1
Lowland panmixia versus highland disjunction: Genetic and bioacoustic differentiation
2
in two species of East African White-eye birds
3
Jan Christian Habel1*, Werner Ulrich2, Gustav Peters3, Martin Husemann1,4, Luc Lens5
4
5
6
1
7
Technische Universität München, D-85350 Freising-Weihenstephan, Germany
8
2
9
Toruń, Poland
Terrestrial Ecology Research Group, Department of Ecology and Ecosystem Management,
Chair of Ecology and Biogeography, Nicolaus Copernicus University in Toruń, PL-87100
10
3
Zoologisches Forschungsmuseum Alexander Koenig, D-53113 Bonn, Germany
11
4
Baylor University, Biology Department, Waco, TX-76798, USA
12
5
Terrestrial Ecology Unit, Department of Biology, Ghent University, B-9000 Ghent, Belgium
13
14
*Corresponding author:
15
Jan Christian Habel, Terrestrial Ecology Research Group, Department of Ecology and
16
Ecosystem Management, Technische Universität München, Hans-Carl-von-Carlowitz-Platz 2,
17
D-85350 Freising-Weihenstephan, Germany
18
E-Mail: Janchristianhabel@gmx.de
19
20
Running title: Genetic and bioacoustic analyses of East African White-eyes
21
1
22
ABSTRACT
23
East-African mountain forest species often occur in small and isolated populations, whereas
24
species inhabiting the dry lowland savannahs exist in large and interconnected population
25
networks. Taxa with closely related highland and lowland species, such as the East-African
26
White-eye birds, allow testing for the potential effects of the two contrasting distribution
27
patterns, mountain disjunction versus lowland panmixia. In this study, we compare the
28
population genetic and bioacoustic differentiation of two representatives of the genus
29
Zosterops: Zosterops poliogaster is exclusively found in forests at higher elevations; in
30
comparison, Zosterops abyssinicus, only occurs in the dry and warm lowland savannahs. Both
31
species were analysed across a similar geographical scale. Population genetic differentiation
32
was inferred using the same set of 15 polymorphic microsatellite loci for both species. In
33
addition, we quantitatively analyzed bioacoustic traits. Both data sets indicate a strong
34
population differentiation among populations of the highland species, but an absence of
35
differentiation in the lowland species. In addition, the lowland Z. abyssinicus was
36
characterised by a twofold higher genetic diversity than detected for the highland
37
Z. poliogaster. These two contrasting intraspecific population structures may reflect the
38
opposite ecology and distribution of these species: the strong population isolation of
39
Z. poliogaster resulting from long-term restriction to the cool and moist mountain forests at
40
higher elevations has led to strong differentiation among local populations and resulted in a
41
comparatively low level of intraspecific variability. In contrast, population panmixia in the
42
lowland Z. abyssinicus provides a high level of gene flow allowing the maintenance of high
43
genetic diversity and avoiding strong population structuring. These findings need to be
44
considered when planning conservation actions.
45
46
Keywords: Bioacoustics, cloud forest, differentiation, disjunction, diversity, panmixia,
47
population genetics, savannah
2
48
INTRODUCTION
49
The mountain cloud forests of the Eastern Afromontane biodiversity hotspot provide a
50
heterogeneous habitat consisting of a mosaic of moist and cool highland archipelagos and
51
highly disjunct cloud forest patches surrounded by dry and warm lowland savannah (White
52
1978; Burgess et al. 2007). The biota of the cloud forests are characterised by long-term
53
geographic isolation and have experienced relatively stable climatic conditions (Fjeldså and
54
Lovett 1997; Measey and Tolley 2011; Tolley et al. 2011). These two factors have supported
55
the accumulation of a large number of endemic species, often being restricted to single or few
56
mountain ranges (Rodgers and Homewood 1982; Burgess et al. 1998; Ehrich et al. 2007). In
57
contrast, the lowland savannah provides a large and interconnected habitat situation – and
58
thus an opposite habitat scenario.
59
60
Long lasting isolation of populations often results in strong differentiation patterns (Endler
61
1982; Hendry et al. 2000; Smith et al. 2001; Nosil 2012). Such differentiation can be driven
62
either by stochastic or directional processes. Neutral genetic drift for example strongly
63
depends on the restriction of gene flow and plays a larger role when population sizes are
64
small. Such populations are typically characterized by low levels of variability and high
65
differentiation of isolated populations is common (reviewed in Habel et al. 2013a). Apart
66
from stochastic processes, differences in the local selective environment (e.g. diverging
67
habitat structures, predators, species interactions) may lead to local adaptation even in the face
68
of ongoing gene flow (Danley et al. 2000). Divergent local selection is largely independent of
69
the size of a population and can lead to strong population differentiation (Nosil 2012). Drift
70
might be an important factor in the isolated mountain populations with rather small effective
71
population sizes (Habel & Husemann unpublished data). In contrast, random processes likely
72
play a rather limited role in the lowland species, as high population sizes and high rates of
73
gene flow prevent intraspecific differentiation.
3
74
75
In this study we investigate the effects of the two different habitat distribution types –
76
disjunction and interconnectivity - on the intraspecific differentiation patterns of two bird
77
species. Two representatives of the genus Zosterops serve as our model taxa: while the
78
lowland-inhabiting Abyssinian White-eye Zosterops abyssinicus has a fairly continuous
79
distribution across the widespread savannahs, the highland-dwelling Mountain White-eye
80
Zosterops poliogaster occurs in very isolated mountain forest patches. We used an identical
81
set of polymorphic microsatellites for both species and analysed characteristics of the
82
sonograms of their contact calls to test for potential intraspecific signatures of these two
83
contrasting distribution settings. Based on this multi-marker approach we test the following
84
hypotheses:
85
(i)
86
87
diverse accompanied by a lack of intraspecific differentiation.
(ii)
88
89
The interconnected populations of the lowland savannah species are genetically
The isolated mountain populations of Z. poliogaster are genetically distinct and
exhibit low genetic diversity as a result of small isolated populations.
