (Mig a ion) is eine A de Tie bewegung, die ypische weise zwischen B u - und Übe win e ungsgebie en mi eine gewissen Regelmäßigkei in Bezug au Zei punk und Aus ich ung s a inde . Saisonale Wande ungen sind sowohl ökologisch als auch e olu ionä seh bedeu sam, da sie eine Ve hal ensanpassung an die jah eszei lichen Ve ände ungen de Umwel da s ellen und die Ve eilung de Tie e bei de Paa ung und Fo p lanzung bes immen. Sei den An ängen de Ve hal ens o schung wu den Mechanismus, En wicklung, ökologische Funk ion und E olu ion de saisonalen Mig a ion bei Vögeln eingehend un e such . In diese A bei be asse ich mich mi de F age, wie sich saisonales Zug e hal en au eine mik oe olu ionä en Zei skala en wickeln kann. Meine Un e suchungen okussie en sich hie bei au die Eu asische Mönchsg asmücke Syl ia a icapilla. Die Mönchsg asmücke is eine in Eu opa und No da ika wei e b ei e e Sing ogela , de en B u gebie sich on de Ibe ischen Halbinsel übe den Kaukasus und Skandina ien bis hin zu den maka onesischen und medi e anen Inseln e s eck . Die Mönchsg asmücke is eine ideale A , um die Mik oe olu ion des Zug e hal ens zu un e suchen, da sie en lang ih es Ve b ei ungsgebie es eine e e bba e Va ia ion im popula ions- ypischen Zug e hal ens in Bezug au Zugneigung, En e nung, O ien ie ung und Zei punk au weis . Mi hil e on Popula ionsgenomik und Epigenomik un e suche ich die E olu ionsgeschich e, sowie die molekula en Mechanismen de Di e si izie ung de saisonalen Zugmus e . Die Anwendung de Popula ionsgenomik au das Sys em de Mönchsg asmücke ziel da au ab, mögliche Ang i spunk e genomische Selek ion im Zusammenhang mi de Di e genz des Wande e hal ens zu iden i izie en und die Geschich e on Popula ionen mi un e schiedlichen Zugs a egien zu e s ehen. Ich beginne dami , die Auswi kungen de Rekombina ions a e ans elle on Selek ion und ech en demog a ischen E eignissen au diese un eschciedlichen Ansä ze popula ionsgenomische Analysen zu bewe en, indem ich sowohl Simula ionen, als auchempi ische Analysen on Mönchsg asmückengenomen e wende. Ich s elle es , dass genomische Regionen mi nied igen Rekombina ions a en dazu neigen, un e schiedliche Mus e de gene ischen Va ia ion au zuweisen, weil die Va ianz de zug undeliegenden gene ischen Abs ammung abnimm , und dies nich du ch selek i e P ozesse beding is . Ich s elle auch es , dass das Vo handensein on Genom egionen mi hohe Rekombina ios a e die Demog a ie- In e enz mi Me hoden beein luss , die au dem Abs ammungskombina ionsg aphen basie en, eine S uk u , die die gene ische Abs ammung en lang ekombinie ende Ch omosomen au zeig , weil die zug unde liegenden Genealogien nich du ch aus eichend iele Mu a ionen in s a k ekombinie enden Regionen ep äsen ie we den können. Du ch empi ische Validie ung on Genom-Scans und Demog a ie-In e enz un e Ve wendung on Mönchsg asmücken-Genomda en and ich eine polymo phe In e sion mi a iablem F equenzg adien en zwischen Popula ionen mi un e schiedlichen Zug e hal ens, sowiezwischen kon inen alen und Inseln Popula ionen, die bei de angenommenen E olu ionsgeschich e de Mönchsg asmücken-Popula ionen wah scheinlich nich neu al 6 is . Du ch den Ve gleich popula ionsgene ische Simula ionen mi de Demog a ie de Mönchsg asmücke und den beobach e en In e sionshäu igkei en s elle ich es , dass die In e sion eine nega i en equenzabhängigen Selek ion un e lieg , wobei sich die op imalen In e sionshäu igkei en zwischen Kon inen und Insel un e scheiden. Auße dem habe ich un e such , wie eschiedene Mönchsg asmückenpopula ionen e olu ionsgeschich lich en s anden sind. Ich s elle es , dass sich die E olu ionsgeschich e de Mönchsg asmückenpopula ionen in zwei Phasen au eilen. Eine anges amm e U sp ungspopula ion ziehende Mönchsg asnücken spal e e sich wäh end eines de pleis ozänen Gle sche zyklen in eine Kaukasus- (ös liche) und eine wes liche Re ugienpopula ion au . Die Popula ion expandie e und spal en sich nach dem le z en glazialen Maximum au . Wi beobach enmindes ens ie unabhängige Übe gangse eignisse on Zug- zu S and ogelpopula ionen. Ich konzen ie e mich au ein Paa on Zug- und S and ogelpopula ionen und un e suche die un e schiedliche epigene ische Regulie ung zwischen den Jah eszei en und Popula ionen im Hypo halamus, eine Hi n egion, die ü den im Fokus s ehenden e olu ionä en Übe gang om Zug zum S and ogel in F age komm . Du ch die Kombina ion eines ”Common Ga den” Expe imen es und de Quan i izie ung de Ch oma inzugänglichkei einzelne Zellen s elle ich es , dass das Ausmaß de saisonalen Ve ände ung de Ch oma inzugänglichkei in einigen genomischen Regionen bei S and ögeln im Ve gleich zu Zug ögelnge inge is . Abschließend asse ich die E kenn nisse aus de Popula ionsgenomik und de Einzelzell- Epigenomik im Hinblick au die Phylogeog aphie und die E olu ion de phäno ypischen Plas izi ä zusammen. Diese A bei s ell dahe einen Fo sch i im ganzhei lichen Ve s ändnis de F age da , wie sich das Ve hal en on Tie en du ch his o ische Ve ände ungen de Umwel übe die Modi ika ion de di e en iellen epigene ischen Regula ion en wickeln kann. 7 Acknowledgemen s Fi s and o emos , I would like o hank my supe iso , Mi iam Lied ogel, o he men o ing. I lea ned wha i means o be a good men o and a good scien is om you, wi h a lo o cha ac e is ics which I had no hough his impo an be o e. You will o e e be my e e ence poin and bible o how o communica e p o essionally wi hou losing pe sonali y. I could no hank you mo e o he us and eedom you ga e me and gene ous suppo on my scien i ic communica ion and ne wo king. I always el sa e when I had o disag ee wi h you, which assu ed my psychological secu i y a wo k, and you mo i a ed me o jump in o new hings. Con inuing my PhD was ne e a ques ion hanks o you. You helped me ealise ha I like doing science. I would also like o hank Julien Du heil o being in my hesis ad iso y and examina ion commi ees and o his scien i ic eedback on my p ojec . I decided no o mo e o Wilhelmsha en ollowing Mi iam la gely because o you p esence in Plön, and you p o ed my decision was igh . On op o ha , hank you o o e ing o sha e he 10x Ch omium machine, which was a big u ning poin o my epigenomics p ojec . I would like o hank Hin ich Schulenbu g o being in my hesis ad iso y and examina ion commi ees. I el honou ed when you o e ed o be in my PhD TAC a e my mas e ’s de ense. Also, I lea ned om you he impo ance o exp essing g a i ude o people o ganising e en s when you did i o me in he IMPRS e ea I co-o ganised du ing he in e -lockdown phase. I would like o hank Linda Oden hal-Hesse, Anja Guen he , Ma ké a Kaucká, And ea Pa icia Mu illo Rincon, Die ha d Tau z, Be nha d Haubold, Tobias Kaise , Nicole Thomsen and Ke s in Schä e o scien i ic and echnical discussion, and suppo on my expe imen s and ca ee . Linda, I eally liked ha you a e commi ed o be accessible o communica ion wi h s uden s om no only you own g oup bu also om o he g oups by pa icipa ing in mos o he IMPRS e ea s and Aqua i s. Anja, hank you o exposing me o mixed e ec s modelling in he ea ly phase o my PhD, which was used in di e en places o my p ojec s. And hank you bo h o in i ing me o he mouse Ch is mas pa y despi e my oo hless s udy sys em. Ma ké a and And ea, hank you o gene ous ma e ialis ic and echnical suppo on 10x scATAC, wi hou you help i would no ha e been possible. Die ha d, hank you o you gene ous suppo on consumables o my scATAC expe imen (which you may no be e en awa e o ) when he concep o budge disappea ed om my pe cep ion, and o cou se, o scien i ic and ca ee discussion and eedback. Be nha d, hank you o always being a ailable o my algo i hmic and bioin o ma ic ques ions, and making me an AWK lo e . Tobias, hank you o including me in he Clunio mee ing e en a e I became he las one om he Beha iou al Genomics G oup in Plön. Nicole and Ke s in, hank you o helping me na iga e in he lab. You we e always a ailable and help ul. I would lo e o wo k wi h echnicians in he u u e who a e as in o science as you a e. Many hanks o all cu en and pas membe s o Beha iou al Genomics G oup in Plön and Wilhelmsha en. Co inna, hanks a lo o le ing me be in ol ed in he obin ieldwo k. I admi e you ha d wo k and you a i ude owa ds science on op o being a bi d lo e . Geo g, I enjoy discussing wi h you, and wa ching you quickly lea n biology om a di e en backg ound was e y inspi ing. Ka en, hanks o le ing me be in ol ed in he ecombina ion p ojec and spending ime in he o ice o discussion. Juan and And ea, I app ecia e ha you we e app oachable o ques ions and discussion in he ea ly phase o my PhD. Ma hias, hanks o a lo o discussion despi e he dis ance, and making me eel we me many mo e imes han we ac ually did. Joe and Robe , I admi e you p oac i eness and you a e my models o good pos docs. 8 Many hanks o Ca olina o exci ing (and dep essing. . . hey can coexis ) discussion on in e sions. I admi e you ha d wo k and I eally liked d opping by a you o ice when you we e in o a andom cha , and I would lo e o be as app oachable and likeable as you a e. Jule, you b ough me en husiasm in science and helped me wi h my in e nal peace and he communi y eeling ou side science. I admi e you ha d wo k, ac i e commi men and empa hy. Alec and Deme is, hanks a lo o le ing me in e up you so equen ly o discussion, and in i ing me o ba becue, dinne and canoeing. You a e he mos impo an discussion pa ne s in my PhD and a majo d i e o me o go ou . Ch is in and Mayo, you made Plön my home. I am no he mos exp essi e pe son bu he e I sec e ly hank you wo o making my pe sonal li e he e colou ul, as e ul, game- ul and Joey- ul. Nikhil, hanks o a lo o discussion on concep s and phenomena o biology. I admi e you since e a i ude owa ds science wi h you open-mindedness and how you a e (o a leas you beha e) always so laid-back. Fe nanda, hank you o making ou o ice 164 (wi hou my name on he pla e un il he end) li ely by b inging plan s, ins alling he whi eboa d, and ini ia ing con e sa ions. Louk and S ella, hanks o he discussion on my scATAC analysis and making me sligh ly home-sick when each o you a elled in Japan. Thank you, Ian and Jinyang, o joining ou PopClub o discuss popula ion gene ics including he ARG. Special hanks o Angela Donne o helping me wi h non-scien i ic oubles, and he IT eam o making my analyses possible. I hank he IMPRS o E olu iona y Biology o gene ous unding. My scien i ic ac i i y was also suppo ed by scien is s ou side he ins i u e. I hank Ki a Delmo e o gene ously le ing me wo k on demog aphy in e ence as my i s e e p ojec o he big pape e en be o e mee ing in pe son. This p ojec opened he doo o popula ion gene ics including heo e ical pa s and e en ually des ined my whole p ojec . I hank Ja ie Pé ez-T is and Juan Ca los Ille a o p o iding me wi h eedback. A pape wi h many au ho s is no he easies ask, and I am so g a e ul ha hey always ead he manusc ip and ga e me eedback. I hank Sean S ankowski, Da ia Shipilina, Kon ad Lohse and Jochen Wol o being suppo i e and c i ical abou my heo y-o ien ed empi ical popula ion genomic p ojec s which we discussed in di e en places om an online con e ence o Lucca, London, (nea ) Vienna, Wilhelmsha en and Plön. I hank Hi ohisa Ebina o occasionally discussing di e se opics o science and ca ee . I hank Masakazu Hoshino o chee ing me up on my p ojec and sha ing his in e es ing da a. I hank Nao O a o sha ing he expe ience o lab isi s a Max Planck Ins i u e o O ni hology du ing my bachelo ’s. This was when I s a ed hinking o going ou o Japan o a pos -g ad s udy. I hank Kazuhi o Wada o p iming my jou ney on beha iou al e olu ion and showing me his philosophy o doing unique science. Finally I hank my amily, especially my pa en s, o no being an obs acle o my decision o ollow he non-canonical pa h o becoming a scien is . I migh sound awkwa d o hank hem o no doing some hing, bu I am awa e h ough anecdo es ha his ac ually could ha e been a p oblem, a guably o a di e en ex en be ween gende s. I am g a e ul ha hey ga e me he libe y o pu sue my in e es , and I belie e hey would ha e ea ed me he same way e en i I we e ano he gende . I hank my g andpa en s o always being chee ul and a i ma i e o my decision, mos ly ia Zoom, on which hey s uggle o se up he audio by o ge ing o unplug he headphones. 9 mig a ion, bu how he mig a o y jou ney is ca ied ou (e.g. ajec o y, ele a ion, s opo e s) canno be measu ed. These limi a ions a e deal wi h by newe sys ems desc ibed in he ollowing pa ag aphs. O e all, he e is a ade-o be ween esolu ion and scalabili y among hese me hods. A a local scale, adio eleme y can be u ilised o moni o p esence/absence o agged bi ds. This is pa icula ly use ul o pheno ype mig an s and esiden s in a pa ial mig an popula ion. Fo example, by pheno yping mig an and esiden indi iduals in a pa ial mig an popula ion o Eu opean blackbi ds Tu dus me ula, Zúñiga e al. (2017) ound ha mig an s had highe p obabili y o win e su i al han esiden s, which complemen s he esigh ing- based esul s by G is e al. (2017), collec i ely cap u ing cos s in ep oduc ion and bene i s in su i al o mig a ion. Ligh le el geoloca o s a e small elec onic a chi al acking de ices, which can be i o bi ds (Fig. 2) o eco d he ime and ligh le el (C oxall e al., 2005; Delmo e e al., 2020a; Van Do en e al., 2021). Based on e ie ed imes o sun ise and sunse , he la i ude and longi ude can be calcula ed. In addi ion o geoloca o s, logge s wi h mul iple senso s (a ligh senso (as in geoloca o s), accele ome e (beha iou al ac i i y) and empe a u e-compensa ed ba ome ic p essu e senso (al i ude)) can be used o eco d mig a o y and o he beha iou wi h local en i onmen al condi ions (Bäckman e al., 2017; Sjöbe g e al., 2018, 2021). Being ligh weigh ed, hese logge s allow measu emen o long-dis ance mig a ion beha iou o small bi ds including songbi ds. Ye , e ie al o eco ded da a om hese logge s equi es ecap u ing. Hence possible causes o lack o da a (e.g. comple e e u n mig a ion o di e en des ina ions, ailu e o mig a ion, and dea h un ela ed o mig a ion) canno be dis inguished. 16 Figu e 2: Ligh le el geoloca o s. A geoloca o is i ed o a Eu asian blackcap Syl ia a icapilla. Sa elli e eleme y, including global posi ioning sys em (GPS) and A gos ansmi e s, allows o ack mo emen o indi idual animals wi hou ecap u ing (Aikens e al., 2024; Kays e al., 2015). Cu en ly hese ansmi e s a e s ill oo hea y (e.g. an ICARUS ansmi e weighs 5 g) o many animals, including small songbi ds, and oo cos ly o acking a popula ion, species, and communi y le els. Finally, use o mili a y and wea he su eillance ada s (Hilge loh, 1989; Van Do en e al., 2015) allows measu emen o he ime, speed, and di ec ion o mig a ion o a la ge numbe o bi ds in na u e wi hou agging — hence wi hou cap u ing. Ye he esolu ion o bi d iden i ica ion is limi ed o g oups o bi ds wi h simila body size (e.g. passe ines and wade s), and obse a ions a di e en loca ions canno be di ec ly compa ed due o he lack o ags. Field expe imen s The expe imen al manipula ion and ield wo k was al eady combined in one o he i s expe imen al s udies by Rowan (1930). To es he e ec o pho ope iodism on o ien a ion o seasonal mig a ion, c ows we e ea ed wi h longe day leng hs and eleased in he ield. The o ien a ion o mig a ion was measu ed by ecap u ing: [T]he bi ds we e su ep i iously eleased on he mo ning o he 9 h [o No embe 1924] and he ac announced in he e ening newspape s and o e he adios, wi h an 17 appeal o anyone and e e yone o hun c ows on Thanksgi ing Day (and he ea e ) and o mail he bi ds in o he Depa men o Zoology a he Uni e si y. (Rowan, 1930). O he ypes o ield expe imen s include displacemen . Fo example, o in es iga e de elopmen o mig a o y o ien a ion, Pe deck (1958) caugh and inged mo e han 11,000 s a lings (S u nus ulga is) in he Ne he lands and eleased hem in Swi ze land. While adul s we e ecap u ed wi hin hei ypical win e ing ange in no hwes e n Eu ope, ju eniles on hei i s mig a ion we e ecap u ed in sou hwes e n Eu ope including sou he n F ance and Spain ollowing he di ec ion which would ha e led o no mal win e ing si es, demons a ing ha mig a o y o ien a ion in ol es bo h inna e ec o o ien a ion and lea ned goal o ien a- ion. Combining s a e-o - he-a acking and hese ield expe imen s is in o ma i e o he mechanisms and de elopmen o mig a ion in na u e, ye in e p e a ion is o en challenging as he en i onmen al ac o s a e no con olled. Expe imen s in cap i i y Since he p esumably i s obse a ion by Naumann in he 18 h cen u y, i has been ecognised ha cap i e bi ds exhibi beha iou al ac i i y du ing mig a o y season, cha ac e ised wi h hopping and lu e ing wings (Be hold, 1993). As a p oxy o mig a o y beha iou , his ele a ed ac i i y called Zugun uhe (mig a o y es lessness) is eco ded in measu emen cages equipped wi h mic oswi ches o in a ed mo ion senso s wi h an e en eco ding sys em (Ba ell & Gwinne , 2005; Be hold, 1993, 1996; Be hold e al., 2000; Wil schko, 1968). P oxy o o ien a ion o seasonal mig a ion can be measu ed in speci ic measu emen cages called Emlen unnels (Emlen & Emlen, 1966. Fig. 3) and ci cula cages (Wil schko, 1968). In he Emlen unnel se up, indi idual bi ds a e placed in a unnel lined wi h co ec ion pape , and he amoun o o ien ed hopping is eco ded as sc a ches on he pape (Be hold, 1993). In ci cula cages, beha iou al ac i i y is eco ded a each pe ch a anged in he cage (Be hold, 1993). These eco ding sys ems ha e been u ilised as s anda d me hods o quan i ying mig a- o y beha iou in con olled condi ions, which is equi ed o es po en ial en i onmen al 18 (e.g. e e ence cues o compass sys ems o o ien a ion, nu i ion and pho ope iodism o induc ion) and endogenous (e.g. neu al basis o compass, biological clock, neu oendoc ine basis o mig a ion onse and o se , and gene ic basis o mig a ion) ac o s egula ing seasonal mig a ion. Fo example, Emlen (1967) and Wil schko (1968) s udied espec i ely he s a and magne ic compasses as po en ial mechanisms o mig a o y o ien a ion by measu ing o ien a ion o Zugun uhe using he Emlen unnel in a plane a ium (Fig. 3C) and using he ci cula cage in magne ic coils (Fig. 3D). The neu al en i y o he magne ic compass unde lying noc u nal mig a o y o ien a ion was es ed by Zapka e al. (2009), whe e he amoun and o ien a ion o Zugun uhe we e measu ed using Eu opean obins E i hacus ubecula which ecei ed lesion a a candida e b ain egion, he clus e N. This e ealed ha he clus e N is essen ial o magne ic compass o ien a ion bu no o he sun compass o he u ge o mig a e. To unde s and how he iming o mig a ion is egula ed, Gwinne (1990) conduc ed expe imen s simila o he Rowan’s s udy on pho ope iodism (Rowan, 1925, 1930, 1932) bu o e much longe du a ions unde sys ema ically con olled ligh -da k cycles and wi h eco ding o Zugun uhe in cap i i y, which e ealed ha he iming o mig a ion is egula ed by an endogenous ci cannual clock en ained by exogenous pho ope iod. Finally, he gene ic basis o seasonal mig a ion was s udied by measu ing Zugun uhe o Eu asian blackcaps Syl ia a icapilla om a pa ial mig an popula ion (Pulido e al., 1996) o hyb ids be ween mig an and esiden blackcaps (Be hold, 1993, 1996; Be hold & Que ne , 1981), e ealing he polygenic basis o a ia ion in mig a o y beha iou by s anding gene ic a ia ion. 19 Figu e 3: O ien a ion cages.A. Emlen unnel (Emlen & Emlen, 1966). B. Ci cula cage (Wil schko, 1968). C. Emlen unnels wi h a plane a ium p ojec o . D. Ci cula cage in magne ic coils. A om Emlen & Emlen (1966); B and D om Wil schko (1968); C om Emlen (1967). Labo a o y me hods, genome sequencing, and mo e In he ield si es and in cap i i y, biological samples such as blood, gonads and b ains, can be collec ed o physiological (Wing ield e al., 1990), his ological and molecula (Ras ogi e al., 2013) analyses. On op o hese, a new se o me hods ha e been applied o answe mechanisms and e olu ion o mig a ion since he de elopmen o high- h oughpu genome sequencing. Fi s , genome sequencing o mul iple indi iduals allows popula ion genomic analyses o in es iga e he s uc u e and his o y o popula ions, po en ial a ge s o selec ion, and associa ion be ween geno ype and pheno ype and/o en i onmen (Du heil, 2010). These me hods enable us o add ess he e olu iona y ques ions (gene ic a chi ec u e and epea abili y). Second, ansc ip omics and epigene ics a e used o quan i y ansc ip ion ac i i y and epigene ic 20 s a es (DNA me hyla ion, his one ail modi ica ions, and ch oma in emodelling) as a measu e o gene exp ession and egula ion, o en compa ing be ween issues, ages, seasons, condi ions, popula ions, o species (Bakken e al., 2021; Bendesky e al., 2017; Colqui e al., 2021; Hu e al., 2022; Jhanwa e al., 2021; Me i e al., 2020). These app oaches aim o in eg a e ou unde s anding o how molecula mechanisms in he cell a e egula ed di e en ly by endogenous and exogenous ac o s by seasons, condi ions, o e olu iona y ime. The e a e es ablished and s a e-o - he-a me hods which ha e no been ully applied in bi d mig a ion s udies. Fi s , compa ed o o he beha iou al and physiological adap a ions o seasons, such as hibe na ion (D ew e al., 2007; Junkins e al., 2022), he e a e limi ed s udies on he neu al egula ion con olling seasonal mig a ion. Neu al co ela es o mig a ion has been in es iga ed h ough his ological assay o immedia e ea ly genes (IEGs) using in si u hyb idisa ion (Mou i sen e al., 2005; Ras ogi e al., 2013; Zapka e al., 2010), ye ecoding o neu al ac i i y wi h elec ophysiology (Takahashi e al., 2022) is limi ed and eco ding in beha ing animals (Ho mann e al., 2019) and manipula ion o local neu al ac i i y wi h in usion (Benicho & Vallen in, 2020; Wa en e al., 2011), cell ype-speci ic molecula gene ic manipula ion (Sánchez-Valpues a e al., 2019), o op ogene ics (Singh Al a ado e al., 2021; Tanaka e al., 2018) ha e no been applied o a ian mig a ion. Second, cu en de elopmen o single cell ansc ip omics and epigenomics (Bakken e al., 2021; Mickelsen e al., 2019; Mo i e al., 2018; Sha e e al., 2022) ha e no been applied. These app oaches a e pa icula ly impo an o unde s and he mechanism o mig a ion, because di e en cells likely con ibu e di e en ly o seasonal mig a ion (Ras ogi e al., 2013). Finally, app oaches o e olu iona y de elopmen al biology (e o-de o) (Ca oll, 2005; Gilbe , 2015) ha e been limi ed in s udies o bi d mig a ion. In e o-de o, he co e ques ion is he molecula mechanisms o de elopmen unde lying di e en pheno ypes exp essed be ween species using a common gene ic oolki . The sca ece applica ion o e o-de o in bi d mig a ion has o do wi h he di icul y in molecula gene ic manipula ion in bi ds, and he likely polygenic na u e o he con ol o seasonal mig a ion (Pulido e al., 1996; Pulido, 2007). 