Available via license: CC BY 4.0
Content may be subject to copyright.
Ecology a nd Evolution . 201 8 ;1–1 3 .
|
1
www.ecolevol.org
1 | INTRODUCTION
Morphological and genetic studies have strongly suppor ted rec-
ognition of t wo African elephant species: the African savanna ele-
phant (Loxodonta africana) and African forest elephant (L. cyclotis)
(Comstock et al., 2002; Groves & Grubb, 2000; Ishida et al., 2011;
Roca, Georgiadis, Pecon- Slattery, & O’Brien, 2001; Rohland et al.,
2010). While many studies have indicated that the forest elephant is
a species distinct from the savanna elephant, the analysis of genetic
diversit y below the species level has been limited. Mitochondrial
DNA (mtDNA) patterns have been examined in African forest el-
ephants across their range (Debruyne, 2005; Debruyne, Van Holt,
Barriel, & Tassy, 2003; Eggert, Rasner, & Woodruff, 2002; Ishida,
Georgiadis, Hondo, & Roca, 2013; Johnson et al., 2007; Nyakaana,
Arctander, & Siegismund, 2002). Five distinct mitochondrial sub-
clades have been detected among forest elephants, each of which
Received:15September2017
|
Revised:15March2018
|
Accepted:19March2018
DOI: 10.100 2/ece3.4062
ORIGINAL RESEARCH
Evolutionary and demographic processes shaping geographic
patterns of genetic diversity in a keystone species, the African
forest elephant (Loxodonta cyclotis)
Yasuko Ishida1 | Natalie A. Gugala1 | Nicholas J. Georgiadis2 | Alfred L. Roca1,3
This is an op en access article under t he terms of t he Creat ive Commons Attr ibutio n License , which pe rmits u se, dist ributi on and rep roduc tion in any m edium,
provide d the orig inal work is proper ly cited.
© 2018 The Aut hors. Ecology an d Evolution pu blished by John Wiley & Sons Ltd .
1Department of Animal Sciences, University
of Illinois at Urbana-Champaign, Urbana,
Illinois
2Puget Sound Institute, Universit y of
Washington, Tacoma, Washington
3Carl R . Woese Ins titute for Genomic
Biolog y, Universi ty of Illinois at Urb ana-
Champaign, Urbana, Illinois
Correspondence
Alfred L . Roca and Yasuko Ishida ,
Department of Animal Sciences, University
of Illinois at Urbana-Champaign, Urbana, IL.
Emails: roca@illinois.edu; yishida@illinois.
edu
Funding information
USFWS African Elephant Conservation
Fund, Grant/Award Number: AFE-0778-
F12AP01143andAFE1606-F16AP009 09
Abstract
The past processes that have shaped geographic patterns of genetic diversity may be
difficult to infer from current patterns. However, in species with sex differences in
dispersal, differing phylogeographic patterns between mitochondrial (mt) and nu-
clear (nu) DNA may provide contrasting insights into past events. Forest elephants
(Loxodonta cyclotis) were impacted by climate and habitat change during the
Pleistocene, which likely shaped phylogeographic patterns in mitochondrial (mt)
DNA that have persisted due to limited female dispersal. By contrast, the nuclear (nu)
DNA phylogeography of forest elephants in Central Africa has not been determined.
We therefore examined the population structure of Central African forest elephants
bygenot yping94individualsfromsixlocalitiesat21microsatelliteloci.Betweenfor-
est elephants in western and eastern Congolian forests, there was only modest ge-
netic differentiation, a pattern highly discordant with that of mtDNA. Nuclear genetic
patterns are consistent with isolation by distance. Alternatively, male- mediated gene
flow may have reduced the previous regional differentiation in Central Africa sug-
gested by mtDNA patterns, which likely reflect forest fragmentation during the
Pleistocene. In species like elephants, male- mediated gene flow erases the nuclear
genetic signatures of past climate and habitat changes, but these continue to persist
as patterns in mtDNA because females do not disperse. Conservation implications of
these results are discussed.
KEYWORDS
Congolian forest block, conservation, isolation by distance, landscape genetics, microsatellites
2
|
ISHIDA et Al.
has a different geographically restricted distribution (Ishida et al.,
2013). However, several factors can lead to discordant patterns in
the phylogeography of nuclear and mitochondrial DNA markers,
both with in and across spe cies (Petit & Exco ffier, 2009; Toews &
Brelsford, 2012), and in elephant s, there is evidence that female
philopatry and male- biased dispersal combine to produce incon-
gruent mitonuclear patterns (Debruyne, 2005; Roca, Georgiadis, &
O’Brien, 2005).
Field studies have shown strong evidence that, despite living in
a fission–fusion society, female elephants remain with their clos-
est kin after they mature (Archie, Moss, & Alberts, 2011). Genetic
analyses have suppor ted female philopatry by demonstrating al-
most complete uniformity of mtDNA haplotypes within families
(Archie, Fitzpatrick, Moss, & Alberts, 2011). By contrast, male el-
ephants upon reaching maturity disperse from their natal herds
(Lee, Poole, Njiraini, Sayialel, & Moss, 2011) and enter periods of
musth characterized by competitive interactions with other males
for reproductive access to females (Poole, Lee, Njiraini, & Moss,
2011). The dispersal of male elephants from their natal social groups
thus mediates nuclear gene flow (Ishida et al., 2011; Nyakaana &
Arctander,1999;Rocaetal.,2005,2015).Thephylogeographicpat-
terns revealed by analyses of Y- chromosome sequences is similar to
the pattern for other nuclear markers, but different from patterns
shown by mtDNA, supporting the role of males in establishing nu-
clear phylogeographic patterns (Roca, Georgiadis, & O’Brien, 2007;
Roca et al., 2005).
Because mitochondrial phylogeographic patterns are often
discordant from nuclear patterns in species in which only males
disperse(Petit&Excoffier,2009;Toews&Brelsford,2012),includ-
ing elephants (Roca et al., 2005, 2007), there is a strong need to
analyze nuclear markers among forest elephants to examine more
completely their evolutionary history and population structure.
Furthermore, Central Africa, the region in which most forest ele-
phants live, has suffered from the highest levels of elephant poach-
ing of any subregion within the continent (CITES, 2012), and has
been the main source for the illegal trade in elephant bushmeat
and ivory (Wasser et al., 2015). Forest elephant numbers declined
by ca. 62% bet ween 2002 and 2011, to <10% of their estimated
historical population size, mainly due to illegal poaching for their
tusks (Maisels et al., 2013). There is thus a strong need to exam-
ine fine- scale population substructure within forest elephants
using nuclear markers, for proper conservation management of the
species.
Here, we use microsatellite markers to examine nuclear genetic
structure in the forest elephant. Ninety- three individuals from five
localities in Central Africa and one individual from Sierra Leone were
genotyped using 21 microsatellite markers. We examined nuclear
genetic markers for geographic differences among forest elephant
localities. We discuss the extent to which regional populations may
or may not be genetically distinctive, and the implications of these
findings for forest elephant conservation. We also specifically ex-
amine the degree of discordance between the phylogeographic pat-
terns inferred using microsatellite markers and patterns previously
reported for forest elephant mtDNA across the same tropical forest
localities within Central Africa.
2 | MATERIALS AND METHODS
2.1 | Samples
This stu dy was conducte d under the Univer sity of Illinois In stitutional
Animal Care and Use Committee (IACUC)- approved protocol num-
ber 15053. Samples were collected in full compliance with required
Convention on International Trade in Endangered Species of Wild
Fauna and Flora and other institutional permits. Wild African for-
est elephants (L. cyclotis) were sampled from six localities (Figure 1).
Tissue samples were collected primarily by biopsy darting from ele-
phants in Lope (LO) in Gabon, Odzala (OD) in the Republic of Congo,
Dzanga Sangha (DS) in the Central African Republic, and Garamba
(GR) in the Democratic Republic of Congo. Dung samples of el-
ephants were collected from the Bili Forest (BF) in the Democratic
Republic of Congo. A blood sample was obtained from a forest
FIGURE1 The map shows the sampling locations of forest elephants. Abbreviations are as follows: DS—Dzanga Sangha, Central African
Republic; OD—Odzala, Republic of Congo; BF—Bili Forest, Democratic Republic of Congo; LO- Lope, Gabon; and SL—Sierra Leone (one zoo
individual). GR—Garamba in Democratic Republic of Congo is located in the Guinea–Congolian/Sudanian transition zone of vegetation
(Olson et al., 2001) that historically included a mix ture of forest and secondary grasslands suitable for both African elephant species (Groves
& Grubb, 2000)
DS
OD
LO
GR
BF
SL
1,000 km
|
3
ISHIDA et A l.
elephant kept at the Paris Zoo that originated in Sierra Leone (SL).
These localities represent different geographic regions: Sierra Leone
(SL) is located in West Africa; the others are in the Congolian forest
block, with LO, OD, and DS to the west, and BF and GR to the east
(Figure 1).
