bs_bs_banner
Biological Journal of the Linnean Society, 2015, 114, 828–836. With 2 figures
Effects of population size and isolation on the genetic
structure of the East African mountain white-eye
Zosterops poliogaster (Aves)
MARTIN HUSEMANN1,2, LAURENCE COUSSEAU3, LUCA BORGHESIO4, LUC LENS3
and JAN CHRISTIAN HABEL1*
1
Terrestrial Ecology Research Group, Department of Ecology and Ecosystem Management, Technische
Universität München, D-85354 Freising-Weihenstephan, Germany
2
General Zoology, Institute of Biology/Zoology, University of Halle, D-06120 Halle, Germany
3
Terrestrial Ecology Unit, Department of Biology, Ghent University, B-9000 Ghent, Belgium
4
C. Re Umberto 42, I-10128 Torino, Italy
Received 15 October 2014; revised 20 November 2014; accepted for publication 20 November 2014
Habitat size, quality and isolation determine the genetic structure and diversity of populations and may influence
their evolutionary potential and vulnerability to stochastic events. Small and isolated populations are subject to
strong genetic drift and can lose much of their genetic diversity due to stochastic fixation and loss of alleles. The
mountain white-eye Zosterops poliogaster, a cloud forest bird species, is exclusively found in the high mountains
of East Africa. We analysed 13 polymorphic microsatellites for 213 individuals of this species that were sampled
at different points in time in three mountain massifs differing in habitat size, isolation and habitat degradation.
We analysed the genetic differentiation among mountain populations and estimated the effective population sizes.
Our results indicate three mountain-specific genetic clusters. Time cohorts did not show genetic divergences,
suggesting that populations are large enough to prevent strong drift effects. Effective population sizes were higher
in larger and geographically interconnected habitat patches. Our findings underline the relevance of ecological
barriers even for mobile species and show the importance of investigating different estimators of population size,
including both approaches based on single and multiple time-points of sampling, for the inference of the
demographic status of a population. © 2015 The Linnean Society of London, Biological Journal of the Linnean
Society, 2015, 114, 828–836.
ADDITIONAL KEYWORDS: cloud forest – effective population size – fragmentation – habitat isolation –
habitat size – metapopulation dynamics – microsatellites.
INTRODUCTION
Habitat size, habitat quality and isolation affect the
demographic and genetic structure of local populations as well as their spatial structure (Fahrig, 2003).
Large and interconnected habitats have the potential to harbour genetically diverse populations
(Frankham, 2005). In contrast, small and isolated
habitat fragments provide limited resources for small
local populations that are often characterized by low
*Corresponding author. E-mail: [email protected]
828
genetic diversity due to stochastic drift effects
(Frankham, 2005). As a result of demographic and
genetic stochasticity, small populations in isolated
habitats are often of conservation concern.
One way to test for the effects of geographical
isolation and patch size on the genetic makeup of
local populations is to test for genetic differentiation
and to estimate their effective population size (Ne)
(according the definition by Wright, 1938). Estimating
Ne helps to determine how vulnerable a population is
to stochastic effects (Charlesworth, 2009) and thus
provides important information on the conservation
status of a population or species (Luikart et al., 2010).
© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, 114, 828–836
FRAGMENTATION GENETICS OF A MOUNTAIN BIRD
However, robustly estimating Ne in natural populations is challenging as the best approaches require
large numbers of samples collected from the same
population, best sampled at multiple points in time
(Barker, 2011; Habel et al., 2014a). In cases where
such samples are available, spatio-temporal genetic
analyses can give insights into the mechanisms
underlying population differentiation, and can inform
scientists and conservationists on the threat status of
a local population (Palstra & Ruzzante, 2008; Luikart
et al., 2010; Phillipsen et al., 2011; Horreo et al.,
2013).
In this study, we analyse the degree of population
differentiation and estimate the effective population
sizes of the mountain white-eye Zosterops poliogaster.
This bird species is exclusively found at higher elevations across the East African highlands (Moreau,
1957; Mulwa et al., 2007) and differs morphologically,
genetically and bioacoustically from its lowland congeners and within the same taxon among distinct
mountain populations (Moreau, 1957; Zimmerman
et al., 1996; Redman et al., 2009; Cox et al., 2014;
Habel et al., 2013, 2014b; Husemann, Ulrich & Habel,
2014).
