Published version:
Molecular Phylogenetics and Evolution
Volume 53, Issue 3, December 2009, Pages 835-847
doi:10.1016/j.ympev.2009.08.001
Copyright © 2009 Elsevier Inc. All rights reserved.
1
Phylogenetic Relationships of Flowerpeckers (Aves: Dicaeidae): Novel Insights
2
into the Evolution of a Tropical Passerine Clade
3
4
5
Árpád S. Nyári1*, A. Townsend Peterson1, Nathan H. Rice2, and Robert G. Moyle1
6
7
8
9
10
1
The University of Kansas Natural History Museum and Biodiversity Institute, Dyche Hall, 1345
Jayhawk Blvd., Lawrence, KS 66045, USA.
11
12
2
Academy of Natural Sciences, 1900 Benjamin Franklin Parkway, Philadelphia, PA 19103, USA
1
email: arpi@ku.edu
13
14
15
16
17
18
19
20
Abstract:
21
Understanding the relationships and evolution of flowerpeckers has been
22
challenging, particularly as no phylogenetic study has as yet assessed the group. Here,
23
we present a first such analysis of this clade based on sequences of two mitochondrial
24
genes and one nuclear intron. Our analyses offer strong support for monophyly of the
25
Dicaeidae. Within the family, 4 Dicaeum species (D. chrysorrheum, D. melanoxanthum,
26
D. agile, and D. everetii) had closer affinity to Prionochilus, although tests of alternative
27
topologies could not reject reciprocal monophyly of the two genera. Across the family,
28
overall bill shape trends from more stout bills basally to more slender and medium bills,
29
whereas sexual dichromatism and plumage patterns show much more homoplasy.
30
Taxonomically, generic allocations may need to be changed to reflect historical
31
relationships better.
32
33
Keywords: Dicaeidae, Dicaeum, Prionochilus, flowerpeckers, molecular phylogeny,
34
monophyly, taxonomy, morphological evolution.
35
2
doi:10.1016/j.ympev.2009.08.001
36
1.
Introduction
37
Flowerpeckers (Aves, Dicaeidae) are a family of small passerine birds inhabiting
38
mature forests, forest edges, scrublands, as well as anthropomorphically modified
39
landscapes in some cases. They range from southern Asia south and east through the
40
Philippine Islands, Indonesia, and New Guinea and surrounding islands, with a single
41
species occurring across Australia. Flowerpeckers have seen considerable taxonomic
42
attention, with previous revisions based on bill morphology, tongue structure, plumage
43
characters, and natural history (Sharpe 1885, Mayr and Amadon 1947, Paynter 1967).
44
These studies encountered great difficulty characterizing the relationships among
45
flowerpeckers and establishing the limits of the clade, although it was clear that sunbirds
46
(Nectariniidae) were the sister lineage to flowerpeckers. DNA-DNA hybridization studies
47
(Sibley and Ahlquist 1983, 1990, Sibley and Monroe 1990, 1993) and more recent DNA
48
sequence-based studies (Barker et al. 2002, 2004, Ericsson and Johansson 2003,
49
Johansson et al. 2008) supported this sister relationship, but also concluded that several
50
genera (Pardalotus, Melanocharis, Rhamphocaris) previously thought closely related
51
(Mayr and Amadon 1947) are in fact only distantly related.
52
Flowerpeckers are currently divided into two genera (Prionochilus and Dicaeum),
53
with 44 recognized species (Cheke et al., 2001, Dickinson 2003). The generic
54
subdivision is based on a single morphological character, the length of the outermost
55
primary, which is elongated in Prionochilus but vestigial in Dicaeum; the only
56
documented exception is D. melanoxanthum (Mayr and Amadon 1947, Cheke et al.
57
2001), which also has an elongated outermost primary. A recent phylogenetic analysis of
58
characters derived from vocalizations of 36 species of flowerpeckers suggests a basal
59
placement of Prionochilus (Iddi et al. 2006). Most flowerpeckers are sexually
60
dichromatic, have stouter bills than sunbirds, and display a broad variety of tongue
3
doi:10.1016/j.ympev.2009.08.001
61
structures (Cheke et al. 2001). Their diet is a mix of small fruits and insects, but they
62
display a notable reliance on mistletoe fruits (Loranthaceae and Viscaceae). The current
63
understanding of the variation, evolution, and biogeography of the family is based on
64
different interpretations of these character sets (Mayr and Amadon 1947).
65
Molecular characters can greatly enrich phylogenetic analyses of such difficult
66
groups by providing abundant, independent, and objective characters on which to base
67
quantitative analyses. The resulting phylogenetic frameworks can provide insights not
68
only into relationships among members of the group of interest, but can also offer a
69
historical framework upon which to base hypotheses of character evolution and
70
evolutionary history (Harvey and Pagel 1991, Lanyon 1993). As such, we aim to
71
establish explicit hypotheses of phylogenetic relationships among flowerpeckers to
72
provide a basis for investigating the evolution of bill size and shape, tongue structure,
73
length of the outermost primary, sexual dimorphism, and plumage patterns. Specifically,
74
in this contribution, we explore 3 key issues: (1) monophyly of flowerpeckers, (2)
75
monophyly of the genera Dicaeum and Prionochilus, and (3) general biogeographic
76
patterns and character evolution within the Dicaeidae.
77
78
2.