21 Summa y Mechanisms, de elopmen , unc ion, and e olu ion o mig a ion a e s udied by combining inging, acking, expe imen al manipula ion o endogenous and exogenous ac o s, s anda dised eco ding o beha iou in cap i i y, labo a o y, and compu a ional me hods. Same me hods can be used o answe di e en ypes o ques ions o seasonal mig a ion. Con e sely, he same ques ion o mig a ion can be app oached using di e en ypes o me hods. The s udy sys em: Eu asian blackcaps In my hesis, I add ess wo o he ou undamen al ques ions: e olu ion and mechanisms. Speci ically, I aim o unde s and he e olu iona y ansi ion o mig a o y pheno ypes and i s molecula mechanisms a a mic o-e olu iona y ime scale. The Eu asian blackcap Syl ia a icapilla is an ideal sys em o add ess hese ques ions. The blackcaps is a songbi d species commonly b eeding h oughou Eu ope, no he n A ica and Maca onesian and Medi e anean islands (Shi ihai e al., 2010). Al hough he blackcap was one o many species shown o be mig an s in he i s decades o inging (Fig. 1), his species is iconic in s udies on di e en ial con ol o mig a o y o ien a ion and p opensi y be ween popula ions, because hey exhibi la ge a ia ion in seasonal mig a ion among popula ions. Mos o he con inen al popula ions a e soli a y noc u nal mig an s (Shi ihai e al., 2010). The dis ance o mig a ion a ies la i udinally, long o sho dis ance mig an s om no he n o sou he n b eeding popula ions (Be hold, 1988; Shi ihai e al., 2010). The o ien a ion o au umn mig a ion ( om he b eeding g ound o win e ing a ea) a e spli longi udinally a a mig a o y di ide in cen al Eu ope: wes e n b eeding popula ions mig a e sou hwes , whe eas eas e n popula ions mig a e sou heas (Helbig, 1991a, 1991b, 1996). The e a e mig an s wi h a no el mig a o y o ien a ion o no hwes win e ing in he B i ish Isles (Be hold e al., 1992; Helbig, 1991b; Langslow, 1979), which o igina e om a wide ange o he con inen al Eu ope (Delmo e e al., 2020a; Langslow, 1979). In addi ion o his la ge a ia ion in mig a o y pheno ypes among mig an popula ions, he e a e mul iple esiden popula ions in sou he n Ibe ia and no he n A ica and on Maca onesian and Medi e anean islands. Pheno ypic a ia ion in mig a o y o ien a ion, dis ance, iming, and p opensi y a e he i able wi h polygenic 22 basis (Pulido e al., 1996). P oblems add essed in his hesis To unde s and e olu iona y ansi ion o popula ion- ypical mig a o y pheno ypes and i s molecula mechanisms in blackcaps, I use popula ion genomics and b ain epigene ics. Chap e s 2, 3: How does ecombina ion a e a ec popula ion genomics me hods? The gene al aims o popula ion genomics a e o unde s and wha oles neu al and selec i e p ocesses play in shaping pa e ns o gene ic a ia ion in a popula ion o genomes, and o cha ac e ise how hese p ocesses ope a e in na u al popula ions based on obse ed pa e ns o genomic a ia ion (Du heil, 2010; Hahn, 2018; Wakeley, 2008). Genome scans o gene ic a ia ion and demog aphy in e ence a e wo o common popula ion genomic me hods o iden i y a ge s o selec i e p ocesses in he genome and o econs uc popula ion his o y (Du heil, 2010). These me hods ha e been widely applied in a ious s udy sys ems including bi ds o answe e olu iona y ques ions (Bu i e al., 2015; Delmo e e al., 2020b; Nadachowska-B zyska e al., 2016; Vijay e al., 2017). Howe e , h ough my empi ical analyses in blackcaps, i became e iden ha hese wo app oaches a e bo h a ec ed by a common ac o : ecombina ion a e. Recombina ion a e depic s he expec ed numbe o c osso e e en s occu ing in a gi en segmen o he genome pe meiosis (i.e. pe gene a ion) (Hudson & Kaplan, 1985). Two mu a ions on a ch omosome wi h no ecombina ion would be inhe i ed oge he : hus he pa e n o hei inhe i ance in a popula ion would be pe ec ly co ela ed. Inhe i ance o wo mu a ions on a ecombining ch omosome a e no as co ela ed, because ecombina ion can happen be ween he wo loci, and i i does, he wo mu a ions a e no on he same ch omosome in he o sp ing ha inhe i he ecombined ch omosome. In a popula ion o a ixed size, he geno ypes a he wo loci become less co ela ed as he expec ed numbe o ecombina ions be ween he loci pe gene a ion inc eases. This is he case as he physical dis ance be ween he loci in base pai s ( o be conside ed) become longe . In o he wo ds, geno ypes o loci in a popula ion a e co ela ed i hey a e close o each o he , and less so as hey a e u he 23 apa . Some popula ion genomics me hods make use o his co ela ion o geno ypes be ween loci along a ecombining ch omosome o model he ances al p ocess along he ch omosome, whe eas (in e p e a ion o ) o he me hods assume ha he uni o he genome scan is la ge enough o assume geno ypes be ween mos combina ions o loci a e independen . Impo an ly, how apidly his decay occu s along he ch omosome (i.e. he local ecombina ion a e) a ies along ch omosomes and among o ganisms. In chap e 2, I add ess how genomic egions wi h educed ecombina ion a e can cause dis inc pa e ns o gene ic a ia ion, which is o en in e p e ed as a signa u e o selec i e p ocesses, using popula ion gene ic simula ion and empi ical popula ion genomic analysis in blackcaps. In chap e 3, I add ess how he p esence o high- ecombining genomic egions a ec s demog aphy in e ence me hods which use locally co ela ed geno ypes along he ch omosome o model he ances al p ocess, wi h popula ion genomic simula ion and empi ical analysis in blackcaps. Chap e 4: How did selec ion on a balanced in e sion shi upon popula ion spli ? Wi hin chap e 2, I iden i y a ch omosomal in e sion in blackcaps wi h a g adien in equency ac oss popula ions (cline). In e sion clines a e commonly in e p e ed as he e ec o locally a ying selec ion and balancing selec ion. Howe e , he mode o selec ion and pa ame e alues o he ype o selec ion ope a ing ha e been a ely measu ed. In chap e 4, I ex end he empi ical popula ion genomic analyses in chap e s 2 and 3 o e alua e and measu e he selec i e p ocess ac ing on a polymo phic in e sion, using a amewo k o es ima ing e olu iona y models and pa ame e alues based on ex ensi e simula ion. Chap e 5: Wha is he epigene ic basis unde lying e olu iona y ansi ion om mig an o esiden ? Th ough demog aphy in e ence conduc ed in chap e s 3 and 4, I ound ha he e we e mul iple independen and simul aneous ansi ions om mig an s o esiden s in blackcaps. To unde s and he cellula and molecula mechanisms o di e en ial egula ion unde lying his 24 ansi ion, I e alua e single cell ch oma in accessibili y o he hypo halamus o mig an and esiden blackcaps in mig a o y and non-mig a o y seasons. Chap e 6: How does seasonal mig a ion e ol e? Finally, I syn hesise he p eceding chap e s. I discuss he pa e n and cause o blackcap popula ion his o y, and how mig a o y beha iou e ol es. 25 2 Dis inc Pa e ns o Gene ic Va ia ion a Low-Recombining Genomic Regions Rep esen Haplo ype S uc u e “God g an me he se eni y o accep he hings I canno change, cou age o change he hings I can, and wisdom always o ell he di e ence.” Among he hings Billy Pilg im could no change we e he pas , he p esen , and he u u e. – Ku Vonnegu , Slaugh e house-Fi e (1969) 32 Dis inc pa e ns o gene ic a ia ion a low- ecombining genomic egions ep esen haplo ype s uc u e Jun Ishigohoka1,* Ka en Bascón-Ca dozo1And ea Bou s1Janina Fuß2 A ang Rhie3Jacquelyn Moun cas le4Be ina Haase4William Chow5 Joanna Collins5Ke s in Howe5Ma cela Uliano-Sil a5Oli ie Fed igo4 E ich D. Ja is4,6,7 Ja ie Pé ez-T is8Juan Ca los Ille a9 Mi iam Lied ogel1,10,* 1Max Planck Ins i u e o E olu iona y Biology, Plön, Ge many 2Ins i u e o Clinical Molecula Biology (IKMB), Kiel Uni e si y, Kiel, Ge many 3 Genome In o ma ics Sec ion, Compu a ional and S a is ical Genomics B anch, Na ional Human Genome Resea ch Ins i u e, Na ional Ins i u es o Heal h, Be hesda, MD, USA 4The Ve eb a e Genome Lab, Rocke elle Uni e si y, New Yo k, NY, USA 5Wellcome Sange Ins i u e, Camb idge, UK 6Labo a o y o Neu ogene ics o Language, Rocke elle Uni e si y, New Yo k, NY, USA 7The Howa ds Hughes Medical Ins i u e, Che y Chase, MD, USA 8 Depa men o Biodi e si y, Ecology and E olu ion, Complu ense Uni e si y o Mad id, Mad id, Spain 9 Biodi e si y Resea ch Ins i u e (CSIC-O iedo Uni e si y-P incipali y o As u ias), O iedo Uni e si y, Mie es, Spain 10Ins i u e o A ian Resea ch, Wilhelmsha en, Ge many * Co espondence: Jun Ishigohoka <ishigohoka@e olbio.mpg.de>,Mi iam Lied ogel <lied o- gel@e olbio.mpg.de> 33 Abs ac Gene ic a ia ion o he en i e genome ep esen s popula ion s uc u e, ye indi idual loci can show dis inc pa e ns. Such de ia ions iden i ied h ough genome scans ha e o en been a ibu ed o e ec s o selec ion ins ead o andomness. This in e p e a ion assumes ha long enough genomic in e als a e age ou andomness in unde lying genealogies, which ep esen local gene ic ances ies. Howe e , an al e na i e explana ion o dis inc pa e ns has no been ully add essed: oo ew genealogies o a e age ou he e ec o andomness. Speci ically, dis inc pa e ns o gene ic a ia ion may be due o educed local ecombina ion a e, which educes he numbe o genealogies in a genomic window. He e, we associa e dis inc pa e ns o local gene ic a ia ion wi h educed ecombina ion a es in a songbi d, he Eu asian blackcap (Syl ia a icapilla), using genome sequences and ecombina ion maps. We ind ha dis inc pa e ns o local gene ic a ia ion e lec haplo ype s uc u e a low- ecombining egions ei he sha ed in mos popula ions o ound only in a ew popula ions. A he o me species- wide low- ecombining egions, gene ic a ia ion depic s conspicuous haplo ypes seg ega ing in mul iple popula ions. A he la e popula ion-speci ic low- ecombining egions, gene ic a ia ion ep esen s a iance among c yp ic haplo ypes wi hin he low- ecombining popula ions. Wi h simula ions, we con i m ha hese dis inc pa e ns o haplo ype s uc u e e ol e due o educed ecombina ion a e, on which he e ec s o selec ion can be o e laid. Ou esul s highligh ha dis inc pa e ns o gene ic a ia ion can eme ge h ough e olu ion o educed local ecombina ion a e. Recombina ion landscape as an e ol able ai he e o e plays an impo an ole de e mining he he e ogeneous dis ibu ion o gene ic a ia ion along he genome. 34 In oduc ion Pa e ns o gene ic a ia ion in he genome ep esen ances ies o sequences and a e in luenced by popula ion his o y. While genome-wide gene ic a ia ion ep esen s popula ion s uc u e (McVean, 2009; Pa e son e al., 2006), andomness in genealogies also con ibu es o luc ua ion o local gene ic a ia ion along ecombining ch omosomes. Speci ically, genealogies can di e be ween loci e en unde he same popula ion his o y (Du heil e al., 2009; Ma in & Van Belleghem, 2017; McVean & Ca din, 2005; Pamilo & Nei, 1988; Wakeley, 2008, 2020; Wiu & Hein, 1999). This is because ealisa ion o a genealogy unde a gi en popula ion his o y is a p obabilis ic p ocess: an ances al haplo ype o a se o indi iduals a one locus is no necessa ily a common ances o o he same se o indi iduals a ano he locus (Shipilina e al., 2023). Pa e ns o local gene ic a ia ion along he genome end o con o m wi h he popula ion s uc u e wi h andom luc ua ion (Fig. 1). 35 ...... ... ... Time pop1 pop2 B C F eely ecombining egion Low- ecombining egion Ano he low- ecombining egion Selec ion agains gene low DA Possible genealogies Unde lying local genealogies Realised gene ic a ia ion Recombina ion ① ② ③ ④ ① ② ③ ④ Figu e 1: Dis inc pa e ns o gene ic a ia ion can be due o educed ecombina ion a e. Popula ion his o y (A) a ec s he dis ibu ion o possible genealogies (B) om which local genealogies a e d awn (C). The numbe o genealogies in a genomic in e al wi h a ixed physical leng h depends on he local ecombina ion a e (C). Mu a ions occu ing on he genealogies (no shown) de e mine he pa e ns o ealised gene ic a ia ion. The ealised gene ic a ia ion can be summa ised and isualised wi h a ious me hods such as PCA (D). (1) In eely ecombining neu al egions, mu a ions ep esen many genealogies and hence he pa e n o gene ic a ia ion con e ges o he popula ion s uc u e. (2, 3) In low- ecombining neu al egions, mu a ions ep esen ew genealogies co e ing he egion leading o pa e ns o gene ic a ia ion dis inc om he popula ion s uc u e. (3) Due o andomness in sampling o genealogies, some o such dis inc pa e ns can be simila o pa e ns expec ed a a ge s o selec i e ac o s (c. . 4). (4) A a ge s o selec ion, dis ibu ion o possible genealogies is di e en om ha a neu al egions, which is depic ed as a di e en se o possible genealogies in Band he do ed a ow. In e ence o popula ion s uc u e as well as o he genome-wide analyses based on gene ic a ia ion ake ad an age o a su icien numbe o unlinked a iable si es (e.g. single nucleo ide polymo phisms (SNPs)) o elimina e he e ec o andomness. One o he mos common me hods o summa ise popula ion s uc u e based on his app oach is p incipal componen analysis (PCA) applied on a whole-genome geno ype able (McVean, 2009; P ice e al., 2006). In a whole-genome PCA, a ia ion among indi iduals based on a iable si es o he en i e genome a e usually p ojec ed on o a ew majo axes (some analyses use many mo e axes), and 36 he dis ances among indi iduals on hese educed dimensions ep esen gene ic di e ences. Summa ising popula ion s uc u e and o he ela ed measu es using he en i e genome has been p o en o be an e ec i e app oach o elimina e andom luc ua ion o genealogies along he genome (Bha ia e al., 2013; Cao e al., 2020; Fedo o a e al., 2013; Pe e , 2022; Shao e al., 2023). Howe e , some undamen al biological ques ions conce n selec i e ac o s ha sys ema - ically bias he shape o genealogies a a genomic local scale, shi ing he expec ed pa e ns o gene ic a ia ion om he popula ion s uc u e. Fo example, pa e ns o local gene ic a ia ion a e dis inc unde selec ion agains gene low (Fig. 1C4), posi i e selec ion and adap i e in og ession because hey a ec coalescence a e, opology, and b anch leng hs o he unde lying genealogies (Hejase e al., 2020; Ma in e al., 2015; Se e e al., 2020; Speidel e al., 2019; Wol & Elleg en, 2017). Empi ically, genome scans o popula ion gene ic summa y s a is ics ha e been commonly used o iden i y egions wi h dis inc pa e ns o gene ic a ia ion (Delmo e e al., 2018; I win e al., 2018; Kawakami e al., 2017; Roes i e al., 2013; Rougemon e al., 2021). Many o hese ha e iden i ied egions wi h dis inc pa e ns, such as ele a ed di e en ia ion and educed di e si y, wi hin low- ecombining genomic egions (Ge aldes e al., 2011; Kawakami e al., 2017; Renau e al., 2013; Roes i e al., 2013, 2013; Rougemon e al., 2021). Dis inc pa e ns a low- ecombining egions can in luence he ch omosome-wide (Knie e al., 2016; Nea sey e al., 2010) and e en genome-wide popula ion s uc u e (Mé o e al., 2021). These associa ions be ween dis inc pa e ns o gene ic a ia ion a “ou lie egions” o “genomic islands” and educed ecombina ion a e is o en in e p e ed as linked selec ion (Bu i e al., 2015; Bu i, 2017; Delmo e e al., 2015, 2018; I win e al., 2018; Kawakami e al., 2017; Roes i e al., 2013; Rougemon e al., 2021; Van Do en e al., 2017). Howe e , a non-selec i e explana ion is equally concei able and ye o en o e looked: he ocal genomic egion may con ain oo ew unde lying genealogies o a genome scan o elimina e he e ec o andom luc ua ion simply due o low ecombina ion a e, which is ep esen ed as he dis inc pa e ns o gene ic a ia ion (Booke e al., 2020; Lo e hos, 2019). Speci ically, i has no been well s udied wha aspec s o dis inc pa e ns o gene ic a ia ion can be explained by educed ecombina ion a e, and wha o he aspec s e lec he e ec o selec ion. 37 We add ess he e ec o educed ecombina ion a e on local gene ic a ia ion using a songbi d species, Eu asian blackcap (Syl ia a icapilla, he ea e “blackcap”), which is cha ac e ised by a iabili y in seasonal mig a ion ac oss i s dis ibu ion ange (Be hold, 1988, 1991; Delmo e e al., 2020a; Helbig, 1991). Popula ions wi h di e ged mig a o y pheno ypes spli as ecen ly as ~30,000 yea s ago, likely co esponding o he las glacial pe iod and now exhibi popula ion s uc u e (Fig. 2A-C, Sup. Fig. 1) (Delmo e e al., 2020b). Due o hei ecen spli and ela i ely la ge e ec i e popula ion size, gene ic di e en ia ion is e y low among blackcap popula ions (Delmo e e al., 2020b). The p esence o popula ion s uc u e albei wi h he low le els o di e en ia ion makes he blackcap a pe ec sys em o in es iga e local de ia ions o gene ic a ia ion: e en he sligh es e ec s o ac o s ha change local gene ic a ia ion a e likely de ec able because such e ec s a e no obscu ed by popula ion s uc u e. In addi ion, ine-scale ecombina ion maps o mul iple popula ions a e a ailable o his species (Bascón-Ca dozo e al., 2022a), acili a ing in es iga ion o he ela ionship be ween changes in he ecombina ion landscape and locally dis inc pa e ns o gene ic a ia ion. By le e aging a la ge-scale genomic e-sequencing da ase , we i s sys ema ically explo e dis inc pa e ns o local gene ic a ia ion along he blackcap genome, and compa e hese wi h genomic egions exhibi ing educed ecombina ion a e. We u he in es iga e he pa e ns o gene ic a ia ion in ou lie egions and associa e hem wi h he p e alence o ecombina ion supp ession ac oss popula ions. We also conduc simula ions o analyse how educed local ecombina ion a e in he en i e species and in a subpopula ion wi h and wi hou selec ion a ec s pa e ns o gene ic a ia ion h ough ime. Finally, we p opose a model o local gene ic a ia ion ep esen ing haplo ype s uc u e co esponding o e olu iona y changes in local ecombina ion a e. 38 Resul s Ch omosome-le el e e ence assembly To allow popula ion genomic analyses in he blackcap sys em, we gene a ed a ch omosome-le el e e ence genome using he Ve eb a e Genomes P ojec pipeline 1.5 (Rhie e al., 2021). We collec ed blood o a emale blackcap om Ta i a, Spain popula ion. We gene a ed con igs om Pacbio long eads, so ed haplo ypes, and sca olded hem wi h 10X Genomics linked eads, Bionano Genomics op ical mapping, and A ima Genomics Hi-C linked eads. Base call e o s we e polished wi h bo h PacBio long eads and A ow sho eads o achie e abo e Q40 accu acy (no mo e han 1 e o e e y 10,000 bp). Manual cu a ion iden i ied 33 au osomes and Z and W ch omosomes (plus 1 unlocalised W). Au osomes we e named in dec easing o de o size, and all had coun e pa s in he commonly used VGP e e ence zeb a inch assembly (Sup. Table 2). The inal 1.1 Gb assembly had 99.14% assigned o ch omosomes, wi h a con ig N50 o 7.4 Mb, and sca old N50 o 73 Mb, indica ing a high-quali y assembly ha ul ills he VGP s anda d me ics. The p ima y and al e na e haplo ype assemblies a e p o ided unde NCBI BioP ojec PRJNA558064, accession numbe s GCA_009819655.1 and GCA_009819715.1. De ia ion o gene ic a ia ion coincides wi h low- ecombining egions To in es iga e he genome-wide dis ibu ion o gene ic a ia ion, we mapped sho eads o he whole-genomes o 179 blackcaps including 69 newly sequenced indi iduals (Sup. Table 1) on a de no o-assembled e e ence genome gene a ed h ough he Ve eb a e Genomes P ojec (VGP, Rhie e al., 2021), and called SNPs (Ma e ials and Me hods). To cha ac e ise genome-wide gene ic a ia ion, we pe o med PCA using SNPs in all au osomes, e ealing popula ion s uc u e. While PC1 and PC2 ep esen ed di e en ia ion o island popula ions (Fig. 2B), PC3 ep esen ed s uc u e wi hin con inen al popula ions wi h di e en mig a o y pheno ypes (Fig. 2C). To iden i y genomic egions wi h pa e ns o gene ic a ia ion dis inc om he popula ion s uc u e, we pe o med local PCA using los uc (Li & Ralph, 2019). B ie ly, los uc pe o ms PCA in sliding genomic windows and dissimila i y o PCA among windows a e summa ised wi h mul idimensionali y scaling (MDS). Dis inc pa e ns o gene ic a ia ion o windows ela i e o he backg ound a e ep esen ed by ex eme alues along 39 he MDS axes. Mul iple windows wi h co ela ed pa e ns o gene ic a ia ion dis inc om he popula ion s uc u e a e ep esen ed by ex eme alues along he same MDS axis. This app oach allowed sys ema ic and unbiased explo a ion una ec ed by ou de ini ion o popula ions o he blackcaps. We pe o med los uc on bo h geno ype and phased haplo ype da a wi h window size o 1,000 SNPs. We iden i ied ou lie windows by applying h eshold MDS alues ( he mode o he dis ibu ion ± 0.3). We u he iden i ied genomic egions wi h dis inc pa e ns o gene ic a ia ion by inding genomic in e als longe han 100 kb wi h a leas i e ou lie windows based on he same MDS axis and me ging he in e als based on he geno ype- and phased haplo ype-based app oaches. This yielded 32 genomic egions wi h dis inc pa e ns o a ia ion (he ea e “ou lie egions”, Fig. 2D, Sup. Table 3, Sup. Fig. 3). Thei size anged om 0.12 o 8.11 Mb (mean and median o 0.71 and 0.29 Mb), and each egion con ained 5,000 o 356,000 SNPs. Compa ing he genomic dis ibu ion o hese ou lie egions o popula ion-le el ecombina ion maps, we ound ha low- ecombining egions (nominally ecombina ion a e lowe han he 20 pe cen ile o each ch omosome) we e signi ican ly en iched in he ou lie egions (pe mu a ion es s wi h n = 1,000, p- alue = 0.000 (Sup. Fig. 10)). Among hese 32 ou lie egions, 19 coincided wi h egions in which ecombina ion a e was educed in mos es ed popula ions (“species-wide” low- ecombining egions), 11 coincided wi h egions in which ecombina ion a e was educed in one o wo popula ions (“popula ion-speci ic” low- ecombining egions), and wo did no coincide wi h low- ecombining egions in any popula ion (Fig. 2E, F, Sup. Fig. 9). 40 Ou lie s o e lapping species−wide low− ec. (19) Ou lie s o e lapping only popula ion−speci ic low− ec. (11) Ou lie s wi hou low− ec. (2) Ch omosomal backg ound con _medlong con _sho con _ esiden Cana y Madei a Azo es CapeVe de Mallo ca C e e n = 1 n = 5 n = 10 A −0.05 0.05 0.15 −0.30 −0.20 −0.10 0.00 PC1 (3.3%) PC2 (2.3%) B −0.05 0.05 0.15 −0.2 −0.1 0.0 0.1 PC1 (3.3%) PC3 (1.8%) C 140 120 100 80 60 40 20 0 1 2 3 Z 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 Ch omosome Posi ion [Mb] D 0 50 100 150 0 10 20 30 40 50 Posi ion [Mb] Recombina ion a e [cM/Mb] Ch omosome 1 con _medlong Azo es Cape Ve de Ou lie s (spp.−wide) Ou lie s (pop.