One way to understand potential forces driving this
diversification is to estimate the effective population
sizes, as these determine the relative importance drift
may play. Small effective population sizes (below 500
individuals) allow for larger effects of stochastic
829
events and drift. The availability of large numbers of
individuals collected during the past allows us to
estimate Ne based on multiple time points. We
genotyped 13 polymorphic microsatellites in four
Z. poliogaster populations collected at multiple time
points at three mountain blocks. The locations represent forests differing in size, geographical isolation
and habitat quality. Based on the data we (i) test for
potential differentiation among the local populations
and (ii) estimate the effective population size for each
population. We finally (iii) interpret our data against
the background of future conservation management.
MATERIAL AND METHODS
STUDY SPECIES
Zosterops poliogaster is a small passerine bird (Fig. 1)
which occurs in the mountain cloud forests of East
Africa. It requires cool and moist climatic conditions
and is hence restricted to distinct mountains
(> 1000 m, Moreau, 1957; > 850 m, Mulwa et al.,
2007). This disjunct distribution over long time
periods caused strong differentiation on molecular,
morphological and bioacoustic levels (Moreau, 1957;
Zimmerman et al., 1996; Redman et al., 2009; Habel
et al., 2013, 2014b; Cox et al., 2014; Husemann et al.,
2014). Some mountain populations have recently
taxonomically been treated as distinct subspecies or
Figure 1. Sampling sites in Kenya with a detailed view of the two Taita Hills fragments and a picture of the study
organism, Zosterops poliogaster photographed on Mt Kulal (J.C.H.).
© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, 114, 828–836
830
M. HUSEMANN ET AL.
even species, e.g. Z. p. silvanus or Z. silvanus in the
Taita Hills (Collar, Stattersfield & Crosby, 1994), Z. p.
mbuluensis in the Chyulu Hills (Zimmerman et al.,
1996), or Z. p. kulalensis or Z. kulalensis found at Mt
Kulal (Collar et al., 1994).
STUDY
REGION AND HABITAT
We sampled populations in four forest patches from
three mountain regions, including Mt Kulal in the
north and the Taita Hills and Chyulu Hills in the
south (Fig. 1). The habitats differ in size and spatial
configuration. The Taita Hills constitute a mountain
massif divided into two sections by a 5-km-wide
valley, the larger Dabida section with the forest patch
‘Ngangao’ and the smaller Mbololo section with the
forest patch ‘Mbololo’ (Fig. 1, inset). The cloud forest
of the Taita Hills has been heavily deforested and
disturbed in recent decades (Pellikka et al., 2009).
The Chyulu Hills, located 75 km north-west of the
Taita Hills, form a comparatively large mountain
range which is covered by relatively intact cloud
forest. The third mountain range, Mt Kulal in the
north of Kenya, still harbours large patches of wellpreserved cloud forest (Wass, 1995). This mountain
is highly isolated from other populations of
Z. poliogaster (closest in the Central Kenyan Highlands) (Zimmerman et al., 1996). As Z. poliogaster
also occurs in secondary habitats such as gardens
and exotic tree plantations outside of pristine
cloud forest (J.C.H., L.L. and L.B., pers. observ.),
we calculated the spatial extent of the required
moist and cool climatic niche available above 850 m,
where Z. poliogaster could be observed in previous
studies (Mulwa et al., 2007) (in the following called
‘suitable habitat’). Details of habitat characteristics
and GPS coordinates for each forest patch are given
in Table 1.
MOLECULAR
ANALYSES
We sampled 213 individuals of Z. poliogaster that had
been collected between 1938 and 2010 using mistnets. The individuals were released at the site of
capture (except for museum specimens collected in
1938 in the Chyulu Hills). Sample size ranged from
10 to 35 individuals per population and the respective
time point. Temporal cohorts are separated by 13
years or more. Feathers or blood samples were stored
dried under dark and cool conditions, in pure ethanol
or dimethyl sulphoxide at −20 °C or at room temperature until DNA extraction. Museum specimens collected in the Chyulu Hills were kindly provided by the
National Museums of Kenya, Nairobi. From these we
used toe pads that are known to yield high-quality
DNA (Fulton, Wagner & Shapiro, 2012). Specific information on the type of sample – feather, blood, tissue
and storage conditions – are given for each individual
in the Supporting Information, Table S1.