Materials and Methods
79
2.1
Taxon sampling and molecular markers
80
Flowerpeckers present a particular challenge regarding taxon sampling for
81
molecular phylogenetic analysis, owing largely to the broad distribution of the family
82
across many countries, as well as to the rarity of some species, (considering the
83
conservation status of several Philippine species where 5 species are Near Threatened,
84
two Vulnerable, and one Critically Endangered; BirdLife International 2000). We base
85
this study on vouchered samples of flowerpecker species collected by us and others
4
doi:10.1016/j.ympev.2009.08.001
86
throughout the family’s range. Our ingroup sampling includes 30 species, covering most
87
key Dicaeum taxa in terms of phenotypic diversity (Table 1), and including all six
88
members of Prionochilus. To provide outgroup information for rooting and polarization of
89
character transformations, we used a broad selection of sunbird genera, as well as a
90
leafbird (Chloropsis hardwickei) as a next closest lineage (Barker et al. 2002).
91
Total genomic DNA was extracted from frozen or alcohol-preserved tissue
92
samples using standard Qiagen tissue extraction protocols (Qiagen, Valencia, CA). Our
93
molecular markers included established mitochondrial protein-coding genes and a
94
nuclear
95
dehydrogenase subunit 2 (ND2; 1034 bp), ND subunit 3 (ND3; 351 bp), and the fifth
96
nuclear intron of the transforming growth factor ß2 (TGFb2; 542 bp aligned) were
97
amplified using the primers L5215 – H6313 (Sorenson et al., 1999), L10755 – H11151
98
(Chesser 1999), and TGF5 and TGF6 (Primmer et al., 2002), respectively.
intron.
The
mitochondrial
genes
nicotinamide
adenine
dinucleotide
99
For lack of fresh tissue material, to represent two species (Table 1), we used a
100
small fragment of toepad from museum study skin specimens as a source of DNA. We
101
chose to include only these two species from ancient DNA sources for this study mainly
102
because of their distinct plumage patterns, but also because of the relative recency of
103
the museum specimens (early 1980s). For these samples, we employed rigorous
104
laboratory protocols intended to minimize contamination and potential amplification of
105
nuclear pseudogenes (Mundy et al. 1997). Owing to degradation of cellular DNA that is
106
characteristic of museum study skins (Cooper et al. 1996, Willerslev and Cooper 2005)
107
we designed specific primers that amplify smaller fragments of each marker for these
108
two taxa. Primer sequences are available upon request from the senior author.
109
All PCR amplifications were performed in 25 µl reactions using PureTaq™ RTG
110
PCR beads (GE Healthcare Bio-Sciences Corp.). Amplified double-stranded PCR
5
doi:10.1016/j.ympev.2009.08.001
111
products were cleaned with ExoSAP-IT™ (GE Healthcare Bio-Sciences Corp.), and
112
visualized on high-melt agarose gels stained with ethidium bromide. Purified PCR
113
products were cycle-sequenced with ABI Prism BigDye™ v3.1 terminator chemistry,
114
using the same primers as for each PCR reaction. Cycle-sequenced products were
115
further purified using Sephadex™ spin columns (GE Healthcare Bio-Sciences Corp.),
116
and finally sequenced on an ABI 3130 automated sequencer. Sequences of both strands
117
of each gene were examined and aligned in Sequencher 4.1 (GeneCodes Corp.), and a
118
final data matrix of contiguous sequences assembled using ClustalX 1.83 (Thompson et
119
al. 1997).
120
121
2.2.
Phylogenetic reconstruction and analyses
122
An incongruence length difference (ILD) test (Farris et al., 1994) was performed
123
through the program PAUP*4b10 (Swofford, 2000), by running 1000 heuristic replicates
124
of the partition homogeneity test, to test for potential incongruence in phylogenetic signal
125
between genes.
126
Phylogenetic reconstruction was performed via maximum likelihood (ML), as
127
implemented in the software PAUP*, with TBR branch-swapping and 100 random taxon-
128
addition replicates. ModelTest 3.7 (Posada and Crandall, 1998) was used to determine
129
the most appropriate model of sequence evolution via the Akaike Information Criterion
130
(AIC). Nodal support was assessed via nonparametric bootstrapping with 100 random-
131
addition replicates.
132
We also conducted Bayesian phylogenetic analyses (BA) using Markov Chain
133
Monte Carlo (MCMC) tree searches through the program MrBayes 3.1.2 (Ronquist and
134
Huelsenbeck 2003). The concatenated dataset was partitioned by gene and codon
135
positions for the nuclear intron and mitochondrial genes, respectively. ModelTest 3.7
6
doi:10.1016/j.ympev.2009.08.001
136
(Posada and Crandall, 1998) was again used to establish the best-suited substitution
137
model according to the AIC (Table 2). Two independent runs of 107 generations were
138
conducted using the respective models of sequence evolution, with default chain heating
139
conditions, sampling every 100 generations. Evaluation of stationarity was conducted by
140
plotting posterior probabilities from the two runs in the program Tracer (Rambaut and
141
Drummond 2007). Topologies sampled from the first 25% of generations were discarded
142
as an initial “burn-in”, so a total of 75,000 trees were summarized to produce a single
143
50% majority-rule consensus tree.
144
To compare current taxonomic treatments at the generic level, tests of
145
topological constraints on the dataset were conducted by enforcing reciprocal
146
monophyly between the Dicaeum and Prionochilus clades. Evaluations of the constraint
147
against the ML tree was carried out in PAUP* with the Shimodaira-Hasegawa (SH) test,
148
using RELL approximations (Shimodaira and Hasegawa 1999).
149
We explored evolutionary patterns in some important morphological characters,
150
which we scored through examination of specimens and, where necessary for lack of
151
specimen material, interpretation of illustrations in field guides (Cheke et al., 2001,
152
Robson et al. 2005). The following phenotypic character sets were scored in discrete
153
states: general bill shape – (A) slender and longer bills, (B) medium length and
154
thickness, or (C) short and stout bills; marked sexual dichromatism – (A) present or (B)
155
absent; presence of carotenoids or other pigments (indicated by yellow, orange, or red
156
color patches) – (A) present or (B) drab and lacking carotenoids; and presence of
157
marked streaked patterns on breast or flanks independent of pigmentation – (A) present
158
or (B) absent.
