842
q 2006 The Society for the Study of Evolution. All rights reserved.
Evolution, 60(4), 2006, pp. 842?855
DIVERGENT TIMING AND PATTERNS OF SPECIES ACCUMULATION IN LOWLAND
AND HIGHLAND NEOTROPICAL BIRDS
JASON T. WEIR
Department of Zoology, University of British Columbia, Vancouver, British Columbia V6T 1Z4, Canada
E-mail: weir@zoology.ubc.ca
Abstract. Late Pliocene and Pleistocene climatic instability has been invoked to explain the buildup of Neotropical
biodiversity, although other theories date Neotropical diversification to earlier periods. If these climatic fluctuations
drove Neotropical diversification, then a large proportion of species should date to this period and faunas should
exhibit accelerated rates of speciation. However, the unique role of recent climatic fluctuations in promoting diver-
sification could be rejected if late Pliocene and Pleistocene rates declined. To test these temporal predictions, dateable
molecular phylogenies for 27 avian taxa were used to contrast the timing and rates of diversification in lowland and
highland Neotropical faunas. Trends in diversification rates were analyzed in two ways. First, rates within taxa were
analyzed for increasing or decreasing speciation rates through time. There was a significant trend within lowland taxa
towards decreasing speciation rates, but no significant trend was observed within most highland taxa. Second, fauna
wide diversification rates through time were estimated during one-million-year intervals by combining rates across
taxa. In the lowlands, rates were highest during the late Miocene and then decreased towards the present. The decline
in rates observed both within taxa and for the fauna as a whole probably resulted from density dependent cladogenesis.
In the highlands, faunawide rates did not vary greatly before the Pleistocene but did increase significantly during the
last one million years of the Pleistocene following the onset of severe glacial cycles in the Andes. These contrasting
patterns of species accumulation suggest that lowland and highland regions were affected differently by recent climatic
fluctuations. Evidently, habitat alterations associated with global climate change were not enough to promote an
increase in the rate of diversification in lowland faunas. In contrast, direct fragmentation of habitats by glaciers and
severe altitudinal migration of montane vegetation zones during climatic cycles may have resulted in the late Pleistocene
increase in highland diversification rates. This increase resulted in a fauna with one third of its species dating to the
last one million years.
Key words. Climate change, diversification rates, Neotropics, Pleistocene speciation, refuge hypothesis.
Received May 16, 2005. Accepted January 20, 2006.
Understanding the historical processes driving the diver-
sification of contemporary faunas is a major aim of bioge-
ography, yet the timing and rate of diversification in some
of the most species rich faunas are poorly understood. Species
diversity is highest in the Neotropics (Rosenzweig 1995). For
example, approximately three thousand species of birds occur
there (Haffer 1990), more than all other tropical regions com-
bined. This pattern is repeated in many other groups. A num-
ber of theories have been proposed to explain the origin of
this diversity (see review in Haffer 1997). These theories
differ in their view of what processes promoted speciation
and of the age of species in Neotropical faunas, but no con-
sensus has been reached.
Originally, Neotropical forests and climates were believed
to have been stable through most of their Cenozoic history
(Richards 1952; Fisher 1960; Schwabe 1969). This stability
was thought to have promoted low extinction rates and al-
lowed for the gradual buildup of high species diversity (Dar-
lington 1957; Sanders 1969; Schwabe 1969). Under this view,
species in Neotropical faunas were thought to be relatively
old. This theory was challenged when it became apparent that
intense climatic fluctuations during the Northern Hemisphere
ice ages also affected climate in the Neotropics. The tem-
perate latitude model of glacial refugia (Rand 1948; Mengel
1964) was applied to explain Neotropical diversification. The
resulting refuge hypothesis and its variant forms (e.g., river
refuge hypothesis) predicted that the majority of current Neo-
tropical species diversified during recent episodes of climatic
fluctuation when Neotropical habitats were believed to have
been repeatedly fragmented (Haffer 1969, 1974, 1997; Van-
zolini and Williams 1970; Brown et al. 1974; Prance 1978;
Simpson and Haffer 1978; Cerqueira 1982; Whitmore and
Prance 1987; Capparella 1991; Ayres and Cluttonbrock 1992;
Haffer and Prance 2001). Although the refuge hypothesis has
been invoked most often to explain diversification in lowland
wet-forest habitats, climatic fluctuations may have also frag-
mented other Neotropical habitats in both lowland (Meave
et al. 1991; Meave and Kellman 1994) and highland faunas
(Steyermark and Dunsterville 1980).
Climatic fluctuations have occurred throughout the history
of the Neotropics and may have contributed to diversification
at any period (Haffer 1997; Haffer and Prance 2001). How-
ever, the glacial cycles of the late Pliocene and Pleistocene
produced the most intense fluctuations. Beginning about 2.5
million years ago (mya, Bloemendal and Demenocal 1989;
Hooghiemstra 1989; Andriessen et al. 1993; van der Hammen
and Hooghiemstra 1997; Ravelo et al. 2004; Liu and Herbert
2004), these fluctuations persisted through the late Pliocene
(2.5 to 2.0 mya) and early Pleistocene (2.0 to 1.0 mya) and
culminated in a series of severe glacial cycles during the late
Pleistocene (; 1.0 mya to recent; Bennett 1990; Hooghiem-
stra et al. 1993). The intensity of these late Pleistocene glacial
cycles led most proponents of the refuge hypothesis to predict
that the majority of Neotropical species dated to this time.
Recently some proponents have extended this prediction to
earlier time periods (Haffer 1997; Haffer and Prance 2001).
Nevertheless, if climate fluctuations drove Neotropical di-
versification, then we would expect the rate of speciation to
increase during time periods when the duration and intensity
of fluctuations were greatest. Speciation rates should increase
843SPECIES ACCUMULATION IN NEOTROPICAL BIRDS
at the onset of glacial cycles 2.5 mya and again at the onset
of the late Pleistocene 1.0 mya.
Several other theories have endeavored to link Neotropical
diversification to specific geological events that mostly pre-
date the climatic cycles of the late Pliocene and Pleistocene.
In the lowlands, events such as uplift of montane barriers in
northwestern South America (Sick 1967), formation of the
Amazon drainage system (Sick 1967; Capparela 1988; Hoorn
et al. 1995; Aleixo 2004; Rossetti et al. 2005), marine in-
cursions (Hoorn 1993, 1994; Irion et al. 1995; Webb 1995;
Rasanen et al. 1995; Nores 1999, 2004; Gregory-Wodzicki
2000), or freshwater lake barriers (Vonhof et al. 2003; Ros-
setti et al. 2005) occurred primarily during the late Miocene
(10 to 5 mya) and early Pliocene (5 to 2.5 mya) and are
thought to have promoted diversification. The effect of the
these geological events may have been temporary (e.g., ma-
rine incursions) resulting in a burst of diversification at the
time of the event or their effect may persist to the present
(e.g., mountain and river barriers; see Bates et al. 2004) re-
sulting in ongoing opportunities for speciation. Due to the
overlap in the predictions of the timing of diversification, it
is difficult to investigate the potential role played by any one
geological event.
In the highlands, rapid uplift of the Andes and other high-
land regions occurred during the last 10 million years
(Hooghiemstra and van der Hammen 1998). For instance, 60
to 80% of the current height of the central and northern Andes
resulted from uplift during this time period (Gregory-Wod-
zicki 2000) and the Talamanca highlands of Central America
formed within the last five million years (Grafe et al. 2002).
This dynamic history of uplift may have provided ongoing
opportunities for diversification of highland species to the
present. Fjeldsa and Lovett (Fjeldsa and Lovett 1997a,b) pro-
posed that highland regions were the main source of diver-
sification for the Neotropics and that highland species dis-
persed to lowland faunas where they were preserved from
extinction.
Only those hypothesis that stress climatic fluctuations as
driving diversification predict that Neotropical faunas should
be recently derived with an increase in diversification rates
near the recent. With the advent of molecular dating tech-
niques it is now possible to test these predictions. Several
molecular based reviews of Neotropical speciation in birds
are available but are incomplete and have not addressed pat-
terns in rates of diversification through time. The review by
Moritz et al. (2000) suggested that Pleistocene speciation was
rare in Neotropical birds and other vertebrates, with the ma-
jority of species dating to the Pliocene and Miocene. How-
ever, their conclusions were based on only a few genera, and
further sampling may find greater support for diversification
near the recent. In contrast, a review of speciation in Andean
birds suggested a protracted history of diversification from
the Miocene to the present with substantial numbers of spe-
cies dating to the late Pliocene and Pleistocene (Garcia-Mo-
reno and Fjeldsa 2000). The larger sampling design in the
Andean study suggests that further sampling of lowland avian
genera may provide a more complete picture of Neotropical
diversification.
I compared the timing and rate of diversification in lowland
and highland avian radiations of the Neotropics. To make
this comparison, patterns of species accumulation were an-
alyzed from mitochondrial DNA phylogenies for 16 lowland
and 11 highland radiations. Patterns in the rate of diversifi-
cation through time were used to determine peak periods of
diversification for the faunas in each region and to test for
increasing, decreasing, or constant diversification rates
through time. The separation of Neotropical taxa into lowland
and highland faunas is useful because both regions experi-
enced different geological histories. In addition, climatic fluc-
tuations were more intense in highland regions where exten-
sive glaciation directly fragmented high elevation habitats.
In contrast, lowland faunas did not experience direct frag-
mentation by glaciers but habitats may have been fragmented
due to fluctuations in temperature and rainfall that accom-
panied them (Hooghiemstra and van der Hammen 1998; Bush
and Silman 2004). If recent climatic fluctuations drove Neo-
tropical diversification, then the majority of species should
date to the late Pliocene and Pleistocene. Additionally, spe-
ciation rates should increase through this period and peak
during the last one million years when climatic fluctuations
were most intense. In contrast, if events that predate the cli-
matic fluctuations were instrumental in Neotropical diver-
sification then we would not expect an increase in diversi-
fication during recent periods of climatic instability and a
large proportion of species should date to the Miocene and
early Pliocene.