DNA extraction was performed using a Qiagen
DNeasy Tissue Extraction Kit following the manufacturer’s protocol for blood and tissue, as well as the
user-developed protocol for feathers (De Volo et al.,
2008). For PCR reactions we used the Thermozym
Mastermix (Molzym). PCR was carried out with a
thermal cycler (Corbett Research, CG1-96) under
primer-specific conditions. The following 13 primers
were used: Cu28, LZ44, LZ41, LZ22, LZ45, LZ14,
LZ54, LZ4, LZ35, Mme12, LZ18, LZ50 and LZ2
(Habel et al., 2013, 2014a, b). Further information on
DNA amplification and detection of fragment lengths
are given in Habel et al. (2013). MicroChecker v.2.2.3
(Van Oosterhout et al., 2004) was used to test the
quality of the genetic data set to check for genotyping
errors due to stutter bands, null alleles and large
allele dropout. We tested for deviations from Hardy–
Weinberg equilibrium and linkage disequilibrium
Table 1. Characteristics of sampled forest habitats with indication of forest size, area located over 850 m (‘suitable
habitat’, for details see Material and Methods), geographical distance to the next Z. poliogaster population, sample size
and sampling year (in parentheses)
Mountain
Location
GPS
Habitat size
(forest ha)
Space (km2)
above 850 m
Taita Hills –
Dabida,
Ngangao
Taita Hills –
Mbololo
3°21′S,
38°22′E
92
38.2
3°21′S,
38°25′E
220
14.1
Chyulu Hills
2°41′S,
37°54′E
2°41′N,
36°56′E
2000
293.7
2000
67.3
Mt Kulal
Distance to closest adjoining
population (km)
Nc
N (year)
12 km from Mbololo but the
nearest population is Vuria
(8 km away)
12 km from Ngangao
478
29 (1990)
21 (2009)
651
75 km from the Taita Hills
n.a.
300 km from Mt Kenya
8000–28 000
20
10
31
23
35
14
30
(1990)
(2000)
(2009)
(1938)
(2010)
(1997)
(2010)
© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, 114, 828–836
FRAGMENTATION GENETICS OF A MOUNTAIN BIRD
using Arlequin v.3.5.1.2 (Excoffier & Lischer, 2010).
Raw data are given in the Supporting Information,
Table S2.
To partition genetic variance among populations,
i.e. mountain ranges, between temporal cohorts and
among individuals within populations as well as
within individuals we conducted non-hierarchical
and hierarchical analyses of molecular variance
(AMOVAs).
Individual-based assignment tests were performed
with the program Structure v.3.1 (Hubisz et al., 2009).
With this approach we are able to detect the most
probable number of genetic clusters without a priori
definition. We carried out a total of 90 runs (10 each
for one to nine clusters), i.e. K = 1 to 9 (the maximal
number of cohorts). For each run, burn-in and simulation lengths were 300 000 and 500 000, respectively.
We calculated the ad-hoc statistic ΔK, based on the
rate of change in the log probability of data between
successive K values to determine the best K (Evanno,
Regnaut & Goudet, 2005).
We employed both single sample and temporal
approaches to estimate Ne from the genetic data. The
single sample method used is based on linkage disequilibrium and adjusts for missing data (LDNe
implemented in NeEstimator v.2.01, Do et al., 2013).
The temporal methods applied all use multiple time
points but have different assumptions and use different algorithms: (1) TM3 assumes closed populations,
is based on coalescence and uses Bayesian statistics
to calculate the likelihoods of specific Ne (Berthier
et al., 2002); (2) TempoFS estimates genetic drift
between temporally spaced samples using the Fs
measure of allele frequency change (Jorde & Ryman,
2007); (3) MLNe (Wang & Whitlock, 2003) uses a
maximum-likelihood approach to estimate drift
between temporally spaced populations – here we
used the option of no gene flow as populations are
strongly isolated (Habel et al., 2014a, b). Lastly, we
employed the two moment-based estimators implemented in (4) NeEstimator (Pollak, 1983) and (5)
MLNe. As all approaches have different assumptions
and may perform differently in specific situations
(Barker, 2011; Holleley et al., 2014), we calculated the
harmonic means of all temporal estimates (excluding
infinite estimates) as our final estimate for Ne for
each population, as recommended by Waples (2005)
and Johnstone et al. (2012).
RESULTS
Patch size based on the spatial extent of the required
moist and cool climatic niche (see above) and geographical isolation differed among the three studied
mountain ranges. The Chyulu Hills and Mt. Kulal
both harboured large forest patches (2000 ha each)
831
and the largest extent of climatically suitable habitat
(Chyulu Hills: 293.7 km2, Mt. Kulal: 67.3 km2).
Within the Taita Hills, Ngangao forest was characterized by the smallest patch size (92 ha) and an
intermediate extent of surrounding climatically suitable habitat (38.2 km2), whereas Mbololo forest was
characterized by an intermediate patch size (220 ha),
but the smallest extent of surrounding climatically
suitable habitat (14.1 km2) (Table 1).
