Content uploaded by Jácint Tökölyi
Author content
All content in this area was uploaded by Jácint Tökölyi on Sep 02, 2014
Content may be subject to copyright.
Ornis Hungarica 2014. 22(1): 15–35.
doi: 10.2478/orhu-2014-0008
Phylogeny, historical biogeography and the
evolution of migration in accipitrid birds of
prey (Aves: Accipitriformes)
Jenő nagy
1
& Jácint tökölyi
2,*
Jenő Nagy & Jácint Tökölyi 2014. Phylogeny, historical biogeography and the evolution of
mig ration in accipitrid birds of prey (Aves: Accipitriformes). – Ornis Hungarica 22(1): 15–35.
Abstract Migration plays a fundamental part in the life of most temperate bird species. The re
gu lar, largescale seasonal movements that characterize temperate migration systems appear to
have originated in parallel with the postglacial northern expansion of tropical species. Migratoriness is also in-
uenced by a number of ecological factors, such as the ability to survive harsh winters. Hence, understanding
the origins and evolution of migration requires integration of the biogeographic history and ecology of birds in a
phylogenetic context. We used molecular dating and ancestral state reconstruction to infer the origins and evolu-
tionary changes in migratory behavior and ancestral area reconstruction to investigate historical patterns of range
evolution in accipitrid birds of prey (Accipitriformes). Migration evolved multiple times in birds of prey, the ear-
liest of which occurred in true hawks (Accipitrinae), during the middle Miocene period, according to our analy-
ses. In most cases, a tropical ancestral distribution was inferred for the nonmigratory ancestors of migratory line-
ages. Results from directional evolutionary tests indicate that migration evolved in the tropics and then increased
the rate of colonization of temperate habitats, suggesting that temperate species might be descendants of tropi-
cal ones that dispersed into these seasonal habitats. Finally, we found that diet generalization predicts migratori-
ness in this group.
Keywords: ancestral area reconstruction, annual cycle, comparative, diet specialization, diurnal birds of prey, mo-
lecular dating, seasonality
Összefoglalás A legtöbb mérsékelt övi madárfaj életciklusában alapvető szerepet tölt be a vonulás. A rendszeres, nagy
kiterjedésű mozgások, melyek a mérsékelt övi vonulási rendszereket jellemzik, egyes feltételezések szerint a trópusi
fajok posztglaciális, északi irányú terjeszkedésével párhuzamosan jelentek meg. Ezen felül a vonulás előfordulását
számos ökológiai tényező is befolyásolhatja, mint például a környezet szezonalitásának mértéke vagy a téli túlélést
befolyásoló tényezők. A vonulás eredete és evolúciója ezért csak úgy érthető meg, ha a madarak biogeográai törté-
netiségét és ökológiáját logenetikai kontextusban tanulmányozzuk. Jelen vizsgálatban a vágómadáralakúak (Acci-
pitriformes) vonulásának evolúcióját elemeztük komparatív módszerekkel. Első lépésben létrehoztunk egy fosszilis
adatok alapján datált molekuláris törzsfát, amelyen jellegrekonstrukciót végeztünk és rekonstruáltuk a fajok ősi elter-
jedési területét. Az elemzéseink alapján a vonulás többször alakult ki a ragadozók esetében, legkorábban a héjafor-
mákon (Accipitrinae) belül, vélhetően a Miocén közepén. A legtöbb esetben a vonuló leszármazási vonalak nem vo-
nuló őseinél trópusi elterjedésre következtethetünk. A direkcionális evolúciós teszt alapján a vonulás a trópusokon
jelent meg és megnövelte a mérsékelt égöv kolonizációjának rátáját. Eszerint tehát a mérsékelt övi ragadozómadár fa-
jok vonuló trópusi fajok leszármazottainak tekinthetők, melyek az erősen szezonális, északi élőhelyek irányába ter-
jeszkedtek. Végezetül negatív kapcsolatot találtunk a vonulás megjelenése és a táplálékspecializáció mértéke között.
Kulcsszavak: éves ciklus, jellegrekonstrukció, komparatív, molekuláris datálás, nappali ragadozómadarak, sze-
zonalitás, táplálékspecializáció
1
MTA-ELTE-MTM Ecology Research Group, 1117 Budapest, Pázmány Péter sétány 1/C, Hungary, e-mail: jenon-
agy.off@gmail.com
2
MTA-DE “Lendület” Behavioural Ecology Research Group, Department of Evolutionary Zoology and Human
Biology, University of Debrecen, 4032 Debrecen, Egyetem tér 1., Hungary
*
corresponding author: jtokolyi@vocs.unideb.hu
ORNIS HUNGARICA 2014. 22(1)
16
Introduction
Birds originated, according to our current
knowledge of the fossil record, about 150
200 million years ago during the geologi-
cal era of Jurassic (Padian & Chiappe 1998).
The appearance of powered ight, pro bab
ly in combination with several other avi-
an fea tures such as warmbloodedness and
the presence of extensive parental care has
fuelled the diversication of this group of
vertebrates, which seems to have accele
rated around, or shortly after the Cretaceous
Pa leo gene boundary (Ericson et al. 2006,
Brown et al. 2008, Jetz et al. 2012).
The widespread occurrence of birds is
greatly facilitated by their excellent dis-
persal capabilities. This is perhaps most
clearly seen in migratory birds, which can
travel thousands of kilometres on continen-
tal scale within a single year. Migration is
a characteristic feature of birds that is ex-
tremely common especially in species in-
habiting the Northern Hemisphere tempe
rate zone and the Arctic, but it also occurs,
although in less extreme forms, in other re-
gions of the globe in the form of intratropi-
cal migration systems (Alerstam 1993, Ber
thold 2001, Newton 2008).
Migration itself profoundly inuences the
distribution, ecology and diversication of
birds (Newton 2008), hence it is not surpri
sing that a great deal of information has ac-
cumulated on its internal, proximate deter-
minants and its phenology ever since Hans
Christian Mortensen started ringing birds at
the very end of the 19
th
century. These stu dies
revealed that the migratory phenotype is de-
termined by a set of complex and tightly re
gu lated mechanisms (Gwinner 1990), which
includes, among others (1) sensory elements
underlying orientation and navigation, (2)
mechanisms responsible for the regulation of
migratory restlessness (‘zugunruhe’) du ring
the annual cycle and (3) a range of physio-
logical adaptations that cover the metabolic
requirements of longdistance ights during
migration. Understanding how such a comp
lex phenotype could have evolved is a major
challenge in ornithology.
While a wealth of information has accu-
mulated on the details of the process of mig
ration, comparatively little is known about
how migration originated and evolved in
birds. This is not surprising, since behavio-
ral traits, such as the migratory ha bits of a
species do not fossilize and hence our cur-
rent ideas of it are strictly inferred from
phylogenetic or biogeographic studies. Cur-
rent theories of the evolution of migra-
tion can be divided into two groups (Rap-
pole & Jones 2002, Bruderer & Salewski
2008). The ‘tropical origin’ hypothesis pro-
poses that migratory birds derive from spe-
cies inhabiting regions where environmen-
tal factors were constant during the year,
so there was no need for migration. These
species could have colonized more sea-
sonal, northern habitats, which, during the
summer months provided appropriate con-
ditions for reproduction. However, during
winter food availability decreased, hence
these birds were forced to return to southern
latitudes (Rappole & Jones 2002, Bruderer
& Salewski 2008). Cox (1985) developed
a steppingstone model of this hypothesis.
According to this model, resourcelimita-
tion due to competition for food forced cer-
tain tropical resident species to expand their
range to the subtropics. These birds conti nu
ed to return to the tropics during the winter,
resulting in the formation of partial mig rant
species. These partial migrants then continu
ed to spread to higher latitudes where they
were able to breed successfully while still
returning to the original area in the winter
17
J. Nagy & J. Tökölyi
(Stiles 1980, Cox 1985). Thus, this hypo the
sis predicts that migratory species evolved
from tropical ancestors.