METHODS
Phylogenetic Analysis
The Neotropical zoogeographic region extends from cen-
tral Mexico to the southern tip of South America. In this
analysis I excluded the Caribbean and other Neotropical is-
lands because I was interested in analyzing rates of diver-
sification within continental faunas. I included all terrestrial
taxa possessing five or more species in highland or lowland
regions for which mitochondrial DNA sequences were avail-
able for at least 75% of species (Table 1). In some cases
recent molecular phylogenetic studies have demonstrated that
two or more genera together formed a monophyletic group
but individually were paraphyletic. These were analyzed as
a single taxon (Troglodytes and Thryorchilus; Crax and Noth-
ocrax; Psarocolius, Cacicus, and Ocyalus; Geositta and Geo-
bates). In addition, a monophyletic assemblage of South
American blackbirds (Macroagelaius, Gymnomystax, Hypo-
pyrrhus, Lampropsar, Gnorimopsar, Curaeus, Amblyramphus,
Chrysomus, Xanthopsar, Pseudoleistes, Oreopsar, and Age-
laioides) and Neotropical swallows (Progne, Phaeoprogne,
Notiochelidon, Atticora, Neochelidon, and Stelgidopteryx)
were also included and each was analyzed as a single taxon
because many of their respective genera were paraphyletic
or they did not have enough species to allow for separate
analysis.
Wide taxonomic and ecological coverage are included in
the sample of Neotropical taxa used in this analysis. In ad-
dition, taxon size ranged from taxa with only five species to
one of the largest Neotropical genera, Tangara, with 49 spe-
cies distributed in both highland and lowland regions. Nev-
ertheless, this sample is constrained to currently available
phylogenies that may not represent a completely random sam-
844 JASON T. WEIR
TA
B
L
E
1.
Lo
w
la
nd
a
n
d
hi
gh
la
nd
ta
xa
a
n
a
ly
ze
d.
R
es
ul
ts
o
f
th
e
g
st
at
is
tic
,t
he
e
x
pe
ct
ed
n
u
m
be
r
o
f
lin
ea
ge
s
fo
r
e
a
c
h
c
la
de
,a
n
d
ph
yl
og
en
et
ic
so
u
rc
e
s
fr
om
w
hi
ch
th
e
m
a
jor
ity
o
f
D
N
A
se
qu
en
ce
s
w
e
re
o
bt
ai
ne
d
fo
r
e
a
c
h
ph
yl
og
en
et
ic
a
n
a
ly
si
s.
Ta
x
o
n
N
um
be
r
o
f
sp
ec
ie
s
R
eg
io
n1
A
va
ila
bl
e2
Sy
m
pa
tri
c3
Y
o
u
n
ge
st
sp
ec
ie
s4
g
(P
-
v
a
lu
e)
b5
(sp
ec
ies
/li
ne
ag
e
pe
r
M
yr
)
d5
(sp
ec
ies
/li
ne
ag
e
pe
r
M
yr
)
Ex
pe
ct
ed
ph
yl
og
ro
up
s
pe
r
sp
ec
ie
s6
Ph
yl
og
en
et
ic
so
u
rc
e
Lo
w
la
nd
Am
az
on
a
16
14
6
1.
05
2
2.
12
(0
.0
33
)
0.
35
0.
00
11
.6
R
us
se
llo
a
n
d
A
m
at
o
20
04
B
la
ck
bi
rd
c
la
de
14
13
9
0.
27
2
3.
26
(0
.0
01
)
0.
25
0.
00
63
1.
0
La
ny
on
a
n
d
O
m
la
nd
19
99
C
ra
x
13
13
3
0.
73
2
1.
80
(0
.07
2)
0.
65
0.
00
0.
2
Pe
re
ira
a
n
d
B
ak
er
20
04
D
en
dr
oc
in
cl
a
7
7
3
1.
38
0.
07
(0
.94
4)
0.
51
0.
60
0.
7
th
is
st
ud
y
Ic
te
ru
s
16
16
6
0.
33
2
2.
06
(0
.0
39
)
0.
22
0.
00
2.
3
O
m
la
nd
e
t
a
l.
19
99
M
yi
ar
ch
us
11
10
5
0.
31
0.
27
(0
.84
2)
0.
40
0.
43
2
0.
5
Jo
se
ph
e
t
a
l.
20
03
N
yc
tib
iu
s
6
5
5
10
.1
0
2
0.
34
(0
.73
0)
*
0.
11
0.
00
6.
3
M
ar
ia
ux
a
n
d
B
ra
un
19
96
Pi
on
op
si
tta
9
8
3
1.
38
1.
15
(0
.24
8)
0.
32
0.
42
2
0.
1
Eb
er
ha
rd
a
n
d
B
er
m
in
gh
am
20
05
Ps
ar
oc
ol
iu
s
19
15
11
1.
10
2
1.
79
(0
.06
8)
0.
26
0.
00
53
.7
Pr
ic
e
a
n
d
La
ny
on
20
02
,2
00
4
Pt
er
og
lo
ss
us
12
12
4
0.
91
2
0.
11
(0
.91
2)
0.
49
0.
24
0.
3
Eb
er
ha
rd
a
n
d
B
er
m
in
gh
am
20
05
Ra
m
ph
oc
el
us
7
6
2
1.
95
0.
27
(0
.78
5)
0.
80
1.
08
0.
4
H
ac
ke
tt
19
96
Sw
al
lo
w
c
la
de
12
12
8
0.
40
2
2.
02
(0
.0
43
)
0.
12
0.
00
1.
8
Sh
el
do
n
e
t
a
l.
20
05
Ta
ch
yc
in
et
a
5
5
2
2.
34
0.
56
(0
.57
6)
1.
37
2.
19
0.
7
W
hi
tti
ng
ha
m
e
t
a
l.
20
02
Ta
ng
ar
a
22
19
8
1.
87
2
2.
11
(0
.0
32
)
0.
30
0.
00
78
1.
2
B
ur
ns
a
n
d
N
ao
ki
20
04
Tr
og
on
14
11
5
3.
69
2
1.
18
(0
.22
9)
0.
13
0.
00
0.
8
Es
pi
no
sa
de
lo
s
19
98
,t
hi
s
st
ud
y
Ve
ni
lio
rn
is
10
8
3
3.
30
2
0.
37
(0
.70
7)
0.
64
0.
39
2.
6
M
oo
re
e
t
a
l.
20
05
Xi
ph
or
hy
nc
hu
s
13
13
6
1.
63
2
1.
26
(0
.20
8)
0.
25
0.
00
7.
6
A
le
ix
o
20
02
Z-
te
st
2
3.
92
(,
0.
00
01
)
A
ve
ra
ge
12
11
5
1.
93
0.
42
0.
32
88
.3
H
ig
hl
an
d
An
ai
re
te
s
7
6
5
1.
21
2
1.
07
(0
.28
2)
0.
30
0.
00
0.
4
R
oy
e
t
a
l.
19
99
C
ar
du
el
is
10
8
5
0.
59
1.
04
(0
.29
0)
4.
28
5.
59
0.
0
v
a
n
de
n
El
ze
n
e
t
a
l.
20
01
C
in
cl
od
es
11
11
3
0.
42
2
0.
62
(0
.53
5)
0.
49
0.
08
0.
0
C
he
ss
er
20
04
C
ra
ni
ol
eu
ca
11
10
3
0.
34
2
1.
15
(0
.25
1)
0.
83
0.
00
0.
8
G
ar
ci
a-
M
or
en
o
e
t
a
l.
19
99
a
G
eo
si
tta
9
9
4
4.
14
0.
21
(0
.83
4)
0.
33
0.
31
0.
5
C
he
vi
ro
n
e
t
a
l.
20
05
a
H
em
is
pi
ng
us
14
12
7
0.
57
2
1.
72
(0
.08
1)
0.
12
0.
00
1.
2
G
ar
ci
a-
M
or
en
o
e
t
a
l.
20
01
M
et
al
lu
ra
10
9
2
0.
34
2
0.
85
(0
.39
4)
0.
24
0.
00
0.
3
G
ar
ci
a-
M
or
en
o
e
t
a
l.
19
99
b
M
us
ci
sa
xi
co
la
12
12
6
0.
33
4.
07
(0
.0
00
)
5.
31
6.
64
2
0.
6
C
he
ss
er
20
00
O
ch
th
oe
ca
11
11
7
2.
19
2
0.
13
(0
.89
7)
0.
32
0.
17
0.
7
G
ar
ci
a-
M
or
en
o
e
t
a
l.
19
98
Ta
ng
ar
a
26
23
13
0.
29
2
3.
17
(0
.0
32
)
0.
20
0.
00
0.
7
B
ur
ns
a
n
d
N
ao
ki
20
04
Tr
og
lo
dy
te
s
7
6
3
3.
29
2
0.
04
(0
.96
8)
0.
24
0.
03
0.
5
R
ic
e
e
t
a
l.
19
99
Z-
te
st
2
1.
03
(0
.30
0)
A
ve
ra
ge
12
11
5
1.
25
1.
15
1.
17
0.
4
1
Sp
ec
ie
s
in
e
a
c
h
ge
og
ra
ph
ic
re
gi
on
o
f
in
te
re
st
.
2
N
um
be
r
o
f
sp
ec
ie
s
fo
r
w
hi
ch
D
N
A
sa
m
pl
es
w
e
re
a
v
a
ila
bl
e.
3
M
ax
im
um
n
u
m
be
r
o
f
re
gi
on
al
ly
sy
m
pa
tr
ic
sp
ec
ie
s
in
ge
og
ra
ph
ic
re
gi
on
.
4
A
pp
ro
xi
m
at
io
n
o
f
th
e
la
g
tim
e
to
sp
ec
ia
tio
n
e
st
im
at
ed
a
s
yo
un
ge
st
si
st
er
sp
ec
ie
s
w
ith
in
th
e
ge
og
ra
ph
ic
re
gi
on
o
f
in
te
re
st
in
e
a
c
h
ta
xo
n.
5
Sp
ec
ia
tio
n
(b
)a
n
d
e
x
tin
ct
io
n
(d
)r
a
te
s
e
st
im
at
ed
u
si
ng
a
bi
rth
de
at
h
m
o
de
l.
6
Ex
pe
ct
ed
n
u
m
be
r
o
f
lin
ea
ge
s
a
t
pr
es
en
t
if
in
iti
al
ra
te
s
o
f
di
ve
rs
ifi
ca
tio
n
re
m
a
in
ed
c
o
n
st
an
t.
In
iti
al
di
ve
rs
ifi
ca
tio
n
ra
te
s
w
e
re
c
a
lc
ul
at
ed
fr
om
th
e
fir
st
fiv
e
n
o
de
s
in
e
a
c
h
ph
yl
og
en
y.
Se
e
te
xt
fo
r
de
ta
ils
.
*
A
ss
um
in
g
a
2%
m
o
le
cu
la
r
c
lo
ck
,N
yc
tib
iu
s
do
es
n
o
t
sp
ec
ia
te
du
rin
g
th
e
la
st
10
m
ill
io
n
ye
ar
s.
