Augus ijnen e al., Sci. Ad . 10, eadl0989 (2024) 17 Ap il 2024
Science Ad AnceS | ReSeARch AR icle
1 o 13
EVOLUTIONARY BIOLOGY
A mac oe olu iona y ole o ch omosomal usion and
ission in E ebia bu e lies
Hannah Augus ijnen1*, Li io Bä sche 1, Ma in Cesanek2, Tina in Chkha ish ili3, Vlad Dincă4,
Gio gi Iankosh ili3, Ko a Ogawa5,6, Roge Vila7, Se aina Klop s ein8,9, Ju iaan M. de Vos1,
Kay Lucek1,10*
The impac o la ge- scale ch omosomal ea angemen s, such as usions and issions, on specia ion is a long-
s anding conund um. We assessed whe he bu s s o change in ch omosome numbe s esul ing om ch omo-
somal usion o ission a e ela ed o inc eased specia ion a es in E ebia, one o he mos species- ich and
ka yo ypically a iable bu e ly g oups. We es ablished a genome- based phylogeny and used s a e- dependen
bi h- dea h models o in e ajec o ies o ka yo ype e olu ion. We demons a ed ha a es o anagene ic ch o-
mosomal changes (i.e., along phylogene ic b anches) exceed cladogene ic changes (i.e., a specia ion e en s), bu ,
when cladogene ic changes occu , hey a e mos ly associa ed wi h ch omosomal issions a he han usions. We
ound ha he ela i e impo ance o usion and ission di e s among E ebia clades o di e en ages and ha es-
pecially in younge , mo e ka yo ypically di e se clades, specia ion is mo e equen ly associa ed wi h cladoge-
ne ic ch omosomal changes. O e all, ou esul s imply ha ch omosomal usions and issions ha e con as ing
mac oe olu iona y oles and ha la ge- scale ch omosomal ea angemen s a e associa ed wi h bu s s o species
di e si ica ion.
INTRODUCTION
The e olu ion o ba ie s o gene low is a c i ical equi emen o he
p og ess o specia ion (1). Al hough se e al ba ie s may con ibu e
o he p ocess, hei ela i e impo ance o en emains unknown, es-
pecially a a mac oe olu iona y scale (2). Ch omosomal specia ion
heo y sugges s ha la ge- scale ch omosomal ea angemen s, such
as usions and issions, a e able o p omo e specia ion. They may allow
o he buildup o gene ic incompa ibili ies be ween lineages ei he by
causing hyb id dys unc ion (3, 4) o by supp essing ecombina ion in
ea anged sec ions o he genome (5–7). The ele ance o ch omo-
somal specia ion has been c i icized because o he expec ed “unde -
dominance” o ch omosomal ea angemen s, whe eby s ong hyb id
i ness disad an ages ensu e ha ixa ion o no el ka yo ypes is di i-
cul , ende ing ba ie o ma ion unlikely (5, 6). Con e sely, i he e -
ec s o ch omosomal ea angemen s on hyb ids we e mino , hen
ixa ion would be possible, bu he esul ing ba ie s would emain
shallow (5, 6). These heo ies we e de eloped o monocen ic ch o-
mosomes, whe eas he ch omosomes o se e al majo o ganismal
g oups, such as bu e lies and sedges (8), a e holocen ic, i.e., hey
ha e cen ome e- like s uc u es sp ead ac oss hei ch omosomes
a he han concen a ed in a single cen ome e (9).
Holocen ic ch omosomes may be mo e likely o o e come he
a o emen ioned unde dominance pa adox, as ea anged ch omo-
somes can e ain kine ocho e unc ionali y and so lead o he c ea ion
o meio ic mul i alen s ha a e only sligh ly dele e ious in hyb ids
(9–11). C osses be ween closely ela ed holocen ic species wi h di -
e en ka yo ypes may emain iable (11) and do no necessa ily esul
in ep oduc i e isola ion (10, 12), al hough his may no be he case
when sex ch omosomes a e ea anged (13). In addi ion, some holo-
cen ic clades ha e e ol ed mechanisms o acili a e p ope ch omo-
some seg ega ion e en when ch omosomes a e ea anged (14, 15),
which has been sugges ed o p omo e ch omosomal specia ion (9).
Empi ical e idence o a link be ween specia ion and ch omosomal
ea angemen s, especially o ch omosomal usions and issions, is
spa se o bo h mono- and holocen ic clades (8, 16). Howe e , he
ac ha many holocen ic g oups wi hin plan s and in e eb a es a e
e y species- ich sugges s ha ch omosomal ea angemen s could
ha e d i en di e si ica ion in some o hem (9, 17).
Lepidop e a is one o he la ges axonomic g oups wi h holocen-
ic ch omosomes, comp ising mo e han 160,000 species o bu e -
lies and mo hs (18). While some gene a wi hin Lepidop e a a e
ex emely di e se in ch omosome numbe s, some imes di e ing by a
coun o mo e han 200 e en wi hin a single genus (19, 20), mos o h-
e s ha e conse ed ch omosome numbe s, o en close o he in e ed
ances al ka yo ype [haploid numbe (n)=31]. Compa a i e phylo-
gene ic analyses indica e a posi i e associa ion be ween he a e o
specia ion and ka yo ype e olu ion o se e al o he mos ka yo ypi-
cally di e se bu e ly gene a (16).
E ebia is one o he mos speciose o all Palea c ic bu e ly gene a,
consis ing o a ound 90 o 100 species ha mainly inhabi cold moun-
ainous egions, wi h he majo i y o di e si y ound in Eu ope, whe e
closely ela ed species o en o m na ow zones o seconda y con ac
wi h li le gene low (21, 22). No ably, E ebia is also one o he gene a
wi h he highes known ka yo ype di e si y among bu e lies (23),
al hough his di e si y di e s be ween clades wi hin he genus. Mos
ka yo ypic a ia ion can be ound in he compa a i ely young ynda-
us clade (n=8 o 51) (Fig.1B and ableS1), whe e phylogene ic ela-
ionships ha e emained unclea (24). As ynda us clade species a e
mos ly ound in moun ainous egions (23), glacial ange expansions
1depa men o en i onmen al Science, Uni e si y o Basel, 4056 Basel, Swi ze -
land. 2Slo ak en omological Socie y, Slo ak Academy o Sciences, B a isla a 1,
Slo akia. 3ins i u e o ecology, ilia S a e Uni e si y, bilisi 0162, Geo gia. 4ecology
and Gene ics Resea ch Uni , Uni e si y o Oulu, 90570 Oulu, Finland. 5Facul y o
Social and cul u al S udies, Kyushu Uni e si y, Fukuoka 819- 0395, Japan. 6insec
Sciences and c ea i e en omology cen e , Kyushu Uni e si y, Fukuoka 819- 0395,
Japan. 7ins i u de Biologia e olu i a (cSic- Uni . Pompeu Fab a), 08003 Ba celona,
Spain. 8ins i u e o ecology and e olu ion, Uni e si y o Be n, 3012 Be n, Swi ze land.
