RESEARCH ARTICLE Open Access
Pa e ns o gene low and selec ion ac oss
mul iple species o Ac ocephalus wa ble s:
oo p in s o pa allel selec ion on he
Z ch omosome
Radka Rei o á
1*
, Ve onika Maje o á
1
,Jiří Rei
2,3
, Ma kus Ahola
4,5
, An e o Lindholm
6
and Pe P ocházka
7
Abs ac
Backg ound: Unde s anding he mechanisms and selec i e o ces leading o adap i e adia ions and o igin o
biodi e si y is a majo goal o e olu iona y biology. Ac ocephalus wa ble s a e small passe ines ha unde wen an
adap i e adia ion in he las app oxima ely 10 million yea s ha ga e ise o 37 ex an species, many o which s ill
hyb idize in na u e. Ac ocephalus wa ble s ha e se ed as model o ganisms o a wide a ie y o ecological and
beha io al s udies, ye ou knowledge o mechanisms and selec i e o ces d i ing hei adia ion is limi ed. He e we
s udied pa e ns o in e speci ic gene low and selec ion ac oss h ee Eu opean Ac ocephalus wa ble s o ge a i s
insigh in o mechanisms o adia ion o his a ian g oup.
Resul s: We analyzed nucleo ide a ia ion a eigh nuclea loci in h ee hyb idizing Ac ocephalus species wi h o e lapping
b eeding anges in Eu ope. Using an isola ion-wi h-mig a ion model o mul iple popula ions, we ound
e idence o unidi ec ional gene low om A. sci paceus oA. palus is and om A. palus is o A. dume o um.
Gene low was highe be ween gene ically mo e closely ela ed A. sci paceus and A. palus is han be ween
ecologically mo e simila A. palus is and A. dume o um, sugges ing ha g adual accumula ion o in insic
ba ie s a he han di e gen ecological selec ion a e mo e e icien in es ic ing in e speci ic gene low in
Ac ocephalus wa ble s. Al hough le els o gene ic di e en ia ion be ween di e en species pai s we e in
gene al no co ela ed, we ound signa u es o appa en ly independen ins ances o posi i e selec ion a he
same woZ-linkedlociinmul iplespecies.
Conclusions: Ou s udy b ings he i s e idence ha gene low occu ed du ing Ac ocephalus adia ion and no only
be ween sis e species. In e speci ic gene low could hus be an impo an sou ce o gene ic a ia ion in indi idual
Ac ocephalus species and could ha e accele a ed adap i e e olu ion and specia ion a e in his a ian g oup by c ea ing
no el gene ic combina ions and new pheno ypes. Independen ins ances o posi i e selec ion a he same loci in
mul iple species indica e an in e es ing possibili y ha he same loci migh ha e con ibu ed o ep oduc i e isola ion in
se e al specia ion e en s.
Keywo ds: Adap i e adia ion, Specia ion, Gene low, Pa allel adap i e e olu ion, Z ch omosome, Ac ocephalus wa ble s
* Co espondence: [email p o ec ed]
1
Depa men o Zoology, Facul y o Science, Cha les Uni e si y in P ague,
P ague, Czech Republic
Full lis o au ho in o ma ion is a ailable a he end o he a icle
© 2016 The Au ho (s). Open Access This a icle is dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0
In e na ional License (h p://c ea i ecommons.o g/licenses/by/4.0/), which pe mi s un es ic ed use, dis ibu ion, and
ep oduc ion in any medium, p o ided you gi e app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o
he C ea i e Commons license, and indica e i changes we e made. The C ea i e Commons Public Domain Dedica ion wai e
(h p://c ea i ecommons.o g/publicdomain/ze o/1.0/) applies o he da a made a ailable in his a icle, unless o he wise s a ed.
Rei o á e al. BMC E olu iona y Biology (2016) 16:130
DOI 10.1186/s12862-016-0692-2
Backg ound
In e speci ic gene low is an impo an e olu iona y
o ce. I may en ich gene ic a ia ion o indi idual
species, acili a e he o igin o new pheno ypes, lead o
he o igin o new species, o , con e sely, o species
usion [1]. I has been sugges ed ha high specia ion
a es and apid pheno ypic changes obse ed du ing
adap i e adia ions may be acili a ed by inc eased gen-
e ic a ia ion and no el gene ic combina ions p oduced
by in og essi e hyb idiza ion among he species [2, 3].
Pa e ns o in e speci ic gene low can also p o ide an
impo an insigh in o he gene ic basis o ep oduc i e
isola ion [4]. In species wi h incomple e ep oduc i e
isola ion, gene low occu s a loci wi h alleles ha a e
neu al o bene icial on he genomic backg ound o he
o he species, bu is limi ed a loci ha bo ing alleles
in ol ed in ep oduc i e isola ion [5].