(iii)
The genetic and acoustic patterns are concordant.
90
4
91
MATERIAL AND METHODS
92
Study species
93
The genus Zosterops is known for its high speciation rates and great colonization potential
94
(Warren et al. 2006; Moyle et al. 2009; Melo et al. 2011; Oatley et al. 2012). In East Africa
95
the genus occurs with several species with divergent ecology, some restricted to cloud forests
96
whereas others occur mostly in savannah habitats (Redman et al. 2009). Zosterops
97
poliogaster, also known as the Mountain White-eye, is restricted to evergreen cloud forests at
98
higher elevations in East Africa (e.g. in the Taita Hills from 850 to 1700 meters, Mulwa et al.
99
2007). The species can be found from Eritrea and Somalia in the north to Tanzania, e.g. in the
100
highlands of Ethiopia, in the high mountain systems of Kenya, in the Eastern Arc Mountains
101
of northern Tanzania, and on Mt. Elgon in Uganda (Zimmermann et al. 1999; Redman et al.
102
2009). It requires moist and cool climatic conditions, which has led to a highly disjunct
103
geographic distribution. This caused the evolution of various morphologically distinct, locally
104
endemic population clusters (Borghesio and Laiolo 2004; Mulwa et al. 2007; Redman et al.
105
2009). The closely related lowland species Abyssinian White-eye Zosterops abyssinicus is
106
widely distributed across the dry savannahs of East Africa and can be easily distinguished
107
from Z. poliogaster by various morphological characters, especially a much smaller eye ring
108
size and yellower plumage coloration (Redman et al. 2009).
109
110
Study area
111
We collected DNA samples for molecular analyses and recorded contact calls for several
112
population groups across Kenya. The following sites were sampled for Z. poliogaster: Mt.
113
Kulal, Aberdares, two forest fragments in the Chyulu Hills (Satellite and Simba Valley), and
114
four forest fragments located in three separate mountain isolates in the Taita Hills (Chawia and
115
Ngangao forests on Mt. Dabida, Mbololo forest on Mt. Mbololo, and Kasigau forest on Mt. Kasigau)
116
(Fig. 1). Zosterops abyssinicus populations were sampled over large parts of the Kenyan
5
117
lowland savannahs, including some localities being in close geographic proximity to the
118
sampled highland populations of Z. poliogaster: The following sites were sampled for
119
Z. abyssinicus: Mt. Nyiru, Mumoni Hills, Hunters Lodge, Kibwezi and Mtito Andei (both at
120
the foothills of Chyulu Hills), as well as at Dembwa and Mwatate (both at the foothills of the
121
Taita Hills). However, not for all of the populations both data types (molecular and
122
bioacoustic) were available. Details on sampling locations, number of DNA samples and
123
number of recorded and analysed contact calls are listed in Table 1. Sampling sites are
124
displayed in Fig. 1.
125
126
Molecular data
127
Individuals were captured with mist nets and ringed individually during 2009-2012. Blood
128
samples were taken from the birds´ brachial vein using a sterile needle and a 20µl capillary.
129
Blood was stored in pure ethanol and frozen at -20°C until DNA extraction. Alternatively we
130
extracted DNA from feathers. DNA extraction, standard PCR procedure and fragment length
131
detection was performed for both species as described in Habel et al. (2013b). We analysed
132
the same 15 polymorphic microsatellites for both species (ZL44, ZL41, Cu28, ZL18, ZL50,
133
ZL49, ZL14, ZL22, ZL45, Mme12, ZL35, ZL04, ZL54, ZL37, ZL2) for a total of 257
134
individuals from 10 sites. Sample sizes ranged from 9 to 35 individuals per population, with a
135
mean number of 25.7 individuals per population. Part of the molecular data was taken from a
136
previous study (Habel et al. 2013b, indicated in Table 1).
137
138
The microsatellite data was tested for distortion through stutter bands, large allele dropout or
139
null alleles (Selkoe and Toonen 2006) using MICROCHECKER vers. 2.2.3 (Van Oosterhout et
140
al. 2004). Parameters of genetic variability including mean number of alleles A, percentage of
141
expected heterozygosity He, percentage of observed heterozygosity Ho, tests on Hardy-
142
Weinberg equilibrium (HWE), and linkage disequilibrium were calculated with ARLEQUIN
6
143
vers. 3.1 (Excoffier et al. 2005). Allelic richness (AR) based on the lowest number of
144
individuals sampled (here on 9 individuals) was calculated with FSTAT vers. 2.9.3.2 (Goudet
145
1995). We further calculated the percentage of private alleles (AP) being restricted to single
146
mountain areas for Z. poliogaster, and single populations in Z. abyssinicus.
147
148
STRUCTURE vers. 2.3.3 (Pritchard et al. 2000) was used to infer the most probable number of
149
genetic clusters for both species separately. For each species the total number of populations
150
was defined as the maximum number of clusters (K=8 for Z. poliogaster, K=2 for
151
Z. abyssinicus). We used the batch run function to carry out a total of 80 runs (10 each for one
152
to 8 clusters, i.e. K = 1 to K = 8) for Z. poliogaster, which allowed us to quantify deviations
153
among multiple different runs for a fixed K and to calculate means and standard deviations.
154
For Z. abyssinicus we carried out a total of 20 runs (10 each for one to two clusters, i.e. K = 1
155
to K = 2). For both species the burn-in and simulation length were 150,000 and 500,000 runs,
156
respectively. Since the log probability values for the different K-values were shown to yield
157
inaccurate numbers of genetic clusters in some cases, we calculated ∆K (Evanno et al. 2005)
158
to infer the most likely number of groups (Kalinowski 2011).
159
160
Analyses of molecular variance (AMOVAs) were carried out using F-statistics as well as the
161
microsatellite-specific R-statistics (Slatkin 1995). For the mountain forest species Z.