Several lines of evidence support the tro
pi cal origin of longdistance migrants. Jo-
seph et al. (1999), in a study of waders,
analyzed the evolution of breeding and win-
tering distribution of 16 species from the
genus Charadrius using phylogenetic me
thods. By reconstructing the hypothesized
distribution of the ancestors of these birds,
they showed that species that are migratory
today derive from ancestors whose bree ding
and nonbreeding ranges were located in the
tropical zone. Another similar study inves-
tigated the evolution of migration in Ca-
tharus thrushes (Outlaw et al. 2003). This
study showed that North American (migra-
tory) thrushes are sister to tropical species,
and the ancestral area for the whole lineage
was inferred to be in the Neotropics, pro-
viding further support for the “tropical ori-
gin” hypothesis. This pattern is not restric
ted to interspecic comparisons but is also
seen among populations differing in migra-
tory status. For instance, in a study of North
American Chipping Sparrows (Spizaella
passerina), Milá et al. (2006) have shown
that the northern, longdistance migrant
populations descend from nonmigrato-
ry Mexican populations, which colonized
North America after the last glacial maxi-
mum 18,000 years ago. Thus, longdistance
migration and colonization of temperate re-
gions in this species developed in tandem.
Other hypotheses on the origin of migra-
tion emphasize the importance of chan ges
on the breeding territories of birds with a
northern distribution (‘northern origin’ hy-
pothesis) (Bell 2000, 2005, Bruderer &
Salewski 2008). According to these suppo-
sitions, climatic or other ecological chan ges
(e.g. global cooling) could have led to the
evolution of migration by forcing nonmig
ratory temperate and arctic species to leave
there home ranges during the winter (Bell
2000, 2005), resulting in migratory strate-
gies which allowed the survival of popu-
lations in a strongly seasonal milieu (Bell
2000, 2005). Thus, this hypothesis predicts
that migration evolved from temperate resi
dent species, a prediction that has received
relatively low support to date. It is clear,
however, that migration can evolve with-
out the expansion of the breeding ranges,
as examplied by the large number of in-
tratropical migrants (e.g. Boyle & Conway
2007, Boyle et al. 2011). Comparative stu
dies of the occurrence of migration among
some of these tropical taxa, such as the pas-
serine group Tyranni revealed that a num-
ber of ecological traits, specically diet and
habitat, predict whether a species is migra-
tory or not in the tropics (Boyle & Conway
2007, Boyle et al
. 2011). Thus, yearround
variation and predictability of food sources
(Boyle & Conway 2007, Boyle et al. 2011),
as well as the ability to exploit these sour
ces (Bell 2011, Boyle et al. 2011) appears
to predispose some avian taxa for migration.
As the examples above suggest, the evolu-
tion of migration in birds is a complex prob-
lem that requires an integrative approach
combining aspects of the historical biogeo
graphy (range expansions), ecology (habi-
tat, food availability) and behavioral ecology
(diet specialization) of birds. Yet, phyloge-
netic studies often target only one of these
aspects, while ignoring others. Here, we de-
scribe an attempt for such an integrative app
roach using accipitrid birds of prey (Accipi
triformes) as a model group. This taxon is
ideal for our purpose because it includes both
tropical and temperate species and there is
wide variation in migratory behavior, habi-
tat and diet within the group (FergusonLees
ORNIS HUNGARICA 2014. 22(1)
18
& Christie 2001). Specically, we investigate
the following: (1) the phylogeny and histori
cal biogeography of diurnal birds of prey;
(2) the evolutionary origins of migration in
raptors in a phylogenetic context and (3)
the ecological and behavioral traits that are
associa ted with migratoriness in this group.
The traits we investigate are body mass, diet
and habitat. Body mass inuences nearly all
aspects of birds’ life and could be a crucial
factor determining which species can sur-
vive periods of food shortage and/or cold
wheather (e.g. Newton 2008). Therefore, we
hypothesize that larger birds are more like-
ly to become residents (or vice versa: resi-
dents might be selected to become larger)
(Tökölyi & Barta 2011). Diet can be impor-
tant for two reasons: rst, raptors feeding on
warmblooded prey or carcass are more like-
ly to survive the winter at temperate latitudes
and hence be nonmigratory. Second, species
with a generalist diet should be more likely
to subsist during periods of resource shortage
by nding alternative food sources, hence
we predict that food generalists are more fre-
quent in mig rants. Lastly, habitat type was
hypothesized to be important in the evolu-
tion of tropicaltemperate migration systems
because it could have determined the availa-
bility of sui table corridors for tropicaltem-
perate dispersal routes (Rappole & Jones
2002). Rappole and Jones (2002) noted that
the majo rity of longdistance migrants in the
Nearctic spend the winter in forests, whereas
almost none of the Palearctic/Asian migrants
do so. They proposed that the lack of forested
habi tats in North Africa could have acted as
a dispersal barrier, effectively ltering range
expansions from south. Therefore, we tested
whether habitat type is associated with mig
ratoriness in birds of prey.
Methods
Phylogenetic reconstruction and
molecular dating
The list of genes used to reconstruct the
molecular phylogeny of birds of prey for
Table 1. Gene sequences used to reconstruct the phylogeny of Accipitriformes, their availability,
length and the most appropriate evolutionary model of sequence evolution applying to
them, as evaluated by jModelTest
1. táblázat A vágómadár-alakúak törzsfa-rekonstrukciójához használt génszekvenciák neve, elérhető-
sége (fajok száma), hossza és a jModelTest által meghatározott legmegfelelőbb evolúciós
modell
Gene (Abbreviation) No species No. bp Model
12S ribosomal RNA (12S) 74 900 TIM2+I+G
16S ribosomal RNA (16S) 53 1527 GTR+I+G
ATP synthase F0 subunit 6 (ATP6) 56 684 TrN+I+G
ATP synthase F0 subunit 8 (ATP8) 56 168 TrN+I+G
β-bronigen intron 7 (BFI7) 69 922 TVM+G
Cytochrome c oxidase subunit 1 (COX1) 86 1551 TIM2+I+G
Cytochrome b (CYTB) 164 1146 TIM3+I+G
NADH-ubiquinone oxidoreductase subunit 2 (ND2) 151 1047 GTR+I+G
NADH-ubiquinone oxidoreductase subunit 2 (ND6) 66 525 GTR+I+G
Recombinase activating gene 1 (RAG1) 87 2872 GTR+I+G
19
J. Nagy & J. Tökölyi
which genetic data is available is shown in
Table 1. All sequences were retrieved from
GenBank (http://ncbi.nlm.nih.gov/). Se-
quences were aligned using MAFFTLINSI
(Katoh et al. 2005) with default parameters
and alignments were visually checked. Two
alignments (12S and 16S) contained mul-
tiple indels and were run through Gblocks
(Castresana 2000) to remove poorly aligned
positions. Sequence management was done
in the R statistical environment (R Develop-
ment Core Team 2012) using functions from
libraries ape (Paradis et al. 2004) and seqinr
(Charif & Lobry 2007).
Alignments were concatenated and spe-
cies with few data (<500 nucleotides) were
removed. The median sequence length for
the remaining 180 species was 1,038 base
pairs (range 5192,872) (Table 1). This taxo
nomic sample represents approximately
70% of extant species.
The resulting sparse supermatrix was
used to reconstruct the phylogenetic rela-
tionships of the 180 species. First, RAxML
(Stamatakis 2006) was used to obtain a
starting tree for phylogeny estimation. We
used a rapid bootstrap analysis with 100
bootstrap replicates followed by a search for
the bestsco ring maximum likelihood tree,
using the GTR+I+G model of evolution.
The Secretary Bird Sagittarius serpentinus
(Sagit tariidae) was used as an outgroup in
this process; the sister relationship between
Sagit tariidae and the rest of the Accipitri-
formes (Pandionidae and Accipitridae) is
well supported from molecular phylogene
tic studies (e.g. Ericson et al. 2006, Brown
et al. 2008, Hackett et al. 2008).
Next, the bestscoring tree obtained from
this analysis was used as a starting tree in
a Bayesian MCMC analysis (implemented
in BEAST; Drummond & Rambaut 2007)
to simultaneously reconstruct the phylo ge
ny and divergence times of birds of prey.
The ten gene segments were partitioned
separately and each gene segment was as-
signed its own bestt evolutionary mo del,
as evaluated by Akaike Information Cri-
terion (AIC) in the software jModelTest 2
(Guindon & Gascuel 2003, Darriba et al.
2012) (Table 1).
Molecular dating was done using an un-
correlated relaxed molecular clocks app
roach, which takes into account variation in
the rate of molecular evolution among line
ages (Drummond et al. 2006). Three fos-
sil constraints were used to date the phy-
logenetic tree (following do Amaral et al.