159
160
3.
Results
7
doi:10.1016/j.ympev.2009.08.001
161
3.1.
Sequence data characteristics
162
High-quality sequences of the mitochondrial genes and the nuclear intron were
163
obtained for all modern DNA samples. Sequences were also obtained from the ancient
164
DNA samples of D. agile and D. everetti for read lengths ≤300 bp. These short
165
fragments were then concatenated to assemble the TGFb2, ND2, and ND3 genes in
166
their entirety through amplification and sequencing of 3, 4, and 2 fragments respectively.
167
Of the 4 ND2 fragments, we could not resolve clean sequence for a 200 bp region in the
168
third quarter of the gene, so this region was coded as missing for all phylogenetic
169
analyses.
170
Sequence alignments for all taxa and genes were straightforward. The
171
mitochondrial data showed no insertions, deletions, or anomalous stop-codon
172
placements, and base composition was typical of this marker (Table 2), suggesting true
173
mitochondrial origin, as opposed to corresponding to nuclear pseudogenes (Sorenson
174
and Quinn, 1998). Gaps and insertions were noted in the nuclear sequences; although
175
these were not coded separately in our analyses, we do make note of several
176
synapomorphic features in the discussions that follow.
177
The complete molecular dataset thus comprised 41 taxa (of which 30 were
178
ingroup taxa) and 1925 aligned bases (Table 2). Average pairwise distance (uncorrected
179
P) between the ingroup and members of the Nectariniidae derived from the entire
180
dataset was 15%, while within flowerpeckers average pairwise distance ranged from 2%
181
between D. concolor and D. pygmaeum, to 14% between D. agile and D. aeneum.
182
183
3. 2.
Phylogenetic analyses
184
The partition homogenetiy test revealed no significant difference in phylogenetic
185
signal (P = 0.84) between the three genes, so all subsequent analyses were performed
8
doi:10.1016/j.ympev.2009.08.001
186
on the combined dataset. We nevertheless carried out separate tree searches based on
187
each individual gene data set (results not shown), and found no strongly supported
188
conflicts between the topologies inferred.
189
ML analyses produced a single topology (likelihood score of -lnL = 21345.47240),
190
which was broadly congruent with the consensus tree inferred via BA (Fig. 1). Ingroup
191
monophyly was strongly supported under both ML and BA. Two main clades were
192
recovered within flowerpeckers: one (Clade A) containing all members of Prionochilus
193
plus Dicaeum chrysorrheum, D. melanoxanthum, D. agile, and D. everetti, hereafter
194
referred to as the four “odd” Dicaeum; a second (Clade B) included the remaining
195
Dicaeum species. Nodal support contrasted between these two main subdivisions:
196
Clade A had only 59% bootstrap support and 0.89 posterior probability under ML and
197
BA, respectively, while Clade B was unambiguously and strongly supported in both
198
analyses (node 5, Fig.1).
199
Phylogenetic patterns within Clade A (Fig.1) indicated paraphyly of Dicaeum as
200
presently conceived, albeit with only modest support (node 2, Fig.1). ML analyses, but
201
not BA, supported monophyly of Prionochilus.
202
Prionochilus olivaceus was especially difficult to resolve, as neither reconstruction
203
method obtained significant support for affinities with either the remainder of
204
Prionochilus or the four odd Dicaeum. On the other hand, strong support was obtained
205
for groupings among the four odd Dicaeum species, in which D. chrysorrheum was
206
placed as sister to D. melanoxanthum, D. agile, and D. everetti (node 3, Fig. 1). The rest
207
of Prionochilus (minus P. olivaceus) was also strongly supported as a monophyletic
208
group (node 4, Fig. 1), within which P. maculatus branched basally, followed by P.
209
thoracicus, P. xanthopygius, and then a sister-pairing of P. plateni and P. percussus.
9
In both analyses, the placement of
doi:10.1016/j.ympev.2009.08.001
210
Within Clade B (node 5, Fig. 1), Dicaeum anthonyi and D. bicolor were well-
211
supported as sister taxa and branched basally, albeit with low support. Relationships
212
recovered within the remainder of this clade were mixed, with many short internodes
213
characterized by relatively weak support. Strong support although was recovered for the
214
group belonging to node 6 (Fig. 1), and for nodes linking D. pygmaeum and D. concolor,
215
and D. hirundinaceum and D. geelvinkianum.
216
Only a singe alternative topology was available for testing against the ML tree,
217
and because it recovered a monophyletic Prionochilus, only the monophyly of Dicaeum
218
was therefore tested. Constraining monophyly of Dicaeum during an ML search yielded
219
a topology with both genera reciprocally monophyletic (-lnL = 21349.20942).
220
Shimodaira-Hasegawa test returned a non-significant result for this topology (p = 0.248),
221
indicating that our data could not reject monophyly of both genera.
The
222
Additional features of our molecular data provided some important insights,
223
through informative insertions and deletions in TGFb2. The taxonomic utility of insertions
224
and deletions in nuclear introns has been acknowledged by previous studies (Johnson et
225
al. 2001, Moyle 2004, Moyle and Marks 2006), so here we comment on several that
226
provide a qualitative assessment of nodal support. A 10 bp deletion in position 327-336
227
and a 2 bp deletion in position 417-418 were common to all taxa in Clade B (Fig. 1). A 1
228
bp deletion (position 251) was shared by the core Prionochilus clade (i.e., excluding P.