We found no significant linkage disequilibrium and
deviations from Hardy–Weinberg equilibrium for any
of the loci, and only marginal effects due to null
alleles and/or large allele dropout for loci ZL2
(Ngangao population), ZL4 (Ngangao and Mt. Kulal
population) and ZL18 (Ngangao population) in both
temporal cohorts. AMOVAs showed that the majority
of genetic variance was attributed to among-group
variation, with a high proportion of the molecular
variance explaining the differentiation among the
three mountain regions (1.1283; FCT = 0.4751,
P < 0.0001); the strongest divergence was found
between the geographically adjoining Taita Hills and
Chyulu Hills (1.3467; FCT = 0.5352, P < 0.0001), followed by the split between the two most distant
mountain ranges Taita Hills and Mt. Kulal (1.1400;
FCT = 0.4734, P < 0.0001). The smallest divergence
was detected between the two geographically strongly
isolated mountain ranges Chyulu Hills and Mt. Kulal
(0.5590; FCT = 0.2952, P < 0.0001). The two populations sampled in the Taita Hills (Dabida and Mbololo)
showed low, but significant, genetic differentiation
(0.0490; FCT = 0.0417, P < 0.01) (Table 2A). Except for
the population from Mt. Kulal, no genetic divergence
was detected between temporal cohorts (Table 2B).
Population-specific FIS values showed no differences
between time cohorts (Table 2C).
The results from the Bayesian Structure analyses
are in congruence with our AMOVA results. We
detected a first split between the Taita Hills and all
other mountain clusters (for K = 2) (data not shown).
A structuring into three mountain clusters (K = 3)
was supported by a high ΔK value (Fig. 2, Table 3).
The estimates of local Ne differed strongly between
populations (Table 4). However, the variation
obtained from various estimators for the same population was large. The Chyulu Hills population was
estimated to be largest with a harmonic mean of
almost 15 000 individuals. However, this population
is probably larger as most estimators yielded infinite
Ne estimates. The Ngangao population of the Taita
Hills was estimated to be second largest with a harmonic mean of ∼540 individuals. Here the estimates
ranged between ∼300 individuals for the momentbased methods and infinite for the single sample
LDNe estimator. The Mbololo (Taita Hills) and Mt.
Kulal populations had comparatively small and
© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, 114, 828–836
832
M. HUSEMANN ET AL.
Table 2. Genetic differentiation at different spatial scales as estimated from non-hierarchical and hierarchical AMOVAs
– variance values are given in the top line with respective F statistics in parentheses below; A, hierarchical variance
analyses based on pre-defined mountain clusters; B, non-hierarchical variance analyses testing for potential differentiation between time cohorts within populations; C, population-specific molecular variance found among individuals within
populations at specific time points
A, hierarchical variance analyses – spatial clusters
Among
populations (FCT)
Group
Taita Hills vs. Chyulu Hills vs. Mt. Kulal
Taita Hills vs. Chyulu Hills
Taita Hills vs. Mt. Kulal
Chyulu Hills vs. Mt. Kulal
Taita Hills Dabida vs. Mbololo
1.1283
(0.4751***)
1.3467
(0.5352***)
1.1400
(0.4734***)
0.5590
(0.2952***)
0.0490
(0.0417**)
Among populations
within groups (RSC)
Within
individuals
0.0362
(0.0291***)
0.0332
(0.0284***)
0.0503
(0.0397***)
0.0140
(0.0105)
0.0187
(0.0166)
1.0657
0.9645
1.1355
1.1275
1.0090
B, non-hierarchical variance analyses – temporal clusters
Population
Taita Hills – Ngangao
Taita Hills – Mbololo
Chyulu Hills
Mt. Kulal
Among
populations (FST)
Among individuals
within populations (FIS)
Within
individuals
0.0153
(0.0134)
0.0205
(0.0185)
−0.0148
(−0.0126)
0.0558
(0.0360*)
0.0304
(0.0269)
0.1557
(0.1428***)
0.3094
(0.2603***)
0.0399
(0.0267)
1.1000
0.9344
0.8793
1.4545
C, non-hierarchical variance analyses – within single cohorts
Population
Taita Hills – Ngangao – 1990
Taita Hills – Ngangao – 2000
Taita Hills – Ngangao – 2009
Taita Hills – Mbololo – 1990
Taita Hills – Mbololo – 2009
Chyulu Hills – 1938
Chyulu Hills – 2010
Mt. Kulal – 1997
Mt. Kulal – 2010
FIS
P
0.09810
−0.09029
0.11765
0.10000
−0.06195
0.16903
0.29035
−0.20168
0.09270
0.144673
0.850440
0.125122
0.450635
0.765396
0.067449
< 0.000001
0.992180
0.099707
*P < 0.05; **P < 0.01; ***P < 0.001.