2009 and references therein): (1) the mini
mum age of divergence between Pandio-
nidae and Accipitridae was set to 37 Mya,
based on the oldest known fossil belon ging
to Pandionidae (Harrison & Walker 1976);
(2) the maximum age of divergence for Bu-
teo galapagoensis was set to 4 Mya and
(3) the maximum age of divergence for B.
soli tarius was set to 5.1 Mya. B. galapa go-
ensis and B. solitarius are both island spe-
cies (restricted to the Galapagos Islands
and Hawaii, respectively) and the latter two
age constraints are based on the assump-
tion that these species cannot be older then
the islands which they inhabit (do Ama
ral et al. 2009). Two independent BEAST
runs were performed, each allowed to run
for 50,000,000 generations with a thinning
interval of 5,000 generations. Convergence
was evaluated by checking effective sample
size (ESS) of parameters in Tracer (Ram-
baut & Drummond 2012). All parame
ters had ESS values >100 and most were
>>200. The two runs were combined (after
removing 10% burnin) and resampled at in-
tervals of 10,000 generations to yield 9,000
trees that represents a sample of the pos-
terior distribution of phylogenetic trees. A
ORNIS HUNGARICA 2014. 22(1)
20
maximum clade credibility tree was gene
rated from this sample in TreeAnnotator
(Rambaut & Drummond 2012).
To visualize diversication rate through
time, we created a lineagesthroughtime
plot for 100 trees selected randomly from
the posterior sample and the maximum
clade credibility tree.
Ancestral area reconstruction
We collected breeding season distribution
data on 180 species from FergusonLees
and Christie (2001). Species were scored
as present/absent in the following biogeo-
graphical realms: Nearctic, Palearctic, Neo
tropical, Afrotropical, Malagasy, Indoma-
layan, Australasian. The delimitation of
these realms is based on FergusonLees and
Christie (2001).
We inferred ancestral ranges based on
this distribution data by employing pro ba
bilistic historical biogeography methods
using the BioGeoBEARS R package (Matz-
ke 2013). These methods model geographic
range evolution by assuming different forms
of anagenetic and cladogenetic changes in
geographic distribution during speciation
events: dispersal, extinction, vicariance,
sympatric speciation and founderevent spe-
ciation. We evaluated which of the traditio
nally used historical biogeographic models
best ts range evolution in birds of prey by
calculating and comparing six models using
Akaike Information Criterion. These mo
dels are the DispersalVicariance (DIVA)
model (Ronquist 1997), the DispersalEx-
tinctionCladogenesis (DEC) model (Ree et
al. 2005, Ree & Smith 2008) and the Bay
Area model (Landis et al
. 2013), toge ther
with the combination of these three with
founderevent speciation. The three base-
line models all assume dispersal, extinction,
sympatric speciation and vicariance as pos-
sible range evolution mechanisms but differ
in the way they treat sympatric and vicariant
speciation events: the DIVA model allows
narrowscale sympatry but both narrow and
widespread vicariance. The DEC model as-
sumes narrowscale and subset sympatry,
but only narrowscale vicariance whereas
BayArea assumes narrowscale and wide
scale sympatry to occur (Matzke 2013).
We used the best t of these models to esti-
mate the most likely ancient distributions at
each node (ancestor state) of the phylogeny.
The method also gives a relative probabili-
ty, ranging from 0 to 1, which gives the pro
bability that the node was in the given state.
The higher this value, the higher is our con-
dence in the actual reconstuction is cor-
rect and uncertainity in the ancestral range
reconstruction is low. These analyzes were
done using the maximum clade credibility
tree.
Life history data
All data, with the exception of body mass
information, were collected from Fergu-
sonLees and Christie (2001), complemen
ted from the Global Raptor Information
Network (2013), if necessary. Migrato-
ry behavior was categorized based on pre-
vious phylogenetic studies (e.g. Kondo &
Omland 2007, do Amaral et al. 2009) as:
(1) nonmig ratory (no seasonal movements
pre sent), (2) partially migratory (part of the
populations, or part of the individuals with-
in the species perform regular seasonal
movements) and (3) completely migratory
(all populations and individuals migratory).
We used this variable to infer rates of evo-
lution to and from complete migration (see
below). However, the number of completely
migratory species was relatively low in our
21
J. Nagy & J. Tökölyi
sample (N=13), therefore, migration was bi-
narized in all other analyzes as either migra-
tory or nonmigratory.
Information on body mass was obtained
from Dunning (2008), and in a few cases
from FergusonLees and Christie (2001).
We used the average of male and female
body masses (logtransformed) when they
were available; however, in 8 cases data on
male or female body mass was available on-
ly. For 29 species no reliable body mass da-
ta could be found.
Diet (winter diet) was categorized follow-
ing Roulin and Wink (2004). These authors
assigned a relative importance value ran
ging from 1 to 9 to each of nine food cate
gories (live birds, mammals, reptiles, sh,
amphibians, crustaceans, insects, worms
and carrion) based on descriptions of indivi
dual species’ diet in FergusonLees and
Christie (2001). Food types that do not ap-
pear in the diet of a species received a score
of 9, whereas the most important food type
received a score of 1. From these values, we
calculated reliance on warmblooded prey
and carrion as the minimum of the impor-
tance scores received for bird or mammal
prey or carrion. Diet specicity was esti-
mated by counting the number of food types
in the diet of a given species that received a
score <9.
Finally we classied habitat type as open
or closed based on descriptions in Fergu-
sonLees and Christie (2001).
Comparative analyzes
We performed Bayesian ancestral state re-
construction in BEAST (Drummond &
Rambaut 2007) to infer the migratory be-
havior at ancestral nodes in the phylogeny
of birds of prey. Migratory behavior was re-
coded as a binary variable for this analysis
(as either migratory or nonmigratory, thus
complete and partial migrants were colla
ted). Bayesian ancestral state reconstruction
takes into account phylogenetic uncertain-
ty and calculates the probability that a gi
ven node was migratory or nonmigratory,
based on the trait values of its ancestor and
descendants. By taking into account uncer-
tainity in phylogenetic reconstruction, this
method is substantially better than parsimo-
nybased reconstructions, whose outcome
is conditional on a single (possibly errone-
ous) topology. In addition, Bayesian ances-
tral state reconstruction also takes into ac-
count differences in branch lengths, which
makes them more realistic than parsimo-
nybased methods.
Next, we investigated how changes in mi-
gratory behavior occurred on the phyloge-
ny by estimating transition rates between
the three levels of migratory behavior (i.e.
the rate of transition from migratory to par-
tial or complete migrant and vice versa, and
the rate of transition from partial to comp
lete migrant and vice versa). This analysis
was done using the MultiState module of
BayesTraits 1.0 (Pagel et al. 2004).
We determined whether body size (log
transformed), habitat type, diet breadth,
reliance on warmblooded prey and geo-
graphical location (Old vs. New World) af-
fects migratory behavior by constructing a
multivariate phylogenetic generalized line-
ar mixed models as implemented in the MC-
MCglmm package in R (Hadeld & Naka
gawa 2010), with these traits as dependent
variables. We also included the interaction
between habitat and geographical occur-
rence to model the differences in habitat use
among Old World and New World migrants.
Lastly, we tested the correlated evolu-
tion among migration and explanatory vari
ables found to be signicant in the multiva
ORNIS HUNGARICA 2014. 22(1)
22
riate analyses using the Discrete module of
BayesTraits 1.0 (Pagel & Meade 2006). This
method evaluates transition rates among
pairs of binary traits on a phylogeny revea
ling details of correlated evolution among
traits. For example, when analy zing the cor-
related evolution between migration and di-
et specicity one can ask whe ther mig ration
is more likely evolve in gene ralist (or spe-
cialist) lineages or vice versa: does a gene
ralist (or specialist) diet evolve more likely
in migrants? Since this analysis can hand
le only binary traits we dichotomized diet
specicity as specialist (<5 food types con-
sumed) or generalist (at least 5 food types
consumed).
