Identifying the Sister Species to the Rapid Capuchino Seedeater Radiation (Passeriformes: Sporophila)

Abstract

Within the Neotropical genus Sporophila, a group of eight species known colloquially as southern capuchinos shows remarkable phenotypic variation (mainly male coloration and song) despite lack of mitochondrial DNA monophyly and extremely low differentiation in other putatively neutral genetic markers. Previous studies have interpreted this to reflect recent common ancestry and perhaps ongoing hybridization and introgression. A recent taxonomic revision of the only polytypic southern capuchino species S. bouvreuil (with four previously recognized subspecies), prompted the designation of S. bouvreuil and S. pileata as two distinct species based on plumage color and geographic distribution. Here we use DNA sequence and microsatellite data to corroborate these new species designations and explore for the first time the relationship between these taxa and the remaining southern capuchinos. Phylogenetic and population genetic analyses show that S. bouvreuil and not S. minuta, as was previously thought, is the sister species to the core radiation of which S. pileata is part. Our data suggest that the ancestor of the southern capuchinos derives from northern South America and began to radiate during the lower to middle Pleistocene into at least eight species within the grasslands of northeastern Argentina, eastern Paraguay and southern Brazil. Consistent with earlier studies, we could not distinguish among southern capuchino species using neutral genetic markers, an expected signature of a rapid and recent radiation.

Full text

IDENTIFYING THE SISTER SPECIES TO THE RAPID CAPUCHINO SEEDEATER
RADIATION (PASSERIFORMES: SPOROPHILA)
Leonardo Campagna,
1,2,4
Luís Fábio siLveira,
3
pabLo L. Tubaro,
2
and sTephen C. Lougheed
1
1
Department of Biology, Queen’s University, Kingston, Ontario K7L 3N6, Canada;
2
División de Ornitología, Museo Argentino de Ciencias Naturales “Bernardino Rivadavia” (MACN), Avenida Ángel Gallardo 470,
Ciudad de Buenos Aires, Buenos Aires C1405DJR, Argentina; and
3
Seção de Aves, Museu de Zoologia, Universidade de São Paulo, Caixa Postal 42.494, CEP 04218-970, São Paulo, SP, Brazil
A.—Within the Neotropical genus Sporophila, a group of eight species known colloquially as “southern capuchinos”
shows remarkable phenotypic variation despite lack of (species level) mitochondrial DNA monophyly and extremely low differentiation
in other putatively neutral genetic markers. Previous studies have interpreted this as reflecting recent common ancestry and, perhaps,
ongoing hybridization and introgression. A recent taxonomic revision of the only polytypic southern capuchino species, Sporophila
bouvreuil (with four previously recognized subspecies), prompted the designation of S. bouvreuil and S. pileata as two distinct species
on the basis of plumage color and geographic distribution. We used DNA sequence and microsatellite data to corroborate these new
species designations and explored for the first time the relationship between these taxa and the remaining southern capuchinos.
Phylogenetic and population genetic analyses showed that S. bouvreuil and not S. minuta, as was previously thought, is the sister
species to the core radiation of which S. pileata is part. Our data suggest that the ancestor of the southern capuchinos is derived from
northern South America and began to radiate during the lower to middle Pleistocene into at least eight species within the grasslands
of northeastern Argentina, eastern Paraguay, and southern Brazil. Consistent with earlier studies, we could not distinguish among
southern capuchino species using neutral genetic markers, an expected signature of a rapid and recent radiation. Received  April ,
accepted  July .
Key words: grassland birds, Neotropics, recent radiation, Sporophila bouvreuil, S. pileata.
Identificación de la Especie Hermana de la Radiación Rápida de los Capuchinos del Sur (Sporophila, Passeriformes)
R.—El género Neotropical Sporophila contiene un grupo de ocho especies conocido coloquialmente como los capuchinos del
sur. Dicho grupo muestra una marcada variación fenotípica (principalmente en la coloración y el canto de los machos) que contrasta con
la falta de monofilia a nivel de especie y bajos niveles de diferenciación en marcadores genéticos neutros. Este patrón ha sido atribuido al
origen reciente del grupo y a la hibridación e introgresión entre especies. Una revisión taxonómica reciente de la única especie politípica,
S. bouvreuil (la cuál contenía cuatro subespecies), designó a S. bouvreuil y S. pileata como nuevas especies en base a patrones de coloración y
distribución geográfica. En el presente estudio utilizamos secuencias de ADN y frecuencias de alelos de ADN microsatélite para corroborar
la designación de dichas especies y estudiar por primera vez la relación entre ellas y con respecto a los demás capuchinos del sur. Utilizando
análisis filogenéticos y herramientas de genética de poblaciones mostramos que S. bouvreuil y no S. minuta, como se creía anteriormente, es
la especie hermana del grupo, al cual pertenece S. pileata. Posiblemente el ancestro de los capuchinos del sur provino del norte de América
del Sur y comenzó a radiar durante el Pleistoceno inferior o medio en al menos ocho especies en los pastizales del noreste Argentino, este
de Paraguay, y sur de Brasil. Como en estudios previos, no fue posible distinguir entre capuchinos del sur utilizando marcadores genéticos
neutros, un resultado esperable en el contexto de una radiación rápida y reciente.
—
645
—
e Auk 130(4):645−655, 2013
 e American Ornithologists’ Union, 2013.
Printed in USA.
e Auk, Vol. 
, Number , pages −. ISSN -, electronic ISSN -.   by e American Ornithologists’ Union. All rights reserved. Please direct all
requests for permission to photocopy or reproduce article content through the University of California Press’s Rights and Permissions website, http://www.ucpressjournals.
com/reprintInfo.asp. DOI: ./auk..
4
Present address: Fuller Evolutionary Biology Program, Cornell Laboratory of Ornithology, 159 Sapsucker Woods Road, Ithaca, New York 14850,
USA. E-mail: leocampagna@gmail.com.
T  N genus Sporophila harbors within it a
remarkable radiation of  granivorous species, colloquially known
as capuchinos or caboclinhos in Spanish and Portuguese, respec-
tively, but lacking a common name in English (Ridgely and Tudor
, Rising et al. , Remsen et al. ). Eight of the capuchino
species are endemic to central and southern South America,
where they are predominantly sympatric and often syntopic, and
appear to have radiated rapidly during the Pleistocene (Ridgely
and Tudor ; Lijtmaer et al. ; Campagna et al. , ).
ese eight species (S. cinnamomea, S. hypochroma, S. hypoxantha,