162
poliogaster we conducted hierarchical variance analyses using pre-defined clusters, according
163
to our Structure results (highest clustering probability for K = 4, see results) and according the
164
five mountain isolates where Z. poliogaster was collected (Taita Hills (Mt. Kasigau), Taita
165
Hills (Chawia, Ngangao and Mbololo), Chyulu Hills, Aberdares, Mt. Kulal). For the lowland Z.
166
abyssinicus we analysed the genetic differentiation between the two local populations.
167
Genetic distances were calculated using the Cavallis-Sforza & Edwards (1967) algorithm
168
implemented in ARLEQUIN.
7
169
170
To test the level and direction of gene flow among locations and genetic clusters we used
171
BAYESASS vers. 1.3 (Wilson and Rannala 2003). This programme relies on multi-locus
172
genotypes and a Markov Chain Monte Carlo (MCMC) algorithm to estimate proportions of
173
non-migrants and the source of migrants for each sampling site (Wilson and Rannala 2003).
174
We performed runs of the algorithm with 9 × 106 iterations; 3 × 106 iterations were discarded
175
as burn-in. Delta values of m = 0.30, P = 0.15, and F = 0.15 yielded an average number of
176
changes in the accepted range. We tested for potential gene flow among all sampling sites
177
including both taxa.
178
179
Bioacoustics
180
Contact calls of foraging bird flocks of both study species were recorded during February-
181
April in the years 2010 and 2013 using a Sennheiser ME67 directional microphone
182
(Sennheiser, Hanover, Germany). The frequency curve 3 was selected to filter lower
183
frequencies during recording. A digital Zoom-H4 recorder was used to save the calls.
184
Microphone and recorder were linked with a Sennheiser K6CL Speisemodul. The input level
185
of 100% was operated manually and not changed during recording. Contact calls of the birds
186
are known to be used for maintaining flock structure, as well as for mate recognition
187
(Robertson 1996; Condo & Watanabe 2009). We recorded at a distance of approximately five
188
meters between microphone and bird flocks. Individuals mostly emit these calls in series and
189
regular intervals. As both species generally occur in large flocks (sizes ranging from few
190
individuals to some tens), and often multiple individuals emit calls simultaneously, our
191
dataset may contain some repeated recordings from the same individuals. Calls were recorded
192
between 6:00 am and 6:00 pm for a period of three days at each site. Recordings were saved
193
as stereo wav-files.
194
8
195
We selected high quality calls using PRAAT vers. 5.2.15 (Boersma 2002) and excluded
196
recordings affected by strong background noise from further analyses. After quality control, a
197
total number of 1997 bird calls were kept for further analyses (a mean of 154 calls per site (±
198
92 SD), ranging from 17 to 345 calls per site, Table 1). For each call we measured the
199
following characters: Starting frequency (sometimes identical to lowest frequency), first peak
200
(sometimes identical to highest frequency), end frequency (sometimes identical with lowest
201
frequency), highest and lowest frequency, total length (duration) of each call and the
202
frequency range (by subtracting the lowest frequency from the highest frequency). The
203
scoring of sonograms was performed blind to site (i.e. population) and species and was done
204
by the same person (JCH). Selected dynamic range (bit depth) was identical for all calls
205
analysed.
206
207
We used Principal Component Analysis (PCA) and applied K-means clustering to group
208
populations based on the six bird call characteristics (Ding and He 2004). The resulting
209
groupings were used to predefine units for subsequent one-way ANOVA and pair-wise post-
210
hoc Tukey tests on the six bird call characteristics. Furthermore, correlation between
211
geographic distances and the similarity of bird calls were tested using a Mantel test. All
212
statistical analyses were performed with STATISTICA vers. 7 (StatSoft, Tulsa, OK, USA).
213
9
214
RESULTS
215
Molecular data
216
Genetic diversity strongly differs between the lowland and the highland species, with about
217
twice the amount of genetic diversity found in the lowland species Z. abyssinicus when
218
compared to the highland species Z. poliogaster. Within the mountain clusters of Z.
219
poliogaster we detected the highest genetic diversity in the population from the Aberdares,
220
and the lowest values in the Taita Hills populations. Population specific values are given in
221
Table 2.
222
223
The ΔK values of the STRUCTURE analysis indicated the highest probability for a clustering
224
into four groups (K = 4) for the highland species Z. poliogaster. The four clusters were in
225
accordance with the following mountain isolates: Mt. Kulal, Chyulu Hills, Taita Hills
226
(Chawia, Ngangao and Mbololo) and Taita Hills (Mt. Kasigau). The latter mountain isolate
227
being geologically assigned to the Taita Hills showed a strong genetic admixture with
228
individuals from the other parts of the Taita Hills (Fig. 2). In contrast to Z. poliogaster, the
229
lowland species Z. abyssinicus showed no genetic clustering. Structure plots for both species
230
are shown in Fig. 2. Probability values and respective ∆K for each K are given in Appendix
231
S1.
232
233
Analyses of molecular variance (AMOVAs) revealed that a similar portion of the total
234
variance is explained by the differences between the highland Z. poliogaster and the lowland
235
Z. abyssinicus (18.1769, RCT = 0.2868, p < 0.0001) as is by the differences between local
236
populations of the highland species Z. poliogaster (19.3574, RST = 0.3958, p < 0.0001;
237
0.9625). When applying AMOVA to the STRUCTURE results we found that much of the total
238
genetic variance is explained by the divergence among the four clusters (24.8257, RCT =
239
0.4426, p < 0.0001). We detected a weak, but significant genetic differentiation among the
10
240
neighbouring Z. poliogaster populations analysed for the Taita Hills (including the forest
241
patches Ngangao, Chawia, and Mbololo, the latter one separated only by a valley). No
242
significant genetic differentiation could be detected between the two Z. abyssinicus
243
populations (-1.3293, RST = -0.0085, p > 0.05). The detailed results are given in Table 3,
244
respective values obtained from F-statistics are given in Appendix S2.