9li e Sciences, na u al his o y Museum Basel, 4051 Basel, Swi ze land. 10ins i u e o
Biology, Uni e si y o neuchâ el, 2000 neuchâ el, Swi ze land.
*co esponding au ho . email: hannah. augus ijnen@ uni . ch (h.A.); kay. lucek@
unine. ch (K.l.)
copy igh © 2024 he
Au ho s, some igh s
ese ed; exclusi e
licensee Ame ican
Associa ion o he
Ad ancemen o
Science. no claim o
o iginal U.S.
Go e nmen Wo ks.
dis ibu ed unde a
c ea i e commons
A ibu ion
noncomme cial
license 4.0 (cc BY- nc).
Downloaded om h ps://www.science.o g a Cen o de Biologia Molecula Se e o Ochoa on No embe 11, 2024
Augus ijnen e al., Sci. Ad . 10, eadl0989 (2024) 17 Ap il 2024
Science Ad AnceS | ReSeARch AR icle
2 o 13
and con ac ions [e.g., (25)] may ha e led o popula ion bo lenecks
and so p omo ed he ixa ion o he clade’s la ge a ie y o ea anged
ka yo ypes h ough d i , as has been ound o o he bu e lies (26).
He e, we le e aged ka yo ype di e si y ac oss E ebia o es o i s
ole in species di e si ica ion. Speci ically, we i s quan i ied he
o e all impac o ch omosomal usion and ission on di e si ica ion
in E ebia using phylogenomic in e ence and Bayesian s a e- dependen
bi h- dea h models. We hen assessed he associa ion be ween di e -
si ica ion and ch omosomal usion and ission ac oss E ebia clades o
di e en ages and ka yo ype di e si y. We hypo hesized ha ch o-
mosomal specia ion has played a subs an ial ole in he di e si ica-
ion o E ebia and ha mo e ka yo ypically di e se clades show an
Fig. 1. Sample dis ibu ion and ela ionships wi hin he Palea c ic genus E ebia. (A) Map o he no he n hemisphe e indica ing sampling loca ions o E ebia specimens used,
colo ed by clade. (B) Known ch omosome numbe s o E ebia species, g ouped by clade [ om (21)]. (C) ime- calib a ed phylogeny o E ebia calcula ed in M Bayes. he ossil
V. ame indica was used o calib a e he oo o he ee (i.e., s em lineage o Sa y ini), while he ossil L. co bie i was placed a he c own node o Sa y ini. clade names a e based on
(21) wi h he excep ion o medusa and plu o. Fo each clade a ep esen a i e pheno ype is shown. F om op o bo om, hese a e E ebia medusa, E ebia plu o, E ebia p onoe, E ebia
epiph on, E ebia ynda us, E ebia ligea, E ebia magdalena, E ebia embla, and E ebia pa menio.
Downloaded om h ps://www.science.o g a Cen o de Biologia Molecula Se e o Ochoa on No embe 11, 2024
Augus ijnen e al., Sci. Ad . 10, eadl0989 (2024) 17 Ap il 2024
Science Ad AnceS | ReSeARch AR icle
3 o 13
inc eased signal o ch omosomal changes and highe associa ed spe-
cia ion a es.
RESULTS
The di e si ica ion o E ebia
Al hough app oxima ely 90 o 100 E ebia species a e ecognized, he
phylogene ic esolu ion o o me s udies on E ebia was limi ed, espe-
cially o e olu iona y younge axa, as hese s udies included ei he
ew genes (21) o ew species (27). Using whole- genome esequencing
da a o 82 E ebia species, 57% o which a e ka yo yped (Fig.1, A and
B, and ableS1), we cons uc ed a nea ly ully esol ed species- le el
coalescen - based phylogeny based on 2920 indi idual maximum like-
lihood (ML) gene ees ( ig.S1). B anch suppo was e y high o e all
(>0.9 ASTRAL consensus o 96.5% o all nodes), and we u he
alida ed he ela ionships among axa ollowing (28) ( igs.S2 o S6).
We used he esul ing opology o cons ain a molecula clock da ing
analysis in M Bayes using ossil calib a ions o he s em and c own
nodes o Sa y ini and a subse o genes selec ed o minimal missing
da a, especially among ou g oups (Fig.1C). We con i med he mono-
phyly o p e iously de ined (21) clades ynda us [2.41 million yea s
(Ma) ago; 95% highes pos e io densi y (HPD) in e al, 1.18 o
3.43 Ma ago], epiph on (2.47 Ma ago; 95% HPD, 1.49 o 4.21 Ma ago),
and p onoe (3.10 Ma ago; 95% HPD, 2.07 o 4.98 Ma ago), as well as
he classic axonomic clades (29) ligea (3.60 Ma ago; 95% HPD, 2.14
o 5.81 Ma ago), medusa (3.08 Ma ago; 95% HPD, 1.87 o 5.38 Ma
ago), and plu o (2.94 Ma ago; 95% HPD, 1.64 o 4.93 Ma ago). We es-
ima ed he age o E ebia o be 20.16 Ma old (95% HPD, 12.98 o 31.63
Ma), wi h he i s majo spli be ween he mos ly non- Eu opean
embla, magdalena, and pa menio clades and all o he E ebia a 7.14
Ma ago (95% HPD, 4.62 o 11.77 Ma ago).
Cladogenesis and ch omosomes
We i ed a Ch omoSSE model (30) o decompose a es o ch omo-
somal usion and ission in o hei anagene ic (ch omosomal change
along a b anch) and cladogene ic (ch omosomal change a a specia-
ion e en ) componen s (Figs.2 and 3). Ch omoSSE in e s anagene ic
pa ame e s h ough a con inuous- ime Ma ko p ocess based on a Q
ma ix, which desc ibes ins an aneous a es o change in e ms o
ch omosome numbe s (30). Cladogene ic pa ame e s, de ined as
a es o cladogenesis (i) wi h ch omosomal usions, (ii) wi h ch omo-
somal issions, and (iii) wi hou ch omosomal change, a e es ima ed
ia a bi h- dea h p ocess (30). Ou models, unless o he wise speci-
ied, a e based on he hypo hesis ha ch omosomal usions and is-
sions e ol e bo h ana- and cladogene ically.