−spec.) E 0 5 10 15 0 10 20 30 40 50 Posi ion [Mb] Recombina ion a e [cM/Mb] Ch omosome 14 F Figu e 2: Local PCA ou lie s coincide wi h species-wide and popula ion-speci ic low- ecombining egions A. Geog aphic loca ion o blackcap popula ions included in his s udy. Each poin on he map ep esen s a sampling loca ion whe e mul iple indi iduals we e sampled. Popula ions we e de ined based on he geog aphic loca ion, mig a o y pheno ype, and genomic-wide popula ion s uc u e. B, C. Genome-wide PCA illus a ing popula ion s uc u e. D. Dis ibu ion o ou lie egions based on local PCA using los uc .E, F In e ed ecombina ion a es along wo exempli ied ch omosomes (ch omosomes 1 and 14) in h ee blackcap popula ions (con _medlong, Azo es, and Cape Ve de). In D-F, pu ple and g een shades espec i ely indica e posi ions o ou lie s ha coincide wi h species-wide and popula ion-speci ic low- ecombining egions. The wo g een shades in Fbo h o e ap wi h Azo es and Cape Ve de-speci ic low- ecombining egions. con _medlong: medium and long dis ance mig an popula ion b eeding on he con inen ; con _sho : sho dis ance mig an popula ion b eeding on he con inen ; con _ es: esiden (non-mig an ) popula ion b eeding on he con inen . All island popula ions (Cana y, Madei a, Azo es, Cape Ve de, Mallo ca and C e e) a e esiden . 41 A −0.15 0.00 0.10 −0.15 0.00 0.15 PC1 (3.2%) PC2 (2.6%) = 0.00 B −0.15 −0.05 0.05 0.15 −0.20 −0.05 0.10 PC1 (2.8%) PC2 (2.5%) = 0.00 C −0.2 0.0 0.2 −0.1 0.1 0.3 PC1 (3.7%) PC2 (3.5%) = 0.61 −0.10 0.00 0.10 0.20 −0.15 0.00 0.10 PC1 (3%) PC2 (2.7%) = 0.61 −0.1 0.1 0.2 −0.1 0.1 0.3 PC1 (3%) PC2 (2.9%) = 0.72 −0.15 0.00 0.10 −0.15 0.00 0.10 PC1 (2.7%) PC2 (2.6%) = 0.72 pop1 pop2 pop3 Time [N gene a ions] Rec. supp ession N NN N = 1,000 Low- ec.No mal ec. Figu e 5: Simula ion o a popula ion-speci ic low- ecombining egion. A. Simula ed scena io. Simula ed genome con ained wo ch omosomes, one wi h a popula ion-speci ic low- ecombining egion and he o he wi hou . B, C. PCA showing pa e ns o gene ic a ia ion a he popula ion-speci ic low- ecombining egion (B) and he no mally ecombining ch omosome (C) a h ee ime poin s in one exempli ied simula ion eplica e. E ec o selec ion on pa e ns o gene ic a ia ion Selec ion is known o cause dis inc pa e ns o gene ic a ia ion (Nielsen, 2005). To es whe he he ou lie egions based on los uc iden i ied in he blackcap genome a e also a ge s o selec ion, we measu ed nucleo ide di e si y ( π ) and Tajima’s D in each popula ion, as well as a io be ween non-synonymous and synonymous subs i u ions ( dN/dS ) o anno a ed genes. Many species-wide low- ecombining egions showed educed nucleo ide di e si y (Sup. 48 Fig. 28; Sup. Table 9) and Tajima’s D (Sup. Fig. 29; Sup. Table 10), sugges ing ha hey a e unde ei he posi i e o pu i ying selec ion. Mos genes wi hin ou lie egions had dN/dS below 0 (Sup. Fig. 30) wi h a ew genes wi h posi i e dN/dS, indica ing ha mos genes a e unde pu i ying selec ion and a ew o he s a e unde posi i e selec ion. Fu he mo e, sequence analysis indica ed ha some bu no all species-wide low- ecombining ou lie egions coincide wi h pu a i e pe icen ome ic egions wi h en ichmen o long andem epea s (Sup. Figs. 33, 34). These esul s indica e ha he ou lie egions may expe ience e ec s o selec ion in addi ion o educed ecombina ion a es. We asked whe he he dis inc pa e ns o local gene ic a ia ion a he ou lie egions obse ed in blackcaps ep esen he e ec o selec ion ins ead o educed ecombina ion a es. Speci ically, we add essed whea he he dis inc pa e ns o gene ic a ia ion ep esen ing haplo ype s uc u e could be caused by (i) pu i ying o (ii) posi i e selec ion alone o i hey p ima ily ep esen he e ec o educed ecombina ion a e. To his end, we used SLiM o simula e pu i ying and posi i e selec ion wi h and wi hou educ ion in ecombina ion a e, and in es iga ed local gene ic a ia ion o e ime by PCA. Fi s , o in es iga e he e ec o pu i ying selec ion, we simula ed wo ch omosomes wi h and wi hou a species-wide low- ecombining egion unde he same demog aphic his o y as he neu al scena io (Fig. 4A) bu wi h di e en s eng h o pu i ying selec ion by in oducing mu a ions wi h di e en a ios be ween he a es o neu al and dele e ious mu a ions (Ma e ials and Me hods). Dis inc pa e ns o gene ic a ia ion ep esen ing haplo ype s uc u e e ol ed only in scena ios whe e ecombina ion a e was educed i espec i e o he dis ibu ion o i ness e ec s (DFE) (Sup. Fig. 31). S onge pu i ying selec ion (DFE wi h mo e equen dele e ious mu a ions in ou simula ion) dec eased he ime o dis inc pa e ns o gene ic a ia ion a low- ecombining egions o be o e aken by popula ion s uc u e (Sup. Fig. 31A, C). Second, o in es iga e he e ec o posi i e selec ion, we simula ed a ch omosome wi h o wi hou a species-wide low- ecombining egion unde he same demog aphic his o y, and in oduced a bene icial mu a ion 100 gene a ions a e he popula ion spli in one popula ion (Sup. Fig. 32A) o 100 gene a ions be o e he spli in he ances al popula ion (Sup. Fig. 32D). Fo simula ions in which he bene icial mu a ion pe sis ed, we eco ded he pa e ns o local gene ic a ia ion by 49 PCA o e ime. Al hough posi i e selec ion a ec ed pa e ns o gene ic a ia ion compa ed o he neu al scena io, dis inc pa e ns o gene ic a ia ion ep esen ing disc e e haplo ypes we e unique o scena ios wi h educed ecombina ion a e in bo h cases (Sup. Fig. 32B-E). These esul s indica e ha dis inc pa e ns o gene ic a ia ion ep esen ed in local PCA, as in he blackcap ou lie egions, p ima ily e lec haplo ype s uc u e due o educed ecombina ion a e, on which he e ec o selec ion can be o e laid. Discussion Dis inc pa e ns o gene ic a ia ion a low- ecombining egions: Genealog- ical in e p e a ions Genealogical noise, genealogical bias, and mu a ional noise A numbe o empi ical popula ion genomics s udies ha e iden i ied ecologically and e olu ion- a ily impo an genomic egions by loca ing ou lie egions wi h dis inc pa e ns o gene ic a ia ion (Jones e al., 2012; Lamichhaney e al., 2016; Lawniczak e al., 2010; Lundbe g e al., 2021; Malinsky e al., 2015). Genomic windows in such s udies a e assumed o be bo h la ge enough o elimina e he e ec o andom luc ua ion in local gene ic a ia ion and small enough o cap u e he localised signa u es o selec ion. We showed empi ically ha genomic egions wi h dis inc pa e ns o gene ic a ia ion iden i ied by a popula ion genomic scan based on p incipal componen analysis (PCA) highly o e lap wi h low- ecombining genomic egions (Fig. 2). Wi h simula ions, we showed ha al hough selec ion may a ec he amoun and pa e n o local gene ic a ia ion a ound he a ge locus, he dis inc pa e ns o gene ic a ia ion ep esen ed by PCA a low- ecombining egions can be p ima ily explained by haplo ype s uc u e due o educed ecombina ion a e (Figs. 4,5). We discuss ou indings om he pe spec i e o unde lying genealogies. We i s de ine h ee e ms: (1) genealogical noise, (2) genealogical bias, and (3) mu a ional noise. (1) By “genealogical noise” we e e o he ac ha gene genealogies a y along he genome ollowing a null dis ibu ion gi en a popula ion his o y (Du heil e al., 2009; Ma in & Van Belleghem, 2017; McVean & Ca din, 2005; Wakeley, 2008, 2020; Wiu & Hein, 1999). 50 (2) By “genealogical bias” we e e o he ac ha selec i e p ocesses can sys ema ically shi he dis ibu ion o local genealogies away om he null dis ibu ion. Fo example, genealogies unde posi i e selec ion, selec ion agains gene low, adap i e in og ession, and balancing selec ion a e biased due o bu s s o coalescence, as e lineage so ing, and in oduc ion and main enance o long b anches (Ba on & E he idge, 2004; Gue e o e al., 2012; Hejase e al., 2020; Ma in e al., 2019; Se e e al., 2020; Speidel e al., 2019; Taylo , 2013). On op o hese, (3) andomness in he p ocess o mu a ion causes addi ional noise in ealised gene ic a ia ion (Ralph e al., 2020), which we call “mu a ional noise”. Fo example, he i s and he second hal es o a ch omosomal in e al wi h a single genealogy can s ill ha e sligh ly di e en pa e ns o gene ic a ia ion because hey ep esen some ini e numbe s o di e en mu a ions. Species-wide low- ecombining egions We showed in blackcaps ha some dis inc pa e ns o gene ic a ia ion a e associa ed wi h species-wide low- ecombining egions (Fig. 2). This is in line wi h p e ious s udies epo ing nega i e co ela ion be ween ecombina ion a e and gene ic di e en ia ion (Bu i e al., 2015; Bu i, 2017; Delmo e e al., 2015, 2018; I win e al., 2018; Kawakami e al., 2017; Roes i e al., 2013; Rougemon e al., 2021; Van Do en e al., 2017). To in es iga e wha ac o s a ec dis inc pa e ns o gen ic a ia ion a low- ecombining egions (Fig. 3) in mo e de ail, we pe o med simula ions o low- ecombining egions wi h and wi hou selec ion, and demons a ed ha haplo ype s uc u e unde lies he dis inc pa e ns which pe sis s only ansien ly un il he e ec o he popula ion s uc u e eme ges (Figs. 4,5). This ansiency e lec s a shi om local gene ic a ia ion p ima ily ep esen ing haplo ype s uc u e (Lo e hos, 2019; Ma & Amos, 2012) o ha ep esen ing popula ion s uc u e, which can be in e p e ed based on he unde lying genealogies. Low- ecombining egions ha e ew unde lying genealogies pe in e al o a ixed physical leng h and haplo ype s uc u e a such egions ends o e lec hei basal b anches because basal b anches end o be longe han pe iphe al b anches (Wakeley, 2008). A a ime poin soon a e a popula ion spli e en , pe iphe al b anches co e ing mo e ecen imes han he popula ion spli ha bou ewe mu a ions han basal b anches. The e o e, he ealised pa e n o gene ic a ia ion a his s age has he g ea es con ibu ions by mu a ions 51 on he long basal b anches undi e en ia ed among popula ions (i.e. consis ing s anding gene ic a ia ion), ep esen ing a ew ances al haplo ypes ha descend he cu en sample. As ime passes a e he popula ion spli , he p opo ion o mu a ions ha ha e occu ed a e he popula ion spli inc eases while some ances al haplo ypes can be los by chance (i.e. d i ), inc easing he con ibu ion o popula ion s uc u e on gene ic a ia ion. This ype o dis inc pa e ns o gene ic a ia ion a ises p edominan ly in low- ecombining egions bu less so in no mally ecombining egions. This is because haplo ype s uc u e ep esen ing a ew ances al lineages would become less p ominen wi h ecombina ion as di e en segmen s o a cu en haplo ype can ollow dis inc ances ies and hus he genealogical noise is e ec i ely a e aged ou . Some low- ecombining egions may ha e genealogies wi h much sho e basal b anches han o he low- ecombining egions because he a iance in he basal b anch leng h is g ea e han pe iphe al b anches (Wakeley, 2008). The o e - ep esen a ion o a ew ances al haplo ypes in gene ic a ia ion equi es long basal b anches in he unde lying genealogies, and hus low- ecombining egions wi h ela i ely sho basal b anches canno accommoda e su icien mu a ions o ep esen dis inc ances al haplo ypes. This dec eases he ela i e con ibu ion o genealogical noise compa ed o mu a ional noise (Supplemen a y No es 1.1). Dis inc pa e ns o gene ic a ia ion wi h a ying le els o clus e ing o indi iduals in PCA in ou empi ical esul s (Sup. Fig. 6) may co espond o di e en a ios be ween genealogical and mu a ional noise due o la ge a iance in he basal b anch leng hs o unde lying genealogies. Speci ically, some ou lie egions wi h mix u e o indi iduals om mul iple popula ions wi hou dis inc clus e s and popula ion subdi ision in PCA may ha e unde lying genealogies wi h sho basal b anches leading o g ea e con ibu ions o mu a ional noise on he ealised gene ic a ia ion. Popula ion-speci ic low- ecombining egions We bo h empi ically and wi h simula ions showed ha popula ion-speci ic low- ecombining egions exhibi dis inc pa e ns o gene ic a ia ion in which indi iduals o low- ecombining and no mally ecombining popula ions ha e di e en a iance in gene ic dis ances (Fig. 3C, Fig. 5). This unequal a iance in low- ecombining and no mally ecombining popula ions can 52 be in e p e ed based on he unde lying genealogies (Sup. Fig. 35). We conside he ances y o cu en samples o low- ecombining and no mally ecombining popula ions and spli he ances y a he ime T when he popula ion-speci ic ecombina ion supp ession ini ia ed (Sup. Fig. 35A). A ime T , he e we e n1 and n2 ances al haplo ypes ha descend all cu en samples in low- ecombining and no mally ecombining popula ions. A imes olde han T , he ances o s o he n1 and n2 haplo ypes may eely ecombine wi hin each se , making he gene ic dis ances among ances al haplo ypes wi hin each popula ion close o equidis an (Sup. Fig. 35B). A e he ini ia ion o he popula ion-speci ic educ ion in ecombina ion a e, he ances y o one cu en sequence o he low- ecombining popula ion can be aced back o ei he one o he n1 ances al haplo ypes p esen a he ime T (Sup. Fig. 35A). On he con a y, he ances y o one cu en sequence o he no mally ecombining popula ion can be aced back o mul iple ances al haplo ypes o he n2 sequences because o he p esence o ecombina ion (Sup. Fig. 35A). F om he pe spec i e o mu a ions, in he low- ecombining popula ion, mu a ions ha a ose on he same haplo ype end o be linked un il he p esen ime because o he supp essed ecombina ion. On he o he hand, in he no mally ecombining popula ion, mu a ions ha a ose on he same ances al haplo ype less likely s ay linked un il he p esen ime because ecombina ion can dissocia e hem. Because shu ling o haplo ypes educes he a iance o gene ic dis ances among sequences, popula ion-speci ic educ ion in ecombina ion a es leads o g ea e a iance in low- ecombining popula ion han in no mally ecombining popula ion as obse ed in ou empi ical esul s and simula ions. In sho , because o he di e en ecombina ion a es be ween he popula ions, genealogical noise is mo e e icien ly elimina ed in he no mally ecombining popula ion han in he low- ecombining popula ion. The haplo ype s uc u e a popula ion-speci ic low- ecombining egion is only c yp ic and less appa en han in species-wide low- ecombining egions because o he s anding mu a ions coexis on he same haplo ype, which a e olde han he ini ia ion o he popula ion-speci ic ecombina ion supp ession (Sup. Fig. 27). The ele a ed PC loadings a linked mu a ions o igina ing in he low- ecombining popula ion could be in o ma i e o s udy e olu iona y change in local ecombina ion a e: he ages o such mu a ions mapped on in e ed genealogies migh be use ul o es ima e he iming a which he popula ion-speci ic ecombina ion supp ession 53 ini ia ed. In ou empi ical analyses in blackcaps, we de ec ed he e ec o popula ion-speci ic educ ion o ecombina ion a e in Azo es and Cape Ve de island popula ions (Fig. 3C, Sup. Fig. 7). I emains unclea why educed ecombina ion a e in ce ain popula ions bu no o he s is e lec ed as dis inc pa e ns o gene ic a ia ion by los uc . The ecen spli o Azo es and Cape Ve de popula ions om o he popula ions, accompanied by educ ion in popula ion size and he le el o isola ion (Delmo e e al., 2020b) may ha e con ibu ed o mo e e icien sp ead o educed ecombina ion a e. Recombina ion landscape as a d i e o e olu ion o local gene ic a ia ion Species-wide and popula ion-speci ic ecombina ion supp ession unde lying dis inc pa e ns o local gene ic a ia ion a e p obably no independen : educ ion in ecombina ion a es ha ini ia es o ma ion o haplo ype blocks likely o igina es om one popula ion and may sp ead o mul iple popula ions. Fo example, local ecombina ion a e may be ini ially educed in one popula ion in which a seg ega ing in e sion o igina es be o e i may sp ead in mul iple popula ions by gene low (Fa ia e al., 2019). In line wi h his iew o ecombina ion map as an e ol able ai di e ging ac oss popula ions acco ding o subdi ision, ecen s udies ind ha di e gence in local ecombina ion a e among popula ions is co ela ed wi h gene ic di e gence (Bascón-Ca dozo e al., 2022a; Roes i e al., 2013; Spence & Song, 2019). Fu u e wo k on he e ec s o ansi ion om popula ion-speci ic o species-wide supp ession o ecombina ion will ill he gap be ween he wo s a es. Besides sp ead o ecombina ion supp ession ac oss popula ions, he e a e o he pa hs along which pa e ns o local gene ic a ia ion may change o e ime. Fi s , change in equency o one haplo ypic a ian by d i o gene low and selec ion and accumula ion o no el mu a ions may shi he dis inc pa e n o gene ic a ia ion (Rubin e al., 2022). Second, an inc ease in ecombina ion a e in he egion may esol e he dis inc pa e n o gene ic a ia ion and esul in eme gence o he popula ion s uc u e, because ecombina ion b eaks down disc e e haplo ypes and gene a es mixed ypes whe eby educing he a iance o gene ic a ia ion (Hudson, 1983). These wo ypes o shi s in dis inc pa e ns o gene ic 54 a ia ion a e no mu ually exclusi e. Fo example, ixa ion o an in e sion esul s in ele a ed ecombina ion a e (Smukowski Heil e al., 2015; S e ison e al., 2011) because he e a e no longe non- ecombining he e ozygo es in he popula ion. Due o esumed ecombina ion, pa e ns o local gene ic a ia ion in such egions a e expec ed o e lec popula ion s uc u e e en ually. The ques ion o how long i akes o an ou lie egion wi h dis inc pa e ns o gene ic a ia ion o disappea a e hese e en s should be ocally s udied in he u u e. In Fig. 6A, we illus a e a model o he e olu ion o local gene ic a ia ion ha changes acco ding p ima ily o he e olu ion o local ecombina ion a es. Local gene ic a ia ion can become dis inc om he popula ion s uc u e i s by ep esen ing eme ging haplo ype s uc u e associa ed wi h popula ion-speci ic ecombina ion supp ession o o he ypes o haplo ype blocks (e.g. in e sions) in one popula ion. I his ecombina ion supp ession sp eads h oughou all popula ions, hen local gene ic a ia ion will s a o e lec species-wide haplo ype s uc u e. Once he ela i e con ibu ion o haplo ype s uc u e on local gene ic a ia ion is educed by di e en ia ion o disappea s by ele a ed ecombina ion a es, hen gene ic a ia ion e u ns o e lec he popula ion s uc u e and consequen ly he ou lie egion disappea s. The e ec o selec ion on local gene ic a ia ion may be o e laid on op (Supplemen a y No es 1.2). 55 Species-wide ecombina ion supp ession In e sion PC1 PC2 Unde lying local ARGLocal gene ic a ia ion Popula ion-speci ic ecombina ion supp ession Inc ease in ecombina ion a e Di e en ia ion Non- ecombining Non- ecombining Non- ecombining Non- ecombining Non- ecombining A Conse ed ecombina ion map Rec. a e Regions wi h dis inc a ia ion Genomic posi ion Genomic posi ion Di e ged ecombina ion maps B ① ② ③ Figu e 6: E olu iona y changes in local ecombina ion a e in luence e olu ion o local gene ic a ia ion. A. Local gene ic a ia ion is shown in hypo he ical PCA plo s. Thei unde lying genealogies a e shown in simpli ied ances al ecombina ion g aphs (ARGs,(G i i hs & Ma jo am, 1997; e iewed in Lewanski e al., 2024)), on which black do s ep esen ances al ecombina ion e en s con ibu ing o he sampled sequences. Poin s in PCA depic diploid indi iduals, while hose on he ARGs ep esen haploid sequences. Two colou s o hese poin s (blue and o ange) indica e wo popula ions. (1) Local gene ic a ia ion conco dan o popula ion s uc u e. Gene ic a ia ion shows sepa a ion o indi iduals om wo popula ions. ARG shows ha ecombina ion is supp essed in nei he popula ion. (2) Popula ion-speci ic ecombina ion supp ession in he blue popula ion. ARG shows ha ecombina ion is supp essed in he blue popula ion. (3) Species-wide ecombina ion supp ession. Top: A case in which he e a e ew mu a ions ep esen ing he basal spli s o he unde lying genealogy a species-wide low- ecombining egion. Middle: A case in which he e a e wo haplo ypic a ian s a he species-wide low- ecombining egion. I his is due o p esence o an in e sion ( igh ARG), ecombina ion is supp essed be ween bu no wi hin he wo clades ep esen ing wo alleles. Bo om: A case in which he e a e h ee haplo ypic a ian s a he species-wide low- ecombining egion. B E olu ion o ecombina ion map in luences di e ence in genomic dis ibu ions o dis inc pa e ns o gene ic a ia ion be ween species/popula ions. 56 Implica ions Finally, we discuss echnical and biological implica ions o ou s udy. The echnical implica ion conce ns in e p e a ion o genome scans based on local gene ic a ia ion. A numbe o me hods based on local gene ic a ia ion ha e been used o de ec loci in ol ed in di e en kinds o selec i e p ocesses. Fo example, FST (di e en ia ion), dXY (di e gence), and o he popula ion pa ame e s a e in e ed o de ec genomic islands o specia ion (Delmo e e al., 2018; Hejase e al., 2020; Huang e al., 2020; Malinsky e al., 2015). Reduced di e si y ( π ) is a signa u e o selec ion (Delmo e e al., 2018; I win e al., 2018; P acana e al., 2017), and by combining i wi h a ia ion among popula ions, loci associa ed wi h popula ion-speci ic selec ion can be also in e ed (Yi e al., 2010). Ta ge s o adap i e in og ession ha e been iden i ied by applying s a is ics based on ABBA-BABA es , which is ela ed o gene ic a ia ion (Pe e , 2016, 2022), in sliding windows (K on o s e al., 2013; Ma in e al., 2015; Pa e son e al., 2012; Reich e al., 2009). Howe e , he e a e con ounding ac o s ha a ec in e ence o hese s a is ics. Fo example, i has been shown ha low di e si y can cause ele a ion in some o hese s a is ics (C uickshank & Hahn, 2014; Noo & Benne , 2009). In addi ion o educed di e si y, his s udy and o he s (Booke e al., 2020; Lo e hos, 2019; Renau e al., 2013) show ha educed ecombina ion a e also causes dis inc pa e ns o gene ic a ia ion which can lead o e oneous iden i ica ion o egions unde in luence o selec i e ac o s. Examining ecombina ion a es a iden i ied egions and compa ing hem o o he egions a e necessa y o a oid his. Fo ins ance, appa en ou lie s in only ew (pai s o ) popula ions a a low- ecombining egion may e lec high a iance, while high a iance a low- ecombining egions alone canno explain signals occu ing in many (quasi-) independen popula ions o species a a low- ecombining egion. Fu he mo e, co obo a ing me hods based on di e en aspec s o dis inc pa e ns o a ia ion, such as si e equency spec um (DeGio gio e al., 2016; Fay & Wu, 2000; Tajima, 1989), LD (Sabe i e al., 2002, 2007; Voigh e al., 2006), in e ed genealogies (Hejase e al., 2020; Speidel e al., 2019; S e n e al., 2019), local landscape o a ia ion (Se e e al., 2020), and si es o mu a ions in genes (Nei & Gojobo i, 1986), as well as app oaches wi h explici simula ion based on in e ed demog aphy (Hage e al., 2022), may be in o ma i e. The biological implica ion is abou e olu ion o ecombina ion a es and gene ic a ia ion 57 We de ined low- ecombining egions and e alua ed o e laps be ween ou lie egions and low- ecombining egions in he ollowing ou s eps. 1. de ine low- ecombining egions o each popula ion ecombina ion map, 2. es o associa ion be ween all ou lie egions and he low- ecombining egions o each popula ion, 3. de ine species-wide and popula ion-speci ic low- ecombining egions, and 4. label ou lie egions wi h species-wide o popula ion-speci ic low- ecombining egions o no o e lap wi h any low- ecombining egion. 