It
se
e
m
s
u
n
lik
el
y
th
at
th
e
sp
ec
ia
tio
n
pr
oc
es
s
(ev
olu
tio
n
o
f
re
pr
od
uc
tiv
e
is
ol
at
io
n)
w
o
u
ld
re
qu
ire
m
o
re
th
an
a
fe
w
m
ill
io
n
ye
ar
s,
th
us
th
is
lo
ng
tim
e
pe
rio
d
(g
n
)p
ro
ba
bl
y
re
pr
es
en
ts
a
ge
nu
in
e
sl
ow
do
w
n
in
th
e
ra
te
o
fs
pl
itt
in
g.
H
ow
ev
er
,
it
is
n
o
t
kn
ow
n
if
in
tra
sp
ec
ifi
c
sp
lit
tin
g
e
v
e
n
ts
ha
ve
o
c
c
u
rr
e
d
a
lo
ng
g n
.
If
g n
is
n
o
t
e
x
c
lu
de
d,
g
e
qu
al
s
2
2.
3
(P
5
0.
02
).
845SPECIES ACCUMULATION IN NEOTROPICAL BIRDS
ple of Neotropical taxa. Phylogenetic analyses have mostly
been confined to regions amenable to genetic sampling. Spe-
cies restricted to countries such as Colombia and Venezuela
are poorly represented in the phylogenies included here. I do
not expect these potential biases to greatly affect the patterns
of diversification uncovered in this study. Nevertheless, fur-
ther sampling undoubtedly will provide a more complete un-
derstanding of Neotropical diversification.
For each taxon, phylogenetic trees were generated and cal-
ibrated so that branch lengths were proportional to time as
follows. Protein coding mitochondrial DNA sequences were
obtained from Genbank or were sequenced for this project
(see Table 1 in Supplementary Material available online at:
http://dx.doi.org/10.1554/05-272.1.s1). Phylogenetic analy-
ses were performed with multiple outgroups in MrBayes ver-
sion 3.0b4 (Huelsenbeck and Ronquist 2001) under the GTR-
g model of evolution. All Bayesian analysis were run for two
million generations and were sampled every 200 generations.
The first 500,000 generations were excluded as the burn-in
period and trees sampled from the remaining 1.5 million
generations were used to construct majority-rule consensus
cladograms. Parameters of the GTR-g model were then es-
timated from the Bayesian consensus trees using maximum
likelihood in PAUP* ver. 4.0b10 (Swofford 2002). These
parameter estimates were used to obtain maximum-likelihood
estimates of branch lengths along the Bayesian consensus
topologies. The only exception was that for Tangara I used
a published Bayesian topology (fig. 2 in Burns and Naoki
2004) and then calculated branch lengths along it using max-
imum likelihood. Penalized likelihood methods, implemented
in r8s (Sanderson 2003), were then used to create ultrametric
trees that allow for local rate variation in the molecular clock.
The cross validation routine implemented in r8s was used to
estimate the appropriate value of the smoothing parameter
for each tree. Branch lengths generated using penalized like-
lihood are proportional to time, but require calibration. The
timing of the basal most split within each taxon was used as
a calibration point. The timing of this split was estimated
with maximum likelihood in PAUP* by determining branch
lengths under the assumption of a global clock and applying
a molecular clock calibration to date this split.
Uncertainty if the rate of molecular evolution is constant
through time and across taxonomic groups needs to be ac-
counted for when dating splitting events. Calibrations ob-
tained for several orders of birds suggest an avian molecular
clock of approximately 2% per million years for protein-
coding mitochondrial DNA (see note 11 in Klicka and Zink
1997). Nevertheless, the validity of this rate has been chal-
lenged due to inconsistent phylogenetic methods used to ar-
rive at this calibration (Lovette 2004a). Moreover, this rate
may not be valid for splitting events near the recent (Garcia-
Moreno 2004; Penny 2005; Ho et al. 2005). To address these
issues, I recalibrated published avian calibrations using GTR-
gamma distances. I obtained additional calibrations using fos-
sil material and island ages (Weir, unpubl. data). A total of
47 avian calibrations from 19 families were obtained for the
mitochondrial cytochrome b gene. Some calibrations ob-
tained for splitting events less than 0.5 mya were much higher
than the 2% rate. However, calibrations obtained for splitting
events between 0.5 and 11 mya closely clustered around a
rate of 2.0%. I used this rate throughout this study.
The resulting calibrated, clocklike trees provide useful
sources of information for analyzing both the timing and rate
of diversification within each geographic region. Nodes in
such phylogenies provide estimates of the dates when species
diverged (population splitting). Node ages actually measure
the coalescence times of DNA haplotypes which may predate
population subdivision. The discrepancy results due to the
presence of polymorphism within populations at the time of
splitting. Assuming that ancestral levels of polymorphism are
similar to current levels, then the mean divergence within
current populations can be used to correct splitting dates.
This is done by subtracting the mean intraspecific divergence
from coalescent dates (Nei and Li 1979; Avise et al. 1998).
I estimated the average intraspecific GTR-gamma divergence
between individuals of a species (see Table 2 in Supple-
mentary Material available online). If species possessed ge-
netic subdivisions then I estimated average divergence be-
tween individuals at the phylogroup level following Avise et
al. (1998). These estimates were derived from available pop-
ulation level phylogenetic datasets for Neotropical birds and
often come from different taxa then those analyzed here.
Nevertheless, these corrections are assumed to be reflective
of the Neotropical avifauna as a whole. Throughout this
study, coalescence dates are reported and are used as a max-
imum estimate of the age at which population divergence
occurred. Estimates of mean intraspecific divergences are
then used to explore the magnitude of the discrepancy be-
tween node ages and splitting ages.
Ancestor State Reconstructions
I analyzed the timing and rate of diversification in the
Neotropics for lowland and highland faunas separately. The
division between the lowlands and highlands was drawn at
1000 m, the approximate upper limit of the tropical lowland
habitats and the lower limit of subtropical montane habitats.
Neotropical species whose elevational distributions were pre-
dominantly above or below 1000 m were assigned to highland
and lowland faunas, respectively. However, in Patagonia, al-
pine habitats typical of the high Andes further north descend
to sea level. The few species included in this dataset that
occur there were considered to belong to the highland avi-
fauna.
Ancestor state reconstruction was used to classify interior
nodes to their appropriate faunas. Species in each tree were
classified as highland, lowland, Caribbean, North American,
or other. Ancestor state reconstructions either assigned nodes
to one of these faunas or designated them as dispersal events
from one fauna to another. Dispersal events between faunas
occur at nodes in which each of the sister lineages occur in
different faunas (Fig. 1a). A splitting event within a fauna
occurs at nodes in which each of the daughter lineages occur
within the fauna. Mesquite (Maddison and Maddison 2003)
was used to obtain the most parsimonious ancestor state re-
construction for each phylogenetic tree (see Supplementary
Material available online at: http://dx.doi.org/10.1554/
05-272.1.s2). For several taxa (Amazona, Icterus, Tangara),
multiple most parsimonious reconstructions were obtained.
846 JASON T. WEIR
FIG. 1. Example of how patterns of species accumulation are ob-
tained from a reconstructed phylogenetic tree. (A) a reconstructed
phylogenetic tree containing species within (black) and outside of
(grey) a geographic region of interest. For lineage through time
(LTT) analysis, only node ages representing splitting events within
the geographic region of interest (black circles) or that represent
colonization from an outside source followed by speciation (black
arrow) were used. Nodes leading to the formation of a clade outside
of the region of interest were excluded (grey arrows) because they
do not represent the addition of a new lineage to the fauna. (B)
Internode distances for geographic region of interest labelled g2 to
g8. (C) LTT plot for taxa in the geographic region of interest. Values
of the g statistic are equal to zero when the slope of the LTT plot
is constant through time. Downturns in the slope produce negative
g-values whereas upturns produce positive values.
In such cases a maximum-likelihood method of ancestor state
reconstruction was used to differentiate between the com-
peting alternatives. For maximum-likelihood reconstructions
of geographic origin, a punctuation model is most appropriate
because it places character change at the time of splitting (at
the node) rather than along branches. This model was im-
plemented by constraining all internode branches to have
equal length (Pagel 1994). Node reconstructions with the
highest likelihood were chosen. These methods of recon-
structing ancestral states assume that the most parsimonious
or the most likely reconstruction is the correct one. This
assumption is probably valid for phylogenetic trees with few
transitions between states (most taxa in this study). More
uncertainty exists when frequent transitions occur (e.g., Tan-
gara). In addition, I assume that transitions occur at nodes.
However, extinction eliminates nodes and may result in tran-
sitions being pushed back to earlier nodes in a tree.
Timing and Rates of Diversification
Analysis within taxa. To investigate the timing and rate
of diversification within taxa, I constructed plots of the log
number of lineages (species) through time (lineages through
time or LTT plots; Fig 1c) for each taxon. Under a null
hypothesis of a constant speciation rate with no extinction
(pure birth model), the number of lineages increases expo-
nentially through time (Yule 1924; Nee 2004) and forms a
straight line on an LTT plot with slope equal to the speciation
rate. I used a method similar to that of Pybus and Harvey
(2000) to test the overall fit of a LTT plot to the pure birth
expectation of a constant slope. The g-statistic they develop
compares the relative position of node ages in a phylogenetic
tree to that expected under the pure birth model. For a phy-
logeny with n taxa, let g1 be the distance between the root
of the tree and the first node, let g2, g3,. . . gn21 be the in-
ternode distances, and gn be the distance between the most
recent node and the present (Fig 1b). The statistic I use here
is identical to that developed by Pybus and Harvey (2000),
except that it excludes gn. This last interval should be ex-
cluded from real phylogenies because unlike the simulated
phylogenetic trees used by Pybus and Harvey there is no
splitting event at the present and thus the interval gn is not
drawn from the same distribution as other internode distances
(i.e., the next splitting event may occur in the future or it
may have already occurred but is not taxonomically recog-
nized as a species). The statistic follows
n22 i1 Skg 2
O O k
1 2 1 2[ ]n 2 m 2 1 2i5m k5m
g 5 ,
1
S
!12(n 2 m 2 1)
n21
S 5 jg (1)
O j
1 2j5m
where S is the sum of the branch lengths in the phylogeny
(excluding the interval gn) and m is the number of initial
lineages. Under the pure birth expectation of exponential
growth, g approaches a standard normal distribution with
mean equal to 0.