Mos wo k explo ing pa e ns and a es o in e speci ic
gene low ocused on pai s o closely ela ed sis e
species [6] because, un il ecen ly [7], models allowing
he analysis o in e speci ic gene low we e limi ed o
only wo species. Se e al gene al indings eme ged om
hese s udies. Fi s , in he case o specia ion d i en by
in insic ba ie s, in e speci ic gene low is o en educed
on he sex ch omosomes (i.e., XY in he e ogame ic male
o ganisms and ZW in he e ogame ic emale o ganisms)
[8–11] and in genomic egions wi h low ecombina ion
a es [12], indica ing ha gene ic incompa ibili ies end
o p e e en ially accumula e in hese genomic egions.
Second, in he case o specia ion d i en by di e gen
ecological selec ion, gene low is co ela ed wi h eco-
logical di e gence [13] and is educed a ound genes
unde lying ecological di e ences [14]. In e speci ic gene
low ac oss la ge sec ions o he phylogene ic ee is,
howe e , much less s udied. Such s udies would be
pa icula ly in e es ing in ins ances o apid adia ions,
whe e lineage di e gence occu s on a sho e imescale
han he comple ion o ep oduc i e isola ion [15–18].
Knowledge o pa e ns and a es o in e speci ic gene
low in such sys ems allows o add ess ques ions, which
a e di icul o e en impossible o s udy using models o
only wo sis e species. One can, o example, examine
whe he he same genes con ibu e o ep oduc i e isola-
ion in di e en specia ion e en s, assess he impo ance
o a ious ep oduc i e ba ie s in es ic ing he gene
low and de e mine he o de in which di e en ep o-
duc i e ba ie s a ise du ing lineage di e gence. This
in o ma ion can help us o unde s and why lineage di-
e gence occu s so quickly in some axonomic g oups,
while no in o he s.
Ac ocephalus wa ble s a e small passe ines ha unde -
wen an adap i e adia ion du ing he las app oxima ely
10 million yea s [19], which ga e ise o 37 ex an
species occupying mainly Eu asia, A ica, Aus alia and
Paci ic islands [20]. The membe s o his genus a e
pheno ypically qui e uni o m. They, howe e , display a
g ea di e si y in ecology and beha iou , and ha e been
used ex ensi ely as models in ecological and beha io al
esea ch [20]. In e speci ic hyb idiza ion occasionally
occu s in Ac ocephalus wa ble s. I has been epo ed
no only be ween sis e species (e.g., A. sci paceus ×A.
palus is,A. a undinaceus ×A. s en o eus), bu also be-
ween mo e dis an ly ela ed species (e.g., A. palus is ×
A. dume o um,A. palus is ×A. schoenobaenus,A.
sci paceus ×A. a undinaceus) (see Addi ional ile 1 o
all epo ed cases o in e speci ic hyb idiza ion in
Ac ocephalus wa ble s and e e ences). In some o hese
cases, hyb idiza ion occu s be ween species o e y
di e en body size (A. sci paceus ×A. a undinaceus)o
e en be ween species belonging o di e en subgene a
(e.g. A. palus is ×A. schoenobaenus). So a e y li le is
known abou he deg ee o pos zygo ic ep oduc i e iso-
la ion be ween he species. In some cases, e.g. be ween
sis e species A. a undinaceus and A. s en o eus, pos zy-
go ic isola ion seems o be qui e s ong as no backc oss
hyb ids ha e been obse ed in a hyb idizing popula ion
[21]. In o he cases (e.g. be ween A. palus is and A.
dume o um), howe e , obse a ion o e ile F
1
hyb ids
[22] sugges s ha pos zygo ic isola ion is incomple e and
in e speci ic gene low migh occu be ween he species.
He e we pe o med he i s s udy explo ing pa e ns
and a es o in e speci ic gene low in his a ian g oup,
ocusing on h ee species o he subgenus No iocichla:A.
sci paceus,A. palus is and A. dume o um. All h ee
species a e e y simila in mo phology, bu ha e di e -
en habi a equi emen s, mig a ion s a egies and song
pa e ns [23]. A. sci paceus b eeds ac oss Eu ope,
no he n A ica and wes e n Asia ypically in eed beds
and win e s in sub-Saha an A ica. I s sis e species, A.
palus is, has simila geog aphical dis ibu ion, bu occu-
pies mos ly damp he baceous ege a ion and win e s in
sou heas e n A ica. Males o each species sing di e en
songs, ne e heless, in A. palus is, mixed singe s ha
inco po a e song ph ases om A. sci paceus o i s epe -
oi e a e known [24]. Bo h species occasionally o m
mixed pai s [24] and a e able o p oduce iable hyb id
o sp ing (e.g. [25]; see Addi ional ile 1 o a comple e
lis o e e ences). A. dume o um, he sis e species o
he A. sci paceus and A. palus is clade [26], b eeds in
Asia and No heas e n Eu ope in a ious he baceous o
bushy ege a ion and win e s in India. In Eas e n Eu ope
and Wes e n Asia, i o en co-occu s wi h A. palus is a
he same si es and hyb idiza ion be ween he wo species
has been obse ed (e.g. [22, 27]; see Addi ional ile 1 o
a comple e lis o e e ences). The exac equency o
hyb idiza ion be ween A. palus is and A. sci paceus and
be ween A. palus is and A. dume o um is ha d o es i-
ma e owing o mo phological simila i y o he species
Rei o á e al. BMC E olu iona y Biology (2016) 16:130 Page 2 o 12
and di icul y o dis inguish hyb ids wi hou gene ic
analysis. The o al numbe o obse ed mixed pai s is,
howe e , compa able in bo h species pai s (Addi ional
ile 1). The cases o hyb idiza ion be ween A. dume-
o um and A. sci paceus, which a e ecologically mo e
di e gen ,a e no known, al hough one pu a i e hyb id
indi idual has been epo ed [28].