229
olivaceus). Dicaeum sanguinoletum, D. pygmaeum, and D. concolor shared a 6 bp
230
deletion (postion 402-407). In the outgroup, a 3 bp deletion (postition 194-196) was
231
shared by the sunbird and spiderhunter clade with the exception of Arachnothera
232
magna, and a 4 bp deletion (positions 359-362) united all Aethopyga species included in
233
this study.
234
10
doi:10.1016/j.ympev.2009.08.001
235
4.
Discussion
236
4.1.
Flowerpecker taxonomy
237
This paper presents a first quantitative evaluation of phylogenetic relationships
238
within the flowerpeckers. Analyses of the three-gene dataset recovered a monophyletic
239
Dicaeidae, with two main subclades (node 1, Fig. 1). This topology is congruent with
240
current taxonomic opinion that divides the family into two genera (Paynter 1967, Cheke
241
et al. 2001, Dickinson 2003), but with some notable differences concerning membership
242
of the two genera.
243
Although it has long been recognized that flowerpeckers contained two discrete
244
lineages (Mayr and Amadon 1947, Cheke et al. 2001), reliable character evidence
245
supporting this division has been more elusive. The morphological character that
246
presently unites Prionochilus, an overall longer outermost primary, is shared by D.
247
melanoxanthum, which confounded earlier treatments (Sharpe 1885, Mayr and Amadon
248
1947, Paynter 1967). Our molecular dataset supports the phylogenetic affinity of D.
249
melanoxanthum with Prionochilus, although three other Dicaeum are also grouped with
250
D. melanoxanthum (node 3, Fig. 1) that do not possess an elongated outermost primary
251
feather (Mayr and Amadon 1947). With our present dataset, we cannot rule out
252
statistically the present two-clade arrangement as an appropriate treatment: specifically,
253
our topological test did not have the statistical power to reject the constrained hypothesis
254
of reciprocal monophyly of the genera. Two deletions in the nuclear intron (Fig. 1)
255
support the distinctiveness of the core set of Dicaeum species as a well-supported
256
clade. This polyphyletic Dicaeum arrangement parallels doubts previously voiced based
257
on morphological evidence (Mayr and Amadon 1947, Cheke et al. 2001). In addition to
258
exceptions to the pattern of outermost primary length mentioned above, other characters
11
doi:10.1016/j.ympev.2009.08.001
259
(e.g., bill structure and tongue anatomy) also disagreed with the present binary
260
Dicaeum-Prionochilus split (Cheke et al. 2001, Morioka 1992).
261
Given our current phylogenetic hypothesis, three options are available for
262
integrating these results into formal taxonomy (particularly if supported by further
263
molecular and detailed morphological evidence): (1) recognize only a single, very
264
inclusive genus (Dicaeum would have formal nomenclatural priority in this case); (2)
265
submerge the 4 odd Dicaeum taxa into a more inclusive Prionochilus; or (3) erect a third
266
genus within the family to refer to the 4 odd Dicaeum, in which case the name
267
Pachyglossa Blyth 1843 would apply. Based on our results, we suggest that the
268
taxonomy should be amended to submerge the 4 odd Dicaeum species under the genus
269
Prionochilus. Because of rather low statistical support for this arrangement, perhaps the
270
most prudent arrangement is to retain the current taxonomy of this family.
271
Phylogenetic patterns within the core Dicaeum (Clade B) are better delineated,
272
albeit not without uncertainties (Fig. 1). Weak support exists for the hypothesis that the
273
clade D. anthonyi + D. bicolor forms the basal lineage within this clade, followed by D.
274
aureolimbatum very weakly supported as sister to a strongly associated D. trigonostigma
275
+ D. australe. Further up the tree, the Mindanao endemic D. nigrilore was inferred to be
276
sister to the more widespread Philippine endemic D. hypoleucum, with moderate nodal
277
support. Subsequent speciation events were not well resolved with our dataset, allowing
278
only limited inference regarding their exact relationships (node 6, Fig. 1). For example,
279
D. concolor + D. pygmaeum is a well-supported clade, but BA placed this clade basally
280
and ML placed them more distal in the tree. The Bornean endemic D. monticolum
281
received moderate support as sister to the more widespread D. ignipectum. Five
282
members of the Australo-Papuan radiation of Dicaeum contain a strongly supported D.
283
geelvinkianum sister to D. hirundinaceum, while the position of the morphologically
12
doi:10.1016/j.ympev.2009.08.001
284
distinctive D. tristrami within this group was ambiguous, receiving low support relative to
285
D. eximium and D. aeneum.
286
287
4. 2.
Character evolution
288
Incomplete taxon sampling and equivocal support for many relationships in the
289
phylogeny preclude formal analyses of character state evolution, but it is nonetheless
290
instructive to examine the distribution of certain character types across the phylogeny
291
(Fig. 2). Bill shape shows some degree of concordance with phylogenetic pattern in the
292
Dicaeidae, with stout-billed species at the base of the tree, and slender- and medium-
293
billed species being more derived. The transition from stout bill to more slender bills
294
occurs within Clade B: the stout-billed species D. anthonyi and D. bicolor are placed as
295
sister to the remaining (slender- and medium-billed) species. However, two stouter-billed
296
species, D. tristrami and D. eximium, are mixed in with more slender-billed species in
297
this clade, pointing towards the possibility of reversals in this trend.
298
In contrast, sexual dichromatism and overall plumage patterns do not seem to
299
correspond well with phylogeny (Fig. 2). Although all streaked species are in Clade A (P.
300
olivaceus, P. maculatus, D. chrysorrheum, D. agile, D. everetti), they do not form a
301
natural group. Indeed, more generally, the more colorful, carotenoid-bearing species are
302
intermixed with species with drab-colored plumages. The suite of taxa sister to D.