similar harmonic means with 168 and 107 individuals, respectively. The variation around the mean for
Mbololo was again large with point estimates
between 52 and 911 individuals and low for Mt. Kulal
with estimates ranging between 73 and 145 individuals. However, no infinite values of point estimates
were obtained for either population. Large confidence
intervals around the point estimates further reduced
the power of these results. Yet, we feel that it is still
worth presenting all results rather than biased
picking of one estimator. No consistent deviation from
the means was found for any of these methods.
© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, 114, 828–836
EVOLUTION ON MOUNTAIN TOPS
AMOVA and Bayesian Structure analysis indicate
highly distinct genetic clusters for each mountain
range (Fig. 2). Levels of genetic divergence, however,
do not correspond with geographical distances among
populations: the Taita Hills are geographically close to
the Chyulu Hills (about 80 km), but are genetically
highly distinct from all other population clusters. In
14 818.16
106.93
∞ (50.5–∞)
136.3 (19.8–∞)
38250 (548.81–∞)
116.99 (48.29–620.16)
∞
137.02
9189.4 (333–10000)
73.3 (42.2–130.6)
∞ (685.9–∞)
144.9 (45.4–1189.3)
∞
82 (56–160)
168.02
51.9 (9.6–∞)
911.28 (194.72–∞)
149.19
319 (57–∞)
819 (0–10000)
254.3 (74–2727.4)
539.36
Single sample
∞ (227.9–∞)
Temporal
38249 (389.52–∞)
Temporal
281.03
Temporal
415 (129–∞)
Temporal
5280.9 (320.1–10000)
MLNe
LDNe
Ne
harmonic
mean
Momentbased
MLNe
© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, 114, 828–836
Infinite estimates were excluded from harmonic mean calculation. Values in parentheses are 95% confidence intervals.
DISCUSSION
DISTINCT
1990 (20), 2000 (10),
2009 (31)
1950 (23), 2010 (35)
1997 (14), 2010 (30)
ln(Pr) is the natural logarithm of the probability calculated for each value of K with Structure. SD is the standard deviation calculated from ten independent runs. The
bold value represents the value of K with the highest
likelihood. The ad-hoc statistic ΔK is not applicable for
K = 1, cannot be calculated for the highest K number and
is not proper for K = 2 (hence, values in parentheses; cf.
Hausdorf & Hennig 2010).
Temporal
324.6 (83–∞)
–
(3316.35)
7.09
0.86
0.65
0.32
0.28
0.25
–
1990 (29), 2009 (21)
−3352.69 ± 0.06
(−2363.8 ± 0.23)
−2140.79 ± 139.24
−2037.02 ± 130.65
−2017.78 ± 137.30
−2972.81 ± 3256.62
−2883.46 ± 2953.55
−1967.06 ± 36.30
−1958.84 ± 73.55
Method
Taita Hills –
Ngangao
Taita Hills –
Mbololo
Chyulu Hills
Mt Kulal
1
2
3
4
5
6
7
8
9
TempoFS
ΔK
TM3
ln(Pr) (± SD)
Temporal samples
K
Moment-based
NeEstimator
Table 3. Results of the Structure analysis for different
numbers of given groups (K = 1–9) analysed based on all
individuals
Approach/
population
Figure 2. Bayesian structure analyses of populations
from the mountain white-eye Zosterops poliogaster performed for K = 1–9. The result for K = 3 supported by a
high ΔK value is presented, distinguishing the mountain
populations of Taita Hills, Chyulu Hills and Mt Kulal. TH,
Taita Hills. Respective ΔK values are given in Table 3.
Table 4. Ne estimates resulting from different temporal and single sample methods and the harmonic mean of all point estimates, with indication of the year
of sampling
FRAGMENTATION GENETICS OF A MOUNTAIN BIRD
833
834
M. HUSEMANN ET AL.
contrast, both the Taita and the Chyulu Hills are geographically distant from Mt. Kulal (> 600 km), but the
genetic split is much shallower than between the Taita
and Chyulu individuals. The strongest split was found
between the Taita Hills samples and all other populations analysed. This might reflect the geological ages
of these mountain blocks: the Taita Hills are part of
the geologically very old Eastern Arc Mountains, while
the Chyulu Hills and Mt. Kulal are geologically much
younger (White, 1983). The effects of geological ages
(and subsequent time spans available for evolutionary
processes) have been demonstrated in a previous
study on the family Zosteropidae (Cox et al., 2014).