1. Fo he ecombina ion map o each o he ou popula ions (med_sw, con _ es, Azo es, Cape Ve de), we de ined low- ecombining egions as he se o 10 kb windows wi h ecombina ion a e lowe han 20 pe cen ile o each ch omosome. This mild h eshold was se o accoun o la ge a ia ion in he ecombina ion landscapes among ch omosomes and o cap u e popula ion-speci ic educ ion in ecombina ion a e which could be wi h weake educ ion in ecombina ion a e han a species-wide low- ecombining egions. 2. Fo he se o low- ecombining egions o each popula ion, we pe o med a pe mu a ion es by shu ling obse ed ou lie egions wi hin he ch omosome and coun ed he o al leng h o o e lap wi h (any) low- ecombining egions (in bp) using BEDTools . We epea ed his 1,000 imes, and compa ed he empi ical null dis ibu ion o he o e lap leng h (in bp) wi h obse ed o e laps. 3. To de ine species-wide and popula ion-speci ic low- ecombining egions, a all posi ions along he genome we coun ed he numbe o popula ion ecombina ion maps sha ing low- ecombining egions. I a egion was labelled low- ecombining in h ee o ou popula ions a s ep 1, we de ined i o be species-wide low- ecombining egion. I a egion was labelled low- ecombining in one o wo popula ions, we de ined i o be a popula ion-speci ic low- ecombining egion, eco ding which popula ions we e low- ecombining. 4. We i s labelled ou lie egions o e lapping species-wide low- ecombining egions. We in e sec ed he species-wide low- ecombining egions de ined in s ep 3 and ou lie egions. We labelled an ou lie egion wi h species-wide low- ecombining i i had a co e age o species-wide low- ecombining egions g ea e han 0.5. Two excep ions we e ou lie _12_3 and ou lie _30_1, which a e pu a i e in e sions. They had co e age o species-wide low- ecombining egions o 0.30 and 0.23 bu his is la gely due o he e oka yo ype- 64 speci ic ecombina ion supp ession and inclusion o homoka yo ypes in ecombina ion a e in e ence. Because hese pu a i e in e sions we e seg ega ed in mos popula ions, we de ined hem o be species-wide low- ecombining egions. We nex labelled ou lie egions o e lapping popula ion-speci ic low- ecombining egions. We in e sec ed he popula ion-speci ic low- ecombining egion de ined in s ep 3 wi h ou lie egions excluding hose o e lapping wi h species-wide low- ecombining egions. We labelled an ou lie egion o e lapping wi h popula ion-speci ic low- ecombining egions i i had a co e age g ea e han 0.01 o any (pai o ) popula ion(s). Finally, he emaining ou lie egions we e labelled no educ ion in ecombina ion a e. To cha ac e ise geno ype-speci ic LD and ecombina ion landscape a he i e ou lie egions wi h h ee clus e s o indi iduals in PCA, we applied c ools --geno- 2 and Py ho (Spence & Song, 2019) o ou empi ical da a using each geno ype (AA, AB, and BB in Sup. Fig. 11) sepa a ely. Valida ion o his p ocedu e is desc ibed in “Simula ion: Valida ion o LD-based in e ence o ecombina ion landscape using non- andomly chosen samples”. In e sion b eakpoin s Th ee clus e s o indi iduals obse ed in PCA wi h geno ype-speci ic LD a wo ou lie egions on ch omosomes 12 and 30 we e indica i e o polymo phic in e sion (Ma & Amos, 2012; Ruiz- A enas e al., 2019). To u he cha ac e ise whe he hey ep esen polymo phic in e sions, we in ended o loca e b eakpoin s by wo independen app oaches. So -clip eads We a emp ed o iden i y posi ions whe e p esence o so -clipping o mapped eads is associa ed wi h PCA-based geno ype o he pu a i e in e sions. Fi s , we ex ac ed ocal egions a ound bounda ies o he ou lie s (Sup. Table 4) om ead mapping ile o all indi iduals using SAM ools (Danecek e al., 2021). Nex , we iden i ied so clip eads in each ex ac ed egion using samex ac clip (Lindenbaum, 2015), and ob ained e e ence posi ion co esponding o he posi ion o so clipping in mapped eads using a cus om sc ip . A all ex ac ed so -clip posi ions, we coun ed he numbe o eads ha swi ch o so -clip (“so -clip dep h”), as well as he dep h o mapped eads, using SAM ools . A each o all posi ions wi h a leas one ead suppo ing so -clip swi ch, we calcula ed p opo ion o eads 65 wi h so -clip swi ch ela i e o all mapped eads (dep h o he posi ion) o each indi idual (“so -clip p opo ion”). This esul ed in “posi ion-by-indi idual” ma ix whose en y depic s he p opo ion o so -clip in all eads mapped a he ocal posi ion o he ocal indi idual. Using his ma ix, we i a linea model ( so -clipp opo ion ∼PCA −basedgeno ype ) in R a each posi ion ea ing geno ypes AA, AB, and BB as 0, 1, and 2. Based on he signi icance o geno ype and R2 o he linea models, we gene a ed a lis o 14 posi ions a which so -clip p opo ion was signi ican ly associa ed wi h geno ype o he pu a i e in e sions. We isualised he dis ibu ion o he so -clip p opo ion a hese posi ions (Sup. Fig. 15) and selec ed six posi ions o which he so -clip p opo ion o BB was high enough and ha o AB was a ound a hal o BB based on he assump ion ha so clip eads co e ing an in e sion b eakpoin should o igina e om haplo ype B and non-so clip eads should o igina e om haplo ype A (Sup. Table 5). To in es iga e whe he some o hese six posi ions ep esen in e sion b eakpoin s, we asked whe he he so -clipped segmen s o he eads ha e homologous sequences a he o he end o he ou lie egions. We ex ac ed so -clipped segmen s o eads mapped a he ocal six posi ions in AB and BB indi iduals using a cus om sc ip , and e-mapped hese segmen s (ins ead o he en i e eads) o he blackcap e e ence using BWA mem . We compu ed he dep h o mapped segmen s in each posi ion using SAM ools (Sup. Table 5). 10x linked ead We used an independen se o blackcap indi iduals (he ea e “10x indi iduals”) whose genomes we e sequenced wi h he 10x linked- ead echnology (Delmo e e al., 2023, NCBI BioP ojec PRJEB65115). We geno yped he 10x indi iduals a he wo pu a i e in e sion loci (i.e. AA, AB, o BB) based on geno ypes a diagnos ic SNP posi ions. We s a ed by de e mining diagnos ic SNP posi ions using ou Illumina sho ead-based esequence da a. Because usable diagnos ic SNP posi ions should ha e geno ypes pe ec ly associa ed wi h PCA-based geno ype, we ocused on posi ions a which FST was 1 be ween AA and BB, and all AB we e he e ozygous, using VCF ools and BCF ools . We also eco ded mapping be ween an allele a he diagnos ic posi ions and a geno ype o he pu a i e in e sion (“A- and B-diagnos ic alleles”, e.g. G o haplo ype A, T o haplo ype B). We hen coun ed he numbe o si es wi h A- and B-diagnos ic allele in each o 10x samples. To con e coo dina es o 10x assemblies o he e e ence coo dina e, we mapped 66 he 10x pseudo-haplo yped assemblies o he blackcap e e ence using minimap2 (Li, 2018). To de e mine he pu a i e in e sion geno ype in he 10x indi iduals, we coun ed he numbe o posi ions wi h A-diagnos ic and B-diagnos ic alleles o each 10x pseudo-haplo ype, and calcula ed he p opo ion o si es wi h A-diagnos ic and B-diagnos ic si es. In p inciple, an AA and a BB indi idual espec i ely a e expec ed o ha e p opo ion o 100% and 0% o A-diagnos ic si es in bo h o wo pseudo-haplo ypes, while an AB indi idual is expec ed o ha e 100% o A-diagnos ic si es in one pseudo-haplo ype and 0% o he o he . Fo geno yping, we se he ollowing h ee h esholds. 1. Missingness a he diagnos ic posi ions is less han 10%, a e emo ing posi ions wi h non-unique minimap2 mapping (i.e. a leas 90% o all diagnos ic posi ions should ha e dep h o 1x). 2. Mo e han 90% o all diagnos ic si es should ag ee pe pseudo-haplo ype. 3. The second c i e ion should be ul illed o bo h pseudo-haplo ypes o an indi idual. We iden i ied wo BB indi iduals o each o he pu a i e in e sions on ch omosomes 12 and 30. The e we e no AB indi iduals passing he abo e h eshold, indica ing 10x pseudo- haplo yping is no accu a e in sepa a ing wo di e ged non- ecombining alleles a a long ange in an indi idual ha has bo h. To iden i y b eakpoin s, we aligned he pseudo-haplo ype assemblies o hese BB indi iduals as well as one AA indi idual o each pu a i e in e sion o he blackcap e e ence using Nucme 4 (Ma çais e al., 2018), and gene a ed do plo s (Sup. Fig. 16). Sequence analysis a b eakpoin o pu a i e in e sion on ch omosome 12 10x con igs o pseudo-haplo ype B aligned nex o he pu a i e b eakpoin posi ion o blackcap e e ence ch omosome 12 had an un-aligned lanking sequence. To cha ac e ise he DNA sequence o hese lanking segmen s, we ex ac ed he lanking sequences using SAM ools , aligned he sequences o hemsel es using minimap2 , and gene a ed sel -do plo s (Sup. Fig. 17), e ealing p esence o andem epea s. To iden i y uni o andem epea s wi hin he lanking sequences, we an TandemRepea sFinde agains hese ex ac ed sequences, esul ing in ou consensus uni sequences o 144 bp based on wo con igs om wo indi iduals. To 67 con i m ha he ou consensus sequences ep esen he same andem epea (because he uni o iden ical andem epea can ha e di e en phases), we an BLASTn ( e sion 2.10.1, Al schul e al., 1990) wi h each consensus as que y agains dime s o he consensus. To in es iga e whe he he andem epea ound a he pu a i e b eakpoin o ch omosome 12 in haplo ype B is p esen in ch omosome 12 and o he ch omosomes o he e e ence and co esponding posi ion o haplo ype A, we an BLASTn wi h he 144 bp consensus o he andem epea uni as he que y agains blackcap e e ence and a con ig o an AA indi idual ha spans he b eakpoin posi ion, and coun ed how many copies we e ound in each e e ence ch omosome/sca old and he 10x con ig (Sup. Fig. 18). Selec ion in blackcaps To es o selec ion in di e en ou lie egions and o compa e hem wi h he genome-wide base line, we compu ed nucleo ide di e si y ( π ) and Tajima’s D in 10 kb sliding windows pe popula ion using PopGenome (P ei e e al., 2014) and VCF ools (Danecek e al., 2011) espec i ely. The e ec s o he ou lie egions on hese s a is ics we e es ed using a linea mixed e ec s model ( nlme::lme (Pinhei o e al., 2021)) and a gene alised linea mixed e ec s model wi h a Gamma dis ibu ion ( lme4::glme (Ba es e al., 2015)). To es o selec ion in genes dN/dS we e compu ed ollowing he coun ing me hod by Nei & Gojobo i (1986). Gene anno a ion o he blackcap was ob ained om Bascón-Ca dozo e al. (2022b). Tandem epea s wi hin and ou side ou lie egions To cha ac e ise co ela ion be ween ou lie egions wi h dis inc pa e ns o gene ic a ia ion and andem epea s, we iden i ied andem epea s in he e e ence genome and compa ed he dis ibu ion o he andem epea s wi h genomic egions wi h dis inc pa e ns o gene ic a i- a ion. Fi s , TandemRepea sFinde (Benson, 1999) was un on he blackcap e e ence genome wi h he pa ame e se ecommended on he documen a ion ( </pa h/ o/ as a> 2 7 7 80 10 50 500 - -d -m -h ). The ou pu was o ma ed and summa ised o isualisa ion using cus om sc ip s. B ie ly, dis ibu ion o andem epea s wi h a di e en uni size along he genome was summa ised in 100 kb sliding windows in blocks o epea uni sizes o 10 bp s ep (Sup. Fig. 33). Tandem epea s wi h he six longes epea uni size we e ex ac ed pe 68 ch omosome, and copy numbe o each andem epea was coun ed (Sup. Fig. 34). Nex , we es ed whe he he numbe o andem epea s wi h long epea uni we e en iched in ou lie egions a species-wide and popula ion-speci ic low- ecombining egions. We ex ac ed andem epea s wi h epea uni size g ea e han o equal o 150 bp, and coun ed he numbe o andem epea s (ins ead o o al copy numbe ) wi hin and ou side ou lie egions. We pe o med Fishe ’s exac es s o es independence be ween he numbe o long andem epea s and he mode o ecombina ion supp ession (species-wide/popula ion-speci ic) (Sup. Table 7) using ishe . es unc ion in R. Simula ion Valida ion o LD-based in e ence o ecombina ion landscape using non- andomly chosen samples We asked whe he LD-based ecombina ion map in e ence using indi iduals chosen based on he ka yo ype ins ead o a andom is in o ma i e o he unde lying mode o ecombina ion supp ession. To his end, we simula ed wo 5 Mb-long ch omosomes wi h neu al mu a ion a e o 4 . 6 × 10 −8 in a popula ion o 1,000 indi iduals in SLiM . The pu pose o hese simula ions was o in es iga e he e ec o an in e sion and addi ional ecombina ion supp ession on ecombina ion a e in e ence and LD in gene al, a he han in es iga ing he e ec s speci ic o blackcap demog aphy. As such, we kep he popula ion size smalle han he blackcap e ec i e popula ion size and he mu a ion a e g ea e han assumed in o de o minimise he ime and compu a ional esou ce o simula ions. We in oduced a mu a ion (in e sion ma ke ) on one ch omosome a 1 Mb posi ion a he 50 h gene a ion. We simula ed an in e sion on he ch omosome by supp essing ecombina ion in an in e al om 1 Mb o 4 Mb posi ion i he in e sion ma ke si e was he e ozygous. We de ined addi ional supp ession acco ding o di e en scena ios (models 1-6 in Sup. Table 6). We applied nega i e equency-dependen selec ion ( i ness o in e sion is 1 − ( pin − 0 . 2) whe e pin is he equency o he in e sion allele). 1,000 gene a ions a e he in e sion e en , we eco ded he mu a ions in all samples, making a VCF ile including all samples. Al hough 1,000 gene a ions is ela i ely sho gi en he popula ion size o 1,000, he haplo ype s uc u e a he in e sion locus was s able in es 69 uns o model-1 (in e sion equency o 0.2 wi hou addi ional ecombina ion supp ession). Based on he geno ype a 1 Mb posi ion, we andomly chose 10 samples o each in e sion geno ype. Py ho was un o es ima e ecombina ion a es using he chosen 10 samples, wi h he block penal y 50 and window size 50. The in e ed ecombina ion maps a e in Sup. Fig. 13. E ec s o ecombina ion supp ession model on ecombina ion a e in e ence a an in e sion Th ee clus e s o indi iduals obse ed in PCA a i e ou lie egions indica e p esence o dis inc haplo ypes. Polymo phic in e sions a e known o show his pa e n due o supp ession o ecombina ion be ween he no mal and in e ed alleles (Wellen eu he & Be na chez, 2018). To es whe he some o he i e ou lie egions ep esen polymo phic in e sions, we in ended o in e ecombina ion a es using AA, AB, and BB indi iduals sepa a ely based on linkage disequilib ium (LD) pa e ns. Be o e add essing his in blackcaps empi ically, we assessed how di e en ypes o ecombina ion supp ession a a haplo ype block a ec in e ence o ecombina ion landscape using a se o indi iduals wi h a ce ain combina ion o haplo ypes. To in es iga e he e ec o a geno ype-speci ic supp ession o ecombina ion on LD-based in e ence o ecombina ion a e, we simula ed di e en modes o ecombina ion supp ession using SLiM e sion 3.5 (Halle & Messe , 2019) unde six scena ios lis ed in Sup. Table 6. Speci ically, we pe o med 1,000 eplica es o o wa d- ime simula ions o wo 500 kb-long ch omosomes wi h neu al mu a ion a e o 1 × 10 −7 [pe si e pe gene a ion] and ecombina ion a e o 1 × 10 −6 [pe si e pe gene a ion] in a popula ion o 1,000 diploid indi iduals unde he W igh -Fishe model (We downscaled he popula ion size and upscaled mu a ion a e o minimise he ime and compu a ional esou ce o simula ion). We in oduced a mu a ion (in e sion ma ke ) on one ch omosome a 100 kb posi ion a he 50 h gene a ion. We modelled an in e sion by supp essing ecombina ion in an in e al om 100 kb o 400 kb posi ion i he in e sion ma ke si e was he e ozygous. We de ined addi ional supp ession acco ding o di e en scena ios (models 1-6). To allow o he in e sion o emain in he popula ion, we applied nega i e equency-dependen selec ion ( i ness o in e sion is 1 − ( pin − 0 . 2) o models 1-3 and 1 − ( pin − 0 . 8) o models 4-6 whe e pin is he equency o he in e sion 70 allele). 1,000 gene a ions a e he in e sion e en , we eco ded he mu a ions in all samples, making a VCF ile including all indi iduals. Al hough 1,000 gene a ions is ela i ely sho gi en he popula ion size o 1,000, he haplo ype s uc u e a he in e sion locus was s able in es uns o model-1 (in e sion equency o 0.2 wi hou addi ional ecombina ion supp ession). Based on he geno ype a he ma ke , we andomly sampled 10 indi iduals o each in e sion geno ype. Py ho was un o es ima e ecombina ion a es using he sampled 10 indi iduals, wi h he block penal y 50 and window size 50. The in e ed ecombina ion landscape is in Sup. Fig. 13. Coalescen simula ion o species-wide educ ion o ecombina ion a e To disce n he e ec o educed ecombina ion a e, demog aphic his o y, and unequal sample sizes among popula ion on ou lie egions iden i ied by los uc , we pe o med neu al coalescen simula ions using msp ime e sion 1.2.0 (Baumdicke e al., 2022). We simula ed a 1-Mb long ecombining ch omosome wi h a mu a ion a e o 4 . 6 × 10 −9 [pe si e pe gene a ion]. We implemen ed 11 models di e ing in he ecombina ion maps, popula ion subdi ision, and demog aphic his o y (Sup. Fig. 19). In models 1-3, he ecombina ion a e was se o 4 . 6 × 10 −9 [pe si e pe gene a ion] h oughou he en i e ch omosome, and hey di e in popula ion subdi ision (model 1: panmic ic, model 2, subdi ision o i e equal popula ions wi hou gene low, model 3: subdi ision o equally-sized popula ions wi h gene low be ween wo pai s o popula ions (symme ic mig a ion a e o 0.025 [pe gene a ion])). In models wi h i e popula ions, we dis ibu ed he sample o 100 indi iduals unequally, as in ou blackcap da ase (50, 20, 10, 10, 10 indi iduals o i e popula ions). In models 4-7, we in oduced educed ecombina ion a e in he middle o he ch omosome (0.4 o 0.6 Mb) wi h he same demog aphic his o ies as models 2 and 3. In addi ion o he uni o m ecombina ion map, we p epa ed wo ecombina ion maps wi h educed ecombina ion a e: “low- ec” wi h one-hund e h he backg ound ecombina ion a e, and “no- ec” wi h ecombina ion a e o 0. In models 8-11, we used he same wo ecombina ion maps wi h educed ecombina ion a e in he middle, wi h di e en demog aphy: 10 imes inc ease in e ec i e popula ion size in one popula ion, and 10 imes desc ease in e ec i e popula ion size in h ee popula ions, which oughly e lec s in e ed demog aphy o blackcap popula ions (Delmo e e al., 2020b). Fo each model, we an 71 1,000 eplica es o simula ions and eco ded SNPs in VCF o ma . To iden i y ou lie egions, we an los uc he same way as in he empi ical analysis. To e alua e how educed ecombina ion a e a ec s he mean and a iance o popula ion gene ic summa y s a is ics, we compu ed nucleo ide di e si y ( π ), Tajima’s D, and FST , using VCFTools . The ou lie s de ec ed by los uc a e in Sup. Fig. 20. The summa y s a is ics a e in Sup. Figs. 21, 22, 23. Fo wa d simula ion o species-wide educ ion o ecombina ion a e To in es iga e how species-wide low- ecombining egions a ec pa e ns o local gene ic a ia ion depic ed in local PCA, we pe o med o wa d simula ion wi h SLiM e sion 4.0.1 (Halle & Messe , 2022). We simula ed 100 eplica es o wo 500 kb-long ch omosomes wi h neu al mu a ion a e o 1 × 10 −7 [pe si e pe gene a ion] and ecombina ion a e o 1 × 10 −6 [pe si e pe gene a ion] excep o an in e al om 100 o 400 [kb] o he i s ch omosome whe e ecombina ion a e was se o 1 × 10 −9 , which is 1/1000 o he no mally ecombining ch omosome. Fi s , we an a bu n-in o 4,000 gene a ions o an ances al popula ion o 1,000 diploids. A e he bu n-in, we made h ee popula ions o 1,000 diploids (pop1, pop2, and pop3) spli om he ances al popula ion. We sampled 50 indi iduals pe popula ion e e y 20 gene a ions o e 1,000 gene a ions a e he popula ion spli and eco ded SNPs in VCF. Fo each ime poin o each o 100 simula ion eplica es, we pe o med PCA wi h PLINK , using SNPs ei he wi hin 100 o 400 [kb] o he i s ch omosome (pop1-speci ic supp ession) o he no mally ecombining ch omosome. We in es iga ed how educed ecombina ion a e a ec s ep esen a ion o popula ion subdi ision in local PCA. To e alua e whe he he indi iduals om di e en popula ions we e dis ibu ed di e en ly in local PCA a he low- ecombining egion, we pe o med Fasano- F anceschini es (Fasano & F anceschini, 1987), which is a mul i-dimensional ex ension o Kolmogo o -Smi no es , in h ee pai s o popula ions (pop1-pop2, pop1-pop3, pop2-pop3). We coun ed he numbe o signi ican pai s o popula ions (0, 1, 2, o 3) o each ime poin o each eplica e. We compa ed be ween he low- ecombining and no mally ecombining egions he numbe o pai s o popula ions wi h dis inc dis ibu ion in PCA (Sup. Fig. 31). 