245
246
We detected low migration rates among the mountain populations of Z. poliogaster, but some
247
gene flow within single mountain massifs, like in the Chyulu Hills (from Satellite to Simba
248
Valley, 18.5%) and within the Taita Hills (from Mbololo to Ngangao, 30.8%). The population
249
from Mt. Kasigau (about 50 km south of the other Taita Hills populations (Chawia, Ngangao
250
and Mbololo)) seemed to have received a relatively large proportion of its gene pool from the
251
Mbololo site of the Taita Hills (27.9%), which is in line with the results obtained from
252
STRUCTURE analyses and our AMOVA calculations. Gene flow estimates between the two Z.
253
abyssinicus populations was high (from Kibwezi to Nyiru, 27.4%) despite the large
254
geographic distance (> 600 km). No interspecific gene flow was detected between Z.
255
abyssinicus and Z. poliogaster. A Mantel test showed that the genetic distance (based on
256
Cavallis-Sofrza & Edwards (1967) algorithm) is not significantly correlated with the
257
geographic distance (no isolation by distance).
258
259
Bioacoustics
260
The first two axes of a Principal Component Analysis (PCA) explained 43.0% (PC1call) and
261
28.7% (PC2call) of the total variance. PC1call was highly correlated with the maximum
262
frequency (r = 0.64) and differences in maximum frequency (r = 0.49) while PC2call was
263
correlated with the starting frequency (r = 0.54) and the lowest frequency (r = 0.55).
264
11
265
An ANOVA revealed significant differences in the total length of calls and in call frequencies
266
(start frequency, highest frequency and range) between Z. poliogaster and Z. abyssinicus (Fig.
267
3), (PC1, PC2, t-test: p < 0.001). PCA further indicated strong differences in the starting
268
frequencies (scoring highest with PC2call) among local populations of both Zosterops species
269
(Fig. 3). Both analyses showed a high overlap in Z. abyssinicus call patterns among the 6
270
populations recorded (Fig. 3). Pairwise post-hoc Tukey comparisons of sites corroborated the
271
PC clustering and showed that most Z. abyssinicus populations are not significantly
272
differentiated in their calls (Table 4). In contrast, we found significant differences among the
273
local mountain populations of Z. poliogaster (ANOVA: p < 0.001) (Fig. 3, Table 4). For Z.
274
poliogaster, population differentiation in call patterns showed a strong correlation with the
275
geographic distance (Mantel test: r = 0.48, p < 0.001). A main split was detected between the
276
populations from Mt. Kulal and Aberdares which formed a group and individuals from the
277
Chyulu Hills, Taita Hills (Chawia, Ngangao and Mbololo) and Taita Hills (Mt. Kasigau) (Fig.
278
3B). This pattern is congruent with the main split derived from our molecular data. Further,
279
the calls of Z. poliogaster showed a low variance within single populations. In contrast, the
280
range of calls in Z. abyssinicus was comparatively broad. A Mantel test used to compare
281
genetic distances and acoustic distances among populations did not show a significant
282
correlation (r = 0.09, p > 0.36). Pairwise distances (genetic and geographic) are given in
283
Appendix S3.
284
12
285
DISCUSSION
286
The signatures of panmixia and disjunction
287
Our genetic and bioacoustic data indicate strong differentiation among local populations of Z.
288
poliogaster from isolated mountains. These deep genetic and bioacoustic splits found in
289
conspecific populations of Z. poliogaster even exceeded the level of divergence between the
290
highland and lowland species. In parallel, the genetic and bioacoustic variability found within
291
each population of Z. poliogaster is rather low. In contrast to these strong intraspecific
292
signatures observed for Z. poliogaster, the lowland Z. abyssinicus showed a contrasting
293
pattern: no differentiation neither genetically, nor bioacoustically, accompanied by high levels
294
of intra-population variation.
295
296
The genetic and bioacoustic patterns reflect the two opposite distribution settings of the taxa:
297
the lowland species has a continuous distribution across the lowland savannahs. This suggests
298
a panmictic or metapopulation-like structure with interconnected local populations. The high
299
intraspecific variability is maintained in an interconnected population network (i.e. high rates
300
of gene flow) of large populations. The opposite scenario applies to Z. poliogaster, where
301
local populations show strong genetic and bioacoustic differentiation, but comparatively low
302
levels of genetic and bioacoustic variability within single populations or groups. This might
303
be the result of long term isolation at geographically separated mountain exclaves with rather
304
small population sizes. The combination of population differentiation and low intra-
305
population variation is a typical feature found in species with highly disjunct occurrences
306
(Habel et al. 2013a).
307
308
Complex differentiation with patterns in Zosterops poliogaster
309
The bioacoustic and genetic differentiation patterns of Z. poliogaster differ from each other.
310
While our molecular data showed no correlation between the genetic and geographical
13
311
distance, the bioacoustic data showed similarities in geographically close populations,
312
whereas geographically distant populations strongly differed from each other (Mantel test: r =
313
0.48, p < 0.001). Our genetic data revealed two main clusters within Z. poliogaster: the
314
southern Kenyan populations including all Taita Hills populations (including Mt. Kasigau),
315
and all other populations from the Central Kenyan Highlands, including northern Kenya. This
316
split separates individuals from the Taita Hills from individuals from the Chyulu Hills, less
317
than 60 km apart. In parallel, individuals from the Chyulu Hills are genetically similar to
318
individuals from Mt. Kulal (more than 600 km distant). Yet, the population from the latter
319
mountain massif shows similarities in their contact calls with individuals from the Taita Hills.
320
321
The differences in genetic and acoustic differentiation patterns indicate that different
322
evolutionary forces might drive their divergence. Assuming that the molecular markers used
323
in this study represent neutral loci, the significant population differentiation of Z. poliogaster
324
indicates that drift might play a strong role for their divergence. Long-lasting geographic
325
isolation accompanied by comparatively small habitat sizes, and consequently population
326
sizes, make populations more vulnerable to stochastic events and can lead to an accelerated
327
fixation of alleles. Similar complex differentiation patterns on mountain islands were
328
observed for other taxa in parts of the Eastern Afromontane region, e.g. in chameleons
329
(Measey and Tolley 2011).