EFFECTIVE
POPULATION SIZES
The population from Chyulu Hills was estimated to
have the highest effective population size, which was
expected given the comparatively large, stable and
fairly intact state of its mountain forest habitat. High
local breeding densities, as inferred from Ne, may
reflect a still intact forest habitat with high habitat
quality. The Chyulu Hills represent one of the most
intact mountain cloud forests of Kenya and have been
protected by National Park status for many years.
Populations from the nearby Taita Hills had much
lower effective population size estimates. While the
forest fragment of Mbololo is larger than that of
Ngangao, the latter forms part of a meta-population
network that comprises several other small indigenous forest remnants and exotic tree plantations
(Pellikka et al., 2009). In contrast, Mbololo is isolated
from other Taita forest fragments by a deep valley
and is also characterized by a smaller extent of surrounding area above 850 m (critical elevation threshold, Table 1, see above). Our estimates on the effective
population sizes in combination with the two contrasting spatial configurations of the two forest
patches (Ngagao, small but part of a forest patchwork; Mbololo, large but geographically isolated from
other forest patches) highlight the importance of
habitat connectivity to maintain high effective population sizes and genetic diversity.
Mt. Kulal still harbours large and intact stretches of
highland forest and is characterized by a large climatically suitable habitat (Table 1). Observations indicate
a large census population size of Z. poliogaster (L.B.).
However, our models yielded a rather low mean effective population size of Ne = 107. This can be explained
as being due to several non-exclusive factors: first, the
comparatively strong geographical isolation of Mt.
Kulal may have driven a loss of genetic information
due to genetic drift effects in combination with a lack
of genetic refreshment from migrating individuals
from other populations. However, the large census
population size should limit the effects of genetic drift.
Second, bottlenecks might have occurred in the past,
followed by rapid population expansion, potentially
explaining the deviation of effective and census population sizes. Third, the mating system and female
fecundity of a species can lead to large deviations of
census and effective population sizes (Nunney, 1996;
Ardren & Kapuscinski, 2003; Watts et al., 2007). If, for
example, variance in female fecundity is high, the
ratios of both population size estimates can become
increasingly small (Nunney, 1996). Yet, very little is
known about the breeding system and mate choice in
our study species. Fourth and lastly, the low effective
population size found for Z. poliogaster on Mt. Kulal
might be a result of ascertainment bias of the markers
or bias in the sampling of genotypes, i.e. due to low
sample sizes or non-independent samples (Palstra &
Ruzzante, 2008).
Sample sizes and intervals differed between populations and time points; yet, this supposedly has no
large effect on Ne estimation (Palstra & Ruzzante,
2008). In addition, some studies have indicated that,
for example, single sample estimators may underestimate the real Ne (Barker, 2011; Holleley et al., 2014).
Our data clearly indicate the importance of investigating multiple estimators of population size and if
possible use different sources of population data when
estimating the size of a population. If only a single
estimator is used, the results and interpretation may
be easier, but can be misleading and may show a
incorrect picture of the real status quo. Therefore,
while sometimes resulting in confusing variation, it is
important to consider different approaches in studies
of effective and census population sizes.
TRANSLATING
OUR DATA INTO CONSERVATION ACTION:
THE WHY AND HOW
Our data set highlights that almost each mountain
top harbours distinct local populations with unique
alleles (and specific contact calls and morphological
characters, Habel et al., 2013a, b; Husemann et al.,
2014). This suggests the existence of distinct evolutionary units and underlines the relevance of preserving each single population cluster, consistent with the
concept of Evolutionary Significant Units (Moritz,
1994) or the Conservation Units concept (Vogler &
Desalle, 1994; Fraser & Bernatchez, 2001). In this
example, conserving the birds’ genetic and phenotypic
variability means protecting all local occurrences. The
second part of our study, i.e. analyses of the effective
population sizes, underlines that connectivity plays a
pivotal role in the conservation of intraspecific variability in Z. poliogaster. Neither in the Mbololo forest
fragment nor on Mt. Kulal do the estimated effective
population sizes reflect the comparatively large
extent of forest habitats and large census population
© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, 114, 828–836
FRAGMENTATION GENETICS OF A MOUNTAIN BIRD
sizes, but both regions are highly affected as a result
of geographical isolation. Our data show that deviations from census population sizes and Ne can give
valuable insight into long-term trends of populations
and highlight the relevance of corridors when translating theory into conservation management.