Results
Phylogeny and diversication of
Accipitriformes
Figure 1 shows the relationship between
major lineages of raptors. Our analysis re-
covered the relationships among major line
ages of birds of prey described in previ-
ous molecular phylogenetic studies (Wink
& SauerGürth 2004, Lerner & Mindell
2005, Grifths et al. 2007). Crown group
Accipit riformes (i.e. the split between Sa-
git tarius and the rest of the species) is in-
ferred to have originated ~44 million years
ago (95% highest posterior density interval:
56.4 – 37.4), during the Eocene period. The
Figure 1. Simplied phylogeny showing major sublineages of Accipitriformes
1. ábra Egyszerűsített törzsfa a vágómadár-alakúak főbb csoportjainak logenetikai viszonyairól
23
J. Nagy & J. Tökölyi
split between Pandion and the rest of Ac-
cipitriformes occurred ~39.5 million years
ago (95% highest posterior density interval:
49.5–37). The two earliest branchings with-
in Accipitridae resulted in the appearance of
elanid kites (Elaninae) ~34.7 million years
ago (95% highest posterior density interval:
44.3–29.4) and the group containing Gy-
paetinae and Perninae (~27.7 million years
ago; 95% highest posterior density interval:
35.3–23.2). Then, around the start of the
Miocene period the diversication of rap-
tors accelerated and continued at a high rate
until recent times (Figure 2).
Ancestral area reconstruction
The best model describing range evolution
in birds of prey was the DEC model con-
taining founderevent speciation. Based on
this model, a southern origin was inferred
for all raptor subfamilies. These analyses
suggest that Accipitrinae, Aegypiinae and
Gypaetinae have an Afrotropical origin,
whereas Elaninae and Perninae derive from
the Neotropics. Buteoninae and Harpiinae
had a joint Afrotropical/Neotropical dis-
tribution according to our reconstruction,
whereas Aquilinae have an Afrotropical/
Neotropical and Indomalayan origin, al-
though we note that the accuracy of these
reconstructions is quite low (<0.3) (Table
2). Circeatinae were assigned an Indoma-
layan origin with relatively high probabili
ty (0.87) (Table 2). Lastly, the most likely
ancestral distribution of Haliaeetinae was
Australasia.
Figure 2. Lineages-through-time plot showing the pattern of diversication and accumulation of rap-
tor species through time as reconstructed by our multi-gene relaxed molecular dating ana-
lysis. Grey lines show random trees (N=100) from the posterior sample of dated ultrametric
trees obtained from BEAST; the black line denotes the maximum clade credibility tree
2. ábra A vágómadár-alakúak diverzikációja a molekuláris datálás alapján. A szürke vonalak a
törzsfa-rekonstrukció során létrehozott poszterior mintából 100 véletlenszerűen kiválasz-
tott ultrametrikus fa, a fekete vonal pedig az összegzett mintából számolt ultrametrikus fa
ágainak számát mutatja az idő függvényében
ORNIS HUNGARICA 2014. 22(1)
24
Subfamily Genera
Node age
(Mya; 95%
HPD interval)
Ancestral
distribution
Prob.
Accipitrinae
Accipiter, Circus, Kaupifalco, Melierax,
Urotriorchis
15.1; 23.0 At 0.79
Aegypiinae
Aegypius, Gyps, Necrosyrtes, Sarcogyps,
Torgos, Trigonoceps
7.4; 12.6 At 0.78
Aquilinae
Aquila, Hieraaetus, Ictinaetus,
Lophaetus, Nisaetus, Oroaetus,
Polemaetus, Spizaetus, Spizastur,
Stephanoaetus
9.8; 15.4 NtAtIm 0.26
Buteoninae
Busarellus, Butastur, Buteo, Buteogallus,
Geranoaetus, Geranospiza,
Harpyhaliaetus, Ictinia, Leucopternis,
Parabuteo, Rostrhamus
10.3; 16.7 NtAt 0.24
Circaetinae
Circaetus, Dryotriorchis, Pithecophaga,
Spilornis, Terathopius
13.8; 22.5 Im 0.87
Elaninae Elanus, Gampsonyx 14.4; 29.4 Nt 0.15
Gypaetinae
Gypaetus, Gypohierax, Neophron,
Polyboroides
18.5; 29.0 At 0.99
Haliaeetinae
Haliaeetus, Haliastur, Ichthyophaga,
Milvus
9.8; 15.7 Au 0.30
Harpiinae
Harpia, Harpyopsis, Macheiramphus,
Morphnus
13.6; 24.4 NtAt 0,20
Perninae
Aviceda, Chondrohierax, Elanoides,
Eutriorchis, Hamirostra, Leptodon,
Lophoictinia, Pernis
18.7; 29.4 Nt 0.15
Table 2.
Geographic origin of major sublineages of Accipitriformes as inferred from the ancestral
area reconstruction following a dispersal-extinction-cladogenesis-founder-event speci-
ation (DEC+J) model of geographic area evolution. This table lists the major subfamilies
of Accipitriformes, the genera they contain, the inferred age of these groups (age of the
crown group, i.e. the most recent common ancestor of all extant species), shown as 95%
highest posterior density interval (HPD) and the most likely ancestral area inferred for the
crown group, along with the probability that the node was in the given state. Geogra phic
distribution is abbreviated as: Pa – Palaearctic; Na – Nearctic; At – Afrotropical; Nt – Neo-
tropical; Im – Indomalayan; Au – Australasian
2. táblázat A vágómadár-alakúak főbb csoportjainak rekonstruált ősi elterjedési területe. A táblázat
a főbb alcsaládok nevét, az ezekbe sorolt génuszokat, az alcsaládok molekuláris datálás
által becsült korát, a legvalószínűbb ősi elterjedési területet, illetve ennek a valószínűsé-
gét mutatja. A földrajzi elterjedések rövidítése: Pa – Palearktisz; Na – Nearktisz; At – Afrot-
ropisz; Nt – Neotropisz; Im – Indomaláj; Au – Ausztrálázsiai
25
J. Nagy & J. Tökölyi
Species Distribution
Ancestor’s age
(Mya; 95% HPD
interval)
Ancestor’s
posterior
node support
Prob. of
ancestor
being non-
migratory
Ancestor’s
distribution
Prob.
Accipiter gularis Pa 0; 0.5 0.95 0.98 PaIm 0.56
Accipiter nisus PaAtIm 2.4; 5.2 0.42 0.84 NtPaAtIm 0.21
Accipiter soloensis PaIm 6.3; 13.3 0.92 0.96 ImAu 0.18
Accipiter striatus NaNt 1.3; 3.3 0.53 0.84 NaNtAt 0.51
Aegypius monachus PaAt 2; 4 1.00 0.98 At 0.46
Aquila chrysaetos NaPaAt 3.6; 6.2 1.00 0.99 At 0.91
Aquila heliaca Pa 0.5; 1.6 0.94 0.99 Pa 1.00
Aquila nipalensis Pa 1.6; 3.5 1.00 0.97 Pa 0.99
Hieraaetus wahlbergi At 3.8; 6.6 1.00 0.59 At 1.00
Butastur indicus PaIm 2.1; 5.1 0.83 0.56 Im 0.32
Butastur rupennis At 2.9; 6.4 1.00 0.57 At 0.29
Buteogallus anthracinus NaNt 0; 0.1 1.00 1.00 Nt 0.95
Buteogallus meridionalis Nt 2.7; 4.5 1.00 1.00 Nt 1.00
Buteo lineatus Na 1.1; 2.1 1.00 0.79 NaNt 0.38
Buteo nitidus NaNt 3; 4.8 0.32 0.66 Na 0.71
Buteo platypterus NaNt 2.8; 4.5 0.99 0.70 Na 0.72
Buteo polyosoma Nt 0.4; 1.1 1.00 0.99 Nt
1.00
Buteo swainsoni Na 0.2; 0.4 1.00 0.99 Nt 0.44
Circaetus fasciolatus At 1.5; 3.9 1.00 0.91 At 1.00
Circaetus gallicus PaAtIm 2.8; 6.1 1.00 0.78 At 0.78
Circaetus pectoralis At 4.7; 8.8 1.00 0.79 At 0.99
Elanoides forcatus NaNt 15.5; 25.1 0.53 0.84 NtPa 0.22
Gypohierax angolensis At 18; 27.8 0.43 0.88 At 0.90
Gyps coprotheres At 0.3; 0.8 0.35 1.00 At 0.40
Gyps fulvus PaAt 0.3; 0.9 0.66 1.00 At 0.96
Haliaeetus leucogaster ImAu 0; 0.3 1.00 0.89 Au 0.87
Haliaeetus vocifer At 0.7; 1.9 1.00 0.91 At 1.00
Haliastur sphenurus Au 2.3; 4.8 1.00 0.70 Au 0.37
Hieraaetus ayresii At 2.4; 4.7 1.00 0.51 At 0.88
Hieraaetus pennatus PaAt 0.7; 1.7 1.00 0.53 PaAtAu 0.28
Macheiramphus alcinus At 13.6; 24.4 0.99 1.00 NtAt 0.20
Neophron percnopterus PaAtIm 14.4; 24.7 1.00 0.87 At 0.42
Pandion haliaetus NaNtPaAtImAu 37; 49.5 1.00 0.91 At 0.37
Parabuteo unicinctus NaNt 4.5; 7.5 1.00 0.98 Nt 0.95
Polyboroides typus At 18.5; 29 0.95 0.91 At 0.99
Rostrhamus sociabilis NaNt 7.5; 12 0.96 0.99 Nt 0,96
Table/táblázat 3A
ORNIS HUNGARICA 2014. 22(1)
26
Node
number
Distribution
Ancestor’s age
(Mya; 95% HPD
interval)
Ancestor’s
posterior node
support
Prob. of
ancestor
being non-
migratory
Ancestor’s
distribution
Prob.