In total, 94 fo rest elepha nts from six lo calities (Fig ure1) were
successfully genot yped at the 21 microsatellite loci (SL: n = 1, LO:
n = 15, OD: n = 3, DS: n = 53, BF: n = 3, GR: n = 19)(Tables1andS1).
As only one sample was available from SL, it was not included in
some statistical analyses. Garamba has historically included mixed
forest and savanna habitats suitable for both species of African ele-
phant (Groves & Grubb, 2000). Most of our samples from Garamba
are forest elephants, although a few are hybrids of savanna and for-
est elephants based on nuclear genotypes (Comstock et al., 2002;
Groves & Grubb, 200 0; Ishida et al., 2011; Roca et al., 2001; Rohland
et al., 2010).
In addition to the forest elephants, 15 African savanna ele-
phants (L. africana) were genotyped, one each from 15 localities
(CH—Chobe and MA—Mashatu in Botswana; BE—Benoue and
WA—Waza in Cameroon; AB—Aberdares, AM—Amboseli, and KE—
Central Kenya/Laikipia in Kenya; NA—Northern Namibia/Etosha;
KR—Kruger in South Africa; NG–Ngorongoro, SE–Serengeti, and
TA—Tarangire in Tanzania; HW—Hwange, SW—Sengwa, and ZZ—
Zambezi in Zimbabwe). Savanna elephant samples were included
in light of previous findings that forest and savanna elephants are
genetically distinct species with a narrow region of hybridization
(Comstock et al., 20 02; Groves & Grubb, 20 00; Ishida et al., 2011;
Roca et al., 2001; Rohland et al., 2010). Because our samples con-
tain forest elephants from Garamba where a few hybrids of sa-
vanna and forest elephants have been identified based on nuclear
genotypes (Comstock et al., 2002; Ishida et al., 2011; Mondol
et al., 2015; Roca et al., 20 01), it was determined that the micro-
satellites would be amplified in savanna elephants, as they would
be needed to identif y hybrids of savanna and forest elephants.
Details on the sampling and DNA extraction have been previously
published (Ishida et al., 2011).
2.2 | Microsatellite genotyping
Allelic variation was examined at 21 microsatellite loci. These
markers have been previously developed by Gugala et al. (Gugala,
Ishida, Georgiadis, & Roca, 2016). Primer sequences are listed in
Table S1. All forward primers included the M13 forward sequence
(TGTAA AACGACGGCCAGT ) at the 5′ end. The PCR pr imer mix
consiste d of a 5′ FAM- or VIC-fluorescent-labeled M13 for ward
primer (to label the PCR amplicon), along with the forward primer
(with M13 forward sequence at the 5′ end) and reverse prim-
ers. The PCR mix included 1× PCR buffer II (Life Technologies,
Carlsbad, CA , USA), 2 mmol/l MgCl2, 200 μmol/l of each dNTP (Life
Technologies) with 0.04 units/μl final concentration of AmpliTaq
Gold DNA Polymerase (Life Technologies) along with 1.2 μl of the
primer mix. For the DNA samples from BF that had been extrac ted
from dung, 1 μg/μl final concentration of bovine serum albumin
(New England BioLabs Inc.) was also included. The PCR cycling pro-
gram consi sted of an initial 95°C fo r 10min; with cycles of 15s
denaturingat95°C,followedby30sannealingat60,58,56,54,or
52°C(two cycleseachtemperature); or 50°C(last 30 cycles),fol-
lowedby45s extensionat72°C;withafinalextension of30min
at72°C.For locus Lcy-M45,the PCR cycling program was modi-
fied as described previously (Gugala et al., 2016). PCR amplicons
were visualized on a 1.5%–2% agarose gel with ethidium bromide
under ultraviolet light. Amplicons of two different loci labeled with
different fluorescent dyes (FAM and VIC) were diluted and mixed
depending on the intensity of the signal on the agarose gel photo-
graph. Fragment analysis was conducted on the ABI 3730XL capil-
lary sequencer at the University of Illinois at Urbana- Champaign
High- Throughput Sequencing and Genotyping Unit. The soft ware
Genemapper version 3.7 (Life Technologies) was used to call alleles.
Relying on a standard of known size, the binning function of the
software Genemapper was used to determine fragment lengths,
following the procedures indicated in the manual. For the DNA
samples extracted from dung (BF), at least four independent ampli-
fications were repeated to confirm homozygotes and three amplifi-
cations for heterozygotes (Allentoft et al., 2011).
2.3 | Characterization of microsatellites
Arlequin version 3.5.1.3 (Excoffier & Lischer, 2010) and GenAlEx 6.5
(Peakall & Smouse, 2012) were used to calculate expected heterozy-
gosity (He) and observed heterozygosit y (Ho). The sof tware GenAlEx
6.5 was also used to calculate Shannon’s diversity index (I) and to
make allele f requency distribution histograms for each locus for each
locality. Tests for Hardy–Weinberg equilibrium (HWE) and linkage
disequilibrium (LD) were conducted on the forest elephant micro-
satellite data. A Markov chain algorithm was used to test for HWE
using 10,000 dememorization steps, 1,00 0 batches, and 10,000 it-
erations per batch using the software Genepop 4. 2 (Rousset, 2008),
and 1,000,0 00 steps and 100,000 dememorization steps were used
with Arlequin version 3.5.1.3. LD was examined using 10,000 de-
memorization steps, 1,000 batches and 10,000 iterations per batch
TABLE1 Comparison of pairwise FST values calculated using
nuclear and mitochondrial DNA
LO (17) OD (3) DS (54) BF (0) GR (20)
LO (15) 0.54 0.87 NA 0 .76
OD (3) 0.02 0.49 NA 0.38
DS (53) 0.02 0.02 NA 0.61
BF (3) 0.06 0.06 0.04 NA
GR(19) 0.05 0.07 0.03 0.00
Pairwise FST is shown for comparisons between localities using nuclear
microsatellites (below diagonal) and mtDNA (above diagonal). FST values
of mtDNA are from Ishida et al. (2013). The values that are significant are
indicated in boldface. Sample sizes are in parenthesis for mtDNA (top
row) and microsatellites (first column). Localities corresponding to the
abbreviations are shown in Figure 1. FST values calculated using mtDNA
are not shown for BF as comparable mtDNA sequences are not available
fo r B F.
4
|
ISHIDA et Al.
for each pairwise comparison between loci for Genepop 4.2 and
10,000 permutations for Arlequin version 3.5.1.3.
2.4 | Population genetic analyses
Analysis of molecular variation (AMOVA) was conducted in Arlequin
version 3.5.1.3 (Excof fier & Lischer, 2010) using 10,0 00 permuta-
tions. For each pair of localities, Arlequin was also used to calculate
pairwise FST values and statistical support using 10,00 0 permuta-
tions, and to calculate the inbreeding coefficient for each locality,
except SL, as only one sample was available from SL.
STRUCTURE 2.3.4 (Pritchard, Stephens, & Donnelly, 2000),
which applies a model- based clustering algorithm to multilocus gen-
otype data, was used to infer population structure using datasets
that included or excluded the savanna elephants. STRUCTURE was
run eight times for each value of K from 1–10, without the use of
prior information on localit y, under the admixture- correlated model,
with each iteration using at least 1 million Markov chain Monte
Carlo generations following a burn- in of at least 100,000 steps.
The uppermost hierarchical level of population structure was ex-
amined using an ad hoc statistic ∆K based on the rate of change in
the log probability of the data for a given K between successive K
values, implemented in Structure Harvester (Earl & vonHoldt, 2012).
Average coef ficients were estimated for each K value that was es-
timated to be uppermost, and for lower values of K, employing the
Greedy algorithm with 1,000 random input orders as implemented
in the program CLUMPP version 1.1.2 (Jakobsson & Rosenberg,
2007). These outputs were visualized using DISTRUCT version 1.1
(Rosenberg, 2004). After identif ying the hybrid elephants in GR, we
also conducted STRUCTURE analyses excluding them to remove
the influence of savanna elephant genotypes. We also conducted
STRUCTURE analyses to compare each pair of localities.
Factorial correspondence analyses (FCA) were implemented
in GENETIX version 4.05 (Belkhir, Borsa, Chikhi, Raufaste, &
Bonhomme, 200 4) to graphically plot the distribution of genetic
variation for each locality with forest elephants (excluding the hy-
brids). Principal coordinate analyses (PCoAs) were implemented in
GenAlEx 6.5 (Peakall & Smouse, 2012) to visualize the genetic rela-
tionships among individual elephants, both including and excluding
the savanna elephants and hybrid GR elephants.
2.5 | Isolation by distance
The coordinate information of each localit y was estimated using the
LatLong.net website (http://www.latlong.net/), and the pairwise
distances between each locality were calculated using the coordi-
nate calculators and distance tools in GPS Visualizer (http://www.
gpsvisualizer.com/calculators). To examine the relationship between
genetic distances and geographic distances among forest elephants
at the five localities in Central Africa, a spatial autocorrelation anal-
ysis was implemented in GenAlEx 6.5 (Peakall & Smouse, 2012).