We ound ha ch omosomal changes h ough anagene ic usion
(0.636 e en s pe species pe million yea s; 95% HPD, 0.134 o 1.118)
occu ed a a highe a e han anagene ic ission (0.212 e en s pe
species pe million yea s; 95% HPD, 0.002 o 0.550; Fig.3 and a-
bleS2). Howe e , mos in e ed specia ion (cladogene ic) e en s in
E ebia coincide wi h ch omosomal change, ei he wi h cladogene ic
ch omosomal usion (0.196 e en s pe species pe million yea s; 95%
HPD, 0.074 o 0.340) o wi h cladogene ic ission (0.328 e en s pe
species pe million yea s; 95% HPD, 0.056 o 0.561; Fig.3), while he
specia ion a e wi hou ch omosomal change was lowe (0.127 e en s
pe species pe million yea s; 95% HPD, 0.001 o 0.320). The ela i e
ex inc ion a e ac oss E ebia was 0.273 e en s pe species pe million
yea s (95% HPD: 0.006–0.511) and o al specia ion (summed specia-
ion a es o cladogene ic usion, cladogene ic ission, and wi hou
ch omosomal change) was 0.651 e en s pe species pe million yea s
(95% HPD, 0.471 o 0.839).
We in e ed ha he ka yo ype a he oo o E ebia was likely
n=22 o 23 (Fig.2). To explo e he in luence o he oo alue, we
epea ed his analysis cons aining he oo ka yo ype o n= 29,
which is he modal ka yo ype o E ebia and he ch omosome numbe
o i s closes ela i es (31). We ob ained e y simila pa ame e alues
whe he we es ima ed o cons ained he oo alue, al hough ixing
he oo a n=29 led o a highe es ima ed a e o anagene ic usion
( ableS3). Simila esul s we e ob ained when cons aining he oo
alue o n=29 o 33 ( ableS3), he eby allowing i o a y a ound he
ances al ka yo ype o all bu e lies (32). Consequen ly, by es ima ing
he oo alue, we ob ained conse a i e es ima es.
To examine he scena io whe e ch omosomal usions and is-
sions would no con ibu e o specia ion in E ebia, we i ed an al-
e na i e Ch omoSSE model whe e specia ion (cladogenesis) was
cons ained o be un ela ed o ch omosomal change, wi h usions
and issions e ol ing only by anagenesis ( able S3). Ou ini ial
model p o ides a be e i o he da a han his al e na i e ( a-
ble S4), again con i ming ha usions and issions e ol e bo h
cladogene ically and anagene ically.
Clade- speci ic ch omosomal changes
To es whe he ana- o cladogene ic ch omosomal change a es
would be highe in clades ha show a highe ka yo ype di e si y
(Fig.1B), we an Ch omoSSE o he six clades wi h su icien ch o-
mosome coun da a and species (Figs.1B and 4A and ableS5). We
ound ha he a es o bo h ana- and cladogene ic ch omosomal
change di e ed ac oss clades unde Ch omoSSE: Anagene ic usions
di e ed signi ican ly among clades (K uskal- Wallis es , χ25=3303.5,
P<0.001; pe mu a ion es wi h 1000 i e a ions, P<0.001; Fig.4B),
wi h pos e io dis ibu ion means anging om compa a i ely low,
0.084 (ligea), o mode a e, 0.296 (p onoe), e en s pe species pe mil-
lion yea s. This was also ue o anagene ic issions (χ25=1582.6,
P<0.001; pe mu a ion es , P<0.001; Fig.4C), wi h mean a es be-
ween 0.065 (ligea) and 0.168 (p onoe) e en s pe species pe million
yea s. To al anagene ic ch omosomal change, de ined as he sum o
a es o anagene ic usions and issions, anged om 0.148 (ligea) o
0.464 (p onoe) e en s pe species pe million yea s, which, o all
clades, is less han he 0.848 (95% HPD, 0.420 o 1.295) ound o he
o e all Ch omoSSE analysis o E ebia ( ableS2).
The a e o cladogene ic ch omosomal usion showed a highe de-
g ee o di e en ia ion among clades han o he ana- o cladogene ic
pa ame e s (χ25=6958.5, P<0.001; pe mu a ion es , P<0.001;
Fig.4F), anging om 0.327 (epiph on) o 1.263 ( ynda us) e en s
pe species pe million yea s. The a e o cladogene ic ch omosomal
ission was likewise a iable be ween clades (χ25=2365.5, P<0.001;
pe mu a ion es , P<0.001; Fig.4G), anging be ween 0.113 (ligea)
and 0.594 ( ynda us) e en s pe species pe million yea s. The
summed a es o cladogene ic usion and ission anged om 0.538
(ligea) o 1.856 ( ynda us) e en s pe species pe million yea s. These
a es we e gene ally highe han o o e all E ebia (0.524 e en s pe
species pe million yea s), bu he pos e io dis ibu ions o mos
clades o e lap wi h he 95% HPD (0.304 o 0.783) o he o e all pa-
ame e ( ableS2). The excep ion was he ka yo ypically di e se yn-
da us clade (95% HPD, 0.906 o 2.898), whose a e exceed ha o
E ebia. While he pa ame e alues we e e y simila be ween some
clades, hei p opo ional con ibu ions o a es o specia ion di e
(Fig.4). Fo example, medusa, plu o, and epiph on show simila a es
Downloaded om h ps://www.science.o g a Cen o de Biologia Molecula Se e o Ochoa on No embe 11, 2024
Augus ijnen e al., Sci. Ad . 10, eadl0989 (2024) 17 Ap il 2024
Science Ad AnceS | ReSeARch AR icle
4 o 13
Fig. 2. Summa y o he ch omosome e olu ion model o E ebia, implemen ed in Ch omoSSE. es ima ed ances al ch omosome numbe s o E ebia, in e ed using
he phylogenomic opology o ig.S1. ch omosome numbe s a e indica ed p opo ionally by he colo o he pie cha s a he b anch nodes. he pie cha s a he “shoul-
de s” o each node ep esen he in e ed ch omosomal s a e immedia ely a e a specia ion e en . he wo in e ed ch omosome numbe s wi h highes pos e io dis i-
bu ion o he ances o o each clade a e depic ed a op o below he node ha s a s ha clade. Fo deepe nodes wi hin he ee, he econs uc ed ances al cha ac e
s a es a e likewise p esen ed. Fo each ex an species in he phylogeny, he ka yo ype is shown be o e he name. When he ka yo ype o a species is no known, he spe-
cies is deno ed wi h a “?.” clades a e named as in Fig.1. he pos e io p obabili ies o econs uc ed ances al ch omosome s a es a e isualized in ig.S7.
Downloaded om h ps://www.science.o g a Cen o de Biologia Molecula Se e o Ochoa on No embe 11, 2024
Augus ijnen e al., Sci. Ad . 10, eadl0989 (2024) 17 Ap il 2024
Science Ad AnceS | ReSeARch AR icle
5 o 13
o cladogene ic ch omosomal issions bu hey accoun o 19.8, 18.9,
and 30.0% o he o al cladogene ic e en s, espec i ely.