303
concolor and D. pygmaeum all present some form of scarlet-red color patch in their
304
plumage, the only exception being the rather drab D. tristrami. This relationship is
305
weakly resolved by our phylogenetic reconstructions, where resolution at the base of this
306
clade collapses the nodes supporting the more basal placement of the drab colored D.
307
concolor and D. pygmaeum. Plumage characters did unite some closely related taxa; for
308
example, our analyses consistently placed two taxa as sister species within this scarlet-
13
doi:10.1016/j.ympev.2009.08.001
309
red colored clade – D. cruentatum and D. igniferum, with both species sharing more
310
extensive coloration dorsally, extending beyond their nape, and onto the mantle (Fig. 2).
311
A character that has not received detailed attention beyond a purely descriptive
312
designation appears to be iris coloration. The majority of members of our clade A have
313
either red or orange-brownish irides, whereas members of clade B have more distinctly
314
brownish irides. Of course, one member of the latter clade defies the trend: males of the
315
only Australian flowerpecker, D. hirundinaceum, have distinct red irides. The anatomical
316
and structural underpinnings of iris color have not yet been investigated, and the
317
evolutionary significance of these differences will have to await further detailed
318
investigations.
319
Given that our molecular phylogenetic analysis lacks several extant members of
320
the family, we cannot elaborate further on the conclusiveness of evolutionary pathways
321
of these characters. Clarifications of these uncertainties will have to await molecular data
322
and a more robust phylogeny for all extant species, coupled with in-depth analysis of a
323
broader suite of morphological characters.
324
325
4. 3.
Biogeographic patterns
326
Two main biogeographic radiations can be recognized within flowerpeckers: one
327
tied to the Philippine Islands and the other, more recent, around New Guinea and its
328
surrounding islands including mainland Australia (Fig. 2). Without a doubt, the
329
morphologically most variable flowerpecker species, D. trigonostigma, with 17
330
recognized subspecies (Cheke et al. 2001), occurs over a broader area, from mainland
331
India east to the Philippines, but not crossing Wallace’s Line. Other wide-ranging
332
species include D. agile, D. chrysorrheum, D. concolor, and D. ignipectum. The
14
doi:10.1016/j.ympev.2009.08.001
333
Philippines also host the most members of the family, with 12 species endemic to the
334
archipelago (Cheke et al. 2001, Kennedy et al. 2000).
335
Wallace’s Line has played an influential role in shaping present distributions of
336
several clades of birds (Clode and O’Brian 2001, Schulte et al. 2003). Our phylogenetic
337
framework indicates that almost all members of Clade A (Fig. 2) presently occur in the
338
Indo-Malayan region, without crossing Wallace’s Line. The single exception is D. agile,
339
with populations on Sumba, Flores, Alor, and Timor (Cheke et al. 2001). The Australo-
340
Malayan members of the family appear to have diverged more recently and more
341
rapidly, as indicated by short internodes linking the species in clade 6 (Fig.1). D.
342
tristrami, from Makira Island, is one of the most interesting and distinctive members of
343
this clade: although part of the core group of sexually dichromatic species with some
344
scarlet-red coloration in the plumage, this species has dull, brownish plumage, and no
345
sexual dichromatism. The precise placement of this species proved difficult with our
346
current dataset, but it is nevertheless closely associated with clades that recently
347
dispersed into Papua-New Guinea, adjacent islands, and Australia.
348
Smith and Fillardi (2007) surveyed a suite of species distributed throughout the
349
Solomon Islands, including D. aeneum. Based on mitochondrial markers, they reported
350
significant genetic structuring among island groups, with a maximum uncorrected
351
pairwise distance of 7% within D. aeneum. That study represents a first look into
352
flowerpecker molecular phylogeography, providing important insights into historical
353
biogeographic patterns and processes in the archipelago, setting the stage for additional
354
exploratory phylogeographic research of other flowerpecker species.
355
356
Conclusions
15
doi:10.1016/j.ympev.2009.08.001
357
Our taxon sampling contained about 70% of recognized flowerpecker species,
358
and it nevertheless proved sufficient to elucidate major patterns of relationships and
359
character evolution within the family. Our analyses provide a much-needed molecular-
360
based framework for reevaluating the generic taxonomy within this family, pointing out
361
that four Dicaeum species are actually more closely related to Prionochilus. Statistical
362
support for the generic dichotomy within flowerpeckers received only weak confirmation,
363
pointing towards the need of including additional molecular markers as well as
364
complementing the taxon sampling throughout the family. The diverse morphological
365
characters presently providing the basis for taxonomic and evolutionary inferences
366
concerning flowerpeckers thus remain a challenging body of evidence for future
367
research.
368
16
doi:10.1016/j.ympev.2009.08.001
369
References
370
BirdLife International. 2000. Threatened birds of the world. Barcelona and Cambridge,
371
UK, Lynx Edicions and BirdLife International.
372
Barker, F.K, Barrowclough, G.F., Groth, J.G. 2002. A phylogenetic hypothesis for
373
passerine birds: taxonomic implications of an analysis of nuclear DNA sequence
374
data. Proc. R. Soc. Lond. B. 269, 295-308.
375
Barker, F.K., Cibois, A., Schikler, P., Feinstein, J., Cracraft, J. 2004. Phylogeny and
376
diversification of the largest avian radiation. Proc. Nat. Acad. Sci. 101, 11040-
377
11045.
378
379
380
381
Cheke, R.A., Mann, C. F., Allen, R. 2001. Sunbirds. A guide to the Sunbirds,
Flowerpeckers and Sugarbirds of the World. Yale University Press.
Chesser, R. T. 1999. Molecular systematics of the rhinocryptid genus Pteroptochos. /
Condor 101, 439 - 446.