330
331
In turn, acoustic behaviour generally is considered to be under sexual or natural selection
332
(Brooks et al. 2005). The contact calls of Zosterops are known to serve as mate recognition
333
character and social signal for maintaining flock structure (Robertson 1996), and therefore
334
might be under sexual or natural selection. However, if selection was the driving force of
335
acoustic differentiation in this system one would expect that contact calls would be stabilized
336
to ensure the recognition of conspecifics. Interestingly, in our system local populations
14
337
strongly differ in acoustic traits. Therefore, it is possible that divergent ecological conditions
338
might change the acoustic environment at a location leading to different local optima.
339
However, if selection is not as strong on contact calls as expected or only specific acoustic
340
parameters are under selection drift could explain acoustic population divergence as well. If
341
drift as a random process acted on two different traits simultaneously one would expect
342
divergent differentiation patterns, such as observed for Z. poliogaster.
343
344
Implementations for conservation
345
The lowland species seems to exist in a large and interconnected population network which
346
guarantees species persistence and the maintenance of high and homogeneously distributed
347
intraspecific diversity. For the long-term persistence of such species, the creation of single,
348
large conservation areas are adequate. In contrast, the geographically disjunct distribution of
349
the highland congener Z. poliogaster with various population groups representing unique
350
molecular and bioacoustic characteristics restricted to single mountain massifs poses a
351
conservation challenge (Moritz 1994). To conserve the entire intraspecific variation, a large
352
proportion of the distribution range has to be taken into consideration when integrating
353
several small conservation areas (see the debate on adapted conservation strategies, the
354
SLOSS-debate, Virolainen et al. 1998; Stockhausen & Lipcius 2001). The preservation of
355
such isolated and small populations might be further complicated due to ongoing
356
deforestation in major parts of Kenya (Pellikka et al. 2009).
357
15
358
Acknowledgements
359
This study was funded by the German Academic Exchange Service (DAAD). We thank Titus
360
Iboma, Onesmus M. Kioko and Ronald K. Mulwa (NMK Nairobi, Kenya) for field assistance.
361
We thank Krystal Tolley and one anonymous referee for improving a previous version of this
362
article.
363
16
364
References
365
Abbott R, Albach D, Ansell S, Arntzen JW, Baird SJ, Bierne N, Boughman J, Brelsford A,
366
Buerkle CA, Buggs R, Butlin RK, Dieckmann U, Eroukhmanoff F, Grill A, Cahan SH,
367
Hermansen JS, Hewitt G, Hudson AG, Jiggins C, Jones J, Keller B, Marczewski T,
368
Mallet J, Martinez-Rodriguez P, Möst M, Mullen S, Nichols R, Nolte AW, Parisod C,
369
Pfennig K, Rice AM, Ritchie MG, Seifert B, Smadja CM, Stelkens R, Szymura JM,
370
Väinölä R, Wolf JBW, Zinner D (2013) Hybridization and speciation. J Evol Biol
371
26:229–246.
372
373
374
375
376
377
378
379
380
381
Audacity Development Team (2010) Audacity, http://audacity.sourceforge.net/ (accessed
5.1.2012)
Boersma PPG (2002) Praat - a system for doing phonetics by computer. Glot International
5:341-345.
Borghesio L, Laiolo P (2004) Habitat use and feeding ecology of Kulal White-eye Zosterops
kulalensis. Bird Conserv Internat 14:11–24.
Brooks T, Lens L, De Meyer M, Waiyaki E, Wilder C (1998) Avian biogeography of the
Taita Hills, Kenya. J East African Nat Hist 87:189-194.
Brooks R, Hunt J, Blows JW, Smith MJ, Bussiére LF, Jennions MD (2005) Experimental
evidence for multivariate stabilizing sexual selection. Evolution 59:871-880.
382
Burgess ND, Butynski TM, Cordeiro NJ, Doggart NH, Fjeldsa J, Howell KM, Kilahama FB,
383
Loader SP, Lovett JC, Mbilinyi B, Menegon M, Moyer DC, Nashanda E, Perkin A,
384
Rovero F, Stanley WT, Stuart SN (2007) The biological importance of the Eastern Arc
385
Mts. of Tanzania and Kenya. Biol Conserv 34:209–231.
386
387
388
389
390
391
392
Catchpole CK (1987) Bird song, sexual selection and female choice. Trends Ecol Evol 2:94–
97.
Danley PD, Markert JA, Arnegard ME, Kocher TD (2000) Divergence with gene flow in the
rock-dwelling cichlids of Lake Malawi. Evolution 54:1725-1737.
Ding C, He X (2004) K-means clustering via principal component analysis. Proceedings of
the twenty-first international conference on Machine learning, 29.
Ehrich D, Gaudeul M, Assefa A, Koch MA, Mummenhoff K, Nemomissa S, Intrabiodiv
393
consortium, Brochmann C (2007) Genetic consequences of Pleistocene range shifts:
394
contras between the Arctic, the Alps and the East African mountains. Mol Ecol
395
16:2542–2559.
396
397
Endler JA (1982) Pelistocene forest refuges: fact or fancy, in: Biological diversification in the
tropics, Prance GT (ed) Columbia University Press, New York.
17
398
399
400
401
402
Evanno G, Regnaut S, Goudet J (2005) Detecting the number of clusters of individuals using
the software STRUCTURE: a simulation study. Mol Ecol 14:2611–2620.
Excoffier L, Laval G, Schneider S (2005) Arlequin ver. 3.0: an integrated software package
for population genetics data analysis. Evol Bioinf Online 1:47–50.
Fjeldså J, Lovett JC (1997) Geographical patterns Geographical patterns of old and young
403
species in African forest biota: the significance of specific montane areas as
404
evolutionary centres. Biodiv Conserv 6:325-346.
405
Fuchs J, Fjeldså J, Bowie RCK (2011) Diversification across an altitudinal gradient in the
406
Tiny Greenbul (Phyllastrephus debilis) from the Eastern Arc Mountains of Africa. BMC
407
Evol Biol 11:117.