To ge he i s insigh in o mechanisms and selec i e
o ces d i ing Ac ocephalus adia ion, we analyzed
nucleo ide a ia ion a ou au osomal and ou loci on
he Z ch omosome in sympa ic popula ions o A.
sci paceus,A. palus is and A. dume o um. Using an
isola ion-wi h-mig a ion model o mul iple popula ions
we es ed whe he in e speci ic gene low occu ed
among he s udied species and i so, whe he i was
highe be ween mo e gene ically simila o be ween
mo e ecologically simila species. In addi ion, using se -
e al es s o neu ali y, we explo ed pa e ns o posi i e
selec ion in all h ee species.
Me hods
S udy a ea and samples
We analyzed samples om 22 indi iduals o A. sci paceus,
26 indi iduals o A. palus is and 25 indi iduals o A.
dume o um. All bi ds we e caugh in 2010 du ing he
b eeding season (June and ea ly July) in sou he n Finland,
whe e he b eeding anges o all h ee species o e lap
(Fig. 1). Coo dina es o he si es, whe e indi idual bi ds
we e cap u ed, a e shown in Addi ional ile 2. F om each
indi idual, a blood sample was collec ed by b achial ein
punc u e and s o ed in pu e e hanol o u he ex ac ion
o genomic DNA. In addi ion, a sample om one indi id-
ual o A. schoenobaenus was used as he ou g oup.
Molecula sex de e mina ion and sequencing
GenomicDNAwaspu i iedbyDNeasyTissueKi
(Qiagen) acco ding o manu ac u e ’sins uc ions.Sex
o indi idual bi ds was de e mined ollowing he me hod
o G i i hs e al. [29] using P2 and P8 p ime s. These
p ime s ampli y a agmen o wo homologous genes
loca ed on he Z and W ch omosomes. The ampli ied
agmen s om he Z and W ch omosomes di e in
leng h, which can be isualized on aga ose gel. We u he
ampli ied and sequenced in ons o ou au osomal and
ou Z-linked loci. The ou au osomal loci a e loca ed on
di e en ch omosomes in he Zeb a Finch (Taeniopygia
gu a a) genome. O he ou Z-linked genes, PPWD1 and
ADAMTS6 a e close physically (68 kb), and he e o e may
no ha e independen e olu iona y his o ies. Howe e , he
o he wo loci on he Z ch omosome a e sepa a ed om
each o he and om PPWD1 and ADAMTS6 by > 19 Mb.
P ime s o PCR ampli ica ion we e pa ially ob ained
om published s udies and pa ially designed by ou sel es.
All p ime s we e designed in conse ed exonic egions o
he Zeb a Finch o chicken (Gallus gallus)genomein
o de o ampli y in onic sequences. P ime sequences
Fig. 1 B eeding anges o he h ee s udied Ac ocephalus wa ble s, A. sci paceus,A. palus is and A. dume o um. Sampling si es a e indica ed by
da k g ey ci cles
Rei o á e al. BMC E olu iona y Biology (2016) 16:130 Page 3 o 12
and leng hs o he ob ained PCR p oduc s a e p o-
ided in Addi ional ile 3. PCR condi ions a e same
as in S o cho á e al. [9]. All PCR p oduc s we e
sequenced in bo h di ec ions wi h he PCR p ime s
using Sange sequencing a he DNA sequencing la-
bo a o y a Cha les Uni e si y in P ague.
Da a analyses
Sequences we e manually edi ed using CodonCode
Aligne so wa e (CodonCode Co po a ion, Dedham,
MA). Alignmen s we e gene a ed by Clus alW as imple-
men ed in he p og am BioEdi [30]. All alignmen s we e
isually checked and manually adjus ed. Exonic se-
quences as well as indel polymo phisms we e excluded
om he analyses. Indi iduals and/o posi ions wi h
missing da a (mos ly a he ends o he sequences) we e
elimina ed om he da ase o ge sequences o he same
leng h o each locus. Diploid sequences a each locus
we e u he sepa a ed in o wo haplo ypes using he
p og am PHASE, e sion 2.1.1 [31] wi h he ollowing
pa ame e s: numbe o i e a ions = 10,000, hinning
in e al = 1, bu nin = 1000. We used he de aul ecom-
bina ion model, which is he gene al model o a ying
ecombina ion a e. Fo each da ase , we applied he
algo i hm i e imes wi h di e en andom seeds, and
we checked o consis ency o esul s ac oss independen
uns. We ob ained iden ical haplo ypes ac oss all uns
o all loci. The inal numbe o haploid sequences o
each locus and he leng h o sequences a e shown in
Table 1.