Departures from the pure birth model can be detected by
a g-value that is either too large or too small. Values of g
. 0 indicate that internode distances are shorter than expected
towards the recent, which is also reflected in an upturn in
the LTT plot (Fig. 1c). Simulations of phylogenetic trees
demonstrate that this can result if the rate of speciation in-
creased through time (see Fig.1c in Supplementary Material
available online). Values of g , 0 indicate that internode
847SPECIES ACCUMULATION IN NEOTROPICAL BIRDS
distances are longer than expected towards the recent, which
is reflected in a downturn in the LTT plot (Fig. 1c). This can
result if the rate of speciation has declined through time (see
Fig. 1f in Supplementary Material available online). Values
of g greater than 1.96 or less than 21.96 are significantly
different at the 5% level from the pure birth expectation.
Extinction may also result in departure from the pure birth
model. Simulations using a variety of extinction rates dem-
onstrate that constant or increasing rates of extinction usually
increased and more rarely decreased g-values slightly, but
not significantly (see Fig. 1 in Supplementary Material avail-
able online). Significantly positive and negative values of g
were only obtained in simulations where speciation rate in-
creased or decreased respectively.
Some of the taxa included in this study possessed one or
more clades distributed outside of the geographic region of
interest (Fig. 1a, gray arrows). These were simply pruned so
that all resulting nodes represented diversification events
within the region of interest. However, a small proportion of
taxa (four of 26) exhibited a more complex biogeographic
history in which a clade distributed outside of a region back
colonized into the region (Fig. 1a, black arrow). Lineages
resulting from such back-colonization events usually pos-
sessed at least one node that represented splitting within the
region of interest and were included in the LTT analysis. The
node at which the back colonization occurred represents the
addition of a new lineage to the fauna following immigration.
This node was included in the analysis even though it does
not represent splitting within the fauna. Exclusion of this
node resulted in similar LTT plots and values of g.
Missing taxa may result in biased diversification patterns.
As many as 23% of species were missing from phylogenetic
trees (Table 1). Missing taxa may result in artificial down-
turns in LTT plots. The effect of missing taxa can be corrected
for in the g statistic using Monte Carlo simulation (Pybus
and Harvey 2000). The following method assumes that miss-
ing taxa are randomly distributed on the tree. Ten thousand
pure birth trees with n tips were simulated in Phyl-O-Gen
(Rambaut 2002) and k tips were randomly pruned from each,
where n is the number of species in a taxon and k is the
number of species sampled. The g-statistic was calculated
for each simulated tree. Average values of g in simulated
pure birth trees equal zero. When tips are deleted in simulated
trees, average values are less than zero. The difference is
proportional to the expected discrepancy in actual calculated
g-values. Calculated values of g were corrected by subtract-
ing the mean value in simulated trees. The resulting corrected
values were only marginally greater than calculated values
suggesting that missing taxa did not have a large effect. The
distribution of values in the simulated datasets were used to
determine the level of significance.
Lineages through time plots use splitting times from phy-
logenetic reconstructions of species level taxa. However,
there is a lag time between lineage splitting and the time
when lineages are recognized as separate species. As a result,
older lineages are more likely to be recognized as distinct
species today than younger lineages. It follows that there are
likely to be recent lineage splitting events in the tree that are
not recorded because the resulting taxa are not recognized as
distinct species (phylogroups hereafter). Some of these phy-
logroups will evolve to become species in the future and these
particular lineages really should be included in LTT plots
and the g-statistic. Failure to include such splits may also
result in an artificial downturn in LTT plots towards the pres-
ent. Most phylogroups are likely to be recent in age and do
not confound the LTT analysis because they date to the time
interval gn which is excluded from LTT plots and the g-
statistic. Nevertheless, some of these splits may predate gn.
This is especially true when the interval gn, is short. Because
detailed intraspecific sampling was lacking for most of the
species in this dataset, I was not able to determine the effect
of missing phylogroups on diversification rates.
Fauna wide trends in the mode of diversification within taxa
were analyzed using a combined Z test (Whitlock 2005):
l
Z 5 g ?l . (2)
O p
@p51
Under the null model of pure birth, Z has a standard normal
distribution where l is the number of taxa being combined
and gp is the g-statistic for taxon p. Values of Z greater than
1.96 or less than 21.96 are significantly different at the 5%
level from the pure birth expectation. The Z test of combined
g-values identifies trends towards negative or positive values
across a series of taxa.
I tested for density dependent cladogenesis in lowland and
highland taxa. Density dependent cladogenesis may occur if
speciation rates slow through time as ecological niches be-
come progressively occupied. Alternatively, if the processes
that promote speciation diminish through time, the speciation
rates will slow in a correlated fashion irrespective of species
density. Negative values of g reflect a slowdown in speciation
through time. If a fauna experiences density dependent clad-
ogenesis, then a negative relationship should exist between
g and the maximum number of sympatric species. I tested
for this relationship using a regression analysis. The maxi-
mum number of regionally sympatric species in each taxon
was determined by overlaying range maps for each species
and determining the geographic location with the highest
density of species.
Finally, extinction rates were estimated directly from phy-
logenetic trees. Equation 17 in Nee et al. (1994) gives the
likelihood of an internode distance for a given extinction and
speciation rate. Following methods similar to Barraclough
and Vogler (2002), I used the ??optim?? function in R (R
Development Core Team 2005) to obtain estimates of ex-
tinction and speciation rates that maximized the likelihood
of internode distances g2 to gn21 (Fig.1) for each tree. The
utility of this estimate is limited because it assumes rates are
constant, when in reality rates may vary.
Faunawide analysis. To illustrate fauna wide rates of di-
versification during different time periods, I used the Kendall/
Moran estimator to calculate the net diversification rate dur-
ing million year intervals for lowland and highland regions
separately (Kendall 1949; Moran 1951; Hey 1992; Baldwin
and Sanderson 1998; Nee 2001). For each of a series of
phylogenetic trees, the per lineage diversification rate during
a time window t is
b(t) 5 (n 2 m) / S (3)
848 JASON T. WEIR
where n and m are the number of lineages at the end and
beginning of the time period t and S is the sum of branch
lengths (excluding the time interval gn in each taxon; see Eq.
1) occurring within t. A single rate of diversification during
each one million year time interval was obtained by summing
n 2 m and S across all phylogenies in lowland and highland
regions separately. The variance of the estimate b(t) provided
by Nee (2001) is
2Var 5 b / (n 2 m) (4)
and was used to determine 95% confidence intervals (eq. 17
in Nee 2001). Diversification rates were calculated back to
eight mya (late Miocene), because not enough nodes were
available before this period.
This faunawide analysis is useful for uncovering patterns
in net diversification rates through time. If Pleistocene cli-
matic fluctuations were a major factor in promoting specia-
tion, then diversification rates should increase during the late
Pliocene and early Pleistocene and again during the late Pleis-
tocene when climatic fluctuations were most intense.
This faunawide analysis of diversification rates assumes
that rates (b) are constant across taxa. To test for rate con-
stancy across taxa, I compared overall diversification rates
within each taxa using the joint scaling test borrowed from
quantitative genetics (Lynch and Walsh 1998, pp. 216). This
test compares observed values of a parameter, in this case b,
calculated for each of k taxa with the expected value of the
parameter b? if all taxa shared the same value. The expected
value is
T 21 21 T 21b? 5 (M V M) M V b (5)
where V is the covariance matrix with diagonal elements
equal to the variance of each bk (Eq. 4) and M is a matrix
with one column of length k with each element equal to one.
The statistic follows a chi-squared distribution with k 2 1
degrees of freedom
k 2(b? 2 b )i i2
x 5 . (6)
O Var(b? )i51 i
Constant diversification rates across taxa were rejected for
both lowland (x2 5 26.6, df 5 16, P 5 0.05) and highland
(x2 5 24.57, df 5 10, P 5 0.006) datasets. To address the
error associated with significantly different diversification
rates across taxa, I systematically removed outlier taxa with
extremely high values of b. When Crax was excluded, con-
stant rates in lowland faunas could not be rejected (x2 5
21.07, df 5 15, P 5 0.13). When Carduelis, Cranioleuca,
and Muscisaxicola were excluded, constant rates in highland
faunas could not be rejected (x2 5 8.7, df 5 7, P 5 0.27).
For both lowland and highland datasets, rates were analyzed
through time using both the complete datasets and datasets
with outliers excluded.
To determine if faunawide estimates of diversification rate
b increased during the late Pliocene and Pleistocene, rates
before and after 2.5 and 1.0 mya were compared using the
joint-scaling test (Eq. 6). Rates were also compared between
highland and lowland faunas before and after 2.5 mya to
determine if diversification rates were different in each fauna.
RESULTS
A total of 198 lowland and 146 highland species were
included in the 27 taxa. Ancestor state reconstructions re-
covered 313 nodes with both descendents in the Neotropics.
Fifty-two percent of nodes were reconstructed as divergence
events within the lowland fauna (Fig. 2a), 33% within the
highland fauna (Fig. 2b), and 15% as interchange events be-
tween these faunas (Fig. 3). In the lowlands, 26% of nodes
(43% of terminal species) dated (coalesced) to the glacial
periods of the late Pliocene and Pleistocene and 5% (12% of
species) to the late Pleistocene. The frequency of nodes de-
creased over the past 1.5 million years similar to simulations
in which speciation rates declined through time (see Fig. 1f
in Supplementary Material available online). In contrast,
highland faunas had 42% and 21% of nodes (43% and 27%
of species) dating to these periods, respectively. The shape
of the distribution of nodes had a strong upturn near the recent
that appeared intermediate between simulations in which spe-
ciation or extinction rates increased towards the present (see
Fig. 1 in Supplementary Material available online). Thus,
even though widespread diversification occurred during the
periods of climatic instability in both faunas, only the pattern
in the highland fauna was consistent with an increase in di-
versification rate during glacial periods (though extinction
may have contributed to this pattern).
Nodes representing dispersal events between highland and
lowland regions were most frequent during the last one mil-
lion years and during the late Miocene and Pliocene (Fig. 3).
Figure 3 also includes dates for intraspecific dispersal events
for species distributed in both faunas. A few additional in-
traspecific interchange events are unrecorded because the rel-
evant sequence data was not available. These are expected
to date near the recent.
These dates represent coalescent dates. The actual dates of
population splitting may occur after the coalescent dates if
populations possessed polymorphism at the time of splitting.
Current levels of intraspecific polymorphism are low, sug-
gesting that on average lowland and highland coalescent dates
predate actual population splitting by only 0.35 and 0.2 mil-
lion years, respectively (see Table 2 in Supplementary Ma-
terial available online). These corrections are similar to those
reported for Northern Hemisphere taxa (Moore 1995). Ap-
plying these corrections did not greatly change any of the
results of this study.