408
409
410
411
412
413
414
Goudet J (1995) Fstat (version 1.2): a computer program to calculate F-statistics. Heredity
86:485–486.
Habel JC, Cox S, Gassert F, Meyer J, Lens L (2013b) Population genetics of four East
African Mountain White-eye congeners. Conserv Genet.
Habel JC, Rödder D, Schmitt T, Lens L (2013a) The genetic signature of ecologically
diverging grassland lepidopterans. Biodiv Conserv 22:2401-2411.
Hendry AP, Wenburg JK, Bentzen P, Volk EC, Quinn TP (2000) Rapid evolution of
415
reproductive isolation in the wild: evidence from introduced salmon. Science 290:515-
416
518.
417
Kalinowski ST (2011) The computer program STRUCTURE does not reliably identify the
418
main genetic clusters within species: simulations and implications for human population
419
structure. Heredity 106:625–632.
420
421
Kondom N, Watanabe S (2009) Contact calls: Information and social function. Jap Psych Res
51:197-208.
422
Measey GJ, Tolley KA (2011) Sequential fragmentation of Pleistocene forests in an East
423
Africa biodiversity hotspot: Chameleons as a model to track forest history. PLoSONE
424
6:e26606.
425
Melo M, Warren BH, Jones PJ (2011) Rapid parallel evolution of aberrant traits in the
426
diversification of the Gulf of Guinea white-eyes (Aves, Zosteropidae). Mol Ecol
427
20:4953-4967.
428
429
Moritz C (1994) Defining evolutionary significant units for conservation. Trends Ecol Evol
9:373-375.
18
430
Moyle RG, Filardi CE, Smith CE, Diamond J (2009) Explosive Pleistocene diversification
431
and hemispheric expansion of a “great speciator”. Proc Nat Acad Sci USA 106:1863–
432
1868.
433
Mulwa RK, Bennun LA, Ogol CKPO, Lens L (2007) Population status and distribution of
434
Taita White-eye Zosterops silvanus in the fragmented forests of Taita Hills and Mount
435
Kasigau, Kenya. Bird Conserv Int 17:141–150.
436
Nosil P (2012) Ecological Speciation. Oxford Series in Ecology and Evolution. 304pp.
437
Oatley G, Voelker G, Crowe TM, Bowie RCK (2012) A multi-locus phylogeny reveals a
438
complex pattern of diversification related to climate and habitat heterogeneity in
439
southern African white-eyes. Mol Phyl Evol 64:633-644.
440
441
442
Panhuis TM, Bulin R, Zuk M, Tregenza T (2001) Sexual selection and speciation. Trends
Ecol Evol 16:364–371.
Pellikka PKE, Lötjönen M, Siljander M, Lens L (2009 ) Airborne remote sensing of
443
spatiotemporal change (1955–2004) in indigenous and exotic forest cover in the Taita
444
Hills, Kenya. Int J Appl Earth Observ Geoinf 11:221-232.
445
446
447
Potvin DA, Parris KM (2012) Song convergence in multiple urban populations of silvereyes
(Zosterops lateralis). Ecol Evol 2:1977-1984.
Potvin DA, Parris KM, Mulder RA (2013) Limited genetic differentiation between
448
acoustically divergent populations of urban and rural silvereyes (Zosterops lateralis).
449
Evol Ecol 27:381-391.
450
451
452
453
454
455
456
457
458
459
460
461
462
463
Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using
multilocus genotype data. Genetics 155:945–955.
Redman N, Stevenson T, Fanshawe J (2009) Birds of the Horn of Africa: Ethiopia, Eritrea,
Djibouti, Somalia, and Socotra. Princeten Field Guide, 488 pp.
Robertson BC (1996) Vocal mate recognition in a monogamous, flock-forming bird, the
silvereye, Zosterops lateralis. Anim Behav 51:303-311.
Rodgers WA, Homewood KM (1982) Species richness and endemism in the Usambara
mountains forests, Tanzania. Biol J Linn Soc 18:197–242.
Selkoe T, Toonen RJ (2006) Microsatellites for ecologists: a practical guide to using and
evaluating microsatellite markers. Ecol Letters 9:615-629.
Servedio MR, Noor MAF (2003) The role of reinforcement in speciation: theory and data.
Ann Rev Ecol Evol Sys 34:339-364.
Slatkin M (1995) A measure of population subdivision based on microsatellite allele
frequencies. Genetics 139:457-462.
19
464
Smith TB, Schneider CJ, Holder K (2001) Refugial isolation versus ecological gradients-
465
testing alternative mechanisms of evolutionary divergence in four rainforest vertebrates.
466
Genetica 112-113:383-298.
467
468
469
Stockhausen WT, Lipcius RN (2001) Single large or several small marine reserves for the
Caribbean spiny lobster? Marine Freshw Res 52:1605-1614.
Tolley KA, Tilbury CR, Measey GJ, Menegon M, Branch WR, Matthee CA (2011) Ancient
470
forest fragmentation or recent radiation? Testing refugial speciation models in
471
chameleons within an African biodiversity hotspot. J Biogeogr 38:1748-1760.
472
Van Oosterhout C, Hutchinson WF, Wills DPM, Shipley P (2004) MICRO-CHECKER
473
(version 2.2.3): software for identifying and correcting genotyping errors in
474
microsatellite data. Mol Ecol Notes 4:535–538.
475
Virolainen KM, Suomi T, Suhonen J, Kuitunen M (1998) Conservation of vascular plants in
476
single large and several small mires: species richness, rarity and taxonomic diversity. J
477
Appl Ecol 35:700–707.
478
Warren BH, Bermingham E, Prys-Jones R, Thebaud C (2006) Immigration, species radiation
479
and extinction in a highly diverse songbird lineage: White-eyes on Indian Ocean islands.
480
Mol Ecol 15:3769–3786.