72 Fo wa d simula ion o popula ion-speci ic educ ion o ecombina ion a e To in es iga e how e olu ion o low- ecombining egions in popula ion(s) a ec pa e ns o local gene ic a ia ion depic ed in local PCA, we pe o med o wa d simula ion wi h SLiM e sion 4.0.1. We simula ed wo 500kb-long ch omosomes wi h neu al mu a ion a e and ecombina ion a e o 1 × 10 −7 [pe si e pe gene a ion] and 1 × 10 −6 [pe si e pe gene a ion]. Fi s , we an a bu n-in o 4,000 gene a ions o an ances al popula ion o 1,000 diploids. A e he bu n-in, we made h ee popula ions o 1,000 diploids (pop1, pop2, and pop3) spli om he ances al popula ion, a e which gene low be ween all pai s o popula ions we e se o 0.0025. We in oduced ecombina ion supp ession in pop1 om 100 o 400 [kb] o he i s ch omosome in wo scena ios. In he i s scena io, ecombina ion supp ession was in oduced a he same ime o he spli . In he second scena io, ecombina ion supp ession was in oduced 4,000 gene a ions a e he popula ion spli e en , allowing he h ee popula ions o di e en ia e be o e popula ion-speci ic ecombina ion supp ession was in oduced in pop1. We sampled 50 indi iduals pe popula ion e e y 20 gene a ions o e 1,000 gene a ions a e he in oduc ion o he popula ion-speci ic supp ession o ecombina ion and eco ded SNPs in VCF. Fo each ime poin o each o 1,000 simula ion eplica es, we pe o med PCA wi h PLINK , using SNPs ei he wi hin 100 o 400 [kb] o he i s ch omosome (pop1-speci ic supp ession) o he no mally ecombining ch omosome. To cha ac e ise ac o s ep esen ed in he p ima y axes o dis inc local PCA a popula ion- speci ic low- ecombining egions, we pe o med one eplica e o SLiM simula ion wi h he same scena ios o models 1 and 2 eco ding he ull ances y and mu a ions in ee sequence, wi h an inc eased du a ion o bu n-in (40,000 gene a ions) o make su e ha all lineages a sampling ime coalesce. We loaded he ee sequence wi h mu a ions in ski (Kellehe e al., 2018) and sampled 50 diploids pe popula ion, and sa ed SNPs in VCF. Using he VCF iles o each ime poin o each model, we pe o med PCA using PLINK a he popula ion-speci ic low- ecombining egion, and de e mined one ime poin pe model showing ypical sp ead o indi iduals om he low- ecombining popula ion in PCA (Sup. Fig. 27A, E). Fo hese PCAs we iden i ied 5% SNPs wi h he highes loadings o he i s wo PC axes. We analysed hese mu a ions on he unde lying genealogies using ski . Speci ically, we in es iga ed 73 ecombina ion. Molecula Ecology,20(22), 4722–4736. h ps://doi.o g/10.1111/j.1365- 294X.2011.05285.x Ghu ye, J., Rhie, A., Walenz, B. P., Schmi , A., Sel a aj, S., Pop, M., Phillippy, A. M., & Ko en, S. (2019). In eg a ing Hi-C links wi h assembly g aphs o ch omosome-scale assembly. PLOS Compu a ional Biology,15(8), e1007273. h ps://doi.o g/10.1371/jou nal.pcbi.1007273 G i i hs, R. C., & Ma jo am, P. (1997). An ances al ecombina ion g aph. In P. Donnelly & S. Ta a é (Eds.), P og ess in Popula ion Gene ics and Human E olu ion (Vol. 87, pp. 257–270). Sp inge . h p://lamas ex.o g/ ecomb/ima.pd Guan, D., McCa hy, S. A., Wood, J., Howe, K., Wang, Y., & Du bin, R. (2020). Iden i ying and emo ing haplo ypic duplica ion in p ima y genome assemblies. Bioin o ma ics, 36(9), 2896–2898. h ps://doi.o g/10.1093/bioin o ma ics/b aa025 Gue e o, R. F., Rousse , F., & Ki kpa ick, M. (2012). Coalescen pa e ns o ch omosomal in e sions in di e gen popula ions. Philosophical T ansac ions o he Royal Socie y B: Biological Sciences,367(1587), 430–438. h ps://doi.o g/10.1098/ s b.2011.0246 Hage , E. R., Ha ingmeye , O. S., Woold idge, T. B., Theingi, S., Gable, J. T., McFadden, S., Neugebo en, B., Tu ne , K. M., Jensen, J. D., & Hoeks a, H. E. (2022). A ch omosomal in e sion con ibu es o di e gence in mul iple ai s be ween dee mouse eco ypes. Science,377 (6604), 399–405. h ps://doi.o g/10.1126/science.abg0718 Halle , B. C., & Messe , P. W. (2019). SLiM 3: Fo wa d Gene ic Simula ions Beyond he W igh –Fishe Model. Molecula Biology and E olu ion,36(3), 632–637. h ps: //doi.o g/10.1093/molbe /msy228 Halle , B. C., & Messe , P. W. (2022). SLiM 4: Mul ispecies Eco-E olu iona y Modeling. The Ame ican Na u alis , E000–E000. h ps://doi.o g/10.1086/723601 Hejase, H. A., Salman-Minko , A., Campagna, L., Hubisz, M. J., Lo e e, I. J., G onau, I., & Siepel, A. (2020). Genomic islands o di e en ia ion in a apid a ian adia ion ha e been d i en by ecen selec i e sweeps. P oceedings o he Na ional Academy o Sciences,117(48), 30554–30565. h ps://doi.o g/10.1073/pnas.2015987117 Helbig, A. J. (1991). Inhe i ance o mig a o y di ec ion in a bi d species: A c oss-b eeding expe imen wi h SE- and SW-mig a ing blackcaps (Syl ia a icapilla). Beha io al Ecology and Sociobiology,28(1), 9–12. h ps://doi.o g/10.1007/BF00172133 Howe, K., Chow, W., Collins, J., Pelan, S., Poin on, D.-L., Sims, Y., To ance, J., T acey, A., & Wood, J. (2021). Signi ican ly imp o ing he quali y o genome assemblies h ough cu a ion. GigaScience,10(1), giaa153. h ps://doi.o g/10.1093/gigascience/giaa153 Huang, K., And ew, R. L., Owens, G. L., Os e ik, K. L., & Riesebe g, L. H. (2020). Mul iple ch omosomal in e sions con ibu e o adap i e di e gence o a dune sun lowe eco ype. Molecula Ecology,29(14), 2535–2549. h ps://doi.o g/h ps://doi.o g/10.111 1/mec.15428 Hudson, R. R. (1983). P ope ies o a neu al allele model wi h in agenic ecombina ion. Theo e ical Popula ion Biology,23(2), 183–201. h ps://doi.o g/10.1016/0040-5809(83 )90013-8 80 I win, D. E., Milá, B., Toews, D. P. L., B els o d, A., Kenyon, H. L., Po e , A. N., G ossen, C., Delmo e, K. E., Alcaide, M., & I win, J. H. (2018). A compa ison o genomic islands o di e en ia ion ac oss h ee young a ian species pai s. Molecula Ecology,27(23), 4839–4855. h ps://doi.o g/h ps://doi.o g/10.1111/mec.14858 Jones, F. C., G abhe , M. G., Chan, Y. F., Russell, P., Mauceli, E., Johnson, J., Swo o d, R., Pi un, M., Zody, M. C., Whi e, S., Bi ney, E., Sea le, S., Schmu z, J., G imwood, J., Dickson, M. C., Mye s, R. M., Mille , C. T., Summe s, B. R., Knech , A. K., e al. (2012). The genomic basis o adap i e e olu ion in h eespine s icklebacks. Na u e, 484(7392), 55–61. h ps://doi.o g/10.1038/na u e10944 Kawakami, T., Mugal, C. F., Suh, A., Na e , A., Bu i, R., Smeds, L., & Elleg en, H. (2017). Whole-genome pa e ns o linkage disequilib ium ac oss lyca che popula ions cla i y he causes and consequences o ine-scale ecombina ion a e a ia ion in bi ds. Molecula Ecology,26(16), 4158–4172. h ps://doi.o g/10.1111/mec.14197 Kellehe , J., Tho n on, K. R., Ashande , J., & Ralph, P. L. (2018). E icien pedig ee eco ding o as popula ion gene ics simula ion. PLOS Compu a ional Biology,14 (11), e1006581. h ps://doi.o g/10.1371/jou nal.pcbi.1006581 Knie , U., Hemm ich-S anisak, G., Wi ig, M., F anke, A., G i i h, S. C., Kempenae s, B., & Fo s meie , W. (2016). Fi ness consequences o polymo phic in e sions in he zeb a inch genome. Genome Biology,17(1), 199. h ps://doi.o g/10.1186/s13059-016-1056-3 Kong, A., Tho lei sson, G., Gudbja sson, D. F., Masson, G., Sigu dsson, A., Jonasdo i , A., Wal e s, G. B., Jonasdo i , A., Gyl ason, A., K is insson, K. T., Gudjonsson, S. A., F igge, M. L., Helgason, A., Tho s einsdo i , U., & S e ansson, K. (2010). Fine-scale ecombina ion a e di e ences be ween sexes, popula ions and indi iduals. Na u e, 467(7319), 1099–1103. h ps://doi.o g/10.1038/na u e09525 K on o s , M. R., Hansen, M. E. B., C aw o d, N. G., Gallan , J. R., Zhang, W., Kula hinal, R. J., Kapan, D. D., & Mullen, S. P. (2013). Hyb idiza ion Re eals he E ol ing Genomic A chi ec u e o Specia ion. Cell Repo s,5(3), 666–677. h ps://doi.o g/10.1 016/j.cel ep.2013.09.042 Lamichhaney, S., Han, F., Be glund, J., Wang, C., Almén, M. S., Webs e , M. T., G an , B. R., G an , P. R., & Ande sson, L. (2016). A beak size locus in Da win’s inches acili a ed cha ac e displacemen du ing a d ough . Science,352(6284), 470–474. h ps://doi.o g/10.1126/science.aad8786 Lawniczak, M. K. N., Em ich, S. J., Holloway, A. K., Regie , A. P., Olson, M., Whi e, B., Redmond, S., Ful on, L., Appelbaum, E., God ey, J., Fa me , C., Chinwalla, A., Yang, S.-P., Minx, P., Nelson, J., Kyung, K., Walenz, B. P., Ga cia-He nandez, E., Aguia , M., e al. (2010). Widesp ead Di e gence Be ween Incipien Anopheles gambiae Species Re ealed by Whole Genome Sequences. Science,330(6003), 512–514. h ps://doi.o g/10.1126/science.1195755 Lewanski, A. L., G undle , M. C., & B adbu d, G. S. (2024). The e a o he ARG: An in oduc ion o ances al ecombina ion g aphs and hei signi icance in empi ical e olu iona y genomics. PLOS Gene ics,20 (1), e1011110. h ps://doi.o g/10.1371/jou nal.pgen.1011110 81 Li, H. (2013). Aligning sequence eads, clone sequences and assembly con igs wi h BWA- MEM [P ep in ]. a Xi . h ps://doi.o g/h ps://doi.o g/10.48550/a Xi .1303.3997 Li, H. (2018). Minimap2: Pai wise alignmen o nucleo ide sequences. Bioin o ma ics, 34(18), 3094–3100. h ps://doi.o g/10.1093/bioin o ma ics/b y191 Li, H., & Ralph, P. (2019). Local PCA Shows How he E ec o Popula ion S uc u e Di e s Along he Genome. Gene ics,211(1), 289–304. h ps://doi.o g/10.1534/gene i cs.118.301747 Lindenbaum, P. (2015). JVa ki : Ja a-based u ili ies o Bioin o ma ics.h ps://doi.o g/ 10.6084/m9. igsha e.1425030. 1 Lo e hos, K. E. (2019). The E ec o Neu al Recombina ion Va ia ion on Genome Scans o Selec ion. G3 GenesGenomesGene ics,9(6), 1851–1867. h ps://doi.o g/10.1534/g3 .119.400088 Lundbe g, M., Mackin osh, A., Pe i, A., & Bensch, S. (2021). In e sions main ain di e ences be ween mig a o y pheno ypes o a songbi d. bioRxi . h ps://doi.o g/10.110 1/2021.04.05.438456 Ma, J., & Amos, C. I. (2012). In es iga ion o In e sion Polymo phisms in he Human Genome Using P incipal Componen s Analysis. PLOS ONE,7(7), e40224. h ps: //doi.o g/10.1371/jou nal.pone.0040224 Malinsky, M., Challis, R. J., Tye s, A. M., Schi els, S., Te ai, Y., Nga unga, B. P., Miska, E. A., Du bin, R., Genne , M. J., & Tu ne , G. F. (2015). Genomic islands o specia ion sepa a e cichlid ecomo phs in an Eas A ican c a e lake. Science,350(6267), 1493–1498. h ps://doi.o g/10.1126/science.aac9927 Ma çais, G., Delche , A. L., Phillippy, A. M., Cos on, R., Salzbe g, S. L., & Zimin, A. (2018). MUMme 4: A as and e sa ile genome alignmen sys em. PLOS Compu a ional Biology,14(1), e1005944. h ps://doi.o g/10.1371/jou nal.pcbi.1005944 Ma in, S. H., Da ey, J. W., & Jiggins, C. D. (2015). E alua ing he Use o ABBA–BABA S a is ics o Loca e In og essed Loci. Molecula Biology and E olu ion,32(1), 244–257. h ps://doi.o g/10.1093/molbe /msu269 Ma in, S. H., Da ey, J. W., Salaza , C., & Jiggins, C. D. (2019). Recombina ion a e a ia ion shapes ba ie s o in og ession ac oss bu e ly genomes. PLoS Biology,17(2), 1–28. h ps://doi.o g/10.1371/jou nal.pbio.2006288 Ma in, S. H., & Van Belleghem, S. M. (2017). Explo ing E olu iona y Rela ionships Ac oss he Genome Using Topology Weigh ing. Gene ics,206(1), 429–438. h ps: //doi.o g/10.1534/gene ics.116.194720 McKenna, A., Hanna, M., Banks, E., Si achenko, A., Cibulskis, K., Ke ny sky, A., Ga imella, K., Al shule , D., Gab iel, S., Daly, M., & DeP is o, M. A. (2010). The Genome Analysis Toolki : A MapReduce amewo k o analyzing nex -gene a ion DNA sequencing da a. Genome Resea ch,20(9), 1297–1303. h ps://doi.o g/10.1101/g .107524.110 McVean, G. (2009). A Genealogical In e p e a ion o P incipal Componen s Analysis. PLOS Gene ics,5(10), e1000686. h ps://doi.o g/10.1371/jou nal.pgen.1000686 82 McVean, G. A. T., & Ca din, N. J. (2005). App oxima ing he coalescen wi h ecombina ion. Philosophical T ansac ions o he Royal Socie y B: Biological Sciences,360(1459), 1387– 1393. h ps://doi.o g/10.1098/ s b.2005.1673 Mé o , C., Be dan, E. L., Cayuela, H., Djambazian, H., Fe chaud, A.-L., Lapo e, M., No mandeau, E., Ragoussis, J., Wellen eu he , M., & Be na chez, L. (2021). Locally Adap i e In e sions Modula e Gene ic Va ia ion a Di e en Geog aphic Scales in a Seaweed Fly. Molecula Biology and E olu ion,38(9), 3953–3971. h ps://doi.o g/10.1 093/molbe /msab143 Mye s, S., F eeman, C., Au on, A., Donnelly, P., & McVean, G. (2008). A common sequence mo i associa ed wi h ecombina ion ho spo s and genome ins abili y in humans. Na u e Gene ics,40(9), 1124–1129. h ps://doi.o g/10.1038/ng.213 Nea sey, D. E., Lawniczak, M. K. N., Pa k, D. J., Redmond, S. N., Coulibaly, M. B., T ao é, S. F., Sagnon, N., Cos an ini, C., Johnson, C., Wiegand, R. C., Collins, F. H., Lande , E. S., Wi h, D. F., Ka a os, F. C., Besansky, N. J., Ch is ophides, G. K., & Muska i ch, M. A. T. (2010). SNP Geno yping De ines Complex Gene-Flow Bounda ies Among A ican Mala ia Vec o Mosqui oes. Science,330(6003), 514–517. h ps://doi.o g/10.1126/science.1193036 Nei, M., & Gojobo i, T. (1986). Simple me hods o es ima ing he numbe s o synonymous and nonsynonymous nucleo ide subs i u ions. Molecula Biology and E olu ion,3(5), 418–426. h ps://doi.o g/10.1093/ox o djou nals.molbe .a040410 Nielsen, R. (2005). Molecula Signa u es o Na u al Selec ion. Annual Re iew o Gene ics, 39(1), 197–218. h ps://doi.o g/10.1146/annu e .gene .39.073003.112420 Noo , M., & Benne , S. (2009). Islands o specia ion o mi ages in he dese ? Examining he ole o es ic ed ecombina ion in main aining species. He edi y,103, 439–444. Okonechniko , K., Conesa, A., & Ga cía-Alcalde, F. (2016). Qualimap 2: Ad anced mul i- sample quali y con ol o high- h oughpu sequencing da a. Bioin o ma ics,32(2), 292–294. h ps://doi.o g/10.1093/bioin o ma ics/b 566 Pamilo, P., & Nei, M. (1988). Rela ionships be ween gene ees and species ees. Molecula Biology and E olu ion,5(5), 568–583. h ps://doi.o g/10.1093/ox o djou nals.molbe . a040517 Pa e son, N., Moo jani, P., Luo, Y., Mallick, S., Rohland, N., Zhan, Y., Genscho eck, T., Webs e , T., & Reich, D. (2012). Ancien Admix u e in Human His o y. Gene ics, 192(3), 1065–1093. h ps://doi.o g/10.1534/gene ics.112.145037 Pa e son, N., P ice, A. L., & Reich, D. (2006). Popula ion S uc u e and Eigenanalysis. PLOS Gene ics,2(12), e190. h ps://doi.o g/10.1371/jou nal.pgen.0020190 Pe e , B. M. (2016). Admix u e, Popula ion S uc u e, and F-S a is ics. Gene ics,202(4), 1485–1501. h ps://doi.o g/10.1534/gene ics.115.183913 Pe e , B. M. (2022). A geome ic ela ionship o F2, F3 and F4-s a is ics wi h p incipal componen analysis. Philosophical T ansac ions o he Royal Socie y B: Biological Sciences,377(1852), 20200413. h ps://doi.o g/10.1098/ s b.2020.0413 83 P ei e , B., Wi elsbue ge , U., Ramos-Onsins, S. E., & Le che , M. J. (2014). PopGenome: An E icien Swiss A my Kni e o Popula ion Genomic Analyses in R. Molecula Biology and E olu ion,31, 1929–1936. h ps://doi.o g/10.1093/molbe /msu136 Pinhei o, J., Ba es, D., DebRoy, S., Sa ka , D., & R Co e Team. (2021). Nlme: Linea and Nonlinea Mixed E ec s Models.h ps://CRAN.R-p ojec .o g/package=nlme P acana, R., P iyam, A., Le an is, I., Nichols, R. A., & Wu m, Y. (2017). The i e an social ch omosome supe gene a ian Sb shows low di e si y bu high di e gence om SB. Molecula Ecology,26(11), 2864–2879. h ps://doi.o g/10.1111/mec.14054 P ice, A. L., Pa e son, N. J., Plenge, R. M., Weinbla , M. E., Shadick, N. A., & Reich, D. (2006). P incipal componen s analysis co ec s o s a i ica ion in genome-wide associa ion s udies. Na u e Gene ics,38(8), 904–909. h ps://doi.o g/10.1038/ng1847 Pu cell, S., Neale, B., Todd-B own, K., Thomas, L., Fe ei a, M. A. R., Bende , D., Malle , J., Skla , P., Bakke , P. I. W. de, Daly, M. J., & Sham, P. C. (2007). PLINK: A Tool Se o Whole-Genome Associa ion and Popula ion-Based Linkage Analyses. The Ame ican Jou nal o Human Gene ics,81(3), 559–575. h ps://doi.o g/10.1086/519795 Ralph, P., Tho n on, K., & Kellehe , J. (2020). E icien ly Summa izing Rela ionships in La ge Samples: A Gene al Duali y Be ween S a is ics o Genealogies and Genomes. Gene ics,215(3), 779–797. h ps://doi.o g/10.1534/gene ics.120.303253 Reich, D., Thanga aj, K., Pa e son, N., P ice, A. L., & Singh, L. (2009). Recons uc ing Indian popula ion his o y. Na u e,461(7263), 489–494. h ps://doi.o g/10.1038/na u e 08365 Renau , S., G assa, C. J., Yeaman, S., Moye s, B. T., Lai, Z., Kane, N. C., Bowe s, J. E., Bu ke, J. M., & Riesebe g, L. H. (2013). Genomic islands o di e gence a e no a ec ed by geog aphy o specia ion in sun lowe s. Na u e Communica ions,4(1), 1827. h ps://doi.o g/10.1038/ncomms2833 Rhie, A., McCa hy, S. A., Fed igo, O., Damas, J., Fo men i, G., Ko en, S., Uliano-Sil a, M., Chow, W., Fung ammasan, A., Kim, J., Lee, C., Ko, B. J., Chaisson, M., Gedman, G. L., Can in, L. J., Thibaud-Nissen, F., Hagge y, L., Bis a, I., Smi h, M., e al. (2021). Towa ds comple e and e o - ee genome assemblies o all e eb a e species. Na u e, 592(7856), 737–746. h ps://doi.o g/10.1038/s41586-021-03451-0 Roes i, M., Mose , D., & Be ne , D. (2013). Recombina ion in he h eespine s ickleback genome—pa e ns and consequences. Molecula Ecology,22(11), 3014–3027. h ps: //doi.o g/h ps://doi.o g/10.1111/mec.12322 Rougemon , Q., Xue eb, A., Dallai e, X., Moo e, J.-S., No mandeau, E., Rondeau, E. B., Wi hle , R. E., Van Doo nik, D. M., C ane, P. A., Naish, K. A., Ga za, J. C., Beacham, T. D., Koop, B. F., & Be na chez, L. (2021). Long-dis ance mig a ion is a majo ac o d i ing local adap a ion a con inen al scale in Coho salmon. Molecula Ecology,n/a(n/a). h ps://doi.o g/10.1111/mec.16339 Rubin, C.-J., Enbody, E. D., Dob e a, M. P., Abzhano , A., Da is, B. W., Lamichhaney, S., Pe e sson, M., Sendell-P ice, A. T., Sp ehn, C. G., Valle, C. A., Vasco, K., Walle man, O., G an , B. R., G an , P. R., & Ande sson, L. (2022). Rapid adap i e adia ion 84 o Da win’s inches depends on ances al gene ic modules. Science Ad ances,8(27), eabm5982. h ps://doi.o g/10.1126/sciad .abm5982 Ruiz-A enas, C., Cáce es, A., López-Sánchez, M., Tolosana, I., Pé ez-Ju ado, L., & González, J. R. (2019). sco eIn Hap: In e sion geno yping o genome-wide associa ion s udies. PLOS Gene ics,15(7), e1008203. h ps://doi.o g/10.1371/jou nal.pgen.1008203 Sabe i, P. C., Reich, D. E., Higgins, J. M., Le ine, H. Z. P., Rich e , D. J., Scha ne , S. F., Gab iel, S. B., Pla ko, J. V., Pa e son, N. J., McDonald, G. J., Acke man, H. C., Campbell, S. J., Al shule , D., Coope , R., Kwia kowski, D., Wa d, R., & Lande , E. S. (2002). De ec ing ecen posi i e selec ion in he human genome om haplo ype s uc u e. Na u e,419(6909), 832–837. h ps://doi.o g/10.1038/na u e01140 Sabe i, P. C., Va illy, P., F y, B., Lohmuelle , J., Hos e e , E., Co sapas, C., Xie, X., By ne, E. H., McCa oll, S. A., Gaude , R., Scha ne , S. F., Lande , E. S., F aze , K. A., Ballinge , D. G., Cox, D. R., Hinds, D. A., S u e, L. L., Gibbs, R. A., Belmon , J. W., e al. (2007). Genome-wide de ec ion and cha ac e iza ion o posi i e selec ion in human popula ions. Na u e,449(7164), 913–918. h ps://doi.o g/10.1038/na u e06250 Se e , D., Mousse , S., Cheng, X., Nielsen, R., DeGio gio, M., & He misson, J. (2020). VolcanoFinde : Genomic scans o adap i e in og ession. PLOS Gene ics,16(6), e1008867. h ps://doi.o g/10.1371/jou nal.pgen.1008867 Shao, C., Sun, S., Liu, K., Wang, J., Li, S., Liu, Q., Deagle, B. E., Seim, I., Biscon in, A., Wang, Q., Liu, X., Kawaguchi, S., Liu, Y., Ja man, S., Wang, Y., Wang, H.-Y., Huang, G., Hu, J., Feng, B., e al. (2023). The eno mous epe i i e An a c ic k ill genome e eals en i onmen al adap a ions and popula ion insigh s. Cell,186(6), 1279–1294.e19. h ps://doi.o g/10.1016/j.cell.2023.02.005 Shipilina, D., Pal, A., S ankowski, S., Chan, Y. F., & Ba on, N. H. (2023). On he o igin and s uc u e o haplo ype blocks. Molecula Ecology,32(6), 1441–1457. h ps: //doi.o g/10.1111/mec.16793 Singhal, S., Le le , E. M., Sanna eddy, K., Tu ne , I., Venn, O., Hoope , D. M., S and, A. I., Li, Q., Raney, B., Balak ishnan, C. N., G i i h, S. C., McVean, G., & P zewo ski, M. (2015). S able ecombina ion ho spo s in bi ds. Science,350(6263), 928–932. h ps://doi.o g/10.1126/science.aad0843 Smeds, L., Q a ns öm, A., & Elleg en, H. (2016). Di ec es ima e o he a e o ge mline mu a ion in a bi d. Genome Resea ch,26(9), 1211–1218. h ps://doi.o g/10.1101/g .2 04669.116 Smukowski Heil, C. S., Ellison, C., Dubin, M., & Noo , M. A. F. (2015). Recombining wi hou Ho spo s: A Comp ehensi e E olu iona y Po ai o Recombina ion in Two Closely Rela ed Species o D osophila. Genome Biology and E olu ion,7(10), 2829–2842. h ps://doi.o g/10.1093/gbe/e 182 Speidel, L., Fo es , M., Shi, S., & Mye s, S. R. (2019). A me hod o genome-wide genealogy es ima ion o housands o samples. Na u e Gene ics,51(9), 1321–1329. h ps://doi.o g/10.1038/s41588-019-0484-x 85 Spence, J. P., & Song, Y. S. (2019). In e ence and analysis o popula ion-speci ic ine-scale ecombina ion maps ac oss 26 di e se human popula ions. Science Ad ances,5(10), eaaw9206. h ps://doi.o g/10.1126/sciad .aaw9206 S e n, A. J., Wil on, P. R., & Nielsen, R. (2019). An app oxima e ull-likelihood me hod o in e ing selec ion and allele equency ajec o ies om DNA sequence da a. PLoS Gene ics,15(9), 1–32. h ps://doi.o g/10.1371/jou nal.pgen.1008384 S e ison, L. S., Hoehn, K. B., & Noo , M. A. F. (2011). E ec s o In e sions on Wi hin- and Be ween-Species Recombina ion and Di e gence. Genome Biology and E olu ion, 3, 830–841. h ps://doi.o g/10.1093/gbe/e 081 S e ison, L. S., Woe ne , A. E., Kidd, J. M., Kelley, J. L., Vee amah, K. R., McManus, K. F., G ea Ape Genome P ojec , Bus aman e, C. D., Hamme , M. F., & Wall, J. D. (2016). The Time Scale o Recombina ion Ra e E olu ion in G ea Apes. Molecula Biology and E olu ion,33(4), 928–945. h ps://doi.o g/10.1093/molbe /ms 331 Tajima, F. (1989). S a is ical me hod o es ing he neu al mu a ion hypo hesis by DNA polymo phism. Gene ics,123(3), 585–595. h ps://doi.o g/PMC1203831 Taylo , J. E. (2013). The e ec o luc ua ing selec ion on he genealogy a a linked si e. Theo e ical Popula ion Biology,87, 34–50. h ps://doi.o g/10.1016/j. pb.2013.03.004 Todesco, M., Owens, G. L., Be co ich, N., Léga é, J.-S., Soudi, S., Bu ge, D. O., Huang, K., Os e ik, K. L., D ummond, E. B. M., Ime o ski, I., Lande, K., Pascual-Robles, M. A., Nana a i, M., Jahani, M., Cheung, W., S a on, S. E., Muños, S., Nielsen, R., Dono an, L. A., e al. (2020). Massi e haplo ypes unde lie eco ypic di e en ia ion in sun lowe s. Na u e,584(7822), 602–607. h ps://doi.o g/10.1038/s41586-020-2467-6 Van Do en, B. M., Campagna, L., Helm, B., Ille a, J. C., Lo e e, I. J., & Lied ogel, M. (2017). Co ela ed pa e ns o gene ic di e si y and di e en ia ion ac oss an a ian amily. Molecula Ecology,26(15), 3982–3997. h ps://doi.o g/10.1111/mec.14083 Voigh , B. F., Kuda a alli, S., Wen, X., & P i cha d, J. K. (2006). A Map o Recen Posi i e Selec ion in he Human Genome. PLOS Biology,4(3), e72. h ps://doi.o g/10 .1371/jou nal.pbio.0040072 Wakeley, J. (2008). Coalescen Theo y: An In oduc ion (1s ed.). W. H. F eeman. Wakeley, J. (2020). De elopmen s in coalescen heo y om single loci o ch omosomes. Theo e ical Popula ion Biology,133, 56–64. h ps://doi.o g/10.1016/j. pb.2020.02.002 Wellen eu he , M., & Be na chez, L. (2018). Eco-E olu iona y Genomics o Ch omosomal In e sions. T ends in Ecology & E olu ion,33(6), 427–440. h ps://doi.o g/10.1016/j. ee.2018.04.002 Wiu , C., & Hein, J. (1999). Recombina ion as a Poin P ocess along Sequences. Theo e ical Popula ion Biology,55(3), 248–259. h ps://doi.o g/10.1006/ pbi.1998.1403 Wol , J. B. W., & Elleg en, H. (2017). Making sense o genomic islands o di e en ia ion in ligh o specia ion. Na u e Re iews Gene ics,18(2), 87–100. h ps://doi.o g/10.1038/ n g.2016.133 86 Yi, X., Liang, Y., Hue a-Sanchez, E., Jin, X., Cuo, Z. X. P., Pool, J. E., Xu, X., Jiang, H., Vinckenbosch, N., Ko neliussen, T. S., Zheng, H., Liu, T., He, W., Li, K., Luo, R., Nie, X., Wu, H., Zhao, M., Cao, H., e al. (2010). Sequencing o 50 Human Exomes Re eals Adap a ion o High Al i ude. Science,329(5987), 75–78. h ps://doi.o g/10.1126/science.1190371 87 Supplemen a y In o ma ion Dis inc pa e ns o gene ic a ia ion a low- ecombining genomic egions ep esen haplo ype s uc u e Jun Ishigohoka1,* Ka en Bascón-Ca dozo1And ea Bou s1Janina Fuß2 A ang Rhie3Jacquelyn Moun cas le4Be ina Haase4William Chow5 Joanna Collins5Ke s in Howe5Ma cela Uliano-Sil a5Oli ie Fed igo4 E ich D. Ja is4,6,7 Ja ie Pé ez-T is8Juan Ca los Ille a9 Mi iam Lied ogel1,10,* 1Max Planck Ins i u e o E olu iona y Biology, Plön, Ge many 2Ins i u e o Clinical Molecula Biology (IKMB), Kiel Uni e si y, Kiel, Ge many 3 Genome In o ma ics Sec ion, Compu a ional and S a is ical Genomics B anch, Na ional Human Genome Resea ch Ins i u e, Na ional Ins i u es o Heal h, Be hesda, MD, USA 4The Ve eb a e Genome Lab, Rocke elle Uni e si y, New Yo k, NY, USA 5Wellcome Sange Ins i u e, Camb idge, UK 6Labo a o y o Neu ogene ics o Language, Rocke elle Uni e si y, New Yo k, NY, USA 7The Howa ds Hughes Medical Ins i u e, Che y Chase, MD, USA 8 Depa men o Biodi e si y, Ecology and E olu ion, Complu ense Uni e si y o Mad id, Mad id, Spain 9Resea ch Uni o Biodi e si y (UO-CSIC-PA), O iedo Uni e si y, Mie es, Spain 10Ins i u e o A ian Resea ch, Wilhelmsha en, Ge many * Co espondence: Jun Ishigohoka <ishigohoka@e olbio.mpg.de>,Mi iam Lied ogel <lied o- gel@e olbio.mpg.de> 88 Con en s 1 Supplemen a y No es 1.1 Discussion: Con ibu ion o genealogical and mu a ional noise on local gene ic a ia ion ........................................ 