Basic popula ion gene ic analyses o polymo phism,
di e gence, ecombina ion, and es s o neu ali y based
on he allele equency spec um we e pe o med wi h
he p og am DnaSP [32]. Hudson-K ei man-Aguadé
(HKA) es s o posi i e selec ion [33] we e pe o med
using he HKA p og am (h ps://bio.cs . emple.edu/
~hey/so wa e/so wa e.h m#HKA). This p og am com-
pa es he a io o polymo phism o di e gence a mul-
iple loci. Loci unde posi i e selec ion a e expec ed o
show lowe le els o polymo phism ela i e o di e -
gence. Fo each locus, we se he in o ma ion whe he
he locus is au osomal o Z-linked, based on which
he HKA p og am accoun s o lowe e ec i e popula-
ion size and hus he lowe nucleo ide a ia ion on he Z
ch omosome. Genealogical ela ionships among haplo-
ypes o each locus we e econs uc ed wi h Ne wo k
so wa e [34] using Median joining algo i hm.
The da a we e i o he isola ion-wi h-mig a ion
model (IM) o mul iple popula ions [7] using he p o-
g am IMa2. The p og am es ima es se e al demog aphic
pa ame e s based on Ma ko chain Mon e Ca lo simula-
ions o genealogies. These pa ame e s include he e -
ec i e popula ion size o he cu en and ances al
popula ions, mig a ion a es be ween he popula ions
and he popula ion-spli imes. Because IMa2 assumes
no ecombina ion wi hin loci, we de e mined he longes
egion wi hou obse ed ecombina ion o each locus
using he p og am IMgc [35]. This p og am emo es
ei he si es o haplo ypes o p oduce he mos
in o ma ion- ich con iguous DNA sequence segmen
ha passes he ou -game e es . Non ecombinan
egions ep esen ed 84 % o he leng h o each locus, on
a e age, and we e used as an inpu o he IMa2
p og am. Fo each locus, we p o ided he in o ma ion
whe he i is au osomal o Z-linked, based on which
IMa2 p og am accoun s o di e en e ec i e popula ion
sizes be ween au osomes and he Z ch omosome. We
an he p og am h ee imes wi h iden ical s a ing
condi ions, wi h he excep ion o he andom numbe
seed, o assess con e gence. To acili a e mixing o he
Ma ko chains, we used Me opolis coupling wi h 30
chains and a geome ic hea ing model. Uppe bounds
o he p io dis ibu ions o pa ame e alues we e se
as sugges ed in IMa2 documen a ion (10 o popula ion
size pa ame e s, 3 o mig a ion a es pa ame e s and 5
o spli ing ime). All uns began wi h a bu n-in pe iod
o 100,000 s eps and we e allowed o con inue o 7–8
million s eps. We we e able o achie e adequa e mixing
o he Ma ko chains as indica ed by end line plo s, e -
ec i e sample size (ESS) alues ( o all h ee uns, mos
ESS alues we e highe han 20,000 and no ESS alue
was lowe han 50), and e y simila pa ame e es ima es
in he i s hal and he second hal o he un.
Independen uns con e ged o he same esul (e.g.,
maximum-likelihood es ima es and ma ginal pos e io
p obabili y dis ibu ions o he demog aphic pa ame e s
we e essen ially he same o all h ee uns). To pe -
o m likelihood- a io es s o nes ed models, we com-
bined esul s o he h ee independen IMa2 uns
( oge he con aining mo e han 200,000 genealogies)
in a single L mode un.
All es ima ed pa ame e s o he IM model a e scaled
o he mu a ion a e. To con e hese pa ame e s o
biologically meaning ul quan i ies (i.e., N
e
, e ec i e
popula ion size in numbe o indi iduals; m, mig a ion
a e pe yea ; 2 Nm, popula ion mig a ion a e; , di e -
gence ime in yea s) we calcula ed he neu al mu a ion
a e o each locus (μ) using di e gence o he ou g oup.
The neu al mu a ion a e pe yea was calcula ed o
each locus by using he o mula D=2μ
1
whe e Dis he
a e age pai wise di e gence, Dxy, be ween he h ee
s udied species and he ou g oup, and
1
is he di e -
gence ime. We assumed ha he di e gence ime
be ween A. sci paceus/A. palus is/A. dume o um and A.
schoenobaenus is 5.9 million yea s. This was es ima ed
on he basis o cy och ome b sequence di e gence,
which is 12.4 % [36], and assuming app oxima ely 2.1 %
sequence di e gence pe million yea s [37]. The
Rei o á e al. BMC E olu iona y Biology (2016) 16:130 Page 4 o 12
geome ic mean o he locus-speci ic mu a ion a es was
hen used o he pa ame e con e sion. A gene a ion
ime o one yea (Ac ocephalus wa ble s ep oduce only
once pe yea ) was assumed o es ima e mu a ion a es
pe gene a ion.