ACKNOWLEDGEMENTS
This study was funded by the German Academic
Exchange Service (DAAD) (M.H. and J.C.H.). We
thank Titus Iboma, Onesmus M. Kioko and Ronald K.
Mulwa (NMK Nairobi, Kenya) for field assistance. We
are grateful for the comments of two anonymous
referees that helped to improve a previous version of
this article.
REFERENCES
Ardren WR, Kapuscinski AR. 2003. Demographic and
genetic estimates of effective population size (Ne) reveals
genetic compensation in steelhead trout. Molecular Ecology
12: 35–49.
Barker JSF. 2011. Effective population size of natural populations of Drosophila buzzatii, with a comparative evaluation of nine methods of estimation. Molecular Ecology 20:
4452–4471.
Berthier P, Beaumont MA, Cornuet JM, Luikart G. 2002.
Likelihood-based estimation of the effective population size
using temporal changes in allele frequencies: a genealogical
approach. Genetics 160: 741–751.
Charlesworth B. 2009. Effective population size and patterns of molecular evolution and variation. Nature Reviews
Genetics 10: 195–205.
Collar NJ, Stattersfield AJ, Crosby MJ. 1994. Birds to
watch 2. The world list of threatened birds. Birdlife Conservation Series No. 4. Cambridge: Birdlife International.
Cox S, Prys-Jones R, Habel JC, Amakobe B, Day J. 2014.
Recent independent colonisations of East African skyislands by montane white-eyes (Aves: Zosteropidae). Molecular Ecology 23: 4103–4118.
De Volo SB, Reynolds RT, Doublas MR, Antolin MF.
2008. An improved extraction method to increase DNA yield
from molted feathers. Condor 110: 762–766.
Do C, Waples RS, Peel D, Macbeth GM, Tillett BJ,
Ovenden JR. 2013. NeEstimator V2: reimplementation of
software for the estimation of contemporary effective population size (Ne) from genetic data. Molecular Ecology
Resources 1: 209–214.
Evanno G, Regnaut S, Goudet J. 2005. Detecting the
number of clusters of individuals using the software
STRUCTURE: a simulation study. Molecular Ecology 14:
2611–2620.
Excoffier L, Lischer HE. 2010. Arlequin suite ver 3.5:
a new series of programs to perform population genetics
analyses under Linux and Windows. Molecular Ecology 10:
564–567.
835
Fahrig L. 2003. Effects of habitat fragmentation on biodiversity. Annual Reviews on Ecology, Evolution and Systematics
34: 487–515.
Frankham R. 2005. Genetics and extinction. Biological Conservation 126: 131–140.
Fraser DJ, Bernatchez L. 2001. Adaptive evolutionary conservation: towards a unified concept for defining conservation units. Molecular Ecology 10: 2741–2752.
Fulton TL, Wagner SM, Shapiro B. 2012. Case study:
recovery of ancient nuclear DNA from toe pads of the extinct
passenger pigeon. Methods in Molecular Biology 840: 29–35.
Habel JC, Cox S, Gassert F, Meyer J, Lens L. 2013.
Population genetics of four East African Mountain Whiteeye congenerics. Conservation Genetics 14: 1019–1028.
Habel JC, Husemann M, Finger A, Danley PD, Zachos
FE. 2014a. The relevance of time series in molecular
ecology and conservation biology. Biological Reviews 89:
484–492.
Habel JC, Mulwa RK, Gassert F, Rödder D, Ulrich W,
Borghesio L, Husemann M, Lens L. 2014b. Population
signatures of large-scale, long-term disjunction and smallscale, short-term habitat fragmentation in an Afromontane
forest bird. Heredity 113: 205–214.
Hausdorf B, Hennig C. 2010. Species delimitation using
dominant and codominant multilocus markers. Systematic
Biology 59: 491–503.
Holleley CE, Nichols RA, Whitehead MR, Adamack AT,
Gunn MR, Sherwin WB. 2014. Testing single-sample estimators of effective population size in genetically structured
populations. Conservation Genetics 15: 23–35.
Horreo LJ, Palacin C, Alonso JC, Mila B. 2013. A link
between historical population decline in the threatened
great buzzard and human expansion in Iberia: evidence
from genetic and demographic data. Biological Journal of
the Linnean Society 110: 518–527.
Hubisz MJ, Falush D, Stephens M, Pritchard JK. 2009.
Inferring weak population structure with the assistance of
sample group information. Molecular Ecology Resources 9:
1322–1332.