The spatial autocorrelation analysis divided the pairwise distances
intofour ordinalclasses andused 9,999random permutationsand
9,999 bootstrap iterations. Isolation-by-distance (IBD) analyses
were conducted to test the relationships between genetic differ-
ences bet ween each pair of localities in Central Africa and the geo-
graphic distance between them. These analyses used the Isolation
By Distance Web Service Version 3.23 (Jensen, Bohonak, & Kelley,
2005). Two different measures of genetic distance were calculated:
FST and Rousset’s distance FST/(1- FST). Mantel tests were run with
30,000 randomizations (one- tailed test). For Slatkin’s similarity
index, we used the recommended log- transformation of both M and
geographic distance. As we had only one sample from West Africa
(from Sierra Leone, SL), we excluded this sample from analyses.
2.6 | Phylogenetic analyses
We inferred the phylogenetic relationships among localities using
the neighbor–joining (NJ) method implemented in POPTREEW
(Takezaki, Nei, & Tamura, 2014). As the number of samples was
different among localities, we used DST (Nei, 1972)andFST (Latter,
1972)withsamplesizebiascorrec tioninadditiontoDA (Nei, Tajima,
&Tateno,1983)(forwhichsamplesizebiascorrectionwasnotavail-
able) to calculate genetic distances among localities. Support for the
nodes in each analysis was assessed using 10,0 00 bootstrap pseu-
doreplicates. To exclude the influence of savanna elephant geno-
types on GR, forest–savanna hybrid elephants were not included
in these analyses. The program FigTree v1.4.2 (Rambaut, 2014) was
used to draw trees. To assess the influence of the small sample size
of SL (n = 1), we also conducted additional analyses using only one
sample from an alternative location (LO). The single LO sample was
chosen randomly by RESEARCH RANDOMIZER (https://www.rand-
omizer.org/). Three iterations were run in which a single sample from
LO was chosen at random, and the NJ tree was reconstruc ted with
the single LO sample.
3 | RESULTS
3.1 | Characterization of forest elephants using
microsatellites
Although 21 microsatellite markers were genotyped, three were re-
moved before analyses. Marker Lc y- M4 had a low genotyping suc-
cess rate due potentially to null alleles. In marker Lcy- M15, a 1- bp
indel was detected in some savanna elephants; this marker was
removed from the forest elephant analyses as some elephants in
Garamba are savanna–forest elephant hybrids. The marker Lcy- M52
showed a significant deviation from HWE even after Bonferroni cor-
rection (p < .0026), and was monomorphic in three localities. The re-
maining 18 microsatellite loci (Table S2) did not show deviation from
HWE in forest elephants and were used in the population analyses.
No significant linkage disequilibrium was detected between pairs of
loci after Bonferroni correction.
The mean number of alleles, the mean Ho, and the mean He of the
18 markers am ong forest e lephants we re 7.44±0.79, 0.58±0.05 ,
and 0.61 ± 0.05 respectively. The allele frequencies for each locus
|
5
ISHIDA et A l.
are shown in Figure S1 and allele number, heterozygosit y, and other
information for each marker are listed in Table S2. The markers did
not show high diversity in savanna elephant s. This would be ex-
pected for two reasons. First, the markers had been designed based
only on the presence of polymorphisms among forest elephants
(Gugala et al., 2016) that are 4–7 million years divergent from sa-
vanna elephants (Brandt, Ishida, Georgiadis, & Roca, 2012; Rohland
et al., 2010). Additionally, savanna elephant s are known to have re-
duced nuclear genetic diversity relative to forest elephants (Roca
et al., 2001; Rohland et al., 2010). In savanna elephants, allele num-
bers for the 18 microsatellite loci ranged from 1 to 4, with the mean
of2.39±0.24.ThemeanHo was 0. 21 ± 0.04, and the mean He was
0.25 ± 0.05. Allele numbers, heterozygosity, and other information
for each marker are listed for the savanna elephants in Table S3.
3.1.1 | Analyses of population structure across
forest elephant localities
Population genetic analyses involved only forest elephants, except
as otherwise indicated. Many analyses were run separately for data-
sets that included or excluded elephants in GR that were identified
as hybrids between forest and savanna elephants, although the out-
comes of these analyses were not greatly affected by including or
excluding hybrids. Analysis of molecular variance (AMOVA) found
that only 3.04% of the variance was accounted for by differences
among localities (Table S4). FIT was 0.054 (p < .005) with significant
deviation from HWE while FIS was 0.024 but did not deviate signifi-
cantly from HWE. The value calculated for FST was low at 0.030, and
this value was statistically significant (p < .001).
FIGURE2 Bayesian clustering
approach implemented in STRUCTURE
(Pritchard et al., 2000) using 18
microsatellite genotypes, including both
forest and savanna elephants or only
forest elephants. (a) When both savanna
and forest elephants were included, the
uppermost K value was estimated as four
(Earl & vonHoldt, 2012). At K = 2, the
forest (Loxodonta cyclotis) and savanna
(L. africana) elephants were almost
completely separated into different
partitions, with a few hybrids in GR,
consistent with previous reports (Ishida
et al., 2011). At higher levels of K, the
additional partitions do not completely
separate elephants from different
localities, although forest elephants from
the eastern Congolian forest (BF, GR)
show different patterns in their partitions
than localities further west. (b) When
only forest elephants were analyzed,
the uppermost K value was estimated
as three. Partitioning within the forest
elephants resembles that seen in panel A,
with a distinctive but incomplete pattern
of partitioning between eastern and
western localities. Note that additional
analyses (Figure S2) excluding the forest–
savanna hybrid elephants from GR did not
greatly af fect the patterns of partitioning
(a)
(b)
Loxodontacyclotis L. africana
SL
LO
OD
DS
BF
GR
Laf
K = 4
SL
LO
OD
DS
BF
GR
Laf
K = 3
SL
L
O
O
D
D
S
BF
G
R
L
af
K = 2
L
O
D
G
L
K = 3
GR0021GR0020
K = 2
SL
LO
OD
DS
BF
GR
SL
LO
O
D
DS
BF
G
R
6
|
ISHIDA et Al.
For each pair of forest localities, FST was also calculated ( Tables 1
and S5) for pairs of localities (excluding SL for which the sample size
was one). We identified five elephants in Garamba as hybrids using
STRUCTURE (see below) and removed these five elephants from
the analyses. Pairwise FST values were high when each forest local-
ity was compared to savanna elephants, which for these analyses
were grouped together (Laf in Table S5). In the comparisons involv-
ing a forest locality and the grouped savanna elephants, pairwise
FST ranged from 0.40 to 0.65, with a statistically significant result
for each comparison. FST was also calculated between each pair of
localities containing forest elephants (Tables 1 and S5), with values
ranging from zero (BF and GR) to 0.07 (GR and OD). FST values were
low between pairs of forest elephant localities although there were
modest but statistically significant dif ferences between some lo-
calities in the eastern and western regions of the Congolian forest
block (Tables 1 and S5). Localities with larger sample sizes (DS, GR,
and LO) tended to yield statistically significant values. Even when
the pair wise differences were found to be statistically significant,
the low values for FST suggested that genetic differentiation among
forest elephants in the Congolian forest block is modest. When sam-
ples from localities in the western forest block were combined and
compared to samples combined across the eastern Congolian forest
block, the calculated FST was low (0.035), although this value was
statistically highly supported (p < .001). By contrast, FST values that
had been previously estimated using only mitochondrial DNA (Ishida
et al., 2013) were much higher (Table 1). The FST values calculated
FIGURE3 Analyses of forest elephants grouped by locality. Five forest–savanna elephant hybrids from GR were excluded from the
analyses. (a) Factorial correspondence analyses implemented in GENETIX version 4.05 (Belkhir et al., 20 04) were used to graphically
represent the distribution of genetic variation among forest elephant localities. Coordinate 1 explained 30.81% of the genetic variation
and separated the Congolian block forest elephants into western (DS, OD, and LO; indicated using darker green) and eastern (BF and GR;
indicatedusinglightgreen)groups.Coordinate3explained19.51%ofthegeneticvarianceandseparatedasingleWestAfricanGuinean
forest block elephant originating in Sierra Leone (SL) from the Central African Congolian forest block elephants. (b) Neighbor–joining trees
based on DA (top), sample size bias- corrected DST (middle), and sample size bias- corrected FST (bottom) showed consistent topologies.
Western Congolian forest block localities (DS, OD, and LO, highlighted in darker green) and eastern Congolian forest block localities (BF and
GR,highlightedwithlightgreen)groupedseparately.Bootstrapvalues≥70%areshown.Onallthreetrees,acladeconsistingofthetwo
localities in the eastern Congolian forest block (BF and GR) showed relatively high bootstrap support. Interestingly, a West Afric an elephant
from Sierra Leone was separated from the other localities by a long branch, which proved robust regardless of method used to calculate
distance or attempts to account for the limited sample size (see Figure S6). Although the distant placement of the Sierra Leone elephant is
intriguing, we caution that no strong conclusion can be drawn from a single individual
(a) (b)
SL
DS
LO
OD
LO
DS OD
SL
BF
GR
GR
BF
85
86
72
0.03
0.02
0.01
|
7
ISHIDA et A l.
using mtDNA ranged from 0.38 to 0.87 for each pair of localities,
while pairwise FST determined using microsatellite markers was 0.07
or less.