646 — Campagna eT aL. — auk, voL. 130
phylogenetic relationships between S. bouvreuil and the species
that comprise the southern capuchino radiation? Is S. bouvreuil
part of the radiation, or is S. bouvreuil and not S. minuta the sister
species to the southern capuchinos? If S. bouvreuil is the sister
lineage, we can then narrow down the geographic and temporal
scenarios for the southern capuchino radiation.
Methods
Sampling and data set.—We augmented the data set of Campagna
et al. () with genetic data from new individuals of S. bouvreuil
(n = ) and S. pileata (n = ). In total, the augmented data set
includes the eight southern capuchino species (n =  of which
 belong to S. pileata), S. bouvreuil (n = ), S. minuta (n = ),
S. castaneiventris (n = ), and  individuals from four closely
related outgroup species (numbers of samples per southern
capuchino species are summarized for each marker in Table ; for
other details, see Table S in the online supplementary material).
Most tissue samples were from vouchered males in adult plumage
(with study skin, skeleton, or specimen in ethanol deposited at the
Museo Argentino de Ciencias Naturales “Bernardino Rivadavia,”
the Museu de Zoologia da Universidade de São Paulo, or another
institution; for details, see Table S). For individuals for which
we have blood samples, a picture was taken to serve as a digital
voucher before the bird was released and is available upon request.
Genetic markers.—Samples in Campagna et al.’s () data
set were genotyped for six previously published microsatellite loci
(Esc, Hanotte et al. ; Mcy, Double et al. ; Pdo,
Neumann and Wetton ; Gf, Gf, and Gf, Petren ).
is data set also includes DNA sequences from eight markers for
most individuals from which fresh tissue samples were available
(i.e., not taken from museum study-skin toe pads). ese mark-
ers included three mitochondrial regions (cytochrome b [cyt b],
 base pairs [bp]; cytochrome c oxidase I [COI],  bp; mito-
chondrial control region [CR], ~, bp), one autosomal intron
(intron  of the β-fibrinogen gene: Fib, ~ bp), two previously
described nuclear sequences of mitochondrial origin (Numt:
~ bp; Numt: ~ bp; see Sato et al. ), and two Z-linked
markers (chromodomain-helicase-DNA binding protein: CHDZ:
~ bp; intron  of the muscle skeletal receptor tyrosine kinase
gene: MUSK, ~ bp) (for details, see Tables  and S).
For the present study, we genotyped the new S. bouvreuil and
S. pileata samples for the six aforementioned DNA microsatellite
loci, and amplified and sequenced COI, cyt b, CR, and CHDZ. DNA
extraction, amplification, sequencing, and genotyping followed pro-
cedures described by Campagna et al. (). We generated additional
sequences ( COI,  cyt b, and two CR) for some southern capu-
chinos to assemble a data set of  individuals for which most had
been sequenced for all three mitochondrial regions. Table S provides
details and GenBank accession numbers. All sequences were aligned
using BIOEDIT, version .. (Hall ).
Phylogenetic analyses.—e model of nucleotide evolu-
tion was selected for each marker using JMODELTEST, version
.. (Darriba et al. ). We used three strategies to explore
the phylogenetic affinities among Sporophila species. First
we obtained a Bayesian gene tree using MRBAYES, version
. (Ronquist etal. ), and data from the three mitochon-
drial markers (COI, cyt b, and CR). We placed each marker in
a separate unlinked partition, and for each we used the model
S. melanogaster, S. nigrorufa, S. pileata, S. palustris, and S. ruficollis)
can show eclipse plumage and gather in mixed flocks when not
breeding (Ridgely and Tudor ). Evidence from molecular stud-
ies shows extremely low neutral genetic differentiation among
these taxa, interpreted to be the consequence of recent common
ancestry and, perhaps, ongoing hybridization and introgression
(Lijtmaer etal. ; Campagna et al. , ). Phylogenetic af-
finities among these eight species (hereafter “southern capuchinos”)
remain unresolved, despite the marked phenotypic differences
that exist mainly in male reproductive plumage and vocalizations
(Campagna et al. ). Southern capuchinos are strikingly sexu-
ally dimorphic; females are brown and olive and hard to distin-
guish among species, whereas males show distinct reproductive
plumage patterns that are generally based on the color cinnamon
(Ridgely and Tudor ). Plumage differences in the UV-portion
of the spectrum are found between some female southern capuchi-
nos (Benites et al. ). Aside from these phenotypic differences,
southern capuchinos are extremely similar in size and shape (Meyer
de Schauensee , Ridgely and Tudor , Ouellet ).
Recent studies of southern capuchino species have led to
the discovery of various alternative color morphs of already
recognized species and, in some cases, have prompted taxonomic
changes. Traditionally, S. zelichi was considered a distinct species,
but, primarily on the basis of overall similarity in song and habi-
tat use, it is now regarded as a color morph of S. palustris and has
been subsumed within the latter taxon (Areta , Remsen et al.
). Similarly, a series of alternative adult male color morphs
that share song types with different southern capuchinos have
been described (for S. melanogaster, see Repenning etal. ; for
S. hypoxantha, see Areta and Repenning ; for S. ruficollis, see
Areta et al. ). Finally, Machado and Silveira () conducted
a detailed taxonomic revision of the four subspecies (S. bouvreuil
bouvreuil, S. b. pileata, S. b. saturata, and S. b. crypta) from the only
polytypic species of the group. ey diagnosed only two species
based on plumage patterns: S. bouvreuil (which now includes S. b.
bouvreuil, S. b. saturata, and S. b. crypta) and S. pileata (includ-
ing the former S. b. pileata) (Machado and Silveira , Remsen
et al. ). ese newly diagnosed taxa are predominantly allo-
patric, with a small area of range overlap in southeastern Brazil
(Machado and Silveira ) where they appear to maintain their
phenotypic integrity.
Sporophila bouvreuil is found in open areas from northern
South America to central and southern Brazil, whereas S. pileata
occurs in central and southeastern Brazil, Paraguay, and north-
eastern Argentina (Machado and Silveira ; Fig. ). us, the
distribution of S. bouvreuil is geographically intermediate to that
of other southern capuchino species (including S. pileata) and
S. minuta, the putative sister species of the southern capuchino
radiation (Campagna et al. , ; Fig. ). Previous studies also
found the former S. b. bouvreuil (now included in S. bouvreuil) to
be more closely allied to the southern capuchino radiation than
is S. minuta (Campagna et al. , ). However, these studies
included mostly S. b. pileata (now S. pileata) and only one individ-
ual S. b. bouvreuil (now included in S. bouvreuil); thus, tests of the
monophyly of the latter taxon could not be performed.
Here, we use putatively neutral mitochondrial and nuclear
sequences as well as DNA microsatellites to address the following
questions. () Is there genetic evidence supporting the designation
of S. bouvreuil and S. pileata as separate species? () What are the