481
482
483
484
485
486
White F (1978) The afromontane region, in: Werger MJA (ed) Biogeography and ecology of
southern Africa. The Hague, Junk Publishers: 463-513.
Wilson GA, Rannala B (2003) Bayesian inference of recent migration rates using multilocus
genotypes. Genetics 163:1177-1191.
Zimmermann DA, Turner DA, Person DJ (1996) Birds of Kenya and Northern Tanzania.
London: Chrostopher Helm.
487
20
488
Figure 1: Sampling sites of the highland species Zosterops poliogaster (grey / green dots) and
489
lowland species Zosterops abyssinicus (white / yellow dots). Site numbers and names
490
coincide with other tables and figures. Country boarder between Kenya and Tanzania is
491
marked as black line. Zosterops poliogaster: 1 = Mt. Kulal, 2 = Aberdares, 3 = CH-Satellite, 4
492
= CH-Simba valley, 5 = TH-Mbololo, 6 = TH-Ngangao, 7 = TH-Chawia, 8 = TH-Mt.
493
Kasigau; Zosterops abyssinicus: 9 = Mt. Nyiru, 10 = Mumoni Hills, 11 = Hunters Lodge, 12 =
494
Kibwezi, 13 = Mtito Andei, 14 = TH-Dembwa, 15 = TH-Mwatate. Abbreviations: CH =
495
Chyulu Hills, TH = Taita Hills.
496
21
497
Figure 2: Structure analysis of A) Zosterops poliogaster and B) Zosterops abyssinicus
498
performed with the Structure software (Pritchard et al. 2000). Structure plots are given for K
499
= 4 for Z. poliogaster determined as most appropriate by the highest ∆K (Evanno et al. 2005)
500
(see Appendix S1), and for Z. abyssinicus for K = 2 (maximum number of populations
501
analysed).
502
503
A)
504
505
B)
506
507
22
Figure 3: The first two axes of Principal Component Analysis (PC1: 43.0% of variance
509
explained, PC2: 28.8%) of contact call parameters of A) Zosterops abyssinicus, B) Zosterops
510
poliogaster, and C) contact calls of both species.
PCA 2
508
4
3
2
1
0
-1
-2
-3
-4
Mumoni
Kibwezi
Mtito Andei
Mwatate
Z. abyssinicus
PCA 2
-4
-3
-2
-1
0
1
2
3
4
B
4
3
2
1
0
-1
-2
-3
-4
Mt Kulal
Z. poliogaster
Ngangao
Aberdares
Mt Kasigau
Satellite
Mbololo
Simba
-4
PCA 2
A
Dembwa
Hunters Lodge
4
3
2
1
0
-1
-2
-3
-4
-3
-2
-1
0
1
2
3
4
C
Z. abyssinicus
Z. poliogaster
-4
-3
-2
-1
0
PCA 1
1
2
3
4
511
23
512
Table 1: Sampling sites for Zosterops poliogaster and Zosterops abyssinicus. Given is the
513
running number for each site (coinciding with other figures and tables), locality name,
514
number of recorded bird calls and number of sampled individuals for microsatellite analyses.
515
Abbreviations: CH = Chyulu Hills, TH = Taita Hills, * = data taken from Habel et al. (2013b).
516
Site Locality
Zosterops poliogaster
1
Mt. Kulal
2
Aberdares
3
CH-Satellite
4
CH-Simba valley
5
TH-Mbololo
6
TH-Ngangao
7
TH-Chawia
8
TH-Mt. Kasigau
Zosterops abyssinicus
9
Mt. Nyiru
10 Mumoni Hills
11 Hunters Lodge
12 Kibwezi
13 Mtito Andei
14 TH-Dembwa
15 TH-Mwatate
N µSat
N Calls
30*
25
28*
9
31
21*
26
21*
17
205
111
248
172
193
72
31
35*
-
121
78
263
38
138
345
517
24
518
Table 2: Parameters of genetic diversity obtained for Zosterops poliogaster and Z.
519
abyssinicus. Given are the locality and five parameters of genetic diversity: mean number of
520
alleles (A), allelic richness (AR), number of private alleles occurring restricted to single
521
mountain massifs (AP), percentage of expected heterozygosity (He), and percentage of
522
observed heterozygosity (Ho). Abbreviations: SD = Standard deviation, CH = Chyulu Hills,
523
TH = Taita Hills.
Locality
Zosterops poliogaster
Mt. Kulal
Aberdares
CH-Satellite
CH-Simba valley
TH-Mbololo
TH-Ngangao
TH-Chawia
TH-Mt. Kasigau
Mean (±SD)
Zosterops abyssinicus
Mt. Nyeri
Kibwezi
Mean (±SD)
A
AR
AP [%]
He [%]
Ho [%]
2.31
3.82
2.07
1.13
1.73
1.53
2.53
1.20
2.04
(±0.87)
2.21
2.73
2.06
1.74
1.74
1.72
2.10
1.5
1.98
(±0.38)
10.26
10.34
13.16
0.00
6.45
6.67
12.50
3.85
7.90
(±4.52)
29.91
35.96
26.00
16.48
20.83
18.72
25.15
18.44
23.94
(±6.64)
25.51
33.99
13.83
13.33
15.64
15.45
19.64
17.81
19.40
(±7.08)
4.10
4.07
4.09
(±0.02)
-
-
-
-
45.30
41.48
43.39
(±2.70)
40.70
34.74
37.72
(±4.21)
524
25
525
Table 3: Analyses of molecular variance (AMOVA) for the two species Zosterops poliogaster
526
and Z. abyssinicus. Given are results on potential differentiation between the two species,
527
differentiation within Z. poliogaster populations (mountain massifs). Variance values in top
528
line with respective R statistic values (in parenthesis below). Abbreviations: *: p < 0.05; **: p
529
< 0.01; ***: p < 0.001. Values obtained from F-statistics are given in Appendix S3.