1.2 Discussion: E ec s o selec ion on dis inc gene ic a ia ion ........... 1.3 Resul s and discussion: Sp ead o indi iduals o low- ecombining popula ions in local PCA ....................................... 2 Supplemen a y Tables 3 Supplemen a y Figu es 3.1 Whole-genome PCA ................................. 3.2 Local PCA and ecombina ion map ......................... 3.3 Pu a i e in e sions .................................. 3.3.1 PCA ...................................... 3.3.2 LD and geno ype-speci ic ecombina ion map ............... 3.3.3 B eakpoin analysis ............................. 3.4 E ec o educed ecombina ion a e on pa e n o local gene ic a ia ion ... 3.4.1 E ec o demog aphy and local ecombina ion a e (coalescen simula ion) 3.4.2 Species-wide educ ion o local ecombina ion a e ( o wa d simula ion) 3.4.3 Popula ion-speci ic educ ion o local ecombina ion a e ( o wa d simu- la ion) ..................................... 3.5 E ec o selec ion ................................... 3.5.1 E ec s o selec ion in blackcap genome ................... 3.5.2 E ec s o selec ion a low- ecombining egions ( o wa d simula ion) ... 3.6 Pe icen ome ic egions in ou lie egions ..................... 3.7 Genealogical in e p e a ion ............................. 4 Re e ences 89 3 Supplemen a y Figu es 3.1 Whole-genome PCA −0.05 0.05 0.15 −0.30 −0.15 0.00 PC1 (3.3 %) PC2 (2.3 %) −0.05 0.05 0.15 −0.2 0.0 0.1 PC1 (3.3 %) PC3 (1.8 %) −0.05 0.05 0.15 −0.2 0.0 0.1 PC1 (3.3 %) PC4 (1.7 %) −0.05 0.05 0.15 −0.2 0.0 0.2 PC1 (3.3 %) PC5 (1.6 %) −0.30 −0.15 0.00 −0.2 0.0 0.1 PC2 (2.3 %) PC3 (1.8 %) −0.30 −0.15 0.00 −0.2 0.0 0.1 PC2 (2.3 %) PC4 (1.7 %) −0.30 −0.15 0.00 −0.2 0.0 0.2 PC2 (2.3 %) PC5 (1.6 %) −0.2 0.0 0.1 −0.2 0.0 0.1 PC3 (1.8 %) PC4 (1.7 %) −0.2 0.0 0.1 −0.2 0.0 0.2 PC3 (1.8 %) PC5 (1.6 %) −0.2 0.0 0.1 −0.2 0.0 0.2 PC4 (1.7 %) PC5 (1.6 %) con _medlong con _sho con _ esiden Cana y Madei a Azo es CapeVe de Mallo ca C e e Supplemen a y Figu e 1: Popula ion s uc u e analysed wi h whole-genome PCA. Supple- men a y da a ela ed o Fig. 2B, C. The i s i e PCs a e shown. 96 3.2 Local PCA and ecombina ion map Supplemen a y Figu e 2: Dis ibu ion o MDS alues o windows o a ch omosome in los uc .Each panel shows he dis ibu ion o MDS alues o windows in an exempli ied ch omosome (ch omosoem 1) in local PCA using los uc based on geno ypes. Red lines show median o MDS alues. Genomic egions wi h a leas 10 windows wi h MDS alues away om he median by 0.3 (blue lines) we e de ined as “ou lie s” (see Ma e ials and Me hods). 97 1 2 3 Z 4 5 6 7 8 9 10 11 12 13 Me ged 20 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 20 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 Ch omosome MDS axis Supplemen a y Figu e 3: Genomic egions wi h dis inc pa e ns o gene ic a ia ion iden i ied wi h los uc .Supplemen a y con en ela ed o Fig. 2D. The y-axis shows he MDS axes along which p ojec ed componen o gene ic a ia ion is de ia ed in geno ype-based ( op hal , cyan) and haplo ype-based (bo om hal , pu ple) analysis o los uc . The x-axis shows he genomic posi ion. The bo om ow (colou s co enspond o species-wide, popula ion-speci ic, and no low- ecombining in Fig. 2D) shows coo dina es o ou lie in e als which we de e mined as inal ou lie s by me ging esul s o he geno ype- and haplo ype-based analyses. 98 Supplemen a y Figu e 4: Consis ency o geno ype- and haplo ype-based los uc .To answe whe he geno ype- and haplo ype-based local PCA using los uc we e consis en , Euclidean dis ance o windows in he 20 dimensional space was compa ed be ween hese app oaches. Red poin s depic windows wi h de ia ed MDS alue along a MDS axis in a leas one o geno ype- o haplo ype- based analyses. 99 Supplemen a y Figu e 5: Robus ness o los uc .To add ess whe he los uc is obus o he ch omosomal backg ound and he leng h o he ch omosome, we pe o med los uc using a i icially syn hesised ch omosomes. Blackcap ch omosomes 1 and 2 we e spli in o hal es in he middle, and ch omosomes 20, 21, 28 we e joined in o a single ch omosome. los uc was pe o med o hese spli /joined ch omosomes and esul s we e compa ed wi h single ch omosome analysis. A, B, E-G. Summa y o los uc analysis based on a single ch omosome (black) and spli /joined ch omosome ( ed). C, D, H, I. MDS o single ch omosome (C, D) and joined ch omosome (H,I)los uc . Red poin s depic windows wi h MDS alue beyond he h eshold. Pe -ch omosome analysis was mos ly consis en wi h spli /joined ch omosome analysis. In egions wi h inconsis en esul s, MDS alues s ill show sub- h eshold de ia ion. 100 ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ●●● ● ● ● ●● ●●● ● ● ●● ● ●● ● ● ● ● ● ● ●●● ●● ● ● ● ●● ● ● ● ● ●● −0.3 −0.2 −0.1 0.0 −0.3 −0.2 −0.1 0.0 0.1 PC1 9.5 % PC2 7.1 % ch _1 56264813 − 57679298 1.41 Mb ● ●● ●●● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ●● ● ●● ● ● ● ●● ● ● ● ● ● ● ●● ●● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ●●● ● ● ●●● ● ● ● ●● ● ●● ● ● ●● ●●● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ●● ● ● ● −0.15 −0.05 0.00 0.05 0.10 −0.3 −0.2 −0.1 0.0 0.1 0.2 PC1 10.5 % PC2 3.9 % ch _1 76931638 − 77195852 0.26 Mb ● ●● ● ●● ●● ● ● ● ●●● ●● ● ● ● ● ● ●● ● ● ● ● ● ●● ● ● ● ● ●● ●●● ● ● ● ● ● ● ●●● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●●● ●● ● ●●● ●●●● ● ●● ● ● ●● ● ●● ●● ● ● ●● ●● ● ● ●● ●● ● ● ● ● ● −0.15 −0.05 0.05 0.10 0.15 −0.4 −0.3 −0.2 −0.1 0.0 PC1 8.8 % PC2 6.7 % ch _2 17876149 − 18183386 0.31 Mb ●● ● ●● ●● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●● ●● ● ● ●● ●● ●● ● ●● ● ●● ● ●● ●● ● ● ●●● ●● ● ●● ● ● ● ● ● ● ● ●● ● ●● ● ● ●● ● ●● ● ●● ● ● ●● ● ● ● ● ●● ●● ● ● ●● ● ● ● ● ● ● −0.3 −0.2 −0.1 0.0 −0.4 −0.2 0.0 0.1 0.2 PC1 16.2 % PC2 10.3 % ch _2 18301735 − 19824380 1.52 Mb ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ●● ●● ●● ● ● ● ● ●● ●●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● −0.1 0.0 0.1 0.2 0.3 0.4 −0.1 0.0 0.1 0.2 PC1 5.1 % PC2 4.2 % ch _Z 14585806 − 14978909 0.39 Mb ● ● ● ● ● ● ●●● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●●●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ●● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● −0.3 −0.2 −0.1 0.0 −0.4 −0.2 0.0 0.2 0.4 PC1 15.1 % PC2 8.1 % ch _Z 23428532 − 24628650 1.2 Mb ● ●●● ●●● ● ● ●● ● ● ●●● ● ● ●● ● ●●● ● ●● ●● ● ● ● ● ● ● ● ●●● ● ●● ●● ● ●● ● ●● ●● ●● ●● ●● ●● ● ● ●● ●● ● ● ●●● ● ● ●●● ● ●● ●● ●● ●● ● ● ● ● ● ● ● ● ●● ●●● ●● ● ●● ●● −0.20 −0.10 0.00 0.05 0.10 0.0 0.1 0.2 0.3 0.4 0.5 PC1 10.6 % PC2 7.8 % ch _Z 47230132 − 47412498 0.18 Mb ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −0.4 −0.3 −0.2 −0.1 0.0 −0.2 −0.1 0.0 0.1 PC1 10.8 % PC2 8.5 % ch _5 68065278 − 68294444 0.23 Mb ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −0.6 −0.5 −0.4 −0.3 −0.2 −0.1 0.0 −0.2 −0.1 0.0 0.1 PC1 14.5 % PC2 6.7 % ch _6 5687266 − 6323968 0.64 Mb ● ● ● ● ●● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ●●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● 0.0 0.1 0.2 0.3 0.4 0.5 0.6 −0.2 0.0 0.2 0.4 PC1 12.2 % PC2 11.5 % ch _8 30282658 − 30636429 0.35 Mb ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ●● ● ● ● ● −0.15 −0.05 0.00 0.05 0.10 −0.20 −0.10 0.00 0.10 PC1 14.5 % PC2 12.1 % ch _10 11602449 − 13202422 1.6 Mb ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●● ●● ●●● ●● ● ● ●● ● ● ● ● ● ●● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −0.15 −0.05 0.00 0.05 0.10 −0.4 −0.3 −0.2 −0.1 0.0 0.1 PC1 17.8 % PC2 8.8 % ch _12 60 − 208254 0.21 Mb ● ● ●● ● ● ● ● ● ● ● ●● ● ● ●● ● ●● ●● ● ● ● ● ● ●● ● ● ● ● ● ●● ●● ● ● ● ● ●● ●● ● ● ●● ● ● ● ●●● ●● ●● ● ● ● ● ● ●● ● ● ● ●● ● ● ●● ● ● ●● ● ● ●● ● ●● ● ● ●● ● ●● ● ● ●● ●● ●● ●● ●● ●● ● ● ● ●● −0.05 0.00 0.05 0.10 0.15 0.20 −0.2 −0.1 0.0 0.1 0.2 PC1 21.7 % PC2 2.6 % ch _12 14118029 − 22229395 8.11 Mb ● ● ●●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ●● ● ●● ● ● ● ● ● ●● ● ● ●● ● ● ● ●● ● ● ● ● ● ●● ●● ●●● ●● ●● ● ● ●● ● ●● ● ● ● ● ● ● ●●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ●● ● ●●● ●● ● ● ● ● ● ● ●● −0.15 −0.05 0.00 0.05 0.10 −0.3 −0.2 −0.1 0.0 0.1 PC1 10.4 % PC2 8.0 % ch _14 42 − 207189 0.21 Mb ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ●● ● ●● ● ● ● ● ● ● −0.3 −0.2 −0.1 0.0 0.1 −0.2 −0.1 0.0 0.1 PC1 12.6 % PC2 11.0 % ch _15 15846845 − 16049116 0.2 Mb ●● ● ●●● ● ●● ●●● ● ● ● ● ● ●● ● ●● ● ● ●● ● ●● ●●● ● ● ● ● ● ● ●●● ● ●● ●●●● ● ● ●● ● ●● ● ●●● ● ● ● ●● ● ● ● ● ● ●●● ● ● ●●● ●● ● ● ●● ● ● ● ●● ●● ● ●● ●● ●●●● ● ● ● ● ● ● ● ● ● ● ● ● ●● −0.20 −0.10 0.00 0.05 0.10 −0.1 0.0 0.1 0.2 0.3 0.4 PC1 8.7 % PC2 7.5 % ch _16 1510565 − 1842318 0.33 Mb ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ●●● ● ● ●● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● −0.20 −0.10 0.00 0.05 −0.15 −0.05 0.05 0.15 PC1 18.1 % PC2 10.9 % ch _17 13936616 − 14179092 0.24 Mb ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● −0.25 −0.15 −0.05 0.05 −0.2 −0.1 0.0 0.1 0.2 PC1 19.2 % PC2 10.6 % ch _20 25273 − 338180 0.31 Mb ●● ●● ● ●● ● ● ●● ●● ●● ●● ● ● ●● ●● ● ●● ● ●● ● ● ● ●● ● ● ●● ● ● ●● ● ● ●● ● ● ● ● ● ●● ● ● ● ● ● ●● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ●● ● ● ● ●● ● ● ●● ● ●● ● ● ●● ● ● ● ●● ● ● ●● ●● ● ●● ●● ● −0.20 −0.10 0.00 0.05 −0.3 −0.2 −0.1 0.0 PC1 9.8 % PC2 3.1 % ch _30 71 − 1471845 1.47 Mb ● ● ● con _medlong con _sho con _ esiden Cana y Madei a Azo es CapeVe de Mallo ca C e e Supplemen a y Figu e 6: PCA o ou lie s o e lapping species-wide low- ecombining egions. Supplemen a y da a ela ed o Fig. 3. PCA plo s ep esen pa e ns o gene ic a ia ion a 19 ou lie egions in he blackcap genome o e lapping species-wide low- ecombining egions (da a poin s ep esen blackcap indi iduals and colou s depic popula ions). The pa e ns we e dis inc om popula ion s uc u e (Fig. 2B, C). 101 ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ●● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● −0.3 −0.2 −0.1 0.0 −0.4 −0.2 0.0 0.2 PC1 4.2 % PC2 2.2 % ch _2 114618117 − 114874768 CapeVe de 0.26 Mb ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ●● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● −0.3 −0.2 −0.1 0.0 −0.4 −0.2 0.0 0.1 0.2 PC1 5.4 % PC2 2.8 % ch _3 108194891 − 108406259 Azo es 0.21 Mb ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −0.30 −0.20 −0.10 0.00 −0.2 0.0 0.2 0.4 PC1 22.0 % PC2 2.4 % ch _4 11647462 − 11966907 med_sw 0.32 Mb ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● −0.3 −0.2 −0.1 0.0 −0.6 −0.4 −0.2 0.0 0.2 PC1 4.7 % PC2 2.1 % ch _12 1850449 − 2310703 CapeVe de 0.46 Mb ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● −0.4 −0.3 −0.2 −0.1 0.0 −0.3 −0.2 −0.1 0.0 0.1 0.2 PC1 4.5 % PC2 2.7 % ch _14 14789429 − 15024671 Azo es;CapeVe de 0.24 Mb ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● −0.4 −0.3 −0.2 −0.1 0.0 −0.4 −0.2 0.0 0.1 0.2 PC1 4.8 % PC2 2.4 % ch _14 15956521 − 16206154 Azo es;CapeVe de 0.25 Mb ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −0.3 −0.2 −0.1 0.0 −0.2 0.0 0.1 0.2 0.3 0.4 0.5 PC1 4.9 % PC2 2.3 % ch _15 13458326 − 14146280 Azo es;CapeVe de 0.69 Mb ●● ●● ● ● ● ● ● ●● ● ● ● ● ● ●● ●● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ●● ●● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ●●● ● ●● ● ●● −0.1 0.0 0.1 0.2 0.3 −0.4 −0.2 0.0 0.2 PC1 4.6 % PC2 3.0 % ch _20 1499744 − 1627522 Azo es;CapeVe de 0.13 Mb ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0.0 0.1 0.2 0.3 0.4 −0.4 −0.3 −0.2 −0.1 0.0 0.1 PC1 4.9 % PC2 2.7 % ch _21 611082 − 757333 Azo es 0.15 Mb ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● −0.3 −0.2 −0.1 0.0 −0.6 −0.4 −0.2 0.0 0.2 PC1 5.6 % PC2 1.9 % ch _21 3003123 − 3440115 Azo es;CapeVe de 0.44 Mb ● ● ●●● ● ● ● ●● ● ● ● ●●● ● ●●●● ● ●● ● ● ● ●● ● ● ●● ● ● ●● ● ● ● ●● ● ● ●● ●● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ●● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ●● −0.05 0.05 0.15 0.25 −0.4 −0.3 −0.2 −0.1 0.0 0.1 PC1 6.3 % PC2 2.9 % ch _28 939875 − 1143638 Azo es;CapeVe de 0.2 Mb ● ● ● con _medlong con _sho con _ esiden Cana y Madei a Azo es CapeVe de Mallo ca C e e Supplemen a y Figu e 7: PCA o ou lie s o e lapping only wi h popula ion-speci ic low- ecombining egions. Supplemen a y da a ela ed o Fig. 3. PCA plo s ep esen pa e ns o gene ic a ia ion a 11 ou lie egions in he blackcap genome o e lapping only wi h popula ion-speci ic low- ecombining egions (da a poin s ep esen blackcap indi iduals and colou s depic popula ions). The pa e ns we e dis inc om popula ion s uc u e (Fig. 2B, C). Popula ion labels on op o each panel depic he popula ion(s) in which low- ecombining egions a e ound wi hin he ou lie . ● ● ● ● ● ●● ● ● ●● ● ● ●● ● ● ●● ●● ● ● ● ●● ● ● ● ● ● ● ● ●● ●● ● ● ●● ● ●● ● ●● ● ● ●● ● ● ●● ● ● ● ●● ●●● ● ●● ● ●● ●● ● ● ● ● ●● ● ● ● ●● ●●● ● ● ● ● ● ●● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● −0.15 −0.05 0.05 0.15 −0.1 0.0 0.1 0.2 0.3 PC1 5.4 % PC2 3.1 % ch _3 1317746 − 1464754 0.15 Mb ● ● ● ● ● ● ●● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ●● ● 0.0 0.1 0.2 0.3 −0.2 −0.1 0.0 0.1 0.2 0.3 0.4 PC1 5.6 % PC2 3.0 % ch _6 62340070 − 62456329 0.12 Mb ● ● ● con _medlong con _sho con _ esiden Cana y Madei a Azo es CapeVe de Mallo ca C e e Supplemen a y Figu e 8: PCA o ou lie egions wi hou o e laps wi h low- ecombining egions. Supplemen a y da a ela ed o Fig. 3. PCA plo s ep esen pa e ns o gene ic a ia ion a wo ou lie egions in he blackcap genome (da a poin s ep esen blackcap indi iduals and colou s depic popula ions). 102 0 50 100 150 0 10 20 30 40 50 ch _1 0 20 40 60 80 100 0 10 20 30 40 50 ch _2 0 20 40 60 80 100 0 10 20 30 40 50 ch _3 0 20 40 60 80 0 10 20 30 40 50 ch _Z 0 20 40 60 0 10 20 30 40 50 ch _4 0 20 40 60 0 10 20 30 40 50 ch _5 0 10 20 30 40 50 60 0 10 20 30 40 50 ch _6 0 10 20 30 40 0 10 20 30 40 50 ch _7 0 5 10 15 20 25 30 35 0 10 20 30 40 50 ch _8 0 5 10 15 20 25 30 0 10 20 30 40 50 ch _9 0 5 10 15 20 25 0 10 20 30 40 50 ch _10 0 5 10 15 20 0 10 20 30 40 50 ch _11 0 5 10 15 20 0 10 20 30 40 50 ch _12 0 5 10 15 20 0 10 20 30 40 50 ch _13 0 5 10 15 0 10 20 30 40 50 ch _14 0 5 10 15 0 10 20 30 40 50 ch _15 0 5 10 15 0 10 20 30 40 50 ch _16 0 2 4 6 8 10 12 14 0 10 20 30 40 50 ch _17 0 2 4 6 8 10 12 0 10 20 30 40 50 ch _18 0 2 4 6 8 10 0 10 20 30 40 50 ch _19 0 2 4 6 8 10 0 10 20 30 40 50 ch _20 0 2 4 6 8 10 0 10 20 30 40 50 ch _21 0 2 4 6 0 10 20 30 40 50 ch _22 0246 0 10 20 30 40 50 ch _23 01234567 0 10 20 30 40 50 ch _24 01234567 0 10 20 30 40 50 ch _25 0 1 2 3 4 5 0 10 20 30 40 50 ch _26 012345 0 10 20 30 40 50 ch _27 0 1 2 3 4 0 10 20 30 40 50 ch _28 0.0 0.5 1.0 1.5 2.0 0 10 20 30 40 50 ch _29 0.0 0.5 1.0 1.5 2.0 0 10 20 30 40 50 ch _30 0.0 0.1 0.2 0.3 0.4 0.5 0 10 20 30 40 50 ch _31 0.0 0.1 0.2 0.3 0.4 0 10 20 30 40 50 ch _32 0.00 0.10 0.20 0.30 0 10 20 30 40 50 ch _33 med_sw con _ es Azo es CapeVe de Species−wide Pop−speci ic No low− ec. Posi ion [Mb] Recombina ion a e [cM/Mb] Supplemen a y Figu e 9: Recombina ion landscape and los uc ou lie s. Supplemen a y da a ela ed o Fig. 2E, F. Fou lines depic ecombina ion maps in e ed o ou blackcap popula ions. Backg ound shades depic posi ions o ou lie s iden i ied by local PCA using los uc . 103 con _medlong O e laps wi h low− ec egions [Mb] F equency 0 2 4 6 8 10 12 0 20 40 60 80 120 p = 0.000 con _ es O e laps wi h low− ec egions [Mb] F equency 0 2 4 6 8 10 12 0 20 60 100 140 p = 0.000 Azo es O e laps wi h low− ec egions [Mb] F equency 0 2 4 6 8 10 12 0 50 100 150 p = 0.000 CapeVe de O e laps wi h low− ec egions [Mb] F equency 0 2 4 6 8 10 12 0 50 100 150 p = 0.000 Supplemen a y Figu e 10: Pe mu a ion es s o he numbe o o e laps be ween local PCA ou lie s and low- ecombining egions. To es whe he he 32 ou lie egions in he blackcap genome based on local PCA signi ican ly o e lap wi h low- ecombining egions, pe mu a ion es s we e pe o med (n = 1,000). In each pe mu a ion, in e als o obse ed ou lie egions we e shu led in each ch omosome, and he o al leng h o o e laps wi h low- ecombining egions (below 20 pe cen ile pe ch omosome) was eco ded (see Ma e ials and Me hods). Red lines ep esen obse ed leng h o o e laps. 104 3.3 Pu a i e in e sions 3.3.1 PCA −0.6 −0.4 −0.2 0.0 −0.2 −0.1 0.0 0.1 PC1 (14.5%) PC2 (6.7%) ou lie _6_1 A −0.15 −0.05 0.05 −0.3 −0.1 0.1 PC1 (10.4%) PC2 (8%) ou lie _14_1 B −0.05 0.05 0.15 0.25 −0.4 −0.2 0.0 PC1 (6.3%) PC2 (2.9%) ou lie _28_1 C −0.20 −0.10 0.00 −0.3 −0.2 −0.1 0.0 PC1 (9.8%) PC2 (3.1%) ou lie _30_1 D con _medlong con _sho con _ esiden Cana y Madei a Azo es CapeVe de Mallo ca C e e AA AA AA AA AB AB AB AB BB BB BB BB Supplemen a y Figu e 11: PCA a ou lie s wi h h ee clus e s o indi iduals. Supplemen a y da a ela ed o Fig. 3A ( op). Fi e ou lie egions ound in he blackcap genome show pa e ns o gene ic a ia ion in PCA wi h h ee clus e s o indi iduals, some o which may ep esen polymo phic in e sions (Huang e al., 2020; Ma & Amos, 2012; Todesco e al., 2020). We in es iga ed his possibili y in Sup. Figs. 12,13,14. In each egion, we named he majo and mino alleles A and B, and he h ee geno ypes AA, AB, and BB. 105 AAF1880_279 ch _1 ch _2 ch _3 ch _Z ch _4 ch _5 ch _6 ch _7 ch _8 ch _9 ch _10 ch _11 ch _12 ch _W ch _13 ch _14 ch _15 ch _16 ch _17 ch _18 ch _19 ch _20 ch _21 ch _22 ch _23 ch _24 ch _25 ch _26 ch _27 ch _28 ch _29 ch _30 ch _31 ch _32 ch _w_unlocalised ch _33 sca old_018 sca old_019 sca old_022 sca old_025 sca old_044 sca old_067 sca old_089 N. copies 0 20 40 60 80 100 120 140 Supplemen a y Figu e 18: Tandem epea s ound a pu a i e in e sion b eakpoin o ch o- mosome 12 a e absen in ances al allele bu p esen in o he ch omosomes. Supplemen a y da a ela ed o Sup. Fig. 17. A 144 bp-long andem epea was iden i ied in lanking sequence o he 10x con igs aligned nex o he pu a i e in e sion b eakpoin o ou lie _12_3 (De ailed in Ma e ials and Me hods). To in es iga e whe he his epea is p esen elsewhe e in he pseudohaplo ype o A allele and he e e ence, BLASTn was pe o med wi h he consensus sequence o he andem epea uni as he que y and con ig 279 o AAF1880 (A haplo ype con ig o 10x) and he blackcap e e ence as a ge . The ba plo depic s he numbe o hi s in each ch omosome/con ig. The esul shows he andem epea is p esen in o he ch omosomes bu no on he A allele o ch omosome 12. 112 3.4 E ec o educed ecombina ion a e on pa e n o local gene ic a ia ion 3.4.1 E ec o demog aphy and local ecombina ion a e (coalescen simula ion) Model 1 Uni o m ec. 106 100 Model 2 Uni o m ec. 106 106106106106106 50 20 10 10 10 Model 3 Uni o m ec. 106 106106106106106 50 20 10 10 10 Model 4 Low- ec. 106 106106106106106 50 20 10 10 10 Model 5 No- ec. 106 106106106106106 50 20 10 10 10 106 106106106106106 50 20 10 10 10 Model 6 Low- ec. 106 106106106106106 50 20 10 10 10 Model 7 No- ec. 106 107106105105105 50 20 10 10 10 Model 8 Low- ec. 106 107106105105105 50 20 10 10 10 Model 9 Low- ec. 106 107106105105105 50 20 10 10 10 Model 10 No- ec. 106 107106105105105 50 20 10 10 10 Model 11 No- ec. A 0.01 0.00 B Supplemen a y Figu e 19: Demog aphy models and ecombina ion maps o neu al coalescen simula ions. Supplemen a y da a ela ed o Sup. Table. 8. To in es iga e he e ec s o local ecombina ion a e and demog aphy on gene ic a ia ion, we implemen ed 11 scena ios o demog aphic his o y and ecombina ion landscapes ([ ab:sup.msp_models]) and simula ed SNPs unde coalescen wi h ecombina ion wi h msp ime 1,000 imes. A. 11 models o demog aphic his o y. The numbe s shown on he popula ions depic he e ec i e popula ion size, and he numbe s below he popula ions depic he numbe s o sampled diploid indi iduals. No e ha demog aphies o models 2, 4 and 5, models 3, 6 and 7, models 8 and 9, and models 10 and 11 a e espec i ely he same, di e ing by he ecombina ion map (B). B. Th ee ecombina ion maps di e ing by he ecombina ion a e in he middle o he ch omosome. Wi hin he middle in e al (0.4 o 0.6 Mb), ecombina ion a e is educe o 1/100 o he backg ound egion in he “low- ecombining” scena ios, while in he “no- ecombining” scena ios ecombina ion a e is educed o 0. 113 0.0 0.2 0.4 0.6 0.8 1.0 0 200 400 600 800 1000 Model 1 Uni . ec. a e, no pop. s uc. 0.0 0.2 0.4 0.6 0.8 1.0 0 200 400 600 800 1000 Model 2 Uni . ec. a e, pop. s uc., no gene low. 0.0 0.2 0.4 0.6 0.8 1.0 0 200 400 600 800 1000 Model 3 Uni . ec. a e, pop. s uc., gene low. 0.0 0.2 0.4 0.6 0.8 1.0 0 200 400 600 800 1000 Model 4 Low ec. egion, pop. s uc., no gene low 0.0 0.2 0.4 0.6 0.8 1.0 0 200 400 600 800 1000 Model 5 Non− ec. egion, pop. s uc., no gene low 0.0 0.2 0.4 0.6 0.8 1.0 0 200 400 600 800 1000 Model 6 Low ec. egion, pop. s uc., gene low 0.0 0.2 0.4 0.6 0.8 1.0 0 200 400 600 800 1000 Model 7 Non− ec. egion, pop. s uc., gene low 0.0 0.2 0.4 0.6 0.8 1.0 0 200 400 600 800 1000 Model 8 Low ec. egion, pop. s uc., di . dem., no gene low. 0.0 0.2 0.4 0.6 0.8 1.0 0 200 400 600 800 1000 Model 9 Low ec. egion, pop. s uc., di . dem., gene low. 0.0 0.2 0.4 0.6 0.8 1.0 0 200 400 600 800 1000 Model 10 Non− ec. egion, pop. s uc., di . dem., no gene low. 0.0 0.2 0.4 0.6 0.8 1.0 0 200 400 600 800 1000 Model 11 Non− ec. egion, pop. s uc., di . dem., gene low. Local PCA ou lie in e al Posi ion [Mb] Simula ion eplica es Supplemen a y Figu e 20: Reduced ecombina ion a e, bu no demog aphy, causes dis inc pa e ns o gene ic a ia ion. Supplemen a y da a ela ed o Sup. Fig. 19 and Sup. Table. 8. To iden i y ou lie egions, los uc was pe o med o each o 1,000 eplica es o he 11 scena ios in Sup. Fig. 19. The esul s show ha no ou lie egions we e de ec ed wi hou educed ecombina ion a e (models 1-3), and ou lie s we e always de ec ed wi h educed ecombina ion a es (models 4-11). This is i espec i e o he p esence o popula ion s uc u e, unequal demog aphic his o y, and unbalanced sample size among popula ions. These esul s indica e ha educed ecombina ion a e, bu no demog aphy, causes dis inc pa e ns o gene ic a ia ion. 114 0.0 0.2 0.4 0.6 0.8 1.0 0.00 0.01 0.02 0.03 0.04 0.05 Model 4 0.0 0.2 0.4 0.6 0.8 1.0 0.00 0.01 0.02 0.03 0.04 0.05 Model 5 0.0 0.2 0.4 0.6 0.8 1.0 0.00 0.01 0.02 0.03 0.04 0.05 Model 6 0.0 0.2 0.4 0.6 0.8 1.0 0.00 0.01 0.02 0.03 0.04 0.05 Model 7 0.0 0.2 0.4 0.6 0.8 1.0 0.00 0.01 0.02 0.03 0.04 0.05 Model 8 0.0 0.2 0.4 0.6 0.8 1.0 0.00 0.01 0.02 0.03 0.04 0.05 Model 9 0.0 0.2 0.4 0.6 0.8 1.0 0.00 0.01 0.02 0.03 0.04 0.05 Model 10 0.0 0.2 0.4 0.6 0.8 1.0 0.00 0.01 0.02 0.03 0.04 0.05 Model 11 popA popB popC popD popE Posi ion [Mb] π Supplemen a y Figu e 21: Reduced ecombina ion a e inc eases a iance o summa y s a is ics. Supplemen a y da a ela ed o Sup. Fig. 19 and Sup. Table. 