oCTober 2013 — CapuChino seedeaTer radiaTion — 647
of nucleotide evolution chosen with JMODELTEST (COI and
cyt b: HKY+I+G, Hasegawa et al. ; CR: GTR+I+G, Tavaré
). e Bayesian analysis consisted of two independent,
simultaneous runs with four chains each (under default pri-
ors for all parameters) for  × 

generations, sampling trees
every  generations. At this point, both runs had reached a
stationary state and converged, which was confirmed using the
“cumulative” and “compare” functions in the software AWTY
(Wilgenbusch et al. ). A % majority rule consensus was
obtained from the combined posterior tree distribution after
discarding the initial % of trees as burn-in.
e second strategy combined mitochondrial and nu-
clear data to estimate a species tree using the Bayesian coales-
cent approach implemented in *BEAST (Heled and Drummond
), included in the BEAUTI/BEAST, version .., package
(Drummond et al. ). e concatenated mitochondrial genes
(COI, cyt b, and CR) and each of the five nuclear markers were
placed in different partitions under the model of nucleotide evo-
lution chosen using JMODELTEST (CHDZ and Fib: HKY;
MUSK: HKY+I; Numt: HKY+G; Numt: GTR+G; and mito-
chondrial: GTR+I+G). We used a Yule speciation model and a
piecewise linear and constant-root population size model. We im-
plemented a relaxed uncorrelated log normal clock and ran the
analysis for  × 

generations (sampling every ,). We used
TRACER, version . (Rambaut and Drummond ), to confirm
convergence in parameter estimates and that effective sample
sizes exceeded . Finally, we used TREE ANNOTATOR, ver-
sion .., from the BEAUTI/BEAST package to summarize the
. × 

(%) post-burn-in sampled trees. We obtained a max-
imum-clade-credibility tree using mean node heights; posterior
Fig. 1. Approximate range map for Sporophila minuta (the distribution of this species outside of South America is not represented here), S. bouvreuil,
and the eight species that comprise the southern capuchino radiation (the distribution of S. pileata is shown separately) following Ridgely and Tudor
(1989), Machado and Silveira (2010), and Somenzari et al. (2011). Circles represent sampling locations. Examples of adult male S. pileata and S. bou-
vreuil are shown in the right panels, together with a schematic representation of the phylogenetic affinities among capuchinos.

648 — Campagna eT aL. — auk, voL. 130
probability indicated node support. Finally we used DENSITREE,
version .. (Bouckaert ), to overlay a subsample ( × 

)
of the post-burn-in trees. We also estimated species trees using
individual markers to assess the relative contribution of each to
the multigene topology.
e third strategy also combined mitochondrial (concate-
nated COI, cyt b, and CR sequences) and nuclear data (CHDZ,
Fib, MUSK, Numt, and Numt sequences placed in separate
partitions) to perform species delimitation using the Bayesian
framework implemented in BP&P, version .b (Rannala and
Yang , Yang and Rannala ). BP&P generates a posterior
distribution of species assignments while accommodating the
effects of incomplete lineage sorting on gene trees. We used the
topology generated by *BEAST as a guide tree, from which BP&P
collapses each internal node while calculating the probabilities of
models containing different numbers of species. We used a gamma
prior for each population size parameter (θ) and the age of the root
in the species tree (τ). For the gamma distributions, we speci-
fied a shape parameter (α = ) and a scale parameter (β = ,)
in both cases. Other divergence time parameters were assigned
default Dirichlet priors. We conducted multiple analyses imple-
menting reversible-jump Markov chain Monte Carlo (rjMCMC)
algorithms  (using ε = , , , or ) or  (trying all combinations
of α = , ., or  and m = ., , or ) (Yang and Rannala ). e
analyses were run for , generations, sampling every  and
discarding the first , as burn-in.
F-statistics.—We explored the degree of genetic divergence
among Sporophila species by calculating all pairwise F
ST
or Φ
ST
values with ARLEQUIN, version . (Excoffier and Lischer );
significance was tested using , random permutations with
sequential Bonferroni corrections (Rice ). We subsequently
displayed these matrices graphically by building neighbor-joining
trees with the program NEIGHBOR provided in PHYLIP, ver-
sion .. (Felsenstein ). Φ
ST
and F
ST
neighbor-joining trees
were constructed for both the mitochondrial and the microsatellite
data. For these trees, we included only species for which at least five
individuals had been sequenced or genotyped. We also calculated
F
ST
based on a single nucleotide polymorphism in CHDZ, between
S. bouvreuil and all the southern capuchinos pooled. Finally, analyses
of molecular variance (AMOVA) were performed in ARLEQUIN,
grouping samples by either species or sampling locality.
Genetic distance and ordination analyses.—We calculated
inter-individual genetic distances using either sequence data or
the six DNA microsatellite loci. For the former, we estimated
uncorrected p-distances in MEGA, version  (Tamura et al. ),
whereas for the latter we computed D
SW
genetic distances (Shriver
et al. ) in POPULATIONS, version .. (Langella ).
Pairwise matrices of inter-individual genetic distances were dis-
played using principal coordinate analyses (PCoA) computed in
GENALEX, version . (Peakall and Smouse ).
Bayesian clustering analysis.—We used the Bayesian approach
implemented in STRUCTURE, version .. (Pritchard et al.
TabLe 1. Sample sizes for each capuchino species and molecular marker (see text) used in this study. Genetic divergence was estimated independently
for each marker; the average divergence between the southern capuchinos and Sporophila bouvreuil and the highest divergence between two south-
ern capuchino species is shown. For sequence data (mitochondrial and nuclear), we calculated average percent p-distances and standard deviations,
and for the DNA microsatellite data we computed pairwise F
ST
values between species.
Species COI Cyt b CR Numt2 Numt3 Fib5 CHD1Z MUSK
DNA mic-
rosatellites
Sporophila
castaneiventris
3 3 3 1 3 – – – 5
S. minuta
7 7 7 6 7 – – – 7
S. bouvreuil
19 17 19 – 1 – 17 – 20
S. cinnamomea
3 3 3 2 2 3 3 3 14
S. hypochroma
2 2 2 1 1 2 2 1 26
S. hypoxantha
30 30 42 8 33 21 4 5 62
S. melanogaster
7 7 7 4 7 7 7 1 7
S. nigrorufa
1 – – – – – – – 2
S. pileata
10 10 10 6 6 6 8 4 37
S. palustris
10 10 10 4 11 11 5 3 15
S. ruficollis
7 7 7 7 7 7 4 4 44
Southern
capuchinos vs.
S. bouvreuil
1.63 ± 0.51 1.09 ± 0.23 0.89 ± 0.33 – 0.82 ± 0.31 – 0.18 ± 0.17 – 0.041
Maximum
among south-
ern capuchino
species
0.97 ± 0.44
a
0.61 ± 0.24
b
0.60 ± 0.31
c
0.17 ± 0.18
d
1.98 ± 0.09
e
0.11 ± 0.23
f
0.062 ± 0.12
g
0.54 ± 0.16
e
0.013
h
a
S. melanogaster vs. S. pileata.
b
S. melanogaster vs. S. palustris.
c
S. melanogaster vs. S. hypoxantha.
d
S. ruficollis vs. S. hypoxantha.
e
S. cinnamomea vs. S. hypochroma.
f
S. cinnamomea vs. S. ruficollis.
g
S. pileata vs. S. hypoxantha.
h
S. melanogaster vs. S. cinnamomea.