Analyses using Structure Harvester suggested that the upper-
most clustering level was K = 4 when we included savanna ele-
phants and K = 3 when we analyzed using only forest elephants data
(Figure 2). The savanna and forest elephants fell into two distinct
partitions, with hybrids detected in GR (Figure 2a), consistent with
previous report s (Ishida et al., 2011). At higher levels of K, the addi-
tional partitioning tended to occur between the three localities on
the western side of the Congolian forest block (LO, OD, and DS) and
the two localities in the eastern side of the Congolian forest block
(BF and GR), although partitioning between west and east was in-
complete (Figure 2a: K = 3, Figure 2b: K = 2). The STRUCTURE analy-
ses excluding 5 hybrid GR elephants produced similar results (Figure
S2). We also conduc ted STRUCTURE analyses for each pair of lo-
calities. We detected differences but incomplete partitioning be-
tween GR and the two western localities, DS and LO in the pairwise
comparisons (Figure S3). Different patterns of par titioning were not
observed between BF and other localities in these pair wise compar-
isons, presumably due to the small sample size for BF (Figure S3).
A principal coordinates analysis (PCoA) conducted using
GenAlEx showed that 25.03% of the genetic variance was explained
by coordinate 1, which revealed clear separation between savanna
and forest elephant (Figure S5a), except for a GR elephant (GR0021)
that was also identified as a forest–savanna elephant hybrid using
STRUCTURE (Figure 2a). Factorial correspondence analyses (FCA)
impleme nted using GENETI X (Belkhir et al., 2004) also d emonstrated
distinc tiveness between forest and savanna elephants (Figure S4a).
Coordinate 1 explained 64.32% of the genetic variance and clearly
separated savanna and forest elephant s. In this FCA, western
Congolian forest elephants separated from eastern Congolian forest
elephants along Coordinate 2 (Figure S4a).
In PCoAs using only forest elephant data, distinctiveness among
localities was not evident when every individual elephant was
plotted (Figure S5b). However, when forest genotypes were com-
bined within each locality, FCA separated localities into two groups
(Figures 3a, S 4a). Coordinate 1 of the FCA explained 30.81% of the
genetic variance, separating forest elephant localities into a western
Congolian forest group (DS, OD, and LO) and an eastern Congolian
forest group (BF and GR). The single sample from Sierra Leone clus-
tered with BF and GR along coordinate 1, but coordinate 3 separated
SL from all other forest elephant localities (Figure 3a). Although co-
ordin at e3explain ed on ly19.51%ofthegenet ic va ri an ceandSL con-
sisted of only one elephant sample, this is an intriguing result given
that SL consisted of a single sample from the only one of our local-
ities within the Guinean forest block of West Africa, which is not
contiguous with the Congolian forest block of Central Africa.
Neighbor–joining (NJ) analyses of forest elephant genotypes
grouped by localit y produced consistent topologies using three
genetic distance calculations: DA, bias- corrected DST, and bias-
corrected FST. The lat ter two methods correct for biases caused by
sample size differences among localities. The analyses involving DA
tended to show a longer branch for the localities with small sample
sizes (Figures 3b, S 4b). This tendency was not evident using bias-
corrected DST and FST (Figures 3b, S4b). In all three trees, savanna
elephants (Laf) formed a lineage distinct from forest elephants
(Figure S4b). Among forest elephants, the eastern Congolian forest
localities (DS, OD, and LO) formed a clade that was distinct from a
clade formed by the western Congolian forest localities (BF and GR),
while SL (in the Guinean forest block) had a long branch separat-
ing it from all other forest elephant localities (Figures 3b, S4b). To
examine whether the separation of SL on the tree was merely due
to it having the smallest sample size of n = 1, we reran the analyses
while also limiting the sample size of LO to a single individual (Figure
S6). Reducing the sample size of LO to one individual affected the
trees based on DA, with the terminal branch length of LO becoming
relatively longer. By contrast, trees based on DST and FST that were
corrected for sample size bias consistently showed a relatively long
branch for SL but not for LO when a single individual was used for
each locality (Figure S6). This was true for three dif ferent individuals
from LO, randomly chosen and used in separate analyses in which
the sample size of LO was limited to one. For the two bias- corrected
methods, the long branch length was consistently present for SL
with n = 1, but not for LO with n = 1, which may suggest that the
long separation between SL and the other populations may be due
to actual genetic differences, and not be a mere artifact of the small
sample size.
3.1.2 | Evidence for isolation by distance among
forest elephants in the Congolian forest block
We examined the degree to which genetic differences among for-
est elephants at different localities varied with the geographic
distances separating them. Geographic distances were computed
between each pair of elephant s, with the distances placed into
quartiles (x- axis in Figure 4a) to implement a spatial autocorrela-
tion analysis (Peakall & Smouse, 2012). Genetic distances between
pairs of elephants were also determined (y- axis in Figure 4a). A
spatial autocorrelation analysis showed a correlation between
genetic distance and geographic distance. Forest elephants that
were close to each other geographically were also more similar
genetically than were the geographically distant forest elephants
(Figure 4a).
Suppor t for isolation by distance (Jensen et al., 2005) was exam-
ined. When the single sample from Sierra Leone (SL) was included in
the analyses, only a marginally significant correlation between ge-
netic distance and geographic distance was detected for FST (r = .45,
p = .049, Mantel test ) and for Rousset ’sd istance (r = .45, p = .052,
Mantel test) (Figure S7). However, as SL included only one sample,
and SL is from a different and geographically discontinuous forest
block, a more conser vative approach excluded the single sample
from SL for these analyses. Within the Congolian forest block, isola-
tion by dist ance received strong suppor t, as determined using both
FST (r = .86, p < .01, Mantel test) (Figure 4b) and Rousset’s distance
(r = .85, p < .01, Mantel test) (Figure 4c).
8
|
ISHIDA et Al.
4 | DISCUSSION
4.1 | Current population structure among forest
elephants
Nuclear genetic dif ferences were detected among Central African
forest locations. Specifically, genetic partitioning by STRUCTURE
identified that elephants in western (LO, OD, and DS) and in eastern
(BF and GR) Congolian forests were detectably different, although
partitioning was far from complete (Figure 2). The same groupings
were also evident in clustering analyses (Figure 3). However, as indi-
cated in the PCoA, microsatellite profiles from individuals in eastern
and western Central Africa showed a great deal of overlap, and FST
values were modest. All of these result s point to only limited genetic
differentiation between eastern and western localities within the
Central African forests.
Hybrids between forest and savanna elephants have been docu-
mented, but only within relatively narrow transition zones bet ween
forest and savanna habitats (Comstock et al., 2002; Ishida et al.,
2011; Mondol et al., 2015; Roca et al., 2001). One impor tant point
regards the number of markers needed to estimate the relative
contributions of the t wo species to hybrid individuals. In a previous
analysis (Ishida et al., 2011), the hybrid individual shown to have
the greatest proportion of savanna elephant alleles was GR0020,
whereas the current study identified GR0021 as having the highest
savanna elephant contribution (GR0020 had the second highest pro-
portion; Figures 2a and S5a). This likely reflect s stochasticity in the
genomic distribution of genetic markers from forest or savanna lin-
eages in hybrids, and in the proportion of each genotype attributed
to either lineage by the assignment software. The use of a larger
number of microsatellite markers, combining the novel markers used
here with those previously developed, is likely to provide greater
FIGURE4 The relationship between genetic distance and geographic distance among forest elephants from different localities. (a)
Spatial autocorrelation analysis found that genetic distance was correlated with geographic distance (r is the spatial autocorrelation
coefficient; Uistheupper95%randomizationlimitsofr; Listhelower95%randomizationlimitsofr). (b) Results consistent with potential
isolation by distance were obtained by comparing genetic distance (FST) to geographic distance (km) in pairwise comparisons of forest
elephant localities (r = .86, p < .01). (c) Result s consistent with potential isolation by distance were obtained when genetic distances were
calculated using Rousset’s distance FST/(1- FST) and compared to geographic distances (km) in pairwise comparisons of the genotypes of
forest elephants grouped by localit y (r = .85, p < .01). Five forest–savanna elephant hybrids from GR were excluded from the analysis shown
in each panel; Sierra Leone (SL) was not used in the pairwise comparisons because the sample size was one. For results including the SL
sample, see Figure S7
0.04
–0.02
–0.04
0.00
0.02
r
r
U
–0.06 161284
Distance class (end point)
L
0.070
0.062
0.080000
0.071000
(c)(b)
(a)
0.046
0.054
0.053000
0.062000
0.014
0.022
0.030
0.038
Genetic distance (
FST)
0.017000
0.026000
0.035000
0.044000
Genetic distance ((FST/(1-FST)))
0599 1199 1799 2399 2999
–0.010
–0.002
0.006
–0.010000
–0.001000
0.008000
Geographic distance
0599 1199 1799 2399 2999
Geographic distance
|
9
ISHIDA et A l.
precision in estimating the degree to which hybrids received alleles
fromonelineageor theother(Boecklen&Howard,1997).Itwould
also be likely to further increase the accuracy and precision of esti-
mating the provenance of confiscated ivory using nuclear markers
(Wasser et al., 2015).