Group
All Z. abyssinicus vs all Z.
poliogaster
Z. poliogaster
(according K = 4)
Z. poliogaster
five mountain areas
Among groups
or species (RCT)
18.1769
(0.2868***)
24.8257
(0.4426***)
19.3574
(0.3958***)
Among populations
within groups (RSC)
14.9725
(0.3313***)
1.04681
(0.0335)
0.8133
(0.0275)
Within
individuals
17.1037
17.1037
14.1379
530
26
531
Table 4: Results of ANOVA to infer differences in calls between the two species Zosterops
532
poliogaster and Z. abyssinicus, and among local populations.
Variable
Species
Sites
Error
SS
93547
8613
24812
df
1
11
1983
F
2792
62.6
-
P
<0.001
<0.001
-
27
533
Table 5: Significance levels of post-hoc Tukey tests for a one-way ANOVA applied to the K-means classification of site membership based on the
534
raw bird call patterns.
Z. poliogaster
Z. abyssinicus
Site
Kibwesi
Mtito Andei
Mumoni
Dembwa
Mwatate
CH-Satellite
CH-Simba valley
TH-Mt. Kasigau
TH-Mbololo
TH-Ngangao
Aberdares
Mt. Kulal
Zosterops abyssinicus
Zosterops poliogaster
Hunters
Mtito
CHCH-Simba
Kibwezi
Mumoni Dembwa Mwatate
Lodge
Andei
Satellite
valley
0.33
<0.001
0.42
0.49
0.99
0.31
0.95
0.98
0.08
0.99
0.99
0.21
0.03
0.93
0.99
0.49
<0.001 <0.001 <0.001
<0.001
<0.001
0.98
0.62
0.06
0.99
0.99
0.99
<0.001
0.47
0.99
0.37
0.99
0.99
0.98
<0.001
0.99
<0.001
<0.001
0.99
<0.001
<0.001
<0.001
<0.001
<0.001
0.13
0.99
0.63
0.99
0.79
0.12
<0.001
0.36
<0.001
<0.001 <0.001 <0.001
<0.001
<0.001
<0.001
<0.001
0.05
<0.001 <0.001 <0.001
<0.001
0.01
0.55
<0.001
TH-Mt.
THTHKasigau Mbololo Ngangao
0.07
0.99
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
<0.001
28
Aberdares
0.99
535
Appendix S1: Results of the STRUCTURE analysis. Ln(Pr) calculated using the STRUCTURE
536
software (white dots) that K is the correct number of populations. The ad-hoc statistic ∆K
537
(black cubes) based on the rate of change in the log probability of data between successive K
538
values (Evanno et al. 2005), is not applicable for K=1, and from the equation given in the
539
methods section it is obvious that it cannot be calculated for the highest K number either
540
(because data for K=1 are needed); we further ignore the high ΔK value obtained for K=2 as
541
suggested by Hausdorf and Hennig (2010).
180
Ln Probability of Data
1
-3000
2
3
4
5
6
7
160
140
120
100
-3500
80
60
-4000
Delta K
-2500
40
20
0
-4500
-20
542
543
29
544
Appendix S2: Analyses of molecular variance (AMOVA) to detect genetic differentiation
545
between the two species, Zosterops poliogaster and Z. abyssinicus, and the differentiation
546
within Z. poliogaster (mountain massifs). Variance values in top line with respective F
547
statistic values (in parenthesis below). Abbreviations: *: p < 0.05; **: p < 0.01; ***: p <
548
0.001. Respective R-statistics is given Table 3.
Group
All Z. abyssinicus vs all Z.
poliogaster
Z. poliogaster
(according K = 4)
Z. poliogaster
five mountain areas
Among groups
or species (FCT)
0.2440
(0.0815*)
0.4493
(0.1589***)
1.0142
(0.36595***)
Among populations
within groups (FSC)
0.8066
(0.2933***)
0.3524
(0.1482***)
0.0945
(0.0538***)
Within
individuals
1.7004
1.6833
1.4255
549
30
550
Appendix S3: Matrix of pairwise geographic distances (in km) among all locations and matrix of pairwise genetic distances based on Cavallis-
551
Sforza & Edwards (1976) and pairwise Rst-values among all populations analysed for Zosterops poliogaster.
Geographic distances
Mt. Kulal
Aberdares
CH-Satellite
CH-Simba Valley
TH-Mbololo
TH-Ngangao
TH-Chawia
TH-Mt. Kasigau
Mt.
Kulal
Aberdares
-
277.29
Cavalli-Sforza & Edwards (1976)
0.000
0.313
Mt. Kulal
0.000
Aberdares
CH-Satellite
CH-Simba Valley
TH-Mbololo
TH-Ngangao
TH-Chawia
TH-Mt. Kasigau
Pairwise Rst-values
Mt. Kulal
Aberdares
CH-Satellite
CH-Simba Valley
TH-Mbololo
0.00000
0.06739
0.12212
0.36561
0.68427
0.00000
0.16422
0.10798
0.53206
CHSatellite
CHSimba V.
THMbololo
THNgangao
THChawia
THMt. Kasigau
554.91
319.90
497.69
252.35
69.68
555.11
316.96
9.25
65.26
553.17
313.77
13.43
61.74
4.52
562.99
322.99
15.84
70.72
8.98
9.83
621.53
386.89
67.71
134.91
69.93
73.20
64.29
-
0.381
0.383
0.000
0.388
0.402
0.158
0.000
0.451
0.407
0.443
0.424
0.000
0.449
0.416
0.448
0.442
0.122
0.000
0.444
0.402
0.419
0.430
0.194
0.149
0.000
0.452
0.437
0.479
0.480
0.222
0.171
0.235
0.000
0.00000
0.05774
0.04359
0.00000
0.72115
0.00000
31
TH-Ngangao
TH-Chawia
TH-Mt. Kasigau
0.68516
0.57333
0.80641
0.52915
0.49116
0.58173
0.07260
0.08063
0.11373
0.71991
0.57565
0.87978
0.02303
0.03690
0.27484
0.00000
0.02020
0.15387
0.00000
0.20306
0.00000
552
32