oCTober 2013 — CapuChino seedeaTer radiaTion — 649
), and our DNA microsatellite data to assign individuals to
different genetic populations (K). Before conducting this analysis,
for each species we assessed Hardy-Weinberg equilibrium (HWE)
and linkage disequilibrium (LD) using ARLEQUIN and sequential
Bonferroni corrections. e following locus–species combinations
were not in HWE: Pdo in S. palustris and S. pileata, Mcy in
S. pileata and S. bouvreuil, and Gf in S. bouvreuil. Mcy and
Gf were in LD in S. cinnamomea, S. hypochroma, S. hypoxantha,
and S. ruficollis. Deviations from HWE and LD had been observed
before for some of these loci (Campagna et al. ) and could be
the product of population-level genetic structure (Wahlund ),
ongoing hybridization and introgression, or technical difficulties
such as undetected allele dropout or null alleles. us, our mic-
rosatellite data may not fit the model assumed by STRUCTURE,
potentially leading to the overestimation of the number of genetic
populations. However, the STRUCTURE analyses found very lit-
tle species-level structure in our data, confirming results from
F-statistics (see below), which suggests that our results were not
spurious.
e STRUCTURE analysis was conducted using the admix-
ture ancestry model, correlated allele frequencies, and, in separate
analyses, both with and without locprior (i.e., a prior indicat-
ing species identity). We included the eight southern capuchino
species and S. bouvreuil, exploring values of K between  and 
(based on previous results that showed lack of differentiation in
these markers among southern capuchino species; Campagna
etal. ). We performed  iterations per value of K, each with
. × 

generations, discarding the initial % as burn-in. e
most likely value of K was determined using Evanno et al.’s ()
method implemented in STRUCTURE HARVESTER, version
.. (Earl and vonHoldt ).
Divergence time estimations.—We calculated the age of the
mitochondrial ancestor (time to most recent common ancestor,
TMRCA) between the southern capuchinos and S. bouvreuil using
BEAUti/BEAST. An estimation of absolute time was reached us-
ing cyt b data and a calibration of .% divergence per million
years (Weir and Schluter ). e BEAST analysis was run with
a random starting tree for  × 