The distinctiveness of elephants from West Africa has been pro-
posed (Eggert et al., 2002), but based largely on mitochondrial DNA
data, which can be misleading in elephants (due to maternal inher-
itance and female philopatr y) (Ishida et al., 2011, 2013). The single
individual from West Africa (Sierra Leone) in this study appeared
to anchor one of the axes in the FCA analysis, and also generated a
long branch in the phylogeny, when compared to other isolated in-
dividuals (Figures 3 and S4). While suggestive, this is not conclusive
evidence for the distinctiveness of West African elephants (a larger
sample set is required). However, the Benin/Dahomey Gap sepa-
rating the Congolian from Guinean forest blocks may hinder gene
flow in forest elephants, as it has affected the distribution of other
forest- dwelling taxa (Linder, 2014). For this reason, we previously
recommended that a conservative approach would treat elephants
on either side of the Gap as deserving of separate conservation sta-
tus (Roca et al., 2015). Whether Guinean and Congolian forest ele-
phants form genetically distinctive groups (based on nuclear DNA
analysis) remains one of the most important unanswered questions
in elephant conservation genetics (Roca et al., 2015).
4.2 | Discordant mitonuclear patterns and the
role of range expansion
Mitochondrial DNA patterns have previously been examined in
African forest elephants (Debruyne, 2005; Debruyne et al., 2003;
Egger t et al., 20 02; Ishida et al., 2013; Johnson et al., 20 07; Ny akaana
et al., 2002), revealing five mitochondrial subclades with distinctive
geographically restricted distributions (Ishida et al., 2013). In a pre-
vious study (Ishida et al., 2013), pairwise FST values calculated using
mtDNA were quite high, ranging from 0.38 to 0.87 when estimated
pairwise between localities (Table 1). However, in taxa with male-
biased dispersal, phylogeographic patterns are often discordant
betweennuclearand mitochondrial DNA (Petit & Excof fier, 2009;
Toews & Brelsford, 2012), and discordant mitonuclear patterns have
been reported among living and extinct elephantid species (Enk
et al., 2011; Lei, Brenneman, Schmitt, & Louis, 2012; Meyer et al.,
2017; Palkopoulou et al., 2015; Roca, 2015; Roca et al., 2005). This
is consistent with the current analysis, which found nuclear genetic
differentiation among forest elephants to be much lower than what
might be inferred using mtDNA alone, with all values of FST≤0.07
among forest elephant populations across Central Africa (Table 1).
We would note that the high mutation rate of mtDNA would not ac-
count for the discrepant patterns, because a faster- evolving marker
would only reveal greater resolution than a slower one; by contrast
mtDNA demonstrates a strikingly different phylogeographic pat-
tern than nuclear DNA in forest elephant s (Figure 2 vs. Figure S8)
as in other elephantids (Enk et al., 2011; Lei et al., 2012; Meyer et al.,
2017; Palkopoulou et al., 2015; Roca et al., 2005, 2007, 2015).
Mitonuclear discordant patterns in most cases have been at-
tributed to adaptive introgression of mtDNA , demographic dispar-
ities and sex- biased asymmetries, with some studies also implicating
habitat changes and hybrid zone movements ( Toews & Brelsford,
2012). In the case of forest elephants, adaptive introgression of
mtDNA is unlikely, not only because selective sweeps are unlikely
to occur in markers carried only by the nondispersing sex (Petit &
Excoffier,2009),butbecause mtDNAshowsgreaterdifferentiation
across the forest elephant range than does nuDNA ( Table 1). Instead,
sex- based differences in gene flow appear to be responsible for the
discordant mitonuclear patterns, with some impact likely due to
changes in habitat across geological time. Because female elephants
are matrilocal and remain with their natal social group (Archie et al.,
2007; Hollister- Smith et al., 2007), this behavior can account for the
persistence of geographic structuring in forest elephant mtDNA
(Ishida et al., 2013). By contrast, male elephants disperse from their
natal social groups and mediate nuclear gene flow across the land-
scape(Nyakaana&Arctander,1999;Rocaetal.,20 05,2015).
In forest elephant s, the mitonuclear patterns were likely im-
pacted by habitat changes across geological time, and discordant
mitonuclear patterns may provide a means for studying their range
expansion after the end of the last glacial period. Genetic patterns
largely depend on the demographic and ecological characteristics of
a species (C astric & Bernatchez, 2003). Spatial patterns of genetic
diversit y may also reflect past changes in climate and habitats that
expanded and contracted the ranges of species, sometimes at a
fast pace (Hewitt, 2000). Current spatial genetic diversity may re-
flect such past events rather than species demography, with geo-
graphic differences in genetic diversity reflecting the effects of past
climate- driven range dynamics (Hewitt, 2000). Range expansion can
lead to patchiness after migration, due to long- distance movements
followed by population expansions (i.e., a leptokurtic distribution of
dispersal distances during colonization) (Ibrahim, Nichols, & Hewitt,
1996;Klopfstein,Currat,&Excoffier,2006;McInerny,Turner,Wong,
Travis,&Benton, 2009).These compounded foundation processes
can lead to increased genetic differentiation, although such effects
are negatively correlated with migration rate, because migration
decreases lags in colonization and reduces the strength of founder
effects(Klopfsteinetal.,2006;McInernyetal.,2009).
In many species, it is often difficult or impossible to infer pro-
cesses that are not directly observable from the current spatial ge-
netic structure, especially as various processes may create similar
patterns(McIntire&Fajardo,2009).However,inelephantsextreme
sex differences in dispersal may allow for the study of both current
demographic effects through the examination of male- mediated nu-
clear patterns (this study), and for the study of ancient landscape
effects through analysis of mitochondrial pat terns mediated by fe-
males (Figure S8) (Ishida et al., 2013), which may retain signatures
of leptokurtic dispersal and compounded foundation processes.
Pleistocene glacial cycles caused habitat changes that temporarily
isolated populations of some species, with repeated cycles of isola-
tion followed by expansion and contraction (Hewitt, 200 0). During
periods of spatial expansion, alleles present at the expanding edge
10
|
ISHIDA et Al.
of the species range can reach high frequencies (Klopfstein et al.,
2006), with expanding populations potentially subject to iterated
founder effects (Klopfstein et al., 2006). During expansions, rare
long- distance dispersal events followed by exponential population
growth can generate long- term patchiness in population structure
(Hewitt , 2000; Ibr ahim etal., 1996). Such ancient e vents may be
preserved in elephant mitochondrial geographic patterns, which are
likely to be stable due to female philopatr y.
Further, studies of forest elephants would also avoid a common
pitfall of using too small a geographic scale (Jenkins et al., 2010)
while benefiting from a very large number of museum samples that
are available for mtDNA analyses and also have precise provenance
information. Species refugia during glacial cycles are better charac-
terized in Europe and North America than elsewhere, and the forest
elephant may provide novel insights into the impact of global glacial
cycles in the African tropics (Hewitt, 200 0).
4.3 | The potential role of isolation by distance
Distinguishing between discrete population genetic structure
(Figures 2 and 3) and isolation by distance (Figure 4) can be difficult
(Meirmans, 2012). Models of isolation by distance are often used
to approach the balance between drift and dispersal (Jenkins et al.,
2010;Wright,1943).Limitedmigrationpermitsgeneticdrift,increas-
ing population differentiation and leading to a correlation between
neutral genetic differentiation and geographic distance (Jenkins
etal.,2010;Wright,1943).IBDdevelopsincontinuouslydistributed
species when divergence accumulates due to genetic drift between
locations separated by geographic distances large enough to over-
comethehomogenizingeffectsofgeneflow(Chesser,1983;Crispo
&Hendry,2005;Hewitt,20 00;Jenkinsetal.,2010;Wright,1943),
and is common in natural populations (Jenkins et al., 2010). IBD can
remain at equilibrium, after sufficient time has elapsed for genetic
patterns to be established and stabilized (Castric & Bernatchez,
2003).
Forest elephants have been contiguously distributed across
Central Africa from the start of the Holocene (Plana, 20 04), until
atleast themid-to late 1900s (Douglas-Hamilton, 1987). The cor-
relation between genetic and geographic distances among localities
in Central Africa (Figure 4) may be attributable to isolation by dis-
tance (IBD)(Jenkinsetal., 2010; Wright, 1943).However,because
forest elephant populations expanded from discrete glacial refugia,
it would be difficult to distinguish a role of IBD from the persistence
of discrete population structure (Meirmans, 2012) that would have
been diminished but perhaps not erased by postglacial gene flow.