8. Using he same da a as Sup. Fig. 19, we asked how educed ecombina ion a es a ec mean and a iance o nucleo ide di e si y. Solid lines show he meean o he i s 100 eplica es (ou o 1,000 simula ed) and do ed lines show he s anda d de ia ion. 115 0.0 0.2 0.4 0.6 0.8 1.0 −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 Model 4 0.0 0.2 0.4 0.6 0.8 1.0 −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 Model 5 0.0 0.2 0.4 0.6 0.8 1.0 −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 Model 6 0.0 0.2 0.4 0.6 0.8 1.0 −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 Model 7 0.0 0.2 0.4 0.6 0.8 1.0 −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 Model 8 0.0 0.2 0.4 0.6 0.8 1.0 −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 Model 9 0.0 0.2 0.4 0.6 0.8 1.0 −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 Model 10 0.0 0.2 0.4 0.6 0.8 1.0 −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 Model 11 popA popB popC popD popE Posi ion [Mb] Tajima's D Supplemen a y Figu e 22: Reduced ecombina ion a e inc eases a iance o summa y s a is ics. Supplemen a y da a ela ed o Sup. Fig. 19 and Sup. Table. 8. Using he same da a as Sup. Fig. 19, we asked how educed ecombina ion a es a ec mean and a iance o Tajima’s D. Solid lines show he meean o he i s 100 eplica es (ou o 1,000 simula ed) and do ed lines show he s anda d de ia ion. 116 0.0 0.2 0.4 0.6 0.8 1.0 0.00 0.05 0.10 0.15 Model 4 0.0 0.2 0.4 0.6 0.8 1.0 0.00 0.05 0.10 0.15 Model 5 0.0 0.2 0.4 0.6 0.8 1.0 0.00 0.05 0.10 0.15 Model 6 0.0 0.2 0.4 0.6 0.8 1.0 0.00 0.05 0.10 0.15 Model 7 0.0 0.2 0.4 0.6 0.8 1.0 0.00 0.05 0.10 0.15 Model 8 0.0 0.2 0.4 0.6 0.8 1.0 0.00 0.05 0.10 0.15 Model 9 0.0 0.2 0.4 0.6 0.8 1.0 0.00 0.05 0.10 0.15 Model 10 0.0 0.2 0.4 0.6 0.8 1.0 0.00 0.05 0.10 0.15 Model 11 popA−popB popA−popC popA−popD popB−popC popB−popD popC−popD Posi ion [Mb] FST Supplemen a y Figu e 23: Reduced ecombina ion a e inc eases a iance o summa y s a is ics. Supplemen a y da a ela ed o Sup. Fig. 19 and Sup. Table. 8. Using he same da a as Sup. Fig. 19, we asked how educed ecombina ion a es a ec mean and a iance o FST be ween i e popula ion pai s (o all 10 pai s). Solid lines show he meean o he i s 100 eplica es (ou o 1,000 simula ed) and do ed lines show he s anda d de ia ion. 117 3.4.2 Species-wide educ ion o local ecombina ion a e ( o wa d simula ion) −0.25 −0.15 −0.05 0.05 −0.3 −0.1 0.1 sim 00 = 0 −0.25 −0.15 −0.05 0.05 −0.2 0.0 0.2 sim 00 = 100 −0.30 −0.20 −0.10 0.00 −0.1 0.1 0.2 0.3 sim 00 = 200 0.0 0.1 0.2 0.3 −0.6 −0.2 0.0 sim 00 = 300 −0.05 0.05 0.15 0.25 −0.3 −0.1 0.1 sim 00 = 400 0.0 0.1 0.2 0.3 −0.6 −0.4 −0.2 0.0 sim 00 = 500 −0.05 0.05 0.15 −0.2 0.0 0.1 sim 00 = 600 −0.10 0.00 0.10 −0.15 0.00 0.10 sim 00 = 700 −0.10 0.00 0.10 −0.10 0.05 0.15 sim 00 = 800 −0.10 0.00 0.10 0.20 −0.20 −0.05 0.05 sim 00 = 900 PC1 PC2 −0.3 −0.2 −0.1 0.0 −0.2 0.0 0.2 0.4 sim 01 = 0 −0.30 −0.20 −0.10 0.00 −0.15 0.00 0.10 sim 01 = 100 0.0 0.1 0.2 0.3 0.4 −0.20 −0.05 0.05 sim 01 = 200 −0.4 −0.2 0.0 −0.4 0.0 0.2 0.4 sim 01 = 300 0.0 0.1 0.2 0.3 −0.1 0.0 0.1 0.2 0.3 sim 01 = 400 −0.3 −0.2 −0.1 0.0 0.1 −0.10 0.00 0.10 0.20 sim 01 = 500 −0.15 −0.05 0.05 0.0 0.1 0.2 0.3 sim 01 = 600 −0.15 −0.05 0.05 0.0 0.1 0.2 0.3 sim 01 = 700 −0.05 0.05 0.15 0.25 −0.6 −0.2 0.0 sim 01 = 800 −0.05 0.05 0.15 −0.15 −0.05 0.05 sim 01 = 900 PC1 PC2 −0.15 −0.05 0.05 −0.10 0.05 0.15 0.25 sim 02 = 0 −0.15 −0.05 0.05 −0.10 0.05 0.15 sim 02 = 100 −0.10 0.00 0.10 0.20 −0.1 0.0 0.1 0.2 sim 02 = 200 −0.10 0.00 0.10 −0.1 0.1 0.2 0.3 sim 02 = 300 −0.20 −0.10 0.00 −0.3 −0.1 0.1 sim 02 = 400 −0.10 0.00 0.10 0.20 −0.3 −0.1 0.1 sim 02 = 500 −0.15 −0.05 0.05 −0.4 −0.2 0.0 sim 02 = 600 −0.10 0.00 0.05 0.10 −0.4 −0.2 0.0 sim 02 = 700 −0.10 0.00 0.10 −0.3 −0.1 sim 02 = 800 −0.10 0.00 0.10 −0.2 0.0 0.1 sim 02 = 900 PC1 PC2 −0.10 0.00 0.05 −0.10 0.05 0.15 sim 03 = 0 −0.05 0.05 0.15 −0.2 0.0 0.2 sim 03 = 100 0.0 0.1 0.2 0.3 −0.2 0.0 0.1 0.2 sim 03 = 200 −0.05 0.05 0.15 0.25 −0.1 0.1 0.3 sim 03 = 300 −0.05 0.05 0.15 −0.2 0.0 0.2 sim 03 = 400 −0.05 0.05 0.15 −0.2 0.0 0.2 sim 03 = 500 −0.20 −0.10 0.00 −0.10 0.00 0.10 0.20 sim 03 = 600 0.0 0.1 0.2 0.3 −0.05 0.05 0.15 sim 03 = 700 0.0 0.1 0.2 0.3 −0.2 0.0 0.1 0.2 sim 03 = 800 −0.8 −0.4 0.0 −0.4 −0.2 0.0 0.2 sim 03 = 900 PC1 PC2 −0.10 0.00 0.10 −0.4 −0.2 0.0 sim 04 = 0 −0.15 −0.05 0.05 −0.3 −0.1 0.1 sim 04 = 100 −0.05 0.05 −0.3 −0.1 0.1 sim 04 = 200 −0.15 −0.05 0.05 −0.4 −0.2 0.0 sim 04 = 300 −0.15 −0.05 0.05 −0.15 0.00 0.10 0.20 sim 04 = 400 −0.05 0.05 0.15 −0.10 0.05 0.15 sim 04 = 500 −0.10 0.00 0.10 −0.15 0.00 0.10 sim 04 = 600 −0.10 0.00 0.10 −0.2 0.0 0.1 0.2 sim 04 = 700 −0.10 0.00 0.05 −0.20 −0.05 0.10 sim 04 = 800 −0.10 0.00 0.05 −0.15 0.00 0.10 sim 04 = 900 PC1 PC2 −0.10 0.00 0.10 0.20 −0.10 0.00 0.10 sim 05 = 0 −0.3 −0.2 −0.1 0.0 0.1 −0.2 0.0 0.1 0.2 sim 05 = 100 −0.05 0.05 0.15 0.25 −0.20 −0.05 0.05 sim 05 = 200 −0.05 0.05 0.15 0.25 −0.2 0.0 0.1 sim 05 = 300 −0.05 0.05 0.15 0.25 −0.1 0.1 0.2 sim 05 = 400 −0.05 0.05 0.15 −0.25 −0.10 0.05 sim 05 = 500 −0.1 0.0 0.1 0.2 0.3 −0.15 0.00 0.10 sim 05 = 600 −0.25 −0.15 −0.05 0.05 −0.10 0.00 0.10 0.20 sim 05 = 700 −0.20 −0.10 0.00 0.10 −0.20 −0.05 0.05 sim 05 = 800 −0.20 −0.10 0.00 0.10 −0.1 0.0 0.1 0.2 sim 05 = 900 PC1 PC2 −0.10 0.00 0.10 0.20 −0.15 −0.05 0.05 sim 06 = 0 −0.20 −0.10 0.00 0.10 −0.15 0.00 0.10 0.20 sim 06 = 100 −0.10 0.00 0.10 −0.10 0.00 0.10 0.20 sim 06 = 200 −0.15 −0.05 0.05 −0.15 0.00 0.10 sim 06 = 300 −0.10 0.00 0.10 −0.1 0.1 0.2 sim 06 = 400 −0.20 −0.10 0.00 0.10 −0.3 −0.1 sim 06 = 500 −0.10 0.00 0.10 0.20 −0.1 0.1 0.3 sim 06 = 600 −0.15 −0.05 0.05 0.15 −0.1 0.1 0.3 sim 06 = 700 −0.30 −0.20 −0.10 0.00 −0.10 0.00 0.10 sim 06 = 800 −0.15 −0.05 0.05 −0.15 −0.05 0.05 sim 06 = 900 PC1 PC2 −0.20 −0.10 0.00 −0.05 0.10 0.25 sim 07 = 0 0.0 0.1 0.2 0.3 −0.3 −0.1 0.1 sim 07 = 100 −0.3 −0.2 −0.1 0.0 −0.15 0.00 0.10 sim 07 = 200 −0.25 −0.15 −0.05 0.05 −0.25 −0.10 0.05 sim 07 = 300 −0.05 0.05 0.15 −0.15 0.00 0.10 sim 07 = 400 −0.5 −0.3 −0.1 −0.3 −0.1 0.1 sim 07 = 500 −0.3 −0.2 −0.1 0.0 −0.05 0.05 0.15 sim 07 = 600 −0.25 −0.15 −0.05 0.05 0.0 0.2 0.4 sim 07 = 700 −0.05 0.05 0.15 0.25 −0.15 −0.05 0.05 sim 07 = 800 −0.20 −0.10 0.00 −0.5 −0.3 −0.1 sim 07 = 900 PC1 PC2 −0.15 −0.05 0.05 0.15 −0.15 −0.05 0.05 sim 08 = 0 −0.15 −0.05 0.05 −0.10 0.00 0.10 sim 08 = 100 −0.10 0.00 0.10 −0.10 0.00 0.10 sim 08 = 200 −0.10 0.00 0.10 −0.10 0.00 0.10 0.20 sim 08 = 300 −0.10 0.00 0.10 −0.15 −0.05 0.05 0.15 sim 08 = 400 −0.15 −0.05 0.05 −0.10 0.00 0.10 sim 08 = 500 −0.15 −0.05 0.05 −0.20 −0.05 0.05 sim 08 = 600 −0.15 −0.05 0.05 −0.20 −0.05 0.10 sim 08 = 700 −0.15 −0.05 0.05 −0.15 −0.05 0.05 sim 08 = 800 −0.10 0.00 0.10 −0.15 −0.05 0.05 sim 08 = 900 PC1 PC2 −0.10 0.00 0.10 0.20 −0.20 −0.05 0.05 sim 09 = 0 −0.10 0.00 0.10 0.20 −0.20 −0.05 0.10 sim 09 = 100 −0.20 −0.10 0.00 −0.10 0.05 0.15 sim 09 = 200 −0.15 −0.05 0.05 −0.10 0.00 0.10 sim 09 = 300 −0.15 −0.05 0.05 −0.05 0.05 0.15 sim 09 = 400 −0.15 −0.05 0.05 −0.20 −0.05 0.10 sim 09 = 500 −0.15 −0.05 0.05 −0.25 −0.10 0.05 sim 09 = 600 −0.15 −0.05 0.05 −0.10 0.05 0.15 sim 09 = 700 −0.05 0.05 0.15 0.0 0.1 0.2 sim 09 = 800 −0.05 0.05 0.15 −0.1 0.0 0.1 0.2 0.3 sim 09 = 900 PC1 PC2 Supplemen a y Figu e 24: PCA a a species-wide low- ecombining egion a en ime poin s o en exempli ied simula ion eplica es. Supplemen a y da a ela ed o Fig. 4. To in es iga e he e ec s o species-wide educ ion in local ecombina ion a e, we simula ed one ances al popula ion o 1,000 diploids wi h a low- ecombining genomic egion ha spli s in o h ee subpopula ions (pop1, pop2, pop3. Fig. 4A). All mu a ions we e neu al. We sampled indi iduals o e ime a e he popula ion spli and conduc ed PCA in he low- ecombining genomic egion. 10 ows ep esen 10 simula ion eplica es (ou o 100, see Ma e ials and Me hods). 10 columns ep esen 10 ime poin s. Da a poin s wi h h ee di e en clou s depic indi iduals om h ee di e en popula ions. In addi ion o Fig. 4B and C, hese exempli ied esul s show high a iabili y in ealised gene ic a ia ion a low- ecombining egions ac oss eplica es wi h h ee o six clus e s wi h di e en deg ees o mix u e o indi iduals in PCA and ansi ioning om haplo ype s uc u e o popula ion s uc u e o e ime. 118 −0.15 −0.05 0.05 −0.10 0.00 0.10 0.20 PC1 (13.9%) PC2 (9.4%) AA BB CC AB AC BC A 0.0 0.1 0.2 0.3 0.4 0.5 BC AC AB CC BB AA Posi ion [Mb] 150 diploid (300 haploid) genomes B BC AC AB CC BB AA A B C C B C A B A C C B B A A Geno ype Haplo ype Ma ke Supplemen a y Figu e 25: Clus e s o indi iduals in local PCA ep esen haplo ype s uc u e. Supplemen a y da a ela ed o Figs. 3, 4. A. Local PCA o a species-wide low- ecombining egion simula ed showing six clus e s o indi iduals ( he same as Fig. 4C, =0). B. Geno ypes a hinned mu a ion si es. Cyan, magen a, and yellow co espond o A-, B-, and C-speci ic mu a ions. The dis ibu ion o hese mu a ions in homozygous (AA, BB, CC) and he e ozygous(AB, AC, BC) indi iduals sugges s ha he clus e s o indi iduals in PCA ep esen combina ion o haplo ypes possessed by diploid indi iduals. 119 3.4.3 Popula ion-speci ic educ ion o local ecombina ion a e ( o wa d simula- ion) Time [N gene a ions] Recombina ion supp ession N = 1,000 NNN Supplemen a y Figu e 26: Summa y o model 1 o popula ion-speci ic ecombina ion supp ession. Supplemen a y da a ela ed o Fig. 5. A. Simula ed scena io. Simula ed genome con ained wo ch omosomes, one wi h a popula ion-speci ic low- ecombining egion and he o he wi hou . B, C. PCA showing pa e ns o gene ic a ia ion a he popula ion-speci ic low- ecombining egion (B) and he no mally ecombining ch omosome (C) a h ee ime poin s in one exempli ied simula ion eplica e. 120 Supplemen a y Figu e 27: Dis inc local PCAs a popula ion-speci ic low- ecombining egion ep esen c yp ic haplo ype s uc u e. To cha ac e ise ac o s ep esen ed in he p ima y axes o dis inc local PCA a popula ion-speci ic low- ecombining egions, we pe o med one eplica e o SLiM simula ion wi h he scena ios o models 1 (A-D) and model 2 (E-H) eco ding he ull ances y and mu a ions in ee sequence in addi ion o VCF iles (De ailed in Ma e ials and Me hods). We pe o med PCA, and iden i ied mu a ions wi h he highes con ibu ions o he PC1 and PC2. We analysed he ee sequence o add ess whe he mu a ions ha occu ed in ce ain popula ion (e.g. ances al popula ion, low- ecombining popula ion) we e en iched in he se o mu a ions con ibu ing o he PC1 and PC2. A, E. Local PCA a popula ion-speci ic low- ecombining egion. B, F. No isible haplo ype s uc u e we e ound a popula ion-speci ic low- ecombining egion. C, G. Mu a ions o igina ing om he low- ecombining popula ion a e he popula ion-speci ic ecombina ion supp ession we e en iched in mu a ions wi h high loading o he dis inc pa e n o local PCA. D, H. Mu a ions o igina ing om he low- ecombining popula ion wi h high PCA loading sha e common genealogical edges. Due o ecombina ion supp ession, he mu a ions on he same edge a e on he same haplo ype in he cu en sample. 121 Supplemen a y Figu e 34: Genomic dis ibu ion o andem epea s wi h long epea uni . In each ch omosome, six andem epea s wi h he longes epea uni s a e shown. Shades indica e genomic egions wi h dis inc pa e ns o gene ic a ia ion. 128 3.7 Genealogical in e p e a ion Recombina ion Ances al haplo ypes p esen a ime T Diploid indi idual A B PC1 PC2 12 3 4 5 67 8 T Ances ies Diploid ID Time F eely ecombining Recombina ion supp ession in pop1 Pas P esen Popula ion 123 4 56 7 8 pop1 pop2 Supplemen a y Figu e 35: Genealogical in e p e a ion o he e ec o popula ion-speci ic ecombina ion supp ession on local gene ic a ia ion. A. An ances al ecombina ion g aph (ARG) ep esen ing ances ies o 16 hypo he ical haploid sequences o 8 diploid sampled indi iduals om wo popula ions. Thei ances ies can be aced back o n1 = n2 = 3 ances al haplo ypes ( he same se o simplici y) p esen a ime T when popula ion-speci ic ecombina ion supp ession ini ia ed in pop1. The ances ies o hese n1 and n2 ances al haplo ypes eely ecombine a imes olde han T . A he bo om, he ances ies o each cu en haplo ype a e shown. Closed and open ci cles ep esen p esence and absence o con ibu ion om he espec i e ances al haplo ype. B. A hypo he ical PCA ep esen ing he pa e n o gene ic a ia ion o he ocal egion. Indi iduals om he low- ecombining popula ion (pop1) a e sp ead in he hypo he ical PCA because each diploid has a combina ion o disc e e ances ies. The indi iduals o no mally ecombining popula ions (pop2) a e clus e ed a ound he cen e because hey ha e mixed haplo ypes due o con inued ecombina ion a e T. 129 4 Re e ences Huang, K., And ew, R. L., Owens, G. L., Os e ik, K. L., & Riesebe g, L. H. (2020). Mul iple ch omosomal in e sions con ibu e o adap i e di e gence o a dune sun lowe eco ype. Molecula Ecology,29 (14), 2535–2549. h ps://doi.o g/h ps://doi.o g/10.111 1/mec.15428 Le oy, T., Anselme i, Y., Tilak, M.-K., Bé a d, S., Csukonyi, L., Gab ielli, M., Sco na acca, C., Milá, B., Thébaud, C., & Nabholz, B. (2021). A bi d’s whi e-eye iew on a ian sex ch omosome e olu ion. Pee Communi y Jou nal,1.h ps://doi.o g/10.24072/pcjou n al.70 Ma, J., & Amos, C. I. (2012). In es iga ion o In e sion Polymo phisms in he Human Genome Using P incipal Componen s Analysis. PLOS ONE,7(7), e40224. h ps: //doi.o g/10.1371/jou nal.pone.0040224 Pala, I., Nau in, S., S e ande , M., Hasselquis , D., Bensch, S., & Hansson, B. (2012). E idence o a neo-sex ch omosome in bi ds. He edi y,108 (3), 264–272. h ps://doi.o g/ 10.1038/hdy.2011.70 Shipilina, D., Pal, A., S ankowski, S., Chan, Y. F., & Ba on, N. H. (2023). On he o igin and s uc u e o haplo ype blocks. Molecula Ecology,32 (6), 1441–1457. h ps: //doi.o g/10.1111/mec.16793 Sigeman, H., Ponnikas, S., & Hansson, B. (2020). Whole-genome analysis ac oss 10 songbi d amilies wi hin Syl ioidea e eals a no el au osome–sex ch omosome usion. Biology Le e s,16 (4), 20200082. h ps://doi.o g/10.1098/ sbl.2020.0082 Sigeman, H., S andh, M., P oux-Wé a, E., Ku sche a, V. E., Ponnikas, S., Zhang, H., Lundbe g, M., Sole , L., Bunikis, I., Ta ka, M., Hasselquis , D., Nys ed , B., Wes - e dahl, H., & Hansson, B. (2021). A ian Neo-Sex Ch omosomes Re eal Dynamics o Recombina ion Supp ession and W Degene a ion. Molecula Biology and E olu ion, 38(12), 5275–5291. h ps://doi.o g/10.1093/molbe /msab277 S ephan, W. (2019). Selec i e Sweeps. Gene ics,211 (1), 5–13. h ps://doi.o g/10.1534/ge ne ics.118.301319 Todesco, M., Owens, G. L., Be co ich, N., Léga é, J.-S., Soudi, S., Bu ge, D. O., Huang, K., Os e ik, K. L., D ummond, E. B. M., Ime o ski, I., Lande, K., Pascual-Robles, M. A., Nana a i, M., Jahani, M., Cheung, W., S a on, S. E., Muños, S., Nielsen, R., Dono an, L. A., e al. (2020). Massi e haplo ypes unde lie eco ypic di e en ia ion in sun lowe s. Na u e,584 (7822), 602–607. h ps://doi.o g/10.1038/s41586-020-2467-6 130 3 High-Recombining Genomic Regions A ec Demog aphy In e ence “I ’s—i ’s a a iable.” Kaplan was shaking, whi e-lipped and pale. “Some hing om which no in e ence can be made. The man om he pas . The machines can’ deal wi h him. The a iable man!” – Philip K. Dick, The Va iable Man (1953) 131 High- ecombining genomic egions a ec demog aphy in e ence Jun Ishigohoka1,* Mi iam Lied ogel1,2,3,* 1 MPRG Beha iou al Genomics, Max Planck Ins i u e o E olu iona y Biology, 24306 Plön, Ge many 2Ins i u e o A ian Resea ch, An de Vogelwa e 21, 26386 Wilhelmsha en, Ge many 3 Depa men o Biology and En i onmen al Sciences, Ca l on Ossie zky Uni e si ä Oldenbu g, Amme lände Hee s aße 114-118, 26129 Oldenbu g, Ge many * Co espondence: Jun Ishigohoka <ishigohoka@e olbio.mpg.de>,Mi iam Lied ogel <lied o- gel@e olbio.mpg.de> Keywo ds: demog aphy in e ence, ances al ecombina ion g aph, ecombina ion a e, popu- la ion genomics, non-model species, Syl ia a icapilla. Running i le: Recombina ion a e and demog aphy in e ence. 132 Abs ac In e ence o popula ion his o y o non-model species is impo an in e olu iona y and conse - a ion biology. Mul iple me hods o popula ion genomics, including hose o in e popula ion his o y, a e based on he ances al ecombina ion g aph (ARG). These me hods use obse ed mu a ions o model local genealogies changing along ch omosomes. B eakpoin s a which genealogies change e ec i ely ep esen he posi ions o his o ical ecombina ion e en s. How- e e , in e ence o unde lying genealogies is di icul in egions wi h high ecombina ion a e ela i e o mu a ion a e. This is because genealogies co e genomic in e als ha a e oo sho o accommoda e su icien ly many mu a ions in o ma i e o he s uc u e o he un- de lying genealogies. Despi e he p e alence o high- ecombining genomic egions in some non-model o ganisms, such as bi ds, i s e ec on ARG-based demog aphy in e ence has no been well s udied. He e, we use popula ion genomics simula ions o in es iga e he impac o high- ecombining egions on ARG-based demog aphy in e ence. We demons a e ha in e ence o e ec i e popula ion size and he ime o popula ion spli e en s is sys ema ically a ec ed when high- ecombining egions co e wide b ead hs o he ch omosomes. We also show ha excluding high- ecombining genomic egions can p ac ically mi iga e his e ec . Finally, we con i m he ele ance o ou indings in empi ical analysis by con as ing demog aphy in e ences applied o a bi d species, he Eu asian blackcap (Syl ia a icapilla), using di e en pa s o he genome wi h high and low ecombina ion a es. Ou esul s sugges ha demog aphy in e ence using ARG-based me hods should be ca ied ou wi h cau ion when applied in species whose e e ence genomes con ain long s e ches o high- ecombining egions. 133 In oduc ion Popula ion his o y a ec s he pa e ns o gene ic a ia ion, and con e sely obse ed gene ic a ia ion in genomes allows in e ence o his o ical demog aphic pa ame e s. The inc easing a ailabili y o genome da a o a ious species a a popula ion le el has acili a ed de elopmen and applica ion o a numbe o popula ion genomics me hods o demog aphy in e ence (Exco ie e al., 2013; Gu enkuns e al., 2009; Ha is & Nielsen, 2013; Li & Du bin, 2011; Liu & Fu, 2020; Schi els & Du bin, 2014; Te ho s e al., 2017). These app oaches a e ypically i s applied o human da a o unde s and he popula ion his o y o ou own species and o alida ion o he new me hods (Exco ie e al., 2013; Gu enkuns e al., 2009; Ha is & Nielsen, 2013; Li & Du bin, 2011; Liu & Fu, 2020; Schi els & Du bin, 2014; Te ho s e al., 2017), bu he ea e adop ed o o he species including domes ica ed and wild o ganisms o answe e olu iona y ques ions (Alonso-Blanco e al., 2016; G oenen e al., 2012; Lanie e al., 2015; Liu e al., 2014; Nadachowska-B zyska e al., 2016) and o assis conse a ion e o s (Dussex e al., 2021; Hohenlohe e al., 2021; Li e al., 2014; Pacheco e al., 2022). Despi e he wide applica ion o demog aphy in e ence me hods in non-model o ganisms, hei pe o mance ou side he pa ame e space o humans has no been well e alua ed. Some me hods o demog aphy in e ence a e based on he ances al ecombina ion g aph (ARG) (Li & Du bin, 2011; Schi els & Du bin, 2014; Speidel e al., 2019; Te ho s e al., 2017). The ARG is a s uc u e ha desc ibes he ull ances ies o sampled genomes along ecombining ch omosomes (G i i hs & Ma jo am, 1997). I essen ially consis s o a se ies o ma ginal genealogical ees changing in he opology and b anch leng hs along he ch omosome, and hei b eakpoin s e ec i ely ep esen his o ical ecombina ions con ibu ing o he sampled genomes (Fig. 1). The ull ARG p o ides ich in o ma ion on he popula ion his o y (i.e. all coalescence and ecombina ion e en s h ough ime and mu a ions mapped on b anches), making ARG-based me hods a powe ul popula ion genomics app oach o s udy e olu iona y p ocesses (Hubisz e al., 2020; Schae e e al., 2021; Speidel e al., 2019; S e n e al., 2019; Wohns e al., 2022). In p ac ice, howe e , ARG-based me hods depend on in e ence o he ARG (Igna ie a e al., 2021; Kellehe e al., 2019; Mi zaei & Wu, 2017; Rasmussen e al., 2014; Speidel e al., 2019; Wohns e al., 2022), o ep esen a ions o unde lying genealogies (Li & 134 Du bin, 2011; Schi els & Du bin, 2014; Te ho s e al., 2017), which in u n elies on obse ed mu a ions. Impo an ly, he p esence o mu a ions ep esen ing an ARG b anch depends on ecombina ion and mu a ion a es. I an ances al haplo ype b eaks by a ecombina ion be o e accommoda ing mu a ions, he co esponding b anch on he ARG is no ep esen ed by any mu a ions (Fig. 1B) (Hayman e al., 2023; Shipilina e al., 2023). The e o e, high ecombina ion a es ( ela i e o he mu a ion a e) makes i di icul o accu a ely in e he unde lying ARG, limi ing he pe o mance o he ARG-based app oach (Sellinge e al., 2020, 2021; Te ho s e al., 2017). ... Recombina ion Mu a ion A B Posi ion Figu e 1: The p esence o mu a ions ep esen ing ARG b anches depends on ecombina ion a e. A. When ecombina ion a e is mode a ely low, b anches o ARG a e ep esen ed by mu a ions. This allows in e ence o he unde lying ARG based on obse ed mu a ions. B. When ecombina ion a e is high, many b anches o ARG a e no ep esen ed by any mu a ions. We ask whe he his a ec s ARG-based demog aphy in e ence. The impac o high ecombina ion a e on ARG-based demog aphy in e ence is p esumably negligible in humans (Li & Du bin, 2011; Te ho s e al., 2017) whe e ecombina ion a e is low excep o na ow ecombina ion ho spo s (Mye s e al., 2010; S e ison e al., 2016). Howe e , his ype o ecombina ion landscape is no uni e sal o all o ganisms (Au on e al., 2013; Bake e al., 2017; Lam & Keeney, 2015; Singhal e al., 2015), including species wi h ecological and 135 e olu iona y ele ance o o conse a ion conce n. The di e ence in ecombina ion landscapes can be pa ially a ibu ed o he p esence and absence o PRDM9, a ansc ip ion ac o ha de e mines he genomic posi ion o ecombina ion ho spo s. PRDM9 in oduces his one modi ica ions o ec ui he molecula machine y ini ia ing DNA double-s and b eaks (DSBs), which is equi ed o meio ic ecombina ion (Bake e al., 2015; Bauda e al., 2010; Paigen & Pe ko , 2018). PRDM9 has been los independen ly a leas hi een imes in e eb a es (Ca assim e al., 2022), which shi ed he ecombina ion ho spo s om apidly e ol ing PRDM9 mo i s (Bake e al., 2015; Mye s e al., 2010; Oli e e al., 2009) o genome ea u es such as ansc ip ion s a si es and CpG-islands (Au on e al., 2013; Bake e al., 2017; Kawakami e al., 2017; Paigen & Pe ko , 2018; Singhal e al., 2015). Ho spo s o PRDM9-independen ecombina ion in bi ds (Bascón-Ca dozo e al., 2024; Kawakami e al., 2017; Singhal e al., 