Specia ion a es (cladogenesis) wi hou ch omosomal change di -
e ed among clades (χ25= 4144.1, P< 0.001; pe mu a ion es ,
P<0.001; Fig.4E) anging om 0.145 (p onoe) o 0.423 (ligea) e en s
pe species pe million yea s bu ne e accoun ed o a majo i y o he
o al specia ion in a clade ( hough nea ly so in ligea; Fig.4, E and F,
and ableS2).
Rela i e ex inc ion a es we e highes o ynda us a 0.745 and
a ied be ween clades (χ25= 5089.6, P< 0.001; pe mu a ion es ,
P<0.001; Fig.4D), wi h he lowes a es o p onoe (0.197) and ligea
(0.252). To al specia ion ela ed di ec ly o a es o cladogene ic e en s
and likewise a ied be ween clades (χ25=5807.1, P<0.001; pe mu a-
ion es , P<0.001; Fig.4H), wi h ynda us ha ing he highes deg ee
o o al specia ion (2.191; 95% HPD, 1.283 o 3.363 e en s pe species
pe million yea s). The co ela ion be ween ela i e ex inc ion and o-
al specia ion was 0.38 ( 1,14998=50.7, P<0.001).
We also epea ed his analysis while equalizing he maximum
ch omosome numbe allowed in each clade (n=56) o minimize he
po en ial e ec s o di e ing Q ma ices be ween models. The esul ing
pa ame e es ima es a e la gely simila o he models whe e maxi-
mum ch omosome numbe s we e allowed o be ailo ed o each clade,
wi h all 95% HPD o e lapping ( ableS6). Howe e , unning all clades
wi h 56 maximum ch omosomes does lead o lowe es ima es o o al
specia ion ( ableS6) and o lowe es ima es o cladogene ic ch omo-
somal usion. None heless, in all cases, he model wi h 56 maximum
ch omosomes was no p e e ed o e he o iginal one ( ableS7).
As o he o e all phylogeny, we i ed al e na i e Ch omoSSE
models o each clade, whe e usions and issions we e cons ained
o be only anagene ic. Fo epiph on, medusa, plu o, and, especially,
ynda us, he o e all model wi h combined cladogene ic and anagen-
e ic e olu ion o ch omosome numbe s was p e e ed ( ableS8). In
p onoe, whe e anagene ic changes a e mos equen o all clades ( a-
bleS2), bu cladogene ic usions a e likewise common, he e idence is
mixed, wi h Akaike’s in o ma ion c i e ion h ough Ma ko chain
Mon e Ca lo (AICM) p e e ing he model whe e specia ion does no
ela e o ch omosomal change and Bayes ac o s (BFs) inding no di -
e ence be ween model i s ( ableS8). Fo ligea, he clade wi h he leas
ch omosomal a ia ion (Fig.1B and ableS1), he al e na i e model
was p e e ed ( ableS8).
A ale o wo models
To assess o which deg ee he combined in e ence o specia ion and
ex inc ion a es in Ch omoSSE migh a ec ou es ima ions o cla-
dogene ic and anagene ic pa ame e s, we alida ed he esul s o
Ch omoSSE by ca ying ou simila analyses in Ch omEVOL (33)
ollowing (30), which allows o model usion- and ission- associa ed
cladogene ic changes bu does no include an ex inc ion pa am-
e e . We expec ed po en ial di e ences be ween he models in
clades wi h highe ex inc ion and specia ion a es. Al hough he
esul s o Ch omoSSE and Ch omE ol we e la gely simila o
nea ly all clades ( ableS9, Fig.4, and ig.S8), he ma ginal likeli-
hoods and AICMs o Ch omE ol models we e gene ally highe ,
likely owing o a la ge pa ame e space o Ch omoSSE compa ed
o Ch omEVOL. The di e ence in ma ginal likelihood was mo e
p onounced o he o e all ee han o indi idual clades (di e ence
in BFs anging om o 10.002 (plu o) o 40.411 (medusa); ablesS4
and S10). In con as o he o he E ebia clades, we ound ha he
Ch omoSSE model leads o a be e ela i e i o he da a han he
Ch omEVOL one (BF=8.857, AICM=34.025) o ynda us,
he clade wi h he highes in e ed ex inc ion a es (Fig.4D). An ex-
plana ion o he di e ence be ween Ch omoSSE and Ch omEVOL
could be ha simple models wi h ewe pa ame e s (Ch omEVOL)
esul in highe ma ginal likelihoods han mo e complex models
0.00.3 0.60.9 1.21.5 1.82.1
E en s pe species pe Ma
To al specia ion
Rela i e ex inc ion
Specia ion (cladogenesis)
wi hou ch omosomal change
Specia ion (cladogenesis)
wi h ch omosomal usion
Specia ion (cladogenesis)
wi h ch omosomal ission
Anagene ic
ch omosomal usion
Anagene ic
ch omosomal ission
Fig. 3. A summa y o he pos e io dis ibu ions o in e ed Ch omoSSE pa ame e s o he o e all E ebia analysis. Shown a e he pos e io densi ies pe pa ame e ,
he aw da a poin s, and he 95% hPd in e al as p esen ed by he bold black line unde nea h each densi y plo ( ableS2). All pa ame e a es a e exp essed in e en s pe
species pe million yea s (Ma).