Patterns in LTT plots also suggest that the timing of di-
versification was different in lowland and highland taxa (Fig.
2). Many lowland taxa had very steep slopes between eight
and four mya, suggesting rapid diversification during this
period. Only one lowland taxon experienced rapid diversi-
fication, primarily within the late Pliocene and Pleistocene
(Crax). The remaining lowland taxa exhibited slower, but
relatively constant rates of diversification through time. In
contrast, Pleistocene diversification was most prevalent in
highland taxa (Fig. 2b). In the LTT plots, four of the 11
highland taxa (Cranioleuca, Carduelis, Cinclodes, and Mus-
cisaxicola) displayed steep slopes during the Pleistocene sug-
gesting rapid rates of speciation during this period. The re-
maining seven taxa diversified primarily before the Pleisto-
849SPECIES ACCUMULATION IN NEOTROPICAL BIRDS
FIG. 2. Lineage through time (LTT) plots and faunawide rates of diversification through time for Neotropical lowland (A) and highland
taxa (B). Maximum-likelihood estimates of rates of diversification are plotted with 95% confidence intervals. Two highland datasets
were used to analyze faunawide rates: all highland genera (black) and Carduelis, Cranioleuca and Muscisaxicola excluded (gray). Numbers
refer to the following taxa: (1) Crax, (2) Muscisaxicola, (3) Cranioleuca, (4) Carduelis, and (5) Cinclodes. The levels of shading on
histograms and pie charts increase from light to dark for warm periods of the Miocene and early Pliocene (in yellow online), mild ice
ages of the late Pliocene and early Pleistocene (in blue online) and severe ice ages of the late Pleistocene (in red online).
FIG. 3. Nodes representing interchange events between highland
and lowland faunas as reconstructed from ancestor state reconstruc-
tions for 27 Neotropical taxa. Dispersal from lowland to highland
faunas (gray) and highland to lowland faunas (black).
cene and had less steep slopes but, unlike lowland taxa, were
not aggregated during any given time period.
Lineage through time plots and the g-statistic further sug-
gest that the rate of diversification decreased through time
in most lowland taxa but remained constant in most highland
taxa. Most lowland taxa displayed a downturn towards the
recent in their LTT plots, consistent with a decrease in spe-
ciation rates through time (Fig. 2a). Values of the g-statistic
were likewise negative in 12 of 17 lowland taxa (Table 1)
and were significantly negative in five taxa. No taxa had
significantly positive values. The Z test of combined g-values
(Eq. 2) rejected the pure birth process (Table 1) for the low-
land avifauna as a whole, suggesting a significant faunawide
trend towards decreasing speciation rates through time.
In contrast to the lowlands, the relatively constant slopes
in LTT plots for most highland genera were reflected in g-
values closer to zero, the pure-birth expectation. One genus
had a significantly positive g-value and one had a signifi-
cantly negative value. The remaining genera were not sig-
nificantly different from the pure birth expectation. The Z
test of combined g-values was negative, but failed to reject
the pure birth process for the avifauna as a whole (Table 1),
850 JASON T. WEIR
FIG. 4. The relationship between the maximum number of re-
gionally sympatric species in each taxa and values of the corrected
g-statistic for (A) lowland and (B) highland Neotropical faunas.
suggesting that the mode of diversification within highland
taxa is not significantly different from exponential growth.
In the lowlands, values of the g-statistic were negatively
correlated with the maximum number of regionally sympatric
species in each taxon (Fig. 4; r2 5 0.57, b 5 20.360 60.083
SE, T 5 24.28, P , 0.0007 two-tailed), suggesting that the
rate of speciation declined as the number of sympatric species
increased. This relationship remained significant after cor-
recting for taxon size and taxon age (age of first split within
taxon) in a multiple repression (b 5 20.297 60.127 SE,T
5 22.34, P , 0.036 two-tailed). However, highland taxa
exhibited no significant relationship between the values of
the g statistic and the maximum number of sympatric species
per taxon (Fig. 4; r2 5 0.19, b 5 20.192 60.187 SE, T 5
21.03, P 5 0.34, two-tailed).
Maximum-likelihood estimates of extinction rates (d) were
low relative to speciation rates (b) for most taxa in both
lowland and highland faunas (Table 1). At face value, these
results suggest that extinction has probably not contributed
significantly to observed patterns of species accumulation in
either fauna. However, extinction rates estimated from re-
constructed phylogenetic trees should be viewed with caution
because these estimates may be biased. For instance, trees
that display a downturn in LTT plots (or negative g-values)
exhibit negative extinction rates when maximum-likelihood
searches are not constrained, but extinction rates of zero when
they are constrained to positive values (values reported in
Table 1 were constrained). In addition, these estimates of
extinction assume a constant rate of extinction through time.
Although separate rates of extinction for different time pe-
riods can be estimated from large phylogenetic trees (Bar-
raclough and Vogler 2002), most of the taxa included here
did not have enough nodes.
The Z test of combined g-values detects common trends
toward increasing or decreasing rates through time in a series
of taxa. However, because this test lacks a temporal timescale
a significant trend does not suggest that all taxa experienced
increasing or decreasing rates concordantly. For example,
two taxa may exhibit similar g-values, but if the timing and
rate of diversification differ between them, then the resulting
net patterns of diversification may give different results. This
is best illustrated in the highland fauna where several taxa
exhibit very steep slopes during the Pleistocene period in
their LTT plots (Carduelis and Cranioleuca) yet had similar
g-values to taxa that did not have steep slopes and diverged
mostly before the Pleistocene (Fig. 2). Likewise, the signif-
icant trend towards negative values of g in the lowlands does
not suggest that diversification rates decline over the same
time periods in lowland taxa. Although LTT plots for many
lowland taxa do appear to decline somewhat concordantly,
this is not true of all taxa (Fig. 2a).
The Kendal-Moran estimator was used to determine faun-
awide values of diversification rate during million-year in-
tervals. In the lowlands, both the full dataset and the dataset
with Crax excluded exhibited very similar rates during each
time period, thus only the full dataset was used. Faunawide
diversification rates declined steadily through time from a
high of 0.35 species per lineage/Myr between seven and eight
mya to 0.16 species per lineage/Myr between the recent and
one mya (Fig. 2a). Diversification rates were almost signif-
icantly lower after 2.5 mya (x2 5 3.60, df 5 1, P 5 0.058),
but no difference was found before and after 1.0 mya (x2 5
2.34, df 5 1, P 5 0.126).
In highland taxa, faunawide rates during the late Pliocene
and early Pleistocene were not different from previous rates
(x2 5 0.54, df 5 1, P , 0.46). However, rates significantly
doubled (as high as 0.52 species per lineage/Myr) during the
late Pleistocene (x2 5 7.16, df 5 1, P , 0.008; Fig. 2b).
This late Pleistocene increase was not significant when the
three taxa (Carduelis, Cranioleuca, and Muscisaxicola) with
the highest birth rates and which diversified primarily within
the Pleistocene were excluded (x2 , 0.001, df 5 1, P 5
0.98). In addition, late Pliocene and Pleistocene rates in the
highlands were significantly higher than those in the lowlands
when using the complete highland dataset (x2 5 5.1, df 5
1, P , 0.024), but when highland outliers were excluded
lowland rates were higher (x2 5 6.51, df 5 1, P 5 0.01).
No difference in rates between these faunas occurred before
the Pleistocene (all highland taxa included x2 5 1.88, df 5
1, P 5 0.17, outliers excluded x2 5 2.58, df 5 1, P 5 0.11).
These data suggest that recent climatic fluctuations had an
effect on faunawide diversification in the highlands but not
the lowlands. However, the faunawide increase in highland
faunas resulted from elevated late Pleistocene rates in a subset
of highland taxa.
DISCUSSION
Lowland and highland faunas exhibited divergent patterns
of species accumulation, suggesting that different processes
resulted in their diversification. In the highlands, faunawide
diversification rates increased throughout the Pliocene and
Pleistocene and culminated in a late Pleistocene diversifi-
cation rate more than double previous values. This increase
in rate resulted in a burst of diversification during the last
one million years (Fig. 2b) consistent with the hypothesis
that climatic fluctuations resulted in a recent build-up of spe-
cies in this fauna. In contrast, lowland diversification rates
slowed through time (Fig. 2a) and were lowest during the
late Pleistocene. Though the faunawide slow down was not
quite significant, this result demonstrates that rates did not
851SPECIES ACCUMULATION IN NEOTROPICAL BIRDS
increase through the Pleistocene as expected if climatic fluc-
tuations drove lowland diversification.
What processes promoted these divergent patterns of spe-
cies accumulation in lowland and highland faunas? In low-
land taxa, g-values revealed a significant trend towards de-
creasing diversification rates (Table 1). This pattern is con-
sistent with the faunawide decrease in diversification rate and
suggests that a decline in rates within taxa resulted in the
faunawide pattern. Simulations demonstrate that a decline in
the rate of speciation was more likely to generate such a
strong pattern than an increase in the rate of extinction (see
Fig. 1 in Supplementary Material available online). Likewise,
estimates of extinction rates, although potentially biased,
were low in most taxa (Table 1). The decrease in speciation
rates may simply reflect the lack of geographic opportunity
for speciation towards the recent. Alternatively, the decrease
may reflect density dependent cladogenesis. The significantly
negative relationship between the value of g and the number
of sympatric species in each taxon (even after correcting for
taxon age and size) suggests that density dependent clado-
genesis is responsible for the slowdown in species accu-
mulation in lowland faunas. These preliminary results are
consistent with the view that low extinction rates have al-
lowed the accumulation of high species diversity and suggest
that the number of species in lowland faunas may be ap-
proaching their capacity. Possible low extinction rates for
birds contrast preliminary findings from Amazonian paleo-
pollen records in which species diversity may have declined
from the Miocene to the present (van der Hammen and Absy
1994; Hooghiemstra and van der Hammen 1998; van der
Hammen and Hooghiemstra 2000; Willis and Niklas 2004).
Further estimates of extinction rates from other taxonomic
groups are needed in order to establish the role of extinction
in the build-up of high species diversity in the Neotropics.