4.4 | Conservation implications
Given the endangered status of forest elephants, and their role as
a keystone species, discussion of the conser vation implications of
our results is warranted. An import ant step for the conservation of
forest elephants would be universal recognition of its status as a
separate species in need of species- specific conservation measures
(Roca et al., 2015). Central Africa, the region in which most forest el-
ephants live, has suf fered from the highest levels of elephant poach-
ing of any subregion (CITES, 2012) and has been the main source for
the illegal trade in elephant bushmeat and ivory ( Wasser et al., 2015)
leading to massive declines in their numbers (Maisels et al., 2013).
The increase in human numbers and activities has caused fragmen-
tation of elephant habitats and range (Blake et al., 2007, 2008). Such
fragmentation reduces genetic connectivit y, which can lead to loss
of genetic variation, and ultimately to inbreeding and increased drift.
Recognition both of species divisions and of population genetic
patterns below the species level is essential for the maintenance of
biodiversity, and an important conser vation principle is to retain
populat ions representi ng existing genet ic variation (Mori tz, 1994;
Ryder, 1986). Our findings suggest that West African, western
Congolian, and eastern Congolian forest elephants should be man-
aged separately. For species such as the forest elephant that exhibit
patterns indicative of limited population structure (Figures 2 and
3) and possibly isolation by distance (Figure 4), Chesser (Chesser,
1983)hassuggesteddividingtherangeintomanagementunits,with
greater genetic exchanges within units than across units in order to
balance the need for connectivity with the need to prevent loss of al-
leles. Within regions, it is important to prevent ex treme habitat frag-
mentation and retain connec tivity so that gene flow can continue
amongpopulations(Chesser,1983).Shouldthedestructionofele-
phants cease while habitats remain, populations could expand from
multiple locations. Should active management become necessary for
recolonization, the best source populations for translocations would
be those that are geographically close, as these would be most simi-
lar to any extirpated population (Monsen & Blouin, 20 04).
Forest elephants play a critical role in shaping their ecosystem,
maintaining tree diversity, dispersing seeds in greater quantities
and distances than most other fauna, and improving rates of seed
germination following passage through the gut (Campos- Arceiz &
Blake, 2011). In the central Congo Basin, seventy- eight percent of
the larger tree species in the rain forest are dependent on forest ele-
phants for seed dispersal (Blake et al., 2007). Thus, it is important to
consider their ecological role in seed germination and dispersal when
determining conservation priorities. Terrestrial ecoregions of the
world have been mapped to identif y areas of high biodiversity and
representative communities (Olson et al., 2001). The range encom-
passed by Central African forest elephants includes five tropical for-
est ecoregions, three ecoregions of forest–savanna mosaic, as well
as the Albertine Rift montane forests and Cameroonian highlands
forests(FigureS9)(Olson etal., 2001). Given limitedgeneticdiffer-
entiation among elephants across the Congo Basin, these ecoregions
could ser ve as management units for the forest elephants, reflecting
the dependence of many plant species on elephants for seed ger-
mination and dispersal. Within regions, the extirpation of local ele-
phant populations should be minimized, to minimize effects on the
regional flora.
An impor tant further consideration is the impact of forest ele-
phant conservation on those plant species that are rare or regionally
restricted. A survey of 5,881 species of plants across sub- Saharan
|
11
ISHIDA et A l.
Africa has been used to map endemism richness, which is defined as
the sum of species present at a geographic loc ation, but with the occur-
rence of each plant species inversely weighted by the size of its range
(Figure S10) (Linder, 2014; Linder et al., 2012). The endemic richness of
plants has been found to be highly congruent with the endemic rich-
ness of frogs, snakes, birds, and mammals, which when combined indi-
cated that the Benin Gap has influenced these patterns (Linder, 2014).
Congruence across these groups has been attributed to (1) a similar
influence across ver tebrates of the vegetation and flora, (2) common
responses to the same climatic parameters, and (3) a common underly-
ing histor y (Linder, 2014). For plants, regions with highest levels of en-
demism have been identified (Linder, 2014; Linder et al., 2012) and are
indicated in Figure S10. These regions of endemism may be considered
when setting conservation priorities for forest elephants, although a
more precise mapping of endemism among those plants that are de-
pendent on elephants would be helpful. Conserving elephants in the
regions rich in plant endemism would directly benefit the conservation
of many plant species dependent on elephants, and indirectly benefit
other vertebrate and nondependent plant species for which levels of
endemism are geographically congruent.
ACKNOWLEDGMENTS
The work was funded by USFWS African Elephant Conservation
Fund Grants AFE-0778-F12AP01143 and AFE1606-F16AP00909.
The study was condu cted under the Universit y of Illinois Institu tional
Animal Care and Use Committee (IACUC) approved protocol number
12040. Samples were impor ted using appropriate CITES permits.
For technic al or other assistance, we thank M. Malask y, R. Hanson,
and the UIUC High- Throughput Sequencing and Genotyping Unit.
We are grateful to the governments of Bot swana, Cameroon,
Central African Republic, Democratic Republic of Congo, Gabon,
Kenya, Namibia, Republic of Congo, South Africa, Tanzania, and
Zimbabwe for permission to collect samples. For help with sam-
ple collection, we thank A. Turkalo, M. Fay, L. White, R. Weladji,
W. Karesh, M. Lindeque, W. Versvelt, K. Hillman Smith, F. Smith,
M. Tchamba, S. Gartlan, P. Aarhaug, A. M. Austmyr, Bakari, Jibrila,
J. Pelleteret, L. White, M. Habibou, M. W. Beskreo, D. Pierre, C.
Tutin, M. Fernandez, R. Barnes, B. Powell, G. Doungoubé, M. Storey,
M. Phillips, B. Mwasaga, A. Mackanga- Missandzou, K. Amman, K.
Comstock, M. Keele, D. Olson, B. York, and A. Baker at the Burnet
Park Zoo, M. Bush at the National Zoological Park, and A. Lécu at
Zoo de Vincennes (Paris Zoo).
CONFLICT OF INTEREST
None declared.
AUTHOR CONTRIBUTIONS
YI and ALR designed the study. YI and NAG performed experiment s
and analyses. NJG provided samples. YI, NJG, and ALR contributed
to writing the manuscript.
DATA ACCESSIBILITY
The sequences of microsatellites used in this manuscript were estab-
lishedanddepositedatNCBIGenBank(KU947083–KU947105),and
primer sequences are available in Gugala et al. (Gugala et al., 2016).
ORCID
Alfred L. Roca http://orcid.org/0000-0001-9217-5593
REFERENCES
Allentof t, M. E., Oskam, C., Houston, J., Hale, M. L ., Gilbert, T. P.,
Rasmussen, M., … Bunce, M. (2011). Profiling the dead: G enerating
microsatellite dat a from fossil bones of extinct megafauna–pro-
tocols, problems , and prospects. PLoS ONE, 6, e16670. https://doi.
org/10.1371/journal.pone.0016670
Archie, E. A., Fitzpatrick, C. L., Moss, C. J., & Alberts, S. C . (2011) The
population genetics of the Amboseli and Kilimanjaro elephants.
In C. J. Moss, H. Croze & P. C. Lee(Eds.), The A mboseli elephant s:
A Long-Term Perspective on a Long-Lived Mammal (pp. 37–47).
Chicago, IL: University of Chicago Press. https://doi.org/10.7208/
chicago/9780226542263.001.0001
Archie, E. A., Hollister-Smith, J. A., Poole, J. H., Lee, P. C., Moss, C. J.,
Maldonado, J. E., … Alberts, D. C. (2007). Behavioural inbreeding
avoidance in wild African elephants. Molecular Ecology, 16, 4138–
4148. ht tps://doi .org/10.1111/ j.13 65-294X.20 07.034 83.x
Archie, E. A., Moss, C. J., & Alber ts, S. C . (2011). Friends and relations:
Kinship and the nature of female elephant social relationships.
In C. J. Moss, H. Croze & P. C. Lee (Eds.), The Amboseli Elephants.
A Long-Term Perspective on a Long-Lived Mammal (pp. 238–245).
Chicago, IL: University of Chicago Press. https://doi.org/10.7208/
chicago/9780226542263.001.0001
Belkhir, K., Borsa, P., Chikhi, L., Raufaste, N., & Bonhomme, F. (2004).