generations, assumed constant
population sizes and a relaxed uncorrelated lognormal clock,
and implemented the HKY+I+G model of nucleotide evolution.
Results were inspected for convergence and adequate effective
samples sizes in TRACER.
Results
e mitochondrial Bayesian tree (based on COI, cyt b, and CR
sequences; Fig. A) suggests that S. bouvreuil is phylogenetically
distinguishable from S. pileata and is indeed the sister species
to the southern capuchino radiation. Most S. bouvreuil form a
highly supported clade that is sister to all individuals involved in
the southern capuchino radiation. Using cyt b data and a clock
calibration of .% divergence Ma
–
, we estimated the age of the
common ancestor between S. bouvreuil and the southern capu-
chinos to be . Ma before present (% high posterior density
interval: .–.). e clade composed of the southern capu-
chinos and S. bouvreuil is, in turn, sister to S. minuta (Fig. A).
Two of  individual S. bouvreuil (marked with arrows in Fig. A)
carry “southern capuchino” mitochondrial haplotypes. Both
individuals are male S. bouvreuil with adult reproductive
plumage (see Fig. S in the online supplementary material).
Southern capuchino species for which more than one individ-
ual was sampled (S. nigrorufa being the exception) show lack of
species-level monophyly at these loci. However, some southern
capuchino species have significant differences in mitochondrial
haplotype frequency (measured using Φ
ST
calculations per-
formed on concatenated COI, cyt b, and CR sequences; Fig. B;
for Φ
ST
values obtained from individual mitochondrial markers,
see Fig. S in the online supplementary material). Although Φ
ST
values between southern capuchino species were generally <.,
comparisons with S. bouvreuil were larger and were in all cases
statistically significant (Fig. B; for comparisons using genetic
distances, see Table ). e PCoA derived from inter-individual
P-distances (Fig. C) also illustrates the relationship among
individual S. minuta, S. bouvreuil, and southern capuchinos.
Southern capuchinos cluster into two groups: individuals that
belong to the main clade found within the radiation (Fig. A) and
a separate cluster comprised of all those remaining. As shown in
Figure A, one S. bouvreuil is found in each southern capuchino
cluster. A small percentage of the variation among southern
capuchinos is attributable to differentiation among either spe-
cies or sampling locality (AMOVA: .%, P = . for species;
.%, P = . for sampling locality).
Similar results were obtained from the multilocus species-tree
analysis (Fig. ) that included  individuals from nine species and
data from six unlinked genetic markers. e seven southern capu-
chino species included in this analysis comprised a highly supported
clade that was sister to S. bouvreuil. Phylogenetic affinities among
southern capuchino species were again uncertain, with low posterior
probabilities for all clades within this group (ranging from . in the
clade including S. hypochroma, S. melanogaster, and S. cinnamomea
to . between S. hypoxantha and S. ruficollis). Alternative topolo-
gies to that of the consensus tree (in white) are visible in gray in
the cloudogram generated from a subsample of the posterior tree
distribution (Fig. ). Darker shades of gray imply larger numbers of
trees with that topology, thus resembling the consensus tree. Trees
derived from nuclear data alone did not resolve differences among
the southern capuchinos, and Figure  is similar to the species tree
obtained from mitochondrial DNA alone (for species trees derived
from individual markers, see Fig. S in the online supplementary
material). However, there was a single nucleotide polymorphism in
CHDZ that was close to fixation between individual S. bouvreuil and
southern capuchinos (F
ST
= .), and the DNA microsatellite data (see
below) also support our conclusion that the differentiation between
southern capuchinos and S. bouvreuil is not solely attributable to
mitochondrial markers. Results from the BP&P analysis varied across
runs that used different algorithms and combinations of priors but
were generally consistent with the results obtained using *BEAST.
e data allowed us to distinguish S. minuta from the remaining
species regardless of the conditions under which the program was
run, and in some cases to distinguish between S. bouvreuil and the
southern capuchinos. However, differences among southern capu-
chino species were generally not supported.
The F
ST
neighbor-joining tree based on DNA microsatellite
data also shows higher values between southern capuchinos
and S. bouvreuil than among the former species (the largest
inter-capuchino F
ST
value was . between S. cinnamomea

650 — Campagna eT aL. — auk, voL. 130
Fig. 2. (A) Bayesian tree for Sporophila derived from mitochondrial DNA sequence data (112 individuals; ≈2,650 base pairs from cytochrome c oxidase
I [COI], cytochrome b [cyt b], and mitochondrial control region [CR]) with posterior probabilities indicating node support. Arrows indicate two male
S. bouvreuil that fall within the southern capuchino clade. Posterior probabilities of nodes with low support were omitted for clarity. When individuals
belonging to the same species form a clade, species name is mentioned only once. (B) Neighbor-joining trees built using pairwise Φ
ST
calculations
derived from concatenated COI, cyt b, and CR data. Comparisons between southern capuchino species that were statistically significant after sequen-
tial Bonferroni correction are indicated by asterisks (α = 0.05). All comparisons between southern capuchinos and S. bouvreuil or S. minuta were
statistically significant (as was the comparison between S. bouvreuil and S. minuta). (C) PCoA derived from inter-individual p-distances calculated using
concatenated COI, cyt b, and CR data. Parentheses indicate the percentage of variation explained by each axis.