2015), dogs (Au on e al., 2013) and pe como ph ish (Bake e al., 2017) appea o be wide han PRDM9-dependen ho spo s in p ima es (Du bin e al., 2010; Mye s e al., 2010; S e ison e al., 2016). On op o he ecombina ion landscape, he a e age ecombina ion a e is highly a iable be ween ch omosomes and species (S apley e al., 2017). These di e ences in meio ic ecombina ion could po en ially impac modelling o he local ARGs in non-human species. In his s udy, we ask how ecombina ion landscapes a ec ARG-based demog aphy in e ence. To his end, we simula e genome da a unde a simple demog aphic his o y wi h a ious ecombina ion maps, and e alua e he accu acy o demog aphy in e ence by di e en ARG-based me hods. Speci ically, we ocus on wo ARG-based me hods, MSMC2 (Malaspinas e al., 2016; Wang e al., 2020) and Rela e (Speidel e al., 2019), di e ing in he way he ARG is modelled. While Rela e in e s a se ies o ma ginal genealogies along he genome wi h hei opology and b anch leng hs collec i ely ep esen ing he ull ARG o he sample, MSMC2 models he dis ibu ion o he coalescence imes be ween pai s o sampled sequences along he genome based on he sequen ially Ma ko ian coalescen (SMC, McVean & Ca din (2005)). To demons a e he ele ance o ou indings based on simula ions, we ansla e ou indings o empi ical da a o a non-model o ganism wi h wide high- ecombining genomic egions. To his end, we use whole-genome esequencing (WGR) da a and ine-scale ecombina ion maps o a songbi d species, he Eu asian blackcap (Syl ia a icapilla), and con as ARG-based 136 demog aphy in e ences using genomic egions di e ing in ecombina ion a es. Resul s Simula ions o di e en ecombina ion maps To in es iga e he e ec o he ecombina ion landscape on ARG-based demog aphy in e ence, we used msp ime (Kellehe e al., 2018) o simula e a simple demog aphic his o y (Fig. 2A) wi h i e di e en ecombina ion maps (Fig. 2B). In all simula ions, h ee subpopula ions (pop1, pop2, and pop3) spli om a cons an -sized ances al popula ion o 1 million diploids 10,000 gene a ions be o e he p esen , a e which hey ollowed di e en ajec o ies o e ec i e popula ion size (cons an (pop1), exponen ial inc ease (pop2), and exponen ial dec ease (pop3)). Unde his demog aphy model, we simula ed 16 Mb-long ch omosomes (Fig. 2B) wi h a cons an mu a ion a e and he e ogeneous ecombina ion a e along he ch omosome (colo -coded om o ange (no high- ecombining egions) o da k ed (ex eme high- ecombining egions)). Speci ically, he 10 Mb s e ch in he middle o he ch omosome (“middle”) had a ecombina ion a e one- en h he mu a ion a e in all i e scena ios, while he ecombina ion a e inc eased in a s epwise manne a he 3 Mb ends o he ch omosome. To es whe he masking high- ecombining egions imp o es ARG-based demog aphy in e ence, we applied masks accoun ing o he leng h o ch omosomes in wo ways: masking a o al o 6 Mb wi hin he cen al pa o he ch omosome wi hou ele a ed ecombina ion a e (“con ol”, Fig. 2C op) o masking he 3 Mb ends o he ch omosome co e ing he en i e high- ecombining egions (Fig. 2C bo om). Mean ecombina ion-mu a ion a ios o he i e scena ios o he con ol condi ion (a e applying he masks) we e 0.1, 0.25, 1, 4, and 10 (Fig. 2B, C)). We in e ed he demog aphy wi h wo ARG-based me hods, MSMC2 and Rela e . We also in e ed demog aphy using S ai way plo 2 , an SFS-based me hod expec ed o be una ec ed by ele a ed local ecombina ion a e. In summa y, we es ed he pe o mance o demog aphy in e ence in g ids o i e scena ios (di e ing in ecombina ion maps), h ee demog aphy in e ence me hods ( wo ARG-based and one SFS-based), and wo condi ions o masking (masking high- ecombining and backg ound egions). 137 in es iga e he e ec o ecombina ion a e on ARG-based me hods o demog aphy in e ence wi h he blackcap da ase , we spli he blackcap genome in o low- and high- ecombining hal es, based on local ecombina ion a es cha ac e ized in a p e ious s udy (Bascón-Ca dozo e al., 2024) (see Ma e ials and Me hods o de ails). We pe o med demog aphy in e ence (e ec i e popula ion size and CCR) wi h MSMC2 and Rela e o each hal sepa a ely. In e ence by MSMC2 showed appa en e ec s o high- ecombining egions consis en wi h ou simula ion. His o ical e ec i e popula ion size in e ed using he high- ecombining hal had cha ac e is ic wa e-shaped ajec o y in he deep pas o he skyline plo (Fig. 5B) compa ed o ha using he low- ecombining hal (Fig. 5A). The appa en spli ime be ween popula ions based on e ec i e popula ion size was olde using he high- ecombining hal (Fig. 5B) han using he low- ecombining hal (Fig. 5A). In line wi h his, di ec compa ison o in e ed CCR o pai s o blackcap popula ions be ween he high- and low- ecombining hal es e ealed in e ence o sys ema ically olde spli ime using he high- ecombining hal han he low- ecombining hal (Fig. 5C, D). These di e ences a e consis en wi h ou simula ion s udy (Figs. 3,4), indica ing ha he e ec o high- ecombining egions on in e ence by MSMC2 is ele an o empi ical analysis. In con as , Rela e was obus o he di e ence in ecombina ion a e be ween he wo condi ions (Sup. Fig. 8). This di e ence be ween MSMC2 and Rela e sugges s di e en le els o obus ness o ARG-based me hods o he p esence o high- ecombining egions, which is in line wi h ou simula ion s udy. 144 103104105106 104 105 106 Time [yea s be o e p esen ] Ne A Low− ec. hal con _medlong con _ esiden isl_Cana y isl_CapeVe de isl_Azo es 103104105106 104 105 106 Time [yea s be o e p esen ] Ne B High− ec. hal 103104105106 0.00 0.25 0.50 0.75 1.00 High Low Time [yea s be o e p esen ] CCR C 0 50 100 150 200 0 50 100 150 200 Time [x 1,000 yea s be o e p esen ] Low High Time [x 1,000 yea s be o e p esen ] D Uppe bound Lowe bound Figu e 5: High- ecombining egions can a ec demog aphy in e ence in empi ical analysis. A, B. In e ence o his o ical e ec i e popula ion size by MSMC2 . Resul s o i e exempli ied blackcap popula ions a e shown using he lowe (A) o he highe (B) hal o he genome based on local ecombina ion a es. C. In e ence o ela i e c oss-coalescence a e ( CCR) wi h MSMC2 be ween Azo es popula ion and each o all o he ou popula ions in Aand Busing he lowe (solid lines) and he highe (do ed lines) hal o he genome based on local ecombina ion a es. D. Compa ison o spli imes in e ed by MSMC2 using he lowe highe hal es o he genome based on ecombina ion a es. Segmen s ep esen in e ence be ween 45 pai s o 10 popula ions. Two ends o a segmen ep esen he lowe and uppe bounda ies o wo consecu i e disc e ized epochs be ween which CCR c osses he h eshold o 0.5. 145 Discussion Ou esul s sugges ha demog aphy in e ence using ARG-based me hods should be ca ied ou wi h cau ion in o ganisms ha a e likely o ha bo ecombina ion landscapes dis inc om humans, o which hese me hods we e ini ially de eloped. In many animals wi h unc ional PRDM9, including humans, ecombina ion e en s a e concen a ed in na ow ecombina ion ho spo s (Mye s e al., 2010; S e ison e al., 2016). Thus, we expec ha ARG-based me hods will be obus . O he species ha e high- ecombining egions mo e widely dis ibu ed a ound genomic ea u es along he genome (Bake e al., 2017; Kawakami e al., 2017; Singhal e al., 2015) and hus ARG-based me hods can be mo e suscep ible o he e ec o high- ecombining egions. Using simula ions we demons a ed ha masking high- ecombining egions imp o es he ARG-based demog aphy in e ence in such cases. In p ac ice, howe e , his aises ano he ques ion o how o de ine egions o mask, which can be challenging due o mul iple ac o s. Fi s , de ining a h eshold alue o he ecombina ion a e using in e ed ecombina ion maps may be p oblema ic. This is because me hods o in e ence o ine-scale ecombina ion maps can be inaccu a e, especially when he ecombina ion a e is highe han he mu a ion a e (Raynaud e al., 2023; Spence & Song, 2019). Second, long-enough con iguous ch omosomal segmen s a e essen ial in ARG-based me hods (Sellinge e al., 2021), hence masking e e y high- ecombining egion can be p oblema ic as i migh spli he genome in o pieces oo small o ARG-based me hods o be applied. An addi ional ac o o ake in o accoun is a ia ion among ch omosomes. Fo example, in mul iple axa, he ch omosome leng hs can subs an ially a y, and he ecombina ion a e is nega i ely co ela ed wi h he ch omosome leng h (Bascón-Ca dozo e al., 2024; Kawakami e al., 2014; Ma in e al., 2019; Singhal e al., 2015), po en ially leading o di e en applicabili y o ARG-based me hods among ch omosomes. Finally, addi ional masks may be necessa y o demog aphy in e ence i la ge blocks ha do no ep esen neu al e olu ion exis in he genome. Fo example, la ge polymo phic in e sions unde long- e m balancing selec ion (Gi aldo-Deck e al., 2022; Hage e al., 2022; Ha ingmeye & Hoeks a, 2022; Kim e al., 2017; Knie e al., 2016, 2017; Küppe e al., 2015; Lamichhaney e al., 2015; Mé o e al., 2021) may be excluded, which may lea e li le da a o in e ence in species wi h small genomes. We sugges o un simula ions ailo ed o he species 146 unde s udy o assess whe he ARG-based me hods can be used wi h some con idence. This is especially necessa y in species wi hou unc ional PRDM9, wi h b oad high- ecombining egions, high genome-wide mean ecombina ion a es, small genomes, highly he e ogeneous ch omosomes, and la ge s uc u al a ia ions. An SFS-based me hod o demog aphy in e ence, S ai way plo 2 , pe o med well wi hou high- ecombining egions and e en be e wi hou high- ecombining egions in ou simula ion s udy. We p opose ha his accu acy unde he p esence o high- ecombining egions can ep esen a gene al cha ac e is ic o SFS-based me hods ha hey bene i om high- ecombining egions, om which ARG-based me hods su e . The p oblem o high- ecombining egions o ARG-based me hods is he ac ha b anches o genealogies a e no ep esen ed by mu a ions (Hayman e al., 2023). In o he wo ds om he pe spec i e o mu a ions, ARG-based me hods su e om independence o mu a ions in a local genomic window. This independence o mu a ions, howe e , is he assump ion o compu e SFS (Gu enkuns e al., 2009), allowing SFS-based me hods o pe o m accu a ely wi h high- ecombining egions. Localized e o s in he in e ence o ( ep esen a ion o ) genealogies wi hin high- ecombining egions in ou s udy indica e ha he issue o high- ecombining egions in ARG-based me hods is no speci ic o demog aphy in e ence bu can be c i ical in o he applica ions, including in e ence o selec ion (Hejase e al., 2020; Speidel e al., 2019; S e n e al., 2019). In egions wi h high ecombina ion a es, in e ed genealogies may be oo inaccu a e o pe o m ARG-based selec ion es s, while SFS-based me hods may be used on he local a ia ion da a (Fay & Wu, 2000; Tajima, 1989). Combining ARG- and SFS-based app oaches, gi ing hem complemen a y weigh s acco ding o he local ecombina ion a e, may make he mos o he a ia ion da a in demog aphy in e ence. In his s udy, we demons a ed ha he ecombina ion landscape can in luence ARG-based app oaches o popula ion genomics. Al hough he ue ARGs should ha e ich in o ma ion on he popula ion his o y and e olu iona y p ocesses, he e ec s o e o s in in e ed local ARGs wi hin egions o ele a ed ecombina ion a e a e, in some cases, no negligible. Ou indings a e likely ele an no only o bi ds bu likely in a wide ange o species, because PRDM9 has been los a leas hi een imes independen ly in e eb a es ( i e clades o 147 ay- inned ish, ou clades o amphibians, a clade o liza ds, he en i e clade o bi ds and c ocodiles, and wo clades o mammals (dogs and pla ypus) (Ca assim e al., 2022)). In addi ion o he ecombina ion a e, o he ac o s, such as he genomic landscape and spec a o mu a ion (Jiang e al., 2021; Mon oe e al., 2022; Sasani e al., 2022; Wu e al., 2020), local e ec i e popula ion size ( e lec ing selec ion: Nielsen, 2005; Bu i, 2017; Elleg en & Gal ie , 2016), and e ec i e mig a ion a e ( e lec ing ba ie s o gene low: Wes am e al., 2022) a e dis ibu ed non-uni o mly along he genome, and hey may simila ly a ec popula ion genomics summa y s a is ics and in e ences. No el app oaches join ly modelling he e ogenei y o some o hese ac o s a e eme ging (Ba oso & Du heil, 2023; Ko mann e al., 2023; Lae sch e al., 2023). None heless, we highligh ha e alua ing he pe o mance and limi a ion o popula ion genomics me hods unde non-canonical pa ame e space ele an in indi idual cases is necessa y o d aw meaning ul in e p e a ions. Ma e ials and Me hods Simula ion s udy Simula ion To in es iga e he e ec o high- ecombining egions on demog aphy in e ence, we simula ed ARGs and mu a ions wi h msp ime e sion 1.2.0 (Baumdicke e al., 2022) unde he s anda d neu al coalescen wi h ecombina ion (Hudson, 1983). The demog aphy model consis ed o an ances al popula ion o 1,000,000 diploids spli ing in o h ee popula ions (pop1, pop2, and pop3) a 10,000 gene a ions be o e he p esen ime. The popula ion size o pop1 was cons an a 10,000, and exponen ial inc ease and dec ease o 10 olds o e 10,000 gene a ions we e in oduced in pop2 and pop3 a e he spli e en . The mu a ion a e was se o 4 . 6 × 10 −9 pe gene a ion pe si e. We p epa ed h ee se s o ecombina ion maps, each o which consis s o i e scena ios. The i s se (“s epwise”) was 16 Mb long, and he ecombina ion a e was se o one en h he mu a ion a e (4 . 6 × 10 −10 ) a he cen al 10 Mb, wi h a s ep-wise inc ease in ecombina ion a e a 3 Mb ends o he ch omosome (Fig. 2B), such ha he mean ecombina ion a e (a e 148 masking 6 Mb o he middle) we e 0.1, 0.25, 1, 4, and 10 imes he mu a ion a e. The second se (“na ow high- ec.”) was 11 Mb long, and he ecombina ion a e was se o 4 . 6 × 10 −10 h oughou he ch omosome, excep a 1 Mb segmen in he middle, whe e ecombina ion a e was ele a ed such ha he mean ecombina ion a e we e 0.1, 0.25, 1, 4, and 10 imes he mu a ion a e Sup. Fig. 5A. The hi d se (“uni o m”) consis ed o i e uni o m ecombina ion maps o 10 Mb wi h ecombina ion a e o 0.1, 0.25, 1, 4, and 10 imes he mu a ion a e Sup. Fig. 5B. Fo he i s se , we simula ed 10 eplica es o 150 diploid indi iduals (50 indi iduals pe popula ion). Fo he second and hi d se s, we simula ed one eplica e. We eco ded he ue ARGs in T eeSeq o ma , and also eco ded haplo ype da a in VCF o ma using ski e sion 0.4.1 (Kellehe e al., 2018). Demog aphy in e ence MSMC2 Fo he na ow high- ec. and uni o m scena ios, we used ou diploid indi iduals om each popula ion o demog aphy in e ence wi h MSMC2 (Malaspinas e al., 2016; Wang e al., 2020). Fo he s epwise scena io, we ea ed en simula ions as en independen ch omosomes, and downsampled ou diploid indi iduals (eigh haploid sequences) pe popula ion wi hou eplacemen en imes as en “ eplica es” (No e ha hey a e no ue independen eplica es because hey we e sampled om a common ARG o each ch omosome). Inpu mul ihe sep iles we e gene a ed om he VCF ile and masks o each ch omosome o each downsample o each eplica e using gen a e_mul ihe sep.py o msmc- ools (Schi els & Wang, 2020). We an MSMC2 o each popula ion o popula ion pai o in e his o ical coalescence a es. The es ima es o his o ical e ec i e popula ion size we e ob ained as he in e se o he in e ed coalescence a e o each popula ion, scaled wi h he ue mu a ion a e o 4 . 6 × 10 −9 . The CCR was ob ained by di iding he be ween-popula ions coalescence a e wi h he a e age wi hin-popula ion coalescence a e. Fo isualiza ion in Figs. 3,4, we compu ed mean and s anda d de ia ion o he in e ed e ec i e popula ion size and CCR wi h a cus om sc ip . Rela e Fo he s epwise scena io, we ea ed en simula ions as en independen eplica es. We applied il e ing o a iable si es based on he posi ion acco ding o masking condi ions using BCFTools e sion 1.9 (Danecek e al., 2021). We in e ed ARGs om he masked 149 VCF using Rela e e sion 1.1.6 (Speidel e al., 2019) speci ying he ue mu a ion a e, ue ecombina ion maps, and haploid popula ion size o 2,000,000, and in e ed demog aphy wi h wo i e a ions. The es ima es o his o ical e ec i e popula ion size we e ob ained as he in e se o he in e ed coalescence a e o each popula ion, scaled wi h he ue mu a ion a e o 4 . 6 × 10 −9 . The CCR was ob ained by di iding he coalescence a e be ween popula ions wi h he a e age wi hin-popula ion coalescence a e. Fo isualiza ion in Figs. 3,4, we compu ed mean and s anda d de ia ion o he in e ed e ec i e popula ion size and CCR wi h a cus om sc ip . S ai way plo 2 We an S ai way plo 2 e sion 2.1 (Liu & Fu, 2020) o he s epwise scena io. We ea ed en simula ions as en independen eplica es. We spli he VCF by popula ion applying masks wi h VCFTools e sion 0.1.16 (Danecek e al., 2011). We compu ed he un olded SFS and p epa ed bluep in con igu a ion iles using cus om sc ip s, and an S ai way plo 2 wi h de aul pa ame e alues. Fo isualiza ion in Fig. 3, we compu ed mean and s anda d de ia ion o he in e ed e ec i e popula ion size wi h a cus om sc ip . Coalescence ime analysis MSMC2 We ocused on wo haploids o he i s ch omosome (simula ion un) o he i s downsample in pop1, and compa ed ue TMRCA eco ded in he ue ARG (in T eeSeq o ma ) and in e ence by MSMC2 . We ex ac ed he ue TMRCA o he ocal pai o haploid genomes in T eeSeq wi h ski . To ob ain in e ence by MSMC2 , we an he decode p og am o MSMC2 wi h decode -m 0.0092 - 0.00736 -I 0,1 - 32 -s 1000 . Based on he ou pu o decode , we eco ded he index o epoch wi h he highes p obabili y o each window. We aligned ue and in e ed TMRCA ea ing an in e sec ed ange as a uni , and compu ed Spea man’s co ela ion coe icien in R e sion 4.3.1 (R Co e Team, 2023). Rela e We ocused on TMRCA o he en i e genealogy o 300 haploid genomes o he i s simula ion eplica e. We ex ac ed TMRCA along he ch omosome om he ue ARG in T eeSeq using ski . To ob ain TMRCA along he ch omosome o he ARG in e ed by Rela e , we con e ed he genealogies (in mu and anc o ma ) o T eeSeq using Rela eFileFo ma s 150 p og am in Rela e , and ex ac ed TMRCA along he ch omosome using ski . We aligned ue and in e ed TMRCA ea ing an in e sec ed ange as a uni , and compu ed Spea man’s co ela ion coe icien in R. Empi ical s udy Da a We used phased whole-genome esequencing (WGR) da a o 179 blackcaps (Ishigohoka e al., 2023), and unphased i e ga den wa ble s and h ee A ican hill babble s (Delmo e e al., 2020). We compu ed mean ecombina ion a e in 10-kb sliding windows along he blackcap genome based on (Bascón-Ca dozo e al., 2024). Demog aphy in e ence MSMC2 We i s applied callabili y masks o he blackcap genome and de ined high- and low- ecombining hal es o he genome o each popula ion o popula ion pai . Speci ically, we chose o each popula ion a mos ou indi iduals wi h mean ead dep h o a leas 15x, excluding pai s o ela ed indi iduals based on kinship coe icien (Manichaikul e al., 2010) compu ed using ela edness2 op ion in VCFTools . We c ea ed a mask ile pe indi idual using bamCalle .py o msmc- ools (Schi els & Wang, 2020) and me ged hem o each popula ion o popula ion pai using bed ools me ge (Quinlan & Hall, 2010). The mask o each popula ion o popula ion pai was applied on he blackcap ecombina ion map (Bascón- Ca dozo e al., 2024), and we o de ed genomic in e als wi hin he unmasked egions acco ding o he ecombina ion a e. Regions in he i s and he second hal es we e de ined as he lowe - and highe - ecombining hal es. A e de ining he egions o be used o in e ence, inpu mul ihe sep iles we e gene a ed om he phased VCF and he mask ile using gen a e_mul ihe sep.py o msmc- ools (Schi els & Wang, 2020). We an MSMC2 o each popula ion o popula ion pai o in e his o ical coalescence a es. The es ima es o his o ical e ec i e popula ion size we e ob ained as he in e se o he in e ed coalescence a e o each popula ion, scaled wi h a mu a ion a e o 4 . 6 × 10 −9 es ima ed in he colla ed lyca che (Smeds e al., 2016). The CCR was ob ained 151 by di iding he be ween-popula ion coalescence a e wi h he a e age wi hin-popula ion coalescence a e. Rela e Rela e equi es haplo ype da a wi h pola ized mu a ions. We pola ized biallelic SNPs in blackcaps using allele equencies in wo ou g oup species, ga den wa ble s (n=5) and A ican hill babble s (n=3). Speci ically, a e emo ing SNPs wi h mo e han wo alleles including he h ee species, we spli blackcap SNPs in o he ollowing i e ca ego ies. 1. Si es a which all ga den wa ble s had missing geno ype. 2. Si es ixed in ga den wa ble s 3. Si es seg ega ed among ga den wa ble s and missing in all A ican hill babble s 4. Si es seg ega ed among bo h ga den wa ble s and A ican hill babble s 5. Si es seg ega ed among ga den wa ble s and ixed in A ican hill babble s Fo each ca ego y we applied he ollowing heu is ics o pola ize mu a ions. Fo si es o he i s ype, we de ined he mino allele among blackcaps o be he de i ed s a e (i.e. he majo allele is he ances al s a e). Fo si es o he second ype, we de ined he allele possessed by ga den wa ble o be he ances al s a e. Fo si es o he hi d o ou h ype, we de ined he mino allele among blackcaps o be he de i ed s a e (i.e. he majo allele is he ances al s a e). Fo si es o he i h ype, we de ined he allele possessed by A ican hill babble o be he ances al s a e. We de ined low- and high- ecombining hal es o he genome in BED o ma based on he blackcap ecombina ion map (Bascón-Ca dozo e al., 2024). Based on hese BED iles o mask high/low- ecombining hal and epea s e ie ed om UCSC Genome B owse acks (Raney e al., 2023) o he blackcap assembly (GenBank: GCA_009819655.1), we made a mask ile in FASTA o ma o each condi ion using BEDTools mask as a . Using he phased and pola ized VCF, ecombina ion map, and he mask, we an Rela e o in e genealogies wi h mu a ion a e o 4 . 6 × 10 −9 and e ec i e popula ion size o 500,000. We in e ed demog aphy om he genealogies using Rela eCoalescenceRa e p og am o Rela e wi h mu a ion a e o 4 . 6 × 10 −9 and i e imes o i e a ions. The es ima es o his o ical e ec i e popula ion size we e ob ained as he in e se o he in e ed coalescence a e o each popula ion wi h escaling o 152 ime by gene a ion ime o 2 yea s (Delmo e e al., 2020). The CCR was ob ained by di iding he be ween-popula ion coalescence a e wi h he a e age wi hin-popula ion coalescence a e. Acknowledgmen s This wo k was suppo ed by he Max Planck Socie y (Max Planck Resea ch G oup g an MFFALIMN0001 o ML), and he DFG (p ojec Na 05 wi hin SFB 1372 – Magne o ecep ion and Na iga ion in Ve eb a es (395940726) o ML). We hank Julien Du heil, Linda Oden hal- Hesse, and Die ha d Tau z o eedback. Da a a ailabili y Sc ip s used o he simula ions and inpu da a, p ocessed ou pu da a and sc ip s o he empi ical analyses a e ound in Zenodo (h ps://doi.o g/10.5281/zenodo.10613446). Con lic o in e es The au ho s decla e no con lic o in e es . 153