Resul s
Le els o in aspeci ic polymo phism and es s o neu ali y
Fo each o he eigh sequenced loci, we ob ained 86 –128
compu a ionally econs uc ed haploid sequences (Table 1).
Le els o nucleo ide a ia ion we e gene ally high in all
Table 1 Polymo phism s a is ics o eigh s udied loci
Locus Ch
a
L
b
Species
c
N
d
S
e
π(%)
θ(%)
g
TD
h
FLD
i
D (%)
j
17483 A 484 A.s. 44 9 0.323 0.427 −0.701 −0.566 3.273
A.p. 32 13 0.805 0.667 0.675 0.149 3.648
A.d. 48 10 0.268 0.466 −1.225 −1.641 3.594
All 124 32 0.731 1.226 3.494
21281 A 355 A.s. 28 29 1.868 2.099 −0.403 −1.197 2.998
A.p. 20 24 1.893 1.906 −0.025 −0.200 2.634
A.d. 38 22 1.071 1.475 −0.922 −0.823 3.136
All 86 50 1.905 2.802 2.974
24972 A 636 A.s. 24 29 0.999 1.221 −0.683 −1.224 2.732
A.p. 38 35 0.819 1.310 −1.314 −2.431 2.557
A.d. 38 30 0.861 1.123 −0.809 −0.404 2.578
All 100 71 1.031 2.156 2.607
RPL5-4 A 390 A.s. 38 4 0.101 0.244 −1.420 −2.135 3.893
A.p. 50 15 0.994 0.859 0.484 1.112 4.359
A.d. 48 12 0.591 0.693 −0.438 0.407 4.049
All 136 35 1.383 1.636 4.120
ADAMTS6 Z 488 A.s. 35 0 0.000 0.000 / / 1.025
A.p. 43 2 0.019 0.095 −1.480 −2.462 0.829
A.d. 42 7 0.206 0.333 −1.050 −1.002 0.937
All 120 10 0.171 0.382 0.924
PPWD1 Z 566 A.s. 34 14 0.550 0.605 −0.295 0.228 2.520
A.p. 45 9 0.318 0.364 −0.358 0.712 2.493
A.d. 13 15 0.716 0.854 −0.676 −0.795 2.202
All 92 39 0.663 1.353 2.462
TG401 Z 829 A.s. 39 0 0.000 0.000 / / 2.051
A.p. 45 4 0.027 0.110 −1.764 −2.240 1.697
A.d. 44 1 0.005 0.028 −1.115 −1.803 2.536
All 128 18 0.712 0.400 2.093
TG1505 Z 532 A.s. 39 0 0.000 0.000 / / 1.880
A.p. 43 0 0.000 0.000 / / 2.444
A.d. 42 0 0.000 0.000 / / 2.632
All 124 6 0.508 0.209 2.330
a
Au osome (A), Ch omosome Z (Z)
b
Leng h o sequence (bp)
c
A. sci paceus (A.s.), A. palus is (A.p.), A. dume o um (A.d.)
d
Numbe o haploid sequences
e
Numbe o seg ega ing si es
A e age numbe o nucleo ide di e ences
g
P opo ion o polymo phic si es
h
Tajima’sD
i
Fu and Li’sD
j
Di e gence o ou g oup measu ed as a e age pai wise di e gence, Dxy
Rei o á e al. BMC E olu iona y Biology (2016) 16:130 Page 5 o 12

h ee s udied species, sugges ing a high e ec i e
popula ion size o he species. When a e aged o e
all loci, π=0.480 % and θ= 0.575 % in A. sci paceus,
π=0.609 % and θ= 0.664 % in A. palus is,andπ=
0.465 % and θ=0.622 % in A. dume o um.We,how-
e e , obse ed a subs an ial he e ogenei y in he le els
o nucleo ide a ia ion among loci. No ably, wi hin-
species nucleo ide a ia ion was ze o o e y low in
h ee ou o ou Z-linked loci: ADAMTS6,TG401
and TG1505. Fo ADAMTS6 and TG401,πand θ
eached 0 % in A. sci paceus and we e also qui e low
in he emaining wo species. Fo TG1505, πand θ
eached 0 % e en in all h ee species (Table 1). This
can be also seen on haplo ype ne wo ks showing ze o
o e y low haplo ype di e si y wi hin he species in
he h ee loci (Fig. 2).
To es whe he his he e ogenei y in nucleo ide
a ia ion could be a ibu ed o he ac ion o posi i e
selec ion a some loci, we pe o med a mul ilocus HKA
es . The es was applied sepa a ely o each species, in
each case using a single sequence om A. schoenobaenus
as he ou g oup. The null model was ejec ed in all h ee
species (Table 2), sugges ing ha posi i e selec ion
a ec s le els o nucleo ide a ia ion a one o mo e loci
in all h ee species. A close inspec ion o le els o poly-
mo phism and di e gence o indi idual loci e ealed
ha loci TG401 in A. sci paceus and A. palus is, and
TG1505 in A. sci paceus,A. palus is and A. dume o um
showed he highes de ia ions om expec ed alues
(Addi ional ile 4). When hese wo genes we e excluded
om he HKA es , he null model was no ejec ed in
A. palus is and A. dume o um, bu was s ill ejec ed in
A. sci paceus (da a no shown). This sugges s ha a
ecen selec i e sweep migh ha e occu ed a hese wo
loci independen ly in mul iple species.