oCTober 2013 — CapuChino seedeaTer radiaTion — 651
and S. melanogaster; Fig. A and Table ). Only comparisons
between southern capuchinos and S. bouvreuil were statisti-
cally significant. Similar results can be observed in the biplot
from a PCoA of inter-individual D
SW
distances (Fig. B). We
found no significant differentiation among southern capuchino
species (AMOVA: –.%, P = .), and only a small fraction
of the total variation was attributable to sampling locality
(AMOVA: .%, P = .). For the STRUCTURE analysis
using microsatellites, the most likely scenario was K =  with
S. bouvreuil assigned to a different cluster than all southern
capuchinos (online supplementary material Fig. S). However,
this result was obtained only when species information was
used as prior, which suggests that the overall signal in these
data is weak (F
ST
= . between S. bouvreuil and all southern
capuchinos pooled; Table ).
discussion
Our analyses of mitochondrial and nuclear DNA sequence data
together with evidence derived from DNA microsatellites sup-
port the recent taxonomic change that elevated S. bouvreuil and
S. pileata to species status (Remsen et al. ) based on evidence
from adult male plumage and geographic distribution of these
taxa (Machado and Silveira , ). ese species can be dis-
tinguished using neutral markers and there are, in fact, other
southern capuchino species that are closer phylogenetically to
S. pileata (e.g., S. palustris) than is S. bouvreuil, despite the similarity
in plumage between S. palustris and S. bouvreuil (see Fig. ). Our
analyses also show that S. bouvreuil, and not S. minuta as was
previously thought (Campagna et al. , ), is the sister spe-
cies to the southern capuchino radiation. e aforementioned
studies included only one S. bouvreuil, precluding tests for species-
level monophyly. Two of the  S. bouvreuil in the present study
have mitochondrial COI, cyt b, and CR sequences that place them
within the southern capuchino clade and not with the remaining
S. bouvreuil. ese two individuals are adult males that have
typical S. bouvreuil reproductive plumage and were assigned to
the S. bouvreuil genetic cluster using DNA microsatellite data.
Nuclear DNA sequences alone did not provide sufficient resolution
to distinguish among southern capuchinos and S. bouvreuil, pos-
sibly as a consequence of lower substitution rates, larger effective
population size, and incomplete lineage sorting. However, when
data from the same individuals were incorporated into the species
tree estimation using *BEAST that explicitly models the effects of
incomplete lineage sorting and ancestral polymorphism (Heled
and Drummond ), both the southern capuchino clade and
the clade involving these species and S. bouvreuil received poste-
rior probability support of  (a similar result was obtained from
the BP&P analysis). is implies that incomplete lineage sorting
could be the reason that a small proportion of S. bouvreuil share
mitochondrial haplotypes with the southern capuchinos, a pattern
that could also have been generated through hybridization. e
methodology used to estimate the species tree assumes that admix-
ture (i.e., hybridization and introgression) does not occur among
individuals of the different species (Heled and Drummond ).
us, regardless of the results from the species tree analysis, we
cannot rule out the possibility of hybridization and introgression
between a female of a species belonging to the southern capuchino
clade and male S. bouvreuil, resulting in mitochondrial introgres-
sion from the former into the latter.
Identifying S. bouvreuil as the sister species to the southern
capuchino radiation allows us to reevaluate the timing and
geographic context of the radiation. Using data from cyt b and
S. minuta as the closest species to the southern capuchinos,
Campagna et al. () used TMRCA to estimate that the radia-
tion began in the Pleistocene. Here, with nearly double the number
of sequences and an improved phylogeny (with S. bouvreuil as the
sister species to the southern capuchinos), we obtained a similar
TMRCA estimate, perhaps as a consequence of the short internode
distance between S. minuta and S. bouvreuil. e % high poste-
rior density estimate of the age of the common ancestor between
S. bouvreuil and the southern capuchinos encompasses the lower
to middle Pleistocene. us, it is possible that Pleistocene climatic
changes and concomitant fluctuations in the distribution of rain-
forest over open areas (Clapperton , Servant et al. , Ledru
et al. ) contributed to isolating populations and promoting
the southern capuchino radiation. Having excluded S. bouvreuil
from the core radiation, we can delimit the northern boundar-
ies of the range within which the southern capuchino radiation
likely occurred (see Fig. ). Our data suggest that the ancestor of
the southern capuchinos came from northern South America and
radiated rapidly into at least eight species within the grasslands
of northeastern Argentina, eastern Paraguay, and southern Brazil
(the area that shows the highest species concentration; for indi-
vidual species maps, see Ridgely and Tudor ).
Fig. 3. Species tree for Sporophila inferred from mitochondrial (concate-
nated COI, cyt b, and CR) and nuclear (Numt2, Numt3, CHD1Z, MUSK,
and Fib5) sequence data from 95 individuals belonging to nine species.
The consensus tree (in white) was superimposed onto a cloudogram de-
rived from 20,000 post-burn-in trees. Each tree is represented in gray,
with darker shades implying greater degree of overlap. Node support
within the southern capuchino radiation was low (0.4–0.7) and, thus,
was omitted for simplicity (see Fig. S3 in the online supplementary mate-
rial for details).

652 — Campagna eT aL. — auk, voL. 130
Although the diversification of the southern capuchinos is
relatively recent compared with other avian radiations (e.g., for war-
blers in genus Dendroica, see Lovette and Bermingham ; or
for Hawaiian honeycreepers in tribe Drepanidini, see Lerner etal.
), the time elapsed since the lower to middle Pleistocene has
been sufficient to show striking phylogeographic structure in other
Neotropical avian taxa (e.g., for Zonotrichia capensis, see Lougheed
et al. ). In the present study, we did not find species-level mono-
phyly or clear affinities among the eight southern capuchino species,
consistent with results from previous studies (Lijtmaer et al. ;
Campagna et al. , ), even in those analyses that account
for the effect of ancestral polymorphism and incomplete lineage
sorting. is does not mean that southern capuchinos should not
be considered good biological species, and various studies have
identified differences in male and female plumage and song that are
maintained in sympatry (Benites etal. , Areta , Campagna
et al. ). We interpret this genetic pattern as the consequence
of a rapid and ongoing radiation. Using Bayesian simulations that
implement the isolation with migration model (IMa; Hey ),
Campagna et al. () obtained results consistent with gene flow
among southern capuchino species. us, ongoing hybridization
and introgression since these species split from a common ancestor
could preclude us from reconstructing their phylogenetic affini-
ties (both *BEAST and BP&P assume lack of admixture). It is also
possible that insufficient time has elapsed for stochastic sorting of
neutral genetic markers to have occurred (McKay and Zink );
incomplete lineage sorting would be accentuated if the speciation
events that led to the eight southern capuchino species occurred
rapidly over a short time span.
A similar challenge in reconstructing phylogenetic affini-
ties has been documented for Darwin’s ground finches (Freeland
and Boag , Sato et al. ). In contrast to Darwin’s finches
(Geospiza spp.), southern capuchino seedeaters exhibit similar
morphology, including bill dimensions, differing mainly in male
plumage and song (Campagna et al. ). ese differences in key
aspects of the avian mate-recognition system (Price ) suggest
Fig. 4. (A) Neighbor-joining tree for Sporophila derived from F
ST
calculations based on allele frequencies at six DNA microsatellite loci. (B) PCoA
displaying an inter-individual pairwise D
SW
distance matrix (percentage of variation explained by each axis in parenthesis).