185 NaNtPaAtImAu 14.4; 29.4 1.00 0.77 Nt 0.15
217 Nt 11.7; 18.2 1.00 0.63 Nt 0.71
251 NaPa 7.1; 11.9 1.00 0.93 Im 0.10
255 PaAtImAu 5.3; 9.3 1.00 0.69 Au 0.38
261 NaNt 9.5; 14.5 0.90 0.99 Nt 0.93
344 Pa 2.1; 4.4 0.98 1.00 Pa 0.81
359 Pa 15.5; 25.1 0.84 0.53 NtPa 0.22
Table 3. Possible cases of independent appearance of migration in Accipitriformes. This table lists
migratory species (A) or nodes that were inferred to be migratory with a probability >0.8
(B) with non-migratory ancestors. The probability that the ancestor was non-migratory,
as inferred from ancestral character estimation, is shown along each possible case. Note
that only those cases are listed where this probability is >0.5, i.e. the node is more like-
ly to be non-migratory than migratory. The greater this value, the higher is our con-
dence that migration appeared on this branch. Also shown are the ancestral nodes’ pos-
terior support, the 95% highest posterior density (HPD) interval of the nodes’ age, along
with the most likely distribution of these nodes and the probability that the node was
in this state, as inferred in the ancestral area reconstruction. Geographic distribution is
abbrevia ted as: Pa – Palaearctic; Na – Nearctic; At – Afrotropical; Nt – Neotropical; Im – In-
domalayan; Au – Australasian. Node numbers appearing in (B) are shown in Figure 3
3. táblázat A vágómadár-alakúak törzsfejlődése során a vonulás egymástól függetlenül többször
megjelent. Ez a táblázat azokat a vonuló fajokat (A) vagy közös ősöket (B) mutatja, ame-
lyek nagy valószínűséggel vonulók voltak (>0.8 valószínűséggel). Mindegyik esetben fel
van tüntetve: (i) a vonuló faj (ős) jelenlegi (becsült) előfordulása, (ii) a közvetlen ősük be-
csült kora, (iii) a közvetlen ős logenetikai helyzetének pontossága, (iv) annak a valószí-
nűsége, hogy a közvetlen ős nem-vonuló volt, (v) a közvetlen ős rekonstruált ősi elterje-
dési területe, illetve (vi) annak a valószínűsége, hogy az ősi elterjedés ténylegesen ebben
az állapotban volt. Csak azok az esetek szerepelnek, ahol a közvetlen ős legalább 0.5 való-
színűséggel helytülő volt. Minél nagyobb ez az érték, annál bizonyosabb, hogy a vonulás
ténylegesen ezen az ágon alakulhatott ki. A földrajzi elterjedések rövidítése: Pa – Paleark-
tisz; Na – Nearktisz; At – Afrotropisz; Nt – Neotropisz; Im – Indomaláj; Au – Ausztrálázsiai.
A (B)-ben szereplő számok közös ősöket jelölnek, amelyeknek helyzete a 3. ábrán látható
Table/táblázat 3B
Figure 3. Bayesian ancestral state reconstruction of migratory behavior in Accipitriformes. Tip labels
show migratory behavior in extant species (white: non-migratory; black: partial or com-
plete migrant). Pie charts labelling the nodes show the probability that the given species
was migratory (black) or non-migratory (white). A completely black chart indicates that the
ancestor was migratory with high posterior probability. The size of the charts is proportio-
nal to the posterior support of specic nodes: small charts indicate high uncertainity in phy-
logenetic reconstruction
3. ábra Vonulási viselkedés rekonstrukciója a vágómadár-alakúak törzsfáján. A fajnevek melletti
négyzetek színe az illető fajok vonulási viselkedését jelöli (fehér – helytülő; fekete – részle-
ges vagy teljes vonuló). A közös ősöket jelölő kördiagrammok mutatják annak a valószínű-
ségét, hogy az illető faj vonuló volt (fekete). A kördiagrammok mérete egyenesen arányos
az illető csomópontok rekonstrukciójának pontosságával
27
Evolution of migration in Accipitri-
formes
Table 3 shows possible cases of independent
appearance of migratory behavior in Accipit
riformes. Since the presence of migration
is quite variable among species, our recon-
struction of ancestral states involves conside
rable uncertainty in some nodes. On the other
hand, in several cases (e.g. Buteo hawks, Ac-
cipiter hawks or Haliaeetus eagles), close-
ly related species are all migratory, with the
consequence that the most likely state for the
ancestor of these species is being migratory.
Two important patterns can be seen
from this list of evolutionary events. First,
J. Nagy & J. Tökölyi
Figure/ábra 3A
ORNIS HUNGARICA 2014. 22(1)
28
Figure/ábra 3B
29
J. Nagy & J. Tökölyi
most of the independent events leading to
the appearance of migration involve single
species (Table 3A). In these situations, the
upper bound for the time of appearance of
migration is the age of that particular spe-
cies (i.e. when it split off from its ances-
tor). Hence, migration could have appeared
in these lineages recently; this possibility
is supported by the observation that most
of these upper bounds are not older than
the beginning of the Pliocene epoch (~5.3
Mya). On the other hand, nodes that are re-
constructed as migratory provide lower esti-
mates for the origin of migration (Table 3B).
The oldest of these nodes is the common an-
cestor of the Goshawk and the harriers, with
an estimated age of 12.9 Mya (95% highest
posterior density interval: 16.3–10.1 Mya).
Second, a tropical or partly tropical an-
cestral area was inferred for the ancestor of
most migratory lineages. There are very few
exceptions from this pattern. For instance,
our results suggest that Aquila heliaca and
A. nipalensis (both migratory) evolved with-
in the Palearctic from nonmigratory ances-
tors. Similarly, two migratory Buteo hawks
(B. platypterus and B. nitidus) seem to have
evolved in the Nearctic from nonmigratory
ancestors, although in these latter cases pos-
terior support for the phylogenetic recon-
struction is quite low (0.32 for the ancestor
of B. nitidus). In sum, our results provide
broad support for a tropical origin of migra-
tory species.
Comparative analyzes
Analysis of the transition rates using
BayesTraits Multistate module revealed
that both complete and partial migrations
evolved from a nonmigratory state, but
evolutionary transitions between partial and
complete migration or vice versa are very
low (Figure 4), suggesting that partial mig
ration is not a transitional state between
complete migration or lack of migration.
Multivariate analysis of the ecological fac-
tors inuencing migration suggests that only
diet breadth and habitat are associated with
migration (Table 4). However, when this
multivariate model is simplied by back-
ward elimination of nonsignicant parame-
ters, habitat type does not remain signicant.
Hence, the only ecological factor associated
signicantly with migration was winter diet
breadth: species with a more generalist win-
ter diet are more likely to be migratory.
Finally, we performed directional tests be-
tween migratory behavior on one hand and
geographic distribution and diet breadth,
res pectively, on the other. In the rst case,
we found that evolutionary transitions to
Figure 4. Evolutionary tran-
sition rates be-
tween levels of
migratoriness in
birds of prey
4. ábra Vonulási stratégi-
ák közötti evolúci-
ós tranzíciós ráták
ORNIS HUNGARICA 2014. 22(1)
30
mig ration occured with a higher rate in lin-
eages with a tropical distribution (9.08 vs.