Downloaded om h ps://www.science.o g a Cen o de Biologia Molecula Se e o Ochoa on No embe 11, 2024

Augus ijnen e al., Sci. Ad . 10, eadl0989 (2024) 17 Ap il 2024
Science Ad AnceS | ReSeARch AR icle
6 o 13
0.00
0.25
0.50
0.75
1.00
0.0 0.4 0.8 1.2 1.8 2.0
E en s pe species pe Ma
A
Cladogenesis wi hou ch omosomal change
E
Ch omoSSECh omEVOL
Cladogene ic ch omosomal usions
F
Ch omoSSECh omEVOL
0.0 0.25 0.50 0.75 1.00 1.25
E en s pe species pe Ma
Cladogene ic ch omosomal issions
G
Ch omoSSECh omEVOL
0.0 0.5 1.0 1.52.0 2.5 3.0
H
0.0 0.5 1.0 1.52.0 2.5 3.03.5 4.0
To al specia ion
E en s pe species pe Ma
0.0 0.2 0.4 0.6 0.8 1.0
Rela i e ex inc ion
D
E en s pe species pe Ma
Anagene ic ch omosomal usions
B
Ch omoSSE Ch omEVOL
0.0 0.2 0.4 0.6 0.8
medusa
plu o
p onoe
epiph on
ynda us
ligea
C
Anagene ic ch omosomal issions
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Ch omoSSE Ch omEVOL
E en s pe species pe Ma
E en s pe species pe Ma
Rela i e e en s pe species pe Ma
Rela i e e en s pe species pe Ma
Rela i e e en s pe species pe Ma
0.00
0.25
0.50
0.75
1.00
0.00
1.00
2.00
0.00
0.25
0.50
0.75
1.00
0.00
0.25
0.50
0.75
1.00
E en s pe species pe Ma
E en s pe species pe Ma
E en s pe species pe Ma
Fig. 4. Summa y o ana- and cladogene ic ances al ch omosome es ima ions o six E ebia clades. (A) O e iew o he phylogeny o e ebia wi h ele an clades
highligh ed. (B o H) dis ibu ions o he pos e io p obabili y es ima es o (B) anagene ic usions, (c) anagene ic issions, (d) ela i e ex inc ion a es, (e) cladogene ic
changes un ela ed o ch omosomal change, (F) cladogene ic ch omosomal usions, (G) cladogene ic ch omosomal issions, and (h) o al specia ion a es. Fo (B) and (c)
and (e) o (G), compa isons be ween s anda dized pa ame e es ima ions o clade speci ic s a e- dependen bi h- dea h models (ch omoSSe) and bi h- dea h indepen-
den models (ch ome Ol) a e shown. each do indica es he mean o he pos e io p obabili y space. All pa ame e a es a e exp essed in e en s pe species pe million
yea s. An al e na i e isualiza ion, pe clade, no pe pa ame e , is gi en in ig.S8.
Downloaded om h ps://www.science.o g a Cen o de Biologia Molecula Se e o Ochoa on No embe 11, 2024
Augus ijnen e al., Sci. Ad . 10, eadl0989 (2024) 17 Ap il 2024
Science Ad AnceS | ReSeARch AR icle
7 o 13
which may su e om o e pa ame e iza ion, pa icula ly in ela-
i ely small da ase s. In clades wi h low o medium in e ed ela i e
ex inc ion a es (e.g., medusa), he mo e complex Ch omoSSE
model was no p e e ed o e he ela i ely simple Ch omE ol. In
ynda us, by con as , i appea s ha modeling ex inc ion in an-
dem wi h ch omosomal e olu ion as Ch omoSSE was designed o
do was su icien o o e come he po en ial issue o o e pa ame-
e iza ion.
Because Ch omE ol es ima es cladogene ic pa ame e s as p o-
po ions o he o al a e o cladogene ic change, cladogene ic pa am-
e e s om he Ch omoSSE analysis we e di ided by he mean o al
specia ion a e (Fig.4G), allowing us o compa e a es be ween mod-
els. Fo he en i e ee, a es o cladogenesis wi hou ch omosomal
change, cladogene ic usions and cladogene ic issions o e lapped in
hei pos e io dis ibu ions, bu pos e io means di e ed be ween
he models (K uskal- Wallis and pe mu a ion es s, all P< 0.001;
ig.S9), in cong uence wi h he ela i ely high di e ence be ween
BFs. Howe e , he a es o ch omosomal ission–associa ed specia-
ion emained highe han hose o ch omosomal usion unde bo h
models ( ableS9). Ra es o anagene ic issions showed a highe o e -
lap in pos e io dis ibu ion and did no di e signi ican ly in means
(χ25=0.017, P=0.896; pe mu a ion es , P=0.208; ig.S9). Ra es
o anagene ic usions did di e be ween models (χ25= 2925.2,
P<0.001; pe mu a ion es , P<0.001; ig.S9). This di e ence was
mos likely d i en by ynda us clade (see Fig.4B) and may be ela ed
o i s high ela i e ex inc ion a es.
To assess he e ec s o missing ka yo ype da a on ou in e ences,
we an Ch omoSSE and Ch omEVOL again, using simila pa ame-
e s, on a da ase limi ed o axa o which ch omosome numbe s a e
a ailable ( ig.S10). The esul s o Ch omEVOL and Ch omoSSE o
his da ase we e highly cong uen wi h ou o e all E ebia analyses
( ableS3). Fo Ch omoSSE, howe e , ch omosomal issions ended o
be assigned mo e as anagene ic ( ableS3), which is consis en wi h
analyses using less comple e phylogenies (34).
DISCUSSION
Compa ed o o he ea angemen s, such as in e sions, much mo e
emains unknown abou he ole o ch omosomal usion and ission in
specia ion (8). Theo y p edic s ha usions and issions could ha e
majo impac s on he specia ion p ocess (3, 5), pa icula ly in species
whose ch omosomes a e holocen ic and he e o e may no ini ially be
as s ongly a ec ed by he unde dominance pa adox (8, 11, 15). He e,
we ound phylogenomic e idence o an associa ion o ch omosomal
usion and ission wi h highe specia ion a es a mac oe olu iona y
scale, p o iding indi ec e idence o hei in ol emen du ing specia-
ion. Ch omosome- associa ed cladogene ic e en s we e p e alen in
E ebia and we e associa ed wi h highe a es o specia ion han
specia ion wi hou ch omosomal change (Fig.2B), ou pe o ming al-
e na i e models wi hou ch omosomal change ( ableS4). Such clado-
gene ic e en s coincide wi h spli s be ween lineages and, in heo y,
ep esen cases whe e ch omosomal ea angemen s may ha e a mo e
causa i e in ol emen in specia ion (30). Al e na i ely, cladogene ic
ch omosomal e en s could ep esen a di ec consequence o specia-
ion i he buildup o ep oduc i e isola ion sepa a es di e ging ka yo-
ypical lineages wi hin he same popula ion (8). Howe e , a causal ole
o ch omosomal usions and issions may be especially likely when
hey hemsel es ac as in insic ba ie s o gene low, e.g., by esul ing
in hyb id dys unc ion (3, 4), by physically b inging oge he si es
unde selec ion (35), o i sex ch omosomes a e in ol ed (13). Ch o-
mosomal usions and issions may he e o e accompany apid di e si-
ica ion o lineages in o species adia ions, e.g., (36), o ins ance, he e,
in E ebia (21).
F om a phylogene ic pe spec i e, ch omosome- associa ed specia-
ion should mainly conce n cladogenesis, i.e., he spli ing o an an-
ces al species in o wo new lineages. Howe e , ecen heo y p edic s
ha anagene ic ch omosomal changes could also con ibu e o spe-
cia ion by g adually building up di e en ia ion in ka yo ypes o e
ime, po en ially leading o inc eased ep oduc i e isola ion (8). Mos
changes in ch omosome numbe in he o e all E ebia phylogeny ap-
pea o ha e occu ed anagene ically (Fig.2). Ch omosomal usion
and ission could he e o e ac as ba ie s o gene low, simila o o h-
e ea angemen s (35), including in e sions, e.g., (37). Fusion and
ission could supp ess ecombina ion wi hin o a ound ea anged
sec ions o he genome be ween hyb ids (38) and so p e en he
b eakup o linkage disequilib ium be ween locally co- adap ed genes
(5), p omo ing he g adual accumula ion o di e ences h ough ime
(39, 40). Specia ion may hen be comple ed h ough addi ional p o-
cesses (41), such as ein o cemen upon seconda y con ac , as has
been sugges ed o o he bu e lies (42).