oCTober 2013 — CapuChino seedeaTer radiaTion — 653
sexual selection as a driver of speciation in the group (Campagna
et al. ), and future work should focus on the possible role of
these phenotypic differences as mechanisms of reproductive
isolation. We expect that identification of genes that underpin
phenotypic differences among southern capuchinos will help us
understand the relationship among species and provide insight
into the mechanisms that promoted speciation in the group. It
is worth noting that the same patches and colors, combined into
different patterns, are implicated in male plumage differences
among southern capuchinos (for representative illustrations,
see Ridgely and Tudor ; for alternative color morphs within
some species, see Areta , Repenning et al. , Areta and
Repenning , Areta et al. ). Certainly, plumage attributes
seem to be evolutionary malleable in the group, possibly meaning
that a limited suite of genes and mutations underlie these differ-
ences in male coloration. Our future work will focus on employing
next-generation sequencing to recover large numbers of loci and
search for those loci that show patterns consistent with selection.
AcknowledgMents
Supplementary material is available with the online version
of this article at dx.doi.org/./auk... We are
indebted to É. Machado and to the staff and students from
Museu de Zoologia, Universidade de São Paulo; to W. Lemos
Morais Neto and the staff from Fazenda Fartura; and to
B. Ehlers (UPS Brazil) for their help in collecting the specimens
used in this study. We thank the Instituto Chico Mendes de
Conservação da Biodiversidade (Brazil) for granting col-
lection permits and the Canadian Food Inspection Agency
for providing import permits. Photographs in Figure  were
kindly provided by E. Endrigo (S. bouvreuil) and R. Güller
(S. pileata). L.C. thanks the Senate Advisory Research Com-
mittee at Queen’s University for postdoctoral funding. This
study was funded by grants to L.F.S. from the Fundação de
Amparo à Pesquisa no Estado de São Paulo and Conselho
Nacional de Desenvolvimento Científico e Tecnológico (CNPq
/-: Evolução da Fauna de Vertebrados Terrestres
Brasileiros do Cretáceo ao Presente: Paleontologia e Filogenia).
The study was also funded by grants to P.L.T. from Agencia
Nacional de Promoción Científica y Tecnológica (PICT
-, Argentina), Consejo Nacional de Investigaciones
Científicas y Técnicas (PIP --, Argentina), Uni-
versidad de Buenos Aires (UBACyT -, Argentina),
International Development Research Centre (Canada), and the
Richard Lounsbery Foundation (USA), and by a Natural Sci-
ences and Engineering Research Council of Canada Discovery
Grant to S.C.L. We thank L. Joseph and two anonymous
reviewers for valuable feedback on the manuscript.
liteR AtuRe cited
A, J. I.
. e Entre Ríos Seedeater (Sporophila zelichi): A
species that never was. Journal of Field Ornithology :–.
A, J. I.
. Winter songs reveal geographic origin of three
migratory seedeaters (Sporophila spp.) in southern Neotropical
grasslands.
Wilson Journal of Ornithology
:–.
A, J. I., J. I. N, L. P,  I. R.
. Unrav-
elling the ecological radiation of the capuchinos: Systematics of
Dark-throated Seedeater Sporophila ruficollis and description of
a new dark-collared form. Bulletin of the British Ornithologists’
Club :–.
A, J. I.,  M. R.
.
Systematics of the Tawny-
bellied Seedeater (
Sporophila hypoxantha
). II. Taxonomy and
evolutionary implications of the existence of a new tawny morph.
Condor :–.
B, P., M. D. E, D. A. L, S. C. L,
 P. L. T.
. Analysis from avian visual perspec-
tive reveals plumage colour differences among females of
capuchino seedeaters (Sporophila). Journal of Avian Biology
:–.
B, R. R.
. DensiTree: Making sense of sets of phyloge-
netic trees. Bioinformatics :–.
C, L., P. B, S. C. L, D. A. L, A. S.
DG, M. D. E,  P. L. T.
. Rapid phe-
notypic evolution during incipient speciation in a continental avian
radiation. Proceedings of the Royal Society of London, Series B
:–.
C, L., D. A. L, K. C. K, A. S. B, P. D.
H, S. C. L,  P. L. T.
. DNA bar-
codes provide new evidence of a recent radiation in the genus
Sporophila (Aves: Passeriformes). Molecular Ecology Resources
:–.
C, C. M.
. Nature of environmental changes in South
America at the last glacial maximum. Palaeogeography, Palaeocli-
matology, Palaeoecology :–.
D, D., G. L. T, R. D,  D. P.
.
jModelTest : More models, new heuristics and parallel computing.
Nature Methods :.
D, M. C., D. D, T. B,  A. C.
.
Finding the fathers in the least faithful bird: A microsatellite-based
genotyping system for the Superb Fairy-Wren Malurus cyaneus.
Molecular Ecology :–.
D, A. J., M. A. S, D. X,  A. R.
.
Bayesian phylogenetics with BEAUti and the BEAST ..
Molecular
Biology and Evolution
:–.
E, D. A.,  B. M. H.
. STRUCTURE HAR-
VESTER: A website and program for visualizing STRUCTURE
output and implementing the Evanno method. Conservation
Genetics Resources :–.
E, G., S. R,  J. G.
. Detecting the
number of clusters of individuals using the software STRUC-
TURE: A simulation study. Molecular Ecology :–.
E, L.,  H. E. L. L.
. ARLEQUIN suite, version
.: A new series of programs to perform population genetics anal-
yses under Linux and Windows. Molecular Ecology Resources
:–.
F, J.
. PHYLIP (Phylogeny Inference Package),
version ... Distributed by the author, Department of Genome
Sciences, University of Washington, Seattle.
F, J. R.,  P. T. B.
. e mitochondrial and nuclear
genetic homogeneity of the phenotypically diverse Darwin’s ground
finches. Evolution :–.
H, T. A. . BioEdit: A user-friendly biological sequence align-
ment editor and analysis program for Windows //NT.
Nucleic Acids Symposium Series :–.
H, O., C. Z, A. P, C. G, A. D, 
T. B.
. Isolation and characterization of microsatellite