382
Clode, D., O’Brien, R. 2001. Why Wallace drew the line: A re-analysis of Wallace’s bird
383
collections in the Malay Archipelago and the origins of biogeography. In Faunal
384
and Floral Migrations and Evolution in SE Asia-Australasia, Ed. I. Metcalfe, CRC
385
Press.
386
387
388
389
Cooper, A. J., Rhymer, J., James, H., Olsen, S., McIntosh, C., Sorenson, M., Fleischer,
R. 1996. Ancient DNA and island endemics. Nature 381, 484.
Dickinson, E.C. (ed.). 2003. The Howard and Moore Complete Checklist of the Birds of
the World. 3rd Edition. Princeton University Press, Princeton, N.J.
390
Ericsson, P. G. P., Johansson, U. S. 2003. Phylogeny of Passerida (Aves:
391
Passeriformes) based on nuclear and mitochondrial sequence data. Mol.
392
Phylogenet. Evol. 29, 126-138.
393
Harvey, P. H., Pagel, M. D. 1991. The comparative method in evolutionary biology.
17
doi:10.1016/j.ympev.2009.08.001
394
395
396
Oxford University Press, New York.
Farris, J.S., Kallersjo, M., Kluge, A.G., Bult, C., 1994. Testing significance of
incongruence. Cladistics 10, 315–319.
397
Iddi, K. N., Cheke, R. A., Kenyon, L. 2006. Phylogenies of the sunbirds (Nectariniidae)
398
and flowerpeckers (Dicaeidae) based on analyses of vocalizations. Journal of
399
Ornithology 147 Suppl.1, 1-297.
400
Johansson, U. S., Fjeldså, J., Bowie, R. C. K. 2008. Phylogenetic relationships within
401
Passerida (Aves: Passeriformes): a review and a new molecular phlogeny based
402
on three nuclear intron markers. Mol. Phylogenet. Evol. 48, 858-876.
403
Johnson, K.P., de Kort, S., Dinwoodey, K., Mateman, A.C., ten Cate, C., Lessells, C.M.,
404
Clayton, D.H., 2001. A molecular phylogeny of the dove genera Streptopelia and
405
Columba. Auk 118, 874–887.
406
407
408
409
Kennedy, R. S., Gonzalez, P. C., Dickinson, E. C,, Miranda, H. C. Jr, Fisher, T. H. 2000.
A guide to the birds of the Philippines. Oxford: Oxford University Press.
Lanyon, S. M. 1993. Phylogenetic frameworks--towards a firmer foundation for the
comparative approach. Biol. Journ. Linn. Soc. 49, 45-61.
410
Mayr, E., Amadon, D. 1947. A review of the Dicaeidae. Amer. Mus. Novit. 1360.
411
Morioka, H. 1992. Tongues of two species of Prionochilus from the Philippines, with
412
notes on feeding habits of flowerpeckers (Dicaeidae). Japanese J. Orn. 40, 85-
413
92.
414
415
Moyle, R. G. 2004. Phylogenetics of barbets (Aves: Piciformes) based on nuclear and
mitochondrial DNA sequence data. Mol. Phylogenet. Evol. 30, 187–200.
416
Moyle, R. G., Marks, B. 2006. Phylogenetic relationships of the bulbuls (Aves:
417
Pycnonotiodae) based on mitochondrial and nuclear DNA sequence data. Mol.
418
Phylogenet. Evol. 40, 687-695.
18
doi:10.1016/j.ympev.2009.08.001
419
Mundy, N. I., Unitt, P., Woodruff, D. S. 1997. Skin from feet of museum specimens as a
420
non-destructive source of DNA for avian genotyping. Auk 114, 126 - 129.
421
Paynter, R. A. (ed). 1967. Check-list of birds of the world: a continuation of the work of
422
James L. Peters, vol. 12. Museum of Comparative Zoology, Cambridge,
423
Massachusetts, USA.
424
425
Posada, D., Crandall, K.A., 1998. Modeltest: testing the model of DNA substitution.
Bioinformatics 14, 817–818.
426
Primmer, C. R., T. Borge, J. Lindell, Saetre, G. P. 2002. Single-nucleotide polymorphism
427
characterization in species with limited available sequence information: high
428
nucleotide diversity revealed in the avian genome. Mol. Ecol. 11, 603–612.
429
430
Rambaut, A., Drummond, A. J. 2007. Tracer v1.4. Available from: <http://
beast.bio.ed.ac.uk/Tracer>
431
Robson, C. 2005. Birds of Southeast Asia. Princeton University Press, NJ, USA.0
432
Ronquist, F., Huelsenbeck, J.P., 2003. MrBayes3: Bayesian phylogenetic inference
433
434
435
436
437
438
439
440
441
442
443
under mixed models. Bioinformatics 19, 1572–1574.
Sibley, C.G., Ahlquist, J.E. 1983. The phylogeny and classification of birds based on the
data of DNA-DNA hybridization. Current Orn. 1, 245-292.
Sibley, C.G., Ahlquist, J.E. 1990. Phylogeny and classification of the birds of the world.
Yale University Press, New Haven.
Sibley, C.G., Monroe, B.L. 1990. Distribution and Taxonomy of birds of the world. Yale
University Press, New Haven.
Sibley, C.G., Monroe, B.L. 1993. A supplement to distribution and taxonomy of birds of
the world. Yale University Press, New Haven.
Smith, C. E., Filardi, C. E. 2007. Patterns of molecular and morphological variation in
some Solomon Island land birds. Auk 124, 479-493.
19
doi:10.1016/j.ympev.2009.08.001
444
445
Sorenson M. D, Quinn T. W. 1998. Numts: a challenge for avian systematics and
population biology. Auk 115, 214–221.