Husemann M, Ulrich W, Habel JC. 2014. Effects from
distinct and overlapping occurrences in two East African
White-eye congeners on the evolution of their bird calls.
BMC Evolutionary Biology 14: 115.
Johnstone DL, O’Connell MF, Palstra FP, Ruzzante DE.
2012. Mature male pall contribution to the effective size of
an anadromous Atlantic salmon (Salmo salar) population
over 30 years. Molecular Ecology 22: 2394–2407.
Jorde PE, Ryman N. 2007. Unbiased estimator for
genetic drift and effective population size. Genetics 177:
927–935.
Luikart G, Ryman N, Tallmon DA, Schwartz MK,
Allendorf FW. 2010. Estimation of census and effective
population sizes: the increasing usefulness of DNA-based
approaches. Conservation Genetics 11: 355–373.
Moreau RE. 1957. Variation in the Western Zosteropidae
(Aves). Bulletin of the British Museum 4: 309–433.
Moritz C. 1994. Defining ‘evolutionary significant units’ for
conservation. Trends in Ecology and Evolution 9: 373–375.
© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, 114, 828–836
836
M. HUSEMANN ET AL.
Mulwa RK, Bennun LA, Ogol CPK, Lens L. 2007. Population status and distribution of Taita White-eye Zosterops
silvanus in the fragmented forests of Taita Hills and Mount
Kasigau, Kenya. Bird Conservation International 17: 141–
150.
Nunney L. 1996. The influence of variation in female fecundity on effective population size. Biological Journal of the
Linnean Society 59: 411–425.
Palstra FP, Ruzzante DE. 2008. Genetic estimates of contemporary effective population size: what can they tell us
about the importance of genetic stochasticity for wild
population persistence. Molecular Ecology 17: 3428–
3447.
Pellikka PKE, Lötjönen M, Siljander M, Lens L. 2009.
Airborne remote sensing of spatiotemporal change (1955–
2004) in indigenous and exotic forest cover in the Taita
Hills, Kenya. International Journal of Applied Earth Observational Geoinformatics 11: 221–232.
Phillipsen IC, Funk WC, Hoffman EA, Monsen KJ,
Blouin MS. 2011. Comparative analyses of effective population size within and among species: ranid frogs as a case
study. Evolution 65: 2927–2945.
Pollak E. 1983. A new method for estimating the effective
population size from allele frequency changes. Genetics 104:
531–548.
Redman N, Stevenson T, Fanshawe J. 2009. Birds of the
Horn of Africa: Ethiopia, Eritrea, Djibouti, Somalia, and
Socotra. Princeton Field Guide 2009, London.
Van Oosterhout C, Hutchinson WF, Wills DPM, Shipley
P. 2004. Micro-checker: software for identifying and correcting genotyping errors in microsatellite data. Molecular
Ecology Notes 4: 535–538.
Vogler AP, Desalle R. 1994. Diagnosing units of conservation management. Conservation Biology 8: 354–363.
Wang J, Whitlock MC. 2003. Estimating effective population
size and migration rates from genetic samples over space
and time. Genetics 163: 429–446.
Waples RS. 2005. Genetic estimates of contemporary effective
population size: to what time periods do estimates apply?
Molecular Ecology 14: 3335–3352.
Wass P. 1995. Kenya’s indigenous forests. Status, management and conservation. Gland: The World Conservation
Union.
Watts PC, Saccheri IJ, Kemp SJ, Thompson DJ. 2007.
Effective population sizes and migration rates in fragmented populations of an endangered insect (Coenagrion
mercuriale: Odonata). Journal of Animal Ecology 76: 790–
800.
White F. 1983. The vegetation of Africa: a descriptive memoir
to accompany the UNESCO/AETFAT/UNSO vegetation
map of Africa. Paris, France: United Nations Scientific and
Cultural Organization.
Wright S. 1938. Size of a population and breeding structure
in relation to evolution. Science 87: 430–431.
Zimmerman DA, Turner DA, Person DJ. 1996. Birds of
Kenya and Northern Tanzania. London: Christopher Helm.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the online version of this article at the publisher’s web-site:
Table S1. Details about each sampled individual used in our study. Given are the location, year of sampling,
source from where DNA was extracted, conservation conditions after sampling, collector and an individual
abbreviation for each sample.
Table S2. Raw data used for the analyses. Given are location with year, abbreviation of the respective sample
(coinciding with Table S1) and the 13 polymorphic microsatellites (fragment length of allele A and B).
© 2015 The Linnean Society of London, Biological Journal of the Linnean Society, 2015, 114, 828–836