654 — Campagna eT aL. — auk, voL. 130
loci in a passerine bird: e Reed Bunting Emberiza schoeniclus.
Molecular Ecology :–.
H, M., H. K,  T. A. Y. . Dating of the
human–ape splitting by a molecular clock of mitochondrial
DNA. Journal of Molecular Evolution :–.
H, J.,  A. J. D.
. Bayesian inference of
species trees from multilocus data.
Molecular Biology and Evo-
lution
:–.
H, J.
. Isolation with migration models for more than two
populations. Molecular Biology and Evolution :–.
L, O.
. Populations ..: A population genetic
software. CNRS UPR. [Online.] Available at bioinformatics.
org/~tryphon/populations/.
L, M.-P., D.-D. R, F. W. C, J., C. R,
I. K,  L. M.
. Paleoclimate changes
during the last , yr from a record in the Brazilian Atlantic
rainforest region and interhemispheric comparison. Quater-
nary Research :–.
L, H. R. L., M. M, H. F. J, M. H, 
R. C. F.
. Multilocus resolution of phylogeny and
timescale in the extant adaptive radiation of Hawaiian honey-
creepers.
Current Biology
:–.
L, D. A., N. M. S, P. L. T,  S. C.
L.
. Molecular phylogenetics and diversification
of the genus Sporophila (Aves: Passeriformes). Molecular Phylo-
genetics and Evolution :–.
L, S. C., L. C, J. A. D, P. L. T, D. A.
L,  P. H.
. Continental phylogeogra-
phy of an ecologically and morphologically diverse Neotropical
songbird, Zonotrichia capensis.
BMC Evolutionary Biology
:.
L, I. J.,  E. B.
. Explosive speciation
in the New World Dendroica warblers.
Proceedings of the Royal
Society of London, Series B :–.
M, E.,  L. F. S.
. Geographical and sea-
sonal distributions of the seedeaters Sporophila bouvreuil and
Sporophila pileata (Aves: Emberizidae). Papeis Avulsos de
Zoología :–.
M, E.,  L. F. S.
. Plumage variability and
taxonomy of the Capped Seedeater Sporophila bouvreuil (Aves:
Passeriformes: Emberizidae). Zootaxa :–.
MK, B. D.,  R. M. Z.
. e causes of mitochondrial
DNA gene tree paraphyly in birds. Molecular Phylogenetics and
Evolution :–.
M  S, R. M.
. A review of the genus
Sporophila. Proceedings of the Academy of Natural Sciences of
Philadelphia :–.
N, K.,  J. H. W.
. Highly polymorphic
microsatellites in the House Sparrow Passer domesticus. Molec-
ular Ecology :–.
O, H.
. Speciation, zoogeography and taxonomic prob-
lems in the Neotropical genus Sporophila (Aves: Emberizinae).
Bulletin of the British Ornithologists’ Club :–.
P, R.,  P. E. S.
. GENALEX : Genetic
analysis in Excel. Population genetic software for teaching and
research. Molecular Ecology Notes :–.
P, K.
. Microsatellite primers from Geospiza fortis and
cross-species amplification in Darwin’s finches. Molecular Ecol-
ogy :–.
P, T.
. Speciation in Birds. Roberts, Englewood, Colorado.
P, J. K., M. S,  P. D.
. Infer-
ence of population structure using multilocus genotype data.
Genetics :–.
R, A.,  A. J. D.
. TRACER, version ..
[Online.] Available at beast.bio.ed.ac.uk/Tracer.
R, B.,  Z. Y.
. Bayes estimation of species
divergence times and ancestral population sizes using DNA
sequences from multiple loci. Genetics
:–.
R, J. V., J., C. D. C, A. J, M. N,
J. F. P, J. P-E, M. B. R, F. G. S,
D. F. S,  K. J. Z.
. A classification of the
bird species of South America. American Ornithologists’
Union. [Online.] Available at www.museum.lsu.edu/~Remsen/
SACCBaseline.html.
R, M., C. R,  C. S. F.
. Another
color morph of Sporophila Seedeater from capuchinos group
(Aves, Emberizidae). Iheringia, Série Zoologia :–.
R, W. R.
. Analyzing tables of statistical tests. Evolution
:–.
R, R. S.,  G. T.
. e Birds of South America.
University of Texas Press, Austin.
R, J. D., A. J, J. L. C, P. G. R,  S. C.
M.
. Family Emberizidae (Buntings and New World
Sparrows). Pages – in Handbook of the Birds of the
World, vol. : Tanagers to New World Blackbirds (J. del Hoyo,
A. Elliott, and D. A. Christie, Eds.). Lynx Edicions, Barcelona,
Spain.
R, F., M. T, P.   M, D. L. A,
A. D, S. H, B. L, L. L, M. A. S,
 J. P. H.
. MrBayes .: Efficient Bayesian
phylogenetic inference and model choice across a large model
space.
Systematic Biology
:–.
S, A., C. O’U, F. F, P. R. G, B. R. G,
H. T,  J. K.
. Phylogeny of Darwin’s finches
as revealed by mtDNA sequences. Proceedings of the National
Academy of Sciences USA :–.
S, A., H. T, C. O’U, P. R. G, B. R. G, 
J. K.
. On the origin of Darwin’s finches. Molecular
Biology and Evolution :–.
S, M., J. M, B. T, M.-L. A, P. B, M.
F,  M.-P. L.
. Tropical forest changes
during the Late Quaternary in African and South American
lowlands. Global and Planetary Change :–.
S, M. D., L. J, E. B, R. D, R. E. F,
 R. C.
. A novel measure of genetic dis-
tance for highly polymorphic tandem repeat loci. Molecular
Biology and Evolution :–.
S, M., L. F. S, V. Q. P, M. A. R,
F. S,  V. C. 
.
Birds of an Amazonia-
Cerrado ecotone in southern Pará, Brazil, and the efficiency of
associating multiple methods in avifaunal inventories
.
Revista
Brasileira de Ornitologia 
:

–

.
T, K., D. P, N. P, G. S, M. N,
 S. K.
. MEGA: Molecular evolutionary genetics
analysis using maximum likelihood, evolutionary distance, and
maximum parsimony methods.
Molecular Biology and Evolu-
tion
:–.

oCTober 2013 — CapuChino seedeaTer radiaTion — 655
T, S. . Some probabilistic and statistical problems in the
analysis of DNA sequences. Pages – in Some Mathemati-
cal Questions in Biology: DNA Sequence Analysis (R. M. Miura,
Ed.). American Mathematical Society, Providence, Rhode Island.
W, S.
. Zusammensetzung von Populationen und Kor-
relationserscheinungen vom Standpunkt der Vererbungslehre
aus betrachtet.
Hereditas
:–.
W, J. T.,  D. S.
. Calibrating the avian molec-
ular clock. Molecular Ecology :–.
W, J. C., D. L. W,  D. L. S. .
AWTY: A system for graphical exploration of MCMC conver-
gence in Bayesian phylogenetic inference.
[Online.] Available
at
ceb.csit.fsu.edu/awty
.
Y, Z.,  B. R.
. Bayesian species delimitation
using multilocus sequence data.
Proceedings of the National
Academy of Sciences USA :–.
Associate Editor: L. Joseph