446
Sorenson, M.D., J.C. Ast, D.E. Dimcheff, T. Yuri, D. P. Mindell. 1999. Primers for a PCR-
447
based approach to mitochondrial genome sequencing in birds and other
448
vertebrates. Mol. Phylogenet. Evol. 12, 105-114.
449
450
Swofford, D. S. 2002. Phylogenetic Analysis Using Parsimony (*and other methods),
Version 4. Sinauer, Sunderland, MA.
451
Shimodaira, H., Hasegawa, M. 1999. Multiple comparisons of log-likelihoods with
452
applications to phylogenetic inference, Molecular Biology and Evolution 16, 1114-
453
1116.
454
455
Sharpe, R. Bowdler. 1885. Catalogue of the birds of the British Museum. London, vol.
10, XIII + 682 pp.
456
Schulte II, J. A., Melville, J., Larson, A. 2003. Molecular phylogenetic evidence for
457
ancient divergence of lizard taxa on either side of Wallace’s Line. Proc. R. Soc.
458
Lond. B. 270, 597-603.
459
Thompson, J. D., Gibson, T. J., Plewniak, F., Jeanmougin, F., Higgins, D. G. 1997. The
460
CLUSTAL_X windows interface: flexible strategies for multiple sequence
461
aligment aided by quality analysis tools. Nucleic Acids Research 25: 4876–4882.
462
Willerslev, E., Cooper, A. 2005. Ancient DNA. Proc. R. Soc. Lond. B. 272, 3-16.
20
doi:10.1016/j.ympev.2009.08.001
Fig. 1.
Maximum likelihood (ML, left) and Bayesian analysis (BA, right) views of phylogenetic patterns implied by analyses of
the complete molecular dataset. Support values are indicated by percent bootstrap (ML, left) and posterior probability values (BA,
right) above or at right of each node. Values <50 recovered by each method are not indicated at nodes. Synapomorphic deletions in
the TGFb2 gene for the ingroup are indicated as black circles on branches, and are detailed further in the text. Clade letters and
node numbers are referenced throughout the main text.
21
doi:10.1016/j.ympev.2009.08.001
22
doi:10.1016/j.ympev.2009.08.001
Fig. 2.
Morphological characters coded as discrete states linked to each terminal ingroup taxon as recovered by a strict
consensus tree of the BA and ML analysis: overall bill shape – slender [clear], medium length and thickness [grey], shorter and stout
bill [black]; sexual dichromatism – absent [white] or present [black]; plumage carotenoids – drab and lacking carotenoids [white],
presence of carotenoids or other pigments [black]; streaked patterns on breast or flanks – absent [white] or present [black]. The
fourth box column indicates species distribution relative to Wallace’s line – within Indo-Malayan region [white] or outside the region
[black]. A distributional extent for each species is given in the rightmost panel.
23
doi:10.1016/j.ympev.2009.08.001
24
doi:10.1016/j.ympev.2009.08.001
Table 1.
Taxonomic sampling, voucher sources, and GenBank accession numbers included in this study. Species with an
asterisk indicate museum specimens from which ancient DNA has been extracted from toe pads.
Taxon
Voucher source1
Sample #
Locality
Ingroup
Dicaeum aeneum
Dicaeum anthonyi
Dicaeum agile *
Dicaeum aureolimbatum
Dicaeum australe
Dicaeum bicolor
Dicaeum celebicum
Dicaeum chrysorrheum
Dicaeum concolor
Dicaeum cruentatum
Dicaeum everetti *
Dicaeum eximium
Dicaeum geelvinkianum
Dicaeum hirundinaceum
Dicaeum hypoleucum
Dicaeum igniferum
Dicaeum ignipectum
Dicaeum melanoxanthum
Dicaeum monticolum
Dicaeum nigrilore
Dicaeum pygmaeum
Dicaeum sanguinoletum
Dicaeum trigonostigma
Dicaeum tristrami
Prionochilus maculatus
Prionochilus olivaceus
Prionochilus plateni
Prionochilus percussus
Prionochilus xanthopygius
Prionochilus thoracicus
AMNH
FMNH
WFVZ
AMNH
ZMUC
FMNH
AMNH
USNM
KUNHM
UWBM
WFVZ
KUNHM
KUNHM
KUNHM
KUNHM
WAM
ZMUC
AMNH
KUNHM
FMNH
KUNHM
WAM
ANSP
KUNHM
KUNHM
FMNH
KUNHM
UWBM
KUNHM
LSUMNS
DOT 6657
429296
41288
DOT 12599
130804
357606
12585
B06156
10365
83538
41290
5312
5143
8821
18070
22677
116133
DOT 5717
17745
357592
10959
22871
1194
12802
12405
357587
12606
67503
12333
47198
Outgroup
Aethopyga duyvenbodei
Aethopyga boltoni
Aehopyga siparaja
Aethopyga primigenius
Aethopyga christinae
ZMUC
FMNH
FMNH
FMNH
KUNHM
123924
357631
358592
357624
10276
25
TGFb2
ND2
ND3
Solomon Islands, Isabel Island
Philippines, Luzon, Kalinga
Malaysia, Sabah, Tawau
Indonesia, Sulawesi, Poso
Philippines, Luzon
Philippines, Mindanao, Bukidnon
Indonesia, Sulawesi, Poso