We also looked o e idence o selec ion by compa ing
he dis ibu ion o allele equencies wi h he expec a-
ions unde a neu al equilib ium model using Tajima’s
D and Fu-Li’s D es s (Table 1). These es s we e pe -
o med o all s udied loci sepa a ely in each species.
The e was a endency owa d nega i e alues o Tajima’s
D in all h ee species (17 o 19 alues we e nega i e),
which may e lec mild popula ion expansions. None he-
less, none o he 19 alues o Tajima’s D wi hin species
we e signi ican ly di e en om he neu al expec a ion
o 0. No signi ican de ia ions om neu al expec a ions
we e ei he ound in Fu-Li’s D es s. These es s,
howe e , could no be pe o med o loci wi h ze o
wi hin-species nucleo ide a ia ion.
Es ima ion o di e gence imes, e ec i e popula ion sizes
and a es o in e speci ic gene low
Using IM model o mul iple popula ions, we es ima ed
ha he ime since di e gence be ween A. sci paceus and
A. palus is is 1.1 Mya and be ween he common ances-
o o hese species and A. dume o um 2.5 Mya. The
es ima ed N
e
was simila o all h ee species, 429,802
o A. sci paceus, 541,781 o A. palus is, and 535,194
o A. dume o um. N
e
o he common ances o o A.
sci paceus and A. palus is was subs an ially lowe ,
80,691. N
e
o he common ances o o all h ee species
could no be es ima ed accu a ely since he pos e io
p obabili y dis ibu ion o his pa ame e was e y la
(Fig. 3). The es ima es o mig a ion a es be ween A.
sci paceus and A. palus is we e ela i ely high, al hough
only in one di ec ion (2 Nm = 0.238 om A. sci paceus
o A. palus is,2Nm < 0.001 in he opposi e di ec ion).
Mig a ion a es be ween A. palus is and A. dume o um
we e lowe (2 Nm = 0.062 om A. palus is o A. dume-
o um,2Nm < 0.001 in he opposi e di ec ion) and no
mig a ion was de ec ed be ween A. sci paceus and A.
dume o um (2 Nm < 0.001 in bo h di ec ions). Es ima es
o mig a ion a es be ween A. dume o um and he com-
mon ances o o A. sci paceus and A. palus is we e qui e
high, hough again only in one di ec ion (2 Nm =0.95
om A. sci paceus and A. palus is ances o o A. dume-
o um,2Nm < 0.01 in he opposi e di ec ion); howe e ,
hese es ima es migh no be accu a e since pos e io
p obabili y dis ibu ions o hese pa ame e we e la
(Fig. 3). Maximum-likelihood es ima es (MLE) and 95 %
highes pos e io densi y (HPD) in e als o all model pa-
ame e s a e gi en in Table 3, and he ma ginal pos e io
p obabili y dis ibu ions a e shown in Fig. 3.
To es whe he ou es ima ed model wi h gene low
i s signi ican ly be e he da a han models wi hou
gene low be ween pa icula species, we pe o med
log-likelihood a io es s o nes ed models [38] as
implemen ed in he IMa2 p og am. We es ed ou
nes ed models (Table 4) whe e bo h mig a ion pa am-
e e s be ween (1) A. sci paceus and A. palus is,(2)
A. sci paceus and A. dume o um,(3)A. palus is and
A. dume o um,and(4)A. dume o um and he com-
mon ances o o A. sci paceus and A. palus is we e se o
ze o. The ull model wi h gene low was a signi ican ly
be e i o he da a han all nes ed models, excep he
second one (Table 4).
We we e u he in e es ed in whe he pa e ns o
in e speci ic gene low be ween A. sci paceus and A.
palus is, and be ween A. palus is and A. dume o um
a e simila o no . As IMa2 does no allow o es ima e
locus-speci ic mig a ion a es, we compa ed F
ST
alues,
which e lec be ween-species di e en ia ion and a e
co ela ed wi h le els o gene low in hyb idizing axa
(e.g., [9]). We should, howe e , no e ha F
ST
alues
migh be also a ec ed by he ac ion o posi i e selec ion
ha leads o a educed wi hin-species nucleo ide
a ia ion [39]. No s a is ically signi ican co ela ion in
F
ST
be ween he wo species pai s was ound ( = 0.47,
Rei o á e al. BMC E olu iona y Biology (2016) 16:130 Page 6 o 12
p> 0.05), sugges ing ha he pa e ns o in e speci ic
gene low and/o posi i e selec ion a e in gene al di e -
en o he wo species pai s.