0.73 in lineages with a nontropical distri-
bution) (Figure 5A). Furthermore, transi-
tions to a nontropical distribution are much
more likely in migratory (30.01) than in
nonmig ratory lineages (0.15). The tran-
sition rates also revealed that switches in
the geogra phic distribution from a tropical
to a nontropical distribution or vice versa
are virtually lacking in nonmigratory line-
ages, whereas they occur at a relatively high
rate in migrants (Figure 5A). In the second
case, transition rates indicate that migration
Parameter
estimate
lower 95%
condence
interval
upper 95%
condence
interval
P-value
Habitat (Forest/Open) 1.30 0.13 2.27 0.02
Old/New World 0.78 -0.75 1.75 0.16
log(body mass) -0.06 -0.51 0.33 0.78
Diet generalism 0.44 0.21 0.75 <0.01
Reliance on warm-blooded prey or carcass 0.00 -0.21 0.23 0.99
Habitat : Old/New World interaction -1.39 -3.35 0.40 0.18
Table 4.
Factors aecting the occurrence of migration in diurnal birds of prey (N=151 species); pa-
rameter estimates, their 95% condence intervals and P-values from a multivariate mixed
eect models controlling for phylogeny
4. táblázat A vonulás előfordulását befolyásoló tényezők vágómadár-alakúaknál (N=151 faj); loge-
netikai viszonyokra kontrollált többváltozós kevert lineáris modellből származó becsült
értékek, azok 95%-os kondenciaintervalluma és a P-értékek
Figure 5. Results from pairwise directional test between migratory behavior (migratory or non-mi-
gratory) and (A) geographic distribution: tropical (species not present in the Nearctic or the
Palearctic) or non-tropical (species present in the Nearctic or the Palearctic); (B) diet speci-
city: specialist (consumes <5 types of food) or generalist (at least 5 food types consumed).
The graph show transition rates among pairs of traits indicating the rate with which these
evolutionary changes are inferred to have occurred on the phylogeny
5. ábra A vonulási viselkedés és a földrajzi elterjedés (A) illetve a táplálékspecializáció (B) közötti di-
rekcionális tesztek eredményei. A vonulási viselkedés kódolása: vonuló (részlegesen vagy
teljesen) vagy nem vonuló. A földrajzi elterjedés kódolása: trópusi (nem fordul elő sem a Pa-
learktiszban sem a Nearktiszban) vagy nem trópusi (az előző ellentettje). A táplálékspeciali-
záció kódolása: specialista (<5 tápláléktípust fogyaszt) vagy generalista (az előző ellentett-
je). Az ábra a különböző jellegpárok közötti evolúciós tranzíciós rátákat mutatja
31
J. Nagy & J. Tökölyi
is more likely to arise in specialist lineages
and that a generalist diet is more likely to
evolve in migrants than in nonmigrants.
Hence, it appears that diet breadth evolves
as in response to the selective environments
imposed by migratoriness, rather than pre-
disposing species for migration.
Discussion
Phylogeny of Accipitriformes
The phylogenetic relationships among and
within major lineages of Accipitriformes
have been extensively studied before (e.g.
Wink & SauerGürth 2004, Helbig et al.
2005, Lerner & Mindell 2005, Griffths et
al. 2007, do Amaral et al. 2009). Here, we
combined all available genetic information
to produce a multigene phylogeny of Ac-
cipitriformes with a broad taxonomic sam-
pling, including approximately twothirds of
extant species of accipitrid birds of prey. The
phylogenetic hypothesis obtained from this
analysis is broadly congruent with previous
reports showing that several traditionally es-
tablished clades are in fact polyphyletic or
paraphyletic. For instance, Old World vul-
tures form a polyphyletic clade comprised
of: (1) Gypaetinae which is monophylet-
ic with Perninae and includes the Bearded
Vulture Gypaetus barbatus and the Egyptian
Vulture Neophron percnopterus and (2) Ae-
gypiinae which contains all remaining Old
World vultures and is the sister clade of Cir-
caetinae. Accipiter
hawks are paraphyle tic
and should include harriers (Circus spp.),
which are closely related to the clade con-
taining goshawks (see also Breman et al.
2013 for a more detailed analysis). In addi-
tion, we also observed widespread paraphy-
ly in aquiline eagles and buteonine hawks, as
repor ted previously (Helbig et al. 2005, do
Amaral et al. 2009).
On the other hand, we also observed seve
ral discrepancies in the higher level relation-
ships of Accipitridae between our reconstruc-
tions and those of obtained from previous
studies (e.g. Lerner & Mindell 2005, Grif-
ths et al. 2007). For instance, we recovered
Aqui linae and Harpiinae as sister clades, al-
beit with relatively low support (posterior
probability: 0.45). Harpagus kites were in-
ferred as the sister group of the clade con-
taining Buteoninae and Haliaeetinae with
relatively high support (posterior probability:
0.83). Lastly, the sister relationship between
the clade containing Aquilinae and Harpii-
nae on one hand and Buteoninae, Haliaeeti-
nae and Accipitrinae on the other was strong-
ly supported (posterior probability: 1).
Historical biogeography and evolution of
migration in Accipitriformes
Our ancestral state reconstruction suggests
that migratory behavior in birds of prey
evolved multiple times. Most of these ap-
pear to be relatively recent events (occur-
ring during the Pliocene or Pleistocene, i.e.
<5 million years ago). In one case howe ver,
migratory behavior appears to be much more
ancient. In true hawks (Accipitrinae) migra-
tion appears to have evolved approxi mately
1412 million years ago, during the middle
of the Miocene period. By compari son, do
Amaral et al. (2009) reconstructed the ori-
gin of migration in one Buteo clade at app
roximately 5 million years ago, a result that
is supportedby our analyses (Figure 3B).
More direct estimates based on the age dis-
tribution of fossilized individuals (speci-
cally, the lack of juveniles) found at Olduvai
Gorge, in Tanzania, suggest that this site was
a wintering location of shorebirds belon
ORNIS HUNGARICA 2014. 22(1)
32
ging to Charadriidae 1.91.74 million years
ago, implying that migration was present at
this time (Louchart 2008). Hence, our esti-
mate for the origin of migration in accipitrid
hawks appears to be one of the oldest dates
published so far. Such estimates are impor-
tant (yet remarkably lacking) if we are to un-
derstand the evolution of migration in a con-
stantly changing spatiotemporal context at a
global scale (Louchart 2008).
The middle of the Miocene period saw a
series of global cooling events (Zachos et
al. 2001), which resulted in the expansion
of grasslands and contraction of forest habi
tats, possibly opening new niches for birds
of prey. Our analyses suggest that accipitrid
hawks appeared shortly before this period,
probably in the Afrotropical realm and colo-
nized other parts of the world shortly there-
after. Since most extant species belonging
to this lineage are at least partly migrato-
ry today, it is likely that their ancestor al-
so performed seasonal migratory move-
ments. Alternatively, migration could have
evolved separately in these lineages due
to similar selective environments (i.e. as a
consequence of convergent evolution ra
ther than shared phylogenetic background).
Since migration is a phylogenetically labile
trait that can evolve very quickly (see e.g.
Zink 2011), independent evolution in mul-
tiple lineages experiencing similar selective
environments is a plausible scenario for the
occurrence of migratory behavior in clusters
of closely related species. However, this ex-
planation is clearly less parsimonious in ex-
plaining the evolution of migration in true
hawks, since this group contains both tem-
peratetropical and intratropical migrants
on different continents, which would im-
ply simultaneous, independent selection for
mig ration in a wide variety of different en-
vironments on different parts of the world.
Joint reconstruction of ancestral distribu-
tion and migratory behavior suggests that in
raptors, migration appeared mostly in spe-
cies with a southern origin. This is further
strengthened by our directional analyses,
which suggest that migration is more like-
ly to evolve in tropical species and that mig
ratory raptors are more likely to switch to a
nontropical breeding range, hence sugges
ting that migratory behavior and range ex-
pansions are evolving in parallel. Interes
tingly, we also found that the transition rate
from a tropical to nontropical distribution
(and vice versa) is very low in nonmigra-
tory raptors, but not in migrants, suggesting
that migration greatly enhanced range ex-
pansions in this group of birds.
The southern origin of migratory raptors is
in line with previous studies obtaining simi-
lar results in a variety of taxonomic groups
(Joseph et al. 1999, Outlaw et al. 2003, Milá
et al
. 2006). We have to emphasize, howe ver,
that this result helps little in understanding
the evolution of migration in birds of prey.