Could the observed slowdown in speciation rates through
time in lowland taxa be an artifact of not sampling intraspe-
cific splits (i.e., phylogroups) in this study? Indeed, a number
of phylogenetic studies have uncovered genetically distinct
lineages within many currently named lowland species (Al-
eixo 2002, 2004; Marks et al. 2002; Joseph et al. 2003; Burns
and Naoki 2004; Joseph et al. 2004; Lovette 2004b; Armenta
et al. 2005; Cheviron et al. 2005b). However, detailed pop-
ulation level sampling was not available for most species in
my dataset. To address this question, I determined the initial
diversification rate (slope) for each taxon from its LTT plot
using only the first five nodes. I extended this rate to the
present to determine the number of expected lineages if di-
versification rates had remain constant. In highland taxa, the
expected number of lineages was very similar to the actual
number in all taxa, with 0.4 additional lineages expected
within each species on average (Table 1). However, in many
lowland taxa the expected number of lineages was much high-
er than the actual number (88 additional lineages per species
on average). It is doubtful that the actual number of lineages
in many lowland species approaches, let alone surpasses their
pure birth expectation. Thus, it appears that even if unrec-
ognized lineages were included in the LTT plots, there still
would be no evidence for a Pleistocene increase in lowland
diversification rates. Detailed phylogenies that include all
genetically distinct lineages regardless of taxonomic status
are needed for confirmation.
In contrast to the lowlands, LTT analysis suggests that
most highland taxa did not exhibit a significant trend away
from the pure birth expectation. This is not unreasonable
given that continual uplift of highland regions could provide
ongoing opportunities for speciation. Likewise, evidence for
density dependence was lacking (Fig. 4), further suggesting
that ecological opportunity is not currently a limiting factor
in highland diversification. This pattern of constant diver-
sification rates within taxa did not match the faunawide pat-
tern of increasing diversification rates during the late Pleis-
tocene. This apparent discrepancy is best explained by the
faster speciation rate in taxa that diversified primarily within
the Pleistocene (slopes in LTT plots for Carduelis, Crani-
oleuca, Cinclodes, and Muscisaxicola are steeper than in taxa
which diversified primarily before the Pleistocene, Fig. 2b).
Thus, it appears that only a subset of highland taxa strongly
contributed to the faunawide increase in diversification rates
during the late Pleistocene.
Could extinction also generate the apparent late Pleisto-
cene increase in diversification rate? Simulations that in-
cluded extinction often did result in a recent upturn, although
this effect was less pronounced than in models with an in-
crease in speciation rate (see Fig. 1 in Supplementary Ma-
terial available online). Nevertheless, estimates from high-
land phylogenetic trees (Table 1) suggest that extinction rates
(estimates assume constant rates through time) are low in
most highland taxa. Rapid speciation rates in a subset of
highland taxa probably drove the late Pleistocene increase.
However, until better estimates of extinction rates (i.e., non-
constant extinction rates) are obtained, the role of extinction
cannot be ruled out entirely.
These differences in species accumulation through time
resulted in differently aged faunas (see Fig. 1 in Supple-
mentary Material available online). The most striking dif-
ference is the abundance of highland species and scarcity of
lowland species dating to the late Pleistocene. Less than one-
fifth of lowland species date to this period even after cor-
recting for ancestral polymorphism. Yet, global climatic fluc-
tuations were most intense during this period. If these cli-
matic fluctuations were not enough to promote widespread
fragmentation and speciation in the lowlands, then it is un-
likely that the weaker climatic fluctuations of the late Plio-
cene and early Pleistocene were important in lowland diver-
sification either.
In contrast, the late Pleistocene increase in highland di-
versification rate resulted in a fauna with one-third of its
species dating to the last million years. Unlike lowland re-
gions, extensive evidence suggests that widespread alteration
of highland habitats occurred repeatedly during the late Pleis-
tocene. Extensive glaciation occurred throughout highland
regions (Hooghiemstra and van der Hammen 2004). Glaciers
undoubtedly fragmented the ranges of many Andean species
by providing a hard barrier between populations displaced
along the eastern and western slopes. In addition, cooling
resulted in an elevational migration of habitat zones to lower
altitudes resulting in an elevational compression of some
zones and expansion of others (van?t Veer and Hooghiemstra
2000; Hooghiemstra and van der Hammen 2004). The cor-
852 JASON T. WEIR
relation between the onset of severe glaciation in the Neo-
tropics about 0.8 to 0.9 mya (Bennett 1990; Hooghiemstra
et al. 1993) and the late Pleistocene increase in speciation
rates suggests a causal link.
Interchange between highland and lowland faunas also
played an important role in Neotropical diversification (Fig.
3). Dispersal events from the lowlands to the highlands oc-
curred primarily during the late Miocene and early Pliocene
when extensive uplift of the central and northern Andes pro-
vided new elevation zones and habitats. Thirty-three nodes
represent dispersal from lowlands to highlands compared to
146 nodes that represent divergence within highland faunas.
This fairly large proportion suggests that dispersal from low-
land regions contributed importantly to the build-up of spe-
cies diversity in the highlands. In contrast, Fjeldsa and Lovett
(1997a,b) envisioned dispersal out of highland regions as a
major source of species diversity for lowland faunas. Whereas
16 dispersal events from highland to lowland faunas are re-
constructed in this dataset, compared to the 198 splitting
events that occurred within the lowlands, dispersal from high-
land faunas represents only a small contribution to lowland
diversity. Rather, diversification within lowland regions was
the predominant mode by which species accumulated in this
fauna. Moreover, most dispersal events from the highlands
into the lowlands occurred during the last one million years
(Fig. 3) and correlate with major glacial cycles in the Andes
(Bennett 1990; Hooghiemstra et al. 1993). Glacial lowering
of elevational zones resulted in mixed floras (Colinvaux et
al. 1996, 2000; van der Hammen and Hooghiemstra 2000;
Bush et al. 2004) and presumably mixed faunas near the base
of highland regions that included both lowland and highland
components. This mixing may have facilitated adaptation to
and subsequent invasion of lowland regions (Rull 2005).
These results are inconsistent with the once prevalent view
that late Pliocene and Pleistocene climatic fluctuations drove
the recent buildup of species diversity in lowland Neotropical
faunas. Many Nearctic avian taxa also display decreasing
speciation rates through time (Zink and Slowinski 1995; Zink
et al. 2004), suggesting that the processes that promoted spe-
ciation in both faunas occurred primarily before the onset of
the late Pliocene and Pleistocene ice ages. Nevertheless, this
study suggests that the proportion of species of glacial age
is much higher in highland regions of the Neotropics where
expanding and retracting glaciers directly fragmented habi-
tats. A late Pleistocene increase in faunawide rates of diver-
sification correlates with the onset of severe glaciation in
highland regions and resulted in a fauna in which one third
of extant species are less than a million years old. Likewise,
ice sheets directly fragmented the high latitude boreal forests
of the Nearctic where a greater proportion of avian sister
species date to the Pleistocene than in subboreal regions
(Weir and Schluter 2004). Together, these studies suggest
that diversification rates in faunas distributed closest to the
expanding and retracting glaciers were most heavily impacted
by climatic fluctuations whereas faunas distributed further
from the glaciers were impacted to a lesser degree. Further
sampling of other glaciated regions is necessary to determine
the generality of this pattern.
ACKNOWLEDGMENTS
This manuscript is greatly indebted to previous molecular
based phylogenetic studies of birds (sources in Table 1). The
manuscript benefited greatly from discussions with D. Schlu-
ter who provided comments on statistical procedures. D.
Schluter, T. Vines, L. Harmon, I. Lovette, T. Price, and two
anonymous reviewers provided comments on earlier versions
of this manuscript. M. Whitlock, D. Irwin, M. Doebeli, A.
Albert, K. Marchinko, A. Waldron, and A. MacColl also
provided useful ideas. Tissue samples were provided by the
Field Museum of Natural History, Chicago, or collected by
the author. Financial support was supplied by a Natural Sci-
ences and Engineering Research Council doctoral fellowship
and a Smithsonian Tropical Research Institute Short Term
Fellowship to J. W and grants from Natural Sciences and
Engineering Research Council and Canada Foundation for
Innovation to D. Schluter.
LITERATURE CITED
Aleixo, A. 2002. Molecular systematics and the role of the ??var-
zea?????terra-firme?? ecotone in the diversification of Xiphor-
hynchus woodcreepers (Aves : Dendrocolaptidae). Auk 119:
621?640.
???. 2004. Historical diversification of a Terra-firme forest bird
superspecies: A phylogeographic perspective on the role of dif-
ferent hypotheses of Amazonian diversification. Evolution 58:
1303?1317.
Andriessen, P. A. M., K. F. Helmens, H. Hooghiemstra, P. A. Rie-
zebos, and T. van der Hammen. 1993. Absolute chronology of
the Pliocene-Quaternary sediment sequence of the Bogota area,
Colombia. Quat. Sci. Rev. 12:483?501.
Armenta, J. K., J. D. Weckstein, and D. F. Lane. 2005. Geographic
variation in mitochondrial DNA sequences of an Amazonian
nonpasserine: the black-spotted barbet complex. Condor 107:
527?536.
Avise, J. C., D. Walker, and G. C. Johns. 1998. Speciation durations
and Pleistocene effects on vertebrate phylogeography. Proc. R.
Soc. Lond. B 265:1707?1712.
Ayres, J. M., and T. H. Cluttonbrock. 1992. River boundaries and
species range size in Amazonian primates. Am. Nat. 140:
531?537.
Baldwin, B. G., and M. J. Sanderson. 1998. Age and rate of di-
versification of the Hawaiian silversword alliance (Compositae)
one. Proc. Natl. Acad. Sci. USA 95:9402?9406.
Barraclough, T. G., and A. P. Vogler. 2002. Recent diversification
rates in North American tiger beetles estimated from a dated
mtDNA phylogenetic tree. Mol. Biol. Evol. 19:1706?1716.
Bates, J. M., J. Haffer, and E. Grismer. 2004. Avian mitochondrial
DNA sequence divergence across a headwater stream of the Rio
Tapajos, a major Amazonian river. J. Ornithol. 145:199?205.
Bennett, K. D. 1990. Milankovitch Cycles and their effects on spe-
cies in ecological and evolutionary time. Paleobiology 16:11?21.
Bloemendal, J., and P. Demenocal. 1989. Evidence for a change in
the periodicity of tropical climate cycles at 2.4 myr from whole-
core magnetic-susceptibility measurements. Nature 342:
897?900.
Brown, K. S., P. M. Sheppard, and J. R. G. Turner. 1974. Quaternary
refugia in tropical America: evidence from race formation in
Heliconius butterflies 305. Proc. R. Soc. Lond. B 187:369?378.
Burns, K. J., and K. Naoki. 2004. Molecular phylogenetics and
biogeography of Neotropical tanagers in the genus Tangara.
Mol. Phylogenet. Evol. 32:838?854.
Bush, M. B., and M. R. Silman. 2004. Observations on Late Pleis-
tocene cooling and precipitation in the lowland Neotropics. J.