Discussion
Adap i e adia ions ep esen an ou s anding sys em
o s udying mechanisms o specia ion and adap i e
di e gence. Al hough di e gen ecological selec ion
seems o be an impo an d i e o many adap i e
adia ions [40], i has been sugges ed ha special gen-
omic p ope ies, such as gene duplica ions, ac i a ion
o ansposable elemen s o p esence o gene low
among species, can also con ibu e o high specia ion
a es and apid pheno ypic changes du ing adap i e
adia ions [41].
Ac ocephalus wa ble s ep esen an in e es ing model
sys em o s udying mechanisms o adia ion in bi ds,
al hough hei mainly con inen al adia ion was ela i ely
slow compa ed o he well-known cases o adap i e
adia ions on islands, such as Da win inches [13, 17].
Ou es ima es o di e gence imes a e app oxima ely 1.1
Mya o he sis e species A. sci paceus and A. palus is,
and 2.5 Mya o hei common ances o and A. dume-
o um. Despi e his ela i ely deep di e gence, many
cu en ly exis ing Ac ocephalus species s ill hyb idize in
Table 2 HKA es o posi i e selec ion
Species
a
Sum o de ia ions
b
D
c
P- alue
d
P- alue
e
A.s. 28.8754 7 0.00015 0.00000
A.p. 17.4805 7 0.01455 0.00310
A.d. 24.4626 7 0.00094 0.00000
a
A. sci paceus (A.s.), A. palus is (A.p.), A. dume o um (A.d.)
b
Coun ed acco ding o he o mula ∑(obse ed - expec ed)
2
/ a iance)
c
Deg ee o eedom
d
P obabili y om chi-squa e dis ibu ion (signi ican alues a e indica ed in bold)
e
P obabili y om simula ions (no. o simula ion 10 000; signi ican alues a e
indica ed in bold)
A
21281
17483
RPL5-424972
B
TG401 TG1505
PPWD1ADAMTS6
Fig. 2 Haplo ype ne wo ks o ou au osomal (a) and ou Z-linked (b)loci. Size o he ci cles a e p opo ional o he numbe o haplo ypes. A. sci paceus is
indica ed in ed, A. palus is in g een and A. dume o um in g ey. In he case o 21281 and 24972, haplo ype ne wo ks we e cons uc ed
using only sequences wi h no ecombina ion wi in loci ob ained wi h p og am IMgc (see Me hods)
Rei o á e al. BMC E olu iona y Biology (2016) 16:130 Page 7 o 12
na u e (Addi ional ile 1). This hyb idiza ion migh in
p inciple lead o gene low among he species i F
1
hyb ids
a e e ile. Howe e , since e y li le is known abou he
deg ee o pos zygo ic isola ion in his a ian g oup, i is
ha d o p edic whe he gene low could occu among
Ac ocephalus species and how s ong i migh be.
Fig. 3 The ma ginal pos e io p obabili y dis ibu ions o he demog aphic pa ame e s o he IM model. E ec i e popula ion sizes o A. sci paceus (N
s
),
A. palus is (N
p
), A. dume o um (N
d
), he common ances o o A. sci paceus and A. palus is (N
sp
), and he common ances o o all h ee species (Na).
Di e gence ime be ween A. sci paceus and A. palus is (
0
) and be ween A. dume o um and he common ances o o A. sci paceus and A. palus is (
1
).
Mig a ion a es pe yea (m
s>p
indica es mig a ion om A. palus is in o A. sci paceus)
Rei o á e al. BMC E olu iona y Biology (2016) 16:130 Page 8 o 12
He e we p o ided he i s e idence ha in e speci ic
gene low occu ed du ing Ac ocephalus adia ion and
no only be ween sis e species. We ound e idence o
gene low be ween A. sci paceus and A. palus is, which
a e he sis e species wi h somewha di e en ecological
equi emen s, and be ween A. palus is and A. dume-
o um, which ha e simila ecological equi emen s and
o en co-occu a he same si es whe e hei anges
o e lap. Signi ican gene low was de ec ed also be ween
A. dume o um and he ances o o A. sci paceus and A.
palus is. No gene low was, howe e , de ec ed be ween
A. sci paceus and A. dume o um. Al hough he isola ion-
wi h-mig a ion model es ima es no only con empo a y
gene low, bu a e age gene low among popula ions
since he ime o di e gence [7], i is no ewo hy ha
hyb idiza ion cu en ly occu s be ween bo h pai s o
species o which gene low has been de ec ed, bu no
be ween A. sci paceus and A. palus is o which no
clea cases o in e speci ic hyb idiza ion a e known.
In e es ingly, he es ima ed le els o gene low we e in
all cases asymme ic. Gene low occu ed om A.
sci paceus o A. palus is (2 Nm = 0.238), om A. palus-
is o A. dume o um (2 Nm = 0.062), and om he A.
sci paceus/A. palus is ances o o A. dume o um (2
Nm = 0.95). Ze o gene low was de ec ed in he opposi e
di ec ions. Such unidi ec ional gene low is expec ed when
a popula ion o one species expands o he a ea al eady
occupied by a ela ed species and in e b eeding is no
p e en ed be ween he wo species. In og ession o neu-
al alleles hen occu s almos exclusi ely om he local
in o he in ading species [42]. Un o una ely, his o y o
b eeding ange changes a e no well known o he s udied
species, bu his o ically ecen b eeding ange expan-
sion o A. dume o um o eas e n and no he n Eu ope
whe e i encoun e ed A. palus is [20] is consis en
wi h his scena io.