As we have shown, all major lineages with-
in Accipitridae trace back their origin to one
of the southern biogeographic realms. Hence,
both migratory and nonmigratory species
currently inhabiting the temperate zone des
cend from the tropics. Our analyses do sug-
gest, however, that migration is more likely
to emerge in the tropics than in the tempe
rate zone, and that migratory birds are more
likely to disperse and switch from a tropi-
cal distribution to a nontropical one. Hence,
the relationship between colonization of the
temperate zone and the evolution of migra-
tion could be the reverse of what traditional-
ly is assumed, i.e. migratory birds (intratropi
cal migrants) more likely to colonize novel
habitats and expand to North. This hypothe-
sis could be tested in the future by investiga
ting the ecological and behavioral traits pro-
33
J. Nagy & J. Tökölyi
moting the colonization of temperate habitats
in a broader sample of birds.
Since all birds of prey are of tropical an-
cestry, their ancestors must have undergone
range expansions to the temperate region.
Yet, not all of these species became migra-
tory. We found that winter diet specialization
predicts the occurrence of migration in acci
pitrid birds of prey, with migratory species
relying on more variable diets. The ability
to feed on a wide source of food types could
greatly enhance the probability that a species
survives the winter in the temperate zone
where food availability is much lower during
the winter (Newton 2008). Hence, this could
at least partly explain interspecic differen
ces in migratoriness. Interestingly, we found
no association between migration and reli-
ance on warmblooded prey or carcass, sug-
gesting that these food sources alone might
not be enough to sustain most species in the
temperate zone during winter. For instance,
two of the four vulture species that occur in
Europe (the Griffon Vulture Gyps fulvus, and
the Egyptian Vulture) are migratory, despite
the fact that their major food source – carcass
–
is most likely available yearround. How-
ever, these food sources might also show sea-
sonal uctuations (e.g. Kendall et al. 2012).
Alternatively, other factors, such as selection
for early breeding or extended breeding sea-
son could generate differences in migrato-
riness between populations or species (e.g.
Tökölyi & Barta 2011, Camacho 2013). Fur-
ther work is required to clarify ecological de-
terminants of migration in birds of prey.
Acknowledgements
We would like to express our gratitude to
the two anonymous referees for their useful
comments on our manuscript. J. N. was sup-
ported by a scholarship from the Universi-
ty of Debrecen’s Talent Development Prog
ram (DETEP). J. T. was supported by the
TÁMOP 4.2.2.C11/1/KONV20120010
project; the project is conanced by the Eu-
ropean Social Fund and the European Re-
gional Development Fund. Computations
were partly run on Hungarian National In-
frastructure Institute supercomputers (http://
niif.hu). We are grateful to József Büki, head
of András Keve Library for Ornithology and
Nature Conservation for his help in acces
sing literature in an early part of this study.
Alerstam, T. 1993. Bird migration. – Cambridge Uni-
versity Press
do Amaral, F. R., Sheldon, F. H., Gamauf, A., Har-
ing, E., Riesing, M., Silveira, L. F. & Wajntal, A.
2009. Patterns and processes of diversication in a
widespread and ecologically diverse avian group,
the buteonine hawks (Aves, Accipitridae). – Mo-
lecular Phylogenetics and Evolution 53: 703–715.
DOI:10.1016/j.ympev.2009.07.020
Bell, C. P. 2000. Process in the evolution of bird mig
ration and pattern in avian ecogeography. – Jour-
nal of Avian Biology 31: 258–265. DOI:10.1034/
j.1600048X.2000.310218.x
Bell, C. P. 2005. The origin and development of bird mig
ration: comments on Rappole and Jones, and an al-
ternative evolutionary model. – Ardea 93: 115–123.
Bell, C. P. 2011. Resource buffering and the evolution
of bird migration. – Evolutionary Ecology 25: 91–
106. DOI:10.1007/s1068201093834
Berthold, P. 2001. Bird migration: a general survey. –
Oxford University Press
Boyle, W. A. & Conway, C. J. 2007. Why migrate?
A test of the evolutionary precursor hypothe-
sis. – The American Naturalist 169: 344–359.
DOI:10.1086/511335
Boyle, W. A., Conway, C. J. & Bronstein, J. L. 2011.
Why do some, but not all, tropical birds migrate?
A comparative study of diet breadth and fruit pre
ference. – Evolutionary Ecology 25: 219–236.
DOI:10.1007/s1068201094034
Breman, F. C., Jordaens, K., Sonet, G., Nagy, Z. T.,
Van Houdt, J. & Louette, M. 2013. DNA barco
References
ORNIS HUNGARICA 2014. 22(1)
34
ding and evolutionary relationships in Accipiter
Brisson, 1760 (Aves, Falconiformes: Accipitri-
dae) with a focus on African and Eurasian repre-
sentatives. – Journal of Ornithology 154: 265–287.
DOI:10.1007/s1033601208925
Brown, J. W., Rest, J. S., García–Moreno, J., Soren-
son, M. D. & Mindell, D. P. 2008. Strong mito-
chondrial DNA support for a Cretaceous origin
of modern avian lineages. – BMC Biology 6: 6.
DOI:10.1186/1741700766
Bruderer, B. & Salewski, V. 2008. Evolution of bird
migration in a biogeographical context. – Journal
of Biogeography 35: 1951–1959. DOI:10.1111/
j.13652699.2008.01992.x
Camacho, C. 2013. Tropical phenology in temperate
regions: extended breeding season in a longdis-
tance migrant. – The Condor 115: 830–837.
DOI:10.1525/cond.2013.120192
Castresana, J. 2000. Selection of conserved blocks from
multiple alignments for their use in phylogenetic
analysis. – Molecular Biology and Evolution 17:
540–552.
Charif, D. & Lobry, J. R. 2007. SeqinR 1.02: A con-
tributed package to the R project for statistical
compu ting devoted to biological sequences ret
rieval and analysis. pp. 207–232. – In: Bastolla,
U., Porto, M., Roman, H. E. & Vendruscolo, M.
(eds.) Structural approaches to sequence evolution.
– Springer, Berlin
Cox, G. W. 1985. The evolution of avian migration sys-
tems between temperate and tropical regions of the
New World. – The American Naturalist 126: 451–
474.
Darriba, D., Taboada, D. L., Doallo, R. & Posada, D.
2012. jModelTest 2: more models, new heuristics
and parallel computing. – Nature Methods 9: 772.
DOI:10.1038/nmeth.2109
Drummond, A. J., Ho, S. Y., Phillips, M. J. & Rambaut,
A. 2006. Relaxed phylogenetics and dating with
condence. – PLoS Biology 4: e88. DOI:10.1371/
journal.pbio.0040088
Drummond, A. & Rambaut, A. 2007. BEAST: bayesian
evolutionary analysis by sampling trees. – BMC
Evolutionary Biology 7: 214. DOI:10.1186/1471
2148-7-214
Dunning, J. B. 2008. CRC handbook of avian body
masses. – CRC Press, Boca Raton
Ericson, P. G. P., Anderson, C. L., Britton, T., Elzanows-
ki, A., Johansson, U. S., Källersjö, M., Ohlson, J. I.,
Parsons, T. J., Zuccon, D. & Mayr. G. 2006. Diver-
sication of Neoaves: integration of molecular se-
quence data and fossils. – Biology Letters 2: 543–
547. DOI:10.1098/rsbl.2006.0523
FergusonLees, J. & Christie, D. A. 2001. Raptors of the
World. – Houghton Mifin Company, Boston
Global Raptor Information Network. 2013. Species ac-
counts. – Downloaded from http://www.globalrap-
tors.org.
Grifths, C. S., Barrowclough, G. F., Groth, J. G. &
Mertz. L. A. 2007. Phylogeny, diversity, and clas-
sication of the Accipitridae based on DNA se-
quences of the RAG1 exon. – Journal of Avian
Biology 38: 587–602. DOI:10.1111/j.2007.0908
8857.03971.x
Guindon, S. & Gascuel, O. 2003. A simple, fast, and
accurate algorithm to estimate large phylogenies
by maximum likelihood. – Systematic Biology 52:
696–704. DOI:10.1080/10635150390235520
Gwinner, E. 1990. Bird migration: physiology and eco-
physiology. – SpringerVerlag, Berlin
Hackett, S. J., Kimball, R. T., Reddy, S., Bowie, R. C.