Genetix 4.05. laboratoire génome, populations, interactions, CNRS UMR
5000. Montpellier, France: Université de Montpellier II.
Blake, S., Deem, S. L., Strindberg, S., Maisels, F., Momont , L., Isia, I.-B., …
Kock, M. D. (2008). Roadless wilderness area determines forest ele-
phant movements in the Congo Basin. PLoS ONE, 3, e354 6. https://
doi.org/10.1371/journal.pone.0003546
Blake, S., Strindberg, S., Boudjan , P., Makombo, C., Bila-Isia, I., Ilambu,
O., … Maisels, F. (2007). Forest elephant crisis in the Congo
basin. PLoS Biology, 5, e111. https://doi.org/10.1371/journal.
pbi o.0 050111
Boecklen,W.J.,&Howard,D.J.(1997).Geneticanalysisofhybridzones:
Number s of markers and power of resolution. Ecology, 78, 2611–
2616. https://doi.org/10.1890/0012-9658(1997)078[2611:GAOHZ
N]2.0.CO;2
Brandt, A. L ., Ishida, Y., Georgiadis, N. J., & Roca, A . L. (2012). Forest
elephant mitochondrial genomes reveal that elephantid diversifica-
tion in Afr ica tracked cli mate transition s. Molecular Ecology, 21, 1175–
1189.https://doi.o rg /10.1111/j.1365-294X .2 012 .05461.x
Campos-Arceiz, A ., & Blake, S. (2011). Megagardeners of the forest –
the role of elephant s in seed dispers al. Acta Oecologica, 37, 542–553.
https://doi.org/10.1016/j.actao.2011.01.014
Castric, V., & Bernatchez, L. (2003). The rise and fall of isolation by dis-
tance in the anadromous brook charr (Salvelinus fontinalis Mitchill).
Genetics, 16 3,983–996.
Chesser, R. K. (1983). Isolation by dist ance: Relationship to manage-
ment of genetic resources. In C . Schonewald-Cox, S. Chambers, B.
MacBride, & L. Thomas (Eds.), Genetics and Conservation: A Reference
for Managing Wild Animal and Plant Populations (pp. 66–77). Menlo
Park, C A: Benjamin Cummings Publ. Co..
12
|
ISHIDA et Al.
CITES (2012). Elephant conser vation, illegal k illing and ivory trade. In
Sixty-second meeting of the Standing Committee, (pp. SC62 Doc. 46.61
(Rev.61)–p.61-p.29).Geneva,Switzerland.
Comstock, K. E., Georgiadis, N., Pecon-Slattery, J., Roca, A . L., Ostrander,
E. A., O’Brien, S. J., & Wasser, S. K. (2002). Patterns of molecular ge-
netic variation among African elephant populations. Molecular Ecology,
11,2489–2498.https://doi.org/10.1046/j.1365-294X. 2002.01615.x
Crispo, E ., & Hendr y, A. P. (2005). Does time since colonization influ-
ence isolation by distance? A meta- analysis. Conservation Genetics, 6,
665–6 82.htt ps://doi.org/10.10 07/s10592-0 05-9 026-4
Debru yne, R. (20 05). A case stu dy of apparent c onflict bet ween molecu lar
phylogenies: The interrelationships of African elephants. Cladistics,
21,31–50.https://doi.org/10.1111/j.1096-0031.2004.00044.x
Debruyne, R., Van Holt, A ., Barriel, V., & Tassy, P. (2003). Status of
the so- called African pygmy elephant (Loxodonta pumilio [NOACK
1906]): Phyloge ny of cytochro me b and mitochon drial control re -
gion sequences. Comptes Rendus Biologies, 326,687–697.https://doi.
org /10.1016/S1631-0691(03)00158-6
Douglas-Hamilton, I. (1987). African elephants: Population trends
and their causes . Oryx, 21, 11–24. https://doi.org /10.1017/
S0030605300020433
Earl, D., & vonHoldt , B. (2012). STRUCTURE HARVESTER: A website
and program for visualizing S TRUCTURE output and implementing
the Evanno method. Conservation Genetics Resources, 4, 359–361.
htt ps://doi.org /10.10 07/s12686-011-954 8-7
Egger t, L. S., Rasner, C. A., & Woodruff, D. S. (2002). The evolution and
phylogeography of the African elephant inferred from mitochondrial
DNA sequence and nuclear microsatellite markers. Proceedings of the
Royal Societ y of London Series B: Biological Sciences, 269,1993–2006.
https://doi.org/10.1098/rspb.2002.2070
Enk, J., Devault, A ., Debruyne, R., King, C. E., Treangen, T., O’Rourke,
D., … Poinar, H. (2011). Complete Columbian mammoth mitogenome
suggests interbreeding with woolly mammoths. Genome Biology, 12,
R51. https://doi.org/10.1186/gb-2011-12-5-r51
Excoff ier, L., & Lischer, H. E. (2010). A rlequin suite ver 3.5: A new series
of programs to perform population genetics analyses under Linux
and Windows. Molecular Ecolog y Resources, 10, 56 4–567. https://doi.
org /10.1111/j.1755- 0998. 2010.028 47.x
Groves, C . P., & Grubb, P. (2000). Do Loxodonta cyclotis and L. africana
interbreed? Elephant, 2, 4–7.
Gugala, N. A., Ishida, Y., Georgiadis, N. J., & Roca, A. L . (2016).
Development and characterization of micros atellite markers in the
African forest elephant (Loxodonta cyclotis). BMC Research Notes, 9,
364. https://doi.org/10.1186/s13104-016-2167-3
Hewitt, G. (20 00). The genetic legacy of the Quaternary ice ages. Nature,
405,907–913.https://doi.org/10.1038/3501600 0
Hollister-Smith, J. A., Poole, J. H ., Archie, E. A., Vance, E. A ., Georgiadis,
N. J., Moss , C. J., & Albert s, S. C. (2007). A ge, musth and pa-
ternit y success in wild male African elephants, Loxodonta afri-
cana. Animal Behaviour, 74, 287–296. https://doi.or g/10.1016/j.
anbehav.2006.12.008
Ibrahim, K. M.,Nichols, R. A., & Hewit t, G. M. (1996).Spatial patterns
of genetic variation generated by different forms of dispersal during
range expansion. Heredity, 77, 282–291. ht tps ://doi.org/10.1038/
hdy.1996 .142
Ishida, Y., Georgiadis, N. J., Hondo, T., & Roca, A . L. (2013).
Triangulating the provenance of African elephants using mito-
chondrial DNA. Evolutionary Applications, 6, 253–265. https://doi.
org /10.1111/j.1752-4571. 201 2. 00 286 .x
Ishida, Y., Ol eksyk, T. K., Geo rgiadis, N. J., Dav id, V. A., Zhao, K., St ephens,
R. M., … Roca, A. L. (2011). Reconciling apparent conflict s between
mitochondrial and nuclear phylogenies in African elephants. PLoS
ONE, 6, e20642. https://doi.org/10.1371/journal.pone.0020642
Jakobsson, M., & Rosenberg, N. A. (2007). CLUMPP: A cluster match-
ing and permutation progr am for dealing with label switching and
multimodality in analysis of population structure. Bioinformatics, 23,
1801–1806.https://doi.org/10.1093/bioinformatics/btm233
Jenkins, D. G., Carey, M., Czerniewska, J., Fletcher, J., Hether, T., Jones,
A., … Tursi, R . (2010). A meta- anal ysis of isolation by dis tance: Relic or
reference standard for landscape genetics? Ecography, 33, 315–320.
Jensen, J. L., Bohonak, A . J., & Kelley, S. T. (2005). Isolation
by distance, web service. BMC Genetics, 6, 13. https://doi.
org /10.118 6/1471-2156 -6-13
Johnson, M. B., Clifford, S. L., G oossens, B., Nyakaana, S., Curran,
B., White, L. J. T., … Bruford, M. W. (2007). Complex phylogeo-
graphic histor y of central African forest elephants and it s implic a-
tions for taxonomy. BMC Evolutionary Biology, 7, 244. https://doi.
org /10.118 6/1471-2148-7-244
Klopfs tein, S., Currat , M., & Excoffier, L. (2006). The fate of mut ations
surfing on the wave of a range expansion. Molecular Biology and
Evolution, 23,482–490.https://doi.org/10.1093/molbev/msj057
Latter,B.D.(1972).Selectionin finitepopulationswithmultiple alleles.
3. Genetic divergence with centripetal selection and mutation.
Genetics, 70,475–490.
Lee, P. C., Poole , J. H., Njiraini , N., Sayialel, C . N., & Moss, C . J. (2011). Male
social dy namics: Independence and beyond. In C. J. Moss, H. Croze &
P. C. Lee (Eds.), The A mboseli Elephants. A Long-Term Perspective on a
Long-Lived Mammal (pp. 260–271). Chicago, IL: University of Chicago
Press.https://doi.org/10.7208/chicago/9780226542263.001.0001
Lei, R. , Brenneman, R . A., Schm itt, D. L., & Lo uis, E. E. Jr (2012). G enetic di-
versity in North American captive Asian elephants. Journal of Zoolog y,
286,38–47.https://doi.org/10.1111/j.1469-7998.2011.0 0851.x
Linder, H. P. (2014). The evolution of African plant diversity. Frontiers in
Ecology and Evolution, 2, 38.