Ou analyses sugges ha he e olu iona y impac s o ch omo-
somal usions and issions di e , as usions we e mo e likely o be ana-
gene ic ac oss he en i e E ebia phylogeny. Con e sely, issions we e
mo e likely o be associa ed wi h cladogene ic e en s, leading o high-
e cladogene ic specia ion a es when issions a e in ol ed (Fig.2).
Reduc ions in ch omosome numbe h ough usion e en s we e mo e
common han issions in E ebia, as well as o he g oups o bu e lies
(16) and holocen ic o ganisms, e.g., (43). Ch omosomal issions, al-
hough a e , could he e o e be associa ed wi h highe specia ion
a es i hey mo e o en esul in somewha dele e ious meio ic mul i-
alen s (16, 44). Mo e s udies a e needed o con i m his hypo hesis,
and his e ec may be pa ially mi iga ed in Lepidop e a due o hei
holocen ic ch omosomes (15, 45). Sho e ch omosomes ha e also
been shown o ha e an inc eased likelihood o be in ol ed in usions
(32), sugges ing ha longe ch omosomes may be mo e e olu iona y
s able. The appa en highe e olu iona y s abili y o used ch omo-
somes and he ins abili y o issions could explain why clades wi h
highe ch omosome numbe s show highe cladogene ic specia ion
a es in E ebia (Figs.2 and 4) and o he bu e lies (16). The highe
o e all deg ee o anagene ic usions obse ed in his s udy migh con-
ibu e o he buildup o ep oduc i e isola ion by g adually educing
ecombina ion a es, as has been ound in mice (46) and o he bu -
e lies (47). Genome- wide c ossing- o e a es co espondingly seem
o be conside ably highe when ch omosomes a e sho o e ol e o
become sho e (48), bu see (49).
While he e a e some examples o s udies ocusing on he mac o-
e olu iona y impac o ch omosomal usions and issions, as well as o he
ea angemen s (50–52), unde s anding pa e ns o ch omosome-
associa ed specia ion may equi e addi ional analysis a a ine axo-
nomic scale (43). He e, we assessed he ela i e con ibu ions o
ch omosomal usion and ission be ween clades o di e ing ages and
ka yo ype di e si y, o assess he impac o hese ea angemen s on
specia ion. We iden i ied a con inuum o ch omosome- associa ed spe-
cia ion a es, anging om he young, ka yo ypically e y di e se yn-
da us clade ha showed he highes a es o specia ion ela ed o
ch omosomal change (Fig.4 and ig.S8) o ka yo ypically mo e con-
se ed clades, such as ligea, whe e he p opo ion o specia ion un e-
la ed o ch omosomal change was much highe (Fig.4E) and a model
Downloaded om h ps://www.science.o g a Cen o de Biologia Molecula Se e o Ochoa on No embe 11, 2024
Augus ijnen e al., Sci. Ad . 10, eadl0989 (2024) 17 Ap il 2024
Science Ad AnceS | ReSeARch AR icle
8 o 13
wi hou ch omosomal specia ion was e en p e e ed ( ableS8). O he
clades ell along his con inuum, whe e a es o specia ion wi h ch o-
mosomal usion we e gene ally high and anagenesis appea s o be less
impo an a he subclade le el han o he o e all ee (Fig.4). No a-
bly, he clades plu o and p onoe, which a e second o ynda us in ch o-
mosome a iabili y (Fig. 1B), also show he lowes p opo ion o
specia ion ha is no d i en by ch omosomal usion and ission
(Fig.4E), al hough hey emain behind ynda us in absolu e alues
( ableS2).
The high ka yo ype di e si y and associa ed high specia ion a es
o he ynda us clade may, in pa , be explained by hei ecology. Spe-
cies o he ynda us clade occu almos exclusi ely in Alpine a eas
(23), whe eas o he E ebia clades a e ecologically mo e di e se (24,
53). Du ing glacial cycles, epea ed ange expansions and con ac-
ions ac oss ela i ely small geog aphic a eas ha e caused popula ion
subdi isions o many E ebia species, e.g., (54, 55), including he yn-
da us clade (25). Fo he la e , d i and o he s ochas ic p ocesses
could ha e p omo ed he ixa ion o no el ch omosomal ea ange-
men s (56). Na u al selec ion may also ha e played a ole in he ixa-
ion o usions, al hough i is no known o which deg ee (26). Fu he
in es iga ion o bu e ly lineages wi h widely a ying ch omosome
numbe s (16, 57) will be equi ed o de e mine he unde lying mech-
anisms o he usions and issions ha a ec specia ion (8).
While s a e- dependen specia ion and ex inc ion (SSE) models al-
low o unp eceden ed phylogene ic insigh s in o mac oe olu iona y
aspec s o specia ion (58–60), hei eliabili y has been pa ially ques-
ioned (61). They may, o example, su e om excess alse- posi i e
a es because shi s in di e si ica ion a es ac oss he phylogeny may
be assigned o a s udied ai in an SSE model e en when hese shi s
a e caused by an unde ec ed and un ela ed ai (61, 62). This could
lead o an o e es ima ion o in e ed pa ame e s. Howe e , simula-
ions ha e indica ed ha Ch omoSSE is mo e likely o unde es ima e,
a he han o e es ima e, cladogene ic changes, implying ha ou in-
e ences a e ins ead a he conse a i e (30).
Fu he mo e, as he model canno de ec “c yp ic” ch omosomal
ea angemen s, e.g., usion and ission e en s ha coun e balance
each o he and, he e o e, do no lead o changes in ch omosome
numbe , ou da a may e en ep esen an unde es ima ion o he e -
ec s o ch omosomal ea angemen s on specia ion. Fo example,
Mackin osh e al. (63) ound nine possible usion and ission e en s
ha may ha e con ibu ed o specia ion be ween wo B en his bu -
e lies ha o he wise di e li le in hei ch omosome numbe s.
O he ea angemen s, such as in e sions, may simila ly a ec ch o-
mosome e olu ion and e en con ibu e o specia ion (64). Howe e ,
ch omosome- le el assemblies o all E ebia would be equi ed o
s udy hei impac in his sys em.