K., Braun, E. L., Braun, M. J., Chojnowski, J. L.,
Cox, W. A., Han, K., Harshman, J., Huddleston, C.
J., Marks, B. D., Miglia, K. J., Moore, W. S., Shel-
don, F. H., Steadman, D. W., Witt, C. C. & Yuri, T.
2008. A phylogenomic study of birds reveals their
evolutionary history. – Science 320: 1763–1768.
DOI:10.1126/science.1157704
Hadeld, J. D. & Nakagawa, S. 2010. General quantita-
tive genetic methods for comparative biology: phy-
logenies, taxonomies, and multitrait models for
continuous and categorical characters. – Journal of
Evolutionary Biology 23: 494–508. DOI:10.1111/
j.14209101.2009.01915.x
Harrison, C. J. O. & Walker, C. A. 1976. Birds of
the British upper Eocene. – Zoological Jour-
nal of the Linnean Society 59: 323–351.
DOI:10.1111/j.10963642.1976.tb01017.x
Helbig, A. J., Kocum, A., Seibold, I. & Braun, M. J.
2005. A multigene phylogeny of aquiline eagles
(Aves: Accipitriformes) reveals extensive para-
phyly at the genus level. – Molecular Phylogene
tics and Evolution 35: 147–164. DOI:10.1016/j.
ympev.2004.10.003
Jetz, W., Thomas, G. H., Joy, J. B., Hartmann, K.
& Mooers, A. O. 2012. The global diversity of
birds in space and time. – Nature 491: 444–448.
DOI:10.1038/nature11631
Joseph, L., Lessa, E. P. & Christidis, L. 1999. Phyloge-
ny and biogeography in the evolution of migration:
shorebirds of the Charadrius complex. – Journal of
Biogeography 26: 329–342. DOI:10.1046/j.1365
2699.1999.00269.x
Katoh, K., Kuma, K., Toh, H. & Miyata, T. 2005.
MAFFT Version 5: improvement in accuracy of
multiple sequence alignment. – Nucleic Acids Re-
search 33: 511–518. DOI:10.1093/nar/gki198
Kendall, C., Virani, M. Z., Kirui, P., Thomsett, S. &
Githiru, M. 2012. Mechanisms of coexistence in
vultures: understanding the patterns of vulture
35
J. Nagy & J. Tökölyi
abundance at carcasses in Masai Mara National Re-
serve, Kenya. – The Condor 114: 523–531.
Kondo, B. & Omland, K. E. 2007. Ancestral state re-
construction of migration: multistate analysis re-
veals rapid changes in New World orioles (Icterus
spp.). – The Auk 124: 410–419. DOI:10.1642/0004
8038(2007)124[410:ASROMM]2.0.CO;2
Landis, M. J., Matzke, N. J., Moore, B. R. & Huelsen-
beck, J. P. 2013. Bayesian analysis of biogeogra-
phy when the number of areas is large. – Systematic
Biology 62: 789–804. DOI:10.1093/sysbio/syt040
Lerner, H. R.L. & Mindell, D. P. 2005. Phylogeny of
eagles, old world vultures, and other Accipitridae
based on nuclear and mitochondrial DNA. – Mo-
lecular Phylogenetics and Evolution 37: 327–346.
DOI:10.1016/j.ympev.2005.04.010
Louchart, A. 2008. Emergence of long distance bird
mig rations: a new model integrating global climate
changes. – Naturwissenschaften 95: 1109–1119.
DOI:10.1007/s0011400804353
Matzke, N. J. 2013. Probabilistic historical biogeogra-
phy: new models for founderevent speciation, im-
perfect detection, and fossils allow improved accu-
racy and modeltesting. – Ph.D. thesis, Department
Integrative Biology and Designated Emphasis in
Computational and Genomic Biology, Universi-
ty of California, Berkeley. Available at http://phy-
lo.wikidot.com/localfiles/biogeobears/Matzke_
PhD_FINAL_v5_w_refs.pdf.
Milá, B., Smith, T. B. & Wayne, R. K. 2006. Postgla-
cial population expansion drives the evolution of
long–distance migration in a songbird. – Evolution
60: 2403–2409. DOI:10.1111/j.00143820.2006.
tb01875.x
Newton, I. 2008. The migration ecology of birds. – Aca
demic Press, London
Outlaw, D. C., Voelker, G., Milá, B. & Girman, D. J.
2003. Evolution of longdistance migration in and
historical biogeography of Catharus thrushes: a
molecular phylogenetic approache. – The Auk 120:
299–310. DOI:10.1642/00048038(2003)120[0299:
EOLMIA]2.0.CO;2
Padian, K. & Chiappe, L. M. 1998. The origin and early
evolution of birds. – Biological Reviews 73: 1–42.
DOI:10.1111/j.1469185X.1997.tb00024.x
Pagel, M., Meade, A. & Barker, D. 2004. Bayesian
estimation of ancestral character states on phy-
logenies. – Systematic Biology 53: 673–684.
DOI:10.1080/10635150490522232
Pagel, M. & Meade, A. 2006. Bayesian analysis of cor-
related evolution of discrete characters by rever sib
lejump Markov Chain Monte Carlo. – The Ameri-
can Naturalist 167: 808–825. DOI:10.1086/503444
Paradis, E., Claude, J. & Strimmer, K. 2004. APE: ana
lyses of phylogenetics and evolution in R language.
– Bioinformatics 20: 289–290. DOI:10.1093/bioin-
formatics/btg412
Rappole, J. H. & Jones, P. 2002. Evolution of Old and
New World migration systems. – Ardea 90: 525–
537.
R Development Core Team. 2012. R: a language and
environment for statistical computing. – R Founda-
tion for Statistical Computing, Vienna. Available at
http://www. R–project.org.
Ree, R. H., Moore, B. R., Webb, C. O. & Donoghue,
M. J. 2005. A likelihood framework for infer-
ring the evolution of geographic range on phy-
logenetic trees. – Evolution 59: 2299–2311.
DOI:10.1111/j.00143820.2005.tb00940.x
Ree, R. H. & Smith, S. A. 2008. Maximum likelihood in-
ference of geographic range evolution by dispersal,
local extinction, and cladogenesis. – Systematic Bio
logy 57: 4–14. DOI:10.1080/10635150701883881
Ronquist, F. 1997. Dispersalvicariance analysis: a new
approach to the quantication of historical bio-
geography. – Systematic Biology 46: 195–203.
DOI:10.1093/sysbio/46.1.195
Roulin, A. & Wink, M. 2004. Predatorprey relation-
ships and the evolution of colour polymorphism: a
comparative analysis in diurnal raptors. – Biologi
cal Journal of the Linnean Society 81: 565–578.
DOI: 10.1111/j.10958312.2004.00308.x
Stamatakis, A. 2006. RAxMLVIHPC: maximum like-
lihoodbased phylogenetic analyses with thousands
of taxa and mixed models. – Bioinformatics 22:
2688–2690. DOI:10.1093/bioinformatics/btl446
Stiles, F. G. 1980. Evolutionary implications of habi-
tat relations between permanent and winter resident
landbirds in Costa Rica. pp. 421–435. – In: Keast,
A. & Morton, E. S. (eds.) Migrant Birds in the Neo-
tropics. – Smithsonian Institution Press, Washing-
ton D.C.
Tökölyi, J. & Barta, Z. 2011. Breeding phenology de-
termines evolutionary transitions in migratory be-
haviour in nches and allies. – Oikos 120: 184–
193. DOI:10.1111/j.16000706.2010.18592.x
Wink, M. & Sauer–Gürth, H. 2004. Phylogenetic re-
lationships in diurnal raptors based on nucleotide
sequences of mitochondrial and nuclear marker
genes. pp. 483–498. – In: MottaJunior, J. C., Bue-
no, A., Chancellor, R. & Meyburg, B. U. (eds.)
Raptors worldwide. WWGBP, Berlin
Zachos, J., Pagani, M., Sloan, L., Thomas, E. & Billups,
K. 2001. Trends, rhythms, and aberrations in global
climate 65 Ma to present. – Science 292: 686–693.
DOI:10.1126/science.1059412
Zink, R. M. 2011. The evolution of avian migration.
– Biological Journal of the Linnean Society 104:
237–250. DOI:10.1111/j.10958312.2011.01752.x