Myanmar, Sagaing
China, Guangxi, Shinwandashan Nature Preserve
Singapore
Malaysia, Sabah, Kota Kinabalu
New Guinea, East New Britain, Tientop Village
New Guinea, Simbu, 10 km SSE Haia
Australia, WA, 40 km W Mumbinia Station
Philippines, Luzon, Mt. Labo
Indonesia, Flores, Kelimutu
Philippine Islands, Luzon, Isabela
Nepal, Malde
Malaysia, Sabah, 58 km SSE Sipitang
Philippines, Mindanao, Bukidnon
Philippines, Cagayan, Longog
Indonesia, Sumba, Waeelonda
Malaysia, Sabah, Mendolong
Solomon Islands, Makira, 17.5 km S Kira Kira
Malaysia, Sarawak, 25 km S Bintulu
Philippines, Mindanao
Philippines, Palawan, Puerto Princesa
Indonesia, Lingkung, Kayutanam
Malaysia, Sarawak, 25 km S Bintulu
Malaysia, Sabah, 8 km W Beaufort
GQ145327
GQ145325
GQ145348
GQ145330
GQ145347
GQ145326
GQ145331
GQ145319
GQ145339
GQ145320
GQ145349
GQ145345
GQ145321
GQ145309
GQ145328
GQ145329
GQ145346
GQ145312
GQ145341
GQ145338
GQ145344
GQ145323
GQ145308
GQ145342
GQ145317
GQ145324
GQ145335
GQ145311
GQ145322
GQ145314
GQ145285
GQ145283
GQ145306
GQ145288
GQ145305
GQ145284
GQ145289
GQ145277
GQ145297
GQ145278
GQ145307
GQ145303
GQ145279
GQ145267
GQ145286
GQ145287
GQ145304
GQ145270
GQ145299
GQ145296
GQ145302
GQ145281
GQ145266
GQ145300
GQ145275
GQ145282
GQ145293
GQ145269
GQ145280
GQ145272
GQ145243
GQ145241
GQ145264
GQ145246
GQ145263
GQ145242
GQ145247
GQ145235
GQ145255
GQ145236
GQ145265
GQ145261
GQ145237
GQ145225
GQ145244
GQ145245
GQ145262
GQ145228
GQ145257
GQ145254
GQ145260
GQ145239
GQ145224
GQ145258
GQ145233
GQ145240
GQ145251
GQ145227
GQ145238
GQ145230
Indonesia, Gunung Sahengbalira
Philippines, Mindanao, Bukidnon
Philippines, Sibuyan, Lambingan Falls
Philippines, Mindanao, Bukidnon
China, Guangxi, Shinwandashan Nature Preserve
GQ145337
GQ145343
GQ145336
GQ145334
GQ145332
GQ145295
GQ145301
GQ145294
GQ145292
GQ145290
GQ145253
GQ145259
GQ145252
GQ145250
GQ145248
doi:10.1016/j.ympev.2009.08.001
Anthreptes rectirostris
Arachnothera magna
Arachnothera longirostra
Cinnyris superbus
Cyanomitra obscura
Nectarinia jugularis
Chloropsis hardwickei
KUNHM
KUNHM
KUNHM
KUNHM
KUNHM
FMNH
KUNHM
8499
10401
12706
8678
8563
358566
10019
Equatorial Guinea, Centro Sur, Monte Alen National Park
China, Guangxi, Shinwandashan Nature Preserve
Philippines, Palawan, Puerto Princesa
Equatorial Guinea, Centro Sur, Monte Alen National Park
Equatorial Guinea, Centro Sur, Monte Alen National Park
Philippine Islands, Sibuyan
China, Guangxi, Diding Headwater Nature Preserve
GQ145315
GQ145316
GQ145310
GQ145333
GQ145340
GQ145313
GQ145318
GQ145273
GQ145274
GQ145268
GQ145291
GQ145298
GQ145271
GQ145276
GQ145231
GQ145232
GQ145226
GQ145249
GQ145256
GQ145229
GQ145234
Institutional abbreviations are as follows: Academy of Natural Sciences, Philadelphia (ANSP), American Museum of Natural History
(AMNH), Field Museum of Natural History (FMNH), The University of Kansas Natural History Museum (KUNHM), Louisiana State
University Museum of Natural Science (LSUMNS), National Museum of Natural History, Smithsonian Institution (USNM), Western
Australian Museum (WAM), Western Foundation of Vertebrate Zoology (WFVZ), University of Washington Burke Museum (UWBM),
Zoological Museum, University of Copenhagen (ZMUC).
26
doi:10.1016/j.ympev.2009.08.001
Table 2.
Characteristics and parameter estimates of the 3 genes and the mitochondrial codon partitions included in the
analyses.
Length
Number of variable sites
Parsimony informative sites
Model
Freq. A
Freq. C
Freq. G
Freq. T
r [A-C]
r [A-G]
r [A-T]
r [C-G]
r [C-T]
r [G-T]

Ts/Tv
TGFb2
542bp (aligned)
162
58
TVM+G
0.2443
0.2288
0.2308
0.2961
0.6469
2.9891
0.5544
1.2623
2.9891
1.0000
1.2194
1.7759
ND2
1032bp
581
495
GTR+I+G
0.3541
0.3870
0.0716
0.1873
0.1870
7.3444
0.1799
0.0839
3.3325
1.0000
0.7708
7.4509
27
ND3
351bp
180
151
TrN+I+G
0.3419
0.4012
0.0713
0.1856
1.0000
19.7322
1.0000
1.0000
11.2652
1.0000
0.9629
5.2834
mtDNA 1st
461bp
207
157
TVM+I+G
0.3317
0.2924
0.1781
0.1977
0.3617
3.2252
0.3758
0.1153
3.2252
1.0000
1.1350
3.9619
doi:10.1016/j.ympev.2009.08.001
mtDNA 2nd
461bp
138
59
GTR+I+G
0.1639
0.3299
0.1036
0.4026
4.8158
65.4046
0.5403
2.7387
18.4150
1.0000
0.4718
7.6983
mtDNA 3rd
461bp
452
430
GTR+G
0.4388
0.3790
0.0584
0.1238
0.1125
10.5714
0.4878
0.0000
8.3023
1.0000
1.4953
10.8125