Linder, H. P., de Kler k, H. M., Born , J., Burgess, N. D., F jeldså, J., & Rah bek,
C. (2012). The partitioning of Africa: Statistically defined biogeo-
graphical regions in sub- Saharan Africa. Journal of Biogeography, 39,
1189–1205.h tt ps://doi.org /10.1111/j.1365 -2699.2012.0 2728 .x
Maisels, F., Strindberg, S., Blake, S., Wittemyer, G., Hart, J., Williamson,
E. A., … Warren, Y. (2013). Devastating decline of forest elephant s in
central Africa. PLoS ONE, 8,e59469.https://doi.org/10.1371/journal.
pone.0059469
McInerny, G. J., Turner, J. R. G., Wong, H. Y., Travis, J. M. J., & B enton, T.
G.(2009).Howrangeshift sinducedbyclimatechangeaffectneutral
evolution. Proceed ings of the Royal Society B: Biological Sciences, 276,
1527–1534.https://doi.org/10.1098/rspb.2008.1567
McIntire,E. J. B., & Fajardo, A. (2009). Beyond description: The active
and effective way to infer processes from spatial patterns. Ecolog y,
90,46–56.https://doi.org/10.1890/07-2096.1
Meirmans, P. G. (2012). The trouble with isolation by dis-
tance. Molecular Ecology, 21, 2839–2846. https://doi.
org /10.1111/j.1365 -294X .2012.0 5578 .x
Meyer, M., Palkopoulou, E., Baleka, S., Stiller, M., Penkman, K . E. H., Alt,
K. W., … Hofreiter, M. (2017). Palaeogenomes of Eurasian straight-
tusked elephants challenge the current view of elephant evolution.
eLife, 6, e 25 413.
Mondol, S., Moltke, I., Har t, J., Keigwin, M., Brown, L ., Stephens, M., &
Wasser, S. K. (2015). New evidence for hybrid zones of forest and
savanna elephant s in Central and West Africa . Molecular Ecology, 24,
6134–6147. htt ps://doi.org/10.1111/me c.13472
Monsen, K. J., & Blouin, M. S. (2004). Extreme isolation by distance in
a montane frog Rana cascadae. Conservation Genetics, 5, 827–835.
htt ps://doi.org /10.10 07/s10592-004-1981-z
Moritz, C. (1994). Defining ‘Evolutionarily Significant Units’ for con-
servation. Trends in Ecology and Evolution, 9, 373–375. https://doi.
org /10.1016/0169-5347(94)9 00 57-4
Nei, M. (1972). Genetic distance bet ween populations. The American
Naturalist, 106,283–292.https://doi.org/10.1086/282771
Nei, M., Tajima,F.,&Tateno, Y.(1983). Accuracyofestimatedphyloge-
netic trees from molecular data. II . Gene frequenc y data. Journal
|
13
ISHIDA et A l.
of Molecular Evolution, 19, 153–170. https://doi.org/10.1007/
BF0230 0753
Nyakaana,S.,&Arctander,P.(1999).Populationgeneticstructureofthe
African elephant in Uganda based on variation at mitochondrial and
nuclear loci: Evidence for male- biased gene flow. Molecular Ecology,
8,1105–1115.https://doi.org/10.1046/j.1365-294x.1999.00661.x
Nyakaana, S., Arctander, P., & Siegismund, H. R. (2002). Population struc-
ture of the African s avannah elephant inferred from mitochondrial
control region sequences and nuclear microsatellite loci. Heredity, 89,
90–98 .ht tps ://doi.org/10.1038/sj.hdy.68 00110
Olson, D. M., Dinerstein, E ., Wikramanayake, E. D., Burgess, N.
D., Powell, G . V. N., Under wood, E. C., … Kassem, K. R . (2001).
Terrestrial ecoregions of the world: A new map of life on ear th: A
new global map of terrestrial ecoregions provides an innovative tool
for conserving biodiversity. BioScience, 51, 933–938. https://doi.
org/10.1641/0006-3568(2001)051[0933:TEOTWA]2.0.CO;2
Palkopoulou, E., Mallick, S., Skoglund, P., Enk, J., Rohland, N., Li, H., …
Dalén, L . (2015). Complete genomes reveal signatures of demo-
graphic and genetic declines in the woolly mammoth. Current Biol ogy,
25,1395–1400.https://doi.org/10.1016/j.cub.2015.04.007
Peakall, R., & Smouse, P. E. (2012). GenAlEx 6 .5: Genetic analysis in
Excel. Population genetic software for teaching and research–an
update. Bioinformatics, 28, 2537–2539. https://doi.org/10.1093/
bioinformatics/bts460
Petit, R . J., & Excof fier, L. (2009). Gene flow and species delimita-
tion. Trends in Ecology a nd Evolution, 24, 386–393. https://doi.
org/10.1016/j.tree.2009.02.011
Plana, V. (2004). Mechanisms and tempo of evolution in the African
Guineo- Congolian rainforest. Philosophical Transactions of the
Royal Societ y B: Biological Sciences, 359, 1585–1594. https://doi.
org /10.1098/rst b.200 4.1535
Poole, J. H., Lee, P. C., Njir aini, N., & Moss, C. J. (2011). Longevity, com-
petition, and musth: A long- term perspective on male reproduc tive
strategies. In C . J. Moss, H . Croze & P. C. Lee (Eds.), The Amboseli
elephants. A Long-Term Perspec tive on a Long-Lived Mammal (pp.
272–286). Chicago, IL: University of Chicago Press. ht tps://doi.
org/10.7208/chicago/9780226542263.001.0001
Pritchard, J. K., Stephens, M., & Donnelly, P. (2000). Inference of pop-
ulation structure using multilocus genotype data. Genetics, 155,
945–9 59.
Rambaut, A. (2014). FigTree v1.4.2. http://tree.bio.ed.ac.uk/software/
figtree/
Roca, A . L. (2015). Evolution: The Island of misfit mammoths. Current
Biology, 25,R549–R551.https://doi.org/10.1016/j.cub.2015.05.006
Roca, A . L., Georgiadis, N., & O’ Brien, S. J. (2005). Cytonuclear genomic
dissociation in African elephant species. Nature Genetics, 37,96–100.
htt ps://doi.org /10.1038/ng14 85
Roca, A . L., Georgiadis, N., & O’ Brien, S. J. (2007). Cyto- nuclear ge-
nomic dissociation and the African elephant species question.
Quaternary International, 169 –17 0 , 4–16. http s://doi.o rg/10.1016/j .
quaint.2006.08.008
Roca, A . L., Georgiadis, N., Pecon-Slattery, J., & O’Brien, S. J. (2001).
Genetic evidence for two species of elephant in Africa. Science, 293,
1473–1477.https://doi.org/10.1126/science.1059936
Roca, A . L., Ishida , Y. , Brandt, A . L., Benjam in, N. R., Zhao , K., & Georgia dis,
N. J. (2015). Elephant natural history: A genomic perspective. Annual
Review of Animal Biosciences, 3, 139–167. https://doi .org/10.1146/
annurev-animal-022114-110838
Rohland, N., Reich, D., Mallick, S., Meyer, M., Green, R. E., Georgiadis, N.
J., … Hofreiter, M. (2010). Genomic DNA sequences from mastodon
and woolly mammoth reveal deep speciation of forest and savanna
elephants. PLoS Biolog y, 8, e100056 4.
Rosenberg, N. A . (200 4). DISTRUCT: A program for the graphical display
of population structure. Molecular Ecology Notes, 4, 137–138.
Rousset , F. (2008). genep op’007: A compl ete re- im plementat ion of the ge-
nepop sof tware for Windows and Linux. Molecular Ecology Resources,
8,103–106.https://doi.org/10.1111/j.1471-8286.2007.01931.x
Ryder,O.A.(1986).Speciesconservationandsystematics:Thedilemma
of subspecies. Trends in Ecology and Evolution, 1, 9–10. ht tps://d oi.
org /10.1016/0169-5347(86)90059-5
Takezaki, N., Nei, M., & Tamura, K. (2014). POP TREEW: Web version of
POPTREE for constructing population trees from allele frequency
data and computing some other quantities. Molecular Biology and
Evolution, 31,1622–1624.https://doi.org/10.1093/molbev/msu093
Toews, D. P., & Brelsford, A. (2012). The biogeography of mitochondrial
and nuclear discordance in animals. Molecular Ecology, 21, 3 907–
3930.https://doi.org/10.1111/j.1365-294X.2012.05664.x
Wasser, S. K., Brown, L., Mailand, C., Mondol, S., Clark, W., Laurie, C., &
Weir, B. S. (2015). Genetic assignment of large seizures of elep hant
ivory reveals Africa’s major poaching hotspot s. Science, 349, 84–87.
https://doi.org/10.1126/science.aaa2457
Wright,S.(1943).Isolationbydistance.Genetics, 28, 1 14–1 3 8 .
SUPPORTING INFORMATION
Additional Supporting Information may be found online in the
supporting information tab for this article.
How to cite this article: Ishida Y, Gugala NA, Georgiadis NJ,
Roca AL. Evolutionary and demographic processes shaping
geographic patterns of genetic diversit y in a keystone
species, the African forest elephant (Loxodonta cyclotis). Ecol
Evol. 2018;00:1–13. https://doi.org/10.1002/ece3.4062