Quat. Sci. 19:677?684.
Bush, M. B., P. E. De Oliveira, P. A. Colinvaux, M. C. Miller, and
J. E. Moreno. 2004. Amazonian paleoecological histories: one
853SPECIES ACCUMULATION IN NEOTROPICAL BIRDS
hill, three watersheds. Palaeogeogr. Palaeoclimatol. Palaeoecol.
214:359?393.
Capparela, A. P. 1988. Genetic variation in Neotropical birds: im-
plications for the speciation process. Acta XIX Congr. Int. Or-
nithol. 101:189?208.
???. 1991. Neotropical avian diversity and riverine barriers.
Acta XX Congr. Int. Ornithol. 1:307?316.
Cerqueira, R. 1982. South American landscapes and their mammals.
Pp. 53?75 in M. A. Mares and H. H. Genoways, eds. Mammalian
biology in South America. Pymatuning Laboratory of Ecology,
University of Pittsburgh, Linesville, Pa.
Chesser, R. T. 2000. Evolution in the high Andes: the phylogenetics
of Muscisaxicola ground-tyrants. Mol. Phylogenet. Evol. 15:
369?380.
???. 2004. Systematics, evolution, and biogeography of the
South American ovenbird genus Cinclodes. Auk 121:752?766.
Cheviron, Z. A., A. P. Capparella, and F. Vuilleumier. 2005a. Mo-
lecular phylogenetic relationships among the Geositta miners
(Furnariidae) and biogeographic implications for avian specia-
tion in Fuego-Patagonia. Auk 122:158?174.
Cheviron, Z. A., S. J. Hackett, and A. P. Capparella. 2005b. Com-
plex evolutionary history of a Neotropical lowland forest bird
(Lepidothrix coronata) and its implications for historical hy-
potheses of the origin of Neotropical avian diversity. Mol. Phy-
logenet. Evol. 36:338?357.
Colinvaux, P. A., P. E. DeOliveira, J. E. Moreno, M. C. Miller, and
M. B. Bush. 1996. A long pollen record from lowland Amazonia:
Forest and cooling in glacial times. Science 274:85?88.
Colinvaux, P. A., P. E. De Oliveira, and M. B. Bush. 2000. Am-
azonian and Neotropical plant communities on glacial time-
scales: the failure of the aridity and refuge hypotheses. Quat.
Sci. Rev. 19:141?169.
Cox, D. R., and P. A. W. Lewis. 1966. The statistical analysis of
series of events. Methuen, London.
Darlington, P. J. 1957. Zoogeography: the geographical distribution
of animals. Wiley, New York.
Eberhard, J. R., and E. Bermingham. 2005. Phylogeny and com-
parative biogeography of Pionopsitta parrots and Pteroglossus
toucans. Mol. Phylogenet. Evol. 36:288?304.
Espinosa, de los, M. A. 1998. Phylogenetic relationships among the
trogons. Auk 115:937?954.
Fisher, A. G. 1960. Latitudinal variations in organic diversity. Evo-
lution 14:64?81.
Fjeldsa, J., and J. C. Lovett. 1997a. Geographical patterns of old
and young species in African forest biota: the significance of
specific montane areas as evolutionary centres. Biodivers. Con-
serv. 6:325?346.
???. 1997b. Biodiversity and environmental stability. Biodivers.
Conserv. 6:315?323.
Garcia-Moreno, J. 2004. Is there a universal mtDNA clock for birds?
J. Avian Biol. 35:465?468.
Garcia-Moreno, J., and J. Fjeldsa. 2000. Chronology and mode of
speciation in the Andean avifauna. Bonn. Zool. Monogr. 46:
25?46.
Garcia-Moreno, J., P. Arctander, and J. Fjeldsa. 1998. Pre-Pleis-
tocene differentiation among chat-tyrants. Condor 100:629?640.
???. 1999a. A case of rapid diversification in the Neotropics:
phylogenetic relationships among Cranioleuca spinetails (Aves,
Furnariidae). Mol. Phylogenet. Evol. 12:273?281.
???. 1999b. Strong diversification at the treeline among Me-
tallura hummingbirds. Auk 116:702?711.
Garcia-Moreno, J., J. Ohlson, and J. Fjeldsa. 2001. MtDNA se-
quences support monophyly of Hemispingus tanagers. Mol. Phy-
logenet. Evol. 21:424?435.
Grafe, K., W. Frisch, I. M. Villa, and M. Meschede. 2002. Geo-
dynamic evolution of southern Costa Rica related to low-angle
subduction of the Cocos Ridge: constraints from thermochron-
ology. Tectonophysics 348:187?204.
Gregory-Wodzicki, K. M. 2000. Uplift history of the Central and
Northern Andes: a review. Geol. Soc. Ame. Bull. 112:
1091?1105.
Hackett, S. J. 1996. Molecular phylogenetics and biogeography of
tanagers in the genus Ramphocelus (Aves). Mol. Phylogenet.
Evol. 5:368?382.
Haffer, J. 1969. Speciation in Amazonian forest birds. Science 165:
131?137.
???. 1974. Avian speciation in tropical South America, with a
systematic survey of the toucans (Ramphastidae) and jacamars
(Galbulidae), Publ. Nuttall Ornithol. Club no. 14.
???. 1990. Avian species richness in tropical South America.
Stud. Neotrop. Fauna Environ. 25:157?183.
???. 1997. Alternative models of vertebrate speciation in Ama-
zonia: an overview. Biodivers. Conserv. 6:451?476.
Haffer, J., and G. T. Prance. 2001. Climatic forcing of evolution
in Amazonia during the Cenozoic: on the refuge theory of biotic
differentiation. Amazoniana 16:579?605.
Hey, J. 1992. Using phylogenetic trees to study speciation and
extinction. Evolution 46:627?640.
Ho, S. Y. W., M. J. Phillips, A. Cooper, and A. J. Drummond. 2005.
Time dependency of molecular rate estimates and systematic
overestimation of recent divergence times. Mol. Biol. Evol. 22:
1561?1568.
Hooghiemstra, H. 1989. Quaternary and Upper-Pliocene glaciations
and forest development in the tropical Andes: evidence from a
long high-resolution pollen record from the sedimentary basin
of Bogota, Colombia. Palaeogeogr. Palaeoclimatol. Palaeoecol.
72:11?26.
Hooghiemstra, H., and T. van der Hammen. 1998. Neogene and
Quaternary development of the Neotropical rain forest: the forest
refugia hypothesis, and a literature overview. Earth-Sci. Rev.
44:147?183.
???. 2004. Quaternary Ice-Age dynamics in the Colombian An-
des: developing an understanding of our legacy. Philos. Trans.
R. Soc. Lond. B 359:173?180.
Hooghiemstra, H., J. L. Melice, A. Berger, and N. J. Shackleton.
1993. Frequency-spectra and paleoclimatic variability of the
high-resolution 30-1450-Ka Funza I pollen record (Eastern Cor-
dillera, Colombia). Quat. Sci. Rev. 12:141?156.
Hoorn, C. 1993. Marine incursions and the influence of Andean
tectonics on the Miocene depositional history of northwestern
Amazonia: results of a palynostratigraphic study. Palaeogeogr.
Palaeoclimatol. Palaeoecol. 105:267?309.
???. 1994. Fluvial paleoenvironments in the intracratonic Ama-
zonas basin (Early Miocene?Early Middle Miocene, Colombia).
Palaeogeogr. Palaeoclimatol. Palaeoecol. 109:1?54.
Hoorn, C., J. Guerrero, G. A. Sarmiento, and M. A. Lorente. 1995.
Andean tectonics as a cause for changing drainage patterns in
Miocene northern South America. Geology 23:237?240.
Huelsenbeck, J. P., and F. Ronquist. 2001. MRBAYES: Bayesian
inference of phylogenetic trees. Bioinformatics 17:754?755.
Irion, G., J. Muller, J. N. deMello, and W. J. Junk. 1995. Quaternary
geology of the Amazonian lowland. Geo-Marine Let. 15:
172?178.
Joseph, L., T. Wilke, and D. Alpers. 2003. Independent evolution
of migration on the South American landscape in a long-distance
temperate-tropical migratory bird, Swainson?s flycatcher (Myiar-
chus swainsoni). J. Biogeogr. 30:925?937.
Joseph, L., T. Wilke, E. Bermingham, D. Alpers, and R. Ricklefs.
2004. Towards a phylogenetic framework for the evolution of
shakes, rattles, and rolls in Myiarchus tyrant-flycatchers (Aves:
Passeriformes:Tyrannidae). Mol. Phylogenet. Evol. 31:139?152.
Kendall, P. G. 1949. Stochastic processes and population growth.
J. R. Statis. Soc. B 11:230?264.
Klicka, J., and R. M. Zink. 1997. The importance of recent ice ages
in speciation: a failed paradigm. Science 277:1666?1669.
Lanyon, S. M., and K. E. Omland. 1999. A molecular phylogeny
of the blackbirds (Icteridae): Five lineages revealed by cyto-
chrome-b sequence data. Auk 116:629?639.
Liu, Z. H., and T. D. Herbert. 2004. High-latitude influence on the
eastern equatorial Pacific climate in the early Pleistocene epoch.
Nature 427:720?723.
Lovette, I. J. 2004a. Mitochondrial dating and mixed-support for
the ??2% rule?? in birds. Auk 121:1?6.
???. 2004b. Molecular phylogeny and plumage signal evolution
854 JASON T. WEIR
in a trans Andean and circum Amazonian avian species complex.
Mol. Phylogenet. Evol. 32:512?523.
Lynch, M., and B. Walsh. 1998. Genetics and analysis of quanti-
tative traits. Sinauer, Sunderland, Ma.
Maddison, W. P., and D. R. Maddison. 2003. Mesquite: a modular
system for evolutionary analysis. Ver. 1.0. http://mesquitepo-
ject.org.
Mariaux, J., and M. J. Braun. 1996. A molecular phylogenetic sur-
vey of the nightjars and allies (Caprimulgiformes) with special
emphasis on the potoos (Nyctibiidae). Mol. Phylogenet. Evol.
6:228?244.
Marks, B. D., S. J. Hackett, and A. P. Capparella. 2002. Historical
relationships among Neotropical lowland forest areas of ende-
mism as determined by mitochondrial DNA sequence variation
within the wedge-billed woodcreeper (Aves:Dendrocolaptidae:
Glyphorynchus spirurus). Mol. Phylogenet. Evol. 24:153?167.
Meave, J., and M. Kellman. 1994. Maintenance of rain-forest di-
versity in riparian forests of tropical savannas: implications for
species conservation during Pleistocene drought. J. Biogeogr.