We also highligh ha a solid axonomic amewo k, as p esen ed
he e, is highly p e e able o he co ec in e p e a ion o SSE- based
in e ences (34). Limi ing he da ase o only ka yo yped species e-
u ns simila alues o mos pa ame e s ( ableS3) bu ends owa d
highe unce ain y (la ge 95% HPD in e als) and o e es ima ion o
anagene ic pa ame e s, pa icula ly issions, which is expec ed o
incomple e phylogenies (34). Howe e , gi en he axonomic com-
plexi y o E ebia (21), i is likely ha some c yp ic species may exis ,
again indica ing ha ou es ima es o di e si ica ion may be a he
conse a i e. In addi ion, hyb idiza ion in E ebia is no common,
and closely ela ed species end o o m e y na ow con ac zones
wi h li le o no gene low, e.g., (22, 65), as is he case o o he Eu o-
pean bu e lies (66). Al hough he possibili y o gene low among
ou sequenced indi iduals in he dis an pas canno be ully excluded,
such gene low is unlikely and would no ha e a ec ed he o e all
opology o ou phylogeny.
We u he con i med he alidi y o ou in e ences by i ing
Ch omE ol (33) models ha do no es ima e ex inc ion and may hus
no su e he same po en ial pi alls as SSE ype models. We ound
e y simila esul s as o he SSE- based analyses (Figs.2B and 4), wi h
he excep ion o in e ed a es o anagene ic usions in he ynda us
clade, which we e es ima ed o be much highe o Ch omE ol. This
di e ence may be due o he in luence o he high ela i e ex inc ion
a es in his clade (Fig.4H), as Ch omE ol does no conside any
unobse ed specia ion ha may ha e esul ed in ex inc ion, whe eas
Ch omoSSE does (30). We u he es ima ed hidden backg ound
specia ion and ex inc ion a es wi hou conside ing ch omosomal
change using MiSSE (59) and ound ha , o ex an species, hese
a es do no appea o a y much (Supplemen a y Me hods and
igs.S12 and S13). Consequen ly, ou econs uc ion o ch omosom-
al change ac oss he deepe E ebia ee is unlikely o ha e been in lu-
enced by hidden specia ion.
He e, we used s a e- o - he- a phylogenomic models in one o he
mos ka yologically di e se g oups o bu e lies, p o iding e idence
o a mac oe olu iona y impac o majo ch omosomal ea ange-
men s ha equally occu in many o he animal, e.g., (51, 67) and plan
(50, 52) g oups. O e all, we p o ide e idence ha specia ion a es a e
highe wi h inc eased ch omosomal changes. Simila in e ences o
he impac s o ch omosomal ea angemen s a e o en ca ied ou a
highe axonomic le els o ac oss as e olu iona y imescales, e.g.,
(68, 69), po en ially masking ine- scaled pa e ns in younge clades.
Ou s udy b idges hese o me in es iga ions and mic oe olu iona y
s udies ha ocus on one species o compa e sibling species, e.g., (10,
51, 63, 68), by demons a ing wi hin- genus di e ences o ch omo-
somal usion- and ission- ela ed specia ion. We highligh ha ch o-
mosomal specia ion may be mo e ele an in clades wi h mo e
di e si y in ch omosome numbe s. In his genomic e a, high- quali y
e e ence genomes can be gene a ed and used o build phylogenomic
amewo ks, which will enable us o u he un a el he complexi ies
o ch omosomal e olu ion and specia ion.
MATERIALS AND METHODS
Da a collec ion
Adul specimens o 83 E ebia species we e collec ed be ween 2009 and
2021 ( ableS1). Whene e possible, species we e no sampled om
egions whe e hey migh hyb idize wi h o he , closely ela ed E ebia
species. Bodies we e ei he s o ed in e hanol a −20°C (n=53) wi h
wings sepa a ed o pinned a oom empe a u e (n=30). Fo he la -
e , he wings we e cu and s o ed sepa a ely be o e DNA ex ac ion.
DNA was ex ac ed om ho ax issue using a Qiagen Blood and Tis-
sue ki (Qiagen AG, Homb ech ikon, Swi ze land) ollowing he s an-
da d manu ac u e ’s p o ocol. Pai ed- end sequencing lib a ies we e
cons uc ed a he Depa men o Biosys ems Science and Enginee -
ing o ETH Zü ich in Basel, ollowed by sequencing on an Illumina
No aSeq 6000. Samples we e sequenced on wo S1 low cells. Species
iden i y o each specimen was con i med using an in silico DNA ba -
coding app oach. In sho , we used he pipeline o (70) wi h s anda d
se ings, o ex ac eads ha showed a high k- me simila i y o E ebia
mi ochond ial ba code sequences. Ba code e e ence sequences o
all a ailable species we e ob ained om he Ba code o Li e Da abase
(www.boldsys ems.o g). Fo each indi idual, we hen pe o med a
Downloaded om h ps://www.science.o g a Cen o de Biologia Molecula Se e o Ochoa on No embe 11, 2024
Augus ijnen e al., Sci. Ad . 10, eadl0989 (2024) 17 Ap il 2024
Science Ad AnceS | ReSeARch AR icle
9 o 13
de no o assembly o he il e ed eads wi h SPAdes 3.13.0 (71) and
mapped he con igs back o he ba code da abase wi h Yass 1.14 (72).
On he basis o he assembled ba code sequence, he species iden i y
o all indi iduals in ou s udy was co ec ly assigned.
Demul iplexed aw sequence eads we e p ocessed using as p
(73), imming poly- G ails. Re ained eads we e mapped o he
ch omosome- esol ed E ebia ligea e e ence genome (74), using bwa
0.7.17 (75). A e age mapping co e age was 38% bu a ied among
species, e lec ing phylogene ic dis ance o he e e ence genome
( ig.S11). One species (E ebia a amen a ia) was omi ed because o
e y low co e age (2.5%). SAM ools .1.13 (76) was hen used o e-
mo e unmapped, unpai ed, o duplica ed eads. A pileup ile o each
sample was gene a ed wi h BCF ools .1.12 (77) mpileup, ollowed by
a ian calling in BCF ools call (78). Indi idual Va ian Call Fo ma
(VCF) iles we e hen me ged and subsequen ly il e ed o emo e (i)
non- biallelic single- nucleo ide polymo phisms (SNPs), (ii) inse ions
and dele ions and adjacen SNPs wi hin 5base pai s (bp), (iii) SNPs
wi h quali y sco e<30, (i ) SNPs wi h mo e han 80% missing da a,
( ) SNPs wi h dep hs < 4 o >25, ( i) SNPs wi h mino allele equen-
cies (MAFs) < 0.03, and ( ii) SNPs alling wi hin epe i i e pa s o
he genome as iden i ied by Repea Maske 4.0.9 (79). This esul ed in
a da ase con aining 1.87 million SNPs om 82 E ebia samples. The
aw eads o hese samples ha e been deposi ed on Na ional Cen e
o Bio echnology In o ma ion (NCBI) unde he accession numbe
PRJNA1000734.