An impo an issue in specia ion esea ch is o unde -
s and e olu iona y o ces ha a e esponsible o he
es ablishmen o ep oduc i e ba ie s be ween he spe-
cies. He e we es ed whe he gene low is highe be ween
gene ically mo e simila , bu ecologically somewha
di e en , A. sci paceus and A. palus is, o be ween
ecologically mo e simila , bu gene ically mo e dis an ,
A. palus is and A. dume o um. Ou esul s show ha
gene low is mo e han h ee imes highe be ween
gene ically mo e simila A. sci paceus and A. palus is
han be ween A. palus is and A. dume o um, al hough
he la e wo species ha e highe chance o mee in he
b eeding si es and e y likely mo e o en hyb idize.
Lowe le els o gene low be ween A. palus is and A.
dume o um migh be caused by his o ically mo e ecen
con ac be ween he species [23], bu migh also sugges
Table 3 Maximum-likelihood es ima es (MLE) and 95 % highes
pos e io densi y (HPD) in e als o demog aphic pa ame e s o
isola ion wi h mig a ion model
Pa ame e MLE HPD95Lo
a
HPD95Hi
b
N
s
c
429802 258540 663640
N
p
c
541781 344171 825022
N
d
c
535194 360638 772326
N
sp
c
80691 0 627412
m
s>p
d
1.14E-09 0 5.09E-07
m
p>s
d
1.92E-07 7.97E-09 5.43E-07
m
s>d
d
1.14E-09 0 1.22E-07
m
d>s
d
1.14E-09 0 1.17E-07
m
p>d
d
1.14E-09 0 1.24E-07
m
d>p
d
5.35E-08 0 2.17E-07
m
d>sp
d
8.84E-07 1.81E-07 2.28E-06
m
sp>d
d
1.14E-09 0 2.10E-06
2N
s
m
s>p
e
0.00081 - -
2N
s
m
s>d
e
0.00039 - -
2N
p
m
p>s
e
0.23800 - -
2N
p
m
p>d
e
0.00054 - -
2N
d
m
d>s
e
0.00044 - -
2N
d
m
d>p
e
0.06182 - -
2N
d
m
d>sp
e
0.94860 - -
2N
sp
m
sp>d
e
0.00750 - -
0
1116496 668580 2217184
1
2546534 1657289 6583046
a
Lowe 95 % HPD
b
Highe 95 % HPD
c
E ec i e popula ion sizes o A. sci paceus (N
s
), A. palus is (N
p
), A. dume o um
(N
d
), and he common ances o o A. sci paceus and A. palus is (N
sp
) in numbe s
o indi iduals
d
Mig a ion a es pe yea (species a e ma ked in he same way as abo e, e.g.,
m
s>p
indica es mig a ion om A. palus is in o A. sci paceus)
e
The popula ion mig a ion a e (species a e ma ked in he same way as abo e,
e.g., 2N
s
m
s>p
indica es popula ion mig a ion a e om A. palus is o A. sci paceus)
Di e gence ime be ween A. sci paceus and A. palus is (
0
) and be ween A.
dume o um and he common ances o o A. sci paceus and A. palus is (
1
)in
numbe s o yea s
Table 4 Log-likelihood a io es s o nes ed models
Model
a
log(P)
b
2LLR
c
d
d
P
e
m
s>p
=0,m
p>s
=0 −40.75 106.5 2 <0.0000
m
s>d
=0,m
d>s
= 0 12.49 0 2 1.0000
m
p>d
=0,m
d>p
= 0 9.265 6.448 2 0.0255
m
p>sp
=0,m
sp>d
= 0 9.333 6.313 2 0.0273
a
Fou nes ed models wi h ze o mig a ion a es be ween (1) A. sci paceus and
A. palus is, (2) A. sci paceus and A. dume o um, (3) A. palus is and A.
dume o um, and (4) A. dume o um and he common ances o o A. sci paceus
and A. palus is we e compa ed o he es ima ed ull model
b
Es ima es o he pos e io densi y unc ion unde he ull model
c
Log-likelihood a io s a is ics calcula ed as he di e ence be ween he highes
pos e io p obabili y o he ull model and he highes pos e io p obabili y
o he nes ed model
d
The deg ees o eedom. Models in which mig a ion a leas in one di ec ion
is equal o 0 ha e dis ibu ions o 2LLR ha a e a mix u e and χ2 mix u e
dis ibu ion was hus used o calcula e he P- alue
e
The p obabili y o achie ing he es s a is ics by chance unde he null model
(signi ican alues in bold)
Rei o á e al. BMC E olu iona y Biology (2016) 16:130 Page 9 o 12