21:121?135.
Meave, J., M. Kellman, A. Macdougall, and J. Rosales. 1991. Ri-
parian habitats as tropical forest refugia. Glob. Ecol. Biogeogr.
Lett. 1:69?76.
Mengel, R. M. 1964. The probably history of species formation in
some northern wood warblers (Parulidae). Living Bird 3:9?43.
Moore, W. S. 1995. Inferring phylogenies from mtDNA variation:
mitochondrial gene trees vs. nuclear gene trees. Evolution 49:
718?726.
Moore, W. S., A. C. Weibel, and A. Agius. 2005. Mitochondrial
DNA phylogeny of the woodpecker genus Veniliornis (Picidae,
Picinae) and related genera implies convergent evolution of
plumage patterns. Biol. J. Linn. Soc. In press.
Moran, P. A. P. 1951. Estimation methods for evolutive processes.
J. R. Statis. Soc. B 13:141?146.
Moritz, C., J. L. Patton, C. J. Schneider, and T. B. Smith. 2000.
Diversification of rainforest faunas: An integrated molecular ap-
proach. Annu. Rev. Ecol. Syst. 31:533?563.
Nee, S. 2001. Inferring speciation rates from phylogenies. Evolution
55:661?668.
???. 2004. Extinct meets extant: simple models in paleontology
and molecular phylogenetics. Paleobiology 30:172?178.
Nee, S., E. C. Holmes, R. M. May, and P. H. Harvey. 1994. Ex-
tinction rates can be estimated from molecular phylogenies. Phi-
los. Trans. R. Soc. Lond. B 344:77?82.
Nei, M., and W. H. Li. 1979. Mathematical model for studying
genetic variation in terms of restriction endonucleases. Proc.
Natl. Acad. Sci. USA 76:5269?5273.
Nores, M. 1999. An alternative hypothesis for the origin of Ama-
zonian bird diversity. J. Biogeogr. 26:475?485.
???. 2004. The implications of Tertiary and Quaternary sea level
rise events for avian distribution patterns in the lowlands of
northern South America. Glob. Ecol. Biogeogr. 13:149?161.
Omland, K. E., S. M. Lanyon, and S. J. Fritz. 1999. A molecular
phylogeny of the new world orioles (Icterus): the importance of
dense taxon sampling. Mol. Phylogenet. Evol. 12:224?239.
Pagel, M. 1994. Detecting correlated evolution on phylogenies: a
general method for the comparative analysis of discrete char-
acters. Proc. R. Soc. Lond. B 255:37?45.
Penny, D. 2005. Evolutionary biology: relativity for molecular
clocks. Nature 436:183?184.
Pereira, S. L., and A. J. Baker. 2004. Vicariant speciation of cu-
rassows (Aves, Cracidae): a hypothesis based on mitochondrial
DNA phylogeny. Auk 121:682?694.
Prance, G. T. 1978. Origin and evolution of Amazon flora. Inter-
ciencia 3:207?222.
Price, J. J., and S. M. Lanyon. 2002. A robust phylogeny of the
oropendolas: polyphyly revealed by mitochondrial sequence
data. Auk 119:335?348.
???. 2004. Song and molecular data identify congruent but novel
affinities of the Green Oropendola (Psarocolius viridis). Auk
121:224?229.
Pybus, O. G., and P. H. Harvey. 2000. Testing macro-evolutionary
models using incomplete molecular phylogenies. Proc. R. Soc.
Lond. B 267:2267?2272.
R Development Core Team. 2005. R: a language and environment
for statistical computing. http://www.R-project.org.
Rambaut, A. 2002. Phyl-O-Gen v1. 1 Available at http://
evolve.zoo.ox.ac.uk/.
Rand, A. L. 1948. Glaciation, an isolating factor in speciation.
Evolution 2:314?321.
Rasanen, M. E., A. M. Linna, J. C. R. Santos, and F. R. Negri.
1995. Late Miocene tidal deposits in the Amazonian foreland
basin. Science 269:386?390.
Ravelo, A. C., D. H. Andreasen, M. Lyle, A. O. Lyle, and M. W.
Wara. 2004. Regional climate shifts caused by gradual global
cooling in the Pliocene epoch. Nature 429:263?267.
Rice, N. H., A. T. Peterson, and G. Escalona-Segura. 1999. Phy-
logenetic patterns in montane Troglodytes wrens. Condor 101:
446?451.
Richards, P. W. 1952. The tropical rain forest: an ecological study.
Cambridge Univ. Press, Cambridge, U.K.
Rosenzweig, M. L. 1995. Species diversity in space and time. Cam-
bridge Univ. Press, Cambridge, U.K.
Rossetti, D. D., P. M. de Toledo, and A. M. Goes. 2005. New
geological framework for western Amazonia (Brazil) and im-
plications for biogeography and evolution. Quat. Res. 63:78?89.
Roy, M. S., J. C. Torres-Mura, and F. Hertel. 1999. Molecular
phylogeny and evolutionary history of the tit-tyrants (Aves:Tyr-
annidae). Mol. Phylogenet. Evol. 11:67?76.
Rull, V. 2005. Biotic diversification in the Guyana highlands: a
proposal. J. Biogeogr. 32:921?927.
Russello, M. A., and G. Amato. 2004. A molecular phylogeny of
Amazona: implications for Neotropical parrot biogeography, tax-
onomy, and conservation. Mol. Phylogenet. Evol. 30:421?437.
Sanders, H. L. 1969. Benthic marine diversity and stability-time
hypothesis. Brookhaven Symp. Biol. 71?81.
Sanderson, M. J. 2003. r8s: inferring absolute rates of molecular
evolution and divergence times in the absence of a molecular
clock. Bioinformatics 19:301?302.
Schwabe, G. H. 1969. Towards an ecological characterization of
the South American continent. Pp. 113?136 in E. J. Fittkau et
al., ed. Biogeography and ecology in South America. Dr. W.
Junk, Dordrecht, The Netherlands.
Sheldon, F. H., L. A. Whittingham, R. G. Moyle, B. Slikas, and D.
W. Winkler. 2005. Phylogeny of swallows (Aves:Hirundinidae)
estimated from nuclear and mitochondrial DNA sequences. Mol.
Phylogenet. Evol. 35:254?270.
Sick, H. 1967. Rios e enchentes na Amazonia como obstaculo para
a avifauna. Pp. 495?520 in H. Lent, ed. Atlas do simposio sobre
a biota Amazonica. vol. 5 (Zoologia). Conselho Nacional de
Pesquisas, Rio de Janeiro.
Simpson, B. B., and J. Haffer. 1978. Speciation patterns in Ama-
zonian forest biota. Annu. Rev. Ecol. Syst. 9:497?518.
Steyermark, J. A., and G. C. K. Dunsterville. 1980. The lowland
floral element on the summit of Cerro-Guaiquinima and other
cerros of the Guayana Highland of Venezuela. J. Biogeogr. 7:
285?303.
Swofford, D. L. 2002. PAUP*4.0b10: phylogenetic analysis using
parsimony (*and other methods). Sinauer, Sunderland, MA.
van den Elzen, R., J. Guillen, V. Ruiz-del-Valle, L. M. Allende, E.
Lowy, J. Zamora, and A. Arnaiz-Villena. 2001. Both morpho-
logical and molecular characters support speciation of South
American siskins by sexual selection. Cell. Mol. Life Sci. 58:
2117?2128.
van der Hammen, T., and M. L. Absy. 1994. Amazonia during the
last glacial. Palaeogeogr. Palaeoclimatol. Palaeoecol. 109:
247?261.
van der Hammen, T., and H. Hooghiemstra. 1997. Chronostratig-
raphy and correlation of the Pliocene and Quaternary of Colom-
bia. Quat. Int. 40:81?91.
???. 2000. Neogene and Quaternary history of vegetation, cli-
mate, and plant diversity in Amazonia. Quat. Sci. Rev. 19:
725?742.
van?t Veer, R., and H. Hooghiemstra. 2000. Montane forest evo-
lution during the last 650,000 yr in Colombia: a multivariate
855SPECIES ACCUMULATION IN NEOTROPICAL BIRDS
approach based on pollen record Funza-I 35. J. Quat. Sci. 15:
329?346.
Vanzolini, P. E., and E. E. Williams. 1970. South American anoles:
the geographic differentiation and evolution of the Anolis chry-
solepis species group (Sauria: Iguanidae). Arq. Zool. Sao Paulo
19:1?298.
Vonhof, H. B., F. P. Wesselingh, R. J. G. Kaandorp, G. R. Davies,
J. E. van Hinte, J. Guerrero, M. Rasanen, L. Romero-Pittman,
and A. Ranzi. 2003. Paleogeography of Miocene Western Ama-
zonia: isotopic composition of molluscan shells constrains the
influence of marine incursions. Geol. Soc. Am. Bull. 115:
983?993.
Webb, S. D. 1995. Biological implications of the Middle Miocene
Amazon seaway. Science 269:361?362.
Weir, J. T., and D. Schluter. 2004. Ice sheets promote speciation
in boreal birds. Proc. R. Soc. Lond. B 271:1881?1887.
Whitlock, M. C. 2005. Combining probability from independent
tests: the weighted Z-method is superior to Fisher?s approach.
J. Evol. Biol. 18:1368?1373.
Whitmore, T. C., and G. T. Prance. 1987. Biogeography and Qua-
ternary history in tropical America. Clarendon Press, Oxford,
U.K.
Whittingham, L. A., B. Slikas, D. W. Winkler, and F. H. Sheldon.
2002. Phylogeny of the tree swallow genus, Tachycineta (Aves :
Hirundinidae), by Bayesian analysis of mitochondrial DNA se-
quences. Mol. Phylogenet. Evol. 22:430?441.
Willis, K. J., and K. J. Niklas. 2004. The role of Quaternary en-
vironmental change in plant macroevolution: the exception or
the rule? Philos. Trans. R. Soc. Lond. B 359:159?172.
Yule, G. U. 1924. A mathematical theory of evolution based on the
conclusions of Dr. J. C. Willis, FRS. Philos. Trans. R. Soc. Lond.
B 213:21?87.
Zink, R. M., and J. B. Slowinski. 1995. Evidence from molecular
systematics for decreased avian diversification in the Pleistocene
epoch 10. Proc. Natl. Acad. Sci. USA. 92:5832?5835.
Zink, R. M., J. Klicka, and B. R. Barber. 2004. The tempo of avian
diversification during the Quaternary. Philos. Trans. R. Soc.
Lond. B 359:215?219.
Corresponding Editor: J. Hey