Ou g oup da a we e aken om ch omosome- esol ed assemblies
o o he Nymphalid bu e lies ha we e publicly a ailable a he ime
o analysis: he Sa y ini Pa a ge aege ia (80), Aphan opus hype an us
(81), and mo e dis an ou g oups Aglais u icae (82), Vanessa a alan a
(83), and Vanessa ca dui (84). To ob ain single copy o hologs (SCOs)
among all assemblies, each assembly including E. ligea was anno a ed
wi h WebAugus us (85, 86) in wo s eps: Fi s , gene p edic ion was
un using he s anda d Augus us species pa ame e s o Heliconius
melpomene. Then, pai wise SCOs be ween E. ligea and all o he spe-
cies we e iden i ied wi h O ho inde (87). These SCOs we e subse-
quen ly used o gene p edic ion in a second un o WebAugus us.
A o al o 4505 SCOs among E. ligea and ou selec ed ou g oups
we e iden i ied, o which 2920 SCOs wi h ≥20 SNPs we e e ained o
downs eam analyses. Exons o each SCO we e conca ena ed in o a
single coding egion o c ea e 2920 sepa a e VCF iles ex ac ed om
he E ebia da ase wi h BED ools (88). VCF iles we e hen ansposed
in o FASTA o ma , and he same exons we e ex ac ed o each ou -
g oup species and aligned o he E ebia sequence iles. Each esul ing
gene sequence ile was aligned wi h MAFFT .7.467 (89).
Phylogenomic analyses
ML gene ees we e es ima ed in IQTREE2 (90), which i s es i-
ma es he op imal subs i u ion model o a gene alignmen h ough
ModelFinde (91). Gene ees we e es ima ed wi h ul a as boo s ap
app oxima ion o 1000 i e a ions. A species consensus ee was hen
in e ed using coalescen me hods in ASTRAL- III (92). Be o e his
analysis, b anches wi h <10% boo s ap suppo we e collapsed o
imp o e he accu acy o ASTRAL- III (92). The esul ing species ee
had a no malized qua e sco e o 0.64, indica ing 64% ag eemen be-
ween gene ee opologies ( e lec ing an in e media e amoun o in-
comple e lineage so ing) and high o e all suppo ( ig.S1).
The obus ness o he phylogeny was u he alida ed ollowing
(28), i.e., selec ing loci based on phylogene ically in o ma i e pa am-
e e s o educe incong uency (28, 93, 94). Fi e alida ion da ase s
we e c ea ed, each con aining he 600 gene ees ha (i) had he high-
es a e age boo s ap in IQTREE; (ii) showed he highes clocklike-
ness, indica ing how well a gene ee app oaches an ul ame ic ee
o ma ion; (iii) had he lowes CG e sus AT con en s; (i ) had he
highes p opo ion o pa simony in o ma i e si es; and ( ) showed
he lowes mu a ion sa u a ion po en ial. Da ase s (i), (ii), and ( )
we e gene a ed wi h a modi ied ee_p ops.R sc ip om (95). Da a-
se s (iii) and (i ) we e gene a ed wi h FASconCAT (96). The sa u a ion
po en ial was calcula ed h ough eg ession slopes, wi h lowe mu a-
ional sa u a ion po en ial indica ing a lowe deg ee o amino acid
subs i u ions and a lowe sensi i i y o di e ences in e olu iona y
a es (93, 97). Consensus opologies o each da ase based on IQTREE
gene ees we e gene a ed wi h ASTRAL- III. The a e age no malized
qua e sco e was simila o he o e all da ase wi h 0.68 ( ange, 0.61
o 0.77). Compa isons be ween opologies o he o e all da ase and
he alida ion opologies indica ed only mino di e ences conce ning
ela ionships wi hin clades and conce ned he placemen o E ebia ni-
phonica and E ebia ae hiops, which a e no used in he clade- le el
analyses. All clades used o u he analyses we e s ongly suppo ed
in all alida ion da ase s ( igs.S2 o S6). The phylogeny based on he
o e all da ase , he ea e e e ed o as he All gene opology, was
he e o e used in downs eam analyses o cons ain da ing analyses
using ossil calib a ion.
Di e gence ime es ima ion
Mos s a e- dependen bi h- dea h (SSE) models equi e ul ame ic
ees because a es a e exp essed in ela i e o absolu e uni s o ime.
The la e allows o he exp ession o s a e- dependen a es o di e -
si ica ion in e ms o e en s pe species pe million yea s and is he e-
o e mo e in o ma i e o in e p e . We in e ed an ul ame ic ime
ee based on he 56 genes wi h <1% missing da a, o a o al o 82,923 bp,
wi h he opology cons ained acco ding o he All gene opology. We
chose o minimize missing da a o he da ing analysis as i occu ed
p ima ily in ou g oups, which could bias b anch leng hs and ela ion-
ships among ou g oups, which would, in u n, a ec he calib a ions,
which we e placed among he ou g oups.
No ossils a e known o E ebia and ew o Sa y ini o e all. Le he
co bie i (98), which has p e iously been used o da ing Sa y ini phy-
logenies (99), was he e o e selec ed. As he wo ou g oup species
wi hin Sa y ini, P. aege ia and A. hype an us, span he b ead h o he
clade, hei spli om E ebia coincides wi h he c own age o Sa y ini.
Thus, he age o L. co bie i (25.0 Ma ago) was aken as a conse a i e
minimal bound o he age o Sa y ini, wi h he median age se a 41.8 Ma
ago, ollowing he in e ence o (99). The oo o he ee ep esen s he
age o Nymphalidae and s em age o Sa y ini, p e iously es ima ed o
be 69.4 Ma ago (59.0 o 80.2 Ma ago) (100), which was aken as he
median age o his node. The minimum age o Nymphalidae was
bounded using he ossil Vanessa ame indica, as i s placemen wi hin
Vanessa is deba ed (101), and we conside ed placing his ossil close
o he base o Nymphalidae o be he mo e conse a i e app oach.
Calcula ions o he di e gence imes we e ca ied ou by node
da ing in M Bayes .3.2.7 (102). Pa i ionFinde (103) was used o
pa i ion he da ase and selec he pe - pa i ion bes subs i u ion
models. The in gamma model o among- si e a e a ia ion was he
bes model o all pa i ions, and we u he allowed o in eg a ion
o e he ull gene alized ime- e e sible (GTR) subs i u ion model
space. Ha d cons ain s we e placed on all nodes based on he All
gene opology, wi h he clade Sa y ini cons ained o E ebia, P. aege ia,
and A. hype an us o es ima e i s node age based on an o se
Downloaded om h ps://www.science.o g a Cen o de Biologia Molecula Se e o Ochoa on No embe 11, 2024